{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.055, "global_step": 55000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.000000000000001e-07, "loss": 0.931, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 0.6816, "step": 2 }, { "epoch": 0.0, "learning_rate": 1.5e-06, "loss": 1.2796, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.0000000000000003e-06, "loss": 1.119, "step": 4 }, { "epoch": 0.0, "learning_rate": 2.5e-06, "loss": 0.8744, "step": 5 }, { "epoch": 0.0, "learning_rate": 3e-06, "loss": 0.1561, "step": 6 }, { "epoch": 0.0, "learning_rate": 3.5000000000000004e-06, "loss": 1.0062, "step": 7 }, { "epoch": 0.0, "learning_rate": 4.000000000000001e-06, "loss": 1.1478, "step": 8 }, { "epoch": 0.0, "learning_rate": 4.5e-06, "loss": 0.5617, "step": 9 }, { "epoch": 0.0, "learning_rate": 5e-06, "loss": 0.5931, "step": 10 }, { "epoch": 0.0, "learning_rate": 5.500000000000001e-06, "loss": 0.6567, "step": 11 }, { "epoch": 0.0, "learning_rate": 6e-06, "loss": 0.6785, "step": 12 }, { "epoch": 0.0, "learning_rate": 6.5000000000000004e-06, "loss": 1.0113, "step": 13 }, { "epoch": 0.0, "learning_rate": 7.000000000000001e-06, "loss": 0.8829, "step": 14 }, { "epoch": 0.0, "learning_rate": 7.5e-06, "loss": 0.8225, "step": 15 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-06, "loss": 1.1247, "step": 16 }, { "epoch": 0.0, "learning_rate": 8.500000000000002e-06, "loss": 1.168, "step": 17 }, { "epoch": 0.0, "learning_rate": 9e-06, "loss": 1.0633, "step": 18 }, { "epoch": 0.0, "learning_rate": 9.5e-06, "loss": 1.0441, "step": 19 }, { "epoch": 0.0, "learning_rate": 1e-05, "loss": 1.2362, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.05e-05, "loss": 1.1747, "step": 21 }, { "epoch": 0.0, "learning_rate": 1.1000000000000001e-05, "loss": 0.9891, "step": 22 }, { "epoch": 0.0, "learning_rate": 1.1500000000000002e-05, "loss": 0.8767, "step": 23 }, { "epoch": 0.0, "learning_rate": 1.2e-05, "loss": 0.9538, "step": 24 }, { "epoch": 0.0, "learning_rate": 1.25e-05, "loss": 0.9734, "step": 25 }, { "epoch": 0.0, "learning_rate": 1.3000000000000001e-05, "loss": 0.839, "step": 26 }, { "epoch": 0.0, "learning_rate": 1.3500000000000001e-05, "loss": 0.7487, "step": 27 }, { "epoch": 0.0, "learning_rate": 1.4000000000000001e-05, "loss": 0.9824, "step": 28 }, { "epoch": 0.0, "learning_rate": 1.45e-05, "loss": 1.0037, "step": 29 }, { "epoch": 0.0, "learning_rate": 1.5e-05, "loss": 1.32, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.55e-05, "loss": 0.9389, "step": 31 }, { "epoch": 0.0, "learning_rate": 1.6000000000000003e-05, "loss": 0.9922, "step": 32 }, { "epoch": 0.0, "learning_rate": 1.65e-05, "loss": 0.9843, "step": 33 }, { "epoch": 0.0, "learning_rate": 1.7000000000000003e-05, "loss": 1.0919, "step": 34 }, { "epoch": 0.0, "learning_rate": 1.75e-05, "loss": 0.8617, "step": 35 }, { "epoch": 0.0, "learning_rate": 1.8e-05, "loss": 1.0183, "step": 36 }, { "epoch": 0.0, "learning_rate": 1.85e-05, "loss": 0.9565, "step": 37 }, { "epoch": 0.0, "learning_rate": 1.9e-05, "loss": 0.79, "step": 38 }, { "epoch": 0.0, "learning_rate": 1.9500000000000003e-05, "loss": 1.2117, "step": 39 }, { "epoch": 0.0, "learning_rate": 2e-05, "loss": 1.111, "step": 40 }, { "epoch": 0.0, "learning_rate": 2.05e-05, "loss": 1.8326, "step": 41 }, { "epoch": 0.0, "learning_rate": 2.1e-05, "loss": 1.7969, "step": 42 }, { "epoch": 0.0, "learning_rate": 2.15e-05, "loss": 1.0872, "step": 43 }, { "epoch": 0.0, "learning_rate": 2.2000000000000003e-05, "loss": 1.0279, "step": 44 }, { "epoch": 0.0, "learning_rate": 2.25e-05, "loss": 1.2225, "step": 45 }, { "epoch": 0.0, "learning_rate": 2.3000000000000003e-05, "loss": 0.9716, "step": 46 }, { "epoch": 0.0, "learning_rate": 2.35e-05, "loss": 1.2187, "step": 47 }, { "epoch": 0.0, "learning_rate": 2.4e-05, "loss": 0.871, "step": 48 }, { "epoch": 0.0, "learning_rate": 2.45e-05, "loss": 0.8476, "step": 49 }, { "epoch": 0.0, "learning_rate": 2.5e-05, "loss": 0.8047, "step": 50 }, { "epoch": 0.0, "learning_rate": 2.5500000000000003e-05, "loss": 1.2494, "step": 51 }, { "epoch": 0.0, "learning_rate": 2.6000000000000002e-05, "loss": 1.0532, "step": 52 }, { "epoch": 0.0, "learning_rate": 2.6500000000000004e-05, "loss": 1.0323, "step": 53 }, { "epoch": 0.0, "learning_rate": 2.7000000000000002e-05, "loss": 1.0223, "step": 54 }, { "epoch": 0.0, "learning_rate": 2.7500000000000004e-05, "loss": 0.9494, "step": 55 }, { "epoch": 0.0, "learning_rate": 2.8000000000000003e-05, "loss": 1.0272, "step": 56 }, { "epoch": 0.0, "learning_rate": 2.8499999999999998e-05, "loss": 1.2829, "step": 57 }, { "epoch": 0.0, "learning_rate": 2.9e-05, "loss": 1.2682, "step": 58 }, { "epoch": 0.0, "learning_rate": 2.95e-05, "loss": 1.6736, "step": 59 }, { "epoch": 0.0, "learning_rate": 3e-05, "loss": 2.3082, "step": 60 }, { "epoch": 0.0, "learning_rate": 3.05e-05, "loss": 1.1302, "step": 61 }, { "epoch": 0.0, "learning_rate": 3.1e-05, "loss": 0.7822, "step": 62 }, { "epoch": 0.0, "learning_rate": 3.15e-05, "loss": 0.9861, "step": 63 }, { "epoch": 0.0, "learning_rate": 3.2000000000000005e-05, "loss": 0.9721, "step": 64 }, { "epoch": 0.0, "learning_rate": 3.2500000000000004e-05, "loss": 0.6753, "step": 65 }, { "epoch": 0.0, "learning_rate": 3.3e-05, "loss": 1.047, "step": 66 }, { "epoch": 0.0, "learning_rate": 3.35e-05, "loss": 1.0422, "step": 67 }, { "epoch": 0.0, "learning_rate": 3.4000000000000007e-05, "loss": 0.9134, "step": 68 }, { "epoch": 0.0, "learning_rate": 3.45e-05, "loss": 1.5718, "step": 69 }, { "epoch": 0.0, "learning_rate": 3.5e-05, "loss": 1.0625, "step": 70 }, { "epoch": 0.0, "learning_rate": 3.55e-05, "loss": 1.0779, "step": 71 }, { "epoch": 0.0, "learning_rate": 3.6e-05, "loss": 1.1702, "step": 72 }, { "epoch": 0.0, "learning_rate": 3.65e-05, "loss": 1.1024, "step": 73 }, { "epoch": 0.0, "learning_rate": 3.7e-05, "loss": 1.0382, "step": 74 }, { "epoch": 0.0, "learning_rate": 3.7500000000000003e-05, "loss": 1.0676, "step": 75 }, { "epoch": 0.0, "learning_rate": 3.8e-05, "loss": 1.2035, "step": 76 }, { "epoch": 0.0, "learning_rate": 3.85e-05, "loss": 1.0576, "step": 77 }, { "epoch": 0.0, "learning_rate": 3.9000000000000006e-05, "loss": 1.1175, "step": 78 }, { "epoch": 0.0, "learning_rate": 3.9500000000000005e-05, "loss": 1.0663, "step": 79 }, { "epoch": 0.0, "learning_rate": 4e-05, "loss": 1.2545, "step": 80 }, { "epoch": 0.0, "learning_rate": 4.05e-05, "loss": 0.7244, "step": 81 }, { "epoch": 0.0, "learning_rate": 4.1e-05, "loss": 1.1455, "step": 82 }, { "epoch": 0.0, "learning_rate": 4.15e-05, "loss": 1.1514, "step": 83 }, { "epoch": 0.0, "learning_rate": 4.2e-05, "loss": 1.0883, "step": 84 }, { "epoch": 0.0, "learning_rate": 4.25e-05, "loss": 0.8759, "step": 85 }, { "epoch": 0.0, "learning_rate": 4.3e-05, "loss": 1.0519, "step": 86 }, { "epoch": 0.0, "learning_rate": 4.35e-05, "loss": 1.1014, "step": 87 }, { "epoch": 0.0, "learning_rate": 4.4000000000000006e-05, "loss": 0.918, "step": 88 }, { "epoch": 0.0, "learning_rate": 4.4500000000000004e-05, "loss": 1.0174, "step": 89 }, { "epoch": 0.0, "learning_rate": 4.5e-05, "loss": 1.0974, "step": 90 }, { "epoch": 0.0, "learning_rate": 4.55e-05, "loss": 1.0889, "step": 91 }, { "epoch": 0.0, "learning_rate": 4.600000000000001e-05, "loss": 1.0881, "step": 92 }, { "epoch": 0.0, "learning_rate": 4.6500000000000005e-05, "loss": 0.8923, "step": 93 }, { "epoch": 0.0, "learning_rate": 4.7e-05, "loss": 1.0121, "step": 94 }, { "epoch": 0.0, "learning_rate": 4.75e-05, "loss": 1.3273, "step": 95 }, { "epoch": 0.0, "learning_rate": 4.8e-05, "loss": 1.0579, "step": 96 }, { "epoch": 0.0, "learning_rate": 4.85e-05, "loss": 1.149, "step": 97 }, { "epoch": 0.0, "learning_rate": 4.9e-05, "loss": 1.4682, "step": 98 }, { "epoch": 0.0, "learning_rate": 4.9500000000000004e-05, "loss": 1.4034, "step": 99 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 0.8914, "step": 100 }, { "epoch": 0.0, "learning_rate": 4.999999999987661e-05, "loss": 1.0526, "step": 101 }, { "epoch": 0.0, "learning_rate": 4.999999999950642e-05, "loss": 1.2849, "step": 102 }, { "epoch": 0.0, "learning_rate": 4.999999999888945e-05, "loss": 1.2292, "step": 103 }, { "epoch": 0.0, "learning_rate": 4.999999999802569e-05, "loss": 1.0693, "step": 104 }, { "epoch": 0.0, "learning_rate": 4.9999999996915136e-05, "loss": 0.952, "step": 105 }, { "epoch": 0.0, "learning_rate": 4.999999999555779e-05, "loss": 1.1392, "step": 106 }, { "epoch": 0.0, "learning_rate": 4.9999999993953664e-05, "loss": 1.0248, "step": 107 }, { "epoch": 0.0, "learning_rate": 4.999999999210274e-05, "loss": 1.2939, "step": 108 }, { "epoch": 0.0, "learning_rate": 4.9999999990005034e-05, "loss": 1.2429, "step": 109 }, { "epoch": 0.0, "learning_rate": 4.999999998766053e-05, "loss": 1.1394, "step": 110 }, { "epoch": 0.0, "learning_rate": 4.999999998506924e-05, "loss": 1.3022, "step": 111 }, { "epoch": 0.0, "learning_rate": 4.999999998223116e-05, "loss": 0.9869, "step": 112 }, { "epoch": 0.0, "learning_rate": 4.9999999979146295e-05, "loss": 1.2054, "step": 113 }, { "epoch": 0.0, "learning_rate": 4.999999997581464e-05, "loss": 1.1226, "step": 114 }, { "epoch": 0.0, "learning_rate": 4.999999997223619e-05, "loss": 1.0491, "step": 115 }, { "epoch": 0.0, "learning_rate": 4.999999996841095e-05, "loss": 0.8425, "step": 116 }, { "epoch": 0.0, "learning_rate": 4.9999999964338926e-05, "loss": 1.305, "step": 117 }, { "epoch": 0.0, "learning_rate": 4.999999996002011e-05, "loss": 1.0233, "step": 118 }, { "epoch": 0.0, "learning_rate": 4.99999999554545e-05, "loss": 0.5656, "step": 119 }, { "epoch": 0.0, "learning_rate": 4.9999999950642105e-05, "loss": 1.0478, "step": 120 }, { "epoch": 0.0, "learning_rate": 4.9999999945582926e-05, "loss": 1.191, "step": 121 }, { "epoch": 0.0, "learning_rate": 4.9999999940276956e-05, "loss": 1.3638, "step": 122 }, { "epoch": 0.0, "learning_rate": 4.999999993472419e-05, "loss": 0.355, "step": 123 }, { "epoch": 0.0, "learning_rate": 4.9999999928924634e-05, "loss": 1.095, "step": 124 }, { "epoch": 0.0, "learning_rate": 4.9999999922878296e-05, "loss": 1.151, "step": 125 }, { "epoch": 0.0, "learning_rate": 4.999999991658516e-05, "loss": 1.4362, "step": 126 }, { "epoch": 0.0, "learning_rate": 4.999999991004524e-05, "loss": 1.0988, "step": 127 }, { "epoch": 0.0, "learning_rate": 4.999999990325853e-05, "loss": 0.8029, "step": 128 }, { "epoch": 0.0, "learning_rate": 4.9999999896225036e-05, "loss": 1.2147, "step": 129 }, { "epoch": 0.0, "learning_rate": 4.999999988894474e-05, "loss": 1.1023, "step": 130 }, { "epoch": 0.0, "learning_rate": 4.9999999881417667e-05, "loss": 0.9649, "step": 131 }, { "epoch": 0.0, "learning_rate": 4.99999998736438e-05, "loss": 1.059, "step": 132 }, { "epoch": 0.0, "learning_rate": 4.999999986562314e-05, "loss": 1.4863, "step": 133 }, { "epoch": 0.0, "learning_rate": 4.99999998573557e-05, "loss": 1.4881, "step": 134 }, { "epoch": 0.0, "learning_rate": 4.999999984884146e-05, "loss": 1.0374, "step": 135 }, { "epoch": 0.0, "learning_rate": 4.9999999840080426e-05, "loss": 1.0476, "step": 136 }, { "epoch": 0.0, "learning_rate": 4.999999983107262e-05, "loss": 1.0538, "step": 137 }, { "epoch": 0.0, "learning_rate": 4.999999982181801e-05, "loss": 0.8671, "step": 138 }, { "epoch": 0.0, "learning_rate": 4.999999981231661e-05, "loss": 1.1986, "step": 139 }, { "epoch": 0.0, "learning_rate": 4.9999999802568433e-05, "loss": 1.2, "step": 140 }, { "epoch": 0.0, "learning_rate": 4.999999979257346e-05, "loss": 1.1648, "step": 141 }, { "epoch": 0.0, "learning_rate": 4.99999997823317e-05, "loss": 1.0579, "step": 142 }, { "epoch": 0.0, "learning_rate": 4.999999977184314e-05, "loss": 1.0379, "step": 143 }, { "epoch": 0.0, "learning_rate": 4.99999997611078e-05, "loss": 0.5544, "step": 144 }, { "epoch": 0.0, "learning_rate": 4.999999975012567e-05, "loss": 0.9992, "step": 145 }, { "epoch": 0.0, "learning_rate": 4.999999973889675e-05, "loss": 1.2599, "step": 146 }, { "epoch": 0.0, "learning_rate": 4.999999972742104e-05, "loss": 1.3262, "step": 147 }, { "epoch": 0.0, "learning_rate": 4.9999999715698544e-05, "loss": 1.3029, "step": 148 }, { "epoch": 0.0, "learning_rate": 4.999999970372925e-05, "loss": 1.238, "step": 149 }, { "epoch": 0.0, "learning_rate": 4.999999969151317e-05, "loss": 1.1567, "step": 150 }, { "epoch": 0.0, "learning_rate": 4.99999996790503e-05, "loss": 1.1385, "step": 151 }, { "epoch": 0.0, "learning_rate": 4.9999999666340646e-05, "loss": 1.068, "step": 152 }, { "epoch": 0.0, "learning_rate": 4.99999996533842e-05, "loss": 1.2784, "step": 153 }, { "epoch": 0.0, "learning_rate": 4.999999964018096e-05, "loss": 0.7641, "step": 154 }, { "epoch": 0.0, "learning_rate": 4.999999962673094e-05, "loss": 0.6886, "step": 155 }, { "epoch": 0.0, "learning_rate": 4.999999961303412e-05, "loss": 1.1172, "step": 156 }, { "epoch": 0.0, "learning_rate": 4.999999959909052e-05, "loss": 0.847, "step": 157 }, { "epoch": 0.0, "learning_rate": 4.999999958490012e-05, "loss": 0.8203, "step": 158 }, { "epoch": 0.0, "learning_rate": 4.999999957046294e-05, "loss": 1.3131, "step": 159 }, { "epoch": 0.0, "learning_rate": 4.999999955577897e-05, "loss": 1.4838, "step": 160 }, { "epoch": 0.0, "learning_rate": 4.99999995408482e-05, "loss": 1.1396, "step": 161 }, { "epoch": 0.0, "learning_rate": 4.9999999525670646e-05, "loss": 1.3872, "step": 162 }, { "epoch": 0.0, "learning_rate": 4.999999951024631e-05, "loss": 1.9415, "step": 163 }, { "epoch": 0.0, "learning_rate": 4.999999949457518e-05, "loss": 1.5973, "step": 164 }, { "epoch": 0.0, "learning_rate": 4.9999999478657254e-05, "loss": 1.2385, "step": 165 }, { "epoch": 0.0, "learning_rate": 4.9999999462492545e-05, "loss": 1.2302, "step": 166 }, { "epoch": 0.0, "learning_rate": 4.9999999446081045e-05, "loss": 1.08, "step": 167 }, { "epoch": 0.0, "learning_rate": 4.999999942942276e-05, "loss": 1.3935, "step": 168 }, { "epoch": 0.0, "learning_rate": 4.999999941251768e-05, "loss": 1.2663, "step": 169 }, { "epoch": 0.0, "learning_rate": 4.9999999395365814e-05, "loss": 1.3094, "step": 170 }, { "epoch": 0.0, "learning_rate": 4.9999999377967155e-05, "loss": 1.2008, "step": 171 }, { "epoch": 0.0, "learning_rate": 4.999999936032172e-05, "loss": 0.8156, "step": 172 }, { "epoch": 0.0, "learning_rate": 4.9999999342429474e-05, "loss": 4.8589, "step": 173 }, { "epoch": 0.0, "learning_rate": 4.9999999324290446e-05, "loss": 6.0248, "step": 174 }, { "epoch": 0.0, "learning_rate": 4.9999999305904634e-05, "loss": 5.9313, "step": 175 }, { "epoch": 0.0, "learning_rate": 4.999999928727202e-05, "loss": 5.9243, "step": 176 }, { "epoch": 0.0, "learning_rate": 4.999999926839264e-05, "loss": 5.7681, "step": 177 }, { "epoch": 0.0, "learning_rate": 4.999999924926645e-05, "loss": 5.6746, "step": 178 }, { "epoch": 0.0, "learning_rate": 4.9999999229893476e-05, "loss": 5.7433, "step": 179 }, { "epoch": 0.0, "learning_rate": 4.999999921027372e-05, "loss": 5.7031, "step": 180 }, { "epoch": 0.0, "learning_rate": 4.9999999190407165e-05, "loss": 5.5989, "step": 181 }, { "epoch": 0.0, "learning_rate": 4.999999917029382e-05, "loss": 5.24, "step": 182 }, { "epoch": 0.0, "learning_rate": 4.9999999149933695e-05, "loss": 5.0161, "step": 183 }, { "epoch": 0.0, "learning_rate": 4.999999912932677e-05, "loss": 4.415, "step": 184 }, { "epoch": 0.0, "learning_rate": 4.999999910847306e-05, "loss": 5.2371, "step": 185 }, { "epoch": 0.0, "learning_rate": 4.999999908737256e-05, "loss": 5.5717, "step": 186 }, { "epoch": 0.0, "learning_rate": 4.9999999066025276e-05, "loss": 5.6388, "step": 187 }, { "epoch": 0.0, "learning_rate": 4.99999990444312e-05, "loss": 5.6072, "step": 188 }, { "epoch": 0.0, "learning_rate": 4.999999902259033e-05, "loss": 5.6016, "step": 189 }, { "epoch": 0.0, "learning_rate": 4.9999999000502676e-05, "loss": 5.5708, "step": 190 }, { "epoch": 0.0, "learning_rate": 4.999999897816823e-05, "loss": 5.5852, "step": 191 }, { "epoch": 0.0, "learning_rate": 4.999999895558699e-05, "loss": 5.572, "step": 192 }, { "epoch": 0.0, "learning_rate": 4.999999893275897e-05, "loss": 5.529, "step": 193 }, { "epoch": 0.0, "learning_rate": 4.9999998909684156e-05, "loss": 5.5539, "step": 194 }, { "epoch": 0.0, "learning_rate": 4.999999888636255e-05, "loss": 5.563, "step": 195 }, { "epoch": 0.0, "learning_rate": 4.9999998862794154e-05, "loss": 2.3107, "step": 196 }, { "epoch": 0.0, "learning_rate": 4.999999883897897e-05, "loss": 1.4454, "step": 197 }, { "epoch": 0.0, "learning_rate": 4.9999998814917e-05, "loss": 1.4917, "step": 198 }, { "epoch": 0.0, "learning_rate": 4.999999879060824e-05, "loss": 1.0891, "step": 199 }, { "epoch": 0.0, "learning_rate": 4.9999998766052684e-05, "loss": 0.7431, "step": 200 }, { "epoch": 0.0, "learning_rate": 4.999999874125034e-05, "loss": 1.0081, "step": 201 }, { "epoch": 0.0, "learning_rate": 4.999999871620121e-05, "loss": 1.4269, "step": 202 }, { "epoch": 0.0, "learning_rate": 4.999999869090529e-05, "loss": 1.1632, "step": 203 }, { "epoch": 0.0, "learning_rate": 4.9999998665362585e-05, "loss": 0.9132, "step": 204 }, { "epoch": 0.0, "learning_rate": 4.999999863957309e-05, "loss": 0.708, "step": 205 }, { "epoch": 0.0, "learning_rate": 4.99999986135368e-05, "loss": 1.159, "step": 206 }, { "epoch": 0.0, "learning_rate": 4.999999858725372e-05, "loss": 1.2212, "step": 207 }, { "epoch": 0.0, "learning_rate": 4.9999998560723855e-05, "loss": 1.1131, "step": 208 }, { "epoch": 0.0, "learning_rate": 4.99999985339472e-05, "loss": 0.9072, "step": 209 }, { "epoch": 0.0, "learning_rate": 4.999999850692375e-05, "loss": 1.3964, "step": 210 }, { "epoch": 0.0, "learning_rate": 4.999999847965352e-05, "loss": 1.2253, "step": 211 }, { "epoch": 0.0, "learning_rate": 4.999999845213649e-05, "loss": 0.8362, "step": 212 }, { "epoch": 0.0, "learning_rate": 4.999999842437267e-05, "loss": 0.6663, "step": 213 }, { "epoch": 0.0, "learning_rate": 4.999999839636207e-05, "loss": 1.006, "step": 214 }, { "epoch": 0.0, "learning_rate": 4.9999998368104686e-05, "loss": 1.6042, "step": 215 }, { "epoch": 0.0, "learning_rate": 4.9999998339600496e-05, "loss": 1.1701, "step": 216 }, { "epoch": 0.0, "learning_rate": 4.999999831084953e-05, "loss": 1.1874, "step": 217 }, { "epoch": 0.0, "learning_rate": 4.999999828185177e-05, "loss": 1.1069, "step": 218 }, { "epoch": 0.0, "learning_rate": 4.999999825260722e-05, "loss": 0.9765, "step": 219 }, { "epoch": 0.0, "learning_rate": 4.9999998223115875e-05, "loss": 0.8295, "step": 220 }, { "epoch": 0.0, "learning_rate": 4.999999819337774e-05, "loss": 1.012, "step": 221 }, { "epoch": 0.0, "learning_rate": 4.999999816339282e-05, "loss": 1.1255, "step": 222 }, { "epoch": 0.0, "learning_rate": 4.999999813316111e-05, "loss": 1.2156, "step": 223 }, { "epoch": 0.0, "learning_rate": 4.9999998102682616e-05, "loss": 1.0133, "step": 224 }, { "epoch": 0.0, "learning_rate": 4.999999807195733e-05, "loss": 1.111, "step": 225 }, { "epoch": 0.0, "learning_rate": 4.999999804098525e-05, "loss": 1.0167, "step": 226 }, { "epoch": 0.0, "learning_rate": 4.999999800976639e-05, "loss": 1.2424, "step": 227 }, { "epoch": 0.0, "learning_rate": 4.999999797830073e-05, "loss": 1.1881, "step": 228 }, { "epoch": 0.0, "learning_rate": 4.999999794658828e-05, "loss": 1.2338, "step": 229 }, { "epoch": 0.0, "learning_rate": 4.999999791462905e-05, "loss": 1.277, "step": 230 }, { "epoch": 0.0, "learning_rate": 4.999999788242302e-05, "loss": 1.1273, "step": 231 }, { "epoch": 0.0, "learning_rate": 4.999999784997021e-05, "loss": 1.1403, "step": 232 }, { "epoch": 0.0, "learning_rate": 4.99999978172706e-05, "loss": 1.1259, "step": 233 }, { "epoch": 0.0, "learning_rate": 4.999999778432421e-05, "loss": 1.1319, "step": 234 }, { "epoch": 0.0, "learning_rate": 4.999999775113103e-05, "loss": 1.1755, "step": 235 }, { "epoch": 0.0, "learning_rate": 4.999999771769106e-05, "loss": 0.8809, "step": 236 }, { "epoch": 0.0, "learning_rate": 4.999999768400429e-05, "loss": 0.8313, "step": 237 }, { "epoch": 0.0, "learning_rate": 4.999999765007075e-05, "loss": 0.4808, "step": 238 }, { "epoch": 0.0, "learning_rate": 4.999999761589041e-05, "loss": 0.3866, "step": 239 }, { "epoch": 0.0, "learning_rate": 4.999999758146328e-05, "loss": 0.4625, "step": 240 }, { "epoch": 0.0, "learning_rate": 4.999999754678936e-05, "loss": 1.005, "step": 241 }, { "epoch": 0.0, "learning_rate": 4.999999751186866e-05, "loss": 0.478, "step": 242 }, { "epoch": 0.0, "learning_rate": 4.999999747670115e-05, "loss": 0.4184, "step": 243 }, { "epoch": 0.0, "learning_rate": 4.999999744128687e-05, "loss": 0.3674, "step": 244 }, { "epoch": 0.0, "learning_rate": 4.999999740562579e-05, "loss": 0.3536, "step": 245 }, { "epoch": 0.0, "learning_rate": 4.999999736971793e-05, "loss": 0.3212, "step": 246 }, { "epoch": 0.0, "learning_rate": 4.999999733356327e-05, "loss": 0.3954, "step": 247 }, { "epoch": 0.0, "learning_rate": 4.9999997297161825e-05, "loss": 0.5134, "step": 248 }, { "epoch": 0.0, "learning_rate": 4.999999726051359e-05, "loss": 0.2877, "step": 249 }, { "epoch": 0.0, "learning_rate": 4.999999722361857e-05, "loss": 0.3694, "step": 250 }, { "epoch": 0.0, "learning_rate": 4.999999718647676e-05, "loss": 0.4496, "step": 251 }, { "epoch": 0.0, "learning_rate": 4.9999997149088154e-05, "loss": 0.2565, "step": 252 }, { "epoch": 0.0, "learning_rate": 4.999999711145276e-05, "loss": 0.3561, "step": 253 }, { "epoch": 0.0, "learning_rate": 4.999999707357058e-05, "loss": 0.3123, "step": 254 }, { "epoch": 0.0, "learning_rate": 4.99999970354416e-05, "loss": 0.2779, "step": 255 }, { "epoch": 0.0, "learning_rate": 4.9999996997065845e-05, "loss": 0.6054, "step": 256 }, { "epoch": 0.0, "learning_rate": 4.99999969584433e-05, "loss": 1.317, "step": 257 }, { "epoch": 0.0, "learning_rate": 4.999999691957395e-05, "loss": 1.1632, "step": 258 }, { "epoch": 0.0, "learning_rate": 4.9999996880457834e-05, "loss": 1.1206, "step": 259 }, { "epoch": 0.0, "learning_rate": 4.999999684109491e-05, "loss": 1.2338, "step": 260 }, { "epoch": 0.0, "learning_rate": 4.9999996801485205e-05, "loss": 1.0508, "step": 261 }, { "epoch": 0.0, "learning_rate": 4.999999676162871e-05, "loss": 1.3887, "step": 262 }, { "epoch": 0.0, "learning_rate": 4.9999996721525424e-05, "loss": 1.1023, "step": 263 }, { "epoch": 0.0, "learning_rate": 4.999999668117534e-05, "loss": 1.3006, "step": 264 }, { "epoch": 0.0, "learning_rate": 4.9999996640578476e-05, "loss": 1.1455, "step": 265 }, { "epoch": 0.0, "learning_rate": 4.999999659973482e-05, "loss": 1.1464, "step": 266 }, { "epoch": 0.0, "learning_rate": 4.9999996558644376e-05, "loss": 1.1131, "step": 267 }, { "epoch": 0.0, "learning_rate": 4.999999651730715e-05, "loss": 1.2394, "step": 268 }, { "epoch": 0.0, "learning_rate": 4.9999996475723124e-05, "loss": 1.2348, "step": 269 }, { "epoch": 0.0, "learning_rate": 4.9999996433892314e-05, "loss": 1.2294, "step": 270 }, { "epoch": 0.0, "learning_rate": 4.9999996391814705e-05, "loss": 0.9849, "step": 271 }, { "epoch": 0.0, "learning_rate": 4.999999634949032e-05, "loss": 1.1667, "step": 272 }, { "epoch": 0.0, "learning_rate": 4.999999630691914e-05, "loss": 1.28, "step": 273 }, { "epoch": 0.0, "learning_rate": 4.9999996264101164e-05, "loss": 1.1602, "step": 274 }, { "epoch": 0.0, "learning_rate": 4.99999962210364e-05, "loss": 1.2452, "step": 275 }, { "epoch": 0.0, "learning_rate": 4.9999996177724864e-05, "loss": 1.4849, "step": 276 }, { "epoch": 0.0, "learning_rate": 4.999999613416652e-05, "loss": 1.1768, "step": 277 }, { "epoch": 0.0, "learning_rate": 4.999999609036139e-05, "loss": 1.0437, "step": 278 }, { "epoch": 0.0, "learning_rate": 4.999999604630947e-05, "loss": 0.8973, "step": 279 }, { "epoch": 0.0, "learning_rate": 4.9999996002010766e-05, "loss": 1.0841, "step": 280 }, { "epoch": 0.0, "learning_rate": 4.9999995957465276e-05, "loss": 0.7193, "step": 281 }, { "epoch": 0.0, "learning_rate": 4.999999591267299e-05, "loss": 1.3937, "step": 282 }, { "epoch": 0.0, "learning_rate": 4.999999586763391e-05, "loss": 1.2132, "step": 283 }, { "epoch": 0.0, "learning_rate": 4.9999995822348044e-05, "loss": 1.2343, "step": 284 }, { "epoch": 0.0, "learning_rate": 4.9999995776815395e-05, "loss": 1.1092, "step": 285 }, { "epoch": 0.0, "learning_rate": 4.9999995731035955e-05, "loss": 1.2969, "step": 286 }, { "epoch": 0.0, "learning_rate": 4.9999995685009716e-05, "loss": 0.8917, "step": 287 }, { "epoch": 0.0, "learning_rate": 4.99999956387367e-05, "loss": 0.9051, "step": 288 }, { "epoch": 0.0, "learning_rate": 4.9999995592216884e-05, "loss": 1.0114, "step": 289 }, { "epoch": 0.0, "learning_rate": 4.9999995545450284e-05, "loss": 0.9757, "step": 290 }, { "epoch": 0.0, "learning_rate": 4.999999549843689e-05, "loss": 1.237, "step": 291 }, { "epoch": 0.0, "learning_rate": 4.999999545117671e-05, "loss": 1.1462, "step": 292 }, { "epoch": 0.0, "learning_rate": 4.999999540366974e-05, "loss": 1.1132, "step": 293 }, { "epoch": 0.0, "learning_rate": 4.999999535591598e-05, "loss": 1.1785, "step": 294 }, { "epoch": 0.0, "learning_rate": 4.999999530791544e-05, "loss": 1.0634, "step": 295 }, { "epoch": 0.0, "learning_rate": 4.9999995259668096e-05, "loss": 1.3001, "step": 296 }, { "epoch": 0.0, "learning_rate": 4.999999521117397e-05, "loss": 1.7609, "step": 297 }, { "epoch": 0.0, "learning_rate": 4.999999516243306e-05, "loss": 0.939, "step": 298 }, { "epoch": 0.0, "learning_rate": 4.9999995113445354e-05, "loss": 1.0552, "step": 299 }, { "epoch": 0.0, "learning_rate": 4.999999506421086e-05, "loss": 0.9572, "step": 300 }, { "epoch": 0.0, "learning_rate": 4.999999501472957e-05, "loss": 0.925, "step": 301 }, { "epoch": 0.0, "learning_rate": 4.9999994965001495e-05, "loss": 0.8796, "step": 302 }, { "epoch": 0.0, "learning_rate": 4.999999491502664e-05, "loss": 0.9126, "step": 303 }, { "epoch": 0.0, "learning_rate": 4.999999486480498e-05, "loss": 1.2216, "step": 304 }, { "epoch": 0.0, "learning_rate": 4.999999481433654e-05, "loss": 1.1434, "step": 305 }, { "epoch": 0.0, "learning_rate": 4.99999947636213e-05, "loss": 1.1265, "step": 306 }, { "epoch": 0.0, "learning_rate": 4.999999471265929e-05, "loss": 1.1126, "step": 307 }, { "epoch": 0.0, "learning_rate": 4.999999466145048e-05, "loss": 1.0995, "step": 308 }, { "epoch": 0.0, "learning_rate": 4.9999994609994876e-05, "loss": 0.9912, "step": 309 }, { "epoch": 0.0, "learning_rate": 4.9999994558292486e-05, "loss": 1.1196, "step": 310 }, { "epoch": 0.0, "learning_rate": 4.9999994506343305e-05, "loss": 1.2063, "step": 311 }, { "epoch": 0.0, "learning_rate": 4.999999445414734e-05, "loss": 1.1179, "step": 312 }, { "epoch": 0.0, "learning_rate": 4.999999440170459e-05, "loss": 1.292, "step": 313 }, { "epoch": 0.0, "learning_rate": 4.999999434901503e-05, "loss": 1.1431, "step": 314 }, { "epoch": 0.0, "learning_rate": 4.99999942960787e-05, "loss": 1.0945, "step": 315 }, { "epoch": 0.0, "learning_rate": 4.9999994242895574e-05, "loss": 1.1347, "step": 316 }, { "epoch": 0.0, "learning_rate": 4.999999418946566e-05, "loss": 1.0005, "step": 317 }, { "epoch": 0.0, "learning_rate": 4.9999994135788956e-05, "loss": 1.0945, "step": 318 }, { "epoch": 0.0, "learning_rate": 4.9999994081865456e-05, "loss": 1.1316, "step": 319 }, { "epoch": 0.0, "learning_rate": 4.999999402769517e-05, "loss": 1.1023, "step": 320 }, { "epoch": 0.0, "learning_rate": 4.99999939732781e-05, "loss": 1.2254, "step": 321 }, { "epoch": 0.0, "learning_rate": 4.999999391861424e-05, "loss": 1.1501, "step": 322 }, { "epoch": 0.0, "learning_rate": 4.999999386370359e-05, "loss": 0.8656, "step": 323 }, { "epoch": 0.0, "learning_rate": 4.999999380854615e-05, "loss": 1.197, "step": 324 }, { "epoch": 0.0, "learning_rate": 4.9999993753141917e-05, "loss": 1.109, "step": 325 }, { "epoch": 0.0, "learning_rate": 4.9999993697490896e-05, "loss": 1.1346, "step": 326 }, { "epoch": 0.0, "learning_rate": 4.999999364159309e-05, "loss": 1.0394, "step": 327 }, { "epoch": 0.0, "learning_rate": 4.999999358544849e-05, "loss": 1.27, "step": 328 }, { "epoch": 0.0, "learning_rate": 4.99999935290571e-05, "loss": 1.2734, "step": 329 }, { "epoch": 0.0, "learning_rate": 4.999999347241893e-05, "loss": 1.0557, "step": 330 }, { "epoch": 0.0, "learning_rate": 4.999999341553396e-05, "loss": 1.1114, "step": 331 }, { "epoch": 0.0, "learning_rate": 4.99999933584022e-05, "loss": 1.4432, "step": 332 }, { "epoch": 0.0, "learning_rate": 4.9999993301023654e-05, "loss": 0.9807, "step": 333 }, { "epoch": 0.0, "learning_rate": 4.999999324339832e-05, "loss": 1.1951, "step": 334 }, { "epoch": 0.0, "learning_rate": 4.999999318552619e-05, "loss": 1.0012, "step": 335 }, { "epoch": 0.0, "learning_rate": 4.9999993127407284e-05, "loss": 1.1595, "step": 336 }, { "epoch": 0.0, "learning_rate": 4.9999993069041584e-05, "loss": 1.1289, "step": 337 }, { "epoch": 0.0, "learning_rate": 4.999999301042909e-05, "loss": 1.2733, "step": 338 }, { "epoch": 0.0, "learning_rate": 4.999999295156981e-05, "loss": 1.1768, "step": 339 }, { "epoch": 0.0, "learning_rate": 4.999999289246373e-05, "loss": 1.2367, "step": 340 }, { "epoch": 0.0, "learning_rate": 4.999999283311088e-05, "loss": 1.0992, "step": 341 }, { "epoch": 0.0, "learning_rate": 4.999999277351122e-05, "loss": 1.0069, "step": 342 }, { "epoch": 0.0, "learning_rate": 4.9999992713664784e-05, "loss": 0.9432, "step": 343 }, { "epoch": 0.0, "learning_rate": 4.999999265357155e-05, "loss": 1.1886, "step": 344 }, { "epoch": 0.0, "learning_rate": 4.999999259323154e-05, "loss": 1.1029, "step": 345 }, { "epoch": 0.0, "learning_rate": 4.999999253264473e-05, "loss": 0.7799, "step": 346 }, { "epoch": 0.0, "learning_rate": 4.999999247181113e-05, "loss": 1.2012, "step": 347 }, { "epoch": 0.0, "learning_rate": 4.999999241073074e-05, "loss": 0.8971, "step": 348 }, { "epoch": 0.0, "learning_rate": 4.999999234940357e-05, "loss": 1.1842, "step": 349 }, { "epoch": 0.0, "learning_rate": 4.99999922878296e-05, "loss": 1.114, "step": 350 }, { "epoch": 0.0, "learning_rate": 4.999999222600885e-05, "loss": 1.3876, "step": 351 }, { "epoch": 0.0, "learning_rate": 4.99999921639413e-05, "loss": 1.2368, "step": 352 }, { "epoch": 0.0, "learning_rate": 4.9999992101626966e-05, "loss": 1.2026, "step": 353 }, { "epoch": 0.0, "learning_rate": 4.9999992039065844e-05, "loss": 1.1897, "step": 354 }, { "epoch": 0.0, "learning_rate": 4.999999197625794e-05, "loss": 1.0612, "step": 355 }, { "epoch": 0.0, "learning_rate": 4.999999191320324e-05, "loss": 1.0526, "step": 356 }, { "epoch": 0.0, "learning_rate": 4.999999184990174e-05, "loss": 1.1144, "step": 357 }, { "epoch": 0.0, "learning_rate": 4.999999178635346e-05, "loss": 1.3094, "step": 358 }, { "epoch": 0.0, "learning_rate": 4.999999172255839e-05, "loss": 0.4662, "step": 359 }, { "epoch": 0.0, "learning_rate": 4.999999165851653e-05, "loss": 0.2525, "step": 360 }, { "epoch": 0.0, "learning_rate": 4.9999991594227885e-05, "loss": 0.8735, "step": 361 }, { "epoch": 0.0, "learning_rate": 4.9999991529692444e-05, "loss": 1.0231, "step": 362 }, { "epoch": 0.0, "learning_rate": 4.9999991464910226e-05, "loss": 1.2666, "step": 363 }, { "epoch": 0.0, "learning_rate": 4.999999139988121e-05, "loss": 1.1313, "step": 364 }, { "epoch": 0.0, "learning_rate": 4.99999913346054e-05, "loss": 1.2996, "step": 365 }, { "epoch": 0.0, "learning_rate": 4.99999912690828e-05, "loss": 1.1301, "step": 366 }, { "epoch": 0.0, "learning_rate": 4.999999120331342e-05, "loss": 1.1243, "step": 367 }, { "epoch": 0.0, "learning_rate": 4.999999113729724e-05, "loss": 1.3049, "step": 368 }, { "epoch": 0.0, "learning_rate": 4.999999107103428e-05, "loss": 1.3117, "step": 369 }, { "epoch": 0.0, "learning_rate": 4.999999100452453e-05, "loss": 1.1549, "step": 370 }, { "epoch": 0.0, "learning_rate": 4.999999093776798e-05, "loss": 1.0233, "step": 371 }, { "epoch": 0.0, "learning_rate": 4.999999087076465e-05, "loss": 1.3463, "step": 372 }, { "epoch": 0.0, "learning_rate": 4.999999080351453e-05, "loss": 1.0907, "step": 373 }, { "epoch": 0.0, "learning_rate": 4.999999073601762e-05, "loss": 0.9733, "step": 374 }, { "epoch": 0.0, "learning_rate": 4.999999066827392e-05, "loss": 1.4368, "step": 375 }, { "epoch": 0.0, "learning_rate": 4.999999060028343e-05, "loss": 5.9099, "step": 376 }, { "epoch": 0.0, "learning_rate": 4.9999990532046156e-05, "loss": 3.7277, "step": 377 }, { "epoch": 0.0, "learning_rate": 4.9999990463562086e-05, "loss": 1.1872, "step": 378 }, { "epoch": 0.0, "learning_rate": 4.999999039483123e-05, "loss": 1.0093, "step": 379 }, { "epoch": 0.0, "learning_rate": 4.9999990325853577e-05, "loss": 1.0568, "step": 380 }, { "epoch": 0.0, "learning_rate": 4.9999990256629145e-05, "loss": 0.831, "step": 381 }, { "epoch": 0.0, "learning_rate": 4.999999018715792e-05, "loss": 0.9143, "step": 382 }, { "epoch": 0.0, "learning_rate": 4.999999011743991e-05, "loss": 1.6355, "step": 383 }, { "epoch": 0.0, "learning_rate": 4.99999900474751e-05, "loss": 1.4826, "step": 384 }, { "epoch": 0.0, "learning_rate": 4.99999899772635e-05, "loss": 1.0458, "step": 385 }, { "epoch": 0.0, "learning_rate": 4.999998990680512e-05, "loss": 1.1318, "step": 386 }, { "epoch": 0.0, "learning_rate": 4.9999989836099946e-05, "loss": 1.0718, "step": 387 }, { "epoch": 0.0, "learning_rate": 4.9999989765147994e-05, "loss": 0.8531, "step": 388 }, { "epoch": 0.0, "learning_rate": 4.999998969394924e-05, "loss": 0.9618, "step": 389 }, { "epoch": 0.0, "learning_rate": 4.9999989622503695e-05, "loss": 1.2255, "step": 390 }, { "epoch": 0.0, "learning_rate": 4.999998955081137e-05, "loss": 1.3624, "step": 391 }, { "epoch": 0.0, "learning_rate": 4.999998947887225e-05, "loss": 1.1627, "step": 392 }, { "epoch": 0.0, "learning_rate": 4.999998940668634e-05, "loss": 1.2721, "step": 393 }, { "epoch": 0.0, "learning_rate": 4.999998933425365e-05, "loss": 1.1621, "step": 394 }, { "epoch": 0.0, "learning_rate": 4.9999989261574154e-05, "loss": 1.2259, "step": 395 }, { "epoch": 0.0, "learning_rate": 4.999998918864788e-05, "loss": 1.1092, "step": 396 }, { "epoch": 0.0, "learning_rate": 4.999998911547481e-05, "loss": 1.205, "step": 397 }, { "epoch": 0.0, "learning_rate": 4.9999989042054954e-05, "loss": 1.164, "step": 398 }, { "epoch": 0.0, "learning_rate": 4.999998896838831e-05, "loss": 1.083, "step": 399 }, { "epoch": 0.0, "learning_rate": 4.999998889447488e-05, "loss": 1.0954, "step": 400 }, { "epoch": 0.0, "learning_rate": 4.999998882031466e-05, "loss": 0.9265, "step": 401 }, { "epoch": 0.0, "learning_rate": 4.9999988745907645e-05, "loss": 0.8312, "step": 402 }, { "epoch": 0.0, "learning_rate": 4.999998867125384e-05, "loss": 0.8345, "step": 403 }, { "epoch": 0.0, "learning_rate": 4.999998859635325e-05, "loss": 0.7232, "step": 404 }, { "epoch": 0.0, "learning_rate": 4.9999988521205864e-05, "loss": 0.6289, "step": 405 }, { "epoch": 0.0, "learning_rate": 4.99999884458117e-05, "loss": 0.9681, "step": 406 }, { "epoch": 0.0, "learning_rate": 4.9999988370170735e-05, "loss": 1.2402, "step": 407 }, { "epoch": 0.0, "learning_rate": 4.999998829428299e-05, "loss": 1.1044, "step": 408 }, { "epoch": 0.0, "learning_rate": 4.999998821814845e-05, "loss": 0.9407, "step": 409 }, { "epoch": 0.0, "learning_rate": 4.999998814176713e-05, "loss": 1.0668, "step": 410 }, { "epoch": 0.0, "learning_rate": 4.999998806513901e-05, "loss": 1.1896, "step": 411 }, { "epoch": 0.0, "learning_rate": 4.9999987988264104e-05, "loss": 1.1691, "step": 412 }, { "epoch": 0.0, "learning_rate": 4.999998791114241e-05, "loss": 1.2159, "step": 413 }, { "epoch": 0.0, "learning_rate": 4.999998783377392e-05, "loss": 1.3894, "step": 414 }, { "epoch": 0.0, "learning_rate": 4.999998775615865e-05, "loss": 1.329, "step": 415 }, { "epoch": 0.0, "learning_rate": 4.9999987678296585e-05, "loss": 1.0454, "step": 416 }, { "epoch": 0.0, "learning_rate": 4.999998760018773e-05, "loss": 0.9511, "step": 417 }, { "epoch": 0.0, "learning_rate": 4.999998752183209e-05, "loss": 0.4759, "step": 418 }, { "epoch": 0.0, "learning_rate": 4.9999987443229654e-05, "loss": 0.8928, "step": 419 }, { "epoch": 0.0, "learning_rate": 4.9999987364380435e-05, "loss": 0.9391, "step": 420 }, { "epoch": 0.0, "learning_rate": 4.9999987285284424e-05, "loss": 1.1387, "step": 421 }, { "epoch": 0.0, "learning_rate": 4.999998720594162e-05, "loss": 1.1293, "step": 422 }, { "epoch": 0.0, "learning_rate": 4.9999987126352034e-05, "loss": 1.2704, "step": 423 }, { "epoch": 0.0, "learning_rate": 4.999998704651566e-05, "loss": 1.137, "step": 424 }, { "epoch": 0.0, "learning_rate": 4.999998696643249e-05, "loss": 0.9722, "step": 425 }, { "epoch": 0.0, "learning_rate": 4.9999986886102536e-05, "loss": 1.2422, "step": 426 }, { "epoch": 0.0, "learning_rate": 4.999998680552579e-05, "loss": 1.0776, "step": 427 }, { "epoch": 0.0, "learning_rate": 4.999998672470225e-05, "loss": 0.8233, "step": 428 }, { "epoch": 0.0, "learning_rate": 4.999998664363192e-05, "loss": 1.2957, "step": 429 }, { "epoch": 0.0, "learning_rate": 4.999998656231481e-05, "loss": 1.0972, "step": 430 }, { "epoch": 0.0, "learning_rate": 4.999998648075091e-05, "loss": 1.2265, "step": 431 }, { "epoch": 0.0, "learning_rate": 4.999998639894021e-05, "loss": 1.3584, "step": 432 }, { "epoch": 0.0, "learning_rate": 4.999998631688273e-05, "loss": 1.9795, "step": 433 }, { "epoch": 0.0, "learning_rate": 4.999998623457846e-05, "loss": 1.4203, "step": 434 }, { "epoch": 0.0, "learning_rate": 4.99999861520274e-05, "loss": 1.595, "step": 435 }, { "epoch": 0.0, "learning_rate": 4.999998606922955e-05, "loss": 0.6731, "step": 436 }, { "epoch": 0.0, "learning_rate": 4.999998598618491e-05, "loss": 1.2655, "step": 437 }, { "epoch": 0.0, "learning_rate": 4.9999985902893484e-05, "loss": 1.1741, "step": 438 }, { "epoch": 0.0, "learning_rate": 4.9999985819355266e-05, "loss": 1.0423, "step": 439 }, { "epoch": 0.0, "learning_rate": 4.999998573557026e-05, "loss": 1.1478, "step": 440 }, { "epoch": 0.0, "learning_rate": 4.999998565153846e-05, "loss": 1.044, "step": 441 }, { "epoch": 0.0, "learning_rate": 4.999998556725987e-05, "loss": 0.7539, "step": 442 }, { "epoch": 0.0, "learning_rate": 4.999998548273449e-05, "loss": 0.685, "step": 443 }, { "epoch": 0.0, "learning_rate": 4.999998539796233e-05, "loss": 5.6911, "step": 444 }, { "epoch": 0.0, "learning_rate": 4.999998531294338e-05, "loss": 5.9401, "step": 445 }, { "epoch": 0.0, "learning_rate": 4.999998522767764e-05, "loss": 6.034, "step": 446 }, { "epoch": 0.0, "learning_rate": 4.99999851421651e-05, "loss": 5.9444, "step": 447 }, { "epoch": 0.0, "learning_rate": 4.999998505640578e-05, "loss": 5.9153, "step": 448 }, { "epoch": 0.0, "learning_rate": 4.999998497039967e-05, "loss": 5.8, "step": 449 }, { "epoch": 0.0, "learning_rate": 4.9999984884146765e-05, "loss": 5.7652, "step": 450 }, { "epoch": 0.0, "learning_rate": 4.9999984797647076e-05, "loss": 5.5309, "step": 451 }, { "epoch": 0.0, "learning_rate": 4.9999984710900596e-05, "loss": 5.6267, "step": 452 }, { "epoch": 0.0, "learning_rate": 4.9999984623907324e-05, "loss": 1.8665, "step": 453 }, { "epoch": 0.0, "learning_rate": 4.9999984536667274e-05, "loss": 1.5152, "step": 454 }, { "epoch": 0.0, "learning_rate": 4.9999984449180425e-05, "loss": 0.5595, "step": 455 }, { "epoch": 0.0, "learning_rate": 4.9999984361446786e-05, "loss": 1.2043, "step": 456 }, { "epoch": 0.0, "learning_rate": 4.999998427346636e-05, "loss": 1.1112, "step": 457 }, { "epoch": 0.0, "learning_rate": 4.9999984185239145e-05, "loss": 1.4293, "step": 458 }, { "epoch": 0.0, "learning_rate": 4.999998409676514e-05, "loss": 1.0783, "step": 459 }, { "epoch": 0.0, "learning_rate": 4.9999984008044345e-05, "loss": 1.049, "step": 460 }, { "epoch": 0.0, "learning_rate": 4.999998391907677e-05, "loss": 1.029, "step": 461 }, { "epoch": 0.0, "learning_rate": 4.999998382986239e-05, "loss": 0.8213, "step": 462 }, { "epoch": 0.0, "learning_rate": 4.999998374040123e-05, "loss": 1.2395, "step": 463 }, { "epoch": 0.0, "learning_rate": 4.9999983650693274e-05, "loss": 1.3614, "step": 464 }, { "epoch": 0.0, "learning_rate": 4.999998356073854e-05, "loss": 1.1796, "step": 465 }, { "epoch": 0.0, "learning_rate": 4.9999983470537e-05, "loss": 1.2114, "step": 466 }, { "epoch": 0.0, "learning_rate": 4.999998338008869e-05, "loss": 0.973, "step": 467 }, { "epoch": 0.0, "learning_rate": 4.999998328939358e-05, "loss": 0.9211, "step": 468 }, { "epoch": 0.0, "learning_rate": 4.9999983198451684e-05, "loss": 1.0192, "step": 469 }, { "epoch": 0.0, "learning_rate": 4.999998310726299e-05, "loss": 1.1802, "step": 470 }, { "epoch": 0.0, "learning_rate": 4.999998301582752e-05, "loss": 0.9868, "step": 471 }, { "epoch": 0.0, "learning_rate": 4.999998292414525e-05, "loss": 1.195, "step": 472 }, { "epoch": 0.0, "learning_rate": 4.99999828322162e-05, "loss": 1.1628, "step": 473 }, { "epoch": 0.0, "learning_rate": 4.999998274004035e-05, "loss": 0.9806, "step": 474 }, { "epoch": 0.0, "learning_rate": 4.999998264761772e-05, "loss": 1.0956, "step": 475 }, { "epoch": 0.0, "learning_rate": 4.99999825549483e-05, "loss": 1.3023, "step": 476 }, { "epoch": 0.0, "learning_rate": 4.9999982462032085e-05, "loss": 1.203, "step": 477 }, { "epoch": 0.0, "learning_rate": 4.999998236886908e-05, "loss": 1.5488, "step": 478 }, { "epoch": 0.0, "learning_rate": 4.999998227545929e-05, "loss": 1.3148, "step": 479 }, { "epoch": 0.0, "learning_rate": 4.999998218180271e-05, "loss": 1.2602, "step": 480 }, { "epoch": 0.0, "learning_rate": 4.9999982087899346e-05, "loss": 1.1743, "step": 481 }, { "epoch": 0.0, "learning_rate": 4.999998199374918e-05, "loss": 0.9299, "step": 482 }, { "epoch": 0.0, "learning_rate": 4.999998189935224e-05, "loss": 0.7564, "step": 483 }, { "epoch": 0.0, "learning_rate": 4.9999981804708504e-05, "loss": 1.0182, "step": 484 }, { "epoch": 0.0, "learning_rate": 4.999998170981798e-05, "loss": 1.1082, "step": 485 }, { "epoch": 0.0, "learning_rate": 4.999998161468066e-05, "loss": 1.3019, "step": 486 }, { "epoch": 0.0, "learning_rate": 4.999998151929656e-05, "loss": 0.8937, "step": 487 }, { "epoch": 0.0, "learning_rate": 4.999998142366566e-05, "loss": 0.8715, "step": 488 }, { "epoch": 0.0, "learning_rate": 4.999998132778798e-05, "loss": 0.6235, "step": 489 }, { "epoch": 0.0, "learning_rate": 4.999998123166351e-05, "loss": 0.5427, "step": 490 }, { "epoch": 0.0, "learning_rate": 4.999998113529224e-05, "loss": 0.9652, "step": 491 }, { "epoch": 0.0, "learning_rate": 4.999998103867419e-05, "loss": 1.0241, "step": 492 }, { "epoch": 0.0, "learning_rate": 4.999998094180935e-05, "loss": 1.2874, "step": 493 }, { "epoch": 0.0, "learning_rate": 4.999998084469772e-05, "loss": 1.1236, "step": 494 }, { "epoch": 0.0, "learning_rate": 4.9999980747339305e-05, "loss": 1.2312, "step": 495 }, { "epoch": 0.0, "learning_rate": 4.999998064973409e-05, "loss": 1.1679, "step": 496 }, { "epoch": 0.0, "learning_rate": 4.9999980551882095e-05, "loss": 0.7927, "step": 497 }, { "epoch": 0.0, "learning_rate": 4.9999980453783306e-05, "loss": 0.7081, "step": 498 }, { "epoch": 0.0, "learning_rate": 4.999998035543773e-05, "loss": 1.2299, "step": 499 }, { "epoch": 0.0, "learning_rate": 4.999998025684536e-05, "loss": 1.1899, "step": 500 }, { "epoch": 0.0, "eval_loss": 1.1330914497375488, "eval_runtime": 85.2447, "eval_samples_per_second": 16.247, "eval_steps_per_second": 4.071, "step": 500 }, { "epoch": 0.0, "learning_rate": 4.999998015800621e-05, "loss": 0.9488, "step": 501 }, { "epoch": 0.0, "learning_rate": 4.999998005892026e-05, "loss": 0.9233, "step": 502 }, { "epoch": 0.0, "learning_rate": 4.999997995958753e-05, "loss": 1.0061, "step": 503 }, { "epoch": 0.0, "learning_rate": 4.9999979860008006e-05, "loss": 0.8307, "step": 504 }, { "epoch": 0.0, "learning_rate": 4.99999797601817e-05, "loss": 0.9544, "step": 505 }, { "epoch": 0.0, "learning_rate": 4.9999979660108597e-05, "loss": 0.5917, "step": 506 }, { "epoch": 0.0, "learning_rate": 4.9999979559788704e-05, "loss": 0.5971, "step": 507 }, { "epoch": 0.0, "learning_rate": 4.999997945922202e-05, "loss": 1.3349, "step": 508 }, { "epoch": 0.0, "learning_rate": 4.999997935840855e-05, "loss": 1.0993, "step": 509 }, { "epoch": 0.0, "learning_rate": 4.99999792573483e-05, "loss": 1.0986, "step": 510 }, { "epoch": 0.0, "learning_rate": 4.9999979156041255e-05, "loss": 1.0709, "step": 511 }, { "epoch": 0.0, "learning_rate": 4.999997905448741e-05, "loss": 1.1506, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.999997895268679e-05, "loss": 0.9141, "step": 513 }, { "epoch": 0.0, "learning_rate": 4.999997885063937e-05, "loss": 1.5013, "step": 514 }, { "epoch": 0.0, "learning_rate": 4.999997874834517e-05, "loss": 0.9547, "step": 515 }, { "epoch": 0.0, "learning_rate": 4.999997864580418e-05, "loss": 1.09, "step": 516 }, { "epoch": 0.0, "learning_rate": 4.999997854301639e-05, "loss": 1.1179, "step": 517 }, { "epoch": 0.0, "learning_rate": 4.999997843998182e-05, "loss": 1.0737, "step": 518 }, { "epoch": 0.0, "learning_rate": 4.999997833670046e-05, "loss": 0.7466, "step": 519 }, { "epoch": 0.0, "learning_rate": 4.999997823317231e-05, "loss": 0.9024, "step": 520 }, { "epoch": 0.0, "learning_rate": 4.999997812939737e-05, "loss": 0.9933, "step": 521 }, { "epoch": 0.0, "learning_rate": 4.999997802537564e-05, "loss": 0.8115, "step": 522 }, { "epoch": 0.0, "learning_rate": 4.999997792110712e-05, "loss": 0.8077, "step": 523 }, { "epoch": 0.0, "learning_rate": 4.9999977816591816e-05, "loss": 0.6865, "step": 524 }, { "epoch": 0.0, "learning_rate": 4.999997771182972e-05, "loss": 0.5826, "step": 525 }, { "epoch": 0.0, "learning_rate": 4.9999977606820834e-05, "loss": 1.1181, "step": 526 }, { "epoch": 0.0, "learning_rate": 4.999997750156515e-05, "loss": 1.4692, "step": 527 }, { "epoch": 0.0, "learning_rate": 4.999997739606269e-05, "loss": 1.0833, "step": 528 }, { "epoch": 0.0, "learning_rate": 4.999997729031344e-05, "loss": 1.1319, "step": 529 }, { "epoch": 0.0, "learning_rate": 4.999997718431739e-05, "loss": 1.252, "step": 530 }, { "epoch": 0.0, "learning_rate": 4.999997707807456e-05, "loss": 1.0724, "step": 531 }, { "epoch": 0.0, "learning_rate": 4.999997697158494e-05, "loss": 1.2738, "step": 532 }, { "epoch": 0.0, "learning_rate": 4.999997686484853e-05, "loss": 0.9913, "step": 533 }, { "epoch": 0.0, "learning_rate": 4.999997675786533e-05, "loss": 1.1738, "step": 534 }, { "epoch": 0.0, "learning_rate": 4.999997665063534e-05, "loss": 1.3109, "step": 535 }, { "epoch": 0.0, "learning_rate": 4.999997654315855e-05, "loss": 1.3323, "step": 536 }, { "epoch": 0.0, "learning_rate": 4.999997643543499e-05, "loss": 1.1117, "step": 537 }, { "epoch": 0.0, "learning_rate": 4.9999976327464635e-05, "loss": 1.0758, "step": 538 }, { "epoch": 0.0, "learning_rate": 4.999997621924749e-05, "loss": 1.029, "step": 539 }, { "epoch": 0.0, "learning_rate": 4.999997611078355e-05, "loss": 1.2414, "step": 540 }, { "epoch": 0.0, "learning_rate": 4.999997600207282e-05, "loss": 1.1829, "step": 541 }, { "epoch": 0.0, "learning_rate": 4.9999975893115315e-05, "loss": 1.037, "step": 542 }, { "epoch": 0.0, "learning_rate": 4.9999975783911016e-05, "loss": 0.9917, "step": 543 }, { "epoch": 0.0, "learning_rate": 4.999997567445992e-05, "loss": 0.9406, "step": 544 }, { "epoch": 0.0, "learning_rate": 4.999997556476204e-05, "loss": 1.1532, "step": 545 }, { "epoch": 0.0, "learning_rate": 4.999997545481736e-05, "loss": 0.9999, "step": 546 }, { "epoch": 0.0, "learning_rate": 4.9999975344625906e-05, "loss": 1.0536, "step": 547 }, { "epoch": 0.0, "learning_rate": 4.9999975234187656e-05, "loss": 1.1719, "step": 548 }, { "epoch": 0.0, "learning_rate": 4.9999975123502615e-05, "loss": 1.0755, "step": 549 }, { "epoch": 0.0, "learning_rate": 4.999997501257079e-05, "loss": 0.1063, "step": 550 }, { "epoch": 0.0, "learning_rate": 4.999997490139217e-05, "loss": 0.0709, "step": 551 }, { "epoch": 0.0, "learning_rate": 4.999997478996676e-05, "loss": 0.2467, "step": 552 }, { "epoch": 0.0, "learning_rate": 4.999997467829457e-05, "loss": 0.2245, "step": 553 }, { "epoch": 0.0, "learning_rate": 4.9999974566375585e-05, "loss": 0.1978, "step": 554 }, { "epoch": 0.0, "learning_rate": 4.999997445420981e-05, "loss": 0.1696, "step": 555 }, { "epoch": 0.0, "learning_rate": 4.9999974341797254e-05, "loss": 0.1546, "step": 556 }, { "epoch": 0.0, "learning_rate": 4.9999974229137894e-05, "loss": 0.1919, "step": 557 }, { "epoch": 0.0, "learning_rate": 4.999997411623175e-05, "loss": 0.2029, "step": 558 }, { "epoch": 0.0, "learning_rate": 4.999997400307882e-05, "loss": 0.1537, "step": 559 }, { "epoch": 0.0, "learning_rate": 4.99999738896791e-05, "loss": 0.1449, "step": 560 }, { "epoch": 0.0, "learning_rate": 4.9999973776032594e-05, "loss": 0.7279, "step": 561 }, { "epoch": 0.0, "learning_rate": 4.99999736621393e-05, "loss": 0.6953, "step": 562 }, { "epoch": 0.0, "learning_rate": 4.999997354799921e-05, "loss": 0.6295, "step": 563 }, { "epoch": 0.0, "learning_rate": 4.999997343361233e-05, "loss": 0.5171, "step": 564 }, { "epoch": 0.0, "learning_rate": 4.9999973318978664e-05, "loss": 0.486, "step": 565 }, { "epoch": 0.0, "learning_rate": 4.999997320409821e-05, "loss": 1.2449, "step": 566 }, { "epoch": 0.0, "learning_rate": 4.999997308897097e-05, "loss": 1.4136, "step": 567 }, { "epoch": 0.0, "learning_rate": 4.9999972973596935e-05, "loss": 1.2475, "step": 568 }, { "epoch": 0.0, "learning_rate": 4.999997285797611e-05, "loss": 1.3221, "step": 569 }, { "epoch": 0.0, "learning_rate": 4.9999972742108495e-05, "loss": 1.1538, "step": 570 }, { "epoch": 0.0, "learning_rate": 4.9999972625994095e-05, "loss": 0.8241, "step": 571 }, { "epoch": 0.0, "learning_rate": 4.999997250963291e-05, "loss": 1.1044, "step": 572 }, { "epoch": 0.0, "learning_rate": 4.9999972393024926e-05, "loss": 1.2559, "step": 573 }, { "epoch": 0.0, "learning_rate": 4.999997227617016e-05, "loss": 1.0469, "step": 574 }, { "epoch": 0.0, "learning_rate": 4.99999721590686e-05, "loss": 1.2542, "step": 575 }, { "epoch": 0.0, "learning_rate": 4.999997204172025e-05, "loss": 1.1862, "step": 576 }, { "epoch": 0.0, "learning_rate": 4.999997192412512e-05, "loss": 1.3234, "step": 577 }, { "epoch": 0.0, "learning_rate": 4.9999971806283185e-05, "loss": 0.8016, "step": 578 }, { "epoch": 0.0, "learning_rate": 4.999997168819447e-05, "loss": 1.3645, "step": 579 }, { "epoch": 0.0, "learning_rate": 4.999997156985897e-05, "loss": 1.1961, "step": 580 }, { "epoch": 0.0, "learning_rate": 4.999997145127668e-05, "loss": 1.2315, "step": 581 }, { "epoch": 0.0, "learning_rate": 4.9999971332447594e-05, "loss": 1.0403, "step": 582 }, { "epoch": 0.0, "learning_rate": 4.999997121337172e-05, "loss": 1.1351, "step": 583 }, { "epoch": 0.0, "learning_rate": 4.9999971094049066e-05, "loss": 1.1116, "step": 584 }, { "epoch": 0.0, "learning_rate": 4.999997097447961e-05, "loss": 1.2778, "step": 585 }, { "epoch": 0.0, "learning_rate": 4.999997085466337e-05, "loss": 1.1626, "step": 586 }, { "epoch": 0.0, "learning_rate": 4.999997073460034e-05, "loss": 1.0976, "step": 587 }, { "epoch": 0.0, "learning_rate": 4.9999970614290527e-05, "loss": 1.3518, "step": 588 }, { "epoch": 0.0, "learning_rate": 4.999997049373392e-05, "loss": 1.1483, "step": 589 }, { "epoch": 0.0, "learning_rate": 4.999997037293053e-05, "loss": 1.1493, "step": 590 }, { "epoch": 0.0, "learning_rate": 4.999997025188034e-05, "loss": 1.1073, "step": 591 }, { "epoch": 0.0, "learning_rate": 4.999997013058337e-05, "loss": 1.3351, "step": 592 }, { "epoch": 0.0, "learning_rate": 4.9999970009039604e-05, "loss": 1.0137, "step": 593 }, { "epoch": 0.0, "learning_rate": 4.999996988724905e-05, "loss": 1.159, "step": 594 }, { "epoch": 0.0, "learning_rate": 4.999996976521171e-05, "loss": 1.0414, "step": 595 }, { "epoch": 0.0, "learning_rate": 4.999996964292758e-05, "loss": 1.2767, "step": 596 }, { "epoch": 0.0, "learning_rate": 4.9999969520396664e-05, "loss": 1.0204, "step": 597 }, { "epoch": 0.0, "learning_rate": 4.9999969397618954e-05, "loss": 0.8231, "step": 598 }, { "epoch": 0.0, "learning_rate": 4.999996927459445e-05, "loss": 1.2339, "step": 599 }, { "epoch": 0.0, "learning_rate": 4.9999969151323165e-05, "loss": 1.1286, "step": 600 }, { "epoch": 0.0, "learning_rate": 4.999996902780509e-05, "loss": 1.3251, "step": 601 }, { "epoch": 0.0, "learning_rate": 4.9999968904040224e-05, "loss": 1.1275, "step": 602 }, { "epoch": 0.0, "learning_rate": 4.999996878002857e-05, "loss": 0.9398, "step": 603 }, { "epoch": 0.0, "learning_rate": 4.9999968655770124e-05, "loss": 1.1224, "step": 604 }, { "epoch": 0.0, "learning_rate": 4.999996853126489e-05, "loss": 1.0039, "step": 605 }, { "epoch": 0.0, "learning_rate": 4.9999968406512865e-05, "loss": 1.0953, "step": 606 }, { "epoch": 0.0, "learning_rate": 4.999996828151405e-05, "loss": 1.0682, "step": 607 }, { "epoch": 0.0, "learning_rate": 4.999996815626846e-05, "loss": 0.981, "step": 608 }, { "epoch": 0.0, "learning_rate": 4.999996803077607e-05, "loss": 1.135, "step": 609 }, { "epoch": 0.0, "learning_rate": 4.999996790503689e-05, "loss": 2.0639, "step": 610 }, { "epoch": 0.0, "learning_rate": 4.9999967779050915e-05, "loss": 1.208, "step": 611 }, { "epoch": 0.0, "learning_rate": 4.999996765281816e-05, "loss": 0.9878, "step": 612 }, { "epoch": 0.0, "learning_rate": 4.9999967526338615e-05, "loss": 1.0357, "step": 613 }, { "epoch": 0.0, "learning_rate": 4.999996739961228e-05, "loss": 1.0447, "step": 614 }, { "epoch": 0.0, "learning_rate": 4.9999967272639156e-05, "loss": 1.0037, "step": 615 }, { "epoch": 0.0, "learning_rate": 4.999996714541924e-05, "loss": 1.256, "step": 616 }, { "epoch": 0.0, "learning_rate": 4.9999967017952544e-05, "loss": 1.0181, "step": 617 }, { "epoch": 0.0, "learning_rate": 4.999996689023905e-05, "loss": 0.8774, "step": 618 }, { "epoch": 0.0, "learning_rate": 4.9999966762278766e-05, "loss": 1.0415, "step": 619 }, { "epoch": 0.0, "learning_rate": 4.9999966634071696e-05, "loss": 1.1236, "step": 620 }, { "epoch": 0.0, "learning_rate": 4.9999966505617835e-05, "loss": 1.1345, "step": 621 }, { "epoch": 0.0, "learning_rate": 4.999996637691718e-05, "loss": 1.103, "step": 622 }, { "epoch": 0.0, "learning_rate": 4.9999966247969745e-05, "loss": 0.8407, "step": 623 }, { "epoch": 0.0, "learning_rate": 4.999996611877552e-05, "loss": 1.2822, "step": 624 }, { "epoch": 0.0, "learning_rate": 4.999996598933451e-05, "loss": 1.3973, "step": 625 }, { "epoch": 0.0, "learning_rate": 4.9999965859646704e-05, "loss": 1.1597, "step": 626 }, { "epoch": 0.0, "learning_rate": 4.999996572971211e-05, "loss": 1.9025, "step": 627 }, { "epoch": 0.0, "learning_rate": 4.9999965599530726e-05, "loss": 1.2152, "step": 628 }, { "epoch": 0.0, "learning_rate": 4.999996546910255e-05, "loss": 1.157, "step": 629 }, { "epoch": 0.0, "learning_rate": 4.999996533842759e-05, "loss": 1.3811, "step": 630 }, { "epoch": 0.0, "learning_rate": 4.999996520750584e-05, "loss": 1.2883, "step": 631 }, { "epoch": 0.0, "learning_rate": 4.99999650763373e-05, "loss": 1.4924, "step": 632 }, { "epoch": 0.0, "learning_rate": 4.9999964944921974e-05, "loss": 1.0566, "step": 633 }, { "epoch": 0.0, "learning_rate": 4.999996481325985e-05, "loss": 1.3231, "step": 634 }, { "epoch": 0.0, "learning_rate": 4.999996468135094e-05, "loss": 1.3814, "step": 635 }, { "epoch": 0.0, "learning_rate": 4.999996454919524e-05, "loss": 1.0348, "step": 636 }, { "epoch": 0.0, "learning_rate": 4.9999964416792764e-05, "loss": 1.1868, "step": 637 }, { "epoch": 0.0, "learning_rate": 4.999996428414349e-05, "loss": 1.2078, "step": 638 }, { "epoch": 0.0, "learning_rate": 4.999996415124743e-05, "loss": 1.1857, "step": 639 }, { "epoch": 0.0, "learning_rate": 4.9999964018104574e-05, "loss": 1.083, "step": 640 }, { "epoch": 0.0, "learning_rate": 4.999996388471493e-05, "loss": 1.2147, "step": 641 }, { "epoch": 0.0, "learning_rate": 4.99999637510785e-05, "loss": 3.2249, "step": 642 }, { "epoch": 0.0, "learning_rate": 4.999996361719528e-05, "loss": 3.0262, "step": 643 }, { "epoch": 0.0, "learning_rate": 4.999996348306527e-05, "loss": 1.0029, "step": 644 }, { "epoch": 0.0, "learning_rate": 4.999996334868847e-05, "loss": 1.5653, "step": 645 }, { "epoch": 0.0, "learning_rate": 4.9999963214064885e-05, "loss": 1.0408, "step": 646 }, { "epoch": 0.0, "learning_rate": 4.9999963079194505e-05, "loss": 1.037, "step": 647 }, { "epoch": 0.0, "learning_rate": 4.999996294407734e-05, "loss": 1.2011, "step": 648 }, { "epoch": 0.0, "learning_rate": 4.9999962808713385e-05, "loss": 1.1739, "step": 649 }, { "epoch": 0.0, "learning_rate": 4.999996267310264e-05, "loss": 1.3935, "step": 650 }, { "epoch": 0.0, "learning_rate": 4.999996253724511e-05, "loss": 1.3371, "step": 651 }, { "epoch": 0.0, "learning_rate": 4.999996240114079e-05, "loss": 1.3247, "step": 652 }, { "epoch": 0.0, "learning_rate": 4.999996226478968e-05, "loss": 1.2976, "step": 653 }, { "epoch": 0.0, "learning_rate": 4.9999962128191774e-05, "loss": 1.1112, "step": 654 }, { "epoch": 0.0, "learning_rate": 4.999996199134709e-05, "loss": 1.1392, "step": 655 }, { "epoch": 0.0, "learning_rate": 4.9999961854255606e-05, "loss": 0.9464, "step": 656 }, { "epoch": 0.0, "learning_rate": 4.9999961716917345e-05, "loss": 1.1398, "step": 657 }, { "epoch": 0.0, "learning_rate": 4.999996157933228e-05, "loss": 1.2702, "step": 658 }, { "epoch": 0.0, "learning_rate": 4.9999961441500436e-05, "loss": 1.046, "step": 659 }, { "epoch": 0.0, "learning_rate": 4.999996130342181e-05, "loss": 1.203, "step": 660 }, { "epoch": 0.0, "learning_rate": 4.999996116509637e-05, "loss": 1.1421, "step": 661 }, { "epoch": 0.0, "learning_rate": 4.999996102652417e-05, "loss": 0.8796, "step": 662 }, { "epoch": 0.0, "learning_rate": 4.9999960887705165e-05, "loss": 1.0829, "step": 663 }, { "epoch": 0.0, "learning_rate": 4.999996074863938e-05, "loss": 1.2028, "step": 664 }, { "epoch": 0.0, "learning_rate": 4.999996060932679e-05, "loss": 1.2499, "step": 665 }, { "epoch": 0.0, "learning_rate": 4.9999960469767427e-05, "loss": 1.5727, "step": 666 }, { "epoch": 0.0, "learning_rate": 4.9999960329961264e-05, "loss": 1.4672, "step": 667 }, { "epoch": 0.0, "learning_rate": 4.999996018990832e-05, "loss": 1.3084, "step": 668 }, { "epoch": 0.0, "learning_rate": 4.999996004960858e-05, "loss": 1.1889, "step": 669 }, { "epoch": 0.0, "learning_rate": 4.9999959909062054e-05, "loss": 1.1393, "step": 670 }, { "epoch": 0.0, "learning_rate": 4.999995976826874e-05, "loss": 1.2297, "step": 671 }, { "epoch": 0.0, "learning_rate": 4.999995962722864e-05, "loss": 1.2911, "step": 672 }, { "epoch": 0.0, "learning_rate": 4.999995948594174e-05, "loss": 1.0032, "step": 673 }, { "epoch": 0.0, "learning_rate": 4.9999959344408066e-05, "loss": 0.9387, "step": 674 }, { "epoch": 0.0, "learning_rate": 4.999995920262759e-05, "loss": 0.9681, "step": 675 }, { "epoch": 0.0, "learning_rate": 4.999995906060033e-05, "loss": 0.9953, "step": 676 }, { "epoch": 0.0, "learning_rate": 4.999995891832628e-05, "loss": 1.0853, "step": 677 }, { "epoch": 0.0, "learning_rate": 4.9999958775805446e-05, "loss": 0.9147, "step": 678 }, { "epoch": 0.0, "learning_rate": 4.999995863303781e-05, "loss": 1.2149, "step": 679 }, { "epoch": 0.0, "learning_rate": 4.99999584900234e-05, "loss": 0.9156, "step": 680 }, { "epoch": 0.0, "learning_rate": 4.99999583467622e-05, "loss": 1.2145, "step": 681 }, { "epoch": 0.0, "learning_rate": 4.99999582032542e-05, "loss": 1.2697, "step": 682 }, { "epoch": 0.0, "learning_rate": 4.999995805949942e-05, "loss": 1.5116, "step": 683 }, { "epoch": 0.0, "learning_rate": 4.999995791549784e-05, "loss": 1.0923, "step": 684 }, { "epoch": 0.0, "learning_rate": 4.9999957771249483e-05, "loss": 1.0256, "step": 685 }, { "epoch": 0.0, "learning_rate": 4.9999957626754336e-05, "loss": 1.2579, "step": 686 }, { "epoch": 0.0, "learning_rate": 4.999995748201239e-05, "loss": 1.8896, "step": 687 }, { "epoch": 0.0, "learning_rate": 4.999995733702366e-05, "loss": 1.9751, "step": 688 }, { "epoch": 0.0, "learning_rate": 4.999995719178815e-05, "loss": 2.1633, "step": 689 }, { "epoch": 0.0, "learning_rate": 4.999995704630584e-05, "loss": 1.6439, "step": 690 }, { "epoch": 0.0, "learning_rate": 4.999995690057674e-05, "loss": 0.7501, "step": 691 }, { "epoch": 0.0, "learning_rate": 4.9999956754600865e-05, "loss": 0.9558, "step": 692 }, { "epoch": 0.0, "learning_rate": 4.9999956608378184e-05, "loss": 1.0196, "step": 693 }, { "epoch": 0.0, "learning_rate": 4.9999956461908725e-05, "loss": 1.3408, "step": 694 }, { "epoch": 0.0, "learning_rate": 4.9999956315192475e-05, "loss": 1.1365, "step": 695 }, { "epoch": 0.0, "learning_rate": 4.999995616822943e-05, "loss": 1.2901, "step": 696 }, { "epoch": 0.0, "learning_rate": 4.99999560210196e-05, "loss": 1.348, "step": 697 }, { "epoch": 0.0, "learning_rate": 4.999995587356298e-05, "loss": 1.3301, "step": 698 }, { "epoch": 0.0, "learning_rate": 4.999995572585957e-05, "loss": 1.2014, "step": 699 }, { "epoch": 0.0, "learning_rate": 4.9999955577909376e-05, "loss": 1.1904, "step": 700 }, { "epoch": 0.0, "learning_rate": 4.999995542971238e-05, "loss": 1.4346, "step": 701 }, { "epoch": 0.0, "learning_rate": 4.999995528126862e-05, "loss": 1.1291, "step": 702 }, { "epoch": 0.0, "learning_rate": 4.999995513257805e-05, "loss": 1.0225, "step": 703 }, { "epoch": 0.0, "learning_rate": 4.9999954983640696e-05, "loss": 1.902, "step": 704 }, { "epoch": 0.0, "learning_rate": 4.999995483445655e-05, "loss": 1.111, "step": 705 }, { "epoch": 0.0, "learning_rate": 4.999995468502562e-05, "loss": 1.0966, "step": 706 }, { "epoch": 0.0, "learning_rate": 4.9999954535347906e-05, "loss": 0.9121, "step": 707 }, { "epoch": 0.0, "learning_rate": 4.99999543854234e-05, "loss": 0.9493, "step": 708 }, { "epoch": 0.0, "learning_rate": 4.99999542352521e-05, "loss": 0.9592, "step": 709 }, { "epoch": 0.0, "learning_rate": 4.999995408483401e-05, "loss": 0.8201, "step": 710 }, { "epoch": 0.0, "learning_rate": 4.999995393416913e-05, "loss": 1.1763, "step": 711 }, { "epoch": 0.0, "learning_rate": 4.999995378325747e-05, "loss": 1.1213, "step": 712 }, { "epoch": 0.0, "learning_rate": 4.9999953632099015e-05, "loss": 1.0797, "step": 713 }, { "epoch": 0.0, "learning_rate": 4.999995348069377e-05, "loss": 0.9418, "step": 714 }, { "epoch": 0.0, "learning_rate": 4.999995332904174e-05, "loss": 1.2825, "step": 715 }, { "epoch": 0.0, "learning_rate": 4.9999953177142915e-05, "loss": 1.2253, "step": 716 }, { "epoch": 0.0, "learning_rate": 4.9999953024997305e-05, "loss": 1.0951, "step": 717 }, { "epoch": 0.0, "learning_rate": 4.9999952872604903e-05, "loss": 1.2137, "step": 718 }, { "epoch": 0.0, "learning_rate": 4.999995271996572e-05, "loss": 1.1174, "step": 719 }, { "epoch": 0.0, "learning_rate": 4.999995256707975e-05, "loss": 1.2376, "step": 720 }, { "epoch": 0.0, "learning_rate": 4.999995241394698e-05, "loss": 1.2504, "step": 721 }, { "epoch": 0.0, "learning_rate": 4.999995226056742e-05, "loss": 1.1348, "step": 722 }, { "epoch": 0.0, "learning_rate": 4.999995210694107e-05, "loss": 1.1022, "step": 723 }, { "epoch": 0.0, "learning_rate": 4.999995195306794e-05, "loss": 1.1405, "step": 724 }, { "epoch": 0.0, "learning_rate": 4.999995179894802e-05, "loss": 1.237, "step": 725 }, { "epoch": 0.0, "learning_rate": 4.9999951644581314e-05, "loss": 1.009, "step": 726 }, { "epoch": 0.0, "learning_rate": 4.999995148996781e-05, "loss": 1.4083, "step": 727 }, { "epoch": 0.0, "learning_rate": 4.999995133510752e-05, "loss": 1.8452, "step": 728 }, { "epoch": 0.0, "learning_rate": 4.999995118000044e-05, "loss": 1.0214, "step": 729 }, { "epoch": 0.0, "learning_rate": 4.999995102464657e-05, "loss": 1.0336, "step": 730 }, { "epoch": 0.0, "learning_rate": 4.9999950869045916e-05, "loss": 1.1379, "step": 731 }, { "epoch": 0.0, "learning_rate": 4.9999950713198474e-05, "loss": 1.0476, "step": 732 }, { "epoch": 0.0, "learning_rate": 4.9999950557104234e-05, "loss": 1.3398, "step": 733 }, { "epoch": 0.0, "learning_rate": 4.9999950400763216e-05, "loss": 1.0744, "step": 734 }, { "epoch": 0.0, "learning_rate": 4.9999950244175406e-05, "loss": 1.0007, "step": 735 }, { "epoch": 0.0, "learning_rate": 4.9999950087340805e-05, "loss": 1.3092, "step": 736 }, { "epoch": 0.0, "learning_rate": 4.999994993025941e-05, "loss": 1.1256, "step": 737 }, { "epoch": 0.0, "learning_rate": 4.9999949772931235e-05, "loss": 1.6756, "step": 738 }, { "epoch": 0.0, "learning_rate": 4.9999949615356266e-05, "loss": 1.9275, "step": 739 }, { "epoch": 0.0, "learning_rate": 4.999994945753451e-05, "loss": 0.9195, "step": 740 }, { "epoch": 0.0, "learning_rate": 4.999994929946597e-05, "loss": 1.3474, "step": 741 }, { "epoch": 0.0, "learning_rate": 4.9999949141150624e-05, "loss": 1.4102, "step": 742 }, { "epoch": 0.0, "learning_rate": 4.999994898258851e-05, "loss": 1.3585, "step": 743 }, { "epoch": 0.0, "learning_rate": 4.999994882377959e-05, "loss": 1.2699, "step": 744 }, { "epoch": 0.0, "learning_rate": 4.999994866472389e-05, "loss": 1.0139, "step": 745 }, { "epoch": 0.0, "learning_rate": 4.99999485054214e-05, "loss": 0.8573, "step": 746 }, { "epoch": 0.0, "learning_rate": 4.9999948345872114e-05, "loss": 0.7978, "step": 747 }, { "epoch": 0.0, "learning_rate": 4.999994818607605e-05, "loss": 0.9753, "step": 748 }, { "epoch": 0.0, "learning_rate": 4.999994802603319e-05, "loss": 1.3202, "step": 749 }, { "epoch": 0.0, "learning_rate": 4.999994786574354e-05, "loss": 0.3981, "step": 750 }, { "epoch": 0.0, "learning_rate": 4.99999477052071e-05, "loss": 0.3395, "step": 751 }, { "epoch": 0.0, "learning_rate": 4.9999947544423885e-05, "loss": 0.3216, "step": 752 }, { "epoch": 0.0, "learning_rate": 4.9999947383393876e-05, "loss": 0.215, "step": 753 }, { "epoch": 0.0, "learning_rate": 4.9999947222117074e-05, "loss": 0.2558, "step": 754 }, { "epoch": 0.0, "learning_rate": 4.9999947060593475e-05, "loss": 0.1257, "step": 755 }, { "epoch": 0.0, "learning_rate": 4.99999468988231e-05, "loss": 0.096, "step": 756 }, { "epoch": 0.0, "learning_rate": 4.999994673680592e-05, "loss": 0.0708, "step": 757 }, { "epoch": 0.0, "learning_rate": 4.999994657454197e-05, "loss": 0.0575, "step": 758 }, { "epoch": 0.0, "learning_rate": 4.999994641203122e-05, "loss": 0.0431, "step": 759 }, { "epoch": 0.0, "learning_rate": 4.9999946249273687e-05, "loss": 0.1335, "step": 760 }, { "epoch": 0.0, "learning_rate": 4.9999946086269365e-05, "loss": 0.1145, "step": 761 }, { "epoch": 0.0, "learning_rate": 4.9999945923018245e-05, "loss": 0.3766, "step": 762 }, { "epoch": 0.0, "learning_rate": 4.999994575952035e-05, "loss": 1.2983, "step": 763 }, { "epoch": 0.0, "learning_rate": 4.999994559577565e-05, "loss": 1.3596, "step": 764 }, { "epoch": 0.0, "learning_rate": 4.999994543178417e-05, "loss": 1.3108, "step": 765 }, { "epoch": 0.0, "learning_rate": 4.99999452675459e-05, "loss": 1.0224, "step": 766 }, { "epoch": 0.0, "learning_rate": 4.999994510306085e-05, "loss": 1.2363, "step": 767 }, { "epoch": 0.0, "learning_rate": 4.9999944938329e-05, "loss": 1.2409, "step": 768 }, { "epoch": 0.0, "learning_rate": 4.999994477335036e-05, "loss": 1.1756, "step": 769 }, { "epoch": 0.0, "learning_rate": 4.9999944608124935e-05, "loss": 0.8086, "step": 770 }, { "epoch": 0.0, "learning_rate": 4.9999944442652726e-05, "loss": 1.2566, "step": 771 }, { "epoch": 0.0, "learning_rate": 4.999994427693372e-05, "loss": 1.3554, "step": 772 }, { "epoch": 0.0, "learning_rate": 4.9999944110967925e-05, "loss": 1.1814, "step": 773 }, { "epoch": 0.0, "learning_rate": 4.999994394475535e-05, "loss": 1.1509, "step": 774 }, { "epoch": 0.0, "learning_rate": 4.999994377829598e-05, "loss": 0.5566, "step": 775 }, { "epoch": 0.0, "learning_rate": 4.999994361158982e-05, "loss": 1.0299, "step": 776 }, { "epoch": 0.0, "learning_rate": 4.9999943444636874e-05, "loss": 1.093, "step": 777 }, { "epoch": 0.0, "learning_rate": 4.999994327743713e-05, "loss": 1.3153, "step": 778 }, { "epoch": 0.0, "learning_rate": 4.999994310999061e-05, "loss": 1.3283, "step": 779 }, { "epoch": 0.0, "learning_rate": 4.9999942942297296e-05, "loss": 1.1775, "step": 780 }, { "epoch": 0.0, "learning_rate": 4.999994277435719e-05, "loss": 1.3292, "step": 781 }, { "epoch": 0.0, "learning_rate": 4.9999942606170296e-05, "loss": 1.416, "step": 782 }, { "epoch": 0.0, "learning_rate": 4.9999942437736616e-05, "loss": 1.2101, "step": 783 }, { "epoch": 0.0, "learning_rate": 4.9999942269056143e-05, "loss": 1.0418, "step": 784 }, { "epoch": 0.0, "learning_rate": 4.9999942100128887e-05, "loss": 1.1471, "step": 785 }, { "epoch": 0.0, "learning_rate": 4.999994193095484e-05, "loss": 1.1467, "step": 786 }, { "epoch": 0.0, "learning_rate": 4.9999941761534005e-05, "loss": 1.0736, "step": 787 }, { "epoch": 0.0, "learning_rate": 4.999994159186638e-05, "loss": 1.2616, "step": 788 }, { "epoch": 0.0, "learning_rate": 4.999994142195197e-05, "loss": 1.2858, "step": 789 }, { "epoch": 0.0, "learning_rate": 4.999994125179076e-05, "loss": 1.3641, "step": 790 }, { "epoch": 0.0, "learning_rate": 4.999994108138277e-05, "loss": 1.2877, "step": 791 }, { "epoch": 0.0, "learning_rate": 4.999994091072799e-05, "loss": 0.8191, "step": 792 }, { "epoch": 0.0, "learning_rate": 4.999994073982642e-05, "loss": 0.6352, "step": 793 }, { "epoch": 0.0, "learning_rate": 4.999994056867806e-05, "loss": 1.0221, "step": 794 }, { "epoch": 0.0, "learning_rate": 4.9999940397282906e-05, "loss": 0.8679, "step": 795 }, { "epoch": 0.0, "learning_rate": 4.9999940225640976e-05, "loss": 1.6629, "step": 796 }, { "epoch": 0.0, "learning_rate": 4.999994005375225e-05, "loss": 1.2877, "step": 797 }, { "epoch": 0.0, "learning_rate": 4.999993988161673e-05, "loss": 1.117, "step": 798 }, { "epoch": 0.0, "learning_rate": 4.999993970923443e-05, "loss": 1.016, "step": 799 }, { "epoch": 0.0, "learning_rate": 4.999993953660534e-05, "loss": 1.0511, "step": 800 }, { "epoch": 0.0, "learning_rate": 4.999993936372946e-05, "loss": 1.0768, "step": 801 }, { "epoch": 0.0, "learning_rate": 4.999993919060679e-05, "loss": 1.0474, "step": 802 }, { "epoch": 0.0, "learning_rate": 4.999993901723733e-05, "loss": 0.9152, "step": 803 }, { "epoch": 0.0, "learning_rate": 4.999993884362108e-05, "loss": 1.1368, "step": 804 }, { "epoch": 0.0, "learning_rate": 4.9999938669758045e-05, "loss": 1.0106, "step": 805 }, { "epoch": 0.0, "learning_rate": 4.999993849564822e-05, "loss": 1.1477, "step": 806 }, { "epoch": 0.0, "learning_rate": 4.9999938321291604e-05, "loss": 1.2261, "step": 807 }, { "epoch": 0.0, "learning_rate": 4.99999381466882e-05, "loss": 0.8828, "step": 808 }, { "epoch": 0.0, "learning_rate": 4.9999937971838004e-05, "loss": 0.9267, "step": 809 }, { "epoch": 0.0, "learning_rate": 4.9999937796741033e-05, "loss": 0.9857, "step": 810 }, { "epoch": 0.0, "learning_rate": 4.9999937621397265e-05, "loss": 0.7243, "step": 811 }, { "epoch": 0.0, "learning_rate": 4.99999374458067e-05, "loss": 0.2533, "step": 812 }, { "epoch": 0.0, "learning_rate": 4.999993726996935e-05, "loss": 0.1383, "step": 813 }, { "epoch": 0.0, "learning_rate": 4.9999937093885217e-05, "loss": 0.1169, "step": 814 }, { "epoch": 0.0, "learning_rate": 4.999993691755429e-05, "loss": 0.5441, "step": 815 }, { "epoch": 0.0, "learning_rate": 4.999993674097658e-05, "loss": 1.1295, "step": 816 }, { "epoch": 0.0, "learning_rate": 4.999993656415208e-05, "loss": 1.2738, "step": 817 }, { "epoch": 0.0, "learning_rate": 4.999993638708079e-05, "loss": 1.1248, "step": 818 }, { "epoch": 0.0, "learning_rate": 4.99999362097627e-05, "loss": 1.1047, "step": 819 }, { "epoch": 0.0, "learning_rate": 4.999993603219784e-05, "loss": 1.0343, "step": 820 }, { "epoch": 0.0, "learning_rate": 4.999993585438618e-05, "loss": 1.3274, "step": 821 }, { "epoch": 0.0, "learning_rate": 4.999993567632773e-05, "loss": 1.233, "step": 822 }, { "epoch": 0.0, "learning_rate": 4.99999354980225e-05, "loss": 0.9655, "step": 823 }, { "epoch": 0.0, "learning_rate": 4.999993531947047e-05, "loss": 1.2446, "step": 824 }, { "epoch": 0.0, "learning_rate": 4.999993514067166e-05, "loss": 1.204, "step": 825 }, { "epoch": 0.0, "learning_rate": 4.999993496162606e-05, "loss": 1.334, "step": 826 }, { "epoch": 0.0, "learning_rate": 4.9999934782333666e-05, "loss": 1.5279, "step": 827 }, { "epoch": 0.0, "learning_rate": 4.999993460279448e-05, "loss": 0.6742, "step": 828 }, { "epoch": 0.0, "learning_rate": 4.999993442300851e-05, "loss": 0.4579, "step": 829 }, { "epoch": 0.0, "learning_rate": 4.999993424297576e-05, "loss": 0.55, "step": 830 }, { "epoch": 0.0, "learning_rate": 4.9999934062696215e-05, "loss": 0.8013, "step": 831 }, { "epoch": 0.0, "learning_rate": 4.999993388216988e-05, "loss": 0.9383, "step": 832 }, { "epoch": 0.0, "learning_rate": 4.999993370139675e-05, "loss": 1.1955, "step": 833 }, { "epoch": 0.0, "learning_rate": 4.999993352037684e-05, "loss": 1.0137, "step": 834 }, { "epoch": 0.0, "learning_rate": 4.999993333911014e-05, "loss": 0.9991, "step": 835 }, { "epoch": 0.0, "learning_rate": 4.9999933157596645e-05, "loss": 0.9199, "step": 836 }, { "epoch": 0.0, "learning_rate": 4.999993297583637e-05, "loss": 0.4115, "step": 837 }, { "epoch": 0.0, "learning_rate": 4.99999327938293e-05, "loss": 0.3562, "step": 838 }, { "epoch": 0.0, "learning_rate": 4.999993261157544e-05, "loss": 1.0957, "step": 839 }, { "epoch": 0.0, "learning_rate": 4.99999324290748e-05, "loss": 0.9675, "step": 840 }, { "epoch": 0.0, "learning_rate": 4.999993224632736e-05, "loss": 0.9914, "step": 841 }, { "epoch": 0.0, "learning_rate": 4.9999932063333144e-05, "loss": 0.9066, "step": 842 }, { "epoch": 0.0, "learning_rate": 4.9999931880092135e-05, "loss": 1.0886, "step": 843 }, { "epoch": 0.0, "learning_rate": 4.999993169660433e-05, "loss": 0.969, "step": 844 }, { "epoch": 0.0, "learning_rate": 4.999993151286974e-05, "loss": 1.0533, "step": 845 }, { "epoch": 0.0, "learning_rate": 4.999993132888836e-05, "loss": 1.0183, "step": 846 }, { "epoch": 0.0, "learning_rate": 4.99999311446602e-05, "loss": 1.1548, "step": 847 }, { "epoch": 0.0, "learning_rate": 4.999993096018524e-05, "loss": 0.987, "step": 848 }, { "epoch": 0.0, "learning_rate": 4.99999307754635e-05, "loss": 1.4268, "step": 849 }, { "epoch": 0.0, "learning_rate": 4.9999930590494956e-05, "loss": 1.5397, "step": 850 }, { "epoch": 0.0, "learning_rate": 4.9999930405279635e-05, "loss": 0.9337, "step": 851 }, { "epoch": 0.0, "learning_rate": 4.999993021981752e-05, "loss": 1.3076, "step": 852 }, { "epoch": 0.0, "learning_rate": 4.999993003410863e-05, "loss": 1.1003, "step": 853 }, { "epoch": 0.0, "learning_rate": 4.9999929848152936e-05, "loss": 1.0851, "step": 854 }, { "epoch": 0.0, "learning_rate": 4.999992966195046e-05, "loss": 1.2169, "step": 855 }, { "epoch": 0.0, "learning_rate": 4.99999294755012e-05, "loss": 1.2069, "step": 856 }, { "epoch": 0.0, "learning_rate": 4.999992928880514e-05, "loss": 1.184, "step": 857 }, { "epoch": 0.0, "learning_rate": 4.99999291018623e-05, "loss": 1.4656, "step": 858 }, { "epoch": 0.0, "learning_rate": 4.999992891467267e-05, "loss": 1.2775, "step": 859 }, { "epoch": 0.0, "learning_rate": 4.999992872723624e-05, "loss": 1.6209, "step": 860 }, { "epoch": 0.0, "learning_rate": 4.9999928539553034e-05, "loss": 1.8363, "step": 861 }, { "epoch": 0.0, "learning_rate": 4.999992835162304e-05, "loss": 1.6992, "step": 862 }, { "epoch": 0.0, "learning_rate": 4.9999928163446254e-05, "loss": 2.3404, "step": 863 }, { "epoch": 0.0, "learning_rate": 4.999992797502268e-05, "loss": 1.5144, "step": 864 }, { "epoch": 0.0, "learning_rate": 4.999992778635231e-05, "loss": 1.5439, "step": 865 }, { "epoch": 0.0, "learning_rate": 4.9999927597435156e-05, "loss": 1.3937, "step": 866 }, { "epoch": 0.0, "learning_rate": 4.999992740827121e-05, "loss": 1.6777, "step": 867 }, { "epoch": 0.0, "learning_rate": 4.999992721886048e-05, "loss": 1.1788, "step": 868 }, { "epoch": 0.0, "learning_rate": 4.999992702920296e-05, "loss": 1.2119, "step": 869 }, { "epoch": 0.0, "learning_rate": 4.999992683929865e-05, "loss": 1.3642, "step": 870 }, { "epoch": 0.0, "learning_rate": 4.999992664914755e-05, "loss": 0.9385, "step": 871 }, { "epoch": 0.0, "learning_rate": 4.999992645874967e-05, "loss": 1.0575, "step": 872 }, { "epoch": 0.0, "learning_rate": 4.999992626810499e-05, "loss": 1.091, "step": 873 }, { "epoch": 0.0, "learning_rate": 4.999992607721353e-05, "loss": 1.205, "step": 874 }, { "epoch": 0.0, "learning_rate": 4.9999925886075275e-05, "loss": 0.9686, "step": 875 }, { "epoch": 0.0, "learning_rate": 4.9999925694690233e-05, "loss": 1.0848, "step": 876 }, { "epoch": 0.0, "learning_rate": 4.999992550305841e-05, "loss": 1.1814, "step": 877 }, { "epoch": 0.0, "learning_rate": 4.999992531117978e-05, "loss": 0.8181, "step": 878 }, { "epoch": 0.0, "learning_rate": 4.999992511905438e-05, "loss": 1.0436, "step": 879 }, { "epoch": 0.0, "learning_rate": 4.999992492668219e-05, "loss": 1.1664, "step": 880 }, { "epoch": 0.0, "learning_rate": 4.9999924734063195e-05, "loss": 1.5344, "step": 881 }, { "epoch": 0.0, "learning_rate": 4.999992454119743e-05, "loss": 1.1005, "step": 882 }, { "epoch": 0.0, "learning_rate": 4.9999924348084864e-05, "loss": 1.3829, "step": 883 }, { "epoch": 0.0, "learning_rate": 4.999992415472551e-05, "loss": 1.0473, "step": 884 }, { "epoch": 0.0, "learning_rate": 4.999992396111938e-05, "loss": 0.9251, "step": 885 }, { "epoch": 0.0, "learning_rate": 4.9999923767266444e-05, "loss": 1.0935, "step": 886 }, { "epoch": 0.0, "learning_rate": 4.999992357316673e-05, "loss": 1.082, "step": 887 }, { "epoch": 0.0, "learning_rate": 4.9999923378820224e-05, "loss": 0.3298, "step": 888 }, { "epoch": 0.0, "learning_rate": 4.999992318422693e-05, "loss": 0.5366, "step": 889 }, { "epoch": 0.0, "learning_rate": 4.999992298938685e-05, "loss": 1.022, "step": 890 }, { "epoch": 0.0, "learning_rate": 4.999992279429997e-05, "loss": 0.8881, "step": 891 }, { "epoch": 0.0, "learning_rate": 4.9999922598966313e-05, "loss": 0.8658, "step": 892 }, { "epoch": 0.0, "learning_rate": 4.9999922403385864e-05, "loss": 1.2105, "step": 893 }, { "epoch": 0.0, "learning_rate": 4.999992220755862e-05, "loss": 1.1332, "step": 894 }, { "epoch": 0.0, "learning_rate": 4.9999922011484604e-05, "loss": 5.3122, "step": 895 }, { "epoch": 0.0, "learning_rate": 4.9999921815163786e-05, "loss": 3.7951, "step": 896 }, { "epoch": 0.0, "learning_rate": 4.999992161859618e-05, "loss": 5.5103, "step": 897 }, { "epoch": 0.0, "learning_rate": 4.9999921421781784e-05, "loss": 3.4476, "step": 898 }, { "epoch": 0.0, "learning_rate": 4.9999921224720605e-05, "loss": 2.2574, "step": 899 }, { "epoch": 0.0, "learning_rate": 4.9999921027412635e-05, "loss": 1.302, "step": 900 }, { "epoch": 0.0, "learning_rate": 4.9999920829857874e-05, "loss": 1.1378, "step": 901 }, { "epoch": 0.0, "learning_rate": 4.999992063205633e-05, "loss": 1.1153, "step": 902 }, { "epoch": 0.0, "learning_rate": 4.9999920434008e-05, "loss": 1.1502, "step": 903 }, { "epoch": 0.0, "learning_rate": 4.9999920235712874e-05, "loss": 1.266, "step": 904 }, { "epoch": 0.0, "learning_rate": 4.9999920037170953e-05, "loss": 1.4186, "step": 905 }, { "epoch": 0.0, "learning_rate": 4.9999919838382255e-05, "loss": 0.9976, "step": 906 }, { "epoch": 0.0, "learning_rate": 4.999991963934676e-05, "loss": 1.2569, "step": 907 }, { "epoch": 0.0, "learning_rate": 4.999991944006448e-05, "loss": 1.2904, "step": 908 }, { "epoch": 0.0, "learning_rate": 4.9999919240535416e-05, "loss": 0.6151, "step": 909 }, { "epoch": 0.0, "learning_rate": 4.999991904075956e-05, "loss": 1.2937, "step": 910 }, { "epoch": 0.0, "learning_rate": 4.9999918840736915e-05, "loss": 1.1849, "step": 911 }, { "epoch": 0.0, "learning_rate": 4.999991864046748e-05, "loss": 1.1805, "step": 912 }, { "epoch": 0.0, "learning_rate": 4.9999918439951255e-05, "loss": 1.2076, "step": 913 }, { "epoch": 0.0, "learning_rate": 4.9999918239188245e-05, "loss": 1.1705, "step": 914 }, { "epoch": 0.0, "learning_rate": 4.9999918038178436e-05, "loss": 0.8461, "step": 915 }, { "epoch": 0.0, "learning_rate": 4.999991783692185e-05, "loss": 1.127, "step": 916 }, { "epoch": 0.0, "learning_rate": 4.999991763541848e-05, "loss": 1.2225, "step": 917 }, { "epoch": 0.0, "learning_rate": 4.999991743366831e-05, "loss": 1.0974, "step": 918 }, { "epoch": 0.0, "learning_rate": 4.999991723167135e-05, "loss": 0.53, "step": 919 }, { "epoch": 0.0, "learning_rate": 4.9999917029427614e-05, "loss": 0.7431, "step": 920 }, { "epoch": 0.0, "learning_rate": 4.9999916826937076e-05, "loss": 1.4294, "step": 921 }, { "epoch": 0.0, "learning_rate": 4.999991662419976e-05, "loss": 1.2934, "step": 922 }, { "epoch": 0.0, "learning_rate": 4.999991642121565e-05, "loss": 0.9691, "step": 923 }, { "epoch": 0.0, "learning_rate": 4.9999916217984755e-05, "loss": 0.9543, "step": 924 }, { "epoch": 0.0, "learning_rate": 4.9999916014507065e-05, "loss": 1.2091, "step": 925 }, { "epoch": 0.0, "learning_rate": 4.999991581078259e-05, "loss": 1.0221, "step": 926 }, { "epoch": 0.0, "learning_rate": 4.9999915606811323e-05, "loss": 1.0449, "step": 927 }, { "epoch": 0.0, "learning_rate": 4.999991540259327e-05, "loss": 0.9845, "step": 928 }, { "epoch": 0.0, "learning_rate": 4.9999915198128436e-05, "loss": 1.1451, "step": 929 }, { "epoch": 0.0, "learning_rate": 4.99999149934168e-05, "loss": 1.0454, "step": 930 }, { "epoch": 0.0, "learning_rate": 4.9999914788458376e-05, "loss": 1.2146, "step": 931 }, { "epoch": 0.0, "learning_rate": 4.999991458325318e-05, "loss": 1.3721, "step": 932 }, { "epoch": 0.0, "learning_rate": 4.9999914377801184e-05, "loss": 1.1491, "step": 933 }, { "epoch": 0.0, "learning_rate": 4.99999141721024e-05, "loss": 0.996, "step": 934 }, { "epoch": 0.0, "learning_rate": 4.999991396615682e-05, "loss": 1.0118, "step": 935 }, { "epoch": 0.0, "learning_rate": 4.999991375996446e-05, "loss": 0.6254, "step": 936 }, { "epoch": 0.0, "learning_rate": 4.9999913553525315e-05, "loss": 0.995, "step": 937 }, { "epoch": 0.0, "learning_rate": 4.999991334683937e-05, "loss": 1.1517, "step": 938 }, { "epoch": 0.0, "learning_rate": 4.9999913139906645e-05, "loss": 1.1732, "step": 939 }, { "epoch": 0.0, "learning_rate": 4.999991293272713e-05, "loss": 0.7338, "step": 940 }, { "epoch": 0.0, "learning_rate": 4.999991272530082e-05, "loss": 1.2936, "step": 941 }, { "epoch": 0.0, "learning_rate": 4.999991251762773e-05, "loss": 1.0981, "step": 942 }, { "epoch": 0.0, "learning_rate": 4.9999912309707855e-05, "loss": 1.2161, "step": 943 }, { "epoch": 0.0, "learning_rate": 4.999991210154118e-05, "loss": 1.5329, "step": 944 }, { "epoch": 0.0, "learning_rate": 4.999991189312772e-05, "loss": 1.1148, "step": 945 }, { "epoch": 0.0, "learning_rate": 4.999991168446747e-05, "loss": 1.5379, "step": 946 }, { "epoch": 0.0, "learning_rate": 4.999991147556043e-05, "loss": 1.166, "step": 947 }, { "epoch": 0.0, "learning_rate": 4.9999911266406606e-05, "loss": 1.0611, "step": 948 }, { "epoch": 0.0, "learning_rate": 4.9999911057006e-05, "loss": 1.2562, "step": 949 }, { "epoch": 0.0, "learning_rate": 4.99999108473586e-05, "loss": 1.1501, "step": 950 }, { "epoch": 0.0, "learning_rate": 4.999991063746441e-05, "loss": 1.2442, "step": 951 }, { "epoch": 0.0, "learning_rate": 4.999991042732343e-05, "loss": 0.8504, "step": 952 }, { "epoch": 0.0, "learning_rate": 4.999991021693566e-05, "loss": 0.6655, "step": 953 }, { "epoch": 0.0, "learning_rate": 4.99999100063011e-05, "loss": 0.8098, "step": 954 }, { "epoch": 0.0, "learning_rate": 4.999990979541976e-05, "loss": 0.6979, "step": 955 }, { "epoch": 0.0, "learning_rate": 4.999990958429163e-05, "loss": 0.6262, "step": 956 }, { "epoch": 0.0, "learning_rate": 4.99999093729167e-05, "loss": 0.9136, "step": 957 }, { "epoch": 0.0, "learning_rate": 4.999990916129499e-05, "loss": 0.6292, "step": 958 }, { "epoch": 0.0, "learning_rate": 4.999990894942649e-05, "loss": 0.7384, "step": 959 }, { "epoch": 0.0, "learning_rate": 4.999990873731121e-05, "loss": 0.6144, "step": 960 }, { "epoch": 0.0, "learning_rate": 4.999990852494913e-05, "loss": 0.8134, "step": 961 }, { "epoch": 0.0, "learning_rate": 4.999990831234026e-05, "loss": 0.6764, "step": 962 }, { "epoch": 0.0, "learning_rate": 4.999990809948461e-05, "loss": 1.2664, "step": 963 }, { "epoch": 0.0, "learning_rate": 4.999990788638217e-05, "loss": 1.2815, "step": 964 }, { "epoch": 0.0, "learning_rate": 4.999990767303293e-05, "loss": 1.1905, "step": 965 }, { "epoch": 0.0, "learning_rate": 4.999990745943692e-05, "loss": 1.2767, "step": 966 }, { "epoch": 0.0, "learning_rate": 4.9999907245594114e-05, "loss": 1.1545, "step": 967 }, { "epoch": 0.0, "learning_rate": 4.999990703150451e-05, "loss": 1.0269, "step": 968 }, { "epoch": 0.0, "learning_rate": 4.999990681716813e-05, "loss": 1.0386, "step": 969 }, { "epoch": 0.0, "learning_rate": 4.9999906602584956e-05, "loss": 1.2325, "step": 970 }, { "epoch": 0.0, "learning_rate": 4.999990638775499e-05, "loss": 1.1714, "step": 971 }, { "epoch": 0.0, "learning_rate": 4.999990617267824e-05, "loss": 1.1213, "step": 972 }, { "epoch": 0.0, "learning_rate": 4.99999059573547e-05, "loss": 1.087, "step": 973 }, { "epoch": 0.0, "learning_rate": 4.999990574178438e-05, "loss": 0.8446, "step": 974 }, { "epoch": 0.0, "learning_rate": 4.999990552596726e-05, "loss": 1.1204, "step": 975 }, { "epoch": 0.0, "learning_rate": 4.999990530990336e-05, "loss": 1.1727, "step": 976 }, { "epoch": 0.0, "learning_rate": 4.9999905093592664e-05, "loss": 1.0612, "step": 977 }, { "epoch": 0.0, "learning_rate": 4.999990487703518e-05, "loss": 1.2293, "step": 978 }, { "epoch": 0.0, "learning_rate": 4.999990466023091e-05, "loss": 1.2555, "step": 979 }, { "epoch": 0.0, "learning_rate": 4.9999904443179855e-05, "loss": 1.0958, "step": 980 }, { "epoch": 0.0, "learning_rate": 4.9999904225882e-05, "loss": 1.0161, "step": 981 }, { "epoch": 0.0, "learning_rate": 4.999990400833737e-05, "loss": 1.1351, "step": 982 }, { "epoch": 0.0, "learning_rate": 4.999990379054594e-05, "loss": 1.3182, "step": 983 }, { "epoch": 0.0, "learning_rate": 4.9999903572507736e-05, "loss": 1.3075, "step": 984 }, { "epoch": 0.0, "learning_rate": 4.999990335422273e-05, "loss": 1.2306, "step": 985 }, { "epoch": 0.0, "learning_rate": 4.999990313569094e-05, "loss": 1.2427, "step": 986 }, { "epoch": 0.0, "learning_rate": 4.999990291691237e-05, "loss": 0.4983, "step": 987 }, { "epoch": 0.0, "learning_rate": 4.9999902697886994e-05, "loss": 0.4873, "step": 988 }, { "epoch": 0.0, "learning_rate": 4.999990247861484e-05, "loss": 0.4938, "step": 989 }, { "epoch": 0.0, "learning_rate": 4.99999022590959e-05, "loss": 0.9492, "step": 990 }, { "epoch": 0.0, "learning_rate": 4.9999902039330167e-05, "loss": 1.1465, "step": 991 }, { "epoch": 0.0, "learning_rate": 4.999990181931765e-05, "loss": 1.27, "step": 992 }, { "epoch": 0.0, "learning_rate": 4.999990159905834e-05, "loss": 0.4938, "step": 993 }, { "epoch": 0.0, "learning_rate": 4.9999901378552236e-05, "loss": 1.0486, "step": 994 }, { "epoch": 0.0, "learning_rate": 4.9999901157799356e-05, "loss": 1.1557, "step": 995 }, { "epoch": 0.0, "learning_rate": 4.999990093679968e-05, "loss": 1.3443, "step": 996 }, { "epoch": 0.0, "learning_rate": 4.9999900715553216e-05, "loss": 1.1414, "step": 997 }, { "epoch": 0.0, "learning_rate": 4.999990049405996e-05, "loss": 1.1585, "step": 998 }, { "epoch": 0.0, "learning_rate": 4.999990027231992e-05, "loss": 1.1172, "step": 999 }, { "epoch": 0.0, "learning_rate": 4.999990005033309e-05, "loss": 1.2645, "step": 1000 }, { "epoch": 0.0, "eval_loss": 1.1082977056503296, "eval_runtime": 84.9058, "eval_samples_per_second": 16.312, "eval_steps_per_second": 4.087, "step": 1000 }, { "epoch": 0.0, "learning_rate": 4.999989982809948e-05, "loss": 1.1184, "step": 1001 }, { "epoch": 0.0, "learning_rate": 4.999989960561907e-05, "loss": 1.1318, "step": 1002 }, { "epoch": 0.0, "learning_rate": 4.999989938289188e-05, "loss": 1.0316, "step": 1003 }, { "epoch": 0.0, "learning_rate": 4.9999899159917895e-05, "loss": 1.1357, "step": 1004 }, { "epoch": 0.0, "learning_rate": 4.999989893669712e-05, "loss": 1.1337, "step": 1005 }, { "epoch": 0.0, "learning_rate": 4.999989871322956e-05, "loss": 0.9274, "step": 1006 }, { "epoch": 0.0, "learning_rate": 4.999989848951522e-05, "loss": 0.9013, "step": 1007 }, { "epoch": 0.0, "learning_rate": 4.999989826555408e-05, "loss": 0.8428, "step": 1008 }, { "epoch": 0.0, "learning_rate": 4.9999898041346155e-05, "loss": 0.8037, "step": 1009 }, { "epoch": 0.0, "learning_rate": 4.9999897816891436e-05, "loss": 0.7398, "step": 1010 }, { "epoch": 0.0, "learning_rate": 4.999989759218994e-05, "loss": 0.6844, "step": 1011 }, { "epoch": 0.0, "learning_rate": 4.999989736724165e-05, "loss": 0.577, "step": 1012 }, { "epoch": 0.0, "learning_rate": 4.999989714204657e-05, "loss": 0.6148, "step": 1013 }, { "epoch": 0.0, "learning_rate": 4.999989691660471e-05, "loss": 0.8637, "step": 1014 }, { "epoch": 0.0, "learning_rate": 4.9999896690916045e-05, "loss": 1.3844, "step": 1015 }, { "epoch": 0.0, "learning_rate": 4.99998964649806e-05, "loss": 1.4286, "step": 1016 }, { "epoch": 0.0, "learning_rate": 4.999989623879837e-05, "loss": 0.8812, "step": 1017 }, { "epoch": 0.0, "learning_rate": 4.999989601236935e-05, "loss": 0.8998, "step": 1018 }, { "epoch": 0.0, "learning_rate": 4.999989578569354e-05, "loss": 1.0472, "step": 1019 }, { "epoch": 0.0, "learning_rate": 4.999989555877094e-05, "loss": 1.1919, "step": 1020 }, { "epoch": 0.0, "learning_rate": 4.9999895331601556e-05, "loss": 1.1963, "step": 1021 }, { "epoch": 0.0, "learning_rate": 4.999989510418538e-05, "loss": 1.198, "step": 1022 }, { "epoch": 0.0, "learning_rate": 4.999989487652242e-05, "loss": 1.0706, "step": 1023 }, { "epoch": 0.0, "learning_rate": 4.9999894648612666e-05, "loss": 1.2298, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.999989442045613e-05, "loss": 1.3413, "step": 1025 }, { "epoch": 0.0, "learning_rate": 4.99998941920528e-05, "loss": 0.837, "step": 1026 }, { "epoch": 0.0, "learning_rate": 4.999989396340268e-05, "loss": 1.1898, "step": 1027 }, { "epoch": 0.0, "learning_rate": 4.9999893734505775e-05, "loss": 1.1632, "step": 1028 }, { "epoch": 0.0, "learning_rate": 4.9999893505362086e-05, "loss": 1.014, "step": 1029 }, { "epoch": 0.0, "learning_rate": 4.99998932759716e-05, "loss": 0.9097, "step": 1030 }, { "epoch": 0.0, "learning_rate": 4.999989304633433e-05, "loss": 1.3816, "step": 1031 }, { "epoch": 0.0, "learning_rate": 4.9999892816450274e-05, "loss": 1.1644, "step": 1032 }, { "epoch": 0.0, "learning_rate": 4.9999892586319425e-05, "loss": 1.205, "step": 1033 }, { "epoch": 0.0, "learning_rate": 4.999989235594179e-05, "loss": 1.3464, "step": 1034 }, { "epoch": 0.0, "learning_rate": 4.9999892125317366e-05, "loss": 1.2883, "step": 1035 }, { "epoch": 0.0, "learning_rate": 4.999989189444615e-05, "loss": 1.085, "step": 1036 }, { "epoch": 0.0, "learning_rate": 4.9999891663328155e-05, "loss": 1.1615, "step": 1037 }, { "epoch": 0.0, "learning_rate": 4.999989143196336e-05, "loss": 1.2407, "step": 1038 }, { "epoch": 0.0, "learning_rate": 4.9999891200351784e-05, "loss": 1.1057, "step": 1039 }, { "epoch": 0.0, "learning_rate": 4.999989096849342e-05, "loss": 1.2035, "step": 1040 }, { "epoch": 0.0, "learning_rate": 4.999989073638827e-05, "loss": 1.2299, "step": 1041 }, { "epoch": 0.0, "learning_rate": 4.9999890504036315e-05, "loss": 1.3304, "step": 1042 }, { "epoch": 0.0, "learning_rate": 4.999989027143759e-05, "loss": 1.0313, "step": 1043 }, { "epoch": 0.0, "learning_rate": 4.999989003859207e-05, "loss": 1.3824, "step": 1044 }, { "epoch": 0.0, "learning_rate": 4.999988980549976e-05, "loss": 1.0822, "step": 1045 }, { "epoch": 0.0, "learning_rate": 4.9999889572160665e-05, "loss": 1.9536, "step": 1046 }, { "epoch": 0.0, "learning_rate": 4.999988933857478e-05, "loss": 1.9319, "step": 1047 }, { "epoch": 0.0, "learning_rate": 4.99998891047421e-05, "loss": 1.1924, "step": 1048 }, { "epoch": 0.0, "learning_rate": 4.999988887066265e-05, "loss": 0.978, "step": 1049 }, { "epoch": 0.0, "learning_rate": 4.99998886363364e-05, "loss": 1.021, "step": 1050 }, { "epoch": 0.0, "learning_rate": 4.999988840176336e-05, "loss": 1.1141, "step": 1051 }, { "epoch": 0.0, "learning_rate": 4.999988816694353e-05, "loss": 1.2393, "step": 1052 }, { "epoch": 0.0, "learning_rate": 4.999988793187692e-05, "loss": 1.2058, "step": 1053 }, { "epoch": 0.0, "learning_rate": 4.9999887696563514e-05, "loss": 1.3326, "step": 1054 }, { "epoch": 0.0, "learning_rate": 4.9999887461003326e-05, "loss": 1.24, "step": 1055 }, { "epoch": 0.0, "learning_rate": 4.999988722519634e-05, "loss": 1.3852, "step": 1056 }, { "epoch": 0.0, "learning_rate": 4.9999886989142576e-05, "loss": 0.7316, "step": 1057 }, { "epoch": 0.0, "learning_rate": 4.999988675284202e-05, "loss": 1.0525, "step": 1058 }, { "epoch": 0.0, "learning_rate": 4.9999886516294666e-05, "loss": 1.064, "step": 1059 }, { "epoch": 0.0, "learning_rate": 4.999988627950054e-05, "loss": 1.0082, "step": 1060 }, { "epoch": 0.0, "learning_rate": 4.999988604245962e-05, "loss": 0.9874, "step": 1061 }, { "epoch": 0.0, "learning_rate": 4.999988580517191e-05, "loss": 0.9579, "step": 1062 }, { "epoch": 0.0, "learning_rate": 4.99998855676374e-05, "loss": 1.0129, "step": 1063 }, { "epoch": 0.0, "learning_rate": 4.9999885329856125e-05, "loss": 1.2541, "step": 1064 }, { "epoch": 0.0, "learning_rate": 4.999988509182805e-05, "loss": 1.1398, "step": 1065 }, { "epoch": 0.0, "learning_rate": 4.999988485355318e-05, "loss": 0.6827, "step": 1066 }, { "epoch": 0.0, "learning_rate": 4.9999884615031536e-05, "loss": 0.9613, "step": 1067 }, { "epoch": 0.0, "learning_rate": 4.999988437626309e-05, "loss": 1.0136, "step": 1068 }, { "epoch": 0.0, "learning_rate": 4.9999884137247864e-05, "loss": 0.9416, "step": 1069 }, { "epoch": 0.0, "learning_rate": 4.9999883897985844e-05, "loss": 1.2144, "step": 1070 }, { "epoch": 0.0, "learning_rate": 4.9999883658477046e-05, "loss": 1.235, "step": 1071 }, { "epoch": 0.0, "learning_rate": 4.999988341872145e-05, "loss": 0.8769, "step": 1072 }, { "epoch": 0.0, "learning_rate": 4.999988317871906e-05, "loss": 0.8905, "step": 1073 }, { "epoch": 0.0, "learning_rate": 4.9999882938469896e-05, "loss": 1.2377, "step": 1074 }, { "epoch": 0.0, "learning_rate": 4.999988269797394e-05, "loss": 1.2858, "step": 1075 }, { "epoch": 0.0, "learning_rate": 4.999988245723119e-05, "loss": 1.1109, "step": 1076 }, { "epoch": 0.0, "learning_rate": 4.9999882216241656e-05, "loss": 1.2624, "step": 1077 }, { "epoch": 0.0, "learning_rate": 4.999988197500533e-05, "loss": 0.8667, "step": 1078 }, { "epoch": 0.0, "learning_rate": 4.999988173352222e-05, "loss": 1.0818, "step": 1079 }, { "epoch": 0.0, "learning_rate": 4.999988149179232e-05, "loss": 0.8727, "step": 1080 }, { "epoch": 0.0, "learning_rate": 4.9999881249815634e-05, "loss": 1.1173, "step": 1081 }, { "epoch": 0.0, "learning_rate": 4.9999881007592156e-05, "loss": 1.2649, "step": 1082 }, { "epoch": 0.0, "learning_rate": 4.999988076512189e-05, "loss": 1.3832, "step": 1083 }, { "epoch": 0.0, "learning_rate": 4.999988052240484e-05, "loss": 1.2634, "step": 1084 }, { "epoch": 0.0, "learning_rate": 4.9999880279440995e-05, "loss": 1.2364, "step": 1085 }, { "epoch": 0.0, "learning_rate": 4.9999880036230365e-05, "loss": 1.3662, "step": 1086 }, { "epoch": 0.0, "learning_rate": 4.999987979277295e-05, "loss": 1.0146, "step": 1087 }, { "epoch": 0.0, "learning_rate": 4.9999879549068737e-05, "loss": 1.1297, "step": 1088 }, { "epoch": 0.0, "learning_rate": 4.9999879305117745e-05, "loss": 1.1345, "step": 1089 }, { "epoch": 0.0, "learning_rate": 4.999987906091996e-05, "loss": 0.7416, "step": 1090 }, { "epoch": 0.0, "learning_rate": 4.9999878816475395e-05, "loss": 0.8146, "step": 1091 }, { "epoch": 0.0, "learning_rate": 4.9999878571784037e-05, "loss": 0.7677, "step": 1092 }, { "epoch": 0.0, "learning_rate": 4.9999878326845886e-05, "loss": 0.7257, "step": 1093 }, { "epoch": 0.0, "learning_rate": 4.999987808166096e-05, "loss": 0.7448, "step": 1094 }, { "epoch": 0.0, "learning_rate": 4.9999877836229224e-05, "loss": 0.7, "step": 1095 }, { "epoch": 0.0, "learning_rate": 4.999987759055071e-05, "loss": 0.6121, "step": 1096 }, { "epoch": 0.0, "learning_rate": 4.999987734462541e-05, "loss": 0.7789, "step": 1097 }, { "epoch": 0.0, "learning_rate": 4.999987709845333e-05, "loss": 0.649, "step": 1098 }, { "epoch": 0.0, "learning_rate": 4.999987685203445e-05, "loss": 0.8089, "step": 1099 }, { "epoch": 0.0, "learning_rate": 4.9999876605368786e-05, "loss": 0.7264, "step": 1100 }, { "epoch": 0.0, "learning_rate": 4.999987635845633e-05, "loss": 0.6354, "step": 1101 }, { "epoch": 0.0, "learning_rate": 4.999987611129709e-05, "loss": 0.7173, "step": 1102 }, { "epoch": 0.0, "learning_rate": 4.999987586389106e-05, "loss": 0.7339, "step": 1103 }, { "epoch": 0.0, "learning_rate": 4.999987561623824e-05, "loss": 0.7132, "step": 1104 }, { "epoch": 0.0, "learning_rate": 4.9999875368338634e-05, "loss": 1.0778, "step": 1105 }, { "epoch": 0.0, "learning_rate": 4.999987512019224e-05, "loss": 1.1789, "step": 1106 }, { "epoch": 0.0, "learning_rate": 4.999987487179906e-05, "loss": 1.0053, "step": 1107 }, { "epoch": 0.0, "learning_rate": 4.999987462315908e-05, "loss": 1.1175, "step": 1108 }, { "epoch": 0.0, "learning_rate": 4.999987437427233e-05, "loss": 1.2642, "step": 1109 }, { "epoch": 0.0, "learning_rate": 4.999987412513878e-05, "loss": 1.1858, "step": 1110 }, { "epoch": 0.0, "learning_rate": 4.999987387575844e-05, "loss": 1.1409, "step": 1111 }, { "epoch": 0.0, "learning_rate": 4.9999873626131323e-05, "loss": 1.4153, "step": 1112 }, { "epoch": 0.0, "learning_rate": 4.9999873376257404e-05, "loss": 1.3314, "step": 1113 }, { "epoch": 0.0, "learning_rate": 4.999987312613671e-05, "loss": 1.2541, "step": 1114 }, { "epoch": 0.0, "learning_rate": 4.9999872875769216e-05, "loss": 1.2764, "step": 1115 }, { "epoch": 0.0, "learning_rate": 4.999987262515494e-05, "loss": 1.5257, "step": 1116 }, { "epoch": 0.0, "learning_rate": 4.999987237429388e-05, "loss": 1.2683, "step": 1117 }, { "epoch": 0.0, "learning_rate": 4.999987212318602e-05, "loss": 0.6983, "step": 1118 }, { "epoch": 0.0, "learning_rate": 4.9999871871831385e-05, "loss": 1.0439, "step": 1119 }, { "epoch": 0.0, "learning_rate": 4.999987162022995e-05, "loss": 1.1931, "step": 1120 }, { "epoch": 0.0, "learning_rate": 4.999987136838173e-05, "loss": 0.7782, "step": 1121 }, { "epoch": 0.0, "learning_rate": 4.9999871116286724e-05, "loss": 1.1135, "step": 1122 }, { "epoch": 0.0, "learning_rate": 4.999987086394493e-05, "loss": 1.1923, "step": 1123 }, { "epoch": 0.0, "learning_rate": 4.999987061135635e-05, "loss": 1.0822, "step": 1124 }, { "epoch": 0.0, "learning_rate": 4.999987035852098e-05, "loss": 1.1745, "step": 1125 }, { "epoch": 0.0, "learning_rate": 4.999987010543882e-05, "loss": 0.9733, "step": 1126 }, { "epoch": 0.0, "learning_rate": 4.999986985210987e-05, "loss": 0.8489, "step": 1127 }, { "epoch": 0.0, "learning_rate": 4.999986959853414e-05, "loss": 1.1524, "step": 1128 }, { "epoch": 0.0, "learning_rate": 4.9999869344711615e-05, "loss": 1.1779, "step": 1129 }, { "epoch": 0.0, "learning_rate": 4.99998690906423e-05, "loss": 1.19, "step": 1130 }, { "epoch": 0.0, "learning_rate": 4.99998688363262e-05, "loss": 1.0589, "step": 1131 }, { "epoch": 0.0, "learning_rate": 4.999986858176332e-05, "loss": 1.1153, "step": 1132 }, { "epoch": 0.0, "learning_rate": 4.9999868326953635e-05, "loss": 1.199, "step": 1133 }, { "epoch": 0.0, "learning_rate": 4.9999868071897174e-05, "loss": 0.846, "step": 1134 }, { "epoch": 0.0, "learning_rate": 4.999986781659392e-05, "loss": 0.9576, "step": 1135 }, { "epoch": 0.0, "learning_rate": 4.999986756104388e-05, "loss": 0.5289, "step": 1136 }, { "epoch": 0.0, "learning_rate": 4.999986730524705e-05, "loss": 0.4011, "step": 1137 }, { "epoch": 0.0, "learning_rate": 4.999986704920343e-05, "loss": 0.2992, "step": 1138 }, { "epoch": 0.0, "learning_rate": 4.999986679291303e-05, "loss": 0.2196, "step": 1139 }, { "epoch": 0.0, "learning_rate": 4.999986653637584e-05, "loss": 0.1587, "step": 1140 }, { "epoch": 0.0, "learning_rate": 4.9999866279591856e-05, "loss": 0.2478, "step": 1141 }, { "epoch": 0.0, "learning_rate": 4.9999866022561084e-05, "loss": 0.2585, "step": 1142 }, { "epoch": 0.0, "learning_rate": 4.999986576528353e-05, "loss": 0.2091, "step": 1143 }, { "epoch": 0.0, "learning_rate": 4.999986550775918e-05, "loss": 0.3503, "step": 1144 }, { "epoch": 0.0, "learning_rate": 4.9999865249988045e-05, "loss": 0.1784, "step": 1145 }, { "epoch": 0.0, "learning_rate": 4.999986499197013e-05, "loss": 0.2116, "step": 1146 }, { "epoch": 0.0, "learning_rate": 4.999986473370542e-05, "loss": 0.1828, "step": 1147 }, { "epoch": 0.0, "learning_rate": 4.999986447519392e-05, "loss": 0.2068, "step": 1148 }, { "epoch": 0.0, "learning_rate": 4.999986421643563e-05, "loss": 0.2104, "step": 1149 }, { "epoch": 0.0, "learning_rate": 4.999986395743056e-05, "loss": 1.1287, "step": 1150 }, { "epoch": 0.0, "learning_rate": 4.999986369817869e-05, "loss": 1.2888, "step": 1151 }, { "epoch": 0.0, "learning_rate": 4.999986343868004e-05, "loss": 1.1025, "step": 1152 }, { "epoch": 0.0, "learning_rate": 4.99998631789346e-05, "loss": 1.352, "step": 1153 }, { "epoch": 0.0, "learning_rate": 4.999986291894238e-05, "loss": 1.4697, "step": 1154 }, { "epoch": 0.0, "learning_rate": 4.999986265870336e-05, "loss": 1.2039, "step": 1155 }, { "epoch": 0.0, "learning_rate": 4.9999862398217556e-05, "loss": 1.298, "step": 1156 }, { "epoch": 0.0, "learning_rate": 4.9999862137484965e-05, "loss": 1.1438, "step": 1157 }, { "epoch": 0.0, "learning_rate": 4.999986187650558e-05, "loss": 1.1965, "step": 1158 }, { "epoch": 0.0, "learning_rate": 4.999986161527942e-05, "loss": 1.2826, "step": 1159 }, { "epoch": 0.0, "learning_rate": 4.9999861353806464e-05, "loss": 0.9909, "step": 1160 }, { "epoch": 0.0, "learning_rate": 4.999986109208672e-05, "loss": 1.1301, "step": 1161 }, { "epoch": 0.0, "learning_rate": 4.9999860830120186e-05, "loss": 1.2068, "step": 1162 }, { "epoch": 0.0, "learning_rate": 4.9999860567906866e-05, "loss": 1.205, "step": 1163 }, { "epoch": 0.0, "learning_rate": 4.9999860305446755e-05, "loss": 0.9465, "step": 1164 }, { "epoch": 0.0, "learning_rate": 4.9999860042739866e-05, "loss": 1.0319, "step": 1165 }, { "epoch": 0.0, "learning_rate": 4.999985977978618e-05, "loss": 1.2249, "step": 1166 }, { "epoch": 0.0, "learning_rate": 4.9999859516585706e-05, "loss": 1.1404, "step": 1167 }, { "epoch": 0.0, "learning_rate": 4.999985925313845e-05, "loss": 1.1021, "step": 1168 }, { "epoch": 0.0, "learning_rate": 4.99998589894444e-05, "loss": 0.7697, "step": 1169 }, { "epoch": 0.0, "learning_rate": 4.999985872550356e-05, "loss": 1.2475, "step": 1170 }, { "epoch": 0.0, "learning_rate": 4.9999858461315936e-05, "loss": 1.346, "step": 1171 }, { "epoch": 0.0, "learning_rate": 4.999985819688152e-05, "loss": 1.2817, "step": 1172 }, { "epoch": 0.0, "learning_rate": 4.9999857932200326e-05, "loss": 0.905, "step": 1173 }, { "epoch": 0.0, "learning_rate": 4.999985766727233e-05, "loss": 0.8015, "step": 1174 }, { "epoch": 0.0, "learning_rate": 4.999985740209756e-05, "loss": 0.9413, "step": 1175 }, { "epoch": 0.0, "learning_rate": 4.9999857136675994e-05, "loss": 1.1069, "step": 1176 }, { "epoch": 0.0, "learning_rate": 4.999985687100764e-05, "loss": 1.0106, "step": 1177 }, { "epoch": 0.0, "learning_rate": 4.9999856605092496e-05, "loss": 1.1865, "step": 1178 }, { "epoch": 0.0, "learning_rate": 4.9999856338930566e-05, "loss": 1.0042, "step": 1179 }, { "epoch": 0.0, "learning_rate": 4.999985607252186e-05, "loss": 1.2633, "step": 1180 }, { "epoch": 0.0, "learning_rate": 4.9999855805866345e-05, "loss": 1.1735, "step": 1181 }, { "epoch": 0.0, "learning_rate": 4.9999855538964055e-05, "loss": 1.2769, "step": 1182 }, { "epoch": 0.0, "learning_rate": 4.999985527181497e-05, "loss": 1.1381, "step": 1183 }, { "epoch": 0.0, "learning_rate": 4.9999855004419105e-05, "loss": 1.0372, "step": 1184 }, { "epoch": 0.0, "learning_rate": 4.999985473677644e-05, "loss": 1.5588, "step": 1185 }, { "epoch": 0.0, "learning_rate": 4.9999854468887e-05, "loss": 1.1077, "step": 1186 }, { "epoch": 0.0, "learning_rate": 4.999985420075077e-05, "loss": 1.1749, "step": 1187 }, { "epoch": 0.0, "learning_rate": 4.999985393236775e-05, "loss": 0.679, "step": 1188 }, { "epoch": 0.0, "learning_rate": 4.999985366373794e-05, "loss": 1.0461, "step": 1189 }, { "epoch": 0.0, "learning_rate": 4.999985339486134e-05, "loss": 1.2824, "step": 1190 }, { "epoch": 0.0, "learning_rate": 4.999985312573795e-05, "loss": 1.0176, "step": 1191 }, { "epoch": 0.0, "learning_rate": 4.9999852856367776e-05, "loss": 1.2261, "step": 1192 }, { "epoch": 0.0, "learning_rate": 4.999985258675082e-05, "loss": 1.4413, "step": 1193 }, { "epoch": 0.0, "learning_rate": 4.9999852316887064e-05, "loss": 1.1552, "step": 1194 }, { "epoch": 0.0, "learning_rate": 4.999985204677653e-05, "loss": 1.1012, "step": 1195 }, { "epoch": 0.0, "learning_rate": 4.999985177641921e-05, "loss": 1.3743, "step": 1196 }, { "epoch": 0.0, "learning_rate": 4.999985150581509e-05, "loss": 1.2877, "step": 1197 }, { "epoch": 0.0, "learning_rate": 4.999985123496419e-05, "loss": 1.0601, "step": 1198 }, { "epoch": 0.0, "learning_rate": 4.9999850963866504e-05, "loss": 1.0229, "step": 1199 }, { "epoch": 0.0, "learning_rate": 4.999985069252202e-05, "loss": 0.5861, "step": 1200 }, { "epoch": 0.0, "learning_rate": 4.999985042093076e-05, "loss": 1.0952, "step": 1201 }, { "epoch": 0.0, "learning_rate": 4.9999850149092705e-05, "loss": 0.8716, "step": 1202 }, { "epoch": 0.0, "learning_rate": 4.999984987700787e-05, "loss": 1.3299, "step": 1203 }, { "epoch": 0.0, "learning_rate": 4.999984960467623e-05, "loss": 1.3952, "step": 1204 }, { "epoch": 0.0, "learning_rate": 4.9999849332097816e-05, "loss": 1.201, "step": 1205 }, { "epoch": 0.0, "learning_rate": 4.999984905927261e-05, "loss": 0.9686, "step": 1206 }, { "epoch": 0.0, "learning_rate": 4.999984878620061e-05, "loss": 1.1564, "step": 1207 }, { "epoch": 0.0, "learning_rate": 4.999984851288183e-05, "loss": 0.7878, "step": 1208 }, { "epoch": 0.0, "learning_rate": 4.9999848239316264e-05, "loss": 1.445, "step": 1209 }, { "epoch": 0.0, "learning_rate": 4.999984796550391e-05, "loss": 1.3462, "step": 1210 }, { "epoch": 0.0, "learning_rate": 4.999984769144476e-05, "loss": 1.3696, "step": 1211 }, { "epoch": 0.0, "learning_rate": 4.999984741713883e-05, "loss": 1.0053, "step": 1212 }, { "epoch": 0.0, "learning_rate": 4.999984714258611e-05, "loss": 1.0261, "step": 1213 }, { "epoch": 0.0, "learning_rate": 4.99998468677866e-05, "loss": 1.2284, "step": 1214 }, { "epoch": 0.0, "learning_rate": 4.99998465927403e-05, "loss": 1.2134, "step": 1215 }, { "epoch": 0.0, "learning_rate": 4.999984631744722e-05, "loss": 1.3078, "step": 1216 }, { "epoch": 0.0, "learning_rate": 4.9999846041907336e-05, "loss": 1.2526, "step": 1217 }, { "epoch": 0.0, "learning_rate": 4.999984576612068e-05, "loss": 1.0028, "step": 1218 }, { "epoch": 0.0, "learning_rate": 4.999984549008723e-05, "loss": 0.9113, "step": 1219 }, { "epoch": 0.0, "learning_rate": 4.9999845213807e-05, "loss": 0.8816, "step": 1220 }, { "epoch": 0.0, "learning_rate": 4.999984493727997e-05, "loss": 0.8277, "step": 1221 }, { "epoch": 0.0, "learning_rate": 4.999984466050616e-05, "loss": 0.8075, "step": 1222 }, { "epoch": 0.0, "learning_rate": 4.999984438348556e-05, "loss": 0.7724, "step": 1223 }, { "epoch": 0.0, "learning_rate": 4.999984410621817e-05, "loss": 0.9919, "step": 1224 }, { "epoch": 0.0, "learning_rate": 4.9999843828703994e-05, "loss": 1.1758, "step": 1225 }, { "epoch": 0.0, "learning_rate": 4.9999843550943026e-05, "loss": 1.1239, "step": 1226 }, { "epoch": 0.0, "learning_rate": 4.999984327293528e-05, "loss": 1.1041, "step": 1227 }, { "epoch": 0.0, "learning_rate": 4.9999842994680734e-05, "loss": 1.1451, "step": 1228 }, { "epoch": 0.0, "learning_rate": 4.9999842716179405e-05, "loss": 1.1521, "step": 1229 }, { "epoch": 0.0, "learning_rate": 4.999984243743129e-05, "loss": 1.0867, "step": 1230 }, { "epoch": 0.0, "learning_rate": 4.9999842158436385e-05, "loss": 1.0515, "step": 1231 }, { "epoch": 0.0, "learning_rate": 4.99998418791947e-05, "loss": 1.0314, "step": 1232 }, { "epoch": 0.0, "learning_rate": 4.999984159970621e-05, "loss": 0.9818, "step": 1233 }, { "epoch": 0.0, "learning_rate": 4.9999841319970945e-05, "loss": 1.0535, "step": 1234 }, { "epoch": 0.0, "learning_rate": 4.9999841039988894e-05, "loss": 1.0195, "step": 1235 }, { "epoch": 0.0, "learning_rate": 4.9999840759760044e-05, "loss": 1.1967, "step": 1236 }, { "epoch": 0.0, "learning_rate": 4.9999840479284416e-05, "loss": 1.082, "step": 1237 }, { "epoch": 0.0, "learning_rate": 4.9999840198562e-05, "loss": 1.3885, "step": 1238 }, { "epoch": 0.0, "learning_rate": 4.9999839917592786e-05, "loss": 0.8606, "step": 1239 }, { "epoch": 0.0, "learning_rate": 4.99998396363768e-05, "loss": 1.3506, "step": 1240 }, { "epoch": 0.0, "learning_rate": 4.999983935491401e-05, "loss": 1.1122, "step": 1241 }, { "epoch": 0.0, "learning_rate": 4.999983907320444e-05, "loss": 1.3343, "step": 1242 }, { "epoch": 0.0, "learning_rate": 4.999983879124808e-05, "loss": 1.3739, "step": 1243 }, { "epoch": 0.0, "learning_rate": 4.999983850904494e-05, "loss": 1.3176, "step": 1244 }, { "epoch": 0.0, "learning_rate": 4.9999838226594994e-05, "loss": 0.8282, "step": 1245 }, { "epoch": 0.0, "learning_rate": 4.9999837943898276e-05, "loss": 0.6075, "step": 1246 }, { "epoch": 0.0, "learning_rate": 4.999983766095477e-05, "loss": 0.7719, "step": 1247 }, { "epoch": 0.0, "learning_rate": 4.9999837377764467e-05, "loss": 0.5532, "step": 1248 }, { "epoch": 0.0, "learning_rate": 4.999983709432738e-05, "loss": 0.5228, "step": 1249 }, { "epoch": 0.0, "learning_rate": 4.999983681064351e-05, "loss": 0.6887, "step": 1250 }, { "epoch": 0.0, "learning_rate": 4.999983652671285e-05, "loss": 1.1635, "step": 1251 }, { "epoch": 0.0, "learning_rate": 4.99998362425354e-05, "loss": 1.174, "step": 1252 }, { "epoch": 0.0, "learning_rate": 4.9999835958111165e-05, "loss": 1.3225, "step": 1253 }, { "epoch": 0.0, "learning_rate": 4.9999835673440136e-05, "loss": 1.2051, "step": 1254 }, { "epoch": 0.0, "learning_rate": 4.999983538852232e-05, "loss": 1.1052, "step": 1255 }, { "epoch": 0.0, "learning_rate": 4.999983510335772e-05, "loss": 1.0838, "step": 1256 }, { "epoch": 0.0, "learning_rate": 4.999983481794633e-05, "loss": 0.8987, "step": 1257 }, { "epoch": 0.0, "learning_rate": 4.999983453228816e-05, "loss": 0.5425, "step": 1258 }, { "epoch": 0.0, "learning_rate": 4.999983424638319e-05, "loss": 0.5464, "step": 1259 }, { "epoch": 0.0, "learning_rate": 4.999983396023144e-05, "loss": 0.5096, "step": 1260 }, { "epoch": 0.0, "learning_rate": 4.99998336738329e-05, "loss": 1.0058, "step": 1261 }, { "epoch": 0.0, "learning_rate": 4.9999833387187575e-05, "loss": 1.2524, "step": 1262 }, { "epoch": 0.0, "learning_rate": 4.9999833100295456e-05, "loss": 1.0138, "step": 1263 }, { "epoch": 0.0, "learning_rate": 4.999983281315656e-05, "loss": 1.1427, "step": 1264 }, { "epoch": 0.0, "learning_rate": 4.999983252577086e-05, "loss": 1.548, "step": 1265 }, { "epoch": 0.0, "learning_rate": 4.999983223813838e-05, "loss": 1.1719, "step": 1266 }, { "epoch": 0.0, "learning_rate": 4.999983195025911e-05, "loss": 1.4455, "step": 1267 }, { "epoch": 0.0, "learning_rate": 4.999983166213306e-05, "loss": 1.4792, "step": 1268 }, { "epoch": 0.0, "learning_rate": 4.999983137376022e-05, "loss": 1.1389, "step": 1269 }, { "epoch": 0.0, "learning_rate": 4.9999831085140594e-05, "loss": 1.0004, "step": 1270 }, { "epoch": 0.0, "learning_rate": 4.999983079627417e-05, "loss": 0.4547, "step": 1271 }, { "epoch": 0.0, "learning_rate": 4.999983050716096e-05, "loss": 0.8922, "step": 1272 }, { "epoch": 0.0, "learning_rate": 4.999983021780097e-05, "loss": 0.8002, "step": 1273 }, { "epoch": 0.0, "learning_rate": 4.999982992819419e-05, "loss": 1.1497, "step": 1274 }, { "epoch": 0.0, "learning_rate": 4.9999829638340625e-05, "loss": 0.9905, "step": 1275 }, { "epoch": 0.0, "learning_rate": 4.9999829348240264e-05, "loss": 1.1683, "step": 1276 }, { "epoch": 0.0, "learning_rate": 4.9999829057893124e-05, "loss": 1.0706, "step": 1277 }, { "epoch": 0.0, "learning_rate": 4.9999828767299187e-05, "loss": 1.1382, "step": 1278 }, { "epoch": 0.0, "learning_rate": 4.9999828476458464e-05, "loss": 1.2152, "step": 1279 }, { "epoch": 0.0, "learning_rate": 4.999982818537096e-05, "loss": 1.0906, "step": 1280 }, { "epoch": 0.0, "learning_rate": 4.999982789403666e-05, "loss": 1.0177, "step": 1281 }, { "epoch": 0.0, "learning_rate": 4.9999827602455574e-05, "loss": 1.2553, "step": 1282 }, { "epoch": 0.0, "learning_rate": 4.9999827310627706e-05, "loss": 0.9395, "step": 1283 }, { "epoch": 0.0, "learning_rate": 4.9999827018553046e-05, "loss": 1.0483, "step": 1284 }, { "epoch": 0.0, "learning_rate": 4.99998267262316e-05, "loss": 1.2686, "step": 1285 }, { "epoch": 0.0, "learning_rate": 4.9999826433663366e-05, "loss": 1.2582, "step": 1286 }, { "epoch": 0.0, "learning_rate": 4.999982614084834e-05, "loss": 1.1988, "step": 1287 }, { "epoch": 0.0, "learning_rate": 4.999982584778653e-05, "loss": 1.1987, "step": 1288 }, { "epoch": 0.0, "learning_rate": 4.999982555447793e-05, "loss": 0.9949, "step": 1289 }, { "epoch": 0.0, "learning_rate": 4.9999825260922554e-05, "loss": 1.3115, "step": 1290 }, { "epoch": 0.0, "learning_rate": 4.9999824967120373e-05, "loss": 1.182, "step": 1291 }, { "epoch": 0.0, "learning_rate": 4.9999824673071415e-05, "loss": 1.107, "step": 1292 }, { "epoch": 0.0, "learning_rate": 4.999982437877566e-05, "loss": 1.0447, "step": 1293 }, { "epoch": 0.0, "learning_rate": 4.9999824084233125e-05, "loss": 0.838, "step": 1294 }, { "epoch": 0.0, "learning_rate": 4.99998237894438e-05, "loss": 1.0267, "step": 1295 }, { "epoch": 0.0, "learning_rate": 4.9999823494407695e-05, "loss": 0.9289, "step": 1296 }, { "epoch": 0.0, "learning_rate": 4.999982319912479e-05, "loss": 1.1514, "step": 1297 }, { "epoch": 0.0, "learning_rate": 4.9999822903595106e-05, "loss": 1.3189, "step": 1298 }, { "epoch": 0.0, "learning_rate": 4.999982260781863e-05, "loss": 1.1155, "step": 1299 }, { "epoch": 0.0, "learning_rate": 4.9999822311795364e-05, "loss": 1.1044, "step": 1300 }, { "epoch": 0.0, "learning_rate": 4.999982201552531e-05, "loss": 1.2472, "step": 1301 }, { "epoch": 0.0, "learning_rate": 4.999982171900848e-05, "loss": 0.9781, "step": 1302 }, { "epoch": 0.0, "learning_rate": 4.999982142224485e-05, "loss": 0.716, "step": 1303 }, { "epoch": 0.0, "learning_rate": 4.999982112523444e-05, "loss": 0.5798, "step": 1304 }, { "epoch": 0.0, "learning_rate": 4.999982082797724e-05, "loss": 1.3287, "step": 1305 }, { "epoch": 0.0, "learning_rate": 4.999982053047324e-05, "loss": 1.0971, "step": 1306 }, { "epoch": 0.0, "learning_rate": 4.999982023272247e-05, "loss": 1.2485, "step": 1307 }, { "epoch": 0.0, "learning_rate": 4.99998199347249e-05, "loss": 0.9496, "step": 1308 }, { "epoch": 0.0, "learning_rate": 4.999981963648055e-05, "loss": 1.0331, "step": 1309 }, { "epoch": 0.0, "learning_rate": 4.999981933798941e-05, "loss": 1.045, "step": 1310 }, { "epoch": 0.0, "learning_rate": 4.999981903925148e-05, "loss": 0.9116, "step": 1311 }, { "epoch": 0.0, "learning_rate": 4.9999818740266766e-05, "loss": 1.256, "step": 1312 }, { "epoch": 0.0, "learning_rate": 4.999981844103526e-05, "loss": 1.1436, "step": 1313 }, { "epoch": 0.0, "learning_rate": 4.999981814155698e-05, "loss": 1.1256, "step": 1314 }, { "epoch": 0.0, "learning_rate": 4.9999817841831896e-05, "loss": 1.0737, "step": 1315 }, { "epoch": 0.0, "learning_rate": 4.999981754186003e-05, "loss": 1.2136, "step": 1316 }, { "epoch": 0.0, "learning_rate": 4.999981724164138e-05, "loss": 1.008, "step": 1317 }, { "epoch": 0.0, "learning_rate": 4.9999816941175935e-05, "loss": 0.9908, "step": 1318 }, { "epoch": 0.0, "learning_rate": 4.99998166404637e-05, "loss": 1.1998, "step": 1319 }, { "epoch": 0.0, "learning_rate": 4.999981633950469e-05, "loss": 1.0553, "step": 1320 }, { "epoch": 0.0, "learning_rate": 4.9999816038298885e-05, "loss": 1.2117, "step": 1321 }, { "epoch": 0.0, "learning_rate": 4.999981573684629e-05, "loss": 0.8267, "step": 1322 }, { "epoch": 0.0, "learning_rate": 4.999981543514691e-05, "loss": 1.5073, "step": 1323 }, { "epoch": 0.0, "learning_rate": 4.9999815133200745e-05, "loss": 1.2936, "step": 1324 }, { "epoch": 0.0, "learning_rate": 4.999981483100778e-05, "loss": 1.1269, "step": 1325 }, { "epoch": 0.0, "learning_rate": 4.999981452856805e-05, "loss": 1.1422, "step": 1326 }, { "epoch": 0.0, "learning_rate": 4.9999814225881515e-05, "loss": 1.0733, "step": 1327 }, { "epoch": 0.0, "learning_rate": 4.9999813922948205e-05, "loss": 1.1965, "step": 1328 }, { "epoch": 0.0, "learning_rate": 4.9999813619768096e-05, "loss": 1.5187, "step": 1329 }, { "epoch": 0.0, "learning_rate": 4.9999813316341196e-05, "loss": 1.0381, "step": 1330 }, { "epoch": 0.0, "learning_rate": 4.999981301266752e-05, "loss": 1.0388, "step": 1331 }, { "epoch": 0.0, "learning_rate": 4.9999812708747054e-05, "loss": 1.3295, "step": 1332 }, { "epoch": 0.0, "learning_rate": 4.999981240457979e-05, "loss": 1.1589, "step": 1333 }, { "epoch": 0.0, "learning_rate": 4.9999812100165754e-05, "loss": 1.2435, "step": 1334 }, { "epoch": 0.0, "learning_rate": 4.9999811795504916e-05, "loss": 1.0928, "step": 1335 }, { "epoch": 0.0, "learning_rate": 4.99998114905973e-05, "loss": 1.1308, "step": 1336 }, { "epoch": 0.0, "learning_rate": 4.9999811185442894e-05, "loss": 1.06, "step": 1337 }, { "epoch": 0.0, "learning_rate": 4.9999810880041695e-05, "loss": 1.6996, "step": 1338 }, { "epoch": 0.0, "learning_rate": 4.999981057439372e-05, "loss": 2.2645, "step": 1339 }, { "epoch": 0.0, "learning_rate": 4.999981026849895e-05, "loss": 2.2996, "step": 1340 }, { "epoch": 0.0, "learning_rate": 4.999980996235739e-05, "loss": 1.1277, "step": 1341 }, { "epoch": 0.0, "learning_rate": 4.9999809655969047e-05, "loss": 1.1166, "step": 1342 }, { "epoch": 0.0, "learning_rate": 4.999980934933392e-05, "loss": 1.1923, "step": 1343 }, { "epoch": 0.0, "learning_rate": 4.9999809042452e-05, "loss": 1.2633, "step": 1344 }, { "epoch": 0.0, "learning_rate": 4.999980873532329e-05, "loss": 0.8925, "step": 1345 }, { "epoch": 0.0, "learning_rate": 4.9999808427947795e-05, "loss": 0.9157, "step": 1346 }, { "epoch": 0.0, "learning_rate": 4.999980812032551e-05, "loss": 1.1102, "step": 1347 }, { "epoch": 0.0, "learning_rate": 4.999980781245645e-05, "loss": 1.0015, "step": 1348 }, { "epoch": 0.0, "learning_rate": 4.999980750434058e-05, "loss": 1.0523, "step": 1349 }, { "epoch": 0.0, "learning_rate": 4.9999807195977946e-05, "loss": 1.1134, "step": 1350 }, { "epoch": 0.0, "learning_rate": 4.9999806887368505e-05, "loss": 0.9844, "step": 1351 }, { "epoch": 0.0, "learning_rate": 4.9999806578512287e-05, "loss": 1.3819, "step": 1352 }, { "epoch": 0.0, "learning_rate": 4.9999806269409276e-05, "loss": 1.3509, "step": 1353 }, { "epoch": 0.0, "learning_rate": 4.999980596005949e-05, "loss": 0.937, "step": 1354 }, { "epoch": 0.0, "learning_rate": 4.999980565046291e-05, "loss": 1.0986, "step": 1355 }, { "epoch": 0.0, "learning_rate": 4.999980534061953e-05, "loss": 0.9307, "step": 1356 }, { "epoch": 0.0, "learning_rate": 4.9999805030529374e-05, "loss": 1.0425, "step": 1357 }, { "epoch": 0.0, "learning_rate": 4.9999804720192426e-05, "loss": 1.0921, "step": 1358 }, { "epoch": 0.0, "learning_rate": 4.99998044096087e-05, "loss": 1.4279, "step": 1359 }, { "epoch": 0.0, "learning_rate": 4.9999804098778176e-05, "loss": 1.0246, "step": 1360 }, { "epoch": 0.0, "learning_rate": 4.999980378770087e-05, "loss": 1.2811, "step": 1361 }, { "epoch": 0.0, "learning_rate": 4.9999803476376774e-05, "loss": 1.2623, "step": 1362 }, { "epoch": 0.0, "learning_rate": 4.999980316480589e-05, "loss": 1.1786, "step": 1363 }, { "epoch": 0.0, "learning_rate": 4.999980285298822e-05, "loss": 1.7484, "step": 1364 }, { "epoch": 0.0, "learning_rate": 4.999980254092376e-05, "loss": 1.5065, "step": 1365 }, { "epoch": 0.0, "learning_rate": 4.999980222861251e-05, "loss": 1.063, "step": 1366 }, { "epoch": 0.0, "learning_rate": 4.9999801916054475e-05, "loss": 1.1653, "step": 1367 }, { "epoch": 0.0, "learning_rate": 4.999980160324966e-05, "loss": 1.2328, "step": 1368 }, { "epoch": 0.0, "learning_rate": 4.999980129019805e-05, "loss": 1.1489, "step": 1369 }, { "epoch": 0.0, "learning_rate": 4.9999800976899654e-05, "loss": 1.5534, "step": 1370 }, { "epoch": 0.0, "learning_rate": 4.999980066335447e-05, "loss": 2.7763, "step": 1371 }, { "epoch": 0.0, "learning_rate": 4.99998003495625e-05, "loss": 2.337, "step": 1372 }, { "epoch": 0.0, "learning_rate": 4.999980003552375e-05, "loss": 2.4659, "step": 1373 }, { "epoch": 0.0, "learning_rate": 4.99997997212382e-05, "loss": 2.3945, "step": 1374 }, { "epoch": 0.0, "learning_rate": 4.9999799406705864e-05, "loss": 2.1941, "step": 1375 }, { "epoch": 0.0, "learning_rate": 4.9999799091926743e-05, "loss": 1.6053, "step": 1376 }, { "epoch": 0.0, "learning_rate": 4.999979877690084e-05, "loss": 1.0788, "step": 1377 }, { "epoch": 0.0, "learning_rate": 4.999979846162814e-05, "loss": 1.3246, "step": 1378 }, { "epoch": 0.0, "learning_rate": 4.999979814610865e-05, "loss": 1.4813, "step": 1379 }, { "epoch": 0.0, "learning_rate": 4.999979783034239e-05, "loss": 1.3126, "step": 1380 }, { "epoch": 0.0, "learning_rate": 4.999979751432933e-05, "loss": 1.178, "step": 1381 }, { "epoch": 0.0, "learning_rate": 4.999979719806949e-05, "loss": 1.5922, "step": 1382 }, { "epoch": 0.0, "learning_rate": 4.999979688156285e-05, "loss": 1.3641, "step": 1383 }, { "epoch": 0.0, "learning_rate": 4.9999796564809434e-05, "loss": 1.3358, "step": 1384 }, { "epoch": 0.0, "learning_rate": 4.999979624780923e-05, "loss": 0.9263, "step": 1385 }, { "epoch": 0.0, "learning_rate": 4.999979593056223e-05, "loss": 1.2803, "step": 1386 }, { "epoch": 0.0, "learning_rate": 4.999979561306844e-05, "loss": 0.8805, "step": 1387 }, { "epoch": 0.0, "learning_rate": 4.999979529532788e-05, "loss": 1.5963, "step": 1388 }, { "epoch": 0.0, "learning_rate": 4.999979497734052e-05, "loss": 1.3765, "step": 1389 }, { "epoch": 0.0, "learning_rate": 4.999979465910637e-05, "loss": 1.1817, "step": 1390 }, { "epoch": 0.0, "learning_rate": 4.999979434062544e-05, "loss": 1.3105, "step": 1391 }, { "epoch": 0.0, "learning_rate": 4.999979402189772e-05, "loss": 1.043, "step": 1392 }, { "epoch": 0.0, "learning_rate": 4.9999793702923216e-05, "loss": 1.0101, "step": 1393 }, { "epoch": 0.0, "learning_rate": 4.999979338370192e-05, "loss": 1.3187, "step": 1394 }, { "epoch": 0.0, "learning_rate": 4.999979306423384e-05, "loss": 1.1142, "step": 1395 }, { "epoch": 0.0, "learning_rate": 4.999979274451897e-05, "loss": 0.9806, "step": 1396 }, { "epoch": 0.0, "learning_rate": 4.9999792424557315e-05, "loss": 1.1195, "step": 1397 }, { "epoch": 0.0, "learning_rate": 4.999979210434886e-05, "loss": 1.2506, "step": 1398 }, { "epoch": 0.0, "learning_rate": 4.999979178389363e-05, "loss": 1.13, "step": 1399 }, { "epoch": 0.0, "learning_rate": 4.9999791463191616e-05, "loss": 1.2353, "step": 1400 }, { "epoch": 0.0, "learning_rate": 4.99997911422428e-05, "loss": 1.1788, "step": 1401 }, { "epoch": 0.0, "learning_rate": 4.999979082104721e-05, "loss": 1.0935, "step": 1402 }, { "epoch": 0.0, "learning_rate": 4.9999790499604835e-05, "loss": 1.0114, "step": 1403 }, { "epoch": 0.0, "learning_rate": 4.999979017791566e-05, "loss": 1.1398, "step": 1404 }, { "epoch": 0.0, "learning_rate": 4.999978985597971e-05, "loss": 1.2195, "step": 1405 }, { "epoch": 0.0, "learning_rate": 4.999978953379696e-05, "loss": 1.1132, "step": 1406 }, { "epoch": 0.0, "learning_rate": 4.9999789211367435e-05, "loss": 1.2819, "step": 1407 }, { "epoch": 0.0, "learning_rate": 4.9999788888691114e-05, "loss": 0.621, "step": 1408 }, { "epoch": 0.0, "learning_rate": 4.9999788565768e-05, "loss": 0.3403, "step": 1409 }, { "epoch": 0.0, "learning_rate": 4.999978824259811e-05, "loss": 0.3945, "step": 1410 }, { "epoch": 0.0, "learning_rate": 4.999978791918143e-05, "loss": 1.0303, "step": 1411 }, { "epoch": 0.0, "learning_rate": 4.9999787595517965e-05, "loss": 1.1808, "step": 1412 }, { "epoch": 0.0, "learning_rate": 4.9999787271607714e-05, "loss": 1.3936, "step": 1413 }, { "epoch": 0.0, "learning_rate": 4.9999786947450665e-05, "loss": 2.0301, "step": 1414 }, { "epoch": 0.0, "learning_rate": 4.999978662304684e-05, "loss": 1.1428, "step": 1415 }, { "epoch": 0.0, "learning_rate": 4.999978629839622e-05, "loss": 0.9583, "step": 1416 }, { "epoch": 0.0, "learning_rate": 4.9999785973498816e-05, "loss": 1.1777, "step": 1417 }, { "epoch": 0.0, "learning_rate": 4.999978564835462e-05, "loss": 1.103, "step": 1418 }, { "epoch": 0.0, "learning_rate": 4.999978532296364e-05, "loss": 1.017, "step": 1419 }, { "epoch": 0.0, "learning_rate": 4.999978499732587e-05, "loss": 1.0472, "step": 1420 }, { "epoch": 0.0, "learning_rate": 4.999978467144132e-05, "loss": 0.9334, "step": 1421 }, { "epoch": 0.0, "learning_rate": 4.999978434530998e-05, "loss": 1.0641, "step": 1422 }, { "epoch": 0.0, "learning_rate": 4.999978401893185e-05, "loss": 1.1676, "step": 1423 }, { "epoch": 0.0, "learning_rate": 4.999978369230693e-05, "loss": 0.9384, "step": 1424 }, { "epoch": 0.0, "learning_rate": 4.999978336543523e-05, "loss": 0.9575, "step": 1425 }, { "epoch": 0.0, "learning_rate": 4.999978303831674e-05, "loss": 1.1896, "step": 1426 }, { "epoch": 0.0, "learning_rate": 4.999978271095146e-05, "loss": 1.1317, "step": 1427 }, { "epoch": 0.0, "learning_rate": 4.9999782383339396e-05, "loss": 1.0979, "step": 1428 }, { "epoch": 0.0, "learning_rate": 4.999978205548054e-05, "loss": 1.0988, "step": 1429 }, { "epoch": 0.0, "learning_rate": 4.99997817273749e-05, "loss": 1.1114, "step": 1430 }, { "epoch": 0.0, "learning_rate": 4.999978139902247e-05, "loss": 2.2097, "step": 1431 }, { "epoch": 0.0, "learning_rate": 4.999978107042326e-05, "loss": 1.0582, "step": 1432 }, { "epoch": 0.0, "learning_rate": 4.9999780741577265e-05, "loss": 0.8381, "step": 1433 }, { "epoch": 0.0, "learning_rate": 4.999978041248447e-05, "loss": 0.8595, "step": 1434 }, { "epoch": 0.0, "learning_rate": 4.999978008314489e-05, "loss": 0.8939, "step": 1435 }, { "epoch": 0.0, "learning_rate": 4.999977975355853e-05, "loss": 1.2282, "step": 1436 }, { "epoch": 0.0, "learning_rate": 4.999977942372538e-05, "loss": 1.3273, "step": 1437 }, { "epoch": 0.0, "learning_rate": 4.999977909364544e-05, "loss": 1.1084, "step": 1438 }, { "epoch": 0.0, "learning_rate": 4.999977876331872e-05, "loss": 1.0852, "step": 1439 }, { "epoch": 0.0, "learning_rate": 4.9999778432745204e-05, "loss": 1.3899, "step": 1440 }, { "epoch": 0.0, "learning_rate": 4.99997781019249e-05, "loss": 1.2291, "step": 1441 }, { "epoch": 0.0, "learning_rate": 4.9999777770857816e-05, "loss": 1.1377, "step": 1442 }, { "epoch": 0.0, "learning_rate": 4.9999777439543935e-05, "loss": 1.0589, "step": 1443 }, { "epoch": 0.0, "learning_rate": 4.9999777107983276e-05, "loss": 1.0669, "step": 1444 }, { "epoch": 0.0, "learning_rate": 4.999977677617582e-05, "loss": 0.9086, "step": 1445 }, { "epoch": 0.0, "learning_rate": 4.999977644412159e-05, "loss": 1.2046, "step": 1446 }, { "epoch": 0.0, "learning_rate": 4.999977611182056e-05, "loss": 1.195, "step": 1447 }, { "epoch": 0.0, "learning_rate": 4.999977577927275e-05, "loss": 1.3642, "step": 1448 }, { "epoch": 0.0, "learning_rate": 4.999977544647816e-05, "loss": 1.078, "step": 1449 }, { "epoch": 0.0, "learning_rate": 4.999977511343677e-05, "loss": 1.2531, "step": 1450 }, { "epoch": 0.0, "learning_rate": 4.99997747801486e-05, "loss": 1.2787, "step": 1451 }, { "epoch": 0.0, "learning_rate": 4.9999774446613637e-05, "loss": 0.9946, "step": 1452 }, { "epoch": 0.0, "learning_rate": 4.999977411283189e-05, "loss": 1.0959, "step": 1453 }, { "epoch": 0.0, "learning_rate": 4.9999773778803354e-05, "loss": 1.214, "step": 1454 }, { "epoch": 0.0, "learning_rate": 4.999977344452803e-05, "loss": 1.1483, "step": 1455 }, { "epoch": 0.0, "learning_rate": 4.999977311000592e-05, "loss": 1.084, "step": 1456 }, { "epoch": 0.0, "learning_rate": 4.999977277523703e-05, "loss": 0.9704, "step": 1457 }, { "epoch": 0.0, "learning_rate": 4.9999772440221344e-05, "loss": 0.7158, "step": 1458 }, { "epoch": 0.0, "learning_rate": 4.999977210495887e-05, "loss": 1.0292, "step": 1459 }, { "epoch": 0.0, "learning_rate": 4.999977176944961e-05, "loss": 1.1296, "step": 1460 }, { "epoch": 0.0, "learning_rate": 4.9999771433693564e-05, "loss": 1.0778, "step": 1461 }, { "epoch": 0.0, "learning_rate": 4.999977109769074e-05, "loss": 3.3233, "step": 1462 }, { "epoch": 0.0, "learning_rate": 4.9999770761441115e-05, "loss": 3.9099, "step": 1463 }, { "epoch": 0.0, "learning_rate": 4.9999770424944707e-05, "loss": 1.3, "step": 1464 }, { "epoch": 0.0, "learning_rate": 4.9999770088201514e-05, "loss": 0.973, "step": 1465 }, { "epoch": 0.0, "learning_rate": 4.9999769751211536e-05, "loss": 1.1086, "step": 1466 }, { "epoch": 0.0, "learning_rate": 4.999976941397476e-05, "loss": 0.8508, "step": 1467 }, { "epoch": 0.0, "learning_rate": 4.999976907649121e-05, "loss": 1.0619, "step": 1468 }, { "epoch": 0.0, "learning_rate": 4.999976873876087e-05, "loss": 1.0714, "step": 1469 }, { "epoch": 0.0, "learning_rate": 4.999976840078374e-05, "loss": 2.7562, "step": 1470 }, { "epoch": 0.0, "learning_rate": 4.999976806255982e-05, "loss": 0.9887, "step": 1471 }, { "epoch": 0.0, "learning_rate": 4.999976772408912e-05, "loss": 0.974, "step": 1472 }, { "epoch": 0.0, "learning_rate": 4.9999767385371624e-05, "loss": 1.245, "step": 1473 }, { "epoch": 0.0, "learning_rate": 4.999976704640734e-05, "loss": 1.5144, "step": 1474 }, { "epoch": 0.0, "learning_rate": 4.999976670719628e-05, "loss": 1.0975, "step": 1475 }, { "epoch": 0.0, "learning_rate": 4.999976636773843e-05, "loss": 1.8456, "step": 1476 }, { "epoch": 0.0, "learning_rate": 4.999976602803379e-05, "loss": 1.646, "step": 1477 }, { "epoch": 0.0, "learning_rate": 4.9999765688082356e-05, "loss": 1.3266, "step": 1478 }, { "epoch": 0.0, "learning_rate": 4.999976534788414e-05, "loss": 1.3988, "step": 1479 }, { "epoch": 0.0, "learning_rate": 4.9999765007439144e-05, "loss": 1.3974, "step": 1480 }, { "epoch": 0.0, "learning_rate": 4.999976466674735e-05, "loss": 1.1976, "step": 1481 }, { "epoch": 0.0, "learning_rate": 4.999976432580878e-05, "loss": 1.5675, "step": 1482 }, { "epoch": 0.0, "learning_rate": 4.9999763984623415e-05, "loss": 1.0349, "step": 1483 }, { "epoch": 0.0, "learning_rate": 4.9999763643191264e-05, "loss": 1.3132, "step": 1484 }, { "epoch": 0.0, "learning_rate": 4.999976330151233e-05, "loss": 1.1919, "step": 1485 }, { "epoch": 0.0, "learning_rate": 4.99997629595866e-05, "loss": 0.9964, "step": 1486 }, { "epoch": 0.0, "learning_rate": 4.999976261741409e-05, "loss": 1.2105, "step": 1487 }, { "epoch": 0.0, "learning_rate": 4.9999762274994794e-05, "loss": 0.8005, "step": 1488 }, { "epoch": 0.0, "learning_rate": 4.9999761932328706e-05, "loss": 1.0002, "step": 1489 }, { "epoch": 0.0, "learning_rate": 4.9999761589415834e-05, "loss": 1.3317, "step": 1490 }, { "epoch": 0.0, "learning_rate": 4.999976124625617e-05, "loss": 1.5958, "step": 1491 }, { "epoch": 0.0, "learning_rate": 4.999976090284972e-05, "loss": 0.9418, "step": 1492 }, { "epoch": 0.0, "learning_rate": 4.9999760559196494e-05, "loss": 1.1749, "step": 1493 }, { "epoch": 0.0, "learning_rate": 4.999976021529647e-05, "loss": 1.041, "step": 1494 }, { "epoch": 0.0, "learning_rate": 4.9999759871149666e-05, "loss": 1.3281, "step": 1495 }, { "epoch": 0.0, "learning_rate": 4.9999759526756064e-05, "loss": 0.9078, "step": 1496 }, { "epoch": 0.0, "learning_rate": 4.9999759182115685e-05, "loss": 1.1638, "step": 1497 }, { "epoch": 0.0, "learning_rate": 4.9999758837228514e-05, "loss": 1.2198, "step": 1498 }, { "epoch": 0.0, "learning_rate": 4.999975849209456e-05, "loss": 1.4662, "step": 1499 }, { "epoch": 0.0, "learning_rate": 4.999975814671382e-05, "loss": 1.384, "step": 1500 }, { "epoch": 0.0, "eval_loss": 1.1025360822677612, "eval_runtime": 85.1844, "eval_samples_per_second": 16.259, "eval_steps_per_second": 4.074, "step": 1500 }, { "epoch": 0.0, "learning_rate": 4.9999757801086286e-05, "loss": 1.0703, "step": 1501 }, { "epoch": 0.0, "learning_rate": 4.999975745521196e-05, "loss": 1.1068, "step": 1502 }, { "epoch": 0.0, "learning_rate": 4.999975710909086e-05, "loss": 1.5193, "step": 1503 }, { "epoch": 0.0, "learning_rate": 4.999975676272297e-05, "loss": 1.0456, "step": 1504 }, { "epoch": 0.0, "learning_rate": 4.999975641610829e-05, "loss": 1.1337, "step": 1505 }, { "epoch": 0.0, "learning_rate": 4.999975606924682e-05, "loss": 1.0031, "step": 1506 }, { "epoch": 0.0, "learning_rate": 4.999975572213856e-05, "loss": 1.2249, "step": 1507 }, { "epoch": 0.0, "learning_rate": 4.999975537478353e-05, "loss": 1.0518, "step": 1508 }, { "epoch": 0.0, "learning_rate": 4.99997550271817e-05, "loss": 1.1407, "step": 1509 }, { "epoch": 0.0, "learning_rate": 4.999975467933309e-05, "loss": 1.0663, "step": 1510 }, { "epoch": 0.0, "learning_rate": 4.999975433123768e-05, "loss": 1.0735, "step": 1511 }, { "epoch": 0.0, "learning_rate": 4.9999753982895494e-05, "loss": 1.2828, "step": 1512 }, { "epoch": 0.0, "learning_rate": 4.9999753634306525e-05, "loss": 1.3558, "step": 1513 }, { "epoch": 0.0, "learning_rate": 4.999975328547076e-05, "loss": 1.0236, "step": 1514 }, { "epoch": 0.0, "learning_rate": 4.9999752936388205e-05, "loss": 0.9801, "step": 1515 }, { "epoch": 0.0, "learning_rate": 4.999975258705887e-05, "loss": 1.0956, "step": 1516 }, { "epoch": 0.0, "learning_rate": 4.999975223748275e-05, "loss": 1.2429, "step": 1517 }, { "epoch": 0.0, "learning_rate": 4.9999751887659833e-05, "loss": 1.0751, "step": 1518 }, { "epoch": 0.0, "learning_rate": 4.9999751537590136e-05, "loss": 1.5672, "step": 1519 }, { "epoch": 0.0, "learning_rate": 4.999975118727365e-05, "loss": 1.2634, "step": 1520 }, { "epoch": 0.0, "learning_rate": 4.999975083671038e-05, "loss": 1.2086, "step": 1521 }, { "epoch": 0.0, "learning_rate": 4.999975048590032e-05, "loss": 1.0235, "step": 1522 }, { "epoch": 0.0, "learning_rate": 4.9999750134843476e-05, "loss": 1.1494, "step": 1523 }, { "epoch": 0.0, "learning_rate": 4.999974978353984e-05, "loss": 1.0878, "step": 1524 }, { "epoch": 0.0, "learning_rate": 4.999974943198942e-05, "loss": 1.1574, "step": 1525 }, { "epoch": 0.0, "learning_rate": 4.9999749080192217e-05, "loss": 1.267, "step": 1526 }, { "epoch": 0.0, "learning_rate": 4.999974872814822e-05, "loss": 1.0234, "step": 1527 }, { "epoch": 0.0, "learning_rate": 4.999974837585744e-05, "loss": 1.2157, "step": 1528 }, { "epoch": 0.0, "learning_rate": 4.999974802331987e-05, "loss": 1.3503, "step": 1529 }, { "epoch": 0.0, "learning_rate": 4.999974767053552e-05, "loss": 1.2035, "step": 1530 }, { "epoch": 0.0, "learning_rate": 4.9999747317504375e-05, "loss": 1.1835, "step": 1531 }, { "epoch": 0.0, "learning_rate": 4.999974696422645e-05, "loss": 1.3205, "step": 1532 }, { "epoch": 0.0, "learning_rate": 4.999974661070174e-05, "loss": 1.2915, "step": 1533 }, { "epoch": 0.0, "learning_rate": 4.999974625693023e-05, "loss": 1.0228, "step": 1534 }, { "epoch": 0.0, "learning_rate": 4.9999745902911946e-05, "loss": 1.0057, "step": 1535 }, { "epoch": 0.0, "learning_rate": 4.999974554864687e-05, "loss": 0.9307, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.9999745194135e-05, "loss": 1.2788, "step": 1537 }, { "epoch": 0.0, "learning_rate": 4.999974483937635e-05, "loss": 1.3145, "step": 1538 }, { "epoch": 0.0, "learning_rate": 4.9999744484370914e-05, "loss": 1.4141, "step": 1539 }, { "epoch": 0.0, "learning_rate": 4.999974412911869e-05, "loss": 0.8175, "step": 1540 }, { "epoch": 0.0, "learning_rate": 4.999974377361968e-05, "loss": 1.0975, "step": 1541 }, { "epoch": 0.0, "learning_rate": 4.999974341787388e-05, "loss": 1.3513, "step": 1542 }, { "epoch": 0.0, "learning_rate": 4.999974306188129e-05, "loss": 1.0626, "step": 1543 }, { "epoch": 0.0, "learning_rate": 4.999974270564192e-05, "loss": 1.0888, "step": 1544 }, { "epoch": 0.0, "learning_rate": 4.9999742349155766e-05, "loss": 1.0065, "step": 1545 }, { "epoch": 0.0, "learning_rate": 4.999974199242282e-05, "loss": 1.0015, "step": 1546 }, { "epoch": 0.0, "learning_rate": 4.999974163544309e-05, "loss": 1.1223, "step": 1547 }, { "epoch": 0.0, "learning_rate": 4.999974127821656e-05, "loss": 1.1169, "step": 1548 }, { "epoch": 0.0, "learning_rate": 4.9999740920743256e-05, "loss": 1.3053, "step": 1549 }, { "epoch": 0.0, "learning_rate": 4.999974056302317e-05, "loss": 1.1333, "step": 1550 }, { "epoch": 0.0, "learning_rate": 4.9999740205056286e-05, "loss": 0.7307, "step": 1551 }, { "epoch": 0.0, "learning_rate": 4.999973984684262e-05, "loss": 1.3703, "step": 1552 }, { "epoch": 0.0, "learning_rate": 4.999973948838216e-05, "loss": 1.4943, "step": 1553 }, { "epoch": 0.0, "learning_rate": 4.999973912967492e-05, "loss": 1.1022, "step": 1554 }, { "epoch": 0.0, "learning_rate": 4.9999738770720895e-05, "loss": 0.9436, "step": 1555 }, { "epoch": 0.0, "learning_rate": 4.999973841152008e-05, "loss": 1.1686, "step": 1556 }, { "epoch": 0.0, "learning_rate": 4.9999738052072474e-05, "loss": 1.0699, "step": 1557 }, { "epoch": 0.0, "learning_rate": 4.999973769237809e-05, "loss": 1.023, "step": 1558 }, { "epoch": 0.0, "learning_rate": 4.9999737332436914e-05, "loss": 1.1993, "step": 1559 }, { "epoch": 0.0, "learning_rate": 4.999973697224895e-05, "loss": 0.9739, "step": 1560 }, { "epoch": 0.0, "learning_rate": 4.9999736611814195e-05, "loss": 1.2015, "step": 1561 }, { "epoch": 0.0, "learning_rate": 4.999973625113266e-05, "loss": 1.3037, "step": 1562 }, { "epoch": 0.0, "learning_rate": 4.9999735890204344e-05, "loss": 1.1989, "step": 1563 }, { "epoch": 0.0, "learning_rate": 4.999973552902923e-05, "loss": 0.9803, "step": 1564 }, { "epoch": 0.0, "learning_rate": 4.999973516760733e-05, "loss": 1.1173, "step": 1565 }, { "epoch": 0.0, "learning_rate": 4.999973480593865e-05, "loss": 0.8269, "step": 1566 }, { "epoch": 0.0, "learning_rate": 4.999973444402318e-05, "loss": 1.2442, "step": 1567 }, { "epoch": 0.0, "learning_rate": 4.9999734081860925e-05, "loss": 1.1248, "step": 1568 }, { "epoch": 0.0, "learning_rate": 4.999973371945188e-05, "loss": 1.2366, "step": 1569 }, { "epoch": 0.0, "learning_rate": 4.9999733356796046e-05, "loss": 1.3797, "step": 1570 }, { "epoch": 0.0, "learning_rate": 4.9999732993893426e-05, "loss": 1.4179, "step": 1571 }, { "epoch": 0.0, "learning_rate": 4.999973263074402e-05, "loss": 1.1726, "step": 1572 }, { "epoch": 0.0, "learning_rate": 4.999973226734783e-05, "loss": 0.4636, "step": 1573 }, { "epoch": 0.0, "learning_rate": 4.999973190370485e-05, "loss": 0.8854, "step": 1574 }, { "epoch": 0.0, "learning_rate": 4.9999731539815086e-05, "loss": 1.2157, "step": 1575 }, { "epoch": 0.0, "learning_rate": 4.9999731175678536e-05, "loss": 1.6862, "step": 1576 }, { "epoch": 0.0, "learning_rate": 4.99997308112952e-05, "loss": 0.8838, "step": 1577 }, { "epoch": 0.0, "learning_rate": 4.9999730446665075e-05, "loss": 1.1561, "step": 1578 }, { "epoch": 0.0, "learning_rate": 4.999973008178816e-05, "loss": 0.9012, "step": 1579 }, { "epoch": 0.0, "learning_rate": 4.999972971666446e-05, "loss": 1.02, "step": 1580 }, { "epoch": 0.0, "learning_rate": 4.999972935129398e-05, "loss": 1.1196, "step": 1581 }, { "epoch": 0.0, "learning_rate": 4.99997289856767e-05, "loss": 1.2262, "step": 1582 }, { "epoch": 0.0, "learning_rate": 4.9999728619812644e-05, "loss": 1.259, "step": 1583 }, { "epoch": 0.0, "learning_rate": 4.9999728253701796e-05, "loss": 1.1746, "step": 1584 }, { "epoch": 0.0, "learning_rate": 4.999972788734416e-05, "loss": 1.0513, "step": 1585 }, { "epoch": 0.0, "learning_rate": 4.9999727520739744e-05, "loss": 1.0383, "step": 1586 }, { "epoch": 0.0, "learning_rate": 4.9999727153888535e-05, "loss": 1.0011, "step": 1587 }, { "epoch": 0.0, "learning_rate": 4.999972678679055e-05, "loss": 1.0647, "step": 1588 }, { "epoch": 0.0, "learning_rate": 4.999972641944576e-05, "loss": 0.817, "step": 1589 }, { "epoch": 0.0, "learning_rate": 4.99997260518542e-05, "loss": 1.0495, "step": 1590 }, { "epoch": 0.0, "learning_rate": 4.999972568401584e-05, "loss": 1.1049, "step": 1591 }, { "epoch": 0.0, "learning_rate": 4.99997253159307e-05, "loss": 0.9831, "step": 1592 }, { "epoch": 0.0, "learning_rate": 4.999972494759878e-05, "loss": 0.9951, "step": 1593 }, { "epoch": 0.0, "learning_rate": 4.9999724579020055e-05, "loss": 0.8695, "step": 1594 }, { "epoch": 0.0, "learning_rate": 4.999972421019456e-05, "loss": 1.1208, "step": 1595 }, { "epoch": 0.0, "learning_rate": 4.999972384112227e-05, "loss": 1.3532, "step": 1596 }, { "epoch": 0.0, "learning_rate": 4.99997234718032e-05, "loss": 1.1476, "step": 1597 }, { "epoch": 0.0, "learning_rate": 4.999972310223734e-05, "loss": 1.2644, "step": 1598 }, { "epoch": 0.0, "learning_rate": 4.999972273242469e-05, "loss": 1.1528, "step": 1599 }, { "epoch": 0.0, "learning_rate": 4.999972236236525e-05, "loss": 1.3417, "step": 1600 }, { "epoch": 0.0, "learning_rate": 4.9999721992059034e-05, "loss": 1.1045, "step": 1601 }, { "epoch": 0.0, "learning_rate": 4.9999721621506026e-05, "loss": 1.084, "step": 1602 }, { "epoch": 0.0, "learning_rate": 4.9999721250706234e-05, "loss": 1.1412, "step": 1603 }, { "epoch": 0.0, "learning_rate": 4.999972087965965e-05, "loss": 0.6463, "step": 1604 }, { "epoch": 0.0, "learning_rate": 4.999972050836628e-05, "loss": 0.5842, "step": 1605 }, { "epoch": 0.0, "learning_rate": 4.999972013682612e-05, "loss": 0.4859, "step": 1606 }, { "epoch": 0.0, "learning_rate": 4.999971976503919e-05, "loss": 1.1645, "step": 1607 }, { "epoch": 0.0, "learning_rate": 4.999971939300546e-05, "loss": 0.9786, "step": 1608 }, { "epoch": 0.0, "learning_rate": 4.9999719020724945e-05, "loss": 1.294, "step": 1609 }, { "epoch": 0.0, "learning_rate": 4.999971864819764e-05, "loss": 1.2126, "step": 1610 }, { "epoch": 0.0, "learning_rate": 4.9999718275423555e-05, "loss": 1.1962, "step": 1611 }, { "epoch": 0.0, "learning_rate": 4.999971790240268e-05, "loss": 1.073, "step": 1612 }, { "epoch": 0.0, "learning_rate": 4.999971752913501e-05, "loss": 1.1562, "step": 1613 }, { "epoch": 0.0, "learning_rate": 4.999971715562057e-05, "loss": 1.1068, "step": 1614 }, { "epoch": 0.0, "learning_rate": 4.999971678185933e-05, "loss": 1.1081, "step": 1615 }, { "epoch": 0.0, "learning_rate": 4.999971640785131e-05, "loss": 1.0775, "step": 1616 }, { "epoch": 0.0, "learning_rate": 4.99997160335965e-05, "loss": 1.083, "step": 1617 }, { "epoch": 0.0, "learning_rate": 4.9999715659094906e-05, "loss": 1.0268, "step": 1618 }, { "epoch": 0.0, "learning_rate": 4.9999715284346524e-05, "loss": 0.9439, "step": 1619 }, { "epoch": 0.0, "learning_rate": 4.999971490935136e-05, "loss": 0.7094, "step": 1620 }, { "epoch": 0.0, "learning_rate": 4.9999714534109405e-05, "loss": 1.4053, "step": 1621 }, { "epoch": 0.0, "learning_rate": 4.999971415862066e-05, "loss": 0.924, "step": 1622 }, { "epoch": 0.0, "learning_rate": 4.9999713782885134e-05, "loss": 1.1412, "step": 1623 }, { "epoch": 0.0, "learning_rate": 4.9999713406902814e-05, "loss": 1.0323, "step": 1624 }, { "epoch": 0.0, "learning_rate": 4.9999713030673717e-05, "loss": 1.2048, "step": 1625 }, { "epoch": 0.0, "learning_rate": 4.999971265419783e-05, "loss": 0.7828, "step": 1626 }, { "epoch": 0.0, "learning_rate": 4.9999712277475154e-05, "loss": 0.9369, "step": 1627 }, { "epoch": 0.0, "learning_rate": 4.9999711900505695e-05, "loss": 1.1362, "step": 1628 }, { "epoch": 0.0, "learning_rate": 4.9999711523289445e-05, "loss": 1.134, "step": 1629 }, { "epoch": 0.0, "learning_rate": 4.999971114582641e-05, "loss": 1.2234, "step": 1630 }, { "epoch": 0.0, "learning_rate": 4.999971076811659e-05, "loss": 1.2499, "step": 1631 }, { "epoch": 0.0, "learning_rate": 4.999971039015998e-05, "loss": 1.2591, "step": 1632 }, { "epoch": 0.0, "learning_rate": 4.999971001195659e-05, "loss": 1.4315, "step": 1633 }, { "epoch": 0.0, "learning_rate": 4.999970963350641e-05, "loss": 1.0759, "step": 1634 }, { "epoch": 0.0, "learning_rate": 4.999970925480944e-05, "loss": 1.169, "step": 1635 }, { "epoch": 0.0, "learning_rate": 4.999970887586568e-05, "loss": 1.393, "step": 1636 }, { "epoch": 0.0, "learning_rate": 4.999970849667515e-05, "loss": 1.3423, "step": 1637 }, { "epoch": 0.0, "learning_rate": 4.9999708117237814e-05, "loss": 1.0546, "step": 1638 }, { "epoch": 0.0, "learning_rate": 4.99997077375537e-05, "loss": 0.8396, "step": 1639 }, { "epoch": 0.0, "learning_rate": 4.999970735762281e-05, "loss": 1.1582, "step": 1640 }, { "epoch": 0.0, "learning_rate": 4.999970697744511e-05, "loss": 1.021, "step": 1641 }, { "epoch": 0.0, "learning_rate": 4.999970659702064e-05, "loss": 1.0423, "step": 1642 }, { "epoch": 0.0, "learning_rate": 4.999970621634938e-05, "loss": 0.8419, "step": 1643 }, { "epoch": 0.0, "learning_rate": 4.999970583543133e-05, "loss": 1.0889, "step": 1644 }, { "epoch": 0.0, "learning_rate": 4.9999705454266496e-05, "loss": 1.2187, "step": 1645 }, { "epoch": 0.0, "learning_rate": 4.999970507285488e-05, "loss": 1.3234, "step": 1646 }, { "epoch": 0.0, "learning_rate": 4.999970469119647e-05, "loss": 1.1936, "step": 1647 }, { "epoch": 0.0, "learning_rate": 4.999970430929128e-05, "loss": 1.1664, "step": 1648 }, { "epoch": 0.0, "learning_rate": 4.9999703927139296e-05, "loss": 0.8755, "step": 1649 }, { "epoch": 0.0, "learning_rate": 4.999970354474053e-05, "loss": 0.4793, "step": 1650 }, { "epoch": 0.0, "learning_rate": 4.999970316209498e-05, "loss": 0.4663, "step": 1651 }, { "epoch": 0.0, "learning_rate": 4.999970277920264e-05, "loss": 1.0177, "step": 1652 }, { "epoch": 0.0, "learning_rate": 4.999970239606352e-05, "loss": 0.8776, "step": 1653 }, { "epoch": 0.0, "learning_rate": 4.999970201267761e-05, "loss": 1.1567, "step": 1654 }, { "epoch": 0.0, "learning_rate": 4.99997016290449e-05, "loss": 1.1212, "step": 1655 }, { "epoch": 0.0, "learning_rate": 4.999970124516542e-05, "loss": 1.0755, "step": 1656 }, { "epoch": 0.0, "learning_rate": 4.999970086103915e-05, "loss": 1.2144, "step": 1657 }, { "epoch": 0.0, "learning_rate": 4.999970047666609e-05, "loss": 1.1866, "step": 1658 }, { "epoch": 0.0, "learning_rate": 4.9999700092046245e-05, "loss": 0.9839, "step": 1659 }, { "epoch": 0.0, "learning_rate": 4.9999699707179614e-05, "loss": 1.3086, "step": 1660 }, { "epoch": 0.0, "learning_rate": 4.99996993220662e-05, "loss": 1.105, "step": 1661 }, { "epoch": 0.0, "learning_rate": 4.999969893670599e-05, "loss": 1.253, "step": 1662 }, { "epoch": 0.0, "learning_rate": 4.9999698551099e-05, "loss": 1.546, "step": 1663 }, { "epoch": 0.0, "learning_rate": 4.999969816524522e-05, "loss": 1.0515, "step": 1664 }, { "epoch": 0.0, "learning_rate": 4.999969777914466e-05, "loss": 1.2819, "step": 1665 }, { "epoch": 0.0, "learning_rate": 4.9999697392797306e-05, "loss": 0.9853, "step": 1666 }, { "epoch": 0.0, "learning_rate": 4.9999697006203175e-05, "loss": 1.0811, "step": 1667 }, { "epoch": 0.0, "learning_rate": 4.9999696619362245e-05, "loss": 1.0727, "step": 1668 }, { "epoch": 0.0, "learning_rate": 4.999969623227454e-05, "loss": 1.1194, "step": 1669 }, { "epoch": 0.0, "learning_rate": 4.9999695844940045e-05, "loss": 0.961, "step": 1670 }, { "epoch": 0.0, "learning_rate": 4.999969545735876e-05, "loss": 1.0931, "step": 1671 }, { "epoch": 0.0, "learning_rate": 4.999969506953069e-05, "loss": 1.2467, "step": 1672 }, { "epoch": 0.0, "learning_rate": 4.999969468145583e-05, "loss": 1.2632, "step": 1673 }, { "epoch": 0.0, "learning_rate": 4.9999694293134195e-05, "loss": 1.3062, "step": 1674 }, { "epoch": 0.0, "learning_rate": 4.9999693904565765e-05, "loss": 1.0255, "step": 1675 }, { "epoch": 0.0, "learning_rate": 4.999969351575055e-05, "loss": 0.4736, "step": 1676 }, { "epoch": 0.0, "learning_rate": 4.999969312668855e-05, "loss": 0.4632, "step": 1677 }, { "epoch": 0.0, "learning_rate": 4.999969273737976e-05, "loss": 1.1215, "step": 1678 }, { "epoch": 0.0, "learning_rate": 4.9999692347824186e-05, "loss": 1.2863, "step": 1679 }, { "epoch": 0.0, "learning_rate": 4.9999691958021826e-05, "loss": 1.2319, "step": 1680 }, { "epoch": 0.0, "learning_rate": 4.999969156797268e-05, "loss": 1.1239, "step": 1681 }, { "epoch": 0.0, "learning_rate": 4.9999691177676745e-05, "loss": 1.1725, "step": 1682 }, { "epoch": 0.0, "learning_rate": 4.9999690787134024e-05, "loss": 1.3194, "step": 1683 }, { "epoch": 0.0, "learning_rate": 4.9999690396344525e-05, "loss": 1.767, "step": 1684 }, { "epoch": 0.0, "learning_rate": 4.999969000530823e-05, "loss": 1.1727, "step": 1685 }, { "epoch": 0.0, "learning_rate": 4.9999689614025146e-05, "loss": 0.942, "step": 1686 }, { "epoch": 0.0, "learning_rate": 4.9999689222495286e-05, "loss": 0.8949, "step": 1687 }, { "epoch": 0.0, "learning_rate": 4.999968883071863e-05, "loss": 1.0356, "step": 1688 }, { "epoch": 0.0, "learning_rate": 4.999968843869519e-05, "loss": 1.1545, "step": 1689 }, { "epoch": 0.0, "learning_rate": 4.999968804642496e-05, "loss": 1.0806, "step": 1690 }, { "epoch": 0.0, "learning_rate": 4.999968765390796e-05, "loss": 1.6976, "step": 1691 }, { "epoch": 0.0, "learning_rate": 4.999968726114416e-05, "loss": 1.4155, "step": 1692 }, { "epoch": 0.0, "learning_rate": 4.999968686813358e-05, "loss": 1.2434, "step": 1693 }, { "epoch": 0.0, "learning_rate": 4.9999686474876205e-05, "loss": 1.0439, "step": 1694 }, { "epoch": 0.0, "learning_rate": 4.9999686081372046e-05, "loss": 0.9256, "step": 1695 }, { "epoch": 0.0, "learning_rate": 4.99996856876211e-05, "loss": 1.2005, "step": 1696 }, { "epoch": 0.0, "learning_rate": 4.9999685293623376e-05, "loss": 1.0603, "step": 1697 }, { "epoch": 0.0, "learning_rate": 4.999968489937886e-05, "loss": 1.3092, "step": 1698 }, { "epoch": 0.0, "learning_rate": 4.999968450488756e-05, "loss": 1.3336, "step": 1699 }, { "epoch": 0.0, "learning_rate": 4.999968411014947e-05, "loss": 1.0654, "step": 1700 }, { "epoch": 0.0, "learning_rate": 4.999968371516459e-05, "loss": 1.1481, "step": 1701 }, { "epoch": 0.0, "learning_rate": 4.999968331993293e-05, "loss": 0.9474, "step": 1702 }, { "epoch": 0.0, "learning_rate": 4.999968292445448e-05, "loss": 1.5143, "step": 1703 }, { "epoch": 0.0, "learning_rate": 4.9999682528729254e-05, "loss": 2.0976, "step": 1704 }, { "epoch": 0.0, "learning_rate": 4.999968213275723e-05, "loss": 1.2388, "step": 1705 }, { "epoch": 0.0, "learning_rate": 4.999968173653843e-05, "loss": 1.3075, "step": 1706 }, { "epoch": 0.0, "learning_rate": 4.9999681340072835e-05, "loss": 1.5414, "step": 1707 }, { "epoch": 0.0, "learning_rate": 4.999968094336045e-05, "loss": 0.8223, "step": 1708 }, { "epoch": 0.0, "learning_rate": 4.999968054640128e-05, "loss": 1.1105, "step": 1709 }, { "epoch": 0.0, "learning_rate": 4.999968014919534e-05, "loss": 1.3593, "step": 1710 }, { "epoch": 0.0, "learning_rate": 4.99996797517426e-05, "loss": 1.2691, "step": 1711 }, { "epoch": 0.0, "learning_rate": 4.999967935404307e-05, "loss": 0.9297, "step": 1712 }, { "epoch": 0.0, "learning_rate": 4.999967895609676e-05, "loss": 1.17, "step": 1713 }, { "epoch": 0.0, "learning_rate": 4.999967855790366e-05, "loss": 0.9769, "step": 1714 }, { "epoch": 0.0, "learning_rate": 4.999967815946378e-05, "loss": 0.8034, "step": 1715 }, { "epoch": 0.0, "learning_rate": 4.9999677760777114e-05, "loss": 0.9674, "step": 1716 }, { "epoch": 0.0, "learning_rate": 4.999967736184366e-05, "loss": 0.9551, "step": 1717 }, { "epoch": 0.0, "learning_rate": 4.999967696266341e-05, "loss": 1.3977, "step": 1718 }, { "epoch": 0.0, "learning_rate": 4.9999676563236384e-05, "loss": 0.9968, "step": 1719 }, { "epoch": 0.0, "learning_rate": 4.999967616356257e-05, "loss": 0.7371, "step": 1720 }, { "epoch": 0.0, "learning_rate": 4.999967576364197e-05, "loss": 0.9805, "step": 1721 }, { "epoch": 0.0, "learning_rate": 4.9999675363474584e-05, "loss": 1.1072, "step": 1722 }, { "epoch": 0.0, "learning_rate": 4.999967496306041e-05, "loss": 1.0521, "step": 1723 }, { "epoch": 0.0, "learning_rate": 4.999967456239945e-05, "loss": 0.7374, "step": 1724 }, { "epoch": 0.0, "learning_rate": 4.999967416149171e-05, "loss": 1.0237, "step": 1725 }, { "epoch": 0.0, "learning_rate": 4.9999673760337175e-05, "loss": 1.0984, "step": 1726 }, { "epoch": 0.0, "learning_rate": 4.999967335893585e-05, "loss": 1.1678, "step": 1727 }, { "epoch": 0.0, "learning_rate": 4.999967295728775e-05, "loss": 1.3822, "step": 1728 }, { "epoch": 0.0, "learning_rate": 4.999967255539286e-05, "loss": 1.2259, "step": 1729 }, { "epoch": 0.0, "learning_rate": 4.999967215325118e-05, "loss": 1.2165, "step": 1730 }, { "epoch": 0.0, "learning_rate": 4.999967175086272e-05, "loss": 1.2102, "step": 1731 }, { "epoch": 0.0, "learning_rate": 4.9999671348227465e-05, "loss": 1.3913, "step": 1732 }, { "epoch": 0.0, "learning_rate": 4.999967094534543e-05, "loss": 0.938, "step": 1733 }, { "epoch": 0.0, "learning_rate": 4.9999670542216614e-05, "loss": 1.6322, "step": 1734 }, { "epoch": 0.0, "learning_rate": 4.9999670138841005e-05, "loss": 1.2323, "step": 1735 }, { "epoch": 0.0, "learning_rate": 4.9999669735218605e-05, "loss": 1.0992, "step": 1736 }, { "epoch": 0.0, "learning_rate": 4.999966933134943e-05, "loss": 1.1555, "step": 1737 }, { "epoch": 0.0, "learning_rate": 4.9999668927233456e-05, "loss": 1.0509, "step": 1738 }, { "epoch": 0.0, "learning_rate": 4.999966852287071e-05, "loss": 1.0388, "step": 1739 }, { "epoch": 0.0, "learning_rate": 4.999966811826117e-05, "loss": 1.2675, "step": 1740 }, { "epoch": 0.0, "learning_rate": 4.999966771340484e-05, "loss": 1.161, "step": 1741 }, { "epoch": 0.0, "learning_rate": 4.9999667308301736e-05, "loss": 1.75, "step": 1742 }, { "epoch": 0.0, "learning_rate": 4.9999666902951835e-05, "loss": 0.7144, "step": 1743 }, { "epoch": 0.0, "learning_rate": 4.999966649735515e-05, "loss": 0.7856, "step": 1744 }, { "epoch": 0.0, "learning_rate": 4.999966609151168e-05, "loss": 1.1215, "step": 1745 }, { "epoch": 0.0, "learning_rate": 4.9999665685421425e-05, "loss": 1.0481, "step": 1746 }, { "epoch": 0.0, "learning_rate": 4.999966527908438e-05, "loss": 1.1639, "step": 1747 }, { "epoch": 0.0, "learning_rate": 4.999966487250055e-05, "loss": 0.5334, "step": 1748 }, { "epoch": 0.0, "learning_rate": 4.999966446566994e-05, "loss": 1.1725, "step": 1749 }, { "epoch": 0.0, "learning_rate": 4.999966405859253e-05, "loss": 1.1146, "step": 1750 }, { "epoch": 0.0, "learning_rate": 4.9999663651268346e-05, "loss": 1.1789, "step": 1751 }, { "epoch": 0.0, "learning_rate": 4.9999663243697376e-05, "loss": 1.1226, "step": 1752 }, { "epoch": 0.0, "learning_rate": 4.9999662835879615e-05, "loss": 1.1025, "step": 1753 }, { "epoch": 0.0, "learning_rate": 4.999966242781507e-05, "loss": 1.0359, "step": 1754 }, { "epoch": 0.0, "learning_rate": 4.999966201950374e-05, "loss": 1.102, "step": 1755 }, { "epoch": 0.0, "learning_rate": 4.999966161094562e-05, "loss": 1.283, "step": 1756 }, { "epoch": 0.0, "learning_rate": 4.9999661202140715e-05, "loss": 1.1886, "step": 1757 }, { "epoch": 0.0, "learning_rate": 4.999966079308902e-05, "loss": 1.2364, "step": 1758 }, { "epoch": 0.0, "learning_rate": 4.9999660383790546e-05, "loss": 1.0793, "step": 1759 }, { "epoch": 0.0, "learning_rate": 4.999965997424529e-05, "loss": 0.7617, "step": 1760 }, { "epoch": 0.0, "learning_rate": 4.999965956445324e-05, "loss": 1.2649, "step": 1761 }, { "epoch": 0.0, "learning_rate": 4.99996591544144e-05, "loss": 1.3203, "step": 1762 }, { "epoch": 0.0, "learning_rate": 4.9999658744128785e-05, "loss": 0.9644, "step": 1763 }, { "epoch": 0.0, "learning_rate": 4.999965833359638e-05, "loss": 1.1444, "step": 1764 }, { "epoch": 0.0, "learning_rate": 4.9999657922817186e-05, "loss": 1.0832, "step": 1765 }, { "epoch": 0.0, "learning_rate": 4.999965751179121e-05, "loss": 1.0513, "step": 1766 }, { "epoch": 0.0, "learning_rate": 4.999965710051844e-05, "loss": 0.9997, "step": 1767 }, { "epoch": 0.0, "learning_rate": 4.9999656688998895e-05, "loss": 1.2584, "step": 1768 }, { "epoch": 0.0, "learning_rate": 4.999965627723256e-05, "loss": 1.4774, "step": 1769 }, { "epoch": 0.0, "learning_rate": 4.999965586521943e-05, "loss": 1.3373, "step": 1770 }, { "epoch": 0.0, "learning_rate": 4.999965545295952e-05, "loss": 0.9803, "step": 1771 }, { "epoch": 0.0, "learning_rate": 4.999965504045283e-05, "loss": 0.8081, "step": 1772 }, { "epoch": 0.0, "learning_rate": 4.9999654627699345e-05, "loss": 1.2502, "step": 1773 }, { "epoch": 0.0, "learning_rate": 4.999965421469908e-05, "loss": 0.9243, "step": 1774 }, { "epoch": 0.0, "learning_rate": 4.9999653801452024e-05, "loss": 0.9536, "step": 1775 }, { "epoch": 0.0, "learning_rate": 4.9999653387958186e-05, "loss": 1.033, "step": 1776 }, { "epoch": 0.0, "learning_rate": 4.999965297421756e-05, "loss": 1.0354, "step": 1777 }, { "epoch": 0.0, "learning_rate": 4.999965256023015e-05, "loss": 1.0212, "step": 1778 }, { "epoch": 0.0, "learning_rate": 4.9999652145995944e-05, "loss": 1.3128, "step": 1779 }, { "epoch": 0.0, "learning_rate": 4.999965173151496e-05, "loss": 0.8684, "step": 1780 }, { "epoch": 0.0, "learning_rate": 4.99996513167872e-05, "loss": 1.0906, "step": 1781 }, { "epoch": 0.0, "learning_rate": 4.999965090181264e-05, "loss": 1.2967, "step": 1782 }, { "epoch": 0.0, "learning_rate": 4.9999650486591295e-05, "loss": 1.1864, "step": 1783 }, { "epoch": 0.0, "learning_rate": 4.999965007112317e-05, "loss": 1.2853, "step": 1784 }, { "epoch": 0.0, "learning_rate": 4.999964965540825e-05, "loss": 1.3813, "step": 1785 }, { "epoch": 0.0, "learning_rate": 4.9999649239446553e-05, "loss": 1.1605, "step": 1786 }, { "epoch": 0.0, "learning_rate": 4.999964882323807e-05, "loss": 1.12, "step": 1787 }, { "epoch": 0.0, "learning_rate": 4.9999648406782795e-05, "loss": 1.1585, "step": 1788 }, { "epoch": 0.0, "learning_rate": 4.999964799008074e-05, "loss": 0.8351, "step": 1789 }, { "epoch": 0.0, "learning_rate": 4.99996475731319e-05, "loss": 1.0039, "step": 1790 }, { "epoch": 0.0, "learning_rate": 4.999964715593627e-05, "loss": 0.9833, "step": 1791 }, { "epoch": 0.0, "learning_rate": 4.9999646738493855e-05, "loss": 1.0115, "step": 1792 }, { "epoch": 0.0, "learning_rate": 4.999964632080465e-05, "loss": 0.8991, "step": 1793 }, { "epoch": 0.0, "learning_rate": 4.9999645902868666e-05, "loss": 0.9605, "step": 1794 }, { "epoch": 0.0, "learning_rate": 4.9999645484685884e-05, "loss": 0.9477, "step": 1795 }, { "epoch": 0.0, "learning_rate": 4.999964506625633e-05, "loss": 0.8547, "step": 1796 }, { "epoch": 0.0, "learning_rate": 4.999964464757998e-05, "loss": 0.9197, "step": 1797 }, { "epoch": 0.0, "learning_rate": 4.999964422865685e-05, "loss": 1.047, "step": 1798 }, { "epoch": 0.0, "learning_rate": 4.999964380948694e-05, "loss": 0.7114, "step": 1799 }, { "epoch": 0.0, "learning_rate": 4.9999643390070225e-05, "loss": 1.0012, "step": 1800 }, { "epoch": 0.0, "learning_rate": 4.999964297040674e-05, "loss": 1.1069, "step": 1801 }, { "epoch": 0.0, "learning_rate": 4.9999642550496467e-05, "loss": 1.0559, "step": 1802 }, { "epoch": 0.0, "learning_rate": 4.999964213033941e-05, "loss": 1.1771, "step": 1803 }, { "epoch": 0.0, "learning_rate": 4.9999641709935556e-05, "loss": 0.9095, "step": 1804 }, { "epoch": 0.0, "learning_rate": 4.999964128928493e-05, "loss": 1.3731, "step": 1805 }, { "epoch": 0.0, "learning_rate": 4.999964086838751e-05, "loss": 1.2707, "step": 1806 }, { "epoch": 0.0, "learning_rate": 4.99996404472433e-05, "loss": 1.3383, "step": 1807 }, { "epoch": 0.0, "learning_rate": 4.999964002585231e-05, "loss": 1.1379, "step": 1808 }, { "epoch": 0.0, "learning_rate": 4.999963960421453e-05, "loss": 1.5423, "step": 1809 }, { "epoch": 0.0, "learning_rate": 4.999963918232997e-05, "loss": 1.1511, "step": 1810 }, { "epoch": 0.0, "learning_rate": 4.999963876019862e-05, "loss": 1.3019, "step": 1811 }, { "epoch": 0.0, "learning_rate": 4.999963833782049e-05, "loss": 1.1668, "step": 1812 }, { "epoch": 0.0, "learning_rate": 4.999963791519556e-05, "loss": 0.9874, "step": 1813 }, { "epoch": 0.0, "learning_rate": 4.9999637492323856e-05, "loss": 1.431, "step": 1814 }, { "epoch": 0.0, "learning_rate": 4.9999637069205366e-05, "loss": 1.8638, "step": 1815 }, { "epoch": 0.0, "learning_rate": 4.9999636645840084e-05, "loss": 1.1395, "step": 1816 }, { "epoch": 0.0, "learning_rate": 4.999963622222803e-05, "loss": 1.125, "step": 1817 }, { "epoch": 0.0, "learning_rate": 4.9999635798369174e-05, "loss": 1.0607, "step": 1818 }, { "epoch": 0.0, "learning_rate": 4.9999635374263545e-05, "loss": 1.1098, "step": 1819 }, { "epoch": 0.0, "learning_rate": 4.9999634949911124e-05, "loss": 0.6851, "step": 1820 }, { "epoch": 0.0, "learning_rate": 4.999963452531191e-05, "loss": 1.1234, "step": 1821 }, { "epoch": 0.0, "learning_rate": 4.999963410046592e-05, "loss": 1.3048, "step": 1822 }, { "epoch": 0.0, "learning_rate": 4.999963367537313e-05, "loss": 1.1247, "step": 1823 }, { "epoch": 0.0, "learning_rate": 4.9999633250033574e-05, "loss": 1.0778, "step": 1824 }, { "epoch": 0.0, "learning_rate": 4.999963282444722e-05, "loss": 0.8994, "step": 1825 }, { "epoch": 0.0, "learning_rate": 4.999963239861409e-05, "loss": 1.1417, "step": 1826 }, { "epoch": 0.0, "learning_rate": 4.9999631972534166e-05, "loss": 1.0574, "step": 1827 }, { "epoch": 0.0, "learning_rate": 4.999963154620746e-05, "loss": 1.1599, "step": 1828 }, { "epoch": 0.0, "learning_rate": 4.999963111963396e-05, "loss": 1.238, "step": 1829 }, { "epoch": 0.0, "learning_rate": 4.999963069281368e-05, "loss": 1.6336, "step": 1830 }, { "epoch": 0.0, "learning_rate": 4.9999630265746616e-05, "loss": 2.0859, "step": 1831 }, { "epoch": 0.0, "learning_rate": 4.9999629838432765e-05, "loss": 1.2952, "step": 1832 }, { "epoch": 0.0, "learning_rate": 4.999962941087213e-05, "loss": 1.384, "step": 1833 }, { "epoch": 0.0, "learning_rate": 4.999962898306471e-05, "loss": 0.9765, "step": 1834 }, { "epoch": 0.0, "learning_rate": 4.999962855501049e-05, "loss": 1.0935, "step": 1835 }, { "epoch": 0.0, "learning_rate": 4.99996281267095e-05, "loss": 1.1919, "step": 1836 }, { "epoch": 0.0, "learning_rate": 4.9999627698161725e-05, "loss": 0.9998, "step": 1837 }, { "epoch": 0.0, "learning_rate": 4.999962726936715e-05, "loss": 1.155, "step": 1838 }, { "epoch": 0.0, "learning_rate": 4.9999626840325794e-05, "loss": 1.1796, "step": 1839 }, { "epoch": 0.0, "learning_rate": 4.999962641103766e-05, "loss": 1.3332, "step": 1840 }, { "epoch": 0.0, "learning_rate": 4.999962598150274e-05, "loss": 1.0477, "step": 1841 }, { "epoch": 0.0, "learning_rate": 4.999962555172103e-05, "loss": 1.112, "step": 1842 }, { "epoch": 0.0, "learning_rate": 4.9999625121692536e-05, "loss": 0.9866, "step": 1843 }, { "epoch": 0.0, "learning_rate": 4.9999624691417254e-05, "loss": 0.7711, "step": 1844 }, { "epoch": 0.0, "learning_rate": 4.999962426089518e-05, "loss": 0.9799, "step": 1845 }, { "epoch": 0.0, "learning_rate": 4.999962383012633e-05, "loss": 0.6033, "step": 1846 }, { "epoch": 0.0, "learning_rate": 4.9999623399110694e-05, "loss": 0.7475, "step": 1847 }, { "epoch": 0.0, "learning_rate": 4.999962296784827e-05, "loss": 0.7878, "step": 1848 }, { "epoch": 0.0, "learning_rate": 4.999962253633906e-05, "loss": 1.272, "step": 1849 }, { "epoch": 0.0, "learning_rate": 4.9999622104583065e-05, "loss": 1.2431, "step": 1850 }, { "epoch": 0.0, "learning_rate": 4.999962167258029e-05, "loss": 1.1266, "step": 1851 }, { "epoch": 0.0, "learning_rate": 4.999962124033072e-05, "loss": 1.2129, "step": 1852 }, { "epoch": 0.0, "learning_rate": 4.9999620807834366e-05, "loss": 1.1271, "step": 1853 }, { "epoch": 0.0, "learning_rate": 4.999962037509123e-05, "loss": 1.1641, "step": 1854 }, { "epoch": 0.0, "learning_rate": 4.99996199421013e-05, "loss": 1.5963, "step": 1855 }, { "epoch": 0.0, "learning_rate": 4.999961950886459e-05, "loss": 1.5356, "step": 1856 }, { "epoch": 0.0, "learning_rate": 4.99996190753811e-05, "loss": 1.118, "step": 1857 }, { "epoch": 0.0, "learning_rate": 4.999961864165081e-05, "loss": 1.4149, "step": 1858 }, { "epoch": 0.0, "learning_rate": 4.9999618207673746e-05, "loss": 0.9634, "step": 1859 }, { "epoch": 0.0, "learning_rate": 4.999961777344989e-05, "loss": 1.0974, "step": 1860 }, { "epoch": 0.0, "learning_rate": 4.999961733897925e-05, "loss": 1.1586, "step": 1861 }, { "epoch": 0.0, "learning_rate": 4.999961690426183e-05, "loss": 1.1899, "step": 1862 }, { "epoch": 0.0, "learning_rate": 4.999961646929762e-05, "loss": 1.4185, "step": 1863 }, { "epoch": 0.0, "learning_rate": 4.999961603408663e-05, "loss": 0.9067, "step": 1864 }, { "epoch": 0.0, "learning_rate": 4.999961559862885e-05, "loss": 1.2734, "step": 1865 }, { "epoch": 0.0, "learning_rate": 4.9999615162924276e-05, "loss": 1.0768, "step": 1866 }, { "epoch": 0.0, "learning_rate": 4.9999614726972925e-05, "loss": 1.469, "step": 1867 }, { "epoch": 0.0, "learning_rate": 4.999961429077479e-05, "loss": 1.3085, "step": 1868 }, { "epoch": 0.0, "learning_rate": 4.999961385432986e-05, "loss": 1.064, "step": 1869 }, { "epoch": 0.0, "learning_rate": 4.999961341763816e-05, "loss": 1.2152, "step": 1870 }, { "epoch": 0.0, "learning_rate": 4.999961298069966e-05, "loss": 0.6752, "step": 1871 }, { "epoch": 0.0, "learning_rate": 4.999961254351438e-05, "loss": 0.4043, "step": 1872 }, { "epoch": 0.0, "learning_rate": 4.999961210608232e-05, "loss": 0.7948, "step": 1873 }, { "epoch": 0.0, "learning_rate": 4.9999611668403455e-05, "loss": 1.1583, "step": 1874 }, { "epoch": 0.0, "learning_rate": 4.999961123047782e-05, "loss": 1.4685, "step": 1875 }, { "epoch": 0.0, "learning_rate": 4.99996107923054e-05, "loss": 1.2818, "step": 1876 }, { "epoch": 0.0, "learning_rate": 4.999961035388619e-05, "loss": 1.0971, "step": 1877 }, { "epoch": 0.0, "learning_rate": 4.99996099152202e-05, "loss": 1.4049, "step": 1878 }, { "epoch": 0.0, "learning_rate": 4.9999609476307415e-05, "loss": 1.1807, "step": 1879 }, { "epoch": 0.0, "learning_rate": 4.999960903714785e-05, "loss": 1.435, "step": 1880 }, { "epoch": 0.0, "learning_rate": 4.99996085977415e-05, "loss": 1.14, "step": 1881 }, { "epoch": 0.0, "learning_rate": 4.999960815808836e-05, "loss": 1.0288, "step": 1882 }, { "epoch": 0.0, "learning_rate": 4.9999607718188435e-05, "loss": 0.9891, "step": 1883 }, { "epoch": 0.0, "learning_rate": 4.999960727804173e-05, "loss": 1.1035, "step": 1884 }, { "epoch": 0.0, "learning_rate": 4.999960683764824e-05, "loss": 1.2492, "step": 1885 }, { "epoch": 0.0, "learning_rate": 4.999960639700796e-05, "loss": 1.1905, "step": 1886 }, { "epoch": 0.0, "learning_rate": 4.999960595612089e-05, "loss": 1.0795, "step": 1887 }, { "epoch": 0.0, "learning_rate": 4.999960551498704e-05, "loss": 1.0844, "step": 1888 }, { "epoch": 0.0, "learning_rate": 4.9999605073606406e-05, "loss": 1.2944, "step": 1889 }, { "epoch": 0.0, "learning_rate": 4.9999604631978986e-05, "loss": 1.1076, "step": 1890 }, { "epoch": 0.0, "learning_rate": 4.9999604190104774e-05, "loss": 1.2012, "step": 1891 }, { "epoch": 0.0, "learning_rate": 4.9999603747983784e-05, "loss": 1.2852, "step": 1892 }, { "epoch": 0.0, "learning_rate": 4.9999603305616e-05, "loss": 1.0921, "step": 1893 }, { "epoch": 0.0, "learning_rate": 4.999960286300144e-05, "loss": 1.2454, "step": 1894 }, { "epoch": 0.0, "learning_rate": 4.999960242014009e-05, "loss": 1.0453, "step": 1895 }, { "epoch": 0.0, "learning_rate": 4.9999601977031957e-05, "loss": 1.1153, "step": 1896 }, { "epoch": 0.0, "learning_rate": 4.9999601533677036e-05, "loss": 1.0698, "step": 1897 }, { "epoch": 0.0, "learning_rate": 4.999960109007533e-05, "loss": 1.2448, "step": 1898 }, { "epoch": 0.0, "learning_rate": 4.999960064622684e-05, "loss": 1.2356, "step": 1899 }, { "epoch": 0.0, "learning_rate": 4.9999600202131567e-05, "loss": 0.9474, "step": 1900 }, { "epoch": 0.0, "learning_rate": 4.99995997577895e-05, "loss": 1.1506, "step": 1901 }, { "epoch": 0.0, "learning_rate": 4.999959931320065e-05, "loss": 1.3618, "step": 1902 }, { "epoch": 0.0, "learning_rate": 4.9999598868365014e-05, "loss": 1.0169, "step": 1903 }, { "epoch": 0.0, "learning_rate": 4.99995984232826e-05, "loss": 1.2114, "step": 1904 }, { "epoch": 0.0, "learning_rate": 4.9999597977953395e-05, "loss": 1.1541, "step": 1905 }, { "epoch": 0.0, "learning_rate": 4.9999597532377405e-05, "loss": 0.8533, "step": 1906 }, { "epoch": 0.0, "learning_rate": 4.999959708655463e-05, "loss": 1.1301, "step": 1907 }, { "epoch": 0.0, "learning_rate": 4.999959664048507e-05, "loss": 0.9514, "step": 1908 }, { "epoch": 0.0, "learning_rate": 4.999959619416872e-05, "loss": 1.0316, "step": 1909 }, { "epoch": 0.0, "learning_rate": 4.9999595747605585e-05, "loss": 1.1069, "step": 1910 }, { "epoch": 0.0, "learning_rate": 4.999959530079567e-05, "loss": 0.9771, "step": 1911 }, { "epoch": 0.0, "learning_rate": 4.9999594853738966e-05, "loss": 0.8837, "step": 1912 }, { "epoch": 0.0, "learning_rate": 4.999959440643548e-05, "loss": 0.7769, "step": 1913 }, { "epoch": 0.0, "learning_rate": 4.99995939588852e-05, "loss": 0.7168, "step": 1914 }, { "epoch": 0.0, "learning_rate": 4.999959351108815e-05, "loss": 0.6889, "step": 1915 }, { "epoch": 0.0, "learning_rate": 4.9999593063044306e-05, "loss": 0.5915, "step": 1916 }, { "epoch": 0.0, "learning_rate": 4.999959261475367e-05, "loss": 0.6089, "step": 1917 }, { "epoch": 0.0, "learning_rate": 4.9999592166216256e-05, "loss": 0.6052, "step": 1918 }, { "epoch": 0.0, "learning_rate": 4.999959171743205e-05, "loss": 0.5808, "step": 1919 }, { "epoch": 0.0, "learning_rate": 4.999959126840107e-05, "loss": 0.5909, "step": 1920 }, { "epoch": 0.0, "learning_rate": 4.99995908191233e-05, "loss": 0.5403, "step": 1921 }, { "epoch": 0.0, "learning_rate": 4.9999590369598734e-05, "loss": 0.6327, "step": 1922 }, { "epoch": 0.0, "learning_rate": 4.99995899198274e-05, "loss": 0.5225, "step": 1923 }, { "epoch": 0.0, "learning_rate": 4.999958946980927e-05, "loss": 1.0703, "step": 1924 }, { "epoch": 0.0, "learning_rate": 4.9999589019544355e-05, "loss": 1.2725, "step": 1925 }, { "epoch": 0.0, "learning_rate": 4.999958856903265e-05, "loss": 1.1882, "step": 1926 }, { "epoch": 0.0, "learning_rate": 4.9999588118274174e-05, "loss": 1.03, "step": 1927 }, { "epoch": 0.0, "learning_rate": 4.99995876672689e-05, "loss": 1.3146, "step": 1928 }, { "epoch": 0.0, "learning_rate": 4.999958721601684e-05, "loss": 1.1833, "step": 1929 }, { "epoch": 0.0, "learning_rate": 4.9999586764518e-05, "loss": 1.0596, "step": 1930 }, { "epoch": 0.0, "learning_rate": 4.999958631277238e-05, "loss": 1.2622, "step": 1931 }, { "epoch": 0.0, "learning_rate": 4.999958586077996e-05, "loss": 0.862, "step": 1932 }, { "epoch": 0.0, "learning_rate": 4.999958540854077e-05, "loss": 1.0546, "step": 1933 }, { "epoch": 0.0, "learning_rate": 4.999958495605479e-05, "loss": 0.5355, "step": 1934 }, { "epoch": 0.0, "learning_rate": 4.999958450332202e-05, "loss": 1.4974, "step": 1935 }, { "epoch": 0.0, "learning_rate": 4.9999584050342466e-05, "loss": 1.2706, "step": 1936 }, { "epoch": 0.0, "learning_rate": 4.999958359711613e-05, "loss": 1.0658, "step": 1937 }, { "epoch": 0.0, "learning_rate": 4.9999583143643e-05, "loss": 1.0819, "step": 1938 }, { "epoch": 0.0, "learning_rate": 4.99995826899231e-05, "loss": 1.0302, "step": 1939 }, { "epoch": 0.0, "learning_rate": 4.99995822359564e-05, "loss": 1.0842, "step": 1940 }, { "epoch": 0.0, "learning_rate": 4.999958178174292e-05, "loss": 1.1691, "step": 1941 }, { "epoch": 0.0, "learning_rate": 4.999958132728266e-05, "loss": 1.2302, "step": 1942 }, { "epoch": 0.0, "learning_rate": 4.9999580872575605e-05, "loss": 0.9427, "step": 1943 }, { "epoch": 0.0, "learning_rate": 4.9999580417621775e-05, "loss": 1.3759, "step": 1944 }, { "epoch": 0.0, "learning_rate": 4.999957996242115e-05, "loss": 1.1939, "step": 1945 }, { "epoch": 0.0, "learning_rate": 4.999957950697374e-05, "loss": 0.8473, "step": 1946 }, { "epoch": 0.0, "learning_rate": 4.9999579051279555e-05, "loss": 1.031, "step": 1947 }, { "epoch": 0.0, "learning_rate": 4.999957859533857e-05, "loss": 0.941, "step": 1948 }, { "epoch": 0.0, "learning_rate": 4.999957813915082e-05, "loss": 0.9295, "step": 1949 }, { "epoch": 0.0, "learning_rate": 4.9999577682716265e-05, "loss": 1.6521, "step": 1950 }, { "epoch": 0.0, "learning_rate": 4.9999577226034935e-05, "loss": 1.2613, "step": 1951 }, { "epoch": 0.0, "learning_rate": 4.999957676910681e-05, "loss": 1.4755, "step": 1952 }, { "epoch": 0.0, "learning_rate": 4.999957631193191e-05, "loss": 1.5328, "step": 1953 }, { "epoch": 0.0, "learning_rate": 4.999957585451022e-05, "loss": 1.3601, "step": 1954 }, { "epoch": 0.0, "learning_rate": 4.9999575396841746e-05, "loss": 1.2172, "step": 1955 }, { "epoch": 0.0, "learning_rate": 4.9999574938926485e-05, "loss": 1.0954, "step": 1956 }, { "epoch": 0.0, "learning_rate": 4.9999574480764446e-05, "loss": 1.0489, "step": 1957 }, { "epoch": 0.0, "learning_rate": 4.999957402235561e-05, "loss": 1.1921, "step": 1958 }, { "epoch": 0.0, "learning_rate": 4.99995735637e-05, "loss": 1.1143, "step": 1959 }, { "epoch": 0.0, "learning_rate": 4.99995731047976e-05, "loss": 1.2846, "step": 1960 }, { "epoch": 0.0, "learning_rate": 4.999957264564841e-05, "loss": 1.1944, "step": 1961 }, { "epoch": 0.0, "learning_rate": 4.999957218625244e-05, "loss": 1.2594, "step": 1962 }, { "epoch": 0.0, "learning_rate": 4.9999571726609686e-05, "loss": 0.9978, "step": 1963 }, { "epoch": 0.0, "learning_rate": 4.999957126672015e-05, "loss": 0.971, "step": 1964 }, { "epoch": 0.0, "learning_rate": 4.999957080658382e-05, "loss": 1.1946, "step": 1965 }, { "epoch": 0.0, "learning_rate": 4.999957034620071e-05, "loss": 1.0838, "step": 1966 }, { "epoch": 0.0, "learning_rate": 4.999956988557081e-05, "loss": 1.233, "step": 1967 }, { "epoch": 0.0, "learning_rate": 4.9999569424694124e-05, "loss": 1.2316, "step": 1968 }, { "epoch": 0.0, "learning_rate": 4.999956896357066e-05, "loss": 1.1695, "step": 1969 }, { "epoch": 0.0, "learning_rate": 4.999956850220041e-05, "loss": 1.1943, "step": 1970 }, { "epoch": 0.0, "learning_rate": 4.999956804058337e-05, "loss": 1.1467, "step": 1971 }, { "epoch": 0.0, "learning_rate": 4.9999567578719545e-05, "loss": 1.1771, "step": 1972 }, { "epoch": 0.0, "learning_rate": 4.999956711660894e-05, "loss": 1.071, "step": 1973 }, { "epoch": 0.0, "learning_rate": 4.9999566654251544e-05, "loss": 1.0812, "step": 1974 }, { "epoch": 0.0, "learning_rate": 4.999956619164736e-05, "loss": 0.6495, "step": 1975 }, { "epoch": 0.0, "learning_rate": 4.99995657287964e-05, "loss": 0.8987, "step": 1976 }, { "epoch": 0.0, "learning_rate": 4.9999565265698656e-05, "loss": 1.7146, "step": 1977 }, { "epoch": 0.0, "learning_rate": 4.999956480235412e-05, "loss": 0.9659, "step": 1978 }, { "epoch": 0.0, "learning_rate": 4.99995643387628e-05, "loss": 0.9835, "step": 1979 }, { "epoch": 0.0, "learning_rate": 4.999956387492469e-05, "loss": 1.2594, "step": 1980 }, { "epoch": 0.0, "learning_rate": 4.9999563410839806e-05, "loss": 1.1015, "step": 1981 }, { "epoch": 0.0, "learning_rate": 4.999956294650813e-05, "loss": 1.5359, "step": 1982 }, { "epoch": 0.0, "learning_rate": 4.999956248192967e-05, "loss": 1.1858, "step": 1983 }, { "epoch": 0.0, "learning_rate": 4.999956201710443e-05, "loss": 2.232, "step": 1984 }, { "epoch": 0.0, "learning_rate": 4.999956155203239e-05, "loss": 1.0845, "step": 1985 }, { "epoch": 0.0, "learning_rate": 4.9999561086713576e-05, "loss": 1.1301, "step": 1986 }, { "epoch": 0.0, "learning_rate": 4.9999560621147975e-05, "loss": 3.6122, "step": 1987 }, { "epoch": 0.0, "learning_rate": 4.999956015533559e-05, "loss": 1.1042, "step": 1988 }, { "epoch": 0.0, "learning_rate": 4.999955968927642e-05, "loss": 1.5292, "step": 1989 }, { "epoch": 0.0, "learning_rate": 4.999955922297047e-05, "loss": 1.2125, "step": 1990 }, { "epoch": 0.0, "learning_rate": 4.999955875641772e-05, "loss": 1.2061, "step": 1991 }, { "epoch": 0.0, "learning_rate": 4.9999558289618196e-05, "loss": 1.3218, "step": 1992 }, { "epoch": 0.0, "learning_rate": 4.9999557822571887e-05, "loss": 1.2492, "step": 1993 }, { "epoch": 0.0, "learning_rate": 4.999955735527879e-05, "loss": 1.6787, "step": 1994 }, { "epoch": 0.0, "learning_rate": 4.9999556887738906e-05, "loss": 1.187, "step": 1995 }, { "epoch": 0.0, "learning_rate": 4.999955641995224e-05, "loss": 1.8543, "step": 1996 }, { "epoch": 0.0, "learning_rate": 4.999955595191879e-05, "loss": 0.6718, "step": 1997 }, { "epoch": 0.0, "learning_rate": 4.999955548363855e-05, "loss": 0.9185, "step": 1998 }, { "epoch": 0.0, "learning_rate": 4.9999555015111534e-05, "loss": 1.3068, "step": 1999 }, { "epoch": 0.0, "learning_rate": 4.9999554546337724e-05, "loss": 1.372, "step": 2000 }, { "epoch": 0.0, "eval_loss": 1.09769606590271, "eval_runtime": 85.4918, "eval_samples_per_second": 16.2, "eval_steps_per_second": 4.059, "step": 2000 }, { "epoch": 0.0, "learning_rate": 4.9999554077317136e-05, "loss": 1.1012, "step": 2001 }, { "epoch": 0.0, "learning_rate": 4.9999553608049757e-05, "loss": 1.2555, "step": 2002 }, { "epoch": 0.0, "learning_rate": 4.999955313853559e-05, "loss": 1.185, "step": 2003 }, { "epoch": 0.0, "learning_rate": 4.999955266877465e-05, "loss": 1.1243, "step": 2004 }, { "epoch": 0.0, "learning_rate": 4.999955219876692e-05, "loss": 1.296, "step": 2005 }, { "epoch": 0.0, "learning_rate": 4.99995517285124e-05, "loss": 1.1022, "step": 2006 }, { "epoch": 0.0, "learning_rate": 4.99995512580111e-05, "loss": 1.2041, "step": 2007 }, { "epoch": 0.0, "learning_rate": 4.9999550787263014e-05, "loss": 1.0095, "step": 2008 }, { "epoch": 0.0, "learning_rate": 4.999955031626814e-05, "loss": 1.1557, "step": 2009 }, { "epoch": 0.0, "learning_rate": 4.9999549845026484e-05, "loss": 1.3879, "step": 2010 }, { "epoch": 0.0, "learning_rate": 4.9999549373538035e-05, "loss": 0.9205, "step": 2011 }, { "epoch": 0.0, "learning_rate": 4.9999548901802815e-05, "loss": 0.1468, "step": 2012 }, { "epoch": 0.0, "learning_rate": 4.99995484298208e-05, "loss": 0.268, "step": 2013 }, { "epoch": 0.0, "learning_rate": 4.9999547957592e-05, "loss": 0.6679, "step": 2014 }, { "epoch": 0.0, "learning_rate": 4.999954748511641e-05, "loss": 0.8559, "step": 2015 }, { "epoch": 0.0, "learning_rate": 4.9999547012394054e-05, "loss": 1.0634, "step": 2016 }, { "epoch": 0.0, "learning_rate": 4.99995465394249e-05, "loss": 1.2137, "step": 2017 }, { "epoch": 0.0, "learning_rate": 4.999954606620896e-05, "loss": 1.2434, "step": 2018 }, { "epoch": 0.0, "learning_rate": 4.999954559274624e-05, "loss": 1.0491, "step": 2019 }, { "epoch": 0.0, "learning_rate": 4.999954511903673e-05, "loss": 1.0955, "step": 2020 }, { "epoch": 0.0, "learning_rate": 4.999954464508044e-05, "loss": 1.3627, "step": 2021 }, { "epoch": 0.0, "learning_rate": 4.9999544170877366e-05, "loss": 0.8931, "step": 2022 }, { "epoch": 0.0, "learning_rate": 4.99995436964275e-05, "loss": 1.1627, "step": 2023 }, { "epoch": 0.0, "learning_rate": 4.999954322173085e-05, "loss": 1.2295, "step": 2024 }, { "epoch": 0.0, "learning_rate": 4.999954274678742e-05, "loss": 1.1959, "step": 2025 }, { "epoch": 0.0, "learning_rate": 4.99995422715972e-05, "loss": 1.1635, "step": 2026 }, { "epoch": 0.0, "learning_rate": 4.9999541796160196e-05, "loss": 0.9611, "step": 2027 }, { "epoch": 0.0, "learning_rate": 4.999954132047642e-05, "loss": 0.7987, "step": 2028 }, { "epoch": 0.0, "learning_rate": 4.9999540844545846e-05, "loss": 1.4583, "step": 2029 }, { "epoch": 0.0, "learning_rate": 4.999954036836849e-05, "loss": 1.2441, "step": 2030 }, { "epoch": 0.0, "learning_rate": 4.999953989194435e-05, "loss": 1.5323, "step": 2031 }, { "epoch": 0.0, "learning_rate": 4.9999539415273415e-05, "loss": 1.1934, "step": 2032 }, { "epoch": 0.0, "learning_rate": 4.999953893835571e-05, "loss": 1.0853, "step": 2033 }, { "epoch": 0.0, "learning_rate": 4.999953846119121e-05, "loss": 1.1559, "step": 2034 }, { "epoch": 0.0, "learning_rate": 4.9999537983779934e-05, "loss": 1.0736, "step": 2035 }, { "epoch": 0.0, "learning_rate": 4.999953750612186e-05, "loss": 1.0305, "step": 2036 }, { "epoch": 0.0, "learning_rate": 4.9999537028217015e-05, "loss": 1.2131, "step": 2037 }, { "epoch": 0.0, "learning_rate": 4.999953655006537e-05, "loss": 1.1931, "step": 2038 }, { "epoch": 0.0, "learning_rate": 4.999953607166695e-05, "loss": 1.179, "step": 2039 }, { "epoch": 0.0, "learning_rate": 4.999953559302174e-05, "loss": 1.0621, "step": 2040 }, { "epoch": 0.0, "learning_rate": 4.999953511412976e-05, "loss": 1.3458, "step": 2041 }, { "epoch": 0.0, "learning_rate": 4.999953463499098e-05, "loss": 1.1783, "step": 2042 }, { "epoch": 0.0, "learning_rate": 4.999953415560542e-05, "loss": 1.159, "step": 2043 }, { "epoch": 0.0, "learning_rate": 4.9999533675973076e-05, "loss": 1.3375, "step": 2044 }, { "epoch": 0.0, "learning_rate": 4.9999533196093946e-05, "loss": 0.6313, "step": 2045 }, { "epoch": 0.0, "learning_rate": 4.999953271596803e-05, "loss": 0.9884, "step": 2046 }, { "epoch": 0.0, "learning_rate": 4.9999532235595325e-05, "loss": 1.2119, "step": 2047 }, { "epoch": 0.0, "learning_rate": 4.999953175497585e-05, "loss": 1.2758, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.999953127410957e-05, "loss": 0.9148, "step": 2049 }, { "epoch": 0.0, "learning_rate": 4.999953079299652e-05, "loss": 1.0894, "step": 2050 }, { "epoch": 0.0, "learning_rate": 4.999953031163668e-05, "loss": 0.7798, "step": 2051 }, { "epoch": 0.0, "learning_rate": 4.9999529830030056e-05, "loss": 0.7435, "step": 2052 }, { "epoch": 0.0, "learning_rate": 4.999952934817664e-05, "loss": 0.8528, "step": 2053 }, { "epoch": 0.0, "learning_rate": 4.999952886607645e-05, "loss": 1.2994, "step": 2054 }, { "epoch": 0.0, "learning_rate": 4.999952838372948e-05, "loss": 1.0807, "step": 2055 }, { "epoch": 0.0, "learning_rate": 4.9999527901135715e-05, "loss": 1.2836, "step": 2056 }, { "epoch": 0.0, "learning_rate": 4.999952741829517e-05, "loss": 1.291, "step": 2057 }, { "epoch": 0.0, "learning_rate": 4.999952693520783e-05, "loss": 1.1041, "step": 2058 }, { "epoch": 0.0, "learning_rate": 4.999952645187371e-05, "loss": 1.0205, "step": 2059 }, { "epoch": 0.0, "learning_rate": 4.999952596829281e-05, "loss": 0.9736, "step": 2060 }, { "epoch": 0.0, "learning_rate": 4.999952548446512e-05, "loss": 1.2984, "step": 2061 }, { "epoch": 0.0, "learning_rate": 4.999952500039065e-05, "loss": 1.2229, "step": 2062 }, { "epoch": 0.0, "learning_rate": 4.99995245160694e-05, "loss": 1.6288, "step": 2063 }, { "epoch": 0.0, "learning_rate": 4.999952403150135e-05, "loss": 1.4125, "step": 2064 }, { "epoch": 0.0, "learning_rate": 4.999952354668652e-05, "loss": 1.3746, "step": 2065 }, { "epoch": 0.0, "learning_rate": 4.999952306162492e-05, "loss": 1.1113, "step": 2066 }, { "epoch": 0.0, "learning_rate": 4.9999522576316516e-05, "loss": 1.2301, "step": 2067 }, { "epoch": 0.0, "learning_rate": 4.999952209076134e-05, "loss": 1.1858, "step": 2068 }, { "epoch": 0.0, "learning_rate": 4.999952160495937e-05, "loss": 1.0311, "step": 2069 }, { "epoch": 0.0, "learning_rate": 4.999952111891062e-05, "loss": 1.2018, "step": 2070 }, { "epoch": 0.0, "learning_rate": 4.9999520632615084e-05, "loss": 1.0876, "step": 2071 }, { "epoch": 0.0, "learning_rate": 4.9999520146072766e-05, "loss": 1.0011, "step": 2072 }, { "epoch": 0.0, "learning_rate": 4.999951965928366e-05, "loss": 1.1342, "step": 2073 }, { "epoch": 0.0, "learning_rate": 4.9999519172247775e-05, "loss": 0.8343, "step": 2074 }, { "epoch": 0.0, "learning_rate": 4.9999518684965096e-05, "loss": 1.3023, "step": 2075 }, { "epoch": 0.0, "learning_rate": 4.999951819743563e-05, "loss": 1.0484, "step": 2076 }, { "epoch": 0.0, "learning_rate": 4.99995177096594e-05, "loss": 0.8818, "step": 2077 }, { "epoch": 0.0, "learning_rate": 4.999951722163637e-05, "loss": 0.5781, "step": 2078 }, { "epoch": 0.0, "learning_rate": 4.999951673336656e-05, "loss": 1.1657, "step": 2079 }, { "epoch": 0.0, "learning_rate": 4.9999516244849956e-05, "loss": 1.2215, "step": 2080 }, { "epoch": 0.0, "learning_rate": 4.999951575608658e-05, "loss": 1.0539, "step": 2081 }, { "epoch": 0.0, "learning_rate": 4.999951526707641e-05, "loss": 1.2303, "step": 2082 }, { "epoch": 0.0, "learning_rate": 4.999951477781946e-05, "loss": 1.5738, "step": 2083 }, { "epoch": 0.0, "learning_rate": 4.999951428831572e-05, "loss": 1.1878, "step": 2084 }, { "epoch": 0.0, "learning_rate": 4.9999513798565204e-05, "loss": 1.1655, "step": 2085 }, { "epoch": 0.0, "learning_rate": 4.999951330856789e-05, "loss": 1.0752, "step": 2086 }, { "epoch": 0.0, "learning_rate": 4.99995128183238e-05, "loss": 0.9155, "step": 2087 }, { "epoch": 0.0, "learning_rate": 4.999951232783293e-05, "loss": 1.0812, "step": 2088 }, { "epoch": 0.0, "learning_rate": 4.999951183709527e-05, "loss": 1.0981, "step": 2089 }, { "epoch": 0.0, "learning_rate": 4.9999511346110827e-05, "loss": 1.0767, "step": 2090 }, { "epoch": 0.0, "learning_rate": 4.999951085487959e-05, "loss": 0.9912, "step": 2091 }, { "epoch": 0.0, "learning_rate": 4.999951036340158e-05, "loss": 1.015, "step": 2092 }, { "epoch": 0.0, "learning_rate": 4.999950987167679e-05, "loss": 1.6259, "step": 2093 }, { "epoch": 0.0, "learning_rate": 4.99995093797052e-05, "loss": 1.4019, "step": 2094 }, { "epoch": 0.0, "learning_rate": 4.999950888748684e-05, "loss": 1.5632, "step": 2095 }, { "epoch": 0.0, "learning_rate": 4.9999508395021685e-05, "loss": 1.2167, "step": 2096 }, { "epoch": 0.0, "learning_rate": 4.999950790230975e-05, "loss": 1.2337, "step": 2097 }, { "epoch": 0.0, "learning_rate": 4.999950740935103e-05, "loss": 1.116, "step": 2098 }, { "epoch": 0.0, "learning_rate": 4.999950691614552e-05, "loss": 1.1899, "step": 2099 }, { "epoch": 0.0, "learning_rate": 4.999950642269323e-05, "loss": 1.1664, "step": 2100 }, { "epoch": 0.0, "learning_rate": 4.9999505928994154e-05, "loss": 0.9768, "step": 2101 }, { "epoch": 0.0, "learning_rate": 4.99995054350483e-05, "loss": 1.3383, "step": 2102 }, { "epoch": 0.0, "learning_rate": 4.999950494085565e-05, "loss": 1.09, "step": 2103 }, { "epoch": 0.0, "learning_rate": 4.9999504446416225e-05, "loss": 1.6185, "step": 2104 }, { "epoch": 0.0, "learning_rate": 4.999950395173001e-05, "loss": 1.2097, "step": 2105 }, { "epoch": 0.0, "learning_rate": 4.999950345679702e-05, "loss": 0.9709, "step": 2106 }, { "epoch": 0.0, "learning_rate": 4.999950296161723e-05, "loss": 1.4045, "step": 2107 }, { "epoch": 0.0, "learning_rate": 4.9999502466190664e-05, "loss": 1.187, "step": 2108 }, { "epoch": 0.0, "learning_rate": 4.999950197051731e-05, "loss": 1.2063, "step": 2109 }, { "epoch": 0.0, "learning_rate": 4.999950147459718e-05, "loss": 1.1091, "step": 2110 }, { "epoch": 0.0, "learning_rate": 4.9999500978430256e-05, "loss": 1.2056, "step": 2111 }, { "epoch": 0.0, "learning_rate": 4.999950048201655e-05, "loss": 0.812, "step": 2112 }, { "epoch": 0.0, "learning_rate": 4.9999499985356055e-05, "loss": 0.5464, "step": 2113 }, { "epoch": 0.0, "learning_rate": 4.9999499488448785e-05, "loss": 1.2549, "step": 2114 }, { "epoch": 0.0, "learning_rate": 4.999949899129472e-05, "loss": 1.3707, "step": 2115 }, { "epoch": 0.0, "learning_rate": 4.999949849389388e-05, "loss": 1.2616, "step": 2116 }, { "epoch": 0.0, "learning_rate": 4.999949799624626e-05, "loss": 1.1822, "step": 2117 }, { "epoch": 0.0, "learning_rate": 4.999949749835184e-05, "loss": 1.0703, "step": 2118 }, { "epoch": 0.0, "learning_rate": 4.999949700021065e-05, "loss": 1.0636, "step": 2119 }, { "epoch": 0.0, "learning_rate": 4.999949650182266e-05, "loss": 1.1332, "step": 2120 }, { "epoch": 0.0, "learning_rate": 4.99994960031879e-05, "loss": 1.0944, "step": 2121 }, { "epoch": 0.0, "learning_rate": 4.999949550430635e-05, "loss": 1.0199, "step": 2122 }, { "epoch": 0.0, "learning_rate": 4.9999495005178015e-05, "loss": 0.8863, "step": 2123 }, { "epoch": 0.0, "learning_rate": 4.999949450580289e-05, "loss": 1.181, "step": 2124 }, { "epoch": 0.0, "learning_rate": 4.9999494006180994e-05, "loss": 1.0959, "step": 2125 }, { "epoch": 0.0, "learning_rate": 4.99994935063123e-05, "loss": 1.0288, "step": 2126 }, { "epoch": 0.0, "learning_rate": 4.999949300619683e-05, "loss": 1.1203, "step": 2127 }, { "epoch": 0.0, "learning_rate": 4.999949250583457e-05, "loss": 1.0955, "step": 2128 }, { "epoch": 0.0, "learning_rate": 4.9999492005225535e-05, "loss": 1.0168, "step": 2129 }, { "epoch": 0.0, "learning_rate": 4.99994915043697e-05, "loss": 0.8773, "step": 2130 }, { "epoch": 0.0, "learning_rate": 4.99994910032671e-05, "loss": 0.4928, "step": 2131 }, { "epoch": 0.0, "learning_rate": 4.99994905019177e-05, "loss": 0.8741, "step": 2132 }, { "epoch": 0.0, "learning_rate": 4.999949000032153e-05, "loss": 1.1476, "step": 2133 }, { "epoch": 0.0, "learning_rate": 4.9999489498478555e-05, "loss": 1.126, "step": 2134 }, { "epoch": 0.0, "learning_rate": 4.999948899638881e-05, "loss": 0.8158, "step": 2135 }, { "epoch": 0.0, "learning_rate": 4.999948849405228e-05, "loss": 0.7589, "step": 2136 }, { "epoch": 0.0, "learning_rate": 4.9999487991468964e-05, "loss": 0.7096, "step": 2137 }, { "epoch": 0.0, "learning_rate": 4.999948748863886e-05, "loss": 0.9201, "step": 2138 }, { "epoch": 0.0, "learning_rate": 4.999948698556197e-05, "loss": 0.8196, "step": 2139 }, { "epoch": 0.0, "learning_rate": 4.99994864822383e-05, "loss": 1.1322, "step": 2140 }, { "epoch": 0.0, "learning_rate": 4.999948597866785e-05, "loss": 1.1709, "step": 2141 }, { "epoch": 0.0, "learning_rate": 4.999948547485061e-05, "loss": 1.1259, "step": 2142 }, { "epoch": 0.0, "learning_rate": 4.999948497078658e-05, "loss": 1.9433, "step": 2143 }, { "epoch": 0.0, "learning_rate": 4.999948446647578e-05, "loss": 1.3849, "step": 2144 }, { "epoch": 0.0, "learning_rate": 4.999948396191818e-05, "loss": 1.5158, "step": 2145 }, { "epoch": 0.0, "learning_rate": 4.9999483457113807e-05, "loss": 1.5195, "step": 2146 }, { "epoch": 0.0, "learning_rate": 4.9999482952062646e-05, "loss": 1.4063, "step": 2147 }, { "epoch": 0.0, "learning_rate": 4.99994824467647e-05, "loss": 2.0924, "step": 2148 }, { "epoch": 0.0, "learning_rate": 4.999948194121997e-05, "loss": 1.1986, "step": 2149 }, { "epoch": 0.0, "learning_rate": 4.9999481435428456e-05, "loss": 0.603, "step": 2150 }, { "epoch": 0.0, "learning_rate": 4.9999480929390164e-05, "loss": 0.5798, "step": 2151 }, { "epoch": 0.0, "learning_rate": 4.999948042310508e-05, "loss": 0.2908, "step": 2152 }, { "epoch": 0.0, "learning_rate": 4.999947991657321e-05, "loss": 0.2309, "step": 2153 }, { "epoch": 0.0, "learning_rate": 4.999947940979456e-05, "loss": 1.0838, "step": 2154 }, { "epoch": 0.0, "learning_rate": 4.9999478902769126e-05, "loss": 1.4886, "step": 2155 }, { "epoch": 0.0, "learning_rate": 4.99994783954969e-05, "loss": 1.1981, "step": 2156 }, { "epoch": 0.0, "learning_rate": 4.9999477887977895e-05, "loss": 1.1871, "step": 2157 }, { "epoch": 0.0, "learning_rate": 4.999947738021211e-05, "loss": 1.2981, "step": 2158 }, { "epoch": 0.0, "learning_rate": 4.999947687219954e-05, "loss": 0.7443, "step": 2159 }, { "epoch": 0.0, "learning_rate": 4.9999476363940176e-05, "loss": 1.2174, "step": 2160 }, { "epoch": 0.0, "learning_rate": 4.999947585543404e-05, "loss": 0.9883, "step": 2161 }, { "epoch": 0.0, "learning_rate": 4.999947534668111e-05, "loss": 0.7254, "step": 2162 }, { "epoch": 0.0, "learning_rate": 4.99994748376814e-05, "loss": 1.322, "step": 2163 }, { "epoch": 0.0, "learning_rate": 4.999947432843491e-05, "loss": 1.4698, "step": 2164 }, { "epoch": 0.0, "learning_rate": 4.9999473818941636e-05, "loss": 1.3622, "step": 2165 }, { "epoch": 0.0, "learning_rate": 4.9999473309201565e-05, "loss": 0.9578, "step": 2166 }, { "epoch": 0.0, "learning_rate": 4.999947279921472e-05, "loss": 1.0192, "step": 2167 }, { "epoch": 0.0, "learning_rate": 4.9999472288981083e-05, "loss": 0.8398, "step": 2168 }, { "epoch": 0.0, "learning_rate": 4.999947177850067e-05, "loss": 0.6333, "step": 2169 }, { "epoch": 0.0, "learning_rate": 4.9999471267773476e-05, "loss": 0.5143, "step": 2170 }, { "epoch": 0.0, "learning_rate": 4.999947075679949e-05, "loss": 0.8075, "step": 2171 }, { "epoch": 0.0, "learning_rate": 4.9999470245578717e-05, "loss": 0.9884, "step": 2172 }, { "epoch": 0.0, "learning_rate": 4.999946973411117e-05, "loss": 1.4855, "step": 2173 }, { "epoch": 0.0, "learning_rate": 4.999946922239683e-05, "loss": 1.0709, "step": 2174 }, { "epoch": 0.0, "learning_rate": 4.999946871043571e-05, "loss": 1.4966, "step": 2175 }, { "epoch": 0.0, "learning_rate": 4.999946819822781e-05, "loss": 1.3703, "step": 2176 }, { "epoch": 0.0, "learning_rate": 4.999946768577312e-05, "loss": 0.8544, "step": 2177 }, { "epoch": 0.0, "learning_rate": 4.999946717307164e-05, "loss": 0.8872, "step": 2178 }, { "epoch": 0.0, "learning_rate": 4.9999466660123386e-05, "loss": 1.1566, "step": 2179 }, { "epoch": 0.0, "learning_rate": 4.999946614692834e-05, "loss": 1.0904, "step": 2180 }, { "epoch": 0.0, "learning_rate": 4.9999465633486514e-05, "loss": 0.8584, "step": 2181 }, { "epoch": 0.0, "learning_rate": 4.999946511979791e-05, "loss": 1.3786, "step": 2182 }, { "epoch": 0.0, "learning_rate": 4.999946460586251e-05, "loss": 1.7855, "step": 2183 }, { "epoch": 0.0, "learning_rate": 4.999946409168033e-05, "loss": 1.0695, "step": 2184 }, { "epoch": 0.0, "learning_rate": 4.999946357725137e-05, "loss": 1.0416, "step": 2185 }, { "epoch": 0.0, "learning_rate": 4.9999463062575616e-05, "loss": 0.7917, "step": 2186 }, { "epoch": 0.0, "learning_rate": 4.9999462547653086e-05, "loss": 4.7444, "step": 2187 }, { "epoch": 0.0, "learning_rate": 4.999946203248377e-05, "loss": 6.0795, "step": 2188 }, { "epoch": 0.0, "learning_rate": 4.999946151706767e-05, "loss": 5.9444, "step": 2189 }, { "epoch": 0.0, "learning_rate": 4.999946100140479e-05, "loss": 5.8905, "step": 2190 }, { "epoch": 0.0, "learning_rate": 4.999946048549512e-05, "loss": 5.7887, "step": 2191 }, { "epoch": 0.0, "learning_rate": 4.9999459969338666e-05, "loss": 1.2573, "step": 2192 }, { "epoch": 0.0, "learning_rate": 4.9999459452935435e-05, "loss": 0.9757, "step": 2193 }, { "epoch": 0.0, "learning_rate": 4.999945893628541e-05, "loss": 0.6567, "step": 2194 }, { "epoch": 0.0, "learning_rate": 4.9999458419388604e-05, "loss": 0.8967, "step": 2195 }, { "epoch": 0.0, "learning_rate": 4.999945790224502e-05, "loss": 1.2589, "step": 2196 }, { "epoch": 0.0, "learning_rate": 4.999945738485464e-05, "loss": 1.1499, "step": 2197 }, { "epoch": 0.0, "learning_rate": 4.9999456867217486e-05, "loss": 0.9498, "step": 2198 }, { "epoch": 0.0, "learning_rate": 4.999945634933355e-05, "loss": 1.3002, "step": 2199 }, { "epoch": 0.0, "learning_rate": 4.999945583120282e-05, "loss": 1.4875, "step": 2200 }, { "epoch": 0.0, "learning_rate": 4.999945531282532e-05, "loss": 1.1735, "step": 2201 }, { "epoch": 0.0, "learning_rate": 4.999945479420102e-05, "loss": 0.207, "step": 2202 }, { "epoch": 0.0, "learning_rate": 4.9999454275329945e-05, "loss": 0.1521, "step": 2203 }, { "epoch": 0.0, "learning_rate": 4.9999453756212075e-05, "loss": 0.1431, "step": 2204 }, { "epoch": 0.0, "learning_rate": 4.999945323684744e-05, "loss": 0.0981, "step": 2205 }, { "epoch": 0.0, "learning_rate": 4.999945271723601e-05, "loss": 0.082, "step": 2206 }, { "epoch": 0.0, "learning_rate": 4.9999452197377796e-05, "loss": 0.0651, "step": 2207 }, { "epoch": 0.0, "learning_rate": 4.999945167727279e-05, "loss": 0.0447, "step": 2208 }, { "epoch": 0.0, "learning_rate": 4.999945115692101e-05, "loss": 0.0372, "step": 2209 }, { "epoch": 0.0, "learning_rate": 4.999945063632245e-05, "loss": 0.0263, "step": 2210 }, { "epoch": 0.0, "learning_rate": 4.99994501154771e-05, "loss": 0.0216, "step": 2211 }, { "epoch": 0.0, "learning_rate": 4.999944959438496e-05, "loss": 0.0183, "step": 2212 }, { "epoch": 0.0, "learning_rate": 4.999944907304604e-05, "loss": 0.0182, "step": 2213 }, { "epoch": 0.0, "learning_rate": 4.9999448551460345e-05, "loss": 0.0174, "step": 2214 }, { "epoch": 0.0, "learning_rate": 4.999944802962786e-05, "loss": 0.0185, "step": 2215 }, { "epoch": 0.0, "learning_rate": 4.999944750754858e-05, "loss": 0.0171, "step": 2216 }, { "epoch": 0.0, "learning_rate": 4.999944698522253e-05, "loss": 0.0146, "step": 2217 }, { "epoch": 0.0, "learning_rate": 4.9999446462649694e-05, "loss": 0.0241, "step": 2218 }, { "epoch": 0.0, "learning_rate": 4.999944593983007e-05, "loss": 0.0217, "step": 2219 }, { "epoch": 0.0, "learning_rate": 4.999944541676367e-05, "loss": 0.0212, "step": 2220 }, { "epoch": 0.0, "learning_rate": 4.9999444893450476e-05, "loss": 0.02, "step": 2221 }, { "epoch": 0.0, "learning_rate": 4.99994443698905e-05, "loss": 0.019, "step": 2222 }, { "epoch": 0.0, "learning_rate": 4.999944384608375e-05, "loss": 0.019, "step": 2223 }, { "epoch": 0.0, "learning_rate": 4.99994433220302e-05, "loss": 0.017, "step": 2224 }, { "epoch": 0.0, "learning_rate": 4.999944279772988e-05, "loss": 0.0168, "step": 2225 }, { "epoch": 0.0, "learning_rate": 4.9999442273182765e-05, "loss": 0.0184, "step": 2226 }, { "epoch": 0.0, "learning_rate": 4.999944174838887e-05, "loss": 0.0185, "step": 2227 }, { "epoch": 0.0, "learning_rate": 4.9999441223348195e-05, "loss": 0.0151, "step": 2228 }, { "epoch": 0.0, "learning_rate": 4.999944069806073e-05, "loss": 0.0121, "step": 2229 }, { "epoch": 0.0, "learning_rate": 4.9999440172526494e-05, "loss": 0.0186, "step": 2230 }, { "epoch": 0.0, "learning_rate": 4.999943964674546e-05, "loss": 0.0138, "step": 2231 }, { "epoch": 0.0, "learning_rate": 4.9999439120717646e-05, "loss": 0.1738, "step": 2232 }, { "epoch": 0.0, "learning_rate": 4.999943859444305e-05, "loss": 0.8234, "step": 2233 }, { "epoch": 0.0, "learning_rate": 4.9999438067921666e-05, "loss": 0.9426, "step": 2234 }, { "epoch": 0.0, "learning_rate": 4.99994375411535e-05, "loss": 1.2646, "step": 2235 }, { "epoch": 0.0, "learning_rate": 4.999943701413855e-05, "loss": 1.3297, "step": 2236 }, { "epoch": 0.0, "learning_rate": 4.9999436486876814e-05, "loss": 1.3769, "step": 2237 }, { "epoch": 0.0, "learning_rate": 4.9999435959368303e-05, "loss": 1.1826, "step": 2238 }, { "epoch": 0.0, "learning_rate": 4.9999435431613e-05, "loss": 1.0608, "step": 2239 }, { "epoch": 0.0, "learning_rate": 4.9999434903610914e-05, "loss": 1.1587, "step": 2240 }, { "epoch": 0.0, "learning_rate": 4.999943437536205e-05, "loss": 1.1704, "step": 2241 }, { "epoch": 0.0, "learning_rate": 4.999943384686639e-05, "loss": 1.0112, "step": 2242 }, { "epoch": 0.0, "learning_rate": 4.999943331812396e-05, "loss": 1.4902, "step": 2243 }, { "epoch": 0.0, "learning_rate": 4.999943278913474e-05, "loss": 0.7887, "step": 2244 }, { "epoch": 0.0, "learning_rate": 4.999943225989873e-05, "loss": 1.4997, "step": 2245 }, { "epoch": 0.0, "learning_rate": 4.9999431730415944e-05, "loss": 1.2394, "step": 2246 }, { "epoch": 0.0, "learning_rate": 4.999943120068638e-05, "loss": 1.1772, "step": 2247 }, { "epoch": 0.0, "learning_rate": 4.999943067071002e-05, "loss": 1.0988, "step": 2248 }, { "epoch": 0.0, "learning_rate": 4.999943014048688e-05, "loss": 1.0802, "step": 2249 }, { "epoch": 0.0, "learning_rate": 4.9999429610016954e-05, "loss": 1.0132, "step": 2250 }, { "epoch": 0.0, "learning_rate": 4.999942907930025e-05, "loss": 0.8749, "step": 2251 }, { "epoch": 0.0, "learning_rate": 4.999942854833676e-05, "loss": 1.2088, "step": 2252 }, { "epoch": 0.0, "learning_rate": 4.999942801712648e-05, "loss": 1.4583, "step": 2253 }, { "epoch": 0.0, "learning_rate": 4.999942748566942e-05, "loss": 1.1602, "step": 2254 }, { "epoch": 0.0, "learning_rate": 4.999942695396558e-05, "loss": 1.3084, "step": 2255 }, { "epoch": 0.0, "learning_rate": 4.999942642201495e-05, "loss": 1.084, "step": 2256 }, { "epoch": 0.0, "learning_rate": 4.999942588981754e-05, "loss": 1.3865, "step": 2257 }, { "epoch": 0.0, "learning_rate": 4.999942535737335e-05, "loss": 1.3727, "step": 2258 }, { "epoch": 0.0, "learning_rate": 4.999942482468237e-05, "loss": 0.8791, "step": 2259 }, { "epoch": 0.0, "learning_rate": 4.9999424291744615e-05, "loss": 1.2743, "step": 2260 }, { "epoch": 0.0, "learning_rate": 4.999942375856006e-05, "loss": 1.3113, "step": 2261 }, { "epoch": 0.0, "learning_rate": 4.999942322512873e-05, "loss": 1.4118, "step": 2262 }, { "epoch": 0.0, "learning_rate": 4.999942269145062e-05, "loss": 0.969, "step": 2263 }, { "epoch": 0.0, "learning_rate": 4.999942215752572e-05, "loss": 0.9726, "step": 2264 }, { "epoch": 0.0, "learning_rate": 4.999942162335404e-05, "loss": 1.5553, "step": 2265 }, { "epoch": 0.0, "learning_rate": 4.9999421088935585e-05, "loss": 1.2159, "step": 2266 }, { "epoch": 0.0, "learning_rate": 4.999942055427033e-05, "loss": 1.1292, "step": 2267 }, { "epoch": 0.0, "learning_rate": 4.9999420019358304e-05, "loss": 0.9931, "step": 2268 }, { "epoch": 0.0, "learning_rate": 4.9999419484199486e-05, "loss": 1.0924, "step": 2269 }, { "epoch": 0.0, "learning_rate": 4.999941894879388e-05, "loss": 1.0195, "step": 2270 }, { "epoch": 0.0, "learning_rate": 4.99994184131415e-05, "loss": 1.1812, "step": 2271 }, { "epoch": 0.0, "learning_rate": 4.999941787724234e-05, "loss": 1.09, "step": 2272 }, { "epoch": 0.0, "learning_rate": 4.999941734109639e-05, "loss": 1.1697, "step": 2273 }, { "epoch": 0.0, "learning_rate": 4.9999416804703646e-05, "loss": 0.9411, "step": 2274 }, { "epoch": 0.0, "learning_rate": 4.9999416268064133e-05, "loss": 1.2827, "step": 2275 }, { "epoch": 0.0, "learning_rate": 4.9999415731177836e-05, "loss": 1.2085, "step": 2276 }, { "epoch": 0.0, "learning_rate": 4.999941519404475e-05, "loss": 1.2183, "step": 2277 }, { "epoch": 0.0, "learning_rate": 4.999941465666488e-05, "loss": 0.8818, "step": 2278 }, { "epoch": 0.0, "learning_rate": 4.999941411903822e-05, "loss": 0.7588, "step": 2279 }, { "epoch": 0.0, "learning_rate": 4.999941358116479e-05, "loss": 1.3433, "step": 2280 }, { "epoch": 0.0, "learning_rate": 4.999941304304457e-05, "loss": 1.3804, "step": 2281 }, { "epoch": 0.0, "learning_rate": 4.9999412504677565e-05, "loss": 1.2872, "step": 2282 }, { "epoch": 0.0, "learning_rate": 4.999941196606378e-05, "loss": 1.1368, "step": 2283 }, { "epoch": 0.0, "learning_rate": 4.9999411427203206e-05, "loss": 1.0265, "step": 2284 }, { "epoch": 0.0, "learning_rate": 4.999941088809585e-05, "loss": 1.0054, "step": 2285 }, { "epoch": 0.0, "learning_rate": 4.9999410348741715e-05, "loss": 1.2381, "step": 2286 }, { "epoch": 0.0, "learning_rate": 4.99994098091408e-05, "loss": 0.956, "step": 2287 }, { "epoch": 0.0, "learning_rate": 4.9999409269293084e-05, "loss": 1.1399, "step": 2288 }, { "epoch": 0.0, "learning_rate": 4.99994087291986e-05, "loss": 1.4057, "step": 2289 }, { "epoch": 0.0, "learning_rate": 4.999940818885732e-05, "loss": 1.1755, "step": 2290 }, { "epoch": 0.0, "learning_rate": 4.999940764826927e-05, "loss": 1.1603, "step": 2291 }, { "epoch": 0.0, "learning_rate": 4.999940710743443e-05, "loss": 1.2242, "step": 2292 }, { "epoch": 0.0, "learning_rate": 4.99994065663528e-05, "loss": 0.7225, "step": 2293 }, { "epoch": 0.0, "learning_rate": 4.99994060250244e-05, "loss": 0.2577, "step": 2294 }, { "epoch": 0.0, "learning_rate": 4.9999405483449206e-05, "loss": 0.1954, "step": 2295 }, { "epoch": 0.0, "learning_rate": 4.9999404941627234e-05, "loss": 0.1391, "step": 2296 }, { "epoch": 0.0, "learning_rate": 4.999940439955848e-05, "loss": 0.1212, "step": 2297 }, { "epoch": 0.0, "learning_rate": 4.999940385724293e-05, "loss": 0.0721, "step": 2298 }, { "epoch": 0.0, "learning_rate": 4.99994033146806e-05, "loss": 0.0758, "step": 2299 }, { "epoch": 0.0, "learning_rate": 4.999940277187149e-05, "loss": 0.0615, "step": 2300 }, { "epoch": 0.0, "learning_rate": 4.99994022288156e-05, "loss": 0.0825, "step": 2301 }, { "epoch": 0.0, "learning_rate": 4.999940168551293e-05, "loss": 0.0567, "step": 2302 }, { "epoch": 0.0, "learning_rate": 4.999940114196347e-05, "loss": 0.0523, "step": 2303 }, { "epoch": 0.0, "learning_rate": 4.999940059816723e-05, "loss": 0.0547, "step": 2304 }, { "epoch": 0.0, "learning_rate": 4.99994000541242e-05, "loss": 0.4968, "step": 2305 }, { "epoch": 0.0, "learning_rate": 4.999939950983439e-05, "loss": 1.3247, "step": 2306 }, { "epoch": 0.0, "learning_rate": 4.99993989652978e-05, "loss": 1.3734, "step": 2307 }, { "epoch": 0.0, "learning_rate": 4.999939842051442e-05, "loss": 1.1558, "step": 2308 }, { "epoch": 0.0, "learning_rate": 4.999939787548426e-05, "loss": 1.2981, "step": 2309 }, { "epoch": 0.0, "learning_rate": 4.999939733020732e-05, "loss": 1.5171, "step": 2310 }, { "epoch": 0.0, "learning_rate": 4.999939678468359e-05, "loss": 1.6943, "step": 2311 }, { "epoch": 0.0, "learning_rate": 4.999939623891308e-05, "loss": 1.5977, "step": 2312 }, { "epoch": 0.0, "learning_rate": 4.999939569289578e-05, "loss": 1.1567, "step": 2313 }, { "epoch": 0.0, "learning_rate": 4.99993951466317e-05, "loss": 1.3617, "step": 2314 }, { "epoch": 0.0, "learning_rate": 4.999939460012084e-05, "loss": 1.2513, "step": 2315 }, { "epoch": 0.0, "learning_rate": 4.9999394053363196e-05, "loss": 1.3635, "step": 2316 }, { "epoch": 0.0, "learning_rate": 4.999939350635877e-05, "loss": 1.0933, "step": 2317 }, { "epoch": 0.0, "learning_rate": 4.9999392959107555e-05, "loss": 1.3159, "step": 2318 }, { "epoch": 0.0, "learning_rate": 4.999939241160956e-05, "loss": 1.1969, "step": 2319 }, { "epoch": 0.0, "learning_rate": 4.999939186386478e-05, "loss": 1.2352, "step": 2320 }, { "epoch": 0.0, "learning_rate": 4.999939131587322e-05, "loss": 1.3023, "step": 2321 }, { "epoch": 0.0, "learning_rate": 4.999939076763487e-05, "loss": 1.5029, "step": 2322 }, { "epoch": 0.0, "learning_rate": 4.9999390219149745e-05, "loss": 1.002, "step": 2323 }, { "epoch": 0.0, "learning_rate": 4.999938967041783e-05, "loss": 0.6181, "step": 2324 }, { "epoch": 0.0, "learning_rate": 4.999938912143913e-05, "loss": 0.8718, "step": 2325 }, { "epoch": 0.0, "learning_rate": 4.999938857221366e-05, "loss": 0.7674, "step": 2326 }, { "epoch": 0.0, "learning_rate": 4.9999388022741386e-05, "loss": 1.4088, "step": 2327 }, { "epoch": 0.0, "learning_rate": 4.999938747302234e-05, "loss": 1.239, "step": 2328 }, { "epoch": 0.0, "learning_rate": 4.9999386923056515e-05, "loss": 0.9995, "step": 2329 }, { "epoch": 0.0, "learning_rate": 4.99993863728439e-05, "loss": 1.0201, "step": 2330 }, { "epoch": 0.0, "learning_rate": 4.9999385822384505e-05, "loss": 1.0667, "step": 2331 }, { "epoch": 0.0, "learning_rate": 4.999938527167832e-05, "loss": 1.2822, "step": 2332 }, { "epoch": 0.0, "learning_rate": 4.999938472072536e-05, "loss": 0.7923, "step": 2333 }, { "epoch": 0.0, "learning_rate": 4.999938416952561e-05, "loss": 1.1033, "step": 2334 }, { "epoch": 0.0, "learning_rate": 4.999938361807908e-05, "loss": 1.1808, "step": 2335 }, { "epoch": 0.0, "learning_rate": 4.999938306638576e-05, "loss": 1.1299, "step": 2336 }, { "epoch": 0.0, "learning_rate": 4.999938251444567e-05, "loss": 1.1362, "step": 2337 }, { "epoch": 0.0, "learning_rate": 4.999938196225879e-05, "loss": 1.1539, "step": 2338 }, { "epoch": 0.0, "learning_rate": 4.9999381409825124e-05, "loss": 0.6446, "step": 2339 }, { "epoch": 0.0, "learning_rate": 4.999938085714468e-05, "loss": 1.214, "step": 2340 }, { "epoch": 0.0, "learning_rate": 4.9999380304217446e-05, "loss": 1.4017, "step": 2341 }, { "epoch": 0.0, "learning_rate": 4.999937975104343e-05, "loss": 1.9942, "step": 2342 }, { "epoch": 0.0, "learning_rate": 4.9999379197622636e-05, "loss": 1.5023, "step": 2343 }, { "epoch": 0.0, "learning_rate": 4.999937864395505e-05, "loss": 1.3173, "step": 2344 }, { "epoch": 0.0, "learning_rate": 4.9999378090040694e-05, "loss": 1.2286, "step": 2345 }, { "epoch": 0.0, "learning_rate": 4.9999377535879546e-05, "loss": 1.4841, "step": 2346 }, { "epoch": 0.0, "learning_rate": 4.999937698147161e-05, "loss": 0.9893, "step": 2347 }, { "epoch": 0.0, "learning_rate": 4.99993764268169e-05, "loss": 1.1911, "step": 2348 }, { "epoch": 0.0, "learning_rate": 4.99993758719154e-05, "loss": 1.119, "step": 2349 }, { "epoch": 0.0, "learning_rate": 4.999937531676712e-05, "loss": 1.0328, "step": 2350 }, { "epoch": 0.0, "learning_rate": 4.9999374761372055e-05, "loss": 1.1877, "step": 2351 }, { "epoch": 0.0, "learning_rate": 4.999937420573021e-05, "loss": 1.46, "step": 2352 }, { "epoch": 0.0, "learning_rate": 4.999937364984158e-05, "loss": 0.8874, "step": 2353 }, { "epoch": 0.0, "learning_rate": 4.9999373093706165e-05, "loss": 1.2046, "step": 2354 }, { "epoch": 0.0, "learning_rate": 4.999937253732397e-05, "loss": 0.9197, "step": 2355 }, { "epoch": 0.0, "learning_rate": 4.9999371980694985e-05, "loss": 1.5218, "step": 2356 }, { "epoch": 0.0, "learning_rate": 4.9999371423819225e-05, "loss": 0.7522, "step": 2357 }, { "epoch": 0.0, "learning_rate": 4.9999370866696674e-05, "loss": 1.3164, "step": 2358 }, { "epoch": 0.0, "learning_rate": 4.999937030932735e-05, "loss": 1.1041, "step": 2359 }, { "epoch": 0.0, "learning_rate": 4.999936975171124e-05, "loss": 1.2685, "step": 2360 }, { "epoch": 0.0, "learning_rate": 4.999936919384833e-05, "loss": 1.3134, "step": 2361 }, { "epoch": 0.0, "learning_rate": 4.999936863573866e-05, "loss": 1.016, "step": 2362 }, { "epoch": 0.0, "learning_rate": 4.999936807738219e-05, "loss": 0.9835, "step": 2363 }, { "epoch": 0.0, "learning_rate": 4.999936751877895e-05, "loss": 0.7403, "step": 2364 }, { "epoch": 0.0, "learning_rate": 4.9999366959928915e-05, "loss": 0.5867, "step": 2365 }, { "epoch": 0.0, "learning_rate": 4.9999366400832106e-05, "loss": 0.8516, "step": 2366 }, { "epoch": 0.0, "learning_rate": 4.999936584148851e-05, "loss": 1.3296, "step": 2367 }, { "epoch": 0.0, "learning_rate": 4.9999365281898134e-05, "loss": 1.0289, "step": 2368 }, { "epoch": 0.0, "learning_rate": 4.999936472206097e-05, "loss": 0.4003, "step": 2369 }, { "epoch": 0.0, "learning_rate": 4.999936416197702e-05, "loss": 1.0497, "step": 2370 }, { "epoch": 0.0, "learning_rate": 4.9999363601646296e-05, "loss": 1.4006, "step": 2371 }, { "epoch": 0.0, "learning_rate": 4.999936304106878e-05, "loss": 0.778, "step": 2372 }, { "epoch": 0.0, "learning_rate": 4.999936248024448e-05, "loss": 0.7985, "step": 2373 }, { "epoch": 0.0, "learning_rate": 4.99993619191734e-05, "loss": 1.0902, "step": 2374 }, { "epoch": 0.0, "learning_rate": 4.9999361357855544e-05, "loss": 1.1798, "step": 2375 }, { "epoch": 0.0, "learning_rate": 4.999936079629091e-05, "loss": 1.3368, "step": 2376 }, { "epoch": 0.0, "learning_rate": 4.9999360234479474e-05, "loss": 1.3867, "step": 2377 }, { "epoch": 0.0, "learning_rate": 4.999935967242126e-05, "loss": 1.1832, "step": 2378 }, { "epoch": 0.0, "learning_rate": 4.999935911011627e-05, "loss": 1.3546, "step": 2379 }, { "epoch": 0.0, "learning_rate": 4.999935854756449e-05, "loss": 1.3045, "step": 2380 }, { "epoch": 0.0, "learning_rate": 4.999935798476593e-05, "loss": 1.2272, "step": 2381 }, { "epoch": 0.0, "learning_rate": 4.999935742172059e-05, "loss": 1.219, "step": 2382 }, { "epoch": 0.0, "learning_rate": 4.999935685842846e-05, "loss": 1.1794, "step": 2383 }, { "epoch": 0.0, "learning_rate": 4.9999356294889554e-05, "loss": 0.8872, "step": 2384 }, { "epoch": 0.0, "learning_rate": 4.9999355731103855e-05, "loss": 1.0011, "step": 2385 }, { "epoch": 0.0, "learning_rate": 4.9999355167071385e-05, "loss": 1.196, "step": 2386 }, { "epoch": 0.0, "learning_rate": 4.9999354602792124e-05, "loss": 1.1468, "step": 2387 }, { "epoch": 0.0, "learning_rate": 4.9999354038266084e-05, "loss": 0.9644, "step": 2388 }, { "epoch": 0.0, "learning_rate": 4.999935347349326e-05, "loss": 0.9631, "step": 2389 }, { "epoch": 0.0, "learning_rate": 4.999935290847365e-05, "loss": 1.2356, "step": 2390 }, { "epoch": 0.0, "learning_rate": 4.999935234320726e-05, "loss": 1.1133, "step": 2391 }, { "epoch": 0.0, "learning_rate": 4.999935177769408e-05, "loss": 1.3155, "step": 2392 }, { "epoch": 0.0, "learning_rate": 4.999935121193413e-05, "loss": 1.4621, "step": 2393 }, { "epoch": 0.0, "learning_rate": 4.999935064592739e-05, "loss": 0.9106, "step": 2394 }, { "epoch": 0.0, "learning_rate": 4.9999350079673855e-05, "loss": 1.099, "step": 2395 }, { "epoch": 0.0, "learning_rate": 4.999934951317356e-05, "loss": 1.2296, "step": 2396 }, { "epoch": 0.0, "learning_rate": 4.999934894642646e-05, "loss": 0.8653, "step": 2397 }, { "epoch": 0.0, "learning_rate": 4.9999348379432595e-05, "loss": 1.285, "step": 2398 }, { "epoch": 0.0, "learning_rate": 4.999934781219193e-05, "loss": 1.5817, "step": 2399 }, { "epoch": 0.0, "learning_rate": 4.9999347244704494e-05, "loss": 1.3223, "step": 2400 }, { "epoch": 0.0, "learning_rate": 4.999934667697027e-05, "loss": 1.2201, "step": 2401 }, { "epoch": 0.0, "learning_rate": 4.999934610898927e-05, "loss": 1.035, "step": 2402 }, { "epoch": 0.0, "learning_rate": 4.9999345540761484e-05, "loss": 1.4577, "step": 2403 }, { "epoch": 0.0, "learning_rate": 4.999934497228691e-05, "loss": 0.9728, "step": 2404 }, { "epoch": 0.0, "learning_rate": 4.9999344403565556e-05, "loss": 0.9041, "step": 2405 }, { "epoch": 0.0, "learning_rate": 4.9999343834597425e-05, "loss": 0.8214, "step": 2406 }, { "epoch": 0.0, "learning_rate": 4.99993432653825e-05, "loss": 0.936, "step": 2407 }, { "epoch": 0.0, "learning_rate": 4.9999342695920795e-05, "loss": 0.9635, "step": 2408 }, { "epoch": 0.0, "learning_rate": 4.999934212621231e-05, "loss": 1.1556, "step": 2409 }, { "epoch": 0.0, "learning_rate": 4.999934155625704e-05, "loss": 1.254, "step": 2410 }, { "epoch": 0.0, "learning_rate": 4.9999340986054985e-05, "loss": 1.209, "step": 2411 }, { "epoch": 0.0, "learning_rate": 4.999934041560615e-05, "loss": 1.0671, "step": 2412 }, { "epoch": 0.0, "learning_rate": 4.999933984491054e-05, "loss": 0.8076, "step": 2413 }, { "epoch": 0.0, "learning_rate": 4.999933927396814e-05, "loss": 1.0531, "step": 2414 }, { "epoch": 0.0, "learning_rate": 4.999933870277895e-05, "loss": 0.9312, "step": 2415 }, { "epoch": 0.0, "learning_rate": 4.999933813134299e-05, "loss": 1.3744, "step": 2416 }, { "epoch": 0.0, "learning_rate": 4.9999337559660244e-05, "loss": 1.0359, "step": 2417 }, { "epoch": 0.0, "learning_rate": 4.99993369877307e-05, "loss": 0.8227, "step": 2418 }, { "epoch": 0.0, "learning_rate": 4.999933641555439e-05, "loss": 1.1006, "step": 2419 }, { "epoch": 0.0, "learning_rate": 4.9999335843131286e-05, "loss": 0.8073, "step": 2420 }, { "epoch": 0.0, "learning_rate": 4.999933527046141e-05, "loss": 1.2168, "step": 2421 }, { "epoch": 0.0, "learning_rate": 4.9999334697544745e-05, "loss": 0.9359, "step": 2422 }, { "epoch": 0.0, "learning_rate": 4.99993341243813e-05, "loss": 1.2773, "step": 2423 }, { "epoch": 0.0, "learning_rate": 4.999933355097107e-05, "loss": 1.1156, "step": 2424 }, { "epoch": 0.0, "learning_rate": 4.9999332977314063e-05, "loss": 1.1494, "step": 2425 }, { "epoch": 0.0, "learning_rate": 4.999933240341026e-05, "loss": 0.8976, "step": 2426 }, { "epoch": 0.0, "learning_rate": 4.999933182925969e-05, "loss": 1.162, "step": 2427 }, { "epoch": 0.0, "learning_rate": 4.9999331254862326e-05, "loss": 0.8506, "step": 2428 }, { "epoch": 0.0, "learning_rate": 4.999933068021818e-05, "loss": 0.6927, "step": 2429 }, { "epoch": 0.0, "learning_rate": 4.999933010532725e-05, "loss": 1.1664, "step": 2430 }, { "epoch": 0.0, "learning_rate": 4.999932953018954e-05, "loss": 0.9944, "step": 2431 }, { "epoch": 0.0, "learning_rate": 4.9999328954805046e-05, "loss": 1.021, "step": 2432 }, { "epoch": 0.0, "learning_rate": 4.999932837917377e-05, "loss": 0.7567, "step": 2433 }, { "epoch": 0.0, "learning_rate": 4.999932780329572e-05, "loss": 1.3012, "step": 2434 }, { "epoch": 0.0, "learning_rate": 4.999932722717088e-05, "loss": 1.0829, "step": 2435 }, { "epoch": 0.0, "learning_rate": 4.999932665079925e-05, "loss": 1.6098, "step": 2436 }, { "epoch": 0.0, "learning_rate": 4.999932607418084e-05, "loss": 1.2366, "step": 2437 }, { "epoch": 0.0, "learning_rate": 4.999932549731565e-05, "loss": 1.191, "step": 2438 }, { "epoch": 0.0, "learning_rate": 4.9999324920203685e-05, "loss": 1.2385, "step": 2439 }, { "epoch": 0.0, "learning_rate": 4.999932434284492e-05, "loss": 0.7776, "step": 2440 }, { "epoch": 0.0, "learning_rate": 4.999932376523939e-05, "loss": 0.5032, "step": 2441 }, { "epoch": 0.0, "learning_rate": 4.999932318738707e-05, "loss": 0.4156, "step": 2442 }, { "epoch": 0.0, "learning_rate": 4.9999322609287964e-05, "loss": 0.4252, "step": 2443 }, { "epoch": 0.0, "learning_rate": 4.9999322030942074e-05, "loss": 0.3237, "step": 2444 }, { "epoch": 0.0, "learning_rate": 4.9999321452349406e-05, "loss": 0.452, "step": 2445 }, { "epoch": 0.0, "learning_rate": 4.999932087350996e-05, "loss": 0.4207, "step": 2446 }, { "epoch": 0.0, "learning_rate": 4.999932029442372e-05, "loss": 0.2601, "step": 2447 }, { "epoch": 0.0, "learning_rate": 4.9999319715090706e-05, "loss": 0.4093, "step": 2448 }, { "epoch": 0.0, "learning_rate": 4.9999319135510905e-05, "loss": 1.6132, "step": 2449 }, { "epoch": 0.0, "learning_rate": 4.999931855568432e-05, "loss": 1.3601, "step": 2450 }, { "epoch": 0.0, "learning_rate": 4.999931797561096e-05, "loss": 1.1905, "step": 2451 }, { "epoch": 0.0, "learning_rate": 4.999931739529081e-05, "loss": 1.187, "step": 2452 }, { "epoch": 0.0, "learning_rate": 4.9999316814723876e-05, "loss": 1.2807, "step": 2453 }, { "epoch": 0.0, "learning_rate": 4.999931623391016e-05, "loss": 1.5622, "step": 2454 }, { "epoch": 0.0, "learning_rate": 4.9999315652849663e-05, "loss": 1.0686, "step": 2455 }, { "epoch": 0.0, "learning_rate": 4.999931507154238e-05, "loss": 1.002, "step": 2456 }, { "epoch": 0.0, "learning_rate": 4.9999314489988325e-05, "loss": 1.3571, "step": 2457 }, { "epoch": 0.0, "learning_rate": 4.9999313908187476e-05, "loss": 1.0784, "step": 2458 }, { "epoch": 0.0, "learning_rate": 4.9999313326139855e-05, "loss": 1.0493, "step": 2459 }, { "epoch": 0.0, "learning_rate": 4.9999312743845436e-05, "loss": 1.3248, "step": 2460 }, { "epoch": 0.0, "learning_rate": 4.9999312161304245e-05, "loss": 1.2138, "step": 2461 }, { "epoch": 0.0, "learning_rate": 4.999931157851627e-05, "loss": 1.3235, "step": 2462 }, { "epoch": 0.0, "learning_rate": 4.999931099548151e-05, "loss": 1.1581, "step": 2463 }, { "epoch": 0.0, "learning_rate": 4.999931041219997e-05, "loss": 1.0672, "step": 2464 }, { "epoch": 0.0, "learning_rate": 4.999930982867165e-05, "loss": 1.2655, "step": 2465 }, { "epoch": 0.0, "learning_rate": 4.999930924489654e-05, "loss": 1.191, "step": 2466 }, { "epoch": 0.0, "learning_rate": 4.999930866087466e-05, "loss": 1.1727, "step": 2467 }, { "epoch": 0.0, "learning_rate": 4.999930807660598e-05, "loss": 1.3226, "step": 2468 }, { "epoch": 0.0, "learning_rate": 4.999930749209053e-05, "loss": 1.5273, "step": 2469 }, { "epoch": 0.0, "learning_rate": 4.9999306907328294e-05, "loss": 1.1624, "step": 2470 }, { "epoch": 0.0, "learning_rate": 4.999930632231927e-05, "loss": 0.857, "step": 2471 }, { "epoch": 0.0, "learning_rate": 4.9999305737063474e-05, "loss": 1.1461, "step": 2472 }, { "epoch": 0.0, "learning_rate": 4.999930515156089e-05, "loss": 1.1581, "step": 2473 }, { "epoch": 0.0, "learning_rate": 4.999930456581152e-05, "loss": 1.0932, "step": 2474 }, { "epoch": 0.0, "learning_rate": 4.999930397981537e-05, "loss": 1.0128, "step": 2475 }, { "epoch": 0.0, "learning_rate": 4.999930339357244e-05, "loss": 1.2064, "step": 2476 }, { "epoch": 0.0, "learning_rate": 4.9999302807082726e-05, "loss": 1.0275, "step": 2477 }, { "epoch": 0.0, "learning_rate": 4.999930222034622e-05, "loss": 0.8178, "step": 2478 }, { "epoch": 0.0, "learning_rate": 4.999930163336295e-05, "loss": 1.0451, "step": 2479 }, { "epoch": 0.0, "learning_rate": 4.999930104613289e-05, "loss": 1.4416, "step": 2480 }, { "epoch": 0.0, "learning_rate": 4.999930045865604e-05, "loss": 0.8916, "step": 2481 }, { "epoch": 0.0, "learning_rate": 4.999929987093241e-05, "loss": 1.195, "step": 2482 }, { "epoch": 0.0, "learning_rate": 4.9999299282962e-05, "loss": 1.3613, "step": 2483 }, { "epoch": 0.0, "learning_rate": 4.999929869474481e-05, "loss": 1.3498, "step": 2484 }, { "epoch": 0.0, "learning_rate": 4.999929810628083e-05, "loss": 1.0009, "step": 2485 }, { "epoch": 0.0, "learning_rate": 4.9999297517570076e-05, "loss": 1.3047, "step": 2486 }, { "epoch": 0.0, "learning_rate": 4.999929692861253e-05, "loss": 0.9789, "step": 2487 }, { "epoch": 0.0, "learning_rate": 4.9999296339408206e-05, "loss": 1.2014, "step": 2488 }, { "epoch": 0.0, "learning_rate": 4.9999295749957104e-05, "loss": 1.2294, "step": 2489 }, { "epoch": 0.0, "learning_rate": 4.999929516025922e-05, "loss": 1.0622, "step": 2490 }, { "epoch": 0.0, "learning_rate": 4.9999294570314545e-05, "loss": 1.1546, "step": 2491 }, { "epoch": 0.0, "learning_rate": 4.999929398012309e-05, "loss": 1.2043, "step": 2492 }, { "epoch": 0.0, "learning_rate": 4.9999293389684855e-05, "loss": 1.0917, "step": 2493 }, { "epoch": 0.0, "learning_rate": 4.999929279899984e-05, "loss": 1.0515, "step": 2494 }, { "epoch": 0.0, "learning_rate": 4.999929220806804e-05, "loss": 1.1921, "step": 2495 }, { "epoch": 0.0, "learning_rate": 4.999929161688946e-05, "loss": 1.1185, "step": 2496 }, { "epoch": 0.0, "learning_rate": 4.999929102546409e-05, "loss": 1.4151, "step": 2497 }, { "epoch": 0.0, "learning_rate": 4.9999290433791946e-05, "loss": 1.1235, "step": 2498 }, { "epoch": 0.0, "learning_rate": 4.999928984187301e-05, "loss": 0.8122, "step": 2499 }, { "epoch": 0.0, "learning_rate": 4.9999289249707296e-05, "loss": 1.1508, "step": 2500 }, { "epoch": 0.0, "eval_loss": 1.0854716300964355, "eval_runtime": 83.847, "eval_samples_per_second": 16.518, "eval_steps_per_second": 4.138, "step": 2500 }, { "epoch": 0.0, "learning_rate": 4.9999288657294804e-05, "loss": 1.0851, "step": 2501 }, { "epoch": 0.0, "learning_rate": 4.999928806463553e-05, "loss": 0.8544, "step": 2502 }, { "epoch": 0.0, "learning_rate": 4.999928747172946e-05, "loss": 0.7983, "step": 2503 }, { "epoch": 0.0, "learning_rate": 4.9999286878576626e-05, "loss": 1.1961, "step": 2504 }, { "epoch": 0.0, "learning_rate": 4.9999286285176995e-05, "loss": 1.0891, "step": 2505 }, { "epoch": 0.0, "learning_rate": 4.9999285691530586e-05, "loss": 1.0302, "step": 2506 }, { "epoch": 0.0, "learning_rate": 4.99992850976374e-05, "loss": 1.1088, "step": 2507 }, { "epoch": 0.0, "learning_rate": 4.999928450349743e-05, "loss": 1.0297, "step": 2508 }, { "epoch": 0.0, "learning_rate": 4.999928390911067e-05, "loss": 1.2901, "step": 2509 }, { "epoch": 0.0, "learning_rate": 4.999928331447714e-05, "loss": 1.0585, "step": 2510 }, { "epoch": 0.0, "learning_rate": 4.999928271959682e-05, "loss": 1.1662, "step": 2511 }, { "epoch": 0.0, "learning_rate": 4.9999282124469714e-05, "loss": 0.6689, "step": 2512 }, { "epoch": 0.0, "learning_rate": 4.999928152909583e-05, "loss": 0.6319, "step": 2513 }, { "epoch": 0.0, "learning_rate": 4.999928093347516e-05, "loss": 0.9722, "step": 2514 }, { "epoch": 0.0, "learning_rate": 4.9999280337607715e-05, "loss": 0.8079, "step": 2515 }, { "epoch": 0.0, "learning_rate": 4.999927974149348e-05, "loss": 1.1531, "step": 2516 }, { "epoch": 0.0, "learning_rate": 4.999927914513247e-05, "loss": 0.886, "step": 2517 }, { "epoch": 0.0, "learning_rate": 4.999927854852467e-05, "loss": 0.9439, "step": 2518 }, { "epoch": 0.0, "learning_rate": 4.9999277951670096e-05, "loss": 1.2437, "step": 2519 }, { "epoch": 0.0, "learning_rate": 4.9999277354568735e-05, "loss": 1.2504, "step": 2520 }, { "epoch": 0.0, "learning_rate": 4.999927675722059e-05, "loss": 1.299, "step": 2521 }, { "epoch": 0.0, "learning_rate": 4.9999276159625665e-05, "loss": 0.8641, "step": 2522 }, { "epoch": 0.0, "learning_rate": 4.9999275561783956e-05, "loss": 0.6387, "step": 2523 }, { "epoch": 0.0, "learning_rate": 4.999927496369547e-05, "loss": 1.1639, "step": 2524 }, { "epoch": 0.0, "learning_rate": 4.999927436536019e-05, "loss": 1.0768, "step": 2525 }, { "epoch": 0.0, "learning_rate": 4.999927376677814e-05, "loss": 1.2307, "step": 2526 }, { "epoch": 0.0, "learning_rate": 4.99992731679493e-05, "loss": 1.1291, "step": 2527 }, { "epoch": 0.0, "learning_rate": 4.999927256887368e-05, "loss": 1.1474, "step": 2528 }, { "epoch": 0.0, "learning_rate": 4.9999271969551284e-05, "loss": 1.1732, "step": 2529 }, { "epoch": 0.0, "learning_rate": 4.99992713699821e-05, "loss": 1.2119, "step": 2530 }, { "epoch": 0.0, "learning_rate": 4.9999270770166136e-05, "loss": 1.0065, "step": 2531 }, { "epoch": 0.0, "learning_rate": 4.9999270170103385e-05, "loss": 1.0581, "step": 2532 }, { "epoch": 0.0, "learning_rate": 4.9999269569793855e-05, "loss": 1.0729, "step": 2533 }, { "epoch": 0.0, "learning_rate": 4.999926896923755e-05, "loss": 1.1211, "step": 2534 }, { "epoch": 0.0, "learning_rate": 4.999926836843445e-05, "loss": 1.17, "step": 2535 }, { "epoch": 0.0, "learning_rate": 4.999926776738457e-05, "loss": 1.1372, "step": 2536 }, { "epoch": 0.0, "learning_rate": 4.999926716608791e-05, "loss": 1.7751, "step": 2537 }, { "epoch": 0.0, "learning_rate": 4.9999266564544465e-05, "loss": 1.0604, "step": 2538 }, { "epoch": 0.0, "learning_rate": 4.999926596275424e-05, "loss": 1.3684, "step": 2539 }, { "epoch": 0.0, "learning_rate": 4.999926536071724e-05, "loss": 1.2431, "step": 2540 }, { "epoch": 0.0, "learning_rate": 4.999926475843345e-05, "loss": 1.0332, "step": 2541 }, { "epoch": 0.0, "learning_rate": 4.999926415590288e-05, "loss": 1.3016, "step": 2542 }, { "epoch": 0.0, "learning_rate": 4.9999263553125525e-05, "loss": 1.0958, "step": 2543 }, { "epoch": 0.0, "learning_rate": 4.999926295010139e-05, "loss": 1.142, "step": 2544 }, { "epoch": 0.0, "learning_rate": 4.999926234683047e-05, "loss": 1.189, "step": 2545 }, { "epoch": 0.0, "learning_rate": 4.9999261743312774e-05, "loss": 1.0052, "step": 2546 }, { "epoch": 0.0, "learning_rate": 4.999926113954829e-05, "loss": 1.0415, "step": 2547 }, { "epoch": 0.0, "learning_rate": 4.9999260535537026e-05, "loss": 1.2109, "step": 2548 }, { "epoch": 0.0, "learning_rate": 4.999925993127899e-05, "loss": 1.2191, "step": 2549 }, { "epoch": 0.0, "learning_rate": 4.999925932677415e-05, "loss": 1.1117, "step": 2550 }, { "epoch": 0.0, "learning_rate": 4.9999258722022545e-05, "loss": 1.0124, "step": 2551 }, { "epoch": 0.0, "learning_rate": 4.9999258117024154e-05, "loss": 1.0574, "step": 2552 }, { "epoch": 0.0, "learning_rate": 4.999925751177898e-05, "loss": 1.0316, "step": 2553 }, { "epoch": 0.0, "learning_rate": 4.9999256906287016e-05, "loss": 0.5798, "step": 2554 }, { "epoch": 0.0, "learning_rate": 4.999925630054828e-05, "loss": 0.5607, "step": 2555 }, { "epoch": 0.0, "learning_rate": 4.9999255694562766e-05, "loss": 0.5936, "step": 2556 }, { "epoch": 0.0, "learning_rate": 4.999925508833046e-05, "loss": 1.1024, "step": 2557 }, { "epoch": 0.0, "learning_rate": 4.999925448185138e-05, "loss": 1.2206, "step": 2558 }, { "epoch": 0.0, "learning_rate": 4.9999253875125505e-05, "loss": 0.9331, "step": 2559 }, { "epoch": 0.0, "learning_rate": 4.9999253268152856e-05, "loss": 0.9955, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.999925266093343e-05, "loss": 0.8683, "step": 2561 }, { "epoch": 0.0, "learning_rate": 4.999925205346721e-05, "loss": 0.6031, "step": 2562 }, { "epoch": 0.0, "learning_rate": 4.999925144575421e-05, "loss": 0.9559, "step": 2563 }, { "epoch": 0.0, "learning_rate": 4.999925083779444e-05, "loss": 0.6049, "step": 2564 }, { "epoch": 0.0, "learning_rate": 4.999925022958788e-05, "loss": 1.2506, "step": 2565 }, { "epoch": 0.0, "learning_rate": 4.9999249621134534e-05, "loss": 1.506, "step": 2566 }, { "epoch": 0.0, "learning_rate": 4.999924901243441e-05, "loss": 1.3293, "step": 2567 }, { "epoch": 0.0, "learning_rate": 4.999924840348751e-05, "loss": 1.9336, "step": 2568 }, { "epoch": 0.0, "learning_rate": 4.999924779429382e-05, "loss": 0.9746, "step": 2569 }, { "epoch": 0.0, "learning_rate": 4.999924718485335e-05, "loss": 0.761, "step": 2570 }, { "epoch": 0.0, "learning_rate": 4.9999246575166095e-05, "loss": 0.308, "step": 2571 }, { "epoch": 0.0, "learning_rate": 4.999924596523206e-05, "loss": 1.4471, "step": 2572 }, { "epoch": 0.0, "learning_rate": 4.9999245355051245e-05, "loss": 1.4313, "step": 2573 }, { "epoch": 0.0, "learning_rate": 4.999924474462365e-05, "loss": 0.8333, "step": 2574 }, { "epoch": 0.0, "learning_rate": 4.999924413394926e-05, "loss": 0.7514, "step": 2575 }, { "epoch": 0.0, "learning_rate": 4.9999243523028105e-05, "loss": 1.001, "step": 2576 }, { "epoch": 0.0, "learning_rate": 4.999924291186016e-05, "loss": 1.1125, "step": 2577 }, { "epoch": 0.0, "learning_rate": 4.9999242300445435e-05, "loss": 1.2343, "step": 2578 }, { "epoch": 0.0, "learning_rate": 4.999924168878393e-05, "loss": 1.1152, "step": 2579 }, { "epoch": 0.0, "learning_rate": 4.999924107687564e-05, "loss": 0.926, "step": 2580 }, { "epoch": 0.0, "learning_rate": 4.9999240464720564e-05, "loss": 1.0037, "step": 2581 }, { "epoch": 0.0, "learning_rate": 4.999923985231871e-05, "loss": 1.2988, "step": 2582 }, { "epoch": 0.0, "learning_rate": 4.9999239239670074e-05, "loss": 0.8706, "step": 2583 }, { "epoch": 0.0, "learning_rate": 4.999923862677466e-05, "loss": 0.9814, "step": 2584 }, { "epoch": 0.0, "learning_rate": 4.999923801363246e-05, "loss": 0.9953, "step": 2585 }, { "epoch": 0.0, "learning_rate": 4.999923740024347e-05, "loss": 0.7982, "step": 2586 }, { "epoch": 0.0, "learning_rate": 4.999923678660771e-05, "loss": 0.2924, "step": 2587 }, { "epoch": 0.0, "learning_rate": 4.999923617272516e-05, "loss": 0.2321, "step": 2588 }, { "epoch": 0.0, "learning_rate": 4.9999235558595836e-05, "loss": 0.2196, "step": 2589 }, { "epoch": 0.0, "learning_rate": 4.999923494421973e-05, "loss": 0.6108, "step": 2590 }, { "epoch": 0.0, "learning_rate": 4.999923432959683e-05, "loss": 1.3284, "step": 2591 }, { "epoch": 0.0, "learning_rate": 4.9999233714727164e-05, "loss": 0.884, "step": 2592 }, { "epoch": 0.0, "learning_rate": 4.9999233099610706e-05, "loss": 0.7834, "step": 2593 }, { "epoch": 0.0, "learning_rate": 4.999923248424747e-05, "loss": 0.7138, "step": 2594 }, { "epoch": 0.0, "learning_rate": 4.999923186863745e-05, "loss": 1.0278, "step": 2595 }, { "epoch": 0.0, "learning_rate": 4.999923125278065e-05, "loss": 1.1555, "step": 2596 }, { "epoch": 0.0, "learning_rate": 4.999923063667706e-05, "loss": 1.1282, "step": 2597 }, { "epoch": 0.0, "learning_rate": 4.9999230020326706e-05, "loss": 1.1194, "step": 2598 }, { "epoch": 0.0, "learning_rate": 4.9999229403729554e-05, "loss": 1.0122, "step": 2599 }, { "epoch": 0.0, "learning_rate": 4.999922878688563e-05, "loss": 1.1518, "step": 2600 }, { "epoch": 0.0, "learning_rate": 4.999922816979492e-05, "loss": 1.2748, "step": 2601 }, { "epoch": 0.0, "learning_rate": 4.999922755245743e-05, "loss": 0.9259, "step": 2602 }, { "epoch": 0.0, "learning_rate": 4.999922693487315e-05, "loss": 0.8864, "step": 2603 }, { "epoch": 0.0, "learning_rate": 4.999922631704209e-05, "loss": 1.0629, "step": 2604 }, { "epoch": 0.0, "learning_rate": 4.999922569896426e-05, "loss": 1.0549, "step": 2605 }, { "epoch": 0.0, "learning_rate": 4.999922508063964e-05, "loss": 1.0743, "step": 2606 }, { "epoch": 0.0, "learning_rate": 4.9999224462068236e-05, "loss": 1.0465, "step": 2607 }, { "epoch": 0.0, "learning_rate": 4.9999223843250054e-05, "loss": 1.3165, "step": 2608 }, { "epoch": 0.0, "learning_rate": 4.999922322418509e-05, "loss": 1.1523, "step": 2609 }, { "epoch": 0.0, "learning_rate": 4.9999222604873344e-05, "loss": 1.0081, "step": 2610 }, { "epoch": 0.0, "learning_rate": 4.9999221985314814e-05, "loss": 0.9723, "step": 2611 }, { "epoch": 0.0, "learning_rate": 4.99992213655095e-05, "loss": 1.3478, "step": 2612 }, { "epoch": 0.0, "learning_rate": 4.999922074545741e-05, "loss": 1.2355, "step": 2613 }, { "epoch": 0.0, "learning_rate": 4.999922012515854e-05, "loss": 1.1177, "step": 2614 }, { "epoch": 0.0, "learning_rate": 4.9999219504612884e-05, "loss": 0.9973, "step": 2615 }, { "epoch": 0.0, "learning_rate": 4.9999218883820445e-05, "loss": 1.0677, "step": 2616 }, { "epoch": 0.0, "learning_rate": 4.999921826278122e-05, "loss": 0.9187, "step": 2617 }, { "epoch": 0.0, "learning_rate": 4.999921764149522e-05, "loss": 1.0991, "step": 2618 }, { "epoch": 0.0, "learning_rate": 4.999921701996244e-05, "loss": 1.2343, "step": 2619 }, { "epoch": 0.0, "learning_rate": 4.999921639818288e-05, "loss": 1.2575, "step": 2620 }, { "epoch": 0.0, "learning_rate": 4.999921577615653e-05, "loss": 1.0961, "step": 2621 }, { "epoch": 0.0, "learning_rate": 4.9999215153883405e-05, "loss": 1.0279, "step": 2622 }, { "epoch": 0.0, "learning_rate": 4.999921453136349e-05, "loss": 1.2034, "step": 2623 }, { "epoch": 0.0, "learning_rate": 4.99992139085968e-05, "loss": 4.8598, "step": 2624 }, { "epoch": 0.0, "learning_rate": 4.999921328558333e-05, "loss": 3.16, "step": 2625 }, { "epoch": 0.0, "learning_rate": 4.9999212662323075e-05, "loss": 1.0854, "step": 2626 }, { "epoch": 0.0, "learning_rate": 4.999921203881604e-05, "loss": 1.1011, "step": 2627 }, { "epoch": 0.0, "learning_rate": 4.999921141506222e-05, "loss": 1.0632, "step": 2628 }, { "epoch": 0.0, "learning_rate": 4.999921079106162e-05, "loss": 1.1298, "step": 2629 }, { "epoch": 0.0, "learning_rate": 4.9999210166814236e-05, "loss": 1.1682, "step": 2630 }, { "epoch": 0.0, "learning_rate": 4.999920954232008e-05, "loss": 1.1885, "step": 2631 }, { "epoch": 0.0, "learning_rate": 4.999920891757913e-05, "loss": 1.7747, "step": 2632 }, { "epoch": 0.0, "learning_rate": 4.999920829259141e-05, "loss": 1.326, "step": 2633 }, { "epoch": 0.0, "learning_rate": 4.9999207667356896e-05, "loss": 1.2759, "step": 2634 }, { "epoch": 0.0, "learning_rate": 4.999920704187561e-05, "loss": 0.9534, "step": 2635 }, { "epoch": 0.0, "learning_rate": 4.9999206416147535e-05, "loss": 1.0857, "step": 2636 }, { "epoch": 0.0, "learning_rate": 4.999920579017269e-05, "loss": 0.8371, "step": 2637 }, { "epoch": 0.0, "learning_rate": 4.999920516395105e-05, "loss": 1.1331, "step": 2638 }, { "epoch": 0.0, "learning_rate": 4.9999204537482634e-05, "loss": 0.9473, "step": 2639 }, { "epoch": 0.0, "learning_rate": 4.9999203910767435e-05, "loss": 1.1575, "step": 2640 }, { "epoch": 0.0, "learning_rate": 4.999920328380546e-05, "loss": 1.06, "step": 2641 }, { "epoch": 0.0, "learning_rate": 4.99992026565967e-05, "loss": 0.4564, "step": 2642 }, { "epoch": 0.0, "learning_rate": 4.999920202914116e-05, "loss": 1.0863, "step": 2643 }, { "epoch": 0.0, "learning_rate": 4.999920140143883e-05, "loss": 1.8009, "step": 2644 }, { "epoch": 0.0, "learning_rate": 4.9999200773489726e-05, "loss": 1.1431, "step": 2645 }, { "epoch": 0.0, "learning_rate": 4.999920014529383e-05, "loss": 0.7587, "step": 2646 }, { "epoch": 0.0, "learning_rate": 4.999919951685117e-05, "loss": 0.8565, "step": 2647 }, { "epoch": 0.0, "learning_rate": 4.999919888816171e-05, "loss": 1.1586, "step": 2648 }, { "epoch": 0.0, "learning_rate": 4.999919825922549e-05, "loss": 0.9109, "step": 2649 }, { "epoch": 0.0, "learning_rate": 4.999919763004247e-05, "loss": 1.2484, "step": 2650 }, { "epoch": 0.0, "learning_rate": 4.9999197000612675e-05, "loss": 1.5153, "step": 2651 }, { "epoch": 0.0, "learning_rate": 4.999919637093609e-05, "loss": 1.4671, "step": 2652 }, { "epoch": 0.0, "learning_rate": 4.999919574101274e-05, "loss": 1.179, "step": 2653 }, { "epoch": 0.0, "learning_rate": 4.99991951108426e-05, "loss": 1.284, "step": 2654 }, { "epoch": 0.0, "learning_rate": 4.999919448042568e-05, "loss": 0.8949, "step": 2655 }, { "epoch": 0.0, "learning_rate": 4.999919384976198e-05, "loss": 1.2341, "step": 2656 }, { "epoch": 0.0, "learning_rate": 4.9999193218851496e-05, "loss": 1.2149, "step": 2657 }, { "epoch": 0.0, "learning_rate": 4.9999192587694226e-05, "loss": 0.8877, "step": 2658 }, { "epoch": 0.0, "learning_rate": 4.999919195629018e-05, "loss": 0.9975, "step": 2659 }, { "epoch": 0.0, "learning_rate": 4.999919132463935e-05, "loss": 1.095, "step": 2660 }, { "epoch": 0.0, "learning_rate": 4.9999190692741735e-05, "loss": 0.7553, "step": 2661 }, { "epoch": 0.0, "learning_rate": 4.999919006059735e-05, "loss": 0.8629, "step": 2662 }, { "epoch": 0.0, "learning_rate": 4.9999189428206174e-05, "loss": 0.904, "step": 2663 }, { "epoch": 0.0, "learning_rate": 4.9999188795568216e-05, "loss": 1.3531, "step": 2664 }, { "epoch": 0.0, "learning_rate": 4.999918816268349e-05, "loss": 1.2525, "step": 2665 }, { "epoch": 0.0, "learning_rate": 4.999918752955197e-05, "loss": 0.8583, "step": 2666 }, { "epoch": 0.0, "learning_rate": 4.999918689617367e-05, "loss": 1.2082, "step": 2667 }, { "epoch": 0.0, "learning_rate": 4.999918626254859e-05, "loss": 1.3743, "step": 2668 }, { "epoch": 0.0, "learning_rate": 4.999918562867672e-05, "loss": 1.4376, "step": 2669 }, { "epoch": 0.0, "learning_rate": 4.999918499455808e-05, "loss": 1.577, "step": 2670 }, { "epoch": 0.0, "learning_rate": 4.999918436019266e-05, "loss": 1.177, "step": 2671 }, { "epoch": 0.0, "learning_rate": 4.999918372558046e-05, "loss": 1.1328, "step": 2672 }, { "epoch": 0.0, "learning_rate": 4.999918309072147e-05, "loss": 1.1072, "step": 2673 }, { "epoch": 0.0, "learning_rate": 4.999918245561569e-05, "loss": 1.086, "step": 2674 }, { "epoch": 0.0, "learning_rate": 4.9999181820263144e-05, "loss": 1.1498, "step": 2675 }, { "epoch": 0.0, "learning_rate": 4.999918118466381e-05, "loss": 1.0412, "step": 2676 }, { "epoch": 0.0, "learning_rate": 4.9999180548817704e-05, "loss": 1.4574, "step": 2677 }, { "epoch": 0.0, "learning_rate": 4.99991799127248e-05, "loss": 0.992, "step": 2678 }, { "epoch": 0.0, "learning_rate": 4.999917927638513e-05, "loss": 0.3997, "step": 2679 }, { "epoch": 0.0, "learning_rate": 4.999917863979867e-05, "loss": 0.8342, "step": 2680 }, { "epoch": 0.0, "learning_rate": 4.9999178002965436e-05, "loss": 1.1407, "step": 2681 }, { "epoch": 0.0, "learning_rate": 4.999917736588542e-05, "loss": 1.3863, "step": 2682 }, { "epoch": 0.0, "learning_rate": 4.999917672855861e-05, "loss": 1.0188, "step": 2683 }, { "epoch": 0.0, "learning_rate": 4.9999176090985035e-05, "loss": 1.1638, "step": 2684 }, { "epoch": 0.0, "learning_rate": 4.999917545316467e-05, "loss": 1.0254, "step": 2685 }, { "epoch": 0.0, "learning_rate": 4.9999174815097524e-05, "loss": 1.1862, "step": 2686 }, { "epoch": 0.0, "learning_rate": 4.999917417678359e-05, "loss": 1.2064, "step": 2687 }, { "epoch": 0.0, "learning_rate": 4.999917353822289e-05, "loss": 0.6489, "step": 2688 }, { "epoch": 0.0, "learning_rate": 4.99991728994154e-05, "loss": 0.8799, "step": 2689 }, { "epoch": 0.0, "learning_rate": 4.999917226036113e-05, "loss": 1.1395, "step": 2690 }, { "epoch": 0.0, "learning_rate": 4.9999171621060076e-05, "loss": 1.3707, "step": 2691 }, { "epoch": 0.0, "learning_rate": 4.999917098151225e-05, "loss": 1.0209, "step": 2692 }, { "epoch": 0.0, "learning_rate": 4.9999170341717626e-05, "loss": 1.2067, "step": 2693 }, { "epoch": 0.0, "learning_rate": 4.999916970167623e-05, "loss": 1.3174, "step": 2694 }, { "epoch": 0.0, "learning_rate": 4.999916906138806e-05, "loss": 1.1166, "step": 2695 }, { "epoch": 0.0, "learning_rate": 4.99991684208531e-05, "loss": 1.2999, "step": 2696 }, { "epoch": 0.0, "learning_rate": 4.999916778007136e-05, "loss": 1.196, "step": 2697 }, { "epoch": 0.0, "learning_rate": 4.9999167139042834e-05, "loss": 1.3003, "step": 2698 }, { "epoch": 0.0, "learning_rate": 4.999916649776754e-05, "loss": 1.1829, "step": 2699 }, { "epoch": 0.0, "learning_rate": 4.999916585624546e-05, "loss": 1.0895, "step": 2700 }, { "epoch": 0.0, "learning_rate": 4.999916521447659e-05, "loss": 0.8053, "step": 2701 }, { "epoch": 0.0, "learning_rate": 4.9999164572460945e-05, "loss": 1.4009, "step": 2702 }, { "epoch": 0.0, "learning_rate": 4.999916393019852e-05, "loss": 1.1801, "step": 2703 }, { "epoch": 0.0, "learning_rate": 4.999916328768931e-05, "loss": 1.2571, "step": 2704 }, { "epoch": 0.0, "learning_rate": 4.9999162644933315e-05, "loss": 1.1818, "step": 2705 }, { "epoch": 0.0, "learning_rate": 4.999916200193055e-05, "loss": 1.4611, "step": 2706 }, { "epoch": 0.0, "learning_rate": 4.9999161358680995e-05, "loss": 1.211, "step": 2707 }, { "epoch": 0.0, "learning_rate": 4.999916071518467e-05, "loss": 1.0912, "step": 2708 }, { "epoch": 0.0, "learning_rate": 4.9999160071441555e-05, "loss": 1.0762, "step": 2709 }, { "epoch": 0.0, "learning_rate": 4.9999159427451655e-05, "loss": 1.2525, "step": 2710 }, { "epoch": 0.0, "learning_rate": 4.999915878321498e-05, "loss": 1.1884, "step": 2711 }, { "epoch": 0.0, "learning_rate": 4.999915813873152e-05, "loss": 1.2574, "step": 2712 }, { "epoch": 0.0, "learning_rate": 4.999915749400128e-05, "loss": 1.4895, "step": 2713 }, { "epoch": 0.0, "learning_rate": 4.999915684902427e-05, "loss": 0.8249, "step": 2714 }, { "epoch": 0.0, "learning_rate": 4.9999156203800464e-05, "loss": 0.9618, "step": 2715 }, { "epoch": 0.0, "learning_rate": 4.999915555832988e-05, "loss": 1.293, "step": 2716 }, { "epoch": 0.0, "learning_rate": 4.9999154912612516e-05, "loss": 1.3095, "step": 2717 }, { "epoch": 0.0, "learning_rate": 4.999915426664837e-05, "loss": 1.0918, "step": 2718 }, { "epoch": 0.0, "learning_rate": 4.999915362043744e-05, "loss": 1.1069, "step": 2719 }, { "epoch": 0.0, "learning_rate": 4.999915297397974e-05, "loss": 1.0256, "step": 2720 }, { "epoch": 0.0, "learning_rate": 4.999915232727526e-05, "loss": 1.2754, "step": 2721 }, { "epoch": 0.0, "learning_rate": 4.999915168032399e-05, "loss": 1.1649, "step": 2722 }, { "epoch": 0.0, "learning_rate": 4.9999151033125934e-05, "loss": 1.2403, "step": 2723 }, { "epoch": 0.0, "learning_rate": 4.99991503856811e-05, "loss": 0.715, "step": 2724 }, { "epoch": 0.0, "learning_rate": 4.999914973798949e-05, "loss": 1.0529, "step": 2725 }, { "epoch": 0.0, "learning_rate": 4.99991490900511e-05, "loss": 1.1878, "step": 2726 }, { "epoch": 0.0, "learning_rate": 4.999914844186592e-05, "loss": 0.9524, "step": 2727 }, { "epoch": 0.0, "learning_rate": 4.9999147793433965e-05, "loss": 0.9358, "step": 2728 }, { "epoch": 0.0, "learning_rate": 4.999914714475523e-05, "loss": 0.5873, "step": 2729 }, { "epoch": 0.0, "learning_rate": 4.9999146495829716e-05, "loss": 0.9812, "step": 2730 }, { "epoch": 0.0, "learning_rate": 4.999914584665741e-05, "loss": 1.1639, "step": 2731 }, { "epoch": 0.0, "learning_rate": 4.999914519723834e-05, "loss": 1.1829, "step": 2732 }, { "epoch": 0.0, "learning_rate": 4.999914454757247e-05, "loss": 1.0936, "step": 2733 }, { "epoch": 0.0, "learning_rate": 4.999914389765984e-05, "loss": 1.0867, "step": 2734 }, { "epoch": 0.0, "learning_rate": 4.999914324750041e-05, "loss": 0.9176, "step": 2735 }, { "epoch": 0.0, "learning_rate": 4.999914259709421e-05, "loss": 1.2482, "step": 2736 }, { "epoch": 0.0, "learning_rate": 4.999914194644123e-05, "loss": 1.1445, "step": 2737 }, { "epoch": 0.0, "learning_rate": 4.999914129554146e-05, "loss": 1.02, "step": 2738 }, { "epoch": 0.0, "learning_rate": 4.9999140644394914e-05, "loss": 1.0444, "step": 2739 }, { "epoch": 0.0, "learning_rate": 4.999913999300159e-05, "loss": 2.032, "step": 2740 }, { "epoch": 0.0, "learning_rate": 4.999913934136148e-05, "loss": 1.6777, "step": 2741 }, { "epoch": 0.0, "learning_rate": 4.999913868947459e-05, "loss": 1.1249, "step": 2742 }, { "epoch": 0.0, "learning_rate": 4.999913803734092e-05, "loss": 1.2135, "step": 2743 }, { "epoch": 0.0, "learning_rate": 4.999913738496047e-05, "loss": 1.1091, "step": 2744 }, { "epoch": 0.0, "learning_rate": 4.999913673233323e-05, "loss": 0.9304, "step": 2745 }, { "epoch": 0.0, "learning_rate": 4.999913607945922e-05, "loss": 1.0457, "step": 2746 }, { "epoch": 0.0, "learning_rate": 4.999913542633843e-05, "loss": 1.1601, "step": 2747 }, { "epoch": 0.0, "learning_rate": 4.999913477297085e-05, "loss": 1.213, "step": 2748 }, { "epoch": 0.0, "learning_rate": 4.99991341193565e-05, "loss": 1.1778, "step": 2749 }, { "epoch": 0.0, "learning_rate": 4.999913346549536e-05, "loss": 1.1547, "step": 2750 }, { "epoch": 0.0, "learning_rate": 4.999913281138744e-05, "loss": 1.2153, "step": 2751 }, { "epoch": 0.0, "learning_rate": 4.999913215703275e-05, "loss": 1.0718, "step": 2752 }, { "epoch": 0.0, "learning_rate": 4.9999131502431266e-05, "loss": 1.0972, "step": 2753 }, { "epoch": 0.0, "learning_rate": 4.9999130847583006e-05, "loss": 1.3415, "step": 2754 }, { "epoch": 0.0, "learning_rate": 4.999913019248796e-05, "loss": 1.0715, "step": 2755 }, { "epoch": 0.0, "learning_rate": 4.999912953714614e-05, "loss": 0.9399, "step": 2756 }, { "epoch": 0.0, "learning_rate": 4.999912888155753e-05, "loss": 1.0309, "step": 2757 }, { "epoch": 0.0, "learning_rate": 4.9999128225722155e-05, "loss": 1.2394, "step": 2758 }, { "epoch": 0.0, "learning_rate": 4.9999127569639985e-05, "loss": 1.0969, "step": 2759 }, { "epoch": 0.0, "learning_rate": 4.9999126913311044e-05, "loss": 1.1535, "step": 2760 }, { "epoch": 0.0, "learning_rate": 4.999912625673532e-05, "loss": 1.1519, "step": 2761 }, { "epoch": 0.0, "learning_rate": 4.999912559991281e-05, "loss": 1.7157, "step": 2762 }, { "epoch": 0.0, "learning_rate": 4.999912494284352e-05, "loss": 1.1947, "step": 2763 }, { "epoch": 0.0, "learning_rate": 4.999912428552745e-05, "loss": 0.9904, "step": 2764 }, { "epoch": 0.0, "learning_rate": 4.99991236279646e-05, "loss": 0.857, "step": 2765 }, { "epoch": 0.0, "learning_rate": 4.999912297015497e-05, "loss": 1.1261, "step": 2766 }, { "epoch": 0.0, "learning_rate": 4.999912231209855e-05, "loss": 1.2156, "step": 2767 }, { "epoch": 0.0, "learning_rate": 4.999912165379537e-05, "loss": 1.2385, "step": 2768 }, { "epoch": 0.0, "learning_rate": 4.99991209952454e-05, "loss": 1.002, "step": 2769 }, { "epoch": 0.0, "learning_rate": 4.999912033644864e-05, "loss": 1.1895, "step": 2770 }, { "epoch": 0.0, "learning_rate": 4.9999119677405105e-05, "loss": 1.2197, "step": 2771 }, { "epoch": 0.0, "learning_rate": 4.999911901811479e-05, "loss": 1.1403, "step": 2772 }, { "epoch": 0.0, "learning_rate": 4.999911835857769e-05, "loss": 1.2452, "step": 2773 }, { "epoch": 0.0, "learning_rate": 4.999911769879382e-05, "loss": 0.8898, "step": 2774 }, { "epoch": 0.0, "learning_rate": 4.9999117038763155e-05, "loss": 0.8207, "step": 2775 }, { "epoch": 0.0, "learning_rate": 4.999911637848572e-05, "loss": 0.6392, "step": 2776 }, { "epoch": 0.0, "learning_rate": 4.99991157179615e-05, "loss": 0.7881, "step": 2777 }, { "epoch": 0.0, "learning_rate": 4.99991150571905e-05, "loss": 1.3896, "step": 2778 }, { "epoch": 0.0, "learning_rate": 4.999911439617272e-05, "loss": 1.2835, "step": 2779 }, { "epoch": 0.0, "learning_rate": 4.9999113734908156e-05, "loss": 1.1221, "step": 2780 }, { "epoch": 0.0, "learning_rate": 4.999911307339682e-05, "loss": 0.4653, "step": 2781 }, { "epoch": 0.0, "learning_rate": 4.999911241163869e-05, "loss": 0.9905, "step": 2782 }, { "epoch": 0.0, "learning_rate": 4.9999111749633785e-05, "loss": 1.1892, "step": 2783 }, { "epoch": 0.0, "learning_rate": 4.9999111087382105e-05, "loss": 1.2269, "step": 2784 }, { "epoch": 0.0, "learning_rate": 4.999911042488365e-05, "loss": 1.7996, "step": 2785 }, { "epoch": 0.0, "learning_rate": 4.99991097621384e-05, "loss": 1.5628, "step": 2786 }, { "epoch": 0.0, "learning_rate": 4.999910909914637e-05, "loss": 0.7233, "step": 2787 }, { "epoch": 0.0, "learning_rate": 4.999910843590756e-05, "loss": 0.6555, "step": 2788 }, { "epoch": 0.0, "learning_rate": 4.999910777242197e-05, "loss": 1.121, "step": 2789 }, { "epoch": 0.0, "learning_rate": 4.99991071086896e-05, "loss": 1.4148, "step": 2790 }, { "epoch": 0.0, "learning_rate": 4.9999106444710454e-05, "loss": 1.2223, "step": 2791 }, { "epoch": 0.0, "learning_rate": 4.9999105780484524e-05, "loss": 1.1463, "step": 2792 }, { "epoch": 0.0, "learning_rate": 4.9999105116011815e-05, "loss": 1.0044, "step": 2793 }, { "epoch": 0.0, "learning_rate": 4.999910445129232e-05, "loss": 1.2547, "step": 2794 }, { "epoch": 0.0, "learning_rate": 4.999910378632605e-05, "loss": 1.0682, "step": 2795 }, { "epoch": 0.0, "learning_rate": 4.9999103121113e-05, "loss": 1.0591, "step": 2796 }, { "epoch": 0.0, "learning_rate": 4.999910245565317e-05, "loss": 1.3869, "step": 2797 }, { "epoch": 0.0, "learning_rate": 4.9999101789946555e-05, "loss": 1.2375, "step": 2798 }, { "epoch": 0.0, "learning_rate": 4.9999101123993165e-05, "loss": 1.1, "step": 2799 }, { "epoch": 0.0, "learning_rate": 4.999910045779299e-05, "loss": 0.9864, "step": 2800 }, { "epoch": 0.0, "learning_rate": 4.999909979134603e-05, "loss": 1.2769, "step": 2801 }, { "epoch": 0.0, "learning_rate": 4.9999099124652294e-05, "loss": 1.2261, "step": 2802 }, { "epoch": 0.0, "learning_rate": 4.999909845771178e-05, "loss": 1.41, "step": 2803 }, { "epoch": 0.0, "learning_rate": 4.9999097790524485e-05, "loss": 1.2601, "step": 2804 }, { "epoch": 0.0, "learning_rate": 4.9999097123090413e-05, "loss": 1.0879, "step": 2805 }, { "epoch": 0.0, "learning_rate": 4.999909645540955e-05, "loss": 1.2381, "step": 2806 }, { "epoch": 0.0, "learning_rate": 4.999909578748191e-05, "loss": 1.0534, "step": 2807 }, { "epoch": 0.0, "learning_rate": 4.999909511930749e-05, "loss": 1.2444, "step": 2808 }, { "epoch": 0.0, "learning_rate": 4.9999094450886294e-05, "loss": 1.0326, "step": 2809 }, { "epoch": 0.0, "learning_rate": 4.9999093782218306e-05, "loss": 1.0756, "step": 2810 }, { "epoch": 0.0, "learning_rate": 4.999909311330355e-05, "loss": 0.9788, "step": 2811 }, { "epoch": 0.0, "learning_rate": 4.999909244414201e-05, "loss": 1.0639, "step": 2812 }, { "epoch": 0.0, "learning_rate": 4.999909177473369e-05, "loss": 1.063, "step": 2813 }, { "epoch": 0.0, "learning_rate": 4.999909110507859e-05, "loss": 1.05, "step": 2814 }, { "epoch": 0.0, "learning_rate": 4.99990904351767e-05, "loss": 0.9695, "step": 2815 }, { "epoch": 0.0, "learning_rate": 4.999908976502804e-05, "loss": 0.9567, "step": 2816 }, { "epoch": 0.0, "learning_rate": 4.99990890946326e-05, "loss": 1.1835, "step": 2817 }, { "epoch": 0.0, "learning_rate": 4.999908842399037e-05, "loss": 1.1377, "step": 2818 }, { "epoch": 0.0, "learning_rate": 4.999908775310137e-05, "loss": 1.3342, "step": 2819 }, { "epoch": 0.0, "learning_rate": 4.999908708196559e-05, "loss": 0.7403, "step": 2820 }, { "epoch": 0.0, "learning_rate": 4.999908641058302e-05, "loss": 1.149, "step": 2821 }, { "epoch": 0.0, "learning_rate": 4.999908573895367e-05, "loss": 1.3792, "step": 2822 }, { "epoch": 0.0, "learning_rate": 4.999908506707754e-05, "loss": 1.1136, "step": 2823 }, { "epoch": 0.0, "learning_rate": 4.999908439495464e-05, "loss": 1.1038, "step": 2824 }, { "epoch": 0.0, "learning_rate": 4.999908372258495e-05, "loss": 1.0534, "step": 2825 }, { "epoch": 0.0, "learning_rate": 4.999908304996849e-05, "loss": 1.4179, "step": 2826 }, { "epoch": 0.0, "learning_rate": 4.999908237710523e-05, "loss": 0.6761, "step": 2827 }, { "epoch": 0.0, "learning_rate": 4.999908170399521e-05, "loss": 1.2538, "step": 2828 }, { "epoch": 0.0, "learning_rate": 4.99990810306384e-05, "loss": 0.8459, "step": 2829 }, { "epoch": 0.0, "learning_rate": 4.999908035703481e-05, "loss": 1.1891, "step": 2830 }, { "epoch": 0.0, "learning_rate": 4.999907968318444e-05, "loss": 1.0343, "step": 2831 }, { "epoch": 0.0, "learning_rate": 4.999907900908729e-05, "loss": 0.8952, "step": 2832 }, { "epoch": 0.0, "learning_rate": 4.999907833474336e-05, "loss": 1.2053, "step": 2833 }, { "epoch": 0.0, "learning_rate": 4.999907766015265e-05, "loss": 0.9051, "step": 2834 }, { "epoch": 0.0, "learning_rate": 4.999907698531516e-05, "loss": 1.2077, "step": 2835 }, { "epoch": 0.0, "learning_rate": 4.999907631023089e-05, "loss": 1.2348, "step": 2836 }, { "epoch": 0.0, "learning_rate": 4.999907563489984e-05, "loss": 0.9451, "step": 2837 }, { "epoch": 0.0, "learning_rate": 4.9999074959322e-05, "loss": 1.019, "step": 2838 }, { "epoch": 0.0, "learning_rate": 4.99990742834974e-05, "loss": 1.1164, "step": 2839 }, { "epoch": 0.0, "learning_rate": 4.9999073607425995e-05, "loss": 0.9338, "step": 2840 }, { "epoch": 0.0, "learning_rate": 4.999907293110783e-05, "loss": 1.0339, "step": 2841 }, { "epoch": 0.0, "learning_rate": 4.999907225454287e-05, "loss": 0.8994, "step": 2842 }, { "epoch": 0.0, "learning_rate": 4.9999071577731135e-05, "loss": 0.8476, "step": 2843 }, { "epoch": 0.0, "learning_rate": 4.999907090067262e-05, "loss": 0.8179, "step": 2844 }, { "epoch": 0.0, "learning_rate": 4.999907022336733e-05, "loss": 0.8631, "step": 2845 }, { "epoch": 0.0, "learning_rate": 4.999906954581526e-05, "loss": 1.1398, "step": 2846 }, { "epoch": 0.0, "learning_rate": 4.99990688680164e-05, "loss": 0.8859, "step": 2847 }, { "epoch": 0.0, "learning_rate": 4.999906818997077e-05, "loss": 1.5999, "step": 2848 }, { "epoch": 0.0, "learning_rate": 4.999906751167835e-05, "loss": 1.3751, "step": 2849 }, { "epoch": 0.0, "learning_rate": 4.9999066833139155e-05, "loss": 1.2111, "step": 2850 }, { "epoch": 0.0, "learning_rate": 4.999906615435318e-05, "loss": 1.1603, "step": 2851 }, { "epoch": 0.0, "learning_rate": 4.9999065475320424e-05, "loss": 1.0992, "step": 2852 }, { "epoch": 0.0, "learning_rate": 4.999906479604088e-05, "loss": 1.5599, "step": 2853 }, { "epoch": 0.0, "learning_rate": 4.9999064116514574e-05, "loss": 1.2482, "step": 2854 }, { "epoch": 0.0, "learning_rate": 4.999906343674147e-05, "loss": 0.7927, "step": 2855 }, { "epoch": 0.0, "learning_rate": 4.999906275672159e-05, "loss": 1.3003, "step": 2856 }, { "epoch": 0.0, "learning_rate": 4.999906207645494e-05, "loss": 0.9895, "step": 2857 }, { "epoch": 0.0, "learning_rate": 4.99990613959415e-05, "loss": 1.076, "step": 2858 }, { "epoch": 0.0, "learning_rate": 4.9999060715181276e-05, "loss": 1.2395, "step": 2859 }, { "epoch": 0.0, "learning_rate": 4.999906003417428e-05, "loss": 1.2853, "step": 2860 }, { "epoch": 0.0, "learning_rate": 4.999905935292051e-05, "loss": 0.9141, "step": 2861 }, { "epoch": 0.0, "learning_rate": 4.999905867141995e-05, "loss": 1.4543, "step": 2862 }, { "epoch": 0.0, "learning_rate": 4.999905798967261e-05, "loss": 0.7374, "step": 2863 }, { "epoch": 0.0, "learning_rate": 4.999905730767849e-05, "loss": 1.238, "step": 2864 }, { "epoch": 0.0, "learning_rate": 4.999905662543759e-05, "loss": 1.0451, "step": 2865 }, { "epoch": 0.0, "learning_rate": 4.999905594294991e-05, "loss": 1.1267, "step": 2866 }, { "epoch": 0.0, "learning_rate": 4.999905526021546e-05, "loss": 1.1472, "step": 2867 }, { "epoch": 0.0, "learning_rate": 4.999905457723422e-05, "loss": 1.1766, "step": 2868 }, { "epoch": 0.0, "learning_rate": 4.99990538940062e-05, "loss": 1.3368, "step": 2869 }, { "epoch": 0.0, "learning_rate": 4.99990532105314e-05, "loss": 1.1428, "step": 2870 }, { "epoch": 0.0, "learning_rate": 4.999905252680982e-05, "loss": 1.3337, "step": 2871 }, { "epoch": 0.0, "learning_rate": 4.999905184284146e-05, "loss": 1.0391, "step": 2872 }, { "epoch": 0.0, "learning_rate": 4.999905115862632e-05, "loss": 1.5314, "step": 2873 }, { "epoch": 0.0, "learning_rate": 4.99990504741644e-05, "loss": 1.2044, "step": 2874 }, { "epoch": 0.0, "learning_rate": 4.99990497894557e-05, "loss": 1.0157, "step": 2875 }, { "epoch": 0.0, "learning_rate": 4.999904910450022e-05, "loss": 1.0111, "step": 2876 }, { "epoch": 0.0, "learning_rate": 4.9999048419297956e-05, "loss": 0.7298, "step": 2877 }, { "epoch": 0.0, "learning_rate": 4.999904773384892e-05, "loss": 0.9956, "step": 2878 }, { "epoch": 0.0, "learning_rate": 4.99990470481531e-05, "loss": 0.5304, "step": 2879 }, { "epoch": 0.0, "learning_rate": 4.99990463622105e-05, "loss": 0.5684, "step": 2880 }, { "epoch": 0.0, "learning_rate": 4.999904567602112e-05, "loss": 1.1679, "step": 2881 }, { "epoch": 0.0, "learning_rate": 4.9999044989584954e-05, "loss": 0.807, "step": 2882 }, { "epoch": 0.0, "learning_rate": 4.999904430290202e-05, "loss": 0.5727, "step": 2883 }, { "epoch": 0.0, "learning_rate": 4.99990436159723e-05, "loss": 0.6624, "step": 2884 }, { "epoch": 0.0, "learning_rate": 4.999904292879579e-05, "loss": 1.0362, "step": 2885 }, { "epoch": 0.0, "learning_rate": 4.999904224137251e-05, "loss": 1.2741, "step": 2886 }, { "epoch": 0.0, "learning_rate": 4.9999041553702455e-05, "loss": 1.2492, "step": 2887 }, { "epoch": 0.0, "learning_rate": 4.999904086578561e-05, "loss": 1.405, "step": 2888 }, { "epoch": 0.0, "learning_rate": 4.999904017762199e-05, "loss": 0.9898, "step": 2889 }, { "epoch": 0.0, "learning_rate": 4.999903948921159e-05, "loss": 1.2159, "step": 2890 }, { "epoch": 0.0, "learning_rate": 4.999903880055441e-05, "loss": 0.9097, "step": 2891 }, { "epoch": 0.0, "learning_rate": 4.999903811165045e-05, "loss": 1.3631, "step": 2892 }, { "epoch": 0.0, "learning_rate": 4.9999037422499703e-05, "loss": 6.0654, "step": 2893 }, { "epoch": 0.0, "learning_rate": 4.999903673310219e-05, "loss": 5.9507, "step": 2894 }, { "epoch": 0.0, "learning_rate": 4.999903604345789e-05, "loss": 5.8381, "step": 2895 }, { "epoch": 0.0, "learning_rate": 4.999903535356681e-05, "loss": 5.6796, "step": 2896 }, { "epoch": 0.0, "learning_rate": 4.999903466342895e-05, "loss": 5.9238, "step": 2897 }, { "epoch": 0.0, "learning_rate": 4.999903397304431e-05, "loss": 4.1682, "step": 2898 }, { "epoch": 0.0, "learning_rate": 4.999903328241289e-05, "loss": 1.2678, "step": 2899 }, { "epoch": 0.0, "learning_rate": 4.999903259153469e-05, "loss": 1.1465, "step": 2900 }, { "epoch": 0.0, "learning_rate": 4.999903190040971e-05, "loss": 1.0097, "step": 2901 }, { "epoch": 0.0, "learning_rate": 4.999903120903795e-05, "loss": 1.0424, "step": 2902 }, { "epoch": 0.0, "learning_rate": 4.999903051741941e-05, "loss": 1.1693, "step": 2903 }, { "epoch": 0.0, "learning_rate": 4.999902982555409e-05, "loss": 0.7356, "step": 2904 }, { "epoch": 0.0, "learning_rate": 4.999902913344199e-05, "loss": 1.0111, "step": 2905 }, { "epoch": 0.0, "learning_rate": 4.999902844108311e-05, "loss": 1.2294, "step": 2906 }, { "epoch": 0.0, "learning_rate": 4.999902774847745e-05, "loss": 0.8252, "step": 2907 }, { "epoch": 0.0, "learning_rate": 4.9999027055625014e-05, "loss": 0.9026, "step": 2908 }, { "epoch": 0.0, "learning_rate": 4.999902636252579e-05, "loss": 1.0901, "step": 2909 }, { "epoch": 0.0, "learning_rate": 4.999902566917979e-05, "loss": 1.0077, "step": 2910 }, { "epoch": 0.0, "learning_rate": 4.9999024975587015e-05, "loss": 1.1404, "step": 2911 }, { "epoch": 0.0, "learning_rate": 4.999902428174745e-05, "loss": 0.9697, "step": 2912 }, { "epoch": 0.0, "learning_rate": 4.999902358766112e-05, "loss": 1.0877, "step": 2913 }, { "epoch": 0.0, "learning_rate": 4.9999022893327994e-05, "loss": 1.2615, "step": 2914 }, { "epoch": 0.0, "learning_rate": 4.99990221987481e-05, "loss": 1.2332, "step": 2915 }, { "epoch": 0.0, "learning_rate": 4.999902150392142e-05, "loss": 1.2203, "step": 2916 }, { "epoch": 0.0, "learning_rate": 4.9999020808847965e-05, "loss": 1.2718, "step": 2917 }, { "epoch": 0.0, "learning_rate": 4.9999020113527726e-05, "loss": 1.2264, "step": 2918 }, { "epoch": 0.0, "learning_rate": 4.9999019417960716e-05, "loss": 1.016, "step": 2919 }, { "epoch": 0.0, "learning_rate": 4.9999018722146914e-05, "loss": 0.9891, "step": 2920 }, { "epoch": 0.0, "learning_rate": 4.9999018026086334e-05, "loss": 1.0894, "step": 2921 }, { "epoch": 0.0, "learning_rate": 4.999901732977898e-05, "loss": 0.9991, "step": 2922 }, { "epoch": 0.0, "learning_rate": 4.999901663322484e-05, "loss": 1.234, "step": 2923 }, { "epoch": 0.0, "learning_rate": 4.999901593642393e-05, "loss": 1.3263, "step": 2924 }, { "epoch": 0.0, "learning_rate": 4.9999015239376235e-05, "loss": 0.9724, "step": 2925 }, { "epoch": 0.0, "learning_rate": 4.999901454208176e-05, "loss": 1.0791, "step": 2926 }, { "epoch": 0.0, "learning_rate": 4.99990138445405e-05, "loss": 1.2334, "step": 2927 }, { "epoch": 0.0, "learning_rate": 4.999901314675246e-05, "loss": 0.8321, "step": 2928 }, { "epoch": 0.0, "learning_rate": 4.9999012448717655e-05, "loss": 1.2932, "step": 2929 }, { "epoch": 0.0, "learning_rate": 4.999901175043605e-05, "loss": 1.3547, "step": 2930 }, { "epoch": 0.0, "learning_rate": 4.999901105190768e-05, "loss": 0.7176, "step": 2931 }, { "epoch": 0.0, "learning_rate": 4.999901035313252e-05, "loss": 0.7461, "step": 2932 }, { "epoch": 0.0, "learning_rate": 4.9999009654110594e-05, "loss": 1.4159, "step": 2933 }, { "epoch": 0.0, "learning_rate": 4.999900895484188e-05, "loss": 1.0933, "step": 2934 }, { "epoch": 0.0, "learning_rate": 4.999900825532639e-05, "loss": 1.0124, "step": 2935 }, { "epoch": 0.0, "learning_rate": 4.999900755556411e-05, "loss": 1.2458, "step": 2936 }, { "epoch": 0.0, "learning_rate": 4.9999006855555064e-05, "loss": 1.2437, "step": 2937 }, { "epoch": 0.0, "learning_rate": 4.999900615529923e-05, "loss": 0.7774, "step": 2938 }, { "epoch": 0.0, "learning_rate": 4.999900545479662e-05, "loss": 1.2332, "step": 2939 }, { "epoch": 0.0, "learning_rate": 4.999900475404723e-05, "loss": 1.1314, "step": 2940 }, { "epoch": 0.0, "learning_rate": 4.999900405305106e-05, "loss": 0.9918, "step": 2941 }, { "epoch": 0.0, "learning_rate": 4.99990033518081e-05, "loss": 0.5156, "step": 2942 }, { "epoch": 0.0, "learning_rate": 4.9999002650318374e-05, "loss": 0.386, "step": 2943 }, { "epoch": 0.0, "learning_rate": 4.999900194858187e-05, "loss": 0.3806, "step": 2944 }, { "epoch": 0.0, "learning_rate": 4.9999001246598575e-05, "loss": 0.2731, "step": 2945 }, { "epoch": 0.0, "learning_rate": 4.99990005443685e-05, "loss": 0.311, "step": 2946 }, { "epoch": 0.0, "learning_rate": 4.999899984189166e-05, "loss": 0.1946, "step": 2947 }, { "epoch": 0.0, "learning_rate": 4.9998999139168026e-05, "loss": 0.1996, "step": 2948 }, { "epoch": 0.0, "learning_rate": 4.999899843619762e-05, "loss": 0.1449, "step": 2949 }, { "epoch": 0.0, "learning_rate": 4.9998997732980434e-05, "loss": 0.1897, "step": 2950 }, { "epoch": 0.0, "learning_rate": 4.999899702951647e-05, "loss": 0.0818, "step": 2951 }, { "epoch": 0.0, "learning_rate": 4.999899632580572e-05, "loss": 0.153, "step": 2952 }, { "epoch": 0.0, "learning_rate": 4.9998995621848195e-05, "loss": 0.1676, "step": 2953 }, { "epoch": 0.0, "learning_rate": 4.999899491764389e-05, "loss": 0.0627, "step": 2954 }, { "epoch": 0.0, "learning_rate": 4.99989942131928e-05, "loss": 0.1163, "step": 2955 }, { "epoch": 0.0, "learning_rate": 4.999899350849494e-05, "loss": 0.1337, "step": 2956 }, { "epoch": 0.0, "learning_rate": 4.999899280355029e-05, "loss": 0.139, "step": 2957 }, { "epoch": 0.0, "learning_rate": 4.999899209835887e-05, "loss": 0.0755, "step": 2958 }, { "epoch": 0.0, "learning_rate": 4.999899139292067e-05, "loss": 0.0569, "step": 2959 }, { "epoch": 0.0, "learning_rate": 4.999899068723568e-05, "loss": 0.0567, "step": 2960 }, { "epoch": 0.0, "learning_rate": 4.999898998130392e-05, "loss": 0.2035, "step": 2961 }, { "epoch": 0.0, "learning_rate": 4.999898927512538e-05, "loss": 0.2997, "step": 2962 }, { "epoch": 0.0, "learning_rate": 4.9998988568700056e-05, "loss": 0.3259, "step": 2963 }, { "epoch": 0.0, "learning_rate": 4.9998987862027956e-05, "loss": 0.1894, "step": 2964 }, { "epoch": 0.0, "learning_rate": 4.999898715510908e-05, "loss": 0.0832, "step": 2965 }, { "epoch": 0.0, "learning_rate": 4.999898644794342e-05, "loss": 1.3086, "step": 2966 }, { "epoch": 0.0, "learning_rate": 4.999898574053098e-05, "loss": 1.2232, "step": 2967 }, { "epoch": 0.0, "learning_rate": 4.999898503287176e-05, "loss": 0.9243, "step": 2968 }, { "epoch": 0.0, "learning_rate": 4.999898432496576e-05, "loss": 1.2614, "step": 2969 }, { "epoch": 0.0, "learning_rate": 4.9998983616812984e-05, "loss": 1.134, "step": 2970 }, { "epoch": 0.0, "learning_rate": 4.9998982908413425e-05, "loss": 1.37, "step": 2971 }, { "epoch": 0.0, "learning_rate": 4.999898219976709e-05, "loss": 1.1254, "step": 2972 }, { "epoch": 0.0, "learning_rate": 4.999898149087397e-05, "loss": 1.1534, "step": 2973 }, { "epoch": 0.0, "learning_rate": 4.999898078173408e-05, "loss": 1.2074, "step": 2974 }, { "epoch": 0.0, "learning_rate": 4.99989800723474e-05, "loss": 0.8483, "step": 2975 }, { "epoch": 0.0, "learning_rate": 4.999897936271395e-05, "loss": 1.3476, "step": 2976 }, { "epoch": 0.0, "learning_rate": 4.999897865283372e-05, "loss": 1.29, "step": 2977 }, { "epoch": 0.0, "learning_rate": 4.9998977942706705e-05, "loss": 1.2638, "step": 2978 }, { "epoch": 0.0, "learning_rate": 4.999897723233291e-05, "loss": 0.9091, "step": 2979 }, { "epoch": 0.0, "learning_rate": 4.999897652171234e-05, "loss": 1.1177, "step": 2980 }, { "epoch": 0.0, "learning_rate": 4.999897581084499e-05, "loss": 1.0874, "step": 2981 }, { "epoch": 0.0, "learning_rate": 4.999897509973086e-05, "loss": 1.0801, "step": 2982 }, { "epoch": 0.0, "learning_rate": 4.999897438836995e-05, "loss": 1.0494, "step": 2983 }, { "epoch": 0.0, "learning_rate": 4.999897367676226e-05, "loss": 1.1651, "step": 2984 }, { "epoch": 0.0, "learning_rate": 4.999897296490779e-05, "loss": 0.8595, "step": 2985 }, { "epoch": 0.0, "learning_rate": 4.999897225280654e-05, "loss": 0.8869, "step": 2986 }, { "epoch": 0.0, "learning_rate": 4.9998971540458514e-05, "loss": 1.2285, "step": 2987 }, { "epoch": 0.0, "learning_rate": 4.999897082786371e-05, "loss": 1.128, "step": 2988 }, { "epoch": 0.0, "learning_rate": 4.9998970115022126e-05, "loss": 1.3749, "step": 2989 }, { "epoch": 0.0, "learning_rate": 4.999896940193376e-05, "loss": 1.3221, "step": 2990 }, { "epoch": 0.0, "learning_rate": 4.999896868859861e-05, "loss": 1.254, "step": 2991 }, { "epoch": 0.0, "learning_rate": 4.9998967975016695e-05, "loss": 1.1455, "step": 2992 }, { "epoch": 0.0, "learning_rate": 4.999896726118799e-05, "loss": 1.0821, "step": 2993 }, { "epoch": 0.0, "learning_rate": 4.9998966547112506e-05, "loss": 1.1394, "step": 2994 }, { "epoch": 0.0, "learning_rate": 4.999896583279025e-05, "loss": 0.8994, "step": 2995 }, { "epoch": 0.0, "learning_rate": 4.9998965118221206e-05, "loss": 1.2445, "step": 2996 }, { "epoch": 0.0, "learning_rate": 4.9998964403405386e-05, "loss": 1.1261, "step": 2997 }, { "epoch": 0.0, "learning_rate": 4.999896368834279e-05, "loss": 1.1851, "step": 2998 }, { "epoch": 0.0, "learning_rate": 4.999896297303341e-05, "loss": 1.5682, "step": 2999 }, { "epoch": 0.0, "learning_rate": 4.9998962257477256e-05, "loss": 0.9643, "step": 3000 }, { "epoch": 0.0, "eval_loss": 1.0737282037734985, "eval_runtime": 85.3847, "eval_samples_per_second": 16.221, "eval_steps_per_second": 4.064, "step": 3000 }, { "epoch": 0.0, "learning_rate": 4.999896154167432e-05, "loss": 1.2074, "step": 3001 }, { "epoch": 0.0, "learning_rate": 4.9998960825624593e-05, "loss": 1.3158, "step": 3002 }, { "epoch": 0.0, "learning_rate": 4.9998960109328105e-05, "loss": 1.0965, "step": 3003 }, { "epoch": 0.0, "learning_rate": 4.9998959392784825e-05, "loss": 1.0788, "step": 3004 }, { "epoch": 0.0, "learning_rate": 4.999895867599478e-05, "loss": 0.8845, "step": 3005 }, { "epoch": 0.0, "learning_rate": 4.9998957958957946e-05, "loss": 0.793, "step": 3006 }, { "epoch": 0.0, "learning_rate": 4.999895724167433e-05, "loss": 0.9951, "step": 3007 }, { "epoch": 0.0, "learning_rate": 4.999895652414393e-05, "loss": 1.0859, "step": 3008 }, { "epoch": 0.0, "learning_rate": 4.999895580636677e-05, "loss": 0.8108, "step": 3009 }, { "epoch": 0.0, "learning_rate": 4.9998955088342816e-05, "loss": 0.5787, "step": 3010 }, { "epoch": 0.0, "learning_rate": 4.999895437007209e-05, "loss": 0.5608, "step": 3011 }, { "epoch": 0.0, "learning_rate": 4.999895365155458e-05, "loss": 0.5967, "step": 3012 }, { "epoch": 0.0, "learning_rate": 4.99989529327903e-05, "loss": 0.4456, "step": 3013 }, { "epoch": 0.0, "learning_rate": 4.999895221377923e-05, "loss": 0.5504, "step": 3014 }, { "epoch": 0.0, "learning_rate": 4.999895149452138e-05, "loss": 0.8343, "step": 3015 }, { "epoch": 0.0, "learning_rate": 4.999895077501676e-05, "loss": 1.6915, "step": 3016 }, { "epoch": 0.0, "learning_rate": 4.999895005526536e-05, "loss": 1.2349, "step": 3017 }, { "epoch": 0.0, "learning_rate": 4.9998949335267174e-05, "loss": 1.139, "step": 3018 }, { "epoch": 0.0, "learning_rate": 4.999894861502221e-05, "loss": 1.1852, "step": 3019 }, { "epoch": 0.0, "learning_rate": 4.999894789453047e-05, "loss": 1.2533, "step": 3020 }, { "epoch": 0.0, "learning_rate": 4.999894717379195e-05, "loss": 1.2747, "step": 3021 }, { "epoch": 0.0, "learning_rate": 4.9998946452806655e-05, "loss": 1.1085, "step": 3022 }, { "epoch": 0.0, "learning_rate": 4.9998945731574574e-05, "loss": 1.2659, "step": 3023 }, { "epoch": 0.0, "learning_rate": 4.999894501009572e-05, "loss": 0.7887, "step": 3024 }, { "epoch": 0.0, "learning_rate": 4.9998944288370084e-05, "loss": 1.1709, "step": 3025 }, { "epoch": 0.0, "learning_rate": 4.999894356639767e-05, "loss": 1.1869, "step": 3026 }, { "epoch": 0.0, "learning_rate": 4.9998942844178475e-05, "loss": 1.1321, "step": 3027 }, { "epoch": 0.0, "learning_rate": 4.99989421217125e-05, "loss": 1.2509, "step": 3028 }, { "epoch": 0.0, "learning_rate": 4.9998941398999754e-05, "loss": 1.3264, "step": 3029 }, { "epoch": 0.0, "learning_rate": 4.999894067604022e-05, "loss": 1.0859, "step": 3030 }, { "epoch": 0.0, "learning_rate": 4.9998939952833915e-05, "loss": 0.9069, "step": 3031 }, { "epoch": 0.0, "learning_rate": 4.9998939229380825e-05, "loss": 1.1984, "step": 3032 }, { "epoch": 0.0, "learning_rate": 4.999893850568095e-05, "loss": 0.9934, "step": 3033 }, { "epoch": 0.0, "learning_rate": 4.9998937781734304e-05, "loss": 0.997, "step": 3034 }, { "epoch": 0.0, "learning_rate": 4.999893705754089e-05, "loss": 1.177, "step": 3035 }, { "epoch": 0.0, "learning_rate": 4.999893633310068e-05, "loss": 1.0475, "step": 3036 }, { "epoch": 0.0, "learning_rate": 4.9998935608413685e-05, "loss": 0.9387, "step": 3037 }, { "epoch": 0.0, "learning_rate": 4.999893488347993e-05, "loss": 1.0551, "step": 3038 }, { "epoch": 0.0, "learning_rate": 4.9998934158299385e-05, "loss": 1.1771, "step": 3039 }, { "epoch": 0.0, "learning_rate": 4.9998933432872065e-05, "loss": 1.2724, "step": 3040 }, { "epoch": 0.0, "learning_rate": 4.9998932707197966e-05, "loss": 1.2089, "step": 3041 }, { "epoch": 0.0, "learning_rate": 4.999893198127708e-05, "loss": 1.0284, "step": 3042 }, { "epoch": 0.0, "learning_rate": 4.999893125510943e-05, "loss": 0.9842, "step": 3043 }, { "epoch": 0.0, "learning_rate": 4.999893052869499e-05, "loss": 1.0936, "step": 3044 }, { "epoch": 0.0, "learning_rate": 4.999892980203378e-05, "loss": 1.0716, "step": 3045 }, { "epoch": 0.0, "learning_rate": 4.9998929075125784e-05, "loss": 1.3936, "step": 3046 }, { "epoch": 0.0, "learning_rate": 4.999892834797101e-05, "loss": 1.2452, "step": 3047 }, { "epoch": 0.0, "learning_rate": 4.999892762056946e-05, "loss": 1.0022, "step": 3048 }, { "epoch": 0.0, "learning_rate": 4.9998926892921124e-05, "loss": 1.0688, "step": 3049 }, { "epoch": 0.0, "learning_rate": 4.999892616502602e-05, "loss": 1.0423, "step": 3050 }, { "epoch": 0.0, "learning_rate": 4.9998925436884126e-05, "loss": 1.1952, "step": 3051 }, { "epoch": 0.0, "learning_rate": 4.999892470849546e-05, "loss": 1.157, "step": 3052 }, { "epoch": 0.0, "learning_rate": 4.9998923979860015e-05, "loss": 1.149, "step": 3053 }, { "epoch": 0.0, "learning_rate": 4.999892325097779e-05, "loss": 0.8852, "step": 3054 }, { "epoch": 0.0, "learning_rate": 4.9998922521848786e-05, "loss": 1.442, "step": 3055 }, { "epoch": 0.0, "learning_rate": 4.9998921792473005e-05, "loss": 1.1477, "step": 3056 }, { "epoch": 0.0, "learning_rate": 4.999892106285044e-05, "loss": 1.0485, "step": 3057 }, { "epoch": 0.0, "learning_rate": 4.99989203329811e-05, "loss": 0.9521, "step": 3058 }, { "epoch": 0.0, "learning_rate": 4.999891960286498e-05, "loss": 1.3072, "step": 3059 }, { "epoch": 0.0, "learning_rate": 4.999891887250208e-05, "loss": 1.0286, "step": 3060 }, { "epoch": 0.0, "learning_rate": 4.99989181418924e-05, "loss": 1.2539, "step": 3061 }, { "epoch": 0.0, "learning_rate": 4.999891741103595e-05, "loss": 1.1245, "step": 3062 }, { "epoch": 0.0, "learning_rate": 4.999891667993272e-05, "loss": 0.932, "step": 3063 }, { "epoch": 0.0, "learning_rate": 4.9998915948582705e-05, "loss": 1.4209, "step": 3064 }, { "epoch": 0.0, "learning_rate": 4.999891521698591e-05, "loss": 1.5105, "step": 3065 }, { "epoch": 0.0, "learning_rate": 4.999891448514234e-05, "loss": 1.2014, "step": 3066 }, { "epoch": 0.0, "learning_rate": 4.999891375305199e-05, "loss": 1.3503, "step": 3067 }, { "epoch": 0.0, "learning_rate": 4.999891302071486e-05, "loss": 0.9712, "step": 3068 }, { "epoch": 0.0, "learning_rate": 4.999891228813095e-05, "loss": 1.1086, "step": 3069 }, { "epoch": 0.0, "learning_rate": 4.999891155530026e-05, "loss": 1.3851, "step": 3070 }, { "epoch": 0.0, "learning_rate": 4.9998910822222804e-05, "loss": 1.1959, "step": 3071 }, { "epoch": 0.0, "learning_rate": 4.999891008889856e-05, "loss": 0.9626, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.999890935532754e-05, "loss": 1.0227, "step": 3073 }, { "epoch": 0.0, "learning_rate": 4.9998908621509736e-05, "loss": 1.2477, "step": 3074 }, { "epoch": 0.0, "learning_rate": 4.999890788744516e-05, "loss": 1.1234, "step": 3075 }, { "epoch": 0.0, "learning_rate": 4.99989071531338e-05, "loss": 1.2972, "step": 3076 }, { "epoch": 0.0, "learning_rate": 4.9998906418575666e-05, "loss": 1.0932, "step": 3077 }, { "epoch": 0.0, "learning_rate": 4.999890568377075e-05, "loss": 1.2099, "step": 3078 }, { "epoch": 0.0, "learning_rate": 4.9998904948719054e-05, "loss": 1.0578, "step": 3079 }, { "epoch": 0.0, "learning_rate": 4.999890421342058e-05, "loss": 1.0771, "step": 3080 }, { "epoch": 0.0, "learning_rate": 4.999890347787533e-05, "loss": 1.2784, "step": 3081 }, { "epoch": 0.0, "learning_rate": 4.99989027420833e-05, "loss": 1.0203, "step": 3082 }, { "epoch": 0.0, "learning_rate": 4.999890200604449e-05, "loss": 1.3236, "step": 3083 }, { "epoch": 0.0, "learning_rate": 4.99989012697589e-05, "loss": 0.9522, "step": 3084 }, { "epoch": 0.0, "learning_rate": 4.999890053322654e-05, "loss": 0.9891, "step": 3085 }, { "epoch": 0.0, "learning_rate": 4.99988997964474e-05, "loss": 1.0306, "step": 3086 }, { "epoch": 0.0, "learning_rate": 4.999889905942147e-05, "loss": 1.2939, "step": 3087 }, { "epoch": 0.0, "learning_rate": 4.999889832214877e-05, "loss": 1.1802, "step": 3088 }, { "epoch": 0.0, "learning_rate": 4.999889758462929e-05, "loss": 1.0999, "step": 3089 }, { "epoch": 0.0, "learning_rate": 4.999889684686303e-05, "loss": 1.2702, "step": 3090 }, { "epoch": 0.0, "learning_rate": 4.999889610884999e-05, "loss": 1.2539, "step": 3091 }, { "epoch": 0.0, "learning_rate": 4.999889537059018e-05, "loss": 0.94, "step": 3092 }, { "epoch": 0.0, "learning_rate": 4.999889463208358e-05, "loss": 1.1902, "step": 3093 }, { "epoch": 0.0, "learning_rate": 4.999889389333021e-05, "loss": 0.987, "step": 3094 }, { "epoch": 0.0, "learning_rate": 4.999889315433006e-05, "loss": 1.141, "step": 3095 }, { "epoch": 0.0, "learning_rate": 4.9998892415083134e-05, "loss": 1.2143, "step": 3096 }, { "epoch": 0.0, "learning_rate": 4.999889167558942e-05, "loss": 0.8637, "step": 3097 }, { "epoch": 0.0, "learning_rate": 4.999889093584893e-05, "loss": 1.0951, "step": 3098 }, { "epoch": 0.0, "learning_rate": 4.999889019586167e-05, "loss": 1.2597, "step": 3099 }, { "epoch": 0.0, "learning_rate": 4.9998889455627625e-05, "loss": 0.9368, "step": 3100 }, { "epoch": 0.0, "learning_rate": 4.9998888715146794e-05, "loss": 1.5598, "step": 3101 }, { "epoch": 0.0, "learning_rate": 4.999888797441919e-05, "loss": 1.1202, "step": 3102 }, { "epoch": 0.0, "learning_rate": 4.999888723344482e-05, "loss": 1.0475, "step": 3103 }, { "epoch": 0.0, "learning_rate": 4.999888649222366e-05, "loss": 1.0778, "step": 3104 }, { "epoch": 0.0, "learning_rate": 4.9998885750755726e-05, "loss": 1.1118, "step": 3105 }, { "epoch": 0.0, "learning_rate": 4.9998885009041005e-05, "loss": 0.9362, "step": 3106 }, { "epoch": 0.0, "learning_rate": 4.999888426707951e-05, "loss": 1.0847, "step": 3107 }, { "epoch": 0.0, "learning_rate": 4.9998883524871244e-05, "loss": 1.0924, "step": 3108 }, { "epoch": 0.0, "learning_rate": 4.999888278241619e-05, "loss": 0.9675, "step": 3109 }, { "epoch": 0.0, "learning_rate": 4.999888203971437e-05, "loss": 1.0317, "step": 3110 }, { "epoch": 0.0, "learning_rate": 4.999888129676575e-05, "loss": 1.1201, "step": 3111 }, { "epoch": 0.0, "learning_rate": 4.999888055357037e-05, "loss": 1.1086, "step": 3112 }, { "epoch": 0.0, "learning_rate": 4.9998879810128205e-05, "loss": 1.2691, "step": 3113 }, { "epoch": 0.0, "learning_rate": 4.999887906643926e-05, "loss": 1.0789, "step": 3114 }, { "epoch": 0.0, "learning_rate": 4.999887832250354e-05, "loss": 0.8669, "step": 3115 }, { "epoch": 0.0, "learning_rate": 4.9998877578321046e-05, "loss": 0.9003, "step": 3116 }, { "epoch": 0.0, "learning_rate": 4.999887683389177e-05, "loss": 0.9656, "step": 3117 }, { "epoch": 0.0, "learning_rate": 4.9998876089215705e-05, "loss": 0.7094, "step": 3118 }, { "epoch": 0.0, "learning_rate": 4.999887534429287e-05, "loss": 0.9274, "step": 3119 }, { "epoch": 0.0, "learning_rate": 4.999887459912326e-05, "loss": 1.2837, "step": 3120 }, { "epoch": 0.0, "learning_rate": 4.9998873853706875e-05, "loss": 0.894, "step": 3121 }, { "epoch": 0.0, "learning_rate": 4.99988731080437e-05, "loss": 0.8587, "step": 3122 }, { "epoch": 0.0, "learning_rate": 4.9998872362133754e-05, "loss": 1.2183, "step": 3123 }, { "epoch": 0.0, "learning_rate": 4.999887161597703e-05, "loss": 1.1431, "step": 3124 }, { "epoch": 0.0, "learning_rate": 4.999887086957352e-05, "loss": 1.3707, "step": 3125 }, { "epoch": 0.0, "learning_rate": 4.9998870122923235e-05, "loss": 1.0031, "step": 3126 }, { "epoch": 0.0, "learning_rate": 4.999886937602618e-05, "loss": 0.9755, "step": 3127 }, { "epoch": 0.0, "learning_rate": 4.999886862888234e-05, "loss": 1.0886, "step": 3128 }, { "epoch": 0.0, "learning_rate": 4.999886788149172e-05, "loss": 0.8562, "step": 3129 }, { "epoch": 0.0, "learning_rate": 4.999886713385432e-05, "loss": 0.464, "step": 3130 }, { "epoch": 0.0, "learning_rate": 4.999886638597015e-05, "loss": 0.6532, "step": 3131 }, { "epoch": 0.0, "learning_rate": 4.99988656378392e-05, "loss": 1.1652, "step": 3132 }, { "epoch": 0.0, "learning_rate": 4.9998864889461466e-05, "loss": 1.2203, "step": 3133 }, { "epoch": 0.0, "learning_rate": 4.9998864140836956e-05, "loss": 1.025, "step": 3134 }, { "epoch": 0.0, "learning_rate": 4.999886339196567e-05, "loss": 1.2002, "step": 3135 }, { "epoch": 0.0, "learning_rate": 4.999886264284761e-05, "loss": 1.3056, "step": 3136 }, { "epoch": 0.0, "learning_rate": 4.9998861893482764e-05, "loss": 1.0095, "step": 3137 }, { "epoch": 0.0, "learning_rate": 4.9998861143871135e-05, "loss": 1.3763, "step": 3138 }, { "epoch": 0.0, "learning_rate": 4.9998860394012734e-05, "loss": 1.0908, "step": 3139 }, { "epoch": 0.0, "learning_rate": 4.9998859643907556e-05, "loss": 1.0629, "step": 3140 }, { "epoch": 0.0, "learning_rate": 4.99988588935556e-05, "loss": 1.3534, "step": 3141 }, { "epoch": 0.0, "learning_rate": 4.9998858142956866e-05, "loss": 0.9142, "step": 3142 }, { "epoch": 0.0, "learning_rate": 4.9998857392111354e-05, "loss": 0.9733, "step": 3143 }, { "epoch": 0.0, "learning_rate": 4.9998856641019064e-05, "loss": 1.0786, "step": 3144 }, { "epoch": 0.0, "learning_rate": 4.9998855889679996e-05, "loss": 0.9734, "step": 3145 }, { "epoch": 0.0, "learning_rate": 4.999885513809415e-05, "loss": 1.4096, "step": 3146 }, { "epoch": 0.0, "learning_rate": 4.999885438626152e-05, "loss": 1.1417, "step": 3147 }, { "epoch": 0.0, "learning_rate": 4.999885363418212e-05, "loss": 1.0368, "step": 3148 }, { "epoch": 0.0, "learning_rate": 4.999885288185593e-05, "loss": 1.1928, "step": 3149 }, { "epoch": 0.0, "learning_rate": 4.9998852129282966e-05, "loss": 0.9801, "step": 3150 }, { "epoch": 0.0, "learning_rate": 4.999885137646324e-05, "loss": 0.9239, "step": 3151 }, { "epoch": 0.0, "learning_rate": 4.999885062339672e-05, "loss": 0.866, "step": 3152 }, { "epoch": 0.0, "learning_rate": 4.999884987008342e-05, "loss": 0.8846, "step": 3153 }, { "epoch": 0.0, "learning_rate": 4.999884911652335e-05, "loss": 0.8873, "step": 3154 }, { "epoch": 0.0, "learning_rate": 4.99988483627165e-05, "loss": 1.2745, "step": 3155 }, { "epoch": 0.0, "learning_rate": 4.999884760866287e-05, "loss": 1.1213, "step": 3156 }, { "epoch": 0.0, "learning_rate": 4.9998846854362465e-05, "loss": 0.9286, "step": 3157 }, { "epoch": 0.0, "learning_rate": 4.9998846099815276e-05, "loss": 0.4891, "step": 3158 }, { "epoch": 0.0, "learning_rate": 4.999884534502132e-05, "loss": 1.1473, "step": 3159 }, { "epoch": 0.0, "learning_rate": 4.999884458998057e-05, "loss": 0.6707, "step": 3160 }, { "epoch": 0.0, "learning_rate": 4.999884383469306e-05, "loss": 1.0229, "step": 3161 }, { "epoch": 0.0, "learning_rate": 4.999884307915875e-05, "loss": 1.3011, "step": 3162 }, { "epoch": 0.0, "learning_rate": 4.999884232337768e-05, "loss": 0.8993, "step": 3163 }, { "epoch": 0.0, "learning_rate": 4.999884156734983e-05, "loss": 1.1166, "step": 3164 }, { "epoch": 0.0, "learning_rate": 4.9998840811075195e-05, "loss": 1.1444, "step": 3165 }, { "epoch": 0.0, "learning_rate": 4.999884005455378e-05, "loss": 1.1135, "step": 3166 }, { "epoch": 0.0, "learning_rate": 4.999883929778559e-05, "loss": 0.9868, "step": 3167 }, { "epoch": 0.0, "learning_rate": 4.999883854077063e-05, "loss": 1.1363, "step": 3168 }, { "epoch": 0.0, "learning_rate": 4.999883778350889e-05, "loss": 1.2838, "step": 3169 }, { "epoch": 0.0, "learning_rate": 4.999883702600037e-05, "loss": 1.428, "step": 3170 }, { "epoch": 0.0, "learning_rate": 4.9998836268245067e-05, "loss": 1.1863, "step": 3171 }, { "epoch": 0.0, "learning_rate": 4.9998835510242993e-05, "loss": 1.1418, "step": 3172 }, { "epoch": 0.0, "learning_rate": 4.9998834751994136e-05, "loss": 0.9341, "step": 3173 }, { "epoch": 0.0, "learning_rate": 4.99988339934985e-05, "loss": 1.0903, "step": 3174 }, { "epoch": 0.0, "learning_rate": 4.9998833234756086e-05, "loss": 1.192, "step": 3175 }, { "epoch": 0.0, "learning_rate": 4.99988324757669e-05, "loss": 1.1706, "step": 3176 }, { "epoch": 0.0, "learning_rate": 4.999883171653093e-05, "loss": 1.1696, "step": 3177 }, { "epoch": 0.0, "learning_rate": 4.999883095704819e-05, "loss": 1.2754, "step": 3178 }, { "epoch": 0.0, "learning_rate": 4.9998830197318665e-05, "loss": 0.8601, "step": 3179 }, { "epoch": 0.0, "learning_rate": 4.999882943734237e-05, "loss": 1.2475, "step": 3180 }, { "epoch": 0.0, "learning_rate": 4.999882867711928e-05, "loss": 1.006, "step": 3181 }, { "epoch": 0.0, "learning_rate": 4.999882791664943e-05, "loss": 1.1577, "step": 3182 }, { "epoch": 0.0, "learning_rate": 4.999882715593279e-05, "loss": 0.969, "step": 3183 }, { "epoch": 0.0, "learning_rate": 4.999882639496938e-05, "loss": 1.0363, "step": 3184 }, { "epoch": 0.0, "learning_rate": 4.999882563375919e-05, "loss": 1.0267, "step": 3185 }, { "epoch": 0.0, "learning_rate": 4.9998824872302216e-05, "loss": 1.2042, "step": 3186 }, { "epoch": 0.0, "learning_rate": 4.999882411059847e-05, "loss": 1.5013, "step": 3187 }, { "epoch": 0.0, "learning_rate": 4.999882334864795e-05, "loss": 1.2098, "step": 3188 }, { "epoch": 0.0, "learning_rate": 4.999882258645065e-05, "loss": 1.028, "step": 3189 }, { "epoch": 0.0, "learning_rate": 4.999882182400657e-05, "loss": 0.9882, "step": 3190 }, { "epoch": 0.0, "learning_rate": 4.999882106131571e-05, "loss": 1.0897, "step": 3191 }, { "epoch": 0.0, "learning_rate": 4.999882029837808e-05, "loss": 1.2585, "step": 3192 }, { "epoch": 0.0, "learning_rate": 4.9998819535193666e-05, "loss": 1.1727, "step": 3193 }, { "epoch": 0.0, "learning_rate": 4.9998818771762465e-05, "loss": 1.0733, "step": 3194 }, { "epoch": 0.0, "learning_rate": 4.9998818008084505e-05, "loss": 1.1633, "step": 3195 }, { "epoch": 0.0, "learning_rate": 4.9998817244159754e-05, "loss": 1.2635, "step": 3196 }, { "epoch": 0.0, "learning_rate": 4.999881647998823e-05, "loss": 0.9241, "step": 3197 }, { "epoch": 0.0, "learning_rate": 4.9998815715569925e-05, "loss": 1.1062, "step": 3198 }, { "epoch": 0.0, "learning_rate": 4.9998814950904846e-05, "loss": 1.1206, "step": 3199 }, { "epoch": 0.0, "learning_rate": 4.999881418599299e-05, "loss": 1.0057, "step": 3200 }, { "epoch": 0.0, "learning_rate": 4.9998813420834356e-05, "loss": 1.1484, "step": 3201 }, { "epoch": 0.0, "learning_rate": 4.999881265542894e-05, "loss": 1.0997, "step": 3202 }, { "epoch": 0.0, "learning_rate": 4.999881188977675e-05, "loss": 1.0839, "step": 3203 }, { "epoch": 0.0, "learning_rate": 4.999881112387778e-05, "loss": 1.5233, "step": 3204 }, { "epoch": 0.0, "learning_rate": 4.9998810357732026e-05, "loss": 1.1323, "step": 3205 }, { "epoch": 0.0, "learning_rate": 4.999880959133951e-05, "loss": 1.2861, "step": 3206 }, { "epoch": 0.0, "learning_rate": 4.999880882470021e-05, "loss": 0.8339, "step": 3207 }, { "epoch": 0.0, "learning_rate": 4.999880805781413e-05, "loss": 1.2662, "step": 3208 }, { "epoch": 0.0, "learning_rate": 4.999880729068127e-05, "loss": 1.0926, "step": 3209 }, { "epoch": 0.0, "learning_rate": 4.999880652330163e-05, "loss": 0.9723, "step": 3210 }, { "epoch": 0.0, "learning_rate": 4.999880575567522e-05, "loss": 0.9267, "step": 3211 }, { "epoch": 0.0, "learning_rate": 4.999880498780203e-05, "loss": 1.014, "step": 3212 }, { "epoch": 0.0, "learning_rate": 4.999880421968206e-05, "loss": 1.7897, "step": 3213 }, { "epoch": 0.0, "learning_rate": 4.999880345131531e-05, "loss": 0.9724, "step": 3214 }, { "epoch": 0.0, "learning_rate": 4.99988026827018e-05, "loss": 1.209, "step": 3215 }, { "epoch": 0.0, "learning_rate": 4.9998801913841494e-05, "loss": 0.9055, "step": 3216 }, { "epoch": 0.0, "learning_rate": 4.9998801144734406e-05, "loss": 1.1476, "step": 3217 }, { "epoch": 0.0, "learning_rate": 4.999880037538055e-05, "loss": 1.0604, "step": 3218 }, { "epoch": 0.0, "learning_rate": 4.999879960577992e-05, "loss": 1.1971, "step": 3219 }, { "epoch": 0.0, "learning_rate": 4.999879883593251e-05, "loss": 2.9813, "step": 3220 }, { "epoch": 0.0, "learning_rate": 4.999879806583832e-05, "loss": 6.0157, "step": 3221 }, { "epoch": 0.0, "learning_rate": 4.9998797295497356e-05, "loss": 5.9594, "step": 3222 }, { "epoch": 0.0, "learning_rate": 4.9998796524909606e-05, "loss": 2.482, "step": 3223 }, { "epoch": 0.0, "learning_rate": 4.999879575407509e-05, "loss": 1.1952, "step": 3224 }, { "epoch": 0.0, "learning_rate": 4.9998794982993794e-05, "loss": 1.3123, "step": 3225 }, { "epoch": 0.0, "learning_rate": 4.999879421166571e-05, "loss": 1.3332, "step": 3226 }, { "epoch": 0.0, "learning_rate": 4.999879344009085e-05, "loss": 1.144, "step": 3227 }, { "epoch": 0.0, "learning_rate": 4.999879266826922e-05, "loss": 1.2806, "step": 3228 }, { "epoch": 0.0, "learning_rate": 4.999879189620081e-05, "loss": 1.1283, "step": 3229 }, { "epoch": 0.0, "learning_rate": 4.9998791123885624e-05, "loss": 1.2873, "step": 3230 }, { "epoch": 0.0, "learning_rate": 4.999879035132366e-05, "loss": 1.1551, "step": 3231 }, { "epoch": 0.0, "learning_rate": 4.9998789578514914e-05, "loss": 1.0223, "step": 3232 }, { "epoch": 0.0, "learning_rate": 4.99987888054594e-05, "loss": 1.0714, "step": 3233 }, { "epoch": 0.0, "learning_rate": 4.99987880321571e-05, "loss": 1.1004, "step": 3234 }, { "epoch": 0.0, "learning_rate": 4.9998787258608026e-05, "loss": 1.0357, "step": 3235 }, { "epoch": 0.0, "learning_rate": 4.999878648481218e-05, "loss": 0.8907, "step": 3236 }, { "epoch": 0.0, "learning_rate": 4.999878571076955e-05, "loss": 0.9581, "step": 3237 }, { "epoch": 0.0, "learning_rate": 4.999878493648014e-05, "loss": 1.0673, "step": 3238 }, { "epoch": 0.0, "learning_rate": 4.9998784161943954e-05, "loss": 1.2363, "step": 3239 }, { "epoch": 0.0, "learning_rate": 4.9998783387160986e-05, "loss": 1.3043, "step": 3240 }, { "epoch": 0.0, "learning_rate": 4.999878261213125e-05, "loss": 0.9974, "step": 3241 }, { "epoch": 0.0, "learning_rate": 4.999878183685473e-05, "loss": 0.7341, "step": 3242 }, { "epoch": 0.0, "learning_rate": 4.9998781061331434e-05, "loss": 0.7755, "step": 3243 }, { "epoch": 0.0, "learning_rate": 4.999878028556136e-05, "loss": 0.951, "step": 3244 }, { "epoch": 0.0, "learning_rate": 4.999877950954451e-05, "loss": 0.7519, "step": 3245 }, { "epoch": 0.0, "learning_rate": 4.999877873328089e-05, "loss": 1.0954, "step": 3246 }, { "epoch": 0.0, "learning_rate": 4.999877795677048e-05, "loss": 1.0842, "step": 3247 }, { "epoch": 0.0, "learning_rate": 4.99987771800133e-05, "loss": 1.1799, "step": 3248 }, { "epoch": 0.0, "learning_rate": 4.999877640300934e-05, "loss": 1.0409, "step": 3249 }, { "epoch": 0.0, "learning_rate": 4.999877562575861e-05, "loss": 1.2298, "step": 3250 }, { "epoch": 0.0, "learning_rate": 4.999877484826109e-05, "loss": 1.1191, "step": 3251 }, { "epoch": 0.0, "learning_rate": 4.9998774070516796e-05, "loss": 0.9785, "step": 3252 }, { "epoch": 0.0, "learning_rate": 4.999877329252573e-05, "loss": 1.2481, "step": 3253 }, { "epoch": 0.0, "learning_rate": 4.999877251428788e-05, "loss": 1.4826, "step": 3254 }, { "epoch": 0.0, "learning_rate": 4.9998771735803264e-05, "loss": 1.1167, "step": 3255 }, { "epoch": 0.0, "learning_rate": 4.999877095707186e-05, "loss": 1.1116, "step": 3256 }, { "epoch": 0.0, "learning_rate": 4.9998770178093676e-05, "loss": 0.9439, "step": 3257 }, { "epoch": 0.0, "learning_rate": 4.9998769398868725e-05, "loss": 1.2351, "step": 3258 }, { "epoch": 0.0, "learning_rate": 4.9998768619396995e-05, "loss": 1.0686, "step": 3259 }, { "epoch": 0.0, "learning_rate": 4.999876783967848e-05, "loss": 1.5349, "step": 3260 }, { "epoch": 0.0, "learning_rate": 4.9998767059713196e-05, "loss": 1.1874, "step": 3261 }, { "epoch": 0.0, "learning_rate": 4.9998766279501133e-05, "loss": 1.3539, "step": 3262 }, { "epoch": 0.0, "learning_rate": 4.9998765499042286e-05, "loss": 1.2347, "step": 3263 }, { "epoch": 0.0, "learning_rate": 4.9998764718336673e-05, "loss": 1.2608, "step": 3264 }, { "epoch": 0.0, "learning_rate": 4.9998763937384277e-05, "loss": 1.4032, "step": 3265 }, { "epoch": 0.0, "learning_rate": 4.99987631561851e-05, "loss": 1.3431, "step": 3266 }, { "epoch": 0.0, "learning_rate": 4.999876237473915e-05, "loss": 1.027, "step": 3267 }, { "epoch": 0.0, "learning_rate": 4.9998761593046425e-05, "loss": 1.5965, "step": 3268 }, { "epoch": 0.0, "learning_rate": 4.9998760811106916e-05, "loss": 1.0208, "step": 3269 }, { "epoch": 0.0, "learning_rate": 4.9998760028920636e-05, "loss": 1.1091, "step": 3270 }, { "epoch": 0.0, "learning_rate": 4.999875924648758e-05, "loss": 1.2286, "step": 3271 }, { "epoch": 0.0, "learning_rate": 4.9998758463807735e-05, "loss": 0.8837, "step": 3272 }, { "epoch": 0.0, "learning_rate": 4.999875768088113e-05, "loss": 0.7764, "step": 3273 }, { "epoch": 0.0, "learning_rate": 4.999875689770773e-05, "loss": 0.7657, "step": 3274 }, { "epoch": 0.0, "learning_rate": 4.999875611428756e-05, "loss": 0.9198, "step": 3275 }, { "epoch": 0.0, "learning_rate": 4.999875533062062e-05, "loss": 1.1006, "step": 3276 }, { "epoch": 0.0, "learning_rate": 4.999875454670689e-05, "loss": 1.0076, "step": 3277 }, { "epoch": 0.0, "learning_rate": 4.999875376254639e-05, "loss": 1.3636, "step": 3278 }, { "epoch": 0.0, "learning_rate": 4.999875297813912e-05, "loss": 1.1701, "step": 3279 }, { "epoch": 0.0, "learning_rate": 4.999875219348506e-05, "loss": 1.1995, "step": 3280 }, { "epoch": 0.0, "learning_rate": 4.999875140858423e-05, "loss": 0.9329, "step": 3281 }, { "epoch": 0.0, "learning_rate": 4.999875062343662e-05, "loss": 1.2552, "step": 3282 }, { "epoch": 0.0, "learning_rate": 4.9998749838042235e-05, "loss": 1.1106, "step": 3283 }, { "epoch": 0.0, "learning_rate": 4.999874905240107e-05, "loss": 1.3376, "step": 3284 }, { "epoch": 0.0, "learning_rate": 4.999874826651313e-05, "loss": 1.1603, "step": 3285 }, { "epoch": 0.0, "learning_rate": 4.9998747480378413e-05, "loss": 1.4663, "step": 3286 }, { "epoch": 0.0, "learning_rate": 4.9998746693996915e-05, "loss": 1.3069, "step": 3287 }, { "epoch": 0.0, "learning_rate": 4.999874590736865e-05, "loss": 1.1898, "step": 3288 }, { "epoch": 0.0, "learning_rate": 4.99987451204936e-05, "loss": 1.2227, "step": 3289 }, { "epoch": 0.0, "learning_rate": 4.999874433337177e-05, "loss": 1.0331, "step": 3290 }, { "epoch": 0.0, "learning_rate": 4.999874354600317e-05, "loss": 1.115, "step": 3291 }, { "epoch": 0.0, "learning_rate": 4.999874275838779e-05, "loss": 1.0944, "step": 3292 }, { "epoch": 0.0, "learning_rate": 4.999874197052563e-05, "loss": 1.2408, "step": 3293 }, { "epoch": 0.0, "learning_rate": 4.9998741182416695e-05, "loss": 1.0281, "step": 3294 }, { "epoch": 0.0, "learning_rate": 4.9998740394060987e-05, "loss": 1.1958, "step": 3295 }, { "epoch": 0.0, "learning_rate": 4.99987396054585e-05, "loss": 0.6619, "step": 3296 }, { "epoch": 0.0, "learning_rate": 4.9998738816609235e-05, "loss": 1.2319, "step": 3297 }, { "epoch": 0.0, "learning_rate": 4.999873802751319e-05, "loss": 1.0441, "step": 3298 }, { "epoch": 0.0, "learning_rate": 4.999873723817037e-05, "loss": 1.2682, "step": 3299 }, { "epoch": 0.0, "learning_rate": 4.999873644858077e-05, "loss": 1.1649, "step": 3300 }, { "epoch": 0.0, "learning_rate": 4.99987356587444e-05, "loss": 0.9821, "step": 3301 }, { "epoch": 0.0, "learning_rate": 4.9998734868661255e-05, "loss": 0.9071, "step": 3302 }, { "epoch": 0.0, "learning_rate": 4.999873407833133e-05, "loss": 0.9467, "step": 3303 }, { "epoch": 0.0, "learning_rate": 4.9998733287754626e-05, "loss": 1.3175, "step": 3304 }, { "epoch": 0.0, "learning_rate": 4.9998732496931144e-05, "loss": 1.2642, "step": 3305 }, { "epoch": 0.0, "learning_rate": 4.9998731705860884e-05, "loss": 1.2276, "step": 3306 }, { "epoch": 0.0, "learning_rate": 4.9998730914543846e-05, "loss": 1.0944, "step": 3307 }, { "epoch": 0.0, "learning_rate": 4.999873012298004e-05, "loss": 0.7492, "step": 3308 }, { "epoch": 0.0, "learning_rate": 4.9998729331169444e-05, "loss": 1.5178, "step": 3309 }, { "epoch": 0.0, "learning_rate": 4.9998728539112086e-05, "loss": 1.1938, "step": 3310 }, { "epoch": 0.0, "learning_rate": 4.9998727746807936e-05, "loss": 1.2197, "step": 3311 }, { "epoch": 0.0, "learning_rate": 4.999872695425702e-05, "loss": 1.0732, "step": 3312 }, { "epoch": 0.0, "learning_rate": 4.9998726161459324e-05, "loss": 0.9876, "step": 3313 }, { "epoch": 0.0, "learning_rate": 4.999872536841485e-05, "loss": 0.8108, "step": 3314 }, { "epoch": 0.0, "learning_rate": 4.99987245751236e-05, "loss": 1.0706, "step": 3315 }, { "epoch": 0.0, "learning_rate": 4.999872378158558e-05, "loss": 1.2819, "step": 3316 }, { "epoch": 0.0, "learning_rate": 4.999872298780077e-05, "loss": 1.1775, "step": 3317 }, { "epoch": 0.0, "learning_rate": 4.999872219376919e-05, "loss": 1.0579, "step": 3318 }, { "epoch": 0.0, "learning_rate": 4.9998721399490835e-05, "loss": 1.0045, "step": 3319 }, { "epoch": 0.0, "learning_rate": 4.99987206049657e-05, "loss": 1.0541, "step": 3320 }, { "epoch": 0.0, "learning_rate": 4.999871981019378e-05, "loss": 1.5234, "step": 3321 }, { "epoch": 0.0, "learning_rate": 4.99987190151751e-05, "loss": 1.2823, "step": 3322 }, { "epoch": 0.0, "learning_rate": 4.999871821990964e-05, "loss": 1.5049, "step": 3323 }, { "epoch": 0.0, "learning_rate": 4.999871742439739e-05, "loss": 1.2312, "step": 3324 }, { "epoch": 0.0, "learning_rate": 4.999871662863837e-05, "loss": 1.2007, "step": 3325 }, { "epoch": 0.0, "learning_rate": 4.999871583263258e-05, "loss": 1.0416, "step": 3326 }, { "epoch": 0.0, "learning_rate": 4.9998715036380005e-05, "loss": 1.4341, "step": 3327 }, { "epoch": 0.0, "learning_rate": 4.999871423988066e-05, "loss": 0.9972, "step": 3328 }, { "epoch": 0.0, "learning_rate": 4.999871344313453e-05, "loss": 1.1302, "step": 3329 }, { "epoch": 0.0, "learning_rate": 4.999871264614163e-05, "loss": 1.2175, "step": 3330 }, { "epoch": 0.0, "learning_rate": 4.9998711848901945e-05, "loss": 0.804, "step": 3331 }, { "epoch": 0.0, "learning_rate": 4.999871105141549e-05, "loss": 0.8429, "step": 3332 }, { "epoch": 0.0, "learning_rate": 4.9998710253682255e-05, "loss": 0.966, "step": 3333 }, { "epoch": 0.0, "learning_rate": 4.999870945570225e-05, "loss": 1.0244, "step": 3334 }, { "epoch": 0.0, "learning_rate": 4.9998708657475466e-05, "loss": 1.2352, "step": 3335 }, { "epoch": 0.0, "learning_rate": 4.99987078590019e-05, "loss": 1.2682, "step": 3336 }, { "epoch": 0.0, "learning_rate": 4.999870706028156e-05, "loss": 1.2687, "step": 3337 }, { "epoch": 0.0, "learning_rate": 4.9998706261314444e-05, "loss": 1.1056, "step": 3338 }, { "epoch": 0.0, "learning_rate": 4.999870546210055e-05, "loss": 0.7855, "step": 3339 }, { "epoch": 0.0, "learning_rate": 4.999870466263988e-05, "loss": 1.506, "step": 3340 }, { "epoch": 0.0, "learning_rate": 4.999870386293244e-05, "loss": 1.3185, "step": 3341 }, { "epoch": 0.0, "learning_rate": 4.9998703062978215e-05, "loss": 1.0603, "step": 3342 }, { "epoch": 0.0, "learning_rate": 4.999870226277721e-05, "loss": 0.8978, "step": 3343 }, { "epoch": 0.0, "learning_rate": 4.9998701462329437e-05, "loss": 0.7992, "step": 3344 }, { "epoch": 0.0, "learning_rate": 4.9998700661634884e-05, "loss": 1.1016, "step": 3345 }, { "epoch": 0.0, "learning_rate": 4.999869986069355e-05, "loss": 1.0676, "step": 3346 }, { "epoch": 0.0, "learning_rate": 4.999869905950545e-05, "loss": 1.3322, "step": 3347 }, { "epoch": 0.0, "learning_rate": 4.9998698258070564e-05, "loss": 0.8448, "step": 3348 }, { "epoch": 0.0, "learning_rate": 4.999869745638891e-05, "loss": 1.2603, "step": 3349 }, { "epoch": 0.0, "learning_rate": 4.999869665446047e-05, "loss": 2.0377, "step": 3350 }, { "epoch": 0.0, "learning_rate": 4.999869585228526e-05, "loss": 1.1663, "step": 3351 }, { "epoch": 0.0, "learning_rate": 4.9998695049863263e-05, "loss": 1.3606, "step": 3352 }, { "epoch": 0.0, "learning_rate": 4.99986942471945e-05, "loss": 1.2747, "step": 3353 }, { "epoch": 0.0, "learning_rate": 4.999869344427896e-05, "loss": 1.1279, "step": 3354 }, { "epoch": 0.0, "learning_rate": 4.999869264111664e-05, "loss": 1.0564, "step": 3355 }, { "epoch": 0.0, "learning_rate": 4.999869183770754e-05, "loss": 0.5569, "step": 3356 }, { "epoch": 0.0, "learning_rate": 4.999869103405167e-05, "loss": 0.9706, "step": 3357 }, { "epoch": 0.0, "learning_rate": 4.999869023014903e-05, "loss": 0.992, "step": 3358 }, { "epoch": 0.0, "learning_rate": 4.99986894259996e-05, "loss": 0.9301, "step": 3359 }, { "epoch": 0.0, "learning_rate": 4.9998688621603395e-05, "loss": 1.0552, "step": 3360 }, { "epoch": 0.0, "learning_rate": 4.999868781696042e-05, "loss": 1.2832, "step": 3361 }, { "epoch": 0.0, "learning_rate": 4.9998687012070664e-05, "loss": 1.0131, "step": 3362 }, { "epoch": 0.0, "learning_rate": 4.999868620693413e-05, "loss": 1.0057, "step": 3363 }, { "epoch": 0.0, "learning_rate": 4.999868540155083e-05, "loss": 0.7254, "step": 3364 }, { "epoch": 0.0, "learning_rate": 4.999868459592074e-05, "loss": 0.9847, "step": 3365 }, { "epoch": 0.0, "learning_rate": 4.999868379004388e-05, "loss": 1.0289, "step": 3366 }, { "epoch": 0.0, "learning_rate": 4.9998682983920245e-05, "loss": 1.1107, "step": 3367 }, { "epoch": 0.0, "learning_rate": 4.9998682177549826e-05, "loss": 0.4372, "step": 3368 }, { "epoch": 0.0, "learning_rate": 4.9998681370932635e-05, "loss": 0.9341, "step": 3369 }, { "epoch": 0.0, "learning_rate": 4.9998680564068674e-05, "loss": 1.035, "step": 3370 }, { "epoch": 0.0, "learning_rate": 4.999867975695793e-05, "loss": 1.0862, "step": 3371 }, { "epoch": 0.0, "learning_rate": 4.999867894960041e-05, "loss": 1.0916, "step": 3372 }, { "epoch": 0.0, "learning_rate": 4.9998678141996115e-05, "loss": 1.4372, "step": 3373 }, { "epoch": 0.0, "learning_rate": 4.999867733414504e-05, "loss": 1.4361, "step": 3374 }, { "epoch": 0.0, "learning_rate": 4.99986765260472e-05, "loss": 1.2444, "step": 3375 }, { "epoch": 0.0, "learning_rate": 4.999867571770257e-05, "loss": 1.3493, "step": 3376 }, { "epoch": 0.0, "learning_rate": 4.999867490911117e-05, "loss": 1.1362, "step": 3377 }, { "epoch": 0.0, "learning_rate": 4.999867410027299e-05, "loss": 1.1176, "step": 3378 }, { "epoch": 0.0, "learning_rate": 4.999867329118804e-05, "loss": 1.4117, "step": 3379 }, { "epoch": 0.0, "learning_rate": 4.9998672481856305e-05, "loss": 1.6114, "step": 3380 }, { "epoch": 0.0, "learning_rate": 4.99986716722778e-05, "loss": 1.5723, "step": 3381 }, { "epoch": 0.0, "learning_rate": 4.9998670862452515e-05, "loss": 1.3304, "step": 3382 }, { "epoch": 0.0, "learning_rate": 4.9998670052380454e-05, "loss": 1.0998, "step": 3383 }, { "epoch": 0.0, "learning_rate": 4.9998669242061614e-05, "loss": 1.1452, "step": 3384 }, { "epoch": 0.0, "learning_rate": 4.9998668431496004e-05, "loss": 1.1085, "step": 3385 }, { "epoch": 0.0, "learning_rate": 4.9998667620683615e-05, "loss": 1.2093, "step": 3386 }, { "epoch": 0.0, "learning_rate": 4.999866680962445e-05, "loss": 1.1872, "step": 3387 }, { "epoch": 0.0, "learning_rate": 4.9998665998318504e-05, "loss": 1.1485, "step": 3388 }, { "epoch": 0.0, "learning_rate": 4.9998665186765795e-05, "loss": 0.9688, "step": 3389 }, { "epoch": 0.0, "learning_rate": 4.9998664374966294e-05, "loss": 1.3017, "step": 3390 }, { "epoch": 0.0, "learning_rate": 4.999866356292003e-05, "loss": 1.1592, "step": 3391 }, { "epoch": 0.0, "learning_rate": 4.999866275062698e-05, "loss": 1.1275, "step": 3392 }, { "epoch": 0.0, "learning_rate": 4.999866193808716e-05, "loss": 0.9936, "step": 3393 }, { "epoch": 0.0, "learning_rate": 4.999866112530056e-05, "loss": 1.0442, "step": 3394 }, { "epoch": 0.0, "learning_rate": 4.999866031226719e-05, "loss": 0.9005, "step": 3395 }, { "epoch": 0.0, "learning_rate": 4.9998659498987035e-05, "loss": 0.7956, "step": 3396 }, { "epoch": 0.0, "learning_rate": 4.999865868546011e-05, "loss": 0.771, "step": 3397 }, { "epoch": 0.0, "learning_rate": 4.9998657871686406e-05, "loss": 0.7461, "step": 3398 }, { "epoch": 0.0, "learning_rate": 4.9998657057665924e-05, "loss": 0.6217, "step": 3399 }, { "epoch": 0.0, "learning_rate": 4.9998656243398664e-05, "loss": 0.5203, "step": 3400 }, { "epoch": 0.0, "learning_rate": 4.999865542888463e-05, "loss": 0.596, "step": 3401 }, { "epoch": 0.0, "learning_rate": 4.9998654614123824e-05, "loss": 0.6091, "step": 3402 }, { "epoch": 0.0, "learning_rate": 4.999865379911624e-05, "loss": 0.6239, "step": 3403 }, { "epoch": 0.0, "learning_rate": 4.999865298386188e-05, "loss": 0.5065, "step": 3404 }, { "epoch": 0.0, "learning_rate": 4.9998652168360736e-05, "loss": 0.6384, "step": 3405 }, { "epoch": 0.0, "learning_rate": 4.999865135261282e-05, "loss": 0.645, "step": 3406 }, { "epoch": 0.0, "learning_rate": 4.999865053661814e-05, "loss": 1.1136, "step": 3407 }, { "epoch": 0.0, "learning_rate": 4.999864972037667e-05, "loss": 0.9981, "step": 3408 }, { "epoch": 0.0, "learning_rate": 4.999864890388843e-05, "loss": 1.2167, "step": 3409 }, { "epoch": 0.0, "learning_rate": 4.999864808715341e-05, "loss": 1.2866, "step": 3410 }, { "epoch": 0.0, "learning_rate": 4.9998647270171616e-05, "loss": 1.0734, "step": 3411 }, { "epoch": 0.0, "learning_rate": 4.999864645294305e-05, "loss": 1.1872, "step": 3412 }, { "epoch": 0.0, "learning_rate": 4.99986456354677e-05, "loss": 1.9395, "step": 3413 }, { "epoch": 0.0, "learning_rate": 4.999864481774558e-05, "loss": 1.3458, "step": 3414 }, { "epoch": 0.0, "learning_rate": 4.999864399977668e-05, "loss": 1.2487, "step": 3415 }, { "epoch": 0.0, "learning_rate": 4.999864318156101e-05, "loss": 1.4208, "step": 3416 }, { "epoch": 0.0, "learning_rate": 4.9998642363098556e-05, "loss": 1.1311, "step": 3417 }, { "epoch": 0.0, "learning_rate": 4.999864154438933e-05, "loss": 1.2627, "step": 3418 }, { "epoch": 0.0, "learning_rate": 4.9998640725433326e-05, "loss": 1.3189, "step": 3419 }, { "epoch": 0.0, "learning_rate": 4.999863990623055e-05, "loss": 1.2433, "step": 3420 }, { "epoch": 0.0, "learning_rate": 4.9998639086781e-05, "loss": 1.1867, "step": 3421 }, { "epoch": 0.0, "learning_rate": 4.9998638267084664e-05, "loss": 0.8289, "step": 3422 }, { "epoch": 0.0, "learning_rate": 4.999863744714156e-05, "loss": 1.4112, "step": 3423 }, { "epoch": 0.0, "learning_rate": 4.9998636626951674e-05, "loss": 1.2278, "step": 3424 }, { "epoch": 0.0, "learning_rate": 4.999863580651501e-05, "loss": 0.9466, "step": 3425 }, { "epoch": 0.0, "learning_rate": 4.999863498583158e-05, "loss": 1.1715, "step": 3426 }, { "epoch": 0.0, "learning_rate": 4.999863416490137e-05, "loss": 1.4013, "step": 3427 }, { "epoch": 0.0, "learning_rate": 4.9998633343724374e-05, "loss": 1.0322, "step": 3428 }, { "epoch": 0.0, "learning_rate": 4.9998632522300614e-05, "loss": 1.1111, "step": 3429 }, { "epoch": 0.0, "learning_rate": 4.9998631700630076e-05, "loss": 0.8467, "step": 3430 }, { "epoch": 0.0, "learning_rate": 4.999863087871276e-05, "loss": 1.0402, "step": 3431 }, { "epoch": 0.0, "learning_rate": 4.999863005654867e-05, "loss": 0.9011, "step": 3432 }, { "epoch": 0.0, "learning_rate": 4.999862923413781e-05, "loss": 1.1107, "step": 3433 }, { "epoch": 0.0, "learning_rate": 4.9998628411480165e-05, "loss": 0.7305, "step": 3434 }, { "epoch": 0.0, "learning_rate": 4.9998627588575744e-05, "loss": 0.5344, "step": 3435 }, { "epoch": 0.0, "learning_rate": 4.999862676542455e-05, "loss": 0.5887, "step": 3436 }, { "epoch": 0.0, "learning_rate": 4.9998625942026575e-05, "loss": 0.53, "step": 3437 }, { "epoch": 0.0, "learning_rate": 4.9998625118381834e-05, "loss": 0.6023, "step": 3438 }, { "epoch": 0.0, "learning_rate": 4.9998624294490315e-05, "loss": 0.5898, "step": 3439 }, { "epoch": 0.0, "learning_rate": 4.999862347035201e-05, "loss": 0.5584, "step": 3440 }, { "epoch": 0.0, "learning_rate": 4.999862264596694e-05, "loss": 0.5397, "step": 3441 }, { "epoch": 0.0, "learning_rate": 4.999862182133509e-05, "loss": 0.732, "step": 3442 }, { "epoch": 0.0, "learning_rate": 4.9998620996456465e-05, "loss": 0.588, "step": 3443 }, { "epoch": 0.0, "learning_rate": 4.999862017133106e-05, "loss": 0.5927, "step": 3444 }, { "epoch": 0.0, "learning_rate": 4.999861934595889e-05, "loss": 0.5751, "step": 3445 }, { "epoch": 0.0, "learning_rate": 4.999861852033994e-05, "loss": 0.5968, "step": 3446 }, { "epoch": 0.0, "learning_rate": 4.99986176944742e-05, "loss": 0.6897, "step": 3447 }, { "epoch": 0.0, "learning_rate": 4.9998616868361694e-05, "loss": 0.4709, "step": 3448 }, { "epoch": 0.0, "learning_rate": 4.9998616042002415e-05, "loss": 0.5125, "step": 3449 }, { "epoch": 0.0, "learning_rate": 4.999861521539636e-05, "loss": 0.4163, "step": 3450 }, { "epoch": 0.0, "learning_rate": 4.999861438854353e-05, "loss": 0.4144, "step": 3451 }, { "epoch": 0.0, "learning_rate": 4.999861356144392e-05, "loss": 0.4661, "step": 3452 }, { "epoch": 0.0, "learning_rate": 4.999861273409754e-05, "loss": 0.3836, "step": 3453 }, { "epoch": 0.0, "learning_rate": 4.999861190650438e-05, "loss": 0.3678, "step": 3454 }, { "epoch": 0.0, "learning_rate": 4.999861107866444e-05, "loss": 0.4695, "step": 3455 }, { "epoch": 0.0, "learning_rate": 4.999861025057773e-05, "loss": 0.3993, "step": 3456 }, { "epoch": 0.0, "learning_rate": 4.999860942224425e-05, "loss": 0.3604, "step": 3457 }, { "epoch": 0.0, "learning_rate": 4.999860859366398e-05, "loss": 0.4166, "step": 3458 }, { "epoch": 0.0, "learning_rate": 4.999860776483695e-05, "loss": 0.4779, "step": 3459 }, { "epoch": 0.0, "learning_rate": 4.999860693576313e-05, "loss": 0.4345, "step": 3460 }, { "epoch": 0.0, "learning_rate": 4.999860610644253e-05, "loss": 0.4439, "step": 3461 }, { "epoch": 0.0, "learning_rate": 4.9998605276875175e-05, "loss": 0.3738, "step": 3462 }, { "epoch": 0.0, "learning_rate": 4.999860444706103e-05, "loss": 0.5384, "step": 3463 }, { "epoch": 0.0, "learning_rate": 4.999860361700012e-05, "loss": 1.3531, "step": 3464 }, { "epoch": 0.0, "learning_rate": 4.9998602786692426e-05, "loss": 1.1958, "step": 3465 }, { "epoch": 0.0, "learning_rate": 4.9998601956137955e-05, "loss": 1.3425, "step": 3466 }, { "epoch": 0.0, "learning_rate": 4.999860112533671e-05, "loss": 1.3543, "step": 3467 }, { "epoch": 0.0, "learning_rate": 4.999860029428869e-05, "loss": 0.9571, "step": 3468 }, { "epoch": 0.0, "learning_rate": 4.99985994629939e-05, "loss": 1.2027, "step": 3469 }, { "epoch": 0.0, "learning_rate": 4.999859863145233e-05, "loss": 1.2444, "step": 3470 }, { "epoch": 0.0, "learning_rate": 4.999859779966398e-05, "loss": 1.1483, "step": 3471 }, { "epoch": 0.0, "learning_rate": 4.999859696762886e-05, "loss": 1.2692, "step": 3472 }, { "epoch": 0.0, "learning_rate": 4.9998596135346956e-05, "loss": 1.1072, "step": 3473 }, { "epoch": 0.0, "learning_rate": 4.999859530281828e-05, "loss": 1.0419, "step": 3474 }, { "epoch": 0.0, "learning_rate": 4.999859447004284e-05, "loss": 1.103, "step": 3475 }, { "epoch": 0.0, "learning_rate": 4.9998593637020615e-05, "loss": 1.3038, "step": 3476 }, { "epoch": 0.0, "learning_rate": 4.999859280375161e-05, "loss": 1.1934, "step": 3477 }, { "epoch": 0.0, "learning_rate": 4.9998591970235836e-05, "loss": 1.0241, "step": 3478 }, { "epoch": 0.0, "learning_rate": 4.9998591136473286e-05, "loss": 1.088, "step": 3479 }, { "epoch": 0.0, "learning_rate": 4.999859030246395e-05, "loss": 1.1071, "step": 3480 }, { "epoch": 0.0, "learning_rate": 4.999858946820786e-05, "loss": 1.1374, "step": 3481 }, { "epoch": 0.0, "learning_rate": 4.999858863370498e-05, "loss": 0.9614, "step": 3482 }, { "epoch": 0.0, "learning_rate": 4.999858779895532e-05, "loss": 1.1881, "step": 3483 }, { "epoch": 0.0, "learning_rate": 4.9998586963958894e-05, "loss": 1.2886, "step": 3484 }, { "epoch": 0.0, "learning_rate": 4.999858612871569e-05, "loss": 0.7944, "step": 3485 }, { "epoch": 0.0, "learning_rate": 4.999858529322571e-05, "loss": 1.3246, "step": 3486 }, { "epoch": 0.0, "learning_rate": 4.9998584457488955e-05, "loss": 1.3053, "step": 3487 }, { "epoch": 0.0, "learning_rate": 4.9998583621505424e-05, "loss": 1.3665, "step": 3488 }, { "epoch": 0.0, "learning_rate": 4.9998582785275114e-05, "loss": 1.2066, "step": 3489 }, { "epoch": 0.0, "learning_rate": 4.999858194879804e-05, "loss": 1.1096, "step": 3490 }, { "epoch": 0.0, "learning_rate": 4.999858111207418e-05, "loss": 1.3171, "step": 3491 }, { "epoch": 0.0, "learning_rate": 4.9998580275103546e-05, "loss": 0.9422, "step": 3492 }, { "epoch": 0.0, "learning_rate": 4.999857943788614e-05, "loss": 1.0387, "step": 3493 }, { "epoch": 0.0, "learning_rate": 4.999857860042195e-05, "loss": 1.2342, "step": 3494 }, { "epoch": 0.0, "learning_rate": 4.9998577762711e-05, "loss": 1.0393, "step": 3495 }, { "epoch": 0.0, "learning_rate": 4.999857692475326e-05, "loss": 0.9846, "step": 3496 }, { "epoch": 0.0, "learning_rate": 4.9998576086548756e-05, "loss": 1.7033, "step": 3497 }, { "epoch": 0.0, "learning_rate": 4.9998575248097466e-05, "loss": 1.2261, "step": 3498 }, { "epoch": 0.0, "learning_rate": 4.9998574409399404e-05, "loss": 0.8879, "step": 3499 }, { "epoch": 0.0, "learning_rate": 4.9998573570454564e-05, "loss": 0.8896, "step": 3500 }, { "epoch": 0.0, "eval_loss": 1.0793774127960205, "eval_runtime": 85.2881, "eval_samples_per_second": 16.239, "eval_steps_per_second": 4.069, "step": 3500 }, { "epoch": 0.0, "learning_rate": 4.999857273126296e-05, "loss": 0.9776, "step": 3501 }, { "epoch": 0.0, "learning_rate": 4.999857189182457e-05, "loss": 1.1962, "step": 3502 }, { "epoch": 0.0, "learning_rate": 4.999857105213941e-05, "loss": 0.9708, "step": 3503 }, { "epoch": 0.0, "learning_rate": 4.999857021220747e-05, "loss": 0.9686, "step": 3504 }, { "epoch": 0.0, "learning_rate": 4.999856937202876e-05, "loss": 1.0573, "step": 3505 }, { "epoch": 0.0, "learning_rate": 4.999856853160327e-05, "loss": 1.1712, "step": 3506 }, { "epoch": 0.0, "learning_rate": 4.999856769093101e-05, "loss": 1.1592, "step": 3507 }, { "epoch": 0.0, "learning_rate": 4.999856685001196e-05, "loss": 0.8742, "step": 3508 }, { "epoch": 0.0, "learning_rate": 4.999856600884615e-05, "loss": 0.9963, "step": 3509 }, { "epoch": 0.0, "learning_rate": 4.999856516743356e-05, "loss": 1.009, "step": 3510 }, { "epoch": 0.0, "learning_rate": 4.99985643257742e-05, "loss": 1.2888, "step": 3511 }, { "epoch": 0.0, "learning_rate": 4.999856348386806e-05, "loss": 1.0399, "step": 3512 }, { "epoch": 0.0, "learning_rate": 4.999856264171514e-05, "loss": 1.138, "step": 3513 }, { "epoch": 0.0, "learning_rate": 4.999856179931545e-05, "loss": 0.9903, "step": 3514 }, { "epoch": 0.0, "learning_rate": 4.9998560956668985e-05, "loss": 0.8786, "step": 3515 }, { "epoch": 0.0, "learning_rate": 4.9998560113775746e-05, "loss": 1.2258, "step": 3516 }, { "epoch": 0.0, "learning_rate": 4.999855927063573e-05, "loss": 1.2219, "step": 3517 }, { "epoch": 0.0, "learning_rate": 4.999855842724893e-05, "loss": 1.0808, "step": 3518 }, { "epoch": 0.0, "learning_rate": 4.9998557583615365e-05, "loss": 1.2123, "step": 3519 }, { "epoch": 0.0, "learning_rate": 4.999855673973503e-05, "loss": 1.3638, "step": 3520 }, { "epoch": 0.0, "learning_rate": 4.9998555895607904e-05, "loss": 1.071, "step": 3521 }, { "epoch": 0.0, "learning_rate": 4.999855505123402e-05, "loss": 0.5962, "step": 3522 }, { "epoch": 0.0, "learning_rate": 4.9998554206613345e-05, "loss": 0.9059, "step": 3523 }, { "epoch": 0.0, "learning_rate": 4.99985533617459e-05, "loss": 0.9626, "step": 3524 }, { "epoch": 0.0, "learning_rate": 4.999855251663169e-05, "loss": 1.3906, "step": 3525 }, { "epoch": 0.0, "learning_rate": 4.999855167127069e-05, "loss": 1.3815, "step": 3526 }, { "epoch": 0.0, "learning_rate": 4.9998550825662925e-05, "loss": 1.3475, "step": 3527 }, { "epoch": 0.0, "learning_rate": 4.999854997980838e-05, "loss": 1.2545, "step": 3528 }, { "epoch": 0.0, "learning_rate": 4.9998549133707064e-05, "loss": 1.2959, "step": 3529 }, { "epoch": 0.0, "learning_rate": 4.9998548287358974e-05, "loss": 0.9743, "step": 3530 }, { "epoch": 0.0, "learning_rate": 4.99985474407641e-05, "loss": 1.2155, "step": 3531 }, { "epoch": 0.0, "learning_rate": 4.999854659392245e-05, "loss": 1.107, "step": 3532 }, { "epoch": 0.0, "learning_rate": 4.9998545746834034e-05, "loss": 1.0236, "step": 3533 }, { "epoch": 0.0, "learning_rate": 4.999854489949884e-05, "loss": 1.0437, "step": 3534 }, { "epoch": 0.0, "learning_rate": 4.9998544051916865e-05, "loss": 1.1194, "step": 3535 }, { "epoch": 0.0, "learning_rate": 4.999854320408812e-05, "loss": 1.0564, "step": 3536 }, { "epoch": 0.0, "learning_rate": 4.9998542356012604e-05, "loss": 1.0349, "step": 3537 }, { "epoch": 0.0, "learning_rate": 4.999854150769031e-05, "loss": 1.2829, "step": 3538 }, { "epoch": 0.0, "learning_rate": 4.999854065912124e-05, "loss": 0.9777, "step": 3539 }, { "epoch": 0.0, "learning_rate": 4.9998539810305394e-05, "loss": 1.2097, "step": 3540 }, { "epoch": 0.0, "learning_rate": 4.999853896124277e-05, "loss": 0.8308, "step": 3541 }, { "epoch": 0.0, "learning_rate": 4.999853811193338e-05, "loss": 0.91, "step": 3542 }, { "epoch": 0.0, "learning_rate": 4.999853726237721e-05, "loss": 0.9906, "step": 3543 }, { "epoch": 0.0, "learning_rate": 4.999853641257426e-05, "loss": 1.122, "step": 3544 }, { "epoch": 0.0, "learning_rate": 4.999853556252454e-05, "loss": 0.9985, "step": 3545 }, { "epoch": 0.0, "learning_rate": 4.999853471222805e-05, "loss": 0.9434, "step": 3546 }, { "epoch": 0.0, "learning_rate": 4.9998533861684784e-05, "loss": 1.1515, "step": 3547 }, { "epoch": 0.0, "learning_rate": 4.999853301089473e-05, "loss": 1.1774, "step": 3548 }, { "epoch": 0.0, "learning_rate": 4.999853215985791e-05, "loss": 0.9875, "step": 3549 }, { "epoch": 0.0, "learning_rate": 4.999853130857432e-05, "loss": 1.1445, "step": 3550 }, { "epoch": 0.0, "learning_rate": 4.999853045704394e-05, "loss": 1.3954, "step": 3551 }, { "epoch": 0.0, "learning_rate": 4.9998529605266805e-05, "loss": 1.2126, "step": 3552 }, { "epoch": 0.0, "learning_rate": 4.999852875324288e-05, "loss": 0.8698, "step": 3553 }, { "epoch": 0.0, "learning_rate": 4.999852790097219e-05, "loss": 1.1348, "step": 3554 }, { "epoch": 0.0, "learning_rate": 4.9998527048454716e-05, "loss": 0.7995, "step": 3555 }, { "epoch": 0.0, "learning_rate": 4.999852619569047e-05, "loss": 1.3086, "step": 3556 }, { "epoch": 0.0, "learning_rate": 4.999852534267945e-05, "loss": 1.1718, "step": 3557 }, { "epoch": 0.0, "learning_rate": 4.999852448942166e-05, "loss": 1.0632, "step": 3558 }, { "epoch": 0.0, "learning_rate": 4.999852363591709e-05, "loss": 1.0728, "step": 3559 }, { "epoch": 0.0, "learning_rate": 4.999852278216575e-05, "loss": 1.2243, "step": 3560 }, { "epoch": 0.0, "learning_rate": 4.999852192816762e-05, "loss": 1.1648, "step": 3561 }, { "epoch": 0.0, "learning_rate": 4.999852107392273e-05, "loss": 0.9032, "step": 3562 }, { "epoch": 0.0, "learning_rate": 4.999852021943106e-05, "loss": 1.2605, "step": 3563 }, { "epoch": 0.0, "learning_rate": 4.999851936469262e-05, "loss": 0.9793, "step": 3564 }, { "epoch": 0.0, "learning_rate": 4.99985185097074e-05, "loss": 1.275, "step": 3565 }, { "epoch": 0.0, "learning_rate": 4.999851765447541e-05, "loss": 1.0971, "step": 3566 }, { "epoch": 0.0, "learning_rate": 4.9998516798996634e-05, "loss": 1.2443, "step": 3567 }, { "epoch": 0.0, "learning_rate": 4.9998515943271096e-05, "loss": 1.1871, "step": 3568 }, { "epoch": 0.0, "learning_rate": 4.999851508729877e-05, "loss": 1.0477, "step": 3569 }, { "epoch": 0.0, "learning_rate": 4.9998514231079686e-05, "loss": 1.0575, "step": 3570 }, { "epoch": 0.0, "learning_rate": 4.9998513374613814e-05, "loss": 0.9334, "step": 3571 }, { "epoch": 0.0, "learning_rate": 4.999851251790118e-05, "loss": 1.1136, "step": 3572 }, { "epoch": 0.0, "learning_rate": 4.9998511660941756e-05, "loss": 1.2688, "step": 3573 }, { "epoch": 0.0, "learning_rate": 4.9998510803735564e-05, "loss": 1.1204, "step": 3574 }, { "epoch": 0.0, "learning_rate": 4.999850994628259e-05, "loss": 1.1439, "step": 3575 }, { "epoch": 0.0, "learning_rate": 4.999850908858286e-05, "loss": 1.1018, "step": 3576 }, { "epoch": 0.0, "learning_rate": 4.999850823063634e-05, "loss": 1.3292, "step": 3577 }, { "epoch": 0.0, "learning_rate": 4.999850737244305e-05, "loss": 0.9875, "step": 3578 }, { "epoch": 0.0, "learning_rate": 4.999850651400298e-05, "loss": 1.0082, "step": 3579 }, { "epoch": 0.0, "learning_rate": 4.999850565531614e-05, "loss": 1.189, "step": 3580 }, { "epoch": 0.0, "learning_rate": 4.999850479638252e-05, "loss": 2.1198, "step": 3581 }, { "epoch": 0.0, "learning_rate": 4.999850393720213e-05, "loss": 1.5383, "step": 3582 }, { "epoch": 0.0, "learning_rate": 4.999850307777497e-05, "loss": 1.1514, "step": 3583 }, { "epoch": 0.0, "learning_rate": 4.999850221810103e-05, "loss": 0.9158, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.999850135818032e-05, "loss": 1.1748, "step": 3585 }, { "epoch": 0.0, "learning_rate": 4.9998500498012824e-05, "loss": 1.2605, "step": 3586 }, { "epoch": 0.0, "learning_rate": 4.999849963759856e-05, "loss": 1.286, "step": 3587 }, { "epoch": 0.0, "learning_rate": 4.999849877693752e-05, "loss": 1.0966, "step": 3588 }, { "epoch": 0.0, "learning_rate": 4.9998497916029714e-05, "loss": 1.1221, "step": 3589 }, { "epoch": 0.0, "learning_rate": 4.999849705487513e-05, "loss": 1.2224, "step": 3590 }, { "epoch": 0.0, "learning_rate": 4.9998496193473765e-05, "loss": 0.6984, "step": 3591 }, { "epoch": 0.0, "learning_rate": 4.9998495331825623e-05, "loss": 1.261, "step": 3592 }, { "epoch": 0.0, "learning_rate": 4.999849446993071e-05, "loss": 1.4161, "step": 3593 }, { "epoch": 0.0, "learning_rate": 4.999849360778903e-05, "loss": 1.2171, "step": 3594 }, { "epoch": 0.0, "learning_rate": 4.999849274540057e-05, "loss": 0.8938, "step": 3595 }, { "epoch": 0.0, "learning_rate": 4.999849188276534e-05, "loss": 1.21, "step": 3596 }, { "epoch": 0.0, "learning_rate": 4.999849101988333e-05, "loss": 1.4424, "step": 3597 }, { "epoch": 0.0, "learning_rate": 4.999849015675454e-05, "loss": 1.1366, "step": 3598 }, { "epoch": 0.0, "learning_rate": 4.999848929337898e-05, "loss": 0.9923, "step": 3599 }, { "epoch": 0.0, "learning_rate": 4.999848842975665e-05, "loss": 1.1938, "step": 3600 }, { "epoch": 0.0, "learning_rate": 4.9998487565887544e-05, "loss": 1.095, "step": 3601 }, { "epoch": 0.0, "learning_rate": 4.999848670177166e-05, "loss": 1.1377, "step": 3602 }, { "epoch": 0.0, "learning_rate": 4.9998485837409004e-05, "loss": 1.1227, "step": 3603 }, { "epoch": 0.0, "learning_rate": 4.999848497279957e-05, "loss": 1.1024, "step": 3604 }, { "epoch": 0.0, "learning_rate": 4.999848410794337e-05, "loss": 0.7742, "step": 3605 }, { "epoch": 0.0, "learning_rate": 4.999848324284039e-05, "loss": 1.2553, "step": 3606 }, { "epoch": 0.0, "learning_rate": 4.999848237749063e-05, "loss": 0.9677, "step": 3607 }, { "epoch": 0.0, "learning_rate": 4.99984815118941e-05, "loss": 1.0944, "step": 3608 }, { "epoch": 0.0, "learning_rate": 4.99984806460508e-05, "loss": 0.93, "step": 3609 }, { "epoch": 0.0, "learning_rate": 4.999847977996072e-05, "loss": 1.1312, "step": 3610 }, { "epoch": 0.0, "learning_rate": 4.999847891362387e-05, "loss": 1.0116, "step": 3611 }, { "epoch": 0.0, "learning_rate": 4.9998478047040245e-05, "loss": 0.9964, "step": 3612 }, { "epoch": 0.0, "learning_rate": 4.9998477180209834e-05, "loss": 0.875, "step": 3613 }, { "epoch": 0.0, "learning_rate": 4.999847631313266e-05, "loss": 0.8657, "step": 3614 }, { "epoch": 0.0, "learning_rate": 4.999847544580871e-05, "loss": 1.2225, "step": 3615 }, { "epoch": 0.0, "learning_rate": 4.9998474578237986e-05, "loss": 0.9426, "step": 3616 }, { "epoch": 0.0, "learning_rate": 4.9998473710420484e-05, "loss": 1.1022, "step": 3617 }, { "epoch": 0.0, "learning_rate": 4.999847284235622e-05, "loss": 1.1475, "step": 3618 }, { "epoch": 0.0, "learning_rate": 4.9998471974045165e-05, "loss": 1.1483, "step": 3619 }, { "epoch": 0.0, "learning_rate": 4.999847110548734e-05, "loss": 0.9323, "step": 3620 }, { "epoch": 0.0, "learning_rate": 4.999847023668275e-05, "loss": 0.9708, "step": 3621 }, { "epoch": 0.0, "learning_rate": 4.9998469367631375e-05, "loss": 0.8721, "step": 3622 }, { "epoch": 0.0, "learning_rate": 4.999846849833323e-05, "loss": 0.7354, "step": 3623 }, { "epoch": 0.0, "learning_rate": 4.999846762878831e-05, "loss": 0.7569, "step": 3624 }, { "epoch": 0.0, "learning_rate": 4.999846675899662e-05, "loss": 0.6744, "step": 3625 }, { "epoch": 0.0, "learning_rate": 4.999846588895815e-05, "loss": 0.6684, "step": 3626 }, { "epoch": 0.0, "learning_rate": 4.99984650186729e-05, "loss": 0.7758, "step": 3627 }, { "epoch": 0.0, "learning_rate": 4.9998464148140886e-05, "loss": 0.566, "step": 3628 }, { "epoch": 0.0, "learning_rate": 4.9998463277362095e-05, "loss": 0.6658, "step": 3629 }, { "epoch": 0.0, "learning_rate": 4.9998462406336526e-05, "loss": 0.5957, "step": 3630 }, { "epoch": 0.0, "learning_rate": 4.9998461535064186e-05, "loss": 0.5347, "step": 3631 }, { "epoch": 0.0, "learning_rate": 4.9998460663545075e-05, "loss": 0.5478, "step": 3632 }, { "epoch": 0.0, "learning_rate": 4.9998459791779186e-05, "loss": 0.4344, "step": 3633 }, { "epoch": 0.0, "learning_rate": 4.999845891976652e-05, "loss": 0.8633, "step": 3634 }, { "epoch": 0.0, "learning_rate": 4.999845804750709e-05, "loss": 1.2405, "step": 3635 }, { "epoch": 0.0, "learning_rate": 4.999845717500087e-05, "loss": 1.6678, "step": 3636 }, { "epoch": 0.0, "learning_rate": 4.999845630224789e-05, "loss": 1.3302, "step": 3637 }, { "epoch": 0.0, "learning_rate": 4.9998455429248125e-05, "loss": 1.0957, "step": 3638 }, { "epoch": 0.0, "learning_rate": 4.9998454556001595e-05, "loss": 0.8661, "step": 3639 }, { "epoch": 0.0, "learning_rate": 4.999845368250829e-05, "loss": 0.9106, "step": 3640 }, { "epoch": 0.0, "learning_rate": 4.99984528087682e-05, "loss": 0.9173, "step": 3641 }, { "epoch": 0.0, "learning_rate": 4.9998451934781344e-05, "loss": 1.0613, "step": 3642 }, { "epoch": 0.0, "learning_rate": 4.9998451060547716e-05, "loss": 0.9675, "step": 3643 }, { "epoch": 0.0, "learning_rate": 4.9998450186067317e-05, "loss": 1.0819, "step": 3644 }, { "epoch": 0.0, "learning_rate": 4.999844931134013e-05, "loss": 1.2227, "step": 3645 }, { "epoch": 0.0, "learning_rate": 4.999844843636618e-05, "loss": 1.172, "step": 3646 }, { "epoch": 0.0, "learning_rate": 4.999844756114545e-05, "loss": 1.0692, "step": 3647 }, { "epoch": 0.0, "learning_rate": 4.9998446685677946e-05, "loss": 0.88, "step": 3648 }, { "epoch": 0.0, "learning_rate": 4.999844580996367e-05, "loss": 1.0156, "step": 3649 }, { "epoch": 0.0, "learning_rate": 4.999844493400262e-05, "loss": 1.1947, "step": 3650 }, { "epoch": 0.0, "learning_rate": 4.99984440577948e-05, "loss": 1.1163, "step": 3651 }, { "epoch": 0.0, "learning_rate": 4.99984431813402e-05, "loss": 1.0156, "step": 3652 }, { "epoch": 0.0, "learning_rate": 4.999844230463883e-05, "loss": 1.1083, "step": 3653 }, { "epoch": 0.0, "learning_rate": 4.999844142769068e-05, "loss": 1.0469, "step": 3654 }, { "epoch": 0.0, "learning_rate": 4.999844055049576e-05, "loss": 0.8485, "step": 3655 }, { "epoch": 0.0, "learning_rate": 4.9998439673054066e-05, "loss": 0.9361, "step": 3656 }, { "epoch": 0.0, "learning_rate": 4.99984387953656e-05, "loss": 1.2245, "step": 3657 }, { "epoch": 0.0, "learning_rate": 4.999843791743035e-05, "loss": 1.2246, "step": 3658 }, { "epoch": 0.0, "learning_rate": 4.9998437039248336e-05, "loss": 1.329, "step": 3659 }, { "epoch": 0.0, "learning_rate": 4.999843616081955e-05, "loss": 1.7445, "step": 3660 }, { "epoch": 0.0, "learning_rate": 4.9998435282143986e-05, "loss": 1.8381, "step": 3661 }, { "epoch": 0.0, "learning_rate": 4.9998434403221644e-05, "loss": 1.6972, "step": 3662 }, { "epoch": 0.0, "learning_rate": 4.999843352405254e-05, "loss": 1.7102, "step": 3663 }, { "epoch": 0.0, "learning_rate": 4.999843264463665e-05, "loss": 1.7059, "step": 3664 }, { "epoch": 0.0, "learning_rate": 4.9998431764973985e-05, "loss": 1.6628, "step": 3665 }, { "epoch": 0.0, "learning_rate": 4.999843088506455e-05, "loss": 1.7004, "step": 3666 }, { "epoch": 0.0, "learning_rate": 4.999843000490834e-05, "loss": 1.675, "step": 3667 }, { "epoch": 0.0, "learning_rate": 4.999842912450536e-05, "loss": 1.6609, "step": 3668 }, { "epoch": 0.0, "learning_rate": 4.99984282438556e-05, "loss": 1.6648, "step": 3669 }, { "epoch": 0.0, "learning_rate": 4.999842736295907e-05, "loss": 1.6662, "step": 3670 }, { "epoch": 0.0, "learning_rate": 4.999842648181576e-05, "loss": 1.6543, "step": 3671 }, { "epoch": 0.0, "learning_rate": 4.999842560042569e-05, "loss": 1.6691, "step": 3672 }, { "epoch": 0.0, "learning_rate": 4.9998424718788837e-05, "loss": 1.6507, "step": 3673 }, { "epoch": 0.0, "learning_rate": 4.9998423836905207e-05, "loss": 1.6278, "step": 3674 }, { "epoch": 0.0, "learning_rate": 4.999842295477481e-05, "loss": 1.6509, "step": 3675 }, { "epoch": 0.0, "learning_rate": 4.999842207239763e-05, "loss": 1.6284, "step": 3676 }, { "epoch": 0.0, "learning_rate": 4.999842118977369e-05, "loss": 1.6695, "step": 3677 }, { "epoch": 0.0, "learning_rate": 4.999842030690297e-05, "loss": 1.6526, "step": 3678 }, { "epoch": 0.0, "learning_rate": 4.999841942378547e-05, "loss": 1.6508, "step": 3679 }, { "epoch": 0.0, "learning_rate": 4.99984185404212e-05, "loss": 1.6496, "step": 3680 }, { "epoch": 0.0, "learning_rate": 4.9998417656810156e-05, "loss": 1.6646, "step": 3681 }, { "epoch": 0.0, "learning_rate": 4.9998416772952336e-05, "loss": 1.6578, "step": 3682 }, { "epoch": 0.0, "learning_rate": 4.999841588884775e-05, "loss": 1.6691, "step": 3683 }, { "epoch": 0.0, "learning_rate": 4.999841500449638e-05, "loss": 1.6716, "step": 3684 }, { "epoch": 0.0, "learning_rate": 4.999841411989824e-05, "loss": 1.658, "step": 3685 }, { "epoch": 0.0, "learning_rate": 4.999841323505333e-05, "loss": 1.6474, "step": 3686 }, { "epoch": 0.0, "learning_rate": 4.999841234996164e-05, "loss": 1.6518, "step": 3687 }, { "epoch": 0.0, "learning_rate": 4.999841146462319e-05, "loss": 1.6826, "step": 3688 }, { "epoch": 0.0, "learning_rate": 4.999841057903795e-05, "loss": 1.6797, "step": 3689 }, { "epoch": 0.0, "learning_rate": 4.999840969320595e-05, "loss": 1.6301, "step": 3690 }, { "epoch": 0.0, "learning_rate": 4.999840880712716e-05, "loss": 1.6173, "step": 3691 }, { "epoch": 0.0, "learning_rate": 4.999840792080161e-05, "loss": 1.6426, "step": 3692 }, { "epoch": 0.0, "learning_rate": 4.999840703422928e-05, "loss": 1.6242, "step": 3693 }, { "epoch": 0.0, "learning_rate": 4.9998406147410176e-05, "loss": 1.64, "step": 3694 }, { "epoch": 0.0, "learning_rate": 4.99984052603443e-05, "loss": 1.6469, "step": 3695 }, { "epoch": 0.0, "learning_rate": 4.9998404373031646e-05, "loss": 1.6164, "step": 3696 }, { "epoch": 0.0, "learning_rate": 4.9998403485472225e-05, "loss": 1.6305, "step": 3697 }, { "epoch": 0.0, "learning_rate": 4.9998402597666025e-05, "loss": 1.6101, "step": 3698 }, { "epoch": 0.0, "learning_rate": 4.9998401709613054e-05, "loss": 1.617, "step": 3699 }, { "epoch": 0.0, "learning_rate": 4.999840082131331e-05, "loss": 1.6174, "step": 3700 }, { "epoch": 0.0, "learning_rate": 4.999839993276679e-05, "loss": 1.6436, "step": 3701 }, { "epoch": 0.0, "learning_rate": 4.99983990439735e-05, "loss": 1.6182, "step": 3702 }, { "epoch": 0.0, "learning_rate": 4.999839815493343e-05, "loss": 1.5683, "step": 3703 }, { "epoch": 0.0, "learning_rate": 4.999839726564659e-05, "loss": 1.559, "step": 3704 }, { "epoch": 0.0, "learning_rate": 4.999839637611298e-05, "loss": 1.5921, "step": 3705 }, { "epoch": 0.0, "learning_rate": 4.999839548633259e-05, "loss": 1.5961, "step": 3706 }, { "epoch": 0.0, "learning_rate": 4.999839459630543e-05, "loss": 1.5753, "step": 3707 }, { "epoch": 0.0, "learning_rate": 4.99983937060315e-05, "loss": 1.568, "step": 3708 }, { "epoch": 0.0, "learning_rate": 4.999839281551079e-05, "loss": 1.5881, "step": 3709 }, { "epoch": 0.0, "learning_rate": 4.999839192474331e-05, "loss": 1.5924, "step": 3710 }, { "epoch": 0.0, "learning_rate": 4.9998391033729056e-05, "loss": 1.5891, "step": 3711 }, { "epoch": 0.0, "learning_rate": 4.9998390142468026e-05, "loss": 1.5866, "step": 3712 }, { "epoch": 0.0, "learning_rate": 4.9998389250960225e-05, "loss": 1.5519, "step": 3713 }, { "epoch": 0.0, "learning_rate": 4.999838835920565e-05, "loss": 1.5676, "step": 3714 }, { "epoch": 0.0, "learning_rate": 4.99983874672043e-05, "loss": 1.5594, "step": 3715 }, { "epoch": 0.0, "learning_rate": 4.999838657495618e-05, "loss": 1.5622, "step": 3716 }, { "epoch": 0.0, "learning_rate": 4.999838568246128e-05, "loss": 1.5621, "step": 3717 }, { "epoch": 0.0, "learning_rate": 4.999838478971961e-05, "loss": 1.5302, "step": 3718 }, { "epoch": 0.0, "learning_rate": 4.999838389673117e-05, "loss": 1.5501, "step": 3719 }, { "epoch": 0.0, "learning_rate": 4.9998383003495955e-05, "loss": 1.5491, "step": 3720 }, { "epoch": 0.0, "learning_rate": 4.9998382110013964e-05, "loss": 1.5433, "step": 3721 }, { "epoch": 0.0, "learning_rate": 4.9998381216285195e-05, "loss": 1.5156, "step": 3722 }, { "epoch": 0.0, "learning_rate": 4.9998380322309654e-05, "loss": 1.5524, "step": 3723 }, { "epoch": 0.0, "learning_rate": 4.999837942808735e-05, "loss": 1.5503, "step": 3724 }, { "epoch": 0.0, "learning_rate": 4.999837853361826e-05, "loss": 1.5909, "step": 3725 }, { "epoch": 0.0, "learning_rate": 4.9998377638902406e-05, "loss": 1.5548, "step": 3726 }, { "epoch": 0.0, "learning_rate": 4.9998376743939774e-05, "loss": 1.5458, "step": 3727 }, { "epoch": 0.0, "learning_rate": 4.999837584873037e-05, "loss": 1.5581, "step": 3728 }, { "epoch": 0.0, "learning_rate": 4.999837495327419e-05, "loss": 1.5439, "step": 3729 }, { "epoch": 0.0, "learning_rate": 4.999837405757124e-05, "loss": 1.5403, "step": 3730 }, { "epoch": 0.0, "learning_rate": 4.9998373161621515e-05, "loss": 1.5434, "step": 3731 }, { "epoch": 0.0, "learning_rate": 4.999837226542502e-05, "loss": 1.5112, "step": 3732 }, { "epoch": 0.0, "learning_rate": 4.999837136898175e-05, "loss": 1.5424, "step": 3733 }, { "epoch": 0.0, "learning_rate": 4.99983704722917e-05, "loss": 1.5404, "step": 3734 }, { "epoch": 0.0, "learning_rate": 4.999836957535488e-05, "loss": 1.5589, "step": 3735 }, { "epoch": 0.0, "learning_rate": 4.999836867817129e-05, "loss": 1.5588, "step": 3736 }, { "epoch": 0.0, "learning_rate": 4.9998367780740925e-05, "loss": 1.5285, "step": 3737 }, { "epoch": 0.0, "learning_rate": 4.999836688306378e-05, "loss": 1.5774, "step": 3738 }, { "epoch": 0.0, "learning_rate": 4.999836598513987e-05, "loss": 1.5473, "step": 3739 }, { "epoch": 0.0, "learning_rate": 4.999836508696919e-05, "loss": 1.5688, "step": 3740 }, { "epoch": 0.0, "learning_rate": 4.999836418855173e-05, "loss": 1.5648, "step": 3741 }, { "epoch": 0.0, "learning_rate": 4.9998363289887496e-05, "loss": 1.4973, "step": 3742 }, { "epoch": 0.0, "learning_rate": 4.999836239097649e-05, "loss": 1.5512, "step": 3743 }, { "epoch": 0.0, "learning_rate": 4.9998361491818716e-05, "loss": 1.5802, "step": 3744 }, { "epoch": 0.0, "learning_rate": 4.999836059241416e-05, "loss": 1.5288, "step": 3745 }, { "epoch": 0.0, "learning_rate": 4.999835969276284e-05, "loss": 1.5351, "step": 3746 }, { "epoch": 0.0, "learning_rate": 4.999835879286474e-05, "loss": 1.5493, "step": 3747 }, { "epoch": 0.0, "learning_rate": 4.9998357892719866e-05, "loss": 1.5846, "step": 3748 }, { "epoch": 0.0, "learning_rate": 4.999835699232822e-05, "loss": 1.6383, "step": 3749 }, { "epoch": 0.0, "learning_rate": 4.9998356091689804e-05, "loss": 1.2697, "step": 3750 }, { "epoch": 0.0, "learning_rate": 4.9998355190804616e-05, "loss": 0.8748, "step": 3751 }, { "epoch": 0.0, "learning_rate": 4.9998354289672644e-05, "loss": 0.9201, "step": 3752 }, { "epoch": 0.0, "learning_rate": 4.9998353388293907e-05, "loss": 0.8053, "step": 3753 }, { "epoch": 0.0, "learning_rate": 4.99983524866684e-05, "loss": 0.4354, "step": 3754 }, { "epoch": 0.0, "learning_rate": 4.999835158479611e-05, "loss": 0.7464, "step": 3755 }, { "epoch": 0.0, "learning_rate": 4.9998350682677055e-05, "loss": 1.2396, "step": 3756 }, { "epoch": 0.0, "learning_rate": 4.9998349780311226e-05, "loss": 1.0182, "step": 3757 }, { "epoch": 0.0, "learning_rate": 4.999834887769862e-05, "loss": 0.9524, "step": 3758 }, { "epoch": 0.0, "learning_rate": 4.999834797483924e-05, "loss": 0.3241, "step": 3759 }, { "epoch": 0.0, "learning_rate": 4.99983470717331e-05, "loss": 0.3452, "step": 3760 }, { "epoch": 0.0, "learning_rate": 4.999834616838017e-05, "loss": 1.3963, "step": 3761 }, { "epoch": 0.0, "learning_rate": 4.9998345264780475e-05, "loss": 1.3835, "step": 3762 }, { "epoch": 0.0, "learning_rate": 4.9998344360934e-05, "loss": 3.2567, "step": 3763 }, { "epoch": 0.0, "learning_rate": 4.9998343456840765e-05, "loss": 2.4853, "step": 3764 }, { "epoch": 0.0, "learning_rate": 4.999834255250075e-05, "loss": 2.1915, "step": 3765 }, { "epoch": 0.0, "learning_rate": 4.999834164791396e-05, "loss": 2.4211, "step": 3766 }, { "epoch": 0.0, "learning_rate": 4.9998340743080397e-05, "loss": 2.2489, "step": 3767 }, { "epoch": 0.0, "learning_rate": 4.9998339838000065e-05, "loss": 2.1991, "step": 3768 }, { "epoch": 0.0, "learning_rate": 4.999833893267295e-05, "loss": 2.1998, "step": 3769 }, { "epoch": 0.0, "learning_rate": 4.9998338027099074e-05, "loss": 2.2579, "step": 3770 }, { "epoch": 0.0, "learning_rate": 4.9998337121278414e-05, "loss": 2.0534, "step": 3771 }, { "epoch": 0.0, "learning_rate": 4.9998336215211e-05, "loss": 2.1723, "step": 3772 }, { "epoch": 0.0, "learning_rate": 4.9998335308896796e-05, "loss": 1.879, "step": 3773 }, { "epoch": 0.0, "learning_rate": 4.999833440233582e-05, "loss": 1.6274, "step": 3774 }, { "epoch": 0.0, "learning_rate": 4.999833349552807e-05, "loss": 1.8824, "step": 3775 }, { "epoch": 0.0, "learning_rate": 4.999833258847355e-05, "loss": 1.5347, "step": 3776 }, { "epoch": 0.0, "learning_rate": 4.999833168117226e-05, "loss": 1.5394, "step": 3777 }, { "epoch": 0.0, "learning_rate": 4.9998330773624194e-05, "loss": 1.6869, "step": 3778 }, { "epoch": 0.0, "learning_rate": 4.999832986582935e-05, "loss": 1.8133, "step": 3779 }, { "epoch": 0.0, "learning_rate": 4.9998328957787745e-05, "loss": 1.4076, "step": 3780 }, { "epoch": 0.0, "learning_rate": 4.999832804949936e-05, "loss": 1.6906, "step": 3781 }, { "epoch": 0.0, "learning_rate": 4.9998327140964205e-05, "loss": 1.7414, "step": 3782 }, { "epoch": 0.0, "learning_rate": 4.999832623218227e-05, "loss": 1.3289, "step": 3783 }, { "epoch": 0.0, "learning_rate": 4.999832532315357e-05, "loss": 1.5075, "step": 3784 }, { "epoch": 0.0, "learning_rate": 4.99983244138781e-05, "loss": 1.2666, "step": 3785 }, { "epoch": 0.0, "learning_rate": 4.999832350435584e-05, "loss": 1.4103, "step": 3786 }, { "epoch": 0.0, "learning_rate": 4.9998322594586824e-05, "loss": 1.4504, "step": 3787 }, { "epoch": 0.0, "learning_rate": 4.999832168457103e-05, "loss": 1.4091, "step": 3788 }, { "epoch": 0.0, "learning_rate": 4.999832077430846e-05, "loss": 1.3284, "step": 3789 }, { "epoch": 0.0, "learning_rate": 4.9998319863799115e-05, "loss": 1.4356, "step": 3790 }, { "epoch": 0.0, "learning_rate": 4.9998318953043e-05, "loss": 1.2981, "step": 3791 }, { "epoch": 0.0, "learning_rate": 4.999831804204012e-05, "loss": 1.5094, "step": 3792 }, { "epoch": 0.0, "learning_rate": 4.9998317130790464e-05, "loss": 1.5671, "step": 3793 }, { "epoch": 0.0, "learning_rate": 4.999831621929403e-05, "loss": 1.4988, "step": 3794 }, { "epoch": 0.0, "learning_rate": 4.9998315307550826e-05, "loss": 1.4861, "step": 3795 }, { "epoch": 0.0, "learning_rate": 4.999831439556084e-05, "loss": 1.5498, "step": 3796 }, { "epoch": 0.0, "learning_rate": 4.999831348332409e-05, "loss": 1.2615, "step": 3797 }, { "epoch": 0.0, "learning_rate": 4.999831257084057e-05, "loss": 1.5546, "step": 3798 }, { "epoch": 0.0, "learning_rate": 4.9998311658110274e-05, "loss": 1.5435, "step": 3799 }, { "epoch": 0.0, "learning_rate": 4.99983107451332e-05, "loss": 1.6826, "step": 3800 }, { "epoch": 0.0, "learning_rate": 4.999830983190936e-05, "loss": 1.4174, "step": 3801 }, { "epoch": 0.0, "learning_rate": 4.999830891843875e-05, "loss": 1.6398, "step": 3802 }, { "epoch": 0.0, "learning_rate": 4.999830800472136e-05, "loss": 1.4567, "step": 3803 }, { "epoch": 0.0, "learning_rate": 4.99983070907572e-05, "loss": 1.6013, "step": 3804 }, { "epoch": 0.0, "learning_rate": 4.9998306176546266e-05, "loss": 1.7008, "step": 3805 }, { "epoch": 0.0, "learning_rate": 4.999830526208856e-05, "loss": 1.6242, "step": 3806 }, { "epoch": 0.0, "learning_rate": 4.9998304347384085e-05, "loss": 1.5235, "step": 3807 }, { "epoch": 0.0, "learning_rate": 4.999830343243283e-05, "loss": 1.3879, "step": 3808 }, { "epoch": 0.0, "learning_rate": 4.9998302517234805e-05, "loss": 1.4885, "step": 3809 }, { "epoch": 0.0, "learning_rate": 4.999830160179001e-05, "loss": 1.5374, "step": 3810 }, { "epoch": 0.0, "learning_rate": 4.999830068609844e-05, "loss": 1.377, "step": 3811 }, { "epoch": 0.0, "learning_rate": 4.9998299770160095e-05, "loss": 1.2827, "step": 3812 }, { "epoch": 0.0, "learning_rate": 4.9998298853974985e-05, "loss": 1.3945, "step": 3813 }, { "epoch": 0.0, "learning_rate": 4.99982979375431e-05, "loss": 1.3208, "step": 3814 }, { "epoch": 0.0, "learning_rate": 4.999829702086444e-05, "loss": 1.276, "step": 3815 }, { "epoch": 0.0, "learning_rate": 4.9998296103939e-05, "loss": 1.1584, "step": 3816 }, { "epoch": 0.0, "learning_rate": 4.99982951867668e-05, "loss": 1.002, "step": 3817 }, { "epoch": 0.0, "learning_rate": 4.9998294269347826e-05, "loss": 0.9674, "step": 3818 }, { "epoch": 0.0, "learning_rate": 4.999829335168207e-05, "loss": 0.9041, "step": 3819 }, { "epoch": 0.0, "learning_rate": 4.999829243376955e-05, "loss": 0.6312, "step": 3820 }, { "epoch": 0.0, "learning_rate": 4.9998291515610254e-05, "loss": 0.6914, "step": 3821 }, { "epoch": 0.0, "learning_rate": 4.999829059720418e-05, "loss": 0.5184, "step": 3822 }, { "epoch": 0.0, "learning_rate": 4.999828967855134e-05, "loss": 0.3013, "step": 3823 }, { "epoch": 0.0, "learning_rate": 4.9998288759651727e-05, "loss": 0.2505, "step": 3824 }, { "epoch": 0.0, "learning_rate": 4.999828784050534e-05, "loss": 0.2985, "step": 3825 }, { "epoch": 0.0, "learning_rate": 4.999828692111218e-05, "loss": 0.899, "step": 3826 }, { "epoch": 0.0, "learning_rate": 4.999828600147225e-05, "loss": 1.2663, "step": 3827 }, { "epoch": 0.0, "learning_rate": 4.999828508158555e-05, "loss": 1.2636, "step": 3828 }, { "epoch": 0.0, "learning_rate": 4.999828416145207e-05, "loss": 1.131, "step": 3829 }, { "epoch": 0.0, "learning_rate": 4.999828324107182e-05, "loss": 1.1377, "step": 3830 }, { "epoch": 0.0, "learning_rate": 4.99982823204448e-05, "loss": 1.1853, "step": 3831 }, { "epoch": 0.0, "learning_rate": 4.9998281399571015e-05, "loss": 1.2119, "step": 3832 }, { "epoch": 0.0, "learning_rate": 4.999828047845044e-05, "loss": 1.1701, "step": 3833 }, { "epoch": 0.0, "learning_rate": 4.999827955708311e-05, "loss": 1.2334, "step": 3834 }, { "epoch": 0.0, "learning_rate": 4.999827863546899e-05, "loss": 0.8946, "step": 3835 }, { "epoch": 0.0, "learning_rate": 4.999827771360811e-05, "loss": 1.0699, "step": 3836 }, { "epoch": 0.0, "learning_rate": 4.999827679150045e-05, "loss": 1.108, "step": 3837 }, { "epoch": 0.0, "learning_rate": 4.999827586914602e-05, "loss": 1.0536, "step": 3838 }, { "epoch": 0.0, "learning_rate": 4.999827494654482e-05, "loss": 1.1607, "step": 3839 }, { "epoch": 0.0, "learning_rate": 4.9998274023696845e-05, "loss": 1.0009, "step": 3840 }, { "epoch": 0.0, "learning_rate": 4.99982731006021e-05, "loss": 1.0169, "step": 3841 }, { "epoch": 0.0, "learning_rate": 4.9998272177260586e-05, "loss": 1.2903, "step": 3842 }, { "epoch": 0.0, "learning_rate": 4.999827125367229e-05, "loss": 1.0584, "step": 3843 }, { "epoch": 0.0, "learning_rate": 4.999827032983723e-05, "loss": 1.0524, "step": 3844 }, { "epoch": 0.0, "learning_rate": 4.999826940575539e-05, "loss": 0.9983, "step": 3845 }, { "epoch": 0.0, "learning_rate": 4.9998268481426786e-05, "loss": 0.9049, "step": 3846 }, { "epoch": 0.0, "learning_rate": 4.999826755685141e-05, "loss": 1.0455, "step": 3847 }, { "epoch": 0.0, "learning_rate": 4.999826663202926e-05, "loss": 1.2578, "step": 3848 }, { "epoch": 0.0, "learning_rate": 4.9998265706960325e-05, "loss": 1.1371, "step": 3849 }, { "epoch": 0.0, "learning_rate": 4.999826478164463e-05, "loss": 1.0095, "step": 3850 }, { "epoch": 0.0, "learning_rate": 4.9998263856082164e-05, "loss": 1.1335, "step": 3851 }, { "epoch": 0.0, "learning_rate": 4.999826293027292e-05, "loss": 1.1164, "step": 3852 }, { "epoch": 0.0, "learning_rate": 4.99982620042169e-05, "loss": 0.9897, "step": 3853 }, { "epoch": 0.0, "learning_rate": 4.9998261077914114e-05, "loss": 1.1324, "step": 3854 }, { "epoch": 0.0, "learning_rate": 4.999826015136456e-05, "loss": 0.9199, "step": 3855 }, { "epoch": 0.0, "learning_rate": 4.999825922456822e-05, "loss": 0.6164, "step": 3856 }, { "epoch": 0.0, "learning_rate": 4.9998258297525124e-05, "loss": 1.0912, "step": 3857 }, { "epoch": 0.0, "learning_rate": 4.999825737023525e-05, "loss": 1.1302, "step": 3858 }, { "epoch": 0.0, "learning_rate": 4.99982564426986e-05, "loss": 0.8905, "step": 3859 }, { "epoch": 0.0, "learning_rate": 4.999825551491518e-05, "loss": 0.9626, "step": 3860 }, { "epoch": 0.0, "learning_rate": 4.9998254586884985e-05, "loss": 0.9578, "step": 3861 }, { "epoch": 0.0, "learning_rate": 4.999825365860802e-05, "loss": 0.9679, "step": 3862 }, { "epoch": 0.0, "learning_rate": 4.999825273008428e-05, "loss": 1.0489, "step": 3863 }, { "epoch": 0.0, "learning_rate": 4.999825180131377e-05, "loss": 0.7817, "step": 3864 }, { "epoch": 0.0, "learning_rate": 4.999825087229649e-05, "loss": 0.5133, "step": 3865 }, { "epoch": 0.0, "learning_rate": 4.9998249943032435e-05, "loss": 0.5381, "step": 3866 }, { "epoch": 0.0, "learning_rate": 4.999824901352161e-05, "loss": 0.5507, "step": 3867 }, { "epoch": 0.0, "learning_rate": 4.999824808376401e-05, "loss": 0.8889, "step": 3868 }, { "epoch": 0.0, "learning_rate": 4.999824715375965e-05, "loss": 0.5353, "step": 3869 }, { "epoch": 0.0, "learning_rate": 4.9998246223508506e-05, "loss": 0.484, "step": 3870 }, { "epoch": 0.0, "learning_rate": 4.9998245293010585e-05, "loss": 0.3894, "step": 3871 }, { "epoch": 0.0, "learning_rate": 4.99982443622659e-05, "loss": 0.3145, "step": 3872 }, { "epoch": 0.0, "learning_rate": 4.999824343127444e-05, "loss": 1.4916, "step": 3873 }, { "epoch": 0.0, "learning_rate": 4.999824250003621e-05, "loss": 0.9999, "step": 3874 }, { "epoch": 0.0, "learning_rate": 4.9998241568551205e-05, "loss": 0.8343, "step": 3875 }, { "epoch": 0.0, "learning_rate": 4.9998240636819436e-05, "loss": 1.1243, "step": 3876 }, { "epoch": 0.0, "learning_rate": 4.999823970484089e-05, "loss": 1.0399, "step": 3877 }, { "epoch": 0.0, "learning_rate": 4.999823877261557e-05, "loss": 0.8224, "step": 3878 }, { "epoch": 0.0, "learning_rate": 4.999823784014347e-05, "loss": 1.0171, "step": 3879 }, { "epoch": 0.0, "learning_rate": 4.999823690742461e-05, "loss": 1.9519, "step": 3880 }, { "epoch": 0.0, "learning_rate": 4.999823597445897e-05, "loss": 1.0492, "step": 3881 }, { "epoch": 0.0, "learning_rate": 4.999823504124657e-05, "loss": 1.2882, "step": 3882 }, { "epoch": 0.0, "learning_rate": 4.999823410778739e-05, "loss": 1.1063, "step": 3883 }, { "epoch": 0.0, "learning_rate": 4.9998233174081434e-05, "loss": 0.9714, "step": 3884 }, { "epoch": 0.0, "learning_rate": 4.999823224012871e-05, "loss": 0.9269, "step": 3885 }, { "epoch": 0.0, "learning_rate": 4.9998231305929215e-05, "loss": 1.3882, "step": 3886 }, { "epoch": 0.0, "learning_rate": 4.999823037148295e-05, "loss": 1.2153, "step": 3887 }, { "epoch": 0.0, "learning_rate": 4.9998229436789905e-05, "loss": 0.9109, "step": 3888 }, { "epoch": 0.0, "learning_rate": 4.9998228501850096e-05, "loss": 0.9884, "step": 3889 }, { "epoch": 0.0, "learning_rate": 4.9998227566663516e-05, "loss": 1.0662, "step": 3890 }, { "epoch": 0.0, "learning_rate": 4.999822663123016e-05, "loss": 0.9719, "step": 3891 }, { "epoch": 0.0, "learning_rate": 4.999822569555003e-05, "loss": 1.3314, "step": 3892 }, { "epoch": 0.0, "learning_rate": 4.999822475962313e-05, "loss": 1.1694, "step": 3893 }, { "epoch": 0.0, "learning_rate": 4.999822382344946e-05, "loss": 1.3414, "step": 3894 }, { "epoch": 0.0, "learning_rate": 4.999822288702901e-05, "loss": 1.2034, "step": 3895 }, { "epoch": 0.0, "learning_rate": 4.99982219503618e-05, "loss": 1.2633, "step": 3896 }, { "epoch": 0.0, "learning_rate": 4.999822101344781e-05, "loss": 1.0826, "step": 3897 }, { "epoch": 0.0, "learning_rate": 4.999822007628705e-05, "loss": 0.9299, "step": 3898 }, { "epoch": 0.0, "learning_rate": 4.999821913887952e-05, "loss": 1.712, "step": 3899 }, { "epoch": 0.0, "learning_rate": 4.9998218201225214e-05, "loss": 1.2466, "step": 3900 }, { "epoch": 0.0, "learning_rate": 4.9998217263324144e-05, "loss": 1.0934, "step": 3901 }, { "epoch": 0.0, "learning_rate": 4.9998216325176296e-05, "loss": 0.7847, "step": 3902 }, { "epoch": 0.0, "learning_rate": 4.999821538678168e-05, "loss": 1.2325, "step": 3903 }, { "epoch": 0.0, "learning_rate": 4.999821444814029e-05, "loss": 1.2659, "step": 3904 }, { "epoch": 0.0, "learning_rate": 4.9998213509252126e-05, "loss": 0.9979, "step": 3905 }, { "epoch": 0.0, "learning_rate": 4.999821257011719e-05, "loss": 1.2732, "step": 3906 }, { "epoch": 0.0, "learning_rate": 4.999821163073548e-05, "loss": 0.6001, "step": 3907 }, { "epoch": 0.0, "learning_rate": 4.999821069110701e-05, "loss": 1.0395, "step": 3908 }, { "epoch": 0.0, "learning_rate": 4.9998209751231755e-05, "loss": 0.9985, "step": 3909 }, { "epoch": 0.0, "learning_rate": 4.999820881110974e-05, "loss": 0.8775, "step": 3910 }, { "epoch": 0.0, "learning_rate": 4.999820787074094e-05, "loss": 1.2523, "step": 3911 }, { "epoch": 0.0, "learning_rate": 4.999820693012538e-05, "loss": 1.4449, "step": 3912 }, { "epoch": 0.0, "learning_rate": 4.999820598926304e-05, "loss": 1.343, "step": 3913 }, { "epoch": 0.0, "learning_rate": 4.999820504815394e-05, "loss": 1.4019, "step": 3914 }, { "epoch": 0.0, "learning_rate": 4.999820410679805e-05, "loss": 1.1844, "step": 3915 }, { "epoch": 0.0, "learning_rate": 4.99982031651954e-05, "loss": 1.3271, "step": 3916 }, { "epoch": 0.0, "learning_rate": 4.999820222334598e-05, "loss": 1.2497, "step": 3917 }, { "epoch": 0.0, "learning_rate": 4.9998201281249776e-05, "loss": 0.9437, "step": 3918 }, { "epoch": 0.0, "learning_rate": 4.999820033890681e-05, "loss": 0.9859, "step": 3919 }, { "epoch": 0.0, "learning_rate": 4.9998199396317076e-05, "loss": 1.2831, "step": 3920 }, { "epoch": 0.0, "learning_rate": 4.999819845348056e-05, "loss": 1.0626, "step": 3921 }, { "epoch": 0.0, "learning_rate": 4.9998197510397284e-05, "loss": 1.1052, "step": 3922 }, { "epoch": 0.0, "learning_rate": 4.999819656706722e-05, "loss": 0.8837, "step": 3923 }, { "epoch": 0.0, "learning_rate": 4.99981956234904e-05, "loss": 1.1141, "step": 3924 }, { "epoch": 0.0, "learning_rate": 4.99981946796668e-05, "loss": 0.9994, "step": 3925 }, { "epoch": 0.0, "learning_rate": 4.999819373559643e-05, "loss": 1.1416, "step": 3926 }, { "epoch": 0.0, "learning_rate": 4.99981927912793e-05, "loss": 1.0897, "step": 3927 }, { "epoch": 0.0, "learning_rate": 4.999819184671538e-05, "loss": 1.2305, "step": 3928 }, { "epoch": 0.0, "learning_rate": 4.99981909019047e-05, "loss": 1.2005, "step": 3929 }, { "epoch": 0.0, "learning_rate": 4.999818995684724e-05, "loss": 0.9295, "step": 3930 }, { "epoch": 0.0, "learning_rate": 4.9998189011543014e-05, "loss": 1.2663, "step": 3931 }, { "epoch": 0.0, "learning_rate": 4.999818806599202e-05, "loss": 1.0269, "step": 3932 }, { "epoch": 0.0, "learning_rate": 4.999818712019425e-05, "loss": 1.0723, "step": 3933 }, { "epoch": 0.0, "learning_rate": 4.999818617414971e-05, "loss": 1.3211, "step": 3934 }, { "epoch": 0.0, "learning_rate": 4.99981852278584e-05, "loss": 1.2296, "step": 3935 }, { "epoch": 0.0, "learning_rate": 4.999818428132031e-05, "loss": 1.2031, "step": 3936 }, { "epoch": 0.0, "learning_rate": 4.999818333453545e-05, "loss": 0.8096, "step": 3937 }, { "epoch": 0.0, "learning_rate": 4.999818238750382e-05, "loss": 1.1207, "step": 3938 }, { "epoch": 0.0, "learning_rate": 4.9998181440225425e-05, "loss": 1.106, "step": 3939 }, { "epoch": 0.0, "learning_rate": 4.999818049270025e-05, "loss": 1.1441, "step": 3940 }, { "epoch": 0.0, "learning_rate": 4.9998179544928314e-05, "loss": 1.1043, "step": 3941 }, { "epoch": 0.0, "learning_rate": 4.99981785969096e-05, "loss": 0.7397, "step": 3942 }, { "epoch": 0.0, "learning_rate": 4.999817764864411e-05, "loss": 0.4566, "step": 3943 }, { "epoch": 0.0, "learning_rate": 4.999817670013186e-05, "loss": 0.3368, "step": 3944 }, { "epoch": 0.0, "learning_rate": 4.999817575137283e-05, "loss": 0.2477, "step": 3945 }, { "epoch": 0.0, "learning_rate": 4.9998174802367024e-05, "loss": 0.1339, "step": 3946 }, { "epoch": 0.0, "learning_rate": 4.999817385311446e-05, "loss": 0.1126, "step": 3947 }, { "epoch": 0.0, "learning_rate": 4.9998172903615116e-05, "loss": 1.0654, "step": 3948 }, { "epoch": 0.0, "learning_rate": 4.9998171953869e-05, "loss": 1.0436, "step": 3949 }, { "epoch": 0.0, "learning_rate": 4.999817100387612e-05, "loss": 1.3441, "step": 3950 }, { "epoch": 0.0, "learning_rate": 4.999817005363646e-05, "loss": 1.4435, "step": 3951 }, { "epoch": 0.0, "learning_rate": 4.9998169103150026e-05, "loss": 1.1704, "step": 3952 }, { "epoch": 0.0, "learning_rate": 4.999816815241683e-05, "loss": 0.2951, "step": 3953 }, { "epoch": 0.0, "learning_rate": 4.9998167201436864e-05, "loss": 1.0192, "step": 3954 }, { "epoch": 0.0, "learning_rate": 4.9998166250210116e-05, "loss": 1.0236, "step": 3955 }, { "epoch": 0.0, "learning_rate": 4.9998165298736604e-05, "loss": 1.1952, "step": 3956 }, { "epoch": 0.0, "learning_rate": 4.999816434701632e-05, "loss": 1.1598, "step": 3957 }, { "epoch": 0.0, "learning_rate": 4.999816339504926e-05, "loss": 1.2215, "step": 3958 }, { "epoch": 0.0, "learning_rate": 4.999816244283543e-05, "loss": 1.4891, "step": 3959 }, { "epoch": 0.0, "learning_rate": 4.9998161490374836e-05, "loss": 1.1688, "step": 3960 }, { "epoch": 0.0, "learning_rate": 4.999816053766746e-05, "loss": 1.1998, "step": 3961 }, { "epoch": 0.0, "learning_rate": 4.999815958471332e-05, "loss": 1.111, "step": 3962 }, { "epoch": 0.0, "learning_rate": 4.999815863151241e-05, "loss": 1.1179, "step": 3963 }, { "epoch": 0.0, "learning_rate": 4.999815767806473e-05, "loss": 1.0286, "step": 3964 }, { "epoch": 0.0, "learning_rate": 4.999815672437027e-05, "loss": 1.1219, "step": 3965 }, { "epoch": 0.0, "learning_rate": 4.9998155770429045e-05, "loss": 1.2234, "step": 3966 }, { "epoch": 0.0, "learning_rate": 4.999815481624104e-05, "loss": 1.1453, "step": 3967 }, { "epoch": 0.0, "learning_rate": 4.9998153861806276e-05, "loss": 1.1588, "step": 3968 }, { "epoch": 0.0, "learning_rate": 4.999815290712473e-05, "loss": 0.6494, "step": 3969 }, { "epoch": 0.0, "learning_rate": 4.999815195219643e-05, "loss": 0.6897, "step": 3970 }, { "epoch": 0.0, "learning_rate": 4.999815099702134e-05, "loss": 1.1793, "step": 3971 }, { "epoch": 0.0, "learning_rate": 4.9998150041599484e-05, "loss": 1.0342, "step": 3972 }, { "epoch": 0.0, "learning_rate": 4.9998149085930854e-05, "loss": 1.1482, "step": 3973 }, { "epoch": 0.0, "learning_rate": 4.999814813001547e-05, "loss": 1.2897, "step": 3974 }, { "epoch": 0.0, "learning_rate": 4.9998147173853296e-05, "loss": 0.7381, "step": 3975 }, { "epoch": 0.0, "learning_rate": 4.999814621744435e-05, "loss": 1.1291, "step": 3976 }, { "epoch": 0.0, "learning_rate": 4.9998145260788645e-05, "loss": 1.1105, "step": 3977 }, { "epoch": 0.0, "learning_rate": 4.999814430388616e-05, "loss": 1.2495, "step": 3978 }, { "epoch": 0.0, "learning_rate": 4.999814334673692e-05, "loss": 1.0232, "step": 3979 }, { "epoch": 0.0, "learning_rate": 4.999814238934088e-05, "loss": 1.4272, "step": 3980 }, { "epoch": 0.0, "learning_rate": 4.999814143169809e-05, "loss": 1.3486, "step": 3981 }, { "epoch": 0.0, "learning_rate": 4.999814047380853e-05, "loss": 1.285, "step": 3982 }, { "epoch": 0.0, "learning_rate": 4.999813951567219e-05, "loss": 1.105, "step": 3983 }, { "epoch": 0.0, "learning_rate": 4.999813855728908e-05, "loss": 0.9492, "step": 3984 }, { "epoch": 0.0, "learning_rate": 4.9998137598659204e-05, "loss": 1.1968, "step": 3985 }, { "epoch": 0.0, "learning_rate": 4.999813663978255e-05, "loss": 0.9647, "step": 3986 }, { "epoch": 0.0, "learning_rate": 4.999813568065913e-05, "loss": 1.1586, "step": 3987 }, { "epoch": 0.0, "learning_rate": 4.999813472128894e-05, "loss": 1.2225, "step": 3988 }, { "epoch": 0.0, "learning_rate": 4.999813376167197e-05, "loss": 1.072, "step": 3989 }, { "epoch": 0.0, "learning_rate": 4.999813280180824e-05, "loss": 1.1902, "step": 3990 }, { "epoch": 0.0, "learning_rate": 4.999813184169774e-05, "loss": 1.1988, "step": 3991 }, { "epoch": 0.0, "learning_rate": 4.999813088134045e-05, "loss": 1.206, "step": 3992 }, { "epoch": 0.0, "learning_rate": 4.9998129920736405e-05, "loss": 1.1791, "step": 3993 }, { "epoch": 0.0, "learning_rate": 4.9998128959885595e-05, "loss": 1.5142, "step": 3994 }, { "epoch": 0.0, "learning_rate": 4.9998127998788006e-05, "loss": 1.4691, "step": 3995 }, { "epoch": 0.0, "learning_rate": 4.9998127037443646e-05, "loss": 1.4362, "step": 3996 }, { "epoch": 0.0, "learning_rate": 4.999812607585251e-05, "loss": 1.2838, "step": 3997 }, { "epoch": 0.0, "learning_rate": 4.9998125114014606e-05, "loss": 0.9268, "step": 3998 }, { "epoch": 0.0, "learning_rate": 4.999812415192994e-05, "loss": 0.9915, "step": 3999 }, { "epoch": 0.0, "learning_rate": 4.999812318959849e-05, "loss": 1.1468, "step": 4000 }, { "epoch": 0.0, "eval_loss": 1.09718918800354, "eval_runtime": 106.6196, "eval_samples_per_second": 12.99, "eval_steps_per_second": 3.255, "step": 4000 }, { "epoch": 0.0, "learning_rate": 4.999812222702027e-05, "loss": 0.8135, "step": 4001 }, { "epoch": 0.0, "learning_rate": 4.9998121264195284e-05, "loss": 1.2887, "step": 4002 }, { "epoch": 0.0, "learning_rate": 4.999812030112353e-05, "loss": 0.9391, "step": 4003 }, { "epoch": 0.0, "learning_rate": 4.9998119337805e-05, "loss": 1.1028, "step": 4004 }, { "epoch": 0.0, "learning_rate": 4.99981183742397e-05, "loss": 1.8483, "step": 4005 }, { "epoch": 0.0, "learning_rate": 4.999811741042764e-05, "loss": 0.5112, "step": 4006 }, { "epoch": 0.0, "learning_rate": 4.9998116446368794e-05, "loss": 0.9238, "step": 4007 }, { "epoch": 0.0, "learning_rate": 4.999811548206318e-05, "loss": 1.0704, "step": 4008 }, { "epoch": 0.0, "learning_rate": 4.9998114517510795e-05, "loss": 0.8911, "step": 4009 }, { "epoch": 0.0, "learning_rate": 4.9998113552711645e-05, "loss": 0.7326, "step": 4010 }, { "epoch": 0.0, "learning_rate": 4.999811258766572e-05, "loss": 0.5943, "step": 4011 }, { "epoch": 0.0, "learning_rate": 4.9998111622373025e-05, "loss": 0.9056, "step": 4012 }, { "epoch": 0.0, "learning_rate": 4.9998110656833555e-05, "loss": 1.0, "step": 4013 }, { "epoch": 0.0, "learning_rate": 4.999810969104733e-05, "loss": 1.0892, "step": 4014 }, { "epoch": 0.0, "learning_rate": 4.9998108725014314e-05, "loss": 0.989, "step": 4015 }, { "epoch": 0.0, "learning_rate": 4.999810775873454e-05, "loss": 1.1427, "step": 4016 }, { "epoch": 0.0, "learning_rate": 4.999810679220799e-05, "loss": 1.3094, "step": 4017 }, { "epoch": 0.0, "learning_rate": 4.999810582543466e-05, "loss": 1.5354, "step": 4018 }, { "epoch": 0.0, "learning_rate": 4.999810485841457e-05, "loss": 1.2071, "step": 4019 }, { "epoch": 0.0, "learning_rate": 4.999810389114772e-05, "loss": 1.0141, "step": 4020 }, { "epoch": 0.0, "learning_rate": 4.9998102923634084e-05, "loss": 1.1229, "step": 4021 }, { "epoch": 0.0, "learning_rate": 4.999810195587368e-05, "loss": 1.4388, "step": 4022 }, { "epoch": 0.0, "learning_rate": 4.9998100987866505e-05, "loss": 1.651, "step": 4023 }, { "epoch": 0.0, "learning_rate": 4.999810001961256e-05, "loss": 1.1134, "step": 4024 }, { "epoch": 0.0, "learning_rate": 4.999809905111185e-05, "loss": 1.1017, "step": 4025 }, { "epoch": 0.0, "learning_rate": 4.9998098082364365e-05, "loss": 1.0174, "step": 4026 }, { "epoch": 0.0, "learning_rate": 4.99980971133701e-05, "loss": 0.3147, "step": 4027 }, { "epoch": 0.0, "learning_rate": 4.999809614412908e-05, "loss": 1.336, "step": 4028 }, { "epoch": 0.0, "learning_rate": 4.999809517464128e-05, "loss": 0.9984, "step": 4029 }, { "epoch": 0.0, "learning_rate": 4.999809420490671e-05, "loss": 0.9577, "step": 4030 }, { "epoch": 0.0, "learning_rate": 4.9998093234925366e-05, "loss": 1.3655, "step": 4031 }, { "epoch": 0.0, "learning_rate": 4.9998092264697264e-05, "loss": 0.8947, "step": 4032 }, { "epoch": 0.0, "learning_rate": 4.999809129422238e-05, "loss": 1.1589, "step": 4033 }, { "epoch": 0.0, "learning_rate": 4.999809032350073e-05, "loss": 1.2562, "step": 4034 }, { "epoch": 0.0, "learning_rate": 4.999808935253231e-05, "loss": 0.9964, "step": 4035 }, { "epoch": 0.0, "learning_rate": 4.999808838131712e-05, "loss": 1.249, "step": 4036 }, { "epoch": 0.0, "learning_rate": 4.9998087409855156e-05, "loss": 1.2319, "step": 4037 }, { "epoch": 0.0, "learning_rate": 4.999808643814642e-05, "loss": 1.2463, "step": 4038 }, { "epoch": 0.0, "learning_rate": 4.999808546619091e-05, "loss": 1.2636, "step": 4039 }, { "epoch": 0.0, "learning_rate": 4.999808449398864e-05, "loss": 1.3586, "step": 4040 }, { "epoch": 0.0, "learning_rate": 4.99980835215396e-05, "loss": 1.1718, "step": 4041 }, { "epoch": 0.0, "learning_rate": 4.999808254884378e-05, "loss": 1.187, "step": 4042 }, { "epoch": 0.0, "learning_rate": 4.99980815759012e-05, "loss": 1.0817, "step": 4043 }, { "epoch": 0.0, "learning_rate": 4.999808060271184e-05, "loss": 1.162, "step": 4044 }, { "epoch": 0.0, "learning_rate": 4.9998079629275705e-05, "loss": 1.1909, "step": 4045 }, { "epoch": 0.0, "learning_rate": 4.999807865559282e-05, "loss": 1.1345, "step": 4046 }, { "epoch": 0.0, "learning_rate": 4.999807768166315e-05, "loss": 0.9697, "step": 4047 }, { "epoch": 0.0, "learning_rate": 4.999807670748671e-05, "loss": 0.9829, "step": 4048 }, { "epoch": 0.0, "learning_rate": 4.99980757330635e-05, "loss": 1.131, "step": 4049 }, { "epoch": 0.0, "learning_rate": 4.999807475839352e-05, "loss": 1.06, "step": 4050 }, { "epoch": 0.0, "learning_rate": 4.999807378347677e-05, "loss": 1.0493, "step": 4051 }, { "epoch": 0.0, "learning_rate": 4.999807280831324e-05, "loss": 0.7613, "step": 4052 }, { "epoch": 0.0, "learning_rate": 4.999807183290296e-05, "loss": 3.0747, "step": 4053 }, { "epoch": 0.0, "learning_rate": 4.999807085724589e-05, "loss": 4.4983, "step": 4054 }, { "epoch": 0.0, "learning_rate": 4.9998069881342064e-05, "loss": 4.1956, "step": 4055 }, { "epoch": 0.0, "learning_rate": 4.999806890519146e-05, "loss": 4.0042, "step": 4056 }, { "epoch": 0.0, "learning_rate": 4.999806792879409e-05, "loss": 3.661, "step": 4057 }, { "epoch": 0.0, "learning_rate": 4.9998066952149945e-05, "loss": 3.8345, "step": 4058 }, { "epoch": 0.0, "learning_rate": 4.9998065975259034e-05, "loss": 3.6504, "step": 4059 }, { "epoch": 0.0, "learning_rate": 4.9998064998121344e-05, "loss": 1.5604, "step": 4060 }, { "epoch": 0.0, "learning_rate": 4.999806402073689e-05, "loss": 0.8872, "step": 4061 }, { "epoch": 0.0, "learning_rate": 4.9998063043105666e-05, "loss": 0.9651, "step": 4062 }, { "epoch": 0.0, "learning_rate": 4.999806206522767e-05, "loss": 1.0188, "step": 4063 }, { "epoch": 0.0, "learning_rate": 4.999806108710291e-05, "loss": 1.4743, "step": 4064 }, { "epoch": 0.0, "learning_rate": 4.999806010873137e-05, "loss": 1.2101, "step": 4065 }, { "epoch": 0.0, "learning_rate": 4.999805913011307e-05, "loss": 1.4814, "step": 4066 }, { "epoch": 0.0, "learning_rate": 4.999805815124799e-05, "loss": 0.996, "step": 4067 }, { "epoch": 0.0, "learning_rate": 4.9998057172136144e-05, "loss": 0.9714, "step": 4068 }, { "epoch": 0.0, "learning_rate": 4.999805619277753e-05, "loss": 0.9029, "step": 4069 }, { "epoch": 0.0, "learning_rate": 4.999805521317214e-05, "loss": 1.2617, "step": 4070 }, { "epoch": 0.0, "learning_rate": 4.999805423331998e-05, "loss": 0.703, "step": 4071 }, { "epoch": 0.0, "learning_rate": 4.999805325322106e-05, "loss": 1.2638, "step": 4072 }, { "epoch": 0.0, "learning_rate": 4.999805227287536e-05, "loss": 1.2354, "step": 4073 }, { "epoch": 0.0, "learning_rate": 4.9998051292282885e-05, "loss": 1.0022, "step": 4074 }, { "epoch": 0.0, "learning_rate": 4.999805031144365e-05, "loss": 1.1542, "step": 4075 }, { "epoch": 0.0, "learning_rate": 4.999804933035764e-05, "loss": 0.9708, "step": 4076 }, { "epoch": 0.0, "learning_rate": 4.999804834902486e-05, "loss": 1.0911, "step": 4077 }, { "epoch": 0.0, "learning_rate": 4.999804736744531e-05, "loss": 1.4731, "step": 4078 }, { "epoch": 0.0, "learning_rate": 4.9998046385619e-05, "loss": 1.1589, "step": 4079 }, { "epoch": 0.0, "learning_rate": 4.9998045403545904e-05, "loss": 1.025, "step": 4080 }, { "epoch": 0.0, "learning_rate": 4.999804442122604e-05, "loss": 0.9713, "step": 4081 }, { "epoch": 0.0, "learning_rate": 4.999804343865941e-05, "loss": 1.3273, "step": 4082 }, { "epoch": 0.0, "learning_rate": 4.999804245584602e-05, "loss": 1.2429, "step": 4083 }, { "epoch": 0.0, "learning_rate": 4.999804147278585e-05, "loss": 0.9776, "step": 4084 }, { "epoch": 0.0, "learning_rate": 4.99980404894789e-05, "loss": 0.8755, "step": 4085 }, { "epoch": 0.0, "learning_rate": 4.99980395059252e-05, "loss": 1.3534, "step": 4086 }, { "epoch": 0.0, "learning_rate": 4.999803852212471e-05, "loss": 1.581, "step": 4087 }, { "epoch": 0.0, "learning_rate": 4.999803753807747e-05, "loss": 2.4373, "step": 4088 }, { "epoch": 0.0, "learning_rate": 4.999803655378344e-05, "loss": 3.114, "step": 4089 }, { "epoch": 0.0, "learning_rate": 4.999803556924265e-05, "loss": 1.136, "step": 4090 }, { "epoch": 0.0, "learning_rate": 4.9998034584455096e-05, "loss": 1.281, "step": 4091 }, { "epoch": 0.0, "learning_rate": 4.999803359942076e-05, "loss": 1.1828, "step": 4092 }, { "epoch": 0.0, "learning_rate": 4.999803261413966e-05, "loss": 1.0111, "step": 4093 }, { "epoch": 0.0, "learning_rate": 4.999803162861179e-05, "loss": 1.7322, "step": 4094 }, { "epoch": 0.0, "learning_rate": 4.999803064283715e-05, "loss": 1.4425, "step": 4095 }, { "epoch": 0.0, "learning_rate": 4.999802965681574e-05, "loss": 1.3707, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.9998028670547556e-05, "loss": 1.4249, "step": 4097 }, { "epoch": 0.0, "learning_rate": 4.999802768403261e-05, "loss": 1.0745, "step": 4098 }, { "epoch": 0.0, "learning_rate": 4.999802669727089e-05, "loss": 0.8969, "step": 4099 }, { "epoch": 0.0, "learning_rate": 4.9998025710262395e-05, "loss": 1.0658, "step": 4100 }, { "epoch": 0.0, "learning_rate": 4.9998024723007134e-05, "loss": 0.9743, "step": 4101 }, { "epoch": 0.0, "learning_rate": 4.999802373550511e-05, "loss": 1.1153, "step": 4102 }, { "epoch": 0.0, "learning_rate": 4.99980227477563e-05, "loss": 1.1346, "step": 4103 }, { "epoch": 0.0, "learning_rate": 4.999802175976073e-05, "loss": 0.8344, "step": 4104 }, { "epoch": 0.0, "learning_rate": 4.9998020771518394e-05, "loss": 0.1926, "step": 4105 }, { "epoch": 0.0, "learning_rate": 4.9998019783029284e-05, "loss": 0.2064, "step": 4106 }, { "epoch": 0.0, "learning_rate": 4.99980187942934e-05, "loss": 0.3738, "step": 4107 }, { "epoch": 0.0, "learning_rate": 4.999801780531075e-05, "loss": 0.3192, "step": 4108 }, { "epoch": 0.0, "learning_rate": 4.999801681608133e-05, "loss": 0.3141, "step": 4109 }, { "epoch": 0.0, "learning_rate": 4.999801582660514e-05, "loss": 0.302, "step": 4110 }, { "epoch": 0.0, "learning_rate": 4.999801483688218e-05, "loss": 0.3193, "step": 4111 }, { "epoch": 0.0, "learning_rate": 4.999801384691245e-05, "loss": 0.2444, "step": 4112 }, { "epoch": 0.0, "learning_rate": 4.999801285669595e-05, "loss": 0.2959, "step": 4113 }, { "epoch": 0.0, "learning_rate": 4.999801186623268e-05, "loss": 0.1291, "step": 4114 }, { "epoch": 0.0, "learning_rate": 4.999801087552264e-05, "loss": 0.3326, "step": 4115 }, { "epoch": 0.0, "learning_rate": 4.9998009884565834e-05, "loss": 1.5303, "step": 4116 }, { "epoch": 0.0, "learning_rate": 4.9998008893362254e-05, "loss": 0.9613, "step": 4117 }, { "epoch": 0.0, "learning_rate": 4.9998007901911904e-05, "loss": 1.2479, "step": 4118 }, { "epoch": 0.0, "learning_rate": 4.999800691021479e-05, "loss": 1.2053, "step": 4119 }, { "epoch": 0.0, "learning_rate": 4.9998005918270897e-05, "loss": 0.9582, "step": 4120 }, { "epoch": 0.0, "learning_rate": 4.999800492608024e-05, "loss": 1.4681, "step": 4121 }, { "epoch": 0.0, "learning_rate": 4.999800393364281e-05, "loss": 1.3266, "step": 4122 }, { "epoch": 0.0, "learning_rate": 4.999800294095861e-05, "loss": 1.107, "step": 4123 }, { "epoch": 0.0, "learning_rate": 4.999800194802765e-05, "loss": 1.2981, "step": 4124 }, { "epoch": 0.0, "learning_rate": 4.999800095484991e-05, "loss": 1.0812, "step": 4125 }, { "epoch": 0.0, "learning_rate": 4.99979999614254e-05, "loss": 0.9215, "step": 4126 }, { "epoch": 0.0, "learning_rate": 4.999799896775412e-05, "loss": 1.056, "step": 4127 }, { "epoch": 0.0, "learning_rate": 4.999799797383607e-05, "loss": 0.9254, "step": 4128 }, { "epoch": 0.0, "learning_rate": 4.999799697967126e-05, "loss": 1.2472, "step": 4129 }, { "epoch": 0.0, "learning_rate": 4.999799598525967e-05, "loss": 0.6347, "step": 4130 }, { "epoch": 0.0, "learning_rate": 4.999799499060132e-05, "loss": 0.6053, "step": 4131 }, { "epoch": 0.0, "learning_rate": 4.999799399569619e-05, "loss": 0.8091, "step": 4132 }, { "epoch": 0.0, "learning_rate": 4.9997993000544295e-05, "loss": 1.1355, "step": 4133 }, { "epoch": 0.0, "learning_rate": 4.999799200514563e-05, "loss": 1.5105, "step": 4134 }, { "epoch": 0.0, "learning_rate": 4.99979910095002e-05, "loss": 1.204, "step": 4135 }, { "epoch": 0.0, "learning_rate": 4.9997990013607995e-05, "loss": 1.1201, "step": 4136 }, { "epoch": 0.0, "learning_rate": 4.9997989017469026e-05, "loss": 1.3261, "step": 4137 }, { "epoch": 0.0, "learning_rate": 4.999798802108328e-05, "loss": 1.1437, "step": 4138 }, { "epoch": 0.0, "learning_rate": 4.999798702445077e-05, "loss": 1.2163, "step": 4139 }, { "epoch": 0.0, "learning_rate": 4.9997986027571485e-05, "loss": 1.0872, "step": 4140 }, { "epoch": 0.0, "learning_rate": 4.999798503044543e-05, "loss": 1.0482, "step": 4141 }, { "epoch": 0.0, "learning_rate": 4.9997984033072606e-05, "loss": 1.1133, "step": 4142 }, { "epoch": 0.0, "learning_rate": 4.999798303545302e-05, "loss": 2.1336, "step": 4143 }, { "epoch": 0.0, "learning_rate": 4.999798203758666e-05, "loss": 1.3565, "step": 4144 }, { "epoch": 0.0, "learning_rate": 4.999798103947353e-05, "loss": 0.8247, "step": 4145 }, { "epoch": 0.0, "learning_rate": 4.9997980041113634e-05, "loss": 1.0643, "step": 4146 }, { "epoch": 0.0, "learning_rate": 4.999797904250696e-05, "loss": 1.3419, "step": 4147 }, { "epoch": 0.0, "learning_rate": 4.999797804365353e-05, "loss": 1.2586, "step": 4148 }, { "epoch": 0.0, "learning_rate": 4.999797704455331e-05, "loss": 2.8076, "step": 4149 }, { "epoch": 0.0, "learning_rate": 4.9997976045206344e-05, "loss": 1.7105, "step": 4150 }, { "epoch": 0.0, "learning_rate": 4.999797504561259e-05, "loss": 1.0705, "step": 4151 }, { "epoch": 0.0, "learning_rate": 4.9997974045772075e-05, "loss": 1.1643, "step": 4152 }, { "epoch": 0.0, "learning_rate": 4.999797304568479e-05, "loss": 1.1731, "step": 4153 }, { "epoch": 0.0, "learning_rate": 4.9997972045350735e-05, "loss": 1.028, "step": 4154 }, { "epoch": 0.0, "learning_rate": 4.999797104476991e-05, "loss": 1.0386, "step": 4155 }, { "epoch": 0.0, "learning_rate": 4.999797004394231e-05, "loss": 1.1987, "step": 4156 }, { "epoch": 0.0, "learning_rate": 4.999796904286795e-05, "loss": 1.2542, "step": 4157 }, { "epoch": 0.0, "learning_rate": 4.9997968041546814e-05, "loss": 1.1764, "step": 4158 }, { "epoch": 0.0, "learning_rate": 4.999796703997891e-05, "loss": 1.6205, "step": 4159 }, { "epoch": 0.0, "learning_rate": 4.999796603816424e-05, "loss": 1.5423, "step": 4160 }, { "epoch": 0.0, "learning_rate": 4.99979650361028e-05, "loss": 1.539, "step": 4161 }, { "epoch": 0.0, "learning_rate": 4.999796403379459e-05, "loss": 1.297, "step": 4162 }, { "epoch": 0.0, "learning_rate": 4.999796303123961e-05, "loss": 1.1458, "step": 4163 }, { "epoch": 0.0, "learning_rate": 4.999796202843786e-05, "loss": 1.2349, "step": 4164 }, { "epoch": 0.0, "learning_rate": 4.9997961025389336e-05, "loss": 1.3741, "step": 4165 }, { "epoch": 0.0, "learning_rate": 4.999796002209405e-05, "loss": 1.2188, "step": 4166 }, { "epoch": 0.0, "learning_rate": 4.999795901855199e-05, "loss": 1.3562, "step": 4167 }, { "epoch": 0.0, "learning_rate": 4.999795801476317e-05, "loss": 0.3673, "step": 4168 }, { "epoch": 0.0, "learning_rate": 4.999795701072757e-05, "loss": 0.2597, "step": 4169 }, { "epoch": 0.0, "learning_rate": 4.99979560064452e-05, "loss": 0.6811, "step": 4170 }, { "epoch": 0.0, "learning_rate": 4.999795500191607e-05, "loss": 1.1625, "step": 4171 }, { "epoch": 0.0, "learning_rate": 4.9997953997140165e-05, "loss": 1.1949, "step": 4172 }, { "epoch": 0.0, "learning_rate": 4.9997952992117494e-05, "loss": 1.0916, "step": 4173 }, { "epoch": 0.0, "learning_rate": 4.9997951986848045e-05, "loss": 0.7343, "step": 4174 }, { "epoch": 0.0, "learning_rate": 4.999795098133184e-05, "loss": 0.7105, "step": 4175 }, { "epoch": 0.0, "learning_rate": 4.999794997556886e-05, "loss": 0.6363, "step": 4176 }, { "epoch": 0.0, "learning_rate": 4.9997948969559105e-05, "loss": 0.6748, "step": 4177 }, { "epoch": 0.0, "learning_rate": 4.9997947963302585e-05, "loss": 0.8273, "step": 4178 }, { "epoch": 0.0, "learning_rate": 4.999794695679929e-05, "loss": 1.3885, "step": 4179 }, { "epoch": 0.0, "learning_rate": 4.999794595004924e-05, "loss": 1.1066, "step": 4180 }, { "epoch": 0.0, "learning_rate": 4.999794494305241e-05, "loss": 1.0197, "step": 4181 }, { "epoch": 0.0, "learning_rate": 4.999794393580882e-05, "loss": 1.0707, "step": 4182 }, { "epoch": 0.0, "learning_rate": 4.999794292831845e-05, "loss": 1.0189, "step": 4183 }, { "epoch": 0.0, "learning_rate": 4.9997941920581316e-05, "loss": 1.2503, "step": 4184 }, { "epoch": 0.0, "learning_rate": 4.999794091259741e-05, "loss": 1.1458, "step": 4185 }, { "epoch": 0.0, "learning_rate": 4.999793990436674e-05, "loss": 1.1863, "step": 4186 }, { "epoch": 0.0, "learning_rate": 4.9997938895889294e-05, "loss": 1.2241, "step": 4187 }, { "epoch": 0.0, "learning_rate": 4.999793788716508e-05, "loss": 0.8661, "step": 4188 }, { "epoch": 0.0, "learning_rate": 4.9997936878194107e-05, "loss": 0.5257, "step": 4189 }, { "epoch": 0.0, "learning_rate": 4.999793586897635e-05, "loss": 0.9842, "step": 4190 }, { "epoch": 0.0, "learning_rate": 4.999793485951184e-05, "loss": 1.1282, "step": 4191 }, { "epoch": 0.0, "learning_rate": 4.999793384980055e-05, "loss": 1.2302, "step": 4192 }, { "epoch": 0.0, "learning_rate": 4.999793283984249e-05, "loss": 1.2022, "step": 4193 }, { "epoch": 0.0, "learning_rate": 4.9997931829637665e-05, "loss": 1.1716, "step": 4194 }, { "epoch": 0.0, "learning_rate": 4.9997930819186075e-05, "loss": 1.0513, "step": 4195 }, { "epoch": 0.0, "learning_rate": 4.999792980848771e-05, "loss": 1.0939, "step": 4196 }, { "epoch": 0.0, "learning_rate": 4.9997928797542575e-05, "loss": 1.15, "step": 4197 }, { "epoch": 0.0, "learning_rate": 4.999792778635067e-05, "loss": 1.2474, "step": 4198 }, { "epoch": 0.0, "learning_rate": 4.9997926774912004e-05, "loss": 1.2844, "step": 4199 }, { "epoch": 0.0, "learning_rate": 4.999792576322656e-05, "loss": 1.1551, "step": 4200 }, { "epoch": 0.0, "learning_rate": 4.9997924751294355e-05, "loss": 1.2846, "step": 4201 }, { "epoch": 0.0, "learning_rate": 4.9997923739115374e-05, "loss": 0.9903, "step": 4202 }, { "epoch": 0.0, "learning_rate": 4.999792272668963e-05, "loss": 1.1966, "step": 4203 }, { "epoch": 0.0, "learning_rate": 4.999792171401712e-05, "loss": 1.275, "step": 4204 }, { "epoch": 0.0, "learning_rate": 4.9997920701097836e-05, "loss": 1.0189, "step": 4205 }, { "epoch": 0.0, "learning_rate": 4.999791968793178e-05, "loss": 1.1909, "step": 4206 }, { "epoch": 0.0, "learning_rate": 4.9997918674518953e-05, "loss": 1.2318, "step": 4207 }, { "epoch": 0.0, "learning_rate": 4.999791766085936e-05, "loss": 1.2186, "step": 4208 }, { "epoch": 0.0, "learning_rate": 4.9997916646953e-05, "loss": 0.8141, "step": 4209 }, { "epoch": 0.0, "learning_rate": 4.9997915632799876e-05, "loss": 1.029, "step": 4210 }, { "epoch": 0.0, "learning_rate": 4.9997914618399974e-05, "loss": 1.1777, "step": 4211 }, { "epoch": 0.0, "learning_rate": 4.999791360375331e-05, "loss": 1.1809, "step": 4212 }, { "epoch": 0.0, "learning_rate": 4.999791258885988e-05, "loss": 1.2198, "step": 4213 }, { "epoch": 0.0, "learning_rate": 4.999791157371967e-05, "loss": 1.2193, "step": 4214 }, { "epoch": 0.0, "learning_rate": 4.9997910558332696e-05, "loss": 0.9905, "step": 4215 }, { "epoch": 0.0, "learning_rate": 4.999790954269895e-05, "loss": 1.2994, "step": 4216 }, { "epoch": 0.0, "learning_rate": 4.9997908526818436e-05, "loss": 1.0537, "step": 4217 }, { "epoch": 0.0, "learning_rate": 4.999790751069116e-05, "loss": 1.5454, "step": 4218 }, { "epoch": 0.0, "learning_rate": 4.999790649431711e-05, "loss": 1.1544, "step": 4219 }, { "epoch": 0.0, "learning_rate": 4.99979054776963e-05, "loss": 0.9599, "step": 4220 }, { "epoch": 0.0, "learning_rate": 4.9997904460828704e-05, "loss": 1.0451, "step": 4221 }, { "epoch": 0.0, "learning_rate": 4.999790344371435e-05, "loss": 1.054, "step": 4222 }, { "epoch": 0.0, "learning_rate": 4.9997902426353224e-05, "loss": 1.161, "step": 4223 }, { "epoch": 0.0, "learning_rate": 4.999790140874534e-05, "loss": 0.8822, "step": 4224 }, { "epoch": 0.0, "learning_rate": 4.999790039089067e-05, "loss": 1.0428, "step": 4225 }, { "epoch": 0.0, "learning_rate": 4.9997899372789244e-05, "loss": 1.132, "step": 4226 }, { "epoch": 0.0, "learning_rate": 4.9997898354441044e-05, "loss": 1.0825, "step": 4227 }, { "epoch": 0.0, "learning_rate": 4.999789733584608e-05, "loss": 1.1674, "step": 4228 }, { "epoch": 0.0, "learning_rate": 4.999789631700434e-05, "loss": 1.1649, "step": 4229 }, { "epoch": 0.0, "learning_rate": 4.999789529791583e-05, "loss": 0.691, "step": 4230 }, { "epoch": 0.0, "learning_rate": 4.999789427858056e-05, "loss": 1.1995, "step": 4231 }, { "epoch": 0.0, "learning_rate": 4.9997893258998517e-05, "loss": 0.9215, "step": 4232 }, { "epoch": 0.0, "learning_rate": 4.999789223916971e-05, "loss": 1.2847, "step": 4233 }, { "epoch": 0.0, "learning_rate": 4.999789121909413e-05, "loss": 1.1242, "step": 4234 }, { "epoch": 0.0, "learning_rate": 4.999789019877178e-05, "loss": 0.6223, "step": 4235 }, { "epoch": 0.0, "learning_rate": 4.999788917820266e-05, "loss": 0.8613, "step": 4236 }, { "epoch": 0.0, "learning_rate": 4.999788815738678e-05, "loss": 0.9802, "step": 4237 }, { "epoch": 0.0, "learning_rate": 4.999788713632412e-05, "loss": 1.3161, "step": 4238 }, { "epoch": 0.0, "learning_rate": 4.99978861150147e-05, "loss": 1.0592, "step": 4239 }, { "epoch": 0.0, "learning_rate": 4.999788509345851e-05, "loss": 1.2151, "step": 4240 }, { "epoch": 0.0, "learning_rate": 4.9997884071655546e-05, "loss": 1.0687, "step": 4241 }, { "epoch": 0.0, "learning_rate": 4.999788304960582e-05, "loss": 1.2194, "step": 4242 }, { "epoch": 0.0, "learning_rate": 4.999788202730932e-05, "loss": 1.1696, "step": 4243 }, { "epoch": 0.0, "learning_rate": 4.999788100476606e-05, "loss": 0.9879, "step": 4244 }, { "epoch": 0.0, "learning_rate": 4.9997879981976016e-05, "loss": 0.6206, "step": 4245 }, { "epoch": 0.0, "learning_rate": 4.999787895893922e-05, "loss": 1.3536, "step": 4246 }, { "epoch": 0.0, "learning_rate": 4.999787793565565e-05, "loss": 1.2582, "step": 4247 }, { "epoch": 0.0, "learning_rate": 4.999787691212531e-05, "loss": 1.3572, "step": 4248 }, { "epoch": 0.0, "learning_rate": 4.9997875888348195e-05, "loss": 1.0565, "step": 4249 }, { "epoch": 0.0, "learning_rate": 4.999787486432432e-05, "loss": 1.3061, "step": 4250 }, { "epoch": 0.0, "learning_rate": 4.999787384005367e-05, "loss": 1.1021, "step": 4251 }, { "epoch": 0.0, "learning_rate": 4.999787281553626e-05, "loss": 0.7918, "step": 4252 }, { "epoch": 0.0, "learning_rate": 4.9997871790772075e-05, "loss": 1.0945, "step": 4253 }, { "epoch": 0.0, "learning_rate": 4.999787076576112e-05, "loss": 1.3962, "step": 4254 }, { "epoch": 0.0, "learning_rate": 4.999786974050341e-05, "loss": 1.0041, "step": 4255 }, { "epoch": 0.0, "learning_rate": 4.999786871499892e-05, "loss": 0.9295, "step": 4256 }, { "epoch": 0.0, "learning_rate": 4.9997867689247664e-05, "loss": 1.2329, "step": 4257 }, { "epoch": 0.0, "learning_rate": 4.999786666324964e-05, "loss": 1.1603, "step": 4258 }, { "epoch": 0.0, "learning_rate": 4.999786563700485e-05, "loss": 1.1955, "step": 4259 }, { "epoch": 0.0, "learning_rate": 4.999786461051329e-05, "loss": 1.642, "step": 4260 }, { "epoch": 0.0, "learning_rate": 4.999786358377495e-05, "loss": 1.9116, "step": 4261 }, { "epoch": 0.0, "learning_rate": 4.999786255678985e-05, "loss": 1.9218, "step": 4262 }, { "epoch": 0.0, "learning_rate": 4.999786152955799e-05, "loss": 1.4183, "step": 4263 }, { "epoch": 0.0, "learning_rate": 4.999786050207935e-05, "loss": 1.0229, "step": 4264 }, { "epoch": 0.0, "learning_rate": 4.999785947435395e-05, "loss": 0.7571, "step": 4265 }, { "epoch": 0.0, "learning_rate": 4.999785844638178e-05, "loss": 0.6904, "step": 4266 }, { "epoch": 0.0, "learning_rate": 4.999785741816284e-05, "loss": 1.4207, "step": 4267 }, { "epoch": 0.0, "learning_rate": 4.999785638969713e-05, "loss": 0.8324, "step": 4268 }, { "epoch": 0.0, "learning_rate": 4.999785536098466e-05, "loss": 1.1555, "step": 4269 }, { "epoch": 0.0, "learning_rate": 4.999785433202541e-05, "loss": 1.3096, "step": 4270 }, { "epoch": 0.0, "learning_rate": 4.99978533028194e-05, "loss": 0.7022, "step": 4271 }, { "epoch": 0.0, "learning_rate": 4.9997852273366616e-05, "loss": 0.5015, "step": 4272 }, { "epoch": 0.0, "learning_rate": 4.999785124366706e-05, "loss": 0.6592, "step": 4273 }, { "epoch": 0.0, "learning_rate": 4.9997850213720745e-05, "loss": 0.8913, "step": 4274 }, { "epoch": 0.0, "learning_rate": 4.9997849183527664e-05, "loss": 1.3936, "step": 4275 }, { "epoch": 0.0, "learning_rate": 4.9997848153087804e-05, "loss": 1.2355, "step": 4276 }, { "epoch": 0.0, "learning_rate": 4.999784712240119e-05, "loss": 0.8601, "step": 4277 }, { "epoch": 0.0, "learning_rate": 4.999784609146779e-05, "loss": 1.0973, "step": 4278 }, { "epoch": 0.0, "learning_rate": 4.999784506028763e-05, "loss": 0.974, "step": 4279 }, { "epoch": 0.0, "learning_rate": 4.999784402886071e-05, "loss": 1.2342, "step": 4280 }, { "epoch": 0.0, "learning_rate": 4.999784299718701e-05, "loss": 1.1617, "step": 4281 }, { "epoch": 0.0, "learning_rate": 4.9997841965266546e-05, "loss": 1.1919, "step": 4282 }, { "epoch": 0.0, "learning_rate": 4.9997840933099315e-05, "loss": 1.281, "step": 4283 }, { "epoch": 0.0, "learning_rate": 4.999783990068532e-05, "loss": 1.119, "step": 4284 }, { "epoch": 0.0, "learning_rate": 4.9997838868024546e-05, "loss": 1.3177, "step": 4285 }, { "epoch": 0.0, "learning_rate": 4.999783783511701e-05, "loss": 0.8571, "step": 4286 }, { "epoch": 0.0, "learning_rate": 4.9997836801962706e-05, "loss": 0.7544, "step": 4287 }, { "epoch": 0.0, "learning_rate": 4.999783576856163e-05, "loss": 0.5614, "step": 4288 }, { "epoch": 0.0, "learning_rate": 4.9997834734913795e-05, "loss": 1.0438, "step": 4289 }, { "epoch": 0.0, "learning_rate": 4.9997833701019186e-05, "loss": 1.2406, "step": 4290 }, { "epoch": 0.0, "learning_rate": 4.9997832666877806e-05, "loss": 1.1987, "step": 4291 }, { "epoch": 0.0, "learning_rate": 4.999783163248967e-05, "loss": 1.4006, "step": 4292 }, { "epoch": 0.0, "learning_rate": 4.999783059785475e-05, "loss": 0.9797, "step": 4293 }, { "epoch": 0.0, "learning_rate": 4.999782956297307e-05, "loss": 1.0625, "step": 4294 }, { "epoch": 0.0, "learning_rate": 4.999782852784463e-05, "loss": 1.2627, "step": 4295 }, { "epoch": 0.0, "learning_rate": 4.9997827492469406e-05, "loss": 1.0369, "step": 4296 }, { "epoch": 0.0, "learning_rate": 4.999782645684742e-05, "loss": 1.4083, "step": 4297 }, { "epoch": 0.0, "learning_rate": 4.999782542097867e-05, "loss": 1.428, "step": 4298 }, { "epoch": 0.0, "learning_rate": 4.9997824384863145e-05, "loss": 1.4295, "step": 4299 }, { "epoch": 0.0, "learning_rate": 4.999782334850086e-05, "loss": 1.0226, "step": 4300 }, { "epoch": 0.0, "learning_rate": 4.99978223118918e-05, "loss": 1.0865, "step": 4301 }, { "epoch": 0.0, "learning_rate": 4.999782127503597e-05, "loss": 1.0492, "step": 4302 }, { "epoch": 0.0, "learning_rate": 4.999782023793338e-05, "loss": 1.5096, "step": 4303 }, { "epoch": 0.0, "learning_rate": 4.999781920058402e-05, "loss": 1.2298, "step": 4304 }, { "epoch": 0.0, "learning_rate": 4.999781816298789e-05, "loss": 1.4012, "step": 4305 }, { "epoch": 0.0, "learning_rate": 4.9997817125144995e-05, "loss": 1.0163, "step": 4306 }, { "epoch": 0.0, "learning_rate": 4.999781608705533e-05, "loss": 0.8868, "step": 4307 }, { "epoch": 0.0, "learning_rate": 4.99978150487189e-05, "loss": 1.0451, "step": 4308 }, { "epoch": 0.0, "learning_rate": 4.99978140101357e-05, "loss": 1.1224, "step": 4309 }, { "epoch": 0.0, "learning_rate": 4.999781297130573e-05, "loss": 0.59, "step": 4310 }, { "epoch": 0.0, "learning_rate": 4.999781193222899e-05, "loss": 0.5647, "step": 4311 }, { "epoch": 0.0, "learning_rate": 4.999781089290549e-05, "loss": 2.8877, "step": 4312 }, { "epoch": 0.0, "learning_rate": 4.999780985333522e-05, "loss": 2.3249, "step": 4313 }, { "epoch": 0.0, "learning_rate": 4.999780881351818e-05, "loss": 0.5768, "step": 4314 }, { "epoch": 0.0, "learning_rate": 4.999780777345438e-05, "loss": 0.9765, "step": 4315 }, { "epoch": 0.0, "learning_rate": 4.999780673314379e-05, "loss": 1.2209, "step": 4316 }, { "epoch": 0.0, "learning_rate": 4.9997805692586455e-05, "loss": 1.1851, "step": 4317 }, { "epoch": 0.0, "learning_rate": 4.999780465178234e-05, "loss": 1.0895, "step": 4318 }, { "epoch": 0.0, "learning_rate": 4.999780361073146e-05, "loss": 1.0949, "step": 4319 }, { "epoch": 0.0, "learning_rate": 4.999780256943381e-05, "loss": 1.3861, "step": 4320 }, { "epoch": 0.0, "learning_rate": 4.99978015278894e-05, "loss": 1.0276, "step": 4321 }, { "epoch": 0.0, "learning_rate": 4.999780048609822e-05, "loss": 1.2675, "step": 4322 }, { "epoch": 0.0, "learning_rate": 4.999779944406027e-05, "loss": 1.2956, "step": 4323 }, { "epoch": 0.0, "learning_rate": 4.9997798401775556e-05, "loss": 1.1, "step": 4324 }, { "epoch": 0.0, "learning_rate": 4.999779735924407e-05, "loss": 1.1209, "step": 4325 }, { "epoch": 0.0, "learning_rate": 4.9997796316465814e-05, "loss": 0.9514, "step": 4326 }, { "epoch": 0.0, "learning_rate": 4.999779527344079e-05, "loss": 1.148, "step": 4327 }, { "epoch": 0.0, "learning_rate": 4.9997794230169e-05, "loss": 0.991, "step": 4328 }, { "epoch": 0.0, "learning_rate": 4.999779318665045e-05, "loss": 1.1714, "step": 4329 }, { "epoch": 0.0, "learning_rate": 4.999779214288512e-05, "loss": 1.024, "step": 4330 }, { "epoch": 0.0, "learning_rate": 4.999779109887304e-05, "loss": 1.0824, "step": 4331 }, { "epoch": 0.0, "learning_rate": 4.9997790054614174e-05, "loss": 1.2253, "step": 4332 }, { "epoch": 0.0, "learning_rate": 4.9997789010108546e-05, "loss": 0.9916, "step": 4333 }, { "epoch": 0.0, "learning_rate": 4.9997787965356154e-05, "loss": 1.2658, "step": 4334 }, { "epoch": 0.0, "learning_rate": 4.999778692035699e-05, "loss": 1.0458, "step": 4335 }, { "epoch": 0.0, "learning_rate": 4.9997785875111056e-05, "loss": 1.299, "step": 4336 }, { "epoch": 0.0, "learning_rate": 4.9997784829618364e-05, "loss": 0.933, "step": 4337 }, { "epoch": 0.0, "learning_rate": 4.99977837838789e-05, "loss": 1.2263, "step": 4338 }, { "epoch": 0.0, "learning_rate": 4.999778273789266e-05, "loss": 1.5378, "step": 4339 }, { "epoch": 0.0, "learning_rate": 4.999778169165966e-05, "loss": 1.2621, "step": 4340 }, { "epoch": 0.0, "learning_rate": 4.9997780645179896e-05, "loss": 1.4907, "step": 4341 }, { "epoch": 0.0, "learning_rate": 4.999777959845336e-05, "loss": 0.6645, "step": 4342 }, { "epoch": 0.0, "learning_rate": 4.9997778551480056e-05, "loss": 0.8306, "step": 4343 }, { "epoch": 0.0, "learning_rate": 4.999777750425999e-05, "loss": 0.6357, "step": 4344 }, { "epoch": 0.0, "learning_rate": 4.999777645679316e-05, "loss": 0.7529, "step": 4345 }, { "epoch": 0.0, "learning_rate": 4.999777540907955e-05, "loss": 0.9319, "step": 4346 }, { "epoch": 0.0, "learning_rate": 4.9997774361119175e-05, "loss": 0.7715, "step": 4347 }, { "epoch": 0.0, "learning_rate": 4.999777331291203e-05, "loss": 0.8799, "step": 4348 }, { "epoch": 0.0, "learning_rate": 4.999777226445813e-05, "loss": 0.8824, "step": 4349 }, { "epoch": 0.0, "learning_rate": 4.999777121575745e-05, "loss": 1.3207, "step": 4350 }, { "epoch": 0.0, "learning_rate": 4.999777016681001e-05, "loss": 1.0959, "step": 4351 }, { "epoch": 0.0, "learning_rate": 4.99977691176158e-05, "loss": 0.6392, "step": 4352 }, { "epoch": 0.0, "learning_rate": 4.999776806817482e-05, "loss": 1.5281, "step": 4353 }, { "epoch": 0.0, "learning_rate": 4.999776701848707e-05, "loss": 1.1428, "step": 4354 }, { "epoch": 0.0, "learning_rate": 4.9997765968552565e-05, "loss": 1.2032, "step": 4355 }, { "epoch": 0.0, "learning_rate": 4.999776491837128e-05, "loss": 1.1605, "step": 4356 }, { "epoch": 0.0, "learning_rate": 4.999776386794324e-05, "loss": 1.1478, "step": 4357 }, { "epoch": 0.0, "learning_rate": 4.999776281726842e-05, "loss": 1.1389, "step": 4358 }, { "epoch": 0.0, "learning_rate": 4.999776176634684e-05, "loss": 1.0784, "step": 4359 }, { "epoch": 0.0, "learning_rate": 4.999776071517849e-05, "loss": 1.0957, "step": 4360 }, { "epoch": 0.0, "learning_rate": 4.9997759663763375e-05, "loss": 1.1626, "step": 4361 }, { "epoch": 0.0, "learning_rate": 4.9997758612101484e-05, "loss": 1.1153, "step": 4362 }, { "epoch": 0.0, "learning_rate": 4.9997757560192836e-05, "loss": 1.3685, "step": 4363 }, { "epoch": 0.0, "learning_rate": 4.9997756508037416e-05, "loss": 1.077, "step": 4364 }, { "epoch": 0.0, "learning_rate": 4.999775545563523e-05, "loss": 1.1683, "step": 4365 }, { "epoch": 0.0, "learning_rate": 4.999775440298628e-05, "loss": 1.2724, "step": 4366 }, { "epoch": 0.0, "learning_rate": 4.999775335009056e-05, "loss": 1.0369, "step": 4367 }, { "epoch": 0.0, "learning_rate": 4.999775229694807e-05, "loss": 0.9634, "step": 4368 }, { "epoch": 0.0, "learning_rate": 4.999775124355881e-05, "loss": 1.1885, "step": 4369 }, { "epoch": 0.0, "learning_rate": 4.9997750189922785e-05, "loss": 1.1612, "step": 4370 }, { "epoch": 0.0, "learning_rate": 4.999774913604e-05, "loss": 1.0395, "step": 4371 }, { "epoch": 0.0, "learning_rate": 4.9997748081910446e-05, "loss": 0.9069, "step": 4372 }, { "epoch": 0.0, "learning_rate": 4.999774702753412e-05, "loss": 1.0056, "step": 4373 }, { "epoch": 0.0, "learning_rate": 4.999774597291102e-05, "loss": 1.2466, "step": 4374 }, { "epoch": 0.0, "learning_rate": 4.999774491804117e-05, "loss": 0.7215, "step": 4375 }, { "epoch": 0.0, "learning_rate": 4.999774386292454e-05, "loss": 1.1564, "step": 4376 }, { "epoch": 0.0, "learning_rate": 4.999774280756116e-05, "loss": 1.313, "step": 4377 }, { "epoch": 0.0, "learning_rate": 4.999774175195099e-05, "loss": 1.1211, "step": 4378 }, { "epoch": 0.0, "learning_rate": 4.999774069609406e-05, "loss": 0.9678, "step": 4379 }, { "epoch": 0.0, "learning_rate": 4.999773963999037e-05, "loss": 1.1849, "step": 4380 }, { "epoch": 0.0, "learning_rate": 4.999773858363991e-05, "loss": 0.7452, "step": 4381 }, { "epoch": 0.0, "learning_rate": 4.9997737527042675e-05, "loss": 1.5299, "step": 4382 }, { "epoch": 0.0, "learning_rate": 4.999773647019868e-05, "loss": 0.9208, "step": 4383 }, { "epoch": 0.0, "learning_rate": 4.9997735413107916e-05, "loss": 0.8891, "step": 4384 }, { "epoch": 0.0, "learning_rate": 4.999773435577039e-05, "loss": 1.143, "step": 4385 }, { "epoch": 0.0, "learning_rate": 4.999773329818609e-05, "loss": 1.2671, "step": 4386 }, { "epoch": 0.0, "learning_rate": 4.999773224035502e-05, "loss": 1.1009, "step": 4387 }, { "epoch": 0.0, "learning_rate": 4.9997731182277196e-05, "loss": 1.114, "step": 4388 }, { "epoch": 0.0, "learning_rate": 4.999773012395259e-05, "loss": 1.0124, "step": 4389 }, { "epoch": 0.0, "learning_rate": 4.999772906538123e-05, "loss": 1.0004, "step": 4390 }, { "epoch": 0.0, "learning_rate": 4.9997728006563097e-05, "loss": 1.1445, "step": 4391 }, { "epoch": 0.0, "learning_rate": 4.99977269474982e-05, "loss": 1.0378, "step": 4392 }, { "epoch": 0.0, "learning_rate": 4.999772588818653e-05, "loss": 1.1042, "step": 4393 }, { "epoch": 0.0, "learning_rate": 4.99977248286281e-05, "loss": 1.2522, "step": 4394 }, { "epoch": 0.0, "learning_rate": 4.99977237688229e-05, "loss": 1.1237, "step": 4395 }, { "epoch": 0.0, "learning_rate": 4.999772270877093e-05, "loss": 1.1444, "step": 4396 }, { "epoch": 0.0, "learning_rate": 4.99977216484722e-05, "loss": 1.9499, "step": 4397 }, { "epoch": 0.0, "learning_rate": 4.9997720587926686e-05, "loss": 3.5417, "step": 4398 }, { "epoch": 0.0, "learning_rate": 4.9997719527134423e-05, "loss": 1.2548, "step": 4399 }, { "epoch": 0.0, "learning_rate": 4.999771846609539e-05, "loss": 1.3411, "step": 4400 }, { "epoch": 0.0, "learning_rate": 4.9997717404809586e-05, "loss": 1.3086, "step": 4401 }, { "epoch": 0.0, "learning_rate": 4.999771634327701e-05, "loss": 1.2577, "step": 4402 }, { "epoch": 0.0, "learning_rate": 4.9997715281497676e-05, "loss": 1.2029, "step": 4403 }, { "epoch": 0.0, "learning_rate": 4.999771421947157e-05, "loss": 1.2334, "step": 4404 }, { "epoch": 0.0, "learning_rate": 4.99977131571987e-05, "loss": 1.2065, "step": 4405 }, { "epoch": 0.0, "learning_rate": 4.999771209467907e-05, "loss": 1.1416, "step": 4406 }, { "epoch": 0.0, "learning_rate": 4.9997711031912665e-05, "loss": 1.1523, "step": 4407 }, { "epoch": 0.0, "learning_rate": 4.999770996889949e-05, "loss": 1.1319, "step": 4408 }, { "epoch": 0.0, "learning_rate": 4.999770890563955e-05, "loss": 1.4375, "step": 4409 }, { "epoch": 0.0, "learning_rate": 4.9997707842132845e-05, "loss": 1.2708, "step": 4410 }, { "epoch": 0.0, "learning_rate": 4.9997706778379376e-05, "loss": 1.4632, "step": 4411 }, { "epoch": 0.0, "learning_rate": 4.999770571437914e-05, "loss": 1.3789, "step": 4412 }, { "epoch": 0.0, "learning_rate": 4.999770465013213e-05, "loss": 1.0032, "step": 4413 }, { "epoch": 0.0, "learning_rate": 4.999770358563836e-05, "loss": 0.7754, "step": 4414 }, { "epoch": 0.0, "learning_rate": 4.999770252089782e-05, "loss": 1.0801, "step": 4415 }, { "epoch": 0.0, "learning_rate": 4.999770145591052e-05, "loss": 1.0647, "step": 4416 }, { "epoch": 0.0, "learning_rate": 4.999770039067644e-05, "loss": 1.2998, "step": 4417 }, { "epoch": 0.0, "learning_rate": 4.99976993251956e-05, "loss": 1.0556, "step": 4418 }, { "epoch": 0.0, "learning_rate": 4.9997698259467996e-05, "loss": 1.1464, "step": 4419 }, { "epoch": 0.0, "learning_rate": 4.999769719349363e-05, "loss": 1.1842, "step": 4420 }, { "epoch": 0.0, "learning_rate": 4.999769612727249e-05, "loss": 1.059, "step": 4421 }, { "epoch": 0.0, "learning_rate": 4.999769506080458e-05, "loss": 1.0721, "step": 4422 }, { "epoch": 0.0, "learning_rate": 4.999769399408991e-05, "loss": 1.0131, "step": 4423 }, { "epoch": 0.0, "learning_rate": 4.999769292712847e-05, "loss": 1.7383, "step": 4424 }, { "epoch": 0.0, "learning_rate": 4.999769185992026e-05, "loss": 1.2293, "step": 4425 }, { "epoch": 0.0, "learning_rate": 4.999769079246529e-05, "loss": 0.9657, "step": 4426 }, { "epoch": 0.0, "learning_rate": 4.999768972476355e-05, "loss": 1.0584, "step": 4427 }, { "epoch": 0.0, "learning_rate": 4.999768865681505e-05, "loss": 1.2185, "step": 4428 }, { "epoch": 0.0, "learning_rate": 4.9997687588619765e-05, "loss": 1.156, "step": 4429 }, { "epoch": 0.0, "learning_rate": 4.999768652017773e-05, "loss": 1.4052, "step": 4430 }, { "epoch": 0.0, "learning_rate": 4.999768545148893e-05, "loss": 1.2739, "step": 4431 }, { "epoch": 0.0, "learning_rate": 4.999768438255336e-05, "loss": 1.1099, "step": 4432 }, { "epoch": 0.0, "learning_rate": 4.999768331337101e-05, "loss": 1.3071, "step": 4433 }, { "epoch": 0.0, "learning_rate": 4.9997682243941906e-05, "loss": 0.9026, "step": 4434 }, { "epoch": 0.0, "learning_rate": 4.999768117426603e-05, "loss": 1.1763, "step": 4435 }, { "epoch": 0.0, "learning_rate": 4.9997680104343396e-05, "loss": 0.6924, "step": 4436 }, { "epoch": 0.0, "learning_rate": 4.999767903417399e-05, "loss": 1.1063, "step": 4437 }, { "epoch": 0.0, "learning_rate": 4.9997677963757815e-05, "loss": 6.2614, "step": 4438 }, { "epoch": 0.0, "learning_rate": 4.999767689309488e-05, "loss": 6.0042, "step": 4439 }, { "epoch": 0.0, "learning_rate": 4.999767582218518e-05, "loss": 5.9976, "step": 4440 }, { "epoch": 0.0, "learning_rate": 4.99976747510287e-05, "loss": 5.9647, "step": 4441 }, { "epoch": 0.0, "learning_rate": 4.999767367962547e-05, "loss": 5.9497, "step": 4442 }, { "epoch": 0.0, "learning_rate": 4.999767260797546e-05, "loss": 5.9563, "step": 4443 }, { "epoch": 0.0, "learning_rate": 4.999767153607869e-05, "loss": 5.8703, "step": 4444 }, { "epoch": 0.0, "learning_rate": 4.999767046393515e-05, "loss": 5.8784, "step": 4445 }, { "epoch": 0.0, "learning_rate": 4.999766939154484e-05, "loss": 5.8673, "step": 4446 }, { "epoch": 0.0, "learning_rate": 4.999766831890777e-05, "loss": 5.9005, "step": 4447 }, { "epoch": 0.0, "learning_rate": 4.999766724602394e-05, "loss": 4.7112, "step": 4448 }, { "epoch": 0.0, "learning_rate": 4.9997666172893334e-05, "loss": 1.3249, "step": 4449 }, { "epoch": 0.0, "learning_rate": 4.9997665099515965e-05, "loss": 1.5641, "step": 4450 }, { "epoch": 0.0, "learning_rate": 4.9997664025891824e-05, "loss": 1.8465, "step": 4451 }, { "epoch": 0.0, "learning_rate": 4.9997662952020926e-05, "loss": 1.9022, "step": 4452 }, { "epoch": 0.0, "learning_rate": 4.999766187790326e-05, "loss": 1.2422, "step": 4453 }, { "epoch": 0.0, "learning_rate": 4.999766080353882e-05, "loss": 1.1355, "step": 4454 }, { "epoch": 0.0, "learning_rate": 4.999765972892762e-05, "loss": 1.0728, "step": 4455 }, { "epoch": 0.0, "learning_rate": 4.999765865406965e-05, "loss": 0.7308, "step": 4456 }, { "epoch": 0.0, "learning_rate": 4.9997657578964915e-05, "loss": 1.221, "step": 4457 }, { "epoch": 0.0, "learning_rate": 4.999765650361341e-05, "loss": 1.0683, "step": 4458 }, { "epoch": 0.0, "learning_rate": 4.999765542801514e-05, "loss": 1.0572, "step": 4459 }, { "epoch": 0.0, "learning_rate": 4.999765435217011e-05, "loss": 1.3279, "step": 4460 }, { "epoch": 0.0, "learning_rate": 4.9997653276078316e-05, "loss": 1.1065, "step": 4461 }, { "epoch": 0.0, "learning_rate": 4.999765219973975e-05, "loss": 1.2609, "step": 4462 }, { "epoch": 0.0, "learning_rate": 4.999765112315441e-05, "loss": 1.0018, "step": 4463 }, { "epoch": 0.0, "learning_rate": 4.9997650046322315e-05, "loss": 1.2095, "step": 4464 }, { "epoch": 0.0, "learning_rate": 4.999764896924345e-05, "loss": 1.0187, "step": 4465 }, { "epoch": 0.0, "learning_rate": 4.999764789191782e-05, "loss": 1.28, "step": 4466 }, { "epoch": 0.0, "learning_rate": 4.999764681434542e-05, "loss": 1.14, "step": 4467 }, { "epoch": 0.0, "learning_rate": 4.999764573652626e-05, "loss": 0.8774, "step": 4468 }, { "epoch": 0.0, "learning_rate": 4.999764465846033e-05, "loss": 1.1061, "step": 4469 }, { "epoch": 0.0, "learning_rate": 4.999764358014763e-05, "loss": 1.3555, "step": 4470 }, { "epoch": 0.0, "learning_rate": 4.999764250158817e-05, "loss": 1.042, "step": 4471 }, { "epoch": 0.0, "learning_rate": 4.9997641422781936e-05, "loss": 1.2689, "step": 4472 }, { "epoch": 0.0, "learning_rate": 4.9997640343728945e-05, "loss": 1.1008, "step": 4473 }, { "epoch": 0.0, "learning_rate": 4.999763926442919e-05, "loss": 1.4017, "step": 4474 }, { "epoch": 0.0, "learning_rate": 4.9997638184882655e-05, "loss": 0.9575, "step": 4475 }, { "epoch": 0.0, "learning_rate": 4.9997637105089364e-05, "loss": 1.3155, "step": 4476 }, { "epoch": 0.0, "learning_rate": 4.99976360250493e-05, "loss": 1.0595, "step": 4477 }, { "epoch": 0.0, "learning_rate": 4.999763494476248e-05, "loss": 1.228, "step": 4478 }, { "epoch": 0.0, "learning_rate": 4.999763386422889e-05, "loss": 0.9576, "step": 4479 }, { "epoch": 0.0, "learning_rate": 4.9997632783448527e-05, "loss": 1.1748, "step": 4480 }, { "epoch": 0.0, "learning_rate": 4.9997631702421406e-05, "loss": 1.2856, "step": 4481 }, { "epoch": 0.0, "learning_rate": 4.9997630621147515e-05, "loss": 1.1432, "step": 4482 }, { "epoch": 0.0, "learning_rate": 4.999762953962686e-05, "loss": 1.0681, "step": 4483 }, { "epoch": 0.0, "learning_rate": 4.999762845785944e-05, "loss": 1.021, "step": 4484 }, { "epoch": 0.0, "learning_rate": 4.999762737584525e-05, "loss": 1.1034, "step": 4485 }, { "epoch": 0.0, "learning_rate": 4.999762629358429e-05, "loss": 0.9276, "step": 4486 }, { "epoch": 0.0, "learning_rate": 4.999762521107657e-05, "loss": 0.9006, "step": 4487 }, { "epoch": 0.0, "learning_rate": 4.9997624128322086e-05, "loss": 0.8601, "step": 4488 }, { "epoch": 0.0, "learning_rate": 4.999762304532083e-05, "loss": 1.2345, "step": 4489 }, { "epoch": 0.0, "learning_rate": 4.9997621962072816e-05, "loss": 1.0603, "step": 4490 }, { "epoch": 0.0, "learning_rate": 4.999762087857803e-05, "loss": 1.2632, "step": 4491 }, { "epoch": 0.0, "learning_rate": 4.999761979483648e-05, "loss": 1.0961, "step": 4492 }, { "epoch": 0.0, "learning_rate": 4.999761871084816e-05, "loss": 1.4085, "step": 4493 }, { "epoch": 0.0, "learning_rate": 4.999761762661308e-05, "loss": 0.9822, "step": 4494 }, { "epoch": 0.0, "learning_rate": 4.9997616542131235e-05, "loss": 0.8745, "step": 4495 }, { "epoch": 0.0, "learning_rate": 4.999761545740262e-05, "loss": 1.2333, "step": 4496 }, { "epoch": 0.0, "learning_rate": 4.9997614372427236e-05, "loss": 1.2191, "step": 4497 }, { "epoch": 0.0, "learning_rate": 4.9997613287205094e-05, "loss": 1.3622, "step": 4498 }, { "epoch": 0.0, "learning_rate": 4.999761220173618e-05, "loss": 1.6796, "step": 4499 }, { "epoch": 0.0, "learning_rate": 4.99976111160205e-05, "loss": 1.0056, "step": 4500 }, { "epoch": 0.0, "eval_loss": 1.0784711837768555, "eval_runtime": 85.0826, "eval_samples_per_second": 16.278, "eval_steps_per_second": 4.078, "step": 4500 }, { "epoch": 0.0, "learning_rate": 4.999761003005806e-05, "loss": 1.3483, "step": 4501 }, { "epoch": 0.0, "learning_rate": 4.999760894384885e-05, "loss": 1.1076, "step": 4502 }, { "epoch": 0.0, "learning_rate": 4.9997607857392875e-05, "loss": 0.3563, "step": 4503 }, { "epoch": 0.0, "learning_rate": 4.999760677069013e-05, "loss": 0.8917, "step": 4504 }, { "epoch": 0.0, "learning_rate": 4.9997605683740625e-05, "loss": 1.2865, "step": 4505 }, { "epoch": 0.0, "learning_rate": 4.9997604596544354e-05, "loss": 1.0649, "step": 4506 }, { "epoch": 0.0, "learning_rate": 4.999760350910132e-05, "loss": 0.8918, "step": 4507 }, { "epoch": 0.0, "learning_rate": 4.999760242141151e-05, "loss": 1.1132, "step": 4508 }, { "epoch": 0.0, "learning_rate": 4.999760133347494e-05, "loss": 1.0425, "step": 4509 }, { "epoch": 0.0, "learning_rate": 4.9997600245291604e-05, "loss": 1.1745, "step": 4510 }, { "epoch": 0.0, "learning_rate": 4.9997599156861504e-05, "loss": 0.932, "step": 4511 }, { "epoch": 0.0, "learning_rate": 4.999759806818464e-05, "loss": 1.0716, "step": 4512 }, { "epoch": 0.0, "learning_rate": 4.9997596979261e-05, "loss": 0.7812, "step": 4513 }, { "epoch": 0.0, "learning_rate": 4.99975958900906e-05, "loss": 0.8982, "step": 4514 }, { "epoch": 0.0, "learning_rate": 4.999759480067344e-05, "loss": 1.2918, "step": 4515 }, { "epoch": 0.0, "learning_rate": 4.99975937110095e-05, "loss": 0.9576, "step": 4516 }, { "epoch": 0.0, "learning_rate": 4.999759262109881e-05, "loss": 1.1978, "step": 4517 }, { "epoch": 0.0, "learning_rate": 4.9997591530941345e-05, "loss": 1.0162, "step": 4518 }, { "epoch": 0.0, "learning_rate": 4.999759044053712e-05, "loss": 1.1299, "step": 4519 }, { "epoch": 0.0, "learning_rate": 4.999758934988612e-05, "loss": 1.3023, "step": 4520 }, { "epoch": 0.0, "learning_rate": 4.999758825898837e-05, "loss": 1.0354, "step": 4521 }, { "epoch": 0.0, "learning_rate": 4.999758716784384e-05, "loss": 1.2752, "step": 4522 }, { "epoch": 0.0, "learning_rate": 4.999758607645255e-05, "loss": 0.9105, "step": 4523 }, { "epoch": 0.0, "learning_rate": 4.999758498481449e-05, "loss": 1.1427, "step": 4524 }, { "epoch": 0.0, "learning_rate": 4.9997583892929676e-05, "loss": 1.123, "step": 4525 }, { "epoch": 0.0, "learning_rate": 4.999758280079809e-05, "loss": 1.0269, "step": 4526 }, { "epoch": 0.0, "learning_rate": 4.999758170841973e-05, "loss": 1.4752, "step": 4527 }, { "epoch": 0.0, "learning_rate": 4.9997580615794615e-05, "loss": 1.9687, "step": 4528 }, { "epoch": 0.0, "learning_rate": 4.9997579522922735e-05, "loss": 1.4191, "step": 4529 }, { "epoch": 0.0, "learning_rate": 4.9997578429804084e-05, "loss": 1.8628, "step": 4530 }, { "epoch": 0.0, "learning_rate": 4.999757733643867e-05, "loss": 2.0473, "step": 4531 }, { "epoch": 0.0, "learning_rate": 4.999757624282648e-05, "loss": 1.7376, "step": 4532 }, { "epoch": 0.0, "learning_rate": 4.999757514896754e-05, "loss": 1.7128, "step": 4533 }, { "epoch": 0.0, "learning_rate": 4.999757405486183e-05, "loss": 1.2395, "step": 4534 }, { "epoch": 0.0, "learning_rate": 4.9997572960509355e-05, "loss": 0.7994, "step": 4535 }, { "epoch": 0.0, "learning_rate": 4.999757186591011e-05, "loss": 1.5496, "step": 4536 }, { "epoch": 0.0, "learning_rate": 4.9997570771064095e-05, "loss": 1.3805, "step": 4537 }, { "epoch": 0.0, "learning_rate": 4.999756967597132e-05, "loss": 1.1317, "step": 4538 }, { "epoch": 0.0, "learning_rate": 4.9997568580631785e-05, "loss": 1.0239, "step": 4539 }, { "epoch": 0.0, "learning_rate": 4.999756748504548e-05, "loss": 0.9977, "step": 4540 }, { "epoch": 0.0, "learning_rate": 4.9997566389212416e-05, "loss": 0.7617, "step": 4541 }, { "epoch": 0.0, "learning_rate": 4.999756529313258e-05, "loss": 0.7574, "step": 4542 }, { "epoch": 0.0, "learning_rate": 4.9997564196805976e-05, "loss": 0.4952, "step": 4543 }, { "epoch": 0.0, "learning_rate": 4.999756310023261e-05, "loss": 0.4071, "step": 4544 }, { "epoch": 0.0, "learning_rate": 4.999756200341248e-05, "loss": 0.3267, "step": 4545 }, { "epoch": 0.0, "learning_rate": 4.999756090634558e-05, "loss": 0.2456, "step": 4546 }, { "epoch": 0.0, "learning_rate": 4.999755980903191e-05, "loss": 0.1915, "step": 4547 }, { "epoch": 0.0, "learning_rate": 4.9997558711471486e-05, "loss": 0.2422, "step": 4548 }, { "epoch": 0.0, "learning_rate": 4.999755761366429e-05, "loss": 0.1811, "step": 4549 }, { "epoch": 0.0, "learning_rate": 4.999755651561033e-05, "loss": 0.1229, "step": 4550 }, { "epoch": 0.0, "learning_rate": 4.9997555417309613e-05, "loss": 0.1014, "step": 4551 }, { "epoch": 0.0, "learning_rate": 4.999755431876212e-05, "loss": 0.5357, "step": 4552 }, { "epoch": 0.0, "learning_rate": 4.9997553219967865e-05, "loss": 1.2375, "step": 4553 }, { "epoch": 0.0, "learning_rate": 4.999755212092685e-05, "loss": 1.3645, "step": 4554 }, { "epoch": 0.0, "learning_rate": 4.999755102163906e-05, "loss": 1.3012, "step": 4555 }, { "epoch": 0.0, "learning_rate": 4.999754992210451e-05, "loss": 1.2112, "step": 4556 }, { "epoch": 0.0, "learning_rate": 4.9997548822323195e-05, "loss": 1.3208, "step": 4557 }, { "epoch": 0.0, "learning_rate": 4.9997547722295113e-05, "loss": 1.0551, "step": 4558 }, { "epoch": 0.0, "learning_rate": 4.999754662202026e-05, "loss": 1.1901, "step": 4559 }, { "epoch": 0.0, "learning_rate": 4.9997545521498656e-05, "loss": 0.8277, "step": 4560 }, { "epoch": 0.0, "learning_rate": 4.999754442073028e-05, "loss": 0.3503, "step": 4561 }, { "epoch": 0.0, "learning_rate": 4.9997543319715135e-05, "loss": 1.1548, "step": 4562 }, { "epoch": 0.0, "learning_rate": 4.999754221845323e-05, "loss": 1.1193, "step": 4563 }, { "epoch": 0.0, "learning_rate": 4.9997541116944556e-05, "loss": 1.0134, "step": 4564 }, { "epoch": 0.0, "learning_rate": 4.9997540015189116e-05, "loss": 1.3742, "step": 4565 }, { "epoch": 0.0, "learning_rate": 4.999753891318691e-05, "loss": 1.2755, "step": 4566 }, { "epoch": 0.0, "learning_rate": 4.9997537810937944e-05, "loss": 1.4311, "step": 4567 }, { "epoch": 0.0, "learning_rate": 4.999753670844222e-05, "loss": 1.3954, "step": 4568 }, { "epoch": 0.0, "learning_rate": 4.999753560569972e-05, "loss": 1.3045, "step": 4569 }, { "epoch": 0.0, "learning_rate": 4.999753450271045e-05, "loss": 2.1417, "step": 4570 }, { "epoch": 0.0, "learning_rate": 4.999753339947442e-05, "loss": 1.2134, "step": 4571 }, { "epoch": 0.0, "learning_rate": 4.999753229599163e-05, "loss": 1.1868, "step": 4572 }, { "epoch": 0.0, "learning_rate": 4.999753119226207e-05, "loss": 1.072, "step": 4573 }, { "epoch": 0.0, "learning_rate": 4.999753008828575e-05, "loss": 0.9794, "step": 4574 }, { "epoch": 0.0, "learning_rate": 4.999752898406266e-05, "loss": 1.0879, "step": 4575 }, { "epoch": 0.0, "learning_rate": 4.99975278795928e-05, "loss": 1.0484, "step": 4576 }, { "epoch": 0.0, "learning_rate": 4.999752677487618e-05, "loss": 0.909, "step": 4577 }, { "epoch": 0.0, "learning_rate": 4.99975256699128e-05, "loss": 0.8647, "step": 4578 }, { "epoch": 0.0, "learning_rate": 4.9997524564702645e-05, "loss": 1.1287, "step": 4579 }, { "epoch": 0.0, "learning_rate": 4.999752345924573e-05, "loss": 1.1584, "step": 4580 }, { "epoch": 0.0, "learning_rate": 4.9997522353542056e-05, "loss": 1.4009, "step": 4581 }, { "epoch": 0.0, "learning_rate": 4.9997521247591614e-05, "loss": 0.9556, "step": 4582 }, { "epoch": 0.0, "learning_rate": 4.99975201413944e-05, "loss": 0.6594, "step": 4583 }, { "epoch": 0.0, "learning_rate": 4.9997519034950425e-05, "loss": 1.0026, "step": 4584 }, { "epoch": 0.0, "learning_rate": 4.9997517928259683e-05, "loss": 1.068, "step": 4585 }, { "epoch": 0.0, "learning_rate": 4.9997516821322184e-05, "loss": 1.2799, "step": 4586 }, { "epoch": 0.0, "learning_rate": 4.999751571413791e-05, "loss": 1.2804, "step": 4587 }, { "epoch": 0.0, "learning_rate": 4.999751460670688e-05, "loss": 1.1973, "step": 4588 }, { "epoch": 0.0, "learning_rate": 4.999751349902908e-05, "loss": 1.1968, "step": 4589 }, { "epoch": 0.0, "learning_rate": 4.999751239110452e-05, "loss": 1.2127, "step": 4590 }, { "epoch": 0.0, "learning_rate": 4.999751128293318e-05, "loss": 1.3084, "step": 4591 }, { "epoch": 0.0, "learning_rate": 4.99975101745151e-05, "loss": 1.0746, "step": 4592 }, { "epoch": 0.0, "learning_rate": 4.9997509065850234e-05, "loss": 1.1344, "step": 4593 }, { "epoch": 0.0, "learning_rate": 4.999750795693861e-05, "loss": 1.1177, "step": 4594 }, { "epoch": 0.0, "learning_rate": 4.999750684778022e-05, "loss": 1.4068, "step": 4595 }, { "epoch": 0.0, "learning_rate": 4.999750573837507e-05, "loss": 1.1157, "step": 4596 }, { "epoch": 0.0, "learning_rate": 4.999750462872315e-05, "loss": 1.0543, "step": 4597 }, { "epoch": 0.0, "learning_rate": 4.999750351882447e-05, "loss": 0.8104, "step": 4598 }, { "epoch": 0.0, "learning_rate": 4.999750240867902e-05, "loss": 1.3103, "step": 4599 }, { "epoch": 0.0, "learning_rate": 4.9997501298286806e-05, "loss": 1.2957, "step": 4600 }, { "epoch": 0.0, "learning_rate": 4.999750018764783e-05, "loss": 1.1328, "step": 4601 }, { "epoch": 0.0, "learning_rate": 4.999749907676209e-05, "loss": 0.9242, "step": 4602 }, { "epoch": 0.0, "learning_rate": 4.999749796562958e-05, "loss": 1.1808, "step": 4603 }, { "epoch": 0.0, "learning_rate": 4.999749685425031e-05, "loss": 1.2292, "step": 4604 }, { "epoch": 0.0, "learning_rate": 4.9997495742624275e-05, "loss": 0.935, "step": 4605 }, { "epoch": 0.0, "learning_rate": 4.9997494630751474e-05, "loss": 1.0585, "step": 4606 }, { "epoch": 0.0, "learning_rate": 4.99974935186319e-05, "loss": 0.9551, "step": 4607 }, { "epoch": 0.0, "learning_rate": 4.999749240626558e-05, "loss": 0.9226, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.999749129365248e-05, "loss": 1.1205, "step": 4609 }, { "epoch": 0.0, "learning_rate": 4.999749018079262e-05, "loss": 1.2519, "step": 4610 }, { "epoch": 0.0, "learning_rate": 4.999748906768599e-05, "loss": 1.063, "step": 4611 }, { "epoch": 0.0, "learning_rate": 4.99974879543326e-05, "loss": 1.1115, "step": 4612 }, { "epoch": 0.0, "learning_rate": 4.999748684073245e-05, "loss": 1.1365, "step": 4613 }, { "epoch": 0.0, "learning_rate": 4.999748572688553e-05, "loss": 1.1795, "step": 4614 }, { "epoch": 0.0, "learning_rate": 4.999748461279185e-05, "loss": 0.8403, "step": 4615 }, { "epoch": 0.0, "learning_rate": 4.99974834984514e-05, "loss": 0.9619, "step": 4616 }, { "epoch": 0.0, "learning_rate": 4.9997482383864186e-05, "loss": 1.5095, "step": 4617 }, { "epoch": 0.0, "learning_rate": 4.999748126903021e-05, "loss": 1.1162, "step": 4618 }, { "epoch": 0.0, "learning_rate": 4.999748015394947e-05, "loss": 0.8096, "step": 4619 }, { "epoch": 0.0, "learning_rate": 4.9997479038621964e-05, "loss": 1.1393, "step": 4620 }, { "epoch": 0.0, "learning_rate": 4.9997477923047684e-05, "loss": 1.0588, "step": 4621 }, { "epoch": 0.0, "learning_rate": 4.999747680722665e-05, "loss": 1.0214, "step": 4622 }, { "epoch": 0.0, "learning_rate": 4.9997475691158856e-05, "loss": 1.0753, "step": 4623 }, { "epoch": 0.0, "learning_rate": 4.999747457484429e-05, "loss": 1.0879, "step": 4624 }, { "epoch": 0.0, "learning_rate": 4.9997473458282964e-05, "loss": 1.1705, "step": 4625 }, { "epoch": 0.0, "learning_rate": 4.999747234147487e-05, "loss": 1.4716, "step": 4626 }, { "epoch": 0.0, "learning_rate": 4.999747122442001e-05, "loss": 0.9731, "step": 4627 }, { "epoch": 0.0, "learning_rate": 4.9997470107118383e-05, "loss": 0.9341, "step": 4628 }, { "epoch": 0.0, "learning_rate": 4.9997468989569994e-05, "loss": 1.1177, "step": 4629 }, { "epoch": 0.0, "learning_rate": 4.999746787177485e-05, "loss": 1.4303, "step": 4630 }, { "epoch": 0.0, "learning_rate": 4.999746675373292e-05, "loss": 1.0908, "step": 4631 }, { "epoch": 0.0, "learning_rate": 4.999746563544425e-05, "loss": 1.4172, "step": 4632 }, { "epoch": 0.0, "learning_rate": 4.99974645169088e-05, "loss": 0.9805, "step": 4633 }, { "epoch": 0.0, "learning_rate": 4.9997463398126596e-05, "loss": 1.1843, "step": 4634 }, { "epoch": 0.0, "learning_rate": 4.999746227909762e-05, "loss": 1.1419, "step": 4635 }, { "epoch": 0.0, "learning_rate": 4.999746115982188e-05, "loss": 0.9037, "step": 4636 }, { "epoch": 0.0, "learning_rate": 4.9997460040299376e-05, "loss": 0.9535, "step": 4637 }, { "epoch": 0.0, "learning_rate": 4.9997458920530113e-05, "loss": 1.1616, "step": 4638 }, { "epoch": 0.0, "learning_rate": 4.999745780051408e-05, "loss": 0.9401, "step": 4639 }, { "epoch": 0.0, "learning_rate": 4.999745668025128e-05, "loss": 0.9385, "step": 4640 }, { "epoch": 0.0, "learning_rate": 4.999745555974173e-05, "loss": 1.5485, "step": 4641 }, { "epoch": 0.0, "learning_rate": 4.99974544389854e-05, "loss": 1.2252, "step": 4642 }, { "epoch": 0.0, "learning_rate": 4.999745331798231e-05, "loss": 1.1477, "step": 4643 }, { "epoch": 0.0, "learning_rate": 4.9997452196732454e-05, "loss": 1.497, "step": 4644 }, { "epoch": 0.0, "learning_rate": 4.9997451075235834e-05, "loss": 0.9156, "step": 4645 }, { "epoch": 0.0, "learning_rate": 4.999744995349245e-05, "loss": 0.6902, "step": 4646 }, { "epoch": 0.0, "learning_rate": 4.999744883150231e-05, "loss": 0.6409, "step": 4647 }, { "epoch": 0.0, "learning_rate": 4.99974477092654e-05, "loss": 0.8875, "step": 4648 }, { "epoch": 0.0, "learning_rate": 4.999744658678172e-05, "loss": 1.017, "step": 4649 }, { "epoch": 0.0, "learning_rate": 4.999744546405128e-05, "loss": 1.4372, "step": 4650 }, { "epoch": 0.0, "learning_rate": 4.9997444341074076e-05, "loss": 1.0174, "step": 4651 }, { "epoch": 0.0, "learning_rate": 4.999744321785011e-05, "loss": 1.1617, "step": 4652 }, { "epoch": 0.0, "learning_rate": 4.999744209437938e-05, "loss": 1.3952, "step": 4653 }, { "epoch": 0.0, "learning_rate": 4.999744097066188e-05, "loss": 1.1098, "step": 4654 }, { "epoch": 0.0, "learning_rate": 4.999743984669763e-05, "loss": 1.1052, "step": 4655 }, { "epoch": 0.0, "learning_rate": 4.99974387224866e-05, "loss": 0.9504, "step": 4656 }, { "epoch": 0.0, "learning_rate": 4.9997437598028806e-05, "loss": 0.9464, "step": 4657 }, { "epoch": 0.0, "learning_rate": 4.9997436473324256e-05, "loss": 1.1921, "step": 4658 }, { "epoch": 0.0, "learning_rate": 4.999743534837294e-05, "loss": 1.0369, "step": 4659 }, { "epoch": 0.0, "learning_rate": 4.999743422317486e-05, "loss": 1.0909, "step": 4660 }, { "epoch": 0.0, "learning_rate": 4.999743309773002e-05, "loss": 0.9977, "step": 4661 }, { "epoch": 0.0, "learning_rate": 4.99974319720384e-05, "loss": 1.1727, "step": 4662 }, { "epoch": 0.0, "learning_rate": 4.999743084610003e-05, "loss": 0.9632, "step": 4663 }, { "epoch": 0.0, "learning_rate": 4.999742971991489e-05, "loss": 1.3036, "step": 4664 }, { "epoch": 0.0, "learning_rate": 4.999742859348299e-05, "loss": 1.1178, "step": 4665 }, { "epoch": 0.0, "learning_rate": 4.999742746680433e-05, "loss": 1.66, "step": 4666 }, { "epoch": 0.0, "learning_rate": 4.99974263398789e-05, "loss": 0.2656, "step": 4667 }, { "epoch": 0.0, "learning_rate": 4.999742521270671e-05, "loss": 0.4481, "step": 4668 }, { "epoch": 0.0, "learning_rate": 4.999742408528775e-05, "loss": 0.273, "step": 4669 }, { "epoch": 0.0, "learning_rate": 4.999742295762202e-05, "loss": 1.0934, "step": 4670 }, { "epoch": 0.0, "learning_rate": 4.999742182970953e-05, "loss": 0.6311, "step": 4671 }, { "epoch": 0.0, "learning_rate": 4.9997420701550286e-05, "loss": 0.6401, "step": 4672 }, { "epoch": 0.0, "learning_rate": 4.999741957314427e-05, "loss": 1.102, "step": 4673 }, { "epoch": 0.0, "learning_rate": 4.9997418444491496e-05, "loss": 0.8887, "step": 4674 }, { "epoch": 0.0, "learning_rate": 4.999741731559196e-05, "loss": 1.336, "step": 4675 }, { "epoch": 0.0, "learning_rate": 4.999741618644565e-05, "loss": 1.7969, "step": 4676 }, { "epoch": 0.0, "learning_rate": 4.999741505705258e-05, "loss": 1.2247, "step": 4677 }, { "epoch": 0.0, "learning_rate": 4.999741392741274e-05, "loss": 1.0536, "step": 4678 }, { "epoch": 0.0, "learning_rate": 4.9997412797526144e-05, "loss": 1.0859, "step": 4679 }, { "epoch": 0.0, "learning_rate": 4.999741166739278e-05, "loss": 1.4554, "step": 4680 }, { "epoch": 0.0, "learning_rate": 4.999741053701266e-05, "loss": 1.3994, "step": 4681 }, { "epoch": 0.0, "learning_rate": 4.999740940638576e-05, "loss": 1.5389, "step": 4682 }, { "epoch": 0.0, "learning_rate": 4.9997408275512115e-05, "loss": 1.4944, "step": 4683 }, { "epoch": 0.0, "learning_rate": 4.9997407144391697e-05, "loss": 1.5548, "step": 4684 }, { "epoch": 0.0, "learning_rate": 4.9997406013024514e-05, "loss": 1.2535, "step": 4685 }, { "epoch": 0.0, "learning_rate": 4.999740488141057e-05, "loss": 1.2727, "step": 4686 }, { "epoch": 0.0, "learning_rate": 4.999740374954986e-05, "loss": 0.9823, "step": 4687 }, { "epoch": 0.0, "learning_rate": 4.999740261744239e-05, "loss": 1.006, "step": 4688 }, { "epoch": 0.0, "learning_rate": 4.999740148508815e-05, "loss": 1.1687, "step": 4689 }, { "epoch": 0.0, "learning_rate": 4.9997400352487154e-05, "loss": 0.9116, "step": 4690 }, { "epoch": 0.0, "learning_rate": 4.9997399219639384e-05, "loss": 1.3216, "step": 4691 }, { "epoch": 0.0, "learning_rate": 4.999739808654486e-05, "loss": 1.1733, "step": 4692 }, { "epoch": 0.0, "learning_rate": 4.9997396953203566e-05, "loss": 1.2683, "step": 4693 }, { "epoch": 0.0, "learning_rate": 4.999739581961551e-05, "loss": 1.9192, "step": 4694 }, { "epoch": 0.0, "learning_rate": 4.999739468578069e-05, "loss": 1.497, "step": 4695 }, { "epoch": 0.0, "learning_rate": 4.999739355169911e-05, "loss": 1.1082, "step": 4696 }, { "epoch": 0.0, "learning_rate": 4.9997392417370756e-05, "loss": 1.5696, "step": 4697 }, { "epoch": 0.0, "learning_rate": 4.999739128279565e-05, "loss": 0.9002, "step": 4698 }, { "epoch": 0.0, "learning_rate": 4.999739014797378e-05, "loss": 0.6441, "step": 4699 }, { "epoch": 0.0, "learning_rate": 4.9997389012905136e-05, "loss": 1.2714, "step": 4700 }, { "epoch": 0.0, "learning_rate": 4.999738787758974e-05, "loss": 1.1863, "step": 4701 }, { "epoch": 0.0, "learning_rate": 4.999738674202757e-05, "loss": 1.1173, "step": 4702 }, { "epoch": 0.0, "learning_rate": 4.9997385606218644e-05, "loss": 1.4007, "step": 4703 }, { "epoch": 0.0, "learning_rate": 4.999738447016295e-05, "loss": 1.0387, "step": 4704 }, { "epoch": 0.0, "learning_rate": 4.999738333386049e-05, "loss": 1.1106, "step": 4705 }, { "epoch": 0.0, "learning_rate": 4.999738219731127e-05, "loss": 1.189, "step": 4706 }, { "epoch": 0.0, "learning_rate": 4.999738106051529e-05, "loss": 1.2896, "step": 4707 }, { "epoch": 0.0, "learning_rate": 4.999737992347254e-05, "loss": 1.2141, "step": 4708 }, { "epoch": 0.0, "learning_rate": 4.999737878618304e-05, "loss": 1.0269, "step": 4709 }, { "epoch": 0.0, "learning_rate": 4.999737764864676e-05, "loss": 1.1955, "step": 4710 }, { "epoch": 0.0, "learning_rate": 4.999737651086372e-05, "loss": 1.3306, "step": 4711 }, { "epoch": 0.0, "learning_rate": 4.999737537283392e-05, "loss": 0.8874, "step": 4712 }, { "epoch": 0.0, "learning_rate": 4.9997374234557354e-05, "loss": 1.1114, "step": 4713 }, { "epoch": 0.0, "learning_rate": 4.999737309603403e-05, "loss": 1.0591, "step": 4714 }, { "epoch": 0.0, "learning_rate": 4.9997371957263935e-05, "loss": 1.1689, "step": 4715 }, { "epoch": 0.0, "learning_rate": 4.999737081824708e-05, "loss": 1.0724, "step": 4716 }, { "epoch": 0.0, "learning_rate": 4.999736967898346e-05, "loss": 1.063, "step": 4717 }, { "epoch": 0.0, "learning_rate": 4.999736853947308e-05, "loss": 1.1839, "step": 4718 }, { "epoch": 0.0, "learning_rate": 4.999736739971593e-05, "loss": 1.0832, "step": 4719 }, { "epoch": 0.0, "learning_rate": 4.999736625971202e-05, "loss": 1.253, "step": 4720 }, { "epoch": 0.0, "learning_rate": 4.999736511946135e-05, "loss": 0.9533, "step": 4721 }, { "epoch": 0.0, "learning_rate": 4.999736397896391e-05, "loss": 0.6984, "step": 4722 }, { "epoch": 0.0, "learning_rate": 4.999736283821971e-05, "loss": 1.1763, "step": 4723 }, { "epoch": 0.0, "learning_rate": 4.999736169722874e-05, "loss": 1.289, "step": 4724 }, { "epoch": 0.0, "learning_rate": 4.999736055599102e-05, "loss": 1.1467, "step": 4725 }, { "epoch": 0.0, "learning_rate": 4.999735941450653e-05, "loss": 0.9113, "step": 4726 }, { "epoch": 0.0, "learning_rate": 4.9997358272775273e-05, "loss": 1.0437, "step": 4727 }, { "epoch": 0.0, "learning_rate": 4.999735713079726e-05, "loss": 1.0288, "step": 4728 }, { "epoch": 0.0, "learning_rate": 4.999735598857248e-05, "loss": 1.0476, "step": 4729 }, { "epoch": 0.0, "learning_rate": 4.999735484610094e-05, "loss": 0.4842, "step": 4730 }, { "epoch": 0.0, "learning_rate": 4.9997353703382625e-05, "loss": 1.1656, "step": 4731 }, { "epoch": 0.0, "learning_rate": 4.9997352560417556e-05, "loss": 1.1774, "step": 4732 }, { "epoch": 0.0, "learning_rate": 4.999735141720572e-05, "loss": 1.4356, "step": 4733 }, { "epoch": 0.0, "learning_rate": 4.999735027374712e-05, "loss": 0.9103, "step": 4734 }, { "epoch": 0.0, "learning_rate": 4.999734913004177e-05, "loss": 1.6251, "step": 4735 }, { "epoch": 0.0, "learning_rate": 4.999734798608964e-05, "loss": 1.2213, "step": 4736 }, { "epoch": 0.0, "learning_rate": 4.9997346841890755e-05, "loss": 1.0874, "step": 4737 }, { "epoch": 0.0, "learning_rate": 4.9997345697445106e-05, "loss": 1.3342, "step": 4738 }, { "epoch": 0.0, "learning_rate": 4.999734455275269e-05, "loss": 1.1486, "step": 4739 }, { "epoch": 0.0, "learning_rate": 4.9997343407813514e-05, "loss": 1.2113, "step": 4740 }, { "epoch": 0.0, "learning_rate": 4.999734226262757e-05, "loss": 1.1141, "step": 4741 }, { "epoch": 0.0, "learning_rate": 4.999734111719487e-05, "loss": 1.1153, "step": 4742 }, { "epoch": 0.0, "learning_rate": 4.99973399715154e-05, "loss": 1.1899, "step": 4743 }, { "epoch": 0.0, "learning_rate": 4.999733882558918e-05, "loss": 1.1211, "step": 4744 }, { "epoch": 0.0, "learning_rate": 4.999733767941618e-05, "loss": 1.0558, "step": 4745 }, { "epoch": 0.0, "learning_rate": 4.9997336532996426e-05, "loss": 1.1537, "step": 4746 }, { "epoch": 0.0, "learning_rate": 4.99973353863299e-05, "loss": 1.1311, "step": 4747 }, { "epoch": 0.0, "learning_rate": 4.9997334239416623e-05, "loss": 0.7318, "step": 4748 }, { "epoch": 0.0, "learning_rate": 4.999733309225658e-05, "loss": 0.8849, "step": 4749 }, { "epoch": 0.0, "learning_rate": 4.999733194484977e-05, "loss": 1.0484, "step": 4750 }, { "epoch": 0.0, "learning_rate": 4.99973307971962e-05, "loss": 1.1667, "step": 4751 }, { "epoch": 0.0, "learning_rate": 4.9997329649295866e-05, "loss": 1.5542, "step": 4752 }, { "epoch": 0.0, "learning_rate": 4.9997328501148764e-05, "loss": 1.0831, "step": 4753 }, { "epoch": 0.0, "learning_rate": 4.9997327352754905e-05, "loss": 1.4354, "step": 4754 }, { "epoch": 0.0, "learning_rate": 4.999732620411428e-05, "loss": 1.2352, "step": 4755 }, { "epoch": 0.0, "learning_rate": 4.999732505522689e-05, "loss": 1.4719, "step": 4756 }, { "epoch": 0.0, "learning_rate": 4.999732390609274e-05, "loss": 1.057, "step": 4757 }, { "epoch": 0.0, "learning_rate": 4.999732275671183e-05, "loss": 1.0866, "step": 4758 }, { "epoch": 0.0, "learning_rate": 4.9997321607084154e-05, "loss": 1.104, "step": 4759 }, { "epoch": 0.0, "learning_rate": 4.9997320457209714e-05, "loss": 1.0629, "step": 4760 }, { "epoch": 0.0, "learning_rate": 4.999731930708851e-05, "loss": 1.6223, "step": 4761 }, { "epoch": 0.0, "learning_rate": 4.999731815672054e-05, "loss": 1.3682, "step": 4762 }, { "epoch": 0.0, "learning_rate": 4.9997317006105816e-05, "loss": 1.2582, "step": 4763 }, { "epoch": 0.0, "learning_rate": 4.999731585524432e-05, "loss": 0.9874, "step": 4764 }, { "epoch": 0.0, "learning_rate": 4.9997314704136065e-05, "loss": 0.8081, "step": 4765 }, { "epoch": 0.0, "learning_rate": 4.999731355278105e-05, "loss": 1.0638, "step": 4766 }, { "epoch": 0.0, "learning_rate": 4.9997312401179276e-05, "loss": 1.5015, "step": 4767 }, { "epoch": 0.0, "learning_rate": 4.999731124933073e-05, "loss": 1.5185, "step": 4768 }, { "epoch": 0.0, "learning_rate": 4.999731009723542e-05, "loss": 1.2712, "step": 4769 }, { "epoch": 0.0, "learning_rate": 4.999730894489335e-05, "loss": 1.0055, "step": 4770 }, { "epoch": 0.0, "learning_rate": 4.999730779230452e-05, "loss": 1.085, "step": 4771 }, { "epoch": 0.0, "learning_rate": 4.9997306639468926e-05, "loss": 1.2079, "step": 4772 }, { "epoch": 0.0, "learning_rate": 4.999730548638657e-05, "loss": 1.1686, "step": 4773 }, { "epoch": 0.0, "learning_rate": 4.999730433305745e-05, "loss": 1.0924, "step": 4774 }, { "epoch": 0.0, "learning_rate": 4.999730317948156e-05, "loss": 1.3168, "step": 4775 }, { "epoch": 0.0, "learning_rate": 4.999730202565892e-05, "loss": 1.1525, "step": 4776 }, { "epoch": 0.0, "learning_rate": 4.9997300871589506e-05, "loss": 1.1351, "step": 4777 }, { "epoch": 0.0, "learning_rate": 4.999729971727334e-05, "loss": 1.0702, "step": 4778 }, { "epoch": 0.0, "learning_rate": 4.9997298562710405e-05, "loss": 1.0449, "step": 4779 }, { "epoch": 0.0, "learning_rate": 4.99972974079007e-05, "loss": 1.6182, "step": 4780 }, { "epoch": 0.0, "learning_rate": 4.9997296252844246e-05, "loss": 1.2316, "step": 4781 }, { "epoch": 0.0, "learning_rate": 4.999729509754102e-05, "loss": 1.1996, "step": 4782 }, { "epoch": 0.0, "learning_rate": 4.999729394199103e-05, "loss": 1.1948, "step": 4783 }, { "epoch": 0.0, "learning_rate": 4.9997292786194285e-05, "loss": 0.9327, "step": 4784 }, { "epoch": 0.0, "learning_rate": 4.9997291630150776e-05, "loss": 1.0551, "step": 4785 }, { "epoch": 0.0, "learning_rate": 4.99972904738605e-05, "loss": 1.0891, "step": 4786 }, { "epoch": 0.0, "learning_rate": 4.999728931732346e-05, "loss": 1.0666, "step": 4787 }, { "epoch": 0.0, "learning_rate": 4.999728816053966e-05, "loss": 1.1281, "step": 4788 }, { "epoch": 0.0, "learning_rate": 4.99972870035091e-05, "loss": 1.5918, "step": 4789 }, { "epoch": 0.0, "learning_rate": 4.999728584623178e-05, "loss": 1.0612, "step": 4790 }, { "epoch": 0.0, "learning_rate": 4.999728468870769e-05, "loss": 1.0924, "step": 4791 }, { "epoch": 0.0, "learning_rate": 4.9997283530936836e-05, "loss": 1.1297, "step": 4792 }, { "epoch": 0.0, "learning_rate": 4.9997282372919225e-05, "loss": 1.176, "step": 4793 }, { "epoch": 0.0, "learning_rate": 4.999728121465485e-05, "loss": 1.1923, "step": 4794 }, { "epoch": 0.0, "learning_rate": 4.999728005614371e-05, "loss": 1.0567, "step": 4795 }, { "epoch": 0.0, "learning_rate": 4.999727889738581e-05, "loss": 1.1293, "step": 4796 }, { "epoch": 0.0, "learning_rate": 4.999727773838115e-05, "loss": 0.6926, "step": 4797 }, { "epoch": 0.0, "learning_rate": 4.999727657912972e-05, "loss": 1.2504, "step": 4798 }, { "epoch": 0.0, "learning_rate": 4.9997275419631535e-05, "loss": 0.7057, "step": 4799 }, { "epoch": 0.0, "learning_rate": 4.999727425988659e-05, "loss": 0.9938, "step": 4800 }, { "epoch": 0.0, "learning_rate": 4.999727309989487e-05, "loss": 1.1662, "step": 4801 }, { "epoch": 0.0, "learning_rate": 4.999727193965639e-05, "loss": 1.3126, "step": 4802 }, { "epoch": 0.0, "learning_rate": 4.999727077917115e-05, "loss": 1.1409, "step": 4803 }, { "epoch": 0.0, "learning_rate": 4.999726961843915e-05, "loss": 1.003, "step": 4804 }, { "epoch": 0.0, "learning_rate": 4.9997268457460386e-05, "loss": 1.1619, "step": 4805 }, { "epoch": 0.0, "learning_rate": 4.9997267296234865e-05, "loss": 1.1168, "step": 4806 }, { "epoch": 0.0, "learning_rate": 4.999726613476257e-05, "loss": 1.2055, "step": 4807 }, { "epoch": 0.0, "learning_rate": 4.999726497304352e-05, "loss": 1.1727, "step": 4808 }, { "epoch": 0.0, "learning_rate": 4.999726381107771e-05, "loss": 1.1429, "step": 4809 }, { "epoch": 0.0, "learning_rate": 4.999726264886513e-05, "loss": 1.2134, "step": 4810 }, { "epoch": 0.0, "learning_rate": 4.9997261486405787e-05, "loss": 0.9725, "step": 4811 }, { "epoch": 0.0, "learning_rate": 4.999726032369969e-05, "loss": 1.3631, "step": 4812 }, { "epoch": 0.0, "learning_rate": 4.999725916074683e-05, "loss": 1.3103, "step": 4813 }, { "epoch": 0.0, "learning_rate": 4.99972579975472e-05, "loss": 1.0344, "step": 4814 }, { "epoch": 0.0, "learning_rate": 4.999725683410082e-05, "loss": 1.7591, "step": 4815 }, { "epoch": 0.0, "learning_rate": 4.999725567040766e-05, "loss": 1.6107, "step": 4816 }, { "epoch": 0.0, "learning_rate": 4.999725450646775e-05, "loss": 1.2663, "step": 4817 }, { "epoch": 0.0, "learning_rate": 4.9997253342281075e-05, "loss": 1.1365, "step": 4818 }, { "epoch": 0.0, "learning_rate": 4.999725217784764e-05, "loss": 0.8303, "step": 4819 }, { "epoch": 0.0, "learning_rate": 4.9997251013167435e-05, "loss": 0.4453, "step": 4820 }, { "epoch": 0.0, "learning_rate": 4.9997249848240474e-05, "loss": 1.1691, "step": 4821 }, { "epoch": 0.0, "learning_rate": 4.999724868306675e-05, "loss": 1.5815, "step": 4822 }, { "epoch": 0.0, "learning_rate": 4.999724751764626e-05, "loss": 1.109, "step": 4823 }, { "epoch": 0.0, "learning_rate": 4.999724635197901e-05, "loss": 0.8534, "step": 4824 }, { "epoch": 0.0, "learning_rate": 4.9997245186065e-05, "loss": 1.1309, "step": 4825 }, { "epoch": 0.0, "learning_rate": 4.999724401990423e-05, "loss": 1.1172, "step": 4826 }, { "epoch": 0.0, "learning_rate": 4.999724285349669e-05, "loss": 1.7788, "step": 4827 }, { "epoch": 0.0, "learning_rate": 4.999724168684239e-05, "loss": 1.3092, "step": 4828 }, { "epoch": 0.0, "learning_rate": 4.9997240519941335e-05, "loss": 1.0972, "step": 4829 }, { "epoch": 0.0, "learning_rate": 4.9997239352793515e-05, "loss": 0.6636, "step": 4830 }, { "epoch": 0.0, "learning_rate": 4.999723818539892e-05, "loss": 0.448, "step": 4831 }, { "epoch": 0.0, "learning_rate": 4.9997237017757576e-05, "loss": 0.4685, "step": 4832 }, { "epoch": 0.0, "learning_rate": 4.999723584986947e-05, "loss": 0.3573, "step": 4833 }, { "epoch": 0.0, "learning_rate": 4.999723468173459e-05, "loss": 0.8482, "step": 4834 }, { "epoch": 0.0, "learning_rate": 4.999723351335296e-05, "loss": 1.3927, "step": 4835 }, { "epoch": 0.0, "learning_rate": 4.9997232344724565e-05, "loss": 1.117, "step": 4836 }, { "epoch": 0.0, "learning_rate": 4.999723117584941e-05, "loss": 1.0969, "step": 4837 }, { "epoch": 0.0, "learning_rate": 4.9997230006727486e-05, "loss": 1.0169, "step": 4838 }, { "epoch": 0.0, "learning_rate": 4.99972288373588e-05, "loss": 0.8773, "step": 4839 }, { "epoch": 0.0, "learning_rate": 4.9997227667743364e-05, "loss": 1.0116, "step": 4840 }, { "epoch": 0.0, "learning_rate": 4.9997226497881156e-05, "loss": 1.153, "step": 4841 }, { "epoch": 0.0, "learning_rate": 4.9997225327772184e-05, "loss": 0.7223, "step": 4842 }, { "epoch": 0.0, "learning_rate": 4.9997224157416454e-05, "loss": 0.5854, "step": 4843 }, { "epoch": 0.0, "learning_rate": 4.9997222986813966e-05, "loss": 1.1327, "step": 4844 }, { "epoch": 0.0, "learning_rate": 4.99972218159647e-05, "loss": 1.3386, "step": 4845 }, { "epoch": 0.0, "learning_rate": 4.999722064486869e-05, "loss": 0.6762, "step": 4846 }, { "epoch": 0.0, "learning_rate": 4.999721947352591e-05, "loss": 1.5002, "step": 4847 }, { "epoch": 0.0, "learning_rate": 4.9997218301936365e-05, "loss": 1.6148, "step": 4848 }, { "epoch": 0.0, "learning_rate": 4.999721713010007e-05, "loss": 1.7661, "step": 4849 }, { "epoch": 0.0, "learning_rate": 4.9997215958016995e-05, "loss": 1.9748, "step": 4850 }, { "epoch": 0.0, "learning_rate": 4.999721478568718e-05, "loss": 1.3449, "step": 4851 }, { "epoch": 0.0, "learning_rate": 4.999721361311058e-05, "loss": 1.1963, "step": 4852 }, { "epoch": 0.0, "learning_rate": 4.9997212440287234e-05, "loss": 1.0218, "step": 4853 }, { "epoch": 0.0, "learning_rate": 4.999721126721712e-05, "loss": 1.3188, "step": 4854 }, { "epoch": 0.0, "learning_rate": 4.999721009390025e-05, "loss": 1.0898, "step": 4855 }, { "epoch": 0.0, "learning_rate": 4.9997208920336606e-05, "loss": 1.1505, "step": 4856 }, { "epoch": 0.0, "learning_rate": 4.999720774652621e-05, "loss": 1.1612, "step": 4857 }, { "epoch": 0.0, "learning_rate": 4.9997206572469046e-05, "loss": 1.2213, "step": 4858 }, { "epoch": 0.0, "learning_rate": 4.9997205398165126e-05, "loss": 1.2131, "step": 4859 }, { "epoch": 0.0, "learning_rate": 4.999720422361445e-05, "loss": 1.0219, "step": 4860 }, { "epoch": 0.0, "learning_rate": 4.9997203048817e-05, "loss": 1.2768, "step": 4861 }, { "epoch": 0.0, "learning_rate": 4.9997201873772794e-05, "loss": 1.0464, "step": 4862 }, { "epoch": 0.0, "learning_rate": 4.9997200698481816e-05, "loss": 1.0345, "step": 4863 }, { "epoch": 0.0, "learning_rate": 4.999719952294409e-05, "loss": 1.2539, "step": 4864 }, { "epoch": 0.0, "learning_rate": 4.999719834715959e-05, "loss": 1.0661, "step": 4865 }, { "epoch": 0.0, "learning_rate": 4.999719717112834e-05, "loss": 1.0236, "step": 4866 }, { "epoch": 0.0, "learning_rate": 4.999719599485032e-05, "loss": 1.1617, "step": 4867 }, { "epoch": 0.0, "learning_rate": 4.9997194818325544e-05, "loss": 0.8693, "step": 4868 }, { "epoch": 0.0, "learning_rate": 4.9997193641554e-05, "loss": 1.0996, "step": 4869 }, { "epoch": 0.0, "learning_rate": 4.9997192464535705e-05, "loss": 1.2102, "step": 4870 }, { "epoch": 0.0, "learning_rate": 4.999719128727063e-05, "loss": 0.6804, "step": 4871 }, { "epoch": 0.0, "learning_rate": 4.999719010975881e-05, "loss": 0.9678, "step": 4872 }, { "epoch": 0.0, "learning_rate": 4.999718893200022e-05, "loss": 0.9744, "step": 4873 }, { "epoch": 0.0, "learning_rate": 4.9997187753994876e-05, "loss": 1.253, "step": 4874 }, { "epoch": 0.0, "learning_rate": 4.999718657574276e-05, "loss": 1.1902, "step": 4875 }, { "epoch": 0.0, "learning_rate": 4.9997185397243885e-05, "loss": 0.7328, "step": 4876 }, { "epoch": 0.0, "learning_rate": 4.9997184218498247e-05, "loss": 0.7258, "step": 4877 }, { "epoch": 0.0, "learning_rate": 4.999718303950586e-05, "loss": 0.5045, "step": 4878 }, { "epoch": 0.0, "learning_rate": 4.99971818602667e-05, "loss": 0.601, "step": 4879 }, { "epoch": 0.0, "learning_rate": 4.999718068078078e-05, "loss": 0.4128, "step": 4880 }, { "epoch": 0.0, "learning_rate": 4.9997179501048095e-05, "loss": 0.7225, "step": 4881 }, { "epoch": 0.0, "learning_rate": 4.9997178321068655e-05, "loss": 1.0695, "step": 4882 }, { "epoch": 0.0, "learning_rate": 4.9997177140842457e-05, "loss": 1.2155, "step": 4883 }, { "epoch": 0.0, "learning_rate": 4.999717596036949e-05, "loss": 1.5828, "step": 4884 }, { "epoch": 0.0, "learning_rate": 4.999717477964976e-05, "loss": 0.9714, "step": 4885 }, { "epoch": 0.0, "learning_rate": 4.999717359868327e-05, "loss": 1.0561, "step": 4886 }, { "epoch": 0.0, "learning_rate": 4.999717241747002e-05, "loss": 1.0278, "step": 4887 }, { "epoch": 0.0, "learning_rate": 4.999717123601001e-05, "loss": 0.9739, "step": 4888 }, { "epoch": 0.0, "learning_rate": 4.9997170054303236e-05, "loss": 1.0747, "step": 4889 }, { "epoch": 0.0, "learning_rate": 4.99971688723497e-05, "loss": 1.3003, "step": 4890 }, { "epoch": 0.0, "learning_rate": 4.999716769014941e-05, "loss": 1.0654, "step": 4891 }, { "epoch": 0.0, "learning_rate": 4.999716650770234e-05, "loss": 1.1268, "step": 4892 }, { "epoch": 0.0, "learning_rate": 4.999716532500852e-05, "loss": 1.0623, "step": 4893 }, { "epoch": 0.0, "learning_rate": 4.999716414206794e-05, "loss": 1.1484, "step": 4894 }, { "epoch": 0.0, "learning_rate": 4.99971629588806e-05, "loss": 1.3828, "step": 4895 }, { "epoch": 0.0, "learning_rate": 4.9997161775446496e-05, "loss": 1.2322, "step": 4896 }, { "epoch": 0.0, "learning_rate": 4.999716059176563e-05, "loss": 1.1974, "step": 4897 }, { "epoch": 0.0, "learning_rate": 4.9997159407838e-05, "loss": 1.1084, "step": 4898 }, { "epoch": 0.0, "learning_rate": 4.999715822366361e-05, "loss": 1.2828, "step": 4899 }, { "epoch": 0.0, "learning_rate": 4.9997157039242466e-05, "loss": 1.0407, "step": 4900 }, { "epoch": 0.0, "learning_rate": 4.999715585457455e-05, "loss": 1.1489, "step": 4901 }, { "epoch": 0.0, "learning_rate": 4.999715466965988e-05, "loss": 1.4901, "step": 4902 }, { "epoch": 0.0, "learning_rate": 4.999715348449845e-05, "loss": 1.0606, "step": 4903 }, { "epoch": 0.0, "learning_rate": 4.999715229909025e-05, "loss": 1.0927, "step": 4904 }, { "epoch": 0.0, "learning_rate": 4.999715111343529e-05, "loss": 1.3185, "step": 4905 }, { "epoch": 0.0, "learning_rate": 4.999714992753358e-05, "loss": 1.3352, "step": 4906 }, { "epoch": 0.0, "learning_rate": 4.99971487413851e-05, "loss": 1.4179, "step": 4907 }, { "epoch": 0.0, "learning_rate": 4.9997147554989856e-05, "loss": 1.1385, "step": 4908 }, { "epoch": 0.0, "learning_rate": 4.999714636834785e-05, "loss": 1.1543, "step": 4909 }, { "epoch": 0.0, "learning_rate": 4.9997145181459095e-05, "loss": 1.3222, "step": 4910 }, { "epoch": 0.0, "learning_rate": 4.999714399432356e-05, "loss": 1.5961, "step": 4911 }, { "epoch": 0.0, "learning_rate": 4.9997142806941276e-05, "loss": 1.1299, "step": 4912 }, { "epoch": 0.0, "learning_rate": 4.9997141619312234e-05, "loss": 1.1121, "step": 4913 }, { "epoch": 0.0, "learning_rate": 4.999714043143642e-05, "loss": 1.0932, "step": 4914 }, { "epoch": 0.0, "learning_rate": 4.999713924331385e-05, "loss": 1.041, "step": 4915 }, { "epoch": 0.0, "learning_rate": 4.999713805494453e-05, "loss": 1.1886, "step": 4916 }, { "epoch": 0.0, "learning_rate": 4.9997136866328434e-05, "loss": 0.8678, "step": 4917 }, { "epoch": 0.0, "learning_rate": 4.9997135677465576e-05, "loss": 0.9387, "step": 4918 }, { "epoch": 0.0, "learning_rate": 4.999713448835596e-05, "loss": 1.246, "step": 4919 }, { "epoch": 0.0, "learning_rate": 4.999713329899959e-05, "loss": 1.2206, "step": 4920 }, { "epoch": 0.0, "learning_rate": 4.999713210939645e-05, "loss": 0.9171, "step": 4921 }, { "epoch": 0.0, "learning_rate": 4.9997130919546555e-05, "loss": 1.0763, "step": 4922 }, { "epoch": 0.0, "learning_rate": 4.9997129729449896e-05, "loss": 1.3132, "step": 4923 }, { "epoch": 0.0, "learning_rate": 4.999712853910647e-05, "loss": 0.8297, "step": 4924 }, { "epoch": 0.0, "learning_rate": 4.999712734851629e-05, "loss": 1.1485, "step": 4925 }, { "epoch": 0.0, "learning_rate": 4.9997126157679344e-05, "loss": 0.9347, "step": 4926 }, { "epoch": 0.0, "learning_rate": 4.999712496659564e-05, "loss": 1.1448, "step": 4927 }, { "epoch": 0.0, "learning_rate": 4.999712377526518e-05, "loss": 1.4814, "step": 4928 }, { "epoch": 0.0, "learning_rate": 4.9997122583687954e-05, "loss": 0.9763, "step": 4929 }, { "epoch": 0.0, "learning_rate": 4.9997121391863964e-05, "loss": 1.3074, "step": 4930 }, { "epoch": 0.0, "learning_rate": 4.999712019979322e-05, "loss": 1.1433, "step": 4931 }, { "epoch": 0.0, "learning_rate": 4.999711900747571e-05, "loss": 1.2763, "step": 4932 }, { "epoch": 0.0, "learning_rate": 4.999711781491144e-05, "loss": 0.9959, "step": 4933 }, { "epoch": 0.0, "learning_rate": 4.999711662210041e-05, "loss": 0.9883, "step": 4934 }, { "epoch": 0.0, "learning_rate": 4.9997115429042615e-05, "loss": 1.2426, "step": 4935 }, { "epoch": 0.0, "learning_rate": 4.9997114235738066e-05, "loss": 1.2025, "step": 4936 }, { "epoch": 0.0, "learning_rate": 4.999711304218675e-05, "loss": 1.0928, "step": 4937 }, { "epoch": 0.0, "learning_rate": 4.999711184838868e-05, "loss": 0.9726, "step": 4938 }, { "epoch": 0.0, "learning_rate": 4.999711065434384e-05, "loss": 1.0069, "step": 4939 }, { "epoch": 0.0, "learning_rate": 4.9997109460052244e-05, "loss": 1.2129, "step": 4940 }, { "epoch": 0.0, "learning_rate": 4.9997108265513886e-05, "loss": 0.8876, "step": 4941 }, { "epoch": 0.0, "learning_rate": 4.9997107070728764e-05, "loss": 1.1348, "step": 4942 }, { "epoch": 0.0, "learning_rate": 4.9997105875696884e-05, "loss": 1.8116, "step": 4943 }, { "epoch": 0.0, "learning_rate": 4.9997104680418247e-05, "loss": 1.475, "step": 4944 }, { "epoch": 0.0, "learning_rate": 4.999710348489285e-05, "loss": 0.8913, "step": 4945 }, { "epoch": 0.0, "learning_rate": 4.9997102289120685e-05, "loss": 1.3088, "step": 4946 }, { "epoch": 0.0, "learning_rate": 4.999710109310176e-05, "loss": 1.3785, "step": 4947 }, { "epoch": 0.0, "learning_rate": 4.999709989683607e-05, "loss": 0.842, "step": 4948 }, { "epoch": 0.0, "learning_rate": 4.999709870032363e-05, "loss": 1.0386, "step": 4949 }, { "epoch": 0.0, "learning_rate": 4.999709750356443e-05, "loss": 1.2069, "step": 4950 }, { "epoch": 0.0, "learning_rate": 4.999709630655846e-05, "loss": 1.0357, "step": 4951 }, { "epoch": 0.0, "learning_rate": 4.9997095109305735e-05, "loss": 0.5687, "step": 4952 }, { "epoch": 0.0, "learning_rate": 4.9997093911806245e-05, "loss": 0.929, "step": 4953 }, { "epoch": 0.0, "learning_rate": 4.999709271406e-05, "loss": 1.0482, "step": 4954 }, { "epoch": 0.0, "learning_rate": 4.9997091516066985e-05, "loss": 1.1471, "step": 4955 }, { "epoch": 0.0, "learning_rate": 4.9997090317827215e-05, "loss": 1.1224, "step": 4956 }, { "epoch": 0.0, "learning_rate": 4.999708911934069e-05, "loss": 1.1038, "step": 4957 }, { "epoch": 0.0, "learning_rate": 4.9997087920607395e-05, "loss": 1.1545, "step": 4958 }, { "epoch": 0.0, "learning_rate": 4.999708672162734e-05, "loss": 1.0817, "step": 4959 }, { "epoch": 0.0, "learning_rate": 4.9997085522400525e-05, "loss": 0.8527, "step": 4960 }, { "epoch": 0.0, "learning_rate": 4.999708432292696e-05, "loss": 1.2293, "step": 4961 }, { "epoch": 0.0, "learning_rate": 4.9997083123206624e-05, "loss": 1.2644, "step": 4962 }, { "epoch": 0.0, "learning_rate": 4.999708192323952e-05, "loss": 1.1254, "step": 4963 }, { "epoch": 0.0, "learning_rate": 4.999708072302567e-05, "loss": 1.0951, "step": 4964 }, { "epoch": 0.0, "learning_rate": 4.999707952256505e-05, "loss": 0.8272, "step": 4965 }, { "epoch": 0.0, "learning_rate": 4.999707832185767e-05, "loss": 1.4679, "step": 4966 }, { "epoch": 0.0, "learning_rate": 4.999707712090354e-05, "loss": 1.8612, "step": 4967 }, { "epoch": 0.0, "learning_rate": 4.999707591970264e-05, "loss": 1.3245, "step": 4968 }, { "epoch": 0.0, "learning_rate": 4.9997074718254976e-05, "loss": 1.3916, "step": 4969 }, { "epoch": 0.0, "learning_rate": 4.999707351656056e-05, "loss": 1.0122, "step": 4970 }, { "epoch": 0.0, "learning_rate": 4.9997072314619376e-05, "loss": 0.8691, "step": 4971 }, { "epoch": 0.0, "learning_rate": 4.999707111243144e-05, "loss": 1.1927, "step": 4972 }, { "epoch": 0.0, "learning_rate": 4.999706990999674e-05, "loss": 0.959, "step": 4973 }, { "epoch": 0.0, "learning_rate": 4.999706870731528e-05, "loss": 1.1453, "step": 4974 }, { "epoch": 0.0, "learning_rate": 4.999706750438705e-05, "loss": 1.2579, "step": 4975 }, { "epoch": 0.0, "learning_rate": 4.999706630121207e-05, "loss": 1.1374, "step": 4976 }, { "epoch": 0.0, "learning_rate": 4.999706509779033e-05, "loss": 1.4763, "step": 4977 }, { "epoch": 0.0, "learning_rate": 4.999706389412182e-05, "loss": 1.2916, "step": 4978 }, { "epoch": 0.0, "learning_rate": 4.999706269020656e-05, "loss": 1.1468, "step": 4979 }, { "epoch": 0.0, "learning_rate": 4.999706148604453e-05, "loss": 0.9316, "step": 4980 }, { "epoch": 0.0, "learning_rate": 4.999706028163575e-05, "loss": 0.7751, "step": 4981 }, { "epoch": 0.0, "learning_rate": 4.99970590769802e-05, "loss": 1.495, "step": 4982 }, { "epoch": 0.0, "learning_rate": 4.9997057872077894e-05, "loss": 1.0799, "step": 4983 }, { "epoch": 0.0, "learning_rate": 4.999705666692883e-05, "loss": 1.0632, "step": 4984 }, { "epoch": 0.0, "learning_rate": 4.9997055461533006e-05, "loss": 1.1453, "step": 4985 }, { "epoch": 0.0, "learning_rate": 4.999705425589042e-05, "loss": 0.639, "step": 4986 }, { "epoch": 0.0, "learning_rate": 4.999705305000106e-05, "loss": 0.8285, "step": 4987 }, { "epoch": 0.0, "learning_rate": 4.9997051843864965e-05, "loss": 1.1244, "step": 4988 }, { "epoch": 0.0, "learning_rate": 4.999705063748209e-05, "loss": 0.9547, "step": 4989 }, { "epoch": 0.0, "learning_rate": 4.999704943085246e-05, "loss": 0.4331, "step": 4990 }, { "epoch": 0.0, "learning_rate": 4.999704822397607e-05, "loss": 0.1439, "step": 4991 }, { "epoch": 0.0, "learning_rate": 4.9997047016852924e-05, "loss": 1.0579, "step": 4992 }, { "epoch": 0.0, "learning_rate": 4.9997045809483013e-05, "loss": 1.2167, "step": 4993 }, { "epoch": 0.0, "learning_rate": 4.9997044601866345e-05, "loss": 1.0966, "step": 4994 }, { "epoch": 0.0, "learning_rate": 4.999704339400291e-05, "loss": 1.6641, "step": 4995 }, { "epoch": 0.0, "learning_rate": 4.999704218589272e-05, "loss": 1.1495, "step": 4996 }, { "epoch": 0.0, "learning_rate": 4.999704097753577e-05, "loss": 1.1242, "step": 4997 }, { "epoch": 0.0, "learning_rate": 4.999703976893207e-05, "loss": 1.1382, "step": 4998 }, { "epoch": 0.0, "learning_rate": 4.999703856008159e-05, "loss": 1.2479, "step": 4999 }, { "epoch": 0.01, "learning_rate": 4.9997037350984355e-05, "loss": 1.1091, "step": 5000 }, { "epoch": 0.01, "eval_loss": 1.0793377161026, "eval_runtime": 84.714, "eval_samples_per_second": 16.349, "eval_steps_per_second": 4.096, "step": 5000 }, { "epoch": 0.01, "learning_rate": 4.999703614164036e-05, "loss": 1.0244, "step": 5001 }, { "epoch": 0.01, "learning_rate": 4.999703493204961e-05, "loss": 1.2538, "step": 5002 }, { "epoch": 0.01, "learning_rate": 4.99970337222121e-05, "loss": 0.9095, "step": 5003 }, { "epoch": 0.01, "learning_rate": 4.9997032512127826e-05, "loss": 1.175, "step": 5004 }, { "epoch": 0.01, "learning_rate": 4.999703130179679e-05, "loss": 0.9529, "step": 5005 }, { "epoch": 0.01, "learning_rate": 4.9997030091219e-05, "loss": 1.0424, "step": 5006 }, { "epoch": 0.01, "learning_rate": 4.999702888039445e-05, "loss": 0.7228, "step": 5007 }, { "epoch": 0.01, "learning_rate": 4.999702766932314e-05, "loss": 1.1532, "step": 5008 }, { "epoch": 0.01, "learning_rate": 4.999702645800506e-05, "loss": 0.8094, "step": 5009 }, { "epoch": 0.01, "learning_rate": 4.9997025246440236e-05, "loss": 1.148, "step": 5010 }, { "epoch": 0.01, "learning_rate": 4.999702403462864e-05, "loss": 1.2554, "step": 5011 }, { "epoch": 0.01, "learning_rate": 4.9997022822570286e-05, "loss": 1.2795, "step": 5012 }, { "epoch": 0.01, "learning_rate": 4.999702161026517e-05, "loss": 1.2422, "step": 5013 }, { "epoch": 0.01, "learning_rate": 4.999702039771329e-05, "loss": 1.2922, "step": 5014 }, { "epoch": 0.01, "learning_rate": 4.999701918491466e-05, "loss": 1.2931, "step": 5015 }, { "epoch": 0.01, "learning_rate": 4.999701797186926e-05, "loss": 1.2758, "step": 5016 }, { "epoch": 0.01, "learning_rate": 4.999701675857711e-05, "loss": 1.284, "step": 5017 }, { "epoch": 0.01, "learning_rate": 4.99970155450382e-05, "loss": 1.0614, "step": 5018 }, { "epoch": 0.01, "learning_rate": 4.999701433125252e-05, "loss": 1.303, "step": 5019 }, { "epoch": 0.01, "learning_rate": 4.9997013117220095e-05, "loss": 1.4622, "step": 5020 }, { "epoch": 0.01, "learning_rate": 4.99970119029409e-05, "loss": 1.2645, "step": 5021 }, { "epoch": 0.01, "learning_rate": 4.9997010688414945e-05, "loss": 1.2847, "step": 5022 }, { "epoch": 0.01, "learning_rate": 4.999700947364223e-05, "loss": 1.1669, "step": 5023 }, { "epoch": 0.01, "learning_rate": 4.999700825862276e-05, "loss": 0.8389, "step": 5024 }, { "epoch": 0.01, "learning_rate": 4.999700704335653e-05, "loss": 1.2053, "step": 5025 }, { "epoch": 0.01, "learning_rate": 4.999700582784353e-05, "loss": 0.8173, "step": 5026 }, { "epoch": 0.01, "learning_rate": 4.9997004612083775e-05, "loss": 0.9234, "step": 5027 }, { "epoch": 0.01, "learning_rate": 4.9997003396077266e-05, "loss": 1.0486, "step": 5028 }, { "epoch": 0.01, "learning_rate": 4.999700217982399e-05, "loss": 1.0777, "step": 5029 }, { "epoch": 0.01, "learning_rate": 4.999700096332396e-05, "loss": 1.1065, "step": 5030 }, { "epoch": 0.01, "learning_rate": 4.9996999746577164e-05, "loss": 0.9859, "step": 5031 }, { "epoch": 0.01, "learning_rate": 4.999699852958361e-05, "loss": 1.3586, "step": 5032 }, { "epoch": 0.01, "learning_rate": 4.99969973123433e-05, "loss": 1.192, "step": 5033 }, { "epoch": 0.01, "learning_rate": 4.999699609485622e-05, "loss": 0.9577, "step": 5034 }, { "epoch": 0.01, "learning_rate": 4.9996994877122396e-05, "loss": 0.8894, "step": 5035 }, { "epoch": 0.01, "learning_rate": 4.99969936591418e-05, "loss": 0.8399, "step": 5036 }, { "epoch": 0.01, "learning_rate": 4.999699244091445e-05, "loss": 0.7582, "step": 5037 }, { "epoch": 0.01, "learning_rate": 4.999699122244034e-05, "loss": 1.0065, "step": 5038 }, { "epoch": 0.01, "learning_rate": 4.9996990003719466e-05, "loss": 0.5715, "step": 5039 }, { "epoch": 0.01, "learning_rate": 4.999698878475184e-05, "loss": 0.5834, "step": 5040 }, { "epoch": 0.01, "learning_rate": 4.9996987565537445e-05, "loss": 0.8216, "step": 5041 }, { "epoch": 0.01, "learning_rate": 4.9996986346076294e-05, "loss": 1.2597, "step": 5042 }, { "epoch": 0.01, "learning_rate": 4.9996985126368386e-05, "loss": 1.2998, "step": 5043 }, { "epoch": 0.01, "learning_rate": 4.9996983906413714e-05, "loss": 1.0993, "step": 5044 }, { "epoch": 0.01, "learning_rate": 4.9996982686212284e-05, "loss": 0.9802, "step": 5045 }, { "epoch": 0.01, "learning_rate": 4.999698146576409e-05, "loss": 1.2207, "step": 5046 }, { "epoch": 0.01, "learning_rate": 4.999698024506915e-05, "loss": 0.8932, "step": 5047 }, { "epoch": 0.01, "learning_rate": 4.9996979024127434e-05, "loss": 0.8309, "step": 5048 }, { "epoch": 0.01, "learning_rate": 4.999697780293897e-05, "loss": 0.8186, "step": 5049 }, { "epoch": 0.01, "learning_rate": 4.999697658150374e-05, "loss": 1.1159, "step": 5050 }, { "epoch": 0.01, "learning_rate": 4.999697535982175e-05, "loss": 1.0577, "step": 5051 }, { "epoch": 0.01, "learning_rate": 4.9996974137893005e-05, "loss": 0.9394, "step": 5052 }, { "epoch": 0.01, "learning_rate": 4.99969729157175e-05, "loss": 1.1254, "step": 5053 }, { "epoch": 0.01, "learning_rate": 4.999697169329523e-05, "loss": 1.2276, "step": 5054 }, { "epoch": 0.01, "learning_rate": 4.9996970470626204e-05, "loss": 0.9111, "step": 5055 }, { "epoch": 0.01, "learning_rate": 4.9996969247710426e-05, "loss": 0.6828, "step": 5056 }, { "epoch": 0.01, "learning_rate": 4.999696802454788e-05, "loss": 0.4971, "step": 5057 }, { "epoch": 0.01, "learning_rate": 4.999696680113857e-05, "loss": 0.3282, "step": 5058 }, { "epoch": 0.01, "learning_rate": 4.9996965577482506e-05, "loss": 0.7744, "step": 5059 }, { "epoch": 0.01, "learning_rate": 4.999696435357968e-05, "loss": 1.1189, "step": 5060 }, { "epoch": 0.01, "learning_rate": 4.99969631294301e-05, "loss": 0.795, "step": 5061 }, { "epoch": 0.01, "learning_rate": 4.9996961905033755e-05, "loss": 1.2361, "step": 5062 }, { "epoch": 0.01, "learning_rate": 4.999696068039065e-05, "loss": 1.1867, "step": 5063 }, { "epoch": 0.01, "learning_rate": 4.9996959455500794e-05, "loss": 1.2989, "step": 5064 }, { "epoch": 0.01, "learning_rate": 4.999695823036417e-05, "loss": 1.2148, "step": 5065 }, { "epoch": 0.01, "learning_rate": 4.9996957004980796e-05, "loss": 0.9647, "step": 5066 }, { "epoch": 0.01, "learning_rate": 4.999695577935065e-05, "loss": 1.4714, "step": 5067 }, { "epoch": 0.01, "learning_rate": 4.9996954553473754e-05, "loss": 1.3342, "step": 5068 }, { "epoch": 0.01, "learning_rate": 4.999695332735009e-05, "loss": 1.1808, "step": 5069 }, { "epoch": 0.01, "learning_rate": 4.9996952100979674e-05, "loss": 1.1745, "step": 5070 }, { "epoch": 0.01, "learning_rate": 4.99969508743625e-05, "loss": 1.0786, "step": 5071 }, { "epoch": 0.01, "learning_rate": 4.999694964749856e-05, "loss": 1.0861, "step": 5072 }, { "epoch": 0.01, "learning_rate": 4.9996948420387866e-05, "loss": 1.1887, "step": 5073 }, { "epoch": 0.01, "learning_rate": 4.999694719303041e-05, "loss": 1.0008, "step": 5074 }, { "epoch": 0.01, "learning_rate": 4.99969459654262e-05, "loss": 1.1709, "step": 5075 }, { "epoch": 0.01, "learning_rate": 4.999694473757522e-05, "loss": 1.1251, "step": 5076 }, { "epoch": 0.01, "learning_rate": 4.999694350947749e-05, "loss": 1.036, "step": 5077 }, { "epoch": 0.01, "learning_rate": 4.9996942281133e-05, "loss": 1.4192, "step": 5078 }, { "epoch": 0.01, "learning_rate": 4.999694105254174e-05, "loss": 0.9987, "step": 5079 }, { "epoch": 0.01, "learning_rate": 4.999693982370373e-05, "loss": 1.015, "step": 5080 }, { "epoch": 0.01, "learning_rate": 4.999693859461896e-05, "loss": 0.9053, "step": 5081 }, { "epoch": 0.01, "learning_rate": 4.9996937365287434e-05, "loss": 1.1885, "step": 5082 }, { "epoch": 0.01, "learning_rate": 4.999693613570914e-05, "loss": 1.2264, "step": 5083 }, { "epoch": 0.01, "learning_rate": 4.9996934905884094e-05, "loss": 1.0292, "step": 5084 }, { "epoch": 0.01, "learning_rate": 4.9996933675812284e-05, "loss": 0.6453, "step": 5085 }, { "epoch": 0.01, "learning_rate": 4.9996932445493716e-05, "loss": 1.0079, "step": 5086 }, { "epoch": 0.01, "learning_rate": 4.999693121492839e-05, "loss": 1.0922, "step": 5087 }, { "epoch": 0.01, "learning_rate": 4.99969299841163e-05, "loss": 1.1088, "step": 5088 }, { "epoch": 0.01, "learning_rate": 4.999692875305746e-05, "loss": 1.046, "step": 5089 }, { "epoch": 0.01, "learning_rate": 4.9996927521751854e-05, "loss": 1.7119, "step": 5090 }, { "epoch": 0.01, "learning_rate": 4.99969262901995e-05, "loss": 1.4077, "step": 5091 }, { "epoch": 0.01, "learning_rate": 4.999692505840038e-05, "loss": 1.9513, "step": 5092 }, { "epoch": 0.01, "learning_rate": 4.999692382635449e-05, "loss": 1.406, "step": 5093 }, { "epoch": 0.01, "learning_rate": 4.999692259406186e-05, "loss": 1.2516, "step": 5094 }, { "epoch": 0.01, "learning_rate": 4.999692136152245e-05, "loss": 1.1783, "step": 5095 }, { "epoch": 0.01, "learning_rate": 4.99969201287363e-05, "loss": 0.7014, "step": 5096 }, { "epoch": 0.01, "learning_rate": 4.9996918895703384e-05, "loss": 0.8977, "step": 5097 }, { "epoch": 0.01, "learning_rate": 4.9996917662423704e-05, "loss": 0.8868, "step": 5098 }, { "epoch": 0.01, "learning_rate": 4.999691642889727e-05, "loss": 1.018, "step": 5099 }, { "epoch": 0.01, "learning_rate": 4.999691519512407e-05, "loss": 1.0971, "step": 5100 }, { "epoch": 0.01, "learning_rate": 4.999691396110412e-05, "loss": 1.1603, "step": 5101 }, { "epoch": 0.01, "learning_rate": 4.999691272683741e-05, "loss": 0.1859, "step": 5102 }, { "epoch": 0.01, "learning_rate": 4.999691149232394e-05, "loss": 0.1624, "step": 5103 }, { "epoch": 0.01, "learning_rate": 4.9996910257563715e-05, "loss": 0.1466, "step": 5104 }, { "epoch": 0.01, "learning_rate": 4.999690902255672e-05, "loss": 0.1275, "step": 5105 }, { "epoch": 0.01, "learning_rate": 4.999690778730297e-05, "loss": 0.1011, "step": 5106 }, { "epoch": 0.01, "learning_rate": 4.9996906551802465e-05, "loss": 0.0959, "step": 5107 }, { "epoch": 0.01, "learning_rate": 4.9996905316055195e-05, "loss": 0.2996, "step": 5108 }, { "epoch": 0.01, "learning_rate": 4.9996904080061175e-05, "loss": 1.068, "step": 5109 }, { "epoch": 0.01, "learning_rate": 4.999690284382039e-05, "loss": 1.4908, "step": 5110 }, { "epoch": 0.01, "learning_rate": 4.999690160733285e-05, "loss": 1.1142, "step": 5111 }, { "epoch": 0.01, "learning_rate": 4.999690037059854e-05, "loss": 1.3284, "step": 5112 }, { "epoch": 0.01, "learning_rate": 4.999689913361749e-05, "loss": 1.0132, "step": 5113 }, { "epoch": 0.01, "learning_rate": 4.9996897896389666e-05, "loss": 1.1871, "step": 5114 }, { "epoch": 0.01, "learning_rate": 4.9996896658915086e-05, "loss": 0.8212, "step": 5115 }, { "epoch": 0.01, "learning_rate": 4.999689542119375e-05, "loss": 0.6875, "step": 5116 }, { "epoch": 0.01, "learning_rate": 4.9996894183225653e-05, "loss": 1.264, "step": 5117 }, { "epoch": 0.01, "learning_rate": 4.99968929450108e-05, "loss": 1.0435, "step": 5118 }, { "epoch": 0.01, "learning_rate": 4.999689170654919e-05, "loss": 1.2933, "step": 5119 }, { "epoch": 0.01, "learning_rate": 4.9996890467840815e-05, "loss": 1.3859, "step": 5120 }, { "epoch": 0.01, "learning_rate": 4.999688922888568e-05, "loss": 0.7856, "step": 5121 }, { "epoch": 0.01, "learning_rate": 4.999688798968379e-05, "loss": 0.5347, "step": 5122 }, { "epoch": 0.01, "learning_rate": 4.999688675023515e-05, "loss": 0.4216, "step": 5123 }, { "epoch": 0.01, "learning_rate": 4.999688551053974e-05, "loss": 1.0338, "step": 5124 }, { "epoch": 0.01, "learning_rate": 4.9996884270597576e-05, "loss": 1.2116, "step": 5125 }, { "epoch": 0.01, "learning_rate": 4.999688303040865e-05, "loss": 1.1681, "step": 5126 }, { "epoch": 0.01, "learning_rate": 4.999688178997296e-05, "loss": 1.3969, "step": 5127 }, { "epoch": 0.01, "learning_rate": 4.999688054929053e-05, "loss": 1.1493, "step": 5128 }, { "epoch": 0.01, "learning_rate": 4.999687930836132e-05, "loss": 1.4235, "step": 5129 }, { "epoch": 0.01, "learning_rate": 4.999687806718536e-05, "loss": 0.9103, "step": 5130 }, { "epoch": 0.01, "learning_rate": 4.9996876825762646e-05, "loss": 0.8201, "step": 5131 }, { "epoch": 0.01, "learning_rate": 4.999687558409317e-05, "loss": 1.1494, "step": 5132 }, { "epoch": 0.01, "learning_rate": 4.9996874342176934e-05, "loss": 1.1671, "step": 5133 }, { "epoch": 0.01, "learning_rate": 4.999687310001394e-05, "loss": 1.1026, "step": 5134 }, { "epoch": 0.01, "learning_rate": 4.9996871857604186e-05, "loss": 1.0921, "step": 5135 }, { "epoch": 0.01, "learning_rate": 4.9996870614947675e-05, "loss": 1.3104, "step": 5136 }, { "epoch": 0.01, "learning_rate": 4.999686937204441e-05, "loss": 1.1806, "step": 5137 }, { "epoch": 0.01, "learning_rate": 4.999686812889438e-05, "loss": 1.0932, "step": 5138 }, { "epoch": 0.01, "learning_rate": 4.99968668854976e-05, "loss": 0.892, "step": 5139 }, { "epoch": 0.01, "learning_rate": 4.999686564185405e-05, "loss": 0.6406, "step": 5140 }, { "epoch": 0.01, "learning_rate": 4.999686439796374e-05, "loss": 0.4112, "step": 5141 }, { "epoch": 0.01, "learning_rate": 4.999686315382668e-05, "loss": 0.5258, "step": 5142 }, { "epoch": 0.01, "learning_rate": 4.999686190944286e-05, "loss": 0.7285, "step": 5143 }, { "epoch": 0.01, "learning_rate": 4.999686066481229e-05, "loss": 1.1035, "step": 5144 }, { "epoch": 0.01, "learning_rate": 4.9996859419934944e-05, "loss": 0.8942, "step": 5145 }, { "epoch": 0.01, "learning_rate": 4.9996858174810843e-05, "loss": 1.0368, "step": 5146 }, { "epoch": 0.01, "learning_rate": 4.999685692943999e-05, "loss": 0.9548, "step": 5147 }, { "epoch": 0.01, "learning_rate": 4.999685568382238e-05, "loss": 1.049, "step": 5148 }, { "epoch": 0.01, "learning_rate": 4.999685443795801e-05, "loss": 0.7284, "step": 5149 }, { "epoch": 0.01, "learning_rate": 4.9996853191846885e-05, "loss": 1.198, "step": 5150 }, { "epoch": 0.01, "learning_rate": 4.999685194548899e-05, "loss": 1.3568, "step": 5151 }, { "epoch": 0.01, "learning_rate": 4.999685069888434e-05, "loss": 1.1284, "step": 5152 }, { "epoch": 0.01, "learning_rate": 4.999684945203293e-05, "loss": 0.92, "step": 5153 }, { "epoch": 0.01, "learning_rate": 4.999684820493478e-05, "loss": 1.5527, "step": 5154 }, { "epoch": 0.01, "learning_rate": 4.999684695758986e-05, "loss": 1.4233, "step": 5155 }, { "epoch": 0.01, "learning_rate": 4.9996845709998173e-05, "loss": 1.3858, "step": 5156 }, { "epoch": 0.01, "learning_rate": 4.999684446215973e-05, "loss": 0.9554, "step": 5157 }, { "epoch": 0.01, "learning_rate": 4.999684321407454e-05, "loss": 1.1443, "step": 5158 }, { "epoch": 0.01, "learning_rate": 4.999684196574258e-05, "loss": 1.0987, "step": 5159 }, { "epoch": 0.01, "learning_rate": 4.9996840717163875e-05, "loss": 0.961, "step": 5160 }, { "epoch": 0.01, "learning_rate": 4.9996839468338396e-05, "loss": 0.8515, "step": 5161 }, { "epoch": 0.01, "learning_rate": 4.999683821926617e-05, "loss": 1.2304, "step": 5162 }, { "epoch": 0.01, "learning_rate": 4.999683696994718e-05, "loss": 1.3475, "step": 5163 }, { "epoch": 0.01, "learning_rate": 4.999683572038143e-05, "loss": 1.1656, "step": 5164 }, { "epoch": 0.01, "learning_rate": 4.999683447056893e-05, "loss": 1.0646, "step": 5165 }, { "epoch": 0.01, "learning_rate": 4.9996833220509665e-05, "loss": 1.1677, "step": 5166 }, { "epoch": 0.01, "learning_rate": 4.999683197020365e-05, "loss": 1.1873, "step": 5167 }, { "epoch": 0.01, "learning_rate": 4.9996830719650865e-05, "loss": 1.361, "step": 5168 }, { "epoch": 0.01, "learning_rate": 4.9996829468851325e-05, "loss": 1.0357, "step": 5169 }, { "epoch": 0.01, "learning_rate": 4.9996828217805034e-05, "loss": 1.2511, "step": 5170 }, { "epoch": 0.01, "learning_rate": 4.999682696651198e-05, "loss": 1.0471, "step": 5171 }, { "epoch": 0.01, "learning_rate": 4.9996825714972166e-05, "loss": 1.0212, "step": 5172 }, { "epoch": 0.01, "learning_rate": 4.9996824463185596e-05, "loss": 0.9291, "step": 5173 }, { "epoch": 0.01, "learning_rate": 4.999682321115227e-05, "loss": 0.6947, "step": 5174 }, { "epoch": 0.01, "learning_rate": 4.999682195887219e-05, "loss": 0.5708, "step": 5175 }, { "epoch": 0.01, "learning_rate": 4.999682070634534e-05, "loss": 0.3236, "step": 5176 }, { "epoch": 0.01, "learning_rate": 4.999681945357174e-05, "loss": 0.3532, "step": 5177 }, { "epoch": 0.01, "learning_rate": 4.999681820055138e-05, "loss": 0.3641, "step": 5178 }, { "epoch": 0.01, "learning_rate": 4.9996816947284256e-05, "loss": 0.749, "step": 5179 }, { "epoch": 0.01, "learning_rate": 4.999681569377038e-05, "loss": 1.5034, "step": 5180 }, { "epoch": 0.01, "learning_rate": 4.9996814440009744e-05, "loss": 1.1136, "step": 5181 }, { "epoch": 0.01, "learning_rate": 4.9996813186002355e-05, "loss": 0.9345, "step": 5182 }, { "epoch": 0.01, "learning_rate": 4.99968119317482e-05, "loss": 1.3294, "step": 5183 }, { "epoch": 0.01, "learning_rate": 4.9996810677247296e-05, "loss": 1.0999, "step": 5184 }, { "epoch": 0.01, "learning_rate": 4.999680942249962e-05, "loss": 0.8845, "step": 5185 }, { "epoch": 0.01, "learning_rate": 4.999680816750521e-05, "loss": 1.214, "step": 5186 }, { "epoch": 0.01, "learning_rate": 4.9996806912264024e-05, "loss": 1.164, "step": 5187 }, { "epoch": 0.01, "learning_rate": 4.999680565677608e-05, "loss": 0.9096, "step": 5188 }, { "epoch": 0.01, "learning_rate": 4.999680440104138e-05, "loss": 0.6272, "step": 5189 }, { "epoch": 0.01, "learning_rate": 4.9996803145059925e-05, "loss": 1.0698, "step": 5190 }, { "epoch": 0.01, "learning_rate": 4.999680188883171e-05, "loss": 1.1675, "step": 5191 }, { "epoch": 0.01, "learning_rate": 4.999680063235674e-05, "loss": 1.1066, "step": 5192 }, { "epoch": 0.01, "learning_rate": 4.999679937563501e-05, "loss": 1.0981, "step": 5193 }, { "epoch": 0.01, "learning_rate": 4.999679811866652e-05, "loss": 0.9234, "step": 5194 }, { "epoch": 0.01, "learning_rate": 4.9996796861451275e-05, "loss": 0.9038, "step": 5195 }, { "epoch": 0.01, "learning_rate": 4.999679560398927e-05, "loss": 1.1016, "step": 5196 }, { "epoch": 0.01, "learning_rate": 4.9996794346280505e-05, "loss": 1.4688, "step": 5197 }, { "epoch": 0.01, "learning_rate": 4.999679308832499e-05, "loss": 1.0326, "step": 5198 }, { "epoch": 0.01, "learning_rate": 4.999679183012271e-05, "loss": 0.9382, "step": 5199 }, { "epoch": 0.01, "learning_rate": 4.999679057167368e-05, "loss": 1.2755, "step": 5200 }, { "epoch": 0.01, "learning_rate": 4.999678931297789e-05, "loss": 1.295, "step": 5201 }, { "epoch": 0.01, "learning_rate": 4.999678805403534e-05, "loss": 1.2337, "step": 5202 }, { "epoch": 0.01, "learning_rate": 4.999678679484603e-05, "loss": 1.0238, "step": 5203 }, { "epoch": 0.01, "learning_rate": 4.999678553540996e-05, "loss": 1.218, "step": 5204 }, { "epoch": 0.01, "learning_rate": 4.999678427572714e-05, "loss": 1.5987, "step": 5205 }, { "epoch": 0.01, "learning_rate": 4.999678301579755e-05, "loss": 1.1188, "step": 5206 }, { "epoch": 0.01, "learning_rate": 4.9996781755621216e-05, "loss": 1.1451, "step": 5207 }, { "epoch": 0.01, "learning_rate": 4.999678049519812e-05, "loss": 1.0454, "step": 5208 }, { "epoch": 0.01, "learning_rate": 4.999677923452827e-05, "loss": 1.0136, "step": 5209 }, { "epoch": 0.01, "learning_rate": 4.999677797361165e-05, "loss": 1.1909, "step": 5210 }, { "epoch": 0.01, "learning_rate": 4.9996776712448285e-05, "loss": 0.9326, "step": 5211 }, { "epoch": 0.01, "learning_rate": 4.999677545103816e-05, "loss": 1.2454, "step": 5212 }, { "epoch": 0.01, "learning_rate": 4.999677418938127e-05, "loss": 0.8697, "step": 5213 }, { "epoch": 0.01, "learning_rate": 4.999677292747763e-05, "loss": 1.1012, "step": 5214 }, { "epoch": 0.01, "learning_rate": 4.999677166532723e-05, "loss": 0.8678, "step": 5215 }, { "epoch": 0.01, "learning_rate": 4.999677040293007e-05, "loss": 1.119, "step": 5216 }, { "epoch": 0.01, "learning_rate": 4.9996769140286156e-05, "loss": 1.3689, "step": 5217 }, { "epoch": 0.01, "learning_rate": 4.9996767877395485e-05, "loss": 1.0955, "step": 5218 }, { "epoch": 0.01, "learning_rate": 4.999676661425805e-05, "loss": 1.2735, "step": 5219 }, { "epoch": 0.01, "learning_rate": 4.9996765350873864e-05, "loss": 1.2391, "step": 5220 }, { "epoch": 0.01, "learning_rate": 4.999676408724292e-05, "loss": 0.8565, "step": 5221 }, { "epoch": 0.01, "learning_rate": 4.999676282336522e-05, "loss": 0.9062, "step": 5222 }, { "epoch": 0.01, "learning_rate": 4.999676155924076e-05, "loss": 1.0853, "step": 5223 }, { "epoch": 0.01, "learning_rate": 4.9996760294869536e-05, "loss": 1.4615, "step": 5224 }, { "epoch": 0.01, "learning_rate": 4.999675903025156e-05, "loss": 0.9734, "step": 5225 }, { "epoch": 0.01, "learning_rate": 4.999675776538682e-05, "loss": 1.1386, "step": 5226 }, { "epoch": 0.01, "learning_rate": 4.999675650027534e-05, "loss": 1.1606, "step": 5227 }, { "epoch": 0.01, "learning_rate": 4.999675523491709e-05, "loss": 1.2047, "step": 5228 }, { "epoch": 0.01, "learning_rate": 4.999675396931208e-05, "loss": 1.385, "step": 5229 }, { "epoch": 0.01, "learning_rate": 4.999675270346032e-05, "loss": 1.4627, "step": 5230 }, { "epoch": 0.01, "learning_rate": 4.99967514373618e-05, "loss": 0.9382, "step": 5231 }, { "epoch": 0.01, "learning_rate": 4.999675017101653e-05, "loss": 0.9209, "step": 5232 }, { "epoch": 0.01, "learning_rate": 4.9996748904424484e-05, "loss": 0.8517, "step": 5233 }, { "epoch": 0.01, "learning_rate": 4.99967476375857e-05, "loss": 0.8616, "step": 5234 }, { "epoch": 0.01, "learning_rate": 4.999674637050015e-05, "loss": 1.0328, "step": 5235 }, { "epoch": 0.01, "learning_rate": 4.999674510316784e-05, "loss": 1.0035, "step": 5236 }, { "epoch": 0.01, "learning_rate": 4.999674383558877e-05, "loss": 0.9074, "step": 5237 }, { "epoch": 0.01, "learning_rate": 4.9996742567762956e-05, "loss": 0.9313, "step": 5238 }, { "epoch": 0.01, "learning_rate": 4.9996741299690374e-05, "loss": 1.2877, "step": 5239 }, { "epoch": 0.01, "learning_rate": 4.9996740031371035e-05, "loss": 1.197, "step": 5240 }, { "epoch": 0.01, "learning_rate": 4.999673876280494e-05, "loss": 1.0386, "step": 5241 }, { "epoch": 0.01, "learning_rate": 4.999673749399209e-05, "loss": 1.2689, "step": 5242 }, { "epoch": 0.01, "learning_rate": 4.9996736224932486e-05, "loss": 1.027, "step": 5243 }, { "epoch": 0.01, "learning_rate": 4.999673495562612e-05, "loss": 1.3627, "step": 5244 }, { "epoch": 0.01, "learning_rate": 4.9996733686072996e-05, "loss": 1.1203, "step": 5245 }, { "epoch": 0.01, "learning_rate": 4.999673241627312e-05, "loss": 1.0976, "step": 5246 }, { "epoch": 0.01, "learning_rate": 4.9996731146226476e-05, "loss": 1.2692, "step": 5247 }, { "epoch": 0.01, "learning_rate": 4.999672987593308e-05, "loss": 0.8317, "step": 5248 }, { "epoch": 0.01, "learning_rate": 4.999672860539293e-05, "loss": 0.8718, "step": 5249 }, { "epoch": 0.01, "learning_rate": 4.9996727334606016e-05, "loss": 0.961, "step": 5250 }, { "epoch": 0.01, "learning_rate": 4.999672606357236e-05, "loss": 1.0874, "step": 5251 }, { "epoch": 0.01, "learning_rate": 4.999672479229193e-05, "loss": 1.4313, "step": 5252 }, { "epoch": 0.01, "learning_rate": 4.999672352076475e-05, "loss": 1.0807, "step": 5253 }, { "epoch": 0.01, "learning_rate": 4.999672224899081e-05, "loss": 1.7528, "step": 5254 }, { "epoch": 0.01, "learning_rate": 4.999672097697012e-05, "loss": 1.6995, "step": 5255 }, { "epoch": 0.01, "learning_rate": 4.9996719704702675e-05, "loss": 1.2513, "step": 5256 }, { "epoch": 0.01, "learning_rate": 4.999671843218846e-05, "loss": 1.2308, "step": 5257 }, { "epoch": 0.01, "learning_rate": 4.99967171594275e-05, "loss": 1.3474, "step": 5258 }, { "epoch": 0.01, "learning_rate": 4.999671588641978e-05, "loss": 1.0074, "step": 5259 }, { "epoch": 0.01, "learning_rate": 4.9996714613165294e-05, "loss": 1.315, "step": 5260 }, { "epoch": 0.01, "learning_rate": 4.999671333966406e-05, "loss": 1.0775, "step": 5261 }, { "epoch": 0.01, "learning_rate": 4.9996712065916065e-05, "loss": 1.1714, "step": 5262 }, { "epoch": 0.01, "learning_rate": 4.999671079192131e-05, "loss": 1.1032, "step": 5263 }, { "epoch": 0.01, "learning_rate": 4.9996709517679805e-05, "loss": 0.8256, "step": 5264 }, { "epoch": 0.01, "learning_rate": 4.999670824319154e-05, "loss": 1.2036, "step": 5265 }, { "epoch": 0.01, "learning_rate": 4.999670696845652e-05, "loss": 1.1551, "step": 5266 }, { "epoch": 0.01, "learning_rate": 4.999670569347474e-05, "loss": 1.088, "step": 5267 }, { "epoch": 0.01, "learning_rate": 4.999670441824621e-05, "loss": 1.1427, "step": 5268 }, { "epoch": 0.01, "learning_rate": 4.99967031427709e-05, "loss": 1.334, "step": 5269 }, { "epoch": 0.01, "learning_rate": 4.999670186704886e-05, "loss": 0.8837, "step": 5270 }, { "epoch": 0.01, "learning_rate": 4.999670059108005e-05, "loss": 1.1194, "step": 5271 }, { "epoch": 0.01, "learning_rate": 4.9996699314864494e-05, "loss": 1.1684, "step": 5272 }, { "epoch": 0.01, "learning_rate": 4.9996698038402166e-05, "loss": 0.5408, "step": 5273 }, { "epoch": 0.01, "learning_rate": 4.9996696761693095e-05, "loss": 0.364, "step": 5274 }, { "epoch": 0.01, "learning_rate": 4.999669548473726e-05, "loss": 1.16, "step": 5275 }, { "epoch": 0.01, "learning_rate": 4.9996694207534665e-05, "loss": 0.7768, "step": 5276 }, { "epoch": 0.01, "learning_rate": 4.999669293008532e-05, "loss": 1.2486, "step": 5277 }, { "epoch": 0.01, "learning_rate": 4.999669165238922e-05, "loss": 0.9353, "step": 5278 }, { "epoch": 0.01, "learning_rate": 4.999669037444635e-05, "loss": 1.4989, "step": 5279 }, { "epoch": 0.01, "learning_rate": 4.999668909625674e-05, "loss": 1.5388, "step": 5280 }, { "epoch": 0.01, "learning_rate": 4.9996687817820366e-05, "loss": 1.2112, "step": 5281 }, { "epoch": 0.01, "learning_rate": 4.999668653913723e-05, "loss": 0.9339, "step": 5282 }, { "epoch": 0.01, "learning_rate": 4.999668526020734e-05, "loss": 1.2608, "step": 5283 }, { "epoch": 0.01, "learning_rate": 4.99966839810307e-05, "loss": 1.1584, "step": 5284 }, { "epoch": 0.01, "learning_rate": 4.99966827016073e-05, "loss": 0.8799, "step": 5285 }, { "epoch": 0.01, "learning_rate": 4.999668142193714e-05, "loss": 1.1249, "step": 5286 }, { "epoch": 0.01, "learning_rate": 4.9996680142020226e-05, "loss": 0.9992, "step": 5287 }, { "epoch": 0.01, "learning_rate": 4.999667886185655e-05, "loss": 0.7114, "step": 5288 }, { "epoch": 0.01, "learning_rate": 4.9996677581446125e-05, "loss": 1.2043, "step": 5289 }, { "epoch": 0.01, "learning_rate": 4.999667630078894e-05, "loss": 0.8246, "step": 5290 }, { "epoch": 0.01, "learning_rate": 4.9996675019885e-05, "loss": 0.8577, "step": 5291 }, { "epoch": 0.01, "learning_rate": 4.99966737387343e-05, "loss": 0.9255, "step": 5292 }, { "epoch": 0.01, "learning_rate": 4.999667245733684e-05, "loss": 1.7593, "step": 5293 }, { "epoch": 0.01, "learning_rate": 4.999667117569263e-05, "loss": 1.4936, "step": 5294 }, { "epoch": 0.01, "learning_rate": 4.9996669893801665e-05, "loss": 1.2999, "step": 5295 }, { "epoch": 0.01, "learning_rate": 4.999666861166393e-05, "loss": 1.7748, "step": 5296 }, { "epoch": 0.01, "learning_rate": 4.9996667329279455e-05, "loss": 4.5491, "step": 5297 }, { "epoch": 0.01, "learning_rate": 4.9996666046648214e-05, "loss": 1.2272, "step": 5298 }, { "epoch": 0.01, "learning_rate": 4.9996664763770215e-05, "loss": 1.1773, "step": 5299 }, { "epoch": 0.01, "learning_rate": 4.9996663480645465e-05, "loss": 1.0385, "step": 5300 }, { "epoch": 0.01, "learning_rate": 4.999666219727396e-05, "loss": 0.9458, "step": 5301 }, { "epoch": 0.01, "learning_rate": 4.99966609136557e-05, "loss": 1.2472, "step": 5302 }, { "epoch": 0.01, "learning_rate": 4.999665962979068e-05, "loss": 1.0255, "step": 5303 }, { "epoch": 0.01, "learning_rate": 4.9996658345678896e-05, "loss": 0.7737, "step": 5304 }, { "epoch": 0.01, "learning_rate": 4.9996657061320365e-05, "loss": 1.0313, "step": 5305 }, { "epoch": 0.01, "learning_rate": 4.9996655776715076e-05, "loss": 1.3828, "step": 5306 }, { "epoch": 0.01, "learning_rate": 4.999665449186303e-05, "loss": 0.9499, "step": 5307 }, { "epoch": 0.01, "learning_rate": 4.9996653206764225e-05, "loss": 1.0777, "step": 5308 }, { "epoch": 0.01, "learning_rate": 4.999665192141867e-05, "loss": 1.1835, "step": 5309 }, { "epoch": 0.01, "learning_rate": 4.999665063582635e-05, "loss": 0.9606, "step": 5310 }, { "epoch": 0.01, "learning_rate": 4.999664934998728e-05, "loss": 1.0589, "step": 5311 }, { "epoch": 0.01, "learning_rate": 4.999664806390145e-05, "loss": 1.1687, "step": 5312 }, { "epoch": 0.01, "learning_rate": 4.999664677756887e-05, "loss": 0.9827, "step": 5313 }, { "epoch": 0.01, "learning_rate": 4.9996645490989524e-05, "loss": 0.867, "step": 5314 }, { "epoch": 0.01, "learning_rate": 4.999664420416342e-05, "loss": 1.2698, "step": 5315 }, { "epoch": 0.01, "learning_rate": 4.999664291709057e-05, "loss": 1.3419, "step": 5316 }, { "epoch": 0.01, "learning_rate": 4.9996641629770955e-05, "loss": 1.3358, "step": 5317 }, { "epoch": 0.01, "learning_rate": 4.9996640342204595e-05, "loss": 0.9363, "step": 5318 }, { "epoch": 0.01, "learning_rate": 4.999663905439147e-05, "loss": 0.6667, "step": 5319 }, { "epoch": 0.01, "learning_rate": 4.999663776633159e-05, "loss": 0.7523, "step": 5320 }, { "epoch": 0.01, "learning_rate": 4.9996636478024955e-05, "loss": 1.2398, "step": 5321 }, { "epoch": 0.01, "learning_rate": 4.9996635189471564e-05, "loss": 1.123, "step": 5322 }, { "epoch": 0.01, "learning_rate": 4.9996633900671416e-05, "loss": 1.1395, "step": 5323 }, { "epoch": 0.01, "learning_rate": 4.999663261162451e-05, "loss": 1.1425, "step": 5324 }, { "epoch": 0.01, "learning_rate": 4.999663132233085e-05, "loss": 1.1347, "step": 5325 }, { "epoch": 0.01, "learning_rate": 4.999663003279043e-05, "loss": 0.7952, "step": 5326 }, { "epoch": 0.01, "learning_rate": 4.999662874300326e-05, "loss": 1.2129, "step": 5327 }, { "epoch": 0.01, "learning_rate": 4.999662745296933e-05, "loss": 1.2426, "step": 5328 }, { "epoch": 0.01, "learning_rate": 4.999662616268864e-05, "loss": 1.0447, "step": 5329 }, { "epoch": 0.01, "learning_rate": 4.9996624872161204e-05, "loss": 1.2094, "step": 5330 }, { "epoch": 0.01, "learning_rate": 4.9996623581387e-05, "loss": 1.3758, "step": 5331 }, { "epoch": 0.01, "learning_rate": 4.9996622290366054e-05, "loss": 1.2301, "step": 5332 }, { "epoch": 0.01, "learning_rate": 4.999662099909834e-05, "loss": 1.4833, "step": 5333 }, { "epoch": 0.01, "learning_rate": 4.9996619707583874e-05, "loss": 1.1072, "step": 5334 }, { "epoch": 0.01, "learning_rate": 4.999661841582265e-05, "loss": 1.3784, "step": 5335 }, { "epoch": 0.01, "learning_rate": 4.999661712381467e-05, "loss": 1.0261, "step": 5336 }, { "epoch": 0.01, "learning_rate": 4.999661583155994e-05, "loss": 0.9837, "step": 5337 }, { "epoch": 0.01, "learning_rate": 4.999661453905845e-05, "loss": 1.0831, "step": 5338 }, { "epoch": 0.01, "learning_rate": 4.9996613246310205e-05, "loss": 1.2405, "step": 5339 }, { "epoch": 0.01, "learning_rate": 4.99966119533152e-05, "loss": 0.954, "step": 5340 }, { "epoch": 0.01, "learning_rate": 4.999661066007344e-05, "loss": 1.0908, "step": 5341 }, { "epoch": 0.01, "learning_rate": 4.999660936658492e-05, "loss": 1.2137, "step": 5342 }, { "epoch": 0.01, "learning_rate": 4.999660807284966e-05, "loss": 1.3878, "step": 5343 }, { "epoch": 0.01, "learning_rate": 4.999660677886763e-05, "loss": 0.8494, "step": 5344 }, { "epoch": 0.01, "learning_rate": 4.999660548463885e-05, "loss": 1.1138, "step": 5345 }, { "epoch": 0.01, "learning_rate": 4.999660419016331e-05, "loss": 1.2996, "step": 5346 }, { "epoch": 0.01, "learning_rate": 4.9996602895441016e-05, "loss": 2.2952, "step": 5347 }, { "epoch": 0.01, "learning_rate": 4.999660160047197e-05, "loss": 3.8883, "step": 5348 }, { "epoch": 0.01, "learning_rate": 4.999660030525616e-05, "loss": 1.2231, "step": 5349 }, { "epoch": 0.01, "learning_rate": 4.99965990097936e-05, "loss": 1.2186, "step": 5350 }, { "epoch": 0.01, "learning_rate": 4.999659771408428e-05, "loss": 1.0336, "step": 5351 }, { "epoch": 0.01, "learning_rate": 4.99965964181282e-05, "loss": 1.1406, "step": 5352 }, { "epoch": 0.01, "learning_rate": 4.999659512192537e-05, "loss": 1.2545, "step": 5353 }, { "epoch": 0.01, "learning_rate": 4.999659382547579e-05, "loss": 0.7299, "step": 5354 }, { "epoch": 0.01, "learning_rate": 4.999659252877945e-05, "loss": 0.9641, "step": 5355 }, { "epoch": 0.01, "learning_rate": 4.999659123183636e-05, "loss": 0.8678, "step": 5356 }, { "epoch": 0.01, "learning_rate": 4.9996589934646496e-05, "loss": 1.1279, "step": 5357 }, { "epoch": 0.01, "learning_rate": 4.9996588637209885e-05, "loss": 0.6869, "step": 5358 }, { "epoch": 0.01, "learning_rate": 4.999658733952652e-05, "loss": 0.8084, "step": 5359 }, { "epoch": 0.01, "learning_rate": 4.99965860415964e-05, "loss": 1.1045, "step": 5360 }, { "epoch": 0.01, "learning_rate": 4.9996584743419525e-05, "loss": 1.1457, "step": 5361 }, { "epoch": 0.01, "learning_rate": 4.999658344499589e-05, "loss": 1.0041, "step": 5362 }, { "epoch": 0.01, "learning_rate": 4.999658214632551e-05, "loss": 1.2116, "step": 5363 }, { "epoch": 0.01, "learning_rate": 4.9996580847408366e-05, "loss": 1.0101, "step": 5364 }, { "epoch": 0.01, "learning_rate": 4.9996579548244465e-05, "loss": 0.4227, "step": 5365 }, { "epoch": 0.01, "learning_rate": 4.9996578248833806e-05, "loss": 0.871, "step": 5366 }, { "epoch": 0.01, "learning_rate": 4.99965769491764e-05, "loss": 1.2822, "step": 5367 }, { "epoch": 0.01, "learning_rate": 4.9996575649272235e-05, "loss": 3.2187, "step": 5368 }, { "epoch": 0.01, "learning_rate": 4.999657434912131e-05, "loss": 1.2285, "step": 5369 }, { "epoch": 0.01, "learning_rate": 4.9996573048723633e-05, "loss": 1.0293, "step": 5370 }, { "epoch": 0.01, "learning_rate": 4.99965717480792e-05, "loss": 1.2138, "step": 5371 }, { "epoch": 0.01, "learning_rate": 4.999657044718801e-05, "loss": 1.1882, "step": 5372 }, { "epoch": 0.01, "learning_rate": 4.9996569146050066e-05, "loss": 1.2422, "step": 5373 }, { "epoch": 0.01, "learning_rate": 4.9996567844665366e-05, "loss": 1.2266, "step": 5374 }, { "epoch": 0.01, "learning_rate": 4.999656654303391e-05, "loss": 1.0654, "step": 5375 }, { "epoch": 0.01, "learning_rate": 4.99965652411557e-05, "loss": 1.1164, "step": 5376 }, { "epoch": 0.01, "learning_rate": 4.9996563939030734e-05, "loss": 1.0866, "step": 5377 }, { "epoch": 0.01, "learning_rate": 4.999656263665901e-05, "loss": 1.1725, "step": 5378 }, { "epoch": 0.01, "learning_rate": 4.999656133404053e-05, "loss": 0.8584, "step": 5379 }, { "epoch": 0.01, "learning_rate": 4.99965600311753e-05, "loss": 1.1981, "step": 5380 }, { "epoch": 0.01, "learning_rate": 4.9996558728063314e-05, "loss": 1.1273, "step": 5381 }, { "epoch": 0.01, "learning_rate": 4.999655742470457e-05, "loss": 1.2806, "step": 5382 }, { "epoch": 0.01, "learning_rate": 4.999655612109907e-05, "loss": 1.8106, "step": 5383 }, { "epoch": 0.01, "learning_rate": 4.999655481724681e-05, "loss": 1.6897, "step": 5384 }, { "epoch": 0.01, "learning_rate": 4.9996553513147806e-05, "loss": 1.2792, "step": 5385 }, { "epoch": 0.01, "learning_rate": 4.9996552208802035e-05, "loss": 1.7448, "step": 5386 }, { "epoch": 0.01, "learning_rate": 4.999655090420951e-05, "loss": 1.6907, "step": 5387 }, { "epoch": 0.01, "learning_rate": 4.999654959937023e-05, "loss": 1.7045, "step": 5388 }, { "epoch": 0.01, "learning_rate": 4.999654829428421e-05, "loss": 1.6589, "step": 5389 }, { "epoch": 0.01, "learning_rate": 4.9996546988951414e-05, "loss": 1.6445, "step": 5390 }, { "epoch": 0.01, "learning_rate": 4.9996545683371875e-05, "loss": 1.6535, "step": 5391 }, { "epoch": 0.01, "learning_rate": 4.999654437754557e-05, "loss": 1.6749, "step": 5392 }, { "epoch": 0.01, "learning_rate": 4.9996543071472524e-05, "loss": 1.65, "step": 5393 }, { "epoch": 0.01, "learning_rate": 4.999654176515271e-05, "loss": 1.6437, "step": 5394 }, { "epoch": 0.01, "learning_rate": 4.999654045858615e-05, "loss": 1.6235, "step": 5395 }, { "epoch": 0.01, "learning_rate": 4.999653915177283e-05, "loss": 1.6279, "step": 5396 }, { "epoch": 0.01, "learning_rate": 4.999653784471276e-05, "loss": 1.645, "step": 5397 }, { "epoch": 0.01, "learning_rate": 4.999653653740592e-05, "loss": 1.6092, "step": 5398 }, { "epoch": 0.01, "learning_rate": 4.9996535229852335e-05, "loss": 1.6206, "step": 5399 }, { "epoch": 0.01, "learning_rate": 4.999653392205199e-05, "loss": 1.6062, "step": 5400 }, { "epoch": 0.01, "learning_rate": 4.9996532614004896e-05, "loss": 1.6103, "step": 5401 }, { "epoch": 0.01, "learning_rate": 4.999653130571105e-05, "loss": 1.5852, "step": 5402 }, { "epoch": 0.01, "learning_rate": 4.999652999717044e-05, "loss": 1.5889, "step": 5403 }, { "epoch": 0.01, "learning_rate": 4.999652868838308e-05, "loss": 1.632, "step": 5404 }, { "epoch": 0.01, "learning_rate": 4.999652737934896e-05, "loss": 1.5914, "step": 5405 }, { "epoch": 0.01, "learning_rate": 4.999652607006808e-05, "loss": 1.5206, "step": 5406 }, { "epoch": 0.01, "learning_rate": 4.999652476054046e-05, "loss": 1.5644, "step": 5407 }, { "epoch": 0.01, "learning_rate": 4.999652345076608e-05, "loss": 1.3976, "step": 5408 }, { "epoch": 0.01, "learning_rate": 4.999652214074494e-05, "loss": 1.5316, "step": 5409 }, { "epoch": 0.01, "learning_rate": 4.999652083047705e-05, "loss": 1.5824, "step": 5410 }, { "epoch": 0.01, "learning_rate": 4.9996519519962396e-05, "loss": 1.4277, "step": 5411 }, { "epoch": 0.01, "learning_rate": 4.9996518209200994e-05, "loss": 1.5752, "step": 5412 }, { "epoch": 0.01, "learning_rate": 4.9996516898192835e-05, "loss": 1.5158, "step": 5413 }, { "epoch": 0.01, "learning_rate": 4.999651558693792e-05, "loss": 1.5819, "step": 5414 }, { "epoch": 0.01, "learning_rate": 4.9996514275436256e-05, "loss": 1.4186, "step": 5415 }, { "epoch": 0.01, "learning_rate": 4.999651296368783e-05, "loss": 1.3164, "step": 5416 }, { "epoch": 0.01, "learning_rate": 4.999651165169266e-05, "loss": 1.4611, "step": 5417 }, { "epoch": 0.01, "learning_rate": 4.999651033945072e-05, "loss": 1.3652, "step": 5418 }, { "epoch": 0.01, "learning_rate": 4.999650902696203e-05, "loss": 1.2459, "step": 5419 }, { "epoch": 0.01, "learning_rate": 4.9996507714226584e-05, "loss": 1.4631, "step": 5420 }, { "epoch": 0.01, "learning_rate": 4.999650640124439e-05, "loss": 1.4849, "step": 5421 }, { "epoch": 0.01, "learning_rate": 4.999650508801543e-05, "loss": 1.5499, "step": 5422 }, { "epoch": 0.01, "learning_rate": 4.9996503774539724e-05, "loss": 1.5977, "step": 5423 }, { "epoch": 0.01, "learning_rate": 4.9996502460817264e-05, "loss": 1.5931, "step": 5424 }, { "epoch": 0.01, "learning_rate": 4.999650114684804e-05, "loss": 1.5806, "step": 5425 }, { "epoch": 0.01, "learning_rate": 4.999649983263207e-05, "loss": 1.544, "step": 5426 }, { "epoch": 0.01, "learning_rate": 4.9996498518169345e-05, "loss": 1.6057, "step": 5427 }, { "epoch": 0.01, "learning_rate": 4.9996497203459855e-05, "loss": 1.6214, "step": 5428 }, { "epoch": 0.01, "learning_rate": 4.999649588850363e-05, "loss": 1.1424, "step": 5429 }, { "epoch": 0.01, "learning_rate": 4.999649457330063e-05, "loss": 1.5469, "step": 5430 }, { "epoch": 0.01, "learning_rate": 4.9996493257850885e-05, "loss": 1.6572, "step": 5431 }, { "epoch": 0.01, "learning_rate": 4.999649194215438e-05, "loss": 1.6586, "step": 5432 }, { "epoch": 0.01, "learning_rate": 4.999649062621112e-05, "loss": 1.6485, "step": 5433 }, { "epoch": 0.01, "learning_rate": 4.999648931002111e-05, "loss": 1.6322, "step": 5434 }, { "epoch": 0.01, "learning_rate": 4.999648799358434e-05, "loss": 1.6366, "step": 5435 }, { "epoch": 0.01, "learning_rate": 4.999648667690082e-05, "loss": 1.6318, "step": 5436 }, { "epoch": 0.01, "learning_rate": 4.9996485359970544e-05, "loss": 1.3802, "step": 5437 }, { "epoch": 0.01, "learning_rate": 4.999648404279351e-05, "loss": 1.635, "step": 5438 }, { "epoch": 0.01, "learning_rate": 4.999648272536973e-05, "loss": 1.6398, "step": 5439 }, { "epoch": 0.01, "learning_rate": 4.999648140769918e-05, "loss": 1.6355, "step": 5440 }, { "epoch": 0.01, "learning_rate": 4.999648008978189e-05, "loss": 1.455, "step": 5441 }, { "epoch": 0.01, "learning_rate": 4.999647877161784e-05, "loss": 1.6273, "step": 5442 }, { "epoch": 0.01, "learning_rate": 4.999647745320704e-05, "loss": 1.6324, "step": 5443 }, { "epoch": 0.01, "learning_rate": 4.999647613454947e-05, "loss": 1.6379, "step": 5444 }, { "epoch": 0.01, "learning_rate": 4.9996474815645156e-05, "loss": 1.6175, "step": 5445 }, { "epoch": 0.01, "learning_rate": 4.999647349649409e-05, "loss": 1.4542, "step": 5446 }, { "epoch": 0.01, "learning_rate": 4.9996472177096264e-05, "loss": 1.3211, "step": 5447 }, { "epoch": 0.01, "learning_rate": 4.999647085745169e-05, "loss": 1.626, "step": 5448 }, { "epoch": 0.01, "learning_rate": 4.9996469537560356e-05, "loss": 0.9563, "step": 5449 }, { "epoch": 0.01, "learning_rate": 4.9996468217422265e-05, "loss": 1.6406, "step": 5450 }, { "epoch": 0.01, "learning_rate": 4.9996466897037424e-05, "loss": 1.5165, "step": 5451 }, { "epoch": 0.01, "learning_rate": 4.9996465576405825e-05, "loss": 1.6257, "step": 5452 }, { "epoch": 0.01, "learning_rate": 4.9996464255527475e-05, "loss": 1.6199, "step": 5453 }, { "epoch": 0.01, "learning_rate": 4.9996462934402374e-05, "loss": 1.6214, "step": 5454 }, { "epoch": 0.01, "learning_rate": 4.999646161303051e-05, "loss": 1.6068, "step": 5455 }, { "epoch": 0.01, "learning_rate": 4.999646029141189e-05, "loss": 1.6175, "step": 5456 }, { "epoch": 0.01, "learning_rate": 4.9996458969546525e-05, "loss": 1.6042, "step": 5457 }, { "epoch": 0.01, "learning_rate": 4.99964576474344e-05, "loss": 1.3519, "step": 5458 }, { "epoch": 0.01, "learning_rate": 4.9996456325075525e-05, "loss": 1.6099, "step": 5459 }, { "epoch": 0.01, "learning_rate": 4.9996455002469885e-05, "loss": 1.6056, "step": 5460 }, { "epoch": 0.01, "learning_rate": 4.999645367961751e-05, "loss": 1.6091, "step": 5461 }, { "epoch": 0.01, "learning_rate": 4.999645235651836e-05, "loss": 1.6222, "step": 5462 }, { "epoch": 0.01, "learning_rate": 4.999645103317247e-05, "loss": 1.601, "step": 5463 }, { "epoch": 0.01, "learning_rate": 4.999644970957981e-05, "loss": 1.5557, "step": 5464 }, { "epoch": 0.01, "learning_rate": 4.999644838574041e-05, "loss": 1.6104, "step": 5465 }, { "epoch": 0.01, "learning_rate": 4.999644706165425e-05, "loss": 1.616, "step": 5466 }, { "epoch": 0.01, "learning_rate": 4.9996445737321334e-05, "loss": 1.611, "step": 5467 }, { "epoch": 0.01, "learning_rate": 4.999644441274166e-05, "loss": 1.6151, "step": 5468 }, { "epoch": 0.01, "learning_rate": 4.999644308791524e-05, "loss": 1.2763, "step": 5469 }, { "epoch": 0.01, "learning_rate": 4.999644176284206e-05, "loss": 1.4887, "step": 5470 }, { "epoch": 0.01, "learning_rate": 4.999644043752213e-05, "loss": 1.363, "step": 5471 }, { "epoch": 0.01, "learning_rate": 4.999643911195545e-05, "loss": 1.6844, "step": 5472 }, { "epoch": 0.01, "learning_rate": 4.9996437786142e-05, "loss": 1.5473, "step": 5473 }, { "epoch": 0.01, "learning_rate": 4.9996436460081815e-05, "loss": 1.5769, "step": 5474 }, { "epoch": 0.01, "learning_rate": 4.9996435133774865e-05, "loss": 1.3247, "step": 5475 }, { "epoch": 0.01, "learning_rate": 4.999643380722116e-05, "loss": 0.9342, "step": 5476 }, { "epoch": 0.01, "learning_rate": 4.9996432480420705e-05, "loss": 1.0491, "step": 5477 }, { "epoch": 0.01, "learning_rate": 4.999643115337349e-05, "loss": 1.2968, "step": 5478 }, { "epoch": 0.01, "learning_rate": 4.999642982607953e-05, "loss": 1.0755, "step": 5479 }, { "epoch": 0.01, "learning_rate": 4.99964284985388e-05, "loss": 1.312, "step": 5480 }, { "epoch": 0.01, "learning_rate": 4.9996427170751334e-05, "loss": 1.1861, "step": 5481 }, { "epoch": 0.01, "learning_rate": 4.99964258427171e-05, "loss": 1.1286, "step": 5482 }, { "epoch": 0.01, "learning_rate": 4.999642451443612e-05, "loss": 1.3266, "step": 5483 }, { "epoch": 0.01, "learning_rate": 4.999642318590838e-05, "loss": 1.486, "step": 5484 }, { "epoch": 0.01, "learning_rate": 4.999642185713389e-05, "loss": 0.9346, "step": 5485 }, { "epoch": 0.01, "learning_rate": 4.9996420528112645e-05, "loss": 0.8062, "step": 5486 }, { "epoch": 0.01, "learning_rate": 4.9996419198844643e-05, "loss": 1.1346, "step": 5487 }, { "epoch": 0.01, "learning_rate": 4.999641786932989e-05, "loss": 1.3305, "step": 5488 }, { "epoch": 0.01, "learning_rate": 4.999641653956839e-05, "loss": 1.0559, "step": 5489 }, { "epoch": 0.01, "learning_rate": 4.999641520956013e-05, "loss": 0.8833, "step": 5490 }, { "epoch": 0.01, "learning_rate": 4.999641387930511e-05, "loss": 1.0513, "step": 5491 }, { "epoch": 0.01, "learning_rate": 4.999641254880334e-05, "loss": 1.0562, "step": 5492 }, { "epoch": 0.01, "learning_rate": 4.999641121805482e-05, "loss": 1.0458, "step": 5493 }, { "epoch": 0.01, "learning_rate": 4.9996409887059536e-05, "loss": 1.0736, "step": 5494 }, { "epoch": 0.01, "learning_rate": 4.999640855581751e-05, "loss": 1.0123, "step": 5495 }, { "epoch": 0.01, "learning_rate": 4.999640722432872e-05, "loss": 1.218, "step": 5496 }, { "epoch": 0.01, "learning_rate": 4.999640589259318e-05, "loss": 1.1162, "step": 5497 }, { "epoch": 0.01, "learning_rate": 4.999640456061089e-05, "loss": 0.9334, "step": 5498 }, { "epoch": 0.01, "learning_rate": 4.999640322838184e-05, "loss": 1.1247, "step": 5499 }, { "epoch": 0.01, "learning_rate": 4.999640189590604e-05, "loss": 1.2491, "step": 5500 }, { "epoch": 0.01, "eval_loss": 1.0697647333145142, "eval_runtime": 83.3184, "eval_samples_per_second": 16.623, "eval_steps_per_second": 4.165, "step": 5500 }, { "epoch": 0.01, "learning_rate": 4.999640056318349e-05, "loss": 1.0866, "step": 5501 }, { "epoch": 0.01, "learning_rate": 4.9996399230214175e-05, "loss": 1.1621, "step": 5502 }, { "epoch": 0.01, "learning_rate": 4.999639789699811e-05, "loss": 1.9449, "step": 5503 }, { "epoch": 0.01, "learning_rate": 4.999639656353529e-05, "loss": 1.3423, "step": 5504 }, { "epoch": 0.01, "learning_rate": 4.999639522982572e-05, "loss": 1.0964, "step": 5505 }, { "epoch": 0.01, "learning_rate": 4.99963938958694e-05, "loss": 1.1129, "step": 5506 }, { "epoch": 0.01, "learning_rate": 4.9996392561666313e-05, "loss": 1.3408, "step": 5507 }, { "epoch": 0.01, "learning_rate": 4.999639122721649e-05, "loss": 0.9812, "step": 5508 }, { "epoch": 0.01, "learning_rate": 4.99963898925199e-05, "loss": 0.3642, "step": 5509 }, { "epoch": 0.01, "learning_rate": 4.9996388557576556e-05, "loss": 0.8257, "step": 5510 }, { "epoch": 0.01, "learning_rate": 4.9996387222386466e-05, "loss": 1.2286, "step": 5511 }, { "epoch": 0.01, "learning_rate": 4.999638588694961e-05, "loss": 1.333, "step": 5512 }, { "epoch": 0.01, "learning_rate": 4.9996384551266014e-05, "loss": 1.2729, "step": 5513 }, { "epoch": 0.01, "learning_rate": 4.999638321533566e-05, "loss": 1.1928, "step": 5514 }, { "epoch": 0.01, "learning_rate": 4.999638187915855e-05, "loss": 1.3171, "step": 5515 }, { "epoch": 0.01, "learning_rate": 4.999638054273468e-05, "loss": 1.176, "step": 5516 }, { "epoch": 0.01, "learning_rate": 4.999637920606407e-05, "loss": 0.9842, "step": 5517 }, { "epoch": 0.01, "learning_rate": 4.999637786914669e-05, "loss": 1.0741, "step": 5518 }, { "epoch": 0.01, "learning_rate": 4.999637653198257e-05, "loss": 1.2538, "step": 5519 }, { "epoch": 0.01, "learning_rate": 4.9996375194571694e-05, "loss": 0.947, "step": 5520 }, { "epoch": 0.01, "learning_rate": 4.999637385691406e-05, "loss": 3.6488, "step": 5521 }, { "epoch": 0.01, "learning_rate": 4.999637251900968e-05, "loss": 2.0481, "step": 5522 }, { "epoch": 0.01, "learning_rate": 4.999637118085854e-05, "loss": 1.2494, "step": 5523 }, { "epoch": 0.01, "learning_rate": 4.9996369842460646e-05, "loss": 0.9502, "step": 5524 }, { "epoch": 0.01, "learning_rate": 4.9996368503816e-05, "loss": 1.3348, "step": 5525 }, { "epoch": 0.01, "learning_rate": 4.99963671649246e-05, "loss": 1.2725, "step": 5526 }, { "epoch": 0.01, "learning_rate": 4.9996365825786446e-05, "loss": 0.4649, "step": 5527 }, { "epoch": 0.01, "learning_rate": 4.999636448640154e-05, "loss": 0.2174, "step": 5528 }, { "epoch": 0.01, "learning_rate": 4.999636314676988e-05, "loss": 0.1566, "step": 5529 }, { "epoch": 0.01, "learning_rate": 4.999636180689146e-05, "loss": 0.0937, "step": 5530 }, { "epoch": 0.01, "learning_rate": 4.999636046676629e-05, "loss": 0.0804, "step": 5531 }, { "epoch": 0.01, "learning_rate": 4.9996359126394374e-05, "loss": 0.0901, "step": 5532 }, { "epoch": 0.01, "learning_rate": 4.999635778577569e-05, "loss": 0.1146, "step": 5533 }, { "epoch": 0.01, "learning_rate": 4.999635644491027e-05, "loss": 0.0508, "step": 5534 }, { "epoch": 0.01, "learning_rate": 4.9996355103798085e-05, "loss": 0.0421, "step": 5535 }, { "epoch": 0.01, "learning_rate": 4.999635376243915e-05, "loss": 0.0781, "step": 5536 }, { "epoch": 0.01, "learning_rate": 4.9996352420833456e-05, "loss": 0.061, "step": 5537 }, { "epoch": 0.01, "learning_rate": 4.999635107898102e-05, "loss": 0.2096, "step": 5538 }, { "epoch": 0.01, "learning_rate": 4.999634973688182e-05, "loss": 0.1037, "step": 5539 }, { "epoch": 0.01, "learning_rate": 4.999634839453588e-05, "loss": 0.204, "step": 5540 }, { "epoch": 0.01, "learning_rate": 4.9996347051943174e-05, "loss": 0.1553, "step": 5541 }, { "epoch": 0.01, "learning_rate": 4.999634570910372e-05, "loss": 0.2328, "step": 5542 }, { "epoch": 0.01, "learning_rate": 4.999634436601751e-05, "loss": 0.1369, "step": 5543 }, { "epoch": 0.01, "learning_rate": 4.9996343022684544e-05, "loss": 0.1622, "step": 5544 }, { "epoch": 0.01, "learning_rate": 4.999634167910483e-05, "loss": 0.0946, "step": 5545 }, { "epoch": 0.01, "learning_rate": 4.999634033527836e-05, "loss": 0.812, "step": 5546 }, { "epoch": 0.01, "learning_rate": 4.9996338991205136e-05, "loss": 1.375, "step": 5547 }, { "epoch": 0.01, "learning_rate": 4.999633764688516e-05, "loss": 1.1506, "step": 5548 }, { "epoch": 0.01, "learning_rate": 4.9996336302318425e-05, "loss": 1.2242, "step": 5549 }, { "epoch": 0.01, "learning_rate": 4.999633495750494e-05, "loss": 1.2195, "step": 5550 }, { "epoch": 0.01, "learning_rate": 4.999633361244471e-05, "loss": 1.1318, "step": 5551 }, { "epoch": 0.01, "learning_rate": 4.999633226713773e-05, "loss": 1.2764, "step": 5552 }, { "epoch": 0.01, "learning_rate": 4.999633092158398e-05, "loss": 1.1542, "step": 5553 }, { "epoch": 0.01, "learning_rate": 4.999632957578349e-05, "loss": 1.4049, "step": 5554 }, { "epoch": 0.01, "learning_rate": 4.999632822973623e-05, "loss": 1.8587, "step": 5555 }, { "epoch": 0.01, "learning_rate": 4.999632688344223e-05, "loss": 1.7064, "step": 5556 }, { "epoch": 0.01, "learning_rate": 4.999632553690148e-05, "loss": 1.2494, "step": 5557 }, { "epoch": 0.01, "learning_rate": 4.999632419011396e-05, "loss": 1.3254, "step": 5558 }, { "epoch": 0.01, "learning_rate": 4.9996322843079703e-05, "loss": 0.8808, "step": 5559 }, { "epoch": 0.01, "learning_rate": 4.999632149579869e-05, "loss": 1.0223, "step": 5560 }, { "epoch": 0.01, "learning_rate": 4.999632014827092e-05, "loss": 0.8446, "step": 5561 }, { "epoch": 0.01, "learning_rate": 4.99963188004964e-05, "loss": 1.0649, "step": 5562 }, { "epoch": 0.01, "learning_rate": 4.999631745247513e-05, "loss": 0.8702, "step": 5563 }, { "epoch": 0.01, "learning_rate": 4.9996316104207094e-05, "loss": 1.0529, "step": 5564 }, { "epoch": 0.01, "learning_rate": 4.999631475569232e-05, "loss": 1.352, "step": 5565 }, { "epoch": 0.01, "learning_rate": 4.9996313406930786e-05, "loss": 1.5323, "step": 5566 }, { "epoch": 0.01, "learning_rate": 4.999631205792249e-05, "loss": 1.368, "step": 5567 }, { "epoch": 0.01, "learning_rate": 4.9996310708667454e-05, "loss": 1.0888, "step": 5568 }, { "epoch": 0.01, "learning_rate": 4.999630935916566e-05, "loss": 0.9347, "step": 5569 }, { "epoch": 0.01, "learning_rate": 4.999630800941711e-05, "loss": 0.9903, "step": 5570 }, { "epoch": 0.01, "learning_rate": 4.999630665942182e-05, "loss": 1.0515, "step": 5571 }, { "epoch": 0.01, "learning_rate": 4.9996305309179767e-05, "loss": 1.1067, "step": 5572 }, { "epoch": 0.01, "learning_rate": 4.999630395869096e-05, "loss": 1.2955, "step": 5573 }, { "epoch": 0.01, "learning_rate": 4.9996302607955404e-05, "loss": 1.2846, "step": 5574 }, { "epoch": 0.01, "learning_rate": 4.999630125697309e-05, "loss": 1.0916, "step": 5575 }, { "epoch": 0.01, "learning_rate": 4.9996299905744025e-05, "loss": 1.1508, "step": 5576 }, { "epoch": 0.01, "learning_rate": 4.999629855426821e-05, "loss": 1.2063, "step": 5577 }, { "epoch": 0.01, "learning_rate": 4.999629720254564e-05, "loss": 1.0702, "step": 5578 }, { "epoch": 0.01, "learning_rate": 4.999629585057632e-05, "loss": 1.0311, "step": 5579 }, { "epoch": 0.01, "learning_rate": 4.999629449836024e-05, "loss": 1.3127, "step": 5580 }, { "epoch": 0.01, "learning_rate": 4.999629314589741e-05, "loss": 1.3089, "step": 5581 }, { "epoch": 0.01, "learning_rate": 4.999629179318783e-05, "loss": 1.0694, "step": 5582 }, { "epoch": 0.01, "learning_rate": 4.99962904402315e-05, "loss": 1.1529, "step": 5583 }, { "epoch": 0.01, "learning_rate": 4.999628908702841e-05, "loss": 1.1264, "step": 5584 }, { "epoch": 0.01, "learning_rate": 4.9996287733578575e-05, "loss": 1.1999, "step": 5585 }, { "epoch": 0.01, "learning_rate": 4.999628637988198e-05, "loss": 1.1031, "step": 5586 }, { "epoch": 0.01, "learning_rate": 4.9996285025938636e-05, "loss": 0.7861, "step": 5587 }, { "epoch": 0.01, "learning_rate": 4.999628367174853e-05, "loss": 0.4965, "step": 5588 }, { "epoch": 0.01, "learning_rate": 4.999628231731168e-05, "loss": 0.8337, "step": 5589 }, { "epoch": 0.01, "learning_rate": 4.9996280962628075e-05, "loss": 0.8052, "step": 5590 }, { "epoch": 0.01, "learning_rate": 4.999627960769772e-05, "loss": 1.2772, "step": 5591 }, { "epoch": 0.01, "learning_rate": 4.9996278252520614e-05, "loss": 0.9121, "step": 5592 }, { "epoch": 0.01, "learning_rate": 4.999627689709675e-05, "loss": 1.7697, "step": 5593 }, { "epoch": 0.01, "learning_rate": 4.9996275541426136e-05, "loss": 1.6754, "step": 5594 }, { "epoch": 0.01, "learning_rate": 4.999627418550877e-05, "loss": 0.9327, "step": 5595 }, { "epoch": 0.01, "learning_rate": 4.999627282934465e-05, "loss": 0.7492, "step": 5596 }, { "epoch": 0.01, "learning_rate": 4.9996271472933773e-05, "loss": 0.6132, "step": 5597 }, { "epoch": 0.01, "learning_rate": 4.999627011627615e-05, "loss": 0.7812, "step": 5598 }, { "epoch": 0.01, "learning_rate": 4.999626875937177e-05, "loss": 1.0239, "step": 5599 }, { "epoch": 0.01, "learning_rate": 4.999626740222064e-05, "loss": 1.3136, "step": 5600 }, { "epoch": 0.01, "learning_rate": 4.999626604482276e-05, "loss": 1.5179, "step": 5601 }, { "epoch": 0.01, "learning_rate": 4.999626468717812e-05, "loss": 1.271, "step": 5602 }, { "epoch": 0.01, "learning_rate": 4.999626332928673e-05, "loss": 1.0892, "step": 5603 }, { "epoch": 0.01, "learning_rate": 4.99962619711486e-05, "loss": 1.113, "step": 5604 }, { "epoch": 0.01, "learning_rate": 4.9996260612763704e-05, "loss": 1.1502, "step": 5605 }, { "epoch": 0.01, "learning_rate": 4.999625925413205e-05, "loss": 1.1396, "step": 5606 }, { "epoch": 0.01, "learning_rate": 4.999625789525365e-05, "loss": 1.56, "step": 5607 }, { "epoch": 0.01, "learning_rate": 4.99962565361285e-05, "loss": 1.2087, "step": 5608 }, { "epoch": 0.01, "learning_rate": 4.99962551767566e-05, "loss": 1.268, "step": 5609 }, { "epoch": 0.01, "learning_rate": 4.999625381713794e-05, "loss": 1.2953, "step": 5610 }, { "epoch": 0.01, "learning_rate": 4.999625245727253e-05, "loss": 1.222, "step": 5611 }, { "epoch": 0.01, "learning_rate": 4.999625109716037e-05, "loss": 1.1168, "step": 5612 }, { "epoch": 0.01, "learning_rate": 4.999624973680146e-05, "loss": 1.1631, "step": 5613 }, { "epoch": 0.01, "learning_rate": 4.9996248376195784e-05, "loss": 1.4415, "step": 5614 }, { "epoch": 0.01, "learning_rate": 4.999624701534337e-05, "loss": 1.1527, "step": 5615 }, { "epoch": 0.01, "learning_rate": 4.99962456542442e-05, "loss": 1.1625, "step": 5616 }, { "epoch": 0.01, "learning_rate": 4.9996244292898275e-05, "loss": 1.3437, "step": 5617 }, { "epoch": 0.01, "learning_rate": 4.9996242931305606e-05, "loss": 0.7632, "step": 5618 }, { "epoch": 0.01, "learning_rate": 4.999624156946617e-05, "loss": 1.2084, "step": 5619 }, { "epoch": 0.01, "learning_rate": 4.999624020737999e-05, "loss": 0.9181, "step": 5620 }, { "epoch": 0.01, "learning_rate": 4.999623884504706e-05, "loss": 1.1415, "step": 5621 }, { "epoch": 0.01, "learning_rate": 4.9996237482467374e-05, "loss": 1.2355, "step": 5622 }, { "epoch": 0.01, "learning_rate": 4.9996236119640937e-05, "loss": 1.2088, "step": 5623 }, { "epoch": 0.01, "learning_rate": 4.999623475656774e-05, "loss": 0.5592, "step": 5624 }, { "epoch": 0.01, "learning_rate": 4.99962333932478e-05, "loss": 1.0354, "step": 5625 }, { "epoch": 0.01, "learning_rate": 4.9996232029681114e-05, "loss": 0.9641, "step": 5626 }, { "epoch": 0.01, "learning_rate": 4.999623066586766e-05, "loss": 1.1484, "step": 5627 }, { "epoch": 0.01, "learning_rate": 4.999622930180746e-05, "loss": 1.1756, "step": 5628 }, { "epoch": 0.01, "learning_rate": 4.999622793750051e-05, "loss": 1.022, "step": 5629 }, { "epoch": 0.01, "learning_rate": 4.9996226572946806e-05, "loss": 1.2777, "step": 5630 }, { "epoch": 0.01, "learning_rate": 4.9996225208146355e-05, "loss": 1.4627, "step": 5631 }, { "epoch": 0.01, "learning_rate": 4.999622384309914e-05, "loss": 1.3966, "step": 5632 }, { "epoch": 0.01, "learning_rate": 4.999622247780518e-05, "loss": 1.1712, "step": 5633 }, { "epoch": 0.01, "learning_rate": 4.999622111226447e-05, "loss": 1.1774, "step": 5634 }, { "epoch": 0.01, "learning_rate": 4.999621974647701e-05, "loss": 1.6404, "step": 5635 }, { "epoch": 0.01, "learning_rate": 4.999621838044279e-05, "loss": 1.4061, "step": 5636 }, { "epoch": 0.01, "learning_rate": 4.999621701416182e-05, "loss": 1.254, "step": 5637 }, { "epoch": 0.01, "learning_rate": 4.99962156476341e-05, "loss": 1.154, "step": 5638 }, { "epoch": 0.01, "learning_rate": 4.9996214280859624e-05, "loss": 0.999, "step": 5639 }, { "epoch": 0.01, "learning_rate": 4.99962129138384e-05, "loss": 1.443, "step": 5640 }, { "epoch": 0.01, "learning_rate": 4.999621154657043e-05, "loss": 1.3345, "step": 5641 }, { "epoch": 0.01, "learning_rate": 4.99962101790557e-05, "loss": 1.4813, "step": 5642 }, { "epoch": 0.01, "learning_rate": 4.999620881129422e-05, "loss": 0.4938, "step": 5643 }, { "epoch": 0.01, "learning_rate": 4.999620744328598e-05, "loss": 0.8844, "step": 5644 }, { "epoch": 0.01, "learning_rate": 4.9996206075030996e-05, "loss": 1.2019, "step": 5645 }, { "epoch": 0.01, "learning_rate": 4.999620470652926e-05, "loss": 1.3681, "step": 5646 }, { "epoch": 0.01, "learning_rate": 4.999620333778077e-05, "loss": 1.3369, "step": 5647 }, { "epoch": 0.01, "learning_rate": 4.999620196878553e-05, "loss": 0.9395, "step": 5648 }, { "epoch": 0.01, "learning_rate": 4.999620059954353e-05, "loss": 1.4062, "step": 5649 }, { "epoch": 0.01, "learning_rate": 4.9996199230054784e-05, "loss": 1.3625, "step": 5650 }, { "epoch": 0.01, "learning_rate": 4.9996197860319295e-05, "loss": 1.1258, "step": 5651 }, { "epoch": 0.01, "learning_rate": 4.999619649033704e-05, "loss": 0.9185, "step": 5652 }, { "epoch": 0.01, "learning_rate": 4.999619512010804e-05, "loss": 1.1147, "step": 5653 }, { "epoch": 0.01, "learning_rate": 4.999619374963229e-05, "loss": 1.2958, "step": 5654 }, { "epoch": 0.01, "learning_rate": 4.9996192378909786e-05, "loss": 0.9991, "step": 5655 }, { "epoch": 0.01, "learning_rate": 4.999619100794053e-05, "loss": 0.9079, "step": 5656 }, { "epoch": 0.01, "learning_rate": 4.999618963672452e-05, "loss": 1.1405, "step": 5657 }, { "epoch": 0.01, "learning_rate": 4.999618826526176e-05, "loss": 1.2317, "step": 5658 }, { "epoch": 0.01, "learning_rate": 4.999618689355224e-05, "loss": 1.2066, "step": 5659 }, { "epoch": 0.01, "learning_rate": 4.9996185521595984e-05, "loss": 1.3058, "step": 5660 }, { "epoch": 0.01, "learning_rate": 4.9996184149392967e-05, "loss": 0.8117, "step": 5661 }, { "epoch": 0.01, "learning_rate": 4.99961827769432e-05, "loss": 0.6044, "step": 5662 }, { "epoch": 0.01, "learning_rate": 4.999618140424668e-05, "loss": 1.3677, "step": 5663 }, { "epoch": 0.01, "learning_rate": 4.999618003130341e-05, "loss": 1.2567, "step": 5664 }, { "epoch": 0.01, "learning_rate": 4.999617865811338e-05, "loss": 1.283, "step": 5665 }, { "epoch": 0.01, "learning_rate": 4.999617728467661e-05, "loss": 1.1749, "step": 5666 }, { "epoch": 0.01, "learning_rate": 4.999617591099308e-05, "loss": 0.912, "step": 5667 }, { "epoch": 0.01, "learning_rate": 4.99961745370628e-05, "loss": 1.248, "step": 5668 }, { "epoch": 0.01, "learning_rate": 4.999617316288577e-05, "loss": 1.3435, "step": 5669 }, { "epoch": 0.01, "learning_rate": 4.9996171788461996e-05, "loss": 1.1158, "step": 5670 }, { "epoch": 0.01, "learning_rate": 4.999617041379146e-05, "loss": 0.7348, "step": 5671 }, { "epoch": 0.01, "learning_rate": 4.999616903887417e-05, "loss": 1.1834, "step": 5672 }, { "epoch": 0.01, "learning_rate": 4.999616766371014e-05, "loss": 1.0043, "step": 5673 }, { "epoch": 0.01, "learning_rate": 4.9996166288299354e-05, "loss": 1.0576, "step": 5674 }, { "epoch": 0.01, "learning_rate": 4.9996164912641804e-05, "loss": 1.4277, "step": 5675 }, { "epoch": 0.01, "learning_rate": 4.999616353673752e-05, "loss": 0.969, "step": 5676 }, { "epoch": 0.01, "learning_rate": 4.999616216058647e-05, "loss": 0.719, "step": 5677 }, { "epoch": 0.01, "learning_rate": 4.999616078418868e-05, "loss": 1.0071, "step": 5678 }, { "epoch": 0.01, "learning_rate": 4.999615940754413e-05, "loss": 1.0275, "step": 5679 }, { "epoch": 0.01, "learning_rate": 4.9996158030652834e-05, "loss": 1.1718, "step": 5680 }, { "epoch": 0.01, "learning_rate": 4.9996156653514786e-05, "loss": 1.0183, "step": 5681 }, { "epoch": 0.01, "learning_rate": 4.999615527612999e-05, "loss": 1.116, "step": 5682 }, { "epoch": 0.01, "learning_rate": 4.999615389849843e-05, "loss": 1.2664, "step": 5683 }, { "epoch": 0.01, "learning_rate": 4.999615252062013e-05, "loss": 1.2419, "step": 5684 }, { "epoch": 0.01, "learning_rate": 4.999615114249507e-05, "loss": 1.298, "step": 5685 }, { "epoch": 0.01, "learning_rate": 4.999614976412327e-05, "loss": 1.3061, "step": 5686 }, { "epoch": 0.01, "learning_rate": 4.99961483855047e-05, "loss": 1.0749, "step": 5687 }, { "epoch": 0.01, "learning_rate": 4.99961470066394e-05, "loss": 0.6447, "step": 5688 }, { "epoch": 0.01, "learning_rate": 4.999614562752734e-05, "loss": 0.4944, "step": 5689 }, { "epoch": 0.01, "learning_rate": 4.9996144248168525e-05, "loss": 0.7365, "step": 5690 }, { "epoch": 0.01, "learning_rate": 4.999614286856296e-05, "loss": 1.5031, "step": 5691 }, { "epoch": 0.01, "learning_rate": 4.999614148871065e-05, "loss": 1.3854, "step": 5692 }, { "epoch": 0.01, "learning_rate": 4.9996140108611576e-05, "loss": 1.0011, "step": 5693 }, { "epoch": 0.01, "learning_rate": 4.999613872826576e-05, "loss": 1.2076, "step": 5694 }, { "epoch": 0.01, "learning_rate": 4.9996137347673186e-05, "loss": 0.9505, "step": 5695 }, { "epoch": 0.01, "learning_rate": 4.999613596683387e-05, "loss": 1.2965, "step": 5696 }, { "epoch": 0.01, "learning_rate": 4.99961345857478e-05, "loss": 1.14, "step": 5697 }, { "epoch": 0.01, "learning_rate": 4.9996133204414974e-05, "loss": 1.3744, "step": 5698 }, { "epoch": 0.01, "learning_rate": 4.9996131822835403e-05, "loss": 1.2792, "step": 5699 }, { "epoch": 0.01, "learning_rate": 4.999613044100907e-05, "loss": 1.2024, "step": 5700 }, { "epoch": 0.01, "learning_rate": 4.9996129058936e-05, "loss": 1.1759, "step": 5701 }, { "epoch": 0.01, "learning_rate": 4.999612767661617e-05, "loss": 1.1439, "step": 5702 }, { "epoch": 0.01, "learning_rate": 4.999612629404959e-05, "loss": 1.1925, "step": 5703 }, { "epoch": 0.01, "learning_rate": 4.9996124911236255e-05, "loss": 0.9742, "step": 5704 }, { "epoch": 0.01, "learning_rate": 4.999612352817617e-05, "loss": 1.1068, "step": 5705 }, { "epoch": 0.01, "learning_rate": 4.999612214486934e-05, "loss": 0.5886, "step": 5706 }, { "epoch": 0.01, "learning_rate": 4.9996120761315756e-05, "loss": 0.991, "step": 5707 }, { "epoch": 0.01, "learning_rate": 4.999611937751542e-05, "loss": 1.0096, "step": 5708 }, { "epoch": 0.01, "learning_rate": 4.9996117993468336e-05, "loss": 0.9139, "step": 5709 }, { "epoch": 0.01, "learning_rate": 4.999611660917449e-05, "loss": 1.185, "step": 5710 }, { "epoch": 0.01, "learning_rate": 4.999611522463391e-05, "loss": 1.2934, "step": 5711 }, { "epoch": 0.01, "learning_rate": 4.999611383984657e-05, "loss": 0.9533, "step": 5712 }, { "epoch": 0.01, "learning_rate": 4.999611245481247e-05, "loss": 1.0366, "step": 5713 }, { "epoch": 0.01, "learning_rate": 4.999611106953164e-05, "loss": 0.8981, "step": 5714 }, { "epoch": 0.01, "learning_rate": 4.999610968400404e-05, "loss": 0.8346, "step": 5715 }, { "epoch": 0.01, "learning_rate": 4.999610829822969e-05, "loss": 1.0552, "step": 5716 }, { "epoch": 0.01, "learning_rate": 4.999610691220859e-05, "loss": 1.1626, "step": 5717 }, { "epoch": 0.01, "learning_rate": 4.999610552594075e-05, "loss": 1.4871, "step": 5718 }, { "epoch": 0.01, "learning_rate": 4.999610413942616e-05, "loss": 1.0045, "step": 5719 }, { "epoch": 0.01, "learning_rate": 4.999610275266481e-05, "loss": 1.1701, "step": 5720 }, { "epoch": 0.01, "learning_rate": 4.999610136565671e-05, "loss": 1.1458, "step": 5721 }, { "epoch": 0.01, "learning_rate": 4.999609997840185e-05, "loss": 1.342, "step": 5722 }, { "epoch": 0.01, "learning_rate": 4.9996098590900255e-05, "loss": 1.2274, "step": 5723 }, { "epoch": 0.01, "learning_rate": 4.999609720315191e-05, "loss": 0.5639, "step": 5724 }, { "epoch": 0.01, "learning_rate": 4.9996095815156794e-05, "loss": 0.4767, "step": 5725 }, { "epoch": 0.01, "learning_rate": 4.9996094426914943e-05, "loss": 1.0088, "step": 5726 }, { "epoch": 0.01, "learning_rate": 4.999609303842634e-05, "loss": 1.4282, "step": 5727 }, { "epoch": 0.01, "learning_rate": 4.999609164969098e-05, "loss": 0.7006, "step": 5728 }, { "epoch": 0.01, "learning_rate": 4.9996090260708874e-05, "loss": 1.1023, "step": 5729 }, { "epoch": 0.01, "learning_rate": 4.999608887148002e-05, "loss": 1.2316, "step": 5730 }, { "epoch": 0.01, "learning_rate": 4.999608748200441e-05, "loss": 1.4644, "step": 5731 }, { "epoch": 0.01, "learning_rate": 4.9996086092282046e-05, "loss": 0.9136, "step": 5732 }, { "epoch": 0.01, "learning_rate": 4.999608470231294e-05, "loss": 1.2339, "step": 5733 }, { "epoch": 0.01, "learning_rate": 4.9996083312097076e-05, "loss": 1.3003, "step": 5734 }, { "epoch": 0.01, "learning_rate": 4.999608192163446e-05, "loss": 1.3405, "step": 5735 }, { "epoch": 0.01, "learning_rate": 4.99960805309251e-05, "loss": 1.1211, "step": 5736 }, { "epoch": 0.01, "learning_rate": 4.9996079139968986e-05, "loss": 1.167, "step": 5737 }, { "epoch": 0.01, "learning_rate": 4.9996077748766125e-05, "loss": 1.0671, "step": 5738 }, { "epoch": 0.01, "learning_rate": 4.999607635731651e-05, "loss": 1.2298, "step": 5739 }, { "epoch": 0.01, "learning_rate": 4.9996074965620144e-05, "loss": 1.0296, "step": 5740 }, { "epoch": 0.01, "learning_rate": 4.9996073573677024e-05, "loss": 1.244, "step": 5741 }, { "epoch": 0.01, "learning_rate": 4.9996072181487154e-05, "loss": 1.1463, "step": 5742 }, { "epoch": 0.01, "learning_rate": 4.999607078905054e-05, "loss": 1.0963, "step": 5743 }, { "epoch": 0.01, "learning_rate": 4.999606939636717e-05, "loss": 1.1444, "step": 5744 }, { "epoch": 0.01, "learning_rate": 4.999606800343705e-05, "loss": 1.1883, "step": 5745 }, { "epoch": 0.01, "learning_rate": 4.999606661026018e-05, "loss": 1.0707, "step": 5746 }, { "epoch": 0.01, "learning_rate": 4.9996065216836563e-05, "loss": 1.1605, "step": 5747 }, { "epoch": 0.01, "learning_rate": 4.999606382316618e-05, "loss": 1.0791, "step": 5748 }, { "epoch": 0.01, "learning_rate": 4.999606242924907e-05, "loss": 0.7398, "step": 5749 }, { "epoch": 0.01, "learning_rate": 4.9996061035085196e-05, "loss": 1.1299, "step": 5750 }, { "epoch": 0.01, "learning_rate": 4.999605964067457e-05, "loss": 1.389, "step": 5751 }, { "epoch": 0.01, "learning_rate": 4.9996058246017195e-05, "loss": 1.3896, "step": 5752 }, { "epoch": 0.01, "learning_rate": 4.999605685111307e-05, "loss": 0.9519, "step": 5753 }, { "epoch": 0.01, "learning_rate": 4.99960554559622e-05, "loss": 1.1582, "step": 5754 }, { "epoch": 0.01, "learning_rate": 4.999605406056457e-05, "loss": 1.1379, "step": 5755 }, { "epoch": 0.01, "learning_rate": 4.9996052664920195e-05, "loss": 1.1026, "step": 5756 }, { "epoch": 0.01, "learning_rate": 4.999605126902907e-05, "loss": 0.9483, "step": 5757 }, { "epoch": 0.01, "learning_rate": 4.999604987289119e-05, "loss": 1.1033, "step": 5758 }, { "epoch": 0.01, "learning_rate": 4.999604847650656e-05, "loss": 1.5733, "step": 5759 }, { "epoch": 0.01, "learning_rate": 4.999604707987519e-05, "loss": 1.185, "step": 5760 }, { "epoch": 0.01, "learning_rate": 4.999604568299706e-05, "loss": 1.3739, "step": 5761 }, { "epoch": 0.01, "learning_rate": 4.999604428587218e-05, "loss": 1.0264, "step": 5762 }, { "epoch": 0.01, "learning_rate": 4.999604288850055e-05, "loss": 1.1799, "step": 5763 }, { "epoch": 0.01, "learning_rate": 4.9996041490882175e-05, "loss": 1.2137, "step": 5764 }, { "epoch": 0.01, "learning_rate": 4.999604009301704e-05, "loss": 2.0663, "step": 5765 }, { "epoch": 0.01, "learning_rate": 4.9996038694905163e-05, "loss": 1.1117, "step": 5766 }, { "epoch": 0.01, "learning_rate": 4.9996037296546535e-05, "loss": 1.0114, "step": 5767 }, { "epoch": 0.01, "learning_rate": 4.999603589794115e-05, "loss": 0.7678, "step": 5768 }, { "epoch": 0.01, "learning_rate": 4.999603449908902e-05, "loss": 0.844, "step": 5769 }, { "epoch": 0.01, "learning_rate": 4.9996033099990136e-05, "loss": 0.9896, "step": 5770 }, { "epoch": 0.01, "learning_rate": 4.999603170064451e-05, "loss": 0.3288, "step": 5771 }, { "epoch": 0.01, "learning_rate": 4.999603030105212e-05, "loss": 0.2919, "step": 5772 }, { "epoch": 0.01, "learning_rate": 4.9996028901212987e-05, "loss": 0.9983, "step": 5773 }, { "epoch": 0.01, "learning_rate": 4.999602750112711e-05, "loss": 1.1373, "step": 5774 }, { "epoch": 0.01, "learning_rate": 4.999602610079447e-05, "loss": 1.32, "step": 5775 }, { "epoch": 0.01, "learning_rate": 4.999602470021509e-05, "loss": 1.1605, "step": 5776 }, { "epoch": 0.01, "learning_rate": 4.999602329938896e-05, "loss": 1.2259, "step": 5777 }, { "epoch": 0.01, "learning_rate": 4.9996021898316074e-05, "loss": 0.7605, "step": 5778 }, { "epoch": 0.01, "learning_rate": 4.999602049699644e-05, "loss": 1.2776, "step": 5779 }, { "epoch": 0.01, "learning_rate": 4.999601909543006e-05, "loss": 1.2831, "step": 5780 }, { "epoch": 0.01, "learning_rate": 4.9996017693616924e-05, "loss": 1.0186, "step": 5781 }, { "epoch": 0.01, "learning_rate": 4.999601629155703e-05, "loss": 1.2855, "step": 5782 }, { "epoch": 0.01, "learning_rate": 4.99960148892504e-05, "loss": 1.0829, "step": 5783 }, { "epoch": 0.01, "learning_rate": 4.9996013486697016e-05, "loss": 1.1834, "step": 5784 }, { "epoch": 0.01, "learning_rate": 4.999601208389688e-05, "loss": 1.0825, "step": 5785 }, { "epoch": 0.01, "learning_rate": 4.9996010680850004e-05, "loss": 1.3013, "step": 5786 }, { "epoch": 0.01, "learning_rate": 4.9996009277556364e-05, "loss": 1.1277, "step": 5787 }, { "epoch": 0.01, "learning_rate": 4.999600787401598e-05, "loss": 1.436, "step": 5788 }, { "epoch": 0.01, "learning_rate": 4.999600647022885e-05, "loss": 1.2099, "step": 5789 }, { "epoch": 0.01, "learning_rate": 4.999600506619496e-05, "loss": 1.4521, "step": 5790 }, { "epoch": 0.01, "learning_rate": 4.9996003661914326e-05, "loss": 1.0093, "step": 5791 }, { "epoch": 0.01, "learning_rate": 4.9996002257386945e-05, "loss": 1.2973, "step": 5792 }, { "epoch": 0.01, "learning_rate": 4.99960008526128e-05, "loss": 1.0852, "step": 5793 }, { "epoch": 0.01, "learning_rate": 4.9995999447591926e-05, "loss": 1.2553, "step": 5794 }, { "epoch": 0.01, "learning_rate": 4.9995998042324286e-05, "loss": 0.214, "step": 5795 }, { "epoch": 0.01, "learning_rate": 4.99959966368099e-05, "loss": 0.1374, "step": 5796 }, { "epoch": 0.01, "learning_rate": 4.999599523104876e-05, "loss": 0.4754, "step": 5797 }, { "epoch": 0.01, "learning_rate": 4.9995993825040876e-05, "loss": 0.4334, "step": 5798 }, { "epoch": 0.01, "learning_rate": 4.999599241878624e-05, "loss": 0.441, "step": 5799 }, { "epoch": 0.01, "learning_rate": 4.999599101228486e-05, "loss": 0.3289, "step": 5800 }, { "epoch": 0.01, "learning_rate": 4.999598960553672e-05, "loss": 0.3035, "step": 5801 }, { "epoch": 0.01, "learning_rate": 4.999598819854184e-05, "loss": 0.2614, "step": 5802 }, { "epoch": 0.01, "learning_rate": 4.999598679130021e-05, "loss": 0.2639, "step": 5803 }, { "epoch": 0.01, "learning_rate": 4.9995985383811824e-05, "loss": 0.3609, "step": 5804 }, { "epoch": 0.01, "learning_rate": 4.999598397607669e-05, "loss": 0.5269, "step": 5805 }, { "epoch": 0.01, "learning_rate": 4.9995982568094804e-05, "loss": 0.2979, "step": 5806 }, { "epoch": 0.01, "learning_rate": 4.999598115986617e-05, "loss": 0.3059, "step": 5807 }, { "epoch": 0.01, "learning_rate": 4.999597975139079e-05, "loss": 0.6379, "step": 5808 }, { "epoch": 0.01, "learning_rate": 4.9995978342668656e-05, "loss": 0.9649, "step": 5809 }, { "epoch": 0.01, "learning_rate": 4.9995976933699774e-05, "loss": 0.9759, "step": 5810 }, { "epoch": 0.01, "learning_rate": 4.999597552448414e-05, "loss": 0.6076, "step": 5811 }, { "epoch": 0.01, "learning_rate": 4.999597411502176e-05, "loss": 1.2914, "step": 5812 }, { "epoch": 0.01, "learning_rate": 4.999597270531262e-05, "loss": 1.1692, "step": 5813 }, { "epoch": 0.01, "learning_rate": 4.9995971295356744e-05, "loss": 1.1402, "step": 5814 }, { "epoch": 0.01, "learning_rate": 4.999596988515411e-05, "loss": 1.1395, "step": 5815 }, { "epoch": 0.01, "learning_rate": 4.999596847470473e-05, "loss": 1.0746, "step": 5816 }, { "epoch": 0.01, "learning_rate": 4.9995967064008596e-05, "loss": 1.2578, "step": 5817 }, { "epoch": 0.01, "learning_rate": 4.9995965653065714e-05, "loss": 1.0937, "step": 5818 }, { "epoch": 0.01, "learning_rate": 4.999596424187609e-05, "loss": 1.114, "step": 5819 }, { "epoch": 0.01, "learning_rate": 4.9995962830439704e-05, "loss": 1.1062, "step": 5820 }, { "epoch": 0.01, "learning_rate": 4.9995961418756576e-05, "loss": 0.3034, "step": 5821 }, { "epoch": 0.01, "learning_rate": 4.99959600068267e-05, "loss": 1.3744, "step": 5822 }, { "epoch": 0.01, "learning_rate": 4.9995958594650074e-05, "loss": 1.0629, "step": 5823 }, { "epoch": 0.01, "learning_rate": 4.9995957182226693e-05, "loss": 1.1352, "step": 5824 }, { "epoch": 0.01, "learning_rate": 4.999595576955656e-05, "loss": 1.1774, "step": 5825 }, { "epoch": 0.01, "learning_rate": 4.9995954356639686e-05, "loss": 0.9296, "step": 5826 }, { "epoch": 0.01, "learning_rate": 4.999595294347605e-05, "loss": 1.1025, "step": 5827 }, { "epoch": 0.01, "learning_rate": 4.999595153006568e-05, "loss": 1.1803, "step": 5828 }, { "epoch": 0.01, "learning_rate": 4.9995950116408555e-05, "loss": 1.2585, "step": 5829 }, { "epoch": 0.01, "learning_rate": 4.999594870250468e-05, "loss": 0.8967, "step": 5830 }, { "epoch": 0.01, "learning_rate": 4.999594728835405e-05, "loss": 1.0024, "step": 5831 }, { "epoch": 0.01, "learning_rate": 4.9995945873956686e-05, "loss": 1.1998, "step": 5832 }, { "epoch": 0.01, "learning_rate": 4.999594445931256e-05, "loss": 1.1819, "step": 5833 }, { "epoch": 0.01, "learning_rate": 4.9995943044421685e-05, "loss": 1.3396, "step": 5834 }, { "epoch": 0.01, "learning_rate": 4.999594162928406e-05, "loss": 1.2523, "step": 5835 }, { "epoch": 0.01, "learning_rate": 4.999594021389969e-05, "loss": 1.3324, "step": 5836 }, { "epoch": 0.01, "learning_rate": 4.9995938798268573e-05, "loss": 1.0069, "step": 5837 }, { "epoch": 0.01, "learning_rate": 4.99959373823907e-05, "loss": 1.0306, "step": 5838 }, { "epoch": 0.01, "learning_rate": 4.999593596626608e-05, "loss": 0.9872, "step": 5839 }, { "epoch": 0.01, "learning_rate": 4.9995934549894704e-05, "loss": 1.0294, "step": 5840 }, { "epoch": 0.01, "learning_rate": 4.999593313327659e-05, "loss": 1.067, "step": 5841 }, { "epoch": 0.01, "learning_rate": 4.999593171641172e-05, "loss": 1.1862, "step": 5842 }, { "epoch": 0.01, "learning_rate": 4.9995930299300104e-05, "loss": 1.0927, "step": 5843 }, { "epoch": 0.01, "learning_rate": 4.999592888194173e-05, "loss": 1.2252, "step": 5844 }, { "epoch": 0.01, "learning_rate": 4.9995927464336625e-05, "loss": 1.2999, "step": 5845 }, { "epoch": 0.01, "learning_rate": 4.999592604648475e-05, "loss": 0.9361, "step": 5846 }, { "epoch": 0.01, "learning_rate": 4.9995924628386136e-05, "loss": 1.0318, "step": 5847 }, { "epoch": 0.01, "learning_rate": 4.9995923210040776e-05, "loss": 1.0358, "step": 5848 }, { "epoch": 0.01, "learning_rate": 4.9995921791448664e-05, "loss": 0.9336, "step": 5849 }, { "epoch": 0.01, "learning_rate": 4.99959203726098e-05, "loss": 0.7572, "step": 5850 }, { "epoch": 0.01, "learning_rate": 4.999591895352419e-05, "loss": 0.7159, "step": 5851 }, { "epoch": 0.01, "learning_rate": 4.9995917534191825e-05, "loss": 1.3437, "step": 5852 }, { "epoch": 0.01, "learning_rate": 4.9995916114612717e-05, "loss": 1.0615, "step": 5853 }, { "epoch": 0.01, "learning_rate": 4.999591469478686e-05, "loss": 1.2398, "step": 5854 }, { "epoch": 0.01, "learning_rate": 4.9995913274714254e-05, "loss": 4.6326, "step": 5855 }, { "epoch": 0.01, "learning_rate": 4.9995911854394894e-05, "loss": 3.5142, "step": 5856 }, { "epoch": 0.01, "learning_rate": 4.999591043382878e-05, "loss": 3.019, "step": 5857 }, { "epoch": 0.01, "learning_rate": 4.999590901301593e-05, "loss": 2.6754, "step": 5858 }, { "epoch": 0.01, "learning_rate": 4.999590759195633e-05, "loss": 2.7023, "step": 5859 }, { "epoch": 0.01, "learning_rate": 4.9995906170649976e-05, "loss": 2.5355, "step": 5860 }, { "epoch": 0.01, "learning_rate": 4.9995904749096875e-05, "loss": 2.4428, "step": 5861 }, { "epoch": 0.01, "learning_rate": 4.999590332729702e-05, "loss": 2.4379, "step": 5862 }, { "epoch": 0.01, "learning_rate": 4.999590190525042e-05, "loss": 2.3608, "step": 5863 }, { "epoch": 0.01, "learning_rate": 4.999590048295707e-05, "loss": 2.1782, "step": 5864 }, { "epoch": 0.01, "learning_rate": 4.9995899060416974e-05, "loss": 2.2068, "step": 5865 }, { "epoch": 0.01, "learning_rate": 4.999589763763013e-05, "loss": 2.152, "step": 5866 }, { "epoch": 0.01, "learning_rate": 4.9995896214596525e-05, "loss": 2.0718, "step": 5867 }, { "epoch": 0.01, "learning_rate": 4.999589479131618e-05, "loss": 2.2563, "step": 5868 }, { "epoch": 0.01, "learning_rate": 4.9995893367789086e-05, "loss": 2.1127, "step": 5869 }, { "epoch": 0.01, "learning_rate": 4.999589194401525e-05, "loss": 2.1021, "step": 5870 }, { "epoch": 0.01, "learning_rate": 4.999589051999465e-05, "loss": 2.1175, "step": 5871 }, { "epoch": 0.01, "learning_rate": 4.999588909572731e-05, "loss": 1.365, "step": 5872 }, { "epoch": 0.01, "learning_rate": 4.999588767121322e-05, "loss": 1.2585, "step": 5873 }, { "epoch": 0.01, "learning_rate": 4.999588624645238e-05, "loss": 1.2671, "step": 5874 }, { "epoch": 0.01, "learning_rate": 4.9995884821444797e-05, "loss": 0.9925, "step": 5875 }, { "epoch": 0.01, "learning_rate": 4.999588339619046e-05, "loss": 0.9696, "step": 5876 }, { "epoch": 0.01, "learning_rate": 4.999588197068937e-05, "loss": 1.2444, "step": 5877 }, { "epoch": 0.01, "learning_rate": 4.999588054494154e-05, "loss": 0.7877, "step": 5878 }, { "epoch": 0.01, "learning_rate": 4.9995879118946956e-05, "loss": 1.2257, "step": 5879 }, { "epoch": 0.01, "learning_rate": 4.999587769270562e-05, "loss": 1.4817, "step": 5880 }, { "epoch": 0.01, "learning_rate": 4.9995876266217544e-05, "loss": 1.2465, "step": 5881 }, { "epoch": 0.01, "learning_rate": 4.9995874839482715e-05, "loss": 1.3474, "step": 5882 }, { "epoch": 0.01, "learning_rate": 4.9995873412501135e-05, "loss": 1.6764, "step": 5883 }, { "epoch": 0.01, "learning_rate": 4.9995871985272804e-05, "loss": 0.9174, "step": 5884 }, { "epoch": 0.01, "learning_rate": 4.999587055779773e-05, "loss": 0.7554, "step": 5885 }, { "epoch": 0.01, "learning_rate": 4.999586913007591e-05, "loss": 0.7866, "step": 5886 }, { "epoch": 0.01, "learning_rate": 4.9995867702107334e-05, "loss": 1.0494, "step": 5887 }, { "epoch": 0.01, "learning_rate": 4.9995866273892014e-05, "loss": 1.0432, "step": 5888 }, { "epoch": 0.01, "learning_rate": 4.999586484542994e-05, "loss": 1.3386, "step": 5889 }, { "epoch": 0.01, "learning_rate": 4.999586341672112e-05, "loss": 1.6875, "step": 5890 }, { "epoch": 0.01, "learning_rate": 4.999586198776556e-05, "loss": 1.0927, "step": 5891 }, { "epoch": 0.01, "learning_rate": 4.999586055856324e-05, "loss": 0.8526, "step": 5892 }, { "epoch": 0.01, "learning_rate": 4.9995859129114175e-05, "loss": 0.8085, "step": 5893 }, { "epoch": 0.01, "learning_rate": 4.9995857699418357e-05, "loss": 0.9571, "step": 5894 }, { "epoch": 0.01, "learning_rate": 4.99958562694758e-05, "loss": 0.8421, "step": 5895 }, { "epoch": 0.01, "learning_rate": 4.999585483928649e-05, "loss": 0.9443, "step": 5896 }, { "epoch": 0.01, "learning_rate": 4.999585340885043e-05, "loss": 0.76, "step": 5897 }, { "epoch": 0.01, "learning_rate": 4.9995851978167626e-05, "loss": 0.7921, "step": 5898 }, { "epoch": 0.01, "learning_rate": 4.999585054723807e-05, "loss": 0.5879, "step": 5899 }, { "epoch": 0.01, "learning_rate": 4.999584911606176e-05, "loss": 0.6805, "step": 5900 }, { "epoch": 0.01, "learning_rate": 4.999584768463871e-05, "loss": 0.7923, "step": 5901 }, { "epoch": 0.01, "learning_rate": 4.99958462529689e-05, "loss": 0.6167, "step": 5902 }, { "epoch": 0.01, "learning_rate": 4.999584482105236e-05, "loss": 0.9948, "step": 5903 }, { "epoch": 0.01, "learning_rate": 4.999584338888906e-05, "loss": 0.3271, "step": 5904 }, { "epoch": 0.01, "learning_rate": 4.999584195647901e-05, "loss": 1.1693, "step": 5905 }, { "epoch": 0.01, "learning_rate": 4.999584052382221e-05, "loss": 0.9345, "step": 5906 }, { "epoch": 0.01, "learning_rate": 4.9995839090918673e-05, "loss": 1.5019, "step": 5907 }, { "epoch": 0.01, "learning_rate": 4.9995837657768376e-05, "loss": 1.1689, "step": 5908 }, { "epoch": 0.01, "learning_rate": 4.999583622437134e-05, "loss": 1.1574, "step": 5909 }, { "epoch": 0.01, "learning_rate": 4.999583479072755e-05, "loss": 1.1267, "step": 5910 }, { "epoch": 0.01, "learning_rate": 4.9995833356837014e-05, "loss": 1.2111, "step": 5911 }, { "epoch": 0.01, "learning_rate": 4.999583192269973e-05, "loss": 1.0961, "step": 5912 }, { "epoch": 0.01, "learning_rate": 4.9995830488315696e-05, "loss": 1.2409, "step": 5913 }, { "epoch": 0.01, "learning_rate": 4.9995829053684914e-05, "loss": 1.0839, "step": 5914 }, { "epoch": 0.01, "learning_rate": 4.999582761880738e-05, "loss": 1.189, "step": 5915 }, { "epoch": 0.01, "learning_rate": 4.9995826183683104e-05, "loss": 0.9599, "step": 5916 }, { "epoch": 0.01, "learning_rate": 4.9995824748312076e-05, "loss": 1.2058, "step": 5917 }, { "epoch": 0.01, "learning_rate": 4.9995823312694304e-05, "loss": 1.1992, "step": 5918 }, { "epoch": 0.01, "learning_rate": 4.9995821876829775e-05, "loss": 0.9585, "step": 5919 }, { "epoch": 0.01, "learning_rate": 4.999582044071851e-05, "loss": 1.1868, "step": 5920 }, { "epoch": 0.01, "learning_rate": 4.9995819004360484e-05, "loss": 1.0446, "step": 5921 }, { "epoch": 0.01, "learning_rate": 4.9995817567755715e-05, "loss": 1.0714, "step": 5922 }, { "epoch": 0.01, "learning_rate": 4.99958161309042e-05, "loss": 1.1892, "step": 5923 }, { "epoch": 0.01, "learning_rate": 4.999581469380593e-05, "loss": 1.0839, "step": 5924 }, { "epoch": 0.01, "learning_rate": 4.9995813256460925e-05, "loss": 1.4469, "step": 5925 }, { "epoch": 0.01, "learning_rate": 4.9995811818869166e-05, "loss": 1.3577, "step": 5926 }, { "epoch": 0.01, "learning_rate": 4.999581038103065e-05, "loss": 1.2339, "step": 5927 }, { "epoch": 0.01, "learning_rate": 4.99958089429454e-05, "loss": 1.2053, "step": 5928 }, { "epoch": 0.01, "learning_rate": 4.999580750461339e-05, "loss": 1.0734, "step": 5929 }, { "epoch": 0.01, "learning_rate": 4.999580606603463e-05, "loss": 1.1296, "step": 5930 }, { "epoch": 0.01, "learning_rate": 4.999580462720914e-05, "loss": 1.1232, "step": 5931 }, { "epoch": 0.01, "learning_rate": 4.999580318813689e-05, "loss": 1.139, "step": 5932 }, { "epoch": 0.01, "learning_rate": 4.999580174881789e-05, "loss": 1.1215, "step": 5933 }, { "epoch": 0.01, "learning_rate": 4.999580030925214e-05, "loss": 0.9538, "step": 5934 }, { "epoch": 0.01, "learning_rate": 4.9995798869439647e-05, "loss": 0.9857, "step": 5935 }, { "epoch": 0.01, "learning_rate": 4.9995797429380407e-05, "loss": 1.0086, "step": 5936 }, { "epoch": 0.01, "learning_rate": 4.999579598907442e-05, "loss": 1.2054, "step": 5937 }, { "epoch": 0.01, "learning_rate": 4.999579454852168e-05, "loss": 1.0337, "step": 5938 }, { "epoch": 0.01, "learning_rate": 4.99957931077222e-05, "loss": 1.5357, "step": 5939 }, { "epoch": 0.01, "learning_rate": 4.999579166667596e-05, "loss": 1.2708, "step": 5940 }, { "epoch": 0.01, "learning_rate": 4.9995790225382985e-05, "loss": 1.1459, "step": 5941 }, { "epoch": 0.01, "learning_rate": 4.999578878384325e-05, "loss": 2.3755, "step": 5942 }, { "epoch": 0.01, "learning_rate": 4.999578734205678e-05, "loss": 1.2678, "step": 5943 }, { "epoch": 0.01, "learning_rate": 4.999578590002355e-05, "loss": 1.4369, "step": 5944 }, { "epoch": 0.01, "learning_rate": 4.999578445774358e-05, "loss": 1.0646, "step": 5945 }, { "epoch": 0.01, "learning_rate": 4.999578301521686e-05, "loss": 1.3442, "step": 5946 }, { "epoch": 0.01, "learning_rate": 4.999578157244339e-05, "loss": 1.3115, "step": 5947 }, { "epoch": 0.01, "learning_rate": 4.999578012942318e-05, "loss": 1.0278, "step": 5948 }, { "epoch": 0.01, "learning_rate": 4.9995778686156217e-05, "loss": 0.9083, "step": 5949 }, { "epoch": 0.01, "learning_rate": 4.99957772426425e-05, "loss": 1.231, "step": 5950 }, { "epoch": 0.01, "learning_rate": 4.9995775798882036e-05, "loss": 1.0653, "step": 5951 }, { "epoch": 0.01, "learning_rate": 4.999577435487484e-05, "loss": 0.9043, "step": 5952 }, { "epoch": 0.01, "learning_rate": 4.999577291062087e-05, "loss": 1.0906, "step": 5953 }, { "epoch": 0.01, "learning_rate": 4.999577146612018e-05, "loss": 0.9863, "step": 5954 }, { "epoch": 0.01, "learning_rate": 4.9995770021372726e-05, "loss": 1.097, "step": 5955 }, { "epoch": 0.01, "learning_rate": 4.999576857637852e-05, "loss": 1.0197, "step": 5956 }, { "epoch": 0.01, "learning_rate": 4.999576713113758e-05, "loss": 1.1653, "step": 5957 }, { "epoch": 0.01, "learning_rate": 4.9995765685649885e-05, "loss": 1.1057, "step": 5958 }, { "epoch": 0.01, "learning_rate": 4.999576423991544e-05, "loss": 1.0866, "step": 5959 }, { "epoch": 0.01, "learning_rate": 4.999576279393425e-05, "loss": 1.7562, "step": 5960 }, { "epoch": 0.01, "learning_rate": 4.999576134770631e-05, "loss": 1.2862, "step": 5961 }, { "epoch": 0.01, "learning_rate": 4.999575990123163e-05, "loss": 1.2652, "step": 5962 }, { "epoch": 0.01, "learning_rate": 4.99957584545102e-05, "loss": 1.1406, "step": 5963 }, { "epoch": 0.01, "learning_rate": 4.9995757007542024e-05, "loss": 1.3136, "step": 5964 }, { "epoch": 0.01, "learning_rate": 4.99957555603271e-05, "loss": 1.0399, "step": 5965 }, { "epoch": 0.01, "learning_rate": 4.999575411286542e-05, "loss": 1.0532, "step": 5966 }, { "epoch": 0.01, "learning_rate": 4.9995752665157005e-05, "loss": 1.0086, "step": 5967 }, { "epoch": 0.01, "learning_rate": 4.999575121720183e-05, "loss": 1.0175, "step": 5968 }, { "epoch": 0.01, "learning_rate": 4.999574976899991e-05, "loss": 1.0899, "step": 5969 }, { "epoch": 0.01, "learning_rate": 4.999574832055125e-05, "loss": 1.1038, "step": 5970 }, { "epoch": 0.01, "learning_rate": 4.9995746871855836e-05, "loss": 1.323, "step": 5971 }, { "epoch": 0.01, "learning_rate": 4.9995745422913674e-05, "loss": 1.26, "step": 5972 }, { "epoch": 0.01, "learning_rate": 4.9995743973724774e-05, "loss": 1.1442, "step": 5973 }, { "epoch": 0.01, "learning_rate": 4.9995742524289116e-05, "loss": 1.0892, "step": 5974 }, { "epoch": 0.01, "learning_rate": 4.9995741074606715e-05, "loss": 1.3687, "step": 5975 }, { "epoch": 0.01, "learning_rate": 4.999573962467756e-05, "loss": 1.1966, "step": 5976 }, { "epoch": 0.01, "learning_rate": 4.9995738174501666e-05, "loss": 1.1295, "step": 5977 }, { "epoch": 0.01, "learning_rate": 4.9995736724079025e-05, "loss": 1.2428, "step": 5978 }, { "epoch": 0.01, "learning_rate": 4.999573527340963e-05, "loss": 0.745, "step": 5979 }, { "epoch": 0.01, "learning_rate": 4.99957338224935e-05, "loss": 1.1817, "step": 5980 }, { "epoch": 0.01, "learning_rate": 4.9995732371330605e-05, "loss": 1.0781, "step": 5981 }, { "epoch": 0.01, "learning_rate": 4.999573091992097e-05, "loss": 0.9299, "step": 5982 }, { "epoch": 0.01, "learning_rate": 4.999572946826459e-05, "loss": 0.9942, "step": 5983 }, { "epoch": 0.01, "learning_rate": 4.999572801636147e-05, "loss": 1.1955, "step": 5984 }, { "epoch": 0.01, "learning_rate": 4.9995726564211585e-05, "loss": 1.2949, "step": 5985 }, { "epoch": 0.01, "learning_rate": 4.9995725111814964e-05, "loss": 1.2096, "step": 5986 }, { "epoch": 0.01, "learning_rate": 4.999572365917159e-05, "loss": 1.4178, "step": 5987 }, { "epoch": 0.01, "learning_rate": 4.999572220628148e-05, "loss": 1.6477, "step": 5988 }, { "epoch": 0.01, "learning_rate": 4.999572075314462e-05, "loss": 1.3698, "step": 5989 }, { "epoch": 0.01, "learning_rate": 4.9995719299761e-05, "loss": 0.9212, "step": 5990 }, { "epoch": 0.01, "learning_rate": 4.999571784613064e-05, "loss": 1.149, "step": 5991 }, { "epoch": 0.01, "learning_rate": 4.9995716392253535e-05, "loss": 0.8656, "step": 5992 }, { "epoch": 0.01, "learning_rate": 4.999571493812968e-05, "loss": 1.2262, "step": 5993 }, { "epoch": 0.01, "learning_rate": 4.999571348375908e-05, "loss": 0.8487, "step": 5994 }, { "epoch": 0.01, "learning_rate": 4.9995712029141734e-05, "loss": 0.3838, "step": 5995 }, { "epoch": 0.01, "learning_rate": 4.999571057427764e-05, "loss": 1.2446, "step": 5996 }, { "epoch": 0.01, "learning_rate": 4.999570911916679e-05, "loss": 0.8315, "step": 5997 }, { "epoch": 0.01, "learning_rate": 4.999570766380921e-05, "loss": 1.1521, "step": 5998 }, { "epoch": 0.01, "learning_rate": 4.9995706208204875e-05, "loss": 1.2743, "step": 5999 }, { "epoch": 0.01, "learning_rate": 4.999570475235379e-05, "loss": 1.1449, "step": 6000 }, { "epoch": 0.01, "eval_loss": 1.0794674158096313, "eval_runtime": 86.1885, "eval_samples_per_second": 16.069, "eval_steps_per_second": 4.026, "step": 6000 }, { "epoch": 0.01, "learning_rate": 4.9995703296255955e-05, "loss": 1.0349, "step": 6001 }, { "epoch": 0.01, "learning_rate": 4.999570183991138e-05, "loss": 1.0534, "step": 6002 }, { "epoch": 0.01, "learning_rate": 4.999570038332006e-05, "loss": 1.3352, "step": 6003 }, { "epoch": 0.01, "learning_rate": 4.999569892648198e-05, "loss": 1.2638, "step": 6004 }, { "epoch": 0.01, "learning_rate": 4.999569746939716e-05, "loss": 1.185, "step": 6005 }, { "epoch": 0.01, "learning_rate": 4.9995696012065593e-05, "loss": 1.1458, "step": 6006 }, { "epoch": 0.01, "learning_rate": 4.9995694554487286e-05, "loss": 1.2757, "step": 6007 }, { "epoch": 0.01, "learning_rate": 4.999569309666223e-05, "loss": 1.2362, "step": 6008 }, { "epoch": 0.01, "learning_rate": 4.9995691638590426e-05, "loss": 1.2912, "step": 6009 }, { "epoch": 0.01, "learning_rate": 4.999569018027187e-05, "loss": 1.4617, "step": 6010 }, { "epoch": 0.01, "learning_rate": 4.999568872170657e-05, "loss": 1.0121, "step": 6011 }, { "epoch": 0.01, "learning_rate": 4.999568726289452e-05, "loss": 1.678, "step": 6012 }, { "epoch": 0.01, "learning_rate": 4.999568580383572e-05, "loss": 0.97, "step": 6013 }, { "epoch": 0.01, "learning_rate": 4.9995684344530186e-05, "loss": 0.9885, "step": 6014 }, { "epoch": 0.01, "learning_rate": 4.99956828849779e-05, "loss": 0.9913, "step": 6015 }, { "epoch": 0.01, "learning_rate": 4.999568142517886e-05, "loss": 1.0645, "step": 6016 }, { "epoch": 0.01, "learning_rate": 4.999567996513308e-05, "loss": 1.6482, "step": 6017 }, { "epoch": 0.01, "learning_rate": 4.9995678504840554e-05, "loss": 0.8483, "step": 6018 }, { "epoch": 0.01, "learning_rate": 4.9995677044301284e-05, "loss": 0.9729, "step": 6019 }, { "epoch": 0.01, "learning_rate": 4.9995675583515256e-05, "loss": 0.7855, "step": 6020 }, { "epoch": 0.01, "learning_rate": 4.999567412248249e-05, "loss": 1.9137, "step": 6021 }, { "epoch": 0.01, "learning_rate": 4.9995672661202975e-05, "loss": 1.5124, "step": 6022 }, { "epoch": 0.01, "learning_rate": 4.999567119967671e-05, "loss": 1.4559, "step": 6023 }, { "epoch": 0.01, "learning_rate": 4.9995669737903704e-05, "loss": 1.3439, "step": 6024 }, { "epoch": 0.01, "learning_rate": 4.999566827588394e-05, "loss": 1.2469, "step": 6025 }, { "epoch": 0.01, "learning_rate": 4.999566681361745e-05, "loss": 1.208, "step": 6026 }, { "epoch": 0.01, "learning_rate": 4.9995665351104193e-05, "loss": 1.1692, "step": 6027 }, { "epoch": 0.01, "learning_rate": 4.99956638883442e-05, "loss": 1.1804, "step": 6028 }, { "epoch": 0.01, "learning_rate": 4.999566242533746e-05, "loss": 1.3017, "step": 6029 }, { "epoch": 0.01, "learning_rate": 4.9995660962083966e-05, "loss": 1.2244, "step": 6030 }, { "epoch": 0.01, "learning_rate": 4.999565949858373e-05, "loss": 1.324, "step": 6031 }, { "epoch": 0.01, "learning_rate": 4.999565803483675e-05, "loss": 1.6463, "step": 6032 }, { "epoch": 0.01, "learning_rate": 4.999565657084302e-05, "loss": 1.201, "step": 6033 }, { "epoch": 0.01, "learning_rate": 4.999565510660255e-05, "loss": 0.8303, "step": 6034 }, { "epoch": 0.01, "learning_rate": 4.9995653642115326e-05, "loss": 0.3958, "step": 6035 }, { "epoch": 0.01, "learning_rate": 4.999565217738136e-05, "loss": 0.8038, "step": 6036 }, { "epoch": 0.01, "learning_rate": 4.999565071240064e-05, "loss": 1.1105, "step": 6037 }, { "epoch": 0.01, "learning_rate": 4.999564924717318e-05, "loss": 1.0792, "step": 6038 }, { "epoch": 0.01, "learning_rate": 4.999564778169897e-05, "loss": 0.9225, "step": 6039 }, { "epoch": 0.01, "learning_rate": 4.9995646315978016e-05, "loss": 0.7761, "step": 6040 }, { "epoch": 0.01, "learning_rate": 4.9995644850010315e-05, "loss": 0.7188, "step": 6041 }, { "epoch": 0.01, "learning_rate": 4.9995643383795864e-05, "loss": 0.854, "step": 6042 }, { "epoch": 0.01, "learning_rate": 4.9995641917334674e-05, "loss": 1.1032, "step": 6043 }, { "epoch": 0.01, "learning_rate": 4.9995640450626734e-05, "loss": 0.8793, "step": 6044 }, { "epoch": 0.01, "learning_rate": 4.9995638983672044e-05, "loss": 0.5732, "step": 6045 }, { "epoch": 0.01, "learning_rate": 4.999563751647061e-05, "loss": 1.0618, "step": 6046 }, { "epoch": 0.01, "learning_rate": 4.999563604902243e-05, "loss": 1.3174, "step": 6047 }, { "epoch": 0.01, "learning_rate": 4.999563458132751e-05, "loss": 1.0841, "step": 6048 }, { "epoch": 0.01, "learning_rate": 4.999563311338583e-05, "loss": 0.999, "step": 6049 }, { "epoch": 0.01, "learning_rate": 4.9995631645197414e-05, "loss": 1.0769, "step": 6050 }, { "epoch": 0.01, "learning_rate": 4.999563017676225e-05, "loss": 1.007, "step": 6051 }, { "epoch": 0.01, "learning_rate": 4.999562870808034e-05, "loss": 1.052, "step": 6052 }, { "epoch": 0.01, "learning_rate": 4.9995627239151675e-05, "loss": 1.1273, "step": 6053 }, { "epoch": 0.01, "learning_rate": 4.9995625769976274e-05, "loss": 1.076, "step": 6054 }, { "epoch": 0.01, "learning_rate": 4.999562430055412e-05, "loss": 0.9607, "step": 6055 }, { "epoch": 0.01, "learning_rate": 4.999562283088523e-05, "loss": 1.3641, "step": 6056 }, { "epoch": 0.01, "learning_rate": 4.9995621360969586e-05, "loss": 1.1847, "step": 6057 }, { "epoch": 0.01, "learning_rate": 4.99956198908072e-05, "loss": 1.026, "step": 6058 }, { "epoch": 0.01, "learning_rate": 4.999561842039806e-05, "loss": 1.3139, "step": 6059 }, { "epoch": 0.01, "learning_rate": 4.999561694974218e-05, "loss": 1.0536, "step": 6060 }, { "epoch": 0.01, "learning_rate": 4.999561547883955e-05, "loss": 0.956, "step": 6061 }, { "epoch": 0.01, "learning_rate": 4.9995614007690176e-05, "loss": 1.3575, "step": 6062 }, { "epoch": 0.01, "learning_rate": 4.999561253629406e-05, "loss": 1.0711, "step": 6063 }, { "epoch": 0.01, "learning_rate": 4.9995611064651195e-05, "loss": 1.0042, "step": 6064 }, { "epoch": 0.01, "learning_rate": 4.999560959276158e-05, "loss": 0.8043, "step": 6065 }, { "epoch": 0.01, "learning_rate": 4.999560812062522e-05, "loss": 1.4357, "step": 6066 }, { "epoch": 0.01, "learning_rate": 4.999560664824212e-05, "loss": 1.1579, "step": 6067 }, { "epoch": 0.01, "learning_rate": 4.9995605175612264e-05, "loss": 1.1137, "step": 6068 }, { "epoch": 0.01, "learning_rate": 4.9995603702735675e-05, "loss": 1.3972, "step": 6069 }, { "epoch": 0.01, "learning_rate": 4.999560222961233e-05, "loss": 0.9841, "step": 6070 }, { "epoch": 0.01, "learning_rate": 4.9995600756242236e-05, "loss": 1.4509, "step": 6071 }, { "epoch": 0.01, "learning_rate": 4.999559928262541e-05, "loss": 1.2235, "step": 6072 }, { "epoch": 0.01, "learning_rate": 4.999559780876182e-05, "loss": 1.0759, "step": 6073 }, { "epoch": 0.01, "learning_rate": 4.99955963346515e-05, "loss": 1.0943, "step": 6074 }, { "epoch": 0.01, "learning_rate": 4.9995594860294424e-05, "loss": 0.8859, "step": 6075 }, { "epoch": 0.01, "learning_rate": 4.99955933856906e-05, "loss": 1.0972, "step": 6076 }, { "epoch": 0.01, "learning_rate": 4.9995591910840036e-05, "loss": 1.0955, "step": 6077 }, { "epoch": 0.01, "learning_rate": 4.999559043574272e-05, "loss": 1.2344, "step": 6078 }, { "epoch": 0.01, "learning_rate": 4.9995588960398665e-05, "loss": 1.1871, "step": 6079 }, { "epoch": 0.01, "learning_rate": 4.999558748480786e-05, "loss": 1.0826, "step": 6080 }, { "epoch": 0.01, "learning_rate": 4.999558600897032e-05, "loss": 1.0216, "step": 6081 }, { "epoch": 0.01, "learning_rate": 4.999558453288602e-05, "loss": 0.9615, "step": 6082 }, { "epoch": 0.01, "learning_rate": 4.999558305655499e-05, "loss": 1.3907, "step": 6083 }, { "epoch": 0.01, "learning_rate": 4.999558157997719e-05, "loss": 1.0384, "step": 6084 }, { "epoch": 0.01, "learning_rate": 4.999558010315266e-05, "loss": 1.2994, "step": 6085 }, { "epoch": 0.01, "learning_rate": 4.999557862608139e-05, "loss": 1.2716, "step": 6086 }, { "epoch": 0.01, "learning_rate": 4.999557714876336e-05, "loss": 1.1883, "step": 6087 }, { "epoch": 0.01, "learning_rate": 4.999557567119859e-05, "loss": 1.1904, "step": 6088 }, { "epoch": 0.01, "learning_rate": 4.9995574193387077e-05, "loss": 1.3526, "step": 6089 }, { "epoch": 0.01, "learning_rate": 4.9995572715328814e-05, "loss": 1.2066, "step": 6090 }, { "epoch": 0.01, "learning_rate": 4.999557123702381e-05, "loss": 0.7818, "step": 6091 }, { "epoch": 0.01, "learning_rate": 4.9995569758472056e-05, "loss": 0.7816, "step": 6092 }, { "epoch": 0.01, "learning_rate": 4.9995568279673554e-05, "loss": 1.0388, "step": 6093 }, { "epoch": 0.01, "learning_rate": 4.9995566800628315e-05, "loss": 1.3529, "step": 6094 }, { "epoch": 0.01, "learning_rate": 4.9995565321336325e-05, "loss": 2.0495, "step": 6095 }, { "epoch": 0.01, "learning_rate": 4.999556384179759e-05, "loss": 1.6708, "step": 6096 }, { "epoch": 0.01, "learning_rate": 4.9995562362012106e-05, "loss": 1.2122, "step": 6097 }, { "epoch": 0.01, "learning_rate": 4.999556088197988e-05, "loss": 1.3476, "step": 6098 }, { "epoch": 0.01, "learning_rate": 4.999555940170091e-05, "loss": 1.5162, "step": 6099 }, { "epoch": 0.01, "learning_rate": 4.9995557921175186e-05, "loss": 1.7973, "step": 6100 }, { "epoch": 0.01, "learning_rate": 4.9995556440402725e-05, "loss": 1.6417, "step": 6101 }, { "epoch": 0.01, "learning_rate": 4.999555495938352e-05, "loss": 1.5791, "step": 6102 }, { "epoch": 0.01, "learning_rate": 4.9995553478117563e-05, "loss": 0.623, "step": 6103 }, { "epoch": 0.01, "learning_rate": 4.999555199660486e-05, "loss": 1.1451, "step": 6104 }, { "epoch": 0.01, "learning_rate": 4.999555051484541e-05, "loss": 1.2129, "step": 6105 }, { "epoch": 0.01, "learning_rate": 4.999554903283922e-05, "loss": 0.9466, "step": 6106 }, { "epoch": 0.01, "learning_rate": 4.9995547550586284e-05, "loss": 0.4574, "step": 6107 }, { "epoch": 0.01, "learning_rate": 4.99955460680866e-05, "loss": 0.6577, "step": 6108 }, { "epoch": 0.01, "learning_rate": 4.999554458534017e-05, "loss": 1.2173, "step": 6109 }, { "epoch": 0.01, "learning_rate": 4.9995543102346995e-05, "loss": 1.0804, "step": 6110 }, { "epoch": 0.01, "learning_rate": 4.999554161910708e-05, "loss": 1.148, "step": 6111 }, { "epoch": 0.01, "learning_rate": 4.999554013562041e-05, "loss": 1.3866, "step": 6112 }, { "epoch": 0.01, "learning_rate": 4.999553865188701e-05, "loss": 1.078, "step": 6113 }, { "epoch": 0.01, "learning_rate": 4.999553716790685e-05, "loss": 0.9918, "step": 6114 }, { "epoch": 0.01, "learning_rate": 4.999553568367995e-05, "loss": 0.9809, "step": 6115 }, { "epoch": 0.01, "learning_rate": 4.99955341992063e-05, "loss": 0.9585, "step": 6116 }, { "epoch": 0.01, "learning_rate": 4.999553271448591e-05, "loss": 1.3153, "step": 6117 }, { "epoch": 0.01, "learning_rate": 4.999553122951878e-05, "loss": 1.0406, "step": 6118 }, { "epoch": 0.01, "learning_rate": 4.999552974430489e-05, "loss": 1.2737, "step": 6119 }, { "epoch": 0.01, "learning_rate": 4.999552825884426e-05, "loss": 1.3108, "step": 6120 }, { "epoch": 0.01, "learning_rate": 4.999552677313689e-05, "loss": 1.1334, "step": 6121 }, { "epoch": 0.01, "learning_rate": 4.9995525287182775e-05, "loss": 1.6797, "step": 6122 }, { "epoch": 0.01, "learning_rate": 4.999552380098191e-05, "loss": 1.2596, "step": 6123 }, { "epoch": 0.01, "learning_rate": 4.99955223145343e-05, "loss": 0.8459, "step": 6124 }, { "epoch": 0.01, "learning_rate": 4.999552082783995e-05, "loss": 1.0164, "step": 6125 }, { "epoch": 0.01, "learning_rate": 4.999551934089884e-05, "loss": 0.5692, "step": 6126 }, { "epoch": 0.01, "learning_rate": 4.9995517853711e-05, "loss": 0.6399, "step": 6127 }, { "epoch": 0.01, "learning_rate": 4.999551636627641e-05, "loss": 1.1292, "step": 6128 }, { "epoch": 0.01, "learning_rate": 4.999551487859507e-05, "loss": 0.9345, "step": 6129 }, { "epoch": 0.01, "learning_rate": 4.999551339066699e-05, "loss": 1.1914, "step": 6130 }, { "epoch": 0.01, "learning_rate": 4.999551190249217e-05, "loss": 1.0834, "step": 6131 }, { "epoch": 0.01, "learning_rate": 4.99955104140706e-05, "loss": 1.3149, "step": 6132 }, { "epoch": 0.01, "learning_rate": 4.999550892540229e-05, "loss": 1.2369, "step": 6133 }, { "epoch": 0.01, "learning_rate": 4.999550743648722e-05, "loss": 1.0676, "step": 6134 }, { "epoch": 0.01, "learning_rate": 4.9995505947325415e-05, "loss": 1.2102, "step": 6135 }, { "epoch": 0.01, "learning_rate": 4.9995504457916864e-05, "loss": 1.1858, "step": 6136 }, { "epoch": 0.01, "learning_rate": 4.999550296826157e-05, "loss": 0.9689, "step": 6137 }, { "epoch": 0.01, "learning_rate": 4.999550147835953e-05, "loss": 1.1618, "step": 6138 }, { "epoch": 0.01, "learning_rate": 4.9995499988210744e-05, "loss": 1.1267, "step": 6139 }, { "epoch": 0.01, "learning_rate": 4.999549849781521e-05, "loss": 0.9775, "step": 6140 }, { "epoch": 0.01, "learning_rate": 4.999549700717293e-05, "loss": 1.2295, "step": 6141 }, { "epoch": 0.01, "learning_rate": 4.9995495516283914e-05, "loss": 1.0337, "step": 6142 }, { "epoch": 0.01, "learning_rate": 4.999549402514815e-05, "loss": 0.7293, "step": 6143 }, { "epoch": 0.01, "learning_rate": 4.999549253376563e-05, "loss": 0.9217, "step": 6144 }, { "epoch": 0.01, "learning_rate": 4.999549104213638e-05, "loss": 0.9321, "step": 6145 }, { "epoch": 0.01, "learning_rate": 4.9995489550260374e-05, "loss": 0.7627, "step": 6146 }, { "epoch": 0.01, "learning_rate": 4.999548805813763e-05, "loss": 0.8338, "step": 6147 }, { "epoch": 0.01, "learning_rate": 4.999548656576814e-05, "loss": 0.8278, "step": 6148 }, { "epoch": 0.01, "learning_rate": 4.9995485073151905e-05, "loss": 1.0563, "step": 6149 }, { "epoch": 0.01, "learning_rate": 4.999548358028893e-05, "loss": 1.2953, "step": 6150 }, { "epoch": 0.01, "learning_rate": 4.9995482087179196e-05, "loss": 0.9959, "step": 6151 }, { "epoch": 0.01, "learning_rate": 4.999548059382273e-05, "loss": 1.1194, "step": 6152 }, { "epoch": 0.01, "learning_rate": 4.999547910021951e-05, "loss": 1.236, "step": 6153 }, { "epoch": 0.01, "learning_rate": 4.999547760636955e-05, "loss": 1.0783, "step": 6154 }, { "epoch": 0.01, "learning_rate": 4.999547611227284e-05, "loss": 1.3052, "step": 6155 }, { "epoch": 0.01, "learning_rate": 4.9995474617929395e-05, "loss": 0.9864, "step": 6156 }, { "epoch": 0.01, "learning_rate": 4.99954731233392e-05, "loss": 1.2228, "step": 6157 }, { "epoch": 0.01, "learning_rate": 4.999547162850226e-05, "loss": 1.0359, "step": 6158 }, { "epoch": 0.01, "learning_rate": 4.9995470133418576e-05, "loss": 0.8855, "step": 6159 }, { "epoch": 0.01, "learning_rate": 4.999546863808815e-05, "loss": 1.1619, "step": 6160 }, { "epoch": 0.01, "learning_rate": 4.999546714251098e-05, "loss": 1.0703, "step": 6161 }, { "epoch": 0.01, "learning_rate": 4.999546564668705e-05, "loss": 1.0541, "step": 6162 }, { "epoch": 0.01, "learning_rate": 4.999546415061639e-05, "loss": 1.301, "step": 6163 }, { "epoch": 0.01, "learning_rate": 4.9995462654298984e-05, "loss": 1.2866, "step": 6164 }, { "epoch": 0.01, "learning_rate": 4.999546115773483e-05, "loss": 1.3945, "step": 6165 }, { "epoch": 0.01, "learning_rate": 4.999545966092393e-05, "loss": 0.7452, "step": 6166 }, { "epoch": 0.01, "learning_rate": 4.9995458163866294e-05, "loss": 0.3172, "step": 6167 }, { "epoch": 0.01, "learning_rate": 4.9995456666561904e-05, "loss": 0.2841, "step": 6168 }, { "epoch": 0.01, "learning_rate": 4.999545516901078e-05, "loss": 0.2657, "step": 6169 }, { "epoch": 0.01, "learning_rate": 4.99954536712129e-05, "loss": 1.2747, "step": 6170 }, { "epoch": 0.01, "learning_rate": 4.999545217316828e-05, "loss": 0.9224, "step": 6171 }, { "epoch": 0.01, "learning_rate": 4.999545067487691e-05, "loss": 1.1763, "step": 6172 }, { "epoch": 0.01, "learning_rate": 4.99954491763388e-05, "loss": 1.1748, "step": 6173 }, { "epoch": 0.01, "learning_rate": 4.999544767755395e-05, "loss": 1.0592, "step": 6174 }, { "epoch": 0.01, "learning_rate": 4.999544617852235e-05, "loss": 0.9498, "step": 6175 }, { "epoch": 0.01, "learning_rate": 4.999544467924401e-05, "loss": 1.3276, "step": 6176 }, { "epoch": 0.01, "learning_rate": 4.9995443179718914e-05, "loss": 1.1507, "step": 6177 }, { "epoch": 0.01, "learning_rate": 4.9995441679947084e-05, "loss": 0.4551, "step": 6178 }, { "epoch": 0.01, "learning_rate": 4.999544017992851e-05, "loss": 0.8058, "step": 6179 }, { "epoch": 0.01, "learning_rate": 4.9995438679663184e-05, "loss": 0.847, "step": 6180 }, { "epoch": 0.01, "learning_rate": 4.999543717915112e-05, "loss": 1.1359, "step": 6181 }, { "epoch": 0.01, "learning_rate": 4.999543567839231e-05, "loss": 0.873, "step": 6182 }, { "epoch": 0.01, "learning_rate": 4.999543417738676e-05, "loss": 0.9266, "step": 6183 }, { "epoch": 0.01, "learning_rate": 4.9995432676134455e-05, "loss": 0.8686, "step": 6184 }, { "epoch": 0.01, "learning_rate": 4.999543117463542e-05, "loss": 1.9153, "step": 6185 }, { "epoch": 0.01, "learning_rate": 4.9995429672889626e-05, "loss": 1.4137, "step": 6186 }, { "epoch": 0.01, "learning_rate": 4.99954281708971e-05, "loss": 1.2259, "step": 6187 }, { "epoch": 0.01, "learning_rate": 4.999542666865782e-05, "loss": 1.1254, "step": 6188 }, { "epoch": 0.01, "learning_rate": 4.9995425166171805e-05, "loss": 1.0492, "step": 6189 }, { "epoch": 0.01, "learning_rate": 4.999542366343903e-05, "loss": 1.0931, "step": 6190 }, { "epoch": 0.01, "learning_rate": 4.999542216045953e-05, "loss": 1.1489, "step": 6191 }, { "epoch": 0.01, "learning_rate": 4.9995420657233274e-05, "loss": 1.3093, "step": 6192 }, { "epoch": 0.01, "learning_rate": 4.999541915376028e-05, "loss": 1.1749, "step": 6193 }, { "epoch": 0.01, "learning_rate": 4.999541765004053e-05, "loss": 1.2643, "step": 6194 }, { "epoch": 0.01, "learning_rate": 4.999541614607405e-05, "loss": 1.4281, "step": 6195 }, { "epoch": 0.01, "learning_rate": 4.9995414641860816e-05, "loss": 1.284, "step": 6196 }, { "epoch": 0.01, "learning_rate": 4.999541313740085e-05, "loss": 1.1549, "step": 6197 }, { "epoch": 0.01, "learning_rate": 4.999541163269412e-05, "loss": 1.0731, "step": 6198 }, { "epoch": 0.01, "learning_rate": 4.999541012774066e-05, "loss": 0.891, "step": 6199 }, { "epoch": 0.01, "learning_rate": 4.999540862254046e-05, "loss": 0.7303, "step": 6200 }, { "epoch": 0.01, "learning_rate": 4.999540711709351e-05, "loss": 0.8618, "step": 6201 }, { "epoch": 0.01, "learning_rate": 4.999540561139981e-05, "loss": 0.6789, "step": 6202 }, { "epoch": 0.01, "learning_rate": 4.999540410545937e-05, "loss": 0.9251, "step": 6203 }, { "epoch": 0.01, "learning_rate": 4.999540259927219e-05, "loss": 0.9536, "step": 6204 }, { "epoch": 0.01, "learning_rate": 4.9995401092838255e-05, "loss": 0.9837, "step": 6205 }, { "epoch": 0.01, "learning_rate": 4.999539958615759e-05, "loss": 0.7343, "step": 6206 }, { "epoch": 0.01, "learning_rate": 4.999539807923017e-05, "loss": 0.9371, "step": 6207 }, { "epoch": 0.01, "learning_rate": 4.9995396572056015e-05, "loss": 0.9017, "step": 6208 }, { "epoch": 0.01, "learning_rate": 4.999539506463511e-05, "loss": 0.9413, "step": 6209 }, { "epoch": 0.01, "learning_rate": 4.999539355696746e-05, "loss": 1.1699, "step": 6210 }, { "epoch": 0.01, "learning_rate": 4.999539204905307e-05, "loss": 1.277, "step": 6211 }, { "epoch": 0.01, "learning_rate": 4.9995390540891935e-05, "loss": 1.2824, "step": 6212 }, { "epoch": 0.01, "learning_rate": 4.9995389032484055e-05, "loss": 1.1497, "step": 6213 }, { "epoch": 0.01, "learning_rate": 4.999538752382943e-05, "loss": 1.1358, "step": 6214 }, { "epoch": 0.01, "learning_rate": 4.999538601492806e-05, "loss": 0.8836, "step": 6215 }, { "epoch": 0.01, "learning_rate": 4.999538450577995e-05, "loss": 0.3426, "step": 6216 }, { "epoch": 0.01, "learning_rate": 4.99953829963851e-05, "loss": 0.9849, "step": 6217 }, { "epoch": 0.01, "learning_rate": 4.99953814867435e-05, "loss": 0.6789, "step": 6218 }, { "epoch": 0.01, "learning_rate": 4.999537997685515e-05, "loss": 0.6195, "step": 6219 }, { "epoch": 0.01, "learning_rate": 4.9995378466720066e-05, "loss": 0.9728, "step": 6220 }, { "epoch": 0.01, "learning_rate": 4.999537695633823e-05, "loss": 1.2863, "step": 6221 }, { "epoch": 0.01, "learning_rate": 4.9995375445709656e-05, "loss": 1.1435, "step": 6222 }, { "epoch": 0.01, "learning_rate": 4.999537393483434e-05, "loss": 1.0346, "step": 6223 }, { "epoch": 0.01, "learning_rate": 4.999537242371228e-05, "loss": 1.0292, "step": 6224 }, { "epoch": 0.01, "learning_rate": 4.9995370912343474e-05, "loss": 1.1251, "step": 6225 }, { "epoch": 0.01, "learning_rate": 4.999536940072793e-05, "loss": 1.1762, "step": 6226 }, { "epoch": 0.01, "learning_rate": 4.999536788886563e-05, "loss": 1.1432, "step": 6227 }, { "epoch": 0.01, "learning_rate": 4.999536637675659e-05, "loss": 2.3195, "step": 6228 }, { "epoch": 0.01, "learning_rate": 4.999536486440081e-05, "loss": 1.9143, "step": 6229 }, { "epoch": 0.01, "learning_rate": 4.999536335179829e-05, "loss": 1.2253, "step": 6230 }, { "epoch": 0.01, "learning_rate": 4.999536183894902e-05, "loss": 1.1785, "step": 6231 }, { "epoch": 0.01, "learning_rate": 4.999536032585301e-05, "loss": 1.0539, "step": 6232 }, { "epoch": 0.01, "learning_rate": 4.9995358812510254e-05, "loss": 0.9327, "step": 6233 }, { "epoch": 0.01, "learning_rate": 4.9995357298920754e-05, "loss": 0.5802, "step": 6234 }, { "epoch": 0.01, "learning_rate": 4.999535578508451e-05, "loss": 1.1302, "step": 6235 }, { "epoch": 0.01, "learning_rate": 4.9995354271001524e-05, "loss": 1.0786, "step": 6236 }, { "epoch": 0.01, "learning_rate": 4.999535275667179e-05, "loss": 1.0324, "step": 6237 }, { "epoch": 0.01, "learning_rate": 4.999535124209532e-05, "loss": 1.0391, "step": 6238 }, { "epoch": 0.01, "learning_rate": 4.9995349727272104e-05, "loss": 0.9431, "step": 6239 }, { "epoch": 0.01, "learning_rate": 4.999534821220214e-05, "loss": 0.9996, "step": 6240 }, { "epoch": 0.01, "learning_rate": 4.999534669688544e-05, "loss": 1.2086, "step": 6241 }, { "epoch": 0.01, "learning_rate": 4.999534518132199e-05, "loss": 1.4165, "step": 6242 }, { "epoch": 0.01, "learning_rate": 4.99953436655118e-05, "loss": 1.1405, "step": 6243 }, { "epoch": 0.01, "learning_rate": 4.999534214945486e-05, "loss": 1.2725, "step": 6244 }, { "epoch": 0.01, "learning_rate": 4.999534063315118e-05, "loss": 1.151, "step": 6245 }, { "epoch": 0.01, "learning_rate": 4.999533911660076e-05, "loss": 1.1695, "step": 6246 }, { "epoch": 0.01, "learning_rate": 4.999533759980359e-05, "loss": 1.0715, "step": 6247 }, { "epoch": 0.01, "learning_rate": 4.9995336082759683e-05, "loss": 1.0895, "step": 6248 }, { "epoch": 0.01, "learning_rate": 4.999533456546903e-05, "loss": 1.3081, "step": 6249 }, { "epoch": 0.01, "learning_rate": 4.999533304793164e-05, "loss": 1.5056, "step": 6250 }, { "epoch": 0.01, "learning_rate": 4.99953315301475e-05, "loss": 1.3054, "step": 6251 }, { "epoch": 0.01, "learning_rate": 4.999533001211662e-05, "loss": 1.1802, "step": 6252 }, { "epoch": 0.01, "learning_rate": 4.999532849383899e-05, "loss": 1.2632, "step": 6253 }, { "epoch": 0.01, "learning_rate": 4.999532697531462e-05, "loss": 1.1067, "step": 6254 }, { "epoch": 0.01, "learning_rate": 4.999532545654351e-05, "loss": 1.1343, "step": 6255 }, { "epoch": 0.01, "learning_rate": 4.999532393752565e-05, "loss": 1.2952, "step": 6256 }, { "epoch": 0.01, "learning_rate": 4.999532241826105e-05, "loss": 1.4313, "step": 6257 }, { "epoch": 0.01, "learning_rate": 4.999532089874971e-05, "loss": 1.7484, "step": 6258 }, { "epoch": 0.01, "learning_rate": 4.999531937899162e-05, "loss": 1.668, "step": 6259 }, { "epoch": 0.01, "learning_rate": 4.99953178589868e-05, "loss": 1.3347, "step": 6260 }, { "epoch": 0.01, "learning_rate": 4.999531633873522e-05, "loss": 1.0661, "step": 6261 }, { "epoch": 0.01, "learning_rate": 4.999531481823691e-05, "loss": 0.9182, "step": 6262 }, { "epoch": 0.01, "learning_rate": 4.999531329749185e-05, "loss": 1.0667, "step": 6263 }, { "epoch": 0.01, "learning_rate": 4.999531177650005e-05, "loss": 1.1956, "step": 6264 }, { "epoch": 0.01, "learning_rate": 4.99953102552615e-05, "loss": 1.3031, "step": 6265 }, { "epoch": 0.01, "learning_rate": 4.999530873377621e-05, "loss": 1.0669, "step": 6266 }, { "epoch": 0.01, "learning_rate": 4.9995307212044186e-05, "loss": 1.2452, "step": 6267 }, { "epoch": 0.01, "learning_rate": 4.999530569006541e-05, "loss": 1.4003, "step": 6268 }, { "epoch": 0.01, "learning_rate": 4.9995304167839893e-05, "loss": 1.3337, "step": 6269 }, { "epoch": 0.01, "learning_rate": 4.999530264536763e-05, "loss": 1.2999, "step": 6270 }, { "epoch": 0.01, "learning_rate": 4.9995301122648624e-05, "loss": 1.2661, "step": 6271 }, { "epoch": 0.01, "learning_rate": 4.999529959968288e-05, "loss": 0.9675, "step": 6272 }, { "epoch": 0.01, "learning_rate": 4.9995298076470385e-05, "loss": 1.1439, "step": 6273 }, { "epoch": 0.01, "learning_rate": 4.9995296553011157e-05, "loss": 1.282, "step": 6274 }, { "epoch": 0.01, "learning_rate": 4.999529502930518e-05, "loss": 1.2966, "step": 6275 }, { "epoch": 0.01, "learning_rate": 4.999529350535246e-05, "loss": 1.1969, "step": 6276 }, { "epoch": 0.01, "learning_rate": 4.9995291981153e-05, "loss": 1.2212, "step": 6277 }, { "epoch": 0.01, "learning_rate": 4.999529045670679e-05, "loss": 0.8573, "step": 6278 }, { "epoch": 0.01, "learning_rate": 4.9995288932013845e-05, "loss": 1.2187, "step": 6279 }, { "epoch": 0.01, "learning_rate": 4.999528740707415e-05, "loss": 1.0123, "step": 6280 }, { "epoch": 0.01, "learning_rate": 4.999528588188772e-05, "loss": 1.3662, "step": 6281 }, { "epoch": 0.01, "learning_rate": 4.999528435645454e-05, "loss": 1.915, "step": 6282 }, { "epoch": 0.01, "learning_rate": 4.999528283077463e-05, "loss": 0.9141, "step": 6283 }, { "epoch": 0.01, "learning_rate": 4.9995281304847964e-05, "loss": 1.4028, "step": 6284 }, { "epoch": 0.01, "learning_rate": 4.999527977867455e-05, "loss": 1.143, "step": 6285 }, { "epoch": 0.01, "learning_rate": 4.9995278252254406e-05, "loss": 1.1971, "step": 6286 }, { "epoch": 0.01, "learning_rate": 4.999527672558751e-05, "loss": 1.0219, "step": 6287 }, { "epoch": 0.01, "learning_rate": 4.999527519867388e-05, "loss": 0.7101, "step": 6288 }, { "epoch": 0.01, "learning_rate": 4.9995273671513496e-05, "loss": 0.8893, "step": 6289 }, { "epoch": 0.01, "learning_rate": 4.999527214410639e-05, "loss": 1.2082, "step": 6290 }, { "epoch": 0.01, "learning_rate": 4.999527061645252e-05, "loss": 1.1676, "step": 6291 }, { "epoch": 0.01, "learning_rate": 4.999526908855191e-05, "loss": 0.8196, "step": 6292 }, { "epoch": 0.01, "learning_rate": 4.999526756040457e-05, "loss": 1.2572, "step": 6293 }, { "epoch": 0.01, "learning_rate": 4.999526603201048e-05, "loss": 1.1991, "step": 6294 }, { "epoch": 0.01, "learning_rate": 4.9995264503369645e-05, "loss": 1.1967, "step": 6295 }, { "epoch": 0.01, "learning_rate": 4.9995262974482066e-05, "loss": 0.994, "step": 6296 }, { "epoch": 0.01, "learning_rate": 4.9995261445347744e-05, "loss": 1.1037, "step": 6297 }, { "epoch": 0.01, "learning_rate": 4.9995259915966684e-05, "loss": 1.2701, "step": 6298 }, { "epoch": 0.01, "learning_rate": 4.999525838633888e-05, "loss": 0.863, "step": 6299 }, { "epoch": 0.01, "learning_rate": 4.999525685646433e-05, "loss": 1.1585, "step": 6300 }, { "epoch": 0.01, "learning_rate": 4.999525532634304e-05, "loss": 1.436, "step": 6301 }, { "epoch": 0.01, "learning_rate": 4.999525379597501e-05, "loss": 0.9212, "step": 6302 }, { "epoch": 0.01, "learning_rate": 4.999525226536024e-05, "loss": 1.8257, "step": 6303 }, { "epoch": 0.01, "learning_rate": 4.9995250734498713e-05, "loss": 0.5991, "step": 6304 }, { "epoch": 0.01, "learning_rate": 4.999524920339046e-05, "loss": 1.1361, "step": 6305 }, { "epoch": 0.01, "learning_rate": 4.999524767203545e-05, "loss": 1.1269, "step": 6306 }, { "epoch": 0.01, "learning_rate": 4.999524614043371e-05, "loss": 1.1514, "step": 6307 }, { "epoch": 0.01, "learning_rate": 4.999524460858522e-05, "loss": 1.3033, "step": 6308 }, { "epoch": 0.01, "learning_rate": 4.999524307648999e-05, "loss": 0.9704, "step": 6309 }, { "epoch": 0.01, "learning_rate": 4.999524154414802e-05, "loss": 1.1856, "step": 6310 }, { "epoch": 0.01, "learning_rate": 4.9995240011559304e-05, "loss": 1.4247, "step": 6311 }, { "epoch": 0.01, "learning_rate": 4.999523847872385e-05, "loss": 1.0136, "step": 6312 }, { "epoch": 0.01, "learning_rate": 4.999523694564165e-05, "loss": 1.3243, "step": 6313 }, { "epoch": 0.01, "learning_rate": 4.99952354123127e-05, "loss": 0.9522, "step": 6314 }, { "epoch": 0.01, "learning_rate": 4.999523387873701e-05, "loss": 1.2351, "step": 6315 }, { "epoch": 0.01, "learning_rate": 4.999523234491459e-05, "loss": 1.0311, "step": 6316 }, { "epoch": 0.01, "learning_rate": 4.999523081084542e-05, "loss": 1.3266, "step": 6317 }, { "epoch": 0.01, "learning_rate": 4.999522927652951e-05, "loss": 1.2291, "step": 6318 }, { "epoch": 0.01, "learning_rate": 4.999522774196686e-05, "loss": 1.4014, "step": 6319 }, { "epoch": 0.01, "learning_rate": 4.9995226207157455e-05, "loss": 1.4934, "step": 6320 }, { "epoch": 0.01, "learning_rate": 4.9995224672101315e-05, "loss": 1.2602, "step": 6321 }, { "epoch": 0.01, "learning_rate": 4.999522313679843e-05, "loss": 1.2214, "step": 6322 }, { "epoch": 0.01, "learning_rate": 4.999522160124881e-05, "loss": 1.1039, "step": 6323 }, { "epoch": 0.01, "learning_rate": 4.9995220065452445e-05, "loss": 1.1839, "step": 6324 }, { "epoch": 0.01, "learning_rate": 4.9995218529409336e-05, "loss": 1.3264, "step": 6325 }, { "epoch": 0.01, "learning_rate": 4.999521699311949e-05, "loss": 1.4254, "step": 6326 }, { "epoch": 0.01, "learning_rate": 4.999521545658289e-05, "loss": 0.9513, "step": 6327 }, { "epoch": 0.01, "learning_rate": 4.9995213919799556e-05, "loss": 1.7214, "step": 6328 }, { "epoch": 0.01, "learning_rate": 4.999521238276947e-05, "loss": 2.0432, "step": 6329 }, { "epoch": 0.01, "learning_rate": 4.9995210845492654e-05, "loss": 1.57, "step": 6330 }, { "epoch": 0.01, "learning_rate": 4.9995209307969094e-05, "loss": 1.2588, "step": 6331 }, { "epoch": 0.01, "learning_rate": 4.9995207770198796e-05, "loss": 1.3624, "step": 6332 }, { "epoch": 0.01, "learning_rate": 4.999520623218175e-05, "loss": 1.1314, "step": 6333 }, { "epoch": 0.01, "learning_rate": 4.9995204693917955e-05, "loss": 0.9187, "step": 6334 }, { "epoch": 0.01, "learning_rate": 4.999520315540742e-05, "loss": 1.0166, "step": 6335 }, { "epoch": 0.01, "learning_rate": 4.999520161665016e-05, "loss": 0.987, "step": 6336 }, { "epoch": 0.01, "learning_rate": 4.999520007764614e-05, "loss": 1.1031, "step": 6337 }, { "epoch": 0.01, "learning_rate": 4.9995198538395385e-05, "loss": 1.2838, "step": 6338 }, { "epoch": 0.01, "learning_rate": 4.999519699889788e-05, "loss": 0.6294, "step": 6339 }, { "epoch": 0.01, "learning_rate": 4.999519545915364e-05, "loss": 0.9347, "step": 6340 }, { "epoch": 0.01, "learning_rate": 4.999519391916266e-05, "loss": 0.9834, "step": 6341 }, { "epoch": 0.01, "learning_rate": 4.999519237892493e-05, "loss": 1.083, "step": 6342 }, { "epoch": 0.01, "learning_rate": 4.999519083844047e-05, "loss": 1.3272, "step": 6343 }, { "epoch": 0.01, "learning_rate": 4.9995189297709254e-05, "loss": 1.0224, "step": 6344 }, { "epoch": 0.01, "learning_rate": 4.99951877567313e-05, "loss": 1.2144, "step": 6345 }, { "epoch": 0.01, "learning_rate": 4.999518621550661e-05, "loss": 1.216, "step": 6346 }, { "epoch": 0.01, "learning_rate": 4.999518467403518e-05, "loss": 1.0545, "step": 6347 }, { "epoch": 0.01, "learning_rate": 4.9995183132317e-05, "loss": 1.6558, "step": 6348 }, { "epoch": 0.01, "learning_rate": 4.999518159035208e-05, "loss": 1.3072, "step": 6349 }, { "epoch": 0.01, "learning_rate": 4.9995180048140424e-05, "loss": 1.5042, "step": 6350 }, { "epoch": 0.01, "learning_rate": 4.9995178505682015e-05, "loss": 1.491, "step": 6351 }, { "epoch": 0.01, "learning_rate": 4.999517696297688e-05, "loss": 1.4511, "step": 6352 }, { "epoch": 0.01, "learning_rate": 4.999517542002499e-05, "loss": 1.2005, "step": 6353 }, { "epoch": 0.01, "learning_rate": 4.999517387682636e-05, "loss": 0.913, "step": 6354 }, { "epoch": 0.01, "learning_rate": 4.999517233338099e-05, "loss": 0.9846, "step": 6355 }, { "epoch": 0.01, "learning_rate": 4.999517078968888e-05, "loss": 0.995, "step": 6356 }, { "epoch": 0.01, "learning_rate": 4.999516924575002e-05, "loss": 1.0608, "step": 6357 }, { "epoch": 0.01, "learning_rate": 4.9995167701564425e-05, "loss": 0.9245, "step": 6358 }, { "epoch": 0.01, "learning_rate": 4.999516615713209e-05, "loss": 0.947, "step": 6359 }, { "epoch": 0.01, "learning_rate": 4.9995164612453014e-05, "loss": 0.8862, "step": 6360 }, { "epoch": 0.01, "learning_rate": 4.99951630675272e-05, "loss": 0.9554, "step": 6361 }, { "epoch": 0.01, "learning_rate": 4.9995161522354626e-05, "loss": 1.039, "step": 6362 }, { "epoch": 0.01, "learning_rate": 4.999515997693532e-05, "loss": 1.1183, "step": 6363 }, { "epoch": 0.01, "learning_rate": 4.999515843126928e-05, "loss": 1.1794, "step": 6364 }, { "epoch": 0.01, "learning_rate": 4.999515688535649e-05, "loss": 1.4328, "step": 6365 }, { "epoch": 0.01, "learning_rate": 4.999515533919696e-05, "loss": 0.9539, "step": 6366 }, { "epoch": 0.01, "learning_rate": 4.9995153792790696e-05, "loss": 0.9478, "step": 6367 }, { "epoch": 0.01, "learning_rate": 4.999515224613768e-05, "loss": 0.9248, "step": 6368 }, { "epoch": 0.01, "learning_rate": 4.9995150699237924e-05, "loss": 0.7609, "step": 6369 }, { "epoch": 0.01, "learning_rate": 4.999514915209143e-05, "loss": 1.0044, "step": 6370 }, { "epoch": 0.01, "learning_rate": 4.999514760469819e-05, "loss": 1.1496, "step": 6371 }, { "epoch": 0.01, "learning_rate": 4.999514605705822e-05, "loss": 0.828, "step": 6372 }, { "epoch": 0.01, "learning_rate": 4.999514450917149e-05, "loss": 1.1735, "step": 6373 }, { "epoch": 0.01, "learning_rate": 4.999514296103803e-05, "loss": 1.0782, "step": 6374 }, { "epoch": 0.01, "learning_rate": 4.9995141412657834e-05, "loss": 1.0884, "step": 6375 }, { "epoch": 0.01, "learning_rate": 4.9995139864030884e-05, "loss": 0.8416, "step": 6376 }, { "epoch": 0.01, "learning_rate": 4.99951383151572e-05, "loss": 1.1527, "step": 6377 }, { "epoch": 0.01, "learning_rate": 4.9995136766036774e-05, "loss": 1.0447, "step": 6378 }, { "epoch": 0.01, "learning_rate": 4.99951352166696e-05, "loss": 1.3278, "step": 6379 }, { "epoch": 0.01, "learning_rate": 4.9995133667055694e-05, "loss": 0.9146, "step": 6380 }, { "epoch": 0.01, "learning_rate": 4.999513211719504e-05, "loss": 1.1876, "step": 6381 }, { "epoch": 0.01, "learning_rate": 4.9995130567087644e-05, "loss": 1.1195, "step": 6382 }, { "epoch": 0.01, "learning_rate": 4.999512901673351e-05, "loss": 0.8774, "step": 6383 }, { "epoch": 0.01, "learning_rate": 4.999512746613263e-05, "loss": 1.2939, "step": 6384 }, { "epoch": 0.01, "learning_rate": 4.999512591528501e-05, "loss": 1.1973, "step": 6385 }, { "epoch": 0.01, "learning_rate": 4.999512436419066e-05, "loss": 1.1773, "step": 6386 }, { "epoch": 0.01, "learning_rate": 4.999512281284956e-05, "loss": 1.3966, "step": 6387 }, { "epoch": 0.01, "learning_rate": 4.999512126126171e-05, "loss": 0.8522, "step": 6388 }, { "epoch": 0.01, "learning_rate": 4.999511970942713e-05, "loss": 1.1953, "step": 6389 }, { "epoch": 0.01, "learning_rate": 4.999511815734581e-05, "loss": 1.2966, "step": 6390 }, { "epoch": 0.01, "learning_rate": 4.999511660501775e-05, "loss": 1.1662, "step": 6391 }, { "epoch": 0.01, "learning_rate": 4.999511505244294e-05, "loss": 1.4128, "step": 6392 }, { "epoch": 0.01, "learning_rate": 4.999511349962139e-05, "loss": 1.075, "step": 6393 }, { "epoch": 0.01, "learning_rate": 4.99951119465531e-05, "loss": 1.1241, "step": 6394 }, { "epoch": 0.01, "learning_rate": 4.999511039323807e-05, "loss": 1.1761, "step": 6395 }, { "epoch": 0.01, "learning_rate": 4.99951088396763e-05, "loss": 1.3527, "step": 6396 }, { "epoch": 0.01, "learning_rate": 4.999510728586778e-05, "loss": 1.2662, "step": 6397 }, { "epoch": 0.01, "learning_rate": 4.999510573181253e-05, "loss": 0.9103, "step": 6398 }, { "epoch": 0.01, "learning_rate": 4.999510417751053e-05, "loss": 0.9209, "step": 6399 }, { "epoch": 0.01, "learning_rate": 4.9995102622961796e-05, "loss": 0.6531, "step": 6400 }, { "epoch": 0.01, "learning_rate": 4.999510106816632e-05, "loss": 0.145, "step": 6401 }, { "epoch": 0.01, "learning_rate": 4.99950995131241e-05, "loss": 0.0826, "step": 6402 }, { "epoch": 0.01, "learning_rate": 4.999509795783515e-05, "loss": 0.0694, "step": 6403 }, { "epoch": 0.01, "learning_rate": 4.999509640229944e-05, "loss": 0.2337, "step": 6404 }, { "epoch": 0.01, "learning_rate": 4.9995094846517e-05, "loss": 0.2951, "step": 6405 }, { "epoch": 0.01, "learning_rate": 4.9995093290487814e-05, "loss": 1.0009, "step": 6406 }, { "epoch": 0.01, "learning_rate": 4.999509173421189e-05, "loss": 0.9331, "step": 6407 }, { "epoch": 0.01, "learning_rate": 4.999509017768923e-05, "loss": 1.3308, "step": 6408 }, { "epoch": 0.01, "learning_rate": 4.9995088620919825e-05, "loss": 1.3025, "step": 6409 }, { "epoch": 0.01, "learning_rate": 4.999508706390368e-05, "loss": 0.9508, "step": 6410 }, { "epoch": 0.01, "learning_rate": 4.999508550664079e-05, "loss": 0.854, "step": 6411 }, { "epoch": 0.01, "learning_rate": 4.999508394913116e-05, "loss": 0.9445, "step": 6412 }, { "epoch": 0.01, "learning_rate": 4.9995082391374795e-05, "loss": 1.2569, "step": 6413 }, { "epoch": 0.01, "learning_rate": 4.999508083337168e-05, "loss": 1.3479, "step": 6414 }, { "epoch": 0.01, "learning_rate": 4.9995079275121825e-05, "loss": 1.3147, "step": 6415 }, { "epoch": 0.01, "learning_rate": 4.999507771662524e-05, "loss": 1.3282, "step": 6416 }, { "epoch": 0.01, "learning_rate": 4.9995076157881906e-05, "loss": 1.3538, "step": 6417 }, { "epoch": 0.01, "learning_rate": 4.999507459889183e-05, "loss": 1.3551, "step": 6418 }, { "epoch": 0.01, "learning_rate": 4.999507303965502e-05, "loss": 1.2779, "step": 6419 }, { "epoch": 0.01, "learning_rate": 4.999507148017146e-05, "loss": 1.3151, "step": 6420 }, { "epoch": 0.01, "learning_rate": 4.999506992044116e-05, "loss": 1.1586, "step": 6421 }, { "epoch": 0.01, "learning_rate": 4.9995068360464125e-05, "loss": 0.8012, "step": 6422 }, { "epoch": 0.01, "learning_rate": 4.999506680024034e-05, "loss": 0.5621, "step": 6423 }, { "epoch": 0.01, "learning_rate": 4.999506523976982e-05, "loss": 0.6656, "step": 6424 }, { "epoch": 0.01, "learning_rate": 4.999506367905257e-05, "loss": 0.6819, "step": 6425 }, { "epoch": 0.01, "learning_rate": 4.9995062118088564e-05, "loss": 0.9057, "step": 6426 }, { "epoch": 0.01, "learning_rate": 4.999506055687783e-05, "loss": 1.3069, "step": 6427 }, { "epoch": 0.01, "learning_rate": 4.999505899542034e-05, "loss": 0.8545, "step": 6428 }, { "epoch": 0.01, "learning_rate": 4.999505743371612e-05, "loss": 1.0369, "step": 6429 }, { "epoch": 0.01, "learning_rate": 4.999505587176516e-05, "loss": 1.0941, "step": 6430 }, { "epoch": 0.01, "learning_rate": 4.9995054309567455e-05, "loss": 1.3182, "step": 6431 }, { "epoch": 0.01, "learning_rate": 4.9995052747123006e-05, "loss": 1.4654, "step": 6432 }, { "epoch": 0.01, "learning_rate": 4.999505118443183e-05, "loss": 1.0973, "step": 6433 }, { "epoch": 0.01, "learning_rate": 4.99950496214939e-05, "loss": 0.9623, "step": 6434 }, { "epoch": 0.01, "learning_rate": 4.9995048058309236e-05, "loss": 1.0791, "step": 6435 }, { "epoch": 0.01, "learning_rate": 4.999504649487783e-05, "loss": 1.4668, "step": 6436 }, { "epoch": 0.01, "learning_rate": 4.9995044931199676e-05, "loss": 1.3924, "step": 6437 }, { "epoch": 0.01, "learning_rate": 4.999504336727478e-05, "loss": 0.6485, "step": 6438 }, { "epoch": 0.01, "learning_rate": 4.999504180310316e-05, "loss": 1.164, "step": 6439 }, { "epoch": 0.01, "learning_rate": 4.9995040238684785e-05, "loss": 1.101, "step": 6440 }, { "epoch": 0.01, "learning_rate": 4.999503867401968e-05, "loss": 1.0594, "step": 6441 }, { "epoch": 0.01, "learning_rate": 4.999503710910783e-05, "loss": 1.0415, "step": 6442 }, { "epoch": 0.01, "learning_rate": 4.9995035543949245e-05, "loss": 1.0195, "step": 6443 }, { "epoch": 0.01, "learning_rate": 4.999503397854391e-05, "loss": 0.9997, "step": 6444 }, { "epoch": 0.01, "learning_rate": 4.999503241289184e-05, "loss": 1.1615, "step": 6445 }, { "epoch": 0.01, "learning_rate": 4.999503084699303e-05, "loss": 0.9525, "step": 6446 }, { "epoch": 0.01, "learning_rate": 4.999502928084747e-05, "loss": 1.1436, "step": 6447 }, { "epoch": 0.01, "learning_rate": 4.999502771445518e-05, "loss": 0.9101, "step": 6448 }, { "epoch": 0.01, "learning_rate": 4.999502614781615e-05, "loss": 1.2262, "step": 6449 }, { "epoch": 0.01, "learning_rate": 4.9995024580930375e-05, "loss": 0.9769, "step": 6450 }, { "epoch": 0.01, "learning_rate": 4.9995023013797856e-05, "loss": 0.797, "step": 6451 }, { "epoch": 0.01, "learning_rate": 4.9995021446418614e-05, "loss": 1.1696, "step": 6452 }, { "epoch": 0.01, "learning_rate": 4.9995019878792614e-05, "loss": 1.226, "step": 6453 }, { "epoch": 0.01, "learning_rate": 4.9995018310919877e-05, "loss": 1.2363, "step": 6454 }, { "epoch": 0.01, "learning_rate": 4.99950167428004e-05, "loss": 1.0814, "step": 6455 }, { "epoch": 0.01, "learning_rate": 4.999501517443419e-05, "loss": 1.03, "step": 6456 }, { "epoch": 0.01, "learning_rate": 4.999501360582123e-05, "loss": 1.0106, "step": 6457 }, { "epoch": 0.01, "learning_rate": 4.9995012036961534e-05, "loss": 0.7034, "step": 6458 }, { "epoch": 0.01, "learning_rate": 4.9995010467855097e-05, "loss": 0.5473, "step": 6459 }, { "epoch": 0.01, "learning_rate": 4.999500889850193e-05, "loss": 0.5888, "step": 6460 }, { "epoch": 0.01, "learning_rate": 4.999500732890201e-05, "loss": 0.2767, "step": 6461 }, { "epoch": 0.01, "learning_rate": 4.999500575905535e-05, "loss": 0.2416, "step": 6462 }, { "epoch": 0.01, "learning_rate": 4.9995004188961954e-05, "loss": 0.1992, "step": 6463 }, { "epoch": 0.01, "learning_rate": 4.9995002618621816e-05, "loss": 0.4523, "step": 6464 }, { "epoch": 0.01, "learning_rate": 4.999500104803494e-05, "loss": 0.262, "step": 6465 }, { "epoch": 0.01, "learning_rate": 4.999499947720132e-05, "loss": 0.177, "step": 6466 }, { "epoch": 0.01, "learning_rate": 4.9994997906120966e-05, "loss": 0.1315, "step": 6467 }, { "epoch": 0.01, "learning_rate": 4.9994996334793866e-05, "loss": 0.1851, "step": 6468 }, { "epoch": 0.01, "learning_rate": 4.999499476322003e-05, "loss": 0.3109, "step": 6469 }, { "epoch": 0.01, "learning_rate": 4.9994993191399453e-05, "loss": 0.8243, "step": 6470 }, { "epoch": 0.01, "learning_rate": 4.9994991619332134e-05, "loss": 1.1854, "step": 6471 }, { "epoch": 0.01, "learning_rate": 4.999499004701808e-05, "loss": 1.1112, "step": 6472 }, { "epoch": 0.01, "learning_rate": 4.9994988474457285e-05, "loss": 1.0755, "step": 6473 }, { "epoch": 0.01, "learning_rate": 4.999498690164974e-05, "loss": 1.141, "step": 6474 }, { "epoch": 0.01, "learning_rate": 4.9994985328595465e-05, "loss": 1.4014, "step": 6475 }, { "epoch": 0.01, "learning_rate": 4.9994983755294446e-05, "loss": 1.1114, "step": 6476 }, { "epoch": 0.01, "learning_rate": 4.999498218174668e-05, "loss": 1.0627, "step": 6477 }, { "epoch": 0.01, "learning_rate": 4.999498060795219e-05, "loss": 1.2869, "step": 6478 }, { "epoch": 0.01, "learning_rate": 4.999497903391096e-05, "loss": 1.2929, "step": 6479 }, { "epoch": 0.01, "learning_rate": 4.9994977459622976e-05, "loss": 1.0122, "step": 6480 }, { "epoch": 0.01, "learning_rate": 4.999497588508826e-05, "loss": 0.7117, "step": 6481 }, { "epoch": 0.01, "learning_rate": 4.99949743103068e-05, "loss": 0.7217, "step": 6482 }, { "epoch": 0.01, "learning_rate": 4.999497273527861e-05, "loss": 1.2543, "step": 6483 }, { "epoch": 0.01, "learning_rate": 4.999497116000367e-05, "loss": 1.2273, "step": 6484 }, { "epoch": 0.01, "learning_rate": 4.9994969584481994e-05, "loss": 1.0454, "step": 6485 }, { "epoch": 0.01, "learning_rate": 4.9994968008713574e-05, "loss": 1.1889, "step": 6486 }, { "epoch": 0.01, "learning_rate": 4.999496643269842e-05, "loss": 1.0451, "step": 6487 }, { "epoch": 0.01, "learning_rate": 4.9994964856436524e-05, "loss": 1.039, "step": 6488 }, { "epoch": 0.01, "learning_rate": 4.9994963279927886e-05, "loss": 0.8729, "step": 6489 }, { "epoch": 0.01, "learning_rate": 4.999496170317252e-05, "loss": 0.7305, "step": 6490 }, { "epoch": 0.01, "learning_rate": 4.99949601261704e-05, "loss": 1.1235, "step": 6491 }, { "epoch": 0.01, "learning_rate": 4.999495854892154e-05, "loss": 1.2587, "step": 6492 }, { "epoch": 0.01, "learning_rate": 4.999495697142595e-05, "loss": 1.1407, "step": 6493 }, { "epoch": 0.01, "learning_rate": 4.9994955393683616e-05, "loss": 1.2415, "step": 6494 }, { "epoch": 0.01, "learning_rate": 4.999495381569455e-05, "loss": 1.3505, "step": 6495 }, { "epoch": 0.01, "learning_rate": 4.999495223745873e-05, "loss": 1.1493, "step": 6496 }, { "epoch": 0.01, "learning_rate": 4.999495065897618e-05, "loss": 1.2276, "step": 6497 }, { "epoch": 0.01, "learning_rate": 4.999494908024688e-05, "loss": 0.9747, "step": 6498 }, { "epoch": 0.01, "learning_rate": 4.999494750127085e-05, "loss": 1.2284, "step": 6499 }, { "epoch": 0.01, "learning_rate": 4.999494592204808e-05, "loss": 1.2709, "step": 6500 }, { "epoch": 0.01, "eval_loss": 1.0758188962936401, "eval_runtime": 94.9288, "eval_samples_per_second": 14.59, "eval_steps_per_second": 3.655, "step": 6500 }, { "epoch": 0.01, "learning_rate": 4.9994944342578564e-05, "loss": 1.2316, "step": 6501 }, { "epoch": 0.01, "learning_rate": 4.999494276286232e-05, "loss": 1.0427, "step": 6502 }, { "epoch": 0.01, "learning_rate": 4.9994941182899326e-05, "loss": 1.5905, "step": 6503 }, { "epoch": 0.01, "learning_rate": 4.9994939602689595e-05, "loss": 0.9685, "step": 6504 }, { "epoch": 0.01, "learning_rate": 4.9994938022233126e-05, "loss": 1.0835, "step": 6505 }, { "epoch": 0.01, "learning_rate": 4.999493644152991e-05, "loss": 1.1555, "step": 6506 }, { "epoch": 0.01, "learning_rate": 4.999493486057997e-05, "loss": 1.1553, "step": 6507 }, { "epoch": 0.01, "learning_rate": 4.9994933279383275e-05, "loss": 1.2677, "step": 6508 }, { "epoch": 0.01, "learning_rate": 4.999493169793985e-05, "loss": 1.3516, "step": 6509 }, { "epoch": 0.01, "learning_rate": 4.999493011624968e-05, "loss": 1.1805, "step": 6510 }, { "epoch": 0.01, "learning_rate": 4.999492853431278e-05, "loss": 1.0536, "step": 6511 }, { "epoch": 0.01, "learning_rate": 4.999492695212913e-05, "loss": 0.7601, "step": 6512 }, { "epoch": 0.01, "learning_rate": 4.9994925369698744e-05, "loss": 1.234, "step": 6513 }, { "epoch": 0.01, "learning_rate": 4.999492378702162e-05, "loss": 1.1395, "step": 6514 }, { "epoch": 0.01, "learning_rate": 4.9994922204097764e-05, "loss": 1.0047, "step": 6515 }, { "epoch": 0.01, "learning_rate": 4.999492062092716e-05, "loss": 1.2291, "step": 6516 }, { "epoch": 0.01, "learning_rate": 4.9994919037509814e-05, "loss": 1.1659, "step": 6517 }, { "epoch": 0.01, "learning_rate": 4.999491745384573e-05, "loss": 1.0848, "step": 6518 }, { "epoch": 0.01, "learning_rate": 4.999491586993491e-05, "loss": 1.1353, "step": 6519 }, { "epoch": 0.01, "learning_rate": 4.999491428577735e-05, "loss": 0.9852, "step": 6520 }, { "epoch": 0.01, "learning_rate": 4.9994912701373046e-05, "loss": 1.0104, "step": 6521 }, { "epoch": 0.01, "learning_rate": 4.999491111672201e-05, "loss": 1.2367, "step": 6522 }, { "epoch": 0.01, "learning_rate": 4.999490953182423e-05, "loss": 1.0964, "step": 6523 }, { "epoch": 0.01, "learning_rate": 4.9994907946679716e-05, "loss": 0.6671, "step": 6524 }, { "epoch": 0.01, "learning_rate": 4.9994906361288454e-05, "loss": 1.2634, "step": 6525 }, { "epoch": 0.01, "learning_rate": 4.999490477565046e-05, "loss": 1.1709, "step": 6526 }, { "epoch": 0.01, "learning_rate": 4.999490318976573e-05, "loss": 0.6647, "step": 6527 }, { "epoch": 0.01, "learning_rate": 4.999490160363425e-05, "loss": 0.5002, "step": 6528 }, { "epoch": 0.01, "learning_rate": 4.999490001725604e-05, "loss": 1.1257, "step": 6529 }, { "epoch": 0.01, "learning_rate": 4.999489843063109e-05, "loss": 1.2702, "step": 6530 }, { "epoch": 0.01, "learning_rate": 4.9994896843759396e-05, "loss": 1.1731, "step": 6531 }, { "epoch": 0.01, "learning_rate": 4.999489525664096e-05, "loss": 1.0127, "step": 6532 }, { "epoch": 0.01, "learning_rate": 4.999489366927579e-05, "loss": 1.1434, "step": 6533 }, { "epoch": 0.01, "learning_rate": 4.9994892081663886e-05, "loss": 1.0425, "step": 6534 }, { "epoch": 0.01, "learning_rate": 4.999489049380524e-05, "loss": 1.4964, "step": 6535 }, { "epoch": 0.01, "learning_rate": 4.999488890569986e-05, "loss": 1.5346, "step": 6536 }, { "epoch": 0.01, "learning_rate": 4.999488731734773e-05, "loss": 1.4415, "step": 6537 }, { "epoch": 0.01, "learning_rate": 4.9994885728748866e-05, "loss": 1.3784, "step": 6538 }, { "epoch": 0.01, "learning_rate": 4.9994884139903254e-05, "loss": 1.4162, "step": 6539 }, { "epoch": 0.01, "learning_rate": 4.999488255081092e-05, "loss": 1.6596, "step": 6540 }, { "epoch": 0.01, "learning_rate": 4.999488096147183e-05, "loss": 1.2537, "step": 6541 }, { "epoch": 0.01, "learning_rate": 4.999487937188601e-05, "loss": 0.9764, "step": 6542 }, { "epoch": 0.01, "learning_rate": 4.9994877782053453e-05, "loss": 1.1744, "step": 6543 }, { "epoch": 0.01, "learning_rate": 4.9994876191974155e-05, "loss": 1.3614, "step": 6544 }, { "epoch": 0.01, "learning_rate": 4.999487460164812e-05, "loss": 1.3881, "step": 6545 }, { "epoch": 0.01, "learning_rate": 4.999487301107534e-05, "loss": 1.0576, "step": 6546 }, { "epoch": 0.01, "learning_rate": 4.999487142025583e-05, "loss": 1.1086, "step": 6547 }, { "epoch": 0.01, "learning_rate": 4.9994869829189575e-05, "loss": 1.1733, "step": 6548 }, { "epoch": 0.01, "learning_rate": 4.999486823787658e-05, "loss": 1.1962, "step": 6549 }, { "epoch": 0.01, "learning_rate": 4.999486664631685e-05, "loss": 1.044, "step": 6550 }, { "epoch": 0.01, "learning_rate": 4.9994865054510374e-05, "loss": 0.9311, "step": 6551 }, { "epoch": 0.01, "learning_rate": 4.9994863462457164e-05, "loss": 0.9222, "step": 6552 }, { "epoch": 0.01, "learning_rate": 4.9994861870157216e-05, "loss": 1.154, "step": 6553 }, { "epoch": 0.01, "learning_rate": 4.999486027761053e-05, "loss": 1.0585, "step": 6554 }, { "epoch": 0.01, "learning_rate": 4.999485868481711e-05, "loss": 1.2625, "step": 6555 }, { "epoch": 0.01, "learning_rate": 4.999485709177695e-05, "loss": 1.243, "step": 6556 }, { "epoch": 0.01, "learning_rate": 4.999485549849004e-05, "loss": 1.2738, "step": 6557 }, { "epoch": 0.01, "learning_rate": 4.99948539049564e-05, "loss": 1.4934, "step": 6558 }, { "epoch": 0.01, "learning_rate": 4.999485231117602e-05, "loss": 1.1014, "step": 6559 }, { "epoch": 0.01, "learning_rate": 4.99948507171489e-05, "loss": 1.0812, "step": 6560 }, { "epoch": 0.01, "learning_rate": 4.999484912287504e-05, "loss": 1.0902, "step": 6561 }, { "epoch": 0.01, "learning_rate": 4.9994847528354446e-05, "loss": 1.2505, "step": 6562 }, { "epoch": 0.01, "learning_rate": 4.999484593358711e-05, "loss": 1.286, "step": 6563 }, { "epoch": 0.01, "learning_rate": 4.9994844338573047e-05, "loss": 1.432, "step": 6564 }, { "epoch": 0.01, "learning_rate": 4.999484274331223e-05, "loss": 1.2748, "step": 6565 }, { "epoch": 0.01, "learning_rate": 4.999484114780468e-05, "loss": 1.4985, "step": 6566 }, { "epoch": 0.01, "learning_rate": 4.999483955205039e-05, "loss": 1.5063, "step": 6567 }, { "epoch": 0.01, "learning_rate": 4.999483795604936e-05, "loss": 1.1407, "step": 6568 }, { "epoch": 0.01, "learning_rate": 4.9994836359801594e-05, "loss": 1.2549, "step": 6569 }, { "epoch": 0.01, "learning_rate": 4.999483476330709e-05, "loss": 1.1062, "step": 6570 }, { "epoch": 0.01, "learning_rate": 4.9994833166565846e-05, "loss": 1.0548, "step": 6571 }, { "epoch": 0.01, "learning_rate": 4.999483156957787e-05, "loss": 1.0383, "step": 6572 }, { "epoch": 0.01, "learning_rate": 4.999482997234315e-05, "loss": 1.208, "step": 6573 }, { "epoch": 0.01, "learning_rate": 4.999482837486169e-05, "loss": 0.8818, "step": 6574 }, { "epoch": 0.01, "learning_rate": 4.9994826777133494e-05, "loss": 0.8663, "step": 6575 }, { "epoch": 0.01, "learning_rate": 4.9994825179158554e-05, "loss": 1.0257, "step": 6576 }, { "epoch": 0.01, "learning_rate": 4.9994823580936884e-05, "loss": 0.9159, "step": 6577 }, { "epoch": 0.01, "learning_rate": 4.9994821982468476e-05, "loss": 1.2628, "step": 6578 }, { "epoch": 0.01, "learning_rate": 4.9994820383753324e-05, "loss": 1.2665, "step": 6579 }, { "epoch": 0.01, "learning_rate": 4.999481878479143e-05, "loss": 0.9341, "step": 6580 }, { "epoch": 0.01, "learning_rate": 4.999481718558281e-05, "loss": 1.3301, "step": 6581 }, { "epoch": 0.01, "learning_rate": 4.999481558612744e-05, "loss": 0.5321, "step": 6582 }, { "epoch": 0.01, "learning_rate": 4.999481398642534e-05, "loss": 0.5291, "step": 6583 }, { "epoch": 0.01, "learning_rate": 4.99948123864765e-05, "loss": 0.745, "step": 6584 }, { "epoch": 0.01, "learning_rate": 4.9994810786280924e-05, "loss": 1.0752, "step": 6585 }, { "epoch": 0.01, "learning_rate": 4.99948091858386e-05, "loss": 1.2102, "step": 6586 }, { "epoch": 0.01, "learning_rate": 4.999480758514954e-05, "loss": 0.5819, "step": 6587 }, { "epoch": 0.01, "learning_rate": 4.9994805984213746e-05, "loss": 0.9426, "step": 6588 }, { "epoch": 0.01, "learning_rate": 4.9994804383031215e-05, "loss": 1.0205, "step": 6589 }, { "epoch": 0.01, "learning_rate": 4.9994802781601946e-05, "loss": 1.0928, "step": 6590 }, { "epoch": 0.01, "learning_rate": 4.999480117992593e-05, "loss": 1.1653, "step": 6591 }, { "epoch": 0.01, "learning_rate": 4.999479957800319e-05, "loss": 1.3666, "step": 6592 }, { "epoch": 0.01, "learning_rate": 4.999479797583371e-05, "loss": 1.3309, "step": 6593 }, { "epoch": 0.01, "learning_rate": 4.999479637341748e-05, "loss": 0.9612, "step": 6594 }, { "epoch": 0.01, "learning_rate": 4.9994794770754515e-05, "loss": 1.2552, "step": 6595 }, { "epoch": 0.01, "learning_rate": 4.999479316784482e-05, "loss": 1.3941, "step": 6596 }, { "epoch": 0.01, "learning_rate": 4.999479156468838e-05, "loss": 1.3753, "step": 6597 }, { "epoch": 0.01, "learning_rate": 4.9994789961285204e-05, "loss": 1.4621, "step": 6598 }, { "epoch": 0.01, "learning_rate": 4.999478835763529e-05, "loss": 0.9614, "step": 6599 }, { "epoch": 0.01, "learning_rate": 4.9994786753738644e-05, "loss": 1.1641, "step": 6600 }, { "epoch": 0.01, "learning_rate": 4.9994785149595244e-05, "loss": 1.1776, "step": 6601 }, { "epoch": 0.01, "learning_rate": 4.999478354520513e-05, "loss": 1.2815, "step": 6602 }, { "epoch": 0.01, "learning_rate": 4.999478194056826e-05, "loss": 0.6628, "step": 6603 }, { "epoch": 0.01, "learning_rate": 4.9994780335684656e-05, "loss": 0.8115, "step": 6604 }, { "epoch": 0.01, "learning_rate": 4.9994778730554314e-05, "loss": 1.2254, "step": 6605 }, { "epoch": 0.01, "learning_rate": 4.999477712517723e-05, "loss": 1.3675, "step": 6606 }, { "epoch": 0.01, "learning_rate": 4.999477551955342e-05, "loss": 1.2994, "step": 6607 }, { "epoch": 0.01, "learning_rate": 4.999477391368286e-05, "loss": 0.9946, "step": 6608 }, { "epoch": 0.01, "learning_rate": 4.9994772307565566e-05, "loss": 1.0522, "step": 6609 }, { "epoch": 0.01, "learning_rate": 4.999477070120153e-05, "loss": 1.0744, "step": 6610 }, { "epoch": 0.01, "learning_rate": 4.9994769094590764e-05, "loss": 1.4907, "step": 6611 }, { "epoch": 0.01, "learning_rate": 4.999476748773326e-05, "loss": 1.18, "step": 6612 }, { "epoch": 0.01, "learning_rate": 4.9994765880629014e-05, "loss": 0.9152, "step": 6613 }, { "epoch": 0.01, "learning_rate": 4.9994764273278036e-05, "loss": 1.127, "step": 6614 }, { "epoch": 0.01, "learning_rate": 4.9994762665680314e-05, "loss": 1.1186, "step": 6615 }, { "epoch": 0.01, "learning_rate": 4.9994761057835855e-05, "loss": 1.0582, "step": 6616 }, { "epoch": 0.01, "learning_rate": 4.999475944974466e-05, "loss": 1.4079, "step": 6617 }, { "epoch": 0.01, "learning_rate": 4.9994757841406724e-05, "loss": 1.3987, "step": 6618 }, { "epoch": 0.01, "learning_rate": 4.999475623282205e-05, "loss": 1.1375, "step": 6619 }, { "epoch": 0.01, "learning_rate": 4.9994754623990645e-05, "loss": 1.0294, "step": 6620 }, { "epoch": 0.01, "learning_rate": 4.99947530149125e-05, "loss": 1.163, "step": 6621 }, { "epoch": 0.01, "learning_rate": 4.9994751405587616e-05, "loss": 0.1906, "step": 6622 }, { "epoch": 0.01, "learning_rate": 4.9994749796015995e-05, "loss": 0.6676, "step": 6623 }, { "epoch": 0.01, "learning_rate": 4.999474818619763e-05, "loss": 1.3446, "step": 6624 }, { "epoch": 0.01, "learning_rate": 4.9994746576132536e-05, "loss": 1.1012, "step": 6625 }, { "epoch": 0.01, "learning_rate": 4.9994744965820703e-05, "loss": 1.0001, "step": 6626 }, { "epoch": 0.01, "learning_rate": 4.999474335526213e-05, "loss": 1.0914, "step": 6627 }, { "epoch": 0.01, "learning_rate": 4.999474174445682e-05, "loss": 0.9416, "step": 6628 }, { "epoch": 0.01, "learning_rate": 4.999474013340477e-05, "loss": 1.1493, "step": 6629 }, { "epoch": 0.01, "learning_rate": 4.999473852210599e-05, "loss": 1.2614, "step": 6630 }, { "epoch": 0.01, "learning_rate": 4.999473691056046e-05, "loss": 1.3037, "step": 6631 }, { "epoch": 0.01, "learning_rate": 4.9994735298768206e-05, "loss": 0.8561, "step": 6632 }, { "epoch": 0.01, "learning_rate": 4.9994733686729205e-05, "loss": 1.146, "step": 6633 }, { "epoch": 0.01, "learning_rate": 4.9994732074443474e-05, "loss": 1.0051, "step": 6634 }, { "epoch": 0.01, "learning_rate": 4.9994730461911e-05, "loss": 1.1089, "step": 6635 }, { "epoch": 0.01, "learning_rate": 4.999472884913179e-05, "loss": 1.3316, "step": 6636 }, { "epoch": 0.01, "learning_rate": 4.9994727236105845e-05, "loss": 1.1779, "step": 6637 }, { "epoch": 0.01, "learning_rate": 4.999472562283316e-05, "loss": 1.2062, "step": 6638 }, { "epoch": 0.01, "learning_rate": 4.9994724009313734e-05, "loss": 1.2757, "step": 6639 }, { "epoch": 0.01, "learning_rate": 4.999472239554758e-05, "loss": 1.1214, "step": 6640 }, { "epoch": 0.01, "learning_rate": 4.9994720781534673e-05, "loss": 1.0394, "step": 6641 }, { "epoch": 0.01, "learning_rate": 4.9994719167275044e-05, "loss": 1.045, "step": 6642 }, { "epoch": 0.01, "learning_rate": 4.999471755276867e-05, "loss": 0.9549, "step": 6643 }, { "epoch": 0.01, "learning_rate": 4.999471593801556e-05, "loss": 0.9414, "step": 6644 }, { "epoch": 0.01, "learning_rate": 4.999471432301572e-05, "loss": 1.0907, "step": 6645 }, { "epoch": 0.01, "learning_rate": 4.9994712707769134e-05, "loss": 1.0651, "step": 6646 }, { "epoch": 0.01, "learning_rate": 4.999471109227582e-05, "loss": 0.9648, "step": 6647 }, { "epoch": 0.01, "learning_rate": 4.999470947653576e-05, "loss": 1.051, "step": 6648 }, { "epoch": 0.01, "learning_rate": 4.999470786054896e-05, "loss": 0.9602, "step": 6649 }, { "epoch": 0.01, "learning_rate": 4.9994706244315426e-05, "loss": 1.0035, "step": 6650 }, { "epoch": 0.01, "learning_rate": 4.999470462783516e-05, "loss": 0.6232, "step": 6651 }, { "epoch": 0.01, "learning_rate": 4.999470301110815e-05, "loss": 0.6789, "step": 6652 }, { "epoch": 0.01, "learning_rate": 4.9994701394134405e-05, "loss": 0.8334, "step": 6653 }, { "epoch": 0.01, "learning_rate": 4.999469977691392e-05, "loss": 1.5565, "step": 6654 }, { "epoch": 0.01, "learning_rate": 4.999469815944671e-05, "loss": 1.1051, "step": 6655 }, { "epoch": 0.01, "learning_rate": 4.9994696541732756e-05, "loss": 1.239, "step": 6656 }, { "epoch": 0.01, "learning_rate": 4.999469492377205e-05, "loss": 1.2687, "step": 6657 }, { "epoch": 0.01, "learning_rate": 4.9994693305564634e-05, "loss": 1.0522, "step": 6658 }, { "epoch": 0.01, "learning_rate": 4.999469168711046e-05, "loss": 1.0463, "step": 6659 }, { "epoch": 0.01, "learning_rate": 4.9994690068409556e-05, "loss": 1.0889, "step": 6660 }, { "epoch": 0.01, "learning_rate": 4.999468844946192e-05, "loss": 0.9888, "step": 6661 }, { "epoch": 0.01, "learning_rate": 4.999468683026754e-05, "loss": 1.3083, "step": 6662 }, { "epoch": 0.01, "learning_rate": 4.999468521082642e-05, "loss": 0.9751, "step": 6663 }, { "epoch": 0.01, "learning_rate": 4.999468359113857e-05, "loss": 0.9221, "step": 6664 }, { "epoch": 0.01, "learning_rate": 4.9994681971203985e-05, "loss": 0.7942, "step": 6665 }, { "epoch": 0.01, "learning_rate": 4.999468035102266e-05, "loss": 0.6858, "step": 6666 }, { "epoch": 0.01, "learning_rate": 4.99946787305946e-05, "loss": 1.0173, "step": 6667 }, { "epoch": 0.01, "learning_rate": 4.99946771099198e-05, "loss": 0.4216, "step": 6668 }, { "epoch": 0.01, "learning_rate": 4.999467548899826e-05, "loss": 0.2308, "step": 6669 }, { "epoch": 0.01, "learning_rate": 4.999467386782999e-05, "loss": 0.1515, "step": 6670 }, { "epoch": 0.01, "learning_rate": 4.999467224641498e-05, "loss": 0.0879, "step": 6671 }, { "epoch": 0.01, "learning_rate": 4.999467062475323e-05, "loss": 0.0501, "step": 6672 }, { "epoch": 0.01, "learning_rate": 4.999466900284474e-05, "loss": 0.0293, "step": 6673 }, { "epoch": 0.01, "learning_rate": 4.999466738068952e-05, "loss": 0.0257, "step": 6674 }, { "epoch": 0.01, "learning_rate": 4.999466575828756e-05, "loss": 0.0204, "step": 6675 }, { "epoch": 0.01, "learning_rate": 4.999466413563887e-05, "loss": 0.0162, "step": 6676 }, { "epoch": 0.01, "learning_rate": 4.999466251274344e-05, "loss": 0.0418, "step": 6677 }, { "epoch": 0.01, "learning_rate": 4.9994660889601274e-05, "loss": 0.0339, "step": 6678 }, { "epoch": 0.01, "learning_rate": 4.9994659266212364e-05, "loss": 0.0208, "step": 6679 }, { "epoch": 0.01, "learning_rate": 4.9994657642576724e-05, "loss": 0.0155, "step": 6680 }, { "epoch": 0.01, "learning_rate": 4.9994656018694346e-05, "loss": 0.0171, "step": 6681 }, { "epoch": 0.01, "learning_rate": 4.999465439456523e-05, "loss": 0.012, "step": 6682 }, { "epoch": 0.01, "learning_rate": 4.999465277018938e-05, "loss": 0.2595, "step": 6683 }, { "epoch": 0.01, "learning_rate": 4.999465114556679e-05, "loss": 0.4083, "step": 6684 }, { "epoch": 0.01, "learning_rate": 4.999464952069747e-05, "loss": 0.3142, "step": 6685 }, { "epoch": 0.01, "learning_rate": 4.9994647895581406e-05, "loss": 0.191, "step": 6686 }, { "epoch": 0.01, "learning_rate": 4.9994646270218604e-05, "loss": 0.1369, "step": 6687 }, { "epoch": 0.01, "learning_rate": 4.9994644644609065e-05, "loss": 0.0933, "step": 6688 }, { "epoch": 0.01, "learning_rate": 4.9994643018752796e-05, "loss": 0.0805, "step": 6689 }, { "epoch": 0.01, "learning_rate": 4.999464139264979e-05, "loss": 0.0519, "step": 6690 }, { "epoch": 0.01, "learning_rate": 4.9994639766300046e-05, "loss": 0.0392, "step": 6691 }, { "epoch": 0.01, "learning_rate": 4.9994638139703564e-05, "loss": 0.0669, "step": 6692 }, { "epoch": 0.01, "learning_rate": 4.9994636512860346e-05, "loss": 0.0817, "step": 6693 }, { "epoch": 0.01, "learning_rate": 4.999463488577039e-05, "loss": 0.0558, "step": 6694 }, { "epoch": 0.01, "learning_rate": 4.99946332584337e-05, "loss": 0.0373, "step": 6695 }, { "epoch": 0.01, "learning_rate": 4.999463163085027e-05, "loss": 0.0279, "step": 6696 }, { "epoch": 0.01, "learning_rate": 4.9994630003020106e-05, "loss": 0.0529, "step": 6697 }, { "epoch": 0.01, "learning_rate": 4.999462837494321e-05, "loss": 0.7594, "step": 6698 }, { "epoch": 0.01, "learning_rate": 4.999462674661957e-05, "loss": 1.082, "step": 6699 }, { "epoch": 0.01, "learning_rate": 4.99946251180492e-05, "loss": 1.4619, "step": 6700 }, { "epoch": 0.01, "learning_rate": 4.999462348923208e-05, "loss": 1.1149, "step": 6701 }, { "epoch": 0.01, "learning_rate": 4.999462186016825e-05, "loss": 1.2684, "step": 6702 }, { "epoch": 0.01, "learning_rate": 4.9994620230857656e-05, "loss": 0.9541, "step": 6703 }, { "epoch": 0.01, "learning_rate": 4.999461860130034e-05, "loss": 1.1064, "step": 6704 }, { "epoch": 0.01, "learning_rate": 4.999461697149629e-05, "loss": 1.3397, "step": 6705 }, { "epoch": 0.01, "learning_rate": 4.999461534144549e-05, "loss": 1.1285, "step": 6706 }, { "epoch": 0.01, "learning_rate": 4.9994613711147964e-05, "loss": 1.1078, "step": 6707 }, { "epoch": 0.01, "learning_rate": 4.9994612080603706e-05, "loss": 1.2667, "step": 6708 }, { "epoch": 0.01, "learning_rate": 4.9994610449812704e-05, "loss": 1.1437, "step": 6709 }, { "epoch": 0.01, "learning_rate": 4.9994608818774965e-05, "loss": 1.0543, "step": 6710 }, { "epoch": 0.01, "learning_rate": 4.9994607187490495e-05, "loss": 1.1706, "step": 6711 }, { "epoch": 0.01, "learning_rate": 4.999460555595928e-05, "loss": 1.1993, "step": 6712 }, { "epoch": 0.01, "learning_rate": 4.999460392418134e-05, "loss": 1.1689, "step": 6713 }, { "epoch": 0.01, "learning_rate": 4.9994602292156655e-05, "loss": 0.9531, "step": 6714 }, { "epoch": 0.01, "learning_rate": 4.9994600659885237e-05, "loss": 0.4351, "step": 6715 }, { "epoch": 0.01, "learning_rate": 4.999459902736709e-05, "loss": 0.3077, "step": 6716 }, { "epoch": 0.01, "learning_rate": 4.999459739460219e-05, "loss": 1.1547, "step": 6717 }, { "epoch": 0.01, "learning_rate": 4.999459576159057e-05, "loss": 1.0603, "step": 6718 }, { "epoch": 0.01, "learning_rate": 4.999459412833221e-05, "loss": 1.2145, "step": 6719 }, { "epoch": 0.01, "learning_rate": 4.999459249482711e-05, "loss": 1.0326, "step": 6720 }, { "epoch": 0.01, "learning_rate": 4.999459086107527e-05, "loss": 4.1597, "step": 6721 }, { "epoch": 0.01, "learning_rate": 4.9994589227076706e-05, "loss": 4.1984, "step": 6722 }, { "epoch": 0.01, "learning_rate": 4.9994587592831396e-05, "loss": 3.4404, "step": 6723 }, { "epoch": 0.01, "learning_rate": 4.9994585958339354e-05, "loss": 4.4833, "step": 6724 }, { "epoch": 0.01, "learning_rate": 4.999458432360058e-05, "loss": 4.0146, "step": 6725 }, { "epoch": 0.01, "learning_rate": 4.999458268861506e-05, "loss": 1.5474, "step": 6726 }, { "epoch": 0.01, "learning_rate": 4.9994581053382814e-05, "loss": 0.9322, "step": 6727 }, { "epoch": 0.01, "learning_rate": 4.999457941790382e-05, "loss": 0.4904, "step": 6728 }, { "epoch": 0.01, "learning_rate": 4.9994577782178095e-05, "loss": 0.3933, "step": 6729 }, { "epoch": 0.01, "learning_rate": 4.9994576146205644e-05, "loss": 0.3312, "step": 6730 }, { "epoch": 0.01, "learning_rate": 4.999457450998645e-05, "loss": 0.2082, "step": 6731 }, { "epoch": 0.01, "learning_rate": 4.9994572873520515e-05, "loss": 1.03, "step": 6732 }, { "epoch": 0.01, "learning_rate": 4.999457123680785e-05, "loss": 1.4107, "step": 6733 }, { "epoch": 0.01, "learning_rate": 4.999456959984845e-05, "loss": 1.2318, "step": 6734 }, { "epoch": 0.01, "learning_rate": 4.999456796264231e-05, "loss": 0.5894, "step": 6735 }, { "epoch": 0.01, "learning_rate": 4.999456632518944e-05, "loss": 0.5863, "step": 6736 }, { "epoch": 0.01, "learning_rate": 4.999456468748983e-05, "loss": 1.0186, "step": 6737 }, { "epoch": 0.01, "learning_rate": 4.999456304954349e-05, "loss": 1.0029, "step": 6738 }, { "epoch": 0.01, "learning_rate": 4.999456141135041e-05, "loss": 0.848, "step": 6739 }, { "epoch": 0.01, "learning_rate": 4.999455977291059e-05, "loss": 1.549, "step": 6740 }, { "epoch": 0.01, "learning_rate": 4.9994558134224035e-05, "loss": 1.1585, "step": 6741 }, { "epoch": 0.01, "learning_rate": 4.999455649529074e-05, "loss": 1.1006, "step": 6742 }, { "epoch": 0.01, "learning_rate": 4.999455485611072e-05, "loss": 1.3462, "step": 6743 }, { "epoch": 0.01, "learning_rate": 4.999455321668396e-05, "loss": 1.1244, "step": 6744 }, { "epoch": 0.01, "learning_rate": 4.999455157701046e-05, "loss": 1.3523, "step": 6745 }, { "epoch": 0.01, "learning_rate": 4.999454993709023e-05, "loss": 1.0386, "step": 6746 }, { "epoch": 0.01, "learning_rate": 4.999454829692327e-05, "loss": 1.4276, "step": 6747 }, { "epoch": 0.01, "learning_rate": 4.9994546656509565e-05, "loss": 1.3696, "step": 6748 }, { "epoch": 0.01, "learning_rate": 4.9994545015849125e-05, "loss": 1.2155, "step": 6749 }, { "epoch": 0.01, "learning_rate": 4.999454337494195e-05, "loss": 1.1756, "step": 6750 }, { "epoch": 0.01, "learning_rate": 4.999454173378804e-05, "loss": 1.3096, "step": 6751 }, { "epoch": 0.01, "learning_rate": 4.99945400923874e-05, "loss": 1.0379, "step": 6752 }, { "epoch": 0.01, "learning_rate": 4.999453845074001e-05, "loss": 1.0734, "step": 6753 }, { "epoch": 0.01, "learning_rate": 4.9994536808845906e-05, "loss": 0.8223, "step": 6754 }, { "epoch": 0.01, "learning_rate": 4.999453516670505e-05, "loss": 0.8731, "step": 6755 }, { "epoch": 0.01, "learning_rate": 4.999453352431746e-05, "loss": 0.7837, "step": 6756 }, { "epoch": 0.01, "learning_rate": 4.9994531881683136e-05, "loss": 0.9755, "step": 6757 }, { "epoch": 0.01, "learning_rate": 4.999453023880209e-05, "loss": 1.4549, "step": 6758 }, { "epoch": 0.01, "learning_rate": 4.999452859567428e-05, "loss": 1.209, "step": 6759 }, { "epoch": 0.01, "learning_rate": 4.999452695229976e-05, "loss": 0.7333, "step": 6760 }, { "epoch": 0.01, "learning_rate": 4.99945253086785e-05, "loss": 1.0987, "step": 6761 }, { "epoch": 0.01, "learning_rate": 4.999452366481049e-05, "loss": 1.315, "step": 6762 }, { "epoch": 0.01, "learning_rate": 4.9994522020695757e-05, "loss": 1.3147, "step": 6763 }, { "epoch": 0.01, "learning_rate": 4.9994520376334284e-05, "loss": 1.0884, "step": 6764 }, { "epoch": 0.01, "learning_rate": 4.999451873172608e-05, "loss": 1.1147, "step": 6765 }, { "epoch": 0.01, "learning_rate": 4.999451708687114e-05, "loss": 1.1792, "step": 6766 }, { "epoch": 0.01, "learning_rate": 4.999451544176946e-05, "loss": 1.3728, "step": 6767 }, { "epoch": 0.01, "learning_rate": 4.999451379642105e-05, "loss": 1.1996, "step": 6768 }, { "epoch": 0.01, "learning_rate": 4.99945121508259e-05, "loss": 0.7847, "step": 6769 }, { "epoch": 0.01, "learning_rate": 4.999451050498402e-05, "loss": 0.4295, "step": 6770 }, { "epoch": 0.01, "learning_rate": 4.9994508858895406e-05, "loss": 0.6031, "step": 6771 }, { "epoch": 0.01, "learning_rate": 4.999450721256005e-05, "loss": 0.8572, "step": 6772 }, { "epoch": 0.01, "learning_rate": 4.999450556597796e-05, "loss": 1.1249, "step": 6773 }, { "epoch": 0.01, "learning_rate": 4.999450391914914e-05, "loss": 0.8992, "step": 6774 }, { "epoch": 0.01, "learning_rate": 4.999450227207358e-05, "loss": 1.182, "step": 6775 }, { "epoch": 0.01, "learning_rate": 4.9994500624751285e-05, "loss": 1.1329, "step": 6776 }, { "epoch": 0.01, "learning_rate": 4.9994498977182255e-05, "loss": 1.2008, "step": 6777 }, { "epoch": 0.01, "learning_rate": 4.9994497329366494e-05, "loss": 1.1274, "step": 6778 }, { "epoch": 0.01, "learning_rate": 4.999449568130399e-05, "loss": 4.991, "step": 6779 }, { "epoch": 0.01, "learning_rate": 4.999449403299475e-05, "loss": 2.3363, "step": 6780 }, { "epoch": 0.01, "learning_rate": 4.999449238443879e-05, "loss": 1.1702, "step": 6781 }, { "epoch": 0.01, "learning_rate": 4.9994490735636076e-05, "loss": 1.3478, "step": 6782 }, { "epoch": 0.01, "learning_rate": 4.9994489086586636e-05, "loss": 1.3045, "step": 6783 }, { "epoch": 0.01, "learning_rate": 4.9994487437290464e-05, "loss": 0.9367, "step": 6784 }, { "epoch": 0.01, "learning_rate": 4.9994485787747556e-05, "loss": 1.5058, "step": 6785 }, { "epoch": 0.01, "learning_rate": 4.999448413795791e-05, "loss": 1.0185, "step": 6786 }, { "epoch": 0.01, "learning_rate": 4.9994482487921526e-05, "loss": 1.3135, "step": 6787 }, { "epoch": 0.01, "learning_rate": 4.999448083763841e-05, "loss": 1.1197, "step": 6788 }, { "epoch": 0.01, "learning_rate": 4.999447918710857e-05, "loss": 0.989, "step": 6789 }, { "epoch": 0.01, "learning_rate": 4.999447753633198e-05, "loss": 1.0585, "step": 6790 }, { "epoch": 0.01, "learning_rate": 4.9994475885308654e-05, "loss": 1.1427, "step": 6791 }, { "epoch": 0.01, "learning_rate": 4.9994474234038604e-05, "loss": 1.2104, "step": 6792 }, { "epoch": 0.01, "learning_rate": 4.999447258252181e-05, "loss": 1.1049, "step": 6793 }, { "epoch": 0.01, "learning_rate": 4.999447093075829e-05, "loss": 1.2225, "step": 6794 }, { "epoch": 0.01, "learning_rate": 4.999446927874803e-05, "loss": 1.271, "step": 6795 }, { "epoch": 0.01, "learning_rate": 4.999446762649104e-05, "loss": 1.402, "step": 6796 }, { "epoch": 0.01, "learning_rate": 4.9994465973987304e-05, "loss": 1.1503, "step": 6797 }, { "epoch": 0.01, "learning_rate": 4.999446432123684e-05, "loss": 1.2106, "step": 6798 }, { "epoch": 0.01, "learning_rate": 4.999446266823964e-05, "loss": 1.0387, "step": 6799 }, { "epoch": 0.01, "learning_rate": 4.9994461014995706e-05, "loss": 1.1426, "step": 6800 }, { "epoch": 0.01, "learning_rate": 4.999445936150504e-05, "loss": 1.159, "step": 6801 }, { "epoch": 0.01, "learning_rate": 4.999445770776764e-05, "loss": 1.0738, "step": 6802 }, { "epoch": 0.01, "learning_rate": 4.99944560537835e-05, "loss": 1.0195, "step": 6803 }, { "epoch": 0.01, "learning_rate": 4.999445439955263e-05, "loss": 1.1082, "step": 6804 }, { "epoch": 0.01, "learning_rate": 4.9994452745075015e-05, "loss": 1.1216, "step": 6805 }, { "epoch": 0.01, "learning_rate": 4.999445109035068e-05, "loss": 0.9553, "step": 6806 }, { "epoch": 0.01, "learning_rate": 4.9994449435379595e-05, "loss": 0.9602, "step": 6807 }, { "epoch": 0.01, "learning_rate": 4.999444778016178e-05, "loss": 1.156, "step": 6808 }, { "epoch": 0.01, "learning_rate": 4.999444612469724e-05, "loss": 0.9277, "step": 6809 }, { "epoch": 0.01, "learning_rate": 4.999444446898596e-05, "loss": 0.9691, "step": 6810 }, { "epoch": 0.01, "learning_rate": 4.999444281302794e-05, "loss": 1.2191, "step": 6811 }, { "epoch": 0.01, "learning_rate": 4.999444115682319e-05, "loss": 1.2958, "step": 6812 }, { "epoch": 0.01, "learning_rate": 4.999443950037171e-05, "loss": 1.5685, "step": 6813 }, { "epoch": 0.01, "learning_rate": 4.9994437843673484e-05, "loss": 1.3699, "step": 6814 }, { "epoch": 0.01, "learning_rate": 4.999443618672853e-05, "loss": 1.0349, "step": 6815 }, { "epoch": 0.01, "learning_rate": 4.9994434529536847e-05, "loss": 1.0596, "step": 6816 }, { "epoch": 0.01, "learning_rate": 4.999443287209842e-05, "loss": 0.853, "step": 6817 }, { "epoch": 0.01, "learning_rate": 4.999443121441326e-05, "loss": 0.9858, "step": 6818 }, { "epoch": 0.01, "learning_rate": 4.999442955648137e-05, "loss": 1.1507, "step": 6819 }, { "epoch": 0.01, "learning_rate": 4.9994427898302744e-05, "loss": 1.123, "step": 6820 }, { "epoch": 0.01, "learning_rate": 4.999442623987739e-05, "loss": 0.9609, "step": 6821 }, { "epoch": 0.01, "learning_rate": 4.999442458120529e-05, "loss": 1.2603, "step": 6822 }, { "epoch": 0.01, "learning_rate": 4.9994422922286455e-05, "loss": 0.9679, "step": 6823 }, { "epoch": 0.01, "learning_rate": 4.999442126312089e-05, "loss": 1.1821, "step": 6824 }, { "epoch": 0.01, "learning_rate": 4.99944196037086e-05, "loss": 1.0709, "step": 6825 }, { "epoch": 0.01, "learning_rate": 4.9994417944049563e-05, "loss": 1.222, "step": 6826 }, { "epoch": 0.01, "learning_rate": 4.9994416284143796e-05, "loss": 1.2295, "step": 6827 }, { "epoch": 0.01, "learning_rate": 4.999441462399129e-05, "loss": 0.9857, "step": 6828 }, { "epoch": 0.01, "learning_rate": 4.9994412963592057e-05, "loss": 0.7959, "step": 6829 }, { "epoch": 0.01, "learning_rate": 4.9994411302946084e-05, "loss": 0.9756, "step": 6830 }, { "epoch": 0.01, "learning_rate": 4.999440964205338e-05, "loss": 1.1994, "step": 6831 }, { "epoch": 0.01, "learning_rate": 4.999440798091394e-05, "loss": 1.4401, "step": 6832 }, { "epoch": 0.01, "learning_rate": 4.999440631952777e-05, "loss": 0.9325, "step": 6833 }, { "epoch": 0.01, "learning_rate": 4.999440465789486e-05, "loss": 0.7858, "step": 6834 }, { "epoch": 0.01, "learning_rate": 4.9994402996015223e-05, "loss": 0.8887, "step": 6835 }, { "epoch": 0.01, "learning_rate": 4.999440133388884e-05, "loss": 1.0236, "step": 6836 }, { "epoch": 0.01, "learning_rate": 4.9994399671515734e-05, "loss": 1.1955, "step": 6837 }, { "epoch": 0.01, "learning_rate": 4.999439800889589e-05, "loss": 0.9023, "step": 6838 }, { "epoch": 0.01, "learning_rate": 4.999439634602931e-05, "loss": 1.3026, "step": 6839 }, { "epoch": 0.01, "learning_rate": 4.9994394682916005e-05, "loss": 1.0394, "step": 6840 }, { "epoch": 0.01, "learning_rate": 4.999439301955595e-05, "loss": 0.9914, "step": 6841 }, { "epoch": 0.01, "learning_rate": 4.999439135594917e-05, "loss": 1.3485, "step": 6842 }, { "epoch": 0.01, "learning_rate": 4.999438969209565e-05, "loss": 1.207, "step": 6843 }, { "epoch": 0.01, "learning_rate": 4.999438802799541e-05, "loss": 0.8097, "step": 6844 }, { "epoch": 0.01, "learning_rate": 4.999438636364843e-05, "loss": 1.2124, "step": 6845 }, { "epoch": 0.01, "learning_rate": 4.9994384699054706e-05, "loss": 1.2324, "step": 6846 }, { "epoch": 0.01, "learning_rate": 4.999438303421425e-05, "loss": 1.0811, "step": 6847 }, { "epoch": 0.01, "learning_rate": 4.999438136912707e-05, "loss": 0.9873, "step": 6848 }, { "epoch": 0.01, "learning_rate": 4.999437970379315e-05, "loss": 1.0589, "step": 6849 }, { "epoch": 0.01, "learning_rate": 4.99943780382125e-05, "loss": 1.2425, "step": 6850 }, { "epoch": 0.01, "learning_rate": 4.999437637238511e-05, "loss": 0.6193, "step": 6851 }, { "epoch": 0.01, "learning_rate": 4.9994374706310985e-05, "loss": 0.4063, "step": 6852 }, { "epoch": 0.01, "learning_rate": 4.9994373039990136e-05, "loss": 0.3648, "step": 6853 }, { "epoch": 0.01, "learning_rate": 4.999437137342254e-05, "loss": 0.3846, "step": 6854 }, { "epoch": 0.01, "learning_rate": 4.9994369706608226e-05, "loss": 0.3362, "step": 6855 }, { "epoch": 0.01, "learning_rate": 4.9994368039547165e-05, "loss": 0.3245, "step": 6856 }, { "epoch": 0.01, "learning_rate": 4.999436637223937e-05, "loss": 0.3606, "step": 6857 }, { "epoch": 0.01, "learning_rate": 4.999436470468485e-05, "loss": 0.3314, "step": 6858 }, { "epoch": 0.01, "learning_rate": 4.999436303688359e-05, "loss": 0.4419, "step": 6859 }, { "epoch": 0.01, "learning_rate": 4.9994361368835595e-05, "loss": 0.9462, "step": 6860 }, { "epoch": 0.01, "learning_rate": 4.9994359700540874e-05, "loss": 1.5972, "step": 6861 }, { "epoch": 0.01, "learning_rate": 4.999435803199941e-05, "loss": 1.2746, "step": 6862 }, { "epoch": 0.01, "learning_rate": 4.999435636321122e-05, "loss": 1.2198, "step": 6863 }, { "epoch": 0.01, "learning_rate": 4.9994354694176295e-05, "loss": 0.869, "step": 6864 }, { "epoch": 0.01, "learning_rate": 4.999435302489463e-05, "loss": 0.4223, "step": 6865 }, { "epoch": 0.01, "learning_rate": 4.999435135536624e-05, "loss": 0.3452, "step": 6866 }, { "epoch": 0.01, "learning_rate": 4.999434968559111e-05, "loss": 0.6081, "step": 6867 }, { "epoch": 0.01, "learning_rate": 4.9994348015569246e-05, "loss": 1.0303, "step": 6868 }, { "epoch": 0.01, "learning_rate": 4.999434634530065e-05, "loss": 1.0843, "step": 6869 }, { "epoch": 0.01, "learning_rate": 4.999434467478532e-05, "loss": 0.9482, "step": 6870 }, { "epoch": 0.01, "learning_rate": 4.999434300402326e-05, "loss": 0.6892, "step": 6871 }, { "epoch": 0.01, "learning_rate": 4.999434133301446e-05, "loss": 0.5223, "step": 6872 }, { "epoch": 0.01, "learning_rate": 4.999433966175893e-05, "loss": 1.151, "step": 6873 }, { "epoch": 0.01, "learning_rate": 4.999433799025667e-05, "loss": 1.3094, "step": 6874 }, { "epoch": 0.01, "learning_rate": 4.9994336318507667e-05, "loss": 1.0836, "step": 6875 }, { "epoch": 0.01, "learning_rate": 4.9994334646511934e-05, "loss": 1.6371, "step": 6876 }, { "epoch": 0.01, "learning_rate": 4.999433297426948e-05, "loss": 1.6683, "step": 6877 }, { "epoch": 0.01, "learning_rate": 4.999433130178027e-05, "loss": 1.7682, "step": 6878 }, { "epoch": 0.01, "learning_rate": 4.9994329629044345e-05, "loss": 1.4243, "step": 6879 }, { "epoch": 0.01, "learning_rate": 4.9994327956061684e-05, "loss": 1.7879, "step": 6880 }, { "epoch": 0.01, "learning_rate": 4.9994326282832285e-05, "loss": 1.7397, "step": 6881 }, { "epoch": 0.01, "learning_rate": 4.999432460935615e-05, "loss": 1.1371, "step": 6882 }, { "epoch": 0.01, "learning_rate": 4.999432293563328e-05, "loss": 1.1943, "step": 6883 }, { "epoch": 0.01, "learning_rate": 4.999432126166368e-05, "loss": 1.2871, "step": 6884 }, { "epoch": 0.01, "learning_rate": 4.999431958744735e-05, "loss": 1.2688, "step": 6885 }, { "epoch": 0.01, "learning_rate": 4.9994317912984284e-05, "loss": 1.5491, "step": 6886 }, { "epoch": 0.01, "learning_rate": 4.999431623827449e-05, "loss": 1.1822, "step": 6887 }, { "epoch": 0.01, "learning_rate": 4.9994314563317955e-05, "loss": 1.2313, "step": 6888 }, { "epoch": 0.01, "learning_rate": 4.9994312888114685e-05, "loss": 1.0805, "step": 6889 }, { "epoch": 0.01, "learning_rate": 4.999431121266469e-05, "loss": 1.0169, "step": 6890 }, { "epoch": 0.01, "learning_rate": 4.999430953696795e-05, "loss": 1.4716, "step": 6891 }, { "epoch": 0.01, "learning_rate": 4.999430786102449e-05, "loss": 1.3546, "step": 6892 }, { "epoch": 0.01, "learning_rate": 4.999430618483429e-05, "loss": 1.2015, "step": 6893 }, { "epoch": 0.01, "learning_rate": 4.999430450839736e-05, "loss": 0.8801, "step": 6894 }, { "epoch": 0.01, "learning_rate": 4.999430283171369e-05, "loss": 1.2549, "step": 6895 }, { "epoch": 0.01, "learning_rate": 4.99943011547833e-05, "loss": 1.0718, "step": 6896 }, { "epoch": 0.01, "learning_rate": 4.999429947760617e-05, "loss": 1.2091, "step": 6897 }, { "epoch": 0.01, "learning_rate": 4.9994297800182296e-05, "loss": 1.1126, "step": 6898 }, { "epoch": 0.01, "learning_rate": 4.9994296122511694e-05, "loss": 0.8508, "step": 6899 }, { "epoch": 0.01, "learning_rate": 4.999429444459437e-05, "loss": 1.2032, "step": 6900 }, { "epoch": 0.01, "learning_rate": 4.99942927664303e-05, "loss": 1.1239, "step": 6901 }, { "epoch": 0.01, "learning_rate": 4.999429108801951e-05, "loss": 1.1106, "step": 6902 }, { "epoch": 0.01, "learning_rate": 4.999428940936197e-05, "loss": 1.2155, "step": 6903 }, { "epoch": 0.01, "learning_rate": 4.9994287730457715e-05, "loss": 1.0073, "step": 6904 }, { "epoch": 0.01, "learning_rate": 4.9994286051306715e-05, "loss": 0.7326, "step": 6905 }, { "epoch": 0.01, "learning_rate": 4.9994284371908985e-05, "loss": 1.1721, "step": 6906 }, { "epoch": 0.01, "learning_rate": 4.9994282692264525e-05, "loss": 1.2096, "step": 6907 }, { "epoch": 0.01, "learning_rate": 4.999428101237332e-05, "loss": 0.9284, "step": 6908 }, { "epoch": 0.01, "learning_rate": 4.999427933223539e-05, "loss": 1.0399, "step": 6909 }, { "epoch": 0.01, "learning_rate": 4.999427765185073e-05, "loss": 1.6415, "step": 6910 }, { "epoch": 0.01, "learning_rate": 4.999427597121934e-05, "loss": 1.1169, "step": 6911 }, { "epoch": 0.01, "learning_rate": 4.999427429034121e-05, "loss": 1.2315, "step": 6912 }, { "epoch": 0.01, "learning_rate": 4.999427260921635e-05, "loss": 1.3721, "step": 6913 }, { "epoch": 0.01, "learning_rate": 4.999427092784476e-05, "loss": 1.1605, "step": 6914 }, { "epoch": 0.01, "learning_rate": 4.9994269246226425e-05, "loss": 1.0359, "step": 6915 }, { "epoch": 0.01, "learning_rate": 4.999426756436137e-05, "loss": 1.2378, "step": 6916 }, { "epoch": 0.01, "learning_rate": 4.9994265882249576e-05, "loss": 1.1071, "step": 6917 }, { "epoch": 0.01, "learning_rate": 4.999426419989105e-05, "loss": 0.9853, "step": 6918 }, { "epoch": 0.01, "learning_rate": 4.999426251728579e-05, "loss": 0.9752, "step": 6919 }, { "epoch": 0.01, "learning_rate": 4.99942608344338e-05, "loss": 1.0607, "step": 6920 }, { "epoch": 0.01, "learning_rate": 4.999425915133508e-05, "loss": 1.1958, "step": 6921 }, { "epoch": 0.01, "learning_rate": 4.999425746798962e-05, "loss": 1.3917, "step": 6922 }, { "epoch": 0.01, "learning_rate": 4.999425578439743e-05, "loss": 0.859, "step": 6923 }, { "epoch": 0.01, "learning_rate": 4.999425410055851e-05, "loss": 0.5413, "step": 6924 }, { "epoch": 0.01, "learning_rate": 4.999425241647285e-05, "loss": 0.7141, "step": 6925 }, { "epoch": 0.01, "learning_rate": 4.999425073214047e-05, "loss": 0.6767, "step": 6926 }, { "epoch": 0.01, "learning_rate": 4.999424904756135e-05, "loss": 1.3069, "step": 6927 }, { "epoch": 0.01, "learning_rate": 4.999424736273549e-05, "loss": 1.1934, "step": 6928 }, { "epoch": 0.01, "learning_rate": 4.99942456776629e-05, "loss": 1.2299, "step": 6929 }, { "epoch": 0.01, "learning_rate": 4.9994243992343585e-05, "loss": 0.6811, "step": 6930 }, { "epoch": 0.01, "learning_rate": 4.999424230677754e-05, "loss": 0.117, "step": 6931 }, { "epoch": 0.01, "learning_rate": 4.999424062096475e-05, "loss": 0.1434, "step": 6932 }, { "epoch": 0.01, "learning_rate": 4.999423893490524e-05, "loss": 0.0817, "step": 6933 }, { "epoch": 0.01, "learning_rate": 4.9994237248598984e-05, "loss": 0.0638, "step": 6934 }, { "epoch": 0.01, "learning_rate": 4.999423556204601e-05, "loss": 0.234, "step": 6935 }, { "epoch": 0.01, "learning_rate": 4.999423387524629e-05, "loss": 0.0815, "step": 6936 }, { "epoch": 0.01, "learning_rate": 4.999423218819984e-05, "loss": 0.3098, "step": 6937 }, { "epoch": 0.01, "learning_rate": 4.999423050090667e-05, "loss": 0.9545, "step": 6938 }, { "epoch": 0.01, "learning_rate": 4.999422881336675e-05, "loss": 1.0009, "step": 6939 }, { "epoch": 0.01, "learning_rate": 4.9994227125580106e-05, "loss": 0.3902, "step": 6940 }, { "epoch": 0.01, "learning_rate": 4.999422543754673e-05, "loss": 0.1013, "step": 6941 }, { "epoch": 0.01, "learning_rate": 4.999422374926662e-05, "loss": 3.111, "step": 6942 }, { "epoch": 0.01, "learning_rate": 4.999422206073978e-05, "loss": 5.4106, "step": 6943 }, { "epoch": 0.01, "learning_rate": 4.99942203719662e-05, "loss": 4.7675, "step": 6944 }, { "epoch": 0.01, "learning_rate": 4.999421868294591e-05, "loss": 4.1092, "step": 6945 }, { "epoch": 0.01, "learning_rate": 4.999421699367887e-05, "loss": 0.9569, "step": 6946 }, { "epoch": 0.01, "learning_rate": 4.999421530416509e-05, "loss": 1.146, "step": 6947 }, { "epoch": 0.01, "learning_rate": 4.999421361440458e-05, "loss": 0.9052, "step": 6948 }, { "epoch": 0.01, "learning_rate": 4.999421192439735e-05, "loss": 0.9786, "step": 6949 }, { "epoch": 0.01, "learning_rate": 4.999421023414338e-05, "loss": 1.1034, "step": 6950 }, { "epoch": 0.01, "learning_rate": 4.9994208543642685e-05, "loss": 1.2519, "step": 6951 }, { "epoch": 0.01, "learning_rate": 4.9994206852895255e-05, "loss": 1.1061, "step": 6952 }, { "epoch": 0.01, "learning_rate": 4.999420516190109e-05, "loss": 0.933, "step": 6953 }, { "epoch": 0.01, "learning_rate": 4.9994203470660184e-05, "loss": 0.9922, "step": 6954 }, { "epoch": 0.01, "learning_rate": 4.9994201779172556e-05, "loss": 0.9366, "step": 6955 }, { "epoch": 0.01, "learning_rate": 4.999420008743819e-05, "loss": 1.1203, "step": 6956 }, { "epoch": 0.01, "learning_rate": 4.99941983954571e-05, "loss": 1.1666, "step": 6957 }, { "epoch": 0.01, "learning_rate": 4.9994196703229276e-05, "loss": 1.0677, "step": 6958 }, { "epoch": 0.01, "learning_rate": 4.999419501075472e-05, "loss": 1.1357, "step": 6959 }, { "epoch": 0.01, "learning_rate": 4.9994193318033425e-05, "loss": 1.1193, "step": 6960 }, { "epoch": 0.01, "learning_rate": 4.99941916250654e-05, "loss": 1.3423, "step": 6961 }, { "epoch": 0.01, "learning_rate": 4.999418993185065e-05, "loss": 1.2765, "step": 6962 }, { "epoch": 0.01, "learning_rate": 4.999418823838916e-05, "loss": 0.8398, "step": 6963 }, { "epoch": 0.01, "learning_rate": 4.9994186544680943e-05, "loss": 0.3674, "step": 6964 }, { "epoch": 0.01, "learning_rate": 4.999418485072599e-05, "loss": 0.4106, "step": 6965 }, { "epoch": 0.01, "learning_rate": 4.999418315652431e-05, "loss": 0.5539, "step": 6966 }, { "epoch": 0.01, "learning_rate": 4.99941814620759e-05, "loss": 1.1158, "step": 6967 }, { "epoch": 0.01, "learning_rate": 4.9994179767380746e-05, "loss": 1.0879, "step": 6968 }, { "epoch": 0.01, "learning_rate": 4.999417807243887e-05, "loss": 0.9524, "step": 6969 }, { "epoch": 0.01, "learning_rate": 4.999417637725026e-05, "loss": 0.9135, "step": 6970 }, { "epoch": 0.01, "learning_rate": 4.999417468181491e-05, "loss": 0.9985, "step": 6971 }, { "epoch": 0.01, "learning_rate": 4.999417298613284e-05, "loss": 1.372, "step": 6972 }, { "epoch": 0.01, "learning_rate": 4.999417129020403e-05, "loss": 1.4207, "step": 6973 }, { "epoch": 0.01, "learning_rate": 4.999416959402849e-05, "loss": 1.0777, "step": 6974 }, { "epoch": 0.01, "learning_rate": 4.999416789760622e-05, "loss": 1.1585, "step": 6975 }, { "epoch": 0.01, "learning_rate": 4.999416620093722e-05, "loss": 1.0714, "step": 6976 }, { "epoch": 0.01, "learning_rate": 4.999416450402148e-05, "loss": 1.1122, "step": 6977 }, { "epoch": 0.01, "learning_rate": 4.9994162806859014e-05, "loss": 1.3217, "step": 6978 }, { "epoch": 0.01, "learning_rate": 4.999416110944982e-05, "loss": 1.0181, "step": 6979 }, { "epoch": 0.01, "learning_rate": 4.999415941179388e-05, "loss": 1.3369, "step": 6980 }, { "epoch": 0.01, "learning_rate": 4.999415771389122e-05, "loss": 1.1256, "step": 6981 }, { "epoch": 0.01, "learning_rate": 4.999415601574183e-05, "loss": 1.1828, "step": 6982 }, { "epoch": 0.01, "learning_rate": 4.99941543173457e-05, "loss": 1.3077, "step": 6983 }, { "epoch": 0.01, "learning_rate": 4.999415261870285e-05, "loss": 0.9496, "step": 6984 }, { "epoch": 0.01, "learning_rate": 4.999415091981326e-05, "loss": 1.2384, "step": 6985 }, { "epoch": 0.01, "learning_rate": 4.9994149220676935e-05, "loss": 0.9296, "step": 6986 }, { "epoch": 0.01, "learning_rate": 4.999414752129389e-05, "loss": 1.3176, "step": 6987 }, { "epoch": 0.01, "learning_rate": 4.99941458216641e-05, "loss": 0.3747, "step": 6988 }, { "epoch": 0.01, "learning_rate": 4.9994144121787586e-05, "loss": 0.1655, "step": 6989 }, { "epoch": 0.01, "learning_rate": 4.9994142421664336e-05, "loss": 0.2205, "step": 6990 }, { "epoch": 0.01, "learning_rate": 4.999414072129436e-05, "loss": 0.2001, "step": 6991 }, { "epoch": 0.01, "learning_rate": 4.999413902067764e-05, "loss": 0.2171, "step": 6992 }, { "epoch": 0.01, "learning_rate": 4.999413731981421e-05, "loss": 0.8767, "step": 6993 }, { "epoch": 0.01, "learning_rate": 4.999413561870403e-05, "loss": 1.1458, "step": 6994 }, { "epoch": 0.01, "learning_rate": 4.9994133917347124e-05, "loss": 1.0598, "step": 6995 }, { "epoch": 0.01, "learning_rate": 4.999413221574348e-05, "loss": 0.576, "step": 6996 }, { "epoch": 0.01, "learning_rate": 4.999413051389312e-05, "loss": 0.6216, "step": 6997 }, { "epoch": 0.01, "learning_rate": 4.999412881179602e-05, "loss": 0.5405, "step": 6998 }, { "epoch": 0.01, "learning_rate": 4.9994127109452185e-05, "loss": 0.9278, "step": 6999 }, { "epoch": 0.01, "learning_rate": 4.999412540686162e-05, "loss": 0.918, "step": 7000 }, { "epoch": 0.01, "eval_loss": 1.0878294706344604, "eval_runtime": 84.2959, "eval_samples_per_second": 16.43, "eval_steps_per_second": 4.116, "step": 7000 }, { "epoch": 0.01, "learning_rate": 4.999412370402433e-05, "loss": 1.0767, "step": 7001 }, { "epoch": 0.01, "learning_rate": 4.9994122000940305e-05, "loss": 1.2599, "step": 7002 }, { "epoch": 0.01, "learning_rate": 4.9994120297609543e-05, "loss": 1.2067, "step": 7003 }, { "epoch": 0.01, "learning_rate": 4.999411859403206e-05, "loss": 1.2545, "step": 7004 }, { "epoch": 0.01, "learning_rate": 4.999411689020783e-05, "loss": 1.119, "step": 7005 }, { "epoch": 0.01, "learning_rate": 4.999411518613688e-05, "loss": 1.1328, "step": 7006 }, { "epoch": 0.01, "learning_rate": 4.99941134818192e-05, "loss": 1.1584, "step": 7007 }, { "epoch": 0.01, "learning_rate": 4.999411177725478e-05, "loss": 1.1062, "step": 7008 }, { "epoch": 0.01, "learning_rate": 4.9994110072443635e-05, "loss": 1.0393, "step": 7009 }, { "epoch": 0.01, "learning_rate": 4.999410836738576e-05, "loss": 0.9956, "step": 7010 }, { "epoch": 0.01, "learning_rate": 4.9994106662081155e-05, "loss": 1.1625, "step": 7011 }, { "epoch": 0.01, "learning_rate": 4.9994104956529805e-05, "loss": 0.7091, "step": 7012 }, { "epoch": 0.01, "learning_rate": 4.999410325073174e-05, "loss": 1.0893, "step": 7013 }, { "epoch": 0.01, "learning_rate": 4.999410154468694e-05, "loss": 1.008, "step": 7014 }, { "epoch": 0.01, "learning_rate": 4.99940998383954e-05, "loss": 1.069, "step": 7015 }, { "epoch": 0.01, "learning_rate": 4.9994098131857135e-05, "loss": 1.0903, "step": 7016 }, { "epoch": 0.01, "learning_rate": 4.9994096425072147e-05, "loss": 1.1329, "step": 7017 }, { "epoch": 0.01, "learning_rate": 4.999409471804042e-05, "loss": 1.099, "step": 7018 }, { "epoch": 0.01, "learning_rate": 4.999409301076196e-05, "loss": 0.8939, "step": 7019 }, { "epoch": 0.01, "learning_rate": 4.999409130323677e-05, "loss": 1.1571, "step": 7020 }, { "epoch": 0.01, "learning_rate": 4.9994089595464846e-05, "loss": 1.0265, "step": 7021 }, { "epoch": 0.01, "learning_rate": 4.99940878874462e-05, "loss": 1.7861, "step": 7022 }, { "epoch": 0.01, "learning_rate": 4.999408617918081e-05, "loss": 1.5894, "step": 7023 }, { "epoch": 0.01, "learning_rate": 4.99940844706687e-05, "loss": 1.3719, "step": 7024 }, { "epoch": 0.01, "learning_rate": 4.999408276190986e-05, "loss": 1.1614, "step": 7025 }, { "epoch": 0.01, "learning_rate": 4.999408105290427e-05, "loss": 1.1558, "step": 7026 }, { "epoch": 0.01, "learning_rate": 4.9994079343651965e-05, "loss": 1.1941, "step": 7027 }, { "epoch": 0.01, "learning_rate": 4.9994077634152934e-05, "loss": 1.3192, "step": 7028 }, { "epoch": 0.01, "learning_rate": 4.9994075924407165e-05, "loss": 1.0487, "step": 7029 }, { "epoch": 0.01, "learning_rate": 4.9994074214414666e-05, "loss": 1.1189, "step": 7030 }, { "epoch": 0.01, "learning_rate": 4.999407250417543e-05, "loss": 1.2138, "step": 7031 }, { "epoch": 0.01, "learning_rate": 4.999407079368947e-05, "loss": 0.9906, "step": 7032 }, { "epoch": 0.01, "learning_rate": 4.999406908295677e-05, "loss": 1.2017, "step": 7033 }, { "epoch": 0.01, "learning_rate": 4.999406737197735e-05, "loss": 1.2657, "step": 7034 }, { "epoch": 0.01, "learning_rate": 4.99940656607512e-05, "loss": 1.2592, "step": 7035 }, { "epoch": 0.01, "learning_rate": 4.9994063949278316e-05, "loss": 1.2166, "step": 7036 }, { "epoch": 0.01, "learning_rate": 4.999406223755869e-05, "loss": 1.1549, "step": 7037 }, { "epoch": 0.01, "learning_rate": 4.9994060525592347e-05, "loss": 0.8705, "step": 7038 }, { "epoch": 0.01, "learning_rate": 4.9994058813379266e-05, "loss": 1.2059, "step": 7039 }, { "epoch": 0.01, "learning_rate": 4.999405710091946e-05, "loss": 1.0886, "step": 7040 }, { "epoch": 0.01, "learning_rate": 4.9994055388212926e-05, "loss": 0.1385, "step": 7041 }, { "epoch": 0.01, "learning_rate": 4.999405367525965e-05, "loss": 0.1784, "step": 7042 }, { "epoch": 0.01, "learning_rate": 4.999405196205965e-05, "loss": 1.1128, "step": 7043 }, { "epoch": 0.01, "learning_rate": 4.999405024861292e-05, "loss": 1.0444, "step": 7044 }, { "epoch": 0.01, "learning_rate": 4.9994048534919454e-05, "loss": 0.9553, "step": 7045 }, { "epoch": 0.01, "learning_rate": 4.999404682097926e-05, "loss": 1.2984, "step": 7046 }, { "epoch": 0.01, "learning_rate": 4.9994045106792334e-05, "loss": 1.1051, "step": 7047 }, { "epoch": 0.01, "learning_rate": 4.999404339235868e-05, "loss": 1.0497, "step": 7048 }, { "epoch": 0.01, "learning_rate": 4.999404167767829e-05, "loss": 0.6193, "step": 7049 }, { "epoch": 0.01, "learning_rate": 4.9994039962751176e-05, "loss": 1.0276, "step": 7050 }, { "epoch": 0.01, "learning_rate": 4.999403824757733e-05, "loss": 1.0596, "step": 7051 }, { "epoch": 0.01, "learning_rate": 4.999403653215675e-05, "loss": 0.7923, "step": 7052 }, { "epoch": 0.01, "learning_rate": 4.999403481648944e-05, "loss": 1.1786, "step": 7053 }, { "epoch": 0.01, "learning_rate": 4.9994033100575404e-05, "loss": 1.2927, "step": 7054 }, { "epoch": 0.01, "learning_rate": 4.9994031384414635e-05, "loss": 1.0812, "step": 7055 }, { "epoch": 0.01, "learning_rate": 4.9994029668007135e-05, "loss": 1.1084, "step": 7056 }, { "epoch": 0.01, "learning_rate": 4.9994027951352905e-05, "loss": 1.2078, "step": 7057 }, { "epoch": 0.01, "learning_rate": 4.9994026234451944e-05, "loss": 0.9318, "step": 7058 }, { "epoch": 0.01, "learning_rate": 4.999402451730425e-05, "loss": 1.1528, "step": 7059 }, { "epoch": 0.01, "learning_rate": 4.999402279990984e-05, "loss": 1.398, "step": 7060 }, { "epoch": 0.01, "learning_rate": 4.9994021082268685e-05, "loss": 1.0382, "step": 7061 }, { "epoch": 0.01, "learning_rate": 4.9994019364380795e-05, "loss": 0.694, "step": 7062 }, { "epoch": 0.01, "learning_rate": 4.999401764624618e-05, "loss": 0.7811, "step": 7063 }, { "epoch": 0.01, "learning_rate": 4.999401592786484e-05, "loss": 1.1132, "step": 7064 }, { "epoch": 0.01, "learning_rate": 4.999401420923676e-05, "loss": 0.9212, "step": 7065 }, { "epoch": 0.01, "learning_rate": 4.9994012490361965e-05, "loss": 0.4159, "step": 7066 }, { "epoch": 0.01, "learning_rate": 4.999401077124043e-05, "loss": 0.5159, "step": 7067 }, { "epoch": 0.01, "learning_rate": 4.9994009051872156e-05, "loss": 1.156, "step": 7068 }, { "epoch": 0.01, "learning_rate": 4.9994007332257166e-05, "loss": 0.8949, "step": 7069 }, { "epoch": 0.01, "learning_rate": 4.999400561239544e-05, "loss": 0.7439, "step": 7070 }, { "epoch": 0.01, "learning_rate": 4.999400389228698e-05, "loss": 1.122, "step": 7071 }, { "epoch": 0.01, "learning_rate": 4.99940021719318e-05, "loss": 1.1978, "step": 7072 }, { "epoch": 0.01, "learning_rate": 4.999400045132988e-05, "loss": 1.0629, "step": 7073 }, { "epoch": 0.01, "learning_rate": 4.999399873048123e-05, "loss": 0.4868, "step": 7074 }, { "epoch": 0.01, "learning_rate": 4.999399700938586e-05, "loss": 1.0516, "step": 7075 }, { "epoch": 0.01, "learning_rate": 4.999399528804375e-05, "loss": 1.1544, "step": 7076 }, { "epoch": 0.01, "learning_rate": 4.9993993566454913e-05, "loss": 1.2188, "step": 7077 }, { "epoch": 0.01, "learning_rate": 4.999399184461935e-05, "loss": 0.8363, "step": 7078 }, { "epoch": 0.01, "learning_rate": 4.999399012253705e-05, "loss": 0.7173, "step": 7079 }, { "epoch": 0.01, "learning_rate": 4.999398840020802e-05, "loss": 0.535, "step": 7080 }, { "epoch": 0.01, "learning_rate": 4.999398667763226e-05, "loss": 0.6233, "step": 7081 }, { "epoch": 0.01, "learning_rate": 4.999398495480977e-05, "loss": 0.4848, "step": 7082 }, { "epoch": 0.01, "learning_rate": 4.999398323174055e-05, "loss": 1.0284, "step": 7083 }, { "epoch": 0.01, "learning_rate": 4.9993981508424605e-05, "loss": 1.0248, "step": 7084 }, { "epoch": 0.01, "learning_rate": 4.999397978486193e-05, "loss": 1.3033, "step": 7085 }, { "epoch": 0.01, "learning_rate": 4.999397806105252e-05, "loss": 1.3483, "step": 7086 }, { "epoch": 0.01, "learning_rate": 4.9993976336996385e-05, "loss": 1.154, "step": 7087 }, { "epoch": 0.01, "learning_rate": 4.999397461269351e-05, "loss": 1.2915, "step": 7088 }, { "epoch": 0.01, "learning_rate": 4.9993972888143914e-05, "loss": 1.0423, "step": 7089 }, { "epoch": 0.01, "learning_rate": 4.999397116334758e-05, "loss": 1.1034, "step": 7090 }, { "epoch": 0.01, "learning_rate": 4.999396943830453e-05, "loss": 0.9868, "step": 7091 }, { "epoch": 0.01, "learning_rate": 4.9993967713014736e-05, "loss": 0.2809, "step": 7092 }, { "epoch": 0.01, "learning_rate": 4.999396598747822e-05, "loss": 0.3869, "step": 7093 }, { "epoch": 0.01, "learning_rate": 4.999396426169497e-05, "loss": 1.2138, "step": 7094 }, { "epoch": 0.01, "learning_rate": 4.9993962535665e-05, "loss": 1.1352, "step": 7095 }, { "epoch": 0.01, "learning_rate": 4.999396080938828e-05, "loss": 0.7317, "step": 7096 }, { "epoch": 0.01, "learning_rate": 4.999395908286485e-05, "loss": 1.0277, "step": 7097 }, { "epoch": 0.01, "learning_rate": 4.9993957356094676e-05, "loss": 1.166, "step": 7098 }, { "epoch": 0.01, "learning_rate": 4.999395562907778e-05, "loss": 1.1118, "step": 7099 }, { "epoch": 0.01, "learning_rate": 4.999395390181416e-05, "loss": 1.0271, "step": 7100 }, { "epoch": 0.01, "learning_rate": 4.999395217430381e-05, "loss": 1.1238, "step": 7101 }, { "epoch": 0.01, "learning_rate": 4.999395044654671e-05, "loss": 0.994, "step": 7102 }, { "epoch": 0.01, "learning_rate": 4.99939487185429e-05, "loss": 1.1137, "step": 7103 }, { "epoch": 0.01, "learning_rate": 4.999394699029235e-05, "loss": 1.2896, "step": 7104 }, { "epoch": 0.01, "learning_rate": 4.9993945261795074e-05, "loss": 1.1667, "step": 7105 }, { "epoch": 0.01, "learning_rate": 4.999394353305107e-05, "loss": 1.2486, "step": 7106 }, { "epoch": 0.01, "learning_rate": 4.999394180406034e-05, "loss": 1.0502, "step": 7107 }, { "epoch": 0.01, "learning_rate": 4.999394007482287e-05, "loss": 1.0564, "step": 7108 }, { "epoch": 0.01, "learning_rate": 4.999393834533867e-05, "loss": 2.3266, "step": 7109 }, { "epoch": 0.01, "learning_rate": 4.999393661560775e-05, "loss": 1.1655, "step": 7110 }, { "epoch": 0.01, "learning_rate": 4.99939348856301e-05, "loss": 1.1056, "step": 7111 }, { "epoch": 0.01, "learning_rate": 4.999393315540572e-05, "loss": 0.9551, "step": 7112 }, { "epoch": 0.01, "learning_rate": 4.9993931424934604e-05, "loss": 1.2253, "step": 7113 }, { "epoch": 0.01, "learning_rate": 4.9993929694216754e-05, "loss": 1.2538, "step": 7114 }, { "epoch": 0.01, "learning_rate": 4.9993927963252186e-05, "loss": 1.1117, "step": 7115 }, { "epoch": 0.01, "learning_rate": 4.999392623204089e-05, "loss": 1.1508, "step": 7116 }, { "epoch": 0.01, "learning_rate": 4.999392450058285e-05, "loss": 1.3282, "step": 7117 }, { "epoch": 0.01, "learning_rate": 4.9993922768878094e-05, "loss": 1.2517, "step": 7118 }, { "epoch": 0.01, "learning_rate": 4.9993921036926605e-05, "loss": 0.919, "step": 7119 }, { "epoch": 0.01, "learning_rate": 4.9993919304728385e-05, "loss": 1.1984, "step": 7120 }, { "epoch": 0.01, "learning_rate": 4.9993917572283434e-05, "loss": 1.0078, "step": 7121 }, { "epoch": 0.01, "learning_rate": 4.999391583959175e-05, "loss": 1.0748, "step": 7122 }, { "epoch": 0.01, "learning_rate": 4.999391410665335e-05, "loss": 1.374, "step": 7123 }, { "epoch": 0.01, "learning_rate": 4.999391237346821e-05, "loss": 1.2001, "step": 7124 }, { "epoch": 0.01, "learning_rate": 4.999391064003635e-05, "loss": 1.3737, "step": 7125 }, { "epoch": 0.01, "learning_rate": 4.999390890635775e-05, "loss": 1.2602, "step": 7126 }, { "epoch": 0.01, "learning_rate": 4.999390717243242e-05, "loss": 1.1102, "step": 7127 }, { "epoch": 0.01, "learning_rate": 4.9993905438260366e-05, "loss": 1.1347, "step": 7128 }, { "epoch": 0.01, "learning_rate": 4.9993903703841584e-05, "loss": 0.888, "step": 7129 }, { "epoch": 0.01, "learning_rate": 4.999390196917607e-05, "loss": 0.718, "step": 7130 }, { "epoch": 0.01, "learning_rate": 4.999390023426383e-05, "loss": 0.4841, "step": 7131 }, { "epoch": 0.01, "learning_rate": 4.999389849910486e-05, "loss": 0.3925, "step": 7132 }, { "epoch": 0.01, "learning_rate": 4.9993896763699154e-05, "loss": 0.3584, "step": 7133 }, { "epoch": 0.01, "learning_rate": 4.999389502804673e-05, "loss": 0.2742, "step": 7134 }, { "epoch": 0.01, "learning_rate": 4.999389329214756e-05, "loss": 0.5736, "step": 7135 }, { "epoch": 0.01, "learning_rate": 4.999389155600168e-05, "loss": 1.1826, "step": 7136 }, { "epoch": 0.01, "learning_rate": 4.999388981960906e-05, "loss": 1.0934, "step": 7137 }, { "epoch": 0.01, "learning_rate": 4.999388808296971e-05, "loss": 1.0974, "step": 7138 }, { "epoch": 0.01, "learning_rate": 4.999388634608364e-05, "loss": 1.0185, "step": 7139 }, { "epoch": 0.01, "learning_rate": 4.9993884608950836e-05, "loss": 1.0648, "step": 7140 }, { "epoch": 0.01, "learning_rate": 4.99938828715713e-05, "loss": 1.1015, "step": 7141 }, { "epoch": 0.01, "learning_rate": 4.999388113394504e-05, "loss": 1.2112, "step": 7142 }, { "epoch": 0.01, "learning_rate": 4.999387939607205e-05, "loss": 1.2462, "step": 7143 }, { "epoch": 0.01, "learning_rate": 4.999387765795232e-05, "loss": 1.2127, "step": 7144 }, { "epoch": 0.01, "learning_rate": 4.9993875919585874e-05, "loss": 1.1119, "step": 7145 }, { "epoch": 0.01, "learning_rate": 4.9993874180972694e-05, "loss": 1.1006, "step": 7146 }, { "epoch": 0.01, "learning_rate": 4.999387244211279e-05, "loss": 1.3689, "step": 7147 }, { "epoch": 0.01, "learning_rate": 4.999387070300615e-05, "loss": 1.1952, "step": 7148 }, { "epoch": 0.01, "learning_rate": 4.999386896365278e-05, "loss": 1.1658, "step": 7149 }, { "epoch": 0.01, "learning_rate": 4.999386722405269e-05, "loss": 1.2052, "step": 7150 }, { "epoch": 0.01, "learning_rate": 4.999386548420586e-05, "loss": 1.1605, "step": 7151 }, { "epoch": 0.01, "learning_rate": 4.999386374411231e-05, "loss": 0.5754, "step": 7152 }, { "epoch": 0.01, "learning_rate": 4.999386200377203e-05, "loss": 1.0458, "step": 7153 }, { "epoch": 0.01, "learning_rate": 4.9993860263185015e-05, "loss": 1.1622, "step": 7154 }, { "epoch": 0.01, "learning_rate": 4.999385852235128e-05, "loss": 1.0383, "step": 7155 }, { "epoch": 0.01, "learning_rate": 4.999385678127081e-05, "loss": 0.9998, "step": 7156 }, { "epoch": 0.01, "learning_rate": 4.9993855039943613e-05, "loss": 0.9812, "step": 7157 }, { "epoch": 0.01, "learning_rate": 4.999385329836969e-05, "loss": 1.1803, "step": 7158 }, { "epoch": 0.01, "learning_rate": 4.999385155654903e-05, "loss": 1.0188, "step": 7159 }, { "epoch": 0.01, "learning_rate": 4.9993849814481644e-05, "loss": 0.9353, "step": 7160 }, { "epoch": 0.01, "learning_rate": 4.999384807216754e-05, "loss": 0.4585, "step": 7161 }, { "epoch": 0.01, "learning_rate": 4.999384632960669e-05, "loss": 0.2882, "step": 7162 }, { "epoch": 0.01, "learning_rate": 4.999384458679912e-05, "loss": 0.2708, "step": 7163 }, { "epoch": 0.01, "learning_rate": 4.9993842843744823e-05, "loss": 0.2178, "step": 7164 }, { "epoch": 0.01, "learning_rate": 4.99938411004438e-05, "loss": 0.3011, "step": 7165 }, { "epoch": 0.01, "learning_rate": 4.999383935689605e-05, "loss": 0.2672, "step": 7166 }, { "epoch": 0.01, "learning_rate": 4.999383761310156e-05, "loss": 0.1637, "step": 7167 }, { "epoch": 0.01, "learning_rate": 4.999383586906035e-05, "loss": 0.156, "step": 7168 }, { "epoch": 0.01, "learning_rate": 4.99938341247724e-05, "loss": 0.1112, "step": 7169 }, { "epoch": 0.01, "learning_rate": 4.9993832380237735e-05, "loss": 0.0966, "step": 7170 }, { "epoch": 0.01, "learning_rate": 4.999383063545633e-05, "loss": 0.071, "step": 7171 }, { "epoch": 0.01, "learning_rate": 4.9993828890428206e-05, "loss": 0.6251, "step": 7172 }, { "epoch": 0.01, "learning_rate": 4.999382714515335e-05, "loss": 0.5264, "step": 7173 }, { "epoch": 0.01, "learning_rate": 4.999382539963177e-05, "loss": 0.9327, "step": 7174 }, { "epoch": 0.01, "learning_rate": 4.999382365386345e-05, "loss": 1.2255, "step": 7175 }, { "epoch": 0.01, "learning_rate": 4.999382190784841e-05, "loss": 1.1865, "step": 7176 }, { "epoch": 0.01, "learning_rate": 4.9993820161586636e-05, "loss": 1.1373, "step": 7177 }, { "epoch": 0.01, "learning_rate": 4.999381841507814e-05, "loss": 0.9976, "step": 7178 }, { "epoch": 0.01, "learning_rate": 4.999381666832291e-05, "loss": 1.1255, "step": 7179 }, { "epoch": 0.01, "learning_rate": 4.9993814921320956e-05, "loss": 1.2367, "step": 7180 }, { "epoch": 0.01, "learning_rate": 4.9993813174072275e-05, "loss": 1.0031, "step": 7181 }, { "epoch": 0.01, "learning_rate": 4.9993811426576864e-05, "loss": 1.0087, "step": 7182 }, { "epoch": 0.01, "learning_rate": 4.999380967883472e-05, "loss": 1.1194, "step": 7183 }, { "epoch": 0.01, "learning_rate": 4.999380793084585e-05, "loss": 1.1359, "step": 7184 }, { "epoch": 0.01, "learning_rate": 4.999380618261025e-05, "loss": 1.5818, "step": 7185 }, { "epoch": 0.01, "learning_rate": 4.9993804434127926e-05, "loss": 1.4419, "step": 7186 }, { "epoch": 0.01, "learning_rate": 4.999380268539887e-05, "loss": 1.1241, "step": 7187 }, { "epoch": 0.01, "learning_rate": 4.999380093642309e-05, "loss": 1.1122, "step": 7188 }, { "epoch": 0.01, "learning_rate": 4.999379918720058e-05, "loss": 1.3919, "step": 7189 }, { "epoch": 0.01, "learning_rate": 4.999379743773134e-05, "loss": 1.2955, "step": 7190 }, { "epoch": 0.01, "learning_rate": 4.9993795688015375e-05, "loss": 1.2744, "step": 7191 }, { "epoch": 0.01, "learning_rate": 4.999379393805268e-05, "loss": 1.4393, "step": 7192 }, { "epoch": 0.01, "learning_rate": 4.999379218784326e-05, "loss": 1.3003, "step": 7193 }, { "epoch": 0.01, "learning_rate": 4.999379043738711e-05, "loss": 1.6819, "step": 7194 }, { "epoch": 0.01, "learning_rate": 4.9993788686684226e-05, "loss": 1.8338, "step": 7195 }, { "epoch": 0.01, "learning_rate": 4.999378693573462e-05, "loss": 1.4827, "step": 7196 }, { "epoch": 0.01, "learning_rate": 4.999378518453828e-05, "loss": 0.9903, "step": 7197 }, { "epoch": 0.01, "learning_rate": 4.999378343309522e-05, "loss": 1.3554, "step": 7198 }, { "epoch": 0.01, "learning_rate": 4.999378168140543e-05, "loss": 1.2596, "step": 7199 }, { "epoch": 0.01, "learning_rate": 4.99937799294689e-05, "loss": 1.0637, "step": 7200 }, { "epoch": 0.01, "learning_rate": 4.999377817728566e-05, "loss": 1.4157, "step": 7201 }, { "epoch": 0.01, "learning_rate": 4.999377642485568e-05, "loss": 1.2872, "step": 7202 }, { "epoch": 0.01, "learning_rate": 4.999377467217897e-05, "loss": 5.7376, "step": 7203 }, { "epoch": 0.01, "learning_rate": 4.999377291925554e-05, "loss": 5.4455, "step": 7204 }, { "epoch": 0.01, "learning_rate": 4.9993771166085376e-05, "loss": 1.0564, "step": 7205 }, { "epoch": 0.01, "learning_rate": 4.999376941266849e-05, "loss": 0.7091, "step": 7206 }, { "epoch": 0.01, "learning_rate": 4.999376765900487e-05, "loss": 1.0153, "step": 7207 }, { "epoch": 0.01, "learning_rate": 4.9993765905094535e-05, "loss": 1.5328, "step": 7208 }, { "epoch": 0.01, "learning_rate": 4.999376415093746e-05, "loss": 0.9528, "step": 7209 }, { "epoch": 0.01, "learning_rate": 4.999376239653366e-05, "loss": 1.1227, "step": 7210 }, { "epoch": 0.01, "learning_rate": 4.999376064188313e-05, "loss": 0.9175, "step": 7211 }, { "epoch": 0.01, "learning_rate": 4.999375888698587e-05, "loss": 1.1229, "step": 7212 }, { "epoch": 0.01, "learning_rate": 4.999375713184189e-05, "loss": 1.1384, "step": 7213 }, { "epoch": 0.01, "learning_rate": 4.999375537645118e-05, "loss": 1.18, "step": 7214 }, { "epoch": 0.01, "learning_rate": 4.9993753620813744e-05, "loss": 1.2178, "step": 7215 }, { "epoch": 0.01, "learning_rate": 4.999375186492957e-05, "loss": 1.4092, "step": 7216 }, { "epoch": 0.01, "learning_rate": 4.999375010879868e-05, "loss": 0.9861, "step": 7217 }, { "epoch": 0.01, "learning_rate": 4.9993748352421055e-05, "loss": 1.0413, "step": 7218 }, { "epoch": 0.01, "learning_rate": 4.99937465957967e-05, "loss": 1.1335, "step": 7219 }, { "epoch": 0.01, "learning_rate": 4.9993744838925625e-05, "loss": 0.6983, "step": 7220 }, { "epoch": 0.01, "learning_rate": 4.999374308180782e-05, "loss": 0.8774, "step": 7221 }, { "epoch": 0.01, "learning_rate": 4.9993741324443286e-05, "loss": 1.1254, "step": 7222 }, { "epoch": 0.01, "learning_rate": 4.999373956683202e-05, "loss": 0.427, "step": 7223 }, { "epoch": 0.01, "learning_rate": 4.999373780897404e-05, "loss": 0.367, "step": 7224 }, { "epoch": 0.01, "learning_rate": 4.999373605086932e-05, "loss": 0.356, "step": 7225 }, { "epoch": 0.01, "learning_rate": 4.999373429251788e-05, "loss": 0.3391, "step": 7226 }, { "epoch": 0.01, "learning_rate": 4.99937325339197e-05, "loss": 0.377, "step": 7227 }, { "epoch": 0.01, "learning_rate": 4.99937307750748e-05, "loss": 0.323, "step": 7228 }, { "epoch": 0.01, "learning_rate": 4.9993729015983174e-05, "loss": 0.293, "step": 7229 }, { "epoch": 0.01, "learning_rate": 4.999372725664482e-05, "loss": 0.2573, "step": 7230 }, { "epoch": 0.01, "learning_rate": 4.999372549705974e-05, "loss": 0.9169, "step": 7231 }, { "epoch": 0.01, "learning_rate": 4.9993723737227924e-05, "loss": 0.7361, "step": 7232 }, { "epoch": 0.01, "learning_rate": 4.999372197714939e-05, "loss": 0.9766, "step": 7233 }, { "epoch": 0.01, "learning_rate": 4.999372021682413e-05, "loss": 1.1306, "step": 7234 }, { "epoch": 0.01, "learning_rate": 4.999371845625213e-05, "loss": 1.3027, "step": 7235 }, { "epoch": 0.01, "learning_rate": 4.9993716695433415e-05, "loss": 0.7542, "step": 7236 }, { "epoch": 0.01, "learning_rate": 4.999371493436796e-05, "loss": 1.1492, "step": 7237 }, { "epoch": 0.01, "learning_rate": 4.999371317305579e-05, "loss": 1.1588, "step": 7238 }, { "epoch": 0.01, "learning_rate": 4.999371141149689e-05, "loss": 0.913, "step": 7239 }, { "epoch": 0.01, "learning_rate": 4.999370964969126e-05, "loss": 1.1554, "step": 7240 }, { "epoch": 0.01, "learning_rate": 4.999370788763891e-05, "loss": 1.176, "step": 7241 }, { "epoch": 0.01, "learning_rate": 4.999370612533982e-05, "loss": 1.0936, "step": 7242 }, { "epoch": 0.01, "learning_rate": 4.999370436279401e-05, "loss": 1.0287, "step": 7243 }, { "epoch": 0.01, "learning_rate": 4.999370260000147e-05, "loss": 1.0089, "step": 7244 }, { "epoch": 0.01, "learning_rate": 4.99937008369622e-05, "loss": 1.0935, "step": 7245 }, { "epoch": 0.01, "learning_rate": 4.9993699073676206e-05, "loss": 1.1778, "step": 7246 }, { "epoch": 0.01, "learning_rate": 4.9993697310143486e-05, "loss": 0.9498, "step": 7247 }, { "epoch": 0.01, "learning_rate": 4.999369554636404e-05, "loss": 1.0388, "step": 7248 }, { "epoch": 0.01, "learning_rate": 4.999369378233787e-05, "loss": 1.2277, "step": 7249 }, { "epoch": 0.01, "learning_rate": 4.999369201806496e-05, "loss": 1.0388, "step": 7250 }, { "epoch": 0.01, "learning_rate": 4.9993690253545335e-05, "loss": 1.2701, "step": 7251 }, { "epoch": 0.01, "learning_rate": 4.9993688488778976e-05, "loss": 1.0766, "step": 7252 }, { "epoch": 0.01, "learning_rate": 4.99936867237659e-05, "loss": 0.5179, "step": 7253 }, { "epoch": 0.01, "learning_rate": 4.999368495850608e-05, "loss": 0.7307, "step": 7254 }, { "epoch": 0.01, "learning_rate": 4.999368319299954e-05, "loss": 1.1805, "step": 7255 }, { "epoch": 0.01, "learning_rate": 4.999368142724628e-05, "loss": 0.9093, "step": 7256 }, { "epoch": 0.01, "learning_rate": 4.9993679661246284e-05, "loss": 1.1146, "step": 7257 }, { "epoch": 0.01, "learning_rate": 4.999367789499957e-05, "loss": 1.3203, "step": 7258 }, { "epoch": 0.01, "learning_rate": 4.9993676128506116e-05, "loss": 1.4558, "step": 7259 }, { "epoch": 0.01, "learning_rate": 4.9993674361765947e-05, "loss": 0.7744, "step": 7260 }, { "epoch": 0.01, "learning_rate": 4.999367259477905e-05, "loss": 0.8, "step": 7261 }, { "epoch": 0.01, "learning_rate": 4.9993670827545416e-05, "loss": 0.8254, "step": 7262 }, { "epoch": 0.01, "learning_rate": 4.999366906006506e-05, "loss": 0.9704, "step": 7263 }, { "epoch": 0.01, "learning_rate": 4.9993667292337984e-05, "loss": 1.1857, "step": 7264 }, { "epoch": 0.01, "learning_rate": 4.9993665524364175e-05, "loss": 1.0887, "step": 7265 }, { "epoch": 0.01, "learning_rate": 4.9993663756143636e-05, "loss": 1.0521, "step": 7266 }, { "epoch": 0.01, "learning_rate": 4.9993661987676373e-05, "loss": 1.1798, "step": 7267 }, { "epoch": 0.01, "learning_rate": 4.9993660218962394e-05, "loss": 0.878, "step": 7268 }, { "epoch": 0.01, "learning_rate": 4.999365845000167e-05, "loss": 1.4214, "step": 7269 }, { "epoch": 0.01, "learning_rate": 4.999365668079423e-05, "loss": 1.3913, "step": 7270 }, { "epoch": 0.01, "learning_rate": 4.999365491134006e-05, "loss": 1.4542, "step": 7271 }, { "epoch": 0.01, "learning_rate": 4.999365314163916e-05, "loss": 1.3582, "step": 7272 }, { "epoch": 0.01, "learning_rate": 4.9993651371691543e-05, "loss": 1.3169, "step": 7273 }, { "epoch": 0.01, "learning_rate": 4.999364960149719e-05, "loss": 0.9608, "step": 7274 }, { "epoch": 0.01, "learning_rate": 4.9993647831056114e-05, "loss": 1.1639, "step": 7275 }, { "epoch": 0.01, "learning_rate": 4.9993646060368303e-05, "loss": 1.1301, "step": 7276 }, { "epoch": 0.01, "learning_rate": 4.9993644289433776e-05, "loss": 1.0542, "step": 7277 }, { "epoch": 0.01, "learning_rate": 4.999364251825252e-05, "loss": 1.14, "step": 7278 }, { "epoch": 0.01, "learning_rate": 4.999364074682453e-05, "loss": 1.0161, "step": 7279 }, { "epoch": 0.01, "learning_rate": 4.9993638975149824e-05, "loss": 0.9362, "step": 7280 }, { "epoch": 0.01, "learning_rate": 4.999363720322839e-05, "loss": 0.9912, "step": 7281 }, { "epoch": 0.01, "learning_rate": 4.999363543106023e-05, "loss": 0.5056, "step": 7282 }, { "epoch": 0.01, "learning_rate": 4.999363365864533e-05, "loss": 0.7172, "step": 7283 }, { "epoch": 0.01, "learning_rate": 4.999363188598372e-05, "loss": 1.1353, "step": 7284 }, { "epoch": 0.01, "learning_rate": 4.999363011307537e-05, "loss": 0.7113, "step": 7285 }, { "epoch": 0.01, "learning_rate": 4.9993628339920304e-05, "loss": 0.8874, "step": 7286 }, { "epoch": 0.01, "learning_rate": 4.9993626566518505e-05, "loss": 0.7167, "step": 7287 }, { "epoch": 0.01, "learning_rate": 4.999362479286999e-05, "loss": 1.0762, "step": 7288 }, { "epoch": 0.01, "learning_rate": 4.9993623018974736e-05, "loss": 1.2297, "step": 7289 }, { "epoch": 0.01, "learning_rate": 4.999362124483276e-05, "loss": 1.1268, "step": 7290 }, { "epoch": 0.01, "learning_rate": 4.999361947044405e-05, "loss": 1.0713, "step": 7291 }, { "epoch": 0.01, "learning_rate": 4.9993617695808626e-05, "loss": 0.9842, "step": 7292 }, { "epoch": 0.01, "learning_rate": 4.9993615920926465e-05, "loss": 1.1844, "step": 7293 }, { "epoch": 0.01, "learning_rate": 4.9993614145797586e-05, "loss": 1.1295, "step": 7294 }, { "epoch": 0.01, "learning_rate": 4.9993612370421977e-05, "loss": 1.3198, "step": 7295 }, { "epoch": 0.01, "learning_rate": 4.9993610594799643e-05, "loss": 1.1933, "step": 7296 }, { "epoch": 0.01, "learning_rate": 4.999360881893058e-05, "loss": 1.1275, "step": 7297 }, { "epoch": 0.01, "learning_rate": 4.999360704281479e-05, "loss": 0.9787, "step": 7298 }, { "epoch": 0.01, "learning_rate": 4.9993605266452275e-05, "loss": 0.9672, "step": 7299 }, { "epoch": 0.01, "learning_rate": 4.999360348984303e-05, "loss": 1.1196, "step": 7300 }, { "epoch": 0.01, "learning_rate": 4.999360171298707e-05, "loss": 1.2345, "step": 7301 }, { "epoch": 0.01, "learning_rate": 4.999359993588437e-05, "loss": 1.1671, "step": 7302 }, { "epoch": 0.01, "learning_rate": 4.999359815853495e-05, "loss": 0.6403, "step": 7303 }, { "epoch": 0.01, "learning_rate": 4.9993596380938815e-05, "loss": 1.2603, "step": 7304 }, { "epoch": 0.01, "learning_rate": 4.999359460309594e-05, "loss": 1.2867, "step": 7305 }, { "epoch": 0.01, "learning_rate": 4.999359282500634e-05, "loss": 1.336, "step": 7306 }, { "epoch": 0.01, "learning_rate": 4.999359104667002e-05, "loss": 1.2513, "step": 7307 }, { "epoch": 0.01, "learning_rate": 4.999358926808696e-05, "loss": 1.0588, "step": 7308 }, { "epoch": 0.01, "learning_rate": 4.999358748925719e-05, "loss": 1.2615, "step": 7309 }, { "epoch": 0.01, "learning_rate": 4.999358571018069e-05, "loss": 1.0442, "step": 7310 }, { "epoch": 0.01, "learning_rate": 4.999358393085746e-05, "loss": 1.1173, "step": 7311 }, { "epoch": 0.01, "learning_rate": 4.99935821512875e-05, "loss": 0.9781, "step": 7312 }, { "epoch": 0.01, "learning_rate": 4.999358037147082e-05, "loss": 1.3245, "step": 7313 }, { "epoch": 0.01, "learning_rate": 4.999357859140741e-05, "loss": 1.2609, "step": 7314 }, { "epoch": 0.01, "learning_rate": 4.999357681109727e-05, "loss": 0.613, "step": 7315 }, { "epoch": 0.01, "learning_rate": 4.999357503054042e-05, "loss": 0.7257, "step": 7316 }, { "epoch": 0.01, "learning_rate": 4.999357324973683e-05, "loss": 1.2196, "step": 7317 }, { "epoch": 0.01, "learning_rate": 4.999357146868652e-05, "loss": 1.1914, "step": 7318 }, { "epoch": 0.01, "learning_rate": 4.999356968738948e-05, "loss": 1.2337, "step": 7319 }, { "epoch": 0.01, "learning_rate": 4.9993567905845716e-05, "loss": 1.654, "step": 7320 }, { "epoch": 0.01, "learning_rate": 4.999356612405522e-05, "loss": 1.7418, "step": 7321 }, { "epoch": 0.01, "learning_rate": 4.999356434201801e-05, "loss": 1.6275, "step": 7322 }, { "epoch": 0.01, "learning_rate": 4.999356255973407e-05, "loss": 1.6289, "step": 7323 }, { "epoch": 0.01, "learning_rate": 4.99935607772034e-05, "loss": 1.5845, "step": 7324 }, { "epoch": 0.01, "learning_rate": 4.9993558994426005e-05, "loss": 1.559, "step": 7325 }, { "epoch": 0.01, "learning_rate": 4.9993557211401884e-05, "loss": 1.5541, "step": 7326 }, { "epoch": 0.01, "learning_rate": 4.999355542813105e-05, "loss": 1.5677, "step": 7327 }, { "epoch": 0.01, "learning_rate": 4.999355364461347e-05, "loss": 1.5506, "step": 7328 }, { "epoch": 0.01, "learning_rate": 4.9993551860849174e-05, "loss": 1.5629, "step": 7329 }, { "epoch": 0.01, "learning_rate": 4.999355007683815e-05, "loss": 1.5482, "step": 7330 }, { "epoch": 0.01, "learning_rate": 4.9993548292580406e-05, "loss": 1.5299, "step": 7331 }, { "epoch": 0.01, "learning_rate": 4.999354650807593e-05, "loss": 1.5587, "step": 7332 }, { "epoch": 0.01, "learning_rate": 4.999354472332473e-05, "loss": 1.4898, "step": 7333 }, { "epoch": 0.01, "learning_rate": 4.9993542938326805e-05, "loss": 1.5037, "step": 7334 }, { "epoch": 0.01, "learning_rate": 4.999354115308215e-05, "loss": 1.1999, "step": 7335 }, { "epoch": 0.01, "learning_rate": 4.999353936759077e-05, "loss": 1.4152, "step": 7336 }, { "epoch": 0.01, "learning_rate": 4.999353758185267e-05, "loss": 1.5625, "step": 7337 }, { "epoch": 0.01, "learning_rate": 4.999353579586784e-05, "loss": 1.3958, "step": 7338 }, { "epoch": 0.01, "learning_rate": 4.999353400963629e-05, "loss": 1.3154, "step": 7339 }, { "epoch": 0.01, "learning_rate": 4.9993532223158015e-05, "loss": 1.3841, "step": 7340 }, { "epoch": 0.01, "learning_rate": 4.9993530436433e-05, "loss": 1.541, "step": 7341 }, { "epoch": 0.01, "learning_rate": 4.999352864946128e-05, "loss": 1.4549, "step": 7342 }, { "epoch": 0.01, "learning_rate": 4.999352686224281e-05, "loss": 1.405, "step": 7343 }, { "epoch": 0.01, "learning_rate": 4.999352507477764e-05, "loss": 1.5148, "step": 7344 }, { "epoch": 0.01, "learning_rate": 4.9993523287065725e-05, "loss": 1.5382, "step": 7345 }, { "epoch": 0.01, "learning_rate": 4.9993521499107096e-05, "loss": 1.3132, "step": 7346 }, { "epoch": 0.01, "learning_rate": 4.9993519710901736e-05, "loss": 1.3158, "step": 7347 }, { "epoch": 0.01, "learning_rate": 4.999351792244965e-05, "loss": 1.178, "step": 7348 }, { "epoch": 0.01, "learning_rate": 4.999351613375084e-05, "loss": 1.2829, "step": 7349 }, { "epoch": 0.01, "learning_rate": 4.99935143448053e-05, "loss": 1.3511, "step": 7350 }, { "epoch": 0.01, "learning_rate": 4.9993512555613045e-05, "loss": 1.2952, "step": 7351 }, { "epoch": 0.01, "learning_rate": 4.999351076617406e-05, "loss": 1.3797, "step": 7352 }, { "epoch": 0.01, "learning_rate": 4.9993508976488344e-05, "loss": 1.5269, "step": 7353 }, { "epoch": 0.01, "learning_rate": 4.999350718655591e-05, "loss": 1.5188, "step": 7354 }, { "epoch": 0.01, "learning_rate": 4.999350539637675e-05, "loss": 1.4032, "step": 7355 }, { "epoch": 0.01, "learning_rate": 4.999350360595086e-05, "loss": 1.6251, "step": 7356 }, { "epoch": 0.01, "learning_rate": 4.999350181527824e-05, "loss": 1.6092, "step": 7357 }, { "epoch": 0.01, "learning_rate": 4.999350002435891e-05, "loss": 1.2604, "step": 7358 }, { "epoch": 0.01, "learning_rate": 4.999349823319284e-05, "loss": 1.0308, "step": 7359 }, { "epoch": 0.01, "learning_rate": 4.999349644178005e-05, "loss": 1.3368, "step": 7360 }, { "epoch": 0.01, "learning_rate": 4.999349465012054e-05, "loss": 1.6052, "step": 7361 }, { "epoch": 0.01, "learning_rate": 4.9993492858214304e-05, "loss": 1.3298, "step": 7362 }, { "epoch": 0.01, "learning_rate": 4.999349106606134e-05, "loss": 1.5927, "step": 7363 }, { "epoch": 0.01, "learning_rate": 4.9993489273661646e-05, "loss": 1.3533, "step": 7364 }, { "epoch": 0.01, "learning_rate": 4.999348748101523e-05, "loss": 1.6058, "step": 7365 }, { "epoch": 0.01, "learning_rate": 4.9993485688122086e-05, "loss": 1.5934, "step": 7366 }, { "epoch": 0.01, "learning_rate": 4.999348389498223e-05, "loss": 1.3531, "step": 7367 }, { "epoch": 0.01, "learning_rate": 4.999348210159563e-05, "loss": 1.366, "step": 7368 }, { "epoch": 0.01, "learning_rate": 4.999348030796232e-05, "loss": 1.3443, "step": 7369 }, { "epoch": 0.01, "learning_rate": 4.999347851408228e-05, "loss": 1.5869, "step": 7370 }, { "epoch": 0.01, "learning_rate": 4.999347671995551e-05, "loss": 1.5461, "step": 7371 }, { "epoch": 0.01, "learning_rate": 4.999347492558202e-05, "loss": 1.593, "step": 7372 }, { "epoch": 0.01, "learning_rate": 4.999347313096181e-05, "loss": 1.4488, "step": 7373 }, { "epoch": 0.01, "learning_rate": 4.999347133609486e-05, "loss": 1.613, "step": 7374 }, { "epoch": 0.01, "learning_rate": 4.99934695409812e-05, "loss": 1.4047, "step": 7375 }, { "epoch": 0.01, "learning_rate": 4.999346774562081e-05, "loss": 1.6145, "step": 7376 }, { "epoch": 0.01, "learning_rate": 4.9993465950013694e-05, "loss": 1.3279, "step": 7377 }, { "epoch": 0.01, "learning_rate": 4.999346415415985e-05, "loss": 1.509, "step": 7378 }, { "epoch": 0.01, "learning_rate": 4.999346235805928e-05, "loss": 1.4429, "step": 7379 }, { "epoch": 0.01, "learning_rate": 4.999346056171199e-05, "loss": 1.5884, "step": 7380 }, { "epoch": 0.01, "learning_rate": 4.999345876511798e-05, "loss": 1.5937, "step": 7381 }, { "epoch": 0.01, "learning_rate": 4.9993456968277245e-05, "loss": 1.607, "step": 7382 }, { "epoch": 0.01, "learning_rate": 4.9993455171189775e-05, "loss": 1.5965, "step": 7383 }, { "epoch": 0.01, "learning_rate": 4.999345337385558e-05, "loss": 1.5968, "step": 7384 }, { "epoch": 0.01, "learning_rate": 4.9993451576274665e-05, "loss": 1.5513, "step": 7385 }, { "epoch": 0.01, "learning_rate": 4.9993449778447025e-05, "loss": 1.222, "step": 7386 }, { "epoch": 0.01, "learning_rate": 4.999344798037267e-05, "loss": 1.3366, "step": 7387 }, { "epoch": 0.01, "learning_rate": 4.999344618205157e-05, "loss": 1.3441, "step": 7388 }, { "epoch": 0.01, "learning_rate": 4.9993444383483754e-05, "loss": 1.6091, "step": 7389 }, { "epoch": 0.01, "learning_rate": 4.999344258466921e-05, "loss": 1.3621, "step": 7390 }, { "epoch": 0.01, "learning_rate": 4.999344078560795e-05, "loss": 1.5896, "step": 7391 }, { "epoch": 0.01, "learning_rate": 4.999343898629997e-05, "loss": 1.5968, "step": 7392 }, { "epoch": 0.01, "learning_rate": 4.999343718674525e-05, "loss": 1.3465, "step": 7393 }, { "epoch": 0.01, "learning_rate": 4.999343538694382e-05, "loss": 1.5791, "step": 7394 }, { "epoch": 0.01, "learning_rate": 4.9993433586895656e-05, "loss": 1.3193, "step": 7395 }, { "epoch": 0.01, "learning_rate": 4.9993431786600765e-05, "loss": 1.5984, "step": 7396 }, { "epoch": 0.01, "learning_rate": 4.9993429986059156e-05, "loss": 1.0591, "step": 7397 }, { "epoch": 0.01, "learning_rate": 4.999342818527082e-05, "loss": 1.6761, "step": 7398 }, { "epoch": 0.01, "learning_rate": 4.9993426384235754e-05, "loss": 1.3947, "step": 7399 }, { "epoch": 0.01, "learning_rate": 4.999342458295397e-05, "loss": 1.146, "step": 7400 }, { "epoch": 0.01, "learning_rate": 4.9993422781425456e-05, "loss": 1.0649, "step": 7401 }, { "epoch": 0.01, "learning_rate": 4.999342097965022e-05, "loss": 1.2759, "step": 7402 }, { "epoch": 0.01, "learning_rate": 4.9993419177628264e-05, "loss": 1.3245, "step": 7403 }, { "epoch": 0.01, "learning_rate": 4.999341737535958e-05, "loss": 1.3419, "step": 7404 }, { "epoch": 0.01, "learning_rate": 4.999341557284417e-05, "loss": 1.2456, "step": 7405 }, { "epoch": 0.01, "learning_rate": 4.999341377008204e-05, "loss": 1.2134, "step": 7406 }, { "epoch": 0.01, "learning_rate": 4.999341196707318e-05, "loss": 1.0812, "step": 7407 }, { "epoch": 0.01, "learning_rate": 4.99934101638176e-05, "loss": 0.9918, "step": 7408 }, { "epoch": 0.01, "learning_rate": 4.99934083603153e-05, "loss": 0.9686, "step": 7409 }, { "epoch": 0.01, "learning_rate": 4.999340655656627e-05, "loss": 1.0861, "step": 7410 }, { "epoch": 0.01, "learning_rate": 4.999340475257051e-05, "loss": 1.0647, "step": 7411 }, { "epoch": 0.01, "learning_rate": 4.9993402948328025e-05, "loss": 1.311, "step": 7412 }, { "epoch": 0.01, "learning_rate": 4.999340114383882e-05, "loss": 1.3783, "step": 7413 }, { "epoch": 0.01, "learning_rate": 4.99933993391029e-05, "loss": 1.2996, "step": 7414 }, { "epoch": 0.01, "learning_rate": 4.999339753412025e-05, "loss": 1.0855, "step": 7415 }, { "epoch": 0.01, "learning_rate": 4.9993395728890866e-05, "loss": 1.189, "step": 7416 }, { "epoch": 0.01, "learning_rate": 4.999339392341477e-05, "loss": 1.3219, "step": 7417 }, { "epoch": 0.01, "learning_rate": 4.999339211769195e-05, "loss": 0.8666, "step": 7418 }, { "epoch": 0.01, "learning_rate": 4.9993390311722396e-05, "loss": 1.1259, "step": 7419 }, { "epoch": 0.01, "learning_rate": 4.999338850550612e-05, "loss": 1.1693, "step": 7420 }, { "epoch": 0.01, "learning_rate": 4.999338669904312e-05, "loss": 1.8688, "step": 7421 }, { "epoch": 0.01, "learning_rate": 4.9993384892333405e-05, "loss": 1.2217, "step": 7422 }, { "epoch": 0.01, "learning_rate": 4.999338308537695e-05, "loss": 0.9835, "step": 7423 }, { "epoch": 0.01, "learning_rate": 4.999338127817379e-05, "loss": 1.1256, "step": 7424 }, { "epoch": 0.01, "learning_rate": 4.999337947072389e-05, "loss": 1.0414, "step": 7425 }, { "epoch": 0.01, "learning_rate": 4.9993377663027276e-05, "loss": 0.865, "step": 7426 }, { "epoch": 0.01, "learning_rate": 4.9993375855083934e-05, "loss": 1.0807, "step": 7427 }, { "epoch": 0.01, "learning_rate": 4.999337404689386e-05, "loss": 1.2146, "step": 7428 }, { "epoch": 0.01, "learning_rate": 4.999337223845707e-05, "loss": 1.1836, "step": 7429 }, { "epoch": 0.01, "learning_rate": 4.999337042977355e-05, "loss": 1.1424, "step": 7430 }, { "epoch": 0.01, "learning_rate": 4.999336862084332e-05, "loss": 1.043, "step": 7431 }, { "epoch": 0.01, "learning_rate": 4.999336681166636e-05, "loss": 1.2966, "step": 7432 }, { "epoch": 0.01, "learning_rate": 4.999336500224268e-05, "loss": 1.5118, "step": 7433 }, { "epoch": 0.01, "learning_rate": 4.999336319257226e-05, "loss": 1.3014, "step": 7434 }, { "epoch": 0.01, "learning_rate": 4.9993361382655125e-05, "loss": 1.1033, "step": 7435 }, { "epoch": 0.01, "learning_rate": 4.999335957249127e-05, "loss": 1.3528, "step": 7436 }, { "epoch": 0.01, "learning_rate": 4.999335776208068e-05, "loss": 1.185, "step": 7437 }, { "epoch": 0.01, "learning_rate": 4.999335595142338e-05, "loss": 1.1749, "step": 7438 }, { "epoch": 0.01, "learning_rate": 4.9993354140519346e-05, "loss": 1.037, "step": 7439 }, { "epoch": 0.01, "learning_rate": 4.9993352329368596e-05, "loss": 0.9035, "step": 7440 }, { "epoch": 0.01, "learning_rate": 4.9993350517971115e-05, "loss": 1.3873, "step": 7441 }, { "epoch": 0.01, "learning_rate": 4.9993348706326916e-05, "loss": 1.1147, "step": 7442 }, { "epoch": 0.01, "learning_rate": 4.999334689443599e-05, "loss": 1.264, "step": 7443 }, { "epoch": 0.01, "learning_rate": 4.9993345082298335e-05, "loss": 2.0064, "step": 7444 }, { "epoch": 0.01, "learning_rate": 4.9993343269913966e-05, "loss": 1.4161, "step": 7445 }, { "epoch": 0.01, "learning_rate": 4.999334145728287e-05, "loss": 0.7776, "step": 7446 }, { "epoch": 0.01, "learning_rate": 4.999333964440505e-05, "loss": 0.6111, "step": 7447 }, { "epoch": 0.01, "learning_rate": 4.999333783128051e-05, "loss": 0.6493, "step": 7448 }, { "epoch": 0.01, "learning_rate": 4.9993336017909244e-05, "loss": 0.6455, "step": 7449 }, { "epoch": 0.01, "learning_rate": 4.999333420429125e-05, "loss": 0.602, "step": 7450 }, { "epoch": 0.01, "learning_rate": 4.999333239042653e-05, "loss": 0.603, "step": 7451 }, { "epoch": 0.01, "learning_rate": 4.999333057631509e-05, "loss": 0.5794, "step": 7452 }, { "epoch": 0.01, "learning_rate": 4.999332876195693e-05, "loss": 0.5581, "step": 7453 }, { "epoch": 0.01, "learning_rate": 4.9993326947352046e-05, "loss": 0.5843, "step": 7454 }, { "epoch": 0.01, "learning_rate": 4.999332513250043e-05, "loss": 0.5669, "step": 7455 }, { "epoch": 0.01, "learning_rate": 4.9993323317402094e-05, "loss": 0.5727, "step": 7456 }, { "epoch": 0.01, "learning_rate": 4.999332150205705e-05, "loss": 0.5761, "step": 7457 }, { "epoch": 0.01, "learning_rate": 4.999331968646526e-05, "loss": 0.5629, "step": 7458 }, { "epoch": 0.01, "learning_rate": 4.999331787062676e-05, "loss": 0.5799, "step": 7459 }, { "epoch": 0.01, "learning_rate": 4.999331605454153e-05, "loss": 0.5618, "step": 7460 }, { "epoch": 0.01, "learning_rate": 4.9993314238209585e-05, "loss": 0.8993, "step": 7461 }, { "epoch": 0.01, "learning_rate": 4.9993312421630904e-05, "loss": 1.4123, "step": 7462 }, { "epoch": 0.01, "learning_rate": 4.999331060480551e-05, "loss": 1.0034, "step": 7463 }, { "epoch": 0.01, "learning_rate": 4.999330878773339e-05, "loss": 1.3514, "step": 7464 }, { "epoch": 0.01, "learning_rate": 4.999330697041454e-05, "loss": 1.3569, "step": 7465 }, { "epoch": 0.01, "learning_rate": 4.9993305152848974e-05, "loss": 1.0425, "step": 7466 }, { "epoch": 0.01, "learning_rate": 4.999330333503668e-05, "loss": 1.0502, "step": 7467 }, { "epoch": 0.01, "learning_rate": 4.9993301516977666e-05, "loss": 1.1806, "step": 7468 }, { "epoch": 0.01, "learning_rate": 4.9993299698671933e-05, "loss": 1.0627, "step": 7469 }, { "epoch": 0.01, "learning_rate": 4.999329788011947e-05, "loss": 0.986, "step": 7470 }, { "epoch": 0.01, "learning_rate": 4.999329606132028e-05, "loss": 0.9962, "step": 7471 }, { "epoch": 0.01, "learning_rate": 4.999329424227437e-05, "loss": 1.2365, "step": 7472 }, { "epoch": 0.01, "learning_rate": 4.9993292422981744e-05, "loss": 1.1068, "step": 7473 }, { "epoch": 0.01, "learning_rate": 4.9993290603442386e-05, "loss": 1.3851, "step": 7474 }, { "epoch": 0.01, "learning_rate": 4.999328878365631e-05, "loss": 1.3683, "step": 7475 }, { "epoch": 0.01, "learning_rate": 4.999328696362351e-05, "loss": 1.3053, "step": 7476 }, { "epoch": 0.01, "learning_rate": 4.999328514334398e-05, "loss": 0.757, "step": 7477 }, { "epoch": 0.01, "learning_rate": 4.999328332281774e-05, "loss": 0.6441, "step": 7478 }, { "epoch": 0.01, "learning_rate": 4.9993281502044765e-05, "loss": 0.9481, "step": 7479 }, { "epoch": 0.01, "learning_rate": 4.999327968102507e-05, "loss": 1.1469, "step": 7480 }, { "epoch": 0.01, "learning_rate": 4.999327785975866e-05, "loss": 1.0571, "step": 7481 }, { "epoch": 0.01, "learning_rate": 4.999327603824552e-05, "loss": 0.7546, "step": 7482 }, { "epoch": 0.01, "learning_rate": 4.999327421648565e-05, "loss": 0.6546, "step": 7483 }, { "epoch": 0.01, "learning_rate": 4.999327239447907e-05, "loss": 1.1642, "step": 7484 }, { "epoch": 0.01, "learning_rate": 4.999327057222576e-05, "loss": 1.0698, "step": 7485 }, { "epoch": 0.01, "learning_rate": 4.999326874972573e-05, "loss": 1.1704, "step": 7486 }, { "epoch": 0.01, "learning_rate": 4.999326692697897e-05, "loss": 1.2712, "step": 7487 }, { "epoch": 0.01, "learning_rate": 4.999326510398549e-05, "loss": 0.9271, "step": 7488 }, { "epoch": 0.01, "learning_rate": 4.9993263280745296e-05, "loss": 1.1715, "step": 7489 }, { "epoch": 0.01, "learning_rate": 4.9993261457258364e-05, "loss": 1.149, "step": 7490 }, { "epoch": 0.01, "learning_rate": 4.999325963352472e-05, "loss": 1.0354, "step": 7491 }, { "epoch": 0.01, "learning_rate": 4.999325780954435e-05, "loss": 1.1546, "step": 7492 }, { "epoch": 0.01, "learning_rate": 4.9993255985317254e-05, "loss": 1.2658, "step": 7493 }, { "epoch": 0.01, "learning_rate": 4.999325416084344e-05, "loss": 1.1783, "step": 7494 }, { "epoch": 0.01, "learning_rate": 4.9993252336122904e-05, "loss": 1.0157, "step": 7495 }, { "epoch": 0.01, "learning_rate": 4.999325051115564e-05, "loss": 1.0256, "step": 7496 }, { "epoch": 0.01, "learning_rate": 4.999324868594166e-05, "loss": 1.5544, "step": 7497 }, { "epoch": 0.01, "learning_rate": 4.9993246860480944e-05, "loss": 1.1515, "step": 7498 }, { "epoch": 0.01, "learning_rate": 4.999324503477352e-05, "loss": 1.2725, "step": 7499 }, { "epoch": 0.01, "learning_rate": 4.999324320881936e-05, "loss": 1.0838, "step": 7500 }, { "epoch": 0.01, "eval_loss": 1.067463994026184, "eval_runtime": 84.948, "eval_samples_per_second": 16.304, "eval_steps_per_second": 4.085, "step": 7500 }, { "epoch": 0.01, "learning_rate": 4.99932413826185e-05, "loss": 1.2704, "step": 7501 }, { "epoch": 0.01, "learning_rate": 4.999323955617089e-05, "loss": 1.2755, "step": 7502 }, { "epoch": 0.01, "learning_rate": 4.9993237729476574e-05, "loss": 1.0452, "step": 7503 }, { "epoch": 0.01, "learning_rate": 4.999323590253553e-05, "loss": 1.1438, "step": 7504 }, { "epoch": 0.01, "learning_rate": 4.999323407534776e-05, "loss": 1.1556, "step": 7505 }, { "epoch": 0.01, "learning_rate": 4.999323224791327e-05, "loss": 1.3714, "step": 7506 }, { "epoch": 0.01, "learning_rate": 4.999323042023206e-05, "loss": 1.3532, "step": 7507 }, { "epoch": 0.01, "learning_rate": 4.9993228592304124e-05, "loss": 0.9893, "step": 7508 }, { "epoch": 0.01, "learning_rate": 4.999322676412947e-05, "loss": 0.9889, "step": 7509 }, { "epoch": 0.01, "learning_rate": 4.999322493570809e-05, "loss": 1.1921, "step": 7510 }, { "epoch": 0.01, "learning_rate": 4.999322310703999e-05, "loss": 1.367, "step": 7511 }, { "epoch": 0.01, "learning_rate": 4.9993221278125164e-05, "loss": 0.8645, "step": 7512 }, { "epoch": 0.01, "learning_rate": 4.999321944896361e-05, "loss": 0.8964, "step": 7513 }, { "epoch": 0.01, "learning_rate": 4.9993217619555345e-05, "loss": 1.1824, "step": 7514 }, { "epoch": 0.01, "learning_rate": 4.999321578990035e-05, "loss": 1.1325, "step": 7515 }, { "epoch": 0.01, "learning_rate": 4.999321395999864e-05, "loss": 0.7096, "step": 7516 }, { "epoch": 0.01, "learning_rate": 4.99932121298502e-05, "loss": 1.5663, "step": 7517 }, { "epoch": 0.01, "learning_rate": 4.999321029945504e-05, "loss": 1.1504, "step": 7518 }, { "epoch": 0.01, "learning_rate": 4.9993208468813156e-05, "loss": 1.3688, "step": 7519 }, { "epoch": 0.01, "learning_rate": 4.999320663792455e-05, "loss": 1.1016, "step": 7520 }, { "epoch": 0.01, "learning_rate": 4.999320480678923e-05, "loss": 1.009, "step": 7521 }, { "epoch": 0.01, "learning_rate": 4.999320297540718e-05, "loss": 1.3089, "step": 7522 }, { "epoch": 0.01, "learning_rate": 4.99932011437784e-05, "loss": 1.3837, "step": 7523 }, { "epoch": 0.01, "learning_rate": 4.999319931190291e-05, "loss": 0.9332, "step": 7524 }, { "epoch": 0.01, "learning_rate": 4.99931974797807e-05, "loss": 1.2406, "step": 7525 }, { "epoch": 0.01, "learning_rate": 4.999319564741176e-05, "loss": 1.1936, "step": 7526 }, { "epoch": 0.01, "learning_rate": 4.999319381479609e-05, "loss": 1.6506, "step": 7527 }, { "epoch": 0.01, "learning_rate": 4.999319198193371e-05, "loss": 0.859, "step": 7528 }, { "epoch": 0.01, "learning_rate": 4.9993190148824604e-05, "loss": 1.2011, "step": 7529 }, { "epoch": 0.01, "learning_rate": 4.9993188315468776e-05, "loss": 0.672, "step": 7530 }, { "epoch": 0.01, "learning_rate": 4.9993186481866225e-05, "loss": 0.8646, "step": 7531 }, { "epoch": 0.01, "learning_rate": 4.9993184648016956e-05, "loss": 0.9417, "step": 7532 }, { "epoch": 0.01, "learning_rate": 4.999318281392096e-05, "loss": 1.31, "step": 7533 }, { "epoch": 0.01, "learning_rate": 4.999318097957824e-05, "loss": 1.1293, "step": 7534 }, { "epoch": 0.01, "learning_rate": 4.99931791449888e-05, "loss": 1.1137, "step": 7535 }, { "epoch": 0.01, "learning_rate": 4.999317731015264e-05, "loss": 1.2222, "step": 7536 }, { "epoch": 0.01, "learning_rate": 4.999317547506976e-05, "loss": 0.9789, "step": 7537 }, { "epoch": 0.01, "learning_rate": 4.9993173639740153e-05, "loss": 1.0231, "step": 7538 }, { "epoch": 0.01, "learning_rate": 4.999317180416383e-05, "loss": 0.7891, "step": 7539 }, { "epoch": 0.01, "learning_rate": 4.999316996834078e-05, "loss": 1.3479, "step": 7540 }, { "epoch": 0.01, "learning_rate": 4.9993168132271005e-05, "loss": 1.1883, "step": 7541 }, { "epoch": 0.01, "learning_rate": 4.999316629595451e-05, "loss": 1.1731, "step": 7542 }, { "epoch": 0.01, "learning_rate": 4.9993164459391296e-05, "loss": 0.9551, "step": 7543 }, { "epoch": 0.01, "learning_rate": 4.9993162622581356e-05, "loss": 1.1011, "step": 7544 }, { "epoch": 0.01, "learning_rate": 4.99931607855247e-05, "loss": 1.026, "step": 7545 }, { "epoch": 0.01, "learning_rate": 4.999315894822132e-05, "loss": 1.1553, "step": 7546 }, { "epoch": 0.01, "learning_rate": 4.999315711067122e-05, "loss": 1.0465, "step": 7547 }, { "epoch": 0.01, "learning_rate": 4.999315527287439e-05, "loss": 0.8922, "step": 7548 }, { "epoch": 0.01, "learning_rate": 4.999315343483084e-05, "loss": 1.1713, "step": 7549 }, { "epoch": 0.01, "learning_rate": 4.999315159654058e-05, "loss": 1.1499, "step": 7550 }, { "epoch": 0.01, "learning_rate": 4.999314975800358e-05, "loss": 1.4271, "step": 7551 }, { "epoch": 0.01, "learning_rate": 4.9993147919219876e-05, "loss": 1.0735, "step": 7552 }, { "epoch": 0.01, "learning_rate": 4.9993146080189436e-05, "loss": 1.0994, "step": 7553 }, { "epoch": 0.01, "learning_rate": 4.999314424091228e-05, "loss": 1.016, "step": 7554 }, { "epoch": 0.01, "learning_rate": 4.9993142401388404e-05, "loss": 0.9924, "step": 7555 }, { "epoch": 0.01, "learning_rate": 4.99931405616178e-05, "loss": 1.2288, "step": 7556 }, { "epoch": 0.01, "learning_rate": 4.999313872160048e-05, "loss": 1.1756, "step": 7557 }, { "epoch": 0.01, "learning_rate": 4.999313688133643e-05, "loss": 1.1061, "step": 7558 }, { "epoch": 0.01, "learning_rate": 4.999313504082567e-05, "loss": 1.1704, "step": 7559 }, { "epoch": 0.01, "learning_rate": 4.999313320006819e-05, "loss": 0.9434, "step": 7560 }, { "epoch": 0.01, "learning_rate": 4.999313135906398e-05, "loss": 0.926, "step": 7561 }, { "epoch": 0.01, "learning_rate": 4.999312951781305e-05, "loss": 1.0371, "step": 7562 }, { "epoch": 0.01, "learning_rate": 4.9993127676315396e-05, "loss": 0.769, "step": 7563 }, { "epoch": 0.01, "learning_rate": 4.999312583457102e-05, "loss": 1.1982, "step": 7564 }, { "epoch": 0.01, "learning_rate": 4.9993123992579923e-05, "loss": 1.1619, "step": 7565 }, { "epoch": 0.01, "learning_rate": 4.999312215034211e-05, "loss": 1.3264, "step": 7566 }, { "epoch": 0.01, "learning_rate": 4.999312030785757e-05, "loss": 1.7573, "step": 7567 }, { "epoch": 0.01, "learning_rate": 4.9993118465126307e-05, "loss": 0.8679, "step": 7568 }, { "epoch": 0.01, "learning_rate": 4.999311662214833e-05, "loss": 1.2439, "step": 7569 }, { "epoch": 0.01, "learning_rate": 4.999311477892363e-05, "loss": 1.1812, "step": 7570 }, { "epoch": 0.01, "learning_rate": 4.99931129354522e-05, "loss": 1.0988, "step": 7571 }, { "epoch": 0.01, "learning_rate": 4.999311109173406e-05, "loss": 1.2373, "step": 7572 }, { "epoch": 0.01, "learning_rate": 4.999310924776919e-05, "loss": 1.9465, "step": 7573 }, { "epoch": 0.01, "learning_rate": 4.9993107403557606e-05, "loss": 1.4244, "step": 7574 }, { "epoch": 0.01, "learning_rate": 4.99931055590993e-05, "loss": 1.3393, "step": 7575 }, { "epoch": 0.01, "learning_rate": 4.9993103714394265e-05, "loss": 1.2637, "step": 7576 }, { "epoch": 0.01, "learning_rate": 4.99931018694425e-05, "loss": 1.245, "step": 7577 }, { "epoch": 0.01, "learning_rate": 4.9993100024244035e-05, "loss": 0.4278, "step": 7578 }, { "epoch": 0.01, "learning_rate": 4.999309817879884e-05, "loss": 1.1911, "step": 7579 }, { "epoch": 0.01, "learning_rate": 4.999309633310692e-05, "loss": 0.746, "step": 7580 }, { "epoch": 0.01, "learning_rate": 4.999309448716828e-05, "loss": 1.1593, "step": 7581 }, { "epoch": 0.01, "learning_rate": 4.9993092640982925e-05, "loss": 1.1381, "step": 7582 }, { "epoch": 0.01, "learning_rate": 4.999309079455085e-05, "loss": 1.0607, "step": 7583 }, { "epoch": 0.01, "learning_rate": 4.999308894787204e-05, "loss": 1.0424, "step": 7584 }, { "epoch": 0.01, "learning_rate": 4.999308710094652e-05, "loss": 1.1306, "step": 7585 }, { "epoch": 0.01, "learning_rate": 4.9993085253774276e-05, "loss": 1.1298, "step": 7586 }, { "epoch": 0.01, "learning_rate": 4.999308340635531e-05, "loss": 0.9243, "step": 7587 }, { "epoch": 0.01, "learning_rate": 4.999308155868962e-05, "loss": 0.8916, "step": 7588 }, { "epoch": 0.01, "learning_rate": 4.999307971077721e-05, "loss": 0.6633, "step": 7589 }, { "epoch": 0.01, "learning_rate": 4.999307786261808e-05, "loss": 1.0586, "step": 7590 }, { "epoch": 0.01, "learning_rate": 4.999307601421224e-05, "loss": 1.1927, "step": 7591 }, { "epoch": 0.01, "learning_rate": 4.999307416555966e-05, "loss": 0.7294, "step": 7592 }, { "epoch": 0.01, "learning_rate": 4.999307231666037e-05, "loss": 0.5216, "step": 7593 }, { "epoch": 0.01, "learning_rate": 4.999307046751436e-05, "loss": 1.1658, "step": 7594 }, { "epoch": 0.01, "learning_rate": 4.999306861812162e-05, "loss": 1.1383, "step": 7595 }, { "epoch": 0.01, "learning_rate": 4.999306676848217e-05, "loss": 0.7021, "step": 7596 }, { "epoch": 0.01, "learning_rate": 4.999306491859599e-05, "loss": 1.2803, "step": 7597 }, { "epoch": 0.01, "learning_rate": 4.9993063068463084e-05, "loss": 1.1629, "step": 7598 }, { "epoch": 0.01, "learning_rate": 4.999306121808347e-05, "loss": 0.7347, "step": 7599 }, { "epoch": 0.01, "learning_rate": 4.999305936745713e-05, "loss": 1.154, "step": 7600 }, { "epoch": 0.01, "learning_rate": 4.999305751658407e-05, "loss": 1.0441, "step": 7601 }, { "epoch": 0.01, "learning_rate": 4.999305566546429e-05, "loss": 1.0921, "step": 7602 }, { "epoch": 0.01, "learning_rate": 4.999305381409779e-05, "loss": 1.0165, "step": 7603 }, { "epoch": 0.01, "learning_rate": 4.999305196248456e-05, "loss": 0.9674, "step": 7604 }, { "epoch": 0.01, "learning_rate": 4.9993050110624614e-05, "loss": 1.2161, "step": 7605 }, { "epoch": 0.01, "learning_rate": 4.999304825851796e-05, "loss": 0.8684, "step": 7606 }, { "epoch": 0.01, "learning_rate": 4.999304640616457e-05, "loss": 1.3836, "step": 7607 }, { "epoch": 0.01, "learning_rate": 4.999304455356446e-05, "loss": 1.0503, "step": 7608 }, { "epoch": 0.01, "learning_rate": 4.999304270071763e-05, "loss": 1.0424, "step": 7609 }, { "epoch": 0.01, "learning_rate": 4.9993040847624085e-05, "loss": 0.7922, "step": 7610 }, { "epoch": 0.01, "learning_rate": 4.9993038994283816e-05, "loss": 1.1937, "step": 7611 }, { "epoch": 0.01, "learning_rate": 4.999303714069682e-05, "loss": 1.0771, "step": 7612 }, { "epoch": 0.01, "learning_rate": 4.999303528686311e-05, "loss": 0.7628, "step": 7613 }, { "epoch": 0.01, "learning_rate": 4.999303343278268e-05, "loss": 1.0711, "step": 7614 }, { "epoch": 0.01, "learning_rate": 4.999303157845553e-05, "loss": 0.9608, "step": 7615 }, { "epoch": 0.01, "learning_rate": 4.9993029723881655e-05, "loss": 1.0593, "step": 7616 }, { "epoch": 0.01, "learning_rate": 4.9993027869061063e-05, "loss": 1.1047, "step": 7617 }, { "epoch": 0.01, "learning_rate": 4.999302601399375e-05, "loss": 0.8421, "step": 7618 }, { "epoch": 0.01, "learning_rate": 4.9993024158679716e-05, "loss": 0.4275, "step": 7619 }, { "epoch": 0.01, "learning_rate": 4.999302230311896e-05, "loss": 0.4701, "step": 7620 }, { "epoch": 0.01, "learning_rate": 4.999302044731149e-05, "loss": 1.1274, "step": 7621 }, { "epoch": 0.01, "learning_rate": 4.999301859125729e-05, "loss": 1.1504, "step": 7622 }, { "epoch": 0.01, "learning_rate": 4.999301673495637e-05, "loss": 1.0298, "step": 7623 }, { "epoch": 0.01, "learning_rate": 4.9993014878408734e-05, "loss": 1.2913, "step": 7624 }, { "epoch": 0.01, "learning_rate": 4.999301302161438e-05, "loss": 1.1684, "step": 7625 }, { "epoch": 0.01, "learning_rate": 4.99930111645733e-05, "loss": 1.1173, "step": 7626 }, { "epoch": 0.01, "learning_rate": 4.99930093072855e-05, "loss": 1.2299, "step": 7627 }, { "epoch": 0.01, "learning_rate": 4.999300744975098e-05, "loss": 1.2688, "step": 7628 }, { "epoch": 0.01, "learning_rate": 4.999300559196974e-05, "loss": 0.7244, "step": 7629 }, { "epoch": 0.01, "learning_rate": 4.999300373394178e-05, "loss": 1.1436, "step": 7630 }, { "epoch": 0.01, "learning_rate": 4.99930018756671e-05, "loss": 1.3276, "step": 7631 }, { "epoch": 0.01, "learning_rate": 4.9993000017145696e-05, "loss": 1.2979, "step": 7632 }, { "epoch": 0.01, "learning_rate": 4.999299815837758e-05, "loss": 1.2512, "step": 7633 }, { "epoch": 0.01, "learning_rate": 4.999299629936274e-05, "loss": 1.1245, "step": 7634 }, { "epoch": 0.01, "learning_rate": 4.999299444010117e-05, "loss": 1.122, "step": 7635 }, { "epoch": 0.01, "learning_rate": 4.999299258059289e-05, "loss": 1.0844, "step": 7636 }, { "epoch": 0.01, "learning_rate": 4.9992990720837886e-05, "loss": 0.9044, "step": 7637 }, { "epoch": 0.01, "learning_rate": 4.9992988860836163e-05, "loss": 1.0386, "step": 7638 }, { "epoch": 0.01, "learning_rate": 4.9992987000587724e-05, "loss": 1.1474, "step": 7639 }, { "epoch": 0.01, "learning_rate": 4.999298514009256e-05, "loss": 1.2707, "step": 7640 }, { "epoch": 0.01, "learning_rate": 4.9992983279350674e-05, "loss": 0.9895, "step": 7641 }, { "epoch": 0.01, "learning_rate": 4.999298141836207e-05, "loss": 1.152, "step": 7642 }, { "epoch": 0.01, "learning_rate": 4.999297955712674e-05, "loss": 1.1866, "step": 7643 }, { "epoch": 0.01, "learning_rate": 4.9992977695644705e-05, "loss": 1.3486, "step": 7644 }, { "epoch": 0.01, "learning_rate": 4.9992975833915936e-05, "loss": 0.8838, "step": 7645 }, { "epoch": 0.01, "learning_rate": 4.999297397194045e-05, "loss": 1.1444, "step": 7646 }, { "epoch": 0.01, "learning_rate": 4.999297210971825e-05, "loss": 1.7278, "step": 7647 }, { "epoch": 0.01, "learning_rate": 4.999297024724933e-05, "loss": 1.3322, "step": 7648 }, { "epoch": 0.01, "learning_rate": 4.999296838453368e-05, "loss": 1.0926, "step": 7649 }, { "epoch": 0.01, "learning_rate": 4.999296652157132e-05, "loss": 1.049, "step": 7650 }, { "epoch": 0.01, "learning_rate": 4.999296465836223e-05, "loss": 1.2226, "step": 7651 }, { "epoch": 0.01, "learning_rate": 4.999296279490642e-05, "loss": 1.1249, "step": 7652 }, { "epoch": 0.01, "learning_rate": 4.99929609312039e-05, "loss": 1.1267, "step": 7653 }, { "epoch": 0.01, "learning_rate": 4.9992959067254655e-05, "loss": 1.1067, "step": 7654 }, { "epoch": 0.01, "learning_rate": 4.999295720305869e-05, "loss": 1.2643, "step": 7655 }, { "epoch": 0.01, "learning_rate": 4.999295533861601e-05, "loss": 1.0183, "step": 7656 }, { "epoch": 0.01, "learning_rate": 4.9992953473926595e-05, "loss": 1.0721, "step": 7657 }, { "epoch": 0.01, "learning_rate": 4.999295160899047e-05, "loss": 1.1275, "step": 7658 }, { "epoch": 0.01, "learning_rate": 4.9992949743807625e-05, "loss": 1.1531, "step": 7659 }, { "epoch": 0.01, "learning_rate": 4.999294787837807e-05, "loss": 1.0552, "step": 7660 }, { "epoch": 0.01, "learning_rate": 4.999294601270178e-05, "loss": 1.0566, "step": 7661 }, { "epoch": 0.01, "learning_rate": 4.9992944146778776e-05, "loss": 1.3829, "step": 7662 }, { "epoch": 0.01, "learning_rate": 4.9992942280609054e-05, "loss": 1.2207, "step": 7663 }, { "epoch": 0.01, "learning_rate": 4.9992940414192616e-05, "loss": 1.0252, "step": 7664 }, { "epoch": 0.01, "learning_rate": 4.999293854752945e-05, "loss": 0.7548, "step": 7665 }, { "epoch": 0.01, "learning_rate": 4.999293668061956e-05, "loss": 1.0979, "step": 7666 }, { "epoch": 0.01, "learning_rate": 4.999293481346296e-05, "loss": 1.8026, "step": 7667 }, { "epoch": 0.01, "learning_rate": 4.999293294605964e-05, "loss": 1.3354, "step": 7668 }, { "epoch": 0.01, "learning_rate": 4.99929310784096e-05, "loss": 1.3458, "step": 7669 }, { "epoch": 0.01, "learning_rate": 4.9992929210512834e-05, "loss": 1.0722, "step": 7670 }, { "epoch": 0.01, "learning_rate": 4.9992927342369356e-05, "loss": 0.8646, "step": 7671 }, { "epoch": 0.01, "learning_rate": 4.9992925473979155e-05, "loss": 1.4073, "step": 7672 }, { "epoch": 0.01, "learning_rate": 4.999292360534223e-05, "loss": 1.0497, "step": 7673 }, { "epoch": 0.01, "learning_rate": 4.999292173645859e-05, "loss": 0.5976, "step": 7674 }, { "epoch": 0.01, "learning_rate": 4.999291986732823e-05, "loss": 1.0724, "step": 7675 }, { "epoch": 0.01, "learning_rate": 4.999291799795115e-05, "loss": 1.0314, "step": 7676 }, { "epoch": 0.01, "learning_rate": 4.999291612832735e-05, "loss": 1.1537, "step": 7677 }, { "epoch": 0.01, "learning_rate": 4.9992914258456836e-05, "loss": 1.3026, "step": 7678 }, { "epoch": 0.01, "learning_rate": 4.9992912388339595e-05, "loss": 0.8799, "step": 7679 }, { "epoch": 0.01, "learning_rate": 4.999291051797563e-05, "loss": 1.2481, "step": 7680 }, { "epoch": 0.01, "learning_rate": 4.9992908647364956e-05, "loss": 1.0373, "step": 7681 }, { "epoch": 0.01, "learning_rate": 4.9992906776507565e-05, "loss": 1.3132, "step": 7682 }, { "epoch": 0.01, "learning_rate": 4.999290490540345e-05, "loss": 1.036, "step": 7683 }, { "epoch": 0.01, "learning_rate": 4.999290303405261e-05, "loss": 0.9312, "step": 7684 }, { "epoch": 0.01, "learning_rate": 4.9992901162455054e-05, "loss": 1.787, "step": 7685 }, { "epoch": 0.01, "learning_rate": 4.999289929061078e-05, "loss": 1.1312, "step": 7686 }, { "epoch": 0.01, "learning_rate": 4.9992897418519784e-05, "loss": 1.0114, "step": 7687 }, { "epoch": 0.01, "learning_rate": 4.999289554618207e-05, "loss": 1.1453, "step": 7688 }, { "epoch": 0.01, "learning_rate": 4.999289367359764e-05, "loss": 1.143, "step": 7689 }, { "epoch": 0.01, "learning_rate": 4.999289180076649e-05, "loss": 1.1481, "step": 7690 }, { "epoch": 0.01, "learning_rate": 4.9992889927688614e-05, "loss": 0.7681, "step": 7691 }, { "epoch": 0.01, "learning_rate": 4.9992888054364026e-05, "loss": 0.9479, "step": 7692 }, { "epoch": 0.01, "learning_rate": 4.999288618079272e-05, "loss": 1.0966, "step": 7693 }, { "epoch": 0.01, "learning_rate": 4.9992884306974684e-05, "loss": 1.1311, "step": 7694 }, { "epoch": 0.01, "learning_rate": 4.999288243290994e-05, "loss": 1.9402, "step": 7695 }, { "epoch": 0.01, "learning_rate": 4.9992880558598475e-05, "loss": 0.9394, "step": 7696 }, { "epoch": 0.01, "learning_rate": 4.999287868404029e-05, "loss": 1.3466, "step": 7697 }, { "epoch": 0.01, "learning_rate": 4.999287680923538e-05, "loss": 0.9499, "step": 7698 }, { "epoch": 0.01, "learning_rate": 4.9992874934183756e-05, "loss": 1.1177, "step": 7699 }, { "epoch": 0.01, "learning_rate": 4.999287305888541e-05, "loss": 1.1028, "step": 7700 }, { "epoch": 0.01, "learning_rate": 4.999287118334035e-05, "loss": 1.7157, "step": 7701 }, { "epoch": 0.01, "learning_rate": 4.9992869307548565e-05, "loss": 1.1457, "step": 7702 }, { "epoch": 0.01, "learning_rate": 4.999286743151006e-05, "loss": 0.9201, "step": 7703 }, { "epoch": 0.01, "learning_rate": 4.999286555522484e-05, "loss": 0.9718, "step": 7704 }, { "epoch": 0.01, "learning_rate": 4.99928636786929e-05, "loss": 1.4427, "step": 7705 }, { "epoch": 0.01, "learning_rate": 4.999286180191425e-05, "loss": 1.1255, "step": 7706 }, { "epoch": 0.01, "learning_rate": 4.999285992488887e-05, "loss": 1.0025, "step": 7707 }, { "epoch": 0.01, "learning_rate": 4.9992858047616776e-05, "loss": 0.9853, "step": 7708 }, { "epoch": 0.01, "learning_rate": 4.999285617009796e-05, "loss": 0.8761, "step": 7709 }, { "epoch": 0.01, "learning_rate": 4.999285429233242e-05, "loss": 0.9686, "step": 7710 }, { "epoch": 0.01, "learning_rate": 4.9992852414320165e-05, "loss": 1.4701, "step": 7711 }, { "epoch": 0.01, "learning_rate": 4.9992850536061196e-05, "loss": 1.2018, "step": 7712 }, { "epoch": 0.01, "learning_rate": 4.999284865755551e-05, "loss": 1.1506, "step": 7713 }, { "epoch": 0.01, "learning_rate": 4.99928467788031e-05, "loss": 1.236, "step": 7714 }, { "epoch": 0.01, "learning_rate": 4.999284489980397e-05, "loss": 1.0325, "step": 7715 }, { "epoch": 0.01, "learning_rate": 4.9992843020558124e-05, "loss": 1.0386, "step": 7716 }, { "epoch": 0.01, "learning_rate": 4.999284114106556e-05, "loss": 1.0008, "step": 7717 }, { "epoch": 0.01, "learning_rate": 4.999283926132627e-05, "loss": 0.8698, "step": 7718 }, { "epoch": 0.01, "learning_rate": 4.9992837381340265e-05, "loss": 0.6912, "step": 7719 }, { "epoch": 0.01, "learning_rate": 4.999283550110754e-05, "loss": 0.861, "step": 7720 }, { "epoch": 0.01, "learning_rate": 4.99928336206281e-05, "loss": 0.6516, "step": 7721 }, { "epoch": 0.01, "learning_rate": 4.9992831739901946e-05, "loss": 0.9092, "step": 7722 }, { "epoch": 0.01, "learning_rate": 4.9992829858929064e-05, "loss": 0.9458, "step": 7723 }, { "epoch": 0.01, "learning_rate": 4.9992827977709464e-05, "loss": 1.2814, "step": 7724 }, { "epoch": 0.01, "learning_rate": 4.999282609624315e-05, "loss": 1.0289, "step": 7725 }, { "epoch": 0.01, "learning_rate": 4.9992824214530114e-05, "loss": 1.1974, "step": 7726 }, { "epoch": 0.01, "learning_rate": 4.9992822332570363e-05, "loss": 0.9542, "step": 7727 }, { "epoch": 0.01, "learning_rate": 4.999282045036389e-05, "loss": 1.0783, "step": 7728 }, { "epoch": 0.01, "learning_rate": 4.99928185679107e-05, "loss": 1.1048, "step": 7729 }, { "epoch": 0.01, "learning_rate": 4.999281668521078e-05, "loss": 1.5022, "step": 7730 }, { "epoch": 0.01, "learning_rate": 4.9992814802264164e-05, "loss": 0.9485, "step": 7731 }, { "epoch": 0.01, "learning_rate": 4.999281291907082e-05, "loss": 1.0487, "step": 7732 }, { "epoch": 0.01, "learning_rate": 4.999281103563075e-05, "loss": 0.9813, "step": 7733 }, { "epoch": 0.01, "learning_rate": 4.9992809151943967e-05, "loss": 1.4835, "step": 7734 }, { "epoch": 0.01, "learning_rate": 4.999280726801047e-05, "loss": 0.905, "step": 7735 }, { "epoch": 0.01, "learning_rate": 4.999280538383024e-05, "loss": 1.3004, "step": 7736 }, { "epoch": 0.01, "learning_rate": 4.999280349940331e-05, "loss": 1.1508, "step": 7737 }, { "epoch": 0.01, "learning_rate": 4.999280161472965e-05, "loss": 0.7234, "step": 7738 }, { "epoch": 0.01, "learning_rate": 4.999279972980927e-05, "loss": 0.7698, "step": 7739 }, { "epoch": 0.01, "learning_rate": 4.999279784464218e-05, "loss": 1.1764, "step": 7740 }, { "epoch": 0.01, "learning_rate": 4.999279595922837e-05, "loss": 1.2074, "step": 7741 }, { "epoch": 0.01, "learning_rate": 4.999279407356784e-05, "loss": 1.1786, "step": 7742 }, { "epoch": 0.01, "learning_rate": 4.999279218766059e-05, "loss": 1.2808, "step": 7743 }, { "epoch": 0.01, "learning_rate": 4.999279030150662e-05, "loss": 1.3466, "step": 7744 }, { "epoch": 0.01, "learning_rate": 4.999278841510593e-05, "loss": 0.9158, "step": 7745 }, { "epoch": 0.01, "learning_rate": 4.999278652845854e-05, "loss": 1.2887, "step": 7746 }, { "epoch": 0.01, "learning_rate": 4.999278464156441e-05, "loss": 1.0082, "step": 7747 }, { "epoch": 0.01, "learning_rate": 4.999278275442357e-05, "loss": 0.1872, "step": 7748 }, { "epoch": 0.01, "learning_rate": 4.999278086703601e-05, "loss": 1.0771, "step": 7749 }, { "epoch": 0.01, "learning_rate": 4.9992778979401734e-05, "loss": 1.3316, "step": 7750 }, { "epoch": 0.01, "learning_rate": 4.9992777091520735e-05, "loss": 1.2208, "step": 7751 }, { "epoch": 0.01, "learning_rate": 4.9992775203393026e-05, "loss": 1.2078, "step": 7752 }, { "epoch": 0.01, "learning_rate": 4.999277331501859e-05, "loss": 1.1975, "step": 7753 }, { "epoch": 0.01, "learning_rate": 4.9992771426397444e-05, "loss": 0.9454, "step": 7754 }, { "epoch": 0.01, "learning_rate": 4.999276953752958e-05, "loss": 1.2826, "step": 7755 }, { "epoch": 0.01, "learning_rate": 4.999276764841499e-05, "loss": 1.4365, "step": 7756 }, { "epoch": 0.01, "learning_rate": 4.9992765759053686e-05, "loss": 1.0539, "step": 7757 }, { "epoch": 0.01, "learning_rate": 4.999276386944567e-05, "loss": 1.0879, "step": 7758 }, { "epoch": 0.01, "learning_rate": 4.999276197959093e-05, "loss": 0.7525, "step": 7759 }, { "epoch": 0.01, "learning_rate": 4.999276008948947e-05, "loss": 1.0603, "step": 7760 }, { "epoch": 0.01, "learning_rate": 4.9992758199141293e-05, "loss": 1.0382, "step": 7761 }, { "epoch": 0.01, "learning_rate": 4.9992756308546394e-05, "loss": 1.4256, "step": 7762 }, { "epoch": 0.01, "learning_rate": 4.999275441770479e-05, "loss": 1.701, "step": 7763 }, { "epoch": 0.01, "learning_rate": 4.999275252661646e-05, "loss": 1.3756, "step": 7764 }, { "epoch": 0.01, "learning_rate": 4.999275063528141e-05, "loss": 1.5012, "step": 7765 }, { "epoch": 0.01, "learning_rate": 4.999274874369965e-05, "loss": 1.3719, "step": 7766 }, { "epoch": 0.01, "learning_rate": 4.9992746851871165e-05, "loss": 0.9829, "step": 7767 }, { "epoch": 0.01, "learning_rate": 4.999274495979596e-05, "loss": 1.0734, "step": 7768 }, { "epoch": 0.01, "learning_rate": 4.999274306747404e-05, "loss": 1.3196, "step": 7769 }, { "epoch": 0.01, "learning_rate": 4.9992741174905404e-05, "loss": 1.0388, "step": 7770 }, { "epoch": 0.01, "learning_rate": 4.999273928209005e-05, "loss": 1.3656, "step": 7771 }, { "epoch": 0.01, "learning_rate": 4.9992737389027977e-05, "loss": 1.1544, "step": 7772 }, { "epoch": 0.01, "learning_rate": 4.999273549571919e-05, "loss": 0.8635, "step": 7773 }, { "epoch": 0.01, "learning_rate": 4.9992733602163674e-05, "loss": 1.4682, "step": 7774 }, { "epoch": 0.01, "learning_rate": 4.999273170836145e-05, "loss": 1.3557, "step": 7775 }, { "epoch": 0.01, "learning_rate": 4.999272981431251e-05, "loss": 1.0475, "step": 7776 }, { "epoch": 0.01, "learning_rate": 4.9992727920016844e-05, "loss": 1.0465, "step": 7777 }, { "epoch": 0.01, "learning_rate": 4.9992726025474466e-05, "loss": 1.0858, "step": 7778 }, { "epoch": 0.01, "learning_rate": 4.999272413068538e-05, "loss": 1.0212, "step": 7779 }, { "epoch": 0.01, "learning_rate": 4.999272223564956e-05, "loss": 1.0863, "step": 7780 }, { "epoch": 0.01, "learning_rate": 4.9992720340367025e-05, "loss": 0.8423, "step": 7781 }, { "epoch": 0.01, "learning_rate": 4.999271844483778e-05, "loss": 0.7978, "step": 7782 }, { "epoch": 0.01, "learning_rate": 4.999271654906181e-05, "loss": 1.2256, "step": 7783 }, { "epoch": 0.01, "learning_rate": 4.999271465303913e-05, "loss": 0.8826, "step": 7784 }, { "epoch": 0.01, "learning_rate": 4.999271275676972e-05, "loss": 0.8992, "step": 7785 }, { "epoch": 0.01, "learning_rate": 4.999271086025361e-05, "loss": 1.2419, "step": 7786 }, { "epoch": 0.01, "learning_rate": 4.9992708963490764e-05, "loss": 1.3075, "step": 7787 }, { "epoch": 0.01, "learning_rate": 4.9992707066481216e-05, "loss": 0.6001, "step": 7788 }, { "epoch": 0.01, "learning_rate": 4.999270516922494e-05, "loss": 0.762, "step": 7789 }, { "epoch": 0.01, "learning_rate": 4.999270327172195e-05, "loss": 0.955, "step": 7790 }, { "epoch": 0.01, "learning_rate": 4.9992701373972245e-05, "loss": 1.1614, "step": 7791 }, { "epoch": 0.01, "learning_rate": 4.9992699475975817e-05, "loss": 1.1102, "step": 7792 }, { "epoch": 0.01, "learning_rate": 4.999269757773268e-05, "loss": 1.5231, "step": 7793 }, { "epoch": 0.01, "learning_rate": 4.9992695679242815e-05, "loss": 0.9765, "step": 7794 }, { "epoch": 0.01, "learning_rate": 4.999269378050624e-05, "loss": 0.8491, "step": 7795 }, { "epoch": 0.01, "learning_rate": 4.9992691881522945e-05, "loss": 1.0887, "step": 7796 }, { "epoch": 0.01, "learning_rate": 4.999268998229294e-05, "loss": 1.1864, "step": 7797 }, { "epoch": 0.01, "learning_rate": 4.999268808281621e-05, "loss": 1.0629, "step": 7798 }, { "epoch": 0.01, "learning_rate": 4.999268618309276e-05, "loss": 1.305, "step": 7799 }, { "epoch": 0.01, "learning_rate": 4.99926842831226e-05, "loss": 1.5089, "step": 7800 }, { "epoch": 0.01, "learning_rate": 4.999268238290572e-05, "loss": 1.2749, "step": 7801 }, { "epoch": 0.01, "learning_rate": 4.999268048244212e-05, "loss": 1.1039, "step": 7802 }, { "epoch": 0.01, "learning_rate": 4.999267858173181e-05, "loss": 1.1027, "step": 7803 }, { "epoch": 0.01, "learning_rate": 4.9992676680774773e-05, "loss": 1.5077, "step": 7804 }, { "epoch": 0.01, "learning_rate": 4.9992674779571024e-05, "loss": 1.3028, "step": 7805 }, { "epoch": 0.01, "learning_rate": 4.9992672878120564e-05, "loss": 1.2525, "step": 7806 }, { "epoch": 0.01, "learning_rate": 4.9992670976423374e-05, "loss": 1.3055, "step": 7807 }, { "epoch": 0.01, "learning_rate": 4.999266907447947e-05, "loss": 1.0859, "step": 7808 }, { "epoch": 0.01, "learning_rate": 4.999266717228886e-05, "loss": 0.8956, "step": 7809 }, { "epoch": 0.01, "learning_rate": 4.999266526985152e-05, "loss": 0.9879, "step": 7810 }, { "epoch": 0.01, "learning_rate": 4.999266336716747e-05, "loss": 1.046, "step": 7811 }, { "epoch": 0.01, "learning_rate": 4.99926614642367e-05, "loss": 1.4059, "step": 7812 }, { "epoch": 0.01, "learning_rate": 4.9992659561059216e-05, "loss": 1.278, "step": 7813 }, { "epoch": 0.01, "learning_rate": 4.9992657657635014e-05, "loss": 1.0187, "step": 7814 }, { "epoch": 0.01, "learning_rate": 4.9992655753964094e-05, "loss": 1.1006, "step": 7815 }, { "epoch": 0.01, "learning_rate": 4.999265385004646e-05, "loss": 1.1579, "step": 7816 }, { "epoch": 0.01, "learning_rate": 4.9992651945882105e-05, "loss": 1.092, "step": 7817 }, { "epoch": 0.01, "learning_rate": 4.9992650041471034e-05, "loss": 0.9898, "step": 7818 }, { "epoch": 0.01, "learning_rate": 4.999264813681325e-05, "loss": 1.1935, "step": 7819 }, { "epoch": 0.01, "learning_rate": 4.999264623190874e-05, "loss": 0.9533, "step": 7820 }, { "epoch": 0.01, "learning_rate": 4.999264432675752e-05, "loss": 1.0438, "step": 7821 }, { "epoch": 0.01, "learning_rate": 4.9992642421359583e-05, "loss": 1.2504, "step": 7822 }, { "epoch": 0.01, "learning_rate": 4.999264051571493e-05, "loss": 0.7561, "step": 7823 }, { "epoch": 0.01, "learning_rate": 4.9992638609823556e-05, "loss": 1.0104, "step": 7824 }, { "epoch": 0.01, "learning_rate": 4.999263670368547e-05, "loss": 1.2529, "step": 7825 }, { "epoch": 0.01, "learning_rate": 4.999263479730066e-05, "loss": 1.1248, "step": 7826 }, { "epoch": 0.01, "learning_rate": 4.999263289066914e-05, "loss": 1.0451, "step": 7827 }, { "epoch": 0.01, "learning_rate": 4.99926309837909e-05, "loss": 1.1424, "step": 7828 }, { "epoch": 0.01, "learning_rate": 4.999262907666594e-05, "loss": 0.9508, "step": 7829 }, { "epoch": 0.01, "learning_rate": 4.999262716929427e-05, "loss": 1.1694, "step": 7830 }, { "epoch": 0.01, "learning_rate": 4.999262526167588e-05, "loss": 1.0555, "step": 7831 }, { "epoch": 0.01, "learning_rate": 4.9992623353810774e-05, "loss": 1.0308, "step": 7832 }, { "epoch": 0.01, "learning_rate": 4.9992621445698956e-05, "loss": 0.9165, "step": 7833 }, { "epoch": 0.01, "learning_rate": 4.9992619537340414e-05, "loss": 1.1295, "step": 7834 }, { "epoch": 0.01, "learning_rate": 4.999261762873516e-05, "loss": 0.9402, "step": 7835 }, { "epoch": 0.01, "learning_rate": 4.999261571988319e-05, "loss": 1.0604, "step": 7836 }, { "epoch": 0.01, "learning_rate": 4.99926138107845e-05, "loss": 1.0774, "step": 7837 }, { "epoch": 0.01, "learning_rate": 4.9992611901439096e-05, "loss": 0.8003, "step": 7838 }, { "epoch": 0.01, "learning_rate": 4.999260999184697e-05, "loss": 0.8127, "step": 7839 }, { "epoch": 0.01, "learning_rate": 4.999260808200813e-05, "loss": 0.5362, "step": 7840 }, { "epoch": 0.01, "learning_rate": 4.999260617192258e-05, "loss": 0.4679, "step": 7841 }, { "epoch": 0.01, "learning_rate": 4.999260426159031e-05, "loss": 1.2592, "step": 7842 }, { "epoch": 0.01, "learning_rate": 4.999260235101132e-05, "loss": 1.0108, "step": 7843 }, { "epoch": 0.01, "learning_rate": 4.9992600440185614e-05, "loss": 0.7332, "step": 7844 }, { "epoch": 0.01, "learning_rate": 4.99925985291132e-05, "loss": 1.0624, "step": 7845 }, { "epoch": 0.01, "learning_rate": 4.999259661779406e-05, "loss": 0.7453, "step": 7846 }, { "epoch": 0.01, "learning_rate": 4.9992594706228204e-05, "loss": 0.8955, "step": 7847 }, { "epoch": 0.01, "learning_rate": 4.999259279441564e-05, "loss": 0.8746, "step": 7848 }, { "epoch": 0.01, "learning_rate": 4.999259088235635e-05, "loss": 1.0011, "step": 7849 }, { "epoch": 0.01, "learning_rate": 4.999258897005035e-05, "loss": 0.8298, "step": 7850 }, { "epoch": 0.01, "learning_rate": 4.999258705749763e-05, "loss": 0.9943, "step": 7851 }, { "epoch": 0.01, "learning_rate": 4.9992585144698197e-05, "loss": 1.1311, "step": 7852 }, { "epoch": 0.01, "learning_rate": 4.9992583231652046e-05, "loss": 0.849, "step": 7853 }, { "epoch": 0.01, "learning_rate": 4.999258131835918e-05, "loss": 0.9639, "step": 7854 }, { "epoch": 0.01, "learning_rate": 4.999257940481959e-05, "loss": 1.1424, "step": 7855 }, { "epoch": 0.01, "learning_rate": 4.999257749103329e-05, "loss": 1.0167, "step": 7856 }, { "epoch": 0.01, "learning_rate": 4.999257557700028e-05, "loss": 1.1265, "step": 7857 }, { "epoch": 0.01, "learning_rate": 4.999257366272054e-05, "loss": 0.9859, "step": 7858 }, { "epoch": 0.01, "learning_rate": 4.9992571748194096e-05, "loss": 1.2524, "step": 7859 }, { "epoch": 0.01, "learning_rate": 4.999256983342093e-05, "loss": 1.0596, "step": 7860 }, { "epoch": 0.01, "learning_rate": 4.9992567918401046e-05, "loss": 0.7429, "step": 7861 }, { "epoch": 0.01, "learning_rate": 4.9992566003134456e-05, "loss": 1.1427, "step": 7862 }, { "epoch": 0.01, "learning_rate": 4.999256408762114e-05, "loss": 1.18, "step": 7863 }, { "epoch": 0.01, "learning_rate": 4.999256217186111e-05, "loss": 1.3083, "step": 7864 }, { "epoch": 0.01, "learning_rate": 4.999256025585436e-05, "loss": 1.0804, "step": 7865 }, { "epoch": 0.01, "learning_rate": 4.9992558339600904e-05, "loss": 1.179, "step": 7866 }, { "epoch": 0.01, "learning_rate": 4.999255642310073e-05, "loss": 1.2044, "step": 7867 }, { "epoch": 0.01, "learning_rate": 4.999255450635384e-05, "loss": 0.9728, "step": 7868 }, { "epoch": 0.01, "learning_rate": 4.999255258936023e-05, "loss": 1.3624, "step": 7869 }, { "epoch": 0.01, "learning_rate": 4.9992550672119894e-05, "loss": 1.4786, "step": 7870 }, { "epoch": 0.01, "learning_rate": 4.9992548754632865e-05, "loss": 1.0779, "step": 7871 }, { "epoch": 0.01, "learning_rate": 4.99925468368991e-05, "loss": 1.2894, "step": 7872 }, { "epoch": 0.01, "learning_rate": 4.999254491891863e-05, "loss": 1.2372, "step": 7873 }, { "epoch": 0.01, "learning_rate": 4.999254300069144e-05, "loss": 1.0072, "step": 7874 }, { "epoch": 0.01, "learning_rate": 4.9992541082217536e-05, "loss": 1.0588, "step": 7875 }, { "epoch": 0.01, "learning_rate": 4.999253916349692e-05, "loss": 1.3491, "step": 7876 }, { "epoch": 0.01, "learning_rate": 4.999253724452958e-05, "loss": 1.5759, "step": 7877 }, { "epoch": 0.01, "learning_rate": 4.999253532531553e-05, "loss": 0.7801, "step": 7878 }, { "epoch": 0.01, "learning_rate": 4.9992533405854756e-05, "loss": 1.0617, "step": 7879 }, { "epoch": 0.01, "learning_rate": 4.999253148614727e-05, "loss": 1.0505, "step": 7880 }, { "epoch": 0.01, "learning_rate": 4.999252956619308e-05, "loss": 1.1608, "step": 7881 }, { "epoch": 0.01, "learning_rate": 4.9992527645992164e-05, "loss": 1.2429, "step": 7882 }, { "epoch": 0.01, "learning_rate": 4.999252572554454e-05, "loss": 1.1126, "step": 7883 }, { "epoch": 0.01, "learning_rate": 4.999252380485019e-05, "loss": 0.9354, "step": 7884 }, { "epoch": 0.01, "learning_rate": 4.9992521883909127e-05, "loss": 1.076, "step": 7885 }, { "epoch": 0.01, "learning_rate": 4.999251996272135e-05, "loss": 1.2101, "step": 7886 }, { "epoch": 0.01, "learning_rate": 4.999251804128685e-05, "loss": 1.2844, "step": 7887 }, { "epoch": 0.01, "learning_rate": 4.999251611960565e-05, "loss": 1.1769, "step": 7888 }, { "epoch": 0.01, "learning_rate": 4.999251419767772e-05, "loss": 1.2664, "step": 7889 }, { "epoch": 0.01, "learning_rate": 4.9992512275503076e-05, "loss": 1.0522, "step": 7890 }, { "epoch": 0.01, "learning_rate": 4.999251035308172e-05, "loss": 1.0897, "step": 7891 }, { "epoch": 0.01, "learning_rate": 4.9992508430413654e-05, "loss": 1.0961, "step": 7892 }, { "epoch": 0.01, "learning_rate": 4.999250650749887e-05, "loss": 1.3068, "step": 7893 }, { "epoch": 0.01, "learning_rate": 4.9992504584337365e-05, "loss": 0.4432, "step": 7894 }, { "epoch": 0.01, "learning_rate": 4.9992502660929155e-05, "loss": 0.7508, "step": 7895 }, { "epoch": 0.01, "learning_rate": 4.9992500737274214e-05, "loss": 0.9558, "step": 7896 }, { "epoch": 0.01, "learning_rate": 4.999249881337257e-05, "loss": 0.8569, "step": 7897 }, { "epoch": 0.01, "learning_rate": 4.99924968892242e-05, "loss": 0.9947, "step": 7898 }, { "epoch": 0.01, "learning_rate": 4.999249496482913e-05, "loss": 1.2156, "step": 7899 }, { "epoch": 0.01, "learning_rate": 4.999249304018733e-05, "loss": 1.2722, "step": 7900 }, { "epoch": 0.01, "learning_rate": 4.9992491115298824e-05, "loss": 0.9461, "step": 7901 }, { "epoch": 0.01, "learning_rate": 4.9992489190163595e-05, "loss": 0.937, "step": 7902 }, { "epoch": 0.01, "learning_rate": 4.999248726478165e-05, "loss": 1.1194, "step": 7903 }, { "epoch": 0.01, "learning_rate": 4.9992485339152993e-05, "loss": 1.1772, "step": 7904 }, { "epoch": 0.01, "learning_rate": 4.999248341327763e-05, "loss": 1.2923, "step": 7905 }, { "epoch": 0.01, "learning_rate": 4.999248148715554e-05, "loss": 1.09, "step": 7906 }, { "epoch": 0.01, "learning_rate": 4.9992479560786744e-05, "loss": 2.6286, "step": 7907 }, { "epoch": 0.01, "learning_rate": 4.999247763417122e-05, "loss": 3.4125, "step": 7908 }, { "epoch": 0.01, "learning_rate": 4.999247570730899e-05, "loss": 1.1838, "step": 7909 }, { "epoch": 0.01, "learning_rate": 4.999247378020004e-05, "loss": 1.379, "step": 7910 }, { "epoch": 0.01, "learning_rate": 4.999247185284438e-05, "loss": 1.099, "step": 7911 }, { "epoch": 0.01, "learning_rate": 4.9992469925242e-05, "loss": 1.0116, "step": 7912 }, { "epoch": 0.01, "learning_rate": 4.999246799739291e-05, "loss": 1.0267, "step": 7913 }, { "epoch": 0.01, "learning_rate": 4.999246606929711e-05, "loss": 1.4954, "step": 7914 }, { "epoch": 0.01, "learning_rate": 4.999246414095458e-05, "loss": 1.6992, "step": 7915 }, { "epoch": 0.01, "learning_rate": 4.9992462212365345e-05, "loss": 1.3373, "step": 7916 }, { "epoch": 0.01, "learning_rate": 4.999246028352939e-05, "loss": 0.8221, "step": 7917 }, { "epoch": 0.01, "learning_rate": 4.999245835444672e-05, "loss": 1.0465, "step": 7918 }, { "epoch": 0.01, "learning_rate": 4.999245642511734e-05, "loss": 1.1051, "step": 7919 }, { "epoch": 0.01, "learning_rate": 4.9992454495541244e-05, "loss": 1.0927, "step": 7920 }, { "epoch": 0.01, "learning_rate": 4.9992452565718426e-05, "loss": 1.1845, "step": 7921 }, { "epoch": 0.01, "learning_rate": 4.99924506356489e-05, "loss": 1.0775, "step": 7922 }, { "epoch": 0.01, "learning_rate": 4.9992448705332654e-05, "loss": 1.3101, "step": 7923 }, { "epoch": 0.01, "learning_rate": 4.99924467747697e-05, "loss": 1.2129, "step": 7924 }, { "epoch": 0.01, "learning_rate": 4.9992444843960027e-05, "loss": 1.3299, "step": 7925 }, { "epoch": 0.01, "learning_rate": 4.999244291290364e-05, "loss": 1.1885, "step": 7926 }, { "epoch": 0.01, "learning_rate": 4.999244098160053e-05, "loss": 0.8202, "step": 7927 }, { "epoch": 0.01, "learning_rate": 4.9992439050050715e-05, "loss": 1.169, "step": 7928 }, { "epoch": 0.01, "learning_rate": 4.999243711825419e-05, "loss": 0.9432, "step": 7929 }, { "epoch": 0.01, "learning_rate": 4.999243518621094e-05, "loss": 1.1669, "step": 7930 }, { "epoch": 0.01, "learning_rate": 4.999243325392098e-05, "loss": 1.0737, "step": 7931 }, { "epoch": 0.01, "learning_rate": 4.9992431321384294e-05, "loss": 1.1336, "step": 7932 }, { "epoch": 0.01, "learning_rate": 4.99924293886009e-05, "loss": 1.4216, "step": 7933 }, { "epoch": 0.01, "learning_rate": 4.99924274555708e-05, "loss": 1.34, "step": 7934 }, { "epoch": 0.01, "learning_rate": 4.999242552229398e-05, "loss": 0.8725, "step": 7935 }, { "epoch": 0.01, "learning_rate": 4.999242358877044e-05, "loss": 1.1615, "step": 7936 }, { "epoch": 0.01, "learning_rate": 4.9992421655000194e-05, "loss": 1.2269, "step": 7937 }, { "epoch": 0.01, "learning_rate": 4.999241972098323e-05, "loss": 1.1951, "step": 7938 }, { "epoch": 0.01, "learning_rate": 4.9992417786719545e-05, "loss": 1.058, "step": 7939 }, { "epoch": 0.01, "learning_rate": 4.999241585220915e-05, "loss": 1.3327, "step": 7940 }, { "epoch": 0.01, "learning_rate": 4.999241391745205e-05, "loss": 1.2085, "step": 7941 }, { "epoch": 0.01, "learning_rate": 4.999241198244822e-05, "loss": 0.9829, "step": 7942 }, { "epoch": 0.01, "learning_rate": 4.9992410047197685e-05, "loss": 1.2403, "step": 7943 }, { "epoch": 0.01, "learning_rate": 4.999240811170043e-05, "loss": 1.1999, "step": 7944 }, { "epoch": 0.01, "learning_rate": 4.999240617595646e-05, "loss": 0.9268, "step": 7945 }, { "epoch": 0.01, "learning_rate": 4.9992404239965786e-05, "loss": 1.1453, "step": 7946 }, { "epoch": 0.01, "learning_rate": 4.9992402303728395e-05, "loss": 0.9478, "step": 7947 }, { "epoch": 0.01, "learning_rate": 4.999240036724427e-05, "loss": 0.852, "step": 7948 }, { "epoch": 0.01, "learning_rate": 4.9992398430513454e-05, "loss": 1.402, "step": 7949 }, { "epoch": 0.01, "learning_rate": 4.999239649353591e-05, "loss": 1.0803, "step": 7950 }, { "epoch": 0.01, "learning_rate": 4.999239455631166e-05, "loss": 1.0025, "step": 7951 }, { "epoch": 0.01, "learning_rate": 4.999239261884069e-05, "loss": 1.3306, "step": 7952 }, { "epoch": 0.01, "learning_rate": 4.999239068112302e-05, "loss": 1.1896, "step": 7953 }, { "epoch": 0.01, "learning_rate": 4.999238874315861e-05, "loss": 0.9684, "step": 7954 }, { "epoch": 0.01, "learning_rate": 4.99923868049475e-05, "loss": 1.3475, "step": 7955 }, { "epoch": 0.01, "learning_rate": 4.9992384866489676e-05, "loss": 1.0657, "step": 7956 }, { "epoch": 0.01, "learning_rate": 4.9992382927785135e-05, "loss": 1.1366, "step": 7957 }, { "epoch": 0.01, "learning_rate": 4.9992380988833884e-05, "loss": 1.3813, "step": 7958 }, { "epoch": 0.01, "learning_rate": 4.9992379049635916e-05, "loss": 1.1277, "step": 7959 }, { "epoch": 0.01, "learning_rate": 4.999237711019124e-05, "loss": 1.2634, "step": 7960 }, { "epoch": 0.01, "learning_rate": 4.999237517049984e-05, "loss": 1.0193, "step": 7961 }, { "epoch": 0.01, "learning_rate": 4.999237323056173e-05, "loss": 1.0104, "step": 7962 }, { "epoch": 0.01, "learning_rate": 4.99923712903769e-05, "loss": 1.8146, "step": 7963 }, { "epoch": 0.01, "learning_rate": 4.999236934994536e-05, "loss": 0.9461, "step": 7964 }, { "epoch": 0.01, "learning_rate": 4.9992367409267105e-05, "loss": 1.1662, "step": 7965 }, { "epoch": 0.01, "learning_rate": 4.999236546834214e-05, "loss": 0.9277, "step": 7966 }, { "epoch": 0.01, "learning_rate": 4.999236352717046e-05, "loss": 1.0035, "step": 7967 }, { "epoch": 0.01, "learning_rate": 4.999236158575206e-05, "loss": 0.9615, "step": 7968 }, { "epoch": 0.01, "learning_rate": 4.999235964408695e-05, "loss": 0.8536, "step": 7969 }, { "epoch": 0.01, "learning_rate": 4.999235770217513e-05, "loss": 1.0511, "step": 7970 }, { "epoch": 0.01, "learning_rate": 4.999235576001659e-05, "loss": 0.8384, "step": 7971 }, { "epoch": 0.01, "learning_rate": 4.9992353817611336e-05, "loss": 0.7573, "step": 7972 }, { "epoch": 0.01, "learning_rate": 4.999235187495938e-05, "loss": 1.2414, "step": 7973 }, { "epoch": 0.01, "learning_rate": 4.9992349932060696e-05, "loss": 1.2377, "step": 7974 }, { "epoch": 0.01, "learning_rate": 4.99923479889153e-05, "loss": 1.0569, "step": 7975 }, { "epoch": 0.01, "learning_rate": 4.9992346045523194e-05, "loss": 0.9342, "step": 7976 }, { "epoch": 0.01, "learning_rate": 4.999234410188437e-05, "loss": 1.0519, "step": 7977 }, { "epoch": 0.01, "learning_rate": 4.999234215799884e-05, "loss": 1.0635, "step": 7978 }, { "epoch": 0.01, "learning_rate": 4.9992340213866585e-05, "loss": 0.9977, "step": 7979 }, { "epoch": 0.01, "learning_rate": 4.999233826948763e-05, "loss": 1.084, "step": 7980 }, { "epoch": 0.01, "learning_rate": 4.999233632486194e-05, "loss": 1.1179, "step": 7981 }, { "epoch": 0.01, "learning_rate": 4.999233437998956e-05, "loss": 1.1019, "step": 7982 }, { "epoch": 0.01, "learning_rate": 4.999233243487045e-05, "loss": 0.6534, "step": 7983 }, { "epoch": 0.01, "learning_rate": 4.999233048950464e-05, "loss": 1.2761, "step": 7984 }, { "epoch": 0.01, "learning_rate": 4.9992328543892096e-05, "loss": 0.9804, "step": 7985 }, { "epoch": 0.01, "learning_rate": 4.999232659803286e-05, "loss": 0.9749, "step": 7986 }, { "epoch": 0.01, "learning_rate": 4.9992324651926895e-05, "loss": 1.5774, "step": 7987 }, { "epoch": 0.01, "learning_rate": 4.999232270557422e-05, "loss": 1.1378, "step": 7988 }, { "epoch": 0.01, "learning_rate": 4.999232075897483e-05, "loss": 0.9541, "step": 7989 }, { "epoch": 0.01, "learning_rate": 4.9992318812128734e-05, "loss": 1.1965, "step": 7990 }, { "epoch": 0.01, "learning_rate": 4.9992316865035924e-05, "loss": 0.8706, "step": 7991 }, { "epoch": 0.01, "learning_rate": 4.999231491769639e-05, "loss": 0.7041, "step": 7992 }, { "epoch": 0.01, "learning_rate": 4.9992312970110153e-05, "loss": 0.6082, "step": 7993 }, { "epoch": 0.01, "learning_rate": 4.999231102227719e-05, "loss": 1.1324, "step": 7994 }, { "epoch": 0.01, "learning_rate": 4.999230907419753e-05, "loss": 1.0204, "step": 7995 }, { "epoch": 0.01, "learning_rate": 4.999230712587115e-05, "loss": 0.9703, "step": 7996 }, { "epoch": 0.01, "learning_rate": 4.999230517729805e-05, "loss": 1.0797, "step": 7997 }, { "epoch": 0.01, "learning_rate": 4.999230322847824e-05, "loss": 1.1712, "step": 7998 }, { "epoch": 0.01, "learning_rate": 4.9992301279411716e-05, "loss": 1.0651, "step": 7999 }, { "epoch": 0.01, "learning_rate": 4.999229933009848e-05, "loss": 0.9631, "step": 8000 }, { "epoch": 0.01, "eval_loss": 1.054444432258606, "eval_runtime": 84.4105, "eval_samples_per_second": 16.408, "eval_steps_per_second": 4.111, "step": 8000 }, { "epoch": 0.01, "learning_rate": 4.9992297380538534e-05, "loss": 1.1107, "step": 8001 }, { "epoch": 0.01, "learning_rate": 4.999229543073187e-05, "loss": 1.0455, "step": 8002 }, { "epoch": 0.01, "learning_rate": 4.999229348067849e-05, "loss": 1.0669, "step": 8003 }, { "epoch": 0.01, "learning_rate": 4.99922915303784e-05, "loss": 1.3531, "step": 8004 }, { "epoch": 0.01, "learning_rate": 4.9992289579831595e-05, "loss": 1.266, "step": 8005 }, { "epoch": 0.01, "learning_rate": 4.999228762903808e-05, "loss": 1.009, "step": 8006 }, { "epoch": 0.01, "learning_rate": 4.999228567799785e-05, "loss": 0.7729, "step": 8007 }, { "epoch": 0.01, "learning_rate": 4.999228372671091e-05, "loss": 1.1114, "step": 8008 }, { "epoch": 0.01, "learning_rate": 4.999228177517725e-05, "loss": 0.8466, "step": 8009 }, { "epoch": 0.01, "learning_rate": 4.999227982339688e-05, "loss": 1.5055, "step": 8010 }, { "epoch": 0.01, "learning_rate": 4.9992277871369806e-05, "loss": 1.2978, "step": 8011 }, { "epoch": 0.01, "learning_rate": 4.999227591909601e-05, "loss": 1.2087, "step": 8012 }, { "epoch": 0.01, "learning_rate": 4.99922739665755e-05, "loss": 1.776, "step": 8013 }, { "epoch": 0.01, "learning_rate": 4.999227201380827e-05, "loss": 1.6039, "step": 8014 }, { "epoch": 0.01, "learning_rate": 4.999227006079434e-05, "loss": 1.1489, "step": 8015 }, { "epoch": 0.01, "learning_rate": 4.999226810753369e-05, "loss": 1.2858, "step": 8016 }, { "epoch": 0.01, "learning_rate": 4.999226615402632e-05, "loss": 1.231, "step": 8017 }, { "epoch": 0.01, "learning_rate": 4.9992264200272254e-05, "loss": 1.1254, "step": 8018 }, { "epoch": 0.01, "learning_rate": 4.999226224627146e-05, "loss": 1.0541, "step": 8019 }, { "epoch": 0.01, "learning_rate": 4.999226029202396e-05, "loss": 1.0562, "step": 8020 }, { "epoch": 0.01, "learning_rate": 4.9992258337529745e-05, "loss": 1.3727, "step": 8021 }, { "epoch": 0.01, "learning_rate": 4.999225638278882e-05, "loss": 1.0765, "step": 8022 }, { "epoch": 0.01, "learning_rate": 4.999225442780118e-05, "loss": 1.0304, "step": 8023 }, { "epoch": 0.01, "learning_rate": 4.9992252472566825e-05, "loss": 0.962, "step": 8024 }, { "epoch": 0.01, "learning_rate": 4.999225051708576e-05, "loss": 0.9971, "step": 8025 }, { "epoch": 0.01, "learning_rate": 4.999224856135798e-05, "loss": 1.0339, "step": 8026 }, { "epoch": 0.01, "learning_rate": 4.9992246605383485e-05, "loss": 1.0693, "step": 8027 }, { "epoch": 0.01, "learning_rate": 4.9992244649162276e-05, "loss": 1.1078, "step": 8028 }, { "epoch": 0.01, "learning_rate": 4.999224269269436e-05, "loss": 1.3662, "step": 8029 }, { "epoch": 0.01, "learning_rate": 4.999224073597973e-05, "loss": 1.4999, "step": 8030 }, { "epoch": 0.01, "learning_rate": 4.9992238779018386e-05, "loss": 1.6307, "step": 8031 }, { "epoch": 0.01, "learning_rate": 4.9992236821810336e-05, "loss": 1.8442, "step": 8032 }, { "epoch": 0.01, "learning_rate": 4.9992234864355556e-05, "loss": 1.3682, "step": 8033 }, { "epoch": 0.01, "learning_rate": 4.999223290665408e-05, "loss": 1.2394, "step": 8034 }, { "epoch": 0.01, "learning_rate": 4.9992230948705884e-05, "loss": 1.0143, "step": 8035 }, { "epoch": 0.01, "learning_rate": 4.999222899051097e-05, "loss": 0.952, "step": 8036 }, { "epoch": 0.01, "learning_rate": 4.999222703206936e-05, "loss": 0.9989, "step": 8037 }, { "epoch": 0.01, "learning_rate": 4.999222507338102e-05, "loss": 1.2007, "step": 8038 }, { "epoch": 0.01, "learning_rate": 4.999222311444597e-05, "loss": 1.1594, "step": 8039 }, { "epoch": 0.01, "learning_rate": 4.999222115526422e-05, "loss": 1.6533, "step": 8040 }, { "epoch": 0.01, "learning_rate": 4.999221919583574e-05, "loss": 1.2455, "step": 8041 }, { "epoch": 0.01, "learning_rate": 4.999221723616056e-05, "loss": 1.1915, "step": 8042 }, { "epoch": 0.01, "learning_rate": 4.999221527623866e-05, "loss": 0.9804, "step": 8043 }, { "epoch": 0.01, "learning_rate": 4.9992213316070054e-05, "loss": 1.0802, "step": 8044 }, { "epoch": 0.01, "learning_rate": 4.999221135565473e-05, "loss": 1.2212, "step": 8045 }, { "epoch": 0.01, "learning_rate": 4.99922093949927e-05, "loss": 1.0551, "step": 8046 }, { "epoch": 0.01, "learning_rate": 4.9992207434083946e-05, "loss": 1.2246, "step": 8047 }, { "epoch": 0.01, "learning_rate": 4.9992205472928485e-05, "loss": 1.357, "step": 8048 }, { "epoch": 0.01, "learning_rate": 4.9992203511526314e-05, "loss": 1.0874, "step": 8049 }, { "epoch": 0.01, "learning_rate": 4.999220154987743e-05, "loss": 1.0338, "step": 8050 }, { "epoch": 0.01, "learning_rate": 4.9992199587981835e-05, "loss": 1.3102, "step": 8051 }, { "epoch": 0.01, "learning_rate": 4.999219762583952e-05, "loss": 0.9823, "step": 8052 }, { "epoch": 0.01, "learning_rate": 4.99921956634505e-05, "loss": 1.0771, "step": 8053 }, { "epoch": 0.01, "learning_rate": 4.999219370081477e-05, "loss": 0.9605, "step": 8054 }, { "epoch": 0.01, "learning_rate": 4.999219173793232e-05, "loss": 0.8811, "step": 8055 }, { "epoch": 0.01, "learning_rate": 4.9992189774803163e-05, "loss": 3.6941, "step": 8056 }, { "epoch": 0.01, "learning_rate": 4.999218781142729e-05, "loss": 6.1163, "step": 8057 }, { "epoch": 0.01, "learning_rate": 4.999218584780471e-05, "loss": 4.5305, "step": 8058 }, { "epoch": 0.01, "learning_rate": 4.999218388393541e-05, "loss": 5.9376, "step": 8059 }, { "epoch": 0.01, "learning_rate": 4.9992181919819403e-05, "loss": 5.7547, "step": 8060 }, { "epoch": 0.01, "learning_rate": 4.9992179955456676e-05, "loss": 5.3154, "step": 8061 }, { "epoch": 0.01, "learning_rate": 4.9992177990847245e-05, "loss": 3.2164, "step": 8062 }, { "epoch": 0.01, "learning_rate": 4.99921760259911e-05, "loss": 1.0417, "step": 8063 }, { "epoch": 0.01, "learning_rate": 4.999217406088824e-05, "loss": 1.563, "step": 8064 }, { "epoch": 0.01, "learning_rate": 4.999217209553867e-05, "loss": 1.1756, "step": 8065 }, { "epoch": 0.01, "learning_rate": 4.999217012994239e-05, "loss": 1.4225, "step": 8066 }, { "epoch": 0.01, "learning_rate": 4.99921681640994e-05, "loss": 1.0189, "step": 8067 }, { "epoch": 0.01, "learning_rate": 4.9992166198009685e-05, "loss": 0.7917, "step": 8068 }, { "epoch": 0.01, "learning_rate": 4.999216423167327e-05, "loss": 1.1507, "step": 8069 }, { "epoch": 0.01, "learning_rate": 4.9992162265090137e-05, "loss": 1.0214, "step": 8070 }, { "epoch": 0.01, "learning_rate": 4.9992160298260294e-05, "loss": 0.8676, "step": 8071 }, { "epoch": 0.01, "learning_rate": 4.999215833118374e-05, "loss": 1.2787, "step": 8072 }, { "epoch": 0.01, "learning_rate": 4.999215636386048e-05, "loss": 1.5972, "step": 8073 }, { "epoch": 0.01, "learning_rate": 4.999215439629049e-05, "loss": 1.3106, "step": 8074 }, { "epoch": 0.01, "learning_rate": 4.9992152428473806e-05, "loss": 1.2539, "step": 8075 }, { "epoch": 0.01, "learning_rate": 4.9992150460410405e-05, "loss": 1.1408, "step": 8076 }, { "epoch": 0.01, "learning_rate": 4.999214849210029e-05, "loss": 1.209, "step": 8077 }, { "epoch": 0.01, "learning_rate": 4.999214652354346e-05, "loss": 1.2816, "step": 8078 }, { "epoch": 0.01, "learning_rate": 4.9992144554739915e-05, "loss": 1.1278, "step": 8079 }, { "epoch": 0.01, "learning_rate": 4.999214258568967e-05, "loss": 1.2074, "step": 8080 }, { "epoch": 0.01, "learning_rate": 4.99921406163927e-05, "loss": 1.1355, "step": 8081 }, { "epoch": 0.01, "learning_rate": 4.999213864684903e-05, "loss": 0.933, "step": 8082 }, { "epoch": 0.01, "learning_rate": 4.9992136677058646e-05, "loss": 1.1354, "step": 8083 }, { "epoch": 0.01, "learning_rate": 4.9992134707021544e-05, "loss": 1.0388, "step": 8084 }, { "epoch": 0.01, "learning_rate": 4.999213273673774e-05, "loss": 3.3175, "step": 8085 }, { "epoch": 0.01, "learning_rate": 4.9992130766207214e-05, "loss": 1.2026, "step": 8086 }, { "epoch": 0.01, "learning_rate": 4.999212879542998e-05, "loss": 1.2803, "step": 8087 }, { "epoch": 0.01, "learning_rate": 4.999212682440604e-05, "loss": 1.1053, "step": 8088 }, { "epoch": 0.01, "learning_rate": 4.9992124853135384e-05, "loss": 1.3526, "step": 8089 }, { "epoch": 0.01, "learning_rate": 4.9992122881618006e-05, "loss": 1.3599, "step": 8090 }, { "epoch": 0.01, "learning_rate": 4.9992120909853925e-05, "loss": 0.9951, "step": 8091 }, { "epoch": 0.01, "learning_rate": 4.9992118937843134e-05, "loss": 1.264, "step": 8092 }, { "epoch": 0.01, "learning_rate": 4.999211696558563e-05, "loss": 0.7493, "step": 8093 }, { "epoch": 0.01, "learning_rate": 4.999211499308142e-05, "loss": 1.1717, "step": 8094 }, { "epoch": 0.01, "learning_rate": 4.9992113020330486e-05, "loss": 1.0709, "step": 8095 }, { "epoch": 0.01, "learning_rate": 4.999211104733285e-05, "loss": 1.1257, "step": 8096 }, { "epoch": 0.01, "learning_rate": 4.99921090740885e-05, "loss": 0.7699, "step": 8097 }, { "epoch": 0.01, "learning_rate": 4.999210710059743e-05, "loss": 2.4386, "step": 8098 }, { "epoch": 0.01, "learning_rate": 4.999210512685966e-05, "loss": 1.8469, "step": 8099 }, { "epoch": 0.01, "learning_rate": 4.999210315287518e-05, "loss": 1.0731, "step": 8100 }, { "epoch": 0.01, "learning_rate": 4.999210117864398e-05, "loss": 1.175, "step": 8101 }, { "epoch": 0.01, "learning_rate": 4.999209920416607e-05, "loss": 1.1062, "step": 8102 }, { "epoch": 0.01, "learning_rate": 4.9992097229441456e-05, "loss": 0.981, "step": 8103 }, { "epoch": 0.01, "learning_rate": 4.9992095254470115e-05, "loss": 0.514, "step": 8104 }, { "epoch": 0.01, "learning_rate": 4.999209327925208e-05, "loss": 0.6874, "step": 8105 }, { "epoch": 0.01, "learning_rate": 4.9992091303787324e-05, "loss": 1.045, "step": 8106 }, { "epoch": 0.01, "learning_rate": 4.999208932807586e-05, "loss": 1.1178, "step": 8107 }, { "epoch": 0.01, "learning_rate": 4.9992087352117684e-05, "loss": 1.1132, "step": 8108 }, { "epoch": 0.01, "learning_rate": 4.999208537591279e-05, "loss": 0.8647, "step": 8109 }, { "epoch": 0.01, "learning_rate": 4.99920833994612e-05, "loss": 1.5936, "step": 8110 }, { "epoch": 0.01, "learning_rate": 4.9992081422762885e-05, "loss": 0.9714, "step": 8111 }, { "epoch": 0.01, "learning_rate": 4.999207944581786e-05, "loss": 0.8233, "step": 8112 }, { "epoch": 0.01, "learning_rate": 4.999207746862613e-05, "loss": 1.031, "step": 8113 }, { "epoch": 0.01, "learning_rate": 4.999207549118768e-05, "loss": 0.4453, "step": 8114 }, { "epoch": 0.01, "learning_rate": 4.999207351350252e-05, "loss": 0.3886, "step": 8115 }, { "epoch": 0.01, "learning_rate": 4.9992071535570665e-05, "loss": 1.0795, "step": 8116 }, { "epoch": 0.01, "learning_rate": 4.9992069557392085e-05, "loss": 0.8687, "step": 8117 }, { "epoch": 0.01, "learning_rate": 4.9992067578966794e-05, "loss": 1.0988, "step": 8118 }, { "epoch": 0.01, "learning_rate": 4.9992065600294794e-05, "loss": 1.2491, "step": 8119 }, { "epoch": 0.01, "learning_rate": 4.999206362137608e-05, "loss": 1.3787, "step": 8120 }, { "epoch": 0.01, "learning_rate": 4.9992061642210655e-05, "loss": 1.224, "step": 8121 }, { "epoch": 0.01, "learning_rate": 4.9992059662798524e-05, "loss": 1.234, "step": 8122 }, { "epoch": 0.01, "learning_rate": 4.999205768313968e-05, "loss": 1.1286, "step": 8123 }, { "epoch": 0.01, "learning_rate": 4.999205570323412e-05, "loss": 0.8692, "step": 8124 }, { "epoch": 0.01, "learning_rate": 4.9992053723081856e-05, "loss": 1.1534, "step": 8125 }, { "epoch": 0.01, "learning_rate": 4.999205174268288e-05, "loss": 1.1841, "step": 8126 }, { "epoch": 0.01, "learning_rate": 4.999204976203719e-05, "loss": 1.2964, "step": 8127 }, { "epoch": 0.01, "learning_rate": 4.999204778114479e-05, "loss": 1.3143, "step": 8128 }, { "epoch": 0.01, "learning_rate": 4.999204580000568e-05, "loss": 1.3841, "step": 8129 }, { "epoch": 0.01, "learning_rate": 4.999204381861985e-05, "loss": 1.3286, "step": 8130 }, { "epoch": 0.01, "learning_rate": 4.9992041836987323e-05, "loss": 1.1715, "step": 8131 }, { "epoch": 0.01, "learning_rate": 4.999203985510808e-05, "loss": 1.2359, "step": 8132 }, { "epoch": 0.01, "learning_rate": 4.999203787298212e-05, "loss": 1.3515, "step": 8133 }, { "epoch": 0.01, "learning_rate": 4.999203589060946e-05, "loss": 1.2288, "step": 8134 }, { "epoch": 0.01, "learning_rate": 4.999203390799008e-05, "loss": 1.0974, "step": 8135 }, { "epoch": 0.01, "learning_rate": 4.9992031925123994e-05, "loss": 1.0515, "step": 8136 }, { "epoch": 0.01, "learning_rate": 4.9992029942011197e-05, "loss": 1.1022, "step": 8137 }, { "epoch": 0.01, "learning_rate": 4.999202795865169e-05, "loss": 1.0654, "step": 8138 }, { "epoch": 0.01, "learning_rate": 4.9992025975045464e-05, "loss": 1.0587, "step": 8139 }, { "epoch": 0.01, "learning_rate": 4.9992023991192535e-05, "loss": 0.9211, "step": 8140 }, { "epoch": 0.01, "learning_rate": 4.99920220070929e-05, "loss": 1.0319, "step": 8141 }, { "epoch": 0.01, "learning_rate": 4.999202002274654e-05, "loss": 1.1965, "step": 8142 }, { "epoch": 0.01, "learning_rate": 4.999201803815348e-05, "loss": 1.1252, "step": 8143 }, { "epoch": 0.01, "learning_rate": 4.9992016053313706e-05, "loss": 1.1956, "step": 8144 }, { "epoch": 0.01, "learning_rate": 4.999201406822722e-05, "loss": 1.0656, "step": 8145 }, { "epoch": 0.01, "learning_rate": 4.999201208289403e-05, "loss": 1.1604, "step": 8146 }, { "epoch": 0.01, "learning_rate": 4.9992010097314124e-05, "loss": 0.9163, "step": 8147 }, { "epoch": 0.01, "learning_rate": 4.999200811148751e-05, "loss": 1.1694, "step": 8148 }, { "epoch": 0.01, "learning_rate": 4.999200612541418e-05, "loss": 1.1699, "step": 8149 }, { "epoch": 0.01, "learning_rate": 4.999200413909414e-05, "loss": 1.3381, "step": 8150 }, { "epoch": 0.01, "learning_rate": 4.9992002152527395e-05, "loss": 1.2251, "step": 8151 }, { "epoch": 0.01, "learning_rate": 4.999200016571394e-05, "loss": 1.0544, "step": 8152 }, { "epoch": 0.01, "learning_rate": 4.999199817865377e-05, "loss": 1.4957, "step": 8153 }, { "epoch": 0.01, "learning_rate": 4.999199619134689e-05, "loss": 1.6528, "step": 8154 }, { "epoch": 0.01, "learning_rate": 4.9991994203793303e-05, "loss": 1.618, "step": 8155 }, { "epoch": 0.01, "learning_rate": 4.9991992215993e-05, "loss": 1.1554, "step": 8156 }, { "epoch": 0.01, "learning_rate": 4.9991990227945996e-05, "loss": 1.1553, "step": 8157 }, { "epoch": 0.01, "learning_rate": 4.999198823965228e-05, "loss": 1.0059, "step": 8158 }, { "epoch": 0.01, "learning_rate": 4.999198625111184e-05, "loss": 1.2516, "step": 8159 }, { "epoch": 0.01, "learning_rate": 4.9991984262324696e-05, "loss": 1.2199, "step": 8160 }, { "epoch": 0.01, "learning_rate": 4.9991982273290846e-05, "loss": 1.0004, "step": 8161 }, { "epoch": 0.01, "learning_rate": 4.9991980284010287e-05, "loss": 1.0749, "step": 8162 }, { "epoch": 0.01, "learning_rate": 4.999197829448302e-05, "loss": 0.9781, "step": 8163 }, { "epoch": 0.01, "learning_rate": 4.999197630470903e-05, "loss": 0.3949, "step": 8164 }, { "epoch": 0.01, "learning_rate": 4.999197431468834e-05, "loss": 1.1743, "step": 8165 }, { "epoch": 0.01, "learning_rate": 4.999197232442093e-05, "loss": 0.8457, "step": 8166 }, { "epoch": 0.01, "learning_rate": 4.999197033390682e-05, "loss": 0.8697, "step": 8167 }, { "epoch": 0.01, "learning_rate": 4.9991968343146e-05, "loss": 1.1964, "step": 8168 }, { "epoch": 0.01, "learning_rate": 4.999196635213846e-05, "loss": 0.86, "step": 8169 }, { "epoch": 0.01, "learning_rate": 4.999196436088422e-05, "loss": 1.0926, "step": 8170 }, { "epoch": 0.01, "learning_rate": 4.999196236938327e-05, "loss": 1.0888, "step": 8171 }, { "epoch": 0.01, "learning_rate": 4.999196037763561e-05, "loss": 0.7222, "step": 8172 }, { "epoch": 0.01, "learning_rate": 4.999195838564123e-05, "loss": 1.1379, "step": 8173 }, { "epoch": 0.01, "learning_rate": 4.999195639340015e-05, "loss": 1.0324, "step": 8174 }, { "epoch": 0.01, "learning_rate": 4.9991954400912355e-05, "loss": 1.1589, "step": 8175 }, { "epoch": 0.01, "learning_rate": 4.9991952408177846e-05, "loss": 1.0928, "step": 8176 }, { "epoch": 0.01, "learning_rate": 4.999195041519663e-05, "loss": 0.9772, "step": 8177 }, { "epoch": 0.01, "learning_rate": 4.999194842196871e-05, "loss": 1.2171, "step": 8178 }, { "epoch": 0.01, "learning_rate": 4.999194642849408e-05, "loss": 1.1504, "step": 8179 }, { "epoch": 0.01, "learning_rate": 4.9991944434772734e-05, "loss": 1.19, "step": 8180 }, { "epoch": 0.01, "learning_rate": 4.999194244080467e-05, "loss": 1.1003, "step": 8181 }, { "epoch": 0.01, "learning_rate": 4.9991940446589916e-05, "loss": 1.0247, "step": 8182 }, { "epoch": 0.01, "learning_rate": 4.999193845212844e-05, "loss": 1.1476, "step": 8183 }, { "epoch": 0.01, "learning_rate": 4.999193645742025e-05, "loss": 1.2773, "step": 8184 }, { "epoch": 0.01, "learning_rate": 4.9991934462465364e-05, "loss": 1.5125, "step": 8185 }, { "epoch": 0.01, "learning_rate": 4.999193246726376e-05, "loss": 1.5328, "step": 8186 }, { "epoch": 0.01, "learning_rate": 4.999193047181544e-05, "loss": 0.9536, "step": 8187 }, { "epoch": 0.01, "learning_rate": 4.999192847612042e-05, "loss": 0.8888, "step": 8188 }, { "epoch": 0.01, "learning_rate": 4.9991926480178684e-05, "loss": 1.2458, "step": 8189 }, { "epoch": 0.01, "learning_rate": 4.9991924483990246e-05, "loss": 1.0091, "step": 8190 }, { "epoch": 0.01, "learning_rate": 4.999192248755509e-05, "loss": 0.1777, "step": 8191 }, { "epoch": 0.01, "learning_rate": 4.999192049087323e-05, "loss": 0.1513, "step": 8192 }, { "epoch": 0.01, "learning_rate": 4.9991918493944655e-05, "loss": 0.2262, "step": 8193 }, { "epoch": 0.01, "learning_rate": 4.9991916496769376e-05, "loss": 1.1995, "step": 8194 }, { "epoch": 0.01, "learning_rate": 4.999191449934738e-05, "loss": 1.1373, "step": 8195 }, { "epoch": 0.01, "learning_rate": 4.9991912501678686e-05, "loss": 1.0447, "step": 8196 }, { "epoch": 0.01, "learning_rate": 4.9991910503763276e-05, "loss": 1.281, "step": 8197 }, { "epoch": 0.01, "learning_rate": 4.999190850560115e-05, "loss": 0.925, "step": 8198 }, { "epoch": 0.01, "learning_rate": 4.999190650719232e-05, "loss": 1.2642, "step": 8199 }, { "epoch": 0.01, "learning_rate": 4.999190450853678e-05, "loss": 1.0075, "step": 8200 }, { "epoch": 0.01, "learning_rate": 4.9991902509634534e-05, "loss": 1.1623, "step": 8201 }, { "epoch": 0.01, "learning_rate": 4.999190051048557e-05, "loss": 0.9961, "step": 8202 }, { "epoch": 0.01, "learning_rate": 4.999189851108991e-05, "loss": 1.1757, "step": 8203 }, { "epoch": 0.01, "learning_rate": 4.999189651144753e-05, "loss": 1.3607, "step": 8204 }, { "epoch": 0.01, "learning_rate": 4.999189451155844e-05, "loss": 1.2876, "step": 8205 }, { "epoch": 0.01, "learning_rate": 4.999189251142265e-05, "loss": 1.4278, "step": 8206 }, { "epoch": 0.01, "learning_rate": 4.999189051104014e-05, "loss": 1.3018, "step": 8207 }, { "epoch": 0.01, "learning_rate": 4.999188851041092e-05, "loss": 1.1068, "step": 8208 }, { "epoch": 0.01, "learning_rate": 4.9991886509535e-05, "loss": 1.1355, "step": 8209 }, { "epoch": 0.01, "learning_rate": 4.9991884508412364e-05, "loss": 0.9839, "step": 8210 }, { "epoch": 0.01, "learning_rate": 4.999188250704302e-05, "loss": 1.1312, "step": 8211 }, { "epoch": 0.01, "learning_rate": 4.999188050542697e-05, "loss": 0.9444, "step": 8212 }, { "epoch": 0.01, "learning_rate": 4.99918785035642e-05, "loss": 0.7048, "step": 8213 }, { "epoch": 0.01, "learning_rate": 4.999187650145474e-05, "loss": 0.9837, "step": 8214 }, { "epoch": 0.01, "learning_rate": 4.999187449909856e-05, "loss": 1.1194, "step": 8215 }, { "epoch": 0.01, "learning_rate": 4.999187249649566e-05, "loss": 1.1048, "step": 8216 }, { "epoch": 0.01, "learning_rate": 4.999187049364607e-05, "loss": 0.9613, "step": 8217 }, { "epoch": 0.01, "learning_rate": 4.999186849054975e-05, "loss": 1.1043, "step": 8218 }, { "epoch": 0.01, "learning_rate": 4.999186648720674e-05, "loss": 1.1788, "step": 8219 }, { "epoch": 0.01, "learning_rate": 4.999186448361701e-05, "loss": 1.0607, "step": 8220 }, { "epoch": 0.01, "learning_rate": 4.999186247978058e-05, "loss": 1.3091, "step": 8221 }, { "epoch": 0.01, "learning_rate": 4.999186047569743e-05, "loss": 1.0853, "step": 8222 }, { "epoch": 0.01, "learning_rate": 4.9991858471367584e-05, "loss": 0.5792, "step": 8223 }, { "epoch": 0.01, "learning_rate": 4.999185646679101e-05, "loss": 0.6422, "step": 8224 }, { "epoch": 0.01, "learning_rate": 4.999185446196774e-05, "loss": 2.3584, "step": 8225 }, { "epoch": 0.01, "learning_rate": 4.999185245689776e-05, "loss": 2.8526, "step": 8226 }, { "epoch": 0.01, "learning_rate": 4.999185045158107e-05, "loss": 2.4602, "step": 8227 }, { "epoch": 0.01, "learning_rate": 4.999184844601767e-05, "loss": 2.0139, "step": 8228 }, { "epoch": 0.01, "learning_rate": 4.999184644020756e-05, "loss": 1.1379, "step": 8229 }, { "epoch": 0.01, "learning_rate": 4.999184443415075e-05, "loss": 1.3758, "step": 8230 }, { "epoch": 0.01, "learning_rate": 4.999184242784722e-05, "loss": 1.0097, "step": 8231 }, { "epoch": 0.01, "learning_rate": 4.9991840421296984e-05, "loss": 1.2272, "step": 8232 }, { "epoch": 0.01, "learning_rate": 4.9991838414500034e-05, "loss": 1.3314, "step": 8233 }, { "epoch": 0.01, "learning_rate": 4.999183640745638e-05, "loss": 0.9933, "step": 8234 }, { "epoch": 0.01, "learning_rate": 4.999183440016603e-05, "loss": 0.9973, "step": 8235 }, { "epoch": 0.01, "learning_rate": 4.9991832392628955e-05, "loss": 1.0917, "step": 8236 }, { "epoch": 0.01, "learning_rate": 4.999183038484517e-05, "loss": 0.9456, "step": 8237 }, { "epoch": 0.01, "learning_rate": 4.999182837681469e-05, "loss": 1.409, "step": 8238 }, { "epoch": 0.01, "learning_rate": 4.9991826368537485e-05, "loss": 1.0647, "step": 8239 }, { "epoch": 0.01, "learning_rate": 4.9991824360013584e-05, "loss": 0.9981, "step": 8240 }, { "epoch": 0.01, "learning_rate": 4.9991822351242965e-05, "loss": 1.1877, "step": 8241 }, { "epoch": 0.01, "learning_rate": 4.9991820342225644e-05, "loss": 1.3668, "step": 8242 }, { "epoch": 0.01, "learning_rate": 4.999181833296162e-05, "loss": 1.2226, "step": 8243 }, { "epoch": 0.01, "learning_rate": 4.999181632345087e-05, "loss": 1.1048, "step": 8244 }, { "epoch": 0.01, "learning_rate": 4.999181431369342e-05, "loss": 0.9423, "step": 8245 }, { "epoch": 0.01, "learning_rate": 4.999181230368926e-05, "loss": 1.2384, "step": 8246 }, { "epoch": 0.01, "learning_rate": 4.9991810293438395e-05, "loss": 1.7832, "step": 8247 }, { "epoch": 0.01, "learning_rate": 4.999180828294082e-05, "loss": 0.9334, "step": 8248 }, { "epoch": 0.01, "learning_rate": 4.999180627219653e-05, "loss": 1.1745, "step": 8249 }, { "epoch": 0.01, "learning_rate": 4.999180426120554e-05, "loss": 0.8047, "step": 8250 }, { "epoch": 0.01, "learning_rate": 4.999180224996784e-05, "loss": 1.095, "step": 8251 }, { "epoch": 0.01, "learning_rate": 4.999180023848343e-05, "loss": 0.8205, "step": 8252 }, { "epoch": 0.01, "learning_rate": 4.999179822675231e-05, "loss": 1.1301, "step": 8253 }, { "epoch": 0.01, "learning_rate": 4.999179621477449e-05, "loss": 1.3518, "step": 8254 }, { "epoch": 0.01, "learning_rate": 4.9991794202549954e-05, "loss": 1.0477, "step": 8255 }, { "epoch": 0.01, "learning_rate": 4.999179219007871e-05, "loss": 0.4998, "step": 8256 }, { "epoch": 0.01, "learning_rate": 4.9991790177360755e-05, "loss": 1.0548, "step": 8257 }, { "epoch": 0.01, "learning_rate": 4.999178816439609e-05, "loss": 1.2165, "step": 8258 }, { "epoch": 0.01, "learning_rate": 4.999178615118473e-05, "loss": 0.8835, "step": 8259 }, { "epoch": 0.01, "learning_rate": 4.999178413772665e-05, "loss": 1.2443, "step": 8260 }, { "epoch": 0.01, "learning_rate": 4.999178212402186e-05, "loss": 0.9742, "step": 8261 }, { "epoch": 0.01, "learning_rate": 4.999178011007037e-05, "loss": 1.2147, "step": 8262 }, { "epoch": 0.01, "learning_rate": 4.999177809587217e-05, "loss": 1.1196, "step": 8263 }, { "epoch": 0.01, "learning_rate": 4.9991776081427255e-05, "loss": 1.172, "step": 8264 }, { "epoch": 0.01, "learning_rate": 4.999177406673564e-05, "loss": 1.1525, "step": 8265 }, { "epoch": 0.01, "learning_rate": 4.9991772051797306e-05, "loss": 0.9987, "step": 8266 }, { "epoch": 0.01, "learning_rate": 4.999177003661227e-05, "loss": 1.1509, "step": 8267 }, { "epoch": 0.01, "learning_rate": 4.999176802118053e-05, "loss": 1.2156, "step": 8268 }, { "epoch": 0.01, "learning_rate": 4.999176600550208e-05, "loss": 1.0428, "step": 8269 }, { "epoch": 0.01, "learning_rate": 4.999176398957692e-05, "loss": 1.0568, "step": 8270 }, { "epoch": 0.01, "learning_rate": 4.999176197340505e-05, "loss": 1.1073, "step": 8271 }, { "epoch": 0.01, "learning_rate": 4.999175995698647e-05, "loss": 1.2529, "step": 8272 }, { "epoch": 0.01, "learning_rate": 4.9991757940321186e-05, "loss": 1.3799, "step": 8273 }, { "epoch": 0.01, "learning_rate": 4.999175592340919e-05, "loss": 1.3727, "step": 8274 }, { "epoch": 0.01, "learning_rate": 4.9991753906250494e-05, "loss": 1.2017, "step": 8275 }, { "epoch": 0.01, "learning_rate": 4.9991751888845086e-05, "loss": 0.9024, "step": 8276 }, { "epoch": 0.01, "learning_rate": 4.999174987119297e-05, "loss": 0.7042, "step": 8277 }, { "epoch": 0.01, "learning_rate": 4.999174785329415e-05, "loss": 0.962, "step": 8278 }, { "epoch": 0.01, "learning_rate": 4.999174583514861e-05, "loss": 1.1191, "step": 8279 }, { "epoch": 0.01, "learning_rate": 4.9991743816756367e-05, "loss": 1.3123, "step": 8280 }, { "epoch": 0.01, "learning_rate": 4.999174179811742e-05, "loss": 0.7662, "step": 8281 }, { "epoch": 0.01, "learning_rate": 4.999173977923176e-05, "loss": 0.8643, "step": 8282 }, { "epoch": 0.01, "learning_rate": 4.99917377600994e-05, "loss": 1.2898, "step": 8283 }, { "epoch": 0.01, "learning_rate": 4.9991735740720324e-05, "loss": 1.2237, "step": 8284 }, { "epoch": 0.01, "learning_rate": 4.9991733721094545e-05, "loss": 1.3147, "step": 8285 }, { "epoch": 0.01, "learning_rate": 4.9991731701222056e-05, "loss": 1.1869, "step": 8286 }, { "epoch": 0.01, "learning_rate": 4.999172968110286e-05, "loss": 1.265, "step": 8287 }, { "epoch": 0.01, "learning_rate": 4.999172766073695e-05, "loss": 1.3549, "step": 8288 }, { "epoch": 0.01, "learning_rate": 4.999172564012434e-05, "loss": 1.2528, "step": 8289 }, { "epoch": 0.01, "learning_rate": 4.999172361926502e-05, "loss": 1.1335, "step": 8290 }, { "epoch": 0.01, "learning_rate": 4.9991721598158996e-05, "loss": 1.05, "step": 8291 }, { "epoch": 0.01, "learning_rate": 4.999171957680626e-05, "loss": 1.0144, "step": 8292 }, { "epoch": 0.01, "learning_rate": 4.999171755520682e-05, "loss": 0.9581, "step": 8293 }, { "epoch": 0.01, "learning_rate": 4.9991715533360674e-05, "loss": 1.1402, "step": 8294 }, { "epoch": 0.01, "learning_rate": 4.999171351126781e-05, "loss": 1.1666, "step": 8295 }, { "epoch": 0.01, "learning_rate": 4.999171148892824e-05, "loss": 0.9455, "step": 8296 }, { "epoch": 0.01, "learning_rate": 4.9991709466341974e-05, "loss": 1.3822, "step": 8297 }, { "epoch": 0.01, "learning_rate": 4.999170744350899e-05, "loss": 1.3498, "step": 8298 }, { "epoch": 0.01, "learning_rate": 4.99917054204293e-05, "loss": 1.2778, "step": 8299 }, { "epoch": 0.01, "learning_rate": 4.999170339710291e-05, "loss": 1.682, "step": 8300 }, { "epoch": 0.01, "learning_rate": 4.99917013735298e-05, "loss": 1.2964, "step": 8301 }, { "epoch": 0.01, "learning_rate": 4.999169934970999e-05, "loss": 1.1784, "step": 8302 }, { "epoch": 0.01, "learning_rate": 4.999169732564347e-05, "loss": 1.1179, "step": 8303 }, { "epoch": 0.01, "learning_rate": 4.999169530133024e-05, "loss": 1.1623, "step": 8304 }, { "epoch": 0.01, "learning_rate": 4.999169327677031e-05, "loss": 1.2041, "step": 8305 }, { "epoch": 0.01, "learning_rate": 4.9991691251963665e-05, "loss": 1.1615, "step": 8306 }, { "epoch": 0.01, "learning_rate": 4.999168922691032e-05, "loss": 1.0726, "step": 8307 }, { "epoch": 0.01, "learning_rate": 4.999168720161026e-05, "loss": 0.8384, "step": 8308 }, { "epoch": 0.01, "learning_rate": 4.99916851760635e-05, "loss": 0.9102, "step": 8309 }, { "epoch": 0.01, "learning_rate": 4.999168315027003e-05, "loss": 0.8444, "step": 8310 }, { "epoch": 0.01, "learning_rate": 4.999168112422985e-05, "loss": 0.7279, "step": 8311 }, { "epoch": 0.01, "learning_rate": 4.999167909794296e-05, "loss": 0.5946, "step": 8312 }, { "epoch": 0.01, "learning_rate": 4.999167707140937e-05, "loss": 0.8794, "step": 8313 }, { "epoch": 0.01, "learning_rate": 4.999167504462907e-05, "loss": 1.1536, "step": 8314 }, { "epoch": 0.01, "learning_rate": 4.999167301760206e-05, "loss": 0.9535, "step": 8315 }, { "epoch": 0.01, "learning_rate": 4.9991670990328346e-05, "loss": 1.1651, "step": 8316 }, { "epoch": 0.01, "learning_rate": 4.999166896280793e-05, "loss": 0.8064, "step": 8317 }, { "epoch": 0.01, "learning_rate": 4.9991666935040793e-05, "loss": 0.8431, "step": 8318 }, { "epoch": 0.01, "learning_rate": 4.9991664907026955e-05, "loss": 1.3013, "step": 8319 }, { "epoch": 0.01, "learning_rate": 4.999166287876641e-05, "loss": 1.0501, "step": 8320 }, { "epoch": 0.01, "learning_rate": 4.999166085025916e-05, "loss": 0.9354, "step": 8321 }, { "epoch": 0.01, "learning_rate": 4.999165882150521e-05, "loss": 0.9781, "step": 8322 }, { "epoch": 0.01, "learning_rate": 4.9991656792504535e-05, "loss": 0.8342, "step": 8323 }, { "epoch": 0.01, "learning_rate": 4.9991654763257166e-05, "loss": 1.1559, "step": 8324 }, { "epoch": 0.01, "learning_rate": 4.999165273376309e-05, "loss": 1.1642, "step": 8325 }, { "epoch": 0.01, "learning_rate": 4.99916507040223e-05, "loss": 1.2031, "step": 8326 }, { "epoch": 0.01, "learning_rate": 4.9991648674034805e-05, "loss": 1.2286, "step": 8327 }, { "epoch": 0.01, "learning_rate": 4.99916466438006e-05, "loss": 1.1127, "step": 8328 }, { "epoch": 0.01, "learning_rate": 4.99916446133197e-05, "loss": 1.3282, "step": 8329 }, { "epoch": 0.01, "learning_rate": 4.999164258259208e-05, "loss": 1.2678, "step": 8330 }, { "epoch": 0.01, "learning_rate": 4.9991640551617766e-05, "loss": 1.3532, "step": 8331 }, { "epoch": 0.01, "learning_rate": 4.9991638520396736e-05, "loss": 1.1879, "step": 8332 }, { "epoch": 0.01, "learning_rate": 4.9991636488928996e-05, "loss": 1.2062, "step": 8333 }, { "epoch": 0.01, "learning_rate": 4.999163445721455e-05, "loss": 1.2149, "step": 8334 }, { "epoch": 0.01, "learning_rate": 4.999163242525341e-05, "loss": 1.1491, "step": 8335 }, { "epoch": 0.01, "learning_rate": 4.9991630393045554e-05, "loss": 1.0911, "step": 8336 }, { "epoch": 0.01, "learning_rate": 4.9991628360590986e-05, "loss": 1.0811, "step": 8337 }, { "epoch": 0.01, "learning_rate": 4.9991626327889715e-05, "loss": 1.2723, "step": 8338 }, { "epoch": 0.01, "learning_rate": 4.999162429494174e-05, "loss": 0.9685, "step": 8339 }, { "epoch": 0.01, "learning_rate": 4.9991622261747056e-05, "loss": 0.91, "step": 8340 }, { "epoch": 0.01, "learning_rate": 4.999162022830567e-05, "loss": 1.0759, "step": 8341 }, { "epoch": 0.01, "learning_rate": 4.9991618194617577e-05, "loss": 1.2971, "step": 8342 }, { "epoch": 0.01, "learning_rate": 4.999161616068276e-05, "loss": 1.0533, "step": 8343 }, { "epoch": 0.01, "learning_rate": 4.9991614126501256e-05, "loss": 0.9694, "step": 8344 }, { "epoch": 0.01, "learning_rate": 4.9991612092073034e-05, "loss": 1.0244, "step": 8345 }, { "epoch": 0.01, "learning_rate": 4.999161005739812e-05, "loss": 1.284, "step": 8346 }, { "epoch": 0.01, "learning_rate": 4.9991608022476486e-05, "loss": 1.184, "step": 8347 }, { "epoch": 0.01, "learning_rate": 4.999160598730815e-05, "loss": 1.2793, "step": 8348 }, { "epoch": 0.01, "learning_rate": 4.9991603951893104e-05, "loss": 0.9557, "step": 8349 }, { "epoch": 0.01, "learning_rate": 4.999160191623136e-05, "loss": 1.1593, "step": 8350 }, { "epoch": 0.01, "learning_rate": 4.9991599880322895e-05, "loss": 1.137, "step": 8351 }, { "epoch": 0.01, "learning_rate": 4.9991597844167736e-05, "loss": 0.6558, "step": 8352 }, { "epoch": 0.01, "learning_rate": 4.9991595807765866e-05, "loss": 1.1623, "step": 8353 }, { "epoch": 0.01, "learning_rate": 4.9991593771117286e-05, "loss": 1.269, "step": 8354 }, { "epoch": 0.01, "learning_rate": 4.999159173422201e-05, "loss": 1.1349, "step": 8355 }, { "epoch": 0.01, "learning_rate": 4.9991589697080015e-05, "loss": 1.1626, "step": 8356 }, { "epoch": 0.01, "learning_rate": 4.999158765969132e-05, "loss": 1.279, "step": 8357 }, { "epoch": 0.01, "learning_rate": 4.999158562205592e-05, "loss": 1.2476, "step": 8358 }, { "epoch": 0.01, "learning_rate": 4.999158358417381e-05, "loss": 1.1814, "step": 8359 }, { "epoch": 0.01, "learning_rate": 4.9991581546045e-05, "loss": 1.097, "step": 8360 }, { "epoch": 0.01, "learning_rate": 4.9991579507669475e-05, "loss": 1.1707, "step": 8361 }, { "epoch": 0.01, "learning_rate": 4.999157746904725e-05, "loss": 1.0193, "step": 8362 }, { "epoch": 0.01, "learning_rate": 4.9991575430178316e-05, "loss": 0.79, "step": 8363 }, { "epoch": 0.01, "learning_rate": 4.9991573391062674e-05, "loss": 1.1222, "step": 8364 }, { "epoch": 0.01, "learning_rate": 4.999157135170033e-05, "loss": 1.1173, "step": 8365 }, { "epoch": 0.01, "learning_rate": 4.9991569312091274e-05, "loss": 1.5073, "step": 8366 }, { "epoch": 0.01, "learning_rate": 4.9991567272235516e-05, "loss": 0.9067, "step": 8367 }, { "epoch": 0.01, "learning_rate": 4.9991565232133054e-05, "loss": 0.6524, "step": 8368 }, { "epoch": 0.01, "learning_rate": 4.9991563191783875e-05, "loss": 0.7113, "step": 8369 }, { "epoch": 0.01, "learning_rate": 4.9991561151188e-05, "loss": 0.7702, "step": 8370 }, { "epoch": 0.01, "learning_rate": 4.9991559110345413e-05, "loss": 1.1213, "step": 8371 }, { "epoch": 0.01, "learning_rate": 4.9991557069256124e-05, "loss": 1.0213, "step": 8372 }, { "epoch": 0.01, "learning_rate": 4.999155502792013e-05, "loss": 1.1007, "step": 8373 }, { "epoch": 0.01, "learning_rate": 4.999155298633743e-05, "loss": 1.0572, "step": 8374 }, { "epoch": 0.01, "learning_rate": 4.999155094450802e-05, "loss": 0.9129, "step": 8375 }, { "epoch": 0.01, "learning_rate": 4.9991548902431906e-05, "loss": 1.0989, "step": 8376 }, { "epoch": 0.01, "learning_rate": 4.9991546860109086e-05, "loss": 0.918, "step": 8377 }, { "epoch": 0.01, "learning_rate": 4.999154481753956e-05, "loss": 1.1115, "step": 8378 }, { "epoch": 0.01, "learning_rate": 4.999154277472333e-05, "loss": 1.1857, "step": 8379 }, { "epoch": 0.01, "learning_rate": 4.999154073166039e-05, "loss": 1.066, "step": 8380 }, { "epoch": 0.01, "learning_rate": 4.9991538688350745e-05, "loss": 0.7562, "step": 8381 }, { "epoch": 0.01, "learning_rate": 4.9991536644794394e-05, "loss": 1.1435, "step": 8382 }, { "epoch": 0.01, "learning_rate": 4.999153460099134e-05, "loss": 0.8951, "step": 8383 }, { "epoch": 0.01, "learning_rate": 4.9991532556941576e-05, "loss": 1.1395, "step": 8384 }, { "epoch": 0.01, "learning_rate": 4.999153051264511e-05, "loss": 0.9641, "step": 8385 }, { "epoch": 0.01, "learning_rate": 4.999152846810193e-05, "loss": 1.1511, "step": 8386 }, { "epoch": 0.01, "learning_rate": 4.999152642331205e-05, "loss": 0.9422, "step": 8387 }, { "epoch": 0.01, "learning_rate": 4.9991524378275464e-05, "loss": 1.1244, "step": 8388 }, { "epoch": 0.01, "learning_rate": 4.999152233299218e-05, "loss": 1.1136, "step": 8389 }, { "epoch": 0.01, "learning_rate": 4.999152028746218e-05, "loss": 1.1821, "step": 8390 }, { "epoch": 0.01, "learning_rate": 4.9991518241685476e-05, "loss": 1.1484, "step": 8391 }, { "epoch": 0.01, "learning_rate": 4.999151619566207e-05, "loss": 0.9898, "step": 8392 }, { "epoch": 0.01, "learning_rate": 4.9991514149391955e-05, "loss": 0.9929, "step": 8393 }, { "epoch": 0.01, "learning_rate": 4.9991512102875137e-05, "loss": 1.2335, "step": 8394 }, { "epoch": 0.01, "learning_rate": 4.9991510056111614e-05, "loss": 0.8588, "step": 8395 }, { "epoch": 0.01, "learning_rate": 4.999150800910138e-05, "loss": 0.5444, "step": 8396 }, { "epoch": 0.01, "learning_rate": 4.9991505961844446e-05, "loss": 0.6808, "step": 8397 }, { "epoch": 0.01, "learning_rate": 4.99915039143408e-05, "loss": 0.3225, "step": 8398 }, { "epoch": 0.01, "learning_rate": 4.999150186659045e-05, "loss": 0.5494, "step": 8399 }, { "epoch": 0.01, "learning_rate": 4.9991499818593404e-05, "loss": 0.6306, "step": 8400 }, { "epoch": 0.01, "learning_rate": 4.999149777034964e-05, "loss": 1.2615, "step": 8401 }, { "epoch": 0.01, "learning_rate": 4.9991495721859174e-05, "loss": 1.2836, "step": 8402 }, { "epoch": 0.01, "learning_rate": 4.9991493673122004e-05, "loss": 1.2379, "step": 8403 }, { "epoch": 0.01, "learning_rate": 4.999149162413813e-05, "loss": 1.2976, "step": 8404 }, { "epoch": 0.01, "learning_rate": 4.999148957490755e-05, "loss": 0.9837, "step": 8405 }, { "epoch": 0.01, "learning_rate": 4.999148752543026e-05, "loss": 1.1341, "step": 8406 }, { "epoch": 0.01, "learning_rate": 4.999148547570627e-05, "loss": 0.6938, "step": 8407 }, { "epoch": 0.01, "learning_rate": 4.9991483425735575e-05, "loss": 1.1328, "step": 8408 }, { "epoch": 0.01, "learning_rate": 4.9991481375518165e-05, "loss": 1.086, "step": 8409 }, { "epoch": 0.01, "learning_rate": 4.999147932505406e-05, "loss": 1.2981, "step": 8410 }, { "epoch": 0.01, "learning_rate": 4.9991477274343246e-05, "loss": 1.4599, "step": 8411 }, { "epoch": 0.01, "learning_rate": 4.999147522338573e-05, "loss": 1.4413, "step": 8412 }, { "epoch": 0.01, "learning_rate": 4.99914731721815e-05, "loss": 0.7609, "step": 8413 }, { "epoch": 0.01, "learning_rate": 4.999147112073057e-05, "loss": 0.608, "step": 8414 }, { "epoch": 0.01, "learning_rate": 4.9991469069032934e-05, "loss": 0.3055, "step": 8415 }, { "epoch": 0.01, "learning_rate": 4.99914670170886e-05, "loss": 0.4677, "step": 8416 }, { "epoch": 0.01, "learning_rate": 4.999146496489755e-05, "loss": 0.4433, "step": 8417 }, { "epoch": 0.01, "learning_rate": 4.9991462912459805e-05, "loss": 0.3726, "step": 8418 }, { "epoch": 0.01, "learning_rate": 4.9991460859775346e-05, "loss": 0.2769, "step": 8419 }, { "epoch": 0.01, "learning_rate": 4.9991458806844184e-05, "loss": 0.3368, "step": 8420 }, { "epoch": 0.01, "learning_rate": 4.999145675366632e-05, "loss": 0.3348, "step": 8421 }, { "epoch": 0.01, "learning_rate": 4.999145470024175e-05, "loss": 0.3991, "step": 8422 }, { "epoch": 0.01, "learning_rate": 4.999145264657047e-05, "loss": 0.9928, "step": 8423 }, { "epoch": 0.01, "learning_rate": 4.999145059265249e-05, "loss": 1.1942, "step": 8424 }, { "epoch": 0.01, "learning_rate": 4.999144853848781e-05, "loss": 1.1893, "step": 8425 }, { "epoch": 0.01, "learning_rate": 4.999144648407641e-05, "loss": 0.9798, "step": 8426 }, { "epoch": 0.01, "learning_rate": 4.999144442941832e-05, "loss": 0.7083, "step": 8427 }, { "epoch": 0.01, "learning_rate": 4.9991442374513516e-05, "loss": 0.942, "step": 8428 }, { "epoch": 0.01, "learning_rate": 4.999144031936201e-05, "loss": 1.0437, "step": 8429 }, { "epoch": 0.01, "learning_rate": 4.99914382639638e-05, "loss": 0.7463, "step": 8430 }, { "epoch": 0.01, "learning_rate": 4.999143620831888e-05, "loss": 0.7416, "step": 8431 }, { "epoch": 0.01, "learning_rate": 4.999143415242726e-05, "loss": 1.4202, "step": 8432 }, { "epoch": 0.01, "learning_rate": 4.9991432096288935e-05, "loss": 2.1545, "step": 8433 }, { "epoch": 0.01, "learning_rate": 4.9991430039903904e-05, "loss": 2.3004, "step": 8434 }, { "epoch": 0.01, "learning_rate": 4.9991427983272164e-05, "loss": 1.0673, "step": 8435 }, { "epoch": 0.01, "learning_rate": 4.9991425926393726e-05, "loss": 1.0957, "step": 8436 }, { "epoch": 0.01, "learning_rate": 4.9991423869268585e-05, "loss": 0.9421, "step": 8437 }, { "epoch": 0.01, "learning_rate": 4.9991421811896734e-05, "loss": 0.3061, "step": 8438 }, { "epoch": 0.01, "learning_rate": 4.999141975427817e-05, "loss": 1.0725, "step": 8439 }, { "epoch": 0.01, "learning_rate": 4.9991417696412915e-05, "loss": 1.3373, "step": 8440 }, { "epoch": 0.01, "learning_rate": 4.999141563830095e-05, "loss": 1.6996, "step": 8441 }, { "epoch": 0.01, "learning_rate": 4.999141357994228e-05, "loss": 1.1349, "step": 8442 }, { "epoch": 0.01, "learning_rate": 4.999141152133691e-05, "loss": 1.2943, "step": 8443 }, { "epoch": 0.01, "learning_rate": 4.999140946248483e-05, "loss": 1.1657, "step": 8444 }, { "epoch": 0.01, "learning_rate": 4.999140740338604e-05, "loss": 1.0395, "step": 8445 }, { "epoch": 0.01, "learning_rate": 4.9991405344040555e-05, "loss": 1.4422, "step": 8446 }, { "epoch": 0.01, "learning_rate": 4.999140328444836e-05, "loss": 1.5469, "step": 8447 }, { "epoch": 0.01, "learning_rate": 4.999140122460946e-05, "loss": 1.4338, "step": 8448 }, { "epoch": 0.01, "learning_rate": 4.999139916452386e-05, "loss": 1.5494, "step": 8449 }, { "epoch": 0.01, "learning_rate": 4.999139710419155e-05, "loss": 1.0026, "step": 8450 }, { "epoch": 0.01, "learning_rate": 4.9991395043612544e-05, "loss": 0.7611, "step": 8451 }, { "epoch": 0.01, "learning_rate": 4.9991392982786825e-05, "loss": 0.9159, "step": 8452 }, { "epoch": 0.01, "learning_rate": 4.999139092171441e-05, "loss": 1.0654, "step": 8453 }, { "epoch": 0.01, "learning_rate": 4.999138886039528e-05, "loss": 1.3249, "step": 8454 }, { "epoch": 0.01, "learning_rate": 4.999138679882945e-05, "loss": 1.4767, "step": 8455 }, { "epoch": 0.01, "learning_rate": 4.999138473701692e-05, "loss": 1.1754, "step": 8456 }, { "epoch": 0.01, "learning_rate": 4.999138267495769e-05, "loss": 1.2345, "step": 8457 }, { "epoch": 0.01, "learning_rate": 4.999138061265174e-05, "loss": 1.1498, "step": 8458 }, { "epoch": 0.01, "learning_rate": 4.9991378550099093e-05, "loss": 1.229, "step": 8459 }, { "epoch": 0.01, "learning_rate": 4.999137648729975e-05, "loss": 1.3139, "step": 8460 }, { "epoch": 0.01, "learning_rate": 4.999137442425369e-05, "loss": 0.9882, "step": 8461 }, { "epoch": 0.01, "learning_rate": 4.999137236096093e-05, "loss": 1.2208, "step": 8462 }, { "epoch": 0.01, "learning_rate": 4.9991370297421466e-05, "loss": 1.1347, "step": 8463 }, { "epoch": 0.01, "learning_rate": 4.999136823363529e-05, "loss": 1.2023, "step": 8464 }, { "epoch": 0.01, "learning_rate": 4.999136616960243e-05, "loss": 1.6499, "step": 8465 }, { "epoch": 0.01, "learning_rate": 4.999136410532285e-05, "loss": 1.1429, "step": 8466 }, { "epoch": 0.01, "learning_rate": 4.999136204079656e-05, "loss": 1.0542, "step": 8467 }, { "epoch": 0.01, "learning_rate": 4.9991359976023575e-05, "loss": 1.0439, "step": 8468 }, { "epoch": 0.01, "learning_rate": 4.999135791100389e-05, "loss": 0.9817, "step": 8469 }, { "epoch": 0.01, "learning_rate": 4.9991355845737496e-05, "loss": 0.9858, "step": 8470 }, { "epoch": 0.01, "learning_rate": 4.9991353780224405e-05, "loss": 1.2853, "step": 8471 }, { "epoch": 0.01, "learning_rate": 4.9991351714464596e-05, "loss": 1.2696, "step": 8472 }, { "epoch": 0.01, "learning_rate": 4.999134964845809e-05, "loss": 1.5929, "step": 8473 }, { "epoch": 0.01, "learning_rate": 4.999134758220488e-05, "loss": 1.0557, "step": 8474 }, { "epoch": 0.01, "learning_rate": 4.999134551570497e-05, "loss": 1.1169, "step": 8475 }, { "epoch": 0.01, "learning_rate": 4.999134344895835e-05, "loss": 0.8522, "step": 8476 }, { "epoch": 0.01, "learning_rate": 4.999134138196503e-05, "loss": 1.2347, "step": 8477 }, { "epoch": 0.01, "learning_rate": 4.9991339314725e-05, "loss": 1.1851, "step": 8478 }, { "epoch": 0.01, "learning_rate": 4.999133724723827e-05, "loss": 1.2193, "step": 8479 }, { "epoch": 0.01, "learning_rate": 4.999133517950484e-05, "loss": 0.9337, "step": 8480 }, { "epoch": 0.01, "learning_rate": 4.9991333111524695e-05, "loss": 1.1871, "step": 8481 }, { "epoch": 0.01, "learning_rate": 4.999133104329786e-05, "loss": 1.1762, "step": 8482 }, { "epoch": 0.01, "learning_rate": 4.9991328974824314e-05, "loss": 1.2306, "step": 8483 }, { "epoch": 0.01, "learning_rate": 4.9991326906104064e-05, "loss": 0.8215, "step": 8484 }, { "epoch": 0.01, "learning_rate": 4.999132483713711e-05, "loss": 1.1186, "step": 8485 }, { "epoch": 0.01, "learning_rate": 4.9991322767923455e-05, "loss": 1.0112, "step": 8486 }, { "epoch": 0.01, "learning_rate": 4.999132069846309e-05, "loss": 1.1783, "step": 8487 }, { "epoch": 0.01, "learning_rate": 4.9991318628756026e-05, "loss": 1.3128, "step": 8488 }, { "epoch": 0.01, "learning_rate": 4.999131655880226e-05, "loss": 1.1166, "step": 8489 }, { "epoch": 0.01, "learning_rate": 4.999131448860178e-05, "loss": 0.995, "step": 8490 }, { "epoch": 0.01, "learning_rate": 4.999131241815461e-05, "loss": 1.2413, "step": 8491 }, { "epoch": 0.01, "learning_rate": 4.999131034746073e-05, "loss": 1.0379, "step": 8492 }, { "epoch": 0.01, "learning_rate": 4.9991308276520145e-05, "loss": 1.1876, "step": 8493 }, { "epoch": 0.01, "learning_rate": 4.9991306205332855e-05, "loss": 1.2118, "step": 8494 }, { "epoch": 0.01, "learning_rate": 4.999130413389887e-05, "loss": 1.316, "step": 8495 }, { "epoch": 0.01, "learning_rate": 4.999130206221817e-05, "loss": 0.973, "step": 8496 }, { "epoch": 0.01, "learning_rate": 4.999129999029078e-05, "loss": 1.3465, "step": 8497 }, { "epoch": 0.01, "learning_rate": 4.999129791811667e-05, "loss": 0.9186, "step": 8498 }, { "epoch": 0.01, "learning_rate": 4.9991295845695865e-05, "loss": 1.2283, "step": 8499 }, { "epoch": 0.01, "learning_rate": 4.999129377302836e-05, "loss": 1.1846, "step": 8500 }, { "epoch": 0.01, "eval_loss": 1.0542309284210205, "eval_runtime": 99.0013, "eval_samples_per_second": 13.99, "eval_steps_per_second": 3.505, "step": 8500 }, { "epoch": 0.01, "learning_rate": 4.999129170011415e-05, "loss": 0.6902, "step": 8501 }, { "epoch": 0.01, "learning_rate": 4.999128962695323e-05, "loss": 0.8112, "step": 8502 }, { "epoch": 0.01, "learning_rate": 4.9991287553545614e-05, "loss": 1.0512, "step": 8503 }, { "epoch": 0.01, "learning_rate": 4.999128547989129e-05, "loss": 1.0738, "step": 8504 }, { "epoch": 0.01, "learning_rate": 4.999128340599026e-05, "loss": 1.0696, "step": 8505 }, { "epoch": 0.01, "learning_rate": 4.999128133184253e-05, "loss": 0.9503, "step": 8506 }, { "epoch": 0.01, "learning_rate": 4.99912792574481e-05, "loss": 1.1886, "step": 8507 }, { "epoch": 0.01, "learning_rate": 4.999127718280697e-05, "loss": 1.2016, "step": 8508 }, { "epoch": 0.01, "learning_rate": 4.999127510791913e-05, "loss": 1.1265, "step": 8509 }, { "epoch": 0.01, "learning_rate": 4.999127303278458e-05, "loss": 1.1183, "step": 8510 }, { "epoch": 0.01, "learning_rate": 4.9991270957403334e-05, "loss": 1.6124, "step": 8511 }, { "epoch": 0.01, "learning_rate": 4.999126888177539e-05, "loss": 1.1347, "step": 8512 }, { "epoch": 0.01, "learning_rate": 4.999126680590074e-05, "loss": 1.2228, "step": 8513 }, { "epoch": 0.01, "learning_rate": 4.9991264729779374e-05, "loss": 1.1211, "step": 8514 }, { "epoch": 0.01, "learning_rate": 4.999126265341132e-05, "loss": 1.1596, "step": 8515 }, { "epoch": 0.01, "learning_rate": 4.999126057679655e-05, "loss": 0.8949, "step": 8516 }, { "epoch": 0.01, "learning_rate": 4.999125849993509e-05, "loss": 0.7996, "step": 8517 }, { "epoch": 0.01, "learning_rate": 4.999125642282692e-05, "loss": 0.7547, "step": 8518 }, { "epoch": 0.01, "learning_rate": 4.999125434547205e-05, "loss": 0.7877, "step": 8519 }, { "epoch": 0.01, "learning_rate": 4.9991252267870475e-05, "loss": 0.7012, "step": 8520 }, { "epoch": 0.01, "learning_rate": 4.99912501900222e-05, "loss": 0.7266, "step": 8521 }, { "epoch": 0.01, "learning_rate": 4.999124811192721e-05, "loss": 0.7987, "step": 8522 }, { "epoch": 0.01, "learning_rate": 4.999124603358553e-05, "loss": 0.6141, "step": 8523 }, { "epoch": 0.01, "learning_rate": 4.999124395499715e-05, "loss": 0.6994, "step": 8524 }, { "epoch": 0.01, "learning_rate": 4.9991241876162053e-05, "loss": 0.6509, "step": 8525 }, { "epoch": 0.01, "learning_rate": 4.999123979708026e-05, "loss": 0.8349, "step": 8526 }, { "epoch": 0.01, "learning_rate": 4.9991237717751757e-05, "loss": 0.3005, "step": 8527 }, { "epoch": 0.01, "learning_rate": 4.9991235638176556e-05, "loss": 1.373, "step": 8528 }, { "epoch": 0.01, "learning_rate": 4.999123355835466e-05, "loss": 1.057, "step": 8529 }, { "epoch": 0.01, "learning_rate": 4.999123147828605e-05, "loss": 1.438, "step": 8530 }, { "epoch": 0.01, "learning_rate": 4.999122939797075e-05, "loss": 1.1025, "step": 8531 }, { "epoch": 0.01, "learning_rate": 4.9991227317408735e-05, "loss": 1.017, "step": 8532 }, { "epoch": 0.01, "learning_rate": 4.9991225236600024e-05, "loss": 1.1234, "step": 8533 }, { "epoch": 0.01, "learning_rate": 4.999122315554461e-05, "loss": 1.251, "step": 8534 }, { "epoch": 0.01, "learning_rate": 4.9991221074242486e-05, "loss": 1.1081, "step": 8535 }, { "epoch": 0.01, "learning_rate": 4.999121899269366e-05, "loss": 1.1494, "step": 8536 }, { "epoch": 0.01, "learning_rate": 4.9991216910898134e-05, "loss": 1.2125, "step": 8537 }, { "epoch": 0.01, "learning_rate": 4.9991214828855907e-05, "loss": 1.5262, "step": 8538 }, { "epoch": 0.01, "learning_rate": 4.999121274656698e-05, "loss": 0.9857, "step": 8539 }, { "epoch": 0.01, "learning_rate": 4.999121066403134e-05, "loss": 1.1487, "step": 8540 }, { "epoch": 0.01, "learning_rate": 4.9991208581249e-05, "loss": 1.0883, "step": 8541 }, { "epoch": 0.01, "learning_rate": 4.999120649821997e-05, "loss": 1.5234, "step": 8542 }, { "epoch": 0.01, "learning_rate": 4.999120441494422e-05, "loss": 2.2477, "step": 8543 }, { "epoch": 0.01, "learning_rate": 4.999120233142178e-05, "loss": 2.9172, "step": 8544 }, { "epoch": 0.01, "learning_rate": 4.999120024765264e-05, "loss": 2.8604, "step": 8545 }, { "epoch": 0.01, "learning_rate": 4.999119816363679e-05, "loss": 2.2149, "step": 8546 }, { "epoch": 0.01, "learning_rate": 4.999119607937424e-05, "loss": 0.9414, "step": 8547 }, { "epoch": 0.01, "learning_rate": 4.999119399486498e-05, "loss": 1.1388, "step": 8548 }, { "epoch": 0.01, "learning_rate": 4.999119191010902e-05, "loss": 0.8004, "step": 8549 }, { "epoch": 0.01, "learning_rate": 4.999118982510637e-05, "loss": 1.2408, "step": 8550 }, { "epoch": 0.01, "learning_rate": 4.9991187739857e-05, "loss": 1.2502, "step": 8551 }, { "epoch": 0.01, "learning_rate": 4.999118565436094e-05, "loss": 1.3305, "step": 8552 }, { "epoch": 0.01, "learning_rate": 4.9991183568618175e-05, "loss": 0.9993, "step": 8553 }, { "epoch": 0.01, "learning_rate": 4.99911814826287e-05, "loss": 0.1552, "step": 8554 }, { "epoch": 0.01, "learning_rate": 4.9991179396392526e-05, "loss": 0.1205, "step": 8555 }, { "epoch": 0.01, "learning_rate": 4.999117730990965e-05, "loss": 0.1116, "step": 8556 }, { "epoch": 0.01, "learning_rate": 4.999117522318008e-05, "loss": 1.049, "step": 8557 }, { "epoch": 0.01, "learning_rate": 4.9991173136203795e-05, "loss": 1.0186, "step": 8558 }, { "epoch": 0.01, "learning_rate": 4.9991171048980816e-05, "loss": 1.154, "step": 8559 }, { "epoch": 0.01, "learning_rate": 4.999116896151114e-05, "loss": 1.5328, "step": 8560 }, { "epoch": 0.01, "learning_rate": 4.999116687379475e-05, "loss": 1.1713, "step": 8561 }, { "epoch": 0.01, "learning_rate": 4.999116478583166e-05, "loss": 1.2756, "step": 8562 }, { "epoch": 0.01, "learning_rate": 4.999116269762187e-05, "loss": 1.2244, "step": 8563 }, { "epoch": 0.01, "learning_rate": 4.9991160609165374e-05, "loss": 1.2125, "step": 8564 }, { "epoch": 0.01, "learning_rate": 4.999115852046218e-05, "loss": 1.0166, "step": 8565 }, { "epoch": 0.01, "learning_rate": 4.9991156431512285e-05, "loss": 0.795, "step": 8566 }, { "epoch": 0.01, "learning_rate": 4.9991154342315685e-05, "loss": 1.2471, "step": 8567 }, { "epoch": 0.01, "learning_rate": 4.999115225287239e-05, "loss": 0.8797, "step": 8568 }, { "epoch": 0.01, "learning_rate": 4.9991150163182374e-05, "loss": 0.3646, "step": 8569 }, { "epoch": 0.01, "learning_rate": 4.999114807324568e-05, "loss": 1.0403, "step": 8570 }, { "epoch": 0.01, "learning_rate": 4.999114598306227e-05, "loss": 1.2882, "step": 8571 }, { "epoch": 0.01, "learning_rate": 4.999114389263216e-05, "loss": 1.0653, "step": 8572 }, { "epoch": 0.01, "learning_rate": 4.9991141801955347e-05, "loss": 1.1562, "step": 8573 }, { "epoch": 0.01, "learning_rate": 4.999113971103183e-05, "loss": 1.317, "step": 8574 }, { "epoch": 0.01, "learning_rate": 4.9991137619861615e-05, "loss": 1.2079, "step": 8575 }, { "epoch": 0.01, "learning_rate": 4.99911355284447e-05, "loss": 1.2094, "step": 8576 }, { "epoch": 0.01, "learning_rate": 4.999113343678108e-05, "loss": 1.0105, "step": 8577 }, { "epoch": 0.01, "learning_rate": 4.999113134487075e-05, "loss": 0.6547, "step": 8578 }, { "epoch": 0.01, "learning_rate": 4.999112925271373e-05, "loss": 0.8973, "step": 8579 }, { "epoch": 0.01, "learning_rate": 4.999112716031001e-05, "loss": 1.044, "step": 8580 }, { "epoch": 0.01, "learning_rate": 4.999112506765958e-05, "loss": 1.189, "step": 8581 }, { "epoch": 0.01, "learning_rate": 4.999112297476245e-05, "loss": 1.3591, "step": 8582 }, { "epoch": 0.01, "learning_rate": 4.999112088161862e-05, "loss": 1.0588, "step": 8583 }, { "epoch": 0.01, "learning_rate": 4.9991118788228084e-05, "loss": 0.9702, "step": 8584 }, { "epoch": 0.01, "learning_rate": 4.9991116694590856e-05, "loss": 0.9531, "step": 8585 }, { "epoch": 0.01, "learning_rate": 4.999111460070691e-05, "loss": 0.8241, "step": 8586 }, { "epoch": 0.01, "learning_rate": 4.9991112506576276e-05, "loss": 1.1639, "step": 8587 }, { "epoch": 0.01, "learning_rate": 4.999111041219894e-05, "loss": 1.12, "step": 8588 }, { "epoch": 0.01, "learning_rate": 4.99911083175749e-05, "loss": 1.3793, "step": 8589 }, { "epoch": 0.01, "learning_rate": 4.999110622270415e-05, "loss": 0.9946, "step": 8590 }, { "epoch": 0.01, "learning_rate": 4.999110412758671e-05, "loss": 0.8347, "step": 8591 }, { "epoch": 0.01, "learning_rate": 4.999110203222256e-05, "loss": 0.4633, "step": 8592 }, { "epoch": 0.01, "learning_rate": 4.999109993661171e-05, "loss": 0.2479, "step": 8593 }, { "epoch": 0.01, "learning_rate": 4.9991097840754165e-05, "loss": 0.373, "step": 8594 }, { "epoch": 0.01, "learning_rate": 4.999109574464991e-05, "loss": 0.8449, "step": 8595 }, { "epoch": 0.01, "learning_rate": 4.999109364829896e-05, "loss": 1.3162, "step": 8596 }, { "epoch": 0.01, "learning_rate": 4.99910915517013e-05, "loss": 1.2068, "step": 8597 }, { "epoch": 0.01, "learning_rate": 4.999108945485695e-05, "loss": 1.4203, "step": 8598 }, { "epoch": 0.01, "learning_rate": 4.999108735776589e-05, "loss": 1.4168, "step": 8599 }, { "epoch": 0.01, "learning_rate": 4.999108526042813e-05, "loss": 1.3545, "step": 8600 }, { "epoch": 0.01, "learning_rate": 4.9991083162843666e-05, "loss": 0.9488, "step": 8601 }, { "epoch": 0.01, "learning_rate": 4.999108106501251e-05, "loss": 1.1733, "step": 8602 }, { "epoch": 0.01, "learning_rate": 4.9991078966934645e-05, "loss": 1.254, "step": 8603 }, { "epoch": 0.01, "learning_rate": 4.999107686861008e-05, "loss": 1.0556, "step": 8604 }, { "epoch": 0.01, "learning_rate": 4.9991074770038817e-05, "loss": 1.1763, "step": 8605 }, { "epoch": 0.01, "learning_rate": 4.9991072671220844e-05, "loss": 1.1645, "step": 8606 }, { "epoch": 0.01, "learning_rate": 4.9991070572156174e-05, "loss": 1.0256, "step": 8607 }, { "epoch": 0.01, "learning_rate": 4.99910684728448e-05, "loss": 1.0187, "step": 8608 }, { "epoch": 0.01, "learning_rate": 4.999106637328673e-05, "loss": 1.0299, "step": 8609 }, { "epoch": 0.01, "learning_rate": 4.9991064273481966e-05, "loss": 0.8712, "step": 8610 }, { "epoch": 0.01, "learning_rate": 4.999106217343049e-05, "loss": 1.0586, "step": 8611 }, { "epoch": 0.01, "learning_rate": 4.999106007313231e-05, "loss": 0.9963, "step": 8612 }, { "epoch": 0.01, "learning_rate": 4.999105797258744e-05, "loss": 1.1931, "step": 8613 }, { "epoch": 0.01, "learning_rate": 4.9991055871795854e-05, "loss": 1.1627, "step": 8614 }, { "epoch": 0.01, "learning_rate": 4.999105377075758e-05, "loss": 1.0302, "step": 8615 }, { "epoch": 0.01, "learning_rate": 4.99910516694726e-05, "loss": 1.0384, "step": 8616 }, { "epoch": 0.01, "learning_rate": 4.9991049567940914e-05, "loss": 1.2491, "step": 8617 }, { "epoch": 0.01, "learning_rate": 4.9991047466162534e-05, "loss": 0.9819, "step": 8618 }, { "epoch": 0.01, "learning_rate": 4.999104536413746e-05, "loss": 0.6744, "step": 8619 }, { "epoch": 0.01, "learning_rate": 4.9991043261865664e-05, "loss": 1.2716, "step": 8620 }, { "epoch": 0.01, "learning_rate": 4.999104115934719e-05, "loss": 1.1201, "step": 8621 }, { "epoch": 0.01, "learning_rate": 4.9991039056581994e-05, "loss": 0.8674, "step": 8622 }, { "epoch": 0.01, "learning_rate": 4.999103695357011e-05, "loss": 1.0039, "step": 8623 }, { "epoch": 0.01, "learning_rate": 4.999103485031152e-05, "loss": 1.1552, "step": 8624 }, { "epoch": 0.01, "learning_rate": 4.999103274680623e-05, "loss": 1.206, "step": 8625 }, { "epoch": 0.01, "learning_rate": 4.999103064305424e-05, "loss": 1.3881, "step": 8626 }, { "epoch": 0.01, "learning_rate": 4.999102853905555e-05, "loss": 0.9618, "step": 8627 }, { "epoch": 0.01, "learning_rate": 4.9991026434810154e-05, "loss": 1.0148, "step": 8628 }, { "epoch": 0.01, "learning_rate": 4.9991024330318064e-05, "loss": 0.8362, "step": 8629 }, { "epoch": 0.01, "learning_rate": 4.999102222557927e-05, "loss": 0.9687, "step": 8630 }, { "epoch": 0.01, "learning_rate": 4.999102012059378e-05, "loss": 1.278, "step": 8631 }, { "epoch": 0.01, "learning_rate": 4.999101801536158e-05, "loss": 0.9909, "step": 8632 }, { "epoch": 0.01, "learning_rate": 4.999101590988268e-05, "loss": 0.8792, "step": 8633 }, { "epoch": 0.01, "learning_rate": 4.999101380415708e-05, "loss": 1.0751, "step": 8634 }, { "epoch": 0.01, "learning_rate": 4.999101169818479e-05, "loss": 1.0353, "step": 8635 }, { "epoch": 0.01, "learning_rate": 4.999100959196579e-05, "loss": 1.1233, "step": 8636 }, { "epoch": 0.01, "learning_rate": 4.999100748550009e-05, "loss": 1.1861, "step": 8637 }, { "epoch": 0.01, "learning_rate": 4.999100537878769e-05, "loss": 1.0026, "step": 8638 }, { "epoch": 0.01, "learning_rate": 4.9991003271828585e-05, "loss": 1.0155, "step": 8639 }, { "epoch": 0.01, "learning_rate": 4.9991001164622784e-05, "loss": 1.0189, "step": 8640 }, { "epoch": 0.01, "learning_rate": 4.9990999057170286e-05, "loss": 1.4193, "step": 8641 }, { "epoch": 0.01, "learning_rate": 4.999099694947108e-05, "loss": 0.6558, "step": 8642 }, { "epoch": 0.01, "learning_rate": 4.999099484152518e-05, "loss": 0.3156, "step": 8643 }, { "epoch": 0.01, "learning_rate": 4.999099273333258e-05, "loss": 0.2008, "step": 8644 }, { "epoch": 0.01, "learning_rate": 4.9990990624893274e-05, "loss": 0.4412, "step": 8645 }, { "epoch": 0.01, "learning_rate": 4.9990988516207265e-05, "loss": 1.0611, "step": 8646 }, { "epoch": 0.01, "learning_rate": 4.999098640727456e-05, "loss": 1.0295, "step": 8647 }, { "epoch": 0.01, "learning_rate": 4.999098429809516e-05, "loss": 1.2576, "step": 8648 }, { "epoch": 0.01, "learning_rate": 4.999098218866905e-05, "loss": 1.1036, "step": 8649 }, { "epoch": 0.01, "learning_rate": 4.999098007899624e-05, "loss": 1.1413, "step": 8650 }, { "epoch": 0.01, "learning_rate": 4.9990977969076734e-05, "loss": 1.3274, "step": 8651 }, { "epoch": 0.01, "learning_rate": 4.9990975858910526e-05, "loss": 0.8323, "step": 8652 }, { "epoch": 0.01, "learning_rate": 4.9990973748497614e-05, "loss": 1.0233, "step": 8653 }, { "epoch": 0.01, "learning_rate": 4.999097163783801e-05, "loss": 0.6758, "step": 8654 }, { "epoch": 0.01, "learning_rate": 4.99909695269317e-05, "loss": 1.018, "step": 8655 }, { "epoch": 0.01, "learning_rate": 4.999096741577869e-05, "loss": 1.0897, "step": 8656 }, { "epoch": 0.01, "learning_rate": 4.9990965304378986e-05, "loss": 0.9251, "step": 8657 }, { "epoch": 0.01, "learning_rate": 4.999096319273258e-05, "loss": 0.9358, "step": 8658 }, { "epoch": 0.01, "learning_rate": 4.999096108083946e-05, "loss": 1.3181, "step": 8659 }, { "epoch": 0.01, "learning_rate": 4.999095896869965e-05, "loss": 1.065, "step": 8660 }, { "epoch": 0.01, "learning_rate": 4.9990956856313144e-05, "loss": 1.2166, "step": 8661 }, { "epoch": 0.01, "learning_rate": 4.999095474367993e-05, "loss": 1.2498, "step": 8662 }, { "epoch": 0.01, "learning_rate": 4.9990952630800023e-05, "loss": 1.0486, "step": 8663 }, { "epoch": 0.01, "learning_rate": 4.999095051767341e-05, "loss": 1.0942, "step": 8664 }, { "epoch": 0.01, "learning_rate": 4.9990948404300095e-05, "loss": 1.3692, "step": 8665 }, { "epoch": 0.01, "learning_rate": 4.9990946290680086e-05, "loss": 1.2716, "step": 8666 }, { "epoch": 0.01, "learning_rate": 4.9990944176813374e-05, "loss": 1.1882, "step": 8667 }, { "epoch": 0.01, "learning_rate": 4.9990942062699965e-05, "loss": 1.1674, "step": 8668 }, { "epoch": 0.01, "learning_rate": 4.999093994833985e-05, "loss": 0.8575, "step": 8669 }, { "epoch": 0.01, "learning_rate": 4.999093783373304e-05, "loss": 0.8663, "step": 8670 }, { "epoch": 0.01, "learning_rate": 4.999093571887953e-05, "loss": 1.1732, "step": 8671 }, { "epoch": 0.01, "learning_rate": 4.999093360377932e-05, "loss": 1.4423, "step": 8672 }, { "epoch": 0.01, "learning_rate": 4.999093148843241e-05, "loss": 1.1841, "step": 8673 }, { "epoch": 0.01, "learning_rate": 4.99909293728388e-05, "loss": 1.1292, "step": 8674 }, { "epoch": 0.01, "learning_rate": 4.999092725699849e-05, "loss": 0.7134, "step": 8675 }, { "epoch": 0.01, "learning_rate": 4.999092514091148e-05, "loss": 1.5599, "step": 8676 }, { "epoch": 0.01, "learning_rate": 4.9990923024577764e-05, "loss": 0.8031, "step": 8677 }, { "epoch": 0.01, "learning_rate": 4.9990920907997355e-05, "loss": 1.0661, "step": 8678 }, { "epoch": 0.01, "learning_rate": 4.999091879117024e-05, "loss": 1.6182, "step": 8679 }, { "epoch": 0.01, "learning_rate": 4.999091667409643e-05, "loss": 1.2834, "step": 8680 }, { "epoch": 0.01, "learning_rate": 4.9990914556775926e-05, "loss": 0.929, "step": 8681 }, { "epoch": 0.01, "learning_rate": 4.999091243920872e-05, "loss": 0.9997, "step": 8682 }, { "epoch": 0.01, "learning_rate": 4.9990910321394804e-05, "loss": 0.9971, "step": 8683 }, { "epoch": 0.01, "learning_rate": 4.9990908203334194e-05, "loss": 1.1674, "step": 8684 }, { "epoch": 0.01, "learning_rate": 4.999090608502688e-05, "loss": 1.2116, "step": 8685 }, { "epoch": 0.01, "learning_rate": 4.999090396647287e-05, "loss": 1.0061, "step": 8686 }, { "epoch": 0.01, "learning_rate": 4.9990901847672165e-05, "loss": 1.3171, "step": 8687 }, { "epoch": 0.01, "learning_rate": 4.9990899728624755e-05, "loss": 1.1013, "step": 8688 }, { "epoch": 0.01, "learning_rate": 4.999089760933065e-05, "loss": 1.5107, "step": 8689 }, { "epoch": 0.01, "learning_rate": 4.999089548978984e-05, "loss": 0.9969, "step": 8690 }, { "epoch": 0.01, "learning_rate": 4.999089337000233e-05, "loss": 1.0103, "step": 8691 }, { "epoch": 0.01, "learning_rate": 4.999089124996813e-05, "loss": 1.3219, "step": 8692 }, { "epoch": 0.01, "learning_rate": 4.999088912968722e-05, "loss": 1.202, "step": 8693 }, { "epoch": 0.01, "learning_rate": 4.999088700915961e-05, "loss": 0.8583, "step": 8694 }, { "epoch": 0.01, "learning_rate": 4.999088488838531e-05, "loss": 0.652, "step": 8695 }, { "epoch": 0.01, "learning_rate": 4.999088276736431e-05, "loss": 0.5651, "step": 8696 }, { "epoch": 0.01, "learning_rate": 4.999088064609661e-05, "loss": 0.5008, "step": 8697 }, { "epoch": 0.01, "learning_rate": 4.99908785245822e-05, "loss": 1.1029, "step": 8698 }, { "epoch": 0.01, "learning_rate": 4.99908764028211e-05, "loss": 1.2576, "step": 8699 }, { "epoch": 0.01, "learning_rate": 4.9990874280813294e-05, "loss": 1.2909, "step": 8700 }, { "epoch": 0.01, "learning_rate": 4.999087215855879e-05, "loss": 1.0101, "step": 8701 }, { "epoch": 0.01, "learning_rate": 4.999087003605759e-05, "loss": 1.3033, "step": 8702 }, { "epoch": 0.01, "learning_rate": 4.999086791330969e-05, "loss": 1.1506, "step": 8703 }, { "epoch": 0.01, "learning_rate": 4.9990865790315086e-05, "loss": 1.0781, "step": 8704 }, { "epoch": 0.01, "learning_rate": 4.999086366707379e-05, "loss": 1.2098, "step": 8705 }, { "epoch": 0.01, "learning_rate": 4.999086154358579e-05, "loss": 1.1135, "step": 8706 }, { "epoch": 0.01, "learning_rate": 4.9990859419851095e-05, "loss": 0.985, "step": 8707 }, { "epoch": 0.01, "learning_rate": 4.99908572958697e-05, "loss": 1.0368, "step": 8708 }, { "epoch": 0.01, "learning_rate": 4.99908551716416e-05, "loss": 1.3571, "step": 8709 }, { "epoch": 0.01, "learning_rate": 4.999085304716681e-05, "loss": 1.4024, "step": 8710 }, { "epoch": 0.01, "learning_rate": 4.9990850922445306e-05, "loss": 1.2478, "step": 8711 }, { "epoch": 0.01, "learning_rate": 4.9990848797477116e-05, "loss": 1.0357, "step": 8712 }, { "epoch": 0.01, "learning_rate": 4.999084667226223e-05, "loss": 1.0545, "step": 8713 }, { "epoch": 0.01, "learning_rate": 4.999084454680063e-05, "loss": 0.9262, "step": 8714 }, { "epoch": 0.01, "learning_rate": 4.9990842421092344e-05, "loss": 0.9385, "step": 8715 }, { "epoch": 0.01, "learning_rate": 4.999084029513735e-05, "loss": 1.1249, "step": 8716 }, { "epoch": 0.01, "learning_rate": 4.999083816893566e-05, "loss": 1.2108, "step": 8717 }, { "epoch": 0.01, "learning_rate": 4.9990836042487275e-05, "loss": 0.9183, "step": 8718 }, { "epoch": 0.01, "learning_rate": 4.999083391579219e-05, "loss": 1.0582, "step": 8719 }, { "epoch": 0.01, "learning_rate": 4.9990831788850404e-05, "loss": 1.0237, "step": 8720 }, { "epoch": 0.01, "learning_rate": 4.9990829661661916e-05, "loss": 1.0782, "step": 8721 }, { "epoch": 0.01, "learning_rate": 4.999082753422673e-05, "loss": 1.018, "step": 8722 }, { "epoch": 0.01, "learning_rate": 4.999082540654485e-05, "loss": 1.084, "step": 8723 }, { "epoch": 0.01, "learning_rate": 4.999082327861627e-05, "loss": 1.151, "step": 8724 }, { "epoch": 0.01, "learning_rate": 4.9990821150440986e-05, "loss": 0.7537, "step": 8725 }, { "epoch": 0.01, "learning_rate": 4.999081902201901e-05, "loss": 1.154, "step": 8726 }, { "epoch": 0.01, "learning_rate": 4.999081689335032e-05, "loss": 1.5061, "step": 8727 }, { "epoch": 0.01, "learning_rate": 4.9990814764434944e-05, "loss": 1.2121, "step": 8728 }, { "epoch": 0.01, "learning_rate": 4.9990812635272876e-05, "loss": 0.9162, "step": 8729 }, { "epoch": 0.01, "learning_rate": 4.99908105058641e-05, "loss": 1.1134, "step": 8730 }, { "epoch": 0.01, "learning_rate": 4.9990808376208624e-05, "loss": 1.0876, "step": 8731 }, { "epoch": 0.01, "learning_rate": 4.999080624630645e-05, "loss": 0.8893, "step": 8732 }, { "epoch": 0.01, "learning_rate": 4.9990804116157586e-05, "loss": 1.0382, "step": 8733 }, { "epoch": 0.01, "learning_rate": 4.9990801985762014e-05, "loss": 1.1064, "step": 8734 }, { "epoch": 0.01, "learning_rate": 4.999079985511974e-05, "loss": 1.2253, "step": 8735 }, { "epoch": 0.01, "learning_rate": 4.9990797724230776e-05, "loss": 1.4358, "step": 8736 }, { "epoch": 0.01, "learning_rate": 4.999079559309511e-05, "loss": 1.6728, "step": 8737 }, { "epoch": 0.01, "learning_rate": 4.999079346171275e-05, "loss": 1.3647, "step": 8738 }, { "epoch": 0.01, "learning_rate": 4.999079133008369e-05, "loss": 1.1633, "step": 8739 }, { "epoch": 0.01, "learning_rate": 4.9990789198207925e-05, "loss": 1.621, "step": 8740 }, { "epoch": 0.01, "learning_rate": 4.9990787066085463e-05, "loss": 1.548, "step": 8741 }, { "epoch": 0.01, "learning_rate": 4.9990784933716306e-05, "loss": 1.3156, "step": 8742 }, { "epoch": 0.01, "learning_rate": 4.9990782801100444e-05, "loss": 1.0951, "step": 8743 }, { "epoch": 0.01, "learning_rate": 4.999078066823789e-05, "loss": 1.5253, "step": 8744 }, { "epoch": 0.01, "learning_rate": 4.999077853512863e-05, "loss": 0.2305, "step": 8745 }, { "epoch": 0.01, "learning_rate": 4.999077640177269e-05, "loss": 0.3264, "step": 8746 }, { "epoch": 0.01, "learning_rate": 4.999077426817004e-05, "loss": 0.3777, "step": 8747 }, { "epoch": 0.01, "learning_rate": 4.999077213432068e-05, "loss": 0.0945, "step": 8748 }, { "epoch": 0.01, "learning_rate": 4.999077000022464e-05, "loss": 0.0586, "step": 8749 }, { "epoch": 0.01, "learning_rate": 4.9990767865881896e-05, "loss": 0.0599, "step": 8750 }, { "epoch": 0.01, "learning_rate": 4.999076573129245e-05, "loss": 1.2068, "step": 8751 }, { "epoch": 0.01, "learning_rate": 4.99907635964563e-05, "loss": 1.1719, "step": 8752 }, { "epoch": 0.01, "learning_rate": 4.999076146137347e-05, "loss": 1.2801, "step": 8753 }, { "epoch": 0.01, "learning_rate": 4.9990759326043924e-05, "loss": 1.1722, "step": 8754 }, { "epoch": 0.01, "learning_rate": 4.999075719046769e-05, "loss": 1.0336, "step": 8755 }, { "epoch": 0.01, "learning_rate": 4.999075505464476e-05, "loss": 1.0087, "step": 8756 }, { "epoch": 0.01, "learning_rate": 4.999075291857512e-05, "loss": 1.2859, "step": 8757 }, { "epoch": 0.01, "learning_rate": 4.9990750782258785e-05, "loss": 1.125, "step": 8758 }, { "epoch": 0.01, "learning_rate": 4.9990748645695764e-05, "loss": 1.0779, "step": 8759 }, { "epoch": 0.01, "learning_rate": 4.999074650888603e-05, "loss": 1.3599, "step": 8760 }, { "epoch": 0.01, "learning_rate": 4.99907443718296e-05, "loss": 1.9964, "step": 8761 }, { "epoch": 0.01, "learning_rate": 4.999074223452648e-05, "loss": 3.0764, "step": 8762 }, { "epoch": 0.01, "learning_rate": 4.999074009697666e-05, "loss": 0.4825, "step": 8763 }, { "epoch": 0.01, "learning_rate": 4.999073795918013e-05, "loss": 0.9809, "step": 8764 }, { "epoch": 0.01, "learning_rate": 4.999073582113692e-05, "loss": 1.3737, "step": 8765 }, { "epoch": 0.01, "learning_rate": 4.9990733682847e-05, "loss": 1.3429, "step": 8766 }, { "epoch": 0.01, "learning_rate": 4.9990731544310386e-05, "loss": 1.1932, "step": 8767 }, { "epoch": 0.01, "learning_rate": 4.9990729405527074e-05, "loss": 1.2749, "step": 8768 }, { "epoch": 0.01, "learning_rate": 4.999072726649706e-05, "loss": 1.0659, "step": 8769 }, { "epoch": 0.01, "learning_rate": 4.9990725127220355e-05, "loss": 1.0079, "step": 8770 }, { "epoch": 0.01, "learning_rate": 4.9990722987696947e-05, "loss": 1.2343, "step": 8771 }, { "epoch": 0.01, "learning_rate": 4.999072084792684e-05, "loss": 0.8325, "step": 8772 }, { "epoch": 0.01, "learning_rate": 4.999071870791005e-05, "loss": 1.0094, "step": 8773 }, { "epoch": 0.01, "learning_rate": 4.999071656764654e-05, "loss": 1.2184, "step": 8774 }, { "epoch": 0.01, "learning_rate": 4.999071442713634e-05, "loss": 0.8084, "step": 8775 }, { "epoch": 0.01, "learning_rate": 4.999071228637945e-05, "loss": 0.5433, "step": 8776 }, { "epoch": 0.01, "learning_rate": 4.9990710145375854e-05, "loss": 0.7831, "step": 8777 }, { "epoch": 0.01, "learning_rate": 4.999070800412556e-05, "loss": 1.2789, "step": 8778 }, { "epoch": 0.01, "learning_rate": 4.9990705862628575e-05, "loss": 1.3226, "step": 8779 }, { "epoch": 0.01, "learning_rate": 4.999070372088489e-05, "loss": 0.8119, "step": 8780 }, { "epoch": 0.01, "learning_rate": 4.999070157889451e-05, "loss": 1.311, "step": 8781 }, { "epoch": 0.01, "learning_rate": 4.9990699436657424e-05, "loss": 1.1202, "step": 8782 }, { "epoch": 0.01, "learning_rate": 4.999069729417364e-05, "loss": 1.1229, "step": 8783 }, { "epoch": 0.01, "learning_rate": 4.999069515144317e-05, "loss": 1.1352, "step": 8784 }, { "epoch": 0.01, "learning_rate": 4.9990693008465996e-05, "loss": 0.9946, "step": 8785 }, { "epoch": 0.01, "learning_rate": 4.999069086524212e-05, "loss": 1.2596, "step": 8786 }, { "epoch": 0.01, "learning_rate": 4.999068872177155e-05, "loss": 1.1081, "step": 8787 }, { "epoch": 0.01, "learning_rate": 4.999068657805428e-05, "loss": 0.9151, "step": 8788 }, { "epoch": 0.01, "learning_rate": 4.9990684434090316e-05, "loss": 0.9901, "step": 8789 }, { "epoch": 0.01, "learning_rate": 4.999068228987965e-05, "loss": 1.2045, "step": 8790 }, { "epoch": 0.01, "learning_rate": 4.99906801454223e-05, "loss": 0.8862, "step": 8791 }, { "epoch": 0.01, "learning_rate": 4.999067800071824e-05, "loss": 1.1898, "step": 8792 }, { "epoch": 0.01, "learning_rate": 4.9990675855767484e-05, "loss": 0.9776, "step": 8793 }, { "epoch": 0.01, "learning_rate": 4.999067371057003e-05, "loss": 1.017, "step": 8794 }, { "epoch": 0.01, "learning_rate": 4.999067156512588e-05, "loss": 1.2088, "step": 8795 }, { "epoch": 0.01, "learning_rate": 4.999066941943503e-05, "loss": 1.3442, "step": 8796 }, { "epoch": 0.01, "learning_rate": 4.999066727349749e-05, "loss": 1.1206, "step": 8797 }, { "epoch": 0.01, "learning_rate": 4.9990665127313246e-05, "loss": 1.4715, "step": 8798 }, { "epoch": 0.01, "learning_rate": 4.9990662980882305e-05, "loss": 1.6074, "step": 8799 }, { "epoch": 0.01, "learning_rate": 4.9990660834204673e-05, "loss": 1.2598, "step": 8800 }, { "epoch": 0.01, "learning_rate": 4.999065868728033e-05, "loss": 1.1772, "step": 8801 }, { "epoch": 0.01, "learning_rate": 4.999065654010931e-05, "loss": 1.1079, "step": 8802 }, { "epoch": 0.01, "learning_rate": 4.9990654392691586e-05, "loss": 0.9349, "step": 8803 }, { "epoch": 0.01, "learning_rate": 4.9990652245027154e-05, "loss": 0.7483, "step": 8804 }, { "epoch": 0.01, "learning_rate": 4.999065009711603e-05, "loss": 0.6774, "step": 8805 }, { "epoch": 0.01, "learning_rate": 4.9990647948958215e-05, "loss": 0.6102, "step": 8806 }, { "epoch": 0.01, "learning_rate": 4.999064580055369e-05, "loss": 0.4901, "step": 8807 }, { "epoch": 0.01, "learning_rate": 4.999064365190248e-05, "loss": 0.6469, "step": 8808 }, { "epoch": 0.01, "learning_rate": 4.999064150300457e-05, "loss": 0.7031, "step": 8809 }, { "epoch": 0.01, "learning_rate": 4.9990639353859956e-05, "loss": 1.2631, "step": 8810 }, { "epoch": 0.01, "learning_rate": 4.9990637204468654e-05, "loss": 0.9765, "step": 8811 }, { "epoch": 0.01, "learning_rate": 4.9990635054830656e-05, "loss": 1.6765, "step": 8812 }, { "epoch": 0.01, "learning_rate": 4.9990632904945955e-05, "loss": 1.2318, "step": 8813 }, { "epoch": 0.01, "learning_rate": 4.9990630754814564e-05, "loss": 1.4459, "step": 8814 }, { "epoch": 0.01, "learning_rate": 4.999062860443646e-05, "loss": 1.3727, "step": 8815 }, { "epoch": 0.01, "learning_rate": 4.999062645381167e-05, "loss": 1.4865, "step": 8816 }, { "epoch": 0.01, "learning_rate": 4.999062430294019e-05, "loss": 1.0549, "step": 8817 }, { "epoch": 0.01, "learning_rate": 4.9990622151822005e-05, "loss": 1.2053, "step": 8818 }, { "epoch": 0.01, "learning_rate": 4.9990620000457123e-05, "loss": 1.2681, "step": 8819 }, { "epoch": 0.01, "learning_rate": 4.9990617848845545e-05, "loss": 1.2183, "step": 8820 }, { "epoch": 0.01, "learning_rate": 4.999061569698727e-05, "loss": 1.1343, "step": 8821 }, { "epoch": 0.01, "learning_rate": 4.99906135448823e-05, "loss": 1.234, "step": 8822 }, { "epoch": 0.01, "learning_rate": 4.999061139253063e-05, "loss": 1.0511, "step": 8823 }, { "epoch": 0.01, "learning_rate": 4.999060923993226e-05, "loss": 1.2056, "step": 8824 }, { "epoch": 0.01, "learning_rate": 4.9990607087087206e-05, "loss": 1.6887, "step": 8825 }, { "epoch": 0.01, "learning_rate": 4.999060493399544e-05, "loss": 1.3319, "step": 8826 }, { "epoch": 0.01, "learning_rate": 4.9990602780656993e-05, "loss": 1.0182, "step": 8827 }, { "epoch": 0.01, "learning_rate": 4.9990600627071835e-05, "loss": 1.0378, "step": 8828 }, { "epoch": 0.01, "learning_rate": 4.9990598473239994e-05, "loss": 1.4682, "step": 8829 }, { "epoch": 0.01, "learning_rate": 4.999059631916144e-05, "loss": 1.3639, "step": 8830 }, { "epoch": 0.01, "learning_rate": 4.99905941648362e-05, "loss": 1.1836, "step": 8831 }, { "epoch": 0.01, "learning_rate": 4.999059201026426e-05, "loss": 1.3155, "step": 8832 }, { "epoch": 0.01, "learning_rate": 4.999058985544562e-05, "loss": 1.128, "step": 8833 }, { "epoch": 0.01, "learning_rate": 4.99905877003803e-05, "loss": 1.2607, "step": 8834 }, { "epoch": 0.01, "learning_rate": 4.999058554506827e-05, "loss": 1.1759, "step": 8835 }, { "epoch": 0.01, "learning_rate": 4.999058338950954e-05, "loss": 0.9747, "step": 8836 }, { "epoch": 0.01, "learning_rate": 4.999058123370412e-05, "loss": 1.1861, "step": 8837 }, { "epoch": 0.01, "learning_rate": 4.9990579077652e-05, "loss": 0.6354, "step": 8838 }, { "epoch": 0.01, "learning_rate": 4.9990576921353187e-05, "loss": 0.6298, "step": 8839 }, { "epoch": 0.01, "learning_rate": 4.9990574764807675e-05, "loss": 0.5622, "step": 8840 }, { "epoch": 0.01, "learning_rate": 4.999057260801547e-05, "loss": 0.3542, "step": 8841 }, { "epoch": 0.01, "learning_rate": 4.9990570450976564e-05, "loss": 0.4076, "step": 8842 }, { "epoch": 0.01, "learning_rate": 4.999056829369096e-05, "loss": 0.5068, "step": 8843 }, { "epoch": 0.01, "learning_rate": 4.9990566136158665e-05, "loss": 0.6871, "step": 8844 }, { "epoch": 0.01, "learning_rate": 4.999056397837968e-05, "loss": 1.1034, "step": 8845 }, { "epoch": 0.01, "learning_rate": 4.9990561820353986e-05, "loss": 1.3705, "step": 8846 }, { "epoch": 0.01, "learning_rate": 4.9990559662081606e-05, "loss": 1.0481, "step": 8847 }, { "epoch": 0.01, "learning_rate": 4.999055750356252e-05, "loss": 1.5051, "step": 8848 }, { "epoch": 0.01, "learning_rate": 4.999055534479674e-05, "loss": 1.2912, "step": 8849 }, { "epoch": 0.01, "learning_rate": 4.999055318578427e-05, "loss": 1.1822, "step": 8850 }, { "epoch": 0.01, "learning_rate": 4.99905510265251e-05, "loss": 1.1203, "step": 8851 }, { "epoch": 0.01, "learning_rate": 4.999054886701923e-05, "loss": 0.8883, "step": 8852 }, { "epoch": 0.01, "learning_rate": 4.999054670726667e-05, "loss": 1.2353, "step": 8853 }, { "epoch": 0.01, "learning_rate": 4.999054454726741e-05, "loss": 0.9927, "step": 8854 }, { "epoch": 0.01, "learning_rate": 4.999054238702145e-05, "loss": 0.6175, "step": 8855 }, { "epoch": 0.01, "learning_rate": 4.99905402265288e-05, "loss": 1.0129, "step": 8856 }, { "epoch": 0.01, "learning_rate": 4.999053806578946e-05, "loss": 1.0805, "step": 8857 }, { "epoch": 0.01, "learning_rate": 4.999053590480342e-05, "loss": 1.1087, "step": 8858 }, { "epoch": 0.01, "learning_rate": 4.999053374357068e-05, "loss": 1.2239, "step": 8859 }, { "epoch": 0.01, "learning_rate": 4.999053158209124e-05, "loss": 1.129, "step": 8860 }, { "epoch": 0.01, "learning_rate": 4.9990529420365106e-05, "loss": 0.9646, "step": 8861 }, { "epoch": 0.01, "learning_rate": 4.999052725839228e-05, "loss": 1.0584, "step": 8862 }, { "epoch": 0.01, "learning_rate": 4.9990525096172755e-05, "loss": 1.085, "step": 8863 }, { "epoch": 0.01, "learning_rate": 4.999052293370654e-05, "loss": 1.0483, "step": 8864 }, { "epoch": 0.01, "learning_rate": 4.9990520770993624e-05, "loss": 1.0508, "step": 8865 }, { "epoch": 0.01, "learning_rate": 4.999051860803401e-05, "loss": 1.2326, "step": 8866 }, { "epoch": 0.01, "learning_rate": 4.9990516444827706e-05, "loss": 1.1117, "step": 8867 }, { "epoch": 0.01, "learning_rate": 4.99905142813747e-05, "loss": 0.9983, "step": 8868 }, { "epoch": 0.01, "learning_rate": 4.9990512117675e-05, "loss": 0.9904, "step": 8869 }, { "epoch": 0.01, "learning_rate": 4.999050995372861e-05, "loss": 1.2398, "step": 8870 }, { "epoch": 0.01, "learning_rate": 4.9990507789535525e-05, "loss": 1.4326, "step": 8871 }, { "epoch": 0.01, "learning_rate": 4.9990505625095735e-05, "loss": 1.0302, "step": 8872 }, { "epoch": 0.01, "learning_rate": 4.9990503460409255e-05, "loss": 1.2675, "step": 8873 }, { "epoch": 0.01, "learning_rate": 4.999050129547608e-05, "loss": 0.8434, "step": 8874 }, { "epoch": 0.01, "learning_rate": 4.9990499130296204e-05, "loss": 0.3367, "step": 8875 }, { "epoch": 0.01, "learning_rate": 4.999049696486964e-05, "loss": 1.0727, "step": 8876 }, { "epoch": 0.01, "learning_rate": 4.999049479919637e-05, "loss": 1.4354, "step": 8877 }, { "epoch": 0.01, "learning_rate": 4.999049263327641e-05, "loss": 1.0382, "step": 8878 }, { "epoch": 0.01, "learning_rate": 4.999049046710975e-05, "loss": 0.8141, "step": 8879 }, { "epoch": 0.01, "learning_rate": 4.99904883006964e-05, "loss": 0.7363, "step": 8880 }, { "epoch": 0.01, "learning_rate": 4.9990486134036354e-05, "loss": 1.6423, "step": 8881 }, { "epoch": 0.01, "learning_rate": 4.999048396712962e-05, "loss": 2.6529, "step": 8882 }, { "epoch": 0.01, "learning_rate": 4.999048179997618e-05, "loss": 1.0003, "step": 8883 }, { "epoch": 0.01, "learning_rate": 4.999047963257604e-05, "loss": 0.9851, "step": 8884 }, { "epoch": 0.01, "learning_rate": 4.9990477464929214e-05, "loss": 0.9985, "step": 8885 }, { "epoch": 0.01, "learning_rate": 4.999047529703569e-05, "loss": 1.1115, "step": 8886 }, { "epoch": 0.01, "learning_rate": 4.999047312889547e-05, "loss": 1.3173, "step": 8887 }, { "epoch": 0.01, "learning_rate": 4.9990470960508555e-05, "loss": 0.7853, "step": 8888 }, { "epoch": 0.01, "learning_rate": 4.999046879187494e-05, "loss": 0.6068, "step": 8889 }, { "epoch": 0.01, "learning_rate": 4.999046662299464e-05, "loss": 0.5536, "step": 8890 }, { "epoch": 0.01, "learning_rate": 4.999046445386763e-05, "loss": 0.5109, "step": 8891 }, { "epoch": 0.01, "learning_rate": 4.999046228449394e-05, "loss": 0.4583, "step": 8892 }, { "epoch": 0.01, "learning_rate": 4.999046011487355e-05, "loss": 0.4123, "step": 8893 }, { "epoch": 0.01, "learning_rate": 4.999045794500646e-05, "loss": 0.4004, "step": 8894 }, { "epoch": 0.01, "learning_rate": 4.999045577489268e-05, "loss": 0.3739, "step": 8895 }, { "epoch": 0.01, "learning_rate": 4.9990453604532205e-05, "loss": 0.3816, "step": 8896 }, { "epoch": 0.01, "learning_rate": 4.999045143392503e-05, "loss": 0.4391, "step": 8897 }, { "epoch": 0.01, "learning_rate": 4.9990449263071156e-05, "loss": 1.1637, "step": 8898 }, { "epoch": 0.01, "learning_rate": 4.99904470919706e-05, "loss": 1.235, "step": 8899 }, { "epoch": 0.01, "learning_rate": 4.9990444920623334e-05, "loss": 0.955, "step": 8900 }, { "epoch": 0.01, "learning_rate": 4.999044274902939e-05, "loss": 0.7153, "step": 8901 }, { "epoch": 0.01, "learning_rate": 4.999044057718873e-05, "loss": 0.9503, "step": 8902 }, { "epoch": 0.01, "learning_rate": 4.9990438405101394e-05, "loss": 1.3083, "step": 8903 }, { "epoch": 0.01, "learning_rate": 4.999043623276736e-05, "loss": 0.7842, "step": 8904 }, { "epoch": 0.01, "learning_rate": 4.999043406018662e-05, "loss": 0.6211, "step": 8905 }, { "epoch": 0.01, "learning_rate": 4.999043188735919e-05, "loss": 0.996, "step": 8906 }, { "epoch": 0.01, "learning_rate": 4.999042971428507e-05, "loss": 0.991, "step": 8907 }, { "epoch": 0.01, "learning_rate": 4.999042754096425e-05, "loss": 1.1821, "step": 8908 }, { "epoch": 0.01, "learning_rate": 4.999042536739673e-05, "loss": 1.0599, "step": 8909 }, { "epoch": 0.01, "learning_rate": 4.999042319358253e-05, "loss": 1.1233, "step": 8910 }, { "epoch": 0.01, "learning_rate": 4.999042101952162e-05, "loss": 1.0006, "step": 8911 }, { "epoch": 0.01, "learning_rate": 4.9990418845214026e-05, "loss": 1.2382, "step": 8912 }, { "epoch": 0.01, "learning_rate": 4.999041667065973e-05, "loss": 1.105, "step": 8913 }, { "epoch": 0.01, "learning_rate": 4.999041449585874e-05, "loss": 1.1125, "step": 8914 }, { "epoch": 0.01, "learning_rate": 4.999041232081106e-05, "loss": 1.1714, "step": 8915 }, { "epoch": 0.01, "learning_rate": 4.999041014551668e-05, "loss": 1.079, "step": 8916 }, { "epoch": 0.01, "learning_rate": 4.9990407969975606e-05, "loss": 0.4372, "step": 8917 }, { "epoch": 0.01, "learning_rate": 4.999040579418784e-05, "loss": 0.5173, "step": 8918 }, { "epoch": 0.01, "learning_rate": 4.999040361815338e-05, "loss": 1.5945, "step": 8919 }, { "epoch": 0.01, "learning_rate": 4.999040144187222e-05, "loss": 1.1579, "step": 8920 }, { "epoch": 0.01, "learning_rate": 4.9990399265344366e-05, "loss": 0.8848, "step": 8921 }, { "epoch": 0.01, "learning_rate": 4.999039708856982e-05, "loss": 1.0272, "step": 8922 }, { "epoch": 0.01, "learning_rate": 4.999039491154858e-05, "loss": 1.1456, "step": 8923 }, { "epoch": 0.01, "learning_rate": 4.9990392734280644e-05, "loss": 0.9839, "step": 8924 }, { "epoch": 0.01, "learning_rate": 4.999039055676601e-05, "loss": 0.9463, "step": 8925 }, { "epoch": 0.01, "learning_rate": 4.999038837900469e-05, "loss": 1.1473, "step": 8926 }, { "epoch": 0.01, "learning_rate": 4.9990386200996666e-05, "loss": 1.6757, "step": 8927 }, { "epoch": 0.01, "learning_rate": 4.999038402274195e-05, "loss": 2.2957, "step": 8928 }, { "epoch": 0.01, "learning_rate": 4.999038184424054e-05, "loss": 0.9495, "step": 8929 }, { "epoch": 0.01, "learning_rate": 4.9990379665492446e-05, "loss": 1.2136, "step": 8930 }, { "epoch": 0.01, "learning_rate": 4.999037748649764e-05, "loss": 1.3899, "step": 8931 }, { "epoch": 0.01, "learning_rate": 4.999037530725615e-05, "loss": 1.2546, "step": 8932 }, { "epoch": 0.01, "learning_rate": 4.999037312776796e-05, "loss": 1.3167, "step": 8933 }, { "epoch": 0.01, "learning_rate": 4.999037094803308e-05, "loss": 1.0479, "step": 8934 }, { "epoch": 0.01, "learning_rate": 4.999036876805151e-05, "loss": 0.8051, "step": 8935 }, { "epoch": 0.01, "learning_rate": 4.9990366587823236e-05, "loss": 0.6917, "step": 8936 }, { "epoch": 0.01, "learning_rate": 4.999036440734827e-05, "loss": 0.6909, "step": 8937 }, { "epoch": 0.01, "learning_rate": 4.999036222662661e-05, "loss": 0.7093, "step": 8938 }, { "epoch": 0.01, "learning_rate": 4.9990360045658266e-05, "loss": 1.1164, "step": 8939 }, { "epoch": 0.01, "learning_rate": 4.999035786444322e-05, "loss": 1.2123, "step": 8940 }, { "epoch": 0.01, "learning_rate": 4.999035568298147e-05, "loss": 1.0801, "step": 8941 }, { "epoch": 0.01, "learning_rate": 4.9990353501273034e-05, "loss": 1.0261, "step": 8942 }, { "epoch": 0.01, "learning_rate": 4.999035131931791e-05, "loss": 0.8396, "step": 8943 }, { "epoch": 0.01, "learning_rate": 4.999034913711608e-05, "loss": 1.0039, "step": 8944 }, { "epoch": 0.01, "learning_rate": 4.9990346954667566e-05, "loss": 1.102, "step": 8945 }, { "epoch": 0.01, "learning_rate": 4.999034477197235e-05, "loss": 0.9566, "step": 8946 }, { "epoch": 0.01, "learning_rate": 4.999034258903045e-05, "loss": 1.3018, "step": 8947 }, { "epoch": 0.01, "learning_rate": 4.999034040584184e-05, "loss": 1.2828, "step": 8948 }, { "epoch": 0.01, "learning_rate": 4.999033822240655e-05, "loss": 1.1951, "step": 8949 }, { "epoch": 0.01, "learning_rate": 4.9990336038724564e-05, "loss": 1.3283, "step": 8950 }, { "epoch": 0.01, "learning_rate": 4.999033385479588e-05, "loss": 1.3253, "step": 8951 }, { "epoch": 0.01, "learning_rate": 4.99903316706205e-05, "loss": 1.4602, "step": 8952 }, { "epoch": 0.01, "learning_rate": 4.999032948619843e-05, "loss": 0.837, "step": 8953 }, { "epoch": 0.01, "learning_rate": 4.999032730152966e-05, "loss": 1.0146, "step": 8954 }, { "epoch": 0.01, "learning_rate": 4.99903251166142e-05, "loss": 1.0879, "step": 8955 }, { "epoch": 0.01, "learning_rate": 4.999032293145205e-05, "loss": 1.4743, "step": 8956 }, { "epoch": 0.01, "learning_rate": 4.9990320746043204e-05, "loss": 1.0642, "step": 8957 }, { "epoch": 0.01, "learning_rate": 4.9990318560387664e-05, "loss": 1.0941, "step": 8958 }, { "epoch": 0.01, "learning_rate": 4.999031637448542e-05, "loss": 0.8432, "step": 8959 }, { "epoch": 0.01, "learning_rate": 4.9990314188336495e-05, "loss": 1.264, "step": 8960 }, { "epoch": 0.01, "learning_rate": 4.999031200194088e-05, "loss": 0.8801, "step": 8961 }, { "epoch": 0.01, "learning_rate": 4.999030981529856e-05, "loss": 0.9694, "step": 8962 }, { "epoch": 0.01, "learning_rate": 4.999030762840955e-05, "loss": 1.2519, "step": 8963 }, { "epoch": 0.01, "learning_rate": 4.9990305441273844e-05, "loss": 1.1655, "step": 8964 }, { "epoch": 0.01, "learning_rate": 4.999030325389145e-05, "loss": 1.7452, "step": 8965 }, { "epoch": 0.01, "learning_rate": 4.9990301066262355e-05, "loss": 1.1885, "step": 8966 }, { "epoch": 0.01, "learning_rate": 4.9990298878386566e-05, "loss": 1.1091, "step": 8967 }, { "epoch": 0.01, "learning_rate": 4.9990296690264086e-05, "loss": 1.0145, "step": 8968 }, { "epoch": 0.01, "learning_rate": 4.999029450189492e-05, "loss": 1.1637, "step": 8969 }, { "epoch": 0.01, "learning_rate": 4.999029231327905e-05, "loss": 0.8908, "step": 8970 }, { "epoch": 0.01, "learning_rate": 4.999029012441649e-05, "loss": 1.0207, "step": 8971 }, { "epoch": 0.01, "learning_rate": 4.999028793530724e-05, "loss": 1.0373, "step": 8972 }, { "epoch": 0.01, "learning_rate": 4.999028574595129e-05, "loss": 1.2048, "step": 8973 }, { "epoch": 0.01, "learning_rate": 4.999028355634865e-05, "loss": 1.6187, "step": 8974 }, { "epoch": 0.01, "learning_rate": 4.9990281366499314e-05, "loss": 1.1053, "step": 8975 }, { "epoch": 0.01, "learning_rate": 4.999027917640329e-05, "loss": 1.4927, "step": 8976 }, { "epoch": 0.01, "learning_rate": 4.999027698606056e-05, "loss": 1.1898, "step": 8977 }, { "epoch": 0.01, "learning_rate": 4.999027479547115e-05, "loss": 1.4062, "step": 8978 }, { "epoch": 0.01, "learning_rate": 4.999027260463503e-05, "loss": 0.9272, "step": 8979 }, { "epoch": 0.01, "learning_rate": 4.9990270413552235e-05, "loss": 0.9686, "step": 8980 }, { "epoch": 0.01, "learning_rate": 4.999026822222274e-05, "loss": 1.0166, "step": 8981 }, { "epoch": 0.01, "learning_rate": 4.9990266030646555e-05, "loss": 1.2849, "step": 8982 }, { "epoch": 0.01, "learning_rate": 4.999026383882367e-05, "loss": 0.7406, "step": 8983 }, { "epoch": 0.01, "learning_rate": 4.99902616467541e-05, "loss": 0.7619, "step": 8984 }, { "epoch": 0.01, "learning_rate": 4.999025945443783e-05, "loss": 1.2153, "step": 8985 }, { "epoch": 0.01, "learning_rate": 4.999025726187486e-05, "loss": 0.8999, "step": 8986 }, { "epoch": 0.01, "learning_rate": 4.9990255069065215e-05, "loss": 0.9631, "step": 8987 }, { "epoch": 0.01, "learning_rate": 4.999025287600886e-05, "loss": 0.8158, "step": 8988 }, { "epoch": 0.01, "learning_rate": 4.999025068270582e-05, "loss": 1.0232, "step": 8989 }, { "epoch": 0.01, "learning_rate": 4.999024848915609e-05, "loss": 1.1381, "step": 8990 }, { "epoch": 0.01, "learning_rate": 4.999024629535966e-05, "loss": 0.9671, "step": 8991 }, { "epoch": 0.01, "learning_rate": 4.999024410131653e-05, "loss": 1.0486, "step": 8992 }, { "epoch": 0.01, "learning_rate": 4.999024190702672e-05, "loss": 1.0093, "step": 8993 }, { "epoch": 0.01, "learning_rate": 4.9990239712490215e-05, "loss": 0.9589, "step": 8994 }, { "epoch": 0.01, "learning_rate": 4.9990237517707014e-05, "loss": 0.9503, "step": 8995 }, { "epoch": 0.01, "learning_rate": 4.999023532267712e-05, "loss": 1.0971, "step": 8996 }, { "epoch": 0.01, "learning_rate": 4.999023312740053e-05, "loss": 1.1159, "step": 8997 }, { "epoch": 0.01, "learning_rate": 4.9990230931877246e-05, "loss": 0.8094, "step": 8998 }, { "epoch": 0.01, "learning_rate": 4.999022873610728e-05, "loss": 1.1743, "step": 8999 }, { "epoch": 0.01, "learning_rate": 4.9990226540090615e-05, "loss": 1.1199, "step": 9000 }, { "epoch": 0.01, "eval_loss": 1.0580137968063354, "eval_runtime": 83.5773, "eval_samples_per_second": 16.571, "eval_steps_per_second": 4.152, "step": 9000 }, { "epoch": 0.01, "learning_rate": 4.9990224343827255e-05, "loss": 0.9495, "step": 9001 }, { "epoch": 0.01, "learning_rate": 4.99902221473172e-05, "loss": 1.0997, "step": 9002 }, { "epoch": 0.01, "learning_rate": 4.999021995056046e-05, "loss": 1.0835, "step": 9003 }, { "epoch": 0.01, "learning_rate": 4.9990217753557015e-05, "loss": 1.1702, "step": 9004 }, { "epoch": 0.01, "learning_rate": 4.999021555630689e-05, "loss": 1.4746, "step": 9005 }, { "epoch": 0.01, "learning_rate": 4.9990213358810065e-05, "loss": 4.1693, "step": 9006 }, { "epoch": 0.01, "learning_rate": 4.9990211161066545e-05, "loss": 1.4417, "step": 9007 }, { "epoch": 0.01, "learning_rate": 4.9990208963076335e-05, "loss": 1.0626, "step": 9008 }, { "epoch": 0.01, "learning_rate": 4.999020676483943e-05, "loss": 1.1296, "step": 9009 }, { "epoch": 0.01, "learning_rate": 4.999020456635583e-05, "loss": 1.0832, "step": 9010 }, { "epoch": 0.01, "learning_rate": 4.999020236762555e-05, "loss": 1.0376, "step": 9011 }, { "epoch": 0.01, "learning_rate": 4.999020016864857e-05, "loss": 1.1048, "step": 9012 }, { "epoch": 0.01, "learning_rate": 4.99901979694249e-05, "loss": 1.0899, "step": 9013 }, { "epoch": 0.01, "learning_rate": 4.999019576995453e-05, "loss": 1.1138, "step": 9014 }, { "epoch": 0.01, "learning_rate": 4.9990193570237476e-05, "loss": 1.126, "step": 9015 }, { "epoch": 0.01, "learning_rate": 4.999019137027372e-05, "loss": 1.1415, "step": 9016 }, { "epoch": 0.01, "learning_rate": 4.999018917006328e-05, "loss": 1.0015, "step": 9017 }, { "epoch": 0.01, "learning_rate": 4.9990186969606146e-05, "loss": 0.7915, "step": 9018 }, { "epoch": 0.01, "learning_rate": 4.9990184768902314e-05, "loss": 0.869, "step": 9019 }, { "epoch": 0.01, "learning_rate": 4.999018256795179e-05, "loss": 0.9655, "step": 9020 }, { "epoch": 0.01, "learning_rate": 4.999018036675458e-05, "loss": 1.0987, "step": 9021 }, { "epoch": 0.01, "learning_rate": 4.999017816531067e-05, "loss": 1.1674, "step": 9022 }, { "epoch": 0.01, "learning_rate": 4.9990175963620066e-05, "loss": 1.2031, "step": 9023 }, { "epoch": 0.01, "learning_rate": 4.999017376168278e-05, "loss": 0.9145, "step": 9024 }, { "epoch": 0.01, "learning_rate": 4.999017155949879e-05, "loss": 0.8055, "step": 9025 }, { "epoch": 0.01, "learning_rate": 4.999016935706812e-05, "loss": 0.8942, "step": 9026 }, { "epoch": 0.01, "learning_rate": 4.9990167154390746e-05, "loss": 1.1422, "step": 9027 }, { "epoch": 0.01, "learning_rate": 4.999016495146669e-05, "loss": 1.146, "step": 9028 }, { "epoch": 0.01, "learning_rate": 4.999016274829593e-05, "loss": 1.1712, "step": 9029 }, { "epoch": 0.01, "learning_rate": 4.9990160544878485e-05, "loss": 1.0368, "step": 9030 }, { "epoch": 0.01, "learning_rate": 4.999015834121435e-05, "loss": 0.8829, "step": 9031 }, { "epoch": 0.01, "learning_rate": 4.999015613730352e-05, "loss": 1.1787, "step": 9032 }, { "epoch": 0.01, "learning_rate": 4.9990153933145995e-05, "loss": 1.5084, "step": 9033 }, { "epoch": 0.01, "learning_rate": 4.999015172874178e-05, "loss": 0.6153, "step": 9034 }, { "epoch": 0.01, "learning_rate": 4.999014952409087e-05, "loss": 1.1142, "step": 9035 }, { "epoch": 0.01, "learning_rate": 4.999014731919327e-05, "loss": 1.2513, "step": 9036 }, { "epoch": 0.01, "learning_rate": 4.999014511404898e-05, "loss": 1.0711, "step": 9037 }, { "epoch": 0.01, "learning_rate": 4.999014290865799e-05, "loss": 0.9331, "step": 9038 }, { "epoch": 0.01, "learning_rate": 4.9990140703020315e-05, "loss": 1.4594, "step": 9039 }, { "epoch": 0.01, "learning_rate": 4.9990138497135954e-05, "loss": 1.106, "step": 9040 }, { "epoch": 0.01, "learning_rate": 4.9990136291004884e-05, "loss": 1.03, "step": 9041 }, { "epoch": 0.01, "learning_rate": 4.999013408462714e-05, "loss": 1.2, "step": 9042 }, { "epoch": 0.01, "learning_rate": 4.999013187800269e-05, "loss": 0.9152, "step": 9043 }, { "epoch": 0.01, "learning_rate": 4.999012967113156e-05, "loss": 0.9454, "step": 9044 }, { "epoch": 0.01, "learning_rate": 4.9990127464013723e-05, "loss": 1.101, "step": 9045 }, { "epoch": 0.01, "learning_rate": 4.9990125256649204e-05, "loss": 1.1816, "step": 9046 }, { "epoch": 0.01, "learning_rate": 4.999012304903799e-05, "loss": 1.1778, "step": 9047 }, { "epoch": 0.01, "learning_rate": 4.999012084118009e-05, "loss": 1.3422, "step": 9048 }, { "epoch": 0.01, "learning_rate": 4.999011863307549e-05, "loss": 1.1377, "step": 9049 }, { "epoch": 0.01, "learning_rate": 4.99901164247242e-05, "loss": 1.1026, "step": 9050 }, { "epoch": 0.01, "learning_rate": 4.9990114216126216e-05, "loss": 1.2882, "step": 9051 }, { "epoch": 0.01, "learning_rate": 4.999011200728155e-05, "loss": 1.2365, "step": 9052 }, { "epoch": 0.01, "learning_rate": 4.999010979819018e-05, "loss": 1.1337, "step": 9053 }, { "epoch": 0.01, "learning_rate": 4.999010758885213e-05, "loss": 0.9899, "step": 9054 }, { "epoch": 0.01, "learning_rate": 4.9990105379267386e-05, "loss": 1.2565, "step": 9055 }, { "epoch": 0.01, "learning_rate": 4.999010316943594e-05, "loss": 1.5078, "step": 9056 }, { "epoch": 0.01, "learning_rate": 4.999010095935781e-05, "loss": 1.4685, "step": 9057 }, { "epoch": 0.01, "learning_rate": 4.9990098749032986e-05, "loss": 1.4006, "step": 9058 }, { "epoch": 0.01, "learning_rate": 4.999009653846147e-05, "loss": 1.1974, "step": 9059 }, { "epoch": 0.01, "learning_rate": 4.999009432764326e-05, "loss": 1.3823, "step": 9060 }, { "epoch": 0.01, "learning_rate": 4.999009211657836e-05, "loss": 0.6458, "step": 9061 }, { "epoch": 0.01, "learning_rate": 4.999008990526677e-05, "loss": 1.0128, "step": 9062 }, { "epoch": 0.01, "learning_rate": 4.999008769370849e-05, "loss": 1.1074, "step": 9063 }, { "epoch": 0.01, "learning_rate": 4.999008548190351e-05, "loss": 1.2405, "step": 9064 }, { "epoch": 0.01, "learning_rate": 4.999008326985185e-05, "loss": 1.6292, "step": 9065 }, { "epoch": 0.01, "learning_rate": 4.999008105755349e-05, "loss": 1.1772, "step": 9066 }, { "epoch": 0.01, "learning_rate": 4.9990078845008444e-05, "loss": 1.118, "step": 9067 }, { "epoch": 0.01, "learning_rate": 4.999007663221671e-05, "loss": 1.3651, "step": 9068 }, { "epoch": 0.01, "learning_rate": 4.999007441917828e-05, "loss": 2.1085, "step": 9069 }, { "epoch": 0.01, "learning_rate": 4.9990072205893155e-05, "loss": 1.1137, "step": 9070 }, { "epoch": 0.01, "learning_rate": 4.999006999236134e-05, "loss": 1.2094, "step": 9071 }, { "epoch": 0.01, "learning_rate": 4.9990067778582837e-05, "loss": 0.8978, "step": 9072 }, { "epoch": 0.01, "learning_rate": 4.999006556455764e-05, "loss": 0.7341, "step": 9073 }, { "epoch": 0.01, "learning_rate": 4.9990063350285745e-05, "loss": 1.0128, "step": 9074 }, { "epoch": 0.01, "learning_rate": 4.999006113576717e-05, "loss": 1.4113, "step": 9075 }, { "epoch": 0.01, "learning_rate": 4.9990058921001895e-05, "loss": 1.2249, "step": 9076 }, { "epoch": 0.01, "learning_rate": 4.9990056705989934e-05, "loss": 1.3765, "step": 9077 }, { "epoch": 0.01, "learning_rate": 4.9990054490731284e-05, "loss": 1.348, "step": 9078 }, { "epoch": 0.01, "learning_rate": 4.999005227522594e-05, "loss": 1.5085, "step": 9079 }, { "epoch": 0.01, "learning_rate": 4.99900500594739e-05, "loss": 1.3146, "step": 9080 }, { "epoch": 0.01, "learning_rate": 4.9990047843475174e-05, "loss": 1.1631, "step": 9081 }, { "epoch": 0.01, "learning_rate": 4.999004562722975e-05, "loss": 1.3994, "step": 9082 }, { "epoch": 0.01, "learning_rate": 4.999004341073764e-05, "loss": 1.1258, "step": 9083 }, { "epoch": 0.01, "learning_rate": 4.999004119399884e-05, "loss": 1.2075, "step": 9084 }, { "epoch": 0.01, "learning_rate": 4.9990038977013356e-05, "loss": 1.3821, "step": 9085 }, { "epoch": 0.01, "learning_rate": 4.999003675978117e-05, "loss": 1.0025, "step": 9086 }, { "epoch": 0.01, "learning_rate": 4.9990034542302294e-05, "loss": 1.1581, "step": 9087 }, { "epoch": 0.01, "learning_rate": 4.999003232457673e-05, "loss": 1.0726, "step": 9088 }, { "epoch": 0.01, "learning_rate": 4.9990030106604466e-05, "loss": 1.0915, "step": 9089 }, { "epoch": 0.01, "learning_rate": 4.9990027888385524e-05, "loss": 1.2369, "step": 9090 }, { "epoch": 0.01, "learning_rate": 4.9990025669919885e-05, "loss": 1.0006, "step": 9091 }, { "epoch": 0.01, "learning_rate": 4.9990023451207557e-05, "loss": 1.1857, "step": 9092 }, { "epoch": 0.01, "learning_rate": 4.999002123224854e-05, "loss": 1.0311, "step": 9093 }, { "epoch": 0.01, "learning_rate": 4.999001901304282e-05, "loss": 1.1941, "step": 9094 }, { "epoch": 0.01, "learning_rate": 4.9990016793590425e-05, "loss": 1.2601, "step": 9095 }, { "epoch": 0.01, "learning_rate": 4.999001457389133e-05, "loss": 1.0129, "step": 9096 }, { "epoch": 0.01, "learning_rate": 4.9990012353945545e-05, "loss": 1.2379, "step": 9097 }, { "epoch": 0.01, "learning_rate": 4.999001013375307e-05, "loss": 1.4153, "step": 9098 }, { "epoch": 0.01, "learning_rate": 4.99900079133139e-05, "loss": 1.368, "step": 9099 }, { "epoch": 0.01, "learning_rate": 4.9990005692628044e-05, "loss": 1.4062, "step": 9100 }, { "epoch": 0.01, "learning_rate": 4.999000347169549e-05, "loss": 0.7764, "step": 9101 }, { "epoch": 0.01, "learning_rate": 4.999000125051626e-05, "loss": 0.9343, "step": 9102 }, { "epoch": 0.01, "learning_rate": 4.998999902909033e-05, "loss": 1.0394, "step": 9103 }, { "epoch": 0.01, "learning_rate": 4.998999680741771e-05, "loss": 0.9598, "step": 9104 }, { "epoch": 0.01, "learning_rate": 4.9989994585498404e-05, "loss": 1.0693, "step": 9105 }, { "epoch": 0.01, "learning_rate": 4.9989992363332396e-05, "loss": 0.4744, "step": 9106 }, { "epoch": 0.01, "learning_rate": 4.9989990140919706e-05, "loss": 0.3535, "step": 9107 }, { "epoch": 0.01, "learning_rate": 4.998998791826032e-05, "loss": 0.3838, "step": 9108 }, { "epoch": 0.01, "learning_rate": 4.998998569535425e-05, "loss": 0.2679, "step": 9109 }, { "epoch": 0.01, "learning_rate": 4.998998347220148e-05, "loss": 0.2629, "step": 9110 }, { "epoch": 0.01, "learning_rate": 4.998998124880203e-05, "loss": 0.2073, "step": 9111 }, { "epoch": 0.01, "learning_rate": 4.9989979025155886e-05, "loss": 0.7131, "step": 9112 }, { "epoch": 0.01, "learning_rate": 4.998997680126305e-05, "loss": 0.7333, "step": 9113 }, { "epoch": 0.01, "learning_rate": 4.998997457712352e-05, "loss": 1.3434, "step": 9114 }, { "epoch": 0.01, "learning_rate": 4.99899723527373e-05, "loss": 1.5403, "step": 9115 }, { "epoch": 0.01, "learning_rate": 4.9989970128104394e-05, "loss": 1.2184, "step": 9116 }, { "epoch": 0.01, "learning_rate": 4.99899679032248e-05, "loss": 1.1485, "step": 9117 }, { "epoch": 0.01, "learning_rate": 4.998996567809851e-05, "loss": 1.2869, "step": 9118 }, { "epoch": 0.01, "learning_rate": 4.998996345272553e-05, "loss": 1.7247, "step": 9119 }, { "epoch": 0.01, "learning_rate": 4.9989961227105865e-05, "loss": 1.1431, "step": 9120 }, { "epoch": 0.01, "learning_rate": 4.99899590012395e-05, "loss": 1.1697, "step": 9121 }, { "epoch": 0.01, "learning_rate": 4.998995677512646e-05, "loss": 1.182, "step": 9122 }, { "epoch": 0.01, "learning_rate": 4.9989954548766715e-05, "loss": 1.1294, "step": 9123 }, { "epoch": 0.01, "learning_rate": 4.9989952322160284e-05, "loss": 1.0839, "step": 9124 }, { "epoch": 0.01, "learning_rate": 4.998995009530716e-05, "loss": 0.7609, "step": 9125 }, { "epoch": 0.01, "learning_rate": 4.998994786820736e-05, "loss": 1.1856, "step": 9126 }, { "epoch": 0.01, "learning_rate": 4.9989945640860856e-05, "loss": 1.1002, "step": 9127 }, { "epoch": 0.01, "learning_rate": 4.998994341326766e-05, "loss": 0.9823, "step": 9128 }, { "epoch": 0.01, "learning_rate": 4.9989941185427777e-05, "loss": 1.3698, "step": 9129 }, { "epoch": 0.01, "learning_rate": 4.998993895734121e-05, "loss": 0.9, "step": 9130 }, { "epoch": 0.01, "learning_rate": 4.9989936729007945e-05, "loss": 1.2307, "step": 9131 }, { "epoch": 0.01, "learning_rate": 4.9989934500427994e-05, "loss": 1.1759, "step": 9132 }, { "epoch": 0.01, "learning_rate": 4.998993227160135e-05, "loss": 0.8295, "step": 9133 }, { "epoch": 0.01, "learning_rate": 4.998993004252802e-05, "loss": 1.1786, "step": 9134 }, { "epoch": 0.01, "learning_rate": 4.9989927813207995e-05, "loss": 1.3618, "step": 9135 }, { "epoch": 0.01, "learning_rate": 4.9989925583641285e-05, "loss": 1.3421, "step": 9136 }, { "epoch": 0.01, "learning_rate": 4.9989923353827885e-05, "loss": 1.1763, "step": 9137 }, { "epoch": 0.01, "learning_rate": 4.998992112376779e-05, "loss": 1.0261, "step": 9138 }, { "epoch": 0.01, "learning_rate": 4.998991889346101e-05, "loss": 1.1373, "step": 9139 }, { "epoch": 0.01, "learning_rate": 4.998991666290754e-05, "loss": 1.1484, "step": 9140 }, { "epoch": 0.01, "learning_rate": 4.998991443210738e-05, "loss": 1.1353, "step": 9141 }, { "epoch": 0.01, "learning_rate": 4.998991220106052e-05, "loss": 1.166, "step": 9142 }, { "epoch": 0.01, "learning_rate": 4.9989909969766976e-05, "loss": 1.2454, "step": 9143 }, { "epoch": 0.01, "learning_rate": 4.998990773822675e-05, "loss": 1.2827, "step": 9144 }, { "epoch": 0.01, "learning_rate": 4.998990550643983e-05, "loss": 1.0495, "step": 9145 }, { "epoch": 0.01, "learning_rate": 4.998990327440621e-05, "loss": 1.1319, "step": 9146 }, { "epoch": 0.01, "learning_rate": 4.998990104212591e-05, "loss": 1.2523, "step": 9147 }, { "epoch": 0.01, "learning_rate": 4.9989898809598924e-05, "loss": 1.0989, "step": 9148 }, { "epoch": 0.01, "learning_rate": 4.9989896576825246e-05, "loss": 1.2735, "step": 9149 }, { "epoch": 0.01, "learning_rate": 4.998989434380487e-05, "loss": 1.3394, "step": 9150 }, { "epoch": 0.01, "learning_rate": 4.998989211053782e-05, "loss": 1.1166, "step": 9151 }, { "epoch": 0.01, "learning_rate": 4.9989889877024064e-05, "loss": 1.2377, "step": 9152 }, { "epoch": 0.01, "learning_rate": 4.9989887643263625e-05, "loss": 1.1499, "step": 9153 }, { "epoch": 0.01, "learning_rate": 4.99898854092565e-05, "loss": 1.1686, "step": 9154 }, { "epoch": 0.01, "learning_rate": 4.998988317500267e-05, "loss": 1.0717, "step": 9155 }, { "epoch": 0.01, "learning_rate": 4.9989880940502165e-05, "loss": 1.1107, "step": 9156 }, { "epoch": 0.01, "learning_rate": 4.998987870575497e-05, "loss": 1.4478, "step": 9157 }, { "epoch": 0.01, "learning_rate": 4.998987647076108e-05, "loss": 1.3044, "step": 9158 }, { "epoch": 0.01, "learning_rate": 4.998987423552051e-05, "loss": 1.1052, "step": 9159 }, { "epoch": 0.01, "learning_rate": 4.998987200003324e-05, "loss": 0.9653, "step": 9160 }, { "epoch": 0.01, "learning_rate": 4.998986976429928e-05, "loss": 1.0159, "step": 9161 }, { "epoch": 0.01, "learning_rate": 4.998986752831864e-05, "loss": 1.042, "step": 9162 }, { "epoch": 0.01, "learning_rate": 4.99898652920913e-05, "loss": 1.109, "step": 9163 }, { "epoch": 0.01, "learning_rate": 4.998986305561728e-05, "loss": 1.3906, "step": 9164 }, { "epoch": 0.01, "learning_rate": 4.9989860818896564e-05, "loss": 1.334, "step": 9165 }, { "epoch": 0.01, "learning_rate": 4.9989858581929164e-05, "loss": 1.0409, "step": 9166 }, { "epoch": 0.01, "learning_rate": 4.998985634471507e-05, "loss": 0.8628, "step": 9167 }, { "epoch": 0.01, "learning_rate": 4.998985410725429e-05, "loss": 0.5957, "step": 9168 }, { "epoch": 0.01, "learning_rate": 4.998985186954682e-05, "loss": 1.0405, "step": 9169 }, { "epoch": 0.01, "learning_rate": 4.998984963159265e-05, "loss": 1.1866, "step": 9170 }, { "epoch": 0.01, "learning_rate": 4.998984739339181e-05, "loss": 1.0571, "step": 9171 }, { "epoch": 0.01, "learning_rate": 4.998984515494427e-05, "loss": 1.0295, "step": 9172 }, { "epoch": 0.01, "learning_rate": 4.998984291625004e-05, "loss": 1.2393, "step": 9173 }, { "epoch": 0.01, "learning_rate": 4.998984067730912e-05, "loss": 0.9602, "step": 9174 }, { "epoch": 0.01, "learning_rate": 4.9989838438121515e-05, "loss": 1.2002, "step": 9175 }, { "epoch": 0.01, "learning_rate": 4.998983619868722e-05, "loss": 1.4526, "step": 9176 }, { "epoch": 0.01, "learning_rate": 4.998983395900624e-05, "loss": 1.1412, "step": 9177 }, { "epoch": 0.01, "learning_rate": 4.998983171907856e-05, "loss": 1.3989, "step": 9178 }, { "epoch": 0.01, "learning_rate": 4.99898294789042e-05, "loss": 1.0866, "step": 9179 }, { "epoch": 0.01, "learning_rate": 4.9989827238483144e-05, "loss": 1.22, "step": 9180 }, { "epoch": 0.01, "learning_rate": 4.99898249978154e-05, "loss": 1.0868, "step": 9181 }, { "epoch": 0.01, "learning_rate": 4.998982275690097e-05, "loss": 0.6667, "step": 9182 }, { "epoch": 0.01, "learning_rate": 4.998982051573986e-05, "loss": 0.258, "step": 9183 }, { "epoch": 0.01, "learning_rate": 4.9989818274332047e-05, "loss": 0.5565, "step": 9184 }, { "epoch": 0.01, "learning_rate": 4.998981603267755e-05, "loss": 1.094, "step": 9185 }, { "epoch": 0.01, "learning_rate": 4.998981379077636e-05, "loss": 1.0107, "step": 9186 }, { "epoch": 0.01, "learning_rate": 4.998981154862849e-05, "loss": 0.566, "step": 9187 }, { "epoch": 0.01, "learning_rate": 4.9989809306233925e-05, "loss": 1.1785, "step": 9188 }, { "epoch": 0.01, "learning_rate": 4.998980706359268e-05, "loss": 0.5826, "step": 9189 }, { "epoch": 0.01, "learning_rate": 4.9989804820704735e-05, "loss": 0.9039, "step": 9190 }, { "epoch": 0.01, "learning_rate": 4.99898025775701e-05, "loss": 1.2496, "step": 9191 }, { "epoch": 0.01, "learning_rate": 4.9989800334188786e-05, "loss": 1.0865, "step": 9192 }, { "epoch": 0.01, "learning_rate": 4.998979809056078e-05, "loss": 0.7945, "step": 9193 }, { "epoch": 0.01, "learning_rate": 4.9989795846686084e-05, "loss": 0.9995, "step": 9194 }, { "epoch": 0.01, "learning_rate": 4.998979360256469e-05, "loss": 1.2338, "step": 9195 }, { "epoch": 0.01, "learning_rate": 4.998979135819662e-05, "loss": 1.2236, "step": 9196 }, { "epoch": 0.01, "learning_rate": 4.998978911358186e-05, "loss": 1.3668, "step": 9197 }, { "epoch": 0.01, "learning_rate": 4.998978686872041e-05, "loss": 1.1045, "step": 9198 }, { "epoch": 0.01, "learning_rate": 4.998978462361227e-05, "loss": 1.1917, "step": 9199 }, { "epoch": 0.01, "learning_rate": 4.998978237825744e-05, "loss": 1.1984, "step": 9200 }, { "epoch": 0.01, "learning_rate": 4.9989780132655916e-05, "loss": 1.2166, "step": 9201 }, { "epoch": 0.01, "learning_rate": 4.9989777886807714e-05, "loss": 1.0184, "step": 9202 }, { "epoch": 0.01, "learning_rate": 4.998977564071282e-05, "loss": 1.1177, "step": 9203 }, { "epoch": 0.01, "learning_rate": 4.9989773394371235e-05, "loss": 1.1258, "step": 9204 }, { "epoch": 0.01, "learning_rate": 4.9989771147782964e-05, "loss": 1.5137, "step": 9205 }, { "epoch": 0.01, "learning_rate": 4.998976890094801e-05, "loss": 1.0125, "step": 9206 }, { "epoch": 0.01, "learning_rate": 4.998976665386635e-05, "loss": 1.0696, "step": 9207 }, { "epoch": 0.01, "learning_rate": 4.998976440653802e-05, "loss": 1.1876, "step": 9208 }, { "epoch": 0.01, "learning_rate": 4.9989762158962995e-05, "loss": 1.1406, "step": 9209 }, { "epoch": 0.01, "learning_rate": 4.998975991114128e-05, "loss": 0.9748, "step": 9210 }, { "epoch": 0.01, "learning_rate": 4.998975766307288e-05, "loss": 0.9761, "step": 9211 }, { "epoch": 0.01, "learning_rate": 4.998975541475779e-05, "loss": 1.0726, "step": 9212 }, { "epoch": 0.01, "learning_rate": 4.998975316619601e-05, "loss": 0.8541, "step": 9213 }, { "epoch": 0.01, "learning_rate": 4.998975091738755e-05, "loss": 1.1607, "step": 9214 }, { "epoch": 0.01, "learning_rate": 4.9989748668332385e-05, "loss": 1.1494, "step": 9215 }, { "epoch": 0.01, "learning_rate": 4.9989746419030546e-05, "loss": 1.4115, "step": 9216 }, { "epoch": 0.01, "learning_rate": 4.998974416948201e-05, "loss": 1.1737, "step": 9217 }, { "epoch": 0.01, "learning_rate": 4.99897419196868e-05, "loss": 1.1807, "step": 9218 }, { "epoch": 0.01, "learning_rate": 4.998973966964489e-05, "loss": 1.0111, "step": 9219 }, { "epoch": 0.01, "learning_rate": 4.998973741935628e-05, "loss": 0.8739, "step": 9220 }, { "epoch": 0.01, "learning_rate": 4.998973516882101e-05, "loss": 0.6346, "step": 9221 }, { "epoch": 0.01, "learning_rate": 4.998973291803903e-05, "loss": 1.0811, "step": 9222 }, { "epoch": 0.01, "learning_rate": 4.998973066701037e-05, "loss": 0.8113, "step": 9223 }, { "epoch": 0.01, "learning_rate": 4.998972841573502e-05, "loss": 1.1547, "step": 9224 }, { "epoch": 0.01, "learning_rate": 4.998972616421298e-05, "loss": 1.2342, "step": 9225 }, { "epoch": 0.01, "learning_rate": 4.9989723912444257e-05, "loss": 1.4094, "step": 9226 }, { "epoch": 0.01, "learning_rate": 4.998972166042884e-05, "loss": 1.1467, "step": 9227 }, { "epoch": 0.01, "learning_rate": 4.9989719408166745e-05, "loss": 0.7494, "step": 9228 }, { "epoch": 0.01, "learning_rate": 4.998971715565796e-05, "loss": 1.2098, "step": 9229 }, { "epoch": 0.01, "learning_rate": 4.998971490290247e-05, "loss": 1.2357, "step": 9230 }, { "epoch": 0.01, "learning_rate": 4.998971264990031e-05, "loss": 1.1795, "step": 9231 }, { "epoch": 0.01, "learning_rate": 4.9989710396651456e-05, "loss": 0.8123, "step": 9232 }, { "epoch": 0.01, "learning_rate": 4.998970814315592e-05, "loss": 0.8651, "step": 9233 }, { "epoch": 0.01, "learning_rate": 4.998970588941369e-05, "loss": 1.2199, "step": 9234 }, { "epoch": 0.01, "learning_rate": 4.998970363542477e-05, "loss": 0.5233, "step": 9235 }, { "epoch": 0.01, "learning_rate": 4.998970138118917e-05, "loss": 1.163, "step": 9236 }, { "epoch": 0.01, "learning_rate": 4.998969912670688e-05, "loss": 1.7315, "step": 9237 }, { "epoch": 0.01, "learning_rate": 4.99896968719779e-05, "loss": 1.5246, "step": 9238 }, { "epoch": 0.01, "learning_rate": 4.998969461700223e-05, "loss": 1.2901, "step": 9239 }, { "epoch": 0.01, "learning_rate": 4.998969236177987e-05, "loss": 1.0149, "step": 9240 }, { "epoch": 0.01, "learning_rate": 4.998969010631083e-05, "loss": 0.9027, "step": 9241 }, { "epoch": 0.01, "learning_rate": 4.9989687850595105e-05, "loss": 1.275, "step": 9242 }, { "epoch": 0.01, "learning_rate": 4.9989685594632686e-05, "loss": 1.1306, "step": 9243 }, { "epoch": 0.01, "learning_rate": 4.998968333842358e-05, "loss": 1.2844, "step": 9244 }, { "epoch": 0.01, "learning_rate": 4.998968108196778e-05, "loss": 1.2132, "step": 9245 }, { "epoch": 0.01, "learning_rate": 4.99896788252653e-05, "loss": 1.1061, "step": 9246 }, { "epoch": 0.01, "learning_rate": 4.9989676568316134e-05, "loss": 0.9193, "step": 9247 }, { "epoch": 0.01, "learning_rate": 4.998967431112027e-05, "loss": 1.0906, "step": 9248 }, { "epoch": 0.01, "learning_rate": 4.998967205367773e-05, "loss": 0.9511, "step": 9249 }, { "epoch": 0.01, "learning_rate": 4.99896697959885e-05, "loss": 1.1308, "step": 9250 }, { "epoch": 0.01, "learning_rate": 4.998966753805258e-05, "loss": 0.7206, "step": 9251 }, { "epoch": 0.01, "learning_rate": 4.9989665279869977e-05, "loss": 1.072, "step": 9252 }, { "epoch": 0.01, "learning_rate": 4.998966302144068e-05, "loss": 1.1586, "step": 9253 }, { "epoch": 0.01, "learning_rate": 4.99896607627647e-05, "loss": 1.266, "step": 9254 }, { "epoch": 0.01, "learning_rate": 4.998965850384203e-05, "loss": 1.3535, "step": 9255 }, { "epoch": 0.01, "learning_rate": 4.998965624467267e-05, "loss": 1.1896, "step": 9256 }, { "epoch": 0.01, "learning_rate": 4.998965398525663e-05, "loss": 0.433, "step": 9257 }, { "epoch": 0.01, "learning_rate": 4.99896517255939e-05, "loss": 0.373, "step": 9258 }, { "epoch": 0.01, "learning_rate": 4.9989649465684485e-05, "loss": 0.321, "step": 9259 }, { "epoch": 0.01, "learning_rate": 4.998964720552838e-05, "loss": 0.4569, "step": 9260 }, { "epoch": 0.01, "learning_rate": 4.998964494512558e-05, "loss": 0.3219, "step": 9261 }, { "epoch": 0.01, "learning_rate": 4.99896426844761e-05, "loss": 0.9249, "step": 9262 }, { "epoch": 0.01, "learning_rate": 4.9989640423579936e-05, "loss": 0.8442, "step": 9263 }, { "epoch": 0.01, "learning_rate": 4.998963816243708e-05, "loss": 1.4251, "step": 9264 }, { "epoch": 0.01, "learning_rate": 4.9989635901047536e-05, "loss": 1.3881, "step": 9265 }, { "epoch": 0.01, "learning_rate": 4.998963363941131e-05, "loss": 1.1606, "step": 9266 }, { "epoch": 0.01, "learning_rate": 4.99896313775284e-05, "loss": 1.3775, "step": 9267 }, { "epoch": 0.01, "learning_rate": 4.998962911539879e-05, "loss": 1.3913, "step": 9268 }, { "epoch": 0.01, "learning_rate": 4.9989626853022505e-05, "loss": 1.0168, "step": 9269 }, { "epoch": 0.01, "learning_rate": 4.998962459039953e-05, "loss": 1.1377, "step": 9270 }, { "epoch": 0.01, "learning_rate": 4.998962232752986e-05, "loss": 1.2716, "step": 9271 }, { "epoch": 0.01, "learning_rate": 4.998962006441351e-05, "loss": 1.3371, "step": 9272 }, { "epoch": 0.01, "learning_rate": 4.998961780105047e-05, "loss": 1.0409, "step": 9273 }, { "epoch": 0.01, "learning_rate": 4.9989615537440746e-05, "loss": 1.2988, "step": 9274 }, { "epoch": 0.01, "learning_rate": 4.998961327358433e-05, "loss": 0.8929, "step": 9275 }, { "epoch": 0.01, "learning_rate": 4.9989611009481236e-05, "loss": 1.1388, "step": 9276 }, { "epoch": 0.01, "learning_rate": 4.998960874513145e-05, "loss": 0.7059, "step": 9277 }, { "epoch": 0.01, "learning_rate": 4.998960648053498e-05, "loss": 0.6566, "step": 9278 }, { "epoch": 0.01, "learning_rate": 4.998960421569182e-05, "loss": 0.6692, "step": 9279 }, { "epoch": 0.01, "learning_rate": 4.998960195060197e-05, "loss": 1.1011, "step": 9280 }, { "epoch": 0.01, "learning_rate": 4.998959968526544e-05, "loss": 0.6837, "step": 9281 }, { "epoch": 0.01, "learning_rate": 4.998959741968222e-05, "loss": 1.2752, "step": 9282 }, { "epoch": 0.01, "learning_rate": 4.9989595153852306e-05, "loss": 1.2245, "step": 9283 }, { "epoch": 0.01, "learning_rate": 4.998959288777572e-05, "loss": 1.2041, "step": 9284 }, { "epoch": 0.01, "learning_rate": 4.998959062145244e-05, "loss": 1.539, "step": 9285 }, { "epoch": 0.01, "learning_rate": 4.9989588354882465e-05, "loss": 0.8351, "step": 9286 }, { "epoch": 0.01, "learning_rate": 4.9989586088065814e-05, "loss": 1.0497, "step": 9287 }, { "epoch": 0.01, "learning_rate": 4.998958382100247e-05, "loss": 1.0906, "step": 9288 }, { "epoch": 0.01, "learning_rate": 4.998958155369244e-05, "loss": 0.7888, "step": 9289 }, { "epoch": 0.01, "learning_rate": 4.9989579286135735e-05, "loss": 0.9287, "step": 9290 }, { "epoch": 0.01, "learning_rate": 4.998957701833233e-05, "loss": 1.2553, "step": 9291 }, { "epoch": 0.01, "learning_rate": 4.998957475028225e-05, "loss": 1.2213, "step": 9292 }, { "epoch": 0.01, "learning_rate": 4.998957248198547e-05, "loss": 1.154, "step": 9293 }, { "epoch": 0.01, "learning_rate": 4.9989570213442014e-05, "loss": 0.9525, "step": 9294 }, { "epoch": 0.01, "learning_rate": 4.9989567944651864e-05, "loss": 1.2354, "step": 9295 }, { "epoch": 0.01, "learning_rate": 4.998956567561504e-05, "loss": 1.3407, "step": 9296 }, { "epoch": 0.01, "learning_rate": 4.9989563406331516e-05, "loss": 1.2901, "step": 9297 }, { "epoch": 0.01, "learning_rate": 4.998956113680131e-05, "loss": 1.5461, "step": 9298 }, { "epoch": 0.01, "learning_rate": 4.9989558867024414e-05, "loss": 0.7708, "step": 9299 }, { "epoch": 0.01, "learning_rate": 4.9989556597000835e-05, "loss": 1.1066, "step": 9300 }, { "epoch": 0.01, "learning_rate": 4.998955432673057e-05, "loss": 1.3445, "step": 9301 }, { "epoch": 0.01, "learning_rate": 4.9989552056213615e-05, "loss": 1.076, "step": 9302 }, { "epoch": 0.01, "learning_rate": 4.9989549785449986e-05, "loss": 1.2831, "step": 9303 }, { "epoch": 0.01, "learning_rate": 4.998954751443966e-05, "loss": 1.116, "step": 9304 }, { "epoch": 0.01, "learning_rate": 4.998954524318265e-05, "loss": 0.6092, "step": 9305 }, { "epoch": 0.01, "learning_rate": 4.998954297167895e-05, "loss": 1.1854, "step": 9306 }, { "epoch": 0.01, "learning_rate": 4.9989540699928575e-05, "loss": 1.0404, "step": 9307 }, { "epoch": 0.01, "learning_rate": 4.9989538427931504e-05, "loss": 1.1735, "step": 9308 }, { "epoch": 0.01, "learning_rate": 4.9989536155687744e-05, "loss": 1.2108, "step": 9309 }, { "epoch": 0.01, "learning_rate": 4.998953388319731e-05, "loss": 1.1294, "step": 9310 }, { "epoch": 0.01, "learning_rate": 4.998953161046017e-05, "loss": 1.0185, "step": 9311 }, { "epoch": 0.01, "learning_rate": 4.998952933747636e-05, "loss": 0.8502, "step": 9312 }, { "epoch": 0.01, "learning_rate": 4.9989527064245864e-05, "loss": 0.98, "step": 9313 }, { "epoch": 0.01, "learning_rate": 4.9989524790768674e-05, "loss": 1.2549, "step": 9314 }, { "epoch": 0.01, "learning_rate": 4.998952251704481e-05, "loss": 1.2172, "step": 9315 }, { "epoch": 0.01, "learning_rate": 4.998952024307425e-05, "loss": 0.9761, "step": 9316 }, { "epoch": 0.01, "learning_rate": 4.9989517968857e-05, "loss": 1.0534, "step": 9317 }, { "epoch": 0.01, "learning_rate": 4.998951569439308e-05, "loss": 1.0157, "step": 9318 }, { "epoch": 0.01, "learning_rate": 4.998951341968245e-05, "loss": 4.9269, "step": 9319 }, { "epoch": 0.01, "learning_rate": 4.998951114472516e-05, "loss": 4.9151, "step": 9320 }, { "epoch": 0.01, "learning_rate": 4.9989508869521165e-05, "loss": 1.274, "step": 9321 }, { "epoch": 0.01, "learning_rate": 4.998950659407049e-05, "loss": 0.8477, "step": 9322 }, { "epoch": 0.01, "learning_rate": 4.998950431837314e-05, "loss": 0.8863, "step": 9323 }, { "epoch": 0.01, "learning_rate": 4.9989502042429085e-05, "loss": 1.1951, "step": 9324 }, { "epoch": 0.01, "learning_rate": 4.9989499766238354e-05, "loss": 1.0261, "step": 9325 }, { "epoch": 0.01, "learning_rate": 4.998949748980094e-05, "loss": 1.0344, "step": 9326 }, { "epoch": 0.01, "learning_rate": 4.9989495213116836e-05, "loss": 1.8301, "step": 9327 }, { "epoch": 0.01, "learning_rate": 4.998949293618605e-05, "loss": 1.6782, "step": 9328 }, { "epoch": 0.01, "learning_rate": 4.998949065900857e-05, "loss": 1.6676, "step": 9329 }, { "epoch": 0.01, "learning_rate": 4.998948838158441e-05, "loss": 1.5011, "step": 9330 }, { "epoch": 0.01, "learning_rate": 4.998948610391357e-05, "loss": 1.6872, "step": 9331 }, { "epoch": 0.01, "learning_rate": 4.998948382599604e-05, "loss": 1.1545, "step": 9332 }, { "epoch": 0.01, "learning_rate": 4.998948154783182e-05, "loss": 1.0702, "step": 9333 }, { "epoch": 0.01, "learning_rate": 4.9989479269420915e-05, "loss": 1.3011, "step": 9334 }, { "epoch": 0.01, "learning_rate": 4.9989476990763326e-05, "loss": 0.8697, "step": 9335 }, { "epoch": 0.01, "learning_rate": 4.9989474711859054e-05, "loss": 0.5981, "step": 9336 }, { "epoch": 0.01, "learning_rate": 4.998947243270809e-05, "loss": 0.7164, "step": 9337 }, { "epoch": 0.01, "learning_rate": 4.998947015331045e-05, "loss": 1.2006, "step": 9338 }, { "epoch": 0.01, "learning_rate": 4.998946787366612e-05, "loss": 1.3245, "step": 9339 }, { "epoch": 0.01, "learning_rate": 4.99894655937751e-05, "loss": 1.077, "step": 9340 }, { "epoch": 0.01, "learning_rate": 4.99894633136374e-05, "loss": 1.1189, "step": 9341 }, { "epoch": 0.01, "learning_rate": 4.9989461033253016e-05, "loss": 1.1796, "step": 9342 }, { "epoch": 0.01, "learning_rate": 4.998945875262194e-05, "loss": 1.222, "step": 9343 }, { "epoch": 0.01, "learning_rate": 4.9989456471744184e-05, "loss": 1.2311, "step": 9344 }, { "epoch": 0.01, "learning_rate": 4.9989454190619744e-05, "loss": 1.1154, "step": 9345 }, { "epoch": 0.01, "learning_rate": 4.9989451909248614e-05, "loss": 0.9449, "step": 9346 }, { "epoch": 0.01, "learning_rate": 4.99894496276308e-05, "loss": 1.1974, "step": 9347 }, { "epoch": 0.01, "learning_rate": 4.99894473457663e-05, "loss": 1.0124, "step": 9348 }, { "epoch": 0.01, "learning_rate": 4.998944506365512e-05, "loss": 1.0151, "step": 9349 }, { "epoch": 0.01, "learning_rate": 4.998944278129725e-05, "loss": 0.7458, "step": 9350 }, { "epoch": 0.01, "learning_rate": 4.998944049869269e-05, "loss": 0.9251, "step": 9351 }, { "epoch": 0.01, "learning_rate": 4.998943821584145e-05, "loss": 1.1292, "step": 9352 }, { "epoch": 0.01, "learning_rate": 4.998943593274352e-05, "loss": 1.2313, "step": 9353 }, { "epoch": 0.01, "learning_rate": 4.9989433649398913e-05, "loss": 0.7314, "step": 9354 }, { "epoch": 0.01, "learning_rate": 4.9989431365807615e-05, "loss": 1.3168, "step": 9355 }, { "epoch": 0.01, "learning_rate": 4.9989429081969634e-05, "loss": 0.7289, "step": 9356 }, { "epoch": 0.01, "learning_rate": 4.998942679788497e-05, "loss": 1.2072, "step": 9357 }, { "epoch": 0.01, "learning_rate": 4.9989424513553615e-05, "loss": 1.1549, "step": 9358 }, { "epoch": 0.01, "learning_rate": 4.998942222897558e-05, "loss": 1.0921, "step": 9359 }, { "epoch": 0.01, "learning_rate": 4.998941994415086e-05, "loss": 0.9914, "step": 9360 }, { "epoch": 0.01, "learning_rate": 4.998941765907945e-05, "loss": 1.1754, "step": 9361 }, { "epoch": 0.01, "learning_rate": 4.998941537376136e-05, "loss": 1.2133, "step": 9362 }, { "epoch": 0.01, "learning_rate": 4.998941308819659e-05, "loss": 1.4551, "step": 9363 }, { "epoch": 0.01, "learning_rate": 4.998941080238513e-05, "loss": 1.2509, "step": 9364 }, { "epoch": 0.01, "learning_rate": 4.998940851632698e-05, "loss": 1.1247, "step": 9365 }, { "epoch": 0.01, "learning_rate": 4.9989406230022154e-05, "loss": 1.0529, "step": 9366 }, { "epoch": 0.01, "learning_rate": 4.998940394347064e-05, "loss": 1.0516, "step": 9367 }, { "epoch": 0.01, "learning_rate": 4.998940165667243e-05, "loss": 0.8707, "step": 9368 }, { "epoch": 0.01, "learning_rate": 4.998939936962754e-05, "loss": 1.4619, "step": 9369 }, { "epoch": 0.01, "learning_rate": 4.998939708233598e-05, "loss": 1.1269, "step": 9370 }, { "epoch": 0.01, "learning_rate": 4.9989394794797716e-05, "loss": 1.0765, "step": 9371 }, { "epoch": 0.01, "learning_rate": 4.998939250701278e-05, "loss": 1.1936, "step": 9372 }, { "epoch": 0.01, "learning_rate": 4.9989390218981157e-05, "loss": 1.3124, "step": 9373 }, { "epoch": 0.01, "learning_rate": 4.9989387930702845e-05, "loss": 1.4301, "step": 9374 }, { "epoch": 0.01, "learning_rate": 4.998938564217786e-05, "loss": 0.4058, "step": 9375 }, { "epoch": 0.01, "learning_rate": 4.9989383353406174e-05, "loss": 0.9463, "step": 9376 }, { "epoch": 0.01, "learning_rate": 4.998938106438781e-05, "loss": 1.3724, "step": 9377 }, { "epoch": 0.01, "learning_rate": 4.998937877512276e-05, "loss": 0.7495, "step": 9378 }, { "epoch": 0.01, "learning_rate": 4.998937648561103e-05, "loss": 1.0482, "step": 9379 }, { "epoch": 0.01, "learning_rate": 4.998937419585261e-05, "loss": 1.1149, "step": 9380 }, { "epoch": 0.01, "learning_rate": 4.998937190584751e-05, "loss": 1.0348, "step": 9381 }, { "epoch": 0.01, "learning_rate": 4.9989369615595724e-05, "loss": 0.7258, "step": 9382 }, { "epoch": 0.01, "learning_rate": 4.998936732509725e-05, "loss": 0.2154, "step": 9383 }, { "epoch": 0.01, "learning_rate": 4.9989365034352096e-05, "loss": 0.3616, "step": 9384 }, { "epoch": 0.01, "learning_rate": 4.998936274336026e-05, "loss": 1.0633, "step": 9385 }, { "epoch": 0.01, "learning_rate": 4.998936045212173e-05, "loss": 1.0139, "step": 9386 }, { "epoch": 0.01, "learning_rate": 4.9989358160636523e-05, "loss": 1.0686, "step": 9387 }, { "epoch": 0.01, "learning_rate": 4.998935586890463e-05, "loss": 1.367, "step": 9388 }, { "epoch": 0.01, "learning_rate": 4.998935357692605e-05, "loss": 0.723, "step": 9389 }, { "epoch": 0.01, "learning_rate": 4.998935128470079e-05, "loss": 0.5576, "step": 9390 }, { "epoch": 0.01, "learning_rate": 4.9989348992228845e-05, "loss": 0.7907, "step": 9391 }, { "epoch": 0.01, "learning_rate": 4.998934669951021e-05, "loss": 1.1041, "step": 9392 }, { "epoch": 0.01, "learning_rate": 4.9989344406544894e-05, "loss": 1.063, "step": 9393 }, { "epoch": 0.01, "learning_rate": 4.9989342113332894e-05, "loss": 1.1615, "step": 9394 }, { "epoch": 0.01, "learning_rate": 4.998933981987422e-05, "loss": 1.0552, "step": 9395 }, { "epoch": 0.01, "learning_rate": 4.9989337526168844e-05, "loss": 1.0791, "step": 9396 }, { "epoch": 0.01, "learning_rate": 4.9989335232216795e-05, "loss": 1.0922, "step": 9397 }, { "epoch": 0.01, "learning_rate": 4.9989332938018055e-05, "loss": 1.0177, "step": 9398 }, { "epoch": 0.01, "learning_rate": 4.998933064357263e-05, "loss": 1.3498, "step": 9399 }, { "epoch": 0.01, "learning_rate": 4.9989328348880534e-05, "loss": 1.2368, "step": 9400 }, { "epoch": 0.01, "learning_rate": 4.9989326053941745e-05, "loss": 0.96, "step": 9401 }, { "epoch": 0.01, "learning_rate": 4.998932375875627e-05, "loss": 0.6169, "step": 9402 }, { "epoch": 0.01, "learning_rate": 4.998932146332411e-05, "loss": 0.5365, "step": 9403 }, { "epoch": 0.01, "learning_rate": 4.9989319167645274e-05, "loss": 0.4449, "step": 9404 }, { "epoch": 0.01, "learning_rate": 4.9989316871719746e-05, "loss": 0.9199, "step": 9405 }, { "epoch": 0.01, "learning_rate": 4.998931457554754e-05, "loss": 0.6908, "step": 9406 }, { "epoch": 0.01, "learning_rate": 4.998931227912864e-05, "loss": 0.4082, "step": 9407 }, { "epoch": 0.01, "learning_rate": 4.998930998246307e-05, "loss": 0.6992, "step": 9408 }, { "epoch": 0.01, "learning_rate": 4.998930768555081e-05, "loss": 0.5228, "step": 9409 }, { "epoch": 0.01, "learning_rate": 4.998930538839186e-05, "loss": 1.0521, "step": 9410 }, { "epoch": 0.01, "learning_rate": 4.9989303090986234e-05, "loss": 0.9689, "step": 9411 }, { "epoch": 0.01, "learning_rate": 4.9989300793333924e-05, "loss": 0.9596, "step": 9412 }, { "epoch": 0.01, "learning_rate": 4.9989298495434925e-05, "loss": 0.6542, "step": 9413 }, { "epoch": 0.01, "learning_rate": 4.998929619728925e-05, "loss": 1.0697, "step": 9414 }, { "epoch": 0.01, "learning_rate": 4.998929389889688e-05, "loss": 1.5153, "step": 9415 }, { "epoch": 0.01, "learning_rate": 4.9989291600257834e-05, "loss": 1.2395, "step": 9416 }, { "epoch": 0.01, "learning_rate": 4.99892893013721e-05, "loss": 1.1297, "step": 9417 }, { "epoch": 0.01, "learning_rate": 4.998928700223968e-05, "loss": 1.3725, "step": 9418 }, { "epoch": 0.01, "learning_rate": 4.998928470286058e-05, "loss": 1.166, "step": 9419 }, { "epoch": 0.01, "learning_rate": 4.99892824032348e-05, "loss": 1.2635, "step": 9420 }, { "epoch": 0.01, "learning_rate": 4.998928010336234e-05, "loss": 1.3999, "step": 9421 }, { "epoch": 0.01, "learning_rate": 4.998927780324318e-05, "loss": 1.1487, "step": 9422 }, { "epoch": 0.01, "learning_rate": 4.998927550287735e-05, "loss": 1.0197, "step": 9423 }, { "epoch": 0.01, "learning_rate": 4.998927320226483e-05, "loss": 1.1196, "step": 9424 }, { "epoch": 0.01, "learning_rate": 4.9989270901405636e-05, "loss": 1.0885, "step": 9425 }, { "epoch": 0.01, "learning_rate": 4.9989268600299756e-05, "loss": 1.2092, "step": 9426 }, { "epoch": 0.01, "learning_rate": 4.9989266298947187e-05, "loss": 1.1147, "step": 9427 }, { "epoch": 0.01, "learning_rate": 4.9989263997347934e-05, "loss": 0.9987, "step": 9428 }, { "epoch": 0.01, "learning_rate": 4.9989261695502e-05, "loss": 0.9774, "step": 9429 }, { "epoch": 0.01, "learning_rate": 4.998925939340938e-05, "loss": 1.0896, "step": 9430 }, { "epoch": 0.01, "learning_rate": 4.998925709107008e-05, "loss": 1.327, "step": 9431 }, { "epoch": 0.01, "learning_rate": 4.99892547884841e-05, "loss": 1.1995, "step": 9432 }, { "epoch": 0.01, "learning_rate": 4.998925248565143e-05, "loss": 1.1151, "step": 9433 }, { "epoch": 0.01, "learning_rate": 4.998925018257207e-05, "loss": 0.2965, "step": 9434 }, { "epoch": 0.01, "learning_rate": 4.998924787924604e-05, "loss": 1.1415, "step": 9435 }, { "epoch": 0.01, "learning_rate": 4.9989245575673314e-05, "loss": 1.1174, "step": 9436 }, { "epoch": 0.01, "learning_rate": 4.998924327185391e-05, "loss": 0.9524, "step": 9437 }, { "epoch": 0.01, "learning_rate": 4.998924096778783e-05, "loss": 1.2086, "step": 9438 }, { "epoch": 0.01, "learning_rate": 4.998923866347506e-05, "loss": 1.1131, "step": 9439 }, { "epoch": 0.01, "learning_rate": 4.998923635891561e-05, "loss": 1.2125, "step": 9440 }, { "epoch": 0.01, "learning_rate": 4.998923405410948e-05, "loss": 1.3067, "step": 9441 }, { "epoch": 0.01, "learning_rate": 4.998923174905666e-05, "loss": 1.2938, "step": 9442 }, { "epoch": 0.01, "learning_rate": 4.998922944375716e-05, "loss": 1.057, "step": 9443 }, { "epoch": 0.01, "learning_rate": 4.998922713821097e-05, "loss": 1.1474, "step": 9444 }, { "epoch": 0.01, "learning_rate": 4.99892248324181e-05, "loss": 1.0418, "step": 9445 }, { "epoch": 0.01, "learning_rate": 4.9989222526378555e-05, "loss": 0.9933, "step": 9446 }, { "epoch": 0.01, "learning_rate": 4.998922022009232e-05, "loss": 1.1512, "step": 9447 }, { "epoch": 0.01, "learning_rate": 4.998921791355941e-05, "loss": 0.9912, "step": 9448 }, { "epoch": 0.01, "learning_rate": 4.9989215606779804e-05, "loss": 1.1426, "step": 9449 }, { "epoch": 0.01, "learning_rate": 4.998921329975352e-05, "loss": 1.3306, "step": 9450 }, { "epoch": 0.01, "learning_rate": 4.998921099248056e-05, "loss": 1.353, "step": 9451 }, { "epoch": 0.01, "learning_rate": 4.998920868496091e-05, "loss": 1.0916, "step": 9452 }, { "epoch": 0.01, "learning_rate": 4.9989206377194575e-05, "loss": 1.0523, "step": 9453 }, { "epoch": 0.01, "learning_rate": 4.998920406918156e-05, "loss": 1.2159, "step": 9454 }, { "epoch": 0.01, "learning_rate": 4.9989201760921865e-05, "loss": 1.2316, "step": 9455 }, { "epoch": 0.01, "learning_rate": 4.998919945241548e-05, "loss": 1.054, "step": 9456 }, { "epoch": 0.01, "learning_rate": 4.9989197143662416e-05, "loss": 1.178, "step": 9457 }, { "epoch": 0.01, "learning_rate": 4.998919483466267e-05, "loss": 0.9162, "step": 9458 }, { "epoch": 0.01, "learning_rate": 4.998919252541624e-05, "loss": 1.32, "step": 9459 }, { "epoch": 0.01, "learning_rate": 4.998919021592313e-05, "loss": 0.9844, "step": 9460 }, { "epoch": 0.01, "learning_rate": 4.9989187906183335e-05, "loss": 1.0082, "step": 9461 }, { "epoch": 0.01, "learning_rate": 4.9989185596196864e-05, "loss": 1.0474, "step": 9462 }, { "epoch": 0.01, "learning_rate": 4.99891832859637e-05, "loss": 1.077, "step": 9463 }, { "epoch": 0.01, "learning_rate": 4.9989180975483865e-05, "loss": 1.0237, "step": 9464 }, { "epoch": 0.01, "learning_rate": 4.998917866475733e-05, "loss": 1.1396, "step": 9465 }, { "epoch": 0.01, "learning_rate": 4.998917635378413e-05, "loss": 1.1195, "step": 9466 }, { "epoch": 0.01, "learning_rate": 4.9989174042564234e-05, "loss": 1.1072, "step": 9467 }, { "epoch": 0.01, "learning_rate": 4.9989171731097664e-05, "loss": 1.8758, "step": 9468 }, { "epoch": 0.01, "learning_rate": 4.998916941938441e-05, "loss": 1.7659, "step": 9469 }, { "epoch": 0.01, "learning_rate": 4.9989167107424475e-05, "loss": 1.2814, "step": 9470 }, { "epoch": 0.01, "learning_rate": 4.998916479521785e-05, "loss": 1.4703, "step": 9471 }, { "epoch": 0.01, "learning_rate": 4.9989162482764546e-05, "loss": 1.1845, "step": 9472 }, { "epoch": 0.01, "learning_rate": 4.998916017006456e-05, "loss": 1.1917, "step": 9473 }, { "epoch": 0.01, "learning_rate": 4.998915785711789e-05, "loss": 1.2706, "step": 9474 }, { "epoch": 0.01, "learning_rate": 4.9989155543924534e-05, "loss": 1.229, "step": 9475 }, { "epoch": 0.01, "learning_rate": 4.9989153230484506e-05, "loss": 0.9546, "step": 9476 }, { "epoch": 0.01, "learning_rate": 4.998915091679779e-05, "loss": 0.955, "step": 9477 }, { "epoch": 0.01, "learning_rate": 4.9989148602864394e-05, "loss": 1.3085, "step": 9478 }, { "epoch": 0.01, "learning_rate": 4.998914628868432e-05, "loss": 1.2113, "step": 9479 }, { "epoch": 0.01, "learning_rate": 4.998914397425755e-05, "loss": 0.991, "step": 9480 }, { "epoch": 0.01, "learning_rate": 4.9989141659584106e-05, "loss": 1.1268, "step": 9481 }, { "epoch": 0.01, "learning_rate": 4.998913934466398e-05, "loss": 1.2348, "step": 9482 }, { "epoch": 0.01, "learning_rate": 4.998913702949717e-05, "loss": 1.2291, "step": 9483 }, { "epoch": 0.01, "learning_rate": 4.998913471408368e-05, "loss": 1.0154, "step": 9484 }, { "epoch": 0.01, "learning_rate": 4.998913239842351e-05, "loss": 1.1843, "step": 9485 }, { "epoch": 0.01, "learning_rate": 4.998913008251665e-05, "loss": 1.2066, "step": 9486 }, { "epoch": 0.01, "learning_rate": 4.998912776636311e-05, "loss": 1.1451, "step": 9487 }, { "epoch": 0.01, "learning_rate": 4.998912544996289e-05, "loss": 0.9843, "step": 9488 }, { "epoch": 0.01, "learning_rate": 4.998912313331599e-05, "loss": 0.9165, "step": 9489 }, { "epoch": 0.01, "learning_rate": 4.998912081642241e-05, "loss": 1.1365, "step": 9490 }, { "epoch": 0.01, "learning_rate": 4.998911849928214e-05, "loss": 1.0598, "step": 9491 }, { "epoch": 0.01, "learning_rate": 4.998911618189519e-05, "loss": 1.1168, "step": 9492 }, { "epoch": 0.01, "learning_rate": 4.998911386426156e-05, "loss": 1.021, "step": 9493 }, { "epoch": 0.01, "learning_rate": 4.998911154638125e-05, "loss": 1.054, "step": 9494 }, { "epoch": 0.01, "learning_rate": 4.998910922825425e-05, "loss": 0.9778, "step": 9495 }, { "epoch": 0.01, "learning_rate": 4.998910690988058e-05, "loss": 1.0052, "step": 9496 }, { "epoch": 0.01, "learning_rate": 4.998910459126022e-05, "loss": 1.1659, "step": 9497 }, { "epoch": 0.01, "learning_rate": 4.9989102272393176e-05, "loss": 1.3724, "step": 9498 }, { "epoch": 0.01, "learning_rate": 4.998909995327946e-05, "loss": 2.5844, "step": 9499 }, { "epoch": 0.01, "learning_rate": 4.9989097633919055e-05, "loss": 1.0749, "step": 9500 }, { "epoch": 0.01, "eval_loss": 1.0519160032272339, "eval_runtime": 86.0341, "eval_samples_per_second": 16.098, "eval_steps_per_second": 4.033, "step": 9500 }, { "epoch": 0.01, "learning_rate": 4.998909531431197e-05, "loss": 1.1633, "step": 9501 }, { "epoch": 0.01, "learning_rate": 4.9989092994458195e-05, "loss": 0.8895, "step": 9502 }, { "epoch": 0.01, "learning_rate": 4.998909067435775e-05, "loss": 1.1799, "step": 9503 }, { "epoch": 0.01, "learning_rate": 4.9989088354010616e-05, "loss": 0.8809, "step": 9504 }, { "epoch": 0.01, "learning_rate": 4.9989086033416806e-05, "loss": 0.9955, "step": 9505 }, { "epoch": 0.01, "learning_rate": 4.998908371257631e-05, "loss": 1.0974, "step": 9506 }, { "epoch": 0.01, "learning_rate": 4.9989081391489135e-05, "loss": 1.0556, "step": 9507 }, { "epoch": 0.01, "learning_rate": 4.9989079070155275e-05, "loss": 0.9914, "step": 9508 }, { "epoch": 0.01, "learning_rate": 4.998907674857474e-05, "loss": 1.2065, "step": 9509 }, { "epoch": 0.01, "learning_rate": 4.998907442674752e-05, "loss": 1.0535, "step": 9510 }, { "epoch": 0.01, "learning_rate": 4.998907210467361e-05, "loss": 1.067, "step": 9511 }, { "epoch": 0.01, "learning_rate": 4.9989069782353024e-05, "loss": 1.2242, "step": 9512 }, { "epoch": 0.01, "learning_rate": 4.998906745978576e-05, "loss": 1.039, "step": 9513 }, { "epoch": 0.01, "learning_rate": 4.998906513697181e-05, "loss": 0.7001, "step": 9514 }, { "epoch": 0.01, "learning_rate": 4.998906281391118e-05, "loss": 0.5576, "step": 9515 }, { "epoch": 0.01, "learning_rate": 4.998906049060387e-05, "loss": 0.5156, "step": 9516 }, { "epoch": 0.01, "learning_rate": 4.9989058167049884e-05, "loss": 0.303, "step": 9517 }, { "epoch": 0.01, "learning_rate": 4.998905584324921e-05, "loss": 0.194, "step": 9518 }, { "epoch": 0.01, "learning_rate": 4.998905351920186e-05, "loss": 1.6149, "step": 9519 }, { "epoch": 0.01, "learning_rate": 4.9989051194907815e-05, "loss": 5.9522, "step": 9520 }, { "epoch": 0.01, "learning_rate": 4.9989048870367096e-05, "loss": 5.3922, "step": 9521 }, { "epoch": 0.01, "learning_rate": 4.99890465455797e-05, "loss": 5.8952, "step": 9522 }, { "epoch": 0.01, "learning_rate": 4.998904422054562e-05, "loss": 5.8887, "step": 9523 }, { "epoch": 0.01, "learning_rate": 4.998904189526486e-05, "loss": 5.8202, "step": 9524 }, { "epoch": 0.01, "learning_rate": 4.998903956973741e-05, "loss": 5.7676, "step": 9525 }, { "epoch": 0.01, "learning_rate": 4.998903724396329e-05, "loss": 5.8913, "step": 9526 }, { "epoch": 0.01, "learning_rate": 4.998903491794248e-05, "loss": 5.7219, "step": 9527 }, { "epoch": 0.01, "learning_rate": 4.9989032591675e-05, "loss": 5.3908, "step": 9528 }, { "epoch": 0.01, "learning_rate": 4.998903026516083e-05, "loss": 5.8612, "step": 9529 }, { "epoch": 0.01, "learning_rate": 4.998902793839998e-05, "loss": 5.416, "step": 9530 }, { "epoch": 0.01, "learning_rate": 4.998902561139245e-05, "loss": 5.6954, "step": 9531 }, { "epoch": 0.01, "learning_rate": 4.9989023284138234e-05, "loss": 5.797, "step": 9532 }, { "epoch": 0.01, "learning_rate": 4.9989020956637344e-05, "loss": 5.6869, "step": 9533 }, { "epoch": 0.01, "learning_rate": 4.998901862888977e-05, "loss": 5.8472, "step": 9534 }, { "epoch": 0.01, "learning_rate": 4.9989016300895514e-05, "loss": 5.7293, "step": 9535 }, { "epoch": 0.01, "learning_rate": 4.9989013972654575e-05, "loss": 5.6829, "step": 9536 }, { "epoch": 0.01, "learning_rate": 4.998901164416696e-05, "loss": 5.8825, "step": 9537 }, { "epoch": 0.01, "learning_rate": 4.998900931543267e-05, "loss": 5.7073, "step": 9538 }, { "epoch": 0.01, "learning_rate": 4.9989006986451685e-05, "loss": 5.5586, "step": 9539 }, { "epoch": 0.01, "learning_rate": 4.998900465722403e-05, "loss": 5.7517, "step": 9540 }, { "epoch": 0.01, "learning_rate": 4.9989002327749686e-05, "loss": 4.9932, "step": 9541 }, { "epoch": 0.01, "learning_rate": 4.998899999802866e-05, "loss": 5.8724, "step": 9542 }, { "epoch": 0.01, "learning_rate": 4.998899766806096e-05, "loss": 5.8379, "step": 9543 }, { "epoch": 0.01, "learning_rate": 4.998899533784658e-05, "loss": 5.7381, "step": 9544 }, { "epoch": 0.01, "learning_rate": 4.998899300738551e-05, "loss": 5.697, "step": 9545 }, { "epoch": 0.01, "learning_rate": 4.9988990676677774e-05, "loss": 5.7577, "step": 9546 }, { "epoch": 0.01, "learning_rate": 4.998898834572334e-05, "loss": 5.6929, "step": 9547 }, { "epoch": 0.01, "learning_rate": 4.998898601452223e-05, "loss": 5.6557, "step": 9548 }, { "epoch": 0.01, "learning_rate": 4.9988983683074445e-05, "loss": 5.7832, "step": 9549 }, { "epoch": 0.01, "learning_rate": 4.998898135137998e-05, "loss": 5.6743, "step": 9550 }, { "epoch": 0.01, "learning_rate": 4.998897901943883e-05, "loss": 5.754, "step": 9551 }, { "epoch": 0.01, "learning_rate": 4.9988976687250996e-05, "loss": 5.1066, "step": 9552 }, { "epoch": 0.01, "learning_rate": 4.998897435481649e-05, "loss": 1.4271, "step": 9553 }, { "epoch": 0.01, "learning_rate": 4.9988972022135304e-05, "loss": 1.3299, "step": 9554 }, { "epoch": 0.01, "learning_rate": 4.998896968920743e-05, "loss": 0.9807, "step": 9555 }, { "epoch": 0.01, "learning_rate": 4.998896735603288e-05, "loss": 1.043, "step": 9556 }, { "epoch": 0.01, "learning_rate": 4.998896502261164e-05, "loss": 0.8069, "step": 9557 }, { "epoch": 0.01, "learning_rate": 4.998896268894373e-05, "loss": 1.1061, "step": 9558 }, { "epoch": 0.01, "learning_rate": 4.998896035502914e-05, "loss": 1.2951, "step": 9559 }, { "epoch": 0.01, "learning_rate": 4.998895802086787e-05, "loss": 0.7709, "step": 9560 }, { "epoch": 0.01, "learning_rate": 4.998895568645992e-05, "loss": 0.327, "step": 9561 }, { "epoch": 0.01, "learning_rate": 4.998895335180528e-05, "loss": 0.6259, "step": 9562 }, { "epoch": 0.01, "learning_rate": 4.998895101690396e-05, "loss": 1.0665, "step": 9563 }, { "epoch": 0.01, "learning_rate": 4.9988948681755965e-05, "loss": 1.0852, "step": 9564 }, { "epoch": 0.01, "learning_rate": 4.998894634636129e-05, "loss": 0.8451, "step": 9565 }, { "epoch": 0.01, "learning_rate": 4.998894401071993e-05, "loss": 1.0385, "step": 9566 }, { "epoch": 0.01, "learning_rate": 4.99889416748319e-05, "loss": 1.1364, "step": 9567 }, { "epoch": 0.01, "learning_rate": 4.998893933869718e-05, "loss": 0.9963, "step": 9568 }, { "epoch": 0.01, "learning_rate": 4.998893700231578e-05, "loss": 1.2045, "step": 9569 }, { "epoch": 0.01, "learning_rate": 4.9988934665687704e-05, "loss": 1.1008, "step": 9570 }, { "epoch": 0.01, "learning_rate": 4.9988932328812945e-05, "loss": 1.0951, "step": 9571 }, { "epoch": 0.01, "learning_rate": 4.998892999169151e-05, "loss": 1.003, "step": 9572 }, { "epoch": 0.01, "learning_rate": 4.998892765432339e-05, "loss": 1.1747, "step": 9573 }, { "epoch": 0.01, "learning_rate": 4.998892531670859e-05, "loss": 1.0342, "step": 9574 }, { "epoch": 0.01, "learning_rate": 4.9988922978847115e-05, "loss": 0.7845, "step": 9575 }, { "epoch": 0.01, "learning_rate": 4.9988920640738954e-05, "loss": 1.0553, "step": 9576 }, { "epoch": 0.01, "learning_rate": 4.9988918302384116e-05, "loss": 0.9495, "step": 9577 }, { "epoch": 0.01, "learning_rate": 4.99889159637826e-05, "loss": 1.1584, "step": 9578 }, { "epoch": 0.01, "learning_rate": 4.99889136249344e-05, "loss": 0.8282, "step": 9579 }, { "epoch": 0.01, "learning_rate": 4.9988911285839523e-05, "loss": 1.1773, "step": 9580 }, { "epoch": 0.01, "learning_rate": 4.998890894649796e-05, "loss": 1.0359, "step": 9581 }, { "epoch": 0.01, "learning_rate": 4.998890660690973e-05, "loss": 1.1669, "step": 9582 }, { "epoch": 0.01, "learning_rate": 4.9988904267074804e-05, "loss": 0.765, "step": 9583 }, { "epoch": 0.01, "learning_rate": 4.9988901926993204e-05, "loss": 1.1385, "step": 9584 }, { "epoch": 0.01, "learning_rate": 4.998889958666493e-05, "loss": 1.0772, "step": 9585 }, { "epoch": 0.01, "learning_rate": 4.998889724608997e-05, "loss": 1.0362, "step": 9586 }, { "epoch": 0.01, "learning_rate": 4.998889490526833e-05, "loss": 1.168, "step": 9587 }, { "epoch": 0.01, "learning_rate": 4.998889256420001e-05, "loss": 1.2161, "step": 9588 }, { "epoch": 0.01, "learning_rate": 4.998889022288501e-05, "loss": 1.0458, "step": 9589 }, { "epoch": 0.01, "learning_rate": 4.998888788132333e-05, "loss": 1.1914, "step": 9590 }, { "epoch": 0.01, "learning_rate": 4.998888553951498e-05, "loss": 1.0189, "step": 9591 }, { "epoch": 0.01, "learning_rate": 4.998888319745994e-05, "loss": 0.2751, "step": 9592 }, { "epoch": 0.01, "learning_rate": 4.9988880855158226e-05, "loss": 0.9459, "step": 9593 }, { "epoch": 0.01, "learning_rate": 4.998887851260983e-05, "loss": 1.2398, "step": 9594 }, { "epoch": 0.01, "learning_rate": 4.998887616981475e-05, "loss": 1.1093, "step": 9595 }, { "epoch": 0.01, "learning_rate": 4.9988873826772994e-05, "loss": 1.1416, "step": 9596 }, { "epoch": 0.01, "learning_rate": 4.9988871483484555e-05, "loss": 0.9976, "step": 9597 }, { "epoch": 0.01, "learning_rate": 4.998886913994945e-05, "loss": 0.247, "step": 9598 }, { "epoch": 0.01, "learning_rate": 4.998886679616765e-05, "loss": 0.4209, "step": 9599 }, { "epoch": 0.01, "learning_rate": 4.9988864452139174e-05, "loss": 1.2544, "step": 9600 }, { "epoch": 0.01, "learning_rate": 4.998886210786402e-05, "loss": 0.9261, "step": 9601 }, { "epoch": 0.01, "learning_rate": 4.998885976334219e-05, "loss": 1.3198, "step": 9602 }, { "epoch": 0.01, "learning_rate": 4.998885741857368e-05, "loss": 1.3817, "step": 9603 }, { "epoch": 0.01, "learning_rate": 4.9988855073558484e-05, "loss": 0.9208, "step": 9604 }, { "epoch": 0.01, "learning_rate": 4.998885272829661e-05, "loss": 1.2108, "step": 9605 }, { "epoch": 0.01, "learning_rate": 4.9988850382788055e-05, "loss": 1.4153, "step": 9606 }, { "epoch": 0.01, "learning_rate": 4.9988848037032825e-05, "loss": 0.9594, "step": 9607 }, { "epoch": 0.01, "learning_rate": 4.998884569103092e-05, "loss": 1.1309, "step": 9608 }, { "epoch": 0.01, "learning_rate": 4.9988843344782324e-05, "loss": 1.1749, "step": 9609 }, { "epoch": 0.01, "learning_rate": 4.998884099828706e-05, "loss": 0.9763, "step": 9610 }, { "epoch": 0.01, "learning_rate": 4.998883865154511e-05, "loss": 1.2261, "step": 9611 }, { "epoch": 0.01, "learning_rate": 4.9988836304556487e-05, "loss": 1.0687, "step": 9612 }, { "epoch": 0.01, "learning_rate": 4.998883395732118e-05, "loss": 1.0324, "step": 9613 }, { "epoch": 0.01, "learning_rate": 4.9988831609839195e-05, "loss": 1.3793, "step": 9614 }, { "epoch": 0.01, "learning_rate": 4.998882926211053e-05, "loss": 0.9648, "step": 9615 }, { "epoch": 0.01, "learning_rate": 4.998882691413518e-05, "loss": 1.2078, "step": 9616 }, { "epoch": 0.01, "learning_rate": 4.998882456591316e-05, "loss": 1.2542, "step": 9617 }, { "epoch": 0.01, "learning_rate": 4.9988822217444455e-05, "loss": 1.3269, "step": 9618 }, { "epoch": 0.01, "learning_rate": 4.9988819868729076e-05, "loss": 1.1625, "step": 9619 }, { "epoch": 0.01, "learning_rate": 4.9988817519767014e-05, "loss": 1.1604, "step": 9620 }, { "epoch": 0.01, "learning_rate": 4.9988815170558276e-05, "loss": 1.0048, "step": 9621 }, { "epoch": 0.01, "learning_rate": 4.9988812821102854e-05, "loss": 0.9177, "step": 9622 }, { "epoch": 0.01, "learning_rate": 4.9988810471400756e-05, "loss": 0.7633, "step": 9623 }, { "epoch": 0.01, "learning_rate": 4.998880812145198e-05, "loss": 0.6968, "step": 9624 }, { "epoch": 0.01, "learning_rate": 4.998880577125652e-05, "loss": 1.1713, "step": 9625 }, { "epoch": 0.01, "learning_rate": 4.998880342081439e-05, "loss": 1.2567, "step": 9626 }, { "epoch": 0.01, "learning_rate": 4.9988801070125574e-05, "loss": 0.8987, "step": 9627 }, { "epoch": 0.01, "learning_rate": 4.9988798719190074e-05, "loss": 1.107, "step": 9628 }, { "epoch": 0.01, "learning_rate": 4.9988796368007905e-05, "loss": 1.1169, "step": 9629 }, { "epoch": 0.01, "learning_rate": 4.998879401657905e-05, "loss": 1.0902, "step": 9630 }, { "epoch": 0.01, "learning_rate": 4.998879166490353e-05, "loss": 1.0164, "step": 9631 }, { "epoch": 0.01, "learning_rate": 4.998878931298132e-05, "loss": 0.9741, "step": 9632 }, { "epoch": 0.01, "learning_rate": 4.998878696081244e-05, "loss": 0.9065, "step": 9633 }, { "epoch": 0.01, "learning_rate": 4.9988784608396865e-05, "loss": 1.2073, "step": 9634 }, { "epoch": 0.01, "learning_rate": 4.9988782255734624e-05, "loss": 1.1442, "step": 9635 }, { "epoch": 0.01, "learning_rate": 4.99887799028257e-05, "loss": 1.2982, "step": 9636 }, { "epoch": 0.01, "learning_rate": 4.998877754967009e-05, "loss": 1.0833, "step": 9637 }, { "epoch": 0.01, "learning_rate": 4.998877519626782e-05, "loss": 1.2827, "step": 9638 }, { "epoch": 0.01, "learning_rate": 4.998877284261886e-05, "loss": 1.3037, "step": 9639 }, { "epoch": 0.01, "learning_rate": 4.998877048872322e-05, "loss": 1.1347, "step": 9640 }, { "epoch": 0.01, "learning_rate": 4.99887681345809e-05, "loss": 1.0885, "step": 9641 }, { "epoch": 0.01, "learning_rate": 4.9988765780191906e-05, "loss": 1.1442, "step": 9642 }, { "epoch": 0.01, "learning_rate": 4.9988763425556234e-05, "loss": 0.748, "step": 9643 }, { "epoch": 0.01, "learning_rate": 4.9988761070673886e-05, "loss": 1.2061, "step": 9644 }, { "epoch": 0.01, "learning_rate": 4.998875871554485e-05, "loss": 1.2798, "step": 9645 }, { "epoch": 0.01, "learning_rate": 4.998875636016914e-05, "loss": 1.0119, "step": 9646 }, { "epoch": 0.01, "learning_rate": 4.998875400454676e-05, "loss": 1.0513, "step": 9647 }, { "epoch": 0.01, "learning_rate": 4.998875164867769e-05, "loss": 1.0932, "step": 9648 }, { "epoch": 0.01, "learning_rate": 4.998874929256194e-05, "loss": 1.0462, "step": 9649 }, { "epoch": 0.01, "learning_rate": 4.998874693619952e-05, "loss": 0.8771, "step": 9650 }, { "epoch": 0.01, "learning_rate": 4.998874457959042e-05, "loss": 0.9754, "step": 9651 }, { "epoch": 0.01, "learning_rate": 4.998874222273464e-05, "loss": 1.2265, "step": 9652 }, { "epoch": 0.01, "learning_rate": 4.998873986563218e-05, "loss": 1.1009, "step": 9653 }, { "epoch": 0.01, "learning_rate": 4.9988737508283044e-05, "loss": 1.1044, "step": 9654 }, { "epoch": 0.01, "learning_rate": 4.998873515068723e-05, "loss": 1.0872, "step": 9655 }, { "epoch": 0.01, "learning_rate": 4.998873279284474e-05, "loss": 1.3058, "step": 9656 }, { "epoch": 0.01, "learning_rate": 4.9988730434755564e-05, "loss": 0.9225, "step": 9657 }, { "epoch": 0.01, "learning_rate": 4.9988728076419714e-05, "loss": 0.9627, "step": 9658 }, { "epoch": 0.01, "learning_rate": 4.998872571783719e-05, "loss": 1.1607, "step": 9659 }, { "epoch": 0.01, "learning_rate": 4.9988723359007983e-05, "loss": 1.1587, "step": 9660 }, { "epoch": 0.01, "learning_rate": 4.99887209999321e-05, "loss": 1.2425, "step": 9661 }, { "epoch": 0.01, "learning_rate": 4.9988718640609535e-05, "loss": 1.0909, "step": 9662 }, { "epoch": 0.01, "learning_rate": 4.9988716281040296e-05, "loss": 1.0877, "step": 9663 }, { "epoch": 0.01, "learning_rate": 4.998871392122438e-05, "loss": 1.1519, "step": 9664 }, { "epoch": 0.01, "learning_rate": 4.998871156116178e-05, "loss": 1.5058, "step": 9665 }, { "epoch": 0.01, "learning_rate": 4.998870920085251e-05, "loss": 1.6072, "step": 9666 }, { "epoch": 0.01, "learning_rate": 4.998870684029655e-05, "loss": 1.698, "step": 9667 }, { "epoch": 0.01, "learning_rate": 4.998870447949392e-05, "loss": 1.301, "step": 9668 }, { "epoch": 0.01, "learning_rate": 4.998870211844462e-05, "loss": 1.3828, "step": 9669 }, { "epoch": 0.01, "learning_rate": 4.9988699757148625e-05, "loss": 1.5221, "step": 9670 }, { "epoch": 0.01, "learning_rate": 4.998869739560597e-05, "loss": 1.0533, "step": 9671 }, { "epoch": 0.01, "learning_rate": 4.998869503381662e-05, "loss": 0.8912, "step": 9672 }, { "epoch": 0.01, "learning_rate": 4.99886926717806e-05, "loss": 0.7509, "step": 9673 }, { "epoch": 0.01, "learning_rate": 4.99886903094979e-05, "loss": 0.8431, "step": 9674 }, { "epoch": 0.01, "learning_rate": 4.9988687946968526e-05, "loss": 0.188, "step": 9675 }, { "epoch": 0.01, "learning_rate": 4.9988685584192475e-05, "loss": 0.6945, "step": 9676 }, { "epoch": 0.01, "learning_rate": 4.998868322116975e-05, "loss": 1.049, "step": 9677 }, { "epoch": 0.01, "learning_rate": 4.9988680857900336e-05, "loss": 1.0678, "step": 9678 }, { "epoch": 0.01, "learning_rate": 4.998867849438425e-05, "loss": 1.2841, "step": 9679 }, { "epoch": 0.01, "learning_rate": 4.9988676130621485e-05, "loss": 1.0755, "step": 9680 }, { "epoch": 0.01, "learning_rate": 4.9988673766612045e-05, "loss": 0.8511, "step": 9681 }, { "epoch": 0.01, "learning_rate": 4.998867140235592e-05, "loss": 1.0256, "step": 9682 }, { "epoch": 0.01, "learning_rate": 4.998866903785312e-05, "loss": 1.1308, "step": 9683 }, { "epoch": 0.01, "learning_rate": 4.9988666673103654e-05, "loss": 0.9747, "step": 9684 }, { "epoch": 0.01, "learning_rate": 4.99886643081075e-05, "loss": 1.1631, "step": 9685 }, { "epoch": 0.01, "learning_rate": 4.998866194286467e-05, "loss": 1.09, "step": 9686 }, { "epoch": 0.01, "learning_rate": 4.998865957737516e-05, "loss": 1.8164, "step": 9687 }, { "epoch": 0.01, "learning_rate": 4.998865721163898e-05, "loss": 1.7847, "step": 9688 }, { "epoch": 0.01, "learning_rate": 4.998865484565611e-05, "loss": 1.7774, "step": 9689 }, { "epoch": 0.01, "learning_rate": 4.998865247942657e-05, "loss": 1.3281, "step": 9690 }, { "epoch": 0.01, "learning_rate": 4.9988650112950354e-05, "loss": 1.3932, "step": 9691 }, { "epoch": 0.01, "learning_rate": 4.9988647746227454e-05, "loss": 1.4631, "step": 9692 }, { "epoch": 0.01, "learning_rate": 4.998864537925789e-05, "loss": 1.2774, "step": 9693 }, { "epoch": 0.01, "learning_rate": 4.998864301204164e-05, "loss": 1.2605, "step": 9694 }, { "epoch": 0.01, "learning_rate": 4.998864064457871e-05, "loss": 0.9692, "step": 9695 }, { "epoch": 0.01, "learning_rate": 4.9988638276869106e-05, "loss": 1.1142, "step": 9696 }, { "epoch": 0.01, "learning_rate": 4.998863590891283e-05, "loss": 1.2291, "step": 9697 }, { "epoch": 0.01, "learning_rate": 4.9988633540709867e-05, "loss": 1.0415, "step": 9698 }, { "epoch": 0.01, "learning_rate": 4.998863117226023e-05, "loss": 1.2165, "step": 9699 }, { "epoch": 0.01, "learning_rate": 4.998862880356392e-05, "loss": 1.092, "step": 9700 }, { "epoch": 0.01, "learning_rate": 4.998862643462092e-05, "loss": 1.5197, "step": 9701 }, { "epoch": 0.01, "learning_rate": 4.998862406543125e-05, "loss": 1.032, "step": 9702 }, { "epoch": 0.01, "learning_rate": 4.998862169599491e-05, "loss": 1.2112, "step": 9703 }, { "epoch": 0.01, "learning_rate": 4.998861932631189e-05, "loss": 1.0574, "step": 9704 }, { "epoch": 0.01, "learning_rate": 4.998861695638219e-05, "loss": 1.1226, "step": 9705 }, { "epoch": 0.01, "learning_rate": 4.998861458620582e-05, "loss": 1.1803, "step": 9706 }, { "epoch": 0.01, "learning_rate": 4.998861221578276e-05, "loss": 1.3568, "step": 9707 }, { "epoch": 0.01, "learning_rate": 4.998860984511303e-05, "loss": 1.6537, "step": 9708 }, { "epoch": 0.01, "learning_rate": 4.998860747419663e-05, "loss": 1.0766, "step": 9709 }, { "epoch": 0.01, "learning_rate": 4.998860510303354e-05, "loss": 1.119, "step": 9710 }, { "epoch": 0.01, "learning_rate": 4.9988602731623776e-05, "loss": 1.1105, "step": 9711 }, { "epoch": 0.01, "learning_rate": 4.998860035996734e-05, "loss": 0.682, "step": 9712 }, { "epoch": 0.01, "learning_rate": 4.998859798806422e-05, "loss": 1.0323, "step": 9713 }, { "epoch": 0.01, "learning_rate": 4.998859561591443e-05, "loss": 0.7753, "step": 9714 }, { "epoch": 0.01, "learning_rate": 4.998859324351796e-05, "loss": 1.0854, "step": 9715 }, { "epoch": 0.01, "learning_rate": 4.998859087087482e-05, "loss": 0.9504, "step": 9716 }, { "epoch": 0.01, "learning_rate": 4.9988588497984986e-05, "loss": 1.2355, "step": 9717 }, { "epoch": 0.01, "learning_rate": 4.998858612484849e-05, "loss": 1.2058, "step": 9718 }, { "epoch": 0.01, "learning_rate": 4.9988583751465314e-05, "loss": 1.6141, "step": 9719 }, { "epoch": 0.01, "learning_rate": 4.9988581377835464e-05, "loss": 1.2754, "step": 9720 }, { "epoch": 0.01, "learning_rate": 4.998857900395894e-05, "loss": 0.976, "step": 9721 }, { "epoch": 0.01, "learning_rate": 4.998857662983573e-05, "loss": 1.2198, "step": 9722 }, { "epoch": 0.01, "learning_rate": 4.998857425546584e-05, "loss": 0.9265, "step": 9723 }, { "epoch": 0.01, "learning_rate": 4.998857188084928e-05, "loss": 1.2309, "step": 9724 }, { "epoch": 0.01, "learning_rate": 4.998856950598604e-05, "loss": 0.8348, "step": 9725 }, { "epoch": 0.01, "learning_rate": 4.998856713087613e-05, "loss": 1.1515, "step": 9726 }, { "epoch": 0.01, "learning_rate": 4.998856475551954e-05, "loss": 1.0123, "step": 9727 }, { "epoch": 0.01, "learning_rate": 4.998856237991627e-05, "loss": 1.0243, "step": 9728 }, { "epoch": 0.01, "learning_rate": 4.998856000406633e-05, "loss": 1.0206, "step": 9729 }, { "epoch": 0.01, "learning_rate": 4.998855762796971e-05, "loss": 1.1601, "step": 9730 }, { "epoch": 0.01, "learning_rate": 4.998855525162641e-05, "loss": 0.9926, "step": 9731 }, { "epoch": 0.01, "learning_rate": 4.9988552875036446e-05, "loss": 0.7021, "step": 9732 }, { "epoch": 0.01, "learning_rate": 4.9988550498199796e-05, "loss": 0.964, "step": 9733 }, { "epoch": 0.01, "learning_rate": 4.998854812111647e-05, "loss": 0.972, "step": 9734 }, { "epoch": 0.01, "learning_rate": 4.998854574378646e-05, "loss": 1.144, "step": 9735 }, { "epoch": 0.01, "learning_rate": 4.9988543366209784e-05, "loss": 1.2402, "step": 9736 }, { "epoch": 0.01, "learning_rate": 4.9988540988386436e-05, "loss": 1.0957, "step": 9737 }, { "epoch": 0.01, "learning_rate": 4.9988538610316405e-05, "loss": 1.081, "step": 9738 }, { "epoch": 0.01, "learning_rate": 4.998853623199969e-05, "loss": 1.106, "step": 9739 }, { "epoch": 0.01, "learning_rate": 4.998853385343631e-05, "loss": 0.6211, "step": 9740 }, { "epoch": 0.01, "learning_rate": 4.998853147462625e-05, "loss": 1.0139, "step": 9741 }, { "epoch": 0.01, "learning_rate": 4.998852909556951e-05, "loss": 1.1727, "step": 9742 }, { "epoch": 0.01, "learning_rate": 4.9988526716266106e-05, "loss": 0.9099, "step": 9743 }, { "epoch": 0.01, "learning_rate": 4.998852433671602e-05, "loss": 0.9932, "step": 9744 }, { "epoch": 0.01, "learning_rate": 4.9988521956919245e-05, "loss": 0.7721, "step": 9745 }, { "epoch": 0.01, "learning_rate": 4.99885195768758e-05, "loss": 1.2619, "step": 9746 }, { "epoch": 0.01, "learning_rate": 4.9988517196585685e-05, "loss": 1.2188, "step": 9747 }, { "epoch": 0.01, "learning_rate": 4.998851481604889e-05, "loss": 1.2085, "step": 9748 }, { "epoch": 0.01, "learning_rate": 4.998851243526542e-05, "loss": 0.9917, "step": 9749 }, { "epoch": 0.01, "learning_rate": 4.998851005423528e-05, "loss": 1.3981, "step": 9750 }, { "epoch": 0.01, "learning_rate": 4.9988507672958464e-05, "loss": 1.2426, "step": 9751 }, { "epoch": 0.01, "learning_rate": 4.9988505291434965e-05, "loss": 1.0428, "step": 9752 }, { "epoch": 0.01, "learning_rate": 4.998850290966479e-05, "loss": 1.1453, "step": 9753 }, { "epoch": 0.01, "learning_rate": 4.9988500527647944e-05, "loss": 1.0549, "step": 9754 }, { "epoch": 0.01, "learning_rate": 4.998849814538441e-05, "loss": 1.0695, "step": 9755 }, { "epoch": 0.01, "learning_rate": 4.998849576287421e-05, "loss": 1.1746, "step": 9756 }, { "epoch": 0.01, "learning_rate": 4.9988493380117336e-05, "loss": 1.2758, "step": 9757 }, { "epoch": 0.01, "learning_rate": 4.998849099711378e-05, "loss": 1.1487, "step": 9758 }, { "epoch": 0.01, "learning_rate": 4.998848861386355e-05, "loss": 1.1116, "step": 9759 }, { "epoch": 0.01, "learning_rate": 4.998848623036665e-05, "loss": 1.1157, "step": 9760 }, { "epoch": 0.01, "learning_rate": 4.998848384662307e-05, "loss": 1.0451, "step": 9761 }, { "epoch": 0.01, "learning_rate": 4.9988481462632806e-05, "loss": 0.8861, "step": 9762 }, { "epoch": 0.01, "learning_rate": 4.998847907839588e-05, "loss": 1.2327, "step": 9763 }, { "epoch": 0.01, "learning_rate": 4.9988476693912266e-05, "loss": 1.5421, "step": 9764 }, { "epoch": 0.01, "learning_rate": 4.998847430918199e-05, "loss": 0.8174, "step": 9765 }, { "epoch": 0.01, "learning_rate": 4.9988471924205026e-05, "loss": 0.9378, "step": 9766 }, { "epoch": 0.01, "learning_rate": 4.998846953898139e-05, "loss": 1.1805, "step": 9767 }, { "epoch": 0.01, "learning_rate": 4.998846715351108e-05, "loss": 1.1691, "step": 9768 }, { "epoch": 0.01, "learning_rate": 4.99884647677941e-05, "loss": 1.0429, "step": 9769 }, { "epoch": 0.01, "learning_rate": 4.998846238183044e-05, "loss": 1.0347, "step": 9770 }, { "epoch": 0.01, "learning_rate": 4.9988459995620096e-05, "loss": 1.0842, "step": 9771 }, { "epoch": 0.01, "learning_rate": 4.9988457609163083e-05, "loss": 1.4174, "step": 9772 }, { "epoch": 0.01, "learning_rate": 4.9988455222459395e-05, "loss": 1.3438, "step": 9773 }, { "epoch": 0.01, "learning_rate": 4.998845283550902e-05, "loss": 1.3013, "step": 9774 }, { "epoch": 0.01, "learning_rate": 4.998845044831199e-05, "loss": 0.4824, "step": 9775 }, { "epoch": 0.01, "learning_rate": 4.998844806086828e-05, "loss": 0.2471, "step": 9776 }, { "epoch": 0.01, "learning_rate": 4.9988445673177884e-05, "loss": 0.2471, "step": 9777 }, { "epoch": 0.01, "learning_rate": 4.998844328524081e-05, "loss": 0.1561, "step": 9778 }, { "epoch": 0.01, "learning_rate": 4.998844089705707e-05, "loss": 0.1945, "step": 9779 }, { "epoch": 0.01, "learning_rate": 4.9988438508626664e-05, "loss": 0.1443, "step": 9780 }, { "epoch": 0.01, "learning_rate": 4.9988436119949565e-05, "loss": 0.1177, "step": 9781 }, { "epoch": 0.01, "learning_rate": 4.99884337310258e-05, "loss": 0.1291, "step": 9782 }, { "epoch": 0.01, "learning_rate": 4.998843134185536e-05, "loss": 0.2359, "step": 9783 }, { "epoch": 0.01, "learning_rate": 4.9988428952438236e-05, "loss": 0.3172, "step": 9784 }, { "epoch": 0.01, "learning_rate": 4.998842656277445e-05, "loss": 0.2112, "step": 9785 }, { "epoch": 0.01, "learning_rate": 4.998842417286398e-05, "loss": 0.8715, "step": 9786 }, { "epoch": 0.01, "learning_rate": 4.998842178270684e-05, "loss": 1.3531, "step": 9787 }, { "epoch": 0.01, "learning_rate": 4.998841939230302e-05, "loss": 1.3317, "step": 9788 }, { "epoch": 0.01, "learning_rate": 4.998841700165252e-05, "loss": 1.3058, "step": 9789 }, { "epoch": 0.01, "learning_rate": 4.998841461075535e-05, "loss": 1.2345, "step": 9790 }, { "epoch": 0.01, "learning_rate": 4.9988412219611516e-05, "loss": 0.7695, "step": 9791 }, { "epoch": 0.01, "learning_rate": 4.998840982822099e-05, "loss": 0.8632, "step": 9792 }, { "epoch": 0.01, "learning_rate": 4.9988407436583796e-05, "loss": 1.1626, "step": 9793 }, { "epoch": 0.01, "learning_rate": 4.9988405044699925e-05, "loss": 1.0613, "step": 9794 }, { "epoch": 0.01, "learning_rate": 4.998840265256939e-05, "loss": 1.042, "step": 9795 }, { "epoch": 0.01, "learning_rate": 4.998840026019217e-05, "loss": 0.3696, "step": 9796 }, { "epoch": 0.01, "learning_rate": 4.998839786756827e-05, "loss": 0.77, "step": 9797 }, { "epoch": 0.01, "learning_rate": 4.9988395474697705e-05, "loss": 1.2609, "step": 9798 }, { "epoch": 0.01, "learning_rate": 4.998839308158046e-05, "loss": 1.0351, "step": 9799 }, { "epoch": 0.01, "learning_rate": 4.998839068821654e-05, "loss": 2.8369, "step": 9800 }, { "epoch": 0.01, "learning_rate": 4.998838829460595e-05, "loss": 1.3971, "step": 9801 }, { "epoch": 0.01, "learning_rate": 4.998838590074868e-05, "loss": 1.3847, "step": 9802 }, { "epoch": 0.01, "learning_rate": 4.998838350664474e-05, "loss": 1.1103, "step": 9803 }, { "epoch": 0.01, "learning_rate": 4.998838111229411e-05, "loss": 1.404, "step": 9804 }, { "epoch": 0.01, "learning_rate": 4.998837871769683e-05, "loss": 1.1215, "step": 9805 }, { "epoch": 0.01, "learning_rate": 4.998837632285286e-05, "loss": 0.905, "step": 9806 }, { "epoch": 0.01, "learning_rate": 4.998837392776221e-05, "loss": 1.1019, "step": 9807 }, { "epoch": 0.01, "learning_rate": 4.99883715324249e-05, "loss": 0.7928, "step": 9808 }, { "epoch": 0.01, "learning_rate": 4.998836913684091e-05, "loss": 0.9298, "step": 9809 }, { "epoch": 0.01, "learning_rate": 4.998836674101024e-05, "loss": 1.22, "step": 9810 }, { "epoch": 0.01, "learning_rate": 4.99883643449329e-05, "loss": 1.2782, "step": 9811 }, { "epoch": 0.01, "learning_rate": 4.9988361948608885e-05, "loss": 1.0752, "step": 9812 }, { "epoch": 0.01, "learning_rate": 4.99883595520382e-05, "loss": 1.2075, "step": 9813 }, { "epoch": 0.01, "learning_rate": 4.998835715522083e-05, "loss": 1.3142, "step": 9814 }, { "epoch": 0.01, "learning_rate": 4.998835475815679e-05, "loss": 1.3853, "step": 9815 }, { "epoch": 0.01, "learning_rate": 4.998835236084608e-05, "loss": 1.2796, "step": 9816 }, { "epoch": 0.01, "learning_rate": 4.998834996328869e-05, "loss": 1.1469, "step": 9817 }, { "epoch": 0.01, "learning_rate": 4.998834756548463e-05, "loss": 0.8047, "step": 9818 }, { "epoch": 0.01, "learning_rate": 4.9988345167433884e-05, "loss": 1.2134, "step": 9819 }, { "epoch": 0.01, "learning_rate": 4.9988342769136476e-05, "loss": 1.0523, "step": 9820 }, { "epoch": 0.01, "learning_rate": 4.998834037059239e-05, "loss": 1.109, "step": 9821 }, { "epoch": 0.01, "learning_rate": 4.998833797180163e-05, "loss": 1.2216, "step": 9822 }, { "epoch": 0.01, "learning_rate": 4.9988335572764193e-05, "loss": 1.0843, "step": 9823 }, { "epoch": 0.01, "learning_rate": 4.998833317348009e-05, "loss": 1.0799, "step": 9824 }, { "epoch": 0.01, "learning_rate": 4.9988330773949304e-05, "loss": 1.2196, "step": 9825 }, { "epoch": 0.01, "learning_rate": 4.9988328374171845e-05, "loss": 1.1059, "step": 9826 }, { "epoch": 0.01, "learning_rate": 4.9988325974147716e-05, "loss": 1.1167, "step": 9827 }, { "epoch": 0.01, "learning_rate": 4.9988323573876904e-05, "loss": 1.1676, "step": 9828 }, { "epoch": 0.01, "learning_rate": 4.998832117335943e-05, "loss": 0.9388, "step": 9829 }, { "epoch": 0.01, "learning_rate": 4.998831877259528e-05, "loss": 0.8102, "step": 9830 }, { "epoch": 0.01, "learning_rate": 4.9988316371584444e-05, "loss": 0.9359, "step": 9831 }, { "epoch": 0.01, "learning_rate": 4.998831397032694e-05, "loss": 1.1201, "step": 9832 }, { "epoch": 0.01, "learning_rate": 4.998831156882277e-05, "loss": 1.09, "step": 9833 }, { "epoch": 0.01, "learning_rate": 4.998830916707191e-05, "loss": 0.981, "step": 9834 }, { "epoch": 0.01, "learning_rate": 4.998830676507439e-05, "loss": 0.9748, "step": 9835 }, { "epoch": 0.01, "learning_rate": 4.998830436283019e-05, "loss": 1.1448, "step": 9836 }, { "epoch": 0.01, "learning_rate": 4.998830196033932e-05, "loss": 0.9496, "step": 9837 }, { "epoch": 0.01, "learning_rate": 4.998829955760177e-05, "loss": 1.2107, "step": 9838 }, { "epoch": 0.01, "learning_rate": 4.9988297154617546e-05, "loss": 0.8966, "step": 9839 }, { "epoch": 0.01, "learning_rate": 4.998829475138666e-05, "loss": 0.731, "step": 9840 }, { "epoch": 0.01, "learning_rate": 4.998829234790908e-05, "loss": 1.031, "step": 9841 }, { "epoch": 0.01, "learning_rate": 4.998828994418484e-05, "loss": 0.3472, "step": 9842 }, { "epoch": 0.01, "learning_rate": 4.998828754021393e-05, "loss": 0.3417, "step": 9843 }, { "epoch": 0.01, "learning_rate": 4.998828513599634e-05, "loss": 0.3128, "step": 9844 }, { "epoch": 0.01, "learning_rate": 4.998828273153208e-05, "loss": 0.3333, "step": 9845 }, { "epoch": 0.01, "learning_rate": 4.998828032682113e-05, "loss": 0.5605, "step": 9846 }, { "epoch": 0.01, "learning_rate": 4.9988277921863525e-05, "loss": 0.2648, "step": 9847 }, { "epoch": 0.01, "learning_rate": 4.998827551665924e-05, "loss": 0.2204, "step": 9848 }, { "epoch": 0.01, "learning_rate": 4.998827311120828e-05, "loss": 0.2649, "step": 9849 }, { "epoch": 0.01, "learning_rate": 4.9988270705510644e-05, "loss": 1.7215, "step": 9850 }, { "epoch": 0.01, "learning_rate": 4.998826829956634e-05, "loss": 1.6377, "step": 9851 }, { "epoch": 0.01, "learning_rate": 4.9988265893375354e-05, "loss": 1.4358, "step": 9852 }, { "epoch": 0.01, "learning_rate": 4.99882634869377e-05, "loss": 1.1824, "step": 9853 }, { "epoch": 0.01, "learning_rate": 4.9988261080253374e-05, "loss": 1.102, "step": 9854 }, { "epoch": 0.01, "learning_rate": 4.9988258673322376e-05, "loss": 1.1893, "step": 9855 }, { "epoch": 0.01, "learning_rate": 4.99882562661447e-05, "loss": 0.9691, "step": 9856 }, { "epoch": 0.01, "learning_rate": 4.998825385872035e-05, "loss": 0.7626, "step": 9857 }, { "epoch": 0.01, "learning_rate": 4.998825145104933e-05, "loss": 0.5945, "step": 9858 }, { "epoch": 0.01, "learning_rate": 4.9988249043131634e-05, "loss": 1.1725, "step": 9859 }, { "epoch": 0.01, "learning_rate": 4.998824663496727e-05, "loss": 1.2295, "step": 9860 }, { "epoch": 0.01, "learning_rate": 4.9988244226556225e-05, "loss": 1.0714, "step": 9861 }, { "epoch": 0.01, "learning_rate": 4.9988241817898506e-05, "loss": 0.6421, "step": 9862 }, { "epoch": 0.01, "learning_rate": 4.998823940899412e-05, "loss": 1.0365, "step": 9863 }, { "epoch": 0.01, "learning_rate": 4.998823699984306e-05, "loss": 1.0579, "step": 9864 }, { "epoch": 0.01, "learning_rate": 4.9988234590445326e-05, "loss": 0.9681, "step": 9865 }, { "epoch": 0.01, "learning_rate": 4.9988232180800915e-05, "loss": 1.1143, "step": 9866 }, { "epoch": 0.01, "learning_rate": 4.9988229770909835e-05, "loss": 1.2837, "step": 9867 }, { "epoch": 0.01, "learning_rate": 4.998822736077208e-05, "loss": 1.1725, "step": 9868 }, { "epoch": 0.01, "learning_rate": 4.998822495038765e-05, "loss": 1.0829, "step": 9869 }, { "epoch": 0.01, "learning_rate": 4.9988222539756544e-05, "loss": 1.0975, "step": 9870 }, { "epoch": 0.01, "learning_rate": 4.998822012887877e-05, "loss": 1.1761, "step": 9871 }, { "epoch": 0.01, "learning_rate": 4.9988217717754324e-05, "loss": 1.1875, "step": 9872 }, { "epoch": 0.01, "learning_rate": 4.99882153063832e-05, "loss": 1.2365, "step": 9873 }, { "epoch": 0.01, "learning_rate": 4.9988212894765405e-05, "loss": 1.3565, "step": 9874 }, { "epoch": 0.01, "learning_rate": 4.9988210482900935e-05, "loss": 1.0789, "step": 9875 }, { "epoch": 0.01, "learning_rate": 4.99882080707898e-05, "loss": 1.159, "step": 9876 }, { "epoch": 0.01, "learning_rate": 4.9988205658431985e-05, "loss": 1.3166, "step": 9877 }, { "epoch": 0.01, "learning_rate": 4.99882032458275e-05, "loss": 0.9486, "step": 9878 }, { "epoch": 0.01, "learning_rate": 4.998820083297634e-05, "loss": 0.9855, "step": 9879 }, { "epoch": 0.01, "learning_rate": 4.9988198419878506e-05, "loss": 0.9638, "step": 9880 }, { "epoch": 0.01, "learning_rate": 4.9988196006534e-05, "loss": 1.309, "step": 9881 }, { "epoch": 0.01, "learning_rate": 4.998819359294282e-05, "loss": 1.0043, "step": 9882 }, { "epoch": 0.01, "learning_rate": 4.998819117910497e-05, "loss": 1.1235, "step": 9883 }, { "epoch": 0.01, "learning_rate": 4.998818876502044e-05, "loss": 1.2024, "step": 9884 }, { "epoch": 0.01, "learning_rate": 4.998818635068925e-05, "loss": 0.9608, "step": 9885 }, { "epoch": 0.01, "learning_rate": 4.9988183936111374e-05, "loss": 0.9681, "step": 9886 }, { "epoch": 0.01, "learning_rate": 4.998818152128683e-05, "loss": 1.2117, "step": 9887 }, { "epoch": 0.01, "learning_rate": 4.9988179106215614e-05, "loss": 1.1707, "step": 9888 }, { "epoch": 0.01, "learning_rate": 4.998817669089773e-05, "loss": 0.6553, "step": 9889 }, { "epoch": 0.01, "learning_rate": 4.998817427533317e-05, "loss": 1.047, "step": 9890 }, { "epoch": 0.01, "learning_rate": 4.998817185952193e-05, "loss": 0.9099, "step": 9891 }, { "epoch": 0.01, "learning_rate": 4.998816944346403e-05, "loss": 1.2208, "step": 9892 }, { "epoch": 0.01, "learning_rate": 4.998816702715944e-05, "loss": 1.1677, "step": 9893 }, { "epoch": 0.01, "learning_rate": 4.998816461060819e-05, "loss": 1.0616, "step": 9894 }, { "epoch": 0.01, "learning_rate": 4.9988162193810266e-05, "loss": 1.1913, "step": 9895 }, { "epoch": 0.01, "learning_rate": 4.998815977676567e-05, "loss": 1.8925, "step": 9896 }, { "epoch": 0.01, "learning_rate": 4.99881573594744e-05, "loss": 1.3356, "step": 9897 }, { "epoch": 0.01, "learning_rate": 4.998815494193646e-05, "loss": 1.1028, "step": 9898 }, { "epoch": 0.01, "learning_rate": 4.998815252415184e-05, "loss": 1.2409, "step": 9899 }, { "epoch": 0.01, "learning_rate": 4.998815010612056e-05, "loss": 1.1552, "step": 9900 }, { "epoch": 0.01, "learning_rate": 4.998814768784259e-05, "loss": 1.0364, "step": 9901 }, { "epoch": 0.01, "learning_rate": 4.998814526931796e-05, "loss": 0.6646, "step": 9902 }, { "epoch": 0.01, "learning_rate": 4.998814285054666e-05, "loss": 0.9307, "step": 9903 }, { "epoch": 0.01, "learning_rate": 4.998814043152868e-05, "loss": 0.902, "step": 9904 }, { "epoch": 0.01, "learning_rate": 4.998813801226403e-05, "loss": 1.3225, "step": 9905 }, { "epoch": 0.01, "learning_rate": 4.998813559275271e-05, "loss": 1.0705, "step": 9906 }, { "epoch": 0.01, "learning_rate": 4.998813317299471e-05, "loss": 1.1976, "step": 9907 }, { "epoch": 0.01, "learning_rate": 4.998813075299005e-05, "loss": 0.8782, "step": 9908 }, { "epoch": 0.01, "learning_rate": 4.9988128332738706e-05, "loss": 1.282, "step": 9909 }, { "epoch": 0.01, "learning_rate": 4.998812591224069e-05, "loss": 1.3072, "step": 9910 }, { "epoch": 0.01, "learning_rate": 4.998812349149601e-05, "loss": 1.2509, "step": 9911 }, { "epoch": 0.01, "learning_rate": 4.998812107050466e-05, "loss": 1.2104, "step": 9912 }, { "epoch": 0.01, "learning_rate": 4.998811864926662e-05, "loss": 1.1429, "step": 9913 }, { "epoch": 0.01, "learning_rate": 4.9988116227781925e-05, "loss": 1.34, "step": 9914 }, { "epoch": 0.01, "learning_rate": 4.998811380605055e-05, "loss": 0.9676, "step": 9915 }, { "epoch": 0.01, "learning_rate": 4.9988111384072514e-05, "loss": 0.9763, "step": 9916 }, { "epoch": 0.01, "learning_rate": 4.998810896184779e-05, "loss": 1.2044, "step": 9917 }, { "epoch": 0.01, "learning_rate": 4.99881065393764e-05, "loss": 1.0246, "step": 9918 }, { "epoch": 0.01, "learning_rate": 4.998810411665834e-05, "loss": 1.0786, "step": 9919 }, { "epoch": 0.01, "learning_rate": 4.998810169369361e-05, "loss": 1.0917, "step": 9920 }, { "epoch": 0.01, "learning_rate": 4.9988099270482205e-05, "loss": 1.1904, "step": 9921 }, { "epoch": 0.01, "learning_rate": 4.9988096847024124e-05, "loss": 1.2247, "step": 9922 }, { "epoch": 0.01, "learning_rate": 4.998809442331937e-05, "loss": 1.1823, "step": 9923 }, { "epoch": 0.01, "learning_rate": 4.998809199936796e-05, "loss": 1.0782, "step": 9924 }, { "epoch": 0.01, "learning_rate": 4.998808957516986e-05, "loss": 0.9249, "step": 9925 }, { "epoch": 0.01, "learning_rate": 4.9988087150725096e-05, "loss": 1.1004, "step": 9926 }, { "epoch": 0.01, "learning_rate": 4.998808472603366e-05, "loss": 1.1934, "step": 9927 }, { "epoch": 0.01, "learning_rate": 4.998808230109555e-05, "loss": 1.3222, "step": 9928 }, { "epoch": 0.01, "learning_rate": 4.998807987591077e-05, "loss": 1.2111, "step": 9929 }, { "epoch": 0.01, "learning_rate": 4.998807745047932e-05, "loss": 1.0947, "step": 9930 }, { "epoch": 0.01, "learning_rate": 4.9988075024801194e-05, "loss": 1.2274, "step": 9931 }, { "epoch": 0.01, "learning_rate": 4.9988072598876404e-05, "loss": 1.1896, "step": 9932 }, { "epoch": 0.01, "learning_rate": 4.998807017270494e-05, "loss": 1.0705, "step": 9933 }, { "epoch": 0.01, "learning_rate": 4.99880677462868e-05, "loss": 1.2374, "step": 9934 }, { "epoch": 0.01, "learning_rate": 4.998806531962199e-05, "loss": 1.466, "step": 9935 }, { "epoch": 0.01, "learning_rate": 4.9988062892710506e-05, "loss": 0.999, "step": 9936 }, { "epoch": 0.01, "learning_rate": 4.998806046555235e-05, "loss": 1.0264, "step": 9937 }, { "epoch": 0.01, "learning_rate": 4.998805803814752e-05, "loss": 1.4109, "step": 9938 }, { "epoch": 0.01, "learning_rate": 4.998805561049603e-05, "loss": 1.1204, "step": 9939 }, { "epoch": 0.01, "learning_rate": 4.998805318259786e-05, "loss": 1.11, "step": 9940 }, { "epoch": 0.01, "learning_rate": 4.998805075445302e-05, "loss": 0.5099, "step": 9941 }, { "epoch": 0.01, "learning_rate": 4.998804832606151e-05, "loss": 0.4688, "step": 9942 }, { "epoch": 0.01, "learning_rate": 4.998804589742332e-05, "loss": 0.3469, "step": 9943 }, { "epoch": 0.01, "learning_rate": 4.998804346853847e-05, "loss": 0.1816, "step": 9944 }, { "epoch": 0.01, "learning_rate": 4.998804103940694e-05, "loss": 0.1119, "step": 9945 }, { "epoch": 0.01, "learning_rate": 4.9988038610028745e-05, "loss": 0.1391, "step": 9946 }, { "epoch": 0.01, "learning_rate": 4.998803618040388e-05, "loss": 0.7113, "step": 9947 }, { "epoch": 0.01, "learning_rate": 4.998803375053234e-05, "loss": 1.4228, "step": 9948 }, { "epoch": 0.01, "learning_rate": 4.998803132041413e-05, "loss": 1.482, "step": 9949 }, { "epoch": 0.01, "learning_rate": 4.9988028890049244e-05, "loss": 1.5651, "step": 9950 }, { "epoch": 0.01, "learning_rate": 4.9988026459437685e-05, "loss": 1.1674, "step": 9951 }, { "epoch": 0.01, "learning_rate": 4.998802402857946e-05, "loss": 1.149, "step": 9952 }, { "epoch": 0.01, "learning_rate": 4.9988021597474565e-05, "loss": 1.245, "step": 9953 }, { "epoch": 0.01, "learning_rate": 4.9988019166123004e-05, "loss": 1.153, "step": 9954 }, { "epoch": 0.01, "learning_rate": 4.998801673452476e-05, "loss": 0.9997, "step": 9955 }, { "epoch": 0.01, "learning_rate": 4.998801430267985e-05, "loss": 1.0358, "step": 9956 }, { "epoch": 0.01, "learning_rate": 4.998801187058827e-05, "loss": 1.0735, "step": 9957 }, { "epoch": 0.01, "learning_rate": 4.9988009438250017e-05, "loss": 1.0173, "step": 9958 }, { "epoch": 0.01, "learning_rate": 4.9988007005665094e-05, "loss": 1.2286, "step": 9959 }, { "epoch": 0.01, "learning_rate": 4.9988004572833496e-05, "loss": 1.1883, "step": 9960 }, { "epoch": 0.01, "learning_rate": 4.998800213975523e-05, "loss": 1.0479, "step": 9961 }, { "epoch": 0.01, "learning_rate": 4.9987999706430296e-05, "loss": 1.4511, "step": 9962 }, { "epoch": 0.01, "learning_rate": 4.998799727285869e-05, "loss": 1.6647, "step": 9963 }, { "epoch": 0.01, "learning_rate": 4.9987994839040405e-05, "loss": 1.3232, "step": 9964 }, { "epoch": 0.01, "learning_rate": 4.998799240497546e-05, "loss": 1.1381, "step": 9965 }, { "epoch": 0.01, "learning_rate": 4.998798997066384e-05, "loss": 0.5195, "step": 9966 }, { "epoch": 0.01, "learning_rate": 4.9987987536105544e-05, "loss": 1.1474, "step": 9967 }, { "epoch": 0.01, "learning_rate": 4.998798510130058e-05, "loss": 0.6412, "step": 9968 }, { "epoch": 0.01, "learning_rate": 4.998798266624895e-05, "loss": 0.6699, "step": 9969 }, { "epoch": 0.01, "learning_rate": 4.998798023095065e-05, "loss": 1.0617, "step": 9970 }, { "epoch": 0.01, "learning_rate": 4.998797779540567e-05, "loss": 1.1165, "step": 9971 }, { "epoch": 0.01, "learning_rate": 4.9987975359614026e-05, "loss": 1.2524, "step": 9972 }, { "epoch": 0.01, "learning_rate": 4.998797292357571e-05, "loss": 1.064, "step": 9973 }, { "epoch": 0.01, "learning_rate": 4.9987970487290724e-05, "loss": 1.1155, "step": 9974 }, { "epoch": 0.01, "learning_rate": 4.998796805075907e-05, "loss": 1.4537, "step": 9975 }, { "epoch": 0.01, "learning_rate": 4.9987965613980737e-05, "loss": 1.5731, "step": 9976 }, { "epoch": 0.01, "learning_rate": 4.9987963176955735e-05, "loss": 1.3807, "step": 9977 }, { "epoch": 0.01, "learning_rate": 4.998796073968407e-05, "loss": 1.4099, "step": 9978 }, { "epoch": 0.01, "learning_rate": 4.9987958302165724e-05, "loss": 1.3611, "step": 9979 }, { "epoch": 0.01, "learning_rate": 4.9987955864400714e-05, "loss": 1.2911, "step": 9980 }, { "epoch": 0.01, "learning_rate": 4.9987953426389035e-05, "loss": 1.2902, "step": 9981 }, { "epoch": 0.01, "learning_rate": 4.9987950988130686e-05, "loss": 1.2613, "step": 9982 }, { "epoch": 0.01, "learning_rate": 4.998794854962566e-05, "loss": 1.1833, "step": 9983 }, { "epoch": 0.01, "learning_rate": 4.9987946110873965e-05, "loss": 1.0884, "step": 9984 }, { "epoch": 0.01, "learning_rate": 4.99879436718756e-05, "loss": 1.4898, "step": 9985 }, { "epoch": 0.01, "learning_rate": 4.998794123263057e-05, "loss": 1.5612, "step": 9986 }, { "epoch": 0.01, "learning_rate": 4.9987938793138863e-05, "loss": 1.1664, "step": 9987 }, { "epoch": 0.01, "learning_rate": 4.998793635340049e-05, "loss": 1.3252, "step": 9988 }, { "epoch": 0.01, "learning_rate": 4.998793391341544e-05, "loss": 1.0936, "step": 9989 }, { "epoch": 0.01, "learning_rate": 4.998793147318372e-05, "loss": 1.1245, "step": 9990 }, { "epoch": 0.01, "learning_rate": 4.998792903270534e-05, "loss": 1.123, "step": 9991 }, { "epoch": 0.01, "learning_rate": 4.998792659198028e-05, "loss": 1.253, "step": 9992 }, { "epoch": 0.01, "learning_rate": 4.998792415100856e-05, "loss": 1.044, "step": 9993 }, { "epoch": 0.01, "learning_rate": 4.998792170979016e-05, "loss": 1.0668, "step": 9994 }, { "epoch": 0.01, "learning_rate": 4.998791926832509e-05, "loss": 1.3681, "step": 9995 }, { "epoch": 0.01, "learning_rate": 4.9987916826613355e-05, "loss": 1.0164, "step": 9996 }, { "epoch": 0.01, "learning_rate": 4.998791438465494e-05, "loss": 1.3209, "step": 9997 }, { "epoch": 0.01, "learning_rate": 4.9987911942449874e-05, "loss": 0.9155, "step": 9998 }, { "epoch": 0.01, "learning_rate": 4.998790949999812e-05, "loss": 1.1548, "step": 9999 }, { "epoch": 0.01, "learning_rate": 4.998790705729971e-05, "loss": 1.1474, "step": 10000 }, { "epoch": 0.01, "eval_loss": 1.0545191764831543, "eval_runtime": 85.9243, "eval_samples_per_second": 16.119, "eval_steps_per_second": 4.038, "step": 10000 }, { "epoch": 0.01, "learning_rate": 4.998790461435462e-05, "loss": 0.9587, "step": 10001 }, { "epoch": 0.01, "learning_rate": 4.9987902171162856e-05, "loss": 1.1316, "step": 10002 }, { "epoch": 0.01, "learning_rate": 4.9987899727724427e-05, "loss": 1.2127, "step": 10003 }, { "epoch": 0.01, "learning_rate": 4.9987897284039334e-05, "loss": 1.0664, "step": 10004 }, { "epoch": 0.01, "learning_rate": 4.9987894840107565e-05, "loss": 1.0854, "step": 10005 }, { "epoch": 0.01, "learning_rate": 4.9987892395929127e-05, "loss": 1.1176, "step": 10006 }, { "epoch": 0.01, "learning_rate": 4.998788995150402e-05, "loss": 1.1545, "step": 10007 }, { "epoch": 0.01, "learning_rate": 4.998788750683224e-05, "loss": 1.1105, "step": 10008 }, { "epoch": 0.01, "learning_rate": 4.9987885061913794e-05, "loss": 1.8349, "step": 10009 }, { "epoch": 0.01, "learning_rate": 4.998788261674867e-05, "loss": 1.2364, "step": 10010 }, { "epoch": 0.01, "learning_rate": 4.9987880171336884e-05, "loss": 1.3148, "step": 10011 }, { "epoch": 0.01, "learning_rate": 4.998787772567843e-05, "loss": 0.9995, "step": 10012 }, { "epoch": 0.01, "learning_rate": 4.9987875279773296e-05, "loss": 1.2341, "step": 10013 }, { "epoch": 0.01, "learning_rate": 4.998787283362151e-05, "loss": 1.3008, "step": 10014 }, { "epoch": 0.01, "learning_rate": 4.998787038722304e-05, "loss": 0.7189, "step": 10015 }, { "epoch": 0.01, "learning_rate": 4.99878679405779e-05, "loss": 1.0264, "step": 10016 }, { "epoch": 0.01, "learning_rate": 4.998786549368609e-05, "loss": 1.1821, "step": 10017 }, { "epoch": 0.01, "learning_rate": 4.998786304654762e-05, "loss": 0.9963, "step": 10018 }, { "epoch": 0.01, "learning_rate": 4.998786059916247e-05, "loss": 1.0399, "step": 10019 }, { "epoch": 0.01, "learning_rate": 4.998785815153066e-05, "loss": 1.0536, "step": 10020 }, { "epoch": 0.01, "learning_rate": 4.998785570365218e-05, "loss": 1.2832, "step": 10021 }, { "epoch": 0.01, "learning_rate": 4.998785325552702e-05, "loss": 1.0504, "step": 10022 }, { "epoch": 0.01, "learning_rate": 4.99878508071552e-05, "loss": 0.9803, "step": 10023 }, { "epoch": 0.01, "learning_rate": 4.99878483585367e-05, "loss": 0.9373, "step": 10024 }, { "epoch": 0.01, "learning_rate": 4.998784590967154e-05, "loss": 1.2109, "step": 10025 }, { "epoch": 0.01, "learning_rate": 4.9987843460559705e-05, "loss": 1.3152, "step": 10026 }, { "epoch": 0.01, "learning_rate": 4.9987841011201206e-05, "loss": 1.4545, "step": 10027 }, { "epoch": 0.01, "learning_rate": 4.998783856159604e-05, "loss": 1.1669, "step": 10028 }, { "epoch": 0.01, "learning_rate": 4.998783611174419e-05, "loss": 1.2884, "step": 10029 }, { "epoch": 0.01, "learning_rate": 4.998783366164568e-05, "loss": 2.1015, "step": 10030 }, { "epoch": 0.01, "learning_rate": 4.99878312113005e-05, "loss": 1.3097, "step": 10031 }, { "epoch": 0.01, "learning_rate": 4.9987828760708654e-05, "loss": 0.4655, "step": 10032 }, { "epoch": 0.01, "learning_rate": 4.998782630987013e-05, "loss": 0.5985, "step": 10033 }, { "epoch": 0.01, "learning_rate": 4.9987823858784946e-05, "loss": 1.4386, "step": 10034 }, { "epoch": 0.01, "learning_rate": 4.998782140745309e-05, "loss": 1.3263, "step": 10035 }, { "epoch": 0.01, "learning_rate": 4.9987818955874565e-05, "loss": 1.1649, "step": 10036 }, { "epoch": 0.01, "learning_rate": 4.9987816504049364e-05, "loss": 1.2357, "step": 10037 }, { "epoch": 0.01, "learning_rate": 4.998781405197751e-05, "loss": 1.011, "step": 10038 }, { "epoch": 0.01, "learning_rate": 4.9987811599658973e-05, "loss": 1.3485, "step": 10039 }, { "epoch": 0.01, "learning_rate": 4.998780914709377e-05, "loss": 1.03, "step": 10040 }, { "epoch": 0.01, "learning_rate": 4.99878066942819e-05, "loss": 0.361, "step": 10041 }, { "epoch": 0.01, "learning_rate": 4.9987804241223356e-05, "loss": 1.4018, "step": 10042 }, { "epoch": 0.01, "learning_rate": 4.9987801787918144e-05, "loss": 1.2243, "step": 10043 }, { "epoch": 0.01, "learning_rate": 4.998779933436627e-05, "loss": 0.8755, "step": 10044 }, { "epoch": 0.01, "learning_rate": 4.998779688056772e-05, "loss": 1.2429, "step": 10045 }, { "epoch": 0.01, "learning_rate": 4.9987794426522505e-05, "loss": 1.2466, "step": 10046 }, { "epoch": 0.01, "learning_rate": 4.9987791972230615e-05, "loss": 1.1774, "step": 10047 }, { "epoch": 0.01, "learning_rate": 4.998778951769206e-05, "loss": 1.3135, "step": 10048 }, { "epoch": 0.01, "learning_rate": 4.998778706290684e-05, "loss": 1.2678, "step": 10049 }, { "epoch": 0.01, "learning_rate": 4.998778460787494e-05, "loss": 1.1939, "step": 10050 }, { "epoch": 0.01, "learning_rate": 4.998778215259639e-05, "loss": 1.1722, "step": 10051 }, { "epoch": 0.01, "learning_rate": 4.998777969707115e-05, "loss": 1.1416, "step": 10052 }, { "epoch": 0.01, "learning_rate": 4.998777724129925e-05, "loss": 1.092, "step": 10053 }, { "epoch": 0.01, "learning_rate": 4.9987774785280686e-05, "loss": 1.1459, "step": 10054 }, { "epoch": 0.01, "learning_rate": 4.9987772329015447e-05, "loss": 1.047, "step": 10055 }, { "epoch": 0.01, "learning_rate": 4.9987769872503544e-05, "loss": 1.2231, "step": 10056 }, { "epoch": 0.01, "learning_rate": 4.9987767415744966e-05, "loss": 1.2234, "step": 10057 }, { "epoch": 0.01, "learning_rate": 4.9987764958739724e-05, "loss": 0.9417, "step": 10058 }, { "epoch": 0.01, "learning_rate": 4.9987762501487813e-05, "loss": 0.9673, "step": 10059 }, { "epoch": 0.01, "learning_rate": 4.9987760043989226e-05, "loss": 1.2353, "step": 10060 }, { "epoch": 0.01, "learning_rate": 4.998775758624398e-05, "loss": 1.1185, "step": 10061 }, { "epoch": 0.01, "learning_rate": 4.998775512825207e-05, "loss": 1.0758, "step": 10062 }, { "epoch": 0.01, "learning_rate": 4.9987752670013474e-05, "loss": 1.0797, "step": 10063 }, { "epoch": 0.01, "learning_rate": 4.998775021152822e-05, "loss": 1.1433, "step": 10064 }, { "epoch": 0.01, "learning_rate": 4.99877477527963e-05, "loss": 0.9668, "step": 10065 }, { "epoch": 0.01, "learning_rate": 4.9987745293817703e-05, "loss": 0.7362, "step": 10066 }, { "epoch": 0.01, "learning_rate": 4.998774283459244e-05, "loss": 0.916, "step": 10067 }, { "epoch": 0.01, "learning_rate": 4.998774037512052e-05, "loss": 1.0283, "step": 10068 }, { "epoch": 0.01, "learning_rate": 4.998773791540192e-05, "loss": 1.2069, "step": 10069 }, { "epoch": 0.01, "learning_rate": 4.998773545543666e-05, "loss": 0.7493, "step": 10070 }, { "epoch": 0.01, "learning_rate": 4.9987732995224726e-05, "loss": 0.9055, "step": 10071 }, { "epoch": 0.01, "learning_rate": 4.998773053476612e-05, "loss": 0.9794, "step": 10072 }, { "epoch": 0.01, "learning_rate": 4.9987728074060847e-05, "loss": 1.4686, "step": 10073 }, { "epoch": 0.01, "learning_rate": 4.9987725613108906e-05, "loss": 1.1496, "step": 10074 }, { "epoch": 0.01, "learning_rate": 4.9987723151910296e-05, "loss": 1.0109, "step": 10075 }, { "epoch": 0.01, "learning_rate": 4.998772069046502e-05, "loss": 0.953, "step": 10076 }, { "epoch": 0.01, "learning_rate": 4.998771822877308e-05, "loss": 1.1296, "step": 10077 }, { "epoch": 0.01, "learning_rate": 4.998771576683446e-05, "loss": 0.6741, "step": 10078 }, { "epoch": 0.01, "learning_rate": 4.998771330464919e-05, "loss": 0.9119, "step": 10079 }, { "epoch": 0.01, "learning_rate": 4.9987710842217236e-05, "loss": 0.5591, "step": 10080 }, { "epoch": 0.01, "learning_rate": 4.998770837953862e-05, "loss": 0.4912, "step": 10081 }, { "epoch": 0.01, "learning_rate": 4.998770591661334e-05, "loss": 0.3447, "step": 10082 }, { "epoch": 0.01, "learning_rate": 4.998770345344138e-05, "loss": 0.2822, "step": 10083 }, { "epoch": 0.01, "learning_rate": 4.9987700990022764e-05, "loss": 0.1761, "step": 10084 }, { "epoch": 0.01, "learning_rate": 4.998769852635747e-05, "loss": 0.1722, "step": 10085 }, { "epoch": 0.01, "learning_rate": 4.998769606244552e-05, "loss": 0.4746, "step": 10086 }, { "epoch": 0.01, "learning_rate": 4.9987693598286886e-05, "loss": 1.0607, "step": 10087 }, { "epoch": 0.01, "learning_rate": 4.99876911338816e-05, "loss": 1.2912, "step": 10088 }, { "epoch": 0.01, "learning_rate": 4.9987688669229635e-05, "loss": 1.1969, "step": 10089 }, { "epoch": 0.01, "learning_rate": 4.998768620433101e-05, "loss": 1.169, "step": 10090 }, { "epoch": 0.01, "learning_rate": 4.9987683739185706e-05, "loss": 1.293, "step": 10091 }, { "epoch": 0.01, "learning_rate": 4.998768127379374e-05, "loss": 1.08, "step": 10092 }, { "epoch": 0.01, "learning_rate": 4.998767880815511e-05, "loss": 0.8878, "step": 10093 }, { "epoch": 0.01, "learning_rate": 4.9987676342269816e-05, "loss": 1.5998, "step": 10094 }, { "epoch": 0.01, "learning_rate": 4.998767387613784e-05, "loss": 1.1856, "step": 10095 }, { "epoch": 0.01, "learning_rate": 4.9987671409759205e-05, "loss": 1.3133, "step": 10096 }, { "epoch": 0.01, "learning_rate": 4.9987668943133905e-05, "loss": 1.43, "step": 10097 }, { "epoch": 0.01, "learning_rate": 4.998766647626193e-05, "loss": 0.6464, "step": 10098 }, { "epoch": 0.01, "learning_rate": 4.998766400914329e-05, "loss": 1.3154, "step": 10099 }, { "epoch": 0.01, "learning_rate": 4.998766154177798e-05, "loss": 1.3452, "step": 10100 }, { "epoch": 0.01, "learning_rate": 4.998765907416601e-05, "loss": 0.9236, "step": 10101 }, { "epoch": 0.01, "learning_rate": 4.9987656606307365e-05, "loss": 0.8636, "step": 10102 }, { "epoch": 0.01, "learning_rate": 4.9987654138202055e-05, "loss": 1.3781, "step": 10103 }, { "epoch": 0.01, "learning_rate": 4.9987651669850075e-05, "loss": 1.2948, "step": 10104 }, { "epoch": 0.01, "learning_rate": 4.998764920125143e-05, "loss": 1.1604, "step": 10105 }, { "epoch": 0.01, "learning_rate": 4.998764673240612e-05, "loss": 1.1219, "step": 10106 }, { "epoch": 0.01, "learning_rate": 4.998764426331414e-05, "loss": 1.3069, "step": 10107 }, { "epoch": 0.01, "learning_rate": 4.9987641793975496e-05, "loss": 1.5225, "step": 10108 }, { "epoch": 0.01, "learning_rate": 4.9987639324390175e-05, "loss": 1.0722, "step": 10109 }, { "epoch": 0.01, "learning_rate": 4.998763685455819e-05, "loss": 0.8805, "step": 10110 }, { "epoch": 0.01, "learning_rate": 4.998763438447954e-05, "loss": 1.0155, "step": 10111 }, { "epoch": 0.01, "learning_rate": 4.998763191415423e-05, "loss": 1.0353, "step": 10112 }, { "epoch": 0.01, "learning_rate": 4.998762944358224e-05, "loss": 1.3936, "step": 10113 }, { "epoch": 0.01, "learning_rate": 4.998762697276359e-05, "loss": 1.2141, "step": 10114 }, { "epoch": 0.01, "learning_rate": 4.998762450169827e-05, "loss": 1.3697, "step": 10115 }, { "epoch": 0.01, "learning_rate": 4.9987622030386286e-05, "loss": 1.1035, "step": 10116 }, { "epoch": 0.01, "learning_rate": 4.998761955882763e-05, "loss": 1.2115, "step": 10117 }, { "epoch": 0.01, "learning_rate": 4.998761708702231e-05, "loss": 0.8148, "step": 10118 }, { "epoch": 0.01, "learning_rate": 4.998761461497032e-05, "loss": 0.94, "step": 10119 }, { "epoch": 0.01, "learning_rate": 4.998761214267166e-05, "loss": 1.0566, "step": 10120 }, { "epoch": 0.01, "learning_rate": 4.998760967012634e-05, "loss": 1.2239, "step": 10121 }, { "epoch": 0.01, "learning_rate": 4.998760719733435e-05, "loss": 0.9975, "step": 10122 }, { "epoch": 0.01, "learning_rate": 4.99876047242957e-05, "loss": 1.152, "step": 10123 }, { "epoch": 0.01, "learning_rate": 4.998760225101037e-05, "loss": 1.0156, "step": 10124 }, { "epoch": 0.01, "learning_rate": 4.9987599777478376e-05, "loss": 1.3767, "step": 10125 }, { "epoch": 0.01, "learning_rate": 4.998759730369972e-05, "loss": 0.971, "step": 10126 }, { "epoch": 0.01, "learning_rate": 4.9987594829674395e-05, "loss": 1.2178, "step": 10127 }, { "epoch": 0.01, "learning_rate": 4.99875923554024e-05, "loss": 1.136, "step": 10128 }, { "epoch": 0.01, "learning_rate": 4.9987589880883744e-05, "loss": 1.0516, "step": 10129 }, { "epoch": 0.01, "learning_rate": 4.998758740611842e-05, "loss": 1.1713, "step": 10130 }, { "epoch": 0.01, "learning_rate": 4.998758493110642e-05, "loss": 1.0194, "step": 10131 }, { "epoch": 0.01, "learning_rate": 4.998758245584776e-05, "loss": 1.0846, "step": 10132 }, { "epoch": 0.01, "learning_rate": 4.998757998034243e-05, "loss": 0.9837, "step": 10133 }, { "epoch": 0.01, "learning_rate": 4.998757750459044e-05, "loss": 0.9976, "step": 10134 }, { "epoch": 0.01, "learning_rate": 4.998757502859178e-05, "loss": 1.0837, "step": 10135 }, { "epoch": 0.01, "learning_rate": 4.998757255234645e-05, "loss": 1.0913, "step": 10136 }, { "epoch": 0.01, "learning_rate": 4.998757007585446e-05, "loss": 0.5574, "step": 10137 }, { "epoch": 0.01, "learning_rate": 4.9987567599115794e-05, "loss": 0.5928, "step": 10138 }, { "epoch": 0.01, "learning_rate": 4.998756512213046e-05, "loss": 0.7991, "step": 10139 }, { "epoch": 0.01, "learning_rate": 4.9987562644898473e-05, "loss": 1.1505, "step": 10140 }, { "epoch": 0.01, "learning_rate": 4.998756016741981e-05, "loss": 1.046, "step": 10141 }, { "epoch": 0.01, "learning_rate": 4.998755768969449e-05, "loss": 1.133, "step": 10142 }, { "epoch": 0.01, "learning_rate": 4.998755521172249e-05, "loss": 1.1574, "step": 10143 }, { "epoch": 0.01, "learning_rate": 4.998755273350383e-05, "loss": 1.1993, "step": 10144 }, { "epoch": 0.01, "learning_rate": 4.99875502550385e-05, "loss": 1.1066, "step": 10145 }, { "epoch": 0.01, "learning_rate": 4.99875477763265e-05, "loss": 1.1578, "step": 10146 }, { "epoch": 0.01, "learning_rate": 4.998754529736784e-05, "loss": 0.9662, "step": 10147 }, { "epoch": 0.01, "learning_rate": 4.998754281816251e-05, "loss": 1.1373, "step": 10148 }, { "epoch": 0.01, "learning_rate": 4.9987540338710523e-05, "loss": 1.1941, "step": 10149 }, { "epoch": 0.01, "learning_rate": 4.998753785901186e-05, "loss": 0.9031, "step": 10150 }, { "epoch": 0.01, "learning_rate": 4.9987535379066534e-05, "loss": 0.9422, "step": 10151 }, { "epoch": 0.01, "learning_rate": 4.9987532898874545e-05, "loss": 1.0914, "step": 10152 }, { "epoch": 0.01, "learning_rate": 4.998753041843588e-05, "loss": 1.0855, "step": 10153 }, { "epoch": 0.01, "learning_rate": 4.998752793775056e-05, "loss": 1.1082, "step": 10154 }, { "epoch": 0.01, "learning_rate": 4.998752545681856e-05, "loss": 1.0152, "step": 10155 }, { "epoch": 0.01, "learning_rate": 4.998752297563991e-05, "loss": 1.1159, "step": 10156 }, { "epoch": 0.01, "learning_rate": 4.998752049421458e-05, "loss": 1.4043, "step": 10157 }, { "epoch": 0.01, "learning_rate": 4.998751801254259e-05, "loss": 1.177, "step": 10158 }, { "epoch": 0.01, "learning_rate": 4.998751553062393e-05, "loss": 1.1901, "step": 10159 }, { "epoch": 0.01, "learning_rate": 4.9987513048458603e-05, "loss": 0.8716, "step": 10160 }, { "epoch": 0.01, "learning_rate": 4.9987510566046616e-05, "loss": 0.9739, "step": 10161 }, { "epoch": 0.01, "learning_rate": 4.998750808338796e-05, "loss": 0.9237, "step": 10162 }, { "epoch": 0.01, "learning_rate": 4.998750560048264e-05, "loss": 0.8929, "step": 10163 }, { "epoch": 0.01, "learning_rate": 4.998750311733065e-05, "loss": 0.9751, "step": 10164 }, { "epoch": 0.01, "learning_rate": 4.9987500633931996e-05, "loss": 1.2216, "step": 10165 }, { "epoch": 0.01, "learning_rate": 4.9987498150286674e-05, "loss": 1.19, "step": 10166 }, { "epoch": 0.01, "learning_rate": 4.998749566639469e-05, "loss": 1.2974, "step": 10167 }, { "epoch": 0.01, "learning_rate": 4.9987493182256035e-05, "loss": 1.2664, "step": 10168 }, { "epoch": 0.01, "learning_rate": 4.998749069787072e-05, "loss": 0.8888, "step": 10169 }, { "epoch": 0.01, "learning_rate": 4.998748821323873e-05, "loss": 0.9302, "step": 10170 }, { "epoch": 0.01, "learning_rate": 4.998748572836008e-05, "loss": 1.3394, "step": 10171 }, { "epoch": 0.01, "learning_rate": 4.998748324323477e-05, "loss": 0.7946, "step": 10172 }, { "epoch": 0.01, "learning_rate": 4.998748075786278e-05, "loss": 0.5738, "step": 10173 }, { "epoch": 0.01, "learning_rate": 4.9987478272244136e-05, "loss": 0.8554, "step": 10174 }, { "epoch": 0.01, "learning_rate": 4.9987475786378816e-05, "loss": 1.3842, "step": 10175 }, { "epoch": 0.01, "learning_rate": 4.998747330026684e-05, "loss": 1.4482, "step": 10176 }, { "epoch": 0.01, "learning_rate": 4.998747081390819e-05, "loss": 1.4754, "step": 10177 }, { "epoch": 0.01, "learning_rate": 4.9987468327302876e-05, "loss": 1.29, "step": 10178 }, { "epoch": 0.01, "learning_rate": 4.9987465840450905e-05, "loss": 1.1728, "step": 10179 }, { "epoch": 0.01, "learning_rate": 4.9987463353352256e-05, "loss": 1.3689, "step": 10180 }, { "epoch": 0.01, "learning_rate": 4.998746086600695e-05, "loss": 1.229, "step": 10181 }, { "epoch": 0.01, "learning_rate": 4.998745837841498e-05, "loss": 1.9158, "step": 10182 }, { "epoch": 0.01, "learning_rate": 4.998745589057634e-05, "loss": 1.06, "step": 10183 }, { "epoch": 0.01, "learning_rate": 4.998745340249103e-05, "loss": 1.0289, "step": 10184 }, { "epoch": 0.01, "learning_rate": 4.998745091415906e-05, "loss": 1.1272, "step": 10185 }, { "epoch": 0.01, "learning_rate": 4.998744842558042e-05, "loss": 1.09, "step": 10186 }, { "epoch": 0.01, "learning_rate": 4.998744593675512e-05, "loss": 1.127, "step": 10187 }, { "epoch": 0.01, "learning_rate": 4.998744344768315e-05, "loss": 1.1904, "step": 10188 }, { "epoch": 0.01, "learning_rate": 4.9987440958364516e-05, "loss": 0.9387, "step": 10189 }, { "epoch": 0.01, "learning_rate": 4.998743846879922e-05, "loss": 0.9157, "step": 10190 }, { "epoch": 0.01, "learning_rate": 4.9987435978987254e-05, "loss": 1.0315, "step": 10191 }, { "epoch": 0.01, "learning_rate": 4.9987433488928626e-05, "loss": 1.5297, "step": 10192 }, { "epoch": 0.01, "learning_rate": 4.998743099862333e-05, "loss": 1.3291, "step": 10193 }, { "epoch": 0.01, "learning_rate": 4.9987428508071367e-05, "loss": 1.2469, "step": 10194 }, { "epoch": 0.01, "learning_rate": 4.998742601727274e-05, "loss": 0.5735, "step": 10195 }, { "epoch": 0.01, "learning_rate": 4.998742352622745e-05, "loss": 1.1063, "step": 10196 }, { "epoch": 0.01, "learning_rate": 4.9987421034935494e-05, "loss": 1.3326, "step": 10197 }, { "epoch": 0.01, "learning_rate": 4.998741854339687e-05, "loss": 1.0274, "step": 10198 }, { "epoch": 0.01, "learning_rate": 4.998741605161158e-05, "loss": 1.5276, "step": 10199 }, { "epoch": 0.01, "learning_rate": 4.9987413559579636e-05, "loss": 1.4601, "step": 10200 }, { "epoch": 0.01, "learning_rate": 4.9987411067301015e-05, "loss": 0.9156, "step": 10201 }, { "epoch": 0.01, "learning_rate": 4.998740857477573e-05, "loss": 1.0569, "step": 10202 }, { "epoch": 0.01, "learning_rate": 4.9987406082003785e-05, "loss": 0.9294, "step": 10203 }, { "epoch": 0.01, "learning_rate": 4.998740358898517e-05, "loss": 1.1716, "step": 10204 }, { "epoch": 0.01, "learning_rate": 4.998740109571989e-05, "loss": 0.6784, "step": 10205 }, { "epoch": 0.01, "learning_rate": 4.998739860220795e-05, "loss": 0.723, "step": 10206 }, { "epoch": 0.01, "learning_rate": 4.998739610844934e-05, "loss": 0.7166, "step": 10207 }, { "epoch": 0.01, "learning_rate": 4.9987393614444066e-05, "loss": 0.6935, "step": 10208 }, { "epoch": 0.01, "learning_rate": 4.998739112019213e-05, "loss": 1.0554, "step": 10209 }, { "epoch": 0.01, "learning_rate": 4.998738862569352e-05, "loss": 1.0829, "step": 10210 }, { "epoch": 0.01, "learning_rate": 4.9987386130948255e-05, "loss": 0.9208, "step": 10211 }, { "epoch": 0.01, "learning_rate": 4.9987383635956317e-05, "loss": 1.1552, "step": 10212 }, { "epoch": 0.01, "learning_rate": 4.998738114071772e-05, "loss": 1.3641, "step": 10213 }, { "epoch": 0.01, "learning_rate": 4.998737864523246e-05, "loss": 0.9493, "step": 10214 }, { "epoch": 0.01, "learning_rate": 4.998737614950053e-05, "loss": 1.0372, "step": 10215 }, { "epoch": 0.01, "learning_rate": 4.9987373653521936e-05, "loss": 1.0349, "step": 10216 }, { "epoch": 0.01, "learning_rate": 4.9987371157296684e-05, "loss": 1.3293, "step": 10217 }, { "epoch": 0.01, "learning_rate": 4.998736866082475e-05, "loss": 1.1373, "step": 10218 }, { "epoch": 0.01, "learning_rate": 4.9987366164106164e-05, "loss": 1.2073, "step": 10219 }, { "epoch": 0.01, "learning_rate": 4.998736366714091e-05, "loss": 1.0581, "step": 10220 }, { "epoch": 0.01, "learning_rate": 4.9987361169929e-05, "loss": 1.7155, "step": 10221 }, { "epoch": 0.01, "learning_rate": 4.998735867247042e-05, "loss": 1.326, "step": 10222 }, { "epoch": 0.01, "learning_rate": 4.9987356174765165e-05, "loss": 1.0172, "step": 10223 }, { "epoch": 0.01, "learning_rate": 4.998735367681326e-05, "loss": 1.2729, "step": 10224 }, { "epoch": 0.01, "learning_rate": 4.9987351178614686e-05, "loss": 0.8779, "step": 10225 }, { "epoch": 0.01, "learning_rate": 4.998734868016944e-05, "loss": 1.1188, "step": 10226 }, { "epoch": 0.01, "learning_rate": 4.9987346181477535e-05, "loss": 1.2608, "step": 10227 }, { "epoch": 0.01, "learning_rate": 4.9987343682538965e-05, "loss": 1.058, "step": 10228 }, { "epoch": 0.01, "learning_rate": 4.998734118335373e-05, "loss": 1.0546, "step": 10229 }, { "epoch": 0.01, "learning_rate": 4.998733868392184e-05, "loss": 1.0555, "step": 10230 }, { "epoch": 0.01, "learning_rate": 4.998733618424327e-05, "loss": 1.0756, "step": 10231 }, { "epoch": 0.01, "learning_rate": 4.9987333684318045e-05, "loss": 2.1973, "step": 10232 }, { "epoch": 0.01, "learning_rate": 4.9987331184146155e-05, "loss": 1.1023, "step": 10233 }, { "epoch": 0.01, "learning_rate": 4.9987328683727595e-05, "loss": 1.2884, "step": 10234 }, { "epoch": 0.01, "learning_rate": 4.998732618306238e-05, "loss": 0.9701, "step": 10235 }, { "epoch": 0.01, "learning_rate": 4.998732368215049e-05, "loss": 1.4618, "step": 10236 }, { "epoch": 0.01, "learning_rate": 4.998732118099194e-05, "loss": 1.9832, "step": 10237 }, { "epoch": 0.01, "learning_rate": 4.998731867958673e-05, "loss": 1.0656, "step": 10238 }, { "epoch": 0.01, "learning_rate": 4.998731617793485e-05, "loss": 0.9133, "step": 10239 }, { "epoch": 0.01, "learning_rate": 4.9987313676036305e-05, "loss": 0.7601, "step": 10240 }, { "epoch": 0.01, "learning_rate": 4.9987311173891105e-05, "loss": 0.8458, "step": 10241 }, { "epoch": 0.01, "learning_rate": 4.998730867149924e-05, "loss": 1.1284, "step": 10242 }, { "epoch": 0.01, "learning_rate": 4.998730616886069e-05, "loss": 0.9589, "step": 10243 }, { "epoch": 0.01, "learning_rate": 4.998730366597549e-05, "loss": 0.9388, "step": 10244 }, { "epoch": 0.01, "learning_rate": 4.9987301162843634e-05, "loss": 0.2874, "step": 10245 }, { "epoch": 0.01, "learning_rate": 4.998729865946511e-05, "loss": 0.2173, "step": 10246 }, { "epoch": 0.01, "learning_rate": 4.998729615583991e-05, "loss": 0.2438, "step": 10247 }, { "epoch": 0.01, "learning_rate": 4.998729365196806e-05, "loss": 0.173, "step": 10248 }, { "epoch": 0.01, "learning_rate": 4.9987291147849544e-05, "loss": 0.2369, "step": 10249 }, { "epoch": 0.01, "learning_rate": 4.998728864348435e-05, "loss": 0.2054, "step": 10250 }, { "epoch": 0.01, "learning_rate": 4.998728613887251e-05, "loss": 0.205, "step": 10251 }, { "epoch": 0.01, "learning_rate": 4.9987283634013995e-05, "loss": 0.5671, "step": 10252 }, { "epoch": 0.01, "learning_rate": 4.998728112890882e-05, "loss": 0.6497, "step": 10253 }, { "epoch": 0.01, "learning_rate": 4.998727862355698e-05, "loss": 0.658, "step": 10254 }, { "epoch": 0.01, "learning_rate": 4.998727611795848e-05, "loss": 0.645, "step": 10255 }, { "epoch": 0.01, "learning_rate": 4.9987273612113314e-05, "loss": 0.2322, "step": 10256 }, { "epoch": 0.01, "learning_rate": 4.998727110602148e-05, "loss": 0.1367, "step": 10257 }, { "epoch": 0.01, "learning_rate": 4.998726859968299e-05, "loss": 0.352, "step": 10258 }, { "epoch": 0.01, "learning_rate": 4.998726609309783e-05, "loss": 1.0021, "step": 10259 }, { "epoch": 0.01, "learning_rate": 4.998726358626601e-05, "loss": 1.5276, "step": 10260 }, { "epoch": 0.01, "learning_rate": 4.998726107918752e-05, "loss": 1.2101, "step": 10261 }, { "epoch": 0.01, "learning_rate": 4.9987258571862375e-05, "loss": 1.2582, "step": 10262 }, { "epoch": 0.01, "learning_rate": 4.9987256064290554e-05, "loss": 1.0663, "step": 10263 }, { "epoch": 0.01, "learning_rate": 4.998725355647208e-05, "loss": 1.2152, "step": 10264 }, { "epoch": 0.01, "learning_rate": 4.9987251048406945e-05, "loss": 1.0573, "step": 10265 }, { "epoch": 0.01, "learning_rate": 4.9987248540095135e-05, "loss": 1.2546, "step": 10266 }, { "epoch": 0.01, "learning_rate": 4.998724603153667e-05, "loss": 1.1061, "step": 10267 }, { "epoch": 0.01, "learning_rate": 4.998724352273154e-05, "loss": 1.0008, "step": 10268 }, { "epoch": 0.01, "learning_rate": 4.9987241013679745e-05, "loss": 1.1719, "step": 10269 }, { "epoch": 0.01, "learning_rate": 4.9987238504381285e-05, "loss": 0.9274, "step": 10270 }, { "epoch": 0.01, "learning_rate": 4.998723599483616e-05, "loss": 1.1606, "step": 10271 }, { "epoch": 0.01, "learning_rate": 4.998723348504438e-05, "loss": 1.1728, "step": 10272 }, { "epoch": 0.01, "learning_rate": 4.998723097500593e-05, "loss": 1.9997, "step": 10273 }, { "epoch": 0.01, "learning_rate": 4.9987228464720816e-05, "loss": 1.7191, "step": 10274 }, { "epoch": 0.01, "learning_rate": 4.998722595418904e-05, "loss": 1.1818, "step": 10275 }, { "epoch": 0.01, "learning_rate": 4.9987223443410605e-05, "loss": 1.0861, "step": 10276 }, { "epoch": 0.01, "learning_rate": 4.9987220932385506e-05, "loss": 0.8447, "step": 10277 }, { "epoch": 0.01, "learning_rate": 4.9987218421113737e-05, "loss": 0.4378, "step": 10278 }, { "epoch": 0.01, "learning_rate": 4.998721590959531e-05, "loss": 0.3998, "step": 10279 }, { "epoch": 0.01, "learning_rate": 4.998721339783021e-05, "loss": 0.6704, "step": 10280 }, { "epoch": 0.01, "learning_rate": 4.998721088581846e-05, "loss": 1.2137, "step": 10281 }, { "epoch": 0.01, "learning_rate": 4.998720837356004e-05, "loss": 0.7144, "step": 10282 }, { "epoch": 0.01, "learning_rate": 4.9987205861054956e-05, "loss": 1.0599, "step": 10283 }, { "epoch": 0.01, "learning_rate": 4.998720334830321e-05, "loss": 0.772, "step": 10284 }, { "epoch": 0.01, "learning_rate": 4.99872008353048e-05, "loss": 0.9947, "step": 10285 }, { "epoch": 0.01, "learning_rate": 4.998719832205974e-05, "loss": 1.2363, "step": 10286 }, { "epoch": 0.01, "learning_rate": 4.9987195808567996e-05, "loss": 0.9215, "step": 10287 }, { "epoch": 0.01, "learning_rate": 4.9987193294829606e-05, "loss": 1.3128, "step": 10288 }, { "epoch": 0.01, "learning_rate": 4.998719078084454e-05, "loss": 1.0097, "step": 10289 }, { "epoch": 0.01, "learning_rate": 4.998718826661282e-05, "loss": 1.1504, "step": 10290 }, { "epoch": 0.01, "learning_rate": 4.998718575213443e-05, "loss": 1.2552, "step": 10291 }, { "epoch": 0.01, "learning_rate": 4.9987183237409384e-05, "loss": 1.0095, "step": 10292 }, { "epoch": 0.01, "learning_rate": 4.9987180722437674e-05, "loss": 1.0155, "step": 10293 }, { "epoch": 0.01, "learning_rate": 4.99871782072193e-05, "loss": 1.1331, "step": 10294 }, { "epoch": 0.01, "learning_rate": 4.998717569175426e-05, "loss": 1.3524, "step": 10295 }, { "epoch": 0.01, "learning_rate": 4.998717317604256e-05, "loss": 1.3273, "step": 10296 }, { "epoch": 0.01, "learning_rate": 4.998717066008419e-05, "loss": 1.216, "step": 10297 }, { "epoch": 0.01, "learning_rate": 4.9987168143879165e-05, "loss": 1.551, "step": 10298 }, { "epoch": 0.01, "learning_rate": 4.998716562742748e-05, "loss": 0.6469, "step": 10299 }, { "epoch": 0.01, "learning_rate": 4.998716311072912e-05, "loss": 0.8778, "step": 10300 }, { "epoch": 0.01, "learning_rate": 4.998716059378411e-05, "loss": 1.2854, "step": 10301 }, { "epoch": 0.01, "learning_rate": 4.998715807659243e-05, "loss": 1.2548, "step": 10302 }, { "epoch": 0.01, "learning_rate": 4.998715555915409e-05, "loss": 0.8318, "step": 10303 }, { "epoch": 0.01, "learning_rate": 4.9987153041469086e-05, "loss": 0.7398, "step": 10304 }, { "epoch": 0.01, "learning_rate": 4.9987150523537415e-05, "loss": 1.1692, "step": 10305 }, { "epoch": 0.01, "learning_rate": 4.9987148005359095e-05, "loss": 0.6863, "step": 10306 }, { "epoch": 0.01, "learning_rate": 4.9987145486934106e-05, "loss": 1.3325, "step": 10307 }, { "epoch": 0.01, "learning_rate": 4.998714296826245e-05, "loss": 1.3931, "step": 10308 }, { "epoch": 0.01, "learning_rate": 4.998714044934413e-05, "loss": 1.1937, "step": 10309 }, { "epoch": 0.01, "learning_rate": 4.9987137930179154e-05, "loss": 1.1837, "step": 10310 }, { "epoch": 0.01, "learning_rate": 4.998713541076751e-05, "loss": 1.5333, "step": 10311 }, { "epoch": 0.01, "learning_rate": 4.998713289110921e-05, "loss": 0.6901, "step": 10312 }, { "epoch": 0.01, "learning_rate": 4.9987130371204244e-05, "loss": 1.7999, "step": 10313 }, { "epoch": 0.01, "learning_rate": 4.9987127851052615e-05, "loss": 1.1222, "step": 10314 }, { "epoch": 0.01, "learning_rate": 4.998712533065433e-05, "loss": 1.2129, "step": 10315 }, { "epoch": 0.01, "learning_rate": 4.998712281000937e-05, "loss": 1.3231, "step": 10316 }, { "epoch": 0.01, "learning_rate": 4.998712028911776e-05, "loss": 1.2286, "step": 10317 }, { "epoch": 0.01, "learning_rate": 4.998711776797947e-05, "loss": 1.1108, "step": 10318 }, { "epoch": 0.01, "learning_rate": 4.998711524659454e-05, "loss": 1.0662, "step": 10319 }, { "epoch": 0.01, "learning_rate": 4.998711272496294e-05, "loss": 0.9464, "step": 10320 }, { "epoch": 0.01, "learning_rate": 4.998711020308467e-05, "loss": 0.9488, "step": 10321 }, { "epoch": 0.01, "learning_rate": 4.9987107680959745e-05, "loss": 1.0501, "step": 10322 }, { "epoch": 0.01, "learning_rate": 4.998710515858816e-05, "loss": 1.2066, "step": 10323 }, { "epoch": 0.01, "learning_rate": 4.998710263596991e-05, "loss": 1.1524, "step": 10324 }, { "epoch": 0.01, "learning_rate": 4.998710011310499e-05, "loss": 1.0147, "step": 10325 }, { "epoch": 0.01, "learning_rate": 4.998709758999341e-05, "loss": 1.0825, "step": 10326 }, { "epoch": 0.01, "learning_rate": 4.9987095066635175e-05, "loss": 1.1649, "step": 10327 }, { "epoch": 0.01, "learning_rate": 4.998709254303028e-05, "loss": 1.1194, "step": 10328 }, { "epoch": 0.01, "learning_rate": 4.998709001917872e-05, "loss": 1.1093, "step": 10329 }, { "epoch": 0.01, "learning_rate": 4.9987087495080484e-05, "loss": 1.2593, "step": 10330 }, { "epoch": 0.01, "learning_rate": 4.998708497073561e-05, "loss": 1.1496, "step": 10331 }, { "epoch": 0.01, "learning_rate": 4.9987082446144056e-05, "loss": 1.1416, "step": 10332 }, { "epoch": 0.01, "learning_rate": 4.998707992130585e-05, "loss": 1.196, "step": 10333 }, { "epoch": 0.01, "learning_rate": 4.998707739622098e-05, "loss": 0.9892, "step": 10334 }, { "epoch": 0.01, "learning_rate": 4.998707487088944e-05, "loss": 1.2026, "step": 10335 }, { "epoch": 0.01, "learning_rate": 4.998707234531125e-05, "loss": 0.8014, "step": 10336 }, { "epoch": 0.01, "learning_rate": 4.998706981948639e-05, "loss": 2.2308, "step": 10337 }, { "epoch": 0.01, "learning_rate": 4.998706729341487e-05, "loss": 1.0938, "step": 10338 }, { "epoch": 0.01, "learning_rate": 4.998706476709669e-05, "loss": 0.7971, "step": 10339 }, { "epoch": 0.01, "learning_rate": 4.998706224053185e-05, "loss": 1.2027, "step": 10340 }, { "epoch": 0.01, "learning_rate": 4.998705971372034e-05, "loss": 1.139, "step": 10341 }, { "epoch": 0.01, "learning_rate": 4.998705718666218e-05, "loss": 0.6991, "step": 10342 }, { "epoch": 0.01, "learning_rate": 4.998705465935734e-05, "loss": 1.3724, "step": 10343 }, { "epoch": 0.01, "learning_rate": 4.998705213180586e-05, "loss": 1.1193, "step": 10344 }, { "epoch": 0.01, "learning_rate": 4.998704960400771e-05, "loss": 1.005, "step": 10345 }, { "epoch": 0.01, "learning_rate": 4.9987047075962893e-05, "loss": 1.098, "step": 10346 }, { "epoch": 0.01, "learning_rate": 4.998704454767143e-05, "loss": 1.116, "step": 10347 }, { "epoch": 0.01, "learning_rate": 4.9987042019133284e-05, "loss": 1.0914, "step": 10348 }, { "epoch": 0.01, "learning_rate": 4.9987039490348485e-05, "loss": 1.8293, "step": 10349 }, { "epoch": 0.01, "learning_rate": 4.998703696131703e-05, "loss": 1.2251, "step": 10350 }, { "epoch": 0.01, "learning_rate": 4.9987034432038906e-05, "loss": 1.2625, "step": 10351 }, { "epoch": 0.01, "learning_rate": 4.9987031902514126e-05, "loss": 1.3011, "step": 10352 }, { "epoch": 0.01, "learning_rate": 4.998702937274268e-05, "loss": 0.5868, "step": 10353 }, { "epoch": 0.01, "learning_rate": 4.998702684272458e-05, "loss": 0.522, "step": 10354 }, { "epoch": 0.01, "learning_rate": 4.9987024312459815e-05, "loss": 0.915, "step": 10355 }, { "epoch": 0.01, "learning_rate": 4.9987021781948384e-05, "loss": 1.1415, "step": 10356 }, { "epoch": 0.01, "learning_rate": 4.998701925119029e-05, "loss": 1.3289, "step": 10357 }, { "epoch": 0.01, "learning_rate": 4.998701672018554e-05, "loss": 1.1288, "step": 10358 }, { "epoch": 0.01, "learning_rate": 4.9987014188934134e-05, "loss": 1.0588, "step": 10359 }, { "epoch": 0.01, "learning_rate": 4.998701165743606e-05, "loss": 1.0426, "step": 10360 }, { "epoch": 0.01, "learning_rate": 4.9987009125691326e-05, "loss": 1.0733, "step": 10361 }, { "epoch": 0.01, "learning_rate": 4.998700659369993e-05, "loss": 0.9528, "step": 10362 }, { "epoch": 0.01, "learning_rate": 4.9987004061461875e-05, "loss": 1.1307, "step": 10363 }, { "epoch": 0.01, "learning_rate": 4.9987001528977155e-05, "loss": 1.1637, "step": 10364 }, { "epoch": 0.01, "learning_rate": 4.998699899624578e-05, "loss": 1.2464, "step": 10365 }, { "epoch": 0.01, "learning_rate": 4.998699646326774e-05, "loss": 0.776, "step": 10366 }, { "epoch": 0.01, "learning_rate": 4.998699393004304e-05, "loss": 0.906, "step": 10367 }, { "epoch": 0.01, "learning_rate": 4.9986991396571674e-05, "loss": 0.9508, "step": 10368 }, { "epoch": 0.01, "learning_rate": 4.9986988862853654e-05, "loss": 1.0688, "step": 10369 }, { "epoch": 0.01, "learning_rate": 4.9986986328888964e-05, "loss": 1.1141, "step": 10370 }, { "epoch": 0.01, "learning_rate": 4.9986983794677625e-05, "loss": 0.9474, "step": 10371 }, { "epoch": 0.01, "learning_rate": 4.998698126021961e-05, "loss": 0.6913, "step": 10372 }, { "epoch": 0.01, "learning_rate": 4.9986978725514945e-05, "loss": 0.4359, "step": 10373 }, { "epoch": 0.01, "learning_rate": 4.998697619056362e-05, "loss": 1.1969, "step": 10374 }, { "epoch": 0.01, "learning_rate": 4.9986973655365634e-05, "loss": 1.2375, "step": 10375 }, { "epoch": 0.01, "learning_rate": 4.9986971119920975e-05, "loss": 1.0497, "step": 10376 }, { "epoch": 0.01, "learning_rate": 4.998696858422967e-05, "loss": 0.9796, "step": 10377 }, { "epoch": 0.01, "learning_rate": 4.9986966048291694e-05, "loss": 1.0029, "step": 10378 }, { "epoch": 0.01, "learning_rate": 4.9986963512107067e-05, "loss": 1.1816, "step": 10379 }, { "epoch": 0.01, "learning_rate": 4.9986960975675776e-05, "loss": 1.1951, "step": 10380 }, { "epoch": 0.01, "learning_rate": 4.998695843899782e-05, "loss": 1.143, "step": 10381 }, { "epoch": 0.01, "learning_rate": 4.998695590207321e-05, "loss": 1.0644, "step": 10382 }, { "epoch": 0.01, "learning_rate": 4.998695336490193e-05, "loss": 1.1941, "step": 10383 }, { "epoch": 0.01, "learning_rate": 4.9986950827483994e-05, "loss": 1.1132, "step": 10384 }, { "epoch": 0.01, "learning_rate": 4.9986948289819396e-05, "loss": 0.549, "step": 10385 }, { "epoch": 0.01, "learning_rate": 4.998694575190814e-05, "loss": 1.0158, "step": 10386 }, { "epoch": 0.01, "learning_rate": 4.998694321375022e-05, "loss": 1.3903, "step": 10387 }, { "epoch": 0.01, "learning_rate": 4.998694067534564e-05, "loss": 1.4583, "step": 10388 }, { "epoch": 0.01, "learning_rate": 4.9986938136694405e-05, "loss": 1.1652, "step": 10389 }, { "epoch": 0.01, "learning_rate": 4.99869355977965e-05, "loss": 1.2458, "step": 10390 }, { "epoch": 0.01, "learning_rate": 4.998693305865194e-05, "loss": 1.2432, "step": 10391 }, { "epoch": 0.01, "learning_rate": 4.9986930519260724e-05, "loss": 1.3848, "step": 10392 }, { "epoch": 0.01, "learning_rate": 4.9986927979622844e-05, "loss": 1.0191, "step": 10393 }, { "epoch": 0.01, "learning_rate": 4.99869254397383e-05, "loss": 0.9536, "step": 10394 }, { "epoch": 0.01, "learning_rate": 4.9986922899607104e-05, "loss": 1.1373, "step": 10395 }, { "epoch": 0.01, "learning_rate": 4.9986920359229237e-05, "loss": 1.2518, "step": 10396 }, { "epoch": 0.01, "learning_rate": 4.998691781860472e-05, "loss": 1.3977, "step": 10397 }, { "epoch": 0.01, "learning_rate": 4.998691527773353e-05, "loss": 1.0759, "step": 10398 }, { "epoch": 0.01, "learning_rate": 4.998691273661569e-05, "loss": 1.6515, "step": 10399 }, { "epoch": 0.01, "learning_rate": 4.9986910195251186e-05, "loss": 1.5402, "step": 10400 }, { "epoch": 0.01, "learning_rate": 4.9986907653640025e-05, "loss": 1.5662, "step": 10401 }, { "epoch": 0.01, "learning_rate": 4.99869051117822e-05, "loss": 1.4326, "step": 10402 }, { "epoch": 0.01, "learning_rate": 4.9986902569677714e-05, "loss": 1.3334, "step": 10403 }, { "epoch": 0.01, "learning_rate": 4.9986900027326564e-05, "loss": 1.2097, "step": 10404 }, { "epoch": 0.01, "learning_rate": 4.9986897484728766e-05, "loss": 1.2671, "step": 10405 }, { "epoch": 0.01, "learning_rate": 4.99868949418843e-05, "loss": 1.1214, "step": 10406 }, { "epoch": 0.01, "learning_rate": 4.998689239879317e-05, "loss": 1.349, "step": 10407 }, { "epoch": 0.01, "learning_rate": 4.9986889855455386e-05, "loss": 1.3452, "step": 10408 }, { "epoch": 0.01, "learning_rate": 4.998688731187094e-05, "loss": 1.3616, "step": 10409 }, { "epoch": 0.01, "learning_rate": 4.9986884768039844e-05, "loss": 1.3294, "step": 10410 }, { "epoch": 0.01, "learning_rate": 4.9986882223962076e-05, "loss": 1.2834, "step": 10411 }, { "epoch": 0.01, "learning_rate": 4.998687967963766e-05, "loss": 1.1846, "step": 10412 }, { "epoch": 0.01, "learning_rate": 4.998687713506657e-05, "loss": 1.2795, "step": 10413 }, { "epoch": 0.01, "learning_rate": 4.998687459024882e-05, "loss": 1.1784, "step": 10414 }, { "epoch": 0.01, "learning_rate": 4.998687204518442e-05, "loss": 1.3204, "step": 10415 }, { "epoch": 0.01, "learning_rate": 4.998686949987336e-05, "loss": 1.1975, "step": 10416 }, { "epoch": 0.01, "learning_rate": 4.9986866954315634e-05, "loss": 1.1086, "step": 10417 }, { "epoch": 0.01, "learning_rate": 4.998686440851125e-05, "loss": 1.285, "step": 10418 }, { "epoch": 0.01, "learning_rate": 4.9986861862460196e-05, "loss": 1.3251, "step": 10419 }, { "epoch": 0.01, "learning_rate": 4.99868593161625e-05, "loss": 0.9713, "step": 10420 }, { "epoch": 0.01, "learning_rate": 4.9986856769618134e-05, "loss": 1.2093, "step": 10421 }, { "epoch": 0.01, "learning_rate": 4.998685422282711e-05, "loss": 0.9838, "step": 10422 }, { "epoch": 0.01, "learning_rate": 4.998685167578943e-05, "loss": 1.0117, "step": 10423 }, { "epoch": 0.01, "learning_rate": 4.9986849128505084e-05, "loss": 1.0491, "step": 10424 }, { "epoch": 0.01, "learning_rate": 4.998684658097409e-05, "loss": 1.4273, "step": 10425 }, { "epoch": 0.01, "learning_rate": 4.998684403319642e-05, "loss": 1.3291, "step": 10426 }, { "epoch": 0.01, "learning_rate": 4.99868414851721e-05, "loss": 1.0863, "step": 10427 }, { "epoch": 0.01, "learning_rate": 4.998683893690112e-05, "loss": 0.884, "step": 10428 }, { "epoch": 0.01, "learning_rate": 4.9986836388383484e-05, "loss": 1.0212, "step": 10429 }, { "epoch": 0.01, "learning_rate": 4.9986833839619184e-05, "loss": 1.1829, "step": 10430 }, { "epoch": 0.01, "learning_rate": 4.998683129060822e-05, "loss": 0.7684, "step": 10431 }, { "epoch": 0.01, "learning_rate": 4.99868287413506e-05, "loss": 1.2292, "step": 10432 }, { "epoch": 0.01, "learning_rate": 4.998682619184633e-05, "loss": 1.5942, "step": 10433 }, { "epoch": 0.01, "learning_rate": 4.998682364209539e-05, "loss": 3.0241, "step": 10434 }, { "epoch": 0.01, "learning_rate": 4.998682109209779e-05, "loss": 1.0854, "step": 10435 }, { "epoch": 0.01, "learning_rate": 4.998681854185353e-05, "loss": 1.2607, "step": 10436 }, { "epoch": 0.01, "learning_rate": 4.998681599136262e-05, "loss": 1.3714, "step": 10437 }, { "epoch": 0.01, "learning_rate": 4.998681344062504e-05, "loss": 1.0285, "step": 10438 }, { "epoch": 0.01, "learning_rate": 4.998681088964081e-05, "loss": 1.0756, "step": 10439 }, { "epoch": 0.01, "learning_rate": 4.998680833840991e-05, "loss": 1.2608, "step": 10440 }, { "epoch": 0.01, "learning_rate": 4.998680578693236e-05, "loss": 1.3557, "step": 10441 }, { "epoch": 0.01, "learning_rate": 4.998680323520814e-05, "loss": 1.3043, "step": 10442 }, { "epoch": 0.01, "learning_rate": 4.998680068323728e-05, "loss": 0.7495, "step": 10443 }, { "epoch": 0.01, "learning_rate": 4.998679813101975e-05, "loss": 1.3086, "step": 10444 }, { "epoch": 0.01, "learning_rate": 4.9986795578555555e-05, "loss": 1.6532, "step": 10445 }, { "epoch": 0.01, "learning_rate": 4.998679302584471e-05, "loss": 1.1917, "step": 10446 }, { "epoch": 0.01, "learning_rate": 4.99867904728872e-05, "loss": 1.2263, "step": 10447 }, { "epoch": 0.01, "learning_rate": 4.998678791968303e-05, "loss": 1.052, "step": 10448 }, { "epoch": 0.01, "learning_rate": 4.998678536623221e-05, "loss": 1.1604, "step": 10449 }, { "epoch": 0.01, "learning_rate": 4.998678281253472e-05, "loss": 1.1795, "step": 10450 }, { "epoch": 0.01, "learning_rate": 4.998678025859058e-05, "loss": 1.3203, "step": 10451 }, { "epoch": 0.01, "learning_rate": 4.998677770439978e-05, "loss": 1.0137, "step": 10452 }, { "epoch": 0.01, "learning_rate": 4.998677514996232e-05, "loss": 1.2603, "step": 10453 }, { "epoch": 0.01, "learning_rate": 4.998677259527819e-05, "loss": 1.1167, "step": 10454 }, { "epoch": 0.01, "learning_rate": 4.998677004034742e-05, "loss": 1.2954, "step": 10455 }, { "epoch": 0.01, "learning_rate": 4.998676748516997e-05, "loss": 1.1379, "step": 10456 }, { "epoch": 0.01, "learning_rate": 4.998676492974588e-05, "loss": 1.1313, "step": 10457 }, { "epoch": 0.01, "learning_rate": 4.998676237407512e-05, "loss": 1.1246, "step": 10458 }, { "epoch": 0.01, "learning_rate": 4.9986759818157704e-05, "loss": 0.5331, "step": 10459 }, { "epoch": 0.01, "learning_rate": 4.9986757261993636e-05, "loss": 0.9279, "step": 10460 }, { "epoch": 0.01, "learning_rate": 4.99867547055829e-05, "loss": 1.4327, "step": 10461 }, { "epoch": 0.01, "learning_rate": 4.998675214892551e-05, "loss": 1.1658, "step": 10462 }, { "epoch": 0.01, "learning_rate": 4.998674959202146e-05, "loss": 0.872, "step": 10463 }, { "epoch": 0.01, "learning_rate": 4.998674703487075e-05, "loss": 2.5678, "step": 10464 }, { "epoch": 0.01, "learning_rate": 4.998674447747338e-05, "loss": 1.6675, "step": 10465 }, { "epoch": 0.01, "learning_rate": 4.9986741919829363e-05, "loss": 1.2018, "step": 10466 }, { "epoch": 0.01, "learning_rate": 4.998673936193867e-05, "loss": 1.1596, "step": 10467 }, { "epoch": 0.01, "learning_rate": 4.9986736803801326e-05, "loss": 0.9724, "step": 10468 }, { "epoch": 0.01, "learning_rate": 4.998673424541733e-05, "loss": 1.0377, "step": 10469 }, { "epoch": 0.01, "learning_rate": 4.998673168678667e-05, "loss": 0.9609, "step": 10470 }, { "epoch": 0.01, "learning_rate": 4.998672912790935e-05, "loss": 1.1886, "step": 10471 }, { "epoch": 0.01, "learning_rate": 4.9986726568785367e-05, "loss": 1.1853, "step": 10472 }, { "epoch": 0.01, "learning_rate": 4.998672400941474e-05, "loss": 0.8812, "step": 10473 }, { "epoch": 0.01, "learning_rate": 4.9986721449797444e-05, "loss": 0.8519, "step": 10474 }, { "epoch": 0.01, "learning_rate": 4.9986718889933495e-05, "loss": 1.2709, "step": 10475 }, { "epoch": 0.01, "learning_rate": 4.998671632982288e-05, "loss": 1.3796, "step": 10476 }, { "epoch": 0.01, "learning_rate": 4.998671376946561e-05, "loss": 1.5551, "step": 10477 }, { "epoch": 0.01, "learning_rate": 4.998671120886168e-05, "loss": 1.0944, "step": 10478 }, { "epoch": 0.01, "learning_rate": 4.99867086480111e-05, "loss": 1.1696, "step": 10479 }, { "epoch": 0.01, "learning_rate": 4.998670608691385e-05, "loss": 1.1202, "step": 10480 }, { "epoch": 0.01, "learning_rate": 4.998670352556995e-05, "loss": 1.2514, "step": 10481 }, { "epoch": 0.01, "learning_rate": 4.998670096397939e-05, "loss": 0.9168, "step": 10482 }, { "epoch": 0.01, "learning_rate": 4.9986698402142175e-05, "loss": 1.1231, "step": 10483 }, { "epoch": 0.01, "learning_rate": 4.9986695840058295e-05, "loss": 0.9077, "step": 10484 }, { "epoch": 0.01, "learning_rate": 4.998669327772776e-05, "loss": 1.162, "step": 10485 }, { "epoch": 0.01, "learning_rate": 4.998669071515057e-05, "loss": 1.3504, "step": 10486 }, { "epoch": 0.01, "learning_rate": 4.998668815232672e-05, "loss": 1.1128, "step": 10487 }, { "epoch": 0.01, "learning_rate": 4.998668558925621e-05, "loss": 0.9506, "step": 10488 }, { "epoch": 0.01, "learning_rate": 4.998668302593904e-05, "loss": 0.9646, "step": 10489 }, { "epoch": 0.01, "learning_rate": 4.998668046237522e-05, "loss": 0.9416, "step": 10490 }, { "epoch": 0.01, "learning_rate": 4.9986677898564735e-05, "loss": 0.8455, "step": 10491 }, { "epoch": 0.01, "learning_rate": 4.9986675334507594e-05, "loss": 0.8593, "step": 10492 }, { "epoch": 0.01, "learning_rate": 4.9986672770203797e-05, "loss": 0.7396, "step": 10493 }, { "epoch": 0.01, "learning_rate": 4.998667020565334e-05, "loss": 0.7632, "step": 10494 }, { "epoch": 0.01, "learning_rate": 4.998666764085622e-05, "loss": 1.1413, "step": 10495 }, { "epoch": 0.01, "learning_rate": 4.9986665075812455e-05, "loss": 0.995, "step": 10496 }, { "epoch": 0.01, "learning_rate": 4.998666251052202e-05, "loss": 1.4334, "step": 10497 }, { "epoch": 0.01, "learning_rate": 4.9986659944984937e-05, "loss": 0.8874, "step": 10498 }, { "epoch": 0.01, "learning_rate": 4.998665737920118e-05, "loss": 1.0983, "step": 10499 }, { "epoch": 0.01, "learning_rate": 4.998665481317079e-05, "loss": 1.3465, "step": 10500 }, { "epoch": 0.01, "eval_loss": 1.0590955018997192, "eval_runtime": 83.8028, "eval_samples_per_second": 16.527, "eval_steps_per_second": 4.141, "step": 10500 }, { "epoch": 0.01, "learning_rate": 4.998665224689372e-05, "loss": 0.8434, "step": 10501 }, { "epoch": 0.01, "learning_rate": 4.998664968037e-05, "loss": 0.758, "step": 10502 }, { "epoch": 0.01, "learning_rate": 4.998664711359963e-05, "loss": 0.8072, "step": 10503 }, { "epoch": 0.01, "learning_rate": 4.9986644546582596e-05, "loss": 1.0505, "step": 10504 }, { "epoch": 0.01, "learning_rate": 4.99866419793189e-05, "loss": 1.1189, "step": 10505 }, { "epoch": 0.01, "learning_rate": 4.998663941180855e-05, "loss": 1.2439, "step": 10506 }, { "epoch": 0.01, "learning_rate": 4.9986636844051546e-05, "loss": 1.1577, "step": 10507 }, { "epoch": 0.01, "learning_rate": 4.9986634276047875e-05, "loss": 1.0675, "step": 10508 }, { "epoch": 0.01, "learning_rate": 4.998663170779756e-05, "loss": 1.2374, "step": 10509 }, { "epoch": 0.01, "learning_rate": 4.998662913930058e-05, "loss": 1.16, "step": 10510 }, { "epoch": 0.01, "learning_rate": 4.998662657055694e-05, "loss": 1.1783, "step": 10511 }, { "epoch": 0.01, "learning_rate": 4.9986624001566644e-05, "loss": 1.1148, "step": 10512 }, { "epoch": 0.01, "learning_rate": 4.998662143232969e-05, "loss": 1.207, "step": 10513 }, { "epoch": 0.01, "learning_rate": 4.998661886284608e-05, "loss": 1.1054, "step": 10514 }, { "epoch": 0.01, "learning_rate": 4.9986616293115816e-05, "loss": 1.2763, "step": 10515 }, { "epoch": 0.01, "learning_rate": 4.998661372313889e-05, "loss": 0.8169, "step": 10516 }, { "epoch": 0.01, "learning_rate": 4.998661115291531e-05, "loss": 1.0664, "step": 10517 }, { "epoch": 0.01, "learning_rate": 4.998660858244507e-05, "loss": 1.1742, "step": 10518 }, { "epoch": 0.01, "learning_rate": 4.998660601172818e-05, "loss": 0.9612, "step": 10519 }, { "epoch": 0.01, "learning_rate": 4.998660344076462e-05, "loss": 1.2785, "step": 10520 }, { "epoch": 0.01, "learning_rate": 4.9986600869554415e-05, "loss": 1.2946, "step": 10521 }, { "epoch": 0.01, "learning_rate": 4.998659829809754e-05, "loss": 1.2146, "step": 10522 }, { "epoch": 0.01, "learning_rate": 4.998659572639402e-05, "loss": 0.9512, "step": 10523 }, { "epoch": 0.01, "learning_rate": 4.998659315444384e-05, "loss": 1.1009, "step": 10524 }, { "epoch": 0.01, "learning_rate": 4.9986590582247e-05, "loss": 1.1699, "step": 10525 }, { "epoch": 0.01, "learning_rate": 4.9986588009803506e-05, "loss": 1.2749, "step": 10526 }, { "epoch": 0.01, "learning_rate": 4.998658543711335e-05, "loss": 1.187, "step": 10527 }, { "epoch": 0.01, "learning_rate": 4.998658286417653e-05, "loss": 1.0756, "step": 10528 }, { "epoch": 0.01, "learning_rate": 4.9986580290993065e-05, "loss": 1.1111, "step": 10529 }, { "epoch": 0.01, "learning_rate": 4.998657771756294e-05, "loss": 1.0458, "step": 10530 }, { "epoch": 0.01, "learning_rate": 4.998657514388616e-05, "loss": 0.9649, "step": 10531 }, { "epoch": 0.01, "learning_rate": 4.998657256996273e-05, "loss": 0.9339, "step": 10532 }, { "epoch": 0.01, "learning_rate": 4.998656999579263e-05, "loss": 0.6619, "step": 10533 }, { "epoch": 0.01, "learning_rate": 4.998656742137588e-05, "loss": 0.5385, "step": 10534 }, { "epoch": 0.01, "learning_rate": 4.998656484671247e-05, "loss": 1.0374, "step": 10535 }, { "epoch": 0.01, "learning_rate": 4.9986562271802406e-05, "loss": 0.5005, "step": 10536 }, { "epoch": 0.01, "learning_rate": 4.998655969664568e-05, "loss": 0.3531, "step": 10537 }, { "epoch": 0.01, "learning_rate": 4.99865571212423e-05, "loss": 0.5739, "step": 10538 }, { "epoch": 0.01, "learning_rate": 4.9986554545592266e-05, "loss": 0.689, "step": 10539 }, { "epoch": 0.01, "learning_rate": 4.998655196969557e-05, "loss": 1.1641, "step": 10540 }, { "epoch": 0.01, "learning_rate": 4.998654939355222e-05, "loss": 0.9358, "step": 10541 }, { "epoch": 0.01, "learning_rate": 4.9986546817162214e-05, "loss": 1.1991, "step": 10542 }, { "epoch": 0.01, "learning_rate": 4.9986544240525556e-05, "loss": 1.6141, "step": 10543 }, { "epoch": 0.01, "learning_rate": 4.9986541663642236e-05, "loss": 1.8724, "step": 10544 }, { "epoch": 0.01, "learning_rate": 4.998653908651226e-05, "loss": 1.3541, "step": 10545 }, { "epoch": 0.01, "learning_rate": 4.9986536509135626e-05, "loss": 1.0405, "step": 10546 }, { "epoch": 0.01, "learning_rate": 4.998653393151234e-05, "loss": 0.9787, "step": 10547 }, { "epoch": 0.01, "learning_rate": 4.9986531353642386e-05, "loss": 0.9666, "step": 10548 }, { "epoch": 0.01, "learning_rate": 4.9986528775525785e-05, "loss": 0.9589, "step": 10549 }, { "epoch": 0.01, "learning_rate": 4.998652619716253e-05, "loss": 1.0897, "step": 10550 }, { "epoch": 0.01, "learning_rate": 4.998652361855261e-05, "loss": 1.1917, "step": 10551 }, { "epoch": 0.01, "learning_rate": 4.998652103969604e-05, "loss": 1.1123, "step": 10552 }, { "epoch": 0.01, "learning_rate": 4.9986518460592816e-05, "loss": 0.8103, "step": 10553 }, { "epoch": 0.01, "learning_rate": 4.998651588124292e-05, "loss": 1.1595, "step": 10554 }, { "epoch": 0.01, "learning_rate": 4.9986513301646385e-05, "loss": 1.1804, "step": 10555 }, { "epoch": 0.01, "learning_rate": 4.998651072180319e-05, "loss": 1.0873, "step": 10556 }, { "epoch": 0.01, "learning_rate": 4.998650814171334e-05, "loss": 1.1571, "step": 10557 }, { "epoch": 0.01, "learning_rate": 4.9986505561376826e-05, "loss": 1.1775, "step": 10558 }, { "epoch": 0.01, "learning_rate": 4.998650298079366e-05, "loss": 1.4187, "step": 10559 }, { "epoch": 0.01, "learning_rate": 4.9986500399963835e-05, "loss": 1.3483, "step": 10560 }, { "epoch": 0.01, "learning_rate": 4.9986497818887355e-05, "loss": 1.0625, "step": 10561 }, { "epoch": 0.01, "learning_rate": 4.998649523756423e-05, "loss": 1.1701, "step": 10562 }, { "epoch": 0.01, "learning_rate": 4.9986492655994435e-05, "loss": 1.0178, "step": 10563 }, { "epoch": 0.01, "learning_rate": 4.998649007417798e-05, "loss": 1.0857, "step": 10564 }, { "epoch": 0.01, "learning_rate": 4.9986487492114885e-05, "loss": 0.7984, "step": 10565 }, { "epoch": 0.01, "learning_rate": 4.998648490980512e-05, "loss": 1.3078, "step": 10566 }, { "epoch": 0.01, "learning_rate": 4.998648232724871e-05, "loss": 1.1336, "step": 10567 }, { "epoch": 0.01, "learning_rate": 4.998647974444564e-05, "loss": 0.9312, "step": 10568 }, { "epoch": 0.01, "learning_rate": 4.9986477161395904e-05, "loss": 1.1834, "step": 10569 }, { "epoch": 0.01, "learning_rate": 4.998647457809953e-05, "loss": 0.9114, "step": 10570 }, { "epoch": 0.01, "learning_rate": 4.998647199455649e-05, "loss": 1.1802, "step": 10571 }, { "epoch": 0.01, "learning_rate": 4.9986469410766786e-05, "loss": 1.0653, "step": 10572 }, { "epoch": 0.01, "learning_rate": 4.9986466826730435e-05, "loss": 0.5615, "step": 10573 }, { "epoch": 0.01, "learning_rate": 4.9986464242447435e-05, "loss": 0.6149, "step": 10574 }, { "epoch": 0.01, "learning_rate": 4.998646165791777e-05, "loss": 0.8447, "step": 10575 }, { "epoch": 0.01, "learning_rate": 4.998645907314146e-05, "loss": 0.4378, "step": 10576 }, { "epoch": 0.01, "learning_rate": 4.9986456488118483e-05, "loss": 1.224, "step": 10577 }, { "epoch": 0.01, "learning_rate": 4.998645390284885e-05, "loss": 1.3924, "step": 10578 }, { "epoch": 0.01, "learning_rate": 4.998645131733256e-05, "loss": 1.2941, "step": 10579 }, { "epoch": 0.01, "learning_rate": 4.998644873156962e-05, "loss": 1.149, "step": 10580 }, { "epoch": 0.01, "learning_rate": 4.998644614556002e-05, "loss": 1.003, "step": 10581 }, { "epoch": 0.01, "learning_rate": 4.9986443559303774e-05, "loss": 1.0581, "step": 10582 }, { "epoch": 0.01, "learning_rate": 4.998644097280086e-05, "loss": 0.9061, "step": 10583 }, { "epoch": 0.01, "learning_rate": 4.9986438386051295e-05, "loss": 1.0999, "step": 10584 }, { "epoch": 0.01, "learning_rate": 4.998643579905508e-05, "loss": 0.9629, "step": 10585 }, { "epoch": 0.01, "learning_rate": 4.99864332118122e-05, "loss": 1.1814, "step": 10586 }, { "epoch": 0.01, "learning_rate": 4.998643062432267e-05, "loss": 1.1246, "step": 10587 }, { "epoch": 0.01, "learning_rate": 4.998642803658649e-05, "loss": 1.0821, "step": 10588 }, { "epoch": 0.01, "learning_rate": 4.9986425448603646e-05, "loss": 1.1333, "step": 10589 }, { "epoch": 0.01, "learning_rate": 4.998642286037415e-05, "loss": 0.9511, "step": 10590 }, { "epoch": 0.01, "learning_rate": 4.998642027189799e-05, "loss": 1.1535, "step": 10591 }, { "epoch": 0.01, "learning_rate": 4.998641768317518e-05, "loss": 1.1484, "step": 10592 }, { "epoch": 0.01, "learning_rate": 4.998641509420572e-05, "loss": 0.636, "step": 10593 }, { "epoch": 0.01, "learning_rate": 4.99864125049896e-05, "loss": 1.128, "step": 10594 }, { "epoch": 0.01, "learning_rate": 4.9986409915526834e-05, "loss": 0.9554, "step": 10595 }, { "epoch": 0.01, "learning_rate": 4.9986407325817395e-05, "loss": 0.5884, "step": 10596 }, { "epoch": 0.01, "learning_rate": 4.9986404735861314e-05, "loss": 0.849, "step": 10597 }, { "epoch": 0.01, "learning_rate": 4.9986402145658576e-05, "loss": 1.3073, "step": 10598 }, { "epoch": 0.01, "learning_rate": 4.998639955520918e-05, "loss": 1.1573, "step": 10599 }, { "epoch": 0.01, "learning_rate": 4.9986396964513126e-05, "loss": 1.026, "step": 10600 }, { "epoch": 0.01, "learning_rate": 4.998639437357042e-05, "loss": 1.1495, "step": 10601 }, { "epoch": 0.01, "learning_rate": 4.998639178238106e-05, "loss": 0.9721, "step": 10602 }, { "epoch": 0.01, "learning_rate": 4.998638919094505e-05, "loss": 1.2164, "step": 10603 }, { "epoch": 0.01, "learning_rate": 4.9986386599262374e-05, "loss": 1.0202, "step": 10604 }, { "epoch": 0.01, "learning_rate": 4.9986384007333045e-05, "loss": 0.9831, "step": 10605 }, { "epoch": 0.01, "learning_rate": 4.9986381415157066e-05, "loss": 1.0955, "step": 10606 }, { "epoch": 0.01, "learning_rate": 4.998637882273443e-05, "loss": 1.1857, "step": 10607 }, { "epoch": 0.01, "learning_rate": 4.9986376230065134e-05, "loss": 1.3291, "step": 10608 }, { "epoch": 0.01, "learning_rate": 4.9986373637149193e-05, "loss": 1.1299, "step": 10609 }, { "epoch": 0.01, "learning_rate": 4.998637104398659e-05, "loss": 1.1586, "step": 10610 }, { "epoch": 0.01, "learning_rate": 4.998636845057734e-05, "loss": 1.114, "step": 10611 }, { "epoch": 0.01, "learning_rate": 4.9986365856921424e-05, "loss": 1.36, "step": 10612 }, { "epoch": 0.01, "learning_rate": 4.998636326301886e-05, "loss": 1.0706, "step": 10613 }, { "epoch": 0.01, "learning_rate": 4.998636066886963e-05, "loss": 1.2458, "step": 10614 }, { "epoch": 0.01, "learning_rate": 4.998635807447376e-05, "loss": 1.0786, "step": 10615 }, { "epoch": 0.01, "learning_rate": 4.998635547983123e-05, "loss": 1.108, "step": 10616 }, { "epoch": 0.01, "learning_rate": 4.9986352884942044e-05, "loss": 1.1663, "step": 10617 }, { "epoch": 0.01, "learning_rate": 4.998635028980621e-05, "loss": 1.181, "step": 10618 }, { "epoch": 0.01, "learning_rate": 4.9986347694423714e-05, "loss": 0.9882, "step": 10619 }, { "epoch": 0.01, "learning_rate": 4.9986345098794565e-05, "loss": 1.0248, "step": 10620 }, { "epoch": 0.01, "learning_rate": 4.998634250291876e-05, "loss": 0.8304, "step": 10621 }, { "epoch": 0.01, "learning_rate": 4.99863399067963e-05, "loss": 1.1176, "step": 10622 }, { "epoch": 0.01, "learning_rate": 4.998633731042719e-05, "loss": 1.0107, "step": 10623 }, { "epoch": 0.01, "learning_rate": 4.998633471381142e-05, "loss": 1.4044, "step": 10624 }, { "epoch": 0.01, "learning_rate": 4.9986332116949e-05, "loss": 4.4344, "step": 10625 }, { "epoch": 0.01, "learning_rate": 4.998632951983992e-05, "loss": 1.1785, "step": 10626 }, { "epoch": 0.01, "learning_rate": 4.9986326922484186e-05, "loss": 1.0471, "step": 10627 }, { "epoch": 0.01, "learning_rate": 4.99863243248818e-05, "loss": 1.2066, "step": 10628 }, { "epoch": 0.01, "learning_rate": 4.9986321727032756e-05, "loss": 0.9483, "step": 10629 }, { "epoch": 0.01, "learning_rate": 4.998631912893707e-05, "loss": 1.0307, "step": 10630 }, { "epoch": 0.01, "learning_rate": 4.9986316530594716e-05, "loss": 1.0138, "step": 10631 }, { "epoch": 0.01, "learning_rate": 4.9986313932005715e-05, "loss": 0.9078, "step": 10632 }, { "epoch": 0.01, "learning_rate": 4.998631133317005e-05, "loss": 1.1706, "step": 10633 }, { "epoch": 0.01, "learning_rate": 4.998630873408774e-05, "loss": 1.1575, "step": 10634 }, { "epoch": 0.01, "learning_rate": 4.998630613475878e-05, "loss": 1.1129, "step": 10635 }, { "epoch": 0.01, "learning_rate": 4.998630353518315e-05, "loss": 1.542, "step": 10636 }, { "epoch": 0.01, "learning_rate": 4.998630093536088e-05, "loss": 0.982, "step": 10637 }, { "epoch": 0.01, "learning_rate": 4.998629833529195e-05, "loss": 1.1706, "step": 10638 }, { "epoch": 0.01, "learning_rate": 4.998629573497636e-05, "loss": 0.9877, "step": 10639 }, { "epoch": 0.01, "learning_rate": 4.998629313441413e-05, "loss": 1.0951, "step": 10640 }, { "epoch": 0.01, "learning_rate": 4.998629053360524e-05, "loss": 1.1293, "step": 10641 }, { "epoch": 0.01, "learning_rate": 4.998628793254969e-05, "loss": 1.1675, "step": 10642 }, { "epoch": 0.01, "learning_rate": 4.998628533124749e-05, "loss": 1.1676, "step": 10643 }, { "epoch": 0.01, "learning_rate": 4.9986282729698633e-05, "loss": 0.8128, "step": 10644 }, { "epoch": 0.01, "learning_rate": 4.998628012790313e-05, "loss": 1.0571, "step": 10645 }, { "epoch": 0.01, "learning_rate": 4.998627752586096e-05, "loss": 1.369, "step": 10646 }, { "epoch": 0.01, "learning_rate": 4.9986274923572154e-05, "loss": 1.4563, "step": 10647 }, { "epoch": 0.01, "learning_rate": 4.998627232103668e-05, "loss": 1.2891, "step": 10648 }, { "epoch": 0.01, "learning_rate": 4.998626971825455e-05, "loss": 0.8835, "step": 10649 }, { "epoch": 0.01, "learning_rate": 4.998626711522577e-05, "loss": 0.9248, "step": 10650 }, { "epoch": 0.01, "learning_rate": 4.998626451195034e-05, "loss": 0.9569, "step": 10651 }, { "epoch": 0.01, "learning_rate": 4.998626190842825e-05, "loss": 1.1694, "step": 10652 }, { "epoch": 0.01, "learning_rate": 4.998625930465952e-05, "loss": 1.1396, "step": 10653 }, { "epoch": 0.01, "learning_rate": 4.9986256700644125e-05, "loss": 1.131, "step": 10654 }, { "epoch": 0.01, "learning_rate": 4.998625409638208e-05, "loss": 1.0525, "step": 10655 }, { "epoch": 0.01, "learning_rate": 4.998625149187337e-05, "loss": 1.4169, "step": 10656 }, { "epoch": 0.01, "learning_rate": 4.998624888711802e-05, "loss": 1.29, "step": 10657 }, { "epoch": 0.01, "learning_rate": 4.998624628211601e-05, "loss": 0.8785, "step": 10658 }, { "epoch": 0.01, "learning_rate": 4.998624367686735e-05, "loss": 1.1464, "step": 10659 }, { "epoch": 0.01, "learning_rate": 4.998624107137203e-05, "loss": 1.2562, "step": 10660 }, { "epoch": 0.01, "learning_rate": 4.998623846563007e-05, "loss": 1.3613, "step": 10661 }, { "epoch": 0.01, "learning_rate": 4.998623585964144e-05, "loss": 1.4084, "step": 10662 }, { "epoch": 0.01, "learning_rate": 4.998623325340616e-05, "loss": 0.99, "step": 10663 }, { "epoch": 0.01, "learning_rate": 4.9986230646924234e-05, "loss": 1.432, "step": 10664 }, { "epoch": 0.01, "learning_rate": 4.9986228040195646e-05, "loss": 1.4405, "step": 10665 }, { "epoch": 0.01, "learning_rate": 4.9986225433220416e-05, "loss": 1.2277, "step": 10666 }, { "epoch": 0.01, "learning_rate": 4.9986222825998517e-05, "loss": 1.0819, "step": 10667 }, { "epoch": 0.01, "learning_rate": 4.9986220218529974e-05, "loss": 1.0059, "step": 10668 }, { "epoch": 0.01, "learning_rate": 4.9986217610814776e-05, "loss": 0.8576, "step": 10669 }, { "epoch": 0.01, "learning_rate": 4.998621500285292e-05, "loss": 0.6349, "step": 10670 }, { "epoch": 0.01, "learning_rate": 4.998621239464442e-05, "loss": 0.9552, "step": 10671 }, { "epoch": 0.01, "learning_rate": 4.998620978618927e-05, "loss": 1.2087, "step": 10672 }, { "epoch": 0.01, "learning_rate": 4.998620717748745e-05, "loss": 5.6711, "step": 10673 }, { "epoch": 0.01, "learning_rate": 4.998620456853899e-05, "loss": 3.5529, "step": 10674 }, { "epoch": 0.01, "learning_rate": 4.9986201959343873e-05, "loss": 0.8959, "step": 10675 }, { "epoch": 0.01, "learning_rate": 4.99861993499021e-05, "loss": 1.1657, "step": 10676 }, { "epoch": 0.01, "learning_rate": 4.998619674021367e-05, "loss": 1.0985, "step": 10677 }, { "epoch": 0.01, "learning_rate": 4.9986194130278597e-05, "loss": 1.3274, "step": 10678 }, { "epoch": 0.01, "learning_rate": 4.998619152009686e-05, "loss": 1.1938, "step": 10679 }, { "epoch": 0.01, "learning_rate": 4.998618890966847e-05, "loss": 1.4198, "step": 10680 }, { "epoch": 0.01, "learning_rate": 4.998618629899344e-05, "loss": 1.1039, "step": 10681 }, { "epoch": 0.01, "learning_rate": 4.998618368807175e-05, "loss": 1.1376, "step": 10682 }, { "epoch": 0.01, "learning_rate": 4.99861810769034e-05, "loss": 1.0047, "step": 10683 }, { "epoch": 0.01, "learning_rate": 4.998617846548841e-05, "loss": 5.4369, "step": 10684 }, { "epoch": 0.01, "learning_rate": 4.998617585382676e-05, "loss": 4.1474, "step": 10685 }, { "epoch": 0.01, "learning_rate": 4.9986173241918456e-05, "loss": 1.1917, "step": 10686 }, { "epoch": 0.01, "learning_rate": 4.9986170629763505e-05, "loss": 1.261, "step": 10687 }, { "epoch": 0.01, "learning_rate": 4.99861680173619e-05, "loss": 1.0599, "step": 10688 }, { "epoch": 0.01, "learning_rate": 4.998616540471363e-05, "loss": 1.2908, "step": 10689 }, { "epoch": 0.01, "learning_rate": 4.998616279181872e-05, "loss": 0.9679, "step": 10690 }, { "epoch": 0.01, "learning_rate": 4.998616017867716e-05, "loss": 1.0874, "step": 10691 }, { "epoch": 0.01, "learning_rate": 4.998615756528893e-05, "loss": 1.1285, "step": 10692 }, { "epoch": 0.01, "learning_rate": 4.9986154951654066e-05, "loss": 0.9024, "step": 10693 }, { "epoch": 0.01, "learning_rate": 4.9986152337772536e-05, "loss": 0.9533, "step": 10694 }, { "epoch": 0.01, "learning_rate": 4.9986149723644357e-05, "loss": 1.5688, "step": 10695 }, { "epoch": 0.01, "learning_rate": 4.998614710926952e-05, "loss": 1.0835, "step": 10696 }, { "epoch": 0.01, "learning_rate": 4.9986144494648043e-05, "loss": 1.0618, "step": 10697 }, { "epoch": 0.01, "learning_rate": 4.99861418797799e-05, "loss": 0.9249, "step": 10698 }, { "epoch": 0.01, "learning_rate": 4.998613926466511e-05, "loss": 0.9606, "step": 10699 }, { "epoch": 0.01, "learning_rate": 4.9986136649303674e-05, "loss": 1.0427, "step": 10700 }, { "epoch": 0.01, "learning_rate": 4.998613403369557e-05, "loss": 0.5878, "step": 10701 }, { "epoch": 0.01, "learning_rate": 4.998613141784083e-05, "loss": 0.7082, "step": 10702 }, { "epoch": 0.01, "learning_rate": 4.998612880173943e-05, "loss": 0.9104, "step": 10703 }, { "epoch": 0.01, "learning_rate": 4.998612618539138e-05, "loss": 0.8221, "step": 10704 }, { "epoch": 0.01, "learning_rate": 4.998612356879667e-05, "loss": 0.7588, "step": 10705 }, { "epoch": 0.01, "learning_rate": 4.998612095195532e-05, "loss": 0.9248, "step": 10706 }, { "epoch": 0.01, "learning_rate": 4.99861183348673e-05, "loss": 0.8445, "step": 10707 }, { "epoch": 0.01, "learning_rate": 4.998611571753264e-05, "loss": 0.8719, "step": 10708 }, { "epoch": 0.01, "learning_rate": 4.998611309995133e-05, "loss": 0.6463, "step": 10709 }, { "epoch": 0.01, "learning_rate": 4.998611048212337e-05, "loss": 0.558, "step": 10710 }, { "epoch": 0.01, "learning_rate": 4.9986107864048746e-05, "loss": 0.5944, "step": 10711 }, { "epoch": 0.01, "learning_rate": 4.9986105245727476e-05, "loss": 1.0485, "step": 10712 }, { "epoch": 0.01, "learning_rate": 4.998610262715955e-05, "loss": 1.4276, "step": 10713 }, { "epoch": 0.01, "learning_rate": 4.9986100008344974e-05, "loss": 1.1766, "step": 10714 }, { "epoch": 0.01, "learning_rate": 4.998609738928375e-05, "loss": 1.0757, "step": 10715 }, { "epoch": 0.01, "learning_rate": 4.998609476997587e-05, "loss": 1.0755, "step": 10716 }, { "epoch": 0.01, "learning_rate": 4.998609215042134e-05, "loss": 1.106, "step": 10717 }, { "epoch": 0.01, "learning_rate": 4.998608953062015e-05, "loss": 0.9057, "step": 10718 }, { "epoch": 0.01, "learning_rate": 4.998608691057232e-05, "loss": 0.8137, "step": 10719 }, { "epoch": 0.01, "learning_rate": 4.998608429027783e-05, "loss": 1.2992, "step": 10720 }, { "epoch": 0.01, "learning_rate": 4.998608166973669e-05, "loss": 1.1847, "step": 10721 }, { "epoch": 0.01, "learning_rate": 4.9986079048948884e-05, "loss": 1.3237, "step": 10722 }, { "epoch": 0.01, "learning_rate": 4.998607642791444e-05, "loss": 1.5411, "step": 10723 }, { "epoch": 0.01, "learning_rate": 4.998607380663335e-05, "loss": 1.2056, "step": 10724 }, { "epoch": 0.01, "learning_rate": 4.99860711851056e-05, "loss": 1.0274, "step": 10725 }, { "epoch": 0.01, "learning_rate": 4.998606856333119e-05, "loss": 1.3021, "step": 10726 }, { "epoch": 0.01, "learning_rate": 4.9986065941310144e-05, "loss": 1.1686, "step": 10727 }, { "epoch": 0.01, "learning_rate": 4.998606331904244e-05, "loss": 1.3316, "step": 10728 }, { "epoch": 0.01, "learning_rate": 4.9986060696528084e-05, "loss": 1.3269, "step": 10729 }, { "epoch": 0.01, "learning_rate": 4.9986058073767074e-05, "loss": 0.9971, "step": 10730 }, { "epoch": 0.01, "learning_rate": 4.998605545075941e-05, "loss": 0.9264, "step": 10731 }, { "epoch": 0.01, "learning_rate": 4.99860528275051e-05, "loss": 0.9454, "step": 10732 }, { "epoch": 0.01, "learning_rate": 4.9986050204004134e-05, "loss": 1.2553, "step": 10733 }, { "epoch": 0.01, "learning_rate": 4.998604758025652e-05, "loss": 1.1812, "step": 10734 }, { "epoch": 0.01, "learning_rate": 4.998604495626224e-05, "loss": 1.0018, "step": 10735 }, { "epoch": 0.01, "learning_rate": 4.998604233202133e-05, "loss": 0.8275, "step": 10736 }, { "epoch": 0.01, "learning_rate": 4.998603970753376e-05, "loss": 1.2006, "step": 10737 }, { "epoch": 0.01, "learning_rate": 4.998603708279954e-05, "loss": 1.1241, "step": 10738 }, { "epoch": 0.01, "learning_rate": 4.9986034457818664e-05, "loss": 1.2169, "step": 10739 }, { "epoch": 0.01, "learning_rate": 4.9986031832591134e-05, "loss": 1.0624, "step": 10740 }, { "epoch": 0.01, "learning_rate": 4.9986029207116955e-05, "loss": 1.0672, "step": 10741 }, { "epoch": 0.01, "learning_rate": 4.998602658139613e-05, "loss": 0.9789, "step": 10742 }, { "epoch": 0.01, "learning_rate": 4.998602395542864e-05, "loss": 1.0947, "step": 10743 }, { "epoch": 0.01, "learning_rate": 4.9986021329214516e-05, "loss": 1.75, "step": 10744 }, { "epoch": 0.01, "learning_rate": 4.9986018702753734e-05, "loss": 1.4034, "step": 10745 }, { "epoch": 0.01, "learning_rate": 4.9986016076046295e-05, "loss": 0.8919, "step": 10746 }, { "epoch": 0.01, "learning_rate": 4.998601344909221e-05, "loss": 1.254, "step": 10747 }, { "epoch": 0.01, "learning_rate": 4.998601082189147e-05, "loss": 1.0796, "step": 10748 }, { "epoch": 0.01, "learning_rate": 4.998600819444408e-05, "loss": 1.2729, "step": 10749 }, { "epoch": 0.01, "learning_rate": 4.9986005566750036e-05, "loss": 1.1744, "step": 10750 }, { "epoch": 0.01, "learning_rate": 4.998600293880935e-05, "loss": 1.2834, "step": 10751 }, { "epoch": 0.01, "learning_rate": 4.9986000310622004e-05, "loss": 1.1745, "step": 10752 }, { "epoch": 0.01, "learning_rate": 4.9985997682188014e-05, "loss": 1.195, "step": 10753 }, { "epoch": 0.01, "learning_rate": 4.998599505350736e-05, "loss": 1.1958, "step": 10754 }, { "epoch": 0.01, "learning_rate": 4.9985992424580066e-05, "loss": 1.4716, "step": 10755 }, { "epoch": 0.01, "learning_rate": 4.998598979540612e-05, "loss": 1.2591, "step": 10756 }, { "epoch": 0.01, "learning_rate": 4.998598716598552e-05, "loss": 1.2078, "step": 10757 }, { "epoch": 0.01, "learning_rate": 4.998598453631827e-05, "loss": 1.0542, "step": 10758 }, { "epoch": 0.01, "learning_rate": 4.998598190640437e-05, "loss": 1.0827, "step": 10759 }, { "epoch": 0.01, "learning_rate": 4.998597927624382e-05, "loss": 1.4116, "step": 10760 }, { "epoch": 0.01, "learning_rate": 4.9985976645836615e-05, "loss": 1.0502, "step": 10761 }, { "epoch": 0.01, "learning_rate": 4.9985974015182755e-05, "loss": 0.6948, "step": 10762 }, { "epoch": 0.01, "learning_rate": 4.9985971384282246e-05, "loss": 0.5402, "step": 10763 }, { "epoch": 0.01, "learning_rate": 4.9985968753135094e-05, "loss": 0.5075, "step": 10764 }, { "epoch": 0.01, "learning_rate": 4.998596612174129e-05, "loss": 0.4851, "step": 10765 }, { "epoch": 0.01, "learning_rate": 4.998596349010083e-05, "loss": 0.3154, "step": 10766 }, { "epoch": 0.01, "learning_rate": 4.9985960858213724e-05, "loss": 0.3732, "step": 10767 }, { "epoch": 0.01, "learning_rate": 4.9985958226079955e-05, "loss": 0.402, "step": 10768 }, { "epoch": 0.01, "learning_rate": 4.9985955593699544e-05, "loss": 0.3886, "step": 10769 }, { "epoch": 0.01, "learning_rate": 4.9985952961072484e-05, "loss": 1.1539, "step": 10770 }, { "epoch": 0.01, "learning_rate": 4.9985950328198774e-05, "loss": 1.3236, "step": 10771 }, { "epoch": 0.01, "learning_rate": 4.998594769507841e-05, "loss": 1.0445, "step": 10772 }, { "epoch": 0.01, "learning_rate": 4.9985945061711393e-05, "loss": 0.8028, "step": 10773 }, { "epoch": 0.01, "learning_rate": 4.998594242809773e-05, "loss": 0.4059, "step": 10774 }, { "epoch": 0.01, "learning_rate": 4.9985939794237416e-05, "loss": 1.045, "step": 10775 }, { "epoch": 0.01, "learning_rate": 4.998593716013045e-05, "loss": 1.4564, "step": 10776 }, { "epoch": 0.01, "learning_rate": 4.998593452577683e-05, "loss": 1.2097, "step": 10777 }, { "epoch": 0.01, "learning_rate": 4.998593189117656e-05, "loss": 1.0419, "step": 10778 }, { "epoch": 0.01, "learning_rate": 4.9985929256329644e-05, "loss": 1.1569, "step": 10779 }, { "epoch": 0.01, "learning_rate": 4.998592662123607e-05, "loss": 1.1554, "step": 10780 }, { "epoch": 0.01, "learning_rate": 4.9985923985895855e-05, "loss": 1.0237, "step": 10781 }, { "epoch": 0.01, "learning_rate": 4.9985921350308984e-05, "loss": 0.9649, "step": 10782 }, { "epoch": 0.01, "learning_rate": 4.9985918714475463e-05, "loss": 1.0963, "step": 10783 }, { "epoch": 0.01, "learning_rate": 4.998591607839529e-05, "loss": 1.0937, "step": 10784 }, { "epoch": 0.01, "learning_rate": 4.9985913442068475e-05, "loss": 1.3142, "step": 10785 }, { "epoch": 0.01, "learning_rate": 4.998591080549499e-05, "loss": 0.7393, "step": 10786 }, { "epoch": 0.01, "learning_rate": 4.9985908168674875e-05, "loss": 1.1603, "step": 10787 }, { "epoch": 0.01, "learning_rate": 4.99859055316081e-05, "loss": 1.3147, "step": 10788 }, { "epoch": 0.01, "learning_rate": 4.998590289429468e-05, "loss": 1.0646, "step": 10789 }, { "epoch": 0.01, "learning_rate": 4.998590025673461e-05, "loss": 0.9826, "step": 10790 }, { "epoch": 0.01, "learning_rate": 4.9985897618927886e-05, "loss": 0.9658, "step": 10791 }, { "epoch": 0.01, "learning_rate": 4.9985894980874516e-05, "loss": 1.1047, "step": 10792 }, { "epoch": 0.01, "learning_rate": 4.998589234257448e-05, "loss": 0.9545, "step": 10793 }, { "epoch": 0.01, "learning_rate": 4.998588970402781e-05, "loss": 1.2241, "step": 10794 }, { "epoch": 0.01, "learning_rate": 4.998588706523449e-05, "loss": 0.6729, "step": 10795 }, { "epoch": 0.01, "learning_rate": 4.9985884426194516e-05, "loss": 0.9699, "step": 10796 }, { "epoch": 0.01, "learning_rate": 4.998588178690789e-05, "loss": 0.7535, "step": 10797 }, { "epoch": 0.01, "learning_rate": 4.998587914737461e-05, "loss": 1.4064, "step": 10798 }, { "epoch": 0.01, "learning_rate": 4.9985876507594685e-05, "loss": 1.1086, "step": 10799 }, { "epoch": 0.01, "learning_rate": 4.9985873867568114e-05, "loss": 1.1563, "step": 10800 }, { "epoch": 0.01, "learning_rate": 4.998587122729489e-05, "loss": 1.0799, "step": 10801 }, { "epoch": 0.01, "learning_rate": 4.998586858677501e-05, "loss": 0.8134, "step": 10802 }, { "epoch": 0.01, "learning_rate": 4.998586594600849e-05, "loss": 1.1417, "step": 10803 }, { "epoch": 0.01, "learning_rate": 4.998586330499532e-05, "loss": 1.1602, "step": 10804 }, { "epoch": 0.01, "learning_rate": 4.998586066373549e-05, "loss": 1.1683, "step": 10805 }, { "epoch": 0.01, "learning_rate": 4.9985858022229015e-05, "loss": 1.2593, "step": 10806 }, { "epoch": 0.01, "learning_rate": 4.998585538047589e-05, "loss": 1.0802, "step": 10807 }, { "epoch": 0.01, "learning_rate": 4.998585273847612e-05, "loss": 0.6904, "step": 10808 }, { "epoch": 0.01, "learning_rate": 4.9985850096229694e-05, "loss": 0.9784, "step": 10809 }, { "epoch": 0.01, "learning_rate": 4.9985847453736624e-05, "loss": 1.2396, "step": 10810 }, { "epoch": 0.01, "learning_rate": 4.99858448109969e-05, "loss": 1.1279, "step": 10811 }, { "epoch": 0.01, "learning_rate": 4.998584216801052e-05, "loss": 1.1777, "step": 10812 }, { "epoch": 0.01, "learning_rate": 4.99858395247775e-05, "loss": 1.4183, "step": 10813 }, { "epoch": 0.01, "learning_rate": 4.9985836881297826e-05, "loss": 1.5293, "step": 10814 }, { "epoch": 0.01, "learning_rate": 4.99858342375715e-05, "loss": 1.2572, "step": 10815 }, { "epoch": 0.01, "learning_rate": 4.998583159359853e-05, "loss": 1.444, "step": 10816 }, { "epoch": 0.01, "learning_rate": 4.998582894937891e-05, "loss": 1.1602, "step": 10817 }, { "epoch": 0.01, "learning_rate": 4.998582630491264e-05, "loss": 0.947, "step": 10818 }, { "epoch": 0.01, "learning_rate": 4.998582366019972e-05, "loss": 1.0774, "step": 10819 }, { "epoch": 0.01, "learning_rate": 4.998582101524015e-05, "loss": 1.0768, "step": 10820 }, { "epoch": 0.01, "learning_rate": 4.9985818370033933e-05, "loss": 1.3893, "step": 10821 }, { "epoch": 0.01, "learning_rate": 4.9985815724581066e-05, "loss": 1.2026, "step": 10822 }, { "epoch": 0.01, "learning_rate": 4.9985813078881536e-05, "loss": 1.0682, "step": 10823 }, { "epoch": 0.01, "learning_rate": 4.998581043293538e-05, "loss": 1.1521, "step": 10824 }, { "epoch": 0.01, "learning_rate": 4.9985807786742556e-05, "loss": 1.0433, "step": 10825 }, { "epoch": 0.01, "learning_rate": 4.9985805140303085e-05, "loss": 0.8613, "step": 10826 }, { "epoch": 0.01, "learning_rate": 4.998580249361697e-05, "loss": 1.178, "step": 10827 }, { "epoch": 0.01, "learning_rate": 4.998579984668421e-05, "loss": 0.9097, "step": 10828 }, { "epoch": 0.01, "learning_rate": 4.998579719950479e-05, "loss": 1.1461, "step": 10829 }, { "epoch": 0.01, "learning_rate": 4.998579455207873e-05, "loss": 1.397, "step": 10830 }, { "epoch": 0.01, "learning_rate": 4.998579190440601e-05, "loss": 1.2818, "step": 10831 }, { "epoch": 0.01, "learning_rate": 4.9985789256486654e-05, "loss": 1.2913, "step": 10832 }, { "epoch": 0.01, "learning_rate": 4.998578660832064e-05, "loss": 1.5145, "step": 10833 }, { "epoch": 0.01, "learning_rate": 4.998578395990798e-05, "loss": 1.4816, "step": 10834 }, { "epoch": 0.01, "learning_rate": 4.9985781311248667e-05, "loss": 0.6521, "step": 10835 }, { "epoch": 0.01, "learning_rate": 4.998577866234271e-05, "loss": 0.9915, "step": 10836 }, { "epoch": 0.01, "learning_rate": 4.9985776013190105e-05, "loss": 1.2507, "step": 10837 }, { "epoch": 0.01, "learning_rate": 4.9985773363790843e-05, "loss": 1.0145, "step": 10838 }, { "epoch": 0.01, "learning_rate": 4.998577071414494e-05, "loss": 1.146, "step": 10839 }, { "epoch": 0.01, "learning_rate": 4.9985768064252386e-05, "loss": 0.9321, "step": 10840 }, { "epoch": 0.01, "learning_rate": 4.9985765414113184e-05, "loss": 1.0879, "step": 10841 }, { "epoch": 0.01, "learning_rate": 4.9985762763727326e-05, "loss": 1.134, "step": 10842 }, { "epoch": 0.01, "learning_rate": 4.9985760113094825e-05, "loss": 0.8009, "step": 10843 }, { "epoch": 0.01, "learning_rate": 4.9985757462215675e-05, "loss": 1.0375, "step": 10844 }, { "epoch": 0.01, "learning_rate": 4.9985754811089876e-05, "loss": 1.517, "step": 10845 }, { "epoch": 0.01, "learning_rate": 4.998575215971743e-05, "loss": 1.6569, "step": 10846 }, { "epoch": 0.01, "learning_rate": 4.998574950809833e-05, "loss": 1.3168, "step": 10847 }, { "epoch": 0.01, "learning_rate": 4.998574685623258e-05, "loss": 1.1746, "step": 10848 }, { "epoch": 0.01, "learning_rate": 4.998574420412019e-05, "loss": 1.3589, "step": 10849 }, { "epoch": 0.01, "learning_rate": 4.998574155176114e-05, "loss": 1.181, "step": 10850 }, { "epoch": 0.01, "learning_rate": 4.9985738899155454e-05, "loss": 1.0479, "step": 10851 }, { "epoch": 0.01, "learning_rate": 4.998573624630311e-05, "loss": 1.0858, "step": 10852 }, { "epoch": 0.01, "learning_rate": 4.998573359320412e-05, "loss": 1.0836, "step": 10853 }, { "epoch": 0.01, "learning_rate": 4.998573093985848e-05, "loss": 1.3111, "step": 10854 }, { "epoch": 0.01, "learning_rate": 4.99857282862662e-05, "loss": 1.0305, "step": 10855 }, { "epoch": 0.01, "learning_rate": 4.9985725632427263e-05, "loss": 1.1478, "step": 10856 }, { "epoch": 0.01, "learning_rate": 4.9985722978341674e-05, "loss": 0.5696, "step": 10857 }, { "epoch": 0.01, "learning_rate": 4.998572032400944e-05, "loss": 0.5121, "step": 10858 }, { "epoch": 0.01, "learning_rate": 4.998571766943057e-05, "loss": 0.7915, "step": 10859 }, { "epoch": 0.01, "learning_rate": 4.998571501460503e-05, "loss": 1.0314, "step": 10860 }, { "epoch": 0.01, "learning_rate": 4.9985712359532856e-05, "loss": 0.9899, "step": 10861 }, { "epoch": 0.01, "learning_rate": 4.998570970421402e-05, "loss": 1.0597, "step": 10862 }, { "epoch": 0.01, "learning_rate": 4.9985707048648555e-05, "loss": 1.1329, "step": 10863 }, { "epoch": 0.01, "learning_rate": 4.998570439283643e-05, "loss": 0.8242, "step": 10864 }, { "epoch": 0.01, "learning_rate": 4.998570173677766e-05, "loss": 1.0128, "step": 10865 }, { "epoch": 0.01, "learning_rate": 4.998569908047224e-05, "loss": 1.2129, "step": 10866 }, { "epoch": 0.01, "learning_rate": 4.998569642392017e-05, "loss": 1.1209, "step": 10867 }, { "epoch": 0.01, "learning_rate": 4.998569376712145e-05, "loss": 0.6446, "step": 10868 }, { "epoch": 0.01, "learning_rate": 4.9985691110076086e-05, "loss": 0.5986, "step": 10869 }, { "epoch": 0.01, "learning_rate": 4.9985688452784076e-05, "loss": 0.905, "step": 10870 }, { "epoch": 0.01, "learning_rate": 4.998568579524541e-05, "loss": 1.1289, "step": 10871 }, { "epoch": 0.01, "learning_rate": 4.99856831374601e-05, "loss": 0.9742, "step": 10872 }, { "epoch": 0.01, "learning_rate": 4.9985680479428146e-05, "loss": 1.108, "step": 10873 }, { "epoch": 0.01, "learning_rate": 4.998567782114954e-05, "loss": 1.1507, "step": 10874 }, { "epoch": 0.01, "learning_rate": 4.998567516262429e-05, "loss": 1.1444, "step": 10875 }, { "epoch": 0.01, "learning_rate": 4.998567250385239e-05, "loss": 1.1622, "step": 10876 }, { "epoch": 0.01, "learning_rate": 4.998566984483384e-05, "loss": 1.0635, "step": 10877 }, { "epoch": 0.01, "learning_rate": 4.9985667185568643e-05, "loss": 1.2776, "step": 10878 }, { "epoch": 0.01, "learning_rate": 4.998566452605679e-05, "loss": 1.2305, "step": 10879 }, { "epoch": 0.01, "learning_rate": 4.9985661866298296e-05, "loss": 1.2408, "step": 10880 }, { "epoch": 0.01, "learning_rate": 4.998565920629316e-05, "loss": 1.2548, "step": 10881 }, { "epoch": 0.01, "learning_rate": 4.998565654604137e-05, "loss": 1.0354, "step": 10882 }, { "epoch": 0.01, "learning_rate": 4.9985653885542924e-05, "loss": 1.0019, "step": 10883 }, { "epoch": 0.01, "learning_rate": 4.9985651224797846e-05, "loss": 0.9741, "step": 10884 }, { "epoch": 0.01, "learning_rate": 4.998564856380611e-05, "loss": 1.2451, "step": 10885 }, { "epoch": 0.01, "learning_rate": 4.9985645902567735e-05, "loss": 1.3416, "step": 10886 }, { "epoch": 0.01, "learning_rate": 4.99856432410827e-05, "loss": 1.1328, "step": 10887 }, { "epoch": 0.01, "learning_rate": 4.998564057935102e-05, "loss": 1.2498, "step": 10888 }, { "epoch": 0.01, "learning_rate": 4.9985637917372704e-05, "loss": 1.1604, "step": 10889 }, { "epoch": 0.01, "learning_rate": 4.998563525514773e-05, "loss": 1.0942, "step": 10890 }, { "epoch": 0.01, "learning_rate": 4.9985632592676115e-05, "loss": 1.0734, "step": 10891 }, { "epoch": 0.01, "learning_rate": 4.998562992995784e-05, "loss": 1.1091, "step": 10892 }, { "epoch": 0.01, "learning_rate": 4.998562726699293e-05, "loss": 1.1214, "step": 10893 }, { "epoch": 0.01, "learning_rate": 4.9985624603781365e-05, "loss": 0.9871, "step": 10894 }, { "epoch": 0.01, "learning_rate": 4.998562194032316e-05, "loss": 1.0969, "step": 10895 }, { "epoch": 0.01, "learning_rate": 4.9985619276618304e-05, "loss": 1.3956, "step": 10896 }, { "epoch": 0.01, "learning_rate": 4.99856166126668e-05, "loss": 0.9659, "step": 10897 }, { "epoch": 0.01, "learning_rate": 4.9985613948468646e-05, "loss": 1.0178, "step": 10898 }, { "epoch": 0.01, "learning_rate": 4.9985611284023843e-05, "loss": 1.0859, "step": 10899 }, { "epoch": 0.01, "learning_rate": 4.998560861933239e-05, "loss": 1.0746, "step": 10900 }, { "epoch": 0.01, "learning_rate": 4.99856059543943e-05, "loss": 1.1855, "step": 10901 }, { "epoch": 0.01, "learning_rate": 4.998560328920956e-05, "loss": 1.2518, "step": 10902 }, { "epoch": 0.01, "learning_rate": 4.9985600623778174e-05, "loss": 1.0322, "step": 10903 }, { "epoch": 0.01, "learning_rate": 4.998559795810013e-05, "loss": 1.1024, "step": 10904 }, { "epoch": 0.01, "learning_rate": 4.998559529217545e-05, "loss": 1.1613, "step": 10905 }, { "epoch": 0.01, "learning_rate": 4.998559262600412e-05, "loss": 1.0327, "step": 10906 }, { "epoch": 0.01, "learning_rate": 4.998558995958614e-05, "loss": 1.1118, "step": 10907 }, { "epoch": 0.01, "learning_rate": 4.998558729292151e-05, "loss": 0.9098, "step": 10908 }, { "epoch": 0.01, "learning_rate": 4.998558462601024e-05, "loss": 1.2527, "step": 10909 }, { "epoch": 0.01, "learning_rate": 4.998558195885232e-05, "loss": 1.0318, "step": 10910 }, { "epoch": 0.01, "learning_rate": 4.9985579291447754e-05, "loss": 1.4081, "step": 10911 }, { "epoch": 0.01, "learning_rate": 4.998557662379654e-05, "loss": 1.3483, "step": 10912 }, { "epoch": 0.01, "learning_rate": 4.998557395589868e-05, "loss": 1.1597, "step": 10913 }, { "epoch": 0.01, "learning_rate": 4.998557128775417e-05, "loss": 1.2172, "step": 10914 }, { "epoch": 0.01, "learning_rate": 4.998556861936301e-05, "loss": 1.0755, "step": 10915 }, { "epoch": 0.01, "learning_rate": 4.998556595072521e-05, "loss": 0.8813, "step": 10916 }, { "epoch": 0.01, "learning_rate": 4.9985563281840764e-05, "loss": 1.2554, "step": 10917 }, { "epoch": 0.01, "learning_rate": 4.9985560612709667e-05, "loss": 0.6499, "step": 10918 }, { "epoch": 0.01, "learning_rate": 4.998555794333192e-05, "loss": 0.3899, "step": 10919 }, { "epoch": 0.01, "learning_rate": 4.998555527370753e-05, "loss": 0.3813, "step": 10920 }, { "epoch": 0.01, "learning_rate": 4.998555260383649e-05, "loss": 0.4494, "step": 10921 }, { "epoch": 0.01, "learning_rate": 4.998554993371881e-05, "loss": 0.447, "step": 10922 }, { "epoch": 0.01, "learning_rate": 4.998554726335448e-05, "loss": 1.1859, "step": 10923 }, { "epoch": 0.01, "learning_rate": 4.9985544592743504e-05, "loss": 0.9167, "step": 10924 }, { "epoch": 0.01, "learning_rate": 4.998554192188588e-05, "loss": 0.8268, "step": 10925 }, { "epoch": 0.01, "learning_rate": 4.998553925078161e-05, "loss": 0.9506, "step": 10926 }, { "epoch": 0.01, "learning_rate": 4.998553657943069e-05, "loss": 1.0431, "step": 10927 }, { "epoch": 0.01, "learning_rate": 4.998553390783313e-05, "loss": 0.9283, "step": 10928 }, { "epoch": 0.01, "learning_rate": 4.998553123598891e-05, "loss": 1.0289, "step": 10929 }, { "epoch": 0.01, "learning_rate": 4.998552856389806e-05, "loss": 1.1909, "step": 10930 }, { "epoch": 0.01, "learning_rate": 4.998552589156055e-05, "loss": 1.4488, "step": 10931 }, { "epoch": 0.01, "learning_rate": 4.99855232189764e-05, "loss": 1.6222, "step": 10932 }, { "epoch": 0.01, "learning_rate": 4.998552054614561e-05, "loss": 1.1969, "step": 10933 }, { "epoch": 0.01, "learning_rate": 4.9985517873068156e-05, "loss": 0.8274, "step": 10934 }, { "epoch": 0.01, "learning_rate": 4.998551519974407e-05, "loss": 1.3114, "step": 10935 }, { "epoch": 0.01, "learning_rate": 4.998551252617333e-05, "loss": 0.9761, "step": 10936 }, { "epoch": 0.01, "learning_rate": 4.998550985235595e-05, "loss": 1.1793, "step": 10937 }, { "epoch": 0.01, "learning_rate": 4.998550717829192e-05, "loss": 0.4267, "step": 10938 }, { "epoch": 0.01, "learning_rate": 4.998550450398124e-05, "loss": 0.2122, "step": 10939 }, { "epoch": 0.01, "learning_rate": 4.998550182942392e-05, "loss": 0.2233, "step": 10940 }, { "epoch": 0.01, "learning_rate": 4.998549915461995e-05, "loss": 0.1119, "step": 10941 }, { "epoch": 0.01, "learning_rate": 4.998549647956934e-05, "loss": 0.0817, "step": 10942 }, { "epoch": 0.01, "learning_rate": 4.998549380427208e-05, "loss": 0.1277, "step": 10943 }, { "epoch": 0.01, "learning_rate": 4.998549112872817e-05, "loss": 0.0484, "step": 10944 }, { "epoch": 0.01, "learning_rate": 4.998548845293761e-05, "loss": 0.0294, "step": 10945 }, { "epoch": 0.01, "learning_rate": 4.998548577690041e-05, "loss": 0.2159, "step": 10946 }, { "epoch": 0.01, "learning_rate": 4.998548310061656e-05, "loss": 1.1548, "step": 10947 }, { "epoch": 0.01, "learning_rate": 4.9985480424086076e-05, "loss": 1.1134, "step": 10948 }, { "epoch": 0.01, "learning_rate": 4.998547774730893e-05, "loss": 0.8281, "step": 10949 }, { "epoch": 0.01, "learning_rate": 4.998547507028515e-05, "loss": 1.2531, "step": 10950 }, { "epoch": 0.01, "learning_rate": 4.998547239301472e-05, "loss": 0.9875, "step": 10951 }, { "epoch": 0.01, "learning_rate": 4.998546971549764e-05, "loss": 0.9606, "step": 10952 }, { "epoch": 0.01, "learning_rate": 4.9985467037733916e-05, "loss": 0.5359, "step": 10953 }, { "epoch": 0.01, "learning_rate": 4.998546435972354e-05, "loss": 0.8507, "step": 10954 }, { "epoch": 0.01, "learning_rate": 4.9985461681466525e-05, "loss": 1.152, "step": 10955 }, { "epoch": 0.01, "learning_rate": 4.998545900296287e-05, "loss": 1.2994, "step": 10956 }, { "epoch": 0.01, "learning_rate": 4.9985456324212564e-05, "loss": 1.015, "step": 10957 }, { "epoch": 0.01, "learning_rate": 4.998545364521561e-05, "loss": 1.0323, "step": 10958 }, { "epoch": 0.01, "learning_rate": 4.9985450965972014e-05, "loss": 1.0276, "step": 10959 }, { "epoch": 0.01, "learning_rate": 4.9985448286481764e-05, "loss": 0.9741, "step": 10960 }, { "epoch": 0.01, "learning_rate": 4.998544560674487e-05, "loss": 0.5701, "step": 10961 }, { "epoch": 0.01, "learning_rate": 4.998544292676134e-05, "loss": 1.1095, "step": 10962 }, { "epoch": 0.01, "learning_rate": 4.9985440246531155e-05, "loss": 1.0364, "step": 10963 }, { "epoch": 0.01, "learning_rate": 4.998543756605433e-05, "loss": 1.1118, "step": 10964 }, { "epoch": 0.01, "learning_rate": 4.998543488533086e-05, "loss": 1.0536, "step": 10965 }, { "epoch": 0.01, "learning_rate": 4.998543220436074e-05, "loss": 1.1477, "step": 10966 }, { "epoch": 0.01, "learning_rate": 4.998542952314398e-05, "loss": 1.2684, "step": 10967 }, { "epoch": 0.01, "learning_rate": 4.998542684168056e-05, "loss": 1.2744, "step": 10968 }, { "epoch": 0.01, "learning_rate": 4.99854241599705e-05, "loss": 1.2041, "step": 10969 }, { "epoch": 0.01, "learning_rate": 4.99854214780138e-05, "loss": 1.0253, "step": 10970 }, { "epoch": 0.01, "learning_rate": 4.998541879581046e-05, "loss": 0.9462, "step": 10971 }, { "epoch": 0.01, "learning_rate": 4.998541611336047e-05, "loss": 1.1174, "step": 10972 }, { "epoch": 0.01, "learning_rate": 4.998541343066383e-05, "loss": 0.7509, "step": 10973 }, { "epoch": 0.01, "learning_rate": 4.998541074772055e-05, "loss": 1.1098, "step": 10974 }, { "epoch": 0.01, "learning_rate": 4.9985408064530615e-05, "loss": 1.0753, "step": 10975 }, { "epoch": 0.01, "learning_rate": 4.998540538109404e-05, "loss": 0.9133, "step": 10976 }, { "epoch": 0.01, "learning_rate": 4.998540269741082e-05, "loss": 1.1574, "step": 10977 }, { "epoch": 0.01, "learning_rate": 4.9985400013480954e-05, "loss": 1.0544, "step": 10978 }, { "epoch": 0.01, "learning_rate": 4.9985397329304444e-05, "loss": 0.5128, "step": 10979 }, { "epoch": 0.01, "learning_rate": 4.998539464488129e-05, "loss": 0.4893, "step": 10980 }, { "epoch": 0.01, "learning_rate": 4.998539196021149e-05, "loss": 1.188, "step": 10981 }, { "epoch": 0.01, "learning_rate": 4.998538927529504e-05, "loss": 1.3035, "step": 10982 }, { "epoch": 0.01, "learning_rate": 4.9985386590131946e-05, "loss": 1.0466, "step": 10983 }, { "epoch": 0.01, "learning_rate": 4.998538390472222e-05, "loss": 1.345, "step": 10984 }, { "epoch": 0.01, "learning_rate": 4.998538121906583e-05, "loss": 0.963, "step": 10985 }, { "epoch": 0.01, "learning_rate": 4.9985378533162806e-05, "loss": 1.0332, "step": 10986 }, { "epoch": 0.01, "learning_rate": 4.9985375847013136e-05, "loss": 0.6341, "step": 10987 }, { "epoch": 0.01, "learning_rate": 4.998537316061681e-05, "loss": 1.2144, "step": 10988 }, { "epoch": 0.01, "learning_rate": 4.9985370473973856e-05, "loss": 1.1795, "step": 10989 }, { "epoch": 0.01, "learning_rate": 4.9985367787084245e-05, "loss": 0.9995, "step": 10990 }, { "epoch": 0.01, "learning_rate": 4.998536509994799e-05, "loss": 1.219, "step": 10991 }, { "epoch": 0.01, "learning_rate": 4.99853624125651e-05, "loss": 1.206, "step": 10992 }, { "epoch": 0.01, "learning_rate": 4.998535972493555e-05, "loss": 1.2541, "step": 10993 }, { "epoch": 0.01, "learning_rate": 4.9985357037059365e-05, "loss": 2.1262, "step": 10994 }, { "epoch": 0.01, "learning_rate": 4.998535434893653e-05, "loss": 1.2823, "step": 10995 }, { "epoch": 0.01, "learning_rate": 4.998535166056706e-05, "loss": 0.9986, "step": 10996 }, { "epoch": 0.01, "learning_rate": 4.998534897195093e-05, "loss": 1.1551, "step": 10997 }, { "epoch": 0.01, "learning_rate": 4.9985346283088165e-05, "loss": 1.1561, "step": 10998 }, { "epoch": 0.01, "learning_rate": 4.998534359397875e-05, "loss": 1.1731, "step": 10999 }, { "epoch": 0.01, "learning_rate": 4.99853409046227e-05, "loss": 1.0143, "step": 11000 }, { "epoch": 0.01, "eval_loss": 1.0610970258712769, "eval_runtime": 86.5342, "eval_samples_per_second": 16.005, "eval_steps_per_second": 4.01, "step": 11000 }, { "epoch": 0.01, "learning_rate": 4.998533821501999e-05, "loss": 1.0851, "step": 11001 }, { "epoch": 0.01, "learning_rate": 4.998533552517065e-05, "loss": 1.0072, "step": 11002 }, { "epoch": 0.01, "learning_rate": 4.998533283507466e-05, "loss": 1.0418, "step": 11003 }, { "epoch": 0.01, "learning_rate": 4.9985330144732024e-05, "loss": 0.8995, "step": 11004 }, { "epoch": 0.01, "learning_rate": 4.998532745414274e-05, "loss": 0.4762, "step": 11005 }, { "epoch": 0.01, "learning_rate": 4.998532476330682e-05, "loss": 1.2354, "step": 11006 }, { "epoch": 0.01, "learning_rate": 4.998532207222425e-05, "loss": 1.5269, "step": 11007 }, { "epoch": 0.01, "learning_rate": 4.998531938089503e-05, "loss": 1.3113, "step": 11008 }, { "epoch": 0.01, "learning_rate": 4.998531668931918e-05, "loss": 1.17, "step": 11009 }, { "epoch": 0.01, "learning_rate": 4.9985313997496674e-05, "loss": 1.1982, "step": 11010 }, { "epoch": 0.01, "learning_rate": 4.998531130542753e-05, "loss": 0.9628, "step": 11011 }, { "epoch": 0.01, "learning_rate": 4.9985308613111733e-05, "loss": 1.1221, "step": 11012 }, { "epoch": 0.01, "learning_rate": 4.99853059205493e-05, "loss": 1.0852, "step": 11013 }, { "epoch": 0.01, "learning_rate": 4.998530322774022e-05, "loss": 1.0638, "step": 11014 }, { "epoch": 0.01, "learning_rate": 4.998530053468449e-05, "loss": 1.4711, "step": 11015 }, { "epoch": 0.01, "learning_rate": 4.998529784138212e-05, "loss": 1.0756, "step": 11016 }, { "epoch": 0.01, "learning_rate": 4.998529514783311e-05, "loss": 1.2878, "step": 11017 }, { "epoch": 0.01, "learning_rate": 4.9985292454037445e-05, "loss": 1.5491, "step": 11018 }, { "epoch": 0.01, "learning_rate": 4.998528975999515e-05, "loss": 1.0756, "step": 11019 }, { "epoch": 0.01, "learning_rate": 4.99852870657062e-05, "loss": 1.0738, "step": 11020 }, { "epoch": 0.01, "learning_rate": 4.998528437117061e-05, "loss": 0.9079, "step": 11021 }, { "epoch": 0.01, "learning_rate": 4.9985281676388374e-05, "loss": 1.3341, "step": 11022 }, { "epoch": 0.01, "learning_rate": 4.9985278981359495e-05, "loss": 1.7209, "step": 11023 }, { "epoch": 0.01, "learning_rate": 4.9985276286083974e-05, "loss": 1.4991, "step": 11024 }, { "epoch": 0.01, "learning_rate": 4.99852735905618e-05, "loss": 0.9747, "step": 11025 }, { "epoch": 0.01, "learning_rate": 4.998527089479299e-05, "loss": 1.0822, "step": 11026 }, { "epoch": 0.01, "learning_rate": 4.9985268198777535e-05, "loss": 1.0989, "step": 11027 }, { "epoch": 0.01, "learning_rate": 4.998526550251544e-05, "loss": 0.8828, "step": 11028 }, { "epoch": 0.01, "learning_rate": 4.9985262806006696e-05, "loss": 0.5451, "step": 11029 }, { "epoch": 0.01, "learning_rate": 4.998526010925131e-05, "loss": 1.0593, "step": 11030 }, { "epoch": 0.01, "learning_rate": 4.9985257412249275e-05, "loss": 1.2572, "step": 11031 }, { "epoch": 0.01, "learning_rate": 4.9985254715000594e-05, "loss": 1.1468, "step": 11032 }, { "epoch": 0.01, "learning_rate": 4.998525201750528e-05, "loss": 1.3098, "step": 11033 }, { "epoch": 0.01, "learning_rate": 4.998524931976332e-05, "loss": 1.1122, "step": 11034 }, { "epoch": 0.01, "learning_rate": 4.998524662177471e-05, "loss": 0.8548, "step": 11035 }, { "epoch": 0.01, "learning_rate": 4.998524392353946e-05, "loss": 0.9925, "step": 11036 }, { "epoch": 0.01, "learning_rate": 4.9985241225057565e-05, "loss": 1.0917, "step": 11037 }, { "epoch": 0.01, "learning_rate": 4.998523852632903e-05, "loss": 1.0632, "step": 11038 }, { "epoch": 0.01, "learning_rate": 4.9985235827353844e-05, "loss": 1.0606, "step": 11039 }, { "epoch": 0.01, "learning_rate": 4.9985233128132016e-05, "loss": 0.9986, "step": 11040 }, { "epoch": 0.01, "learning_rate": 4.9985230428663546e-05, "loss": 1.0103, "step": 11041 }, { "epoch": 0.01, "learning_rate": 4.9985227728948434e-05, "loss": 0.67, "step": 11042 }, { "epoch": 0.01, "learning_rate": 4.998522502898668e-05, "loss": 1.176, "step": 11043 }, { "epoch": 0.01, "learning_rate": 4.9985222328778275e-05, "loss": 1.2349, "step": 11044 }, { "epoch": 0.01, "learning_rate": 4.9985219628323235e-05, "loss": 1.3533, "step": 11045 }, { "epoch": 0.01, "learning_rate": 4.998521692762154e-05, "loss": 1.4718, "step": 11046 }, { "epoch": 0.01, "learning_rate": 4.998521422667321e-05, "loss": 0.7743, "step": 11047 }, { "epoch": 0.01, "learning_rate": 4.9985211525478235e-05, "loss": 1.2536, "step": 11048 }, { "epoch": 0.01, "learning_rate": 4.998520882403662e-05, "loss": 1.1083, "step": 11049 }, { "epoch": 0.01, "learning_rate": 4.998520612234835e-05, "loss": 1.0355, "step": 11050 }, { "epoch": 0.01, "learning_rate": 4.998520342041345e-05, "loss": 1.0358, "step": 11051 }, { "epoch": 0.01, "learning_rate": 4.998520071823191e-05, "loss": 0.9751, "step": 11052 }, { "epoch": 0.01, "learning_rate": 4.9985198015803715e-05, "loss": 0.9187, "step": 11053 }, { "epoch": 0.01, "learning_rate": 4.998519531312887e-05, "loss": 1.0517, "step": 11054 }, { "epoch": 0.01, "learning_rate": 4.9985192610207396e-05, "loss": 1.2088, "step": 11055 }, { "epoch": 0.01, "learning_rate": 4.9985189907039276e-05, "loss": 1.1475, "step": 11056 }, { "epoch": 0.01, "learning_rate": 4.9985187203624506e-05, "loss": 0.7752, "step": 11057 }, { "epoch": 0.01, "learning_rate": 4.99851844999631e-05, "loss": 0.8341, "step": 11058 }, { "epoch": 0.01, "learning_rate": 4.998518179605505e-05, "loss": 0.8604, "step": 11059 }, { "epoch": 0.01, "learning_rate": 4.998517909190035e-05, "loss": 0.9871, "step": 11060 }, { "epoch": 0.01, "learning_rate": 4.998517638749901e-05, "loss": 0.514, "step": 11061 }, { "epoch": 0.01, "learning_rate": 4.998517368285103e-05, "loss": 1.3165, "step": 11062 }, { "epoch": 0.01, "learning_rate": 4.9985170977956406e-05, "loss": 1.1155, "step": 11063 }, { "epoch": 0.01, "learning_rate": 4.998516827281514e-05, "loss": 0.7402, "step": 11064 }, { "epoch": 0.01, "learning_rate": 4.998516556742723e-05, "loss": 0.7353, "step": 11065 }, { "epoch": 0.01, "learning_rate": 4.998516286179268e-05, "loss": 0.6377, "step": 11066 }, { "epoch": 0.01, "learning_rate": 4.998516015591148e-05, "loss": 0.6585, "step": 11067 }, { "epoch": 0.01, "learning_rate": 4.998515744978364e-05, "loss": 0.6662, "step": 11068 }, { "epoch": 0.01, "learning_rate": 4.998515474340916e-05, "loss": 0.6446, "step": 11069 }, { "epoch": 0.01, "learning_rate": 4.9985152036788036e-05, "loss": 0.6433, "step": 11070 }, { "epoch": 0.01, "learning_rate": 4.9985149329920266e-05, "loss": 0.6352, "step": 11071 }, { "epoch": 0.01, "learning_rate": 4.998514662280585e-05, "loss": 0.6851, "step": 11072 }, { "epoch": 0.01, "learning_rate": 4.99851439154448e-05, "loss": 0.6676, "step": 11073 }, { "epoch": 0.01, "learning_rate": 4.998514120783711e-05, "loss": 0.6724, "step": 11074 }, { "epoch": 0.01, "learning_rate": 4.998513849998277e-05, "loss": 0.6671, "step": 11075 }, { "epoch": 0.01, "learning_rate": 4.9985135791881784e-05, "loss": 0.7058, "step": 11076 }, { "epoch": 0.01, "learning_rate": 4.998513308353416e-05, "loss": 0.7303, "step": 11077 }, { "epoch": 0.01, "learning_rate": 4.998513037493989e-05, "loss": 0.7453, "step": 11078 }, { "epoch": 0.01, "learning_rate": 4.998512766609898e-05, "loss": 0.8886, "step": 11079 }, { "epoch": 0.01, "learning_rate": 4.998512495701142e-05, "loss": 1.077, "step": 11080 }, { "epoch": 0.01, "learning_rate": 4.998512224767723e-05, "loss": 1.3381, "step": 11081 }, { "epoch": 0.01, "learning_rate": 4.998511953809639e-05, "loss": 1.021, "step": 11082 }, { "epoch": 0.01, "learning_rate": 4.9985116828268905e-05, "loss": 1.1992, "step": 11083 }, { "epoch": 0.01, "learning_rate": 4.998511411819479e-05, "loss": 1.2632, "step": 11084 }, { "epoch": 0.01, "learning_rate": 4.998511140787402e-05, "loss": 0.8558, "step": 11085 }, { "epoch": 0.01, "learning_rate": 4.998510869730662e-05, "loss": 0.8319, "step": 11086 }, { "epoch": 0.01, "learning_rate": 4.998510598649256e-05, "loss": 0.8487, "step": 11087 }, { "epoch": 0.01, "learning_rate": 4.9985103275431863e-05, "loss": 1.1844, "step": 11088 }, { "epoch": 0.01, "learning_rate": 4.998510056412453e-05, "loss": 1.1179, "step": 11089 }, { "epoch": 0.01, "learning_rate": 4.998509785257056e-05, "loss": 1.1867, "step": 11090 }, { "epoch": 0.01, "learning_rate": 4.9985095140769933e-05, "loss": 1.2414, "step": 11091 }, { "epoch": 0.01, "learning_rate": 4.998509242872267e-05, "loss": 1.1789, "step": 11092 }, { "epoch": 0.01, "learning_rate": 4.998508971642877e-05, "loss": 1.3805, "step": 11093 }, { "epoch": 0.01, "learning_rate": 4.998508700388822e-05, "loss": 0.9683, "step": 11094 }, { "epoch": 0.01, "learning_rate": 4.998508429110103e-05, "loss": 1.2478, "step": 11095 }, { "epoch": 0.01, "learning_rate": 4.9985081578067194e-05, "loss": 1.7467, "step": 11096 }, { "epoch": 0.01, "learning_rate": 4.998507886478673e-05, "loss": 1.1114, "step": 11097 }, { "epoch": 0.01, "learning_rate": 4.9985076151259604e-05, "loss": 0.9845, "step": 11098 }, { "epoch": 0.01, "learning_rate": 4.998507343748585e-05, "loss": 0.4566, "step": 11099 }, { "epoch": 0.01, "learning_rate": 4.998507072346545e-05, "loss": 0.8858, "step": 11100 }, { "epoch": 0.01, "learning_rate": 4.998506800919841e-05, "loss": 1.0575, "step": 11101 }, { "epoch": 0.01, "learning_rate": 4.998506529468472e-05, "loss": 1.1572, "step": 11102 }, { "epoch": 0.01, "learning_rate": 4.9985062579924393e-05, "loss": 1.1657, "step": 11103 }, { "epoch": 0.01, "learning_rate": 4.998505986491743e-05, "loss": 0.8033, "step": 11104 }, { "epoch": 0.01, "learning_rate": 4.9985057149663816e-05, "loss": 0.7907, "step": 11105 }, { "epoch": 0.01, "learning_rate": 4.998505443416357e-05, "loss": 1.0515, "step": 11106 }, { "epoch": 0.01, "learning_rate": 4.998505171841667e-05, "loss": 1.2635, "step": 11107 }, { "epoch": 0.01, "learning_rate": 4.9985049002423135e-05, "loss": 1.2101, "step": 11108 }, { "epoch": 0.01, "learning_rate": 4.998504628618296e-05, "loss": 1.2433, "step": 11109 }, { "epoch": 0.01, "learning_rate": 4.998504356969613e-05, "loss": 1.0422, "step": 11110 }, { "epoch": 0.01, "learning_rate": 4.998504085296267e-05, "loss": 1.072, "step": 11111 }, { "epoch": 0.01, "learning_rate": 4.998503813598256e-05, "loss": 1.0027, "step": 11112 }, { "epoch": 0.01, "learning_rate": 4.998503541875582e-05, "loss": 1.4205, "step": 11113 }, { "epoch": 0.01, "learning_rate": 4.9985032701282434e-05, "loss": 1.4059, "step": 11114 }, { "epoch": 0.01, "learning_rate": 4.99850299835624e-05, "loss": 1.1584, "step": 11115 }, { "epoch": 0.01, "learning_rate": 4.9985027265595736e-05, "loss": 1.2866, "step": 11116 }, { "epoch": 0.01, "learning_rate": 4.998502454738242e-05, "loss": 1.1668, "step": 11117 }, { "epoch": 0.01, "learning_rate": 4.998502182892247e-05, "loss": 1.2536, "step": 11118 }, { "epoch": 0.01, "learning_rate": 4.9985019110215875e-05, "loss": 1.6954, "step": 11119 }, { "epoch": 0.01, "learning_rate": 4.998501639126264e-05, "loss": 1.1413, "step": 11120 }, { "epoch": 0.01, "learning_rate": 4.998501367206276e-05, "loss": 0.9921, "step": 11121 }, { "epoch": 0.01, "learning_rate": 4.998501095261624e-05, "loss": 0.9011, "step": 11122 }, { "epoch": 0.01, "learning_rate": 4.9985008232923075e-05, "loss": 1.3716, "step": 11123 }, { "epoch": 0.01, "learning_rate": 4.9985005512983275e-05, "loss": 1.0429, "step": 11124 }, { "epoch": 0.01, "learning_rate": 4.9985002792796834e-05, "loss": 3.123, "step": 11125 }, { "epoch": 0.01, "learning_rate": 4.998500007236374e-05, "loss": 2.4726, "step": 11126 }, { "epoch": 0.01, "learning_rate": 4.9984997351684016e-05, "loss": 0.9741, "step": 11127 }, { "epoch": 0.01, "learning_rate": 4.998499463075764e-05, "loss": 0.3929, "step": 11128 }, { "epoch": 0.01, "learning_rate": 4.9984991909584635e-05, "loss": 1.1812, "step": 11129 }, { "epoch": 0.01, "learning_rate": 4.998498918816499e-05, "loss": 1.3151, "step": 11130 }, { "epoch": 0.01, "learning_rate": 4.998498646649869e-05, "loss": 1.2314, "step": 11131 }, { "epoch": 0.01, "learning_rate": 4.998498374458576e-05, "loss": 1.1442, "step": 11132 }, { "epoch": 0.01, "learning_rate": 4.998498102242618e-05, "loss": 1.1083, "step": 11133 }, { "epoch": 0.01, "learning_rate": 4.998497830001997e-05, "loss": 0.9796, "step": 11134 }, { "epoch": 0.01, "learning_rate": 4.998497557736711e-05, "loss": 1.0991, "step": 11135 }, { "epoch": 0.01, "learning_rate": 4.998497285446761e-05, "loss": 1.0666, "step": 11136 }, { "epoch": 0.01, "learning_rate": 4.998497013132147e-05, "loss": 1.2087, "step": 11137 }, { "epoch": 0.01, "learning_rate": 4.998496740792868e-05, "loss": 1.1696, "step": 11138 }, { "epoch": 0.01, "learning_rate": 4.9984964684289266e-05, "loss": 1.259, "step": 11139 }, { "epoch": 0.01, "learning_rate": 4.99849619604032e-05, "loss": 1.1692, "step": 11140 }, { "epoch": 0.01, "learning_rate": 4.99849592362705e-05, "loss": 1.1599, "step": 11141 }, { "epoch": 0.01, "learning_rate": 4.998495651189116e-05, "loss": 1.1001, "step": 11142 }, { "epoch": 0.01, "learning_rate": 4.9984953787265166e-05, "loss": 1.228, "step": 11143 }, { "epoch": 0.01, "learning_rate": 4.998495106239254e-05, "loss": 0.9668, "step": 11144 }, { "epoch": 0.01, "learning_rate": 4.9984948337273274e-05, "loss": 0.8809, "step": 11145 }, { "epoch": 0.01, "learning_rate": 4.998494561190736e-05, "loss": 1.2722, "step": 11146 }, { "epoch": 0.01, "learning_rate": 4.998494288629482e-05, "loss": 1.1455, "step": 11147 }, { "epoch": 0.01, "learning_rate": 4.998494016043562e-05, "loss": 0.994, "step": 11148 }, { "epoch": 0.01, "learning_rate": 4.9984937434329796e-05, "loss": 1.2156, "step": 11149 }, { "epoch": 0.01, "learning_rate": 4.998493470797731e-05, "loss": 1.0701, "step": 11150 }, { "epoch": 0.01, "learning_rate": 4.99849319813782e-05, "loss": 1.0988, "step": 11151 }, { "epoch": 0.01, "learning_rate": 4.9984929254532454e-05, "loss": 0.8265, "step": 11152 }, { "epoch": 0.01, "learning_rate": 4.998492652744006e-05, "loss": 1.4351, "step": 11153 }, { "epoch": 0.01, "learning_rate": 4.9984923800101026e-05, "loss": 0.8209, "step": 11154 }, { "epoch": 0.01, "learning_rate": 4.9984921072515345e-05, "loss": 1.0785, "step": 11155 }, { "epoch": 0.01, "learning_rate": 4.998491834468303e-05, "loss": 1.0431, "step": 11156 }, { "epoch": 0.01, "learning_rate": 4.9984915616604076e-05, "loss": 1.0766, "step": 11157 }, { "epoch": 0.01, "learning_rate": 4.998491288827848e-05, "loss": 1.1995, "step": 11158 }, { "epoch": 0.01, "learning_rate": 4.998491015970624e-05, "loss": 0.9651, "step": 11159 }, { "epoch": 0.01, "learning_rate": 4.9984907430887363e-05, "loss": 0.7261, "step": 11160 }, { "epoch": 0.01, "learning_rate": 4.998490470182184e-05, "loss": 1.0632, "step": 11161 }, { "epoch": 0.01, "learning_rate": 4.998490197250968e-05, "loss": 1.1548, "step": 11162 }, { "epoch": 0.01, "learning_rate": 4.998489924295088e-05, "loss": 1.144, "step": 11163 }, { "epoch": 0.01, "learning_rate": 4.998489651314544e-05, "loss": 1.1877, "step": 11164 }, { "epoch": 0.01, "learning_rate": 4.998489378309337e-05, "loss": 0.7849, "step": 11165 }, { "epoch": 0.01, "learning_rate": 4.998489105279464e-05, "loss": 1.0305, "step": 11166 }, { "epoch": 0.01, "learning_rate": 4.998488832224928e-05, "loss": 1.0139, "step": 11167 }, { "epoch": 0.01, "learning_rate": 4.998488559145728e-05, "loss": 1.0266, "step": 11168 }, { "epoch": 0.01, "learning_rate": 4.9984882860418637e-05, "loss": 1.2371, "step": 11169 }, { "epoch": 0.01, "learning_rate": 4.998488012913336e-05, "loss": 1.2292, "step": 11170 }, { "epoch": 0.01, "learning_rate": 4.998487739760144e-05, "loss": 1.25, "step": 11171 }, { "epoch": 0.01, "learning_rate": 4.998487466582287e-05, "loss": 1.1316, "step": 11172 }, { "epoch": 0.01, "learning_rate": 4.998487193379767e-05, "loss": 1.1257, "step": 11173 }, { "epoch": 0.01, "learning_rate": 4.998486920152583e-05, "loss": 0.9874, "step": 11174 }, { "epoch": 0.01, "learning_rate": 4.9984866469007344e-05, "loss": 0.9263, "step": 11175 }, { "epoch": 0.01, "learning_rate": 4.998486373624222e-05, "loss": 1.2612, "step": 11176 }, { "epoch": 0.01, "learning_rate": 4.998486100323045e-05, "loss": 1.3142, "step": 11177 }, { "epoch": 0.01, "learning_rate": 4.9984858269972055e-05, "loss": 1.019, "step": 11178 }, { "epoch": 0.01, "learning_rate": 4.998485553646701e-05, "loss": 0.9364, "step": 11179 }, { "epoch": 0.01, "learning_rate": 4.9984852802715335e-05, "loss": 0.7494, "step": 11180 }, { "epoch": 0.01, "learning_rate": 4.9984850068717004e-05, "loss": 0.9415, "step": 11181 }, { "epoch": 0.01, "learning_rate": 4.9984847334472045e-05, "loss": 1.0966, "step": 11182 }, { "epoch": 0.01, "learning_rate": 4.998484459998044e-05, "loss": 1.0315, "step": 11183 }, { "epoch": 0.01, "learning_rate": 4.99848418652422e-05, "loss": 1.1631, "step": 11184 }, { "epoch": 0.01, "learning_rate": 4.998483913025732e-05, "loss": 1.348, "step": 11185 }, { "epoch": 0.01, "learning_rate": 4.998483639502579e-05, "loss": 1.2972, "step": 11186 }, { "epoch": 0.01, "learning_rate": 4.998483365954764e-05, "loss": 0.8458, "step": 11187 }, { "epoch": 0.01, "learning_rate": 4.998483092382283e-05, "loss": 0.9921, "step": 11188 }, { "epoch": 0.01, "learning_rate": 4.998482818785139e-05, "loss": 1.1262, "step": 11189 }, { "epoch": 0.01, "learning_rate": 4.998482545163331e-05, "loss": 0.9357, "step": 11190 }, { "epoch": 0.01, "learning_rate": 4.998482271516859e-05, "loss": 1.1249, "step": 11191 }, { "epoch": 0.01, "learning_rate": 4.998481997845723e-05, "loss": 0.9263, "step": 11192 }, { "epoch": 0.01, "learning_rate": 4.9984817241499227e-05, "loss": 1.1577, "step": 11193 }, { "epoch": 0.01, "learning_rate": 4.9984814504294585e-05, "loss": 1.2202, "step": 11194 }, { "epoch": 0.01, "learning_rate": 4.998481176684331e-05, "loss": 1.1899, "step": 11195 }, { "epoch": 0.01, "learning_rate": 4.998480902914539e-05, "loss": 1.3262, "step": 11196 }, { "epoch": 0.01, "learning_rate": 4.998480629120084e-05, "loss": 1.3787, "step": 11197 }, { "epoch": 0.01, "learning_rate": 4.9984803553009635e-05, "loss": 1.1495, "step": 11198 }, { "epoch": 0.01, "learning_rate": 4.9984800814571795e-05, "loss": 1.1057, "step": 11199 }, { "epoch": 0.01, "learning_rate": 4.9984798075887326e-05, "loss": 1.1455, "step": 11200 }, { "epoch": 0.01, "learning_rate": 4.99847953369562e-05, "loss": 1.1896, "step": 11201 }, { "epoch": 0.01, "learning_rate": 4.9984792597778453e-05, "loss": 1.0, "step": 11202 }, { "epoch": 0.01, "learning_rate": 4.998478985835405e-05, "loss": 1.0723, "step": 11203 }, { "epoch": 0.01, "learning_rate": 4.9984787118683025e-05, "loss": 1.1118, "step": 11204 }, { "epoch": 0.01, "learning_rate": 4.998478437876535e-05, "loss": 1.128, "step": 11205 }, { "epoch": 0.01, "learning_rate": 4.998478163860104e-05, "loss": 1.2559, "step": 11206 }, { "epoch": 0.01, "learning_rate": 4.998477889819009e-05, "loss": 1.1666, "step": 11207 }, { "epoch": 0.01, "learning_rate": 4.99847761575325e-05, "loss": 1.0831, "step": 11208 }, { "epoch": 0.01, "learning_rate": 4.998477341662826e-05, "loss": 1.1202, "step": 11209 }, { "epoch": 0.01, "learning_rate": 4.99847706754774e-05, "loss": 0.7362, "step": 11210 }, { "epoch": 0.01, "learning_rate": 4.998476793407989e-05, "loss": 0.8456, "step": 11211 }, { "epoch": 0.01, "learning_rate": 4.998476519243574e-05, "loss": 0.4883, "step": 11212 }, { "epoch": 0.01, "learning_rate": 4.9984762450544955e-05, "loss": 1.166, "step": 11213 }, { "epoch": 0.01, "learning_rate": 4.9984759708407526e-05, "loss": 1.0985, "step": 11214 }, { "epoch": 0.01, "learning_rate": 4.998475696602346e-05, "loss": 0.7811, "step": 11215 }, { "epoch": 0.01, "learning_rate": 4.998475422339276e-05, "loss": 1.166, "step": 11216 }, { "epoch": 0.01, "learning_rate": 4.998475148051542e-05, "loss": 1.0113, "step": 11217 }, { "epoch": 0.01, "learning_rate": 4.998474873739144e-05, "loss": 1.2124, "step": 11218 }, { "epoch": 0.01, "learning_rate": 4.998474599402082e-05, "loss": 1.0753, "step": 11219 }, { "epoch": 0.01, "learning_rate": 4.998474325040355e-05, "loss": 1.2656, "step": 11220 }, { "epoch": 0.01, "learning_rate": 4.9984740506539654e-05, "loss": 1.1407, "step": 11221 }, { "epoch": 0.01, "learning_rate": 4.9984737762429125e-05, "loss": 0.7786, "step": 11222 }, { "epoch": 0.01, "learning_rate": 4.998473501807195e-05, "loss": 0.963, "step": 11223 }, { "epoch": 0.01, "learning_rate": 4.998473227346813e-05, "loss": 1.0386, "step": 11224 }, { "epoch": 0.01, "learning_rate": 4.998472952861768e-05, "loss": 1.1661, "step": 11225 }, { "epoch": 0.01, "learning_rate": 4.998472678352059e-05, "loss": 1.217, "step": 11226 }, { "epoch": 0.01, "learning_rate": 4.9984724038176854e-05, "loss": 0.7382, "step": 11227 }, { "epoch": 0.01, "learning_rate": 4.9984721292586485e-05, "loss": 1.2549, "step": 11228 }, { "epoch": 0.01, "learning_rate": 4.998471854674947e-05, "loss": 0.9207, "step": 11229 }, { "epoch": 0.01, "learning_rate": 4.9984715800665825e-05, "loss": 1.0814, "step": 11230 }, { "epoch": 0.01, "learning_rate": 4.998471305433554e-05, "loss": 1.1673, "step": 11231 }, { "epoch": 0.01, "learning_rate": 4.998471030775862e-05, "loss": 1.0175, "step": 11232 }, { "epoch": 0.01, "learning_rate": 4.998470756093506e-05, "loss": 1.1739, "step": 11233 }, { "epoch": 0.01, "learning_rate": 4.998470481386486e-05, "loss": 0.8953, "step": 11234 }, { "epoch": 0.01, "learning_rate": 4.998470206654802e-05, "loss": 1.0875, "step": 11235 }, { "epoch": 0.01, "learning_rate": 4.998469931898454e-05, "loss": 1.0619, "step": 11236 }, { "epoch": 0.01, "learning_rate": 4.998469657117443e-05, "loss": 0.7886, "step": 11237 }, { "epoch": 0.01, "learning_rate": 4.998469382311767e-05, "loss": 0.75, "step": 11238 }, { "epoch": 0.01, "learning_rate": 4.998469107481428e-05, "loss": 1.1156, "step": 11239 }, { "epoch": 0.01, "learning_rate": 4.998468832626425e-05, "loss": 1.859, "step": 11240 }, { "epoch": 0.01, "learning_rate": 4.998468557746758e-05, "loss": 1.4037, "step": 11241 }, { "epoch": 0.01, "learning_rate": 4.998468282842427e-05, "loss": 1.1006, "step": 11242 }, { "epoch": 0.01, "learning_rate": 4.998468007913432e-05, "loss": 0.8764, "step": 11243 }, { "epoch": 0.01, "learning_rate": 4.998467732959774e-05, "loss": 1.0423, "step": 11244 }, { "epoch": 0.01, "learning_rate": 4.9984674579814515e-05, "loss": 0.9886, "step": 11245 }, { "epoch": 0.01, "learning_rate": 4.9984671829784656e-05, "loss": 0.8288, "step": 11246 }, { "epoch": 0.01, "learning_rate": 4.9984669079508155e-05, "loss": 1.3285, "step": 11247 }, { "epoch": 0.01, "learning_rate": 4.998466632898502e-05, "loss": 1.2376, "step": 11248 }, { "epoch": 0.01, "learning_rate": 4.998466357821524e-05, "loss": 1.1547, "step": 11249 }, { "epoch": 0.01, "learning_rate": 4.998466082719883e-05, "loss": 1.1883, "step": 11250 }, { "epoch": 0.01, "learning_rate": 4.998465807593578e-05, "loss": 0.918, "step": 11251 }, { "epoch": 0.01, "learning_rate": 4.998465532442609e-05, "loss": 1.1977, "step": 11252 }, { "epoch": 0.01, "learning_rate": 4.9984652572669765e-05, "loss": 1.2453, "step": 11253 }, { "epoch": 0.01, "learning_rate": 4.9984649820666794e-05, "loss": 1.1108, "step": 11254 }, { "epoch": 0.01, "learning_rate": 4.9984647068417193e-05, "loss": 1.2474, "step": 11255 }, { "epoch": 0.01, "learning_rate": 4.998464431592095e-05, "loss": 1.0577, "step": 11256 }, { "epoch": 0.01, "learning_rate": 4.998464156317807e-05, "loss": 1.0733, "step": 11257 }, { "epoch": 0.01, "learning_rate": 4.998463881018856e-05, "loss": 0.9653, "step": 11258 }, { "epoch": 0.01, "learning_rate": 4.99846360569524e-05, "loss": 1.0213, "step": 11259 }, { "epoch": 0.01, "learning_rate": 4.998463330346961e-05, "loss": 0.9565, "step": 11260 }, { "epoch": 0.01, "learning_rate": 4.998463054974018e-05, "loss": 1.2118, "step": 11261 }, { "epoch": 0.01, "learning_rate": 4.998462779576412e-05, "loss": 1.1782, "step": 11262 }, { "epoch": 0.01, "learning_rate": 4.9984625041541414e-05, "loss": 1.181, "step": 11263 }, { "epoch": 0.01, "learning_rate": 4.9984622287072066e-05, "loss": 1.0004, "step": 11264 }, { "epoch": 0.01, "learning_rate": 4.998461953235608e-05, "loss": 1.2491, "step": 11265 }, { "epoch": 0.01, "learning_rate": 4.998461677739347e-05, "loss": 1.4297, "step": 11266 }, { "epoch": 0.01, "learning_rate": 4.998461402218421e-05, "loss": 1.1271, "step": 11267 }, { "epoch": 0.01, "learning_rate": 4.9984611266728316e-05, "loss": 1.1888, "step": 11268 }, { "epoch": 0.01, "learning_rate": 4.998460851102578e-05, "loss": 1.2104, "step": 11269 }, { "epoch": 0.01, "learning_rate": 4.9984605755076614e-05, "loss": 1.0469, "step": 11270 }, { "epoch": 0.01, "learning_rate": 4.998460299888081e-05, "loss": 1.071, "step": 11271 }, { "epoch": 0.01, "learning_rate": 4.998460024243836e-05, "loss": 1.1764, "step": 11272 }, { "epoch": 0.01, "learning_rate": 4.9984597485749286e-05, "loss": 1.5474, "step": 11273 }, { "epoch": 0.01, "learning_rate": 4.998459472881356e-05, "loss": 1.2111, "step": 11274 }, { "epoch": 0.01, "learning_rate": 4.998459197163121e-05, "loss": 1.0373, "step": 11275 }, { "epoch": 0.01, "learning_rate": 4.998458921420222e-05, "loss": 0.9903, "step": 11276 }, { "epoch": 0.01, "learning_rate": 4.9984586456526585e-05, "loss": 0.9765, "step": 11277 }, { "epoch": 0.01, "learning_rate": 4.998458369860432e-05, "loss": 1.1196, "step": 11278 }, { "epoch": 0.01, "learning_rate": 4.9984580940435414e-05, "loss": 1.2614, "step": 11279 }, { "epoch": 0.01, "learning_rate": 4.998457818201987e-05, "loss": 0.6967, "step": 11280 }, { "epoch": 0.01, "learning_rate": 4.9984575423357686e-05, "loss": 0.5994, "step": 11281 }, { "epoch": 0.01, "learning_rate": 4.998457266444887e-05, "loss": 0.6136, "step": 11282 }, { "epoch": 0.01, "learning_rate": 4.9984569905293416e-05, "loss": 0.5445, "step": 11283 }, { "epoch": 0.01, "learning_rate": 4.998456714589133e-05, "loss": 0.7206, "step": 11284 }, { "epoch": 0.01, "learning_rate": 4.9984564386242596e-05, "loss": 0.2923, "step": 11285 }, { "epoch": 0.01, "learning_rate": 4.998456162634722e-05, "loss": 0.2527, "step": 11286 }, { "epoch": 0.01, "learning_rate": 4.998455886620523e-05, "loss": 0.1374, "step": 11287 }, { "epoch": 0.01, "learning_rate": 4.998455610581659e-05, "loss": 0.099, "step": 11288 }, { "epoch": 0.01, "learning_rate": 4.998455334518131e-05, "loss": 0.1067, "step": 11289 }, { "epoch": 0.01, "learning_rate": 4.998455058429939e-05, "loss": 0.0849, "step": 11290 }, { "epoch": 0.01, "learning_rate": 4.998454782317085e-05, "loss": 0.0381, "step": 11291 }, { "epoch": 0.01, "learning_rate": 4.998454506179566e-05, "loss": 0.0622, "step": 11292 }, { "epoch": 0.01, "learning_rate": 4.998454230017383e-05, "loss": 0.0453, "step": 11293 }, { "epoch": 0.01, "learning_rate": 4.998453953830538e-05, "loss": 0.2759, "step": 11294 }, { "epoch": 0.01, "learning_rate": 4.998453677619027e-05, "loss": 0.3704, "step": 11295 }, { "epoch": 0.01, "learning_rate": 4.998453401382855e-05, "loss": 0.3534, "step": 11296 }, { "epoch": 0.01, "learning_rate": 4.998453125122017e-05, "loss": 0.3021, "step": 11297 }, { "epoch": 0.01, "learning_rate": 4.998452848836516e-05, "loss": 0.2176, "step": 11298 }, { "epoch": 0.01, "learning_rate": 4.9984525725263515e-05, "loss": 0.8484, "step": 11299 }, { "epoch": 0.01, "learning_rate": 4.998452296191523e-05, "loss": 0.9705, "step": 11300 }, { "epoch": 0.01, "learning_rate": 4.9984520198320314e-05, "loss": 2.57, "step": 11301 }, { "epoch": 0.01, "learning_rate": 4.9984517434478756e-05, "loss": 4.6057, "step": 11302 }, { "epoch": 0.01, "learning_rate": 4.998451467039056e-05, "loss": 4.4689, "step": 11303 }, { "epoch": 0.01, "learning_rate": 4.9984511906055733e-05, "loss": 4.3654, "step": 11304 }, { "epoch": 0.01, "learning_rate": 4.9984509141474276e-05, "loss": 4.1739, "step": 11305 }, { "epoch": 0.01, "learning_rate": 4.998450637664617e-05, "loss": 3.8854, "step": 11306 }, { "epoch": 0.01, "learning_rate": 4.998450361157143e-05, "loss": 3.6601, "step": 11307 }, { "epoch": 0.01, "learning_rate": 4.998450084625006e-05, "loss": 3.5853, "step": 11308 }, { "epoch": 0.01, "learning_rate": 4.9984498080682047e-05, "loss": 3.4867, "step": 11309 }, { "epoch": 0.01, "learning_rate": 4.99844953148674e-05, "loss": 3.2333, "step": 11310 }, { "epoch": 0.01, "learning_rate": 4.998449254880612e-05, "loss": 3.3127, "step": 11311 }, { "epoch": 0.01, "learning_rate": 4.998448978249819e-05, "loss": 3.4235, "step": 11312 }, { "epoch": 0.01, "learning_rate": 4.998448701594364e-05, "loss": 3.0145, "step": 11313 }, { "epoch": 0.01, "learning_rate": 4.998448424914244e-05, "loss": 3.0155, "step": 11314 }, { "epoch": 0.01, "learning_rate": 4.998448148209461e-05, "loss": 2.968, "step": 11315 }, { "epoch": 0.01, "learning_rate": 4.998447871480015e-05, "loss": 2.445, "step": 11316 }, { "epoch": 0.01, "learning_rate": 4.998447594725905e-05, "loss": 0.4463, "step": 11317 }, { "epoch": 0.01, "learning_rate": 4.998447317947131e-05, "loss": 1.2936, "step": 11318 }, { "epoch": 0.01, "learning_rate": 4.998447041143694e-05, "loss": 1.6815, "step": 11319 }, { "epoch": 0.01, "learning_rate": 4.9984467643155924e-05, "loss": 1.3543, "step": 11320 }, { "epoch": 0.01, "learning_rate": 4.998446487462828e-05, "loss": 0.8619, "step": 11321 }, { "epoch": 0.01, "learning_rate": 4.998446210585399e-05, "loss": 1.2308, "step": 11322 }, { "epoch": 0.01, "learning_rate": 4.998445933683308e-05, "loss": 1.9087, "step": 11323 }, { "epoch": 0.01, "learning_rate": 4.998445656756552e-05, "loss": 1.0822, "step": 11324 }, { "epoch": 0.01, "learning_rate": 4.998445379805133e-05, "loss": 0.9731, "step": 11325 }, { "epoch": 0.01, "learning_rate": 4.99844510282905e-05, "loss": 0.9626, "step": 11326 }, { "epoch": 0.01, "learning_rate": 4.998444825828304e-05, "loss": 1.0638, "step": 11327 }, { "epoch": 0.01, "learning_rate": 4.998444548802894e-05, "loss": 1.3116, "step": 11328 }, { "epoch": 0.01, "learning_rate": 4.99844427175282e-05, "loss": 0.529, "step": 11329 }, { "epoch": 0.01, "learning_rate": 4.998443994678084e-05, "loss": 1.1222, "step": 11330 }, { "epoch": 0.01, "learning_rate": 4.9984437175786824e-05, "loss": 1.2004, "step": 11331 }, { "epoch": 0.01, "learning_rate": 4.998443440454619e-05, "loss": 1.0878, "step": 11332 }, { "epoch": 0.01, "learning_rate": 4.9984431633058906e-05, "loss": 1.2296, "step": 11333 }, { "epoch": 0.01, "learning_rate": 4.9984428861324994e-05, "loss": 1.2536, "step": 11334 }, { "epoch": 0.01, "learning_rate": 4.9984426089344446e-05, "loss": 1.0364, "step": 11335 }, { "epoch": 0.01, "learning_rate": 4.9984423317117255e-05, "loss": 1.6685, "step": 11336 }, { "epoch": 0.01, "learning_rate": 4.998442054464344e-05, "loss": 1.0711, "step": 11337 }, { "epoch": 0.01, "learning_rate": 4.998441777192298e-05, "loss": 0.8518, "step": 11338 }, { "epoch": 0.01, "learning_rate": 4.998441499895589e-05, "loss": 0.7944, "step": 11339 }, { "epoch": 0.01, "learning_rate": 4.998441222574216e-05, "loss": 1.0822, "step": 11340 }, { "epoch": 0.01, "learning_rate": 4.99844094522818e-05, "loss": 0.9353, "step": 11341 }, { "epoch": 0.01, "learning_rate": 4.9984406678574796e-05, "loss": 1.3678, "step": 11342 }, { "epoch": 0.01, "learning_rate": 4.998440390462116e-05, "loss": 1.8005, "step": 11343 }, { "epoch": 0.01, "learning_rate": 4.9984401130420886e-05, "loss": 1.3246, "step": 11344 }, { "epoch": 0.01, "learning_rate": 4.998439835597398e-05, "loss": 1.0563, "step": 11345 }, { "epoch": 0.01, "learning_rate": 4.998439558128045e-05, "loss": 1.4994, "step": 11346 }, { "epoch": 0.01, "learning_rate": 4.998439280634027e-05, "loss": 3.3592, "step": 11347 }, { "epoch": 0.01, "learning_rate": 4.998439003115346e-05, "loss": 5.7064, "step": 11348 }, { "epoch": 0.01, "learning_rate": 4.998438725572001e-05, "loss": 5.6699, "step": 11349 }, { "epoch": 0.01, "learning_rate": 4.998438448003992e-05, "loss": 5.6864, "step": 11350 }, { "epoch": 0.01, "learning_rate": 4.998438170411321e-05, "loss": 5.5852, "step": 11351 }, { "epoch": 0.01, "learning_rate": 4.9984378927939856e-05, "loss": 5.5625, "step": 11352 }, { "epoch": 0.01, "learning_rate": 4.9984376151519866e-05, "loss": 5.5607, "step": 11353 }, { "epoch": 0.01, "learning_rate": 4.9984373374853246e-05, "loss": 5.4607, "step": 11354 }, { "epoch": 0.01, "learning_rate": 4.9984370597939985e-05, "loss": 5.2944, "step": 11355 }, { "epoch": 0.01, "learning_rate": 4.998436782078009e-05, "loss": 5.3528, "step": 11356 }, { "epoch": 0.01, "learning_rate": 4.998436504337356e-05, "loss": 1.5605, "step": 11357 }, { "epoch": 0.01, "learning_rate": 4.99843622657204e-05, "loss": 0.9664, "step": 11358 }, { "epoch": 0.01, "learning_rate": 4.9984359487820596e-05, "loss": 0.8769, "step": 11359 }, { "epoch": 0.01, "learning_rate": 4.998435670967416e-05, "loss": 0.7703, "step": 11360 }, { "epoch": 0.01, "learning_rate": 4.9984353931281094e-05, "loss": 0.6933, "step": 11361 }, { "epoch": 0.01, "learning_rate": 4.998435115264138e-05, "loss": 0.6304, "step": 11362 }, { "epoch": 0.01, "learning_rate": 4.998434837375505e-05, "loss": 0.599, "step": 11363 }, { "epoch": 0.01, "learning_rate": 4.9984345594622074e-05, "loss": 0.5209, "step": 11364 }, { "epoch": 0.01, "learning_rate": 4.998434281524246e-05, "loss": 0.5048, "step": 11365 }, { "epoch": 0.01, "learning_rate": 4.9984340035616215e-05, "loss": 0.45, "step": 11366 }, { "epoch": 0.01, "learning_rate": 4.998433725574334e-05, "loss": 0.3872, "step": 11367 }, { "epoch": 0.01, "learning_rate": 4.998433447562383e-05, "loss": 0.3557, "step": 11368 }, { "epoch": 0.01, "learning_rate": 4.998433169525768e-05, "loss": 0.327, "step": 11369 }, { "epoch": 0.01, "learning_rate": 4.99843289146449e-05, "loss": 0.2999, "step": 11370 }, { "epoch": 0.01, "learning_rate": 4.998432613378547e-05, "loss": 0.2661, "step": 11371 }, { "epoch": 0.01, "learning_rate": 4.998432335267942e-05, "loss": 0.2799, "step": 11372 }, { "epoch": 0.01, "learning_rate": 4.9984320571326735e-05, "loss": 0.2492, "step": 11373 }, { "epoch": 0.01, "learning_rate": 4.998431778972741e-05, "loss": 0.2166, "step": 11374 }, { "epoch": 0.01, "learning_rate": 4.9984315007881455e-05, "loss": 0.2357, "step": 11375 }, { "epoch": 0.01, "learning_rate": 4.998431222578886e-05, "loss": 0.217, "step": 11376 }, { "epoch": 0.01, "learning_rate": 4.998430944344963e-05, "loss": 0.2398, "step": 11377 }, { "epoch": 0.01, "learning_rate": 4.998430666086377e-05, "loss": 0.205, "step": 11378 }, { "epoch": 0.01, "learning_rate": 4.998430387803128e-05, "loss": 0.2642, "step": 11379 }, { "epoch": 0.01, "learning_rate": 4.998430109495215e-05, "loss": 0.2823, "step": 11380 }, { "epoch": 0.01, "learning_rate": 4.998429831162638e-05, "loss": 0.7373, "step": 11381 }, { "epoch": 0.01, "learning_rate": 4.998429552805398e-05, "loss": 1.2914, "step": 11382 }, { "epoch": 0.01, "learning_rate": 4.998429274423495e-05, "loss": 0.6993, "step": 11383 }, { "epoch": 0.01, "learning_rate": 4.9984289960169286e-05, "loss": 1.2087, "step": 11384 }, { "epoch": 0.01, "learning_rate": 4.998428717585698e-05, "loss": 1.1759, "step": 11385 }, { "epoch": 0.01, "learning_rate": 4.998428439129804e-05, "loss": 0.8273, "step": 11386 }, { "epoch": 0.01, "learning_rate": 4.9984281606492466e-05, "loss": 1.3036, "step": 11387 }, { "epoch": 0.01, "learning_rate": 4.998427882144027e-05, "loss": 1.8611, "step": 11388 }, { "epoch": 0.01, "learning_rate": 4.998427603614143e-05, "loss": 1.269, "step": 11389 }, { "epoch": 0.01, "learning_rate": 4.9984273250595946e-05, "loss": 1.1731, "step": 11390 }, { "epoch": 0.01, "learning_rate": 4.998427046480384e-05, "loss": 1.1321, "step": 11391 }, { "epoch": 0.01, "learning_rate": 4.99842676787651e-05, "loss": 1.0621, "step": 11392 }, { "epoch": 0.01, "learning_rate": 4.9984264892479724e-05, "loss": 1.1258, "step": 11393 }, { "epoch": 0.01, "learning_rate": 4.998426210594771e-05, "loss": 1.191, "step": 11394 }, { "epoch": 0.01, "learning_rate": 4.998425931916907e-05, "loss": 1.1869, "step": 11395 }, { "epoch": 0.01, "learning_rate": 4.998425653214379e-05, "loss": 1.1527, "step": 11396 }, { "epoch": 0.01, "learning_rate": 4.9984253744871877e-05, "loss": 0.9867, "step": 11397 }, { "epoch": 0.01, "learning_rate": 4.998425095735333e-05, "loss": 0.5327, "step": 11398 }, { "epoch": 0.01, "learning_rate": 4.998424816958815e-05, "loss": 0.9447, "step": 11399 }, { "epoch": 0.01, "learning_rate": 4.998424538157633e-05, "loss": 1.1777, "step": 11400 }, { "epoch": 0.01, "learning_rate": 4.9984242593317885e-05, "loss": 1.1002, "step": 11401 }, { "epoch": 0.01, "learning_rate": 4.99842398048128e-05, "loss": 1.1605, "step": 11402 }, { "epoch": 0.01, "learning_rate": 4.998423701606109e-05, "loss": 1.0985, "step": 11403 }, { "epoch": 0.01, "learning_rate": 4.9984234227062734e-05, "loss": 0.8734, "step": 11404 }, { "epoch": 0.01, "learning_rate": 4.998423143781775e-05, "loss": 1.2233, "step": 11405 }, { "epoch": 0.01, "learning_rate": 4.998422864832614e-05, "loss": 1.0246, "step": 11406 }, { "epoch": 0.01, "learning_rate": 4.998422585858788e-05, "loss": 1.3118, "step": 11407 }, { "epoch": 0.01, "learning_rate": 4.9984223068603e-05, "loss": 1.359, "step": 11408 }, { "epoch": 0.01, "learning_rate": 4.998422027837147e-05, "loss": 1.284, "step": 11409 }, { "epoch": 0.01, "learning_rate": 4.9984217487893324e-05, "loss": 1.2624, "step": 11410 }, { "epoch": 0.01, "learning_rate": 4.998421469716853e-05, "loss": 0.9914, "step": 11411 }, { "epoch": 0.01, "learning_rate": 4.998421190619711e-05, "loss": 0.9909, "step": 11412 }, { "epoch": 0.01, "learning_rate": 4.9984209114979064e-05, "loss": 1.0387, "step": 11413 }, { "epoch": 0.01, "learning_rate": 4.998420632351437e-05, "loss": 1.1785, "step": 11414 }, { "epoch": 0.01, "learning_rate": 4.998420353180305e-05, "loss": 0.9459, "step": 11415 }, { "epoch": 0.01, "learning_rate": 4.99842007398451e-05, "loss": 1.1308, "step": 11416 }, { "epoch": 0.01, "learning_rate": 4.998419794764051e-05, "loss": 1.2018, "step": 11417 }, { "epoch": 0.01, "learning_rate": 4.998419515518929e-05, "loss": 1.0973, "step": 11418 }, { "epoch": 0.01, "learning_rate": 4.9984192362491436e-05, "loss": 1.1917, "step": 11419 }, { "epoch": 0.01, "learning_rate": 4.9984189569546944e-05, "loss": 0.8545, "step": 11420 }, { "epoch": 0.01, "learning_rate": 4.9984186776355824e-05, "loss": 0.8211, "step": 11421 }, { "epoch": 0.01, "learning_rate": 4.998418398291807e-05, "loss": 1.113, "step": 11422 }, { "epoch": 0.01, "learning_rate": 4.998418118923368e-05, "loss": 1.032, "step": 11423 }, { "epoch": 0.01, "learning_rate": 4.998417839530266e-05, "loss": 1.2295, "step": 11424 }, { "epoch": 0.01, "learning_rate": 4.9984175601125005e-05, "loss": 1.0615, "step": 11425 }, { "epoch": 0.01, "learning_rate": 4.998417280670071e-05, "loss": 0.175, "step": 11426 }, { "epoch": 0.01, "learning_rate": 4.998417001202979e-05, "loss": 0.3851, "step": 11427 }, { "epoch": 0.01, "learning_rate": 4.9984167217112235e-05, "loss": 1.0646, "step": 11428 }, { "epoch": 0.01, "learning_rate": 4.998416442194805e-05, "loss": 1.359, "step": 11429 }, { "epoch": 0.01, "learning_rate": 4.998416162653723e-05, "loss": 1.206, "step": 11430 }, { "epoch": 0.01, "learning_rate": 4.998415883087977e-05, "loss": 1.2987, "step": 11431 }, { "epoch": 0.01, "learning_rate": 4.998415603497568e-05, "loss": 0.8914, "step": 11432 }, { "epoch": 0.01, "learning_rate": 4.9984153238824964e-05, "loss": 1.0199, "step": 11433 }, { "epoch": 0.01, "learning_rate": 4.9984150442427614e-05, "loss": 0.9027, "step": 11434 }, { "epoch": 0.01, "learning_rate": 4.998414764578362e-05, "loss": 0.9879, "step": 11435 }, { "epoch": 0.01, "learning_rate": 4.9984144848893006e-05, "loss": 1.1858, "step": 11436 }, { "epoch": 0.01, "learning_rate": 4.9984142051755756e-05, "loss": 1.6896, "step": 11437 }, { "epoch": 0.01, "learning_rate": 4.998413925437186e-05, "loss": 1.8751, "step": 11438 }, { "epoch": 0.01, "learning_rate": 4.998413645674135e-05, "loss": 1.1724, "step": 11439 }, { "epoch": 0.01, "learning_rate": 4.99841336588642e-05, "loss": 1.2096, "step": 11440 }, { "epoch": 0.01, "learning_rate": 4.998413086074042e-05, "loss": 0.9712, "step": 11441 }, { "epoch": 0.01, "learning_rate": 4.9984128062369995e-05, "loss": 1.1346, "step": 11442 }, { "epoch": 0.01, "learning_rate": 4.9984125263752944e-05, "loss": 1.4867, "step": 11443 }, { "epoch": 0.01, "learning_rate": 4.9984122464889264e-05, "loss": 1.1887, "step": 11444 }, { "epoch": 0.01, "learning_rate": 4.998411966577895e-05, "loss": 1.2453, "step": 11445 }, { "epoch": 0.01, "learning_rate": 4.9984116866422004e-05, "loss": 1.0821, "step": 11446 }, { "epoch": 0.01, "learning_rate": 4.9984114066818424e-05, "loss": 0.858, "step": 11447 }, { "epoch": 0.01, "learning_rate": 4.9984111266968215e-05, "loss": 1.3127, "step": 11448 }, { "epoch": 0.01, "learning_rate": 4.9984108466871364e-05, "loss": 1.443, "step": 11449 }, { "epoch": 0.01, "learning_rate": 4.998410566652788e-05, "loss": 1.2599, "step": 11450 }, { "epoch": 0.01, "learning_rate": 4.9984102865937774e-05, "loss": 0.9952, "step": 11451 }, { "epoch": 0.01, "learning_rate": 4.9984100065101036e-05, "loss": 1.2394, "step": 11452 }, { "epoch": 0.01, "learning_rate": 4.998409726401766e-05, "loss": 1.1049, "step": 11453 }, { "epoch": 0.01, "learning_rate": 4.9984094462687646e-05, "loss": 0.6838, "step": 11454 }, { "epoch": 0.01, "learning_rate": 4.998409166111101e-05, "loss": 1.3454, "step": 11455 }, { "epoch": 0.01, "learning_rate": 4.9984088859287734e-05, "loss": 1.0764, "step": 11456 }, { "epoch": 0.01, "learning_rate": 4.9984086057217824e-05, "loss": 1.18, "step": 11457 }, { "epoch": 0.01, "learning_rate": 4.998408325490129e-05, "loss": 0.9963, "step": 11458 }, { "epoch": 0.01, "learning_rate": 4.9984080452338125e-05, "loss": 1.1228, "step": 11459 }, { "epoch": 0.01, "learning_rate": 4.9984077649528315e-05, "loss": 0.8998, "step": 11460 }, { "epoch": 0.01, "learning_rate": 4.998407484647188e-05, "loss": 0.9379, "step": 11461 }, { "epoch": 0.01, "learning_rate": 4.9984072043168815e-05, "loss": 1.5251, "step": 11462 }, { "epoch": 0.01, "learning_rate": 4.998406923961911e-05, "loss": 1.5672, "step": 11463 }, { "epoch": 0.01, "learning_rate": 4.998406643582279e-05, "loss": 1.1538, "step": 11464 }, { "epoch": 0.01, "learning_rate": 4.998406363177982e-05, "loss": 1.1419, "step": 11465 }, { "epoch": 0.01, "learning_rate": 4.998406082749023e-05, "loss": 1.306, "step": 11466 }, { "epoch": 0.01, "learning_rate": 4.9984058022953996e-05, "loss": 1.0548, "step": 11467 }, { "epoch": 0.01, "learning_rate": 4.9984055218171135e-05, "loss": 1.177, "step": 11468 }, { "epoch": 0.01, "learning_rate": 4.9984052413141645e-05, "loss": 1.223, "step": 11469 }, { "epoch": 0.01, "learning_rate": 4.998404960786552e-05, "loss": 1.4315, "step": 11470 }, { "epoch": 0.01, "learning_rate": 4.9984046802342764e-05, "loss": 1.4485, "step": 11471 }, { "epoch": 0.01, "learning_rate": 4.9984043996573374e-05, "loss": 0.9677, "step": 11472 }, { "epoch": 0.01, "learning_rate": 4.9984041190557355e-05, "loss": 1.0573, "step": 11473 }, { "epoch": 0.01, "learning_rate": 4.9984038384294706e-05, "loss": 1.4296, "step": 11474 }, { "epoch": 0.01, "learning_rate": 4.9984035577785416e-05, "loss": 3.2775, "step": 11475 }, { "epoch": 0.01, "learning_rate": 4.99840327710295e-05, "loss": 1.2259, "step": 11476 }, { "epoch": 0.01, "learning_rate": 4.9984029964026954e-05, "loss": 1.2345, "step": 11477 }, { "epoch": 0.01, "learning_rate": 4.998402715677778e-05, "loss": 1.1811, "step": 11478 }, { "epoch": 0.01, "learning_rate": 4.9984024349281964e-05, "loss": 1.2459, "step": 11479 }, { "epoch": 0.01, "learning_rate": 4.998402154153952e-05, "loss": 1.0402, "step": 11480 }, { "epoch": 0.01, "learning_rate": 4.9984018733550445e-05, "loss": 1.2996, "step": 11481 }, { "epoch": 0.01, "learning_rate": 4.9984015925314745e-05, "loss": 1.4455, "step": 11482 }, { "epoch": 0.01, "learning_rate": 4.99840131168324e-05, "loss": 1.0468, "step": 11483 }, { "epoch": 0.01, "learning_rate": 4.998401030810343e-05, "loss": 0.9074, "step": 11484 }, { "epoch": 0.01, "learning_rate": 4.9984007499127825e-05, "loss": 1.6606, "step": 11485 }, { "epoch": 0.01, "learning_rate": 4.99840046899056e-05, "loss": 1.3244, "step": 11486 }, { "epoch": 0.01, "learning_rate": 4.998400188043673e-05, "loss": 0.9966, "step": 11487 }, { "epoch": 0.01, "learning_rate": 4.998399907072123e-05, "loss": 0.7921, "step": 11488 }, { "epoch": 0.01, "learning_rate": 4.998399626075911e-05, "loss": 0.9289, "step": 11489 }, { "epoch": 0.01, "learning_rate": 4.9983993450550345e-05, "loss": 1.4705, "step": 11490 }, { "epoch": 0.01, "learning_rate": 4.998399064009496e-05, "loss": 1.2707, "step": 11491 }, { "epoch": 0.01, "learning_rate": 4.9983987829392936e-05, "loss": 1.1782, "step": 11492 }, { "epoch": 0.01, "learning_rate": 4.998398501844428e-05, "loss": 1.0509, "step": 11493 }, { "epoch": 0.01, "learning_rate": 4.9983982207249e-05, "loss": 1.2076, "step": 11494 }, { "epoch": 0.01, "learning_rate": 4.998397939580708e-05, "loss": 1.2198, "step": 11495 }, { "epoch": 0.01, "learning_rate": 4.998397658411854e-05, "loss": 1.2317, "step": 11496 }, { "epoch": 0.01, "learning_rate": 4.998397377218336e-05, "loss": 1.0691, "step": 11497 }, { "epoch": 0.01, "learning_rate": 4.998397096000155e-05, "loss": 1.4243, "step": 11498 }, { "epoch": 0.01, "learning_rate": 4.99839681475731e-05, "loss": 1.2581, "step": 11499 }, { "epoch": 0.01, "learning_rate": 4.998396533489803e-05, "loss": 1.1266, "step": 11500 }, { "epoch": 0.01, "eval_loss": 1.0503673553466797, "eval_runtime": 83.5269, "eval_samples_per_second": 16.581, "eval_steps_per_second": 4.154, "step": 11500 }, { "epoch": 0.01, "learning_rate": 4.9983962521976326e-05, "loss": 0.9645, "step": 11501 }, { "epoch": 0.01, "learning_rate": 4.9983959708807994e-05, "loss": 1.2468, "step": 11502 }, { "epoch": 0.01, "learning_rate": 4.998395689539303e-05, "loss": 1.0432, "step": 11503 }, { "epoch": 0.01, "learning_rate": 4.998395408173143e-05, "loss": 1.2302, "step": 11504 }, { "epoch": 0.01, "learning_rate": 4.9983951267823206e-05, "loss": 1.081, "step": 11505 }, { "epoch": 0.01, "learning_rate": 4.9983948453668345e-05, "loss": 0.9665, "step": 11506 }, { "epoch": 0.01, "learning_rate": 4.998394563926685e-05, "loss": 0.8637, "step": 11507 }, { "epoch": 0.01, "learning_rate": 4.998394282461874e-05, "loss": 1.2088, "step": 11508 }, { "epoch": 0.01, "learning_rate": 4.9983940009723976e-05, "loss": 1.144, "step": 11509 }, { "epoch": 0.01, "learning_rate": 4.99839371945826e-05, "loss": 1.9551, "step": 11510 }, { "epoch": 0.01, "learning_rate": 4.998393437919459e-05, "loss": 1.3644, "step": 11511 }, { "epoch": 0.01, "learning_rate": 4.998393156355994e-05, "loss": 1.1624, "step": 11512 }, { "epoch": 0.01, "learning_rate": 4.9983928747678663e-05, "loss": 1.0378, "step": 11513 }, { "epoch": 0.01, "learning_rate": 4.998392593155076e-05, "loss": 1.1334, "step": 11514 }, { "epoch": 0.01, "learning_rate": 4.998392311517622e-05, "loss": 1.2368, "step": 11515 }, { "epoch": 0.01, "learning_rate": 4.9983920298555054e-05, "loss": 1.1113, "step": 11516 }, { "epoch": 0.01, "learning_rate": 4.998391748168725e-05, "loss": 1.0115, "step": 11517 }, { "epoch": 0.01, "learning_rate": 4.998391466457282e-05, "loss": 1.3807, "step": 11518 }, { "epoch": 0.01, "learning_rate": 4.9983911847211764e-05, "loss": 1.6326, "step": 11519 }, { "epoch": 0.01, "learning_rate": 4.998390902960407e-05, "loss": 1.2276, "step": 11520 }, { "epoch": 0.01, "learning_rate": 4.998390621174975e-05, "loss": 1.0294, "step": 11521 }, { "epoch": 0.01, "learning_rate": 4.99839033936488e-05, "loss": 0.6542, "step": 11522 }, { "epoch": 0.01, "learning_rate": 4.9983900575301214e-05, "loss": 0.7583, "step": 11523 }, { "epoch": 0.01, "learning_rate": 4.9983897756707e-05, "loss": 0.9786, "step": 11524 }, { "epoch": 0.01, "learning_rate": 4.998389493786616e-05, "loss": 1.024, "step": 11525 }, { "epoch": 0.01, "learning_rate": 4.9983892118778684e-05, "loss": 1.0794, "step": 11526 }, { "epoch": 0.01, "learning_rate": 4.998388929944458e-05, "loss": 1.2192, "step": 11527 }, { "epoch": 0.01, "learning_rate": 4.9983886479863846e-05, "loss": 1.3275, "step": 11528 }, { "epoch": 0.01, "learning_rate": 4.998388366003648e-05, "loss": 0.9568, "step": 11529 }, { "epoch": 0.01, "learning_rate": 4.9983880839962485e-05, "loss": 1.1889, "step": 11530 }, { "epoch": 0.01, "learning_rate": 4.998387801964186e-05, "loss": 0.9586, "step": 11531 }, { "epoch": 0.01, "learning_rate": 4.99838751990746e-05, "loss": 1.2669, "step": 11532 }, { "epoch": 0.01, "learning_rate": 4.9983872378260714e-05, "loss": 1.2115, "step": 11533 }, { "epoch": 0.01, "learning_rate": 4.99838695572002e-05, "loss": 1.479, "step": 11534 }, { "epoch": 0.01, "learning_rate": 4.998386673589305e-05, "loss": 1.3306, "step": 11535 }, { "epoch": 0.01, "learning_rate": 4.9983863914339276e-05, "loss": 1.3134, "step": 11536 }, { "epoch": 0.01, "learning_rate": 4.9983861092538865e-05, "loss": 0.8815, "step": 11537 }, { "epoch": 0.01, "learning_rate": 4.9983858270491826e-05, "loss": 1.1442, "step": 11538 }, { "epoch": 0.01, "learning_rate": 4.998385544819816e-05, "loss": 1.404, "step": 11539 }, { "epoch": 0.01, "learning_rate": 4.998385262565786e-05, "loss": 0.7192, "step": 11540 }, { "epoch": 0.01, "learning_rate": 4.998384980287093e-05, "loss": 1.7457, "step": 11541 }, { "epoch": 0.01, "learning_rate": 4.998384697983738e-05, "loss": 1.2463, "step": 11542 }, { "epoch": 0.01, "learning_rate": 4.998384415655719e-05, "loss": 1.0283, "step": 11543 }, { "epoch": 0.01, "learning_rate": 4.998384133303037e-05, "loss": 1.0004, "step": 11544 }, { "epoch": 0.01, "learning_rate": 4.998383850925692e-05, "loss": 0.9757, "step": 11545 }, { "epoch": 0.01, "learning_rate": 4.9983835685236844e-05, "loss": 1.1777, "step": 11546 }, { "epoch": 0.01, "learning_rate": 4.998383286097014e-05, "loss": 0.9577, "step": 11547 }, { "epoch": 0.01, "learning_rate": 4.99838300364568e-05, "loss": 1.2794, "step": 11548 }, { "epoch": 0.01, "learning_rate": 4.998382721169683e-05, "loss": 1.1624, "step": 11549 }, { "epoch": 0.01, "learning_rate": 4.998382438669024e-05, "loss": 1.4319, "step": 11550 }, { "epoch": 0.01, "learning_rate": 4.9983821561437003e-05, "loss": 1.134, "step": 11551 }, { "epoch": 0.01, "learning_rate": 4.9983818735937146e-05, "loss": 1.0575, "step": 11552 }, { "epoch": 0.01, "learning_rate": 4.998381591019066e-05, "loss": 0.9679, "step": 11553 }, { "epoch": 0.01, "learning_rate": 4.998381308419755e-05, "loss": 1.13, "step": 11554 }, { "epoch": 0.01, "learning_rate": 4.99838102579578e-05, "loss": 1.1842, "step": 11555 }, { "epoch": 0.01, "learning_rate": 4.998380743147143e-05, "loss": 1.129, "step": 11556 }, { "epoch": 0.01, "learning_rate": 4.998380460473842e-05, "loss": 0.9742, "step": 11557 }, { "epoch": 0.01, "learning_rate": 4.998380177775879e-05, "loss": 0.9647, "step": 11558 }, { "epoch": 0.01, "learning_rate": 4.998379895053252e-05, "loss": 1.1006, "step": 11559 }, { "epoch": 0.01, "learning_rate": 4.9983796123059626e-05, "loss": 1.0966, "step": 11560 }, { "epoch": 0.01, "learning_rate": 4.99837932953401e-05, "loss": 1.3214, "step": 11561 }, { "epoch": 0.01, "learning_rate": 4.998379046737395e-05, "loss": 0.9388, "step": 11562 }, { "epoch": 0.01, "learning_rate": 4.998378763916117e-05, "loss": 1.0182, "step": 11563 }, { "epoch": 0.01, "learning_rate": 4.9983784810701756e-05, "loss": 1.1197, "step": 11564 }, { "epoch": 0.01, "learning_rate": 4.9983781981995717e-05, "loss": 0.9419, "step": 11565 }, { "epoch": 0.01, "learning_rate": 4.998377915304304e-05, "loss": 0.6806, "step": 11566 }, { "epoch": 0.01, "learning_rate": 4.9983776323843744e-05, "loss": 0.8233, "step": 11567 }, { "epoch": 0.01, "learning_rate": 4.998377349439781e-05, "loss": 0.8634, "step": 11568 }, { "epoch": 0.01, "learning_rate": 4.9983770664705255e-05, "loss": 0.5917, "step": 11569 }, { "epoch": 0.01, "learning_rate": 4.998376783476607e-05, "loss": 1.2584, "step": 11570 }, { "epoch": 0.01, "learning_rate": 4.998376500458025e-05, "loss": 1.2925, "step": 11571 }, { "epoch": 0.01, "learning_rate": 4.998376217414781e-05, "loss": 1.3929, "step": 11572 }, { "epoch": 0.01, "learning_rate": 4.9983759343468725e-05, "loss": 1.0965, "step": 11573 }, { "epoch": 0.01, "learning_rate": 4.9983756512543025e-05, "loss": 1.5753, "step": 11574 }, { "epoch": 0.01, "learning_rate": 4.9983753681370696e-05, "loss": 1.0878, "step": 11575 }, { "epoch": 0.01, "learning_rate": 4.9983750849951725e-05, "loss": 0.8005, "step": 11576 }, { "epoch": 0.01, "learning_rate": 4.998374801828614e-05, "loss": 0.1209, "step": 11577 }, { "epoch": 0.01, "learning_rate": 4.9983745186373923e-05, "loss": 0.0624, "step": 11578 }, { "epoch": 0.01, "learning_rate": 4.9983742354215066e-05, "loss": 0.0626, "step": 11579 }, { "epoch": 0.01, "learning_rate": 4.998373952180959e-05, "loss": 0.1138, "step": 11580 }, { "epoch": 0.01, "learning_rate": 4.998373668915748e-05, "loss": 0.0397, "step": 11581 }, { "epoch": 0.01, "learning_rate": 4.998373385625874e-05, "loss": 0.0496, "step": 11582 }, { "epoch": 0.01, "learning_rate": 4.998373102311338e-05, "loss": 0.0782, "step": 11583 }, { "epoch": 0.01, "learning_rate": 4.9983728189721384e-05, "loss": 0.4105, "step": 11584 }, { "epoch": 0.01, "learning_rate": 4.998372535608276e-05, "loss": 1.2428, "step": 11585 }, { "epoch": 0.01, "learning_rate": 4.998372252219751e-05, "loss": 0.8774, "step": 11586 }, { "epoch": 0.01, "learning_rate": 4.9983719688065625e-05, "loss": 1.1977, "step": 11587 }, { "epoch": 0.01, "learning_rate": 4.998371685368712e-05, "loss": 1.2278, "step": 11588 }, { "epoch": 0.01, "learning_rate": 4.998371401906198e-05, "loss": 0.7271, "step": 11589 }, { "epoch": 0.01, "learning_rate": 4.998371118419021e-05, "loss": 0.562, "step": 11590 }, { "epoch": 0.01, "learning_rate": 4.9983708349071814e-05, "loss": 1.1148, "step": 11591 }, { "epoch": 0.01, "learning_rate": 4.9983705513706794e-05, "loss": 1.038, "step": 11592 }, { "epoch": 0.01, "learning_rate": 4.998370267809514e-05, "loss": 1.3651, "step": 11593 }, { "epoch": 0.01, "learning_rate": 4.998369984223686e-05, "loss": 1.2701, "step": 11594 }, { "epoch": 0.01, "learning_rate": 4.998369700613195e-05, "loss": 0.9801, "step": 11595 }, { "epoch": 0.01, "learning_rate": 4.9983694169780406e-05, "loss": 1.1135, "step": 11596 }, { "epoch": 0.01, "learning_rate": 4.998369133318224e-05, "loss": 1.1544, "step": 11597 }, { "epoch": 0.01, "learning_rate": 4.998368849633744e-05, "loss": 1.1129, "step": 11598 }, { "epoch": 0.01, "learning_rate": 4.998368565924603e-05, "loss": 1.0495, "step": 11599 }, { "epoch": 0.01, "learning_rate": 4.9983682821907976e-05, "loss": 0.9521, "step": 11600 }, { "epoch": 0.01, "learning_rate": 4.998367998432329e-05, "loss": 1.2354, "step": 11601 }, { "epoch": 0.01, "learning_rate": 4.998367714649198e-05, "loss": 0.9108, "step": 11602 }, { "epoch": 0.01, "learning_rate": 4.998367430841404e-05, "loss": 1.4155, "step": 11603 }, { "epoch": 0.01, "learning_rate": 4.998367147008948e-05, "loss": 1.2151, "step": 11604 }, { "epoch": 0.01, "learning_rate": 4.998366863151828e-05, "loss": 1.066, "step": 11605 }, { "epoch": 0.01, "learning_rate": 4.9983665792700464e-05, "loss": 0.9034, "step": 11606 }, { "epoch": 0.01, "learning_rate": 4.998366295363601e-05, "loss": 1.1208, "step": 11607 }, { "epoch": 0.01, "learning_rate": 4.998366011432493e-05, "loss": 1.1024, "step": 11608 }, { "epoch": 0.01, "learning_rate": 4.9983657274767225e-05, "loss": 1.1969, "step": 11609 }, { "epoch": 0.01, "learning_rate": 4.998365443496289e-05, "loss": 1.2588, "step": 11610 }, { "epoch": 0.01, "learning_rate": 4.998365159491193e-05, "loss": 1.196, "step": 11611 }, { "epoch": 0.01, "learning_rate": 4.998364875461433e-05, "loss": 0.9279, "step": 11612 }, { "epoch": 0.01, "learning_rate": 4.998364591407011e-05, "loss": 1.0814, "step": 11613 }, { "epoch": 0.01, "learning_rate": 4.998364307327927e-05, "loss": 1.2077, "step": 11614 }, { "epoch": 0.01, "learning_rate": 4.9983640232241794e-05, "loss": 0.9932, "step": 11615 }, { "epoch": 0.01, "learning_rate": 4.998363739095769e-05, "loss": 1.1417, "step": 11616 }, { "epoch": 0.01, "learning_rate": 4.998363454942696e-05, "loss": 1.229, "step": 11617 }, { "epoch": 0.01, "learning_rate": 4.9983631707649595e-05, "loss": 1.2227, "step": 11618 }, { "epoch": 0.01, "learning_rate": 4.998362886562561e-05, "loss": 1.1244, "step": 11619 }, { "epoch": 0.01, "learning_rate": 4.9983626023354994e-05, "loss": 1.1602, "step": 11620 }, { "epoch": 0.01, "learning_rate": 4.9983623180837756e-05, "loss": 1.2657, "step": 11621 }, { "epoch": 0.01, "learning_rate": 4.9983620338073883e-05, "loss": 1.2006, "step": 11622 }, { "epoch": 0.01, "learning_rate": 4.998361749506338e-05, "loss": 1.1533, "step": 11623 }, { "epoch": 0.01, "learning_rate": 4.998361465180626e-05, "loss": 1.0951, "step": 11624 }, { "epoch": 0.01, "learning_rate": 4.9983611808302505e-05, "loss": 1.0147, "step": 11625 }, { "epoch": 0.01, "learning_rate": 4.998360896455212e-05, "loss": 1.3346, "step": 11626 }, { "epoch": 0.01, "learning_rate": 4.998360612055511e-05, "loss": 1.5097, "step": 11627 }, { "epoch": 0.01, "learning_rate": 4.9983603276311473e-05, "loss": 0.9217, "step": 11628 }, { "epoch": 0.01, "learning_rate": 4.998360043182121e-05, "loss": 1.0167, "step": 11629 }, { "epoch": 0.01, "learning_rate": 4.998359758708432e-05, "loss": 1.0593, "step": 11630 }, { "epoch": 0.01, "learning_rate": 4.99835947421008e-05, "loss": 0.9394, "step": 11631 }, { "epoch": 0.01, "learning_rate": 4.9983591896870654e-05, "loss": 0.9581, "step": 11632 }, { "epoch": 0.01, "learning_rate": 4.9983589051393884e-05, "loss": 1.3864, "step": 11633 }, { "epoch": 0.01, "learning_rate": 4.998358620567048e-05, "loss": 0.5484, "step": 11634 }, { "epoch": 0.01, "learning_rate": 4.998358335970045e-05, "loss": 0.3537, "step": 11635 }, { "epoch": 0.01, "learning_rate": 4.99835805134838e-05, "loss": 1.1215, "step": 11636 }, { "epoch": 0.01, "learning_rate": 4.9983577667020507e-05, "loss": 1.1501, "step": 11637 }, { "epoch": 0.01, "learning_rate": 4.99835748203106e-05, "loss": 1.1637, "step": 11638 }, { "epoch": 0.01, "learning_rate": 4.998357197335406e-05, "loss": 0.8523, "step": 11639 }, { "epoch": 0.01, "learning_rate": 4.998356912615089e-05, "loss": 1.2117, "step": 11640 }, { "epoch": 0.01, "learning_rate": 4.99835662787011e-05, "loss": 1.1864, "step": 11641 }, { "epoch": 0.01, "learning_rate": 4.998356343100468e-05, "loss": 0.9856, "step": 11642 }, { "epoch": 0.01, "learning_rate": 4.9983560583061625e-05, "loss": 1.0273, "step": 11643 }, { "epoch": 0.01, "learning_rate": 4.998355773487196e-05, "loss": 1.1164, "step": 11644 }, { "epoch": 0.01, "learning_rate": 4.9983554886435654e-05, "loss": 1.209, "step": 11645 }, { "epoch": 0.01, "learning_rate": 4.998355203775272e-05, "loss": 1.2607, "step": 11646 }, { "epoch": 0.01, "learning_rate": 4.998354918882317e-05, "loss": 0.9828, "step": 11647 }, { "epoch": 0.01, "learning_rate": 4.9983546339646984e-05, "loss": 1.1073, "step": 11648 }, { "epoch": 0.01, "learning_rate": 4.998354349022417e-05, "loss": 0.8334, "step": 11649 }, { "epoch": 0.01, "learning_rate": 4.998354064055474e-05, "loss": 1.2648, "step": 11650 }, { "epoch": 0.01, "learning_rate": 4.998353779063867e-05, "loss": 1.2619, "step": 11651 }, { "epoch": 0.01, "learning_rate": 4.998353494047598e-05, "loss": 1.405, "step": 11652 }, { "epoch": 0.01, "learning_rate": 4.998353209006666e-05, "loss": 1.4111, "step": 11653 }, { "epoch": 0.01, "learning_rate": 4.998352923941072e-05, "loss": 0.9003, "step": 11654 }, { "epoch": 0.01, "learning_rate": 4.998352638850814e-05, "loss": 1.1074, "step": 11655 }, { "epoch": 0.01, "learning_rate": 4.9983523537358945e-05, "loss": 0.9703, "step": 11656 }, { "epoch": 0.01, "learning_rate": 4.9983520685963115e-05, "loss": 1.0727, "step": 11657 }, { "epoch": 0.01, "learning_rate": 4.998351783432066e-05, "loss": 1.1307, "step": 11658 }, { "epoch": 0.01, "learning_rate": 4.998351498243159e-05, "loss": 0.7871, "step": 11659 }, { "epoch": 0.01, "learning_rate": 4.998351213029588e-05, "loss": 0.7997, "step": 11660 }, { "epoch": 0.01, "learning_rate": 4.998350927791355e-05, "loss": 0.7349, "step": 11661 }, { "epoch": 0.01, "learning_rate": 4.998350642528459e-05, "loss": 0.934, "step": 11662 }, { "epoch": 0.01, "learning_rate": 4.9983503572409004e-05, "loss": 0.9828, "step": 11663 }, { "epoch": 0.01, "learning_rate": 4.998350071928679e-05, "loss": 1.241, "step": 11664 }, { "epoch": 0.01, "learning_rate": 4.998349786591795e-05, "loss": 1.3398, "step": 11665 }, { "epoch": 0.01, "learning_rate": 4.998349501230248e-05, "loss": 1.0326, "step": 11666 }, { "epoch": 0.01, "learning_rate": 4.998349215844039e-05, "loss": 1.2806, "step": 11667 }, { "epoch": 0.01, "learning_rate": 4.998348930433167e-05, "loss": 1.0945, "step": 11668 }, { "epoch": 0.01, "learning_rate": 4.998348644997633e-05, "loss": 1.1345, "step": 11669 }, { "epoch": 0.01, "learning_rate": 4.9983483595374356e-05, "loss": 1.2521, "step": 11670 }, { "epoch": 0.01, "learning_rate": 4.9983480740525756e-05, "loss": 1.3518, "step": 11671 }, { "epoch": 0.01, "learning_rate": 4.998347788543053e-05, "loss": 1.1941, "step": 11672 }, { "epoch": 0.01, "learning_rate": 4.9983475030088675e-05, "loss": 1.2435, "step": 11673 }, { "epoch": 0.01, "learning_rate": 4.99834721745002e-05, "loss": 1.1946, "step": 11674 }, { "epoch": 0.01, "learning_rate": 4.99834693186651e-05, "loss": 0.9514, "step": 11675 }, { "epoch": 0.01, "learning_rate": 4.9983466462583375e-05, "loss": 0.9711, "step": 11676 }, { "epoch": 0.01, "learning_rate": 4.998346360625501e-05, "loss": 1.1204, "step": 11677 }, { "epoch": 0.01, "learning_rate": 4.998346074968003e-05, "loss": 0.6449, "step": 11678 }, { "epoch": 0.01, "learning_rate": 4.998345789285842e-05, "loss": 0.932, "step": 11679 }, { "epoch": 0.01, "learning_rate": 4.998345503579018e-05, "loss": 1.0239, "step": 11680 }, { "epoch": 0.01, "learning_rate": 4.9983452178475326e-05, "loss": 1.2577, "step": 11681 }, { "epoch": 0.01, "learning_rate": 4.998344932091384e-05, "loss": 1.2697, "step": 11682 }, { "epoch": 0.01, "learning_rate": 4.998344646310573e-05, "loss": 1.0062, "step": 11683 }, { "epoch": 0.01, "learning_rate": 4.998344360505099e-05, "loss": 0.96, "step": 11684 }, { "epoch": 0.01, "learning_rate": 4.998344074674962e-05, "loss": 1.4697, "step": 11685 }, { "epoch": 0.01, "learning_rate": 4.998343788820163e-05, "loss": 1.0183, "step": 11686 }, { "epoch": 0.01, "learning_rate": 4.998343502940701e-05, "loss": 0.9672, "step": 11687 }, { "epoch": 0.01, "learning_rate": 4.998343217036577e-05, "loss": 0.8766, "step": 11688 }, { "epoch": 0.01, "learning_rate": 4.99834293110779e-05, "loss": 1.0352, "step": 11689 }, { "epoch": 0.01, "learning_rate": 4.99834264515434e-05, "loss": 1.0626, "step": 11690 }, { "epoch": 0.01, "learning_rate": 4.9983423591762285e-05, "loss": 0.9342, "step": 11691 }, { "epoch": 0.01, "learning_rate": 4.998342073173454e-05, "loss": 0.9113, "step": 11692 }, { "epoch": 0.01, "learning_rate": 4.9983417871460164e-05, "loss": 1.1392, "step": 11693 }, { "epoch": 0.01, "learning_rate": 4.998341501093917e-05, "loss": 1.147, "step": 11694 }, { "epoch": 0.01, "learning_rate": 4.998341215017154e-05, "loss": 1.2468, "step": 11695 }, { "epoch": 0.01, "learning_rate": 4.998340928915729e-05, "loss": 1.2733, "step": 11696 }, { "epoch": 0.01, "learning_rate": 4.9983406427896416e-05, "loss": 0.771, "step": 11697 }, { "epoch": 0.01, "learning_rate": 4.998340356638892e-05, "loss": 1.2064, "step": 11698 }, { "epoch": 0.01, "learning_rate": 4.998340070463479e-05, "loss": 1.2075, "step": 11699 }, { "epoch": 0.01, "learning_rate": 4.998339784263404e-05, "loss": 1.1184, "step": 11700 }, { "epoch": 0.01, "learning_rate": 4.9983394980386653e-05, "loss": 1.0077, "step": 11701 }, { "epoch": 0.01, "learning_rate": 4.998339211789266e-05, "loss": 1.3163, "step": 11702 }, { "epoch": 0.01, "learning_rate": 4.998338925515202e-05, "loss": 1.1615, "step": 11703 }, { "epoch": 0.01, "learning_rate": 4.998338639216477e-05, "loss": 1.0435, "step": 11704 }, { "epoch": 0.01, "learning_rate": 4.9983383528930896e-05, "loss": 1.4037, "step": 11705 }, { "epoch": 0.01, "learning_rate": 4.9983380665450386e-05, "loss": 1.2341, "step": 11706 }, { "epoch": 0.01, "learning_rate": 4.9983377801723254e-05, "loss": 0.9345, "step": 11707 }, { "epoch": 0.01, "learning_rate": 4.99833749377495e-05, "loss": 0.874, "step": 11708 }, { "epoch": 0.01, "learning_rate": 4.9983372073529124e-05, "loss": 1.1688, "step": 11709 }, { "epoch": 0.01, "learning_rate": 4.998336920906211e-05, "loss": 1.1019, "step": 11710 }, { "epoch": 0.01, "learning_rate": 4.998336634434848e-05, "loss": 1.3786, "step": 11711 }, { "epoch": 0.01, "learning_rate": 4.998336347938822e-05, "loss": 1.439, "step": 11712 }, { "epoch": 0.01, "learning_rate": 4.998336061418134e-05, "loss": 1.339, "step": 11713 }, { "epoch": 0.01, "learning_rate": 4.998335774872783e-05, "loss": 1.1371, "step": 11714 }, { "epoch": 0.01, "learning_rate": 4.99833548830277e-05, "loss": 0.8706, "step": 11715 }, { "epoch": 0.01, "learning_rate": 4.998335201708094e-05, "loss": 0.7482, "step": 11716 }, { "epoch": 0.01, "learning_rate": 4.998334915088756e-05, "loss": 0.9907, "step": 11717 }, { "epoch": 0.01, "learning_rate": 4.9983346284447554e-05, "loss": 0.9305, "step": 11718 }, { "epoch": 0.01, "learning_rate": 4.998334341776092e-05, "loss": 1.0183, "step": 11719 }, { "epoch": 0.01, "learning_rate": 4.9983340550827656e-05, "loss": 1.0315, "step": 11720 }, { "epoch": 0.01, "learning_rate": 4.9983337683647774e-05, "loss": 0.9853, "step": 11721 }, { "epoch": 0.01, "learning_rate": 4.998333481622127e-05, "loss": 0.8974, "step": 11722 }, { "epoch": 0.01, "learning_rate": 4.998333194854813e-05, "loss": 0.9707, "step": 11723 }, { "epoch": 0.01, "learning_rate": 4.9983329080628374e-05, "loss": 1.1358, "step": 11724 }, { "epoch": 0.01, "learning_rate": 4.998332621246199e-05, "loss": 1.0645, "step": 11725 }, { "epoch": 0.01, "learning_rate": 4.9983323344048984e-05, "loss": 1.1729, "step": 11726 }, { "epoch": 0.01, "learning_rate": 4.998332047538935e-05, "loss": 0.9998, "step": 11727 }, { "epoch": 0.01, "learning_rate": 4.998331760648309e-05, "loss": 0.8516, "step": 11728 }, { "epoch": 0.01, "learning_rate": 4.998331473733021e-05, "loss": 0.4824, "step": 11729 }, { "epoch": 0.01, "learning_rate": 4.99833118679307e-05, "loss": 0.5791, "step": 11730 }, { "epoch": 0.01, "learning_rate": 4.998330899828457e-05, "loss": 0.4122, "step": 11731 }, { "epoch": 0.01, "learning_rate": 4.9983306128391814e-05, "loss": 0.4782, "step": 11732 }, { "epoch": 0.01, "learning_rate": 4.998330325825243e-05, "loss": 0.4057, "step": 11733 }, { "epoch": 0.01, "learning_rate": 4.998330038786643e-05, "loss": 0.9686, "step": 11734 }, { "epoch": 0.01, "learning_rate": 4.9983297517233796e-05, "loss": 1.0921, "step": 11735 }, { "epoch": 0.01, "learning_rate": 4.998329464635454e-05, "loss": 0.9476, "step": 11736 }, { "epoch": 0.01, "learning_rate": 4.998329177522866e-05, "loss": 1.1128, "step": 11737 }, { "epoch": 0.01, "learning_rate": 4.998328890385616e-05, "loss": 1.2486, "step": 11738 }, { "epoch": 0.01, "learning_rate": 4.9983286032237025e-05, "loss": 1.1701, "step": 11739 }, { "epoch": 0.01, "learning_rate": 4.9983283160371275e-05, "loss": 1.0324, "step": 11740 }, { "epoch": 0.01, "learning_rate": 4.99832802882589e-05, "loss": 1.2524, "step": 11741 }, { "epoch": 0.01, "learning_rate": 4.9983277415899896e-05, "loss": 0.9121, "step": 11742 }, { "epoch": 0.01, "learning_rate": 4.9983274543294266e-05, "loss": 0.7149, "step": 11743 }, { "epoch": 0.01, "learning_rate": 4.9983271670442014e-05, "loss": 0.7126, "step": 11744 }, { "epoch": 0.01, "learning_rate": 4.998326879734314e-05, "loss": 0.7163, "step": 11745 }, { "epoch": 0.01, "learning_rate": 4.998326592399764e-05, "loss": 0.6952, "step": 11746 }, { "epoch": 0.01, "learning_rate": 4.998326305040551e-05, "loss": 0.6841, "step": 11747 }, { "epoch": 0.01, "learning_rate": 4.998326017656677e-05, "loss": 0.6924, "step": 11748 }, { "epoch": 0.01, "learning_rate": 4.9983257302481404e-05, "loss": 0.7164, "step": 11749 }, { "epoch": 0.01, "learning_rate": 4.99832544281494e-05, "loss": 0.6867, "step": 11750 }, { "epoch": 0.01, "learning_rate": 4.998325155357078e-05, "loss": 0.6912, "step": 11751 }, { "epoch": 0.01, "learning_rate": 4.9983248678745536e-05, "loss": 0.6771, "step": 11752 }, { "epoch": 0.01, "learning_rate": 4.9983245803673665e-05, "loss": 0.6655, "step": 11753 }, { "epoch": 0.01, "learning_rate": 4.998324292835517e-05, "loss": 0.534, "step": 11754 }, { "epoch": 0.01, "learning_rate": 4.998324005279006e-05, "loss": 0.4779, "step": 11755 }, { "epoch": 0.01, "learning_rate": 4.998323717697831e-05, "loss": 0.6134, "step": 11756 }, { "epoch": 0.01, "learning_rate": 4.998323430091995e-05, "loss": 0.6505, "step": 11757 }, { "epoch": 0.01, "learning_rate": 4.9983231424614964e-05, "loss": 0.658, "step": 11758 }, { "epoch": 0.01, "learning_rate": 4.998322854806335e-05, "loss": 0.6802, "step": 11759 }, { "epoch": 0.01, "learning_rate": 4.998322567126511e-05, "loss": 0.6745, "step": 11760 }, { "epoch": 0.01, "learning_rate": 4.998322279422025e-05, "loss": 0.6616, "step": 11761 }, { "epoch": 0.01, "learning_rate": 4.998321991692877e-05, "loss": 0.6549, "step": 11762 }, { "epoch": 0.01, "learning_rate": 4.9983217039390663e-05, "loss": 0.6667, "step": 11763 }, { "epoch": 0.01, "learning_rate": 4.998321416160593e-05, "loss": 0.6433, "step": 11764 }, { "epoch": 0.01, "learning_rate": 4.998321128357457e-05, "loss": 0.6033, "step": 11765 }, { "epoch": 0.01, "learning_rate": 4.9983208405296586e-05, "loss": 0.6027, "step": 11766 }, { "epoch": 0.01, "learning_rate": 4.9983205526771985e-05, "loss": 0.5814, "step": 11767 }, { "epoch": 0.01, "learning_rate": 4.998320264800076e-05, "loss": 0.5885, "step": 11768 }, { "epoch": 0.01, "learning_rate": 4.998319976898292e-05, "loss": 0.5924, "step": 11769 }, { "epoch": 0.01, "learning_rate": 4.9983196889718436e-05, "loss": 0.5976, "step": 11770 }, { "epoch": 0.01, "learning_rate": 4.998319401020733e-05, "loss": 0.584, "step": 11771 }, { "epoch": 0.01, "learning_rate": 4.9983191130449614e-05, "loss": 0.567, "step": 11772 }, { "epoch": 0.01, "learning_rate": 4.998318825044527e-05, "loss": 0.5799, "step": 11773 }, { "epoch": 0.01, "learning_rate": 4.9983185370194305e-05, "loss": 0.6011, "step": 11774 }, { "epoch": 0.01, "learning_rate": 4.998318248969671e-05, "loss": 0.6608, "step": 11775 }, { "epoch": 0.01, "learning_rate": 4.99831796089525e-05, "loss": 0.6653, "step": 11776 }, { "epoch": 0.01, "learning_rate": 4.998317672796166e-05, "loss": 0.6688, "step": 11777 }, { "epoch": 0.01, "learning_rate": 4.99831738467242e-05, "loss": 0.6603, "step": 11778 }, { "epoch": 0.01, "learning_rate": 4.998317096524011e-05, "loss": 0.6453, "step": 11779 }, { "epoch": 0.01, "learning_rate": 4.9983168083509404e-05, "loss": 0.6551, "step": 11780 }, { "epoch": 0.01, "learning_rate": 4.9983165201532066e-05, "loss": 0.6575, "step": 11781 }, { "epoch": 0.01, "learning_rate": 4.998316231930811e-05, "loss": 0.6216, "step": 11782 }, { "epoch": 0.01, "learning_rate": 4.998315943683754e-05, "loss": 0.6309, "step": 11783 }, { "epoch": 0.01, "learning_rate": 4.9983156554120334e-05, "loss": 0.6641, "step": 11784 }, { "epoch": 0.01, "learning_rate": 4.998315367115651e-05, "loss": 0.9049, "step": 11785 }, { "epoch": 0.01, "learning_rate": 4.998315078794606e-05, "loss": 1.3519, "step": 11786 }, { "epoch": 0.01, "learning_rate": 4.998314790448899e-05, "loss": 1.1333, "step": 11787 }, { "epoch": 0.01, "learning_rate": 4.9983145020785296e-05, "loss": 0.67, "step": 11788 }, { "epoch": 0.01, "learning_rate": 4.998314213683497e-05, "loss": 1.1663, "step": 11789 }, { "epoch": 0.01, "learning_rate": 4.998313925263803e-05, "loss": 1.0568, "step": 11790 }, { "epoch": 0.01, "learning_rate": 4.998313636819447e-05, "loss": 1.3779, "step": 11791 }, { "epoch": 0.01, "learning_rate": 4.9983133483504285e-05, "loss": 1.1582, "step": 11792 }, { "epoch": 0.01, "learning_rate": 4.9983130598567475e-05, "loss": 0.9026, "step": 11793 }, { "epoch": 0.01, "learning_rate": 4.9983127713384036e-05, "loss": 1.1377, "step": 11794 }, { "epoch": 0.01, "learning_rate": 4.998312482795399e-05, "loss": 1.2038, "step": 11795 }, { "epoch": 0.01, "learning_rate": 4.9983121942277306e-05, "loss": 1.0308, "step": 11796 }, { "epoch": 0.01, "learning_rate": 4.998311905635401e-05, "loss": 0.9214, "step": 11797 }, { "epoch": 0.01, "learning_rate": 4.998311617018409e-05, "loss": 1.11, "step": 11798 }, { "epoch": 0.01, "learning_rate": 4.998311328376754e-05, "loss": 1.3039, "step": 11799 }, { "epoch": 0.01, "learning_rate": 4.998311039710437e-05, "loss": 1.0181, "step": 11800 }, { "epoch": 0.01, "learning_rate": 4.998310751019457e-05, "loss": 0.7254, "step": 11801 }, { "epoch": 0.01, "learning_rate": 4.998310462303816e-05, "loss": 0.9878, "step": 11802 }, { "epoch": 0.01, "learning_rate": 4.998310173563512e-05, "loss": 1.1497, "step": 11803 }, { "epoch": 0.01, "learning_rate": 4.998309884798546e-05, "loss": 1.106, "step": 11804 }, { "epoch": 0.01, "learning_rate": 4.998309596008918e-05, "loss": 0.331, "step": 11805 }, { "epoch": 0.01, "learning_rate": 4.9983093071946274e-05, "loss": 0.6411, "step": 11806 }, { "epoch": 0.01, "learning_rate": 4.998309018355674e-05, "loss": 1.171, "step": 11807 }, { "epoch": 0.01, "learning_rate": 4.998308729492059e-05, "loss": 1.2667, "step": 11808 }, { "epoch": 0.01, "learning_rate": 4.9983084406037815e-05, "loss": 0.9939, "step": 11809 }, { "epoch": 0.01, "learning_rate": 4.998308151690842e-05, "loss": 1.0861, "step": 11810 }, { "epoch": 0.01, "learning_rate": 4.99830786275324e-05, "loss": 1.1063, "step": 11811 }, { "epoch": 0.01, "learning_rate": 4.9983075737909765e-05, "loss": 1.1264, "step": 11812 }, { "epoch": 0.01, "learning_rate": 4.99830728480405e-05, "loss": 0.9674, "step": 11813 }, { "epoch": 0.01, "learning_rate": 4.998306995792462e-05, "loss": 1.1379, "step": 11814 }, { "epoch": 0.01, "learning_rate": 4.998306706756211e-05, "loss": 0.9703, "step": 11815 }, { "epoch": 0.01, "learning_rate": 4.9983064176952976e-05, "loss": 1.3968, "step": 11816 }, { "epoch": 0.01, "learning_rate": 4.998306128609722e-05, "loss": 1.0956, "step": 11817 }, { "epoch": 0.01, "learning_rate": 4.998305839499485e-05, "loss": 2.2113, "step": 11818 }, { "epoch": 0.01, "learning_rate": 4.9983055503645854e-05, "loss": 2.0404, "step": 11819 }, { "epoch": 0.01, "learning_rate": 4.9983052612050234e-05, "loss": 1.96, "step": 11820 }, { "epoch": 0.01, "learning_rate": 4.9983049720208e-05, "loss": 1.9934, "step": 11821 }, { "epoch": 0.01, "learning_rate": 4.9983046828119127e-05, "loss": 1.0033, "step": 11822 }, { "epoch": 0.01, "learning_rate": 4.9983043935783647e-05, "loss": 1.083, "step": 11823 }, { "epoch": 0.01, "learning_rate": 4.998304104320154e-05, "loss": 0.9911, "step": 11824 }, { "epoch": 0.01, "learning_rate": 4.998303815037281e-05, "loss": 1.1688, "step": 11825 }, { "epoch": 0.01, "learning_rate": 4.9983035257297454e-05, "loss": 1.3553, "step": 11826 }, { "epoch": 0.01, "learning_rate": 4.998303236397548e-05, "loss": 0.9616, "step": 11827 }, { "epoch": 0.01, "learning_rate": 4.998302947040688e-05, "loss": 1.381, "step": 11828 }, { "epoch": 0.01, "learning_rate": 4.998302657659167e-05, "loss": 1.2144, "step": 11829 }, { "epoch": 0.01, "learning_rate": 4.998302368252983e-05, "loss": 0.9461, "step": 11830 }, { "epoch": 0.01, "learning_rate": 4.998302078822137e-05, "loss": 0.5215, "step": 11831 }, { "epoch": 0.01, "learning_rate": 4.9983017893666284e-05, "loss": 0.9689, "step": 11832 }, { "epoch": 0.01, "learning_rate": 4.998301499886458e-05, "loss": 1.4347, "step": 11833 }, { "epoch": 0.01, "learning_rate": 4.998301210381625e-05, "loss": 0.9473, "step": 11834 }, { "epoch": 0.01, "learning_rate": 4.99830092085213e-05, "loss": 1.5753, "step": 11835 }, { "epoch": 0.01, "learning_rate": 4.998300631297973e-05, "loss": 1.5506, "step": 11836 }, { "epoch": 0.01, "learning_rate": 4.9983003417191544e-05, "loss": 1.1735, "step": 11837 }, { "epoch": 0.01, "learning_rate": 4.998300052115673e-05, "loss": 1.0264, "step": 11838 }, { "epoch": 0.01, "learning_rate": 4.998299762487529e-05, "loss": 1.2732, "step": 11839 }, { "epoch": 0.01, "learning_rate": 4.9982994728347235e-05, "loss": 1.0026, "step": 11840 }, { "epoch": 0.01, "learning_rate": 4.998299183157256e-05, "loss": 0.9684, "step": 11841 }, { "epoch": 0.01, "learning_rate": 4.998298893455126e-05, "loss": 1.1025, "step": 11842 }, { "epoch": 0.01, "learning_rate": 4.9982986037283335e-05, "loss": 0.8175, "step": 11843 }, { "epoch": 0.01, "learning_rate": 4.998298313976879e-05, "loss": 0.6195, "step": 11844 }, { "epoch": 0.01, "learning_rate": 4.9982980242007624e-05, "loss": 1.047, "step": 11845 }, { "epoch": 0.01, "learning_rate": 4.998297734399984e-05, "loss": 0.8917, "step": 11846 }, { "epoch": 0.01, "learning_rate": 4.998297444574543e-05, "loss": 1.0824, "step": 11847 }, { "epoch": 0.01, "learning_rate": 4.9982971547244406e-05, "loss": 1.0504, "step": 11848 }, { "epoch": 0.01, "learning_rate": 4.998296864849675e-05, "loss": 0.9859, "step": 11849 }, { "epoch": 0.01, "learning_rate": 4.998296574950249e-05, "loss": 1.2751, "step": 11850 }, { "epoch": 0.01, "learning_rate": 4.9982962850261595e-05, "loss": 1.1871, "step": 11851 }, { "epoch": 0.01, "learning_rate": 4.998295995077408e-05, "loss": 1.3634, "step": 11852 }, { "epoch": 0.01, "learning_rate": 4.9982957051039944e-05, "loss": 1.5256, "step": 11853 }, { "epoch": 0.01, "learning_rate": 4.9982954151059185e-05, "loss": 1.6309, "step": 11854 }, { "epoch": 0.01, "learning_rate": 4.998295125083181e-05, "loss": 1.72, "step": 11855 }, { "epoch": 0.01, "learning_rate": 4.998294835035781e-05, "loss": 2.9109, "step": 11856 }, { "epoch": 0.01, "learning_rate": 4.998294544963719e-05, "loss": 0.969, "step": 11857 }, { "epoch": 0.01, "learning_rate": 4.998294254866995e-05, "loss": 0.7834, "step": 11858 }, { "epoch": 0.01, "learning_rate": 4.9982939647456087e-05, "loss": 1.0156, "step": 11859 }, { "epoch": 0.01, "learning_rate": 4.99829367459956e-05, "loss": 1.1561, "step": 11860 }, { "epoch": 0.01, "learning_rate": 4.9982933844288495e-05, "loss": 1.1978, "step": 11861 }, { "epoch": 0.01, "learning_rate": 4.998293094233477e-05, "loss": 1.1596, "step": 11862 }, { "epoch": 0.01, "learning_rate": 4.998292804013442e-05, "loss": 1.118, "step": 11863 }, { "epoch": 0.01, "learning_rate": 4.9982925137687456e-05, "loss": 1.2925, "step": 11864 }, { "epoch": 0.01, "learning_rate": 4.998292223499387e-05, "loss": 1.0552, "step": 11865 }, { "epoch": 0.01, "learning_rate": 4.998291933205366e-05, "loss": 1.5542, "step": 11866 }, { "epoch": 0.01, "learning_rate": 4.998291642886683e-05, "loss": 1.1159, "step": 11867 }, { "epoch": 0.01, "learning_rate": 4.9982913525433376e-05, "loss": 0.9709, "step": 11868 }, { "epoch": 0.01, "learning_rate": 4.9982910621753306e-05, "loss": 0.993, "step": 11869 }, { "epoch": 0.01, "learning_rate": 4.9982907717826614e-05, "loss": 1.3587, "step": 11870 }, { "epoch": 0.01, "learning_rate": 4.99829048136533e-05, "loss": 1.2833, "step": 11871 }, { "epoch": 0.01, "learning_rate": 4.998290190923337e-05, "loss": 1.1247, "step": 11872 }, { "epoch": 0.01, "learning_rate": 4.998289900456681e-05, "loss": 0.836, "step": 11873 }, { "epoch": 0.01, "learning_rate": 4.9982896099653644e-05, "loss": 0.9815, "step": 11874 }, { "epoch": 0.01, "learning_rate": 4.998289319449384e-05, "loss": 0.985, "step": 11875 }, { "epoch": 0.01, "learning_rate": 4.9982890289087423e-05, "loss": 0.7515, "step": 11876 }, { "epoch": 0.01, "learning_rate": 4.998288738343439e-05, "loss": 1.1292, "step": 11877 }, { "epoch": 0.01, "learning_rate": 4.998288447753473e-05, "loss": 1.2825, "step": 11878 }, { "epoch": 0.01, "learning_rate": 4.998288157138845e-05, "loss": 0.9876, "step": 11879 }, { "epoch": 0.01, "learning_rate": 4.998287866499556e-05, "loss": 1.0972, "step": 11880 }, { "epoch": 0.01, "learning_rate": 4.998287575835604e-05, "loss": 1.5533, "step": 11881 }, { "epoch": 0.01, "learning_rate": 4.99828728514699e-05, "loss": 1.0838, "step": 11882 }, { "epoch": 0.01, "learning_rate": 4.998286994433714e-05, "loss": 0.9633, "step": 11883 }, { "epoch": 0.01, "learning_rate": 4.998286703695776e-05, "loss": 1.0257, "step": 11884 }, { "epoch": 0.01, "learning_rate": 4.998286412933176e-05, "loss": 1.0479, "step": 11885 }, { "epoch": 0.01, "learning_rate": 4.9982861221459135e-05, "loss": 1.1619, "step": 11886 }, { "epoch": 0.01, "learning_rate": 4.99828583133399e-05, "loss": 1.0674, "step": 11887 }, { "epoch": 0.01, "learning_rate": 4.998285540497403e-05, "loss": 1.2695, "step": 11888 }, { "epoch": 0.01, "learning_rate": 4.998285249636155e-05, "loss": 1.2531, "step": 11889 }, { "epoch": 0.01, "learning_rate": 4.998284958750245e-05, "loss": 0.8401, "step": 11890 }, { "epoch": 0.01, "learning_rate": 4.998284667839673e-05, "loss": 0.7986, "step": 11891 }, { "epoch": 0.01, "learning_rate": 4.9982843769044384e-05, "loss": 1.2023, "step": 11892 }, { "epoch": 0.01, "learning_rate": 4.9982840859445424e-05, "loss": 0.6585, "step": 11893 }, { "epoch": 0.01, "learning_rate": 4.998283794959984e-05, "loss": 1.1371, "step": 11894 }, { "epoch": 0.01, "learning_rate": 4.998283503950764e-05, "loss": 1.1359, "step": 11895 }, { "epoch": 0.01, "learning_rate": 4.9982832129168824e-05, "loss": 0.848, "step": 11896 }, { "epoch": 0.01, "learning_rate": 4.998282921858337e-05, "loss": 1.4398, "step": 11897 }, { "epoch": 0.01, "learning_rate": 4.998282630775131e-05, "loss": 1.3705, "step": 11898 }, { "epoch": 0.01, "learning_rate": 4.998282339667263e-05, "loss": 1.0344, "step": 11899 }, { "epoch": 0.01, "learning_rate": 4.9982820485347324e-05, "loss": 1.21, "step": 11900 }, { "epoch": 0.01, "learning_rate": 4.998281757377541e-05, "loss": 1.0661, "step": 11901 }, { "epoch": 0.01, "learning_rate": 4.998281466195687e-05, "loss": 0.636, "step": 11902 }, { "epoch": 0.01, "learning_rate": 4.99828117498917e-05, "loss": 0.464, "step": 11903 }, { "epoch": 0.01, "learning_rate": 4.998280883757992e-05, "loss": 0.4931, "step": 11904 }, { "epoch": 0.01, "learning_rate": 4.9982805925021526e-05, "loss": 0.4886, "step": 11905 }, { "epoch": 0.01, "learning_rate": 4.99828030122165e-05, "loss": 0.4493, "step": 11906 }, { "epoch": 0.01, "learning_rate": 4.998280009916486e-05, "loss": 0.4238, "step": 11907 }, { "epoch": 0.01, "learning_rate": 4.9982797185866605e-05, "loss": 0.3304, "step": 11908 }, { "epoch": 0.01, "learning_rate": 4.998279427232172e-05, "loss": 0.2891, "step": 11909 }, { "epoch": 0.01, "learning_rate": 4.998279135853022e-05, "loss": 0.5425, "step": 11910 }, { "epoch": 0.01, "learning_rate": 4.99827884444921e-05, "loss": 1.1136, "step": 11911 }, { "epoch": 0.01, "learning_rate": 4.9982785530207366e-05, "loss": 1.098, "step": 11912 }, { "epoch": 0.01, "learning_rate": 4.9982782615676004e-05, "loss": 1.3658, "step": 11913 }, { "epoch": 0.01, "learning_rate": 4.9982779700898034e-05, "loss": 1.0967, "step": 11914 }, { "epoch": 0.01, "learning_rate": 4.998277678587343e-05, "loss": 0.8413, "step": 11915 }, { "epoch": 0.01, "learning_rate": 4.998277387060221e-05, "loss": 1.1046, "step": 11916 }, { "epoch": 0.01, "learning_rate": 4.9982770955084376e-05, "loss": 0.5204, "step": 11917 }, { "epoch": 0.01, "learning_rate": 4.9982768039319924e-05, "loss": 0.5353, "step": 11918 }, { "epoch": 0.01, "learning_rate": 4.998276512330885e-05, "loss": 0.7756, "step": 11919 }, { "epoch": 0.01, "learning_rate": 4.9982762207051154e-05, "loss": 1.1151, "step": 11920 }, { "epoch": 0.01, "learning_rate": 4.9982759290546836e-05, "loss": 1.0285, "step": 11921 }, { "epoch": 0.01, "learning_rate": 4.99827563737959e-05, "loss": 0.9638, "step": 11922 }, { "epoch": 0.01, "learning_rate": 4.998275345679835e-05, "loss": 1.0292, "step": 11923 }, { "epoch": 0.01, "learning_rate": 4.998275053955418e-05, "loss": 1.0397, "step": 11924 }, { "epoch": 0.01, "learning_rate": 4.998274762206339e-05, "loss": 0.2139, "step": 11925 }, { "epoch": 0.01, "learning_rate": 4.998274470432599e-05, "loss": 0.1712, "step": 11926 }, { "epoch": 0.01, "learning_rate": 4.9982741786341956e-05, "loss": 0.0851, "step": 11927 }, { "epoch": 0.01, "learning_rate": 4.9982738868111304e-05, "loss": 0.0965, "step": 11928 }, { "epoch": 0.01, "learning_rate": 4.998273594963404e-05, "loss": 0.0471, "step": 11929 }, { "epoch": 0.01, "learning_rate": 4.998273303091016e-05, "loss": 0.172, "step": 11930 }, { "epoch": 0.01, "learning_rate": 4.9982730111939646e-05, "loss": 0.3087, "step": 11931 }, { "epoch": 0.01, "learning_rate": 4.9982727192722525e-05, "loss": 1.0428, "step": 11932 }, { "epoch": 0.01, "learning_rate": 4.998272427325878e-05, "loss": 1.0664, "step": 11933 }, { "epoch": 0.01, "learning_rate": 4.9982721353548424e-05, "loss": 1.4479, "step": 11934 }, { "epoch": 0.01, "learning_rate": 4.9982718433591444e-05, "loss": 0.9686, "step": 11935 }, { "epoch": 0.01, "learning_rate": 4.998271551338784e-05, "loss": 0.7969, "step": 11936 }, { "epoch": 0.01, "learning_rate": 4.9982712592937616e-05, "loss": 0.9167, "step": 11937 }, { "epoch": 0.01, "learning_rate": 4.998270967224078e-05, "loss": 1.133, "step": 11938 }, { "epoch": 0.01, "learning_rate": 4.9982706751297334e-05, "loss": 1.3677, "step": 11939 }, { "epoch": 0.01, "learning_rate": 4.998270383010726e-05, "loss": 0.8491, "step": 11940 }, { "epoch": 0.01, "learning_rate": 4.9982700908670564e-05, "loss": 1.178, "step": 11941 }, { "epoch": 0.01, "learning_rate": 4.998269798698725e-05, "loss": 1.2003, "step": 11942 }, { "epoch": 0.01, "learning_rate": 4.998269506505732e-05, "loss": 1.2625, "step": 11943 }, { "epoch": 0.01, "learning_rate": 4.998269214288077e-05, "loss": 1.0666, "step": 11944 }, { "epoch": 0.01, "learning_rate": 4.99826892204576e-05, "loss": 0.9202, "step": 11945 }, { "epoch": 0.01, "learning_rate": 4.9982686297787816e-05, "loss": 1.2332, "step": 11946 }, { "epoch": 0.01, "learning_rate": 4.998268337487141e-05, "loss": 0.8747, "step": 11947 }, { "epoch": 0.01, "learning_rate": 4.998268045170838e-05, "loss": 0.788, "step": 11948 }, { "epoch": 0.01, "learning_rate": 4.998267752829874e-05, "loss": 0.9752, "step": 11949 }, { "epoch": 0.01, "learning_rate": 4.998267460464249e-05, "loss": 0.8442, "step": 11950 }, { "epoch": 0.01, "learning_rate": 4.99826716807396e-05, "loss": 1.1171, "step": 11951 }, { "epoch": 0.01, "learning_rate": 4.998266875659011e-05, "loss": 0.8666, "step": 11952 }, { "epoch": 0.01, "learning_rate": 4.998266583219399e-05, "loss": 1.3084, "step": 11953 }, { "epoch": 0.01, "learning_rate": 4.998266290755126e-05, "loss": 1.5812, "step": 11954 }, { "epoch": 0.01, "learning_rate": 4.998265998266191e-05, "loss": 1.4263, "step": 11955 }, { "epoch": 0.01, "learning_rate": 4.998265705752593e-05, "loss": 1.3035, "step": 11956 }, { "epoch": 0.01, "learning_rate": 4.9982654132143346e-05, "loss": 1.618, "step": 11957 }, { "epoch": 0.01, "learning_rate": 4.998265120651414e-05, "loss": 1.44, "step": 11958 }, { "epoch": 0.01, "learning_rate": 4.998264828063831e-05, "loss": 0.9215, "step": 11959 }, { "epoch": 0.01, "learning_rate": 4.998264535451587e-05, "loss": 0.9074, "step": 11960 }, { "epoch": 0.01, "learning_rate": 4.998264242814681e-05, "loss": 1.1559, "step": 11961 }, { "epoch": 0.01, "learning_rate": 4.998263950153113e-05, "loss": 0.8698, "step": 11962 }, { "epoch": 0.01, "learning_rate": 4.998263657466883e-05, "loss": 1.4071, "step": 11963 }, { "epoch": 0.01, "learning_rate": 4.9982633647559915e-05, "loss": 1.1193, "step": 11964 }, { "epoch": 0.01, "learning_rate": 4.998263072020438e-05, "loss": 0.8669, "step": 11965 }, { "epoch": 0.01, "learning_rate": 4.998262779260223e-05, "loss": 1.0289, "step": 11966 }, { "epoch": 0.01, "learning_rate": 4.9982624864753456e-05, "loss": 1.0742, "step": 11967 }, { "epoch": 0.01, "learning_rate": 4.998262193665807e-05, "loss": 0.9148, "step": 11968 }, { "epoch": 0.01, "learning_rate": 4.9982619008316065e-05, "loss": 1.1374, "step": 11969 }, { "epoch": 0.01, "learning_rate": 4.998261607972744e-05, "loss": 1.2256, "step": 11970 }, { "epoch": 0.01, "learning_rate": 4.99826131508922e-05, "loss": 0.9827, "step": 11971 }, { "epoch": 0.01, "learning_rate": 4.9982610221810335e-05, "loss": 0.8465, "step": 11972 }, { "epoch": 0.01, "learning_rate": 4.9982607292481856e-05, "loss": 0.9577, "step": 11973 }, { "epoch": 0.01, "learning_rate": 4.998260436290676e-05, "loss": 1.0681, "step": 11974 }, { "epoch": 0.01, "learning_rate": 4.998260143308505e-05, "loss": 1.0878, "step": 11975 }, { "epoch": 0.01, "learning_rate": 4.998259850301672e-05, "loss": 1.1793, "step": 11976 }, { "epoch": 0.01, "learning_rate": 4.9982595572701776e-05, "loss": 1.0593, "step": 11977 }, { "epoch": 0.01, "learning_rate": 4.998259264214021e-05, "loss": 1.1046, "step": 11978 }, { "epoch": 0.01, "learning_rate": 4.998258971133202e-05, "loss": 1.1013, "step": 11979 }, { "epoch": 0.01, "learning_rate": 4.998258678027722e-05, "loss": 1.0016, "step": 11980 }, { "epoch": 0.01, "learning_rate": 4.998258384897581e-05, "loss": 1.2534, "step": 11981 }, { "epoch": 0.01, "learning_rate": 4.9982580917427765e-05, "loss": 1.0986, "step": 11982 }, { "epoch": 0.01, "learning_rate": 4.998257798563312e-05, "loss": 1.0771, "step": 11983 }, { "epoch": 0.01, "learning_rate": 4.998257505359184e-05, "loss": 1.2647, "step": 11984 }, { "epoch": 0.01, "learning_rate": 4.998257212130395e-05, "loss": 0.9088, "step": 11985 }, { "epoch": 0.01, "learning_rate": 4.9982569188769444e-05, "loss": 1.4022, "step": 11986 }, { "epoch": 0.01, "learning_rate": 4.998256625598833e-05, "loss": 1.1231, "step": 11987 }, { "epoch": 0.01, "learning_rate": 4.998256332296059e-05, "loss": 1.1074, "step": 11988 }, { "epoch": 0.01, "learning_rate": 4.998256038968623e-05, "loss": 1.1353, "step": 11989 }, { "epoch": 0.01, "learning_rate": 4.9982557456165244e-05, "loss": 1.07, "step": 11990 }, { "epoch": 0.01, "learning_rate": 4.9982554522397656e-05, "loss": 0.887, "step": 11991 }, { "epoch": 0.01, "learning_rate": 4.9982551588383445e-05, "loss": 0.6104, "step": 11992 }, { "epoch": 0.01, "learning_rate": 4.998254865412262e-05, "loss": 1.2055, "step": 11993 }, { "epoch": 0.01, "learning_rate": 4.998254571961518e-05, "loss": 1.4669, "step": 11994 }, { "epoch": 0.01, "learning_rate": 4.9982542784861115e-05, "loss": 1.1528, "step": 11995 }, { "epoch": 0.01, "learning_rate": 4.998253984986043e-05, "loss": 1.1111, "step": 11996 }, { "epoch": 0.01, "learning_rate": 4.9982536914613135e-05, "loss": 1.1834, "step": 11997 }, { "epoch": 0.01, "learning_rate": 4.9982533979119226e-05, "loss": 0.9809, "step": 11998 }, { "epoch": 0.01, "learning_rate": 4.9982531043378695e-05, "loss": 1.3561, "step": 11999 }, { "epoch": 0.01, "learning_rate": 4.998252810739155e-05, "loss": 1.0595, "step": 12000 }, { "epoch": 0.01, "eval_loss": 1.0549529790878296, "eval_runtime": 95.3857, "eval_samples_per_second": 14.52, "eval_steps_per_second": 3.638, "step": 12000 }, { "epoch": 0.01, "learning_rate": 4.9982525171157786e-05, "loss": 1.407, "step": 12001 }, { "epoch": 0.01, "learning_rate": 4.99825222346774e-05, "loss": 1.2803, "step": 12002 }, { "epoch": 0.01, "learning_rate": 4.99825192979504e-05, "loss": 1.2961, "step": 12003 }, { "epoch": 0.01, "learning_rate": 4.9982516360976794e-05, "loss": 1.1177, "step": 12004 }, { "epoch": 0.01, "learning_rate": 4.9982513423756564e-05, "loss": 1.2042, "step": 12005 }, { "epoch": 0.01, "learning_rate": 4.998251048628971e-05, "loss": 1.0742, "step": 12006 }, { "epoch": 0.01, "learning_rate": 4.9982507548576244e-05, "loss": 1.3132, "step": 12007 }, { "epoch": 0.01, "learning_rate": 4.998250461061617e-05, "loss": 1.3119, "step": 12008 }, { "epoch": 0.01, "learning_rate": 4.998250167240946e-05, "loss": 1.468, "step": 12009 }, { "epoch": 0.01, "learning_rate": 4.998249873395615e-05, "loss": 1.5029, "step": 12010 }, { "epoch": 0.01, "learning_rate": 4.998249579525621e-05, "loss": 0.4397, "step": 12011 }, { "epoch": 0.01, "learning_rate": 4.998249285630967e-05, "loss": 0.9234, "step": 12012 }, { "epoch": 0.01, "learning_rate": 4.99824899171165e-05, "loss": 1.1719, "step": 12013 }, { "epoch": 0.01, "learning_rate": 4.998248697767672e-05, "loss": 1.0474, "step": 12014 }, { "epoch": 0.01, "learning_rate": 4.9982484037990316e-05, "loss": 1.0646, "step": 12015 }, { "epoch": 0.01, "learning_rate": 4.9982481098057304e-05, "loss": 1.647, "step": 12016 }, { "epoch": 0.01, "learning_rate": 4.9982478157877676e-05, "loss": 1.5274, "step": 12017 }, { "epoch": 0.01, "learning_rate": 4.9982475217451426e-05, "loss": 1.4571, "step": 12018 }, { "epoch": 0.01, "learning_rate": 4.998247227677856e-05, "loss": 1.371, "step": 12019 }, { "epoch": 0.01, "learning_rate": 4.998246933585908e-05, "loss": 0.9796, "step": 12020 }, { "epoch": 0.01, "learning_rate": 4.9982466394692984e-05, "loss": 0.8978, "step": 12021 }, { "epoch": 0.01, "learning_rate": 4.998246345328027e-05, "loss": 1.0147, "step": 12022 }, { "epoch": 0.01, "learning_rate": 4.998246051162093e-05, "loss": 1.2353, "step": 12023 }, { "epoch": 0.01, "learning_rate": 4.998245756971499e-05, "loss": 1.0887, "step": 12024 }, { "epoch": 0.01, "learning_rate": 4.998245462756243e-05, "loss": 0.9905, "step": 12025 }, { "epoch": 0.01, "learning_rate": 4.998245168516325e-05, "loss": 1.1723, "step": 12026 }, { "epoch": 0.01, "learning_rate": 4.998244874251745e-05, "loss": 1.1858, "step": 12027 }, { "epoch": 0.01, "learning_rate": 4.9982445799625036e-05, "loss": 1.1936, "step": 12028 }, { "epoch": 0.01, "learning_rate": 4.998244285648601e-05, "loss": 1.1744, "step": 12029 }, { "epoch": 0.01, "learning_rate": 4.998243991310036e-05, "loss": 1.1702, "step": 12030 }, { "epoch": 0.01, "learning_rate": 4.998243696946811e-05, "loss": 1.2338, "step": 12031 }, { "epoch": 0.01, "learning_rate": 4.998243402558923e-05, "loss": 1.0368, "step": 12032 }, { "epoch": 0.01, "learning_rate": 4.9982431081463735e-05, "loss": 1.0198, "step": 12033 }, { "epoch": 0.01, "learning_rate": 4.998242813709163e-05, "loss": 0.8942, "step": 12034 }, { "epoch": 0.01, "learning_rate": 4.99824251924729e-05, "loss": 1.2525, "step": 12035 }, { "epoch": 0.01, "learning_rate": 4.998242224760756e-05, "loss": 1.1418, "step": 12036 }, { "epoch": 0.01, "learning_rate": 4.998241930249561e-05, "loss": 1.1844, "step": 12037 }, { "epoch": 0.01, "learning_rate": 4.998241635713704e-05, "loss": 1.4223, "step": 12038 }, { "epoch": 0.01, "learning_rate": 4.9982413411531845e-05, "loss": 1.1155, "step": 12039 }, { "epoch": 0.01, "learning_rate": 4.998241046568004e-05, "loss": 1.1506, "step": 12040 }, { "epoch": 0.01, "learning_rate": 4.998240751958163e-05, "loss": 1.1888, "step": 12041 }, { "epoch": 0.01, "learning_rate": 4.998240457323658e-05, "loss": 1.2224, "step": 12042 }, { "epoch": 0.01, "learning_rate": 4.9982401626644936e-05, "loss": 1.2334, "step": 12043 }, { "epoch": 0.01, "learning_rate": 4.998239867980667e-05, "loss": 0.8968, "step": 12044 }, { "epoch": 0.01, "learning_rate": 4.998239573272179e-05, "loss": 1.0604, "step": 12045 }, { "epoch": 0.01, "learning_rate": 4.998239278539028e-05, "loss": 0.7093, "step": 12046 }, { "epoch": 0.01, "learning_rate": 4.9982389837812175e-05, "loss": 1.4507, "step": 12047 }, { "epoch": 0.01, "learning_rate": 4.9982386889987444e-05, "loss": 1.3586, "step": 12048 }, { "epoch": 0.01, "learning_rate": 4.99823839419161e-05, "loss": 0.9533, "step": 12049 }, { "epoch": 0.01, "learning_rate": 4.998238099359814e-05, "loss": 0.9154, "step": 12050 }, { "epoch": 0.01, "learning_rate": 4.998237804503356e-05, "loss": 0.9695, "step": 12051 }, { "epoch": 0.01, "learning_rate": 4.998237509622237e-05, "loss": 1.1887, "step": 12052 }, { "epoch": 0.01, "learning_rate": 4.998237214716456e-05, "loss": 1.1882, "step": 12053 }, { "epoch": 0.01, "learning_rate": 4.998236919786014e-05, "loss": 1.0273, "step": 12054 }, { "epoch": 0.01, "learning_rate": 4.9982366248309096e-05, "loss": 1.229, "step": 12055 }, { "epoch": 0.01, "learning_rate": 4.998236329851145e-05, "loss": 1.2101, "step": 12056 }, { "epoch": 0.01, "learning_rate": 4.998236034846718e-05, "loss": 0.9866, "step": 12057 }, { "epoch": 0.01, "learning_rate": 4.998235739817629e-05, "loss": 1.1917, "step": 12058 }, { "epoch": 0.01, "learning_rate": 4.998235444763879e-05, "loss": 0.853, "step": 12059 }, { "epoch": 0.01, "learning_rate": 4.998235149685468e-05, "loss": 1.0745, "step": 12060 }, { "epoch": 0.01, "learning_rate": 4.998234854582395e-05, "loss": 1.1735, "step": 12061 }, { "epoch": 0.01, "learning_rate": 4.99823455945466e-05, "loss": 1.3232, "step": 12062 }, { "epoch": 0.01, "learning_rate": 4.998234264302264e-05, "loss": 1.137, "step": 12063 }, { "epoch": 0.01, "learning_rate": 4.9982339691252066e-05, "loss": 1.0724, "step": 12064 }, { "epoch": 0.01, "learning_rate": 4.9982336739234875e-05, "loss": 1.1275, "step": 12065 }, { "epoch": 0.01, "learning_rate": 4.998233378697107e-05, "loss": 1.189, "step": 12066 }, { "epoch": 0.01, "learning_rate": 4.998233083446065e-05, "loss": 1.0839, "step": 12067 }, { "epoch": 0.01, "learning_rate": 4.998232788170361e-05, "loss": 1.1865, "step": 12068 }, { "epoch": 0.01, "learning_rate": 4.9982324928699964e-05, "loss": 1.2996, "step": 12069 }, { "epoch": 0.01, "learning_rate": 4.9982321975449696e-05, "loss": 1.0822, "step": 12070 }, { "epoch": 0.01, "learning_rate": 4.998231902195282e-05, "loss": 0.9162, "step": 12071 }, { "epoch": 0.01, "learning_rate": 4.9982316068209315e-05, "loss": 1.1151, "step": 12072 }, { "epoch": 0.01, "learning_rate": 4.99823131142192e-05, "loss": 0.9708, "step": 12073 }, { "epoch": 0.01, "learning_rate": 4.998231015998248e-05, "loss": 0.8525, "step": 12074 }, { "epoch": 0.01, "learning_rate": 4.998230720549914e-05, "loss": 1.1532, "step": 12075 }, { "epoch": 0.01, "learning_rate": 4.998230425076919e-05, "loss": 1.061, "step": 12076 }, { "epoch": 0.01, "learning_rate": 4.998230129579261e-05, "loss": 1.1644, "step": 12077 }, { "epoch": 0.01, "learning_rate": 4.998229834056943e-05, "loss": 1.3484, "step": 12078 }, { "epoch": 0.01, "learning_rate": 4.998229538509963e-05, "loss": 1.3026, "step": 12079 }, { "epoch": 0.01, "learning_rate": 4.9982292429383216e-05, "loss": 1.1669, "step": 12080 }, { "epoch": 0.01, "learning_rate": 4.9982289473420186e-05, "loss": 1.1338, "step": 12081 }, { "epoch": 0.01, "learning_rate": 4.998228651721054e-05, "loss": 0.9417, "step": 12082 }, { "epoch": 0.01, "learning_rate": 4.998228356075428e-05, "loss": 1.1192, "step": 12083 }, { "epoch": 0.01, "learning_rate": 4.998228060405141e-05, "loss": 1.1765, "step": 12084 }, { "epoch": 0.01, "learning_rate": 4.998227764710192e-05, "loss": 1.162, "step": 12085 }, { "epoch": 0.01, "learning_rate": 4.9982274689905814e-05, "loss": 1.1069, "step": 12086 }, { "epoch": 0.01, "learning_rate": 4.9982271732463106e-05, "loss": 1.2318, "step": 12087 }, { "epoch": 0.01, "learning_rate": 4.9982268774773776e-05, "loss": 1.0605, "step": 12088 }, { "epoch": 0.01, "learning_rate": 4.9982265816837824e-05, "loss": 0.5878, "step": 12089 }, { "epoch": 0.01, "learning_rate": 4.998226285865527e-05, "loss": 0.6539, "step": 12090 }, { "epoch": 0.01, "learning_rate": 4.998225990022609e-05, "loss": 0.9472, "step": 12091 }, { "epoch": 0.01, "learning_rate": 4.998225694155031e-05, "loss": 1.1295, "step": 12092 }, { "epoch": 0.01, "learning_rate": 4.99822539826279e-05, "loss": 1.0472, "step": 12093 }, { "epoch": 0.01, "learning_rate": 4.998225102345888e-05, "loss": 1.2961, "step": 12094 }, { "epoch": 0.01, "learning_rate": 4.9982248064043255e-05, "loss": 1.1211, "step": 12095 }, { "epoch": 0.01, "learning_rate": 4.998224510438101e-05, "loss": 0.8796, "step": 12096 }, { "epoch": 0.01, "learning_rate": 4.998224214447215e-05, "loss": 0.8588, "step": 12097 }, { "epoch": 0.01, "learning_rate": 4.998223918431668e-05, "loss": 0.6565, "step": 12098 }, { "epoch": 0.01, "learning_rate": 4.9982236223914585e-05, "loss": 1.061, "step": 12099 }, { "epoch": 0.01, "learning_rate": 4.9982233263265884e-05, "loss": 1.2461, "step": 12100 }, { "epoch": 0.01, "learning_rate": 4.998223030237057e-05, "loss": 1.4045, "step": 12101 }, { "epoch": 0.01, "learning_rate": 4.998222734122864e-05, "loss": 1.216, "step": 12102 }, { "epoch": 0.01, "learning_rate": 4.998222437984009e-05, "loss": 1.0652, "step": 12103 }, { "epoch": 0.01, "learning_rate": 4.9982221418204935e-05, "loss": 1.2943, "step": 12104 }, { "epoch": 0.01, "learning_rate": 4.998221845632316e-05, "loss": 1.3192, "step": 12105 }, { "epoch": 0.01, "learning_rate": 4.998221549419477e-05, "loss": 1.3339, "step": 12106 }, { "epoch": 0.01, "learning_rate": 4.998221253181978e-05, "loss": 1.5881, "step": 12107 }, { "epoch": 0.01, "learning_rate": 4.998220956919816e-05, "loss": 1.3818, "step": 12108 }, { "epoch": 0.01, "learning_rate": 4.9982206606329935e-05, "loss": 1.2337, "step": 12109 }, { "epoch": 0.01, "learning_rate": 4.9982203643215094e-05, "loss": 1.2094, "step": 12110 }, { "epoch": 0.01, "learning_rate": 4.998220067985364e-05, "loss": 1.3412, "step": 12111 }, { "epoch": 0.01, "learning_rate": 4.9982197716245574e-05, "loss": 1.1303, "step": 12112 }, { "epoch": 0.01, "learning_rate": 4.998219475239089e-05, "loss": 0.9769, "step": 12113 }, { "epoch": 0.01, "learning_rate": 4.998219178828959e-05, "loss": 1.0795, "step": 12114 }, { "epoch": 0.01, "learning_rate": 4.9982188823941675e-05, "loss": 1.0715, "step": 12115 }, { "epoch": 0.01, "learning_rate": 4.9982185859347156e-05, "loss": 1.1885, "step": 12116 }, { "epoch": 0.01, "learning_rate": 4.9982182894506015e-05, "loss": 0.9226, "step": 12117 }, { "epoch": 0.01, "learning_rate": 4.9982179929418265e-05, "loss": 0.9401, "step": 12118 }, { "epoch": 0.01, "learning_rate": 4.998217696408391e-05, "loss": 0.7499, "step": 12119 }, { "epoch": 0.01, "learning_rate": 4.9982173998502926e-05, "loss": 0.364, "step": 12120 }, { "epoch": 0.01, "learning_rate": 4.998217103267533e-05, "loss": 1.0029, "step": 12121 }, { "epoch": 0.01, "learning_rate": 4.9982168066601126e-05, "loss": 0.9788, "step": 12122 }, { "epoch": 0.01, "learning_rate": 4.998216510028031e-05, "loss": 0.8128, "step": 12123 }, { "epoch": 0.01, "learning_rate": 4.998216213371287e-05, "loss": 0.8819, "step": 12124 }, { "epoch": 0.01, "learning_rate": 4.998215916689883e-05, "loss": 1.0108, "step": 12125 }, { "epoch": 0.01, "learning_rate": 4.998215619983817e-05, "loss": 0.9376, "step": 12126 }, { "epoch": 0.01, "learning_rate": 4.9982153232530895e-05, "loss": 0.7809, "step": 12127 }, { "epoch": 0.01, "learning_rate": 4.998215026497701e-05, "loss": 0.7568, "step": 12128 }, { "epoch": 0.01, "learning_rate": 4.998214729717652e-05, "loss": 1.0232, "step": 12129 }, { "epoch": 0.01, "learning_rate": 4.998214432912941e-05, "loss": 1.3491, "step": 12130 }, { "epoch": 0.01, "learning_rate": 4.998214136083568e-05, "loss": 1.2105, "step": 12131 }, { "epoch": 0.01, "learning_rate": 4.998213839229534e-05, "loss": 1.145, "step": 12132 }, { "epoch": 0.01, "learning_rate": 4.998213542350839e-05, "loss": 1.1217, "step": 12133 }, { "epoch": 0.01, "learning_rate": 4.998213245447483e-05, "loss": 1.166, "step": 12134 }, { "epoch": 0.01, "learning_rate": 4.9982129485194644e-05, "loss": 1.2469, "step": 12135 }, { "epoch": 0.01, "learning_rate": 4.998212651566786e-05, "loss": 1.2689, "step": 12136 }, { "epoch": 0.01, "learning_rate": 4.998212354589445e-05, "loss": 1.0761, "step": 12137 }, { "epoch": 0.01, "learning_rate": 4.9982120575874436e-05, "loss": 0.984, "step": 12138 }, { "epoch": 0.01, "learning_rate": 4.9982117605607805e-05, "loss": 1.2007, "step": 12139 }, { "epoch": 0.01, "learning_rate": 4.9982114635094566e-05, "loss": 1.1728, "step": 12140 }, { "epoch": 0.01, "learning_rate": 4.998211166433471e-05, "loss": 1.186, "step": 12141 }, { "epoch": 0.01, "learning_rate": 4.998210869332824e-05, "loss": 1.2243, "step": 12142 }, { "epoch": 0.01, "learning_rate": 4.9982105722075154e-05, "loss": 1.4051, "step": 12143 }, { "epoch": 0.01, "learning_rate": 4.998210275057547e-05, "loss": 1.0489, "step": 12144 }, { "epoch": 0.01, "learning_rate": 4.9982099778829164e-05, "loss": 0.8759, "step": 12145 }, { "epoch": 0.01, "learning_rate": 4.998209680683624e-05, "loss": 1.2142, "step": 12146 }, { "epoch": 0.01, "learning_rate": 4.9982093834596706e-05, "loss": 1.3355, "step": 12147 }, { "epoch": 0.01, "learning_rate": 4.9982090862110564e-05, "loss": 0.9924, "step": 12148 }, { "epoch": 0.01, "learning_rate": 4.9982087889377806e-05, "loss": 0.9815, "step": 12149 }, { "epoch": 0.01, "learning_rate": 4.9982084916398434e-05, "loss": 0.6959, "step": 12150 }, { "epoch": 0.01, "learning_rate": 4.998208194317245e-05, "loss": 0.4738, "step": 12151 }, { "epoch": 0.01, "learning_rate": 4.9982078969699856e-05, "loss": 0.3914, "step": 12152 }, { "epoch": 0.01, "learning_rate": 4.998207599598065e-05, "loss": 0.6012, "step": 12153 }, { "epoch": 0.01, "learning_rate": 4.998207302201482e-05, "loss": 0.3795, "step": 12154 }, { "epoch": 0.01, "learning_rate": 4.9982070047802394e-05, "loss": 0.3312, "step": 12155 }, { "epoch": 0.01, "learning_rate": 4.998206707334335e-05, "loss": 0.4454, "step": 12156 }, { "epoch": 0.01, "learning_rate": 4.998206409863769e-05, "loss": 0.9173, "step": 12157 }, { "epoch": 0.01, "learning_rate": 4.998206112368542e-05, "loss": 1.0747, "step": 12158 }, { "epoch": 0.01, "learning_rate": 4.998205814848653e-05, "loss": 1.2573, "step": 12159 }, { "epoch": 0.01, "learning_rate": 4.998205517304104e-05, "loss": 1.3131, "step": 12160 }, { "epoch": 0.01, "learning_rate": 4.998205219734893e-05, "loss": 0.9893, "step": 12161 }, { "epoch": 0.01, "learning_rate": 4.998204922141021e-05, "loss": 0.5699, "step": 12162 }, { "epoch": 0.01, "learning_rate": 4.998204624522488e-05, "loss": 0.7396, "step": 12163 }, { "epoch": 0.01, "learning_rate": 4.998204326879294e-05, "loss": 0.9701, "step": 12164 }, { "epoch": 0.01, "learning_rate": 4.998204029211438e-05, "loss": 1.189, "step": 12165 }, { "epoch": 0.01, "learning_rate": 4.998203731518921e-05, "loss": 1.275, "step": 12166 }, { "epoch": 0.01, "learning_rate": 4.998203433801743e-05, "loss": 1.1439, "step": 12167 }, { "epoch": 0.01, "learning_rate": 4.998203136059904e-05, "loss": 0.8104, "step": 12168 }, { "epoch": 0.01, "learning_rate": 4.9982028382934034e-05, "loss": 1.1535, "step": 12169 }, { "epoch": 0.01, "learning_rate": 4.9982025405022416e-05, "loss": 1.0484, "step": 12170 }, { "epoch": 0.01, "learning_rate": 4.998202242686418e-05, "loss": 0.9965, "step": 12171 }, { "epoch": 0.01, "learning_rate": 4.9982019448459346e-05, "loss": 0.8001, "step": 12172 }, { "epoch": 0.01, "learning_rate": 4.998201646980789e-05, "loss": 0.9192, "step": 12173 }, { "epoch": 0.01, "learning_rate": 4.998201349090983e-05, "loss": 1.187, "step": 12174 }, { "epoch": 0.01, "learning_rate": 4.9982010511765154e-05, "loss": 1.1789, "step": 12175 }, { "epoch": 0.01, "learning_rate": 4.9982007532373864e-05, "loss": 0.832, "step": 12176 }, { "epoch": 0.01, "learning_rate": 4.998200455273596e-05, "loss": 1.1056, "step": 12177 }, { "epoch": 0.01, "learning_rate": 4.998200157285145e-05, "loss": 1.0418, "step": 12178 }, { "epoch": 0.01, "learning_rate": 4.998199859272032e-05, "loss": 0.8702, "step": 12179 }, { "epoch": 0.01, "learning_rate": 4.998199561234259e-05, "loss": 1.1537, "step": 12180 }, { "epoch": 0.01, "learning_rate": 4.9981992631718244e-05, "loss": 1.0839, "step": 12181 }, { "epoch": 0.01, "learning_rate": 4.998198965084728e-05, "loss": 0.9476, "step": 12182 }, { "epoch": 0.01, "learning_rate": 4.9981986669729705e-05, "loss": 0.5703, "step": 12183 }, { "epoch": 0.01, "learning_rate": 4.998198368836553e-05, "loss": 0.6009, "step": 12184 }, { "epoch": 0.01, "learning_rate": 4.998198070675473e-05, "loss": 1.3527, "step": 12185 }, { "epoch": 0.01, "learning_rate": 4.998197772489733e-05, "loss": 1.3717, "step": 12186 }, { "epoch": 0.01, "learning_rate": 4.998197474279331e-05, "loss": 1.1882, "step": 12187 }, { "epoch": 0.01, "learning_rate": 4.998197176044268e-05, "loss": 0.9134, "step": 12188 }, { "epoch": 0.01, "learning_rate": 4.998196877784544e-05, "loss": 0.7254, "step": 12189 }, { "epoch": 0.01, "learning_rate": 4.998196579500159e-05, "loss": 0.9603, "step": 12190 }, { "epoch": 0.01, "learning_rate": 4.9981962811911125e-05, "loss": 1.0766, "step": 12191 }, { "epoch": 0.01, "learning_rate": 4.998195982857405e-05, "loss": 1.2593, "step": 12192 }, { "epoch": 0.01, "learning_rate": 4.998195684499036e-05, "loss": 1.2716, "step": 12193 }, { "epoch": 0.01, "learning_rate": 4.998195386116007e-05, "loss": 1.1075, "step": 12194 }, { "epoch": 0.01, "learning_rate": 4.998195087708316e-05, "loss": 1.1545, "step": 12195 }, { "epoch": 0.01, "learning_rate": 4.998194789275964e-05, "loss": 1.0028, "step": 12196 }, { "epoch": 0.01, "learning_rate": 4.99819449081895e-05, "loss": 1.1678, "step": 12197 }, { "epoch": 0.01, "learning_rate": 4.998194192337276e-05, "loss": 1.1907, "step": 12198 }, { "epoch": 0.01, "learning_rate": 4.998193893830941e-05, "loss": 1.3121, "step": 12199 }, { "epoch": 0.01, "learning_rate": 4.9981935952999445e-05, "loss": 0.7019, "step": 12200 }, { "epoch": 0.01, "learning_rate": 4.9981932967442866e-05, "loss": 1.1035, "step": 12201 }, { "epoch": 0.01, "learning_rate": 4.998192998163968e-05, "loss": 1.6263, "step": 12202 }, { "epoch": 0.01, "learning_rate": 4.9981926995589876e-05, "loss": 0.9957, "step": 12203 }, { "epoch": 0.01, "learning_rate": 4.998192400929347e-05, "loss": 0.9056, "step": 12204 }, { "epoch": 0.01, "learning_rate": 4.998192102275045e-05, "loss": 0.8507, "step": 12205 }, { "epoch": 0.01, "learning_rate": 4.998191803596082e-05, "loss": 0.9602, "step": 12206 }, { "epoch": 0.01, "learning_rate": 4.998191504892458e-05, "loss": 0.8681, "step": 12207 }, { "epoch": 0.01, "learning_rate": 4.998191206164172e-05, "loss": 0.7919, "step": 12208 }, { "epoch": 0.01, "learning_rate": 4.998190907411226e-05, "loss": 0.784, "step": 12209 }, { "epoch": 0.01, "learning_rate": 4.9981906086336184e-05, "loss": 0.739, "step": 12210 }, { "epoch": 0.01, "learning_rate": 4.99819030983135e-05, "loss": 0.7427, "step": 12211 }, { "epoch": 0.01, "learning_rate": 4.99819001100442e-05, "loss": 0.7395, "step": 12212 }, { "epoch": 0.01, "learning_rate": 4.99818971215283e-05, "loss": 0.6996, "step": 12213 }, { "epoch": 0.01, "learning_rate": 4.998189413276578e-05, "loss": 0.7456, "step": 12214 }, { "epoch": 0.01, "learning_rate": 4.998189114375665e-05, "loss": 0.7073, "step": 12215 }, { "epoch": 0.01, "learning_rate": 4.9981888154500914e-05, "loss": 0.7032, "step": 12216 }, { "epoch": 0.01, "learning_rate": 4.998188516499855e-05, "loss": 0.6384, "step": 12217 }, { "epoch": 0.01, "learning_rate": 4.998188217524959e-05, "loss": 0.6803, "step": 12218 }, { "epoch": 0.01, "learning_rate": 4.998187918525402e-05, "loss": 0.6206, "step": 12219 }, { "epoch": 0.01, "learning_rate": 4.9981876195011844e-05, "loss": 0.6704, "step": 12220 }, { "epoch": 0.01, "learning_rate": 4.998187320452305e-05, "loss": 1.6482, "step": 12221 }, { "epoch": 0.01, "learning_rate": 4.998187021378765e-05, "loss": 0.9323, "step": 12222 }, { "epoch": 0.01, "learning_rate": 4.998186722280563e-05, "loss": 1.0517, "step": 12223 }, { "epoch": 0.01, "learning_rate": 4.998186423157701e-05, "loss": 0.9098, "step": 12224 }, { "epoch": 0.01, "learning_rate": 4.9981861240101765e-05, "loss": 1.2377, "step": 12225 }, { "epoch": 0.01, "learning_rate": 4.9981858248379925e-05, "loss": 1.1893, "step": 12226 }, { "epoch": 0.01, "learning_rate": 4.998185525641147e-05, "loss": 0.9471, "step": 12227 }, { "epoch": 0.01, "learning_rate": 4.9981852264196405e-05, "loss": 1.2663, "step": 12228 }, { "epoch": 0.01, "learning_rate": 4.9981849271734726e-05, "loss": 1.4243, "step": 12229 }, { "epoch": 0.01, "learning_rate": 4.9981846279026445e-05, "loss": 0.8721, "step": 12230 }, { "epoch": 0.01, "learning_rate": 4.998184328607154e-05, "loss": 1.2458, "step": 12231 }, { "epoch": 0.01, "learning_rate": 4.998184029287004e-05, "loss": 1.2852, "step": 12232 }, { "epoch": 0.01, "learning_rate": 4.9981837299421916e-05, "loss": 1.1038, "step": 12233 }, { "epoch": 0.01, "learning_rate": 4.9981834305727194e-05, "loss": 1.0695, "step": 12234 }, { "epoch": 0.01, "learning_rate": 4.9981831311785856e-05, "loss": 1.1814, "step": 12235 }, { "epoch": 0.01, "learning_rate": 4.998182831759791e-05, "loss": 1.1959, "step": 12236 }, { "epoch": 0.01, "learning_rate": 4.998182532316335e-05, "loss": 0.9837, "step": 12237 }, { "epoch": 0.01, "learning_rate": 4.9981822328482186e-05, "loss": 1.0547, "step": 12238 }, { "epoch": 0.01, "learning_rate": 4.99818193335544e-05, "loss": 1.3387, "step": 12239 }, { "epoch": 0.01, "learning_rate": 4.998181633838001e-05, "loss": 1.2434, "step": 12240 }, { "epoch": 0.01, "learning_rate": 4.998181334295902e-05, "loss": 1.1832, "step": 12241 }, { "epoch": 0.01, "learning_rate": 4.998181034729141e-05, "loss": 2.241, "step": 12242 }, { "epoch": 0.01, "learning_rate": 4.9981807351377194e-05, "loss": 1.0094, "step": 12243 }, { "epoch": 0.01, "learning_rate": 4.9981804355216366e-05, "loss": 0.8128, "step": 12244 }, { "epoch": 0.01, "learning_rate": 4.998180135880893e-05, "loss": 1.1457, "step": 12245 }, { "epoch": 0.01, "learning_rate": 4.998179836215488e-05, "loss": 1.1867, "step": 12246 }, { "epoch": 0.01, "learning_rate": 4.998179536525422e-05, "loss": 1.2103, "step": 12247 }, { "epoch": 0.01, "learning_rate": 4.998179236810696e-05, "loss": 1.129, "step": 12248 }, { "epoch": 0.01, "learning_rate": 4.998178937071308e-05, "loss": 1.0006, "step": 12249 }, { "epoch": 0.01, "learning_rate": 4.998178637307259e-05, "loss": 1.1473, "step": 12250 }, { "epoch": 0.01, "learning_rate": 4.9981783375185495e-05, "loss": 1.2892, "step": 12251 }, { "epoch": 0.01, "learning_rate": 4.998178037705179e-05, "loss": 1.1412, "step": 12252 }, { "epoch": 0.01, "learning_rate": 4.998177737867148e-05, "loss": 0.8711, "step": 12253 }, { "epoch": 0.01, "learning_rate": 4.998177438004455e-05, "loss": 0.9427, "step": 12254 }, { "epoch": 0.01, "learning_rate": 4.998177138117102e-05, "loss": 1.1894, "step": 12255 }, { "epoch": 0.01, "learning_rate": 4.998176838205087e-05, "loss": 1.3187, "step": 12256 }, { "epoch": 0.01, "learning_rate": 4.998176538268412e-05, "loss": 1.5896, "step": 12257 }, { "epoch": 0.01, "learning_rate": 4.9981762383070765e-05, "loss": 1.7242, "step": 12258 }, { "epoch": 0.01, "learning_rate": 4.998175938321079e-05, "loss": 1.3832, "step": 12259 }, { "epoch": 0.01, "learning_rate": 4.99817563831042e-05, "loss": 1.3111, "step": 12260 }, { "epoch": 0.01, "learning_rate": 4.9981753382751015e-05, "loss": 1.1332, "step": 12261 }, { "epoch": 0.01, "learning_rate": 4.998175038215121e-05, "loss": 0.7401, "step": 12262 }, { "epoch": 0.01, "learning_rate": 4.998174738130481e-05, "loss": 1.1572, "step": 12263 }, { "epoch": 0.01, "learning_rate": 4.998174438021179e-05, "loss": 0.963, "step": 12264 }, { "epoch": 0.01, "learning_rate": 4.998174137887216e-05, "loss": 0.9892, "step": 12265 }, { "epoch": 0.01, "learning_rate": 4.9981738377285924e-05, "loss": 2.0799, "step": 12266 }, { "epoch": 0.01, "learning_rate": 4.998173537545308e-05, "loss": 1.0048, "step": 12267 }, { "epoch": 0.01, "learning_rate": 4.998173237337362e-05, "loss": 1.4656, "step": 12268 }, { "epoch": 0.01, "learning_rate": 4.9981729371047555e-05, "loss": 0.895, "step": 12269 }, { "epoch": 0.01, "learning_rate": 4.998172636847488e-05, "loss": 1.1404, "step": 12270 }, { "epoch": 0.01, "learning_rate": 4.99817233656556e-05, "loss": 1.145, "step": 12271 }, { "epoch": 0.01, "learning_rate": 4.998172036258971e-05, "loss": 1.2568, "step": 12272 }, { "epoch": 0.01, "learning_rate": 4.99817173592772e-05, "loss": 1.4198, "step": 12273 }, { "epoch": 0.01, "learning_rate": 4.9981714355718094e-05, "loss": 1.2216, "step": 12274 }, { "epoch": 0.01, "learning_rate": 4.9981711351912374e-05, "loss": 1.1723, "step": 12275 }, { "epoch": 0.01, "learning_rate": 4.9981708347860044e-05, "loss": 0.9818, "step": 12276 }, { "epoch": 0.01, "learning_rate": 4.9981705343561114e-05, "loss": 1.1023, "step": 12277 }, { "epoch": 0.01, "learning_rate": 4.998170233901557e-05, "loss": 0.602, "step": 12278 }, { "epoch": 0.01, "learning_rate": 4.9981699334223406e-05, "loss": 0.888, "step": 12279 }, { "epoch": 0.01, "learning_rate": 4.998169632918465e-05, "loss": 0.8708, "step": 12280 }, { "epoch": 0.01, "learning_rate": 4.998169332389928e-05, "loss": 1.2637, "step": 12281 }, { "epoch": 0.01, "learning_rate": 4.998169031836729e-05, "loss": 1.0338, "step": 12282 }, { "epoch": 0.01, "learning_rate": 4.998168731258871e-05, "loss": 0.7036, "step": 12283 }, { "epoch": 0.01, "learning_rate": 4.99816843065635e-05, "loss": 0.7508, "step": 12284 }, { "epoch": 0.01, "learning_rate": 4.99816813002917e-05, "loss": 0.584, "step": 12285 }, { "epoch": 0.01, "learning_rate": 4.998167829377328e-05, "loss": 0.4087, "step": 12286 }, { "epoch": 0.01, "learning_rate": 4.998167528700826e-05, "loss": 1.1411, "step": 12287 }, { "epoch": 0.01, "learning_rate": 4.998167227999663e-05, "loss": 1.0434, "step": 12288 }, { "epoch": 0.01, "learning_rate": 4.9981669272738386e-05, "loss": 0.7202, "step": 12289 }, { "epoch": 0.01, "learning_rate": 4.998166626523354e-05, "loss": 0.831, "step": 12290 }, { "epoch": 0.01, "learning_rate": 4.998166325748208e-05, "loss": 0.7361, "step": 12291 }, { "epoch": 0.01, "learning_rate": 4.998166024948401e-05, "loss": 0.954, "step": 12292 }, { "epoch": 0.01, "learning_rate": 4.998165724123933e-05, "loss": 1.6453, "step": 12293 }, { "epoch": 0.01, "learning_rate": 4.9981654232748056e-05, "loss": 1.0373, "step": 12294 }, { "epoch": 0.01, "learning_rate": 4.9981651224010164e-05, "loss": 0.9059, "step": 12295 }, { "epoch": 0.01, "learning_rate": 4.998164821502566e-05, "loss": 1.0726, "step": 12296 }, { "epoch": 0.01, "learning_rate": 4.9981645205794555e-05, "loss": 1.3252, "step": 12297 }, { "epoch": 0.01, "learning_rate": 4.998164219631684e-05, "loss": 1.1061, "step": 12298 }, { "epoch": 0.01, "learning_rate": 4.998163918659251e-05, "loss": 0.9898, "step": 12299 }, { "epoch": 0.01, "learning_rate": 4.9981636176621577e-05, "loss": 0.9904, "step": 12300 }, { "epoch": 0.01, "learning_rate": 4.9981633166404033e-05, "loss": 1.2752, "step": 12301 }, { "epoch": 0.01, "learning_rate": 4.998163015593988e-05, "loss": 0.8749, "step": 12302 }, { "epoch": 0.01, "learning_rate": 4.998162714522913e-05, "loss": 1.0561, "step": 12303 }, { "epoch": 0.01, "learning_rate": 4.9981624134271766e-05, "loss": 1.0469, "step": 12304 }, { "epoch": 0.01, "learning_rate": 4.998162112306779e-05, "loss": 1.0939, "step": 12305 }, { "epoch": 0.01, "learning_rate": 4.998161811161721e-05, "loss": 0.8517, "step": 12306 }, { "epoch": 0.01, "learning_rate": 4.9981615099920015e-05, "loss": 1.1539, "step": 12307 }, { "epoch": 0.01, "learning_rate": 4.998161208797621e-05, "loss": 1.1281, "step": 12308 }, { "epoch": 0.01, "learning_rate": 4.9981609075785814e-05, "loss": 0.976, "step": 12309 }, { "epoch": 0.01, "learning_rate": 4.9981606063348794e-05, "loss": 0.6655, "step": 12310 }, { "epoch": 0.01, "learning_rate": 4.998160305066518e-05, "loss": 1.7917, "step": 12311 }, { "epoch": 0.01, "learning_rate": 4.998160003773494e-05, "loss": 1.8101, "step": 12312 }, { "epoch": 0.01, "learning_rate": 4.9981597024558115e-05, "loss": 1.9419, "step": 12313 }, { "epoch": 0.01, "learning_rate": 4.998159401113466e-05, "loss": 1.9558, "step": 12314 }, { "epoch": 0.01, "learning_rate": 4.998159099746461e-05, "loss": 1.7066, "step": 12315 }, { "epoch": 0.01, "learning_rate": 4.9981587983547954e-05, "loss": 1.8039, "step": 12316 }, { "epoch": 0.01, "learning_rate": 4.998158496938468e-05, "loss": 1.7127, "step": 12317 }, { "epoch": 0.01, "learning_rate": 4.998158195497481e-05, "loss": 1.2072, "step": 12318 }, { "epoch": 0.01, "learning_rate": 4.998157894031832e-05, "loss": 0.3727, "step": 12319 }, { "epoch": 0.01, "learning_rate": 4.998157592541523e-05, "loss": 0.2639, "step": 12320 }, { "epoch": 0.01, "learning_rate": 4.998157291026553e-05, "loss": 0.5425, "step": 12321 }, { "epoch": 0.01, "learning_rate": 4.9981569894869226e-05, "loss": 0.5329, "step": 12322 }, { "epoch": 0.01, "learning_rate": 4.9981566879226314e-05, "loss": 0.7372, "step": 12323 }, { "epoch": 0.01, "learning_rate": 4.998156386333679e-05, "loss": 1.1371, "step": 12324 }, { "epoch": 0.01, "learning_rate": 4.998156084720066e-05, "loss": 1.0917, "step": 12325 }, { "epoch": 0.01, "learning_rate": 4.998155783081793e-05, "loss": 1.0032, "step": 12326 }, { "epoch": 0.01, "learning_rate": 4.998155481418858e-05, "loss": 1.1574, "step": 12327 }, { "epoch": 0.01, "learning_rate": 4.998155179731263e-05, "loss": 1.24, "step": 12328 }, { "epoch": 0.01, "learning_rate": 4.998154878019007e-05, "loss": 1.3952, "step": 12329 }, { "epoch": 0.01, "learning_rate": 4.9981545762820903e-05, "loss": 1.1931, "step": 12330 }, { "epoch": 0.01, "learning_rate": 4.998154274520514e-05, "loss": 0.7442, "step": 12331 }, { "epoch": 0.01, "learning_rate": 4.998153972734275e-05, "loss": 1.1324, "step": 12332 }, { "epoch": 0.01, "learning_rate": 4.9981536709233766e-05, "loss": 0.605, "step": 12333 }, { "epoch": 0.01, "learning_rate": 4.998153369087817e-05, "loss": 0.5132, "step": 12334 }, { "epoch": 0.01, "learning_rate": 4.998153067227597e-05, "loss": 0.4633, "step": 12335 }, { "epoch": 0.01, "learning_rate": 4.998152765342716e-05, "loss": 0.4099, "step": 12336 }, { "epoch": 0.01, "learning_rate": 4.998152463433175e-05, "loss": 0.445, "step": 12337 }, { "epoch": 0.01, "learning_rate": 4.998152161498972e-05, "loss": 0.5687, "step": 12338 }, { "epoch": 0.01, "learning_rate": 4.998151859540109e-05, "loss": 1.0694, "step": 12339 }, { "epoch": 0.01, "learning_rate": 4.998151557556585e-05, "loss": 1.2817, "step": 12340 }, { "epoch": 0.01, "learning_rate": 4.9981512555484005e-05, "loss": 1.1015, "step": 12341 }, { "epoch": 0.01, "learning_rate": 4.998150953515556e-05, "loss": 1.2148, "step": 12342 }, { "epoch": 0.01, "learning_rate": 4.99815065145805e-05, "loss": 1.1055, "step": 12343 }, { "epoch": 0.01, "learning_rate": 4.998150349375883e-05, "loss": 0.8014, "step": 12344 }, { "epoch": 0.01, "learning_rate": 4.9981500472690554e-05, "loss": 1.1865, "step": 12345 }, { "epoch": 0.01, "learning_rate": 4.998149745137568e-05, "loss": 0.8474, "step": 12346 }, { "epoch": 0.01, "learning_rate": 4.998149442981419e-05, "loss": 1.171, "step": 12347 }, { "epoch": 0.01, "learning_rate": 4.99814914080061e-05, "loss": 1.1856, "step": 12348 }, { "epoch": 0.01, "learning_rate": 4.9981488385951406e-05, "loss": 1.1312, "step": 12349 }, { "epoch": 0.01, "learning_rate": 4.998148536365009e-05, "loss": 1.0987, "step": 12350 }, { "epoch": 0.01, "learning_rate": 4.998148234110218e-05, "loss": 1.1335, "step": 12351 }, { "epoch": 0.01, "learning_rate": 4.998147931830767e-05, "loss": 1.0597, "step": 12352 }, { "epoch": 0.01, "learning_rate": 4.998147629526654e-05, "loss": 1.0757, "step": 12353 }, { "epoch": 0.01, "learning_rate": 4.9981473271978805e-05, "loss": 0.9938, "step": 12354 }, { "epoch": 0.01, "learning_rate": 4.998147024844446e-05, "loss": 1.2549, "step": 12355 }, { "epoch": 0.01, "learning_rate": 4.998146722466352e-05, "loss": 1.1512, "step": 12356 }, { "epoch": 0.01, "learning_rate": 4.9981464200635966e-05, "loss": 1.1168, "step": 12357 }, { "epoch": 0.01, "learning_rate": 4.998146117636181e-05, "loss": 0.9795, "step": 12358 }, { "epoch": 0.01, "learning_rate": 4.9981458151841045e-05, "loss": 1.1687, "step": 12359 }, { "epoch": 0.01, "learning_rate": 4.998145512707367e-05, "loss": 1.1541, "step": 12360 }, { "epoch": 0.01, "learning_rate": 4.998145210205969e-05, "loss": 1.0797, "step": 12361 }, { "epoch": 0.01, "learning_rate": 4.998144907679911e-05, "loss": 1.126, "step": 12362 }, { "epoch": 0.01, "learning_rate": 4.998144605129191e-05, "loss": 1.1401, "step": 12363 }, { "epoch": 0.01, "learning_rate": 4.9981443025538125e-05, "loss": 0.9566, "step": 12364 }, { "epoch": 0.01, "learning_rate": 4.9981439999537715e-05, "loss": 1.0134, "step": 12365 }, { "epoch": 0.01, "learning_rate": 4.998143697329071e-05, "loss": 1.2251, "step": 12366 }, { "epoch": 0.01, "learning_rate": 4.998143394679709e-05, "loss": 1.0263, "step": 12367 }, { "epoch": 0.01, "learning_rate": 4.998143092005687e-05, "loss": 1.1975, "step": 12368 }, { "epoch": 0.01, "learning_rate": 4.9981427893070045e-05, "loss": 1.7828, "step": 12369 }, { "epoch": 0.01, "learning_rate": 4.9981424865836606e-05, "loss": 0.8206, "step": 12370 }, { "epoch": 0.01, "learning_rate": 4.9981421838356565e-05, "loss": 0.7618, "step": 12371 }, { "epoch": 0.01, "learning_rate": 4.998141881062992e-05, "loss": 0.5565, "step": 12372 }, { "epoch": 0.01, "learning_rate": 4.9981415782656664e-05, "loss": 0.4305, "step": 12373 }, { "epoch": 0.01, "learning_rate": 4.9981412754436805e-05, "loss": 0.3556, "step": 12374 }, { "epoch": 0.01, "learning_rate": 4.998140972597034e-05, "loss": 0.7321, "step": 12375 }, { "epoch": 0.01, "learning_rate": 4.998140669725727e-05, "loss": 0.7104, "step": 12376 }, { "epoch": 0.01, "learning_rate": 4.99814036682976e-05, "loss": 0.4816, "step": 12377 }, { "epoch": 0.01, "learning_rate": 4.9981400639091314e-05, "loss": 0.211, "step": 12378 }, { "epoch": 0.01, "learning_rate": 4.9981397609638426e-05, "loss": 0.319, "step": 12379 }, { "epoch": 0.01, "learning_rate": 4.9981394579938935e-05, "loss": 0.5376, "step": 12380 }, { "epoch": 0.01, "learning_rate": 4.998139154999283e-05, "loss": 0.5669, "step": 12381 }, { "epoch": 0.01, "learning_rate": 4.998138851980013e-05, "loss": 0.6685, "step": 12382 }, { "epoch": 0.01, "learning_rate": 4.998138548936082e-05, "loss": 0.3601, "step": 12383 }, { "epoch": 0.01, "learning_rate": 4.99813824586749e-05, "loss": 0.6418, "step": 12384 }, { "epoch": 0.01, "learning_rate": 4.998137942774237e-05, "loss": 0.9678, "step": 12385 }, { "epoch": 0.01, "learning_rate": 4.998137639656325e-05, "loss": 0.7382, "step": 12386 }, { "epoch": 0.01, "learning_rate": 4.9981373365137515e-05, "loss": 0.8277, "step": 12387 }, { "epoch": 0.01, "learning_rate": 4.998137033346517e-05, "loss": 0.4406, "step": 12388 }, { "epoch": 0.01, "learning_rate": 4.998136730154623e-05, "loss": 0.6967, "step": 12389 }, { "epoch": 0.01, "learning_rate": 4.998136426938068e-05, "loss": 0.4694, "step": 12390 }, { "epoch": 0.01, "learning_rate": 4.998136123696852e-05, "loss": 0.5338, "step": 12391 }, { "epoch": 0.01, "learning_rate": 4.9981358204309755e-05, "loss": 0.6247, "step": 12392 }, { "epoch": 0.01, "learning_rate": 4.9981355171404394e-05, "loss": 0.3445, "step": 12393 }, { "epoch": 0.01, "learning_rate": 4.9981352138252424e-05, "loss": 0.297, "step": 12394 }, { "epoch": 0.01, "learning_rate": 4.998134910485384e-05, "loss": 0.9048, "step": 12395 }, { "epoch": 0.01, "learning_rate": 4.998134607120866e-05, "loss": 1.2166, "step": 12396 }, { "epoch": 0.01, "learning_rate": 4.9981343037316876e-05, "loss": 1.181, "step": 12397 }, { "epoch": 0.01, "learning_rate": 4.998134000317848e-05, "loss": 1.2389, "step": 12398 }, { "epoch": 0.01, "learning_rate": 4.998133696879348e-05, "loss": 1.0067, "step": 12399 }, { "epoch": 0.01, "learning_rate": 4.998133393416187e-05, "loss": 1.1272, "step": 12400 }, { "epoch": 0.01, "learning_rate": 4.998133089928366e-05, "loss": 1.2163, "step": 12401 }, { "epoch": 0.01, "learning_rate": 4.998132786415885e-05, "loss": 0.8942, "step": 12402 }, { "epoch": 0.01, "learning_rate": 4.998132482878743e-05, "loss": 1.1119, "step": 12403 }, { "epoch": 0.01, "learning_rate": 4.99813217931694e-05, "loss": 1.3623, "step": 12404 }, { "epoch": 0.01, "learning_rate": 4.998131875730477e-05, "loss": 0.9595, "step": 12405 }, { "epoch": 0.01, "learning_rate": 4.998131572119354e-05, "loss": 0.7142, "step": 12406 }, { "epoch": 0.01, "learning_rate": 4.99813126848357e-05, "loss": 1.2325, "step": 12407 }, { "epoch": 0.01, "learning_rate": 4.998130964823126e-05, "loss": 1.4484, "step": 12408 }, { "epoch": 0.01, "learning_rate": 4.998130661138021e-05, "loss": 1.345, "step": 12409 }, { "epoch": 0.01, "learning_rate": 4.998130357428255e-05, "loss": 1.3685, "step": 12410 }, { "epoch": 0.01, "learning_rate": 4.9981300536938294e-05, "loss": 1.1315, "step": 12411 }, { "epoch": 0.01, "learning_rate": 4.9981297499347424e-05, "loss": 0.7519, "step": 12412 }, { "epoch": 0.01, "learning_rate": 4.998129446150995e-05, "loss": 1.1386, "step": 12413 }, { "epoch": 0.01, "learning_rate": 4.998129142342588e-05, "loss": 1.145, "step": 12414 }, { "epoch": 0.01, "learning_rate": 4.9981288385095204e-05, "loss": 1.0816, "step": 12415 }, { "epoch": 0.01, "learning_rate": 4.9981285346517914e-05, "loss": 1.1919, "step": 12416 }, { "epoch": 0.01, "learning_rate": 4.998128230769403e-05, "loss": 1.1038, "step": 12417 }, { "epoch": 0.01, "learning_rate": 4.998127926862354e-05, "loss": 1.1895, "step": 12418 }, { "epoch": 0.01, "learning_rate": 4.998127622930644e-05, "loss": 1.1287, "step": 12419 }, { "epoch": 0.01, "learning_rate": 4.9981273189742734e-05, "loss": 1.1634, "step": 12420 }, { "epoch": 0.01, "learning_rate": 4.998127014993242e-05, "loss": 0.6406, "step": 12421 }, { "epoch": 0.01, "learning_rate": 4.998126710987552e-05, "loss": 0.8229, "step": 12422 }, { "epoch": 0.01, "learning_rate": 4.9981264069572e-05, "loss": 1.1455, "step": 12423 }, { "epoch": 0.01, "learning_rate": 4.998126102902188e-05, "loss": 1.0545, "step": 12424 }, { "epoch": 0.01, "learning_rate": 4.998125798822515e-05, "loss": 1.0934, "step": 12425 }, { "epoch": 0.01, "learning_rate": 4.998125494718182e-05, "loss": 1.0143, "step": 12426 }, { "epoch": 0.01, "learning_rate": 4.998125190589189e-05, "loss": 1.1201, "step": 12427 }, { "epoch": 0.01, "learning_rate": 4.998124886435535e-05, "loss": 0.9573, "step": 12428 }, { "epoch": 0.01, "learning_rate": 4.9981245822572206e-05, "loss": 0.8915, "step": 12429 }, { "epoch": 0.01, "learning_rate": 4.998124278054246e-05, "loss": 1.3156, "step": 12430 }, { "epoch": 0.01, "learning_rate": 4.9981239738266106e-05, "loss": 1.2664, "step": 12431 }, { "epoch": 0.01, "learning_rate": 4.998123669574315e-05, "loss": 1.0817, "step": 12432 }, { "epoch": 0.01, "learning_rate": 4.998123365297359e-05, "loss": 0.702, "step": 12433 }, { "epoch": 0.01, "learning_rate": 4.998123060995743e-05, "loss": 1.0154, "step": 12434 }, { "epoch": 0.01, "learning_rate": 4.9981227566694656e-05, "loss": 1.2546, "step": 12435 }, { "epoch": 0.01, "learning_rate": 4.998122452318529e-05, "loss": 1.1136, "step": 12436 }, { "epoch": 0.01, "learning_rate": 4.998122147942931e-05, "loss": 0.9993, "step": 12437 }, { "epoch": 0.01, "learning_rate": 4.9981218435426724e-05, "loss": 1.0578, "step": 12438 }, { "epoch": 0.01, "learning_rate": 4.998121539117754e-05, "loss": 0.7509, "step": 12439 }, { "epoch": 0.01, "learning_rate": 4.998121234668175e-05, "loss": 1.0293, "step": 12440 }, { "epoch": 0.01, "learning_rate": 4.998120930193936e-05, "loss": 1.2657, "step": 12441 }, { "epoch": 0.01, "learning_rate": 4.998120625695036e-05, "loss": 1.0948, "step": 12442 }, { "epoch": 0.01, "learning_rate": 4.9981203211714764e-05, "loss": 1.2965, "step": 12443 }, { "epoch": 0.01, "learning_rate": 4.9981200166232556e-05, "loss": 1.1268, "step": 12444 }, { "epoch": 0.01, "learning_rate": 4.9981197120503746e-05, "loss": 1.1302, "step": 12445 }, { "epoch": 0.01, "learning_rate": 4.9981194074528334e-05, "loss": 1.464, "step": 12446 }, { "epoch": 0.01, "learning_rate": 4.998119102830632e-05, "loss": 1.2115, "step": 12447 }, { "epoch": 0.01, "learning_rate": 4.99811879818377e-05, "loss": 1.1439, "step": 12448 }, { "epoch": 0.01, "learning_rate": 4.9981184935122474e-05, "loss": 1.3663, "step": 12449 }, { "epoch": 0.01, "learning_rate": 4.998118188816064e-05, "loss": 1.1248, "step": 12450 }, { "epoch": 0.01, "learning_rate": 4.9981178840952215e-05, "loss": 1.0585, "step": 12451 }, { "epoch": 0.01, "learning_rate": 4.998117579349718e-05, "loss": 1.1164, "step": 12452 }, { "epoch": 0.01, "learning_rate": 4.998117274579555e-05, "loss": 1.0238, "step": 12453 }, { "epoch": 0.01, "learning_rate": 4.9981169697847294e-05, "loss": 0.7298, "step": 12454 }, { "epoch": 0.01, "learning_rate": 4.998116664965245e-05, "loss": 1.1957, "step": 12455 }, { "epoch": 0.01, "learning_rate": 4.9981163601211e-05, "loss": 1.012, "step": 12456 }, { "epoch": 0.01, "learning_rate": 4.998116055252295e-05, "loss": 1.1416, "step": 12457 }, { "epoch": 0.01, "learning_rate": 4.998115750358829e-05, "loss": 1.1676, "step": 12458 }, { "epoch": 0.01, "learning_rate": 4.9981154454407036e-05, "loss": 1.1385, "step": 12459 }, { "epoch": 0.01, "learning_rate": 4.998115140497918e-05, "loss": 1.0225, "step": 12460 }, { "epoch": 0.01, "learning_rate": 4.998114835530471e-05, "loss": 0.7041, "step": 12461 }, { "epoch": 0.01, "learning_rate": 4.998114530538364e-05, "loss": 0.8163, "step": 12462 }, { "epoch": 0.01, "learning_rate": 4.998114225521596e-05, "loss": 0.5854, "step": 12463 }, { "epoch": 0.01, "learning_rate": 4.9981139204801686e-05, "loss": 0.4704, "step": 12464 }, { "epoch": 0.01, "learning_rate": 4.9981136154140806e-05, "loss": 0.5542, "step": 12465 }, { "epoch": 0.01, "learning_rate": 4.9981133103233325e-05, "loss": 0.5285, "step": 12466 }, { "epoch": 0.01, "learning_rate": 4.998113005207924e-05, "loss": 0.6402, "step": 12467 }, { "epoch": 0.01, "learning_rate": 4.998112700067854e-05, "loss": 0.7608, "step": 12468 }, { "epoch": 0.01, "learning_rate": 4.998112394903125e-05, "loss": 0.8731, "step": 12469 }, { "epoch": 0.01, "learning_rate": 4.9981120897137355e-05, "loss": 0.7127, "step": 12470 }, { "epoch": 0.01, "learning_rate": 4.998111784499686e-05, "loss": 1.0558, "step": 12471 }, { "epoch": 0.01, "learning_rate": 4.998111479260976e-05, "loss": 1.4552, "step": 12472 }, { "epoch": 0.01, "learning_rate": 4.998111173997605e-05, "loss": 0.5671, "step": 12473 }, { "epoch": 0.01, "learning_rate": 4.998110868709574e-05, "loss": 1.531, "step": 12474 }, { "epoch": 0.01, "learning_rate": 4.9981105633968825e-05, "loss": 0.9289, "step": 12475 }, { "epoch": 0.01, "learning_rate": 4.998110258059532e-05, "loss": 0.9232, "step": 12476 }, { "epoch": 0.01, "learning_rate": 4.99810995269752e-05, "loss": 1.111, "step": 12477 }, { "epoch": 0.01, "learning_rate": 4.998109647310848e-05, "loss": 1.0808, "step": 12478 }, { "epoch": 0.01, "learning_rate": 4.9981093418995157e-05, "loss": 1.274, "step": 12479 }, { "epoch": 0.01, "learning_rate": 4.998109036463523e-05, "loss": 1.0235, "step": 12480 }, { "epoch": 0.01, "learning_rate": 4.998108731002871e-05, "loss": 1.1525, "step": 12481 }, { "epoch": 0.01, "learning_rate": 4.998108425517557e-05, "loss": 1.1978, "step": 12482 }, { "epoch": 0.01, "learning_rate": 4.998108120007584e-05, "loss": 1.1724, "step": 12483 }, { "epoch": 0.01, "learning_rate": 4.99810781447295e-05, "loss": 1.2258, "step": 12484 }, { "epoch": 0.01, "learning_rate": 4.9981075089136563e-05, "loss": 1.1161, "step": 12485 }, { "epoch": 0.01, "learning_rate": 4.998107203329702e-05, "loss": 0.6646, "step": 12486 }, { "epoch": 0.01, "learning_rate": 4.9981068977210873e-05, "loss": 0.4699, "step": 12487 }, { "epoch": 0.01, "learning_rate": 4.9981065920878126e-05, "loss": 0.4123, "step": 12488 }, { "epoch": 0.01, "learning_rate": 4.9981062864298776e-05, "loss": 0.3083, "step": 12489 }, { "epoch": 0.01, "learning_rate": 4.998105980747283e-05, "loss": 0.8927, "step": 12490 }, { "epoch": 0.01, "learning_rate": 4.9981056750400265e-05, "loss": 1.4359, "step": 12491 }, { "epoch": 0.01, "learning_rate": 4.998105369308112e-05, "loss": 0.9881, "step": 12492 }, { "epoch": 0.01, "learning_rate": 4.998105063551536e-05, "loss": 1.3489, "step": 12493 }, { "epoch": 0.01, "learning_rate": 4.998104757770299e-05, "loss": 1.2331, "step": 12494 }, { "epoch": 0.01, "learning_rate": 4.998104451964403e-05, "loss": 0.9847, "step": 12495 }, { "epoch": 0.01, "learning_rate": 4.998104146133846e-05, "loss": 1.0687, "step": 12496 }, { "epoch": 0.01, "learning_rate": 4.9981038402786296e-05, "loss": 1.0977, "step": 12497 }, { "epoch": 0.01, "learning_rate": 4.998103534398752e-05, "loss": 1.1294, "step": 12498 }, { "epoch": 0.01, "learning_rate": 4.998103228494215e-05, "loss": 0.8384, "step": 12499 }, { "epoch": 0.01, "learning_rate": 4.9981029225650174e-05, "loss": 0.8916, "step": 12500 }, { "epoch": 0.01, "eval_loss": 1.0588072538375854, "eval_runtime": 85.3376, "eval_samples_per_second": 16.23, "eval_steps_per_second": 4.066, "step": 12500 }, { "epoch": 0.01, "learning_rate": 4.9981026166111597e-05, "loss": 0.9107, "step": 12501 }, { "epoch": 0.01, "learning_rate": 4.998102310632641e-05, "loss": 1.0434, "step": 12502 }, { "epoch": 0.01, "learning_rate": 4.998102004629463e-05, "loss": 0.9114, "step": 12503 }, { "epoch": 0.01, "learning_rate": 4.9981016986016246e-05, "loss": 1.0933, "step": 12504 }, { "epoch": 0.01, "learning_rate": 4.998101392549126e-05, "loss": 1.0656, "step": 12505 }, { "epoch": 0.01, "learning_rate": 4.998101086471967e-05, "loss": 0.8722, "step": 12506 }, { "epoch": 0.01, "learning_rate": 4.998100780370148e-05, "loss": 0.9555, "step": 12507 }, { "epoch": 0.01, "learning_rate": 4.998100474243669e-05, "loss": 1.1079, "step": 12508 }, { "epoch": 0.01, "learning_rate": 4.998100168092529e-05, "loss": 1.1997, "step": 12509 }, { "epoch": 0.01, "learning_rate": 4.99809986191673e-05, "loss": 0.8179, "step": 12510 }, { "epoch": 0.01, "learning_rate": 4.99809955571627e-05, "loss": 0.8281, "step": 12511 }, { "epoch": 0.01, "learning_rate": 4.99809924949115e-05, "loss": 1.0142, "step": 12512 }, { "epoch": 0.01, "learning_rate": 4.998098943241369e-05, "loss": 1.0984, "step": 12513 }, { "epoch": 0.01, "learning_rate": 4.998098636966929e-05, "loss": 1.2993, "step": 12514 }, { "epoch": 0.01, "learning_rate": 4.998098330667829e-05, "loss": 0.802, "step": 12515 }, { "epoch": 0.01, "learning_rate": 4.998098024344068e-05, "loss": 1.1062, "step": 12516 }, { "epoch": 0.01, "learning_rate": 4.9980977179956465e-05, "loss": 1.3741, "step": 12517 }, { "epoch": 0.01, "learning_rate": 4.9980974116225657e-05, "loss": 1.2558, "step": 12518 }, { "epoch": 0.01, "learning_rate": 4.998097105224825e-05, "loss": 0.8686, "step": 12519 }, { "epoch": 0.01, "learning_rate": 4.998096798802423e-05, "loss": 0.5407, "step": 12520 }, { "epoch": 0.01, "learning_rate": 4.998096492355362e-05, "loss": 0.9241, "step": 12521 }, { "epoch": 0.01, "learning_rate": 4.99809618588364e-05, "loss": 1.1364, "step": 12522 }, { "epoch": 0.01, "learning_rate": 4.9980958793872577e-05, "loss": 1.2324, "step": 12523 }, { "epoch": 0.01, "learning_rate": 4.998095572866216e-05, "loss": 0.91, "step": 12524 }, { "epoch": 0.01, "learning_rate": 4.998095266320514e-05, "loss": 1.3726, "step": 12525 }, { "epoch": 0.01, "learning_rate": 4.9980949597501515e-05, "loss": 1.0983, "step": 12526 }, { "epoch": 0.01, "learning_rate": 4.998094653155129e-05, "loss": 1.1695, "step": 12527 }, { "epoch": 0.01, "learning_rate": 4.9980943465354465e-05, "loss": 1.1617, "step": 12528 }, { "epoch": 0.01, "learning_rate": 4.9980940398911045e-05, "loss": 1.1853, "step": 12529 }, { "epoch": 0.01, "learning_rate": 4.998093733222101e-05, "loss": 1.0265, "step": 12530 }, { "epoch": 0.01, "learning_rate": 4.9980934265284384e-05, "loss": 0.9487, "step": 12531 }, { "epoch": 0.01, "learning_rate": 4.998093119810115e-05, "loss": 0.7968, "step": 12532 }, { "epoch": 0.01, "learning_rate": 4.9980928130671316e-05, "loss": 1.2177, "step": 12533 }, { "epoch": 0.01, "learning_rate": 4.998092506299489e-05, "loss": 0.8995, "step": 12534 }, { "epoch": 0.01, "learning_rate": 4.998092199507185e-05, "loss": 1.1374, "step": 12535 }, { "epoch": 0.01, "learning_rate": 4.9980918926902215e-05, "loss": 1.1554, "step": 12536 }, { "epoch": 0.01, "learning_rate": 4.998091585848598e-05, "loss": 1.1725, "step": 12537 }, { "epoch": 0.01, "learning_rate": 4.998091278982314e-05, "loss": 1.2441, "step": 12538 }, { "epoch": 0.01, "learning_rate": 4.9980909720913705e-05, "loss": 0.9656, "step": 12539 }, { "epoch": 0.01, "learning_rate": 4.9980906651757664e-05, "loss": 1.0499, "step": 12540 }, { "epoch": 0.01, "learning_rate": 4.998090358235502e-05, "loss": 1.3264, "step": 12541 }, { "epoch": 0.01, "learning_rate": 4.998090051270578e-05, "loss": 0.9362, "step": 12542 }, { "epoch": 0.01, "learning_rate": 4.998089744280994e-05, "loss": 1.0823, "step": 12543 }, { "epoch": 0.01, "learning_rate": 4.998089437266749e-05, "loss": 1.3716, "step": 12544 }, { "epoch": 0.01, "learning_rate": 4.998089130227844e-05, "loss": 0.5029, "step": 12545 }, { "epoch": 0.01, "learning_rate": 4.99808882316428e-05, "loss": 1.1508, "step": 12546 }, { "epoch": 0.01, "learning_rate": 4.998088516076055e-05, "loss": 1.5744, "step": 12547 }, { "epoch": 0.01, "learning_rate": 4.998088208963171e-05, "loss": 1.0503, "step": 12548 }, { "epoch": 0.01, "learning_rate": 4.998087901825626e-05, "loss": 1.1751, "step": 12549 }, { "epoch": 0.01, "learning_rate": 4.998087594663421e-05, "loss": 1.1362, "step": 12550 }, { "epoch": 0.01, "learning_rate": 4.998087287476556e-05, "loss": 1.1305, "step": 12551 }, { "epoch": 0.01, "learning_rate": 4.998086980265031e-05, "loss": 1.2603, "step": 12552 }, { "epoch": 0.01, "learning_rate": 4.9980866730288464e-05, "loss": 1.2407, "step": 12553 }, { "epoch": 0.01, "learning_rate": 4.998086365768001e-05, "loss": 1.2634, "step": 12554 }, { "epoch": 0.01, "learning_rate": 4.998086058482496e-05, "loss": 0.9507, "step": 12555 }, { "epoch": 0.01, "learning_rate": 4.9980857511723303e-05, "loss": 0.9109, "step": 12556 }, { "epoch": 0.01, "learning_rate": 4.998085443837506e-05, "loss": 0.8525, "step": 12557 }, { "epoch": 0.01, "learning_rate": 4.99808513647802e-05, "loss": 1.0178, "step": 12558 }, { "epoch": 0.01, "learning_rate": 4.998084829093875e-05, "loss": 1.0699, "step": 12559 }, { "epoch": 0.01, "learning_rate": 4.998084521685069e-05, "loss": 1.0353, "step": 12560 }, { "epoch": 0.01, "learning_rate": 4.998084214251604e-05, "loss": 1.2007, "step": 12561 }, { "epoch": 0.01, "learning_rate": 4.998083906793478e-05, "loss": 1.2147, "step": 12562 }, { "epoch": 0.01, "learning_rate": 4.998083599310693e-05, "loss": 1.0938, "step": 12563 }, { "epoch": 0.01, "learning_rate": 4.998083291803247e-05, "loss": 0.8749, "step": 12564 }, { "epoch": 0.01, "learning_rate": 4.998082984271142e-05, "loss": 1.1915, "step": 12565 }, { "epoch": 0.01, "learning_rate": 4.998082676714376e-05, "loss": 1.2303, "step": 12566 }, { "epoch": 0.01, "learning_rate": 4.9980823691329505e-05, "loss": 1.3084, "step": 12567 }, { "epoch": 0.01, "learning_rate": 4.9980820615268655e-05, "loss": 1.021, "step": 12568 }, { "epoch": 0.01, "learning_rate": 4.99808175389612e-05, "loss": 0.8386, "step": 12569 }, { "epoch": 0.01, "learning_rate": 4.998081446240714e-05, "loss": 0.5628, "step": 12570 }, { "epoch": 0.01, "learning_rate": 4.998081138560649e-05, "loss": 0.394, "step": 12571 }, { "epoch": 0.01, "learning_rate": 4.9980808308559225e-05, "loss": 0.3166, "step": 12572 }, { "epoch": 0.01, "learning_rate": 4.998080523126537e-05, "loss": 0.3661, "step": 12573 }, { "epoch": 0.01, "learning_rate": 4.998080215372491e-05, "loss": 0.2504, "step": 12574 }, { "epoch": 0.01, "learning_rate": 4.998079907593786e-05, "loss": 0.1394, "step": 12575 }, { "epoch": 0.01, "learning_rate": 4.99807959979042e-05, "loss": 0.1276, "step": 12576 }, { "epoch": 0.01, "learning_rate": 4.998079291962394e-05, "loss": 0.0873, "step": 12577 }, { "epoch": 0.01, "learning_rate": 4.998078984109709e-05, "loss": 0.1252, "step": 12578 }, { "epoch": 0.01, "learning_rate": 4.998078676232363e-05, "loss": 0.1228, "step": 12579 }, { "epoch": 0.01, "learning_rate": 4.998078368330358e-05, "loss": 0.1049, "step": 12580 }, { "epoch": 0.01, "learning_rate": 4.998078060403692e-05, "loss": 0.2173, "step": 12581 }, { "epoch": 0.01, "learning_rate": 4.9980777524523667e-05, "loss": 0.1332, "step": 12582 }, { "epoch": 0.01, "learning_rate": 4.998077444476381e-05, "loss": 0.0962, "step": 12583 }, { "epoch": 0.01, "learning_rate": 4.998077136475736e-05, "loss": 0.0435, "step": 12584 }, { "epoch": 0.01, "learning_rate": 4.99807682845043e-05, "loss": 0.024, "step": 12585 }, { "epoch": 0.01, "learning_rate": 4.9980765204004645e-05, "loss": 0.0397, "step": 12586 }, { "epoch": 0.01, "learning_rate": 4.9980762123258395e-05, "loss": 0.0451, "step": 12587 }, { "epoch": 0.01, "learning_rate": 4.9980759042265544e-05, "loss": 0.0358, "step": 12588 }, { "epoch": 0.01, "learning_rate": 4.998075596102609e-05, "loss": 0.0299, "step": 12589 }, { "epoch": 0.01, "learning_rate": 4.9980752879540035e-05, "loss": 0.0337, "step": 12590 }, { "epoch": 0.01, "learning_rate": 4.9980749797807385e-05, "loss": 0.0337, "step": 12591 }, { "epoch": 0.01, "learning_rate": 4.998074671582813e-05, "loss": 0.0328, "step": 12592 }, { "epoch": 0.01, "learning_rate": 4.9980743633602286e-05, "loss": 0.0385, "step": 12593 }, { "epoch": 0.01, "learning_rate": 4.998074055112984e-05, "loss": 0.0282, "step": 12594 }, { "epoch": 0.01, "learning_rate": 4.998073746841079e-05, "loss": 0.123, "step": 12595 }, { "epoch": 0.01, "learning_rate": 4.9980734385445135e-05, "loss": 1.1476, "step": 12596 }, { "epoch": 0.01, "learning_rate": 4.998073130223289e-05, "loss": 1.3075, "step": 12597 }, { "epoch": 0.01, "learning_rate": 4.9980728218774045e-05, "loss": 1.1733, "step": 12598 }, { "epoch": 0.01, "learning_rate": 4.9980725135068595e-05, "loss": 1.0228, "step": 12599 }, { "epoch": 0.01, "learning_rate": 4.998072205111655e-05, "loss": 1.1655, "step": 12600 }, { "epoch": 0.01, "learning_rate": 4.998071896691791e-05, "loss": 1.2506, "step": 12601 }, { "epoch": 0.01, "learning_rate": 4.998071588247266e-05, "loss": 1.0348, "step": 12602 }, { "epoch": 0.01, "learning_rate": 4.9980712797780814e-05, "loss": 0.8461, "step": 12603 }, { "epoch": 0.01, "learning_rate": 4.9980709712842375e-05, "loss": 0.7891, "step": 12604 }, { "epoch": 0.01, "learning_rate": 4.9980706627657333e-05, "loss": 0.9184, "step": 12605 }, { "epoch": 0.01, "learning_rate": 4.998070354222569e-05, "loss": 1.1337, "step": 12606 }, { "epoch": 0.01, "learning_rate": 4.998070045654746e-05, "loss": 1.1779, "step": 12607 }, { "epoch": 0.01, "learning_rate": 4.998069737062262e-05, "loss": 1.1481, "step": 12608 }, { "epoch": 0.01, "learning_rate": 4.998069428445118e-05, "loss": 1.2349, "step": 12609 }, { "epoch": 0.01, "learning_rate": 4.998069119803315e-05, "loss": 1.1853, "step": 12610 }, { "epoch": 0.01, "learning_rate": 4.998068811136851e-05, "loss": 1.2069, "step": 12611 }, { "epoch": 0.01, "learning_rate": 4.998068502445728e-05, "loss": 0.9469, "step": 12612 }, { "epoch": 0.01, "learning_rate": 4.998068193729945e-05, "loss": 1.2307, "step": 12613 }, { "epoch": 0.01, "learning_rate": 4.998067884989502e-05, "loss": 0.8482, "step": 12614 }, { "epoch": 0.01, "learning_rate": 4.9980675762243986e-05, "loss": 1.2056, "step": 12615 }, { "epoch": 0.01, "learning_rate": 4.9980672674346366e-05, "loss": 0.9474, "step": 12616 }, { "epoch": 0.01, "learning_rate": 4.998066958620213e-05, "loss": 1.1598, "step": 12617 }, { "epoch": 0.01, "learning_rate": 4.9980666497811307e-05, "loss": 1.2106, "step": 12618 }, { "epoch": 0.01, "learning_rate": 4.998066340917388e-05, "loss": 1.0984, "step": 12619 }, { "epoch": 0.01, "learning_rate": 4.998066032028986e-05, "loss": 1.2588, "step": 12620 }, { "epoch": 0.01, "learning_rate": 4.998065723115924e-05, "loss": 1.129, "step": 12621 }, { "epoch": 0.01, "learning_rate": 4.998065414178202e-05, "loss": 0.9005, "step": 12622 }, { "epoch": 0.01, "learning_rate": 4.99806510521582e-05, "loss": 1.1697, "step": 12623 }, { "epoch": 0.01, "learning_rate": 4.998064796228779e-05, "loss": 1.2553, "step": 12624 }, { "epoch": 0.01, "learning_rate": 4.9980644872170765e-05, "loss": 1.0182, "step": 12625 }, { "epoch": 0.01, "learning_rate": 4.998064178180716e-05, "loss": 1.2193, "step": 12626 }, { "epoch": 0.01, "learning_rate": 4.998063869119695e-05, "loss": 1.0012, "step": 12627 }, { "epoch": 0.01, "learning_rate": 4.9980635600340134e-05, "loss": 1.1328, "step": 12628 }, { "epoch": 0.01, "learning_rate": 4.9980632509236724e-05, "loss": 1.2161, "step": 12629 }, { "epoch": 0.01, "learning_rate": 4.998062941788671e-05, "loss": 0.9976, "step": 12630 }, { "epoch": 0.01, "learning_rate": 4.9980626326290114e-05, "loss": 1.162, "step": 12631 }, { "epoch": 0.01, "learning_rate": 4.998062323444691e-05, "loss": 1.1436, "step": 12632 }, { "epoch": 0.01, "learning_rate": 4.99806201423571e-05, "loss": 1.2007, "step": 12633 }, { "epoch": 0.01, "learning_rate": 4.9980617050020704e-05, "loss": 1.162, "step": 12634 }, { "epoch": 0.01, "learning_rate": 4.9980613957437704e-05, "loss": 1.2691, "step": 12635 }, { "epoch": 0.01, "learning_rate": 4.998061086460811e-05, "loss": 1.2555, "step": 12636 }, { "epoch": 0.01, "learning_rate": 4.998060777153192e-05, "loss": 0.7833, "step": 12637 }, { "epoch": 0.01, "learning_rate": 4.998060467820913e-05, "loss": 0.6888, "step": 12638 }, { "epoch": 0.01, "learning_rate": 4.9980601584639735e-05, "loss": 1.0737, "step": 12639 }, { "epoch": 0.01, "learning_rate": 4.9980598490823746e-05, "loss": 1.1638, "step": 12640 }, { "epoch": 0.01, "learning_rate": 4.998059539676116e-05, "loss": 1.257, "step": 12641 }, { "epoch": 0.01, "learning_rate": 4.998059230245198e-05, "loss": 0.9882, "step": 12642 }, { "epoch": 0.01, "learning_rate": 4.99805892078962e-05, "loss": 1.2478, "step": 12643 }, { "epoch": 0.01, "learning_rate": 4.998058611309382e-05, "loss": 1.2171, "step": 12644 }, { "epoch": 0.01, "learning_rate": 4.998058301804484e-05, "loss": 0.894, "step": 12645 }, { "epoch": 0.01, "learning_rate": 4.9980579922749266e-05, "loss": 1.1506, "step": 12646 }, { "epoch": 0.01, "learning_rate": 4.998057682720709e-05, "loss": 1.0116, "step": 12647 }, { "epoch": 0.01, "learning_rate": 4.9980573731418324e-05, "loss": 1.2914, "step": 12648 }, { "epoch": 0.01, "learning_rate": 4.998057063538295e-05, "loss": 1.2192, "step": 12649 }, { "epoch": 0.01, "learning_rate": 4.998056753910099e-05, "loss": 1.235, "step": 12650 }, { "epoch": 0.01, "learning_rate": 4.998056444257242e-05, "loss": 1.0878, "step": 12651 }, { "epoch": 0.01, "learning_rate": 4.9980561345797264e-05, "loss": 1.2564, "step": 12652 }, { "epoch": 0.01, "learning_rate": 4.99805582487755e-05, "loss": 1.0418, "step": 12653 }, { "epoch": 0.01, "learning_rate": 4.998055515150715e-05, "loss": 1.008, "step": 12654 }, { "epoch": 0.01, "learning_rate": 4.998055205399219e-05, "loss": 1.3084, "step": 12655 }, { "epoch": 0.01, "learning_rate": 4.9980548956230644e-05, "loss": 1.2171, "step": 12656 }, { "epoch": 0.01, "learning_rate": 4.998054585822249e-05, "loss": 0.9538, "step": 12657 }, { "epoch": 0.01, "learning_rate": 4.998054275996774e-05, "loss": 1.1178, "step": 12658 }, { "epoch": 0.01, "learning_rate": 4.99805396614664e-05, "loss": 1.0045, "step": 12659 }, { "epoch": 0.01, "learning_rate": 4.998053656271846e-05, "loss": 1.2685, "step": 12660 }, { "epoch": 0.01, "learning_rate": 4.998053346372392e-05, "loss": 1.0377, "step": 12661 }, { "epoch": 0.01, "learning_rate": 4.9980530364482786e-05, "loss": 0.9865, "step": 12662 }, { "epoch": 0.01, "learning_rate": 4.998052726499505e-05, "loss": 1.273, "step": 12663 }, { "epoch": 0.01, "learning_rate": 4.998052416526072e-05, "loss": 0.9645, "step": 12664 }, { "epoch": 0.01, "learning_rate": 4.99805210652798e-05, "loss": 1.2464, "step": 12665 }, { "epoch": 0.01, "learning_rate": 4.998051796505227e-05, "loss": 1.0677, "step": 12666 }, { "epoch": 0.01, "learning_rate": 4.998051486457815e-05, "loss": 0.9309, "step": 12667 }, { "epoch": 0.01, "learning_rate": 4.998051176385743e-05, "loss": 1.0929, "step": 12668 }, { "epoch": 0.01, "learning_rate": 4.9980508662890114e-05, "loss": 0.8478, "step": 12669 }, { "epoch": 0.01, "learning_rate": 4.998050556167621e-05, "loss": 0.6965, "step": 12670 }, { "epoch": 0.01, "learning_rate": 4.998050246021569e-05, "loss": 0.7587, "step": 12671 }, { "epoch": 0.01, "learning_rate": 4.998049935850858e-05, "loss": 0.8848, "step": 12672 }, { "epoch": 0.01, "learning_rate": 4.9980496256554885e-05, "loss": 0.9169, "step": 12673 }, { "epoch": 0.01, "learning_rate": 4.998049315435458e-05, "loss": 1.0632, "step": 12674 }, { "epoch": 0.01, "learning_rate": 4.998049005190768e-05, "loss": 1.4337, "step": 12675 }, { "epoch": 0.01, "learning_rate": 4.9980486949214186e-05, "loss": 0.9088, "step": 12676 }, { "epoch": 0.01, "learning_rate": 4.9980483846274094e-05, "loss": 1.1794, "step": 12677 }, { "epoch": 0.01, "learning_rate": 4.998048074308741e-05, "loss": 1.1089, "step": 12678 }, { "epoch": 0.01, "learning_rate": 4.998047763965412e-05, "loss": 1.0994, "step": 12679 }, { "epoch": 0.01, "learning_rate": 4.998047453597424e-05, "loss": 1.035, "step": 12680 }, { "epoch": 0.01, "learning_rate": 4.9980471432047757e-05, "loss": 0.5826, "step": 12681 }, { "epoch": 0.01, "learning_rate": 4.998046832787468e-05, "loss": 0.378, "step": 12682 }, { "epoch": 0.01, "learning_rate": 4.9980465223455014e-05, "loss": 0.582, "step": 12683 }, { "epoch": 0.01, "learning_rate": 4.9980462118788744e-05, "loss": 0.6371, "step": 12684 }, { "epoch": 0.01, "learning_rate": 4.998045901387588e-05, "loss": 1.3622, "step": 12685 }, { "epoch": 0.01, "learning_rate": 4.998045590871642e-05, "loss": 0.9852, "step": 12686 }, { "epoch": 0.01, "learning_rate": 4.998045280331036e-05, "loss": 1.2683, "step": 12687 }, { "epoch": 0.01, "learning_rate": 4.9980449697657705e-05, "loss": 1.1576, "step": 12688 }, { "epoch": 0.01, "learning_rate": 4.998044659175845e-05, "loss": 1.333, "step": 12689 }, { "epoch": 0.01, "learning_rate": 4.9980443485612606e-05, "loss": 1.3333, "step": 12690 }, { "epoch": 0.01, "learning_rate": 4.9980440379220163e-05, "loss": 1.0612, "step": 12691 }, { "epoch": 0.01, "learning_rate": 4.9980437272581126e-05, "loss": 1.1116, "step": 12692 }, { "epoch": 0.01, "learning_rate": 4.998043416569549e-05, "loss": 1.3357, "step": 12693 }, { "epoch": 0.01, "learning_rate": 4.9980431058563246e-05, "loss": 0.7662, "step": 12694 }, { "epoch": 0.01, "learning_rate": 4.9980427951184424e-05, "loss": 0.2322, "step": 12695 }, { "epoch": 0.01, "learning_rate": 4.9980424843559e-05, "loss": 0.4027, "step": 12696 }, { "epoch": 0.01, "learning_rate": 4.998042173568698e-05, "loss": 0.4733, "step": 12697 }, { "epoch": 0.01, "learning_rate": 4.998041862756836e-05, "loss": 0.6448, "step": 12698 }, { "epoch": 0.01, "learning_rate": 4.9980415519203144e-05, "loss": 0.1055, "step": 12699 }, { "epoch": 0.01, "learning_rate": 4.998041241059133e-05, "loss": 0.7113, "step": 12700 }, { "epoch": 0.01, "learning_rate": 4.998040930173292e-05, "loss": 0.0967, "step": 12701 }, { "epoch": 0.01, "learning_rate": 4.9980406192627926e-05, "loss": 0.5632, "step": 12702 }, { "epoch": 0.01, "learning_rate": 4.998040308327633e-05, "loss": 0.9085, "step": 12703 }, { "epoch": 0.01, "learning_rate": 4.9980399973678126e-05, "loss": 1.2519, "step": 12704 }, { "epoch": 0.01, "learning_rate": 4.998039686383334e-05, "loss": 1.1727, "step": 12705 }, { "epoch": 0.01, "learning_rate": 4.998039375374195e-05, "loss": 1.2409, "step": 12706 }, { "epoch": 0.01, "learning_rate": 4.9980390643403976e-05, "loss": 1.1744, "step": 12707 }, { "epoch": 0.01, "learning_rate": 4.998038753281939e-05, "loss": 1.0515, "step": 12708 }, { "epoch": 0.01, "learning_rate": 4.998038442198822e-05, "loss": 0.8253, "step": 12709 }, { "epoch": 0.01, "learning_rate": 4.998038131091045e-05, "loss": 0.9936, "step": 12710 }, { "epoch": 0.01, "learning_rate": 4.998037819958608e-05, "loss": 1.1891, "step": 12711 }, { "epoch": 0.01, "learning_rate": 4.998037508801512e-05, "loss": 1.3771, "step": 12712 }, { "epoch": 0.01, "learning_rate": 4.998037197619756e-05, "loss": 1.0651, "step": 12713 }, { "epoch": 0.01, "learning_rate": 4.998036886413341e-05, "loss": 1.07, "step": 12714 }, { "epoch": 0.01, "learning_rate": 4.998036575182266e-05, "loss": 1.0145, "step": 12715 }, { "epoch": 0.01, "learning_rate": 4.9980362639265316e-05, "loss": 1.0349, "step": 12716 }, { "epoch": 0.01, "learning_rate": 4.9980359526461376e-05, "loss": 1.2096, "step": 12717 }, { "epoch": 0.01, "learning_rate": 4.998035641341084e-05, "loss": 0.9221, "step": 12718 }, { "epoch": 0.01, "learning_rate": 4.998035330011371e-05, "loss": 1.2303, "step": 12719 }, { "epoch": 0.01, "learning_rate": 4.998035018656998e-05, "loss": 0.8573, "step": 12720 }, { "epoch": 0.01, "learning_rate": 4.998034707277966e-05, "loss": 1.3344, "step": 12721 }, { "epoch": 0.01, "learning_rate": 4.998034395874274e-05, "loss": 0.8321, "step": 12722 }, { "epoch": 0.01, "learning_rate": 4.998034084445923e-05, "loss": 0.8328, "step": 12723 }, { "epoch": 0.01, "learning_rate": 4.998033772992912e-05, "loss": 0.8378, "step": 12724 }, { "epoch": 0.01, "learning_rate": 4.998033461515242e-05, "loss": 0.7233, "step": 12725 }, { "epoch": 0.01, "learning_rate": 4.998033150012912e-05, "loss": 0.5925, "step": 12726 }, { "epoch": 0.01, "learning_rate": 4.998032838485922e-05, "loss": 0.6885, "step": 12727 }, { "epoch": 0.01, "learning_rate": 4.998032526934273e-05, "loss": 0.9141, "step": 12728 }, { "epoch": 0.01, "learning_rate": 4.998032215357965e-05, "loss": 2.9479, "step": 12729 }, { "epoch": 0.01, "learning_rate": 4.998031903756997e-05, "loss": 1.1008, "step": 12730 }, { "epoch": 0.01, "learning_rate": 4.998031592131369e-05, "loss": 0.9494, "step": 12731 }, { "epoch": 0.01, "learning_rate": 4.998031280481083e-05, "loss": 1.0106, "step": 12732 }, { "epoch": 0.01, "learning_rate": 4.9980309688061354e-05, "loss": 1.2855, "step": 12733 }, { "epoch": 0.01, "learning_rate": 4.998030657106529e-05, "loss": 1.1777, "step": 12734 }, { "epoch": 0.01, "learning_rate": 4.998030345382264e-05, "loss": 1.3056, "step": 12735 }, { "epoch": 0.01, "learning_rate": 4.998030033633339e-05, "loss": 1.0951, "step": 12736 }, { "epoch": 0.01, "learning_rate": 4.9980297218597545e-05, "loss": 1.4119, "step": 12737 }, { "epoch": 0.01, "learning_rate": 4.9980294100615104e-05, "loss": 1.2088, "step": 12738 }, { "epoch": 0.01, "learning_rate": 4.998029098238607e-05, "loss": 1.381, "step": 12739 }, { "epoch": 0.01, "learning_rate": 4.9980287863910435e-05, "loss": 1.1912, "step": 12740 }, { "epoch": 0.01, "learning_rate": 4.998028474518821e-05, "loss": 0.9648, "step": 12741 }, { "epoch": 0.01, "learning_rate": 4.9980281626219394e-05, "loss": 1.0769, "step": 12742 }, { "epoch": 0.01, "learning_rate": 4.998027850700398e-05, "loss": 1.1291, "step": 12743 }, { "epoch": 0.01, "learning_rate": 4.9980275387541966e-05, "loss": 1.2115, "step": 12744 }, { "epoch": 0.01, "learning_rate": 4.998027226783336e-05, "loss": 1.1611, "step": 12745 }, { "epoch": 0.01, "learning_rate": 4.9980269147878165e-05, "loss": 1.1745, "step": 12746 }, { "epoch": 0.01, "learning_rate": 4.998026602767637e-05, "loss": 0.7889, "step": 12747 }, { "epoch": 0.01, "learning_rate": 4.998026290722798e-05, "loss": 0.973, "step": 12748 }, { "epoch": 0.01, "learning_rate": 4.9980259786533e-05, "loss": 0.8657, "step": 12749 }, { "epoch": 0.01, "learning_rate": 4.9980256665591415e-05, "loss": 1.1629, "step": 12750 }, { "epoch": 0.01, "learning_rate": 4.9980253544403245e-05, "loss": 0.835, "step": 12751 }, { "epoch": 0.01, "learning_rate": 4.998025042296848e-05, "loss": 1.2045, "step": 12752 }, { "epoch": 0.01, "learning_rate": 4.998024730128712e-05, "loss": 0.9768, "step": 12753 }, { "epoch": 0.01, "learning_rate": 4.998024417935916e-05, "loss": 0.8842, "step": 12754 }, { "epoch": 0.01, "learning_rate": 4.998024105718461e-05, "loss": 1.3598, "step": 12755 }, { "epoch": 0.01, "learning_rate": 4.9980237934763464e-05, "loss": 1.1881, "step": 12756 }, { "epoch": 0.01, "learning_rate": 4.9980234812095724e-05, "loss": 1.1587, "step": 12757 }, { "epoch": 0.01, "learning_rate": 4.998023168918139e-05, "loss": 1.0493, "step": 12758 }, { "epoch": 0.01, "learning_rate": 4.9980228566020466e-05, "loss": 0.9941, "step": 12759 }, { "epoch": 0.01, "learning_rate": 4.998022544261294e-05, "loss": 1.0987, "step": 12760 }, { "epoch": 0.01, "learning_rate": 4.998022231895883e-05, "loss": 0.8508, "step": 12761 }, { "epoch": 0.01, "learning_rate": 4.998021919505811e-05, "loss": 1.0839, "step": 12762 }, { "epoch": 0.01, "learning_rate": 4.998021607091081e-05, "loss": 1.209, "step": 12763 }, { "epoch": 0.01, "learning_rate": 4.998021294651691e-05, "loss": 1.2368, "step": 12764 }, { "epoch": 0.01, "learning_rate": 4.998020982187642e-05, "loss": 0.9204, "step": 12765 }, { "epoch": 0.01, "learning_rate": 4.998020669698933e-05, "loss": 1.0901, "step": 12766 }, { "epoch": 0.01, "learning_rate": 4.998020357185564e-05, "loss": 1.3258, "step": 12767 }, { "epoch": 0.01, "learning_rate": 4.9980200446475376e-05, "loss": 1.1116, "step": 12768 }, { "epoch": 0.01, "learning_rate": 4.9980197320848496e-05, "loss": 1.1504, "step": 12769 }, { "epoch": 0.01, "learning_rate": 4.998019419497504e-05, "loss": 0.928, "step": 12770 }, { "epoch": 0.01, "learning_rate": 4.998019106885498e-05, "loss": 1.0945, "step": 12771 }, { "epoch": 0.01, "learning_rate": 4.998018794248833e-05, "loss": 1.0755, "step": 12772 }, { "epoch": 0.01, "learning_rate": 4.998018481587509e-05, "loss": 0.9464, "step": 12773 }, { "epoch": 0.01, "learning_rate": 4.9980181689015245e-05, "loss": 0.9836, "step": 12774 }, { "epoch": 0.01, "learning_rate": 4.9980178561908816e-05, "loss": 0.7677, "step": 12775 }, { "epoch": 0.01, "learning_rate": 4.9980175434555784e-05, "loss": 0.9265, "step": 12776 }, { "epoch": 0.01, "learning_rate": 4.9980172306956164e-05, "loss": 0.9324, "step": 12777 }, { "epoch": 0.01, "learning_rate": 4.998016917910995e-05, "loss": 1.0801, "step": 12778 }, { "epoch": 0.01, "learning_rate": 4.998016605101714e-05, "loss": 1.2038, "step": 12779 }, { "epoch": 0.01, "learning_rate": 4.998016292267774e-05, "loss": 0.5574, "step": 12780 }, { "epoch": 0.01, "learning_rate": 4.998015979409175e-05, "loss": 1.1233, "step": 12781 }, { "epoch": 0.01, "learning_rate": 4.998015666525916e-05, "loss": 0.9915, "step": 12782 }, { "epoch": 0.01, "learning_rate": 4.998015353617997e-05, "loss": 1.1283, "step": 12783 }, { "epoch": 0.01, "learning_rate": 4.99801504068542e-05, "loss": 1.2538, "step": 12784 }, { "epoch": 0.01, "learning_rate": 4.998014727728183e-05, "loss": 1.0479, "step": 12785 }, { "epoch": 0.01, "learning_rate": 4.9980144147462866e-05, "loss": 1.1323, "step": 12786 }, { "epoch": 0.01, "learning_rate": 4.998014101739731e-05, "loss": 1.2417, "step": 12787 }, { "epoch": 0.01, "learning_rate": 4.998013788708516e-05, "loss": 1.0424, "step": 12788 }, { "epoch": 0.01, "learning_rate": 4.998013475652642e-05, "loss": 1.068, "step": 12789 }, { "epoch": 0.01, "learning_rate": 4.9980131625721084e-05, "loss": 1.3921, "step": 12790 }, { "epoch": 0.01, "learning_rate": 4.9980128494669155e-05, "loss": 1.0747, "step": 12791 }, { "epoch": 0.01, "learning_rate": 4.998012536337063e-05, "loss": 1.2593, "step": 12792 }, { "epoch": 0.01, "learning_rate": 4.998012223182551e-05, "loss": 1.1864, "step": 12793 }, { "epoch": 0.01, "learning_rate": 4.99801191000338e-05, "loss": 1.0663, "step": 12794 }, { "epoch": 0.01, "learning_rate": 4.99801159679955e-05, "loss": 1.13, "step": 12795 }, { "epoch": 0.01, "learning_rate": 4.99801128357106e-05, "loss": 1.0854, "step": 12796 }, { "epoch": 0.01, "learning_rate": 4.998010970317911e-05, "loss": 1.0491, "step": 12797 }, { "epoch": 0.01, "learning_rate": 4.998010657040103e-05, "loss": 1.0693, "step": 12798 }, { "epoch": 0.01, "learning_rate": 4.998010343737636e-05, "loss": 1.0836, "step": 12799 }, { "epoch": 0.01, "learning_rate": 4.998010030410508e-05, "loss": 1.1157, "step": 12800 }, { "epoch": 0.01, "learning_rate": 4.998009717058722e-05, "loss": 1.3284, "step": 12801 }, { "epoch": 0.01, "learning_rate": 4.998009403682277e-05, "loss": 0.8892, "step": 12802 }, { "epoch": 0.01, "learning_rate": 4.998009090281172e-05, "loss": 1.1005, "step": 12803 }, { "epoch": 0.01, "learning_rate": 4.998008776855408e-05, "loss": 0.7738, "step": 12804 }, { "epoch": 0.01, "learning_rate": 4.998008463404985e-05, "loss": 0.8511, "step": 12805 }, { "epoch": 0.01, "learning_rate": 4.998008149929903e-05, "loss": 0.9628, "step": 12806 }, { "epoch": 0.01, "learning_rate": 4.9980078364301606e-05, "loss": 1.1077, "step": 12807 }, { "epoch": 0.01, "learning_rate": 4.998007522905759e-05, "loss": 0.8475, "step": 12808 }, { "epoch": 0.01, "learning_rate": 4.998007209356699e-05, "loss": 0.328, "step": 12809 }, { "epoch": 0.01, "learning_rate": 4.998006895782978e-05, "loss": 0.5646, "step": 12810 }, { "epoch": 0.01, "learning_rate": 4.9980065821846e-05, "loss": 1.0582, "step": 12811 }, { "epoch": 0.01, "learning_rate": 4.9980062685615615e-05, "loss": 1.2618, "step": 12812 }, { "epoch": 0.01, "learning_rate": 4.9980059549138636e-05, "loss": 1.1225, "step": 12813 }, { "epoch": 0.01, "learning_rate": 4.998005641241507e-05, "loss": 0.8237, "step": 12814 }, { "epoch": 0.01, "learning_rate": 4.998005327544491e-05, "loss": 0.933, "step": 12815 }, { "epoch": 0.01, "learning_rate": 4.9980050138228156e-05, "loss": 0.808, "step": 12816 }, { "epoch": 0.01, "learning_rate": 4.998004700076481e-05, "loss": 0.7977, "step": 12817 }, { "epoch": 0.01, "learning_rate": 4.998004386305487e-05, "loss": 1.1029, "step": 12818 }, { "epoch": 0.01, "learning_rate": 4.998004072509834e-05, "loss": 1.0664, "step": 12819 }, { "epoch": 0.01, "learning_rate": 4.9980037586895225e-05, "loss": 0.919, "step": 12820 }, { "epoch": 0.01, "learning_rate": 4.99800344484455e-05, "loss": 0.8339, "step": 12821 }, { "epoch": 0.01, "learning_rate": 4.99800313097492e-05, "loss": 0.8931, "step": 12822 }, { "epoch": 0.01, "learning_rate": 4.998002817080629e-05, "loss": 1.1265, "step": 12823 }, { "epoch": 0.01, "learning_rate": 4.998002503161681e-05, "loss": 1.0405, "step": 12824 }, { "epoch": 0.01, "learning_rate": 4.9980021892180715e-05, "loss": 1.2704, "step": 12825 }, { "epoch": 0.01, "learning_rate": 4.9980018752498035e-05, "loss": 0.983, "step": 12826 }, { "epoch": 0.01, "learning_rate": 4.998001561256877e-05, "loss": 1.1701, "step": 12827 }, { "epoch": 0.01, "learning_rate": 4.998001247239291e-05, "loss": 0.9842, "step": 12828 }, { "epoch": 0.01, "learning_rate": 4.998000933197046e-05, "loss": 0.9698, "step": 12829 }, { "epoch": 0.01, "learning_rate": 4.998000619130141e-05, "loss": 0.813, "step": 12830 }, { "epoch": 0.01, "learning_rate": 4.998000305038577e-05, "loss": 0.8723, "step": 12831 }, { "epoch": 0.01, "learning_rate": 4.997999990922354e-05, "loss": 1.0961, "step": 12832 }, { "epoch": 0.01, "learning_rate": 4.997999676781472e-05, "loss": 1.2978, "step": 12833 }, { "epoch": 0.01, "learning_rate": 4.997999362615931e-05, "loss": 1.3828, "step": 12834 }, { "epoch": 0.01, "learning_rate": 4.99799904842573e-05, "loss": 1.0502, "step": 12835 }, { "epoch": 0.01, "learning_rate": 4.99799873421087e-05, "loss": 1.2325, "step": 12836 }, { "epoch": 0.01, "learning_rate": 4.997998419971351e-05, "loss": 1.2105, "step": 12837 }, { "epoch": 0.01, "learning_rate": 4.9979981057071725e-05, "loss": 1.4, "step": 12838 }, { "epoch": 0.01, "learning_rate": 4.997997791418336e-05, "loss": 0.9596, "step": 12839 }, { "epoch": 0.01, "learning_rate": 4.997997477104839e-05, "loss": 0.9279, "step": 12840 }, { "epoch": 0.01, "learning_rate": 4.997997162766683e-05, "loss": 1.3523, "step": 12841 }, { "epoch": 0.01, "learning_rate": 4.9979968484038684e-05, "loss": 1.2956, "step": 12842 }, { "epoch": 0.01, "learning_rate": 4.997996534016395e-05, "loss": 1.2156, "step": 12843 }, { "epoch": 0.01, "learning_rate": 4.997996219604261e-05, "loss": 1.2605, "step": 12844 }, { "epoch": 0.01, "learning_rate": 4.997995905167468e-05, "loss": 1.11, "step": 12845 }, { "epoch": 0.01, "learning_rate": 4.997995590706017e-05, "loss": 1.1698, "step": 12846 }, { "epoch": 0.01, "learning_rate": 4.997995276219907e-05, "loss": 1.0783, "step": 12847 }, { "epoch": 0.01, "learning_rate": 4.9979949617091366e-05, "loss": 1.4209, "step": 12848 }, { "epoch": 0.01, "learning_rate": 4.9979946471737075e-05, "loss": 1.0297, "step": 12849 }, { "epoch": 0.01, "learning_rate": 4.997994332613619e-05, "loss": 1.1634, "step": 12850 }, { "epoch": 0.01, "learning_rate": 4.997994018028872e-05, "loss": 1.3217, "step": 12851 }, { "epoch": 0.01, "learning_rate": 4.997993703419466e-05, "loss": 0.9882, "step": 12852 }, { "epoch": 0.01, "learning_rate": 4.9979933887854e-05, "loss": 1.0151, "step": 12853 }, { "epoch": 0.01, "learning_rate": 4.997993074126675e-05, "loss": 1.0105, "step": 12854 }, { "epoch": 0.01, "learning_rate": 4.9979927594432916e-05, "loss": 1.0664, "step": 12855 }, { "epoch": 0.01, "learning_rate": 4.997992444735248e-05, "loss": 1.1174, "step": 12856 }, { "epoch": 0.01, "learning_rate": 4.997992130002546e-05, "loss": 1.2691, "step": 12857 }, { "epoch": 0.01, "learning_rate": 4.997991815245185e-05, "loss": 1.1248, "step": 12858 }, { "epoch": 0.01, "learning_rate": 4.9979915004631644e-05, "loss": 1.1102, "step": 12859 }, { "epoch": 0.01, "learning_rate": 4.997991185656485e-05, "loss": 1.1124, "step": 12860 }, { "epoch": 0.01, "learning_rate": 4.9979908708251465e-05, "loss": 1.1699, "step": 12861 }, { "epoch": 0.01, "learning_rate": 4.9979905559691487e-05, "loss": 1.2475, "step": 12862 }, { "epoch": 0.01, "learning_rate": 4.997990241088491e-05, "loss": 1.0424, "step": 12863 }, { "epoch": 0.01, "learning_rate": 4.997989926183175e-05, "loss": 1.1789, "step": 12864 }, { "epoch": 0.01, "learning_rate": 4.997989611253201e-05, "loss": 0.8412, "step": 12865 }, { "epoch": 0.01, "learning_rate": 4.997989296298566e-05, "loss": 0.9935, "step": 12866 }, { "epoch": 0.01, "learning_rate": 4.997988981319273e-05, "loss": 1.3858, "step": 12867 }, { "epoch": 0.01, "learning_rate": 4.997988666315321e-05, "loss": 1.2902, "step": 12868 }, { "epoch": 0.01, "learning_rate": 4.997988351286709e-05, "loss": 1.7678, "step": 12869 }, { "epoch": 0.01, "learning_rate": 4.997988036233439e-05, "loss": 1.1934, "step": 12870 }, { "epoch": 0.01, "learning_rate": 4.9979877211555094e-05, "loss": 1.1657, "step": 12871 }, { "epoch": 0.01, "learning_rate": 4.9979874060529206e-05, "loss": 0.9137, "step": 12872 }, { "epoch": 0.01, "learning_rate": 4.997987090925672e-05, "loss": 1.0796, "step": 12873 }, { "epoch": 0.01, "learning_rate": 4.997986775773766e-05, "loss": 1.1675, "step": 12874 }, { "epoch": 0.01, "learning_rate": 4.9979864605972e-05, "loss": 1.1648, "step": 12875 }, { "epoch": 0.01, "learning_rate": 4.9979861453959745e-05, "loss": 1.8118, "step": 12876 }, { "epoch": 0.01, "learning_rate": 4.997985830170091e-05, "loss": 0.8699, "step": 12877 }, { "epoch": 0.01, "learning_rate": 4.997985514919548e-05, "loss": 1.2232, "step": 12878 }, { "epoch": 0.01, "learning_rate": 4.997985199644345e-05, "loss": 1.1903, "step": 12879 }, { "epoch": 0.01, "learning_rate": 4.9979848843444845e-05, "loss": 1.1403, "step": 12880 }, { "epoch": 0.01, "learning_rate": 4.997984569019964e-05, "loss": 1.2959, "step": 12881 }, { "epoch": 0.01, "learning_rate": 4.9979842536707845e-05, "loss": 1.0923, "step": 12882 }, { "epoch": 0.01, "learning_rate": 4.997983938296946e-05, "loss": 0.8126, "step": 12883 }, { "epoch": 0.01, "learning_rate": 4.997983622898449e-05, "loss": 1.1587, "step": 12884 }, { "epoch": 0.01, "learning_rate": 4.997983307475292e-05, "loss": 0.941, "step": 12885 }, { "epoch": 0.01, "learning_rate": 4.9979829920274766e-05, "loss": 1.2729, "step": 12886 }, { "epoch": 0.01, "learning_rate": 4.997982676555002e-05, "loss": 1.2418, "step": 12887 }, { "epoch": 0.01, "learning_rate": 4.997982361057868e-05, "loss": 1.1064, "step": 12888 }, { "epoch": 0.01, "learning_rate": 4.997982045536076e-05, "loss": 1.249, "step": 12889 }, { "epoch": 0.01, "learning_rate": 4.997981729989624e-05, "loss": 1.2328, "step": 12890 }, { "epoch": 0.01, "learning_rate": 4.997981414418513e-05, "loss": 1.0787, "step": 12891 }, { "epoch": 0.01, "learning_rate": 4.997981098822744e-05, "loss": 1.0769, "step": 12892 }, { "epoch": 0.01, "learning_rate": 4.997980783202315e-05, "loss": 0.9737, "step": 12893 }, { "epoch": 0.01, "learning_rate": 4.997980467557227e-05, "loss": 0.965, "step": 12894 }, { "epoch": 0.01, "learning_rate": 4.99798015188748e-05, "loss": 0.9079, "step": 12895 }, { "epoch": 0.01, "learning_rate": 4.997979836193075e-05, "loss": 0.7914, "step": 12896 }, { "epoch": 0.01, "learning_rate": 4.99797952047401e-05, "loss": 0.9449, "step": 12897 }, { "epoch": 0.01, "learning_rate": 4.9979792047302863e-05, "loss": 0.9801, "step": 12898 }, { "epoch": 0.01, "learning_rate": 4.997978888961904e-05, "loss": 1.3363, "step": 12899 }, { "epoch": 0.01, "learning_rate": 4.997978573168862e-05, "loss": 1.0618, "step": 12900 }, { "epoch": 0.01, "learning_rate": 4.997978257351161e-05, "loss": 0.2838, "step": 12901 }, { "epoch": 0.01, "learning_rate": 4.9979779415088016e-05, "loss": 0.3652, "step": 12902 }, { "epoch": 0.01, "learning_rate": 4.9979776256417825e-05, "loss": 0.9881, "step": 12903 }, { "epoch": 0.01, "learning_rate": 4.9979773097501046e-05, "loss": 1.7913, "step": 12904 }, { "epoch": 0.01, "learning_rate": 4.9979769938337685e-05, "loss": 1.7617, "step": 12905 }, { "epoch": 0.01, "learning_rate": 4.997976677892773e-05, "loss": 1.3546, "step": 12906 }, { "epoch": 0.01, "learning_rate": 4.997976361927118e-05, "loss": 1.2825, "step": 12907 }, { "epoch": 0.01, "learning_rate": 4.997976045936804e-05, "loss": 1.0407, "step": 12908 }, { "epoch": 0.01, "learning_rate": 4.997975729921831e-05, "loss": 0.9894, "step": 12909 }, { "epoch": 0.01, "learning_rate": 4.9979754138822e-05, "loss": 1.2142, "step": 12910 }, { "epoch": 0.01, "learning_rate": 4.9979750978179094e-05, "loss": 1.1767, "step": 12911 }, { "epoch": 0.01, "learning_rate": 4.99797478172896e-05, "loss": 0.8767, "step": 12912 }, { "epoch": 0.01, "learning_rate": 4.9979744656153516e-05, "loss": 1.2397, "step": 12913 }, { "epoch": 0.01, "learning_rate": 4.997974149477085e-05, "loss": 1.2256, "step": 12914 }, { "epoch": 0.01, "learning_rate": 4.997973833314158e-05, "loss": 0.8299, "step": 12915 }, { "epoch": 0.01, "learning_rate": 4.9979735171265726e-05, "loss": 1.1286, "step": 12916 }, { "epoch": 0.01, "learning_rate": 4.9979732009143286e-05, "loss": 0.9414, "step": 12917 }, { "epoch": 0.01, "learning_rate": 4.997972884677425e-05, "loss": 1.035, "step": 12918 }, { "epoch": 0.01, "learning_rate": 4.9979725684158636e-05, "loss": 1.084, "step": 12919 }, { "epoch": 0.01, "learning_rate": 4.9979722521296424e-05, "loss": 0.8994, "step": 12920 }, { "epoch": 0.01, "learning_rate": 4.9979719358187625e-05, "loss": 1.2614, "step": 12921 }, { "epoch": 0.01, "learning_rate": 4.997971619483224e-05, "loss": 1.0149, "step": 12922 }, { "epoch": 0.01, "learning_rate": 4.9979713031230255e-05, "loss": 1.0084, "step": 12923 }, { "epoch": 0.01, "learning_rate": 4.99797098673817e-05, "loss": 1.2544, "step": 12924 }, { "epoch": 0.01, "learning_rate": 4.997970670328654e-05, "loss": 0.9306, "step": 12925 }, { "epoch": 0.01, "learning_rate": 4.997970353894479e-05, "loss": 1.2647, "step": 12926 }, { "epoch": 0.01, "learning_rate": 4.9979700374356455e-05, "loss": 1.2281, "step": 12927 }, { "epoch": 0.01, "learning_rate": 4.997969720952154e-05, "loss": 0.5042, "step": 12928 }, { "epoch": 0.01, "learning_rate": 4.9979694044440026e-05, "loss": 0.6964, "step": 12929 }, { "epoch": 0.01, "learning_rate": 4.997969087911192e-05, "loss": 1.3595, "step": 12930 }, { "epoch": 0.01, "learning_rate": 4.9979687713537224e-05, "loss": 1.0151, "step": 12931 }, { "epoch": 0.01, "learning_rate": 4.997968454771595e-05, "loss": 1.1689, "step": 12932 }, { "epoch": 0.01, "learning_rate": 4.997968138164808e-05, "loss": 1.1825, "step": 12933 }, { "epoch": 0.01, "learning_rate": 4.997967821533362e-05, "loss": 0.9164, "step": 12934 }, { "epoch": 0.01, "learning_rate": 4.997967504877258e-05, "loss": 1.018, "step": 12935 }, { "epoch": 0.01, "learning_rate": 4.997967188196494e-05, "loss": 0.7397, "step": 12936 }, { "epoch": 0.01, "learning_rate": 4.997966871491072e-05, "loss": 1.758, "step": 12937 }, { "epoch": 0.01, "learning_rate": 4.9979665547609914e-05, "loss": 1.2953, "step": 12938 }, { "epoch": 0.01, "learning_rate": 4.9979662380062506e-05, "loss": 1.0949, "step": 12939 }, { "epoch": 0.01, "learning_rate": 4.9979659212268516e-05, "loss": 1.389, "step": 12940 }, { "epoch": 0.01, "learning_rate": 4.997965604422794e-05, "loss": 1.1682, "step": 12941 }, { "epoch": 0.01, "learning_rate": 4.997965287594077e-05, "loss": 1.4026, "step": 12942 }, { "epoch": 0.01, "learning_rate": 4.997964970740702e-05, "loss": 1.5013, "step": 12943 }, { "epoch": 0.01, "learning_rate": 4.997964653862667e-05, "loss": 1.0865, "step": 12944 }, { "epoch": 0.01, "learning_rate": 4.997964336959974e-05, "loss": 1.0767, "step": 12945 }, { "epoch": 0.01, "learning_rate": 4.9979640200326214e-05, "loss": 1.4635, "step": 12946 }, { "epoch": 0.01, "learning_rate": 4.997963703080611e-05, "loss": 1.2607, "step": 12947 }, { "epoch": 0.01, "learning_rate": 4.997963386103941e-05, "loss": 1.0479, "step": 12948 }, { "epoch": 0.01, "learning_rate": 4.997963069102613e-05, "loss": 1.2824, "step": 12949 }, { "epoch": 0.01, "learning_rate": 4.997962752076625e-05, "loss": 1.3158, "step": 12950 }, { "epoch": 0.01, "learning_rate": 4.9979624350259776e-05, "loss": 1.0888, "step": 12951 }, { "epoch": 0.01, "learning_rate": 4.997962117950673e-05, "loss": 0.8503, "step": 12952 }, { "epoch": 0.01, "learning_rate": 4.997961800850709e-05, "loss": 0.7097, "step": 12953 }, { "epoch": 0.01, "learning_rate": 4.997961483726086e-05, "loss": 0.8083, "step": 12954 }, { "epoch": 0.01, "learning_rate": 4.997961166576805e-05, "loss": 1.0741, "step": 12955 }, { "epoch": 0.01, "learning_rate": 4.997960849402864e-05, "loss": 1.1947, "step": 12956 }, { "epoch": 0.01, "learning_rate": 4.9979605322042644e-05, "loss": 0.9949, "step": 12957 }, { "epoch": 0.01, "learning_rate": 4.9979602149810066e-05, "loss": 1.1383, "step": 12958 }, { "epoch": 0.01, "learning_rate": 4.99795989773309e-05, "loss": 0.9662, "step": 12959 }, { "epoch": 0.01, "learning_rate": 4.997959580460514e-05, "loss": 0.824, "step": 12960 }, { "epoch": 0.01, "learning_rate": 4.997959263163279e-05, "loss": 1.1551, "step": 12961 }, { "epoch": 0.01, "learning_rate": 4.997958945841386e-05, "loss": 1.2205, "step": 12962 }, { "epoch": 0.01, "learning_rate": 4.997958628494834e-05, "loss": 0.4988, "step": 12963 }, { "epoch": 0.01, "learning_rate": 4.997958311123623e-05, "loss": 0.2457, "step": 12964 }, { "epoch": 0.01, "learning_rate": 4.997957993727753e-05, "loss": 0.2279, "step": 12965 }, { "epoch": 0.01, "learning_rate": 4.997957676307224e-05, "loss": 0.1507, "step": 12966 }, { "epoch": 0.01, "learning_rate": 4.997957358862038e-05, "loss": 0.39, "step": 12967 }, { "epoch": 0.01, "learning_rate": 4.997957041392192e-05, "loss": 0.773, "step": 12968 }, { "epoch": 0.01, "learning_rate": 4.997956723897686e-05, "loss": 0.7261, "step": 12969 }, { "epoch": 0.01, "learning_rate": 4.9979564063785226e-05, "loss": 0.7207, "step": 12970 }, { "epoch": 0.01, "learning_rate": 4.9979560888347e-05, "loss": 0.6963, "step": 12971 }, { "epoch": 0.01, "learning_rate": 4.9979557712662196e-05, "loss": 0.5992, "step": 12972 }, { "epoch": 0.01, "learning_rate": 4.9979554536730795e-05, "loss": 0.7173, "step": 12973 }, { "epoch": 0.01, "learning_rate": 4.9979551360552806e-05, "loss": 0.7709, "step": 12974 }, { "epoch": 0.01, "learning_rate": 4.9979548184128235e-05, "loss": 0.6878, "step": 12975 }, { "epoch": 0.01, "learning_rate": 4.997954500745707e-05, "loss": 0.5592, "step": 12976 }, { "epoch": 0.01, "learning_rate": 4.997954183053932e-05, "loss": 0.645, "step": 12977 }, { "epoch": 0.01, "learning_rate": 4.997953865337498e-05, "loss": 0.5575, "step": 12978 }, { "epoch": 0.01, "learning_rate": 4.9979535475964056e-05, "loss": 0.5624, "step": 12979 }, { "epoch": 0.01, "learning_rate": 4.9979532298306544e-05, "loss": 0.562, "step": 12980 }, { "epoch": 0.01, "learning_rate": 4.9979529120402444e-05, "loss": 0.5335, "step": 12981 }, { "epoch": 0.01, "learning_rate": 4.9979525942251756e-05, "loss": 0.4988, "step": 12982 }, { "epoch": 0.01, "learning_rate": 4.997952276385448e-05, "loss": 0.502, "step": 12983 }, { "epoch": 0.01, "learning_rate": 4.997951958521062e-05, "loss": 0.4721, "step": 12984 }, { "epoch": 0.01, "learning_rate": 4.997951640632017e-05, "loss": 0.4373, "step": 12985 }, { "epoch": 0.01, "learning_rate": 4.997951322718314e-05, "loss": 0.4362, "step": 12986 }, { "epoch": 0.01, "learning_rate": 4.997951004779952e-05, "loss": 0.4396, "step": 12987 }, { "epoch": 0.01, "learning_rate": 4.99795068681693e-05, "loss": 0.4634, "step": 12988 }, { "epoch": 0.01, "learning_rate": 4.997950368829251e-05, "loss": 0.5829, "step": 12989 }, { "epoch": 0.01, "learning_rate": 4.997950050816912e-05, "loss": 0.5246, "step": 12990 }, { "epoch": 0.01, "learning_rate": 4.9979497327799145e-05, "loss": 0.4869, "step": 12991 }, { "epoch": 0.01, "learning_rate": 4.997949414718259e-05, "loss": 0.498, "step": 12992 }, { "epoch": 0.01, "learning_rate": 4.997949096631944e-05, "loss": 0.4482, "step": 12993 }, { "epoch": 0.01, "learning_rate": 4.997948778520971e-05, "loss": 0.4193, "step": 12994 }, { "epoch": 0.01, "learning_rate": 4.997948460385339e-05, "loss": 0.4703, "step": 12995 }, { "epoch": 0.01, "learning_rate": 4.997948142225048e-05, "loss": 0.5105, "step": 12996 }, { "epoch": 0.01, "learning_rate": 4.997947824040099e-05, "loss": 0.4898, "step": 12997 }, { "epoch": 0.01, "learning_rate": 4.9979475058304906e-05, "loss": 0.4599, "step": 12998 }, { "epoch": 0.01, "learning_rate": 4.997947187596224e-05, "loss": 0.4422, "step": 12999 }, { "epoch": 0.01, "learning_rate": 4.997946869337298e-05, "loss": 0.4958, "step": 13000 }, { "epoch": 0.01, "eval_loss": 1.0994994640350342, "eval_runtime": 84.1793, "eval_samples_per_second": 16.453, "eval_steps_per_second": 4.122, "step": 13000 }, { "epoch": 0.01, "learning_rate": 4.997946551053714e-05, "loss": 0.5183, "step": 13001 }, { "epoch": 0.01, "learning_rate": 4.997946232745471e-05, "loss": 0.4278, "step": 13002 }, { "epoch": 0.01, "learning_rate": 4.99794591441257e-05, "loss": 0.4494, "step": 13003 }, { "epoch": 0.01, "learning_rate": 4.9979455960550096e-05, "loss": 0.4726, "step": 13004 }, { "epoch": 0.01, "learning_rate": 4.997945277672791e-05, "loss": 0.47, "step": 13005 }, { "epoch": 0.01, "learning_rate": 4.997944959265913e-05, "loss": 0.5201, "step": 13006 }, { "epoch": 0.01, "learning_rate": 4.997944640834377e-05, "loss": 0.5025, "step": 13007 }, { "epoch": 0.01, "learning_rate": 4.9979443223781824e-05, "loss": 0.525, "step": 13008 }, { "epoch": 0.01, "learning_rate": 4.997944003897329e-05, "loss": 0.4876, "step": 13009 }, { "epoch": 0.01, "learning_rate": 4.997943685391817e-05, "loss": 0.8739, "step": 13010 }, { "epoch": 0.01, "learning_rate": 4.997943366861646e-05, "loss": 1.0577, "step": 13011 }, { "epoch": 0.01, "learning_rate": 4.997943048306817e-05, "loss": 1.1702, "step": 13012 }, { "epoch": 0.01, "learning_rate": 4.9979427297273286e-05, "loss": 1.2193, "step": 13013 }, { "epoch": 0.01, "learning_rate": 4.997942411123182e-05, "loss": 1.2447, "step": 13014 }, { "epoch": 0.01, "learning_rate": 4.997942092494376e-05, "loss": 1.5422, "step": 13015 }, { "epoch": 0.01, "learning_rate": 4.997941773840913e-05, "loss": 1.0676, "step": 13016 }, { "epoch": 0.01, "learning_rate": 4.99794145516279e-05, "loss": 1.3999, "step": 13017 }, { "epoch": 0.01, "learning_rate": 4.9979411364600095e-05, "loss": 0.8781, "step": 13018 }, { "epoch": 0.01, "learning_rate": 4.997940817732569e-05, "loss": 1.171, "step": 13019 }, { "epoch": 0.01, "learning_rate": 4.997940498980471e-05, "loss": 1.1286, "step": 13020 }, { "epoch": 0.01, "learning_rate": 4.9979401802037135e-05, "loss": 1.2802, "step": 13021 }, { "epoch": 0.01, "learning_rate": 4.997939861402298e-05, "loss": 0.9592, "step": 13022 }, { "epoch": 0.01, "learning_rate": 4.997939542576223e-05, "loss": 1.0764, "step": 13023 }, { "epoch": 0.01, "learning_rate": 4.997939223725491e-05, "loss": 1.2927, "step": 13024 }, { "epoch": 0.01, "learning_rate": 4.997938904850099e-05, "loss": 1.04, "step": 13025 }, { "epoch": 0.01, "learning_rate": 4.9979385859500497e-05, "loss": 1.0126, "step": 13026 }, { "epoch": 0.01, "learning_rate": 4.9979382670253406e-05, "loss": 1.3136, "step": 13027 }, { "epoch": 0.01, "learning_rate": 4.997937948075973e-05, "loss": 1.3859, "step": 13028 }, { "epoch": 0.01, "learning_rate": 4.997937629101947e-05, "loss": 1.697, "step": 13029 }, { "epoch": 0.01, "learning_rate": 4.997937310103263e-05, "loss": 1.0814, "step": 13030 }, { "epoch": 0.01, "learning_rate": 4.99793699107992e-05, "loss": 0.9613, "step": 13031 }, { "epoch": 0.01, "learning_rate": 4.997936672031918e-05, "loss": 1.0867, "step": 13032 }, { "epoch": 0.01, "learning_rate": 4.997936352959258e-05, "loss": 1.217, "step": 13033 }, { "epoch": 0.01, "learning_rate": 4.997936033861939e-05, "loss": 1.0016, "step": 13034 }, { "epoch": 0.01, "learning_rate": 4.997935714739962e-05, "loss": 0.9677, "step": 13035 }, { "epoch": 0.01, "learning_rate": 4.997935395593326e-05, "loss": 1.2878, "step": 13036 }, { "epoch": 0.01, "learning_rate": 4.997935076422031e-05, "loss": 1.1637, "step": 13037 }, { "epoch": 0.01, "learning_rate": 4.9979347572260785e-05, "loss": 0.9675, "step": 13038 }, { "epoch": 0.01, "learning_rate": 4.997934438005467e-05, "loss": 1.9872, "step": 13039 }, { "epoch": 0.01, "learning_rate": 4.9979341187601965e-05, "loss": 0.9205, "step": 13040 }, { "epoch": 0.01, "learning_rate": 4.997933799490267e-05, "loss": 0.8962, "step": 13041 }, { "epoch": 0.01, "learning_rate": 4.99793348019568e-05, "loss": 1.0186, "step": 13042 }, { "epoch": 0.01, "learning_rate": 4.9979331608764344e-05, "loss": 1.146, "step": 13043 }, { "epoch": 0.01, "learning_rate": 4.99793284153253e-05, "loss": 1.4336, "step": 13044 }, { "epoch": 0.01, "learning_rate": 4.9979325221639675e-05, "loss": 1.1118, "step": 13045 }, { "epoch": 0.01, "learning_rate": 4.997932202770746e-05, "loss": 0.8323, "step": 13046 }, { "epoch": 0.01, "learning_rate": 4.997931883352866e-05, "loss": 2.7557, "step": 13047 }, { "epoch": 0.01, "learning_rate": 4.997931563910328e-05, "loss": 0.3497, "step": 13048 }, { "epoch": 0.01, "learning_rate": 4.99793124444313e-05, "loss": 1.1898, "step": 13049 }, { "epoch": 0.01, "learning_rate": 4.9979309249512754e-05, "loss": 1.3632, "step": 13050 }, { "epoch": 0.01, "learning_rate": 4.9979306054347607e-05, "loss": 1.1675, "step": 13051 }, { "epoch": 0.01, "learning_rate": 4.9979302858935884e-05, "loss": 1.0783, "step": 13052 }, { "epoch": 0.01, "learning_rate": 4.997929966327757e-05, "loss": 1.0931, "step": 13053 }, { "epoch": 0.01, "learning_rate": 4.997929646737267e-05, "loss": 1.1867, "step": 13054 }, { "epoch": 0.01, "learning_rate": 4.9979293271221195e-05, "loss": 1.4483, "step": 13055 }, { "epoch": 0.01, "learning_rate": 4.997929007482313e-05, "loss": 1.4049, "step": 13056 }, { "epoch": 0.01, "learning_rate": 4.997928687817848e-05, "loss": 1.5023, "step": 13057 }, { "epoch": 0.01, "learning_rate": 4.997928368128724e-05, "loss": 0.9887, "step": 13058 }, { "epoch": 0.01, "learning_rate": 4.997928048414941e-05, "loss": 1.3957, "step": 13059 }, { "epoch": 0.01, "learning_rate": 4.997927728676501e-05, "loss": 1.084, "step": 13060 }, { "epoch": 0.01, "learning_rate": 4.997927408913402e-05, "loss": 1.2527, "step": 13061 }, { "epoch": 0.01, "learning_rate": 4.9979270891256445e-05, "loss": 1.4548, "step": 13062 }, { "epoch": 0.01, "learning_rate": 4.997926769313228e-05, "loss": 1.1246, "step": 13063 }, { "epoch": 0.01, "learning_rate": 4.997926449476153e-05, "loss": 0.9704, "step": 13064 }, { "epoch": 0.01, "learning_rate": 4.997926129614421e-05, "loss": 1.164, "step": 13065 }, { "epoch": 0.01, "learning_rate": 4.9979258097280294e-05, "loss": 1.577, "step": 13066 }, { "epoch": 0.01, "learning_rate": 4.997925489816979e-05, "loss": 1.0228, "step": 13067 }, { "epoch": 0.01, "learning_rate": 4.997925169881271e-05, "loss": 0.8215, "step": 13068 }, { "epoch": 0.01, "learning_rate": 4.997924849920904e-05, "loss": 0.9098, "step": 13069 }, { "epoch": 0.01, "learning_rate": 4.9979245299358785e-05, "loss": 1.2712, "step": 13070 }, { "epoch": 0.01, "learning_rate": 4.997924209926195e-05, "loss": 1.2067, "step": 13071 }, { "epoch": 0.01, "learning_rate": 4.997923889891852e-05, "loss": 1.1048, "step": 13072 }, { "epoch": 0.01, "learning_rate": 4.997923569832852e-05, "loss": 1.1711, "step": 13073 }, { "epoch": 0.01, "learning_rate": 4.9979232497491925e-05, "loss": 1.35, "step": 13074 }, { "epoch": 0.01, "learning_rate": 4.997922929640875e-05, "loss": 0.7622, "step": 13075 }, { "epoch": 0.01, "learning_rate": 4.997922609507898e-05, "loss": 1.3351, "step": 13076 }, { "epoch": 0.01, "learning_rate": 4.997922289350264e-05, "loss": 0.9535, "step": 13077 }, { "epoch": 0.01, "learning_rate": 4.9979219691679715e-05, "loss": 0.9206, "step": 13078 }, { "epoch": 0.01, "learning_rate": 4.99792164896102e-05, "loss": 1.1329, "step": 13079 }, { "epoch": 0.01, "learning_rate": 4.99792132872941e-05, "loss": 0.8848, "step": 13080 }, { "epoch": 0.01, "learning_rate": 4.9979210084731414e-05, "loss": 0.7544, "step": 13081 }, { "epoch": 0.01, "learning_rate": 4.997920688192215e-05, "loss": 0.8494, "step": 13082 }, { "epoch": 0.01, "learning_rate": 4.99792036788663e-05, "loss": 1.131, "step": 13083 }, { "epoch": 0.01, "learning_rate": 4.997920047556387e-05, "loss": 0.7298, "step": 13084 }, { "epoch": 0.01, "learning_rate": 4.997919727201484e-05, "loss": 0.1745, "step": 13085 }, { "epoch": 0.01, "learning_rate": 4.9979194068219235e-05, "loss": 0.1724, "step": 13086 }, { "epoch": 0.01, "learning_rate": 4.9979190864177053e-05, "loss": 0.1669, "step": 13087 }, { "epoch": 0.01, "learning_rate": 4.9979187659888284e-05, "loss": 0.1473, "step": 13088 }, { "epoch": 0.01, "learning_rate": 4.9979184455352926e-05, "loss": 0.5057, "step": 13089 }, { "epoch": 0.01, "learning_rate": 4.9979181250570986e-05, "loss": 0.536, "step": 13090 }, { "epoch": 0.01, "learning_rate": 4.9979178045542465e-05, "loss": 0.6683, "step": 13091 }, { "epoch": 0.01, "learning_rate": 4.997917484026735e-05, "loss": 0.6792, "step": 13092 }, { "epoch": 0.01, "learning_rate": 4.9979171634745665e-05, "loss": 0.6064, "step": 13093 }, { "epoch": 0.01, "learning_rate": 4.9979168428977386e-05, "loss": 0.5759, "step": 13094 }, { "epoch": 0.01, "learning_rate": 4.997916522296253e-05, "loss": 0.5447, "step": 13095 }, { "epoch": 0.01, "learning_rate": 4.9979162016701084e-05, "loss": 0.4281, "step": 13096 }, { "epoch": 0.01, "learning_rate": 4.997915881019306e-05, "loss": 0.5687, "step": 13097 }, { "epoch": 0.01, "learning_rate": 4.997915560343845e-05, "loss": 0.5711, "step": 13098 }, { "epoch": 0.01, "learning_rate": 4.997915239643726e-05, "loss": 0.532, "step": 13099 }, { "epoch": 0.01, "learning_rate": 4.9979149189189476e-05, "loss": 0.4495, "step": 13100 }, { "epoch": 0.01, "learning_rate": 4.997914598169512e-05, "loss": 0.5383, "step": 13101 }, { "epoch": 0.01, "learning_rate": 4.997914277395417e-05, "loss": 0.6257, "step": 13102 }, { "epoch": 0.01, "learning_rate": 4.997913956596664e-05, "loss": 0.4699, "step": 13103 }, { "epoch": 0.01, "learning_rate": 4.997913635773253e-05, "loss": 0.4719, "step": 13104 }, { "epoch": 0.01, "learning_rate": 4.9979133149251834e-05, "loss": 0.4495, "step": 13105 }, { "epoch": 0.01, "learning_rate": 4.997912994052456e-05, "loss": 0.4682, "step": 13106 }, { "epoch": 0.01, "learning_rate": 4.997912673155069e-05, "loss": 0.403, "step": 13107 }, { "epoch": 0.01, "learning_rate": 4.997912352233025e-05, "loss": 0.3921, "step": 13108 }, { "epoch": 0.01, "learning_rate": 4.997912031286322e-05, "loss": 0.3934, "step": 13109 }, { "epoch": 0.01, "learning_rate": 4.997911710314961e-05, "loss": 0.4463, "step": 13110 }, { "epoch": 0.01, "learning_rate": 4.997911389318941e-05, "loss": 0.5515, "step": 13111 }, { "epoch": 0.01, "learning_rate": 4.997911068298263e-05, "loss": 0.5984, "step": 13112 }, { "epoch": 0.01, "learning_rate": 4.9979107472529266e-05, "loss": 0.4758, "step": 13113 }, { "epoch": 0.01, "learning_rate": 4.9979104261829324e-05, "loss": 0.462, "step": 13114 }, { "epoch": 0.01, "learning_rate": 4.9979101050882794e-05, "loss": 0.4119, "step": 13115 }, { "epoch": 0.01, "learning_rate": 4.997909783968968e-05, "loss": 0.4225, "step": 13116 }, { "epoch": 0.01, "learning_rate": 4.997909462824999e-05, "loss": 0.4138, "step": 13117 }, { "epoch": 0.01, "learning_rate": 4.997909141656371e-05, "loss": 0.5493, "step": 13118 }, { "epoch": 0.01, "learning_rate": 4.9979088204630845e-05, "loss": 0.5572, "step": 13119 }, { "epoch": 0.01, "learning_rate": 4.997908499245141e-05, "loss": 0.4677, "step": 13120 }, { "epoch": 0.01, "learning_rate": 4.9979081780025376e-05, "loss": 0.4902, "step": 13121 }, { "epoch": 0.01, "learning_rate": 4.997907856735277e-05, "loss": 0.4652, "step": 13122 }, { "epoch": 0.01, "learning_rate": 4.997907535443357e-05, "loss": 0.5826, "step": 13123 }, { "epoch": 0.01, "learning_rate": 4.9979072141267796e-05, "loss": 0.4451, "step": 13124 }, { "epoch": 0.01, "learning_rate": 4.997906892785544e-05, "loss": 0.4576, "step": 13125 }, { "epoch": 0.01, "learning_rate": 4.997906571419649e-05, "loss": 0.4687, "step": 13126 }, { "epoch": 0.01, "learning_rate": 4.997906250029097e-05, "loss": 0.4176, "step": 13127 }, { "epoch": 0.01, "learning_rate": 4.997905928613886e-05, "loss": 0.4297, "step": 13128 }, { "epoch": 0.01, "learning_rate": 4.997905607174017e-05, "loss": 0.4745, "step": 13129 }, { "epoch": 0.01, "learning_rate": 4.99790528570949e-05, "loss": 0.5335, "step": 13130 }, { "epoch": 0.01, "learning_rate": 4.9979049642203045e-05, "loss": 0.482, "step": 13131 }, { "epoch": 0.01, "learning_rate": 4.997904642706461e-05, "loss": 0.5525, "step": 13132 }, { "epoch": 0.01, "learning_rate": 4.9979043211679586e-05, "loss": 0.8318, "step": 13133 }, { "epoch": 0.01, "learning_rate": 4.997903999604798e-05, "loss": 1.1823, "step": 13134 }, { "epoch": 0.01, "learning_rate": 4.9979036780169795e-05, "loss": 1.3099, "step": 13135 }, { "epoch": 0.01, "learning_rate": 4.997903356404503e-05, "loss": 0.9838, "step": 13136 }, { "epoch": 0.01, "learning_rate": 4.997903034767368e-05, "loss": 1.0567, "step": 13137 }, { "epoch": 0.01, "learning_rate": 4.997902713105575e-05, "loss": 1.2956, "step": 13138 }, { "epoch": 0.01, "learning_rate": 4.997902391419123e-05, "loss": 1.2132, "step": 13139 }, { "epoch": 0.01, "learning_rate": 4.997902069708013e-05, "loss": 1.2639, "step": 13140 }, { "epoch": 0.01, "learning_rate": 4.9979017479722446e-05, "loss": 1.2568, "step": 13141 }, { "epoch": 0.01, "learning_rate": 4.997901426211818e-05, "loss": 1.2874, "step": 13142 }, { "epoch": 0.01, "learning_rate": 4.9979011044267345e-05, "loss": 1.105, "step": 13143 }, { "epoch": 0.01, "learning_rate": 4.997900782616991e-05, "loss": 1.1131, "step": 13144 }, { "epoch": 0.01, "learning_rate": 4.9979004607825904e-05, "loss": 1.1554, "step": 13145 }, { "epoch": 0.01, "learning_rate": 4.9979001389235315e-05, "loss": 0.9998, "step": 13146 }, { "epoch": 0.01, "learning_rate": 4.997899817039814e-05, "loss": 1.0174, "step": 13147 }, { "epoch": 0.01, "learning_rate": 4.9978994951314386e-05, "loss": 1.1904, "step": 13148 }, { "epoch": 0.01, "learning_rate": 4.9978991731984046e-05, "loss": 0.8066, "step": 13149 }, { "epoch": 0.01, "learning_rate": 4.9978988512407124e-05, "loss": 0.5018, "step": 13150 }, { "epoch": 0.01, "learning_rate": 4.997898529258362e-05, "loss": 0.3624, "step": 13151 }, { "epoch": 0.01, "learning_rate": 4.9978982072513536e-05, "loss": 0.3332, "step": 13152 }, { "epoch": 0.01, "learning_rate": 4.997897885219687e-05, "loss": 0.2909, "step": 13153 }, { "epoch": 0.01, "learning_rate": 4.9978975631633616e-05, "loss": 0.2502, "step": 13154 }, { "epoch": 0.01, "learning_rate": 4.997897241082379e-05, "loss": 1.0549, "step": 13155 }, { "epoch": 0.01, "learning_rate": 4.9978969189767377e-05, "loss": 1.093, "step": 13156 }, { "epoch": 0.01, "learning_rate": 4.997896596846438e-05, "loss": 1.2906, "step": 13157 }, { "epoch": 0.01, "learning_rate": 4.9978962746914805e-05, "loss": 0.7933, "step": 13158 }, { "epoch": 0.01, "learning_rate": 4.997895952511865e-05, "loss": 2.054, "step": 13159 }, { "epoch": 0.01, "learning_rate": 4.997895630307591e-05, "loss": 1.0205, "step": 13160 }, { "epoch": 0.01, "learning_rate": 4.997895308078658e-05, "loss": 1.2437, "step": 13161 }, { "epoch": 0.01, "learning_rate": 4.997894985825068e-05, "loss": 1.0688, "step": 13162 }, { "epoch": 0.01, "learning_rate": 4.9978946635468196e-05, "loss": 0.919, "step": 13163 }, { "epoch": 0.01, "learning_rate": 4.9978943412439134e-05, "loss": 0.8673, "step": 13164 }, { "epoch": 0.01, "learning_rate": 4.9978940189163483e-05, "loss": 0.4756, "step": 13165 }, { "epoch": 0.01, "learning_rate": 4.997893696564125e-05, "loss": 0.872, "step": 13166 }, { "epoch": 0.01, "learning_rate": 4.9978933741872445e-05, "loss": 1.2752, "step": 13167 }, { "epoch": 0.01, "learning_rate": 4.997893051785705e-05, "loss": 1.132, "step": 13168 }, { "epoch": 0.01, "learning_rate": 4.997892729359508e-05, "loss": 0.8541, "step": 13169 }, { "epoch": 0.01, "learning_rate": 4.997892406908652e-05, "loss": 1.2013, "step": 13170 }, { "epoch": 0.01, "learning_rate": 4.997892084433139e-05, "loss": 1.0825, "step": 13171 }, { "epoch": 0.01, "learning_rate": 4.997891761932967e-05, "loss": 1.1343, "step": 13172 }, { "epoch": 0.01, "learning_rate": 4.997891439408137e-05, "loss": 1.1043, "step": 13173 }, { "epoch": 0.01, "learning_rate": 4.9978911168586486e-05, "loss": 1.1395, "step": 13174 }, { "epoch": 0.01, "learning_rate": 4.997890794284502e-05, "loss": 1.1256, "step": 13175 }, { "epoch": 0.01, "learning_rate": 4.997890471685698e-05, "loss": 1.09, "step": 13176 }, { "epoch": 0.01, "learning_rate": 4.9978901490622354e-05, "loss": 1.0381, "step": 13177 }, { "epoch": 0.01, "learning_rate": 4.997889826414115e-05, "loss": 0.9623, "step": 13178 }, { "epoch": 0.01, "learning_rate": 4.997889503741337e-05, "loss": 0.575, "step": 13179 }, { "epoch": 0.01, "learning_rate": 4.9978891810438995e-05, "loss": 0.7448, "step": 13180 }, { "epoch": 0.01, "learning_rate": 4.997888858321804e-05, "loss": 0.6096, "step": 13181 }, { "epoch": 0.01, "learning_rate": 4.997888535575051e-05, "loss": 1.2684, "step": 13182 }, { "epoch": 0.01, "learning_rate": 4.9978882128036404e-05, "loss": 1.1718, "step": 13183 }, { "epoch": 0.01, "learning_rate": 4.997887890007571e-05, "loss": 1.1802, "step": 13184 }, { "epoch": 0.01, "learning_rate": 4.997887567186844e-05, "loss": 0.7917, "step": 13185 }, { "epoch": 0.01, "learning_rate": 4.997887244341458e-05, "loss": 1.0472, "step": 13186 }, { "epoch": 0.01, "learning_rate": 4.997886921471415e-05, "loss": 1.371, "step": 13187 }, { "epoch": 0.01, "learning_rate": 4.997886598576713e-05, "loss": 1.1039, "step": 13188 }, { "epoch": 0.01, "learning_rate": 4.9978862756573535e-05, "loss": 1.1748, "step": 13189 }, { "epoch": 0.01, "learning_rate": 4.997885952713336e-05, "loss": 1.0792, "step": 13190 }, { "epoch": 0.01, "learning_rate": 4.99788562974466e-05, "loss": 1.1257, "step": 13191 }, { "epoch": 0.01, "learning_rate": 4.997885306751326e-05, "loss": 1.1188, "step": 13192 }, { "epoch": 0.01, "learning_rate": 4.997884983733334e-05, "loss": 1.0585, "step": 13193 }, { "epoch": 0.01, "learning_rate": 4.9978846606906834e-05, "loss": 0.8592, "step": 13194 }, { "epoch": 0.01, "learning_rate": 4.997884337623376e-05, "loss": 1.3862, "step": 13195 }, { "epoch": 0.01, "learning_rate": 4.99788401453141e-05, "loss": 1.1329, "step": 13196 }, { "epoch": 0.01, "learning_rate": 4.9978836914147855e-05, "loss": 1.3172, "step": 13197 }, { "epoch": 0.01, "learning_rate": 4.997883368273503e-05, "loss": 1.1406, "step": 13198 }, { "epoch": 0.01, "learning_rate": 4.997883045107563e-05, "loss": 1.0109, "step": 13199 }, { "epoch": 0.01, "learning_rate": 4.9978827219169643e-05, "loss": 1.282, "step": 13200 }, { "epoch": 0.01, "learning_rate": 4.9978823987017074e-05, "loss": 0.9855, "step": 13201 }, { "epoch": 0.01, "learning_rate": 4.997882075461793e-05, "loss": 0.7936, "step": 13202 }, { "epoch": 0.01, "learning_rate": 4.997881752197221e-05, "loss": 1.0041, "step": 13203 }, { "epoch": 0.01, "learning_rate": 4.99788142890799e-05, "loss": 1.1821, "step": 13204 }, { "epoch": 0.01, "learning_rate": 4.997881105594102e-05, "loss": 1.3851, "step": 13205 }, { "epoch": 0.01, "learning_rate": 4.9978807822555554e-05, "loss": 0.9859, "step": 13206 }, { "epoch": 0.01, "learning_rate": 4.99788045889235e-05, "loss": 0.9828, "step": 13207 }, { "epoch": 0.01, "learning_rate": 4.997880135504488e-05, "loss": 0.9934, "step": 13208 }, { "epoch": 0.01, "learning_rate": 4.997879812091967e-05, "loss": 0.9731, "step": 13209 }, { "epoch": 0.01, "learning_rate": 4.997879488654789e-05, "loss": 1.0061, "step": 13210 }, { "epoch": 0.01, "learning_rate": 4.997879165192951e-05, "loss": 1.2834, "step": 13211 }, { "epoch": 0.01, "learning_rate": 4.997878841706457e-05, "loss": 1.2319, "step": 13212 }, { "epoch": 0.01, "learning_rate": 4.997878518195304e-05, "loss": 1.1768, "step": 13213 }, { "epoch": 0.01, "learning_rate": 4.997878194659493e-05, "loss": 1.0656, "step": 13214 }, { "epoch": 0.01, "learning_rate": 4.997877871099025e-05, "loss": 1.0897, "step": 13215 }, { "epoch": 0.01, "learning_rate": 4.997877547513898e-05, "loss": 1.3611, "step": 13216 }, { "epoch": 0.01, "learning_rate": 4.9978772239041134e-05, "loss": 0.9368, "step": 13217 }, { "epoch": 0.01, "learning_rate": 4.99787690026967e-05, "loss": 1.487, "step": 13218 }, { "epoch": 0.01, "learning_rate": 4.997876576610569e-05, "loss": 1.0307, "step": 13219 }, { "epoch": 0.01, "learning_rate": 4.997876252926811e-05, "loss": 1.1302, "step": 13220 }, { "epoch": 0.01, "learning_rate": 4.997875929218394e-05, "loss": 1.1502, "step": 13221 }, { "epoch": 0.01, "learning_rate": 4.997875605485319e-05, "loss": 1.1711, "step": 13222 }, { "epoch": 0.01, "learning_rate": 4.997875281727587e-05, "loss": 0.9221, "step": 13223 }, { "epoch": 0.01, "learning_rate": 4.997874957945196e-05, "loss": 0.903, "step": 13224 }, { "epoch": 0.01, "learning_rate": 4.997874634138148e-05, "loss": 1.0384, "step": 13225 }, { "epoch": 0.01, "learning_rate": 4.9978743103064415e-05, "loss": 1.1008, "step": 13226 }, { "epoch": 0.01, "learning_rate": 4.997873986450077e-05, "loss": 1.1846, "step": 13227 }, { "epoch": 0.01, "learning_rate": 4.997873662569054e-05, "loss": 1.3003, "step": 13228 }, { "epoch": 0.01, "learning_rate": 4.997873338663374e-05, "loss": 1.1651, "step": 13229 }, { "epoch": 0.01, "learning_rate": 4.9978730147330355e-05, "loss": 0.9041, "step": 13230 }, { "epoch": 0.01, "learning_rate": 4.9978726907780396e-05, "loss": 1.1599, "step": 13231 }, { "epoch": 0.01, "learning_rate": 4.997872366798385e-05, "loss": 1.2287, "step": 13232 }, { "epoch": 0.01, "learning_rate": 4.997872042794073e-05, "loss": 1.3725, "step": 13233 }, { "epoch": 0.01, "learning_rate": 4.9978717187651025e-05, "loss": 1.2067, "step": 13234 }, { "epoch": 0.01, "learning_rate": 4.997871394711474e-05, "loss": 1.1195, "step": 13235 }, { "epoch": 0.01, "learning_rate": 4.997871070633188e-05, "loss": 1.0614, "step": 13236 }, { "epoch": 0.01, "learning_rate": 4.997870746530244e-05, "loss": 0.9316, "step": 13237 }, { "epoch": 0.01, "learning_rate": 4.9978704224026426e-05, "loss": 0.9898, "step": 13238 }, { "epoch": 0.01, "learning_rate": 4.997870098250382e-05, "loss": 0.7764, "step": 13239 }, { "epoch": 0.01, "learning_rate": 4.997869774073465e-05, "loss": 0.5434, "step": 13240 }, { "epoch": 0.01, "learning_rate": 4.997869449871889e-05, "loss": 1.0947, "step": 13241 }, { "epoch": 0.01, "learning_rate": 4.997869125645656e-05, "loss": 1.0978, "step": 13242 }, { "epoch": 0.01, "learning_rate": 4.997868801394764e-05, "loss": 1.242, "step": 13243 }, { "epoch": 0.01, "learning_rate": 4.9978684771192144e-05, "loss": 1.0641, "step": 13244 }, { "epoch": 0.01, "learning_rate": 4.997868152819007e-05, "loss": 1.096, "step": 13245 }, { "epoch": 0.01, "learning_rate": 4.9978678284941426e-05, "loss": 1.0832, "step": 13246 }, { "epoch": 0.01, "learning_rate": 4.997867504144619e-05, "loss": 1.04, "step": 13247 }, { "epoch": 0.01, "learning_rate": 4.997867179770438e-05, "loss": 1.0938, "step": 13248 }, { "epoch": 0.01, "learning_rate": 4.997866855371599e-05, "loss": 1.0092, "step": 13249 }, { "epoch": 0.01, "learning_rate": 4.9978665309481025e-05, "loss": 1.2841, "step": 13250 }, { "epoch": 0.01, "learning_rate": 4.9978662064999485e-05, "loss": 1.3628, "step": 13251 }, { "epoch": 0.01, "learning_rate": 4.997865882027135e-05, "loss": 1.3454, "step": 13252 }, { "epoch": 0.01, "learning_rate": 4.997865557529665e-05, "loss": 1.1328, "step": 13253 }, { "epoch": 0.01, "learning_rate": 4.997865233007537e-05, "loss": 1.2953, "step": 13254 }, { "epoch": 0.01, "learning_rate": 4.99786490846075e-05, "loss": 1.0408, "step": 13255 }, { "epoch": 0.01, "learning_rate": 4.997864583889307e-05, "loss": 1.0113, "step": 13256 }, { "epoch": 0.01, "learning_rate": 4.9978642592932046e-05, "loss": 0.99, "step": 13257 }, { "epoch": 0.01, "learning_rate": 4.997863934672445e-05, "loss": 1.3442, "step": 13258 }, { "epoch": 0.01, "learning_rate": 4.997863610027027e-05, "loss": 1.1026, "step": 13259 }, { "epoch": 0.01, "learning_rate": 4.997863285356952e-05, "loss": 1.1804, "step": 13260 }, { "epoch": 0.01, "learning_rate": 4.997862960662218e-05, "loss": 1.118, "step": 13261 }, { "epoch": 0.01, "learning_rate": 4.9978626359428274e-05, "loss": 0.8905, "step": 13262 }, { "epoch": 0.01, "learning_rate": 4.997862311198778e-05, "loss": 0.859, "step": 13263 }, { "epoch": 0.01, "learning_rate": 4.997861986430071e-05, "loss": 0.9503, "step": 13264 }, { "epoch": 0.01, "learning_rate": 4.9978616616367065e-05, "loss": 0.837, "step": 13265 }, { "epoch": 0.01, "learning_rate": 4.9978613368186837e-05, "loss": 0.8207, "step": 13266 }, { "epoch": 0.01, "learning_rate": 4.997861011976004e-05, "loss": 1.2311, "step": 13267 }, { "epoch": 0.01, "learning_rate": 4.9978606871086656e-05, "loss": 1.1697, "step": 13268 }, { "epoch": 0.01, "learning_rate": 4.997860362216669e-05, "loss": 0.8688, "step": 13269 }, { "epoch": 0.01, "learning_rate": 4.997860037300015e-05, "loss": 0.7533, "step": 13270 }, { "epoch": 0.01, "learning_rate": 4.9978597123587035e-05, "loss": 0.8438, "step": 13271 }, { "epoch": 0.01, "learning_rate": 4.997859387392734e-05, "loss": 1.1314, "step": 13272 }, { "epoch": 0.01, "learning_rate": 4.9978590624021074e-05, "loss": 1.059, "step": 13273 }, { "epoch": 0.01, "learning_rate": 4.9978587373868214e-05, "loss": 1.2934, "step": 13274 }, { "epoch": 0.01, "learning_rate": 4.997858412346879e-05, "loss": 1.2194, "step": 13275 }, { "epoch": 0.01, "learning_rate": 4.997858087282278e-05, "loss": 1.388, "step": 13276 }, { "epoch": 0.01, "learning_rate": 4.997857762193019e-05, "loss": 1.0481, "step": 13277 }, { "epoch": 0.01, "learning_rate": 4.997857437079103e-05, "loss": 0.8995, "step": 13278 }, { "epoch": 0.01, "learning_rate": 4.997857111940529e-05, "loss": 1.1896, "step": 13279 }, { "epoch": 0.01, "learning_rate": 4.997856786777297e-05, "loss": 1.0073, "step": 13280 }, { "epoch": 0.01, "learning_rate": 4.997856461589407e-05, "loss": 0.8425, "step": 13281 }, { "epoch": 0.01, "learning_rate": 4.9978561363768595e-05, "loss": 1.5723, "step": 13282 }, { "epoch": 0.01, "learning_rate": 4.997855811139654e-05, "loss": 1.0289, "step": 13283 }, { "epoch": 0.01, "learning_rate": 4.997855485877792e-05, "loss": 0.913, "step": 13284 }, { "epoch": 0.01, "learning_rate": 4.99785516059127e-05, "loss": 1.1144, "step": 13285 }, { "epoch": 0.01, "learning_rate": 4.997854835280091e-05, "loss": 1.0904, "step": 13286 }, { "epoch": 0.01, "learning_rate": 4.9978545099442555e-05, "loss": 0.9672, "step": 13287 }, { "epoch": 0.01, "learning_rate": 4.997854184583761e-05, "loss": 1.1621, "step": 13288 }, { "epoch": 0.01, "learning_rate": 4.997853859198609e-05, "loss": 0.9643, "step": 13289 }, { "epoch": 0.01, "learning_rate": 4.9978535337887994e-05, "loss": 1.0162, "step": 13290 }, { "epoch": 0.01, "learning_rate": 4.997853208354332e-05, "loss": 1.1104, "step": 13291 }, { "epoch": 0.01, "learning_rate": 4.9978528828952066e-05, "loss": 1.0895, "step": 13292 }, { "epoch": 0.01, "learning_rate": 4.997852557411424e-05, "loss": 0.9818, "step": 13293 }, { "epoch": 0.01, "learning_rate": 4.997852231902983e-05, "loss": 1.1447, "step": 13294 }, { "epoch": 0.01, "learning_rate": 4.997851906369885e-05, "loss": 1.1971, "step": 13295 }, { "epoch": 0.01, "learning_rate": 4.997851580812129e-05, "loss": 0.8869, "step": 13296 }, { "epoch": 0.01, "learning_rate": 4.997851255229714e-05, "loss": 1.4195, "step": 13297 }, { "epoch": 0.01, "learning_rate": 4.997850929622643e-05, "loss": 1.0647, "step": 13298 }, { "epoch": 0.01, "learning_rate": 4.9978506039909135e-05, "loss": 0.9029, "step": 13299 }, { "epoch": 0.01, "learning_rate": 4.997850278334526e-05, "loss": 1.104, "step": 13300 }, { "epoch": 0.01, "learning_rate": 4.9978499526534816e-05, "loss": 1.2008, "step": 13301 }, { "epoch": 0.01, "learning_rate": 4.9978496269477784e-05, "loss": 0.8578, "step": 13302 }, { "epoch": 0.01, "learning_rate": 4.997849301217419e-05, "loss": 1.1522, "step": 13303 }, { "epoch": 0.01, "learning_rate": 4.9978489754624e-05, "loss": 1.3556, "step": 13304 }, { "epoch": 0.01, "learning_rate": 4.997848649682725e-05, "loss": 1.2666, "step": 13305 }, { "epoch": 0.01, "learning_rate": 4.997848323878391e-05, "loss": 1.2673, "step": 13306 }, { "epoch": 0.01, "learning_rate": 4.9978479980494005e-05, "loss": 1.0229, "step": 13307 }, { "epoch": 0.01, "learning_rate": 4.997847672195751e-05, "loss": 1.771, "step": 13308 }, { "epoch": 0.01, "learning_rate": 4.997847346317445e-05, "loss": 0.9576, "step": 13309 }, { "epoch": 0.01, "learning_rate": 4.997847020414481e-05, "loss": 1.0482, "step": 13310 }, { "epoch": 0.01, "learning_rate": 4.997846694486858e-05, "loss": 0.9268, "step": 13311 }, { "epoch": 0.01, "learning_rate": 4.997846368534579e-05, "loss": 1.0142, "step": 13312 }, { "epoch": 0.01, "learning_rate": 4.997846042557642e-05, "loss": 0.9943, "step": 13313 }, { "epoch": 0.01, "learning_rate": 4.997845716556047e-05, "loss": 0.9842, "step": 13314 }, { "epoch": 0.01, "learning_rate": 4.997845390529794e-05, "loss": 1.0571, "step": 13315 }, { "epoch": 0.01, "learning_rate": 4.9978450644788834e-05, "loss": 1.0647, "step": 13316 }, { "epoch": 0.01, "learning_rate": 4.9978447384033155e-05, "loss": 1.3043, "step": 13317 }, { "epoch": 0.01, "learning_rate": 4.99784441230309e-05, "loss": 1.1047, "step": 13318 }, { "epoch": 0.01, "learning_rate": 4.997844086178206e-05, "loss": 1.0826, "step": 13319 }, { "epoch": 0.01, "learning_rate": 4.9978437600286656e-05, "loss": 0.7416, "step": 13320 }, { "epoch": 0.01, "learning_rate": 4.9978434338544665e-05, "loss": 0.9795, "step": 13321 }, { "epoch": 0.01, "learning_rate": 4.9978431076556106e-05, "loss": 1.1324, "step": 13322 }, { "epoch": 0.01, "learning_rate": 4.997842781432096e-05, "loss": 0.931, "step": 13323 }, { "epoch": 0.01, "learning_rate": 4.997842455183925e-05, "loss": 0.9355, "step": 13324 }, { "epoch": 0.01, "learning_rate": 4.997842128911095e-05, "loss": 1.0136, "step": 13325 }, { "epoch": 0.01, "learning_rate": 4.997841802613609e-05, "loss": 1.1904, "step": 13326 }, { "epoch": 0.01, "learning_rate": 4.997841476291464e-05, "loss": 0.8844, "step": 13327 }, { "epoch": 0.01, "learning_rate": 4.997841149944662e-05, "loss": 1.1935, "step": 13328 }, { "epoch": 0.01, "learning_rate": 4.9978408235732025e-05, "loss": 1.3043, "step": 13329 }, { "epoch": 0.01, "learning_rate": 4.997840497177084e-05, "loss": 1.4427, "step": 13330 }, { "epoch": 0.01, "learning_rate": 4.9978401707563096e-05, "loss": 1.121, "step": 13331 }, { "epoch": 0.01, "learning_rate": 4.997839844310876e-05, "loss": 1.0611, "step": 13332 }, { "epoch": 0.01, "learning_rate": 4.997839517840786e-05, "loss": 1.2253, "step": 13333 }, { "epoch": 0.01, "learning_rate": 4.997839191346038e-05, "loss": 1.2705, "step": 13334 }, { "epoch": 0.01, "learning_rate": 4.997838864826633e-05, "loss": 0.8734, "step": 13335 }, { "epoch": 0.01, "learning_rate": 4.997838538282569e-05, "loss": 1.1791, "step": 13336 }, { "epoch": 0.01, "learning_rate": 4.997838211713848e-05, "loss": 1.1229, "step": 13337 }, { "epoch": 0.01, "learning_rate": 4.99783788512047e-05, "loss": 1.0637, "step": 13338 }, { "epoch": 0.01, "learning_rate": 4.997837558502434e-05, "loss": 1.0856, "step": 13339 }, { "epoch": 0.01, "learning_rate": 4.99783723185974e-05, "loss": 1.2392, "step": 13340 }, { "epoch": 0.01, "learning_rate": 4.997836905192389e-05, "loss": 0.953, "step": 13341 }, { "epoch": 0.01, "learning_rate": 4.99783657850038e-05, "loss": 1.0463, "step": 13342 }, { "epoch": 0.01, "learning_rate": 4.997836251783714e-05, "loss": 1.5255, "step": 13343 }, { "epoch": 0.01, "learning_rate": 4.99783592504239e-05, "loss": 1.0337, "step": 13344 }, { "epoch": 0.01, "learning_rate": 4.997835598276408e-05, "loss": 1.0521, "step": 13345 }, { "epoch": 0.01, "learning_rate": 4.997835271485769e-05, "loss": 1.051, "step": 13346 }, { "epoch": 0.01, "learning_rate": 4.997834944670472e-05, "loss": 1.206, "step": 13347 }, { "epoch": 0.01, "learning_rate": 4.997834617830518e-05, "loss": 0.962, "step": 13348 }, { "epoch": 0.01, "learning_rate": 4.997834290965906e-05, "loss": 0.9511, "step": 13349 }, { "epoch": 0.01, "learning_rate": 4.997833964076637e-05, "loss": 1.265, "step": 13350 }, { "epoch": 0.01, "learning_rate": 4.99783363716271e-05, "loss": 1.032, "step": 13351 }, { "epoch": 0.01, "learning_rate": 4.9978333102241247e-05, "loss": 1.1215, "step": 13352 }, { "epoch": 0.01, "learning_rate": 4.9978329832608825e-05, "loss": 0.7762, "step": 13353 }, { "epoch": 0.01, "learning_rate": 4.9978326562729835e-05, "loss": 1.1542, "step": 13354 }, { "epoch": 0.01, "learning_rate": 4.997832329260426e-05, "loss": 1.1663, "step": 13355 }, { "epoch": 0.01, "learning_rate": 4.997832002223211e-05, "loss": 1.0352, "step": 13356 }, { "epoch": 0.01, "learning_rate": 4.9978316751613384e-05, "loss": 0.9754, "step": 13357 }, { "epoch": 0.01, "learning_rate": 4.997831348074808e-05, "loss": 1.048, "step": 13358 }, { "epoch": 0.01, "learning_rate": 4.997831020963621e-05, "loss": 1.1207, "step": 13359 }, { "epoch": 0.01, "learning_rate": 4.997830693827776e-05, "loss": 0.8837, "step": 13360 }, { "epoch": 0.01, "learning_rate": 4.997830366667274e-05, "loss": 1.3253, "step": 13361 }, { "epoch": 0.01, "learning_rate": 4.997830039482114e-05, "loss": 1.2562, "step": 13362 }, { "epoch": 0.01, "learning_rate": 4.997829712272296e-05, "loss": 1.3778, "step": 13363 }, { "epoch": 0.01, "learning_rate": 4.997829385037821e-05, "loss": 0.9268, "step": 13364 }, { "epoch": 0.01, "learning_rate": 4.997829057778689e-05, "loss": 1.0259, "step": 13365 }, { "epoch": 0.01, "learning_rate": 4.997828730494899e-05, "loss": 1.2288, "step": 13366 }, { "epoch": 0.01, "learning_rate": 4.997828403186451e-05, "loss": 1.1526, "step": 13367 }, { "epoch": 0.01, "learning_rate": 4.9978280758533455e-05, "loss": 1.0788, "step": 13368 }, { "epoch": 0.01, "learning_rate": 4.997827748495583e-05, "loss": 1.1302, "step": 13369 }, { "epoch": 0.01, "learning_rate": 4.9978274211131633e-05, "loss": 1.4603, "step": 13370 }, { "epoch": 0.01, "learning_rate": 4.9978270937060854e-05, "loss": 1.3785, "step": 13371 }, { "epoch": 0.01, "learning_rate": 4.9978267662743506e-05, "loss": 1.2116, "step": 13372 }, { "epoch": 0.01, "learning_rate": 4.997826438817958e-05, "loss": 1.2782, "step": 13373 }, { "epoch": 0.01, "learning_rate": 4.9978261113369074e-05, "loss": 1.1429, "step": 13374 }, { "epoch": 0.01, "learning_rate": 4.9978257838312e-05, "loss": 1.1785, "step": 13375 }, { "epoch": 0.01, "learning_rate": 4.9978254563008356e-05, "loss": 0.774, "step": 13376 }, { "epoch": 0.01, "learning_rate": 4.997825128745812e-05, "loss": 1.0027, "step": 13377 }, { "epoch": 0.01, "learning_rate": 4.997824801166132e-05, "loss": 0.9855, "step": 13378 }, { "epoch": 0.01, "learning_rate": 4.997824473561795e-05, "loss": 1.1095, "step": 13379 }, { "epoch": 0.01, "learning_rate": 4.9978241459328004e-05, "loss": 1.1271, "step": 13380 }, { "epoch": 0.01, "learning_rate": 4.997823818279147e-05, "loss": 1.0207, "step": 13381 }, { "epoch": 0.01, "learning_rate": 4.997823490600838e-05, "loss": 1.0327, "step": 13382 }, { "epoch": 0.01, "learning_rate": 4.99782316289787e-05, "loss": 1.0698, "step": 13383 }, { "epoch": 0.01, "learning_rate": 4.997822835170245e-05, "loss": 1.2344, "step": 13384 }, { "epoch": 0.01, "learning_rate": 4.997822507417963e-05, "loss": 1.2526, "step": 13385 }, { "epoch": 0.01, "learning_rate": 4.997822179641023e-05, "loss": 1.2137, "step": 13386 }, { "epoch": 0.01, "learning_rate": 4.997821851839426e-05, "loss": 0.9287, "step": 13387 }, { "epoch": 0.01, "learning_rate": 4.997821524013171e-05, "loss": 1.1097, "step": 13388 }, { "epoch": 0.01, "learning_rate": 4.997821196162259e-05, "loss": 1.0661, "step": 13389 }, { "epoch": 0.01, "learning_rate": 4.99782086828669e-05, "loss": 0.9219, "step": 13390 }, { "epoch": 0.01, "learning_rate": 4.9978205403864634e-05, "loss": 1.1103, "step": 13391 }, { "epoch": 0.01, "learning_rate": 4.997820212461578e-05, "loss": 0.8699, "step": 13392 }, { "epoch": 0.01, "learning_rate": 4.9978198845120364e-05, "loss": 0.889, "step": 13393 }, { "epoch": 0.01, "learning_rate": 4.997819556537837e-05, "loss": 1.1113, "step": 13394 }, { "epoch": 0.01, "learning_rate": 4.997819228538981e-05, "loss": 1.1604, "step": 13395 }, { "epoch": 0.01, "learning_rate": 4.997818900515466e-05, "loss": 1.141, "step": 13396 }, { "epoch": 0.01, "learning_rate": 4.9978185724672954e-05, "loss": 0.5784, "step": 13397 }, { "epoch": 0.01, "learning_rate": 4.9978182443944665e-05, "loss": 1.2137, "step": 13398 }, { "epoch": 0.01, "learning_rate": 4.9978179162969794e-05, "loss": 1.1008, "step": 13399 }, { "epoch": 0.01, "learning_rate": 4.997817588174836e-05, "loss": 0.967, "step": 13400 }, { "epoch": 0.01, "learning_rate": 4.997817260028035e-05, "loss": 1.0273, "step": 13401 }, { "epoch": 0.01, "learning_rate": 4.997816931856577e-05, "loss": 1.1577, "step": 13402 }, { "epoch": 0.01, "learning_rate": 4.9978166036604606e-05, "loss": 1.2207, "step": 13403 }, { "epoch": 0.01, "learning_rate": 4.9978162754396875e-05, "loss": 1.4783, "step": 13404 }, { "epoch": 0.01, "learning_rate": 4.9978159471942563e-05, "loss": 1.1987, "step": 13405 }, { "epoch": 0.01, "learning_rate": 4.997815618924169e-05, "loss": 1.4815, "step": 13406 }, { "epoch": 0.01, "learning_rate": 4.997815290629423e-05, "loss": 1.1812, "step": 13407 }, { "epoch": 0.01, "learning_rate": 4.99781496231002e-05, "loss": 0.9092, "step": 13408 }, { "epoch": 0.01, "learning_rate": 4.9978146339659596e-05, "loss": 1.0899, "step": 13409 }, { "epoch": 0.01, "learning_rate": 4.9978143055972424e-05, "loss": 1.3144, "step": 13410 }, { "epoch": 0.01, "learning_rate": 4.997813977203867e-05, "loss": 1.1962, "step": 13411 }, { "epoch": 0.01, "learning_rate": 4.997813648785835e-05, "loss": 1.2084, "step": 13412 }, { "epoch": 0.01, "learning_rate": 4.9978133203431454e-05, "loss": 1.0493, "step": 13413 }, { "epoch": 0.01, "learning_rate": 4.9978129918757984e-05, "loss": 1.1011, "step": 13414 }, { "epoch": 0.01, "learning_rate": 4.997812663383794e-05, "loss": 1.3777, "step": 13415 }, { "epoch": 0.01, "learning_rate": 4.9978123348671326e-05, "loss": 1.651, "step": 13416 }, { "epoch": 0.01, "learning_rate": 4.9978120063258125e-05, "loss": 1.5813, "step": 13417 }, { "epoch": 0.01, "learning_rate": 4.997811677759836e-05, "loss": 1.5543, "step": 13418 }, { "epoch": 0.01, "learning_rate": 4.9978113491692025e-05, "loss": 1.2562, "step": 13419 }, { "epoch": 0.01, "learning_rate": 4.997811020553911e-05, "loss": 1.6136, "step": 13420 }, { "epoch": 0.01, "learning_rate": 4.9978106919139634e-05, "loss": 1.7078, "step": 13421 }, { "epoch": 0.01, "learning_rate": 4.997810363249357e-05, "loss": 1.173, "step": 13422 }, { "epoch": 0.01, "learning_rate": 4.9978100345600944e-05, "loss": 1.5011, "step": 13423 }, { "epoch": 0.01, "learning_rate": 4.997809705846174e-05, "loss": 1.6022, "step": 13424 }, { "epoch": 0.01, "learning_rate": 4.9978093771075955e-05, "loss": 1.2602, "step": 13425 }, { "epoch": 0.01, "learning_rate": 4.997809048344361e-05, "loss": 1.3651, "step": 13426 }, { "epoch": 0.01, "learning_rate": 4.997808719556468e-05, "loss": 1.3125, "step": 13427 }, { "epoch": 0.01, "learning_rate": 4.9978083907439185e-05, "loss": 1.3091, "step": 13428 }, { "epoch": 0.01, "learning_rate": 4.9978080619067115e-05, "loss": 0.9429, "step": 13429 }, { "epoch": 0.01, "learning_rate": 4.997807733044847e-05, "loss": 1.1973, "step": 13430 }, { "epoch": 0.01, "learning_rate": 4.997807404158325e-05, "loss": 1.0719, "step": 13431 }, { "epoch": 0.01, "learning_rate": 4.997807075247146e-05, "loss": 1.0395, "step": 13432 }, { "epoch": 0.01, "learning_rate": 4.99780674631131e-05, "loss": 1.1366, "step": 13433 }, { "epoch": 0.01, "learning_rate": 4.9978064173508164e-05, "loss": 1.019, "step": 13434 }, { "epoch": 0.01, "learning_rate": 4.997806088365665e-05, "loss": 1.0591, "step": 13435 }, { "epoch": 0.01, "learning_rate": 4.997805759355857e-05, "loss": 1.0975, "step": 13436 }, { "epoch": 0.01, "learning_rate": 4.997805430321392e-05, "loss": 1.2038, "step": 13437 }, { "epoch": 0.01, "learning_rate": 4.997805101262269e-05, "loss": 1.2305, "step": 13438 }, { "epoch": 0.01, "learning_rate": 4.997804772178489e-05, "loss": 1.1124, "step": 13439 }, { "epoch": 0.01, "learning_rate": 4.9978044430700514e-05, "loss": 1.1629, "step": 13440 }, { "epoch": 0.01, "learning_rate": 4.997804113936957e-05, "loss": 1.0923, "step": 13441 }, { "epoch": 0.01, "learning_rate": 4.9978037847792054e-05, "loss": 0.9277, "step": 13442 }, { "epoch": 0.01, "learning_rate": 4.997803455596796e-05, "loss": 0.981, "step": 13443 }, { "epoch": 0.01, "learning_rate": 4.9978031263897295e-05, "loss": 1.1139, "step": 13444 }, { "epoch": 0.01, "learning_rate": 4.997802797158006e-05, "loss": 1.2224, "step": 13445 }, { "epoch": 0.01, "learning_rate": 4.997802467901626e-05, "loss": 0.944, "step": 13446 }, { "epoch": 0.01, "learning_rate": 4.997802138620587e-05, "loss": 1.0513, "step": 13447 }, { "epoch": 0.01, "learning_rate": 4.9978018093148915e-05, "loss": 1.0209, "step": 13448 }, { "epoch": 0.01, "learning_rate": 4.997801479984539e-05, "loss": 1.0556, "step": 13449 }, { "epoch": 0.01, "learning_rate": 4.9978011506295294e-05, "loss": 1.188, "step": 13450 }, { "epoch": 0.01, "learning_rate": 4.9978008212498615e-05, "loss": 1.4315, "step": 13451 }, { "epoch": 0.01, "learning_rate": 4.997800491845538e-05, "loss": 1.6668, "step": 13452 }, { "epoch": 0.01, "learning_rate": 4.997800162416556e-05, "loss": 1.1929, "step": 13453 }, { "epoch": 0.01, "learning_rate": 4.997799832962917e-05, "loss": 1.0631, "step": 13454 }, { "epoch": 0.01, "learning_rate": 4.997799503484621e-05, "loss": 1.1428, "step": 13455 }, { "epoch": 0.01, "learning_rate": 4.997799173981668e-05, "loss": 0.8503, "step": 13456 }, { "epoch": 0.01, "learning_rate": 4.997798844454057e-05, "loss": 0.9871, "step": 13457 }, { "epoch": 0.01, "learning_rate": 4.99779851490179e-05, "loss": 1.2522, "step": 13458 }, { "epoch": 0.01, "learning_rate": 4.997798185324865e-05, "loss": 1.2566, "step": 13459 }, { "epoch": 0.01, "learning_rate": 4.997797855723283e-05, "loss": 0.9995, "step": 13460 }, { "epoch": 0.01, "learning_rate": 4.997797526097043e-05, "loss": 0.896, "step": 13461 }, { "epoch": 0.01, "learning_rate": 4.997797196446147e-05, "loss": 1.0634, "step": 13462 }, { "epoch": 0.01, "learning_rate": 4.997796866770593e-05, "loss": 0.6953, "step": 13463 }, { "epoch": 0.01, "learning_rate": 4.997796537070382e-05, "loss": 1.043, "step": 13464 }, { "epoch": 0.01, "learning_rate": 4.997796207345514e-05, "loss": 1.3509, "step": 13465 }, { "epoch": 0.01, "learning_rate": 4.997795877595989e-05, "loss": 1.317, "step": 13466 }, { "epoch": 0.01, "learning_rate": 4.997795547821806e-05, "loss": 1.2163, "step": 13467 }, { "epoch": 0.01, "learning_rate": 4.997795218022966e-05, "loss": 1.1202, "step": 13468 }, { "epoch": 0.01, "learning_rate": 4.99779488819947e-05, "loss": 1.0508, "step": 13469 }, { "epoch": 0.01, "learning_rate": 4.9977945583513155e-05, "loss": 1.203, "step": 13470 }, { "epoch": 0.01, "learning_rate": 4.997794228478504e-05, "loss": 1.082, "step": 13471 }, { "epoch": 0.01, "learning_rate": 4.9977938985810356e-05, "loss": 0.8471, "step": 13472 }, { "epoch": 0.01, "learning_rate": 4.99779356865891e-05, "loss": 0.4605, "step": 13473 }, { "epoch": 0.01, "learning_rate": 4.997793238712127e-05, "loss": 0.2344, "step": 13474 }, { "epoch": 0.01, "learning_rate": 4.997792908740687e-05, "loss": 0.3221, "step": 13475 }, { "epoch": 0.01, "learning_rate": 4.9977925787445904e-05, "loss": 1.3215, "step": 13476 }, { "epoch": 0.01, "learning_rate": 4.997792248723836e-05, "loss": 1.2194, "step": 13477 }, { "epoch": 0.01, "learning_rate": 4.997791918678425e-05, "loss": 0.6998, "step": 13478 }, { "epoch": 0.01, "learning_rate": 4.997791588608356e-05, "loss": 1.1459, "step": 13479 }, { "epoch": 0.01, "learning_rate": 4.9977912585136304e-05, "loss": 0.9691, "step": 13480 }, { "epoch": 0.01, "learning_rate": 4.997790928394248e-05, "loss": 0.8361, "step": 13481 }, { "epoch": 0.01, "learning_rate": 4.997790598250208e-05, "loss": 1.3147, "step": 13482 }, { "epoch": 0.01, "learning_rate": 4.9977902680815105e-05, "loss": 0.9627, "step": 13483 }, { "epoch": 0.01, "learning_rate": 4.997789937888156e-05, "loss": 0.7577, "step": 13484 }, { "epoch": 0.01, "learning_rate": 4.9977896076701446e-05, "loss": 0.5085, "step": 13485 }, { "epoch": 0.01, "learning_rate": 4.997789277427477e-05, "loss": 0.6398, "step": 13486 }, { "epoch": 0.01, "learning_rate": 4.997788947160151e-05, "loss": 1.1449, "step": 13487 }, { "epoch": 0.01, "learning_rate": 4.997788616868168e-05, "loss": 1.0784, "step": 13488 }, { "epoch": 0.01, "learning_rate": 4.997788286551528e-05, "loss": 0.9939, "step": 13489 }, { "epoch": 0.01, "learning_rate": 4.9977879562102316e-05, "loss": 1.1397, "step": 13490 }, { "epoch": 0.01, "learning_rate": 4.997787625844277e-05, "loss": 1.3888, "step": 13491 }, { "epoch": 0.01, "learning_rate": 4.9977872954536666e-05, "loss": 1.2933, "step": 13492 }, { "epoch": 0.01, "learning_rate": 4.997786965038398e-05, "loss": 1.1572, "step": 13493 }, { "epoch": 0.01, "learning_rate": 4.9977866345984724e-05, "loss": 1.1605, "step": 13494 }, { "epoch": 0.01, "learning_rate": 4.99778630413389e-05, "loss": 1.1386, "step": 13495 }, { "epoch": 0.01, "learning_rate": 4.9977859736446503e-05, "loss": 1.1907, "step": 13496 }, { "epoch": 0.01, "learning_rate": 4.997785643130754e-05, "loss": 1.1242, "step": 13497 }, { "epoch": 0.01, "learning_rate": 4.9977853125922004e-05, "loss": 0.7879, "step": 13498 }, { "epoch": 0.01, "learning_rate": 4.997784982028989e-05, "loss": 0.6311, "step": 13499 }, { "epoch": 0.01, "learning_rate": 4.997784651441121e-05, "loss": 0.883, "step": 13500 }, { "epoch": 0.01, "eval_loss": 1.0440155267715454, "eval_runtime": 86.2254, "eval_samples_per_second": 16.063, "eval_steps_per_second": 4.024, "step": 13500 }, { "epoch": 0.01, "learning_rate": 4.997784320828596e-05, "loss": 0.7417, "step": 13501 }, { "epoch": 0.01, "learning_rate": 4.9977839901914144e-05, "loss": 0.8161, "step": 13502 }, { "epoch": 0.01, "learning_rate": 4.9977836595295744e-05, "loss": 0.7489, "step": 13503 }, { "epoch": 0.01, "learning_rate": 4.997783328843079e-05, "loss": 0.5493, "step": 13504 }, { "epoch": 0.01, "learning_rate": 4.997782998131926e-05, "loss": 0.4742, "step": 13505 }, { "epoch": 0.01, "learning_rate": 4.997782667396115e-05, "loss": 0.4656, "step": 13506 }, { "epoch": 0.01, "learning_rate": 4.997782336635648e-05, "loss": 0.4238, "step": 13507 }, { "epoch": 0.01, "learning_rate": 4.997782005850523e-05, "loss": 0.438, "step": 13508 }, { "epoch": 0.01, "learning_rate": 4.997781675040742e-05, "loss": 0.5162, "step": 13509 }, { "epoch": 0.01, "learning_rate": 4.997781344206303e-05, "loss": 0.4822, "step": 13510 }, { "epoch": 0.01, "learning_rate": 4.997781013347208e-05, "loss": 0.4873, "step": 13511 }, { "epoch": 0.01, "learning_rate": 4.9977806824634554e-05, "loss": 0.5565, "step": 13512 }, { "epoch": 0.01, "learning_rate": 4.9977803515550455e-05, "loss": 0.4265, "step": 13513 }, { "epoch": 0.01, "learning_rate": 4.997780020621978e-05, "loss": 0.4158, "step": 13514 }, { "epoch": 0.01, "learning_rate": 4.9977796896642546e-05, "loss": 0.4371, "step": 13515 }, { "epoch": 0.01, "learning_rate": 4.997779358681874e-05, "loss": 0.6408, "step": 13516 }, { "epoch": 0.01, "learning_rate": 4.997779027674836e-05, "loss": 0.5063, "step": 13517 }, { "epoch": 0.01, "learning_rate": 4.997778696643142e-05, "loss": 0.4855, "step": 13518 }, { "epoch": 0.01, "learning_rate": 4.997778365586789e-05, "loss": 0.5207, "step": 13519 }, { "epoch": 0.01, "learning_rate": 4.997778034505781e-05, "loss": 0.4506, "step": 13520 }, { "epoch": 0.01, "learning_rate": 4.997777703400115e-05, "loss": 0.4622, "step": 13521 }, { "epoch": 0.01, "learning_rate": 4.997777372269792e-05, "loss": 0.4479, "step": 13522 }, { "epoch": 0.01, "learning_rate": 4.9977770411148125e-05, "loss": 0.4662, "step": 13523 }, { "epoch": 0.01, "learning_rate": 4.997776709935176e-05, "loss": 0.4799, "step": 13524 }, { "epoch": 0.01, "learning_rate": 4.997776378730882e-05, "loss": 0.4619, "step": 13525 }, { "epoch": 0.01, "learning_rate": 4.997776047501931e-05, "loss": 0.4242, "step": 13526 }, { "epoch": 0.01, "learning_rate": 4.997775716248323e-05, "loss": 0.4824, "step": 13527 }, { "epoch": 0.01, "learning_rate": 4.997775384970058e-05, "loss": 0.463, "step": 13528 }, { "epoch": 0.01, "learning_rate": 4.9977750536671366e-05, "loss": 0.4151, "step": 13529 }, { "epoch": 0.01, "learning_rate": 4.997774722339557e-05, "loss": 0.416, "step": 13530 }, { "epoch": 0.01, "learning_rate": 4.9977743909873216e-05, "loss": 0.4624, "step": 13531 }, { "epoch": 0.01, "learning_rate": 4.9977740596104285e-05, "loss": 0.4563, "step": 13532 }, { "epoch": 0.01, "learning_rate": 4.997773728208879e-05, "loss": 0.4372, "step": 13533 }, { "epoch": 0.01, "learning_rate": 4.997773396782672e-05, "loss": 0.5665, "step": 13534 }, { "epoch": 0.01, "learning_rate": 4.997773065331809e-05, "loss": 0.4993, "step": 13535 }, { "epoch": 0.01, "learning_rate": 4.997772733856288e-05, "loss": 0.4868, "step": 13536 }, { "epoch": 0.01, "learning_rate": 4.99777240235611e-05, "loss": 0.5029, "step": 13537 }, { "epoch": 0.01, "learning_rate": 4.997772070831276e-05, "loss": 0.51, "step": 13538 }, { "epoch": 0.01, "learning_rate": 4.997771739281784e-05, "loss": 0.5196, "step": 13539 }, { "epoch": 0.01, "learning_rate": 4.997771407707635e-05, "loss": 0.4814, "step": 13540 }, { "epoch": 0.01, "learning_rate": 4.9977710761088304e-05, "loss": 0.4599, "step": 13541 }, { "epoch": 0.01, "learning_rate": 4.997770744485368e-05, "loss": 0.4118, "step": 13542 }, { "epoch": 0.01, "learning_rate": 4.9977704128372483e-05, "loss": 0.4171, "step": 13543 }, { "epoch": 0.01, "learning_rate": 4.9977700811644725e-05, "loss": 0.4187, "step": 13544 }, { "epoch": 0.01, "learning_rate": 4.997769749467039e-05, "loss": 0.4279, "step": 13545 }, { "epoch": 0.01, "learning_rate": 4.997769417744949e-05, "loss": 0.4406, "step": 13546 }, { "epoch": 0.01, "learning_rate": 4.997769085998202e-05, "loss": 0.4483, "step": 13547 }, { "epoch": 0.01, "learning_rate": 4.9977687542267984e-05, "loss": 0.4418, "step": 13548 }, { "epoch": 0.01, "learning_rate": 4.997768422430737e-05, "loss": 0.4313, "step": 13549 }, { "epoch": 0.01, "learning_rate": 4.997768090610019e-05, "loss": 0.4078, "step": 13550 }, { "epoch": 0.01, "learning_rate": 4.9977677587646446e-05, "loss": 0.4018, "step": 13551 }, { "epoch": 0.01, "learning_rate": 4.997767426894613e-05, "loss": 0.5116, "step": 13552 }, { "epoch": 0.01, "learning_rate": 4.997767094999925e-05, "loss": 0.7394, "step": 13553 }, { "epoch": 0.01, "learning_rate": 4.997766763080579e-05, "loss": 0.5174, "step": 13554 }, { "epoch": 0.01, "learning_rate": 4.997766431136577e-05, "loss": 0.4871, "step": 13555 }, { "epoch": 0.01, "learning_rate": 4.9977660991679176e-05, "loss": 0.7955, "step": 13556 }, { "epoch": 0.01, "learning_rate": 4.9977657671746014e-05, "loss": 1.3388, "step": 13557 }, { "epoch": 0.01, "learning_rate": 4.9977654351566285e-05, "loss": 0.621, "step": 13558 }, { "epoch": 0.01, "learning_rate": 4.997765103113998e-05, "loss": 0.3241, "step": 13559 }, { "epoch": 0.01, "learning_rate": 4.9977647710467115e-05, "loss": 0.6977, "step": 13560 }, { "epoch": 0.01, "learning_rate": 4.997764438954768e-05, "loss": 1.4052, "step": 13561 }, { "epoch": 0.01, "learning_rate": 4.9977641068381674e-05, "loss": 1.2271, "step": 13562 }, { "epoch": 0.01, "learning_rate": 4.99776377469691e-05, "loss": 1.379, "step": 13563 }, { "epoch": 0.01, "learning_rate": 4.997763442530996e-05, "loss": 1.7043, "step": 13564 }, { "epoch": 0.01, "learning_rate": 4.997763110340424e-05, "loss": 1.728, "step": 13565 }, { "epoch": 0.01, "learning_rate": 4.997762778125197e-05, "loss": 1.3264, "step": 13566 }, { "epoch": 0.01, "learning_rate": 4.9977624458853116e-05, "loss": 1.2016, "step": 13567 }, { "epoch": 0.01, "learning_rate": 4.99776211362077e-05, "loss": 0.9734, "step": 13568 }, { "epoch": 0.01, "learning_rate": 4.997761781331571e-05, "loss": 1.1336, "step": 13569 }, { "epoch": 0.01, "learning_rate": 4.997761449017716e-05, "loss": 1.1087, "step": 13570 }, { "epoch": 0.01, "learning_rate": 4.997761116679204e-05, "loss": 1.0512, "step": 13571 }, { "epoch": 0.01, "learning_rate": 4.9977607843160346e-05, "loss": 1.1937, "step": 13572 }, { "epoch": 0.01, "learning_rate": 4.9977604519282085e-05, "loss": 1.1724, "step": 13573 }, { "epoch": 0.01, "learning_rate": 4.9977601195157255e-05, "loss": 1.1913, "step": 13574 }, { "epoch": 0.01, "learning_rate": 4.9977597870785864e-05, "loss": 1.0414, "step": 13575 }, { "epoch": 0.01, "learning_rate": 4.997759454616789e-05, "loss": 1.2591, "step": 13576 }, { "epoch": 0.01, "learning_rate": 4.9977591221303366e-05, "loss": 0.7551, "step": 13577 }, { "epoch": 0.01, "learning_rate": 4.997758789619226e-05, "loss": 1.1312, "step": 13578 }, { "epoch": 0.01, "learning_rate": 4.997758457083459e-05, "loss": 0.9839, "step": 13579 }, { "epoch": 0.01, "learning_rate": 4.997758124523035e-05, "loss": 1.0888, "step": 13580 }, { "epoch": 0.01, "learning_rate": 4.997757791937955e-05, "loss": 1.0604, "step": 13581 }, { "epoch": 0.01, "learning_rate": 4.9977574593282174e-05, "loss": 1.04, "step": 13582 }, { "epoch": 0.01, "learning_rate": 4.997757126693823e-05, "loss": 1.0527, "step": 13583 }, { "epoch": 0.01, "learning_rate": 4.997756794034772e-05, "loss": 0.9416, "step": 13584 }, { "epoch": 0.01, "learning_rate": 4.997756461351064e-05, "loss": 1.3833, "step": 13585 }, { "epoch": 0.01, "learning_rate": 4.9977561286427e-05, "loss": 1.0416, "step": 13586 }, { "epoch": 0.01, "learning_rate": 4.9977557959096785e-05, "loss": 0.8595, "step": 13587 }, { "epoch": 0.01, "learning_rate": 4.997755463152e-05, "loss": 0.4039, "step": 13588 }, { "epoch": 0.01, "learning_rate": 4.997755130369664e-05, "loss": 0.312, "step": 13589 }, { "epoch": 0.01, "learning_rate": 4.9977547975626734e-05, "loss": 0.8537, "step": 13590 }, { "epoch": 0.01, "learning_rate": 4.997754464731025e-05, "loss": 0.9302, "step": 13591 }, { "epoch": 0.01, "learning_rate": 4.997754131874719e-05, "loss": 1.1255, "step": 13592 }, { "epoch": 0.01, "learning_rate": 4.997753798993757e-05, "loss": 1.3073, "step": 13593 }, { "epoch": 0.01, "learning_rate": 4.9977534660881384e-05, "loss": 1.0837, "step": 13594 }, { "epoch": 0.01, "learning_rate": 4.997753133157863e-05, "loss": 0.9998, "step": 13595 }, { "epoch": 0.01, "learning_rate": 4.99775280020293e-05, "loss": 1.1223, "step": 13596 }, { "epoch": 0.01, "learning_rate": 4.9977524672233414e-05, "loss": 0.9514, "step": 13597 }, { "epoch": 0.01, "learning_rate": 4.997752134219095e-05, "loss": 1.4154, "step": 13598 }, { "epoch": 0.01, "learning_rate": 4.997751801190193e-05, "loss": 1.2093, "step": 13599 }, { "epoch": 0.01, "learning_rate": 4.997751468136632e-05, "loss": 1.1492, "step": 13600 }, { "epoch": 0.01, "learning_rate": 4.997751135058416e-05, "loss": 1.0651, "step": 13601 }, { "epoch": 0.01, "learning_rate": 4.997750801955543e-05, "loss": 1.1727, "step": 13602 }, { "epoch": 0.01, "learning_rate": 4.997750468828013e-05, "loss": 1.0718, "step": 13603 }, { "epoch": 0.01, "learning_rate": 4.997750135675827e-05, "loss": 1.1734, "step": 13604 }, { "epoch": 0.01, "learning_rate": 4.9977498024989845e-05, "loss": 1.1093, "step": 13605 }, { "epoch": 0.01, "learning_rate": 4.997749469297484e-05, "loss": 0.9622, "step": 13606 }, { "epoch": 0.01, "learning_rate": 4.997749136071327e-05, "loss": 1.1411, "step": 13607 }, { "epoch": 0.01, "learning_rate": 4.997748802820514e-05, "loss": 1.2939, "step": 13608 }, { "epoch": 0.01, "learning_rate": 4.997748469545044e-05, "loss": 1.1855, "step": 13609 }, { "epoch": 0.01, "learning_rate": 4.9977481362449174e-05, "loss": 1.1944, "step": 13610 }, { "epoch": 0.01, "learning_rate": 4.997747802920133e-05, "loss": 1.3632, "step": 13611 }, { "epoch": 0.01, "learning_rate": 4.9977474695706936e-05, "loss": 0.9715, "step": 13612 }, { "epoch": 0.01, "learning_rate": 4.9977471361965965e-05, "loss": 0.9561, "step": 13613 }, { "epoch": 0.01, "learning_rate": 4.9977468027978426e-05, "loss": 1.3744, "step": 13614 }, { "epoch": 0.01, "learning_rate": 4.9977464693744326e-05, "loss": 1.0924, "step": 13615 }, { "epoch": 0.01, "learning_rate": 4.997746135926366e-05, "loss": 1.1654, "step": 13616 }, { "epoch": 0.01, "learning_rate": 4.9977458024536416e-05, "loss": 1.1529, "step": 13617 }, { "epoch": 0.01, "learning_rate": 4.9977454689562605e-05, "loss": 0.9529, "step": 13618 }, { "epoch": 0.01, "learning_rate": 4.9977451354342233e-05, "loss": 1.2269, "step": 13619 }, { "epoch": 0.01, "learning_rate": 4.99774480188753e-05, "loss": 1.3335, "step": 13620 }, { "epoch": 0.01, "learning_rate": 4.997744468316179e-05, "loss": 1.3038, "step": 13621 }, { "epoch": 0.01, "learning_rate": 4.9977441347201725e-05, "loss": 1.1051, "step": 13622 }, { "epoch": 0.01, "learning_rate": 4.997743801099508e-05, "loss": 1.0032, "step": 13623 }, { "epoch": 0.01, "learning_rate": 4.997743467454188e-05, "loss": 1.327, "step": 13624 }, { "epoch": 0.01, "learning_rate": 4.9977431337842105e-05, "loss": 1.2184, "step": 13625 }, { "epoch": 0.01, "learning_rate": 4.9977428000895765e-05, "loss": 1.1238, "step": 13626 }, { "epoch": 0.01, "learning_rate": 4.9977424663702857e-05, "loss": 1.5121, "step": 13627 }, { "epoch": 0.01, "learning_rate": 4.997742132626339e-05, "loss": 1.0176, "step": 13628 }, { "epoch": 0.01, "learning_rate": 4.9977417988577343e-05, "loss": 0.9828, "step": 13629 }, { "epoch": 0.01, "learning_rate": 4.9977414650644745e-05, "loss": 0.8862, "step": 13630 }, { "epoch": 0.01, "learning_rate": 4.997741131246557e-05, "loss": 0.8507, "step": 13631 }, { "epoch": 0.01, "learning_rate": 4.997740797403984e-05, "loss": 0.8304, "step": 13632 }, { "epoch": 0.01, "learning_rate": 4.997740463536753e-05, "loss": 0.7821, "step": 13633 }, { "epoch": 0.01, "learning_rate": 4.997740129644865e-05, "loss": 0.7177, "step": 13634 }, { "epoch": 0.01, "learning_rate": 4.9977397957283215e-05, "loss": 0.8141, "step": 13635 }, { "epoch": 0.01, "learning_rate": 4.9977394617871216e-05, "loss": 0.684, "step": 13636 }, { "epoch": 0.01, "learning_rate": 4.997739127821264e-05, "loss": 0.7407, "step": 13637 }, { "epoch": 0.01, "learning_rate": 4.997738793830751e-05, "loss": 0.973, "step": 13638 }, { "epoch": 0.01, "learning_rate": 4.9977384598155805e-05, "loss": 1.2165, "step": 13639 }, { "epoch": 0.01, "learning_rate": 4.997738125775754e-05, "loss": 1.1518, "step": 13640 }, { "epoch": 0.01, "learning_rate": 4.9977377917112697e-05, "loss": 1.25, "step": 13641 }, { "epoch": 0.01, "learning_rate": 4.99773745762213e-05, "loss": 0.8807, "step": 13642 }, { "epoch": 0.01, "learning_rate": 4.997737123508333e-05, "loss": 1.2195, "step": 13643 }, { "epoch": 0.01, "learning_rate": 4.99773678936988e-05, "loss": 1.1316, "step": 13644 }, { "epoch": 0.01, "learning_rate": 4.99773645520677e-05, "loss": 1.1911, "step": 13645 }, { "epoch": 0.01, "learning_rate": 4.9977361210190033e-05, "loss": 0.9835, "step": 13646 }, { "epoch": 0.01, "learning_rate": 4.99773578680658e-05, "loss": 0.8384, "step": 13647 }, { "epoch": 0.01, "learning_rate": 4.9977354525695e-05, "loss": 1.0689, "step": 13648 }, { "epoch": 0.01, "learning_rate": 4.997735118307764e-05, "loss": 1.3812, "step": 13649 }, { "epoch": 0.01, "learning_rate": 4.9977347840213704e-05, "loss": 1.0354, "step": 13650 }, { "epoch": 0.01, "learning_rate": 4.997734449710322e-05, "loss": 1.2625, "step": 13651 }, { "epoch": 0.01, "learning_rate": 4.9977341153746155e-05, "loss": 1.1085, "step": 13652 }, { "epoch": 0.01, "learning_rate": 4.997733781014253e-05, "loss": 0.6286, "step": 13653 }, { "epoch": 0.01, "learning_rate": 4.9977334466292335e-05, "loss": 1.4704, "step": 13654 }, { "epoch": 0.01, "learning_rate": 4.997733112219557e-05, "loss": 0.4904, "step": 13655 }, { "epoch": 0.01, "learning_rate": 4.997732777785225e-05, "loss": 0.2927, "step": 13656 }, { "epoch": 0.01, "learning_rate": 4.997732443326236e-05, "loss": 0.2228, "step": 13657 }, { "epoch": 0.01, "learning_rate": 4.997732108842591e-05, "loss": 0.7885, "step": 13658 }, { "epoch": 0.01, "learning_rate": 4.997731774334289e-05, "loss": 1.2204, "step": 13659 }, { "epoch": 0.01, "learning_rate": 4.99773143980133e-05, "loss": 1.2167, "step": 13660 }, { "epoch": 0.01, "learning_rate": 4.997731105243715e-05, "loss": 1.1974, "step": 13661 }, { "epoch": 0.01, "learning_rate": 4.9977307706614427e-05, "loss": 1.1804, "step": 13662 }, { "epoch": 0.01, "learning_rate": 4.997730436054515e-05, "loss": 1.1207, "step": 13663 }, { "epoch": 0.01, "learning_rate": 4.9977301014229296e-05, "loss": 0.9005, "step": 13664 }, { "epoch": 0.01, "learning_rate": 4.997729766766688e-05, "loss": 0.9722, "step": 13665 }, { "epoch": 0.01, "learning_rate": 4.997729432085791e-05, "loss": 0.9487, "step": 13666 }, { "epoch": 0.01, "learning_rate": 4.997729097380236e-05, "loss": 1.0496, "step": 13667 }, { "epoch": 0.01, "learning_rate": 4.9977287626500246e-05, "loss": 1.1762, "step": 13668 }, { "epoch": 0.01, "learning_rate": 4.9977284278951575e-05, "loss": 0.9811, "step": 13669 }, { "epoch": 0.01, "learning_rate": 4.9977280931156335e-05, "loss": 0.8552, "step": 13670 }, { "epoch": 0.01, "learning_rate": 4.997727758311453e-05, "loss": 1.0674, "step": 13671 }, { "epoch": 0.01, "learning_rate": 4.997727423482616e-05, "loss": 1.0416, "step": 13672 }, { "epoch": 0.01, "learning_rate": 4.997727088629123e-05, "loss": 1.0856, "step": 13673 }, { "epoch": 0.01, "learning_rate": 4.997726753750972e-05, "loss": 1.3287, "step": 13674 }, { "epoch": 0.01, "learning_rate": 4.997726418848165e-05, "loss": 1.2644, "step": 13675 }, { "epoch": 0.01, "learning_rate": 4.997726083920703e-05, "loss": 0.824, "step": 13676 }, { "epoch": 0.01, "learning_rate": 4.997725748968583e-05, "loss": 1.0309, "step": 13677 }, { "epoch": 0.01, "learning_rate": 4.9977254139918073e-05, "loss": 1.0639, "step": 13678 }, { "epoch": 0.01, "learning_rate": 4.997725078990374e-05, "loss": 1.3336, "step": 13679 }, { "epoch": 0.01, "learning_rate": 4.997724743964285e-05, "loss": 1.3766, "step": 13680 }, { "epoch": 0.01, "learning_rate": 4.99772440891354e-05, "loss": 0.897, "step": 13681 }, { "epoch": 0.01, "learning_rate": 4.997724073838138e-05, "loss": 1.1617, "step": 13682 }, { "epoch": 0.01, "learning_rate": 4.9977237387380796e-05, "loss": 1.2439, "step": 13683 }, { "epoch": 0.01, "learning_rate": 4.997723403613364e-05, "loss": 1.3029, "step": 13684 }, { "epoch": 0.01, "learning_rate": 4.997723068463993e-05, "loss": 1.1788, "step": 13685 }, { "epoch": 0.01, "learning_rate": 4.997722733289964e-05, "loss": 1.1932, "step": 13686 }, { "epoch": 0.01, "learning_rate": 4.99772239809128e-05, "loss": 1.3462, "step": 13687 }, { "epoch": 0.01, "learning_rate": 4.9977220628679395e-05, "loss": 1.4004, "step": 13688 }, { "epoch": 0.01, "learning_rate": 4.9977217276199426e-05, "loss": 1.3485, "step": 13689 }, { "epoch": 0.01, "learning_rate": 4.997721392347289e-05, "loss": 1.3562, "step": 13690 }, { "epoch": 0.01, "learning_rate": 4.9977210570499784e-05, "loss": 1.2233, "step": 13691 }, { "epoch": 0.01, "learning_rate": 4.997720721728011e-05, "loss": 1.0757, "step": 13692 }, { "epoch": 0.01, "learning_rate": 4.9977203863813884e-05, "loss": 0.9933, "step": 13693 }, { "epoch": 0.01, "learning_rate": 4.997720051010108e-05, "loss": 1.1821, "step": 13694 }, { "epoch": 0.01, "learning_rate": 4.997719715614173e-05, "loss": 1.3159, "step": 13695 }, { "epoch": 0.01, "learning_rate": 4.997719380193581e-05, "loss": 1.2139, "step": 13696 }, { "epoch": 0.01, "learning_rate": 4.997719044748331e-05, "loss": 2.1006, "step": 13697 }, { "epoch": 0.01, "learning_rate": 4.9977187092784264e-05, "loss": 1.1615, "step": 13698 }, { "epoch": 0.01, "learning_rate": 4.997718373783864e-05, "loss": 1.2347, "step": 13699 }, { "epoch": 0.01, "learning_rate": 4.997718038264646e-05, "loss": 1.1515, "step": 13700 }, { "epoch": 0.01, "learning_rate": 4.997717702720772e-05, "loss": 0.917, "step": 13701 }, { "epoch": 0.01, "learning_rate": 4.99771736715224e-05, "loss": 0.9968, "step": 13702 }, { "epoch": 0.01, "learning_rate": 4.9977170315590535e-05, "loss": 1.0793, "step": 13703 }, { "epoch": 0.01, "learning_rate": 4.9977166959412095e-05, "loss": 0.9374, "step": 13704 }, { "epoch": 0.01, "learning_rate": 4.9977163602987094e-05, "loss": 1.0369, "step": 13705 }, { "epoch": 0.01, "learning_rate": 4.9977160246315525e-05, "loss": 1.1964, "step": 13706 }, { "epoch": 0.01, "learning_rate": 4.9977156889397394e-05, "loss": 0.9708, "step": 13707 }, { "epoch": 0.01, "learning_rate": 4.99771535322327e-05, "loss": 1.1762, "step": 13708 }, { "epoch": 0.01, "learning_rate": 4.9977150174821444e-05, "loss": 1.1125, "step": 13709 }, { "epoch": 0.01, "learning_rate": 4.9977146817163623e-05, "loss": 1.1945, "step": 13710 }, { "epoch": 0.01, "learning_rate": 4.9977143459259235e-05, "loss": 0.8692, "step": 13711 }, { "epoch": 0.01, "learning_rate": 4.9977140101108286e-05, "loss": 0.5989, "step": 13712 }, { "epoch": 0.01, "learning_rate": 4.9977136742710775e-05, "loss": 0.9482, "step": 13713 }, { "epoch": 0.01, "learning_rate": 4.99771333840667e-05, "loss": 0.6659, "step": 13714 }, { "epoch": 0.01, "learning_rate": 4.997713002517606e-05, "loss": 1.0247, "step": 13715 }, { "epoch": 0.01, "learning_rate": 4.997712666603885e-05, "loss": 1.2016, "step": 13716 }, { "epoch": 0.01, "learning_rate": 4.997712330665508e-05, "loss": 0.8925, "step": 13717 }, { "epoch": 0.01, "learning_rate": 4.9977119947024744e-05, "loss": 1.1947, "step": 13718 }, { "epoch": 0.01, "learning_rate": 4.9977116587147854e-05, "loss": 1.1535, "step": 13719 }, { "epoch": 0.01, "learning_rate": 4.997711322702439e-05, "loss": 1.2114, "step": 13720 }, { "epoch": 0.01, "learning_rate": 4.997710986665437e-05, "loss": 0.9377, "step": 13721 }, { "epoch": 0.01, "learning_rate": 4.997710650603779e-05, "loss": 1.2785, "step": 13722 }, { "epoch": 0.01, "learning_rate": 4.997710314517463e-05, "loss": 1.0253, "step": 13723 }, { "epoch": 0.01, "learning_rate": 4.997709978406492e-05, "loss": 1.0243, "step": 13724 }, { "epoch": 0.01, "learning_rate": 4.9977096422708643e-05, "loss": 1.2286, "step": 13725 }, { "epoch": 0.01, "learning_rate": 4.9977093061105804e-05, "loss": 1.0534, "step": 13726 }, { "epoch": 0.01, "learning_rate": 4.99770896992564e-05, "loss": 1.121, "step": 13727 }, { "epoch": 0.01, "learning_rate": 4.9977086337160436e-05, "loss": 1.4994, "step": 13728 }, { "epoch": 0.01, "learning_rate": 4.99770829748179e-05, "loss": 0.7856, "step": 13729 }, { "epoch": 0.01, "learning_rate": 4.99770796122288e-05, "loss": 0.9881, "step": 13730 }, { "epoch": 0.01, "learning_rate": 4.997707624939315e-05, "loss": 0.8364, "step": 13731 }, { "epoch": 0.01, "learning_rate": 4.997707288631093e-05, "loss": 1.0286, "step": 13732 }, { "epoch": 0.01, "learning_rate": 4.997706952298215e-05, "loss": 1.2626, "step": 13733 }, { "epoch": 0.01, "learning_rate": 4.99770661594068e-05, "loss": 1.2507, "step": 13734 }, { "epoch": 0.01, "learning_rate": 4.9977062795584893e-05, "loss": 1.246, "step": 13735 }, { "epoch": 0.01, "learning_rate": 4.997705943151643e-05, "loss": 1.1015, "step": 13736 }, { "epoch": 0.01, "learning_rate": 4.997705606720139e-05, "loss": 0.9253, "step": 13737 }, { "epoch": 0.01, "learning_rate": 4.997705270263979e-05, "loss": 1.2107, "step": 13738 }, { "epoch": 0.01, "learning_rate": 4.997704933783163e-05, "loss": 1.1612, "step": 13739 }, { "epoch": 0.01, "learning_rate": 4.99770459727769e-05, "loss": 1.1212, "step": 13740 }, { "epoch": 0.01, "learning_rate": 4.997704260747562e-05, "loss": 1.1298, "step": 13741 }, { "epoch": 0.01, "learning_rate": 4.997703924192777e-05, "loss": 1.1805, "step": 13742 }, { "epoch": 0.01, "learning_rate": 4.997703587613336e-05, "loss": 0.9882, "step": 13743 }, { "epoch": 0.01, "learning_rate": 4.997703251009238e-05, "loss": 0.9589, "step": 13744 }, { "epoch": 0.01, "learning_rate": 4.997702914380484e-05, "loss": 1.1904, "step": 13745 }, { "epoch": 0.01, "learning_rate": 4.9977025777270746e-05, "loss": 1.1239, "step": 13746 }, { "epoch": 0.01, "learning_rate": 4.997702241049008e-05, "loss": 1.209, "step": 13747 }, { "epoch": 0.01, "learning_rate": 4.997701904346286e-05, "loss": 0.9033, "step": 13748 }, { "epoch": 0.01, "learning_rate": 4.9977015676189065e-05, "loss": 0.9875, "step": 13749 }, { "epoch": 0.01, "learning_rate": 4.997701230866871e-05, "loss": 1.0884, "step": 13750 }, { "epoch": 0.01, "learning_rate": 4.9977008940901804e-05, "loss": 1.3568, "step": 13751 }, { "epoch": 0.01, "learning_rate": 4.997700557288832e-05, "loss": 1.1509, "step": 13752 }, { "epoch": 0.01, "learning_rate": 4.9977002204628284e-05, "loss": 1.2721, "step": 13753 }, { "epoch": 0.01, "learning_rate": 4.997699883612169e-05, "loss": 1.1617, "step": 13754 }, { "epoch": 0.01, "learning_rate": 4.997699546736853e-05, "loss": 1.172, "step": 13755 }, { "epoch": 0.01, "learning_rate": 4.99769920983688e-05, "loss": 1.154, "step": 13756 }, { "epoch": 0.01, "learning_rate": 4.997698872912251e-05, "loss": 1.2378, "step": 13757 }, { "epoch": 0.01, "learning_rate": 4.9976985359629656e-05, "loss": 1.0012, "step": 13758 }, { "epoch": 0.01, "learning_rate": 4.9976981989890245e-05, "loss": 1.0612, "step": 13759 }, { "epoch": 0.01, "learning_rate": 4.997697861990427e-05, "loss": 1.1235, "step": 13760 }, { "epoch": 0.01, "learning_rate": 4.9976975249671734e-05, "loss": 1.1316, "step": 13761 }, { "epoch": 0.01, "learning_rate": 4.9976971879192633e-05, "loss": 0.9852, "step": 13762 }, { "epoch": 0.01, "learning_rate": 4.997696850846697e-05, "loss": 1.066, "step": 13763 }, { "epoch": 0.01, "learning_rate": 4.9976965137494756e-05, "loss": 1.265, "step": 13764 }, { "epoch": 0.01, "learning_rate": 4.9976961766275965e-05, "loss": 1.0337, "step": 13765 }, { "epoch": 0.01, "learning_rate": 4.997695839481062e-05, "loss": 1.1193, "step": 13766 }, { "epoch": 0.01, "learning_rate": 4.997695502309871e-05, "loss": 1.2403, "step": 13767 }, { "epoch": 0.01, "learning_rate": 4.997695165114023e-05, "loss": 1.0786, "step": 13768 }, { "epoch": 0.01, "learning_rate": 4.9976948278935204e-05, "loss": 0.9102, "step": 13769 }, { "epoch": 0.01, "learning_rate": 4.997694490648361e-05, "loss": 1.4613, "step": 13770 }, { "epoch": 0.01, "learning_rate": 4.997694153378545e-05, "loss": 1.3508, "step": 13771 }, { "epoch": 0.01, "learning_rate": 4.997693816084073e-05, "loss": 1.226, "step": 13772 }, { "epoch": 0.01, "learning_rate": 4.997693478764945e-05, "loss": 1.1583, "step": 13773 }, { "epoch": 0.01, "learning_rate": 4.9976931414211604e-05, "loss": 0.9677, "step": 13774 }, { "epoch": 0.01, "learning_rate": 4.99769280405272e-05, "loss": 1.0938, "step": 13775 }, { "epoch": 0.01, "learning_rate": 4.997692466659624e-05, "loss": 1.1531, "step": 13776 }, { "epoch": 0.01, "learning_rate": 4.997692129241871e-05, "loss": 1.0068, "step": 13777 }, { "epoch": 0.01, "learning_rate": 4.9976917917994615e-05, "loss": 1.1943, "step": 13778 }, { "epoch": 0.01, "learning_rate": 4.997691454332397e-05, "loss": 1.0498, "step": 13779 }, { "epoch": 0.01, "learning_rate": 4.9976911168406754e-05, "loss": 1.2163, "step": 13780 }, { "epoch": 0.01, "learning_rate": 4.997690779324298e-05, "loss": 1.2518, "step": 13781 }, { "epoch": 0.01, "learning_rate": 4.997690441783265e-05, "loss": 1.278, "step": 13782 }, { "epoch": 0.01, "learning_rate": 4.997690104217575e-05, "loss": 1.1028, "step": 13783 }, { "epoch": 0.01, "learning_rate": 4.9976897666272285e-05, "loss": 0.6773, "step": 13784 }, { "epoch": 0.01, "learning_rate": 4.9976894290122265e-05, "loss": 1.1019, "step": 13785 }, { "epoch": 0.01, "learning_rate": 4.997689091372569e-05, "loss": 1.1219, "step": 13786 }, { "epoch": 0.01, "learning_rate": 4.997688753708255e-05, "loss": 1.1479, "step": 13787 }, { "epoch": 0.01, "learning_rate": 4.997688416019284e-05, "loss": 1.0981, "step": 13788 }, { "epoch": 0.01, "learning_rate": 4.997688078305658e-05, "loss": 0.9778, "step": 13789 }, { "epoch": 0.01, "learning_rate": 4.997687740567375e-05, "loss": 1.0922, "step": 13790 }, { "epoch": 0.01, "learning_rate": 4.997687402804436e-05, "loss": 1.0741, "step": 13791 }, { "epoch": 0.01, "learning_rate": 4.997687065016841e-05, "loss": 0.9276, "step": 13792 }, { "epoch": 0.01, "learning_rate": 4.9976867272045904e-05, "loss": 1.2373, "step": 13793 }, { "epoch": 0.01, "learning_rate": 4.997686389367683e-05, "loss": 0.9606, "step": 13794 }, { "epoch": 0.01, "learning_rate": 4.9976860515061196e-05, "loss": 1.0542, "step": 13795 }, { "epoch": 0.01, "learning_rate": 4.997685713619901e-05, "loss": 1.1851, "step": 13796 }, { "epoch": 0.01, "learning_rate": 4.997685375709025e-05, "loss": 1.0827, "step": 13797 }, { "epoch": 0.01, "learning_rate": 4.9976850377734935e-05, "loss": 0.9717, "step": 13798 }, { "epoch": 0.01, "learning_rate": 4.997684699813306e-05, "loss": 1.2958, "step": 13799 }, { "epoch": 0.01, "learning_rate": 4.997684361828462e-05, "loss": 0.897, "step": 13800 }, { "epoch": 0.01, "learning_rate": 4.9976840238189624e-05, "loss": 1.0937, "step": 13801 }, { "epoch": 0.01, "learning_rate": 4.9976836857848065e-05, "loss": 0.9241, "step": 13802 }, { "epoch": 0.01, "learning_rate": 4.9976833477259944e-05, "loss": 0.94, "step": 13803 }, { "epoch": 0.01, "learning_rate": 4.997683009642526e-05, "loss": 1.1254, "step": 13804 }, { "epoch": 0.01, "learning_rate": 4.997682671534402e-05, "loss": 0.9907, "step": 13805 }, { "epoch": 0.01, "learning_rate": 4.997682333401622e-05, "loss": 1.2535, "step": 13806 }, { "epoch": 0.01, "learning_rate": 4.997681995244186e-05, "loss": 1.0171, "step": 13807 }, { "epoch": 0.01, "learning_rate": 4.997681657062093e-05, "loss": 0.9388, "step": 13808 }, { "epoch": 0.01, "learning_rate": 4.997681318855345e-05, "loss": 0.8717, "step": 13809 }, { "epoch": 0.01, "learning_rate": 4.9976809806239404e-05, "loss": 1.3064, "step": 13810 }, { "epoch": 0.01, "learning_rate": 4.99768064236788e-05, "loss": 1.0957, "step": 13811 }, { "epoch": 0.01, "learning_rate": 4.997680304087163e-05, "loss": 1.0988, "step": 13812 }, { "epoch": 0.01, "learning_rate": 4.9976799657817906e-05, "loss": 1.0316, "step": 13813 }, { "epoch": 0.01, "learning_rate": 4.9976796274517627e-05, "loss": 0.9699, "step": 13814 }, { "epoch": 0.01, "learning_rate": 4.997679289097077e-05, "loss": 0.864, "step": 13815 }, { "epoch": 0.01, "learning_rate": 4.997678950717737e-05, "loss": 0.9019, "step": 13816 }, { "epoch": 0.01, "learning_rate": 4.9976786123137395e-05, "loss": 1.2512, "step": 13817 }, { "epoch": 0.01, "learning_rate": 4.997678273885087e-05, "loss": 1.2744, "step": 13818 }, { "epoch": 0.01, "learning_rate": 4.997677935431778e-05, "loss": 1.0352, "step": 13819 }, { "epoch": 0.01, "learning_rate": 4.9976775969538134e-05, "loss": 1.2294, "step": 13820 }, { "epoch": 0.01, "learning_rate": 4.997677258451193e-05, "loss": 1.4391, "step": 13821 }, { "epoch": 0.01, "learning_rate": 4.997676919923916e-05, "loss": 0.9864, "step": 13822 }, { "epoch": 0.01, "learning_rate": 4.997676581371983e-05, "loss": 1.2285, "step": 13823 }, { "epoch": 0.01, "learning_rate": 4.997676242795394e-05, "loss": 1.2572, "step": 13824 }, { "epoch": 0.01, "learning_rate": 4.997675904194149e-05, "loss": 1.0138, "step": 13825 }, { "epoch": 0.01, "learning_rate": 4.997675565568248e-05, "loss": 0.3226, "step": 13826 }, { "epoch": 0.01, "learning_rate": 4.9976752269176914e-05, "loss": 1.0268, "step": 13827 }, { "epoch": 0.01, "learning_rate": 4.9976748882424786e-05, "loss": 1.1284, "step": 13828 }, { "epoch": 0.01, "learning_rate": 4.997674549542609e-05, "loss": 1.2994, "step": 13829 }, { "epoch": 0.01, "learning_rate": 4.9976742108180846e-05, "loss": 0.5558, "step": 13830 }, { "epoch": 0.01, "learning_rate": 4.9976738720689034e-05, "loss": 0.984, "step": 13831 }, { "epoch": 0.01, "learning_rate": 4.997673533295067e-05, "loss": 1.2748, "step": 13832 }, { "epoch": 0.01, "learning_rate": 4.9976731944965735e-05, "loss": 1.2244, "step": 13833 }, { "epoch": 0.01, "learning_rate": 4.997672855673425e-05, "loss": 1.4405, "step": 13834 }, { "epoch": 0.01, "learning_rate": 4.99767251682562e-05, "loss": 1.2221, "step": 13835 }, { "epoch": 0.01, "learning_rate": 4.99767217795316e-05, "loss": 1.0759, "step": 13836 }, { "epoch": 0.01, "learning_rate": 4.997671839056043e-05, "loss": 0.7912, "step": 13837 }, { "epoch": 0.01, "learning_rate": 4.99767150013427e-05, "loss": 0.9559, "step": 13838 }, { "epoch": 0.01, "learning_rate": 4.9976711611878416e-05, "loss": 1.0427, "step": 13839 }, { "epoch": 0.01, "learning_rate": 4.997670822216757e-05, "loss": 1.1443, "step": 13840 }, { "epoch": 0.01, "learning_rate": 4.9976704832210165e-05, "loss": 0.9395, "step": 13841 }, { "epoch": 0.01, "learning_rate": 4.9976701442006195e-05, "loss": 1.1513, "step": 13842 }, { "epoch": 0.01, "learning_rate": 4.997669805155567e-05, "loss": 1.2022, "step": 13843 }, { "epoch": 0.01, "learning_rate": 4.997669466085859e-05, "loss": 1.0524, "step": 13844 }, { "epoch": 0.01, "learning_rate": 4.9976691269914944e-05, "loss": 1.107, "step": 13845 }, { "epoch": 0.01, "learning_rate": 4.997668787872474e-05, "loss": 0.8735, "step": 13846 }, { "epoch": 0.01, "learning_rate": 4.997668448728799e-05, "loss": 1.0215, "step": 13847 }, { "epoch": 0.01, "learning_rate": 4.997668109560466e-05, "loss": 1.1086, "step": 13848 }, { "epoch": 0.01, "learning_rate": 4.997667770367478e-05, "loss": 1.0567, "step": 13849 }, { "epoch": 0.01, "learning_rate": 4.997667431149834e-05, "loss": 1.3272, "step": 13850 }, { "epoch": 0.01, "learning_rate": 4.997667091907534e-05, "loss": 1.423, "step": 13851 }, { "epoch": 0.01, "learning_rate": 4.997666752640578e-05, "loss": 0.9504, "step": 13852 }, { "epoch": 0.01, "learning_rate": 4.997666413348966e-05, "loss": 1.1183, "step": 13853 }, { "epoch": 0.01, "learning_rate": 4.9976660740326985e-05, "loss": 0.8809, "step": 13854 }, { "epoch": 0.01, "learning_rate": 4.997665734691775e-05, "loss": 1.2235, "step": 13855 }, { "epoch": 0.01, "learning_rate": 4.9976653953261955e-05, "loss": 1.0271, "step": 13856 }, { "epoch": 0.01, "learning_rate": 4.99766505593596e-05, "loss": 1.1365, "step": 13857 }, { "epoch": 0.01, "learning_rate": 4.997664716521069e-05, "loss": 1.0548, "step": 13858 }, { "epoch": 0.01, "learning_rate": 4.997664377081521e-05, "loss": 1.1949, "step": 13859 }, { "epoch": 0.01, "learning_rate": 4.997664037617319e-05, "loss": 0.6866, "step": 13860 }, { "epoch": 0.01, "learning_rate": 4.99766369812846e-05, "loss": 0.8584, "step": 13861 }, { "epoch": 0.01, "learning_rate": 4.9976633586149444e-05, "loss": 1.0817, "step": 13862 }, { "epoch": 0.01, "learning_rate": 4.997663019076774e-05, "loss": 1.0412, "step": 13863 }, { "epoch": 0.01, "learning_rate": 4.997662679513948e-05, "loss": 0.7251, "step": 13864 }, { "epoch": 0.01, "learning_rate": 4.997662339926465e-05, "loss": 1.0131, "step": 13865 }, { "epoch": 0.01, "learning_rate": 4.9976620003143265e-05, "loss": 1.0863, "step": 13866 }, { "epoch": 0.01, "learning_rate": 4.997661660677533e-05, "loss": 1.1812, "step": 13867 }, { "epoch": 0.01, "learning_rate": 4.997661321016083e-05, "loss": 0.8374, "step": 13868 }, { "epoch": 0.01, "learning_rate": 4.997660981329977e-05, "loss": 1.0997, "step": 13869 }, { "epoch": 0.01, "learning_rate": 4.9976606416192154e-05, "loss": 1.0978, "step": 13870 }, { "epoch": 0.01, "learning_rate": 4.997660301883797e-05, "loss": 1.1241, "step": 13871 }, { "epoch": 0.01, "learning_rate": 4.9976599621237236e-05, "loss": 0.9462, "step": 13872 }, { "epoch": 0.01, "learning_rate": 4.9976596223389945e-05, "loss": 1.0272, "step": 13873 }, { "epoch": 0.01, "learning_rate": 4.9976592825296086e-05, "loss": 1.0391, "step": 13874 }, { "epoch": 0.01, "learning_rate": 4.997658942695568e-05, "loss": 1.1377, "step": 13875 }, { "epoch": 0.01, "learning_rate": 4.997658602836871e-05, "loss": 0.995, "step": 13876 }, { "epoch": 0.01, "learning_rate": 4.9976582629535184e-05, "loss": 1.0341, "step": 13877 }, { "epoch": 0.01, "learning_rate": 4.99765792304551e-05, "loss": 1.0076, "step": 13878 }, { "epoch": 0.01, "learning_rate": 4.997657583112846e-05, "loss": 0.9456, "step": 13879 }, { "epoch": 0.01, "learning_rate": 4.997657243155526e-05, "loss": 1.1291, "step": 13880 }, { "epoch": 0.01, "learning_rate": 4.99765690317355e-05, "loss": 0.8903, "step": 13881 }, { "epoch": 0.01, "learning_rate": 4.997656563166918e-05, "loss": 1.1877, "step": 13882 }, { "epoch": 0.01, "learning_rate": 4.99765622313563e-05, "loss": 1.0815, "step": 13883 }, { "epoch": 0.01, "learning_rate": 4.997655883079687e-05, "loss": 0.9905, "step": 13884 }, { "epoch": 0.01, "learning_rate": 4.997655542999087e-05, "loss": 1.0751, "step": 13885 }, { "epoch": 0.01, "learning_rate": 4.9976552028938326e-05, "loss": 0.8339, "step": 13886 }, { "epoch": 0.01, "learning_rate": 4.997654862763922e-05, "loss": 1.188, "step": 13887 }, { "epoch": 0.01, "learning_rate": 4.997654522609355e-05, "loss": 1.1361, "step": 13888 }, { "epoch": 0.01, "learning_rate": 4.9976541824301324e-05, "loss": 1.0927, "step": 13889 }, { "epoch": 0.01, "learning_rate": 4.997653842226254e-05, "loss": 1.1153, "step": 13890 }, { "epoch": 0.01, "learning_rate": 4.99765350199772e-05, "loss": 1.0066, "step": 13891 }, { "epoch": 0.01, "learning_rate": 4.997653161744531e-05, "loss": 0.7331, "step": 13892 }, { "epoch": 0.01, "learning_rate": 4.997652821466685e-05, "loss": 0.8633, "step": 13893 }, { "epoch": 0.01, "learning_rate": 4.9976524811641834e-05, "loss": 1.037, "step": 13894 }, { "epoch": 0.01, "learning_rate": 4.997652140837027e-05, "loss": 0.7927, "step": 13895 }, { "epoch": 0.01, "learning_rate": 4.997651800485214e-05, "loss": 0.9617, "step": 13896 }, { "epoch": 0.01, "learning_rate": 4.9976514601087454e-05, "loss": 0.7226, "step": 13897 }, { "epoch": 0.01, "learning_rate": 4.99765111970762e-05, "loss": 1.095, "step": 13898 }, { "epoch": 0.01, "learning_rate": 4.9976507792818404e-05, "loss": 1.0521, "step": 13899 }, { "epoch": 0.01, "learning_rate": 4.9976504388314044e-05, "loss": 1.1347, "step": 13900 }, { "epoch": 0.01, "learning_rate": 4.997650098356313e-05, "loss": 1.0732, "step": 13901 }, { "epoch": 0.01, "learning_rate": 4.997649757856565e-05, "loss": 1.0931, "step": 13902 }, { "epoch": 0.01, "learning_rate": 4.997649417332162e-05, "loss": 1.1045, "step": 13903 }, { "epoch": 0.01, "learning_rate": 4.997649076783103e-05, "loss": 1.1639, "step": 13904 }, { "epoch": 0.01, "learning_rate": 4.997648736209389e-05, "loss": 0.9362, "step": 13905 }, { "epoch": 0.01, "learning_rate": 4.997648395611018e-05, "loss": 1.2209, "step": 13906 }, { "epoch": 0.01, "learning_rate": 4.997648054987992e-05, "loss": 1.1402, "step": 13907 }, { "epoch": 0.01, "learning_rate": 4.99764771434031e-05, "loss": 1.189, "step": 13908 }, { "epoch": 0.01, "learning_rate": 4.997647373667973e-05, "loss": 1.1464, "step": 13909 }, { "epoch": 0.01, "learning_rate": 4.9976470329709794e-05, "loss": 1.0142, "step": 13910 }, { "epoch": 0.01, "learning_rate": 4.99764669224933e-05, "loss": 1.0851, "step": 13911 }, { "epoch": 0.01, "learning_rate": 4.9976463515030256e-05, "loss": 1.3289, "step": 13912 }, { "epoch": 0.01, "learning_rate": 4.9976460107320656e-05, "loss": 1.1941, "step": 13913 }, { "epoch": 0.01, "learning_rate": 4.9976456699364494e-05, "loss": 0.8055, "step": 13914 }, { "epoch": 0.01, "learning_rate": 4.997645329116177e-05, "loss": 0.929, "step": 13915 }, { "epoch": 0.01, "learning_rate": 4.9976449882712495e-05, "loss": 1.5083, "step": 13916 }, { "epoch": 0.01, "learning_rate": 4.997644647401666e-05, "loss": 1.3804, "step": 13917 }, { "epoch": 0.01, "learning_rate": 4.997644306507428e-05, "loss": 1.2218, "step": 13918 }, { "epoch": 0.01, "learning_rate": 4.9976439655885324e-05, "loss": 1.1761, "step": 13919 }, { "epoch": 0.01, "learning_rate": 4.997643624644982e-05, "loss": 0.9478, "step": 13920 }, { "epoch": 0.01, "learning_rate": 4.997643283676776e-05, "loss": 1.2176, "step": 13921 }, { "epoch": 0.01, "learning_rate": 4.9976429426839144e-05, "loss": 1.4068, "step": 13922 }, { "epoch": 0.01, "learning_rate": 4.997642601666397e-05, "loss": 1.0524, "step": 13923 }, { "epoch": 0.01, "learning_rate": 4.997642260624225e-05, "loss": 1.092, "step": 13924 }, { "epoch": 0.01, "learning_rate": 4.9976419195573955e-05, "loss": 0.937, "step": 13925 }, { "epoch": 0.01, "learning_rate": 4.9976415784659114e-05, "loss": 1.1258, "step": 13926 }, { "epoch": 0.01, "learning_rate": 4.997641237349771e-05, "loss": 1.0027, "step": 13927 }, { "epoch": 0.01, "learning_rate": 4.9976408962089756e-05, "loss": 1.1325, "step": 13928 }, { "epoch": 0.01, "learning_rate": 4.997640555043524e-05, "loss": 1.1987, "step": 13929 }, { "epoch": 0.01, "learning_rate": 4.9976402138534173e-05, "loss": 1.3612, "step": 13930 }, { "epoch": 0.01, "learning_rate": 4.997639872638654e-05, "loss": 1.2419, "step": 13931 }, { "epoch": 0.01, "learning_rate": 4.997639531399236e-05, "loss": 1.1688, "step": 13932 }, { "epoch": 0.01, "learning_rate": 4.997639190135162e-05, "loss": 1.1912, "step": 13933 }, { "epoch": 0.01, "learning_rate": 4.997638848846432e-05, "loss": 1.076, "step": 13934 }, { "epoch": 0.01, "learning_rate": 4.9976385075330466e-05, "loss": 1.4147, "step": 13935 }, { "epoch": 0.01, "learning_rate": 4.997638166195006e-05, "loss": 0.9483, "step": 13936 }, { "epoch": 0.01, "learning_rate": 4.9976378248323095e-05, "loss": 1.2198, "step": 13937 }, { "epoch": 0.01, "learning_rate": 4.9976374834449575e-05, "loss": 0.9229, "step": 13938 }, { "epoch": 0.01, "learning_rate": 4.9976371420329494e-05, "loss": 1.1425, "step": 13939 }, { "epoch": 0.01, "learning_rate": 4.997636800596286e-05, "loss": 1.1703, "step": 13940 }, { "epoch": 0.01, "learning_rate": 4.9976364591349675e-05, "loss": 1.1516, "step": 13941 }, { "epoch": 0.01, "learning_rate": 4.9976361176489924e-05, "loss": 0.9866, "step": 13942 }, { "epoch": 0.01, "learning_rate": 4.9976357761383626e-05, "loss": 1.1424, "step": 13943 }, { "epoch": 0.01, "learning_rate": 4.9976354346030766e-05, "loss": 1.1475, "step": 13944 }, { "epoch": 0.01, "learning_rate": 4.9976350930431345e-05, "loss": 0.7793, "step": 13945 }, { "epoch": 0.01, "learning_rate": 4.9976347514585377e-05, "loss": 0.8283, "step": 13946 }, { "epoch": 0.01, "learning_rate": 4.9976344098492854e-05, "loss": 1.0096, "step": 13947 }, { "epoch": 0.01, "learning_rate": 4.997634068215376e-05, "loss": 0.7455, "step": 13948 }, { "epoch": 0.01, "learning_rate": 4.997633726556813e-05, "loss": 1.2219, "step": 13949 }, { "epoch": 0.01, "learning_rate": 4.997633384873593e-05, "loss": 1.2689, "step": 13950 }, { "epoch": 0.01, "learning_rate": 4.9976330431657185e-05, "loss": 0.9038, "step": 13951 }, { "epoch": 0.01, "learning_rate": 4.9976327014331884e-05, "loss": 0.9423, "step": 13952 }, { "epoch": 0.01, "learning_rate": 4.997632359676002e-05, "loss": 1.2045, "step": 13953 }, { "epoch": 0.01, "learning_rate": 4.9976320178941605e-05, "loss": 1.2527, "step": 13954 }, { "epoch": 0.01, "learning_rate": 4.997631676087663e-05, "loss": 1.2215, "step": 13955 }, { "epoch": 0.01, "learning_rate": 4.99763133425651e-05, "loss": 0.9627, "step": 13956 }, { "epoch": 0.01, "learning_rate": 4.9976309924007015e-05, "loss": 1.0087, "step": 13957 }, { "epoch": 0.01, "learning_rate": 4.9976306505202374e-05, "loss": 0.7504, "step": 13958 }, { "epoch": 0.01, "learning_rate": 4.997630308615118e-05, "loss": 1.0684, "step": 13959 }, { "epoch": 0.01, "learning_rate": 4.997629966685343e-05, "loss": 1.1437, "step": 13960 }, { "epoch": 0.01, "learning_rate": 4.997629624730912e-05, "loss": 1.1618, "step": 13961 }, { "epoch": 0.01, "learning_rate": 4.997629282751826e-05, "loss": 1.2139, "step": 13962 }, { "epoch": 0.01, "learning_rate": 4.997628940748084e-05, "loss": 1.0534, "step": 13963 }, { "epoch": 0.01, "learning_rate": 4.9976285987196866e-05, "loss": 1.0269, "step": 13964 }, { "epoch": 0.01, "learning_rate": 4.9976282566666345e-05, "loss": 1.1003, "step": 13965 }, { "epoch": 0.01, "learning_rate": 4.997627914588926e-05, "loss": 1.0463, "step": 13966 }, { "epoch": 0.01, "learning_rate": 4.997627572486562e-05, "loss": 1.1097, "step": 13967 }, { "epoch": 0.01, "learning_rate": 4.997627230359542e-05, "loss": 1.1212, "step": 13968 }, { "epoch": 0.01, "learning_rate": 4.997626888207867e-05, "loss": 1.0862, "step": 13969 }, { "epoch": 0.01, "learning_rate": 4.997626546031537e-05, "loss": 1.2666, "step": 13970 }, { "epoch": 0.01, "learning_rate": 4.997626203830551e-05, "loss": 1.1089, "step": 13971 }, { "epoch": 0.01, "learning_rate": 4.997625861604909e-05, "loss": 1.1409, "step": 13972 }, { "epoch": 0.01, "learning_rate": 4.997625519354612e-05, "loss": 1.0216, "step": 13973 }, { "epoch": 0.01, "learning_rate": 4.99762517707966e-05, "loss": 1.1534, "step": 13974 }, { "epoch": 0.01, "learning_rate": 4.997624834780051e-05, "loss": 1.0097, "step": 13975 }, { "epoch": 0.01, "learning_rate": 4.997624492455788e-05, "loss": 0.8927, "step": 13976 }, { "epoch": 0.01, "learning_rate": 4.9976241501068686e-05, "loss": 1.0883, "step": 13977 }, { "epoch": 0.01, "learning_rate": 4.997623807733295e-05, "loss": 1.0862, "step": 13978 }, { "epoch": 0.01, "learning_rate": 4.997623465335065e-05, "loss": 1.1607, "step": 13979 }, { "epoch": 0.01, "learning_rate": 4.99762312291218e-05, "loss": 1.1474, "step": 13980 }, { "epoch": 0.01, "learning_rate": 4.997622780464638e-05, "loss": 1.0373, "step": 13981 }, { "epoch": 0.01, "learning_rate": 4.997622437992442e-05, "loss": 0.5321, "step": 13982 }, { "epoch": 0.01, "learning_rate": 4.99762209549559e-05, "loss": 0.8732, "step": 13983 }, { "epoch": 0.01, "learning_rate": 4.997621752974082e-05, "loss": 1.2515, "step": 13984 }, { "epoch": 0.01, "learning_rate": 4.99762141042792e-05, "loss": 1.18, "step": 13985 }, { "epoch": 0.01, "learning_rate": 4.9976210678571015e-05, "loss": 0.95, "step": 13986 }, { "epoch": 0.01, "learning_rate": 4.997620725261627e-05, "loss": 1.0746, "step": 13987 }, { "epoch": 0.01, "learning_rate": 4.997620382641498e-05, "loss": 0.9309, "step": 13988 }, { "epoch": 0.01, "learning_rate": 4.997620039996713e-05, "loss": 1.0832, "step": 13989 }, { "epoch": 0.01, "learning_rate": 4.997619697327273e-05, "loss": 1.0483, "step": 13990 }, { "epoch": 0.01, "learning_rate": 4.997619354633177e-05, "loss": 1.0112, "step": 13991 }, { "epoch": 0.01, "learning_rate": 4.997619011914426e-05, "loss": 1.1311, "step": 13992 }, { "epoch": 0.01, "learning_rate": 4.9976186691710204e-05, "loss": 1.0354, "step": 13993 }, { "epoch": 0.01, "learning_rate": 4.997618326402958e-05, "loss": 1.3114, "step": 13994 }, { "epoch": 0.01, "learning_rate": 4.997617983610241e-05, "loss": 1.0396, "step": 13995 }, { "epoch": 0.01, "learning_rate": 4.9976176407928674e-05, "loss": 1.159, "step": 13996 }, { "epoch": 0.01, "learning_rate": 4.997617297950839e-05, "loss": 0.9205, "step": 13997 }, { "epoch": 0.01, "learning_rate": 4.997616955084156e-05, "loss": 1.0981, "step": 13998 }, { "epoch": 0.01, "learning_rate": 4.997616612192817e-05, "loss": 1.3084, "step": 13999 }, { "epoch": 0.01, "learning_rate": 4.9976162692768224e-05, "loss": 1.3947, "step": 14000 }, { "epoch": 0.01, "eval_loss": 1.0414711236953735, "eval_runtime": 85.5056, "eval_samples_per_second": 16.198, "eval_steps_per_second": 4.058, "step": 14000 }, { "epoch": 0.01, "learning_rate": 4.997615926336172e-05, "loss": 1.2238, "step": 14001 }, { "epoch": 0.01, "learning_rate": 4.997615583370867e-05, "loss": 1.2037, "step": 14002 }, { "epoch": 0.01, "learning_rate": 4.997615240380906e-05, "loss": 1.175, "step": 14003 }, { "epoch": 0.01, "learning_rate": 4.99761489736629e-05, "loss": 1.0471, "step": 14004 }, { "epoch": 0.01, "learning_rate": 4.997614554327019e-05, "loss": 0.9915, "step": 14005 }, { "epoch": 0.01, "learning_rate": 4.997614211263092e-05, "loss": 0.9342, "step": 14006 }, { "epoch": 0.01, "learning_rate": 4.9976138681745094e-05, "loss": 1.1484, "step": 14007 }, { "epoch": 0.01, "learning_rate": 4.9976135250612715e-05, "loss": 1.0165, "step": 14008 }, { "epoch": 0.01, "learning_rate": 4.997613181923378e-05, "loss": 0.9837, "step": 14009 }, { "epoch": 0.01, "learning_rate": 4.99761283876083e-05, "loss": 0.9486, "step": 14010 }, { "epoch": 0.01, "learning_rate": 4.997612495573626e-05, "loss": 1.0191, "step": 14011 }, { "epoch": 0.01, "learning_rate": 4.997612152361767e-05, "loss": 1.5051, "step": 14012 }, { "epoch": 0.01, "learning_rate": 4.997611809125252e-05, "loss": 1.4434, "step": 14013 }, { "epoch": 0.01, "learning_rate": 4.997611465864082e-05, "loss": 1.3862, "step": 14014 }, { "epoch": 0.01, "learning_rate": 4.997611122578256e-05, "loss": 1.4054, "step": 14015 }, { "epoch": 0.01, "learning_rate": 4.997610779267776e-05, "loss": 1.4269, "step": 14016 }, { "epoch": 0.01, "learning_rate": 4.99761043593264e-05, "loss": 1.4767, "step": 14017 }, { "epoch": 0.01, "learning_rate": 4.9976100925728485e-05, "loss": 1.5355, "step": 14018 }, { "epoch": 0.01, "learning_rate": 4.997609749188401e-05, "loss": 1.3726, "step": 14019 }, { "epoch": 0.01, "learning_rate": 4.997609405779299e-05, "loss": 1.4025, "step": 14020 }, { "epoch": 0.01, "learning_rate": 4.997609062345542e-05, "loss": 1.4298, "step": 14021 }, { "epoch": 0.01, "learning_rate": 4.997608718887129e-05, "loss": 1.4052, "step": 14022 }, { "epoch": 0.01, "learning_rate": 4.9976083754040606e-05, "loss": 1.2636, "step": 14023 }, { "epoch": 0.01, "learning_rate": 4.997608031896337e-05, "loss": 0.9864, "step": 14024 }, { "epoch": 0.01, "learning_rate": 4.997607688363958e-05, "loss": 1.2486, "step": 14025 }, { "epoch": 0.01, "learning_rate": 4.997607344806924e-05, "loss": 1.2348, "step": 14026 }, { "epoch": 0.01, "learning_rate": 4.997607001225235e-05, "loss": 1.0257, "step": 14027 }, { "epoch": 0.01, "learning_rate": 4.9976066576188895e-05, "loss": 1.235, "step": 14028 }, { "epoch": 0.01, "learning_rate": 4.9976063139878895e-05, "loss": 1.1075, "step": 14029 }, { "epoch": 0.01, "learning_rate": 4.997605970332234e-05, "loss": 1.0367, "step": 14030 }, { "epoch": 0.01, "learning_rate": 4.997605626651923e-05, "loss": 0.9863, "step": 14031 }, { "epoch": 0.01, "learning_rate": 4.997605282946957e-05, "loss": 1.1902, "step": 14032 }, { "epoch": 0.01, "learning_rate": 4.997604939217335e-05, "loss": 0.9979, "step": 14033 }, { "epoch": 0.01, "learning_rate": 4.997604595463059e-05, "loss": 1.7464, "step": 14034 }, { "epoch": 0.01, "learning_rate": 4.997604251684127e-05, "loss": 1.6848, "step": 14035 }, { "epoch": 0.01, "learning_rate": 4.9976039078805395e-05, "loss": 1.3606, "step": 14036 }, { "epoch": 0.01, "learning_rate": 4.997603564052297e-05, "loss": 0.3411, "step": 14037 }, { "epoch": 0.01, "learning_rate": 4.997603220199399e-05, "loss": 0.2637, "step": 14038 }, { "epoch": 0.01, "learning_rate": 4.9976028763218455e-05, "loss": 0.2964, "step": 14039 }, { "epoch": 0.01, "learning_rate": 4.997602532419637e-05, "loss": 0.2212, "step": 14040 }, { "epoch": 0.01, "learning_rate": 4.997602188492774e-05, "loss": 0.3276, "step": 14041 }, { "epoch": 0.01, "learning_rate": 4.997601844541254e-05, "loss": 0.339, "step": 14042 }, { "epoch": 0.01, "learning_rate": 4.99760150056508e-05, "loss": 0.2881, "step": 14043 }, { "epoch": 0.01, "learning_rate": 4.9976011565642504e-05, "loss": 0.191, "step": 14044 }, { "epoch": 0.01, "learning_rate": 4.9976008125387655e-05, "loss": 0.1706, "step": 14045 }, { "epoch": 0.01, "learning_rate": 4.997600468488626e-05, "loss": 0.2245, "step": 14046 }, { "epoch": 0.01, "learning_rate": 4.997600124413831e-05, "loss": 0.27, "step": 14047 }, { "epoch": 0.01, "learning_rate": 4.9975997803143795e-05, "loss": 0.2916, "step": 14048 }, { "epoch": 0.01, "learning_rate": 4.9975994361902735e-05, "loss": 0.2542, "step": 14049 }, { "epoch": 0.01, "learning_rate": 4.997599092041513e-05, "loss": 0.1982, "step": 14050 }, { "epoch": 0.01, "learning_rate": 4.9975987478680966e-05, "loss": 0.4146, "step": 14051 }, { "epoch": 0.01, "learning_rate": 4.9975984036700244e-05, "loss": 0.6991, "step": 14052 }, { "epoch": 0.01, "learning_rate": 4.997598059447297e-05, "loss": 0.6499, "step": 14053 }, { "epoch": 0.01, "learning_rate": 4.9975977151999155e-05, "loss": 0.3083, "step": 14054 }, { "epoch": 0.01, "learning_rate": 4.997597370927878e-05, "loss": 0.2072, "step": 14055 }, { "epoch": 0.01, "learning_rate": 4.9975970266311856e-05, "loss": 0.4152, "step": 14056 }, { "epoch": 0.01, "learning_rate": 4.997596682309838e-05, "loss": 0.7276, "step": 14057 }, { "epoch": 0.01, "learning_rate": 4.997596337963835e-05, "loss": 1.0309, "step": 14058 }, { "epoch": 0.01, "learning_rate": 4.997595993593177e-05, "loss": 1.3804, "step": 14059 }, { "epoch": 0.01, "learning_rate": 4.997595649197864e-05, "loss": 2.1518, "step": 14060 }, { "epoch": 0.01, "learning_rate": 4.997595304777894e-05, "loss": 1.1852, "step": 14061 }, { "epoch": 0.01, "learning_rate": 4.99759496033327e-05, "loss": 0.7476, "step": 14062 }, { "epoch": 0.01, "learning_rate": 4.9975946158639916e-05, "loss": 0.4246, "step": 14063 }, { "epoch": 0.01, "learning_rate": 4.9975942713700575e-05, "loss": 0.7485, "step": 14064 }, { "epoch": 0.01, "learning_rate": 4.997593926851468e-05, "loss": 0.3086, "step": 14065 }, { "epoch": 0.01, "learning_rate": 4.997593582308223e-05, "loss": 0.2025, "step": 14066 }, { "epoch": 0.01, "learning_rate": 4.997593237740323e-05, "loss": 0.1456, "step": 14067 }, { "epoch": 0.01, "learning_rate": 4.997592893147769e-05, "loss": 1.0141, "step": 14068 }, { "epoch": 0.01, "learning_rate": 4.997592548530559e-05, "loss": 1.0374, "step": 14069 }, { "epoch": 0.01, "learning_rate": 4.997592203888693e-05, "loss": 1.0126, "step": 14070 }, { "epoch": 0.01, "learning_rate": 4.9975918592221726e-05, "loss": 1.0492, "step": 14071 }, { "epoch": 0.01, "learning_rate": 4.9975915145309964e-05, "loss": 1.3127, "step": 14072 }, { "epoch": 0.01, "learning_rate": 4.9975911698151654e-05, "loss": 1.4635, "step": 14073 }, { "epoch": 0.01, "learning_rate": 4.9975908250746796e-05, "loss": 1.2857, "step": 14074 }, { "epoch": 0.01, "learning_rate": 4.9975904803095384e-05, "loss": 0.9974, "step": 14075 }, { "epoch": 0.01, "learning_rate": 4.997590135519742e-05, "loss": 1.5194, "step": 14076 }, { "epoch": 0.01, "learning_rate": 4.9975897907052904e-05, "loss": 1.1631, "step": 14077 }, { "epoch": 0.01, "learning_rate": 4.9975894458661836e-05, "loss": 1.0396, "step": 14078 }, { "epoch": 0.01, "learning_rate": 4.9975891010024214e-05, "loss": 1.4006, "step": 14079 }, { "epoch": 0.01, "learning_rate": 4.9975887561140044e-05, "loss": 1.1684, "step": 14080 }, { "epoch": 0.01, "learning_rate": 4.9975884112009326e-05, "loss": 1.2567, "step": 14081 }, { "epoch": 0.01, "learning_rate": 4.9975880662632054e-05, "loss": 1.0493, "step": 14082 }, { "epoch": 0.01, "learning_rate": 4.997587721300823e-05, "loss": 1.4118, "step": 14083 }, { "epoch": 0.01, "learning_rate": 4.997587376313785e-05, "loss": 1.0582, "step": 14084 }, { "epoch": 0.01, "learning_rate": 4.9975870313020926e-05, "loss": 0.9975, "step": 14085 }, { "epoch": 0.01, "learning_rate": 4.997586686265745e-05, "loss": 1.1511, "step": 14086 }, { "epoch": 0.01, "learning_rate": 4.997586341204742e-05, "loss": 0.9706, "step": 14087 }, { "epoch": 0.01, "learning_rate": 4.9975859961190835e-05, "loss": 1.076, "step": 14088 }, { "epoch": 0.01, "learning_rate": 4.99758565100877e-05, "loss": 1.3168, "step": 14089 }, { "epoch": 0.01, "learning_rate": 4.997585305873802e-05, "loss": 0.8597, "step": 14090 }, { "epoch": 0.01, "learning_rate": 4.997584960714179e-05, "loss": 0.7687, "step": 14091 }, { "epoch": 0.01, "learning_rate": 4.9975846155299e-05, "loss": 0.7132, "step": 14092 }, { "epoch": 0.01, "learning_rate": 4.9975842703209665e-05, "loss": 0.6899, "step": 14093 }, { "epoch": 0.01, "learning_rate": 4.997583925087378e-05, "loss": 0.668, "step": 14094 }, { "epoch": 0.01, "learning_rate": 4.997583579829134e-05, "loss": 0.7043, "step": 14095 }, { "epoch": 0.01, "learning_rate": 4.9975832345462345e-05, "loss": 0.6641, "step": 14096 }, { "epoch": 0.01, "learning_rate": 4.997582889238681e-05, "loss": 0.5732, "step": 14097 }, { "epoch": 0.01, "learning_rate": 4.997582543906472e-05, "loss": 0.6086, "step": 14098 }, { "epoch": 0.01, "learning_rate": 4.9975821985496076e-05, "loss": 0.5433, "step": 14099 }, { "epoch": 0.01, "learning_rate": 4.9975818531680886e-05, "loss": 0.78, "step": 14100 }, { "epoch": 0.01, "learning_rate": 4.997581507761914e-05, "loss": 1.2859, "step": 14101 }, { "epoch": 0.01, "learning_rate": 4.9975811623310845e-05, "loss": 0.9706, "step": 14102 }, { "epoch": 0.01, "learning_rate": 4.9975808168756e-05, "loss": 1.0174, "step": 14103 }, { "epoch": 0.01, "learning_rate": 4.997580471395461e-05, "loss": 0.9041, "step": 14104 }, { "epoch": 0.01, "learning_rate": 4.9975801258906665e-05, "loss": 1.0138, "step": 14105 }, { "epoch": 0.01, "learning_rate": 4.997579780361217e-05, "loss": 1.1986, "step": 14106 }, { "epoch": 0.01, "learning_rate": 4.9975794348071114e-05, "loss": 1.0084, "step": 14107 }, { "epoch": 0.01, "learning_rate": 4.997579089228352e-05, "loss": 1.2811, "step": 14108 }, { "epoch": 0.01, "learning_rate": 4.997578743624937e-05, "loss": 1.0046, "step": 14109 }, { "epoch": 0.01, "learning_rate": 4.9975783979968673e-05, "loss": 1.2217, "step": 14110 }, { "epoch": 0.01, "learning_rate": 4.997578052344143e-05, "loss": 1.2312, "step": 14111 }, { "epoch": 0.01, "learning_rate": 4.9975777066667626e-05, "loss": 1.3594, "step": 14112 }, { "epoch": 0.01, "learning_rate": 4.997577360964728e-05, "loss": 0.9843, "step": 14113 }, { "epoch": 0.01, "learning_rate": 4.9975770152380375e-05, "loss": 1.0723, "step": 14114 }, { "epoch": 0.01, "learning_rate": 4.9975766694866924e-05, "loss": 1.1024, "step": 14115 }, { "epoch": 0.01, "learning_rate": 4.9975763237106927e-05, "loss": 1.0341, "step": 14116 }, { "epoch": 0.01, "learning_rate": 4.9975759779100374e-05, "loss": 0.527, "step": 14117 }, { "epoch": 0.01, "learning_rate": 4.9975756320847274e-05, "loss": 0.4008, "step": 14118 }, { "epoch": 0.01, "learning_rate": 4.997575286234763e-05, "loss": 0.41, "step": 14119 }, { "epoch": 0.01, "learning_rate": 4.9975749403601425e-05, "loss": 0.3649, "step": 14120 }, { "epoch": 0.01, "learning_rate": 4.997574594460867e-05, "loss": 0.2659, "step": 14121 }, { "epoch": 0.01, "learning_rate": 4.997574248536937e-05, "loss": 0.1907, "step": 14122 }, { "epoch": 0.01, "learning_rate": 4.9975739025883514e-05, "loss": 0.25, "step": 14123 }, { "epoch": 0.01, "learning_rate": 4.9975735566151115e-05, "loss": 1.0159, "step": 14124 }, { "epoch": 0.01, "learning_rate": 4.997573210617217e-05, "loss": 1.2458, "step": 14125 }, { "epoch": 0.01, "learning_rate": 4.997572864594666e-05, "loss": 1.1723, "step": 14126 }, { "epoch": 0.01, "learning_rate": 4.9975725185474606e-05, "loss": 1.1799, "step": 14127 }, { "epoch": 0.01, "learning_rate": 4.997572172475601e-05, "loss": 1.0264, "step": 14128 }, { "epoch": 0.01, "learning_rate": 4.997571826379086e-05, "loss": 1.2867, "step": 14129 }, { "epoch": 0.01, "learning_rate": 4.997571480257916e-05, "loss": 1.1588, "step": 14130 }, { "epoch": 0.01, "learning_rate": 4.997571134112091e-05, "loss": 1.269, "step": 14131 }, { "epoch": 0.01, "learning_rate": 4.997570787941611e-05, "loss": 1.105, "step": 14132 }, { "epoch": 0.01, "learning_rate": 4.997570441746476e-05, "loss": 1.1432, "step": 14133 }, { "epoch": 0.01, "learning_rate": 4.997570095526686e-05, "loss": 1.0189, "step": 14134 }, { "epoch": 0.01, "learning_rate": 4.997569749282241e-05, "loss": 1.552, "step": 14135 }, { "epoch": 0.01, "learning_rate": 4.9975694030131415e-05, "loss": 1.186, "step": 14136 }, { "epoch": 0.01, "learning_rate": 4.9975690567193864e-05, "loss": 0.9589, "step": 14137 }, { "epoch": 0.01, "learning_rate": 4.997568710400977e-05, "loss": 1.0039, "step": 14138 }, { "epoch": 0.01, "learning_rate": 4.9975683640579126e-05, "loss": 0.9679, "step": 14139 }, { "epoch": 0.01, "learning_rate": 4.997568017690193e-05, "loss": 1.2845, "step": 14140 }, { "epoch": 0.01, "learning_rate": 4.9975676712978183e-05, "loss": 1.432, "step": 14141 }, { "epoch": 0.01, "learning_rate": 4.997567324880789e-05, "loss": 1.1023, "step": 14142 }, { "epoch": 0.01, "learning_rate": 4.9975669784391044e-05, "loss": 1.1681, "step": 14143 }, { "epoch": 0.01, "learning_rate": 4.9975666319727646e-05, "loss": 1.1895, "step": 14144 }, { "epoch": 0.01, "learning_rate": 4.997566285481771e-05, "loss": 1.0524, "step": 14145 }, { "epoch": 0.01, "learning_rate": 4.997565938966121e-05, "loss": 1.0158, "step": 14146 }, { "epoch": 0.01, "learning_rate": 4.9975655924258175e-05, "loss": 0.8595, "step": 14147 }, { "epoch": 0.01, "learning_rate": 4.997565245860858e-05, "loss": 1.0239, "step": 14148 }, { "epoch": 0.01, "learning_rate": 4.9975648992712444e-05, "loss": 1.0904, "step": 14149 }, { "epoch": 0.01, "learning_rate": 4.997564552656976e-05, "loss": 1.1, "step": 14150 }, { "epoch": 0.01, "learning_rate": 4.9975642060180517e-05, "loss": 0.9147, "step": 14151 }, { "epoch": 0.01, "learning_rate": 4.997563859354473e-05, "loss": 0.966, "step": 14152 }, { "epoch": 0.01, "learning_rate": 4.99756351266624e-05, "loss": 1.1773, "step": 14153 }, { "epoch": 0.01, "learning_rate": 4.997563165953351e-05, "loss": 0.8457, "step": 14154 }, { "epoch": 0.01, "learning_rate": 4.997562819215808e-05, "loss": 0.676, "step": 14155 }, { "epoch": 0.01, "learning_rate": 4.9975624724536095e-05, "loss": 1.0086, "step": 14156 }, { "epoch": 0.01, "learning_rate": 4.9975621256667565e-05, "loss": 0.637, "step": 14157 }, { "epoch": 0.01, "learning_rate": 4.997561778855249e-05, "loss": 0.7949, "step": 14158 }, { "epoch": 0.01, "learning_rate": 4.9975614320190856e-05, "loss": 1.2112, "step": 14159 }, { "epoch": 0.01, "learning_rate": 4.997561085158268e-05, "loss": 1.006, "step": 14160 }, { "epoch": 0.01, "learning_rate": 4.997560738272795e-05, "loss": 1.1773, "step": 14161 }, { "epoch": 0.01, "learning_rate": 4.9975603913626675e-05, "loss": 1.1284, "step": 14162 }, { "epoch": 0.01, "learning_rate": 4.9975600444278853e-05, "loss": 1.0709, "step": 14163 }, { "epoch": 0.01, "learning_rate": 4.997559697468448e-05, "loss": 0.9037, "step": 14164 }, { "epoch": 0.01, "learning_rate": 4.997559350484356e-05, "loss": 1.2146, "step": 14165 }, { "epoch": 0.01, "learning_rate": 4.9975590034756095e-05, "loss": 1.0601, "step": 14166 }, { "epoch": 0.01, "learning_rate": 4.997558656442207e-05, "loss": 1.1863, "step": 14167 }, { "epoch": 0.01, "learning_rate": 4.99755830938415e-05, "loss": 1.3709, "step": 14168 }, { "epoch": 0.01, "learning_rate": 4.997557962301439e-05, "loss": 1.3823, "step": 14169 }, { "epoch": 0.01, "learning_rate": 4.997557615194073e-05, "loss": 1.089, "step": 14170 }, { "epoch": 0.01, "learning_rate": 4.997557268062052e-05, "loss": 1.1908, "step": 14171 }, { "epoch": 0.01, "learning_rate": 4.997556920905376e-05, "loss": 1.1416, "step": 14172 }, { "epoch": 0.01, "learning_rate": 4.997556573724045e-05, "loss": 0.6685, "step": 14173 }, { "epoch": 0.01, "learning_rate": 4.99755622651806e-05, "loss": 1.248, "step": 14174 }, { "epoch": 0.01, "learning_rate": 4.997555879287419e-05, "loss": 0.8823, "step": 14175 }, { "epoch": 0.01, "learning_rate": 4.997555532032124e-05, "loss": 1.2691, "step": 14176 }, { "epoch": 0.01, "learning_rate": 4.997555184752174e-05, "loss": 0.9468, "step": 14177 }, { "epoch": 0.01, "learning_rate": 4.9975548374475686e-05, "loss": 1.1383, "step": 14178 }, { "epoch": 0.01, "learning_rate": 4.997554490118309e-05, "loss": 1.1812, "step": 14179 }, { "epoch": 0.01, "learning_rate": 4.9975541427643945e-05, "loss": 1.1472, "step": 14180 }, { "epoch": 0.01, "learning_rate": 4.997553795385825e-05, "loss": 1.1894, "step": 14181 }, { "epoch": 0.01, "learning_rate": 4.9975534479826014e-05, "loss": 1.0304, "step": 14182 }, { "epoch": 0.01, "learning_rate": 4.9975531005547227e-05, "loss": 1.3028, "step": 14183 }, { "epoch": 0.01, "learning_rate": 4.9975527531021885e-05, "loss": 0.9467, "step": 14184 }, { "epoch": 0.01, "learning_rate": 4.997552405625e-05, "loss": 1.227, "step": 14185 }, { "epoch": 0.01, "learning_rate": 4.997552058123157e-05, "loss": 1.0007, "step": 14186 }, { "epoch": 0.01, "learning_rate": 4.997551710596658e-05, "loss": 1.0721, "step": 14187 }, { "epoch": 0.01, "learning_rate": 4.9975513630455064e-05, "loss": 1.1782, "step": 14188 }, { "epoch": 0.01, "learning_rate": 4.997551015469698e-05, "loss": 1.1244, "step": 14189 }, { "epoch": 0.01, "learning_rate": 4.9975506678692365e-05, "loss": 1.1925, "step": 14190 }, { "epoch": 0.01, "learning_rate": 4.997550320244119e-05, "loss": 1.1091, "step": 14191 }, { "epoch": 0.01, "learning_rate": 4.997549972594347e-05, "loss": 1.0063, "step": 14192 }, { "epoch": 0.01, "learning_rate": 4.9975496249199206e-05, "loss": 1.0792, "step": 14193 }, { "epoch": 0.01, "learning_rate": 4.997549277220839e-05, "loss": 0.9771, "step": 14194 }, { "epoch": 0.01, "learning_rate": 4.9975489294971024e-05, "loss": 1.3387, "step": 14195 }, { "epoch": 0.01, "learning_rate": 4.997548581748712e-05, "loss": 1.0673, "step": 14196 }, { "epoch": 0.01, "learning_rate": 4.9975482339756665e-05, "loss": 0.997, "step": 14197 }, { "epoch": 0.01, "learning_rate": 4.997547886177966e-05, "loss": 1.1778, "step": 14198 }, { "epoch": 0.01, "learning_rate": 4.99754753835561e-05, "loss": 1.0958, "step": 14199 }, { "epoch": 0.01, "learning_rate": 4.9975471905086e-05, "loss": 1.2378, "step": 14200 }, { "epoch": 0.01, "learning_rate": 4.997546842636936e-05, "loss": 1.1805, "step": 14201 }, { "epoch": 0.01, "learning_rate": 4.997546494740617e-05, "loss": 1.2093, "step": 14202 }, { "epoch": 0.01, "learning_rate": 4.997546146819643e-05, "loss": 1.1227, "step": 14203 }, { "epoch": 0.01, "learning_rate": 4.9975457988740135e-05, "loss": 1.2625, "step": 14204 }, { "epoch": 0.01, "learning_rate": 4.99754545090373e-05, "loss": 1.0928, "step": 14205 }, { "epoch": 0.01, "learning_rate": 4.997545102908791e-05, "loss": 1.148, "step": 14206 }, { "epoch": 0.01, "learning_rate": 4.997544754889199e-05, "loss": 0.9633, "step": 14207 }, { "epoch": 0.01, "learning_rate": 4.9975444068449515e-05, "loss": 1.0229, "step": 14208 }, { "epoch": 0.01, "learning_rate": 4.997544058776049e-05, "loss": 1.1975, "step": 14209 }, { "epoch": 0.01, "learning_rate": 4.997543710682492e-05, "loss": 0.9083, "step": 14210 }, { "epoch": 0.01, "learning_rate": 4.99754336256428e-05, "loss": 1.2467, "step": 14211 }, { "epoch": 0.01, "learning_rate": 4.9975430144214134e-05, "loss": 1.0118, "step": 14212 }, { "epoch": 0.01, "learning_rate": 4.9975426662538926e-05, "loss": 0.7819, "step": 14213 }, { "epoch": 0.01, "learning_rate": 4.9975423180617164e-05, "loss": 1.1299, "step": 14214 }, { "epoch": 0.01, "learning_rate": 4.997541969844886e-05, "loss": 1.0817, "step": 14215 }, { "epoch": 0.01, "learning_rate": 4.997541621603401e-05, "loss": 1.1529, "step": 14216 }, { "epoch": 0.01, "learning_rate": 4.997541273337261e-05, "loss": 1.2344, "step": 14217 }, { "epoch": 0.01, "learning_rate": 4.997540925046467e-05, "loss": 0.7974, "step": 14218 }, { "epoch": 0.01, "learning_rate": 4.997540576731017e-05, "loss": 1.2864, "step": 14219 }, { "epoch": 0.01, "learning_rate": 4.9975402283909135e-05, "loss": 0.9274, "step": 14220 }, { "epoch": 0.01, "learning_rate": 4.997539880026155e-05, "loss": 1.1259, "step": 14221 }, { "epoch": 0.01, "learning_rate": 4.997539531636741e-05, "loss": 0.8732, "step": 14222 }, { "epoch": 0.01, "learning_rate": 4.9975391832226735e-05, "loss": 1.1752, "step": 14223 }, { "epoch": 0.01, "learning_rate": 4.997538834783951e-05, "loss": 0.9616, "step": 14224 }, { "epoch": 0.01, "learning_rate": 4.997538486320574e-05, "loss": 0.9051, "step": 14225 }, { "epoch": 0.01, "learning_rate": 4.997538137832542e-05, "loss": 1.0937, "step": 14226 }, { "epoch": 0.01, "learning_rate": 4.997537789319855e-05, "loss": 1.0993, "step": 14227 }, { "epoch": 0.01, "learning_rate": 4.9975374407825146e-05, "loss": 1.1841, "step": 14228 }, { "epoch": 0.01, "learning_rate": 4.9975370922205184e-05, "loss": 0.7506, "step": 14229 }, { "epoch": 0.01, "learning_rate": 4.997536743633868e-05, "loss": 1.4471, "step": 14230 }, { "epoch": 0.01, "learning_rate": 4.997536395022563e-05, "loss": 0.5568, "step": 14231 }, { "epoch": 0.01, "learning_rate": 4.997536046386603e-05, "loss": 0.832, "step": 14232 }, { "epoch": 0.01, "learning_rate": 4.997535697725989e-05, "loss": 1.2028, "step": 14233 }, { "epoch": 0.01, "learning_rate": 4.99753534904072e-05, "loss": 1.0788, "step": 14234 }, { "epoch": 0.01, "learning_rate": 4.997535000330796e-05, "loss": 0.7651, "step": 14235 }, { "epoch": 0.01, "learning_rate": 4.9975346515962186e-05, "loss": 0.9412, "step": 14236 }, { "epoch": 0.01, "learning_rate": 4.997534302836986e-05, "loss": 1.28, "step": 14237 }, { "epoch": 0.01, "learning_rate": 4.997533954053099e-05, "loss": 1.1412, "step": 14238 }, { "epoch": 0.01, "learning_rate": 4.9975336052445564e-05, "loss": 0.7331, "step": 14239 }, { "epoch": 0.01, "learning_rate": 4.99753325641136e-05, "loss": 0.9254, "step": 14240 }, { "epoch": 0.01, "learning_rate": 4.9975329075535086e-05, "loss": 0.9398, "step": 14241 }, { "epoch": 0.01, "learning_rate": 4.997532558671003e-05, "loss": 1.412, "step": 14242 }, { "epoch": 0.01, "learning_rate": 4.9975322097638426e-05, "loss": 1.1725, "step": 14243 }, { "epoch": 0.01, "learning_rate": 4.997531860832028e-05, "loss": 1.2248, "step": 14244 }, { "epoch": 0.01, "learning_rate": 4.997531511875558e-05, "loss": 1.0279, "step": 14245 }, { "epoch": 0.01, "learning_rate": 4.9975311628944344e-05, "loss": 1.0378, "step": 14246 }, { "epoch": 0.01, "learning_rate": 4.997530813888655e-05, "loss": 1.2247, "step": 14247 }, { "epoch": 0.01, "learning_rate": 4.997530464858222e-05, "loss": 0.4484, "step": 14248 }, { "epoch": 0.01, "learning_rate": 4.997530115803135e-05, "loss": 0.9846, "step": 14249 }, { "epoch": 0.01, "learning_rate": 4.997529766723392e-05, "loss": 0.8921, "step": 14250 }, { "epoch": 0.01, "learning_rate": 4.9975294176189954e-05, "loss": 1.0155, "step": 14251 }, { "epoch": 0.01, "learning_rate": 4.997529068489943e-05, "loss": 0.9831, "step": 14252 }, { "epoch": 0.01, "learning_rate": 4.9975287193362375e-05, "loss": 1.0879, "step": 14253 }, { "epoch": 0.01, "learning_rate": 4.9975283701578765e-05, "loss": 1.1171, "step": 14254 }, { "epoch": 0.01, "learning_rate": 4.997528020954862e-05, "loss": 1.0931, "step": 14255 }, { "epoch": 0.01, "learning_rate": 4.997527671727192e-05, "loss": 1.1142, "step": 14256 }, { "epoch": 0.01, "learning_rate": 4.997527322474868e-05, "loss": 1.0804, "step": 14257 }, { "epoch": 0.01, "learning_rate": 4.9975269731978894e-05, "loss": 1.1162, "step": 14258 }, { "epoch": 0.01, "learning_rate": 4.997526623896256e-05, "loss": 1.1853, "step": 14259 }, { "epoch": 0.01, "learning_rate": 4.997526274569968e-05, "loss": 1.082, "step": 14260 }, { "epoch": 0.01, "learning_rate": 4.997525925219026e-05, "loss": 0.9478, "step": 14261 }, { "epoch": 0.01, "learning_rate": 4.997525575843429e-05, "loss": 1.033, "step": 14262 }, { "epoch": 0.01, "learning_rate": 4.997525226443178e-05, "loss": 1.1039, "step": 14263 }, { "epoch": 0.01, "learning_rate": 4.997524877018271e-05, "loss": 1.0438, "step": 14264 }, { "epoch": 0.01, "learning_rate": 4.997524527568711e-05, "loss": 1.1779, "step": 14265 }, { "epoch": 0.01, "learning_rate": 4.9975241780944964e-05, "loss": 1.1588, "step": 14266 }, { "epoch": 0.01, "learning_rate": 4.997523828595627e-05, "loss": 1.1946, "step": 14267 }, { "epoch": 0.01, "learning_rate": 4.997523479072103e-05, "loss": 1.1071, "step": 14268 }, { "epoch": 0.01, "learning_rate": 4.997523129523924e-05, "loss": 1.0114, "step": 14269 }, { "epoch": 0.01, "learning_rate": 4.997522779951091e-05, "loss": 0.9748, "step": 14270 }, { "epoch": 0.01, "learning_rate": 4.997522430353604e-05, "loss": 1.3441, "step": 14271 }, { "epoch": 0.01, "learning_rate": 4.997522080731463e-05, "loss": 1.0202, "step": 14272 }, { "epoch": 0.01, "learning_rate": 4.9975217310846653e-05, "loss": 1.1202, "step": 14273 }, { "epoch": 0.01, "learning_rate": 4.997521381413215e-05, "loss": 0.757, "step": 14274 }, { "epoch": 0.01, "learning_rate": 4.99752103171711e-05, "loss": 0.4076, "step": 14275 }, { "epoch": 0.01, "learning_rate": 4.99752068199635e-05, "loss": 0.6337, "step": 14276 }, { "epoch": 0.01, "learning_rate": 4.997520332250936e-05, "loss": 0.3864, "step": 14277 }, { "epoch": 0.01, "learning_rate": 4.997519982480867e-05, "loss": 0.4376, "step": 14278 }, { "epoch": 0.01, "learning_rate": 4.997519632686144e-05, "loss": 0.6838, "step": 14279 }, { "epoch": 0.01, "learning_rate": 4.997519282866766e-05, "loss": 0.4501, "step": 14280 }, { "epoch": 0.01, "learning_rate": 4.997518933022733e-05, "loss": 0.2631, "step": 14281 }, { "epoch": 0.01, "learning_rate": 4.997518583154047e-05, "loss": 0.3098, "step": 14282 }, { "epoch": 0.01, "learning_rate": 4.997518233260706e-05, "loss": 0.1209, "step": 14283 }, { "epoch": 0.01, "learning_rate": 4.9975178833427106e-05, "loss": 0.4564, "step": 14284 }, { "epoch": 0.01, "learning_rate": 4.997517533400061e-05, "loss": 0.7584, "step": 14285 }, { "epoch": 0.01, "learning_rate": 4.9975171834327564e-05, "loss": 0.8234, "step": 14286 }, { "epoch": 0.01, "learning_rate": 4.997516833440797e-05, "loss": 1.2015, "step": 14287 }, { "epoch": 0.01, "learning_rate": 4.997516483424184e-05, "loss": 1.2761, "step": 14288 }, { "epoch": 0.01, "learning_rate": 4.9975161333829165e-05, "loss": 1.1576, "step": 14289 }, { "epoch": 0.01, "learning_rate": 4.9975157833169936e-05, "loss": 0.9933, "step": 14290 }, { "epoch": 0.01, "learning_rate": 4.9975154332264174e-05, "loss": 1.3237, "step": 14291 }, { "epoch": 0.01, "learning_rate": 4.9975150831111864e-05, "loss": 0.9063, "step": 14292 }, { "epoch": 0.01, "learning_rate": 4.997514732971301e-05, "loss": 1.0895, "step": 14293 }, { "epoch": 0.01, "learning_rate": 4.9975143828067615e-05, "loss": 1.091, "step": 14294 }, { "epoch": 0.01, "learning_rate": 4.9975140326175676e-05, "loss": 1.0007, "step": 14295 }, { "epoch": 0.01, "learning_rate": 4.997513682403718e-05, "loss": 1.4644, "step": 14296 }, { "epoch": 0.01, "learning_rate": 4.9975133321652155e-05, "loss": 0.9079, "step": 14297 }, { "epoch": 0.01, "learning_rate": 4.997512981902058e-05, "loss": 1.1363, "step": 14298 }, { "epoch": 0.01, "learning_rate": 4.9975126316142464e-05, "loss": 1.3871, "step": 14299 }, { "epoch": 0.01, "learning_rate": 4.997512281301781e-05, "loss": 1.1419, "step": 14300 }, { "epoch": 0.01, "learning_rate": 4.99751193096466e-05, "loss": 1.207, "step": 14301 }, { "epoch": 0.01, "learning_rate": 4.997511580602885e-05, "loss": 0.5183, "step": 14302 }, { "epoch": 0.01, "learning_rate": 4.997511230216455e-05, "loss": 1.3572, "step": 14303 }, { "epoch": 0.01, "learning_rate": 4.997510879805372e-05, "loss": 0.7195, "step": 14304 }, { "epoch": 0.01, "learning_rate": 4.997510529369633e-05, "loss": 0.7084, "step": 14305 }, { "epoch": 0.01, "learning_rate": 4.997510178909241e-05, "loss": 0.7775, "step": 14306 }, { "epoch": 0.01, "learning_rate": 4.997509828424194e-05, "loss": 1.1075, "step": 14307 }, { "epoch": 0.01, "learning_rate": 4.997509477914493e-05, "loss": 1.1303, "step": 14308 }, { "epoch": 0.01, "learning_rate": 4.997509127380138e-05, "loss": 1.0402, "step": 14309 }, { "epoch": 0.01, "learning_rate": 4.997508776821127e-05, "loss": 1.3226, "step": 14310 }, { "epoch": 0.01, "learning_rate": 4.997508426237463e-05, "loss": 1.282, "step": 14311 }, { "epoch": 0.01, "learning_rate": 4.997508075629145e-05, "loss": 1.2724, "step": 14312 }, { "epoch": 0.01, "learning_rate": 4.997507724996172e-05, "loss": 1.2906, "step": 14313 }, { "epoch": 0.01, "learning_rate": 4.9975073743385444e-05, "loss": 1.2708, "step": 14314 }, { "epoch": 0.01, "learning_rate": 4.997507023656263e-05, "loss": 0.7824, "step": 14315 }, { "epoch": 0.01, "learning_rate": 4.9975066729493264e-05, "loss": 1.1854, "step": 14316 }, { "epoch": 0.01, "learning_rate": 4.9975063222177366e-05, "loss": 0.9226, "step": 14317 }, { "epoch": 0.01, "learning_rate": 4.9975059714614914e-05, "loss": 0.824, "step": 14318 }, { "epoch": 0.01, "learning_rate": 4.997505620680593e-05, "loss": 1.1231, "step": 14319 }, { "epoch": 0.01, "learning_rate": 4.997505269875039e-05, "loss": 1.0022, "step": 14320 }, { "epoch": 0.01, "learning_rate": 4.997504919044831e-05, "loss": 0.5094, "step": 14321 }, { "epoch": 0.01, "learning_rate": 4.99750456818997e-05, "loss": 1.136, "step": 14322 }, { "epoch": 0.01, "learning_rate": 4.9975042173104535e-05, "loss": 1.4898, "step": 14323 }, { "epoch": 0.01, "learning_rate": 4.9975038664062825e-05, "loss": 1.0795, "step": 14324 }, { "epoch": 0.01, "learning_rate": 4.9975035154774574e-05, "loss": 1.8042, "step": 14325 }, { "epoch": 0.01, "learning_rate": 4.997503164523979e-05, "loss": 1.0174, "step": 14326 }, { "epoch": 0.01, "learning_rate": 4.997502813545845e-05, "loss": 1.4137, "step": 14327 }, { "epoch": 0.01, "learning_rate": 4.9975024625430576e-05, "loss": 1.2082, "step": 14328 }, { "epoch": 0.01, "learning_rate": 4.997502111515615e-05, "loss": 1.0201, "step": 14329 }, { "epoch": 0.01, "learning_rate": 4.9975017604635186e-05, "loss": 1.3109, "step": 14330 }, { "epoch": 0.01, "learning_rate": 4.997501409386768e-05, "loss": 1.0131, "step": 14331 }, { "epoch": 0.01, "learning_rate": 4.997501058285363e-05, "loss": 0.9856, "step": 14332 }, { "epoch": 0.01, "learning_rate": 4.997500707159304e-05, "loss": 1.3251, "step": 14333 }, { "epoch": 0.01, "learning_rate": 4.997500356008591e-05, "loss": 1.2006, "step": 14334 }, { "epoch": 0.01, "learning_rate": 4.997500004833223e-05, "loss": 1.5733, "step": 14335 }, { "epoch": 0.01, "learning_rate": 4.997499653633201e-05, "loss": 1.4137, "step": 14336 }, { "epoch": 0.01, "learning_rate": 4.997499302408525e-05, "loss": 1.1408, "step": 14337 }, { "epoch": 0.01, "learning_rate": 4.997498951159194e-05, "loss": 1.2687, "step": 14338 }, { "epoch": 0.01, "learning_rate": 4.997498599885209e-05, "loss": 1.5553, "step": 14339 }, { "epoch": 0.01, "learning_rate": 4.99749824858657e-05, "loss": 0.52, "step": 14340 }, { "epoch": 0.01, "learning_rate": 4.997497897263277e-05, "loss": 0.8567, "step": 14341 }, { "epoch": 0.01, "learning_rate": 4.99749754591533e-05, "loss": 1.132, "step": 14342 }, { "epoch": 0.01, "learning_rate": 4.997497194542727e-05, "loss": 1.0302, "step": 14343 }, { "epoch": 0.01, "learning_rate": 4.997496843145472e-05, "loss": 0.9373, "step": 14344 }, { "epoch": 0.01, "learning_rate": 4.997496491723561e-05, "loss": 1.1295, "step": 14345 }, { "epoch": 0.01, "learning_rate": 4.997496140276997e-05, "loss": 1.3824, "step": 14346 }, { "epoch": 0.01, "learning_rate": 4.9974957888057786e-05, "loss": 1.354, "step": 14347 }, { "epoch": 0.01, "learning_rate": 4.997495437309905e-05, "loss": 1.3216, "step": 14348 }, { "epoch": 0.01, "learning_rate": 4.997495085789378e-05, "loss": 1.06, "step": 14349 }, { "epoch": 0.01, "learning_rate": 4.9974947342441966e-05, "loss": 1.1986, "step": 14350 }, { "epoch": 0.01, "learning_rate": 4.997494382674361e-05, "loss": 1.0807, "step": 14351 }, { "epoch": 0.01, "learning_rate": 4.9974940310798714e-05, "loss": 1.24, "step": 14352 }, { "epoch": 0.01, "learning_rate": 4.9974936794607266e-05, "loss": 1.1952, "step": 14353 }, { "epoch": 0.01, "learning_rate": 4.997493327816929e-05, "loss": 1.5186, "step": 14354 }, { "epoch": 0.01, "learning_rate": 4.997492976148476e-05, "loss": 1.1785, "step": 14355 }, { "epoch": 0.01, "learning_rate": 4.99749262445537e-05, "loss": 0.9407, "step": 14356 }, { "epoch": 0.01, "learning_rate": 4.997492272737608e-05, "loss": 1.2702, "step": 14357 }, { "epoch": 0.01, "learning_rate": 4.9974919209951936e-05, "loss": 1.2608, "step": 14358 }, { "epoch": 0.01, "learning_rate": 4.997491569228124e-05, "loss": 1.2639, "step": 14359 }, { "epoch": 0.01, "learning_rate": 4.997491217436401e-05, "loss": 0.932, "step": 14360 }, { "epoch": 0.01, "learning_rate": 4.9974908656200236e-05, "loss": 1.3442, "step": 14361 }, { "epoch": 0.01, "learning_rate": 4.997490513778992e-05, "loss": 1.2339, "step": 14362 }, { "epoch": 0.01, "learning_rate": 4.997490161913306e-05, "loss": 1.2823, "step": 14363 }, { "epoch": 0.01, "learning_rate": 4.9974898100229655e-05, "loss": 1.1675, "step": 14364 }, { "epoch": 0.01, "learning_rate": 4.997489458107971e-05, "loss": 0.7062, "step": 14365 }, { "epoch": 0.01, "learning_rate": 4.9974891061683226e-05, "loss": 1.0909, "step": 14366 }, { "epoch": 0.01, "learning_rate": 4.997488754204021e-05, "loss": 1.2318, "step": 14367 }, { "epoch": 0.01, "learning_rate": 4.9974884022150634e-05, "loss": 1.6796, "step": 14368 }, { "epoch": 0.01, "learning_rate": 4.997488050201453e-05, "loss": 1.1237, "step": 14369 }, { "epoch": 0.01, "learning_rate": 4.997487698163188e-05, "loss": 1.7231, "step": 14370 }, { "epoch": 0.01, "learning_rate": 4.9974873461002683e-05, "loss": 0.9219, "step": 14371 }, { "epoch": 0.01, "learning_rate": 4.9974869940126954e-05, "loss": 0.7085, "step": 14372 }, { "epoch": 0.01, "learning_rate": 4.9974866419004677e-05, "loss": 1.2581, "step": 14373 }, { "epoch": 0.01, "learning_rate": 4.997486289763586e-05, "loss": 1.0116, "step": 14374 }, { "epoch": 0.01, "learning_rate": 4.99748593760205e-05, "loss": 1.3663, "step": 14375 }, { "epoch": 0.01, "learning_rate": 4.99748558541586e-05, "loss": 1.0475, "step": 14376 }, { "epoch": 0.01, "learning_rate": 4.9974852332050167e-05, "loss": 0.9837, "step": 14377 }, { "epoch": 0.01, "learning_rate": 4.9974848809695185e-05, "loss": 0.8915, "step": 14378 }, { "epoch": 0.01, "learning_rate": 4.997484528709366e-05, "loss": 1.2295, "step": 14379 }, { "epoch": 0.01, "learning_rate": 4.9974841764245594e-05, "loss": 1.2447, "step": 14380 }, { "epoch": 0.01, "learning_rate": 4.997483824115099e-05, "loss": 1.198, "step": 14381 }, { "epoch": 0.01, "learning_rate": 4.9974834717809846e-05, "loss": 1.3973, "step": 14382 }, { "epoch": 0.01, "learning_rate": 4.997483119422216e-05, "loss": 1.1056, "step": 14383 }, { "epoch": 0.01, "learning_rate": 4.997482767038793e-05, "loss": 1.1403, "step": 14384 }, { "epoch": 0.01, "learning_rate": 4.997482414630716e-05, "loss": 1.6749, "step": 14385 }, { "epoch": 0.01, "learning_rate": 4.9974820621979853e-05, "loss": 1.3194, "step": 14386 }, { "epoch": 0.01, "learning_rate": 4.9974817097406e-05, "loss": 0.8325, "step": 14387 }, { "epoch": 0.01, "learning_rate": 4.997481357258561e-05, "loss": 1.031, "step": 14388 }, { "epoch": 0.01, "learning_rate": 4.997481004751868e-05, "loss": 1.1606, "step": 14389 }, { "epoch": 0.01, "learning_rate": 4.997480652220521e-05, "loss": 0.8679, "step": 14390 }, { "epoch": 0.01, "learning_rate": 4.997480299664519e-05, "loss": 0.9806, "step": 14391 }, { "epoch": 0.01, "learning_rate": 4.997479947083864e-05, "loss": 1.0224, "step": 14392 }, { "epoch": 0.01, "learning_rate": 4.997479594478554e-05, "loss": 1.1977, "step": 14393 }, { "epoch": 0.01, "learning_rate": 4.9974792418485904e-05, "loss": 1.0153, "step": 14394 }, { "epoch": 0.01, "learning_rate": 4.997478889193973e-05, "loss": 1.1233, "step": 14395 }, { "epoch": 0.01, "learning_rate": 4.9974785365147013e-05, "loss": 0.9666, "step": 14396 }, { "epoch": 0.01, "learning_rate": 4.997478183810776e-05, "loss": 1.2151, "step": 14397 }, { "epoch": 0.01, "learning_rate": 4.997477831082196e-05, "loss": 1.1125, "step": 14398 }, { "epoch": 0.01, "learning_rate": 4.9974774783289615e-05, "loss": 1.289, "step": 14399 }, { "epoch": 0.01, "learning_rate": 4.997477125551073e-05, "loss": 1.0788, "step": 14400 }, { "epoch": 0.01, "learning_rate": 4.997476772748532e-05, "loss": 0.9865, "step": 14401 }, { "epoch": 0.01, "learning_rate": 4.997476419921335e-05, "loss": 0.8677, "step": 14402 }, { "epoch": 0.01, "learning_rate": 4.9974760670694855e-05, "loss": 1.1269, "step": 14403 }, { "epoch": 0.01, "learning_rate": 4.997475714192982e-05, "loss": 1.0795, "step": 14404 }, { "epoch": 0.01, "learning_rate": 4.997475361291824e-05, "loss": 1.1291, "step": 14405 }, { "epoch": 0.01, "learning_rate": 4.9974750083660114e-05, "loss": 1.2704, "step": 14406 }, { "epoch": 0.01, "learning_rate": 4.997474655415545e-05, "loss": 1.1482, "step": 14407 }, { "epoch": 0.01, "learning_rate": 4.9974743024404244e-05, "loss": 1.0979, "step": 14408 }, { "epoch": 0.01, "learning_rate": 4.9974739494406505e-05, "loss": 0.8416, "step": 14409 }, { "epoch": 0.01, "learning_rate": 4.9974735964162225e-05, "loss": 0.921, "step": 14410 }, { "epoch": 0.01, "learning_rate": 4.99747324336714e-05, "loss": 1.1402, "step": 14411 }, { "epoch": 0.01, "learning_rate": 4.997472890293404e-05, "loss": 1.2226, "step": 14412 }, { "epoch": 0.01, "learning_rate": 4.997472537195014e-05, "loss": 1.346, "step": 14413 }, { "epoch": 0.01, "learning_rate": 4.99747218407197e-05, "loss": 0.9469, "step": 14414 }, { "epoch": 0.01, "learning_rate": 4.997471830924272e-05, "loss": 1.1293, "step": 14415 }, { "epoch": 0.01, "learning_rate": 4.9974714777519196e-05, "loss": 0.9046, "step": 14416 }, { "epoch": 0.01, "learning_rate": 4.997471124554913e-05, "loss": 0.8587, "step": 14417 }, { "epoch": 0.01, "learning_rate": 4.997470771333253e-05, "loss": 1.1661, "step": 14418 }, { "epoch": 0.01, "learning_rate": 4.997470418086939e-05, "loss": 0.9759, "step": 14419 }, { "epoch": 0.01, "learning_rate": 4.997470064815971e-05, "loss": 1.3513, "step": 14420 }, { "epoch": 0.01, "learning_rate": 4.997469711520349e-05, "loss": 0.7998, "step": 14421 }, { "epoch": 0.01, "learning_rate": 4.997469358200073e-05, "loss": 0.4826, "step": 14422 }, { "epoch": 0.01, "learning_rate": 4.9974690048551434e-05, "loss": 0.681, "step": 14423 }, { "epoch": 0.01, "learning_rate": 4.997468651485558e-05, "loss": 0.6375, "step": 14424 }, { "epoch": 0.01, "learning_rate": 4.997468298091321e-05, "loss": 0.532, "step": 14425 }, { "epoch": 0.01, "learning_rate": 4.997467944672429e-05, "loss": 0.3822, "step": 14426 }, { "epoch": 0.01, "learning_rate": 4.997467591228883e-05, "loss": 0.2592, "step": 14427 }, { "epoch": 0.01, "learning_rate": 4.997467237760683e-05, "loss": 0.264, "step": 14428 }, { "epoch": 0.01, "learning_rate": 4.99746688426783e-05, "loss": 0.1805, "step": 14429 }, { "epoch": 0.01, "learning_rate": 4.997466530750322e-05, "loss": 1.1733, "step": 14430 }, { "epoch": 0.01, "learning_rate": 4.9974661772081605e-05, "loss": 1.3639, "step": 14431 }, { "epoch": 0.01, "learning_rate": 4.9974658236413455e-05, "loss": 1.1364, "step": 14432 }, { "epoch": 0.01, "learning_rate": 4.9974654700498757e-05, "loss": 0.8945, "step": 14433 }, { "epoch": 0.01, "learning_rate": 4.997465116433752e-05, "loss": 0.6625, "step": 14434 }, { "epoch": 0.01, "learning_rate": 4.997464762792975e-05, "loss": 0.7756, "step": 14435 }, { "epoch": 0.01, "learning_rate": 4.997464409127544e-05, "loss": 1.1861, "step": 14436 }, { "epoch": 0.01, "learning_rate": 4.9974640554374584e-05, "loss": 1.0582, "step": 14437 }, { "epoch": 0.01, "learning_rate": 4.997463701722719e-05, "loss": 1.1725, "step": 14438 }, { "epoch": 0.01, "learning_rate": 4.9974633479833266e-05, "loss": 1.078, "step": 14439 }, { "epoch": 0.01, "learning_rate": 4.9974629942192796e-05, "loss": 0.9228, "step": 14440 }, { "epoch": 0.01, "learning_rate": 4.9974626404305785e-05, "loss": 0.5655, "step": 14441 }, { "epoch": 0.01, "learning_rate": 4.997462286617224e-05, "loss": 0.4891, "step": 14442 }, { "epoch": 0.01, "learning_rate": 4.997461932779215e-05, "loss": 0.484, "step": 14443 }, { "epoch": 0.01, "learning_rate": 4.997461578916552e-05, "loss": 0.9818, "step": 14444 }, { "epoch": 0.01, "learning_rate": 4.997461225029236e-05, "loss": 1.0619, "step": 14445 }, { "epoch": 0.01, "learning_rate": 4.997460871117267e-05, "loss": 1.129, "step": 14446 }, { "epoch": 0.01, "learning_rate": 4.997460517180642e-05, "loss": 1.2111, "step": 14447 }, { "epoch": 0.01, "learning_rate": 4.997460163219364e-05, "loss": 0.4887, "step": 14448 }, { "epoch": 0.01, "learning_rate": 4.9974598092334325e-05, "loss": 1.174, "step": 14449 }, { "epoch": 0.01, "learning_rate": 4.997459455222847e-05, "loss": 1.7233, "step": 14450 }, { "epoch": 0.01, "learning_rate": 4.997459101187607e-05, "loss": 1.5597, "step": 14451 }, { "epoch": 0.01, "learning_rate": 4.9974587471277137e-05, "loss": 0.8462, "step": 14452 }, { "epoch": 0.01, "learning_rate": 4.997458393043166e-05, "loss": 1.3633, "step": 14453 }, { "epoch": 0.01, "learning_rate": 4.997458038933965e-05, "loss": 0.6299, "step": 14454 }, { "epoch": 0.01, "learning_rate": 4.99745768480011e-05, "loss": 0.1442, "step": 14455 }, { "epoch": 0.01, "learning_rate": 4.9974573306416005e-05, "loss": 0.9584, "step": 14456 }, { "epoch": 0.01, "learning_rate": 4.9974569764584376e-05, "loss": 1.3043, "step": 14457 }, { "epoch": 0.01, "learning_rate": 4.997456622250621e-05, "loss": 0.9104, "step": 14458 }, { "epoch": 0.01, "learning_rate": 4.9974562680181514e-05, "loss": 1.1106, "step": 14459 }, { "epoch": 0.01, "learning_rate": 4.997455913761027e-05, "loss": 1.1653, "step": 14460 }, { "epoch": 0.01, "learning_rate": 4.997455559479249e-05, "loss": 0.9697, "step": 14461 }, { "epoch": 0.01, "learning_rate": 4.997455205172817e-05, "loss": 1.2126, "step": 14462 }, { "epoch": 0.01, "learning_rate": 4.997454850841731e-05, "loss": 0.9386, "step": 14463 }, { "epoch": 0.01, "learning_rate": 4.9974544964859914e-05, "loss": 1.1982, "step": 14464 }, { "epoch": 0.01, "learning_rate": 4.9974541421055985e-05, "loss": 1.0994, "step": 14465 }, { "epoch": 0.01, "learning_rate": 4.9974537877005515e-05, "loss": 1.2372, "step": 14466 }, { "epoch": 0.01, "learning_rate": 4.99745343327085e-05, "loss": 1.0711, "step": 14467 }, { "epoch": 0.01, "learning_rate": 4.997453078816495e-05, "loss": 1.279, "step": 14468 }, { "epoch": 0.01, "learning_rate": 4.997452724337486e-05, "loss": 1.1346, "step": 14469 }, { "epoch": 0.01, "learning_rate": 4.997452369833824e-05, "loss": 1.0648, "step": 14470 }, { "epoch": 0.01, "learning_rate": 4.997452015305508e-05, "loss": 1.367, "step": 14471 }, { "epoch": 0.01, "learning_rate": 4.997451660752538e-05, "loss": 1.1264, "step": 14472 }, { "epoch": 0.01, "learning_rate": 4.997451306174914e-05, "loss": 1.1409, "step": 14473 }, { "epoch": 0.01, "learning_rate": 4.997450951572636e-05, "loss": 0.9711, "step": 14474 }, { "epoch": 0.01, "learning_rate": 4.997450596945705e-05, "loss": 0.9547, "step": 14475 }, { "epoch": 0.01, "learning_rate": 4.99745024229412e-05, "loss": 1.4691, "step": 14476 }, { "epoch": 0.01, "learning_rate": 4.997449887617881e-05, "loss": 1.1123, "step": 14477 }, { "epoch": 0.01, "learning_rate": 4.997449532916988e-05, "loss": 1.3364, "step": 14478 }, { "epoch": 0.01, "learning_rate": 4.997449178191442e-05, "loss": 1.0358, "step": 14479 }, { "epoch": 0.01, "learning_rate": 4.997448823441241e-05, "loss": 1.5155, "step": 14480 }, { "epoch": 0.01, "learning_rate": 4.997448468666388e-05, "loss": 1.4649, "step": 14481 }, { "epoch": 0.01, "learning_rate": 4.9974481138668796e-05, "loss": 1.2042, "step": 14482 }, { "epoch": 0.01, "learning_rate": 4.997447759042718e-05, "loss": 0.8581, "step": 14483 }, { "epoch": 0.01, "learning_rate": 4.997447404193903e-05, "loss": 1.2848, "step": 14484 }, { "epoch": 0.01, "learning_rate": 4.997447049320434e-05, "loss": 0.6986, "step": 14485 }, { "epoch": 0.01, "learning_rate": 4.997446694422311e-05, "loss": 0.9755, "step": 14486 }, { "epoch": 0.01, "learning_rate": 4.997446339499535e-05, "loss": 1.2767, "step": 14487 }, { "epoch": 0.01, "learning_rate": 4.997445984552104e-05, "loss": 1.1844, "step": 14488 }, { "epoch": 0.01, "learning_rate": 4.99744562958002e-05, "loss": 1.7483, "step": 14489 }, { "epoch": 0.01, "learning_rate": 4.997445274583282e-05, "loss": 1.4285, "step": 14490 }, { "epoch": 0.01, "learning_rate": 4.997444919561891e-05, "loss": 1.0301, "step": 14491 }, { "epoch": 0.01, "learning_rate": 4.9974445645158464e-05, "loss": 1.2763, "step": 14492 }, { "epoch": 0.01, "learning_rate": 4.997444209445147e-05, "loss": 1.1808, "step": 14493 }, { "epoch": 0.01, "learning_rate": 4.997443854349795e-05, "loss": 1.2185, "step": 14494 }, { "epoch": 0.01, "learning_rate": 4.997443499229788e-05, "loss": 1.1837, "step": 14495 }, { "epoch": 0.01, "learning_rate": 4.997443144085128e-05, "loss": 0.8031, "step": 14496 }, { "epoch": 0.01, "learning_rate": 4.9974427889158146e-05, "loss": 1.2143, "step": 14497 }, { "epoch": 0.01, "learning_rate": 4.997442433721847e-05, "loss": 1.1006, "step": 14498 }, { "epoch": 0.01, "learning_rate": 4.997442078503226e-05, "loss": 1.2395, "step": 14499 }, { "epoch": 0.01, "learning_rate": 4.997441723259951e-05, "loss": 1.1118, "step": 14500 }, { "epoch": 0.01, "eval_loss": 1.032680630683899, "eval_runtime": 83.9404, "eval_samples_per_second": 16.5, "eval_steps_per_second": 4.134, "step": 14500 }, { "epoch": 0.01, "learning_rate": 4.997441367992023e-05, "loss": 1.2032, "step": 14501 }, { "epoch": 0.01, "learning_rate": 4.9974410126994404e-05, "loss": 0.9744, "step": 14502 }, { "epoch": 0.01, "learning_rate": 4.997440657382204e-05, "loss": 1.0883, "step": 14503 }, { "epoch": 0.01, "learning_rate": 4.997440302040315e-05, "loss": 1.1888, "step": 14504 }, { "epoch": 0.01, "learning_rate": 4.997439946673772e-05, "loss": 1.1701, "step": 14505 }, { "epoch": 0.01, "learning_rate": 4.9974395912825744e-05, "loss": 1.5214, "step": 14506 }, { "epoch": 0.01, "learning_rate": 4.9974392358667243e-05, "loss": 1.174, "step": 14507 }, { "epoch": 0.01, "learning_rate": 4.9974388804262196e-05, "loss": 1.1367, "step": 14508 }, { "epoch": 0.01, "learning_rate": 4.997438524961062e-05, "loss": 0.7751, "step": 14509 }, { "epoch": 0.01, "learning_rate": 4.9974381694712505e-05, "loss": 1.1393, "step": 14510 }, { "epoch": 0.01, "learning_rate": 4.997437813956785e-05, "loss": 1.1476, "step": 14511 }, { "epoch": 0.01, "learning_rate": 4.9974374584176665e-05, "loss": 1.2118, "step": 14512 }, { "epoch": 0.01, "learning_rate": 4.997437102853893e-05, "loss": 1.3143, "step": 14513 }, { "epoch": 0.01, "learning_rate": 4.997436747265467e-05, "loss": 1.0653, "step": 14514 }, { "epoch": 0.01, "learning_rate": 4.9974363916523875e-05, "loss": 1.2056, "step": 14515 }, { "epoch": 0.01, "learning_rate": 4.997436036014654e-05, "loss": 0.8385, "step": 14516 }, { "epoch": 0.01, "learning_rate": 4.997435680352267e-05, "loss": 1.083, "step": 14517 }, { "epoch": 0.01, "learning_rate": 4.997435324665226e-05, "loss": 1.1865, "step": 14518 }, { "epoch": 0.01, "learning_rate": 4.997434968953532e-05, "loss": 1.0839, "step": 14519 }, { "epoch": 0.01, "learning_rate": 4.9974346132171834e-05, "loss": 1.1288, "step": 14520 }, { "epoch": 0.01, "learning_rate": 4.997434257456182e-05, "loss": 1.1401, "step": 14521 }, { "epoch": 0.01, "learning_rate": 4.9974339016705265e-05, "loss": 0.758, "step": 14522 }, { "epoch": 0.01, "learning_rate": 4.997433545860219e-05, "loss": 1.0797, "step": 14523 }, { "epoch": 0.01, "learning_rate": 4.997433190025256e-05, "loss": 1.3489, "step": 14524 }, { "epoch": 0.01, "learning_rate": 4.9974328341656394e-05, "loss": 1.1971, "step": 14525 }, { "epoch": 0.01, "learning_rate": 4.99743247828137e-05, "loss": 0.9595, "step": 14526 }, { "epoch": 0.01, "learning_rate": 4.9974321223724465e-05, "loss": 0.982, "step": 14527 }, { "epoch": 0.01, "learning_rate": 4.997431766438869e-05, "loss": 1.2661, "step": 14528 }, { "epoch": 0.01, "learning_rate": 4.9974314104806394e-05, "loss": 0.7192, "step": 14529 }, { "epoch": 0.01, "learning_rate": 4.997431054497756e-05, "loss": 1.1649, "step": 14530 }, { "epoch": 0.01, "learning_rate": 4.997430698490217e-05, "loss": 1.2051, "step": 14531 }, { "epoch": 0.01, "learning_rate": 4.997430342458026e-05, "loss": 1.1533, "step": 14532 }, { "epoch": 0.01, "learning_rate": 4.9974299864011815e-05, "loss": 0.7428, "step": 14533 }, { "epoch": 0.01, "learning_rate": 4.9974296303196835e-05, "loss": 0.5662, "step": 14534 }, { "epoch": 0.01, "learning_rate": 4.997429274213531e-05, "loss": 1.1641, "step": 14535 }, { "epoch": 0.01, "learning_rate": 4.9974289180827253e-05, "loss": 1.0921, "step": 14536 }, { "epoch": 0.01, "learning_rate": 4.9974285619272665e-05, "loss": 0.9288, "step": 14537 }, { "epoch": 0.01, "learning_rate": 4.9974282057471535e-05, "loss": 1.2262, "step": 14538 }, { "epoch": 0.01, "learning_rate": 4.997427849542388e-05, "loss": 0.985, "step": 14539 }, { "epoch": 0.01, "learning_rate": 4.9974274933129675e-05, "loss": 1.1805, "step": 14540 }, { "epoch": 0.01, "learning_rate": 4.997427137058894e-05, "loss": 1.1029, "step": 14541 }, { "epoch": 0.01, "learning_rate": 4.997426780780168e-05, "loss": 0.5841, "step": 14542 }, { "epoch": 0.01, "learning_rate": 4.997426424476787e-05, "loss": 0.9382, "step": 14543 }, { "epoch": 0.01, "learning_rate": 4.997426068148753e-05, "loss": 0.9929, "step": 14544 }, { "epoch": 0.01, "learning_rate": 4.997425711796066e-05, "loss": 1.0577, "step": 14545 }, { "epoch": 0.01, "learning_rate": 4.997425355418724e-05, "loss": 1.0238, "step": 14546 }, { "epoch": 0.01, "learning_rate": 4.99742499901673e-05, "loss": 1.0698, "step": 14547 }, { "epoch": 0.01, "learning_rate": 4.9974246425900816e-05, "loss": 1.0617, "step": 14548 }, { "epoch": 0.01, "learning_rate": 4.99742428613878e-05, "loss": 0.9446, "step": 14549 }, { "epoch": 0.01, "learning_rate": 4.997423929662825e-05, "loss": 0.715, "step": 14550 }, { "epoch": 0.01, "learning_rate": 4.997423573162216e-05, "loss": 0.6863, "step": 14551 }, { "epoch": 0.01, "learning_rate": 4.997423216636954e-05, "loss": 1.0478, "step": 14552 }, { "epoch": 0.01, "learning_rate": 4.997422860087039e-05, "loss": 1.3427, "step": 14553 }, { "epoch": 0.01, "learning_rate": 4.9974225035124696e-05, "loss": 1.3418, "step": 14554 }, { "epoch": 0.01, "learning_rate": 4.997422146913247e-05, "loss": 1.3523, "step": 14555 }, { "epoch": 0.01, "learning_rate": 4.9974217902893706e-05, "loss": 1.4194, "step": 14556 }, { "epoch": 0.01, "learning_rate": 4.997421433640841e-05, "loss": 1.1343, "step": 14557 }, { "epoch": 0.01, "learning_rate": 4.997421076967658e-05, "loss": 1.0051, "step": 14558 }, { "epoch": 0.01, "learning_rate": 4.997420720269821e-05, "loss": 1.3802, "step": 14559 }, { "epoch": 0.01, "learning_rate": 4.9974203635473304e-05, "loss": 1.1661, "step": 14560 }, { "epoch": 0.01, "learning_rate": 4.997420006800187e-05, "loss": 1.1622, "step": 14561 }, { "epoch": 0.01, "learning_rate": 4.99741965002839e-05, "loss": 1.3033, "step": 14562 }, { "epoch": 0.01, "learning_rate": 4.99741929323194e-05, "loss": 0.9048, "step": 14563 }, { "epoch": 0.01, "learning_rate": 4.997418936410835e-05, "loss": 2.2987, "step": 14564 }, { "epoch": 0.01, "learning_rate": 4.9974185795650776e-05, "loss": 1.6775, "step": 14565 }, { "epoch": 0.01, "learning_rate": 4.997418222694667e-05, "loss": 0.9796, "step": 14566 }, { "epoch": 0.01, "learning_rate": 4.9974178657996024e-05, "loss": 1.2108, "step": 14567 }, { "epoch": 0.01, "learning_rate": 4.9974175088798846e-05, "loss": 0.3747, "step": 14568 }, { "epoch": 0.01, "learning_rate": 4.997417151935513e-05, "loss": 1.1472, "step": 14569 }, { "epoch": 0.01, "learning_rate": 4.997416794966488e-05, "loss": 1.1262, "step": 14570 }, { "epoch": 0.01, "learning_rate": 4.99741643797281e-05, "loss": 0.9259, "step": 14571 }, { "epoch": 0.01, "learning_rate": 4.9974160809544784e-05, "loss": 0.2318, "step": 14572 }, { "epoch": 0.01, "learning_rate": 4.997415723911493e-05, "loss": 0.7396, "step": 14573 }, { "epoch": 0.01, "learning_rate": 4.997415366843855e-05, "loss": 1.0218, "step": 14574 }, { "epoch": 0.01, "learning_rate": 4.997415009751563e-05, "loss": 1.2151, "step": 14575 }, { "epoch": 0.01, "learning_rate": 4.997414652634617e-05, "loss": 1.0314, "step": 14576 }, { "epoch": 0.01, "learning_rate": 4.997414295493018e-05, "loss": 1.1444, "step": 14577 }, { "epoch": 0.01, "learning_rate": 4.9974139383267657e-05, "loss": 1.1257, "step": 14578 }, { "epoch": 0.01, "learning_rate": 4.9974135811358605e-05, "loss": 1.1311, "step": 14579 }, { "epoch": 0.01, "learning_rate": 4.9974132239203006e-05, "loss": 1.2159, "step": 14580 }, { "epoch": 0.01, "learning_rate": 4.997412866680089e-05, "loss": 1.1511, "step": 14581 }, { "epoch": 0.01, "learning_rate": 4.997412509415223e-05, "loss": 1.0625, "step": 14582 }, { "epoch": 0.01, "learning_rate": 4.997412152125703e-05, "loss": 1.8244, "step": 14583 }, { "epoch": 0.01, "learning_rate": 4.9974117948115304e-05, "loss": 1.1516, "step": 14584 }, { "epoch": 0.01, "learning_rate": 4.997411437472704e-05, "loss": 0.9846, "step": 14585 }, { "epoch": 0.01, "learning_rate": 4.997411080109225e-05, "loss": 0.9495, "step": 14586 }, { "epoch": 0.01, "learning_rate": 4.997410722721092e-05, "loss": 1.0236, "step": 14587 }, { "epoch": 0.01, "learning_rate": 4.997410365308306e-05, "loss": 1.0757, "step": 14588 }, { "epoch": 0.01, "learning_rate": 4.9974100078708666e-05, "loss": 1.1619, "step": 14589 }, { "epoch": 0.01, "learning_rate": 4.9974096504087734e-05, "loss": 0.9697, "step": 14590 }, { "epoch": 0.01, "learning_rate": 4.997409292922027e-05, "loss": 1.085, "step": 14591 }, { "epoch": 0.01, "learning_rate": 4.997408935410627e-05, "loss": 1.4168, "step": 14592 }, { "epoch": 0.01, "learning_rate": 4.997408577874574e-05, "loss": 1.0358, "step": 14593 }, { "epoch": 0.01, "learning_rate": 4.997408220313867e-05, "loss": 0.9766, "step": 14594 }, { "epoch": 0.01, "learning_rate": 4.997407862728507e-05, "loss": 1.0095, "step": 14595 }, { "epoch": 0.01, "learning_rate": 4.9974075051184945e-05, "loss": 1.1314, "step": 14596 }, { "epoch": 0.01, "learning_rate": 4.9974071474838275e-05, "loss": 1.4483, "step": 14597 }, { "epoch": 0.01, "learning_rate": 4.997406789824508e-05, "loss": 0.9681, "step": 14598 }, { "epoch": 0.01, "learning_rate": 4.997406432140534e-05, "loss": 1.0439, "step": 14599 }, { "epoch": 0.01, "learning_rate": 4.997406074431907e-05, "loss": 1.0032, "step": 14600 }, { "epoch": 0.01, "learning_rate": 4.997405716698628e-05, "loss": 1.1184, "step": 14601 }, { "epoch": 0.01, "learning_rate": 4.9974053589406946e-05, "loss": 1.1359, "step": 14602 }, { "epoch": 0.01, "learning_rate": 4.997405001158108e-05, "loss": 1.0905, "step": 14603 }, { "epoch": 0.01, "learning_rate": 4.997404643350867e-05, "loss": 1.1548, "step": 14604 }, { "epoch": 0.01, "learning_rate": 4.997404285518974e-05, "loss": 0.956, "step": 14605 }, { "epoch": 0.01, "learning_rate": 4.997403927662428e-05, "loss": 1.1926, "step": 14606 }, { "epoch": 0.01, "learning_rate": 4.997403569781227e-05, "loss": 1.0824, "step": 14607 }, { "epoch": 0.01, "learning_rate": 4.997403211875374e-05, "loss": 1.0924, "step": 14608 }, { "epoch": 0.01, "learning_rate": 4.997402853944868e-05, "loss": 1.0906, "step": 14609 }, { "epoch": 0.01, "learning_rate": 4.997402495989708e-05, "loss": 1.2025, "step": 14610 }, { "epoch": 0.01, "learning_rate": 4.9974021380098944e-05, "loss": 1.3244, "step": 14611 }, { "epoch": 0.01, "learning_rate": 4.997401780005428e-05, "loss": 1.0882, "step": 14612 }, { "epoch": 0.01, "learning_rate": 4.997401421976308e-05, "loss": 1.0168, "step": 14613 }, { "epoch": 0.01, "learning_rate": 4.997401063922535e-05, "loss": 0.8935, "step": 14614 }, { "epoch": 0.01, "learning_rate": 4.9974007058441084e-05, "loss": 1.1162, "step": 14615 }, { "epoch": 0.01, "learning_rate": 4.9974003477410295e-05, "loss": 0.8198, "step": 14616 }, { "epoch": 0.01, "learning_rate": 4.9973999896132964e-05, "loss": 0.7165, "step": 14617 }, { "epoch": 0.01, "learning_rate": 4.997399631460909e-05, "loss": 1.4271, "step": 14618 }, { "epoch": 0.01, "learning_rate": 4.99739927328387e-05, "loss": 1.2403, "step": 14619 }, { "epoch": 0.01, "learning_rate": 4.9973989150821775e-05, "loss": 1.1047, "step": 14620 }, { "epoch": 0.01, "learning_rate": 4.9973985568558315e-05, "loss": 0.9785, "step": 14621 }, { "epoch": 0.01, "learning_rate": 4.997398198604832e-05, "loss": 0.9175, "step": 14622 }, { "epoch": 0.01, "learning_rate": 4.997397840329179e-05, "loss": 0.9679, "step": 14623 }, { "epoch": 0.01, "learning_rate": 4.997397482028874e-05, "loss": 1.4641, "step": 14624 }, { "epoch": 0.01, "learning_rate": 4.997397123703914e-05, "loss": 1.4326, "step": 14625 }, { "epoch": 0.01, "learning_rate": 4.997396765354302e-05, "loss": 0.6138, "step": 14626 }, { "epoch": 0.01, "learning_rate": 4.997396406980036e-05, "loss": 0.5195, "step": 14627 }, { "epoch": 0.01, "learning_rate": 4.9973960485811176e-05, "loss": 0.5505, "step": 14628 }, { "epoch": 0.01, "learning_rate": 4.997395690157546e-05, "loss": 1.2871, "step": 14629 }, { "epoch": 0.01, "learning_rate": 4.9973953317093205e-05, "loss": 1.3319, "step": 14630 }, { "epoch": 0.01, "learning_rate": 4.997394973236442e-05, "loss": 1.171, "step": 14631 }, { "epoch": 0.01, "learning_rate": 4.997394614738909e-05, "loss": 1.0855, "step": 14632 }, { "epoch": 0.01, "learning_rate": 4.997394256216725e-05, "loss": 0.9017, "step": 14633 }, { "epoch": 0.01, "learning_rate": 4.997393897669886e-05, "loss": 0.8217, "step": 14634 }, { "epoch": 0.01, "learning_rate": 4.997393539098395e-05, "loss": 0.5106, "step": 14635 }, { "epoch": 0.01, "learning_rate": 4.99739318050225e-05, "loss": 0.6605, "step": 14636 }, { "epoch": 0.01, "learning_rate": 4.9973928218814525e-05, "loss": 1.1345, "step": 14637 }, { "epoch": 0.01, "learning_rate": 4.9973924632360014e-05, "loss": 1.3363, "step": 14638 }, { "epoch": 0.01, "learning_rate": 4.997392104565897e-05, "loss": 1.2334, "step": 14639 }, { "epoch": 0.01, "learning_rate": 4.9973917458711397e-05, "loss": 1.2053, "step": 14640 }, { "epoch": 0.01, "learning_rate": 4.997391387151728e-05, "loss": 1.3028, "step": 14641 }, { "epoch": 0.01, "learning_rate": 4.997391028407665e-05, "loss": 1.1134, "step": 14642 }, { "epoch": 0.01, "learning_rate": 4.9973906696389475e-05, "loss": 1.096, "step": 14643 }, { "epoch": 0.01, "learning_rate": 4.997390310845578e-05, "loss": 1.1165, "step": 14644 }, { "epoch": 0.01, "learning_rate": 4.9973899520275544e-05, "loss": 1.0725, "step": 14645 }, { "epoch": 0.01, "learning_rate": 4.9973895931848774e-05, "loss": 0.9818, "step": 14646 }, { "epoch": 0.01, "learning_rate": 4.997389234317548e-05, "loss": 0.951, "step": 14647 }, { "epoch": 0.01, "learning_rate": 4.9973888754255646e-05, "loss": 1.2996, "step": 14648 }, { "epoch": 0.01, "learning_rate": 4.997388516508929e-05, "loss": 1.0835, "step": 14649 }, { "epoch": 0.01, "learning_rate": 4.997388157567639e-05, "loss": 1.3664, "step": 14650 }, { "epoch": 0.01, "learning_rate": 4.997387798601696e-05, "loss": 1.3241, "step": 14651 }, { "epoch": 0.01, "learning_rate": 4.997387439611101e-05, "loss": 0.752, "step": 14652 }, { "epoch": 0.01, "learning_rate": 4.997387080595852e-05, "loss": 0.9959, "step": 14653 }, { "epoch": 0.01, "learning_rate": 4.99738672155595e-05, "loss": 1.0986, "step": 14654 }, { "epoch": 0.01, "learning_rate": 4.9973863624913944e-05, "loss": 1.1846, "step": 14655 }, { "epoch": 0.01, "learning_rate": 4.997386003402187e-05, "loss": 1.1293, "step": 14656 }, { "epoch": 0.01, "learning_rate": 4.997385644288325e-05, "loss": 1.0859, "step": 14657 }, { "epoch": 0.01, "learning_rate": 4.9973852851498107e-05, "loss": 1.0269, "step": 14658 }, { "epoch": 0.01, "learning_rate": 4.997384925986643e-05, "loss": 1.2403, "step": 14659 }, { "epoch": 0.01, "learning_rate": 4.9973845667988216e-05, "loss": 0.9942, "step": 14660 }, { "epoch": 0.01, "learning_rate": 4.9973842075863483e-05, "loss": 1.0663, "step": 14661 }, { "epoch": 0.01, "learning_rate": 4.997383848349221e-05, "loss": 1.0268, "step": 14662 }, { "epoch": 0.01, "learning_rate": 4.99738348908744e-05, "loss": 1.1241, "step": 14663 }, { "epoch": 0.01, "learning_rate": 4.9973831298010074e-05, "loss": 1.0425, "step": 14664 }, { "epoch": 0.01, "learning_rate": 4.997382770489921e-05, "loss": 0.7101, "step": 14665 }, { "epoch": 0.01, "learning_rate": 4.997382411154181e-05, "loss": 0.8925, "step": 14666 }, { "epoch": 0.01, "learning_rate": 4.9973820517937886e-05, "loss": 1.0695, "step": 14667 }, { "epoch": 0.01, "learning_rate": 4.997381692408743e-05, "loss": 1.1682, "step": 14668 }, { "epoch": 0.01, "learning_rate": 4.9973813329990444e-05, "loss": 1.0653, "step": 14669 }, { "epoch": 0.01, "learning_rate": 4.9973809735646926e-05, "loss": 1.0438, "step": 14670 }, { "epoch": 0.01, "learning_rate": 4.997380614105687e-05, "loss": 0.848, "step": 14671 }, { "epoch": 0.01, "learning_rate": 4.9973802546220293e-05, "loss": 0.9973, "step": 14672 }, { "epoch": 0.01, "learning_rate": 4.997379895113718e-05, "loss": 1.1369, "step": 14673 }, { "epoch": 0.01, "learning_rate": 4.997379535580754e-05, "loss": 1.0308, "step": 14674 }, { "epoch": 0.01, "learning_rate": 4.997379176023136e-05, "loss": 1.2479, "step": 14675 }, { "epoch": 0.01, "learning_rate": 4.997378816440866e-05, "loss": 0.8342, "step": 14676 }, { "epoch": 0.01, "learning_rate": 4.997378456833942e-05, "loss": 1.2351, "step": 14677 }, { "epoch": 0.01, "learning_rate": 4.9973780972023654e-05, "loss": 1.0975, "step": 14678 }, { "epoch": 0.01, "learning_rate": 4.997377737546136e-05, "loss": 1.1363, "step": 14679 }, { "epoch": 0.01, "learning_rate": 4.997377377865253e-05, "loss": 0.975, "step": 14680 }, { "epoch": 0.01, "learning_rate": 4.997377018159718e-05, "loss": 0.724, "step": 14681 }, { "epoch": 0.01, "learning_rate": 4.9973766584295287e-05, "loss": 0.3843, "step": 14682 }, { "epoch": 0.01, "learning_rate": 4.9973762986746866e-05, "loss": 0.3052, "step": 14683 }, { "epoch": 0.01, "learning_rate": 4.997375938895192e-05, "loss": 0.3796, "step": 14684 }, { "epoch": 0.01, "learning_rate": 4.997375579091044e-05, "loss": 0.3063, "step": 14685 }, { "epoch": 0.01, "learning_rate": 4.997375219262243e-05, "loss": 0.8041, "step": 14686 }, { "epoch": 0.01, "learning_rate": 4.9973748594087886e-05, "loss": 1.13, "step": 14687 }, { "epoch": 0.01, "learning_rate": 4.9973744995306816e-05, "loss": 1.1825, "step": 14688 }, { "epoch": 0.01, "learning_rate": 4.997374139627922e-05, "loss": 0.9396, "step": 14689 }, { "epoch": 0.01, "learning_rate": 4.997373779700508e-05, "loss": 1.1783, "step": 14690 }, { "epoch": 0.01, "learning_rate": 4.997373419748442e-05, "loss": 1.1474, "step": 14691 }, { "epoch": 0.01, "learning_rate": 4.997373059771723e-05, "loss": 1.1786, "step": 14692 }, { "epoch": 0.01, "learning_rate": 4.997372699770351e-05, "loss": 1.0563, "step": 14693 }, { "epoch": 0.01, "learning_rate": 4.997372339744326e-05, "loss": 0.9318, "step": 14694 }, { "epoch": 0.01, "learning_rate": 4.997371979693648e-05, "loss": 0.858, "step": 14695 }, { "epoch": 0.01, "learning_rate": 4.997371619618316e-05, "loss": 0.8403, "step": 14696 }, { "epoch": 0.01, "learning_rate": 4.997371259518332e-05, "loss": 0.8338, "step": 14697 }, { "epoch": 0.01, "learning_rate": 4.997370899393695e-05, "loss": 1.2301, "step": 14698 }, { "epoch": 0.01, "learning_rate": 4.9973705392444045e-05, "loss": 1.0208, "step": 14699 }, { "epoch": 0.01, "learning_rate": 4.997370179070462e-05, "loss": 1.033, "step": 14700 }, { "epoch": 0.01, "learning_rate": 4.9973698188718656e-05, "loss": 0.8073, "step": 14701 }, { "epoch": 0.01, "learning_rate": 4.9973694586486164e-05, "loss": 1.0833, "step": 14702 }, { "epoch": 0.01, "learning_rate": 4.9973690984007145e-05, "loss": 1.1689, "step": 14703 }, { "epoch": 0.01, "learning_rate": 4.997368738128159e-05, "loss": 1.225, "step": 14704 }, { "epoch": 0.01, "learning_rate": 4.997368377830951e-05, "loss": 1.1056, "step": 14705 }, { "epoch": 0.01, "learning_rate": 4.9973680175090896e-05, "loss": 0.8393, "step": 14706 }, { "epoch": 0.01, "learning_rate": 4.997367657162576e-05, "loss": 0.91, "step": 14707 }, { "epoch": 0.01, "learning_rate": 4.9973672967914085e-05, "loss": 0.9734, "step": 14708 }, { "epoch": 0.01, "learning_rate": 4.997366936395589e-05, "loss": 1.1687, "step": 14709 }, { "epoch": 0.01, "learning_rate": 4.9973665759751166e-05, "loss": 0.9984, "step": 14710 }, { "epoch": 0.01, "learning_rate": 4.997366215529991e-05, "loss": 1.1054, "step": 14711 }, { "epoch": 0.01, "learning_rate": 4.997365855060212e-05, "loss": 1.0267, "step": 14712 }, { "epoch": 0.01, "learning_rate": 4.99736549456578e-05, "loss": 1.1135, "step": 14713 }, { "epoch": 0.01, "learning_rate": 4.997365134046695e-05, "loss": 0.9772, "step": 14714 }, { "epoch": 0.01, "learning_rate": 4.997364773502958e-05, "loss": 1.2222, "step": 14715 }, { "epoch": 0.01, "learning_rate": 4.997364412934567e-05, "loss": 0.9664, "step": 14716 }, { "epoch": 0.01, "learning_rate": 4.9973640523415236e-05, "loss": 0.8003, "step": 14717 }, { "epoch": 0.01, "learning_rate": 4.9973636917238274e-05, "loss": 0.6631, "step": 14718 }, { "epoch": 0.01, "learning_rate": 4.9973633310814785e-05, "loss": 0.7529, "step": 14719 }, { "epoch": 0.01, "learning_rate": 4.997362970414476e-05, "loss": 0.9658, "step": 14720 }, { "epoch": 0.01, "learning_rate": 4.997362609722821e-05, "loss": 1.2275, "step": 14721 }, { "epoch": 0.01, "learning_rate": 4.997362249006513e-05, "loss": 1.1193, "step": 14722 }, { "epoch": 0.01, "learning_rate": 4.9973618882655515e-05, "loss": 1.2217, "step": 14723 }, { "epoch": 0.01, "learning_rate": 4.9973615274999377e-05, "loss": 1.0647, "step": 14724 }, { "epoch": 0.01, "learning_rate": 4.997361166709671e-05, "loss": 0.6829, "step": 14725 }, { "epoch": 0.01, "learning_rate": 4.997360805894751e-05, "loss": 1.0552, "step": 14726 }, { "epoch": 0.01, "learning_rate": 4.997360445055179e-05, "loss": 1.0175, "step": 14727 }, { "epoch": 0.01, "learning_rate": 4.997360084190953e-05, "loss": 1.0626, "step": 14728 }, { "epoch": 0.01, "learning_rate": 4.997359723302075e-05, "loss": 0.8943, "step": 14729 }, { "epoch": 0.01, "learning_rate": 4.997359362388544e-05, "loss": 1.1485, "step": 14730 }, { "epoch": 0.01, "learning_rate": 4.9973590014503594e-05, "loss": 1.4467, "step": 14731 }, { "epoch": 0.01, "learning_rate": 4.997358640487523e-05, "loss": 0.9085, "step": 14732 }, { "epoch": 0.01, "learning_rate": 4.9973582795000326e-05, "loss": 1.0796, "step": 14733 }, { "epoch": 0.01, "learning_rate": 4.997357918487891e-05, "loss": 1.1598, "step": 14734 }, { "epoch": 0.01, "learning_rate": 4.997357557451094e-05, "loss": 1.1493, "step": 14735 }, { "epoch": 0.01, "learning_rate": 4.9973571963896456e-05, "loss": 1.0093, "step": 14736 }, { "epoch": 0.01, "learning_rate": 4.997356835303545e-05, "loss": 1.0047, "step": 14737 }, { "epoch": 0.01, "learning_rate": 4.99735647419279e-05, "loss": 1.1759, "step": 14738 }, { "epoch": 0.01, "learning_rate": 4.9973561130573834e-05, "loss": 1.0909, "step": 14739 }, { "epoch": 0.01, "learning_rate": 4.997355751897323e-05, "loss": 0.8766, "step": 14740 }, { "epoch": 0.01, "learning_rate": 4.9973553907126117e-05, "loss": 1.4397, "step": 14741 }, { "epoch": 0.01, "learning_rate": 4.9973550295032453e-05, "loss": 0.9025, "step": 14742 }, { "epoch": 0.01, "learning_rate": 4.9973546682692277e-05, "loss": 1.2771, "step": 14743 }, { "epoch": 0.01, "learning_rate": 4.997354307010556e-05, "loss": 1.2829, "step": 14744 }, { "epoch": 0.01, "learning_rate": 4.997353945727232e-05, "loss": 0.755, "step": 14745 }, { "epoch": 0.01, "learning_rate": 4.9973535844192555e-05, "loss": 1.2094, "step": 14746 }, { "epoch": 0.01, "learning_rate": 4.9973532230866256e-05, "loss": 1.1977, "step": 14747 }, { "epoch": 0.01, "learning_rate": 4.997352861729343e-05, "loss": 1.0674, "step": 14748 }, { "epoch": 0.01, "learning_rate": 4.9973525003474075e-05, "loss": 1.1, "step": 14749 }, { "epoch": 0.01, "learning_rate": 4.9973521389408194e-05, "loss": 1.5613, "step": 14750 }, { "epoch": 0.01, "learning_rate": 4.9973517775095786e-05, "loss": 1.5573, "step": 14751 }, { "epoch": 0.01, "learning_rate": 4.997351416053685e-05, "loss": 1.4173, "step": 14752 }, { "epoch": 0.01, "learning_rate": 4.997351054573138e-05, "loss": 1.2348, "step": 14753 }, { "epoch": 0.01, "learning_rate": 4.997350693067939e-05, "loss": 1.2222, "step": 14754 }, { "epoch": 0.01, "learning_rate": 4.9973503315380866e-05, "loss": 1.1506, "step": 14755 }, { "epoch": 0.01, "learning_rate": 4.997349969983582e-05, "loss": 0.746, "step": 14756 }, { "epoch": 0.01, "learning_rate": 4.997349608404424e-05, "loss": 0.935, "step": 14757 }, { "epoch": 0.01, "learning_rate": 4.997349246800613e-05, "loss": 1.1727, "step": 14758 }, { "epoch": 0.01, "learning_rate": 4.9973488851721504e-05, "loss": 1.2878, "step": 14759 }, { "epoch": 0.01, "learning_rate": 4.9973485235190344e-05, "loss": 1.2053, "step": 14760 }, { "epoch": 0.01, "learning_rate": 4.9973481618412656e-05, "loss": 1.1233, "step": 14761 }, { "epoch": 0.01, "learning_rate": 4.997347800138844e-05, "loss": 1.3196, "step": 14762 }, { "epoch": 0.01, "learning_rate": 4.99734743841177e-05, "loss": 1.0109, "step": 14763 }, { "epoch": 0.01, "learning_rate": 4.997347076660043e-05, "loss": 1.2367, "step": 14764 }, { "epoch": 0.01, "learning_rate": 4.9973467148836626e-05, "loss": 1.101, "step": 14765 }, { "epoch": 0.01, "learning_rate": 4.99734635308263e-05, "loss": 1.1594, "step": 14766 }, { "epoch": 0.01, "learning_rate": 4.997345991256945e-05, "loss": 1.2238, "step": 14767 }, { "epoch": 0.01, "learning_rate": 4.9973456294066066e-05, "loss": 1.1648, "step": 14768 }, { "epoch": 0.01, "learning_rate": 4.9973452675316154e-05, "loss": 1.1207, "step": 14769 }, { "epoch": 0.01, "learning_rate": 4.997344905631972e-05, "loss": 0.9738, "step": 14770 }, { "epoch": 0.01, "learning_rate": 4.997344543707676e-05, "loss": 1.0473, "step": 14771 }, { "epoch": 0.01, "learning_rate": 4.997344181758726e-05, "loss": 0.9454, "step": 14772 }, { "epoch": 0.01, "learning_rate": 4.997343819785125e-05, "loss": 0.8451, "step": 14773 }, { "epoch": 0.01, "learning_rate": 4.9973434577868704e-05, "loss": 1.2283, "step": 14774 }, { "epoch": 0.01, "learning_rate": 4.9973430957639635e-05, "loss": 0.9389, "step": 14775 }, { "epoch": 0.01, "learning_rate": 4.997342733716403e-05, "loss": 1.1241, "step": 14776 }, { "epoch": 0.01, "learning_rate": 4.99734237164419e-05, "loss": 0.9569, "step": 14777 }, { "epoch": 0.01, "learning_rate": 4.997342009547325e-05, "loss": 1.0957, "step": 14778 }, { "epoch": 0.01, "learning_rate": 4.997341647425807e-05, "loss": 0.6861, "step": 14779 }, { "epoch": 0.01, "learning_rate": 4.997341285279637e-05, "loss": 1.0632, "step": 14780 }, { "epoch": 0.01, "learning_rate": 4.997340923108813e-05, "loss": 1.0148, "step": 14781 }, { "epoch": 0.01, "learning_rate": 4.997340560913336e-05, "loss": 1.6072, "step": 14782 }, { "epoch": 0.01, "learning_rate": 4.997340198693208e-05, "loss": 1.3777, "step": 14783 }, { "epoch": 0.01, "learning_rate": 4.997339836448426e-05, "loss": 1.1734, "step": 14784 }, { "epoch": 0.01, "learning_rate": 4.997339474178991e-05, "loss": 0.4705, "step": 14785 }, { "epoch": 0.01, "learning_rate": 4.997339111884904e-05, "loss": 1.7411, "step": 14786 }, { "epoch": 0.01, "learning_rate": 4.997338749566165e-05, "loss": 2.3054, "step": 14787 }, { "epoch": 0.01, "learning_rate": 4.997338387222773e-05, "loss": 2.3027, "step": 14788 }, { "epoch": 0.01, "learning_rate": 4.9973380248547273e-05, "loss": 2.2947, "step": 14789 }, { "epoch": 0.01, "learning_rate": 4.99733766246203e-05, "loss": 2.2729, "step": 14790 }, { "epoch": 0.01, "learning_rate": 4.99733730004468e-05, "loss": 2.2439, "step": 14791 }, { "epoch": 0.01, "learning_rate": 4.9973369376026765e-05, "loss": 2.2121, "step": 14792 }, { "epoch": 0.01, "learning_rate": 4.997336575136021e-05, "loss": 1.6621, "step": 14793 }, { "epoch": 0.01, "learning_rate": 4.997336212644713e-05, "loss": 0.1036, "step": 14794 }, { "epoch": 0.01, "learning_rate": 4.997335850128751e-05, "loss": 0.0036, "step": 14795 }, { "epoch": 0.01, "learning_rate": 4.9973354875881384e-05, "loss": 0.4035, "step": 14796 }, { "epoch": 0.01, "learning_rate": 4.9973351250228716e-05, "loss": 1.617, "step": 14797 }, { "epoch": 0.01, "learning_rate": 4.9973347624329534e-05, "loss": 1.0411, "step": 14798 }, { "epoch": 0.01, "learning_rate": 4.997334399818382e-05, "loss": 1.1408, "step": 14799 }, { "epoch": 0.01, "learning_rate": 4.9973340371791574e-05, "loss": 1.0667, "step": 14800 }, { "epoch": 0.01, "learning_rate": 4.997333674515281e-05, "loss": 1.0492, "step": 14801 }, { "epoch": 0.01, "learning_rate": 4.997333311826751e-05, "loss": 1.1013, "step": 14802 }, { "epoch": 0.01, "learning_rate": 4.997332949113569e-05, "loss": 1.6289, "step": 14803 }, { "epoch": 0.01, "learning_rate": 4.997332586375735e-05, "loss": 1.1567, "step": 14804 }, { "epoch": 0.01, "learning_rate": 4.9973322236132475e-05, "loss": 0.9798, "step": 14805 }, { "epoch": 0.01, "learning_rate": 4.9973318608261075e-05, "loss": 1.0451, "step": 14806 }, { "epoch": 0.01, "learning_rate": 4.997331498014315e-05, "loss": 1.3002, "step": 14807 }, { "epoch": 0.01, "learning_rate": 4.9973311351778693e-05, "loss": 0.9813, "step": 14808 }, { "epoch": 0.01, "learning_rate": 4.9973307723167725e-05, "loss": 0.9681, "step": 14809 }, { "epoch": 0.01, "learning_rate": 4.9973304094310216e-05, "loss": 0.894, "step": 14810 }, { "epoch": 0.01, "learning_rate": 4.997330046520619e-05, "loss": 0.7351, "step": 14811 }, { "epoch": 0.01, "learning_rate": 4.9973296835855636e-05, "loss": 0.6698, "step": 14812 }, { "epoch": 0.01, "learning_rate": 4.997329320625855e-05, "loss": 0.6868, "step": 14813 }, { "epoch": 0.01, "learning_rate": 4.997328957641495e-05, "loss": 1.1907, "step": 14814 }, { "epoch": 0.01, "learning_rate": 4.9973285946324816e-05, "loss": 1.1407, "step": 14815 }, { "epoch": 0.01, "learning_rate": 4.997328231598816e-05, "loss": 1.2337, "step": 14816 }, { "epoch": 0.01, "learning_rate": 4.997327868540498e-05, "loss": 1.1565, "step": 14817 }, { "epoch": 0.01, "learning_rate": 4.9973275054575264e-05, "loss": 1.3925, "step": 14818 }, { "epoch": 0.01, "learning_rate": 4.997327142349904e-05, "loss": 1.2359, "step": 14819 }, { "epoch": 0.01, "learning_rate": 4.997326779217627e-05, "loss": 1.3999, "step": 14820 }, { "epoch": 0.01, "learning_rate": 4.997326416060699e-05, "loss": 0.9791, "step": 14821 }, { "epoch": 0.01, "learning_rate": 4.997326052879118e-05, "loss": 0.7765, "step": 14822 }, { "epoch": 0.01, "learning_rate": 4.9973256896728836e-05, "loss": 1.2074, "step": 14823 }, { "epoch": 0.01, "learning_rate": 4.997325326441997e-05, "loss": 1.1995, "step": 14824 }, { "epoch": 0.01, "learning_rate": 4.997324963186458e-05, "loss": 0.9219, "step": 14825 }, { "epoch": 0.01, "learning_rate": 4.997324599906267e-05, "loss": 1.2767, "step": 14826 }, { "epoch": 0.01, "learning_rate": 4.997324236601423e-05, "loss": 1.108, "step": 14827 }, { "epoch": 0.01, "learning_rate": 4.9973238732719266e-05, "loss": 0.6057, "step": 14828 }, { "epoch": 0.01, "learning_rate": 4.997323509917778e-05, "loss": 0.6592, "step": 14829 }, { "epoch": 0.01, "learning_rate": 4.997323146538977e-05, "loss": 1.0524, "step": 14830 }, { "epoch": 0.01, "learning_rate": 4.9973227831355226e-05, "loss": 1.1696, "step": 14831 }, { "epoch": 0.01, "learning_rate": 4.997322419707416e-05, "loss": 0.9979, "step": 14832 }, { "epoch": 0.01, "learning_rate": 4.997322056254657e-05, "loss": 1.1966, "step": 14833 }, { "epoch": 0.01, "learning_rate": 4.9973216927772455e-05, "loss": 1.2316, "step": 14834 }, { "epoch": 0.01, "learning_rate": 4.997321329275181e-05, "loss": 1.3542, "step": 14835 }, { "epoch": 0.01, "learning_rate": 4.997320965748465e-05, "loss": 1.1776, "step": 14836 }, { "epoch": 0.01, "learning_rate": 4.997320602197096e-05, "loss": 1.0939, "step": 14837 }, { "epoch": 0.01, "learning_rate": 4.997320238621074e-05, "loss": 1.1933, "step": 14838 }, { "epoch": 0.01, "learning_rate": 4.9973198750204e-05, "loss": 0.9844, "step": 14839 }, { "epoch": 0.01, "learning_rate": 4.997319511395074e-05, "loss": 1.0345, "step": 14840 }, { "epoch": 0.01, "learning_rate": 4.997319147745095e-05, "loss": 0.9952, "step": 14841 }, { "epoch": 0.01, "learning_rate": 4.997318784070463e-05, "loss": 1.0174, "step": 14842 }, { "epoch": 0.01, "learning_rate": 4.997318420371179e-05, "loss": 0.8177, "step": 14843 }, { "epoch": 0.01, "learning_rate": 4.997318056647243e-05, "loss": 1.0, "step": 14844 }, { "epoch": 0.01, "learning_rate": 4.997317692898654e-05, "loss": 1.2882, "step": 14845 }, { "epoch": 0.01, "learning_rate": 4.9973173291254126e-05, "loss": 1.1994, "step": 14846 }, { "epoch": 0.01, "learning_rate": 4.997316965327519e-05, "loss": 0.9474, "step": 14847 }, { "epoch": 0.01, "learning_rate": 4.997316601504972e-05, "loss": 1.3273, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.997316237657774e-05, "loss": 0.9667, "step": 14849 }, { "epoch": 0.01, "learning_rate": 4.9973158737859226e-05, "loss": 0.8481, "step": 14850 }, { "epoch": 0.01, "learning_rate": 4.997315509889419e-05, "loss": 1.2891, "step": 14851 }, { "epoch": 0.01, "learning_rate": 4.997315145968263e-05, "loss": 1.218, "step": 14852 }, { "epoch": 0.01, "learning_rate": 4.9973147820224544e-05, "loss": 1.0303, "step": 14853 }, { "epoch": 0.01, "learning_rate": 4.9973144180519924e-05, "loss": 0.9202, "step": 14854 }, { "epoch": 0.01, "learning_rate": 4.99731405405688e-05, "loss": 1.045, "step": 14855 }, { "epoch": 0.01, "learning_rate": 4.9973136900371137e-05, "loss": 1.4301, "step": 14856 }, { "epoch": 0.01, "learning_rate": 4.997313325992695e-05, "loss": 0.9022, "step": 14857 }, { "epoch": 0.01, "learning_rate": 4.997312961923625e-05, "loss": 1.0168, "step": 14858 }, { "epoch": 0.01, "learning_rate": 4.997312597829902e-05, "loss": 1.0202, "step": 14859 }, { "epoch": 0.01, "learning_rate": 4.997312233711526e-05, "loss": 1.1546, "step": 14860 }, { "epoch": 0.01, "learning_rate": 4.9973118695684985e-05, "loss": 1.2422, "step": 14861 }, { "epoch": 0.01, "learning_rate": 4.997311505400818e-05, "loss": 0.9089, "step": 14862 }, { "epoch": 0.01, "learning_rate": 4.997311141208485e-05, "loss": 1.1787, "step": 14863 }, { "epoch": 0.01, "learning_rate": 4.997310776991499e-05, "loss": 1.0265, "step": 14864 }, { "epoch": 0.01, "learning_rate": 4.997310412749862e-05, "loss": 1.1003, "step": 14865 }, { "epoch": 0.01, "learning_rate": 4.997310048483572e-05, "loss": 1.1564, "step": 14866 }, { "epoch": 0.01, "learning_rate": 4.9973096841926294e-05, "loss": 1.2079, "step": 14867 }, { "epoch": 0.01, "learning_rate": 4.997309319877035e-05, "loss": 1.1985, "step": 14868 }, { "epoch": 0.01, "learning_rate": 4.997308955536788e-05, "loss": 0.9917, "step": 14869 }, { "epoch": 0.01, "learning_rate": 4.997308591171888e-05, "loss": 1.0653, "step": 14870 }, { "epoch": 0.01, "learning_rate": 4.997308226782336e-05, "loss": 1.0926, "step": 14871 }, { "epoch": 0.01, "learning_rate": 4.997307862368132e-05, "loss": 0.8831, "step": 14872 }, { "epoch": 0.01, "learning_rate": 4.997307497929276e-05, "loss": 1.342, "step": 14873 }, { "epoch": 0.01, "learning_rate": 4.997307133465766e-05, "loss": 0.8392, "step": 14874 }, { "epoch": 0.01, "learning_rate": 4.997306768977605e-05, "loss": 1.1513, "step": 14875 }, { "epoch": 0.01, "learning_rate": 4.997306404464791e-05, "loss": 1.1801, "step": 14876 }, { "epoch": 0.01, "learning_rate": 4.9973060399273254e-05, "loss": 1.3827, "step": 14877 }, { "epoch": 0.01, "learning_rate": 4.997305675365206e-05, "loss": 1.0725, "step": 14878 }, { "epoch": 0.01, "learning_rate": 4.9973053107784355e-05, "loss": 0.8855, "step": 14879 }, { "epoch": 0.01, "learning_rate": 4.997304946167013e-05, "loss": 1.2834, "step": 14880 }, { "epoch": 0.01, "learning_rate": 4.997304581530937e-05, "loss": 1.0089, "step": 14881 }, { "epoch": 0.01, "learning_rate": 4.997304216870209e-05, "loss": 0.8269, "step": 14882 }, { "epoch": 0.01, "learning_rate": 4.997303852184829e-05, "loss": 1.0531, "step": 14883 }, { "epoch": 0.01, "learning_rate": 4.9973034874747964e-05, "loss": 1.0446, "step": 14884 }, { "epoch": 0.01, "learning_rate": 4.9973031227401115e-05, "loss": 2.3159, "step": 14885 }, { "epoch": 0.01, "learning_rate": 4.9973027579807745e-05, "loss": 1.4127, "step": 14886 }, { "epoch": 0.01, "learning_rate": 4.9973023931967855e-05, "loss": 1.1488, "step": 14887 }, { "epoch": 0.01, "learning_rate": 4.997302028388143e-05, "loss": 1.1259, "step": 14888 }, { "epoch": 0.01, "learning_rate": 4.997301663554849e-05, "loss": 1.1292, "step": 14889 }, { "epoch": 0.01, "learning_rate": 4.997301298696903e-05, "loss": 1.0684, "step": 14890 }, { "epoch": 0.01, "learning_rate": 4.997300933814304e-05, "loss": 0.9875, "step": 14891 }, { "epoch": 0.01, "learning_rate": 4.997300568907053e-05, "loss": 1.0519, "step": 14892 }, { "epoch": 0.01, "learning_rate": 4.9973002039751496e-05, "loss": 1.092, "step": 14893 }, { "epoch": 0.01, "learning_rate": 4.997299839018594e-05, "loss": 1.2802, "step": 14894 }, { "epoch": 0.01, "learning_rate": 4.997299474037386e-05, "loss": 1.1129, "step": 14895 }, { "epoch": 0.01, "learning_rate": 4.997299109031526e-05, "loss": 1.0056, "step": 14896 }, { "epoch": 0.01, "learning_rate": 4.997298744001013e-05, "loss": 1.1671, "step": 14897 }, { "epoch": 0.01, "learning_rate": 4.997298378945849e-05, "loss": 0.9901, "step": 14898 }, { "epoch": 0.01, "learning_rate": 4.997298013866032e-05, "loss": 0.94, "step": 14899 }, { "epoch": 0.01, "learning_rate": 4.997297648761562e-05, "loss": 0.9399, "step": 14900 }, { "epoch": 0.01, "learning_rate": 4.9972972836324405e-05, "loss": 0.7191, "step": 14901 }, { "epoch": 0.01, "learning_rate": 4.997296918478667e-05, "loss": 0.909, "step": 14902 }, { "epoch": 0.01, "learning_rate": 4.997296553300241e-05, "loss": 1.1014, "step": 14903 }, { "epoch": 0.01, "learning_rate": 4.9972961880971624e-05, "loss": 1.0286, "step": 14904 }, { "epoch": 0.01, "learning_rate": 4.9972958228694324e-05, "loss": 1.1349, "step": 14905 }, { "epoch": 0.01, "learning_rate": 4.997295457617049e-05, "loss": 0.6595, "step": 14906 }, { "epoch": 0.01, "learning_rate": 4.9972950923400144e-05, "loss": 0.7242, "step": 14907 }, { "epoch": 0.01, "learning_rate": 4.997294727038326e-05, "loss": 0.7094, "step": 14908 }, { "epoch": 0.01, "learning_rate": 4.997294361711987e-05, "loss": 0.6031, "step": 14909 }, { "epoch": 0.01, "learning_rate": 4.997293996360995e-05, "loss": 0.452, "step": 14910 }, { "epoch": 0.01, "learning_rate": 4.9972936309853504e-05, "loss": 0.3931, "step": 14911 }, { "epoch": 0.01, "learning_rate": 4.997293265585055e-05, "loss": 0.7613, "step": 14912 }, { "epoch": 0.01, "learning_rate": 4.997292900160106e-05, "loss": 0.9411, "step": 14913 }, { "epoch": 0.01, "learning_rate": 4.997292534710505e-05, "loss": 0.8584, "step": 14914 }, { "epoch": 0.01, "learning_rate": 4.997292169236252e-05, "loss": 0.9696, "step": 14915 }, { "epoch": 0.01, "learning_rate": 4.997291803737347e-05, "loss": 0.8084, "step": 14916 }, { "epoch": 0.01, "learning_rate": 4.99729143821379e-05, "loss": 0.9684, "step": 14917 }, { "epoch": 0.01, "learning_rate": 4.9972910726655794e-05, "loss": 1.0021, "step": 14918 }, { "epoch": 0.01, "learning_rate": 4.997290707092718e-05, "loss": 1.1548, "step": 14919 }, { "epoch": 0.01, "learning_rate": 4.997290341495204e-05, "loss": 1.0268, "step": 14920 }, { "epoch": 0.01, "learning_rate": 4.997289975873038e-05, "loss": 0.952, "step": 14921 }, { "epoch": 0.01, "learning_rate": 4.997289610226219e-05, "loss": 0.8096, "step": 14922 }, { "epoch": 0.01, "learning_rate": 4.997289244554749e-05, "loss": 0.567, "step": 14923 }, { "epoch": 0.01, "learning_rate": 4.9972888788586255e-05, "loss": 1.1943, "step": 14924 }, { "epoch": 0.01, "learning_rate": 4.997288513137851e-05, "loss": 1.4243, "step": 14925 }, { "epoch": 0.01, "learning_rate": 4.9972881473924236e-05, "loss": 0.9679, "step": 14926 }, { "epoch": 0.01, "learning_rate": 4.9972877816223446e-05, "loss": 1.1361, "step": 14927 }, { "epoch": 0.01, "learning_rate": 4.997287415827613e-05, "loss": 1.0682, "step": 14928 }, { "epoch": 0.01, "learning_rate": 4.997287050008229e-05, "loss": 1.1659, "step": 14929 }, { "epoch": 0.01, "learning_rate": 4.997286684164193e-05, "loss": 1.0461, "step": 14930 }, { "epoch": 0.01, "learning_rate": 4.9972863182955054e-05, "loss": 1.2201, "step": 14931 }, { "epoch": 0.01, "learning_rate": 4.9972859524021654e-05, "loss": 1.1834, "step": 14932 }, { "epoch": 0.01, "learning_rate": 4.997285586484173e-05, "loss": 1.1125, "step": 14933 }, { "epoch": 0.01, "learning_rate": 4.997285220541529e-05, "loss": 1.0901, "step": 14934 }, { "epoch": 0.01, "learning_rate": 4.997284854574233e-05, "loss": 1.0592, "step": 14935 }, { "epoch": 0.01, "learning_rate": 4.997284488582283e-05, "loss": 1.0942, "step": 14936 }, { "epoch": 0.01, "learning_rate": 4.997284122565682e-05, "loss": 0.9409, "step": 14937 }, { "epoch": 0.01, "learning_rate": 4.997283756524429e-05, "loss": 1.2872, "step": 14938 }, { "epoch": 0.01, "learning_rate": 4.997283390458524e-05, "loss": 1.0326, "step": 14939 }, { "epoch": 0.01, "learning_rate": 4.997283024367967e-05, "loss": 1.243, "step": 14940 }, { "epoch": 0.01, "learning_rate": 4.997282658252758e-05, "loss": 1.1183, "step": 14941 }, { "epoch": 0.01, "learning_rate": 4.997282292112896e-05, "loss": 1.112, "step": 14942 }, { "epoch": 0.01, "learning_rate": 4.997281925948382e-05, "loss": 1.2697, "step": 14943 }, { "epoch": 0.01, "learning_rate": 4.997281559759216e-05, "loss": 1.11, "step": 14944 }, { "epoch": 0.01, "learning_rate": 4.9972811935453985e-05, "loss": 0.5532, "step": 14945 }, { "epoch": 0.01, "learning_rate": 4.9972808273069286e-05, "loss": 1.1706, "step": 14946 }, { "epoch": 0.01, "learning_rate": 4.997280461043806e-05, "loss": 1.176, "step": 14947 }, { "epoch": 0.01, "learning_rate": 4.997280094756032e-05, "loss": 1.0269, "step": 14948 }, { "epoch": 0.01, "learning_rate": 4.997279728443606e-05, "loss": 1.1924, "step": 14949 }, { "epoch": 0.01, "learning_rate": 4.997279362106527e-05, "loss": 1.3168, "step": 14950 }, { "epoch": 0.01, "learning_rate": 4.997278995744797e-05, "loss": 1.3441, "step": 14951 }, { "epoch": 0.01, "learning_rate": 4.997278629358414e-05, "loss": 1.2574, "step": 14952 }, { "epoch": 0.01, "learning_rate": 4.9972782629473796e-05, "loss": 1.0412, "step": 14953 }, { "epoch": 0.01, "learning_rate": 4.9972778965116926e-05, "loss": 1.0197, "step": 14954 }, { "epoch": 0.01, "learning_rate": 4.997277530051354e-05, "loss": 1.0773, "step": 14955 }, { "epoch": 0.01, "learning_rate": 4.997277163566363e-05, "loss": 1.0283, "step": 14956 }, { "epoch": 0.01, "learning_rate": 4.99727679705672e-05, "loss": 1.0716, "step": 14957 }, { "epoch": 0.01, "learning_rate": 4.997276430522425e-05, "loss": 1.0224, "step": 14958 }, { "epoch": 0.01, "learning_rate": 4.997276063963478e-05, "loss": 0.9932, "step": 14959 }, { "epoch": 0.01, "learning_rate": 4.997275697379879e-05, "loss": 0.7764, "step": 14960 }, { "epoch": 0.01, "learning_rate": 4.997275330771628e-05, "loss": 1.1036, "step": 14961 }, { "epoch": 0.01, "learning_rate": 4.997274964138724e-05, "loss": 0.9545, "step": 14962 }, { "epoch": 0.01, "learning_rate": 4.9972745974811696e-05, "loss": 1.0358, "step": 14963 }, { "epoch": 0.01, "learning_rate": 4.9972742307989614e-05, "loss": 1.1154, "step": 14964 }, { "epoch": 0.01, "learning_rate": 4.9972738640921026e-05, "loss": 1.2059, "step": 14965 }, { "epoch": 0.01, "learning_rate": 4.997273497360591e-05, "loss": 1.0128, "step": 14966 }, { "epoch": 0.01, "learning_rate": 4.997273130604428e-05, "loss": 0.8856, "step": 14967 }, { "epoch": 0.01, "learning_rate": 4.9972727638236125e-05, "loss": 0.4957, "step": 14968 }, { "epoch": 0.01, "learning_rate": 4.997272397018145e-05, "loss": 0.2882, "step": 14969 }, { "epoch": 0.01, "learning_rate": 4.997272030188026e-05, "loss": 0.2383, "step": 14970 }, { "epoch": 0.01, "learning_rate": 4.997271663333254e-05, "loss": 0.5858, "step": 14971 }, { "epoch": 0.01, "learning_rate": 4.997271296453831e-05, "loss": 1.2508, "step": 14972 }, { "epoch": 0.01, "learning_rate": 4.997270929549754e-05, "loss": 1.0206, "step": 14973 }, { "epoch": 0.01, "learning_rate": 4.997270562621027e-05, "loss": 1.0349, "step": 14974 }, { "epoch": 0.01, "learning_rate": 4.9972701956676476e-05, "loss": 0.2786, "step": 14975 }, { "epoch": 0.01, "learning_rate": 4.9972698286896156e-05, "loss": 0.2352, "step": 14976 }, { "epoch": 0.01, "learning_rate": 4.997269461686933e-05, "loss": 0.1715, "step": 14977 }, { "epoch": 0.01, "learning_rate": 4.997269094659597e-05, "loss": 0.1294, "step": 14978 }, { "epoch": 0.01, "learning_rate": 4.997268727607609e-05, "loss": 0.1131, "step": 14979 }, { "epoch": 0.01, "learning_rate": 4.99726836053097e-05, "loss": 0.1052, "step": 14980 }, { "epoch": 0.01, "learning_rate": 4.997267993429678e-05, "loss": 0.0949, "step": 14981 }, { "epoch": 0.01, "learning_rate": 4.997267626303735e-05, "loss": 0.5213, "step": 14982 }, { "epoch": 0.01, "learning_rate": 4.99726725915314e-05, "loss": 1.1957, "step": 14983 }, { "epoch": 0.01, "learning_rate": 4.997266891977892e-05, "loss": 0.9684, "step": 14984 }, { "epoch": 0.01, "learning_rate": 4.997266524777993e-05, "loss": 1.0212, "step": 14985 }, { "epoch": 0.01, "learning_rate": 4.9972661575534415e-05, "loss": 1.2718, "step": 14986 }, { "epoch": 0.01, "learning_rate": 4.9972657903042383e-05, "loss": 1.1513, "step": 14987 }, { "epoch": 0.01, "learning_rate": 4.997265423030383e-05, "loss": 1.2228, "step": 14988 }, { "epoch": 0.01, "learning_rate": 4.997265055731876e-05, "loss": 1.1608, "step": 14989 }, { "epoch": 0.01, "learning_rate": 4.9972646884087165e-05, "loss": 0.9096, "step": 14990 }, { "epoch": 0.01, "learning_rate": 4.997264321060906e-05, "loss": 1.2469, "step": 14991 }, { "epoch": 0.01, "learning_rate": 4.9972639536884424e-05, "loss": 1.3127, "step": 14992 }, { "epoch": 0.01, "learning_rate": 4.9972635862913276e-05, "loss": 1.7783, "step": 14993 }, { "epoch": 0.01, "learning_rate": 4.997263218869561e-05, "loss": 0.7732, "step": 14994 }, { "epoch": 0.01, "learning_rate": 4.997262851423142e-05, "loss": 1.0013, "step": 14995 }, { "epoch": 0.01, "learning_rate": 4.997262483952071e-05, "loss": 2.2452, "step": 14996 }, { "epoch": 0.01, "learning_rate": 4.997262116456348e-05, "loss": 6.0102, "step": 14997 }, { "epoch": 0.01, "learning_rate": 4.9972617489359736e-05, "loss": 3.1733, "step": 14998 }, { "epoch": 0.01, "learning_rate": 4.997261381390947e-05, "loss": 1.0003, "step": 14999 }, { "epoch": 0.01, "learning_rate": 4.997261013821268e-05, "loss": 1.2808, "step": 15000 }, { "epoch": 0.01, "eval_loss": 1.0408920049667358, "eval_runtime": 84.1304, "eval_samples_per_second": 16.463, "eval_steps_per_second": 4.125, "step": 15000 }, { "epoch": 0.02, "learning_rate": 4.997260646226938e-05, "loss": 0.9748, "step": 15001 }, { "epoch": 0.02, "learning_rate": 4.997260278607956e-05, "loss": 1.0493, "step": 15002 }, { "epoch": 0.02, "learning_rate": 4.997259910964321e-05, "loss": 1.1049, "step": 15003 }, { "epoch": 0.02, "learning_rate": 4.997259543296036e-05, "loss": 1.2809, "step": 15004 }, { "epoch": 0.02, "learning_rate": 4.997259175603098e-05, "loss": 1.0699, "step": 15005 }, { "epoch": 0.02, "learning_rate": 4.997258807885508e-05, "loss": 1.1148, "step": 15006 }, { "epoch": 0.02, "learning_rate": 4.9972584401432664e-05, "loss": 1.0759, "step": 15007 }, { "epoch": 0.02, "learning_rate": 4.997258072376373e-05, "loss": 1.3592, "step": 15008 }, { "epoch": 0.02, "learning_rate": 4.997257704584827e-05, "loss": 1.1556, "step": 15009 }, { "epoch": 0.02, "learning_rate": 4.99725733676863e-05, "loss": 1.2197, "step": 15010 }, { "epoch": 0.02, "learning_rate": 4.99725696892778e-05, "loss": 1.3116, "step": 15011 }, { "epoch": 0.02, "learning_rate": 4.997256601062279e-05, "loss": 1.1053, "step": 15012 }, { "epoch": 0.02, "learning_rate": 4.997256233172126e-05, "loss": 1.2717, "step": 15013 }, { "epoch": 0.02, "learning_rate": 4.997255865257322e-05, "loss": 1.3174, "step": 15014 }, { "epoch": 0.02, "learning_rate": 4.9972554973178654e-05, "loss": 1.0946, "step": 15015 }, { "epoch": 0.02, "learning_rate": 4.997255129353756e-05, "loss": 0.9896, "step": 15016 }, { "epoch": 0.02, "learning_rate": 4.9972547613649964e-05, "loss": 0.9097, "step": 15017 }, { "epoch": 0.02, "learning_rate": 4.997254393351584e-05, "loss": 0.9884, "step": 15018 }, { "epoch": 0.02, "learning_rate": 4.99725402531352e-05, "loss": 1.2711, "step": 15019 }, { "epoch": 0.02, "learning_rate": 4.997253657250804e-05, "loss": 0.9559, "step": 15020 }, { "epoch": 0.02, "learning_rate": 4.997253289163436e-05, "loss": 1.2035, "step": 15021 }, { "epoch": 0.02, "learning_rate": 4.9972529210514165e-05, "loss": 1.775, "step": 15022 }, { "epoch": 0.02, "learning_rate": 4.997252552914745e-05, "loss": 1.278, "step": 15023 }, { "epoch": 0.02, "learning_rate": 4.9972521847534215e-05, "loss": 1.323, "step": 15024 }, { "epoch": 0.02, "learning_rate": 4.9972518165674466e-05, "loss": 1.3745, "step": 15025 }, { "epoch": 0.02, "learning_rate": 4.99725144835682e-05, "loss": 0.4495, "step": 15026 }, { "epoch": 0.02, "learning_rate": 4.997251080121541e-05, "loss": 1.0237, "step": 15027 }, { "epoch": 0.02, "learning_rate": 4.997250711861611e-05, "loss": 1.0747, "step": 15028 }, { "epoch": 0.02, "learning_rate": 4.997250343577029e-05, "loss": 1.0376, "step": 15029 }, { "epoch": 0.02, "learning_rate": 4.997249975267794e-05, "loss": 1.0421, "step": 15030 }, { "epoch": 0.02, "learning_rate": 4.997249606933908e-05, "loss": 0.9438, "step": 15031 }, { "epoch": 0.02, "learning_rate": 4.9972492385753706e-05, "loss": 1.1047, "step": 15032 }, { "epoch": 0.02, "learning_rate": 4.997248870192181e-05, "loss": 1.4917, "step": 15033 }, { "epoch": 0.02, "learning_rate": 4.9972485017843395e-05, "loss": 0.9421, "step": 15034 }, { "epoch": 0.02, "learning_rate": 4.997248133351847e-05, "loss": 1.058, "step": 15035 }, { "epoch": 0.02, "learning_rate": 4.997247764894702e-05, "loss": 1.0983, "step": 15036 }, { "epoch": 0.02, "learning_rate": 4.9972473964129054e-05, "loss": 1.3499, "step": 15037 }, { "epoch": 0.02, "learning_rate": 4.997247027906457e-05, "loss": 0.8397, "step": 15038 }, { "epoch": 0.02, "learning_rate": 4.9972466593753565e-05, "loss": 1.0555, "step": 15039 }, { "epoch": 0.02, "learning_rate": 4.997246290819605e-05, "loss": 1.1275, "step": 15040 }, { "epoch": 0.02, "learning_rate": 4.997245922239201e-05, "loss": 0.7054, "step": 15041 }, { "epoch": 0.02, "learning_rate": 4.997245553634146e-05, "loss": 0.7308, "step": 15042 }, { "epoch": 0.02, "learning_rate": 4.997245185004438e-05, "loss": 0.7697, "step": 15043 }, { "epoch": 0.02, "learning_rate": 4.99724481635008e-05, "loss": 0.7623, "step": 15044 }, { "epoch": 0.02, "learning_rate": 4.997244447671069e-05, "loss": 0.8263, "step": 15045 }, { "epoch": 0.02, "learning_rate": 4.997244078967406e-05, "loss": 1.3559, "step": 15046 }, { "epoch": 0.02, "learning_rate": 4.9972437102390926e-05, "loss": 1.4373, "step": 15047 }, { "epoch": 0.02, "learning_rate": 4.997243341486127e-05, "loss": 1.274, "step": 15048 }, { "epoch": 0.02, "learning_rate": 4.997242972708508e-05, "loss": 1.3229, "step": 15049 }, { "epoch": 0.02, "learning_rate": 4.9972426039062395e-05, "loss": 1.2814, "step": 15050 }, { "epoch": 0.02, "learning_rate": 4.997242235079318e-05, "loss": 1.1339, "step": 15051 }, { "epoch": 0.02, "learning_rate": 4.997241866227745e-05, "loss": 0.8458, "step": 15052 }, { "epoch": 0.02, "learning_rate": 4.99724149735152e-05, "loss": 1.1498, "step": 15053 }, { "epoch": 0.02, "learning_rate": 4.997241128450644e-05, "loss": 1.2158, "step": 15054 }, { "epoch": 0.02, "learning_rate": 4.997240759525116e-05, "loss": 0.8556, "step": 15055 }, { "epoch": 0.02, "learning_rate": 4.9972403905749364e-05, "loss": 0.9101, "step": 15056 }, { "epoch": 0.02, "learning_rate": 4.997240021600105e-05, "loss": 1.0137, "step": 15057 }, { "epoch": 0.02, "learning_rate": 4.997239652600622e-05, "loss": 1.0061, "step": 15058 }, { "epoch": 0.02, "learning_rate": 4.997239283576487e-05, "loss": 1.145, "step": 15059 }, { "epoch": 0.02, "learning_rate": 4.9972389145277005e-05, "loss": 1.3146, "step": 15060 }, { "epoch": 0.02, "learning_rate": 4.997238545454263e-05, "loss": 1.2995, "step": 15061 }, { "epoch": 0.02, "learning_rate": 4.9972381763561724e-05, "loss": 1.0795, "step": 15062 }, { "epoch": 0.02, "learning_rate": 4.9972378072334306e-05, "loss": 1.1407, "step": 15063 }, { "epoch": 0.02, "learning_rate": 4.9972374380860374e-05, "loss": 1.2046, "step": 15064 }, { "epoch": 0.02, "learning_rate": 4.997237068913992e-05, "loss": 0.9195, "step": 15065 }, { "epoch": 0.02, "learning_rate": 4.997236699717296e-05, "loss": 0.7231, "step": 15066 }, { "epoch": 0.02, "learning_rate": 4.997236330495948e-05, "loss": 0.8774, "step": 15067 }, { "epoch": 0.02, "learning_rate": 4.9972359612499476e-05, "loss": 1.0792, "step": 15068 }, { "epoch": 0.02, "learning_rate": 4.997235591979296e-05, "loss": 0.9843, "step": 15069 }, { "epoch": 0.02, "learning_rate": 4.997235222683993e-05, "loss": 1.1931, "step": 15070 }, { "epoch": 0.02, "learning_rate": 4.997234853364038e-05, "loss": 1.1407, "step": 15071 }, { "epoch": 0.02, "learning_rate": 4.997234484019431e-05, "loss": 1.1714, "step": 15072 }, { "epoch": 0.02, "learning_rate": 4.9972341146501736e-05, "loss": 1.1567, "step": 15073 }, { "epoch": 0.02, "learning_rate": 4.997233745256263e-05, "loss": 1.1103, "step": 15074 }, { "epoch": 0.02, "learning_rate": 4.9972333758377024e-05, "loss": 1.4842, "step": 15075 }, { "epoch": 0.02, "learning_rate": 4.997233006394488e-05, "loss": 0.9396, "step": 15076 }, { "epoch": 0.02, "learning_rate": 4.997232636926624e-05, "loss": 1.2873, "step": 15077 }, { "epoch": 0.02, "learning_rate": 4.997232267434108e-05, "loss": 1.1418, "step": 15078 }, { "epoch": 0.02, "learning_rate": 4.9972318979169394e-05, "loss": 0.9179, "step": 15079 }, { "epoch": 0.02, "learning_rate": 4.99723152837512e-05, "loss": 1.2488, "step": 15080 }, { "epoch": 0.02, "learning_rate": 4.9972311588086484e-05, "loss": 1.002, "step": 15081 }, { "epoch": 0.02, "learning_rate": 4.997230789217526e-05, "loss": 1.2295, "step": 15082 }, { "epoch": 0.02, "learning_rate": 4.997230419601752e-05, "loss": 0.7993, "step": 15083 }, { "epoch": 0.02, "learning_rate": 4.997230049961326e-05, "loss": 0.6249, "step": 15084 }, { "epoch": 0.02, "learning_rate": 4.997229680296248e-05, "loss": 0.5616, "step": 15085 }, { "epoch": 0.02, "learning_rate": 4.9972293106065185e-05, "loss": 1.0836, "step": 15086 }, { "epoch": 0.02, "learning_rate": 4.997228940892138e-05, "loss": 1.234, "step": 15087 }, { "epoch": 0.02, "learning_rate": 4.9972285711531055e-05, "loss": 1.1247, "step": 15088 }, { "epoch": 0.02, "learning_rate": 4.9972282013894214e-05, "loss": 1.014, "step": 15089 }, { "epoch": 0.02, "learning_rate": 4.997227831601086e-05, "loss": 1.0118, "step": 15090 }, { "epoch": 0.02, "learning_rate": 4.997227461788098e-05, "loss": 1.0921, "step": 15091 }, { "epoch": 0.02, "learning_rate": 4.997227091950459e-05, "loss": 1.1307, "step": 15092 }, { "epoch": 0.02, "learning_rate": 4.997226722088169e-05, "loss": 1.0173, "step": 15093 }, { "epoch": 0.02, "learning_rate": 4.997226352201227e-05, "loss": 0.8703, "step": 15094 }, { "epoch": 0.02, "learning_rate": 4.9972259822896334e-05, "loss": 1.2485, "step": 15095 }, { "epoch": 0.02, "learning_rate": 4.997225612353388e-05, "loss": 0.9719, "step": 15096 }, { "epoch": 0.02, "learning_rate": 4.997225242392492e-05, "loss": 1.0862, "step": 15097 }, { "epoch": 0.02, "learning_rate": 4.997224872406944e-05, "loss": 1.1109, "step": 15098 }, { "epoch": 0.02, "learning_rate": 4.997224502396743e-05, "loss": 1.2727, "step": 15099 }, { "epoch": 0.02, "learning_rate": 4.997224132361893e-05, "loss": 0.9653, "step": 15100 }, { "epoch": 0.02, "learning_rate": 4.99722376230239e-05, "loss": 0.8132, "step": 15101 }, { "epoch": 0.02, "learning_rate": 4.9972233922182353e-05, "loss": 1.1867, "step": 15102 }, { "epoch": 0.02, "learning_rate": 4.99722302210943e-05, "loss": 0.5581, "step": 15103 }, { "epoch": 0.02, "learning_rate": 4.9972226519759725e-05, "loss": 1.5255, "step": 15104 }, { "epoch": 0.02, "learning_rate": 4.9972222818178623e-05, "loss": 1.0357, "step": 15105 }, { "epoch": 0.02, "learning_rate": 4.997221911635103e-05, "loss": 1.3428, "step": 15106 }, { "epoch": 0.02, "learning_rate": 4.99722154142769e-05, "loss": 1.1329, "step": 15107 }, { "epoch": 0.02, "learning_rate": 4.997221171195628e-05, "loss": 0.9716, "step": 15108 }, { "epoch": 0.02, "learning_rate": 4.997220800938912e-05, "loss": 0.9562, "step": 15109 }, { "epoch": 0.02, "learning_rate": 4.997220430657546e-05, "loss": 1.0278, "step": 15110 }, { "epoch": 0.02, "learning_rate": 4.9972200603515273e-05, "loss": 1.1521, "step": 15111 }, { "epoch": 0.02, "learning_rate": 4.9972196900208576e-05, "loss": 0.7229, "step": 15112 }, { "epoch": 0.02, "learning_rate": 4.997219319665537e-05, "loss": 1.2424, "step": 15113 }, { "epoch": 0.02, "learning_rate": 4.997218949285564e-05, "loss": 0.7695, "step": 15114 }, { "epoch": 0.02, "learning_rate": 4.99721857888094e-05, "loss": 0.7015, "step": 15115 }, { "epoch": 0.02, "learning_rate": 4.997218208451664e-05, "loss": 0.7679, "step": 15116 }, { "epoch": 0.02, "learning_rate": 4.9972178379977376e-05, "loss": 1.2424, "step": 15117 }, { "epoch": 0.02, "learning_rate": 4.997217467519158e-05, "loss": 1.1658, "step": 15118 }, { "epoch": 0.02, "learning_rate": 4.99721709701593e-05, "loss": 1.114, "step": 15119 }, { "epoch": 0.02, "learning_rate": 4.9972167264880476e-05, "loss": 1.1159, "step": 15120 }, { "epoch": 0.02, "learning_rate": 4.997216355935514e-05, "loss": 1.2718, "step": 15121 }, { "epoch": 0.02, "learning_rate": 4.99721598535833e-05, "loss": 1.0424, "step": 15122 }, { "epoch": 0.02, "learning_rate": 4.997215614756494e-05, "loss": 1.3091, "step": 15123 }, { "epoch": 0.02, "learning_rate": 4.997215244130007e-05, "loss": 0.922, "step": 15124 }, { "epoch": 0.02, "learning_rate": 4.997214873478868e-05, "loss": 1.0213, "step": 15125 }, { "epoch": 0.02, "learning_rate": 4.997214502803078e-05, "loss": 1.1075, "step": 15126 }, { "epoch": 0.02, "learning_rate": 4.997214132102636e-05, "loss": 0.9798, "step": 15127 }, { "epoch": 0.02, "learning_rate": 4.9972137613775425e-05, "loss": 1.6559, "step": 15128 }, { "epoch": 0.02, "learning_rate": 4.997213390627798e-05, "loss": 1.3745, "step": 15129 }, { "epoch": 0.02, "learning_rate": 4.9972130198534024e-05, "loss": 1.0863, "step": 15130 }, { "epoch": 0.02, "learning_rate": 4.9972126490543546e-05, "loss": 1.1254, "step": 15131 }, { "epoch": 0.02, "learning_rate": 4.997212278230656e-05, "loss": 1.329, "step": 15132 }, { "epoch": 0.02, "learning_rate": 4.9972119073823056e-05, "loss": 1.3043, "step": 15133 }, { "epoch": 0.02, "learning_rate": 4.9972115365093043e-05, "loss": 1.277, "step": 15134 }, { "epoch": 0.02, "learning_rate": 4.9972111656116504e-05, "loss": 1.3675, "step": 15135 }, { "epoch": 0.02, "learning_rate": 4.9972107946893464e-05, "loss": 1.0155, "step": 15136 }, { "epoch": 0.02, "learning_rate": 4.99721042374239e-05, "loss": 1.3606, "step": 15137 }, { "epoch": 0.02, "learning_rate": 4.997210052770783e-05, "loss": 1.0277, "step": 15138 }, { "epoch": 0.02, "learning_rate": 4.997209681774524e-05, "loss": 1.312, "step": 15139 }, { "epoch": 0.02, "learning_rate": 4.997209310753614e-05, "loss": 0.7359, "step": 15140 }, { "epoch": 0.02, "learning_rate": 4.997208939708053e-05, "loss": 0.6109, "step": 15141 }, { "epoch": 0.02, "learning_rate": 4.997208568637839e-05, "loss": 1.6273, "step": 15142 }, { "epoch": 0.02, "learning_rate": 4.997208197542975e-05, "loss": 1.1852, "step": 15143 }, { "epoch": 0.02, "learning_rate": 4.997207826423459e-05, "loss": 1.0924, "step": 15144 }, { "epoch": 0.02, "learning_rate": 4.9972074552792917e-05, "loss": 1.1806, "step": 15145 }, { "epoch": 0.02, "learning_rate": 4.997207084110473e-05, "loss": 1.0955, "step": 15146 }, { "epoch": 0.02, "learning_rate": 4.9972067129170036e-05, "loss": 0.8981, "step": 15147 }, { "epoch": 0.02, "learning_rate": 4.9972063416988825e-05, "loss": 0.8413, "step": 15148 }, { "epoch": 0.02, "learning_rate": 4.9972059704561094e-05, "loss": 0.9495, "step": 15149 }, { "epoch": 0.02, "learning_rate": 4.9972055991886856e-05, "loss": 1.217, "step": 15150 }, { "epoch": 0.02, "learning_rate": 4.99720522789661e-05, "loss": 1.4445, "step": 15151 }, { "epoch": 0.02, "learning_rate": 4.997204856579883e-05, "loss": 1.2377, "step": 15152 }, { "epoch": 0.02, "learning_rate": 4.997204485238505e-05, "loss": 1.1629, "step": 15153 }, { "epoch": 0.02, "learning_rate": 4.997204113872476e-05, "loss": 1.1063, "step": 15154 }, { "epoch": 0.02, "learning_rate": 4.997203742481795e-05, "loss": 1.2614, "step": 15155 }, { "epoch": 0.02, "learning_rate": 4.9972033710664626e-05, "loss": 1.0013, "step": 15156 }, { "epoch": 0.02, "learning_rate": 4.997202999626479e-05, "loss": 1.3374, "step": 15157 }, { "epoch": 0.02, "learning_rate": 4.9972026281618444e-05, "loss": 1.1069, "step": 15158 }, { "epoch": 0.02, "learning_rate": 4.997202256672558e-05, "loss": 1.1944, "step": 15159 }, { "epoch": 0.02, "learning_rate": 4.997201885158621e-05, "loss": 0.973, "step": 15160 }, { "epoch": 0.02, "learning_rate": 4.9972015136200326e-05, "loss": 0.4015, "step": 15161 }, { "epoch": 0.02, "learning_rate": 4.9972011420567924e-05, "loss": 1.2352, "step": 15162 }, { "epoch": 0.02, "learning_rate": 4.997200770468901e-05, "loss": 1.2335, "step": 15163 }, { "epoch": 0.02, "learning_rate": 4.997200398856358e-05, "loss": 1.2301, "step": 15164 }, { "epoch": 0.02, "learning_rate": 4.997200027219164e-05, "loss": 1.1323, "step": 15165 }, { "epoch": 0.02, "learning_rate": 4.9971996555573184e-05, "loss": 0.8692, "step": 15166 }, { "epoch": 0.02, "learning_rate": 4.997199283870821e-05, "loss": 1.0918, "step": 15167 }, { "epoch": 0.02, "learning_rate": 4.997198912159674e-05, "loss": 1.2836, "step": 15168 }, { "epoch": 0.02, "learning_rate": 4.997198540423875e-05, "loss": 1.1655, "step": 15169 }, { "epoch": 0.02, "learning_rate": 4.997198168663424e-05, "loss": 1.237, "step": 15170 }, { "epoch": 0.02, "learning_rate": 4.9971977968783226e-05, "loss": 1.0943, "step": 15171 }, { "epoch": 0.02, "learning_rate": 4.997197425068569e-05, "loss": 1.1444, "step": 15172 }, { "epoch": 0.02, "learning_rate": 4.997197053234165e-05, "loss": 1.0765, "step": 15173 }, { "epoch": 0.02, "learning_rate": 4.9971966813751094e-05, "loss": 0.8227, "step": 15174 }, { "epoch": 0.02, "learning_rate": 4.997196309491402e-05, "loss": 1.2364, "step": 15175 }, { "epoch": 0.02, "learning_rate": 4.997195937583043e-05, "loss": 1.2285, "step": 15176 }, { "epoch": 0.02, "learning_rate": 4.9971955656500345e-05, "loss": 1.0951, "step": 15177 }, { "epoch": 0.02, "learning_rate": 4.997195193692373e-05, "loss": 1.1524, "step": 15178 }, { "epoch": 0.02, "learning_rate": 4.9971948217100616e-05, "loss": 1.1948, "step": 15179 }, { "epoch": 0.02, "learning_rate": 4.997194449703099e-05, "loss": 1.1285, "step": 15180 }, { "epoch": 0.02, "learning_rate": 4.997194077671484e-05, "loss": 1.0622, "step": 15181 }, { "epoch": 0.02, "learning_rate": 4.997193705615218e-05, "loss": 1.409, "step": 15182 }, { "epoch": 0.02, "learning_rate": 4.997193333534301e-05, "loss": 1.5899, "step": 15183 }, { "epoch": 0.02, "learning_rate": 4.997192961428733e-05, "loss": 1.1493, "step": 15184 }, { "epoch": 0.02, "learning_rate": 4.997192589298513e-05, "loss": 1.1914, "step": 15185 }, { "epoch": 0.02, "learning_rate": 4.997192217143643e-05, "loss": 1.2167, "step": 15186 }, { "epoch": 0.02, "learning_rate": 4.997191844964121e-05, "loss": 1.1121, "step": 15187 }, { "epoch": 0.02, "learning_rate": 4.997191472759948e-05, "loss": 1.2398, "step": 15188 }, { "epoch": 0.02, "learning_rate": 4.9971911005311234e-05, "loss": 1.0817, "step": 15189 }, { "epoch": 0.02, "learning_rate": 4.9971907282776475e-05, "loss": 1.3304, "step": 15190 }, { "epoch": 0.02, "learning_rate": 4.99719035599952e-05, "loss": 0.9831, "step": 15191 }, { "epoch": 0.02, "learning_rate": 4.997189983696743e-05, "loss": 0.8255, "step": 15192 }, { "epoch": 0.02, "learning_rate": 4.997189611369314e-05, "loss": 1.1217, "step": 15193 }, { "epoch": 0.02, "learning_rate": 4.9971892390172324e-05, "loss": 1.2713, "step": 15194 }, { "epoch": 0.02, "learning_rate": 4.997188866640501e-05, "loss": 1.2356, "step": 15195 }, { "epoch": 0.02, "learning_rate": 4.997188494239118e-05, "loss": 1.2145, "step": 15196 }, { "epoch": 0.02, "learning_rate": 4.997188121813084e-05, "loss": 1.2601, "step": 15197 }, { "epoch": 0.02, "learning_rate": 4.9971877493623994e-05, "loss": 0.9486, "step": 15198 }, { "epoch": 0.02, "learning_rate": 4.9971873768870625e-05, "loss": 1.0815, "step": 15199 }, { "epoch": 0.02, "learning_rate": 4.997187004387075e-05, "loss": 1.1651, "step": 15200 }, { "epoch": 0.02, "learning_rate": 4.997186631862436e-05, "loss": 0.9465, "step": 15201 }, { "epoch": 0.02, "learning_rate": 4.9971862593131465e-05, "loss": 0.9616, "step": 15202 }, { "epoch": 0.02, "learning_rate": 4.9971858867392055e-05, "loss": 0.9271, "step": 15203 }, { "epoch": 0.02, "learning_rate": 4.997185514140613e-05, "loss": 1.215, "step": 15204 }, { "epoch": 0.02, "learning_rate": 4.99718514151737e-05, "loss": 1.2199, "step": 15205 }, { "epoch": 0.02, "learning_rate": 4.997184768869475e-05, "loss": 1.0676, "step": 15206 }, { "epoch": 0.02, "learning_rate": 4.9971843961969286e-05, "loss": 0.9031, "step": 15207 }, { "epoch": 0.02, "learning_rate": 4.997184023499732e-05, "loss": 1.0616, "step": 15208 }, { "epoch": 0.02, "learning_rate": 4.9971836507778844e-05, "loss": 1.3447, "step": 15209 }, { "epoch": 0.02, "learning_rate": 4.997183278031385e-05, "loss": 1.0256, "step": 15210 }, { "epoch": 0.02, "learning_rate": 4.9971829052602346e-05, "loss": 0.9974, "step": 15211 }, { "epoch": 0.02, "learning_rate": 4.997182532464433e-05, "loss": 1.3337, "step": 15212 }, { "epoch": 0.02, "learning_rate": 4.997182159643981e-05, "loss": 1.1154, "step": 15213 }, { "epoch": 0.02, "learning_rate": 4.997181786798877e-05, "loss": 1.0998, "step": 15214 }, { "epoch": 0.02, "learning_rate": 4.997181413929122e-05, "loss": 1.0698, "step": 15215 }, { "epoch": 0.02, "learning_rate": 4.997181041034716e-05, "loss": 1.1894, "step": 15216 }, { "epoch": 0.02, "learning_rate": 4.9971806681156586e-05, "loss": 1.0463, "step": 15217 }, { "epoch": 0.02, "learning_rate": 4.9971802951719505e-05, "loss": 0.9366, "step": 15218 }, { "epoch": 0.02, "learning_rate": 4.997179922203591e-05, "loss": 1.1174, "step": 15219 }, { "epoch": 0.02, "learning_rate": 4.997179549210581e-05, "loss": 1.0023, "step": 15220 }, { "epoch": 0.02, "learning_rate": 4.9971791761929186e-05, "loss": 0.7451, "step": 15221 }, { "epoch": 0.02, "learning_rate": 4.997178803150606e-05, "loss": 0.7231, "step": 15222 }, { "epoch": 0.02, "learning_rate": 4.997178430083643e-05, "loss": 1.0292, "step": 15223 }, { "epoch": 0.02, "learning_rate": 4.997178056992028e-05, "loss": 1.0638, "step": 15224 }, { "epoch": 0.02, "learning_rate": 4.997177683875762e-05, "loss": 0.8664, "step": 15225 }, { "epoch": 0.02, "learning_rate": 4.997177310734845e-05, "loss": 0.93, "step": 15226 }, { "epoch": 0.02, "learning_rate": 4.9971769375692766e-05, "loss": 0.9748, "step": 15227 }, { "epoch": 0.02, "learning_rate": 4.997176564379058e-05, "loss": 1.1762, "step": 15228 }, { "epoch": 0.02, "learning_rate": 4.997176191164187e-05, "loss": 1.0592, "step": 15229 }, { "epoch": 0.02, "learning_rate": 4.9971758179246666e-05, "loss": 1.0584, "step": 15230 }, { "epoch": 0.02, "learning_rate": 4.9971754446604934e-05, "loss": 1.2004, "step": 15231 }, { "epoch": 0.02, "learning_rate": 4.99717507137167e-05, "loss": 0.9591, "step": 15232 }, { "epoch": 0.02, "learning_rate": 4.9971746980581956e-05, "loss": 1.1207, "step": 15233 }, { "epoch": 0.02, "learning_rate": 4.9971743247200696e-05, "loss": 1.0051, "step": 15234 }, { "epoch": 0.02, "learning_rate": 4.9971739513572937e-05, "loss": 1.245, "step": 15235 }, { "epoch": 0.02, "learning_rate": 4.9971735779698656e-05, "loss": 1.0049, "step": 15236 }, { "epoch": 0.02, "learning_rate": 4.997173204557787e-05, "loss": 1.1428, "step": 15237 }, { "epoch": 0.02, "learning_rate": 4.9971728311210575e-05, "loss": 0.9968, "step": 15238 }, { "epoch": 0.02, "learning_rate": 4.997172457659677e-05, "loss": 0.438, "step": 15239 }, { "epoch": 0.02, "learning_rate": 4.9971720841736446e-05, "loss": 1.0456, "step": 15240 }, { "epoch": 0.02, "learning_rate": 4.997171710662962e-05, "loss": 1.1608, "step": 15241 }, { "epoch": 0.02, "learning_rate": 4.997171337127628e-05, "loss": 1.0993, "step": 15242 }, { "epoch": 0.02, "learning_rate": 4.9971709635676434e-05, "loss": 1.2221, "step": 15243 }, { "epoch": 0.02, "learning_rate": 4.997170589983007e-05, "loss": 1.1112, "step": 15244 }, { "epoch": 0.02, "learning_rate": 4.99717021637372e-05, "loss": 1.2677, "step": 15245 }, { "epoch": 0.02, "learning_rate": 4.997169842739782e-05, "loss": 0.9942, "step": 15246 }, { "epoch": 0.02, "learning_rate": 4.9971694690811935e-05, "loss": 1.1507, "step": 15247 }, { "epoch": 0.02, "learning_rate": 4.997169095397953e-05, "loss": 1.0, "step": 15248 }, { "epoch": 0.02, "learning_rate": 4.997168721690062e-05, "loss": 1.147, "step": 15249 }, { "epoch": 0.02, "learning_rate": 4.99716834795752e-05, "loss": 1.1028, "step": 15250 }, { "epoch": 0.02, "learning_rate": 4.997167974200327e-05, "loss": 0.8925, "step": 15251 }, { "epoch": 0.02, "learning_rate": 4.997167600418483e-05, "loss": 1.176, "step": 15252 }, { "epoch": 0.02, "learning_rate": 4.997167226611988e-05, "loss": 0.7811, "step": 15253 }, { "epoch": 0.02, "learning_rate": 4.997166852780842e-05, "loss": 1.1627, "step": 15254 }, { "epoch": 0.02, "learning_rate": 4.9971664789250456e-05, "loss": 1.0467, "step": 15255 }, { "epoch": 0.02, "learning_rate": 4.997166105044597e-05, "loss": 0.9798, "step": 15256 }, { "epoch": 0.02, "learning_rate": 4.997165731139499e-05, "loss": 1.1626, "step": 15257 }, { "epoch": 0.02, "learning_rate": 4.997165357209749e-05, "loss": 0.8546, "step": 15258 }, { "epoch": 0.02, "learning_rate": 4.9971649832553477e-05, "loss": 1.1482, "step": 15259 }, { "epoch": 0.02, "learning_rate": 4.997164609276296e-05, "loss": 1.0805, "step": 15260 }, { "epoch": 0.02, "learning_rate": 4.9971642352725935e-05, "loss": 1.2735, "step": 15261 }, { "epoch": 0.02, "learning_rate": 4.99716386124424e-05, "loss": 1.3956, "step": 15262 }, { "epoch": 0.02, "learning_rate": 4.997163487191235e-05, "loss": 1.3231, "step": 15263 }, { "epoch": 0.02, "learning_rate": 4.99716311311358e-05, "loss": 1.3394, "step": 15264 }, { "epoch": 0.02, "learning_rate": 4.997162739011273e-05, "loss": 1.1015, "step": 15265 }, { "epoch": 0.02, "learning_rate": 4.997162364884316e-05, "loss": 0.8929, "step": 15266 }, { "epoch": 0.02, "learning_rate": 4.997161990732707e-05, "loss": 1.2235, "step": 15267 }, { "epoch": 0.02, "learning_rate": 4.997161616556448e-05, "loss": 1.1731, "step": 15268 }, { "epoch": 0.02, "learning_rate": 4.9971612423555384e-05, "loss": 1.2492, "step": 15269 }, { "epoch": 0.02, "learning_rate": 4.9971608681299767e-05, "loss": 1.1721, "step": 15270 }, { "epoch": 0.02, "learning_rate": 4.997160493879764e-05, "loss": 0.9599, "step": 15271 }, { "epoch": 0.02, "learning_rate": 4.997160119604902e-05, "loss": 1.2362, "step": 15272 }, { "epoch": 0.02, "learning_rate": 4.9971597453053874e-05, "loss": 1.1751, "step": 15273 }, { "epoch": 0.02, "learning_rate": 4.997159370981223e-05, "loss": 1.034, "step": 15274 }, { "epoch": 0.02, "learning_rate": 4.997158996632407e-05, "loss": 1.1111, "step": 15275 }, { "epoch": 0.02, "learning_rate": 4.9971586222589406e-05, "loss": 1.1358, "step": 15276 }, { "epoch": 0.02, "learning_rate": 4.9971582478608234e-05, "loss": 1.0874, "step": 15277 }, { "epoch": 0.02, "learning_rate": 4.997157873438055e-05, "loss": 0.946, "step": 15278 }, { "epoch": 0.02, "learning_rate": 4.9971574989906355e-05, "loss": 1.2169, "step": 15279 }, { "epoch": 0.02, "learning_rate": 4.9971571245185655e-05, "loss": 1.0533, "step": 15280 }, { "epoch": 0.02, "learning_rate": 4.997156750021844e-05, "loss": 1.1441, "step": 15281 }, { "epoch": 0.02, "learning_rate": 4.997156375500472e-05, "loss": 1.3319, "step": 15282 }, { "epoch": 0.02, "learning_rate": 4.9971560009544494e-05, "loss": 1.1559, "step": 15283 }, { "epoch": 0.02, "learning_rate": 4.997155626383776e-05, "loss": 1.1068, "step": 15284 }, { "epoch": 0.02, "learning_rate": 4.997155251788451e-05, "loss": 0.8974, "step": 15285 }, { "epoch": 0.02, "learning_rate": 4.9971548771684764e-05, "loss": 0.7726, "step": 15286 }, { "epoch": 0.02, "learning_rate": 4.9971545025238496e-05, "loss": 0.862, "step": 15287 }, { "epoch": 0.02, "learning_rate": 4.997154127854573e-05, "loss": 0.8907, "step": 15288 }, { "epoch": 0.02, "learning_rate": 4.9971537531606445e-05, "loss": 1.1277, "step": 15289 }, { "epoch": 0.02, "learning_rate": 4.9971533784420656e-05, "loss": 1.3205, "step": 15290 }, { "epoch": 0.02, "learning_rate": 4.997153003698836e-05, "loss": 0.9791, "step": 15291 }, { "epoch": 0.02, "learning_rate": 4.997152628930956e-05, "loss": 1.2524, "step": 15292 }, { "epoch": 0.02, "learning_rate": 4.997152254138424e-05, "loss": 1.2695, "step": 15293 }, { "epoch": 0.02, "learning_rate": 4.9971518793212416e-05, "loss": 1.2791, "step": 15294 }, { "epoch": 0.02, "learning_rate": 4.9971515044794086e-05, "loss": 1.0037, "step": 15295 }, { "epoch": 0.02, "learning_rate": 4.997151129612925e-05, "loss": 0.9392, "step": 15296 }, { "epoch": 0.02, "learning_rate": 4.99715075472179e-05, "loss": 1.1555, "step": 15297 }, { "epoch": 0.02, "learning_rate": 4.9971503798060046e-05, "loss": 1.1924, "step": 15298 }, { "epoch": 0.02, "learning_rate": 4.997150004865568e-05, "loss": 1.4881, "step": 15299 }, { "epoch": 0.02, "learning_rate": 4.9971496299004816e-05, "loss": 0.9699, "step": 15300 }, { "epoch": 0.02, "learning_rate": 4.997149254910744e-05, "loss": 0.8355, "step": 15301 }, { "epoch": 0.02, "learning_rate": 4.9971488798963544e-05, "loss": 1.1416, "step": 15302 }, { "epoch": 0.02, "learning_rate": 4.997148504857315e-05, "loss": 1.1125, "step": 15303 }, { "epoch": 0.02, "learning_rate": 4.9971481297936245e-05, "loss": 1.1312, "step": 15304 }, { "epoch": 0.02, "learning_rate": 4.997147754705284e-05, "loss": 1.1235, "step": 15305 }, { "epoch": 0.02, "learning_rate": 4.997147379592292e-05, "loss": 1.1419, "step": 15306 }, { "epoch": 0.02, "learning_rate": 4.9971470044546485e-05, "loss": 1.0428, "step": 15307 }, { "epoch": 0.02, "learning_rate": 4.997146629292355e-05, "loss": 1.3987, "step": 15308 }, { "epoch": 0.02, "learning_rate": 4.997146254105411e-05, "loss": 1.5465, "step": 15309 }, { "epoch": 0.02, "learning_rate": 4.997145878893816e-05, "loss": 1.1879, "step": 15310 }, { "epoch": 0.02, "learning_rate": 4.99714550365757e-05, "loss": 1.0822, "step": 15311 }, { "epoch": 0.02, "learning_rate": 4.997145128396674e-05, "loss": 1.0474, "step": 15312 }, { "epoch": 0.02, "learning_rate": 4.9971447531111256e-05, "loss": 1.0235, "step": 15313 }, { "epoch": 0.02, "learning_rate": 4.997144377800928e-05, "loss": 1.0694, "step": 15314 }, { "epoch": 0.02, "learning_rate": 4.997144002466079e-05, "loss": 0.8507, "step": 15315 }, { "epoch": 0.02, "learning_rate": 4.9971436271065796e-05, "loss": 1.1211, "step": 15316 }, { "epoch": 0.02, "learning_rate": 4.997143251722429e-05, "loss": 1.103, "step": 15317 }, { "epoch": 0.02, "learning_rate": 4.997142876313628e-05, "loss": 0.9408, "step": 15318 }, { "epoch": 0.02, "learning_rate": 4.9971425008801766e-05, "loss": 1.4217, "step": 15319 }, { "epoch": 0.02, "learning_rate": 4.9971421254220735e-05, "loss": 1.0344, "step": 15320 }, { "epoch": 0.02, "learning_rate": 4.9971417499393204e-05, "loss": 1.1387, "step": 15321 }, { "epoch": 0.02, "learning_rate": 4.997141374431916e-05, "loss": 0.8562, "step": 15322 }, { "epoch": 0.02, "learning_rate": 4.9971409988998615e-05, "loss": 1.0609, "step": 15323 }, { "epoch": 0.02, "learning_rate": 4.9971406233431564e-05, "loss": 0.9183, "step": 15324 }, { "epoch": 0.02, "learning_rate": 4.9971402477618e-05, "loss": 1.1931, "step": 15325 }, { "epoch": 0.02, "learning_rate": 4.997139872155793e-05, "loss": 1.5752, "step": 15326 }, { "epoch": 0.02, "learning_rate": 4.997139496525135e-05, "loss": 1.0926, "step": 15327 }, { "epoch": 0.02, "learning_rate": 4.997139120869826e-05, "loss": 1.2595, "step": 15328 }, { "epoch": 0.02, "learning_rate": 4.9971387451898676e-05, "loss": 1.2706, "step": 15329 }, { "epoch": 0.02, "learning_rate": 4.997138369485258e-05, "loss": 1.8415, "step": 15330 }, { "epoch": 0.02, "learning_rate": 4.997137993755997e-05, "loss": 1.6819, "step": 15331 }, { "epoch": 0.02, "learning_rate": 4.9971376180020856e-05, "loss": 1.5807, "step": 15332 }, { "epoch": 0.02, "learning_rate": 4.997137242223524e-05, "loss": 1.2045, "step": 15333 }, { "epoch": 0.02, "learning_rate": 4.9971368664203115e-05, "loss": 1.2469, "step": 15334 }, { "epoch": 0.02, "learning_rate": 4.997136490592448e-05, "loss": 1.1578, "step": 15335 }, { "epoch": 0.02, "learning_rate": 4.9971361147399346e-05, "loss": 1.1761, "step": 15336 }, { "epoch": 0.02, "learning_rate": 4.99713573886277e-05, "loss": 1.3091, "step": 15337 }, { "epoch": 0.02, "learning_rate": 4.997135362960954e-05, "loss": 1.1823, "step": 15338 }, { "epoch": 0.02, "learning_rate": 4.997134987034489e-05, "loss": 1.0622, "step": 15339 }, { "epoch": 0.02, "learning_rate": 4.997134611083372e-05, "loss": 0.9757, "step": 15340 }, { "epoch": 0.02, "learning_rate": 4.997134235107604e-05, "loss": 1.1082, "step": 15341 }, { "epoch": 0.02, "learning_rate": 4.997133859107187e-05, "loss": 1.1068, "step": 15342 }, { "epoch": 0.02, "learning_rate": 4.9971334830821184e-05, "loss": 0.771, "step": 15343 }, { "epoch": 0.02, "learning_rate": 4.997133107032399e-05, "loss": 1.1108, "step": 15344 }, { "epoch": 0.02, "learning_rate": 4.997132730958029e-05, "loss": 0.6597, "step": 15345 }, { "epoch": 0.02, "learning_rate": 4.997132354859009e-05, "loss": 0.8194, "step": 15346 }, { "epoch": 0.02, "learning_rate": 4.9971319787353377e-05, "loss": 0.6937, "step": 15347 }, { "epoch": 0.02, "learning_rate": 4.997131602587016e-05, "loss": 1.0032, "step": 15348 }, { "epoch": 0.02, "learning_rate": 4.9971312264140435e-05, "loss": 1.3638, "step": 15349 }, { "epoch": 0.02, "learning_rate": 4.9971308502164203e-05, "loss": 1.2591, "step": 15350 }, { "epoch": 0.02, "learning_rate": 4.9971304739941465e-05, "loss": 1.4679, "step": 15351 }, { "epoch": 0.02, "learning_rate": 4.997130097747223e-05, "loss": 1.1335, "step": 15352 }, { "epoch": 0.02, "learning_rate": 4.9971297214756475e-05, "loss": 1.2889, "step": 15353 }, { "epoch": 0.02, "learning_rate": 4.997129345179422e-05, "loss": 1.1222, "step": 15354 }, { "epoch": 0.02, "learning_rate": 4.997128968858546e-05, "loss": 1.2838, "step": 15355 }, { "epoch": 0.02, "learning_rate": 4.997128592513019e-05, "loss": 0.9344, "step": 15356 }, { "epoch": 0.02, "learning_rate": 4.997128216142842e-05, "loss": 0.9273, "step": 15357 }, { "epoch": 0.02, "learning_rate": 4.997127839748014e-05, "loss": 1.144, "step": 15358 }, { "epoch": 0.02, "learning_rate": 4.997127463328535e-05, "loss": 1.1295, "step": 15359 }, { "epoch": 0.02, "learning_rate": 4.9971270868844064e-05, "loss": 1.0321, "step": 15360 }, { "epoch": 0.02, "learning_rate": 4.9971267104156264e-05, "loss": 1.071, "step": 15361 }, { "epoch": 0.02, "learning_rate": 4.997126333922196e-05, "loss": 1.0994, "step": 15362 }, { "epoch": 0.02, "learning_rate": 4.997125957404115e-05, "loss": 0.6077, "step": 15363 }, { "epoch": 0.02, "learning_rate": 4.9971255808613834e-05, "loss": 0.5418, "step": 15364 }, { "epoch": 0.02, "learning_rate": 4.997125204294002e-05, "loss": 1.4358, "step": 15365 }, { "epoch": 0.02, "learning_rate": 4.997124827701969e-05, "loss": 1.2474, "step": 15366 }, { "epoch": 0.02, "learning_rate": 4.997124451085286e-05, "loss": 0.8211, "step": 15367 }, { "epoch": 0.02, "learning_rate": 4.997124074443953e-05, "loss": 0.9611, "step": 15368 }, { "epoch": 0.02, "learning_rate": 4.997123697777968e-05, "loss": 1.064, "step": 15369 }, { "epoch": 0.02, "learning_rate": 4.9971233210873335e-05, "loss": 1.1149, "step": 15370 }, { "epoch": 0.02, "learning_rate": 4.997122944372048e-05, "loss": 1.1558, "step": 15371 }, { "epoch": 0.02, "learning_rate": 4.997122567632112e-05, "loss": 1.1323, "step": 15372 }, { "epoch": 0.02, "learning_rate": 4.997122190867526e-05, "loss": 1.1184, "step": 15373 }, { "epoch": 0.02, "learning_rate": 4.997121814078288e-05, "loss": 1.1691, "step": 15374 }, { "epoch": 0.02, "learning_rate": 4.997121437264401e-05, "loss": 1.3078, "step": 15375 }, { "epoch": 0.02, "learning_rate": 4.9971210604258635e-05, "loss": 1.2348, "step": 15376 }, { "epoch": 0.02, "learning_rate": 4.997120683562675e-05, "loss": 1.1815, "step": 15377 }, { "epoch": 0.02, "learning_rate": 4.997120306674835e-05, "loss": 1.2258, "step": 15378 }, { "epoch": 0.02, "learning_rate": 4.997119929762346e-05, "loss": 0.9017, "step": 15379 }, { "epoch": 0.02, "learning_rate": 4.997119552825206e-05, "loss": 1.1378, "step": 15380 }, { "epoch": 0.02, "learning_rate": 4.997119175863415e-05, "loss": 1.1385, "step": 15381 }, { "epoch": 0.02, "learning_rate": 4.997118798876974e-05, "loss": 1.0615, "step": 15382 }, { "epoch": 0.02, "learning_rate": 4.997118421865882e-05, "loss": 0.6674, "step": 15383 }, { "epoch": 0.02, "learning_rate": 4.99711804483014e-05, "loss": 0.4038, "step": 15384 }, { "epoch": 0.02, "learning_rate": 4.9971176677697476e-05, "loss": 0.4091, "step": 15385 }, { "epoch": 0.02, "learning_rate": 4.997117290684704e-05, "loss": 0.3502, "step": 15386 }, { "epoch": 0.02, "learning_rate": 4.99711691357501e-05, "loss": 0.317, "step": 15387 }, { "epoch": 0.02, "learning_rate": 4.997116536440667e-05, "loss": 0.3268, "step": 15388 }, { "epoch": 0.02, "learning_rate": 4.997116159281672e-05, "loss": 0.7288, "step": 15389 }, { "epoch": 0.02, "learning_rate": 4.997115782098027e-05, "loss": 0.5239, "step": 15390 }, { "epoch": 0.02, "learning_rate": 4.997115404889731e-05, "loss": 0.4891, "step": 15391 }, { "epoch": 0.02, "learning_rate": 4.9971150276567855e-05, "loss": 0.3504, "step": 15392 }, { "epoch": 0.02, "learning_rate": 4.997114650399188e-05, "loss": 0.2326, "step": 15393 }, { "epoch": 0.02, "learning_rate": 4.997114273116942e-05, "loss": 0.2361, "step": 15394 }, { "epoch": 0.02, "learning_rate": 4.9971138958100445e-05, "loss": 1.0379, "step": 15395 }, { "epoch": 0.02, "learning_rate": 4.997113518478497e-05, "loss": 1.0147, "step": 15396 }, { "epoch": 0.02, "learning_rate": 4.997113141122298e-05, "loss": 1.1079, "step": 15397 }, { "epoch": 0.02, "learning_rate": 4.997112763741449e-05, "loss": 1.1869, "step": 15398 }, { "epoch": 0.02, "learning_rate": 4.9971123863359504e-05, "loss": 1.1236, "step": 15399 }, { "epoch": 0.02, "learning_rate": 4.997112008905801e-05, "loss": 1.1269, "step": 15400 }, { "epoch": 0.02, "learning_rate": 4.9971116314510005e-05, "loss": 1.165, "step": 15401 }, { "epoch": 0.02, "learning_rate": 4.9971112539715495e-05, "loss": 0.8438, "step": 15402 }, { "epoch": 0.02, "learning_rate": 4.997110876467449e-05, "loss": 1.2577, "step": 15403 }, { "epoch": 0.02, "learning_rate": 4.9971104989386976e-05, "loss": 1.2677, "step": 15404 }, { "epoch": 0.02, "learning_rate": 4.997110121385296e-05, "loss": 1.0797, "step": 15405 }, { "epoch": 0.02, "learning_rate": 4.9971097438072435e-05, "loss": 0.9922, "step": 15406 }, { "epoch": 0.02, "learning_rate": 4.997109366204541e-05, "loss": 0.9806, "step": 15407 }, { "epoch": 0.02, "learning_rate": 4.9971089885771874e-05, "loss": 0.6245, "step": 15408 }, { "epoch": 0.02, "learning_rate": 4.997108610925184e-05, "loss": 1.2668, "step": 15409 }, { "epoch": 0.02, "learning_rate": 4.9971082332485305e-05, "loss": 1.1568, "step": 15410 }, { "epoch": 0.02, "learning_rate": 4.997107855547226e-05, "loss": 1.2187, "step": 15411 }, { "epoch": 0.02, "learning_rate": 4.9971074778212715e-05, "loss": 0.9575, "step": 15412 }, { "epoch": 0.02, "learning_rate": 4.997107100070666e-05, "loss": 1.3958, "step": 15413 }, { "epoch": 0.02, "learning_rate": 4.997106722295411e-05, "loss": 1.2579, "step": 15414 }, { "epoch": 0.02, "learning_rate": 4.997106344495505e-05, "loss": 1.6806, "step": 15415 }, { "epoch": 0.02, "learning_rate": 4.997105966670949e-05, "loss": 1.4584, "step": 15416 }, { "epoch": 0.02, "learning_rate": 4.997105588821742e-05, "loss": 1.3861, "step": 15417 }, { "epoch": 0.02, "learning_rate": 4.9971052109478855e-05, "loss": 1.1164, "step": 15418 }, { "epoch": 0.02, "learning_rate": 4.997104833049378e-05, "loss": 1.0986, "step": 15419 }, { "epoch": 0.02, "learning_rate": 4.99710445512622e-05, "loss": 1.0733, "step": 15420 }, { "epoch": 0.02, "learning_rate": 4.997104077178412e-05, "loss": 1.006, "step": 15421 }, { "epoch": 0.02, "learning_rate": 4.997103699205954e-05, "loss": 1.0617, "step": 15422 }, { "epoch": 0.02, "learning_rate": 4.997103321208845e-05, "loss": 1.1979, "step": 15423 }, { "epoch": 0.02, "learning_rate": 4.997102943187086e-05, "loss": 1.1973, "step": 15424 }, { "epoch": 0.02, "learning_rate": 4.9971025651406766e-05, "loss": 1.3848, "step": 15425 }, { "epoch": 0.02, "learning_rate": 4.9971021870696165e-05, "loss": 1.2419, "step": 15426 }, { "epoch": 0.02, "learning_rate": 4.9971018089739064e-05, "loss": 0.8851, "step": 15427 }, { "epoch": 0.02, "learning_rate": 4.9971014308535456e-05, "loss": 1.1536, "step": 15428 }, { "epoch": 0.02, "learning_rate": 4.997101052708535e-05, "loss": 1.0217, "step": 15429 }, { "epoch": 0.02, "learning_rate": 4.997100674538874e-05, "loss": 1.0199, "step": 15430 }, { "epoch": 0.02, "learning_rate": 4.9971002963445624e-05, "loss": 1.0635, "step": 15431 }, { "epoch": 0.02, "learning_rate": 4.997099918125601e-05, "loss": 1.0826, "step": 15432 }, { "epoch": 0.02, "learning_rate": 4.9970995398819886e-05, "loss": 0.6998, "step": 15433 }, { "epoch": 0.02, "learning_rate": 4.997099161613726e-05, "loss": 1.1935, "step": 15434 }, { "epoch": 0.02, "learning_rate": 4.9970987833208134e-05, "loss": 0.9907, "step": 15435 }, { "epoch": 0.02, "learning_rate": 4.99709840500325e-05, "loss": 1.1779, "step": 15436 }, { "epoch": 0.02, "learning_rate": 4.997098026661037e-05, "loss": 1.2193, "step": 15437 }, { "epoch": 0.02, "learning_rate": 4.997097648294173e-05, "loss": 0.9302, "step": 15438 }, { "epoch": 0.02, "learning_rate": 4.9970972699026594e-05, "loss": 1.2012, "step": 15439 }, { "epoch": 0.02, "learning_rate": 4.9970968914864944e-05, "loss": 1.0083, "step": 15440 }, { "epoch": 0.02, "learning_rate": 4.997096513045681e-05, "loss": 1.1163, "step": 15441 }, { "epoch": 0.02, "learning_rate": 4.9970961345802156e-05, "loss": 1.3618, "step": 15442 }, { "epoch": 0.02, "learning_rate": 4.9970957560901005e-05, "loss": 1.189, "step": 15443 }, { "epoch": 0.02, "learning_rate": 4.997095377575335e-05, "loss": 1.1081, "step": 15444 }, { "epoch": 0.02, "learning_rate": 4.9970949990359196e-05, "loss": 1.2028, "step": 15445 }, { "epoch": 0.02, "learning_rate": 4.997094620471854e-05, "loss": 1.2225, "step": 15446 }, { "epoch": 0.02, "learning_rate": 4.997094241883137e-05, "loss": 1.2214, "step": 15447 }, { "epoch": 0.02, "learning_rate": 4.997093863269771e-05, "loss": 1.1734, "step": 15448 }, { "epoch": 0.02, "learning_rate": 4.997093484631754e-05, "loss": 1.0583, "step": 15449 }, { "epoch": 0.02, "learning_rate": 4.997093105969087e-05, "loss": 1.1264, "step": 15450 }, { "epoch": 0.02, "learning_rate": 4.9970927272817705e-05, "loss": 1.2098, "step": 15451 }, { "epoch": 0.02, "learning_rate": 4.9970923485698025e-05, "loss": 0.7442, "step": 15452 }, { "epoch": 0.02, "learning_rate": 4.9970919698331846e-05, "loss": 0.9979, "step": 15453 }, { "epoch": 0.02, "learning_rate": 4.9970915910719166e-05, "loss": 1.0807, "step": 15454 }, { "epoch": 0.02, "learning_rate": 4.997091212285999e-05, "loss": 1.0449, "step": 15455 }, { "epoch": 0.02, "learning_rate": 4.99709083347543e-05, "loss": 0.8006, "step": 15456 }, { "epoch": 0.02, "learning_rate": 4.9970904546402113e-05, "loss": 2.4288, "step": 15457 }, { "epoch": 0.02, "learning_rate": 4.9970900757803427e-05, "loss": 1.2356, "step": 15458 }, { "epoch": 0.02, "learning_rate": 4.997089696895824e-05, "loss": 1.3419, "step": 15459 }, { "epoch": 0.02, "learning_rate": 4.997089317986654e-05, "loss": 1.4171, "step": 15460 }, { "epoch": 0.02, "learning_rate": 4.9970889390528345e-05, "loss": 1.4521, "step": 15461 }, { "epoch": 0.02, "learning_rate": 4.997088560094365e-05, "loss": 1.4672, "step": 15462 }, { "epoch": 0.02, "learning_rate": 4.997088181111245e-05, "loss": 1.4953, "step": 15463 }, { "epoch": 0.02, "learning_rate": 4.997087802103475e-05, "loss": 1.1776, "step": 15464 }, { "epoch": 0.02, "learning_rate": 4.997087423071055e-05, "loss": 1.0757, "step": 15465 }, { "epoch": 0.02, "learning_rate": 4.997087044013984e-05, "loss": 0.5758, "step": 15466 }, { "epoch": 0.02, "learning_rate": 4.9970866649322637e-05, "loss": 1.2338, "step": 15467 }, { "epoch": 0.02, "learning_rate": 4.997086285825893e-05, "loss": 1.3703, "step": 15468 }, { "epoch": 0.02, "learning_rate": 4.997085906694872e-05, "loss": 0.9041, "step": 15469 }, { "epoch": 0.02, "learning_rate": 4.9970855275392004e-05, "loss": 1.1656, "step": 15470 }, { "epoch": 0.02, "learning_rate": 4.997085148358879e-05, "loss": 1.1609, "step": 15471 }, { "epoch": 0.02, "learning_rate": 4.997084769153907e-05, "loss": 0.9311, "step": 15472 }, { "epoch": 0.02, "learning_rate": 4.997084389924285e-05, "loss": 1.4685, "step": 15473 }, { "epoch": 0.02, "learning_rate": 4.997084010670013e-05, "loss": 1.0369, "step": 15474 }, { "epoch": 0.02, "learning_rate": 4.997083631391091e-05, "loss": 0.9598, "step": 15475 }, { "epoch": 0.02, "learning_rate": 4.9970832520875186e-05, "loss": 1.0708, "step": 15476 }, { "epoch": 0.02, "learning_rate": 4.997082872759297e-05, "loss": 1.5911, "step": 15477 }, { "epoch": 0.02, "learning_rate": 4.997082493406424e-05, "loss": 1.3801, "step": 15478 }, { "epoch": 0.02, "learning_rate": 4.997082114028901e-05, "loss": 1.0498, "step": 15479 }, { "epoch": 0.02, "learning_rate": 4.9970817346267285e-05, "loss": 1.2106, "step": 15480 }, { "epoch": 0.02, "learning_rate": 4.9970813551999054e-05, "loss": 0.9464, "step": 15481 }, { "epoch": 0.02, "learning_rate": 4.997080975748432e-05, "loss": 0.984, "step": 15482 }, { "epoch": 0.02, "learning_rate": 4.9970805962723085e-05, "loss": 1.0681, "step": 15483 }, { "epoch": 0.02, "learning_rate": 4.9970802167715354e-05, "loss": 1.3086, "step": 15484 }, { "epoch": 0.02, "learning_rate": 4.997079837246112e-05, "loss": 1.1362, "step": 15485 }, { "epoch": 0.02, "learning_rate": 4.997079457696038e-05, "loss": 1.2258, "step": 15486 }, { "epoch": 0.02, "learning_rate": 4.9970790781213146e-05, "loss": 1.4545, "step": 15487 }, { "epoch": 0.02, "learning_rate": 4.997078698521941e-05, "loss": 0.9743, "step": 15488 }, { "epoch": 0.02, "learning_rate": 4.997078318897916e-05, "loss": 1.0882, "step": 15489 }, { "epoch": 0.02, "learning_rate": 4.997077939249243e-05, "loss": 1.4589, "step": 15490 }, { "epoch": 0.02, "learning_rate": 4.9970775595759184e-05, "loss": 1.0599, "step": 15491 }, { "epoch": 0.02, "learning_rate": 4.997077179877944e-05, "loss": 0.3353, "step": 15492 }, { "epoch": 0.02, "learning_rate": 4.997076800155319e-05, "loss": 0.1309, "step": 15493 }, { "epoch": 0.02, "learning_rate": 4.997076420408045e-05, "loss": 0.2341, "step": 15494 }, { "epoch": 0.02, "learning_rate": 4.997076040636121e-05, "loss": 0.9858, "step": 15495 }, { "epoch": 0.02, "learning_rate": 4.997075660839546e-05, "loss": 0.9304, "step": 15496 }, { "epoch": 0.02, "learning_rate": 4.9970752810183214e-05, "loss": 1.1588, "step": 15497 }, { "epoch": 0.02, "learning_rate": 4.997074901172446e-05, "loss": 1.0732, "step": 15498 }, { "epoch": 0.02, "learning_rate": 4.9970745213019214e-05, "loss": 1.1569, "step": 15499 }, { "epoch": 0.02, "learning_rate": 4.997074141406747e-05, "loss": 1.1404, "step": 15500 }, { "epoch": 0.02, "eval_loss": 1.0326253175735474, "eval_runtime": 110.0077, "eval_samples_per_second": 12.59, "eval_steps_per_second": 3.154, "step": 15500 }, { "epoch": 0.02, "learning_rate": 4.997073761486922e-05, "loss": 1.084, "step": 15501 }, { "epoch": 0.02, "learning_rate": 4.9970733815424466e-05, "loss": 1.0516, "step": 15502 }, { "epoch": 0.02, "learning_rate": 4.997073001573321e-05, "loss": 1.1499, "step": 15503 }, { "epoch": 0.02, "learning_rate": 4.997072621579546e-05, "loss": 1.196, "step": 15504 }, { "epoch": 0.02, "learning_rate": 4.997072241561121e-05, "loss": 1.0631, "step": 15505 }, { "epoch": 0.02, "learning_rate": 4.9970718615180455e-05, "loss": 0.8387, "step": 15506 }, { "epoch": 0.02, "learning_rate": 4.997071481450321e-05, "loss": 1.1692, "step": 15507 }, { "epoch": 0.02, "learning_rate": 4.997071101357945e-05, "loss": 1.1049, "step": 15508 }, { "epoch": 0.02, "learning_rate": 4.99707072124092e-05, "loss": 1.1272, "step": 15509 }, { "epoch": 0.02, "learning_rate": 4.997070341099245e-05, "loss": 0.95, "step": 15510 }, { "epoch": 0.02, "learning_rate": 4.997069960932919e-05, "loss": 1.1122, "step": 15511 }, { "epoch": 0.02, "learning_rate": 4.997069580741943e-05, "loss": 1.0699, "step": 15512 }, { "epoch": 0.02, "learning_rate": 4.997069200526318e-05, "loss": 1.2448, "step": 15513 }, { "epoch": 0.02, "learning_rate": 4.9970688202860426e-05, "loss": 1.1439, "step": 15514 }, { "epoch": 0.02, "learning_rate": 4.997068440021117e-05, "loss": 1.1283, "step": 15515 }, { "epoch": 0.02, "learning_rate": 4.9970680597315415e-05, "loss": 0.8256, "step": 15516 }, { "epoch": 0.02, "learning_rate": 4.997067679417316e-05, "loss": 0.7458, "step": 15517 }, { "epoch": 0.02, "learning_rate": 4.997067299078441e-05, "loss": 1.4405, "step": 15518 }, { "epoch": 0.02, "learning_rate": 4.997066918714915e-05, "loss": 1.3147, "step": 15519 }, { "epoch": 0.02, "learning_rate": 4.9970665383267396e-05, "loss": 1.3012, "step": 15520 }, { "epoch": 0.02, "learning_rate": 4.9970661579139146e-05, "loss": 1.0775, "step": 15521 }, { "epoch": 0.02, "learning_rate": 4.997065777476438e-05, "loss": 1.2531, "step": 15522 }, { "epoch": 0.02, "learning_rate": 4.997065397014313e-05, "loss": 1.2214, "step": 15523 }, { "epoch": 0.02, "learning_rate": 4.997065016527538e-05, "loss": 1.1619, "step": 15524 }, { "epoch": 0.02, "learning_rate": 4.9970646360161125e-05, "loss": 1.0365, "step": 15525 }, { "epoch": 0.02, "learning_rate": 4.9970642554800374e-05, "loss": 1.0415, "step": 15526 }, { "epoch": 0.02, "learning_rate": 4.9970638749193116e-05, "loss": 1.0644, "step": 15527 }, { "epoch": 0.02, "learning_rate": 4.9970634943339365e-05, "loss": 1.1692, "step": 15528 }, { "epoch": 0.02, "learning_rate": 4.997063113723911e-05, "loss": 1.0218, "step": 15529 }, { "epoch": 0.02, "learning_rate": 4.997062733089236e-05, "loss": 1.1165, "step": 15530 }, { "epoch": 0.02, "learning_rate": 4.997062352429911e-05, "loss": 1.3149, "step": 15531 }, { "epoch": 0.02, "learning_rate": 4.9970619717459354e-05, "loss": 1.0921, "step": 15532 }, { "epoch": 0.02, "learning_rate": 4.997061591037311e-05, "loss": 1.1878, "step": 15533 }, { "epoch": 0.02, "learning_rate": 4.997061210304036e-05, "loss": 1.0947, "step": 15534 }, { "epoch": 0.02, "learning_rate": 4.9970608295461105e-05, "loss": 1.0979, "step": 15535 }, { "epoch": 0.02, "learning_rate": 4.997060448763536e-05, "loss": 1.2264, "step": 15536 }, { "epoch": 0.02, "learning_rate": 4.997060067956311e-05, "loss": 0.8933, "step": 15537 }, { "epoch": 0.02, "learning_rate": 4.997059687124436e-05, "loss": 1.6329, "step": 15538 }, { "epoch": 0.02, "learning_rate": 4.9970593062679117e-05, "loss": 1.0252, "step": 15539 }, { "epoch": 0.02, "learning_rate": 4.997058925386737e-05, "loss": 1.0822, "step": 15540 }, { "epoch": 0.02, "learning_rate": 4.9970585444809125e-05, "loss": 1.362, "step": 15541 }, { "epoch": 0.02, "learning_rate": 4.9970581635504385e-05, "loss": 1.6719, "step": 15542 }, { "epoch": 0.02, "learning_rate": 4.997057782595314e-05, "loss": 2.0673, "step": 15543 }, { "epoch": 0.02, "learning_rate": 4.99705740161554e-05, "loss": 1.0982, "step": 15544 }, { "epoch": 0.02, "learning_rate": 4.997057020611115e-05, "loss": 1.1138, "step": 15545 }, { "epoch": 0.02, "learning_rate": 4.997056639582042e-05, "loss": 1.0255, "step": 15546 }, { "epoch": 0.02, "learning_rate": 4.997056258528317e-05, "loss": 0.8348, "step": 15547 }, { "epoch": 0.02, "learning_rate": 4.997055877449944e-05, "loss": 1.1316, "step": 15548 }, { "epoch": 0.02, "learning_rate": 4.9970554963469206e-05, "loss": 1.0755, "step": 15549 }, { "epoch": 0.02, "learning_rate": 4.9970551152192466e-05, "loss": 0.81, "step": 15550 }, { "epoch": 0.02, "learning_rate": 4.997054734066924e-05, "loss": 0.8974, "step": 15551 }, { "epoch": 0.02, "learning_rate": 4.9970543528899505e-05, "loss": 0.92, "step": 15552 }, { "epoch": 0.02, "learning_rate": 4.997053971688327e-05, "loss": 0.8723, "step": 15553 }, { "epoch": 0.02, "learning_rate": 4.997053590462054e-05, "loss": 1.2132, "step": 15554 }, { "epoch": 0.02, "learning_rate": 4.997053209211131e-05, "loss": 1.1012, "step": 15555 }, { "epoch": 0.02, "learning_rate": 4.997052827935559e-05, "loss": 1.1338, "step": 15556 }, { "epoch": 0.02, "learning_rate": 4.997052446635336e-05, "loss": 1.1448, "step": 15557 }, { "epoch": 0.02, "learning_rate": 4.997052065310463e-05, "loss": 1.1036, "step": 15558 }, { "epoch": 0.02, "learning_rate": 4.997051683960942e-05, "loss": 0.7476, "step": 15559 }, { "epoch": 0.02, "learning_rate": 4.9970513025867696e-05, "loss": 1.2147, "step": 15560 }, { "epoch": 0.02, "learning_rate": 4.9970509211879474e-05, "loss": 1.139, "step": 15561 }, { "epoch": 0.02, "learning_rate": 4.997050539764475e-05, "loss": 0.7803, "step": 15562 }, { "epoch": 0.02, "learning_rate": 4.9970501583163544e-05, "loss": 0.4161, "step": 15563 }, { "epoch": 0.02, "learning_rate": 4.997049776843583e-05, "loss": 0.5792, "step": 15564 }, { "epoch": 0.02, "learning_rate": 4.9970493953461613e-05, "loss": 0.9664, "step": 15565 }, { "epoch": 0.02, "learning_rate": 4.99704901382409e-05, "loss": 1.0948, "step": 15566 }, { "epoch": 0.02, "learning_rate": 4.997048632277369e-05, "loss": 1.0837, "step": 15567 }, { "epoch": 0.02, "learning_rate": 4.9970482507059987e-05, "loss": 1.0737, "step": 15568 }, { "epoch": 0.02, "learning_rate": 4.9970478691099784e-05, "loss": 1.1318, "step": 15569 }, { "epoch": 0.02, "learning_rate": 4.997047487489308e-05, "loss": 1.0954, "step": 15570 }, { "epoch": 0.02, "learning_rate": 4.997047105843988e-05, "loss": 1.222, "step": 15571 }, { "epoch": 0.02, "learning_rate": 4.9970467241740176e-05, "loss": 1.1029, "step": 15572 }, { "epoch": 0.02, "learning_rate": 4.997046342479398e-05, "loss": 1.2778, "step": 15573 }, { "epoch": 0.02, "learning_rate": 4.997045960760128e-05, "loss": 1.0396, "step": 15574 }, { "epoch": 0.02, "learning_rate": 4.9970455790162094e-05, "loss": 1.2474, "step": 15575 }, { "epoch": 0.02, "learning_rate": 4.9970451972476404e-05, "loss": 0.8859, "step": 15576 }, { "epoch": 0.02, "learning_rate": 4.9970448154544214e-05, "loss": 1.1095, "step": 15577 }, { "epoch": 0.02, "learning_rate": 4.997044433636553e-05, "loss": 1.0876, "step": 15578 }, { "epoch": 0.02, "learning_rate": 4.997044051794034e-05, "loss": 1.0309, "step": 15579 }, { "epoch": 0.02, "learning_rate": 4.997043669926866e-05, "loss": 1.2329, "step": 15580 }, { "epoch": 0.02, "learning_rate": 4.997043288035048e-05, "loss": 0.9731, "step": 15581 }, { "epoch": 0.02, "learning_rate": 4.997042906118581e-05, "loss": 0.5186, "step": 15582 }, { "epoch": 0.02, "learning_rate": 4.997042524177463e-05, "loss": 0.9851, "step": 15583 }, { "epoch": 0.02, "learning_rate": 4.997042142211696e-05, "loss": 1.1017, "step": 15584 }, { "epoch": 0.02, "learning_rate": 4.997041760221279e-05, "loss": 1.1057, "step": 15585 }, { "epoch": 0.02, "learning_rate": 4.997041378206212e-05, "loss": 1.1474, "step": 15586 }, { "epoch": 0.02, "learning_rate": 4.9970409961664955e-05, "loss": 0.9451, "step": 15587 }, { "epoch": 0.02, "learning_rate": 4.99704061410213e-05, "loss": 0.9753, "step": 15588 }, { "epoch": 0.02, "learning_rate": 4.997040232013114e-05, "loss": 1.1145, "step": 15589 }, { "epoch": 0.02, "learning_rate": 4.997039849899448e-05, "loss": 1.0578, "step": 15590 }, { "epoch": 0.02, "learning_rate": 4.997039467761133e-05, "loss": 1.2155, "step": 15591 }, { "epoch": 0.02, "learning_rate": 4.997039085598167e-05, "loss": 1.0917, "step": 15592 }, { "epoch": 0.02, "learning_rate": 4.9970387034105534e-05, "loss": 1.1275, "step": 15593 }, { "epoch": 0.02, "learning_rate": 4.997038321198288e-05, "loss": 1.0806, "step": 15594 }, { "epoch": 0.02, "learning_rate": 4.9970379389613744e-05, "loss": 1.0596, "step": 15595 }, { "epoch": 0.02, "learning_rate": 4.9970375566998105e-05, "loss": 0.3714, "step": 15596 }, { "epoch": 0.02, "learning_rate": 4.997037174413597e-05, "loss": 0.366, "step": 15597 }, { "epoch": 0.02, "learning_rate": 4.9970367921027335e-05, "loss": 0.3177, "step": 15598 }, { "epoch": 0.02, "learning_rate": 4.997036409767221e-05, "loss": 0.3467, "step": 15599 }, { "epoch": 0.02, "learning_rate": 4.997036027407058e-05, "loss": 0.8203, "step": 15600 }, { "epoch": 0.02, "learning_rate": 4.997035645022246e-05, "loss": 1.2714, "step": 15601 }, { "epoch": 0.02, "learning_rate": 4.997035262612784e-05, "loss": 0.5748, "step": 15602 }, { "epoch": 0.02, "learning_rate": 4.9970348801786713e-05, "loss": 0.9596, "step": 15603 }, { "epoch": 0.02, "learning_rate": 4.99703449771991e-05, "loss": 1.3077, "step": 15604 }, { "epoch": 0.02, "learning_rate": 4.997034115236499e-05, "loss": 1.227, "step": 15605 }, { "epoch": 0.02, "learning_rate": 4.997033732728438e-05, "loss": 1.1659, "step": 15606 }, { "epoch": 0.02, "learning_rate": 4.9970333501957276e-05, "loss": 1.1771, "step": 15607 }, { "epoch": 0.02, "learning_rate": 4.997032967638368e-05, "loss": 1.172, "step": 15608 }, { "epoch": 0.02, "learning_rate": 4.9970325850563584e-05, "loss": 0.7679, "step": 15609 }, { "epoch": 0.02, "learning_rate": 4.9970322024496984e-05, "loss": 1.0431, "step": 15610 }, { "epoch": 0.02, "learning_rate": 4.99703181981839e-05, "loss": 1.2105, "step": 15611 }, { "epoch": 0.02, "learning_rate": 4.9970314371624304e-05, "loss": 0.9255, "step": 15612 }, { "epoch": 0.02, "learning_rate": 4.9970310544818225e-05, "loss": 1.1481, "step": 15613 }, { "epoch": 0.02, "learning_rate": 4.9970306717765645e-05, "loss": 0.9807, "step": 15614 }, { "epoch": 0.02, "learning_rate": 4.997030289046657e-05, "loss": 1.1817, "step": 15615 }, { "epoch": 0.02, "learning_rate": 4.9970299062921e-05, "loss": 1.0103, "step": 15616 }, { "epoch": 0.02, "learning_rate": 4.997029523512893e-05, "loss": 0.9556, "step": 15617 }, { "epoch": 0.02, "learning_rate": 4.9970291407090364e-05, "loss": 1.0283, "step": 15618 }, { "epoch": 0.02, "learning_rate": 4.9970287578805303e-05, "loss": 1.0917, "step": 15619 }, { "epoch": 0.02, "learning_rate": 4.997028375027374e-05, "loss": 0.9854, "step": 15620 }, { "epoch": 0.02, "learning_rate": 4.997027992149569e-05, "loss": 0.6732, "step": 15621 }, { "epoch": 0.02, "learning_rate": 4.997027609247114e-05, "loss": 1.2088, "step": 15622 }, { "epoch": 0.02, "learning_rate": 4.9970272263200094e-05, "loss": 0.7174, "step": 15623 }, { "epoch": 0.02, "learning_rate": 4.9970268433682554e-05, "loss": 0.8792, "step": 15624 }, { "epoch": 0.02, "learning_rate": 4.997026460391851e-05, "loss": 1.2889, "step": 15625 }, { "epoch": 0.02, "learning_rate": 4.9970260773907985e-05, "loss": 1.0461, "step": 15626 }, { "epoch": 0.02, "learning_rate": 4.997025694365095e-05, "loss": 1.0169, "step": 15627 }, { "epoch": 0.02, "learning_rate": 4.997025311314743e-05, "loss": 0.9686, "step": 15628 }, { "epoch": 0.02, "learning_rate": 4.997024928239741e-05, "loss": 0.9619, "step": 15629 }, { "epoch": 0.02, "learning_rate": 4.997024545140089e-05, "loss": 0.8603, "step": 15630 }, { "epoch": 0.02, "learning_rate": 4.997024162015787e-05, "loss": 0.802, "step": 15631 }, { "epoch": 0.02, "learning_rate": 4.997023778866837e-05, "loss": 0.7393, "step": 15632 }, { "epoch": 0.02, "learning_rate": 4.9970233956932364e-05, "loss": 0.7595, "step": 15633 }, { "epoch": 0.02, "learning_rate": 4.997023012494987e-05, "loss": 0.7294, "step": 15634 }, { "epoch": 0.02, "learning_rate": 4.9970226292720874e-05, "loss": 0.7022, "step": 15635 }, { "epoch": 0.02, "learning_rate": 4.997022246024538e-05, "loss": 0.6951, "step": 15636 }, { "epoch": 0.02, "learning_rate": 4.99702186275234e-05, "loss": 0.6884, "step": 15637 }, { "epoch": 0.02, "learning_rate": 4.9970214794554916e-05, "loss": 0.6694, "step": 15638 }, { "epoch": 0.02, "learning_rate": 4.9970210961339934e-05, "loss": 0.6479, "step": 15639 }, { "epoch": 0.02, "learning_rate": 4.9970207127878465e-05, "loss": 0.6595, "step": 15640 }, { "epoch": 0.02, "learning_rate": 4.997020329417049e-05, "loss": 0.6397, "step": 15641 }, { "epoch": 0.02, "learning_rate": 4.9970199460216035e-05, "loss": 0.6237, "step": 15642 }, { "epoch": 0.02, "learning_rate": 4.997019562601507e-05, "loss": 0.619, "step": 15643 }, { "epoch": 0.02, "learning_rate": 4.9970191791567624e-05, "loss": 0.6189, "step": 15644 }, { "epoch": 0.02, "learning_rate": 4.997018795687367e-05, "loss": 0.5966, "step": 15645 }, { "epoch": 0.02, "learning_rate": 4.9970184121933226e-05, "loss": 0.6017, "step": 15646 }, { "epoch": 0.02, "learning_rate": 4.997018028674629e-05, "loss": 0.599, "step": 15647 }, { "epoch": 0.02, "learning_rate": 4.9970176451312855e-05, "loss": 0.5918, "step": 15648 }, { "epoch": 0.02, "learning_rate": 4.9970172615632926e-05, "loss": 0.5899, "step": 15649 }, { "epoch": 0.02, "learning_rate": 4.99701687797065e-05, "loss": 0.5678, "step": 15650 }, { "epoch": 0.02, "learning_rate": 4.997016494353358e-05, "loss": 0.5699, "step": 15651 }, { "epoch": 0.02, "learning_rate": 4.9970161107114165e-05, "loss": 0.5593, "step": 15652 }, { "epoch": 0.02, "learning_rate": 4.9970157270448256e-05, "loss": 0.5443, "step": 15653 }, { "epoch": 0.02, "learning_rate": 4.997015343353585e-05, "loss": 0.5405, "step": 15654 }, { "epoch": 0.02, "learning_rate": 4.997014959637696e-05, "loss": 0.5375, "step": 15655 }, { "epoch": 0.02, "learning_rate": 4.997014575897156e-05, "loss": 0.5517, "step": 15656 }, { "epoch": 0.02, "learning_rate": 4.997014192131967e-05, "loss": 0.5406, "step": 15657 }, { "epoch": 0.02, "learning_rate": 4.997013808342129e-05, "loss": 0.5306, "step": 15658 }, { "epoch": 0.02, "learning_rate": 4.997013424527641e-05, "loss": 0.5325, "step": 15659 }, { "epoch": 0.02, "learning_rate": 4.997013040688504e-05, "loss": 0.5294, "step": 15660 }, { "epoch": 0.02, "learning_rate": 4.997012656824717e-05, "loss": 0.5251, "step": 15661 }, { "epoch": 0.02, "learning_rate": 4.997012272936281e-05, "loss": 0.5146, "step": 15662 }, { "epoch": 0.02, "learning_rate": 4.997011889023195e-05, "loss": 0.5038, "step": 15663 }, { "epoch": 0.02, "learning_rate": 4.9970115050854603e-05, "loss": 0.5171, "step": 15664 }, { "epoch": 0.02, "learning_rate": 4.997011121123076e-05, "loss": 0.5031, "step": 15665 }, { "epoch": 0.02, "learning_rate": 4.997010737136042e-05, "loss": 0.4977, "step": 15666 }, { "epoch": 0.02, "learning_rate": 4.997010353124359e-05, "loss": 0.5015, "step": 15667 }, { "epoch": 0.02, "learning_rate": 4.997009969088026e-05, "loss": 0.4928, "step": 15668 }, { "epoch": 0.02, "learning_rate": 4.9970095850270433e-05, "loss": 0.4876, "step": 15669 }, { "epoch": 0.02, "learning_rate": 4.9970092009414117e-05, "loss": 0.4895, "step": 15670 }, { "epoch": 0.02, "learning_rate": 4.99700881683113e-05, "loss": 0.4892, "step": 15671 }, { "epoch": 0.02, "learning_rate": 4.9970084326961996e-05, "loss": 0.483, "step": 15672 }, { "epoch": 0.02, "learning_rate": 4.99700804853662e-05, "loss": 0.4925, "step": 15673 }, { "epoch": 0.02, "learning_rate": 4.99700766435239e-05, "loss": 0.4721, "step": 15674 }, { "epoch": 0.02, "learning_rate": 4.997007280143512e-05, "loss": 0.4789, "step": 15675 }, { "epoch": 0.02, "learning_rate": 4.9970068959099835e-05, "loss": 0.4667, "step": 15676 }, { "epoch": 0.02, "learning_rate": 4.9970065116518064e-05, "loss": 0.4765, "step": 15677 }, { "epoch": 0.02, "learning_rate": 4.9970061273689794e-05, "loss": 0.4686, "step": 15678 }, { "epoch": 0.02, "learning_rate": 4.997005743061502e-05, "loss": 0.4647, "step": 15679 }, { "epoch": 0.02, "learning_rate": 4.9970053587293766e-05, "loss": 0.4689, "step": 15680 }, { "epoch": 0.02, "learning_rate": 4.997004974372602e-05, "loss": 0.473, "step": 15681 }, { "epoch": 0.02, "learning_rate": 4.997004589991178e-05, "loss": 0.4641, "step": 15682 }, { "epoch": 0.02, "learning_rate": 4.9970042055851034e-05, "loss": 1.1291, "step": 15683 }, { "epoch": 0.02, "learning_rate": 4.9970038211543804e-05, "loss": 1.5296, "step": 15684 }, { "epoch": 0.02, "learning_rate": 4.997003436699007e-05, "loss": 1.2582, "step": 15685 }, { "epoch": 0.02, "learning_rate": 4.997003052218986e-05, "loss": 0.8071, "step": 15686 }, { "epoch": 0.02, "learning_rate": 4.9970026677143145e-05, "loss": 1.3081, "step": 15687 }, { "epoch": 0.02, "learning_rate": 4.997002283184994e-05, "loss": 1.1521, "step": 15688 }, { "epoch": 0.02, "learning_rate": 4.9970018986310236e-05, "loss": 1.0356, "step": 15689 }, { "epoch": 0.02, "learning_rate": 4.997001514052404e-05, "loss": 0.9397, "step": 15690 }, { "epoch": 0.02, "learning_rate": 4.9970011294491354e-05, "loss": 1.3001, "step": 15691 }, { "epoch": 0.02, "learning_rate": 4.997000744821218e-05, "loss": 1.0951, "step": 15692 }, { "epoch": 0.02, "learning_rate": 4.99700036016865e-05, "loss": 1.0805, "step": 15693 }, { "epoch": 0.02, "learning_rate": 4.9969999754914335e-05, "loss": 0.788, "step": 15694 }, { "epoch": 0.02, "learning_rate": 4.996999590789567e-05, "loss": 1.1341, "step": 15695 }, { "epoch": 0.02, "learning_rate": 4.996999206063051e-05, "loss": 1.2373, "step": 15696 }, { "epoch": 0.02, "learning_rate": 4.996998821311887e-05, "loss": 0.9787, "step": 15697 }, { "epoch": 0.02, "learning_rate": 4.9969984365360723e-05, "loss": 1.1946, "step": 15698 }, { "epoch": 0.02, "learning_rate": 4.996998051735609e-05, "loss": 0.9751, "step": 15699 }, { "epoch": 0.02, "learning_rate": 4.996997666910497e-05, "loss": 1.1606, "step": 15700 }, { "epoch": 0.02, "learning_rate": 4.996997282060735e-05, "loss": 0.8047, "step": 15701 }, { "epoch": 0.02, "learning_rate": 4.996996897186323e-05, "loss": 1.475, "step": 15702 }, { "epoch": 0.02, "learning_rate": 4.996996512287263e-05, "loss": 1.091, "step": 15703 }, { "epoch": 0.02, "learning_rate": 4.996996127363552e-05, "loss": 1.1386, "step": 15704 }, { "epoch": 0.02, "learning_rate": 4.996995742415194e-05, "loss": 1.4912, "step": 15705 }, { "epoch": 0.02, "learning_rate": 4.996995357442185e-05, "loss": 1.189, "step": 15706 }, { "epoch": 0.02, "learning_rate": 4.996994972444527e-05, "loss": 1.1181, "step": 15707 }, { "epoch": 0.02, "learning_rate": 4.9969945874222204e-05, "loss": 1.01, "step": 15708 }, { "epoch": 0.02, "learning_rate": 4.996994202375264e-05, "loss": 1.1171, "step": 15709 }, { "epoch": 0.02, "learning_rate": 4.996993817303658e-05, "loss": 1.1735, "step": 15710 }, { "epoch": 0.02, "learning_rate": 4.996993432207403e-05, "loss": 1.1781, "step": 15711 }, { "epoch": 0.02, "learning_rate": 4.996993047086499e-05, "loss": 1.0949, "step": 15712 }, { "epoch": 0.02, "learning_rate": 4.996992661940946e-05, "loss": 1.2326, "step": 15713 }, { "epoch": 0.02, "learning_rate": 4.996992276770742e-05, "loss": 1.1842, "step": 15714 }, { "epoch": 0.02, "learning_rate": 4.9969918915758903e-05, "loss": 1.1811, "step": 15715 }, { "epoch": 0.02, "learning_rate": 4.996991506356389e-05, "loss": 0.9913, "step": 15716 }, { "epoch": 0.02, "learning_rate": 4.996991121112239e-05, "loss": 1.2811, "step": 15717 }, { "epoch": 0.02, "learning_rate": 4.9969907358434395e-05, "loss": 1.0541, "step": 15718 }, { "epoch": 0.02, "learning_rate": 4.9969903505499904e-05, "loss": 1.0126, "step": 15719 }, { "epoch": 0.02, "learning_rate": 4.996989965231892e-05, "loss": 1.0806, "step": 15720 }, { "epoch": 0.02, "learning_rate": 4.996989579889145e-05, "loss": 1.4714, "step": 15721 }, { "epoch": 0.02, "learning_rate": 4.996989194521748e-05, "loss": 1.0939, "step": 15722 }, { "epoch": 0.02, "learning_rate": 4.9969888091297024e-05, "loss": 1.6136, "step": 15723 }, { "epoch": 0.02, "learning_rate": 4.996988423713007e-05, "loss": 1.173, "step": 15724 }, { "epoch": 0.02, "learning_rate": 4.996988038271663e-05, "loss": 1.1176, "step": 15725 }, { "epoch": 0.02, "learning_rate": 4.996987652805669e-05, "loss": 1.0242, "step": 15726 }, { "epoch": 0.02, "learning_rate": 4.9969872673150264e-05, "loss": 0.894, "step": 15727 }, { "epoch": 0.02, "learning_rate": 4.9969868817997345e-05, "loss": 0.8595, "step": 15728 }, { "epoch": 0.02, "learning_rate": 4.9969864962597934e-05, "loss": 1.0393, "step": 15729 }, { "epoch": 0.02, "learning_rate": 4.996986110695203e-05, "loss": 1.8482, "step": 15730 }, { "epoch": 0.02, "learning_rate": 4.996985725105964e-05, "loss": 1.6636, "step": 15731 }, { "epoch": 0.02, "learning_rate": 4.996985339492075e-05, "loss": 1.0546, "step": 15732 }, { "epoch": 0.02, "learning_rate": 4.9969849538535374e-05, "loss": 0.9092, "step": 15733 }, { "epoch": 0.02, "learning_rate": 4.99698456819035e-05, "loss": 1.3023, "step": 15734 }, { "epoch": 0.02, "learning_rate": 4.996984182502514e-05, "loss": 1.2776, "step": 15735 }, { "epoch": 0.02, "learning_rate": 4.9969837967900285e-05, "loss": 1.2188, "step": 15736 }, { "epoch": 0.02, "learning_rate": 4.996983411052893e-05, "loss": 1.1586, "step": 15737 }, { "epoch": 0.02, "learning_rate": 4.99698302529111e-05, "loss": 1.073, "step": 15738 }, { "epoch": 0.02, "learning_rate": 4.996982639504677e-05, "loss": 1.191, "step": 15739 }, { "epoch": 0.02, "learning_rate": 4.996982253693595e-05, "loss": 1.5325, "step": 15740 }, { "epoch": 0.02, "learning_rate": 4.996981867857864e-05, "loss": 1.4183, "step": 15741 }, { "epoch": 0.02, "learning_rate": 4.9969814819974834e-05, "loss": 1.0973, "step": 15742 }, { "epoch": 0.02, "learning_rate": 4.9969810961124536e-05, "loss": 1.0266, "step": 15743 }, { "epoch": 0.02, "learning_rate": 4.996980710202775e-05, "loss": 1.138, "step": 15744 }, { "epoch": 0.02, "learning_rate": 4.996980324268447e-05, "loss": 1.094, "step": 15745 }, { "epoch": 0.02, "learning_rate": 4.99697993830947e-05, "loss": 0.8729, "step": 15746 }, { "epoch": 0.02, "learning_rate": 4.996979552325844e-05, "loss": 0.9571, "step": 15747 }, { "epoch": 0.02, "learning_rate": 4.996979166317569e-05, "loss": 1.3084, "step": 15748 }, { "epoch": 0.02, "learning_rate": 4.9969787802846445e-05, "loss": 1.196, "step": 15749 }, { "epoch": 0.02, "learning_rate": 4.996978394227071e-05, "loss": 0.8647, "step": 15750 }, { "epoch": 0.02, "learning_rate": 4.996978008144848e-05, "loss": 1.1199, "step": 15751 }, { "epoch": 0.02, "learning_rate": 4.996977622037976e-05, "loss": 0.9679, "step": 15752 }, { "epoch": 0.02, "learning_rate": 4.996977235906456e-05, "loss": 1.1227, "step": 15753 }, { "epoch": 0.02, "learning_rate": 4.996976849750285e-05, "loss": 1.3058, "step": 15754 }, { "epoch": 0.02, "learning_rate": 4.996976463569467e-05, "loss": 0.9908, "step": 15755 }, { "epoch": 0.02, "learning_rate": 4.996976077363998e-05, "loss": 1.087, "step": 15756 }, { "epoch": 0.02, "learning_rate": 4.996975691133882e-05, "loss": 1.2508, "step": 15757 }, { "epoch": 0.02, "learning_rate": 4.9969753048791146e-05, "loss": 1.0065, "step": 15758 }, { "epoch": 0.02, "learning_rate": 4.9969749185996995e-05, "loss": 0.9778, "step": 15759 }, { "epoch": 0.02, "learning_rate": 4.996974532295635e-05, "loss": 1.1656, "step": 15760 }, { "epoch": 0.02, "learning_rate": 4.996974145966922e-05, "loss": 0.9799, "step": 15761 }, { "epoch": 0.02, "learning_rate": 4.996973759613559e-05, "loss": 1.1699, "step": 15762 }, { "epoch": 0.02, "learning_rate": 4.9969733732355475e-05, "loss": 1.233, "step": 15763 }, { "epoch": 0.02, "learning_rate": 4.996972986832886e-05, "loss": 1.0715, "step": 15764 }, { "epoch": 0.02, "learning_rate": 4.9969726004055765e-05, "loss": 1.1573, "step": 15765 }, { "epoch": 0.02, "learning_rate": 4.996972213953617e-05, "loss": 1.1353, "step": 15766 }, { "epoch": 0.02, "learning_rate": 4.9969718274770095e-05, "loss": 1.2174, "step": 15767 }, { "epoch": 0.02, "learning_rate": 4.996971440975752e-05, "loss": 1.0838, "step": 15768 }, { "epoch": 0.02, "learning_rate": 4.996971054449846e-05, "loss": 0.8773, "step": 15769 }, { "epoch": 0.02, "learning_rate": 4.9969706678992913e-05, "loss": 1.0404, "step": 15770 }, { "epoch": 0.02, "learning_rate": 4.996970281324087e-05, "loss": 1.2448, "step": 15771 }, { "epoch": 0.02, "learning_rate": 4.996969894724234e-05, "loss": 0.8504, "step": 15772 }, { "epoch": 0.02, "learning_rate": 4.996969508099731e-05, "loss": 1.1887, "step": 15773 }, { "epoch": 0.02, "learning_rate": 4.99696912145058e-05, "loss": 1.093, "step": 15774 }, { "epoch": 0.02, "learning_rate": 4.99696873477678e-05, "loss": 1.0875, "step": 15775 }, { "epoch": 0.02, "learning_rate": 4.9969683480783304e-05, "loss": 1.0078, "step": 15776 }, { "epoch": 0.02, "learning_rate": 4.996967961355232e-05, "loss": 1.1218, "step": 15777 }, { "epoch": 0.02, "learning_rate": 4.996967574607485e-05, "loss": 1.436, "step": 15778 }, { "epoch": 0.02, "learning_rate": 4.9969671878350884e-05, "loss": 1.0902, "step": 15779 }, { "epoch": 0.02, "learning_rate": 4.996966801038043e-05, "loss": 0.6765, "step": 15780 }, { "epoch": 0.02, "learning_rate": 4.996966414216349e-05, "loss": 1.1032, "step": 15781 }, { "epoch": 0.02, "learning_rate": 4.996966027370005e-05, "loss": 1.1222, "step": 15782 }, { "epoch": 0.02, "learning_rate": 4.996965640499013e-05, "loss": 0.9782, "step": 15783 }, { "epoch": 0.02, "learning_rate": 4.996965253603372e-05, "loss": 0.9998, "step": 15784 }, { "epoch": 0.02, "learning_rate": 4.996964866683081e-05, "loss": 1.0014, "step": 15785 }, { "epoch": 0.02, "learning_rate": 4.996964479738142e-05, "loss": 1.6226, "step": 15786 }, { "epoch": 0.02, "learning_rate": 4.9969640927685533e-05, "loss": 1.6385, "step": 15787 }, { "epoch": 0.02, "learning_rate": 4.9969637057743156e-05, "loss": 1.5961, "step": 15788 }, { "epoch": 0.02, "learning_rate": 4.99696331875543e-05, "loss": 1.5374, "step": 15789 }, { "epoch": 0.02, "learning_rate": 4.996962931711894e-05, "loss": 1.485, "step": 15790 }, { "epoch": 0.02, "learning_rate": 4.99696254464371e-05, "loss": 1.4111, "step": 15791 }, { "epoch": 0.02, "learning_rate": 4.996962157550877e-05, "loss": 1.4236, "step": 15792 }, { "epoch": 0.02, "learning_rate": 4.996961770433395e-05, "loss": 1.3338, "step": 15793 }, { "epoch": 0.02, "learning_rate": 4.9969613832912636e-05, "loss": 1.2793, "step": 15794 }, { "epoch": 0.02, "learning_rate": 4.996960996124483e-05, "loss": 1.234, "step": 15795 }, { "epoch": 0.02, "learning_rate": 4.996960608933055e-05, "loss": 1.1902, "step": 15796 }, { "epoch": 0.02, "learning_rate": 4.996960221716976e-05, "loss": 1.1726, "step": 15797 }, { "epoch": 0.02, "learning_rate": 4.99695983447625e-05, "loss": 1.1311, "step": 15798 }, { "epoch": 0.02, "learning_rate": 4.996959447210874e-05, "loss": 1.0969, "step": 15799 }, { "epoch": 0.02, "learning_rate": 4.996959059920849e-05, "loss": 1.1718, "step": 15800 }, { "epoch": 0.02, "learning_rate": 4.9969586726061753e-05, "loss": 1.2182, "step": 15801 }, { "epoch": 0.02, "learning_rate": 4.996958285266853e-05, "loss": 1.1148, "step": 15802 }, { "epoch": 0.02, "learning_rate": 4.996957897902881e-05, "loss": 1.2094, "step": 15803 }, { "epoch": 0.02, "learning_rate": 4.996957510514261e-05, "loss": 1.1015, "step": 15804 }, { "epoch": 0.02, "learning_rate": 4.996957123100992e-05, "loss": 0.9884, "step": 15805 }, { "epoch": 0.02, "learning_rate": 4.9969567356630734e-05, "loss": 1.1299, "step": 15806 }, { "epoch": 0.02, "learning_rate": 4.996956348200506e-05, "loss": 1.0892, "step": 15807 }, { "epoch": 0.02, "learning_rate": 4.99695596071329e-05, "loss": 0.977, "step": 15808 }, { "epoch": 0.02, "learning_rate": 4.996955573201425e-05, "loss": 1.168, "step": 15809 }, { "epoch": 0.02, "learning_rate": 4.996955185664911e-05, "loss": 1.2611, "step": 15810 }, { "epoch": 0.02, "learning_rate": 4.996954798103748e-05, "loss": 1.1094, "step": 15811 }, { "epoch": 0.02, "learning_rate": 4.9969544105179364e-05, "loss": 1.2689, "step": 15812 }, { "epoch": 0.02, "learning_rate": 4.996954022907476e-05, "loss": 1.1934, "step": 15813 }, { "epoch": 0.02, "learning_rate": 4.9969536352723664e-05, "loss": 0.558, "step": 15814 }, { "epoch": 0.02, "learning_rate": 4.9969532476126094e-05, "loss": 0.8069, "step": 15815 }, { "epoch": 0.02, "learning_rate": 4.996952859928202e-05, "loss": 1.4292, "step": 15816 }, { "epoch": 0.02, "learning_rate": 4.996952472219146e-05, "loss": 1.057, "step": 15817 }, { "epoch": 0.02, "learning_rate": 4.996952084485441e-05, "loss": 1.5976, "step": 15818 }, { "epoch": 0.02, "learning_rate": 4.996951696727087e-05, "loss": 1.5469, "step": 15819 }, { "epoch": 0.02, "learning_rate": 4.9969513089440844e-05, "loss": 1.0112, "step": 15820 }, { "epoch": 0.02, "learning_rate": 4.996950921136433e-05, "loss": 1.5691, "step": 15821 }, { "epoch": 0.02, "learning_rate": 4.996950533304133e-05, "loss": 1.4388, "step": 15822 }, { "epoch": 0.02, "learning_rate": 4.9969501454471834e-05, "loss": 0.7895, "step": 15823 }, { "epoch": 0.02, "learning_rate": 4.996949757565586e-05, "loss": 0.6242, "step": 15824 }, { "epoch": 0.02, "learning_rate": 4.996949369659339e-05, "loss": 1.5227, "step": 15825 }, { "epoch": 0.02, "learning_rate": 4.996948981728443e-05, "loss": 1.4323, "step": 15826 }, { "epoch": 0.02, "learning_rate": 4.996948593772899e-05, "loss": 1.4847, "step": 15827 }, { "epoch": 0.02, "learning_rate": 4.996948205792705e-05, "loss": 1.0376, "step": 15828 }, { "epoch": 0.02, "learning_rate": 4.9969478177878636e-05, "loss": 1.1002, "step": 15829 }, { "epoch": 0.02, "learning_rate": 4.996947429758373e-05, "loss": 0.8962, "step": 15830 }, { "epoch": 0.02, "learning_rate": 4.996947041704233e-05, "loss": 0.9425, "step": 15831 }, { "epoch": 0.02, "learning_rate": 4.996946653625444e-05, "loss": 1.2343, "step": 15832 }, { "epoch": 0.02, "learning_rate": 4.9969462655220066e-05, "loss": 1.0559, "step": 15833 }, { "epoch": 0.02, "learning_rate": 4.99694587739392e-05, "loss": 1.1196, "step": 15834 }, { "epoch": 0.02, "learning_rate": 4.996945489241186e-05, "loss": 1.2245, "step": 15835 }, { "epoch": 0.02, "learning_rate": 4.9969451010638025e-05, "loss": 1.2001, "step": 15836 }, { "epoch": 0.02, "learning_rate": 4.9969447128617696e-05, "loss": 1.1571, "step": 15837 }, { "epoch": 0.02, "learning_rate": 4.996944324635088e-05, "loss": 1.1728, "step": 15838 }, { "epoch": 0.02, "learning_rate": 4.996943936383758e-05, "loss": 0.9492, "step": 15839 }, { "epoch": 0.02, "learning_rate": 4.996943548107779e-05, "loss": 0.8979, "step": 15840 }, { "epoch": 0.02, "learning_rate": 4.9969431598071513e-05, "loss": 1.3406, "step": 15841 }, { "epoch": 0.02, "learning_rate": 4.996942771481875e-05, "loss": 1.2007, "step": 15842 }, { "epoch": 0.02, "learning_rate": 4.99694238313195e-05, "loss": 1.0421, "step": 15843 }, { "epoch": 0.02, "learning_rate": 4.996941994757376e-05, "loss": 0.7652, "step": 15844 }, { "epoch": 0.02, "learning_rate": 4.996941606358153e-05, "loss": 0.6384, "step": 15845 }, { "epoch": 0.02, "learning_rate": 4.9969412179342824e-05, "loss": 0.7307, "step": 15846 }, { "epoch": 0.02, "learning_rate": 4.9969408294857615e-05, "loss": 1.1159, "step": 15847 }, { "epoch": 0.02, "learning_rate": 4.9969404410125934e-05, "loss": 1.1519, "step": 15848 }, { "epoch": 0.02, "learning_rate": 4.9969400525147745e-05, "loss": 1.179, "step": 15849 }, { "epoch": 0.02, "learning_rate": 4.9969396639923084e-05, "loss": 0.9514, "step": 15850 }, { "epoch": 0.02, "learning_rate": 4.9969392754451936e-05, "loss": 1.0071, "step": 15851 }, { "epoch": 0.02, "learning_rate": 4.996938886873429e-05, "loss": 0.697, "step": 15852 }, { "epoch": 0.02, "learning_rate": 4.9969384982770166e-05, "loss": 0.7484, "step": 15853 }, { "epoch": 0.02, "learning_rate": 4.996938109655955e-05, "loss": 1.0321, "step": 15854 }, { "epoch": 0.02, "learning_rate": 4.996937721010245e-05, "loss": 0.9037, "step": 15855 }, { "epoch": 0.02, "learning_rate": 4.996937332339887e-05, "loss": 0.9142, "step": 15856 }, { "epoch": 0.02, "learning_rate": 4.996936943644879e-05, "loss": 1.0078, "step": 15857 }, { "epoch": 0.02, "learning_rate": 4.996936554925223e-05, "loss": 1.4005, "step": 15858 }, { "epoch": 0.02, "learning_rate": 4.996936166180918e-05, "loss": 1.2878, "step": 15859 }, { "epoch": 0.02, "learning_rate": 4.996935777411964e-05, "loss": 1.5016, "step": 15860 }, { "epoch": 0.02, "learning_rate": 4.996935388618362e-05, "loss": 0.8036, "step": 15861 }, { "epoch": 0.02, "learning_rate": 4.996934999800111e-05, "loss": 1.081, "step": 15862 }, { "epoch": 0.02, "learning_rate": 4.996934610957211e-05, "loss": 0.9565, "step": 15863 }, { "epoch": 0.02, "learning_rate": 4.996934222089663e-05, "loss": 1.0705, "step": 15864 }, { "epoch": 0.02, "learning_rate": 4.9969338331974655e-05, "loss": 0.9327, "step": 15865 }, { "epoch": 0.02, "learning_rate": 4.99693344428062e-05, "loss": 1.0016, "step": 15866 }, { "epoch": 0.02, "learning_rate": 4.996933055339126e-05, "loss": 1.0836, "step": 15867 }, { "epoch": 0.02, "learning_rate": 4.996932666372982e-05, "loss": 0.4842, "step": 15868 }, { "epoch": 0.02, "learning_rate": 4.99693227738219e-05, "loss": 0.452, "step": 15869 }, { "epoch": 0.02, "learning_rate": 4.996931888366751e-05, "loss": 0.3079, "step": 15870 }, { "epoch": 0.02, "learning_rate": 4.9969314993266614e-05, "loss": 0.4565, "step": 15871 }, { "epoch": 0.02, "learning_rate": 4.9969311102619234e-05, "loss": 1.4961, "step": 15872 }, { "epoch": 0.02, "learning_rate": 4.996930721172537e-05, "loss": 0.9223, "step": 15873 }, { "epoch": 0.02, "learning_rate": 4.9969303320585014e-05, "loss": 0.8829, "step": 15874 }, { "epoch": 0.02, "learning_rate": 4.996929942919818e-05, "loss": 1.1984, "step": 15875 }, { "epoch": 0.02, "learning_rate": 4.996929553756486e-05, "loss": 0.9723, "step": 15876 }, { "epoch": 0.02, "learning_rate": 4.996929164568505e-05, "loss": 0.7124, "step": 15877 }, { "epoch": 0.02, "learning_rate": 4.996928775355875e-05, "loss": 1.1402, "step": 15878 }, { "epoch": 0.02, "learning_rate": 4.996928386118597e-05, "loss": 0.9664, "step": 15879 }, { "epoch": 0.02, "learning_rate": 4.9969279968566704e-05, "loss": 1.0777, "step": 15880 }, { "epoch": 0.02, "learning_rate": 4.9969276075700944e-05, "loss": 1.0117, "step": 15881 }, { "epoch": 0.02, "learning_rate": 4.9969272182588705e-05, "loss": 0.7138, "step": 15882 }, { "epoch": 0.02, "learning_rate": 4.996926828922997e-05, "loss": 0.6949, "step": 15883 }, { "epoch": 0.02, "learning_rate": 4.996926439562477e-05, "loss": 0.6052, "step": 15884 }, { "epoch": 0.02, "learning_rate": 4.996926050177306e-05, "loss": 0.6187, "step": 15885 }, { "epoch": 0.02, "learning_rate": 4.996925660767488e-05, "loss": 0.6995, "step": 15886 }, { "epoch": 0.02, "learning_rate": 4.9969252713330204e-05, "loss": 0.6548, "step": 15887 }, { "epoch": 0.02, "learning_rate": 4.996924881873905e-05, "loss": 0.5436, "step": 15888 }, { "epoch": 0.02, "learning_rate": 4.996924492390141e-05, "loss": 0.5614, "step": 15889 }, { "epoch": 0.02, "learning_rate": 4.996924102881728e-05, "loss": 0.4978, "step": 15890 }, { "epoch": 0.02, "learning_rate": 4.9969237133486664e-05, "loss": 0.6292, "step": 15891 }, { "epoch": 0.02, "learning_rate": 4.9969233237909565e-05, "loss": 1.0817, "step": 15892 }, { "epoch": 0.02, "learning_rate": 4.9969229342085974e-05, "loss": 0.889, "step": 15893 }, { "epoch": 0.02, "learning_rate": 4.99692254460159e-05, "loss": 1.2768, "step": 15894 }, { "epoch": 0.02, "learning_rate": 4.9969221549699345e-05, "loss": 1.5298, "step": 15895 }, { "epoch": 0.02, "learning_rate": 4.99692176531363e-05, "loss": 2.4553, "step": 15896 }, { "epoch": 0.02, "learning_rate": 4.996921375632677e-05, "loss": 1.4702, "step": 15897 }, { "epoch": 0.02, "learning_rate": 4.996920985927076e-05, "loss": 0.6525, "step": 15898 }, { "epoch": 0.02, "learning_rate": 4.996920596196826e-05, "loss": 1.0691, "step": 15899 }, { "epoch": 0.02, "learning_rate": 4.9969202064419276e-05, "loss": 1.0772, "step": 15900 }, { "epoch": 0.02, "learning_rate": 4.9969198166623805e-05, "loss": 0.3766, "step": 15901 }, { "epoch": 0.02, "learning_rate": 4.996919426858185e-05, "loss": 0.4076, "step": 15902 }, { "epoch": 0.02, "learning_rate": 4.9969190370293404e-05, "loss": 0.8458, "step": 15903 }, { "epoch": 0.02, "learning_rate": 4.996918647175848e-05, "loss": 1.4799, "step": 15904 }, { "epoch": 0.02, "learning_rate": 4.996918257297707e-05, "loss": 0.9496, "step": 15905 }, { "epoch": 0.02, "learning_rate": 4.9969178673949167e-05, "loss": 0.9569, "step": 15906 }, { "epoch": 0.02, "learning_rate": 4.996917477467479e-05, "loss": 1.0882, "step": 15907 }, { "epoch": 0.02, "learning_rate": 4.996917087515391e-05, "loss": 1.2019, "step": 15908 }, { "epoch": 0.02, "learning_rate": 4.996916697538656e-05, "loss": 1.1008, "step": 15909 }, { "epoch": 0.02, "learning_rate": 4.996916307537273e-05, "loss": 0.9841, "step": 15910 }, { "epoch": 0.02, "learning_rate": 4.9969159175112404e-05, "loss": 2.6722, "step": 15911 }, { "epoch": 0.02, "learning_rate": 4.9969155274605594e-05, "loss": 1.1447, "step": 15912 }, { "epoch": 0.02, "learning_rate": 4.9969151373852305e-05, "loss": 0.8918, "step": 15913 }, { "epoch": 0.02, "learning_rate": 4.996914747285252e-05, "loss": 0.1826, "step": 15914 }, { "epoch": 0.02, "learning_rate": 4.996914357160626e-05, "loss": 0.1544, "step": 15915 }, { "epoch": 0.02, "learning_rate": 4.996913967011351e-05, "loss": 0.1236, "step": 15916 }, { "epoch": 0.02, "learning_rate": 4.9969135768374274e-05, "loss": 0.0733, "step": 15917 }, { "epoch": 0.02, "learning_rate": 4.9969131866388565e-05, "loss": 0.1476, "step": 15918 }, { "epoch": 0.02, "learning_rate": 4.9969127964156356e-05, "loss": 0.2543, "step": 15919 }, { "epoch": 0.02, "learning_rate": 4.9969124061677674e-05, "loss": 0.4413, "step": 15920 }, { "epoch": 0.02, "learning_rate": 4.99691201589525e-05, "loss": 1.0583, "step": 15921 }, { "epoch": 0.02, "learning_rate": 4.9969116255980844e-05, "loss": 1.0857, "step": 15922 }, { "epoch": 0.02, "learning_rate": 4.99691123527627e-05, "loss": 1.2017, "step": 15923 }, { "epoch": 0.02, "learning_rate": 4.9969108449298074e-05, "loss": 1.2358, "step": 15924 }, { "epoch": 0.02, "learning_rate": 4.9969104545586966e-05, "loss": 1.0742, "step": 15925 }, { "epoch": 0.02, "learning_rate": 4.996910064162937e-05, "loss": 2.2286, "step": 15926 }, { "epoch": 0.02, "learning_rate": 4.996909673742529e-05, "loss": 3.0932, "step": 15927 }, { "epoch": 0.02, "learning_rate": 4.996909283297473e-05, "loss": 3.0961, "step": 15928 }, { "epoch": 0.02, "learning_rate": 4.996908892827769e-05, "loss": 3.0471, "step": 15929 }, { "epoch": 0.02, "learning_rate": 4.996908502333415e-05, "loss": 3.085, "step": 15930 }, { "epoch": 0.02, "learning_rate": 4.996908111814413e-05, "loss": 2.2995, "step": 15931 }, { "epoch": 0.02, "learning_rate": 4.996907721270764e-05, "loss": 1.6166, "step": 15932 }, { "epoch": 0.02, "learning_rate": 4.996907330702465e-05, "loss": 0.9383, "step": 15933 }, { "epoch": 0.02, "learning_rate": 4.9969069401095184e-05, "loss": 1.0632, "step": 15934 }, { "epoch": 0.02, "learning_rate": 4.996906549491923e-05, "loss": 1.0955, "step": 15935 }, { "epoch": 0.02, "learning_rate": 4.996906158849679e-05, "loss": 1.2428, "step": 15936 }, { "epoch": 0.02, "learning_rate": 4.996905768182787e-05, "loss": 1.0717, "step": 15937 }, { "epoch": 0.02, "learning_rate": 4.996905377491247e-05, "loss": 1.1718, "step": 15938 }, { "epoch": 0.02, "learning_rate": 4.996904986775057e-05, "loss": 1.1196, "step": 15939 }, { "epoch": 0.02, "learning_rate": 4.9969045960342197e-05, "loss": 1.2579, "step": 15940 }, { "epoch": 0.02, "learning_rate": 4.9969042052687344e-05, "loss": 0.9889, "step": 15941 }, { "epoch": 0.02, "learning_rate": 4.9969038144786004e-05, "loss": 0.9782, "step": 15942 }, { "epoch": 0.02, "learning_rate": 4.996903423663818e-05, "loss": 1.1278, "step": 15943 }, { "epoch": 0.02, "learning_rate": 4.996903032824387e-05, "loss": 0.9426, "step": 15944 }, { "epoch": 0.02, "learning_rate": 4.996902641960308e-05, "loss": 1.1588, "step": 15945 }, { "epoch": 0.02, "learning_rate": 4.99690225107158e-05, "loss": 1.3812, "step": 15946 }, { "epoch": 0.02, "learning_rate": 4.996901860158204e-05, "loss": 1.137, "step": 15947 }, { "epoch": 0.02, "learning_rate": 4.9969014692201796e-05, "loss": 1.0239, "step": 15948 }, { "epoch": 0.02, "learning_rate": 4.996901078257508e-05, "loss": 0.9451, "step": 15949 }, { "epoch": 0.02, "learning_rate": 4.996900687270186e-05, "loss": 1.1793, "step": 15950 }, { "epoch": 0.02, "learning_rate": 4.996900296258217e-05, "loss": 1.1752, "step": 15951 }, { "epoch": 0.02, "learning_rate": 4.996899905221599e-05, "loss": 1.1459, "step": 15952 }, { "epoch": 0.02, "learning_rate": 4.996899514160333e-05, "loss": 0.7601, "step": 15953 }, { "epoch": 0.02, "learning_rate": 4.9968991230744186e-05, "loss": 0.7365, "step": 15954 }, { "epoch": 0.02, "learning_rate": 4.996898731963856e-05, "loss": 1.0842, "step": 15955 }, { "epoch": 0.02, "learning_rate": 4.9968983408286444e-05, "loss": 1.3112, "step": 15956 }, { "epoch": 0.02, "learning_rate": 4.996897949668785e-05, "loss": 1.1491, "step": 15957 }, { "epoch": 0.02, "learning_rate": 4.9968975584842776e-05, "loss": 1.12, "step": 15958 }, { "epoch": 0.02, "learning_rate": 4.996897167275121e-05, "loss": 1.1248, "step": 15959 }, { "epoch": 0.02, "learning_rate": 4.996896776041317e-05, "loss": 0.9892, "step": 15960 }, { "epoch": 0.02, "learning_rate": 4.9968963847828644e-05, "loss": 1.2619, "step": 15961 }, { "epoch": 0.02, "learning_rate": 4.9968959934997634e-05, "loss": 0.9759, "step": 15962 }, { "epoch": 0.02, "learning_rate": 4.9968956021920144e-05, "loss": 1.103, "step": 15963 }, { "epoch": 0.02, "learning_rate": 4.996895210859617e-05, "loss": 1.113, "step": 15964 }, { "epoch": 0.02, "learning_rate": 4.9968948195025704e-05, "loss": 1.1596, "step": 15965 }, { "epoch": 0.02, "learning_rate": 4.996894428120876e-05, "loss": 1.0598, "step": 15966 }, { "epoch": 0.02, "learning_rate": 4.996894036714534e-05, "loss": 1.316, "step": 15967 }, { "epoch": 0.02, "learning_rate": 4.996893645283543e-05, "loss": 2.0774, "step": 15968 }, { "epoch": 0.02, "learning_rate": 4.996893253827904e-05, "loss": 0.9468, "step": 15969 }, { "epoch": 0.02, "learning_rate": 4.9968928623476163e-05, "loss": 1.0569, "step": 15970 }, { "epoch": 0.02, "learning_rate": 4.9968924708426815e-05, "loss": 1.1094, "step": 15971 }, { "epoch": 0.02, "learning_rate": 4.9968920793130965e-05, "loss": 1.1638, "step": 15972 }, { "epoch": 0.02, "learning_rate": 4.996891687758865e-05, "loss": 1.1114, "step": 15973 }, { "epoch": 0.02, "learning_rate": 4.996891296179985e-05, "loss": 1.0219, "step": 15974 }, { "epoch": 0.02, "learning_rate": 4.996890904576455e-05, "loss": 1.0367, "step": 15975 }, { "epoch": 0.02, "learning_rate": 4.996890512948279e-05, "loss": 1.0037, "step": 15976 }, { "epoch": 0.02, "learning_rate": 4.996890121295454e-05, "loss": 1.224, "step": 15977 }, { "epoch": 0.02, "learning_rate": 4.996889729617981e-05, "loss": 0.7068, "step": 15978 }, { "epoch": 0.02, "learning_rate": 4.996889337915859e-05, "loss": 0.9655, "step": 15979 }, { "epoch": 0.02, "learning_rate": 4.9968889461890886e-05, "loss": 1.5248, "step": 15980 }, { "epoch": 0.02, "learning_rate": 4.996888554437671e-05, "loss": 1.0857, "step": 15981 }, { "epoch": 0.02, "learning_rate": 4.9968881626616045e-05, "loss": 1.252, "step": 15982 }, { "epoch": 0.02, "learning_rate": 4.99688777086089e-05, "loss": 0.8674, "step": 15983 }, { "epoch": 0.02, "learning_rate": 4.996887379035527e-05, "loss": 1.0696, "step": 15984 }, { "epoch": 0.02, "learning_rate": 4.9968869871855164e-05, "loss": 1.2208, "step": 15985 }, { "epoch": 0.02, "learning_rate": 4.996886595310857e-05, "loss": 1.0645, "step": 15986 }, { "epoch": 0.02, "learning_rate": 4.99688620341155e-05, "loss": 0.9622, "step": 15987 }, { "epoch": 0.02, "learning_rate": 4.996885811487594e-05, "loss": 0.8759, "step": 15988 }, { "epoch": 0.02, "learning_rate": 4.99688541953899e-05, "loss": 1.0974, "step": 15989 }, { "epoch": 0.02, "learning_rate": 4.996885027565738e-05, "loss": 1.2132, "step": 15990 }, { "epoch": 0.02, "learning_rate": 4.996884635567837e-05, "loss": 0.9603, "step": 15991 }, { "epoch": 0.02, "learning_rate": 4.996884243545289e-05, "loss": 0.777, "step": 15992 }, { "epoch": 0.02, "learning_rate": 4.996883851498092e-05, "loss": 0.8661, "step": 15993 }, { "epoch": 0.02, "learning_rate": 4.996883459426247e-05, "loss": 0.9645, "step": 15994 }, { "epoch": 0.02, "learning_rate": 4.996883067329754e-05, "loss": 0.9234, "step": 15995 }, { "epoch": 0.02, "learning_rate": 4.996882675208613e-05, "loss": 1.0664, "step": 15996 }, { "epoch": 0.02, "learning_rate": 4.996882283062824e-05, "loss": 1.2544, "step": 15997 }, { "epoch": 0.02, "learning_rate": 4.996881890892387e-05, "loss": 0.7092, "step": 15998 }, { "epoch": 0.02, "learning_rate": 4.9968814986973e-05, "loss": 1.0547, "step": 15999 }, { "epoch": 0.02, "learning_rate": 4.9968811064775664e-05, "loss": 1.257, "step": 16000 }, { "epoch": 0.02, "eval_loss": 1.0286349058151245, "eval_runtime": 85.5057, "eval_samples_per_second": 16.198, "eval_steps_per_second": 4.058, "step": 16000 }, { "epoch": 0.02, "learning_rate": 4.996880714233184e-05, "loss": 1.335, "step": 16001 }, { "epoch": 0.02, "learning_rate": 4.996880321964155e-05, "loss": 1.232, "step": 16002 }, { "epoch": 0.02, "learning_rate": 4.996879929670476e-05, "loss": 0.966, "step": 16003 }, { "epoch": 0.02, "learning_rate": 4.9968795373521494e-05, "loss": 1.0119, "step": 16004 }, { "epoch": 0.02, "learning_rate": 4.996879145009174e-05, "loss": 1.1361, "step": 16005 }, { "epoch": 0.02, "learning_rate": 4.996878752641552e-05, "loss": 1.0948, "step": 16006 }, { "epoch": 0.02, "learning_rate": 4.996878360249281e-05, "loss": 1.1098, "step": 16007 }, { "epoch": 0.02, "learning_rate": 4.996877967832362e-05, "loss": 1.0399, "step": 16008 }, { "epoch": 0.02, "learning_rate": 4.996877575390795e-05, "loss": 1.1335, "step": 16009 }, { "epoch": 0.02, "learning_rate": 4.99687718292458e-05, "loss": 1.0527, "step": 16010 }, { "epoch": 0.02, "learning_rate": 4.9968767904337164e-05, "loss": 1.038, "step": 16011 }, { "epoch": 0.02, "learning_rate": 4.996876397918205e-05, "loss": 0.973, "step": 16012 }, { "epoch": 0.02, "learning_rate": 4.996876005378045e-05, "loss": 1.1094, "step": 16013 }, { "epoch": 0.02, "learning_rate": 4.9968756128132376e-05, "loss": 0.95, "step": 16014 }, { "epoch": 0.02, "learning_rate": 4.9968752202237814e-05, "loss": 1.0416, "step": 16015 }, { "epoch": 0.02, "learning_rate": 4.996874827609678e-05, "loss": 1.2394, "step": 16016 }, { "epoch": 0.02, "learning_rate": 4.996874434970925e-05, "loss": 1.0531, "step": 16017 }, { "epoch": 0.02, "learning_rate": 4.9968740423075255e-05, "loss": 1.0012, "step": 16018 }, { "epoch": 0.02, "learning_rate": 4.996873649619477e-05, "loss": 0.9496, "step": 16019 }, { "epoch": 0.02, "learning_rate": 4.996873256906781e-05, "loss": 1.1404, "step": 16020 }, { "epoch": 0.02, "learning_rate": 4.9968728641694364e-05, "loss": 0.7881, "step": 16021 }, { "epoch": 0.02, "learning_rate": 4.996872471407444e-05, "loss": 0.9775, "step": 16022 }, { "epoch": 0.02, "learning_rate": 4.9968720786208036e-05, "loss": 1.135, "step": 16023 }, { "epoch": 0.02, "learning_rate": 4.996871685809515e-05, "loss": 0.7833, "step": 16024 }, { "epoch": 0.02, "learning_rate": 4.996871292973578e-05, "loss": 1.0859, "step": 16025 }, { "epoch": 0.02, "learning_rate": 4.9968709001129935e-05, "loss": 1.2412, "step": 16026 }, { "epoch": 0.02, "learning_rate": 4.996870507227761e-05, "loss": 1.5017, "step": 16027 }, { "epoch": 0.02, "learning_rate": 4.9968701143178795e-05, "loss": 0.8793, "step": 16028 }, { "epoch": 0.02, "learning_rate": 4.996869721383351e-05, "loss": 1.0542, "step": 16029 }, { "epoch": 0.02, "learning_rate": 4.996869328424174e-05, "loss": 0.9377, "step": 16030 }, { "epoch": 0.02, "learning_rate": 4.996868935440349e-05, "loss": 0.8212, "step": 16031 }, { "epoch": 0.02, "learning_rate": 4.996868542431876e-05, "loss": 0.9972, "step": 16032 }, { "epoch": 0.02, "learning_rate": 4.9968681493987544e-05, "loss": 1.0806, "step": 16033 }, { "epoch": 0.02, "learning_rate": 4.996867756340986e-05, "loss": 1.1353, "step": 16034 }, { "epoch": 0.02, "learning_rate": 4.996867363258569e-05, "loss": 1.1027, "step": 16035 }, { "epoch": 0.02, "learning_rate": 4.9968669701515035e-05, "loss": 0.988, "step": 16036 }, { "epoch": 0.02, "learning_rate": 4.99686657701979e-05, "loss": 1.2436, "step": 16037 }, { "epoch": 0.02, "learning_rate": 4.996866183863429e-05, "loss": 1.0596, "step": 16038 }, { "epoch": 0.02, "learning_rate": 4.99686579068242e-05, "loss": 1.1205, "step": 16039 }, { "epoch": 0.02, "learning_rate": 4.996865397476763e-05, "loss": 1.6028, "step": 16040 }, { "epoch": 0.02, "learning_rate": 4.996865004246458e-05, "loss": 1.0554, "step": 16041 }, { "epoch": 0.02, "learning_rate": 4.996864610991505e-05, "loss": 1.0465, "step": 16042 }, { "epoch": 0.02, "learning_rate": 4.996864217711904e-05, "loss": 1.2438, "step": 16043 }, { "epoch": 0.02, "learning_rate": 4.996863824407654e-05, "loss": 0.8443, "step": 16044 }, { "epoch": 0.02, "learning_rate": 4.996863431078757e-05, "loss": 1.7163, "step": 16045 }, { "epoch": 0.02, "learning_rate": 4.996863037725212e-05, "loss": 2.501, "step": 16046 }, { "epoch": 0.02, "learning_rate": 4.996862644347019e-05, "loss": 1.183, "step": 16047 }, { "epoch": 0.02, "learning_rate": 4.9968622509441774e-05, "loss": 1.074, "step": 16048 }, { "epoch": 0.02, "learning_rate": 4.9968618575166885e-05, "loss": 1.0039, "step": 16049 }, { "epoch": 0.02, "learning_rate": 4.9968614640645516e-05, "loss": 1.2442, "step": 16050 }, { "epoch": 0.02, "learning_rate": 4.9968610705877667e-05, "loss": 1.2344, "step": 16051 }, { "epoch": 0.02, "learning_rate": 4.996860677086334e-05, "loss": 1.0345, "step": 16052 }, { "epoch": 0.02, "learning_rate": 4.996860283560253e-05, "loss": 1.1195, "step": 16053 }, { "epoch": 0.02, "learning_rate": 4.9968598900095234e-05, "loss": 1.2332, "step": 16054 }, { "epoch": 0.02, "learning_rate": 4.996859496434147e-05, "loss": 1.3668, "step": 16055 }, { "epoch": 0.02, "learning_rate": 4.9968591028341224e-05, "loss": 1.223, "step": 16056 }, { "epoch": 0.02, "learning_rate": 4.9968587092094496e-05, "loss": 0.9158, "step": 16057 }, { "epoch": 0.02, "learning_rate": 4.996858315560129e-05, "loss": 0.7889, "step": 16058 }, { "epoch": 0.02, "learning_rate": 4.99685792188616e-05, "loss": 0.831, "step": 16059 }, { "epoch": 0.02, "learning_rate": 4.996857528187544e-05, "loss": 0.6155, "step": 16060 }, { "epoch": 0.02, "learning_rate": 4.996857134464279e-05, "loss": 0.4096, "step": 16061 }, { "epoch": 0.02, "learning_rate": 4.996856740716367e-05, "loss": 0.7652, "step": 16062 }, { "epoch": 0.02, "learning_rate": 4.996856346943807e-05, "loss": 1.124, "step": 16063 }, { "epoch": 0.02, "learning_rate": 4.996855953146599e-05, "loss": 1.152, "step": 16064 }, { "epoch": 0.02, "learning_rate": 4.996855559324742e-05, "loss": 1.7036, "step": 16065 }, { "epoch": 0.02, "learning_rate": 4.996855165478238e-05, "loss": 1.3949, "step": 16066 }, { "epoch": 0.02, "learning_rate": 4.996854771607087e-05, "loss": 0.9946, "step": 16067 }, { "epoch": 0.02, "learning_rate": 4.996854377711286e-05, "loss": 1.1107, "step": 16068 }, { "epoch": 0.02, "learning_rate": 4.996853983790839e-05, "loss": 1.0923, "step": 16069 }, { "epoch": 0.02, "learning_rate": 4.996853589845744e-05, "loss": 0.9937, "step": 16070 }, { "epoch": 0.02, "learning_rate": 4.996853195876e-05, "loss": 1.1859, "step": 16071 }, { "epoch": 0.02, "learning_rate": 4.996852801881608e-05, "loss": 0.9965, "step": 16072 }, { "epoch": 0.02, "learning_rate": 4.99685240786257e-05, "loss": 0.9947, "step": 16073 }, { "epoch": 0.02, "learning_rate": 4.996852013818883e-05, "loss": 0.8839, "step": 16074 }, { "epoch": 0.02, "learning_rate": 4.996851619750547e-05, "loss": 1.1034, "step": 16075 }, { "epoch": 0.02, "learning_rate": 4.9968512256575645e-05, "loss": 0.9629, "step": 16076 }, { "epoch": 0.02, "learning_rate": 4.9968508315399335e-05, "loss": 0.7939, "step": 16077 }, { "epoch": 0.02, "learning_rate": 4.996850437397656e-05, "loss": 1.1657, "step": 16078 }, { "epoch": 0.02, "learning_rate": 4.996850043230729e-05, "loss": 0.9332, "step": 16079 }, { "epoch": 0.02, "learning_rate": 4.9968496490391544e-05, "loss": 1.1233, "step": 16080 }, { "epoch": 0.02, "learning_rate": 4.996849254822932e-05, "loss": 1.1291, "step": 16081 }, { "epoch": 0.02, "learning_rate": 4.9968488605820625e-05, "loss": 1.1487, "step": 16082 }, { "epoch": 0.02, "learning_rate": 4.996848466316545e-05, "loss": 1.0192, "step": 16083 }, { "epoch": 0.02, "learning_rate": 4.996848072026379e-05, "loss": 1.3341, "step": 16084 }, { "epoch": 0.02, "learning_rate": 4.996847677711566e-05, "loss": 1.1141, "step": 16085 }, { "epoch": 0.02, "learning_rate": 4.996847283372104e-05, "loss": 1.0351, "step": 16086 }, { "epoch": 0.02, "learning_rate": 4.996846889007995e-05, "loss": 0.921, "step": 16087 }, { "epoch": 0.02, "learning_rate": 4.9968464946192387e-05, "loss": 1.1743, "step": 16088 }, { "epoch": 0.02, "learning_rate": 4.996846100205833e-05, "loss": 1.0212, "step": 16089 }, { "epoch": 0.02, "learning_rate": 4.996845705767781e-05, "loss": 1.0725, "step": 16090 }, { "epoch": 0.02, "learning_rate": 4.99684531130508e-05, "loss": 1.1326, "step": 16091 }, { "epoch": 0.02, "learning_rate": 4.996844916817732e-05, "loss": 1.0872, "step": 16092 }, { "epoch": 0.02, "learning_rate": 4.9968445223057355e-05, "loss": 1.4123, "step": 16093 }, { "epoch": 0.02, "learning_rate": 4.996844127769092e-05, "loss": 1.2065, "step": 16094 }, { "epoch": 0.02, "learning_rate": 4.996843733207801e-05, "loss": 1.1871, "step": 16095 }, { "epoch": 0.02, "learning_rate": 4.9968433386218616e-05, "loss": 0.8718, "step": 16096 }, { "epoch": 0.02, "learning_rate": 4.996842944011274e-05, "loss": 1.1697, "step": 16097 }, { "epoch": 0.02, "learning_rate": 4.99684254937604e-05, "loss": 1.0986, "step": 16098 }, { "epoch": 0.02, "learning_rate": 4.996842154716157e-05, "loss": 1.1475, "step": 16099 }, { "epoch": 0.02, "learning_rate": 4.9968417600316266e-05, "loss": 1.0599, "step": 16100 }, { "epoch": 0.02, "learning_rate": 4.996841365322448e-05, "loss": 1.2229, "step": 16101 }, { "epoch": 0.02, "learning_rate": 4.996840970588622e-05, "loss": 1.251, "step": 16102 }, { "epoch": 0.02, "learning_rate": 4.9968405758301484e-05, "loss": 0.683, "step": 16103 }, { "epoch": 0.02, "learning_rate": 4.9968401810470266e-05, "loss": 1.1209, "step": 16104 }, { "epoch": 0.02, "learning_rate": 4.996839786239257e-05, "loss": 0.8694, "step": 16105 }, { "epoch": 0.02, "learning_rate": 4.9968393914068404e-05, "loss": 1.3749, "step": 16106 }, { "epoch": 0.02, "learning_rate": 4.996838996549775e-05, "loss": 1.0793, "step": 16107 }, { "epoch": 0.02, "learning_rate": 4.996838601668063e-05, "loss": 1.1458, "step": 16108 }, { "epoch": 0.02, "learning_rate": 4.9968382067617026e-05, "loss": 1.2545, "step": 16109 }, { "epoch": 0.02, "learning_rate": 4.996837811830695e-05, "loss": 0.9319, "step": 16110 }, { "epoch": 0.02, "learning_rate": 4.9968374168750386e-05, "loss": 0.6395, "step": 16111 }, { "epoch": 0.02, "learning_rate": 4.996837021894735e-05, "loss": 1.0393, "step": 16112 }, { "epoch": 0.02, "learning_rate": 4.996836626889784e-05, "loss": 0.9462, "step": 16113 }, { "epoch": 0.02, "learning_rate": 4.996836231860185e-05, "loss": 0.8878, "step": 16114 }, { "epoch": 0.02, "learning_rate": 4.996835836805939e-05, "loss": 1.6842, "step": 16115 }, { "epoch": 0.02, "learning_rate": 4.996835441727045e-05, "loss": 1.9949, "step": 16116 }, { "epoch": 0.02, "learning_rate": 4.9968350466235026e-05, "loss": 0.9831, "step": 16117 }, { "epoch": 0.02, "learning_rate": 4.9968346514953124e-05, "loss": 1.1943, "step": 16118 }, { "epoch": 0.02, "learning_rate": 4.996834256342475e-05, "loss": 0.9621, "step": 16119 }, { "epoch": 0.02, "learning_rate": 4.99683386116499e-05, "loss": 1.0894, "step": 16120 }, { "epoch": 0.02, "learning_rate": 4.996833465962857e-05, "loss": 1.1831, "step": 16121 }, { "epoch": 0.02, "learning_rate": 4.996833070736077e-05, "loss": 1.0437, "step": 16122 }, { "epoch": 0.02, "learning_rate": 4.996832675484648e-05, "loss": 0.6832, "step": 16123 }, { "epoch": 0.02, "learning_rate": 4.996832280208572e-05, "loss": 0.7591, "step": 16124 }, { "epoch": 0.02, "learning_rate": 4.996831884907849e-05, "loss": 0.977, "step": 16125 }, { "epoch": 0.02, "learning_rate": 4.9968314895824775e-05, "loss": 0.7077, "step": 16126 }, { "epoch": 0.02, "learning_rate": 4.996831094232459e-05, "loss": 0.6646, "step": 16127 }, { "epoch": 0.02, "learning_rate": 4.996830698857792e-05, "loss": 0.7334, "step": 16128 }, { "epoch": 0.02, "learning_rate": 4.996830303458478e-05, "loss": 1.3824, "step": 16129 }, { "epoch": 0.02, "learning_rate": 4.996829908034516e-05, "loss": 0.8863, "step": 16130 }, { "epoch": 0.02, "learning_rate": 4.996829512585907e-05, "loss": 1.0294, "step": 16131 }, { "epoch": 0.02, "learning_rate": 4.996829117112649e-05, "loss": 1.0788, "step": 16132 }, { "epoch": 0.02, "learning_rate": 4.9968287216147445e-05, "loss": 1.2313, "step": 16133 }, { "epoch": 0.02, "learning_rate": 4.996828326092192e-05, "loss": 0.791, "step": 16134 }, { "epoch": 0.02, "learning_rate": 4.9968279305449916e-05, "loss": 1.3411, "step": 16135 }, { "epoch": 0.02, "learning_rate": 4.996827534973144e-05, "loss": 1.1637, "step": 16136 }, { "epoch": 0.02, "learning_rate": 4.996827139376649e-05, "loss": 1.1389, "step": 16137 }, { "epoch": 0.02, "learning_rate": 4.996826743755506e-05, "loss": 1.0491, "step": 16138 }, { "epoch": 0.02, "learning_rate": 4.9968263481097156e-05, "loss": 0.9974, "step": 16139 }, { "epoch": 0.02, "learning_rate": 4.9968259524392767e-05, "loss": 1.1931, "step": 16140 }, { "epoch": 0.02, "learning_rate": 4.996825556744191e-05, "loss": 1.14, "step": 16141 }, { "epoch": 0.02, "learning_rate": 4.9968251610244575e-05, "loss": 1.0189, "step": 16142 }, { "epoch": 0.02, "learning_rate": 4.9968247652800766e-05, "loss": 1.1904, "step": 16143 }, { "epoch": 0.02, "learning_rate": 4.996824369511048e-05, "loss": 1.0098, "step": 16144 }, { "epoch": 0.02, "learning_rate": 4.9968239737173716e-05, "loss": 1.048, "step": 16145 }, { "epoch": 0.02, "learning_rate": 4.996823577899048e-05, "loss": 1.0062, "step": 16146 }, { "epoch": 0.02, "learning_rate": 4.996823182056076e-05, "loss": 1.0512, "step": 16147 }, { "epoch": 0.02, "learning_rate": 4.996822786188457e-05, "loss": 1.1053, "step": 16148 }, { "epoch": 0.02, "learning_rate": 4.9968223902961906e-05, "loss": 1.4602, "step": 16149 }, { "epoch": 0.02, "learning_rate": 4.9968219943792766e-05, "loss": 0.3209, "step": 16150 }, { "epoch": 0.02, "learning_rate": 4.9968215984377146e-05, "loss": 0.2716, "step": 16151 }, { "epoch": 0.02, "learning_rate": 4.996821202471505e-05, "loss": 0.1903, "step": 16152 }, { "epoch": 0.02, "learning_rate": 4.9968208064806486e-05, "loss": 1.2917, "step": 16153 }, { "epoch": 0.02, "learning_rate": 4.996820410465144e-05, "loss": 1.2429, "step": 16154 }, { "epoch": 0.02, "learning_rate": 4.996820014424992e-05, "loss": 1.0873, "step": 16155 }, { "epoch": 0.02, "learning_rate": 4.996819618360192e-05, "loss": 0.9876, "step": 16156 }, { "epoch": 0.02, "learning_rate": 4.996819222270746e-05, "loss": 1.0216, "step": 16157 }, { "epoch": 0.02, "learning_rate": 4.9968188261566506e-05, "loss": 1.1061, "step": 16158 }, { "epoch": 0.02, "learning_rate": 4.996818430017908e-05, "loss": 1.0066, "step": 16159 }, { "epoch": 0.02, "learning_rate": 4.996818033854519e-05, "loss": 1.181, "step": 16160 }, { "epoch": 0.02, "learning_rate": 4.996817637666481e-05, "loss": 1.3481, "step": 16161 }, { "epoch": 0.02, "learning_rate": 4.996817241453796e-05, "loss": 1.9089, "step": 16162 }, { "epoch": 0.02, "learning_rate": 4.9968168452164645e-05, "loss": 1.3536, "step": 16163 }, { "epoch": 0.02, "learning_rate": 4.996816448954484e-05, "loss": 1.3729, "step": 16164 }, { "epoch": 0.02, "learning_rate": 4.9968160526678565e-05, "loss": 1.4777, "step": 16165 }, { "epoch": 0.02, "learning_rate": 4.9968156563565816e-05, "loss": 1.051, "step": 16166 }, { "epoch": 0.02, "learning_rate": 4.99681526002066e-05, "loss": 1.4189, "step": 16167 }, { "epoch": 0.02, "learning_rate": 4.99681486366009e-05, "loss": 1.304, "step": 16168 }, { "epoch": 0.02, "learning_rate": 4.9968144672748716e-05, "loss": 0.6471, "step": 16169 }, { "epoch": 0.02, "learning_rate": 4.996814070865007e-05, "loss": 1.0248, "step": 16170 }, { "epoch": 0.02, "learning_rate": 4.996813674430495e-05, "loss": 1.2299, "step": 16171 }, { "epoch": 0.02, "learning_rate": 4.996813277971335e-05, "loss": 0.8333, "step": 16172 }, { "epoch": 0.02, "learning_rate": 4.996812881487527e-05, "loss": 0.8823, "step": 16173 }, { "epoch": 0.02, "learning_rate": 4.996812484979072e-05, "loss": 1.1194, "step": 16174 }, { "epoch": 0.02, "learning_rate": 4.99681208844597e-05, "loss": 1.1598, "step": 16175 }, { "epoch": 0.02, "learning_rate": 4.99681169188822e-05, "loss": 1.0874, "step": 16176 }, { "epoch": 0.02, "learning_rate": 4.996811295305823e-05, "loss": 0.8971, "step": 16177 }, { "epoch": 0.02, "learning_rate": 4.996810898698778e-05, "loss": 1.2511, "step": 16178 }, { "epoch": 0.02, "learning_rate": 4.996810502067086e-05, "loss": 1.2893, "step": 16179 }, { "epoch": 0.02, "learning_rate": 4.9968101054107465e-05, "loss": 0.9393, "step": 16180 }, { "epoch": 0.02, "learning_rate": 4.9968097087297586e-05, "loss": 1.1149, "step": 16181 }, { "epoch": 0.02, "learning_rate": 4.996809312024124e-05, "loss": 1.958, "step": 16182 }, { "epoch": 0.02, "learning_rate": 4.996808915293842e-05, "loss": 2.5813, "step": 16183 }, { "epoch": 0.02, "learning_rate": 4.9968085185389125e-05, "loss": 2.4732, "step": 16184 }, { "epoch": 0.02, "learning_rate": 4.9968081217593354e-05, "loss": 2.4139, "step": 16185 }, { "epoch": 0.02, "learning_rate": 4.9968077249551116e-05, "loss": 2.3772, "step": 16186 }, { "epoch": 0.02, "learning_rate": 4.996807328126239e-05, "loss": 1.754, "step": 16187 }, { "epoch": 0.02, "learning_rate": 4.99680693127272e-05, "loss": 0.9631, "step": 16188 }, { "epoch": 0.02, "learning_rate": 4.996806534394554e-05, "loss": 1.1956, "step": 16189 }, { "epoch": 0.02, "learning_rate": 4.996806137491739e-05, "loss": 1.2091, "step": 16190 }, { "epoch": 0.02, "learning_rate": 4.996805740564277e-05, "loss": 1.1417, "step": 16191 }, { "epoch": 0.02, "learning_rate": 4.9968053436121684e-05, "loss": 1.1485, "step": 16192 }, { "epoch": 0.02, "learning_rate": 4.996804946635412e-05, "loss": 1.3441, "step": 16193 }, { "epoch": 0.02, "learning_rate": 4.996804549634009e-05, "loss": 1.2406, "step": 16194 }, { "epoch": 0.02, "learning_rate": 4.996804152607957e-05, "loss": 1.259, "step": 16195 }, { "epoch": 0.02, "learning_rate": 4.996803755557259e-05, "loss": 0.7639, "step": 16196 }, { "epoch": 0.02, "learning_rate": 4.996803358481913e-05, "loss": 1.5595, "step": 16197 }, { "epoch": 0.02, "learning_rate": 4.9968029613819196e-05, "loss": 0.9994, "step": 16198 }, { "epoch": 0.02, "learning_rate": 4.996802564257279e-05, "loss": 0.9487, "step": 16199 }, { "epoch": 0.02, "learning_rate": 4.99680216710799e-05, "loss": 1.0912, "step": 16200 }, { "epoch": 0.02, "learning_rate": 4.9968017699340555e-05, "loss": 1.1955, "step": 16201 }, { "epoch": 0.02, "learning_rate": 4.996801372735472e-05, "loss": 0.8499, "step": 16202 }, { "epoch": 0.02, "learning_rate": 4.9968009755122416e-05, "loss": 1.0375, "step": 16203 }, { "epoch": 0.02, "learning_rate": 4.9968005782643644e-05, "loss": 1.1882, "step": 16204 }, { "epoch": 0.02, "learning_rate": 4.996800180991839e-05, "loss": 0.7487, "step": 16205 }, { "epoch": 0.02, "learning_rate": 4.996799783694667e-05, "loss": 1.0927, "step": 16206 }, { "epoch": 0.02, "learning_rate": 4.996799386372847e-05, "loss": 1.2371, "step": 16207 }, { "epoch": 0.02, "learning_rate": 4.99679898902638e-05, "loss": 1.1254, "step": 16208 }, { "epoch": 0.02, "learning_rate": 4.9967985916552654e-05, "loss": 1.0465, "step": 16209 }, { "epoch": 0.02, "learning_rate": 4.996798194259504e-05, "loss": 1.1878, "step": 16210 }, { "epoch": 0.02, "learning_rate": 4.996797796839095e-05, "loss": 1.2273, "step": 16211 }, { "epoch": 0.02, "learning_rate": 4.996797399394039e-05, "loss": 1.1758, "step": 16212 }, { "epoch": 0.02, "learning_rate": 4.9967970019243346e-05, "loss": 0.9587, "step": 16213 }, { "epoch": 0.02, "learning_rate": 4.996796604429984e-05, "loss": 1.2403, "step": 16214 }, { "epoch": 0.02, "learning_rate": 4.9967962069109855e-05, "loss": 1.0609, "step": 16215 }, { "epoch": 0.02, "learning_rate": 4.996795809367339e-05, "loss": 0.8989, "step": 16216 }, { "epoch": 0.02, "learning_rate": 4.996795411799047e-05, "loss": 0.9804, "step": 16217 }, { "epoch": 0.02, "learning_rate": 4.996795014206106e-05, "loss": 0.4521, "step": 16218 }, { "epoch": 0.02, "learning_rate": 4.996794616588518e-05, "loss": 0.3247, "step": 16219 }, { "epoch": 0.02, "learning_rate": 4.9967942189462833e-05, "loss": 0.4387, "step": 16220 }, { "epoch": 0.02, "learning_rate": 4.9967938212794006e-05, "loss": 0.5697, "step": 16221 }, { "epoch": 0.02, "learning_rate": 4.996793423587872e-05, "loss": 0.4907, "step": 16222 }, { "epoch": 0.02, "learning_rate": 4.9967930258716946e-05, "loss": 0.3862, "step": 16223 }, { "epoch": 0.02, "learning_rate": 4.9967926281308706e-05, "loss": 0.8819, "step": 16224 }, { "epoch": 0.02, "learning_rate": 4.996792230365399e-05, "loss": 1.2234, "step": 16225 }, { "epoch": 0.02, "learning_rate": 4.996791832575281e-05, "loss": 0.7997, "step": 16226 }, { "epoch": 0.02, "learning_rate": 4.996791434760514e-05, "loss": 1.0663, "step": 16227 }, { "epoch": 0.02, "learning_rate": 4.996791036921101e-05, "loss": 1.0899, "step": 16228 }, { "epoch": 0.02, "learning_rate": 4.9967906390570405e-05, "loss": 1.0533, "step": 16229 }, { "epoch": 0.02, "learning_rate": 4.9967902411683334e-05, "loss": 1.3995, "step": 16230 }, { "epoch": 0.02, "learning_rate": 4.9967898432549776e-05, "loss": 1.369, "step": 16231 }, { "epoch": 0.02, "learning_rate": 4.996789445316976e-05, "loss": 1.1766, "step": 16232 }, { "epoch": 0.02, "learning_rate": 4.996789047354326e-05, "loss": 0.9267, "step": 16233 }, { "epoch": 0.02, "learning_rate": 4.996788649367029e-05, "loss": 1.1148, "step": 16234 }, { "epoch": 0.02, "learning_rate": 4.996788251355085e-05, "loss": 1.1133, "step": 16235 }, { "epoch": 0.02, "learning_rate": 4.996787853318494e-05, "loss": 1.035, "step": 16236 }, { "epoch": 0.02, "learning_rate": 4.996787455257255e-05, "loss": 0.9821, "step": 16237 }, { "epoch": 0.02, "learning_rate": 4.9967870571713695e-05, "loss": 1.168, "step": 16238 }, { "epoch": 0.02, "learning_rate": 4.9967866590608366e-05, "loss": 1.21, "step": 16239 }, { "epoch": 0.02, "learning_rate": 4.9967862609256565e-05, "loss": 0.9015, "step": 16240 }, { "epoch": 0.02, "learning_rate": 4.996785862765829e-05, "loss": 1.1564, "step": 16241 }, { "epoch": 0.02, "learning_rate": 4.9967854645813536e-05, "loss": 0.9021, "step": 16242 }, { "epoch": 0.02, "learning_rate": 4.996785066372232e-05, "loss": 0.7426, "step": 16243 }, { "epoch": 0.02, "learning_rate": 4.9967846681384635e-05, "loss": 0.6433, "step": 16244 }, { "epoch": 0.02, "learning_rate": 4.996784269880047e-05, "loss": 0.66, "step": 16245 }, { "epoch": 0.02, "learning_rate": 4.9967838715969835e-05, "loss": 0.5208, "step": 16246 }, { "epoch": 0.02, "learning_rate": 4.996783473289273e-05, "loss": 1.1492, "step": 16247 }, { "epoch": 0.02, "learning_rate": 4.9967830749569143e-05, "loss": 0.9436, "step": 16248 }, { "epoch": 0.02, "learning_rate": 4.996782676599909e-05, "loss": 0.9296, "step": 16249 }, { "epoch": 0.02, "learning_rate": 4.996782278218257e-05, "loss": 1.1591, "step": 16250 }, { "epoch": 0.02, "learning_rate": 4.996781879811958e-05, "loss": 1.0893, "step": 16251 }, { "epoch": 0.02, "learning_rate": 4.996781481381011e-05, "loss": 1.0829, "step": 16252 }, { "epoch": 0.02, "learning_rate": 4.996781082925417e-05, "loss": 1.1987, "step": 16253 }, { "epoch": 0.02, "learning_rate": 4.996780684445176e-05, "loss": 0.7229, "step": 16254 }, { "epoch": 0.02, "learning_rate": 4.9967802859402886e-05, "loss": 1.1515, "step": 16255 }, { "epoch": 0.02, "learning_rate": 4.996779887410753e-05, "loss": 1.0667, "step": 16256 }, { "epoch": 0.02, "learning_rate": 4.996779488856571e-05, "loss": 1.4895, "step": 16257 }, { "epoch": 0.02, "learning_rate": 4.996779090277741e-05, "loss": 0.9377, "step": 16258 }, { "epoch": 0.02, "learning_rate": 4.996778691674264e-05, "loss": 1.115, "step": 16259 }, { "epoch": 0.02, "learning_rate": 4.996778293046141e-05, "loss": 1.0619, "step": 16260 }, { "epoch": 0.02, "learning_rate": 4.996777894393369e-05, "loss": 1.3029, "step": 16261 }, { "epoch": 0.02, "learning_rate": 4.996777495715951e-05, "loss": 1.0012, "step": 16262 }, { "epoch": 0.02, "learning_rate": 4.996777097013886e-05, "loss": 1.0984, "step": 16263 }, { "epoch": 0.02, "learning_rate": 4.9967766982871736e-05, "loss": 1.1476, "step": 16264 }, { "epoch": 0.02, "learning_rate": 4.9967762995358136e-05, "loss": 1.2302, "step": 16265 }, { "epoch": 0.02, "learning_rate": 4.996775900759807e-05, "loss": 1.5285, "step": 16266 }, { "epoch": 0.02, "learning_rate": 4.9967755019591536e-05, "loss": 1.3113, "step": 16267 }, { "epoch": 0.02, "learning_rate": 4.9967751031338524e-05, "loss": 1.1252, "step": 16268 }, { "epoch": 0.02, "learning_rate": 4.996774704283904e-05, "loss": 1.0883, "step": 16269 }, { "epoch": 0.02, "learning_rate": 4.9967743054093086e-05, "loss": 1.1315, "step": 16270 }, { "epoch": 0.02, "learning_rate": 4.996773906510067e-05, "loss": 0.8545, "step": 16271 }, { "epoch": 0.02, "learning_rate": 4.996773507586177e-05, "loss": 1.2811, "step": 16272 }, { "epoch": 0.02, "learning_rate": 4.9967731086376406e-05, "loss": 1.0656, "step": 16273 }, { "epoch": 0.02, "learning_rate": 4.996772709664457e-05, "loss": 0.9933, "step": 16274 }, { "epoch": 0.02, "learning_rate": 4.9967723106666265e-05, "loss": 1.0854, "step": 16275 }, { "epoch": 0.02, "learning_rate": 4.996771911644149e-05, "loss": 1.7743, "step": 16276 }, { "epoch": 0.02, "learning_rate": 4.996771512597024e-05, "loss": 1.2574, "step": 16277 }, { "epoch": 0.02, "learning_rate": 4.996771113525252e-05, "loss": 1.0404, "step": 16278 }, { "epoch": 0.02, "learning_rate": 4.9967707144288334e-05, "loss": 1.3512, "step": 16279 }, { "epoch": 0.02, "learning_rate": 4.996770315307767e-05, "loss": 0.8853, "step": 16280 }, { "epoch": 0.02, "learning_rate": 4.996769916162054e-05, "loss": 0.3981, "step": 16281 }, { "epoch": 0.02, "learning_rate": 4.9967695169916936e-05, "loss": 0.5679, "step": 16282 }, { "epoch": 0.02, "learning_rate": 4.996769117796687e-05, "loss": 1.0688, "step": 16283 }, { "epoch": 0.02, "learning_rate": 4.996768718577032e-05, "loss": 1.0377, "step": 16284 }, { "epoch": 0.02, "learning_rate": 4.996768319332731e-05, "loss": 1.1405, "step": 16285 }, { "epoch": 0.02, "learning_rate": 4.996767920063783e-05, "loss": 1.1835, "step": 16286 }, { "epoch": 0.02, "learning_rate": 4.996767520770187e-05, "loss": 1.2494, "step": 16287 }, { "epoch": 0.02, "learning_rate": 4.996767121451945e-05, "loss": 0.864, "step": 16288 }, { "epoch": 0.02, "learning_rate": 4.996766722109055e-05, "loss": 1.0897, "step": 16289 }, { "epoch": 0.02, "learning_rate": 4.9967663227415185e-05, "loss": 1.0804, "step": 16290 }, { "epoch": 0.02, "learning_rate": 4.9967659233493353e-05, "loss": 1.0939, "step": 16291 }, { "epoch": 0.02, "learning_rate": 4.996765523932505e-05, "loss": 0.8912, "step": 16292 }, { "epoch": 0.02, "learning_rate": 4.996765124491027e-05, "loss": 1.1055, "step": 16293 }, { "epoch": 0.02, "learning_rate": 4.996764725024903e-05, "loss": 1.3556, "step": 16294 }, { "epoch": 0.02, "learning_rate": 4.996764325534131e-05, "loss": 1.1317, "step": 16295 }, { "epoch": 0.02, "learning_rate": 4.996763926018713e-05, "loss": 1.0213, "step": 16296 }, { "epoch": 0.02, "learning_rate": 4.9967635264786465e-05, "loss": 1.0258, "step": 16297 }, { "epoch": 0.02, "learning_rate": 4.996763126913934e-05, "loss": 0.9683, "step": 16298 }, { "epoch": 0.02, "learning_rate": 4.996762727324574e-05, "loss": 1.2127, "step": 16299 }, { "epoch": 0.02, "learning_rate": 4.996762327710568e-05, "loss": 1.1685, "step": 16300 }, { "epoch": 0.02, "learning_rate": 4.9967619280719144e-05, "loss": 1.1319, "step": 16301 }, { "epoch": 0.02, "learning_rate": 4.9967615284086136e-05, "loss": 1.0097, "step": 16302 }, { "epoch": 0.02, "learning_rate": 4.996761128720666e-05, "loss": 1.0898, "step": 16303 }, { "epoch": 0.02, "learning_rate": 4.996760729008072e-05, "loss": 1.5117, "step": 16304 }, { "epoch": 0.02, "learning_rate": 4.99676032927083e-05, "loss": 1.26, "step": 16305 }, { "epoch": 0.02, "learning_rate": 4.9967599295089416e-05, "loss": 1.4611, "step": 16306 }, { "epoch": 0.02, "learning_rate": 4.996759529722407e-05, "loss": 1.0977, "step": 16307 }, { "epoch": 0.02, "learning_rate": 4.9967591299112244e-05, "loss": 0.9306, "step": 16308 }, { "epoch": 0.02, "learning_rate": 4.9967587300753946e-05, "loss": 0.998, "step": 16309 }, { "epoch": 0.02, "learning_rate": 4.996758330214919e-05, "loss": 1.0757, "step": 16310 }, { "epoch": 0.02, "learning_rate": 4.996757930329795e-05, "loss": 1.0549, "step": 16311 }, { "epoch": 0.02, "learning_rate": 4.996757530420025e-05, "loss": 1.2741, "step": 16312 }, { "epoch": 0.02, "learning_rate": 4.996757130485609e-05, "loss": 1.2743, "step": 16313 }, { "epoch": 0.02, "learning_rate": 4.996756730526544e-05, "loss": 1.3359, "step": 16314 }, { "epoch": 0.02, "learning_rate": 4.9967563305428335e-05, "loss": 0.9481, "step": 16315 }, { "epoch": 0.02, "learning_rate": 4.996755930534476e-05, "loss": 1.0658, "step": 16316 }, { "epoch": 0.02, "learning_rate": 4.996755530501471e-05, "loss": 0.5864, "step": 16317 }, { "epoch": 0.02, "learning_rate": 4.996755130443819e-05, "loss": 0.5213, "step": 16318 }, { "epoch": 0.02, "learning_rate": 4.99675473036152e-05, "loss": 0.3932, "step": 16319 }, { "epoch": 0.02, "learning_rate": 4.996754330254575e-05, "loss": 0.4016, "step": 16320 }, { "epoch": 0.02, "learning_rate": 4.9967539301229825e-05, "loss": 0.4387, "step": 16321 }, { "epoch": 0.02, "learning_rate": 4.996753529966743e-05, "loss": 0.3618, "step": 16322 }, { "epoch": 0.02, "learning_rate": 4.9967531297858564e-05, "loss": 0.3892, "step": 16323 }, { "epoch": 0.02, "learning_rate": 4.996752729580324e-05, "loss": 0.4412, "step": 16324 }, { "epoch": 0.02, "learning_rate": 4.996752329350144e-05, "loss": 0.3351, "step": 16325 }, { "epoch": 0.02, "learning_rate": 4.9967519290953165e-05, "loss": 0.7273, "step": 16326 }, { "epoch": 0.02, "learning_rate": 4.996751528815843e-05, "loss": 0.9889, "step": 16327 }, { "epoch": 0.02, "learning_rate": 4.996751128511722e-05, "loss": 1.0277, "step": 16328 }, { "epoch": 0.02, "learning_rate": 4.9967507281829544e-05, "loss": 0.964, "step": 16329 }, { "epoch": 0.02, "learning_rate": 4.99675032782954e-05, "loss": 1.108, "step": 16330 }, { "epoch": 0.02, "learning_rate": 4.996749927451479e-05, "loss": 1.089, "step": 16331 }, { "epoch": 0.02, "learning_rate": 4.9967495270487705e-05, "loss": 1.061, "step": 16332 }, { "epoch": 0.02, "learning_rate": 4.9967491266214155e-05, "loss": 1.014, "step": 16333 }, { "epoch": 0.02, "learning_rate": 4.996748726169413e-05, "loss": 1.1837, "step": 16334 }, { "epoch": 0.02, "learning_rate": 4.996748325692765e-05, "loss": 1.0725, "step": 16335 }, { "epoch": 0.02, "learning_rate": 4.9967479251914695e-05, "loss": 1.2409, "step": 16336 }, { "epoch": 0.02, "learning_rate": 4.996747524665527e-05, "loss": 1.2574, "step": 16337 }, { "epoch": 0.02, "learning_rate": 4.996747124114938e-05, "loss": 1.4221, "step": 16338 }, { "epoch": 0.02, "learning_rate": 4.996746723539702e-05, "loss": 1.1872, "step": 16339 }, { "epoch": 0.02, "learning_rate": 4.996746322939819e-05, "loss": 0.8682, "step": 16340 }, { "epoch": 0.02, "learning_rate": 4.996745922315289e-05, "loss": 1.2092, "step": 16341 }, { "epoch": 0.02, "learning_rate": 4.9967455216661124e-05, "loss": 1.0144, "step": 16342 }, { "epoch": 0.02, "learning_rate": 4.996745120992289e-05, "loss": 0.8675, "step": 16343 }, { "epoch": 0.02, "learning_rate": 4.996744720293819e-05, "loss": 0.9412, "step": 16344 }, { "epoch": 0.02, "learning_rate": 4.996744319570702e-05, "loss": 1.0969, "step": 16345 }, { "epoch": 0.02, "learning_rate": 4.996743918822938e-05, "loss": 1.1246, "step": 16346 }, { "epoch": 0.02, "learning_rate": 4.996743518050527e-05, "loss": 0.9383, "step": 16347 }, { "epoch": 0.02, "learning_rate": 4.99674311725347e-05, "loss": 1.54, "step": 16348 }, { "epoch": 0.02, "learning_rate": 4.996742716431766e-05, "loss": 1.1927, "step": 16349 }, { "epoch": 0.02, "learning_rate": 4.9967423155854145e-05, "loss": 0.9948, "step": 16350 }, { "epoch": 0.02, "learning_rate": 4.996741914714417e-05, "loss": 1.1343, "step": 16351 }, { "epoch": 0.02, "learning_rate": 4.9967415138187726e-05, "loss": 1.1142, "step": 16352 }, { "epoch": 0.02, "learning_rate": 4.996741112898481e-05, "loss": 1.0497, "step": 16353 }, { "epoch": 0.02, "learning_rate": 4.996740711953543e-05, "loss": 1.0757, "step": 16354 }, { "epoch": 0.02, "learning_rate": 4.996740310983958e-05, "loss": 1.3099, "step": 16355 }, { "epoch": 0.02, "learning_rate": 4.996739909989726e-05, "loss": 1.2023, "step": 16356 }, { "epoch": 0.02, "learning_rate": 4.996739508970848e-05, "loss": 0.9294, "step": 16357 }, { "epoch": 0.02, "learning_rate": 4.996739107927323e-05, "loss": 0.7495, "step": 16358 }, { "epoch": 0.02, "learning_rate": 4.996738706859151e-05, "loss": 0.7693, "step": 16359 }, { "epoch": 0.02, "learning_rate": 4.996738305766332e-05, "loss": 1.0183, "step": 16360 }, { "epoch": 0.02, "learning_rate": 4.996737904648866e-05, "loss": 0.9784, "step": 16361 }, { "epoch": 0.02, "learning_rate": 4.996737503506754e-05, "loss": 1.1849, "step": 16362 }, { "epoch": 0.02, "learning_rate": 4.9967371023399955e-05, "loss": 1.1728, "step": 16363 }, { "epoch": 0.02, "learning_rate": 4.99673670114859e-05, "loss": 1.0427, "step": 16364 }, { "epoch": 0.02, "learning_rate": 4.996736299932537e-05, "loss": 0.9394, "step": 16365 }, { "epoch": 0.02, "learning_rate": 4.9967358986918376e-05, "loss": 1.0391, "step": 16366 }, { "epoch": 0.02, "learning_rate": 4.9967354974264926e-05, "loss": 1.1171, "step": 16367 }, { "epoch": 0.02, "learning_rate": 4.9967350961364996e-05, "loss": 0.9364, "step": 16368 }, { "epoch": 0.02, "learning_rate": 4.99673469482186e-05, "loss": 1.5185, "step": 16369 }, { "epoch": 0.02, "learning_rate": 4.9967342934825744e-05, "loss": 1.0152, "step": 16370 }, { "epoch": 0.02, "learning_rate": 4.9967338921186415e-05, "loss": 0.9826, "step": 16371 }, { "epoch": 0.02, "learning_rate": 4.996733490730062e-05, "loss": 0.9392, "step": 16372 }, { "epoch": 0.02, "learning_rate": 4.996733089316836e-05, "loss": 0.7348, "step": 16373 }, { "epoch": 0.02, "learning_rate": 4.996732687878963e-05, "loss": 1.1406, "step": 16374 }, { "epoch": 0.02, "learning_rate": 4.996732286416443e-05, "loss": 0.9823, "step": 16375 }, { "epoch": 0.02, "learning_rate": 4.996731884929277e-05, "loss": 1.1228, "step": 16376 }, { "epoch": 0.02, "learning_rate": 4.996731483417464e-05, "loss": 1.0975, "step": 16377 }, { "epoch": 0.02, "learning_rate": 4.9967310818810046e-05, "loss": 1.0636, "step": 16378 }, { "epoch": 0.02, "learning_rate": 4.996730680319899e-05, "loss": 1.0302, "step": 16379 }, { "epoch": 0.02, "learning_rate": 4.9967302787341455e-05, "loss": 0.9555, "step": 16380 }, { "epoch": 0.02, "learning_rate": 4.996729877123746e-05, "loss": 1.1268, "step": 16381 }, { "epoch": 0.02, "learning_rate": 4.996729475488699e-05, "loss": 0.9097, "step": 16382 }, { "epoch": 0.02, "learning_rate": 4.996729073829007e-05, "loss": 1.0198, "step": 16383 }, { "epoch": 0.02, "learning_rate": 4.996728672144667e-05, "loss": 1.0253, "step": 16384 }, { "epoch": 0.02, "learning_rate": 4.996728270435681e-05, "loss": 0.9535, "step": 16385 }, { "epoch": 0.02, "learning_rate": 4.996727868702048e-05, "loss": 1.2251, "step": 16386 }, { "epoch": 0.02, "learning_rate": 4.996727466943768e-05, "loss": 1.1155, "step": 16387 }, { "epoch": 0.02, "learning_rate": 4.996727065160842e-05, "loss": 1.1432, "step": 16388 }, { "epoch": 0.02, "learning_rate": 4.9967266633532694e-05, "loss": 1.1097, "step": 16389 }, { "epoch": 0.02, "learning_rate": 4.99672626152105e-05, "loss": 1.2736, "step": 16390 }, { "epoch": 0.02, "learning_rate": 4.996725859664184e-05, "loss": 1.1754, "step": 16391 }, { "epoch": 0.02, "learning_rate": 4.996725457782671e-05, "loss": 1.1784, "step": 16392 }, { "epoch": 0.02, "learning_rate": 4.996725055876512e-05, "loss": 0.899, "step": 16393 }, { "epoch": 0.02, "learning_rate": 4.996724653945706e-05, "loss": 0.9968, "step": 16394 }, { "epoch": 0.02, "learning_rate": 4.9967242519902534e-05, "loss": 1.4226, "step": 16395 }, { "epoch": 0.02, "learning_rate": 4.996723850010154e-05, "loss": 1.2693, "step": 16396 }, { "epoch": 0.02, "learning_rate": 4.9967234480054084e-05, "loss": 1.063, "step": 16397 }, { "epoch": 0.02, "learning_rate": 4.996723045976016e-05, "loss": 0.9308, "step": 16398 }, { "epoch": 0.02, "learning_rate": 4.996722643921977e-05, "loss": 0.7395, "step": 16399 }, { "epoch": 0.02, "learning_rate": 4.996722241843291e-05, "loss": 0.8999, "step": 16400 }, { "epoch": 0.02, "learning_rate": 4.9967218397399596e-05, "loss": 1.2738, "step": 16401 }, { "epoch": 0.02, "learning_rate": 4.9967214376119806e-05, "loss": 1.0037, "step": 16402 }, { "epoch": 0.02, "learning_rate": 4.996721035459355e-05, "loss": 1.0335, "step": 16403 }, { "epoch": 0.02, "learning_rate": 4.9967206332820836e-05, "loss": 1.4803, "step": 16404 }, { "epoch": 0.02, "learning_rate": 4.996720231080165e-05, "loss": 1.2902, "step": 16405 }, { "epoch": 0.02, "learning_rate": 4.9967198288536e-05, "loss": 1.1419, "step": 16406 }, { "epoch": 0.02, "learning_rate": 4.996719426602388e-05, "loss": 1.2684, "step": 16407 }, { "epoch": 0.02, "learning_rate": 4.99671902432653e-05, "loss": 1.1004, "step": 16408 }, { "epoch": 0.02, "learning_rate": 4.996718622026025e-05, "loss": 1.6123, "step": 16409 }, { "epoch": 0.02, "learning_rate": 4.9967182197008746e-05, "loss": 1.0305, "step": 16410 }, { "epoch": 0.02, "learning_rate": 4.996717817351077e-05, "loss": 1.3689, "step": 16411 }, { "epoch": 0.02, "learning_rate": 4.9967174149766315e-05, "loss": 1.0398, "step": 16412 }, { "epoch": 0.02, "learning_rate": 4.996717012577541e-05, "loss": 1.0676, "step": 16413 }, { "epoch": 0.02, "learning_rate": 4.996716610153803e-05, "loss": 1.1456, "step": 16414 }, { "epoch": 0.02, "learning_rate": 4.9967162077054195e-05, "loss": 1.1777, "step": 16415 }, { "epoch": 0.02, "learning_rate": 4.9967158052323885e-05, "loss": 0.9724, "step": 16416 }, { "epoch": 0.02, "learning_rate": 4.9967154027347115e-05, "loss": 0.9042, "step": 16417 }, { "epoch": 0.02, "learning_rate": 4.9967150002123886e-05, "loss": 0.8798, "step": 16418 }, { "epoch": 0.02, "learning_rate": 4.9967145976654183e-05, "loss": 1.0784, "step": 16419 }, { "epoch": 0.02, "learning_rate": 4.9967141950938015e-05, "loss": 1.6644, "step": 16420 }, { "epoch": 0.02, "learning_rate": 4.996713792497538e-05, "loss": 1.7673, "step": 16421 }, { "epoch": 0.02, "learning_rate": 4.9967133898766286e-05, "loss": 1.38, "step": 16422 }, { "epoch": 0.02, "learning_rate": 4.9967129872310725e-05, "loss": 1.1686, "step": 16423 }, { "epoch": 0.02, "learning_rate": 4.99671258456087e-05, "loss": 1.0706, "step": 16424 }, { "epoch": 0.02, "learning_rate": 4.996712181866021e-05, "loss": 1.1724, "step": 16425 }, { "epoch": 0.02, "learning_rate": 4.996711779146525e-05, "loss": 0.9545, "step": 16426 }, { "epoch": 0.02, "learning_rate": 4.9967113764023834e-05, "loss": 1.1589, "step": 16427 }, { "epoch": 0.02, "learning_rate": 4.996710973633595e-05, "loss": 0.7277, "step": 16428 }, { "epoch": 0.02, "learning_rate": 4.99671057084016e-05, "loss": 0.7543, "step": 16429 }, { "epoch": 0.02, "learning_rate": 4.996710168022078e-05, "loss": 0.7892, "step": 16430 }, { "epoch": 0.02, "learning_rate": 4.9967097651793496e-05, "loss": 1.1139, "step": 16431 }, { "epoch": 0.02, "learning_rate": 4.996709362311975e-05, "loss": 1.0567, "step": 16432 }, { "epoch": 0.02, "learning_rate": 4.996708959419954e-05, "loss": 1.16, "step": 16433 }, { "epoch": 0.02, "learning_rate": 4.996708556503287e-05, "loss": 1.2519, "step": 16434 }, { "epoch": 0.02, "learning_rate": 4.996708153561973e-05, "loss": 0.7287, "step": 16435 }, { "epoch": 0.02, "learning_rate": 4.996707750596013e-05, "loss": 0.4836, "step": 16436 }, { "epoch": 0.02, "learning_rate": 4.996707347605406e-05, "loss": 0.4589, "step": 16437 }, { "epoch": 0.02, "learning_rate": 4.996706944590153e-05, "loss": 0.6768, "step": 16438 }, { "epoch": 0.02, "learning_rate": 4.9967065415502535e-05, "loss": 1.0248, "step": 16439 }, { "epoch": 0.02, "learning_rate": 4.9967061384857075e-05, "loss": 1.2461, "step": 16440 }, { "epoch": 0.02, "learning_rate": 4.996705735396515e-05, "loss": 1.1995, "step": 16441 }, { "epoch": 0.02, "learning_rate": 4.996705332282676e-05, "loss": 1.167, "step": 16442 }, { "epoch": 0.02, "learning_rate": 4.9967049291441904e-05, "loss": 1.1454, "step": 16443 }, { "epoch": 0.02, "learning_rate": 4.9967045259810586e-05, "loss": 1.2731, "step": 16444 }, { "epoch": 0.02, "learning_rate": 4.996704122793281e-05, "loss": 1.1783, "step": 16445 }, { "epoch": 0.02, "learning_rate": 4.9967037195808565e-05, "loss": 1.2154, "step": 16446 }, { "epoch": 0.02, "learning_rate": 4.996703316343785e-05, "loss": 2.1484, "step": 16447 }, { "epoch": 0.02, "learning_rate": 4.996702913082068e-05, "loss": 1.5237, "step": 16448 }, { "epoch": 0.02, "learning_rate": 4.9967025097957036e-05, "loss": 1.147, "step": 16449 }, { "epoch": 0.02, "learning_rate": 4.9967021064846934e-05, "loss": 0.8201, "step": 16450 }, { "epoch": 0.02, "learning_rate": 4.996701703149037e-05, "loss": 0.9415, "step": 16451 }, { "epoch": 0.02, "learning_rate": 4.9967012997887344e-05, "loss": 1.1502, "step": 16452 }, { "epoch": 0.02, "learning_rate": 4.996700896403784e-05, "loss": 1.1597, "step": 16453 }, { "epoch": 0.02, "learning_rate": 4.996700492994188e-05, "loss": 1.2252, "step": 16454 }, { "epoch": 0.02, "learning_rate": 4.996700089559947e-05, "loss": 1.3019, "step": 16455 }, { "epoch": 0.02, "learning_rate": 4.9966996861010584e-05, "loss": 1.1005, "step": 16456 }, { "epoch": 0.02, "learning_rate": 4.996699282617524e-05, "loss": 1.3154, "step": 16457 }, { "epoch": 0.02, "learning_rate": 4.996698879109342e-05, "loss": 0.9608, "step": 16458 }, { "epoch": 0.02, "learning_rate": 4.996698475576515e-05, "loss": 1.5192, "step": 16459 }, { "epoch": 0.02, "learning_rate": 4.9966980720190404e-05, "loss": 1.0495, "step": 16460 }, { "epoch": 0.02, "learning_rate": 4.9966976684369206e-05, "loss": 0.9472, "step": 16461 }, { "epoch": 0.02, "learning_rate": 4.9966972648301536e-05, "loss": 1.4851, "step": 16462 }, { "epoch": 0.02, "learning_rate": 4.996696861198741e-05, "loss": 1.2061, "step": 16463 }, { "epoch": 0.02, "learning_rate": 4.996696457542682e-05, "loss": 0.9333, "step": 16464 }, { "epoch": 0.02, "learning_rate": 4.9966960538619756e-05, "loss": 1.0153, "step": 16465 }, { "epoch": 0.02, "learning_rate": 4.996695650156624e-05, "loss": 0.9272, "step": 16466 }, { "epoch": 0.02, "learning_rate": 4.996695246426626e-05, "loss": 1.1944, "step": 16467 }, { "epoch": 0.02, "learning_rate": 4.9966948426719805e-05, "loss": 1.1198, "step": 16468 }, { "epoch": 0.02, "learning_rate": 4.99669443889269e-05, "loss": 1.1995, "step": 16469 }, { "epoch": 0.02, "learning_rate": 4.9966940350887526e-05, "loss": 1.0167, "step": 16470 }, { "epoch": 0.02, "learning_rate": 4.996693631260169e-05, "loss": 0.9834, "step": 16471 }, { "epoch": 0.02, "learning_rate": 4.996693227406939e-05, "loss": 1.259, "step": 16472 }, { "epoch": 0.02, "learning_rate": 4.9966928235290636e-05, "loss": 1.9653, "step": 16473 }, { "epoch": 0.02, "learning_rate": 4.9966924196265404e-05, "loss": 1.8612, "step": 16474 }, { "epoch": 0.02, "learning_rate": 4.996692015699372e-05, "loss": 1.0861, "step": 16475 }, { "epoch": 0.02, "learning_rate": 4.996691611747557e-05, "loss": 1.076, "step": 16476 }, { "epoch": 0.02, "learning_rate": 4.9966912077710953e-05, "loss": 1.2206, "step": 16477 }, { "epoch": 0.02, "learning_rate": 4.996690803769988e-05, "loss": 0.8671, "step": 16478 }, { "epoch": 0.02, "learning_rate": 4.9966903997442335e-05, "loss": 0.8373, "step": 16479 }, { "epoch": 0.02, "learning_rate": 4.996689995693834e-05, "loss": 0.9271, "step": 16480 }, { "epoch": 0.02, "learning_rate": 4.996689591618787e-05, "loss": 1.0112, "step": 16481 }, { "epoch": 0.02, "learning_rate": 4.9966891875190944e-05, "loss": 1.0359, "step": 16482 }, { "epoch": 0.02, "learning_rate": 4.996688783394755e-05, "loss": 1.1535, "step": 16483 }, { "epoch": 0.02, "learning_rate": 4.9966883792457704e-05, "loss": 1.0136, "step": 16484 }, { "epoch": 0.02, "learning_rate": 4.9966879750721385e-05, "loss": 1.2872, "step": 16485 }, { "epoch": 0.02, "learning_rate": 4.9966875708738606e-05, "loss": 1.115, "step": 16486 }, { "epoch": 0.02, "learning_rate": 4.996687166650937e-05, "loss": 1.1417, "step": 16487 }, { "epoch": 0.02, "learning_rate": 4.996686762403366e-05, "loss": 1.2457, "step": 16488 }, { "epoch": 0.02, "learning_rate": 4.9966863581311505e-05, "loss": 1.0294, "step": 16489 }, { "epoch": 0.02, "learning_rate": 4.9966859538342874e-05, "loss": 0.5416, "step": 16490 }, { "epoch": 0.02, "learning_rate": 4.9966855495127784e-05, "loss": 0.5301, "step": 16491 }, { "epoch": 0.02, "learning_rate": 4.996685145166623e-05, "loss": 0.5015, "step": 16492 }, { "epoch": 0.02, "learning_rate": 4.996684740795822e-05, "loss": 0.4428, "step": 16493 }, { "epoch": 0.02, "learning_rate": 4.9966843364003744e-05, "loss": 0.4028, "step": 16494 }, { "epoch": 0.02, "learning_rate": 4.99668393198028e-05, "loss": 0.4598, "step": 16495 }, { "epoch": 0.02, "learning_rate": 4.996683527535541e-05, "loss": 0.3943, "step": 16496 }, { "epoch": 0.02, "learning_rate": 4.9966831230661534e-05, "loss": 0.395, "step": 16497 }, { "epoch": 0.02, "learning_rate": 4.9966827185721214e-05, "loss": 0.4142, "step": 16498 }, { "epoch": 0.02, "learning_rate": 4.996682314053442e-05, "loss": 0.7986, "step": 16499 }, { "epoch": 0.02, "learning_rate": 4.996681909510118e-05, "loss": 1.1246, "step": 16500 }, { "epoch": 0.02, "eval_loss": 1.0717031955718994, "eval_runtime": 83.8078, "eval_samples_per_second": 16.526, "eval_steps_per_second": 4.14, "step": 16500 }, { "epoch": 0.02, "learning_rate": 4.9966815049421464e-05, "loss": 0.852, "step": 16501 }, { "epoch": 0.02, "learning_rate": 4.996681100349529e-05, "loss": 0.6896, "step": 16502 }, { "epoch": 0.02, "learning_rate": 4.996680695732266e-05, "loss": 0.9153, "step": 16503 }, { "epoch": 0.02, "learning_rate": 4.9966802910903564e-05, "loss": 1.1552, "step": 16504 }, { "epoch": 0.02, "learning_rate": 4.996679886423801e-05, "loss": 1.3038, "step": 16505 }, { "epoch": 0.02, "learning_rate": 4.996679481732599e-05, "loss": 1.1098, "step": 16506 }, { "epoch": 0.02, "learning_rate": 4.99667907701675e-05, "loss": 1.4375, "step": 16507 }, { "epoch": 0.02, "learning_rate": 4.996678672276257e-05, "loss": 1.2337, "step": 16508 }, { "epoch": 0.02, "learning_rate": 4.996678267511116e-05, "loss": 1.1692, "step": 16509 }, { "epoch": 0.02, "learning_rate": 4.99667786272133e-05, "loss": 0.9022, "step": 16510 }, { "epoch": 0.02, "learning_rate": 4.9966774579068965e-05, "loss": 1.5421, "step": 16511 }, { "epoch": 0.02, "learning_rate": 4.996677053067818e-05, "loss": 2.2308, "step": 16512 }, { "epoch": 0.02, "learning_rate": 4.9966766482040925e-05, "loss": 0.8242, "step": 16513 }, { "epoch": 0.02, "learning_rate": 4.996676243315721e-05, "loss": 0.8223, "step": 16514 }, { "epoch": 0.02, "learning_rate": 4.996675838402704e-05, "loss": 0.7621, "step": 16515 }, { "epoch": 0.02, "learning_rate": 4.99667543346504e-05, "loss": 0.7414, "step": 16516 }, { "epoch": 0.02, "learning_rate": 4.9966750285027306e-05, "loss": 0.8153, "step": 16517 }, { "epoch": 0.02, "learning_rate": 4.9966746235157755e-05, "loss": 1.3378, "step": 16518 }, { "epoch": 0.02, "learning_rate": 4.996674218504174e-05, "loss": 1.1553, "step": 16519 }, { "epoch": 0.02, "learning_rate": 4.9966738134679256e-05, "loss": 1.1391, "step": 16520 }, { "epoch": 0.02, "learning_rate": 4.9966734084070314e-05, "loss": 0.887, "step": 16521 }, { "epoch": 0.02, "learning_rate": 4.996673003321491e-05, "loss": 1.9101, "step": 16522 }, { "epoch": 0.02, "learning_rate": 4.996672598211305e-05, "loss": 1.0085, "step": 16523 }, { "epoch": 0.02, "learning_rate": 4.9966721930764724e-05, "loss": 1.2954, "step": 16524 }, { "epoch": 0.02, "learning_rate": 4.9966717879169943e-05, "loss": 1.31, "step": 16525 }, { "epoch": 0.02, "learning_rate": 4.996671382732869e-05, "loss": 1.0428, "step": 16526 }, { "epoch": 0.02, "learning_rate": 4.996670977524099e-05, "loss": 1.2657, "step": 16527 }, { "epoch": 0.02, "learning_rate": 4.9966705722906826e-05, "loss": 1.2665, "step": 16528 }, { "epoch": 0.02, "learning_rate": 4.9966701670326194e-05, "loss": 1.1931, "step": 16529 }, { "epoch": 0.02, "learning_rate": 4.99666976174991e-05, "loss": 1.201, "step": 16530 }, { "epoch": 0.02, "learning_rate": 4.996669356442556e-05, "loss": 1.0277, "step": 16531 }, { "epoch": 0.02, "learning_rate": 4.996668951110555e-05, "loss": 0.9173, "step": 16532 }, { "epoch": 0.02, "learning_rate": 4.996668545753907e-05, "loss": 1.1702, "step": 16533 }, { "epoch": 0.02, "learning_rate": 4.9966681403726144e-05, "loss": 1.2528, "step": 16534 }, { "epoch": 0.02, "learning_rate": 4.9966677349666755e-05, "loss": 1.4963, "step": 16535 }, { "epoch": 0.02, "learning_rate": 4.99666732953609e-05, "loss": 1.2674, "step": 16536 }, { "epoch": 0.02, "learning_rate": 4.9966669240808586e-05, "loss": 0.9046, "step": 16537 }, { "epoch": 0.02, "learning_rate": 4.996666518600981e-05, "loss": 0.8433, "step": 16538 }, { "epoch": 0.02, "learning_rate": 4.996666113096458e-05, "loss": 0.9715, "step": 16539 }, { "epoch": 0.02, "learning_rate": 4.9966657075672886e-05, "loss": 1.0459, "step": 16540 }, { "epoch": 0.02, "learning_rate": 4.996665302013473e-05, "loss": 1.1493, "step": 16541 }, { "epoch": 0.02, "learning_rate": 4.9966648964350116e-05, "loss": 1.2271, "step": 16542 }, { "epoch": 0.02, "learning_rate": 4.996664490831904e-05, "loss": 1.3862, "step": 16543 }, { "epoch": 0.02, "learning_rate": 4.99666408520415e-05, "loss": 1.4713, "step": 16544 }, { "epoch": 0.02, "learning_rate": 4.996663679551751e-05, "loss": 1.3913, "step": 16545 }, { "epoch": 0.02, "learning_rate": 4.9966632738747054e-05, "loss": 1.3553, "step": 16546 }, { "epoch": 0.02, "learning_rate": 4.996662868173014e-05, "loss": 1.0174, "step": 16547 }, { "epoch": 0.02, "learning_rate": 4.996662462446676e-05, "loss": 1.1201, "step": 16548 }, { "epoch": 0.02, "learning_rate": 4.996662056695692e-05, "loss": 1.2895, "step": 16549 }, { "epoch": 0.02, "learning_rate": 4.996661650920063e-05, "loss": 1.1312, "step": 16550 }, { "epoch": 0.02, "learning_rate": 4.996661245119787e-05, "loss": 1.0577, "step": 16551 }, { "epoch": 0.02, "learning_rate": 4.996660839294866e-05, "loss": 1.1619, "step": 16552 }, { "epoch": 0.02, "learning_rate": 4.996660433445298e-05, "loss": 1.2603, "step": 16553 }, { "epoch": 0.02, "learning_rate": 4.996660027571085e-05, "loss": 1.1966, "step": 16554 }, { "epoch": 0.02, "learning_rate": 4.9966596216722256e-05, "loss": 0.9881, "step": 16555 }, { "epoch": 0.02, "learning_rate": 4.9966592157487205e-05, "loss": 1.0043, "step": 16556 }, { "epoch": 0.02, "learning_rate": 4.996658809800569e-05, "loss": 0.9814, "step": 16557 }, { "epoch": 0.02, "learning_rate": 4.996658403827771e-05, "loss": 0.8964, "step": 16558 }, { "epoch": 0.02, "learning_rate": 4.996657997830328e-05, "loss": 1.0718, "step": 16559 }, { "epoch": 0.02, "learning_rate": 4.9966575918082385e-05, "loss": 1.0221, "step": 16560 }, { "epoch": 0.02, "learning_rate": 4.996657185761503e-05, "loss": 1.3086, "step": 16561 }, { "epoch": 0.02, "learning_rate": 4.996656779690122e-05, "loss": 1.2142, "step": 16562 }, { "epoch": 0.02, "learning_rate": 4.996656373594095e-05, "loss": 1.3019, "step": 16563 }, { "epoch": 0.02, "learning_rate": 4.996655967473422e-05, "loss": 1.0613, "step": 16564 }, { "epoch": 0.02, "learning_rate": 4.996655561328103e-05, "loss": 1.2345, "step": 16565 }, { "epoch": 0.02, "learning_rate": 4.996655155158138e-05, "loss": 0.9696, "step": 16566 }, { "epoch": 0.02, "learning_rate": 4.9966547489635275e-05, "loss": 0.7926, "step": 16567 }, { "epoch": 0.02, "learning_rate": 4.99665434274427e-05, "loss": 1.2845, "step": 16568 }, { "epoch": 0.02, "learning_rate": 4.996653936500367e-05, "loss": 0.9566, "step": 16569 }, { "epoch": 0.02, "learning_rate": 4.996653530231819e-05, "loss": 0.9945, "step": 16570 }, { "epoch": 0.02, "learning_rate": 4.9966531239386244e-05, "loss": 0.7139, "step": 16571 }, { "epoch": 0.02, "learning_rate": 4.996652717620784e-05, "loss": 0.9812, "step": 16572 }, { "epoch": 0.02, "learning_rate": 4.996652311278298e-05, "loss": 1.2039, "step": 16573 }, { "epoch": 0.02, "learning_rate": 4.996651904911166e-05, "loss": 1.1899, "step": 16574 }, { "epoch": 0.02, "learning_rate": 4.996651498519387e-05, "loss": 0.7483, "step": 16575 }, { "epoch": 0.02, "learning_rate": 4.996651092102963e-05, "loss": 1.2013, "step": 16576 }, { "epoch": 0.02, "learning_rate": 4.9966506856618934e-05, "loss": 0.8949, "step": 16577 }, { "epoch": 0.02, "learning_rate": 4.9966502791961774e-05, "loss": 1.3335, "step": 16578 }, { "epoch": 0.02, "learning_rate": 4.9966498727058155e-05, "loss": 1.0725, "step": 16579 }, { "epoch": 0.02, "learning_rate": 4.996649466190808e-05, "loss": 1.1775, "step": 16580 }, { "epoch": 0.02, "learning_rate": 4.996649059651155e-05, "loss": 1.1097, "step": 16581 }, { "epoch": 0.02, "learning_rate": 4.9966486530868554e-05, "loss": 1.0837, "step": 16582 }, { "epoch": 0.02, "learning_rate": 4.99664824649791e-05, "loss": 1.0946, "step": 16583 }, { "epoch": 0.02, "learning_rate": 4.996647839884319e-05, "loss": 1.1962, "step": 16584 }, { "epoch": 0.02, "learning_rate": 4.9966474332460824e-05, "loss": 0.7366, "step": 16585 }, { "epoch": 0.02, "learning_rate": 4.9966470265831995e-05, "loss": 0.7874, "step": 16586 }, { "epoch": 0.02, "learning_rate": 4.9966466198956706e-05, "loss": 1.0273, "step": 16587 }, { "epoch": 0.02, "learning_rate": 4.9966462131834965e-05, "loss": 0.928, "step": 16588 }, { "epoch": 0.02, "learning_rate": 4.996645806446676e-05, "loss": 1.034, "step": 16589 }, { "epoch": 0.02, "learning_rate": 4.9966453996852105e-05, "loss": 1.093, "step": 16590 }, { "epoch": 0.02, "learning_rate": 4.9966449928990985e-05, "loss": 1.3554, "step": 16591 }, { "epoch": 0.02, "learning_rate": 4.9966445860883406e-05, "loss": 0.9442, "step": 16592 }, { "epoch": 0.02, "learning_rate": 4.996644179252937e-05, "loss": 0.7757, "step": 16593 }, { "epoch": 0.02, "learning_rate": 4.996643772392887e-05, "loss": 1.1583, "step": 16594 }, { "epoch": 0.02, "learning_rate": 4.9966433655081926e-05, "loss": 1.064, "step": 16595 }, { "epoch": 0.02, "learning_rate": 4.996642958598851e-05, "loss": 1.2097, "step": 16596 }, { "epoch": 0.02, "learning_rate": 4.996642551664864e-05, "loss": 0.7976, "step": 16597 }, { "epoch": 0.02, "learning_rate": 4.9966421447062316e-05, "loss": 1.1266, "step": 16598 }, { "epoch": 0.02, "learning_rate": 4.9966417377229535e-05, "loss": 1.1103, "step": 16599 }, { "epoch": 0.02, "learning_rate": 4.9966413307150286e-05, "loss": 1.1689, "step": 16600 }, { "epoch": 0.02, "learning_rate": 4.996640923682459e-05, "loss": 1.0684, "step": 16601 }, { "epoch": 0.02, "learning_rate": 4.9966405166252425e-05, "loss": 1.1045, "step": 16602 }, { "epoch": 0.02, "learning_rate": 4.996640109543381e-05, "loss": 1.0252, "step": 16603 }, { "epoch": 0.02, "learning_rate": 4.996639702436874e-05, "loss": 0.9911, "step": 16604 }, { "epoch": 0.02, "learning_rate": 4.996639295305721e-05, "loss": 1.2316, "step": 16605 }, { "epoch": 0.02, "learning_rate": 4.996638888149922e-05, "loss": 0.8862, "step": 16606 }, { "epoch": 0.02, "learning_rate": 4.996638480969478e-05, "loss": 1.215, "step": 16607 }, { "epoch": 0.02, "learning_rate": 4.996638073764387e-05, "loss": 0.9901, "step": 16608 }, { "epoch": 0.02, "learning_rate": 4.9966376665346505e-05, "loss": 1.2669, "step": 16609 }, { "epoch": 0.02, "learning_rate": 4.996637259280269e-05, "loss": 1.2011, "step": 16610 }, { "epoch": 0.02, "learning_rate": 4.996636852001241e-05, "loss": 0.6976, "step": 16611 }, { "epoch": 0.02, "learning_rate": 4.996636444697568e-05, "loss": 0.9118, "step": 16612 }, { "epoch": 0.02, "learning_rate": 4.9966360373692485e-05, "loss": 0.9129, "step": 16613 }, { "epoch": 0.02, "learning_rate": 4.996635630016283e-05, "loss": 0.9218, "step": 16614 }, { "epoch": 0.02, "learning_rate": 4.9966352226386725e-05, "loss": 0.974, "step": 16615 }, { "epoch": 0.02, "learning_rate": 4.996634815236416e-05, "loss": 1.1337, "step": 16616 }, { "epoch": 0.02, "learning_rate": 4.996634407809514e-05, "loss": 0.8914, "step": 16617 }, { "epoch": 0.02, "learning_rate": 4.996634000357966e-05, "loss": 0.9413, "step": 16618 }, { "epoch": 0.02, "learning_rate": 4.9966335928817724e-05, "loss": 1.0699, "step": 16619 }, { "epoch": 0.02, "learning_rate": 4.9966331853809334e-05, "loss": 1.0181, "step": 16620 }, { "epoch": 0.02, "learning_rate": 4.9966327778554484e-05, "loss": 1.0375, "step": 16621 }, { "epoch": 0.02, "learning_rate": 4.9966323703053175e-05, "loss": 0.974, "step": 16622 }, { "epoch": 0.02, "learning_rate": 4.996631962730541e-05, "loss": 1.0828, "step": 16623 }, { "epoch": 0.02, "learning_rate": 4.996631555131119e-05, "loss": 1.3124, "step": 16624 }, { "epoch": 0.02, "learning_rate": 4.996631147507052e-05, "loss": 0.9741, "step": 16625 }, { "epoch": 0.02, "learning_rate": 4.9966307398583377e-05, "loss": 0.8479, "step": 16626 }, { "epoch": 0.02, "learning_rate": 4.9966303321849783e-05, "loss": 1.0124, "step": 16627 }, { "epoch": 0.02, "learning_rate": 4.996629924486974e-05, "loss": 1.3266, "step": 16628 }, { "epoch": 0.02, "learning_rate": 4.996629516764323e-05, "loss": 0.6632, "step": 16629 }, { "epoch": 0.02, "learning_rate": 4.996629109017027e-05, "loss": 0.3679, "step": 16630 }, { "epoch": 0.02, "learning_rate": 4.996628701245085e-05, "loss": 0.3498, "step": 16631 }, { "epoch": 0.02, "learning_rate": 4.996628293448497e-05, "loss": 0.3288, "step": 16632 }, { "epoch": 0.02, "learning_rate": 4.996627885627264e-05, "loss": 0.6746, "step": 16633 }, { "epoch": 0.02, "learning_rate": 4.9966274777813854e-05, "loss": 0.9577, "step": 16634 }, { "epoch": 0.02, "learning_rate": 4.9966270699108605e-05, "loss": 1.0087, "step": 16635 }, { "epoch": 0.02, "learning_rate": 4.996626662015691e-05, "loss": 1.2065, "step": 16636 }, { "epoch": 0.02, "learning_rate": 4.996626254095875e-05, "loss": 1.3431, "step": 16637 }, { "epoch": 0.02, "learning_rate": 4.9966258461514136e-05, "loss": 1.0444, "step": 16638 }, { "epoch": 0.02, "learning_rate": 4.996625438182306e-05, "loss": 1.3579, "step": 16639 }, { "epoch": 0.02, "learning_rate": 4.996625030188553e-05, "loss": 1.3195, "step": 16640 }, { "epoch": 0.02, "learning_rate": 4.996624622170155e-05, "loss": 1.3443, "step": 16641 }, { "epoch": 0.02, "learning_rate": 4.9966242141271114e-05, "loss": 1.2297, "step": 16642 }, { "epoch": 0.02, "learning_rate": 4.9966238060594216e-05, "loss": 1.1234, "step": 16643 }, { "epoch": 0.02, "learning_rate": 4.996623397967086e-05, "loss": 1.2622, "step": 16644 }, { "epoch": 0.02, "learning_rate": 4.9966229898501057e-05, "loss": 1.1222, "step": 16645 }, { "epoch": 0.02, "learning_rate": 4.996622581708479e-05, "loss": 1.1789, "step": 16646 }, { "epoch": 0.02, "learning_rate": 4.996622173542207e-05, "loss": 1.1328, "step": 16647 }, { "epoch": 0.02, "learning_rate": 4.996621765351289e-05, "loss": 1.1746, "step": 16648 }, { "epoch": 0.02, "learning_rate": 4.996621357135726e-05, "loss": 0.8202, "step": 16649 }, { "epoch": 0.02, "learning_rate": 4.996620948895517e-05, "loss": 0.868, "step": 16650 }, { "epoch": 0.02, "learning_rate": 4.9966205406306624e-05, "loss": 0.8718, "step": 16651 }, { "epoch": 0.02, "learning_rate": 4.996620132341162e-05, "loss": 1.286, "step": 16652 }, { "epoch": 0.02, "learning_rate": 4.996619724027017e-05, "loss": 1.1304, "step": 16653 }, { "epoch": 0.02, "learning_rate": 4.9966193156882256e-05, "loss": 0.8905, "step": 16654 }, { "epoch": 0.02, "learning_rate": 4.9966189073247885e-05, "loss": 0.6736, "step": 16655 }, { "epoch": 0.02, "learning_rate": 4.996618498936707e-05, "loss": 0.6963, "step": 16656 }, { "epoch": 0.02, "learning_rate": 4.996618090523979e-05, "loss": 1.2097, "step": 16657 }, { "epoch": 0.02, "learning_rate": 4.996617682086605e-05, "loss": 1.2617, "step": 16658 }, { "epoch": 0.02, "learning_rate": 4.996617273624586e-05, "loss": 0.8113, "step": 16659 }, { "epoch": 0.02, "learning_rate": 4.9966168651379214e-05, "loss": 0.402, "step": 16660 }, { "epoch": 0.02, "learning_rate": 4.9966164566266114e-05, "loss": 0.392, "step": 16661 }, { "epoch": 0.02, "learning_rate": 4.9966160480906554e-05, "loss": 0.4594, "step": 16662 }, { "epoch": 0.02, "learning_rate": 4.996615639530055e-05, "loss": 1.0284, "step": 16663 }, { "epoch": 0.02, "learning_rate": 4.9966152309448076e-05, "loss": 1.4787, "step": 16664 }, { "epoch": 0.02, "learning_rate": 4.996614822334915e-05, "loss": 0.842, "step": 16665 }, { "epoch": 0.02, "learning_rate": 4.996614413700378e-05, "loss": 0.5603, "step": 16666 }, { "epoch": 0.02, "learning_rate": 4.9966140050411944e-05, "loss": 0.983, "step": 16667 }, { "epoch": 0.02, "learning_rate": 4.9966135963573654e-05, "loss": 1.1018, "step": 16668 }, { "epoch": 0.02, "learning_rate": 4.996613187648891e-05, "loss": 0.7569, "step": 16669 }, { "epoch": 0.02, "learning_rate": 4.996612778915771e-05, "loss": 1.3882, "step": 16670 }, { "epoch": 0.02, "learning_rate": 4.996612370158006e-05, "loss": 1.4836, "step": 16671 }, { "epoch": 0.02, "learning_rate": 4.996611961375595e-05, "loss": 0.8851, "step": 16672 }, { "epoch": 0.02, "learning_rate": 4.996611552568539e-05, "loss": 0.5777, "step": 16673 }, { "epoch": 0.02, "learning_rate": 4.996611143736836e-05, "loss": 0.5935, "step": 16674 }, { "epoch": 0.02, "learning_rate": 4.996610734880489e-05, "loss": 1.2951, "step": 16675 }, { "epoch": 0.02, "learning_rate": 4.9966103259994965e-05, "loss": 1.2895, "step": 16676 }, { "epoch": 0.02, "learning_rate": 4.9966099170938574e-05, "loss": 1.1234, "step": 16677 }, { "epoch": 0.02, "learning_rate": 4.996609508163574e-05, "loss": 1.0689, "step": 16678 }, { "epoch": 0.02, "learning_rate": 4.996609099208645e-05, "loss": 1.0915, "step": 16679 }, { "epoch": 0.02, "learning_rate": 4.99660869022907e-05, "loss": 1.3714, "step": 16680 }, { "epoch": 0.02, "learning_rate": 4.9966082812248496e-05, "loss": 1.0338, "step": 16681 }, { "epoch": 0.02, "learning_rate": 4.9966078721959834e-05, "loss": 0.708, "step": 16682 }, { "epoch": 0.02, "learning_rate": 4.996607463142473e-05, "loss": 0.779, "step": 16683 }, { "epoch": 0.02, "learning_rate": 4.996607054064315e-05, "loss": 1.1155, "step": 16684 }, { "epoch": 0.02, "learning_rate": 4.9966066449615134e-05, "loss": 1.1101, "step": 16685 }, { "epoch": 0.02, "learning_rate": 4.996606235834066e-05, "loss": 1.0737, "step": 16686 }, { "epoch": 0.02, "learning_rate": 4.996605826681972e-05, "loss": 0.8443, "step": 16687 }, { "epoch": 0.02, "learning_rate": 4.9966054175052345e-05, "loss": 0.979, "step": 16688 }, { "epoch": 0.02, "learning_rate": 4.996605008303851e-05, "loss": 1.174, "step": 16689 }, { "epoch": 0.02, "learning_rate": 4.996604599077821e-05, "loss": 1.2915, "step": 16690 }, { "epoch": 0.02, "learning_rate": 4.996604189827146e-05, "loss": 0.6875, "step": 16691 }, { "epoch": 0.02, "learning_rate": 4.996603780551826e-05, "loss": 1.2094, "step": 16692 }, { "epoch": 0.02, "learning_rate": 4.996603371251861e-05, "loss": 0.9497, "step": 16693 }, { "epoch": 0.02, "learning_rate": 4.99660296192725e-05, "loss": 0.8918, "step": 16694 }, { "epoch": 0.02, "learning_rate": 4.996602552577994e-05, "loss": 0.9831, "step": 16695 }, { "epoch": 0.02, "learning_rate": 4.996602143204092e-05, "loss": 1.2135, "step": 16696 }, { "epoch": 0.02, "learning_rate": 4.9966017338055446e-05, "loss": 1.0891, "step": 16697 }, { "epoch": 0.02, "learning_rate": 4.996601324382352e-05, "loss": 0.9437, "step": 16698 }, { "epoch": 0.02, "learning_rate": 4.9966009149345136e-05, "loss": 0.9093, "step": 16699 }, { "epoch": 0.02, "learning_rate": 4.99660050546203e-05, "loss": 0.9861, "step": 16700 }, { "epoch": 0.02, "learning_rate": 4.9966000959649016e-05, "loss": 1.2344, "step": 16701 }, { "epoch": 0.02, "learning_rate": 4.996599686443127e-05, "loss": 1.0995, "step": 16702 }, { "epoch": 0.02, "learning_rate": 4.996599276896707e-05, "loss": 1.1801, "step": 16703 }, { "epoch": 0.02, "learning_rate": 4.996598867325642e-05, "loss": 1.3899, "step": 16704 }, { "epoch": 0.02, "learning_rate": 4.996598457729932e-05, "loss": 0.9331, "step": 16705 }, { "epoch": 0.02, "learning_rate": 4.996598048109576e-05, "loss": 1.2834, "step": 16706 }, { "epoch": 0.02, "learning_rate": 4.996597638464575e-05, "loss": 1.0422, "step": 16707 }, { "epoch": 0.02, "learning_rate": 4.996597228794929e-05, "loss": 1.0223, "step": 16708 }, { "epoch": 0.02, "learning_rate": 4.996596819100637e-05, "loss": 1.4528, "step": 16709 }, { "epoch": 0.02, "learning_rate": 4.9965964093817e-05, "loss": 0.9167, "step": 16710 }, { "epoch": 0.02, "learning_rate": 4.996595999638118e-05, "loss": 1.2883, "step": 16711 }, { "epoch": 0.02, "learning_rate": 4.99659558986989e-05, "loss": 1.2901, "step": 16712 }, { "epoch": 0.02, "learning_rate": 4.996595180077017e-05, "loss": 1.1457, "step": 16713 }, { "epoch": 0.02, "learning_rate": 4.996594770259498e-05, "loss": 1.4019, "step": 16714 }, { "epoch": 0.02, "learning_rate": 4.996594360417335e-05, "loss": 1.08, "step": 16715 }, { "epoch": 0.02, "learning_rate": 4.9965939505505254e-05, "loss": 1.0866, "step": 16716 }, { "epoch": 0.02, "learning_rate": 4.996593540659071e-05, "loss": 0.8027, "step": 16717 }, { "epoch": 0.02, "learning_rate": 4.9965931307429714e-05, "loss": 1.0548, "step": 16718 }, { "epoch": 0.02, "learning_rate": 4.9965927208022255e-05, "loss": 0.5563, "step": 16719 }, { "epoch": 0.02, "learning_rate": 4.996592310836835e-05, "loss": 1.0494, "step": 16720 }, { "epoch": 0.02, "learning_rate": 4.9965919008468e-05, "loss": 1.0373, "step": 16721 }, { "epoch": 0.02, "learning_rate": 4.996591490832119e-05, "loss": 1.2256, "step": 16722 }, { "epoch": 0.02, "learning_rate": 4.9965910807927926e-05, "loss": 0.939, "step": 16723 }, { "epoch": 0.02, "learning_rate": 4.996590670728821e-05, "loss": 0.9157, "step": 16724 }, { "epoch": 0.02, "learning_rate": 4.9965902606402035e-05, "loss": 1.4165, "step": 16725 }, { "epoch": 0.02, "learning_rate": 4.9965898505269414e-05, "loss": 0.8425, "step": 16726 }, { "epoch": 0.02, "learning_rate": 4.9965894403890347e-05, "loss": 1.3648, "step": 16727 }, { "epoch": 0.02, "learning_rate": 4.996589030226482e-05, "loss": 1.3553, "step": 16728 }, { "epoch": 0.02, "learning_rate": 4.9965886200392834e-05, "loss": 1.3747, "step": 16729 }, { "epoch": 0.02, "learning_rate": 4.9965882098274395e-05, "loss": 0.946, "step": 16730 }, { "epoch": 0.02, "learning_rate": 4.996587799590952e-05, "loss": 1.4941, "step": 16731 }, { "epoch": 0.02, "learning_rate": 4.996587389329817e-05, "loss": 1.3225, "step": 16732 }, { "epoch": 0.02, "learning_rate": 4.996586979044039e-05, "loss": 1.0894, "step": 16733 }, { "epoch": 0.02, "learning_rate": 4.996586568733615e-05, "loss": 1.0873, "step": 16734 }, { "epoch": 0.02, "learning_rate": 4.9965861583985445e-05, "loss": 0.9497, "step": 16735 }, { "epoch": 0.02, "learning_rate": 4.99658574803883e-05, "loss": 1.3325, "step": 16736 }, { "epoch": 0.02, "learning_rate": 4.9965853376544695e-05, "loss": 1.0551, "step": 16737 }, { "epoch": 0.02, "learning_rate": 4.996584927245465e-05, "loss": 1.0396, "step": 16738 }, { "epoch": 0.02, "learning_rate": 4.9965845168118135e-05, "loss": 1.1581, "step": 16739 }, { "epoch": 0.02, "learning_rate": 4.9965841063535176e-05, "loss": 0.9999, "step": 16740 }, { "epoch": 0.02, "learning_rate": 4.996583695870577e-05, "loss": 1.0565, "step": 16741 }, { "epoch": 0.02, "learning_rate": 4.996583285362991e-05, "loss": 1.0459, "step": 16742 }, { "epoch": 0.02, "learning_rate": 4.996582874830759e-05, "loss": 1.0718, "step": 16743 }, { "epoch": 0.02, "learning_rate": 4.996582464273882e-05, "loss": 1.1664, "step": 16744 }, { "epoch": 0.02, "learning_rate": 4.9965820536923604e-05, "loss": 1.1634, "step": 16745 }, { "epoch": 0.02, "learning_rate": 4.9965816430861936e-05, "loss": 1.6196, "step": 16746 }, { "epoch": 0.02, "learning_rate": 4.9965812324553814e-05, "loss": 1.2276, "step": 16747 }, { "epoch": 0.02, "learning_rate": 4.996580821799924e-05, "loss": 0.9878, "step": 16748 }, { "epoch": 0.02, "learning_rate": 4.9965804111198214e-05, "loss": 1.1393, "step": 16749 }, { "epoch": 0.02, "learning_rate": 4.9965800004150734e-05, "loss": 0.6872, "step": 16750 }, { "epoch": 0.02, "learning_rate": 4.99657958968568e-05, "loss": 1.0776, "step": 16751 }, { "epoch": 0.02, "learning_rate": 4.996579178931642e-05, "loss": 1.2382, "step": 16752 }, { "epoch": 0.02, "learning_rate": 4.996578768152959e-05, "loss": 1.1693, "step": 16753 }, { "epoch": 0.02, "learning_rate": 4.99657835734963e-05, "loss": 1.1698, "step": 16754 }, { "epoch": 0.02, "learning_rate": 4.996577946521656e-05, "loss": 1.1649, "step": 16755 }, { "epoch": 0.02, "learning_rate": 4.996577535669037e-05, "loss": 1.3233, "step": 16756 }, { "epoch": 0.02, "learning_rate": 4.996577124791773e-05, "loss": 1.095, "step": 16757 }, { "epoch": 0.02, "learning_rate": 4.9965767138898636e-05, "loss": 1.08, "step": 16758 }, { "epoch": 0.02, "learning_rate": 4.9965763029633096e-05, "loss": 1.0797, "step": 16759 }, { "epoch": 0.02, "learning_rate": 4.9965758920121096e-05, "loss": 1.0174, "step": 16760 }, { "epoch": 0.02, "learning_rate": 4.996575481036265e-05, "loss": 1.3959, "step": 16761 }, { "epoch": 0.02, "learning_rate": 4.996575070035775e-05, "loss": 1.0393, "step": 16762 }, { "epoch": 0.02, "learning_rate": 4.99657465901064e-05, "loss": 1.3243, "step": 16763 }, { "epoch": 0.02, "learning_rate": 4.9965742479608605e-05, "loss": 1.252, "step": 16764 }, { "epoch": 0.02, "learning_rate": 4.996573836886435e-05, "loss": 0.9156, "step": 16765 }, { "epoch": 0.02, "learning_rate": 4.996573425787364e-05, "loss": 1.3264, "step": 16766 }, { "epoch": 0.02, "learning_rate": 4.996573014663649e-05, "loss": 1.0443, "step": 16767 }, { "epoch": 0.02, "learning_rate": 4.996572603515288e-05, "loss": 1.0448, "step": 16768 }, { "epoch": 0.02, "learning_rate": 4.996572192342282e-05, "loss": 1.1402, "step": 16769 }, { "epoch": 0.02, "learning_rate": 4.996571781144632e-05, "loss": 1.2106, "step": 16770 }, { "epoch": 0.02, "learning_rate": 4.996571369922336e-05, "loss": 1.1338, "step": 16771 }, { "epoch": 0.02, "learning_rate": 4.996570958675395e-05, "loss": 1.1623, "step": 16772 }, { "epoch": 0.02, "learning_rate": 4.996570547403809e-05, "loss": 1.3325, "step": 16773 }, { "epoch": 0.02, "learning_rate": 4.996570136107578e-05, "loss": 1.0579, "step": 16774 }, { "epoch": 0.02, "learning_rate": 4.9965697247867014e-05, "loss": 1.0937, "step": 16775 }, { "epoch": 0.02, "learning_rate": 4.99656931344118e-05, "loss": 1.0413, "step": 16776 }, { "epoch": 0.02, "learning_rate": 4.996568902071014e-05, "loss": 0.9278, "step": 16777 }, { "epoch": 0.02, "learning_rate": 4.996568490676202e-05, "loss": 1.0191, "step": 16778 }, { "epoch": 0.02, "learning_rate": 4.9965680792567455e-05, "loss": 1.1559, "step": 16779 }, { "epoch": 0.02, "learning_rate": 4.9965676678126435e-05, "loss": 0.8959, "step": 16780 }, { "epoch": 0.02, "learning_rate": 4.996567256343897e-05, "loss": 1.3292, "step": 16781 }, { "epoch": 0.02, "learning_rate": 4.996566844850505e-05, "loss": 0.9531, "step": 16782 }, { "epoch": 0.02, "learning_rate": 4.996566433332468e-05, "loss": 1.1523, "step": 16783 }, { "epoch": 0.02, "learning_rate": 4.996566021789786e-05, "loss": 1.0218, "step": 16784 }, { "epoch": 0.02, "learning_rate": 4.996565610222459e-05, "loss": 1.3074, "step": 16785 }, { "epoch": 0.02, "learning_rate": 4.996565198630487e-05, "loss": 1.2885, "step": 16786 }, { "epoch": 0.02, "learning_rate": 4.99656478701387e-05, "loss": 1.1237, "step": 16787 }, { "epoch": 0.02, "learning_rate": 4.996564375372608e-05, "loss": 0.9325, "step": 16788 }, { "epoch": 0.02, "learning_rate": 4.9965639637067007e-05, "loss": 0.8641, "step": 16789 }, { "epoch": 0.02, "learning_rate": 4.9965635520161494e-05, "loss": 1.0525, "step": 16790 }, { "epoch": 0.02, "learning_rate": 4.9965631403009514e-05, "loss": 1.2073, "step": 16791 }, { "epoch": 0.02, "learning_rate": 4.9965627285611096e-05, "loss": 1.2401, "step": 16792 }, { "epoch": 0.02, "learning_rate": 4.9965623167966225e-05, "loss": 0.9488, "step": 16793 }, { "epoch": 0.02, "learning_rate": 4.99656190500749e-05, "loss": 0.9807, "step": 16794 }, { "epoch": 0.02, "learning_rate": 4.996561493193713e-05, "loss": 1.0958, "step": 16795 }, { "epoch": 0.02, "learning_rate": 4.996561081355291e-05, "loss": 1.199, "step": 16796 }, { "epoch": 0.02, "learning_rate": 4.996560669492223e-05, "loss": 1.0049, "step": 16797 }, { "epoch": 0.02, "learning_rate": 4.996560257604511e-05, "loss": 1.1125, "step": 16798 }, { "epoch": 0.02, "learning_rate": 4.996559845692154e-05, "loss": 1.0018, "step": 16799 }, { "epoch": 0.02, "learning_rate": 4.996559433755152e-05, "loss": 0.8581, "step": 16800 }, { "epoch": 0.02, "learning_rate": 4.9965590217935046e-05, "loss": 1.2083, "step": 16801 }, { "epoch": 0.02, "learning_rate": 4.996558609807213e-05, "loss": 1.1268, "step": 16802 }, { "epoch": 0.02, "learning_rate": 4.996558197796275e-05, "loss": 1.0259, "step": 16803 }, { "epoch": 0.02, "learning_rate": 4.996557785760693e-05, "loss": 1.074, "step": 16804 }, { "epoch": 0.02, "learning_rate": 4.996557373700466e-05, "loss": 0.5176, "step": 16805 }, { "epoch": 0.02, "learning_rate": 4.996556961615594e-05, "loss": 0.4992, "step": 16806 }, { "epoch": 0.02, "learning_rate": 4.996556549506077e-05, "loss": 0.4515, "step": 16807 }, { "epoch": 0.02, "learning_rate": 4.996556137371915e-05, "loss": 0.6775, "step": 16808 }, { "epoch": 0.02, "learning_rate": 4.9965557252131084e-05, "loss": 0.9663, "step": 16809 }, { "epoch": 0.02, "learning_rate": 4.9965553130296564e-05, "loss": 0.9702, "step": 16810 }, { "epoch": 0.02, "learning_rate": 4.99655490082156e-05, "loss": 1.4939, "step": 16811 }, { "epoch": 0.02, "learning_rate": 4.996554488588818e-05, "loss": 1.1516, "step": 16812 }, { "epoch": 0.02, "learning_rate": 4.996554076331431e-05, "loss": 1.2307, "step": 16813 }, { "epoch": 0.02, "learning_rate": 4.9965536640494e-05, "loss": 1.1261, "step": 16814 }, { "epoch": 0.02, "learning_rate": 4.9965532517427235e-05, "loss": 1.2522, "step": 16815 }, { "epoch": 0.02, "learning_rate": 4.996552839411401e-05, "loss": 1.0733, "step": 16816 }, { "epoch": 0.02, "learning_rate": 4.996552427055435e-05, "loss": 0.4757, "step": 16817 }, { "epoch": 0.02, "learning_rate": 4.9965520146748244e-05, "loss": 0.3949, "step": 16818 }, { "epoch": 0.02, "learning_rate": 4.996551602269568e-05, "loss": 0.3809, "step": 16819 }, { "epoch": 0.02, "learning_rate": 4.9965511898396664e-05, "loss": 0.3144, "step": 16820 }, { "epoch": 0.02, "learning_rate": 4.9965507773851205e-05, "loss": 0.4285, "step": 16821 }, { "epoch": 0.02, "learning_rate": 4.99655036490593e-05, "loss": 0.3382, "step": 16822 }, { "epoch": 0.02, "learning_rate": 4.996549952402094e-05, "loss": 0.2479, "step": 16823 }, { "epoch": 0.02, "learning_rate": 4.996549539873614e-05, "loss": 0.2332, "step": 16824 }, { "epoch": 0.02, "learning_rate": 4.9965491273204877e-05, "loss": 0.4393, "step": 16825 }, { "epoch": 0.02, "learning_rate": 4.9965487147427174e-05, "loss": 0.4781, "step": 16826 }, { "epoch": 0.02, "learning_rate": 4.9965483021403026e-05, "loss": 0.7115, "step": 16827 }, { "epoch": 0.02, "learning_rate": 4.9965478895132426e-05, "loss": 0.898, "step": 16828 }, { "epoch": 0.02, "learning_rate": 4.996547476861537e-05, "loss": 0.9597, "step": 16829 }, { "epoch": 0.02, "learning_rate": 4.996547064185187e-05, "loss": 1.1202, "step": 16830 }, { "epoch": 0.02, "learning_rate": 4.996546651484192e-05, "loss": 1.0522, "step": 16831 }, { "epoch": 0.02, "learning_rate": 4.996546238758553e-05, "loss": 1.3063, "step": 16832 }, { "epoch": 0.02, "learning_rate": 4.9965458260082686e-05, "loss": 1.0041, "step": 16833 }, { "epoch": 0.02, "learning_rate": 4.996545413233339e-05, "loss": 1.3297, "step": 16834 }, { "epoch": 0.02, "learning_rate": 4.996545000433765e-05, "loss": 1.1705, "step": 16835 }, { "epoch": 0.02, "learning_rate": 4.9965445876095465e-05, "loss": 0.9359, "step": 16836 }, { "epoch": 0.02, "learning_rate": 4.9965441747606824e-05, "loss": 1.1187, "step": 16837 }, { "epoch": 0.02, "learning_rate": 4.996543761887174e-05, "loss": 1.0203, "step": 16838 }, { "epoch": 0.02, "learning_rate": 4.9965433489890204e-05, "loss": 1.0938, "step": 16839 }, { "epoch": 0.02, "learning_rate": 4.996542936066222e-05, "loss": 1.2317, "step": 16840 }, { "epoch": 0.02, "learning_rate": 4.9965425231187786e-05, "loss": 1.1463, "step": 16841 }, { "epoch": 0.02, "learning_rate": 4.99654211014669e-05, "loss": 1.1069, "step": 16842 }, { "epoch": 0.02, "learning_rate": 4.996541697149958e-05, "loss": 0.9595, "step": 16843 }, { "epoch": 0.02, "learning_rate": 4.9965412841285795e-05, "loss": 1.0811, "step": 16844 }, { "epoch": 0.02, "learning_rate": 4.996540871082558e-05, "loss": 1.2748, "step": 16845 }, { "epoch": 0.02, "learning_rate": 4.9965404580118905e-05, "loss": 1.1745, "step": 16846 }, { "epoch": 0.02, "learning_rate": 4.996540044916579e-05, "loss": 1.1534, "step": 16847 }, { "epoch": 0.02, "learning_rate": 4.996539631796622e-05, "loss": 0.5495, "step": 16848 }, { "epoch": 0.02, "learning_rate": 4.99653921865202e-05, "loss": 0.8761, "step": 16849 }, { "epoch": 0.02, "learning_rate": 4.996538805482774e-05, "loss": 0.9127, "step": 16850 }, { "epoch": 0.02, "learning_rate": 4.996538392288883e-05, "loss": 1.0061, "step": 16851 }, { "epoch": 0.02, "learning_rate": 4.996537979070347e-05, "loss": 1.2454, "step": 16852 }, { "epoch": 0.02, "learning_rate": 4.996537565827166e-05, "loss": 1.378, "step": 16853 }, { "epoch": 0.02, "learning_rate": 4.9965371525593405e-05, "loss": 0.9582, "step": 16854 }, { "epoch": 0.02, "learning_rate": 4.99653673926687e-05, "loss": 1.0881, "step": 16855 }, { "epoch": 0.02, "learning_rate": 4.9965363259497556e-05, "loss": 1.0177, "step": 16856 }, { "epoch": 0.02, "learning_rate": 4.9965359126079955e-05, "loss": 1.1509, "step": 16857 }, { "epoch": 0.02, "learning_rate": 4.9965354992415916e-05, "loss": 0.972, "step": 16858 }, { "epoch": 0.02, "learning_rate": 4.996535085850542e-05, "loss": 1.0542, "step": 16859 }, { "epoch": 0.02, "learning_rate": 4.9965346724348485e-05, "loss": 0.17, "step": 16860 }, { "epoch": 0.02, "learning_rate": 4.9965342589945094e-05, "loss": 0.6616, "step": 16861 }, { "epoch": 0.02, "learning_rate": 4.9965338455295264e-05, "loss": 1.6308, "step": 16862 }, { "epoch": 0.02, "learning_rate": 4.996533432039898e-05, "loss": 1.3101, "step": 16863 }, { "epoch": 0.02, "learning_rate": 4.996533018525625e-05, "loss": 1.015, "step": 16864 }, { "epoch": 0.02, "learning_rate": 4.996532604986707e-05, "loss": 0.8646, "step": 16865 }, { "epoch": 0.02, "learning_rate": 4.996532191423145e-05, "loss": 0.749, "step": 16866 }, { "epoch": 0.02, "learning_rate": 4.996531777834938e-05, "loss": 0.7537, "step": 16867 }, { "epoch": 0.02, "learning_rate": 4.9965313642220865e-05, "loss": 1.1897, "step": 16868 }, { "epoch": 0.02, "learning_rate": 4.99653095058459e-05, "loss": 1.0762, "step": 16869 }, { "epoch": 0.02, "learning_rate": 4.996530536922449e-05, "loss": 1.1929, "step": 16870 }, { "epoch": 0.02, "learning_rate": 4.996530123235663e-05, "loss": 1.1974, "step": 16871 }, { "epoch": 0.02, "learning_rate": 4.996529709524232e-05, "loss": 0.9872, "step": 16872 }, { "epoch": 0.02, "learning_rate": 4.996529295788157e-05, "loss": 0.8774, "step": 16873 }, { "epoch": 0.02, "learning_rate": 4.996528882027437e-05, "loss": 0.8457, "step": 16874 }, { "epoch": 0.02, "learning_rate": 4.9965284682420724e-05, "loss": 1.0507, "step": 16875 }, { "epoch": 0.02, "learning_rate": 4.996528054432064e-05, "loss": 1.1527, "step": 16876 }, { "epoch": 0.02, "learning_rate": 4.996527640597409e-05, "loss": 1.0335, "step": 16877 }, { "epoch": 0.02, "learning_rate": 4.9965272267381106e-05, "loss": 0.7664, "step": 16878 }, { "epoch": 0.02, "learning_rate": 4.9965268128541675e-05, "loss": 0.7212, "step": 16879 }, { "epoch": 0.02, "learning_rate": 4.99652639894558e-05, "loss": 0.9155, "step": 16880 }, { "epoch": 0.02, "learning_rate": 4.996525985012347e-05, "loss": 1.0865, "step": 16881 }, { "epoch": 0.02, "learning_rate": 4.99652557105447e-05, "loss": 1.294, "step": 16882 }, { "epoch": 0.02, "learning_rate": 4.996525157071948e-05, "loss": 0.2576, "step": 16883 }, { "epoch": 0.02, "learning_rate": 4.996524743064781e-05, "loss": 0.2085, "step": 16884 }, { "epoch": 0.02, "learning_rate": 4.99652432903297e-05, "loss": 0.124, "step": 16885 }, { "epoch": 0.02, "learning_rate": 4.996523914976514e-05, "loss": 0.6439, "step": 16886 }, { "epoch": 0.02, "learning_rate": 4.996523500895414e-05, "loss": 1.0005, "step": 16887 }, { "epoch": 0.02, "learning_rate": 4.9965230867896686e-05, "loss": 0.9706, "step": 16888 }, { "epoch": 0.02, "learning_rate": 4.996522672659279e-05, "loss": 1.0865, "step": 16889 }, { "epoch": 0.02, "learning_rate": 4.9965222585042446e-05, "loss": 1.2426, "step": 16890 }, { "epoch": 0.02, "learning_rate": 4.9965218443245656e-05, "loss": 1.1496, "step": 16891 }, { "epoch": 0.02, "learning_rate": 4.996521430120242e-05, "loss": 1.3892, "step": 16892 }, { "epoch": 0.02, "learning_rate": 4.996521015891274e-05, "loss": 0.9313, "step": 16893 }, { "epoch": 0.02, "learning_rate": 4.9965206016376607e-05, "loss": 1.0959, "step": 16894 }, { "epoch": 0.02, "learning_rate": 4.9965201873594034e-05, "loss": 1.2665, "step": 16895 }, { "epoch": 0.02, "learning_rate": 4.9965197730565015e-05, "loss": 1.1766, "step": 16896 }, { "epoch": 0.02, "learning_rate": 4.996519358728955e-05, "loss": 0.8894, "step": 16897 }, { "epoch": 0.02, "learning_rate": 4.9965189443767646e-05, "loss": 1.3634, "step": 16898 }, { "epoch": 0.02, "learning_rate": 4.996518529999928e-05, "loss": 1.0216, "step": 16899 }, { "epoch": 0.02, "learning_rate": 4.996518115598448e-05, "loss": 0.8901, "step": 16900 }, { "epoch": 0.02, "learning_rate": 4.996517701172323e-05, "loss": 1.251, "step": 16901 }, { "epoch": 0.02, "learning_rate": 4.996517286721554e-05, "loss": 1.6033, "step": 16902 }, { "epoch": 0.02, "learning_rate": 4.99651687224614e-05, "loss": 1.3355, "step": 16903 }, { "epoch": 0.02, "learning_rate": 4.996516457746081e-05, "loss": 1.1677, "step": 16904 }, { "epoch": 0.02, "learning_rate": 4.996516043221378e-05, "loss": 0.9257, "step": 16905 }, { "epoch": 0.02, "learning_rate": 4.996515628672031e-05, "loss": 1.0758, "step": 16906 }, { "epoch": 0.02, "learning_rate": 4.996515214098038e-05, "loss": 1.1511, "step": 16907 }, { "epoch": 0.02, "learning_rate": 4.996514799499401e-05, "loss": 1.0877, "step": 16908 }, { "epoch": 0.02, "learning_rate": 4.99651438487612e-05, "loss": 1.1926, "step": 16909 }, { "epoch": 0.02, "learning_rate": 4.996513970228194e-05, "loss": 1.0493, "step": 16910 }, { "epoch": 0.02, "learning_rate": 4.996513555555624e-05, "loss": 1.2535, "step": 16911 }, { "epoch": 0.02, "learning_rate": 4.996513140858409e-05, "loss": 1.2452, "step": 16912 }, { "epoch": 0.02, "learning_rate": 4.9965127261365495e-05, "loss": 1.4611, "step": 16913 }, { "epoch": 0.02, "learning_rate": 4.9965123113900456e-05, "loss": 1.1301, "step": 16914 }, { "epoch": 0.02, "learning_rate": 4.996511896618897e-05, "loss": 1.1435, "step": 16915 }, { "epoch": 0.02, "learning_rate": 4.996511481823104e-05, "loss": 1.1146, "step": 16916 }, { "epoch": 0.02, "learning_rate": 4.9965110670026664e-05, "loss": 1.1103, "step": 16917 }, { "epoch": 0.02, "learning_rate": 4.996510652157584e-05, "loss": 0.9252, "step": 16918 }, { "epoch": 0.02, "learning_rate": 4.996510237287858e-05, "loss": 0.7727, "step": 16919 }, { "epoch": 0.02, "learning_rate": 4.996509822393487e-05, "loss": 0.9401, "step": 16920 }, { "epoch": 0.02, "learning_rate": 4.996509407474471e-05, "loss": 1.0591, "step": 16921 }, { "epoch": 0.02, "learning_rate": 4.996508992530812e-05, "loss": 1.1756, "step": 16922 }, { "epoch": 0.02, "learning_rate": 4.996508577562507e-05, "loss": 0.8581, "step": 16923 }, { "epoch": 0.02, "learning_rate": 4.996508162569558e-05, "loss": 1.0644, "step": 16924 }, { "epoch": 0.02, "learning_rate": 4.9965077475519646e-05, "loss": 1.1265, "step": 16925 }, { "epoch": 0.02, "learning_rate": 4.996507332509727e-05, "loss": 1.4137, "step": 16926 }, { "epoch": 0.02, "learning_rate": 4.996506917442844e-05, "loss": 1.1044, "step": 16927 }, { "epoch": 0.02, "learning_rate": 4.996506502351317e-05, "loss": 1.2725, "step": 16928 }, { "epoch": 0.02, "learning_rate": 4.996506087235146e-05, "loss": 0.8745, "step": 16929 }, { "epoch": 0.02, "learning_rate": 4.9965056720943304e-05, "loss": 0.9583, "step": 16930 }, { "epoch": 0.02, "learning_rate": 4.99650525692887e-05, "loss": 0.8965, "step": 16931 }, { "epoch": 0.02, "learning_rate": 4.996504841738765e-05, "loss": 0.8921, "step": 16932 }, { "epoch": 0.02, "learning_rate": 4.996504426524016e-05, "loss": 0.9241, "step": 16933 }, { "epoch": 0.02, "learning_rate": 4.996504011284623e-05, "loss": 1.0459, "step": 16934 }, { "epoch": 0.02, "learning_rate": 4.996503596020585e-05, "loss": 1.1032, "step": 16935 }, { "epoch": 0.02, "learning_rate": 4.9965031807319027e-05, "loss": 1.2464, "step": 16936 }, { "epoch": 0.02, "learning_rate": 4.996502765418576e-05, "loss": 0.8342, "step": 16937 }, { "epoch": 0.02, "learning_rate": 4.996502350080604e-05, "loss": 1.0984, "step": 16938 }, { "epoch": 0.02, "learning_rate": 4.9965019347179885e-05, "loss": 1.4428, "step": 16939 }, { "epoch": 0.02, "learning_rate": 4.9965015193307285e-05, "loss": 1.1296, "step": 16940 }, { "epoch": 0.02, "learning_rate": 4.9965011039188234e-05, "loss": 1.182, "step": 16941 }, { "epoch": 0.02, "learning_rate": 4.996500688482275e-05, "loss": 1.1627, "step": 16942 }, { "epoch": 0.02, "learning_rate": 4.996500273021082e-05, "loss": 1.0323, "step": 16943 }, { "epoch": 0.02, "learning_rate": 4.996499857535244e-05, "loss": 0.8979, "step": 16944 }, { "epoch": 0.02, "learning_rate": 4.996499442024762e-05, "loss": 0.943, "step": 16945 }, { "epoch": 0.02, "learning_rate": 4.9964990264896354e-05, "loss": 1.1286, "step": 16946 }, { "epoch": 0.02, "learning_rate": 4.9964986109298647e-05, "loss": 1.0072, "step": 16947 }, { "epoch": 0.02, "learning_rate": 4.9964981953454494e-05, "loss": 1.1529, "step": 16948 }, { "epoch": 0.02, "learning_rate": 4.99649777973639e-05, "loss": 1.1849, "step": 16949 }, { "epoch": 0.02, "learning_rate": 4.996497364102686e-05, "loss": 1.1461, "step": 16950 }, { "epoch": 0.02, "learning_rate": 4.9964969484443366e-05, "loss": 0.7396, "step": 16951 }, { "epoch": 0.02, "learning_rate": 4.996496532761344e-05, "loss": 1.1525, "step": 16952 }, { "epoch": 0.02, "learning_rate": 4.9964961170537074e-05, "loss": 1.3184, "step": 16953 }, { "epoch": 0.02, "learning_rate": 4.996495701321426e-05, "loss": 0.9543, "step": 16954 }, { "epoch": 0.02, "learning_rate": 4.9964952855645e-05, "loss": 1.0231, "step": 16955 }, { "epoch": 0.02, "learning_rate": 4.99649486978293e-05, "loss": 0.8666, "step": 16956 }, { "epoch": 0.02, "learning_rate": 4.996494453976716e-05, "loss": 1.1742, "step": 16957 }, { "epoch": 0.02, "learning_rate": 4.9964940381458567e-05, "loss": 1.1606, "step": 16958 }, { "epoch": 0.02, "learning_rate": 4.9964936222903535e-05, "loss": 1.1058, "step": 16959 }, { "epoch": 0.02, "learning_rate": 4.996493206410205e-05, "loss": 1.1045, "step": 16960 }, { "epoch": 0.02, "learning_rate": 4.9964927905054135e-05, "loss": 0.7528, "step": 16961 }, { "epoch": 0.02, "learning_rate": 4.996492374575977e-05, "loss": 0.9455, "step": 16962 }, { "epoch": 0.02, "learning_rate": 4.996491958621897e-05, "loss": 1.0045, "step": 16963 }, { "epoch": 0.02, "learning_rate": 4.9964915426431725e-05, "loss": 0.8128, "step": 16964 }, { "epoch": 0.02, "learning_rate": 4.996491126639803e-05, "loss": 1.019, "step": 16965 }, { "epoch": 0.02, "learning_rate": 4.99649071061179e-05, "loss": 1.0857, "step": 16966 }, { "epoch": 0.02, "learning_rate": 4.996490294559132e-05, "loss": 1.3725, "step": 16967 }, { "epoch": 0.02, "learning_rate": 4.99648987848183e-05, "loss": 1.0572, "step": 16968 }, { "epoch": 0.02, "learning_rate": 4.9964894623798834e-05, "loss": 1.2566, "step": 16969 }, { "epoch": 0.02, "learning_rate": 4.996489046253293e-05, "loss": 1.0518, "step": 16970 }, { "epoch": 0.02, "learning_rate": 4.996488630102058e-05, "loss": 0.9972, "step": 16971 }, { "epoch": 0.02, "learning_rate": 4.996488213926179e-05, "loss": 0.872, "step": 16972 }, { "epoch": 0.02, "learning_rate": 4.9964877977256555e-05, "loss": 1.3396, "step": 16973 }, { "epoch": 0.02, "learning_rate": 4.9964873815004876e-05, "loss": 1.1596, "step": 16974 }, { "epoch": 0.02, "learning_rate": 4.996486965250675e-05, "loss": 1.3249, "step": 16975 }, { "epoch": 0.02, "learning_rate": 4.996486548976219e-05, "loss": 1.1562, "step": 16976 }, { "epoch": 0.02, "learning_rate": 4.996486132677118e-05, "loss": 1.0276, "step": 16977 }, { "epoch": 0.02, "learning_rate": 4.996485716353374e-05, "loss": 0.8831, "step": 16978 }, { "epoch": 0.02, "learning_rate": 4.996485300004985e-05, "loss": 0.7286, "step": 16979 }, { "epoch": 0.02, "learning_rate": 4.996484883631951e-05, "loss": 0.8235, "step": 16980 }, { "epoch": 0.02, "learning_rate": 4.9964844672342736e-05, "loss": 1.1816, "step": 16981 }, { "epoch": 0.02, "learning_rate": 4.9964840508119516e-05, "loss": 1.514, "step": 16982 }, { "epoch": 0.02, "learning_rate": 4.996483634364986e-05, "loss": 1.2372, "step": 16983 }, { "epoch": 0.02, "learning_rate": 4.996483217893375e-05, "loss": 1.0974, "step": 16984 }, { "epoch": 0.02, "learning_rate": 4.996482801397121e-05, "loss": 0.9436, "step": 16985 }, { "epoch": 0.02, "learning_rate": 4.996482384876222e-05, "loss": 1.1224, "step": 16986 }, { "epoch": 0.02, "learning_rate": 4.996481968330679e-05, "loss": 1.2766, "step": 16987 }, { "epoch": 0.02, "learning_rate": 4.996481551760492e-05, "loss": 1.1842, "step": 16988 }, { "epoch": 0.02, "learning_rate": 4.996481135165661e-05, "loss": 0.9383, "step": 16989 }, { "epoch": 0.02, "learning_rate": 4.996480718546185e-05, "loss": 1.036, "step": 16990 }, { "epoch": 0.02, "learning_rate": 4.996480301902065e-05, "loss": 1.1436, "step": 16991 }, { "epoch": 0.02, "learning_rate": 4.996479885233301e-05, "loss": 1.1322, "step": 16992 }, { "epoch": 0.02, "learning_rate": 4.996479468539893e-05, "loss": 1.0676, "step": 16993 }, { "epoch": 0.02, "learning_rate": 4.996479051821841e-05, "loss": 1.0792, "step": 16994 }, { "epoch": 0.02, "learning_rate": 4.996478635079144e-05, "loss": 0.6835, "step": 16995 }, { "epoch": 0.02, "learning_rate": 4.996478218311803e-05, "loss": 0.9158, "step": 16996 }, { "epoch": 0.02, "learning_rate": 4.9964778015198176e-05, "loss": 1.0694, "step": 16997 }, { "epoch": 0.02, "learning_rate": 4.996477384703189e-05, "loss": 1.1936, "step": 16998 }, { "epoch": 0.02, "learning_rate": 4.9964769678619157e-05, "loss": 1.3911, "step": 16999 }, { "epoch": 0.02, "learning_rate": 4.9964765509959985e-05, "loss": 0.9993, "step": 17000 }, { "epoch": 0.02, "eval_loss": 1.0266118049621582, "eval_runtime": 85.6314, "eval_samples_per_second": 16.174, "eval_steps_per_second": 4.052, "step": 17000 }, { "epoch": 0.02, "learning_rate": 4.996476134105437e-05, "loss": 1.3432, "step": 17001 }, { "epoch": 0.02, "learning_rate": 4.996475717190231e-05, "loss": 1.0194, "step": 17002 }, { "epoch": 0.02, "learning_rate": 4.996475300250381e-05, "loss": 1.1707, "step": 17003 }, { "epoch": 0.02, "learning_rate": 4.9964748832858865e-05, "loss": 1.2612, "step": 17004 }, { "epoch": 0.02, "learning_rate": 4.9964744662967484e-05, "loss": 1.1914, "step": 17005 }, { "epoch": 0.02, "learning_rate": 4.9964740492829656e-05, "loss": 1.149, "step": 17006 }, { "epoch": 0.02, "learning_rate": 4.99647363224454e-05, "loss": 1.1162, "step": 17007 }, { "epoch": 0.02, "learning_rate": 4.9964732151814685e-05, "loss": 1.111, "step": 17008 }, { "epoch": 0.02, "learning_rate": 4.996472798093754e-05, "loss": 1.2897, "step": 17009 }, { "epoch": 0.02, "learning_rate": 4.996472380981395e-05, "loss": 1.0675, "step": 17010 }, { "epoch": 0.02, "learning_rate": 4.996471963844391e-05, "loss": 1.0948, "step": 17011 }, { "epoch": 0.02, "learning_rate": 4.9964715466827444e-05, "loss": 1.1462, "step": 17012 }, { "epoch": 0.02, "learning_rate": 4.996471129496453e-05, "loss": 1.2166, "step": 17013 }, { "epoch": 0.02, "learning_rate": 4.996470712285518e-05, "loss": 1.055, "step": 17014 }, { "epoch": 0.02, "learning_rate": 4.996470295049938e-05, "loss": 1.1979, "step": 17015 }, { "epoch": 0.02, "learning_rate": 4.996469877789714e-05, "loss": 1.0299, "step": 17016 }, { "epoch": 0.02, "learning_rate": 4.9964694605048465e-05, "loss": 1.0215, "step": 17017 }, { "epoch": 0.02, "learning_rate": 4.9964690431953354e-05, "loss": 0.7385, "step": 17018 }, { "epoch": 0.02, "learning_rate": 4.996468625861179e-05, "loss": 1.0382, "step": 17019 }, { "epoch": 0.02, "learning_rate": 4.9964682085023795e-05, "loss": 1.2014, "step": 17020 }, { "epoch": 0.02, "learning_rate": 4.996467791118935e-05, "loss": 0.9783, "step": 17021 }, { "epoch": 0.02, "learning_rate": 4.9964673737108466e-05, "loss": 0.6325, "step": 17022 }, { "epoch": 0.02, "learning_rate": 4.9964669562781146e-05, "loss": 0.8298, "step": 17023 }, { "epoch": 0.02, "learning_rate": 4.996466538820738e-05, "loss": 1.088, "step": 17024 }, { "epoch": 0.02, "learning_rate": 4.9964661213387176e-05, "loss": 0.8899, "step": 17025 }, { "epoch": 0.02, "learning_rate": 4.996465703832054e-05, "loss": 1.2099, "step": 17026 }, { "epoch": 0.02, "learning_rate": 4.996465286300745e-05, "loss": 1.2739, "step": 17027 }, { "epoch": 0.02, "learning_rate": 4.9964648687447926e-05, "loss": 1.3037, "step": 17028 }, { "epoch": 0.02, "learning_rate": 4.996464451164196e-05, "loss": 1.1223, "step": 17029 }, { "epoch": 0.02, "learning_rate": 4.996464033558955e-05, "loss": 1.3314, "step": 17030 }, { "epoch": 0.02, "learning_rate": 4.99646361592907e-05, "loss": 1.0791, "step": 17031 }, { "epoch": 0.02, "learning_rate": 4.996463198274542e-05, "loss": 0.9956, "step": 17032 }, { "epoch": 0.02, "learning_rate": 4.996462780595369e-05, "loss": 1.0603, "step": 17033 }, { "epoch": 0.02, "learning_rate": 4.996462362891552e-05, "loss": 1.3253, "step": 17034 }, { "epoch": 0.02, "learning_rate": 4.996461945163091e-05, "loss": 1.123, "step": 17035 }, { "epoch": 0.02, "learning_rate": 4.996461527409987e-05, "loss": 0.5553, "step": 17036 }, { "epoch": 0.02, "learning_rate": 4.996461109632238e-05, "loss": 0.9056, "step": 17037 }, { "epoch": 0.02, "learning_rate": 4.996460691829845e-05, "loss": 1.1559, "step": 17038 }, { "epoch": 0.02, "learning_rate": 4.996460274002808e-05, "loss": 1.3572, "step": 17039 }, { "epoch": 0.02, "learning_rate": 4.9964598561511274e-05, "loss": 1.1788, "step": 17040 }, { "epoch": 0.02, "learning_rate": 4.996459438274803e-05, "loss": 1.3355, "step": 17041 }, { "epoch": 0.02, "learning_rate": 4.996459020373834e-05, "loss": 0.7532, "step": 17042 }, { "epoch": 0.02, "learning_rate": 4.996458602448221e-05, "loss": 1.2198, "step": 17043 }, { "epoch": 0.02, "learning_rate": 4.996458184497964e-05, "loss": 0.9523, "step": 17044 }, { "epoch": 0.02, "learning_rate": 4.996457766523063e-05, "loss": 1.1045, "step": 17045 }, { "epoch": 0.02, "learning_rate": 4.996457348523518e-05, "loss": 1.1337, "step": 17046 }, { "epoch": 0.02, "learning_rate": 4.996456930499329e-05, "loss": 1.0515, "step": 17047 }, { "epoch": 0.02, "learning_rate": 4.996456512450497e-05, "loss": 1.319, "step": 17048 }, { "epoch": 0.02, "learning_rate": 4.99645609437702e-05, "loss": 1.2448, "step": 17049 }, { "epoch": 0.02, "learning_rate": 4.9964556762788994e-05, "loss": 1.2194, "step": 17050 }, { "epoch": 0.02, "learning_rate": 4.9964552581561344e-05, "loss": 1.087, "step": 17051 }, { "epoch": 0.02, "learning_rate": 4.996454840008726e-05, "loss": 1.3277, "step": 17052 }, { "epoch": 0.02, "learning_rate": 4.996454421836674e-05, "loss": 0.4406, "step": 17053 }, { "epoch": 0.02, "learning_rate": 4.996454003639977e-05, "loss": 1.022, "step": 17054 }, { "epoch": 0.02, "learning_rate": 4.9964535854186366e-05, "loss": 1.3239, "step": 17055 }, { "epoch": 0.02, "learning_rate": 4.996453167172652e-05, "loss": 1.2189, "step": 17056 }, { "epoch": 0.02, "learning_rate": 4.996452748902024e-05, "loss": 1.2144, "step": 17057 }, { "epoch": 0.02, "learning_rate": 4.996452330606752e-05, "loss": 0.9751, "step": 17058 }, { "epoch": 0.02, "learning_rate": 4.996451912286836e-05, "loss": 1.2124, "step": 17059 }, { "epoch": 0.02, "learning_rate": 4.996451493942276e-05, "loss": 0.9633, "step": 17060 }, { "epoch": 0.02, "learning_rate": 4.996451075573072e-05, "loss": 1.3018, "step": 17061 }, { "epoch": 0.02, "learning_rate": 4.9964506571792236e-05, "loss": 1.5028, "step": 17062 }, { "epoch": 0.02, "learning_rate": 4.9964502387607316e-05, "loss": 1.2864, "step": 17063 }, { "epoch": 0.02, "learning_rate": 4.996449820317596e-05, "loss": 1.2757, "step": 17064 }, { "epoch": 0.02, "learning_rate": 4.9964494018498166e-05, "loss": 0.9664, "step": 17065 }, { "epoch": 0.02, "learning_rate": 4.996448983357393e-05, "loss": 1.1671, "step": 17066 }, { "epoch": 0.02, "learning_rate": 4.996448564840326e-05, "loss": 1.2138, "step": 17067 }, { "epoch": 0.02, "learning_rate": 4.996448146298614e-05, "loss": 0.9621, "step": 17068 }, { "epoch": 0.02, "learning_rate": 4.996447727732259e-05, "loss": 0.6476, "step": 17069 }, { "epoch": 0.02, "learning_rate": 4.99644730914126e-05, "loss": 0.881, "step": 17070 }, { "epoch": 0.02, "learning_rate": 4.996446890525617e-05, "loss": 1.2651, "step": 17071 }, { "epoch": 0.02, "learning_rate": 4.99644647188533e-05, "loss": 1.1601, "step": 17072 }, { "epoch": 0.02, "learning_rate": 4.996446053220399e-05, "loss": 1.3367, "step": 17073 }, { "epoch": 0.02, "learning_rate": 4.996445634530825e-05, "loss": 1.032, "step": 17074 }, { "epoch": 0.02, "learning_rate": 4.9964452158166064e-05, "loss": 1.2032, "step": 17075 }, { "epoch": 0.02, "learning_rate": 4.996444797077745e-05, "loss": 1.0557, "step": 17076 }, { "epoch": 0.02, "learning_rate": 4.996444378314239e-05, "loss": 1.2644, "step": 17077 }, { "epoch": 0.02, "learning_rate": 4.996443959526088e-05, "loss": 1.0396, "step": 17078 }, { "epoch": 0.02, "learning_rate": 4.996443540713295e-05, "loss": 1.4458, "step": 17079 }, { "epoch": 0.02, "learning_rate": 4.996443121875857e-05, "loss": 1.0424, "step": 17080 }, { "epoch": 0.02, "learning_rate": 4.996442703013776e-05, "loss": 1.0034, "step": 17081 }, { "epoch": 0.02, "learning_rate": 4.99644228412705e-05, "loss": 1.1232, "step": 17082 }, { "epoch": 0.02, "learning_rate": 4.996441865215681e-05, "loss": 1.1754, "step": 17083 }, { "epoch": 0.02, "learning_rate": 4.996441446279668e-05, "loss": 0.9483, "step": 17084 }, { "epoch": 0.02, "learning_rate": 4.9964410273190114e-05, "loss": 0.9419, "step": 17085 }, { "epoch": 0.02, "learning_rate": 4.9964406083337114e-05, "loss": 0.9803, "step": 17086 }, { "epoch": 0.02, "learning_rate": 4.996440189323767e-05, "loss": 0.963, "step": 17087 }, { "epoch": 0.02, "learning_rate": 4.996439770289178e-05, "loss": 0.4295, "step": 17088 }, { "epoch": 0.02, "learning_rate": 4.9964393512299465e-05, "loss": 0.9823, "step": 17089 }, { "epoch": 0.02, "learning_rate": 4.99643893214607e-05, "loss": 0.6973, "step": 17090 }, { "epoch": 0.02, "learning_rate": 4.996438513037551e-05, "loss": 0.9756, "step": 17091 }, { "epoch": 0.02, "learning_rate": 4.996438093904388e-05, "loss": 1.2441, "step": 17092 }, { "epoch": 0.02, "learning_rate": 4.99643767474658e-05, "loss": 1.1462, "step": 17093 }, { "epoch": 0.02, "learning_rate": 4.996437255564129e-05, "loss": 1.0648, "step": 17094 }, { "epoch": 0.02, "learning_rate": 4.9964368363570344e-05, "loss": 1.1123, "step": 17095 }, { "epoch": 0.02, "learning_rate": 4.9964364171252966e-05, "loss": 1.2947, "step": 17096 }, { "epoch": 0.02, "learning_rate": 4.996435997868914e-05, "loss": 1.2598, "step": 17097 }, { "epoch": 0.02, "learning_rate": 4.996435578587888e-05, "loss": 1.3251, "step": 17098 }, { "epoch": 0.02, "learning_rate": 4.9964351592822176e-05, "loss": 0.9466, "step": 17099 }, { "epoch": 0.02, "learning_rate": 4.996434739951905e-05, "loss": 0.7017, "step": 17100 }, { "epoch": 0.02, "learning_rate": 4.9964343205969474e-05, "loss": 1.0579, "step": 17101 }, { "epoch": 0.02, "learning_rate": 4.996433901217347e-05, "loss": 1.1884, "step": 17102 }, { "epoch": 0.02, "learning_rate": 4.9964334818131016e-05, "loss": 0.9634, "step": 17103 }, { "epoch": 0.02, "learning_rate": 4.9964330623842125e-05, "loss": 0.8608, "step": 17104 }, { "epoch": 0.02, "learning_rate": 4.996432642930681e-05, "loss": 1.0821, "step": 17105 }, { "epoch": 0.02, "learning_rate": 4.9964322234525045e-05, "loss": 1.1591, "step": 17106 }, { "epoch": 0.02, "learning_rate": 4.996431803949685e-05, "loss": 1.1362, "step": 17107 }, { "epoch": 0.02, "learning_rate": 4.9964313844222215e-05, "loss": 1.2582, "step": 17108 }, { "epoch": 0.02, "learning_rate": 4.996430964870115e-05, "loss": 1.0, "step": 17109 }, { "epoch": 0.02, "learning_rate": 4.9964305452933636e-05, "loss": 0.9845, "step": 17110 }, { "epoch": 0.02, "learning_rate": 4.996430125691969e-05, "loss": 1.1918, "step": 17111 }, { "epoch": 0.02, "learning_rate": 4.9964297060659306e-05, "loss": 1.1198, "step": 17112 }, { "epoch": 0.02, "learning_rate": 4.996429286415248e-05, "loss": 1.141, "step": 17113 }, { "epoch": 0.02, "learning_rate": 4.996428866739923e-05, "loss": 1.5284, "step": 17114 }, { "epoch": 0.02, "learning_rate": 4.996428447039954e-05, "loss": 1.396, "step": 17115 }, { "epoch": 0.02, "learning_rate": 4.9964280273153405e-05, "loss": 1.1226, "step": 17116 }, { "epoch": 0.02, "learning_rate": 4.996427607566084e-05, "loss": 1.2073, "step": 17117 }, { "epoch": 0.02, "learning_rate": 4.996427187792183e-05, "loss": 1.0523, "step": 17118 }, { "epoch": 0.02, "learning_rate": 4.9964267679936395e-05, "loss": 0.8472, "step": 17119 }, { "epoch": 0.02, "learning_rate": 4.996426348170452e-05, "loss": 0.7644, "step": 17120 }, { "epoch": 0.02, "learning_rate": 4.99642592832262e-05, "loss": 1.1019, "step": 17121 }, { "epoch": 0.02, "learning_rate": 4.9964255084501446e-05, "loss": 1.0891, "step": 17122 }, { "epoch": 0.02, "learning_rate": 4.9964250885530265e-05, "loss": 1.1578, "step": 17123 }, { "epoch": 0.02, "learning_rate": 4.996424668631264e-05, "loss": 1.1067, "step": 17124 }, { "epoch": 0.02, "learning_rate": 4.996424248684858e-05, "loss": 1.0518, "step": 17125 }, { "epoch": 0.02, "learning_rate": 4.996423828713808e-05, "loss": 1.1473, "step": 17126 }, { "epoch": 0.02, "learning_rate": 4.996423408718114e-05, "loss": 1.0619, "step": 17127 }, { "epoch": 0.02, "learning_rate": 4.996422988697778e-05, "loss": 1.2103, "step": 17128 }, { "epoch": 0.02, "learning_rate": 4.996422568652796e-05, "loss": 0.8704, "step": 17129 }, { "epoch": 0.02, "learning_rate": 4.996422148583173e-05, "loss": 1.5559, "step": 17130 }, { "epoch": 0.02, "learning_rate": 4.996421728488905e-05, "loss": 1.4091, "step": 17131 }, { "epoch": 0.02, "learning_rate": 4.996421308369993e-05, "loss": 1.4804, "step": 17132 }, { "epoch": 0.02, "learning_rate": 4.996420888226438e-05, "loss": 1.5789, "step": 17133 }, { "epoch": 0.02, "learning_rate": 4.99642046805824e-05, "loss": 1.7038, "step": 17134 }, { "epoch": 0.02, "learning_rate": 4.9964200478653967e-05, "loss": 1.5351, "step": 17135 }, { "epoch": 0.02, "learning_rate": 4.996419627647911e-05, "loss": 1.5582, "step": 17136 }, { "epoch": 0.02, "learning_rate": 4.9964192074057815e-05, "loss": 1.9338, "step": 17137 }, { "epoch": 0.02, "learning_rate": 4.996418787139008e-05, "loss": 1.1169, "step": 17138 }, { "epoch": 0.02, "learning_rate": 4.9964183668475914e-05, "loss": 1.0119, "step": 17139 }, { "epoch": 0.02, "learning_rate": 4.996417946531531e-05, "loss": 0.8378, "step": 17140 }, { "epoch": 0.02, "learning_rate": 4.996417526190827e-05, "loss": 1.1842, "step": 17141 }, { "epoch": 0.02, "learning_rate": 4.996417105825479e-05, "loss": 0.8782, "step": 17142 }, { "epoch": 0.02, "learning_rate": 4.996416685435488e-05, "loss": 0.9028, "step": 17143 }, { "epoch": 0.02, "learning_rate": 4.996416265020853e-05, "loss": 0.3882, "step": 17144 }, { "epoch": 0.02, "learning_rate": 4.996415844581575e-05, "loss": 0.6216, "step": 17145 }, { "epoch": 0.02, "learning_rate": 4.996415424117653e-05, "loss": 1.2481, "step": 17146 }, { "epoch": 0.02, "learning_rate": 4.996415003629088e-05, "loss": 0.5825, "step": 17147 }, { "epoch": 0.02, "learning_rate": 4.996414583115879e-05, "loss": 0.6968, "step": 17148 }, { "epoch": 0.02, "learning_rate": 4.996414162578026e-05, "loss": 1.1528, "step": 17149 }, { "epoch": 0.02, "learning_rate": 4.9964137420155304e-05, "loss": 1.0865, "step": 17150 }, { "epoch": 0.02, "learning_rate": 4.996413321428391e-05, "loss": 0.8462, "step": 17151 }, { "epoch": 0.02, "learning_rate": 4.996412900816608e-05, "loss": 1.1814, "step": 17152 }, { "epoch": 0.02, "learning_rate": 4.996412480180181e-05, "loss": 1.1778, "step": 17153 }, { "epoch": 0.02, "learning_rate": 4.996412059519111e-05, "loss": 0.781, "step": 17154 }, { "epoch": 0.02, "learning_rate": 4.996411638833397e-05, "loss": 1.3372, "step": 17155 }, { "epoch": 0.02, "learning_rate": 4.99641121812304e-05, "loss": 0.9577, "step": 17156 }, { "epoch": 0.02, "learning_rate": 4.996410797388039e-05, "loss": 0.9539, "step": 17157 }, { "epoch": 0.02, "learning_rate": 4.996410376628395e-05, "loss": 1.015, "step": 17158 }, { "epoch": 0.02, "learning_rate": 4.9964099558441066e-05, "loss": 0.6379, "step": 17159 }, { "epoch": 0.02, "learning_rate": 4.996409535035176e-05, "loss": 1.1556, "step": 17160 }, { "epoch": 0.02, "learning_rate": 4.996409114201601e-05, "loss": 1.1133, "step": 17161 }, { "epoch": 0.02, "learning_rate": 4.996408693343383e-05, "loss": 0.962, "step": 17162 }, { "epoch": 0.02, "learning_rate": 4.996408272460521e-05, "loss": 1.1877, "step": 17163 }, { "epoch": 0.02, "learning_rate": 4.9964078515530155e-05, "loss": 1.0683, "step": 17164 }, { "epoch": 0.02, "learning_rate": 4.9964074306208665e-05, "loss": 0.8889, "step": 17165 }, { "epoch": 0.02, "learning_rate": 4.996407009664075e-05, "loss": 0.7492, "step": 17166 }, { "epoch": 0.02, "learning_rate": 4.996406588682638e-05, "loss": 1.2761, "step": 17167 }, { "epoch": 0.02, "learning_rate": 4.9964061676765594e-05, "loss": 0.9403, "step": 17168 }, { "epoch": 0.02, "learning_rate": 4.996405746645837e-05, "loss": 1.046, "step": 17169 }, { "epoch": 0.02, "learning_rate": 4.996405325590471e-05, "loss": 1.205, "step": 17170 }, { "epoch": 0.02, "learning_rate": 4.996404904510461e-05, "loss": 1.1947, "step": 17171 }, { "epoch": 0.02, "learning_rate": 4.9964044834058075e-05, "loss": 1.1725, "step": 17172 }, { "epoch": 0.02, "learning_rate": 4.996404062276512e-05, "loss": 1.0102, "step": 17173 }, { "epoch": 0.02, "learning_rate": 4.996403641122571e-05, "loss": 1.4921, "step": 17174 }, { "epoch": 0.02, "learning_rate": 4.996403219943988e-05, "loss": 1.1074, "step": 17175 }, { "epoch": 0.02, "learning_rate": 4.9964027987407616e-05, "loss": 1.2726, "step": 17176 }, { "epoch": 0.02, "learning_rate": 4.996402377512891e-05, "loss": 0.974, "step": 17177 }, { "epoch": 0.02, "learning_rate": 4.996401956260377e-05, "loss": 0.9307, "step": 17178 }, { "epoch": 0.02, "learning_rate": 4.99640153498322e-05, "loss": 1.0392, "step": 17179 }, { "epoch": 0.02, "learning_rate": 4.996401113681419e-05, "loss": 0.9007, "step": 17180 }, { "epoch": 0.02, "learning_rate": 4.996400692354976e-05, "loss": 0.8559, "step": 17181 }, { "epoch": 0.02, "learning_rate": 4.9964002710038883e-05, "loss": 0.5702, "step": 17182 }, { "epoch": 0.02, "learning_rate": 4.996399849628157e-05, "loss": 0.6818, "step": 17183 }, { "epoch": 0.02, "learning_rate": 4.9963994282277824e-05, "loss": 1.2182, "step": 17184 }, { "epoch": 0.02, "learning_rate": 4.996399006802765e-05, "loss": 1.3155, "step": 17185 }, { "epoch": 0.02, "learning_rate": 4.996398585353104e-05, "loss": 1.0473, "step": 17186 }, { "epoch": 0.02, "learning_rate": 4.9963981638788e-05, "loss": 1.134, "step": 17187 }, { "epoch": 0.02, "learning_rate": 4.996397742379852e-05, "loss": 1.1505, "step": 17188 }, { "epoch": 0.02, "learning_rate": 4.996397320856261e-05, "loss": 1.2495, "step": 17189 }, { "epoch": 0.02, "learning_rate": 4.996396899308026e-05, "loss": 1.2216, "step": 17190 }, { "epoch": 0.02, "learning_rate": 4.9963964777351476e-05, "loss": 1.4167, "step": 17191 }, { "epoch": 0.02, "learning_rate": 4.9963960561376265e-05, "loss": 1.1755, "step": 17192 }, { "epoch": 0.02, "learning_rate": 4.996395634515462e-05, "loss": 1.3357, "step": 17193 }, { "epoch": 0.02, "learning_rate": 4.9963952128686544e-05, "loss": 1.0604, "step": 17194 }, { "epoch": 0.02, "learning_rate": 4.996394791197202e-05, "loss": 1.2739, "step": 17195 }, { "epoch": 0.02, "learning_rate": 4.996394369501108e-05, "loss": 1.4028, "step": 17196 }, { "epoch": 0.02, "learning_rate": 4.99639394778037e-05, "loss": 1.4299, "step": 17197 }, { "epoch": 0.02, "learning_rate": 4.996393526034988e-05, "loss": 1.3966, "step": 17198 }, { "epoch": 0.02, "learning_rate": 4.9963931042649636e-05, "loss": 1.132, "step": 17199 }, { "epoch": 0.02, "learning_rate": 4.996392682470296e-05, "loss": 1.253, "step": 17200 }, { "epoch": 0.02, "learning_rate": 4.9963922606509835e-05, "loss": 1.5545, "step": 17201 }, { "epoch": 0.02, "learning_rate": 4.9963918388070286e-05, "loss": 1.3841, "step": 17202 }, { "epoch": 0.02, "learning_rate": 4.9963914169384305e-05, "loss": 0.8694, "step": 17203 }, { "epoch": 0.02, "learning_rate": 4.996390995045189e-05, "loss": 0.9877, "step": 17204 }, { "epoch": 0.02, "learning_rate": 4.996390573127304e-05, "loss": 0.9831, "step": 17205 }, { "epoch": 0.02, "learning_rate": 4.996390151184776e-05, "loss": 0.9441, "step": 17206 }, { "epoch": 0.02, "learning_rate": 4.996389729217604e-05, "loss": 1.1723, "step": 17207 }, { "epoch": 0.02, "learning_rate": 4.99638930722579e-05, "loss": 0.7799, "step": 17208 }, { "epoch": 0.02, "learning_rate": 4.9963888852093324e-05, "loss": 1.1345, "step": 17209 }, { "epoch": 0.02, "learning_rate": 4.99638846316823e-05, "loss": 1.2506, "step": 17210 }, { "epoch": 0.02, "learning_rate": 4.9963880411024855e-05, "loss": 1.2413, "step": 17211 }, { "epoch": 0.02, "learning_rate": 4.9963876190120975e-05, "loss": 0.963, "step": 17212 }, { "epoch": 0.02, "learning_rate": 4.9963871968970664e-05, "loss": 1.2428, "step": 17213 }, { "epoch": 0.02, "learning_rate": 4.996386774757392e-05, "loss": 1.1133, "step": 17214 }, { "epoch": 0.02, "learning_rate": 4.996386352593074e-05, "loss": 0.9141, "step": 17215 }, { "epoch": 0.02, "learning_rate": 4.9963859304041134e-05, "loss": 1.0773, "step": 17216 }, { "epoch": 0.02, "learning_rate": 4.9963855081905086e-05, "loss": 1.1935, "step": 17217 }, { "epoch": 0.02, "learning_rate": 4.996385085952261e-05, "loss": 0.822, "step": 17218 }, { "epoch": 0.02, "learning_rate": 4.9963846636893706e-05, "loss": 0.9676, "step": 17219 }, { "epoch": 0.02, "learning_rate": 4.996384241401836e-05, "loss": 1.18, "step": 17220 }, { "epoch": 0.02, "learning_rate": 4.996383819089659e-05, "loss": 0.933, "step": 17221 }, { "epoch": 0.02, "learning_rate": 4.9963833967528386e-05, "loss": 1.2033, "step": 17222 }, { "epoch": 0.02, "learning_rate": 4.9963829743913744e-05, "loss": 0.9014, "step": 17223 }, { "epoch": 0.02, "learning_rate": 4.9963825520052676e-05, "loss": 0.9146, "step": 17224 }, { "epoch": 0.02, "learning_rate": 4.996382129594517e-05, "loss": 1.1156, "step": 17225 }, { "epoch": 0.02, "learning_rate": 4.9963817071591236e-05, "loss": 1.0826, "step": 17226 }, { "epoch": 0.02, "learning_rate": 4.9963812846990864e-05, "loss": 0.9781, "step": 17227 }, { "epoch": 0.02, "learning_rate": 4.9963808622144066e-05, "loss": 0.9874, "step": 17228 }, { "epoch": 0.02, "learning_rate": 4.996380439705083e-05, "loss": 1.2103, "step": 17229 }, { "epoch": 0.02, "learning_rate": 4.996380017171117e-05, "loss": 1.0287, "step": 17230 }, { "epoch": 0.02, "learning_rate": 4.996379594612507e-05, "loss": 1.2416, "step": 17231 }, { "epoch": 0.02, "learning_rate": 4.996379172029254e-05, "loss": 1.2759, "step": 17232 }, { "epoch": 0.02, "learning_rate": 4.996378749421358e-05, "loss": 1.0926, "step": 17233 }, { "epoch": 0.02, "learning_rate": 4.996378326788819e-05, "loss": 1.2005, "step": 17234 }, { "epoch": 0.02, "learning_rate": 4.996377904131636e-05, "loss": 0.7497, "step": 17235 }, { "epoch": 0.02, "learning_rate": 4.996377481449811e-05, "loss": 1.0177, "step": 17236 }, { "epoch": 0.02, "learning_rate": 4.996377058743342e-05, "loss": 1.2318, "step": 17237 }, { "epoch": 0.02, "learning_rate": 4.9963766360122296e-05, "loss": 0.7603, "step": 17238 }, { "epoch": 0.02, "learning_rate": 4.996376213256474e-05, "loss": 1.1657, "step": 17239 }, { "epoch": 0.02, "learning_rate": 4.996375790476076e-05, "loss": 1.1482, "step": 17240 }, { "epoch": 0.02, "learning_rate": 4.9963753676710345e-05, "loss": 1.1145, "step": 17241 }, { "epoch": 0.02, "learning_rate": 4.99637494484135e-05, "loss": 1.2936, "step": 17242 }, { "epoch": 0.02, "learning_rate": 4.996374521987022e-05, "loss": 1.0326, "step": 17243 }, { "epoch": 0.02, "learning_rate": 4.996374099108051e-05, "loss": 1.1578, "step": 17244 }, { "epoch": 0.02, "learning_rate": 4.996373676204437e-05, "loss": 1.2512, "step": 17245 }, { "epoch": 0.02, "learning_rate": 4.9963732532761794e-05, "loss": 0.9417, "step": 17246 }, { "epoch": 0.02, "learning_rate": 4.996372830323279e-05, "loss": 1.1779, "step": 17247 }, { "epoch": 0.02, "learning_rate": 4.996372407345735e-05, "loss": 1.2913, "step": 17248 }, { "epoch": 0.02, "learning_rate": 4.9963719843435486e-05, "loss": 1.5138, "step": 17249 }, { "epoch": 0.02, "learning_rate": 4.996371561316718e-05, "loss": 0.879, "step": 17250 }, { "epoch": 0.02, "learning_rate": 4.9963711382652454e-05, "loss": 1.0488, "step": 17251 }, { "epoch": 0.02, "learning_rate": 4.996370715189129e-05, "loss": 0.9471, "step": 17252 }, { "epoch": 0.02, "learning_rate": 4.99637029208837e-05, "loss": 1.1719, "step": 17253 }, { "epoch": 0.02, "learning_rate": 4.996369868962968e-05, "loss": 1.2307, "step": 17254 }, { "epoch": 0.02, "learning_rate": 4.9963694458129216e-05, "loss": 0.8388, "step": 17255 }, { "epoch": 0.02, "learning_rate": 4.996369022638233e-05, "loss": 1.1033, "step": 17256 }, { "epoch": 0.02, "learning_rate": 4.996368599438902e-05, "loss": 0.6805, "step": 17257 }, { "epoch": 0.02, "learning_rate": 4.9963681762149274e-05, "loss": 1.232, "step": 17258 }, { "epoch": 0.02, "learning_rate": 4.9963677529663086e-05, "loss": 1.1412, "step": 17259 }, { "epoch": 0.02, "learning_rate": 4.996367329693048e-05, "loss": 1.1539, "step": 17260 }, { "epoch": 0.02, "learning_rate": 4.9963669063951434e-05, "loss": 1.0914, "step": 17261 }, { "epoch": 0.02, "learning_rate": 4.996366483072596e-05, "loss": 1.2681, "step": 17262 }, { "epoch": 0.02, "learning_rate": 4.9963660597254066e-05, "loss": 1.0189, "step": 17263 }, { "epoch": 0.02, "learning_rate": 4.996365636353573e-05, "loss": 1.4464, "step": 17264 }, { "epoch": 0.02, "learning_rate": 4.996365212957097e-05, "loss": 1.2144, "step": 17265 }, { "epoch": 0.02, "learning_rate": 4.9963647895359774e-05, "loss": 1.2996, "step": 17266 }, { "epoch": 0.02, "learning_rate": 4.996364366090215e-05, "loss": 1.2995, "step": 17267 }, { "epoch": 0.02, "learning_rate": 4.9963639426198096e-05, "loss": 1.2946, "step": 17268 }, { "epoch": 0.02, "learning_rate": 4.9963635191247605e-05, "loss": 1.4179, "step": 17269 }, { "epoch": 0.02, "learning_rate": 4.996363095605069e-05, "loss": 1.0535, "step": 17270 }, { "epoch": 0.02, "learning_rate": 4.996362672060734e-05, "loss": 1.051, "step": 17271 }, { "epoch": 0.02, "learning_rate": 4.996362248491756e-05, "loss": 1.2165, "step": 17272 }, { "epoch": 0.02, "learning_rate": 4.996361824898136e-05, "loss": 1.3238, "step": 17273 }, { "epoch": 0.02, "learning_rate": 4.996361401279872e-05, "loss": 0.9395, "step": 17274 }, { "epoch": 0.02, "learning_rate": 4.9963609776369646e-05, "loss": 1.1245, "step": 17275 }, { "epoch": 0.02, "learning_rate": 4.996360553969415e-05, "loss": 1.1777, "step": 17276 }, { "epoch": 0.02, "learning_rate": 4.9963601302772225e-05, "loss": 0.9452, "step": 17277 }, { "epoch": 0.02, "learning_rate": 4.996359706560386e-05, "loss": 1.0391, "step": 17278 }, { "epoch": 0.02, "learning_rate": 4.996359282818907e-05, "loss": 0.9944, "step": 17279 }, { "epoch": 0.02, "learning_rate": 4.9963588590527855e-05, "loss": 1.0862, "step": 17280 }, { "epoch": 0.02, "learning_rate": 4.99635843526202e-05, "loss": 1.0277, "step": 17281 }, { "epoch": 0.02, "learning_rate": 4.9963580114466124e-05, "loss": 0.9585, "step": 17282 }, { "epoch": 0.02, "learning_rate": 4.9963575876065614e-05, "loss": 1.0803, "step": 17283 }, { "epoch": 0.02, "learning_rate": 4.996357163741867e-05, "loss": 0.9859, "step": 17284 }, { "epoch": 0.02, "learning_rate": 4.996356739852531e-05, "loss": 1.0695, "step": 17285 }, { "epoch": 0.02, "learning_rate": 4.996356315938551e-05, "loss": 0.7614, "step": 17286 }, { "epoch": 0.02, "learning_rate": 4.9963558919999275e-05, "loss": 0.9558, "step": 17287 }, { "epoch": 0.02, "learning_rate": 4.996355468036662e-05, "loss": 0.9884, "step": 17288 }, { "epoch": 0.02, "learning_rate": 4.9963550440487525e-05, "loss": 1.2689, "step": 17289 }, { "epoch": 0.02, "learning_rate": 4.996354620036201e-05, "loss": 0.9861, "step": 17290 }, { "epoch": 0.02, "learning_rate": 4.9963541959990065e-05, "loss": 1.0806, "step": 17291 }, { "epoch": 0.02, "learning_rate": 4.996353771937169e-05, "loss": 0.5441, "step": 17292 }, { "epoch": 0.02, "learning_rate": 4.996353347850688e-05, "loss": 0.146, "step": 17293 }, { "epoch": 0.02, "learning_rate": 4.9963529237395644e-05, "loss": 0.1402, "step": 17294 }, { "epoch": 0.02, "learning_rate": 4.996352499603798e-05, "loss": 0.9284, "step": 17295 }, { "epoch": 0.02, "learning_rate": 4.996352075443389e-05, "loss": 1.1727, "step": 17296 }, { "epoch": 0.02, "learning_rate": 4.9963516512583364e-05, "loss": 0.6947, "step": 17297 }, { "epoch": 0.02, "learning_rate": 4.9963512270486416e-05, "loss": 0.7262, "step": 17298 }, { "epoch": 0.02, "learning_rate": 4.996350802814303e-05, "loss": 0.4485, "step": 17299 }, { "epoch": 0.02, "learning_rate": 4.9963503785553214e-05, "loss": 0.2741, "step": 17300 }, { "epoch": 0.02, "learning_rate": 4.9963499542716975e-05, "loss": 0.4738, "step": 17301 }, { "epoch": 0.02, "learning_rate": 4.99634952996343e-05, "loss": 0.7866, "step": 17302 }, { "epoch": 0.02, "learning_rate": 4.99634910563052e-05, "loss": 0.7192, "step": 17303 }, { "epoch": 0.02, "learning_rate": 4.996348681272968e-05, "loss": 0.6647, "step": 17304 }, { "epoch": 0.02, "learning_rate": 4.9963482568907715e-05, "loss": 0.7507, "step": 17305 }, { "epoch": 0.02, "learning_rate": 4.996347832483933e-05, "loss": 1.1764, "step": 17306 }, { "epoch": 0.02, "learning_rate": 4.996347408052452e-05, "loss": 0.9444, "step": 17307 }, { "epoch": 0.02, "learning_rate": 4.9963469835963276e-05, "loss": 1.1251, "step": 17308 }, { "epoch": 0.02, "learning_rate": 4.99634655911556e-05, "loss": 1.3977, "step": 17309 }, { "epoch": 0.02, "learning_rate": 4.9963461346101494e-05, "loss": 1.1988, "step": 17310 }, { "epoch": 0.02, "learning_rate": 4.9963457100800965e-05, "loss": 1.2108, "step": 17311 }, { "epoch": 0.02, "learning_rate": 4.9963452855254004e-05, "loss": 1.4584, "step": 17312 }, { "epoch": 0.02, "learning_rate": 4.996344860946062e-05, "loss": 0.8228, "step": 17313 }, { "epoch": 0.02, "learning_rate": 4.99634443634208e-05, "loss": 0.9375, "step": 17314 }, { "epoch": 0.02, "learning_rate": 4.996344011713455e-05, "loss": 1.0138, "step": 17315 }, { "epoch": 0.02, "learning_rate": 4.996343587060188e-05, "loss": 0.8541, "step": 17316 }, { "epoch": 0.02, "learning_rate": 4.996343162382278e-05, "loss": 1.1414, "step": 17317 }, { "epoch": 0.02, "learning_rate": 4.9963427376797244e-05, "loss": 0.9219, "step": 17318 }, { "epoch": 0.02, "learning_rate": 4.996342312952529e-05, "loss": 1.2181, "step": 17319 }, { "epoch": 0.02, "learning_rate": 4.99634188820069e-05, "loss": 0.8825, "step": 17320 }, { "epoch": 0.02, "learning_rate": 4.9963414634242084e-05, "loss": 0.9772, "step": 17321 }, { "epoch": 0.02, "learning_rate": 4.996341038623084e-05, "loss": 1.1053, "step": 17322 }, { "epoch": 0.02, "learning_rate": 4.996340613797317e-05, "loss": 1.2224, "step": 17323 }, { "epoch": 0.02, "learning_rate": 4.996340188946907e-05, "loss": 0.9532, "step": 17324 }, { "epoch": 0.02, "learning_rate": 4.996339764071853e-05, "loss": 0.891, "step": 17325 }, { "epoch": 0.02, "learning_rate": 4.996339339172158e-05, "loss": 0.9064, "step": 17326 }, { "epoch": 0.02, "learning_rate": 4.996338914247819e-05, "loss": 1.7453, "step": 17327 }, { "epoch": 0.02, "learning_rate": 4.996338489298839e-05, "loss": 1.1493, "step": 17328 }, { "epoch": 0.02, "learning_rate": 4.9963380643252137e-05, "loss": 0.7362, "step": 17329 }, { "epoch": 0.02, "learning_rate": 4.996337639326947e-05, "loss": 1.2371, "step": 17330 }, { "epoch": 0.02, "learning_rate": 4.996337214304038e-05, "loss": 1.1131, "step": 17331 }, { "epoch": 0.02, "learning_rate": 4.996336789256485e-05, "loss": 0.9624, "step": 17332 }, { "epoch": 0.02, "learning_rate": 4.99633636418429e-05, "loss": 0.9961, "step": 17333 }, { "epoch": 0.02, "learning_rate": 4.996335939087451e-05, "loss": 0.8062, "step": 17334 }, { "epoch": 0.02, "learning_rate": 4.996335513965971e-05, "loss": 0.9296, "step": 17335 }, { "epoch": 0.02, "learning_rate": 4.996335088819847e-05, "loss": 0.9914, "step": 17336 }, { "epoch": 0.02, "learning_rate": 4.99633466364908e-05, "loss": 1.1337, "step": 17337 }, { "epoch": 0.02, "learning_rate": 4.9963342384536713e-05, "loss": 1.2149, "step": 17338 }, { "epoch": 0.02, "learning_rate": 4.996333813233619e-05, "loss": 1.1082, "step": 17339 }, { "epoch": 0.02, "learning_rate": 4.996333387988925e-05, "loss": 1.0661, "step": 17340 }, { "epoch": 0.02, "learning_rate": 4.996332962719587e-05, "loss": 0.9451, "step": 17341 }, { "epoch": 0.02, "learning_rate": 4.996332537425607e-05, "loss": 1.2898, "step": 17342 }, { "epoch": 0.02, "learning_rate": 4.996332112106984e-05, "loss": 1.193, "step": 17343 }, { "epoch": 0.02, "learning_rate": 4.996331686763718e-05, "loss": 1.0381, "step": 17344 }, { "epoch": 0.02, "learning_rate": 4.99633126139581e-05, "loss": 1.0595, "step": 17345 }, { "epoch": 0.02, "learning_rate": 4.996330836003259e-05, "loss": 0.9709, "step": 17346 }, { "epoch": 0.02, "learning_rate": 4.9963304105860654e-05, "loss": 1.5223, "step": 17347 }, { "epoch": 0.02, "learning_rate": 4.996329985144228e-05, "loss": 1.2276, "step": 17348 }, { "epoch": 0.02, "learning_rate": 4.996329559677749e-05, "loss": 1.1893, "step": 17349 }, { "epoch": 0.02, "learning_rate": 4.996329134186627e-05, "loss": 0.9225, "step": 17350 }, { "epoch": 0.02, "learning_rate": 4.996328708670862e-05, "loss": 1.0711, "step": 17351 }, { "epoch": 0.02, "learning_rate": 4.996328283130455e-05, "loss": 0.989, "step": 17352 }, { "epoch": 0.02, "learning_rate": 4.996327857565405e-05, "loss": 1.1306, "step": 17353 }, { "epoch": 0.02, "learning_rate": 4.996327431975712e-05, "loss": 1.1915, "step": 17354 }, { "epoch": 0.02, "learning_rate": 4.996327006361377e-05, "loss": 1.1522, "step": 17355 }, { "epoch": 0.02, "learning_rate": 4.996326580722398e-05, "loss": 0.9981, "step": 17356 }, { "epoch": 0.02, "learning_rate": 4.996326155058777e-05, "loss": 1.1472, "step": 17357 }, { "epoch": 0.02, "learning_rate": 4.9963257293705136e-05, "loss": 1.1636, "step": 17358 }, { "epoch": 0.02, "learning_rate": 4.9963253036576077e-05, "loss": 0.9603, "step": 17359 }, { "epoch": 0.02, "learning_rate": 4.996324877920059e-05, "loss": 0.9564, "step": 17360 }, { "epoch": 0.02, "learning_rate": 4.996324452157867e-05, "loss": 1.0739, "step": 17361 }, { "epoch": 0.02, "learning_rate": 4.996324026371033e-05, "loss": 0.8183, "step": 17362 }, { "epoch": 0.02, "learning_rate": 4.996323600559556e-05, "loss": 1.1497, "step": 17363 }, { "epoch": 0.02, "learning_rate": 4.996323174723436e-05, "loss": 0.8956, "step": 17364 }, { "epoch": 0.02, "learning_rate": 4.996322748862674e-05, "loss": 0.8879, "step": 17365 }, { "epoch": 0.02, "learning_rate": 4.9963223229772696e-05, "loss": 0.7887, "step": 17366 }, { "epoch": 0.02, "learning_rate": 4.996321897067222e-05, "loss": 1.0428, "step": 17367 }, { "epoch": 0.02, "learning_rate": 4.9963214711325314e-05, "loss": 1.2277, "step": 17368 }, { "epoch": 0.02, "learning_rate": 4.996321045173199e-05, "loss": 1.1084, "step": 17369 }, { "epoch": 0.02, "learning_rate": 4.996320619189223e-05, "loss": 0.7871, "step": 17370 }, { "epoch": 0.02, "learning_rate": 4.996320193180606e-05, "loss": 1.0237, "step": 17371 }, { "epoch": 0.02, "learning_rate": 4.996319767147345e-05, "loss": 0.7141, "step": 17372 }, { "epoch": 0.02, "learning_rate": 4.996319341089441e-05, "loss": 0.8453, "step": 17373 }, { "epoch": 0.02, "learning_rate": 4.9963189150068956e-05, "loss": 1.4282, "step": 17374 }, { "epoch": 0.02, "learning_rate": 4.996318488899707e-05, "loss": 1.3896, "step": 17375 }, { "epoch": 0.02, "learning_rate": 4.9963180627678765e-05, "loss": 0.9688, "step": 17376 }, { "epoch": 0.02, "learning_rate": 4.9963176366114024e-05, "loss": 1.2528, "step": 17377 }, { "epoch": 0.02, "learning_rate": 4.996317210430286e-05, "loss": 1.1309, "step": 17378 }, { "epoch": 0.02, "learning_rate": 4.996316784224527e-05, "loss": 1.1008, "step": 17379 }, { "epoch": 0.02, "learning_rate": 4.9963163579941255e-05, "loss": 1.172, "step": 17380 }, { "epoch": 0.02, "learning_rate": 4.996315931739082e-05, "loss": 1.0076, "step": 17381 }, { "epoch": 0.02, "learning_rate": 4.996315505459395e-05, "loss": 1.0946, "step": 17382 }, { "epoch": 0.02, "learning_rate": 4.996315079155066e-05, "loss": 1.2352, "step": 17383 }, { "epoch": 0.02, "learning_rate": 4.996314652826094e-05, "loss": 1.1527, "step": 17384 }, { "epoch": 0.02, "learning_rate": 4.99631422647248e-05, "loss": 1.0632, "step": 17385 }, { "epoch": 0.02, "learning_rate": 4.9963138000942226e-05, "loss": 1.0584, "step": 17386 }, { "epoch": 0.02, "learning_rate": 4.9963133736913236e-05, "loss": 0.9125, "step": 17387 }, { "epoch": 0.02, "learning_rate": 4.9963129472637814e-05, "loss": 0.8828, "step": 17388 }, { "epoch": 0.02, "learning_rate": 4.9963125208115966e-05, "loss": 1.3073, "step": 17389 }, { "epoch": 0.02, "learning_rate": 4.99631209433477e-05, "loss": 1.1409, "step": 17390 }, { "epoch": 0.02, "learning_rate": 4.9963116678333e-05, "loss": 1.2063, "step": 17391 }, { "epoch": 0.02, "learning_rate": 4.9963112413071875e-05, "loss": 1.0319, "step": 17392 }, { "epoch": 0.02, "learning_rate": 4.996310814756433e-05, "loss": 1.2216, "step": 17393 }, { "epoch": 0.02, "learning_rate": 4.9963103881810356e-05, "loss": 1.2572, "step": 17394 }, { "epoch": 0.02, "learning_rate": 4.996309961580996e-05, "loss": 1.2577, "step": 17395 }, { "epoch": 0.02, "learning_rate": 4.996309534956314e-05, "loss": 1.3898, "step": 17396 }, { "epoch": 0.02, "learning_rate": 4.996309108306989e-05, "loss": 1.0667, "step": 17397 }, { "epoch": 0.02, "learning_rate": 4.9963086816330217e-05, "loss": 1.2333, "step": 17398 }, { "epoch": 0.02, "learning_rate": 4.996308254934412e-05, "loss": 1.1824, "step": 17399 }, { "epoch": 0.02, "learning_rate": 4.99630782821116e-05, "loss": 1.2248, "step": 17400 }, { "epoch": 0.02, "learning_rate": 4.996307401463265e-05, "loss": 0.8983, "step": 17401 }, { "epoch": 0.02, "learning_rate": 4.9963069746907274e-05, "loss": 0.6246, "step": 17402 }, { "epoch": 0.02, "learning_rate": 4.9963065478935475e-05, "loss": 0.5794, "step": 17403 }, { "epoch": 0.02, "learning_rate": 4.9963061210717256e-05, "loss": 0.6263, "step": 17404 }, { "epoch": 0.02, "learning_rate": 4.9963056942252606e-05, "loss": 1.0083, "step": 17405 }, { "epoch": 0.02, "learning_rate": 4.9963052673541536e-05, "loss": 0.9281, "step": 17406 }, { "epoch": 0.02, "learning_rate": 4.9963048404584034e-05, "loss": 1.0358, "step": 17407 }, { "epoch": 0.02, "learning_rate": 4.9963044135380113e-05, "loss": 1.281, "step": 17408 }, { "epoch": 0.02, "learning_rate": 4.996303986592977e-05, "loss": 1.1101, "step": 17409 }, { "epoch": 0.02, "learning_rate": 4.9963035596233e-05, "loss": 1.0381, "step": 17410 }, { "epoch": 0.02, "learning_rate": 4.99630313262898e-05, "loss": 0.8644, "step": 17411 }, { "epoch": 0.02, "learning_rate": 4.9963027056100174e-05, "loss": 1.4455, "step": 17412 }, { "epoch": 0.02, "learning_rate": 4.996302278566414e-05, "loss": 1.4179, "step": 17413 }, { "epoch": 0.02, "learning_rate": 4.9963018514981665e-05, "loss": 1.4093, "step": 17414 }, { "epoch": 0.02, "learning_rate": 4.996301424405278e-05, "loss": 1.1314, "step": 17415 }, { "epoch": 0.02, "learning_rate": 4.996300997287745e-05, "loss": 1.0532, "step": 17416 }, { "epoch": 0.02, "learning_rate": 4.9963005701455715e-05, "loss": 1.3066, "step": 17417 }, { "epoch": 0.02, "learning_rate": 4.9963001429787545e-05, "loss": 1.1634, "step": 17418 }, { "epoch": 0.02, "learning_rate": 4.9962997157872956e-05, "loss": 1.0959, "step": 17419 }, { "epoch": 0.02, "learning_rate": 4.9962992885711935e-05, "loss": 1.076, "step": 17420 }, { "epoch": 0.02, "learning_rate": 4.9962988613304495e-05, "loss": 1.2153, "step": 17421 }, { "epoch": 0.02, "learning_rate": 4.9962984340650636e-05, "loss": 1.1288, "step": 17422 }, { "epoch": 0.02, "learning_rate": 4.9962980067750345e-05, "loss": 0.9758, "step": 17423 }, { "epoch": 0.02, "learning_rate": 4.996297579460364e-05, "loss": 1.1316, "step": 17424 }, { "epoch": 0.02, "learning_rate": 4.99629715212105e-05, "loss": 1.2153, "step": 17425 }, { "epoch": 0.02, "learning_rate": 4.9962967247570945e-05, "loss": 1.189, "step": 17426 }, { "epoch": 0.02, "learning_rate": 4.996296297368496e-05, "loss": 1.1286, "step": 17427 }, { "epoch": 0.02, "learning_rate": 4.996295869955255e-05, "loss": 1.1278, "step": 17428 }, { "epoch": 0.02, "learning_rate": 4.996295442517373e-05, "loss": 1.1855, "step": 17429 }, { "epoch": 0.02, "learning_rate": 4.9962950150548464e-05, "loss": 0.951, "step": 17430 }, { "epoch": 0.02, "learning_rate": 4.9962945875676795e-05, "loss": 1.014, "step": 17431 }, { "epoch": 0.02, "learning_rate": 4.996294160055869e-05, "loss": 1.0577, "step": 17432 }, { "epoch": 0.02, "learning_rate": 4.996293732519416e-05, "loss": 1.3129, "step": 17433 }, { "epoch": 0.02, "learning_rate": 4.996293304958321e-05, "loss": 1.0192, "step": 17434 }, { "epoch": 0.02, "learning_rate": 4.9962928773725836e-05, "loss": 1.1545, "step": 17435 }, { "epoch": 0.02, "learning_rate": 4.9962924497622046e-05, "loss": 0.7061, "step": 17436 }, { "epoch": 0.02, "learning_rate": 4.996292022127182e-05, "loss": 1.3276, "step": 17437 }, { "epoch": 0.02, "learning_rate": 4.996291594467518e-05, "loss": 1.2633, "step": 17438 }, { "epoch": 0.02, "learning_rate": 4.9962911667832115e-05, "loss": 1.3611, "step": 17439 }, { "epoch": 0.02, "learning_rate": 4.996290739074262e-05, "loss": 1.0487, "step": 17440 }, { "epoch": 0.02, "learning_rate": 4.996290311340671e-05, "loss": 1.0328, "step": 17441 }, { "epoch": 0.02, "learning_rate": 4.9962898835824374e-05, "loss": 0.5902, "step": 17442 }, { "epoch": 0.02, "learning_rate": 4.996289455799561e-05, "loss": 1.2433, "step": 17443 }, { "epoch": 0.02, "learning_rate": 4.996289027992043e-05, "loss": 1.1824, "step": 17444 }, { "epoch": 0.02, "learning_rate": 4.996288600159882e-05, "loss": 1.2213, "step": 17445 }, { "epoch": 0.02, "learning_rate": 4.99628817230308e-05, "loss": 1.281, "step": 17446 }, { "epoch": 0.02, "learning_rate": 4.996287744421634e-05, "loss": 0.8739, "step": 17447 }, { "epoch": 0.02, "learning_rate": 4.996287316515547e-05, "loss": 1.1386, "step": 17448 }, { "epoch": 0.02, "learning_rate": 4.9962868885848166e-05, "loss": 1.3728, "step": 17449 }, { "epoch": 0.02, "learning_rate": 4.9962864606294444e-05, "loss": 1.2323, "step": 17450 }, { "epoch": 0.02, "learning_rate": 4.9962860326494304e-05, "loss": 1.1179, "step": 17451 }, { "epoch": 0.02, "learning_rate": 4.996285604644774e-05, "loss": 0.8826, "step": 17452 }, { "epoch": 0.02, "learning_rate": 4.996285176615475e-05, "loss": 1.0244, "step": 17453 }, { "epoch": 0.02, "learning_rate": 4.9962847485615336e-05, "loss": 0.6633, "step": 17454 }, { "epoch": 0.02, "learning_rate": 4.99628432048295e-05, "loss": 0.6568, "step": 17455 }, { "epoch": 0.02, "learning_rate": 4.996283892379724e-05, "loss": 0.7697, "step": 17456 }, { "epoch": 0.02, "learning_rate": 4.9962834642518565e-05, "loss": 1.0342, "step": 17457 }, { "epoch": 0.02, "learning_rate": 4.996283036099346e-05, "loss": 1.1591, "step": 17458 }, { "epoch": 0.02, "learning_rate": 4.9962826079221934e-05, "loss": 1.1062, "step": 17459 }, { "epoch": 0.02, "learning_rate": 4.9962821797203984e-05, "loss": 1.3233, "step": 17460 }, { "epoch": 0.02, "learning_rate": 4.9962817514939615e-05, "loss": 1.4055, "step": 17461 }, { "epoch": 0.02, "learning_rate": 4.996281323242882e-05, "loss": 0.9421, "step": 17462 }, { "epoch": 0.02, "learning_rate": 4.99628089496716e-05, "loss": 0.5923, "step": 17463 }, { "epoch": 0.02, "learning_rate": 4.9962804666667974e-05, "loss": 1.2461, "step": 17464 }, { "epoch": 0.02, "learning_rate": 4.996280038341791e-05, "loss": 0.9632, "step": 17465 }, { "epoch": 0.02, "learning_rate": 4.9962796099921425e-05, "loss": 1.1671, "step": 17466 }, { "epoch": 0.02, "learning_rate": 4.996279181617852e-05, "loss": 1.3371, "step": 17467 }, { "epoch": 0.02, "learning_rate": 4.9962787532189195e-05, "loss": 1.2114, "step": 17468 }, { "epoch": 0.02, "learning_rate": 4.996278324795345e-05, "loss": 1.2673, "step": 17469 }, { "epoch": 0.02, "learning_rate": 4.996277896347128e-05, "loss": 1.2273, "step": 17470 }, { "epoch": 0.02, "learning_rate": 4.9962774678742684e-05, "loss": 1.0447, "step": 17471 }, { "epoch": 0.02, "learning_rate": 4.996277039376767e-05, "loss": 1.0823, "step": 17472 }, { "epoch": 0.02, "learning_rate": 4.996276610854623e-05, "loss": 1.1767, "step": 17473 }, { "epoch": 0.02, "learning_rate": 4.9962761823078377e-05, "loss": 1.2755, "step": 17474 }, { "epoch": 0.02, "learning_rate": 4.996275753736409e-05, "loss": 0.9984, "step": 17475 }, { "epoch": 0.02, "learning_rate": 4.996275325140339e-05, "loss": 1.0276, "step": 17476 }, { "epoch": 0.02, "learning_rate": 4.996274896519627e-05, "loss": 1.3435, "step": 17477 }, { "epoch": 0.02, "learning_rate": 4.996274467874272e-05, "loss": 1.2194, "step": 17478 }, { "epoch": 0.02, "learning_rate": 4.996274039204275e-05, "loss": 1.3562, "step": 17479 }, { "epoch": 0.02, "learning_rate": 4.9962736105096365e-05, "loss": 1.0642, "step": 17480 }, { "epoch": 0.02, "learning_rate": 4.996273181790355e-05, "loss": 1.0313, "step": 17481 }, { "epoch": 0.02, "learning_rate": 4.996272753046432e-05, "loss": 0.9993, "step": 17482 }, { "epoch": 0.02, "learning_rate": 4.996272324277866e-05, "loss": 1.1906, "step": 17483 }, { "epoch": 0.02, "learning_rate": 4.996271895484659e-05, "loss": 1.0643, "step": 17484 }, { "epoch": 0.02, "learning_rate": 4.9962714666668096e-05, "loss": 1.1569, "step": 17485 }, { "epoch": 0.02, "learning_rate": 4.9962710378243175e-05, "loss": 1.4357, "step": 17486 }, { "epoch": 0.02, "learning_rate": 4.9962706089571835e-05, "loss": 1.1304, "step": 17487 }, { "epoch": 0.02, "learning_rate": 4.9962701800654076e-05, "loss": 1.5991, "step": 17488 }, { "epoch": 0.02, "learning_rate": 4.99626975114899e-05, "loss": 1.5954, "step": 17489 }, { "epoch": 0.02, "learning_rate": 4.9962693222079295e-05, "loss": 1.3534, "step": 17490 }, { "epoch": 0.02, "learning_rate": 4.9962688932422267e-05, "loss": 0.2291, "step": 17491 }, { "epoch": 0.02, "learning_rate": 4.996268464251882e-05, "loss": 0.1923, "step": 17492 }, { "epoch": 0.02, "learning_rate": 4.996268035236895e-05, "loss": 0.1576, "step": 17493 }, { "epoch": 0.02, "learning_rate": 4.996267606197267e-05, "loss": 0.8841, "step": 17494 }, { "epoch": 0.02, "learning_rate": 4.996267177132996e-05, "loss": 1.2412, "step": 17495 }, { "epoch": 0.02, "learning_rate": 4.9962667480440826e-05, "loss": 1.2512, "step": 17496 }, { "epoch": 0.02, "learning_rate": 4.996266318930528e-05, "loss": 1.2207, "step": 17497 }, { "epoch": 0.02, "learning_rate": 4.996265889792331e-05, "loss": 0.8542, "step": 17498 }, { "epoch": 0.02, "learning_rate": 4.996265460629491e-05, "loss": 0.9536, "step": 17499 }, { "epoch": 0.02, "learning_rate": 4.99626503144201e-05, "loss": 1.121, "step": 17500 }, { "epoch": 0.02, "eval_loss": 1.0300171375274658, "eval_runtime": 114.2587, "eval_samples_per_second": 12.122, "eval_steps_per_second": 3.037, "step": 17500 }, { "epoch": 0.02, "learning_rate": 4.996264602229886e-05, "loss": 1.0129, "step": 17501 }, { "epoch": 0.02, "learning_rate": 4.996264172993121e-05, "loss": 1.2211, "step": 17502 }, { "epoch": 0.02, "learning_rate": 4.996263743731714e-05, "loss": 1.1894, "step": 17503 }, { "epoch": 0.02, "learning_rate": 4.9962633144456636e-05, "loss": 1.2215, "step": 17504 }, { "epoch": 0.02, "learning_rate": 4.9962628851349724e-05, "loss": 1.1546, "step": 17505 }, { "epoch": 0.02, "learning_rate": 4.9962624557996385e-05, "loss": 1.0137, "step": 17506 }, { "epoch": 0.02, "learning_rate": 4.996262026439663e-05, "loss": 0.747, "step": 17507 }, { "epoch": 0.02, "learning_rate": 4.996261597055045e-05, "loss": 1.1863, "step": 17508 }, { "epoch": 0.02, "learning_rate": 4.996261167645785e-05, "loss": 0.9583, "step": 17509 }, { "epoch": 0.02, "learning_rate": 4.996260738211883e-05, "loss": 1.0171, "step": 17510 }, { "epoch": 0.02, "learning_rate": 4.996260308753339e-05, "loss": 1.4778, "step": 17511 }, { "epoch": 0.02, "learning_rate": 4.996259879270153e-05, "loss": 1.4268, "step": 17512 }, { "epoch": 0.02, "learning_rate": 4.996259449762325e-05, "loss": 1.1606, "step": 17513 }, { "epoch": 0.02, "learning_rate": 4.9962590202298554e-05, "loss": 1.011, "step": 17514 }, { "epoch": 0.02, "learning_rate": 4.996258590672743e-05, "loss": 0.9438, "step": 17515 }, { "epoch": 0.02, "learning_rate": 4.9962581610909885e-05, "loss": 0.9229, "step": 17516 }, { "epoch": 0.02, "learning_rate": 4.9962577314845926e-05, "loss": 1.1587, "step": 17517 }, { "epoch": 0.02, "learning_rate": 4.9962573018535555e-05, "loss": 1.0432, "step": 17518 }, { "epoch": 0.02, "learning_rate": 4.9962568721978745e-05, "loss": 0.9585, "step": 17519 }, { "epoch": 0.02, "learning_rate": 4.996256442517553e-05, "loss": 1.192, "step": 17520 }, { "epoch": 0.02, "learning_rate": 4.996256012812589e-05, "loss": 1.2137, "step": 17521 }, { "epoch": 0.02, "learning_rate": 4.996255583082983e-05, "loss": 0.953, "step": 17522 }, { "epoch": 0.02, "learning_rate": 4.996255153328735e-05, "loss": 0.4916, "step": 17523 }, { "epoch": 0.02, "learning_rate": 4.9962547235498446e-05, "loss": 0.9689, "step": 17524 }, { "epoch": 0.02, "learning_rate": 4.996254293746313e-05, "loss": 1.0869, "step": 17525 }, { "epoch": 0.02, "learning_rate": 4.996253863918139e-05, "loss": 1.2203, "step": 17526 }, { "epoch": 0.02, "learning_rate": 4.996253434065323e-05, "loss": 1.1731, "step": 17527 }, { "epoch": 0.02, "learning_rate": 4.9962530041878654e-05, "loss": 0.8759, "step": 17528 }, { "epoch": 0.02, "learning_rate": 4.996252574285765e-05, "loss": 0.9557, "step": 17529 }, { "epoch": 0.02, "learning_rate": 4.996252144359024e-05, "loss": 1.3017, "step": 17530 }, { "epoch": 0.02, "learning_rate": 4.9962517144076396e-05, "loss": 1.0891, "step": 17531 }, { "epoch": 0.02, "learning_rate": 4.996251284431614e-05, "loss": 0.9834, "step": 17532 }, { "epoch": 0.02, "learning_rate": 4.996250854430946e-05, "loss": 1.1332, "step": 17533 }, { "epoch": 0.02, "learning_rate": 4.996250424405637e-05, "loss": 1.5566, "step": 17534 }, { "epoch": 0.02, "learning_rate": 4.996249994355685e-05, "loss": 1.0745, "step": 17535 }, { "epoch": 0.02, "learning_rate": 4.996249564281092e-05, "loss": 0.9592, "step": 17536 }, { "epoch": 0.02, "learning_rate": 4.9962491341818564e-05, "loss": 0.7865, "step": 17537 }, { "epoch": 0.02, "learning_rate": 4.996248704057979e-05, "loss": 0.8038, "step": 17538 }, { "epoch": 0.02, "learning_rate": 4.99624827390946e-05, "loss": 1.1383, "step": 17539 }, { "epoch": 0.02, "learning_rate": 4.996247843736299e-05, "loss": 1.1819, "step": 17540 }, { "epoch": 0.02, "learning_rate": 4.996247413538496e-05, "loss": 1.2586, "step": 17541 }, { "epoch": 0.02, "learning_rate": 4.996246983316051e-05, "loss": 1.1353, "step": 17542 }, { "epoch": 0.02, "learning_rate": 4.996246553068964e-05, "loss": 1.0877, "step": 17543 }, { "epoch": 0.02, "learning_rate": 4.996246122797236e-05, "loss": 1.3518, "step": 17544 }, { "epoch": 0.02, "learning_rate": 4.996245692500865e-05, "loss": 0.7672, "step": 17545 }, { "epoch": 0.02, "learning_rate": 4.996245262179853e-05, "loss": 0.816, "step": 17546 }, { "epoch": 0.02, "learning_rate": 4.9962448318341984e-05, "loss": 1.0396, "step": 17547 }, { "epoch": 0.02, "learning_rate": 4.996244401463902e-05, "loss": 0.8345, "step": 17548 }, { "epoch": 0.02, "learning_rate": 4.996243971068964e-05, "loss": 0.7996, "step": 17549 }, { "epoch": 0.02, "learning_rate": 4.996243540649384e-05, "loss": 1.3654, "step": 17550 }, { "epoch": 0.02, "learning_rate": 4.996243110205162e-05, "loss": 1.2618, "step": 17551 }, { "epoch": 0.02, "learning_rate": 4.996242679736298e-05, "loss": 1.0082, "step": 17552 }, { "epoch": 0.02, "learning_rate": 4.996242249242793e-05, "loss": 1.0619, "step": 17553 }, { "epoch": 0.02, "learning_rate": 4.996241818724645e-05, "loss": 1.2977, "step": 17554 }, { "epoch": 0.02, "learning_rate": 4.996241388181856e-05, "loss": 1.0483, "step": 17555 }, { "epoch": 0.02, "learning_rate": 4.9962409576144254e-05, "loss": 0.9183, "step": 17556 }, { "epoch": 0.02, "learning_rate": 4.996240527022352e-05, "loss": 1.1413, "step": 17557 }, { "epoch": 0.02, "learning_rate": 4.996240096405638e-05, "loss": 0.8811, "step": 17558 }, { "epoch": 0.02, "learning_rate": 4.996239665764281e-05, "loss": 1.2113, "step": 17559 }, { "epoch": 0.02, "learning_rate": 4.9962392350982826e-05, "loss": 1.7992, "step": 17560 }, { "epoch": 0.02, "learning_rate": 4.9962388044076425e-05, "loss": 1.382, "step": 17561 }, { "epoch": 0.02, "learning_rate": 4.99623837369236e-05, "loss": 1.8228, "step": 17562 }, { "epoch": 0.02, "learning_rate": 4.996237942952437e-05, "loss": 1.6389, "step": 17563 }, { "epoch": 0.02, "learning_rate": 4.996237512187871e-05, "loss": 1.0656, "step": 17564 }, { "epoch": 0.02, "learning_rate": 4.9962370813986634e-05, "loss": 0.7273, "step": 17565 }, { "epoch": 0.02, "learning_rate": 4.996236650584813e-05, "loss": 0.9601, "step": 17566 }, { "epoch": 0.02, "learning_rate": 4.996236219746323e-05, "loss": 0.9904, "step": 17567 }, { "epoch": 0.02, "learning_rate": 4.99623578888319e-05, "loss": 1.0461, "step": 17568 }, { "epoch": 0.02, "learning_rate": 4.9962353579954155e-05, "loss": 1.2404, "step": 17569 }, { "epoch": 0.02, "learning_rate": 4.996234927082999e-05, "loss": 0.8456, "step": 17570 }, { "epoch": 0.02, "learning_rate": 4.996234496145941e-05, "loss": 1.2965, "step": 17571 }, { "epoch": 0.02, "learning_rate": 4.996234065184241e-05, "loss": 1.2, "step": 17572 }, { "epoch": 0.02, "learning_rate": 4.996233634197899e-05, "loss": 0.7102, "step": 17573 }, { "epoch": 0.02, "learning_rate": 4.996233203186915e-05, "loss": 1.0808, "step": 17574 }, { "epoch": 0.02, "learning_rate": 4.9962327721512905e-05, "loss": 1.0467, "step": 17575 }, { "epoch": 0.02, "learning_rate": 4.996232341091023e-05, "loss": 1.1065, "step": 17576 }, { "epoch": 0.02, "learning_rate": 4.996231910006114e-05, "loss": 1.1417, "step": 17577 }, { "epoch": 0.02, "learning_rate": 4.996231478896563e-05, "loss": 1.1984, "step": 17578 }, { "epoch": 0.02, "learning_rate": 4.9962310477623714e-05, "loss": 1.3023, "step": 17579 }, { "epoch": 0.02, "learning_rate": 4.996230616603537e-05, "loss": 1.1139, "step": 17580 }, { "epoch": 0.02, "learning_rate": 4.996230185420061e-05, "loss": 2.0879, "step": 17581 }, { "epoch": 0.02, "learning_rate": 4.996229754211944e-05, "loss": 1.7322, "step": 17582 }, { "epoch": 0.02, "learning_rate": 4.996229322979185e-05, "loss": 1.205, "step": 17583 }, { "epoch": 0.02, "learning_rate": 4.9962288917217835e-05, "loss": 1.0267, "step": 17584 }, { "epoch": 0.02, "learning_rate": 4.9962284604397416e-05, "loss": 1.4575, "step": 17585 }, { "epoch": 0.02, "learning_rate": 4.9962280291330565e-05, "loss": 1.1628, "step": 17586 }, { "epoch": 0.02, "learning_rate": 4.99622759780173e-05, "loss": 1.0927, "step": 17587 }, { "epoch": 0.02, "learning_rate": 4.9962271664457626e-05, "loss": 0.9515, "step": 17588 }, { "epoch": 0.02, "learning_rate": 4.996226735065153e-05, "loss": 0.8999, "step": 17589 }, { "epoch": 0.02, "learning_rate": 4.996226303659902e-05, "loss": 1.1061, "step": 17590 }, { "epoch": 0.02, "learning_rate": 4.9962258722300095e-05, "loss": 0.9472, "step": 17591 }, { "epoch": 0.02, "learning_rate": 4.9962254407754744e-05, "loss": 0.8737, "step": 17592 }, { "epoch": 0.02, "learning_rate": 4.996225009296298e-05, "loss": 1.2111, "step": 17593 }, { "epoch": 0.02, "learning_rate": 4.99622457779248e-05, "loss": 1.0294, "step": 17594 }, { "epoch": 0.02, "learning_rate": 4.99622414626402e-05, "loss": 0.6605, "step": 17595 }, { "epoch": 0.02, "learning_rate": 4.9962237147109195e-05, "loss": 0.8284, "step": 17596 }, { "epoch": 0.02, "learning_rate": 4.9962232831331763e-05, "loss": 0.9766, "step": 17597 }, { "epoch": 0.02, "learning_rate": 4.996222851530792e-05, "loss": 1.1289, "step": 17598 }, { "epoch": 0.02, "learning_rate": 4.996222419903765e-05, "loss": 1.2024, "step": 17599 }, { "epoch": 0.02, "learning_rate": 4.996221988252097e-05, "loss": 1.1505, "step": 17600 }, { "epoch": 0.02, "learning_rate": 4.996221556575788e-05, "loss": 0.985, "step": 17601 }, { "epoch": 0.02, "learning_rate": 4.9962211248748366e-05, "loss": 1.0529, "step": 17602 }, { "epoch": 0.02, "learning_rate": 4.9962206931492436e-05, "loss": 1.0801, "step": 17603 }, { "epoch": 0.02, "learning_rate": 4.996220261399009e-05, "loss": 0.9786, "step": 17604 }, { "epoch": 0.02, "learning_rate": 4.9962198296241325e-05, "loss": 1.1808, "step": 17605 }, { "epoch": 0.02, "learning_rate": 4.996219397824615e-05, "loss": 0.9943, "step": 17606 }, { "epoch": 0.02, "learning_rate": 4.996218966000455e-05, "loss": 0.9549, "step": 17607 }, { "epoch": 0.02, "learning_rate": 4.996218534151654e-05, "loss": 1.1674, "step": 17608 }, { "epoch": 0.02, "learning_rate": 4.996218102278212e-05, "loss": 1.1715, "step": 17609 }, { "epoch": 0.02, "learning_rate": 4.996217670380127e-05, "loss": 0.9442, "step": 17610 }, { "epoch": 0.02, "learning_rate": 4.9962172384574016e-05, "loss": 1.1659, "step": 17611 }, { "epoch": 0.02, "learning_rate": 4.996216806510034e-05, "loss": 0.6341, "step": 17612 }, { "epoch": 0.02, "learning_rate": 4.9962163745380245e-05, "loss": 0.158, "step": 17613 }, { "epoch": 0.02, "learning_rate": 4.996215942541374e-05, "loss": 0.8477, "step": 17614 }, { "epoch": 0.02, "learning_rate": 4.996215510520081e-05, "loss": 0.924, "step": 17615 }, { "epoch": 0.02, "learning_rate": 4.996215078474148e-05, "loss": 0.8934, "step": 17616 }, { "epoch": 0.02, "learning_rate": 4.9962146464035724e-05, "loss": 1.1694, "step": 17617 }, { "epoch": 0.02, "learning_rate": 4.996214214308356e-05, "loss": 1.1512, "step": 17618 }, { "epoch": 0.02, "learning_rate": 4.996213782188497e-05, "loss": 1.1294, "step": 17619 }, { "epoch": 0.02, "learning_rate": 4.9962133500439965e-05, "loss": 0.9833, "step": 17620 }, { "epoch": 0.02, "learning_rate": 4.996212917874855e-05, "loss": 1.2548, "step": 17621 }, { "epoch": 0.02, "learning_rate": 4.9962124856810716e-05, "loss": 0.9866, "step": 17622 }, { "epoch": 0.02, "learning_rate": 4.996212053462647e-05, "loss": 1.1241, "step": 17623 }, { "epoch": 0.02, "learning_rate": 4.9962116212195805e-05, "loss": 1.5279, "step": 17624 }, { "epoch": 0.02, "learning_rate": 4.996211188951873e-05, "loss": 1.1799, "step": 17625 }, { "epoch": 0.02, "learning_rate": 4.9962107566595226e-05, "loss": 1.17, "step": 17626 }, { "epoch": 0.02, "learning_rate": 4.996210324342532e-05, "loss": 0.7973, "step": 17627 }, { "epoch": 0.02, "learning_rate": 4.996209892000899e-05, "loss": 1.0882, "step": 17628 }, { "epoch": 0.02, "learning_rate": 4.9962094596346246e-05, "loss": 1.2165, "step": 17629 }, { "epoch": 0.02, "learning_rate": 4.996209027243709e-05, "loss": 1.1187, "step": 17630 }, { "epoch": 0.02, "learning_rate": 4.9962085948281526e-05, "loss": 1.0985, "step": 17631 }, { "epoch": 0.02, "learning_rate": 4.996208162387953e-05, "loss": 1.1231, "step": 17632 }, { "epoch": 0.02, "learning_rate": 4.996207729923113e-05, "loss": 1.06, "step": 17633 }, { "epoch": 0.02, "learning_rate": 4.996207297433632e-05, "loss": 1.1135, "step": 17634 }, { "epoch": 0.02, "learning_rate": 4.996206864919508e-05, "loss": 0.9715, "step": 17635 }, { "epoch": 0.02, "learning_rate": 4.996206432380744e-05, "loss": 0.8571, "step": 17636 }, { "epoch": 0.02, "learning_rate": 4.9962059998173375e-05, "loss": 0.9907, "step": 17637 }, { "epoch": 0.02, "learning_rate": 4.9962055672292894e-05, "loss": 0.9307, "step": 17638 }, { "epoch": 0.02, "learning_rate": 4.996205134616601e-05, "loss": 1.1845, "step": 17639 }, { "epoch": 0.02, "learning_rate": 4.996204701979269e-05, "loss": 1.1976, "step": 17640 }, { "epoch": 0.02, "learning_rate": 4.996204269317297e-05, "loss": 1.2271, "step": 17641 }, { "epoch": 0.02, "learning_rate": 4.9962038366306836e-05, "loss": 1.0322, "step": 17642 }, { "epoch": 0.02, "learning_rate": 4.996203403919428e-05, "loss": 1.2612, "step": 17643 }, { "epoch": 0.02, "learning_rate": 4.9962029711835315e-05, "loss": 1.1587, "step": 17644 }, { "epoch": 0.02, "learning_rate": 4.9962025384229936e-05, "loss": 1.1265, "step": 17645 }, { "epoch": 0.02, "learning_rate": 4.996202105637814e-05, "loss": 0.9862, "step": 17646 }, { "epoch": 0.02, "learning_rate": 4.996201672827993e-05, "loss": 1.0855, "step": 17647 }, { "epoch": 0.02, "learning_rate": 4.996201239993531e-05, "loss": 1.1253, "step": 17648 }, { "epoch": 0.02, "learning_rate": 4.9962008071344266e-05, "loss": 0.9555, "step": 17649 }, { "epoch": 0.02, "learning_rate": 4.9962003742506813e-05, "loss": 1.054, "step": 17650 }, { "epoch": 0.02, "learning_rate": 4.996199941342294e-05, "loss": 1.0531, "step": 17651 }, { "epoch": 0.02, "learning_rate": 4.9961995084092665e-05, "loss": 0.9774, "step": 17652 }, { "epoch": 0.02, "learning_rate": 4.996199075451597e-05, "loss": 1.2955, "step": 17653 }, { "epoch": 0.02, "learning_rate": 4.9961986424692855e-05, "loss": 0.9387, "step": 17654 }, { "epoch": 0.02, "learning_rate": 4.9961982094623336e-05, "loss": 1.0335, "step": 17655 }, { "epoch": 0.02, "learning_rate": 4.99619777643074e-05, "loss": 1.0396, "step": 17656 }, { "epoch": 0.02, "learning_rate": 4.996197343374505e-05, "loss": 1.0175, "step": 17657 }, { "epoch": 0.02, "learning_rate": 4.996196910293628e-05, "loss": 1.3186, "step": 17658 }, { "epoch": 0.02, "learning_rate": 4.9961964771881096e-05, "loss": 1.2502, "step": 17659 }, { "epoch": 0.02, "learning_rate": 4.99619604405795e-05, "loss": 1.0594, "step": 17660 }, { "epoch": 0.02, "learning_rate": 4.996195610903149e-05, "loss": 1.0422, "step": 17661 }, { "epoch": 0.02, "learning_rate": 4.996195177723707e-05, "loss": 0.91, "step": 17662 }, { "epoch": 0.02, "learning_rate": 4.9961947445196237e-05, "loss": 1.1158, "step": 17663 }, { "epoch": 0.02, "learning_rate": 4.996194311290898e-05, "loss": 0.8666, "step": 17664 }, { "epoch": 0.02, "learning_rate": 4.9961938780375314e-05, "loss": 0.8887, "step": 17665 }, { "epoch": 0.02, "learning_rate": 4.9961934447595235e-05, "loss": 1.287, "step": 17666 }, { "epoch": 0.02, "learning_rate": 4.9961930114568744e-05, "loss": 1.2711, "step": 17667 }, { "epoch": 0.02, "learning_rate": 4.996192578129584e-05, "loss": 1.0686, "step": 17668 }, { "epoch": 0.02, "learning_rate": 4.996192144777652e-05, "loss": 0.8081, "step": 17669 }, { "epoch": 0.02, "learning_rate": 4.9961917114010784e-05, "loss": 1.7464, "step": 17670 }, { "epoch": 0.02, "learning_rate": 4.9961912779998644e-05, "loss": 0.95, "step": 17671 }, { "epoch": 0.02, "learning_rate": 4.996190844574008e-05, "loss": 1.005, "step": 17672 }, { "epoch": 0.02, "learning_rate": 4.996190411123511e-05, "loss": 1.1172, "step": 17673 }, { "epoch": 0.02, "learning_rate": 4.996189977648372e-05, "loss": 1.0489, "step": 17674 }, { "epoch": 0.02, "learning_rate": 4.996189544148592e-05, "loss": 0.9513, "step": 17675 }, { "epoch": 0.02, "learning_rate": 4.996189110624171e-05, "loss": 0.7141, "step": 17676 }, { "epoch": 0.02, "learning_rate": 4.996188677075108e-05, "loss": 0.8956, "step": 17677 }, { "epoch": 0.02, "learning_rate": 4.996188243501404e-05, "loss": 1.0659, "step": 17678 }, { "epoch": 0.02, "learning_rate": 4.996187809903059e-05, "loss": 1.1496, "step": 17679 }, { "epoch": 0.02, "learning_rate": 4.996187376280072e-05, "loss": 1.0677, "step": 17680 }, { "epoch": 0.02, "learning_rate": 4.9961869426324444e-05, "loss": 0.8862, "step": 17681 }, { "epoch": 0.02, "learning_rate": 4.996186508960175e-05, "loss": 1.7406, "step": 17682 }, { "epoch": 0.02, "learning_rate": 4.996186075263265e-05, "loss": 1.4485, "step": 17683 }, { "epoch": 0.02, "learning_rate": 4.996185641541713e-05, "loss": 1.1186, "step": 17684 }, { "epoch": 0.02, "learning_rate": 4.9961852077955196e-05, "loss": 1.2264, "step": 17685 }, { "epoch": 0.02, "learning_rate": 4.996184774024685e-05, "loss": 1.192, "step": 17686 }, { "epoch": 0.02, "learning_rate": 4.9961843402292096e-05, "loss": 1.0894, "step": 17687 }, { "epoch": 0.02, "learning_rate": 4.996183906409092e-05, "loss": 1.0665, "step": 17688 }, { "epoch": 0.02, "learning_rate": 4.996183472564334e-05, "loss": 1.2266, "step": 17689 }, { "epoch": 0.02, "learning_rate": 4.996183038694935e-05, "loss": 1.2252, "step": 17690 }, { "epoch": 0.02, "learning_rate": 4.9961826048008944e-05, "loss": 1.0922, "step": 17691 }, { "epoch": 0.02, "learning_rate": 4.996182170882212e-05, "loss": 0.8818, "step": 17692 }, { "epoch": 0.02, "learning_rate": 4.9961817369388886e-05, "loss": 1.0697, "step": 17693 }, { "epoch": 0.02, "learning_rate": 4.9961813029709236e-05, "loss": 1.0744, "step": 17694 }, { "epoch": 0.02, "learning_rate": 4.9961808689783187e-05, "loss": 1.1978, "step": 17695 }, { "epoch": 0.02, "learning_rate": 4.9961804349610705e-05, "loss": 1.1365, "step": 17696 }, { "epoch": 0.02, "learning_rate": 4.9961800009191825e-05, "loss": 1.2275, "step": 17697 }, { "epoch": 0.02, "learning_rate": 4.996179566852653e-05, "loss": 1.0285, "step": 17698 }, { "epoch": 0.02, "learning_rate": 4.996179132761483e-05, "loss": 0.7626, "step": 17699 }, { "epoch": 0.02, "learning_rate": 4.9961786986456707e-05, "loss": 0.7302, "step": 17700 }, { "epoch": 0.02, "learning_rate": 4.996178264505217e-05, "loss": 0.6511, "step": 17701 }, { "epoch": 0.02, "learning_rate": 4.996177830340123e-05, "loss": 1.058, "step": 17702 }, { "epoch": 0.02, "learning_rate": 4.996177396150387e-05, "loss": 1.0377, "step": 17703 }, { "epoch": 0.02, "learning_rate": 4.99617696193601e-05, "loss": 1.0946, "step": 17704 }, { "epoch": 0.02, "learning_rate": 4.996176527696992e-05, "loss": 1.3185, "step": 17705 }, { "epoch": 0.02, "learning_rate": 4.996176093433333e-05, "loss": 1.1071, "step": 17706 }, { "epoch": 0.02, "learning_rate": 4.996175659145032e-05, "loss": 1.4332, "step": 17707 }, { "epoch": 0.02, "learning_rate": 4.996175224832091e-05, "loss": 0.8426, "step": 17708 }, { "epoch": 0.02, "learning_rate": 4.996174790494508e-05, "loss": 0.9583, "step": 17709 }, { "epoch": 0.02, "learning_rate": 4.9961743561322836e-05, "loss": 1.2795, "step": 17710 }, { "epoch": 0.02, "learning_rate": 4.996173921745419e-05, "loss": 0.8293, "step": 17711 }, { "epoch": 0.02, "learning_rate": 4.996173487333912e-05, "loss": 1.5821, "step": 17712 }, { "epoch": 0.02, "learning_rate": 4.996173052897765e-05, "loss": 1.1, "step": 17713 }, { "epoch": 0.02, "learning_rate": 4.996172618436976e-05, "loss": 1.1575, "step": 17714 }, { "epoch": 0.02, "learning_rate": 4.9961721839515466e-05, "loss": 1.0456, "step": 17715 }, { "epoch": 0.02, "learning_rate": 4.996171749441475e-05, "loss": 1.176, "step": 17716 }, { "epoch": 0.02, "learning_rate": 4.996171314906762e-05, "loss": 1.1851, "step": 17717 }, { "epoch": 0.02, "learning_rate": 4.99617088034741e-05, "loss": 0.9399, "step": 17718 }, { "epoch": 0.02, "learning_rate": 4.996170445763415e-05, "loss": 0.9748, "step": 17719 }, { "epoch": 0.02, "learning_rate": 4.9961700111547795e-05, "loss": 1.0271, "step": 17720 }, { "epoch": 0.02, "learning_rate": 4.996169576521502e-05, "loss": 1.2641, "step": 17721 }, { "epoch": 0.02, "learning_rate": 4.9961691418635844e-05, "loss": 1.2566, "step": 17722 }, { "epoch": 0.02, "learning_rate": 4.996168707181025e-05, "loss": 1.2029, "step": 17723 }, { "epoch": 0.02, "learning_rate": 4.996168272473825e-05, "loss": 1.2951, "step": 17724 }, { "epoch": 0.02, "learning_rate": 4.996167837741984e-05, "loss": 1.1431, "step": 17725 }, { "epoch": 0.02, "learning_rate": 4.996167402985501e-05, "loss": 0.9106, "step": 17726 }, { "epoch": 0.02, "learning_rate": 4.996166968204378e-05, "loss": 1.1292, "step": 17727 }, { "epoch": 0.02, "learning_rate": 4.996166533398613e-05, "loss": 0.7149, "step": 17728 }, { "epoch": 0.02, "learning_rate": 4.996166098568207e-05, "loss": 1.0786, "step": 17729 }, { "epoch": 0.02, "learning_rate": 4.99616566371316e-05, "loss": 1.0467, "step": 17730 }, { "epoch": 0.02, "learning_rate": 4.9961652288334727e-05, "loss": 0.9294, "step": 17731 }, { "epoch": 0.02, "learning_rate": 4.996164793929143e-05, "loss": 1.1534, "step": 17732 }, { "epoch": 0.02, "learning_rate": 4.996164359000173e-05, "loss": 1.0862, "step": 17733 }, { "epoch": 0.02, "learning_rate": 4.996163924046562e-05, "loss": 0.9491, "step": 17734 }, { "epoch": 0.02, "learning_rate": 4.9961634890683095e-05, "loss": 1.2404, "step": 17735 }, { "epoch": 0.02, "learning_rate": 4.9961630540654166e-05, "loss": 1.066, "step": 17736 }, { "epoch": 0.02, "learning_rate": 4.996162619037882e-05, "loss": 0.8946, "step": 17737 }, { "epoch": 0.02, "learning_rate": 4.9961621839857064e-05, "loss": 0.2297, "step": 17738 }, { "epoch": 0.02, "learning_rate": 4.99616174890889e-05, "loss": 0.0471, "step": 17739 }, { "epoch": 0.02, "learning_rate": 4.996161313807433e-05, "loss": 0.232, "step": 17740 }, { "epoch": 0.02, "learning_rate": 4.996160878681334e-05, "loss": 1.3398, "step": 17741 }, { "epoch": 0.02, "learning_rate": 4.996160443530594e-05, "loss": 1.2047, "step": 17742 }, { "epoch": 0.02, "learning_rate": 4.996160008355213e-05, "loss": 1.3647, "step": 17743 }, { "epoch": 0.02, "learning_rate": 4.996159573155192e-05, "loss": 1.0151, "step": 17744 }, { "epoch": 0.02, "learning_rate": 4.9961591379305286e-05, "loss": 1.1073, "step": 17745 }, { "epoch": 0.02, "learning_rate": 4.996158702681224e-05, "loss": 0.8109, "step": 17746 }, { "epoch": 0.02, "learning_rate": 4.9961582674072794e-05, "loss": 1.158, "step": 17747 }, { "epoch": 0.02, "learning_rate": 4.996157832108694e-05, "loss": 1.232, "step": 17748 }, { "epoch": 0.02, "learning_rate": 4.996157396785467e-05, "loss": 1.256, "step": 17749 }, { "epoch": 0.02, "learning_rate": 4.996156961437599e-05, "loss": 1.1706, "step": 17750 }, { "epoch": 0.02, "learning_rate": 4.99615652606509e-05, "loss": 1.1536, "step": 17751 }, { "epoch": 0.02, "learning_rate": 4.99615609066794e-05, "loss": 1.0007, "step": 17752 }, { "epoch": 0.02, "learning_rate": 4.9961556552461484e-05, "loss": 0.6541, "step": 17753 }, { "epoch": 0.02, "learning_rate": 4.996155219799717e-05, "loss": 1.1607, "step": 17754 }, { "epoch": 0.02, "learning_rate": 4.996154784328644e-05, "loss": 1.0444, "step": 17755 }, { "epoch": 0.02, "learning_rate": 4.99615434883293e-05, "loss": 0.8942, "step": 17756 }, { "epoch": 0.02, "learning_rate": 4.996153913312575e-05, "loss": 1.1181, "step": 17757 }, { "epoch": 0.02, "learning_rate": 4.996153477767579e-05, "loss": 1.1953, "step": 17758 }, { "epoch": 0.02, "learning_rate": 4.996153042197942e-05, "loss": 1.0063, "step": 17759 }, { "epoch": 0.02, "learning_rate": 4.9961526066036644e-05, "loss": 0.9434, "step": 17760 }, { "epoch": 0.02, "learning_rate": 4.996152170984746e-05, "loss": 1.2352, "step": 17761 }, { "epoch": 0.02, "learning_rate": 4.996151735341186e-05, "loss": 0.9462, "step": 17762 }, { "epoch": 0.02, "learning_rate": 4.996151299672985e-05, "loss": 1.0325, "step": 17763 }, { "epoch": 0.02, "learning_rate": 4.996150863980144e-05, "loss": 0.7396, "step": 17764 }, { "epoch": 0.02, "learning_rate": 4.99615042826266e-05, "loss": 0.9112, "step": 17765 }, { "epoch": 0.02, "learning_rate": 4.996149992520537e-05, "loss": 1.1087, "step": 17766 }, { "epoch": 0.02, "learning_rate": 4.996149556753772e-05, "loss": 1.0122, "step": 17767 }, { "epoch": 0.02, "learning_rate": 4.996149120962367e-05, "loss": 1.1162, "step": 17768 }, { "epoch": 0.02, "learning_rate": 4.996148685146321e-05, "loss": 1.0184, "step": 17769 }, { "epoch": 0.02, "learning_rate": 4.996148249305633e-05, "loss": 1.0254, "step": 17770 }, { "epoch": 0.02, "learning_rate": 4.996147813440305e-05, "loss": 1.1897, "step": 17771 }, { "epoch": 0.02, "learning_rate": 4.996147377550336e-05, "loss": 1.1157, "step": 17772 }, { "epoch": 0.02, "learning_rate": 4.996146941635725e-05, "loss": 1.2131, "step": 17773 }, { "epoch": 0.02, "learning_rate": 4.996146505696475e-05, "loss": 1.1821, "step": 17774 }, { "epoch": 0.02, "learning_rate": 4.996146069732583e-05, "loss": 1.0645, "step": 17775 }, { "epoch": 0.02, "learning_rate": 4.99614563374405e-05, "loss": 1.0137, "step": 17776 }, { "epoch": 0.02, "learning_rate": 4.996145197730876e-05, "loss": 1.1401, "step": 17777 }, { "epoch": 0.02, "learning_rate": 4.996144761693062e-05, "loss": 1.0146, "step": 17778 }, { "epoch": 0.02, "learning_rate": 4.996144325630606e-05, "loss": 0.8929, "step": 17779 }, { "epoch": 0.02, "learning_rate": 4.9961438895435096e-05, "loss": 1.1418, "step": 17780 }, { "epoch": 0.02, "learning_rate": 4.9961434534317726e-05, "loss": 1.2176, "step": 17781 }, { "epoch": 0.02, "learning_rate": 4.9961430172953944e-05, "loss": 1.0159, "step": 17782 }, { "epoch": 0.02, "learning_rate": 4.996142581134376e-05, "loss": 1.3819, "step": 17783 }, { "epoch": 0.02, "learning_rate": 4.996142144948716e-05, "loss": 1.0658, "step": 17784 }, { "epoch": 0.02, "learning_rate": 4.9961417087384146e-05, "loss": 1.0802, "step": 17785 }, { "epoch": 0.02, "learning_rate": 4.996141272503473e-05, "loss": 0.9955, "step": 17786 }, { "epoch": 0.02, "learning_rate": 4.996140836243891e-05, "loss": 1.0343, "step": 17787 }, { "epoch": 0.02, "learning_rate": 4.996140399959668e-05, "loss": 0.9223, "step": 17788 }, { "epoch": 0.02, "learning_rate": 4.9961399636508034e-05, "loss": 0.8904, "step": 17789 }, { "epoch": 0.02, "learning_rate": 4.996139527317298e-05, "loss": 0.9913, "step": 17790 }, { "epoch": 0.02, "learning_rate": 4.996139090959152e-05, "loss": 1.0482, "step": 17791 }, { "epoch": 0.02, "learning_rate": 4.996138654576366e-05, "loss": 0.9714, "step": 17792 }, { "epoch": 0.02, "learning_rate": 4.9961382181689384e-05, "loss": 1.0372, "step": 17793 }, { "epoch": 0.02, "learning_rate": 4.9961377817368704e-05, "loss": 0.998, "step": 17794 }, { "epoch": 0.02, "learning_rate": 4.9961373452801606e-05, "loss": 0.4208, "step": 17795 }, { "epoch": 0.02, "learning_rate": 4.996136908798811e-05, "loss": 0.752, "step": 17796 }, { "epoch": 0.02, "learning_rate": 4.996136472292821e-05, "loss": 1.138, "step": 17797 }, { "epoch": 0.02, "learning_rate": 4.996136035762189e-05, "loss": 1.0157, "step": 17798 }, { "epoch": 0.02, "learning_rate": 4.996135599206916e-05, "loss": 1.2037, "step": 17799 }, { "epoch": 0.02, "learning_rate": 4.996135162627003e-05, "loss": 1.1487, "step": 17800 }, { "epoch": 0.02, "learning_rate": 4.996134726022449e-05, "loss": 0.7653, "step": 17801 }, { "epoch": 0.02, "learning_rate": 4.9961342893932545e-05, "loss": 0.8066, "step": 17802 }, { "epoch": 0.02, "learning_rate": 4.9961338527394185e-05, "loss": 1.0121, "step": 17803 }, { "epoch": 0.02, "learning_rate": 4.9961334160609426e-05, "loss": 1.0457, "step": 17804 }, { "epoch": 0.02, "learning_rate": 4.9961329793578255e-05, "loss": 1.2664, "step": 17805 }, { "epoch": 0.02, "learning_rate": 4.996132542630068e-05, "loss": 1.0643, "step": 17806 }, { "epoch": 0.02, "learning_rate": 4.996132105877669e-05, "loss": 0.9038, "step": 17807 }, { "epoch": 0.02, "learning_rate": 4.9961316691006296e-05, "loss": 0.7782, "step": 17808 }, { "epoch": 0.02, "learning_rate": 4.99613123229895e-05, "loss": 1.0781, "step": 17809 }, { "epoch": 0.02, "learning_rate": 4.9961307954726286e-05, "loss": 0.7864, "step": 17810 }, { "epoch": 0.02, "learning_rate": 4.996130358621667e-05, "loss": 1.3976, "step": 17811 }, { "epoch": 0.02, "learning_rate": 4.996129921746064e-05, "loss": 1.5575, "step": 17812 }, { "epoch": 0.02, "learning_rate": 4.996129484845821e-05, "loss": 1.39, "step": 17813 }, { "epoch": 0.02, "learning_rate": 4.996129047920937e-05, "loss": 1.1701, "step": 17814 }, { "epoch": 0.02, "learning_rate": 4.9961286109714125e-05, "loss": 0.9059, "step": 17815 }, { "epoch": 0.02, "learning_rate": 4.996128173997247e-05, "loss": 1.0271, "step": 17816 }, { "epoch": 0.02, "learning_rate": 4.996127736998441e-05, "loss": 0.9995, "step": 17817 }, { "epoch": 0.02, "learning_rate": 4.996127299974994e-05, "loss": 1.1153, "step": 17818 }, { "epoch": 0.02, "learning_rate": 4.996126862926906e-05, "loss": 1.0012, "step": 17819 }, { "epoch": 0.02, "learning_rate": 4.9961264258541785e-05, "loss": 1.0479, "step": 17820 }, { "epoch": 0.02, "learning_rate": 4.9961259887568095e-05, "loss": 1.0638, "step": 17821 }, { "epoch": 0.02, "learning_rate": 4.9961255516348e-05, "loss": 1.2964, "step": 17822 }, { "epoch": 0.02, "learning_rate": 4.9961251144881494e-05, "loss": 0.9223, "step": 17823 }, { "epoch": 0.02, "learning_rate": 4.996124677316858e-05, "loss": 1.0178, "step": 17824 }, { "epoch": 0.02, "learning_rate": 4.996124240120927e-05, "loss": 1.1145, "step": 17825 }, { "epoch": 0.02, "learning_rate": 4.996123802900354e-05, "loss": 1.0434, "step": 17826 }, { "epoch": 0.02, "learning_rate": 4.996123365655141e-05, "loss": 0.3403, "step": 17827 }, { "epoch": 0.02, "learning_rate": 4.9961229283852875e-05, "loss": 0.2058, "step": 17828 }, { "epoch": 0.02, "learning_rate": 4.996122491090793e-05, "loss": 0.7896, "step": 17829 }, { "epoch": 0.02, "learning_rate": 4.996122053771658e-05, "loss": 0.5549, "step": 17830 }, { "epoch": 0.02, "learning_rate": 4.9961216164278824e-05, "loss": 0.8353, "step": 17831 }, { "epoch": 0.02, "learning_rate": 4.9961211790594656e-05, "loss": 0.5369, "step": 17832 }, { "epoch": 0.02, "learning_rate": 4.996120741666408e-05, "loss": 0.9177, "step": 17833 }, { "epoch": 0.02, "learning_rate": 4.9961203042487105e-05, "loss": 1.225, "step": 17834 }, { "epoch": 0.02, "learning_rate": 4.996119866806372e-05, "loss": 1.0038, "step": 17835 }, { "epoch": 0.02, "learning_rate": 4.996119429339393e-05, "loss": 0.9841, "step": 17836 }, { "epoch": 0.02, "learning_rate": 4.996118991847774e-05, "loss": 1.1686, "step": 17837 }, { "epoch": 0.02, "learning_rate": 4.996118554331513e-05, "loss": 1.0309, "step": 17838 }, { "epoch": 0.02, "learning_rate": 4.996118116790612e-05, "loss": 1.4359, "step": 17839 }, { "epoch": 0.02, "learning_rate": 4.9961176792250705e-05, "loss": 1.8824, "step": 17840 }, { "epoch": 0.02, "learning_rate": 4.996117241634888e-05, "loss": 1.6606, "step": 17841 }, { "epoch": 0.02, "learning_rate": 4.9961168040200656e-05, "loss": 1.631, "step": 17842 }, { "epoch": 0.02, "learning_rate": 4.996116366380602e-05, "loss": 1.4825, "step": 17843 }, { "epoch": 0.02, "learning_rate": 4.9961159287164986e-05, "loss": 1.5155, "step": 17844 }, { "epoch": 0.02, "learning_rate": 4.996115491027753e-05, "loss": 1.5507, "step": 17845 }, { "epoch": 0.02, "learning_rate": 4.996115053314368e-05, "loss": 1.4561, "step": 17846 }, { "epoch": 0.02, "learning_rate": 4.996114615576342e-05, "loss": 1.4281, "step": 17847 }, { "epoch": 0.02, "learning_rate": 4.9961141778136755e-05, "loss": 1.4336, "step": 17848 }, { "epoch": 0.02, "learning_rate": 4.996113740026369e-05, "loss": 1.4138, "step": 17849 }, { "epoch": 0.02, "learning_rate": 4.9961133022144215e-05, "loss": 1.1119, "step": 17850 }, { "epoch": 0.02, "learning_rate": 4.9961128643778333e-05, "loss": 0.9559, "step": 17851 }, { "epoch": 0.02, "learning_rate": 4.996112426516604e-05, "loss": 1.2736, "step": 17852 }, { "epoch": 0.02, "learning_rate": 4.996111988630735e-05, "loss": 1.2252, "step": 17853 }, { "epoch": 0.02, "learning_rate": 4.996111550720225e-05, "loss": 1.1023, "step": 17854 }, { "epoch": 0.02, "learning_rate": 4.996111112785075e-05, "loss": 1.2164, "step": 17855 }, { "epoch": 0.02, "learning_rate": 4.9961106748252834e-05, "loss": 0.8608, "step": 17856 }, { "epoch": 0.02, "learning_rate": 4.996110236840852e-05, "loss": 0.5434, "step": 17857 }, { "epoch": 0.02, "learning_rate": 4.9961097988317796e-05, "loss": 1.5819, "step": 17858 }, { "epoch": 0.02, "learning_rate": 4.996109360798067e-05, "loss": 1.0247, "step": 17859 }, { "epoch": 0.02, "learning_rate": 4.9961089227397137e-05, "loss": 0.9757, "step": 17860 }, { "epoch": 0.02, "learning_rate": 4.99610848465672e-05, "loss": 1.371, "step": 17861 }, { "epoch": 0.02, "learning_rate": 4.996108046549086e-05, "loss": 1.2974, "step": 17862 }, { "epoch": 0.02, "learning_rate": 4.996107608416811e-05, "loss": 0.3844, "step": 17863 }, { "epoch": 0.02, "learning_rate": 4.996107170259896e-05, "loss": 0.1993, "step": 17864 }, { "epoch": 0.02, "learning_rate": 4.99610673207834e-05, "loss": 0.1216, "step": 17865 }, { "epoch": 0.02, "learning_rate": 4.996106293872144e-05, "loss": 0.1464, "step": 17866 }, { "epoch": 0.02, "learning_rate": 4.996105855641307e-05, "loss": 0.9081, "step": 17867 }, { "epoch": 0.02, "learning_rate": 4.996105417385829e-05, "loss": 1.0491, "step": 17868 }, { "epoch": 0.02, "learning_rate": 4.996104979105711e-05, "loss": 1.1273, "step": 17869 }, { "epoch": 0.02, "learning_rate": 4.996104540800953e-05, "loss": 0.9371, "step": 17870 }, { "epoch": 0.02, "learning_rate": 4.9961041024715544e-05, "loss": 0.9948, "step": 17871 }, { "epoch": 0.02, "learning_rate": 4.996103664117515e-05, "loss": 0.7867, "step": 17872 }, { "epoch": 0.02, "learning_rate": 4.9961032257388355e-05, "loss": 1.2941, "step": 17873 }, { "epoch": 0.02, "learning_rate": 4.9961027873355145e-05, "loss": 1.2699, "step": 17874 }, { "epoch": 0.02, "learning_rate": 4.9961023489075544e-05, "loss": 1.3135, "step": 17875 }, { "epoch": 0.02, "learning_rate": 4.996101910454953e-05, "loss": 1.1298, "step": 17876 }, { "epoch": 0.02, "learning_rate": 4.996101471977711e-05, "loss": 1.4404, "step": 17877 }, { "epoch": 0.02, "learning_rate": 4.996101033475828e-05, "loss": 1.2673, "step": 17878 }, { "epoch": 0.02, "learning_rate": 4.996100594949306e-05, "loss": 0.9231, "step": 17879 }, { "epoch": 0.02, "learning_rate": 4.996100156398143e-05, "loss": 1.3769, "step": 17880 }, { "epoch": 0.02, "learning_rate": 4.996099717822339e-05, "loss": 2.7435, "step": 17881 }, { "epoch": 0.02, "learning_rate": 4.9960992792218955e-05, "loss": 1.3202, "step": 17882 }, { "epoch": 0.02, "learning_rate": 4.996098840596811e-05, "loss": 0.9185, "step": 17883 }, { "epoch": 0.02, "learning_rate": 4.996098401947086e-05, "loss": 0.8772, "step": 17884 }, { "epoch": 0.02, "learning_rate": 4.996097963272721e-05, "loss": 1.2196, "step": 17885 }, { "epoch": 0.02, "learning_rate": 4.996097524573715e-05, "loss": 0.8258, "step": 17886 }, { "epoch": 0.02, "learning_rate": 4.9960970858500685e-05, "loss": 0.8416, "step": 17887 }, { "epoch": 0.02, "learning_rate": 4.996096647101782e-05, "loss": 0.919, "step": 17888 }, { "epoch": 0.02, "learning_rate": 4.9960962083288554e-05, "loss": 1.1087, "step": 17889 }, { "epoch": 0.02, "learning_rate": 4.996095769531288e-05, "loss": 1.0873, "step": 17890 }, { "epoch": 0.02, "learning_rate": 4.996095330709081e-05, "loss": 1.1233, "step": 17891 }, { "epoch": 0.02, "learning_rate": 4.996094891862232e-05, "loss": 0.9123, "step": 17892 }, { "epoch": 0.02, "learning_rate": 4.9960944529907435e-05, "loss": 1.1303, "step": 17893 }, { "epoch": 0.02, "learning_rate": 4.996094014094614e-05, "loss": 1.0402, "step": 17894 }, { "epoch": 0.02, "learning_rate": 4.9960935751738455e-05, "loss": 1.1151, "step": 17895 }, { "epoch": 0.02, "learning_rate": 4.9960931362284356e-05, "loss": 0.8309, "step": 17896 }, { "epoch": 0.02, "learning_rate": 4.996092697258385e-05, "loss": 0.9464, "step": 17897 }, { "epoch": 0.02, "learning_rate": 4.996092258263695e-05, "loss": 1.2149, "step": 17898 }, { "epoch": 0.02, "learning_rate": 4.9960918192443636e-05, "loss": 1.0995, "step": 17899 }, { "epoch": 0.02, "learning_rate": 4.996091380200393e-05, "loss": 1.1807, "step": 17900 }, { "epoch": 0.02, "learning_rate": 4.996090941131781e-05, "loss": 1.1365, "step": 17901 }, { "epoch": 0.02, "learning_rate": 4.996090502038529e-05, "loss": 1.0854, "step": 17902 }, { "epoch": 0.02, "learning_rate": 4.996090062920636e-05, "loss": 0.7762, "step": 17903 }, { "epoch": 0.02, "learning_rate": 4.996089623778104e-05, "loss": 1.0915, "step": 17904 }, { "epoch": 0.02, "learning_rate": 4.9960891846109315e-05, "loss": 1.6023, "step": 17905 }, { "epoch": 0.02, "learning_rate": 4.996088745419117e-05, "loss": 1.9308, "step": 17906 }, { "epoch": 0.02, "learning_rate": 4.996088306202664e-05, "loss": 2.0805, "step": 17907 }, { "epoch": 0.02, "learning_rate": 4.9960878669615694e-05, "loss": 1.5449, "step": 17908 }, { "epoch": 0.02, "learning_rate": 4.996087427695836e-05, "loss": 1.3992, "step": 17909 }, { "epoch": 0.02, "learning_rate": 4.996086988405461e-05, "loss": 1.1294, "step": 17910 }, { "epoch": 0.02, "learning_rate": 4.9960865490904466e-05, "loss": 0.9645, "step": 17911 }, { "epoch": 0.02, "learning_rate": 4.99608610975079e-05, "loss": 1.0826, "step": 17912 }, { "epoch": 0.02, "learning_rate": 4.996085670386495e-05, "loss": 1.1494, "step": 17913 }, { "epoch": 0.02, "learning_rate": 4.996085230997559e-05, "loss": 1.1544, "step": 17914 }, { "epoch": 0.02, "learning_rate": 4.9960847915839825e-05, "loss": 1.3492, "step": 17915 }, { "epoch": 0.02, "learning_rate": 4.996084352145767e-05, "loss": 1.1013, "step": 17916 }, { "epoch": 0.02, "learning_rate": 4.9960839126829094e-05, "loss": 1.0455, "step": 17917 }, { "epoch": 0.02, "learning_rate": 4.996083473195412e-05, "loss": 1.0529, "step": 17918 }, { "epoch": 0.02, "learning_rate": 4.9960830336832755e-05, "loss": 1.4955, "step": 17919 }, { "epoch": 0.02, "learning_rate": 4.996082594146497e-05, "loss": 1.1802, "step": 17920 }, { "epoch": 0.02, "learning_rate": 4.9960821545850795e-05, "loss": 1.0798, "step": 17921 }, { "epoch": 0.02, "learning_rate": 4.9960817149990213e-05, "loss": 0.9555, "step": 17922 }, { "epoch": 0.02, "learning_rate": 4.996081275388323e-05, "loss": 0.9913, "step": 17923 }, { "epoch": 0.02, "learning_rate": 4.996080835752984e-05, "loss": 1.2333, "step": 17924 }, { "epoch": 0.02, "learning_rate": 4.9960803960930045e-05, "loss": 1.1927, "step": 17925 }, { "epoch": 0.02, "learning_rate": 4.996079956408386e-05, "loss": 1.3165, "step": 17926 }, { "epoch": 0.02, "learning_rate": 4.9960795166991255e-05, "loss": 1.1601, "step": 17927 }, { "epoch": 0.02, "learning_rate": 4.996079076965226e-05, "loss": 1.1796, "step": 17928 }, { "epoch": 0.02, "learning_rate": 4.9960786372066865e-05, "loss": 1.1237, "step": 17929 }, { "epoch": 0.02, "learning_rate": 4.996078197423506e-05, "loss": 1.1265, "step": 17930 }, { "epoch": 0.02, "learning_rate": 4.996077757615686e-05, "loss": 1.3149, "step": 17931 }, { "epoch": 0.02, "learning_rate": 4.9960773177832254e-05, "loss": 0.7454, "step": 17932 }, { "epoch": 0.02, "learning_rate": 4.996076877926124e-05, "loss": 0.5543, "step": 17933 }, { "epoch": 0.02, "learning_rate": 4.996076438044383e-05, "loss": 1.411, "step": 17934 }, { "epoch": 0.02, "learning_rate": 4.9960759981380015e-05, "loss": 1.4529, "step": 17935 }, { "epoch": 0.02, "learning_rate": 4.99607555820698e-05, "loss": 1.3448, "step": 17936 }, { "epoch": 0.02, "learning_rate": 4.996075118251318e-05, "loss": 1.0683, "step": 17937 }, { "epoch": 0.02, "learning_rate": 4.996074678271016e-05, "loss": 1.2759, "step": 17938 }, { "epoch": 0.02, "learning_rate": 4.996074238266074e-05, "loss": 1.044, "step": 17939 }, { "epoch": 0.02, "learning_rate": 4.996073798236492e-05, "loss": 1.0703, "step": 17940 }, { "epoch": 0.02, "learning_rate": 4.996073358182269e-05, "loss": 1.0292, "step": 17941 }, { "epoch": 0.02, "learning_rate": 4.9960729181034065e-05, "loss": 1.2431, "step": 17942 }, { "epoch": 0.02, "learning_rate": 4.9960724779999034e-05, "loss": 0.6406, "step": 17943 }, { "epoch": 0.02, "learning_rate": 4.9960720378717605e-05, "loss": 0.2434, "step": 17944 }, { "epoch": 0.02, "learning_rate": 4.996071597718978e-05, "loss": 0.3419, "step": 17945 }, { "epoch": 0.02, "learning_rate": 4.9960711575415544e-05, "loss": 0.2989, "step": 17946 }, { "epoch": 0.02, "learning_rate": 4.9960707173394905e-05, "loss": 0.7781, "step": 17947 }, { "epoch": 0.02, "learning_rate": 4.996070277112787e-05, "loss": 1.0198, "step": 17948 }, { "epoch": 0.02, "learning_rate": 4.9960698368614426e-05, "loss": 1.0838, "step": 17949 }, { "epoch": 0.02, "learning_rate": 4.9960693965854585e-05, "loss": 1.1136, "step": 17950 }, { "epoch": 0.02, "learning_rate": 4.996068956284834e-05, "loss": 1.0603, "step": 17951 }, { "epoch": 0.02, "learning_rate": 4.99606851595957e-05, "loss": 0.9599, "step": 17952 }, { "epoch": 0.02, "learning_rate": 4.996068075609666e-05, "loss": 1.1496, "step": 17953 }, { "epoch": 0.02, "learning_rate": 4.996067635235121e-05, "loss": 1.1231, "step": 17954 }, { "epoch": 0.02, "learning_rate": 4.9960671948359363e-05, "loss": 1.2266, "step": 17955 }, { "epoch": 0.02, "learning_rate": 4.996066754412111e-05, "loss": 1.3298, "step": 17956 }, { "epoch": 0.02, "learning_rate": 4.996066313963647e-05, "loss": 1.2169, "step": 17957 }, { "epoch": 0.02, "learning_rate": 4.996065873490541e-05, "loss": 1.2234, "step": 17958 }, { "epoch": 0.02, "learning_rate": 4.996065432992796e-05, "loss": 1.3403, "step": 17959 }, { "epoch": 0.02, "learning_rate": 4.9960649924704105e-05, "loss": 1.1476, "step": 17960 }, { "epoch": 0.02, "learning_rate": 4.9960645519233853e-05, "loss": 0.8511, "step": 17961 }, { "epoch": 0.02, "learning_rate": 4.99606411135172e-05, "loss": 0.8109, "step": 17962 }, { "epoch": 0.02, "learning_rate": 4.996063670755414e-05, "loss": 0.8822, "step": 17963 }, { "epoch": 0.02, "learning_rate": 4.996063230134469e-05, "loss": 1.1449, "step": 17964 }, { "epoch": 0.02, "learning_rate": 4.996062789488883e-05, "loss": 1.0863, "step": 17965 }, { "epoch": 0.02, "learning_rate": 4.996062348818657e-05, "loss": 1.1397, "step": 17966 }, { "epoch": 0.02, "learning_rate": 4.9960619081237916e-05, "loss": 1.0783, "step": 17967 }, { "epoch": 0.02, "learning_rate": 4.996061467404285e-05, "loss": 1.1457, "step": 17968 }, { "epoch": 0.02, "learning_rate": 4.9960610266601396e-05, "loss": 1.4081, "step": 17969 }, { "epoch": 0.02, "learning_rate": 4.996060585891353e-05, "loss": 1.2906, "step": 17970 }, { "epoch": 0.02, "learning_rate": 4.996060145097927e-05, "loss": 1.1079, "step": 17971 }, { "epoch": 0.02, "learning_rate": 4.9960597042798605e-05, "loss": 1.2476, "step": 17972 }, { "epoch": 0.02, "learning_rate": 4.996059263437154e-05, "loss": 0.9978, "step": 17973 }, { "epoch": 0.02, "learning_rate": 4.996058822569808e-05, "loss": 0.7534, "step": 17974 }, { "epoch": 0.02, "learning_rate": 4.9960583816778215e-05, "loss": 0.8802, "step": 17975 }, { "epoch": 0.02, "learning_rate": 4.996057940761195e-05, "loss": 1.303, "step": 17976 }, { "epoch": 0.02, "learning_rate": 4.9960574998199285e-05, "loss": 1.0353, "step": 17977 }, { "epoch": 0.02, "learning_rate": 4.9960570588540225e-05, "loss": 1.2494, "step": 17978 }, { "epoch": 0.02, "learning_rate": 4.9960566178634754e-05, "loss": 1.0256, "step": 17979 }, { "epoch": 0.02, "learning_rate": 4.99605617684829e-05, "loss": 1.1767, "step": 17980 }, { "epoch": 0.02, "learning_rate": 4.996055735808463e-05, "loss": 1.1519, "step": 17981 }, { "epoch": 0.02, "learning_rate": 4.996055294743996e-05, "loss": 1.0401, "step": 17982 }, { "epoch": 0.02, "learning_rate": 4.996054853654889e-05, "loss": 0.8165, "step": 17983 }, { "epoch": 0.02, "learning_rate": 4.996054412541143e-05, "loss": 1.5519, "step": 17984 }, { "epoch": 0.02, "learning_rate": 4.996053971402757e-05, "loss": 2.0726, "step": 17985 }, { "epoch": 0.02, "learning_rate": 4.99605353023973e-05, "loss": 1.8672, "step": 17986 }, { "epoch": 0.02, "learning_rate": 4.996053089052063e-05, "loss": 0.6705, "step": 17987 }, { "epoch": 0.02, "learning_rate": 4.996052647839757e-05, "loss": 1.5385, "step": 17988 }, { "epoch": 0.02, "learning_rate": 4.9960522066028106e-05, "loss": 0.9687, "step": 17989 }, { "epoch": 0.02, "learning_rate": 4.996051765341224e-05, "loss": 1.1204, "step": 17990 }, { "epoch": 0.02, "learning_rate": 4.996051324054997e-05, "loss": 1.0293, "step": 17991 }, { "epoch": 0.02, "learning_rate": 4.996050882744131e-05, "loss": 0.7258, "step": 17992 }, { "epoch": 0.02, "learning_rate": 4.996050441408625e-05, "loss": 0.6446, "step": 17993 }, { "epoch": 0.02, "learning_rate": 4.9960500000484785e-05, "loss": 0.5994, "step": 17994 }, { "epoch": 0.02, "learning_rate": 4.996049558663692e-05, "loss": 0.6836, "step": 17995 }, { "epoch": 0.02, "learning_rate": 4.996049117254266e-05, "loss": 0.574, "step": 17996 }, { "epoch": 0.02, "learning_rate": 4.9960486758202e-05, "loss": 1.1941, "step": 17997 }, { "epoch": 0.02, "learning_rate": 4.9960482343614935e-05, "loss": 0.984, "step": 17998 }, { "epoch": 0.02, "learning_rate": 4.9960477928781473e-05, "loss": 0.6684, "step": 17999 }, { "epoch": 0.02, "learning_rate": 4.996047351370161e-05, "loss": 0.7643, "step": 18000 }, { "epoch": 0.02, "eval_loss": 1.050679326057434, "eval_runtime": 85.592, "eval_samples_per_second": 16.181, "eval_steps_per_second": 4.054, "step": 18000 }, { "epoch": 0.02, "learning_rate": 4.9960469098375354e-05, "loss": 0.9056, "step": 18001 }, { "epoch": 0.02, "learning_rate": 4.99604646828027e-05, "loss": 1.0514, "step": 18002 }, { "epoch": 0.02, "learning_rate": 4.996046026698364e-05, "loss": 1.2171, "step": 18003 }, { "epoch": 0.02, "learning_rate": 4.996045585091818e-05, "loss": 1.337, "step": 18004 }, { "epoch": 0.02, "learning_rate": 4.996045143460633e-05, "loss": 0.8909, "step": 18005 }, { "epoch": 0.02, "learning_rate": 4.9960447018048075e-05, "loss": 1.1762, "step": 18006 }, { "epoch": 0.02, "learning_rate": 4.996044260124342e-05, "loss": 1.0932, "step": 18007 }, { "epoch": 0.02, "learning_rate": 4.9960438184192364e-05, "loss": 0.9789, "step": 18008 }, { "epoch": 0.02, "learning_rate": 4.996043376689491e-05, "loss": 1.0348, "step": 18009 }, { "epoch": 0.02, "learning_rate": 4.9960429349351065e-05, "loss": 0.9626, "step": 18010 }, { "epoch": 0.02, "learning_rate": 4.9960424931560814e-05, "loss": 1.0603, "step": 18011 }, { "epoch": 0.02, "learning_rate": 4.9960420513524165e-05, "loss": 1.0675, "step": 18012 }, { "epoch": 0.02, "learning_rate": 4.996041609524112e-05, "loss": 1.2027, "step": 18013 }, { "epoch": 0.02, "learning_rate": 4.996041167671167e-05, "loss": 0.9694, "step": 18014 }, { "epoch": 0.02, "learning_rate": 4.9960407257935826e-05, "loss": 1.0229, "step": 18015 }, { "epoch": 0.02, "learning_rate": 4.996040283891358e-05, "loss": 0.6471, "step": 18016 }, { "epoch": 0.02, "learning_rate": 4.9960398419644935e-05, "loss": 0.8391, "step": 18017 }, { "epoch": 0.02, "learning_rate": 4.99603940001299e-05, "loss": 1.4049, "step": 18018 }, { "epoch": 0.02, "learning_rate": 4.996038958036846e-05, "loss": 0.9414, "step": 18019 }, { "epoch": 0.02, "learning_rate": 4.9960385160360626e-05, "loss": 0.6825, "step": 18020 }, { "epoch": 0.02, "learning_rate": 4.99603807401064e-05, "loss": 1.7898, "step": 18021 }, { "epoch": 0.02, "learning_rate": 4.9960376319605756e-05, "loss": 1.1204, "step": 18022 }, { "epoch": 0.02, "learning_rate": 4.9960371898858723e-05, "loss": 1.0597, "step": 18023 }, { "epoch": 0.02, "learning_rate": 4.996036747786529e-05, "loss": 1.0251, "step": 18024 }, { "epoch": 0.02, "learning_rate": 4.996036305662546e-05, "loss": 1.3625, "step": 18025 }, { "epoch": 0.02, "learning_rate": 4.9960358635139235e-05, "loss": 1.1888, "step": 18026 }, { "epoch": 0.02, "learning_rate": 4.9960354213406615e-05, "loss": 1.148, "step": 18027 }, { "epoch": 0.02, "learning_rate": 4.996034979142759e-05, "loss": 0.4998, "step": 18028 }, { "epoch": 0.02, "learning_rate": 4.9960345369202166e-05, "loss": 0.6164, "step": 18029 }, { "epoch": 0.02, "learning_rate": 4.996034094673035e-05, "loss": 0.6197, "step": 18030 }, { "epoch": 0.02, "learning_rate": 4.996033652401213e-05, "loss": 1.141, "step": 18031 }, { "epoch": 0.02, "learning_rate": 4.996033210104752e-05, "loss": 1.2347, "step": 18032 }, { "epoch": 0.02, "learning_rate": 4.996032767783651e-05, "loss": 0.958, "step": 18033 }, { "epoch": 0.02, "learning_rate": 4.996032325437909e-05, "loss": 0.6896, "step": 18034 }, { "epoch": 0.02, "learning_rate": 4.9960318830675283e-05, "loss": 1.3073, "step": 18035 }, { "epoch": 0.02, "learning_rate": 4.996031440672508e-05, "loss": 1.0048, "step": 18036 }, { "epoch": 0.02, "learning_rate": 4.996030998252847e-05, "loss": 1.022, "step": 18037 }, { "epoch": 0.02, "learning_rate": 4.9960305558085476e-05, "loss": 1.2817, "step": 18038 }, { "epoch": 0.02, "learning_rate": 4.996030113339607e-05, "loss": 1.0414, "step": 18039 }, { "epoch": 0.02, "learning_rate": 4.9960296708460274e-05, "loss": 1.2339, "step": 18040 }, { "epoch": 0.02, "learning_rate": 4.996029228327808e-05, "loss": 1.0857, "step": 18041 }, { "epoch": 0.02, "learning_rate": 4.996028785784949e-05, "loss": 1.0162, "step": 18042 }, { "epoch": 0.02, "learning_rate": 4.99602834321745e-05, "loss": 1.015, "step": 18043 }, { "epoch": 0.02, "learning_rate": 4.996027900625312e-05, "loss": 0.863, "step": 18044 }, { "epoch": 0.02, "learning_rate": 4.996027458008533e-05, "loss": 1.035, "step": 18045 }, { "epoch": 0.02, "learning_rate": 4.996027015367115e-05, "loss": 0.9516, "step": 18046 }, { "epoch": 0.02, "learning_rate": 4.996026572701057e-05, "loss": 1.1389, "step": 18047 }, { "epoch": 0.02, "learning_rate": 4.9960261300103594e-05, "loss": 1.1366, "step": 18048 }, { "epoch": 0.02, "learning_rate": 4.996025687295022e-05, "loss": 0.784, "step": 18049 }, { "epoch": 0.02, "learning_rate": 4.996025244555045e-05, "loss": 1.0802, "step": 18050 }, { "epoch": 0.02, "learning_rate": 4.9960248017904285e-05, "loss": 0.9317, "step": 18051 }, { "epoch": 0.02, "learning_rate": 4.996024359001172e-05, "loss": 1.2861, "step": 18052 }, { "epoch": 0.02, "learning_rate": 4.9960239161872756e-05, "loss": 1.1905, "step": 18053 }, { "epoch": 0.02, "learning_rate": 4.99602347334874e-05, "loss": 1.2289, "step": 18054 }, { "epoch": 0.02, "learning_rate": 4.996023030485565e-05, "loss": 1.1468, "step": 18055 }, { "epoch": 0.02, "learning_rate": 4.99602258759775e-05, "loss": 1.1557, "step": 18056 }, { "epoch": 0.02, "learning_rate": 4.996022144685294e-05, "loss": 1.2146, "step": 18057 }, { "epoch": 0.02, "learning_rate": 4.9960217017482e-05, "loss": 1.0674, "step": 18058 }, { "epoch": 0.02, "learning_rate": 4.996021258786465e-05, "loss": 1.1627, "step": 18059 }, { "epoch": 0.02, "learning_rate": 4.9960208158000916e-05, "loss": 0.9766, "step": 18060 }, { "epoch": 0.02, "learning_rate": 4.996020372789078e-05, "loss": 0.9162, "step": 18061 }, { "epoch": 0.02, "learning_rate": 4.996019929753425e-05, "loss": 0.7696, "step": 18062 }, { "epoch": 0.02, "learning_rate": 4.996019486693132e-05, "loss": 1.1892, "step": 18063 }, { "epoch": 0.02, "learning_rate": 4.996019043608199e-05, "loss": 1.4385, "step": 18064 }, { "epoch": 0.02, "learning_rate": 4.996018600498627e-05, "loss": 1.0275, "step": 18065 }, { "epoch": 0.02, "learning_rate": 4.996018157364415e-05, "loss": 0.5482, "step": 18066 }, { "epoch": 0.02, "learning_rate": 4.996017714205564e-05, "loss": 0.4382, "step": 18067 }, { "epoch": 0.02, "learning_rate": 4.996017271022073e-05, "loss": 0.6151, "step": 18068 }, { "epoch": 0.02, "learning_rate": 4.996016827813942e-05, "loss": 0.4725, "step": 18069 }, { "epoch": 0.02, "learning_rate": 4.996016384581171e-05, "loss": 0.4936, "step": 18070 }, { "epoch": 0.02, "learning_rate": 4.996015941323762e-05, "loss": 0.3741, "step": 18071 }, { "epoch": 0.02, "learning_rate": 4.996015498041712e-05, "loss": 0.2599, "step": 18072 }, { "epoch": 0.02, "learning_rate": 4.9960150547350224e-05, "loss": 0.2654, "step": 18073 }, { "epoch": 0.02, "learning_rate": 4.9960146114036936e-05, "loss": 1.0072, "step": 18074 }, { "epoch": 0.02, "learning_rate": 4.9960141680477256e-05, "loss": 1.0779, "step": 18075 }, { "epoch": 0.02, "learning_rate": 4.996013724667117e-05, "loss": 1.2817, "step": 18076 }, { "epoch": 0.02, "learning_rate": 4.99601328126187e-05, "loss": 1.3007, "step": 18077 }, { "epoch": 0.02, "learning_rate": 4.9960128378319825e-05, "loss": 1.0265, "step": 18078 }, { "epoch": 0.02, "learning_rate": 4.996012394377456e-05, "loss": 1.1672, "step": 18079 }, { "epoch": 0.02, "learning_rate": 4.996011950898289e-05, "loss": 0.8514, "step": 18080 }, { "epoch": 0.02, "learning_rate": 4.9960115073944835e-05, "loss": 1.0394, "step": 18081 }, { "epoch": 0.02, "learning_rate": 4.996011063866037e-05, "loss": 1.0375, "step": 18082 }, { "epoch": 0.02, "learning_rate": 4.9960106203129525e-05, "loss": 1.1931, "step": 18083 }, { "epoch": 0.02, "learning_rate": 4.996010176735227e-05, "loss": 1.1129, "step": 18084 }, { "epoch": 0.02, "learning_rate": 4.9960097331328635e-05, "loss": 1.2561, "step": 18085 }, { "epoch": 0.02, "learning_rate": 4.99600928950586e-05, "loss": 1.076, "step": 18086 }, { "epoch": 0.02, "learning_rate": 4.9960088458542164e-05, "loss": 1.116, "step": 18087 }, { "epoch": 0.02, "learning_rate": 4.996008402177934e-05, "loss": 1.1238, "step": 18088 }, { "epoch": 0.02, "learning_rate": 4.996007958477011e-05, "loss": 1.007, "step": 18089 }, { "epoch": 0.02, "learning_rate": 4.996007514751449e-05, "loss": 1.0495, "step": 18090 }, { "epoch": 0.02, "learning_rate": 4.9960070710012474e-05, "loss": 1.1175, "step": 18091 }, { "epoch": 0.02, "learning_rate": 4.996006627226406e-05, "loss": 1.0128, "step": 18092 }, { "epoch": 0.02, "learning_rate": 4.9960061834269255e-05, "loss": 1.5412, "step": 18093 }, { "epoch": 0.02, "learning_rate": 4.996005739602806e-05, "loss": 1.4803, "step": 18094 }, { "epoch": 0.02, "learning_rate": 4.996005295754046e-05, "loss": 0.8497, "step": 18095 }, { "epoch": 0.02, "learning_rate": 4.996004851880647e-05, "loss": 1.0703, "step": 18096 }, { "epoch": 0.02, "learning_rate": 4.996004407982608e-05, "loss": 1.214, "step": 18097 }, { "epoch": 0.02, "learning_rate": 4.99600396405993e-05, "loss": 1.089, "step": 18098 }, { "epoch": 0.02, "learning_rate": 4.996003520112613e-05, "loss": 1.1132, "step": 18099 }, { "epoch": 0.02, "learning_rate": 4.9960030761406554e-05, "loss": 1.1375, "step": 18100 }, { "epoch": 0.02, "learning_rate": 4.996002632144059e-05, "loss": 1.0878, "step": 18101 }, { "epoch": 0.02, "learning_rate": 4.996002188122822e-05, "loss": 0.8723, "step": 18102 }, { "epoch": 0.02, "learning_rate": 4.996001744076947e-05, "loss": 0.9465, "step": 18103 }, { "epoch": 0.02, "learning_rate": 4.996001300006432e-05, "loss": 1.0728, "step": 18104 }, { "epoch": 0.02, "learning_rate": 4.996000855911278e-05, "loss": 1.191, "step": 18105 }, { "epoch": 0.02, "learning_rate": 4.996000411791484e-05, "loss": 1.0594, "step": 18106 }, { "epoch": 0.02, "learning_rate": 4.9959999676470503e-05, "loss": 1.0284, "step": 18107 }, { "epoch": 0.02, "learning_rate": 4.9959995234779775e-05, "loss": 0.7396, "step": 18108 }, { "epoch": 0.02, "learning_rate": 4.995999079284265e-05, "loss": 0.8808, "step": 18109 }, { "epoch": 0.02, "learning_rate": 4.995998635065913e-05, "loss": 0.8918, "step": 18110 }, { "epoch": 0.02, "learning_rate": 4.995998190822921e-05, "loss": 1.0108, "step": 18111 }, { "epoch": 0.02, "learning_rate": 4.995997746555291e-05, "loss": 0.8671, "step": 18112 }, { "epoch": 0.02, "learning_rate": 4.99599730226302e-05, "loss": 0.7689, "step": 18113 }, { "epoch": 0.02, "learning_rate": 4.9959968579461116e-05, "loss": 0.8836, "step": 18114 }, { "epoch": 0.02, "learning_rate": 4.9959964136045625e-05, "loss": 0.9026, "step": 18115 }, { "epoch": 0.02, "learning_rate": 4.9959959692383736e-05, "loss": 1.1428, "step": 18116 }, { "epoch": 0.02, "learning_rate": 4.9959955248475455e-05, "loss": 1.1471, "step": 18117 }, { "epoch": 0.02, "learning_rate": 4.995995080432078e-05, "loss": 1.1529, "step": 18118 }, { "epoch": 0.02, "learning_rate": 4.995994635991972e-05, "loss": 1.0676, "step": 18119 }, { "epoch": 0.02, "learning_rate": 4.9959941915272254e-05, "loss": 1.1613, "step": 18120 }, { "epoch": 0.02, "learning_rate": 4.99599374703784e-05, "loss": 1.4425, "step": 18121 }, { "epoch": 0.02, "learning_rate": 4.9959933025238146e-05, "loss": 1.2366, "step": 18122 }, { "epoch": 0.02, "learning_rate": 4.9959928579851514e-05, "loss": 0.8954, "step": 18123 }, { "epoch": 0.02, "learning_rate": 4.995992413421847e-05, "loss": 0.9595, "step": 18124 }, { "epoch": 0.02, "learning_rate": 4.995991968833904e-05, "loss": 1.1951, "step": 18125 }, { "epoch": 0.02, "learning_rate": 4.9959915242213215e-05, "loss": 0.9885, "step": 18126 }, { "epoch": 0.02, "learning_rate": 4.9959910795840996e-05, "loss": 1.1215, "step": 18127 }, { "epoch": 0.02, "learning_rate": 4.9959906349222386e-05, "loss": 1.0994, "step": 18128 }, { "epoch": 0.02, "learning_rate": 4.995990190235738e-05, "loss": 1.4625, "step": 18129 }, { "epoch": 0.02, "learning_rate": 4.9959897455245976e-05, "loss": 1.1744, "step": 18130 }, { "epoch": 0.02, "learning_rate": 4.9959893007888184e-05, "loss": 1.0358, "step": 18131 }, { "epoch": 0.02, "learning_rate": 4.995988856028399e-05, "loss": 0.9786, "step": 18132 }, { "epoch": 0.02, "learning_rate": 4.995988411243342e-05, "loss": 1.3601, "step": 18133 }, { "epoch": 0.02, "learning_rate": 4.995987966433644e-05, "loss": 1.1878, "step": 18134 }, { "epoch": 0.02, "learning_rate": 4.995987521599308e-05, "loss": 1.212, "step": 18135 }, { "epoch": 0.02, "learning_rate": 4.995987076740332e-05, "loss": 1.003, "step": 18136 }, { "epoch": 0.02, "learning_rate": 4.995986631856716e-05, "loss": 1.0741, "step": 18137 }, { "epoch": 0.02, "learning_rate": 4.9959861869484615e-05, "loss": 1.0701, "step": 18138 }, { "epoch": 0.02, "learning_rate": 4.9959857420155675e-05, "loss": 1.0748, "step": 18139 }, { "epoch": 0.02, "learning_rate": 4.9959852970580344e-05, "loss": 1.1109, "step": 18140 }, { "epoch": 0.02, "learning_rate": 4.9959848520758614e-05, "loss": 0.9822, "step": 18141 }, { "epoch": 0.02, "learning_rate": 4.99598440706905e-05, "loss": 1.013, "step": 18142 }, { "epoch": 0.02, "learning_rate": 4.995983962037598e-05, "loss": 1.1869, "step": 18143 }, { "epoch": 0.02, "learning_rate": 4.995983516981508e-05, "loss": 1.1398, "step": 18144 }, { "epoch": 0.02, "learning_rate": 4.9959830719007784e-05, "loss": 1.3199, "step": 18145 }, { "epoch": 0.02, "learning_rate": 4.995982626795409e-05, "loss": 1.2738, "step": 18146 }, { "epoch": 0.02, "learning_rate": 4.995982181665401e-05, "loss": 1.2449, "step": 18147 }, { "epoch": 0.02, "learning_rate": 4.995981736510753e-05, "loss": 0.9902, "step": 18148 }, { "epoch": 0.02, "learning_rate": 4.995981291331466e-05, "loss": 1.2013, "step": 18149 }, { "epoch": 0.02, "learning_rate": 4.995980846127539e-05, "loss": 1.0507, "step": 18150 }, { "epoch": 0.02, "learning_rate": 4.9959804008989744e-05, "loss": 1.3328, "step": 18151 }, { "epoch": 0.02, "learning_rate": 4.9959799556457684e-05, "loss": 1.3154, "step": 18152 }, { "epoch": 0.02, "learning_rate": 4.9959795103679254e-05, "loss": 1.0196, "step": 18153 }, { "epoch": 0.02, "learning_rate": 4.995979065065441e-05, "loss": 1.1392, "step": 18154 }, { "epoch": 0.02, "learning_rate": 4.995978619738319e-05, "loss": 1.0309, "step": 18155 }, { "epoch": 0.02, "learning_rate": 4.995978174386557e-05, "loss": 1.3264, "step": 18156 }, { "epoch": 0.02, "learning_rate": 4.995977729010156e-05, "loss": 1.0503, "step": 18157 }, { "epoch": 0.02, "learning_rate": 4.995977283609116e-05, "loss": 0.8997, "step": 18158 }, { "epoch": 0.02, "learning_rate": 4.9959768381834356e-05, "loss": 1.0872, "step": 18159 }, { "epoch": 0.02, "learning_rate": 4.995976392733117e-05, "loss": 1.1339, "step": 18160 }, { "epoch": 0.02, "learning_rate": 4.995975947258159e-05, "loss": 1.069, "step": 18161 }, { "epoch": 0.02, "learning_rate": 4.9959755017585616e-05, "loss": 1.1728, "step": 18162 }, { "epoch": 0.02, "learning_rate": 4.9959750562343255e-05, "loss": 1.607, "step": 18163 }, { "epoch": 0.02, "learning_rate": 4.99597461068545e-05, "loss": 0.8533, "step": 18164 }, { "epoch": 0.02, "learning_rate": 4.995974165111935e-05, "loss": 1.2901, "step": 18165 }, { "epoch": 0.02, "learning_rate": 4.995973719513781e-05, "loss": 1.2426, "step": 18166 }, { "epoch": 0.02, "learning_rate": 4.995973273890988e-05, "loss": 1.3435, "step": 18167 }, { "epoch": 0.02, "learning_rate": 4.9959728282435546e-05, "loss": 1.2354, "step": 18168 }, { "epoch": 0.02, "learning_rate": 4.9959723825714834e-05, "loss": 0.6667, "step": 18169 }, { "epoch": 0.02, "learning_rate": 4.9959719368747724e-05, "loss": 1.2161, "step": 18170 }, { "epoch": 0.02, "learning_rate": 4.995971491153422e-05, "loss": 1.2082, "step": 18171 }, { "epoch": 0.02, "learning_rate": 4.9959710454074335e-05, "loss": 1.1846, "step": 18172 }, { "epoch": 0.02, "learning_rate": 4.995970599636804e-05, "loss": 1.3826, "step": 18173 }, { "epoch": 0.02, "learning_rate": 4.995970153841537e-05, "loss": 1.3235, "step": 18174 }, { "epoch": 0.02, "learning_rate": 4.9959697080216304e-05, "loss": 0.7302, "step": 18175 }, { "epoch": 0.02, "learning_rate": 4.995969262177085e-05, "loss": 0.9685, "step": 18176 }, { "epoch": 0.02, "learning_rate": 4.995968816307899e-05, "loss": 1.1002, "step": 18177 }, { "epoch": 0.02, "learning_rate": 4.995968370414075e-05, "loss": 0.9935, "step": 18178 }, { "epoch": 0.02, "learning_rate": 4.995967924495611e-05, "loss": 1.0348, "step": 18179 }, { "epoch": 0.02, "learning_rate": 4.995967478552509e-05, "loss": 1.0308, "step": 18180 }, { "epoch": 0.02, "learning_rate": 4.995967032584767e-05, "loss": 1.0874, "step": 18181 }, { "epoch": 0.02, "learning_rate": 4.995966586592387e-05, "loss": 0.8182, "step": 18182 }, { "epoch": 0.02, "learning_rate": 4.9959661405753666e-05, "loss": 0.6662, "step": 18183 }, { "epoch": 0.02, "learning_rate": 4.995965694533707e-05, "loss": 1.0945, "step": 18184 }, { "epoch": 0.02, "learning_rate": 4.99596524846741e-05, "loss": 1.2955, "step": 18185 }, { "epoch": 0.02, "learning_rate": 4.995964802376473e-05, "loss": 1.2857, "step": 18186 }, { "epoch": 0.02, "learning_rate": 4.995964356260896e-05, "loss": 1.087, "step": 18187 }, { "epoch": 0.02, "learning_rate": 4.99596391012068e-05, "loss": 1.0841, "step": 18188 }, { "epoch": 0.02, "learning_rate": 4.995963463955826e-05, "loss": 1.1765, "step": 18189 }, { "epoch": 0.02, "learning_rate": 4.995963017766332e-05, "loss": 1.2735, "step": 18190 }, { "epoch": 0.02, "learning_rate": 4.9959625715522e-05, "loss": 1.0869, "step": 18191 }, { "epoch": 0.02, "learning_rate": 4.995962125313428e-05, "loss": 1.2522, "step": 18192 }, { "epoch": 0.02, "learning_rate": 4.9959616790500165e-05, "loss": 1.2131, "step": 18193 }, { "epoch": 0.02, "learning_rate": 4.995961232761967e-05, "loss": 1.1952, "step": 18194 }, { "epoch": 0.02, "learning_rate": 4.995960786449278e-05, "loss": 0.8877, "step": 18195 }, { "epoch": 0.02, "learning_rate": 4.995960340111949e-05, "loss": 1.2954, "step": 18196 }, { "epoch": 0.02, "learning_rate": 4.9959598937499816e-05, "loss": 1.1065, "step": 18197 }, { "epoch": 0.02, "learning_rate": 4.995959447363375e-05, "loss": 1.0998, "step": 18198 }, { "epoch": 0.02, "learning_rate": 4.99595900095213e-05, "loss": 0.5373, "step": 18199 }, { "epoch": 0.02, "learning_rate": 4.9959585545162454e-05, "loss": 1.0371, "step": 18200 }, { "epoch": 0.02, "learning_rate": 4.995958108055722e-05, "loss": 0.8803, "step": 18201 }, { "epoch": 0.02, "learning_rate": 4.9959576615705596e-05, "loss": 1.131, "step": 18202 }, { "epoch": 0.02, "learning_rate": 4.995957215060758e-05, "loss": 0.8677, "step": 18203 }, { "epoch": 0.02, "learning_rate": 4.995956768526318e-05, "loss": 1.2961, "step": 18204 }, { "epoch": 0.02, "learning_rate": 4.995956321967238e-05, "loss": 1.0677, "step": 18205 }, { "epoch": 0.02, "learning_rate": 4.9959558753835186e-05, "loss": 1.2602, "step": 18206 }, { "epoch": 0.02, "learning_rate": 4.995955428775161e-05, "loss": 1.3877, "step": 18207 }, { "epoch": 0.02, "learning_rate": 4.995954982142165e-05, "loss": 1.1169, "step": 18208 }, { "epoch": 0.02, "learning_rate": 4.995954535484528e-05, "loss": 0.8725, "step": 18209 }, { "epoch": 0.02, "learning_rate": 4.9959540888022536e-05, "loss": 1.2329, "step": 18210 }, { "epoch": 0.02, "learning_rate": 4.99595364209534e-05, "loss": 1.1446, "step": 18211 }, { "epoch": 0.02, "learning_rate": 4.995953195363787e-05, "loss": 0.7766, "step": 18212 }, { "epoch": 0.02, "learning_rate": 4.995952748607595e-05, "loss": 0.5956, "step": 18213 }, { "epoch": 0.02, "learning_rate": 4.995952301826765e-05, "loss": 0.3396, "step": 18214 }, { "epoch": 0.02, "learning_rate": 4.995951855021295e-05, "loss": 0.5136, "step": 18215 }, { "epoch": 0.02, "learning_rate": 4.995951408191186e-05, "loss": 1.4237, "step": 18216 }, { "epoch": 0.02, "learning_rate": 4.995950961336439e-05, "loss": 1.7097, "step": 18217 }, { "epoch": 0.02, "learning_rate": 4.9959505144570515e-05, "loss": 1.4086, "step": 18218 }, { "epoch": 0.02, "learning_rate": 4.995950067553026e-05, "loss": 1.1448, "step": 18219 }, { "epoch": 0.02, "learning_rate": 4.9959496206243616e-05, "loss": 0.9727, "step": 18220 }, { "epoch": 0.02, "learning_rate": 4.9959491736710576e-05, "loss": 1.3568, "step": 18221 }, { "epoch": 0.02, "learning_rate": 4.995948726693115e-05, "loss": 1.5732, "step": 18222 }, { "epoch": 0.02, "learning_rate": 4.995948279690534e-05, "loss": 1.6105, "step": 18223 }, { "epoch": 0.02, "learning_rate": 4.995947832663314e-05, "loss": 1.3131, "step": 18224 }, { "epoch": 0.02, "learning_rate": 4.995947385611454e-05, "loss": 0.9889, "step": 18225 }, { "epoch": 0.02, "learning_rate": 4.995946938534955e-05, "loss": 0.9401, "step": 18226 }, { "epoch": 0.02, "learning_rate": 4.995946491433818e-05, "loss": 1.1841, "step": 18227 }, { "epoch": 0.02, "learning_rate": 4.995946044308042e-05, "loss": 1.1538, "step": 18228 }, { "epoch": 0.02, "learning_rate": 4.9959455971576266e-05, "loss": 0.924, "step": 18229 }, { "epoch": 0.02, "learning_rate": 4.995945149982573e-05, "loss": 0.9382, "step": 18230 }, { "epoch": 0.02, "learning_rate": 4.9959447027828797e-05, "loss": 1.1172, "step": 18231 }, { "epoch": 0.02, "learning_rate": 4.995944255558548e-05, "loss": 1.0739, "step": 18232 }, { "epoch": 0.02, "learning_rate": 4.995943808309577e-05, "loss": 1.1412, "step": 18233 }, { "epoch": 0.02, "learning_rate": 4.9959433610359675e-05, "loss": 1.149, "step": 18234 }, { "epoch": 0.02, "learning_rate": 4.995942913737719e-05, "loss": 0.8393, "step": 18235 }, { "epoch": 0.02, "learning_rate": 4.9959424664148316e-05, "loss": 1.2145, "step": 18236 }, { "epoch": 0.02, "learning_rate": 4.995942019067305e-05, "loss": 1.0294, "step": 18237 }, { "epoch": 0.02, "learning_rate": 4.9959415716951396e-05, "loss": 1.3151, "step": 18238 }, { "epoch": 0.02, "learning_rate": 4.995941124298336e-05, "loss": 1.1017, "step": 18239 }, { "epoch": 0.02, "learning_rate": 4.9959406768768924e-05, "loss": 1.1733, "step": 18240 }, { "epoch": 0.02, "learning_rate": 4.99594022943081e-05, "loss": 1.1012, "step": 18241 }, { "epoch": 0.02, "learning_rate": 4.99593978196009e-05, "loss": 1.0657, "step": 18242 }, { "epoch": 0.02, "learning_rate": 4.99593933446473e-05, "loss": 0.996, "step": 18243 }, { "epoch": 0.02, "learning_rate": 4.995938886944731e-05, "loss": 1.5859, "step": 18244 }, { "epoch": 0.02, "learning_rate": 4.9959384394000944e-05, "loss": 2.5835, "step": 18245 }, { "epoch": 0.02, "learning_rate": 4.995937991830817e-05, "loss": 1.9406, "step": 18246 }, { "epoch": 0.02, "learning_rate": 4.995937544236903e-05, "loss": 1.8655, "step": 18247 }, { "epoch": 0.02, "learning_rate": 4.9959370966183486e-05, "loss": 0.9834, "step": 18248 }, { "epoch": 0.02, "learning_rate": 4.995936648975156e-05, "loss": 1.2702, "step": 18249 }, { "epoch": 0.02, "learning_rate": 4.995936201307324e-05, "loss": 1.4034, "step": 18250 }, { "epoch": 0.02, "learning_rate": 4.9959357536148536e-05, "loss": 1.0864, "step": 18251 }, { "epoch": 0.02, "learning_rate": 4.995935305897744e-05, "loss": 0.9772, "step": 18252 }, { "epoch": 0.02, "learning_rate": 4.995934858155996e-05, "loss": 0.6886, "step": 18253 }, { "epoch": 0.02, "learning_rate": 4.9959344103896094e-05, "loss": 1.026, "step": 18254 }, { "epoch": 0.02, "learning_rate": 4.995933962598584e-05, "loss": 1.0624, "step": 18255 }, { "epoch": 0.02, "learning_rate": 4.995933514782919e-05, "loss": 1.1887, "step": 18256 }, { "epoch": 0.02, "learning_rate": 4.995933066942616e-05, "loss": 1.3504, "step": 18257 }, { "epoch": 0.02, "learning_rate": 4.9959326190776735e-05, "loss": 0.7912, "step": 18258 }, { "epoch": 0.02, "learning_rate": 4.9959321711880924e-05, "loss": 1.21, "step": 18259 }, { "epoch": 0.02, "learning_rate": 4.9959317232738736e-05, "loss": 1.2832, "step": 18260 }, { "epoch": 0.02, "learning_rate": 4.995931275335015e-05, "loss": 0.7327, "step": 18261 }, { "epoch": 0.02, "learning_rate": 4.995930827371517e-05, "loss": 0.9154, "step": 18262 }, { "epoch": 0.02, "learning_rate": 4.995930379383381e-05, "loss": 1.2694, "step": 18263 }, { "epoch": 0.02, "learning_rate": 4.9959299313706064e-05, "loss": 0.9726, "step": 18264 }, { "epoch": 0.02, "learning_rate": 4.995929483333193e-05, "loss": 1.0436, "step": 18265 }, { "epoch": 0.02, "learning_rate": 4.9959290352711404e-05, "loss": 0.9869, "step": 18266 }, { "epoch": 0.02, "learning_rate": 4.9959285871844494e-05, "loss": 1.1556, "step": 18267 }, { "epoch": 0.02, "learning_rate": 4.995928139073119e-05, "loss": 1.0325, "step": 18268 }, { "epoch": 0.02, "learning_rate": 4.995927690937151e-05, "loss": 1.0499, "step": 18269 }, { "epoch": 0.02, "learning_rate": 4.995927242776543e-05, "loss": 1.1624, "step": 18270 }, { "epoch": 0.02, "learning_rate": 4.995926794591297e-05, "loss": 1.084, "step": 18271 }, { "epoch": 0.02, "learning_rate": 4.995926346381412e-05, "loss": 1.0882, "step": 18272 }, { "epoch": 0.02, "learning_rate": 4.995925898146889e-05, "loss": 1.183, "step": 18273 }, { "epoch": 0.02, "learning_rate": 4.995925449887726e-05, "loss": 1.0154, "step": 18274 }, { "epoch": 0.02, "learning_rate": 4.995925001603926e-05, "loss": 1.0948, "step": 18275 }, { "epoch": 0.02, "learning_rate": 4.9959245532954856e-05, "loss": 1.0534, "step": 18276 }, { "epoch": 0.02, "learning_rate": 4.9959241049624076e-05, "loss": 0.9524, "step": 18277 }, { "epoch": 0.02, "learning_rate": 4.99592365660469e-05, "loss": 0.6684, "step": 18278 }, { "epoch": 0.02, "learning_rate": 4.995923208222334e-05, "loss": 1.1587, "step": 18279 }, { "epoch": 0.02, "learning_rate": 4.995922759815339e-05, "loss": 0.8138, "step": 18280 }, { "epoch": 0.02, "learning_rate": 4.9959223113837065e-05, "loss": 0.951, "step": 18281 }, { "epoch": 0.02, "learning_rate": 4.995921862927434e-05, "loss": 1.586, "step": 18282 }, { "epoch": 0.02, "learning_rate": 4.9959214144465236e-05, "loss": 0.6946, "step": 18283 }, { "epoch": 0.02, "learning_rate": 4.995920965940975e-05, "loss": 0.8482, "step": 18284 }, { "epoch": 0.02, "learning_rate": 4.995920517410786e-05, "loss": 0.8691, "step": 18285 }, { "epoch": 0.02, "learning_rate": 4.9959200688559596e-05, "loss": 0.9515, "step": 18286 }, { "epoch": 0.02, "learning_rate": 4.995919620276493e-05, "loss": 1.0254, "step": 18287 }, { "epoch": 0.02, "learning_rate": 4.99591917167239e-05, "loss": 1.084, "step": 18288 }, { "epoch": 0.02, "learning_rate": 4.995918723043647e-05, "loss": 1.1622, "step": 18289 }, { "epoch": 0.02, "learning_rate": 4.995918274390265e-05, "loss": 1.2231, "step": 18290 }, { "epoch": 0.02, "learning_rate": 4.9959178257122456e-05, "loss": 1.2836, "step": 18291 }, { "epoch": 0.02, "learning_rate": 4.995917377009587e-05, "loss": 1.1115, "step": 18292 }, { "epoch": 0.02, "learning_rate": 4.9959169282822895e-05, "loss": 1.2112, "step": 18293 }, { "epoch": 0.02, "learning_rate": 4.9959164795303536e-05, "loss": 1.2341, "step": 18294 }, { "epoch": 0.02, "learning_rate": 4.9959160307537787e-05, "loss": 1.1189, "step": 18295 }, { "epoch": 0.02, "learning_rate": 4.995915581952566e-05, "loss": 1.2342, "step": 18296 }, { "epoch": 0.02, "learning_rate": 4.995915133126714e-05, "loss": 1.0963, "step": 18297 }, { "epoch": 0.02, "learning_rate": 4.995914684276224e-05, "loss": 1.3317, "step": 18298 }, { "epoch": 0.02, "learning_rate": 4.9959142354010945e-05, "loss": 1.0509, "step": 18299 }, { "epoch": 0.02, "learning_rate": 4.995913786501327e-05, "loss": 1.0397, "step": 18300 }, { "epoch": 0.02, "learning_rate": 4.9959133375769204e-05, "loss": 1.0481, "step": 18301 }, { "epoch": 0.02, "learning_rate": 4.995912888627875e-05, "loss": 0.4182, "step": 18302 }, { "epoch": 0.02, "learning_rate": 4.995912439654192e-05, "loss": 0.3065, "step": 18303 }, { "epoch": 0.02, "learning_rate": 4.99591199065587e-05, "loss": 0.2723, "step": 18304 }, { "epoch": 0.02, "learning_rate": 4.9959115416329096e-05, "loss": 0.237, "step": 18305 }, { "epoch": 0.02, "learning_rate": 4.99591109258531e-05, "loss": 0.1672, "step": 18306 }, { "epoch": 0.02, "learning_rate": 4.995910643513072e-05, "loss": 0.1905, "step": 18307 }, { "epoch": 0.02, "learning_rate": 4.9959101944161955e-05, "loss": 0.183, "step": 18308 }, { "epoch": 0.02, "learning_rate": 4.99590974529468e-05, "loss": 0.1746, "step": 18309 }, { "epoch": 0.02, "learning_rate": 4.9959092961485264e-05, "loss": 0.1514, "step": 18310 }, { "epoch": 0.02, "learning_rate": 4.995908846977735e-05, "loss": 0.1341, "step": 18311 }, { "epoch": 0.02, "learning_rate": 4.995908397782304e-05, "loss": 0.1512, "step": 18312 }, { "epoch": 0.02, "learning_rate": 4.995907948562234e-05, "loss": 0.1605, "step": 18313 }, { "epoch": 0.02, "learning_rate": 4.995907499317527e-05, "loss": 0.1645, "step": 18314 }, { "epoch": 0.02, "learning_rate": 4.99590705004818e-05, "loss": 0.1523, "step": 18315 }, { "epoch": 0.02, "learning_rate": 4.9959066007541955e-05, "loss": 0.2021, "step": 18316 }, { "epoch": 0.02, "learning_rate": 4.995906151435572e-05, "loss": 0.6824, "step": 18317 }, { "epoch": 0.02, "learning_rate": 4.99590570209231e-05, "loss": 0.9813, "step": 18318 }, { "epoch": 0.02, "learning_rate": 4.995905252724409e-05, "loss": 1.1523, "step": 18319 }, { "epoch": 0.02, "learning_rate": 4.9959048033318704e-05, "loss": 1.2604, "step": 18320 }, { "epoch": 0.02, "learning_rate": 4.9959043539146925e-05, "loss": 1.0231, "step": 18321 }, { "epoch": 0.02, "learning_rate": 4.995903904472877e-05, "loss": 0.8977, "step": 18322 }, { "epoch": 0.02, "learning_rate": 4.995903455006422e-05, "loss": 1.2779, "step": 18323 }, { "epoch": 0.02, "learning_rate": 4.9959030055153286e-05, "loss": 0.9602, "step": 18324 }, { "epoch": 0.02, "learning_rate": 4.9959025559995974e-05, "loss": 1.1869, "step": 18325 }, { "epoch": 0.02, "learning_rate": 4.995902106459227e-05, "loss": 1.342, "step": 18326 }, { "epoch": 0.02, "learning_rate": 4.995901656894218e-05, "loss": 0.9983, "step": 18327 }, { "epoch": 0.02, "learning_rate": 4.9959012073045716e-05, "loss": 1.0568, "step": 18328 }, { "epoch": 0.02, "learning_rate": 4.995900757690286e-05, "loss": 1.0632, "step": 18329 }, { "epoch": 0.02, "learning_rate": 4.995900308051362e-05, "loss": 1.1853, "step": 18330 }, { "epoch": 0.02, "learning_rate": 4.995899858387799e-05, "loss": 1.0868, "step": 18331 }, { "epoch": 0.02, "learning_rate": 4.995899408699598e-05, "loss": 0.9986, "step": 18332 }, { "epoch": 0.02, "learning_rate": 4.995898958986759e-05, "loss": 1.3138, "step": 18333 }, { "epoch": 0.02, "learning_rate": 4.995898509249281e-05, "loss": 0.8306, "step": 18334 }, { "epoch": 0.02, "learning_rate": 4.9958980594871645e-05, "loss": 1.1192, "step": 18335 }, { "epoch": 0.02, "learning_rate": 4.99589760970041e-05, "loss": 0.8943, "step": 18336 }, { "epoch": 0.02, "learning_rate": 4.995897159889016e-05, "loss": 1.035, "step": 18337 }, { "epoch": 0.02, "learning_rate": 4.9958967100529845e-05, "loss": 0.6405, "step": 18338 }, { "epoch": 0.02, "learning_rate": 4.9958962601923144e-05, "loss": 1.4876, "step": 18339 }, { "epoch": 0.02, "learning_rate": 4.995895810307006e-05, "loss": 1.0796, "step": 18340 }, { "epoch": 0.02, "learning_rate": 4.995895360397059e-05, "loss": 0.8686, "step": 18341 }, { "epoch": 0.02, "learning_rate": 4.995894910462473e-05, "loss": 0.8334, "step": 18342 }, { "epoch": 0.02, "learning_rate": 4.995894460503249e-05, "loss": 1.1744, "step": 18343 }, { "epoch": 0.02, "learning_rate": 4.995894010519387e-05, "loss": 0.961, "step": 18344 }, { "epoch": 0.02, "learning_rate": 4.995893560510886e-05, "loss": 0.956, "step": 18345 }, { "epoch": 0.02, "learning_rate": 4.995893110477747e-05, "loss": 1.0698, "step": 18346 }, { "epoch": 0.02, "learning_rate": 4.99589266041997e-05, "loss": 1.0129, "step": 18347 }, { "epoch": 0.02, "learning_rate": 4.995892210337554e-05, "loss": 1.1725, "step": 18348 }, { "epoch": 0.02, "learning_rate": 4.995891760230499e-05, "loss": 1.346, "step": 18349 }, { "epoch": 0.02, "learning_rate": 4.995891310098807e-05, "loss": 1.2322, "step": 18350 }, { "epoch": 0.02, "learning_rate": 4.9958908599424755e-05, "loss": 1.1883, "step": 18351 }, { "epoch": 0.02, "learning_rate": 4.9958904097615056e-05, "loss": 1.0649, "step": 18352 }, { "epoch": 0.02, "learning_rate": 4.9958899595558987e-05, "loss": 1.2905, "step": 18353 }, { "epoch": 0.02, "learning_rate": 4.995889509325652e-05, "loss": 1.1603, "step": 18354 }, { "epoch": 0.02, "learning_rate": 4.995889059070767e-05, "loss": 1.1534, "step": 18355 }, { "epoch": 0.02, "learning_rate": 4.995888608791244e-05, "loss": 1.0142, "step": 18356 }, { "epoch": 0.02, "learning_rate": 4.995888158487083e-05, "loss": 1.147, "step": 18357 }, { "epoch": 0.02, "learning_rate": 4.9958877081582836e-05, "loss": 1.2824, "step": 18358 }, { "epoch": 0.02, "learning_rate": 4.995887257804845e-05, "loss": 1.151, "step": 18359 }, { "epoch": 0.02, "learning_rate": 4.9958868074267686e-05, "loss": 0.9651, "step": 18360 }, { "epoch": 0.02, "learning_rate": 4.9958863570240536e-05, "loss": 0.9157, "step": 18361 }, { "epoch": 0.02, "learning_rate": 4.995885906596701e-05, "loss": 1.0149, "step": 18362 }, { "epoch": 0.02, "learning_rate": 4.9958854561447096e-05, "loss": 1.0436, "step": 18363 }, { "epoch": 0.02, "learning_rate": 4.995885005668079e-05, "loss": 1.2484, "step": 18364 }, { "epoch": 0.02, "learning_rate": 4.995884555166812e-05, "loss": 1.1382, "step": 18365 }, { "epoch": 0.02, "learning_rate": 4.995884104640905e-05, "loss": 1.1968, "step": 18366 }, { "epoch": 0.02, "learning_rate": 4.9958836540903604e-05, "loss": 1.0643, "step": 18367 }, { "epoch": 0.02, "learning_rate": 4.9958832035151773e-05, "loss": 1.0081, "step": 18368 }, { "epoch": 0.02, "learning_rate": 4.9958827529153565e-05, "loss": 0.8702, "step": 18369 }, { "epoch": 0.02, "learning_rate": 4.9958823022908965e-05, "loss": 1.1349, "step": 18370 }, { "epoch": 0.02, "learning_rate": 4.995881851641798e-05, "loss": 1.021, "step": 18371 }, { "epoch": 0.02, "learning_rate": 4.995881400968062e-05, "loss": 1.4183, "step": 18372 }, { "epoch": 0.02, "learning_rate": 4.995880950269688e-05, "loss": 1.3256, "step": 18373 }, { "epoch": 0.02, "learning_rate": 4.995880499546675e-05, "loss": 1.1291, "step": 18374 }, { "epoch": 0.02, "learning_rate": 4.9958800487990236e-05, "loss": 1.2474, "step": 18375 }, { "epoch": 0.02, "learning_rate": 4.995879598026734e-05, "loss": 1.0276, "step": 18376 }, { "epoch": 0.02, "learning_rate": 4.9958791472298066e-05, "loss": 0.9539, "step": 18377 }, { "epoch": 0.02, "learning_rate": 4.9958786964082406e-05, "loss": 1.1405, "step": 18378 }, { "epoch": 0.02, "learning_rate": 4.995878245562037e-05, "loss": 1.0747, "step": 18379 }, { "epoch": 0.02, "learning_rate": 4.995877794691194e-05, "loss": 1.002, "step": 18380 }, { "epoch": 0.02, "learning_rate": 4.995877343795714e-05, "loss": 0.9923, "step": 18381 }, { "epoch": 0.02, "learning_rate": 4.9958768928755946e-05, "loss": 0.2467, "step": 18382 }, { "epoch": 0.02, "learning_rate": 4.9958764419308375e-05, "loss": 1.0074, "step": 18383 }, { "epoch": 0.02, "learning_rate": 4.995875990961442e-05, "loss": 1.134, "step": 18384 }, { "epoch": 0.02, "learning_rate": 4.9958755399674086e-05, "loss": 1.0755, "step": 18385 }, { "epoch": 0.02, "learning_rate": 4.995875088948737e-05, "loss": 1.3666, "step": 18386 }, { "epoch": 0.02, "learning_rate": 4.995874637905426e-05, "loss": 0.9621, "step": 18387 }, { "epoch": 0.02, "learning_rate": 4.995874186837478e-05, "loss": 1.1432, "step": 18388 }, { "epoch": 0.02, "learning_rate": 4.9958737357448915e-05, "loss": 1.055, "step": 18389 }, { "epoch": 0.02, "learning_rate": 4.995873284627667e-05, "loss": 1.0031, "step": 18390 }, { "epoch": 0.02, "learning_rate": 4.995872833485804e-05, "loss": 1.2682, "step": 18391 }, { "epoch": 0.02, "learning_rate": 4.9958723823193025e-05, "loss": 1.0962, "step": 18392 }, { "epoch": 0.02, "learning_rate": 4.995871931128163e-05, "loss": 1.1938, "step": 18393 }, { "epoch": 0.02, "learning_rate": 4.9958714799123854e-05, "loss": 1.3822, "step": 18394 }, { "epoch": 0.02, "learning_rate": 4.99587102867197e-05, "loss": 1.0939, "step": 18395 }, { "epoch": 0.02, "learning_rate": 4.995870577406916e-05, "loss": 1.2194, "step": 18396 }, { "epoch": 0.02, "learning_rate": 4.995870126117224e-05, "loss": 1.2812, "step": 18397 }, { "epoch": 0.02, "learning_rate": 4.995869674802893e-05, "loss": 1.0675, "step": 18398 }, { "epoch": 0.02, "learning_rate": 4.995869223463925e-05, "loss": 0.4616, "step": 18399 }, { "epoch": 0.02, "learning_rate": 4.9958687721003184e-05, "loss": 1.031, "step": 18400 }, { "epoch": 0.02, "learning_rate": 4.995868320712074e-05, "loss": 1.1872, "step": 18401 }, { "epoch": 0.02, "learning_rate": 4.99586786929919e-05, "loss": 1.2939, "step": 18402 }, { "epoch": 0.02, "learning_rate": 4.995867417861669e-05, "loss": 1.1857, "step": 18403 }, { "epoch": 0.02, "learning_rate": 4.99586696639951e-05, "loss": 1.2095, "step": 18404 }, { "epoch": 0.02, "learning_rate": 4.995866514912713e-05, "loss": 1.1288, "step": 18405 }, { "epoch": 0.02, "learning_rate": 4.9958660634012764e-05, "loss": 1.3285, "step": 18406 }, { "epoch": 0.02, "learning_rate": 4.9958656118652036e-05, "loss": 1.2138, "step": 18407 }, { "epoch": 0.02, "learning_rate": 4.9958651603044916e-05, "loss": 1.2384, "step": 18408 }, { "epoch": 0.02, "learning_rate": 4.995864708719141e-05, "loss": 0.8081, "step": 18409 }, { "epoch": 0.02, "learning_rate": 4.995864257109153e-05, "loss": 1.1014, "step": 18410 }, { "epoch": 0.02, "learning_rate": 4.9958638054745275e-05, "loss": 1.2486, "step": 18411 }, { "epoch": 0.02, "learning_rate": 4.995863353815263e-05, "loss": 1.0324, "step": 18412 }, { "epoch": 0.02, "learning_rate": 4.9958629021313604e-05, "loss": 1.1035, "step": 18413 }, { "epoch": 0.02, "learning_rate": 4.9958624504228195e-05, "loss": 1.2168, "step": 18414 }, { "epoch": 0.02, "learning_rate": 4.995861998689641e-05, "loss": 1.0017, "step": 18415 }, { "epoch": 0.02, "learning_rate": 4.995861546931824e-05, "loss": 1.166, "step": 18416 }, { "epoch": 0.02, "learning_rate": 4.995861095149369e-05, "loss": 1.0983, "step": 18417 }, { "epoch": 0.02, "learning_rate": 4.9958606433422764e-05, "loss": 1.0443, "step": 18418 }, { "epoch": 0.02, "learning_rate": 4.995860191510545e-05, "loss": 1.058, "step": 18419 }, { "epoch": 0.02, "learning_rate": 4.995859739654176e-05, "loss": 0.9717, "step": 18420 }, { "epoch": 0.02, "learning_rate": 4.995859287773169e-05, "loss": 1.1819, "step": 18421 }, { "epoch": 0.02, "learning_rate": 4.995858835867524e-05, "loss": 1.4359, "step": 18422 }, { "epoch": 0.02, "learning_rate": 4.995858383937241e-05, "loss": 1.3861, "step": 18423 }, { "epoch": 0.02, "learning_rate": 4.99585793198232e-05, "loss": 1.0926, "step": 18424 }, { "epoch": 0.02, "learning_rate": 4.99585748000276e-05, "loss": 0.8776, "step": 18425 }, { "epoch": 0.02, "learning_rate": 4.995857027998562e-05, "loss": 2.9384, "step": 18426 }, { "epoch": 0.02, "learning_rate": 4.995856575969727e-05, "loss": 5.9024, "step": 18427 }, { "epoch": 0.02, "learning_rate": 4.995856123916253e-05, "loss": 2.3389, "step": 18428 }, { "epoch": 0.02, "learning_rate": 4.9958556718381415e-05, "loss": 0.8945, "step": 18429 }, { "epoch": 0.02, "learning_rate": 4.995855219735392e-05, "loss": 1.2027, "step": 18430 }, { "epoch": 0.02, "learning_rate": 4.995854767608004e-05, "loss": 0.9737, "step": 18431 }, { "epoch": 0.02, "learning_rate": 4.9958543154559786e-05, "loss": 1.1853, "step": 18432 }, { "epoch": 0.02, "learning_rate": 4.995853863279315e-05, "loss": 1.2295, "step": 18433 }, { "epoch": 0.02, "learning_rate": 4.995853411078013e-05, "loss": 0.9874, "step": 18434 }, { "epoch": 0.02, "learning_rate": 4.9958529588520734e-05, "loss": 1.3611, "step": 18435 }, { "epoch": 0.02, "learning_rate": 4.995852506601496e-05, "loss": 1.277, "step": 18436 }, { "epoch": 0.02, "learning_rate": 4.9958520543262796e-05, "loss": 0.8831, "step": 18437 }, { "epoch": 0.02, "learning_rate": 4.995851602026426e-05, "loss": 1.3356, "step": 18438 }, { "epoch": 0.02, "learning_rate": 4.9958511497019346e-05, "loss": 1.6394, "step": 18439 }, { "epoch": 0.02, "learning_rate": 4.9958506973528044e-05, "loss": 1.4243, "step": 18440 }, { "epoch": 0.02, "learning_rate": 4.995850244979037e-05, "loss": 1.596, "step": 18441 }, { "epoch": 0.02, "learning_rate": 4.995849792580631e-05, "loss": 1.2079, "step": 18442 }, { "epoch": 0.02, "learning_rate": 4.995849340157587e-05, "loss": 1.1491, "step": 18443 }, { "epoch": 0.02, "learning_rate": 4.995848887709905e-05, "loss": 0.8432, "step": 18444 }, { "epoch": 0.02, "learning_rate": 4.995848435237586e-05, "loss": 1.4494, "step": 18445 }, { "epoch": 0.02, "learning_rate": 4.995847982740628e-05, "loss": 1.207, "step": 18446 }, { "epoch": 0.02, "learning_rate": 4.9958475302190325e-05, "loss": 1.0892, "step": 18447 }, { "epoch": 0.02, "learning_rate": 4.995847077672799e-05, "loss": 1.853, "step": 18448 }, { "epoch": 0.02, "learning_rate": 4.995846625101928e-05, "loss": 0.9565, "step": 18449 }, { "epoch": 0.02, "learning_rate": 4.995846172506418e-05, "loss": 0.8053, "step": 18450 }, { "epoch": 0.02, "learning_rate": 4.9958457198862704e-05, "loss": 0.6888, "step": 18451 }, { "epoch": 0.02, "learning_rate": 4.995845267241486e-05, "loss": 0.6319, "step": 18452 }, { "epoch": 0.02, "learning_rate": 4.995844814572062e-05, "loss": 1.2801, "step": 18453 }, { "epoch": 0.02, "learning_rate": 4.995844361878001e-05, "loss": 1.1611, "step": 18454 }, { "epoch": 0.02, "learning_rate": 4.995843909159302e-05, "loss": 1.163, "step": 18455 }, { "epoch": 0.02, "learning_rate": 4.995843456415965e-05, "loss": 1.3396, "step": 18456 }, { "epoch": 0.02, "learning_rate": 4.9958430036479894e-05, "loss": 0.9252, "step": 18457 }, { "epoch": 0.02, "learning_rate": 4.995842550855377e-05, "loss": 1.6883, "step": 18458 }, { "epoch": 0.02, "learning_rate": 4.995842098038126e-05, "loss": 0.9914, "step": 18459 }, { "epoch": 0.02, "learning_rate": 4.995841645196238e-05, "loss": 0.8411, "step": 18460 }, { "epoch": 0.02, "learning_rate": 4.995841192329711e-05, "loss": 0.975, "step": 18461 }, { "epoch": 0.02, "learning_rate": 4.9958407394385464e-05, "loss": 1.2624, "step": 18462 }, { "epoch": 0.02, "learning_rate": 4.9958402865227443e-05, "loss": 1.0267, "step": 18463 }, { "epoch": 0.02, "learning_rate": 4.9958398335823045e-05, "loss": 0.9703, "step": 18464 }, { "epoch": 0.02, "learning_rate": 4.995839380617226e-05, "loss": 1.0218, "step": 18465 }, { "epoch": 0.02, "learning_rate": 4.99583892762751e-05, "loss": 1.4141, "step": 18466 }, { "epoch": 0.02, "learning_rate": 4.995838474613156e-05, "loss": 0.9899, "step": 18467 }, { "epoch": 0.02, "learning_rate": 4.995838021574165e-05, "loss": 1.2778, "step": 18468 }, { "epoch": 0.02, "learning_rate": 4.995837568510535e-05, "loss": 1.2015, "step": 18469 }, { "epoch": 0.02, "learning_rate": 4.995837115422267e-05, "loss": 0.8757, "step": 18470 }, { "epoch": 0.02, "learning_rate": 4.9958366623093624e-05, "loss": 0.9858, "step": 18471 }, { "epoch": 0.02, "learning_rate": 4.995836209171819e-05, "loss": 1.0113, "step": 18472 }, { "epoch": 0.02, "learning_rate": 4.995835756009638e-05, "loss": 1.0319, "step": 18473 }, { "epoch": 0.02, "learning_rate": 4.99583530282282e-05, "loss": 1.15, "step": 18474 }, { "epoch": 0.02, "learning_rate": 4.9958348496113625e-05, "loss": 1.1155, "step": 18475 }, { "epoch": 0.02, "learning_rate": 4.995834396375269e-05, "loss": 1.098, "step": 18476 }, { "epoch": 0.02, "learning_rate": 4.995833943114536e-05, "loss": 1.2586, "step": 18477 }, { "epoch": 0.02, "learning_rate": 4.995833489829166e-05, "loss": 0.9165, "step": 18478 }, { "epoch": 0.02, "learning_rate": 4.995833036519158e-05, "loss": 1.2261, "step": 18479 }, { "epoch": 0.02, "learning_rate": 4.995832583184512e-05, "loss": 1.1286, "step": 18480 }, { "epoch": 0.02, "learning_rate": 4.9958321298252284e-05, "loss": 1.201, "step": 18481 }, { "epoch": 0.02, "learning_rate": 4.995831676441307e-05, "loss": 1.0331, "step": 18482 }, { "epoch": 0.02, "learning_rate": 4.995831223032748e-05, "loss": 1.3341, "step": 18483 }, { "epoch": 0.02, "learning_rate": 4.995830769599551e-05, "loss": 1.3514, "step": 18484 }, { "epoch": 0.02, "learning_rate": 4.9958303161417165e-05, "loss": 1.1229, "step": 18485 }, { "epoch": 0.02, "learning_rate": 4.995829862659244e-05, "loss": 0.8268, "step": 18486 }, { "epoch": 0.02, "learning_rate": 4.9958294091521343e-05, "loss": 0.9977, "step": 18487 }, { "epoch": 0.02, "learning_rate": 4.9958289556203855e-05, "loss": 1.1368, "step": 18488 }, { "epoch": 0.02, "learning_rate": 4.9958285020639995e-05, "loss": 1.09, "step": 18489 }, { "epoch": 0.02, "learning_rate": 4.9958280484829764e-05, "loss": 1.0654, "step": 18490 }, { "epoch": 0.02, "learning_rate": 4.995827594877315e-05, "loss": 0.9309, "step": 18491 }, { "epoch": 0.02, "learning_rate": 4.995827141247016e-05, "loss": 1.1166, "step": 18492 }, { "epoch": 0.02, "learning_rate": 4.995826687592079e-05, "loss": 0.9025, "step": 18493 }, { "epoch": 0.02, "learning_rate": 4.995826233912504e-05, "loss": 1.0814, "step": 18494 }, { "epoch": 0.02, "learning_rate": 4.995825780208292e-05, "loss": 1.0962, "step": 18495 }, { "epoch": 0.02, "learning_rate": 4.9958253264794416e-05, "loss": 1.0873, "step": 18496 }, { "epoch": 0.02, "learning_rate": 4.995824872725954e-05, "loss": 0.961, "step": 18497 }, { "epoch": 0.02, "learning_rate": 4.995824418947828e-05, "loss": 1.0476, "step": 18498 }, { "epoch": 0.02, "learning_rate": 4.9958239651450647e-05, "loss": 0.979, "step": 18499 }, { "epoch": 0.02, "learning_rate": 4.995823511317664e-05, "loss": 1.1472, "step": 18500 }, { "epoch": 0.02, "eval_loss": 1.0176509618759155, "eval_runtime": 83.3027, "eval_samples_per_second": 16.626, "eval_steps_per_second": 4.166, "step": 18500 }, { "epoch": 0.02, "learning_rate": 4.995823057465625e-05, "loss": 1.0562, "step": 18501 }, { "epoch": 0.02, "learning_rate": 4.995822603588949e-05, "loss": 1.1454, "step": 18502 }, { "epoch": 0.02, "learning_rate": 4.995822149687635e-05, "loss": 0.8597, "step": 18503 }, { "epoch": 0.02, "learning_rate": 4.995821695761682e-05, "loss": 0.9547, "step": 18504 }, { "epoch": 0.02, "learning_rate": 4.9958212418110925e-05, "loss": 1.1587, "step": 18505 }, { "epoch": 0.02, "learning_rate": 4.9958207878358656e-05, "loss": 1.0525, "step": 18506 }, { "epoch": 0.02, "learning_rate": 4.995820333836001e-05, "loss": 1.1954, "step": 18507 }, { "epoch": 0.02, "learning_rate": 4.995819879811498e-05, "loss": 1.0153, "step": 18508 }, { "epoch": 0.02, "learning_rate": 4.9958194257623576e-05, "loss": 1.2558, "step": 18509 }, { "epoch": 0.02, "learning_rate": 4.9958189716885795e-05, "loss": 0.6076, "step": 18510 }, { "epoch": 0.02, "learning_rate": 4.995818517590164e-05, "loss": 1.058, "step": 18511 }, { "epoch": 0.02, "learning_rate": 4.9958180634671105e-05, "loss": 1.1497, "step": 18512 }, { "epoch": 0.02, "learning_rate": 4.99581760931942e-05, "loss": 1.3763, "step": 18513 }, { "epoch": 0.02, "learning_rate": 4.995817155147091e-05, "loss": 1.1538, "step": 18514 }, { "epoch": 0.02, "learning_rate": 4.9958167009501245e-05, "loss": 1.1228, "step": 18515 }, { "epoch": 0.02, "learning_rate": 4.995816246728521e-05, "loss": 1.0067, "step": 18516 }, { "epoch": 0.02, "learning_rate": 4.995815792482279e-05, "loss": 0.9833, "step": 18517 }, { "epoch": 0.02, "learning_rate": 4.9958153382113995e-05, "loss": 0.8629, "step": 18518 }, { "epoch": 0.02, "learning_rate": 4.9958148839158824e-05, "loss": 1.029, "step": 18519 }, { "epoch": 0.02, "learning_rate": 4.995814429595728e-05, "loss": 1.0649, "step": 18520 }, { "epoch": 0.02, "learning_rate": 4.995813975250936e-05, "loss": 0.7171, "step": 18521 }, { "epoch": 0.02, "learning_rate": 4.995813520881506e-05, "loss": 0.9066, "step": 18522 }, { "epoch": 0.02, "learning_rate": 4.9958130664874386e-05, "loss": 0.897, "step": 18523 }, { "epoch": 0.02, "learning_rate": 4.995812612068734e-05, "loss": 1.1758, "step": 18524 }, { "epoch": 0.02, "learning_rate": 4.995812157625391e-05, "loss": 1.0875, "step": 18525 }, { "epoch": 0.02, "learning_rate": 4.995811703157411e-05, "loss": 0.9316, "step": 18526 }, { "epoch": 0.02, "learning_rate": 4.995811248664793e-05, "loss": 1.3038, "step": 18527 }, { "epoch": 0.02, "learning_rate": 4.995810794147538e-05, "loss": 1.2434, "step": 18528 }, { "epoch": 0.02, "learning_rate": 4.9958103396056445e-05, "loss": 1.3386, "step": 18529 }, { "epoch": 0.02, "learning_rate": 4.995809885039114e-05, "loss": 1.028, "step": 18530 }, { "epoch": 0.02, "learning_rate": 4.995809430447945e-05, "loss": 0.5326, "step": 18531 }, { "epoch": 0.02, "learning_rate": 4.99580897583214e-05, "loss": 0.5613, "step": 18532 }, { "epoch": 0.02, "learning_rate": 4.995808521191696e-05, "loss": 0.8667, "step": 18533 }, { "epoch": 0.02, "learning_rate": 4.995808066526615e-05, "loss": 0.9885, "step": 18534 }, { "epoch": 0.02, "learning_rate": 4.995807611836897e-05, "loss": 1.0232, "step": 18535 }, { "epoch": 0.02, "learning_rate": 4.9958071571225405e-05, "loss": 1.4324, "step": 18536 }, { "epoch": 0.02, "learning_rate": 4.9958067023835475e-05, "loss": 1.2016, "step": 18537 }, { "epoch": 0.02, "learning_rate": 4.9958062476199166e-05, "loss": 1.1257, "step": 18538 }, { "epoch": 0.02, "learning_rate": 4.995805792831647e-05, "loss": 0.7357, "step": 18539 }, { "epoch": 0.02, "learning_rate": 4.995805338018741e-05, "loss": 0.7083, "step": 18540 }, { "epoch": 0.02, "learning_rate": 4.995804883181198e-05, "loss": 0.8057, "step": 18541 }, { "epoch": 0.02, "learning_rate": 4.995804428319016e-05, "loss": 1.1432, "step": 18542 }, { "epoch": 0.02, "learning_rate": 4.995803973432197e-05, "loss": 1.3129, "step": 18543 }, { "epoch": 0.02, "learning_rate": 4.995803518520741e-05, "loss": 1.2044, "step": 18544 }, { "epoch": 0.02, "learning_rate": 4.9958030635846465e-05, "loss": 1.7206, "step": 18545 }, { "epoch": 0.02, "learning_rate": 4.995802608623915e-05, "loss": 1.2542, "step": 18546 }, { "epoch": 0.02, "learning_rate": 4.9958021536385466e-05, "loss": 1.0481, "step": 18547 }, { "epoch": 0.02, "learning_rate": 4.9958016986285396e-05, "loss": 1.0254, "step": 18548 }, { "epoch": 0.02, "learning_rate": 4.995801243593896e-05, "loss": 0.5418, "step": 18549 }, { "epoch": 0.02, "learning_rate": 4.995800788534615e-05, "loss": 0.2343, "step": 18550 }, { "epoch": 0.02, "learning_rate": 4.995800333450695e-05, "loss": 0.2062, "step": 18551 }, { "epoch": 0.02, "learning_rate": 4.995799878342139e-05, "loss": 1.1097, "step": 18552 }, { "epoch": 0.02, "learning_rate": 4.9957994232089455e-05, "loss": 1.0292, "step": 18553 }, { "epoch": 0.02, "learning_rate": 4.9957989680511144e-05, "loss": 1.0258, "step": 18554 }, { "epoch": 0.02, "learning_rate": 4.995798512868645e-05, "loss": 1.3199, "step": 18555 }, { "epoch": 0.02, "learning_rate": 4.9957980576615385e-05, "loss": 1.0545, "step": 18556 }, { "epoch": 0.02, "learning_rate": 4.9957976024297946e-05, "loss": 1.1292, "step": 18557 }, { "epoch": 0.02, "learning_rate": 4.9957971471734135e-05, "loss": 1.138, "step": 18558 }, { "epoch": 0.02, "learning_rate": 4.9957966918923946e-05, "loss": 1.2978, "step": 18559 }, { "epoch": 0.02, "learning_rate": 4.9957962365867385e-05, "loss": 1.2418, "step": 18560 }, { "epoch": 0.02, "learning_rate": 4.9957957812564447e-05, "loss": 1.5199, "step": 18561 }, { "epoch": 0.02, "learning_rate": 4.9957953259015137e-05, "loss": 1.5399, "step": 18562 }, { "epoch": 0.02, "learning_rate": 4.9957948705219455e-05, "loss": 1.0994, "step": 18563 }, { "epoch": 0.02, "learning_rate": 4.9957944151177395e-05, "loss": 0.994, "step": 18564 }, { "epoch": 0.02, "learning_rate": 4.995793959688896e-05, "loss": 1.4342, "step": 18565 }, { "epoch": 0.02, "learning_rate": 4.9957935042354155e-05, "loss": 1.403, "step": 18566 }, { "epoch": 0.02, "learning_rate": 4.995793048757297e-05, "loss": 1.1099, "step": 18567 }, { "epoch": 0.02, "learning_rate": 4.9957925932545416e-05, "loss": 1.0234, "step": 18568 }, { "epoch": 0.02, "learning_rate": 4.9957921377271486e-05, "loss": 0.8214, "step": 18569 }, { "epoch": 0.02, "learning_rate": 4.995791682175118e-05, "loss": 1.1172, "step": 18570 }, { "epoch": 0.02, "learning_rate": 4.99579122659845e-05, "loss": 1.1485, "step": 18571 }, { "epoch": 0.02, "learning_rate": 4.995790770997145e-05, "loss": 1.0386, "step": 18572 }, { "epoch": 0.02, "learning_rate": 4.9957903153712024e-05, "loss": 1.023, "step": 18573 }, { "epoch": 0.02, "learning_rate": 4.9957898597206223e-05, "loss": 1.1225, "step": 18574 }, { "epoch": 0.02, "learning_rate": 4.9957894040454045e-05, "loss": 1.1434, "step": 18575 }, { "epoch": 0.02, "learning_rate": 4.99578894834555e-05, "loss": 1.091, "step": 18576 }, { "epoch": 0.02, "learning_rate": 4.995788492621057e-05, "loss": 1.3018, "step": 18577 }, { "epoch": 0.02, "learning_rate": 4.995788036871928e-05, "loss": 1.4687, "step": 18578 }, { "epoch": 0.02, "learning_rate": 4.9957875810981615e-05, "loss": 1.2286, "step": 18579 }, { "epoch": 0.02, "learning_rate": 4.9957871252997566e-05, "loss": 0.8739, "step": 18580 }, { "epoch": 0.02, "learning_rate": 4.995786669476715e-05, "loss": 1.2501, "step": 18581 }, { "epoch": 0.02, "learning_rate": 4.995786213629035e-05, "loss": 0.9123, "step": 18582 }, { "epoch": 0.02, "learning_rate": 4.9957857577567197e-05, "loss": 1.2209, "step": 18583 }, { "epoch": 0.02, "learning_rate": 4.9957853018597655e-05, "loss": 1.1597, "step": 18584 }, { "epoch": 0.02, "learning_rate": 4.995784845938175e-05, "loss": 0.8396, "step": 18585 }, { "epoch": 0.02, "learning_rate": 4.9957843899919464e-05, "loss": 1.0583, "step": 18586 }, { "epoch": 0.02, "learning_rate": 4.995783934021081e-05, "loss": 1.734, "step": 18587 }, { "epoch": 0.02, "learning_rate": 4.9957834780255774e-05, "loss": 1.4075, "step": 18588 }, { "epoch": 0.02, "learning_rate": 4.995783022005437e-05, "loss": 1.0102, "step": 18589 }, { "epoch": 0.02, "learning_rate": 4.995782565960659e-05, "loss": 0.9147, "step": 18590 }, { "epoch": 0.02, "learning_rate": 4.995782109891245e-05, "loss": 1.0111, "step": 18591 }, { "epoch": 0.02, "learning_rate": 4.9957816537971924e-05, "loss": 1.3474, "step": 18592 }, { "epoch": 0.02, "learning_rate": 4.995781197678503e-05, "loss": 1.5134, "step": 18593 }, { "epoch": 0.02, "learning_rate": 4.995780741535175e-05, "loss": 1.173, "step": 18594 }, { "epoch": 0.02, "learning_rate": 4.995780285367212e-05, "loss": 1.3334, "step": 18595 }, { "epoch": 0.02, "learning_rate": 4.99577982917461e-05, "loss": 1.3363, "step": 18596 }, { "epoch": 0.02, "learning_rate": 4.995779372957371e-05, "loss": 1.0683, "step": 18597 }, { "epoch": 0.02, "learning_rate": 4.9957789167154954e-05, "loss": 0.9859, "step": 18598 }, { "epoch": 0.02, "learning_rate": 4.995778460448982e-05, "loss": 1.0153, "step": 18599 }, { "epoch": 0.02, "learning_rate": 4.995778004157831e-05, "loss": 0.8705, "step": 18600 }, { "epoch": 0.02, "learning_rate": 4.9957775478420434e-05, "loss": 0.7207, "step": 18601 }, { "epoch": 0.02, "learning_rate": 4.995777091501618e-05, "loss": 1.0778, "step": 18602 }, { "epoch": 0.02, "learning_rate": 4.9957766351365555e-05, "loss": 0.9905, "step": 18603 }, { "epoch": 0.02, "learning_rate": 4.995776178746856e-05, "loss": 1.3214, "step": 18604 }, { "epoch": 0.02, "learning_rate": 4.995775722332519e-05, "loss": 1.0826, "step": 18605 }, { "epoch": 0.02, "learning_rate": 4.995775265893545e-05, "loss": 1.5456, "step": 18606 }, { "epoch": 0.02, "learning_rate": 4.995774809429934e-05, "loss": 1.081, "step": 18607 }, { "epoch": 0.02, "learning_rate": 4.9957743529416844e-05, "loss": 1.1703, "step": 18608 }, { "epoch": 0.02, "learning_rate": 4.995773896428799e-05, "loss": 1.1928, "step": 18609 }, { "epoch": 0.02, "learning_rate": 4.995773439891276e-05, "loss": 1.2327, "step": 18610 }, { "epoch": 0.02, "learning_rate": 4.9957729833291156e-05, "loss": 1.18, "step": 18611 }, { "epoch": 0.02, "learning_rate": 4.995772526742318e-05, "loss": 1.1311, "step": 18612 }, { "epoch": 0.02, "learning_rate": 4.995772070130883e-05, "loss": 1.0516, "step": 18613 }, { "epoch": 0.02, "learning_rate": 4.995771613494811e-05, "loss": 0.9645, "step": 18614 }, { "epoch": 0.02, "learning_rate": 4.995771156834102e-05, "loss": 0.7369, "step": 18615 }, { "epoch": 0.02, "learning_rate": 4.9957707001487556e-05, "loss": 1.0913, "step": 18616 }, { "epoch": 0.02, "learning_rate": 4.995770243438772e-05, "loss": 1.0984, "step": 18617 }, { "epoch": 0.02, "learning_rate": 4.9957697867041515e-05, "loss": 1.3612, "step": 18618 }, { "epoch": 0.02, "learning_rate": 4.995769329944894e-05, "loss": 1.2238, "step": 18619 }, { "epoch": 0.02, "learning_rate": 4.995768873160999e-05, "loss": 0.8196, "step": 18620 }, { "epoch": 0.02, "learning_rate": 4.995768416352466e-05, "loss": 1.0637, "step": 18621 }, { "epoch": 0.02, "learning_rate": 4.995767959519296e-05, "loss": 1.062, "step": 18622 }, { "epoch": 0.02, "learning_rate": 4.9957675026614896e-05, "loss": 0.9911, "step": 18623 }, { "epoch": 0.02, "learning_rate": 4.9957670457790465e-05, "loss": 1.0274, "step": 18624 }, { "epoch": 0.02, "learning_rate": 4.995766588871965e-05, "loss": 1.2646, "step": 18625 }, { "epoch": 0.02, "learning_rate": 4.995766131940247e-05, "loss": 1.1133, "step": 18626 }, { "epoch": 0.02, "learning_rate": 4.995765674983891e-05, "loss": 1.2661, "step": 18627 }, { "epoch": 0.02, "learning_rate": 4.995765218002899e-05, "loss": 1.2347, "step": 18628 }, { "epoch": 0.02, "learning_rate": 4.995764760997269e-05, "loss": 1.2202, "step": 18629 }, { "epoch": 0.02, "learning_rate": 4.9957643039670024e-05, "loss": 1.3274, "step": 18630 }, { "epoch": 0.02, "learning_rate": 4.9957638469120987e-05, "loss": 1.0043, "step": 18631 }, { "epoch": 0.02, "learning_rate": 4.995763389832558e-05, "loss": 1.1143, "step": 18632 }, { "epoch": 0.02, "learning_rate": 4.99576293272838e-05, "loss": 0.8379, "step": 18633 }, { "epoch": 0.02, "learning_rate": 4.9957624755995645e-05, "loss": 1.0156, "step": 18634 }, { "epoch": 0.02, "learning_rate": 4.9957620184461115e-05, "loss": 0.9409, "step": 18635 }, { "epoch": 0.02, "learning_rate": 4.995761561268023e-05, "loss": 1.569, "step": 18636 }, { "epoch": 0.02, "learning_rate": 4.995761104065296e-05, "loss": 1.2287, "step": 18637 }, { "epoch": 0.02, "learning_rate": 4.9957606468379324e-05, "loss": 1.2217, "step": 18638 }, { "epoch": 0.02, "learning_rate": 4.995760189585932e-05, "loss": 1.1593, "step": 18639 }, { "epoch": 0.02, "learning_rate": 4.995759732309294e-05, "loss": 2.1623, "step": 18640 }, { "epoch": 0.02, "learning_rate": 4.995759275008018e-05, "loss": 1.002, "step": 18641 }, { "epoch": 0.02, "learning_rate": 4.995758817682107e-05, "loss": 1.09, "step": 18642 }, { "epoch": 0.02, "learning_rate": 4.995758360331558e-05, "loss": 0.9607, "step": 18643 }, { "epoch": 0.02, "learning_rate": 4.995757902956372e-05, "loss": 0.9771, "step": 18644 }, { "epoch": 0.02, "learning_rate": 4.995757445556548e-05, "loss": 1.1793, "step": 18645 }, { "epoch": 0.02, "learning_rate": 4.9957569881320874e-05, "loss": 1.0006, "step": 18646 }, { "epoch": 0.02, "learning_rate": 4.995756530682991e-05, "loss": 1.1336, "step": 18647 }, { "epoch": 0.02, "learning_rate": 4.995756073209256e-05, "loss": 1.0241, "step": 18648 }, { "epoch": 0.02, "learning_rate": 4.995755615710884e-05, "loss": 1.098, "step": 18649 }, { "epoch": 0.02, "learning_rate": 4.995755158187876e-05, "loss": 1.1592, "step": 18650 }, { "epoch": 0.02, "learning_rate": 4.995754700640231e-05, "loss": 0.6572, "step": 18651 }, { "epoch": 0.02, "learning_rate": 4.995754243067948e-05, "loss": 0.724, "step": 18652 }, { "epoch": 0.02, "learning_rate": 4.9957537854710276e-05, "loss": 0.9599, "step": 18653 }, { "epoch": 0.02, "learning_rate": 4.995753327849472e-05, "loss": 0.9119, "step": 18654 }, { "epoch": 0.02, "learning_rate": 4.995752870203277e-05, "loss": 1.1966, "step": 18655 }, { "epoch": 0.02, "learning_rate": 4.995752412532447e-05, "loss": 1.1889, "step": 18656 }, { "epoch": 0.02, "learning_rate": 4.995751954836979e-05, "loss": 1.1185, "step": 18657 }, { "epoch": 0.02, "learning_rate": 4.9957514971168745e-05, "loss": 0.7278, "step": 18658 }, { "epoch": 0.02, "learning_rate": 4.995751039372133e-05, "loss": 1.0603, "step": 18659 }, { "epoch": 0.02, "learning_rate": 4.995750581602754e-05, "loss": 0.7097, "step": 18660 }, { "epoch": 0.02, "learning_rate": 4.995750123808738e-05, "loss": 0.9824, "step": 18661 }, { "epoch": 0.02, "learning_rate": 4.995749665990085e-05, "loss": 1.2096, "step": 18662 }, { "epoch": 0.02, "learning_rate": 4.995749208146796e-05, "loss": 0.8129, "step": 18663 }, { "epoch": 0.02, "learning_rate": 4.995748750278869e-05, "loss": 1.0414, "step": 18664 }, { "epoch": 0.02, "learning_rate": 4.995748292386305e-05, "loss": 0.9903, "step": 18665 }, { "epoch": 0.02, "learning_rate": 4.995747834469104e-05, "loss": 1.1082, "step": 18666 }, { "epoch": 0.02, "learning_rate": 4.995747376527267e-05, "loss": 1.2643, "step": 18667 }, { "epoch": 0.02, "learning_rate": 4.9957469185607917e-05, "loss": 0.7967, "step": 18668 }, { "epoch": 0.02, "learning_rate": 4.995746460569681e-05, "loss": 1.1204, "step": 18669 }, { "epoch": 0.02, "learning_rate": 4.995746002553932e-05, "loss": 1.2089, "step": 18670 }, { "epoch": 0.02, "learning_rate": 4.9957455445135466e-05, "loss": 0.9445, "step": 18671 }, { "epoch": 0.02, "learning_rate": 4.995745086448524e-05, "loss": 1.356, "step": 18672 }, { "epoch": 0.02, "learning_rate": 4.9957446283588646e-05, "loss": 1.238, "step": 18673 }, { "epoch": 0.02, "learning_rate": 4.9957441702445686e-05, "loss": 1.1101, "step": 18674 }, { "epoch": 0.02, "learning_rate": 4.995743712105635e-05, "loss": 1.0953, "step": 18675 }, { "epoch": 0.02, "learning_rate": 4.9957432539420645e-05, "loss": 1.0968, "step": 18676 }, { "epoch": 0.02, "learning_rate": 4.995742795753858e-05, "loss": 0.8034, "step": 18677 }, { "epoch": 0.02, "learning_rate": 4.995742337541014e-05, "loss": 1.1695, "step": 18678 }, { "epoch": 0.02, "learning_rate": 4.995741879303533e-05, "loss": 0.8475, "step": 18679 }, { "epoch": 0.02, "learning_rate": 4.995741421041415e-05, "loss": 1.0989, "step": 18680 }, { "epoch": 0.02, "learning_rate": 4.99574096275466e-05, "loss": 1.2023, "step": 18681 }, { "epoch": 0.02, "learning_rate": 4.995740504443268e-05, "loss": 1.2671, "step": 18682 }, { "epoch": 0.02, "learning_rate": 4.99574004610724e-05, "loss": 1.2851, "step": 18683 }, { "epoch": 0.02, "learning_rate": 4.995739587746574e-05, "loss": 1.1667, "step": 18684 }, { "epoch": 0.02, "learning_rate": 4.995739129361272e-05, "loss": 1.0814, "step": 18685 }, { "epoch": 0.02, "learning_rate": 4.995738670951333e-05, "loss": 0.9102, "step": 18686 }, { "epoch": 0.02, "learning_rate": 4.995738212516757e-05, "loss": 1.0891, "step": 18687 }, { "epoch": 0.02, "learning_rate": 4.995737754057543e-05, "loss": 1.1878, "step": 18688 }, { "epoch": 0.02, "learning_rate": 4.9957372955736935e-05, "loss": 1.2836, "step": 18689 }, { "epoch": 0.02, "learning_rate": 4.9957368370652066e-05, "loss": 1.1388, "step": 18690 }, { "epoch": 0.02, "learning_rate": 4.995736378532083e-05, "loss": 1.4151, "step": 18691 }, { "epoch": 0.02, "learning_rate": 4.995735919974323e-05, "loss": 1.3109, "step": 18692 }, { "epoch": 0.02, "learning_rate": 4.995735461391925e-05, "loss": 0.5712, "step": 18693 }, { "epoch": 0.02, "learning_rate": 4.995735002784891e-05, "loss": 1.1513, "step": 18694 }, { "epoch": 0.02, "learning_rate": 4.995734544153221e-05, "loss": 1.0729, "step": 18695 }, { "epoch": 0.02, "learning_rate": 4.9957340854969124e-05, "loss": 1.3447, "step": 18696 }, { "epoch": 0.02, "learning_rate": 4.995733626815968e-05, "loss": 0.9536, "step": 18697 }, { "epoch": 0.02, "learning_rate": 4.9957331681103863e-05, "loss": 1.1581, "step": 18698 }, { "epoch": 0.02, "learning_rate": 4.995732709380168e-05, "loss": 1.3243, "step": 18699 }, { "epoch": 0.02, "learning_rate": 4.995732250625313e-05, "loss": 0.7225, "step": 18700 }, { "epoch": 0.02, "learning_rate": 4.9957317918458204e-05, "loss": 1.2341, "step": 18701 }, { "epoch": 0.02, "learning_rate": 4.995731333041692e-05, "loss": 1.076, "step": 18702 }, { "epoch": 0.02, "learning_rate": 4.9957308742129264e-05, "loss": 1.2282, "step": 18703 }, { "epoch": 0.02, "learning_rate": 4.9957304153595243e-05, "loss": 0.7369, "step": 18704 }, { "epoch": 0.02, "learning_rate": 4.995729956481485e-05, "loss": 1.141, "step": 18705 }, { "epoch": 0.02, "learning_rate": 4.995729497578809e-05, "loss": 0.9246, "step": 18706 }, { "epoch": 0.02, "learning_rate": 4.995729038651496e-05, "loss": 1.1081, "step": 18707 }, { "epoch": 0.02, "learning_rate": 4.995728579699547e-05, "loss": 0.9558, "step": 18708 }, { "epoch": 0.02, "learning_rate": 4.99572812072296e-05, "loss": 1.0073, "step": 18709 }, { "epoch": 0.02, "learning_rate": 4.995727661721737e-05, "loss": 1.0935, "step": 18710 }, { "epoch": 0.02, "learning_rate": 4.995727202695877e-05, "loss": 1.0558, "step": 18711 }, { "epoch": 0.02, "learning_rate": 4.9957267436453804e-05, "loss": 1.2808, "step": 18712 }, { "epoch": 0.02, "learning_rate": 4.995726284570247e-05, "loss": 1.0017, "step": 18713 }, { "epoch": 0.02, "learning_rate": 4.995725825470476e-05, "loss": 1.0826, "step": 18714 }, { "epoch": 0.02, "learning_rate": 4.99572536634607e-05, "loss": 0.9219, "step": 18715 }, { "epoch": 0.02, "learning_rate": 4.995724907197026e-05, "loss": 1.0565, "step": 18716 }, { "epoch": 0.02, "learning_rate": 4.995724448023346e-05, "loss": 1.0814, "step": 18717 }, { "epoch": 0.02, "learning_rate": 4.9957239888250285e-05, "loss": 1.1296, "step": 18718 }, { "epoch": 0.02, "learning_rate": 4.995723529602074e-05, "loss": 1.8605, "step": 18719 }, { "epoch": 0.02, "learning_rate": 4.995723070354484e-05, "loss": 4.8837, "step": 18720 }, { "epoch": 0.02, "learning_rate": 4.9957226110822566e-05, "loss": 2.0658, "step": 18721 }, { "epoch": 0.02, "learning_rate": 4.995722151785392e-05, "loss": 0.9957, "step": 18722 }, { "epoch": 0.02, "learning_rate": 4.995721692463892e-05, "loss": 0.8657, "step": 18723 }, { "epoch": 0.02, "learning_rate": 4.995721233117754e-05, "loss": 1.1575, "step": 18724 }, { "epoch": 0.02, "learning_rate": 4.99572077374698e-05, "loss": 1.1779, "step": 18725 }, { "epoch": 0.02, "learning_rate": 4.995720314351569e-05, "loss": 1.089, "step": 18726 }, { "epoch": 0.02, "learning_rate": 4.995719854931521e-05, "loss": 1.3419, "step": 18727 }, { "epoch": 0.02, "learning_rate": 4.995719395486837e-05, "loss": 1.1165, "step": 18728 }, { "epoch": 0.02, "learning_rate": 4.995718936017516e-05, "loss": 1.1278, "step": 18729 }, { "epoch": 0.02, "learning_rate": 4.9957184765235585e-05, "loss": 1.2034, "step": 18730 }, { "epoch": 0.02, "learning_rate": 4.995718017004964e-05, "loss": 0.5961, "step": 18731 }, { "epoch": 0.02, "learning_rate": 4.995717557461733e-05, "loss": 0.6231, "step": 18732 }, { "epoch": 0.02, "learning_rate": 4.995717097893865e-05, "loss": 1.055, "step": 18733 }, { "epoch": 0.02, "learning_rate": 4.995716638301361e-05, "loss": 1.345, "step": 18734 }, { "epoch": 0.02, "learning_rate": 4.99571617868422e-05, "loss": 0.96, "step": 18735 }, { "epoch": 0.02, "learning_rate": 4.995715719042442e-05, "loss": 1.181, "step": 18736 }, { "epoch": 0.02, "learning_rate": 4.995715259376027e-05, "loss": 1.0727, "step": 18737 }, { "epoch": 0.02, "learning_rate": 4.9957147996849764e-05, "loss": 1.2496, "step": 18738 }, { "epoch": 0.02, "learning_rate": 4.995714339969289e-05, "loss": 1.163, "step": 18739 }, { "epoch": 0.02, "learning_rate": 4.9957138802289645e-05, "loss": 1.255, "step": 18740 }, { "epoch": 0.02, "learning_rate": 4.995713420464003e-05, "loss": 1.0586, "step": 18741 }, { "epoch": 0.02, "learning_rate": 4.9957129606744055e-05, "loss": 1.0457, "step": 18742 }, { "epoch": 0.02, "learning_rate": 4.995712500860171e-05, "loss": 0.9793, "step": 18743 }, { "epoch": 0.02, "learning_rate": 4.995712041021301e-05, "loss": 0.9416, "step": 18744 }, { "epoch": 0.02, "learning_rate": 4.9957115811577934e-05, "loss": 1.0545, "step": 18745 }, { "epoch": 0.02, "learning_rate": 4.995711121269649e-05, "loss": 1.1682, "step": 18746 }, { "epoch": 0.02, "learning_rate": 4.9957106613568684e-05, "loss": 1.0444, "step": 18747 }, { "epoch": 0.02, "learning_rate": 4.995710201419451e-05, "loss": 1.0657, "step": 18748 }, { "epoch": 0.02, "learning_rate": 4.995709741457397e-05, "loss": 0.7604, "step": 18749 }, { "epoch": 0.02, "learning_rate": 4.9957092814707075e-05, "loss": 0.9503, "step": 18750 }, { "epoch": 0.02, "learning_rate": 4.99570882145938e-05, "loss": 1.0412, "step": 18751 }, { "epoch": 0.02, "learning_rate": 4.9957083614234165e-05, "loss": 1.2006, "step": 18752 }, { "epoch": 0.02, "learning_rate": 4.995707901362816e-05, "loss": 1.1235, "step": 18753 }, { "epoch": 0.02, "learning_rate": 4.9957074412775796e-05, "loss": 1.0585, "step": 18754 }, { "epoch": 0.02, "learning_rate": 4.995706981167706e-05, "loss": 1.3471, "step": 18755 }, { "epoch": 0.02, "learning_rate": 4.995706521033196e-05, "loss": 1.1442, "step": 18756 }, { "epoch": 0.02, "learning_rate": 4.995706060874049e-05, "loss": 1.2157, "step": 18757 }, { "epoch": 0.02, "learning_rate": 4.9957056006902664e-05, "loss": 1.1927, "step": 18758 }, { "epoch": 0.02, "learning_rate": 4.995705140481847e-05, "loss": 1.063, "step": 18759 }, { "epoch": 0.02, "learning_rate": 4.995704680248791e-05, "loss": 0.7825, "step": 18760 }, { "epoch": 0.02, "learning_rate": 4.9957042199910974e-05, "loss": 1.1334, "step": 18761 }, { "epoch": 0.02, "learning_rate": 4.9957037597087684e-05, "loss": 1.0263, "step": 18762 }, { "epoch": 0.02, "learning_rate": 4.995703299401803e-05, "loss": 1.2272, "step": 18763 }, { "epoch": 0.02, "learning_rate": 4.9957028390702003e-05, "loss": 1.1425, "step": 18764 }, { "epoch": 0.02, "learning_rate": 4.995702378713962e-05, "loss": 1.0721, "step": 18765 }, { "epoch": 0.02, "learning_rate": 4.9957019183330864e-05, "loss": 0.9337, "step": 18766 }, { "epoch": 0.02, "learning_rate": 4.995701457927574e-05, "loss": 0.9464, "step": 18767 }, { "epoch": 0.02, "learning_rate": 4.995700997497425e-05, "loss": 1.1565, "step": 18768 }, { "epoch": 0.02, "learning_rate": 4.995700537042641e-05, "loss": 1.2243, "step": 18769 }, { "epoch": 0.02, "learning_rate": 4.99570007656322e-05, "loss": 1.6535, "step": 18770 }, { "epoch": 0.02, "learning_rate": 4.995699616059162e-05, "loss": 0.9981, "step": 18771 }, { "epoch": 0.02, "learning_rate": 4.9956991555304675e-05, "loss": 1.1624, "step": 18772 }, { "epoch": 0.02, "learning_rate": 4.995698694977137e-05, "loss": 0.9038, "step": 18773 }, { "epoch": 0.02, "learning_rate": 4.995698234399169e-05, "loss": 0.9798, "step": 18774 }, { "epoch": 0.02, "learning_rate": 4.995697773796565e-05, "loss": 1.079, "step": 18775 }, { "epoch": 0.02, "learning_rate": 4.995697313169325e-05, "loss": 0.8937, "step": 18776 }, { "epoch": 0.02, "learning_rate": 4.995696852517448e-05, "loss": 1.2507, "step": 18777 }, { "epoch": 0.02, "learning_rate": 4.995696391840935e-05, "loss": 0.8063, "step": 18778 }, { "epoch": 0.02, "learning_rate": 4.995695931139785e-05, "loss": 0.7078, "step": 18779 }, { "epoch": 0.02, "learning_rate": 4.995695470413999e-05, "loss": 1.1335, "step": 18780 }, { "epoch": 0.02, "learning_rate": 4.995695009663576e-05, "loss": 1.1945, "step": 18781 }, { "epoch": 0.02, "learning_rate": 4.995694548888517e-05, "loss": 1.024, "step": 18782 }, { "epoch": 0.02, "learning_rate": 4.9956940880888216e-05, "loss": 1.0968, "step": 18783 }, { "epoch": 0.02, "learning_rate": 4.99569362726449e-05, "loss": 0.6154, "step": 18784 }, { "epoch": 0.02, "learning_rate": 4.995693166415521e-05, "loss": 1.3035, "step": 18785 }, { "epoch": 0.02, "learning_rate": 4.995692705541916e-05, "loss": 1.192, "step": 18786 }, { "epoch": 0.02, "learning_rate": 4.995692244643675e-05, "loss": 1.5293, "step": 18787 }, { "epoch": 0.02, "learning_rate": 4.995691783720797e-05, "loss": 1.1368, "step": 18788 }, { "epoch": 0.02, "learning_rate": 4.995691322773283e-05, "loss": 1.1486, "step": 18789 }, { "epoch": 0.02, "learning_rate": 4.995690861801132e-05, "loss": 1.0575, "step": 18790 }, { "epoch": 0.02, "learning_rate": 4.995690400804346e-05, "loss": 0.8594, "step": 18791 }, { "epoch": 0.02, "learning_rate": 4.9956899397829224e-05, "loss": 1.0828, "step": 18792 }, { "epoch": 0.02, "learning_rate": 4.9956894787368624e-05, "loss": 1.1661, "step": 18793 }, { "epoch": 0.02, "learning_rate": 4.9956890176661666e-05, "loss": 1.3301, "step": 18794 }, { "epoch": 0.02, "learning_rate": 4.995688556570834e-05, "loss": 1.3199, "step": 18795 }, { "epoch": 0.02, "learning_rate": 4.995688095450864e-05, "loss": 1.1489, "step": 18796 }, { "epoch": 0.02, "learning_rate": 4.99568763430626e-05, "loss": 1.1774, "step": 18797 }, { "epoch": 0.02, "learning_rate": 4.995687173137018e-05, "loss": 0.8964, "step": 18798 }, { "epoch": 0.02, "learning_rate": 4.9956867119431394e-05, "loss": 1.2073, "step": 18799 }, { "epoch": 0.02, "learning_rate": 4.995686250724625e-05, "loss": 0.6838, "step": 18800 }, { "epoch": 0.02, "learning_rate": 4.9956857894814745e-05, "loss": 1.2512, "step": 18801 }, { "epoch": 0.02, "learning_rate": 4.995685328213687e-05, "loss": 1.143, "step": 18802 }, { "epoch": 0.02, "learning_rate": 4.9956848669212644e-05, "loss": 1.6867, "step": 18803 }, { "epoch": 0.02, "learning_rate": 4.995684405604204e-05, "loss": 0.8509, "step": 18804 }, { "epoch": 0.02, "learning_rate": 4.995683944262508e-05, "loss": 1.221, "step": 18805 }, { "epoch": 0.02, "learning_rate": 4.995683482896175e-05, "loss": 1.0642, "step": 18806 }, { "epoch": 0.02, "learning_rate": 4.995683021505206e-05, "loss": 1.1661, "step": 18807 }, { "epoch": 0.02, "learning_rate": 4.995682560089601e-05, "loss": 0.9478, "step": 18808 }, { "epoch": 0.02, "learning_rate": 4.99568209864936e-05, "loss": 1.0575, "step": 18809 }, { "epoch": 0.02, "learning_rate": 4.9956816371844814e-05, "loss": 1.1764, "step": 18810 }, { "epoch": 0.02, "learning_rate": 4.995681175694967e-05, "loss": 1.1719, "step": 18811 }, { "epoch": 0.02, "learning_rate": 4.995680714180817e-05, "loss": 1.2393, "step": 18812 }, { "epoch": 0.02, "learning_rate": 4.9956802526420305e-05, "loss": 0.8329, "step": 18813 }, { "epoch": 0.02, "learning_rate": 4.9956797910786076e-05, "loss": 0.511, "step": 18814 }, { "epoch": 0.02, "learning_rate": 4.995679329490548e-05, "loss": 1.016, "step": 18815 }, { "epoch": 0.02, "learning_rate": 4.995678867877852e-05, "loss": 0.5035, "step": 18816 }, { "epoch": 0.02, "learning_rate": 4.9956784062405205e-05, "loss": 0.7687, "step": 18817 }, { "epoch": 0.02, "learning_rate": 4.995677944578552e-05, "loss": 1.0423, "step": 18818 }, { "epoch": 0.02, "learning_rate": 4.995677482891947e-05, "loss": 1.1893, "step": 18819 }, { "epoch": 0.02, "learning_rate": 4.995677021180707e-05, "loss": 1.1816, "step": 18820 }, { "epoch": 0.02, "learning_rate": 4.99567655944483e-05, "loss": 1.1766, "step": 18821 }, { "epoch": 0.02, "learning_rate": 4.9956760976843165e-05, "loss": 1.2197, "step": 18822 }, { "epoch": 0.02, "learning_rate": 4.995675635899166e-05, "loss": 1.0619, "step": 18823 }, { "epoch": 0.02, "learning_rate": 4.9956751740893805e-05, "loss": 0.9724, "step": 18824 }, { "epoch": 0.02, "learning_rate": 4.9956747122549585e-05, "loss": 1.0751, "step": 18825 }, { "epoch": 0.02, "learning_rate": 4.9956742503959e-05, "loss": 1.2292, "step": 18826 }, { "epoch": 0.02, "learning_rate": 4.995673788512206e-05, "loss": 1.1899, "step": 18827 }, { "epoch": 0.02, "learning_rate": 4.995673326603874e-05, "loss": 1.1267, "step": 18828 }, { "epoch": 0.02, "learning_rate": 4.995672864670907e-05, "loss": 0.8714, "step": 18829 }, { "epoch": 0.02, "learning_rate": 4.995672402713304e-05, "loss": 0.8407, "step": 18830 }, { "epoch": 0.02, "learning_rate": 4.995671940731065e-05, "loss": 0.9007, "step": 18831 }, { "epoch": 0.02, "learning_rate": 4.995671478724189e-05, "loss": 0.8815, "step": 18832 }, { "epoch": 0.02, "learning_rate": 4.9956710166926765e-05, "loss": 1.1909, "step": 18833 }, { "epoch": 0.02, "learning_rate": 4.995670554636529e-05, "loss": 1.2753, "step": 18834 }, { "epoch": 0.02, "learning_rate": 4.9956700925557444e-05, "loss": 0.5933, "step": 18835 }, { "epoch": 0.02, "learning_rate": 4.995669630450323e-05, "loss": 1.0495, "step": 18836 }, { "epoch": 0.02, "learning_rate": 4.995669168320267e-05, "loss": 0.7873, "step": 18837 }, { "epoch": 0.02, "learning_rate": 4.995668706165574e-05, "loss": 1.1398, "step": 18838 }, { "epoch": 0.02, "learning_rate": 4.995668243986245e-05, "loss": 0.7889, "step": 18839 }, { "epoch": 0.02, "learning_rate": 4.995667781782279e-05, "loss": 1.1508, "step": 18840 }, { "epoch": 0.02, "learning_rate": 4.995667319553677e-05, "loss": 1.126, "step": 18841 }, { "epoch": 0.02, "learning_rate": 4.99566685730044e-05, "loss": 1.3882, "step": 18842 }, { "epoch": 0.02, "learning_rate": 4.995666395022566e-05, "loss": 0.9422, "step": 18843 }, { "epoch": 0.02, "learning_rate": 4.9956659327200564e-05, "loss": 1.1636, "step": 18844 }, { "epoch": 0.02, "learning_rate": 4.99566547039291e-05, "loss": 1.0074, "step": 18845 }, { "epoch": 0.02, "learning_rate": 4.995665008041127e-05, "loss": 1.1587, "step": 18846 }, { "epoch": 0.02, "learning_rate": 4.995664545664709e-05, "loss": 1.1707, "step": 18847 }, { "epoch": 0.02, "learning_rate": 4.9956640832636536e-05, "loss": 0.9952, "step": 18848 }, { "epoch": 0.02, "learning_rate": 4.995663620837963e-05, "loss": 1.1831, "step": 18849 }, { "epoch": 0.02, "learning_rate": 4.9956631583876365e-05, "loss": 1.0425, "step": 18850 }, { "epoch": 0.02, "learning_rate": 4.9956626959126726e-05, "loss": 1.3147, "step": 18851 }, { "epoch": 0.02, "learning_rate": 4.9956622334130736e-05, "loss": 1.029, "step": 18852 }, { "epoch": 0.02, "learning_rate": 4.995661770888839e-05, "loss": 1.019, "step": 18853 }, { "epoch": 0.02, "learning_rate": 4.9956613083399675e-05, "loss": 1.1613, "step": 18854 }, { "epoch": 0.02, "learning_rate": 4.99566084576646e-05, "loss": 1.1628, "step": 18855 }, { "epoch": 0.02, "learning_rate": 4.9956603831683155e-05, "loss": 1.1337, "step": 18856 }, { "epoch": 0.02, "learning_rate": 4.9956599205455356e-05, "loss": 1.1482, "step": 18857 }, { "epoch": 0.02, "learning_rate": 4.99565945789812e-05, "loss": 1.0812, "step": 18858 }, { "epoch": 0.02, "learning_rate": 4.995658995226068e-05, "loss": 1.1089, "step": 18859 }, { "epoch": 0.02, "learning_rate": 4.995658532529379e-05, "loss": 0.8698, "step": 18860 }, { "epoch": 0.02, "learning_rate": 4.995658069808056e-05, "loss": 1.0523, "step": 18861 }, { "epoch": 0.02, "learning_rate": 4.9956576070620955e-05, "loss": 0.9829, "step": 18862 }, { "epoch": 0.02, "learning_rate": 4.995657144291499e-05, "loss": 0.9738, "step": 18863 }, { "epoch": 0.02, "learning_rate": 4.995656681496266e-05, "loss": 1.401, "step": 18864 }, { "epoch": 0.02, "learning_rate": 4.995656218676398e-05, "loss": 1.3726, "step": 18865 }, { "epoch": 0.02, "learning_rate": 4.9956557558318926e-05, "loss": 1.3136, "step": 18866 }, { "epoch": 0.02, "learning_rate": 4.995655292962752e-05, "loss": 1.1849, "step": 18867 }, { "epoch": 0.02, "learning_rate": 4.995654830068976e-05, "loss": 1.0544, "step": 18868 }, { "epoch": 0.02, "learning_rate": 4.995654367150563e-05, "loss": 1.0512, "step": 18869 }, { "epoch": 0.02, "learning_rate": 4.995653904207514e-05, "loss": 0.4722, "step": 18870 }, { "epoch": 0.02, "learning_rate": 4.9956534412398294e-05, "loss": 0.3293, "step": 18871 }, { "epoch": 0.02, "learning_rate": 4.995652978247508e-05, "loss": 0.2504, "step": 18872 }, { "epoch": 0.02, "learning_rate": 4.995652515230551e-05, "loss": 0.9653, "step": 18873 }, { "epoch": 0.02, "learning_rate": 4.995652052188958e-05, "loss": 0.6443, "step": 18874 }, { "epoch": 0.02, "learning_rate": 4.9956515891227294e-05, "loss": 0.4432, "step": 18875 }, { "epoch": 0.02, "learning_rate": 4.995651126031865e-05, "loss": 0.4998, "step": 18876 }, { "epoch": 0.02, "learning_rate": 4.995650662916364e-05, "loss": 0.6492, "step": 18877 }, { "epoch": 0.02, "learning_rate": 4.995650199776226e-05, "loss": 0.4128, "step": 18878 }, { "epoch": 0.02, "learning_rate": 4.995649736611453e-05, "loss": 0.3397, "step": 18879 }, { "epoch": 0.02, "learning_rate": 4.995649273422044e-05, "loss": 0.3419, "step": 18880 }, { "epoch": 0.02, "learning_rate": 4.995648810207999e-05, "loss": 0.215, "step": 18881 }, { "epoch": 0.02, "learning_rate": 4.995648346969318e-05, "loss": 0.897, "step": 18882 }, { "epoch": 0.02, "learning_rate": 4.995647883706001e-05, "loss": 1.0589, "step": 18883 }, { "epoch": 0.02, "learning_rate": 4.995647420418048e-05, "loss": 1.0587, "step": 18884 }, { "epoch": 0.02, "learning_rate": 4.995646957105459e-05, "loss": 1.3063, "step": 18885 }, { "epoch": 0.02, "learning_rate": 4.995646493768234e-05, "loss": 1.1808, "step": 18886 }, { "epoch": 0.02, "learning_rate": 4.995646030406373e-05, "loss": 1.1897, "step": 18887 }, { "epoch": 0.02, "learning_rate": 4.9956455670198765e-05, "loss": 1.0648, "step": 18888 }, { "epoch": 0.02, "learning_rate": 4.995645103608743e-05, "loss": 1.3186, "step": 18889 }, { "epoch": 0.02, "learning_rate": 4.9956446401729743e-05, "loss": 1.2679, "step": 18890 }, { "epoch": 0.02, "learning_rate": 4.9956441767125696e-05, "loss": 1.3377, "step": 18891 }, { "epoch": 0.02, "learning_rate": 4.995643713227529e-05, "loss": 0.9258, "step": 18892 }, { "epoch": 0.02, "learning_rate": 4.995643249717852e-05, "loss": 1.0726, "step": 18893 }, { "epoch": 0.02, "learning_rate": 4.99564278618354e-05, "loss": 1.1436, "step": 18894 }, { "epoch": 0.02, "learning_rate": 4.9956423226245907e-05, "loss": 1.1302, "step": 18895 }, { "epoch": 0.02, "learning_rate": 4.995641859041007e-05, "loss": 1.3627, "step": 18896 }, { "epoch": 0.02, "learning_rate": 4.995641395432786e-05, "loss": 1.3557, "step": 18897 }, { "epoch": 0.02, "learning_rate": 4.9956409317999295e-05, "loss": 1.014, "step": 18898 }, { "epoch": 0.02, "learning_rate": 4.995640468142437e-05, "loss": 1.0417, "step": 18899 }, { "epoch": 0.02, "learning_rate": 4.995640004460309e-05, "loss": 1.0157, "step": 18900 }, { "epoch": 0.02, "learning_rate": 4.995639540753546e-05, "loss": 1.1135, "step": 18901 }, { "epoch": 0.02, "learning_rate": 4.995639077022146e-05, "loss": 1.1388, "step": 18902 }, { "epoch": 0.02, "learning_rate": 4.995638613266109e-05, "loss": 0.9336, "step": 18903 }, { "epoch": 0.02, "learning_rate": 4.995638149485438e-05, "loss": 1.1962, "step": 18904 }, { "epoch": 0.02, "learning_rate": 4.99563768568013e-05, "loss": 1.0552, "step": 18905 }, { "epoch": 0.02, "learning_rate": 4.995637221850187e-05, "loss": 1.3947, "step": 18906 }, { "epoch": 0.02, "learning_rate": 4.995636757995608e-05, "loss": 1.2745, "step": 18907 }, { "epoch": 0.02, "learning_rate": 4.995636294116392e-05, "loss": 1.0674, "step": 18908 }, { "epoch": 0.02, "learning_rate": 4.9956358302125404e-05, "loss": 1.2956, "step": 18909 }, { "epoch": 0.02, "learning_rate": 4.995635366284054e-05, "loss": 1.193, "step": 18910 }, { "epoch": 0.02, "learning_rate": 4.995634902330931e-05, "loss": 1.1428, "step": 18911 }, { "epoch": 0.02, "learning_rate": 4.995634438353172e-05, "loss": 1.0531, "step": 18912 }, { "epoch": 0.02, "learning_rate": 4.9956339743507776e-05, "loss": 0.9837, "step": 18913 }, { "epoch": 0.02, "learning_rate": 4.995633510323747e-05, "loss": 0.9115, "step": 18914 }, { "epoch": 0.02, "learning_rate": 4.9956330462720814e-05, "loss": 1.073, "step": 18915 }, { "epoch": 0.02, "learning_rate": 4.995632582195779e-05, "loss": 1.2471, "step": 18916 }, { "epoch": 0.02, "learning_rate": 4.995632118094841e-05, "loss": 1.2539, "step": 18917 }, { "epoch": 0.02, "learning_rate": 4.995631653969267e-05, "loss": 0.8266, "step": 18918 }, { "epoch": 0.02, "learning_rate": 4.995631189819058e-05, "loss": 0.5051, "step": 18919 }, { "epoch": 0.02, "learning_rate": 4.9956307256442126e-05, "loss": 0.6859, "step": 18920 }, { "epoch": 0.02, "learning_rate": 4.9956302614447316e-05, "loss": 0.7548, "step": 18921 }, { "epoch": 0.02, "learning_rate": 4.995629797220614e-05, "loss": 0.5911, "step": 18922 }, { "epoch": 0.02, "learning_rate": 4.995629332971861e-05, "loss": 1.3208, "step": 18923 }, { "epoch": 0.02, "learning_rate": 4.9956288686984725e-05, "loss": 1.0405, "step": 18924 }, { "epoch": 0.02, "learning_rate": 4.9956284044004484e-05, "loss": 0.9442, "step": 18925 }, { "epoch": 0.02, "learning_rate": 4.995627940077788e-05, "loss": 1.1572, "step": 18926 }, { "epoch": 0.02, "learning_rate": 4.995627475730492e-05, "loss": 1.1518, "step": 18927 }, { "epoch": 0.02, "learning_rate": 4.99562701135856e-05, "loss": 1.0818, "step": 18928 }, { "epoch": 0.02, "learning_rate": 4.995626546961992e-05, "loss": 1.1069, "step": 18929 }, { "epoch": 0.02, "learning_rate": 4.995626082540789e-05, "loss": 1.0277, "step": 18930 }, { "epoch": 0.02, "learning_rate": 4.99562561809495e-05, "loss": 1.0615, "step": 18931 }, { "epoch": 0.02, "learning_rate": 4.9956251536244755e-05, "loss": 0.8931, "step": 18932 }, { "epoch": 0.02, "learning_rate": 4.9956246891293645e-05, "loss": 1.1395, "step": 18933 }, { "epoch": 0.02, "learning_rate": 4.995624224609618e-05, "loss": 0.9476, "step": 18934 }, { "epoch": 0.02, "learning_rate": 4.9956237600652356e-05, "loss": 1.0827, "step": 18935 }, { "epoch": 0.02, "learning_rate": 4.995623295496218e-05, "loss": 0.9857, "step": 18936 }, { "epoch": 0.02, "learning_rate": 4.9956228309025644e-05, "loss": 1.2044, "step": 18937 }, { "epoch": 0.02, "learning_rate": 4.995622366284275e-05, "loss": 0.8928, "step": 18938 }, { "epoch": 0.02, "learning_rate": 4.99562190164135e-05, "loss": 0.9947, "step": 18939 }, { "epoch": 0.02, "learning_rate": 4.995621436973789e-05, "loss": 1.0686, "step": 18940 }, { "epoch": 0.02, "learning_rate": 4.995620972281593e-05, "loss": 1.2732, "step": 18941 }, { "epoch": 0.02, "learning_rate": 4.99562050756476e-05, "loss": 1.1725, "step": 18942 }, { "epoch": 0.02, "learning_rate": 4.9956200428232924e-05, "loss": 1.5026, "step": 18943 }, { "epoch": 0.02, "learning_rate": 4.9956195780571884e-05, "loss": 0.9898, "step": 18944 }, { "epoch": 0.02, "learning_rate": 4.995619113266449e-05, "loss": 1.0833, "step": 18945 }, { "epoch": 0.02, "learning_rate": 4.9956186484510736e-05, "loss": 1.0769, "step": 18946 }, { "epoch": 0.02, "learning_rate": 4.995618183611063e-05, "loss": 0.9637, "step": 18947 }, { "epoch": 0.02, "learning_rate": 4.995617718746417e-05, "loss": 0.9886, "step": 18948 }, { "epoch": 0.02, "learning_rate": 4.995617253857134e-05, "loss": 0.9436, "step": 18949 }, { "epoch": 0.02, "learning_rate": 4.995616788943217e-05, "loss": 1.1846, "step": 18950 }, { "epoch": 0.02, "learning_rate": 4.995616324004663e-05, "loss": 0.9475, "step": 18951 }, { "epoch": 0.02, "learning_rate": 4.9956158590414734e-05, "loss": 1.2991, "step": 18952 }, { "epoch": 0.02, "learning_rate": 4.995615394053649e-05, "loss": 1.104, "step": 18953 }, { "epoch": 0.02, "learning_rate": 4.995614929041189e-05, "loss": 0.9206, "step": 18954 }, { "epoch": 0.02, "learning_rate": 4.9956144640040926e-05, "loss": 1.102, "step": 18955 }, { "epoch": 0.02, "learning_rate": 4.9956139989423605e-05, "loss": 1.18, "step": 18956 }, { "epoch": 0.02, "learning_rate": 4.9956135338559926e-05, "loss": 1.0659, "step": 18957 }, { "epoch": 0.02, "learning_rate": 4.99561306874499e-05, "loss": 1.0928, "step": 18958 }, { "epoch": 0.02, "learning_rate": 4.995612603609351e-05, "loss": 1.2373, "step": 18959 }, { "epoch": 0.02, "learning_rate": 4.9956121384490764e-05, "loss": 1.1252, "step": 18960 }, { "epoch": 0.02, "learning_rate": 4.995611673264167e-05, "loss": 1.0529, "step": 18961 }, { "epoch": 0.02, "learning_rate": 4.9956112080546207e-05, "loss": 1.0768, "step": 18962 }, { "epoch": 0.02, "learning_rate": 4.9956107428204395e-05, "loss": 1.1304, "step": 18963 }, { "epoch": 0.02, "learning_rate": 4.9956102775616225e-05, "loss": 0.384, "step": 18964 }, { "epoch": 0.02, "learning_rate": 4.9956098122781704e-05, "loss": 0.2102, "step": 18965 }, { "epoch": 0.02, "learning_rate": 4.995609346970082e-05, "loss": 0.1633, "step": 18966 }, { "epoch": 0.02, "learning_rate": 4.995608881637358e-05, "loss": 0.2497, "step": 18967 }, { "epoch": 0.02, "learning_rate": 4.995608416279999e-05, "loss": 0.9523, "step": 18968 }, { "epoch": 0.02, "learning_rate": 4.9956079508980035e-05, "loss": 1.0711, "step": 18969 }, { "epoch": 0.02, "learning_rate": 4.995607485491374e-05, "loss": 1.0848, "step": 18970 }, { "epoch": 0.02, "learning_rate": 4.995607020060107e-05, "loss": 0.9964, "step": 18971 }, { "epoch": 0.02, "learning_rate": 4.995606554604205e-05, "loss": 1.3506, "step": 18972 }, { "epoch": 0.02, "learning_rate": 4.995606089123668e-05, "loss": 1.05, "step": 18973 }, { "epoch": 0.02, "learning_rate": 4.995605623618495e-05, "loss": 0.9567, "step": 18974 }, { "epoch": 0.02, "learning_rate": 4.995605158088687e-05, "loss": 1.0788, "step": 18975 }, { "epoch": 0.02, "learning_rate": 4.9956046925342424e-05, "loss": 0.9734, "step": 18976 }, { "epoch": 0.02, "learning_rate": 4.9956042269551636e-05, "loss": 1.1051, "step": 18977 }, { "epoch": 0.02, "learning_rate": 4.9956037613514483e-05, "loss": 1.3113, "step": 18978 }, { "epoch": 0.02, "learning_rate": 4.995603295723098e-05, "loss": 1.2998, "step": 18979 }, { "epoch": 0.02, "learning_rate": 4.995602830070112e-05, "loss": 1.332, "step": 18980 }, { "epoch": 0.02, "learning_rate": 4.99560236439249e-05, "loss": 0.9759, "step": 18981 }, { "epoch": 0.02, "learning_rate": 4.995601898690232e-05, "loss": 0.9934, "step": 18982 }, { "epoch": 0.02, "learning_rate": 4.9956014329633394e-05, "loss": 0.9981, "step": 18983 }, { "epoch": 0.02, "learning_rate": 4.9956009672118115e-05, "loss": 0.9694, "step": 18984 }, { "epoch": 0.02, "learning_rate": 4.995600501435648e-05, "loss": 1.2893, "step": 18985 }, { "epoch": 0.02, "learning_rate": 4.995600035634848e-05, "loss": 1.2033, "step": 18986 }, { "epoch": 0.02, "learning_rate": 4.995599569809414e-05, "loss": 1.2731, "step": 18987 }, { "epoch": 0.02, "learning_rate": 4.995599103959343e-05, "loss": 1.0699, "step": 18988 }, { "epoch": 0.02, "learning_rate": 4.995598638084637e-05, "loss": 1.1132, "step": 18989 }, { "epoch": 0.02, "learning_rate": 4.995598172185296e-05, "loss": 1.1168, "step": 18990 }, { "epoch": 0.02, "learning_rate": 4.995597706261319e-05, "loss": 1.3113, "step": 18991 }, { "epoch": 0.02, "learning_rate": 4.995597240312707e-05, "loss": 1.247, "step": 18992 }, { "epoch": 0.02, "learning_rate": 4.995596774339459e-05, "loss": 1.1948, "step": 18993 }, { "epoch": 0.02, "learning_rate": 4.995596308341576e-05, "loss": 1.2353, "step": 18994 }, { "epoch": 0.02, "learning_rate": 4.995595842319056e-05, "loss": 1.1199, "step": 18995 }, { "epoch": 0.02, "learning_rate": 4.995595376271902e-05, "loss": 1.2796, "step": 18996 }, { "epoch": 0.02, "learning_rate": 4.9955949102001125e-05, "loss": 1.0496, "step": 18997 }, { "epoch": 0.02, "learning_rate": 4.995594444103687e-05, "loss": 0.8329, "step": 18998 }, { "epoch": 0.02, "learning_rate": 4.9955939779826264e-05, "loss": 0.2451, "step": 18999 }, { "epoch": 0.02, "learning_rate": 4.995593511836931e-05, "loss": 0.152, "step": 19000 }, { "epoch": 0.02, "eval_loss": 1.0204665660858154, "eval_runtime": 85.1485, "eval_samples_per_second": 16.266, "eval_steps_per_second": 4.075, "step": 19000 }, { "epoch": 0.02, "learning_rate": 4.995593045666599e-05, "loss": 0.1561, "step": 19001 }, { "epoch": 0.02, "learning_rate": 4.995592579471632e-05, "loss": 0.1704, "step": 19002 }, { "epoch": 0.02, "learning_rate": 4.99559211325203e-05, "loss": 0.1196, "step": 19003 }, { "epoch": 0.02, "learning_rate": 4.995591647007792e-05, "loss": 0.1489, "step": 19004 }, { "epoch": 0.02, "learning_rate": 4.995591180738919e-05, "loss": 0.1265, "step": 19005 }, { "epoch": 0.02, "learning_rate": 4.9955907144454096e-05, "loss": 0.2233, "step": 19006 }, { "epoch": 0.02, "learning_rate": 4.9955902481272655e-05, "loss": 0.1087, "step": 19007 }, { "epoch": 0.02, "learning_rate": 4.995589781784486e-05, "loss": 0.0728, "step": 19008 }, { "epoch": 0.02, "learning_rate": 4.9955893154170704e-05, "loss": 0.082, "step": 19009 }, { "epoch": 0.02, "learning_rate": 4.995588849025021e-05, "loss": 0.1207, "step": 19010 }, { "epoch": 0.02, "learning_rate": 4.995588382608335e-05, "loss": 0.1522, "step": 19011 }, { "epoch": 0.02, "learning_rate": 4.995587916167014e-05, "loss": 0.2727, "step": 19012 }, { "epoch": 0.02, "learning_rate": 4.995587449701057e-05, "loss": 0.182, "step": 19013 }, { "epoch": 0.02, "learning_rate": 4.995586983210465e-05, "loss": 0.5141, "step": 19014 }, { "epoch": 0.02, "learning_rate": 4.995586516695238e-05, "loss": 0.1179, "step": 19015 }, { "epoch": 0.02, "learning_rate": 4.995586050155375e-05, "loss": 0.1046, "step": 19016 }, { "epoch": 0.02, "learning_rate": 4.995585583590877e-05, "loss": 0.0833, "step": 19017 }, { "epoch": 0.02, "learning_rate": 4.9955851170017436e-05, "loss": 0.6373, "step": 19018 }, { "epoch": 0.02, "learning_rate": 4.995584650387975e-05, "loss": 0.9522, "step": 19019 }, { "epoch": 0.02, "learning_rate": 4.995584183749571e-05, "loss": 0.8878, "step": 19020 }, { "epoch": 0.02, "learning_rate": 4.995583717086532e-05, "loss": 1.391, "step": 19021 }, { "epoch": 0.02, "learning_rate": 4.995583250398856e-05, "loss": 1.226, "step": 19022 }, { "epoch": 0.02, "learning_rate": 4.995582783686546e-05, "loss": 1.3243, "step": 19023 }, { "epoch": 0.02, "learning_rate": 4.9955823169496006e-05, "loss": 1.3184, "step": 19024 }, { "epoch": 0.02, "learning_rate": 4.99558185018802e-05, "loss": 1.1628, "step": 19025 }, { "epoch": 0.02, "learning_rate": 4.995581383401803e-05, "loss": 1.1769, "step": 19026 }, { "epoch": 0.02, "learning_rate": 4.9955809165909514e-05, "loss": 1.0807, "step": 19027 }, { "epoch": 0.02, "learning_rate": 4.9955804497554646e-05, "loss": 0.9889, "step": 19028 }, { "epoch": 0.02, "learning_rate": 4.995579982895343e-05, "loss": 1.2633, "step": 19029 }, { "epoch": 0.02, "learning_rate": 4.9955795160105856e-05, "loss": 1.0718, "step": 19030 }, { "epoch": 0.02, "learning_rate": 4.995579049101192e-05, "loss": 1.1655, "step": 19031 }, { "epoch": 0.02, "learning_rate": 4.995578582167164e-05, "loss": 1.1402, "step": 19032 }, { "epoch": 0.02, "learning_rate": 4.9955781152085005e-05, "loss": 1.1924, "step": 19033 }, { "epoch": 0.02, "learning_rate": 4.995577648225202e-05, "loss": 1.2321, "step": 19034 }, { "epoch": 0.02, "learning_rate": 4.995577181217268e-05, "loss": 1.0401, "step": 19035 }, { "epoch": 0.02, "learning_rate": 4.995576714184699e-05, "loss": 1.089, "step": 19036 }, { "epoch": 0.02, "learning_rate": 4.995576247127494e-05, "loss": 1.1472, "step": 19037 }, { "epoch": 0.02, "learning_rate": 4.9955757800456546e-05, "loss": 1.1526, "step": 19038 }, { "epoch": 0.02, "learning_rate": 4.995575312939179e-05, "loss": 1.1022, "step": 19039 }, { "epoch": 0.02, "learning_rate": 4.995574845808069e-05, "loss": 1.1399, "step": 19040 }, { "epoch": 0.02, "learning_rate": 4.995574378652323e-05, "loss": 1.1141, "step": 19041 }, { "epoch": 0.02, "learning_rate": 4.9955739114719425e-05, "loss": 1.2824, "step": 19042 }, { "epoch": 0.02, "learning_rate": 4.995573444266926e-05, "loss": 1.1095, "step": 19043 }, { "epoch": 0.02, "learning_rate": 4.9955729770372745e-05, "loss": 1.2525, "step": 19044 }, { "epoch": 0.02, "learning_rate": 4.9955725097829875e-05, "loss": 1.1549, "step": 19045 }, { "epoch": 0.02, "learning_rate": 4.995572042504066e-05, "loss": 1.2527, "step": 19046 }, { "epoch": 0.02, "learning_rate": 4.995571575200508e-05, "loss": 1.0127, "step": 19047 }, { "epoch": 0.02, "learning_rate": 4.9955711078723164e-05, "loss": 1.0015, "step": 19048 }, { "epoch": 0.02, "learning_rate": 4.995570640519489e-05, "loss": 1.0354, "step": 19049 }, { "epoch": 0.02, "learning_rate": 4.995570173142026e-05, "loss": 1.2093, "step": 19050 }, { "epoch": 0.02, "learning_rate": 4.9955697057399274e-05, "loss": 0.7061, "step": 19051 }, { "epoch": 0.02, "learning_rate": 4.995569238313195e-05, "loss": 1.1595, "step": 19052 }, { "epoch": 0.02, "learning_rate": 4.995568770861826e-05, "loss": 1.2386, "step": 19053 }, { "epoch": 0.02, "learning_rate": 4.9955683033858225e-05, "loss": 1.1142, "step": 19054 }, { "epoch": 0.02, "learning_rate": 4.995567835885183e-05, "loss": 1.1172, "step": 19055 }, { "epoch": 0.02, "learning_rate": 4.995567368359909e-05, "loss": 1.045, "step": 19056 }, { "epoch": 0.02, "learning_rate": 4.99556690081e-05, "loss": 1.1164, "step": 19057 }, { "epoch": 0.02, "learning_rate": 4.9955664332354554e-05, "loss": 0.9473, "step": 19058 }, { "epoch": 0.02, "learning_rate": 4.995565965636276e-05, "loss": 1.8171, "step": 19059 }, { "epoch": 0.02, "learning_rate": 4.995565498012461e-05, "loss": 1.7887, "step": 19060 }, { "epoch": 0.02, "learning_rate": 4.995565030364011e-05, "loss": 1.2405, "step": 19061 }, { "epoch": 0.02, "learning_rate": 4.995564562690927e-05, "loss": 1.1586, "step": 19062 }, { "epoch": 0.02, "learning_rate": 4.995564094993206e-05, "loss": 1.2735, "step": 19063 }, { "epoch": 0.02, "learning_rate": 4.9955636272708503e-05, "loss": 1.174, "step": 19064 }, { "epoch": 0.02, "learning_rate": 4.99556315952386e-05, "loss": 1.1701, "step": 19065 }, { "epoch": 0.02, "learning_rate": 4.995562691752234e-05, "loss": 0.9996, "step": 19066 }, { "epoch": 0.02, "learning_rate": 4.995562223955973e-05, "loss": 1.1356, "step": 19067 }, { "epoch": 0.02, "learning_rate": 4.995561756135077e-05, "loss": 1.0385, "step": 19068 }, { "epoch": 0.02, "learning_rate": 4.995561288289546e-05, "loss": 1.0056, "step": 19069 }, { "epoch": 0.02, "learning_rate": 4.9955608204193794e-05, "loss": 1.1518, "step": 19070 }, { "epoch": 0.02, "learning_rate": 4.995560352524579e-05, "loss": 0.8987, "step": 19071 }, { "epoch": 0.02, "learning_rate": 4.995559884605142e-05, "loss": 1.0456, "step": 19072 }, { "epoch": 0.02, "learning_rate": 4.99555941666107e-05, "loss": 1.027, "step": 19073 }, { "epoch": 0.02, "learning_rate": 4.995558948692363e-05, "loss": 1.1552, "step": 19074 }, { "epoch": 0.02, "learning_rate": 4.995558480699021e-05, "loss": 0.9775, "step": 19075 }, { "epoch": 0.02, "learning_rate": 4.995558012681044e-05, "loss": 0.834, "step": 19076 }, { "epoch": 0.02, "learning_rate": 4.995557544638432e-05, "loss": 1.1551, "step": 19077 }, { "epoch": 0.02, "learning_rate": 4.995557076571185e-05, "loss": 3.9377, "step": 19078 }, { "epoch": 0.02, "learning_rate": 4.9955566084793026e-05, "loss": 3.0507, "step": 19079 }, { "epoch": 0.02, "learning_rate": 4.995556140362786e-05, "loss": 3.5665, "step": 19080 }, { "epoch": 0.02, "learning_rate": 4.995555672221633e-05, "loss": 4.0749, "step": 19081 }, { "epoch": 0.02, "learning_rate": 4.9955552040558453e-05, "loss": 3.454, "step": 19082 }, { "epoch": 0.02, "learning_rate": 4.995554735865423e-05, "loss": 3.5411, "step": 19083 }, { "epoch": 0.02, "learning_rate": 4.9955542676503654e-05, "loss": 3.2418, "step": 19084 }, { "epoch": 0.02, "learning_rate": 4.995553799410673e-05, "loss": 3.5707, "step": 19085 }, { "epoch": 0.02, "learning_rate": 4.995553331146345e-05, "loss": 3.0043, "step": 19086 }, { "epoch": 0.02, "learning_rate": 4.995552862857382e-05, "loss": 3.3118, "step": 19087 }, { "epoch": 0.02, "learning_rate": 4.995552394543784e-05, "loss": 5.9625, "step": 19088 }, { "epoch": 0.02, "learning_rate": 4.995551926205552e-05, "loss": 5.9677, "step": 19089 }, { "epoch": 0.02, "learning_rate": 4.995551457842683e-05, "loss": 5.8623, "step": 19090 }, { "epoch": 0.02, "learning_rate": 4.9955509894551805e-05, "loss": 5.8644, "step": 19091 }, { "epoch": 0.02, "learning_rate": 4.995550521043043e-05, "loss": 5.8618, "step": 19092 }, { "epoch": 0.02, "learning_rate": 4.995550052606269e-05, "loss": 5.8618, "step": 19093 }, { "epoch": 0.02, "learning_rate": 4.995549584144862e-05, "loss": 5.823, "step": 19094 }, { "epoch": 0.02, "learning_rate": 4.9955491156588184e-05, "loss": 5.6292, "step": 19095 }, { "epoch": 0.02, "learning_rate": 4.99554864714814e-05, "loss": 0.5963, "step": 19096 }, { "epoch": 0.02, "learning_rate": 4.995548178612828e-05, "loss": 1.1896, "step": 19097 }, { "epoch": 0.02, "learning_rate": 4.9955477100528794e-05, "loss": 1.1945, "step": 19098 }, { "epoch": 0.02, "learning_rate": 4.995547241468297e-05, "loss": 1.0754, "step": 19099 }, { "epoch": 0.02, "learning_rate": 4.995546772859078e-05, "loss": 1.0528, "step": 19100 }, { "epoch": 0.02, "learning_rate": 4.995546304225225e-05, "loss": 1.1762, "step": 19101 }, { "epoch": 0.02, "learning_rate": 4.995545835566737e-05, "loss": 1.0971, "step": 19102 }, { "epoch": 0.02, "learning_rate": 4.9955453668836146e-05, "loss": 1.2426, "step": 19103 }, { "epoch": 0.02, "learning_rate": 4.995544898175856e-05, "loss": 1.2848, "step": 19104 }, { "epoch": 0.02, "learning_rate": 4.9955444294434636e-05, "loss": 1.1273, "step": 19105 }, { "epoch": 0.02, "learning_rate": 4.995543960686435e-05, "loss": 1.1429, "step": 19106 }, { "epoch": 0.02, "learning_rate": 4.995543491904773e-05, "loss": 0.957, "step": 19107 }, { "epoch": 0.02, "learning_rate": 4.995543023098475e-05, "loss": 1.3764, "step": 19108 }, { "epoch": 0.02, "learning_rate": 4.995542554267542e-05, "loss": 1.1152, "step": 19109 }, { "epoch": 0.02, "learning_rate": 4.995542085411975e-05, "loss": 1.1084, "step": 19110 }, { "epoch": 0.02, "learning_rate": 4.9955416165317714e-05, "loss": 0.9628, "step": 19111 }, { "epoch": 0.02, "learning_rate": 4.995541147626934e-05, "loss": 0.8417, "step": 19112 }, { "epoch": 0.02, "learning_rate": 4.995540678697462e-05, "loss": 0.7174, "step": 19113 }, { "epoch": 0.02, "learning_rate": 4.995540209743355e-05, "loss": 1.3198, "step": 19114 }, { "epoch": 0.02, "learning_rate": 4.9955397407646123e-05, "loss": 0.8946, "step": 19115 }, { "epoch": 0.02, "learning_rate": 4.995539271761235e-05, "loss": 0.9679, "step": 19116 }, { "epoch": 0.02, "learning_rate": 4.995538802733223e-05, "loss": 1.2588, "step": 19117 }, { "epoch": 0.02, "learning_rate": 4.995538333680576e-05, "loss": 1.0634, "step": 19118 }, { "epoch": 0.02, "learning_rate": 4.9955378646032936e-05, "loss": 1.4185, "step": 19119 }, { "epoch": 0.02, "learning_rate": 4.995537395501377e-05, "loss": 1.0541, "step": 19120 }, { "epoch": 0.02, "learning_rate": 4.9955369263748246e-05, "loss": 0.975, "step": 19121 }, { "epoch": 0.02, "learning_rate": 4.9955364572236385e-05, "loss": 0.4344, "step": 19122 }, { "epoch": 0.02, "learning_rate": 4.995535988047817e-05, "loss": 0.8041, "step": 19123 }, { "epoch": 0.02, "learning_rate": 4.99553551884736e-05, "loss": 1.0857, "step": 19124 }, { "epoch": 0.02, "learning_rate": 4.995535049622269e-05, "loss": 0.7406, "step": 19125 }, { "epoch": 0.02, "learning_rate": 4.995534580372543e-05, "loss": 1.3219, "step": 19126 }, { "epoch": 0.02, "learning_rate": 4.995534111098182e-05, "loss": 2.2649, "step": 19127 }, { "epoch": 0.02, "learning_rate": 4.995533641799186e-05, "loss": 2.3423, "step": 19128 }, { "epoch": 0.02, "learning_rate": 4.995533172475555e-05, "loss": 2.2207, "step": 19129 }, { "epoch": 0.02, "learning_rate": 4.995532703127289e-05, "loss": 2.2989, "step": 19130 }, { "epoch": 0.02, "learning_rate": 4.995532233754389e-05, "loss": 2.2645, "step": 19131 }, { "epoch": 0.02, "learning_rate": 4.995531764356853e-05, "loss": 2.1723, "step": 19132 }, { "epoch": 0.02, "learning_rate": 4.9955312949346835e-05, "loss": 2.3898, "step": 19133 }, { "epoch": 0.02, "learning_rate": 4.995530825487879e-05, "loss": 2.1433, "step": 19134 }, { "epoch": 0.02, "learning_rate": 4.995530356016439e-05, "loss": 2.0962, "step": 19135 }, { "epoch": 0.02, "learning_rate": 4.995529886520364e-05, "loss": 1.7601, "step": 19136 }, { "epoch": 0.02, "learning_rate": 4.995529416999655e-05, "loss": 1.3007, "step": 19137 }, { "epoch": 0.02, "learning_rate": 4.99552894745431e-05, "loss": 1.2957, "step": 19138 }, { "epoch": 0.02, "learning_rate": 4.9955284778843314e-05, "loss": 1.0063, "step": 19139 }, { "epoch": 0.02, "learning_rate": 4.9955280082897174e-05, "loss": 1.2809, "step": 19140 }, { "epoch": 0.02, "learning_rate": 4.995527538670469e-05, "loss": 0.9682, "step": 19141 }, { "epoch": 0.02, "learning_rate": 4.9955270690265855e-05, "loss": 1.1679, "step": 19142 }, { "epoch": 0.02, "learning_rate": 4.995526599358067e-05, "loss": 1.2805, "step": 19143 }, { "epoch": 0.02, "learning_rate": 4.995526129664914e-05, "loss": 0.9726, "step": 19144 }, { "epoch": 0.02, "learning_rate": 4.9955256599471264e-05, "loss": 1.3276, "step": 19145 }, { "epoch": 0.02, "learning_rate": 4.995525190204704e-05, "loss": 1.1737, "step": 19146 }, { "epoch": 0.02, "learning_rate": 4.995524720437647e-05, "loss": 0.842, "step": 19147 }, { "epoch": 0.02, "learning_rate": 4.9955242506459546e-05, "loss": 1.6157, "step": 19148 }, { "epoch": 0.02, "learning_rate": 4.9955237808296274e-05, "loss": 0.9117, "step": 19149 }, { "epoch": 0.02, "learning_rate": 4.995523310988666e-05, "loss": 0.9877, "step": 19150 }, { "epoch": 0.02, "learning_rate": 4.995522841123069e-05, "loss": 1.1675, "step": 19151 }, { "epoch": 0.02, "learning_rate": 4.995522371232838e-05, "loss": 1.0004, "step": 19152 }, { "epoch": 0.02, "learning_rate": 4.995521901317972e-05, "loss": 1.4656, "step": 19153 }, { "epoch": 0.02, "learning_rate": 4.9955214313784715e-05, "loss": 1.1162, "step": 19154 }, { "epoch": 0.02, "learning_rate": 4.995520961414336e-05, "loss": 1.0988, "step": 19155 }, { "epoch": 0.02, "learning_rate": 4.995520491425566e-05, "loss": 1.0695, "step": 19156 }, { "epoch": 0.02, "learning_rate": 4.9955200214121614e-05, "loss": 0.9797, "step": 19157 }, { "epoch": 0.02, "learning_rate": 4.9955195513741216e-05, "loss": 0.7575, "step": 19158 }, { "epoch": 0.02, "learning_rate": 4.9955190813114473e-05, "loss": 0.8562, "step": 19159 }, { "epoch": 0.02, "learning_rate": 4.995518611224138e-05, "loss": 1.2865, "step": 19160 }, { "epoch": 0.02, "learning_rate": 4.995518141112194e-05, "loss": 1.0898, "step": 19161 }, { "epoch": 0.02, "learning_rate": 4.995517670975616e-05, "loss": 1.1337, "step": 19162 }, { "epoch": 0.02, "learning_rate": 4.995517200814403e-05, "loss": 1.1592, "step": 19163 }, { "epoch": 0.02, "learning_rate": 4.995516730628555e-05, "loss": 1.1954, "step": 19164 }, { "epoch": 0.02, "learning_rate": 4.995516260418073e-05, "loss": 1.1467, "step": 19165 }, { "epoch": 0.02, "learning_rate": 4.995515790182955e-05, "loss": 1.1322, "step": 19166 }, { "epoch": 0.02, "learning_rate": 4.995515319923203e-05, "loss": 1.1098, "step": 19167 }, { "epoch": 0.02, "learning_rate": 4.995514849638817e-05, "loss": 0.9572, "step": 19168 }, { "epoch": 0.02, "learning_rate": 4.995514379329795e-05, "loss": 0.8791, "step": 19169 }, { "epoch": 0.02, "learning_rate": 4.99551390899614e-05, "loss": 1.1627, "step": 19170 }, { "epoch": 0.02, "learning_rate": 4.995513438637849e-05, "loss": 1.0847, "step": 19171 }, { "epoch": 0.02, "learning_rate": 4.995512968254924e-05, "loss": 1.1937, "step": 19172 }, { "epoch": 0.02, "learning_rate": 4.995512497847364e-05, "loss": 1.0026, "step": 19173 }, { "epoch": 0.02, "learning_rate": 4.9955120274151694e-05, "loss": 0.8382, "step": 19174 }, { "epoch": 0.02, "learning_rate": 4.99551155695834e-05, "loss": 1.1517, "step": 19175 }, { "epoch": 0.02, "learning_rate": 4.9955110864768766e-05, "loss": 1.0642, "step": 19176 }, { "epoch": 0.02, "learning_rate": 4.995510615970778e-05, "loss": 1.2226, "step": 19177 }, { "epoch": 0.02, "learning_rate": 4.9955101454400455e-05, "loss": 0.6325, "step": 19178 }, { "epoch": 0.02, "learning_rate": 4.995509674884677e-05, "loss": 1.4666, "step": 19179 }, { "epoch": 0.02, "learning_rate": 4.995509204304675e-05, "loss": 2.118, "step": 19180 }, { "epoch": 0.02, "learning_rate": 4.995508733700038e-05, "loss": 1.3256, "step": 19181 }, { "epoch": 0.02, "learning_rate": 4.9955082630707674e-05, "loss": 1.4319, "step": 19182 }, { "epoch": 0.02, "learning_rate": 4.995507792416861e-05, "loss": 1.1903, "step": 19183 }, { "epoch": 0.02, "learning_rate": 4.99550732173832e-05, "loss": 1.0023, "step": 19184 }, { "epoch": 0.02, "learning_rate": 4.995506851035144e-05, "loss": 1.085, "step": 19185 }, { "epoch": 0.02, "learning_rate": 4.995506380307335e-05, "loss": 0.907, "step": 19186 }, { "epoch": 0.02, "learning_rate": 4.995505909554891e-05, "loss": 0.9016, "step": 19187 }, { "epoch": 0.02, "learning_rate": 4.9955054387778114e-05, "loss": 1.1099, "step": 19188 }, { "epoch": 0.02, "learning_rate": 4.995504967976098e-05, "loss": 1.1104, "step": 19189 }, { "epoch": 0.02, "learning_rate": 4.995504497149749e-05, "loss": 0.7809, "step": 19190 }, { "epoch": 0.02, "learning_rate": 4.9955040262987665e-05, "loss": 1.0954, "step": 19191 }, { "epoch": 0.02, "learning_rate": 4.995503555423149e-05, "loss": 1.0978, "step": 19192 }, { "epoch": 0.02, "learning_rate": 4.995503084522897e-05, "loss": 0.9413, "step": 19193 }, { "epoch": 0.02, "learning_rate": 4.995502613598011e-05, "loss": 1.2441, "step": 19194 }, { "epoch": 0.02, "learning_rate": 4.995502142648489e-05, "loss": 1.2783, "step": 19195 }, { "epoch": 0.02, "learning_rate": 4.9955016716743345e-05, "loss": 0.9133, "step": 19196 }, { "epoch": 0.02, "learning_rate": 4.9955012006755445e-05, "loss": 0.5905, "step": 19197 }, { "epoch": 0.02, "learning_rate": 4.9955007296521194e-05, "loss": 0.6176, "step": 19198 }, { "epoch": 0.02, "learning_rate": 4.99550025860406e-05, "loss": 0.7252, "step": 19199 }, { "epoch": 0.02, "learning_rate": 4.995499787531367e-05, "loss": 0.5752, "step": 19200 }, { "epoch": 0.02, "learning_rate": 4.995499316434038e-05, "loss": 1.1965, "step": 19201 }, { "epoch": 0.02, "learning_rate": 4.9954988453120755e-05, "loss": 0.9561, "step": 19202 }, { "epoch": 0.02, "learning_rate": 4.995498374165478e-05, "loss": 0.9637, "step": 19203 }, { "epoch": 0.02, "learning_rate": 4.995497902994246e-05, "loss": 1.0664, "step": 19204 }, { "epoch": 0.02, "learning_rate": 4.9954974317983806e-05, "loss": 1.0012, "step": 19205 }, { "epoch": 0.02, "learning_rate": 4.995496960577879e-05, "loss": 1.0365, "step": 19206 }, { "epoch": 0.02, "learning_rate": 4.995496489332744e-05, "loss": 0.9338, "step": 19207 }, { "epoch": 0.02, "learning_rate": 4.995496018062974e-05, "loss": 1.1256, "step": 19208 }, { "epoch": 0.02, "learning_rate": 4.99549554676857e-05, "loss": 1.244, "step": 19209 }, { "epoch": 0.02, "learning_rate": 4.995495075449531e-05, "loss": 0.8106, "step": 19210 }, { "epoch": 0.02, "learning_rate": 4.995494604105858e-05, "loss": 0.9954, "step": 19211 }, { "epoch": 0.02, "learning_rate": 4.995494132737551e-05, "loss": 1.1606, "step": 19212 }, { "epoch": 0.02, "learning_rate": 4.995493661344608e-05, "loss": 1.2498, "step": 19213 }, { "epoch": 0.02, "learning_rate": 4.995493189927032e-05, "loss": 1.7133, "step": 19214 }, { "epoch": 0.02, "learning_rate": 4.99549271848482e-05, "loss": 2.2877, "step": 19215 }, { "epoch": 0.02, "learning_rate": 4.9954922470179754e-05, "loss": 2.1447, "step": 19216 }, { "epoch": 0.02, "learning_rate": 4.995491775526495e-05, "loss": 1.3295, "step": 19217 }, { "epoch": 0.02, "learning_rate": 4.9954913040103806e-05, "loss": 1.0012, "step": 19218 }, { "epoch": 0.02, "learning_rate": 4.995490832469632e-05, "loss": 0.9109, "step": 19219 }, { "epoch": 0.02, "learning_rate": 4.9954903609042486e-05, "loss": 0.5036, "step": 19220 }, { "epoch": 0.02, "learning_rate": 4.995489889314231e-05, "loss": 2.0009, "step": 19221 }, { "epoch": 0.02, "learning_rate": 4.995489417699579e-05, "loss": 1.3429, "step": 19222 }, { "epoch": 0.02, "learning_rate": 4.995488946060292e-05, "loss": 1.0764, "step": 19223 }, { "epoch": 0.02, "learning_rate": 4.9954884743963716e-05, "loss": 1.4857, "step": 19224 }, { "epoch": 0.02, "learning_rate": 4.9954880027078156e-05, "loss": 1.2899, "step": 19225 }, { "epoch": 0.02, "learning_rate": 4.995487530994626e-05, "loss": 0.721, "step": 19226 }, { "epoch": 0.02, "learning_rate": 4.995487059256802e-05, "loss": 1.0742, "step": 19227 }, { "epoch": 0.02, "learning_rate": 4.995486587494344e-05, "loss": 0.8446, "step": 19228 }, { "epoch": 0.02, "learning_rate": 4.99548611570725e-05, "loss": 0.753, "step": 19229 }, { "epoch": 0.02, "learning_rate": 4.995485643895523e-05, "loss": 0.4958, "step": 19230 }, { "epoch": 0.02, "learning_rate": 4.9954851720591614e-05, "loss": 0.4861, "step": 19231 }, { "epoch": 0.02, "learning_rate": 4.995484700198165e-05, "loss": 0.4741, "step": 19232 }, { "epoch": 0.02, "learning_rate": 4.995484228312535e-05, "loss": 0.4521, "step": 19233 }, { "epoch": 0.02, "learning_rate": 4.995483756402269e-05, "loss": 1.2927, "step": 19234 }, { "epoch": 0.02, "learning_rate": 4.99548328446737e-05, "loss": 1.245, "step": 19235 }, { "epoch": 0.02, "learning_rate": 4.995482812507837e-05, "loss": 1.1217, "step": 19236 }, { "epoch": 0.02, "learning_rate": 4.995482340523669e-05, "loss": 0.9203, "step": 19237 }, { "epoch": 0.02, "learning_rate": 4.995481868514866e-05, "loss": 1.1989, "step": 19238 }, { "epoch": 0.02, "learning_rate": 4.99548139648143e-05, "loss": 1.0409, "step": 19239 }, { "epoch": 0.02, "learning_rate": 4.995480924423359e-05, "loss": 1.045, "step": 19240 }, { "epoch": 0.02, "learning_rate": 4.995480452340654e-05, "loss": 0.9666, "step": 19241 }, { "epoch": 0.02, "learning_rate": 4.9954799802333144e-05, "loss": 1.1373, "step": 19242 }, { "epoch": 0.02, "learning_rate": 4.99547950810134e-05, "loss": 1.3707, "step": 19243 }, { "epoch": 0.02, "learning_rate": 4.995479035944732e-05, "loss": 1.1284, "step": 19244 }, { "epoch": 0.02, "learning_rate": 4.9954785637634894e-05, "loss": 0.9551, "step": 19245 }, { "epoch": 0.02, "learning_rate": 4.9954780915576124e-05, "loss": 1.0073, "step": 19246 }, { "epoch": 0.02, "learning_rate": 4.995477619327101e-05, "loss": 1.0787, "step": 19247 }, { "epoch": 0.02, "learning_rate": 4.995477147071955e-05, "loss": 0.5148, "step": 19248 }, { "epoch": 0.02, "learning_rate": 4.9954766747921756e-05, "loss": 1.3025, "step": 19249 }, { "epoch": 0.02, "learning_rate": 4.9954762024877616e-05, "loss": 0.984, "step": 19250 }, { "epoch": 0.02, "learning_rate": 4.995475730158713e-05, "loss": 0.9945, "step": 19251 }, { "epoch": 0.02, "learning_rate": 4.99547525780503e-05, "loss": 1.0042, "step": 19252 }, { "epoch": 0.02, "learning_rate": 4.995474785426714e-05, "loss": 0.9965, "step": 19253 }, { "epoch": 0.02, "learning_rate": 4.995474313023763e-05, "loss": 1.1785, "step": 19254 }, { "epoch": 0.02, "learning_rate": 4.9954738405961765e-05, "loss": 1.0819, "step": 19255 }, { "epoch": 0.02, "learning_rate": 4.9954733681439566e-05, "loss": 1.0069, "step": 19256 }, { "epoch": 0.02, "learning_rate": 4.995472895667103e-05, "loss": 0.7242, "step": 19257 }, { "epoch": 0.02, "learning_rate": 4.995472423165615e-05, "loss": 1.3509, "step": 19258 }, { "epoch": 0.02, "learning_rate": 4.995471950639492e-05, "loss": 1.0406, "step": 19259 }, { "epoch": 0.02, "learning_rate": 4.9954714780887355e-05, "loss": 1.0825, "step": 19260 }, { "epoch": 0.02, "learning_rate": 4.995471005513344e-05, "loss": 1.2383, "step": 19261 }, { "epoch": 0.02, "learning_rate": 4.995470532913319e-05, "loss": 1.0412, "step": 19262 }, { "epoch": 0.02, "learning_rate": 4.9954700602886594e-05, "loss": 1.1885, "step": 19263 }, { "epoch": 0.02, "learning_rate": 4.9954695876393655e-05, "loss": 1.029, "step": 19264 }, { "epoch": 0.02, "learning_rate": 4.995469114965438e-05, "loss": 1.0229, "step": 19265 }, { "epoch": 0.02, "learning_rate": 4.995468642266875e-05, "loss": 1.1967, "step": 19266 }, { "epoch": 0.02, "learning_rate": 4.995468169543679e-05, "loss": 1.2703, "step": 19267 }, { "epoch": 0.02, "learning_rate": 4.995467696795848e-05, "loss": 1.3139, "step": 19268 }, { "epoch": 0.02, "learning_rate": 4.995467224023383e-05, "loss": 1.4161, "step": 19269 }, { "epoch": 0.02, "learning_rate": 4.995466751226284e-05, "loss": 1.2393, "step": 19270 }, { "epoch": 0.02, "learning_rate": 4.995466278404551e-05, "loss": 0.8414, "step": 19271 }, { "epoch": 0.02, "learning_rate": 4.995465805558184e-05, "loss": 0.9599, "step": 19272 }, { "epoch": 0.02, "learning_rate": 4.995465332687182e-05, "loss": 1.415, "step": 19273 }, { "epoch": 0.02, "learning_rate": 4.9954648597915464e-05, "loss": 0.8591, "step": 19274 }, { "epoch": 0.02, "learning_rate": 4.9954643868712764e-05, "loss": 0.5772, "step": 19275 }, { "epoch": 0.02, "learning_rate": 4.995463913926373e-05, "loss": 0.9447, "step": 19276 }, { "epoch": 0.02, "learning_rate": 4.995463440956833e-05, "loss": 1.0767, "step": 19277 }, { "epoch": 0.02, "learning_rate": 4.995462967962661e-05, "loss": 0.8706, "step": 19278 }, { "epoch": 0.02, "learning_rate": 4.995462494943854e-05, "loss": 1.2448, "step": 19279 }, { "epoch": 0.02, "learning_rate": 4.995462021900414e-05, "loss": 1.2057, "step": 19280 }, { "epoch": 0.02, "learning_rate": 4.995461548832338e-05, "loss": 0.9464, "step": 19281 }, { "epoch": 0.02, "learning_rate": 4.99546107573963e-05, "loss": 1.2164, "step": 19282 }, { "epoch": 0.02, "learning_rate": 4.995460602622287e-05, "loss": 0.6809, "step": 19283 }, { "epoch": 0.02, "learning_rate": 4.9954601294803096e-05, "loss": 0.9319, "step": 19284 }, { "epoch": 0.02, "learning_rate": 4.995459656313698e-05, "loss": 0.8735, "step": 19285 }, { "epoch": 0.02, "learning_rate": 4.995459183122452e-05, "loss": 1.0454, "step": 19286 }, { "epoch": 0.02, "learning_rate": 4.995458709906572e-05, "loss": 1.0508, "step": 19287 }, { "epoch": 0.02, "learning_rate": 4.995458236666059e-05, "loss": 1.2713, "step": 19288 }, { "epoch": 0.02, "learning_rate": 4.99545776340091e-05, "loss": 1.0254, "step": 19289 }, { "epoch": 0.02, "learning_rate": 4.9954572901111286e-05, "loss": 1.0027, "step": 19290 }, { "epoch": 0.02, "learning_rate": 4.9954568167967125e-05, "loss": 1.1721, "step": 19291 }, { "epoch": 0.02, "learning_rate": 4.9954563434576627e-05, "loss": 1.0835, "step": 19292 }, { "epoch": 0.02, "learning_rate": 4.995455870093978e-05, "loss": 1.1481, "step": 19293 }, { "epoch": 0.02, "learning_rate": 4.99545539670566e-05, "loss": 1.1336, "step": 19294 }, { "epoch": 0.02, "learning_rate": 4.995454923292707e-05, "loss": 1.2392, "step": 19295 }, { "epoch": 0.02, "learning_rate": 4.9954544498551204e-05, "loss": 1.0011, "step": 19296 }, { "epoch": 0.02, "learning_rate": 4.9954539763929e-05, "loss": 0.9099, "step": 19297 }, { "epoch": 0.02, "learning_rate": 4.995453502906045e-05, "loss": 0.9345, "step": 19298 }, { "epoch": 0.02, "learning_rate": 4.995453029394556e-05, "loss": 0.8692, "step": 19299 }, { "epoch": 0.02, "learning_rate": 4.995452555858433e-05, "loss": 0.8648, "step": 19300 }, { "epoch": 0.02, "learning_rate": 4.9954520822976756e-05, "loss": 1.0525, "step": 19301 }, { "epoch": 0.02, "learning_rate": 4.995451608712285e-05, "loss": 1.2489, "step": 19302 }, { "epoch": 0.02, "learning_rate": 4.9954511351022604e-05, "loss": 1.1957, "step": 19303 }, { "epoch": 0.02, "learning_rate": 4.995450661467601e-05, "loss": 1.1293, "step": 19304 }, { "epoch": 0.02, "learning_rate": 4.995450187808308e-05, "loss": 1.4837, "step": 19305 }, { "epoch": 0.02, "learning_rate": 4.9954497141243804e-05, "loss": 1.0879, "step": 19306 }, { "epoch": 0.02, "learning_rate": 4.9954492404158195e-05, "loss": 1.1045, "step": 19307 }, { "epoch": 0.02, "learning_rate": 4.995448766682624e-05, "loss": 1.3895, "step": 19308 }, { "epoch": 0.02, "learning_rate": 4.9954482929247945e-05, "loss": 1.0971, "step": 19309 }, { "epoch": 0.02, "learning_rate": 4.995447819142332e-05, "loss": 1.3849, "step": 19310 }, { "epoch": 0.02, "learning_rate": 4.9954473453352346e-05, "loss": 1.1698, "step": 19311 }, { "epoch": 0.02, "learning_rate": 4.995446871503503e-05, "loss": 0.708, "step": 19312 }, { "epoch": 0.02, "learning_rate": 4.9954463976471375e-05, "loss": 0.8596, "step": 19313 }, { "epoch": 0.02, "learning_rate": 4.995445923766139e-05, "loss": 0.9283, "step": 19314 }, { "epoch": 0.02, "learning_rate": 4.9954454498605055e-05, "loss": 0.9766, "step": 19315 }, { "epoch": 0.02, "learning_rate": 4.995444975930238e-05, "loss": 1.1457, "step": 19316 }, { "epoch": 0.02, "learning_rate": 4.995444501975337e-05, "loss": 0.9783, "step": 19317 }, { "epoch": 0.02, "learning_rate": 4.995444027995802e-05, "loss": 1.2832, "step": 19318 }, { "epoch": 0.02, "learning_rate": 4.995443553991633e-05, "loss": 1.1363, "step": 19319 }, { "epoch": 0.02, "learning_rate": 4.9954430799628294e-05, "loss": 1.2653, "step": 19320 }, { "epoch": 0.02, "learning_rate": 4.9954426059093926e-05, "loss": 1.0693, "step": 19321 }, { "epoch": 0.02, "learning_rate": 4.9954421318313214e-05, "loss": 1.4815, "step": 19322 }, { "epoch": 0.02, "learning_rate": 4.9954416577286165e-05, "loss": 1.1423, "step": 19323 }, { "epoch": 0.02, "learning_rate": 4.995441183601278e-05, "loss": 1.0474, "step": 19324 }, { "epoch": 0.02, "learning_rate": 4.995440709449305e-05, "loss": 0.9858, "step": 19325 }, { "epoch": 0.02, "learning_rate": 4.995440235272698e-05, "loss": 1.1529, "step": 19326 }, { "epoch": 0.02, "learning_rate": 4.995439761071457e-05, "loss": 1.23, "step": 19327 }, { "epoch": 0.02, "learning_rate": 4.995439286845582e-05, "loss": 1.5527, "step": 19328 }, { "epoch": 0.02, "learning_rate": 4.9954388125950734e-05, "loss": 1.0115, "step": 19329 }, { "epoch": 0.02, "learning_rate": 4.995438338319931e-05, "loss": 0.9263, "step": 19330 }, { "epoch": 0.02, "learning_rate": 4.9954378640201546e-05, "loss": 0.9061, "step": 19331 }, { "epoch": 0.02, "learning_rate": 4.995437389695744e-05, "loss": 0.9853, "step": 19332 }, { "epoch": 0.02, "learning_rate": 4.9954369153467e-05, "loss": 1.196, "step": 19333 }, { "epoch": 0.02, "learning_rate": 4.995436440973022e-05, "loss": 0.9523, "step": 19334 }, { "epoch": 0.02, "learning_rate": 4.99543596657471e-05, "loss": 0.8554, "step": 19335 }, { "epoch": 0.02, "learning_rate": 4.995435492151764e-05, "loss": 1.0282, "step": 19336 }, { "epoch": 0.02, "learning_rate": 4.995435017704184e-05, "loss": 1.1775, "step": 19337 }, { "epoch": 0.02, "learning_rate": 4.995434543231971e-05, "loss": 1.0784, "step": 19338 }, { "epoch": 0.02, "learning_rate": 4.995434068735123e-05, "loss": 1.1094, "step": 19339 }, { "epoch": 0.02, "learning_rate": 4.995433594213641e-05, "loss": 1.0715, "step": 19340 }, { "epoch": 0.02, "learning_rate": 4.995433119667526e-05, "loss": 1.1213, "step": 19341 }, { "epoch": 0.02, "learning_rate": 4.995432645096777e-05, "loss": 1.1566, "step": 19342 }, { "epoch": 0.02, "learning_rate": 4.995432170501394e-05, "loss": 1.2043, "step": 19343 }, { "epoch": 0.02, "learning_rate": 4.9954316958813765e-05, "loss": 1.1413, "step": 19344 }, { "epoch": 0.02, "learning_rate": 4.9954312212367264e-05, "loss": 1.124, "step": 19345 }, { "epoch": 0.02, "learning_rate": 4.995430746567441e-05, "loss": 1.0499, "step": 19346 }, { "epoch": 0.02, "learning_rate": 4.995430271873523e-05, "loss": 1.2526, "step": 19347 }, { "epoch": 0.02, "learning_rate": 4.99542979715497e-05, "loss": 1.1216, "step": 19348 }, { "epoch": 0.02, "learning_rate": 4.995429322411784e-05, "loss": 1.119, "step": 19349 }, { "epoch": 0.02, "learning_rate": 4.9954288476439645e-05, "loss": 1.117, "step": 19350 }, { "epoch": 0.02, "learning_rate": 4.9954283728515105e-05, "loss": 1.2026, "step": 19351 }, { "epoch": 0.02, "learning_rate": 4.995427898034423e-05, "loss": 1.2079, "step": 19352 }, { "epoch": 0.02, "learning_rate": 4.995427423192702e-05, "loss": 1.2886, "step": 19353 }, { "epoch": 0.02, "learning_rate": 4.995426948326346e-05, "loss": 1.0737, "step": 19354 }, { "epoch": 0.02, "learning_rate": 4.9954264734353576e-05, "loss": 1.2677, "step": 19355 }, { "epoch": 0.02, "learning_rate": 4.995425998519735e-05, "loss": 1.1583, "step": 19356 }, { "epoch": 0.02, "learning_rate": 4.9954255235794776e-05, "loss": 0.6243, "step": 19357 }, { "epoch": 0.02, "learning_rate": 4.995425048614587e-05, "loss": 0.9922, "step": 19358 }, { "epoch": 0.02, "learning_rate": 4.995424573625063e-05, "loss": 1.06, "step": 19359 }, { "epoch": 0.02, "learning_rate": 4.9954240986109055e-05, "loss": 1.2124, "step": 19360 }, { "epoch": 0.02, "learning_rate": 4.995423623572113e-05, "loss": 1.0914, "step": 19361 }, { "epoch": 0.02, "learning_rate": 4.995423148508688e-05, "loss": 0.83, "step": 19362 }, { "epoch": 0.02, "learning_rate": 4.995422673420629e-05, "loss": 1.0264, "step": 19363 }, { "epoch": 0.02, "learning_rate": 4.9954221983079355e-05, "loss": 0.9966, "step": 19364 }, { "epoch": 0.02, "learning_rate": 4.995421723170609e-05, "loss": 0.8204, "step": 19365 }, { "epoch": 0.02, "learning_rate": 4.995421248008648e-05, "loss": 1.1324, "step": 19366 }, { "epoch": 0.02, "learning_rate": 4.995420772822054e-05, "loss": 1.1546, "step": 19367 }, { "epoch": 0.02, "learning_rate": 4.995420297610826e-05, "loss": 1.258, "step": 19368 }, { "epoch": 0.02, "learning_rate": 4.9954198223749646e-05, "loss": 1.0821, "step": 19369 }, { "epoch": 0.02, "learning_rate": 4.995419347114469e-05, "loss": 1.2707, "step": 19370 }, { "epoch": 0.02, "learning_rate": 4.9954188718293396e-05, "loss": 1.1921, "step": 19371 }, { "epoch": 0.02, "learning_rate": 4.995418396519577e-05, "loss": 1.3018, "step": 19372 }, { "epoch": 0.02, "learning_rate": 4.9954179211851796e-05, "loss": 0.868, "step": 19373 }, { "epoch": 0.02, "learning_rate": 4.995417445826149e-05, "loss": 1.1215, "step": 19374 }, { "epoch": 0.02, "learning_rate": 4.995416970442485e-05, "loss": 0.977, "step": 19375 }, { "epoch": 0.02, "learning_rate": 4.9954164950341875e-05, "loss": 1.0759, "step": 19376 }, { "epoch": 0.02, "learning_rate": 4.995416019601256e-05, "loss": 1.0187, "step": 19377 }, { "epoch": 0.02, "learning_rate": 4.995415544143691e-05, "loss": 0.3381, "step": 19378 }, { "epoch": 0.02, "learning_rate": 4.995415068661492e-05, "loss": 0.8756, "step": 19379 }, { "epoch": 0.02, "learning_rate": 4.9954145931546594e-05, "loss": 0.9663, "step": 19380 }, { "epoch": 0.02, "learning_rate": 4.995414117623193e-05, "loss": 1.0871, "step": 19381 }, { "epoch": 0.02, "learning_rate": 4.9954136420670926e-05, "loss": 1.3809, "step": 19382 }, { "epoch": 0.02, "learning_rate": 4.99541316648636e-05, "loss": 1.5835, "step": 19383 }, { "epoch": 0.02, "learning_rate": 4.995412690880992e-05, "loss": 1.0434, "step": 19384 }, { "epoch": 0.02, "learning_rate": 4.995412215250991e-05, "loss": 1.0567, "step": 19385 }, { "epoch": 0.02, "learning_rate": 4.9954117395963566e-05, "loss": 1.0762, "step": 19386 }, { "epoch": 0.02, "learning_rate": 4.995411263917088e-05, "loss": 1.0567, "step": 19387 }, { "epoch": 0.02, "learning_rate": 4.995410788213186e-05, "loss": 0.8567, "step": 19388 }, { "epoch": 0.02, "learning_rate": 4.995410312484651e-05, "loss": 0.9516, "step": 19389 }, { "epoch": 0.02, "learning_rate": 4.995409836731482e-05, "loss": 1.1749, "step": 19390 }, { "epoch": 0.02, "learning_rate": 4.995409360953679e-05, "loss": 1.2256, "step": 19391 }, { "epoch": 0.02, "learning_rate": 4.995408885151243e-05, "loss": 0.8389, "step": 19392 }, { "epoch": 0.02, "learning_rate": 4.9954084093241725e-05, "loss": 0.8863, "step": 19393 }, { "epoch": 0.02, "learning_rate": 4.995407933472469e-05, "loss": 0.577, "step": 19394 }, { "epoch": 0.02, "learning_rate": 4.9954074575961315e-05, "loss": 0.9026, "step": 19395 }, { "epoch": 0.02, "learning_rate": 4.995406981695161e-05, "loss": 0.9742, "step": 19396 }, { "epoch": 0.02, "learning_rate": 4.995406505769556e-05, "loss": 1.0701, "step": 19397 }, { "epoch": 0.02, "learning_rate": 4.995406029819318e-05, "loss": 1.0514, "step": 19398 }, { "epoch": 0.02, "learning_rate": 4.995405553844446e-05, "loss": 1.2353, "step": 19399 }, { "epoch": 0.02, "learning_rate": 4.995405077844941e-05, "loss": 1.0002, "step": 19400 }, { "epoch": 0.02, "learning_rate": 4.9954046018208026e-05, "loss": 0.8695, "step": 19401 }, { "epoch": 0.02, "learning_rate": 4.99540412577203e-05, "loss": 0.5463, "step": 19402 }, { "epoch": 0.02, "learning_rate": 4.9954036496986236e-05, "loss": 1.0383, "step": 19403 }, { "epoch": 0.02, "learning_rate": 4.9954031736005835e-05, "loss": 1.3333, "step": 19404 }, { "epoch": 0.02, "learning_rate": 4.99540269747791e-05, "loss": 1.028, "step": 19405 }, { "epoch": 0.02, "learning_rate": 4.995402221330604e-05, "loss": 1.0235, "step": 19406 }, { "epoch": 0.02, "learning_rate": 4.9954017451586634e-05, "loss": 1.0162, "step": 19407 }, { "epoch": 0.02, "learning_rate": 4.9954012689620896e-05, "loss": 0.9853, "step": 19408 }, { "epoch": 0.02, "learning_rate": 4.995400792740882e-05, "loss": 0.8552, "step": 19409 }, { "epoch": 0.02, "learning_rate": 4.9954003164950415e-05, "loss": 0.9314, "step": 19410 }, { "epoch": 0.02, "learning_rate": 4.9953998402245665e-05, "loss": 0.6745, "step": 19411 }, { "epoch": 0.02, "learning_rate": 4.9953993639294584e-05, "loss": 1.0115, "step": 19412 }, { "epoch": 0.02, "learning_rate": 4.9953988876097165e-05, "loss": 1.0986, "step": 19413 }, { "epoch": 0.02, "learning_rate": 4.9953984112653416e-05, "loss": 1.1983, "step": 19414 }, { "epoch": 0.02, "learning_rate": 4.995397934896333e-05, "loss": 0.9997, "step": 19415 }, { "epoch": 0.02, "learning_rate": 4.9953974585026905e-05, "loss": 0.5433, "step": 19416 }, { "epoch": 0.02, "learning_rate": 4.995396982084416e-05, "loss": 0.4551, "step": 19417 }, { "epoch": 0.02, "learning_rate": 4.9953965056415065e-05, "loss": 0.983, "step": 19418 }, { "epoch": 0.02, "learning_rate": 4.9953960291739635e-05, "loss": 1.3413, "step": 19419 }, { "epoch": 0.02, "learning_rate": 4.9953955526817874e-05, "loss": 0.9104, "step": 19420 }, { "epoch": 0.02, "learning_rate": 4.9953950761649776e-05, "loss": 1.1702, "step": 19421 }, { "epoch": 0.02, "learning_rate": 4.995394599623534e-05, "loss": 0.9552, "step": 19422 }, { "epoch": 0.02, "learning_rate": 4.9953941230574574e-05, "loss": 1.1938, "step": 19423 }, { "epoch": 0.02, "learning_rate": 4.995393646466748e-05, "loss": 1.0448, "step": 19424 }, { "epoch": 0.02, "learning_rate": 4.995393169851404e-05, "loss": 1.392, "step": 19425 }, { "epoch": 0.02, "learning_rate": 4.995392693211427e-05, "loss": 0.5965, "step": 19426 }, { "epoch": 0.02, "learning_rate": 4.995392216546817e-05, "loss": 0.2987, "step": 19427 }, { "epoch": 0.02, "learning_rate": 4.995391739857572e-05, "loss": 0.3413, "step": 19428 }, { "epoch": 0.02, "learning_rate": 4.995391263143695e-05, "loss": 0.9478, "step": 19429 }, { "epoch": 0.02, "learning_rate": 4.995390786405184e-05, "loss": 1.0214, "step": 19430 }, { "epoch": 0.02, "learning_rate": 4.9953903096420394e-05, "loss": 1.0263, "step": 19431 }, { "epoch": 0.02, "learning_rate": 4.995389832854262e-05, "loss": 1.1779, "step": 19432 }, { "epoch": 0.02, "learning_rate": 4.995389356041851e-05, "loss": 1.4795, "step": 19433 }, { "epoch": 0.02, "learning_rate": 4.9953888792048064e-05, "loss": 1.0849, "step": 19434 }, { "epoch": 0.02, "learning_rate": 4.995388402343128e-05, "loss": 1.2079, "step": 19435 }, { "epoch": 0.02, "learning_rate": 4.995387925456817e-05, "loss": 1.1864, "step": 19436 }, { "epoch": 0.02, "learning_rate": 4.995387448545872e-05, "loss": 1.1666, "step": 19437 }, { "epoch": 0.02, "learning_rate": 4.995386971610293e-05, "loss": 1.2221, "step": 19438 }, { "epoch": 0.02, "learning_rate": 4.995386494650082e-05, "loss": 0.9236, "step": 19439 }, { "epoch": 0.02, "learning_rate": 4.995386017665237e-05, "loss": 0.633, "step": 19440 }, { "epoch": 0.02, "learning_rate": 4.995385540655758e-05, "loss": 0.464, "step": 19441 }, { "epoch": 0.02, "learning_rate": 4.995385063621646e-05, "loss": 0.5513, "step": 19442 }, { "epoch": 0.02, "learning_rate": 4.9953845865629006e-05, "loss": 0.3287, "step": 19443 }, { "epoch": 0.02, "learning_rate": 4.995384109479522e-05, "loss": 0.5287, "step": 19444 }, { "epoch": 0.02, "learning_rate": 4.9953836323715096e-05, "loss": 0.8404, "step": 19445 }, { "epoch": 0.02, "learning_rate": 4.995383155238864e-05, "loss": 1.1239, "step": 19446 }, { "epoch": 0.02, "learning_rate": 4.9953826780815856e-05, "loss": 1.4788, "step": 19447 }, { "epoch": 0.02, "learning_rate": 4.995382200899673e-05, "loss": 1.1838, "step": 19448 }, { "epoch": 0.02, "learning_rate": 4.995381723693128e-05, "loss": 0.8796, "step": 19449 }, { "epoch": 0.02, "learning_rate": 4.995381246461949e-05, "loss": 0.9667, "step": 19450 }, { "epoch": 0.02, "learning_rate": 4.995380769206136e-05, "loss": 1.2038, "step": 19451 }, { "epoch": 0.02, "learning_rate": 4.995380291925691e-05, "loss": 1.1696, "step": 19452 }, { "epoch": 0.02, "learning_rate": 4.9953798146206124e-05, "loss": 1.4705, "step": 19453 }, { "epoch": 0.02, "learning_rate": 4.9953793372908996e-05, "loss": 1.1468, "step": 19454 }, { "epoch": 0.02, "learning_rate": 4.995378859936554e-05, "loss": 1.23, "step": 19455 }, { "epoch": 0.02, "learning_rate": 4.995378382557575e-05, "loss": 1.3248, "step": 19456 }, { "epoch": 0.02, "learning_rate": 4.995377905153963e-05, "loss": 1.3197, "step": 19457 }, { "epoch": 0.02, "learning_rate": 4.995377427725717e-05, "loss": 1.0324, "step": 19458 }, { "epoch": 0.02, "learning_rate": 4.9953769502728384e-05, "loss": 1.0601, "step": 19459 }, { "epoch": 0.02, "learning_rate": 4.995376472795326e-05, "loss": 0.994, "step": 19460 }, { "epoch": 0.02, "learning_rate": 4.99537599529318e-05, "loss": 1.1759, "step": 19461 }, { "epoch": 0.02, "learning_rate": 4.995375517766402e-05, "loss": 1.1833, "step": 19462 }, { "epoch": 0.02, "learning_rate": 4.995375040214989e-05, "loss": 1.0677, "step": 19463 }, { "epoch": 0.02, "learning_rate": 4.9953745626389435e-05, "loss": 1.1962, "step": 19464 }, { "epoch": 0.02, "learning_rate": 4.995374085038265e-05, "loss": 0.7766, "step": 19465 }, { "epoch": 0.02, "learning_rate": 4.995373607412953e-05, "loss": 0.9128, "step": 19466 }, { "epoch": 0.02, "learning_rate": 4.995373129763008e-05, "loss": 0.995, "step": 19467 }, { "epoch": 0.02, "learning_rate": 4.9953726520884294e-05, "loss": 1.2128, "step": 19468 }, { "epoch": 0.02, "learning_rate": 4.995372174389218e-05, "loss": 0.8503, "step": 19469 }, { "epoch": 0.02, "learning_rate": 4.9953716966653725e-05, "loss": 1.0194, "step": 19470 }, { "epoch": 0.02, "learning_rate": 4.995371218916894e-05, "loss": 1.2112, "step": 19471 }, { "epoch": 0.02, "learning_rate": 4.995370741143782e-05, "loss": 1.0213, "step": 19472 }, { "epoch": 0.02, "learning_rate": 4.9953702633460374e-05, "loss": 1.1726, "step": 19473 }, { "epoch": 0.02, "learning_rate": 4.99536978552366e-05, "loss": 1.1736, "step": 19474 }, { "epoch": 0.02, "learning_rate": 4.9953693076766484e-05, "loss": 1.0172, "step": 19475 }, { "epoch": 0.02, "learning_rate": 4.995368829805004e-05, "loss": 1.0514, "step": 19476 }, { "epoch": 0.02, "learning_rate": 4.995368351908726e-05, "loss": 1.0098, "step": 19477 }, { "epoch": 0.02, "learning_rate": 4.995367873987815e-05, "loss": 1.3219, "step": 19478 }, { "epoch": 0.02, "learning_rate": 4.995367396042271e-05, "loss": 1.1564, "step": 19479 }, { "epoch": 0.02, "learning_rate": 4.995366918072093e-05, "loss": 1.027, "step": 19480 }, { "epoch": 0.02, "learning_rate": 4.995366440077282e-05, "loss": 0.9331, "step": 19481 }, { "epoch": 0.02, "learning_rate": 4.995365962057839e-05, "loss": 0.926, "step": 19482 }, { "epoch": 0.02, "learning_rate": 4.995365484013762e-05, "loss": 0.801, "step": 19483 }, { "epoch": 0.02, "learning_rate": 4.995365005945052e-05, "loss": 1.0164, "step": 19484 }, { "epoch": 0.02, "learning_rate": 4.995364527851708e-05, "loss": 1.2393, "step": 19485 }, { "epoch": 0.02, "learning_rate": 4.995364049733732e-05, "loss": 0.6805, "step": 19486 }, { "epoch": 0.02, "learning_rate": 4.9953635715911215e-05, "loss": 0.2952, "step": 19487 }, { "epoch": 0.02, "learning_rate": 4.995363093423878e-05, "loss": 0.2485, "step": 19488 }, { "epoch": 0.02, "learning_rate": 4.9953626152320024e-05, "loss": 0.2909, "step": 19489 }, { "epoch": 0.02, "learning_rate": 4.995362137015493e-05, "loss": 0.9623, "step": 19490 }, { "epoch": 0.02, "learning_rate": 4.995361658774351e-05, "loss": 1.0428, "step": 19491 }, { "epoch": 0.02, "learning_rate": 4.995361180508575e-05, "loss": 1.266, "step": 19492 }, { "epoch": 0.02, "learning_rate": 4.995360702218166e-05, "loss": 1.1301, "step": 19493 }, { "epoch": 0.02, "learning_rate": 4.995360223903124e-05, "loss": 1.6919, "step": 19494 }, { "epoch": 0.02, "learning_rate": 4.995359745563449e-05, "loss": 0.9695, "step": 19495 }, { "epoch": 0.02, "learning_rate": 4.9953592671991414e-05, "loss": 0.9921, "step": 19496 }, { "epoch": 0.02, "learning_rate": 4.995358788810199e-05, "loss": 1.1344, "step": 19497 }, { "epoch": 0.02, "learning_rate": 4.995358310396625e-05, "loss": 1.2144, "step": 19498 }, { "epoch": 0.02, "learning_rate": 4.995357831958417e-05, "loss": 1.3514, "step": 19499 }, { "epoch": 0.02, "learning_rate": 4.995357353495577e-05, "loss": 1.266, "step": 19500 }, { "epoch": 0.02, "eval_loss": 1.0280624628067017, "eval_runtime": 91.79, "eval_samples_per_second": 15.089, "eval_steps_per_second": 3.78, "step": 19500 }, { "epoch": 0.02, "learning_rate": 4.995356875008103e-05, "loss": 1.0593, "step": 19501 }, { "epoch": 0.02, "learning_rate": 4.995356396495996e-05, "loss": 1.1086, "step": 19502 }, { "epoch": 0.02, "learning_rate": 4.995355917959256e-05, "loss": 0.8475, "step": 19503 }, { "epoch": 0.02, "learning_rate": 4.9953554393978825e-05, "loss": 1.234, "step": 19504 }, { "epoch": 0.02, "learning_rate": 4.995354960811876e-05, "loss": 1.1417, "step": 19505 }, { "epoch": 0.02, "learning_rate": 4.995354482201237e-05, "loss": 0.9985, "step": 19506 }, { "epoch": 0.02, "learning_rate": 4.995354003565964e-05, "loss": 1.6862, "step": 19507 }, { "epoch": 0.02, "learning_rate": 4.995353524906058e-05, "loss": 1.2272, "step": 19508 }, { "epoch": 0.02, "learning_rate": 4.9953530462215204e-05, "loss": 1.1184, "step": 19509 }, { "epoch": 0.02, "learning_rate": 4.995352567512348e-05, "loss": 1.0556, "step": 19510 }, { "epoch": 0.02, "learning_rate": 4.9953520887785434e-05, "loss": 1.1329, "step": 19511 }, { "epoch": 0.02, "learning_rate": 4.9953516100201056e-05, "loss": 1.1709, "step": 19512 }, { "epoch": 0.02, "learning_rate": 4.995351131237035e-05, "loss": 0.8374, "step": 19513 }, { "epoch": 0.02, "learning_rate": 4.995350652429331e-05, "loss": 1.1394, "step": 19514 }, { "epoch": 0.02, "learning_rate": 4.995350173596994e-05, "loss": 1.069, "step": 19515 }, { "epoch": 0.02, "learning_rate": 4.995349694740024e-05, "loss": 1.1622, "step": 19516 }, { "epoch": 0.02, "learning_rate": 4.9953492158584206e-05, "loss": 0.7735, "step": 19517 }, { "epoch": 0.02, "learning_rate": 4.995348736952184e-05, "loss": 0.1568, "step": 19518 }, { "epoch": 0.02, "learning_rate": 4.995348258021315e-05, "loss": 0.5787, "step": 19519 }, { "epoch": 0.02, "learning_rate": 4.995347779065813e-05, "loss": 1.0258, "step": 19520 }, { "epoch": 0.02, "learning_rate": 4.995347300085678e-05, "loss": 1.0835, "step": 19521 }, { "epoch": 0.02, "learning_rate": 4.9953468210809094e-05, "loss": 0.9143, "step": 19522 }, { "epoch": 0.02, "learning_rate": 4.9953463420515084e-05, "loss": 1.0433, "step": 19523 }, { "epoch": 0.02, "learning_rate": 4.9953458629974744e-05, "loss": 1.0386, "step": 19524 }, { "epoch": 0.02, "learning_rate": 4.995345383918807e-05, "loss": 1.2294, "step": 19525 }, { "epoch": 0.02, "learning_rate": 4.9953449048155066e-05, "loss": 1.0884, "step": 19526 }, { "epoch": 0.02, "learning_rate": 4.9953444256875734e-05, "loss": 0.7121, "step": 19527 }, { "epoch": 0.02, "learning_rate": 4.995343946535007e-05, "loss": 1.0282, "step": 19528 }, { "epoch": 0.02, "learning_rate": 4.9953434673578084e-05, "loss": 1.2684, "step": 19529 }, { "epoch": 0.02, "learning_rate": 4.995342988155975e-05, "loss": 1.3149, "step": 19530 }, { "epoch": 0.02, "learning_rate": 4.9953425089295105e-05, "loss": 1.1232, "step": 19531 }, { "epoch": 0.02, "learning_rate": 4.9953420296784126e-05, "loss": 1.0037, "step": 19532 }, { "epoch": 0.02, "learning_rate": 4.9953415504026816e-05, "loss": 1.2248, "step": 19533 }, { "epoch": 0.02, "learning_rate": 4.9953410711023175e-05, "loss": 0.9695, "step": 19534 }, { "epoch": 0.02, "learning_rate": 4.99534059177732e-05, "loss": 0.9408, "step": 19535 }, { "epoch": 0.02, "learning_rate": 4.99534011242769e-05, "loss": 0.7978, "step": 19536 }, { "epoch": 0.02, "learning_rate": 4.9953396330534276e-05, "loss": 0.8975, "step": 19537 }, { "epoch": 0.02, "learning_rate": 4.995339153654531e-05, "loss": 0.9662, "step": 19538 }, { "epoch": 0.02, "learning_rate": 4.9953386742310026e-05, "loss": 0.7139, "step": 19539 }, { "epoch": 0.02, "learning_rate": 4.995338194782841e-05, "loss": 1.0453, "step": 19540 }, { "epoch": 0.02, "learning_rate": 4.9953377153100465e-05, "loss": 0.8875, "step": 19541 }, { "epoch": 0.02, "learning_rate": 4.9953372358126186e-05, "loss": 1.003, "step": 19542 }, { "epoch": 0.02, "learning_rate": 4.995336756290558e-05, "loss": 1.6193, "step": 19543 }, { "epoch": 0.02, "learning_rate": 4.995336276743865e-05, "loss": 1.2163, "step": 19544 }, { "epoch": 0.02, "learning_rate": 4.995335797172539e-05, "loss": 1.2068, "step": 19545 }, { "epoch": 0.02, "learning_rate": 4.9953353175765794e-05, "loss": 0.9622, "step": 19546 }, { "epoch": 0.02, "learning_rate": 4.9953348379559875e-05, "loss": 1.0852, "step": 19547 }, { "epoch": 0.02, "learning_rate": 4.9953343583107624e-05, "loss": 1.081, "step": 19548 }, { "epoch": 0.02, "learning_rate": 4.995333878640904e-05, "loss": 1.3372, "step": 19549 }, { "epoch": 0.02, "learning_rate": 4.995333398946414e-05, "loss": 1.0742, "step": 19550 }, { "epoch": 0.02, "learning_rate": 4.99533291922729e-05, "loss": 1.2906, "step": 19551 }, { "epoch": 0.02, "learning_rate": 4.9953324394835336e-05, "loss": 0.888, "step": 19552 }, { "epoch": 0.02, "learning_rate": 4.995331959715144e-05, "loss": 1.2805, "step": 19553 }, { "epoch": 0.02, "learning_rate": 4.995331479922122e-05, "loss": 0.5993, "step": 19554 }, { "epoch": 0.02, "learning_rate": 4.9953310001044665e-05, "loss": 1.1229, "step": 19555 }, { "epoch": 0.02, "learning_rate": 4.995330520262179e-05, "loss": 1.0982, "step": 19556 }, { "epoch": 0.02, "learning_rate": 4.9953300403952576e-05, "loss": 1.1552, "step": 19557 }, { "epoch": 0.02, "learning_rate": 4.9953295605037045e-05, "loss": 1.465, "step": 19558 }, { "epoch": 0.02, "learning_rate": 4.995329080587518e-05, "loss": 1.5656, "step": 19559 }, { "epoch": 0.02, "learning_rate": 4.9953286006466985e-05, "loss": 1.4026, "step": 19560 }, { "epoch": 0.02, "learning_rate": 4.9953281206812455e-05, "loss": 1.1061, "step": 19561 }, { "epoch": 0.02, "learning_rate": 4.995327640691161e-05, "loss": 1.2087, "step": 19562 }, { "epoch": 0.02, "learning_rate": 4.995327160676443e-05, "loss": 1.4417, "step": 19563 }, { "epoch": 0.02, "learning_rate": 4.995326680637092e-05, "loss": 0.5304, "step": 19564 }, { "epoch": 0.02, "learning_rate": 4.995326200573109e-05, "loss": 0.1152, "step": 19565 }, { "epoch": 0.02, "learning_rate": 4.995325720484493e-05, "loss": 0.067, "step": 19566 }, { "epoch": 0.02, "learning_rate": 4.9953252403712434e-05, "loss": 0.0598, "step": 19567 }, { "epoch": 0.02, "learning_rate": 4.9953247602333616e-05, "loss": 0.0576, "step": 19568 }, { "epoch": 0.02, "learning_rate": 4.995324280070847e-05, "loss": 0.0508, "step": 19569 }, { "epoch": 0.02, "learning_rate": 4.9953237998836995e-05, "loss": 0.0933, "step": 19570 }, { "epoch": 0.02, "learning_rate": 4.995323319671919e-05, "loss": 0.1034, "step": 19571 }, { "epoch": 0.02, "learning_rate": 4.995322839435506e-05, "loss": 1.1872, "step": 19572 }, { "epoch": 0.02, "learning_rate": 4.99532235917446e-05, "loss": 0.9402, "step": 19573 }, { "epoch": 0.02, "learning_rate": 4.995321878888782e-05, "loss": 0.9242, "step": 19574 }, { "epoch": 0.02, "learning_rate": 4.99532139857847e-05, "loss": 0.955, "step": 19575 }, { "epoch": 0.02, "learning_rate": 4.995320918243526e-05, "loss": 1.0, "step": 19576 }, { "epoch": 0.02, "learning_rate": 4.995320437883949e-05, "loss": 1.1454, "step": 19577 }, { "epoch": 0.02, "learning_rate": 4.99531995749974e-05, "loss": 1.1186, "step": 19578 }, { "epoch": 0.02, "learning_rate": 4.9953194770908976e-05, "loss": 1.041, "step": 19579 }, { "epoch": 0.02, "learning_rate": 4.9953189966574224e-05, "loss": 1.1514, "step": 19580 }, { "epoch": 0.02, "learning_rate": 4.995318516199314e-05, "loss": 1.1516, "step": 19581 }, { "epoch": 0.02, "learning_rate": 4.995318035716574e-05, "loss": 1.082, "step": 19582 }, { "epoch": 0.02, "learning_rate": 4.9953175552092e-05, "loss": 1.0778, "step": 19583 }, { "epoch": 0.02, "learning_rate": 4.995317074677194e-05, "loss": 1.107, "step": 19584 }, { "epoch": 0.02, "learning_rate": 4.995316594120556e-05, "loss": 0.8963, "step": 19585 }, { "epoch": 0.02, "learning_rate": 4.9953161135392844e-05, "loss": 1.089, "step": 19586 }, { "epoch": 0.02, "learning_rate": 4.99531563293338e-05, "loss": 1.1228, "step": 19587 }, { "epoch": 0.02, "learning_rate": 4.995315152302843e-05, "loss": 1.0707, "step": 19588 }, { "epoch": 0.02, "learning_rate": 4.995314671647674e-05, "loss": 1.3275, "step": 19589 }, { "epoch": 0.02, "learning_rate": 4.995314190967871e-05, "loss": 0.7864, "step": 19590 }, { "epoch": 0.02, "learning_rate": 4.995313710263436e-05, "loss": 0.9707, "step": 19591 }, { "epoch": 0.02, "learning_rate": 4.995313229534369e-05, "loss": 1.0789, "step": 19592 }, { "epoch": 0.02, "learning_rate": 4.995312748780668e-05, "loss": 1.92, "step": 19593 }, { "epoch": 0.02, "learning_rate": 4.9953122680023354e-05, "loss": 0.9433, "step": 19594 }, { "epoch": 0.02, "learning_rate": 4.9953117871993695e-05, "loss": 0.9429, "step": 19595 }, { "epoch": 0.02, "learning_rate": 4.995311306371772e-05, "loss": 1.3022, "step": 19596 }, { "epoch": 0.02, "learning_rate": 4.9953108255195404e-05, "loss": 1.0819, "step": 19597 }, { "epoch": 0.02, "learning_rate": 4.9953103446426766e-05, "loss": 1.0501, "step": 19598 }, { "epoch": 0.02, "learning_rate": 4.99530986374118e-05, "loss": 1.0737, "step": 19599 }, { "epoch": 0.02, "learning_rate": 4.995309382815051e-05, "loss": 1.3264, "step": 19600 }, { "epoch": 0.02, "learning_rate": 4.9953089018642895e-05, "loss": 0.9325, "step": 19601 }, { "epoch": 0.02, "learning_rate": 4.9953084208888954e-05, "loss": 0.9234, "step": 19602 }, { "epoch": 0.02, "learning_rate": 4.995307939888868e-05, "loss": 1.0842, "step": 19603 }, { "epoch": 0.02, "learning_rate": 4.995307458864209e-05, "loss": 1.3292, "step": 19604 }, { "epoch": 0.02, "learning_rate": 4.995306977814916e-05, "loss": 0.8856, "step": 19605 }, { "epoch": 0.02, "learning_rate": 4.9953064967409925e-05, "loss": 1.3047, "step": 19606 }, { "epoch": 0.02, "learning_rate": 4.9953060156424344e-05, "loss": 1.0327, "step": 19607 }, { "epoch": 0.02, "learning_rate": 4.9953055345192446e-05, "loss": 0.8712, "step": 19608 }, { "epoch": 0.02, "learning_rate": 4.995305053371422e-05, "loss": 1.0555, "step": 19609 }, { "epoch": 0.02, "learning_rate": 4.9953045721989665e-05, "loss": 1.2446, "step": 19610 }, { "epoch": 0.02, "learning_rate": 4.995304091001879e-05, "loss": 1.0868, "step": 19611 }, { "epoch": 0.02, "learning_rate": 4.995303609780159e-05, "loss": 1.3396, "step": 19612 }, { "epoch": 0.02, "learning_rate": 4.995303128533806e-05, "loss": 1.0766, "step": 19613 }, { "epoch": 0.02, "learning_rate": 4.995302647262821e-05, "loss": 1.103, "step": 19614 }, { "epoch": 0.02, "learning_rate": 4.995302165967202e-05, "loss": 1.113, "step": 19615 }, { "epoch": 0.02, "learning_rate": 4.995301684646951e-05, "loss": 0.9803, "step": 19616 }, { "epoch": 0.02, "learning_rate": 4.9953012033020685e-05, "loss": 1.0499, "step": 19617 }, { "epoch": 0.02, "learning_rate": 4.995300721932552e-05, "loss": 0.7142, "step": 19618 }, { "epoch": 0.02, "learning_rate": 4.995300240538404e-05, "loss": 1.1143, "step": 19619 }, { "epoch": 0.02, "learning_rate": 4.995299759119623e-05, "loss": 0.9296, "step": 19620 }, { "epoch": 0.02, "learning_rate": 4.99529927767621e-05, "loss": 0.8527, "step": 19621 }, { "epoch": 0.02, "learning_rate": 4.9952987962081636e-05, "loss": 1.0394, "step": 19622 }, { "epoch": 0.02, "learning_rate": 4.995298314715485e-05, "loss": 1.1625, "step": 19623 }, { "epoch": 0.02, "learning_rate": 4.995297833198174e-05, "loss": 1.2135, "step": 19624 }, { "epoch": 0.02, "learning_rate": 4.9952973516562305e-05, "loss": 0.8277, "step": 19625 }, { "epoch": 0.02, "learning_rate": 4.995296870089654e-05, "loss": 1.2074, "step": 19626 }, { "epoch": 0.02, "learning_rate": 4.9952963884984455e-05, "loss": 1.1347, "step": 19627 }, { "epoch": 0.02, "learning_rate": 4.9952959068826044e-05, "loss": 0.6683, "step": 19628 }, { "epoch": 0.02, "learning_rate": 4.995295425242131e-05, "loss": 0.9953, "step": 19629 }, { "epoch": 0.02, "learning_rate": 4.995294943577024e-05, "loss": 1.1722, "step": 19630 }, { "epoch": 0.02, "learning_rate": 4.995294461887286e-05, "loss": 1.0451, "step": 19631 }, { "epoch": 0.02, "learning_rate": 4.995293980172915e-05, "loss": 1.0502, "step": 19632 }, { "epoch": 0.02, "learning_rate": 4.995293498433911e-05, "loss": 1.108, "step": 19633 }, { "epoch": 0.02, "learning_rate": 4.995293016670275e-05, "loss": 0.872, "step": 19634 }, { "epoch": 0.02, "learning_rate": 4.995292534882006e-05, "loss": 0.9183, "step": 19635 }, { "epoch": 0.02, "learning_rate": 4.9952920530691046e-05, "loss": 1.1408, "step": 19636 }, { "epoch": 0.02, "learning_rate": 4.995291571231572e-05, "loss": 0.8893, "step": 19637 }, { "epoch": 0.02, "learning_rate": 4.9952910893694056e-05, "loss": 1.0212, "step": 19638 }, { "epoch": 0.02, "learning_rate": 4.995290607482607e-05, "loss": 0.9591, "step": 19639 }, { "epoch": 0.02, "learning_rate": 4.995290125571176e-05, "loss": 1.1269, "step": 19640 }, { "epoch": 0.02, "learning_rate": 4.995289643635113e-05, "loss": 1.0326, "step": 19641 }, { "epoch": 0.02, "learning_rate": 4.995289161674417e-05, "loss": 1.1609, "step": 19642 }, { "epoch": 0.02, "learning_rate": 4.995288679689088e-05, "loss": 1.0486, "step": 19643 }, { "epoch": 0.02, "learning_rate": 4.9952881976791275e-05, "loss": 1.2846, "step": 19644 }, { "epoch": 0.02, "learning_rate": 4.995287715644534e-05, "loss": 1.0467, "step": 19645 }, { "epoch": 0.02, "learning_rate": 4.995287233585309e-05, "loss": 1.0944, "step": 19646 }, { "epoch": 0.02, "learning_rate": 4.995286751501451e-05, "loss": 1.0255, "step": 19647 }, { "epoch": 0.02, "learning_rate": 4.9952862693929604e-05, "loss": 1.1184, "step": 19648 }, { "epoch": 0.02, "learning_rate": 4.995285787259838e-05, "loss": 0.8052, "step": 19649 }, { "epoch": 0.02, "learning_rate": 4.995285305102082e-05, "loss": 0.7232, "step": 19650 }, { "epoch": 0.02, "learning_rate": 4.995284822919695e-05, "loss": 1.0202, "step": 19651 }, { "epoch": 0.02, "learning_rate": 4.995284340712674e-05, "loss": 1.0011, "step": 19652 }, { "epoch": 0.02, "learning_rate": 4.995283858481022e-05, "loss": 0.8517, "step": 19653 }, { "epoch": 0.02, "learning_rate": 4.995283376224737e-05, "loss": 1.1823, "step": 19654 }, { "epoch": 0.02, "learning_rate": 4.99528289394382e-05, "loss": 1.2, "step": 19655 }, { "epoch": 0.02, "learning_rate": 4.99528241163827e-05, "loss": 0.8898, "step": 19656 }, { "epoch": 0.02, "learning_rate": 4.995281929308088e-05, "loss": 1.0298, "step": 19657 }, { "epoch": 0.02, "learning_rate": 4.995281446953274e-05, "loss": 0.9889, "step": 19658 }, { "epoch": 0.02, "learning_rate": 4.995280964573827e-05, "loss": 1.1419, "step": 19659 }, { "epoch": 0.02, "learning_rate": 4.995280482169747e-05, "loss": 1.0369, "step": 19660 }, { "epoch": 0.02, "learning_rate": 4.995279999741036e-05, "loss": 0.9841, "step": 19661 }, { "epoch": 0.02, "learning_rate": 4.995279517287692e-05, "loss": 1.0522, "step": 19662 }, { "epoch": 0.02, "learning_rate": 4.995279034809716e-05, "loss": 1.0818, "step": 19663 }, { "epoch": 0.02, "learning_rate": 4.9952785523071074e-05, "loss": 1.0004, "step": 19664 }, { "epoch": 0.02, "learning_rate": 4.995278069779867e-05, "loss": 1.2191, "step": 19665 }, { "epoch": 0.02, "learning_rate": 4.9952775872279935e-05, "loss": 1.1579, "step": 19666 }, { "epoch": 0.02, "learning_rate": 4.9952771046514877e-05, "loss": 1.1779, "step": 19667 }, { "epoch": 0.02, "learning_rate": 4.99527662205035e-05, "loss": 1.2012, "step": 19668 }, { "epoch": 0.02, "learning_rate": 4.99527613942458e-05, "loss": 1.2191, "step": 19669 }, { "epoch": 0.02, "learning_rate": 4.995275656774178e-05, "loss": 1.027, "step": 19670 }, { "epoch": 0.02, "learning_rate": 4.9952751740991436e-05, "loss": 1.0522, "step": 19671 }, { "epoch": 0.02, "learning_rate": 4.995274691399476e-05, "loss": 1.0838, "step": 19672 }, { "epoch": 0.02, "learning_rate": 4.995274208675176e-05, "loss": 1.6773, "step": 19673 }, { "epoch": 0.02, "learning_rate": 4.995273725926245e-05, "loss": 1.2025, "step": 19674 }, { "epoch": 0.02, "learning_rate": 4.9952732431526804e-05, "loss": 1.4595, "step": 19675 }, { "epoch": 0.02, "learning_rate": 4.9952727603544844e-05, "loss": 0.6885, "step": 19676 }, { "epoch": 0.02, "learning_rate": 4.995272277531656e-05, "loss": 1.1281, "step": 19677 }, { "epoch": 0.02, "learning_rate": 4.995271794684196e-05, "loss": 1.2027, "step": 19678 }, { "epoch": 0.02, "learning_rate": 4.995271311812102e-05, "loss": 1.0859, "step": 19679 }, { "epoch": 0.02, "learning_rate": 4.995270828915377e-05, "loss": 1.1027, "step": 19680 }, { "epoch": 0.02, "learning_rate": 4.9952703459940187e-05, "loss": 0.9807, "step": 19681 }, { "epoch": 0.02, "learning_rate": 4.995269863048029e-05, "loss": 1.2301, "step": 19682 }, { "epoch": 0.02, "learning_rate": 4.995269380077407e-05, "loss": 0.6885, "step": 19683 }, { "epoch": 0.02, "learning_rate": 4.995268897082153e-05, "loss": 0.9541, "step": 19684 }, { "epoch": 0.02, "learning_rate": 4.995268414062266e-05, "loss": 1.0124, "step": 19685 }, { "epoch": 0.02, "learning_rate": 4.995267931017747e-05, "loss": 0.934, "step": 19686 }, { "epoch": 0.02, "learning_rate": 4.995267447948596e-05, "loss": 1.0919, "step": 19687 }, { "epoch": 0.02, "learning_rate": 4.9952669648548126e-05, "loss": 1.3242, "step": 19688 }, { "epoch": 0.02, "learning_rate": 4.995266481736397e-05, "loss": 1.311, "step": 19689 }, { "epoch": 0.02, "learning_rate": 4.995265998593349e-05, "loss": 1.0763, "step": 19690 }, { "epoch": 0.02, "learning_rate": 4.995265515425669e-05, "loss": 1.1591, "step": 19691 }, { "epoch": 0.02, "learning_rate": 4.9952650322333574e-05, "loss": 1.1842, "step": 19692 }, { "epoch": 0.02, "learning_rate": 4.995264549016413e-05, "loss": 0.8564, "step": 19693 }, { "epoch": 0.02, "learning_rate": 4.995264065774836e-05, "loss": 1.107, "step": 19694 }, { "epoch": 0.02, "learning_rate": 4.995263582508627e-05, "loss": 1.2695, "step": 19695 }, { "epoch": 0.02, "learning_rate": 4.995263099217786e-05, "loss": 1.3649, "step": 19696 }, { "epoch": 0.02, "learning_rate": 4.995262615902313e-05, "loss": 1.7468, "step": 19697 }, { "epoch": 0.02, "learning_rate": 4.995262132562208e-05, "loss": 1.7001, "step": 19698 }, { "epoch": 0.02, "learning_rate": 4.9952616491974704e-05, "loss": 1.7814, "step": 19699 }, { "epoch": 0.02, "learning_rate": 4.995261165808101e-05, "loss": 1.6476, "step": 19700 }, { "epoch": 0.02, "learning_rate": 4.995260682394098e-05, "loss": 1.6183, "step": 19701 }, { "epoch": 0.02, "learning_rate": 4.9952601989554646e-05, "loss": 1.6833, "step": 19702 }, { "epoch": 0.02, "learning_rate": 4.9952597154921986e-05, "loss": 1.5801, "step": 19703 }, { "epoch": 0.02, "learning_rate": 4.9952592320043e-05, "loss": 1.5981, "step": 19704 }, { "epoch": 0.02, "learning_rate": 4.995258748491769e-05, "loss": 1.5753, "step": 19705 }, { "epoch": 0.02, "learning_rate": 4.995258264954607e-05, "loss": 1.5866, "step": 19706 }, { "epoch": 0.02, "learning_rate": 4.9952577813928125e-05, "loss": 1.508, "step": 19707 }, { "epoch": 0.02, "learning_rate": 4.995257297806385e-05, "loss": 1.272, "step": 19708 }, { "epoch": 0.02, "learning_rate": 4.995256814195326e-05, "loss": 0.4184, "step": 19709 }, { "epoch": 0.02, "learning_rate": 4.995256330559635e-05, "loss": 0.3692, "step": 19710 }, { "epoch": 0.02, "learning_rate": 4.9952558468993115e-05, "loss": 0.3675, "step": 19711 }, { "epoch": 0.02, "learning_rate": 4.995255363214356e-05, "loss": 1.2069, "step": 19712 }, { "epoch": 0.02, "learning_rate": 4.995254879504769e-05, "loss": 2.1163, "step": 19713 }, { "epoch": 0.02, "learning_rate": 4.995254395770549e-05, "loss": 2.0282, "step": 19714 }, { "epoch": 0.02, "learning_rate": 4.9952539120116976e-05, "loss": 1.9783, "step": 19715 }, { "epoch": 0.02, "learning_rate": 4.995253428228214e-05, "loss": 2.0105, "step": 19716 }, { "epoch": 0.02, "learning_rate": 4.995252944420098e-05, "loss": 1.9535, "step": 19717 }, { "epoch": 0.02, "learning_rate": 4.99525246058735e-05, "loss": 1.9379, "step": 19718 }, { "epoch": 0.02, "learning_rate": 4.99525197672997e-05, "loss": 1.8932, "step": 19719 }, { "epoch": 0.02, "learning_rate": 4.9952514928479575e-05, "loss": 1.3612, "step": 19720 }, { "epoch": 0.02, "learning_rate": 4.995251008941314e-05, "loss": 0.8687, "step": 19721 }, { "epoch": 0.02, "learning_rate": 4.9952505250100376e-05, "loss": 1.0693, "step": 19722 }, { "epoch": 0.02, "learning_rate": 4.995250041054129e-05, "loss": 1.0527, "step": 19723 }, { "epoch": 0.02, "learning_rate": 4.995249557073589e-05, "loss": 1.0204, "step": 19724 }, { "epoch": 0.02, "learning_rate": 4.995249073068416e-05, "loss": 1.1925, "step": 19725 }, { "epoch": 0.02, "learning_rate": 4.995248589038611e-05, "loss": 1.017, "step": 19726 }, { "epoch": 0.02, "learning_rate": 4.995248104984175e-05, "loss": 1.1953, "step": 19727 }, { "epoch": 0.02, "learning_rate": 4.9952476209051066e-05, "loss": 0.7603, "step": 19728 }, { "epoch": 0.02, "learning_rate": 4.995247136801406e-05, "loss": 0.7274, "step": 19729 }, { "epoch": 0.02, "learning_rate": 4.995246652673073e-05, "loss": 1.0518, "step": 19730 }, { "epoch": 0.02, "learning_rate": 4.995246168520108e-05, "loss": 1.1496, "step": 19731 }, { "epoch": 0.02, "learning_rate": 4.995245684342512e-05, "loss": 1.3168, "step": 19732 }, { "epoch": 0.02, "learning_rate": 4.995245200140283e-05, "loss": 1.0891, "step": 19733 }, { "epoch": 0.02, "learning_rate": 4.995244715913422e-05, "loss": 1.0854, "step": 19734 }, { "epoch": 0.02, "learning_rate": 4.99524423166193e-05, "loss": 0.9309, "step": 19735 }, { "epoch": 0.02, "learning_rate": 4.995243747385805e-05, "loss": 1.1332, "step": 19736 }, { "epoch": 0.02, "learning_rate": 4.995243263085049e-05, "loss": 0.9958, "step": 19737 }, { "epoch": 0.02, "learning_rate": 4.99524277875966e-05, "loss": 0.9782, "step": 19738 }, { "epoch": 0.02, "learning_rate": 4.995242294409639e-05, "loss": 0.7962, "step": 19739 }, { "epoch": 0.02, "learning_rate": 4.9952418100349866e-05, "loss": 0.3554, "step": 19740 }, { "epoch": 0.02, "learning_rate": 4.9952413256357024e-05, "loss": 0.2748, "step": 19741 }, { "epoch": 0.02, "learning_rate": 4.995240841211785e-05, "loss": 0.3307, "step": 19742 }, { "epoch": 0.02, "learning_rate": 4.995240356763237e-05, "loss": 0.6902, "step": 19743 }, { "epoch": 0.02, "learning_rate": 4.995239872290056e-05, "loss": 0.8881, "step": 19744 }, { "epoch": 0.02, "learning_rate": 4.995239387792244e-05, "loss": 1.4307, "step": 19745 }, { "epoch": 0.02, "learning_rate": 4.995238903269799e-05, "loss": 1.3082, "step": 19746 }, { "epoch": 0.02, "learning_rate": 4.995238418722723e-05, "loss": 1.2909, "step": 19747 }, { "epoch": 0.02, "learning_rate": 4.995237934151015e-05, "loss": 1.2271, "step": 19748 }, { "epoch": 0.02, "learning_rate": 4.995237449554674e-05, "loss": 1.0485, "step": 19749 }, { "epoch": 0.02, "learning_rate": 4.9952369649337024e-05, "loss": 1.1645, "step": 19750 }, { "epoch": 0.02, "learning_rate": 4.995236480288098e-05, "loss": 1.5562, "step": 19751 }, { "epoch": 0.02, "learning_rate": 4.995235995617862e-05, "loss": 0.9653, "step": 19752 }, { "epoch": 0.02, "learning_rate": 4.995235510922994e-05, "loss": 0.8649, "step": 19753 }, { "epoch": 0.02, "learning_rate": 4.9952350262034944e-05, "loss": 0.6512, "step": 19754 }, { "epoch": 0.02, "learning_rate": 4.995234541459363e-05, "loss": 1.0656, "step": 19755 }, { "epoch": 0.02, "learning_rate": 4.995234056690599e-05, "loss": 0.9732, "step": 19756 }, { "epoch": 0.02, "learning_rate": 4.995233571897203e-05, "loss": 1.0942, "step": 19757 }, { "epoch": 0.02, "learning_rate": 4.9952330870791755e-05, "loss": 1.0566, "step": 19758 }, { "epoch": 0.02, "learning_rate": 4.995232602236516e-05, "loss": 1.0405, "step": 19759 }, { "epoch": 0.02, "learning_rate": 4.9952321173692255e-05, "loss": 1.0702, "step": 19760 }, { "epoch": 0.02, "learning_rate": 4.995231632477302e-05, "loss": 1.0673, "step": 19761 }, { "epoch": 0.02, "learning_rate": 4.995231147560747e-05, "loss": 1.099, "step": 19762 }, { "epoch": 0.02, "learning_rate": 4.99523066261956e-05, "loss": 1.2589, "step": 19763 }, { "epoch": 0.02, "learning_rate": 4.995230177653741e-05, "loss": 0.5698, "step": 19764 }, { "epoch": 0.02, "learning_rate": 4.9952296926632904e-05, "loss": 0.5423, "step": 19765 }, { "epoch": 0.02, "learning_rate": 4.995229207648208e-05, "loss": 0.7262, "step": 19766 }, { "epoch": 0.02, "learning_rate": 4.9952287226084936e-05, "loss": 1.0165, "step": 19767 }, { "epoch": 0.02, "learning_rate": 4.9952282375441476e-05, "loss": 1.188, "step": 19768 }, { "epoch": 0.02, "learning_rate": 4.995227752455169e-05, "loss": 0.9412, "step": 19769 }, { "epoch": 0.02, "learning_rate": 4.9952272673415604e-05, "loss": 1.111, "step": 19770 }, { "epoch": 0.02, "learning_rate": 4.9952267822033186e-05, "loss": 0.9398, "step": 19771 }, { "epoch": 0.02, "learning_rate": 4.9952262970404443e-05, "loss": 1.2999, "step": 19772 }, { "epoch": 0.02, "learning_rate": 4.995225811852939e-05, "loss": 1.0939, "step": 19773 }, { "epoch": 0.02, "learning_rate": 4.995225326640802e-05, "loss": 0.9866, "step": 19774 }, { "epoch": 0.02, "learning_rate": 4.9952248414040334e-05, "loss": 0.8609, "step": 19775 }, { "epoch": 0.02, "learning_rate": 4.995224356142633e-05, "loss": 1.0768, "step": 19776 }, { "epoch": 0.02, "learning_rate": 4.9952238708566e-05, "loss": 1.1745, "step": 19777 }, { "epoch": 0.02, "learning_rate": 4.995223385545935e-05, "loss": 0.7907, "step": 19778 }, { "epoch": 0.02, "learning_rate": 4.995222900210639e-05, "loss": 0.9543, "step": 19779 }, { "epoch": 0.02, "learning_rate": 4.995222414850711e-05, "loss": 1.2915, "step": 19780 }, { "epoch": 0.02, "learning_rate": 4.9952219294661517e-05, "loss": 1.0397, "step": 19781 }, { "epoch": 0.02, "learning_rate": 4.99522144405696e-05, "loss": 1.0812, "step": 19782 }, { "epoch": 0.02, "learning_rate": 4.995220958623137e-05, "loss": 0.9663, "step": 19783 }, { "epoch": 0.02, "learning_rate": 4.9952204731646816e-05, "loss": 1.1324, "step": 19784 }, { "epoch": 0.02, "learning_rate": 4.995219987681595e-05, "loss": 1.1988, "step": 19785 }, { "epoch": 0.02, "learning_rate": 4.995219502173876e-05, "loss": 1.2214, "step": 19786 }, { "epoch": 0.02, "learning_rate": 4.995219016641526e-05, "loss": 1.0391, "step": 19787 }, { "epoch": 0.02, "learning_rate": 4.995218531084544e-05, "loss": 1.1344, "step": 19788 }, { "epoch": 0.02, "learning_rate": 4.99521804550293e-05, "loss": 0.921, "step": 19789 }, { "epoch": 0.02, "learning_rate": 4.9952175598966844e-05, "loss": 1.0889, "step": 19790 }, { "epoch": 0.02, "learning_rate": 4.9952170742658075e-05, "loss": 0.2578, "step": 19791 }, { "epoch": 0.02, "learning_rate": 4.995216588610298e-05, "loss": 0.2139, "step": 19792 }, { "epoch": 0.02, "learning_rate": 4.995216102930157e-05, "loss": 0.1942, "step": 19793 }, { "epoch": 0.02, "learning_rate": 4.995215617225385e-05, "loss": 0.2236, "step": 19794 }, { "epoch": 0.02, "learning_rate": 4.995215131495981e-05, "loss": 0.2054, "step": 19795 }, { "epoch": 0.02, "learning_rate": 4.995214645741945e-05, "loss": 0.179, "step": 19796 }, { "epoch": 0.02, "learning_rate": 4.9952141599632775e-05, "loss": 0.2465, "step": 19797 }, { "epoch": 0.02, "learning_rate": 4.995213674159977e-05, "loss": 0.6638, "step": 19798 }, { "epoch": 0.02, "learning_rate": 4.9952131883320466e-05, "loss": 1.2053, "step": 19799 }, { "epoch": 0.02, "learning_rate": 4.9952127024794836e-05, "loss": 1.0865, "step": 19800 }, { "epoch": 0.02, "learning_rate": 4.9952122166022895e-05, "loss": 0.7951, "step": 19801 }, { "epoch": 0.02, "learning_rate": 4.995211730700464e-05, "loss": 0.7298, "step": 19802 }, { "epoch": 0.02, "learning_rate": 4.9952112447740056e-05, "loss": 1.1313, "step": 19803 }, { "epoch": 0.02, "learning_rate": 4.9952107588229164e-05, "loss": 1.3023, "step": 19804 }, { "epoch": 0.02, "learning_rate": 4.995210272847195e-05, "loss": 1.1852, "step": 19805 }, { "epoch": 0.02, "learning_rate": 4.995209786846843e-05, "loss": 1.0843, "step": 19806 }, { "epoch": 0.02, "learning_rate": 4.9952093008218584e-05, "loss": 1.0489, "step": 19807 }, { "epoch": 0.02, "learning_rate": 4.995208814772242e-05, "loss": 1.3249, "step": 19808 }, { "epoch": 0.02, "learning_rate": 4.9952083286979946e-05, "loss": 1.2792, "step": 19809 }, { "epoch": 0.02, "learning_rate": 4.9952078425991144e-05, "loss": 1.366, "step": 19810 }, { "epoch": 0.02, "learning_rate": 4.995207356475604e-05, "loss": 1.2116, "step": 19811 }, { "epoch": 0.02, "learning_rate": 4.9952068703274615e-05, "loss": 1.2203, "step": 19812 }, { "epoch": 0.02, "learning_rate": 4.995206384154687e-05, "loss": 1.044, "step": 19813 }, { "epoch": 0.02, "learning_rate": 4.9952058979572805e-05, "loss": 1.3441, "step": 19814 }, { "epoch": 0.02, "learning_rate": 4.995205411735243e-05, "loss": 1.4112, "step": 19815 }, { "epoch": 0.02, "learning_rate": 4.995204925488574e-05, "loss": 1.0541, "step": 19816 }, { "epoch": 0.02, "learning_rate": 4.995204439217274e-05, "loss": 0.9978, "step": 19817 }, { "epoch": 0.02, "learning_rate": 4.995203952921341e-05, "loss": 1.1235, "step": 19818 }, { "epoch": 0.02, "learning_rate": 4.9952034666007775e-05, "loss": 1.1931, "step": 19819 }, { "epoch": 0.02, "learning_rate": 4.9952029802555815e-05, "loss": 1.1176, "step": 19820 }, { "epoch": 0.02, "learning_rate": 4.9952024938857544e-05, "loss": 1.0524, "step": 19821 }, { "epoch": 0.02, "learning_rate": 4.9952020074912956e-05, "loss": 1.034, "step": 19822 }, { "epoch": 0.02, "learning_rate": 4.995201521072205e-05, "loss": 0.928, "step": 19823 }, { "epoch": 0.02, "learning_rate": 4.9952010346284836e-05, "loss": 1.0914, "step": 19824 }, { "epoch": 0.02, "learning_rate": 4.9952005481601296e-05, "loss": 0.9478, "step": 19825 }, { "epoch": 0.02, "learning_rate": 4.9952000616671447e-05, "loss": 1.2157, "step": 19826 }, { "epoch": 0.02, "learning_rate": 4.9951995751495286e-05, "loss": 1.0494, "step": 19827 }, { "epoch": 0.02, "learning_rate": 4.9951990886072796e-05, "loss": 1.0317, "step": 19828 }, { "epoch": 0.02, "learning_rate": 4.9951986020404e-05, "loss": 1.3918, "step": 19829 }, { "epoch": 0.02, "learning_rate": 4.995198115448889e-05, "loss": 1.1973, "step": 19830 }, { "epoch": 0.02, "learning_rate": 4.995197628832746e-05, "loss": 0.9511, "step": 19831 }, { "epoch": 0.02, "learning_rate": 4.995197142191972e-05, "loss": 0.9834, "step": 19832 }, { "epoch": 0.02, "learning_rate": 4.9951966555265665e-05, "loss": 1.1609, "step": 19833 }, { "epoch": 0.02, "learning_rate": 4.9951961688365284e-05, "loss": 0.9816, "step": 19834 }, { "epoch": 0.02, "learning_rate": 4.9951956821218593e-05, "loss": 1.5256, "step": 19835 }, { "epoch": 0.02, "learning_rate": 4.995195195382559e-05, "loss": 1.1088, "step": 19836 }, { "epoch": 0.02, "learning_rate": 4.995194708618627e-05, "loss": 0.4684, "step": 19837 }, { "epoch": 0.02, "learning_rate": 4.995194221830063e-05, "loss": 0.245, "step": 19838 }, { "epoch": 0.02, "learning_rate": 4.9951937350168685e-05, "loss": 0.2415, "step": 19839 }, { "epoch": 0.02, "learning_rate": 4.995193248179042e-05, "loss": 0.1603, "step": 19840 }, { "epoch": 0.02, "learning_rate": 4.995192761316584e-05, "loss": 0.4589, "step": 19841 }, { "epoch": 0.02, "learning_rate": 4.9951922744294944e-05, "loss": 0.9528, "step": 19842 }, { "epoch": 0.02, "learning_rate": 4.9951917875177736e-05, "loss": 0.502, "step": 19843 }, { "epoch": 0.02, "learning_rate": 4.995191300581421e-05, "loss": 1.1688, "step": 19844 }, { "epoch": 0.02, "learning_rate": 4.9951908136204376e-05, "loss": 1.187, "step": 19845 }, { "epoch": 0.02, "learning_rate": 4.9951903266348216e-05, "loss": 1.0579, "step": 19846 }, { "epoch": 0.02, "learning_rate": 4.9951898396245746e-05, "loss": 1.6157, "step": 19847 }, { "epoch": 0.02, "learning_rate": 4.9951893525896966e-05, "loss": 0.3302, "step": 19848 }, { "epoch": 0.02, "learning_rate": 4.995188865530186e-05, "loss": 0.3947, "step": 19849 }, { "epoch": 0.02, "learning_rate": 4.9951883784460454e-05, "loss": 1.3273, "step": 19850 }, { "epoch": 0.02, "learning_rate": 4.995187891337273e-05, "loss": 1.0389, "step": 19851 }, { "epoch": 0.02, "learning_rate": 4.995187404203868e-05, "loss": 1.0853, "step": 19852 }, { "epoch": 0.02, "learning_rate": 4.9951869170458334e-05, "loss": 1.1117, "step": 19853 }, { "epoch": 0.02, "learning_rate": 4.995186429863166e-05, "loss": 1.1337, "step": 19854 }, { "epoch": 0.02, "learning_rate": 4.995185942655868e-05, "loss": 1.119, "step": 19855 }, { "epoch": 0.02, "learning_rate": 4.995185455423937e-05, "loss": 1.3506, "step": 19856 }, { "epoch": 0.02, "learning_rate": 4.9951849681673765e-05, "loss": 1.5422, "step": 19857 }, { "epoch": 0.02, "learning_rate": 4.995184480886183e-05, "loss": 1.4147, "step": 19858 }, { "epoch": 0.02, "learning_rate": 4.995183993580359e-05, "loss": 0.9951, "step": 19859 }, { "epoch": 0.02, "learning_rate": 4.995183506249904e-05, "loss": 0.893, "step": 19860 }, { "epoch": 0.02, "learning_rate": 4.995183018894817e-05, "loss": 1.2312, "step": 19861 }, { "epoch": 0.02, "learning_rate": 4.995182531515099e-05, "loss": 1.1476, "step": 19862 }, { "epoch": 0.02, "learning_rate": 4.995182044110749e-05, "loss": 1.229, "step": 19863 }, { "epoch": 0.02, "learning_rate": 4.9951815566817676e-05, "loss": 1.1509, "step": 19864 }, { "epoch": 0.02, "learning_rate": 4.995181069228155e-05, "loss": 0.9121, "step": 19865 }, { "epoch": 0.02, "learning_rate": 4.9951805817499116e-05, "loss": 1.3532, "step": 19866 }, { "epoch": 0.02, "learning_rate": 4.9951800942470364e-05, "loss": 1.2682, "step": 19867 }, { "epoch": 0.02, "learning_rate": 4.99517960671953e-05, "loss": 1.1546, "step": 19868 }, { "epoch": 0.02, "learning_rate": 4.9951791191673914e-05, "loss": 1.1764, "step": 19869 }, { "epoch": 0.02, "learning_rate": 4.995178631590622e-05, "loss": 0.8395, "step": 19870 }, { "epoch": 0.02, "learning_rate": 4.995178143989222e-05, "loss": 1.2247, "step": 19871 }, { "epoch": 0.02, "learning_rate": 4.99517765636319e-05, "loss": 1.0671, "step": 19872 }, { "epoch": 0.02, "learning_rate": 4.995177168712526e-05, "loss": 1.1357, "step": 19873 }, { "epoch": 0.02, "learning_rate": 4.9951766810372316e-05, "loss": 1.0832, "step": 19874 }, { "epoch": 0.02, "learning_rate": 4.9951761933373054e-05, "loss": 1.1123, "step": 19875 }, { "epoch": 0.02, "learning_rate": 4.995175705612748e-05, "loss": 1.2845, "step": 19876 }, { "epoch": 0.02, "learning_rate": 4.9951752178635597e-05, "loss": 1.3855, "step": 19877 }, { "epoch": 0.02, "learning_rate": 4.99517473008974e-05, "loss": 1.2405, "step": 19878 }, { "epoch": 0.02, "learning_rate": 4.9951742422912884e-05, "loss": 0.6798, "step": 19879 }, { "epoch": 0.02, "learning_rate": 4.9951737544682056e-05, "loss": 1.0642, "step": 19880 }, { "epoch": 0.02, "learning_rate": 4.995173266620492e-05, "loss": 1.1656, "step": 19881 }, { "epoch": 0.02, "learning_rate": 4.995172778748147e-05, "loss": 1.1354, "step": 19882 }, { "epoch": 0.02, "learning_rate": 4.99517229085117e-05, "loss": 1.1622, "step": 19883 }, { "epoch": 0.02, "learning_rate": 4.9951718029295625e-05, "loss": 1.0806, "step": 19884 }, { "epoch": 0.02, "learning_rate": 4.995171314983324e-05, "loss": 1.1213, "step": 19885 }, { "epoch": 0.02, "learning_rate": 4.995170827012453e-05, "loss": 0.7129, "step": 19886 }, { "epoch": 0.02, "learning_rate": 4.995170339016951e-05, "loss": 1.279, "step": 19887 }, { "epoch": 0.02, "learning_rate": 4.995169850996818e-05, "loss": 1.1153, "step": 19888 }, { "epoch": 0.02, "learning_rate": 4.9951693629520544e-05, "loss": 0.8464, "step": 19889 }, { "epoch": 0.02, "learning_rate": 4.9951688748826584e-05, "loss": 0.95, "step": 19890 }, { "epoch": 0.02, "learning_rate": 4.995168386788632e-05, "loss": 1.214, "step": 19891 }, { "epoch": 0.02, "learning_rate": 4.995167898669974e-05, "loss": 1.145, "step": 19892 }, { "epoch": 0.02, "learning_rate": 4.995167410526685e-05, "loss": 1.1607, "step": 19893 }, { "epoch": 0.02, "learning_rate": 4.995166922358764e-05, "loss": 1.3608, "step": 19894 }, { "epoch": 0.02, "learning_rate": 4.995166434166213e-05, "loss": 1.2716, "step": 19895 }, { "epoch": 0.02, "learning_rate": 4.9951659459490294e-05, "loss": 1.1194, "step": 19896 }, { "epoch": 0.02, "learning_rate": 4.995165457707216e-05, "loss": 1.3742, "step": 19897 }, { "epoch": 0.02, "learning_rate": 4.9951649694407705e-05, "loss": 1.259, "step": 19898 }, { "epoch": 0.02, "learning_rate": 4.995164481149694e-05, "loss": 1.1134, "step": 19899 }, { "epoch": 0.02, "learning_rate": 4.995163992833986e-05, "loss": 0.8128, "step": 19900 }, { "epoch": 0.02, "learning_rate": 4.9951635044936474e-05, "loss": 1.1163, "step": 19901 }, { "epoch": 0.02, "learning_rate": 4.995163016128677e-05, "loss": 1.0653, "step": 19902 }, { "epoch": 0.02, "learning_rate": 4.9951625277390754e-05, "loss": 1.3579, "step": 19903 }, { "epoch": 0.02, "learning_rate": 4.9951620393248435e-05, "loss": 1.5247, "step": 19904 }, { "epoch": 0.02, "learning_rate": 4.995161550885979e-05, "loss": 1.563, "step": 19905 }, { "epoch": 0.02, "learning_rate": 4.9951610624224846e-05, "loss": 1.371, "step": 19906 }, { "epoch": 0.02, "learning_rate": 4.995160573934358e-05, "loss": 1.1175, "step": 19907 }, { "epoch": 0.02, "learning_rate": 4.995160085421601e-05, "loss": 0.8112, "step": 19908 }, { "epoch": 0.02, "learning_rate": 4.9951595968842124e-05, "loss": 0.9531, "step": 19909 }, { "epoch": 0.02, "learning_rate": 4.995159108322193e-05, "loss": 1.0515, "step": 19910 }, { "epoch": 0.02, "learning_rate": 4.995158619735542e-05, "loss": 1.0934, "step": 19911 }, { "epoch": 0.02, "learning_rate": 4.99515813112426e-05, "loss": 0.6406, "step": 19912 }, { "epoch": 0.02, "learning_rate": 4.9951576424883475e-05, "loss": 0.9396, "step": 19913 }, { "epoch": 0.02, "learning_rate": 4.995157153827803e-05, "loss": 1.1762, "step": 19914 }, { "epoch": 0.02, "learning_rate": 4.995156665142627e-05, "loss": 1.2595, "step": 19915 }, { "epoch": 0.02, "learning_rate": 4.995156176432821e-05, "loss": 1.1789, "step": 19916 }, { "epoch": 0.02, "learning_rate": 4.9951556876983827e-05, "loss": 0.8944, "step": 19917 }, { "epoch": 0.02, "learning_rate": 4.995155198939314e-05, "loss": 1.191, "step": 19918 }, { "epoch": 0.02, "learning_rate": 4.9951547101556146e-05, "loss": 1.2088, "step": 19919 }, { "epoch": 0.02, "learning_rate": 4.9951542213472834e-05, "loss": 1.366, "step": 19920 }, { "epoch": 0.02, "learning_rate": 4.995153732514322e-05, "loss": 1.1819, "step": 19921 }, { "epoch": 0.02, "learning_rate": 4.9951532436567285e-05, "loss": 1.2097, "step": 19922 }, { "epoch": 0.02, "learning_rate": 4.9951527547745034e-05, "loss": 1.3596, "step": 19923 }, { "epoch": 0.02, "learning_rate": 4.995152265867649e-05, "loss": 1.0346, "step": 19924 }, { "epoch": 0.02, "learning_rate": 4.995151776936162e-05, "loss": 0.9956, "step": 19925 }, { "epoch": 0.02, "learning_rate": 4.995151287980044e-05, "loss": 0.7451, "step": 19926 }, { "epoch": 0.02, "learning_rate": 4.995150798999295e-05, "loss": 1.1893, "step": 19927 }, { "epoch": 0.02, "learning_rate": 4.9951503099939153e-05, "loss": 1.1116, "step": 19928 }, { "epoch": 0.02, "learning_rate": 4.995149820963905e-05, "loss": 0.7205, "step": 19929 }, { "epoch": 0.02, "learning_rate": 4.995149331909263e-05, "loss": 1.1302, "step": 19930 }, { "epoch": 0.02, "learning_rate": 4.99514884282999e-05, "loss": 1.229, "step": 19931 }, { "epoch": 0.02, "learning_rate": 4.995148353726086e-05, "loss": 1.3831, "step": 19932 }, { "epoch": 0.02, "learning_rate": 4.995147864597551e-05, "loss": 0.9814, "step": 19933 }, { "epoch": 0.02, "learning_rate": 4.995147375444385e-05, "loss": 1.176, "step": 19934 }, { "epoch": 0.02, "learning_rate": 4.995146886266587e-05, "loss": 1.0441, "step": 19935 }, { "epoch": 0.02, "learning_rate": 4.995146397064159e-05, "loss": 1.398, "step": 19936 }, { "epoch": 0.02, "learning_rate": 4.9951459078371e-05, "loss": 1.1432, "step": 19937 }, { "epoch": 0.02, "learning_rate": 4.99514541858541e-05, "loss": 1.2002, "step": 19938 }, { "epoch": 0.02, "learning_rate": 4.995144929309089e-05, "loss": 1.0776, "step": 19939 }, { "epoch": 0.02, "learning_rate": 4.995144440008136e-05, "loss": 0.7207, "step": 19940 }, { "epoch": 0.02, "learning_rate": 4.995143950682553e-05, "loss": 1.1236, "step": 19941 }, { "epoch": 0.02, "learning_rate": 4.995143461332339e-05, "loss": 0.9496, "step": 19942 }, { "epoch": 0.02, "learning_rate": 4.995142971957493e-05, "loss": 0.9429, "step": 19943 }, { "epoch": 0.02, "learning_rate": 4.9951424825580174e-05, "loss": 1.1929, "step": 19944 }, { "epoch": 0.02, "learning_rate": 4.9951419931339095e-05, "loss": 1.3029, "step": 19945 }, { "epoch": 0.02, "learning_rate": 4.995141503685171e-05, "loss": 1.1513, "step": 19946 }, { "epoch": 0.02, "learning_rate": 4.995141014211802e-05, "loss": 0.9263, "step": 19947 }, { "epoch": 0.02, "learning_rate": 4.9951405247138015e-05, "loss": 1.1595, "step": 19948 }, { "epoch": 0.02, "learning_rate": 4.99514003519117e-05, "loss": 1.2248, "step": 19949 }, { "epoch": 0.02, "learning_rate": 4.995139545643908e-05, "loss": 1.1268, "step": 19950 }, { "epoch": 0.02, "learning_rate": 4.995139056072015e-05, "loss": 1.0567, "step": 19951 }, { "epoch": 0.02, "learning_rate": 4.995138566475491e-05, "loss": 1.1301, "step": 19952 }, { "epoch": 0.02, "learning_rate": 4.9951380768543354e-05, "loss": 1.1523, "step": 19953 }, { "epoch": 0.02, "learning_rate": 4.9951375872085494e-05, "loss": 1.8745, "step": 19954 }, { "epoch": 0.02, "learning_rate": 4.995137097538133e-05, "loss": 1.0256, "step": 19955 }, { "epoch": 0.02, "learning_rate": 4.995136607843085e-05, "loss": 1.0276, "step": 19956 }, { "epoch": 0.02, "learning_rate": 4.995136118123406e-05, "loss": 0.8394, "step": 19957 }, { "epoch": 0.02, "learning_rate": 4.995135628379096e-05, "loss": 0.9696, "step": 19958 }, { "epoch": 0.02, "learning_rate": 4.9951351386101555e-05, "loss": 0.9729, "step": 19959 }, { "epoch": 0.02, "learning_rate": 4.995134648816583e-05, "loss": 1.1364, "step": 19960 }, { "epoch": 0.02, "learning_rate": 4.995134158998381e-05, "loss": 1.1014, "step": 19961 }, { "epoch": 0.02, "learning_rate": 4.995133669155547e-05, "loss": 1.1283, "step": 19962 }, { "epoch": 0.02, "learning_rate": 4.995133179288083e-05, "loss": 1.0353, "step": 19963 }, { "epoch": 0.02, "learning_rate": 4.9951326893959875e-05, "loss": 0.9008, "step": 19964 }, { "epoch": 0.02, "learning_rate": 4.995132199479261e-05, "loss": 1.1224, "step": 19965 }, { "epoch": 0.02, "learning_rate": 4.9951317095379044e-05, "loss": 1.0088, "step": 19966 }, { "epoch": 0.02, "learning_rate": 4.995131219571916e-05, "loss": 1.0762, "step": 19967 }, { "epoch": 0.02, "learning_rate": 4.995130729581298e-05, "loss": 1.0413, "step": 19968 }, { "epoch": 0.02, "learning_rate": 4.9951302395660477e-05, "loss": 0.8766, "step": 19969 }, { "epoch": 0.02, "learning_rate": 4.995129749526167e-05, "loss": 1.0994, "step": 19970 }, { "epoch": 0.02, "learning_rate": 4.995129259461655e-05, "loss": 1.1821, "step": 19971 }, { "epoch": 0.02, "learning_rate": 4.995128769372513e-05, "loss": 1.0293, "step": 19972 }, { "epoch": 0.02, "learning_rate": 4.995128279258739e-05, "loss": 1.1084, "step": 19973 }, { "epoch": 0.02, "learning_rate": 4.995127789120336e-05, "loss": 1.1738, "step": 19974 }, { "epoch": 0.02, "learning_rate": 4.995127298957301e-05, "loss": 0.6615, "step": 19975 }, { "epoch": 0.02, "learning_rate": 4.9951268087696346e-05, "loss": 1.1627, "step": 19976 }, { "epoch": 0.02, "learning_rate": 4.995126318557338e-05, "loss": 1.1334, "step": 19977 }, { "epoch": 0.02, "learning_rate": 4.9951258283204106e-05, "loss": 1.0451, "step": 19978 }, { "epoch": 0.02, "learning_rate": 4.995125338058853e-05, "loss": 0.8299, "step": 19979 }, { "epoch": 0.02, "learning_rate": 4.995124847772663e-05, "loss": 0.9589, "step": 19980 }, { "epoch": 0.02, "learning_rate": 4.995124357461843e-05, "loss": 1.0443, "step": 19981 }, { "epoch": 0.02, "learning_rate": 4.995123867126393e-05, "loss": 1.3036, "step": 19982 }, { "epoch": 0.02, "learning_rate": 4.995123376766311e-05, "loss": 1.1141, "step": 19983 }, { "epoch": 0.02, "learning_rate": 4.995122886381599e-05, "loss": 1.1289, "step": 19984 }, { "epoch": 0.02, "learning_rate": 4.9951223959722556e-05, "loss": 1.2433, "step": 19985 }, { "epoch": 0.02, "learning_rate": 4.995121905538281e-05, "loss": 1.2589, "step": 19986 }, { "epoch": 0.02, "learning_rate": 4.995121415079676e-05, "loss": 1.1457, "step": 19987 }, { "epoch": 0.02, "learning_rate": 4.995120924596442e-05, "loss": 1.1782, "step": 19988 }, { "epoch": 0.02, "learning_rate": 4.995120434088575e-05, "loss": 1.4586, "step": 19989 }, { "epoch": 0.02, "learning_rate": 4.995119943556078e-05, "loss": 1.1532, "step": 19990 }, { "epoch": 0.02, "learning_rate": 4.99511945299895e-05, "loss": 1.1147, "step": 19991 }, { "epoch": 0.02, "learning_rate": 4.995118962417191e-05, "loss": 0.8136, "step": 19992 }, { "epoch": 0.02, "learning_rate": 4.995118471810802e-05, "loss": 0.7862, "step": 19993 }, { "epoch": 0.02, "learning_rate": 4.995117981179781e-05, "loss": 0.8302, "step": 19994 }, { "epoch": 0.02, "learning_rate": 4.99511749052413e-05, "loss": 0.6526, "step": 19995 }, { "epoch": 0.02, "learning_rate": 4.995116999843849e-05, "loss": 0.9735, "step": 19996 }, { "epoch": 0.02, "learning_rate": 4.995116509138936e-05, "loss": 0.8907, "step": 19997 }, { "epoch": 0.02, "learning_rate": 4.995116018409394e-05, "loss": 0.858, "step": 19998 }, { "epoch": 0.02, "learning_rate": 4.99511552765522e-05, "loss": 0.8701, "step": 19999 }, { "epoch": 0.02, "learning_rate": 4.995115036876415e-05, "loss": 1.1036, "step": 20000 }, { "epoch": 0.02, "eval_loss": 1.050695538520813, "eval_runtime": 85.5766, "eval_samples_per_second": 16.184, "eval_steps_per_second": 4.055, "step": 20000 }, { "epoch": 0.02, "learning_rate": 4.9951145460729795e-05, "loss": 1.0065, "step": 20001 }, { "epoch": 0.02, "learning_rate": 4.995114055244914e-05, "loss": 1.0904, "step": 20002 }, { "epoch": 0.02, "learning_rate": 4.9951135643922164e-05, "loss": 1.1266, "step": 20003 }, { "epoch": 0.02, "learning_rate": 4.995113073514889e-05, "loss": 0.9593, "step": 20004 }, { "epoch": 0.02, "learning_rate": 4.995112582612931e-05, "loss": 1.046, "step": 20005 }, { "epoch": 0.02, "learning_rate": 4.9951120916863426e-05, "loss": 1.0173, "step": 20006 }, { "epoch": 0.02, "learning_rate": 4.995111600735123e-05, "loss": 1.2986, "step": 20007 }, { "epoch": 0.02, "learning_rate": 4.9951111097592724e-05, "loss": 0.9342, "step": 20008 }, { "epoch": 0.02, "learning_rate": 4.9951106187587914e-05, "loss": 1.1265, "step": 20009 }, { "epoch": 0.02, "learning_rate": 4.99511012773368e-05, "loss": 1.1941, "step": 20010 }, { "epoch": 0.02, "learning_rate": 4.995109636683937e-05, "loss": 1.189, "step": 20011 }, { "epoch": 0.02, "learning_rate": 4.995109145609564e-05, "loss": 1.3733, "step": 20012 }, { "epoch": 0.02, "learning_rate": 4.9951086545105606e-05, "loss": 1.0884, "step": 20013 }, { "epoch": 0.02, "learning_rate": 4.9951081633869264e-05, "loss": 0.9061, "step": 20014 }, { "epoch": 0.02, "learning_rate": 4.995107672238661e-05, "loss": 1.0685, "step": 20015 }, { "epoch": 0.02, "learning_rate": 4.995107181065766e-05, "loss": 0.9686, "step": 20016 }, { "epoch": 0.02, "learning_rate": 4.99510668986824e-05, "loss": 0.8972, "step": 20017 }, { "epoch": 0.02, "learning_rate": 4.995106198646082e-05, "loss": 1.0111, "step": 20018 }, { "epoch": 0.02, "learning_rate": 4.995105707399295e-05, "loss": 1.1011, "step": 20019 }, { "epoch": 0.02, "learning_rate": 4.9951052161278765e-05, "loss": 1.2997, "step": 20020 }, { "epoch": 0.02, "learning_rate": 4.9951047248318285e-05, "loss": 1.198, "step": 20021 }, { "epoch": 0.02, "learning_rate": 4.995104233511149e-05, "loss": 1.65, "step": 20022 }, { "epoch": 0.02, "learning_rate": 4.9951037421658386e-05, "loss": 2.1101, "step": 20023 }, { "epoch": 0.02, "learning_rate": 4.995103250795898e-05, "loss": 0.8452, "step": 20024 }, { "epoch": 0.02, "learning_rate": 4.9951027594013266e-05, "loss": 0.653, "step": 20025 }, { "epoch": 0.02, "learning_rate": 4.995102267982125e-05, "loss": 1.1717, "step": 20026 }, { "epoch": 0.02, "learning_rate": 4.9951017765382925e-05, "loss": 1.0334, "step": 20027 }, { "epoch": 0.02, "learning_rate": 4.995101285069829e-05, "loss": 1.0921, "step": 20028 }, { "epoch": 0.02, "learning_rate": 4.9951007935767356e-05, "loss": 1.4456, "step": 20029 }, { "epoch": 0.02, "learning_rate": 4.9951003020590116e-05, "loss": 1.0322, "step": 20030 }, { "epoch": 0.02, "learning_rate": 4.9950998105166565e-05, "loss": 1.2649, "step": 20031 }, { "epoch": 0.02, "learning_rate": 4.995099318949671e-05, "loss": 0.9538, "step": 20032 }, { "epoch": 0.02, "learning_rate": 4.995098827358055e-05, "loss": 1.0172, "step": 20033 }, { "epoch": 0.02, "learning_rate": 4.9950983357418085e-05, "loss": 1.0325, "step": 20034 }, { "epoch": 0.02, "learning_rate": 4.995097844100931e-05, "loss": 1.1101, "step": 20035 }, { "epoch": 0.02, "learning_rate": 4.995097352435424e-05, "loss": 0.8585, "step": 20036 }, { "epoch": 0.02, "learning_rate": 4.995096860745285e-05, "loss": 1.0438, "step": 20037 }, { "epoch": 0.02, "learning_rate": 4.995096369030517e-05, "loss": 0.5976, "step": 20038 }, { "epoch": 0.02, "learning_rate": 4.995095877291117e-05, "loss": 0.7719, "step": 20039 }, { "epoch": 0.02, "learning_rate": 4.995095385527088e-05, "loss": 1.177, "step": 20040 }, { "epoch": 0.02, "learning_rate": 4.995094893738427e-05, "loss": 1.3327, "step": 20041 }, { "epoch": 0.02, "learning_rate": 4.995094401925137e-05, "loss": 0.9201, "step": 20042 }, { "epoch": 0.02, "learning_rate": 4.9950939100872154e-05, "loss": 1.1723, "step": 20043 }, { "epoch": 0.02, "learning_rate": 4.9950934182246636e-05, "loss": 1.1364, "step": 20044 }, { "epoch": 0.02, "learning_rate": 4.995092926337481e-05, "loss": 0.9087, "step": 20045 }, { "epoch": 0.02, "learning_rate": 4.9950924344256676e-05, "loss": 0.9299, "step": 20046 }, { "epoch": 0.02, "learning_rate": 4.995091942489225e-05, "loss": 0.6843, "step": 20047 }, { "epoch": 0.02, "learning_rate": 4.995091450528151e-05, "loss": 0.9089, "step": 20048 }, { "epoch": 0.02, "learning_rate": 4.995090958542446e-05, "loss": 0.9763, "step": 20049 }, { "epoch": 0.02, "learning_rate": 4.995090466532111e-05, "loss": 1.092, "step": 20050 }, { "epoch": 0.02, "learning_rate": 4.995089974497146e-05, "loss": 1.1509, "step": 20051 }, { "epoch": 0.02, "learning_rate": 4.99508948243755e-05, "loss": 0.759, "step": 20052 }, { "epoch": 0.02, "learning_rate": 4.9950889903533236e-05, "loss": 0.7007, "step": 20053 }, { "epoch": 0.02, "learning_rate": 4.995088498244467e-05, "loss": 1.0243, "step": 20054 }, { "epoch": 0.02, "learning_rate": 4.99508800611098e-05, "loss": 1.1457, "step": 20055 }, { "epoch": 0.02, "learning_rate": 4.995087513952862e-05, "loss": 1.1078, "step": 20056 }, { "epoch": 0.02, "learning_rate": 4.995087021770114e-05, "loss": 1.1792, "step": 20057 }, { "epoch": 0.02, "learning_rate": 4.995086529562735e-05, "loss": 0.8517, "step": 20058 }, { "epoch": 0.02, "learning_rate": 4.995086037330726e-05, "loss": 1.0568, "step": 20059 }, { "epoch": 0.02, "learning_rate": 4.9950855450740865e-05, "loss": 0.9297, "step": 20060 }, { "epoch": 0.02, "learning_rate": 4.9950850527928165e-05, "loss": 1.0727, "step": 20061 }, { "epoch": 0.02, "learning_rate": 4.995084560486916e-05, "loss": 1.1094, "step": 20062 }, { "epoch": 0.02, "learning_rate": 4.9950840681563854e-05, "loss": 1.0855, "step": 20063 }, { "epoch": 0.02, "learning_rate": 4.995083575801224e-05, "loss": 0.9035, "step": 20064 }, { "epoch": 0.02, "learning_rate": 4.995083083421432e-05, "loss": 1.1314, "step": 20065 }, { "epoch": 0.02, "learning_rate": 4.995082591017011e-05, "loss": 1.1572, "step": 20066 }, { "epoch": 0.02, "learning_rate": 4.995082098587958e-05, "loss": 1.3336, "step": 20067 }, { "epoch": 0.02, "learning_rate": 4.9950816061342756e-05, "loss": 0.8142, "step": 20068 }, { "epoch": 0.02, "learning_rate": 4.995081113655962e-05, "loss": 0.9544, "step": 20069 }, { "epoch": 0.02, "learning_rate": 4.995080621153019e-05, "loss": 0.9004, "step": 20070 }, { "epoch": 0.02, "learning_rate": 4.9950801286254445e-05, "loss": 0.7088, "step": 20071 }, { "epoch": 0.02, "learning_rate": 4.99507963607324e-05, "loss": 1.2632, "step": 20072 }, { "epoch": 0.02, "learning_rate": 4.995079143496405e-05, "loss": 0.9631, "step": 20073 }, { "epoch": 0.02, "learning_rate": 4.995078650894941e-05, "loss": 0.6777, "step": 20074 }, { "epoch": 0.02, "learning_rate": 4.995078158268845e-05, "loss": 0.9409, "step": 20075 }, { "epoch": 0.02, "learning_rate": 4.995077665618119e-05, "loss": 0.8817, "step": 20076 }, { "epoch": 0.02, "learning_rate": 4.9950771729427624e-05, "loss": 0.9843, "step": 20077 }, { "epoch": 0.02, "learning_rate": 4.995076680242776e-05, "loss": 1.1929, "step": 20078 }, { "epoch": 0.02, "learning_rate": 4.9950761875181597e-05, "loss": 1.1861, "step": 20079 }, { "epoch": 0.02, "learning_rate": 4.995075694768912e-05, "loss": 1.0514, "step": 20080 }, { "epoch": 0.02, "learning_rate": 4.995075201995034e-05, "loss": 1.137, "step": 20081 }, { "epoch": 0.02, "learning_rate": 4.9950747091965265e-05, "loss": 1.1693, "step": 20082 }, { "epoch": 0.02, "learning_rate": 4.9950742163733885e-05, "loss": 1.0282, "step": 20083 }, { "epoch": 0.02, "learning_rate": 4.9950737235256195e-05, "loss": 1.0292, "step": 20084 }, { "epoch": 0.02, "learning_rate": 4.995073230653221e-05, "loss": 1.205, "step": 20085 }, { "epoch": 0.02, "learning_rate": 4.995072737756192e-05, "loss": 1.0222, "step": 20086 }, { "epoch": 0.02, "learning_rate": 4.995072244834532e-05, "loss": 0.9928, "step": 20087 }, { "epoch": 0.02, "learning_rate": 4.995071751888243e-05, "loss": 1.2947, "step": 20088 }, { "epoch": 0.02, "learning_rate": 4.995071258917322e-05, "loss": 1.2765, "step": 20089 }, { "epoch": 0.02, "learning_rate": 4.995070765921772e-05, "loss": 1.2854, "step": 20090 }, { "epoch": 0.02, "learning_rate": 4.9950702729015916e-05, "loss": 1.059, "step": 20091 }, { "epoch": 0.02, "learning_rate": 4.9950697798567803e-05, "loss": 1.0854, "step": 20092 }, { "epoch": 0.02, "learning_rate": 4.9950692867873394e-05, "loss": 1.2086, "step": 20093 }, { "epoch": 0.02, "learning_rate": 4.9950687936932674e-05, "loss": 1.0258, "step": 20094 }, { "epoch": 0.02, "learning_rate": 4.995068300574566e-05, "loss": 1.0001, "step": 20095 }, { "epoch": 0.02, "learning_rate": 4.995067807431234e-05, "loss": 0.8674, "step": 20096 }, { "epoch": 0.02, "learning_rate": 4.995067314263272e-05, "loss": 1.2002, "step": 20097 }, { "epoch": 0.02, "learning_rate": 4.995066821070679e-05, "loss": 1.0499, "step": 20098 }, { "epoch": 0.02, "learning_rate": 4.995066327853456e-05, "loss": 1.2658, "step": 20099 }, { "epoch": 0.02, "learning_rate": 4.9950658346116034e-05, "loss": 1.2492, "step": 20100 }, { "epoch": 0.02, "learning_rate": 4.9950653413451195e-05, "loss": 1.1206, "step": 20101 }, { "epoch": 0.02, "learning_rate": 4.995064848054006e-05, "loss": 1.1142, "step": 20102 }, { "epoch": 0.02, "learning_rate": 4.995064354738263e-05, "loss": 1.0411, "step": 20103 }, { "epoch": 0.02, "learning_rate": 4.995063861397889e-05, "loss": 1.0506, "step": 20104 }, { "epoch": 0.02, "learning_rate": 4.995063368032885e-05, "loss": 0.8861, "step": 20105 }, { "epoch": 0.02, "learning_rate": 4.99506287464325e-05, "loss": 1.3545, "step": 20106 }, { "epoch": 0.02, "learning_rate": 4.995062381228986e-05, "loss": 1.0546, "step": 20107 }, { "epoch": 0.02, "learning_rate": 4.995061887790091e-05, "loss": 1.162, "step": 20108 }, { "epoch": 0.02, "learning_rate": 4.995061394326566e-05, "loss": 1.1609, "step": 20109 }, { "epoch": 0.02, "learning_rate": 4.995060900838411e-05, "loss": 1.2274, "step": 20110 }, { "epoch": 0.02, "learning_rate": 4.9950604073256254e-05, "loss": 1.3624, "step": 20111 }, { "epoch": 0.02, "learning_rate": 4.9950599137882096e-05, "loss": 1.1483, "step": 20112 }, { "epoch": 0.02, "learning_rate": 4.995059420226164e-05, "loss": 1.1703, "step": 20113 }, { "epoch": 0.02, "learning_rate": 4.995058926639488e-05, "loss": 0.8651, "step": 20114 }, { "epoch": 0.02, "learning_rate": 4.995058433028182e-05, "loss": 0.9816, "step": 20115 }, { "epoch": 0.02, "learning_rate": 4.9950579393922455e-05, "loss": 0.9014, "step": 20116 }, { "epoch": 0.02, "learning_rate": 4.995057445731679e-05, "loss": 0.9993, "step": 20117 }, { "epoch": 0.02, "learning_rate": 4.995056952046483e-05, "loss": 1.0268, "step": 20118 }, { "epoch": 0.02, "learning_rate": 4.9950564583366564e-05, "loss": 0.8134, "step": 20119 }, { "epoch": 0.02, "learning_rate": 4.995055964602199e-05, "loss": 1.3194, "step": 20120 }, { "epoch": 0.02, "learning_rate": 4.995055470843112e-05, "loss": 0.8554, "step": 20121 }, { "epoch": 0.02, "learning_rate": 4.9950549770593946e-05, "loss": 1.1572, "step": 20122 }, { "epoch": 0.02, "learning_rate": 4.9950544832510475e-05, "loss": 1.0276, "step": 20123 }, { "epoch": 0.02, "learning_rate": 4.99505398941807e-05, "loss": 1.1513, "step": 20124 }, { "epoch": 0.02, "learning_rate": 4.995053495560463e-05, "loss": 1.0222, "step": 20125 }, { "epoch": 0.02, "learning_rate": 4.995053001678225e-05, "loss": 1.0026, "step": 20126 }, { "epoch": 0.02, "learning_rate": 4.995052507771357e-05, "loss": 1.0999, "step": 20127 }, { "epoch": 0.02, "learning_rate": 4.9950520138398594e-05, "loss": 0.9272, "step": 20128 }, { "epoch": 0.02, "learning_rate": 4.995051519883731e-05, "loss": 0.5443, "step": 20129 }, { "epoch": 0.02, "learning_rate": 4.995051025902974e-05, "loss": 0.4532, "step": 20130 }, { "epoch": 0.02, "learning_rate": 4.995050531897585e-05, "loss": 1.1827, "step": 20131 }, { "epoch": 0.02, "learning_rate": 4.995050037867567e-05, "loss": 1.1481, "step": 20132 }, { "epoch": 0.02, "learning_rate": 4.995049543812918e-05, "loss": 0.9864, "step": 20133 }, { "epoch": 0.02, "learning_rate": 4.99504904973364e-05, "loss": 0.9894, "step": 20134 }, { "epoch": 0.02, "learning_rate": 4.995048555629732e-05, "loss": 1.2225, "step": 20135 }, { "epoch": 0.02, "learning_rate": 4.995048061501193e-05, "loss": 1.2814, "step": 20136 }, { "epoch": 0.02, "learning_rate": 4.995047567348025e-05, "loss": 1.23, "step": 20137 }, { "epoch": 0.02, "learning_rate": 4.9950470731702256e-05, "loss": 1.2809, "step": 20138 }, { "epoch": 0.02, "learning_rate": 4.995046578967797e-05, "loss": 1.4328, "step": 20139 }, { "epoch": 0.02, "learning_rate": 4.995046084740739e-05, "loss": 1.3739, "step": 20140 }, { "epoch": 0.02, "learning_rate": 4.9950455904890494e-05, "loss": 1.2828, "step": 20141 }, { "epoch": 0.02, "learning_rate": 4.995045096212731e-05, "loss": 0.9902, "step": 20142 }, { "epoch": 0.02, "learning_rate": 4.995044601911782e-05, "loss": 1.2222, "step": 20143 }, { "epoch": 0.02, "learning_rate": 4.9950441075862033e-05, "loss": 0.5884, "step": 20144 }, { "epoch": 0.02, "learning_rate": 4.995043613235994e-05, "loss": 1.0651, "step": 20145 }, { "epoch": 0.02, "learning_rate": 4.995043118861155e-05, "loss": 1.1243, "step": 20146 }, { "epoch": 0.02, "learning_rate": 4.995042624461687e-05, "loss": 0.9865, "step": 20147 }, { "epoch": 0.02, "learning_rate": 4.9950421300375874e-05, "loss": 1.0428, "step": 20148 }, { "epoch": 0.02, "learning_rate": 4.995041635588858e-05, "loss": 1.1082, "step": 20149 }, { "epoch": 0.02, "learning_rate": 4.9950411411154996e-05, "loss": 1.3345, "step": 20150 }, { "epoch": 0.02, "learning_rate": 4.9950406466175104e-05, "loss": 1.0685, "step": 20151 }, { "epoch": 0.02, "learning_rate": 4.9950401520948917e-05, "loss": 0.8446, "step": 20152 }, { "epoch": 0.02, "learning_rate": 4.995039657547642e-05, "loss": 0.9485, "step": 20153 }, { "epoch": 0.02, "learning_rate": 4.995039162975763e-05, "loss": 1.1102, "step": 20154 }, { "epoch": 0.02, "learning_rate": 4.9950386683792545e-05, "loss": 1.2435, "step": 20155 }, { "epoch": 0.02, "learning_rate": 4.9950381737581156e-05, "loss": 1.1389, "step": 20156 }, { "epoch": 0.02, "learning_rate": 4.995037679112346e-05, "loss": 1.3285, "step": 20157 }, { "epoch": 0.02, "learning_rate": 4.995037184441948e-05, "loss": 0.9419, "step": 20158 }, { "epoch": 0.02, "learning_rate": 4.9950366897469194e-05, "loss": 1.0364, "step": 20159 }, { "epoch": 0.02, "learning_rate": 4.9950361950272605e-05, "loss": 1.1671, "step": 20160 }, { "epoch": 0.02, "learning_rate": 4.995035700282972e-05, "loss": 1.0167, "step": 20161 }, { "epoch": 0.02, "learning_rate": 4.9950352055140534e-05, "loss": 1.1601, "step": 20162 }, { "epoch": 0.02, "learning_rate": 4.995034710720505e-05, "loss": 1.352, "step": 20163 }, { "epoch": 0.02, "learning_rate": 4.995034215902326e-05, "loss": 1.1218, "step": 20164 }, { "epoch": 0.02, "learning_rate": 4.9950337210595176e-05, "loss": 1.4909, "step": 20165 }, { "epoch": 0.02, "learning_rate": 4.99503322619208e-05, "loss": 1.2785, "step": 20166 }, { "epoch": 0.02, "learning_rate": 4.995032731300011e-05, "loss": 1.038, "step": 20167 }, { "epoch": 0.02, "learning_rate": 4.995032236383314e-05, "loss": 0.8988, "step": 20168 }, { "epoch": 0.02, "learning_rate": 4.995031741441985e-05, "loss": 0.8175, "step": 20169 }, { "epoch": 0.02, "learning_rate": 4.995031246476027e-05, "loss": 0.919, "step": 20170 }, { "epoch": 0.02, "learning_rate": 4.995030751485439e-05, "loss": 0.8404, "step": 20171 }, { "epoch": 0.02, "learning_rate": 4.995030256470221e-05, "loss": 1.2985, "step": 20172 }, { "epoch": 0.02, "learning_rate": 4.995029761430374e-05, "loss": 1.2081, "step": 20173 }, { "epoch": 0.02, "learning_rate": 4.995029266365896e-05, "loss": 1.0379, "step": 20174 }, { "epoch": 0.02, "learning_rate": 4.995028771276789e-05, "loss": 0.9316, "step": 20175 }, { "epoch": 0.02, "learning_rate": 4.995028276163052e-05, "loss": 0.7401, "step": 20176 }, { "epoch": 0.02, "learning_rate": 4.995027781024685e-05, "loss": 0.8672, "step": 20177 }, { "epoch": 0.02, "learning_rate": 4.9950272858616873e-05, "loss": 0.9295, "step": 20178 }, { "epoch": 0.02, "learning_rate": 4.995026790674061e-05, "loss": 1.2057, "step": 20179 }, { "epoch": 0.02, "learning_rate": 4.995026295461804e-05, "loss": 0.9457, "step": 20180 }, { "epoch": 0.02, "learning_rate": 4.995025800224917e-05, "loss": 1.0979, "step": 20181 }, { "epoch": 0.02, "learning_rate": 4.995025304963401e-05, "loss": 1.2415, "step": 20182 }, { "epoch": 0.02, "learning_rate": 4.9950248096772545e-05, "loss": 1.1573, "step": 20183 }, { "epoch": 0.02, "learning_rate": 4.995024314366479e-05, "loss": 1.164, "step": 20184 }, { "epoch": 0.02, "learning_rate": 4.9950238190310724e-05, "loss": 1.1329, "step": 20185 }, { "epoch": 0.02, "learning_rate": 4.9950233236710376e-05, "loss": 0.9043, "step": 20186 }, { "epoch": 0.02, "learning_rate": 4.9950228282863724e-05, "loss": 1.0337, "step": 20187 }, { "epoch": 0.02, "learning_rate": 4.995022332877076e-05, "loss": 1.4155, "step": 20188 }, { "epoch": 0.02, "learning_rate": 4.9950218374431515e-05, "loss": 1.2085, "step": 20189 }, { "epoch": 0.02, "learning_rate": 4.995021341984596e-05, "loss": 1.3255, "step": 20190 }, { "epoch": 0.02, "learning_rate": 4.995020846501411e-05, "loss": 0.9157, "step": 20191 }, { "epoch": 0.02, "learning_rate": 4.995020350993597e-05, "loss": 0.9245, "step": 20192 }, { "epoch": 0.02, "learning_rate": 4.995019855461153e-05, "loss": 1.3599, "step": 20193 }, { "epoch": 0.02, "learning_rate": 4.9950193599040785e-05, "loss": 0.6718, "step": 20194 }, { "epoch": 0.02, "learning_rate": 4.9950188643223744e-05, "loss": 0.2402, "step": 20195 }, { "epoch": 0.02, "learning_rate": 4.9950183687160407e-05, "loss": 0.7368, "step": 20196 }, { "epoch": 0.02, "learning_rate": 4.995017873085077e-05, "loss": 1.0651, "step": 20197 }, { "epoch": 0.02, "learning_rate": 4.995017377429484e-05, "loss": 1.0939, "step": 20198 }, { "epoch": 0.02, "learning_rate": 4.995016881749261e-05, "loss": 1.2186, "step": 20199 }, { "epoch": 0.02, "learning_rate": 4.995016386044408e-05, "loss": 1.2096, "step": 20200 }, { "epoch": 0.02, "learning_rate": 4.995015890314926e-05, "loss": 1.23, "step": 20201 }, { "epoch": 0.02, "learning_rate": 4.9950153945608134e-05, "loss": 1.1173, "step": 20202 }, { "epoch": 0.02, "learning_rate": 4.995014898782072e-05, "loss": 1.1534, "step": 20203 }, { "epoch": 0.02, "learning_rate": 4.9950144029787e-05, "loss": 0.979, "step": 20204 }, { "epoch": 0.02, "learning_rate": 4.9950139071506984e-05, "loss": 1.1825, "step": 20205 }, { "epoch": 0.02, "learning_rate": 4.995013411298067e-05, "loss": 1.1874, "step": 20206 }, { "epoch": 0.02, "learning_rate": 4.995012915420806e-05, "loss": 1.1811, "step": 20207 }, { "epoch": 0.02, "learning_rate": 4.9950124195189155e-05, "loss": 1.1252, "step": 20208 }, { "epoch": 0.02, "learning_rate": 4.995011923592395e-05, "loss": 1.206, "step": 20209 }, { "epoch": 0.02, "learning_rate": 4.995011427641245e-05, "loss": 1.2616, "step": 20210 }, { "epoch": 0.02, "learning_rate": 4.995010931665466e-05, "loss": 1.1561, "step": 20211 }, { "epoch": 0.02, "learning_rate": 4.995010435665056e-05, "loss": 1.6784, "step": 20212 }, { "epoch": 0.02, "learning_rate": 4.9950099396400176e-05, "loss": 1.187, "step": 20213 }, { "epoch": 0.02, "learning_rate": 4.995009443590348e-05, "loss": 1.1191, "step": 20214 }, { "epoch": 0.02, "learning_rate": 4.99500894751605e-05, "loss": 0.9832, "step": 20215 }, { "epoch": 0.02, "learning_rate": 4.9950084514171216e-05, "loss": 0.9797, "step": 20216 }, { "epoch": 0.02, "learning_rate": 4.995007955293564e-05, "loss": 1.0893, "step": 20217 }, { "epoch": 0.02, "learning_rate": 4.995007459145376e-05, "loss": 1.1748, "step": 20218 }, { "epoch": 0.02, "learning_rate": 4.995006962972559e-05, "loss": 1.1114, "step": 20219 }, { "epoch": 0.02, "learning_rate": 4.995006466775112e-05, "loss": 1.2227, "step": 20220 }, { "epoch": 0.02, "learning_rate": 4.995005970553036e-05, "loss": 0.9083, "step": 20221 }, { "epoch": 0.02, "learning_rate": 4.99500547430633e-05, "loss": 0.1676, "step": 20222 }, { "epoch": 0.02, "learning_rate": 4.995004978034994e-05, "loss": 0.2449, "step": 20223 }, { "epoch": 0.02, "learning_rate": 4.9950044817390285e-05, "loss": 0.3735, "step": 20224 }, { "epoch": 0.02, "learning_rate": 4.995003985418434e-05, "loss": 1.2023, "step": 20225 }, { "epoch": 0.02, "learning_rate": 4.995003489073209e-05, "loss": 1.1682, "step": 20226 }, { "epoch": 0.02, "learning_rate": 4.995002992703355e-05, "loss": 1.2588, "step": 20227 }, { "epoch": 0.02, "learning_rate": 4.9950024963088706e-05, "loss": 1.335, "step": 20228 }, { "epoch": 0.02, "learning_rate": 4.995001999889757e-05, "loss": 1.1518, "step": 20229 }, { "epoch": 0.02, "learning_rate": 4.995001503446014e-05, "loss": 1.1241, "step": 20230 }, { "epoch": 0.02, "learning_rate": 4.9950010069776414e-05, "loss": 0.9709, "step": 20231 }, { "epoch": 0.02, "learning_rate": 4.995000510484639e-05, "loss": 1.163, "step": 20232 }, { "epoch": 0.02, "learning_rate": 4.995000013967007e-05, "loss": 1.0282, "step": 20233 }, { "epoch": 0.02, "learning_rate": 4.994999517424746e-05, "loss": 1.606, "step": 20234 }, { "epoch": 0.02, "learning_rate": 4.994999020857854e-05, "loss": 1.2887, "step": 20235 }, { "epoch": 0.02, "learning_rate": 4.994998524266334e-05, "loss": 1.0431, "step": 20236 }, { "epoch": 0.02, "learning_rate": 4.994998027650184e-05, "loss": 1.0777, "step": 20237 }, { "epoch": 0.02, "learning_rate": 4.994997531009404e-05, "loss": 0.995, "step": 20238 }, { "epoch": 0.02, "learning_rate": 4.994997034343994e-05, "loss": 0.9552, "step": 20239 }, { "epoch": 0.02, "learning_rate": 4.994996537653955e-05, "loss": 1.3656, "step": 20240 }, { "epoch": 0.02, "learning_rate": 4.994996040939287e-05, "loss": 1.3398, "step": 20241 }, { "epoch": 0.02, "learning_rate": 4.994995544199989e-05, "loss": 1.2442, "step": 20242 }, { "epoch": 0.02, "learning_rate": 4.994995047436062e-05, "loss": 0.702, "step": 20243 }, { "epoch": 0.02, "learning_rate": 4.994994550647504e-05, "loss": 1.2549, "step": 20244 }, { "epoch": 0.02, "learning_rate": 4.994994053834318e-05, "loss": 1.011, "step": 20245 }, { "epoch": 0.02, "learning_rate": 4.994993556996501e-05, "loss": 1.2791, "step": 20246 }, { "epoch": 0.02, "learning_rate": 4.9949930601340554e-05, "loss": 1.1118, "step": 20247 }, { "epoch": 0.02, "learning_rate": 4.994992563246981e-05, "loss": 1.1697, "step": 20248 }, { "epoch": 0.02, "learning_rate": 4.994992066335276e-05, "loss": 1.3886, "step": 20249 }, { "epoch": 0.02, "learning_rate": 4.994991569398941e-05, "loss": 1.0775, "step": 20250 }, { "epoch": 0.02, "learning_rate": 4.994991072437978e-05, "loss": 1.2219, "step": 20251 }, { "epoch": 0.02, "learning_rate": 4.9949905754523845e-05, "loss": 1.4358, "step": 20252 }, { "epoch": 0.02, "learning_rate": 4.9949900784421614e-05, "loss": 0.98, "step": 20253 }, { "epoch": 0.02, "learning_rate": 4.9949895814073093e-05, "loss": 0.8539, "step": 20254 }, { "epoch": 0.02, "learning_rate": 4.9949890843478275e-05, "loss": 0.9379, "step": 20255 }, { "epoch": 0.02, "learning_rate": 4.994988587263717e-05, "loss": 0.5159, "step": 20256 }, { "epoch": 0.02, "learning_rate": 4.994988090154976e-05, "loss": 0.8544, "step": 20257 }, { "epoch": 0.02, "learning_rate": 4.9949875930216054e-05, "loss": 0.6313, "step": 20258 }, { "epoch": 0.02, "learning_rate": 4.9949870958636056e-05, "loss": 0.9562, "step": 20259 }, { "epoch": 0.02, "learning_rate": 4.994986598680977e-05, "loss": 0.8234, "step": 20260 }, { "epoch": 0.02, "learning_rate": 4.994986101473719e-05, "loss": 1.097, "step": 20261 }, { "epoch": 0.02, "learning_rate": 4.99498560424183e-05, "loss": 1.0396, "step": 20262 }, { "epoch": 0.02, "learning_rate": 4.994985106985313e-05, "loss": 1.1175, "step": 20263 }, { "epoch": 0.02, "learning_rate": 4.9949846097041664e-05, "loss": 0.9861, "step": 20264 }, { "epoch": 0.02, "learning_rate": 4.99498411239839e-05, "loss": 1.259, "step": 20265 }, { "epoch": 0.02, "learning_rate": 4.994983615067984e-05, "loss": 1.0801, "step": 20266 }, { "epoch": 0.02, "learning_rate": 4.994983117712949e-05, "loss": 1.0298, "step": 20267 }, { "epoch": 0.02, "learning_rate": 4.994982620333284e-05, "loss": 1.2786, "step": 20268 }, { "epoch": 0.02, "learning_rate": 4.99498212292899e-05, "loss": 1.1202, "step": 20269 }, { "epoch": 0.02, "learning_rate": 4.994981625500067e-05, "loss": 1.2205, "step": 20270 }, { "epoch": 0.02, "learning_rate": 4.994981128046514e-05, "loss": 0.9688, "step": 20271 }, { "epoch": 0.02, "learning_rate": 4.994980630568332e-05, "loss": 0.9288, "step": 20272 }, { "epoch": 0.02, "learning_rate": 4.99498013306552e-05, "loss": 1.2016, "step": 20273 }, { "epoch": 0.02, "learning_rate": 4.994979635538079e-05, "loss": 1.3092, "step": 20274 }, { "epoch": 0.02, "learning_rate": 4.994979137986009e-05, "loss": 1.02, "step": 20275 }, { "epoch": 0.02, "learning_rate": 4.994978640409308e-05, "loss": 0.9183, "step": 20276 }, { "epoch": 0.02, "learning_rate": 4.9949781428079795e-05, "loss": 1.1259, "step": 20277 }, { "epoch": 0.02, "learning_rate": 4.994977645182021e-05, "loss": 1.1506, "step": 20278 }, { "epoch": 0.02, "learning_rate": 4.9949771475314335e-05, "loss": 0.9443, "step": 20279 }, { "epoch": 0.02, "learning_rate": 4.994976649856216e-05, "loss": 1.1103, "step": 20280 }, { "epoch": 0.02, "learning_rate": 4.994976152156369e-05, "loss": 1.1721, "step": 20281 }, { "epoch": 0.02, "learning_rate": 4.9949756544318924e-05, "loss": 1.0795, "step": 20282 }, { "epoch": 0.02, "learning_rate": 4.9949751566827875e-05, "loss": 1.1299, "step": 20283 }, { "epoch": 0.02, "learning_rate": 4.994974658909053e-05, "loss": 0.9313, "step": 20284 }, { "epoch": 0.02, "learning_rate": 4.9949741611106895e-05, "loss": 1.1373, "step": 20285 }, { "epoch": 0.02, "learning_rate": 4.9949736632876956e-05, "loss": 1.0785, "step": 20286 }, { "epoch": 0.02, "learning_rate": 4.9949731654400727e-05, "loss": 0.9043, "step": 20287 }, { "epoch": 0.02, "learning_rate": 4.994972667567821e-05, "loss": 1.1225, "step": 20288 }, { "epoch": 0.02, "learning_rate": 4.99497216967094e-05, "loss": 1.2617, "step": 20289 }, { "epoch": 0.02, "learning_rate": 4.9949716717494285e-05, "loss": 0.9974, "step": 20290 }, { "epoch": 0.02, "learning_rate": 4.994971173803289e-05, "loss": 0.5454, "step": 20291 }, { "epoch": 0.02, "learning_rate": 4.9949706758325196e-05, "loss": 0.9964, "step": 20292 }, { "epoch": 0.02, "learning_rate": 4.994970177837121e-05, "loss": 0.7908, "step": 20293 }, { "epoch": 0.02, "learning_rate": 4.994969679817093e-05, "loss": 1.1258, "step": 20294 }, { "epoch": 0.02, "learning_rate": 4.994969181772436e-05, "loss": 1.3919, "step": 20295 }, { "epoch": 0.02, "learning_rate": 4.9949686837031496e-05, "loss": 1.4259, "step": 20296 }, { "epoch": 0.02, "learning_rate": 4.994968185609234e-05, "loss": 1.1144, "step": 20297 }, { "epoch": 0.02, "learning_rate": 4.994967687490689e-05, "loss": 0.9763, "step": 20298 }, { "epoch": 0.02, "learning_rate": 4.994967189347515e-05, "loss": 0.743, "step": 20299 }, { "epoch": 0.02, "learning_rate": 4.994966691179711e-05, "loss": 1.0759, "step": 20300 }, { "epoch": 0.02, "learning_rate": 4.9949661929872784e-05, "loss": 1.1583, "step": 20301 }, { "epoch": 0.02, "learning_rate": 4.994965694770216e-05, "loss": 1.1869, "step": 20302 }, { "epoch": 0.02, "learning_rate": 4.994965196528525e-05, "loss": 1.0493, "step": 20303 }, { "epoch": 0.02, "learning_rate": 4.9949646982622044e-05, "loss": 1.0201, "step": 20304 }, { "epoch": 0.02, "learning_rate": 4.9949641999712546e-05, "loss": 0.8446, "step": 20305 }, { "epoch": 0.02, "learning_rate": 4.994963701655676e-05, "loss": 1.0621, "step": 20306 }, { "epoch": 0.02, "learning_rate": 4.994963203315467e-05, "loss": 1.0534, "step": 20307 }, { "epoch": 0.02, "learning_rate": 4.99496270495063e-05, "loss": 1.2442, "step": 20308 }, { "epoch": 0.02, "learning_rate": 4.994962206561163e-05, "loss": 1.3193, "step": 20309 }, { "epoch": 0.02, "learning_rate": 4.994961708147068e-05, "loss": 1.0634, "step": 20310 }, { "epoch": 0.02, "learning_rate": 4.9949612097083426e-05, "loss": 1.3033, "step": 20311 }, { "epoch": 0.02, "learning_rate": 4.994960711244988e-05, "loss": 0.7795, "step": 20312 }, { "epoch": 0.02, "learning_rate": 4.994960212757005e-05, "loss": 1.0243, "step": 20313 }, { "epoch": 0.02, "learning_rate": 4.994959714244392e-05, "loss": 1.2474, "step": 20314 }, { "epoch": 0.02, "learning_rate": 4.99495921570715e-05, "loss": 1.7791, "step": 20315 }, { "epoch": 0.02, "learning_rate": 4.994958717145279e-05, "loss": 1.1594, "step": 20316 }, { "epoch": 0.02, "learning_rate": 4.994958218558778e-05, "loss": 1.1319, "step": 20317 }, { "epoch": 0.02, "learning_rate": 4.9949577199476494e-05, "loss": 1.2082, "step": 20318 }, { "epoch": 0.02, "learning_rate": 4.994957221311891e-05, "loss": 1.3547, "step": 20319 }, { "epoch": 0.02, "learning_rate": 4.9949567226515025e-05, "loss": 1.0767, "step": 20320 }, { "epoch": 0.02, "learning_rate": 4.994956223966486e-05, "loss": 0.8488, "step": 20321 }, { "epoch": 0.02, "learning_rate": 4.99495572525684e-05, "loss": 0.9438, "step": 20322 }, { "epoch": 0.02, "learning_rate": 4.994955226522565e-05, "loss": 1.1886, "step": 20323 }, { "epoch": 0.02, "learning_rate": 4.99495472776366e-05, "loss": 1.1505, "step": 20324 }, { "epoch": 0.02, "learning_rate": 4.9949542289801266e-05, "loss": 1.1204, "step": 20325 }, { "epoch": 0.02, "learning_rate": 4.994953730171964e-05, "loss": 1.0408, "step": 20326 }, { "epoch": 0.02, "learning_rate": 4.994953231339172e-05, "loss": 0.9235, "step": 20327 }, { "epoch": 0.02, "learning_rate": 4.994952732481751e-05, "loss": 1.131, "step": 20328 }, { "epoch": 0.02, "learning_rate": 4.994952233599701e-05, "loss": 0.9634, "step": 20329 }, { "epoch": 0.02, "learning_rate": 4.994951734693022e-05, "loss": 0.9429, "step": 20330 }, { "epoch": 0.02, "learning_rate": 4.9949512357617134e-05, "loss": 1.068, "step": 20331 }, { "epoch": 0.02, "learning_rate": 4.994950736805776e-05, "loss": 1.0612, "step": 20332 }, { "epoch": 0.02, "learning_rate": 4.99495023782521e-05, "loss": 0.7488, "step": 20333 }, { "epoch": 0.02, "learning_rate": 4.994949738820014e-05, "loss": 1.2353, "step": 20334 }, { "epoch": 0.02, "learning_rate": 4.99494923979019e-05, "loss": 1.0548, "step": 20335 }, { "epoch": 0.02, "learning_rate": 4.994948740735735e-05, "loss": 1.2048, "step": 20336 }, { "epoch": 0.02, "learning_rate": 4.994948241656653e-05, "loss": 1.2339, "step": 20337 }, { "epoch": 0.02, "learning_rate": 4.9949477425529414e-05, "loss": 0.9382, "step": 20338 }, { "epoch": 0.02, "learning_rate": 4.9949472434246006e-05, "loss": 1.261, "step": 20339 }, { "epoch": 0.02, "learning_rate": 4.99494674427163e-05, "loss": 1.3775, "step": 20340 }, { "epoch": 0.02, "learning_rate": 4.9949462450940307e-05, "loss": 1.1623, "step": 20341 }, { "epoch": 0.02, "learning_rate": 4.994945745891803e-05, "loss": 1.1393, "step": 20342 }, { "epoch": 0.02, "learning_rate": 4.994945246664945e-05, "loss": 1.0106, "step": 20343 }, { "epoch": 0.02, "learning_rate": 4.9949447474134595e-05, "loss": 1.083, "step": 20344 }, { "epoch": 0.02, "learning_rate": 4.994944248137344e-05, "loss": 0.4456, "step": 20345 }, { "epoch": 0.02, "learning_rate": 4.994943748836599e-05, "loss": 0.5227, "step": 20346 }, { "epoch": 0.02, "learning_rate": 4.994943249511226e-05, "loss": 1.5341, "step": 20347 }, { "epoch": 0.02, "learning_rate": 4.9949427501612234e-05, "loss": 1.0367, "step": 20348 }, { "epoch": 0.02, "learning_rate": 4.994942250786592e-05, "loss": 1.4389, "step": 20349 }, { "epoch": 0.02, "learning_rate": 4.994941751387331e-05, "loss": 1.0551, "step": 20350 }, { "epoch": 0.02, "learning_rate": 4.994941251963442e-05, "loss": 0.9989, "step": 20351 }, { "epoch": 0.02, "learning_rate": 4.994940752514924e-05, "loss": 1.0123, "step": 20352 }, { "epoch": 0.02, "learning_rate": 4.994940253041776e-05, "loss": 0.7292, "step": 20353 }, { "epoch": 0.02, "learning_rate": 4.994939753544e-05, "loss": 0.2389, "step": 20354 }, { "epoch": 0.02, "learning_rate": 4.994939254021594e-05, "loss": 1.0091, "step": 20355 }, { "epoch": 0.02, "learning_rate": 4.9949387544745595e-05, "loss": 1.1038, "step": 20356 }, { "epoch": 0.02, "learning_rate": 4.994938254902896e-05, "loss": 1.1386, "step": 20357 }, { "epoch": 0.02, "learning_rate": 4.994937755306603e-05, "loss": 0.7092, "step": 20358 }, { "epoch": 0.02, "learning_rate": 4.9949372556856824e-05, "loss": 1.0237, "step": 20359 }, { "epoch": 0.02, "learning_rate": 4.994936756040132e-05, "loss": 0.908, "step": 20360 }, { "epoch": 0.02, "learning_rate": 4.994936256369952e-05, "loss": 1.0243, "step": 20361 }, { "epoch": 0.02, "learning_rate": 4.994935756675144e-05, "loss": 1.0143, "step": 20362 }, { "epoch": 0.02, "learning_rate": 4.9949352569557065e-05, "loss": 0.9909, "step": 20363 }, { "epoch": 0.02, "learning_rate": 4.9949347572116405e-05, "loss": 1.0071, "step": 20364 }, { "epoch": 0.02, "learning_rate": 4.9949342574429455e-05, "loss": 1.1139, "step": 20365 }, { "epoch": 0.02, "learning_rate": 4.9949337576496215e-05, "loss": 0.8215, "step": 20366 }, { "epoch": 0.02, "learning_rate": 4.9949332578316685e-05, "loss": 0.9461, "step": 20367 }, { "epoch": 0.02, "learning_rate": 4.994932757989086e-05, "loss": 1.1085, "step": 20368 }, { "epoch": 0.02, "learning_rate": 4.9949322581218755e-05, "loss": 0.9859, "step": 20369 }, { "epoch": 0.02, "learning_rate": 4.994931758230036e-05, "loss": 1.2495, "step": 20370 }, { "epoch": 0.02, "learning_rate": 4.9949312583135664e-05, "loss": 1.1785, "step": 20371 }, { "epoch": 0.02, "learning_rate": 4.994930758372469e-05, "loss": 1.0522, "step": 20372 }, { "epoch": 0.02, "learning_rate": 4.994930258406743e-05, "loss": 1.1065, "step": 20373 }, { "epoch": 0.02, "learning_rate": 4.9949297584163866e-05, "loss": 0.8667, "step": 20374 }, { "epoch": 0.02, "learning_rate": 4.994929258401403e-05, "loss": 0.9229, "step": 20375 }, { "epoch": 0.02, "learning_rate": 4.9949287583617895e-05, "loss": 1.069, "step": 20376 }, { "epoch": 0.02, "learning_rate": 4.994928258297548e-05, "loss": 1.2157, "step": 20377 }, { "epoch": 0.02, "learning_rate": 4.9949277582086764e-05, "loss": 1.5214, "step": 20378 }, { "epoch": 0.02, "learning_rate": 4.9949272580951767e-05, "loss": 1.3871, "step": 20379 }, { "epoch": 0.02, "learning_rate": 4.994926757957048e-05, "loss": 1.3439, "step": 20380 }, { "epoch": 0.02, "learning_rate": 4.99492625779429e-05, "loss": 1.1913, "step": 20381 }, { "epoch": 0.02, "learning_rate": 4.994925757606904e-05, "loss": 1.139, "step": 20382 }, { "epoch": 0.02, "learning_rate": 4.994925257394888e-05, "loss": 1.1586, "step": 20383 }, { "epoch": 0.02, "learning_rate": 4.994924757158244e-05, "loss": 1.0063, "step": 20384 }, { "epoch": 0.02, "learning_rate": 4.994924256896971e-05, "loss": 1.0054, "step": 20385 }, { "epoch": 0.02, "learning_rate": 4.9949237566110697e-05, "loss": 0.9568, "step": 20386 }, { "epoch": 0.02, "learning_rate": 4.9949232563005385e-05, "loss": 0.7738, "step": 20387 }, { "epoch": 0.02, "learning_rate": 4.994922755965379e-05, "loss": 0.801, "step": 20388 }, { "epoch": 0.02, "learning_rate": 4.9949222556055906e-05, "loss": 0.9423, "step": 20389 }, { "epoch": 0.02, "learning_rate": 4.994921755221174e-05, "loss": 1.141, "step": 20390 }, { "epoch": 0.02, "learning_rate": 4.9949212548121274e-05, "loss": 0.4064, "step": 20391 }, { "epoch": 0.02, "learning_rate": 4.9949207543784526e-05, "loss": 0.2079, "step": 20392 }, { "epoch": 0.02, "learning_rate": 4.994920253920149e-05, "loss": 1.0039, "step": 20393 }, { "epoch": 0.02, "learning_rate": 4.994919753437217e-05, "loss": 0.8507, "step": 20394 }, { "epoch": 0.02, "learning_rate": 4.9949192529296554e-05, "loss": 1.1805, "step": 20395 }, { "epoch": 0.02, "learning_rate": 4.994918752397465e-05, "loss": 1.0034, "step": 20396 }, { "epoch": 0.02, "learning_rate": 4.994918251840647e-05, "loss": 1.0121, "step": 20397 }, { "epoch": 0.02, "learning_rate": 4.994917751259199e-05, "loss": 1.1833, "step": 20398 }, { "epoch": 0.02, "learning_rate": 4.994917250653123e-05, "loss": 1.2649, "step": 20399 }, { "epoch": 0.02, "learning_rate": 4.994916750022418e-05, "loss": 1.0091, "step": 20400 }, { "epoch": 0.02, "learning_rate": 4.994916249367084e-05, "loss": 0.8859, "step": 20401 }, { "epoch": 0.02, "learning_rate": 4.994915748687121e-05, "loss": 1.243, "step": 20402 }, { "epoch": 0.02, "learning_rate": 4.99491524798253e-05, "loss": 1.1594, "step": 20403 }, { "epoch": 0.02, "learning_rate": 4.99491474725331e-05, "loss": 1.0349, "step": 20404 }, { "epoch": 0.02, "learning_rate": 4.994914246499461e-05, "loss": 1.0105, "step": 20405 }, { "epoch": 0.02, "learning_rate": 4.9949137457209825e-05, "loss": 1.2083, "step": 20406 }, { "epoch": 0.02, "learning_rate": 4.9949132449178773e-05, "loss": 1.1124, "step": 20407 }, { "epoch": 0.02, "learning_rate": 4.994912744090142e-05, "loss": 1.094, "step": 20408 }, { "epoch": 0.02, "learning_rate": 4.994912243237778e-05, "loss": 1.0577, "step": 20409 }, { "epoch": 0.02, "learning_rate": 4.994911742360785e-05, "loss": 1.2048, "step": 20410 }, { "epoch": 0.02, "learning_rate": 4.994911241459164e-05, "loss": 1.0037, "step": 20411 }, { "epoch": 0.02, "learning_rate": 4.994910740532914e-05, "loss": 1.5609, "step": 20412 }, { "epoch": 0.02, "learning_rate": 4.994910239582035e-05, "loss": 1.313, "step": 20413 }, { "epoch": 0.02, "learning_rate": 4.994909738606528e-05, "loss": 1.2168, "step": 20414 }, { "epoch": 0.02, "learning_rate": 4.9949092376063917e-05, "loss": 1.0983, "step": 20415 }, { "epoch": 0.02, "learning_rate": 4.9949087365816274e-05, "loss": 1.0738, "step": 20416 }, { "epoch": 0.02, "learning_rate": 4.9949082355322336e-05, "loss": 1.2525, "step": 20417 }, { "epoch": 0.02, "learning_rate": 4.994907734458211e-05, "loss": 1.1444, "step": 20418 }, { "epoch": 0.02, "learning_rate": 4.99490723335956e-05, "loss": 0.984, "step": 20419 }, { "epoch": 0.02, "learning_rate": 4.9949067322362805e-05, "loss": 0.7995, "step": 20420 }, { "epoch": 0.02, "learning_rate": 4.9949062310883726e-05, "loss": 0.8166, "step": 20421 }, { "epoch": 0.02, "learning_rate": 4.994905729915836e-05, "loss": 1.0289, "step": 20422 }, { "epoch": 0.02, "learning_rate": 4.9949052287186705e-05, "loss": 0.7589, "step": 20423 }, { "epoch": 0.02, "learning_rate": 4.994904727496876e-05, "loss": 1.2308, "step": 20424 }, { "epoch": 0.02, "learning_rate": 4.994904226250453e-05, "loss": 1.1788, "step": 20425 }, { "epoch": 0.02, "learning_rate": 4.9949037249794014e-05, "loss": 1.2371, "step": 20426 }, { "epoch": 0.02, "learning_rate": 4.9949032236837215e-05, "loss": 0.9588, "step": 20427 }, { "epoch": 0.02, "learning_rate": 4.994902722363413e-05, "loss": 1.2215, "step": 20428 }, { "epoch": 0.02, "learning_rate": 4.994902221018476e-05, "loss": 0.9776, "step": 20429 }, { "epoch": 0.02, "learning_rate": 4.99490171964891e-05, "loss": 1.0478, "step": 20430 }, { "epoch": 0.02, "learning_rate": 4.994901218254715e-05, "loss": 0.6612, "step": 20431 }, { "epoch": 0.02, "learning_rate": 4.9949007168358916e-05, "loss": 1.0912, "step": 20432 }, { "epoch": 0.02, "learning_rate": 4.9949002153924404e-05, "loss": 1.0184, "step": 20433 }, { "epoch": 0.02, "learning_rate": 4.9948997139243595e-05, "loss": 0.7621, "step": 20434 }, { "epoch": 0.02, "learning_rate": 4.994899212431651e-05, "loss": 0.9941, "step": 20435 }, { "epoch": 0.02, "learning_rate": 4.994898710914313e-05, "loss": 1.4804, "step": 20436 }, { "epoch": 0.02, "learning_rate": 4.9948982093723474e-05, "loss": 1.0024, "step": 20437 }, { "epoch": 0.02, "learning_rate": 4.9948977078057524e-05, "loss": 1.078, "step": 20438 }, { "epoch": 0.02, "learning_rate": 4.9948972062145285e-05, "loss": 1.07, "step": 20439 }, { "epoch": 0.02, "learning_rate": 4.994896704598677e-05, "loss": 1.1107, "step": 20440 }, { "epoch": 0.02, "learning_rate": 4.994896202958196e-05, "loss": 1.5577, "step": 20441 }, { "epoch": 0.02, "learning_rate": 4.9948957012930874e-05, "loss": 1.255, "step": 20442 }, { "epoch": 0.02, "learning_rate": 4.99489519960335e-05, "loss": 0.9073, "step": 20443 }, { "epoch": 0.02, "learning_rate": 4.994894697888983e-05, "loss": 1.0193, "step": 20444 }, { "epoch": 0.02, "learning_rate": 4.9948941961499886e-05, "loss": 1.0198, "step": 20445 }, { "epoch": 0.02, "learning_rate": 4.994893694386366e-05, "loss": 1.1247, "step": 20446 }, { "epoch": 0.02, "learning_rate": 4.994893192598114e-05, "loss": 0.8786, "step": 20447 }, { "epoch": 0.02, "learning_rate": 4.9948926907852335e-05, "loss": 0.8272, "step": 20448 }, { "epoch": 0.02, "learning_rate": 4.994892188947724e-05, "loss": 0.8434, "step": 20449 }, { "epoch": 0.02, "learning_rate": 4.994891687085587e-05, "loss": 0.4277, "step": 20450 }, { "epoch": 0.02, "learning_rate": 4.9948911851988214e-05, "loss": 0.3038, "step": 20451 }, { "epoch": 0.02, "learning_rate": 4.9948906832874265e-05, "loss": 0.7202, "step": 20452 }, { "epoch": 0.02, "learning_rate": 4.994890181351404e-05, "loss": 1.1243, "step": 20453 }, { "epoch": 0.02, "learning_rate": 4.994889679390752e-05, "loss": 1.0209, "step": 20454 }, { "epoch": 0.02, "learning_rate": 4.9948891774054724e-05, "loss": 1.2765, "step": 20455 }, { "epoch": 0.02, "learning_rate": 4.9948886753955635e-05, "loss": 1.1735, "step": 20456 }, { "epoch": 0.02, "learning_rate": 4.994888173361027e-05, "loss": 1.4854, "step": 20457 }, { "epoch": 0.02, "learning_rate": 4.994887671301862e-05, "loss": 0.9207, "step": 20458 }, { "epoch": 0.02, "learning_rate": 4.9948871692180674e-05, "loss": 0.8616, "step": 20459 }, { "epoch": 0.02, "learning_rate": 4.994886667109645e-05, "loss": 1.0523, "step": 20460 }, { "epoch": 0.02, "learning_rate": 4.9948861649765945e-05, "loss": 1.0666, "step": 20461 }, { "epoch": 0.02, "learning_rate": 4.994885662818915e-05, "loss": 1.4422, "step": 20462 }, { "epoch": 0.02, "learning_rate": 4.994885160636608e-05, "loss": 4.3306, "step": 20463 }, { "epoch": 0.02, "learning_rate": 4.994884658429671e-05, "loss": 1.1483, "step": 20464 }, { "epoch": 0.02, "learning_rate": 4.994884156198106e-05, "loss": 1.2823, "step": 20465 }, { "epoch": 0.02, "learning_rate": 4.994883653941913e-05, "loss": 1.417, "step": 20466 }, { "epoch": 0.02, "learning_rate": 4.9948831516610914e-05, "loss": 1.1768, "step": 20467 }, { "epoch": 0.02, "learning_rate": 4.994882649355642e-05, "loss": 1.0893, "step": 20468 }, { "epoch": 0.02, "learning_rate": 4.994882147025563e-05, "loss": 1.1775, "step": 20469 }, { "epoch": 0.02, "learning_rate": 4.994881644670857e-05, "loss": 1.2037, "step": 20470 }, { "epoch": 0.02, "learning_rate": 4.994881142291521e-05, "loss": 1.2072, "step": 20471 }, { "epoch": 0.02, "learning_rate": 4.9948806398875575e-05, "loss": 1.1105, "step": 20472 }, { "epoch": 0.02, "learning_rate": 4.9948801374589656e-05, "loss": 1.5748, "step": 20473 }, { "epoch": 0.02, "learning_rate": 4.994879635005745e-05, "loss": 1.3014, "step": 20474 }, { "epoch": 0.02, "learning_rate": 4.994879132527896e-05, "loss": 1.1553, "step": 20475 }, { "epoch": 0.02, "learning_rate": 4.994878630025419e-05, "loss": 1.1796, "step": 20476 }, { "epoch": 0.02, "learning_rate": 4.9948781274983134e-05, "loss": 1.0419, "step": 20477 }, { "epoch": 0.02, "learning_rate": 4.994877624946579e-05, "loss": 1.2093, "step": 20478 }, { "epoch": 0.02, "learning_rate": 4.994877122370216e-05, "loss": 1.057, "step": 20479 }, { "epoch": 0.02, "learning_rate": 4.994876619769225e-05, "loss": 1.323, "step": 20480 }, { "epoch": 0.02, "learning_rate": 4.9948761171436065e-05, "loss": 1.0633, "step": 20481 }, { "epoch": 0.02, "learning_rate": 4.994875614493359e-05, "loss": 1.1679, "step": 20482 }, { "epoch": 0.02, "learning_rate": 4.994875111818483e-05, "loss": 0.966, "step": 20483 }, { "epoch": 0.02, "learning_rate": 4.994874609118979e-05, "loss": 1.1301, "step": 20484 }, { "epoch": 0.02, "learning_rate": 4.9948741063948464e-05, "loss": 0.9325, "step": 20485 }, { "epoch": 0.02, "learning_rate": 4.9948736036460855e-05, "loss": 1.4471, "step": 20486 }, { "epoch": 0.02, "learning_rate": 4.9948731008726957e-05, "loss": 1.1489, "step": 20487 }, { "epoch": 0.02, "learning_rate": 4.994872598074678e-05, "loss": 1.1858, "step": 20488 }, { "epoch": 0.02, "learning_rate": 4.994872095252032e-05, "loss": 1.2127, "step": 20489 }, { "epoch": 0.02, "learning_rate": 4.994871592404758e-05, "loss": 1.0272, "step": 20490 }, { "epoch": 0.02, "learning_rate": 4.9948710895328556e-05, "loss": 1.9511, "step": 20491 }, { "epoch": 0.02, "learning_rate": 4.994870586636324e-05, "loss": 1.0792, "step": 20492 }, { "epoch": 0.02, "learning_rate": 4.994870083715165e-05, "loss": 1.1621, "step": 20493 }, { "epoch": 0.02, "learning_rate": 4.9948695807693774e-05, "loss": 1.2359, "step": 20494 }, { "epoch": 0.02, "learning_rate": 4.9948690777989616e-05, "loss": 0.9695, "step": 20495 }, { "epoch": 0.02, "learning_rate": 4.994868574803917e-05, "loss": 1.0185, "step": 20496 }, { "epoch": 0.02, "learning_rate": 4.994868071784245e-05, "loss": 1.0531, "step": 20497 }, { "epoch": 0.02, "learning_rate": 4.994867568739944e-05, "loss": 1.0265, "step": 20498 }, { "epoch": 0.02, "learning_rate": 4.994867065671015e-05, "loss": 1.1798, "step": 20499 }, { "epoch": 0.02, "learning_rate": 4.994866562577458e-05, "loss": 1.0785, "step": 20500 }, { "epoch": 0.02, "eval_loss": 1.0215940475463867, "eval_runtime": 83.917, "eval_samples_per_second": 16.504, "eval_steps_per_second": 4.135, "step": 20500 }, { "epoch": 0.02, "learning_rate": 4.994866059459272e-05, "loss": 1.1798, "step": 20501 }, { "epoch": 0.02, "learning_rate": 4.994865556316458e-05, "loss": 1.4237, "step": 20502 }, { "epoch": 0.02, "learning_rate": 4.994865053149016e-05, "loss": 1.1493, "step": 20503 }, { "epoch": 0.02, "learning_rate": 4.9948645499569456e-05, "loss": 1.0907, "step": 20504 }, { "epoch": 0.02, "learning_rate": 4.994864046740247e-05, "loss": 0.9533, "step": 20505 }, { "epoch": 0.02, "learning_rate": 4.9948635434989203e-05, "loss": 1.5278, "step": 20506 }, { "epoch": 0.02, "learning_rate": 4.994863040232965e-05, "loss": 1.2102, "step": 20507 }, { "epoch": 0.02, "learning_rate": 4.994862536942382e-05, "loss": 1.0935, "step": 20508 }, { "epoch": 0.02, "learning_rate": 4.994862033627171e-05, "loss": 0.9425, "step": 20509 }, { "epoch": 0.02, "learning_rate": 4.9948615302873306e-05, "loss": 1.1779, "step": 20510 }, { "epoch": 0.02, "learning_rate": 4.994861026922862e-05, "loss": 1.0389, "step": 20511 }, { "epoch": 0.02, "learning_rate": 4.994860523533767e-05, "loss": 1.2344, "step": 20512 }, { "epoch": 0.02, "learning_rate": 4.994860020120042e-05, "loss": 1.1296, "step": 20513 }, { "epoch": 0.02, "learning_rate": 4.994859516681689e-05, "loss": 1.0571, "step": 20514 }, { "epoch": 0.02, "learning_rate": 4.994859013218709e-05, "loss": 0.9802, "step": 20515 }, { "epoch": 0.02, "learning_rate": 4.9948585097311e-05, "loss": 0.9266, "step": 20516 }, { "epoch": 0.02, "learning_rate": 4.994858006218862e-05, "loss": 0.9232, "step": 20517 }, { "epoch": 0.02, "learning_rate": 4.9948575026819966e-05, "loss": 0.6939, "step": 20518 }, { "epoch": 0.02, "learning_rate": 4.994856999120503e-05, "loss": 1.3959, "step": 20519 }, { "epoch": 0.02, "learning_rate": 4.9948564955343816e-05, "loss": 1.2354, "step": 20520 }, { "epoch": 0.02, "learning_rate": 4.994855991923631e-05, "loss": 1.1157, "step": 20521 }, { "epoch": 0.02, "learning_rate": 4.994855488288254e-05, "loss": 1.1328, "step": 20522 }, { "epoch": 0.02, "learning_rate": 4.9948549846282475e-05, "loss": 1.8559, "step": 20523 }, { "epoch": 0.02, "learning_rate": 4.994854480943613e-05, "loss": 0.9372, "step": 20524 }, { "epoch": 0.02, "learning_rate": 4.99485397723435e-05, "loss": 1.2147, "step": 20525 }, { "epoch": 0.02, "learning_rate": 4.994853473500459e-05, "loss": 1.1951, "step": 20526 }, { "epoch": 0.02, "learning_rate": 4.994852969741941e-05, "loss": 1.1203, "step": 20527 }, { "epoch": 0.02, "learning_rate": 4.9948524659587934e-05, "loss": 1.246, "step": 20528 }, { "epoch": 0.02, "learning_rate": 4.9948519621510184e-05, "loss": 1.1379, "step": 20529 }, { "epoch": 0.02, "learning_rate": 4.994851458318615e-05, "loss": 0.9786, "step": 20530 }, { "epoch": 0.02, "learning_rate": 4.9948509544615835e-05, "loss": 1.0192, "step": 20531 }, { "epoch": 0.02, "learning_rate": 4.994850450579924e-05, "loss": 0.9764, "step": 20532 }, { "epoch": 0.02, "learning_rate": 4.9948499466736366e-05, "loss": 1.2278, "step": 20533 }, { "epoch": 0.02, "learning_rate": 4.9948494427427206e-05, "loss": 1.1132, "step": 20534 }, { "epoch": 0.02, "learning_rate": 4.994848938787176e-05, "loss": 1.045, "step": 20535 }, { "epoch": 0.02, "learning_rate": 4.9948484348070044e-05, "loss": 1.0171, "step": 20536 }, { "epoch": 0.02, "learning_rate": 4.994847930802205e-05, "loss": 1.2686, "step": 20537 }, { "epoch": 0.02, "learning_rate": 4.994847426772777e-05, "loss": 0.9343, "step": 20538 }, { "epoch": 0.02, "learning_rate": 4.9948469227187206e-05, "loss": 1.1846, "step": 20539 }, { "epoch": 0.02, "learning_rate": 4.994846418640036e-05, "loss": 0.8699, "step": 20540 }, { "epoch": 0.02, "learning_rate": 4.994845914536724e-05, "loss": 1.2019, "step": 20541 }, { "epoch": 0.02, "learning_rate": 4.994845410408784e-05, "loss": 1.214, "step": 20542 }, { "epoch": 0.02, "learning_rate": 4.994844906256215e-05, "loss": 1.165, "step": 20543 }, { "epoch": 0.02, "learning_rate": 4.9948444020790185e-05, "loss": 0.9873, "step": 20544 }, { "epoch": 0.02, "learning_rate": 4.9948438978771936e-05, "loss": 1.2067, "step": 20545 }, { "epoch": 0.02, "learning_rate": 4.994843393650741e-05, "loss": 1.1716, "step": 20546 }, { "epoch": 0.02, "learning_rate": 4.99484288939966e-05, "loss": 1.0789, "step": 20547 }, { "epoch": 0.02, "learning_rate": 4.994842385123952e-05, "loss": 1.0884, "step": 20548 }, { "epoch": 0.02, "learning_rate": 4.9948418808236155e-05, "loss": 0.8562, "step": 20549 }, { "epoch": 0.02, "learning_rate": 4.9948413764986503e-05, "loss": 1.1415, "step": 20550 }, { "epoch": 0.02, "learning_rate": 4.9948408721490575e-05, "loss": 1.2349, "step": 20551 }, { "epoch": 0.02, "learning_rate": 4.994840367774837e-05, "loss": 1.1789, "step": 20552 }, { "epoch": 0.02, "learning_rate": 4.9948398633759876e-05, "loss": 0.9627, "step": 20553 }, { "epoch": 0.02, "learning_rate": 4.994839358952511e-05, "loss": 0.2376, "step": 20554 }, { "epoch": 0.02, "learning_rate": 4.994838854504406e-05, "loss": 0.0839, "step": 20555 }, { "epoch": 0.02, "learning_rate": 4.994838350031673e-05, "loss": 0.0495, "step": 20556 }, { "epoch": 0.02, "learning_rate": 4.9948378455343124e-05, "loss": 0.0677, "step": 20557 }, { "epoch": 0.02, "learning_rate": 4.994837341012324e-05, "loss": 0.075, "step": 20558 }, { "epoch": 0.02, "learning_rate": 4.994836836465707e-05, "loss": 0.7124, "step": 20559 }, { "epoch": 0.02, "learning_rate": 4.994836331894462e-05, "loss": 0.945, "step": 20560 }, { "epoch": 0.02, "learning_rate": 4.9948358272985894e-05, "loss": 0.8929, "step": 20561 }, { "epoch": 0.02, "learning_rate": 4.9948353226780884e-05, "loss": 0.94, "step": 20562 }, { "epoch": 0.02, "learning_rate": 4.9948348180329596e-05, "loss": 1.283, "step": 20563 }, { "epoch": 0.02, "learning_rate": 4.994834313363204e-05, "loss": 0.8967, "step": 20564 }, { "epoch": 0.02, "learning_rate": 4.9948338086688186e-05, "loss": 0.839, "step": 20565 }, { "epoch": 0.02, "learning_rate": 4.994833303949806e-05, "loss": 0.9629, "step": 20566 }, { "epoch": 0.02, "learning_rate": 4.9948327992061655e-05, "loss": 0.7622, "step": 20567 }, { "epoch": 0.02, "learning_rate": 4.994832294437897e-05, "loss": 0.5821, "step": 20568 }, { "epoch": 0.02, "learning_rate": 4.9948317896450005e-05, "loss": 1.0234, "step": 20569 }, { "epoch": 0.02, "learning_rate": 4.994831284827476e-05, "loss": 1.3592, "step": 20570 }, { "epoch": 0.02, "learning_rate": 4.994830779985324e-05, "loss": 1.2687, "step": 20571 }, { "epoch": 0.02, "learning_rate": 4.994830275118544e-05, "loss": 1.1834, "step": 20572 }, { "epoch": 0.02, "learning_rate": 4.994829770227136e-05, "loss": 0.8404, "step": 20573 }, { "epoch": 0.02, "learning_rate": 4.9948292653111004e-05, "loss": 0.9492, "step": 20574 }, { "epoch": 0.02, "learning_rate": 4.9948287603704365e-05, "loss": 0.9871, "step": 20575 }, { "epoch": 0.02, "learning_rate": 4.994828255405144e-05, "loss": 0.8609, "step": 20576 }, { "epoch": 0.02, "learning_rate": 4.994827750415225e-05, "loss": 1.1345, "step": 20577 }, { "epoch": 0.02, "learning_rate": 4.9948272454006774e-05, "loss": 1.1334, "step": 20578 }, { "epoch": 0.02, "learning_rate": 4.994826740361502e-05, "loss": 1.3962, "step": 20579 }, { "epoch": 0.02, "learning_rate": 4.994826235297698e-05, "loss": 1.6432, "step": 20580 }, { "epoch": 0.02, "learning_rate": 4.9948257302092674e-05, "loss": 1.2254, "step": 20581 }, { "epoch": 0.02, "learning_rate": 4.9948252250962086e-05, "loss": 0.9394, "step": 20582 }, { "epoch": 0.02, "learning_rate": 4.994824719958522e-05, "loss": 1.1011, "step": 20583 }, { "epoch": 0.02, "learning_rate": 4.9948242147962065e-05, "loss": 1.0529, "step": 20584 }, { "epoch": 0.02, "learning_rate": 4.994823709609264e-05, "loss": 1.1848, "step": 20585 }, { "epoch": 0.02, "learning_rate": 4.994823204397694e-05, "loss": 1.1384, "step": 20586 }, { "epoch": 0.02, "learning_rate": 4.9948226991614954e-05, "loss": 1.1495, "step": 20587 }, { "epoch": 0.02, "learning_rate": 4.9948221939006686e-05, "loss": 0.9372, "step": 20588 }, { "epoch": 0.02, "learning_rate": 4.9948216886152155e-05, "loss": 1.1513, "step": 20589 }, { "epoch": 0.02, "learning_rate": 4.9948211833051334e-05, "loss": 0.8972, "step": 20590 }, { "epoch": 0.02, "learning_rate": 4.994820677970424e-05, "loss": 1.0895, "step": 20591 }, { "epoch": 0.02, "learning_rate": 4.994820172611086e-05, "loss": 1.1794, "step": 20592 }, { "epoch": 0.02, "learning_rate": 4.9948196672271205e-05, "loss": 1.3706, "step": 20593 }, { "epoch": 0.02, "learning_rate": 4.994819161818528e-05, "loss": 1.1998, "step": 20594 }, { "epoch": 0.02, "learning_rate": 4.9948186563853074e-05, "loss": 1.2926, "step": 20595 }, { "epoch": 0.02, "learning_rate": 4.994818150927458e-05, "loss": 1.0652, "step": 20596 }, { "epoch": 0.02, "learning_rate": 4.994817645444982e-05, "loss": 1.0558, "step": 20597 }, { "epoch": 0.02, "learning_rate": 4.994817139937878e-05, "loss": 0.9876, "step": 20598 }, { "epoch": 0.02, "learning_rate": 4.9948166344061454e-05, "loss": 1.1109, "step": 20599 }, { "epoch": 0.02, "learning_rate": 4.994816128849786e-05, "loss": 1.2917, "step": 20600 }, { "epoch": 0.02, "learning_rate": 4.9948156232687984e-05, "loss": 0.9461, "step": 20601 }, { "epoch": 0.02, "learning_rate": 4.994815117663183e-05, "loss": 0.7874, "step": 20602 }, { "epoch": 0.02, "learning_rate": 4.9948146120329395e-05, "loss": 1.0423, "step": 20603 }, { "epoch": 0.02, "learning_rate": 4.994814106378069e-05, "loss": 1.3804, "step": 20604 }, { "epoch": 0.02, "learning_rate": 4.99481360069857e-05, "loss": 1.2473, "step": 20605 }, { "epoch": 0.02, "learning_rate": 4.994813094994444e-05, "loss": 1.3151, "step": 20606 }, { "epoch": 0.02, "learning_rate": 4.994812589265689e-05, "loss": 0.7192, "step": 20607 }, { "epoch": 0.02, "learning_rate": 4.994812083512308e-05, "loss": 0.3976, "step": 20608 }, { "epoch": 0.02, "learning_rate": 4.9948115777342984e-05, "loss": 0.2892, "step": 20609 }, { "epoch": 0.02, "learning_rate": 4.994811071931661e-05, "loss": 0.2133, "step": 20610 }, { "epoch": 0.02, "learning_rate": 4.994810566104396e-05, "loss": 0.3406, "step": 20611 }, { "epoch": 0.02, "learning_rate": 4.9948100602525026e-05, "loss": 0.3587, "step": 20612 }, { "epoch": 0.02, "learning_rate": 4.9948095543759824e-05, "loss": 0.8555, "step": 20613 }, { "epoch": 0.02, "learning_rate": 4.9948090484748346e-05, "loss": 0.7441, "step": 20614 }, { "epoch": 0.02, "learning_rate": 4.9948085425490585e-05, "loss": 1.0809, "step": 20615 }, { "epoch": 0.02, "learning_rate": 4.994808036598655e-05, "loss": 1.1309, "step": 20616 }, { "epoch": 0.02, "learning_rate": 4.994807530623623e-05, "loss": 1.0297, "step": 20617 }, { "epoch": 0.02, "learning_rate": 4.994807024623965e-05, "loss": 1.2328, "step": 20618 }, { "epoch": 0.02, "learning_rate": 4.994806518599678e-05, "loss": 0.9495, "step": 20619 }, { "epoch": 0.02, "learning_rate": 4.994806012550764e-05, "loss": 0.8014, "step": 20620 }, { "epoch": 0.02, "learning_rate": 4.994805506477222e-05, "loss": 1.0521, "step": 20621 }, { "epoch": 0.02, "learning_rate": 4.994805000379052e-05, "loss": 1.2195, "step": 20622 }, { "epoch": 0.02, "learning_rate": 4.994804494256255e-05, "loss": 1.0379, "step": 20623 }, { "epoch": 0.02, "learning_rate": 4.994803988108829e-05, "loss": 0.5961, "step": 20624 }, { "epoch": 0.02, "learning_rate": 4.994803481936777e-05, "loss": 1.0857, "step": 20625 }, { "epoch": 0.02, "learning_rate": 4.994802975740097e-05, "loss": 0.9888, "step": 20626 }, { "epoch": 0.02, "learning_rate": 4.994802469518789e-05, "loss": 1.0407, "step": 20627 }, { "epoch": 0.02, "learning_rate": 4.994801963272853e-05, "loss": 1.2679, "step": 20628 }, { "epoch": 0.02, "learning_rate": 4.99480145700229e-05, "loss": 1.2779, "step": 20629 }, { "epoch": 0.02, "learning_rate": 4.994800950707099e-05, "loss": 1.1004, "step": 20630 }, { "epoch": 0.02, "learning_rate": 4.994800444387281e-05, "loss": 1.0953, "step": 20631 }, { "epoch": 0.02, "learning_rate": 4.994799938042834e-05, "loss": 1.261, "step": 20632 }, { "epoch": 0.02, "learning_rate": 4.994799431673761e-05, "loss": 1.0813, "step": 20633 }, { "epoch": 0.02, "learning_rate": 4.9947989252800595e-05, "loss": 0.4703, "step": 20634 }, { "epoch": 0.02, "learning_rate": 4.99479841886173e-05, "loss": 0.4115, "step": 20635 }, { "epoch": 0.02, "learning_rate": 4.9947979124187735e-05, "loss": 1.1072, "step": 20636 }, { "epoch": 0.02, "learning_rate": 4.99479740595119e-05, "loss": 1.0784, "step": 20637 }, { "epoch": 0.02, "learning_rate": 4.994796899458978e-05, "loss": 0.7266, "step": 20638 }, { "epoch": 0.02, "learning_rate": 4.994796392942139e-05, "loss": 0.6355, "step": 20639 }, { "epoch": 0.02, "learning_rate": 4.9947958864006724e-05, "loss": 1.3791, "step": 20640 }, { "epoch": 0.02, "learning_rate": 4.994795379834577e-05, "loss": 1.2974, "step": 20641 }, { "epoch": 0.02, "learning_rate": 4.994794873243855e-05, "loss": 0.9291, "step": 20642 }, { "epoch": 0.02, "learning_rate": 4.9947943666285055e-05, "loss": 1.1866, "step": 20643 }, { "epoch": 0.02, "learning_rate": 4.994793859988529e-05, "loss": 0.9061, "step": 20644 }, { "epoch": 0.02, "learning_rate": 4.9947933533239245e-05, "loss": 1.0333, "step": 20645 }, { "epoch": 0.02, "learning_rate": 4.994792846634692e-05, "loss": 1.0324, "step": 20646 }, { "epoch": 0.02, "learning_rate": 4.994792339920832e-05, "loss": 0.4977, "step": 20647 }, { "epoch": 0.02, "learning_rate": 4.994791833182345e-05, "loss": 0.9781, "step": 20648 }, { "epoch": 0.02, "learning_rate": 4.9947913264192306e-05, "loss": 1.2779, "step": 20649 }, { "epoch": 0.02, "learning_rate": 4.994790819631487e-05, "loss": 0.9247, "step": 20650 }, { "epoch": 0.02, "learning_rate": 4.994790312819118e-05, "loss": 1.1156, "step": 20651 }, { "epoch": 0.02, "learning_rate": 4.9947898059821205e-05, "loss": 1.1309, "step": 20652 }, { "epoch": 0.02, "learning_rate": 4.9947892991204956e-05, "loss": 1.1929, "step": 20653 }, { "epoch": 0.02, "learning_rate": 4.994788792234243e-05, "loss": 1.0149, "step": 20654 }, { "epoch": 0.02, "learning_rate": 4.994788285323363e-05, "loss": 0.7167, "step": 20655 }, { "epoch": 0.02, "learning_rate": 4.994787778387856e-05, "loss": 0.8417, "step": 20656 }, { "epoch": 0.02, "learning_rate": 4.994787271427721e-05, "loss": 1.0342, "step": 20657 }, { "epoch": 0.02, "learning_rate": 4.994786764442958e-05, "loss": 1.3453, "step": 20658 }, { "epoch": 0.02, "learning_rate": 4.9947862574335677e-05, "loss": 1.0122, "step": 20659 }, { "epoch": 0.02, "learning_rate": 4.994785750399551e-05, "loss": 1.1082, "step": 20660 }, { "epoch": 0.02, "learning_rate": 4.994785243340906e-05, "loss": 1.1208, "step": 20661 }, { "epoch": 0.02, "learning_rate": 4.994784736257634e-05, "loss": 1.1874, "step": 20662 }, { "epoch": 0.02, "learning_rate": 4.994784229149734e-05, "loss": 1.1206, "step": 20663 }, { "epoch": 0.02, "learning_rate": 4.994783722017207e-05, "loss": 0.9856, "step": 20664 }, { "epoch": 0.02, "learning_rate": 4.994783214860052e-05, "loss": 0.9931, "step": 20665 }, { "epoch": 0.02, "learning_rate": 4.99478270767827e-05, "loss": 1.3226, "step": 20666 }, { "epoch": 0.02, "learning_rate": 4.99478220047186e-05, "loss": 1.5303, "step": 20667 }, { "epoch": 0.02, "learning_rate": 4.994781693240823e-05, "loss": 1.0815, "step": 20668 }, { "epoch": 0.02, "learning_rate": 4.994781185985159e-05, "loss": 0.9836, "step": 20669 }, { "epoch": 0.02, "learning_rate": 4.994780678704867e-05, "loss": 1.2532, "step": 20670 }, { "epoch": 0.02, "learning_rate": 4.9947801713999476e-05, "loss": 1.0708, "step": 20671 }, { "epoch": 0.02, "learning_rate": 4.994779664070401e-05, "loss": 0.9388, "step": 20672 }, { "epoch": 0.02, "learning_rate": 4.994779156716227e-05, "loss": 0.9802, "step": 20673 }, { "epoch": 0.02, "learning_rate": 4.994778649337425e-05, "loss": 1.5807, "step": 20674 }, { "epoch": 0.02, "learning_rate": 4.994778141933996e-05, "loss": 1.1361, "step": 20675 }, { "epoch": 0.02, "learning_rate": 4.9947776345059394e-05, "loss": 0.9857, "step": 20676 }, { "epoch": 0.02, "learning_rate": 4.994777127053256e-05, "loss": 1.2357, "step": 20677 }, { "epoch": 0.02, "learning_rate": 4.994776619575945e-05, "loss": 1.1093, "step": 20678 }, { "epoch": 0.02, "learning_rate": 4.9947761120740064e-05, "loss": 1.1785, "step": 20679 }, { "epoch": 0.02, "learning_rate": 4.99477560454744e-05, "loss": 1.1368, "step": 20680 }, { "epoch": 0.02, "learning_rate": 4.994775096996247e-05, "loss": 1.0208, "step": 20681 }, { "epoch": 0.02, "learning_rate": 4.9947745894204266e-05, "loss": 1.1131, "step": 20682 }, { "epoch": 0.02, "learning_rate": 4.994774081819978e-05, "loss": 1.0333, "step": 20683 }, { "epoch": 0.02, "learning_rate": 4.994773574194903e-05, "loss": 0.9327, "step": 20684 }, { "epoch": 0.02, "learning_rate": 4.9947730665452006e-05, "loss": 1.1032, "step": 20685 }, { "epoch": 0.02, "learning_rate": 4.9947725588708706e-05, "loss": 1.271, "step": 20686 }, { "epoch": 0.02, "learning_rate": 4.994772051171913e-05, "loss": 1.1184, "step": 20687 }, { "epoch": 0.02, "learning_rate": 4.994771543448328e-05, "loss": 0.5272, "step": 20688 }, { "epoch": 0.02, "learning_rate": 4.994771035700116e-05, "loss": 1.1593, "step": 20689 }, { "epoch": 0.02, "learning_rate": 4.994770527927276e-05, "loss": 1.1369, "step": 20690 }, { "epoch": 0.02, "learning_rate": 4.9947700201298094e-05, "loss": 1.0505, "step": 20691 }, { "epoch": 0.02, "learning_rate": 4.9947695123077156e-05, "loss": 0.7023, "step": 20692 }, { "epoch": 0.02, "learning_rate": 4.994769004460994e-05, "loss": 0.9648, "step": 20693 }, { "epoch": 0.02, "learning_rate": 4.9947684965896456e-05, "loss": 1.2624, "step": 20694 }, { "epoch": 0.02, "learning_rate": 4.9947679886936695e-05, "loss": 1.0662, "step": 20695 }, { "epoch": 0.02, "learning_rate": 4.994767480773066e-05, "loss": 1.0596, "step": 20696 }, { "epoch": 0.02, "learning_rate": 4.994766972827836e-05, "loss": 0.838, "step": 20697 }, { "epoch": 0.02, "learning_rate": 4.994766464857978e-05, "loss": 1.13, "step": 20698 }, { "epoch": 0.02, "learning_rate": 4.994765956863493e-05, "loss": 1.0402, "step": 20699 }, { "epoch": 0.02, "learning_rate": 4.994765448844381e-05, "loss": 1.1582, "step": 20700 }, { "epoch": 0.02, "learning_rate": 4.99476494080064e-05, "loss": 1.8826, "step": 20701 }, { "epoch": 0.02, "learning_rate": 4.994764432732274e-05, "loss": 1.3675, "step": 20702 }, { "epoch": 0.02, "learning_rate": 4.994763924639279e-05, "loss": 1.1388, "step": 20703 }, { "epoch": 0.02, "learning_rate": 4.9947634165216584e-05, "loss": 1.1243, "step": 20704 }, { "epoch": 0.02, "learning_rate": 4.994762908379409e-05, "loss": 1.1758, "step": 20705 }, { "epoch": 0.02, "learning_rate": 4.994762400212533e-05, "loss": 1.0824, "step": 20706 }, { "epoch": 0.02, "learning_rate": 4.9947618920210304e-05, "loss": 1.1573, "step": 20707 }, { "epoch": 0.02, "learning_rate": 4.9947613838048996e-05, "loss": 0.9419, "step": 20708 }, { "epoch": 0.02, "learning_rate": 4.994760875564142e-05, "loss": 1.3454, "step": 20709 }, { "epoch": 0.02, "learning_rate": 4.994760367298757e-05, "loss": 0.9501, "step": 20710 }, { "epoch": 0.02, "learning_rate": 4.9947598590087445e-05, "loss": 0.9463, "step": 20711 }, { "epoch": 0.02, "learning_rate": 4.994759350694106e-05, "loss": 1.1179, "step": 20712 }, { "epoch": 0.02, "learning_rate": 4.9947588423548386e-05, "loss": 0.9973, "step": 20713 }, { "epoch": 0.02, "learning_rate": 4.9947583339909446e-05, "loss": 1.3399, "step": 20714 }, { "epoch": 0.02, "learning_rate": 4.994757825602424e-05, "loss": 0.9823, "step": 20715 }, { "epoch": 0.02, "learning_rate": 4.9947573171892755e-05, "loss": 1.2456, "step": 20716 }, { "epoch": 0.02, "learning_rate": 4.9947568087515005e-05, "loss": 1.1046, "step": 20717 }, { "epoch": 0.02, "learning_rate": 4.994756300289098e-05, "loss": 0.9354, "step": 20718 }, { "epoch": 0.02, "learning_rate": 4.994755791802068e-05, "loss": 1.2016, "step": 20719 }, { "epoch": 0.02, "learning_rate": 4.994755283290411e-05, "loss": 1.1467, "step": 20720 }, { "epoch": 0.02, "learning_rate": 4.994754774754127e-05, "loss": 0.9806, "step": 20721 }, { "epoch": 0.02, "learning_rate": 4.9947542661932156e-05, "loss": 1.1871, "step": 20722 }, { "epoch": 0.02, "learning_rate": 4.9947537576076774e-05, "loss": 1.0602, "step": 20723 }, { "epoch": 0.02, "learning_rate": 4.9947532489975116e-05, "loss": 1.1738, "step": 20724 }, { "epoch": 0.02, "learning_rate": 4.9947527403627194e-05, "loss": 0.9801, "step": 20725 }, { "epoch": 0.02, "learning_rate": 4.994752231703299e-05, "loss": 0.9521, "step": 20726 }, { "epoch": 0.02, "learning_rate": 4.994751723019252e-05, "loss": 1.2318, "step": 20727 }, { "epoch": 0.02, "learning_rate": 4.994751214310578e-05, "loss": 1.1267, "step": 20728 }, { "epoch": 0.02, "learning_rate": 4.9947507055772764e-05, "loss": 1.092, "step": 20729 }, { "epoch": 0.02, "learning_rate": 4.994750196819348e-05, "loss": 1.1567, "step": 20730 }, { "epoch": 0.02, "learning_rate": 4.994749688036793e-05, "loss": 1.1191, "step": 20731 }, { "epoch": 0.02, "learning_rate": 4.99474917922961e-05, "loss": 1.0886, "step": 20732 }, { "epoch": 0.02, "learning_rate": 4.9947486703978e-05, "loss": 1.0951, "step": 20733 }, { "epoch": 0.02, "learning_rate": 4.994748161541363e-05, "loss": 1.0129, "step": 20734 }, { "epoch": 0.02, "learning_rate": 4.994747652660299e-05, "loss": 1.0814, "step": 20735 }, { "epoch": 0.02, "learning_rate": 4.994747143754608e-05, "loss": 1.2244, "step": 20736 }, { "epoch": 0.02, "learning_rate": 4.99474663482429e-05, "loss": 1.23, "step": 20737 }, { "epoch": 0.02, "learning_rate": 4.9947461258693446e-05, "loss": 1.2448, "step": 20738 }, { "epoch": 0.02, "learning_rate": 4.994745616889772e-05, "loss": 1.104, "step": 20739 }, { "epoch": 0.02, "learning_rate": 4.994745107885572e-05, "loss": 1.0084, "step": 20740 }, { "epoch": 0.02, "learning_rate": 4.994744598856746e-05, "loss": 1.0627, "step": 20741 }, { "epoch": 0.02, "learning_rate": 4.994744089803293e-05, "loss": 0.8839, "step": 20742 }, { "epoch": 0.02, "learning_rate": 4.994743580725212e-05, "loss": 0.6835, "step": 20743 }, { "epoch": 0.02, "learning_rate": 4.9947430716225044e-05, "loss": 1.103, "step": 20744 }, { "epoch": 0.02, "learning_rate": 4.9947425624951694e-05, "loss": 0.9938, "step": 20745 }, { "epoch": 0.02, "learning_rate": 4.9947420533432074e-05, "loss": 0.9369, "step": 20746 }, { "epoch": 0.02, "learning_rate": 4.994741544166619e-05, "loss": 0.9632, "step": 20747 }, { "epoch": 0.02, "learning_rate": 4.994741034965403e-05, "loss": 0.8797, "step": 20748 }, { "epoch": 0.02, "learning_rate": 4.9947405257395595e-05, "loss": 1.1272, "step": 20749 }, { "epoch": 0.02, "learning_rate": 4.99474001648909e-05, "loss": 1.2303, "step": 20750 }, { "epoch": 0.02, "learning_rate": 4.994739507213993e-05, "loss": 1.2516, "step": 20751 }, { "epoch": 0.02, "learning_rate": 4.994738997914269e-05, "loss": 1.2519, "step": 20752 }, { "epoch": 0.02, "learning_rate": 4.994738488589918e-05, "loss": 1.3436, "step": 20753 }, { "epoch": 0.02, "learning_rate": 4.9947379792409397e-05, "loss": 1.5545, "step": 20754 }, { "epoch": 0.02, "learning_rate": 4.994737469867335e-05, "loss": 0.9951, "step": 20755 }, { "epoch": 0.02, "learning_rate": 4.994736960469103e-05, "loss": 1.3711, "step": 20756 }, { "epoch": 0.02, "learning_rate": 4.9947364510462444e-05, "loss": 1.37, "step": 20757 }, { "epoch": 0.02, "learning_rate": 4.994735941598758e-05, "loss": 1.3354, "step": 20758 }, { "epoch": 0.02, "learning_rate": 4.994735432126645e-05, "loss": 1.0918, "step": 20759 }, { "epoch": 0.02, "learning_rate": 4.994734922629905e-05, "loss": 1.0659, "step": 20760 }, { "epoch": 0.02, "learning_rate": 4.9947344131085385e-05, "loss": 1.1695, "step": 20761 }, { "epoch": 0.02, "learning_rate": 4.994733903562544e-05, "loss": 1.1368, "step": 20762 }, { "epoch": 0.02, "learning_rate": 4.9947333939919236e-05, "loss": 0.9634, "step": 20763 }, { "epoch": 0.02, "learning_rate": 4.9947328843966754e-05, "loss": 1.0036, "step": 20764 }, { "epoch": 0.02, "learning_rate": 4.994732374776801e-05, "loss": 1.1566, "step": 20765 }, { "epoch": 0.02, "learning_rate": 4.9947318651322994e-05, "loss": 1.5064, "step": 20766 }, { "epoch": 0.02, "learning_rate": 4.99473135546317e-05, "loss": 1.0103, "step": 20767 }, { "epoch": 0.02, "learning_rate": 4.9947308457694155e-05, "loss": 1.0233, "step": 20768 }, { "epoch": 0.02, "learning_rate": 4.9947303360510324e-05, "loss": 1.0324, "step": 20769 }, { "epoch": 0.02, "learning_rate": 4.994729826308023e-05, "loss": 1.0067, "step": 20770 }, { "epoch": 0.02, "learning_rate": 4.994729316540387e-05, "loss": 0.9207, "step": 20771 }, { "epoch": 0.02, "learning_rate": 4.9947288067481234e-05, "loss": 1.109, "step": 20772 }, { "epoch": 0.02, "learning_rate": 4.994728296931234e-05, "loss": 1.1022, "step": 20773 }, { "epoch": 0.02, "learning_rate": 4.9947277870897164e-05, "loss": 0.9956, "step": 20774 }, { "epoch": 0.02, "learning_rate": 4.994727277223572e-05, "loss": 1.0054, "step": 20775 }, { "epoch": 0.02, "learning_rate": 4.9947267673328016e-05, "loss": 1.0118, "step": 20776 }, { "epoch": 0.02, "learning_rate": 4.9947262574174034e-05, "loss": 0.8171, "step": 20777 }, { "epoch": 0.02, "learning_rate": 4.994725747477379e-05, "loss": 0.8109, "step": 20778 }, { "epoch": 0.02, "learning_rate": 4.9947252375127274e-05, "loss": 1.1602, "step": 20779 }, { "epoch": 0.02, "learning_rate": 4.994724727523449e-05, "loss": 1.2367, "step": 20780 }, { "epoch": 0.02, "learning_rate": 4.9947242175095436e-05, "loss": 1.3149, "step": 20781 }, { "epoch": 0.02, "learning_rate": 4.994723707471012e-05, "loss": 1.0764, "step": 20782 }, { "epoch": 0.02, "learning_rate": 4.9947231974078524e-05, "loss": 1.169, "step": 20783 }, { "epoch": 0.02, "learning_rate": 4.994722687320067e-05, "loss": 0.9984, "step": 20784 }, { "epoch": 0.02, "learning_rate": 4.9947221772076534e-05, "loss": 1.0858, "step": 20785 }, { "epoch": 0.02, "learning_rate": 4.9947216670706145e-05, "loss": 1.0889, "step": 20786 }, { "epoch": 0.02, "learning_rate": 4.994721156908948e-05, "loss": 1.1517, "step": 20787 }, { "epoch": 0.02, "learning_rate": 4.994720646722655e-05, "loss": 1.0422, "step": 20788 }, { "epoch": 0.02, "learning_rate": 4.9947201365117344e-05, "loss": 1.7072, "step": 20789 }, { "epoch": 0.02, "learning_rate": 4.9947196262761876e-05, "loss": 2.0981, "step": 20790 }, { "epoch": 0.02, "learning_rate": 4.994719116016014e-05, "loss": 5.9634, "step": 20791 }, { "epoch": 0.02, "learning_rate": 4.9947186057312137e-05, "loss": 5.8435, "step": 20792 }, { "epoch": 0.02, "learning_rate": 4.994718095421786e-05, "loss": 5.8926, "step": 20793 }, { "epoch": 0.02, "learning_rate": 4.994717585087731e-05, "loss": 4.4387, "step": 20794 }, { "epoch": 0.02, "learning_rate": 4.994717074729051e-05, "loss": 1.0005, "step": 20795 }, { "epoch": 0.02, "learning_rate": 4.994716564345743e-05, "loss": 1.1464, "step": 20796 }, { "epoch": 0.02, "learning_rate": 4.9947160539378084e-05, "loss": 1.0787, "step": 20797 }, { "epoch": 0.02, "learning_rate": 4.994715543505247e-05, "loss": 1.4069, "step": 20798 }, { "epoch": 0.02, "learning_rate": 4.994715033048059e-05, "loss": 1.0131, "step": 20799 }, { "epoch": 0.02, "learning_rate": 4.9947145225662436e-05, "loss": 1.1181, "step": 20800 }, { "epoch": 0.02, "learning_rate": 4.994714012059802e-05, "loss": 1.2829, "step": 20801 }, { "epoch": 0.02, "learning_rate": 4.9947135015287336e-05, "loss": 1.1655, "step": 20802 }, { "epoch": 0.02, "learning_rate": 4.994712990973039e-05, "loss": 0.9592, "step": 20803 }, { "epoch": 0.02, "learning_rate": 4.994712480392717e-05, "loss": 0.6747, "step": 20804 }, { "epoch": 0.02, "learning_rate": 4.994711969787769e-05, "loss": 0.9551, "step": 20805 }, { "epoch": 0.02, "learning_rate": 4.9947114591581925e-05, "loss": 1.1996, "step": 20806 }, { "epoch": 0.02, "learning_rate": 4.994710948503991e-05, "loss": 1.004, "step": 20807 }, { "epoch": 0.02, "learning_rate": 4.994710437825162e-05, "loss": 1.195, "step": 20808 }, { "epoch": 0.02, "learning_rate": 4.994709927121706e-05, "loss": 0.9958, "step": 20809 }, { "epoch": 0.02, "learning_rate": 4.994709416393624e-05, "loss": 1.2091, "step": 20810 }, { "epoch": 0.02, "learning_rate": 4.9947089056409147e-05, "loss": 1.1386, "step": 20811 }, { "epoch": 0.02, "learning_rate": 4.9947083948635784e-05, "loss": 1.045, "step": 20812 }, { "epoch": 0.02, "learning_rate": 4.9947078840616166e-05, "loss": 1.1134, "step": 20813 }, { "epoch": 0.02, "learning_rate": 4.994707373235027e-05, "loss": 1.4223, "step": 20814 }, { "epoch": 0.02, "learning_rate": 4.994706862383811e-05, "loss": 1.1985, "step": 20815 }, { "epoch": 0.02, "learning_rate": 4.9947063515079685e-05, "loss": 1.2451, "step": 20816 }, { "epoch": 0.02, "learning_rate": 4.9947058406074995e-05, "loss": 1.0046, "step": 20817 }, { "epoch": 0.02, "learning_rate": 4.9947053296824034e-05, "loss": 0.8655, "step": 20818 }, { "epoch": 0.02, "learning_rate": 4.9947048187326804e-05, "loss": 1.1277, "step": 20819 }, { "epoch": 0.02, "learning_rate": 4.994704307758331e-05, "loss": 0.9173, "step": 20820 }, { "epoch": 0.02, "learning_rate": 4.9947037967593555e-05, "loss": 0.9538, "step": 20821 }, { "epoch": 0.02, "learning_rate": 4.994703285735752e-05, "loss": 0.7732, "step": 20822 }, { "epoch": 0.02, "learning_rate": 4.994702774687523e-05, "loss": 1.0398, "step": 20823 }, { "epoch": 0.02, "learning_rate": 4.9947022636146675e-05, "loss": 1.1848, "step": 20824 }, { "epoch": 0.02, "learning_rate": 4.9947017525171847e-05, "loss": 0.7015, "step": 20825 }, { "epoch": 0.02, "learning_rate": 4.9947012413950755e-05, "loss": 0.8922, "step": 20826 }, { "epoch": 0.02, "learning_rate": 4.9947007302483394e-05, "loss": 1.1208, "step": 20827 }, { "epoch": 0.02, "learning_rate": 4.9947002190769763e-05, "loss": 1.0976, "step": 20828 }, { "epoch": 0.02, "learning_rate": 4.9946997078809876e-05, "loss": 1.2356, "step": 20829 }, { "epoch": 0.02, "learning_rate": 4.994699196660372e-05, "loss": 0.8612, "step": 20830 }, { "epoch": 0.02, "learning_rate": 4.994698685415129e-05, "loss": 1.0709, "step": 20831 }, { "epoch": 0.02, "learning_rate": 4.9946981741452604e-05, "loss": 0.7906, "step": 20832 }, { "epoch": 0.02, "learning_rate": 4.9946976628507645e-05, "loss": 0.9268, "step": 20833 }, { "epoch": 0.02, "learning_rate": 4.994697151531643e-05, "loss": 1.2107, "step": 20834 }, { "epoch": 0.02, "learning_rate": 4.994696640187894e-05, "loss": 1.335, "step": 20835 }, { "epoch": 0.02, "learning_rate": 4.994696128819518e-05, "loss": 1.0986, "step": 20836 }, { "epoch": 0.02, "learning_rate": 4.994695617426516e-05, "loss": 1.1421, "step": 20837 }, { "epoch": 0.02, "learning_rate": 4.994695106008888e-05, "loss": 1.0623, "step": 20838 }, { "epoch": 0.02, "learning_rate": 4.9946945945666327e-05, "loss": 1.326, "step": 20839 }, { "epoch": 0.02, "learning_rate": 4.99469408309975e-05, "loss": 1.2441, "step": 20840 }, { "epoch": 0.02, "learning_rate": 4.994693571608242e-05, "loss": 0.9912, "step": 20841 }, { "epoch": 0.02, "learning_rate": 4.994693060092107e-05, "loss": 0.9385, "step": 20842 }, { "epoch": 0.02, "learning_rate": 4.9946925485513454e-05, "loss": 1.2431, "step": 20843 }, { "epoch": 0.02, "learning_rate": 4.994692036985958e-05, "loss": 1.3795, "step": 20844 }, { "epoch": 0.02, "learning_rate": 4.994691525395943e-05, "loss": 0.9865, "step": 20845 }, { "epoch": 0.02, "learning_rate": 4.994691013781302e-05, "loss": 1.1457, "step": 20846 }, { "epoch": 0.02, "learning_rate": 4.994690502142034e-05, "loss": 0.9403, "step": 20847 }, { "epoch": 0.02, "learning_rate": 4.99468999047814e-05, "loss": 1.2468, "step": 20848 }, { "epoch": 0.02, "learning_rate": 4.9946894787896195e-05, "loss": 0.9696, "step": 20849 }, { "epoch": 0.02, "learning_rate": 4.9946889670764725e-05, "loss": 1.0227, "step": 20850 }, { "epoch": 0.02, "learning_rate": 4.9946884553386984e-05, "loss": 1.1223, "step": 20851 }, { "epoch": 0.02, "learning_rate": 4.994687943576298e-05, "loss": 1.1746, "step": 20852 }, { "epoch": 0.02, "learning_rate": 4.994687431789271e-05, "loss": 1.2417, "step": 20853 }, { "epoch": 0.02, "learning_rate": 4.994686919977618e-05, "loss": 0.9782, "step": 20854 }, { "epoch": 0.02, "learning_rate": 4.994686408141338e-05, "loss": 1.1173, "step": 20855 }, { "epoch": 0.02, "learning_rate": 4.994685896280432e-05, "loss": 1.0773, "step": 20856 }, { "epoch": 0.02, "learning_rate": 4.9946853843948994e-05, "loss": 1.2182, "step": 20857 }, { "epoch": 0.02, "learning_rate": 4.99468487248474e-05, "loss": 0.6974, "step": 20858 }, { "epoch": 0.02, "learning_rate": 4.994684360549954e-05, "loss": 1.0724, "step": 20859 }, { "epoch": 0.02, "learning_rate": 4.994683848590542e-05, "loss": 0.9102, "step": 20860 }, { "epoch": 0.02, "learning_rate": 4.994683336606504e-05, "loss": 0.9819, "step": 20861 }, { "epoch": 0.02, "learning_rate": 4.9946828245978386e-05, "loss": 0.8667, "step": 20862 }, { "epoch": 0.02, "learning_rate": 4.994682312564547e-05, "loss": 0.9272, "step": 20863 }, { "epoch": 0.02, "learning_rate": 4.9946818005066284e-05, "loss": 1.0001, "step": 20864 }, { "epoch": 0.02, "learning_rate": 4.994681288424085e-05, "loss": 1.0791, "step": 20865 }, { "epoch": 0.02, "learning_rate": 4.994680776316914e-05, "loss": 1.0464, "step": 20866 }, { "epoch": 0.02, "learning_rate": 4.994680264185116e-05, "loss": 1.0748, "step": 20867 }, { "epoch": 0.02, "learning_rate": 4.9946797520286924e-05, "loss": 1.1578, "step": 20868 }, { "epoch": 0.02, "learning_rate": 4.994679239847642e-05, "loss": 1.4008, "step": 20869 }, { "epoch": 0.02, "learning_rate": 4.994678727641965e-05, "loss": 1.103, "step": 20870 }, { "epoch": 0.02, "learning_rate": 4.994678215411662e-05, "loss": 1.4026, "step": 20871 }, { "epoch": 0.02, "learning_rate": 4.9946777031567323e-05, "loss": 1.2187, "step": 20872 }, { "epoch": 0.02, "learning_rate": 4.994677190877177e-05, "loss": 1.0401, "step": 20873 }, { "epoch": 0.02, "learning_rate": 4.994676678572995e-05, "loss": 1.1067, "step": 20874 }, { "epoch": 0.02, "learning_rate": 4.9946761662441857e-05, "loss": 0.8927, "step": 20875 }, { "epoch": 0.02, "learning_rate": 4.9946756538907515e-05, "loss": 1.4942, "step": 20876 }, { "epoch": 0.02, "learning_rate": 4.994675141512689e-05, "loss": 1.0961, "step": 20877 }, { "epoch": 0.02, "learning_rate": 4.9946746291100016e-05, "loss": 1.2383, "step": 20878 }, { "epoch": 0.02, "learning_rate": 4.994674116682687e-05, "loss": 0.9971, "step": 20879 }, { "epoch": 0.02, "learning_rate": 4.9946736042307465e-05, "loss": 1.0888, "step": 20880 }, { "epoch": 0.02, "learning_rate": 4.9946730917541795e-05, "loss": 1.0409, "step": 20881 }, { "epoch": 0.02, "learning_rate": 4.994672579252987e-05, "loss": 1.1712, "step": 20882 }, { "epoch": 0.02, "learning_rate": 4.994672066727167e-05, "loss": 0.9231, "step": 20883 }, { "epoch": 0.02, "learning_rate": 4.994671554176721e-05, "loss": 1.1859, "step": 20884 }, { "epoch": 0.02, "learning_rate": 4.9946710416016486e-05, "loss": 1.0145, "step": 20885 }, { "epoch": 0.02, "learning_rate": 4.9946705290019495e-05, "loss": 1.2621, "step": 20886 }, { "epoch": 0.02, "learning_rate": 4.994670016377625e-05, "loss": 1.0818, "step": 20887 }, { "epoch": 0.02, "learning_rate": 4.9946695037286736e-05, "loss": 1.1749, "step": 20888 }, { "epoch": 0.02, "learning_rate": 4.9946689910550956e-05, "loss": 0.9284, "step": 20889 }, { "epoch": 0.02, "learning_rate": 4.994668478356892e-05, "loss": 1.0909, "step": 20890 }, { "epoch": 0.02, "learning_rate": 4.994667965634061e-05, "loss": 0.9114, "step": 20891 }, { "epoch": 0.02, "learning_rate": 4.9946674528866044e-05, "loss": 1.1396, "step": 20892 }, { "epoch": 0.02, "learning_rate": 4.994666940114522e-05, "loss": 1.0086, "step": 20893 }, { "epoch": 0.02, "learning_rate": 4.994666427317812e-05, "loss": 1.1465, "step": 20894 }, { "epoch": 0.02, "learning_rate": 4.994665914496477e-05, "loss": 0.9023, "step": 20895 }, { "epoch": 0.02, "learning_rate": 4.994665401650515e-05, "loss": 1.3454, "step": 20896 }, { "epoch": 0.02, "learning_rate": 4.9946648887799266e-05, "loss": 1.3141, "step": 20897 }, { "epoch": 0.02, "learning_rate": 4.9946643758847125e-05, "loss": 0.9223, "step": 20898 }, { "epoch": 0.02, "learning_rate": 4.9946638629648715e-05, "loss": 0.8794, "step": 20899 }, { "epoch": 0.02, "learning_rate": 4.994663350020405e-05, "loss": 0.8245, "step": 20900 }, { "epoch": 0.02, "learning_rate": 4.994662837051311e-05, "loss": 0.86, "step": 20901 }, { "epoch": 0.02, "learning_rate": 4.994662324057592e-05, "loss": 1.3174, "step": 20902 }, { "epoch": 0.02, "learning_rate": 4.9946618110392465e-05, "loss": 0.5888, "step": 20903 }, { "epoch": 0.02, "learning_rate": 4.994661297996274e-05, "loss": 1.3055, "step": 20904 }, { "epoch": 0.02, "learning_rate": 4.9946607849286766e-05, "loss": 1.1049, "step": 20905 }, { "epoch": 0.02, "learning_rate": 4.9946602718364515e-05, "loss": 0.8983, "step": 20906 }, { "epoch": 0.02, "learning_rate": 4.994659758719601e-05, "loss": 0.8022, "step": 20907 }, { "epoch": 0.02, "learning_rate": 4.994659245578124e-05, "loss": 0.9775, "step": 20908 }, { "epoch": 0.02, "learning_rate": 4.994658732412021e-05, "loss": 1.0935, "step": 20909 }, { "epoch": 0.02, "learning_rate": 4.9946582192212916e-05, "loss": 0.9607, "step": 20910 }, { "epoch": 0.02, "learning_rate": 4.9946577060059357e-05, "loss": 1.119, "step": 20911 }, { "epoch": 0.02, "learning_rate": 4.994657192765954e-05, "loss": 0.9225, "step": 20912 }, { "epoch": 0.02, "learning_rate": 4.994656679501346e-05, "loss": 1.0327, "step": 20913 }, { "epoch": 0.02, "learning_rate": 4.9946561662121115e-05, "loss": 0.8572, "step": 20914 }, { "epoch": 0.02, "learning_rate": 4.994655652898251e-05, "loss": 0.6497, "step": 20915 }, { "epoch": 0.02, "learning_rate": 4.994655139559765e-05, "loss": 0.772, "step": 20916 }, { "epoch": 0.02, "learning_rate": 4.994654626196652e-05, "loss": 1.2574, "step": 20917 }, { "epoch": 0.02, "learning_rate": 4.994654112808913e-05, "loss": 1.2265, "step": 20918 }, { "epoch": 0.02, "learning_rate": 4.9946535993965485e-05, "loss": 1.2345, "step": 20919 }, { "epoch": 0.02, "learning_rate": 4.9946530859595566e-05, "loss": 0.9126, "step": 20920 }, { "epoch": 0.02, "learning_rate": 4.994652572497939e-05, "loss": 2.0427, "step": 20921 }, { "epoch": 0.02, "learning_rate": 4.994652059011695e-05, "loss": 1.504, "step": 20922 }, { "epoch": 0.02, "learning_rate": 4.994651545500826e-05, "loss": 1.0284, "step": 20923 }, { "epoch": 0.02, "learning_rate": 4.9946510319653295e-05, "loss": 1.3978, "step": 20924 }, { "epoch": 0.02, "learning_rate": 4.994650518405208e-05, "loss": 1.101, "step": 20925 }, { "epoch": 0.02, "learning_rate": 4.994650004820459e-05, "loss": 1.1041, "step": 20926 }, { "epoch": 0.02, "learning_rate": 4.994649491211085e-05, "loss": 0.9689, "step": 20927 }, { "epoch": 0.02, "learning_rate": 4.994648977577084e-05, "loss": 1.1889, "step": 20928 }, { "epoch": 0.02, "learning_rate": 4.9946484639184584e-05, "loss": 1.1946, "step": 20929 }, { "epoch": 0.02, "learning_rate": 4.994647950235205e-05, "loss": 1.0847, "step": 20930 }, { "epoch": 0.02, "learning_rate": 4.9946474365273264e-05, "loss": 1.0617, "step": 20931 }, { "epoch": 0.02, "learning_rate": 4.994646922794822e-05, "loss": 1.1662, "step": 20932 }, { "epoch": 0.02, "learning_rate": 4.9946464090376906e-05, "loss": 0.985, "step": 20933 }, { "epoch": 0.02, "learning_rate": 4.994645895255933e-05, "loss": 1.4368, "step": 20934 }, { "epoch": 0.02, "learning_rate": 4.99464538144955e-05, "loss": 1.1823, "step": 20935 }, { "epoch": 0.02, "learning_rate": 4.994644867618541e-05, "loss": 1.4004, "step": 20936 }, { "epoch": 0.02, "learning_rate": 4.994644353762905e-05, "loss": 1.1122, "step": 20937 }, { "epoch": 0.02, "learning_rate": 4.9946438398826436e-05, "loss": 1.186, "step": 20938 }, { "epoch": 0.02, "learning_rate": 4.994643325977756e-05, "loss": 1.205, "step": 20939 }, { "epoch": 0.02, "learning_rate": 4.9946428120482424e-05, "loss": 1.3062, "step": 20940 }, { "epoch": 0.02, "learning_rate": 4.994642298094103e-05, "loss": 1.2884, "step": 20941 }, { "epoch": 0.02, "learning_rate": 4.994641784115337e-05, "loss": 1.0721, "step": 20942 }, { "epoch": 0.02, "learning_rate": 4.994641270111945e-05, "loss": 1.2893, "step": 20943 }, { "epoch": 0.02, "learning_rate": 4.9946407560839284e-05, "loss": 0.955, "step": 20944 }, { "epoch": 0.02, "learning_rate": 4.994640242031284e-05, "loss": 0.9796, "step": 20945 }, { "epoch": 0.02, "learning_rate": 4.9946397279540136e-05, "loss": 0.9761, "step": 20946 }, { "epoch": 0.02, "learning_rate": 4.994639213852118e-05, "loss": 0.8201, "step": 20947 }, { "epoch": 0.02, "learning_rate": 4.994638699725596e-05, "loss": 0.7255, "step": 20948 }, { "epoch": 0.02, "learning_rate": 4.994638185574448e-05, "loss": 1.8115, "step": 20949 }, { "epoch": 0.02, "learning_rate": 4.994637671398674e-05, "loss": 1.7783, "step": 20950 }, { "epoch": 0.02, "learning_rate": 4.994637157198274e-05, "loss": 1.1134, "step": 20951 }, { "epoch": 0.02, "learning_rate": 4.994636642973248e-05, "loss": 1.4805, "step": 20952 }, { "epoch": 0.02, "learning_rate": 4.9946361287235954e-05, "loss": 1.0357, "step": 20953 }, { "epoch": 0.02, "learning_rate": 4.9946356144493175e-05, "loss": 1.0688, "step": 20954 }, { "epoch": 0.02, "learning_rate": 4.994635100150414e-05, "loss": 1.239, "step": 20955 }, { "epoch": 0.02, "learning_rate": 4.994634585826884e-05, "loss": 1.0338, "step": 20956 }, { "epoch": 0.02, "learning_rate": 4.9946340714787275e-05, "loss": 1.1258, "step": 20957 }, { "epoch": 0.02, "learning_rate": 4.994633557105946e-05, "loss": 1.0223, "step": 20958 }, { "epoch": 0.02, "learning_rate": 4.994633042708538e-05, "loss": 1.1425, "step": 20959 }, { "epoch": 0.02, "learning_rate": 4.9946325282865035e-05, "loss": 1.1216, "step": 20960 }, { "epoch": 0.02, "learning_rate": 4.994632013839844e-05, "loss": 1.3265, "step": 20961 }, { "epoch": 0.02, "learning_rate": 4.994631499368558e-05, "loss": 1.1231, "step": 20962 }, { "epoch": 0.02, "learning_rate": 4.994630984872646e-05, "loss": 1.2164, "step": 20963 }, { "epoch": 0.02, "learning_rate": 4.994630470352109e-05, "loss": 0.9801, "step": 20964 }, { "epoch": 0.02, "learning_rate": 4.994629955806946e-05, "loss": 1.1029, "step": 20965 }, { "epoch": 0.02, "learning_rate": 4.994629441237155e-05, "loss": 1.0052, "step": 20966 }, { "epoch": 0.02, "learning_rate": 4.99462892664274e-05, "loss": 0.9723, "step": 20967 }, { "epoch": 0.02, "learning_rate": 4.994628412023699e-05, "loss": 1.3189, "step": 20968 }, { "epoch": 0.02, "learning_rate": 4.994627897380031e-05, "loss": 1.1371, "step": 20969 }, { "epoch": 0.02, "learning_rate": 4.994627382711738e-05, "loss": 1.1727, "step": 20970 }, { "epoch": 0.02, "learning_rate": 4.9946268680188184e-05, "loss": 1.2555, "step": 20971 }, { "epoch": 0.02, "learning_rate": 4.994626353301274e-05, "loss": 1.277, "step": 20972 }, { "epoch": 0.02, "learning_rate": 4.9946258385591026e-05, "loss": 1.3507, "step": 20973 }, { "epoch": 0.02, "learning_rate": 4.9946253237923055e-05, "loss": 1.1635, "step": 20974 }, { "epoch": 0.02, "learning_rate": 4.994624809000883e-05, "loss": 1.1027, "step": 20975 }, { "epoch": 0.02, "learning_rate": 4.994624294184834e-05, "loss": 1.1862, "step": 20976 }, { "epoch": 0.02, "learning_rate": 4.9946237793441594e-05, "loss": 0.8513, "step": 20977 }, { "epoch": 0.02, "learning_rate": 4.994623264478859e-05, "loss": 1.0352, "step": 20978 }, { "epoch": 0.02, "learning_rate": 4.9946227495889334e-05, "loss": 1.2291, "step": 20979 }, { "epoch": 0.02, "learning_rate": 4.994622234674381e-05, "loss": 1.0522, "step": 20980 }, { "epoch": 0.02, "learning_rate": 4.994621719735203e-05, "loss": 1.1142, "step": 20981 }, { "epoch": 0.02, "learning_rate": 4.994621204771399e-05, "loss": 0.9774, "step": 20982 }, { "epoch": 0.02, "learning_rate": 4.994620689782969e-05, "loss": 0.9852, "step": 20983 }, { "epoch": 0.02, "learning_rate": 4.994620174769913e-05, "loss": 1.09, "step": 20984 }, { "epoch": 0.02, "learning_rate": 4.994619659732232e-05, "loss": 0.8063, "step": 20985 }, { "epoch": 0.02, "learning_rate": 4.9946191446699245e-05, "loss": 0.9329, "step": 20986 }, { "epoch": 0.02, "learning_rate": 4.994618629582992e-05, "loss": 1.4585, "step": 20987 }, { "epoch": 0.02, "learning_rate": 4.9946181144714326e-05, "loss": 1.1385, "step": 20988 }, { "epoch": 0.02, "learning_rate": 4.9946175993352486e-05, "loss": 0.9578, "step": 20989 }, { "epoch": 0.02, "learning_rate": 4.9946170841744376e-05, "loss": 0.6807, "step": 20990 }, { "epoch": 0.02, "learning_rate": 4.994616568989001e-05, "loss": 1.2238, "step": 20991 }, { "epoch": 0.02, "learning_rate": 4.994616053778939e-05, "loss": 1.1575, "step": 20992 }, { "epoch": 0.02, "learning_rate": 4.994615538544251e-05, "loss": 1.133, "step": 20993 }, { "epoch": 0.02, "learning_rate": 4.9946150232849374e-05, "loss": 0.9837, "step": 20994 }, { "epoch": 0.02, "learning_rate": 4.9946145080009976e-05, "loss": 1.0795, "step": 20995 }, { "epoch": 0.02, "learning_rate": 4.994613992692432e-05, "loss": 0.9672, "step": 20996 }, { "epoch": 0.02, "learning_rate": 4.994613477359241e-05, "loss": 1.1224, "step": 20997 }, { "epoch": 0.02, "learning_rate": 4.9946129620014245e-05, "loss": 0.8909, "step": 20998 }, { "epoch": 0.02, "learning_rate": 4.994612446618982e-05, "loss": 1.0183, "step": 20999 }, { "epoch": 0.02, "learning_rate": 4.994611931211913e-05, "loss": 1.1505, "step": 21000 }, { "epoch": 0.02, "eval_loss": 1.017486810684204, "eval_runtime": 85.53, "eval_samples_per_second": 16.193, "eval_steps_per_second": 4.057, "step": 21000 }, { "epoch": 0.02, "learning_rate": 4.9946114157802195e-05, "loss": 1.0313, "step": 21001 }, { "epoch": 0.02, "learning_rate": 4.9946109003239e-05, "loss": 1.0053, "step": 21002 }, { "epoch": 0.02, "learning_rate": 4.9946103848429536e-05, "loss": 1.2735, "step": 21003 }, { "epoch": 0.02, "learning_rate": 4.994609869337382e-05, "loss": 1.0377, "step": 21004 }, { "epoch": 0.02, "learning_rate": 4.994609353807185e-05, "loss": 1.2023, "step": 21005 }, { "epoch": 0.02, "learning_rate": 4.994608838252362e-05, "loss": 1.3537, "step": 21006 }, { "epoch": 0.02, "learning_rate": 4.994608322672914e-05, "loss": 1.1081, "step": 21007 }, { "epoch": 0.02, "learning_rate": 4.9946078070688395e-05, "loss": 1.1041, "step": 21008 }, { "epoch": 0.02, "learning_rate": 4.99460729144014e-05, "loss": 0.9523, "step": 21009 }, { "epoch": 0.02, "learning_rate": 4.9946067757868136e-05, "loss": 1.0675, "step": 21010 }, { "epoch": 0.02, "learning_rate": 4.9946062601088625e-05, "loss": 0.8807, "step": 21011 }, { "epoch": 0.02, "learning_rate": 4.994605744406285e-05, "loss": 1.1967, "step": 21012 }, { "epoch": 0.02, "learning_rate": 4.994605228679082e-05, "loss": 1.1674, "step": 21013 }, { "epoch": 0.02, "learning_rate": 4.9946047129272536e-05, "loss": 1.2564, "step": 21014 }, { "epoch": 0.02, "learning_rate": 4.9946041971507994e-05, "loss": 0.8709, "step": 21015 }, { "epoch": 0.02, "learning_rate": 4.9946036813497196e-05, "loss": 0.9843, "step": 21016 }, { "epoch": 0.02, "learning_rate": 4.994603165524014e-05, "loss": 1.355, "step": 21017 }, { "epoch": 0.02, "learning_rate": 4.994602649673683e-05, "loss": 1.0709, "step": 21018 }, { "epoch": 0.02, "learning_rate": 4.994602133798726e-05, "loss": 1.2502, "step": 21019 }, { "epoch": 0.02, "learning_rate": 4.9946016178991434e-05, "loss": 0.9942, "step": 21020 }, { "epoch": 0.02, "learning_rate": 4.994601101974935e-05, "loss": 1.094, "step": 21021 }, { "epoch": 0.02, "learning_rate": 4.994600586026101e-05, "loss": 1.4539, "step": 21022 }, { "epoch": 0.02, "learning_rate": 4.994600070052642e-05, "loss": 0.8906, "step": 21023 }, { "epoch": 0.02, "learning_rate": 4.994599554054557e-05, "loss": 1.0852, "step": 21024 }, { "epoch": 0.02, "learning_rate": 4.9945990380318456e-05, "loss": 1.0328, "step": 21025 }, { "epoch": 0.02, "learning_rate": 4.9945985219845096e-05, "loss": 1.1608, "step": 21026 }, { "epoch": 0.02, "learning_rate": 4.994598005912547e-05, "loss": 0.935, "step": 21027 }, { "epoch": 0.02, "learning_rate": 4.99459748981596e-05, "loss": 1.1968, "step": 21028 }, { "epoch": 0.02, "learning_rate": 4.9945969736947464e-05, "loss": 1.0615, "step": 21029 }, { "epoch": 0.02, "learning_rate": 4.9945964575489066e-05, "loss": 0.7513, "step": 21030 }, { "epoch": 0.02, "learning_rate": 4.9945959413784425e-05, "loss": 1.2473, "step": 21031 }, { "epoch": 0.02, "learning_rate": 4.994595425183353e-05, "loss": 1.0366, "step": 21032 }, { "epoch": 0.02, "learning_rate": 4.994594908963637e-05, "loss": 0.4786, "step": 21033 }, { "epoch": 0.02, "learning_rate": 4.994594392719296e-05, "loss": 1.1505, "step": 21034 }, { "epoch": 0.02, "learning_rate": 4.9945938764503284e-05, "loss": 1.268, "step": 21035 }, { "epoch": 0.02, "learning_rate": 4.994593360156736e-05, "loss": 1.0142, "step": 21036 }, { "epoch": 0.02, "learning_rate": 4.9945928438385184e-05, "loss": 0.8381, "step": 21037 }, { "epoch": 0.02, "learning_rate": 4.994592327495675e-05, "loss": 1.4257, "step": 21038 }, { "epoch": 0.02, "learning_rate": 4.994591811128205e-05, "loss": 1.1497, "step": 21039 }, { "epoch": 0.02, "learning_rate": 4.9945912947361105e-05, "loss": 1.245, "step": 21040 }, { "epoch": 0.02, "learning_rate": 4.99459077831939e-05, "loss": 0.8979, "step": 21041 }, { "epoch": 0.02, "learning_rate": 4.994590261878045e-05, "loss": 0.8957, "step": 21042 }, { "epoch": 0.02, "learning_rate": 4.9945897454120734e-05, "loss": 0.9957, "step": 21043 }, { "epoch": 0.02, "learning_rate": 4.994589228921476e-05, "loss": 1.072, "step": 21044 }, { "epoch": 0.02, "learning_rate": 4.994588712406254e-05, "loss": 1.0411, "step": 21045 }, { "epoch": 0.02, "learning_rate": 4.994588195866406e-05, "loss": 1.0436, "step": 21046 }, { "epoch": 0.02, "learning_rate": 4.9945876793019316e-05, "loss": 1.1942, "step": 21047 }, { "epoch": 0.02, "learning_rate": 4.9945871627128333e-05, "loss": 1.1678, "step": 21048 }, { "epoch": 0.02, "learning_rate": 4.994586646099109e-05, "loss": 1.0374, "step": 21049 }, { "epoch": 0.02, "learning_rate": 4.9945861294607585e-05, "loss": 1.183, "step": 21050 }, { "epoch": 0.02, "learning_rate": 4.994585612797783e-05, "loss": 0.7339, "step": 21051 }, { "epoch": 0.02, "learning_rate": 4.994585096110182e-05, "loss": 0.9819, "step": 21052 }, { "epoch": 0.02, "learning_rate": 4.9945845793979555e-05, "loss": 0.898, "step": 21053 }, { "epoch": 0.02, "learning_rate": 4.9945840626611035e-05, "loss": 1.0127, "step": 21054 }, { "epoch": 0.02, "learning_rate": 4.994583545899626e-05, "loss": 0.9624, "step": 21055 }, { "epoch": 0.02, "learning_rate": 4.994583029113523e-05, "loss": 1.2786, "step": 21056 }, { "epoch": 0.02, "learning_rate": 4.9945825123027945e-05, "loss": 1.4213, "step": 21057 }, { "epoch": 0.02, "learning_rate": 4.9945819954674407e-05, "loss": 1.2429, "step": 21058 }, { "epoch": 0.02, "learning_rate": 4.994581478607461e-05, "loss": 1.0678, "step": 21059 }, { "epoch": 0.02, "learning_rate": 4.994580961722857e-05, "loss": 1.065, "step": 21060 }, { "epoch": 0.02, "learning_rate": 4.9945804448136255e-05, "loss": 1.0489, "step": 21061 }, { "epoch": 0.02, "learning_rate": 4.99457992787977e-05, "loss": 1.2093, "step": 21062 }, { "epoch": 0.02, "learning_rate": 4.9945794109212893e-05, "loss": 1.161, "step": 21063 }, { "epoch": 0.02, "learning_rate": 4.994578893938182e-05, "loss": 1.096, "step": 21064 }, { "epoch": 0.02, "learning_rate": 4.994578376930451e-05, "loss": 0.9711, "step": 21065 }, { "epoch": 0.02, "learning_rate": 4.994577859898093e-05, "loss": 0.6296, "step": 21066 }, { "epoch": 0.02, "learning_rate": 4.99457734284111e-05, "loss": 0.614, "step": 21067 }, { "epoch": 0.02, "learning_rate": 4.9945768257595016e-05, "loss": 1.9061, "step": 21068 }, { "epoch": 0.02, "learning_rate": 4.994576308653268e-05, "loss": 1.7768, "step": 21069 }, { "epoch": 0.02, "learning_rate": 4.994575791522409e-05, "loss": 0.7634, "step": 21070 }, { "epoch": 0.02, "learning_rate": 4.9945752743669246e-05, "loss": 1.1721, "step": 21071 }, { "epoch": 0.02, "learning_rate": 4.994574757186815e-05, "loss": 1.1355, "step": 21072 }, { "epoch": 0.02, "learning_rate": 4.9945742399820795e-05, "loss": 0.7708, "step": 21073 }, { "epoch": 0.02, "learning_rate": 4.9945737227527185e-05, "loss": 1.0259, "step": 21074 }, { "epoch": 0.02, "learning_rate": 4.9945732054987325e-05, "loss": 1.265, "step": 21075 }, { "epoch": 0.02, "learning_rate": 4.99457268822012e-05, "loss": 0.7292, "step": 21076 }, { "epoch": 0.02, "learning_rate": 4.9945721709168844e-05, "loss": 0.7803, "step": 21077 }, { "epoch": 0.02, "learning_rate": 4.9945716535890216e-05, "loss": 1.243, "step": 21078 }, { "epoch": 0.02, "learning_rate": 4.9945711362365345e-05, "loss": 1.183, "step": 21079 }, { "epoch": 0.02, "learning_rate": 4.994570618859421e-05, "loss": 1.0076, "step": 21080 }, { "epoch": 0.02, "learning_rate": 4.994570101457683e-05, "loss": 0.7705, "step": 21081 }, { "epoch": 0.02, "learning_rate": 4.9945695840313196e-05, "loss": 1.2117, "step": 21082 }, { "epoch": 0.02, "learning_rate": 4.99456906658033e-05, "loss": 1.142, "step": 21083 }, { "epoch": 0.02, "learning_rate": 4.994568549104716e-05, "loss": 1.1821, "step": 21084 }, { "epoch": 0.02, "learning_rate": 4.994568031604476e-05, "loss": 1.1167, "step": 21085 }, { "epoch": 0.02, "learning_rate": 4.994567514079612e-05, "loss": 1.0798, "step": 21086 }, { "epoch": 0.02, "learning_rate": 4.994566996530121e-05, "loss": 0.9731, "step": 21087 }, { "epoch": 0.02, "learning_rate": 4.9945664789560054e-05, "loss": 1.0906, "step": 21088 }, { "epoch": 0.02, "learning_rate": 4.994565961357265e-05, "loss": 1.0388, "step": 21089 }, { "epoch": 0.02, "learning_rate": 4.994565443733899e-05, "loss": 1.334, "step": 21090 }, { "epoch": 0.02, "learning_rate": 4.9945649260859076e-05, "loss": 1.0248, "step": 21091 }, { "epoch": 0.02, "learning_rate": 4.99456440841329e-05, "loss": 1.0796, "step": 21092 }, { "epoch": 0.02, "learning_rate": 4.9945638907160485e-05, "loss": 1.0, "step": 21093 }, { "epoch": 0.02, "learning_rate": 4.994563372994181e-05, "loss": 1.1476, "step": 21094 }, { "epoch": 0.02, "learning_rate": 4.994562855247689e-05, "loss": 1.2354, "step": 21095 }, { "epoch": 0.02, "learning_rate": 4.9945623374765704e-05, "loss": 1.214, "step": 21096 }, { "epoch": 0.02, "learning_rate": 4.9945618196808276e-05, "loss": 1.0097, "step": 21097 }, { "epoch": 0.02, "learning_rate": 4.9945613018604585e-05, "loss": 1.1555, "step": 21098 }, { "epoch": 0.02, "learning_rate": 4.994560784015465e-05, "loss": 0.9204, "step": 21099 }, { "epoch": 0.02, "learning_rate": 4.994560266145846e-05, "loss": 1.1744, "step": 21100 }, { "epoch": 0.02, "learning_rate": 4.994559748251602e-05, "loss": 1.0138, "step": 21101 }, { "epoch": 0.02, "learning_rate": 4.9945592303327326e-05, "loss": 1.0485, "step": 21102 }, { "epoch": 0.02, "learning_rate": 4.994558712389238e-05, "loss": 1.0806, "step": 21103 }, { "epoch": 0.02, "learning_rate": 4.994558194421119e-05, "loss": 1.2148, "step": 21104 }, { "epoch": 0.02, "learning_rate": 4.994557676428373e-05, "loss": 0.9992, "step": 21105 }, { "epoch": 0.02, "learning_rate": 4.994557158411003e-05, "loss": 0.9681, "step": 21106 }, { "epoch": 0.02, "learning_rate": 4.994556640369007e-05, "loss": 0.9423, "step": 21107 }, { "epoch": 0.02, "learning_rate": 4.994556122302386e-05, "loss": 1.0427, "step": 21108 }, { "epoch": 0.02, "learning_rate": 4.99455560421114e-05, "loss": 1.0163, "step": 21109 }, { "epoch": 0.02, "learning_rate": 4.994555086095269e-05, "loss": 0.9803, "step": 21110 }, { "epoch": 0.02, "learning_rate": 4.994554567954772e-05, "loss": 0.6944, "step": 21111 }, { "epoch": 0.02, "learning_rate": 4.994554049789651e-05, "loss": 1.1024, "step": 21112 }, { "epoch": 0.02, "learning_rate": 4.9945535315999045e-05, "loss": 1.3173, "step": 21113 }, { "epoch": 0.02, "learning_rate": 4.994553013385532e-05, "loss": 1.1826, "step": 21114 }, { "epoch": 0.02, "learning_rate": 4.994552495146535e-05, "loss": 1.214, "step": 21115 }, { "epoch": 0.02, "learning_rate": 4.994551976882913e-05, "loss": 1.0601, "step": 21116 }, { "epoch": 0.02, "learning_rate": 4.994551458594665e-05, "loss": 1.2321, "step": 21117 }, { "epoch": 0.02, "learning_rate": 4.9945509402817925e-05, "loss": 1.2789, "step": 21118 }, { "epoch": 0.02, "learning_rate": 4.994550421944295e-05, "loss": 1.4251, "step": 21119 }, { "epoch": 0.02, "learning_rate": 4.9945499035821717e-05, "loss": 1.2614, "step": 21120 }, { "epoch": 0.02, "learning_rate": 4.994549385195424e-05, "loss": 1.2809, "step": 21121 }, { "epoch": 0.02, "learning_rate": 4.9945488667840503e-05, "loss": 0.9847, "step": 21122 }, { "epoch": 0.02, "learning_rate": 4.994548348348053e-05, "loss": 1.2301, "step": 21123 }, { "epoch": 0.02, "learning_rate": 4.994547829887429e-05, "loss": 0.8903, "step": 21124 }, { "epoch": 0.02, "learning_rate": 4.99454731140218e-05, "loss": 0.8109, "step": 21125 }, { "epoch": 0.02, "learning_rate": 4.9945467928923064e-05, "loss": 1.1551, "step": 21126 }, { "epoch": 0.02, "learning_rate": 4.994546274357808e-05, "loss": 0.5476, "step": 21127 }, { "epoch": 0.02, "learning_rate": 4.9945457557986844e-05, "loss": 0.5172, "step": 21128 }, { "epoch": 0.02, "learning_rate": 4.9945452372149347e-05, "loss": 0.2947, "step": 21129 }, { "epoch": 0.02, "learning_rate": 4.9945447186065606e-05, "loss": 0.8786, "step": 21130 }, { "epoch": 0.02, "learning_rate": 4.994544199973561e-05, "loss": 0.7853, "step": 21131 }, { "epoch": 0.02, "learning_rate": 4.994543681315937e-05, "loss": 1.1334, "step": 21132 }, { "epoch": 0.02, "learning_rate": 4.9945431626336876e-05, "loss": 0.8781, "step": 21133 }, { "epoch": 0.02, "learning_rate": 4.994542643926813e-05, "loss": 1.1763, "step": 21134 }, { "epoch": 0.02, "learning_rate": 4.994542125195314e-05, "loss": 1.0743, "step": 21135 }, { "epoch": 0.02, "learning_rate": 4.9945416064391894e-05, "loss": 1.3162, "step": 21136 }, { "epoch": 0.02, "learning_rate": 4.9945410876584394e-05, "loss": 1.4024, "step": 21137 }, { "epoch": 0.02, "learning_rate": 4.9945405688530645e-05, "loss": 1.2238, "step": 21138 }, { "epoch": 0.02, "learning_rate": 4.9945400500230646e-05, "loss": 1.0513, "step": 21139 }, { "epoch": 0.02, "learning_rate": 4.99453953116844e-05, "loss": 0.9626, "step": 21140 }, { "epoch": 0.02, "learning_rate": 4.99453901228919e-05, "loss": 1.3468, "step": 21141 }, { "epoch": 0.02, "learning_rate": 4.994538493385315e-05, "loss": 1.1838, "step": 21142 }, { "epoch": 0.02, "learning_rate": 4.994537974456815e-05, "loss": 1.0936, "step": 21143 }, { "epoch": 0.02, "learning_rate": 4.994537455503689e-05, "loss": 0.8794, "step": 21144 }, { "epoch": 0.02, "learning_rate": 4.99453693652594e-05, "loss": 0.962, "step": 21145 }, { "epoch": 0.02, "learning_rate": 4.9945364175235646e-05, "loss": 0.8169, "step": 21146 }, { "epoch": 0.02, "learning_rate": 4.9945358984965645e-05, "loss": 1.2008, "step": 21147 }, { "epoch": 0.02, "learning_rate": 4.9945353794449395e-05, "loss": 1.0444, "step": 21148 }, { "epoch": 0.02, "learning_rate": 4.994534860368689e-05, "loss": 1.0054, "step": 21149 }, { "epoch": 0.02, "learning_rate": 4.994534341267815e-05, "loss": 1.4666, "step": 21150 }, { "epoch": 0.02, "learning_rate": 4.994533822142314e-05, "loss": 1.2751, "step": 21151 }, { "epoch": 0.02, "learning_rate": 4.9945333029921894e-05, "loss": 1.3218, "step": 21152 }, { "epoch": 0.02, "learning_rate": 4.99453278381744e-05, "loss": 1.28, "step": 21153 }, { "epoch": 0.02, "learning_rate": 4.9945322646180644e-05, "loss": 1.2512, "step": 21154 }, { "epoch": 0.02, "learning_rate": 4.994531745394064e-05, "loss": 1.118, "step": 21155 }, { "epoch": 0.02, "learning_rate": 4.994531226145439e-05, "loss": 1.2186, "step": 21156 }, { "epoch": 0.02, "learning_rate": 4.9945307068721894e-05, "loss": 1.0428, "step": 21157 }, { "epoch": 0.02, "learning_rate": 4.994530187574314e-05, "loss": 1.0928, "step": 21158 }, { "epoch": 0.02, "learning_rate": 4.994529668251815e-05, "loss": 1.1575, "step": 21159 }, { "epoch": 0.02, "learning_rate": 4.99452914890469e-05, "loss": 1.1642, "step": 21160 }, { "epoch": 0.02, "learning_rate": 4.994528629532941e-05, "loss": 1.0627, "step": 21161 }, { "epoch": 0.02, "learning_rate": 4.994528110136566e-05, "loss": 0.9902, "step": 21162 }, { "epoch": 0.02, "learning_rate": 4.994527590715567e-05, "loss": 1.2233, "step": 21163 }, { "epoch": 0.02, "learning_rate": 4.994527071269941e-05, "loss": 0.5616, "step": 21164 }, { "epoch": 0.02, "learning_rate": 4.994526551799692e-05, "loss": 0.7776, "step": 21165 }, { "epoch": 0.02, "learning_rate": 4.994526032304818e-05, "loss": 1.1388, "step": 21166 }, { "epoch": 0.02, "learning_rate": 4.994525512785319e-05, "loss": 1.1785, "step": 21167 }, { "epoch": 0.02, "learning_rate": 4.994524993241194e-05, "loss": 1.2019, "step": 21168 }, { "epoch": 0.02, "learning_rate": 4.9945244736724454e-05, "loss": 0.9579, "step": 21169 }, { "epoch": 0.02, "learning_rate": 4.994523954079072e-05, "loss": 1.1858, "step": 21170 }, { "epoch": 0.02, "learning_rate": 4.9945234344610724e-05, "loss": 1.0865, "step": 21171 }, { "epoch": 0.02, "learning_rate": 4.9945229148184494e-05, "loss": 1.0308, "step": 21172 }, { "epoch": 0.02, "learning_rate": 4.9945223951512e-05, "loss": 1.0123, "step": 21173 }, { "epoch": 0.02, "learning_rate": 4.9945218754593274e-05, "loss": 0.9677, "step": 21174 }, { "epoch": 0.02, "learning_rate": 4.994521355742828e-05, "loss": 1.0348, "step": 21175 }, { "epoch": 0.02, "learning_rate": 4.9945208360017056e-05, "loss": 1.0258, "step": 21176 }, { "epoch": 0.02, "learning_rate": 4.994520316235958e-05, "loss": 1.04, "step": 21177 }, { "epoch": 0.02, "learning_rate": 4.994519796445585e-05, "loss": 0.9936, "step": 21178 }, { "epoch": 0.02, "learning_rate": 4.994519276630587e-05, "loss": 0.9591, "step": 21179 }, { "epoch": 0.02, "learning_rate": 4.994518756790965e-05, "loss": 0.9519, "step": 21180 }, { "epoch": 0.02, "learning_rate": 4.9945182369267166e-05, "loss": 0.8666, "step": 21181 }, { "epoch": 0.02, "learning_rate": 4.994517717037844e-05, "loss": 1.0334, "step": 21182 }, { "epoch": 0.02, "learning_rate": 4.9945171971243476e-05, "loss": 0.8654, "step": 21183 }, { "epoch": 0.02, "learning_rate": 4.9945166771862254e-05, "loss": 0.85, "step": 21184 }, { "epoch": 0.02, "learning_rate": 4.994516157223479e-05, "loss": 0.9523, "step": 21185 }, { "epoch": 0.02, "learning_rate": 4.994515637236107e-05, "loss": 0.8205, "step": 21186 }, { "epoch": 0.02, "learning_rate": 4.994515117224111e-05, "loss": 0.9746, "step": 21187 }, { "epoch": 0.02, "learning_rate": 4.99451459718749e-05, "loss": 0.9048, "step": 21188 }, { "epoch": 0.02, "learning_rate": 4.994514077126244e-05, "loss": 0.8659, "step": 21189 }, { "epoch": 0.02, "learning_rate": 4.9945135570403736e-05, "loss": 0.9844, "step": 21190 }, { "epoch": 0.02, "learning_rate": 4.9945130369298775e-05, "loss": 0.8789, "step": 21191 }, { "epoch": 0.02, "learning_rate": 4.994512516794757e-05, "loss": 0.7788, "step": 21192 }, { "epoch": 0.02, "learning_rate": 4.994511996635012e-05, "loss": 0.9424, "step": 21193 }, { "epoch": 0.02, "learning_rate": 4.994511476450642e-05, "loss": 0.8791, "step": 21194 }, { "epoch": 0.02, "learning_rate": 4.9945109562416475e-05, "loss": 0.9264, "step": 21195 }, { "epoch": 0.02, "learning_rate": 4.994510436008029e-05, "loss": 0.9351, "step": 21196 }, { "epoch": 0.02, "learning_rate": 4.994509915749784e-05, "loss": 0.8969, "step": 21197 }, { "epoch": 0.02, "learning_rate": 4.9945093954669156e-05, "loss": 0.8645, "step": 21198 }, { "epoch": 0.02, "learning_rate": 4.994508875159421e-05, "loss": 0.8501, "step": 21199 }, { "epoch": 0.02, "learning_rate": 4.994508354827303e-05, "loss": 0.8759, "step": 21200 }, { "epoch": 0.02, "learning_rate": 4.99450783447056e-05, "loss": 1.0809, "step": 21201 }, { "epoch": 0.02, "learning_rate": 4.994507314089192e-05, "loss": 1.0119, "step": 21202 }, { "epoch": 0.02, "learning_rate": 4.9945067936832e-05, "loss": 0.9412, "step": 21203 }, { "epoch": 0.02, "learning_rate": 4.994506273252583e-05, "loss": 0.9274, "step": 21204 }, { "epoch": 0.02, "learning_rate": 4.99450575279734e-05, "loss": 0.9335, "step": 21205 }, { "epoch": 0.02, "learning_rate": 4.9945052323174734e-05, "loss": 0.8489, "step": 21206 }, { "epoch": 0.02, "learning_rate": 4.994504711812982e-05, "loss": 0.8045, "step": 21207 }, { "epoch": 0.02, "learning_rate": 4.994504191283866e-05, "loss": 0.9587, "step": 21208 }, { "epoch": 0.02, "learning_rate": 4.994503670730125e-05, "loss": 0.8599, "step": 21209 }, { "epoch": 0.02, "learning_rate": 4.99450315015176e-05, "loss": 0.7227, "step": 21210 }, { "epoch": 0.02, "learning_rate": 4.9945026295487696e-05, "loss": 0.8593, "step": 21211 }, { "epoch": 0.02, "learning_rate": 4.9945021089211544e-05, "loss": 0.3204, "step": 21212 }, { "epoch": 0.02, "learning_rate": 4.994501588268915e-05, "loss": 0.9194, "step": 21213 }, { "epoch": 0.02, "learning_rate": 4.994501067592051e-05, "loss": 1.0652, "step": 21214 }, { "epoch": 0.02, "learning_rate": 4.994500546890562e-05, "loss": 1.0129, "step": 21215 }, { "epoch": 0.02, "learning_rate": 4.994500026164449e-05, "loss": 0.8645, "step": 21216 }, { "epoch": 0.02, "learning_rate": 4.99449950541371e-05, "loss": 0.9588, "step": 21217 }, { "epoch": 0.02, "learning_rate": 4.994498984638347e-05, "loss": 1.009, "step": 21218 }, { "epoch": 0.02, "learning_rate": 4.99449846383836e-05, "loss": 0.9509, "step": 21219 }, { "epoch": 0.02, "learning_rate": 4.994497943013748e-05, "loss": 0.9426, "step": 21220 }, { "epoch": 0.02, "learning_rate": 4.9944974221645116e-05, "loss": 0.8943, "step": 21221 }, { "epoch": 0.02, "learning_rate": 4.99449690129065e-05, "loss": 0.9066, "step": 21222 }, { "epoch": 0.02, "learning_rate": 4.994496380392164e-05, "loss": 1.6434, "step": 21223 }, { "epoch": 0.02, "learning_rate": 4.994495859469053e-05, "loss": 0.8702, "step": 21224 }, { "epoch": 0.02, "learning_rate": 4.994495338521318e-05, "loss": 0.918, "step": 21225 }, { "epoch": 0.02, "learning_rate": 4.994494817548958e-05, "loss": 0.8908, "step": 21226 }, { "epoch": 0.02, "learning_rate": 4.9944942965519736e-05, "loss": 0.8662, "step": 21227 }, { "epoch": 0.02, "learning_rate": 4.994493775530365e-05, "loss": 0.9552, "step": 21228 }, { "epoch": 0.02, "learning_rate": 4.994493254484131e-05, "loss": 1.226, "step": 21229 }, { "epoch": 0.02, "learning_rate": 4.994492733413273e-05, "loss": 1.2979, "step": 21230 }, { "epoch": 0.02, "learning_rate": 4.99449221231779e-05, "loss": 1.03, "step": 21231 }, { "epoch": 0.02, "learning_rate": 4.994491691197683e-05, "loss": 1.143, "step": 21232 }, { "epoch": 0.02, "learning_rate": 4.9944911700529506e-05, "loss": 0.8046, "step": 21233 }, { "epoch": 0.02, "learning_rate": 4.994490648883594e-05, "loss": 1.08, "step": 21234 }, { "epoch": 0.02, "learning_rate": 4.9944901276896134e-05, "loss": 0.9966, "step": 21235 }, { "epoch": 0.02, "learning_rate": 4.994489606471008e-05, "loss": 0.5478, "step": 21236 }, { "epoch": 0.02, "learning_rate": 4.994489085227778e-05, "loss": 0.4291, "step": 21237 }, { "epoch": 0.02, "learning_rate": 4.994488563959922e-05, "loss": 1.1664, "step": 21238 }, { "epoch": 0.02, "learning_rate": 4.994488042667443e-05, "loss": 0.7748, "step": 21239 }, { "epoch": 0.02, "learning_rate": 4.99448752135034e-05, "loss": 1.1067, "step": 21240 }, { "epoch": 0.02, "learning_rate": 4.9944870000086114e-05, "loss": 1.3567, "step": 21241 }, { "epoch": 0.02, "learning_rate": 4.994486478642259e-05, "loss": 0.7858, "step": 21242 }, { "epoch": 0.02, "learning_rate": 4.994485957251281e-05, "loss": 0.9862, "step": 21243 }, { "epoch": 0.02, "learning_rate": 4.9944854358356795e-05, "loss": 0.9686, "step": 21244 }, { "epoch": 0.02, "learning_rate": 4.994484914395453e-05, "loss": 1.1012, "step": 21245 }, { "epoch": 0.02, "learning_rate": 4.994484392930602e-05, "loss": 0.9071, "step": 21246 }, { "epoch": 0.02, "learning_rate": 4.994483871441127e-05, "loss": 0.9202, "step": 21247 }, { "epoch": 0.02, "learning_rate": 4.994483349927027e-05, "loss": 0.7442, "step": 21248 }, { "epoch": 0.02, "learning_rate": 4.9944828283883024e-05, "loss": 0.7375, "step": 21249 }, { "epoch": 0.02, "learning_rate": 4.994482306824954e-05, "loss": 0.6829, "step": 21250 }, { "epoch": 0.02, "learning_rate": 4.99448178523698e-05, "loss": 0.8778, "step": 21251 }, { "epoch": 0.02, "learning_rate": 4.994481263624382e-05, "loss": 0.9076, "step": 21252 }, { "epoch": 0.02, "learning_rate": 4.9944807419871605e-05, "loss": 0.782, "step": 21253 }, { "epoch": 0.02, "learning_rate": 4.994480220325314e-05, "loss": 0.7479, "step": 21254 }, { "epoch": 0.02, "learning_rate": 4.994479698638842e-05, "loss": 0.7892, "step": 21255 }, { "epoch": 0.02, "learning_rate": 4.994479176927746e-05, "loss": 0.8916, "step": 21256 }, { "epoch": 0.02, "learning_rate": 4.994478655192026e-05, "loss": 0.8197, "step": 21257 }, { "epoch": 0.02, "learning_rate": 4.994478133431682e-05, "loss": 0.5922, "step": 21258 }, { "epoch": 0.02, "learning_rate": 4.994477611646713e-05, "loss": 0.6607, "step": 21259 }, { "epoch": 0.02, "learning_rate": 4.9944770898371195e-05, "loss": 0.5339, "step": 21260 }, { "epoch": 0.02, "learning_rate": 4.994476568002902e-05, "loss": 0.6611, "step": 21261 }, { "epoch": 0.02, "learning_rate": 4.99447604614406e-05, "loss": 0.6938, "step": 21262 }, { "epoch": 0.02, "learning_rate": 4.994475524260593e-05, "loss": 0.7736, "step": 21263 }, { "epoch": 0.02, "learning_rate": 4.9944750023525025e-05, "loss": 0.9005, "step": 21264 }, { "epoch": 0.02, "learning_rate": 4.9944744804197864e-05, "loss": 0.6891, "step": 21265 }, { "epoch": 0.02, "learning_rate": 4.994473958462447e-05, "loss": 0.6536, "step": 21266 }, { "epoch": 0.02, "learning_rate": 4.994473436480482e-05, "loss": 0.4722, "step": 21267 }, { "epoch": 0.02, "learning_rate": 4.994472914473893e-05, "loss": 0.7118, "step": 21268 }, { "epoch": 0.02, "learning_rate": 4.994472392442681e-05, "loss": 1.021, "step": 21269 }, { "epoch": 0.02, "learning_rate": 4.9944718703868434e-05, "loss": 0.9973, "step": 21270 }, { "epoch": 0.02, "learning_rate": 4.994471348306381e-05, "loss": 1.0374, "step": 21271 }, { "epoch": 0.02, "learning_rate": 4.994470826201295e-05, "loss": 1.1298, "step": 21272 }, { "epoch": 0.02, "learning_rate": 4.994470304071585e-05, "loss": 0.8766, "step": 21273 }, { "epoch": 0.02, "learning_rate": 4.994469781917249e-05, "loss": 0.9596, "step": 21274 }, { "epoch": 0.02, "learning_rate": 4.99446925973829e-05, "loss": 1.1949, "step": 21275 }, { "epoch": 0.02, "learning_rate": 4.9944687375347066e-05, "loss": 0.8763, "step": 21276 }, { "epoch": 0.02, "learning_rate": 4.9944682153064987e-05, "loss": 0.7254, "step": 21277 }, { "epoch": 0.02, "learning_rate": 4.9944676930536665e-05, "loss": 0.3957, "step": 21278 }, { "epoch": 0.02, "learning_rate": 4.99446717077621e-05, "loss": 0.2739, "step": 21279 }, { "epoch": 0.02, "learning_rate": 4.9944666484741286e-05, "loss": 0.1371, "step": 21280 }, { "epoch": 0.02, "learning_rate": 4.9944661261474236e-05, "loss": 0.1692, "step": 21281 }, { "epoch": 0.02, "learning_rate": 4.994465603796094e-05, "loss": 0.1512, "step": 21282 }, { "epoch": 0.02, "learning_rate": 4.99446508142014e-05, "loss": 0.0791, "step": 21283 }, { "epoch": 0.02, "learning_rate": 4.9944645590195616e-05, "loss": 0.0448, "step": 21284 }, { "epoch": 0.02, "learning_rate": 4.9944640365943596e-05, "loss": 0.07, "step": 21285 }, { "epoch": 0.02, "learning_rate": 4.9944635141445326e-05, "loss": 1.1381, "step": 21286 }, { "epoch": 0.02, "learning_rate": 4.994462991670081e-05, "loss": 0.8569, "step": 21287 }, { "epoch": 0.02, "learning_rate": 4.9944624691710065e-05, "loss": 1.1895, "step": 21288 }, { "epoch": 0.02, "learning_rate": 4.994461946647307e-05, "loss": 1.3355, "step": 21289 }, { "epoch": 0.02, "learning_rate": 4.9944614240989826e-05, "loss": 1.318, "step": 21290 }, { "epoch": 0.02, "learning_rate": 4.994460901526034e-05, "loss": 0.9115, "step": 21291 }, { "epoch": 0.02, "learning_rate": 4.994460378928462e-05, "loss": 1.0316, "step": 21292 }, { "epoch": 0.02, "learning_rate": 4.994459856306265e-05, "loss": 1.1097, "step": 21293 }, { "epoch": 0.02, "learning_rate": 4.994459333659444e-05, "loss": 1.0897, "step": 21294 }, { "epoch": 0.02, "learning_rate": 4.9944588109879984e-05, "loss": 1.1458, "step": 21295 }, { "epoch": 0.02, "learning_rate": 4.994458288291929e-05, "loss": 0.9417, "step": 21296 }, { "epoch": 0.02, "learning_rate": 4.9944577655712355e-05, "loss": 1.0377, "step": 21297 }, { "epoch": 0.02, "learning_rate": 4.994457242825917e-05, "loss": 1.9576, "step": 21298 }, { "epoch": 0.02, "learning_rate": 4.9944567200559756e-05, "loss": 1.3086, "step": 21299 }, { "epoch": 0.02, "learning_rate": 4.994456197261409e-05, "loss": 1.1091, "step": 21300 }, { "epoch": 0.02, "learning_rate": 4.994455674442218e-05, "loss": 1.067, "step": 21301 }, { "epoch": 0.02, "learning_rate": 4.994455151598403e-05, "loss": 1.6588, "step": 21302 }, { "epoch": 0.02, "learning_rate": 4.9944546287299646e-05, "loss": 1.4143, "step": 21303 }, { "epoch": 0.02, "learning_rate": 4.994454105836901e-05, "loss": 1.1955, "step": 21304 }, { "epoch": 0.02, "learning_rate": 4.994453582919213e-05, "loss": 1.0721, "step": 21305 }, { "epoch": 0.02, "learning_rate": 4.994453059976901e-05, "loss": 0.9138, "step": 21306 }, { "epoch": 0.02, "learning_rate": 4.994452537009966e-05, "loss": 0.7385, "step": 21307 }, { "epoch": 0.02, "learning_rate": 4.994452014018406e-05, "loss": 0.9718, "step": 21308 }, { "epoch": 0.02, "learning_rate": 4.994451491002221e-05, "loss": 0.5418, "step": 21309 }, { "epoch": 0.02, "learning_rate": 4.994450967961413e-05, "loss": 0.8399, "step": 21310 }, { "epoch": 0.02, "learning_rate": 4.9944504448959805e-05, "loss": 0.9506, "step": 21311 }, { "epoch": 0.02, "learning_rate": 4.994449921805924e-05, "loss": 0.9537, "step": 21312 }, { "epoch": 0.02, "learning_rate": 4.9944493986912426e-05, "loss": 0.9808, "step": 21313 }, { "epoch": 0.02, "learning_rate": 4.9944488755519375e-05, "loss": 1.3362, "step": 21314 }, { "epoch": 0.02, "learning_rate": 4.994448352388008e-05, "loss": 1.1291, "step": 21315 }, { "epoch": 0.02, "learning_rate": 4.994447829199455e-05, "loss": 1.224, "step": 21316 }, { "epoch": 0.02, "learning_rate": 4.9944473059862775e-05, "loss": 1.323, "step": 21317 }, { "epoch": 0.02, "learning_rate": 4.994446782748476e-05, "loss": 1.2361, "step": 21318 }, { "epoch": 0.02, "learning_rate": 4.9944462594860504e-05, "loss": 1.1337, "step": 21319 }, { "epoch": 0.02, "learning_rate": 4.9944457361990004e-05, "loss": 1.1519, "step": 21320 }, { "epoch": 0.02, "learning_rate": 4.994445212887326e-05, "loss": 0.569, "step": 21321 }, { "epoch": 0.02, "learning_rate": 4.9944446895510284e-05, "loss": 1.0727, "step": 21322 }, { "epoch": 0.02, "learning_rate": 4.994444166190106e-05, "loss": 1.3189, "step": 21323 }, { "epoch": 0.02, "learning_rate": 4.9944436428045594e-05, "loss": 0.9177, "step": 21324 }, { "epoch": 0.02, "learning_rate": 4.9944431193943894e-05, "loss": 0.7438, "step": 21325 }, { "epoch": 0.02, "learning_rate": 4.9944425959595946e-05, "loss": 0.4636, "step": 21326 }, { "epoch": 0.02, "learning_rate": 4.994442072500176e-05, "loss": 1.1284, "step": 21327 }, { "epoch": 0.02, "learning_rate": 4.9944415490161334e-05, "loss": 1.3429, "step": 21328 }, { "epoch": 0.02, "learning_rate": 4.994441025507467e-05, "loss": 1.1576, "step": 21329 }, { "epoch": 0.02, "learning_rate": 4.994440501974176e-05, "loss": 1.081, "step": 21330 }, { "epoch": 0.02, "learning_rate": 4.994439978416261e-05, "loss": 1.3083, "step": 21331 }, { "epoch": 0.02, "learning_rate": 4.994439454833722e-05, "loss": 1.1263, "step": 21332 }, { "epoch": 0.02, "learning_rate": 4.994438931226559e-05, "loss": 1.2432, "step": 21333 }, { "epoch": 0.02, "learning_rate": 4.994438407594772e-05, "loss": 0.8692, "step": 21334 }, { "epoch": 0.02, "learning_rate": 4.994437883938361e-05, "loss": 0.9011, "step": 21335 }, { "epoch": 0.02, "learning_rate": 4.994437360257326e-05, "loss": 0.619, "step": 21336 }, { "epoch": 0.02, "learning_rate": 4.994436836551667e-05, "loss": 1.0444, "step": 21337 }, { "epoch": 0.02, "learning_rate": 4.9944363128213833e-05, "loss": 1.13, "step": 21338 }, { "epoch": 0.02, "learning_rate": 4.9944357890664764e-05, "loss": 1.2477, "step": 21339 }, { "epoch": 0.02, "learning_rate": 4.994435265286945e-05, "loss": 1.5688, "step": 21340 }, { "epoch": 0.02, "learning_rate": 4.99443474148279e-05, "loss": 1.1908, "step": 21341 }, { "epoch": 0.02, "learning_rate": 4.99443421765401e-05, "loss": 1.0464, "step": 21342 }, { "epoch": 0.02, "learning_rate": 4.9944336938006075e-05, "loss": 0.8116, "step": 21343 }, { "epoch": 0.02, "learning_rate": 4.99443316992258e-05, "loss": 1.2522, "step": 21344 }, { "epoch": 0.02, "learning_rate": 4.994432646019929e-05, "loss": 1.6597, "step": 21345 }, { "epoch": 0.02, "learning_rate": 4.9944321220926535e-05, "loss": 0.7934, "step": 21346 }, { "epoch": 0.02, "learning_rate": 4.994431598140754e-05, "loss": 0.7231, "step": 21347 }, { "epoch": 0.02, "learning_rate": 4.9944310741642306e-05, "loss": 0.8085, "step": 21348 }, { "epoch": 0.02, "learning_rate": 4.994430550163084e-05, "loss": 1.2041, "step": 21349 }, { "epoch": 0.02, "learning_rate": 4.9944300261373126e-05, "loss": 0.9433, "step": 21350 }, { "epoch": 0.02, "learning_rate": 4.994429502086917e-05, "loss": 0.2579, "step": 21351 }, { "epoch": 0.02, "learning_rate": 4.994428978011898e-05, "loss": 1.0741, "step": 21352 }, { "epoch": 0.02, "learning_rate": 4.994428453912256e-05, "loss": 0.961, "step": 21353 }, { "epoch": 0.02, "learning_rate": 4.994427929787989e-05, "loss": 0.3312, "step": 21354 }, { "epoch": 0.02, "learning_rate": 4.994427405639098e-05, "loss": 1.118, "step": 21355 }, { "epoch": 0.02, "learning_rate": 4.9944268814655826e-05, "loss": 1.3152, "step": 21356 }, { "epoch": 0.02, "learning_rate": 4.994426357267444e-05, "loss": 1.1632, "step": 21357 }, { "epoch": 0.02, "learning_rate": 4.994425833044681e-05, "loss": 0.8464, "step": 21358 }, { "epoch": 0.02, "learning_rate": 4.994425308797295e-05, "loss": 1.1619, "step": 21359 }, { "epoch": 0.02, "learning_rate": 4.994424784525285e-05, "loss": 0.9892, "step": 21360 }, { "epoch": 0.02, "learning_rate": 4.9944242602286494e-05, "loss": 1.1089, "step": 21361 }, { "epoch": 0.02, "learning_rate": 4.994423735907391e-05, "loss": 1.0009, "step": 21362 }, { "epoch": 0.02, "learning_rate": 4.9944232115615094e-05, "loss": 0.9223, "step": 21363 }, { "epoch": 0.02, "learning_rate": 4.994422687191003e-05, "loss": 1.1184, "step": 21364 }, { "epoch": 0.02, "learning_rate": 4.994422162795873e-05, "loss": 1.1963, "step": 21365 }, { "epoch": 0.02, "learning_rate": 4.994421638376119e-05, "loss": 1.1248, "step": 21366 }, { "epoch": 0.02, "learning_rate": 4.994421113931741e-05, "loss": 1.0048, "step": 21367 }, { "epoch": 0.02, "learning_rate": 4.994420589462739e-05, "loss": 0.6783, "step": 21368 }, { "epoch": 0.02, "learning_rate": 4.994420064969114e-05, "loss": 1.1281, "step": 21369 }, { "epoch": 0.02, "learning_rate": 4.994419540450865e-05, "loss": 1.1271, "step": 21370 }, { "epoch": 0.02, "learning_rate": 4.994419015907992e-05, "loss": 0.8529, "step": 21371 }, { "epoch": 0.02, "learning_rate": 4.994418491340494e-05, "loss": 1.0933, "step": 21372 }, { "epoch": 0.02, "learning_rate": 4.994417966748374e-05, "loss": 1.3416, "step": 21373 }, { "epoch": 0.02, "learning_rate": 4.9944174421316286e-05, "loss": 0.6393, "step": 21374 }, { "epoch": 0.02, "learning_rate": 4.99441691749026e-05, "loss": 0.9645, "step": 21375 }, { "epoch": 0.02, "learning_rate": 4.994416392824267e-05, "loss": 1.0948, "step": 21376 }, { "epoch": 0.02, "learning_rate": 4.994415868133651e-05, "loss": 1.3385, "step": 21377 }, { "epoch": 0.02, "learning_rate": 4.994415343418411e-05, "loss": 1.0865, "step": 21378 }, { "epoch": 0.02, "learning_rate": 4.994414818678547e-05, "loss": 1.1815, "step": 21379 }, { "epoch": 0.02, "learning_rate": 4.9944142939140594e-05, "loss": 1.3544, "step": 21380 }, { "epoch": 0.02, "learning_rate": 4.9944137691249476e-05, "loss": 1.1664, "step": 21381 }, { "epoch": 0.02, "learning_rate": 4.994413244311212e-05, "loss": 1.2454, "step": 21382 }, { "epoch": 0.02, "learning_rate": 4.9944127194728527e-05, "loss": 1.1638, "step": 21383 }, { "epoch": 0.02, "learning_rate": 4.99441219460987e-05, "loss": 1.2228, "step": 21384 }, { "epoch": 0.02, "learning_rate": 4.994411669722263e-05, "loss": 0.6998, "step": 21385 }, { "epoch": 0.02, "learning_rate": 4.994411144810033e-05, "loss": 0.9502, "step": 21386 }, { "epoch": 0.02, "learning_rate": 4.994410619873179e-05, "loss": 1.0736, "step": 21387 }, { "epoch": 0.02, "learning_rate": 4.994410094911701e-05, "loss": 1.1984, "step": 21388 }, { "epoch": 0.02, "learning_rate": 4.994409569925599e-05, "loss": 1.1979, "step": 21389 }, { "epoch": 0.02, "learning_rate": 4.994409044914873e-05, "loss": 0.9893, "step": 21390 }, { "epoch": 0.02, "learning_rate": 4.9944085198795245e-05, "loss": 1.0458, "step": 21391 }, { "epoch": 0.02, "learning_rate": 4.994407994819551e-05, "loss": 0.8051, "step": 21392 }, { "epoch": 0.02, "learning_rate": 4.9944074697349544e-05, "loss": 0.941, "step": 21393 }, { "epoch": 0.02, "learning_rate": 4.994406944625733e-05, "loss": 1.0936, "step": 21394 }, { "epoch": 0.02, "learning_rate": 4.994406419491889e-05, "loss": 1.3327, "step": 21395 }, { "epoch": 0.02, "learning_rate": 4.994405894333421e-05, "loss": 1.1432, "step": 21396 }, { "epoch": 0.02, "learning_rate": 4.994405369150329e-05, "loss": 1.0211, "step": 21397 }, { "epoch": 0.02, "learning_rate": 4.994404843942614e-05, "loss": 1.0848, "step": 21398 }, { "epoch": 0.02, "learning_rate": 4.994404318710275e-05, "loss": 1.3489, "step": 21399 }, { "epoch": 0.02, "learning_rate": 4.9944037934533114e-05, "loss": 1.2683, "step": 21400 }, { "epoch": 0.02, "learning_rate": 4.994403268171725e-05, "loss": 1.0208, "step": 21401 }, { "epoch": 0.02, "learning_rate": 4.994402742865515e-05, "loss": 1.1155, "step": 21402 }, { "epoch": 0.02, "learning_rate": 4.9944022175346806e-05, "loss": 1.0715, "step": 21403 }, { "epoch": 0.02, "learning_rate": 4.994401692179223e-05, "loss": 1.2818, "step": 21404 }, { "epoch": 0.02, "learning_rate": 4.994401166799142e-05, "loss": 1.1193, "step": 21405 }, { "epoch": 0.02, "learning_rate": 4.994400641394437e-05, "loss": 1.0586, "step": 21406 }, { "epoch": 0.02, "learning_rate": 4.994400115965108e-05, "loss": 0.8666, "step": 21407 }, { "epoch": 0.02, "learning_rate": 4.994399590511155e-05, "loss": 1.2661, "step": 21408 }, { "epoch": 0.02, "learning_rate": 4.9943990650325796e-05, "loss": 0.9074, "step": 21409 }, { "epoch": 0.02, "learning_rate": 4.99439853952938e-05, "loss": 0.8745, "step": 21410 }, { "epoch": 0.02, "learning_rate": 4.994398014001557e-05, "loss": 1.0076, "step": 21411 }, { "epoch": 0.02, "learning_rate": 4.994397488449109e-05, "loss": 1.0941, "step": 21412 }, { "epoch": 0.02, "learning_rate": 4.9943969628720387e-05, "loss": 1.1989, "step": 21413 }, { "epoch": 0.02, "learning_rate": 4.9943964372703445e-05, "loss": 1.2982, "step": 21414 }, { "epoch": 0.02, "learning_rate": 4.994395911644027e-05, "loss": 1.0213, "step": 21415 }, { "epoch": 0.02, "learning_rate": 4.9943953859930846e-05, "loss": 1.0046, "step": 21416 }, { "epoch": 0.02, "learning_rate": 4.99439486031752e-05, "loss": 1.0804, "step": 21417 }, { "epoch": 0.02, "learning_rate": 4.994394334617331e-05, "loss": 1.077, "step": 21418 }, { "epoch": 0.02, "learning_rate": 4.994393808892519e-05, "loss": 1.0331, "step": 21419 }, { "epoch": 0.02, "learning_rate": 4.9943932831430826e-05, "loss": 1.1398, "step": 21420 }, { "epoch": 0.02, "learning_rate": 4.994392757369023e-05, "loss": 0.9519, "step": 21421 }, { "epoch": 0.02, "learning_rate": 4.99439223157034e-05, "loss": 0.9076, "step": 21422 }, { "epoch": 0.02, "learning_rate": 4.994391705747033e-05, "loss": 1.0135, "step": 21423 }, { "epoch": 0.02, "learning_rate": 4.9943911798991025e-05, "loss": 1.2365, "step": 21424 }, { "epoch": 0.02, "learning_rate": 4.994390654026549e-05, "loss": 1.1807, "step": 21425 }, { "epoch": 0.02, "learning_rate": 4.994390128129372e-05, "loss": 0.921, "step": 21426 }, { "epoch": 0.02, "learning_rate": 4.99438960220757e-05, "loss": 1.1299, "step": 21427 }, { "epoch": 0.02, "learning_rate": 4.994389076261146e-05, "loss": 0.8181, "step": 21428 }, { "epoch": 0.02, "learning_rate": 4.994388550290098e-05, "loss": 1.1476, "step": 21429 }, { "epoch": 0.02, "learning_rate": 4.9943880242944255e-05, "loss": 1.2777, "step": 21430 }, { "epoch": 0.02, "learning_rate": 4.9943874982741304e-05, "loss": 1.1631, "step": 21431 }, { "epoch": 0.02, "learning_rate": 4.9943869722292116e-05, "loss": 1.1281, "step": 21432 }, { "epoch": 0.02, "learning_rate": 4.994386446159669e-05, "loss": 1.5449, "step": 21433 }, { "epoch": 0.02, "learning_rate": 4.9943859200655027e-05, "loss": 1.1031, "step": 21434 }, { "epoch": 0.02, "learning_rate": 4.994385393946714e-05, "loss": 1.1158, "step": 21435 }, { "epoch": 0.02, "learning_rate": 4.9943848678033014e-05, "loss": 1.04, "step": 21436 }, { "epoch": 0.02, "learning_rate": 4.994384341635264e-05, "loss": 0.9918, "step": 21437 }, { "epoch": 0.02, "learning_rate": 4.994383815442605e-05, "loss": 1.0878, "step": 21438 }, { "epoch": 0.02, "learning_rate": 4.994383289225321e-05, "loss": 1.1274, "step": 21439 }, { "epoch": 0.02, "learning_rate": 4.994382762983414e-05, "loss": 1.1882, "step": 21440 }, { "epoch": 0.02, "learning_rate": 4.9943822367168835e-05, "loss": 1.0224, "step": 21441 }, { "epoch": 0.02, "learning_rate": 4.9943817104257295e-05, "loss": 1.0642, "step": 21442 }, { "epoch": 0.02, "learning_rate": 4.994381184109952e-05, "loss": 1.0736, "step": 21443 }, { "epoch": 0.02, "learning_rate": 4.994380657769551e-05, "loss": 1.015, "step": 21444 }, { "epoch": 0.02, "learning_rate": 4.994380131404527e-05, "loss": 1.1421, "step": 21445 }, { "epoch": 0.02, "learning_rate": 4.9943796050148786e-05, "loss": 0.9571, "step": 21446 }, { "epoch": 0.02, "learning_rate": 4.9943790786006074e-05, "loss": 1.1894, "step": 21447 }, { "epoch": 0.02, "learning_rate": 4.9943785521617126e-05, "loss": 1.0957, "step": 21448 }, { "epoch": 0.02, "learning_rate": 4.994378025698194e-05, "loss": 0.5327, "step": 21449 }, { "epoch": 0.02, "learning_rate": 4.994377499210053e-05, "loss": 1.0091, "step": 21450 }, { "epoch": 0.02, "learning_rate": 4.994376972697288e-05, "loss": 1.0431, "step": 21451 }, { "epoch": 0.02, "learning_rate": 4.994376446159899e-05, "loss": 1.2426, "step": 21452 }, { "epoch": 0.02, "learning_rate": 4.994375919597887e-05, "loss": 1.1802, "step": 21453 }, { "epoch": 0.02, "learning_rate": 4.994375393011251e-05, "loss": 0.9929, "step": 21454 }, { "epoch": 0.02, "learning_rate": 4.994374866399992e-05, "loss": 1.2309, "step": 21455 }, { "epoch": 0.02, "learning_rate": 4.9943743397641106e-05, "loss": 1.2299, "step": 21456 }, { "epoch": 0.02, "learning_rate": 4.994373813103604e-05, "loss": 1.1763, "step": 21457 }, { "epoch": 0.02, "learning_rate": 4.9943732864184756e-05, "loss": 1.1846, "step": 21458 }, { "epoch": 0.02, "learning_rate": 4.994372759708723e-05, "loss": 1.2441, "step": 21459 }, { "epoch": 0.02, "learning_rate": 4.994372232974347e-05, "loss": 1.2999, "step": 21460 }, { "epoch": 0.02, "learning_rate": 4.9943717062153476e-05, "loss": 0.8639, "step": 21461 }, { "epoch": 0.02, "learning_rate": 4.9943711794317246e-05, "loss": 0.8381, "step": 21462 }, { "epoch": 0.02, "learning_rate": 4.994370652623479e-05, "loss": 1.1953, "step": 21463 }, { "epoch": 0.02, "learning_rate": 4.994370125790609e-05, "loss": 1.1012, "step": 21464 }, { "epoch": 0.02, "learning_rate": 4.994369598933117e-05, "loss": 1.0871, "step": 21465 }, { "epoch": 0.02, "learning_rate": 4.994369072051e-05, "loss": 0.8221, "step": 21466 }, { "epoch": 0.02, "learning_rate": 4.9943685451442614e-05, "loss": 1.1163, "step": 21467 }, { "epoch": 0.02, "learning_rate": 4.994368018212898e-05, "loss": 1.0373, "step": 21468 }, { "epoch": 0.02, "learning_rate": 4.9943674912569116e-05, "loss": 1.0757, "step": 21469 }, { "epoch": 0.02, "learning_rate": 4.9943669642763027e-05, "loss": 0.9568, "step": 21470 }, { "epoch": 0.02, "learning_rate": 4.9943664372710695e-05, "loss": 1.0369, "step": 21471 }, { "epoch": 0.02, "learning_rate": 4.994365910241213e-05, "loss": 1.0439, "step": 21472 }, { "epoch": 0.02, "learning_rate": 4.994365383186733e-05, "loss": 1.3323, "step": 21473 }, { "epoch": 0.02, "learning_rate": 4.9943648561076303e-05, "loss": 2.0937, "step": 21474 }, { "epoch": 0.02, "learning_rate": 4.994364329003904e-05, "loss": 1.8081, "step": 21475 }, { "epoch": 0.02, "learning_rate": 4.9943638018755544e-05, "loss": 1.2279, "step": 21476 }, { "epoch": 0.02, "learning_rate": 4.994363274722582e-05, "loss": 1.3869, "step": 21477 }, { "epoch": 0.02, "learning_rate": 4.994362747544986e-05, "loss": 1.3084, "step": 21478 }, { "epoch": 0.02, "learning_rate": 4.994362220342766e-05, "loss": 1.1648, "step": 21479 }, { "epoch": 0.02, "learning_rate": 4.9943616931159234e-05, "loss": 1.0312, "step": 21480 }, { "epoch": 0.02, "learning_rate": 4.994361165864457e-05, "loss": 0.9607, "step": 21481 }, { "epoch": 0.02, "learning_rate": 4.994360638588368e-05, "loss": 1.1924, "step": 21482 }, { "epoch": 0.02, "learning_rate": 4.9943601112876556e-05, "loss": 1.3385, "step": 21483 }, { "epoch": 0.02, "learning_rate": 4.99435958396232e-05, "loss": 1.1178, "step": 21484 }, { "epoch": 0.02, "learning_rate": 4.994359056612361e-05, "loss": 1.0333, "step": 21485 }, { "epoch": 0.02, "learning_rate": 4.994358529237778e-05, "loss": 0.9613, "step": 21486 }, { "epoch": 0.02, "learning_rate": 4.994358001838573e-05, "loss": 0.8821, "step": 21487 }, { "epoch": 0.02, "learning_rate": 4.9943574744147436e-05, "loss": 0.778, "step": 21488 }, { "epoch": 0.02, "learning_rate": 4.994356946966291e-05, "loss": 0.8237, "step": 21489 }, { "epoch": 0.02, "learning_rate": 4.994356419493216e-05, "loss": 1.2051, "step": 21490 }, { "epoch": 0.02, "learning_rate": 4.994355891995517e-05, "loss": 1.2017, "step": 21491 }, { "epoch": 0.02, "learning_rate": 4.9943553644731954e-05, "loss": 1.2945, "step": 21492 }, { "epoch": 0.02, "learning_rate": 4.994354836926251e-05, "loss": 1.2878, "step": 21493 }, { "epoch": 0.02, "learning_rate": 4.9943543093546825e-05, "loss": 1.1788, "step": 21494 }, { "epoch": 0.02, "learning_rate": 4.994353781758491e-05, "loss": 0.9545, "step": 21495 }, { "epoch": 0.02, "learning_rate": 4.994353254137676e-05, "loss": 1.2799, "step": 21496 }, { "epoch": 0.02, "learning_rate": 4.994352726492238e-05, "loss": 0.9301, "step": 21497 }, { "epoch": 0.02, "learning_rate": 4.994352198822177e-05, "loss": 0.8946, "step": 21498 }, { "epoch": 0.02, "learning_rate": 4.994351671127493e-05, "loss": 0.99, "step": 21499 }, { "epoch": 0.02, "learning_rate": 4.994351143408185e-05, "loss": 1.1911, "step": 21500 }, { "epoch": 0.02, "eval_loss": 1.0204905271530151, "eval_runtime": 82.6423, "eval_samples_per_second": 16.759, "eval_steps_per_second": 4.199, "step": 21500 }, { "epoch": 0.02, "learning_rate": 4.9943506156642545e-05, "loss": 0.8441, "step": 21501 }, { "epoch": 0.02, "learning_rate": 4.9943500878957e-05, "loss": 0.92, "step": 21502 }, { "epoch": 0.02, "learning_rate": 4.994349560102524e-05, "loss": 1.5358, "step": 21503 }, { "epoch": 0.02, "learning_rate": 4.994349032284723e-05, "loss": 1.3599, "step": 21504 }, { "epoch": 0.02, "learning_rate": 4.9943485044423e-05, "loss": 1.368, "step": 21505 }, { "epoch": 0.02, "learning_rate": 4.994347976575253e-05, "loss": 0.9406, "step": 21506 }, { "epoch": 0.02, "learning_rate": 4.994347448683584e-05, "loss": 1.2045, "step": 21507 }, { "epoch": 0.02, "learning_rate": 4.994346920767291e-05, "loss": 1.2324, "step": 21508 }, { "epoch": 0.02, "learning_rate": 4.9943463928263745e-05, "loss": 1.2825, "step": 21509 }, { "epoch": 0.02, "learning_rate": 4.9943458648608356e-05, "loss": 1.3985, "step": 21510 }, { "epoch": 0.02, "learning_rate": 4.994345336870674e-05, "loss": 1.1605, "step": 21511 }, { "epoch": 0.02, "learning_rate": 4.994344808855888e-05, "loss": 0.8541, "step": 21512 }, { "epoch": 0.02, "learning_rate": 4.9943442808164796e-05, "loss": 1.2308, "step": 21513 }, { "epoch": 0.02, "learning_rate": 4.9943437527524484e-05, "loss": 1.1285, "step": 21514 }, { "epoch": 0.02, "learning_rate": 4.994343224663794e-05, "loss": 1.236, "step": 21515 }, { "epoch": 0.02, "learning_rate": 4.994342696550516e-05, "loss": 0.999, "step": 21516 }, { "epoch": 0.02, "learning_rate": 4.994342168412615e-05, "loss": 0.7226, "step": 21517 }, { "epoch": 0.02, "learning_rate": 4.9943416402500906e-05, "loss": 0.5034, "step": 21518 }, { "epoch": 0.02, "learning_rate": 4.9943411120629435e-05, "loss": 0.744, "step": 21519 }, { "epoch": 0.02, "learning_rate": 4.994340583851173e-05, "loss": 0.6641, "step": 21520 }, { "epoch": 0.02, "learning_rate": 4.99434005561478e-05, "loss": 1.3566, "step": 21521 }, { "epoch": 0.02, "learning_rate": 4.9943395273537635e-05, "loss": 1.2232, "step": 21522 }, { "epoch": 0.02, "learning_rate": 4.994338999068124e-05, "loss": 1.0332, "step": 21523 }, { "epoch": 0.02, "learning_rate": 4.994338470757862e-05, "loss": 1.7143, "step": 21524 }, { "epoch": 0.02, "learning_rate": 4.9943379424229766e-05, "loss": 1.3351, "step": 21525 }, { "epoch": 0.02, "learning_rate": 4.994337414063467e-05, "loss": 1.1161, "step": 21526 }, { "epoch": 0.02, "learning_rate": 4.994336885679336e-05, "loss": 1.1575, "step": 21527 }, { "epoch": 0.02, "learning_rate": 4.9943363572705815e-05, "loss": 1.1081, "step": 21528 }, { "epoch": 0.02, "learning_rate": 4.9943358288372034e-05, "loss": 0.7882, "step": 21529 }, { "epoch": 0.02, "learning_rate": 4.994335300379203e-05, "loss": 1.0401, "step": 21530 }, { "epoch": 0.02, "learning_rate": 4.9943347718965796e-05, "loss": 1.0647, "step": 21531 }, { "epoch": 0.02, "learning_rate": 4.994334243389333e-05, "loss": 1.048, "step": 21532 }, { "epoch": 0.02, "learning_rate": 4.994333714857462e-05, "loss": 1.0985, "step": 21533 }, { "epoch": 0.02, "learning_rate": 4.9943331863009694e-05, "loss": 0.6156, "step": 21534 }, { "epoch": 0.02, "learning_rate": 4.994332657719854e-05, "loss": 1.0441, "step": 21535 }, { "epoch": 0.02, "learning_rate": 4.9943321291141146e-05, "loss": 1.1048, "step": 21536 }, { "epoch": 0.02, "learning_rate": 4.994331600483753e-05, "loss": 0.5624, "step": 21537 }, { "epoch": 0.02, "learning_rate": 4.994331071828768e-05, "loss": 0.113, "step": 21538 }, { "epoch": 0.02, "learning_rate": 4.994330543149161e-05, "loss": 0.0646, "step": 21539 }, { "epoch": 0.02, "learning_rate": 4.994330014444929e-05, "loss": 0.0759, "step": 21540 }, { "epoch": 0.02, "learning_rate": 4.994329485716076e-05, "loss": 0.1033, "step": 21541 }, { "epoch": 0.02, "learning_rate": 4.9943289569625994e-05, "loss": 0.153, "step": 21542 }, { "epoch": 0.02, "learning_rate": 4.994328428184499e-05, "loss": 0.4491, "step": 21543 }, { "epoch": 0.02, "learning_rate": 4.994327899381777e-05, "loss": 0.947, "step": 21544 }, { "epoch": 0.02, "learning_rate": 4.994327370554431e-05, "loss": 1.0054, "step": 21545 }, { "epoch": 0.02, "learning_rate": 4.994326841702462e-05, "loss": 0.9873, "step": 21546 }, { "epoch": 0.02, "learning_rate": 4.994326312825871e-05, "loss": 1.0444, "step": 21547 }, { "epoch": 0.02, "learning_rate": 4.994325783924657e-05, "loss": 1.0811, "step": 21548 }, { "epoch": 0.02, "learning_rate": 4.99432525499882e-05, "loss": 1.2559, "step": 21549 }, { "epoch": 0.02, "learning_rate": 4.994324726048359e-05, "loss": 1.1211, "step": 21550 }, { "epoch": 0.02, "learning_rate": 4.9943241970732757e-05, "loss": 1.2957, "step": 21551 }, { "epoch": 0.02, "learning_rate": 4.994323668073569e-05, "loss": 1.2087, "step": 21552 }, { "epoch": 0.02, "learning_rate": 4.9943231390492405e-05, "loss": 0.7734, "step": 21553 }, { "epoch": 0.02, "learning_rate": 4.994322610000289e-05, "loss": 1.0999, "step": 21554 }, { "epoch": 0.02, "learning_rate": 4.994322080926714e-05, "loss": 0.8471, "step": 21555 }, { "epoch": 0.02, "learning_rate": 4.9943215518285157e-05, "loss": 1.0168, "step": 21556 }, { "epoch": 0.02, "learning_rate": 4.994321022705696e-05, "loss": 0.9685, "step": 21557 }, { "epoch": 0.02, "learning_rate": 4.994320493558252e-05, "loss": 1.2167, "step": 21558 }, { "epoch": 0.02, "learning_rate": 4.994319964386185e-05, "loss": 0.9677, "step": 21559 }, { "epoch": 0.02, "learning_rate": 4.994319435189496e-05, "loss": 1.1441, "step": 21560 }, { "epoch": 0.02, "learning_rate": 4.9943189059681836e-05, "loss": 0.9623, "step": 21561 }, { "epoch": 0.02, "learning_rate": 4.994318376722249e-05, "loss": 1.0047, "step": 21562 }, { "epoch": 0.02, "learning_rate": 4.994317847451691e-05, "loss": 1.1001, "step": 21563 }, { "epoch": 0.02, "learning_rate": 4.99431731815651e-05, "loss": 1.2923, "step": 21564 }, { "epoch": 0.02, "learning_rate": 4.9943167888367074e-05, "loss": 1.0494, "step": 21565 }, { "epoch": 0.02, "learning_rate": 4.99431625949228e-05, "loss": 0.7959, "step": 21566 }, { "epoch": 0.02, "learning_rate": 4.9943157301232315e-05, "loss": 0.9435, "step": 21567 }, { "epoch": 0.02, "learning_rate": 4.994315200729559e-05, "loss": 1.0393, "step": 21568 }, { "epoch": 0.02, "learning_rate": 4.9943146713112646e-05, "loss": 0.9602, "step": 21569 }, { "epoch": 0.02, "learning_rate": 4.994314141868346e-05, "loss": 1.1554, "step": 21570 }, { "epoch": 0.02, "learning_rate": 4.9943136124008054e-05, "loss": 1.2093, "step": 21571 }, { "epoch": 0.02, "learning_rate": 4.994313082908643e-05, "loss": 1.3478, "step": 21572 }, { "epoch": 0.02, "learning_rate": 4.994312553391856e-05, "loss": 0.9962, "step": 21573 }, { "epoch": 0.02, "learning_rate": 4.994312023850447e-05, "loss": 1.2062, "step": 21574 }, { "epoch": 0.02, "learning_rate": 4.994311494284415e-05, "loss": 1.3346, "step": 21575 }, { "epoch": 0.02, "learning_rate": 4.99431096469376e-05, "loss": 1.4078, "step": 21576 }, { "epoch": 0.02, "learning_rate": 4.994310435078483e-05, "loss": 1.3449, "step": 21577 }, { "epoch": 0.02, "learning_rate": 4.9943099054385834e-05, "loss": 1.2704, "step": 21578 }, { "epoch": 0.02, "learning_rate": 4.99430937577406e-05, "loss": 1.1559, "step": 21579 }, { "epoch": 0.02, "learning_rate": 4.994308846084914e-05, "loss": 1.3784, "step": 21580 }, { "epoch": 0.02, "learning_rate": 4.994308316371146e-05, "loss": 1.2439, "step": 21581 }, { "epoch": 0.02, "learning_rate": 4.994307786632755e-05, "loss": 1.356, "step": 21582 }, { "epoch": 0.02, "learning_rate": 4.994307256869741e-05, "loss": 1.3247, "step": 21583 }, { "epoch": 0.02, "learning_rate": 4.994306727082104e-05, "loss": 1.3048, "step": 21584 }, { "epoch": 0.02, "learning_rate": 4.994306197269844e-05, "loss": 1.1362, "step": 21585 }, { "epoch": 0.02, "learning_rate": 4.9943056674329623e-05, "loss": 1.1682, "step": 21586 }, { "epoch": 0.02, "learning_rate": 4.9943051375714575e-05, "loss": 1.0061, "step": 21587 }, { "epoch": 0.02, "learning_rate": 4.99430460768533e-05, "loss": 1.3011, "step": 21588 }, { "epoch": 0.02, "learning_rate": 4.994304077774579e-05, "loss": 1.6349, "step": 21589 }, { "epoch": 0.02, "learning_rate": 4.994303547839206e-05, "loss": 0.9376, "step": 21590 }, { "epoch": 0.02, "learning_rate": 4.99430301787921e-05, "loss": 1.2217, "step": 21591 }, { "epoch": 0.02, "learning_rate": 4.994302487894592e-05, "loss": 1.2279, "step": 21592 }, { "epoch": 0.02, "learning_rate": 4.9943019578853505e-05, "loss": 1.4965, "step": 21593 }, { "epoch": 0.02, "learning_rate": 4.9943014278514866e-05, "loss": 0.8856, "step": 21594 }, { "epoch": 0.02, "learning_rate": 4.994300897793e-05, "loss": 1.0858, "step": 21595 }, { "epoch": 0.02, "learning_rate": 4.994300367709891e-05, "loss": 1.3529, "step": 21596 }, { "epoch": 0.02, "learning_rate": 4.994299837602159e-05, "loss": 1.1038, "step": 21597 }, { "epoch": 0.02, "learning_rate": 4.994299307469804e-05, "loss": 1.4503, "step": 21598 }, { "epoch": 0.02, "learning_rate": 4.994298777312827e-05, "loss": 1.052, "step": 21599 }, { "epoch": 0.02, "learning_rate": 4.994298247131226e-05, "loss": 1.2115, "step": 21600 }, { "epoch": 0.02, "learning_rate": 4.994297716925004e-05, "loss": 1.1085, "step": 21601 }, { "epoch": 0.02, "learning_rate": 4.994297186694158e-05, "loss": 0.9663, "step": 21602 }, { "epoch": 0.02, "learning_rate": 4.994296656438691e-05, "loss": 1.2905, "step": 21603 }, { "epoch": 0.02, "learning_rate": 4.9942961261586e-05, "loss": 0.8859, "step": 21604 }, { "epoch": 0.02, "learning_rate": 4.994295595853886e-05, "loss": 0.8585, "step": 21605 }, { "epoch": 0.02, "learning_rate": 4.99429506552455e-05, "loss": 1.0194, "step": 21606 }, { "epoch": 0.02, "learning_rate": 4.994294535170592e-05, "loss": 1.0628, "step": 21607 }, { "epoch": 0.02, "learning_rate": 4.9942940047920105e-05, "loss": 1.1757, "step": 21608 }, { "epoch": 0.02, "learning_rate": 4.994293474388807e-05, "loss": 0.7947, "step": 21609 }, { "epoch": 0.02, "learning_rate": 4.994292943960981e-05, "loss": 0.7889, "step": 21610 }, { "epoch": 0.02, "learning_rate": 4.9942924135085314e-05, "loss": 1.1843, "step": 21611 }, { "epoch": 0.02, "learning_rate": 4.99429188303146e-05, "loss": 1.0943, "step": 21612 }, { "epoch": 0.02, "learning_rate": 4.9942913525297655e-05, "loss": 1.1487, "step": 21613 }, { "epoch": 0.02, "learning_rate": 4.994290822003449e-05, "loss": 1.2974, "step": 21614 }, { "epoch": 0.02, "learning_rate": 4.99429029145251e-05, "loss": 1.2003, "step": 21615 }, { "epoch": 0.02, "learning_rate": 4.994289760876948e-05, "loss": 1.1206, "step": 21616 }, { "epoch": 0.02, "learning_rate": 4.994289230276763e-05, "loss": 1.2136, "step": 21617 }, { "epoch": 0.02, "learning_rate": 4.994288699651956e-05, "loss": 1.2015, "step": 21618 }, { "epoch": 0.02, "learning_rate": 4.994288169002526e-05, "loss": 1.1269, "step": 21619 }, { "epoch": 0.02, "learning_rate": 4.994287638328474e-05, "loss": 1.1119, "step": 21620 }, { "epoch": 0.02, "learning_rate": 4.994287107629799e-05, "loss": 1.1216, "step": 21621 }, { "epoch": 0.02, "learning_rate": 4.994286576906502e-05, "loss": 1.0699, "step": 21622 }, { "epoch": 0.02, "learning_rate": 4.994286046158582e-05, "loss": 1.1917, "step": 21623 }, { "epoch": 0.02, "learning_rate": 4.9942855153860393e-05, "loss": 0.9727, "step": 21624 }, { "epoch": 0.02, "learning_rate": 4.994284984588875e-05, "loss": 0.7442, "step": 21625 }, { "epoch": 0.02, "learning_rate": 4.9942844537670866e-05, "loss": 0.9184, "step": 21626 }, { "epoch": 0.02, "learning_rate": 4.9942839229206776e-05, "loss": 1.0461, "step": 21627 }, { "epoch": 0.02, "learning_rate": 4.994283392049644e-05, "loss": 1.2155, "step": 21628 }, { "epoch": 0.02, "learning_rate": 4.9942828611539894e-05, "loss": 1.134, "step": 21629 }, { "epoch": 0.02, "learning_rate": 4.9942823302337116e-05, "loss": 1.0879, "step": 21630 }, { "epoch": 0.02, "learning_rate": 4.9942817992888116e-05, "loss": 1.0368, "step": 21631 }, { "epoch": 0.02, "learning_rate": 4.994281268319289e-05, "loss": 1.164, "step": 21632 }, { "epoch": 0.02, "learning_rate": 4.9942807373251436e-05, "loss": 1.1019, "step": 21633 }, { "epoch": 0.02, "learning_rate": 4.994280206306376e-05, "loss": 1.0738, "step": 21634 }, { "epoch": 0.02, "learning_rate": 4.994279675262986e-05, "loss": 1.8306, "step": 21635 }, { "epoch": 0.02, "learning_rate": 4.994279144194974e-05, "loss": 0.9836, "step": 21636 }, { "epoch": 0.02, "learning_rate": 4.994278613102339e-05, "loss": 0.9996, "step": 21637 }, { "epoch": 0.02, "learning_rate": 4.994278081985081e-05, "loss": 1.356, "step": 21638 }, { "epoch": 0.02, "learning_rate": 4.994277550843202e-05, "loss": 0.9421, "step": 21639 }, { "epoch": 0.02, "learning_rate": 4.9942770196766984e-05, "loss": 0.975, "step": 21640 }, { "epoch": 0.02, "learning_rate": 4.994276488485574e-05, "loss": 1.3124, "step": 21641 }, { "epoch": 0.02, "learning_rate": 4.994275957269827e-05, "loss": 1.0569, "step": 21642 }, { "epoch": 0.02, "learning_rate": 4.9942754260294566e-05, "loss": 1.079, "step": 21643 }, { "epoch": 0.02, "learning_rate": 4.994274894764465e-05, "loss": 1.3087, "step": 21644 }, { "epoch": 0.02, "learning_rate": 4.99427436347485e-05, "loss": 1.3121, "step": 21645 }, { "epoch": 0.02, "learning_rate": 4.994273832160613e-05, "loss": 1.0195, "step": 21646 }, { "epoch": 0.02, "learning_rate": 4.9942733008217537e-05, "loss": 1.1651, "step": 21647 }, { "epoch": 0.02, "learning_rate": 4.9942727694582717e-05, "loss": 0.9677, "step": 21648 }, { "epoch": 0.02, "learning_rate": 4.994272238070168e-05, "loss": 0.334, "step": 21649 }, { "epoch": 0.02, "learning_rate": 4.994271706657441e-05, "loss": 0.6542, "step": 21650 }, { "epoch": 0.02, "learning_rate": 4.994271175220092e-05, "loss": 0.851, "step": 21651 }, { "epoch": 0.02, "learning_rate": 4.994270643758121e-05, "loss": 1.1128, "step": 21652 }, { "epoch": 0.02, "learning_rate": 4.994270112271526e-05, "loss": 1.0196, "step": 21653 }, { "epoch": 0.02, "learning_rate": 4.99426958076031e-05, "loss": 1.1413, "step": 21654 }, { "epoch": 0.02, "learning_rate": 4.994269049224471e-05, "loss": 1.0522, "step": 21655 }, { "epoch": 0.02, "learning_rate": 4.9942685176640106e-05, "loss": 1.2215, "step": 21656 }, { "epoch": 0.02, "learning_rate": 4.994267986078928e-05, "loss": 1.7623, "step": 21657 }, { "epoch": 0.02, "learning_rate": 4.9942674544692215e-05, "loss": 0.7826, "step": 21658 }, { "epoch": 0.02, "learning_rate": 4.9942669228348936e-05, "loss": 1.0889, "step": 21659 }, { "epoch": 0.02, "learning_rate": 4.994266391175943e-05, "loss": 1.3004, "step": 21660 }, { "epoch": 0.02, "learning_rate": 4.9942658594923705e-05, "loss": 1.1324, "step": 21661 }, { "epoch": 0.02, "learning_rate": 4.994265327784176e-05, "loss": 1.0685, "step": 21662 }, { "epoch": 0.02, "learning_rate": 4.994264796051358e-05, "loss": 1.2731, "step": 21663 }, { "epoch": 0.02, "learning_rate": 4.994264264293918e-05, "loss": 1.0818, "step": 21664 }, { "epoch": 0.02, "learning_rate": 4.994263732511856e-05, "loss": 1.0772, "step": 21665 }, { "epoch": 0.02, "learning_rate": 4.994263200705172e-05, "loss": 1.3476, "step": 21666 }, { "epoch": 0.02, "learning_rate": 4.994262668873865e-05, "loss": 1.0928, "step": 21667 }, { "epoch": 0.02, "learning_rate": 4.9942621370179354e-05, "loss": 0.9148, "step": 21668 }, { "epoch": 0.02, "learning_rate": 4.9942616051373845e-05, "loss": 1.0421, "step": 21669 }, { "epoch": 0.02, "learning_rate": 4.994261073232211e-05, "loss": 1.1773, "step": 21670 }, { "epoch": 0.02, "learning_rate": 4.994260541302415e-05, "loss": 1.0953, "step": 21671 }, { "epoch": 0.02, "learning_rate": 4.9942600093479964e-05, "loss": 0.9524, "step": 21672 }, { "epoch": 0.02, "learning_rate": 4.994259477368955e-05, "loss": 1.3443, "step": 21673 }, { "epoch": 0.02, "learning_rate": 4.994258945365293e-05, "loss": 0.9674, "step": 21674 }, { "epoch": 0.02, "learning_rate": 4.994258413337008e-05, "loss": 1.0573, "step": 21675 }, { "epoch": 0.02, "learning_rate": 4.9942578812841004e-05, "loss": 1.2206, "step": 21676 }, { "epoch": 0.02, "learning_rate": 4.994257349206571e-05, "loss": 1.0628, "step": 21677 }, { "epoch": 0.02, "learning_rate": 4.994256817104419e-05, "loss": 0.6475, "step": 21678 }, { "epoch": 0.02, "learning_rate": 4.9942562849776454e-05, "loss": 0.4589, "step": 21679 }, { "epoch": 0.02, "learning_rate": 4.994255752826249e-05, "loss": 0.7265, "step": 21680 }, { "epoch": 0.02, "learning_rate": 4.99425522065023e-05, "loss": 0.5564, "step": 21681 }, { "epoch": 0.02, "learning_rate": 4.994254688449589e-05, "loss": 1.2558, "step": 21682 }, { "epoch": 0.02, "learning_rate": 4.994254156224326e-05, "loss": 0.8174, "step": 21683 }, { "epoch": 0.02, "learning_rate": 4.9942536239744405e-05, "loss": 0.7358, "step": 21684 }, { "epoch": 0.02, "learning_rate": 4.994253091699933e-05, "loss": 1.1193, "step": 21685 }, { "epoch": 0.02, "learning_rate": 4.994252559400804e-05, "loss": 1.0256, "step": 21686 }, { "epoch": 0.02, "learning_rate": 4.9942520270770514e-05, "loss": 1.0101, "step": 21687 }, { "epoch": 0.02, "learning_rate": 4.9942514947286774e-05, "loss": 1.1176, "step": 21688 }, { "epoch": 0.02, "learning_rate": 4.994250962355681e-05, "loss": 1.1421, "step": 21689 }, { "epoch": 0.02, "learning_rate": 4.994250429958063e-05, "loss": 1.1085, "step": 21690 }, { "epoch": 0.02, "learning_rate": 4.994249897535822e-05, "loss": 1.0204, "step": 21691 }, { "epoch": 0.02, "learning_rate": 4.9942493650889586e-05, "loss": 1.0841, "step": 21692 }, { "epoch": 0.02, "learning_rate": 4.994248832617474e-05, "loss": 1.118, "step": 21693 }, { "epoch": 0.02, "learning_rate": 4.994248300121366e-05, "loss": 0.8442, "step": 21694 }, { "epoch": 0.02, "learning_rate": 4.994247767600637e-05, "loss": 1.0817, "step": 21695 }, { "epoch": 0.02, "learning_rate": 4.9942472350552854e-05, "loss": 1.0163, "step": 21696 }, { "epoch": 0.02, "learning_rate": 4.9942467024853114e-05, "loss": 1.0365, "step": 21697 }, { "epoch": 0.02, "learning_rate": 4.994246169890716e-05, "loss": 1.2203, "step": 21698 }, { "epoch": 0.02, "learning_rate": 4.994245637271497e-05, "loss": 1.3005, "step": 21699 }, { "epoch": 0.02, "learning_rate": 4.994245104627657e-05, "loss": 1.1164, "step": 21700 }, { "epoch": 0.02, "learning_rate": 4.994244571959195e-05, "loss": 1.271, "step": 21701 }, { "epoch": 0.02, "learning_rate": 4.9942440392661104e-05, "loss": 0.9653, "step": 21702 }, { "epoch": 0.02, "learning_rate": 4.9942435065484036e-05, "loss": 1.1133, "step": 21703 }, { "epoch": 0.02, "learning_rate": 4.994242973806075e-05, "loss": 1.2871, "step": 21704 }, { "epoch": 0.02, "learning_rate": 4.994242441039124e-05, "loss": 1.3675, "step": 21705 }, { "epoch": 0.02, "learning_rate": 4.9942419082475513e-05, "loss": 1.1945, "step": 21706 }, { "epoch": 0.02, "learning_rate": 4.9942413754313563e-05, "loss": 1.0927, "step": 21707 }, { "epoch": 0.02, "learning_rate": 4.99424084259054e-05, "loss": 1.2139, "step": 21708 }, { "epoch": 0.02, "learning_rate": 4.9942403097251e-05, "loss": 1.1224, "step": 21709 }, { "epoch": 0.02, "learning_rate": 4.994239776835039e-05, "loss": 1.9707, "step": 21710 }, { "epoch": 0.02, "learning_rate": 4.9942392439203554e-05, "loss": 1.7993, "step": 21711 }, { "epoch": 0.02, "learning_rate": 4.994238710981049e-05, "loss": 1.3211, "step": 21712 }, { "epoch": 0.02, "learning_rate": 4.994238178017122e-05, "loss": 1.331, "step": 21713 }, { "epoch": 0.02, "learning_rate": 4.994237645028572e-05, "loss": 1.0083, "step": 21714 }, { "epoch": 0.02, "learning_rate": 4.994237112015401e-05, "loss": 1.0146, "step": 21715 }, { "epoch": 0.02, "learning_rate": 4.994236578977607e-05, "loss": 1.0937, "step": 21716 }, { "epoch": 0.02, "learning_rate": 4.994236045915192e-05, "loss": 1.0813, "step": 21717 }, { "epoch": 0.02, "learning_rate": 4.9942355128281536e-05, "loss": 0.9742, "step": 21718 }, { "epoch": 0.02, "learning_rate": 4.994234979716493e-05, "loss": 0.9272, "step": 21719 }, { "epoch": 0.02, "learning_rate": 4.994234446580212e-05, "loss": 1.0401, "step": 21720 }, { "epoch": 0.02, "learning_rate": 4.994233913419307e-05, "loss": 1.0828, "step": 21721 }, { "epoch": 0.02, "learning_rate": 4.994233380233781e-05, "loss": 1.1581, "step": 21722 }, { "epoch": 0.02, "learning_rate": 4.994232847023633e-05, "loss": 1.1312, "step": 21723 }, { "epoch": 0.02, "learning_rate": 4.9942323137888637e-05, "loss": 1.1615, "step": 21724 }, { "epoch": 0.02, "learning_rate": 4.994231780529471e-05, "loss": 1.0344, "step": 21725 }, { "epoch": 0.02, "learning_rate": 4.994231247245458e-05, "loss": 1.169, "step": 21726 }, { "epoch": 0.02, "learning_rate": 4.9942307139368216e-05, "loss": 1.1597, "step": 21727 }, { "epoch": 0.02, "learning_rate": 4.994230180603563e-05, "loss": 0.9474, "step": 21728 }, { "epoch": 0.02, "learning_rate": 4.994229647245683e-05, "loss": 0.9169, "step": 21729 }, { "epoch": 0.02, "learning_rate": 4.9942291138631816e-05, "loss": 1.0681, "step": 21730 }, { "epoch": 0.02, "learning_rate": 4.994228580456057e-05, "loss": 1.2147, "step": 21731 }, { "epoch": 0.02, "learning_rate": 4.994228047024311e-05, "loss": 1.0963, "step": 21732 }, { "epoch": 0.02, "learning_rate": 4.9942275135679434e-05, "loss": 0.9662, "step": 21733 }, { "epoch": 0.02, "learning_rate": 4.9942269800869536e-05, "loss": 1.252, "step": 21734 }, { "epoch": 0.02, "learning_rate": 4.9942264465813416e-05, "loss": 0.84, "step": 21735 }, { "epoch": 0.02, "learning_rate": 4.994225913051108e-05, "loss": 0.9815, "step": 21736 }, { "epoch": 0.02, "learning_rate": 4.994225379496252e-05, "loss": 0.9686, "step": 21737 }, { "epoch": 0.02, "learning_rate": 4.994224845916774e-05, "loss": 1.1927, "step": 21738 }, { "epoch": 0.02, "learning_rate": 4.9942243123126744e-05, "loss": 1.1757, "step": 21739 }, { "epoch": 0.02, "learning_rate": 4.994223778683953e-05, "loss": 0.9197, "step": 21740 }, { "epoch": 0.02, "learning_rate": 4.99422324503061e-05, "loss": 1.1095, "step": 21741 }, { "epoch": 0.02, "learning_rate": 4.994222711352644e-05, "loss": 1.3057, "step": 21742 }, { "epoch": 0.02, "learning_rate": 4.994222177650057e-05, "loss": 1.1465, "step": 21743 }, { "epoch": 0.02, "learning_rate": 4.9942216439228475e-05, "loss": 1.0029, "step": 21744 }, { "epoch": 0.02, "learning_rate": 4.9942211101710166e-05, "loss": 1.141, "step": 21745 }, { "epoch": 0.02, "learning_rate": 4.994220576394564e-05, "loss": 1.1373, "step": 21746 }, { "epoch": 0.02, "learning_rate": 4.994220042593488e-05, "loss": 1.1457, "step": 21747 }, { "epoch": 0.02, "learning_rate": 4.9942195087677916e-05, "loss": 1.0692, "step": 21748 }, { "epoch": 0.02, "learning_rate": 4.994218974917473e-05, "loss": 1.2658, "step": 21749 }, { "epoch": 0.02, "learning_rate": 4.994218441042532e-05, "loss": 1.1392, "step": 21750 }, { "epoch": 0.02, "learning_rate": 4.994217907142969e-05, "loss": 0.8615, "step": 21751 }, { "epoch": 0.02, "learning_rate": 4.994217373218786e-05, "loss": 0.7384, "step": 21752 }, { "epoch": 0.02, "learning_rate": 4.994216839269979e-05, "loss": 0.6546, "step": 21753 }, { "epoch": 0.02, "learning_rate": 4.9942163052965515e-05, "loss": 0.7958, "step": 21754 }, { "epoch": 0.02, "learning_rate": 4.9942157712985016e-05, "loss": 1.1952, "step": 21755 }, { "epoch": 0.02, "learning_rate": 4.99421523727583e-05, "loss": 1.3958, "step": 21756 }, { "epoch": 0.02, "learning_rate": 4.994214703228536e-05, "loss": 1.0041, "step": 21757 }, { "epoch": 0.02, "learning_rate": 4.9942141691566206e-05, "loss": 1.0627, "step": 21758 }, { "epoch": 0.02, "learning_rate": 4.994213635060083e-05, "loss": 1.2193, "step": 21759 }, { "epoch": 0.02, "learning_rate": 4.994213100938924e-05, "loss": 3.3144, "step": 21760 }, { "epoch": 0.02, "learning_rate": 4.9942125667931435e-05, "loss": 1.3413, "step": 21761 }, { "epoch": 0.02, "learning_rate": 4.9942120326227407e-05, "loss": 1.0732, "step": 21762 }, { "epoch": 0.02, "learning_rate": 4.994211498427716e-05, "loss": 1.3135, "step": 21763 }, { "epoch": 0.02, "learning_rate": 4.99421096420807e-05, "loss": 1.0917, "step": 21764 }, { "epoch": 0.02, "learning_rate": 4.9942104299638015e-05, "loss": 1.5893, "step": 21765 }, { "epoch": 0.02, "learning_rate": 4.994209895694912e-05, "loss": 1.214, "step": 21766 }, { "epoch": 0.02, "learning_rate": 4.9942093614014e-05, "loss": 1.0506, "step": 21767 }, { "epoch": 0.02, "learning_rate": 4.9942088270832664e-05, "loss": 1.128, "step": 21768 }, { "epoch": 0.02, "learning_rate": 4.9942082927405107e-05, "loss": 1.293, "step": 21769 }, { "epoch": 0.02, "learning_rate": 4.994207758373134e-05, "loss": 1.1546, "step": 21770 }, { "epoch": 0.02, "learning_rate": 4.994207223981135e-05, "loss": 1.1068, "step": 21771 }, { "epoch": 0.02, "learning_rate": 4.994206689564515e-05, "loss": 0.9333, "step": 21772 }, { "epoch": 0.02, "learning_rate": 4.994206155123272e-05, "loss": 1.0025, "step": 21773 }, { "epoch": 0.02, "learning_rate": 4.994205620657408e-05, "loss": 1.0137, "step": 21774 }, { "epoch": 0.02, "learning_rate": 4.994205086166922e-05, "loss": 1.2749, "step": 21775 }, { "epoch": 0.02, "learning_rate": 4.994204551651815e-05, "loss": 1.4387, "step": 21776 }, { "epoch": 0.02, "learning_rate": 4.994204017112085e-05, "loss": 1.3789, "step": 21777 }, { "epoch": 0.02, "learning_rate": 4.9942034825477345e-05, "loss": 0.2956, "step": 21778 }, { "epoch": 0.02, "learning_rate": 4.994202947958762e-05, "loss": 0.6192, "step": 21779 }, { "epoch": 0.02, "learning_rate": 4.994202413345167e-05, "loss": 0.6336, "step": 21780 }, { "epoch": 0.02, "learning_rate": 4.994201878706951e-05, "loss": 0.3367, "step": 21781 }, { "epoch": 0.02, "learning_rate": 4.9942013440441126e-05, "loss": 0.2924, "step": 21782 }, { "epoch": 0.02, "learning_rate": 4.994200809356653e-05, "loss": 0.2864, "step": 21783 }, { "epoch": 0.02, "learning_rate": 4.994200274644572e-05, "loss": 0.2752, "step": 21784 }, { "epoch": 0.02, "learning_rate": 4.994199739907869e-05, "loss": 0.8606, "step": 21785 }, { "epoch": 0.02, "learning_rate": 4.9941992051465446e-05, "loss": 1.2449, "step": 21786 }, { "epoch": 0.02, "learning_rate": 4.994198670360598e-05, "loss": 1.3306, "step": 21787 }, { "epoch": 0.02, "learning_rate": 4.99419813555003e-05, "loss": 1.2214, "step": 21788 }, { "epoch": 0.02, "learning_rate": 4.99419760071484e-05, "loss": 1.2383, "step": 21789 }, { "epoch": 0.02, "learning_rate": 4.994197065855028e-05, "loss": 1.1973, "step": 21790 }, { "epoch": 0.02, "learning_rate": 4.994196530970596e-05, "loss": 0.9796, "step": 21791 }, { "epoch": 0.02, "learning_rate": 4.994195996061541e-05, "loss": 1.0425, "step": 21792 }, { "epoch": 0.02, "learning_rate": 4.9941954611278645e-05, "loss": 0.9307, "step": 21793 }, { "epoch": 0.02, "learning_rate": 4.9941949261695665e-05, "loss": 0.971, "step": 21794 }, { "epoch": 0.02, "learning_rate": 4.994194391186647e-05, "loss": 1.1617, "step": 21795 }, { "epoch": 0.02, "learning_rate": 4.994193856179106e-05, "loss": 1.0951, "step": 21796 }, { "epoch": 0.02, "learning_rate": 4.9941933211469426e-05, "loss": 1.3445, "step": 21797 }, { "epoch": 0.02, "learning_rate": 4.9941927860901585e-05, "loss": 0.943, "step": 21798 }, { "epoch": 0.02, "learning_rate": 4.994192251008752e-05, "loss": 1.1043, "step": 21799 }, { "epoch": 0.02, "learning_rate": 4.994191715902724e-05, "loss": 1.3865, "step": 21800 }, { "epoch": 0.02, "learning_rate": 4.9941911807720746e-05, "loss": 1.0918, "step": 21801 }, { "epoch": 0.02, "learning_rate": 4.9941906456168035e-05, "loss": 1.1059, "step": 21802 }, { "epoch": 0.02, "learning_rate": 4.994190110436911e-05, "loss": 0.9877, "step": 21803 }, { "epoch": 0.02, "learning_rate": 4.994189575232396e-05, "loss": 1.2638, "step": 21804 }, { "epoch": 0.02, "learning_rate": 4.994189040003261e-05, "loss": 1.3432, "step": 21805 }, { "epoch": 0.02, "learning_rate": 4.994188504749504e-05, "loss": 3.0931, "step": 21806 }, { "epoch": 0.02, "learning_rate": 4.994187969471125e-05, "loss": 1.3332, "step": 21807 }, { "epoch": 0.02, "learning_rate": 4.994187434168124e-05, "loss": 1.0825, "step": 21808 }, { "epoch": 0.02, "learning_rate": 4.994186898840502e-05, "loss": 0.9153, "step": 21809 }, { "epoch": 0.02, "learning_rate": 4.994186363488258e-05, "loss": 1.1746, "step": 21810 }, { "epoch": 0.02, "learning_rate": 4.994185828111393e-05, "loss": 0.9023, "step": 21811 }, { "epoch": 0.02, "learning_rate": 4.994185292709906e-05, "loss": 1.0312, "step": 21812 }, { "epoch": 0.02, "learning_rate": 4.994184757283798e-05, "loss": 1.0223, "step": 21813 }, { "epoch": 0.02, "learning_rate": 4.994184221833068e-05, "loss": 1.2973, "step": 21814 }, { "epoch": 0.02, "learning_rate": 4.994183686357717e-05, "loss": 1.2954, "step": 21815 }, { "epoch": 0.02, "learning_rate": 4.994183150857744e-05, "loss": 0.8376, "step": 21816 }, { "epoch": 0.02, "learning_rate": 4.99418261533315e-05, "loss": 1.1644, "step": 21817 }, { "epoch": 0.02, "learning_rate": 4.994182079783933e-05, "loss": 1.2632, "step": 21818 }, { "epoch": 0.02, "learning_rate": 4.994181544210096e-05, "loss": 0.9657, "step": 21819 }, { "epoch": 0.02, "learning_rate": 4.994181008611637e-05, "loss": 0.8824, "step": 21820 }, { "epoch": 0.02, "learning_rate": 4.994180472988557e-05, "loss": 1.0467, "step": 21821 }, { "epoch": 0.02, "learning_rate": 4.9941799373408546e-05, "loss": 0.9279, "step": 21822 }, { "epoch": 0.02, "learning_rate": 4.99417940166853e-05, "loss": 0.3203, "step": 21823 }, { "epoch": 0.02, "learning_rate": 4.994178865971586e-05, "loss": 0.4087, "step": 21824 }, { "epoch": 0.02, "learning_rate": 4.9941783302500196e-05, "loss": 0.4972, "step": 21825 }, { "epoch": 0.02, "learning_rate": 4.994177794503831e-05, "loss": 0.459, "step": 21826 }, { "epoch": 0.02, "learning_rate": 4.994177258733022e-05, "loss": 0.4399, "step": 21827 }, { "epoch": 0.02, "learning_rate": 4.994176722937591e-05, "loss": 0.446, "step": 21828 }, { "epoch": 0.02, "learning_rate": 4.994176187117539e-05, "loss": 1.1619, "step": 21829 }, { "epoch": 0.02, "learning_rate": 4.994175651272866e-05, "loss": 1.3143, "step": 21830 }, { "epoch": 0.02, "learning_rate": 4.9941751154035695e-05, "loss": 0.9979, "step": 21831 }, { "epoch": 0.02, "learning_rate": 4.9941745795096525e-05, "loss": 1.2747, "step": 21832 }, { "epoch": 0.02, "learning_rate": 4.994174043591115e-05, "loss": 1.079, "step": 21833 }, { "epoch": 0.02, "learning_rate": 4.994173507647956e-05, "loss": 1.125, "step": 21834 }, { "epoch": 0.02, "learning_rate": 4.994172971680174e-05, "loss": 0.9689, "step": 21835 }, { "epoch": 0.02, "learning_rate": 4.994172435687772e-05, "loss": 1.1055, "step": 21836 }, { "epoch": 0.02, "learning_rate": 4.994171899670748e-05, "loss": 1.2078, "step": 21837 }, { "epoch": 0.02, "learning_rate": 4.9941713636291034e-05, "loss": 1.015, "step": 21838 }, { "epoch": 0.02, "learning_rate": 4.994170827562836e-05, "loss": 1.3581, "step": 21839 }, { "epoch": 0.02, "learning_rate": 4.994170291471948e-05, "loss": 0.8426, "step": 21840 }, { "epoch": 0.02, "learning_rate": 4.9941697553564385e-05, "loss": 0.6546, "step": 21841 }, { "epoch": 0.02, "learning_rate": 4.994169219216308e-05, "loss": 0.5401, "step": 21842 }, { "epoch": 0.02, "learning_rate": 4.994168683051555e-05, "loss": 0.8876, "step": 21843 }, { "epoch": 0.02, "learning_rate": 4.994168146862182e-05, "loss": 1.1866, "step": 21844 }, { "epoch": 0.02, "learning_rate": 4.994167610648187e-05, "loss": 1.1873, "step": 21845 }, { "epoch": 0.02, "learning_rate": 4.9941670744095704e-05, "loss": 1.0371, "step": 21846 }, { "epoch": 0.02, "learning_rate": 4.9941665381463335e-05, "loss": 1.1136, "step": 21847 }, { "epoch": 0.02, "learning_rate": 4.994166001858474e-05, "loss": 1.0121, "step": 21848 }, { "epoch": 0.02, "learning_rate": 4.994165465545994e-05, "loss": 0.5295, "step": 21849 }, { "epoch": 0.02, "learning_rate": 4.9941649292088914e-05, "loss": 1.014, "step": 21850 }, { "epoch": 0.02, "learning_rate": 4.994164392847168e-05, "loss": 1.0339, "step": 21851 }, { "epoch": 0.02, "learning_rate": 4.994163856460824e-05, "loss": 0.4855, "step": 21852 }, { "epoch": 0.02, "learning_rate": 4.994163320049858e-05, "loss": 0.713, "step": 21853 }, { "epoch": 0.02, "learning_rate": 4.994162783614271e-05, "loss": 0.8797, "step": 21854 }, { "epoch": 0.02, "learning_rate": 4.994162247154063e-05, "loss": 0.766, "step": 21855 }, { "epoch": 0.02, "learning_rate": 4.994161710669233e-05, "loss": 0.8868, "step": 21856 }, { "epoch": 0.02, "learning_rate": 4.9941611741597816e-05, "loss": 0.8826, "step": 21857 }, { "epoch": 0.02, "learning_rate": 4.994160637625709e-05, "loss": 0.8637, "step": 21858 }, { "epoch": 0.02, "learning_rate": 4.994160101067016e-05, "loss": 1.089, "step": 21859 }, { "epoch": 0.02, "learning_rate": 4.9941595644837004e-05, "loss": 1.1623, "step": 21860 }, { "epoch": 0.02, "learning_rate": 4.9941590278757644e-05, "loss": 1.3446, "step": 21861 }, { "epoch": 0.02, "learning_rate": 4.994158491243207e-05, "loss": 1.1669, "step": 21862 }, { "epoch": 0.02, "learning_rate": 4.994157954586027e-05, "loss": 1.1461, "step": 21863 }, { "epoch": 0.02, "learning_rate": 4.994157417904227e-05, "loss": 0.9994, "step": 21864 }, { "epoch": 0.02, "learning_rate": 4.994156881197806e-05, "loss": 1.2095, "step": 21865 }, { "epoch": 0.02, "learning_rate": 4.9941563444667634e-05, "loss": 0.9657, "step": 21866 }, { "epoch": 0.02, "learning_rate": 4.9941558077110994e-05, "loss": 1.1049, "step": 21867 }, { "epoch": 0.02, "learning_rate": 4.994155270930814e-05, "loss": 0.6798, "step": 21868 }, { "epoch": 0.02, "learning_rate": 4.9941547341259076e-05, "loss": 1.0595, "step": 21869 }, { "epoch": 0.02, "learning_rate": 4.9941541972963804e-05, "loss": 1.3335, "step": 21870 }, { "epoch": 0.02, "learning_rate": 4.994153660442231e-05, "loss": 1.1598, "step": 21871 }, { "epoch": 0.02, "learning_rate": 4.9941531235634606e-05, "loss": 0.9445, "step": 21872 }, { "epoch": 0.02, "learning_rate": 4.9941525866600694e-05, "loss": 1.1211, "step": 21873 }, { "epoch": 0.02, "learning_rate": 4.994152049732057e-05, "loss": 1.1046, "step": 21874 }, { "epoch": 0.02, "learning_rate": 4.994151512779423e-05, "loss": 0.8923, "step": 21875 }, { "epoch": 0.02, "learning_rate": 4.994150975802168e-05, "loss": 1.1848, "step": 21876 }, { "epoch": 0.02, "learning_rate": 4.994150438800291e-05, "loss": 1.2439, "step": 21877 }, { "epoch": 0.02, "learning_rate": 4.994149901773794e-05, "loss": 1.2503, "step": 21878 }, { "epoch": 0.02, "learning_rate": 4.9941493647226756e-05, "loss": 0.9794, "step": 21879 }, { "epoch": 0.02, "learning_rate": 4.9941488276469355e-05, "loss": 0.9108, "step": 21880 }, { "epoch": 0.02, "learning_rate": 4.9941482905465746e-05, "loss": 1.26, "step": 21881 }, { "epoch": 0.02, "learning_rate": 4.994147753421593e-05, "loss": 0.8044, "step": 21882 }, { "epoch": 0.02, "learning_rate": 4.9941472162719894e-05, "loss": 0.8771, "step": 21883 }, { "epoch": 0.02, "learning_rate": 4.9941466790977645e-05, "loss": 0.3538, "step": 21884 }, { "epoch": 0.02, "learning_rate": 4.994146141898919e-05, "loss": 1.0575, "step": 21885 }, { "epoch": 0.02, "learning_rate": 4.994145604675452e-05, "loss": 1.0018, "step": 21886 }, { "epoch": 0.02, "learning_rate": 4.994145067427364e-05, "loss": 1.2002, "step": 21887 }, { "epoch": 0.02, "learning_rate": 4.994144530154655e-05, "loss": 0.9861, "step": 21888 }, { "epoch": 0.02, "learning_rate": 4.994143992857324e-05, "loss": 1.026, "step": 21889 }, { "epoch": 0.02, "learning_rate": 4.9941434555353725e-05, "loss": 1.01, "step": 21890 }, { "epoch": 0.02, "learning_rate": 4.9941429181888e-05, "loss": 0.9993, "step": 21891 }, { "epoch": 0.02, "learning_rate": 4.994142380817606e-05, "loss": 1.0555, "step": 21892 }, { "epoch": 0.02, "learning_rate": 4.9941418434217913e-05, "loss": 1.1267, "step": 21893 }, { "epoch": 0.02, "learning_rate": 4.9941413060013556e-05, "loss": 0.9104, "step": 21894 }, { "epoch": 0.02, "learning_rate": 4.994140768556298e-05, "loss": 1.0472, "step": 21895 }, { "epoch": 0.02, "learning_rate": 4.99414023108662e-05, "loss": 0.7859, "step": 21896 }, { "epoch": 0.02, "learning_rate": 4.994139693592321e-05, "loss": 1.181, "step": 21897 }, { "epoch": 0.02, "learning_rate": 4.9941391560734007e-05, "loss": 1.0757, "step": 21898 }, { "epoch": 0.02, "learning_rate": 4.9941386185298585e-05, "loss": 1.3281, "step": 21899 }, { "epoch": 0.02, "learning_rate": 4.994138080961697e-05, "loss": 1.1034, "step": 21900 }, { "epoch": 0.02, "learning_rate": 4.994137543368913e-05, "loss": 1.2334, "step": 21901 }, { "epoch": 0.02, "learning_rate": 4.994137005751508e-05, "loss": 1.0356, "step": 21902 }, { "epoch": 0.02, "learning_rate": 4.994136468109482e-05, "loss": 1.1099, "step": 21903 }, { "epoch": 0.02, "learning_rate": 4.9941359304428355e-05, "loss": 1.2952, "step": 21904 }, { "epoch": 0.02, "learning_rate": 4.9941353927515674e-05, "loss": 1.0641, "step": 21905 }, { "epoch": 0.02, "learning_rate": 4.994134855035678e-05, "loss": 1.2827, "step": 21906 }, { "epoch": 0.02, "learning_rate": 4.994134317295168e-05, "loss": 0.9622, "step": 21907 }, { "epoch": 0.02, "learning_rate": 4.994133779530037e-05, "loss": 1.0262, "step": 21908 }, { "epoch": 0.02, "learning_rate": 4.994133241740285e-05, "loss": 1.0099, "step": 21909 }, { "epoch": 0.02, "learning_rate": 4.994132703925912e-05, "loss": 1.1109, "step": 21910 }, { "epoch": 0.02, "learning_rate": 4.994132166086918e-05, "loss": 0.4983, "step": 21911 }, { "epoch": 0.02, "learning_rate": 4.994131628223303e-05, "loss": 0.2494, "step": 21912 }, { "epoch": 0.02, "learning_rate": 4.994131090335067e-05, "loss": 0.6109, "step": 21913 }, { "epoch": 0.02, "learning_rate": 4.9941305524222094e-05, "loss": 1.1543, "step": 21914 }, { "epoch": 0.02, "learning_rate": 4.994130014484731e-05, "loss": 0.9951, "step": 21915 }, { "epoch": 0.02, "learning_rate": 4.994129476522632e-05, "loss": 1.0126, "step": 21916 }, { "epoch": 0.02, "learning_rate": 4.9941289385359116e-05, "loss": 1.0468, "step": 21917 }, { "epoch": 0.02, "learning_rate": 4.99412840052457e-05, "loss": 0.7462, "step": 21918 }, { "epoch": 0.02, "learning_rate": 4.9941278624886084e-05, "loss": 0.974, "step": 21919 }, { "epoch": 0.02, "learning_rate": 4.994127324428025e-05, "loss": 1.1896, "step": 21920 }, { "epoch": 0.02, "learning_rate": 4.994126786342821e-05, "loss": 1.1428, "step": 21921 }, { "epoch": 0.02, "learning_rate": 4.994126248232996e-05, "loss": 1.4326, "step": 21922 }, { "epoch": 0.02, "learning_rate": 4.9941257100985495e-05, "loss": 1.4802, "step": 21923 }, { "epoch": 0.02, "learning_rate": 4.994125171939483e-05, "loss": 1.6295, "step": 21924 }, { "epoch": 0.02, "learning_rate": 4.9941246337557945e-05, "loss": 1.1482, "step": 21925 }, { "epoch": 0.02, "learning_rate": 4.9941240955474854e-05, "loss": 1.051, "step": 21926 }, { "epoch": 0.02, "learning_rate": 4.994123557314556e-05, "loss": 1.212, "step": 21927 }, { "epoch": 0.02, "learning_rate": 4.994123019057005e-05, "loss": 1.0036, "step": 21928 }, { "epoch": 0.02, "learning_rate": 4.994122480774833e-05, "loss": 0.9438, "step": 21929 }, { "epoch": 0.02, "learning_rate": 4.9941219424680406e-05, "loss": 0.8709, "step": 21930 }, { "epoch": 0.02, "learning_rate": 4.994121404136627e-05, "loss": 1.2161, "step": 21931 }, { "epoch": 0.02, "learning_rate": 4.9941208657805926e-05, "loss": 1.2513, "step": 21932 }, { "epoch": 0.02, "learning_rate": 4.9941203273999374e-05, "loss": 0.848, "step": 21933 }, { "epoch": 0.02, "learning_rate": 4.994119788994661e-05, "loss": 1.0146, "step": 21934 }, { "epoch": 0.02, "learning_rate": 4.994119250564764e-05, "loss": 1.2515, "step": 21935 }, { "epoch": 0.02, "learning_rate": 4.9941187121102455e-05, "loss": 1.1471, "step": 21936 }, { "epoch": 0.02, "learning_rate": 4.994118173631107e-05, "loss": 1.1343, "step": 21937 }, { "epoch": 0.02, "learning_rate": 4.9941176351273464e-05, "loss": 0.9777, "step": 21938 }, { "epoch": 0.02, "learning_rate": 4.994117096598966e-05, "loss": 0.7114, "step": 21939 }, { "epoch": 0.02, "learning_rate": 4.9941165580459644e-05, "loss": 0.5791, "step": 21940 }, { "epoch": 0.02, "learning_rate": 4.9941160194683414e-05, "loss": 0.5484, "step": 21941 }, { "epoch": 0.02, "learning_rate": 4.994115480866098e-05, "loss": 0.7862, "step": 21942 }, { "epoch": 0.02, "learning_rate": 4.994114942239234e-05, "loss": 1.1782, "step": 21943 }, { "epoch": 0.02, "learning_rate": 4.994114403587749e-05, "loss": 1.2639, "step": 21944 }, { "epoch": 0.02, "learning_rate": 4.9941138649116434e-05, "loss": 1.5894, "step": 21945 }, { "epoch": 0.02, "learning_rate": 4.994113326210916e-05, "loss": 1.1496, "step": 21946 }, { "epoch": 0.02, "learning_rate": 4.9941127874855685e-05, "loss": 1.5786, "step": 21947 }, { "epoch": 0.02, "learning_rate": 4.994112248735601e-05, "loss": 1.2341, "step": 21948 }, { "epoch": 0.02, "learning_rate": 4.9941117099610114e-05, "loss": 1.2032, "step": 21949 }, { "epoch": 0.02, "learning_rate": 4.994111171161801e-05, "loss": 0.8861, "step": 21950 }, { "epoch": 0.02, "learning_rate": 4.99411063233797e-05, "loss": 1.0823, "step": 21951 }, { "epoch": 0.02, "learning_rate": 4.994110093489518e-05, "loss": 1.1882, "step": 21952 }, { "epoch": 0.02, "learning_rate": 4.994109554616446e-05, "loss": 1.2747, "step": 21953 }, { "epoch": 0.02, "learning_rate": 4.9941090157187524e-05, "loss": 0.6245, "step": 21954 }, { "epoch": 0.02, "learning_rate": 4.9941084767964385e-05, "loss": 0.988, "step": 21955 }, { "epoch": 0.02, "learning_rate": 4.994107937849504e-05, "loss": 0.9734, "step": 21956 }, { "epoch": 0.02, "learning_rate": 4.994107398877948e-05, "loss": 1.241, "step": 21957 }, { "epoch": 0.02, "learning_rate": 4.9941068598817716e-05, "loss": 1.2708, "step": 21958 }, { "epoch": 0.02, "learning_rate": 4.994106320860974e-05, "loss": 1.6863, "step": 21959 }, { "epoch": 0.02, "learning_rate": 4.994105781815556e-05, "loss": 1.129, "step": 21960 }, { "epoch": 0.02, "learning_rate": 4.9941052427455174e-05, "loss": 1.1986, "step": 21961 }, { "epoch": 0.02, "learning_rate": 4.994104703650858e-05, "loss": 1.1472, "step": 21962 }, { "epoch": 0.02, "learning_rate": 4.994104164531578e-05, "loss": 1.1469, "step": 21963 }, { "epoch": 0.02, "learning_rate": 4.9941036253876774e-05, "loss": 1.0672, "step": 21964 }, { "epoch": 0.02, "learning_rate": 4.9941030862191554e-05, "loss": 0.9028, "step": 21965 }, { "epoch": 0.02, "learning_rate": 4.9941025470260126e-05, "loss": 1.8202, "step": 21966 }, { "epoch": 0.02, "learning_rate": 4.994102007808249e-05, "loss": 0.8994, "step": 21967 }, { "epoch": 0.02, "learning_rate": 4.9941014685658656e-05, "loss": 1.0473, "step": 21968 }, { "epoch": 0.02, "learning_rate": 4.994100929298861e-05, "loss": 1.0011, "step": 21969 }, { "epoch": 0.02, "learning_rate": 4.994100390007236e-05, "loss": 1.1317, "step": 21970 }, { "epoch": 0.02, "learning_rate": 4.99409985069099e-05, "loss": 1.1085, "step": 21971 }, { "epoch": 0.02, "learning_rate": 4.9940993113501225e-05, "loss": 1.0306, "step": 21972 }, { "epoch": 0.02, "learning_rate": 4.9940987719846355e-05, "loss": 1.2431, "step": 21973 }, { "epoch": 0.02, "learning_rate": 4.994098232594527e-05, "loss": 1.1118, "step": 21974 }, { "epoch": 0.02, "learning_rate": 4.9940976931797977e-05, "loss": 1.1545, "step": 21975 }, { "epoch": 0.02, "learning_rate": 4.994097153740449e-05, "loss": 1.1777, "step": 21976 }, { "epoch": 0.02, "learning_rate": 4.994096614276478e-05, "loss": 1.3271, "step": 21977 }, { "epoch": 0.02, "learning_rate": 4.9940960747878876e-05, "loss": 1.6184, "step": 21978 }, { "epoch": 0.02, "learning_rate": 4.994095535274676e-05, "loss": 0.8815, "step": 21979 }, { "epoch": 0.02, "learning_rate": 4.994094995736844e-05, "loss": 1.5912, "step": 21980 }, { "epoch": 0.02, "learning_rate": 4.9940944561743904e-05, "loss": 0.8405, "step": 21981 }, { "epoch": 0.02, "learning_rate": 4.9940939165873176e-05, "loss": 0.9565, "step": 21982 }, { "epoch": 0.02, "learning_rate": 4.994093376975623e-05, "loss": 0.996, "step": 21983 }, { "epoch": 0.02, "learning_rate": 4.994092837339308e-05, "loss": 1.1973, "step": 21984 }, { "epoch": 0.02, "learning_rate": 4.994092297678373e-05, "loss": 1.0706, "step": 21985 }, { "epoch": 0.02, "learning_rate": 4.994091757992817e-05, "loss": 1.0045, "step": 21986 }, { "epoch": 0.02, "learning_rate": 4.9940912182826405e-05, "loss": 0.976, "step": 21987 }, { "epoch": 0.02, "learning_rate": 4.994090678547843e-05, "loss": 0.9117, "step": 21988 }, { "epoch": 0.02, "learning_rate": 4.9940901387884255e-05, "loss": 0.6756, "step": 21989 }, { "epoch": 0.02, "learning_rate": 4.994089599004387e-05, "loss": 1.0898, "step": 21990 }, { "epoch": 0.02, "learning_rate": 4.994089059195728e-05, "loss": 1.1355, "step": 21991 }, { "epoch": 0.02, "learning_rate": 4.9940885193624475e-05, "loss": 1.2356, "step": 21992 }, { "epoch": 0.02, "learning_rate": 4.994087979504548e-05, "loss": 1.1732, "step": 21993 }, { "epoch": 0.02, "learning_rate": 4.9940874396220265e-05, "loss": 1.0352, "step": 21994 }, { "epoch": 0.02, "learning_rate": 4.994086899714885e-05, "loss": 1.0433, "step": 21995 }, { "epoch": 0.02, "learning_rate": 4.994086359783123e-05, "loss": 0.6878, "step": 21996 }, { "epoch": 0.02, "learning_rate": 4.99408581982674e-05, "loss": 1.1046, "step": 21997 }, { "epoch": 0.02, "learning_rate": 4.994085279845737e-05, "loss": 0.9943, "step": 21998 }, { "epoch": 0.02, "learning_rate": 4.994084739840114e-05, "loss": 1.2319, "step": 21999 }, { "epoch": 0.02, "learning_rate": 4.99408419980987e-05, "loss": 0.9888, "step": 22000 }, { "epoch": 0.02, "eval_loss": 1.0208954811096191, "eval_runtime": 86.4671, "eval_samples_per_second": 16.018, "eval_steps_per_second": 4.013, "step": 22000 }, { "epoch": 0.02, "learning_rate": 4.9940836597550045e-05, "loss": 0.6165, "step": 22001 }, { "epoch": 0.02, "learning_rate": 4.994083119675519e-05, "loss": 0.6036, "step": 22002 }, { "epoch": 0.02, "learning_rate": 4.994082579571413e-05, "loss": 0.5895, "step": 22003 }, { "epoch": 0.02, "learning_rate": 4.994082039442687e-05, "loss": 0.5954, "step": 22004 }, { "epoch": 0.02, "learning_rate": 4.9940814992893396e-05, "loss": 0.6935, "step": 22005 }, { "epoch": 0.02, "learning_rate": 4.9940809591113725e-05, "loss": 1.3611, "step": 22006 }, { "epoch": 0.02, "learning_rate": 4.994080418908784e-05, "loss": 1.0446, "step": 22007 }, { "epoch": 0.02, "learning_rate": 4.994079878681576e-05, "loss": 1.0151, "step": 22008 }, { "epoch": 0.02, "learning_rate": 4.994079338429747e-05, "loss": 1.2993, "step": 22009 }, { "epoch": 0.02, "learning_rate": 4.994078798153297e-05, "loss": 2.0238, "step": 22010 }, { "epoch": 0.02, "learning_rate": 4.994078257852227e-05, "loss": 1.2142, "step": 22011 }, { "epoch": 0.02, "learning_rate": 4.9940777175265366e-05, "loss": 1.1003, "step": 22012 }, { "epoch": 0.02, "learning_rate": 4.9940771771762254e-05, "loss": 1.0433, "step": 22013 }, { "epoch": 0.02, "learning_rate": 4.994076636801294e-05, "loss": 1.1874, "step": 22014 }, { "epoch": 0.02, "learning_rate": 4.9940760964017425e-05, "loss": 1.2904, "step": 22015 }, { "epoch": 0.02, "learning_rate": 4.99407555597757e-05, "loss": 1.1201, "step": 22016 }, { "epoch": 0.02, "learning_rate": 4.9940750155287775e-05, "loss": 1.093, "step": 22017 }, { "epoch": 0.02, "learning_rate": 4.994074475055364e-05, "loss": 1.0782, "step": 22018 }, { "epoch": 0.02, "learning_rate": 4.9940739345573296e-05, "loss": 0.866, "step": 22019 }, { "epoch": 0.02, "learning_rate": 4.994073394034676e-05, "loss": 1.0851, "step": 22020 }, { "epoch": 0.02, "learning_rate": 4.994072853487401e-05, "loss": 1.0589, "step": 22021 }, { "epoch": 0.02, "learning_rate": 4.994072312915506e-05, "loss": 0.606, "step": 22022 }, { "epoch": 0.02, "learning_rate": 4.99407177231899e-05, "loss": 0.7022, "step": 22023 }, { "epoch": 0.02, "learning_rate": 4.994071231697855e-05, "loss": 0.9314, "step": 22024 }, { "epoch": 0.02, "learning_rate": 4.994070691052098e-05, "loss": 1.0438, "step": 22025 }, { "epoch": 0.02, "learning_rate": 4.9940701503817213e-05, "loss": 1.178, "step": 22026 }, { "epoch": 0.02, "learning_rate": 4.994069609686724e-05, "loss": 1.0563, "step": 22027 }, { "epoch": 0.02, "learning_rate": 4.9940690689671065e-05, "loss": 0.9947, "step": 22028 }, { "epoch": 0.02, "learning_rate": 4.994068528222868e-05, "loss": 1.1696, "step": 22029 }, { "epoch": 0.02, "learning_rate": 4.9940679874540095e-05, "loss": 0.881, "step": 22030 }, { "epoch": 0.02, "learning_rate": 4.994067446660531e-05, "loss": 1.0745, "step": 22031 }, { "epoch": 0.02, "learning_rate": 4.994066905842432e-05, "loss": 1.1293, "step": 22032 }, { "epoch": 0.02, "learning_rate": 4.994066364999712e-05, "loss": 1.0681, "step": 22033 }, { "epoch": 0.02, "learning_rate": 4.9940658241323725e-05, "loss": 0.759, "step": 22034 }, { "epoch": 0.02, "learning_rate": 4.994065283240412e-05, "loss": 0.7087, "step": 22035 }, { "epoch": 0.02, "learning_rate": 4.994064742323832e-05, "loss": 1.2128, "step": 22036 }, { "epoch": 0.02, "learning_rate": 4.9940642013826304e-05, "loss": 1.5023, "step": 22037 }, { "epoch": 0.02, "learning_rate": 4.994063660416809e-05, "loss": 1.2635, "step": 22038 }, { "epoch": 0.02, "learning_rate": 4.994063119426368e-05, "loss": 1.0739, "step": 22039 }, { "epoch": 0.02, "learning_rate": 4.994062578411305e-05, "loss": 1.1619, "step": 22040 }, { "epoch": 0.02, "learning_rate": 4.994062037371623e-05, "loss": 0.9069, "step": 22041 }, { "epoch": 0.02, "learning_rate": 4.99406149630732e-05, "loss": 1.0926, "step": 22042 }, { "epoch": 0.02, "learning_rate": 4.994060955218397e-05, "loss": 1.3818, "step": 22043 }, { "epoch": 0.02, "learning_rate": 4.994060414104854e-05, "loss": 1.2179, "step": 22044 }, { "epoch": 0.02, "learning_rate": 4.99405987296669e-05, "loss": 1.0342, "step": 22045 }, { "epoch": 0.02, "learning_rate": 4.994059331803906e-05, "loss": 0.8937, "step": 22046 }, { "epoch": 0.02, "learning_rate": 4.9940587906165014e-05, "loss": 1.1544, "step": 22047 }, { "epoch": 0.02, "learning_rate": 4.994058249404477e-05, "loss": 0.8583, "step": 22048 }, { "epoch": 0.02, "learning_rate": 4.994057708167833e-05, "loss": 1.1914, "step": 22049 }, { "epoch": 0.02, "learning_rate": 4.9940571669065675e-05, "loss": 0.9587, "step": 22050 }, { "epoch": 0.02, "learning_rate": 4.994056625620682e-05, "loss": 0.9812, "step": 22051 }, { "epoch": 0.02, "learning_rate": 4.994056084310176e-05, "loss": 1.1614, "step": 22052 }, { "epoch": 0.02, "learning_rate": 4.9940555429750505e-05, "loss": 1.1222, "step": 22053 }, { "epoch": 0.02, "learning_rate": 4.9940550016153044e-05, "loss": 1.1019, "step": 22054 }, { "epoch": 0.02, "learning_rate": 4.9940544602309375e-05, "loss": 1.0073, "step": 22055 }, { "epoch": 0.02, "learning_rate": 4.9940539188219504e-05, "loss": 0.9672, "step": 22056 }, { "epoch": 0.02, "learning_rate": 4.994053377388344e-05, "loss": 0.9748, "step": 22057 }, { "epoch": 0.02, "learning_rate": 4.994052835930116e-05, "loss": 1.0404, "step": 22058 }, { "epoch": 0.02, "learning_rate": 4.994052294447269e-05, "loss": 1.1456, "step": 22059 }, { "epoch": 0.02, "learning_rate": 4.994051752939801e-05, "loss": 1.0114, "step": 22060 }, { "epoch": 0.02, "learning_rate": 4.994051211407713e-05, "loss": 0.9307, "step": 22061 }, { "epoch": 0.02, "learning_rate": 4.994050669851005e-05, "loss": 1.1907, "step": 22062 }, { "epoch": 0.02, "learning_rate": 4.994050128269676e-05, "loss": 1.0894, "step": 22063 }, { "epoch": 0.02, "learning_rate": 4.994049586663728e-05, "loss": 1.2495, "step": 22064 }, { "epoch": 0.02, "learning_rate": 4.994049045033159e-05, "loss": 1.2588, "step": 22065 }, { "epoch": 0.02, "learning_rate": 4.994048503377971e-05, "loss": 0.9996, "step": 22066 }, { "epoch": 0.02, "learning_rate": 4.994047961698161e-05, "loss": 1.1738, "step": 22067 }, { "epoch": 0.02, "learning_rate": 4.994047419993731e-05, "loss": 1.1268, "step": 22068 }, { "epoch": 0.02, "learning_rate": 4.994046878264682e-05, "loss": 1.1319, "step": 22069 }, { "epoch": 0.02, "learning_rate": 4.994046336511012e-05, "loss": 1.2897, "step": 22070 }, { "epoch": 0.02, "learning_rate": 4.994045794732722e-05, "loss": 0.938, "step": 22071 }, { "epoch": 0.02, "learning_rate": 4.994045252929812e-05, "loss": 0.9807, "step": 22072 }, { "epoch": 0.02, "learning_rate": 4.994044711102281e-05, "loss": 1.1295, "step": 22073 }, { "epoch": 0.02, "learning_rate": 4.994044169250132e-05, "loss": 1.2488, "step": 22074 }, { "epoch": 0.02, "learning_rate": 4.9940436273733604e-05, "loss": 1.2837, "step": 22075 }, { "epoch": 0.02, "learning_rate": 4.99404308547197e-05, "loss": 1.34, "step": 22076 }, { "epoch": 0.02, "learning_rate": 4.994042543545959e-05, "loss": 1.2402, "step": 22077 }, { "epoch": 0.02, "learning_rate": 4.9940420015953285e-05, "loss": 0.9383, "step": 22078 }, { "epoch": 0.02, "learning_rate": 4.994041459620077e-05, "loss": 1.3563, "step": 22079 }, { "epoch": 0.02, "learning_rate": 4.9940409176202055e-05, "loss": 1.0821, "step": 22080 }, { "epoch": 0.02, "learning_rate": 4.994040375595714e-05, "loss": 0.886, "step": 22081 }, { "epoch": 0.02, "learning_rate": 4.9940398335466025e-05, "loss": 1.2144, "step": 22082 }, { "epoch": 0.02, "learning_rate": 4.9940392914728713e-05, "loss": 1.1659, "step": 22083 }, { "epoch": 0.02, "learning_rate": 4.994038749374519e-05, "loss": 1.0736, "step": 22084 }, { "epoch": 0.02, "learning_rate": 4.994038207251548e-05, "loss": 0.1461, "step": 22085 }, { "epoch": 0.02, "learning_rate": 4.9940376651039554e-05, "loss": 0.0973, "step": 22086 }, { "epoch": 0.02, "learning_rate": 4.994037122931744e-05, "loss": 0.0882, "step": 22087 }, { "epoch": 0.02, "learning_rate": 4.994036580734911e-05, "loss": 0.07, "step": 22088 }, { "epoch": 0.02, "learning_rate": 4.9940360385134596e-05, "loss": 0.0834, "step": 22089 }, { "epoch": 0.02, "learning_rate": 4.994035496267387e-05, "loss": 0.0662, "step": 22090 }, { "epoch": 0.02, "learning_rate": 4.9940349539966944e-05, "loss": 0.0606, "step": 22091 }, { "epoch": 0.02, "learning_rate": 4.994034411701383e-05, "loss": 0.0522, "step": 22092 }, { "epoch": 0.02, "learning_rate": 4.99403386938145e-05, "loss": 0.0544, "step": 22093 }, { "epoch": 0.02, "learning_rate": 4.994033327036898e-05, "loss": 0.0373, "step": 22094 }, { "epoch": 0.02, "learning_rate": 4.994032784667725e-05, "loss": 0.0541, "step": 22095 }, { "epoch": 0.02, "learning_rate": 4.9940322422739325e-05, "loss": 0.0627, "step": 22096 }, { "epoch": 0.02, "learning_rate": 4.99403169985552e-05, "loss": 0.0694, "step": 22097 }, { "epoch": 0.02, "learning_rate": 4.994031157412487e-05, "loss": 0.0533, "step": 22098 }, { "epoch": 0.02, "learning_rate": 4.994030614944835e-05, "loss": 0.8092, "step": 22099 }, { "epoch": 0.02, "learning_rate": 4.994030072452562e-05, "loss": 1.3318, "step": 22100 }, { "epoch": 0.02, "learning_rate": 4.994029529935669e-05, "loss": 1.1477, "step": 22101 }, { "epoch": 0.02, "learning_rate": 4.994028987394157e-05, "loss": 1.4543, "step": 22102 }, { "epoch": 0.02, "learning_rate": 4.994028444828024e-05, "loss": 0.8911, "step": 22103 }, { "epoch": 0.02, "learning_rate": 4.994027902237271e-05, "loss": 0.8009, "step": 22104 }, { "epoch": 0.02, "learning_rate": 4.994027359621898e-05, "loss": 0.8756, "step": 22105 }, { "epoch": 0.02, "learning_rate": 4.994026816981906e-05, "loss": 0.9474, "step": 22106 }, { "epoch": 0.02, "learning_rate": 4.994026274317293e-05, "loss": 0.645, "step": 22107 }, { "epoch": 0.02, "learning_rate": 4.99402573162806e-05, "loss": 0.7555, "step": 22108 }, { "epoch": 0.02, "learning_rate": 4.994025188914208e-05, "loss": 0.121, "step": 22109 }, { "epoch": 0.02, "learning_rate": 4.994024646175735e-05, "loss": 0.7596, "step": 22110 }, { "epoch": 0.02, "learning_rate": 4.994024103412642e-05, "loss": 0.9172, "step": 22111 }, { "epoch": 0.02, "learning_rate": 4.99402356062493e-05, "loss": 1.1311, "step": 22112 }, { "epoch": 0.02, "learning_rate": 4.994023017812598e-05, "loss": 1.383, "step": 22113 }, { "epoch": 0.02, "learning_rate": 4.994022474975645e-05, "loss": 1.4239, "step": 22114 }, { "epoch": 0.02, "learning_rate": 4.994021932114073e-05, "loss": 1.2888, "step": 22115 }, { "epoch": 0.02, "learning_rate": 4.994021389227881e-05, "loss": 1.3813, "step": 22116 }, { "epoch": 0.02, "learning_rate": 4.994020846317068e-05, "loss": 1.1023, "step": 22117 }, { "epoch": 0.02, "learning_rate": 4.994020303381636e-05, "loss": 1.1101, "step": 22118 }, { "epoch": 0.02, "learning_rate": 4.994019760421585e-05, "loss": 1.0153, "step": 22119 }, { "epoch": 0.02, "learning_rate": 4.994019217436913e-05, "loss": 1.2262, "step": 22120 }, { "epoch": 0.02, "learning_rate": 4.9940186744276204e-05, "loss": 1.1241, "step": 22121 }, { "epoch": 0.02, "learning_rate": 4.994018131393708e-05, "loss": 0.8321, "step": 22122 }, { "epoch": 0.02, "learning_rate": 4.994017588335177e-05, "loss": 0.818, "step": 22123 }, { "epoch": 0.02, "learning_rate": 4.9940170452520255e-05, "loss": 1.2988, "step": 22124 }, { "epoch": 0.02, "learning_rate": 4.994016502144254e-05, "loss": 1.2263, "step": 22125 }, { "epoch": 0.02, "learning_rate": 4.9940159590118626e-05, "loss": 0.7546, "step": 22126 }, { "epoch": 0.02, "learning_rate": 4.9940154158548516e-05, "loss": 1.0711, "step": 22127 }, { "epoch": 0.02, "learning_rate": 4.99401487267322e-05, "loss": 1.0441, "step": 22128 }, { "epoch": 0.02, "learning_rate": 4.9940143294669695e-05, "loss": 0.8568, "step": 22129 }, { "epoch": 0.02, "learning_rate": 4.994013786236099e-05, "loss": 1.1937, "step": 22130 }, { "epoch": 0.02, "learning_rate": 4.994013242980609e-05, "loss": 1.2498, "step": 22131 }, { "epoch": 0.02, "learning_rate": 4.994012699700498e-05, "loss": 1.0209, "step": 22132 }, { "epoch": 0.02, "learning_rate": 4.994012156395768e-05, "loss": 1.3841, "step": 22133 }, { "epoch": 0.02, "learning_rate": 4.9940116130664173e-05, "loss": 1.1133, "step": 22134 }, { "epoch": 0.02, "learning_rate": 4.9940110697124474e-05, "loss": 1.126, "step": 22135 }, { "epoch": 0.02, "learning_rate": 4.994010526333858e-05, "loss": 0.8245, "step": 22136 }, { "epoch": 0.02, "learning_rate": 4.994009982930648e-05, "loss": 0.8978, "step": 22137 }, { "epoch": 0.02, "learning_rate": 4.9940094395028184e-05, "loss": 1.4016, "step": 22138 }, { "epoch": 0.02, "learning_rate": 4.99400889605037e-05, "loss": 0.8745, "step": 22139 }, { "epoch": 0.02, "learning_rate": 4.994008352573301e-05, "loss": 1.1652, "step": 22140 }, { "epoch": 0.02, "learning_rate": 4.994007809071612e-05, "loss": 1.1048, "step": 22141 }, { "epoch": 0.02, "learning_rate": 4.994007265545303e-05, "loss": 1.181, "step": 22142 }, { "epoch": 0.02, "learning_rate": 4.994006721994374e-05, "loss": 1.021, "step": 22143 }, { "epoch": 0.02, "learning_rate": 4.9940061784188266e-05, "loss": 1.3591, "step": 22144 }, { "epoch": 0.02, "learning_rate": 4.994005634818658e-05, "loss": 1.1281, "step": 22145 }, { "epoch": 0.02, "learning_rate": 4.994005091193871e-05, "loss": 1.172, "step": 22146 }, { "epoch": 0.02, "learning_rate": 4.994004547544463e-05, "loss": 1.4742, "step": 22147 }, { "epoch": 0.02, "learning_rate": 4.994004003870435e-05, "loss": 1.2497, "step": 22148 }, { "epoch": 0.02, "learning_rate": 4.994003460171788e-05, "loss": 1.1395, "step": 22149 }, { "epoch": 0.02, "learning_rate": 4.994002916448521e-05, "loss": 0.9154, "step": 22150 }, { "epoch": 0.02, "learning_rate": 4.994002372700635e-05, "loss": 1.1961, "step": 22151 }, { "epoch": 0.02, "learning_rate": 4.994001828928129e-05, "loss": 1.1869, "step": 22152 }, { "epoch": 0.02, "learning_rate": 4.994001285131002e-05, "loss": 1.1057, "step": 22153 }, { "epoch": 0.02, "learning_rate": 4.994000741309257e-05, "loss": 1.1132, "step": 22154 }, { "epoch": 0.02, "learning_rate": 4.9940001974628915e-05, "loss": 1.2987, "step": 22155 }, { "epoch": 0.02, "learning_rate": 4.993999653591906e-05, "loss": 1.3052, "step": 22156 }, { "epoch": 0.02, "learning_rate": 4.9939991096963014e-05, "loss": 1.0056, "step": 22157 }, { "epoch": 0.02, "learning_rate": 4.993998565776077e-05, "loss": 0.7543, "step": 22158 }, { "epoch": 0.02, "learning_rate": 4.993998021831232e-05, "loss": 1.1718, "step": 22159 }, { "epoch": 0.02, "learning_rate": 4.9939974778617685e-05, "loss": 0.9731, "step": 22160 }, { "epoch": 0.02, "learning_rate": 4.993996933867684e-05, "loss": 1.1318, "step": 22161 }, { "epoch": 0.02, "learning_rate": 4.993996389848981e-05, "loss": 0.9144, "step": 22162 }, { "epoch": 0.02, "learning_rate": 4.993995845805658e-05, "loss": 1.0712, "step": 22163 }, { "epoch": 0.02, "learning_rate": 4.993995301737715e-05, "loss": 1.0625, "step": 22164 }, { "epoch": 0.02, "learning_rate": 4.9939947576451526e-05, "loss": 1.1013, "step": 22165 }, { "epoch": 0.02, "learning_rate": 4.993994213527971e-05, "loss": 1.2061, "step": 22166 }, { "epoch": 0.02, "learning_rate": 4.993993669386169e-05, "loss": 0.9907, "step": 22167 }, { "epoch": 0.02, "learning_rate": 4.993993125219748e-05, "loss": 0.7643, "step": 22168 }, { "epoch": 0.02, "learning_rate": 4.993992581028707e-05, "loss": 1.9165, "step": 22169 }, { "epoch": 0.02, "learning_rate": 4.9939920368130454e-05, "loss": 1.573, "step": 22170 }, { "epoch": 0.02, "learning_rate": 4.993991492572765e-05, "loss": 1.0276, "step": 22171 }, { "epoch": 0.02, "learning_rate": 4.9939909483078655e-05, "loss": 1.1533, "step": 22172 }, { "epoch": 0.02, "learning_rate": 4.9939904040183464e-05, "loss": 1.3832, "step": 22173 }, { "epoch": 0.02, "learning_rate": 4.993989859704207e-05, "loss": 0.8612, "step": 22174 }, { "epoch": 0.02, "learning_rate": 4.993989315365448e-05, "loss": 0.8356, "step": 22175 }, { "epoch": 0.02, "learning_rate": 4.9939887710020696e-05, "loss": 0.271, "step": 22176 }, { "epoch": 0.02, "learning_rate": 4.993988226614072e-05, "loss": 0.2349, "step": 22177 }, { "epoch": 0.02, "learning_rate": 4.9939876822014535e-05, "loss": 0.2051, "step": 22178 }, { "epoch": 0.02, "learning_rate": 4.9939871377642165e-05, "loss": 0.7093, "step": 22179 }, { "epoch": 0.02, "learning_rate": 4.99398659330236e-05, "loss": 1.2577, "step": 22180 }, { "epoch": 0.02, "learning_rate": 4.9939860488158826e-05, "loss": 1.1181, "step": 22181 }, { "epoch": 0.02, "learning_rate": 4.9939855043047864e-05, "loss": 1.1564, "step": 22182 }, { "epoch": 0.02, "learning_rate": 4.993984959769071e-05, "loss": 1.0597, "step": 22183 }, { "epoch": 0.02, "learning_rate": 4.993984415208737e-05, "loss": 0.4073, "step": 22184 }, { "epoch": 0.02, "learning_rate": 4.993983870623782e-05, "loss": 1.2348, "step": 22185 }, { "epoch": 0.02, "learning_rate": 4.9939833260142076e-05, "loss": 0.9688, "step": 22186 }, { "epoch": 0.02, "learning_rate": 4.993982781380013e-05, "loss": 1.0327, "step": 22187 }, { "epoch": 0.02, "learning_rate": 4.9939822367212004e-05, "loss": 1.3823, "step": 22188 }, { "epoch": 0.02, "learning_rate": 4.993981692037767e-05, "loss": 1.1151, "step": 22189 }, { "epoch": 0.02, "learning_rate": 4.993981147329715e-05, "loss": 1.0981, "step": 22190 }, { "epoch": 0.02, "learning_rate": 4.9939806025970425e-05, "loss": 1.1135, "step": 22191 }, { "epoch": 0.02, "learning_rate": 4.993980057839751e-05, "loss": 1.1927, "step": 22192 }, { "epoch": 0.02, "learning_rate": 4.99397951305784e-05, "loss": 1.103, "step": 22193 }, { "epoch": 0.02, "learning_rate": 4.9939789682513097e-05, "loss": 0.885, "step": 22194 }, { "epoch": 0.02, "learning_rate": 4.993978423420159e-05, "loss": 1.1748, "step": 22195 }, { "epoch": 0.02, "learning_rate": 4.99397787856439e-05, "loss": 1.0383, "step": 22196 }, { "epoch": 0.02, "learning_rate": 4.993977333684001e-05, "loss": 1.1571, "step": 22197 }, { "epoch": 0.02, "learning_rate": 4.9939767887789925e-05, "loss": 1.0387, "step": 22198 }, { "epoch": 0.02, "learning_rate": 4.9939762438493644e-05, "loss": 1.1472, "step": 22199 }, { "epoch": 0.02, "learning_rate": 4.993975698895117e-05, "loss": 0.8504, "step": 22200 }, { "epoch": 0.02, "learning_rate": 4.99397515391625e-05, "loss": 1.0299, "step": 22201 }, { "epoch": 0.02, "learning_rate": 4.993974608912764e-05, "loss": 1.1455, "step": 22202 }, { "epoch": 0.02, "learning_rate": 4.9939740638846576e-05, "loss": 1.2034, "step": 22203 }, { "epoch": 0.02, "learning_rate": 4.993973518831932e-05, "loss": 1.2032, "step": 22204 }, { "epoch": 0.02, "learning_rate": 4.993972973754587e-05, "loss": 1.1968, "step": 22205 }, { "epoch": 0.02, "learning_rate": 4.9939724286526226e-05, "loss": 1.2317, "step": 22206 }, { "epoch": 0.02, "learning_rate": 4.993971883526039e-05, "loss": 1.1543, "step": 22207 }, { "epoch": 0.02, "learning_rate": 4.993971338374836e-05, "loss": 1.0025, "step": 22208 }, { "epoch": 0.02, "learning_rate": 4.993970793199013e-05, "loss": 1.1406, "step": 22209 }, { "epoch": 0.02, "learning_rate": 4.993970247998571e-05, "loss": 1.2305, "step": 22210 }, { "epoch": 0.02, "learning_rate": 4.99396970277351e-05, "loss": 0.8869, "step": 22211 }, { "epoch": 0.02, "learning_rate": 4.993969157523829e-05, "loss": 1.0279, "step": 22212 }, { "epoch": 0.02, "learning_rate": 4.993968612249529e-05, "loss": 1.0019, "step": 22213 }, { "epoch": 0.02, "learning_rate": 4.993968066950609e-05, "loss": 1.452, "step": 22214 }, { "epoch": 0.02, "learning_rate": 4.99396752162707e-05, "loss": 1.0182, "step": 22215 }, { "epoch": 0.02, "learning_rate": 4.993966976278911e-05, "loss": 1.1806, "step": 22216 }, { "epoch": 0.02, "learning_rate": 4.993966430906133e-05, "loss": 1.1939, "step": 22217 }, { "epoch": 0.02, "learning_rate": 4.9939658855087354e-05, "loss": 1.0205, "step": 22218 }, { "epoch": 0.02, "learning_rate": 4.993965340086719e-05, "loss": 1.2034, "step": 22219 }, { "epoch": 0.02, "learning_rate": 4.9939647946400836e-05, "loss": 1.1132, "step": 22220 }, { "epoch": 0.02, "learning_rate": 4.9939642491688274e-05, "loss": 1.0666, "step": 22221 }, { "epoch": 0.02, "learning_rate": 4.993963703672953e-05, "loss": 0.7728, "step": 22222 }, { "epoch": 0.02, "learning_rate": 4.993963158152459e-05, "loss": 0.6948, "step": 22223 }, { "epoch": 0.02, "learning_rate": 4.993962612607346e-05, "loss": 1.4592, "step": 22224 }, { "epoch": 0.02, "learning_rate": 4.993962067037613e-05, "loss": 1.2735, "step": 22225 }, { "epoch": 0.02, "learning_rate": 4.9939615214432604e-05, "loss": 0.5582, "step": 22226 }, { "epoch": 0.02, "learning_rate": 4.993960975824289e-05, "loss": 1.1151, "step": 22227 }, { "epoch": 0.02, "learning_rate": 4.993960430180698e-05, "loss": 1.0791, "step": 22228 }, { "epoch": 0.02, "learning_rate": 4.993959884512488e-05, "loss": 1.6951, "step": 22229 }, { "epoch": 0.02, "learning_rate": 4.993959338819658e-05, "loss": 1.4961, "step": 22230 }, { "epoch": 0.02, "learning_rate": 4.9939587931022095e-05, "loss": 1.2747, "step": 22231 }, { "epoch": 0.02, "learning_rate": 4.993958247360141e-05, "loss": 0.8967, "step": 22232 }, { "epoch": 0.02, "learning_rate": 4.993957701593454e-05, "loss": 1.3909, "step": 22233 }, { "epoch": 0.02, "learning_rate": 4.993957155802147e-05, "loss": 1.1862, "step": 22234 }, { "epoch": 0.02, "learning_rate": 4.993956609986221e-05, "loss": 1.2344, "step": 22235 }, { "epoch": 0.02, "learning_rate": 4.993956064145676e-05, "loss": 1.3602, "step": 22236 }, { "epoch": 0.02, "learning_rate": 4.9939555182805115e-05, "loss": 0.623, "step": 22237 }, { "epoch": 0.02, "learning_rate": 4.9939549723907275e-05, "loss": 0.2655, "step": 22238 }, { "epoch": 0.02, "learning_rate": 4.993954426476325e-05, "loss": 0.1426, "step": 22239 }, { "epoch": 0.02, "learning_rate": 4.993953880537302e-05, "loss": 0.5222, "step": 22240 }, { "epoch": 0.02, "learning_rate": 4.99395333457366e-05, "loss": 1.0755, "step": 22241 }, { "epoch": 0.02, "learning_rate": 4.993952788585399e-05, "loss": 1.2242, "step": 22242 }, { "epoch": 0.02, "learning_rate": 4.9939522425725195e-05, "loss": 1.1512, "step": 22243 }, { "epoch": 0.02, "learning_rate": 4.99395169653502e-05, "loss": 0.5723, "step": 22244 }, { "epoch": 0.02, "learning_rate": 4.9939511504729005e-05, "loss": 0.9184, "step": 22245 }, { "epoch": 0.02, "learning_rate": 4.993950604386163e-05, "loss": 1.023, "step": 22246 }, { "epoch": 0.02, "learning_rate": 4.993950058274806e-05, "loss": 1.2333, "step": 22247 }, { "epoch": 0.02, "learning_rate": 4.9939495121388297e-05, "loss": 1.0079, "step": 22248 }, { "epoch": 0.02, "learning_rate": 4.9939489659782343e-05, "loss": 1.4063, "step": 22249 }, { "epoch": 0.02, "learning_rate": 4.9939484197930195e-05, "loss": 1.1359, "step": 22250 }, { "epoch": 0.02, "learning_rate": 4.993947873583186e-05, "loss": 1.1665, "step": 22251 }, { "epoch": 0.02, "learning_rate": 4.9939473273487326e-05, "loss": 0.9153, "step": 22252 }, { "epoch": 0.02, "learning_rate": 4.99394678108966e-05, "loss": 0.9765, "step": 22253 }, { "epoch": 0.02, "learning_rate": 4.993946234805968e-05, "loss": 1.1693, "step": 22254 }, { "epoch": 0.02, "learning_rate": 4.993945688497658e-05, "loss": 1.3216, "step": 22255 }, { "epoch": 0.02, "learning_rate": 4.993945142164728e-05, "loss": 1.1673, "step": 22256 }, { "epoch": 0.02, "learning_rate": 4.9939445958071785e-05, "loss": 0.9825, "step": 22257 }, { "epoch": 0.02, "learning_rate": 4.99394404942501e-05, "loss": 1.0525, "step": 22258 }, { "epoch": 0.02, "learning_rate": 4.9939435030182224e-05, "loss": 1.0972, "step": 22259 }, { "epoch": 0.02, "learning_rate": 4.993942956586816e-05, "loss": 1.0889, "step": 22260 }, { "epoch": 0.02, "learning_rate": 4.9939424101307896e-05, "loss": 1.204, "step": 22261 }, { "epoch": 0.02, "learning_rate": 4.993941863650145e-05, "loss": 1.2045, "step": 22262 }, { "epoch": 0.02, "learning_rate": 4.9939413171448814e-05, "loss": 1.2288, "step": 22263 }, { "epoch": 0.02, "learning_rate": 4.993940770614998e-05, "loss": 1.1883, "step": 22264 }, { "epoch": 0.02, "learning_rate": 4.993940224060496e-05, "loss": 0.8922, "step": 22265 }, { "epoch": 0.02, "learning_rate": 4.993939677481374e-05, "loss": 0.9595, "step": 22266 }, { "epoch": 0.02, "learning_rate": 4.9939391308776334e-05, "loss": 1.1863, "step": 22267 }, { "epoch": 0.02, "learning_rate": 4.993938584249274e-05, "loss": 1.5447, "step": 22268 }, { "epoch": 0.02, "learning_rate": 4.993938037596295e-05, "loss": 1.5893, "step": 22269 }, { "epoch": 0.02, "learning_rate": 4.993937490918697e-05, "loss": 0.8702, "step": 22270 }, { "epoch": 0.02, "learning_rate": 4.99393694421648e-05, "loss": 1.189, "step": 22271 }, { "epoch": 0.02, "learning_rate": 4.993936397489644e-05, "loss": 1.1282, "step": 22272 }, { "epoch": 0.02, "learning_rate": 4.993935850738189e-05, "loss": 1.0422, "step": 22273 }, { "epoch": 0.02, "learning_rate": 4.993935303962115e-05, "loss": 1.0953, "step": 22274 }, { "epoch": 0.02, "learning_rate": 4.993934757161421e-05, "loss": 1.1224, "step": 22275 }, { "epoch": 0.02, "learning_rate": 4.9939342103361086e-05, "loss": 1.0236, "step": 22276 }, { "epoch": 0.02, "learning_rate": 4.9939336634861775e-05, "loss": 0.837, "step": 22277 }, { "epoch": 0.02, "learning_rate": 4.993933116611626e-05, "loss": 1.1181, "step": 22278 }, { "epoch": 0.02, "learning_rate": 4.9939325697124575e-05, "loss": 0.9411, "step": 22279 }, { "epoch": 0.02, "learning_rate": 4.9939320227886685e-05, "loss": 0.9417, "step": 22280 }, { "epoch": 0.02, "learning_rate": 4.993931475840261e-05, "loss": 1.0896, "step": 22281 }, { "epoch": 0.02, "learning_rate": 4.9939309288672334e-05, "loss": 1.225, "step": 22282 }, { "epoch": 0.02, "learning_rate": 4.993930381869588e-05, "loss": 1.1255, "step": 22283 }, { "epoch": 0.02, "learning_rate": 4.993929834847323e-05, "loss": 0.9771, "step": 22284 }, { "epoch": 0.02, "learning_rate": 4.99392928780044e-05, "loss": 0.567, "step": 22285 }, { "epoch": 0.02, "learning_rate": 4.9939287407289365e-05, "loss": 0.6939, "step": 22286 }, { "epoch": 0.02, "learning_rate": 4.993928193632814e-05, "loss": 1.2104, "step": 22287 }, { "epoch": 0.02, "learning_rate": 4.993927646512073e-05, "loss": 0.8033, "step": 22288 }, { "epoch": 0.02, "learning_rate": 4.993927099366714e-05, "loss": 1.0041, "step": 22289 }, { "epoch": 0.02, "learning_rate": 4.9939265521967346e-05, "loss": 1.1004, "step": 22290 }, { "epoch": 0.02, "learning_rate": 4.993926005002137e-05, "loss": 1.037, "step": 22291 }, { "epoch": 0.02, "learning_rate": 4.99392545778292e-05, "loss": 0.9785, "step": 22292 }, { "epoch": 0.02, "learning_rate": 4.993924910539084e-05, "loss": 0.9523, "step": 22293 }, { "epoch": 0.02, "learning_rate": 4.993924363270629e-05, "loss": 1.2054, "step": 22294 }, { "epoch": 0.02, "learning_rate": 4.9939238159775556e-05, "loss": 1.2382, "step": 22295 }, { "epoch": 0.02, "learning_rate": 4.9939232686598624e-05, "loss": 1.3422, "step": 22296 }, { "epoch": 0.02, "learning_rate": 4.993922721317551e-05, "loss": 1.1719, "step": 22297 }, { "epoch": 0.02, "learning_rate": 4.99392217395062e-05, "loss": 1.2362, "step": 22298 }, { "epoch": 0.02, "learning_rate": 4.9939216265590706e-05, "loss": 1.3537, "step": 22299 }, { "epoch": 0.02, "learning_rate": 4.993921079142903e-05, "loss": 0.9478, "step": 22300 }, { "epoch": 0.02, "learning_rate": 4.993920531702114e-05, "loss": 1.0678, "step": 22301 }, { "epoch": 0.02, "learning_rate": 4.993919984236708e-05, "loss": 0.8433, "step": 22302 }, { "epoch": 0.02, "learning_rate": 4.993919436746682e-05, "loss": 0.9548, "step": 22303 }, { "epoch": 0.02, "learning_rate": 4.993918889232039e-05, "loss": 0.9891, "step": 22304 }, { "epoch": 0.02, "learning_rate": 4.9939183416927755e-05, "loss": 1.6143, "step": 22305 }, { "epoch": 0.02, "learning_rate": 4.9939177941288925e-05, "loss": 1.2971, "step": 22306 }, { "epoch": 0.02, "learning_rate": 4.993917246540392e-05, "loss": 0.8234, "step": 22307 }, { "epoch": 0.02, "learning_rate": 4.993916698927272e-05, "loss": 0.8769, "step": 22308 }, { "epoch": 0.02, "learning_rate": 4.9939161512895334e-05, "loss": 1.0315, "step": 22309 }, { "epoch": 0.02, "learning_rate": 4.993915603627176e-05, "loss": 0.9439, "step": 22310 }, { "epoch": 0.02, "learning_rate": 4.993915055940199e-05, "loss": 0.9394, "step": 22311 }, { "epoch": 0.02, "learning_rate": 4.993914508228604e-05, "loss": 1.2126, "step": 22312 }, { "epoch": 0.02, "learning_rate": 4.993913960492389e-05, "loss": 1.1805, "step": 22313 }, { "epoch": 0.02, "learning_rate": 4.9939134127315554e-05, "loss": 1.1278, "step": 22314 }, { "epoch": 0.02, "learning_rate": 4.993912864946104e-05, "loss": 1.0248, "step": 22315 }, { "epoch": 0.02, "learning_rate": 4.993912317136033e-05, "loss": 0.7719, "step": 22316 }, { "epoch": 0.02, "learning_rate": 4.9939117693013434e-05, "loss": 0.6507, "step": 22317 }, { "epoch": 0.02, "learning_rate": 4.993911221442035e-05, "loss": 0.8241, "step": 22318 }, { "epoch": 0.02, "learning_rate": 4.993910673558107e-05, "loss": 1.1863, "step": 22319 }, { "epoch": 0.02, "learning_rate": 4.993910125649561e-05, "loss": 1.3555, "step": 22320 }, { "epoch": 0.02, "learning_rate": 4.9939095777163955e-05, "loss": 1.0136, "step": 22321 }, { "epoch": 0.02, "learning_rate": 4.993909029758612e-05, "loss": 1.168, "step": 22322 }, { "epoch": 0.02, "learning_rate": 4.993908481776209e-05, "loss": 1.2779, "step": 22323 }, { "epoch": 0.02, "learning_rate": 4.993907933769188e-05, "loss": 1.253, "step": 22324 }, { "epoch": 0.02, "learning_rate": 4.9939073857375474e-05, "loss": 0.9574, "step": 22325 }, { "epoch": 0.02, "learning_rate": 4.993906837681288e-05, "loss": 1.0986, "step": 22326 }, { "epoch": 0.02, "learning_rate": 4.9939062896004096e-05, "loss": 0.9036, "step": 22327 }, { "epoch": 0.02, "learning_rate": 4.993905741494913e-05, "loss": 0.7313, "step": 22328 }, { "epoch": 0.02, "learning_rate": 4.993905193364797e-05, "loss": 1.087, "step": 22329 }, { "epoch": 0.02, "learning_rate": 4.993904645210063e-05, "loss": 1.033, "step": 22330 }, { "epoch": 0.02, "learning_rate": 4.99390409703071e-05, "loss": 1.0377, "step": 22331 }, { "epoch": 0.02, "learning_rate": 4.9939035488267385e-05, "loss": 1.1691, "step": 22332 }, { "epoch": 0.02, "learning_rate": 4.9939030005981475e-05, "loss": 0.8152, "step": 22333 }, { "epoch": 0.02, "learning_rate": 4.9939024523449376e-05, "loss": 0.8757, "step": 22334 }, { "epoch": 0.02, "learning_rate": 4.9939019040671096e-05, "loss": 0.9749, "step": 22335 }, { "epoch": 0.02, "learning_rate": 4.993901355764663e-05, "loss": 1.4952, "step": 22336 }, { "epoch": 0.02, "learning_rate": 4.993900807437597e-05, "loss": 1.6821, "step": 22337 }, { "epoch": 0.02, "learning_rate": 4.993900259085913e-05, "loss": 1.5565, "step": 22338 }, { "epoch": 0.02, "learning_rate": 4.99389971070961e-05, "loss": 1.1564, "step": 22339 }, { "epoch": 0.02, "learning_rate": 4.9938991623086875e-05, "loss": 1.1183, "step": 22340 }, { "epoch": 0.02, "learning_rate": 4.9938986138831464e-05, "loss": 0.6658, "step": 22341 }, { "epoch": 0.02, "learning_rate": 4.993898065432988e-05, "loss": 0.3195, "step": 22342 }, { "epoch": 0.02, "learning_rate": 4.99389751695821e-05, "loss": 1.0331, "step": 22343 }, { "epoch": 0.02, "learning_rate": 4.9938969684588134e-05, "loss": 1.1784, "step": 22344 }, { "epoch": 0.02, "learning_rate": 4.9938964199347977e-05, "loss": 1.1786, "step": 22345 }, { "epoch": 0.02, "learning_rate": 4.993895871386163e-05, "loss": 1.2127, "step": 22346 }, { "epoch": 0.02, "learning_rate": 4.99389532281291e-05, "loss": 1.0877, "step": 22347 }, { "epoch": 0.02, "learning_rate": 4.9938947742150386e-05, "loss": 1.0797, "step": 22348 }, { "epoch": 0.02, "learning_rate": 4.9938942255925495e-05, "loss": 1.1066, "step": 22349 }, { "epoch": 0.02, "learning_rate": 4.99389367694544e-05, "loss": 1.0485, "step": 22350 }, { "epoch": 0.02, "learning_rate": 4.993893128273712e-05, "loss": 1.1021, "step": 22351 }, { "epoch": 0.02, "learning_rate": 4.993892579577366e-05, "loss": 1.1608, "step": 22352 }, { "epoch": 0.02, "learning_rate": 4.993892030856402e-05, "loss": 0.7678, "step": 22353 }, { "epoch": 0.02, "learning_rate": 4.9938914821108185e-05, "loss": 0.8466, "step": 22354 }, { "epoch": 0.02, "learning_rate": 4.9938909333406156e-05, "loss": 0.9763, "step": 22355 }, { "epoch": 0.02, "learning_rate": 4.993890384545795e-05, "loss": 1.0229, "step": 22356 }, { "epoch": 0.02, "learning_rate": 4.993889835726356e-05, "loss": 1.1285, "step": 22357 }, { "epoch": 0.02, "learning_rate": 4.993889286882298e-05, "loss": 1.1387, "step": 22358 }, { "epoch": 0.02, "learning_rate": 4.993888738013621e-05, "loss": 1.1313, "step": 22359 }, { "epoch": 0.02, "learning_rate": 4.993888189120326e-05, "loss": 0.8278, "step": 22360 }, { "epoch": 0.02, "learning_rate": 4.9938876402024115e-05, "loss": 1.1724, "step": 22361 }, { "epoch": 0.02, "learning_rate": 4.993887091259879e-05, "loss": 1.5463, "step": 22362 }, { "epoch": 0.02, "learning_rate": 4.9938865422927286e-05, "loss": 1.1391, "step": 22363 }, { "epoch": 0.02, "learning_rate": 4.993885993300959e-05, "loss": 1.194, "step": 22364 }, { "epoch": 0.02, "learning_rate": 4.993885444284571e-05, "loss": 1.1724, "step": 22365 }, { "epoch": 0.02, "learning_rate": 4.9938848952435635e-05, "loss": 0.8531, "step": 22366 }, { "epoch": 0.02, "learning_rate": 4.9938843461779386e-05, "loss": 0.996, "step": 22367 }, { "epoch": 0.02, "learning_rate": 4.993883797087694e-05, "loss": 1.2535, "step": 22368 }, { "epoch": 0.02, "learning_rate": 4.9938832479728315e-05, "loss": 1.2504, "step": 22369 }, { "epoch": 0.02, "learning_rate": 4.99388269883335e-05, "loss": 0.9097, "step": 22370 }, { "epoch": 0.02, "learning_rate": 4.99388214966925e-05, "loss": 0.8968, "step": 22371 }, { "epoch": 0.02, "learning_rate": 4.993881600480532e-05, "loss": 0.8397, "step": 22372 }, { "epoch": 0.02, "learning_rate": 4.9938810512671954e-05, "loss": 0.7432, "step": 22373 }, { "epoch": 0.02, "learning_rate": 4.99388050202924e-05, "loss": 0.9595, "step": 22374 }, { "epoch": 0.02, "learning_rate": 4.9938799527666656e-05, "loss": 1.0751, "step": 22375 }, { "epoch": 0.02, "learning_rate": 4.993879403479473e-05, "loss": 1.315, "step": 22376 }, { "epoch": 0.02, "learning_rate": 4.993878854167663e-05, "loss": 1.271, "step": 22377 }, { "epoch": 0.02, "learning_rate": 4.993878304831233e-05, "loss": 1.112, "step": 22378 }, { "epoch": 0.02, "learning_rate": 4.9938777554701854e-05, "loss": 0.9374, "step": 22379 }, { "epoch": 0.02, "learning_rate": 4.993877206084518e-05, "loss": 1.1241, "step": 22380 }, { "epoch": 0.02, "learning_rate": 4.9938766566742336e-05, "loss": 1.0031, "step": 22381 }, { "epoch": 0.02, "learning_rate": 4.99387610723933e-05, "loss": 1.2828, "step": 22382 }, { "epoch": 0.02, "learning_rate": 4.993875557779808e-05, "loss": 1.1168, "step": 22383 }, { "epoch": 0.02, "learning_rate": 4.993875008295667e-05, "loss": 1.0985, "step": 22384 }, { "epoch": 0.02, "learning_rate": 4.9938744587869086e-05, "loss": 1.2524, "step": 22385 }, { "epoch": 0.02, "learning_rate": 4.993873909253531e-05, "loss": 0.3048, "step": 22386 }, { "epoch": 0.02, "learning_rate": 4.9938733596955354e-05, "loss": 0.3393, "step": 22387 }, { "epoch": 0.02, "learning_rate": 4.993872810112921e-05, "loss": 0.9326, "step": 22388 }, { "epoch": 0.02, "learning_rate": 4.9938722605056875e-05, "loss": 1.8575, "step": 22389 }, { "epoch": 0.02, "learning_rate": 4.9938717108738366e-05, "loss": 1.0155, "step": 22390 }, { "epoch": 0.02, "learning_rate": 4.993871161217367e-05, "loss": 0.8362, "step": 22391 }, { "epoch": 0.02, "learning_rate": 4.993870611536279e-05, "loss": 1.1972, "step": 22392 }, { "epoch": 0.02, "learning_rate": 4.9938700618305723e-05, "loss": 0.9672, "step": 22393 }, { "epoch": 0.02, "learning_rate": 4.993869512100247e-05, "loss": 0.7031, "step": 22394 }, { "epoch": 0.02, "learning_rate": 4.993868962345303e-05, "loss": 0.1834, "step": 22395 }, { "epoch": 0.02, "learning_rate": 4.993868412565741e-05, "loss": 0.1404, "step": 22396 }, { "epoch": 0.02, "learning_rate": 4.993867862761561e-05, "loss": 0.2524, "step": 22397 }, { "epoch": 0.02, "learning_rate": 4.9938673129327616e-05, "loss": 0.8606, "step": 22398 }, { "epoch": 0.02, "learning_rate": 4.993866763079345e-05, "loss": 1.1858, "step": 22399 }, { "epoch": 0.02, "learning_rate": 4.993866213201309e-05, "loss": 0.7368, "step": 22400 }, { "epoch": 0.02, "learning_rate": 4.993865663298656e-05, "loss": 0.599, "step": 22401 }, { "epoch": 0.02, "learning_rate": 4.993865113371383e-05, "loss": 0.6903, "step": 22402 }, { "epoch": 0.02, "learning_rate": 4.9938645634194926e-05, "loss": 0.6282, "step": 22403 }, { "epoch": 0.02, "learning_rate": 4.9938640134429834e-05, "loss": 0.6495, "step": 22404 }, { "epoch": 0.02, "learning_rate": 4.993863463441856e-05, "loss": 0.4325, "step": 22405 }, { "epoch": 0.02, "learning_rate": 4.99386291341611e-05, "loss": 0.6941, "step": 22406 }, { "epoch": 0.02, "learning_rate": 4.993862363365745e-05, "loss": 0.4224, "step": 22407 }, { "epoch": 0.02, "learning_rate": 4.9938618132907634e-05, "loss": 0.6211, "step": 22408 }, { "epoch": 0.02, "learning_rate": 4.993861263191162e-05, "loss": 0.7563, "step": 22409 }, { "epoch": 0.02, "learning_rate": 4.9938607130669424e-05, "loss": 0.7944, "step": 22410 }, { "epoch": 0.02, "learning_rate": 4.9938601629181046e-05, "loss": 0.5879, "step": 22411 }, { "epoch": 0.02, "learning_rate": 4.9938596127446494e-05, "loss": 0.535, "step": 22412 }, { "epoch": 0.02, "learning_rate": 4.9938590625465746e-05, "loss": 0.3786, "step": 22413 }, { "epoch": 0.02, "learning_rate": 4.9938585123238824e-05, "loss": 0.3783, "step": 22414 }, { "epoch": 0.02, "learning_rate": 4.993857962076571e-05, "loss": 0.2556, "step": 22415 }, { "epoch": 0.02, "learning_rate": 4.993857411804642e-05, "loss": 0.6238, "step": 22416 }, { "epoch": 0.02, "learning_rate": 4.9938568615080946e-05, "loss": 0.9536, "step": 22417 }, { "epoch": 0.02, "learning_rate": 4.993856311186928e-05, "loss": 1.3165, "step": 22418 }, { "epoch": 0.02, "learning_rate": 4.993855760841144e-05, "loss": 1.2405, "step": 22419 }, { "epoch": 0.02, "learning_rate": 4.993855210470741e-05, "loss": 1.5025, "step": 22420 }, { "epoch": 0.02, "learning_rate": 4.9938546600757205e-05, "loss": 0.9054, "step": 22421 }, { "epoch": 0.02, "learning_rate": 4.993854109656082e-05, "loss": 0.8825, "step": 22422 }, { "epoch": 0.02, "learning_rate": 4.9938535592118245e-05, "loss": 0.9591, "step": 22423 }, { "epoch": 0.02, "learning_rate": 4.993853008742949e-05, "loss": 1.1853, "step": 22424 }, { "epoch": 0.02, "learning_rate": 4.993852458249455e-05, "loss": 0.9967, "step": 22425 }, { "epoch": 0.02, "learning_rate": 4.993851907731343e-05, "loss": 1.0152, "step": 22426 }, { "epoch": 0.02, "learning_rate": 4.9938513571886124e-05, "loss": 1.058, "step": 22427 }, { "epoch": 0.02, "learning_rate": 4.993850806621264e-05, "loss": 1.1433, "step": 22428 }, { "epoch": 0.02, "learning_rate": 4.993850256029297e-05, "loss": 1.1891, "step": 22429 }, { "epoch": 0.02, "learning_rate": 4.993849705412712e-05, "loss": 1.3641, "step": 22430 }, { "epoch": 0.02, "learning_rate": 4.9938491547715083e-05, "loss": 1.0699, "step": 22431 }, { "epoch": 0.02, "learning_rate": 4.993848604105687e-05, "loss": 1.2606, "step": 22432 }, { "epoch": 0.02, "learning_rate": 4.9938480534152477e-05, "loss": 0.4229, "step": 22433 }, { "epoch": 0.02, "learning_rate": 4.9938475027001894e-05, "loss": 0.3915, "step": 22434 }, { "epoch": 0.02, "learning_rate": 4.993846951960513e-05, "loss": 0.3044, "step": 22435 }, { "epoch": 0.02, "learning_rate": 4.993846401196218e-05, "loss": 0.2466, "step": 22436 }, { "epoch": 0.02, "learning_rate": 4.993845850407306e-05, "loss": 0.138, "step": 22437 }, { "epoch": 0.02, "learning_rate": 4.993845299593775e-05, "loss": 0.0789, "step": 22438 }, { "epoch": 0.02, "learning_rate": 4.993844748755626e-05, "loss": 0.058, "step": 22439 }, { "epoch": 0.02, "learning_rate": 4.993844197892858e-05, "loss": 0.0846, "step": 22440 }, { "epoch": 0.02, "learning_rate": 4.993843647005474e-05, "loss": 1.1509, "step": 22441 }, { "epoch": 0.02, "learning_rate": 4.99384309609347e-05, "loss": 1.1969, "step": 22442 }, { "epoch": 0.02, "learning_rate": 4.993842545156848e-05, "loss": 1.066, "step": 22443 }, { "epoch": 0.02, "learning_rate": 4.993841994195608e-05, "loss": 1.005, "step": 22444 }, { "epoch": 0.02, "learning_rate": 4.99384144320975e-05, "loss": 1.2684, "step": 22445 }, { "epoch": 0.02, "learning_rate": 4.993840892199274e-05, "loss": 1.1741, "step": 22446 }, { "epoch": 0.02, "learning_rate": 4.99384034116418e-05, "loss": 1.0013, "step": 22447 }, { "epoch": 0.02, "learning_rate": 4.9938397901044675e-05, "loss": 1.0643, "step": 22448 }, { "epoch": 0.02, "learning_rate": 4.993839239020137e-05, "loss": 1.4342, "step": 22449 }, { "epoch": 0.02, "learning_rate": 4.993838687911188e-05, "loss": 1.126, "step": 22450 }, { "epoch": 0.02, "learning_rate": 4.993838136777621e-05, "loss": 0.7729, "step": 22451 }, { "epoch": 0.02, "learning_rate": 4.993837585619436e-05, "loss": 1.119, "step": 22452 }, { "epoch": 0.02, "learning_rate": 4.993837034436633e-05, "loss": 1.0293, "step": 22453 }, { "epoch": 0.02, "learning_rate": 4.993836483229212e-05, "loss": 1.2539, "step": 22454 }, { "epoch": 0.02, "learning_rate": 4.993835931997173e-05, "loss": 1.2863, "step": 22455 }, { "epoch": 0.02, "learning_rate": 4.993835380740515e-05, "loss": 0.8858, "step": 22456 }, { "epoch": 0.02, "learning_rate": 4.9938348294592395e-05, "loss": 0.9769, "step": 22457 }, { "epoch": 0.02, "learning_rate": 4.993834278153346e-05, "loss": 1.0828, "step": 22458 }, { "epoch": 0.02, "learning_rate": 4.993833726822834e-05, "loss": 1.1159, "step": 22459 }, { "epoch": 0.02, "learning_rate": 4.993833175467705e-05, "loss": 1.2737, "step": 22460 }, { "epoch": 0.02, "learning_rate": 4.993832624087956e-05, "loss": 1.2609, "step": 22461 }, { "epoch": 0.02, "learning_rate": 4.9938320726835906e-05, "loss": 1.0574, "step": 22462 }, { "epoch": 0.02, "learning_rate": 4.993831521254607e-05, "loss": 1.1267, "step": 22463 }, { "epoch": 0.02, "learning_rate": 4.9938309698010044e-05, "loss": 0.4327, "step": 22464 }, { "epoch": 0.02, "learning_rate": 4.993830418322784e-05, "loss": 0.8185, "step": 22465 }, { "epoch": 0.02, "learning_rate": 4.993829866819947e-05, "loss": 1.2331, "step": 22466 }, { "epoch": 0.02, "learning_rate": 4.993829315292491e-05, "loss": 0.9731, "step": 22467 }, { "epoch": 0.02, "learning_rate": 4.9938287637404165e-05, "loss": 1.0528, "step": 22468 }, { "epoch": 0.02, "learning_rate": 4.9938282121637245e-05, "loss": 1.3864, "step": 22469 }, { "epoch": 0.02, "learning_rate": 4.993827660562414e-05, "loss": 0.9818, "step": 22470 }, { "epoch": 0.02, "learning_rate": 4.993827108936486e-05, "loss": 1.0552, "step": 22471 }, { "epoch": 0.02, "learning_rate": 4.9938265572859394e-05, "loss": 0.9404, "step": 22472 }, { "epoch": 0.02, "learning_rate": 4.9938260056107754e-05, "loss": 1.3186, "step": 22473 }, { "epoch": 0.02, "learning_rate": 4.993825453910993e-05, "loss": 1.7446, "step": 22474 }, { "epoch": 0.02, "learning_rate": 4.993824902186593e-05, "loss": 1.4857, "step": 22475 }, { "epoch": 0.02, "learning_rate": 4.993824350437575e-05, "loss": 1.1708, "step": 22476 }, { "epoch": 0.02, "learning_rate": 4.993823798663938e-05, "loss": 1.0398, "step": 22477 }, { "epoch": 0.02, "learning_rate": 4.993823246865684e-05, "loss": 1.0995, "step": 22478 }, { "epoch": 0.02, "learning_rate": 4.9938226950428125e-05, "loss": 1.0324, "step": 22479 }, { "epoch": 0.02, "learning_rate": 4.993822143195322e-05, "loss": 1.3237, "step": 22480 }, { "epoch": 0.02, "learning_rate": 4.993821591323214e-05, "loss": 1.2396, "step": 22481 }, { "epoch": 0.02, "learning_rate": 4.993821039426488e-05, "loss": 1.3239, "step": 22482 }, { "epoch": 0.02, "learning_rate": 4.993820487505144e-05, "loss": 1.2095, "step": 22483 }, { "epoch": 0.02, "learning_rate": 4.993819935559182e-05, "loss": 1.3323, "step": 22484 }, { "epoch": 0.02, "learning_rate": 4.9938193835886024e-05, "loss": 0.9935, "step": 22485 }, { "epoch": 0.02, "learning_rate": 4.993818831593404e-05, "loss": 0.6312, "step": 22486 }, { "epoch": 0.02, "learning_rate": 4.9938182795735886e-05, "loss": 1.2235, "step": 22487 }, { "epoch": 0.02, "learning_rate": 4.993817727529155e-05, "loss": 1.0446, "step": 22488 }, { "epoch": 0.02, "learning_rate": 4.993817175460103e-05, "loss": 1.1664, "step": 22489 }, { "epoch": 0.02, "learning_rate": 4.993816623366434e-05, "loss": 0.8425, "step": 22490 }, { "epoch": 0.02, "learning_rate": 4.9938160712481465e-05, "loss": 1.0723, "step": 22491 }, { "epoch": 0.02, "learning_rate": 4.993815519105241e-05, "loss": 1.0579, "step": 22492 }, { "epoch": 0.02, "learning_rate": 4.9938149669377175e-05, "loss": 1.1235, "step": 22493 }, { "epoch": 0.02, "learning_rate": 4.993814414745577e-05, "loss": 0.8712, "step": 22494 }, { "epoch": 0.02, "learning_rate": 4.993813862528818e-05, "loss": 1.0406, "step": 22495 }, { "epoch": 0.02, "learning_rate": 4.993813310287441e-05, "loss": 1.0186, "step": 22496 }, { "epoch": 0.02, "learning_rate": 4.993812758021446e-05, "loss": 1.2051, "step": 22497 }, { "epoch": 0.02, "learning_rate": 4.9938122057308334e-05, "loss": 1.0523, "step": 22498 }, { "epoch": 0.02, "learning_rate": 4.993811653415603e-05, "loss": 1.1873, "step": 22499 }, { "epoch": 0.02, "learning_rate": 4.993811101075755e-05, "loss": 1.3007, "step": 22500 }, { "epoch": 0.02, "eval_loss": 1.0252221822738647, "eval_runtime": 85.1725, "eval_samples_per_second": 16.261, "eval_steps_per_second": 4.074, "step": 22500 }, { "epoch": 0.02, "learning_rate": 4.993810548711288e-05, "loss": 1.2051, "step": 22501 }, { "epoch": 0.02, "learning_rate": 4.993809996322205e-05, "loss": 1.1607, "step": 22502 }, { "epoch": 0.02, "learning_rate": 4.9938094439085026e-05, "loss": 0.8914, "step": 22503 }, { "epoch": 0.02, "learning_rate": 4.993808891470183e-05, "loss": 1.1283, "step": 22504 }, { "epoch": 0.02, "learning_rate": 4.993808339007246e-05, "loss": 1.5721, "step": 22505 }, { "epoch": 0.02, "learning_rate": 4.9938077865196904e-05, "loss": 1.2308, "step": 22506 }, { "epoch": 0.02, "learning_rate": 4.993807234007517e-05, "loss": 1.2162, "step": 22507 }, { "epoch": 0.02, "learning_rate": 4.9938066814707265e-05, "loss": 1.1442, "step": 22508 }, { "epoch": 0.02, "learning_rate": 4.9938061289093174e-05, "loss": 1.2297, "step": 22509 }, { "epoch": 0.02, "learning_rate": 4.993805576323291e-05, "loss": 1.6944, "step": 22510 }, { "epoch": 0.02, "learning_rate": 4.9938050237126466e-05, "loss": 1.0412, "step": 22511 }, { "epoch": 0.02, "learning_rate": 4.993804471077384e-05, "loss": 1.1968, "step": 22512 }, { "epoch": 0.02, "learning_rate": 4.9938039184175045e-05, "loss": 1.2116, "step": 22513 }, { "epoch": 0.02, "learning_rate": 4.9938033657330065e-05, "loss": 1.2242, "step": 22514 }, { "epoch": 0.02, "learning_rate": 4.993802813023891e-05, "loss": 0.3192, "step": 22515 }, { "epoch": 0.02, "learning_rate": 4.993802260290158e-05, "loss": 0.1779, "step": 22516 }, { "epoch": 0.02, "learning_rate": 4.993801707531807e-05, "loss": 0.1543, "step": 22517 }, { "epoch": 0.02, "learning_rate": 4.993801154748838e-05, "loss": 0.1213, "step": 22518 }, { "epoch": 0.02, "learning_rate": 4.993800601941251e-05, "loss": 0.0841, "step": 22519 }, { "epoch": 0.02, "learning_rate": 4.993800049109047e-05, "loss": 0.0696, "step": 22520 }, { "epoch": 0.02, "learning_rate": 4.993799496252225e-05, "loss": 0.0594, "step": 22521 }, { "epoch": 0.02, "learning_rate": 4.993798943370785e-05, "loss": 0.0512, "step": 22522 }, { "epoch": 0.02, "learning_rate": 4.993798390464728e-05, "loss": 0.0433, "step": 22523 }, { "epoch": 0.02, "learning_rate": 4.993797837534052e-05, "loss": 0.0422, "step": 22524 }, { "epoch": 0.02, "learning_rate": 4.99379728457876e-05, "loss": 0.0397, "step": 22525 }, { "epoch": 0.02, "learning_rate": 4.9937967315988485e-05, "loss": 0.0448, "step": 22526 }, { "epoch": 0.02, "learning_rate": 4.99379617859432e-05, "loss": 0.0389, "step": 22527 }, { "epoch": 0.02, "learning_rate": 4.993795625565174e-05, "loss": 0.0421, "step": 22528 }, { "epoch": 0.02, "learning_rate": 4.99379507251141e-05, "loss": 0.034, "step": 22529 }, { "epoch": 0.02, "learning_rate": 4.993794519433029e-05, "loss": 0.0353, "step": 22530 }, { "epoch": 0.02, "learning_rate": 4.9937939663300295e-05, "loss": 0.0363, "step": 22531 }, { "epoch": 0.02, "learning_rate": 4.993793413202412e-05, "loss": 0.0348, "step": 22532 }, { "epoch": 0.02, "learning_rate": 4.9937928600501784e-05, "loss": 0.8558, "step": 22533 }, { "epoch": 0.02, "learning_rate": 4.993792306873326e-05, "loss": 1.8739, "step": 22534 }, { "epoch": 0.02, "learning_rate": 4.993791753671856e-05, "loss": 1.8322, "step": 22535 }, { "epoch": 0.02, "learning_rate": 4.9937912004457685e-05, "loss": 1.8161, "step": 22536 }, { "epoch": 0.02, "learning_rate": 4.9937906471950636e-05, "loss": 1.4508, "step": 22537 }, { "epoch": 0.02, "learning_rate": 4.9937900939197405e-05, "loss": 1.0158, "step": 22538 }, { "epoch": 0.02, "learning_rate": 4.9937895406198e-05, "loss": 1.3069, "step": 22539 }, { "epoch": 0.02, "learning_rate": 4.993788987295242e-05, "loss": 1.2899, "step": 22540 }, { "epoch": 0.02, "learning_rate": 4.993788433946066e-05, "loss": 1.3676, "step": 22541 }, { "epoch": 0.02, "learning_rate": 4.9937878805722725e-05, "loss": 1.4553, "step": 22542 }, { "epoch": 0.02, "learning_rate": 4.993787327173861e-05, "loss": 0.9864, "step": 22543 }, { "epoch": 0.02, "learning_rate": 4.9937867737508326e-05, "loss": 1.1751, "step": 22544 }, { "epoch": 0.02, "learning_rate": 4.9937862203031864e-05, "loss": 1.1465, "step": 22545 }, { "epoch": 0.02, "learning_rate": 4.993785666830923e-05, "loss": 1.0132, "step": 22546 }, { "epoch": 0.02, "learning_rate": 4.993785113334041e-05, "loss": 0.9624, "step": 22547 }, { "epoch": 0.02, "learning_rate": 4.993784559812542e-05, "loss": 0.458, "step": 22548 }, { "epoch": 0.02, "learning_rate": 4.9937840062664254e-05, "loss": 0.5249, "step": 22549 }, { "epoch": 0.02, "learning_rate": 4.993783452695691e-05, "loss": 0.6154, "step": 22550 }, { "epoch": 0.02, "learning_rate": 4.993782899100339e-05, "loss": 1.3439, "step": 22551 }, { "epoch": 0.02, "learning_rate": 4.99378234548037e-05, "loss": 1.0807, "step": 22552 }, { "epoch": 0.02, "learning_rate": 4.993781791835783e-05, "loss": 1.0568, "step": 22553 }, { "epoch": 0.02, "learning_rate": 4.993781238166578e-05, "loss": 1.0271, "step": 22554 }, { "epoch": 0.02, "learning_rate": 4.9937806844727565e-05, "loss": 1.301, "step": 22555 }, { "epoch": 0.02, "learning_rate": 4.9937801307543166e-05, "loss": 1.2428, "step": 22556 }, { "epoch": 0.02, "learning_rate": 4.993779577011259e-05, "loss": 1.3241, "step": 22557 }, { "epoch": 0.02, "learning_rate": 4.993779023243584e-05, "loss": 1.1664, "step": 22558 }, { "epoch": 0.02, "learning_rate": 4.9937784694512926e-05, "loss": 1.0907, "step": 22559 }, { "epoch": 0.02, "learning_rate": 4.993777915634383e-05, "loss": 1.214, "step": 22560 }, { "epoch": 0.02, "learning_rate": 4.9937773617928554e-05, "loss": 1.1309, "step": 22561 }, { "epoch": 0.02, "learning_rate": 4.9937768079267106e-05, "loss": 1.2536, "step": 22562 }, { "epoch": 0.02, "learning_rate": 4.9937762540359476e-05, "loss": 0.9744, "step": 22563 }, { "epoch": 0.02, "learning_rate": 4.9937757001205684e-05, "loss": 1.0108, "step": 22564 }, { "epoch": 0.02, "learning_rate": 4.993775146180571e-05, "loss": 0.9428, "step": 22565 }, { "epoch": 0.02, "learning_rate": 4.993774592215956e-05, "loss": 1.266, "step": 22566 }, { "epoch": 0.02, "learning_rate": 4.993774038226724e-05, "loss": 1.2137, "step": 22567 }, { "epoch": 0.02, "learning_rate": 4.9937734842128736e-05, "loss": 0.9608, "step": 22568 }, { "epoch": 0.02, "learning_rate": 4.993772930174406e-05, "loss": 1.1149, "step": 22569 }, { "epoch": 0.02, "learning_rate": 4.993772376111322e-05, "loss": 1.1658, "step": 22570 }, { "epoch": 0.02, "learning_rate": 4.9937718220236194e-05, "loss": 1.1381, "step": 22571 }, { "epoch": 0.02, "learning_rate": 4.9937712679112996e-05, "loss": 1.1104, "step": 22572 }, { "epoch": 0.02, "learning_rate": 4.993770713774363e-05, "loss": 1.0456, "step": 22573 }, { "epoch": 0.02, "learning_rate": 4.993770159612808e-05, "loss": 0.9538, "step": 22574 }, { "epoch": 0.02, "learning_rate": 4.9937696054266355e-05, "loss": 1.0949, "step": 22575 }, { "epoch": 0.02, "learning_rate": 4.9937690512158465e-05, "loss": 0.9993, "step": 22576 }, { "epoch": 0.02, "learning_rate": 4.9937684969804394e-05, "loss": 1.0598, "step": 22577 }, { "epoch": 0.02, "learning_rate": 4.993767942720415e-05, "loss": 0.911, "step": 22578 }, { "epoch": 0.02, "learning_rate": 4.9937673884357726e-05, "loss": 1.1522, "step": 22579 }, { "epoch": 0.02, "learning_rate": 4.993766834126514e-05, "loss": 1.2352, "step": 22580 }, { "epoch": 0.02, "learning_rate": 4.993766279792637e-05, "loss": 0.928, "step": 22581 }, { "epoch": 0.02, "learning_rate": 4.993765725434143e-05, "loss": 1.0037, "step": 22582 }, { "epoch": 0.02, "learning_rate": 4.993765171051031e-05, "loss": 1.1558, "step": 22583 }, { "epoch": 0.02, "learning_rate": 4.993764616643303e-05, "loss": 0.8408, "step": 22584 }, { "epoch": 0.02, "learning_rate": 4.9937640622109566e-05, "loss": 1.0436, "step": 22585 }, { "epoch": 0.02, "learning_rate": 4.993763507753993e-05, "loss": 1.0615, "step": 22586 }, { "epoch": 0.02, "learning_rate": 4.993762953272412e-05, "loss": 0.9861, "step": 22587 }, { "epoch": 0.02, "learning_rate": 4.993762398766213e-05, "loss": 1.1553, "step": 22588 }, { "epoch": 0.02, "learning_rate": 4.9937618442353975e-05, "loss": 1.164, "step": 22589 }, { "epoch": 0.02, "learning_rate": 4.9937612896799644e-05, "loss": 1.1346, "step": 22590 }, { "epoch": 0.02, "learning_rate": 4.993760735099914e-05, "loss": 1.0086, "step": 22591 }, { "epoch": 0.02, "learning_rate": 4.993760180495246e-05, "loss": 0.9379, "step": 22592 }, { "epoch": 0.02, "learning_rate": 4.9937596258659607e-05, "loss": 1.0963, "step": 22593 }, { "epoch": 0.02, "learning_rate": 4.993759071212058e-05, "loss": 1.3239, "step": 22594 }, { "epoch": 0.02, "learning_rate": 4.993758516533538e-05, "loss": 1.1023, "step": 22595 }, { "epoch": 0.02, "learning_rate": 4.993757961830401e-05, "loss": 0.9658, "step": 22596 }, { "epoch": 0.02, "learning_rate": 4.993757407102647e-05, "loss": 0.9631, "step": 22597 }, { "epoch": 0.02, "learning_rate": 4.993756852350274e-05, "loss": 0.8693, "step": 22598 }, { "epoch": 0.02, "learning_rate": 4.993756297573285e-05, "loss": 1.0443, "step": 22599 }, { "epoch": 0.02, "learning_rate": 4.9937557427716785e-05, "loss": 1.0653, "step": 22600 }, { "epoch": 0.02, "learning_rate": 4.993755187945455e-05, "loss": 0.8518, "step": 22601 }, { "epoch": 0.02, "learning_rate": 4.993754633094614e-05, "loss": 1.1191, "step": 22602 }, { "epoch": 0.02, "learning_rate": 4.9937540782191556e-05, "loss": 1.3057, "step": 22603 }, { "epoch": 0.02, "learning_rate": 4.99375352331908e-05, "loss": 1.0854, "step": 22604 }, { "epoch": 0.02, "learning_rate": 4.993752968394387e-05, "loss": 1.1266, "step": 22605 }, { "epoch": 0.02, "learning_rate": 4.9937524134450766e-05, "loss": 1.2959, "step": 22606 }, { "epoch": 0.02, "learning_rate": 4.993751858471149e-05, "loss": 1.373, "step": 22607 }, { "epoch": 0.02, "learning_rate": 4.993751303472604e-05, "loss": 1.1307, "step": 22608 }, { "epoch": 0.02, "learning_rate": 4.993750748449442e-05, "loss": 1.1271, "step": 22609 }, { "epoch": 0.02, "learning_rate": 4.993750193401663e-05, "loss": 1.2049, "step": 22610 }, { "epoch": 0.02, "learning_rate": 4.993749638329266e-05, "loss": 0.8657, "step": 22611 }, { "epoch": 0.02, "learning_rate": 4.9937490832322526e-05, "loss": 0.8164, "step": 22612 }, { "epoch": 0.02, "learning_rate": 4.993748528110621e-05, "loss": 1.0827, "step": 22613 }, { "epoch": 0.02, "learning_rate": 4.993747972964373e-05, "loss": 1.0471, "step": 22614 }, { "epoch": 0.02, "learning_rate": 4.9937474177935076e-05, "loss": 0.9298, "step": 22615 }, { "epoch": 0.02, "learning_rate": 4.9937468625980245e-05, "loss": 1.0358, "step": 22616 }, { "epoch": 0.02, "learning_rate": 4.9937463073779246e-05, "loss": 0.6894, "step": 22617 }, { "epoch": 0.02, "learning_rate": 4.993745752133208e-05, "loss": 1.1969, "step": 22618 }, { "epoch": 0.02, "learning_rate": 4.993745196863873e-05, "loss": 0.6646, "step": 22619 }, { "epoch": 0.02, "learning_rate": 4.993744641569922e-05, "loss": 0.9686, "step": 22620 }, { "epoch": 0.02, "learning_rate": 4.993744086251353e-05, "loss": 1.3989, "step": 22621 }, { "epoch": 0.02, "learning_rate": 4.993743530908167e-05, "loss": 0.5438, "step": 22622 }, { "epoch": 0.02, "learning_rate": 4.9937429755403634e-05, "loss": 0.636, "step": 22623 }, { "epoch": 0.02, "learning_rate": 4.993742420147943e-05, "loss": 1.0628, "step": 22624 }, { "epoch": 0.02, "learning_rate": 4.993741864730906e-05, "loss": 1.2207, "step": 22625 }, { "epoch": 0.02, "learning_rate": 4.993741309289251e-05, "loss": 1.3564, "step": 22626 }, { "epoch": 0.02, "learning_rate": 4.99374075382298e-05, "loss": 1.1934, "step": 22627 }, { "epoch": 0.02, "learning_rate": 4.9937401983320904e-05, "loss": 0.9505, "step": 22628 }, { "epoch": 0.02, "learning_rate": 4.993739642816584e-05, "loss": 1.0725, "step": 22629 }, { "epoch": 0.02, "learning_rate": 4.993739087276461e-05, "loss": 0.9544, "step": 22630 }, { "epoch": 0.02, "learning_rate": 4.99373853171172e-05, "loss": 1.0974, "step": 22631 }, { "epoch": 0.02, "learning_rate": 4.993737976122362e-05, "loss": 1.1036, "step": 22632 }, { "epoch": 0.02, "learning_rate": 4.9937374205083884e-05, "loss": 1.3183, "step": 22633 }, { "epoch": 0.02, "learning_rate": 4.993736864869796e-05, "loss": 1.1828, "step": 22634 }, { "epoch": 0.02, "learning_rate": 4.993736309206587e-05, "loss": 1.1859, "step": 22635 }, { "epoch": 0.02, "learning_rate": 4.993735753518761e-05, "loss": 1.3911, "step": 22636 }, { "epoch": 0.02, "learning_rate": 4.993735197806318e-05, "loss": 1.2263, "step": 22637 }, { "epoch": 0.02, "learning_rate": 4.9937346420692574e-05, "loss": 1.4497, "step": 22638 }, { "epoch": 0.02, "learning_rate": 4.99373408630758e-05, "loss": 1.1906, "step": 22639 }, { "epoch": 0.02, "learning_rate": 4.993733530521286e-05, "loss": 1.1292, "step": 22640 }, { "epoch": 0.02, "learning_rate": 4.993732974710375e-05, "loss": 1.119, "step": 22641 }, { "epoch": 0.02, "learning_rate": 4.993732418874846e-05, "loss": 0.6021, "step": 22642 }, { "epoch": 0.02, "learning_rate": 4.9937318630147e-05, "loss": 1.2337, "step": 22643 }, { "epoch": 0.02, "learning_rate": 4.9937313071299374e-05, "loss": 0.8921, "step": 22644 }, { "epoch": 0.02, "learning_rate": 4.993730751220557e-05, "loss": 1.1524, "step": 22645 }, { "epoch": 0.02, "learning_rate": 4.99373019528656e-05, "loss": 1.1733, "step": 22646 }, { "epoch": 0.02, "learning_rate": 4.993729639327947e-05, "loss": 0.8224, "step": 22647 }, { "epoch": 0.02, "learning_rate": 4.9937290833447155e-05, "loss": 0.9192, "step": 22648 }, { "epoch": 0.02, "learning_rate": 4.9937285273368675e-05, "loss": 1.0595, "step": 22649 }, { "epoch": 0.02, "learning_rate": 4.993727971304402e-05, "loss": 1.2095, "step": 22650 }, { "epoch": 0.02, "learning_rate": 4.9937274152473204e-05, "loss": 1.2862, "step": 22651 }, { "epoch": 0.02, "learning_rate": 4.9937268591656206e-05, "loss": 1.1416, "step": 22652 }, { "epoch": 0.02, "learning_rate": 4.9937263030593054e-05, "loss": 1.0873, "step": 22653 }, { "epoch": 0.02, "learning_rate": 4.993725746928371e-05, "loss": 1.1832, "step": 22654 }, { "epoch": 0.02, "learning_rate": 4.993725190772821e-05, "loss": 1.2774, "step": 22655 }, { "epoch": 0.02, "learning_rate": 4.993724634592654e-05, "loss": 1.099, "step": 22656 }, { "epoch": 0.02, "learning_rate": 4.99372407838787e-05, "loss": 1.3481, "step": 22657 }, { "epoch": 0.02, "learning_rate": 4.993723522158469e-05, "loss": 1.4918, "step": 22658 }, { "epoch": 0.02, "learning_rate": 4.993722965904451e-05, "loss": 1.1506, "step": 22659 }, { "epoch": 0.02, "learning_rate": 4.993722409625815e-05, "loss": 1.1658, "step": 22660 }, { "epoch": 0.02, "learning_rate": 4.9937218533225634e-05, "loss": 0.724, "step": 22661 }, { "epoch": 0.02, "learning_rate": 4.993721296994695e-05, "loss": 1.2022, "step": 22662 }, { "epoch": 0.02, "learning_rate": 4.993720740642208e-05, "loss": 1.2759, "step": 22663 }, { "epoch": 0.02, "learning_rate": 4.993720184265105e-05, "loss": 1.2355, "step": 22664 }, { "epoch": 0.02, "learning_rate": 4.993719627863385e-05, "loss": 1.2328, "step": 22665 }, { "epoch": 0.02, "learning_rate": 4.993719071437048e-05, "loss": 1.0191, "step": 22666 }, { "epoch": 0.02, "learning_rate": 4.9937185149860944e-05, "loss": 1.076, "step": 22667 }, { "epoch": 0.02, "learning_rate": 4.993717958510523e-05, "loss": 1.0552, "step": 22668 }, { "epoch": 0.02, "learning_rate": 4.9937174020103354e-05, "loss": 1.0761, "step": 22669 }, { "epoch": 0.02, "learning_rate": 4.993716845485531e-05, "loss": 1.0696, "step": 22670 }, { "epoch": 0.02, "learning_rate": 4.993716288936109e-05, "loss": 0.9892, "step": 22671 }, { "epoch": 0.02, "learning_rate": 4.993715732362071e-05, "loss": 1.1709, "step": 22672 }, { "epoch": 0.02, "learning_rate": 4.9937151757634155e-05, "loss": 0.9486, "step": 22673 }, { "epoch": 0.02, "learning_rate": 4.993714619140143e-05, "loss": 1.5936, "step": 22674 }, { "epoch": 0.02, "learning_rate": 4.993714062492254e-05, "loss": 1.0812, "step": 22675 }, { "epoch": 0.02, "learning_rate": 4.9937135058197475e-05, "loss": 1.2617, "step": 22676 }, { "epoch": 0.02, "learning_rate": 4.993712949122624e-05, "loss": 0.8049, "step": 22677 }, { "epoch": 0.02, "learning_rate": 4.9937123924008845e-05, "loss": 1.1228, "step": 22678 }, { "epoch": 0.02, "learning_rate": 4.9937118356545274e-05, "loss": 1.129, "step": 22679 }, { "epoch": 0.02, "learning_rate": 4.9937112788835536e-05, "loss": 1.0454, "step": 22680 }, { "epoch": 0.02, "learning_rate": 4.9937107220879636e-05, "loss": 0.8853, "step": 22681 }, { "epoch": 0.02, "learning_rate": 4.993710165267757e-05, "loss": 1.0691, "step": 22682 }, { "epoch": 0.02, "learning_rate": 4.993709608422932e-05, "loss": 1.3264, "step": 22683 }, { "epoch": 0.02, "learning_rate": 4.9937090515534914e-05, "loss": 1.2246, "step": 22684 }, { "epoch": 0.02, "learning_rate": 4.9937084946594334e-05, "loss": 0.8565, "step": 22685 }, { "epoch": 0.02, "learning_rate": 4.993707937740758e-05, "loss": 1.1212, "step": 22686 }, { "epoch": 0.02, "learning_rate": 4.9937073807974665e-05, "loss": 0.9594, "step": 22687 }, { "epoch": 0.02, "learning_rate": 4.993706823829558e-05, "loss": 0.8823, "step": 22688 }, { "epoch": 0.02, "learning_rate": 4.993706266837033e-05, "loss": 1.4009, "step": 22689 }, { "epoch": 0.02, "learning_rate": 4.993705709819891e-05, "loss": 1.297, "step": 22690 }, { "epoch": 0.02, "learning_rate": 4.993705152778132e-05, "loss": 1.189, "step": 22691 }, { "epoch": 0.02, "learning_rate": 4.993704595711756e-05, "loss": 1.0223, "step": 22692 }, { "epoch": 0.02, "learning_rate": 4.9937040386207636e-05, "loss": 1.1296, "step": 22693 }, { "epoch": 0.02, "learning_rate": 4.9937034815051544e-05, "loss": 1.077, "step": 22694 }, { "epoch": 0.02, "learning_rate": 4.993702924364928e-05, "loss": 0.9934, "step": 22695 }, { "epoch": 0.02, "learning_rate": 4.9937023672000856e-05, "loss": 1.1961, "step": 22696 }, { "epoch": 0.02, "learning_rate": 4.993701810010626e-05, "loss": 0.8879, "step": 22697 }, { "epoch": 0.02, "learning_rate": 4.9937012527965495e-05, "loss": 1.2375, "step": 22698 }, { "epoch": 0.02, "learning_rate": 4.993700695557856e-05, "loss": 1.0417, "step": 22699 }, { "epoch": 0.02, "learning_rate": 4.993700138294546e-05, "loss": 1.1838, "step": 22700 }, { "epoch": 0.02, "learning_rate": 4.993699581006619e-05, "loss": 0.7993, "step": 22701 }, { "epoch": 0.02, "learning_rate": 4.993699023694076e-05, "loss": 0.6067, "step": 22702 }, { "epoch": 0.02, "learning_rate": 4.993698466356916e-05, "loss": 0.8747, "step": 22703 }, { "epoch": 0.02, "learning_rate": 4.9936979089951384e-05, "loss": 1.0813, "step": 22704 }, { "epoch": 0.02, "learning_rate": 4.993697351608746e-05, "loss": 1.3342, "step": 22705 }, { "epoch": 0.02, "learning_rate": 4.993696794197735e-05, "loss": 1.3443, "step": 22706 }, { "epoch": 0.02, "learning_rate": 4.993696236762108e-05, "loss": 1.2694, "step": 22707 }, { "epoch": 0.02, "learning_rate": 4.993695679301864e-05, "loss": 1.1238, "step": 22708 }, { "epoch": 0.02, "learning_rate": 4.993695121817004e-05, "loss": 0.9825, "step": 22709 }, { "epoch": 0.02, "learning_rate": 4.993694564307526e-05, "loss": 0.9064, "step": 22710 }, { "epoch": 0.02, "learning_rate": 4.993694006773432e-05, "loss": 1.0396, "step": 22711 }, { "epoch": 0.02, "learning_rate": 4.9936934492147216e-05, "loss": 1.0247, "step": 22712 }, { "epoch": 0.02, "learning_rate": 4.993692891631394e-05, "loss": 1.232, "step": 22713 }, { "epoch": 0.02, "learning_rate": 4.9936923340234496e-05, "loss": 1.4698, "step": 22714 }, { "epoch": 0.02, "learning_rate": 4.9936917763908894e-05, "loss": 1.0355, "step": 22715 }, { "epoch": 0.02, "learning_rate": 4.993691218733712e-05, "loss": 1.1536, "step": 22716 }, { "epoch": 0.02, "learning_rate": 4.993690661051917e-05, "loss": 1.0708, "step": 22717 }, { "epoch": 0.02, "learning_rate": 4.9936901033455066e-05, "loss": 1.2156, "step": 22718 }, { "epoch": 0.02, "learning_rate": 4.993689545614479e-05, "loss": 0.9829, "step": 22719 }, { "epoch": 0.02, "learning_rate": 4.9936889878588356e-05, "loss": 1.1797, "step": 22720 }, { "epoch": 0.02, "learning_rate": 4.9936884300785746e-05, "loss": 1.1594, "step": 22721 }, { "epoch": 0.02, "learning_rate": 4.993687872273698e-05, "loss": 1.2054, "step": 22722 }, { "epoch": 0.02, "learning_rate": 4.9936873144442034e-05, "loss": 1.093, "step": 22723 }, { "epoch": 0.02, "learning_rate": 4.993686756590093e-05, "loss": 1.1236, "step": 22724 }, { "epoch": 0.02, "learning_rate": 4.9936861987113656e-05, "loss": 1.7612, "step": 22725 }, { "epoch": 0.02, "learning_rate": 4.993685640808022e-05, "loss": 1.2637, "step": 22726 }, { "epoch": 0.02, "learning_rate": 4.993685082880061e-05, "loss": 1.1045, "step": 22727 }, { "epoch": 0.02, "learning_rate": 4.9936845249274846e-05, "loss": 1.166, "step": 22728 }, { "epoch": 0.02, "learning_rate": 4.993683966950291e-05, "loss": 1.0644, "step": 22729 }, { "epoch": 0.02, "learning_rate": 4.993683408948481e-05, "loss": 1.6695, "step": 22730 }, { "epoch": 0.02, "learning_rate": 4.993682850922054e-05, "loss": 1.2037, "step": 22731 }, { "epoch": 0.02, "learning_rate": 4.9936822928710104e-05, "loss": 0.8885, "step": 22732 }, { "epoch": 0.02, "learning_rate": 4.9936817347953504e-05, "loss": 0.8305, "step": 22733 }, { "epoch": 0.02, "learning_rate": 4.9936811766950735e-05, "loss": 0.71, "step": 22734 }, { "epoch": 0.02, "learning_rate": 4.99368061857018e-05, "loss": 1.0804, "step": 22735 }, { "epoch": 0.02, "learning_rate": 4.993680060420671e-05, "loss": 1.2445, "step": 22736 }, { "epoch": 0.02, "learning_rate": 4.993679502246544e-05, "loss": 1.0393, "step": 22737 }, { "epoch": 0.02, "learning_rate": 4.993678944047801e-05, "loss": 1.0069, "step": 22738 }, { "epoch": 0.02, "learning_rate": 4.993678385824442e-05, "loss": 1.1425, "step": 22739 }, { "epoch": 0.02, "learning_rate": 4.993677827576466e-05, "loss": 1.4789, "step": 22740 }, { "epoch": 0.02, "learning_rate": 4.9936772693038735e-05, "loss": 1.2097, "step": 22741 }, { "epoch": 0.02, "learning_rate": 4.993676711006665e-05, "loss": 1.133, "step": 22742 }, { "epoch": 0.02, "learning_rate": 4.9936761526848394e-05, "loss": 1.1156, "step": 22743 }, { "epoch": 0.02, "learning_rate": 4.993675594338397e-05, "loss": 1.1182, "step": 22744 }, { "epoch": 0.02, "learning_rate": 4.993675035967339e-05, "loss": 0.815, "step": 22745 }, { "epoch": 0.02, "learning_rate": 4.9936744775716635e-05, "loss": 2.0168, "step": 22746 }, { "epoch": 0.02, "learning_rate": 4.9936739191513715e-05, "loss": 1.0715, "step": 22747 }, { "epoch": 0.02, "learning_rate": 4.993673360706463e-05, "loss": 1.2664, "step": 22748 }, { "epoch": 0.02, "learning_rate": 4.99367280223694e-05, "loss": 0.9799, "step": 22749 }, { "epoch": 0.02, "learning_rate": 4.993672243742798e-05, "loss": 1.0327, "step": 22750 }, { "epoch": 0.02, "learning_rate": 4.993671685224041e-05, "loss": 1.2776, "step": 22751 }, { "epoch": 0.02, "learning_rate": 4.993671126680667e-05, "loss": 1.0275, "step": 22752 }, { "epoch": 0.02, "learning_rate": 4.993670568112676e-05, "loss": 1.2539, "step": 22753 }, { "epoch": 0.02, "learning_rate": 4.9936700095200695e-05, "loss": 1.359, "step": 22754 }, { "epoch": 0.02, "learning_rate": 4.993669450902846e-05, "loss": 0.7562, "step": 22755 }, { "epoch": 0.02, "learning_rate": 4.993668892261006e-05, "loss": 0.8791, "step": 22756 }, { "epoch": 0.02, "learning_rate": 4.99366833359455e-05, "loss": 1.0477, "step": 22757 }, { "epoch": 0.02, "learning_rate": 4.993667774903477e-05, "loss": 0.5487, "step": 22758 }, { "epoch": 0.02, "learning_rate": 4.993667216187787e-05, "loss": 0.7209, "step": 22759 }, { "epoch": 0.02, "learning_rate": 4.993666657447482e-05, "loss": 0.6032, "step": 22760 }, { "epoch": 0.02, "learning_rate": 4.99366609868256e-05, "loss": 0.773, "step": 22761 }, { "epoch": 0.02, "learning_rate": 4.9936655398930213e-05, "loss": 0.7667, "step": 22762 }, { "epoch": 0.02, "learning_rate": 4.9936649810788665e-05, "loss": 0.8161, "step": 22763 }, { "epoch": 0.02, "learning_rate": 4.9936644222400954e-05, "loss": 0.9563, "step": 22764 }, { "epoch": 0.02, "learning_rate": 4.9936638633767076e-05, "loss": 1.283, "step": 22765 }, { "epoch": 0.02, "learning_rate": 4.9936633044887036e-05, "loss": 0.9568, "step": 22766 }, { "epoch": 0.02, "learning_rate": 4.993662745576083e-05, "loss": 1.1037, "step": 22767 }, { "epoch": 0.02, "learning_rate": 4.993662186638846e-05, "loss": 0.899, "step": 22768 }, { "epoch": 0.02, "learning_rate": 4.993661627676993e-05, "loss": 1.0854, "step": 22769 }, { "epoch": 0.02, "learning_rate": 4.993661068690523e-05, "loss": 0.7627, "step": 22770 }, { "epoch": 0.02, "learning_rate": 4.993660509679437e-05, "loss": 1.1997, "step": 22771 }, { "epoch": 0.02, "learning_rate": 4.993659950643735e-05, "loss": 1.1211, "step": 22772 }, { "epoch": 0.02, "learning_rate": 4.993659391583416e-05, "loss": 1.008, "step": 22773 }, { "epoch": 0.02, "learning_rate": 4.993658832498481e-05, "loss": 0.8119, "step": 22774 }, { "epoch": 0.02, "learning_rate": 4.993658273388929e-05, "loss": 0.4886, "step": 22775 }, { "epoch": 0.02, "learning_rate": 4.993657714254761e-05, "loss": 0.5779, "step": 22776 }, { "epoch": 0.02, "learning_rate": 4.9936571550959777e-05, "loss": 0.5804, "step": 22777 }, { "epoch": 0.02, "learning_rate": 4.993656595912577e-05, "loss": 1.0595, "step": 22778 }, { "epoch": 0.02, "learning_rate": 4.99365603670456e-05, "loss": 1.3707, "step": 22779 }, { "epoch": 0.02, "learning_rate": 4.993655477471927e-05, "loss": 1.3533, "step": 22780 }, { "epoch": 0.02, "learning_rate": 4.993654918214678e-05, "loss": 1.0336, "step": 22781 }, { "epoch": 0.02, "learning_rate": 4.993654358932812e-05, "loss": 0.7583, "step": 22782 }, { "epoch": 0.02, "learning_rate": 4.99365379962633e-05, "loss": 0.9931, "step": 22783 }, { "epoch": 0.02, "learning_rate": 4.9936532402952316e-05, "loss": 1.2337, "step": 22784 }, { "epoch": 0.02, "learning_rate": 4.993652680939517e-05, "loss": 0.7133, "step": 22785 }, { "epoch": 0.02, "learning_rate": 4.9936521215591857e-05, "loss": 0.992, "step": 22786 }, { "epoch": 0.02, "learning_rate": 4.993651562154239e-05, "loss": 1.0192, "step": 22787 }, { "epoch": 0.02, "learning_rate": 4.993651002724675e-05, "loss": 0.9643, "step": 22788 }, { "epoch": 0.02, "learning_rate": 4.9936504432704955e-05, "loss": 1.1712, "step": 22789 }, { "epoch": 0.02, "learning_rate": 4.9936498837917e-05, "loss": 1.2167, "step": 22790 }, { "epoch": 0.02, "learning_rate": 4.9936493242882875e-05, "loss": 1.1559, "step": 22791 }, { "epoch": 0.02, "learning_rate": 4.9936487647602594e-05, "loss": 1.1, "step": 22792 }, { "epoch": 0.02, "learning_rate": 4.993648205207614e-05, "loss": 1.1241, "step": 22793 }, { "epoch": 0.02, "learning_rate": 4.9936476456303526e-05, "loss": 1.1747, "step": 22794 }, { "epoch": 0.02, "learning_rate": 4.9936470860284754e-05, "loss": 1.0487, "step": 22795 }, { "epoch": 0.02, "learning_rate": 4.993646526401982e-05, "loss": 1.1629, "step": 22796 }, { "epoch": 0.02, "learning_rate": 4.9936459667508725e-05, "loss": 1.2597, "step": 22797 }, { "epoch": 0.02, "learning_rate": 4.993645407075146e-05, "loss": 1.2533, "step": 22798 }, { "epoch": 0.02, "learning_rate": 4.9936448473748045e-05, "loss": 1.0461, "step": 22799 }, { "epoch": 0.02, "learning_rate": 4.993644287649846e-05, "loss": 1.277, "step": 22800 }, { "epoch": 0.02, "learning_rate": 4.993643727900271e-05, "loss": 1.1808, "step": 22801 }, { "epoch": 0.02, "learning_rate": 4.993643168126081e-05, "loss": 1.0565, "step": 22802 }, { "epoch": 0.02, "learning_rate": 4.9936426083272734e-05, "loss": 1.0575, "step": 22803 }, { "epoch": 0.02, "learning_rate": 4.99364204850385e-05, "loss": 1.0956, "step": 22804 }, { "epoch": 0.02, "learning_rate": 4.993641488655812e-05, "loss": 1.1178, "step": 22805 }, { "epoch": 0.02, "learning_rate": 4.9936409287831556e-05, "loss": 1.4916, "step": 22806 }, { "epoch": 0.02, "learning_rate": 4.993640368885884e-05, "loss": 1.0628, "step": 22807 }, { "epoch": 0.02, "learning_rate": 4.993639808963996e-05, "loss": 0.8721, "step": 22808 }, { "epoch": 0.02, "learning_rate": 4.993639249017492e-05, "loss": 1.106, "step": 22809 }, { "epoch": 0.02, "learning_rate": 4.9936386890463715e-05, "loss": 1.1806, "step": 22810 }, { "epoch": 0.02, "learning_rate": 4.9936381290506354e-05, "loss": 1.1241, "step": 22811 }, { "epoch": 0.02, "learning_rate": 4.9936375690302826e-05, "loss": 1.2374, "step": 22812 }, { "epoch": 0.02, "learning_rate": 4.993637008985314e-05, "loss": 1.1573, "step": 22813 }, { "epoch": 0.02, "learning_rate": 4.993636448915729e-05, "loss": 1.2514, "step": 22814 }, { "epoch": 0.02, "learning_rate": 4.9936358888215286e-05, "loss": 1.3566, "step": 22815 }, { "epoch": 0.02, "learning_rate": 4.993635328702711e-05, "loss": 1.031, "step": 22816 }, { "epoch": 0.02, "learning_rate": 4.993634768559279e-05, "loss": 0.8074, "step": 22817 }, { "epoch": 0.02, "learning_rate": 4.993634208391229e-05, "loss": 1.1266, "step": 22818 }, { "epoch": 0.02, "learning_rate": 4.9936336481985644e-05, "loss": 1.0654, "step": 22819 }, { "epoch": 0.02, "learning_rate": 4.9936330879812825e-05, "loss": 1.3112, "step": 22820 }, { "epoch": 0.02, "learning_rate": 4.9936325277393844e-05, "loss": 0.8976, "step": 22821 }, { "epoch": 0.02, "learning_rate": 4.9936319674728716e-05, "loss": 1.0811, "step": 22822 }, { "epoch": 0.02, "learning_rate": 4.993631407181741e-05, "loss": 1.2365, "step": 22823 }, { "epoch": 0.02, "learning_rate": 4.993630846865996e-05, "loss": 1.4577, "step": 22824 }, { "epoch": 0.02, "learning_rate": 4.993630286525634e-05, "loss": 1.161, "step": 22825 }, { "epoch": 0.02, "learning_rate": 4.993629726160656e-05, "loss": 1.044, "step": 22826 }, { "epoch": 0.02, "learning_rate": 4.993629165771062e-05, "loss": 1.4188, "step": 22827 }, { "epoch": 0.02, "learning_rate": 4.993628605356852e-05, "loss": 1.2865, "step": 22828 }, { "epoch": 0.02, "learning_rate": 4.993628044918026e-05, "loss": 0.6906, "step": 22829 }, { "epoch": 0.02, "learning_rate": 4.9936274844545835e-05, "loss": 1.1752, "step": 22830 }, { "epoch": 0.02, "learning_rate": 4.9936269239665255e-05, "loss": 0.6921, "step": 22831 }, { "epoch": 0.02, "learning_rate": 4.9936263634538514e-05, "loss": 0.9259, "step": 22832 }, { "epoch": 0.02, "learning_rate": 4.993625802916561e-05, "loss": 1.0902, "step": 22833 }, { "epoch": 0.02, "learning_rate": 4.993625242354655e-05, "loss": 0.8924, "step": 22834 }, { "epoch": 0.02, "learning_rate": 4.993624681768133e-05, "loss": 1.0054, "step": 22835 }, { "epoch": 0.02, "learning_rate": 4.993624121156995e-05, "loss": 1.1921, "step": 22836 }, { "epoch": 0.02, "learning_rate": 4.99362356052124e-05, "loss": 1.0673, "step": 22837 }, { "epoch": 0.02, "learning_rate": 4.99362299986087e-05, "loss": 1.1527, "step": 22838 }, { "epoch": 0.02, "learning_rate": 4.9936224391758834e-05, "loss": 0.9995, "step": 22839 }, { "epoch": 0.02, "learning_rate": 4.9936218784662815e-05, "loss": 1.0818, "step": 22840 }, { "epoch": 0.02, "learning_rate": 4.993621317732063e-05, "loss": 1.1909, "step": 22841 }, { "epoch": 0.02, "learning_rate": 4.993620756973229e-05, "loss": 0.9967, "step": 22842 }, { "epoch": 0.02, "learning_rate": 4.9936201961897785e-05, "loss": 0.9477, "step": 22843 }, { "epoch": 0.02, "learning_rate": 4.993619635381712e-05, "loss": 1.1626, "step": 22844 }, { "epoch": 0.02, "learning_rate": 4.9936190745490296e-05, "loss": 1.0329, "step": 22845 }, { "epoch": 0.02, "learning_rate": 4.9936185136917316e-05, "loss": 1.2588, "step": 22846 }, { "epoch": 0.02, "learning_rate": 4.993617952809818e-05, "loss": 1.1594, "step": 22847 }, { "epoch": 0.02, "learning_rate": 4.993617391903288e-05, "loss": 0.7747, "step": 22848 }, { "epoch": 0.02, "learning_rate": 4.993616830972142e-05, "loss": 1.2559, "step": 22849 }, { "epoch": 0.02, "learning_rate": 4.9936162700163805e-05, "loss": 0.9532, "step": 22850 }, { "epoch": 0.02, "learning_rate": 4.9936157090360025e-05, "loss": 1.1414, "step": 22851 }, { "epoch": 0.02, "learning_rate": 4.9936151480310084e-05, "loss": 1.1945, "step": 22852 }, { "epoch": 0.02, "learning_rate": 4.993614587001398e-05, "loss": 1.1865, "step": 22853 }, { "epoch": 0.02, "learning_rate": 4.993614025947173e-05, "loss": 1.5611, "step": 22854 }, { "epoch": 0.02, "learning_rate": 4.993613464868332e-05, "loss": 1.253, "step": 22855 }, { "epoch": 0.02, "learning_rate": 4.993612903764874e-05, "loss": 1.3309, "step": 22856 }, { "epoch": 0.02, "learning_rate": 4.993612342636801e-05, "loss": 1.0304, "step": 22857 }, { "epoch": 0.02, "learning_rate": 4.993611781484112e-05, "loss": 0.7214, "step": 22858 }, { "epoch": 0.02, "learning_rate": 4.993611220306807e-05, "loss": 0.9967, "step": 22859 }, { "epoch": 0.02, "learning_rate": 4.993610659104886e-05, "loss": 1.2314, "step": 22860 }, { "epoch": 0.02, "learning_rate": 4.993610097878349e-05, "loss": 1.1125, "step": 22861 }, { "epoch": 0.02, "learning_rate": 4.9936095366271965e-05, "loss": 1.1133, "step": 22862 }, { "epoch": 0.02, "learning_rate": 4.9936089753514284e-05, "loss": 1.0877, "step": 22863 }, { "epoch": 0.02, "learning_rate": 4.9936084140510434e-05, "loss": 1.1586, "step": 22864 }, { "epoch": 0.02, "learning_rate": 4.993607852726043e-05, "loss": 0.9032, "step": 22865 }, { "epoch": 0.02, "learning_rate": 4.993607291376428e-05, "loss": 1.2117, "step": 22866 }, { "epoch": 0.02, "learning_rate": 4.993606730002195e-05, "loss": 1.1376, "step": 22867 }, { "epoch": 0.02, "learning_rate": 4.9936061686033476e-05, "loss": 1.1882, "step": 22868 }, { "epoch": 0.02, "learning_rate": 4.993605607179884e-05, "loss": 1.163, "step": 22869 }, { "epoch": 0.02, "learning_rate": 4.993605045731804e-05, "loss": 1.1174, "step": 22870 }, { "epoch": 0.02, "learning_rate": 4.993604484259109e-05, "loss": 0.9193, "step": 22871 }, { "epoch": 0.02, "learning_rate": 4.9936039227617984e-05, "loss": 1.3452, "step": 22872 }, { "epoch": 0.02, "learning_rate": 4.993603361239872e-05, "loss": 1.2845, "step": 22873 }, { "epoch": 0.02, "learning_rate": 4.993602799693329e-05, "loss": 1.1429, "step": 22874 }, { "epoch": 0.02, "learning_rate": 4.9936022381221704e-05, "loss": 0.2373, "step": 22875 }, { "epoch": 0.02, "learning_rate": 4.993601676526396e-05, "loss": 0.1727, "step": 22876 }, { "epoch": 0.02, "learning_rate": 4.993601114906006e-05, "loss": 0.146, "step": 22877 }, { "epoch": 0.02, "learning_rate": 4.9936005532610005e-05, "loss": 0.1212, "step": 22878 }, { "epoch": 0.02, "learning_rate": 4.993599991591379e-05, "loss": 0.137, "step": 22879 }, { "epoch": 0.02, "learning_rate": 4.993599429897141e-05, "loss": 0.0669, "step": 22880 }, { "epoch": 0.02, "learning_rate": 4.993598868178289e-05, "loss": 0.1667, "step": 22881 }, { "epoch": 0.02, "learning_rate": 4.993598306434819e-05, "loss": 0.9938, "step": 22882 }, { "epoch": 0.02, "learning_rate": 4.993597744666735e-05, "loss": 0.9944, "step": 22883 }, { "epoch": 0.02, "learning_rate": 4.993597182874035e-05, "loss": 1.2144, "step": 22884 }, { "epoch": 0.02, "learning_rate": 4.9935966210567184e-05, "loss": 1.0508, "step": 22885 }, { "epoch": 0.02, "learning_rate": 4.993596059214786e-05, "loss": 0.8657, "step": 22886 }, { "epoch": 0.02, "learning_rate": 4.993595497348239e-05, "loss": 0.9969, "step": 22887 }, { "epoch": 0.02, "learning_rate": 4.9935949354570755e-05, "loss": 1.3994, "step": 22888 }, { "epoch": 0.02, "learning_rate": 4.9935943735412967e-05, "loss": 1.4817, "step": 22889 }, { "epoch": 0.02, "learning_rate": 4.993593811600902e-05, "loss": 1.0266, "step": 22890 }, { "epoch": 0.02, "learning_rate": 4.993593249635892e-05, "loss": 1.5283, "step": 22891 }, { "epoch": 0.02, "learning_rate": 4.9935926876462655e-05, "loss": 0.8331, "step": 22892 }, { "epoch": 0.02, "learning_rate": 4.9935921256320235e-05, "loss": 1.2384, "step": 22893 }, { "epoch": 0.02, "learning_rate": 4.9935915635931654e-05, "loss": 1.339, "step": 22894 }, { "epoch": 0.02, "learning_rate": 4.9935910015296925e-05, "loss": 0.8456, "step": 22895 }, { "epoch": 0.02, "learning_rate": 4.993590439441604e-05, "loss": 1.1507, "step": 22896 }, { "epoch": 0.02, "learning_rate": 4.993589877328899e-05, "loss": 1.1215, "step": 22897 }, { "epoch": 0.02, "learning_rate": 4.993589315191579e-05, "loss": 1.2436, "step": 22898 }, { "epoch": 0.02, "learning_rate": 4.993588753029643e-05, "loss": 1.1305, "step": 22899 }, { "epoch": 0.02, "learning_rate": 4.9935881908430915e-05, "loss": 1.0648, "step": 22900 }, { "epoch": 0.02, "learning_rate": 4.9935876286319246e-05, "loss": 1.356, "step": 22901 }, { "epoch": 0.02, "learning_rate": 4.9935870663961414e-05, "loss": 1.2557, "step": 22902 }, { "epoch": 0.02, "learning_rate": 4.9935865041357436e-05, "loss": 0.9333, "step": 22903 }, { "epoch": 0.02, "learning_rate": 4.993585941850729e-05, "loss": 1.1992, "step": 22904 }, { "epoch": 0.02, "learning_rate": 4.9935853795410994e-05, "loss": 0.7205, "step": 22905 }, { "epoch": 0.02, "learning_rate": 4.993584817206853e-05, "loss": 0.6744, "step": 22906 }, { "epoch": 0.02, "learning_rate": 4.993584254847993e-05, "loss": 0.9376, "step": 22907 }, { "epoch": 0.02, "learning_rate": 4.993583692464516e-05, "loss": 1.2699, "step": 22908 }, { "epoch": 0.02, "learning_rate": 4.993583130056424e-05, "loss": 1.6732, "step": 22909 }, { "epoch": 0.02, "learning_rate": 4.9935825676237155e-05, "loss": 1.2692, "step": 22910 }, { "epoch": 0.02, "learning_rate": 4.9935820051663926e-05, "loss": 1.1482, "step": 22911 }, { "epoch": 0.02, "learning_rate": 4.9935814426844536e-05, "loss": 1.1489, "step": 22912 }, { "epoch": 0.02, "learning_rate": 4.993580880177899e-05, "loss": 0.9965, "step": 22913 }, { "epoch": 0.02, "learning_rate": 4.9935803176467285e-05, "loss": 1.0552, "step": 22914 }, { "epoch": 0.02, "learning_rate": 4.993579755090943e-05, "loss": 1.7842, "step": 22915 }, { "epoch": 0.02, "learning_rate": 4.9935791925105417e-05, "loss": 1.0552, "step": 22916 }, { "epoch": 0.02, "learning_rate": 4.993578629905525e-05, "loss": 0.9522, "step": 22917 }, { "epoch": 0.02, "learning_rate": 4.993578067275892e-05, "loss": 1.0008, "step": 22918 }, { "epoch": 0.02, "learning_rate": 4.993577504621645e-05, "loss": 1.16, "step": 22919 }, { "epoch": 0.02, "learning_rate": 4.993576941942781e-05, "loss": 1.0426, "step": 22920 }, { "epoch": 0.02, "learning_rate": 4.993576379239302e-05, "loss": 0.9916, "step": 22921 }, { "epoch": 0.02, "learning_rate": 4.9935758165112074e-05, "loss": 0.9738, "step": 22922 }, { "epoch": 0.02, "learning_rate": 4.993575253758497e-05, "loss": 0.9987, "step": 22923 }, { "epoch": 0.02, "learning_rate": 4.993574690981171e-05, "loss": 1.1087, "step": 22924 }, { "epoch": 0.02, "learning_rate": 4.9935741281792306e-05, "loss": 1.1899, "step": 22925 }, { "epoch": 0.02, "learning_rate": 4.993573565352674e-05, "loss": 0.8281, "step": 22926 }, { "epoch": 0.02, "learning_rate": 4.9935730025015016e-05, "loss": 1.1294, "step": 22927 }, { "epoch": 0.02, "learning_rate": 4.993572439625713e-05, "loss": 1.0034, "step": 22928 }, { "epoch": 0.02, "learning_rate": 4.99357187672531e-05, "loss": 0.8759, "step": 22929 }, { "epoch": 0.02, "learning_rate": 4.993571313800292e-05, "loss": 1.1925, "step": 22930 }, { "epoch": 0.02, "learning_rate": 4.9935707508506577e-05, "loss": 1.0938, "step": 22931 }, { "epoch": 0.02, "learning_rate": 4.993570187876408e-05, "loss": 1.0943, "step": 22932 }, { "epoch": 0.02, "learning_rate": 4.993569624877543e-05, "loss": 1.1507, "step": 22933 }, { "epoch": 0.02, "learning_rate": 4.993569061854062e-05, "loss": 1.1888, "step": 22934 }, { "epoch": 0.02, "learning_rate": 4.9935684988059664e-05, "loss": 1.1569, "step": 22935 }, { "epoch": 0.02, "learning_rate": 4.9935679357332544e-05, "loss": 1.0678, "step": 22936 }, { "epoch": 0.02, "learning_rate": 4.993567372635928e-05, "loss": 1.1952, "step": 22937 }, { "epoch": 0.02, "learning_rate": 4.9935668095139855e-05, "loss": 0.9247, "step": 22938 }, { "epoch": 0.02, "learning_rate": 4.993566246367427e-05, "loss": 0.9171, "step": 22939 }, { "epoch": 0.02, "learning_rate": 4.993565683196254e-05, "loss": 1.0093, "step": 22940 }, { "epoch": 0.02, "learning_rate": 4.9935651200004655e-05, "loss": 1.2875, "step": 22941 }, { "epoch": 0.02, "learning_rate": 4.9935645567800615e-05, "loss": 0.9603, "step": 22942 }, { "epoch": 0.02, "learning_rate": 4.993563993535042e-05, "loss": 0.858, "step": 22943 }, { "epoch": 0.02, "learning_rate": 4.9935634302654064e-05, "loss": 0.7886, "step": 22944 }, { "epoch": 0.02, "learning_rate": 4.993562866971157e-05, "loss": 0.9116, "step": 22945 }, { "epoch": 0.02, "learning_rate": 4.99356230365229e-05, "loss": 0.8696, "step": 22946 }, { "epoch": 0.02, "learning_rate": 4.993561740308809e-05, "loss": 1.2192, "step": 22947 }, { "epoch": 0.02, "learning_rate": 4.993561176940713e-05, "loss": 1.1909, "step": 22948 }, { "epoch": 0.02, "learning_rate": 4.9935606135480006e-05, "loss": 1.1797, "step": 22949 }, { "epoch": 0.02, "learning_rate": 4.9935600501306736e-05, "loss": 1.1113, "step": 22950 }, { "epoch": 0.02, "learning_rate": 4.993559486688731e-05, "loss": 1.0455, "step": 22951 }, { "epoch": 0.02, "learning_rate": 4.9935589232221726e-05, "loss": 0.9476, "step": 22952 }, { "epoch": 0.02, "learning_rate": 4.993558359730999e-05, "loss": 1.0316, "step": 22953 }, { "epoch": 0.02, "learning_rate": 4.9935577962152105e-05, "loss": 0.9869, "step": 22954 }, { "epoch": 0.02, "learning_rate": 4.993557232674806e-05, "loss": 1.4799, "step": 22955 }, { "epoch": 0.02, "learning_rate": 4.9935566691097865e-05, "loss": 0.3023, "step": 22956 }, { "epoch": 0.02, "learning_rate": 4.993556105520152e-05, "loss": 0.2833, "step": 22957 }, { "epoch": 0.02, "learning_rate": 4.9935555419059014e-05, "loss": 0.8165, "step": 22958 }, { "epoch": 0.02, "learning_rate": 4.993554978267037e-05, "loss": 1.3387, "step": 22959 }, { "epoch": 0.02, "learning_rate": 4.993554414603555e-05, "loss": 0.8779, "step": 22960 }, { "epoch": 0.02, "learning_rate": 4.993553850915459e-05, "loss": 1.1595, "step": 22961 }, { "epoch": 0.02, "learning_rate": 4.993553287202748e-05, "loss": 1.0014, "step": 22962 }, { "epoch": 0.02, "learning_rate": 4.993552723465421e-05, "loss": 1.1151, "step": 22963 }, { "epoch": 0.02, "learning_rate": 4.993552159703479e-05, "loss": 1.195, "step": 22964 }, { "epoch": 0.02, "learning_rate": 4.993551595916922e-05, "loss": 1.0252, "step": 22965 }, { "epoch": 0.02, "learning_rate": 4.993551032105749e-05, "loss": 0.8012, "step": 22966 }, { "epoch": 0.02, "learning_rate": 4.9935504682699605e-05, "loss": 1.1935, "step": 22967 }, { "epoch": 0.02, "learning_rate": 4.993549904409558e-05, "loss": 1.0364, "step": 22968 }, { "epoch": 0.02, "learning_rate": 4.9935493405245394e-05, "loss": 1.1491, "step": 22969 }, { "epoch": 0.02, "learning_rate": 4.993548776614906e-05, "loss": 0.9459, "step": 22970 }, { "epoch": 0.02, "learning_rate": 4.9935482126806566e-05, "loss": 1.0603, "step": 22971 }, { "epoch": 0.02, "learning_rate": 4.993547648721792e-05, "loss": 0.987, "step": 22972 }, { "epoch": 0.02, "learning_rate": 4.9935470847383125e-05, "loss": 0.8059, "step": 22973 }, { "epoch": 0.02, "learning_rate": 4.993546520730218e-05, "loss": 1.0326, "step": 22974 }, { "epoch": 0.02, "learning_rate": 4.993545956697508e-05, "loss": 0.9547, "step": 22975 }, { "epoch": 0.02, "learning_rate": 4.9935453926401826e-05, "loss": 0.4794, "step": 22976 }, { "epoch": 0.02, "learning_rate": 4.993544828558242e-05, "loss": 0.5014, "step": 22977 }, { "epoch": 0.02, "learning_rate": 4.993544264451686e-05, "loss": 0.3616, "step": 22978 }, { "epoch": 0.02, "learning_rate": 4.993543700320516e-05, "loss": 0.4725, "step": 22979 }, { "epoch": 0.02, "learning_rate": 4.993543136164729e-05, "loss": 1.3509, "step": 22980 }, { "epoch": 0.02, "learning_rate": 4.993542571984328e-05, "loss": 1.0847, "step": 22981 }, { "epoch": 0.02, "learning_rate": 4.9935420077793116e-05, "loss": 1.2071, "step": 22982 }, { "epoch": 0.02, "learning_rate": 4.9935414435496794e-05, "loss": 0.9628, "step": 22983 }, { "epoch": 0.02, "learning_rate": 4.993540879295433e-05, "loss": 0.97, "step": 22984 }, { "epoch": 0.02, "learning_rate": 4.9935403150165705e-05, "loss": 1.2847, "step": 22985 }, { "epoch": 0.02, "learning_rate": 4.993539750713093e-05, "loss": 0.9338, "step": 22986 }, { "epoch": 0.02, "learning_rate": 4.993539186385001e-05, "loss": 1.0953, "step": 22987 }, { "epoch": 0.02, "learning_rate": 4.993538622032293e-05, "loss": 0.8373, "step": 22988 }, { "epoch": 0.02, "learning_rate": 4.993538057654971e-05, "loss": 1.5436, "step": 22989 }, { "epoch": 0.02, "learning_rate": 4.9935374932530324e-05, "loss": 0.91, "step": 22990 }, { "epoch": 0.02, "learning_rate": 4.9935369288264785e-05, "loss": 0.6649, "step": 22991 }, { "epoch": 0.02, "learning_rate": 4.9935363643753105e-05, "loss": 1.2547, "step": 22992 }, { "epoch": 0.02, "learning_rate": 4.993535799899527e-05, "loss": 1.1787, "step": 22993 }, { "epoch": 0.02, "learning_rate": 4.9935352353991296e-05, "loss": 1.0793, "step": 22994 }, { "epoch": 0.02, "learning_rate": 4.993534670874115e-05, "loss": 0.9317, "step": 22995 }, { "epoch": 0.02, "learning_rate": 4.993534106324487e-05, "loss": 1.3777, "step": 22996 }, { "epoch": 0.02, "learning_rate": 4.993533541750243e-05, "loss": 1.0168, "step": 22997 }, { "epoch": 0.02, "learning_rate": 4.9935329771513836e-05, "loss": 1.1855, "step": 22998 }, { "epoch": 0.02, "learning_rate": 4.9935324125279095e-05, "loss": 1.1221, "step": 22999 }, { "epoch": 0.02, "learning_rate": 4.9935318478798206e-05, "loss": 1.1693, "step": 23000 }, { "epoch": 0.02, "eval_loss": 1.0269695520401, "eval_runtime": 85.8819, "eval_samples_per_second": 16.127, "eval_steps_per_second": 4.04, "step": 23000 }, { "epoch": 0.02, "learning_rate": 4.9935312832071156e-05, "loss": 1.469, "step": 23001 }, { "epoch": 0.02, "learning_rate": 4.9935307185097965e-05, "loss": 1.1519, "step": 23002 }, { "epoch": 0.02, "learning_rate": 4.993530153787862e-05, "loss": 1.0409, "step": 23003 }, { "epoch": 0.02, "learning_rate": 4.9935295890413125e-05, "loss": 1.6231, "step": 23004 }, { "epoch": 0.02, "learning_rate": 4.9935290242701484e-05, "loss": 1.2493, "step": 23005 }, { "epoch": 0.02, "learning_rate": 4.993528459474368e-05, "loss": 0.8139, "step": 23006 }, { "epoch": 0.02, "learning_rate": 4.993527894653973e-05, "loss": 0.7774, "step": 23007 }, { "epoch": 0.02, "learning_rate": 4.993527329808963e-05, "loss": 0.6831, "step": 23008 }, { "epoch": 0.02, "learning_rate": 4.993526764939339e-05, "loss": 0.682, "step": 23009 }, { "epoch": 0.02, "learning_rate": 4.9935262000450986e-05, "loss": 0.581, "step": 23010 }, { "epoch": 0.02, "learning_rate": 4.993525635126244e-05, "loss": 0.9875, "step": 23011 }, { "epoch": 0.02, "learning_rate": 4.993525070182774e-05, "loss": 1.0447, "step": 23012 }, { "epoch": 0.02, "learning_rate": 4.9935245052146884e-05, "loss": 1.0847, "step": 23013 }, { "epoch": 0.02, "learning_rate": 4.9935239402219886e-05, "loss": 0.8432, "step": 23014 }, { "epoch": 0.02, "learning_rate": 4.993523375204674e-05, "loss": 0.9486, "step": 23015 }, { "epoch": 0.02, "learning_rate": 4.993522810162743e-05, "loss": 0.9912, "step": 23016 }, { "epoch": 0.02, "learning_rate": 4.9935222450961984e-05, "loss": 0.823, "step": 23017 }, { "epoch": 0.02, "learning_rate": 4.993521680005038e-05, "loss": 1.1972, "step": 23018 }, { "epoch": 0.02, "learning_rate": 4.993521114889264e-05, "loss": 1.4042, "step": 23019 }, { "epoch": 0.02, "learning_rate": 4.993520549748874e-05, "loss": 0.5818, "step": 23020 }, { "epoch": 0.02, "learning_rate": 4.9935199845838685e-05, "loss": 0.9654, "step": 23021 }, { "epoch": 0.02, "learning_rate": 4.9935194193942484e-05, "loss": 0.8589, "step": 23022 }, { "epoch": 0.02, "learning_rate": 4.993518854180013e-05, "loss": 1.1889, "step": 23023 }, { "epoch": 0.02, "learning_rate": 4.993518288941163e-05, "loss": 0.835, "step": 23024 }, { "epoch": 0.02, "learning_rate": 4.993517723677698e-05, "loss": 1.0985, "step": 23025 }, { "epoch": 0.02, "learning_rate": 4.993517158389618e-05, "loss": 1.3575, "step": 23026 }, { "epoch": 0.02, "learning_rate": 4.9935165930769235e-05, "loss": 2.0939, "step": 23027 }, { "epoch": 0.02, "learning_rate": 4.9935160277396134e-05, "loss": 1.0884, "step": 23028 }, { "epoch": 0.02, "learning_rate": 4.9935154623776885e-05, "loss": 1.0751, "step": 23029 }, { "epoch": 0.02, "learning_rate": 4.9935148969911495e-05, "loss": 1.0979, "step": 23030 }, { "epoch": 0.02, "learning_rate": 4.993514331579995e-05, "loss": 0.9313, "step": 23031 }, { "epoch": 0.02, "learning_rate": 4.993513766144225e-05, "loss": 1.0359, "step": 23032 }, { "epoch": 0.02, "learning_rate": 4.9935132006838405e-05, "loss": 1.0045, "step": 23033 }, { "epoch": 0.02, "learning_rate": 4.993512635198842e-05, "loss": 1.1111, "step": 23034 }, { "epoch": 0.02, "learning_rate": 4.993512069689227e-05, "loss": 0.901, "step": 23035 }, { "epoch": 0.02, "learning_rate": 4.993511504154997e-05, "loss": 1.2885, "step": 23036 }, { "epoch": 0.02, "learning_rate": 4.993510938596154e-05, "loss": 1.2137, "step": 23037 }, { "epoch": 0.02, "learning_rate": 4.993510373012695e-05, "loss": 1.1166, "step": 23038 }, { "epoch": 0.02, "learning_rate": 4.993509807404621e-05, "loss": 1.1621, "step": 23039 }, { "epoch": 0.02, "learning_rate": 4.993509241771932e-05, "loss": 0.9863, "step": 23040 }, { "epoch": 0.02, "learning_rate": 4.993508676114629e-05, "loss": 1.0108, "step": 23041 }, { "epoch": 0.02, "learning_rate": 4.99350811043271e-05, "loss": 0.8774, "step": 23042 }, { "epoch": 0.02, "learning_rate": 4.993507544726177e-05, "loss": 1.2331, "step": 23043 }, { "epoch": 0.02, "learning_rate": 4.993506978995028e-05, "loss": 0.9145, "step": 23044 }, { "epoch": 0.02, "learning_rate": 4.993506413239266e-05, "loss": 1.0575, "step": 23045 }, { "epoch": 0.02, "learning_rate": 4.993505847458888e-05, "loss": 1.2772, "step": 23046 }, { "epoch": 0.02, "learning_rate": 4.993505281653895e-05, "loss": 0.9245, "step": 23047 }, { "epoch": 0.02, "learning_rate": 4.9935047158242875e-05, "loss": 1.065, "step": 23048 }, { "epoch": 0.02, "learning_rate": 4.993504149970065e-05, "loss": 0.9264, "step": 23049 }, { "epoch": 0.02, "learning_rate": 4.993503584091228e-05, "loss": 0.9454, "step": 23050 }, { "epoch": 0.02, "learning_rate": 4.9935030181877765e-05, "loss": 0.9317, "step": 23051 }, { "epoch": 0.02, "learning_rate": 4.99350245225971e-05, "loss": 0.9128, "step": 23052 }, { "epoch": 0.02, "learning_rate": 4.9935018863070274e-05, "loss": 0.928, "step": 23053 }, { "epoch": 0.02, "learning_rate": 4.993501320329731e-05, "loss": 0.8554, "step": 23054 }, { "epoch": 0.02, "learning_rate": 4.99350075432782e-05, "loss": 0.8706, "step": 23055 }, { "epoch": 0.02, "learning_rate": 4.993500188301295e-05, "loss": 0.9101, "step": 23056 }, { "epoch": 0.02, "learning_rate": 4.993499622250153e-05, "loss": 0.9994, "step": 23057 }, { "epoch": 0.02, "learning_rate": 4.993499056174398e-05, "loss": 1.0442, "step": 23058 }, { "epoch": 0.02, "learning_rate": 4.9934984900740276e-05, "loss": 1.1151, "step": 23059 }, { "epoch": 0.02, "learning_rate": 4.993497923949042e-05, "loss": 1.0228, "step": 23060 }, { "epoch": 0.02, "learning_rate": 4.993497357799443e-05, "loss": 0.9086, "step": 23061 }, { "epoch": 0.02, "learning_rate": 4.993496791625228e-05, "loss": 1.0301, "step": 23062 }, { "epoch": 0.02, "learning_rate": 4.993496225426398e-05, "loss": 1.2314, "step": 23063 }, { "epoch": 0.02, "learning_rate": 4.993495659202955e-05, "loss": 1.4393, "step": 23064 }, { "epoch": 0.02, "learning_rate": 4.993495092954896e-05, "loss": 1.0543, "step": 23065 }, { "epoch": 0.02, "learning_rate": 4.993494526682222e-05, "loss": 0.4764, "step": 23066 }, { "epoch": 0.02, "learning_rate": 4.993493960384934e-05, "loss": 0.3206, "step": 23067 }, { "epoch": 0.02, "learning_rate": 4.993493394063031e-05, "loss": 0.2417, "step": 23068 }, { "epoch": 0.02, "learning_rate": 4.993492827716513e-05, "loss": 0.2595, "step": 23069 }, { "epoch": 0.02, "learning_rate": 4.99349226134538e-05, "loss": 0.114, "step": 23070 }, { "epoch": 0.02, "learning_rate": 4.993491694949633e-05, "loss": 0.0979, "step": 23071 }, { "epoch": 0.02, "learning_rate": 4.9934911285292716e-05, "loss": 0.1303, "step": 23072 }, { "epoch": 0.02, "learning_rate": 4.993490562084295e-05, "loss": 0.2053, "step": 23073 }, { "epoch": 0.02, "learning_rate": 4.993489995614704e-05, "loss": 0.4402, "step": 23074 }, { "epoch": 0.02, "learning_rate": 4.993489429120498e-05, "loss": 0.9164, "step": 23075 }, { "epoch": 0.02, "learning_rate": 4.993488862601677e-05, "loss": 1.292, "step": 23076 }, { "epoch": 0.02, "learning_rate": 4.9934882960582416e-05, "loss": 1.2109, "step": 23077 }, { "epoch": 0.02, "learning_rate": 4.993487729490192e-05, "loss": 1.0574, "step": 23078 }, { "epoch": 0.02, "learning_rate": 4.9934871628975275e-05, "loss": 1.1563, "step": 23079 }, { "epoch": 0.02, "learning_rate": 4.993486596280248e-05, "loss": 0.9788, "step": 23080 }, { "epoch": 0.02, "learning_rate": 4.993486029638354e-05, "loss": 1.0671, "step": 23081 }, { "epoch": 0.02, "learning_rate": 4.9934854629718455e-05, "loss": 1.2214, "step": 23082 }, { "epoch": 0.02, "learning_rate": 4.9934848962807226e-05, "loss": 1.1954, "step": 23083 }, { "epoch": 0.02, "learning_rate": 4.993484329564985e-05, "loss": 1.3354, "step": 23084 }, { "epoch": 0.02, "learning_rate": 4.9934837628246325e-05, "loss": 1.1561, "step": 23085 }, { "epoch": 0.02, "learning_rate": 4.993483196059665e-05, "loss": 1.3062, "step": 23086 }, { "epoch": 0.02, "learning_rate": 4.993482629270084e-05, "loss": 1.6879, "step": 23087 }, { "epoch": 0.02, "learning_rate": 4.993482062455887e-05, "loss": 1.6889, "step": 23088 }, { "epoch": 0.02, "learning_rate": 4.9934814956170764e-05, "loss": 1.0667, "step": 23089 }, { "epoch": 0.02, "learning_rate": 4.993480928753651e-05, "loss": 1.0738, "step": 23090 }, { "epoch": 0.02, "learning_rate": 4.993480361865611e-05, "loss": 1.0413, "step": 23091 }, { "epoch": 0.02, "learning_rate": 4.9934797949529565e-05, "loss": 0.9875, "step": 23092 }, { "epoch": 0.02, "learning_rate": 4.993479228015687e-05, "loss": 1.1007, "step": 23093 }, { "epoch": 0.02, "learning_rate": 4.993478661053803e-05, "loss": 1.0273, "step": 23094 }, { "epoch": 0.02, "learning_rate": 4.993478094067305e-05, "loss": 1.1697, "step": 23095 }, { "epoch": 0.02, "learning_rate": 4.993477527056192e-05, "loss": 1.8186, "step": 23096 }, { "epoch": 0.02, "learning_rate": 4.993476960020464e-05, "loss": 1.269, "step": 23097 }, { "epoch": 0.02, "learning_rate": 4.9934763929601225e-05, "loss": 1.3217, "step": 23098 }, { "epoch": 0.02, "learning_rate": 4.993475825875166e-05, "loss": 0.7491, "step": 23099 }, { "epoch": 0.02, "learning_rate": 4.9934752587655945e-05, "loss": 1.1563, "step": 23100 }, { "epoch": 0.02, "learning_rate": 4.993474691631409e-05, "loss": 1.6386, "step": 23101 }, { "epoch": 0.02, "learning_rate": 4.993474124472609e-05, "loss": 1.1117, "step": 23102 }, { "epoch": 0.02, "learning_rate": 4.993473557289194e-05, "loss": 1.1203, "step": 23103 }, { "epoch": 0.02, "learning_rate": 4.9934729900811646e-05, "loss": 0.688, "step": 23104 }, { "epoch": 0.02, "learning_rate": 4.993472422848521e-05, "loss": 1.1657, "step": 23105 }, { "epoch": 0.02, "learning_rate": 4.993471855591263e-05, "loss": 0.8521, "step": 23106 }, { "epoch": 0.02, "learning_rate": 4.993471288309389e-05, "loss": 0.8462, "step": 23107 }, { "epoch": 0.02, "learning_rate": 4.9934707210029016e-05, "loss": 1.0982, "step": 23108 }, { "epoch": 0.02, "learning_rate": 4.9934701536718006e-05, "loss": 0.8981, "step": 23109 }, { "epoch": 0.02, "learning_rate": 4.993469586316084e-05, "loss": 1.0856, "step": 23110 }, { "epoch": 0.02, "learning_rate": 4.9934690189357537e-05, "loss": 0.9739, "step": 23111 }, { "epoch": 0.02, "learning_rate": 4.993468451530808e-05, "loss": 1.2552, "step": 23112 }, { "epoch": 0.02, "learning_rate": 4.993467884101248e-05, "loss": 0.7775, "step": 23113 }, { "epoch": 0.02, "learning_rate": 4.993467316647075e-05, "loss": 1.0261, "step": 23114 }, { "epoch": 0.02, "learning_rate": 4.993466749168286e-05, "loss": 1.0804, "step": 23115 }, { "epoch": 0.02, "learning_rate": 4.9934661816648827e-05, "loss": 1.0074, "step": 23116 }, { "epoch": 0.02, "learning_rate": 4.9934656141368655e-05, "loss": 0.9904, "step": 23117 }, { "epoch": 0.02, "learning_rate": 4.993465046584233e-05, "loss": 1.0019, "step": 23118 }, { "epoch": 0.02, "learning_rate": 4.993464479006987e-05, "loss": 1.1856, "step": 23119 }, { "epoch": 0.02, "learning_rate": 4.993463911405126e-05, "loss": 1.1158, "step": 23120 }, { "epoch": 0.02, "learning_rate": 4.9934633437786504e-05, "loss": 1.1002, "step": 23121 }, { "epoch": 0.02, "learning_rate": 4.9934627761275607e-05, "loss": 1.1092, "step": 23122 }, { "epoch": 0.02, "learning_rate": 4.993462208451857e-05, "loss": 0.8211, "step": 23123 }, { "epoch": 0.02, "learning_rate": 4.993461640751538e-05, "loss": 0.993, "step": 23124 }, { "epoch": 0.02, "learning_rate": 4.993461073026605e-05, "loss": 0.9857, "step": 23125 }, { "epoch": 0.02, "learning_rate": 4.9934605052770574e-05, "loss": 1.0475, "step": 23126 }, { "epoch": 0.02, "learning_rate": 4.9934599375028966e-05, "loss": 0.6335, "step": 23127 }, { "epoch": 0.02, "learning_rate": 4.99345936970412e-05, "loss": 0.6242, "step": 23128 }, { "epoch": 0.02, "learning_rate": 4.99345880188073e-05, "loss": 0.5313, "step": 23129 }, { "epoch": 0.02, "learning_rate": 4.993458234032725e-05, "loss": 0.9406, "step": 23130 }, { "epoch": 0.02, "learning_rate": 4.993457666160106e-05, "loss": 0.8033, "step": 23131 }, { "epoch": 0.02, "learning_rate": 4.9934570982628725e-05, "loss": 0.88, "step": 23132 }, { "epoch": 0.02, "learning_rate": 4.993456530341024e-05, "loss": 1.0469, "step": 23133 }, { "epoch": 0.02, "learning_rate": 4.993455962394562e-05, "loss": 0.7635, "step": 23134 }, { "epoch": 0.02, "learning_rate": 4.993455394423485e-05, "loss": 1.0965, "step": 23135 }, { "epoch": 0.02, "learning_rate": 4.9934548264277945e-05, "loss": 1.0838, "step": 23136 }, { "epoch": 0.02, "learning_rate": 4.993454258407489e-05, "loss": 1.1403, "step": 23137 }, { "epoch": 0.02, "learning_rate": 4.99345369036257e-05, "loss": 1.1377, "step": 23138 }, { "epoch": 0.02, "learning_rate": 4.993453122293036e-05, "loss": 1.1707, "step": 23139 }, { "epoch": 0.02, "learning_rate": 4.9934525541988874e-05, "loss": 0.9139, "step": 23140 }, { "epoch": 0.02, "learning_rate": 4.9934519860801244e-05, "loss": 1.2194, "step": 23141 }, { "epoch": 0.02, "learning_rate": 4.993451417936747e-05, "loss": 1.0025, "step": 23142 }, { "epoch": 0.02, "learning_rate": 4.993450849768757e-05, "loss": 1.1549, "step": 23143 }, { "epoch": 0.02, "learning_rate": 4.993450281576151e-05, "loss": 1.1303, "step": 23144 }, { "epoch": 0.02, "learning_rate": 4.993449713358931e-05, "loss": 1.0568, "step": 23145 }, { "epoch": 0.02, "learning_rate": 4.993449145117097e-05, "loss": 0.9952, "step": 23146 }, { "epoch": 0.02, "learning_rate": 4.993448576850649e-05, "loss": 1.2394, "step": 23147 }, { "epoch": 0.02, "learning_rate": 4.9934480085595864e-05, "loss": 0.9275, "step": 23148 }, { "epoch": 0.02, "learning_rate": 4.993447440243909e-05, "loss": 0.9323, "step": 23149 }, { "epoch": 0.02, "learning_rate": 4.993446871903618e-05, "loss": 1.0981, "step": 23150 }, { "epoch": 0.02, "learning_rate": 4.9934463035387125e-05, "loss": 1.2442, "step": 23151 }, { "epoch": 0.02, "learning_rate": 4.993445735149193e-05, "loss": 1.382, "step": 23152 }, { "epoch": 0.02, "learning_rate": 4.993445166735059e-05, "loss": 1.5269, "step": 23153 }, { "epoch": 0.02, "learning_rate": 4.9934445982963105e-05, "loss": 1.1347, "step": 23154 }, { "epoch": 0.02, "learning_rate": 4.993444029832949e-05, "loss": 1.1598, "step": 23155 }, { "epoch": 0.02, "learning_rate": 4.993443461344972e-05, "loss": 0.8793, "step": 23156 }, { "epoch": 0.02, "learning_rate": 4.9934428928323815e-05, "loss": 1.562, "step": 23157 }, { "epoch": 0.02, "learning_rate": 4.993442324295176e-05, "loss": 1.9111, "step": 23158 }, { "epoch": 0.02, "learning_rate": 4.9934417557333565e-05, "loss": 1.5534, "step": 23159 }, { "epoch": 0.02, "learning_rate": 4.9934411871469235e-05, "loss": 0.8926, "step": 23160 }, { "epoch": 0.02, "learning_rate": 4.9934406185358765e-05, "loss": 1.1984, "step": 23161 }, { "epoch": 0.02, "learning_rate": 4.993440049900214e-05, "loss": 1.3421, "step": 23162 }, { "epoch": 0.02, "learning_rate": 4.9934394812399374e-05, "loss": 1.1659, "step": 23163 }, { "epoch": 0.02, "learning_rate": 4.9934389125550473e-05, "loss": 1.0761, "step": 23164 }, { "epoch": 0.02, "learning_rate": 4.993438343845543e-05, "loss": 0.7331, "step": 23165 }, { "epoch": 0.02, "learning_rate": 4.993437775111425e-05, "loss": 0.8914, "step": 23166 }, { "epoch": 0.02, "learning_rate": 4.993437206352692e-05, "loss": 1.5132, "step": 23167 }, { "epoch": 0.02, "learning_rate": 4.9934366375693455e-05, "loss": 0.9486, "step": 23168 }, { "epoch": 0.02, "learning_rate": 4.993436068761384e-05, "loss": 0.9884, "step": 23169 }, { "epoch": 0.02, "learning_rate": 4.993435499928809e-05, "loss": 0.9348, "step": 23170 }, { "epoch": 0.02, "learning_rate": 4.9934349310716196e-05, "loss": 0.9606, "step": 23171 }, { "epoch": 0.02, "learning_rate": 4.993434362189816e-05, "loss": 0.6987, "step": 23172 }, { "epoch": 0.02, "learning_rate": 4.9934337932833985e-05, "loss": 0.9939, "step": 23173 }, { "epoch": 0.02, "learning_rate": 4.993433224352366e-05, "loss": 0.8377, "step": 23174 }, { "epoch": 0.02, "learning_rate": 4.99343265539672e-05, "loss": 1.4567, "step": 23175 }, { "epoch": 0.02, "learning_rate": 4.993432086416461e-05, "loss": 0.8906, "step": 23176 }, { "epoch": 0.02, "learning_rate": 4.9934315174115864e-05, "loss": 1.0628, "step": 23177 }, { "epoch": 0.02, "learning_rate": 4.993430948382098e-05, "loss": 1.4498, "step": 23178 }, { "epoch": 0.02, "learning_rate": 4.9934303793279954e-05, "loss": 1.0737, "step": 23179 }, { "epoch": 0.02, "learning_rate": 4.993429810249279e-05, "loss": 1.1439, "step": 23180 }, { "epoch": 0.02, "learning_rate": 4.9934292411459494e-05, "loss": 1.1084, "step": 23181 }, { "epoch": 0.02, "learning_rate": 4.993428672018005e-05, "loss": 1.1309, "step": 23182 }, { "epoch": 0.02, "learning_rate": 4.9934281028654456e-05, "loss": 0.9647, "step": 23183 }, { "epoch": 0.02, "learning_rate": 4.993427533688273e-05, "loss": 1.1088, "step": 23184 }, { "epoch": 0.02, "learning_rate": 4.993426964486486e-05, "loss": 1.019, "step": 23185 }, { "epoch": 0.02, "learning_rate": 4.993426395260086e-05, "loss": 1.1446, "step": 23186 }, { "epoch": 0.02, "learning_rate": 4.993425826009071e-05, "loss": 1.3041, "step": 23187 }, { "epoch": 0.02, "learning_rate": 4.993425256733442e-05, "loss": 1.0589, "step": 23188 }, { "epoch": 0.02, "learning_rate": 4.993424687433198e-05, "loss": 1.2708, "step": 23189 }, { "epoch": 0.02, "learning_rate": 4.993424118108341e-05, "loss": 1.3329, "step": 23190 }, { "epoch": 0.02, "learning_rate": 4.9934235487588705e-05, "loss": 1.463, "step": 23191 }, { "epoch": 0.02, "learning_rate": 4.993422979384785e-05, "loss": 0.9503, "step": 23192 }, { "epoch": 0.02, "learning_rate": 4.993422409986086e-05, "loss": 1.402, "step": 23193 }, { "epoch": 0.02, "learning_rate": 4.993421840562773e-05, "loss": 1.1898, "step": 23194 }, { "epoch": 0.02, "learning_rate": 4.993421271114845e-05, "loss": 1.3067, "step": 23195 }, { "epoch": 0.02, "learning_rate": 4.993420701642305e-05, "loss": 0.8574, "step": 23196 }, { "epoch": 0.02, "learning_rate": 4.9934201321451494e-05, "loss": 1.1057, "step": 23197 }, { "epoch": 0.02, "learning_rate": 4.9934195626233806e-05, "loss": 1.0861, "step": 23198 }, { "epoch": 0.02, "learning_rate": 4.993418993076997e-05, "loss": 1.071, "step": 23199 }, { "epoch": 0.02, "learning_rate": 4.993418423506e-05, "loss": 1.055, "step": 23200 }, { "epoch": 0.02, "learning_rate": 4.993417853910389e-05, "loss": 0.877, "step": 23201 }, { "epoch": 0.02, "learning_rate": 4.993417284290164e-05, "loss": 1.2662, "step": 23202 }, { "epoch": 0.02, "learning_rate": 4.993416714645325e-05, "loss": 0.9654, "step": 23203 }, { "epoch": 0.02, "learning_rate": 4.993416144975872e-05, "loss": 0.9656, "step": 23204 }, { "epoch": 0.02, "learning_rate": 4.9934155752818046e-05, "loss": 0.8111, "step": 23205 }, { "epoch": 0.02, "learning_rate": 4.993415005563124e-05, "loss": 1.1934, "step": 23206 }, { "epoch": 0.02, "learning_rate": 4.9934144358198294e-05, "loss": 1.1579, "step": 23207 }, { "epoch": 0.02, "learning_rate": 4.9934138660519203e-05, "loss": 1.1547, "step": 23208 }, { "epoch": 0.02, "learning_rate": 4.993413296259398e-05, "loss": 1.155, "step": 23209 }, { "epoch": 0.02, "learning_rate": 4.993412726442261e-05, "loss": 1.1375, "step": 23210 }, { "epoch": 0.02, "learning_rate": 4.993412156600511e-05, "loss": 1.2495, "step": 23211 }, { "epoch": 0.02, "learning_rate": 4.993411586734146e-05, "loss": 1.0842, "step": 23212 }, { "epoch": 0.02, "learning_rate": 4.993411016843168e-05, "loss": 1.1401, "step": 23213 }, { "epoch": 0.02, "learning_rate": 4.993410446927575e-05, "loss": 1.3073, "step": 23214 }, { "epoch": 0.02, "learning_rate": 4.9934098769873695e-05, "loss": 1.1376, "step": 23215 }, { "epoch": 0.02, "learning_rate": 4.99340930702255e-05, "loss": 0.9504, "step": 23216 }, { "epoch": 0.02, "learning_rate": 4.993408737033115e-05, "loss": 1.1906, "step": 23217 }, { "epoch": 0.02, "learning_rate": 4.993408167019068e-05, "loss": 1.2061, "step": 23218 }, { "epoch": 0.02, "learning_rate": 4.993407596980406e-05, "loss": 1.0791, "step": 23219 }, { "epoch": 0.02, "learning_rate": 4.993407026917131e-05, "loss": 1.0531, "step": 23220 }, { "epoch": 0.02, "learning_rate": 4.9934064568292404e-05, "loss": 1.1521, "step": 23221 }, { "epoch": 0.02, "learning_rate": 4.993405886716738e-05, "loss": 1.025, "step": 23222 }, { "epoch": 0.02, "learning_rate": 4.99340531657962e-05, "loss": 1.3181, "step": 23223 }, { "epoch": 0.02, "learning_rate": 4.9934047464178895e-05, "loss": 1.1022, "step": 23224 }, { "epoch": 0.02, "learning_rate": 4.993404176231544e-05, "loss": 0.4759, "step": 23225 }, { "epoch": 0.02, "learning_rate": 4.993403606020586e-05, "loss": 1.0561, "step": 23226 }, { "epoch": 0.02, "learning_rate": 4.993403035785014e-05, "loss": 1.3042, "step": 23227 }, { "epoch": 0.02, "learning_rate": 4.993402465524827e-05, "loss": 1.0135, "step": 23228 }, { "epoch": 0.02, "learning_rate": 4.993401895240027e-05, "loss": 1.0974, "step": 23229 }, { "epoch": 0.02, "learning_rate": 4.993401324930613e-05, "loss": 1.1836, "step": 23230 }, { "epoch": 0.02, "learning_rate": 4.993400754596585e-05, "loss": 1.1168, "step": 23231 }, { "epoch": 0.02, "learning_rate": 4.993400184237944e-05, "loss": 1.0508, "step": 23232 }, { "epoch": 0.02, "learning_rate": 4.9933996138546884e-05, "loss": 1.1428, "step": 23233 }, { "epoch": 0.02, "learning_rate": 4.993399043446819e-05, "loss": 1.2427, "step": 23234 }, { "epoch": 0.02, "learning_rate": 4.9933984730143366e-05, "loss": 0.7216, "step": 23235 }, { "epoch": 0.02, "learning_rate": 4.9933979025572395e-05, "loss": 4.1468, "step": 23236 }, { "epoch": 0.02, "learning_rate": 4.993397332075529e-05, "loss": 5.7936, "step": 23237 }, { "epoch": 0.02, "learning_rate": 4.9933967615692044e-05, "loss": 5.7593, "step": 23238 }, { "epoch": 0.02, "learning_rate": 4.993396191038266e-05, "loss": 5.7522, "step": 23239 }, { "epoch": 0.02, "learning_rate": 4.993395620482715e-05, "loss": 5.2589, "step": 23240 }, { "epoch": 0.02, "learning_rate": 4.993395049902549e-05, "loss": 1.6599, "step": 23241 }, { "epoch": 0.02, "learning_rate": 4.9933944792977703e-05, "loss": 1.0671, "step": 23242 }, { "epoch": 0.02, "learning_rate": 4.993393908668377e-05, "loss": 1.2064, "step": 23243 }, { "epoch": 0.02, "learning_rate": 4.993393338014371e-05, "loss": 1.4545, "step": 23244 }, { "epoch": 0.02, "learning_rate": 4.99339276733575e-05, "loss": 1.3643, "step": 23245 }, { "epoch": 0.02, "learning_rate": 4.993392196632516e-05, "loss": 0.9779, "step": 23246 }, { "epoch": 0.02, "learning_rate": 4.993391625904668e-05, "loss": 0.8743, "step": 23247 }, { "epoch": 0.02, "learning_rate": 4.9933910551522065e-05, "loss": 0.9627, "step": 23248 }, { "epoch": 0.02, "learning_rate": 4.993390484375131e-05, "loss": 1.1667, "step": 23249 }, { "epoch": 0.02, "learning_rate": 4.993389913573442e-05, "loss": 1.6021, "step": 23250 }, { "epoch": 0.02, "learning_rate": 4.993389342747139e-05, "loss": 1.0278, "step": 23251 }, { "epoch": 0.02, "learning_rate": 4.993388771896223e-05, "loss": 1.1331, "step": 23252 }, { "epoch": 0.02, "learning_rate": 4.993388201020693e-05, "loss": 1.1256, "step": 23253 }, { "epoch": 0.02, "learning_rate": 4.993387630120549e-05, "loss": 1.2151, "step": 23254 }, { "epoch": 0.02, "learning_rate": 4.993387059195792e-05, "loss": 0.8404, "step": 23255 }, { "epoch": 0.02, "learning_rate": 4.99338648824642e-05, "loss": 1.0377, "step": 23256 }, { "epoch": 0.02, "learning_rate": 4.993385917272436e-05, "loss": 1.2286, "step": 23257 }, { "epoch": 0.02, "learning_rate": 4.993385346273837e-05, "loss": 0.8818, "step": 23258 }, { "epoch": 0.02, "learning_rate": 4.9933847752506254e-05, "loss": 1.2151, "step": 23259 }, { "epoch": 0.02, "learning_rate": 4.9933842042027994e-05, "loss": 1.0584, "step": 23260 }, { "epoch": 0.02, "learning_rate": 4.99338363313036e-05, "loss": 1.1344, "step": 23261 }, { "epoch": 0.02, "learning_rate": 4.993383062033307e-05, "loss": 0.9592, "step": 23262 }, { "epoch": 0.02, "learning_rate": 4.993382490911641e-05, "loss": 0.8394, "step": 23263 }, { "epoch": 0.02, "learning_rate": 4.9933819197653605e-05, "loss": 0.9241, "step": 23264 }, { "epoch": 0.02, "learning_rate": 4.993381348594467e-05, "loss": 0.9839, "step": 23265 }, { "epoch": 0.02, "learning_rate": 4.9933807773989595e-05, "loss": 0.9413, "step": 23266 }, { "epoch": 0.02, "learning_rate": 4.993380206178838e-05, "loss": 1.6318, "step": 23267 }, { "epoch": 0.02, "learning_rate": 4.9933796349341035e-05, "loss": 5.3795, "step": 23268 }, { "epoch": 0.02, "learning_rate": 4.9933790636647546e-05, "loss": 5.4089, "step": 23269 }, { "epoch": 0.02, "learning_rate": 4.993378492370793e-05, "loss": 5.741, "step": 23270 }, { "epoch": 0.02, "learning_rate": 4.993377921052218e-05, "loss": 5.5781, "step": 23271 }, { "epoch": 0.02, "learning_rate": 4.993377349709028e-05, "loss": 5.1289, "step": 23272 }, { "epoch": 0.02, "learning_rate": 4.9933767783412265e-05, "loss": 5.3855, "step": 23273 }, { "epoch": 0.02, "learning_rate": 4.99337620694881e-05, "loss": 3.9171, "step": 23274 }, { "epoch": 0.02, "learning_rate": 4.9933756355317805e-05, "loss": 1.0562, "step": 23275 }, { "epoch": 0.02, "learning_rate": 4.993375064090137e-05, "loss": 1.0869, "step": 23276 }, { "epoch": 0.02, "learning_rate": 4.99337449262388e-05, "loss": 1.0745, "step": 23277 }, { "epoch": 0.02, "learning_rate": 4.99337392113301e-05, "loss": 1.3202, "step": 23278 }, { "epoch": 0.02, "learning_rate": 4.993373349617526e-05, "loss": 0.9736, "step": 23279 }, { "epoch": 0.02, "learning_rate": 4.993372778077429e-05, "loss": 0.9415, "step": 23280 }, { "epoch": 0.02, "learning_rate": 4.993372206512718e-05, "loss": 0.9958, "step": 23281 }, { "epoch": 0.02, "learning_rate": 4.993371634923393e-05, "loss": 1.1369, "step": 23282 }, { "epoch": 0.02, "learning_rate": 4.9933710633094556e-05, "loss": 1.2525, "step": 23283 }, { "epoch": 0.02, "learning_rate": 4.993370491670904e-05, "loss": 1.0063, "step": 23284 }, { "epoch": 0.02, "learning_rate": 4.993369920007739e-05, "loss": 1.4447, "step": 23285 }, { "epoch": 0.02, "learning_rate": 4.993369348319961e-05, "loss": 1.1304, "step": 23286 }, { "epoch": 0.02, "learning_rate": 4.993368776607569e-05, "loss": 0.9209, "step": 23287 }, { "epoch": 0.02, "learning_rate": 4.993368204870563e-05, "loss": 0.8852, "step": 23288 }, { "epoch": 0.02, "learning_rate": 4.993367633108944e-05, "loss": 0.7293, "step": 23289 }, { "epoch": 0.02, "learning_rate": 4.993367061322712e-05, "loss": 0.3413, "step": 23290 }, { "epoch": 0.02, "learning_rate": 4.9933664895118656e-05, "loss": 0.277, "step": 23291 }, { "epoch": 0.02, "learning_rate": 4.993365917676406e-05, "loss": 0.3108, "step": 23292 }, { "epoch": 0.02, "learning_rate": 4.993365345816334e-05, "loss": 0.2486, "step": 23293 }, { "epoch": 0.02, "learning_rate": 4.993364773931647e-05, "loss": 0.4615, "step": 23294 }, { "epoch": 0.02, "learning_rate": 4.993364202022348e-05, "loss": 0.3067, "step": 23295 }, { "epoch": 0.02, "learning_rate": 4.993363630088435e-05, "loss": 0.38, "step": 23296 }, { "epoch": 0.02, "learning_rate": 4.993363058129908e-05, "loss": 0.4469, "step": 23297 }, { "epoch": 0.02, "learning_rate": 4.993362486146768e-05, "loss": 0.2989, "step": 23298 }, { "epoch": 0.02, "learning_rate": 4.9933619141390146e-05, "loss": 0.5672, "step": 23299 }, { "epoch": 0.02, "learning_rate": 4.9933613421066476e-05, "loss": 1.0456, "step": 23300 }, { "epoch": 0.02, "learning_rate": 4.993360770049668e-05, "loss": 1.089, "step": 23301 }, { "epoch": 0.02, "learning_rate": 4.993360197968074e-05, "loss": 0.8896, "step": 23302 }, { "epoch": 0.02, "learning_rate": 4.993359625861866e-05, "loss": 0.7015, "step": 23303 }, { "epoch": 0.02, "learning_rate": 4.993359053731046e-05, "loss": 1.0919, "step": 23304 }, { "epoch": 0.02, "learning_rate": 4.993358481575612e-05, "loss": 1.232, "step": 23305 }, { "epoch": 0.02, "learning_rate": 4.9933579093955654e-05, "loss": 1.3141, "step": 23306 }, { "epoch": 0.02, "learning_rate": 4.9933573371909045e-05, "loss": 1.7366, "step": 23307 }, { "epoch": 0.02, "learning_rate": 4.993356764961631e-05, "loss": 1.3943, "step": 23308 }, { "epoch": 0.02, "learning_rate": 4.993356192707743e-05, "loss": 1.0011, "step": 23309 }, { "epoch": 0.02, "learning_rate": 4.9933556204292426e-05, "loss": 1.1579, "step": 23310 }, { "epoch": 0.02, "learning_rate": 4.993355048126128e-05, "loss": 1.3984, "step": 23311 }, { "epoch": 0.02, "learning_rate": 4.993354475798401e-05, "loss": 1.0276, "step": 23312 }, { "epoch": 0.02, "learning_rate": 4.99335390344606e-05, "loss": 0.5449, "step": 23313 }, { "epoch": 0.02, "learning_rate": 4.993353331069106e-05, "loss": 0.6579, "step": 23314 }, { "epoch": 0.02, "learning_rate": 4.993352758667539e-05, "loss": 1.2305, "step": 23315 }, { "epoch": 0.02, "learning_rate": 4.993352186241358e-05, "loss": 1.143, "step": 23316 }, { "epoch": 0.02, "learning_rate": 4.993351613790564e-05, "loss": 1.1576, "step": 23317 }, { "epoch": 0.02, "learning_rate": 4.993351041315156e-05, "loss": 1.1882, "step": 23318 }, { "epoch": 0.02, "learning_rate": 4.9933504688151355e-05, "loss": 0.9701, "step": 23319 }, { "epoch": 0.02, "learning_rate": 4.9933498962905015e-05, "loss": 0.7091, "step": 23320 }, { "epoch": 0.02, "learning_rate": 4.993349323741254e-05, "loss": 0.275, "step": 23321 }, { "epoch": 0.02, "learning_rate": 4.993348751167394e-05, "loss": 0.9396, "step": 23322 }, { "epoch": 0.02, "learning_rate": 4.99334817856892e-05, "loss": 1.3934, "step": 23323 }, { "epoch": 0.02, "learning_rate": 4.993347605945833e-05, "loss": 1.0501, "step": 23324 }, { "epoch": 0.02, "learning_rate": 4.993347033298133e-05, "loss": 0.4584, "step": 23325 }, { "epoch": 0.02, "learning_rate": 4.993346460625819e-05, "loss": 0.7367, "step": 23326 }, { "epoch": 0.02, "learning_rate": 4.993345887928892e-05, "loss": 0.7613, "step": 23327 }, { "epoch": 0.02, "learning_rate": 4.9933453152073514e-05, "loss": 0.8282, "step": 23328 }, { "epoch": 0.02, "learning_rate": 4.993344742461198e-05, "loss": 1.2878, "step": 23329 }, { "epoch": 0.02, "learning_rate": 4.993344169690431e-05, "loss": 1.081, "step": 23330 }, { "epoch": 0.02, "learning_rate": 4.9933435968950514e-05, "loss": 0.9425, "step": 23331 }, { "epoch": 0.02, "learning_rate": 4.9933430240750576e-05, "loss": 1.0906, "step": 23332 }, { "epoch": 0.02, "learning_rate": 4.993342451230452e-05, "loss": 1.1445, "step": 23333 }, { "epoch": 0.02, "learning_rate": 4.993341878361232e-05, "loss": 1.0742, "step": 23334 }, { "epoch": 0.02, "learning_rate": 4.9933413054673985e-05, "loss": 1.0938, "step": 23335 }, { "epoch": 0.02, "learning_rate": 4.9933407325489525e-05, "loss": 1.2609, "step": 23336 }, { "epoch": 0.02, "learning_rate": 4.993340159605893e-05, "loss": 0.944, "step": 23337 }, { "epoch": 0.02, "learning_rate": 4.993339586638221e-05, "loss": 1.0673, "step": 23338 }, { "epoch": 0.02, "learning_rate": 4.9933390136459344e-05, "loss": 1.0947, "step": 23339 }, { "epoch": 0.02, "learning_rate": 4.993338440629036e-05, "loss": 1.0315, "step": 23340 }, { "epoch": 0.02, "learning_rate": 4.993337867587524e-05, "loss": 1.1006, "step": 23341 }, { "epoch": 0.02, "learning_rate": 4.993337294521399e-05, "loss": 1.1156, "step": 23342 }, { "epoch": 0.02, "learning_rate": 4.99333672143066e-05, "loss": 1.1108, "step": 23343 }, { "epoch": 0.02, "learning_rate": 4.993336148315309e-05, "loss": 1.1668, "step": 23344 }, { "epoch": 0.02, "learning_rate": 4.993335575175344e-05, "loss": 1.0407, "step": 23345 }, { "epoch": 0.02, "learning_rate": 4.993335002010766e-05, "loss": 1.1181, "step": 23346 }, { "epoch": 0.02, "learning_rate": 4.9933344288215746e-05, "loss": 0.9547, "step": 23347 }, { "epoch": 0.02, "learning_rate": 4.99333385560777e-05, "loss": 1.1953, "step": 23348 }, { "epoch": 0.02, "learning_rate": 4.993333282369353e-05, "loss": 1.2058, "step": 23349 }, { "epoch": 0.02, "learning_rate": 4.993332709106322e-05, "loss": 0.9971, "step": 23350 }, { "epoch": 0.02, "learning_rate": 4.993332135818679e-05, "loss": 0.8413, "step": 23351 }, { "epoch": 0.02, "learning_rate": 4.993331562506422e-05, "loss": 0.6554, "step": 23352 }, { "epoch": 0.02, "learning_rate": 4.993330989169552e-05, "loss": 1.5378, "step": 23353 }, { "epoch": 0.02, "learning_rate": 4.993330415808069e-05, "loss": 1.1317, "step": 23354 }, { "epoch": 0.02, "learning_rate": 4.9933298424219734e-05, "loss": 0.9963, "step": 23355 }, { "epoch": 0.02, "learning_rate": 4.993329269011263e-05, "loss": 0.9579, "step": 23356 }, { "epoch": 0.02, "learning_rate": 4.993328695575941e-05, "loss": 1.16, "step": 23357 }, { "epoch": 0.02, "learning_rate": 4.993328122116006e-05, "loss": 0.8475, "step": 23358 }, { "epoch": 0.02, "learning_rate": 4.993327548631457e-05, "loss": 1.0802, "step": 23359 }, { "epoch": 0.02, "learning_rate": 4.993326975122296e-05, "loss": 1.1835, "step": 23360 }, { "epoch": 0.02, "learning_rate": 4.993326401588521e-05, "loss": 1.1793, "step": 23361 }, { "epoch": 0.02, "learning_rate": 4.993325828030133e-05, "loss": 1.0708, "step": 23362 }, { "epoch": 0.02, "learning_rate": 4.993325254447133e-05, "loss": 0.9444, "step": 23363 }, { "epoch": 0.02, "learning_rate": 4.9933246808395184e-05, "loss": 1.0821, "step": 23364 }, { "epoch": 0.02, "learning_rate": 4.993324107207292e-05, "loss": 1.0358, "step": 23365 }, { "epoch": 0.02, "learning_rate": 4.993323533550452e-05, "loss": 0.9806, "step": 23366 }, { "epoch": 0.02, "learning_rate": 4.993322959868999e-05, "loss": 0.4699, "step": 23367 }, { "epoch": 0.02, "learning_rate": 4.993322386162933e-05, "loss": 1.1496, "step": 23368 }, { "epoch": 0.02, "learning_rate": 4.993321812432254e-05, "loss": 1.1119, "step": 23369 }, { "epoch": 0.02, "learning_rate": 4.993321238676962e-05, "loss": 0.9411, "step": 23370 }, { "epoch": 0.02, "learning_rate": 4.9933206648970564e-05, "loss": 1.6814, "step": 23371 }, { "epoch": 0.02, "learning_rate": 4.993320091092538e-05, "loss": 1.181, "step": 23372 }, { "epoch": 0.02, "learning_rate": 4.9933195172634074e-05, "loss": 0.913, "step": 23373 }, { "epoch": 0.02, "learning_rate": 4.9933189434096635e-05, "loss": 1.1855, "step": 23374 }, { "epoch": 0.02, "learning_rate": 4.993318369531306e-05, "loss": 1.2098, "step": 23375 }, { "epoch": 0.02, "learning_rate": 4.993317795628336e-05, "loss": 1.2998, "step": 23376 }, { "epoch": 0.02, "learning_rate": 4.993317221700753e-05, "loss": 1.1796, "step": 23377 }, { "epoch": 0.02, "learning_rate": 4.993316647748557e-05, "loss": 0.6056, "step": 23378 }, { "epoch": 0.02, "learning_rate": 4.993316073771748e-05, "loss": 0.6432, "step": 23379 }, { "epoch": 0.02, "learning_rate": 4.9933154997703255e-05, "loss": 0.6672, "step": 23380 }, { "epoch": 0.02, "learning_rate": 4.993314925744291e-05, "loss": 0.7372, "step": 23381 }, { "epoch": 0.02, "learning_rate": 4.993314351693643e-05, "loss": 0.503, "step": 23382 }, { "epoch": 0.02, "learning_rate": 4.9933137776183824e-05, "loss": 0.3402, "step": 23383 }, { "epoch": 0.02, "learning_rate": 4.993313203518509e-05, "loss": 0.9668, "step": 23384 }, { "epoch": 0.02, "learning_rate": 4.993312629394022e-05, "loss": 1.2158, "step": 23385 }, { "epoch": 0.02, "learning_rate": 4.993312055244922e-05, "loss": 1.0672, "step": 23386 }, { "epoch": 0.02, "learning_rate": 4.99331148107121e-05, "loss": 1.3195, "step": 23387 }, { "epoch": 0.02, "learning_rate": 4.9933109068728846e-05, "loss": 0.958, "step": 23388 }, { "epoch": 0.02, "learning_rate": 4.993310332649946e-05, "loss": 0.96, "step": 23389 }, { "epoch": 0.02, "learning_rate": 4.993309758402395e-05, "loss": 0.9401, "step": 23390 }, { "epoch": 0.02, "learning_rate": 4.99330918413023e-05, "loss": 0.949, "step": 23391 }, { "epoch": 0.02, "learning_rate": 4.993308609833453e-05, "loss": 1.1524, "step": 23392 }, { "epoch": 0.02, "learning_rate": 4.993308035512063e-05, "loss": 1.1711, "step": 23393 }, { "epoch": 0.02, "learning_rate": 4.9933074611660604e-05, "loss": 1.0838, "step": 23394 }, { "epoch": 0.02, "learning_rate": 4.993306886795445e-05, "loss": 1.1809, "step": 23395 }, { "epoch": 0.02, "learning_rate": 4.993306312400216e-05, "loss": 1.1751, "step": 23396 }, { "epoch": 0.02, "learning_rate": 4.993305737980374e-05, "loss": 1.0431, "step": 23397 }, { "epoch": 0.02, "learning_rate": 4.9933051635359194e-05, "loss": 0.9809, "step": 23398 }, { "epoch": 0.02, "learning_rate": 4.993304589066853e-05, "loss": 1.1022, "step": 23399 }, { "epoch": 0.02, "learning_rate": 4.993304014573172e-05, "loss": 1.1239, "step": 23400 }, { "epoch": 0.02, "learning_rate": 4.993303440054879e-05, "loss": 1.0594, "step": 23401 }, { "epoch": 0.02, "learning_rate": 4.993302865511974e-05, "loss": 1.2254, "step": 23402 }, { "epoch": 0.02, "learning_rate": 4.993302290944455e-05, "loss": 1.397, "step": 23403 }, { "epoch": 0.02, "learning_rate": 4.993301716352323e-05, "loss": 0.9546, "step": 23404 }, { "epoch": 0.02, "learning_rate": 4.993301141735579e-05, "loss": 0.8791, "step": 23405 }, { "epoch": 0.02, "learning_rate": 4.993300567094222e-05, "loss": 0.9978, "step": 23406 }, { "epoch": 0.02, "learning_rate": 4.9932999924282516e-05, "loss": 0.9931, "step": 23407 }, { "epoch": 0.02, "learning_rate": 4.9932994177376697e-05, "loss": 1.2079, "step": 23408 }, { "epoch": 0.02, "learning_rate": 4.993298843022474e-05, "loss": 1.3116, "step": 23409 }, { "epoch": 0.02, "learning_rate": 4.9932982682826654e-05, "loss": 1.3916, "step": 23410 }, { "epoch": 0.02, "learning_rate": 4.9932976935182445e-05, "loss": 1.2177, "step": 23411 }, { "epoch": 0.02, "learning_rate": 4.993297118729211e-05, "loss": 0.9864, "step": 23412 }, { "epoch": 0.02, "learning_rate": 4.9932965439155645e-05, "loss": 0.9315, "step": 23413 }, { "epoch": 0.02, "learning_rate": 4.993295969077305e-05, "loss": 0.6931, "step": 23414 }, { "epoch": 0.02, "learning_rate": 4.993295394214432e-05, "loss": 0.7407, "step": 23415 }, { "epoch": 0.02, "learning_rate": 4.9932948193269475e-05, "loss": 1.1855, "step": 23416 }, { "epoch": 0.02, "learning_rate": 4.99329424441485e-05, "loss": 1.2927, "step": 23417 }, { "epoch": 0.02, "learning_rate": 4.9932936694781393e-05, "loss": 1.2798, "step": 23418 }, { "epoch": 0.02, "learning_rate": 4.993293094516816e-05, "loss": 1.0461, "step": 23419 }, { "epoch": 0.02, "learning_rate": 4.99329251953088e-05, "loss": 0.9474, "step": 23420 }, { "epoch": 0.02, "learning_rate": 4.993291944520331e-05, "loss": 2.0329, "step": 23421 }, { "epoch": 0.02, "learning_rate": 4.99329136948517e-05, "loss": 2.5434, "step": 23422 }, { "epoch": 0.02, "learning_rate": 4.993290794425396e-05, "loss": 0.9622, "step": 23423 }, { "epoch": 0.02, "learning_rate": 4.993290219341009e-05, "loss": 1.571, "step": 23424 }, { "epoch": 0.02, "learning_rate": 4.99328964423201e-05, "loss": 0.7814, "step": 23425 }, { "epoch": 0.02, "learning_rate": 4.993289069098397e-05, "loss": 1.139, "step": 23426 }, { "epoch": 0.02, "learning_rate": 4.993288493940173e-05, "loss": 1.6188, "step": 23427 }, { "epoch": 0.02, "learning_rate": 4.9932879187573346e-05, "loss": 1.4267, "step": 23428 }, { "epoch": 0.02, "learning_rate": 4.9932873435498836e-05, "loss": 1.28, "step": 23429 }, { "epoch": 0.02, "learning_rate": 4.9932867683178205e-05, "loss": 2.7587, "step": 23430 }, { "epoch": 0.02, "learning_rate": 4.9932861930611454e-05, "loss": 1.1619, "step": 23431 }, { "epoch": 0.02, "learning_rate": 4.993285617779857e-05, "loss": 0.9199, "step": 23432 }, { "epoch": 0.02, "learning_rate": 4.993285042473956e-05, "loss": 0.5233, "step": 23433 }, { "epoch": 0.02, "learning_rate": 4.993284467143442e-05, "loss": 0.7261, "step": 23434 }, { "epoch": 0.02, "learning_rate": 4.993283891788316e-05, "loss": 0.5727, "step": 23435 }, { "epoch": 0.02, "learning_rate": 4.993283316408577e-05, "loss": 0.8431, "step": 23436 }, { "epoch": 0.02, "learning_rate": 4.993282741004226e-05, "loss": 1.145, "step": 23437 }, { "epoch": 0.02, "learning_rate": 4.993282165575261e-05, "loss": 0.7237, "step": 23438 }, { "epoch": 0.02, "learning_rate": 4.9932815901216844e-05, "loss": 1.056, "step": 23439 }, { "epoch": 0.02, "learning_rate": 4.993281014643495e-05, "loss": 1.0847, "step": 23440 }, { "epoch": 0.02, "learning_rate": 4.993280439140693e-05, "loss": 1.0575, "step": 23441 }, { "epoch": 0.02, "learning_rate": 4.9932798636132774e-05, "loss": 1.1646, "step": 23442 }, { "epoch": 0.02, "learning_rate": 4.99327928806125e-05, "loss": 0.9696, "step": 23443 }, { "epoch": 0.02, "learning_rate": 4.9932787124846106e-05, "loss": 1.0667, "step": 23444 }, { "epoch": 0.02, "learning_rate": 4.9932781368833584e-05, "loss": 0.8858, "step": 23445 }, { "epoch": 0.02, "learning_rate": 4.993277561257492e-05, "loss": 0.7612, "step": 23446 }, { "epoch": 0.02, "learning_rate": 4.993276985607015e-05, "loss": 0.4589, "step": 23447 }, { "epoch": 0.02, "learning_rate": 4.993276409931925e-05, "loss": 0.2337, "step": 23448 }, { "epoch": 0.02, "learning_rate": 4.9932758342322215e-05, "loss": 0.7243, "step": 23449 }, { "epoch": 0.02, "learning_rate": 4.993275258507906e-05, "loss": 1.1749, "step": 23450 }, { "epoch": 0.02, "learning_rate": 4.993274682758978e-05, "loss": 1.2144, "step": 23451 }, { "epoch": 0.02, "learning_rate": 4.993274106985437e-05, "loss": 1.5626, "step": 23452 }, { "epoch": 0.02, "learning_rate": 4.993273531187285e-05, "loss": 0.9917, "step": 23453 }, { "epoch": 0.02, "learning_rate": 4.9932729553645186e-05, "loss": 0.8859, "step": 23454 }, { "epoch": 0.02, "learning_rate": 4.9932723795171404e-05, "loss": 1.059, "step": 23455 }, { "epoch": 0.02, "learning_rate": 4.99327180364515e-05, "loss": 1.2748, "step": 23456 }, { "epoch": 0.02, "learning_rate": 4.993271227748546e-05, "loss": 1.0928, "step": 23457 }, { "epoch": 0.02, "learning_rate": 4.993270651827331e-05, "loss": 2.7888, "step": 23458 }, { "epoch": 0.02, "learning_rate": 4.993270075881502e-05, "loss": 2.442, "step": 23459 }, { "epoch": 0.02, "learning_rate": 4.993269499911061e-05, "loss": 1.7864, "step": 23460 }, { "epoch": 0.02, "learning_rate": 4.993268923916008e-05, "loss": 1.0732, "step": 23461 }, { "epoch": 0.02, "learning_rate": 4.9932683478963424e-05, "loss": 0.828, "step": 23462 }, { "epoch": 0.02, "learning_rate": 4.9932677718520635e-05, "loss": 1.0951, "step": 23463 }, { "epoch": 0.02, "learning_rate": 4.9932671957831726e-05, "loss": 1.2802, "step": 23464 }, { "epoch": 0.02, "learning_rate": 4.9932666196896696e-05, "loss": 1.3998, "step": 23465 }, { "epoch": 0.02, "learning_rate": 4.993266043571554e-05, "loss": 1.2392, "step": 23466 }, { "epoch": 0.02, "learning_rate": 4.993265467428825e-05, "loss": 1.1793, "step": 23467 }, { "epoch": 0.02, "learning_rate": 4.993264891261485e-05, "loss": 1.2057, "step": 23468 }, { "epoch": 0.02, "learning_rate": 4.9932643150695314e-05, "loss": 0.7938, "step": 23469 }, { "epoch": 0.02, "learning_rate": 4.9932637388529654e-05, "loss": 1.2047, "step": 23470 }, { "epoch": 0.02, "learning_rate": 4.993263162611788e-05, "loss": 0.9026, "step": 23471 }, { "epoch": 0.02, "learning_rate": 4.993262586345997e-05, "loss": 1.1477, "step": 23472 }, { "epoch": 0.02, "learning_rate": 4.993262010055594e-05, "loss": 1.1637, "step": 23473 }, { "epoch": 0.02, "learning_rate": 4.993261433740579e-05, "loss": 0.6325, "step": 23474 }, { "epoch": 0.02, "learning_rate": 4.993260857400951e-05, "loss": 1.0565, "step": 23475 }, { "epoch": 0.02, "learning_rate": 4.993260281036711e-05, "loss": 0.9072, "step": 23476 }, { "epoch": 0.02, "learning_rate": 4.993259704647858e-05, "loss": 0.6164, "step": 23477 }, { "epoch": 0.02, "learning_rate": 4.993259128234393e-05, "loss": 1.0499, "step": 23478 }, { "epoch": 0.02, "learning_rate": 4.993258551796316e-05, "loss": 2.9011, "step": 23479 }, { "epoch": 0.02, "learning_rate": 4.9932579753336256e-05, "loss": 0.9936, "step": 23480 }, { "epoch": 0.02, "learning_rate": 4.9932573988463234e-05, "loss": 0.9732, "step": 23481 }, { "epoch": 0.02, "learning_rate": 4.993256822334409e-05, "loss": 1.0005, "step": 23482 }, { "epoch": 0.02, "learning_rate": 4.9932562457978815e-05, "loss": 0.9007, "step": 23483 }, { "epoch": 0.02, "learning_rate": 4.9932556692367425e-05, "loss": 1.4692, "step": 23484 }, { "epoch": 0.02, "learning_rate": 4.99325509265099e-05, "loss": 1.117, "step": 23485 }, { "epoch": 0.02, "learning_rate": 4.993254516040626e-05, "loss": 1.0016, "step": 23486 }, { "epoch": 0.02, "learning_rate": 4.99325393940565e-05, "loss": 0.8793, "step": 23487 }, { "epoch": 0.02, "learning_rate": 4.993253362746061e-05, "loss": 0.934, "step": 23488 }, { "epoch": 0.02, "learning_rate": 4.9932527860618594e-05, "loss": 0.9091, "step": 23489 }, { "epoch": 0.02, "learning_rate": 4.9932522093530466e-05, "loss": 1.1824, "step": 23490 }, { "epoch": 0.02, "learning_rate": 4.99325163261962e-05, "loss": 1.2191, "step": 23491 }, { "epoch": 0.02, "learning_rate": 4.993251055861582e-05, "loss": 1.0254, "step": 23492 }, { "epoch": 0.02, "learning_rate": 4.993250479078932e-05, "loss": 0.9703, "step": 23493 }, { "epoch": 0.02, "learning_rate": 4.9932499022716686e-05, "loss": 1.2049, "step": 23494 }, { "epoch": 0.02, "learning_rate": 4.9932493254397934e-05, "loss": 0.8612, "step": 23495 }, { "epoch": 0.02, "learning_rate": 4.9932487485833055e-05, "loss": 0.8506, "step": 23496 }, { "epoch": 0.02, "learning_rate": 4.993248171702206e-05, "loss": 1.1419, "step": 23497 }, { "epoch": 0.02, "learning_rate": 4.993247594796494e-05, "loss": 1.047, "step": 23498 }, { "epoch": 0.02, "learning_rate": 4.993247017866169e-05, "loss": 1.1208, "step": 23499 }, { "epoch": 0.02, "learning_rate": 4.993246440911233e-05, "loss": 1.4139, "step": 23500 }, { "epoch": 0.02, "eval_loss": 1.019091248512268, "eval_runtime": 83.505, "eval_samples_per_second": 16.586, "eval_steps_per_second": 4.155, "step": 23500 }, { "epoch": 0.02, "learning_rate": 4.993245863931684e-05, "loss": 1.7553, "step": 23501 }, { "epoch": 0.02, "learning_rate": 4.9932452869275225e-05, "loss": 1.545, "step": 23502 }, { "epoch": 0.02, "learning_rate": 4.9932447098987494e-05, "loss": 1.0281, "step": 23503 }, { "epoch": 0.02, "learning_rate": 4.9932441328453636e-05, "loss": 0.9925, "step": 23504 }, { "epoch": 0.02, "learning_rate": 4.993243555767366e-05, "loss": 1.2199, "step": 23505 }, { "epoch": 0.02, "learning_rate": 4.993242978664756e-05, "loss": 1.6498, "step": 23506 }, { "epoch": 0.02, "learning_rate": 4.993242401537534e-05, "loss": 1.4192, "step": 23507 }, { "epoch": 0.02, "learning_rate": 4.993241824385698e-05, "loss": 0.9705, "step": 23508 }, { "epoch": 0.02, "learning_rate": 4.9932412472092515e-05, "loss": 1.2505, "step": 23509 }, { "epoch": 0.02, "learning_rate": 4.9932406700081926e-05, "loss": 1.1719, "step": 23510 }, { "epoch": 0.02, "learning_rate": 4.993240092782521e-05, "loss": 0.8925, "step": 23511 }, { "epoch": 0.02, "learning_rate": 4.993239515532237e-05, "loss": 1.2526, "step": 23512 }, { "epoch": 0.02, "learning_rate": 4.9932389382573415e-05, "loss": 1.1179, "step": 23513 }, { "epoch": 0.02, "learning_rate": 4.9932383609578336e-05, "loss": 0.7899, "step": 23514 }, { "epoch": 0.02, "learning_rate": 4.993237783633713e-05, "loss": 0.8602, "step": 23515 }, { "epoch": 0.02, "learning_rate": 4.993237206284981e-05, "loss": 1.1267, "step": 23516 }, { "epoch": 0.02, "learning_rate": 4.9932366289116364e-05, "loss": 1.4289, "step": 23517 }, { "epoch": 0.02, "learning_rate": 4.9932360515136796e-05, "loss": 1.0198, "step": 23518 }, { "epoch": 0.02, "learning_rate": 4.99323547409111e-05, "loss": 1.3792, "step": 23519 }, { "epoch": 0.02, "learning_rate": 4.993234896643929e-05, "loss": 1.3194, "step": 23520 }, { "epoch": 0.02, "learning_rate": 4.993234319172136e-05, "loss": 1.0939, "step": 23521 }, { "epoch": 0.02, "learning_rate": 4.9932337416757304e-05, "loss": 1.1187, "step": 23522 }, { "epoch": 0.02, "learning_rate": 4.9932331641547126e-05, "loss": 0.8891, "step": 23523 }, { "epoch": 0.02, "learning_rate": 4.9932325866090834e-05, "loss": 0.9343, "step": 23524 }, { "epoch": 0.02, "learning_rate": 4.9932320090388415e-05, "loss": 0.9085, "step": 23525 }, { "epoch": 0.02, "learning_rate": 4.993231431443987e-05, "loss": 0.9539, "step": 23526 }, { "epoch": 0.02, "learning_rate": 4.993230853824521e-05, "loss": 1.18, "step": 23527 }, { "epoch": 0.02, "learning_rate": 4.9932302761804426e-05, "loss": 1.2567, "step": 23528 }, { "epoch": 0.02, "learning_rate": 4.9932296985117524e-05, "loss": 1.0974, "step": 23529 }, { "epoch": 0.02, "learning_rate": 4.99322912081845e-05, "loss": 1.1708, "step": 23530 }, { "epoch": 0.02, "learning_rate": 4.993228543100535e-05, "loss": 1.4342, "step": 23531 }, { "epoch": 0.02, "learning_rate": 4.993227965358008e-05, "loss": 1.2641, "step": 23532 }, { "epoch": 0.02, "learning_rate": 4.9932273875908696e-05, "loss": 1.1019, "step": 23533 }, { "epoch": 0.02, "learning_rate": 4.9932268097991184e-05, "loss": 1.2149, "step": 23534 }, { "epoch": 0.02, "learning_rate": 4.993226231982756e-05, "loss": 0.5881, "step": 23535 }, { "epoch": 0.02, "learning_rate": 4.993225654141781e-05, "loss": 0.4593, "step": 23536 }, { "epoch": 0.02, "learning_rate": 4.993225076276194e-05, "loss": 0.7895, "step": 23537 }, { "epoch": 0.02, "learning_rate": 4.993224498385994e-05, "loss": 1.0561, "step": 23538 }, { "epoch": 0.02, "learning_rate": 4.993223920471183e-05, "loss": 1.1219, "step": 23539 }, { "epoch": 0.02, "learning_rate": 4.99322334253176e-05, "loss": 1.3474, "step": 23540 }, { "epoch": 0.02, "learning_rate": 4.993222764567724e-05, "loss": 0.7555, "step": 23541 }, { "epoch": 0.02, "learning_rate": 4.993222186579077e-05, "loss": 1.2043, "step": 23542 }, { "epoch": 0.02, "learning_rate": 4.993221608565817e-05, "loss": 1.1644, "step": 23543 }, { "epoch": 0.02, "learning_rate": 4.993221030527946e-05, "loss": 0.4963, "step": 23544 }, { "epoch": 0.02, "learning_rate": 4.993220452465463e-05, "loss": 1.0281, "step": 23545 }, { "epoch": 0.02, "learning_rate": 4.993219874378367e-05, "loss": 0.9727, "step": 23546 }, { "epoch": 0.02, "learning_rate": 4.993219296266659e-05, "loss": 1.0703, "step": 23547 }, { "epoch": 0.02, "learning_rate": 4.99321871813034e-05, "loss": 0.9731, "step": 23548 }, { "epoch": 0.02, "learning_rate": 4.9932181399694076e-05, "loss": 1.2766, "step": 23549 }, { "epoch": 0.02, "learning_rate": 4.993217561783865e-05, "loss": 1.061, "step": 23550 }, { "epoch": 0.02, "learning_rate": 4.993216983573708e-05, "loss": 0.9679, "step": 23551 }, { "epoch": 0.02, "learning_rate": 4.993216405338941e-05, "loss": 0.9538, "step": 23552 }, { "epoch": 0.02, "learning_rate": 4.9932158270795615e-05, "loss": 1.088, "step": 23553 }, { "epoch": 0.02, "learning_rate": 4.99321524879557e-05, "loss": 1.2186, "step": 23554 }, { "epoch": 0.02, "learning_rate": 4.993214670486966e-05, "loss": 1.0643, "step": 23555 }, { "epoch": 0.02, "learning_rate": 4.993214092153752e-05, "loss": 1.2714, "step": 23556 }, { "epoch": 0.02, "learning_rate": 4.993213513795924e-05, "loss": 0.9717, "step": 23557 }, { "epoch": 0.02, "learning_rate": 4.993212935413485e-05, "loss": 0.8208, "step": 23558 }, { "epoch": 0.02, "learning_rate": 4.993212357006433e-05, "loss": 0.3989, "step": 23559 }, { "epoch": 0.02, "learning_rate": 4.9932117785747704e-05, "loss": 0.3641, "step": 23560 }, { "epoch": 0.02, "learning_rate": 4.993211200118495e-05, "loss": 1.021, "step": 23561 }, { "epoch": 0.02, "learning_rate": 4.9932106216376076e-05, "loss": 1.3988, "step": 23562 }, { "epoch": 0.02, "learning_rate": 4.9932100431321085e-05, "loss": 1.0308, "step": 23563 }, { "epoch": 0.02, "learning_rate": 4.993209464601998e-05, "loss": 1.3788, "step": 23564 }, { "epoch": 0.02, "learning_rate": 4.993208886047275e-05, "loss": 1.5581, "step": 23565 }, { "epoch": 0.02, "learning_rate": 4.99320830746794e-05, "loss": 1.0108, "step": 23566 }, { "epoch": 0.02, "learning_rate": 4.9932077288639934e-05, "loss": 0.476, "step": 23567 }, { "epoch": 0.02, "learning_rate": 4.9932071502354346e-05, "loss": 1.0627, "step": 23568 }, { "epoch": 0.02, "learning_rate": 4.993206571582264e-05, "loss": 0.9829, "step": 23569 }, { "epoch": 0.02, "learning_rate": 4.9932059929044816e-05, "loss": 0.9804, "step": 23570 }, { "epoch": 0.02, "learning_rate": 4.993205414202088e-05, "loss": 1.2072, "step": 23571 }, { "epoch": 0.02, "learning_rate": 4.9932048354750816e-05, "loss": 1.2994, "step": 23572 }, { "epoch": 0.02, "learning_rate": 4.993204256723464e-05, "loss": 1.1765, "step": 23573 }, { "epoch": 0.02, "learning_rate": 4.993203677947233e-05, "loss": 1.0123, "step": 23574 }, { "epoch": 0.02, "learning_rate": 4.993203099146392e-05, "loss": 1.1781, "step": 23575 }, { "epoch": 0.02, "learning_rate": 4.993202520320939e-05, "loss": 1.2677, "step": 23576 }, { "epoch": 0.02, "learning_rate": 4.993201941470873e-05, "loss": 1.1516, "step": 23577 }, { "epoch": 0.02, "learning_rate": 4.993201362596196e-05, "loss": 0.9365, "step": 23578 }, { "epoch": 0.02, "learning_rate": 4.993200783696907e-05, "loss": 1.0756, "step": 23579 }, { "epoch": 0.02, "learning_rate": 4.993200204773006e-05, "loss": 1.3574, "step": 23580 }, { "epoch": 0.02, "learning_rate": 4.9931996258244934e-05, "loss": 0.922, "step": 23581 }, { "epoch": 0.02, "learning_rate": 4.993199046851369e-05, "loss": 1.2906, "step": 23582 }, { "epoch": 0.02, "learning_rate": 4.993198467853633e-05, "loss": 0.7179, "step": 23583 }, { "epoch": 0.02, "learning_rate": 4.993197888831284e-05, "loss": 1.1182, "step": 23584 }, { "epoch": 0.02, "learning_rate": 4.9931973097843244e-05, "loss": 0.34, "step": 23585 }, { "epoch": 0.02, "learning_rate": 4.9931967307127525e-05, "loss": 0.7896, "step": 23586 }, { "epoch": 0.02, "learning_rate": 4.993196151616569e-05, "loss": 1.0032, "step": 23587 }, { "epoch": 0.02, "learning_rate": 4.993195572495774e-05, "loss": 0.5165, "step": 23588 }, { "epoch": 0.02, "learning_rate": 4.9931949933503664e-05, "loss": 1.0506, "step": 23589 }, { "epoch": 0.02, "learning_rate": 4.9931944141803476e-05, "loss": 0.7367, "step": 23590 }, { "epoch": 0.02, "learning_rate": 4.993193834985717e-05, "loss": 1.178, "step": 23591 }, { "epoch": 0.02, "learning_rate": 4.993193255766475e-05, "loss": 1.204, "step": 23592 }, { "epoch": 0.02, "learning_rate": 4.9931926765226214e-05, "loss": 1.1325, "step": 23593 }, { "epoch": 0.02, "learning_rate": 4.993192097254154e-05, "loss": 1.0491, "step": 23594 }, { "epoch": 0.02, "learning_rate": 4.993191517961077e-05, "loss": 1.0496, "step": 23595 }, { "epoch": 0.02, "learning_rate": 4.993190938643388e-05, "loss": 1.3832, "step": 23596 }, { "epoch": 0.02, "learning_rate": 4.993190359301087e-05, "loss": 1.4029, "step": 23597 }, { "epoch": 0.02, "learning_rate": 4.993189779934174e-05, "loss": 0.8646, "step": 23598 }, { "epoch": 0.02, "learning_rate": 4.993189200542649e-05, "loss": 1.147, "step": 23599 }, { "epoch": 0.02, "learning_rate": 4.993188621126513e-05, "loss": 1.1318, "step": 23600 }, { "epoch": 0.02, "learning_rate": 4.993188041685765e-05, "loss": 1.6338, "step": 23601 }, { "epoch": 0.02, "learning_rate": 4.9931874622204055e-05, "loss": 1.1445, "step": 23602 }, { "epoch": 0.02, "learning_rate": 4.993186882730434e-05, "loss": 0.9323, "step": 23603 }, { "epoch": 0.02, "learning_rate": 4.993186303215851e-05, "loss": 0.7262, "step": 23604 }, { "epoch": 0.02, "learning_rate": 4.9931857236766564e-05, "loss": 1.0773, "step": 23605 }, { "epoch": 0.02, "learning_rate": 4.99318514411285e-05, "loss": 1.1596, "step": 23606 }, { "epoch": 0.02, "learning_rate": 4.993184564524433e-05, "loss": 1.0628, "step": 23607 }, { "epoch": 0.02, "learning_rate": 4.993183984911403e-05, "loss": 1.1167, "step": 23608 }, { "epoch": 0.02, "learning_rate": 4.993183405273761e-05, "loss": 1.0022, "step": 23609 }, { "epoch": 0.02, "learning_rate": 4.9931828256115074e-05, "loss": 0.9754, "step": 23610 }, { "epoch": 0.02, "learning_rate": 4.993182245924643e-05, "loss": 1.0467, "step": 23611 }, { "epoch": 0.02, "learning_rate": 4.993181666213167e-05, "loss": 1.1302, "step": 23612 }, { "epoch": 0.02, "learning_rate": 4.993181086477079e-05, "loss": 1.6587, "step": 23613 }, { "epoch": 0.02, "learning_rate": 4.993180506716379e-05, "loss": 5.2102, "step": 23614 }, { "epoch": 0.02, "learning_rate": 4.9931799269310676e-05, "loss": 5.2838, "step": 23615 }, { "epoch": 0.02, "learning_rate": 4.993179347121145e-05, "loss": 5.4952, "step": 23616 }, { "epoch": 0.02, "learning_rate": 4.993178767286611e-05, "loss": 5.4624, "step": 23617 }, { "epoch": 0.02, "learning_rate": 4.993178187427464e-05, "loss": 5.7819, "step": 23618 }, { "epoch": 0.02, "learning_rate": 4.993177607543706e-05, "loss": 5.8194, "step": 23619 }, { "epoch": 0.02, "learning_rate": 4.9931770276353376e-05, "loss": 5.9286, "step": 23620 }, { "epoch": 0.02, "learning_rate": 4.993176447702357e-05, "loss": 2.4721, "step": 23621 }, { "epoch": 0.02, "learning_rate": 4.9931758677447634e-05, "loss": 1.0889, "step": 23622 }, { "epoch": 0.02, "learning_rate": 4.9931752877625606e-05, "loss": 1.1775, "step": 23623 }, { "epoch": 0.02, "learning_rate": 4.9931747077557444e-05, "loss": 1.0515, "step": 23624 }, { "epoch": 0.02, "learning_rate": 4.9931741277243174e-05, "loss": 1.1736, "step": 23625 }, { "epoch": 0.02, "learning_rate": 4.9931735476682784e-05, "loss": 1.2036, "step": 23626 }, { "epoch": 0.02, "learning_rate": 4.993172967587628e-05, "loss": 1.0255, "step": 23627 }, { "epoch": 0.02, "learning_rate": 4.993172387482366e-05, "loss": 1.1074, "step": 23628 }, { "epoch": 0.02, "learning_rate": 4.9931718073524924e-05, "loss": 0.8979, "step": 23629 }, { "epoch": 0.02, "learning_rate": 4.993171227198008e-05, "loss": 1.0418, "step": 23630 }, { "epoch": 0.02, "learning_rate": 4.993170647018911e-05, "loss": 0.8295, "step": 23631 }, { "epoch": 0.02, "learning_rate": 4.993170066815203e-05, "loss": 0.5075, "step": 23632 }, { "epoch": 0.02, "learning_rate": 4.993169486586884e-05, "loss": 0.6138, "step": 23633 }, { "epoch": 0.02, "learning_rate": 4.993168906333953e-05, "loss": 0.4799, "step": 23634 }, { "epoch": 0.02, "learning_rate": 4.99316832605641e-05, "loss": 0.5506, "step": 23635 }, { "epoch": 0.02, "learning_rate": 4.993167745754256e-05, "loss": 1.0181, "step": 23636 }, { "epoch": 0.02, "learning_rate": 4.99316716542749e-05, "loss": 1.0466, "step": 23637 }, { "epoch": 0.02, "learning_rate": 4.993166585076113e-05, "loss": 0.9689, "step": 23638 }, { "epoch": 0.02, "learning_rate": 4.993166004700125e-05, "loss": 1.102, "step": 23639 }, { "epoch": 0.02, "learning_rate": 4.993165424299524e-05, "loss": 0.3556, "step": 23640 }, { "epoch": 0.02, "learning_rate": 4.993164843874313e-05, "loss": 0.2803, "step": 23641 }, { "epoch": 0.02, "learning_rate": 4.9931642634244904e-05, "loss": 0.2194, "step": 23642 }, { "epoch": 0.02, "learning_rate": 4.993163682950055e-05, "loss": 0.6376, "step": 23643 }, { "epoch": 0.02, "learning_rate": 4.99316310245101e-05, "loss": 1.0771, "step": 23644 }, { "epoch": 0.02, "learning_rate": 4.993162521927352e-05, "loss": 1.171, "step": 23645 }, { "epoch": 0.02, "learning_rate": 4.993161941379083e-05, "loss": 1.0057, "step": 23646 }, { "epoch": 0.02, "learning_rate": 4.9931613608062026e-05, "loss": 0.9148, "step": 23647 }, { "epoch": 0.02, "learning_rate": 4.993160780208711e-05, "loss": 1.1109, "step": 23648 }, { "epoch": 0.02, "learning_rate": 4.993160199586608e-05, "loss": 1.1471, "step": 23649 }, { "epoch": 0.02, "learning_rate": 4.9931596189398936e-05, "loss": 1.0191, "step": 23650 }, { "epoch": 0.02, "learning_rate": 4.993159038268567e-05, "loss": 0.7647, "step": 23651 }, { "epoch": 0.02, "learning_rate": 4.99315845757263e-05, "loss": 1.1461, "step": 23652 }, { "epoch": 0.02, "learning_rate": 4.993157876852081e-05, "loss": 1.3627, "step": 23653 }, { "epoch": 0.02, "learning_rate": 4.993157296106921e-05, "loss": 1.086, "step": 23654 }, { "epoch": 0.02, "learning_rate": 4.9931567153371495e-05, "loss": 1.1152, "step": 23655 }, { "epoch": 0.02, "learning_rate": 4.993156134542766e-05, "loss": 1.2105, "step": 23656 }, { "epoch": 0.02, "learning_rate": 4.993155553723772e-05, "loss": 0.5528, "step": 23657 }, { "epoch": 0.02, "learning_rate": 4.993154972880166e-05, "loss": 0.2873, "step": 23658 }, { "epoch": 0.02, "learning_rate": 4.993154392011948e-05, "loss": 0.252, "step": 23659 }, { "epoch": 0.02, "learning_rate": 4.99315381111912e-05, "loss": 0.176, "step": 23660 }, { "epoch": 0.02, "learning_rate": 4.99315323020168e-05, "loss": 0.3991, "step": 23661 }, { "epoch": 0.02, "learning_rate": 4.993152649259629e-05, "loss": 0.6287, "step": 23662 }, { "epoch": 0.02, "learning_rate": 4.993152068292966e-05, "loss": 1.0101, "step": 23663 }, { "epoch": 0.02, "learning_rate": 4.993151487301692e-05, "loss": 0.9747, "step": 23664 }, { "epoch": 0.02, "learning_rate": 4.993150906285807e-05, "loss": 0.2485, "step": 23665 }, { "epoch": 0.02, "learning_rate": 4.9931503252453104e-05, "loss": 0.4099, "step": 23666 }, { "epoch": 0.02, "learning_rate": 4.9931497441802024e-05, "loss": 0.2613, "step": 23667 }, { "epoch": 0.02, "learning_rate": 4.993149163090483e-05, "loss": 0.7652, "step": 23668 }, { "epoch": 0.02, "learning_rate": 4.993148581976152e-05, "loss": 1.3451, "step": 23669 }, { "epoch": 0.02, "learning_rate": 4.99314800083721e-05, "loss": 1.252, "step": 23670 }, { "epoch": 0.02, "learning_rate": 4.993147419673657e-05, "loss": 1.0193, "step": 23671 }, { "epoch": 0.02, "learning_rate": 4.993146838485493e-05, "loss": 0.1945, "step": 23672 }, { "epoch": 0.02, "learning_rate": 4.9931462572727166e-05, "loss": 0.5495, "step": 23673 }, { "epoch": 0.02, "learning_rate": 4.9931456760353295e-05, "loss": 0.8914, "step": 23674 }, { "epoch": 0.02, "learning_rate": 4.993145094773332e-05, "loss": 0.9787, "step": 23675 }, { "epoch": 0.02, "learning_rate": 4.993144513486722e-05, "loss": 1.1274, "step": 23676 }, { "epoch": 0.02, "learning_rate": 4.9931439321755e-05, "loss": 0.8015, "step": 23677 }, { "epoch": 0.02, "learning_rate": 4.993143350839669e-05, "loss": 0.8738, "step": 23678 }, { "epoch": 0.02, "learning_rate": 4.993142769479225e-05, "loss": 0.8787, "step": 23679 }, { "epoch": 0.02, "learning_rate": 4.99314218809417e-05, "loss": 1.3542, "step": 23680 }, { "epoch": 0.02, "learning_rate": 4.993141606684504e-05, "loss": 1.1512, "step": 23681 }, { "epoch": 0.02, "learning_rate": 4.993141025250227e-05, "loss": 0.9856, "step": 23682 }, { "epoch": 0.02, "learning_rate": 4.993140443791339e-05, "loss": 1.2006, "step": 23683 }, { "epoch": 0.02, "learning_rate": 4.993139862307839e-05, "loss": 1.1093, "step": 23684 }, { "epoch": 0.02, "learning_rate": 4.993139280799728e-05, "loss": 1.2161, "step": 23685 }, { "epoch": 0.02, "learning_rate": 4.993138699267006e-05, "loss": 0.8431, "step": 23686 }, { "epoch": 0.02, "learning_rate": 4.993138117709672e-05, "loss": 1.0788, "step": 23687 }, { "epoch": 0.02, "learning_rate": 4.993137536127728e-05, "loss": 1.1481, "step": 23688 }, { "epoch": 0.02, "learning_rate": 4.993136954521171e-05, "loss": 1.0185, "step": 23689 }, { "epoch": 0.02, "learning_rate": 4.993136372890005e-05, "loss": 0.9628, "step": 23690 }, { "epoch": 0.02, "learning_rate": 4.993135791234227e-05, "loss": 1.2382, "step": 23691 }, { "epoch": 0.02, "learning_rate": 4.9931352095538374e-05, "loss": 1.0777, "step": 23692 }, { "epoch": 0.02, "learning_rate": 4.9931346278488367e-05, "loss": 0.9592, "step": 23693 }, { "epoch": 0.02, "learning_rate": 4.993134046119225e-05, "loss": 1.1439, "step": 23694 }, { "epoch": 0.02, "learning_rate": 4.993133464365002e-05, "loss": 1.1385, "step": 23695 }, { "epoch": 0.02, "learning_rate": 4.993132882586168e-05, "loss": 1.0657, "step": 23696 }, { "epoch": 0.02, "learning_rate": 4.993132300782723e-05, "loss": 1.2884, "step": 23697 }, { "epoch": 0.02, "learning_rate": 4.993131718954667e-05, "loss": 0.8459, "step": 23698 }, { "epoch": 0.02, "learning_rate": 4.993131137102e-05, "loss": 1.1862, "step": 23699 }, { "epoch": 0.02, "learning_rate": 4.993130555224721e-05, "loss": 1.1447, "step": 23700 }, { "epoch": 0.02, "learning_rate": 4.993129973322831e-05, "loss": 0.9087, "step": 23701 }, { "epoch": 0.02, "learning_rate": 4.99312939139633e-05, "loss": 0.8303, "step": 23702 }, { "epoch": 0.02, "learning_rate": 4.993128809445218e-05, "loss": 1.0653, "step": 23703 }, { "epoch": 0.02, "learning_rate": 4.9931282274694945e-05, "loss": 1.1346, "step": 23704 }, { "epoch": 0.02, "learning_rate": 4.9931276454691604e-05, "loss": 0.9446, "step": 23705 }, { "epoch": 0.02, "learning_rate": 4.993127063444215e-05, "loss": 1.0693, "step": 23706 }, { "epoch": 0.02, "learning_rate": 4.993126481394659e-05, "loss": 1.405, "step": 23707 }, { "epoch": 0.02, "learning_rate": 4.993125899320491e-05, "loss": 1.082, "step": 23708 }, { "epoch": 0.02, "learning_rate": 4.9931253172217125e-05, "loss": 1.0827, "step": 23709 }, { "epoch": 0.02, "learning_rate": 4.993124735098322e-05, "loss": 1.0239, "step": 23710 }, { "epoch": 0.02, "learning_rate": 4.993124152950322e-05, "loss": 1.069, "step": 23711 }, { "epoch": 0.02, "learning_rate": 4.99312357077771e-05, "loss": 1.0926, "step": 23712 }, { "epoch": 0.02, "learning_rate": 4.993122988580487e-05, "loss": 1.0431, "step": 23713 }, { "epoch": 0.02, "learning_rate": 4.993122406358653e-05, "loss": 1.1593, "step": 23714 }, { "epoch": 0.02, "learning_rate": 4.993121824112208e-05, "loss": 0.9782, "step": 23715 }, { "epoch": 0.02, "learning_rate": 4.993121241841152e-05, "loss": 0.9652, "step": 23716 }, { "epoch": 0.02, "learning_rate": 4.993120659545485e-05, "loss": 1.4466, "step": 23717 }, { "epoch": 0.02, "learning_rate": 4.993120077225207e-05, "loss": 1.2158, "step": 23718 }, { "epoch": 0.02, "learning_rate": 4.9931194948803174e-05, "loss": 0.9522, "step": 23719 }, { "epoch": 0.02, "learning_rate": 4.9931189125108166e-05, "loss": 1.0287, "step": 23720 }, { "epoch": 0.02, "learning_rate": 4.9931183301167065e-05, "loss": 1.0403, "step": 23721 }, { "epoch": 0.02, "learning_rate": 4.993117747697984e-05, "loss": 0.7293, "step": 23722 }, { "epoch": 0.02, "learning_rate": 4.9931171652546506e-05, "loss": 1.1244, "step": 23723 }, { "epoch": 0.02, "learning_rate": 4.993116582786706e-05, "loss": 1.0774, "step": 23724 }, { "epoch": 0.02, "learning_rate": 4.993116000294151e-05, "loss": 1.0303, "step": 23725 }, { "epoch": 0.02, "learning_rate": 4.993115417776985e-05, "loss": 0.8624, "step": 23726 }, { "epoch": 0.02, "learning_rate": 4.993114835235208e-05, "loss": 1.089, "step": 23727 }, { "epoch": 0.02, "learning_rate": 4.993114252668819e-05, "loss": 1.1367, "step": 23728 }, { "epoch": 0.02, "learning_rate": 4.99311367007782e-05, "loss": 1.1349, "step": 23729 }, { "epoch": 0.02, "learning_rate": 4.99311308746221e-05, "loss": 0.9834, "step": 23730 }, { "epoch": 0.02, "learning_rate": 4.993112504821989e-05, "loss": 0.6479, "step": 23731 }, { "epoch": 0.02, "learning_rate": 4.9931119221571567e-05, "loss": 0.8246, "step": 23732 }, { "epoch": 0.02, "learning_rate": 4.993111339467714e-05, "loss": 1.1826, "step": 23733 }, { "epoch": 0.02, "learning_rate": 4.99311075675366e-05, "loss": 1.0771, "step": 23734 }, { "epoch": 0.02, "learning_rate": 4.993110174014994e-05, "loss": 0.7828, "step": 23735 }, { "epoch": 0.02, "learning_rate": 4.9931095912517186e-05, "loss": 0.9775, "step": 23736 }, { "epoch": 0.02, "learning_rate": 4.9931090084638324e-05, "loss": 1.1782, "step": 23737 }, { "epoch": 0.02, "learning_rate": 4.993108425651335e-05, "loss": 1.6867, "step": 23738 }, { "epoch": 0.02, "learning_rate": 4.9931078428142256e-05, "loss": 0.9503, "step": 23739 }, { "epoch": 0.02, "learning_rate": 4.9931072599525065e-05, "loss": 0.9879, "step": 23740 }, { "epoch": 0.02, "learning_rate": 4.993106677066176e-05, "loss": 0.8653, "step": 23741 }, { "epoch": 0.02, "learning_rate": 4.993106094155234e-05, "loss": 0.9791, "step": 23742 }, { "epoch": 0.02, "learning_rate": 4.993105511219682e-05, "loss": 1.0325, "step": 23743 }, { "epoch": 0.02, "learning_rate": 4.993104928259519e-05, "loss": 1.5904, "step": 23744 }, { "epoch": 0.02, "learning_rate": 4.993104345274745e-05, "loss": 1.3606, "step": 23745 }, { "epoch": 0.02, "learning_rate": 4.99310376226536e-05, "loss": 1.1178, "step": 23746 }, { "epoch": 0.02, "learning_rate": 4.993103179231364e-05, "loss": 1.1239, "step": 23747 }, { "epoch": 0.02, "learning_rate": 4.993102596172758e-05, "loss": 1.0504, "step": 23748 }, { "epoch": 0.02, "learning_rate": 4.9931020130895405e-05, "loss": 1.0831, "step": 23749 }, { "epoch": 0.02, "learning_rate": 4.993101429981712e-05, "loss": 1.3501, "step": 23750 }, { "epoch": 0.02, "learning_rate": 4.993100846849273e-05, "loss": 1.1157, "step": 23751 }, { "epoch": 0.02, "learning_rate": 4.993100263692223e-05, "loss": 1.5701, "step": 23752 }, { "epoch": 0.02, "learning_rate": 4.9930996805105625e-05, "loss": 1.4327, "step": 23753 }, { "epoch": 0.02, "learning_rate": 4.99309909730429e-05, "loss": 1.0792, "step": 23754 }, { "epoch": 0.02, "learning_rate": 4.9930985140734076e-05, "loss": 1.0079, "step": 23755 }, { "epoch": 0.02, "learning_rate": 4.993097930817915e-05, "loss": 0.8345, "step": 23756 }, { "epoch": 0.02, "learning_rate": 4.9930973475378105e-05, "loss": 0.9985, "step": 23757 }, { "epoch": 0.02, "learning_rate": 4.993096764233096e-05, "loss": 1.067, "step": 23758 }, { "epoch": 0.02, "learning_rate": 4.99309618090377e-05, "loss": 1.1998, "step": 23759 }, { "epoch": 0.02, "learning_rate": 4.993095597549833e-05, "loss": 1.2613, "step": 23760 }, { "epoch": 0.02, "learning_rate": 4.9930950141712864e-05, "loss": 0.9185, "step": 23761 }, { "epoch": 0.02, "learning_rate": 4.993094430768128e-05, "loss": 0.8382, "step": 23762 }, { "epoch": 0.02, "learning_rate": 4.9930938473403594e-05, "loss": 0.9646, "step": 23763 }, { "epoch": 0.02, "learning_rate": 4.99309326388798e-05, "loss": 0.566, "step": 23764 }, { "epoch": 0.02, "learning_rate": 4.993092680410989e-05, "loss": 1.3314, "step": 23765 }, { "epoch": 0.02, "learning_rate": 4.993092096909388e-05, "loss": 1.3131, "step": 23766 }, { "epoch": 0.02, "learning_rate": 4.9930915133831756e-05, "loss": 0.9487, "step": 23767 }, { "epoch": 0.02, "learning_rate": 4.993090929832354e-05, "loss": 1.0897, "step": 23768 }, { "epoch": 0.02, "learning_rate": 4.99309034625692e-05, "loss": 0.9157, "step": 23769 }, { "epoch": 0.02, "learning_rate": 4.993089762656876e-05, "loss": 1.2563, "step": 23770 }, { "epoch": 0.02, "learning_rate": 4.993089179032221e-05, "loss": 0.9168, "step": 23771 }, { "epoch": 0.02, "learning_rate": 4.993088595382955e-05, "loss": 1.2192, "step": 23772 }, { "epoch": 0.02, "learning_rate": 4.993088011709079e-05, "loss": 0.8755, "step": 23773 }, { "epoch": 0.02, "learning_rate": 4.9930874280105924e-05, "loss": 1.1551, "step": 23774 }, { "epoch": 0.02, "learning_rate": 4.993086844287494e-05, "loss": 1.2147, "step": 23775 }, { "epoch": 0.02, "learning_rate": 4.9930862605397855e-05, "loss": 1.0485, "step": 23776 }, { "epoch": 0.02, "learning_rate": 4.993085676767467e-05, "loss": 1.1032, "step": 23777 }, { "epoch": 0.02, "learning_rate": 4.9930850929705365e-05, "loss": 0.8923, "step": 23778 }, { "epoch": 0.02, "learning_rate": 4.993084509148996e-05, "loss": 1.1027, "step": 23779 }, { "epoch": 0.02, "learning_rate": 4.9930839253028446e-05, "loss": 1.2721, "step": 23780 }, { "epoch": 0.02, "learning_rate": 4.993083341432083e-05, "loss": 1.2301, "step": 23781 }, { "epoch": 0.02, "learning_rate": 4.9930827575367106e-05, "loss": 1.1322, "step": 23782 }, { "epoch": 0.02, "learning_rate": 4.9930821736167275e-05, "loss": 1.0031, "step": 23783 }, { "epoch": 0.02, "learning_rate": 4.993081589672134e-05, "loss": 0.8995, "step": 23784 }, { "epoch": 0.02, "learning_rate": 4.9930810057029285e-05, "loss": 1.4999, "step": 23785 }, { "epoch": 0.02, "learning_rate": 4.993080421709114e-05, "loss": 1.0827, "step": 23786 }, { "epoch": 0.02, "learning_rate": 4.993079837690689e-05, "loss": 1.0451, "step": 23787 }, { "epoch": 0.02, "learning_rate": 4.9930792536476514e-05, "loss": 1.1639, "step": 23788 }, { "epoch": 0.02, "learning_rate": 4.993078669580005e-05, "loss": 1.0526, "step": 23789 }, { "epoch": 0.02, "learning_rate": 4.993078085487747e-05, "loss": 1.1766, "step": 23790 }, { "epoch": 0.02, "learning_rate": 4.9930775013708786e-05, "loss": 1.1207, "step": 23791 }, { "epoch": 0.02, "learning_rate": 4.9930769172294e-05, "loss": 1.5984, "step": 23792 }, { "epoch": 0.02, "learning_rate": 4.99307633306331e-05, "loss": 1.1601, "step": 23793 }, { "epoch": 0.02, "learning_rate": 4.99307574887261e-05, "loss": 0.8567, "step": 23794 }, { "epoch": 0.02, "learning_rate": 4.993075164657299e-05, "loss": 1.1053, "step": 23795 }, { "epoch": 0.02, "learning_rate": 4.993074580417378e-05, "loss": 1.0403, "step": 23796 }, { "epoch": 0.02, "learning_rate": 4.9930739961528464e-05, "loss": 0.9976, "step": 23797 }, { "epoch": 0.02, "learning_rate": 4.993073411863703e-05, "loss": 1.0831, "step": 23798 }, { "epoch": 0.02, "learning_rate": 4.993072827549951e-05, "loss": 1.1232, "step": 23799 }, { "epoch": 0.02, "learning_rate": 4.993072243211587e-05, "loss": 0.9499, "step": 23800 }, { "epoch": 0.02, "learning_rate": 4.9930716588486125e-05, "loss": 1.0696, "step": 23801 }, { "epoch": 0.02, "learning_rate": 4.993071074461028e-05, "loss": 1.1445, "step": 23802 }, { "epoch": 0.02, "learning_rate": 4.9930704900488326e-05, "loss": 1.1471, "step": 23803 }, { "epoch": 0.02, "learning_rate": 4.993069905612027e-05, "loss": 1.2165, "step": 23804 }, { "epoch": 0.02, "learning_rate": 4.9930693211506106e-05, "loss": 0.9736, "step": 23805 }, { "epoch": 0.02, "learning_rate": 4.993068736664584e-05, "loss": 1.0538, "step": 23806 }, { "epoch": 0.02, "learning_rate": 4.993068152153947e-05, "loss": 0.9501, "step": 23807 }, { "epoch": 0.02, "learning_rate": 4.993067567618699e-05, "loss": 0.2198, "step": 23808 }, { "epoch": 0.02, "learning_rate": 4.993066983058841e-05, "loss": 0.9532, "step": 23809 }, { "epoch": 0.02, "learning_rate": 4.993066398474371e-05, "loss": 1.0315, "step": 23810 }, { "epoch": 0.02, "learning_rate": 4.993065813865292e-05, "loss": 0.9177, "step": 23811 }, { "epoch": 0.02, "learning_rate": 4.993065229231602e-05, "loss": 0.8108, "step": 23812 }, { "epoch": 0.02, "learning_rate": 4.993064644573302e-05, "loss": 1.2054, "step": 23813 }, { "epoch": 0.02, "learning_rate": 4.993064059890391e-05, "loss": 1.0114, "step": 23814 }, { "epoch": 0.02, "learning_rate": 4.99306347518287e-05, "loss": 1.3438, "step": 23815 }, { "epoch": 0.02, "learning_rate": 4.9930628904507374e-05, "loss": 1.1394, "step": 23816 }, { "epoch": 0.02, "learning_rate": 4.993062305693995e-05, "loss": 1.4066, "step": 23817 }, { "epoch": 0.02, "learning_rate": 4.993061720912642e-05, "loss": 1.3887, "step": 23818 }, { "epoch": 0.02, "learning_rate": 4.993061136106679e-05, "loss": 0.258, "step": 23819 }, { "epoch": 0.02, "learning_rate": 4.993060551276105e-05, "loss": 0.7334, "step": 23820 }, { "epoch": 0.02, "learning_rate": 4.993059966420922e-05, "loss": 1.4757, "step": 23821 }, { "epoch": 0.02, "learning_rate": 4.993059381541127e-05, "loss": 1.0754, "step": 23822 }, { "epoch": 0.02, "learning_rate": 4.993058796636722e-05, "loss": 0.6737, "step": 23823 }, { "epoch": 0.02, "learning_rate": 4.993058211707706e-05, "loss": 0.7908, "step": 23824 }, { "epoch": 0.02, "learning_rate": 4.9930576267540805e-05, "loss": 1.0169, "step": 23825 }, { "epoch": 0.02, "learning_rate": 4.993057041775845e-05, "loss": 1.0715, "step": 23826 }, { "epoch": 0.02, "learning_rate": 4.993056456772997e-05, "loss": 0.8269, "step": 23827 }, { "epoch": 0.02, "learning_rate": 4.993055871745541e-05, "loss": 0.4865, "step": 23828 }, { "epoch": 0.02, "learning_rate": 4.993055286693473e-05, "loss": 0.4388, "step": 23829 }, { "epoch": 0.02, "learning_rate": 4.9930547016167956e-05, "loss": 0.4696, "step": 23830 }, { "epoch": 0.02, "learning_rate": 4.993054116515508e-05, "loss": 0.4639, "step": 23831 }, { "epoch": 0.02, "learning_rate": 4.993053531389609e-05, "loss": 0.3263, "step": 23832 }, { "epoch": 0.02, "learning_rate": 4.9930529462390995e-05, "loss": 0.5169, "step": 23833 }, { "epoch": 0.02, "learning_rate": 4.99305236106398e-05, "loss": 0.4791, "step": 23834 }, { "epoch": 0.02, "learning_rate": 4.9930517758642503e-05, "loss": 1.4377, "step": 23835 }, { "epoch": 0.02, "learning_rate": 4.9930511906399104e-05, "loss": 1.2656, "step": 23836 }, { "epoch": 0.02, "learning_rate": 4.99305060539096e-05, "loss": 1.0563, "step": 23837 }, { "epoch": 0.02, "learning_rate": 4.993050020117399e-05, "loss": 0.92, "step": 23838 }, { "epoch": 0.02, "learning_rate": 4.993049434819228e-05, "loss": 1.0601, "step": 23839 }, { "epoch": 0.02, "learning_rate": 4.993048849496447e-05, "loss": 0.8547, "step": 23840 }, { "epoch": 0.02, "learning_rate": 4.993048264149055e-05, "loss": 1.0293, "step": 23841 }, { "epoch": 0.02, "learning_rate": 4.993047678777053e-05, "loss": 1.0402, "step": 23842 }, { "epoch": 0.02, "learning_rate": 4.993047093380441e-05, "loss": 1.0667, "step": 23843 }, { "epoch": 0.02, "learning_rate": 4.993046507959218e-05, "loss": 1.0693, "step": 23844 }, { "epoch": 0.02, "learning_rate": 4.993045922513385e-05, "loss": 1.0883, "step": 23845 }, { "epoch": 0.02, "learning_rate": 4.993045337042942e-05, "loss": 0.7582, "step": 23846 }, { "epoch": 0.02, "learning_rate": 4.9930447515478875e-05, "loss": 1.0995, "step": 23847 }, { "epoch": 0.02, "learning_rate": 4.993044166028224e-05, "loss": 1.0469, "step": 23848 }, { "epoch": 0.02, "learning_rate": 4.99304358048395e-05, "loss": 1.0867, "step": 23849 }, { "epoch": 0.02, "learning_rate": 4.9930429949150655e-05, "loss": 0.9903, "step": 23850 }, { "epoch": 0.02, "learning_rate": 4.993042409321571e-05, "loss": 1.0323, "step": 23851 }, { "epoch": 0.02, "learning_rate": 4.993041823703466e-05, "loss": 1.0535, "step": 23852 }, { "epoch": 0.02, "learning_rate": 4.9930412380607504e-05, "loss": 1.1562, "step": 23853 }, { "epoch": 0.02, "learning_rate": 4.993040652393425e-05, "loss": 0.9531, "step": 23854 }, { "epoch": 0.02, "learning_rate": 4.993040066701489e-05, "loss": 0.8346, "step": 23855 }, { "epoch": 0.02, "learning_rate": 4.993039480984944e-05, "loss": 1.071, "step": 23856 }, { "epoch": 0.02, "learning_rate": 4.993038895243787e-05, "loss": 0.9997, "step": 23857 }, { "epoch": 0.02, "learning_rate": 4.993038309478021e-05, "loss": 1.2292, "step": 23858 }, { "epoch": 0.02, "learning_rate": 4.993037723687643e-05, "loss": 1.0845, "step": 23859 }, { "epoch": 0.02, "learning_rate": 4.993037137872657e-05, "loss": 1.163, "step": 23860 }, { "epoch": 0.02, "learning_rate": 4.99303655203306e-05, "loss": 1.0931, "step": 23861 }, { "epoch": 0.02, "learning_rate": 4.9930359661688525e-05, "loss": 1.0741, "step": 23862 }, { "epoch": 0.02, "learning_rate": 4.993035380280035e-05, "loss": 1.1054, "step": 23863 }, { "epoch": 0.02, "learning_rate": 4.993034794366607e-05, "loss": 1.2512, "step": 23864 }, { "epoch": 0.02, "learning_rate": 4.9930342084285694e-05, "loss": 0.9502, "step": 23865 }, { "epoch": 0.02, "learning_rate": 4.9930336224659217e-05, "loss": 1.3488, "step": 23866 }, { "epoch": 0.02, "learning_rate": 4.993033036478663e-05, "loss": 1.2018, "step": 23867 }, { "epoch": 0.02, "learning_rate": 4.993032450466795e-05, "loss": 0.9976, "step": 23868 }, { "epoch": 0.02, "learning_rate": 4.993031864430316e-05, "loss": 1.1617, "step": 23869 }, { "epoch": 0.02, "learning_rate": 4.9930312783692276e-05, "loss": 0.9424, "step": 23870 }, { "epoch": 0.02, "learning_rate": 4.993030692283528e-05, "loss": 1.0249, "step": 23871 }, { "epoch": 0.02, "learning_rate": 4.9930301061732196e-05, "loss": 1.0682, "step": 23872 }, { "epoch": 0.02, "learning_rate": 4.9930295200383e-05, "loss": 1.039, "step": 23873 }, { "epoch": 0.02, "learning_rate": 4.993028933878771e-05, "loss": 1.2089, "step": 23874 }, { "epoch": 0.02, "learning_rate": 4.9930283476946315e-05, "loss": 0.9747, "step": 23875 }, { "epoch": 0.02, "learning_rate": 4.993027761485882e-05, "loss": 1.0975, "step": 23876 }, { "epoch": 0.02, "learning_rate": 4.993027175252522e-05, "loss": 0.8476, "step": 23877 }, { "epoch": 0.02, "learning_rate": 4.993026588994553e-05, "loss": 0.937, "step": 23878 }, { "epoch": 0.02, "learning_rate": 4.993026002711972e-05, "loss": 0.8607, "step": 23879 }, { "epoch": 0.02, "learning_rate": 4.9930254164047825e-05, "loss": 0.8889, "step": 23880 }, { "epoch": 0.02, "learning_rate": 4.993024830072982e-05, "loss": 1.0154, "step": 23881 }, { "epoch": 0.02, "learning_rate": 4.993024243716572e-05, "loss": 1.1148, "step": 23882 }, { "epoch": 0.02, "learning_rate": 4.9930236573355515e-05, "loss": 0.9134, "step": 23883 }, { "epoch": 0.02, "learning_rate": 4.993023070929921e-05, "loss": 1.1565, "step": 23884 }, { "epoch": 0.02, "learning_rate": 4.9930224844996806e-05, "loss": 1.2463, "step": 23885 }, { "epoch": 0.02, "learning_rate": 4.993021898044831e-05, "loss": 0.96, "step": 23886 }, { "epoch": 0.02, "learning_rate": 4.99302131156537e-05, "loss": 0.6576, "step": 23887 }, { "epoch": 0.02, "learning_rate": 4.993020725061299e-05, "loss": 1.0451, "step": 23888 }, { "epoch": 0.02, "learning_rate": 4.993020138532618e-05, "loss": 1.1067, "step": 23889 }, { "epoch": 0.02, "learning_rate": 4.993019551979328e-05, "loss": 1.0875, "step": 23890 }, { "epoch": 0.02, "learning_rate": 4.993018965401427e-05, "loss": 1.3352, "step": 23891 }, { "epoch": 0.02, "learning_rate": 4.993018378798916e-05, "loss": 0.8975, "step": 23892 }, { "epoch": 0.02, "learning_rate": 4.993017792171795e-05, "loss": 1.0853, "step": 23893 }, { "epoch": 0.02, "learning_rate": 4.9930172055200644e-05, "loss": 0.9325, "step": 23894 }, { "epoch": 0.02, "learning_rate": 4.993016618843723e-05, "loss": 0.9883, "step": 23895 }, { "epoch": 0.02, "learning_rate": 4.993016032142773e-05, "loss": 0.836, "step": 23896 }, { "epoch": 0.02, "learning_rate": 4.993015445417212e-05, "loss": 0.9883, "step": 23897 }, { "epoch": 0.02, "learning_rate": 4.993014858667041e-05, "loss": 1.0446, "step": 23898 }, { "epoch": 0.02, "learning_rate": 4.99301427189226e-05, "loss": 0.7852, "step": 23899 }, { "epoch": 0.02, "learning_rate": 4.9930136850928686e-05, "loss": 1.119, "step": 23900 }, { "epoch": 0.02, "learning_rate": 4.993013098268868e-05, "loss": 1.0756, "step": 23901 }, { "epoch": 0.02, "learning_rate": 4.993012511420257e-05, "loss": 1.1287, "step": 23902 }, { "epoch": 0.02, "learning_rate": 4.993011924547036e-05, "loss": 1.2119, "step": 23903 }, { "epoch": 0.02, "learning_rate": 4.993011337649206e-05, "loss": 1.1498, "step": 23904 }, { "epoch": 0.02, "learning_rate": 4.993010750726765e-05, "loss": 1.1328, "step": 23905 }, { "epoch": 0.02, "learning_rate": 4.993010163779714e-05, "loss": 1.2905, "step": 23906 }, { "epoch": 0.02, "learning_rate": 4.993009576808054e-05, "loss": 1.1138, "step": 23907 }, { "epoch": 0.02, "learning_rate": 4.993008989811784e-05, "loss": 0.9635, "step": 23908 }, { "epoch": 0.02, "learning_rate": 4.993008402790903e-05, "loss": 1.0723, "step": 23909 }, { "epoch": 0.02, "learning_rate": 4.9930078157454126e-05, "loss": 1.093, "step": 23910 }, { "epoch": 0.02, "learning_rate": 4.993007228675312e-05, "loss": 1.1435, "step": 23911 }, { "epoch": 0.02, "learning_rate": 4.993006641580602e-05, "loss": 0.9883, "step": 23912 }, { "epoch": 0.02, "learning_rate": 4.993006054461282e-05, "loss": 1.2072, "step": 23913 }, { "epoch": 0.02, "learning_rate": 4.993005467317352e-05, "loss": 1.1739, "step": 23914 }, { "epoch": 0.02, "learning_rate": 4.993004880148812e-05, "loss": 1.0008, "step": 23915 }, { "epoch": 0.02, "learning_rate": 4.993004292955662e-05, "loss": 0.9375, "step": 23916 }, { "epoch": 0.02, "learning_rate": 4.993003705737902e-05, "loss": 0.8533, "step": 23917 }, { "epoch": 0.02, "learning_rate": 4.9930031184955325e-05, "loss": 1.1426, "step": 23918 }, { "epoch": 0.02, "learning_rate": 4.9930025312285537e-05, "loss": 1.0799, "step": 23919 }, { "epoch": 0.02, "learning_rate": 4.9930019439369634e-05, "loss": 0.9404, "step": 23920 }, { "epoch": 0.02, "learning_rate": 4.9930013566207644e-05, "loss": 1.2811, "step": 23921 }, { "epoch": 0.02, "learning_rate": 4.9930007692799554e-05, "loss": 1.1519, "step": 23922 }, { "epoch": 0.02, "learning_rate": 4.9930001819145364e-05, "loss": 1.3707, "step": 23923 }, { "epoch": 0.02, "learning_rate": 4.992999594524508e-05, "loss": 1.3059, "step": 23924 }, { "epoch": 0.02, "learning_rate": 4.992999007109869e-05, "loss": 1.1786, "step": 23925 }, { "epoch": 0.02, "learning_rate": 4.9929984196706205e-05, "loss": 1.7966, "step": 23926 }, { "epoch": 0.02, "learning_rate": 4.992997832206762e-05, "loss": 0.9254, "step": 23927 }, { "epoch": 0.02, "learning_rate": 4.9929972447182935e-05, "loss": 0.8747, "step": 23928 }, { "epoch": 0.02, "learning_rate": 4.992996657205216e-05, "loss": 1.0476, "step": 23929 }, { "epoch": 0.02, "learning_rate": 4.9929960696675284e-05, "loss": 0.7788, "step": 23930 }, { "epoch": 0.02, "learning_rate": 4.9929954821052304e-05, "loss": 0.8293, "step": 23931 }, { "epoch": 0.02, "learning_rate": 4.992994894518324e-05, "loss": 0.5478, "step": 23932 }, { "epoch": 0.02, "learning_rate": 4.9929943069068065e-05, "loss": 0.456, "step": 23933 }, { "epoch": 0.02, "learning_rate": 4.992993719270679e-05, "loss": 0.4008, "step": 23934 }, { "epoch": 0.02, "learning_rate": 4.9929931316099424e-05, "loss": 1.0019, "step": 23935 }, { "epoch": 0.02, "learning_rate": 4.992992543924596e-05, "loss": 0.9696, "step": 23936 }, { "epoch": 0.02, "learning_rate": 4.99299195621464e-05, "loss": 1.1781, "step": 23937 }, { "epoch": 0.02, "learning_rate": 4.9929913684800734e-05, "loss": 1.1525, "step": 23938 }, { "epoch": 0.02, "learning_rate": 4.992990780720898e-05, "loss": 1.1141, "step": 23939 }, { "epoch": 0.02, "learning_rate": 4.992990192937112e-05, "loss": 1.0472, "step": 23940 }, { "epoch": 0.02, "learning_rate": 4.9929896051287174e-05, "loss": 0.9839, "step": 23941 }, { "epoch": 0.02, "learning_rate": 4.992989017295712e-05, "loss": 0.909, "step": 23942 }, { "epoch": 0.02, "learning_rate": 4.992988429438097e-05, "loss": 1.163, "step": 23943 }, { "epoch": 0.02, "learning_rate": 4.9929878415558725e-05, "loss": 1.2155, "step": 23944 }, { "epoch": 0.02, "learning_rate": 4.992987253649038e-05, "loss": 0.8846, "step": 23945 }, { "epoch": 0.02, "learning_rate": 4.9929866657175945e-05, "loss": 0.9096, "step": 23946 }, { "epoch": 0.02, "learning_rate": 4.9929860777615414e-05, "loss": 1.2423, "step": 23947 }, { "epoch": 0.02, "learning_rate": 4.992985489780877e-05, "loss": 1.0584, "step": 23948 }, { "epoch": 0.02, "learning_rate": 4.9929849017756045e-05, "loss": 1.1496, "step": 23949 }, { "epoch": 0.02, "learning_rate": 4.992984313745721e-05, "loss": 1.3065, "step": 23950 }, { "epoch": 0.02, "learning_rate": 4.9929837256912295e-05, "loss": 1.0829, "step": 23951 }, { "epoch": 0.02, "learning_rate": 4.992983137612127e-05, "loss": 1.1409, "step": 23952 }, { "epoch": 0.02, "learning_rate": 4.992982549508415e-05, "loss": 1.1532, "step": 23953 }, { "epoch": 0.02, "learning_rate": 4.992981961380093e-05, "loss": 1.1388, "step": 23954 }, { "epoch": 0.02, "learning_rate": 4.9929813732271624e-05, "loss": 1.2693, "step": 23955 }, { "epoch": 0.02, "learning_rate": 4.992980785049621e-05, "loss": 1.2789, "step": 23956 }, { "epoch": 0.02, "learning_rate": 4.992980196847471e-05, "loss": 1.1358, "step": 23957 }, { "epoch": 0.02, "learning_rate": 4.992979608620711e-05, "loss": 0.7598, "step": 23958 }, { "epoch": 0.02, "learning_rate": 4.992979020369341e-05, "loss": 0.8166, "step": 23959 }, { "epoch": 0.02, "learning_rate": 4.9929784320933615e-05, "loss": 1.2738, "step": 23960 }, { "epoch": 0.02, "learning_rate": 4.992977843792772e-05, "loss": 1.0608, "step": 23961 }, { "epoch": 0.02, "learning_rate": 4.992977255467574e-05, "loss": 0.8491, "step": 23962 }, { "epoch": 0.02, "learning_rate": 4.992976667117766e-05, "loss": 1.1542, "step": 23963 }, { "epoch": 0.02, "learning_rate": 4.9929760787433474e-05, "loss": 0.9427, "step": 23964 }, { "epoch": 0.02, "learning_rate": 4.99297549034432e-05, "loss": 0.9826, "step": 23965 }, { "epoch": 0.02, "learning_rate": 4.992974901920683e-05, "loss": 0.7197, "step": 23966 }, { "epoch": 0.02, "learning_rate": 4.992974313472436e-05, "loss": 1.262, "step": 23967 }, { "epoch": 0.02, "learning_rate": 4.99297372499958e-05, "loss": 0.9121, "step": 23968 }, { "epoch": 0.02, "learning_rate": 4.992973136502114e-05, "loss": 1.2549, "step": 23969 }, { "epoch": 0.02, "learning_rate": 4.992972547980038e-05, "loss": 1.747, "step": 23970 }, { "epoch": 0.02, "learning_rate": 4.992971959433354e-05, "loss": 0.9309, "step": 23971 }, { "epoch": 0.02, "learning_rate": 4.992971370862059e-05, "loss": 1.2747, "step": 23972 }, { "epoch": 0.02, "learning_rate": 4.992970782266154e-05, "loss": 1.0841, "step": 23973 }, { "epoch": 0.02, "learning_rate": 4.992970193645641e-05, "loss": 1.1291, "step": 23974 }, { "epoch": 0.02, "learning_rate": 4.992969605000517e-05, "loss": 0.9889, "step": 23975 }, { "epoch": 0.02, "learning_rate": 4.992969016330785e-05, "loss": 1.364, "step": 23976 }, { "epoch": 0.02, "learning_rate": 4.992968427636442e-05, "loss": 1.2789, "step": 23977 }, { "epoch": 0.02, "learning_rate": 4.99296783891749e-05, "loss": 1.0603, "step": 23978 }, { "epoch": 0.02, "learning_rate": 4.992967250173929e-05, "loss": 1.2268, "step": 23979 }, { "epoch": 0.02, "learning_rate": 4.992966661405758e-05, "loss": 1.4013, "step": 23980 }, { "epoch": 0.02, "learning_rate": 4.9929660726129776e-05, "loss": 1.2344, "step": 23981 }, { "epoch": 0.02, "learning_rate": 4.9929654837955876e-05, "loss": 0.8566, "step": 23982 }, { "epoch": 0.02, "learning_rate": 4.9929648949535875e-05, "loss": 1.393, "step": 23983 }, { "epoch": 0.02, "learning_rate": 4.992964306086979e-05, "loss": 1.4354, "step": 23984 }, { "epoch": 0.02, "learning_rate": 4.99296371719576e-05, "loss": 1.0701, "step": 23985 }, { "epoch": 0.02, "learning_rate": 4.992963128279932e-05, "loss": 1.0008, "step": 23986 }, { "epoch": 0.02, "learning_rate": 4.9929625393394944e-05, "loss": 1.2444, "step": 23987 }, { "epoch": 0.02, "learning_rate": 4.9929619503744476e-05, "loss": 1.2604, "step": 23988 }, { "epoch": 0.02, "learning_rate": 4.9929613613847914e-05, "loss": 1.4118, "step": 23989 }, { "epoch": 0.02, "learning_rate": 4.992960772370525e-05, "loss": 1.4226, "step": 23990 }, { "epoch": 0.02, "learning_rate": 4.9929601833316495e-05, "loss": 0.8521, "step": 23991 }, { "epoch": 0.02, "learning_rate": 4.992959594268165e-05, "loss": 0.9982, "step": 23992 }, { "epoch": 0.02, "learning_rate": 4.992959005180071e-05, "loss": 0.9278, "step": 23993 }, { "epoch": 0.02, "learning_rate": 4.992958416067367e-05, "loss": 0.7392, "step": 23994 }, { "epoch": 0.02, "learning_rate": 4.992957826930054e-05, "loss": 1.1162, "step": 23995 }, { "epoch": 0.02, "learning_rate": 4.992957237768132e-05, "loss": 0.9021, "step": 23996 }, { "epoch": 0.02, "learning_rate": 4.9929566485815995e-05, "loss": 0.9197, "step": 23997 }, { "epoch": 0.02, "learning_rate": 4.9929560593704584e-05, "loss": 1.2618, "step": 23998 }, { "epoch": 0.02, "learning_rate": 4.992955470134707e-05, "loss": 1.0041, "step": 23999 }, { "epoch": 0.02, "learning_rate": 4.9929548808743475e-05, "loss": 1.0374, "step": 24000 }, { "epoch": 0.02, "eval_loss": 1.0256379842758179, "eval_runtime": 86.2144, "eval_samples_per_second": 16.065, "eval_steps_per_second": 4.025, "step": 24000 }, { "epoch": 0.02, "learning_rate": 4.992954291589378e-05, "loss": 0.8192, "step": 24001 }, { "epoch": 0.02, "learning_rate": 4.992953702279799e-05, "loss": 0.9872, "step": 24002 }, { "epoch": 0.02, "learning_rate": 4.9929531129456106e-05, "loss": 1.1955, "step": 24003 }, { "epoch": 0.02, "learning_rate": 4.992952523586813e-05, "loss": 1.1084, "step": 24004 }, { "epoch": 0.02, "learning_rate": 4.9929519342034054e-05, "loss": 1.1383, "step": 24005 }, { "epoch": 0.02, "learning_rate": 4.992951344795389e-05, "loss": 1.1065, "step": 24006 }, { "epoch": 0.02, "learning_rate": 4.992950755362763e-05, "loss": 0.8596, "step": 24007 }, { "epoch": 0.02, "learning_rate": 4.9929501659055275e-05, "loss": 1.324, "step": 24008 }, { "epoch": 0.02, "learning_rate": 4.9929495764236835e-05, "loss": 1.0322, "step": 24009 }, { "epoch": 0.02, "learning_rate": 4.992948986917229e-05, "loss": 1.119, "step": 24010 }, { "epoch": 0.02, "learning_rate": 4.9929483973861654e-05, "loss": 1.2774, "step": 24011 }, { "epoch": 0.02, "learning_rate": 4.992947807830493e-05, "loss": 1.9411, "step": 24012 }, { "epoch": 0.02, "learning_rate": 4.992947218250211e-05, "loss": 0.8934, "step": 24013 }, { "epoch": 0.02, "learning_rate": 4.99294662864532e-05, "loss": 0.8308, "step": 24014 }, { "epoch": 0.02, "learning_rate": 4.9929460390158196e-05, "loss": 1.1224, "step": 24015 }, { "epoch": 0.02, "learning_rate": 4.9929454493617094e-05, "loss": 0.81, "step": 24016 }, { "epoch": 0.02, "learning_rate": 4.9929448596829906e-05, "loss": 0.7787, "step": 24017 }, { "epoch": 0.02, "learning_rate": 4.992944269979662e-05, "loss": 0.873, "step": 24018 }, { "epoch": 0.02, "learning_rate": 4.992943680251724e-05, "loss": 1.032, "step": 24019 }, { "epoch": 0.02, "learning_rate": 4.992943090499177e-05, "loss": 1.22, "step": 24020 }, { "epoch": 0.02, "learning_rate": 4.99294250072202e-05, "loss": 0.8855, "step": 24021 }, { "epoch": 0.02, "learning_rate": 4.9929419109202546e-05, "loss": 1.2705, "step": 24022 }, { "epoch": 0.02, "learning_rate": 4.99294132109388e-05, "loss": 0.8653, "step": 24023 }, { "epoch": 0.02, "learning_rate": 4.992940731242895e-05, "loss": 1.1048, "step": 24024 }, { "epoch": 0.02, "learning_rate": 4.992940141367302e-05, "loss": 0.7766, "step": 24025 }, { "epoch": 0.02, "learning_rate": 4.9929395514671e-05, "loss": 1.1276, "step": 24026 }, { "epoch": 0.02, "learning_rate": 4.9929389615422874e-05, "loss": 1.0022, "step": 24027 }, { "epoch": 0.02, "learning_rate": 4.992938371592866e-05, "loss": 1.1292, "step": 24028 }, { "epoch": 0.02, "learning_rate": 4.992937781618836e-05, "loss": 1.5412, "step": 24029 }, { "epoch": 0.02, "learning_rate": 4.992937191620196e-05, "loss": 1.6076, "step": 24030 }, { "epoch": 0.02, "learning_rate": 4.992936601596947e-05, "loss": 1.5389, "step": 24031 }, { "epoch": 0.02, "learning_rate": 4.992936011549089e-05, "loss": 1.1828, "step": 24032 }, { "epoch": 0.02, "learning_rate": 4.9929354214766225e-05, "loss": 1.3197, "step": 24033 }, { "epoch": 0.02, "learning_rate": 4.992934831379545e-05, "loss": 1.143, "step": 24034 }, { "epoch": 0.02, "learning_rate": 4.992934241257859e-05, "loss": 1.0148, "step": 24035 }, { "epoch": 0.02, "learning_rate": 4.992933651111564e-05, "loss": 1.2204, "step": 24036 }, { "epoch": 0.02, "learning_rate": 4.992933060940661e-05, "loss": 0.9826, "step": 24037 }, { "epoch": 0.02, "learning_rate": 4.9929324707451474e-05, "loss": 1.2265, "step": 24038 }, { "epoch": 0.02, "learning_rate": 4.992931880525025e-05, "loss": 1.2869, "step": 24039 }, { "epoch": 0.02, "learning_rate": 4.992931290280293e-05, "loss": 1.0548, "step": 24040 }, { "epoch": 0.02, "learning_rate": 4.992930700010952e-05, "loss": 1.218, "step": 24041 }, { "epoch": 0.02, "learning_rate": 4.992930109717002e-05, "loss": 1.1817, "step": 24042 }, { "epoch": 0.02, "learning_rate": 4.9929295193984426e-05, "loss": 1.2657, "step": 24043 }, { "epoch": 0.02, "learning_rate": 4.992928929055275e-05, "loss": 1.2041, "step": 24044 }, { "epoch": 0.02, "learning_rate": 4.992928338687497e-05, "loss": 1.2043, "step": 24045 }, { "epoch": 0.02, "learning_rate": 4.992927748295111e-05, "loss": 1.1938, "step": 24046 }, { "epoch": 0.02, "learning_rate": 4.992927157878115e-05, "loss": 1.1856, "step": 24047 }, { "epoch": 0.02, "learning_rate": 4.99292656743651e-05, "loss": 1.2109, "step": 24048 }, { "epoch": 0.02, "learning_rate": 4.9929259769702964e-05, "loss": 1.4686, "step": 24049 }, { "epoch": 0.02, "learning_rate": 4.9929253864794735e-05, "loss": 1.1906, "step": 24050 }, { "epoch": 0.02, "learning_rate": 4.992924795964041e-05, "loss": 1.1244, "step": 24051 }, { "epoch": 0.02, "learning_rate": 4.9929242054239997e-05, "loss": 1.1775, "step": 24052 }, { "epoch": 0.02, "learning_rate": 4.9929236148593494e-05, "loss": 0.904, "step": 24053 }, { "epoch": 0.02, "learning_rate": 4.99292302427009e-05, "loss": 1.0554, "step": 24054 }, { "epoch": 0.02, "learning_rate": 4.992922433656221e-05, "loss": 1.0955, "step": 24055 }, { "epoch": 0.02, "learning_rate": 4.992921843017744e-05, "loss": 1.1677, "step": 24056 }, { "epoch": 0.02, "learning_rate": 4.992921252354657e-05, "loss": 1.0176, "step": 24057 }, { "epoch": 0.02, "learning_rate": 4.992920661666961e-05, "loss": 1.1891, "step": 24058 }, { "epoch": 0.02, "learning_rate": 4.992920070954657e-05, "loss": 1.0973, "step": 24059 }, { "epoch": 0.02, "learning_rate": 4.992919480217742e-05, "loss": 1.0989, "step": 24060 }, { "epoch": 0.02, "learning_rate": 4.99291888945622e-05, "loss": 1.1221, "step": 24061 }, { "epoch": 0.02, "learning_rate": 4.9929182986700874e-05, "loss": 1.2773, "step": 24062 }, { "epoch": 0.02, "learning_rate": 4.9929177078593456e-05, "loss": 1.1757, "step": 24063 }, { "epoch": 0.02, "learning_rate": 4.9929171170239965e-05, "loss": 1.1566, "step": 24064 }, { "epoch": 0.02, "learning_rate": 4.9929165261640374e-05, "loss": 1.4818, "step": 24065 }, { "epoch": 0.02, "learning_rate": 4.992915935279469e-05, "loss": 1.0018, "step": 24066 }, { "epoch": 0.02, "learning_rate": 4.9929153443702916e-05, "loss": 1.1401, "step": 24067 }, { "epoch": 0.02, "learning_rate": 4.992914753436506e-05, "loss": 0.9396, "step": 24068 }, { "epoch": 0.02, "learning_rate": 4.9929141624781105e-05, "loss": 1.2054, "step": 24069 }, { "epoch": 0.02, "learning_rate": 4.9929135714951066e-05, "loss": 1.0937, "step": 24070 }, { "epoch": 0.02, "learning_rate": 4.992912980487493e-05, "loss": 1.065, "step": 24071 }, { "epoch": 0.02, "learning_rate": 4.992912389455271e-05, "loss": 0.9308, "step": 24072 }, { "epoch": 0.02, "learning_rate": 4.9929117983984395e-05, "loss": 0.915, "step": 24073 }, { "epoch": 0.02, "learning_rate": 4.9929112073169995e-05, "loss": 1.0242, "step": 24074 }, { "epoch": 0.02, "learning_rate": 4.992910616210951e-05, "loss": 0.8787, "step": 24075 }, { "epoch": 0.02, "learning_rate": 4.992910025080293e-05, "loss": 0.8575, "step": 24076 }, { "epoch": 0.02, "learning_rate": 4.992909433925026e-05, "loss": 0.8885, "step": 24077 }, { "epoch": 0.02, "learning_rate": 4.99290884274515e-05, "loss": 0.8424, "step": 24078 }, { "epoch": 0.02, "learning_rate": 4.9929082515406647e-05, "loss": 0.9289, "step": 24079 }, { "epoch": 0.02, "learning_rate": 4.992907660311571e-05, "loss": 1.2562, "step": 24080 }, { "epoch": 0.02, "learning_rate": 4.9929070690578685e-05, "loss": 0.878, "step": 24081 }, { "epoch": 0.02, "learning_rate": 4.9929064777795564e-05, "loss": 1.2008, "step": 24082 }, { "epoch": 0.02, "learning_rate": 4.992905886476636e-05, "loss": 1.0722, "step": 24083 }, { "epoch": 0.02, "learning_rate": 4.992905295149107e-05, "loss": 0.955, "step": 24084 }, { "epoch": 0.02, "learning_rate": 4.992904703796968e-05, "loss": 1.2274, "step": 24085 }, { "epoch": 0.02, "learning_rate": 4.992904112420221e-05, "loss": 1.19, "step": 24086 }, { "epoch": 0.02, "learning_rate": 4.992903521018865e-05, "loss": 1.2143, "step": 24087 }, { "epoch": 0.02, "learning_rate": 4.9929029295929e-05, "loss": 1.1704, "step": 24088 }, { "epoch": 0.02, "learning_rate": 4.992902338142326e-05, "loss": 1.3001, "step": 24089 }, { "epoch": 0.02, "learning_rate": 4.992901746667142e-05, "loss": 1.0435, "step": 24090 }, { "epoch": 0.02, "learning_rate": 4.992901155167351e-05, "loss": 1.0097, "step": 24091 }, { "epoch": 0.02, "learning_rate": 4.9929005636429504e-05, "loss": 1.0777, "step": 24092 }, { "epoch": 0.02, "learning_rate": 4.992899972093941e-05, "loss": 1.2675, "step": 24093 }, { "epoch": 0.02, "learning_rate": 4.9928993805203225e-05, "loss": 1.0016, "step": 24094 }, { "epoch": 0.02, "learning_rate": 4.9928987889220955e-05, "loss": 0.8967, "step": 24095 }, { "epoch": 0.02, "learning_rate": 4.992898197299259e-05, "loss": 1.115, "step": 24096 }, { "epoch": 0.02, "learning_rate": 4.992897605651814e-05, "loss": 1.0985, "step": 24097 }, { "epoch": 0.02, "learning_rate": 4.9928970139797605e-05, "loss": 1.2262, "step": 24098 }, { "epoch": 0.02, "learning_rate": 4.9928964222830974e-05, "loss": 1.3129, "step": 24099 }, { "epoch": 0.02, "learning_rate": 4.992895830561827e-05, "loss": 0.806, "step": 24100 }, { "epoch": 0.02, "learning_rate": 4.9928952388159466e-05, "loss": 0.7639, "step": 24101 }, { "epoch": 0.02, "learning_rate": 4.9928946470454576e-05, "loss": 0.8922, "step": 24102 }, { "epoch": 0.02, "learning_rate": 4.99289405525036e-05, "loss": 0.793, "step": 24103 }, { "epoch": 0.02, "learning_rate": 4.992893463430653e-05, "loss": 0.7395, "step": 24104 }, { "epoch": 0.02, "learning_rate": 4.9928928715863375e-05, "loss": 0.6735, "step": 24105 }, { "epoch": 0.02, "learning_rate": 4.992892279717413e-05, "loss": 0.7087, "step": 24106 }, { "epoch": 0.02, "learning_rate": 4.9928916878238806e-05, "loss": 1.1764, "step": 24107 }, { "epoch": 0.02, "learning_rate": 4.992891095905739e-05, "loss": 1.1869, "step": 24108 }, { "epoch": 0.02, "learning_rate": 4.992890503962988e-05, "loss": 1.1303, "step": 24109 }, { "epoch": 0.02, "learning_rate": 4.9928899119956284e-05, "loss": 0.9253, "step": 24110 }, { "epoch": 0.02, "learning_rate": 4.9928893200036605e-05, "loss": 0.8383, "step": 24111 }, { "epoch": 0.02, "learning_rate": 4.992888727987084e-05, "loss": 1.2807, "step": 24112 }, { "epoch": 0.02, "learning_rate": 4.992888135945899e-05, "loss": 0.8428, "step": 24113 }, { "epoch": 0.02, "learning_rate": 4.9928875438801035e-05, "loss": 0.8443, "step": 24114 }, { "epoch": 0.02, "learning_rate": 4.9928869517897016e-05, "loss": 0.9686, "step": 24115 }, { "epoch": 0.02, "learning_rate": 4.9928863596746896e-05, "loss": 1.0426, "step": 24116 }, { "epoch": 0.02, "learning_rate": 4.992885767535069e-05, "loss": 1.0709, "step": 24117 }, { "epoch": 0.02, "learning_rate": 4.99288517537084e-05, "loss": 1.2419, "step": 24118 }, { "epoch": 0.02, "learning_rate": 4.992884583182002e-05, "loss": 0.9178, "step": 24119 }, { "epoch": 0.02, "learning_rate": 4.992883990968556e-05, "loss": 1.1018, "step": 24120 }, { "epoch": 0.02, "learning_rate": 4.9928833987305004e-05, "loss": 1.0398, "step": 24121 }, { "epoch": 0.02, "learning_rate": 4.992882806467836e-05, "loss": 1.147, "step": 24122 }, { "epoch": 0.02, "learning_rate": 4.992882214180563e-05, "loss": 1.5939, "step": 24123 }, { "epoch": 0.02, "learning_rate": 4.9928816218686816e-05, "loss": 1.562, "step": 24124 }, { "epoch": 0.02, "learning_rate": 4.9928810295321916e-05, "loss": 1.6745, "step": 24125 }, { "epoch": 0.02, "learning_rate": 4.992880437171093e-05, "loss": 1.4223, "step": 24126 }, { "epoch": 0.02, "learning_rate": 4.9928798447853855e-05, "loss": 1.651, "step": 24127 }, { "epoch": 0.02, "learning_rate": 4.9928792523750694e-05, "loss": 1.5524, "step": 24128 }, { "epoch": 0.02, "learning_rate": 4.9928786599401446e-05, "loss": 1.4367, "step": 24129 }, { "epoch": 0.02, "learning_rate": 4.992878067480611e-05, "loss": 1.455, "step": 24130 }, { "epoch": 0.02, "learning_rate": 4.992877474996469e-05, "loss": 1.5197, "step": 24131 }, { "epoch": 0.02, "learning_rate": 4.992876882487719e-05, "loss": 1.3464, "step": 24132 }, { "epoch": 0.02, "learning_rate": 4.992876289954359e-05, "loss": 1.1148, "step": 24133 }, { "epoch": 0.02, "learning_rate": 4.992875697396392e-05, "loss": 0.8889, "step": 24134 }, { "epoch": 0.02, "learning_rate": 4.992875104813815e-05, "loss": 0.9795, "step": 24135 }, { "epoch": 0.02, "learning_rate": 4.99287451220663e-05, "loss": 1.0111, "step": 24136 }, { "epoch": 0.02, "learning_rate": 4.992873919574837e-05, "loss": 1.0145, "step": 24137 }, { "epoch": 0.02, "learning_rate": 4.992873326918434e-05, "loss": 0.9554, "step": 24138 }, { "epoch": 0.02, "learning_rate": 4.992872734237422e-05, "loss": 1.0446, "step": 24139 }, { "epoch": 0.02, "learning_rate": 4.9928721415318034e-05, "loss": 1.2072, "step": 24140 }, { "epoch": 0.02, "learning_rate": 4.992871548801575e-05, "loss": 1.1016, "step": 24141 }, { "epoch": 0.02, "learning_rate": 4.992870956046739e-05, "loss": 1.3995, "step": 24142 }, { "epoch": 0.02, "learning_rate": 4.9928703632672934e-05, "loss": 1.0819, "step": 24143 }, { "epoch": 0.02, "learning_rate": 4.99286977046324e-05, "loss": 1.0095, "step": 24144 }, { "epoch": 0.02, "learning_rate": 4.992869177634577e-05, "loss": 0.9089, "step": 24145 }, { "epoch": 0.02, "learning_rate": 4.992868584781306e-05, "loss": 0.9564, "step": 24146 }, { "epoch": 0.02, "learning_rate": 4.992867991903427e-05, "loss": 0.9206, "step": 24147 }, { "epoch": 0.02, "learning_rate": 4.9928673990009386e-05, "loss": 0.8088, "step": 24148 }, { "epoch": 0.02, "learning_rate": 4.992866806073842e-05, "loss": 0.8272, "step": 24149 }, { "epoch": 0.02, "learning_rate": 4.992866213122137e-05, "loss": 1.0474, "step": 24150 }, { "epoch": 0.02, "learning_rate": 4.9928656201458236e-05, "loss": 1.2565, "step": 24151 }, { "epoch": 0.02, "learning_rate": 4.992865027144902e-05, "loss": 1.2902, "step": 24152 }, { "epoch": 0.02, "learning_rate": 4.992864434119371e-05, "loss": 1.393, "step": 24153 }, { "epoch": 0.02, "learning_rate": 4.992863841069232e-05, "loss": 0.9702, "step": 24154 }, { "epoch": 0.02, "learning_rate": 4.992863247994484e-05, "loss": 0.7488, "step": 24155 }, { "epoch": 0.02, "learning_rate": 4.992862654895128e-05, "loss": 0.5989, "step": 24156 }, { "epoch": 0.02, "learning_rate": 4.992862061771163e-05, "loss": 1.3384, "step": 24157 }, { "epoch": 0.02, "learning_rate": 4.9928614686225904e-05, "loss": 0.9962, "step": 24158 }, { "epoch": 0.02, "learning_rate": 4.9928608754494086e-05, "loss": 1.2563, "step": 24159 }, { "epoch": 0.02, "learning_rate": 4.992860282251618e-05, "loss": 1.0987, "step": 24160 }, { "epoch": 0.02, "learning_rate": 4.99285968902922e-05, "loss": 1.129, "step": 24161 }, { "epoch": 0.02, "learning_rate": 4.992859095782213e-05, "loss": 1.2527, "step": 24162 }, { "epoch": 0.02, "learning_rate": 4.9928585025105976e-05, "loss": 1.057, "step": 24163 }, { "epoch": 0.02, "learning_rate": 4.992857909214373e-05, "loss": 0.9024, "step": 24164 }, { "epoch": 0.02, "learning_rate": 4.992857315893541e-05, "loss": 0.7036, "step": 24165 }, { "epoch": 0.02, "learning_rate": 4.9928567225481e-05, "loss": 1.0307, "step": 24166 }, { "epoch": 0.02, "learning_rate": 4.992856129178052e-05, "loss": 1.0782, "step": 24167 }, { "epoch": 0.02, "learning_rate": 4.992855535783394e-05, "loss": 1.3696, "step": 24168 }, { "epoch": 0.02, "learning_rate": 4.992854942364128e-05, "loss": 1.2549, "step": 24169 }, { "epoch": 0.02, "learning_rate": 4.992854348920253e-05, "loss": 1.0177, "step": 24170 }, { "epoch": 0.02, "learning_rate": 4.9928537554517704e-05, "loss": 1.2334, "step": 24171 }, { "epoch": 0.02, "learning_rate": 4.992853161958679e-05, "loss": 1.3228, "step": 24172 }, { "epoch": 0.02, "learning_rate": 4.9928525684409794e-05, "loss": 0.9345, "step": 24173 }, { "epoch": 0.02, "learning_rate": 4.9928519748986715e-05, "loss": 1.0616, "step": 24174 }, { "epoch": 0.02, "learning_rate": 4.992851381331754e-05, "loss": 1.0415, "step": 24175 }, { "epoch": 0.02, "learning_rate": 4.99285078774023e-05, "loss": 1.1505, "step": 24176 }, { "epoch": 0.02, "learning_rate": 4.9928501941240966e-05, "loss": 1.1742, "step": 24177 }, { "epoch": 0.02, "learning_rate": 4.992849600483355e-05, "loss": 0.9986, "step": 24178 }, { "epoch": 0.02, "learning_rate": 4.9928490068180054e-05, "loss": 1.0851, "step": 24179 }, { "epoch": 0.02, "learning_rate": 4.9928484131280475e-05, "loss": 0.9781, "step": 24180 }, { "epoch": 0.02, "learning_rate": 4.992847819413481e-05, "loss": 1.1776, "step": 24181 }, { "epoch": 0.02, "learning_rate": 4.992847225674305e-05, "loss": 1.077, "step": 24182 }, { "epoch": 0.02, "learning_rate": 4.9928466319105224e-05, "loss": 0.9565, "step": 24183 }, { "epoch": 0.02, "learning_rate": 4.992846038122131e-05, "loss": 1.3119, "step": 24184 }, { "epoch": 0.02, "learning_rate": 4.992845444309131e-05, "loss": 1.1471, "step": 24185 }, { "epoch": 0.02, "learning_rate": 4.992844850471522e-05, "loss": 0.9968, "step": 24186 }, { "epoch": 0.02, "learning_rate": 4.992844256609306e-05, "loss": 1.1051, "step": 24187 }, { "epoch": 0.02, "learning_rate": 4.992843662722481e-05, "loss": 0.8831, "step": 24188 }, { "epoch": 0.02, "learning_rate": 4.9928430688110474e-05, "loss": 0.9841, "step": 24189 }, { "epoch": 0.02, "learning_rate": 4.992842474875006e-05, "loss": 0.9237, "step": 24190 }, { "epoch": 0.02, "learning_rate": 4.992841880914356e-05, "loss": 0.9932, "step": 24191 }, { "epoch": 0.02, "learning_rate": 4.992841286929099e-05, "loss": 0.8519, "step": 24192 }, { "epoch": 0.02, "learning_rate": 4.9928406929192315e-05, "loss": 1.1374, "step": 24193 }, { "epoch": 0.02, "learning_rate": 4.9928400988847575e-05, "loss": 1.153, "step": 24194 }, { "epoch": 0.02, "learning_rate": 4.992839504825674e-05, "loss": 1.1652, "step": 24195 }, { "epoch": 0.02, "learning_rate": 4.992838910741983e-05, "loss": 1.1358, "step": 24196 }, { "epoch": 0.02, "learning_rate": 4.9928383166336835e-05, "loss": 1.2323, "step": 24197 }, { "epoch": 0.02, "learning_rate": 4.992837722500776e-05, "loss": 1.0047, "step": 24198 }, { "epoch": 0.02, "learning_rate": 4.99283712834326e-05, "loss": 1.1758, "step": 24199 }, { "epoch": 0.02, "learning_rate": 4.992836534161136e-05, "loss": 1.0649, "step": 24200 }, { "epoch": 0.02, "learning_rate": 4.9928359399544034e-05, "loss": 0.5622, "step": 24201 }, { "epoch": 0.02, "learning_rate": 4.9928353457230634e-05, "loss": 1.3003, "step": 24202 }, { "epoch": 0.02, "learning_rate": 4.992834751467114e-05, "loss": 1.079, "step": 24203 }, { "epoch": 0.02, "learning_rate": 4.992834157186557e-05, "loss": 1.06, "step": 24204 }, { "epoch": 0.02, "learning_rate": 4.992833562881392e-05, "loss": 1.2988, "step": 24205 }, { "epoch": 0.02, "learning_rate": 4.9928329685516185e-05, "loss": 1.2342, "step": 24206 }, { "epoch": 0.02, "learning_rate": 4.992832374197237e-05, "loss": 1.203, "step": 24207 }, { "epoch": 0.02, "learning_rate": 4.9928317798182464e-05, "loss": 1.0136, "step": 24208 }, { "epoch": 0.02, "learning_rate": 4.992831185414649e-05, "loss": 1.1984, "step": 24209 }, { "epoch": 0.02, "learning_rate": 4.992830590986442e-05, "loss": 1.2357, "step": 24210 }, { "epoch": 0.02, "learning_rate": 4.992829996533628e-05, "loss": 1.238, "step": 24211 }, { "epoch": 0.02, "learning_rate": 4.992829402056206e-05, "loss": 1.4971, "step": 24212 }, { "epoch": 0.02, "learning_rate": 4.9928288075541754e-05, "loss": 0.8502, "step": 24213 }, { "epoch": 0.02, "learning_rate": 4.992828213027536e-05, "loss": 0.9654, "step": 24214 }, { "epoch": 0.02, "learning_rate": 4.992827618476289e-05, "loss": 0.9637, "step": 24215 }, { "epoch": 0.02, "learning_rate": 4.9928270239004345e-05, "loss": 1.0185, "step": 24216 }, { "epoch": 0.02, "learning_rate": 4.992826429299971e-05, "loss": 1.1469, "step": 24217 }, { "epoch": 0.02, "learning_rate": 4.9928258346748996e-05, "loss": 1.0067, "step": 24218 }, { "epoch": 0.02, "learning_rate": 4.9928252400252195e-05, "loss": 1.2665, "step": 24219 }, { "epoch": 0.02, "learning_rate": 4.992824645350932e-05, "loss": 1.1297, "step": 24220 }, { "epoch": 0.02, "learning_rate": 4.992824050652036e-05, "loss": 1.2041, "step": 24221 }, { "epoch": 0.02, "learning_rate": 4.9928234559285325e-05, "loss": 1.1025, "step": 24222 }, { "epoch": 0.02, "learning_rate": 4.9928228611804204e-05, "loss": 1.2413, "step": 24223 }, { "epoch": 0.02, "learning_rate": 4.992822266407701e-05, "loss": 0.9509, "step": 24224 }, { "epoch": 0.02, "learning_rate": 4.992821671610373e-05, "loss": 0.8809, "step": 24225 }, { "epoch": 0.02, "learning_rate": 4.992821076788437e-05, "loss": 1.0775, "step": 24226 }, { "epoch": 0.02, "learning_rate": 4.992820481941892e-05, "loss": 0.9786, "step": 24227 }, { "epoch": 0.02, "learning_rate": 4.992819887070739e-05, "loss": 1.08, "step": 24228 }, { "epoch": 0.02, "learning_rate": 4.992819292174979e-05, "loss": 1.2905, "step": 24229 }, { "epoch": 0.02, "learning_rate": 4.99281869725461e-05, "loss": 1.5677, "step": 24230 }, { "epoch": 0.02, "learning_rate": 4.992818102309634e-05, "loss": 1.0758, "step": 24231 }, { "epoch": 0.02, "learning_rate": 4.992817507340049e-05, "loss": 0.9225, "step": 24232 }, { "epoch": 0.02, "learning_rate": 4.9928169123458565e-05, "loss": 1.0924, "step": 24233 }, { "epoch": 0.02, "learning_rate": 4.992816317327056e-05, "loss": 1.1451, "step": 24234 }, { "epoch": 0.02, "learning_rate": 4.992815722283647e-05, "loss": 0.9253, "step": 24235 }, { "epoch": 0.02, "learning_rate": 4.99281512721563e-05, "loss": 1.2732, "step": 24236 }, { "epoch": 0.02, "learning_rate": 4.992814532123005e-05, "loss": 0.9871, "step": 24237 }, { "epoch": 0.02, "learning_rate": 4.9928139370057724e-05, "loss": 0.9317, "step": 24238 }, { "epoch": 0.02, "learning_rate": 4.9928133418639316e-05, "loss": 1.1775, "step": 24239 }, { "epoch": 0.02, "learning_rate": 4.992812746697483e-05, "loss": 1.0052, "step": 24240 }, { "epoch": 0.02, "learning_rate": 4.992812151506426e-05, "loss": 0.9033, "step": 24241 }, { "epoch": 0.02, "learning_rate": 4.9928115562907605e-05, "loss": 0.9505, "step": 24242 }, { "epoch": 0.02, "learning_rate": 4.9928109610504884e-05, "loss": 1.3743, "step": 24243 }, { "epoch": 0.02, "learning_rate": 4.9928103657856076e-05, "loss": 1.2599, "step": 24244 }, { "epoch": 0.02, "learning_rate": 4.992809770496119e-05, "loss": 1.2924, "step": 24245 }, { "epoch": 0.02, "learning_rate": 4.992809175182022e-05, "loss": 1.1646, "step": 24246 }, { "epoch": 0.02, "learning_rate": 4.992808579843317e-05, "loss": 1.143, "step": 24247 }, { "epoch": 0.02, "learning_rate": 4.992807984480005e-05, "loss": 1.3022, "step": 24248 }, { "epoch": 0.02, "learning_rate": 4.992807389092084e-05, "loss": 0.9846, "step": 24249 }, { "epoch": 0.02, "learning_rate": 4.992806793679555e-05, "loss": 0.7784, "step": 24250 }, { "epoch": 0.02, "learning_rate": 4.992806198242419e-05, "loss": 1.0458, "step": 24251 }, { "epoch": 0.02, "learning_rate": 4.9928056027806743e-05, "loss": 0.9615, "step": 24252 }, { "epoch": 0.02, "learning_rate": 4.992805007294322e-05, "loss": 1.2047, "step": 24253 }, { "epoch": 0.02, "learning_rate": 4.9928044117833614e-05, "loss": 1.1073, "step": 24254 }, { "epoch": 0.02, "learning_rate": 4.992803816247794e-05, "loss": 1.1555, "step": 24255 }, { "epoch": 0.02, "learning_rate": 4.992803220687618e-05, "loss": 0.9771, "step": 24256 }, { "epoch": 0.02, "learning_rate": 4.9928026251028336e-05, "loss": 1.0688, "step": 24257 }, { "epoch": 0.02, "learning_rate": 4.992802029493442e-05, "loss": 1.1013, "step": 24258 }, { "epoch": 0.02, "learning_rate": 4.992801433859442e-05, "loss": 0.9458, "step": 24259 }, { "epoch": 0.02, "learning_rate": 4.992800838200834e-05, "loss": 1.3607, "step": 24260 }, { "epoch": 0.02, "learning_rate": 4.992800242517619e-05, "loss": 1.7415, "step": 24261 }, { "epoch": 0.02, "learning_rate": 4.9927996468097955e-05, "loss": 1.5598, "step": 24262 }, { "epoch": 0.02, "learning_rate": 4.992799051077364e-05, "loss": 1.0536, "step": 24263 }, { "epoch": 0.02, "learning_rate": 4.992798455320325e-05, "loss": 1.0542, "step": 24264 }, { "epoch": 0.02, "learning_rate": 4.9927978595386784e-05, "loss": 1.1939, "step": 24265 }, { "epoch": 0.02, "learning_rate": 4.992797263732423e-05, "loss": 0.9606, "step": 24266 }, { "epoch": 0.02, "learning_rate": 4.99279666790156e-05, "loss": 1.0534, "step": 24267 }, { "epoch": 0.02, "learning_rate": 4.99279607204609e-05, "loss": 1.1883, "step": 24268 }, { "epoch": 0.02, "learning_rate": 4.992795476166011e-05, "loss": 0.8724, "step": 24269 }, { "epoch": 0.02, "learning_rate": 4.992794880261326e-05, "loss": 1.1025, "step": 24270 }, { "epoch": 0.02, "learning_rate": 4.9927942843320315e-05, "loss": 1.0987, "step": 24271 }, { "epoch": 0.02, "learning_rate": 4.9927936883781294e-05, "loss": 1.1441, "step": 24272 }, { "epoch": 0.02, "learning_rate": 4.99279309239962e-05, "loss": 1.1854, "step": 24273 }, { "epoch": 0.02, "learning_rate": 4.9927924963965025e-05, "loss": 1.282, "step": 24274 }, { "epoch": 0.02, "learning_rate": 4.992791900368777e-05, "loss": 1.107, "step": 24275 }, { "epoch": 0.02, "learning_rate": 4.992791304316444e-05, "loss": 0.934, "step": 24276 }, { "epoch": 0.02, "learning_rate": 4.9927907082395034e-05, "loss": 1.0459, "step": 24277 }, { "epoch": 0.02, "learning_rate": 4.992790112137955e-05, "loss": 1.2678, "step": 24278 }, { "epoch": 0.02, "learning_rate": 4.9927895160117985e-05, "loss": 0.8407, "step": 24279 }, { "epoch": 0.02, "learning_rate": 4.9927889198610344e-05, "loss": 0.9175, "step": 24280 }, { "epoch": 0.02, "learning_rate": 4.992788323685662e-05, "loss": 1.0273, "step": 24281 }, { "epoch": 0.02, "learning_rate": 4.992787727485683e-05, "loss": 1.0564, "step": 24282 }, { "epoch": 0.02, "learning_rate": 4.9927871312610954e-05, "loss": 0.1496, "step": 24283 }, { "epoch": 0.02, "learning_rate": 4.9927865350119e-05, "loss": 0.0947, "step": 24284 }, { "epoch": 0.02, "learning_rate": 4.992785938738097e-05, "loss": 0.0527, "step": 24285 }, { "epoch": 0.02, "learning_rate": 4.9927853424396864e-05, "loss": 0.0438, "step": 24286 }, { "epoch": 0.02, "learning_rate": 4.9927847461166683e-05, "loss": 0.0359, "step": 24287 }, { "epoch": 0.02, "learning_rate": 4.992784149769042e-05, "loss": 0.1177, "step": 24288 }, { "epoch": 0.02, "learning_rate": 4.992783553396808e-05, "loss": 0.3107, "step": 24289 }, { "epoch": 0.02, "learning_rate": 4.992782956999967e-05, "loss": 0.22, "step": 24290 }, { "epoch": 0.02, "learning_rate": 4.992782360578517e-05, "loss": 0.5223, "step": 24291 }, { "epoch": 0.02, "learning_rate": 4.9927817641324606e-05, "loss": 1.0032, "step": 24292 }, { "epoch": 0.02, "learning_rate": 4.9927811676617965e-05, "loss": 0.9656, "step": 24293 }, { "epoch": 0.02, "learning_rate": 4.9927805711665244e-05, "loss": 1.2079, "step": 24294 }, { "epoch": 0.02, "learning_rate": 4.9927799746466444e-05, "loss": 1.0992, "step": 24295 }, { "epoch": 0.02, "learning_rate": 4.992779378102156e-05, "loss": 1.7526, "step": 24296 }, { "epoch": 0.02, "learning_rate": 4.992778781533061e-05, "loss": 1.3603, "step": 24297 }, { "epoch": 0.02, "learning_rate": 4.992778184939358e-05, "loss": 1.2173, "step": 24298 }, { "epoch": 0.02, "learning_rate": 4.9927775883210475e-05, "loss": 1.1452, "step": 24299 }, { "epoch": 0.02, "learning_rate": 4.9927769916781294e-05, "loss": 1.1124, "step": 24300 }, { "epoch": 0.02, "learning_rate": 4.9927763950106034e-05, "loss": 0.5364, "step": 24301 }, { "epoch": 0.02, "learning_rate": 4.99277579831847e-05, "loss": 0.8146, "step": 24302 }, { "epoch": 0.02, "learning_rate": 4.992775201601729e-05, "loss": 1.1357, "step": 24303 }, { "epoch": 0.02, "learning_rate": 4.992774604860379e-05, "loss": 1.3369, "step": 24304 }, { "epoch": 0.02, "learning_rate": 4.992774008094423e-05, "loss": 0.9853, "step": 24305 }, { "epoch": 0.02, "learning_rate": 4.992773411303859e-05, "loss": 1.1325, "step": 24306 }, { "epoch": 0.02, "learning_rate": 4.992772814488688e-05, "loss": 1.0816, "step": 24307 }, { "epoch": 0.02, "learning_rate": 4.9927722176489086e-05, "loss": 0.8392, "step": 24308 }, { "epoch": 0.02, "learning_rate": 4.992771620784522e-05, "loss": 1.082, "step": 24309 }, { "epoch": 0.02, "learning_rate": 4.992771023895527e-05, "loss": 1.1061, "step": 24310 }, { "epoch": 0.02, "learning_rate": 4.992770426981925e-05, "loss": 0.7759, "step": 24311 }, { "epoch": 0.02, "learning_rate": 4.992769830043715e-05, "loss": 1.1955, "step": 24312 }, { "epoch": 0.02, "learning_rate": 4.9927692330808986e-05, "loss": 1.2285, "step": 24313 }, { "epoch": 0.02, "learning_rate": 4.992768636093474e-05, "loss": 1.1905, "step": 24314 }, { "epoch": 0.02, "learning_rate": 4.9927680390814414e-05, "loss": 0.8488, "step": 24315 }, { "epoch": 0.02, "learning_rate": 4.9927674420448014e-05, "loss": 1.0147, "step": 24316 }, { "epoch": 0.02, "learning_rate": 4.992766844983554e-05, "loss": 0.8666, "step": 24317 }, { "epoch": 0.02, "learning_rate": 4.9927662478976995e-05, "loss": 1.3212, "step": 24318 }, { "epoch": 0.02, "learning_rate": 4.992765650787237e-05, "loss": 1.1797, "step": 24319 }, { "epoch": 0.02, "learning_rate": 4.992765053652166e-05, "loss": 1.1304, "step": 24320 }, { "epoch": 0.02, "learning_rate": 4.992764456492489e-05, "loss": 0.9464, "step": 24321 }, { "epoch": 0.02, "learning_rate": 4.9927638593082045e-05, "loss": 0.7597, "step": 24322 }, { "epoch": 0.02, "learning_rate": 4.992763262099311e-05, "loss": 1.1407, "step": 24323 }, { "epoch": 0.02, "learning_rate": 4.992762664865811e-05, "loss": 1.2424, "step": 24324 }, { "epoch": 0.02, "learning_rate": 4.9927620676077034e-05, "loss": 1.0521, "step": 24325 }, { "epoch": 0.02, "learning_rate": 4.9927614703249895e-05, "loss": 1.1115, "step": 24326 }, { "epoch": 0.02, "learning_rate": 4.992760873017667e-05, "loss": 0.933, "step": 24327 }, { "epoch": 0.02, "learning_rate": 4.9927602756857364e-05, "loss": 1.0648, "step": 24328 }, { "epoch": 0.02, "learning_rate": 4.9927596783291985e-05, "loss": 0.9881, "step": 24329 }, { "epoch": 0.02, "learning_rate": 4.992759080948054e-05, "loss": 0.9343, "step": 24330 }, { "epoch": 0.02, "learning_rate": 4.9927584835423015e-05, "loss": 0.8709, "step": 24331 }, { "epoch": 0.02, "learning_rate": 4.992757886111941e-05, "loss": 1.1466, "step": 24332 }, { "epoch": 0.02, "learning_rate": 4.992757288656974e-05, "loss": 1.1672, "step": 24333 }, { "epoch": 0.02, "learning_rate": 4.992756691177399e-05, "loss": 1.2033, "step": 24334 }, { "epoch": 0.02, "learning_rate": 4.992756093673217e-05, "loss": 1.2005, "step": 24335 }, { "epoch": 0.02, "learning_rate": 4.992755496144427e-05, "loss": 0.9952, "step": 24336 }, { "epoch": 0.02, "learning_rate": 4.99275489859103e-05, "loss": 1.2088, "step": 24337 }, { "epoch": 0.02, "learning_rate": 4.9927543010130253e-05, "loss": 1.6756, "step": 24338 }, { "epoch": 0.02, "learning_rate": 4.9927537034104136e-05, "loss": 1.1861, "step": 24339 }, { "epoch": 0.02, "learning_rate": 4.992753105783194e-05, "loss": 1.1627, "step": 24340 }, { "epoch": 0.02, "learning_rate": 4.9927525081313666e-05, "loss": 1.3196, "step": 24341 }, { "epoch": 0.02, "learning_rate": 4.992751910454932e-05, "loss": 1.2791, "step": 24342 }, { "epoch": 0.02, "learning_rate": 4.992751312753891e-05, "loss": 1.0221, "step": 24343 }, { "epoch": 0.02, "learning_rate": 4.992750715028242e-05, "loss": 1.1273, "step": 24344 }, { "epoch": 0.02, "learning_rate": 4.9927501172779856e-05, "loss": 1.034, "step": 24345 }, { "epoch": 0.02, "learning_rate": 4.992749519503122e-05, "loss": 0.861, "step": 24346 }, { "epoch": 0.02, "learning_rate": 4.992748921703651e-05, "loss": 1.1572, "step": 24347 }, { "epoch": 0.02, "learning_rate": 4.9927483238795724e-05, "loss": 0.9568, "step": 24348 }, { "epoch": 0.02, "learning_rate": 4.992747726030886e-05, "loss": 0.3494, "step": 24349 }, { "epoch": 0.02, "learning_rate": 4.992747128157593e-05, "loss": 0.7869, "step": 24350 }, { "epoch": 0.02, "learning_rate": 4.992746530259693e-05, "loss": 1.0582, "step": 24351 }, { "epoch": 0.02, "learning_rate": 4.992745932337185e-05, "loss": 0.7102, "step": 24352 }, { "epoch": 0.02, "learning_rate": 4.992745334390069e-05, "loss": 0.9041, "step": 24353 }, { "epoch": 0.02, "learning_rate": 4.992744736418346e-05, "loss": 0.5888, "step": 24354 }, { "epoch": 0.02, "learning_rate": 4.992744138422017e-05, "loss": 0.887, "step": 24355 }, { "epoch": 0.02, "learning_rate": 4.99274354040108e-05, "loss": 1.117, "step": 24356 }, { "epoch": 0.02, "learning_rate": 4.992742942355535e-05, "loss": 0.7432, "step": 24357 }, { "epoch": 0.02, "learning_rate": 4.992742344285383e-05, "loss": 1.2744, "step": 24358 }, { "epoch": 0.02, "learning_rate": 4.992741746190624e-05, "loss": 0.7411, "step": 24359 }, { "epoch": 0.02, "learning_rate": 4.992741148071258e-05, "loss": 1.1617, "step": 24360 }, { "epoch": 0.02, "learning_rate": 4.9927405499272837e-05, "loss": 0.847, "step": 24361 }, { "epoch": 0.02, "learning_rate": 4.992739951758703e-05, "loss": 1.1165, "step": 24362 }, { "epoch": 0.02, "learning_rate": 4.9927393535655145e-05, "loss": 1.0725, "step": 24363 }, { "epoch": 0.02, "learning_rate": 4.992738755347719e-05, "loss": 1.1321, "step": 24364 }, { "epoch": 0.02, "learning_rate": 4.992738157105316e-05, "loss": 1.4605, "step": 24365 }, { "epoch": 0.02, "learning_rate": 4.9927375588383065e-05, "loss": 1.2717, "step": 24366 }, { "epoch": 0.02, "learning_rate": 4.992736960546689e-05, "loss": 1.2921, "step": 24367 }, { "epoch": 0.02, "learning_rate": 4.992736362230464e-05, "loss": 1.4048, "step": 24368 }, { "epoch": 0.02, "learning_rate": 4.992735763889632e-05, "loss": 1.2178, "step": 24369 }, { "epoch": 0.02, "learning_rate": 4.9927351655241925e-05, "loss": 1.259, "step": 24370 }, { "epoch": 0.02, "learning_rate": 4.9927345671341463e-05, "loss": 0.8468, "step": 24371 }, { "epoch": 0.02, "learning_rate": 4.992733968719493e-05, "loss": 0.4921, "step": 24372 }, { "epoch": 0.02, "learning_rate": 4.992733370280233e-05, "loss": 1.2083, "step": 24373 }, { "epoch": 0.02, "learning_rate": 4.992732771816364e-05, "loss": 0.8192, "step": 24374 }, { "epoch": 0.02, "learning_rate": 4.992732173327889e-05, "loss": 1.2897, "step": 24375 }, { "epoch": 0.02, "learning_rate": 4.992731574814807e-05, "loss": 1.1029, "step": 24376 }, { "epoch": 0.02, "learning_rate": 4.992730976277117e-05, "loss": 1.0785, "step": 24377 }, { "epoch": 0.02, "learning_rate": 4.99273037771482e-05, "loss": 1.0746, "step": 24378 }, { "epoch": 0.02, "learning_rate": 4.992729779127916e-05, "loss": 0.9224, "step": 24379 }, { "epoch": 0.02, "learning_rate": 4.9927291805164055e-05, "loss": 1.0075, "step": 24380 }, { "epoch": 0.02, "learning_rate": 4.992728581880287e-05, "loss": 0.7694, "step": 24381 }, { "epoch": 0.02, "learning_rate": 4.9927279832195615e-05, "loss": 1.1058, "step": 24382 }, { "epoch": 0.02, "learning_rate": 4.992727384534229e-05, "loss": 1.1263, "step": 24383 }, { "epoch": 0.02, "learning_rate": 4.992726785824289e-05, "loss": 0.9603, "step": 24384 }, { "epoch": 0.02, "learning_rate": 4.992726187089742e-05, "loss": 1.2155, "step": 24385 }, { "epoch": 0.02, "learning_rate": 4.992725588330588e-05, "loss": 1.3387, "step": 24386 }, { "epoch": 0.02, "learning_rate": 4.992724989546826e-05, "loss": 1.3723, "step": 24387 }, { "epoch": 0.02, "learning_rate": 4.992724390738458e-05, "loss": 1.0002, "step": 24388 }, { "epoch": 0.02, "learning_rate": 4.9927237919054824e-05, "loss": 1.2258, "step": 24389 }, { "epoch": 0.02, "learning_rate": 4.9927231930479005e-05, "loss": 0.7858, "step": 24390 }, { "epoch": 0.02, "learning_rate": 4.9927225941657106e-05, "loss": 1.2825, "step": 24391 }, { "epoch": 0.02, "learning_rate": 4.992721995258914e-05, "loss": 1.1404, "step": 24392 }, { "epoch": 0.02, "learning_rate": 4.99272139632751e-05, "loss": 0.9279, "step": 24393 }, { "epoch": 0.02, "learning_rate": 4.992720797371498e-05, "loss": 1.1167, "step": 24394 }, { "epoch": 0.02, "learning_rate": 4.9927201983908804e-05, "loss": 1.3172, "step": 24395 }, { "epoch": 0.02, "learning_rate": 4.992719599385656e-05, "loss": 0.9265, "step": 24396 }, { "epoch": 0.02, "learning_rate": 4.9927190003558234e-05, "loss": 1.3116, "step": 24397 }, { "epoch": 0.02, "learning_rate": 4.9927184013013836e-05, "loss": 1.1274, "step": 24398 }, { "epoch": 0.02, "learning_rate": 4.992717802222338e-05, "loss": 1.1781, "step": 24399 }, { "epoch": 0.02, "learning_rate": 4.992717203118684e-05, "loss": 0.9653, "step": 24400 }, { "epoch": 0.02, "learning_rate": 4.9927166039904236e-05, "loss": 1.1877, "step": 24401 }, { "epoch": 0.02, "learning_rate": 4.992716004837556e-05, "loss": 1.15, "step": 24402 }, { "epoch": 0.02, "learning_rate": 4.992715405660081e-05, "loss": 1.2119, "step": 24403 }, { "epoch": 0.02, "learning_rate": 4.992714806458e-05, "loss": 1.0378, "step": 24404 }, { "epoch": 0.02, "learning_rate": 4.992714207231311e-05, "loss": 1.3089, "step": 24405 }, { "epoch": 0.02, "learning_rate": 4.992713607980015e-05, "loss": 1.4753, "step": 24406 }, { "epoch": 0.02, "learning_rate": 4.9927130087041126e-05, "loss": 0.8042, "step": 24407 }, { "epoch": 0.02, "learning_rate": 4.992712409403602e-05, "loss": 0.8297, "step": 24408 }, { "epoch": 0.02, "learning_rate": 4.992711810078487e-05, "loss": 1.0259, "step": 24409 }, { "epoch": 0.02, "learning_rate": 4.9927112107287624e-05, "loss": 1.1014, "step": 24410 }, { "epoch": 0.02, "learning_rate": 4.992710611354432e-05, "loss": 0.9907, "step": 24411 }, { "epoch": 0.02, "learning_rate": 4.992710011955494e-05, "loss": 1.1886, "step": 24412 }, { "epoch": 0.02, "learning_rate": 4.9927094125319496e-05, "loss": 0.9516, "step": 24413 }, { "epoch": 0.02, "learning_rate": 4.992708813083797e-05, "loss": 0.9904, "step": 24414 }, { "epoch": 0.02, "learning_rate": 4.992708213611039e-05, "loss": 1.2044, "step": 24415 }, { "epoch": 0.02, "learning_rate": 4.992707614113673e-05, "loss": 0.9392, "step": 24416 }, { "epoch": 0.02, "learning_rate": 4.992707014591701e-05, "loss": 1.1496, "step": 24417 }, { "epoch": 0.02, "learning_rate": 4.9927064150451205e-05, "loss": 1.4856, "step": 24418 }, { "epoch": 0.02, "learning_rate": 4.992705815473935e-05, "loss": 0.8448, "step": 24419 }, { "epoch": 0.02, "learning_rate": 4.9927052158781415e-05, "loss": 0.8749, "step": 24420 }, { "epoch": 0.02, "learning_rate": 4.992704616257741e-05, "loss": 1.014, "step": 24421 }, { "epoch": 0.02, "learning_rate": 4.992704016612734e-05, "loss": 1.2585, "step": 24422 }, { "epoch": 0.02, "learning_rate": 4.99270341694312e-05, "loss": 1.0814, "step": 24423 }, { "epoch": 0.02, "learning_rate": 4.9927028172488984e-05, "loss": 1.3848, "step": 24424 }, { "epoch": 0.02, "learning_rate": 4.9927022175300704e-05, "loss": 0.945, "step": 24425 }, { "epoch": 0.02, "learning_rate": 4.992701617786636e-05, "loss": 0.6783, "step": 24426 }, { "epoch": 0.02, "learning_rate": 4.9927010180185936e-05, "loss": 0.7239, "step": 24427 }, { "epoch": 0.02, "learning_rate": 4.992700418225945e-05, "loss": 1.0416, "step": 24428 }, { "epoch": 0.02, "learning_rate": 4.992699818408689e-05, "loss": 0.9032, "step": 24429 }, { "epoch": 0.02, "learning_rate": 4.992699218566826e-05, "loss": 1.2013, "step": 24430 }, { "epoch": 0.02, "learning_rate": 4.992698618700358e-05, "loss": 0.8852, "step": 24431 }, { "epoch": 0.02, "learning_rate": 4.992698018809281e-05, "loss": 1.1553, "step": 24432 }, { "epoch": 0.02, "learning_rate": 4.9926974188935985e-05, "loss": 1.0157, "step": 24433 }, { "epoch": 0.02, "learning_rate": 4.992696818953308e-05, "loss": 0.914, "step": 24434 }, { "epoch": 0.02, "learning_rate": 4.9926962189884107e-05, "loss": 1.0104, "step": 24435 }, { "epoch": 0.02, "learning_rate": 4.9926956189989075e-05, "loss": 1.0219, "step": 24436 }, { "epoch": 0.02, "learning_rate": 4.992695018984797e-05, "loss": 2.6764, "step": 24437 }, { "epoch": 0.02, "learning_rate": 4.99269441894608e-05, "loss": 1.0226, "step": 24438 }, { "epoch": 0.02, "learning_rate": 4.992693818882755e-05, "loss": 1.2257, "step": 24439 }, { "epoch": 0.02, "learning_rate": 4.992693218794824e-05, "loss": 1.2859, "step": 24440 }, { "epoch": 0.02, "learning_rate": 4.992692618682286e-05, "loss": 1.3383, "step": 24441 }, { "epoch": 0.02, "learning_rate": 4.992692018545142e-05, "loss": 1.6692, "step": 24442 }, { "epoch": 0.02, "learning_rate": 4.992691418383391e-05, "loss": 1.1903, "step": 24443 }, { "epoch": 0.02, "learning_rate": 4.9926908181970324e-05, "loss": 0.7703, "step": 24444 }, { "epoch": 0.02, "learning_rate": 4.9926902179860674e-05, "loss": 1.1083, "step": 24445 }, { "epoch": 0.02, "learning_rate": 4.992689617750496e-05, "loss": 1.1153, "step": 24446 }, { "epoch": 0.02, "learning_rate": 4.9926890174903166e-05, "loss": 1.2537, "step": 24447 }, { "epoch": 0.02, "learning_rate": 4.9926884172055317e-05, "loss": 0.9862, "step": 24448 }, { "epoch": 0.02, "learning_rate": 4.9926878168961393e-05, "loss": 1.203, "step": 24449 }, { "epoch": 0.02, "learning_rate": 4.992687216562141e-05, "loss": 0.42, "step": 24450 }, { "epoch": 0.02, "learning_rate": 4.992686616203535e-05, "loss": 0.9616, "step": 24451 }, { "epoch": 0.02, "learning_rate": 4.9926860158203225e-05, "loss": 1.6385, "step": 24452 }, { "epoch": 0.02, "learning_rate": 4.9926854154125036e-05, "loss": 1.2251, "step": 24453 }, { "epoch": 0.02, "learning_rate": 4.9926848149800774e-05, "loss": 0.8885, "step": 24454 }, { "epoch": 0.02, "learning_rate": 4.9926842145230446e-05, "loss": 0.9874, "step": 24455 }, { "epoch": 0.02, "learning_rate": 4.992683614041405e-05, "loss": 1.2189, "step": 24456 }, { "epoch": 0.02, "learning_rate": 4.9926830135351596e-05, "loss": 0.8753, "step": 24457 }, { "epoch": 0.02, "learning_rate": 4.992682413004307e-05, "loss": 0.9704, "step": 24458 }, { "epoch": 0.02, "learning_rate": 4.992681812448847e-05, "loss": 1.2693, "step": 24459 }, { "epoch": 0.02, "learning_rate": 4.9926812118687806e-05, "loss": 0.6925, "step": 24460 }, { "epoch": 0.02, "learning_rate": 4.992680611264108e-05, "loss": 0.965, "step": 24461 }, { "epoch": 0.02, "learning_rate": 4.9926800106348284e-05, "loss": 0.7978, "step": 24462 }, { "epoch": 0.02, "learning_rate": 4.992679409980942e-05, "loss": 1.3209, "step": 24463 }, { "epoch": 0.02, "learning_rate": 4.992678809302449e-05, "loss": 1.4389, "step": 24464 }, { "epoch": 0.02, "learning_rate": 4.99267820859935e-05, "loss": 1.2622, "step": 24465 }, { "epoch": 0.02, "learning_rate": 4.992677607871643e-05, "loss": 1.139, "step": 24466 }, { "epoch": 0.02, "learning_rate": 4.99267700711933e-05, "loss": 0.3752, "step": 24467 }, { "epoch": 0.02, "learning_rate": 4.9926764063424105e-05, "loss": 0.3013, "step": 24468 }, { "epoch": 0.02, "learning_rate": 4.992675805540884e-05, "loss": 1.2789, "step": 24469 }, { "epoch": 0.02, "learning_rate": 4.992675204714751e-05, "loss": 0.8498, "step": 24470 }, { "epoch": 0.02, "learning_rate": 4.9926746038640115e-05, "loss": 0.9297, "step": 24471 }, { "epoch": 0.02, "learning_rate": 4.992674002988666e-05, "loss": 1.2087, "step": 24472 }, { "epoch": 0.02, "learning_rate": 4.9926734020887125e-05, "loss": 0.9166, "step": 24473 }, { "epoch": 0.02, "learning_rate": 4.9926728011641534e-05, "loss": 1.0591, "step": 24474 }, { "epoch": 0.02, "learning_rate": 4.9926722002149876e-05, "loss": 1.1709, "step": 24475 }, { "epoch": 0.02, "learning_rate": 4.9926715992412145e-05, "loss": 1.163, "step": 24476 }, { "epoch": 0.02, "learning_rate": 4.992670998242835e-05, "loss": 1.0705, "step": 24477 }, { "epoch": 0.02, "learning_rate": 4.992670397219849e-05, "loss": 1.2218, "step": 24478 }, { "epoch": 0.02, "learning_rate": 4.9926697961722566e-05, "loss": 1.2339, "step": 24479 }, { "epoch": 0.02, "learning_rate": 4.9926691951000576e-05, "loss": 1.131, "step": 24480 }, { "epoch": 0.02, "learning_rate": 4.992668594003252e-05, "loss": 1.1568, "step": 24481 }, { "epoch": 0.02, "learning_rate": 4.9926679928818397e-05, "loss": 1.0256, "step": 24482 }, { "epoch": 0.02, "learning_rate": 4.992667391735821e-05, "loss": 0.9029, "step": 24483 }, { "epoch": 0.02, "learning_rate": 4.992666790565196e-05, "loss": 1.1322, "step": 24484 }, { "epoch": 0.02, "learning_rate": 4.9926661893699635e-05, "loss": 1.4244, "step": 24485 }, { "epoch": 0.02, "learning_rate": 4.9926655881501247e-05, "loss": 1.3805, "step": 24486 }, { "epoch": 0.02, "learning_rate": 4.9926649869056805e-05, "loss": 1.053, "step": 24487 }, { "epoch": 0.02, "learning_rate": 4.992664385636628e-05, "loss": 1.303, "step": 24488 }, { "epoch": 0.02, "learning_rate": 4.99266378434297e-05, "loss": 0.9932, "step": 24489 }, { "epoch": 0.02, "learning_rate": 4.992663183024706e-05, "loss": 1.2436, "step": 24490 }, { "epoch": 0.02, "learning_rate": 4.992662581681835e-05, "loss": 0.926, "step": 24491 }, { "epoch": 0.02, "learning_rate": 4.992661980314357e-05, "loss": 1.2446, "step": 24492 }, { "epoch": 0.02, "learning_rate": 4.992661378922273e-05, "loss": 0.9082, "step": 24493 }, { "epoch": 0.02, "learning_rate": 4.9926607775055824e-05, "loss": 1.2247, "step": 24494 }, { "epoch": 0.02, "learning_rate": 4.992660176064286e-05, "loss": 1.2939, "step": 24495 }, { "epoch": 0.02, "learning_rate": 4.992659574598382e-05, "loss": 1.1958, "step": 24496 }, { "epoch": 0.02, "learning_rate": 4.992658973107872e-05, "loss": 1.2497, "step": 24497 }, { "epoch": 0.02, "learning_rate": 4.992658371592755e-05, "loss": 1.1057, "step": 24498 }, { "epoch": 0.02, "learning_rate": 4.992657770053032e-05, "loss": 1.0389, "step": 24499 }, { "epoch": 0.02, "learning_rate": 4.992657168488703e-05, "loss": 1.1692, "step": 24500 }, { "epoch": 0.02, "eval_loss": 1.0119251012802124, "eval_runtime": 83.4411, "eval_samples_per_second": 16.599, "eval_steps_per_second": 4.159, "step": 24500 }, { "epoch": 0.02, "learning_rate": 4.992656566899766e-05, "loss": 1.1565, "step": 24501 }, { "epoch": 0.02, "learning_rate": 4.992655965286224e-05, "loss": 1.0502, "step": 24502 }, { "epoch": 0.02, "learning_rate": 4.992655363648075e-05, "loss": 0.9798, "step": 24503 }, { "epoch": 0.02, "learning_rate": 4.99265476198532e-05, "loss": 0.919, "step": 24504 }, { "epoch": 0.02, "learning_rate": 4.992654160297958e-05, "loss": 1.1145, "step": 24505 }, { "epoch": 0.02, "learning_rate": 4.99265355858599e-05, "loss": 1.0337, "step": 24506 }, { "epoch": 0.02, "learning_rate": 4.992652956849415e-05, "loss": 1.0843, "step": 24507 }, { "epoch": 0.02, "learning_rate": 4.9926523550882345e-05, "loss": 0.9274, "step": 24508 }, { "epoch": 0.02, "learning_rate": 4.992651753302446e-05, "loss": 1.1041, "step": 24509 }, { "epoch": 0.02, "learning_rate": 4.9926511514920525e-05, "loss": 1.0121, "step": 24510 }, { "epoch": 0.02, "learning_rate": 4.992650549657052e-05, "loss": 0.9965, "step": 24511 }, { "epoch": 0.02, "learning_rate": 4.992649947797445e-05, "loss": 0.8199, "step": 24512 }, { "epoch": 0.02, "learning_rate": 4.992649345913233e-05, "loss": 1.2358, "step": 24513 }, { "epoch": 0.02, "learning_rate": 4.9926487440044134e-05, "loss": 1.5196, "step": 24514 }, { "epoch": 0.02, "learning_rate": 4.992648142070987e-05, "loss": 0.8739, "step": 24515 }, { "epoch": 0.02, "learning_rate": 4.992647540112955e-05, "loss": 1.2081, "step": 24516 }, { "epoch": 0.02, "learning_rate": 4.992646938130316e-05, "loss": 1.113, "step": 24517 }, { "epoch": 0.02, "learning_rate": 4.9926463361230715e-05, "loss": 1.2594, "step": 24518 }, { "epoch": 0.02, "learning_rate": 4.99264573409122e-05, "loss": 1.1732, "step": 24519 }, { "epoch": 0.02, "learning_rate": 4.992645132034763e-05, "loss": 1.0771, "step": 24520 }, { "epoch": 0.02, "learning_rate": 4.9926445299536986e-05, "loss": 1.1254, "step": 24521 }, { "epoch": 0.02, "learning_rate": 4.9926439278480286e-05, "loss": 0.9157, "step": 24522 }, { "epoch": 0.02, "learning_rate": 4.992643325717752e-05, "loss": 0.945, "step": 24523 }, { "epoch": 0.02, "learning_rate": 4.992642723562869e-05, "loss": 1.0745, "step": 24524 }, { "epoch": 0.02, "learning_rate": 4.9926421213833794e-05, "loss": 0.8978, "step": 24525 }, { "epoch": 0.02, "learning_rate": 4.992641519179284e-05, "loss": 1.1127, "step": 24526 }, { "epoch": 0.02, "learning_rate": 4.992640916950582e-05, "loss": 1.1984, "step": 24527 }, { "epoch": 0.02, "learning_rate": 4.992640314697274e-05, "loss": 0.9217, "step": 24528 }, { "epoch": 0.02, "learning_rate": 4.992639712419359e-05, "loss": 1.1399, "step": 24529 }, { "epoch": 0.02, "learning_rate": 4.992639110116839e-05, "loss": 0.7392, "step": 24530 }, { "epoch": 0.02, "learning_rate": 4.9926385077897117e-05, "loss": 0.7058, "step": 24531 }, { "epoch": 0.02, "learning_rate": 4.992637905437978e-05, "loss": 0.6144, "step": 24532 }, { "epoch": 0.02, "learning_rate": 4.992637303061639e-05, "loss": 0.5799, "step": 24533 }, { "epoch": 0.02, "learning_rate": 4.9926367006606925e-05, "loss": 0.576, "step": 24534 }, { "epoch": 0.02, "learning_rate": 4.992636098235141e-05, "loss": 0.6727, "step": 24535 }, { "epoch": 0.02, "learning_rate": 4.992635495784982e-05, "loss": 0.6656, "step": 24536 }, { "epoch": 0.02, "learning_rate": 4.9926348933102175e-05, "loss": 0.6454, "step": 24537 }, { "epoch": 0.02, "learning_rate": 4.992634290810847e-05, "loss": 0.8548, "step": 24538 }, { "epoch": 0.02, "learning_rate": 4.99263368828687e-05, "loss": 0.9307, "step": 24539 }, { "epoch": 0.02, "learning_rate": 4.992633085738286e-05, "loss": 1.1607, "step": 24540 }, { "epoch": 0.02, "learning_rate": 4.992632483165097e-05, "loss": 1.2171, "step": 24541 }, { "epoch": 0.02, "learning_rate": 4.992631880567301e-05, "loss": 1.1779, "step": 24542 }, { "epoch": 0.02, "learning_rate": 4.992631277944899e-05, "loss": 1.2085, "step": 24543 }, { "epoch": 0.02, "learning_rate": 4.992630675297891e-05, "loss": 1.1175, "step": 24544 }, { "epoch": 0.02, "learning_rate": 4.992630072626276e-05, "loss": 0.9825, "step": 24545 }, { "epoch": 0.02, "learning_rate": 4.9926294699300556e-05, "loss": 1.4949, "step": 24546 }, { "epoch": 0.02, "learning_rate": 4.992628867209229e-05, "loss": 1.4488, "step": 24547 }, { "epoch": 0.02, "learning_rate": 4.9926282644637956e-05, "loss": 1.0446, "step": 24548 }, { "epoch": 0.02, "learning_rate": 4.992627661693757e-05, "loss": 1.0881, "step": 24549 }, { "epoch": 0.02, "learning_rate": 4.992627058899112e-05, "loss": 1.4002, "step": 24550 }, { "epoch": 0.02, "learning_rate": 4.99262645607986e-05, "loss": 1.203, "step": 24551 }, { "epoch": 0.02, "learning_rate": 4.9926258532360026e-05, "loss": 0.9141, "step": 24552 }, { "epoch": 0.02, "learning_rate": 4.992625250367539e-05, "loss": 1.1793, "step": 24553 }, { "epoch": 0.02, "learning_rate": 4.992624647474469e-05, "loss": 1.056, "step": 24554 }, { "epoch": 0.02, "learning_rate": 4.992624044556793e-05, "loss": 1.0449, "step": 24555 }, { "epoch": 0.02, "learning_rate": 4.99262344161451e-05, "loss": 1.2581, "step": 24556 }, { "epoch": 0.02, "learning_rate": 4.992622838647622e-05, "loss": 1.0222, "step": 24557 }, { "epoch": 0.02, "learning_rate": 4.9926222356561274e-05, "loss": 1.0478, "step": 24558 }, { "epoch": 0.02, "learning_rate": 4.9926216326400264e-05, "loss": 1.0678, "step": 24559 }, { "epoch": 0.02, "learning_rate": 4.99262102959932e-05, "loss": 0.9311, "step": 24560 }, { "epoch": 0.02, "learning_rate": 4.992620426534007e-05, "loss": 1.2125, "step": 24561 }, { "epoch": 0.02, "learning_rate": 4.992619823444088e-05, "loss": 0.8974, "step": 24562 }, { "epoch": 0.02, "learning_rate": 4.9926192203295634e-05, "loss": 1.0543, "step": 24563 }, { "epoch": 0.02, "learning_rate": 4.9926186171904325e-05, "loss": 1.0005, "step": 24564 }, { "epoch": 0.02, "learning_rate": 4.992618014026696e-05, "loss": 1.0912, "step": 24565 }, { "epoch": 0.02, "learning_rate": 4.992617410838352e-05, "loss": 1.1329, "step": 24566 }, { "epoch": 0.02, "learning_rate": 4.9926168076254035e-05, "loss": 1.211, "step": 24567 }, { "epoch": 0.02, "learning_rate": 4.9926162043878474e-05, "loss": 1.0138, "step": 24568 }, { "epoch": 0.02, "learning_rate": 4.992615601125686e-05, "loss": 1.0469, "step": 24569 }, { "epoch": 0.02, "learning_rate": 4.992614997838919e-05, "loss": 0.9207, "step": 24570 }, { "epoch": 0.02, "learning_rate": 4.992614394527545e-05, "loss": 1.1993, "step": 24571 }, { "epoch": 0.02, "learning_rate": 4.9926137911915655e-05, "loss": 1.0654, "step": 24572 }, { "epoch": 0.02, "learning_rate": 4.99261318783098e-05, "loss": 1.504, "step": 24573 }, { "epoch": 0.02, "learning_rate": 4.9926125844457884e-05, "loss": 1.1942, "step": 24574 }, { "epoch": 0.02, "learning_rate": 4.9926119810359905e-05, "loss": 1.1104, "step": 24575 }, { "epoch": 0.02, "learning_rate": 4.992611377601587e-05, "loss": 0.8931, "step": 24576 }, { "epoch": 0.02, "learning_rate": 4.9926107741425776e-05, "loss": 0.9287, "step": 24577 }, { "epoch": 0.02, "learning_rate": 4.992610170658962e-05, "loss": 1.2326, "step": 24578 }, { "epoch": 0.02, "learning_rate": 4.99260956715074e-05, "loss": 1.0249, "step": 24579 }, { "epoch": 0.02, "learning_rate": 4.9926089636179124e-05, "loss": 0.7135, "step": 24580 }, { "epoch": 0.02, "learning_rate": 4.992608360060479e-05, "loss": 0.621, "step": 24581 }, { "epoch": 0.02, "learning_rate": 4.992607756478439e-05, "loss": 0.8204, "step": 24582 }, { "epoch": 0.02, "learning_rate": 4.9926071528717935e-05, "loss": 1.1137, "step": 24583 }, { "epoch": 0.02, "learning_rate": 4.992606549240542e-05, "loss": 0.9056, "step": 24584 }, { "epoch": 0.02, "learning_rate": 4.9926059455846844e-05, "loss": 1.1476, "step": 24585 }, { "epoch": 0.02, "learning_rate": 4.992605341904221e-05, "loss": 0.3556, "step": 24586 }, { "epoch": 0.02, "learning_rate": 4.9926047381991515e-05, "loss": 0.8778, "step": 24587 }, { "epoch": 0.02, "learning_rate": 4.992604134469476e-05, "loss": 0.8891, "step": 24588 }, { "epoch": 0.02, "learning_rate": 4.992603530715195e-05, "loss": 1.1661, "step": 24589 }, { "epoch": 0.02, "learning_rate": 4.992602926936308e-05, "loss": 1.3675, "step": 24590 }, { "epoch": 0.02, "learning_rate": 4.9926023231328146e-05, "loss": 1.1644, "step": 24591 }, { "epoch": 0.02, "learning_rate": 4.992601719304716e-05, "loss": 1.0584, "step": 24592 }, { "epoch": 0.02, "learning_rate": 4.99260111545201e-05, "loss": 1.1671, "step": 24593 }, { "epoch": 0.02, "learning_rate": 4.9926005115747e-05, "loss": 0.9462, "step": 24594 }, { "epoch": 0.02, "learning_rate": 4.992599907672783e-05, "loss": 0.9978, "step": 24595 }, { "epoch": 0.02, "learning_rate": 4.9925993037462604e-05, "loss": 1.2193, "step": 24596 }, { "epoch": 0.02, "learning_rate": 4.992598699795132e-05, "loss": 0.6275, "step": 24597 }, { "epoch": 0.02, "learning_rate": 4.9925980958193975e-05, "loss": 1.0742, "step": 24598 }, { "epoch": 0.02, "learning_rate": 4.9925974918190564e-05, "loss": 1.2055, "step": 24599 }, { "epoch": 0.02, "learning_rate": 4.992596887794111e-05, "loss": 0.8926, "step": 24600 }, { "epoch": 0.02, "learning_rate": 4.992596283744558e-05, "loss": 1.2732, "step": 24601 }, { "epoch": 0.02, "learning_rate": 4.992595679670401e-05, "loss": 1.0264, "step": 24602 }, { "epoch": 0.02, "learning_rate": 4.9925950755716364e-05, "loss": 0.6964, "step": 24603 }, { "epoch": 0.02, "learning_rate": 4.9925944714482675e-05, "loss": 1.2781, "step": 24604 }, { "epoch": 0.02, "learning_rate": 4.992593867300291e-05, "loss": 1.0458, "step": 24605 }, { "epoch": 0.02, "learning_rate": 4.9925932631277105e-05, "loss": 1.132, "step": 24606 }, { "epoch": 0.02, "learning_rate": 4.992592658930523e-05, "loss": 1.0652, "step": 24607 }, { "epoch": 0.02, "learning_rate": 4.99259205470873e-05, "loss": 0.8006, "step": 24608 }, { "epoch": 0.02, "learning_rate": 4.9925914504623315e-05, "loss": 0.9243, "step": 24609 }, { "epoch": 0.02, "learning_rate": 4.992590846191327e-05, "loss": 0.9712, "step": 24610 }, { "epoch": 0.02, "learning_rate": 4.992590241895716e-05, "loss": 0.9978, "step": 24611 }, { "epoch": 0.02, "learning_rate": 4.9925896375755e-05, "loss": 0.8557, "step": 24612 }, { "epoch": 0.02, "learning_rate": 4.992589033230678e-05, "loss": 1.0034, "step": 24613 }, { "epoch": 0.02, "learning_rate": 4.99258842886125e-05, "loss": 0.884, "step": 24614 }, { "epoch": 0.02, "learning_rate": 4.9925878244672164e-05, "loss": 0.839, "step": 24615 }, { "epoch": 0.02, "learning_rate": 4.992587220048577e-05, "loss": 1.0983, "step": 24616 }, { "epoch": 0.02, "learning_rate": 4.992586615605333e-05, "loss": 1.2438, "step": 24617 }, { "epoch": 0.02, "learning_rate": 4.992586011137481e-05, "loss": 0.9623, "step": 24618 }, { "epoch": 0.02, "learning_rate": 4.992585406645025e-05, "loss": 0.9664, "step": 24619 }, { "epoch": 0.02, "learning_rate": 4.992584802127962e-05, "loss": 1.1733, "step": 24620 }, { "epoch": 0.02, "learning_rate": 4.992584197586294e-05, "loss": 1.3532, "step": 24621 }, { "epoch": 0.02, "learning_rate": 4.99258359302002e-05, "loss": 0.9588, "step": 24622 }, { "epoch": 0.02, "learning_rate": 4.992582988429141e-05, "loss": 1.8154, "step": 24623 }, { "epoch": 0.02, "learning_rate": 4.992582383813655e-05, "loss": 0.7109, "step": 24624 }, { "epoch": 0.02, "learning_rate": 4.9925817791735644e-05, "loss": 0.761, "step": 24625 }, { "epoch": 0.02, "learning_rate": 4.9925811745088675e-05, "loss": 0.9894, "step": 24626 }, { "epoch": 0.02, "learning_rate": 4.992580569819565e-05, "loss": 0.8915, "step": 24627 }, { "epoch": 0.02, "learning_rate": 4.9925799651056574e-05, "loss": 0.9638, "step": 24628 }, { "epoch": 0.02, "learning_rate": 4.9925793603671426e-05, "loss": 0.9968, "step": 24629 }, { "epoch": 0.02, "learning_rate": 4.992578755604023e-05, "loss": 0.9739, "step": 24630 }, { "epoch": 0.02, "learning_rate": 4.992578150816298e-05, "loss": 1.0269, "step": 24631 }, { "epoch": 0.02, "learning_rate": 4.992577546003967e-05, "loss": 1.0936, "step": 24632 }, { "epoch": 0.02, "learning_rate": 4.99257694116703e-05, "loss": 1.0479, "step": 24633 }, { "epoch": 0.02, "learning_rate": 4.992576336305488e-05, "loss": 1.2263, "step": 24634 }, { "epoch": 0.02, "learning_rate": 4.99257573141934e-05, "loss": 0.9192, "step": 24635 }, { "epoch": 0.02, "learning_rate": 4.992575126508586e-05, "loss": 1.0719, "step": 24636 }, { "epoch": 0.02, "learning_rate": 4.992574521573227e-05, "loss": 0.8153, "step": 24637 }, { "epoch": 0.02, "learning_rate": 4.9925739166132626e-05, "loss": 1.0202, "step": 24638 }, { "epoch": 0.02, "learning_rate": 4.9925733116286916e-05, "loss": 1.0385, "step": 24639 }, { "epoch": 0.02, "learning_rate": 4.9925727066195146e-05, "loss": 1.1777, "step": 24640 }, { "epoch": 0.02, "learning_rate": 4.992572101585734e-05, "loss": 1.1301, "step": 24641 }, { "epoch": 0.02, "learning_rate": 4.992571496527346e-05, "loss": 0.521, "step": 24642 }, { "epoch": 0.02, "learning_rate": 4.9925708914443526e-05, "loss": 0.3122, "step": 24643 }, { "epoch": 0.02, "learning_rate": 4.992570286336754e-05, "loss": 0.4167, "step": 24644 }, { "epoch": 0.02, "learning_rate": 4.99256968120455e-05, "loss": 0.9854, "step": 24645 }, { "epoch": 0.02, "learning_rate": 4.99256907604774e-05, "loss": 1.1117, "step": 24646 }, { "epoch": 0.02, "learning_rate": 4.992568470866324e-05, "loss": 0.9005, "step": 24647 }, { "epoch": 0.02, "learning_rate": 4.9925678656603036e-05, "loss": 0.6621, "step": 24648 }, { "epoch": 0.02, "learning_rate": 4.992567260429677e-05, "loss": 1.0432, "step": 24649 }, { "epoch": 0.02, "learning_rate": 4.9925666551744444e-05, "loss": 1.05, "step": 24650 }, { "epoch": 0.02, "learning_rate": 4.992566049894607e-05, "loss": 0.9411, "step": 24651 }, { "epoch": 0.02, "learning_rate": 4.992565444590164e-05, "loss": 1.0306, "step": 24652 }, { "epoch": 0.02, "learning_rate": 4.992564839261115e-05, "loss": 1.1315, "step": 24653 }, { "epoch": 0.02, "learning_rate": 4.9925642339074604e-05, "loss": 1.261, "step": 24654 }, { "epoch": 0.02, "learning_rate": 4.9925636285292e-05, "loss": 1.1779, "step": 24655 }, { "epoch": 0.02, "learning_rate": 4.992563023126334e-05, "loss": 0.8318, "step": 24656 }, { "epoch": 0.02, "learning_rate": 4.992562417698864e-05, "loss": 1.0357, "step": 24657 }, { "epoch": 0.02, "learning_rate": 4.992561812246787e-05, "loss": 1.0619, "step": 24658 }, { "epoch": 0.02, "learning_rate": 4.992561206770105e-05, "loss": 1.4076, "step": 24659 }, { "epoch": 0.02, "learning_rate": 4.992560601268817e-05, "loss": 0.951, "step": 24660 }, { "epoch": 0.02, "learning_rate": 4.9925599957429244e-05, "loss": 1.1716, "step": 24661 }, { "epoch": 0.02, "learning_rate": 4.9925593901924256e-05, "loss": 1.1523, "step": 24662 }, { "epoch": 0.02, "learning_rate": 4.9925587846173215e-05, "loss": 0.7192, "step": 24663 }, { "epoch": 0.02, "learning_rate": 4.9925581790176115e-05, "loss": 0.9731, "step": 24664 }, { "epoch": 0.02, "learning_rate": 4.992557573393297e-05, "loss": 1.1622, "step": 24665 }, { "epoch": 0.02, "learning_rate": 4.992556967744376e-05, "loss": 0.9156, "step": 24666 }, { "epoch": 0.02, "learning_rate": 4.99255636207085e-05, "loss": 0.807, "step": 24667 }, { "epoch": 0.02, "learning_rate": 4.992555756372719e-05, "loss": 0.9207, "step": 24668 }, { "epoch": 0.02, "learning_rate": 4.992555150649981e-05, "loss": 0.3062, "step": 24669 }, { "epoch": 0.02, "learning_rate": 4.992554544902639e-05, "loss": 0.2771, "step": 24670 }, { "epoch": 0.02, "learning_rate": 4.992553939130691e-05, "loss": 0.3224, "step": 24671 }, { "epoch": 0.02, "learning_rate": 4.992553333334138e-05, "loss": 0.9651, "step": 24672 }, { "epoch": 0.02, "learning_rate": 4.992552727512979e-05, "loss": 0.8582, "step": 24673 }, { "epoch": 0.02, "learning_rate": 4.992552121667215e-05, "loss": 0.8406, "step": 24674 }, { "epoch": 0.02, "learning_rate": 4.992551515796845e-05, "loss": 1.1696, "step": 24675 }, { "epoch": 0.02, "learning_rate": 4.99255090990187e-05, "loss": 0.9383, "step": 24676 }, { "epoch": 0.02, "learning_rate": 4.99255030398229e-05, "loss": 1.0347, "step": 24677 }, { "epoch": 0.02, "learning_rate": 4.9925496980381037e-05, "loss": 0.9808, "step": 24678 }, { "epoch": 0.02, "learning_rate": 4.992549092069313e-05, "loss": 1.2078, "step": 24679 }, { "epoch": 0.02, "learning_rate": 4.992548486075916e-05, "loss": 1.2208, "step": 24680 }, { "epoch": 0.02, "learning_rate": 4.992547880057914e-05, "loss": 0.4045, "step": 24681 }, { "epoch": 0.02, "learning_rate": 4.992547274015307e-05, "loss": 1.1093, "step": 24682 }, { "epoch": 0.02, "learning_rate": 4.9925466679480935e-05, "loss": 1.2134, "step": 24683 }, { "epoch": 0.02, "learning_rate": 4.9925460618562756e-05, "loss": 0.7896, "step": 24684 }, { "epoch": 0.02, "learning_rate": 4.9925454557398524e-05, "loss": 0.4962, "step": 24685 }, { "epoch": 0.02, "learning_rate": 4.992544849598824e-05, "loss": 0.5568, "step": 24686 }, { "epoch": 0.02, "learning_rate": 4.9925442434331895e-05, "loss": 0.8523, "step": 24687 }, { "epoch": 0.02, "learning_rate": 4.992543637242949e-05, "loss": 1.1279, "step": 24688 }, { "epoch": 0.02, "learning_rate": 4.992543031028105e-05, "loss": 1.1394, "step": 24689 }, { "epoch": 0.02, "learning_rate": 4.9925424247886544e-05, "loss": 1.688, "step": 24690 }, { "epoch": 0.02, "learning_rate": 4.992541818524599e-05, "loss": 1.5875, "step": 24691 }, { "epoch": 0.02, "learning_rate": 4.992541212235938e-05, "loss": 1.1066, "step": 24692 }, { "epoch": 0.02, "learning_rate": 4.992540605922672e-05, "loss": 1.1852, "step": 24693 }, { "epoch": 0.02, "learning_rate": 4.9925399995848e-05, "loss": 1.0187, "step": 24694 }, { "epoch": 0.02, "learning_rate": 4.992539393222323e-05, "loss": 1.1848, "step": 24695 }, { "epoch": 0.02, "learning_rate": 4.992538786835241e-05, "loss": 1.0521, "step": 24696 }, { "epoch": 0.02, "learning_rate": 4.992538180423554e-05, "loss": 1.072, "step": 24697 }, { "epoch": 0.02, "learning_rate": 4.9925375739872607e-05, "loss": 1.0174, "step": 24698 }, { "epoch": 0.02, "learning_rate": 4.992536967526363e-05, "loss": 0.8966, "step": 24699 }, { "epoch": 0.02, "learning_rate": 4.9925363610408595e-05, "loss": 0.576, "step": 24700 }, { "epoch": 0.02, "learning_rate": 4.992535754530751e-05, "loss": 0.0022, "step": 24701 }, { "epoch": 0.02, "learning_rate": 4.992535147996037e-05, "loss": 0.6625, "step": 24702 }, { "epoch": 0.02, "learning_rate": 4.992534541436718e-05, "loss": 0.0012, "step": 24703 }, { "epoch": 0.02, "learning_rate": 4.9925339348527937e-05, "loss": 0.4065, "step": 24704 }, { "epoch": 0.02, "learning_rate": 4.9925333282442646e-05, "loss": 1.0099, "step": 24705 }, { "epoch": 0.02, "learning_rate": 4.99253272161113e-05, "loss": 0.7672, "step": 24706 }, { "epoch": 0.02, "learning_rate": 4.9925321149533894e-05, "loss": 1.1957, "step": 24707 }, { "epoch": 0.02, "learning_rate": 4.9925315082710445e-05, "loss": 1.0584, "step": 24708 }, { "epoch": 0.02, "learning_rate": 4.992530901564094e-05, "loss": 0.949, "step": 24709 }, { "epoch": 0.02, "learning_rate": 4.992530294832538e-05, "loss": 0.9009, "step": 24710 }, { "epoch": 0.02, "learning_rate": 4.9925296880763775e-05, "loss": 1.4002, "step": 24711 }, { "epoch": 0.02, "learning_rate": 4.992529081295611e-05, "loss": 1.1565, "step": 24712 }, { "epoch": 0.02, "learning_rate": 4.99252847449024e-05, "loss": 0.6269, "step": 24713 }, { "epoch": 0.02, "learning_rate": 4.9925278676602635e-05, "loss": 1.0191, "step": 24714 }, { "epoch": 0.02, "learning_rate": 4.9925272608056816e-05, "loss": 1.0224, "step": 24715 }, { "epoch": 0.02, "learning_rate": 4.992526653926495e-05, "loss": 1.3352, "step": 24716 }, { "epoch": 0.02, "learning_rate": 4.9925260470227026e-05, "loss": 1.1393, "step": 24717 }, { "epoch": 0.02, "learning_rate": 4.9925254400943054e-05, "loss": 1.3045, "step": 24718 }, { "epoch": 0.02, "learning_rate": 4.992524833141303e-05, "loss": 1.1809, "step": 24719 }, { "epoch": 0.02, "learning_rate": 4.9925242261636954e-05, "loss": 1.1072, "step": 24720 }, { "epoch": 0.02, "learning_rate": 4.992523619161483e-05, "loss": 0.8779, "step": 24721 }, { "epoch": 0.02, "learning_rate": 4.992523012134665e-05, "loss": 0.9891, "step": 24722 }, { "epoch": 0.02, "learning_rate": 4.992522405083242e-05, "loss": 1.0986, "step": 24723 }, { "epoch": 0.02, "learning_rate": 4.9925217980072144e-05, "loss": 0.815, "step": 24724 }, { "epoch": 0.02, "learning_rate": 4.99252119090658e-05, "loss": 1.2872, "step": 24725 }, { "epoch": 0.02, "learning_rate": 4.992520583781343e-05, "loss": 0.8402, "step": 24726 }, { "epoch": 0.02, "learning_rate": 4.992519976631499e-05, "loss": 1.0987, "step": 24727 }, { "epoch": 0.02, "learning_rate": 4.99251936945705e-05, "loss": 1.0828, "step": 24728 }, { "epoch": 0.02, "learning_rate": 4.992518762257997e-05, "loss": 0.8079, "step": 24729 }, { "epoch": 0.02, "learning_rate": 4.992518155034338e-05, "loss": 0.6275, "step": 24730 }, { "epoch": 0.02, "learning_rate": 4.9925175477860745e-05, "loss": 1.0216, "step": 24731 }, { "epoch": 0.02, "learning_rate": 4.992516940513205e-05, "loss": 0.8008, "step": 24732 }, { "epoch": 0.02, "learning_rate": 4.9925163332157316e-05, "loss": 1.1504, "step": 24733 }, { "epoch": 0.02, "learning_rate": 4.9925157258936525e-05, "loss": 1.0071, "step": 24734 }, { "epoch": 0.02, "learning_rate": 4.992515118546969e-05, "loss": 1.0931, "step": 24735 }, { "epoch": 0.02, "learning_rate": 4.992514511175679e-05, "loss": 1.1209, "step": 24736 }, { "epoch": 0.02, "learning_rate": 4.992513903779785e-05, "loss": 1.419, "step": 24737 }, { "epoch": 0.02, "learning_rate": 4.9925132963592855e-05, "loss": 2.4437, "step": 24738 }, { "epoch": 0.02, "learning_rate": 4.992512688914181e-05, "loss": 2.3806, "step": 24739 }, { "epoch": 0.02, "learning_rate": 4.992512081444472e-05, "loss": 2.2704, "step": 24740 }, { "epoch": 0.02, "learning_rate": 4.992511473950158e-05, "loss": 2.2244, "step": 24741 }, { "epoch": 0.02, "learning_rate": 4.992510866431238e-05, "loss": 2.1305, "step": 24742 }, { "epoch": 0.02, "learning_rate": 4.992510258887714e-05, "loss": 1.9933, "step": 24743 }, { "epoch": 0.02, "learning_rate": 4.9925096513195846e-05, "loss": 1.8875, "step": 24744 }, { "epoch": 0.02, "learning_rate": 4.99250904372685e-05, "loss": 1.69, "step": 24745 }, { "epoch": 0.02, "learning_rate": 4.992508436109511e-05, "loss": 1.5305, "step": 24746 }, { "epoch": 0.02, "learning_rate": 4.9925078284675665e-05, "loss": 1.503, "step": 24747 }, { "epoch": 0.02, "learning_rate": 4.992507220801017e-05, "loss": 1.4654, "step": 24748 }, { "epoch": 0.02, "learning_rate": 4.9925066131098624e-05, "loss": 1.4004, "step": 24749 }, { "epoch": 0.02, "learning_rate": 4.992506005394103e-05, "loss": 1.4038, "step": 24750 }, { "epoch": 0.02, "learning_rate": 4.9925053976537386e-05, "loss": 1.3233, "step": 24751 }, { "epoch": 0.02, "learning_rate": 4.99250478988877e-05, "loss": 1.3287, "step": 24752 }, { "epoch": 0.02, "learning_rate": 4.992504182099195e-05, "loss": 1.289, "step": 24753 }, { "epoch": 0.02, "learning_rate": 4.9925035742850154e-05, "loss": 1.4036, "step": 24754 }, { "epoch": 0.02, "learning_rate": 4.992502966446232e-05, "loss": 1.286, "step": 24755 }, { "epoch": 0.02, "learning_rate": 4.992502358582842e-05, "loss": 1.4152, "step": 24756 }, { "epoch": 0.02, "learning_rate": 4.992501750694849e-05, "loss": 1.322, "step": 24757 }, { "epoch": 0.02, "learning_rate": 4.99250114278225e-05, "loss": 1.2045, "step": 24758 }, { "epoch": 0.02, "learning_rate": 4.992500534845046e-05, "loss": 1.1986, "step": 24759 }, { "epoch": 0.02, "learning_rate": 4.992499926883237e-05, "loss": 1.16, "step": 24760 }, { "epoch": 0.02, "learning_rate": 4.9924993188968234e-05, "loss": 1.1795, "step": 24761 }, { "epoch": 0.02, "learning_rate": 4.9924987108858043e-05, "loss": 1.238, "step": 24762 }, { "epoch": 0.02, "learning_rate": 4.9924981028501814e-05, "loss": 1.1965, "step": 24763 }, { "epoch": 0.02, "learning_rate": 4.992497494789953e-05, "loss": 0.8114, "step": 24764 }, { "epoch": 0.02, "learning_rate": 4.9924968867051195e-05, "loss": 1.0925, "step": 24765 }, { "epoch": 0.02, "learning_rate": 4.992496278595681e-05, "loss": 0.9681, "step": 24766 }, { "epoch": 0.02, "learning_rate": 4.9924956704616385e-05, "loss": 1.2626, "step": 24767 }, { "epoch": 0.02, "learning_rate": 4.9924950623029904e-05, "loss": 1.1054, "step": 24768 }, { "epoch": 0.02, "learning_rate": 4.992494454119738e-05, "loss": 0.8193, "step": 24769 }, { "epoch": 0.02, "learning_rate": 4.99249384591188e-05, "loss": 1.2712, "step": 24770 }, { "epoch": 0.02, "learning_rate": 4.992493237679418e-05, "loss": 1.3307, "step": 24771 }, { "epoch": 0.02, "learning_rate": 4.9924926294223506e-05, "loss": 0.9741, "step": 24772 }, { "epoch": 0.02, "learning_rate": 4.992492021140678e-05, "loss": 1.1622, "step": 24773 }, { "epoch": 0.02, "learning_rate": 4.9924914128344016e-05, "loss": 1.1095, "step": 24774 }, { "epoch": 0.02, "learning_rate": 4.992490804503519e-05, "loss": 1.3685, "step": 24775 }, { "epoch": 0.02, "learning_rate": 4.992490196148033e-05, "loss": 1.0401, "step": 24776 }, { "epoch": 0.02, "learning_rate": 4.992489587767941e-05, "loss": 1.0276, "step": 24777 }, { "epoch": 0.02, "learning_rate": 4.992488979363245e-05, "loss": 0.583, "step": 24778 }, { "epoch": 0.02, "learning_rate": 4.992488370933944e-05, "loss": 0.9058, "step": 24779 }, { "epoch": 0.02, "learning_rate": 4.992487762480038e-05, "loss": 0.9159, "step": 24780 }, { "epoch": 0.02, "learning_rate": 4.992487154001528e-05, "loss": 1.078, "step": 24781 }, { "epoch": 0.02, "learning_rate": 4.9924865454984125e-05, "loss": 1.3363, "step": 24782 }, { "epoch": 0.02, "learning_rate": 4.9924859369706925e-05, "loss": 1.0105, "step": 24783 }, { "epoch": 0.02, "learning_rate": 4.992485328418367e-05, "loss": 0.9579, "step": 24784 }, { "epoch": 0.02, "learning_rate": 4.992484719841437e-05, "loss": 1.2003, "step": 24785 }, { "epoch": 0.02, "learning_rate": 4.992484111239903e-05, "loss": 1.2073, "step": 24786 }, { "epoch": 0.02, "learning_rate": 4.992483502613764e-05, "loss": 1.2123, "step": 24787 }, { "epoch": 0.02, "learning_rate": 4.99248289396302e-05, "loss": 0.9298, "step": 24788 }, { "epoch": 0.02, "learning_rate": 4.992482285287671e-05, "loss": 1.3311, "step": 24789 }, { "epoch": 0.02, "learning_rate": 4.992481676587717e-05, "loss": 1.2886, "step": 24790 }, { "epoch": 0.02, "learning_rate": 4.99248106786316e-05, "loss": 1.2508, "step": 24791 }, { "epoch": 0.02, "learning_rate": 4.992480459113996e-05, "loss": 0.8291, "step": 24792 }, { "epoch": 0.02, "learning_rate": 4.9924798503402294e-05, "loss": 1.194, "step": 24793 }, { "epoch": 0.02, "learning_rate": 4.9924792415418566e-05, "loss": 1.1389, "step": 24794 }, { "epoch": 0.02, "learning_rate": 4.99247863271888e-05, "loss": 1.5175, "step": 24795 }, { "epoch": 0.02, "learning_rate": 4.992478023871298e-05, "loss": 0.9886, "step": 24796 }, { "epoch": 0.02, "learning_rate": 4.992477414999111e-05, "loss": 1.0344, "step": 24797 }, { "epoch": 0.02, "learning_rate": 4.99247680610232e-05, "loss": 1.4679, "step": 24798 }, { "epoch": 0.02, "learning_rate": 4.992476197180924e-05, "loss": 0.8572, "step": 24799 }, { "epoch": 0.02, "learning_rate": 4.992475588234924e-05, "loss": 0.8238, "step": 24800 }, { "epoch": 0.02, "learning_rate": 4.992474979264319e-05, "loss": 1.404, "step": 24801 }, { "epoch": 0.02, "learning_rate": 4.992474370269109e-05, "loss": 1.3466, "step": 24802 }, { "epoch": 0.02, "learning_rate": 4.992473761249295e-05, "loss": 0.5016, "step": 24803 }, { "epoch": 0.02, "learning_rate": 4.992473152204876e-05, "loss": 1.1906, "step": 24804 }, { "epoch": 0.02, "learning_rate": 4.992472543135852e-05, "loss": 0.546, "step": 24805 }, { "epoch": 0.02, "learning_rate": 4.9924719340422235e-05, "loss": 1.372, "step": 24806 }, { "epoch": 0.02, "learning_rate": 4.992471324923991e-05, "loss": 0.6404, "step": 24807 }, { "epoch": 0.02, "learning_rate": 4.992470715781153e-05, "loss": 0.9759, "step": 24808 }, { "epoch": 0.02, "learning_rate": 4.99247010661371e-05, "loss": 0.7342, "step": 24809 }, { "epoch": 0.02, "learning_rate": 4.9924694974216635e-05, "loss": 0.9617, "step": 24810 }, { "epoch": 0.02, "learning_rate": 4.9924688882050116e-05, "loss": 0.8479, "step": 24811 }, { "epoch": 0.02, "learning_rate": 4.992468278963756e-05, "loss": 1.0129, "step": 24812 }, { "epoch": 0.02, "learning_rate": 4.9924676696978954e-05, "loss": 0.9736, "step": 24813 }, { "epoch": 0.02, "learning_rate": 4.99246706040743e-05, "loss": 1.1784, "step": 24814 }, { "epoch": 0.02, "learning_rate": 4.99246645109236e-05, "loss": 1.0536, "step": 24815 }, { "epoch": 0.02, "learning_rate": 4.992465841752686e-05, "loss": 1.1448, "step": 24816 }, { "epoch": 0.02, "learning_rate": 4.992465232388407e-05, "loss": 0.4279, "step": 24817 }, { "epoch": 0.02, "learning_rate": 4.9924646229995234e-05, "loss": 1.0984, "step": 24818 }, { "epoch": 0.02, "learning_rate": 4.9924640135860354e-05, "loss": 1.0264, "step": 24819 }, { "epoch": 0.02, "learning_rate": 4.9924634041479426e-05, "loss": 1.0276, "step": 24820 }, { "epoch": 0.02, "learning_rate": 4.992462794685245e-05, "loss": 1.2264, "step": 24821 }, { "epoch": 0.02, "learning_rate": 4.9924621851979434e-05, "loss": 0.5969, "step": 24822 }, { "epoch": 0.02, "learning_rate": 4.992461575686037e-05, "loss": 0.6712, "step": 24823 }, { "epoch": 0.02, "learning_rate": 4.992460966149526e-05, "loss": 1.2668, "step": 24824 }, { "epoch": 0.02, "learning_rate": 4.992460356588411e-05, "loss": 1.0837, "step": 24825 }, { "epoch": 0.02, "learning_rate": 4.992459747002691e-05, "loss": 0.9508, "step": 24826 }, { "epoch": 0.02, "learning_rate": 4.9924591373923665e-05, "loss": 0.9767, "step": 24827 }, { "epoch": 0.02, "learning_rate": 4.9924585277574376e-05, "loss": 1.109, "step": 24828 }, { "epoch": 0.02, "learning_rate": 4.992457918097904e-05, "loss": 0.8894, "step": 24829 }, { "epoch": 0.02, "learning_rate": 4.992457308413766e-05, "loss": 1.5066, "step": 24830 }, { "epoch": 0.02, "learning_rate": 4.992456698705024e-05, "loss": 1.1599, "step": 24831 }, { "epoch": 0.02, "learning_rate": 4.9924560889716764e-05, "loss": 1.2186, "step": 24832 }, { "epoch": 0.02, "learning_rate": 4.992455479213726e-05, "loss": 1.219, "step": 24833 }, { "epoch": 0.02, "learning_rate": 4.99245486943117e-05, "loss": 1.1151, "step": 24834 }, { "epoch": 0.02, "learning_rate": 4.99245425962401e-05, "loss": 1.089, "step": 24835 }, { "epoch": 0.02, "learning_rate": 4.992453649792245e-05, "loss": 1.0572, "step": 24836 }, { "epoch": 0.02, "learning_rate": 4.992453039935876e-05, "loss": 1.0204, "step": 24837 }, { "epoch": 0.02, "learning_rate": 4.992452430054902e-05, "loss": 1.1187, "step": 24838 }, { "epoch": 0.02, "learning_rate": 4.9924518201493236e-05, "loss": 0.7574, "step": 24839 }, { "epoch": 0.02, "learning_rate": 4.9924512102191414e-05, "loss": 0.2656, "step": 24840 }, { "epoch": 0.02, "learning_rate": 4.9924506002643545e-05, "loss": 0.2236, "step": 24841 }, { "epoch": 0.02, "learning_rate": 4.992449990284963e-05, "loss": 0.2413, "step": 24842 }, { "epoch": 0.02, "learning_rate": 4.992449380280967e-05, "loss": 0.362, "step": 24843 }, { "epoch": 0.02, "learning_rate": 4.992448770252367e-05, "loss": 0.7748, "step": 24844 }, { "epoch": 0.02, "learning_rate": 4.992448160199162e-05, "loss": 1.1207, "step": 24845 }, { "epoch": 0.02, "learning_rate": 4.992447550121353e-05, "loss": 0.9493, "step": 24846 }, { "epoch": 0.02, "learning_rate": 4.99244694001894e-05, "loss": 0.9208, "step": 24847 }, { "epoch": 0.02, "learning_rate": 4.992446329891922e-05, "loss": 0.9672, "step": 24848 }, { "epoch": 0.02, "learning_rate": 4.9924457197402995e-05, "loss": 0.9119, "step": 24849 }, { "epoch": 0.02, "learning_rate": 4.992445109564073e-05, "loss": 1.0942, "step": 24850 }, { "epoch": 0.02, "learning_rate": 4.992444499363242e-05, "loss": 1.1255, "step": 24851 }, { "epoch": 0.02, "learning_rate": 4.992443889137807e-05, "loss": 0.9548, "step": 24852 }, { "epoch": 0.02, "learning_rate": 4.992443278887767e-05, "loss": 0.989, "step": 24853 }, { "epoch": 0.02, "learning_rate": 4.9924426686131236e-05, "loss": 0.9453, "step": 24854 }, { "epoch": 0.02, "learning_rate": 4.9924420583138756e-05, "loss": 1.0301, "step": 24855 }, { "epoch": 0.02, "learning_rate": 4.992441447990023e-05, "loss": 1.357, "step": 24856 }, { "epoch": 0.02, "learning_rate": 4.9924408376415656e-05, "loss": 0.834, "step": 24857 }, { "epoch": 0.02, "learning_rate": 4.9924402272685044e-05, "loss": 0.9754, "step": 24858 }, { "epoch": 0.02, "learning_rate": 4.9924396168708385e-05, "loss": 1.2009, "step": 24859 }, { "epoch": 0.02, "learning_rate": 4.992439006448569e-05, "loss": 0.9311, "step": 24860 }, { "epoch": 0.02, "learning_rate": 4.9924383960016944e-05, "loss": 0.8317, "step": 24861 }, { "epoch": 0.02, "learning_rate": 4.992437785530216e-05, "loss": 1.2131, "step": 24862 }, { "epoch": 0.02, "learning_rate": 4.992437175034133e-05, "loss": 1.0865, "step": 24863 }, { "epoch": 0.02, "learning_rate": 4.992436564513446e-05, "loss": 1.0464, "step": 24864 }, { "epoch": 0.02, "learning_rate": 4.992435953968155e-05, "loss": 1.1109, "step": 24865 }, { "epoch": 0.02, "learning_rate": 4.992435343398259e-05, "loss": 0.9513, "step": 24866 }, { "epoch": 0.02, "learning_rate": 4.992434732803759e-05, "loss": 1.1218, "step": 24867 }, { "epoch": 0.02, "learning_rate": 4.9924341221846546e-05, "loss": 0.9772, "step": 24868 }, { "epoch": 0.02, "learning_rate": 4.992433511540946e-05, "loss": 1.1315, "step": 24869 }, { "epoch": 0.02, "learning_rate": 4.9924329008726335e-05, "loss": 1.028, "step": 24870 }, { "epoch": 0.02, "learning_rate": 4.9924322901797164e-05, "loss": 1.6592, "step": 24871 }, { "epoch": 0.02, "learning_rate": 4.992431679462195e-05, "loss": 1.175, "step": 24872 }, { "epoch": 0.02, "learning_rate": 4.9924310687200696e-05, "loss": 1.2921, "step": 24873 }, { "epoch": 0.02, "learning_rate": 4.99243045795334e-05, "loss": 1.3849, "step": 24874 }, { "epoch": 0.02, "learning_rate": 4.9924298471620064e-05, "loss": 1.0969, "step": 24875 }, { "epoch": 0.02, "learning_rate": 4.9924292363460676e-05, "loss": 0.4001, "step": 24876 }, { "epoch": 0.02, "learning_rate": 4.9924286255055255e-05, "loss": 0.2367, "step": 24877 }, { "epoch": 0.02, "learning_rate": 4.992428014640379e-05, "loss": 0.657, "step": 24878 }, { "epoch": 0.02, "learning_rate": 4.992427403750629e-05, "loss": 1.054, "step": 24879 }, { "epoch": 0.02, "learning_rate": 4.9924267928362736e-05, "loss": 1.1969, "step": 24880 }, { "epoch": 0.02, "learning_rate": 4.9924261818973144e-05, "loss": 0.9004, "step": 24881 }, { "epoch": 0.02, "learning_rate": 4.9924255709337505e-05, "loss": 0.8162, "step": 24882 }, { "epoch": 0.02, "learning_rate": 4.9924249599455835e-05, "loss": 0.9602, "step": 24883 }, { "epoch": 0.02, "learning_rate": 4.992424348932812e-05, "loss": 1.1458, "step": 24884 }, { "epoch": 0.02, "learning_rate": 4.992423737895436e-05, "loss": 1.1507, "step": 24885 }, { "epoch": 0.02, "learning_rate": 4.9924231268334566e-05, "loss": 1.1043, "step": 24886 }, { "epoch": 0.02, "learning_rate": 4.992422515746872e-05, "loss": 0.9218, "step": 24887 }, { "epoch": 0.02, "learning_rate": 4.992421904635684e-05, "loss": 1.115, "step": 24888 }, { "epoch": 0.02, "learning_rate": 4.9924212934998916e-05, "loss": 1.0292, "step": 24889 }, { "epoch": 0.02, "learning_rate": 4.992420682339495e-05, "loss": 1.059, "step": 24890 }, { "epoch": 0.02, "learning_rate": 4.9924200711544944e-05, "loss": 1.1494, "step": 24891 }, { "epoch": 0.02, "learning_rate": 4.992419459944889e-05, "loss": 1.7477, "step": 24892 }, { "epoch": 0.02, "learning_rate": 4.992418848710681e-05, "loss": 1.6383, "step": 24893 }, { "epoch": 0.02, "learning_rate": 4.992418237451868e-05, "loss": 1.7584, "step": 24894 }, { "epoch": 0.02, "learning_rate": 4.99241762616845e-05, "loss": 0.9766, "step": 24895 }, { "epoch": 0.02, "learning_rate": 4.992417014860429e-05, "loss": 1.1291, "step": 24896 }, { "epoch": 0.02, "learning_rate": 4.9924164035278035e-05, "loss": 1.0837, "step": 24897 }, { "epoch": 0.02, "learning_rate": 4.992415792170575e-05, "loss": 1.2559, "step": 24898 }, { "epoch": 0.02, "learning_rate": 4.992415180788741e-05, "loss": 1.5923, "step": 24899 }, { "epoch": 0.02, "learning_rate": 4.992414569382303e-05, "loss": 1.6998, "step": 24900 }, { "epoch": 0.02, "learning_rate": 4.992413957951262e-05, "loss": 1.0448, "step": 24901 }, { "epoch": 0.02, "learning_rate": 4.992413346495616e-05, "loss": 0.9542, "step": 24902 }, { "epoch": 0.02, "learning_rate": 4.992412735015366e-05, "loss": 1.3388, "step": 24903 }, { "epoch": 0.02, "learning_rate": 4.9924121235105124e-05, "loss": 0.9491, "step": 24904 }, { "epoch": 0.02, "learning_rate": 4.992411511981055e-05, "loss": 0.9821, "step": 24905 }, { "epoch": 0.02, "learning_rate": 4.9924109004269923e-05, "loss": 0.6625, "step": 24906 }, { "epoch": 0.02, "learning_rate": 4.992410288848327e-05, "loss": 1.0136, "step": 24907 }, { "epoch": 0.02, "learning_rate": 4.9924096772450566e-05, "loss": 1.0689, "step": 24908 }, { "epoch": 0.02, "learning_rate": 4.992409065617183e-05, "loss": 1.1115, "step": 24909 }, { "epoch": 0.02, "learning_rate": 4.992408453964705e-05, "loss": 1.0123, "step": 24910 }, { "epoch": 0.02, "learning_rate": 4.992407842287623e-05, "loss": 1.1278, "step": 24911 }, { "epoch": 0.02, "learning_rate": 4.992407230585937e-05, "loss": 1.0213, "step": 24912 }, { "epoch": 0.02, "learning_rate": 4.992406618859647e-05, "loss": 1.0596, "step": 24913 }, { "epoch": 0.02, "learning_rate": 4.9924060071087534e-05, "loss": 1.1565, "step": 24914 }, { "epoch": 0.02, "learning_rate": 4.9924053953332557e-05, "loss": 0.9526, "step": 24915 }, { "epoch": 0.02, "learning_rate": 4.992404783533153e-05, "loss": 1.1842, "step": 24916 }, { "epoch": 0.02, "learning_rate": 4.992404171708448e-05, "loss": 1.0103, "step": 24917 }, { "epoch": 0.02, "learning_rate": 4.992403559859138e-05, "loss": 1.273, "step": 24918 }, { "epoch": 0.02, "learning_rate": 4.992402947985224e-05, "loss": 1.1352, "step": 24919 }, { "epoch": 0.02, "learning_rate": 4.9924023360867065e-05, "loss": 1.0047, "step": 24920 }, { "epoch": 0.02, "learning_rate": 4.9924017241635845e-05, "loss": 1.1872, "step": 24921 }, { "epoch": 0.02, "learning_rate": 4.992401112215859e-05, "loss": 1.0912, "step": 24922 }, { "epoch": 0.02, "learning_rate": 4.99240050024353e-05, "loss": 1.0857, "step": 24923 }, { "epoch": 0.02, "learning_rate": 4.992399888246596e-05, "loss": 0.9572, "step": 24924 }, { "epoch": 0.02, "learning_rate": 4.992399276225059e-05, "loss": 1.0034, "step": 24925 }, { "epoch": 0.02, "learning_rate": 4.992398664178918e-05, "loss": 1.0605, "step": 24926 }, { "epoch": 0.02, "learning_rate": 4.9923980521081723e-05, "loss": 0.9983, "step": 24927 }, { "epoch": 0.02, "learning_rate": 4.9923974400128234e-05, "loss": 1.1913, "step": 24928 }, { "epoch": 0.02, "learning_rate": 4.99239682789287e-05, "loss": 1.0678, "step": 24929 }, { "epoch": 0.02, "learning_rate": 4.992396215748313e-05, "loss": 1.0793, "step": 24930 }, { "epoch": 0.02, "learning_rate": 4.9923956035791524e-05, "loss": 0.9825, "step": 24931 }, { "epoch": 0.02, "learning_rate": 4.992394991385388e-05, "loss": 0.868, "step": 24932 }, { "epoch": 0.02, "learning_rate": 4.992394379167019e-05, "loss": 1.0859, "step": 24933 }, { "epoch": 0.02, "learning_rate": 4.9923937669240474e-05, "loss": 1.1075, "step": 24934 }, { "epoch": 0.02, "learning_rate": 4.992393154656471e-05, "loss": 1.2649, "step": 24935 }, { "epoch": 0.02, "learning_rate": 4.9923925423642905e-05, "loss": 0.6111, "step": 24936 }, { "epoch": 0.02, "learning_rate": 4.992391930047506e-05, "loss": 0.4073, "step": 24937 }, { "epoch": 0.02, "learning_rate": 4.9923913177061186e-05, "loss": 0.514, "step": 24938 }, { "epoch": 0.02, "learning_rate": 4.992390705340127e-05, "loss": 0.5093, "step": 24939 }, { "epoch": 0.02, "learning_rate": 4.9923900929495316e-05, "loss": 0.4732, "step": 24940 }, { "epoch": 0.02, "learning_rate": 4.992389480534332e-05, "loss": 0.3586, "step": 24941 }, { "epoch": 0.02, "learning_rate": 4.992388868094529e-05, "loss": 0.4708, "step": 24942 }, { "epoch": 0.02, "learning_rate": 4.992388255630122e-05, "loss": 0.4855, "step": 24943 }, { "epoch": 0.02, "learning_rate": 4.992387643141112e-05, "loss": 0.5097, "step": 24944 }, { "epoch": 0.02, "learning_rate": 4.992387030627497e-05, "loss": 0.4146, "step": 24945 }, { "epoch": 0.02, "learning_rate": 4.992386418089279e-05, "loss": 0.2928, "step": 24946 }, { "epoch": 0.02, "learning_rate": 4.992385805526456e-05, "loss": 0.2805, "step": 24947 }, { "epoch": 0.02, "learning_rate": 4.992385192939031e-05, "loss": 0.301, "step": 24948 }, { "epoch": 0.02, "learning_rate": 4.992384580327001e-05, "loss": 0.2897, "step": 24949 }, { "epoch": 0.02, "learning_rate": 4.9923839676903674e-05, "loss": 0.2696, "step": 24950 }, { "epoch": 0.02, "learning_rate": 4.99238335502913e-05, "loss": 0.2173, "step": 24951 }, { "epoch": 0.02, "learning_rate": 4.992382742343289e-05, "loss": 0.2524, "step": 24952 }, { "epoch": 0.02, "learning_rate": 4.992382129632844e-05, "loss": 0.189, "step": 24953 }, { "epoch": 0.02, "learning_rate": 4.992381516897796e-05, "loss": 0.2358, "step": 24954 }, { "epoch": 0.02, "learning_rate": 4.992380904138144e-05, "loss": 0.2233, "step": 24955 }, { "epoch": 0.02, "learning_rate": 4.9923802913538874e-05, "loss": 0.2514, "step": 24956 }, { "epoch": 0.02, "learning_rate": 4.9923796785450285e-05, "loss": 0.7897, "step": 24957 }, { "epoch": 0.02, "learning_rate": 4.992379065711564e-05, "loss": 1.0508, "step": 24958 }, { "epoch": 0.02, "learning_rate": 4.992378452853498e-05, "loss": 0.9697, "step": 24959 }, { "epoch": 0.02, "learning_rate": 4.9923778399708266e-05, "loss": 0.8024, "step": 24960 }, { "epoch": 0.02, "learning_rate": 4.992377227063552e-05, "loss": 1.1697, "step": 24961 }, { "epoch": 0.02, "learning_rate": 4.992376614131674e-05, "loss": 1.3325, "step": 24962 }, { "epoch": 0.02, "learning_rate": 4.992376001175192e-05, "loss": 1.0017, "step": 24963 }, { "epoch": 0.02, "learning_rate": 4.992375388194107e-05, "loss": 0.8457, "step": 24964 }, { "epoch": 0.02, "learning_rate": 4.992374775188417e-05, "loss": 1.2833, "step": 24965 }, { "epoch": 0.02, "learning_rate": 4.992374162158124e-05, "loss": 1.2185, "step": 24966 }, { "epoch": 0.02, "learning_rate": 4.992373549103228e-05, "loss": 1.0207, "step": 24967 }, { "epoch": 0.02, "learning_rate": 4.992372936023727e-05, "loss": 1.1256, "step": 24968 }, { "epoch": 0.02, "learning_rate": 4.9923723229196234e-05, "loss": 1.2243, "step": 24969 }, { "epoch": 0.02, "learning_rate": 4.992371709790916e-05, "loss": 1.1101, "step": 24970 }, { "epoch": 0.02, "learning_rate": 4.9923710966376046e-05, "loss": 0.9294, "step": 24971 }, { "epoch": 0.02, "learning_rate": 4.992370483459689e-05, "loss": 0.9307, "step": 24972 }, { "epoch": 0.02, "learning_rate": 4.992369870257171e-05, "loss": 0.9323, "step": 24973 }, { "epoch": 0.02, "learning_rate": 4.992369257030049e-05, "loss": 0.9699, "step": 24974 }, { "epoch": 0.02, "learning_rate": 4.992368643778323e-05, "loss": 1.0039, "step": 24975 }, { "epoch": 0.02, "learning_rate": 4.9923680305019935e-05, "loss": 1.1591, "step": 24976 }, { "epoch": 0.02, "learning_rate": 4.992367417201061e-05, "loss": 0.9625, "step": 24977 }, { "epoch": 0.02, "learning_rate": 4.9923668038755244e-05, "loss": 1.2073, "step": 24978 }, { "epoch": 0.02, "learning_rate": 4.9923661905253836e-05, "loss": 1.2866, "step": 24979 }, { "epoch": 0.02, "learning_rate": 4.992365577150641e-05, "loss": 1.1023, "step": 24980 }, { "epoch": 0.02, "learning_rate": 4.992364963751293e-05, "loss": 1.2063, "step": 24981 }, { "epoch": 0.02, "learning_rate": 4.992364350327342e-05, "loss": 1.1871, "step": 24982 }, { "epoch": 0.02, "learning_rate": 4.992363736878788e-05, "loss": 1.288, "step": 24983 }, { "epoch": 0.02, "learning_rate": 4.9923631234056293e-05, "loss": 0.7788, "step": 24984 }, { "epoch": 0.02, "learning_rate": 4.992362509907868e-05, "loss": 0.5713, "step": 24985 }, { "epoch": 0.02, "learning_rate": 4.992361896385503e-05, "loss": 1.1372, "step": 24986 }, { "epoch": 0.02, "learning_rate": 4.992361282838534e-05, "loss": 1.0635, "step": 24987 }, { "epoch": 0.02, "learning_rate": 4.992360669266962e-05, "loss": 1.2737, "step": 24988 }, { "epoch": 0.02, "learning_rate": 4.992360055670786e-05, "loss": 1.0093, "step": 24989 }, { "epoch": 0.02, "learning_rate": 4.992359442050006e-05, "loss": 0.8532, "step": 24990 }, { "epoch": 0.02, "learning_rate": 4.992358828404624e-05, "loss": 1.0843, "step": 24991 }, { "epoch": 0.02, "learning_rate": 4.992358214734638e-05, "loss": 1.3212, "step": 24992 }, { "epoch": 0.02, "learning_rate": 4.992357601040047e-05, "loss": 1.1594, "step": 24993 }, { "epoch": 0.02, "learning_rate": 4.992356987320854e-05, "loss": 0.9182, "step": 24994 }, { "epoch": 0.02, "learning_rate": 4.992356373577057e-05, "loss": 1.2352, "step": 24995 }, { "epoch": 0.02, "learning_rate": 4.992355759808657e-05, "loss": 1.2275, "step": 24996 }, { "epoch": 0.02, "learning_rate": 4.9923551460156534e-05, "loss": 1.3346, "step": 24997 }, { "epoch": 0.02, "learning_rate": 4.992354532198046e-05, "loss": 1.5031, "step": 24998 }, { "epoch": 0.02, "learning_rate": 4.992353918355835e-05, "loss": 1.3621, "step": 24999 }, { "epoch": 0.03, "learning_rate": 4.9923533044890206e-05, "loss": 0.9476, "step": 25000 }, { "epoch": 0.03, "eval_loss": 1.0197904109954834, "eval_runtime": 85.2004, "eval_samples_per_second": 16.256, "eval_steps_per_second": 4.073, "step": 25000 }, { "epoch": 0.03, "learning_rate": 4.9923526905976034e-05, "loss": 0.8347, "step": 25001 }, { "epoch": 0.03, "learning_rate": 4.992352076681582e-05, "loss": 0.817, "step": 25002 }, { "epoch": 0.03, "learning_rate": 4.992351462740957e-05, "loss": 1.15, "step": 25003 }, { "epoch": 0.03, "learning_rate": 4.9923508487757295e-05, "loss": 0.6978, "step": 25004 }, { "epoch": 0.03, "learning_rate": 4.992350234785897e-05, "loss": 0.5877, "step": 25005 }, { "epoch": 0.03, "learning_rate": 4.992349620771462e-05, "loss": 1.0021, "step": 25006 }, { "epoch": 0.03, "learning_rate": 4.992349006732424e-05, "loss": 0.9068, "step": 25007 }, { "epoch": 0.03, "learning_rate": 4.992348392668783e-05, "loss": 1.0197, "step": 25008 }, { "epoch": 0.03, "learning_rate": 4.9923477785805375e-05, "loss": 0.8391, "step": 25009 }, { "epoch": 0.03, "learning_rate": 4.992347164467689e-05, "loss": 1.0577, "step": 25010 }, { "epoch": 0.03, "learning_rate": 4.992346550330237e-05, "loss": 0.8527, "step": 25011 }, { "epoch": 0.03, "learning_rate": 4.992345936168181e-05, "loss": 0.649, "step": 25012 }, { "epoch": 0.03, "learning_rate": 4.992345321981523e-05, "loss": 0.8938, "step": 25013 }, { "epoch": 0.03, "learning_rate": 4.992344707770261e-05, "loss": 0.7799, "step": 25014 }, { "epoch": 0.03, "learning_rate": 4.9923440935343945e-05, "loss": 0.9586, "step": 25015 }, { "epoch": 0.03, "learning_rate": 4.992343479273926e-05, "loss": 1.0116, "step": 25016 }, { "epoch": 0.03, "learning_rate": 4.9923428649888536e-05, "loss": 1.134, "step": 25017 }, { "epoch": 0.03, "learning_rate": 4.992342250679178e-05, "loss": 1.0014, "step": 25018 }, { "epoch": 0.03, "learning_rate": 4.992341636344899e-05, "loss": 0.7448, "step": 25019 }, { "epoch": 0.03, "learning_rate": 4.9923410219860166e-05, "loss": 0.7103, "step": 25020 }, { "epoch": 0.03, "learning_rate": 4.992340407602531e-05, "loss": 0.8879, "step": 25021 }, { "epoch": 0.03, "learning_rate": 4.992339793194442e-05, "loss": 0.5409, "step": 25022 }, { "epoch": 0.03, "learning_rate": 4.9923391787617495e-05, "loss": 0.7493, "step": 25023 }, { "epoch": 0.03, "learning_rate": 4.992338564304454e-05, "loss": 1.0124, "step": 25024 }, { "epoch": 0.03, "learning_rate": 4.992337949822555e-05, "loss": 1.0126, "step": 25025 }, { "epoch": 0.03, "learning_rate": 4.992337335316053e-05, "loss": 0.9065, "step": 25026 }, { "epoch": 0.03, "learning_rate": 4.992336720784947e-05, "loss": 1.2448, "step": 25027 }, { "epoch": 0.03, "learning_rate": 4.9923361062292386e-05, "loss": 1.074, "step": 25028 }, { "epoch": 0.03, "learning_rate": 4.992335491648926e-05, "loss": 0.7659, "step": 25029 }, { "epoch": 0.03, "learning_rate": 4.99233487704401e-05, "loss": 0.606, "step": 25030 }, { "epoch": 0.03, "learning_rate": 4.992334262414492e-05, "loss": 1.0034, "step": 25031 }, { "epoch": 0.03, "learning_rate": 4.99233364776037e-05, "loss": 0.945, "step": 25032 }, { "epoch": 0.03, "learning_rate": 4.992333033081644e-05, "loss": 1.3674, "step": 25033 }, { "epoch": 0.03, "learning_rate": 4.992332418378316e-05, "loss": 1.0743, "step": 25034 }, { "epoch": 0.03, "learning_rate": 4.992331803650384e-05, "loss": 1.2407, "step": 25035 }, { "epoch": 0.03, "learning_rate": 4.9923311888978493e-05, "loss": 1.0378, "step": 25036 }, { "epoch": 0.03, "learning_rate": 4.992330574120711e-05, "loss": 1.072, "step": 25037 }, { "epoch": 0.03, "learning_rate": 4.9923299593189696e-05, "loss": 0.3395, "step": 25038 }, { "epoch": 0.03, "learning_rate": 4.9923293444926246e-05, "loss": 0.2991, "step": 25039 }, { "epoch": 0.03, "learning_rate": 4.992328729641676e-05, "loss": 0.35, "step": 25040 }, { "epoch": 0.03, "learning_rate": 4.9923281147661246e-05, "loss": 0.3478, "step": 25041 }, { "epoch": 0.03, "learning_rate": 4.992327499865971e-05, "loss": 1.1695, "step": 25042 }, { "epoch": 0.03, "learning_rate": 4.992326884941213e-05, "loss": 0.989, "step": 25043 }, { "epoch": 0.03, "learning_rate": 4.992326269991853e-05, "loss": 1.0827, "step": 25044 }, { "epoch": 0.03, "learning_rate": 4.992325655017889e-05, "loss": 1.0255, "step": 25045 }, { "epoch": 0.03, "learning_rate": 4.992325040019321e-05, "loss": 1.1286, "step": 25046 }, { "epoch": 0.03, "learning_rate": 4.992324424996151e-05, "loss": 0.7944, "step": 25047 }, { "epoch": 0.03, "learning_rate": 4.992323809948377e-05, "loss": 0.8027, "step": 25048 }, { "epoch": 0.03, "learning_rate": 4.9923231948760007e-05, "loss": 1.2326, "step": 25049 }, { "epoch": 0.03, "learning_rate": 4.9923225797790204e-05, "loss": 1.2462, "step": 25050 }, { "epoch": 0.03, "learning_rate": 4.992321964657437e-05, "loss": 0.7767, "step": 25051 }, { "epoch": 0.03, "learning_rate": 4.992321349511251e-05, "loss": 1.1514, "step": 25052 }, { "epoch": 0.03, "learning_rate": 4.992320734340462e-05, "loss": 0.8306, "step": 25053 }, { "epoch": 0.03, "learning_rate": 4.9923201191450696e-05, "loss": 1.2511, "step": 25054 }, { "epoch": 0.03, "learning_rate": 4.992319503925074e-05, "loss": 1.0314, "step": 25055 }, { "epoch": 0.03, "learning_rate": 4.992318888680475e-05, "loss": 1.0569, "step": 25056 }, { "epoch": 0.03, "learning_rate": 4.992318273411273e-05, "loss": 1.1128, "step": 25057 }, { "epoch": 0.03, "learning_rate": 4.992317658117468e-05, "loss": 1.0062, "step": 25058 }, { "epoch": 0.03, "learning_rate": 4.99231704279906e-05, "loss": 0.9499, "step": 25059 }, { "epoch": 0.03, "learning_rate": 4.992316427456048e-05, "loss": 0.8981, "step": 25060 }, { "epoch": 0.03, "learning_rate": 4.992315812088434e-05, "loss": 0.9923, "step": 25061 }, { "epoch": 0.03, "learning_rate": 4.992315196696217e-05, "loss": 0.6737, "step": 25062 }, { "epoch": 0.03, "learning_rate": 4.9923145812793964e-05, "loss": 0.3192, "step": 25063 }, { "epoch": 0.03, "learning_rate": 4.9923139658379726e-05, "loss": 1.0879, "step": 25064 }, { "epoch": 0.03, "learning_rate": 4.992313350371947e-05, "loss": 1.1829, "step": 25065 }, { "epoch": 0.03, "learning_rate": 4.9923127348813166e-05, "loss": 0.9678, "step": 25066 }, { "epoch": 0.03, "learning_rate": 4.9923121193660836e-05, "loss": 0.8177, "step": 25067 }, { "epoch": 0.03, "learning_rate": 4.992311503826248e-05, "loss": 0.7637, "step": 25068 }, { "epoch": 0.03, "learning_rate": 4.9923108882618094e-05, "loss": 1.0813, "step": 25069 }, { "epoch": 0.03, "learning_rate": 4.9923102726727674e-05, "loss": 0.9057, "step": 25070 }, { "epoch": 0.03, "learning_rate": 4.992309657059122e-05, "loss": 0.8536, "step": 25071 }, { "epoch": 0.03, "learning_rate": 4.992309041420874e-05, "loss": 1.1515, "step": 25072 }, { "epoch": 0.03, "learning_rate": 4.992308425758023e-05, "loss": 0.991, "step": 25073 }, { "epoch": 0.03, "learning_rate": 4.992307810070569e-05, "loss": 1.0317, "step": 25074 }, { "epoch": 0.03, "learning_rate": 4.992307194358512e-05, "loss": 0.9014, "step": 25075 }, { "epoch": 0.03, "learning_rate": 4.9923065786218523e-05, "loss": 1.2781, "step": 25076 }, { "epoch": 0.03, "learning_rate": 4.992305962860589e-05, "loss": 0.8905, "step": 25077 }, { "epoch": 0.03, "learning_rate": 4.992305347074723e-05, "loss": 0.8441, "step": 25078 }, { "epoch": 0.03, "learning_rate": 4.9923047312642536e-05, "loss": 0.8423, "step": 25079 }, { "epoch": 0.03, "learning_rate": 4.992304115429182e-05, "loss": 1.0254, "step": 25080 }, { "epoch": 0.03, "learning_rate": 4.9923034995695065e-05, "loss": 1.3564, "step": 25081 }, { "epoch": 0.03, "learning_rate": 4.9923028836852295e-05, "loss": 1.0223, "step": 25082 }, { "epoch": 0.03, "learning_rate": 4.992302267776348e-05, "loss": 0.9651, "step": 25083 }, { "epoch": 0.03, "learning_rate": 4.992301651842864e-05, "loss": 0.6304, "step": 25084 }, { "epoch": 0.03, "learning_rate": 4.992301035884777e-05, "loss": 0.7032, "step": 25085 }, { "epoch": 0.03, "learning_rate": 4.9923004199020874e-05, "loss": 1.0789, "step": 25086 }, { "epoch": 0.03, "learning_rate": 4.992299803894794e-05, "loss": 1.3526, "step": 25087 }, { "epoch": 0.03, "learning_rate": 4.992299187862899e-05, "loss": 1.0528, "step": 25088 }, { "epoch": 0.03, "learning_rate": 4.992298571806401e-05, "loss": 0.5238, "step": 25089 }, { "epoch": 0.03, "learning_rate": 4.992297955725299e-05, "loss": 0.2522, "step": 25090 }, { "epoch": 0.03, "learning_rate": 4.992297339619595e-05, "loss": 0.7892, "step": 25091 }, { "epoch": 0.03, "learning_rate": 4.992296723489287e-05, "loss": 1.1137, "step": 25092 }, { "epoch": 0.03, "learning_rate": 4.9922961073343766e-05, "loss": 1.2157, "step": 25093 }, { "epoch": 0.03, "learning_rate": 4.992295491154864e-05, "loss": 1.3797, "step": 25094 }, { "epoch": 0.03, "learning_rate": 4.992294874950748e-05, "loss": 0.957, "step": 25095 }, { "epoch": 0.03, "learning_rate": 4.992294258722029e-05, "loss": 0.7168, "step": 25096 }, { "epoch": 0.03, "learning_rate": 4.9922936424687074e-05, "loss": 1.0018, "step": 25097 }, { "epoch": 0.03, "learning_rate": 4.9922930261907823e-05, "loss": 0.8612, "step": 25098 }, { "epoch": 0.03, "learning_rate": 4.9922924098882553e-05, "loss": 0.8657, "step": 25099 }, { "epoch": 0.03, "learning_rate": 4.9922917935611244e-05, "loss": 1.0994, "step": 25100 }, { "epoch": 0.03, "learning_rate": 4.9922911772093916e-05, "loss": 0.9604, "step": 25101 }, { "epoch": 0.03, "learning_rate": 4.992290560833055e-05, "loss": 0.9985, "step": 25102 }, { "epoch": 0.03, "learning_rate": 4.992289944432116e-05, "loss": 1.256, "step": 25103 }, { "epoch": 0.03, "learning_rate": 4.992289328006575e-05, "loss": 0.946, "step": 25104 }, { "epoch": 0.03, "learning_rate": 4.99228871155643e-05, "loss": 0.9306, "step": 25105 }, { "epoch": 0.03, "learning_rate": 4.9922880950816826e-05, "loss": 1.3531, "step": 25106 }, { "epoch": 0.03, "learning_rate": 4.992287478582333e-05, "loss": 1.1832, "step": 25107 }, { "epoch": 0.03, "learning_rate": 4.992286862058379e-05, "loss": 1.0711, "step": 25108 }, { "epoch": 0.03, "learning_rate": 4.992286245509824e-05, "loss": 0.7556, "step": 25109 }, { "epoch": 0.03, "learning_rate": 4.992285628936665e-05, "loss": 0.8984, "step": 25110 }, { "epoch": 0.03, "learning_rate": 4.992285012338903e-05, "loss": 0.9954, "step": 25111 }, { "epoch": 0.03, "learning_rate": 4.992284395716539e-05, "loss": 0.5796, "step": 25112 }, { "epoch": 0.03, "learning_rate": 4.992283779069572e-05, "loss": 0.536, "step": 25113 }, { "epoch": 0.03, "learning_rate": 4.992283162398003e-05, "loss": 0.892, "step": 25114 }, { "epoch": 0.03, "learning_rate": 4.99228254570183e-05, "loss": 0.9259, "step": 25115 }, { "epoch": 0.03, "learning_rate": 4.992281928981054e-05, "loss": 0.883, "step": 25116 }, { "epoch": 0.03, "learning_rate": 4.992281312235676e-05, "loss": 0.886, "step": 25117 }, { "epoch": 0.03, "learning_rate": 4.992280695465695e-05, "loss": 1.0124, "step": 25118 }, { "epoch": 0.03, "learning_rate": 4.992280078671112e-05, "loss": 0.9801, "step": 25119 }, { "epoch": 0.03, "learning_rate": 4.992279461851925e-05, "loss": 0.8232, "step": 25120 }, { "epoch": 0.03, "learning_rate": 4.992278845008137e-05, "loss": 0.8465, "step": 25121 }, { "epoch": 0.03, "learning_rate": 4.992278228139745e-05, "loss": 1.2903, "step": 25122 }, { "epoch": 0.03, "learning_rate": 4.992277611246751e-05, "loss": 0.994, "step": 25123 }, { "epoch": 0.03, "learning_rate": 4.992276994329153e-05, "loss": 1.0585, "step": 25124 }, { "epoch": 0.03, "learning_rate": 4.992276377386953e-05, "loss": 0.9633, "step": 25125 }, { "epoch": 0.03, "learning_rate": 4.992275760420151e-05, "loss": 0.9627, "step": 25126 }, { "epoch": 0.03, "learning_rate": 4.992275143428745e-05, "loss": 0.5739, "step": 25127 }, { "epoch": 0.03, "learning_rate": 4.992274526412737e-05, "loss": 1.0005, "step": 25128 }, { "epoch": 0.03, "learning_rate": 4.9922739093721264e-05, "loss": 0.9775, "step": 25129 }, { "epoch": 0.03, "learning_rate": 4.992273292306913e-05, "loss": 1.1425, "step": 25130 }, { "epoch": 0.03, "learning_rate": 4.992272675217097e-05, "loss": 1.1003, "step": 25131 }, { "epoch": 0.03, "learning_rate": 4.992272058102678e-05, "loss": 1.1229, "step": 25132 }, { "epoch": 0.03, "learning_rate": 4.992271440963657e-05, "loss": 1.3476, "step": 25133 }, { "epoch": 0.03, "learning_rate": 4.9922708238000325e-05, "loss": 1.0289, "step": 25134 }, { "epoch": 0.03, "learning_rate": 4.9922702066118065e-05, "loss": 0.832, "step": 25135 }, { "epoch": 0.03, "learning_rate": 4.9922695893989766e-05, "loss": 1.027, "step": 25136 }, { "epoch": 0.03, "learning_rate": 4.9922689721615454e-05, "loss": 0.9691, "step": 25137 }, { "epoch": 0.03, "learning_rate": 4.992268354899511e-05, "loss": 0.4913, "step": 25138 }, { "epoch": 0.03, "learning_rate": 4.9922677376128734e-05, "loss": 0.5484, "step": 25139 }, { "epoch": 0.03, "learning_rate": 4.992267120301634e-05, "loss": 0.3011, "step": 25140 }, { "epoch": 0.03, "learning_rate": 4.992266502965791e-05, "loss": 0.2391, "step": 25141 }, { "epoch": 0.03, "learning_rate": 4.992265885605346e-05, "loss": 0.2796, "step": 25142 }, { "epoch": 0.03, "learning_rate": 4.992265268220298e-05, "loss": 0.2041, "step": 25143 }, { "epoch": 0.03, "learning_rate": 4.9922646508106484e-05, "loss": 0.2022, "step": 25144 }, { "epoch": 0.03, "learning_rate": 4.992264033376395e-05, "loss": 0.2846, "step": 25145 }, { "epoch": 0.03, "learning_rate": 4.9922634159175395e-05, "loss": 0.1546, "step": 25146 }, { "epoch": 0.03, "learning_rate": 4.992262798434082e-05, "loss": 0.1229, "step": 25147 }, { "epoch": 0.03, "learning_rate": 4.992262180926022e-05, "loss": 0.1206, "step": 25148 }, { "epoch": 0.03, "learning_rate": 4.992261563393358e-05, "loss": 0.0549, "step": 25149 }, { "epoch": 0.03, "learning_rate": 4.992260945836092e-05, "loss": 0.0559, "step": 25150 }, { "epoch": 0.03, "learning_rate": 4.992260328254225e-05, "loss": 0.0581, "step": 25151 }, { "epoch": 0.03, "learning_rate": 4.9922597106477544e-05, "loss": 0.056, "step": 25152 }, { "epoch": 0.03, "learning_rate": 4.9922590930166807e-05, "loss": 0.0592, "step": 25153 }, { "epoch": 0.03, "learning_rate": 4.992258475361005e-05, "loss": 0.0559, "step": 25154 }, { "epoch": 0.03, "learning_rate": 4.992257857680727e-05, "loss": 0.0883, "step": 25155 }, { "epoch": 0.03, "learning_rate": 4.992257239975845e-05, "loss": 0.0735, "step": 25156 }, { "epoch": 0.03, "learning_rate": 4.9922566222463624e-05, "loss": 0.0754, "step": 25157 }, { "epoch": 0.03, "learning_rate": 4.9922560044922764e-05, "loss": 0.0941, "step": 25158 }, { "epoch": 0.03, "learning_rate": 4.992255386713588e-05, "loss": 0.0678, "step": 25159 }, { "epoch": 0.03, "learning_rate": 4.9922547689102974e-05, "loss": 0.0667, "step": 25160 }, { "epoch": 0.03, "learning_rate": 4.9922541510824036e-05, "loss": 0.0911, "step": 25161 }, { "epoch": 0.03, "learning_rate": 4.992253533229908e-05, "loss": 0.07, "step": 25162 }, { "epoch": 0.03, "learning_rate": 4.99225291535281e-05, "loss": 0.0411, "step": 25163 }, { "epoch": 0.03, "learning_rate": 4.992252297451109e-05, "loss": 0.0406, "step": 25164 }, { "epoch": 0.03, "learning_rate": 4.9922516795248056e-05, "loss": 0.0354, "step": 25165 }, { "epoch": 0.03, "learning_rate": 4.9922510615738996e-05, "loss": 0.0281, "step": 25166 }, { "epoch": 0.03, "learning_rate": 4.992250443598392e-05, "loss": 0.0319, "step": 25167 }, { "epoch": 0.03, "learning_rate": 4.9922498255982806e-05, "loss": 0.0286, "step": 25168 }, { "epoch": 0.03, "learning_rate": 4.992249207573568e-05, "loss": 0.0182, "step": 25169 }, { "epoch": 0.03, "learning_rate": 4.9922485895242526e-05, "loss": 0.0228, "step": 25170 }, { "epoch": 0.03, "learning_rate": 4.9922479714503344e-05, "loss": 0.0253, "step": 25171 }, { "epoch": 0.03, "learning_rate": 4.992247353351814e-05, "loss": 0.0278, "step": 25172 }, { "epoch": 0.03, "learning_rate": 4.992246735228692e-05, "loss": 0.0333, "step": 25173 }, { "epoch": 0.03, "learning_rate": 4.9922461170809664e-05, "loss": 0.0326, "step": 25174 }, { "epoch": 0.03, "learning_rate": 4.9922454989086386e-05, "loss": 0.0319, "step": 25175 }, { "epoch": 0.03, "learning_rate": 4.992244880711709e-05, "loss": 0.0176, "step": 25176 }, { "epoch": 0.03, "learning_rate": 4.9922442624901765e-05, "loss": 0.0349, "step": 25177 }, { "epoch": 0.03, "learning_rate": 4.9922436442440416e-05, "loss": 0.0359, "step": 25178 }, { "epoch": 0.03, "learning_rate": 4.992243025973304e-05, "loss": 0.0339, "step": 25179 }, { "epoch": 0.03, "learning_rate": 4.992242407677965e-05, "loss": 0.0321, "step": 25180 }, { "epoch": 0.03, "learning_rate": 4.992241789358023e-05, "loss": 0.0295, "step": 25181 }, { "epoch": 0.03, "learning_rate": 4.992241171013479e-05, "loss": 0.0379, "step": 25182 }, { "epoch": 0.03, "learning_rate": 4.9922405526443325e-05, "loss": 0.0321, "step": 25183 }, { "epoch": 0.03, "learning_rate": 4.9922399342505835e-05, "loss": 0.0299, "step": 25184 }, { "epoch": 0.03, "learning_rate": 4.9922393158322326e-05, "loss": 0.0318, "step": 25185 }, { "epoch": 0.03, "learning_rate": 4.992238697389279e-05, "loss": 0.0329, "step": 25186 }, { "epoch": 0.03, "learning_rate": 4.992238078921723e-05, "loss": 0.0286, "step": 25187 }, { "epoch": 0.03, "learning_rate": 4.992237460429564e-05, "loss": 0.0297, "step": 25188 }, { "epoch": 0.03, "learning_rate": 4.992236841912804e-05, "loss": 0.0252, "step": 25189 }, { "epoch": 0.03, "learning_rate": 4.992236223371441e-05, "loss": 0.1481, "step": 25190 }, { "epoch": 0.03, "learning_rate": 4.992235604805476e-05, "loss": 1.1247, "step": 25191 }, { "epoch": 0.03, "learning_rate": 4.9922349862149086e-05, "loss": 0.8362, "step": 25192 }, { "epoch": 0.03, "learning_rate": 4.9922343675997384e-05, "loss": 1.1324, "step": 25193 }, { "epoch": 0.03, "learning_rate": 4.992233748959967e-05, "loss": 1.1421, "step": 25194 }, { "epoch": 0.03, "learning_rate": 4.992233130295592e-05, "loss": 0.9913, "step": 25195 }, { "epoch": 0.03, "learning_rate": 4.9922325116066156e-05, "loss": 1.0372, "step": 25196 }, { "epoch": 0.03, "learning_rate": 4.9922318928930365e-05, "loss": 1.0103, "step": 25197 }, { "epoch": 0.03, "learning_rate": 4.9922312741548554e-05, "loss": 0.7789, "step": 25198 }, { "epoch": 0.03, "learning_rate": 4.9922306553920724e-05, "loss": 0.9592, "step": 25199 }, { "epoch": 0.03, "learning_rate": 4.992230036604686e-05, "loss": 1.1377, "step": 25200 }, { "epoch": 0.03, "learning_rate": 4.992229417792699e-05, "loss": 1.3155, "step": 25201 }, { "epoch": 0.03, "learning_rate": 4.9922287989561087e-05, "loss": 0.858, "step": 25202 }, { "epoch": 0.03, "learning_rate": 4.992228180094916e-05, "loss": 0.8514, "step": 25203 }, { "epoch": 0.03, "learning_rate": 4.9922275612091215e-05, "loss": 0.9601, "step": 25204 }, { "epoch": 0.03, "learning_rate": 4.9922269422987244e-05, "loss": 0.7068, "step": 25205 }, { "epoch": 0.03, "learning_rate": 4.9922263233637254e-05, "loss": 0.6542, "step": 25206 }, { "epoch": 0.03, "learning_rate": 4.9922257044041245e-05, "loss": 0.6096, "step": 25207 }, { "epoch": 0.03, "learning_rate": 4.992225085419921e-05, "loss": 0.6577, "step": 25208 }, { "epoch": 0.03, "learning_rate": 4.992224466411115e-05, "loss": 0.8597, "step": 25209 }, { "epoch": 0.03, "learning_rate": 4.992223847377707e-05, "loss": 0.5305, "step": 25210 }, { "epoch": 0.03, "learning_rate": 4.992223228319697e-05, "loss": 0.5249, "step": 25211 }, { "epoch": 0.03, "learning_rate": 4.992222609237085e-05, "loss": 1.3521, "step": 25212 }, { "epoch": 0.03, "learning_rate": 4.99222199012987e-05, "loss": 1.0136, "step": 25213 }, { "epoch": 0.03, "learning_rate": 4.992221370998054e-05, "loss": 1.115, "step": 25214 }, { "epoch": 0.03, "learning_rate": 4.992220751841635e-05, "loss": 1.0156, "step": 25215 }, { "epoch": 0.03, "learning_rate": 4.9922201326606144e-05, "loss": 1.0157, "step": 25216 }, { "epoch": 0.03, "learning_rate": 4.992219513454992e-05, "loss": 1.0533, "step": 25217 }, { "epoch": 0.03, "learning_rate": 4.9922188942247665e-05, "loss": 1.0771, "step": 25218 }, { "epoch": 0.03, "learning_rate": 4.99221827496994e-05, "loss": 0.9614, "step": 25219 }, { "epoch": 0.03, "learning_rate": 4.99221765569051e-05, "loss": 1.1284, "step": 25220 }, { "epoch": 0.03, "learning_rate": 4.992217036386479e-05, "loss": 1.3159, "step": 25221 }, { "epoch": 0.03, "learning_rate": 4.9922164170578445e-05, "loss": 0.8393, "step": 25222 }, { "epoch": 0.03, "learning_rate": 4.992215797704609e-05, "loss": 0.5453, "step": 25223 }, { "epoch": 0.03, "learning_rate": 4.992215178326771e-05, "loss": 1.0581, "step": 25224 }, { "epoch": 0.03, "learning_rate": 4.992214558924331e-05, "loss": 1.0304, "step": 25225 }, { "epoch": 0.03, "learning_rate": 4.992213939497289e-05, "loss": 1.1248, "step": 25226 }, { "epoch": 0.03, "learning_rate": 4.992213320045645e-05, "loss": 0.6368, "step": 25227 }, { "epoch": 0.03, "learning_rate": 4.992212700569399e-05, "loss": 0.8258, "step": 25228 }, { "epoch": 0.03, "learning_rate": 4.99221208106855e-05, "loss": 1.0213, "step": 25229 }, { "epoch": 0.03, "learning_rate": 4.9922114615431005e-05, "loss": 1.1141, "step": 25230 }, { "epoch": 0.03, "learning_rate": 4.992210841993048e-05, "loss": 1.1015, "step": 25231 }, { "epoch": 0.03, "learning_rate": 4.992210222418393e-05, "loss": 0.5771, "step": 25232 }, { "epoch": 0.03, "learning_rate": 4.992209602819137e-05, "loss": 1.2469, "step": 25233 }, { "epoch": 0.03, "learning_rate": 4.9922089831952784e-05, "loss": 0.7842, "step": 25234 }, { "epoch": 0.03, "learning_rate": 4.9922083635468174e-05, "loss": 1.0711, "step": 25235 }, { "epoch": 0.03, "learning_rate": 4.9922077438737545e-05, "loss": 1.0961, "step": 25236 }, { "epoch": 0.03, "learning_rate": 4.9922071241760904e-05, "loss": 1.0949, "step": 25237 }, { "epoch": 0.03, "learning_rate": 4.992206504453824e-05, "loss": 0.9151, "step": 25238 }, { "epoch": 0.03, "learning_rate": 4.992205884706955e-05, "loss": 1.1718, "step": 25239 }, { "epoch": 0.03, "learning_rate": 4.992205264935485e-05, "loss": 1.1624, "step": 25240 }, { "epoch": 0.03, "learning_rate": 4.992204645139412e-05, "loss": 1.2207, "step": 25241 }, { "epoch": 0.03, "learning_rate": 4.992204025318737e-05, "loss": 1.1188, "step": 25242 }, { "epoch": 0.03, "learning_rate": 4.99220340547346e-05, "loss": 0.7908, "step": 25243 }, { "epoch": 0.03, "learning_rate": 4.992202785603581e-05, "loss": 0.8162, "step": 25244 }, { "epoch": 0.03, "learning_rate": 4.992202165709101e-05, "loss": 0.943, "step": 25245 }, { "epoch": 0.03, "learning_rate": 4.9922015457900186e-05, "loss": 1.1426, "step": 25246 }, { "epoch": 0.03, "learning_rate": 4.992200925846334e-05, "loss": 1.0149, "step": 25247 }, { "epoch": 0.03, "learning_rate": 4.992200305878048e-05, "loss": 1.3231, "step": 25248 }, { "epoch": 0.03, "learning_rate": 4.992199685885159e-05, "loss": 1.0122, "step": 25249 }, { "epoch": 0.03, "learning_rate": 4.9921990658676686e-05, "loss": 0.7944, "step": 25250 }, { "epoch": 0.03, "learning_rate": 4.992198445825576e-05, "loss": 0.9803, "step": 25251 }, { "epoch": 0.03, "learning_rate": 4.992197825758882e-05, "loss": 1.1797, "step": 25252 }, { "epoch": 0.03, "learning_rate": 4.992197205667586e-05, "loss": 1.0542, "step": 25253 }, { "epoch": 0.03, "learning_rate": 4.9921965855516875e-05, "loss": 0.7744, "step": 25254 }, { "epoch": 0.03, "learning_rate": 4.992195965411187e-05, "loss": 1.0356, "step": 25255 }, { "epoch": 0.03, "learning_rate": 4.9921953452460855e-05, "loss": 1.0049, "step": 25256 }, { "epoch": 0.03, "learning_rate": 4.9921947250563814e-05, "loss": 1.2333, "step": 25257 }, { "epoch": 0.03, "learning_rate": 4.992194104842076e-05, "loss": 0.927, "step": 25258 }, { "epoch": 0.03, "learning_rate": 4.992193484603168e-05, "loss": 0.9216, "step": 25259 }, { "epoch": 0.03, "learning_rate": 4.992192864339659e-05, "loss": 0.8418, "step": 25260 }, { "epoch": 0.03, "learning_rate": 4.992192244051547e-05, "loss": 1.0691, "step": 25261 }, { "epoch": 0.03, "learning_rate": 4.992191623738835e-05, "loss": 1.0355, "step": 25262 }, { "epoch": 0.03, "learning_rate": 4.992191003401519e-05, "loss": 1.0515, "step": 25263 }, { "epoch": 0.03, "learning_rate": 4.992190383039602e-05, "loss": 1.0592, "step": 25264 }, { "epoch": 0.03, "learning_rate": 4.9921897626530836e-05, "loss": 1.1024, "step": 25265 }, { "epoch": 0.03, "learning_rate": 4.992189142241962e-05, "loss": 1.0361, "step": 25266 }, { "epoch": 0.03, "learning_rate": 4.99218852180624e-05, "loss": 1.2286, "step": 25267 }, { "epoch": 0.03, "learning_rate": 4.9921879013459153e-05, "loss": 0.8063, "step": 25268 }, { "epoch": 0.03, "learning_rate": 4.992187280860989e-05, "loss": 1.1173, "step": 25269 }, { "epoch": 0.03, "learning_rate": 4.992186660351461e-05, "loss": 1.1498, "step": 25270 }, { "epoch": 0.03, "learning_rate": 4.992186039817331e-05, "loss": 1.1151, "step": 25271 }, { "epoch": 0.03, "learning_rate": 4.992185419258599e-05, "loss": 1.1275, "step": 25272 }, { "epoch": 0.03, "learning_rate": 4.992184798675266e-05, "loss": 1.0718, "step": 25273 }, { "epoch": 0.03, "learning_rate": 4.992184178067331e-05, "loss": 0.5926, "step": 25274 }, { "epoch": 0.03, "learning_rate": 4.992183557434794e-05, "loss": 0.4591, "step": 25275 }, { "epoch": 0.03, "learning_rate": 4.992182936777655e-05, "loss": 0.8349, "step": 25276 }, { "epoch": 0.03, "learning_rate": 4.992182316095915e-05, "loss": 0.9374, "step": 25277 }, { "epoch": 0.03, "learning_rate": 4.992181695389572e-05, "loss": 0.9491, "step": 25278 }, { "epoch": 0.03, "learning_rate": 4.992181074658628e-05, "loss": 0.9518, "step": 25279 }, { "epoch": 0.03, "learning_rate": 4.992180453903082e-05, "loss": 1.1468, "step": 25280 }, { "epoch": 0.03, "learning_rate": 4.992179833122934e-05, "loss": 1.1256, "step": 25281 }, { "epoch": 0.03, "learning_rate": 4.992179212318185e-05, "loss": 1.1141, "step": 25282 }, { "epoch": 0.03, "learning_rate": 4.9921785914888336e-05, "loss": 0.7789, "step": 25283 }, { "epoch": 0.03, "learning_rate": 4.992177970634881e-05, "loss": 1.0126, "step": 25284 }, { "epoch": 0.03, "learning_rate": 4.992177349756326e-05, "loss": 1.0437, "step": 25285 }, { "epoch": 0.03, "learning_rate": 4.9921767288531694e-05, "loss": 1.1684, "step": 25286 }, { "epoch": 0.03, "learning_rate": 4.9921761079254115e-05, "loss": 1.0412, "step": 25287 }, { "epoch": 0.03, "learning_rate": 4.992175486973052e-05, "loss": 1.1102, "step": 25288 }, { "epoch": 0.03, "learning_rate": 4.99217486599609e-05, "loss": 0.7031, "step": 25289 }, { "epoch": 0.03, "learning_rate": 4.992174244994527e-05, "loss": 0.8885, "step": 25290 }, { "epoch": 0.03, "learning_rate": 4.9921736239683624e-05, "loss": 1.207, "step": 25291 }, { "epoch": 0.03, "learning_rate": 4.9921730029175964e-05, "loss": 0.8742, "step": 25292 }, { "epoch": 0.03, "learning_rate": 4.992172381842227e-05, "loss": 0.9332, "step": 25293 }, { "epoch": 0.03, "learning_rate": 4.992171760742257e-05, "loss": 0.8615, "step": 25294 }, { "epoch": 0.03, "learning_rate": 4.992171139617686e-05, "loss": 1.1136, "step": 25295 }, { "epoch": 0.03, "learning_rate": 4.992170518468513e-05, "loss": 1.2426, "step": 25296 }, { "epoch": 0.03, "learning_rate": 4.992169897294738e-05, "loss": 1.0677, "step": 25297 }, { "epoch": 0.03, "learning_rate": 4.992169276096361e-05, "loss": 1.1393, "step": 25298 }, { "epoch": 0.03, "learning_rate": 4.9921686548733824e-05, "loss": 1.0658, "step": 25299 }, { "epoch": 0.03, "learning_rate": 4.992168033625803e-05, "loss": 1.0903, "step": 25300 }, { "epoch": 0.03, "learning_rate": 4.9921674123536214e-05, "loss": 0.9362, "step": 25301 }, { "epoch": 0.03, "learning_rate": 4.9921667910568384e-05, "loss": 0.988, "step": 25302 }, { "epoch": 0.03, "learning_rate": 4.992166169735454e-05, "loss": 1.1563, "step": 25303 }, { "epoch": 0.03, "learning_rate": 4.992165548389467e-05, "loss": 1.131, "step": 25304 }, { "epoch": 0.03, "learning_rate": 4.99216492701888e-05, "loss": 0.942, "step": 25305 }, { "epoch": 0.03, "learning_rate": 4.99216430562369e-05, "loss": 0.8895, "step": 25306 }, { "epoch": 0.03, "learning_rate": 4.992163684203899e-05, "loss": 0.8774, "step": 25307 }, { "epoch": 0.03, "learning_rate": 4.9921630627595065e-05, "loss": 0.8934, "step": 25308 }, { "epoch": 0.03, "learning_rate": 4.9921624412905115e-05, "loss": 1.1833, "step": 25309 }, { "epoch": 0.03, "learning_rate": 4.992161819796915e-05, "loss": 1.1606, "step": 25310 }, { "epoch": 0.03, "learning_rate": 4.992161198278718e-05, "loss": 0.9047, "step": 25311 }, { "epoch": 0.03, "learning_rate": 4.992160576735919e-05, "loss": 1.2609, "step": 25312 }, { "epoch": 0.03, "learning_rate": 4.992159955168518e-05, "loss": 0.7526, "step": 25313 }, { "epoch": 0.03, "learning_rate": 4.992159333576516e-05, "loss": 0.9345, "step": 25314 }, { "epoch": 0.03, "learning_rate": 4.9921587119599125e-05, "loss": 1.1732, "step": 25315 }, { "epoch": 0.03, "learning_rate": 4.992158090318707e-05, "loss": 0.8698, "step": 25316 }, { "epoch": 0.03, "learning_rate": 4.9921574686529e-05, "loss": 1.0294, "step": 25317 }, { "epoch": 0.03, "learning_rate": 4.992156846962492e-05, "loss": 0.7285, "step": 25318 }, { "epoch": 0.03, "learning_rate": 4.992156225247482e-05, "loss": 1.007, "step": 25319 }, { "epoch": 0.03, "learning_rate": 4.992155603507871e-05, "loss": 1.1879, "step": 25320 }, { "epoch": 0.03, "learning_rate": 4.9921549817436586e-05, "loss": 0.8376, "step": 25321 }, { "epoch": 0.03, "learning_rate": 4.992154359954844e-05, "loss": 1.0078, "step": 25322 }, { "epoch": 0.03, "learning_rate": 4.992153738141428e-05, "loss": 0.9619, "step": 25323 }, { "epoch": 0.03, "learning_rate": 4.992153116303411e-05, "loss": 0.8041, "step": 25324 }, { "epoch": 0.03, "learning_rate": 4.9921524944407915e-05, "loss": 1.242, "step": 25325 }, { "epoch": 0.03, "learning_rate": 4.9921518725535716e-05, "loss": 1.06, "step": 25326 }, { "epoch": 0.03, "learning_rate": 4.99215125064175e-05, "loss": 0.8303, "step": 25327 }, { "epoch": 0.03, "learning_rate": 4.992150628705327e-05, "loss": 0.8614, "step": 25328 }, { "epoch": 0.03, "learning_rate": 4.992150006744302e-05, "loss": 1.1408, "step": 25329 }, { "epoch": 0.03, "learning_rate": 4.992149384758675e-05, "loss": 1.0482, "step": 25330 }, { "epoch": 0.03, "learning_rate": 4.992148762748448e-05, "loss": 0.8925, "step": 25331 }, { "epoch": 0.03, "learning_rate": 4.992148140713619e-05, "loss": 0.9556, "step": 25332 }, { "epoch": 0.03, "learning_rate": 4.992147518654188e-05, "loss": 0.7117, "step": 25333 }, { "epoch": 0.03, "learning_rate": 4.992146896570157e-05, "loss": 0.9339, "step": 25334 }, { "epoch": 0.03, "learning_rate": 4.9921462744615235e-05, "loss": 0.9717, "step": 25335 }, { "epoch": 0.03, "learning_rate": 4.9921456523282887e-05, "loss": 0.9243, "step": 25336 }, { "epoch": 0.03, "learning_rate": 4.992145030170452e-05, "loss": 1.0203, "step": 25337 }, { "epoch": 0.03, "learning_rate": 4.992144407988014e-05, "loss": 0.3971, "step": 25338 }, { "epoch": 0.03, "learning_rate": 4.9921437857809755e-05, "loss": 0.578, "step": 25339 }, { "epoch": 0.03, "learning_rate": 4.992143163549335e-05, "loss": 0.4495, "step": 25340 }, { "epoch": 0.03, "learning_rate": 4.9921425412930935e-05, "loss": 0.5035, "step": 25341 }, { "epoch": 0.03, "learning_rate": 4.9921419190122506e-05, "loss": 0.5704, "step": 25342 }, { "epoch": 0.03, "learning_rate": 4.992141296706806e-05, "loss": 0.3427, "step": 25343 }, { "epoch": 0.03, "learning_rate": 4.99214067437676e-05, "loss": 1.1393, "step": 25344 }, { "epoch": 0.03, "learning_rate": 4.992140052022113e-05, "loss": 0.7993, "step": 25345 }, { "epoch": 0.03, "learning_rate": 4.9921394296428636e-05, "loss": 0.8591, "step": 25346 }, { "epoch": 0.03, "learning_rate": 4.992138807239014e-05, "loss": 0.992, "step": 25347 }, { "epoch": 0.03, "learning_rate": 4.992138184810563e-05, "loss": 0.8876, "step": 25348 }, { "epoch": 0.03, "learning_rate": 4.99213756235751e-05, "loss": 1.2374, "step": 25349 }, { "epoch": 0.03, "learning_rate": 4.992136939879856e-05, "loss": 1.3325, "step": 25350 }, { "epoch": 0.03, "learning_rate": 4.992136317377601e-05, "loss": 0.8105, "step": 25351 }, { "epoch": 0.03, "learning_rate": 4.992135694850744e-05, "loss": 0.8143, "step": 25352 }, { "epoch": 0.03, "learning_rate": 4.992135072299286e-05, "loss": 1.0643, "step": 25353 }, { "epoch": 0.03, "learning_rate": 4.992134449723227e-05, "loss": 1.0418, "step": 25354 }, { "epoch": 0.03, "learning_rate": 4.9921338271225665e-05, "loss": 1.0012, "step": 25355 }, { "epoch": 0.03, "learning_rate": 4.9921332044973045e-05, "loss": 0.9529, "step": 25356 }, { "epoch": 0.03, "learning_rate": 4.992132581847442e-05, "loss": 0.7987, "step": 25357 }, { "epoch": 0.03, "learning_rate": 4.992131959172977e-05, "loss": 1.1857, "step": 25358 }, { "epoch": 0.03, "learning_rate": 4.992131336473911e-05, "loss": 1.0198, "step": 25359 }, { "epoch": 0.03, "learning_rate": 4.992130713750244e-05, "loss": 0.8573, "step": 25360 }, { "epoch": 0.03, "learning_rate": 4.9921300910019767e-05, "loss": 0.6895, "step": 25361 }, { "epoch": 0.03, "learning_rate": 4.9921294682291066e-05, "loss": 0.4411, "step": 25362 }, { "epoch": 0.03, "learning_rate": 4.992128845431636e-05, "loss": 0.2818, "step": 25363 }, { "epoch": 0.03, "learning_rate": 4.992128222609564e-05, "loss": 0.4128, "step": 25364 }, { "epoch": 0.03, "learning_rate": 4.992127599762891e-05, "loss": 0.5094, "step": 25365 }, { "epoch": 0.03, "learning_rate": 4.992126976891616e-05, "loss": 0.6093, "step": 25366 }, { "epoch": 0.03, "learning_rate": 4.9921263539957406e-05, "loss": 0.448, "step": 25367 }, { "epoch": 0.03, "learning_rate": 4.992125731075263e-05, "loss": 0.7534, "step": 25368 }, { "epoch": 0.03, "learning_rate": 4.992125108130186e-05, "loss": 0.9877, "step": 25369 }, { "epoch": 0.03, "learning_rate": 4.992124485160507e-05, "loss": 1.4394, "step": 25370 }, { "epoch": 0.03, "learning_rate": 4.992123862166226e-05, "loss": 0.6872, "step": 25371 }, { "epoch": 0.03, "learning_rate": 4.992123239147344e-05, "loss": 1.0095, "step": 25372 }, { "epoch": 0.03, "learning_rate": 4.992122616103861e-05, "loss": 1.1616, "step": 25373 }, { "epoch": 0.03, "learning_rate": 4.992121993035777e-05, "loss": 1.0794, "step": 25374 }, { "epoch": 0.03, "learning_rate": 4.992121369943092e-05, "loss": 0.8517, "step": 25375 }, { "epoch": 0.03, "learning_rate": 4.992120746825805e-05, "loss": 1.1044, "step": 25376 }, { "epoch": 0.03, "learning_rate": 4.9921201236839176e-05, "loss": 1.3176, "step": 25377 }, { "epoch": 0.03, "learning_rate": 4.992119500517429e-05, "loss": 1.0207, "step": 25378 }, { "epoch": 0.03, "learning_rate": 4.992118877326339e-05, "loss": 1.0954, "step": 25379 }, { "epoch": 0.03, "learning_rate": 4.992118254110648e-05, "loss": 1.1591, "step": 25380 }, { "epoch": 0.03, "learning_rate": 4.9921176308703556e-05, "loss": 1.0753, "step": 25381 }, { "epoch": 0.03, "learning_rate": 4.9921170076054624e-05, "loss": 0.8058, "step": 25382 }, { "epoch": 0.03, "learning_rate": 4.992116384315968e-05, "loss": 0.9258, "step": 25383 }, { "epoch": 0.03, "learning_rate": 4.9921157610018724e-05, "loss": 1.0384, "step": 25384 }, { "epoch": 0.03, "learning_rate": 4.992115137663176e-05, "loss": 1.3298, "step": 25385 }, { "epoch": 0.03, "learning_rate": 4.9921145142998774e-05, "loss": 1.078, "step": 25386 }, { "epoch": 0.03, "learning_rate": 4.9921138909119794e-05, "loss": 0.657, "step": 25387 }, { "epoch": 0.03, "learning_rate": 4.992113267499479e-05, "loss": 0.6318, "step": 25388 }, { "epoch": 0.03, "learning_rate": 4.992112644062378e-05, "loss": 0.159, "step": 25389 }, { "epoch": 0.03, "learning_rate": 4.9921120206006755e-05, "loss": 0.1065, "step": 25390 }, { "epoch": 0.03, "learning_rate": 4.992111397114372e-05, "loss": 0.0809, "step": 25391 }, { "epoch": 0.03, "learning_rate": 4.992110773603468e-05, "loss": 0.5427, "step": 25392 }, { "epoch": 0.03, "learning_rate": 4.9921101500679625e-05, "loss": 0.2939, "step": 25393 }, { "epoch": 0.03, "learning_rate": 4.992109526507855e-05, "loss": 0.3915, "step": 25394 }, { "epoch": 0.03, "learning_rate": 4.992108902923148e-05, "loss": 0.5509, "step": 25395 }, { "epoch": 0.03, "learning_rate": 4.99210827931384e-05, "loss": 0.2196, "step": 25396 }, { "epoch": 0.03, "learning_rate": 4.99210765567993e-05, "loss": 0.0504, "step": 25397 }, { "epoch": 0.03, "learning_rate": 4.992107032021419e-05, "loss": 0.058, "step": 25398 }, { "epoch": 0.03, "learning_rate": 4.9921064083383076e-05, "loss": 0.4688, "step": 25399 }, { "epoch": 0.03, "learning_rate": 4.992105784630595e-05, "loss": 0.3038, "step": 25400 }, { "epoch": 0.03, "learning_rate": 4.992105160898281e-05, "loss": 0.0728, "step": 25401 }, { "epoch": 0.03, "learning_rate": 4.992104537141366e-05, "loss": 0.0445, "step": 25402 }, { "epoch": 0.03, "learning_rate": 4.992103913359851e-05, "loss": 0.3594, "step": 25403 }, { "epoch": 0.03, "learning_rate": 4.992103289553734e-05, "loss": 0.4276, "step": 25404 }, { "epoch": 0.03, "learning_rate": 4.992102665723016e-05, "loss": 0.5449, "step": 25405 }, { "epoch": 0.03, "learning_rate": 4.9921020418676975e-05, "loss": 0.9461, "step": 25406 }, { "epoch": 0.03, "learning_rate": 4.992101417987778e-05, "loss": 0.9638, "step": 25407 }, { "epoch": 0.03, "learning_rate": 4.992100794083257e-05, "loss": 1.2127, "step": 25408 }, { "epoch": 0.03, "learning_rate": 4.992100170154135e-05, "loss": 1.145, "step": 25409 }, { "epoch": 0.03, "learning_rate": 4.9920995462004124e-05, "loss": 0.9204, "step": 25410 }, { "epoch": 0.03, "learning_rate": 4.992098922222089e-05, "loss": 1.1378, "step": 25411 }, { "epoch": 0.03, "learning_rate": 4.9920982982191646e-05, "loss": 1.0549, "step": 25412 }, { "epoch": 0.03, "learning_rate": 4.9920976741916395e-05, "loss": 1.057, "step": 25413 }, { "epoch": 0.03, "learning_rate": 4.9920970501395126e-05, "loss": 0.9796, "step": 25414 }, { "epoch": 0.03, "learning_rate": 4.992096426062786e-05, "loss": 0.8747, "step": 25415 }, { "epoch": 0.03, "learning_rate": 4.992095801961457e-05, "loss": 0.984, "step": 25416 }, { "epoch": 0.03, "learning_rate": 4.9920951778355277e-05, "loss": 0.627, "step": 25417 }, { "epoch": 0.03, "learning_rate": 4.992094553684997e-05, "loss": 0.7613, "step": 25418 }, { "epoch": 0.03, "learning_rate": 4.992093929509867e-05, "loss": 0.9423, "step": 25419 }, { "epoch": 0.03, "learning_rate": 4.992093305310135e-05, "loss": 0.7867, "step": 25420 }, { "epoch": 0.03, "learning_rate": 4.9920926810858016e-05, "loss": 0.8356, "step": 25421 }, { "epoch": 0.03, "learning_rate": 4.9920920568368675e-05, "loss": 1.2746, "step": 25422 }, { "epoch": 0.03, "learning_rate": 4.9920914325633336e-05, "loss": 1.0376, "step": 25423 }, { "epoch": 0.03, "learning_rate": 4.9920908082651984e-05, "loss": 1.2483, "step": 25424 }, { "epoch": 0.03, "learning_rate": 4.9920901839424614e-05, "loss": 0.8622, "step": 25425 }, { "epoch": 0.03, "learning_rate": 4.9920895595951245e-05, "loss": 1.0395, "step": 25426 }, { "epoch": 0.03, "learning_rate": 4.992088935223186e-05, "loss": 1.1181, "step": 25427 }, { "epoch": 0.03, "learning_rate": 4.9920883108266476e-05, "loss": 0.7554, "step": 25428 }, { "epoch": 0.03, "learning_rate": 4.9920876864055077e-05, "loss": 0.8817, "step": 25429 }, { "epoch": 0.03, "learning_rate": 4.992087061959767e-05, "loss": 0.862, "step": 25430 }, { "epoch": 0.03, "learning_rate": 4.992086437489426e-05, "loss": 0.912, "step": 25431 }, { "epoch": 0.03, "learning_rate": 4.992085812994484e-05, "loss": 0.9013, "step": 25432 }, { "epoch": 0.03, "learning_rate": 4.9920851884749404e-05, "loss": 1.0225, "step": 25433 }, { "epoch": 0.03, "learning_rate": 4.9920845639307964e-05, "loss": 1.1501, "step": 25434 }, { "epoch": 0.03, "learning_rate": 4.9920839393620525e-05, "loss": 0.6082, "step": 25435 }, { "epoch": 0.03, "learning_rate": 4.992083314768706e-05, "loss": 0.9016, "step": 25436 }, { "epoch": 0.03, "learning_rate": 4.99208269015076e-05, "loss": 0.858, "step": 25437 }, { "epoch": 0.03, "learning_rate": 4.992082065508213e-05, "loss": 1.2233, "step": 25438 }, { "epoch": 0.03, "learning_rate": 4.9920814408410654e-05, "loss": 0.6477, "step": 25439 }, { "epoch": 0.03, "learning_rate": 4.992080816149316e-05, "loss": 0.7057, "step": 25440 }, { "epoch": 0.03, "learning_rate": 4.9920801914329675e-05, "loss": 1.3023, "step": 25441 }, { "epoch": 0.03, "learning_rate": 4.9920795666920164e-05, "loss": 1.0188, "step": 25442 }, { "epoch": 0.03, "learning_rate": 4.9920789419264654e-05, "loss": 0.5966, "step": 25443 }, { "epoch": 0.03, "learning_rate": 4.992078317136314e-05, "loss": 0.5925, "step": 25444 }, { "epoch": 0.03, "learning_rate": 4.992077692321562e-05, "loss": 0.6258, "step": 25445 }, { "epoch": 0.03, "learning_rate": 4.992077067482208e-05, "loss": 0.5038, "step": 25446 }, { "epoch": 0.03, "learning_rate": 4.992076442618254e-05, "loss": 0.589, "step": 25447 }, { "epoch": 0.03, "learning_rate": 4.9920758177297e-05, "loss": 0.4842, "step": 25448 }, { "epoch": 0.03, "learning_rate": 4.992075192816545e-05, "loss": 0.4888, "step": 25449 }, { "epoch": 0.03, "learning_rate": 4.9920745678787885e-05, "loss": 0.398, "step": 25450 }, { "epoch": 0.03, "learning_rate": 4.992073942916432e-05, "loss": 0.2454, "step": 25451 }, { "epoch": 0.03, "learning_rate": 4.9920733179294745e-05, "loss": 0.2128, "step": 25452 }, { "epoch": 0.03, "learning_rate": 4.992072692917916e-05, "loss": 0.2167, "step": 25453 }, { "epoch": 0.03, "learning_rate": 4.992072067881757e-05, "loss": 0.745, "step": 25454 }, { "epoch": 0.03, "learning_rate": 4.9920714428209974e-05, "loss": 0.793, "step": 25455 }, { "epoch": 0.03, "learning_rate": 4.9920708177356366e-05, "loss": 0.8001, "step": 25456 }, { "epoch": 0.03, "learning_rate": 4.9920701926256766e-05, "loss": 4.6841, "step": 25457 }, { "epoch": 0.03, "learning_rate": 4.992069567491115e-05, "loss": 1.0981, "step": 25458 }, { "epoch": 0.03, "learning_rate": 4.992068942331952e-05, "loss": 1.1137, "step": 25459 }, { "epoch": 0.03, "learning_rate": 4.99206831714819e-05, "loss": 0.9468, "step": 25460 }, { "epoch": 0.03, "learning_rate": 4.992067691939826e-05, "loss": 0.6921, "step": 25461 }, { "epoch": 0.03, "learning_rate": 4.9920670667068616e-05, "loss": 0.9783, "step": 25462 }, { "epoch": 0.03, "learning_rate": 4.992066441449297e-05, "loss": 0.9143, "step": 25463 }, { "epoch": 0.03, "learning_rate": 4.992065816167131e-05, "loss": 0.9276, "step": 25464 }, { "epoch": 0.03, "learning_rate": 4.9920651908603654e-05, "loss": 1.2974, "step": 25465 }, { "epoch": 0.03, "learning_rate": 4.992064565528999e-05, "loss": 0.9993, "step": 25466 }, { "epoch": 0.03, "learning_rate": 4.992063940173031e-05, "loss": 0.8887, "step": 25467 }, { "epoch": 0.03, "learning_rate": 4.9920633147924636e-05, "loss": 1.1553, "step": 25468 }, { "epoch": 0.03, "learning_rate": 4.992062689387294e-05, "loss": 1.2363, "step": 25469 }, { "epoch": 0.03, "learning_rate": 4.992062063957526e-05, "loss": 1.0574, "step": 25470 }, { "epoch": 0.03, "learning_rate": 4.992061438503156e-05, "loss": 0.8442, "step": 25471 }, { "epoch": 0.03, "learning_rate": 4.992060813024185e-05, "loss": 1.1638, "step": 25472 }, { "epoch": 0.03, "learning_rate": 4.992060187520615e-05, "loss": 1.197, "step": 25473 }, { "epoch": 0.03, "learning_rate": 4.992059561992444e-05, "loss": 1.3049, "step": 25474 }, { "epoch": 0.03, "learning_rate": 4.992058936439671e-05, "loss": 1.3725, "step": 25475 }, { "epoch": 0.03, "learning_rate": 4.992058310862299e-05, "loss": 0.7517, "step": 25476 }, { "epoch": 0.03, "learning_rate": 4.992057685260326e-05, "loss": 1.2874, "step": 25477 }, { "epoch": 0.03, "learning_rate": 4.992057059633752e-05, "loss": 1.1467, "step": 25478 }, { "epoch": 0.03, "learning_rate": 4.992056433982578e-05, "loss": 1.0138, "step": 25479 }, { "epoch": 0.03, "learning_rate": 4.992055808306803e-05, "loss": 1.0434, "step": 25480 }, { "epoch": 0.03, "learning_rate": 4.992055182606428e-05, "loss": 0.7546, "step": 25481 }, { "epoch": 0.03, "learning_rate": 4.992054556881453e-05, "loss": 0.9228, "step": 25482 }, { "epoch": 0.03, "learning_rate": 4.992053931131876e-05, "loss": 0.9298, "step": 25483 }, { "epoch": 0.03, "learning_rate": 4.992053305357699e-05, "loss": 1.0157, "step": 25484 }, { "epoch": 0.03, "learning_rate": 4.9920526795589216e-05, "loss": 0.7477, "step": 25485 }, { "epoch": 0.03, "learning_rate": 4.9920520537355444e-05, "loss": 1.2381, "step": 25486 }, { "epoch": 0.03, "learning_rate": 4.992051427887566e-05, "loss": 0.9527, "step": 25487 }, { "epoch": 0.03, "learning_rate": 4.992050802014987e-05, "loss": 1.1989, "step": 25488 }, { "epoch": 0.03, "learning_rate": 4.992050176117808e-05, "loss": 1.077, "step": 25489 }, { "epoch": 0.03, "learning_rate": 4.992049550196029e-05, "loss": 0.9295, "step": 25490 }, { "epoch": 0.03, "learning_rate": 4.992048924249648e-05, "loss": 1.1007, "step": 25491 }, { "epoch": 0.03, "learning_rate": 4.9920482982786675e-05, "loss": 0.9844, "step": 25492 }, { "epoch": 0.03, "learning_rate": 4.9920476722830865e-05, "loss": 0.5751, "step": 25493 }, { "epoch": 0.03, "learning_rate": 4.992047046262905e-05, "loss": 0.7737, "step": 25494 }, { "epoch": 0.03, "learning_rate": 4.992046420218123e-05, "loss": 0.7894, "step": 25495 }, { "epoch": 0.03, "learning_rate": 4.992045794148741e-05, "loss": 0.9189, "step": 25496 }, { "epoch": 0.03, "learning_rate": 4.9920451680547575e-05, "loss": 0.8127, "step": 25497 }, { "epoch": 0.03, "learning_rate": 4.992044541936175e-05, "loss": 0.7078, "step": 25498 }, { "epoch": 0.03, "learning_rate": 4.992043915792991e-05, "loss": 1.1538, "step": 25499 }, { "epoch": 0.03, "learning_rate": 4.9920432896252065e-05, "loss": 0.8989, "step": 25500 }, { "epoch": 0.03, "eval_loss": 1.0479439496994019, "eval_runtime": 128.7307, "eval_samples_per_second": 10.759, "eval_steps_per_second": 5.383, "step": 25500 }, { "epoch": 0.03, "learning_rate": 4.9920426634328224e-05, "loss": 0.692, "step": 25501 }, { "epoch": 0.03, "learning_rate": 4.992042037215838e-05, "loss": 0.7568, "step": 25502 }, { "epoch": 0.03, "learning_rate": 4.9920414109742526e-05, "loss": 0.9906, "step": 25503 }, { "epoch": 0.03, "learning_rate": 4.992040784708067e-05, "loss": 1.1287, "step": 25504 }, { "epoch": 0.03, "learning_rate": 4.992040158417281e-05, "loss": 0.8113, "step": 25505 }, { "epoch": 0.03, "learning_rate": 4.9920395321018945e-05, "loss": 0.9368, "step": 25506 }, { "epoch": 0.03, "learning_rate": 4.992038905761908e-05, "loss": 0.6073, "step": 25507 }, { "epoch": 0.03, "learning_rate": 4.9920382793973206e-05, "loss": 0.8541, "step": 25508 }, { "epoch": 0.03, "learning_rate": 4.9920376530081335e-05, "loss": 1.3915, "step": 25509 }, { "epoch": 0.03, "learning_rate": 4.992037026594346e-05, "loss": 1.0817, "step": 25510 }, { "epoch": 0.03, "learning_rate": 4.992036400155957e-05, "loss": 1.0359, "step": 25511 }, { "epoch": 0.03, "learning_rate": 4.992035773692969e-05, "loss": 1.1118, "step": 25512 }, { "epoch": 0.03, "learning_rate": 4.99203514720538e-05, "loss": 1.0692, "step": 25513 }, { "epoch": 0.03, "learning_rate": 4.992034520693191e-05, "loss": 1.0317, "step": 25514 }, { "epoch": 0.03, "learning_rate": 4.9920338941564014e-05, "loss": 1.0124, "step": 25515 }, { "epoch": 0.03, "learning_rate": 4.992033267595012e-05, "loss": 0.8217, "step": 25516 }, { "epoch": 0.03, "learning_rate": 4.9920326410090224e-05, "loss": 0.8563, "step": 25517 }, { "epoch": 0.03, "learning_rate": 4.992032014398432e-05, "loss": 0.8516, "step": 25518 }, { "epoch": 0.03, "learning_rate": 4.992031387763241e-05, "loss": 0.8468, "step": 25519 }, { "epoch": 0.03, "learning_rate": 4.99203076110345e-05, "loss": 1.0821, "step": 25520 }, { "epoch": 0.03, "learning_rate": 4.992030134419059e-05, "loss": 0.6678, "step": 25521 }, { "epoch": 0.03, "learning_rate": 4.9920295077100676e-05, "loss": 0.8539, "step": 25522 }, { "epoch": 0.03, "learning_rate": 4.992028880976476e-05, "loss": 0.8829, "step": 25523 }, { "epoch": 0.03, "learning_rate": 4.9920282542182836e-05, "loss": 1.0373, "step": 25524 }, { "epoch": 0.03, "learning_rate": 4.992027627435492e-05, "loss": 1.1651, "step": 25525 }, { "epoch": 0.03, "learning_rate": 4.9920270006280995e-05, "loss": 0.9403, "step": 25526 }, { "epoch": 0.03, "learning_rate": 4.992026373796107e-05, "loss": 1.0858, "step": 25527 }, { "epoch": 0.03, "learning_rate": 4.992025746939514e-05, "loss": 1.04, "step": 25528 }, { "epoch": 0.03, "learning_rate": 4.9920251200583205e-05, "loss": 0.6519, "step": 25529 }, { "epoch": 0.03, "learning_rate": 4.9920244931525274e-05, "loss": 0.8379, "step": 25530 }, { "epoch": 0.03, "learning_rate": 4.9920238662221337e-05, "loss": 0.929, "step": 25531 }, { "epoch": 0.03, "learning_rate": 4.9920232392671394e-05, "loss": 0.9804, "step": 25532 }, { "epoch": 0.03, "learning_rate": 4.992022612287546e-05, "loss": 1.2467, "step": 25533 }, { "epoch": 0.03, "learning_rate": 4.992021985283352e-05, "loss": 1.1121, "step": 25534 }, { "epoch": 0.03, "learning_rate": 4.9920213582545574e-05, "loss": 0.8523, "step": 25535 }, { "epoch": 0.03, "learning_rate": 4.992020731201162e-05, "loss": 1.1252, "step": 25536 }, { "epoch": 0.03, "learning_rate": 4.992020104123168e-05, "loss": 0.9649, "step": 25537 }, { "epoch": 0.03, "learning_rate": 4.992019477020573e-05, "loss": 1.0461, "step": 25538 }, { "epoch": 0.03, "learning_rate": 4.992018849893378e-05, "loss": 0.8588, "step": 25539 }, { "epoch": 0.03, "learning_rate": 4.9920182227415825e-05, "loss": 0.7761, "step": 25540 }, { "epoch": 0.03, "learning_rate": 4.9920175955651874e-05, "loss": 0.8121, "step": 25541 }, { "epoch": 0.03, "learning_rate": 4.992016968364192e-05, "loss": 1.1637, "step": 25542 }, { "epoch": 0.03, "learning_rate": 4.9920163411385955e-05, "loss": 0.9679, "step": 25543 }, { "epoch": 0.03, "learning_rate": 4.992015713888401e-05, "loss": 0.9611, "step": 25544 }, { "epoch": 0.03, "learning_rate": 4.992015086613604e-05, "loss": 0.9391, "step": 25545 }, { "epoch": 0.03, "learning_rate": 4.992014459314208e-05, "loss": 0.7813, "step": 25546 }, { "epoch": 0.03, "learning_rate": 4.992013831990212e-05, "loss": 0.84, "step": 25547 }, { "epoch": 0.03, "learning_rate": 4.9920132046416156e-05, "loss": 0.8469, "step": 25548 }, { "epoch": 0.03, "learning_rate": 4.992012577268419e-05, "loss": 0.5595, "step": 25549 }, { "epoch": 0.03, "learning_rate": 4.992011949870622e-05, "loss": 0.7381, "step": 25550 }, { "epoch": 0.03, "learning_rate": 4.9920113224482256e-05, "loss": 0.9399, "step": 25551 }, { "epoch": 0.03, "learning_rate": 4.992010695001229e-05, "loss": 0.6042, "step": 25552 }, { "epoch": 0.03, "learning_rate": 4.992010067529632e-05, "loss": 0.824, "step": 25553 }, { "epoch": 0.03, "learning_rate": 4.992009440033436e-05, "loss": 0.7738, "step": 25554 }, { "epoch": 0.03, "learning_rate": 4.992008812512639e-05, "loss": 0.8869, "step": 25555 }, { "epoch": 0.03, "learning_rate": 4.9920081849672415e-05, "loss": 0.9603, "step": 25556 }, { "epoch": 0.03, "learning_rate": 4.9920075573972444e-05, "loss": 0.9571, "step": 25557 }, { "epoch": 0.03, "learning_rate": 4.9920069298026474e-05, "loss": 1.2069, "step": 25558 }, { "epoch": 0.03, "learning_rate": 4.9920063021834506e-05, "loss": 0.2281, "step": 25559 }, { "epoch": 0.03, "learning_rate": 4.992005674539654e-05, "loss": 0.666, "step": 25560 }, { "epoch": 0.03, "learning_rate": 4.9920050468712566e-05, "loss": 0.8789, "step": 25561 }, { "epoch": 0.03, "learning_rate": 4.992004419178259e-05, "loss": 1.0569, "step": 25562 }, { "epoch": 0.03, "learning_rate": 4.992003791460662e-05, "loss": 0.9824, "step": 25563 }, { "epoch": 0.03, "learning_rate": 4.992003163718464e-05, "loss": 0.7829, "step": 25564 }, { "epoch": 0.03, "learning_rate": 4.9920025359516675e-05, "loss": 0.9022, "step": 25565 }, { "epoch": 0.03, "learning_rate": 4.9920019081602695e-05, "loss": 1.1186, "step": 25566 }, { "epoch": 0.03, "learning_rate": 4.992001280344273e-05, "loss": 1.138, "step": 25567 }, { "epoch": 0.03, "learning_rate": 4.992000652503675e-05, "loss": 1.0575, "step": 25568 }, { "epoch": 0.03, "learning_rate": 4.9920000246384784e-05, "loss": 0.7972, "step": 25569 }, { "epoch": 0.03, "learning_rate": 4.991999396748681e-05, "loss": 1.0315, "step": 25570 }, { "epoch": 0.03, "learning_rate": 4.9919987688342837e-05, "loss": 1.0861, "step": 25571 }, { "epoch": 0.03, "learning_rate": 4.991998140895287e-05, "loss": 0.9074, "step": 25572 }, { "epoch": 0.03, "learning_rate": 4.99199751293169e-05, "loss": 1.1712, "step": 25573 }, { "epoch": 0.03, "learning_rate": 4.9919968849434925e-05, "loss": 1.0447, "step": 25574 }, { "epoch": 0.03, "learning_rate": 4.991996256930696e-05, "loss": 0.8542, "step": 25575 }, { "epoch": 0.03, "learning_rate": 4.991995628893299e-05, "loss": 0.9268, "step": 25576 }, { "epoch": 0.03, "learning_rate": 4.991995000831302e-05, "loss": 0.9937, "step": 25577 }, { "epoch": 0.03, "learning_rate": 4.9919943727447054e-05, "loss": 0.9313, "step": 25578 }, { "epoch": 0.03, "learning_rate": 4.9919937446335085e-05, "loss": 1.0095, "step": 25579 }, { "epoch": 0.03, "learning_rate": 4.9919931164977116e-05, "loss": 1.2027, "step": 25580 }, { "epoch": 0.03, "learning_rate": 4.9919924883373156e-05, "loss": 0.9313, "step": 25581 }, { "epoch": 0.03, "learning_rate": 4.99199186015232e-05, "loss": 0.8642, "step": 25582 }, { "epoch": 0.03, "learning_rate": 4.991991231942723e-05, "loss": 0.9955, "step": 25583 }, { "epoch": 0.03, "learning_rate": 4.991990603708527e-05, "loss": 1.305, "step": 25584 }, { "epoch": 0.03, "learning_rate": 4.9919899754497315e-05, "loss": 1.0399, "step": 25585 }, { "epoch": 0.03, "learning_rate": 4.991989347166335e-05, "loss": 1.0497, "step": 25586 }, { "epoch": 0.03, "learning_rate": 4.991988718858339e-05, "loss": 0.7731, "step": 25587 }, { "epoch": 0.03, "learning_rate": 4.991988090525743e-05, "loss": 1.1434, "step": 25588 }, { "epoch": 0.03, "learning_rate": 4.991987462168548e-05, "loss": 1.1553, "step": 25589 }, { "epoch": 0.03, "learning_rate": 4.9919868337867526e-05, "loss": 0.8772, "step": 25590 }, { "epoch": 0.03, "learning_rate": 4.991986205380357e-05, "loss": 1.0083, "step": 25591 }, { "epoch": 0.03, "learning_rate": 4.991985576949362e-05, "loss": 0.9481, "step": 25592 }, { "epoch": 0.03, "learning_rate": 4.9919849484937675e-05, "loss": 0.8152, "step": 25593 }, { "epoch": 0.03, "learning_rate": 4.991984320013573e-05, "loss": 1.0207, "step": 25594 }, { "epoch": 0.03, "learning_rate": 4.9919836915087783e-05, "loss": 1.0346, "step": 25595 }, { "epoch": 0.03, "learning_rate": 4.991983062979384e-05, "loss": 0.9336, "step": 25596 }, { "epoch": 0.03, "learning_rate": 4.99198243442539e-05, "loss": 1.0044, "step": 25597 }, { "epoch": 0.03, "learning_rate": 4.991981805846796e-05, "loss": 0.8974, "step": 25598 }, { "epoch": 0.03, "learning_rate": 4.991981177243602e-05, "loss": 0.9835, "step": 25599 }, { "epoch": 0.03, "learning_rate": 4.991980548615809e-05, "loss": 1.0279, "step": 25600 }, { "epoch": 0.03, "learning_rate": 4.991979919963415e-05, "loss": 1.0142, "step": 25601 }, { "epoch": 0.03, "learning_rate": 4.991979291286422e-05, "loss": 1.5041, "step": 25602 }, { "epoch": 0.03, "learning_rate": 4.9919786625848295e-05, "loss": 0.8546, "step": 25603 }, { "epoch": 0.03, "learning_rate": 4.9919780338586365e-05, "loss": 0.7064, "step": 25604 }, { "epoch": 0.03, "learning_rate": 4.991977405107844e-05, "loss": 0.9156, "step": 25605 }, { "epoch": 0.03, "learning_rate": 4.991976776332452e-05, "loss": 1.0011, "step": 25606 }, { "epoch": 0.03, "learning_rate": 4.99197614753246e-05, "loss": 0.6444, "step": 25607 }, { "epoch": 0.03, "learning_rate": 4.9919755187078685e-05, "loss": 0.6394, "step": 25608 }, { "epoch": 0.03, "learning_rate": 4.9919748898586775e-05, "loss": 0.8335, "step": 25609 }, { "epoch": 0.03, "learning_rate": 4.9919742609848866e-05, "loss": 0.7045, "step": 25610 }, { "epoch": 0.03, "learning_rate": 4.991973632086496e-05, "loss": 0.985, "step": 25611 }, { "epoch": 0.03, "learning_rate": 4.9919730031635045e-05, "loss": 0.7575, "step": 25612 }, { "epoch": 0.03, "learning_rate": 4.9919723742159154e-05, "loss": 1.043, "step": 25613 }, { "epoch": 0.03, "learning_rate": 4.991971745243725e-05, "loss": 0.7751, "step": 25614 }, { "epoch": 0.03, "learning_rate": 4.991971116246935e-05, "loss": 0.6379, "step": 25615 }, { "epoch": 0.03, "learning_rate": 4.991970487225546e-05, "loss": 0.6742, "step": 25616 }, { "epoch": 0.03, "learning_rate": 4.991969858179557e-05, "loss": 0.2412, "step": 25617 }, { "epoch": 0.03, "learning_rate": 4.991969229108969e-05, "loss": 0.1725, "step": 25618 }, { "epoch": 0.03, "learning_rate": 4.99196860001378e-05, "loss": 0.2412, "step": 25619 }, { "epoch": 0.03, "learning_rate": 4.9919679708939924e-05, "loss": 0.5532, "step": 25620 }, { "epoch": 0.03, "learning_rate": 4.991967341749605e-05, "loss": 0.8368, "step": 25621 }, { "epoch": 0.03, "learning_rate": 4.991966712580618e-05, "loss": 0.9496, "step": 25622 }, { "epoch": 0.03, "learning_rate": 4.9919660833870306e-05, "loss": 1.1434, "step": 25623 }, { "epoch": 0.03, "learning_rate": 4.9919654541688435e-05, "loss": 1.1843, "step": 25624 }, { "epoch": 0.03, "learning_rate": 4.991964824926057e-05, "loss": 1.2085, "step": 25625 }, { "epoch": 0.03, "learning_rate": 4.991964195658671e-05, "loss": 0.8527, "step": 25626 }, { "epoch": 0.03, "learning_rate": 4.9919635663666867e-05, "loss": 0.8021, "step": 25627 }, { "epoch": 0.03, "learning_rate": 4.991962937050101e-05, "loss": 0.6328, "step": 25628 }, { "epoch": 0.03, "learning_rate": 4.9919623077089165e-05, "loss": 0.9434, "step": 25629 }, { "epoch": 0.03, "learning_rate": 4.991961678343132e-05, "loss": 0.7023, "step": 25630 }, { "epoch": 0.03, "learning_rate": 4.991961048952748e-05, "loss": 0.968, "step": 25631 }, { "epoch": 0.03, "learning_rate": 4.991960419537765e-05, "loss": 0.4385, "step": 25632 }, { "epoch": 0.03, "learning_rate": 4.991959790098182e-05, "loss": 0.5095, "step": 25633 }, { "epoch": 0.03, "learning_rate": 4.9919591606339995e-05, "loss": 1.0062, "step": 25634 }, { "epoch": 0.03, "learning_rate": 4.991958531145217e-05, "loss": 1.0049, "step": 25635 }, { "epoch": 0.03, "learning_rate": 4.991957901631835e-05, "loss": 0.9749, "step": 25636 }, { "epoch": 0.03, "learning_rate": 4.991957272093854e-05, "loss": 0.8883, "step": 25637 }, { "epoch": 0.03, "learning_rate": 4.9919566425312735e-05, "loss": 1.0521, "step": 25638 }, { "epoch": 0.03, "learning_rate": 4.991956012944093e-05, "loss": 0.79, "step": 25639 }, { "epoch": 0.03, "learning_rate": 4.991955383332313e-05, "loss": 0.8642, "step": 25640 }, { "epoch": 0.03, "learning_rate": 4.991954753695933e-05, "loss": 0.9301, "step": 25641 }, { "epoch": 0.03, "learning_rate": 4.9919541240349543e-05, "loss": 0.4734, "step": 25642 }, { "epoch": 0.03, "learning_rate": 4.991953494349376e-05, "loss": 0.4748, "step": 25643 }, { "epoch": 0.03, "learning_rate": 4.9919528646391976e-05, "loss": 1.0601, "step": 25644 }, { "epoch": 0.03, "learning_rate": 4.99195223490442e-05, "loss": 0.9702, "step": 25645 }, { "epoch": 0.03, "learning_rate": 4.9919516051450434e-05, "loss": 1.0483, "step": 25646 }, { "epoch": 0.03, "learning_rate": 4.991950975361067e-05, "loss": 0.9423, "step": 25647 }, { "epoch": 0.03, "learning_rate": 4.991950345552491e-05, "loss": 0.9092, "step": 25648 }, { "epoch": 0.03, "learning_rate": 4.9919497157193154e-05, "loss": 1.1845, "step": 25649 }, { "epoch": 0.03, "learning_rate": 4.9919490858615406e-05, "loss": 1.1256, "step": 25650 }, { "epoch": 0.03, "learning_rate": 4.991948455979165e-05, "loss": 0.8735, "step": 25651 }, { "epoch": 0.03, "learning_rate": 4.991947826072192e-05, "loss": 0.8359, "step": 25652 }, { "epoch": 0.03, "learning_rate": 4.991947196140618e-05, "loss": 1.1225, "step": 25653 }, { "epoch": 0.03, "learning_rate": 4.991946566184445e-05, "loss": 0.9565, "step": 25654 }, { "epoch": 0.03, "learning_rate": 4.991945936203673e-05, "loss": 0.9137, "step": 25655 }, { "epoch": 0.03, "learning_rate": 4.991945306198301e-05, "loss": 0.8453, "step": 25656 }, { "epoch": 0.03, "learning_rate": 4.99194467616833e-05, "loss": 0.981, "step": 25657 }, { "epoch": 0.03, "learning_rate": 4.9919440461137596e-05, "loss": 0.7872, "step": 25658 }, { "epoch": 0.03, "learning_rate": 4.991943416034589e-05, "loss": 0.7202, "step": 25659 }, { "epoch": 0.03, "learning_rate": 4.99194278593082e-05, "loss": 0.6923, "step": 25660 }, { "epoch": 0.03, "learning_rate": 4.991942155802451e-05, "loss": 0.7891, "step": 25661 }, { "epoch": 0.03, "learning_rate": 4.991941525649483e-05, "loss": 0.7018, "step": 25662 }, { "epoch": 0.03, "learning_rate": 4.991940895471915e-05, "loss": 0.8904, "step": 25663 }, { "epoch": 0.03, "learning_rate": 4.991940265269748e-05, "loss": 1.0819, "step": 25664 }, { "epoch": 0.03, "learning_rate": 4.991939635042981e-05, "loss": 1.227, "step": 25665 }, { "epoch": 0.03, "learning_rate": 4.991939004791616e-05, "loss": 1.1283, "step": 25666 }, { "epoch": 0.03, "learning_rate": 4.991938374515651e-05, "loss": 1.3979, "step": 25667 }, { "epoch": 0.03, "learning_rate": 4.991937744215086e-05, "loss": 1.1434, "step": 25668 }, { "epoch": 0.03, "learning_rate": 4.991937113889922e-05, "loss": 1.0276, "step": 25669 }, { "epoch": 0.03, "learning_rate": 4.9919364835401596e-05, "loss": 0.7966, "step": 25670 }, { "epoch": 0.03, "learning_rate": 4.991935853165797e-05, "loss": 1.2742, "step": 25671 }, { "epoch": 0.03, "learning_rate": 4.9919352227668345e-05, "loss": 0.9509, "step": 25672 }, { "epoch": 0.03, "learning_rate": 4.9919345923432735e-05, "loss": 0.9776, "step": 25673 }, { "epoch": 0.03, "learning_rate": 4.991933961895113e-05, "loss": 1.0581, "step": 25674 }, { "epoch": 0.03, "learning_rate": 4.9919333314223526e-05, "loss": 1.2724, "step": 25675 }, { "epoch": 0.03, "learning_rate": 4.991932700924994e-05, "loss": 1.2349, "step": 25676 }, { "epoch": 0.03, "learning_rate": 4.991932070403035e-05, "loss": 0.8734, "step": 25677 }, { "epoch": 0.03, "learning_rate": 4.9919314398564773e-05, "loss": 0.8264, "step": 25678 }, { "epoch": 0.03, "learning_rate": 4.99193080928532e-05, "loss": 0.8606, "step": 25679 }, { "epoch": 0.03, "learning_rate": 4.991930178689563e-05, "loss": 0.7791, "step": 25680 }, { "epoch": 0.03, "learning_rate": 4.991929548069207e-05, "loss": 1.3218, "step": 25681 }, { "epoch": 0.03, "learning_rate": 4.991928917424253e-05, "loss": 1.1962, "step": 25682 }, { "epoch": 0.03, "learning_rate": 4.9919282867546986e-05, "loss": 1.0835, "step": 25683 }, { "epoch": 0.03, "learning_rate": 4.991927656060545e-05, "loss": 1.2724, "step": 25684 }, { "epoch": 0.03, "learning_rate": 4.991927025341792e-05, "loss": 1.113, "step": 25685 }, { "epoch": 0.03, "learning_rate": 4.99192639459844e-05, "loss": 1.0718, "step": 25686 }, { "epoch": 0.03, "learning_rate": 4.991925763830489e-05, "loss": 1.3142, "step": 25687 }, { "epoch": 0.03, "learning_rate": 4.991925133037938e-05, "loss": 0.9934, "step": 25688 }, { "epoch": 0.03, "learning_rate": 4.991924502220788e-05, "loss": 0.9403, "step": 25689 }, { "epoch": 0.03, "learning_rate": 4.9919238713790394e-05, "loss": 1.0894, "step": 25690 }, { "epoch": 0.03, "learning_rate": 4.9919232405126916e-05, "loss": 0.9014, "step": 25691 }, { "epoch": 0.03, "learning_rate": 4.991922609621744e-05, "loss": 1.1872, "step": 25692 }, { "epoch": 0.03, "learning_rate": 4.991921978706197e-05, "loss": 1.018, "step": 25693 }, { "epoch": 0.03, "learning_rate": 4.991921347766051e-05, "loss": 0.8534, "step": 25694 }, { "epoch": 0.03, "learning_rate": 4.991920716801306e-05, "loss": 1.419, "step": 25695 }, { "epoch": 0.03, "learning_rate": 4.991920085811962e-05, "loss": 1.1202, "step": 25696 }, { "epoch": 0.03, "learning_rate": 4.9919194547980184e-05, "loss": 0.7459, "step": 25697 }, { "epoch": 0.03, "learning_rate": 4.991918823759476e-05, "loss": 1.0924, "step": 25698 }, { "epoch": 0.03, "learning_rate": 4.991918192696334e-05, "loss": 1.0415, "step": 25699 }, { "epoch": 0.03, "learning_rate": 4.991917561608593e-05, "loss": 1.0242, "step": 25700 }, { "epoch": 0.03, "learning_rate": 4.9919169304962534e-05, "loss": 1.2055, "step": 25701 }, { "epoch": 0.03, "learning_rate": 4.991916299359314e-05, "loss": 1.1676, "step": 25702 }, { "epoch": 0.03, "learning_rate": 4.991915668197775e-05, "loss": 0.9963, "step": 25703 }, { "epoch": 0.03, "learning_rate": 4.991915037011638e-05, "loss": 0.7652, "step": 25704 }, { "epoch": 0.03, "learning_rate": 4.991914405800901e-05, "loss": 0.8364, "step": 25705 }, { "epoch": 0.03, "learning_rate": 4.9919137745655656e-05, "loss": 0.9469, "step": 25706 }, { "epoch": 0.03, "learning_rate": 4.991913143305631e-05, "loss": 0.8693, "step": 25707 }, { "epoch": 0.03, "learning_rate": 4.9919125120210966e-05, "loss": 0.9541, "step": 25708 }, { "epoch": 0.03, "learning_rate": 4.9919118807119634e-05, "loss": 0.9792, "step": 25709 }, { "epoch": 0.03, "learning_rate": 4.99191124937823e-05, "loss": 0.9718, "step": 25710 }, { "epoch": 0.03, "learning_rate": 4.991910618019899e-05, "loss": 0.728, "step": 25711 }, { "epoch": 0.03, "learning_rate": 4.991909986636969e-05, "loss": 1.2885, "step": 25712 }, { "epoch": 0.03, "learning_rate": 4.991909355229439e-05, "loss": 0.9237, "step": 25713 }, { "epoch": 0.03, "learning_rate": 4.9919087237973106e-05, "loss": 1.2815, "step": 25714 }, { "epoch": 0.03, "learning_rate": 4.991908092340583e-05, "loss": 1.1122, "step": 25715 }, { "epoch": 0.03, "learning_rate": 4.9919074608592566e-05, "loss": 0.9325, "step": 25716 }, { "epoch": 0.03, "learning_rate": 4.9919068293533305e-05, "loss": 1.1455, "step": 25717 }, { "epoch": 0.03, "learning_rate": 4.991906197822805e-05, "loss": 0.8719, "step": 25718 }, { "epoch": 0.03, "learning_rate": 4.9919055662676814e-05, "loss": 1.1377, "step": 25719 }, { "epoch": 0.03, "learning_rate": 4.991904934687959e-05, "loss": 0.8786, "step": 25720 }, { "epoch": 0.03, "learning_rate": 4.991904303083637e-05, "loss": 1.1421, "step": 25721 }, { "epoch": 0.03, "learning_rate": 4.9919036714547155e-05, "loss": 0.9805, "step": 25722 }, { "epoch": 0.03, "learning_rate": 4.9919030398011956e-05, "loss": 0.8057, "step": 25723 }, { "epoch": 0.03, "learning_rate": 4.991902408123077e-05, "loss": 1.363, "step": 25724 }, { "epoch": 0.03, "learning_rate": 4.991901776420359e-05, "loss": 1.0186, "step": 25725 }, { "epoch": 0.03, "learning_rate": 4.9919011446930414e-05, "loss": 0.7966, "step": 25726 }, { "epoch": 0.03, "learning_rate": 4.9919005129411254e-05, "loss": 1.1076, "step": 25727 }, { "epoch": 0.03, "learning_rate": 4.99189988116461e-05, "loss": 0.9373, "step": 25728 }, { "epoch": 0.03, "learning_rate": 4.9918992493634965e-05, "loss": 0.666, "step": 25729 }, { "epoch": 0.03, "learning_rate": 4.9918986175377836e-05, "loss": 0.8127, "step": 25730 }, { "epoch": 0.03, "learning_rate": 4.9918979856874715e-05, "loss": 0.8135, "step": 25731 }, { "epoch": 0.03, "learning_rate": 4.99189735381256e-05, "loss": 0.93, "step": 25732 }, { "epoch": 0.03, "learning_rate": 4.9918967219130504e-05, "loss": 1.1284, "step": 25733 }, { "epoch": 0.03, "learning_rate": 4.9918960899889414e-05, "loss": 1.3688, "step": 25734 }, { "epoch": 0.03, "learning_rate": 4.991895458040233e-05, "loss": 1.2117, "step": 25735 }, { "epoch": 0.03, "learning_rate": 4.9918948260669265e-05, "loss": 1.0694, "step": 25736 }, { "epoch": 0.03, "learning_rate": 4.991894194069021e-05, "loss": 1.3328, "step": 25737 }, { "epoch": 0.03, "learning_rate": 4.991893562046516e-05, "loss": 1.8201, "step": 25738 }, { "epoch": 0.03, "learning_rate": 4.9918929299994127e-05, "loss": 0.8728, "step": 25739 }, { "epoch": 0.03, "learning_rate": 4.9918922979277105e-05, "loss": 1.2008, "step": 25740 }, { "epoch": 0.03, "learning_rate": 4.9918916658314086e-05, "loss": 1.0079, "step": 25741 }, { "epoch": 0.03, "learning_rate": 4.991891033710509e-05, "loss": 1.0044, "step": 25742 }, { "epoch": 0.03, "learning_rate": 4.991890401565009e-05, "loss": 0.8514, "step": 25743 }, { "epoch": 0.03, "learning_rate": 4.9918897693949116e-05, "loss": 0.5864, "step": 25744 }, { "epoch": 0.03, "learning_rate": 4.991889137200214e-05, "loss": 0.527, "step": 25745 }, { "epoch": 0.03, "learning_rate": 4.991888504980918e-05, "loss": 1.1255, "step": 25746 }, { "epoch": 0.03, "learning_rate": 4.991887872737023e-05, "loss": 1.0253, "step": 25747 }, { "epoch": 0.03, "learning_rate": 4.99188724046853e-05, "loss": 1.0765, "step": 25748 }, { "epoch": 0.03, "learning_rate": 4.991886608175438e-05, "loss": 0.99, "step": 25749 }, { "epoch": 0.03, "learning_rate": 4.991885975857746e-05, "loss": 1.0563, "step": 25750 }, { "epoch": 0.03, "learning_rate": 4.991885343515456e-05, "loss": 2.487, "step": 25751 }, { "epoch": 0.03, "learning_rate": 4.9918847111485666e-05, "loss": 0.9786, "step": 25752 }, { "epoch": 0.03, "learning_rate": 4.991884078757079e-05, "loss": 0.6252, "step": 25753 }, { "epoch": 0.03, "learning_rate": 4.991883446340992e-05, "loss": 0.924, "step": 25754 }, { "epoch": 0.03, "learning_rate": 4.991882813900307e-05, "loss": 1.1609, "step": 25755 }, { "epoch": 0.03, "learning_rate": 4.9918821814350226e-05, "loss": 1.0787, "step": 25756 }, { "epoch": 0.03, "learning_rate": 4.99188154894514e-05, "loss": 1.1311, "step": 25757 }, { "epoch": 0.03, "learning_rate": 4.991880916430658e-05, "loss": 0.9768, "step": 25758 }, { "epoch": 0.03, "learning_rate": 4.991880283891577e-05, "loss": 0.9717, "step": 25759 }, { "epoch": 0.03, "learning_rate": 4.991879651327898e-05, "loss": 1.1451, "step": 25760 }, { "epoch": 0.03, "learning_rate": 4.9918790187396194e-05, "loss": 1.2621, "step": 25761 }, { "epoch": 0.03, "learning_rate": 4.991878386126742e-05, "loss": 1.0463, "step": 25762 }, { "epoch": 0.03, "learning_rate": 4.991877753489266e-05, "loss": 1.0465, "step": 25763 }, { "epoch": 0.03, "learning_rate": 4.991877120827191e-05, "loss": 0.9057, "step": 25764 }, { "epoch": 0.03, "learning_rate": 4.991876488140518e-05, "loss": 0.6581, "step": 25765 }, { "epoch": 0.03, "learning_rate": 4.991875855429246e-05, "loss": 0.7678, "step": 25766 }, { "epoch": 0.03, "learning_rate": 4.991875222693375e-05, "loss": 1.1339, "step": 25767 }, { "epoch": 0.03, "learning_rate": 4.9918745899329056e-05, "loss": 0.9398, "step": 25768 }, { "epoch": 0.03, "learning_rate": 4.991873957147837e-05, "loss": 0.6879, "step": 25769 }, { "epoch": 0.03, "learning_rate": 4.99187332433817e-05, "loss": 0.9775, "step": 25770 }, { "epoch": 0.03, "learning_rate": 4.991872691503904e-05, "loss": 1.0992, "step": 25771 }, { "epoch": 0.03, "learning_rate": 4.99187205864504e-05, "loss": 1.017, "step": 25772 }, { "epoch": 0.03, "learning_rate": 4.991871425761576e-05, "loss": 1.155, "step": 25773 }, { "epoch": 0.03, "learning_rate": 4.991870792853515e-05, "loss": 1.0096, "step": 25774 }, { "epoch": 0.03, "learning_rate": 4.991870159920854e-05, "loss": 1.0484, "step": 25775 }, { "epoch": 0.03, "learning_rate": 4.991869526963594e-05, "loss": 0.9261, "step": 25776 }, { "epoch": 0.03, "learning_rate": 4.991868893981736e-05, "loss": 1.1229, "step": 25777 }, { "epoch": 0.03, "learning_rate": 4.99186826097528e-05, "loss": 1.1667, "step": 25778 }, { "epoch": 0.03, "learning_rate": 4.991867627944224e-05, "loss": 1.3097, "step": 25779 }, { "epoch": 0.03, "learning_rate": 4.99186699488857e-05, "loss": 0.869, "step": 25780 }, { "epoch": 0.03, "learning_rate": 4.991866361808317e-05, "loss": 0.8759, "step": 25781 }, { "epoch": 0.03, "learning_rate": 4.991865728703467e-05, "loss": 1.0684, "step": 25782 }, { "epoch": 0.03, "learning_rate": 4.991865095574016e-05, "loss": 0.915, "step": 25783 }, { "epoch": 0.03, "learning_rate": 4.9918644624199674e-05, "loss": 0.9299, "step": 25784 }, { "epoch": 0.03, "learning_rate": 4.99186382924132e-05, "loss": 0.7329, "step": 25785 }, { "epoch": 0.03, "learning_rate": 4.9918631960380746e-05, "loss": 0.8754, "step": 25786 }, { "epoch": 0.03, "learning_rate": 4.99186256281023e-05, "loss": 0.7077, "step": 25787 }, { "epoch": 0.03, "learning_rate": 4.9918619295577865e-05, "loss": 1.0019, "step": 25788 }, { "epoch": 0.03, "learning_rate": 4.991861296280745e-05, "loss": 0.7738, "step": 25789 }, { "epoch": 0.03, "learning_rate": 4.991860662979104e-05, "loss": 0.6864, "step": 25790 }, { "epoch": 0.03, "learning_rate": 4.991860029652865e-05, "loss": 0.5901, "step": 25791 }, { "epoch": 0.03, "learning_rate": 4.991859396302028e-05, "loss": 0.6398, "step": 25792 }, { "epoch": 0.03, "learning_rate": 4.991858762926592e-05, "loss": 0.9284, "step": 25793 }, { "epoch": 0.03, "learning_rate": 4.991858129526556e-05, "loss": 0.7596, "step": 25794 }, { "epoch": 0.03, "learning_rate": 4.991857496101924e-05, "loss": 0.9285, "step": 25795 }, { "epoch": 0.03, "learning_rate": 4.991856862652692e-05, "loss": 0.8106, "step": 25796 }, { "epoch": 0.03, "learning_rate": 4.991856229178861e-05, "loss": 0.994, "step": 25797 }, { "epoch": 0.03, "learning_rate": 4.991855595680433e-05, "loss": 1.4894, "step": 25798 }, { "epoch": 0.03, "learning_rate": 4.991854962157405e-05, "loss": 1.3267, "step": 25799 }, { "epoch": 0.03, "learning_rate": 4.991854328609778e-05, "loss": 0.6233, "step": 25800 }, { "epoch": 0.03, "learning_rate": 4.991853695037554e-05, "loss": 0.2478, "step": 25801 }, { "epoch": 0.03, "learning_rate": 4.991853061440731e-05, "loss": 0.1664, "step": 25802 }, { "epoch": 0.03, "learning_rate": 4.991852427819309e-05, "loss": 0.1837, "step": 25803 }, { "epoch": 0.03, "learning_rate": 4.991851794173289e-05, "loss": 0.3485, "step": 25804 }, { "epoch": 0.03, "learning_rate": 4.99185116050267e-05, "loss": 0.9135, "step": 25805 }, { "epoch": 0.03, "learning_rate": 4.991850526807453e-05, "loss": 0.8773, "step": 25806 }, { "epoch": 0.03, "learning_rate": 4.9918498930876376e-05, "loss": 1.3027, "step": 25807 }, { "epoch": 0.03, "learning_rate": 4.991849259343223e-05, "loss": 2.0103, "step": 25808 }, { "epoch": 0.03, "learning_rate": 4.9918486255742105e-05, "loss": 1.9137, "step": 25809 }, { "epoch": 0.03, "learning_rate": 4.9918479917805995e-05, "loss": 1.1655, "step": 25810 }, { "epoch": 0.03, "learning_rate": 4.99184735796239e-05, "loss": 1.0848, "step": 25811 }, { "epoch": 0.03, "learning_rate": 4.991846724119581e-05, "loss": 1.3777, "step": 25812 }, { "epoch": 0.03, "learning_rate": 4.991846090252175e-05, "loss": 0.7799, "step": 25813 }, { "epoch": 0.03, "learning_rate": 4.9918454563601704e-05, "loss": 1.4942, "step": 25814 }, { "epoch": 0.03, "learning_rate": 4.991844822443567e-05, "loss": 0.8962, "step": 25815 }, { "epoch": 0.03, "learning_rate": 4.991844188502365e-05, "loss": 0.9722, "step": 25816 }, { "epoch": 0.03, "learning_rate": 4.991843554536564e-05, "loss": 1.0301, "step": 25817 }, { "epoch": 0.03, "learning_rate": 4.991842920546166e-05, "loss": 0.7009, "step": 25818 }, { "epoch": 0.03, "learning_rate": 4.991842286531169e-05, "loss": 1.1413, "step": 25819 }, { "epoch": 0.03, "learning_rate": 4.9918416524915726e-05, "loss": 1.0206, "step": 25820 }, { "epoch": 0.03, "learning_rate": 4.991841018427379e-05, "loss": 0.9912, "step": 25821 }, { "epoch": 0.03, "learning_rate": 4.991840384338586e-05, "loss": 1.1298, "step": 25822 }, { "epoch": 0.03, "learning_rate": 4.991839750225196e-05, "loss": 0.9288, "step": 25823 }, { "epoch": 0.03, "learning_rate": 4.991839116087206e-05, "loss": 0.71, "step": 25824 }, { "epoch": 0.03, "learning_rate": 4.991838481924619e-05, "loss": 0.9027, "step": 25825 }, { "epoch": 0.03, "learning_rate": 4.991837847737433e-05, "loss": 1.3135, "step": 25826 }, { "epoch": 0.03, "learning_rate": 4.991837213525649e-05, "loss": 1.0744, "step": 25827 }, { "epoch": 0.03, "learning_rate": 4.991836579289265e-05, "loss": 1.0782, "step": 25828 }, { "epoch": 0.03, "learning_rate": 4.991835945028285e-05, "loss": 0.4802, "step": 25829 }, { "epoch": 0.03, "learning_rate": 4.991835310742705e-05, "loss": 0.8428, "step": 25830 }, { "epoch": 0.03, "learning_rate": 4.9918346764325274e-05, "loss": 0.9026, "step": 25831 }, { "epoch": 0.03, "learning_rate": 4.991834042097751e-05, "loss": 1.142, "step": 25832 }, { "epoch": 0.03, "learning_rate": 4.9918334077383764e-05, "loss": 0.9227, "step": 25833 }, { "epoch": 0.03, "learning_rate": 4.991832773354404e-05, "loss": 0.7308, "step": 25834 }, { "epoch": 0.03, "learning_rate": 4.9918321389458326e-05, "loss": 0.9323, "step": 25835 }, { "epoch": 0.03, "learning_rate": 4.991831504512663e-05, "loss": 0.9103, "step": 25836 }, { "epoch": 0.03, "learning_rate": 4.9918308700548955e-05, "loss": 0.9816, "step": 25837 }, { "epoch": 0.03, "learning_rate": 4.9918302355725294e-05, "loss": 0.9848, "step": 25838 }, { "epoch": 0.03, "learning_rate": 4.991829601065565e-05, "loss": 0.771, "step": 25839 }, { "epoch": 0.03, "learning_rate": 4.9918289665340026e-05, "loss": 0.7217, "step": 25840 }, { "epoch": 0.03, "learning_rate": 4.991828331977842e-05, "loss": 1.0244, "step": 25841 }, { "epoch": 0.03, "learning_rate": 4.991827697397082e-05, "loss": 1.2314, "step": 25842 }, { "epoch": 0.03, "learning_rate": 4.991827062791725e-05, "loss": 0.8956, "step": 25843 }, { "epoch": 0.03, "learning_rate": 4.991826428161769e-05, "loss": 0.9167, "step": 25844 }, { "epoch": 0.03, "learning_rate": 4.991825793507216e-05, "loss": 0.8992, "step": 25845 }, { "epoch": 0.03, "learning_rate": 4.991825158828063e-05, "loss": 0.9102, "step": 25846 }, { "epoch": 0.03, "learning_rate": 4.9918245241243124e-05, "loss": 1.1132, "step": 25847 }, { "epoch": 0.03, "learning_rate": 4.991823889395963e-05, "loss": 0.9443, "step": 25848 }, { "epoch": 0.03, "learning_rate": 4.991823254643016e-05, "loss": 0.7022, "step": 25849 }, { "epoch": 0.03, "learning_rate": 4.9918226198654715e-05, "loss": 0.8022, "step": 25850 }, { "epoch": 0.03, "learning_rate": 4.991821985063328e-05, "loss": 1.0386, "step": 25851 }, { "epoch": 0.03, "learning_rate": 4.9918213502365864e-05, "loss": 1.1772, "step": 25852 }, { "epoch": 0.03, "learning_rate": 4.991820715385247e-05, "loss": 0.9803, "step": 25853 }, { "epoch": 0.03, "learning_rate": 4.9918200805093086e-05, "loss": 1.204, "step": 25854 }, { "epoch": 0.03, "learning_rate": 4.991819445608772e-05, "loss": 0.7079, "step": 25855 }, { "epoch": 0.03, "learning_rate": 4.9918188106836374e-05, "loss": 0.1484, "step": 25856 }, { "epoch": 0.03, "learning_rate": 4.991818175733905e-05, "loss": 0.3056, "step": 25857 }, { "epoch": 0.03, "learning_rate": 4.991817540759574e-05, "loss": 0.9031, "step": 25858 }, { "epoch": 0.03, "learning_rate": 4.991816905760645e-05, "loss": 1.1465, "step": 25859 }, { "epoch": 0.03, "learning_rate": 4.991816270737118e-05, "loss": 1.2944, "step": 25860 }, { "epoch": 0.03, "learning_rate": 4.9918156356889925e-05, "loss": 0.8954, "step": 25861 }, { "epoch": 0.03, "learning_rate": 4.991815000616269e-05, "loss": 0.9071, "step": 25862 }, { "epoch": 0.03, "learning_rate": 4.9918143655189475e-05, "loss": 1.0322, "step": 25863 }, { "epoch": 0.03, "learning_rate": 4.991813730397028e-05, "loss": 1.1384, "step": 25864 }, { "epoch": 0.03, "learning_rate": 4.99181309525051e-05, "loss": 1.1667, "step": 25865 }, { "epoch": 0.03, "learning_rate": 4.9918124600793935e-05, "loss": 0.9993, "step": 25866 }, { "epoch": 0.03, "learning_rate": 4.99181182488368e-05, "loss": 0.7987, "step": 25867 }, { "epoch": 0.03, "learning_rate": 4.9918111896633674e-05, "loss": 0.7769, "step": 25868 }, { "epoch": 0.03, "learning_rate": 4.991810554418457e-05, "loss": 0.7258, "step": 25869 }, { "epoch": 0.03, "learning_rate": 4.991809919148949e-05, "loss": 0.9819, "step": 25870 }, { "epoch": 0.03, "learning_rate": 4.991809283854842e-05, "loss": 0.9439, "step": 25871 }, { "epoch": 0.03, "learning_rate": 4.991808648536137e-05, "loss": 0.3832, "step": 25872 }, { "epoch": 0.03, "learning_rate": 4.991808013192835e-05, "loss": 1.7812, "step": 25873 }, { "epoch": 0.03, "learning_rate": 4.9918073778249344e-05, "loss": 1.4398, "step": 25874 }, { "epoch": 0.03, "learning_rate": 4.991806742432435e-05, "loss": 1.3273, "step": 25875 }, { "epoch": 0.03, "learning_rate": 4.9918061070153375e-05, "loss": 0.9974, "step": 25876 }, { "epoch": 0.03, "learning_rate": 4.991805471573643e-05, "loss": 0.8572, "step": 25877 }, { "epoch": 0.03, "learning_rate": 4.99180483610735e-05, "loss": 1.1089, "step": 25878 }, { "epoch": 0.03, "learning_rate": 4.991804200616459e-05, "loss": 1.1339, "step": 25879 }, { "epoch": 0.03, "learning_rate": 4.99180356510097e-05, "loss": 1.3066, "step": 25880 }, { "epoch": 0.03, "learning_rate": 4.9918029295608825e-05, "loss": 0.9561, "step": 25881 }, { "epoch": 0.03, "learning_rate": 4.9918022939961974e-05, "loss": 1.0982, "step": 25882 }, { "epoch": 0.03, "learning_rate": 4.991801658406914e-05, "loss": 1.3559, "step": 25883 }, { "epoch": 0.03, "learning_rate": 4.9918010227930325e-05, "loss": 1.2015, "step": 25884 }, { "epoch": 0.03, "learning_rate": 4.991800387154553e-05, "loss": 1.3261, "step": 25885 }, { "epoch": 0.03, "learning_rate": 4.991799751491476e-05, "loss": 1.3883, "step": 25886 }, { "epoch": 0.03, "learning_rate": 4.9917991158038005e-05, "loss": 1.0042, "step": 25887 }, { "epoch": 0.03, "learning_rate": 4.991798480091527e-05, "loss": 0.8556, "step": 25888 }, { "epoch": 0.03, "learning_rate": 4.9917978443546565e-05, "loss": 0.851, "step": 25889 }, { "epoch": 0.03, "learning_rate": 4.991797208593187e-05, "loss": 0.9601, "step": 25890 }, { "epoch": 0.03, "learning_rate": 4.991796572807119e-05, "loss": 1.3565, "step": 25891 }, { "epoch": 0.03, "learning_rate": 4.9917959369964543e-05, "loss": 1.3138, "step": 25892 }, { "epoch": 0.03, "learning_rate": 4.991795301161191e-05, "loss": 1.1688, "step": 25893 }, { "epoch": 0.03, "learning_rate": 4.9917946653013306e-05, "loss": 1.0954, "step": 25894 }, { "epoch": 0.03, "learning_rate": 4.991794029416871e-05, "loss": 0.9156, "step": 25895 }, { "epoch": 0.03, "learning_rate": 4.991793393507814e-05, "loss": 0.8094, "step": 25896 }, { "epoch": 0.03, "learning_rate": 4.9917927575741595e-05, "loss": 1.1329, "step": 25897 }, { "epoch": 0.03, "learning_rate": 4.9917921216159064e-05, "loss": 1.2325, "step": 25898 }, { "epoch": 0.03, "learning_rate": 4.9917914856330554e-05, "loss": 1.2242, "step": 25899 }, { "epoch": 0.03, "learning_rate": 4.991790849625607e-05, "loss": 1.132, "step": 25900 }, { "epoch": 0.03, "learning_rate": 4.99179021359356e-05, "loss": 1.0243, "step": 25901 }, { "epoch": 0.03, "learning_rate": 4.991789577536916e-05, "loss": 0.8993, "step": 25902 }, { "epoch": 0.03, "learning_rate": 4.991788941455674e-05, "loss": 0.7386, "step": 25903 }, { "epoch": 0.03, "learning_rate": 4.9917883053498335e-05, "loss": 0.7659, "step": 25904 }, { "epoch": 0.03, "learning_rate": 4.9917876692193955e-05, "loss": 0.4891, "step": 25905 }, { "epoch": 0.03, "learning_rate": 4.991787033064359e-05, "loss": 0.4536, "step": 25906 }, { "epoch": 0.03, "learning_rate": 4.991786396884725e-05, "loss": 0.3568, "step": 25907 }, { "epoch": 0.03, "learning_rate": 4.991785760680493e-05, "loss": 0.9202, "step": 25908 }, { "epoch": 0.03, "learning_rate": 4.991785124451663e-05, "loss": 0.896, "step": 25909 }, { "epoch": 0.03, "learning_rate": 4.991784488198236e-05, "loss": 1.0383, "step": 25910 }, { "epoch": 0.03, "learning_rate": 4.991783851920211e-05, "loss": 1.0848, "step": 25911 }, { "epoch": 0.03, "learning_rate": 4.991783215617588e-05, "loss": 1.0207, "step": 25912 }, { "epoch": 0.03, "learning_rate": 4.9917825792903666e-05, "loss": 0.8927, "step": 25913 }, { "epoch": 0.03, "learning_rate": 4.991781942938548e-05, "loss": 0.8552, "step": 25914 }, { "epoch": 0.03, "learning_rate": 4.991781306562131e-05, "loss": 0.9227, "step": 25915 }, { "epoch": 0.03, "learning_rate": 4.991780670161117e-05, "loss": 1.1167, "step": 25916 }, { "epoch": 0.03, "learning_rate": 4.991780033735505e-05, "loss": 0.8406, "step": 25917 }, { "epoch": 0.03, "learning_rate": 4.991779397285294e-05, "loss": 0.9079, "step": 25918 }, { "epoch": 0.03, "learning_rate": 4.9917787608104864e-05, "loss": 0.6282, "step": 25919 }, { "epoch": 0.03, "learning_rate": 4.991778124311081e-05, "loss": 0.8382, "step": 25920 }, { "epoch": 0.03, "learning_rate": 4.991777487787077e-05, "loss": 1.0199, "step": 25921 }, { "epoch": 0.03, "learning_rate": 4.991776851238476e-05, "loss": 1.0519, "step": 25922 }, { "epoch": 0.03, "learning_rate": 4.991776214665277e-05, "loss": 1.0039, "step": 25923 }, { "epoch": 0.03, "learning_rate": 4.99177557806748e-05, "loss": 1.2006, "step": 25924 }, { "epoch": 0.03, "learning_rate": 4.991774941445086e-05, "loss": 0.4781, "step": 25925 }, { "epoch": 0.03, "learning_rate": 4.9917743047980936e-05, "loss": 0.2254, "step": 25926 }, { "epoch": 0.03, "learning_rate": 4.991773668126504e-05, "loss": 0.1132, "step": 25927 }, { "epoch": 0.03, "learning_rate": 4.9917730314303156e-05, "loss": 0.1581, "step": 25928 }, { "epoch": 0.03, "learning_rate": 4.99177239470953e-05, "loss": 0.1378, "step": 25929 }, { "epoch": 0.03, "learning_rate": 4.991771757964147e-05, "loss": 0.1471, "step": 25930 }, { "epoch": 0.03, "learning_rate": 4.9917711211941665e-05, "loss": 0.1376, "step": 25931 }, { "epoch": 0.03, "learning_rate": 4.991770484399587e-05, "loss": 0.0799, "step": 25932 }, { "epoch": 0.03, "learning_rate": 4.991769847580411e-05, "loss": 0.1844, "step": 25933 }, { "epoch": 0.03, "learning_rate": 4.991769210736637e-05, "loss": 0.4573, "step": 25934 }, { "epoch": 0.03, "learning_rate": 4.991768573868265e-05, "loss": 0.6143, "step": 25935 }, { "epoch": 0.03, "learning_rate": 4.9917679369752955e-05, "loss": 0.6123, "step": 25936 }, { "epoch": 0.03, "learning_rate": 4.991767300057729e-05, "loss": 0.4783, "step": 25937 }, { "epoch": 0.03, "learning_rate": 4.9917666631155635e-05, "loss": 0.6303, "step": 25938 }, { "epoch": 0.03, "learning_rate": 4.991766026148801e-05, "loss": 0.4937, "step": 25939 }, { "epoch": 0.03, "learning_rate": 4.9917653891574415e-05, "loss": 0.5631, "step": 25940 }, { "epoch": 0.03, "learning_rate": 4.9917647521414834e-05, "loss": 0.5555, "step": 25941 }, { "epoch": 0.03, "learning_rate": 4.9917641151009274e-05, "loss": 0.5362, "step": 25942 }, { "epoch": 0.03, "learning_rate": 4.991763478035775e-05, "loss": 0.5339, "step": 25943 }, { "epoch": 0.03, "learning_rate": 4.991762840946024e-05, "loss": 0.4648, "step": 25944 }, { "epoch": 0.03, "learning_rate": 4.991762203831676e-05, "loss": 0.5782, "step": 25945 }, { "epoch": 0.03, "learning_rate": 4.99176156669273e-05, "loss": 0.5922, "step": 25946 }, { "epoch": 0.03, "learning_rate": 4.991760929529186e-05, "loss": 0.6157, "step": 25947 }, { "epoch": 0.03, "learning_rate": 4.991760292341045e-05, "loss": 0.4761, "step": 25948 }, { "epoch": 0.03, "learning_rate": 4.9917596551283066e-05, "loss": 0.5092, "step": 25949 }, { "epoch": 0.03, "learning_rate": 4.991759017890969e-05, "loss": 0.5757, "step": 25950 }, { "epoch": 0.03, "learning_rate": 4.9917583806290356e-05, "loss": 0.4771, "step": 25951 }, { "epoch": 0.03, "learning_rate": 4.991757743342504e-05, "loss": 0.4629, "step": 25952 }, { "epoch": 0.03, "learning_rate": 4.991757106031375e-05, "loss": 0.5341, "step": 25953 }, { "epoch": 0.03, "learning_rate": 4.991756468695647e-05, "loss": 0.5705, "step": 25954 }, { "epoch": 0.03, "learning_rate": 4.9917558313353235e-05, "loss": 0.5107, "step": 25955 }, { "epoch": 0.03, "learning_rate": 4.991755193950401e-05, "loss": 0.4536, "step": 25956 }, { "epoch": 0.03, "learning_rate": 4.991754556540882e-05, "loss": 0.4818, "step": 25957 }, { "epoch": 0.03, "learning_rate": 4.991753919106765e-05, "loss": 0.5219, "step": 25958 }, { "epoch": 0.03, "learning_rate": 4.991753281648051e-05, "loss": 0.4933, "step": 25959 }, { "epoch": 0.03, "learning_rate": 4.9917526441647385e-05, "loss": 0.5105, "step": 25960 }, { "epoch": 0.03, "learning_rate": 4.991752006656829e-05, "loss": 0.5003, "step": 25961 }, { "epoch": 0.03, "learning_rate": 4.991751369124322e-05, "loss": 0.45, "step": 25962 }, { "epoch": 0.03, "learning_rate": 4.991750731567217e-05, "loss": 0.4207, "step": 25963 }, { "epoch": 0.03, "learning_rate": 4.9917500939855145e-05, "loss": 0.4828, "step": 25964 }, { "epoch": 0.03, "learning_rate": 4.991749456379215e-05, "loss": 0.4453, "step": 25965 }, { "epoch": 0.03, "learning_rate": 4.991748818748317e-05, "loss": 0.4744, "step": 25966 }, { "epoch": 0.03, "learning_rate": 4.9917481810928235e-05, "loss": 0.4235, "step": 25967 }, { "epoch": 0.03, "learning_rate": 4.9917475434127305e-05, "loss": 0.4519, "step": 25968 }, { "epoch": 0.03, "learning_rate": 4.991746905708041e-05, "loss": 0.4124, "step": 25969 }, { "epoch": 0.03, "learning_rate": 4.991746267978754e-05, "loss": 0.3865, "step": 25970 }, { "epoch": 0.03, "learning_rate": 4.9917456302248695e-05, "loss": 0.3994, "step": 25971 }, { "epoch": 0.03, "learning_rate": 4.991744992446388e-05, "loss": 0.3866, "step": 25972 }, { "epoch": 0.03, "learning_rate": 4.991744354643308e-05, "loss": 0.3604, "step": 25973 }, { "epoch": 0.03, "learning_rate": 4.991743716815631e-05, "loss": 0.4321, "step": 25974 }, { "epoch": 0.03, "learning_rate": 4.991743078963357e-05, "loss": 0.462, "step": 25975 }, { "epoch": 0.03, "learning_rate": 4.991742441086484e-05, "loss": 0.4357, "step": 25976 }, { "epoch": 0.03, "learning_rate": 4.9917418031850147e-05, "loss": 0.5157, "step": 25977 }, { "epoch": 0.03, "learning_rate": 4.991741165258949e-05, "loss": 0.5101, "step": 25978 }, { "epoch": 0.03, "learning_rate": 4.991740527308284e-05, "loss": 0.4588, "step": 25979 }, { "epoch": 0.03, "learning_rate": 4.991739889333022e-05, "loss": 0.4946, "step": 25980 }, { "epoch": 0.03, "learning_rate": 4.991739251333164e-05, "loss": 0.4544, "step": 25981 }, { "epoch": 0.03, "learning_rate": 4.9917386133087076e-05, "loss": 0.4413, "step": 25982 }, { "epoch": 0.03, "learning_rate": 4.991737975259654e-05, "loss": 0.4763, "step": 25983 }, { "epoch": 0.03, "learning_rate": 4.9917373371860024e-05, "loss": 0.4482, "step": 25984 }, { "epoch": 0.03, "learning_rate": 4.991736699087754e-05, "loss": 0.4221, "step": 25985 }, { "epoch": 0.03, "learning_rate": 4.991736060964908e-05, "loss": 0.4137, "step": 25986 }, { "epoch": 0.03, "learning_rate": 4.991735422817465e-05, "loss": 0.4145, "step": 25987 }, { "epoch": 0.03, "learning_rate": 4.991734784645424e-05, "loss": 0.3517, "step": 25988 }, { "epoch": 0.03, "learning_rate": 4.991734146448787e-05, "loss": 0.3733, "step": 25989 }, { "epoch": 0.03, "learning_rate": 4.991733508227551e-05, "loss": 0.5152, "step": 25990 }, { "epoch": 0.03, "learning_rate": 4.991732869981718e-05, "loss": 0.5202, "step": 25991 }, { "epoch": 0.03, "learning_rate": 4.9917322317112883e-05, "loss": 0.4793, "step": 25992 }, { "epoch": 0.03, "learning_rate": 4.991731593416261e-05, "loss": 0.471, "step": 25993 }, { "epoch": 0.03, "learning_rate": 4.9917309550966365e-05, "loss": 0.4608, "step": 25994 }, { "epoch": 0.03, "learning_rate": 4.991730316752414e-05, "loss": 0.4078, "step": 25995 }, { "epoch": 0.03, "learning_rate": 4.991729678383595e-05, "loss": 0.4474, "step": 25996 }, { "epoch": 0.03, "learning_rate": 4.991729039990178e-05, "loss": 0.4474, "step": 25997 }, { "epoch": 0.03, "learning_rate": 4.991728401572164e-05, "loss": 0.3836, "step": 25998 }, { "epoch": 0.03, "learning_rate": 4.9917277631295534e-05, "loss": 0.4841, "step": 25999 }, { "epoch": 0.03, "learning_rate": 4.9917271246623446e-05, "loss": 0.4515, "step": 26000 }, { "epoch": 0.03, "eval_loss": 1.0860557556152344, "eval_runtime": 125.5761, "eval_samples_per_second": 11.029, "eval_steps_per_second": 5.519, "step": 26000 }, { "epoch": 0.03, "learning_rate": 4.991726486170539e-05, "loss": 0.4444, "step": 26001 }, { "epoch": 0.03, "learning_rate": 4.991725847654136e-05, "loss": 0.5174, "step": 26002 }, { "epoch": 0.03, "learning_rate": 4.991725209113135e-05, "loss": 0.3774, "step": 26003 }, { "epoch": 0.03, "learning_rate": 4.991724570547538e-05, "loss": 0.4213, "step": 26004 }, { "epoch": 0.03, "learning_rate": 4.9917239319573426e-05, "loss": 0.4221, "step": 26005 }, { "epoch": 0.03, "learning_rate": 4.9917232933425515e-05, "loss": 0.4465, "step": 26006 }, { "epoch": 0.03, "learning_rate": 4.991722654703161e-05, "loss": 0.4283, "step": 26007 }, { "epoch": 0.03, "learning_rate": 4.991722016039175e-05, "loss": 0.4563, "step": 26008 }, { "epoch": 0.03, "learning_rate": 4.991721377350591e-05, "loss": 0.4204, "step": 26009 }, { "epoch": 0.03, "learning_rate": 4.99172073863741e-05, "loss": 0.4499, "step": 26010 }, { "epoch": 0.03, "learning_rate": 4.991720099899632e-05, "loss": 0.509, "step": 26011 }, { "epoch": 0.03, "learning_rate": 4.9917194611372566e-05, "loss": 0.4553, "step": 26012 }, { "epoch": 0.03, "learning_rate": 4.991718822350284e-05, "loss": 0.4346, "step": 26013 }, { "epoch": 0.03, "learning_rate": 4.991718183538714e-05, "loss": 0.4909, "step": 26014 }, { "epoch": 0.03, "learning_rate": 4.991717544702547e-05, "loss": 0.5268, "step": 26015 }, { "epoch": 0.03, "learning_rate": 4.9917169058417835e-05, "loss": 0.4683, "step": 26016 }, { "epoch": 0.03, "learning_rate": 4.991716266956422e-05, "loss": 0.4367, "step": 26017 }, { "epoch": 0.03, "learning_rate": 4.9917156280464627e-05, "loss": 0.4877, "step": 26018 }, { "epoch": 0.03, "learning_rate": 4.991714989111907e-05, "loss": 0.6743, "step": 26019 }, { "epoch": 0.03, "learning_rate": 4.991714350152754e-05, "loss": 0.8407, "step": 26020 }, { "epoch": 0.03, "learning_rate": 4.991713711169004e-05, "loss": 1.0402, "step": 26021 }, { "epoch": 0.03, "learning_rate": 4.991713072160658e-05, "loss": 0.8534, "step": 26022 }, { "epoch": 0.03, "learning_rate": 4.991712433127713e-05, "loss": 0.9654, "step": 26023 }, { "epoch": 0.03, "learning_rate": 4.9917117940701716e-05, "loss": 1.0535, "step": 26024 }, { "epoch": 0.03, "learning_rate": 4.9917111549880325e-05, "loss": 0.9986, "step": 26025 }, { "epoch": 0.03, "learning_rate": 4.991710515881297e-05, "loss": 1.1621, "step": 26026 }, { "epoch": 0.03, "learning_rate": 4.991709876749964e-05, "loss": 0.8025, "step": 26027 }, { "epoch": 0.03, "learning_rate": 4.991709237594034e-05, "loss": 1.4118, "step": 26028 }, { "epoch": 0.03, "learning_rate": 4.991708598413507e-05, "loss": 1.323, "step": 26029 }, { "epoch": 0.03, "learning_rate": 4.991707959208383e-05, "loss": 1.419, "step": 26030 }, { "epoch": 0.03, "learning_rate": 4.9917073199786615e-05, "loss": 1.0978, "step": 26031 }, { "epoch": 0.03, "learning_rate": 4.9917066807243436e-05, "loss": 0.8578, "step": 26032 }, { "epoch": 0.03, "learning_rate": 4.991706041445428e-05, "loss": 1.5992, "step": 26033 }, { "epoch": 0.03, "learning_rate": 4.991705402141916e-05, "loss": 0.8286, "step": 26034 }, { "epoch": 0.03, "learning_rate": 4.9917047628138057e-05, "loss": 0.693, "step": 26035 }, { "epoch": 0.03, "learning_rate": 4.9917041234611e-05, "loss": 0.8298, "step": 26036 }, { "epoch": 0.03, "learning_rate": 4.991703484083796e-05, "loss": 1.204, "step": 26037 }, { "epoch": 0.03, "learning_rate": 4.9917028446818946e-05, "loss": 0.9053, "step": 26038 }, { "epoch": 0.03, "learning_rate": 4.991702205255397e-05, "loss": 1.0469, "step": 26039 }, { "epoch": 0.03, "learning_rate": 4.991701565804303e-05, "loss": 0.9925, "step": 26040 }, { "epoch": 0.03, "learning_rate": 4.9917009263286105e-05, "loss": 1.1052, "step": 26041 }, { "epoch": 0.03, "learning_rate": 4.991700286828322e-05, "loss": 1.1548, "step": 26042 }, { "epoch": 0.03, "learning_rate": 4.991699647303436e-05, "loss": 0.9096, "step": 26043 }, { "epoch": 0.03, "learning_rate": 4.991699007753953e-05, "loss": 0.8057, "step": 26044 }, { "epoch": 0.03, "learning_rate": 4.991698368179874e-05, "loss": 0.6651, "step": 26045 }, { "epoch": 0.03, "learning_rate": 4.991697728581197e-05, "loss": 1.1431, "step": 26046 }, { "epoch": 0.03, "learning_rate": 4.991697088957923e-05, "loss": 1.2183, "step": 26047 }, { "epoch": 0.03, "learning_rate": 4.991696449310051e-05, "loss": 1.0361, "step": 26048 }, { "epoch": 0.03, "learning_rate": 4.9916958096375844e-05, "loss": 0.8701, "step": 26049 }, { "epoch": 0.03, "learning_rate": 4.991695169940519e-05, "loss": 0.6644, "step": 26050 }, { "epoch": 0.03, "learning_rate": 4.991694530218858e-05, "loss": 0.5679, "step": 26051 }, { "epoch": 0.03, "learning_rate": 4.991693890472599e-05, "loss": 0.8478, "step": 26052 }, { "epoch": 0.03, "learning_rate": 4.991693250701743e-05, "loss": 0.8303, "step": 26053 }, { "epoch": 0.03, "learning_rate": 4.9916926109062906e-05, "loss": 1.2064, "step": 26054 }, { "epoch": 0.03, "learning_rate": 4.991691971086241e-05, "loss": 1.2902, "step": 26055 }, { "epoch": 0.03, "learning_rate": 4.9916913312415945e-05, "loss": 1.2241, "step": 26056 }, { "epoch": 0.03, "learning_rate": 4.991690691372352e-05, "loss": 1.374, "step": 26057 }, { "epoch": 0.03, "learning_rate": 4.991690051478511e-05, "loss": 1.188, "step": 26058 }, { "epoch": 0.03, "learning_rate": 4.991689411560074e-05, "loss": 0.965, "step": 26059 }, { "epoch": 0.03, "learning_rate": 4.9916887716170396e-05, "loss": 0.986, "step": 26060 }, { "epoch": 0.03, "learning_rate": 4.991688131649408e-05, "loss": 1.0385, "step": 26061 }, { "epoch": 0.03, "learning_rate": 4.991687491657181e-05, "loss": 0.6881, "step": 26062 }, { "epoch": 0.03, "learning_rate": 4.991686851640356e-05, "loss": 0.7574, "step": 26063 }, { "epoch": 0.03, "learning_rate": 4.991686211598934e-05, "loss": 1.0766, "step": 26064 }, { "epoch": 0.03, "learning_rate": 4.9916855715329156e-05, "loss": 1.1716, "step": 26065 }, { "epoch": 0.03, "learning_rate": 4.9916849314423e-05, "loss": 0.9634, "step": 26066 }, { "epoch": 0.03, "learning_rate": 4.991684291327088e-05, "loss": 0.8014, "step": 26067 }, { "epoch": 0.03, "learning_rate": 4.9916836511872786e-05, "loss": 0.9244, "step": 26068 }, { "epoch": 0.03, "learning_rate": 4.991683011022873e-05, "loss": 0.7757, "step": 26069 }, { "epoch": 0.03, "learning_rate": 4.99168237083387e-05, "loss": 0.9133, "step": 26070 }, { "epoch": 0.03, "learning_rate": 4.99168173062027e-05, "loss": 1.134, "step": 26071 }, { "epoch": 0.03, "learning_rate": 4.991681090382073e-05, "loss": 1.1169, "step": 26072 }, { "epoch": 0.03, "learning_rate": 4.9916804501192804e-05, "loss": 1.3285, "step": 26073 }, { "epoch": 0.03, "learning_rate": 4.991679809831889e-05, "loss": 0.7384, "step": 26074 }, { "epoch": 0.03, "learning_rate": 4.991679169519903e-05, "loss": 0.7999, "step": 26075 }, { "epoch": 0.03, "learning_rate": 4.991678529183319e-05, "loss": 0.9455, "step": 26076 }, { "epoch": 0.03, "learning_rate": 4.9916778888221384e-05, "loss": 1.6089, "step": 26077 }, { "epoch": 0.03, "learning_rate": 4.9916772484363614e-05, "loss": 1.5912, "step": 26078 }, { "epoch": 0.03, "learning_rate": 4.991676608025987e-05, "loss": 0.8181, "step": 26079 }, { "epoch": 0.03, "learning_rate": 4.9916759675910165e-05, "loss": 0.7797, "step": 26080 }, { "epoch": 0.03, "learning_rate": 4.991675327131448e-05, "loss": 1.0226, "step": 26081 }, { "epoch": 0.03, "learning_rate": 4.991674686647284e-05, "loss": 0.5923, "step": 26082 }, { "epoch": 0.03, "learning_rate": 4.991674046138523e-05, "loss": 0.953, "step": 26083 }, { "epoch": 0.03, "learning_rate": 4.991673405605164e-05, "loss": 0.8666, "step": 26084 }, { "epoch": 0.03, "learning_rate": 4.99167276504721e-05, "loss": 0.9239, "step": 26085 }, { "epoch": 0.03, "learning_rate": 4.991672124464658e-05, "loss": 1.1436, "step": 26086 }, { "epoch": 0.03, "learning_rate": 4.9916714838575094e-05, "loss": 1.2577, "step": 26087 }, { "epoch": 0.03, "learning_rate": 4.9916708432257655e-05, "loss": 1.3525, "step": 26088 }, { "epoch": 0.03, "learning_rate": 4.991670202569423e-05, "loss": 0.9972, "step": 26089 }, { "epoch": 0.03, "learning_rate": 4.991669561888485e-05, "loss": 0.9677, "step": 26090 }, { "epoch": 0.03, "learning_rate": 4.9916689211829495e-05, "loss": 0.6115, "step": 26091 }, { "epoch": 0.03, "learning_rate": 4.9916682804528184e-05, "loss": 0.8531, "step": 26092 }, { "epoch": 0.03, "learning_rate": 4.991667639698089e-05, "loss": 1.4811, "step": 26093 }, { "epoch": 0.03, "learning_rate": 4.991666998918764e-05, "loss": 3.8289, "step": 26094 }, { "epoch": 0.03, "learning_rate": 4.991666358114842e-05, "loss": 0.4209, "step": 26095 }, { "epoch": 0.03, "learning_rate": 4.9916657172863234e-05, "loss": 0.1446, "step": 26096 }, { "epoch": 0.03, "learning_rate": 4.9916650764332085e-05, "loss": 0.9051, "step": 26097 }, { "epoch": 0.03, "learning_rate": 4.9916644355554964e-05, "loss": 1.2332, "step": 26098 }, { "epoch": 0.03, "learning_rate": 4.991663794653187e-05, "loss": 1.6801, "step": 26099 }, { "epoch": 0.03, "learning_rate": 4.991663153726283e-05, "loss": 0.8838, "step": 26100 }, { "epoch": 0.03, "learning_rate": 4.9916625127747806e-05, "loss": 1.1228, "step": 26101 }, { "epoch": 0.03, "learning_rate": 4.991661871798682e-05, "loss": 1.0108, "step": 26102 }, { "epoch": 0.03, "learning_rate": 4.991661230797987e-05, "loss": 0.9096, "step": 26103 }, { "epoch": 0.03, "learning_rate": 4.991660589772695e-05, "loss": 1.0328, "step": 26104 }, { "epoch": 0.03, "learning_rate": 4.991659948722807e-05, "loss": 0.9394, "step": 26105 }, { "epoch": 0.03, "learning_rate": 4.991659307648321e-05, "loss": 1.033, "step": 26106 }, { "epoch": 0.03, "learning_rate": 4.991658666549239e-05, "loss": 1.1274, "step": 26107 }, { "epoch": 0.03, "learning_rate": 4.9916580254255616e-05, "loss": 0.9261, "step": 26108 }, { "epoch": 0.03, "learning_rate": 4.991657384277286e-05, "loss": 0.9871, "step": 26109 }, { "epoch": 0.03, "learning_rate": 4.991656743104415e-05, "loss": 1.8349, "step": 26110 }, { "epoch": 0.03, "learning_rate": 4.991656101906947e-05, "loss": 1.0934, "step": 26111 }, { "epoch": 0.03, "learning_rate": 4.9916554606848817e-05, "loss": 1.6266, "step": 26112 }, { "epoch": 0.03, "learning_rate": 4.991654819438221e-05, "loss": 1.6194, "step": 26113 }, { "epoch": 0.03, "learning_rate": 4.991654178166963e-05, "loss": 1.3072, "step": 26114 }, { "epoch": 0.03, "learning_rate": 4.9916535368711095e-05, "loss": 0.8548, "step": 26115 }, { "epoch": 0.03, "learning_rate": 4.991652895550658e-05, "loss": 0.9472, "step": 26116 }, { "epoch": 0.03, "learning_rate": 4.991652254205611e-05, "loss": 1.2554, "step": 26117 }, { "epoch": 0.03, "learning_rate": 4.991651612835967e-05, "loss": 1.2452, "step": 26118 }, { "epoch": 0.03, "learning_rate": 4.991650971441726e-05, "loss": 0.9404, "step": 26119 }, { "epoch": 0.03, "learning_rate": 4.991650330022889e-05, "loss": 0.9088, "step": 26120 }, { "epoch": 0.03, "learning_rate": 4.9916496885794555e-05, "loss": 1.1325, "step": 26121 }, { "epoch": 0.03, "learning_rate": 4.991649047111425e-05, "loss": 1.0911, "step": 26122 }, { "epoch": 0.03, "learning_rate": 4.991648405618799e-05, "loss": 1.2145, "step": 26123 }, { "epoch": 0.03, "learning_rate": 4.991647764101576e-05, "loss": 1.4028, "step": 26124 }, { "epoch": 0.03, "learning_rate": 4.991647122559756e-05, "loss": 1.0197, "step": 26125 }, { "epoch": 0.03, "learning_rate": 4.99164648099334e-05, "loss": 1.0037, "step": 26126 }, { "epoch": 0.03, "learning_rate": 4.9916458394023275e-05, "loss": 0.673, "step": 26127 }, { "epoch": 0.03, "learning_rate": 4.991645197786718e-05, "loss": 1.0895, "step": 26128 }, { "epoch": 0.03, "learning_rate": 4.991644556146513e-05, "loss": 0.9901, "step": 26129 }, { "epoch": 0.03, "learning_rate": 4.991643914481711e-05, "loss": 1.0335, "step": 26130 }, { "epoch": 0.03, "learning_rate": 4.991643272792312e-05, "loss": 1.6946, "step": 26131 }, { "epoch": 0.03, "learning_rate": 4.991642631078318e-05, "loss": 1.2949, "step": 26132 }, { "epoch": 0.03, "learning_rate": 4.991641989339727e-05, "loss": 0.8925, "step": 26133 }, { "epoch": 0.03, "learning_rate": 4.991641347576539e-05, "loss": 0.8859, "step": 26134 }, { "epoch": 0.03, "learning_rate": 4.991640705788754e-05, "loss": 0.8471, "step": 26135 }, { "epoch": 0.03, "learning_rate": 4.9916400639763736e-05, "loss": 0.5979, "step": 26136 }, { "epoch": 0.03, "learning_rate": 4.991639422139397e-05, "loss": 0.6681, "step": 26137 }, { "epoch": 0.03, "learning_rate": 4.9916387802778233e-05, "loss": 0.8831, "step": 26138 }, { "epoch": 0.03, "learning_rate": 4.9916381383916535e-05, "loss": 1.0922, "step": 26139 }, { "epoch": 0.03, "learning_rate": 4.991637496480887e-05, "loss": 1.1613, "step": 26140 }, { "epoch": 0.03, "learning_rate": 4.991636854545524e-05, "loss": 1.1967, "step": 26141 }, { "epoch": 0.03, "learning_rate": 4.991636212585566e-05, "loss": 0.8537, "step": 26142 }, { "epoch": 0.03, "learning_rate": 4.99163557060101e-05, "loss": 1.047, "step": 26143 }, { "epoch": 0.03, "learning_rate": 4.991634928591858e-05, "loss": 0.94, "step": 26144 }, { "epoch": 0.03, "learning_rate": 4.99163428655811e-05, "loss": 1.0826, "step": 26145 }, { "epoch": 0.03, "learning_rate": 4.991633644499765e-05, "loss": 1.0241, "step": 26146 }, { "epoch": 0.03, "learning_rate": 4.9916330024168246e-05, "loss": 1.318, "step": 26147 }, { "epoch": 0.03, "learning_rate": 4.991632360309287e-05, "loss": 1.1236, "step": 26148 }, { "epoch": 0.03, "learning_rate": 4.991631718177153e-05, "loss": 0.5673, "step": 26149 }, { "epoch": 0.03, "learning_rate": 4.991631076020423e-05, "loss": 0.7606, "step": 26150 }, { "epoch": 0.03, "learning_rate": 4.9916304338390976e-05, "loss": 1.5247, "step": 26151 }, { "epoch": 0.03, "learning_rate": 4.991629791633174e-05, "loss": 1.0264, "step": 26152 }, { "epoch": 0.03, "learning_rate": 4.9916291494026556e-05, "loss": 1.2371, "step": 26153 }, { "epoch": 0.03, "learning_rate": 4.9916285071475406e-05, "loss": 0.4817, "step": 26154 }, { "epoch": 0.03, "learning_rate": 4.991627864867829e-05, "loss": 0.7484, "step": 26155 }, { "epoch": 0.03, "learning_rate": 4.991627222563521e-05, "loss": 0.8763, "step": 26156 }, { "epoch": 0.03, "learning_rate": 4.991626580234617e-05, "loss": 0.8567, "step": 26157 }, { "epoch": 0.03, "learning_rate": 4.991625937881116e-05, "loss": 1.2165, "step": 26158 }, { "epoch": 0.03, "learning_rate": 4.9916252955030194e-05, "loss": 1.0351, "step": 26159 }, { "epoch": 0.03, "learning_rate": 4.991624653100327e-05, "loss": 0.5446, "step": 26160 }, { "epoch": 0.03, "learning_rate": 4.991624010673037e-05, "loss": 0.7159, "step": 26161 }, { "epoch": 0.03, "learning_rate": 4.991623368221152e-05, "loss": 0.6068, "step": 26162 }, { "epoch": 0.03, "learning_rate": 4.99162272574467e-05, "loss": 0.4778, "step": 26163 }, { "epoch": 0.03, "learning_rate": 4.991622083243592e-05, "loss": 0.9843, "step": 26164 }, { "epoch": 0.03, "learning_rate": 4.991621440717918e-05, "loss": 0.8191, "step": 26165 }, { "epoch": 0.03, "learning_rate": 4.991620798167648e-05, "loss": 1.2769, "step": 26166 }, { "epoch": 0.03, "learning_rate": 4.991620155592781e-05, "loss": 0.8247, "step": 26167 }, { "epoch": 0.03, "learning_rate": 4.991619512993318e-05, "loss": 0.4417, "step": 26168 }, { "epoch": 0.03, "learning_rate": 4.991618870369259e-05, "loss": 0.2118, "step": 26169 }, { "epoch": 0.03, "learning_rate": 4.991618227720604e-05, "loss": 0.1124, "step": 26170 }, { "epoch": 0.03, "learning_rate": 4.991617585047352e-05, "loss": 0.1774, "step": 26171 }, { "epoch": 0.03, "learning_rate": 4.9916169423495044e-05, "loss": 0.133, "step": 26172 }, { "epoch": 0.03, "learning_rate": 4.9916162996270596e-05, "loss": 0.1695, "step": 26173 }, { "epoch": 0.03, "learning_rate": 4.9916156568800196e-05, "loss": 0.1485, "step": 26174 }, { "epoch": 0.03, "learning_rate": 4.991615014108383e-05, "loss": 0.0928, "step": 26175 }, { "epoch": 0.03, "learning_rate": 4.991614371312151e-05, "loss": 0.1705, "step": 26176 }, { "epoch": 0.03, "learning_rate": 4.991613728491322e-05, "loss": 0.4077, "step": 26177 }, { "epoch": 0.03, "learning_rate": 4.991613085645898e-05, "loss": 0.4945, "step": 26178 }, { "epoch": 0.03, "learning_rate": 4.991612442775877e-05, "loss": 0.5179, "step": 26179 }, { "epoch": 0.03, "learning_rate": 4.991611799881259e-05, "loss": 0.3656, "step": 26180 }, { "epoch": 0.03, "learning_rate": 4.991611156962046e-05, "loss": 0.6605, "step": 26181 }, { "epoch": 0.03, "learning_rate": 4.991610514018237e-05, "loss": 0.5359, "step": 26182 }, { "epoch": 0.03, "learning_rate": 4.991609871049831e-05, "loss": 0.6405, "step": 26183 }, { "epoch": 0.03, "learning_rate": 4.991609228056829e-05, "loss": 0.6068, "step": 26184 }, { "epoch": 0.03, "learning_rate": 4.9916085850392314e-05, "loss": 0.5701, "step": 26185 }, { "epoch": 0.03, "learning_rate": 4.991607941997037e-05, "loss": 0.5832, "step": 26186 }, { "epoch": 0.03, "learning_rate": 4.9916072989302476e-05, "loss": 0.4512, "step": 26187 }, { "epoch": 0.03, "learning_rate": 4.991606655838861e-05, "loss": 0.5943, "step": 26188 }, { "epoch": 0.03, "learning_rate": 4.991606012722879e-05, "loss": 0.4802, "step": 26189 }, { "epoch": 0.03, "learning_rate": 4.9916053695823003e-05, "loss": 0.4161, "step": 26190 }, { "epoch": 0.03, "learning_rate": 4.991604726417126e-05, "loss": 0.3822, "step": 26191 }, { "epoch": 0.03, "learning_rate": 4.991604083227356e-05, "loss": 0.381, "step": 26192 }, { "epoch": 0.03, "learning_rate": 4.9916034400129895e-05, "loss": 0.5058, "step": 26193 }, { "epoch": 0.03, "learning_rate": 4.991602796774027e-05, "loss": 0.5226, "step": 26194 }, { "epoch": 0.03, "learning_rate": 4.991602153510469e-05, "loss": 0.4985, "step": 26195 }, { "epoch": 0.03, "learning_rate": 4.991601510222314e-05, "loss": 0.5016, "step": 26196 }, { "epoch": 0.03, "learning_rate": 4.9916008669095635e-05, "loss": 0.4556, "step": 26197 }, { "epoch": 0.03, "learning_rate": 4.991600223572217e-05, "loss": 0.4792, "step": 26198 }, { "epoch": 0.03, "learning_rate": 4.9915995802102736e-05, "loss": 0.3886, "step": 26199 }, { "epoch": 0.03, "learning_rate": 4.9915989368237346e-05, "loss": 0.4247, "step": 26200 }, { "epoch": 0.03, "learning_rate": 4.9915982934126005e-05, "loss": 0.4942, "step": 26201 }, { "epoch": 0.03, "learning_rate": 4.991597649976869e-05, "loss": 0.4914, "step": 26202 }, { "epoch": 0.03, "learning_rate": 4.991597006516542e-05, "loss": 0.5582, "step": 26203 }, { "epoch": 0.03, "learning_rate": 4.99159636303162e-05, "loss": 0.5923, "step": 26204 }, { "epoch": 0.03, "learning_rate": 4.9915957195221006e-05, "loss": 0.4462, "step": 26205 }, { "epoch": 0.03, "learning_rate": 4.991595075987986e-05, "loss": 0.4232, "step": 26206 }, { "epoch": 0.03, "learning_rate": 4.991594432429275e-05, "loss": 0.4507, "step": 26207 }, { "epoch": 0.03, "learning_rate": 4.9915937888459684e-05, "loss": 0.4294, "step": 26208 }, { "epoch": 0.03, "learning_rate": 4.991593145238066e-05, "loss": 0.4226, "step": 26209 }, { "epoch": 0.03, "learning_rate": 4.9915925016055675e-05, "loss": 0.4418, "step": 26210 }, { "epoch": 0.03, "learning_rate": 4.9915918579484726e-05, "loss": 0.4319, "step": 26211 }, { "epoch": 0.03, "learning_rate": 4.991591214266782e-05, "loss": 0.4723, "step": 26212 }, { "epoch": 0.03, "learning_rate": 4.991590570560496e-05, "loss": 0.3979, "step": 26213 }, { "epoch": 0.03, "learning_rate": 4.991589926829613e-05, "loss": 0.363, "step": 26214 }, { "epoch": 0.03, "learning_rate": 4.991589283074135e-05, "loss": 0.3988, "step": 26215 }, { "epoch": 0.03, "learning_rate": 4.9915886392940606e-05, "loss": 0.3467, "step": 26216 }, { "epoch": 0.03, "learning_rate": 4.991587995489391e-05, "loss": 0.3542, "step": 26217 }, { "epoch": 0.03, "learning_rate": 4.991587351660124e-05, "loss": 0.3993, "step": 26218 }, { "epoch": 0.03, "learning_rate": 4.991586707806263e-05, "loss": 0.4435, "step": 26219 }, { "epoch": 0.03, "learning_rate": 4.9915860639278044e-05, "loss": 0.436, "step": 26220 }, { "epoch": 0.03, "learning_rate": 4.991585420024751e-05, "loss": 0.47, "step": 26221 }, { "epoch": 0.03, "learning_rate": 4.9915847760971014e-05, "loss": 0.5567, "step": 26222 }, { "epoch": 0.03, "learning_rate": 4.9915841321448556e-05, "loss": 0.5489, "step": 26223 }, { "epoch": 0.03, "learning_rate": 4.9915834881680146e-05, "loss": 0.5758, "step": 26224 }, { "epoch": 0.03, "learning_rate": 4.991582844166577e-05, "loss": 0.4574, "step": 26225 }, { "epoch": 0.03, "learning_rate": 4.991582200140544e-05, "loss": 0.4563, "step": 26226 }, { "epoch": 0.03, "learning_rate": 4.991581556089915e-05, "loss": 0.4314, "step": 26227 }, { "epoch": 0.03, "learning_rate": 4.9915809120146906e-05, "loss": 0.43, "step": 26228 }, { "epoch": 0.03, "learning_rate": 4.991580267914869e-05, "loss": 0.3926, "step": 26229 }, { "epoch": 0.03, "learning_rate": 4.9915796237904534e-05, "loss": 0.3889, "step": 26230 }, { "epoch": 0.03, "learning_rate": 4.991578979641441e-05, "loss": 0.4197, "step": 26231 }, { "epoch": 0.03, "learning_rate": 4.9915783354678324e-05, "loss": 0.3731, "step": 26232 }, { "epoch": 0.03, "learning_rate": 4.991577691269629e-05, "loss": 0.3427, "step": 26233 }, { "epoch": 0.03, "learning_rate": 4.9915770470468295e-05, "loss": 0.4582, "step": 26234 }, { "epoch": 0.03, "learning_rate": 4.991576402799434e-05, "loss": 0.5256, "step": 26235 }, { "epoch": 0.03, "learning_rate": 4.9915757585274426e-05, "loss": 0.5136, "step": 26236 }, { "epoch": 0.03, "learning_rate": 4.9915751142308555e-05, "loss": 0.5019, "step": 26237 }, { "epoch": 0.03, "learning_rate": 4.991574469909672e-05, "loss": 0.5632, "step": 26238 }, { "epoch": 0.03, "learning_rate": 4.991573825563893e-05, "loss": 0.5141, "step": 26239 }, { "epoch": 0.03, "learning_rate": 4.9915731811935194e-05, "loss": 0.3972, "step": 26240 }, { "epoch": 0.03, "learning_rate": 4.991572536798549e-05, "loss": 0.4635, "step": 26241 }, { "epoch": 0.03, "learning_rate": 4.9915718923789836e-05, "loss": 0.473, "step": 26242 }, { "epoch": 0.03, "learning_rate": 4.991571247934822e-05, "loss": 0.3901, "step": 26243 }, { "epoch": 0.03, "learning_rate": 4.991570603466064e-05, "loss": 0.4911, "step": 26244 }, { "epoch": 0.03, "learning_rate": 4.9915699589727104e-05, "loss": 0.5677, "step": 26245 }, { "epoch": 0.03, "learning_rate": 4.9915693144547624e-05, "loss": 0.5609, "step": 26246 }, { "epoch": 0.03, "learning_rate": 4.991568669912218e-05, "loss": 0.4634, "step": 26247 }, { "epoch": 0.03, "learning_rate": 4.991568025345077e-05, "loss": 0.3629, "step": 26248 }, { "epoch": 0.03, "learning_rate": 4.991567380753341e-05, "loss": 0.4125, "step": 26249 }, { "epoch": 0.03, "learning_rate": 4.991566736137009e-05, "loss": 0.4555, "step": 26250 }, { "epoch": 0.03, "learning_rate": 4.991566091496081e-05, "loss": 0.4548, "step": 26251 }, { "epoch": 0.03, "learning_rate": 4.9915654468305585e-05, "loss": 0.4108, "step": 26252 }, { "epoch": 0.03, "learning_rate": 4.99156480214044e-05, "loss": 0.4031, "step": 26253 }, { "epoch": 0.03, "learning_rate": 4.9915641574257255e-05, "loss": 0.3731, "step": 26254 }, { "epoch": 0.03, "learning_rate": 4.991563512686415e-05, "loss": 0.3806, "step": 26255 }, { "epoch": 0.03, "learning_rate": 4.991562867922509e-05, "loss": 0.3758, "step": 26256 }, { "epoch": 0.03, "learning_rate": 4.9915622231340074e-05, "loss": 0.4648, "step": 26257 }, { "epoch": 0.03, "learning_rate": 4.991561578320911e-05, "loss": 0.4271, "step": 26258 }, { "epoch": 0.03, "learning_rate": 4.991560933483218e-05, "loss": 0.4682, "step": 26259 }, { "epoch": 0.03, "learning_rate": 4.99156028862093e-05, "loss": 0.5311, "step": 26260 }, { "epoch": 0.03, "learning_rate": 4.9915596437340454e-05, "loss": 0.5002, "step": 26261 }, { "epoch": 0.03, "learning_rate": 4.991558998822566e-05, "loss": 0.4195, "step": 26262 }, { "epoch": 0.03, "learning_rate": 4.9915583538864906e-05, "loss": 0.455, "step": 26263 }, { "epoch": 0.03, "learning_rate": 4.9915577089258195e-05, "loss": 0.5333, "step": 26264 }, { "epoch": 0.03, "learning_rate": 4.9915570639405526e-05, "loss": 0.5352, "step": 26265 }, { "epoch": 0.03, "learning_rate": 4.9915564189306905e-05, "loss": 0.8874, "step": 26266 }, { "epoch": 0.03, "learning_rate": 4.9915557738962334e-05, "loss": 1.104, "step": 26267 }, { "epoch": 0.03, "learning_rate": 4.99155512883718e-05, "loss": 0.9953, "step": 26268 }, { "epoch": 0.03, "learning_rate": 4.991554483753531e-05, "loss": 1.0556, "step": 26269 }, { "epoch": 0.03, "learning_rate": 4.9915538386452864e-05, "loss": 1.2879, "step": 26270 }, { "epoch": 0.03, "learning_rate": 4.991553193512447e-05, "loss": 1.0084, "step": 26271 }, { "epoch": 0.03, "learning_rate": 4.9915525483550105e-05, "loss": 0.7354, "step": 26272 }, { "epoch": 0.03, "learning_rate": 4.99155190317298e-05, "loss": 0.9757, "step": 26273 }, { "epoch": 0.03, "learning_rate": 4.9915512579663534e-05, "loss": 0.886, "step": 26274 }, { "epoch": 0.03, "learning_rate": 4.9915506127351305e-05, "loss": 1.0123, "step": 26275 }, { "epoch": 0.03, "learning_rate": 4.991549967479313e-05, "loss": 1.3076, "step": 26276 }, { "epoch": 0.03, "learning_rate": 4.991549322198899e-05, "loss": 1.104, "step": 26277 }, { "epoch": 0.03, "learning_rate": 4.991548676893891e-05, "loss": 1.0424, "step": 26278 }, { "epoch": 0.03, "learning_rate": 4.991548031564286e-05, "loss": 1.1174, "step": 26279 }, { "epoch": 0.03, "learning_rate": 4.991547386210087e-05, "loss": 1.2435, "step": 26280 }, { "epoch": 0.03, "learning_rate": 4.991546740831291e-05, "loss": 1.1675, "step": 26281 }, { "epoch": 0.03, "learning_rate": 4.9915460954279005e-05, "loss": 1.0732, "step": 26282 }, { "epoch": 0.03, "learning_rate": 4.991545449999914e-05, "loss": 0.8396, "step": 26283 }, { "epoch": 0.03, "learning_rate": 4.991544804547332e-05, "loss": 1.4425, "step": 26284 }, { "epoch": 0.03, "learning_rate": 4.991544159070155e-05, "loss": 0.9062, "step": 26285 }, { "epoch": 0.03, "learning_rate": 4.991543513568382e-05, "loss": 1.0825, "step": 26286 }, { "epoch": 0.03, "learning_rate": 4.991542868042014e-05, "loss": 1.0097, "step": 26287 }, { "epoch": 0.03, "learning_rate": 4.9915422224910504e-05, "loss": 0.9317, "step": 26288 }, { "epoch": 0.03, "learning_rate": 4.99154157691549e-05, "loss": 1.035, "step": 26289 }, { "epoch": 0.03, "learning_rate": 4.9915409313153364e-05, "loss": 1.0449, "step": 26290 }, { "epoch": 0.03, "learning_rate": 4.991540285690586e-05, "loss": 0.9418, "step": 26291 }, { "epoch": 0.03, "learning_rate": 4.9915396400412405e-05, "loss": 0.8218, "step": 26292 }, { "epoch": 0.03, "learning_rate": 4.9915389943673e-05, "loss": 0.9577, "step": 26293 }, { "epoch": 0.03, "learning_rate": 4.9915383486687635e-05, "loss": 0.8339, "step": 26294 }, { "epoch": 0.03, "learning_rate": 4.991537702945631e-05, "loss": 1.0096, "step": 26295 }, { "epoch": 0.03, "learning_rate": 4.991537057197904e-05, "loss": 1.1229, "step": 26296 }, { "epoch": 0.03, "learning_rate": 4.9915364114255815e-05, "loss": 0.9962, "step": 26297 }, { "epoch": 0.03, "learning_rate": 4.991535765628664e-05, "loss": 0.3849, "step": 26298 }, { "epoch": 0.03, "learning_rate": 4.9915351198071505e-05, "loss": 0.2717, "step": 26299 }, { "epoch": 0.03, "learning_rate": 4.991534473961041e-05, "loss": 0.4351, "step": 26300 }, { "epoch": 0.03, "learning_rate": 4.9915338280903376e-05, "loss": 0.2628, "step": 26301 }, { "epoch": 0.03, "learning_rate": 4.991533182195038e-05, "loss": 0.2272, "step": 26302 }, { "epoch": 0.03, "learning_rate": 4.991532536275143e-05, "loss": 0.2606, "step": 26303 }, { "epoch": 0.03, "learning_rate": 4.991531890330653e-05, "loss": 0.1835, "step": 26304 }, { "epoch": 0.03, "learning_rate": 4.991531244361567e-05, "loss": 0.2243, "step": 26305 }, { "epoch": 0.03, "learning_rate": 4.9915305983678864e-05, "loss": 0.1572, "step": 26306 }, { "epoch": 0.03, "learning_rate": 4.991529952349609e-05, "loss": 0.171, "step": 26307 }, { "epoch": 0.03, "learning_rate": 4.9915293063067384e-05, "loss": 0.1675, "step": 26308 }, { "epoch": 0.03, "learning_rate": 4.991528660239271e-05, "loss": 0.8312, "step": 26309 }, { "epoch": 0.03, "learning_rate": 4.9915280141472085e-05, "loss": 1.09, "step": 26310 }, { "epoch": 0.03, "learning_rate": 4.9915273680305515e-05, "loss": 1.16, "step": 26311 }, { "epoch": 0.03, "learning_rate": 4.991526721889298e-05, "loss": 0.7803, "step": 26312 }, { "epoch": 0.03, "learning_rate": 4.99152607572345e-05, "loss": 1.4041, "step": 26313 }, { "epoch": 0.03, "learning_rate": 4.9915254295330065e-05, "loss": 0.9313, "step": 26314 }, { "epoch": 0.03, "learning_rate": 4.991524783317968e-05, "loss": 0.7128, "step": 26315 }, { "epoch": 0.03, "learning_rate": 4.991524137078334e-05, "loss": 0.6648, "step": 26316 }, { "epoch": 0.03, "learning_rate": 4.991523490814105e-05, "loss": 2.4841, "step": 26317 }, { "epoch": 0.03, "learning_rate": 4.99152284452528e-05, "loss": 1.3497, "step": 26318 }, { "epoch": 0.03, "learning_rate": 4.99152219821186e-05, "loss": 0.99, "step": 26319 }, { "epoch": 0.03, "learning_rate": 4.991521551873845e-05, "loss": 0.8246, "step": 26320 }, { "epoch": 0.03, "learning_rate": 4.9915209055112345e-05, "loss": 1.4009, "step": 26321 }, { "epoch": 0.03, "learning_rate": 4.991520259124029e-05, "loss": 0.8341, "step": 26322 }, { "epoch": 0.03, "learning_rate": 4.991519612712228e-05, "loss": 0.9569, "step": 26323 }, { "epoch": 0.03, "learning_rate": 4.9915189662758323e-05, "loss": 0.7435, "step": 26324 }, { "epoch": 0.03, "learning_rate": 4.991518319814841e-05, "loss": 0.615, "step": 26325 }, { "epoch": 0.03, "learning_rate": 4.991517673329254e-05, "loss": 0.9217, "step": 26326 }, { "epoch": 0.03, "learning_rate": 4.991517026819073e-05, "loss": 0.8297, "step": 26327 }, { "epoch": 0.03, "learning_rate": 4.991516380284296e-05, "loss": 0.697, "step": 26328 }, { "epoch": 0.03, "learning_rate": 4.991515733724923e-05, "loss": 0.5306, "step": 26329 }, { "epoch": 0.03, "learning_rate": 4.9915150871409555e-05, "loss": 0.2699, "step": 26330 }, { "epoch": 0.03, "learning_rate": 4.9915144405323933e-05, "loss": 0.6386, "step": 26331 }, { "epoch": 0.03, "learning_rate": 4.9915137938992354e-05, "loss": 0.8911, "step": 26332 }, { "epoch": 0.03, "learning_rate": 4.991513147241482e-05, "loss": 1.2351, "step": 26333 }, { "epoch": 0.03, "learning_rate": 4.991512500559135e-05, "loss": 1.0743, "step": 26334 }, { "epoch": 0.03, "learning_rate": 4.991511853852191e-05, "loss": 0.9627, "step": 26335 }, { "epoch": 0.03, "learning_rate": 4.991511207120653e-05, "loss": 1.0674, "step": 26336 }, { "epoch": 0.03, "learning_rate": 4.99151056036452e-05, "loss": 1.0499, "step": 26337 }, { "epoch": 0.03, "learning_rate": 4.9915099135837904e-05, "loss": 0.5073, "step": 26338 }, { "epoch": 0.03, "learning_rate": 4.9915092667784666e-05, "loss": 0.9845, "step": 26339 }, { "epoch": 0.03, "learning_rate": 4.9915086199485476e-05, "loss": 1.1443, "step": 26340 }, { "epoch": 0.03, "learning_rate": 4.991507973094034e-05, "loss": 0.7054, "step": 26341 }, { "epoch": 0.03, "learning_rate": 4.991507326214925e-05, "loss": 1.1952, "step": 26342 }, { "epoch": 0.03, "learning_rate": 4.9915066793112205e-05, "loss": 1.0224, "step": 26343 }, { "epoch": 0.03, "learning_rate": 4.991506032382921e-05, "loss": 0.9133, "step": 26344 }, { "epoch": 0.03, "learning_rate": 4.9915053854300264e-05, "loss": 1.0111, "step": 26345 }, { "epoch": 0.03, "learning_rate": 4.991504738452537e-05, "loss": 0.9544, "step": 26346 }, { "epoch": 0.03, "learning_rate": 4.991504091450452e-05, "loss": 0.7952, "step": 26347 }, { "epoch": 0.03, "learning_rate": 4.9915034444237724e-05, "loss": 1.3111, "step": 26348 }, { "epoch": 0.03, "learning_rate": 4.991502797372498e-05, "loss": 0.9001, "step": 26349 }, { "epoch": 0.03, "learning_rate": 4.991502150296628e-05, "loss": 1.0702, "step": 26350 }, { "epoch": 0.03, "learning_rate": 4.991501503196163e-05, "loss": 0.8875, "step": 26351 }, { "epoch": 0.03, "learning_rate": 4.991500856071103e-05, "loss": 1.0312, "step": 26352 }, { "epoch": 0.03, "learning_rate": 4.991500208921448e-05, "loss": 0.9989, "step": 26353 }, { "epoch": 0.03, "learning_rate": 4.9914995617471974e-05, "loss": 0.8885, "step": 26354 }, { "epoch": 0.03, "learning_rate": 4.991498914548353e-05, "loss": 0.8667, "step": 26355 }, { "epoch": 0.03, "learning_rate": 4.9914982673249125e-05, "loss": 0.6203, "step": 26356 }, { "epoch": 0.03, "learning_rate": 4.991497620076877e-05, "loss": 0.2184, "step": 26357 }, { "epoch": 0.03, "learning_rate": 4.991496972804247e-05, "loss": 0.2155, "step": 26358 }, { "epoch": 0.03, "learning_rate": 4.991496325507022e-05, "loss": 0.2879, "step": 26359 }, { "epoch": 0.03, "learning_rate": 4.991495678185202e-05, "loss": 0.9102, "step": 26360 }, { "epoch": 0.03, "learning_rate": 4.9914950308387866e-05, "loss": 0.5272, "step": 26361 }, { "epoch": 0.03, "learning_rate": 4.991494383467776e-05, "loss": 0.547, "step": 26362 }, { "epoch": 0.03, "learning_rate": 4.991493736072171e-05, "loss": 0.9327, "step": 26363 }, { "epoch": 0.03, "learning_rate": 4.9914930886519706e-05, "loss": 1.3136, "step": 26364 }, { "epoch": 0.03, "learning_rate": 4.991492441207176e-05, "loss": 0.9932, "step": 26365 }, { "epoch": 0.03, "learning_rate": 4.991491793737786e-05, "loss": 1.0904, "step": 26366 }, { "epoch": 0.03, "learning_rate": 4.9914911462438006e-05, "loss": 1.0791, "step": 26367 }, { "epoch": 0.03, "learning_rate": 4.991490498725221e-05, "loss": 1.0038, "step": 26368 }, { "epoch": 0.03, "learning_rate": 4.9914898511820465e-05, "loss": 0.5293, "step": 26369 }, { "epoch": 0.03, "learning_rate": 4.991489203614276e-05, "loss": 0.9861, "step": 26370 }, { "epoch": 0.03, "learning_rate": 4.991488556021912e-05, "loss": 0.8856, "step": 26371 }, { "epoch": 0.03, "learning_rate": 4.9914879084049515e-05, "loss": 0.9358, "step": 26372 }, { "epoch": 0.03, "learning_rate": 4.991487260763397e-05, "loss": 1.2281, "step": 26373 }, { "epoch": 0.03, "learning_rate": 4.991486613097247e-05, "loss": 1.2161, "step": 26374 }, { "epoch": 0.03, "learning_rate": 4.9914859654065025e-05, "loss": 0.9964, "step": 26375 }, { "epoch": 0.03, "learning_rate": 4.991485317691164e-05, "loss": 0.9988, "step": 26376 }, { "epoch": 0.03, "learning_rate": 4.99148466995123e-05, "loss": 0.9529, "step": 26377 }, { "epoch": 0.03, "learning_rate": 4.9914840221867e-05, "loss": 1.1156, "step": 26378 }, { "epoch": 0.03, "learning_rate": 4.9914833743975765e-05, "loss": 0.9664, "step": 26379 }, { "epoch": 0.03, "learning_rate": 4.991482726583857e-05, "loss": 0.9096, "step": 26380 }, { "epoch": 0.03, "learning_rate": 4.9914820787455435e-05, "loss": 0.8037, "step": 26381 }, { "epoch": 0.03, "learning_rate": 4.991481430882635e-05, "loss": 1.1242, "step": 26382 }, { "epoch": 0.03, "learning_rate": 4.991480782995131e-05, "loss": 0.8462, "step": 26383 }, { "epoch": 0.03, "learning_rate": 4.991480135083033e-05, "loss": 1.0812, "step": 26384 }, { "epoch": 0.03, "learning_rate": 4.99147948714634e-05, "loss": 0.8729, "step": 26385 }, { "epoch": 0.03, "learning_rate": 4.991478839185052e-05, "loss": 0.9842, "step": 26386 }, { "epoch": 0.03, "learning_rate": 4.9914781911991694e-05, "loss": 0.6173, "step": 26387 }, { "epoch": 0.03, "learning_rate": 4.991477543188692e-05, "loss": 0.9025, "step": 26388 }, { "epoch": 0.03, "learning_rate": 4.991476895153619e-05, "loss": 1.1055, "step": 26389 }, { "epoch": 0.03, "learning_rate": 4.991476247093951e-05, "loss": 1.3985, "step": 26390 }, { "epoch": 0.03, "learning_rate": 4.9914755990096896e-05, "loss": 1.0423, "step": 26391 }, { "epoch": 0.03, "learning_rate": 4.991474950900833e-05, "loss": 1.0021, "step": 26392 }, { "epoch": 0.03, "learning_rate": 4.991474302767381e-05, "loss": 1.264, "step": 26393 }, { "epoch": 0.03, "learning_rate": 4.9914736546093346e-05, "loss": 1.1002, "step": 26394 }, { "epoch": 0.03, "learning_rate": 4.991473006426694e-05, "loss": 1.0632, "step": 26395 }, { "epoch": 0.03, "learning_rate": 4.991472358219457e-05, "loss": 0.9114, "step": 26396 }, { "epoch": 0.03, "learning_rate": 4.991471709987626e-05, "loss": 1.0777, "step": 26397 }, { "epoch": 0.03, "learning_rate": 4.9914710617312014e-05, "loss": 0.7861, "step": 26398 }, { "epoch": 0.03, "learning_rate": 4.991470413450181e-05, "loss": 1.4321, "step": 26399 }, { "epoch": 0.03, "learning_rate": 4.991469765144566e-05, "loss": 0.8867, "step": 26400 }, { "epoch": 0.03, "learning_rate": 4.9914691168143555e-05, "loss": 0.9937, "step": 26401 }, { "epoch": 0.03, "learning_rate": 4.9914684684595516e-05, "loss": 0.6895, "step": 26402 }, { "epoch": 0.03, "learning_rate": 4.9914678200801525e-05, "loss": 0.5806, "step": 26403 }, { "epoch": 0.03, "learning_rate": 4.991467171676158e-05, "loss": 0.6155, "step": 26404 }, { "epoch": 0.03, "learning_rate": 4.9914665232475696e-05, "loss": 0.9866, "step": 26405 }, { "epoch": 0.03, "learning_rate": 4.9914658747943865e-05, "loss": 0.768, "step": 26406 }, { "epoch": 0.03, "learning_rate": 4.991465226316608e-05, "loss": 1.1548, "step": 26407 }, { "epoch": 0.03, "learning_rate": 4.9914645778142356e-05, "loss": 1.0235, "step": 26408 }, { "epoch": 0.03, "learning_rate": 4.991463929287268e-05, "loss": 1.4975, "step": 26409 }, { "epoch": 0.03, "learning_rate": 4.991463280735706e-05, "loss": 0.9725, "step": 26410 }, { "epoch": 0.03, "learning_rate": 4.9914626321595495e-05, "loss": 0.7657, "step": 26411 }, { "epoch": 0.03, "learning_rate": 4.991461983558797e-05, "loss": 0.9486, "step": 26412 }, { "epoch": 0.03, "learning_rate": 4.9914613349334514e-05, "loss": 1.0024, "step": 26413 }, { "epoch": 0.03, "learning_rate": 4.991460686283511e-05, "loss": 0.7183, "step": 26414 }, { "epoch": 0.03, "learning_rate": 4.991460037608975e-05, "loss": 0.7768, "step": 26415 }, { "epoch": 0.03, "learning_rate": 4.9914593889098446e-05, "loss": 0.9217, "step": 26416 }, { "epoch": 0.03, "learning_rate": 4.9914587401861205e-05, "loss": 0.8991, "step": 26417 }, { "epoch": 0.03, "learning_rate": 4.9914580914378006e-05, "loss": 0.8451, "step": 26418 }, { "epoch": 0.03, "learning_rate": 4.9914574426648876e-05, "loss": 1.1295, "step": 26419 }, { "epoch": 0.03, "learning_rate": 4.991456793867378e-05, "loss": 0.6534, "step": 26420 }, { "epoch": 0.03, "learning_rate": 4.991456145045275e-05, "loss": 0.8426, "step": 26421 }, { "epoch": 0.03, "learning_rate": 4.991455496198577e-05, "loss": 1.4677, "step": 26422 }, { "epoch": 0.03, "learning_rate": 4.991454847327285e-05, "loss": 1.0968, "step": 26423 }, { "epoch": 0.03, "learning_rate": 4.991454198431398e-05, "loss": 1.138, "step": 26424 }, { "epoch": 0.03, "learning_rate": 4.991453549510916e-05, "loss": 1.223, "step": 26425 }, { "epoch": 0.03, "learning_rate": 4.991452900565841e-05, "loss": 0.9099, "step": 26426 }, { "epoch": 0.03, "learning_rate": 4.9914522515961695e-05, "loss": 0.9461, "step": 26427 }, { "epoch": 0.03, "learning_rate": 4.9914516026019045e-05, "loss": 0.9793, "step": 26428 }, { "epoch": 0.03, "learning_rate": 4.991450953583044e-05, "loss": 1.0168, "step": 26429 }, { "epoch": 0.03, "learning_rate": 4.99145030453959e-05, "loss": 0.9299, "step": 26430 }, { "epoch": 0.03, "learning_rate": 4.9914496554715406e-05, "loss": 1.1653, "step": 26431 }, { "epoch": 0.03, "learning_rate": 4.991449006378898e-05, "loss": 1.3132, "step": 26432 }, { "epoch": 0.03, "learning_rate": 4.991448357261659e-05, "loss": 1.004, "step": 26433 }, { "epoch": 0.03, "learning_rate": 4.991447708119826e-05, "loss": 0.6788, "step": 26434 }, { "epoch": 0.03, "learning_rate": 4.9914470589534e-05, "loss": 1.3737, "step": 26435 }, { "epoch": 0.03, "learning_rate": 4.991446409762378e-05, "loss": 1.2934, "step": 26436 }, { "epoch": 0.03, "learning_rate": 4.991445760546762e-05, "loss": 1.1547, "step": 26437 }, { "epoch": 0.03, "learning_rate": 4.991445111306551e-05, "loss": 0.6727, "step": 26438 }, { "epoch": 0.03, "learning_rate": 4.991444462041746e-05, "loss": 1.095, "step": 26439 }, { "epoch": 0.03, "learning_rate": 4.991443812752347e-05, "loss": 0.7684, "step": 26440 }, { "epoch": 0.03, "learning_rate": 4.9914431634383526e-05, "loss": 1.1079, "step": 26441 }, { "epoch": 0.03, "learning_rate": 4.991442514099764e-05, "loss": 0.9026, "step": 26442 }, { "epoch": 0.03, "learning_rate": 4.991441864736581e-05, "loss": 0.9026, "step": 26443 }, { "epoch": 0.03, "learning_rate": 4.991441215348803e-05, "loss": 1.2051, "step": 26444 }, { "epoch": 0.03, "learning_rate": 4.991440565936432e-05, "loss": 0.9586, "step": 26445 }, { "epoch": 0.03, "learning_rate": 4.9914399164994654e-05, "loss": 0.7224, "step": 26446 }, { "epoch": 0.03, "learning_rate": 4.991439267037904e-05, "loss": 0.6418, "step": 26447 }, { "epoch": 0.03, "learning_rate": 4.9914386175517484e-05, "loss": 0.9031, "step": 26448 }, { "epoch": 0.03, "learning_rate": 4.9914379680409986e-05, "loss": 0.7415, "step": 26449 }, { "epoch": 0.03, "learning_rate": 4.991437318505655e-05, "loss": 1.1432, "step": 26450 }, { "epoch": 0.03, "learning_rate": 4.9914366689457156e-05, "loss": 0.8822, "step": 26451 }, { "epoch": 0.03, "learning_rate": 4.991436019361183e-05, "loss": 1.0814, "step": 26452 }, { "epoch": 0.03, "learning_rate": 4.9914353697520555e-05, "loss": 1.0252, "step": 26453 }, { "epoch": 0.03, "learning_rate": 4.991434720118334e-05, "loss": 1.0778, "step": 26454 }, { "epoch": 0.03, "learning_rate": 4.9914340704600176e-05, "loss": 1.1994, "step": 26455 }, { "epoch": 0.03, "learning_rate": 4.9914334207771074e-05, "loss": 1.0815, "step": 26456 }, { "epoch": 0.03, "learning_rate": 4.991432771069602e-05, "loss": 1.1064, "step": 26457 }, { "epoch": 0.03, "learning_rate": 4.991432121337503e-05, "loss": 0.8877, "step": 26458 }, { "epoch": 0.03, "learning_rate": 4.991431471580809e-05, "loss": 0.6061, "step": 26459 }, { "epoch": 0.03, "learning_rate": 4.991430821799521e-05, "loss": 0.9613, "step": 26460 }, { "epoch": 0.03, "learning_rate": 4.9914301719936386e-05, "loss": 1.0432, "step": 26461 }, { "epoch": 0.03, "learning_rate": 4.9914295221631616e-05, "loss": 1.0025, "step": 26462 }, { "epoch": 0.03, "learning_rate": 4.99142887230809e-05, "loss": 0.6768, "step": 26463 }, { "epoch": 0.03, "learning_rate": 4.9914282224284257e-05, "loss": 1.5083, "step": 26464 }, { "epoch": 0.03, "learning_rate": 4.991427572524165e-05, "loss": 1.1781, "step": 26465 }, { "epoch": 0.03, "learning_rate": 4.991426922595311e-05, "loss": 1.2936, "step": 26466 }, { "epoch": 0.03, "learning_rate": 4.9914262726418634e-05, "loss": 1.0814, "step": 26467 }, { "epoch": 0.03, "learning_rate": 4.9914256226638204e-05, "loss": 1.0378, "step": 26468 }, { "epoch": 0.03, "learning_rate": 4.9914249726611836e-05, "loss": 1.0801, "step": 26469 }, { "epoch": 0.03, "learning_rate": 4.991424322633952e-05, "loss": 0.9379, "step": 26470 }, { "epoch": 0.03, "learning_rate": 4.991423672582126e-05, "loss": 0.8743, "step": 26471 }, { "epoch": 0.03, "learning_rate": 4.991423022505707e-05, "loss": 0.9897, "step": 26472 }, { "epoch": 0.03, "learning_rate": 4.991422372404692e-05, "loss": 0.8483, "step": 26473 }, { "epoch": 0.03, "learning_rate": 4.991421722279084e-05, "loss": 0.8323, "step": 26474 }, { "epoch": 0.03, "learning_rate": 4.991421072128881e-05, "loss": 1.0691, "step": 26475 }, { "epoch": 0.03, "learning_rate": 4.991420421954084e-05, "loss": 0.6878, "step": 26476 }, { "epoch": 0.03, "learning_rate": 4.991419771754693e-05, "loss": 0.8261, "step": 26477 }, { "epoch": 0.03, "learning_rate": 4.9914191215307074e-05, "loss": 0.6114, "step": 26478 }, { "epoch": 0.03, "learning_rate": 4.9914184712821275e-05, "loss": 0.5325, "step": 26479 }, { "epoch": 0.03, "learning_rate": 4.991417821008954e-05, "loss": 0.3695, "step": 26480 }, { "epoch": 0.03, "learning_rate": 4.991417170711186e-05, "loss": 1.0279, "step": 26481 }, { "epoch": 0.03, "learning_rate": 4.991416520388823e-05, "loss": 0.9288, "step": 26482 }, { "epoch": 0.03, "learning_rate": 4.991415870041867e-05, "loss": 1.0958, "step": 26483 }, { "epoch": 0.03, "learning_rate": 4.9914152196703154e-05, "loss": 0.9976, "step": 26484 }, { "epoch": 0.03, "learning_rate": 4.991414569274171e-05, "loss": 0.9787, "step": 26485 }, { "epoch": 0.03, "learning_rate": 4.991413918853431e-05, "loss": 1.2108, "step": 26486 }, { "epoch": 0.03, "learning_rate": 4.9914132684080984e-05, "loss": 1.0233, "step": 26487 }, { "epoch": 0.03, "learning_rate": 4.9914126179381705e-05, "loss": 0.864, "step": 26488 }, { "epoch": 0.03, "learning_rate": 4.991411967443649e-05, "loss": 0.917, "step": 26489 }, { "epoch": 0.03, "learning_rate": 4.991411316924533e-05, "loss": 1.0737, "step": 26490 }, { "epoch": 0.03, "learning_rate": 4.991410666380822e-05, "loss": 0.8926, "step": 26491 }, { "epoch": 0.03, "learning_rate": 4.9914100158125185e-05, "loss": 1.0618, "step": 26492 }, { "epoch": 0.03, "learning_rate": 4.9914093652196204e-05, "loss": 0.8548, "step": 26493 }, { "epoch": 0.03, "learning_rate": 4.991408714602127e-05, "loss": 0.9595, "step": 26494 }, { "epoch": 0.03, "learning_rate": 4.991408063960041e-05, "loss": 0.9115, "step": 26495 }, { "epoch": 0.03, "learning_rate": 4.9914074132933594e-05, "loss": 1.0515, "step": 26496 }, { "epoch": 0.03, "learning_rate": 4.991406762602085e-05, "loss": 0.8174, "step": 26497 }, { "epoch": 0.03, "learning_rate": 4.991406111886216e-05, "loss": 0.9635, "step": 26498 }, { "epoch": 0.03, "learning_rate": 4.991405461145753e-05, "loss": 1.0582, "step": 26499 }, { "epoch": 0.03, "learning_rate": 4.991404810380695e-05, "loss": 1.3509, "step": 26500 }, { "epoch": 0.03, "eval_loss": 1.059253454208374, "eval_runtime": 126.5882, "eval_samples_per_second": 10.941, "eval_steps_per_second": 5.474, "step": 26500 }, { "epoch": 0.03, "learning_rate": 4.991404159591044e-05, "loss": 1.3138, "step": 26501 }, { "epoch": 0.03, "learning_rate": 4.991403508776799e-05, "loss": 1.3593, "step": 26502 }, { "epoch": 0.03, "learning_rate": 4.991402857937959e-05, "loss": 1.3163, "step": 26503 }, { "epoch": 0.03, "learning_rate": 4.991402207074526e-05, "loss": 1.1691, "step": 26504 }, { "epoch": 0.03, "learning_rate": 4.991401556186498e-05, "loss": 1.0829, "step": 26505 }, { "epoch": 0.03, "learning_rate": 4.9914009052738766e-05, "loss": 0.9203, "step": 26506 }, { "epoch": 0.03, "learning_rate": 4.99140025433666e-05, "loss": 1.3035, "step": 26507 }, { "epoch": 0.03, "learning_rate": 4.99139960337485e-05, "loss": 1.0156, "step": 26508 }, { "epoch": 0.03, "learning_rate": 4.9913989523884464e-05, "loss": 0.7801, "step": 26509 }, { "epoch": 0.03, "learning_rate": 4.991398301377449e-05, "loss": 1.0427, "step": 26510 }, { "epoch": 0.03, "learning_rate": 4.9913976503418567e-05, "loss": 0.6046, "step": 26511 }, { "epoch": 0.03, "learning_rate": 4.99139699928167e-05, "loss": 1.2639, "step": 26512 }, { "epoch": 0.03, "learning_rate": 4.9913963481968905e-05, "loss": 0.8549, "step": 26513 }, { "epoch": 0.03, "learning_rate": 4.9913956970875164e-05, "loss": 0.8926, "step": 26514 }, { "epoch": 0.03, "learning_rate": 4.9913950459535486e-05, "loss": 1.1237, "step": 26515 }, { "epoch": 0.03, "learning_rate": 4.9913943947949856e-05, "loss": 1.329, "step": 26516 }, { "epoch": 0.03, "learning_rate": 4.9913937436118296e-05, "loss": 1.0363, "step": 26517 }, { "epoch": 0.03, "learning_rate": 4.99139309240408e-05, "loss": 0.9725, "step": 26518 }, { "epoch": 0.03, "learning_rate": 4.991392441171736e-05, "loss": 1.0001, "step": 26519 }, { "epoch": 0.03, "learning_rate": 4.9913917899147974e-05, "loss": 1.0782, "step": 26520 }, { "epoch": 0.03, "learning_rate": 4.9913911386332656e-05, "loss": 1.1654, "step": 26521 }, { "epoch": 0.03, "learning_rate": 4.99139048732714e-05, "loss": 0.8433, "step": 26522 }, { "epoch": 0.03, "learning_rate": 4.99138983599642e-05, "loss": 0.8733, "step": 26523 }, { "epoch": 0.03, "learning_rate": 4.991389184641107e-05, "loss": 0.6547, "step": 26524 }, { "epoch": 0.03, "learning_rate": 4.9913885332611985e-05, "loss": 0.8079, "step": 26525 }, { "epoch": 0.03, "learning_rate": 4.9913878818566965e-05, "loss": 0.6542, "step": 26526 }, { "epoch": 0.03, "learning_rate": 4.9913872304276014e-05, "loss": 0.5413, "step": 26527 }, { "epoch": 0.03, "learning_rate": 4.991386578973911e-05, "loss": 1.0732, "step": 26528 }, { "epoch": 0.03, "learning_rate": 4.991385927495628e-05, "loss": 0.6916, "step": 26529 }, { "epoch": 0.03, "learning_rate": 4.9913852759927506e-05, "loss": 0.7581, "step": 26530 }, { "epoch": 0.03, "learning_rate": 4.991384624465279e-05, "loss": 0.5425, "step": 26531 }, { "epoch": 0.03, "learning_rate": 4.991383972913214e-05, "loss": 0.8617, "step": 26532 }, { "epoch": 0.03, "learning_rate": 4.9913833213365546e-05, "loss": 0.9565, "step": 26533 }, { "epoch": 0.03, "learning_rate": 4.991382669735302e-05, "loss": 1.348, "step": 26534 }, { "epoch": 0.03, "learning_rate": 4.991382018109455e-05, "loss": 1.1681, "step": 26535 }, { "epoch": 0.03, "learning_rate": 4.991381366459014e-05, "loss": 0.9598, "step": 26536 }, { "epoch": 0.03, "learning_rate": 4.99138071478398e-05, "loss": 0.9269, "step": 26537 }, { "epoch": 0.03, "learning_rate": 4.991380063084351e-05, "loss": 0.6442, "step": 26538 }, { "epoch": 0.03, "learning_rate": 4.991379411360129e-05, "loss": 0.5665, "step": 26539 }, { "epoch": 0.03, "learning_rate": 4.9913787596113126e-05, "loss": 0.7593, "step": 26540 }, { "epoch": 0.03, "learning_rate": 4.9913781078379026e-05, "loss": 0.6642, "step": 26541 }, { "epoch": 0.03, "learning_rate": 4.991377456039898e-05, "loss": 0.8075, "step": 26542 }, { "epoch": 0.03, "learning_rate": 4.991376804217301e-05, "loss": 0.9759, "step": 26543 }, { "epoch": 0.03, "learning_rate": 4.9913761523701094e-05, "loss": 1.0537, "step": 26544 }, { "epoch": 0.03, "learning_rate": 4.9913755004983244e-05, "loss": 0.8617, "step": 26545 }, { "epoch": 0.03, "learning_rate": 4.991374848601945e-05, "loss": 0.9665, "step": 26546 }, { "epoch": 0.03, "learning_rate": 4.9913741966809716e-05, "loss": 1.0809, "step": 26547 }, { "epoch": 0.03, "learning_rate": 4.9913735447354046e-05, "loss": 1.2464, "step": 26548 }, { "epoch": 0.03, "learning_rate": 4.9913728927652445e-05, "loss": 1.1587, "step": 26549 }, { "epoch": 0.03, "learning_rate": 4.99137224077049e-05, "loss": 1.0142, "step": 26550 }, { "epoch": 0.03, "learning_rate": 4.991371588751142e-05, "loss": 1.3479, "step": 26551 }, { "epoch": 0.03, "learning_rate": 4.9913709367072e-05, "loss": 1.1529, "step": 26552 }, { "epoch": 0.03, "learning_rate": 4.991370284638664e-05, "loss": 0.8997, "step": 26553 }, { "epoch": 0.03, "learning_rate": 4.991369632545535e-05, "loss": 0.9531, "step": 26554 }, { "epoch": 0.03, "learning_rate": 4.991368980427812e-05, "loss": 0.9492, "step": 26555 }, { "epoch": 0.03, "learning_rate": 4.991368328285495e-05, "loss": 0.6766, "step": 26556 }, { "epoch": 0.03, "learning_rate": 4.9913676761185843e-05, "loss": 1.1456, "step": 26557 }, { "epoch": 0.03, "learning_rate": 4.99136702392708e-05, "loss": 1.0605, "step": 26558 }, { "epoch": 0.03, "learning_rate": 4.991366371710982e-05, "loss": 0.7864, "step": 26559 }, { "epoch": 0.03, "learning_rate": 4.99136571947029e-05, "loss": 1.0016, "step": 26560 }, { "epoch": 0.03, "learning_rate": 4.991365067205005e-05, "loss": 0.6241, "step": 26561 }, { "epoch": 0.03, "learning_rate": 4.9913644149151256e-05, "loss": 0.8148, "step": 26562 }, { "epoch": 0.03, "learning_rate": 4.991363762600653e-05, "loss": 1.8407, "step": 26563 }, { "epoch": 0.03, "learning_rate": 4.991363110261586e-05, "loss": 1.0944, "step": 26564 }, { "epoch": 0.03, "learning_rate": 4.9913624578979256e-05, "loss": 0.8532, "step": 26565 }, { "epoch": 0.03, "learning_rate": 4.991361805509672e-05, "loss": 1.024, "step": 26566 }, { "epoch": 0.03, "learning_rate": 4.991361153096824e-05, "loss": 0.8708, "step": 26567 }, { "epoch": 0.03, "learning_rate": 4.991360500659383e-05, "loss": 0.738, "step": 26568 }, { "epoch": 0.03, "learning_rate": 4.991359848197348e-05, "loss": 1.0545, "step": 26569 }, { "epoch": 0.03, "learning_rate": 4.9913591957107194e-05, "loss": 0.931, "step": 26570 }, { "epoch": 0.03, "learning_rate": 4.991358543199497e-05, "loss": 0.8857, "step": 26571 }, { "epoch": 0.03, "learning_rate": 4.991357890663681e-05, "loss": 1.04, "step": 26572 }, { "epoch": 0.03, "learning_rate": 4.9913572381032716e-05, "loss": 1.0223, "step": 26573 }, { "epoch": 0.03, "learning_rate": 4.991356585518268e-05, "loss": 0.71, "step": 26574 }, { "epoch": 0.03, "learning_rate": 4.991355932908671e-05, "loss": 1.1412, "step": 26575 }, { "epoch": 0.03, "learning_rate": 4.991355280274481e-05, "loss": 0.8846, "step": 26576 }, { "epoch": 0.03, "learning_rate": 4.991354627615697e-05, "loss": 0.9287, "step": 26577 }, { "epoch": 0.03, "learning_rate": 4.991353974932319e-05, "loss": 0.8072, "step": 26578 }, { "epoch": 0.03, "learning_rate": 4.991353322224348e-05, "loss": 1.0013, "step": 26579 }, { "epoch": 0.03, "learning_rate": 4.991352669491783e-05, "loss": 0.8616, "step": 26580 }, { "epoch": 0.03, "learning_rate": 4.991352016734625e-05, "loss": 1.2211, "step": 26581 }, { "epoch": 0.03, "learning_rate": 4.991351363952873e-05, "loss": 0.8255, "step": 26582 }, { "epoch": 0.03, "learning_rate": 4.9913507111465275e-05, "loss": 0.9199, "step": 26583 }, { "epoch": 0.03, "learning_rate": 4.991350058315588e-05, "loss": 1.0177, "step": 26584 }, { "epoch": 0.03, "learning_rate": 4.991349405460055e-05, "loss": 0.8242, "step": 26585 }, { "epoch": 0.03, "learning_rate": 4.9913487525799294e-05, "loss": 0.9365, "step": 26586 }, { "epoch": 0.03, "learning_rate": 4.991348099675209e-05, "loss": 0.8892, "step": 26587 }, { "epoch": 0.03, "learning_rate": 4.991347446745896e-05, "loss": 1.1601, "step": 26588 }, { "epoch": 0.03, "learning_rate": 4.9913467937919886e-05, "loss": 1.0765, "step": 26589 }, { "epoch": 0.03, "learning_rate": 4.9913461408134886e-05, "loss": 1.088, "step": 26590 }, { "epoch": 0.03, "learning_rate": 4.991345487810395e-05, "loss": 0.626, "step": 26591 }, { "epoch": 0.03, "learning_rate": 4.991344834782707e-05, "loss": 0.9148, "step": 26592 }, { "epoch": 0.03, "learning_rate": 4.991344181730426e-05, "loss": 1.3591, "step": 26593 }, { "epoch": 0.03, "learning_rate": 4.991343528653552e-05, "loss": 1.1861, "step": 26594 }, { "epoch": 0.03, "learning_rate": 4.991342875552084e-05, "loss": 1.084, "step": 26595 }, { "epoch": 0.03, "learning_rate": 4.991342222426023e-05, "loss": 0.7684, "step": 26596 }, { "epoch": 0.03, "learning_rate": 4.9913415692753685e-05, "loss": 0.5724, "step": 26597 }, { "epoch": 0.03, "learning_rate": 4.9913409161001204e-05, "loss": 0.9024, "step": 26598 }, { "epoch": 0.03, "learning_rate": 4.991340262900278e-05, "loss": 0.9817, "step": 26599 }, { "epoch": 0.03, "learning_rate": 4.991339609675842e-05, "loss": 1.0247, "step": 26600 }, { "epoch": 0.03, "learning_rate": 4.991338956426814e-05, "loss": 1.11, "step": 26601 }, { "epoch": 0.03, "learning_rate": 4.9913383031531915e-05, "loss": 1.07, "step": 26602 }, { "epoch": 0.03, "learning_rate": 4.991337649854976e-05, "loss": 0.4502, "step": 26603 }, { "epoch": 0.03, "learning_rate": 4.991336996532167e-05, "loss": 1.0268, "step": 26604 }, { "epoch": 0.03, "learning_rate": 4.991336343184765e-05, "loss": 0.9461, "step": 26605 }, { "epoch": 0.03, "learning_rate": 4.9913356898127684e-05, "loss": 1.0942, "step": 26606 }, { "epoch": 0.03, "learning_rate": 4.991335036416179e-05, "loss": 1.1404, "step": 26607 }, { "epoch": 0.03, "learning_rate": 4.9913343829949966e-05, "loss": 1.3175, "step": 26608 }, { "epoch": 0.03, "learning_rate": 4.9913337295492204e-05, "loss": 1.0565, "step": 26609 }, { "epoch": 0.03, "learning_rate": 4.991333076078851e-05, "loss": 1.235, "step": 26610 }, { "epoch": 0.03, "learning_rate": 4.991332422583888e-05, "loss": 1.1696, "step": 26611 }, { "epoch": 0.03, "learning_rate": 4.991331769064331e-05, "loss": 1.0902, "step": 26612 }, { "epoch": 0.03, "learning_rate": 4.991331115520182e-05, "loss": 0.9244, "step": 26613 }, { "epoch": 0.03, "learning_rate": 4.991330461951439e-05, "loss": 0.8775, "step": 26614 }, { "epoch": 0.03, "learning_rate": 4.991329808358103e-05, "loss": 1.6401, "step": 26615 }, { "epoch": 0.03, "learning_rate": 4.991329154740173e-05, "loss": 1.6216, "step": 26616 }, { "epoch": 0.03, "learning_rate": 4.99132850109765e-05, "loss": 1.0087, "step": 26617 }, { "epoch": 0.03, "learning_rate": 4.991327847430533e-05, "loss": 0.8009, "step": 26618 }, { "epoch": 0.03, "learning_rate": 4.991327193738823e-05, "loss": 0.9941, "step": 26619 }, { "epoch": 0.03, "learning_rate": 4.99132654002252e-05, "loss": 0.9688, "step": 26620 }, { "epoch": 0.03, "learning_rate": 4.9913258862816234e-05, "loss": 0.739, "step": 26621 }, { "epoch": 0.03, "learning_rate": 4.991325232516134e-05, "loss": 1.0117, "step": 26622 }, { "epoch": 0.03, "learning_rate": 4.991324578726051e-05, "loss": 1.0001, "step": 26623 }, { "epoch": 0.03, "learning_rate": 4.991323924911374e-05, "loss": 0.961, "step": 26624 }, { "epoch": 0.03, "learning_rate": 4.991323271072105e-05, "loss": 0.9612, "step": 26625 }, { "epoch": 0.03, "learning_rate": 4.9913226172082414e-05, "loss": 0.9412, "step": 26626 }, { "epoch": 0.03, "learning_rate": 4.991321963319785e-05, "loss": 0.9391, "step": 26627 }, { "epoch": 0.03, "learning_rate": 4.9913213094067356e-05, "loss": 0.9277, "step": 26628 }, { "epoch": 0.03, "learning_rate": 4.991320655469093e-05, "loss": 0.9147, "step": 26629 }, { "epoch": 0.03, "learning_rate": 4.991320001506857e-05, "loss": 1.047, "step": 26630 }, { "epoch": 0.03, "learning_rate": 4.991319347520027e-05, "loss": 0.9517, "step": 26631 }, { "epoch": 0.03, "learning_rate": 4.991318693508605e-05, "loss": 0.9245, "step": 26632 }, { "epoch": 0.03, "learning_rate": 4.991318039472589e-05, "loss": 1.1425, "step": 26633 }, { "epoch": 0.03, "learning_rate": 4.99131738541198e-05, "loss": 1.1568, "step": 26634 }, { "epoch": 0.03, "learning_rate": 4.9913167313267775e-05, "loss": 0.9222, "step": 26635 }, { "epoch": 0.03, "learning_rate": 4.991316077216982e-05, "loss": 0.9955, "step": 26636 }, { "epoch": 0.03, "learning_rate": 4.9913154230825934e-05, "loss": 1.0849, "step": 26637 }, { "epoch": 0.03, "learning_rate": 4.9913147689236107e-05, "loss": 0.8867, "step": 26638 }, { "epoch": 0.03, "learning_rate": 4.9913141147400355e-05, "loss": 0.6078, "step": 26639 }, { "epoch": 0.03, "learning_rate": 4.991313460531867e-05, "loss": 0.6905, "step": 26640 }, { "epoch": 0.03, "learning_rate": 4.991312806299105e-05, "loss": 1.0155, "step": 26641 }, { "epoch": 0.03, "learning_rate": 4.99131215204175e-05, "loss": 0.7934, "step": 26642 }, { "epoch": 0.03, "learning_rate": 4.991311497759803e-05, "loss": 1.1019, "step": 26643 }, { "epoch": 0.03, "learning_rate": 4.9913108434532615e-05, "loss": 0.9067, "step": 26644 }, { "epoch": 0.03, "learning_rate": 4.991310189122127e-05, "loss": 0.7873, "step": 26645 }, { "epoch": 0.03, "learning_rate": 4.9913095347664e-05, "loss": 0.8926, "step": 26646 }, { "epoch": 0.03, "learning_rate": 4.991308880386079e-05, "loss": 0.9431, "step": 26647 }, { "epoch": 0.03, "learning_rate": 4.991308225981165e-05, "loss": 0.7589, "step": 26648 }, { "epoch": 0.03, "learning_rate": 4.991307571551658e-05, "loss": 0.8291, "step": 26649 }, { "epoch": 0.03, "learning_rate": 4.991306917097558e-05, "loss": 0.9818, "step": 26650 }, { "epoch": 0.03, "learning_rate": 4.991306262618865e-05, "loss": 1.209, "step": 26651 }, { "epoch": 0.03, "learning_rate": 4.9913056081155786e-05, "loss": 0.9561, "step": 26652 }, { "epoch": 0.03, "learning_rate": 4.991304953587699e-05, "loss": 0.9088, "step": 26653 }, { "epoch": 0.03, "learning_rate": 4.9913042990352267e-05, "loss": 0.6468, "step": 26654 }, { "epoch": 0.03, "learning_rate": 4.991303644458161e-05, "loss": 1.0286, "step": 26655 }, { "epoch": 0.03, "learning_rate": 4.991302989856502e-05, "loss": 1.1467, "step": 26656 }, { "epoch": 0.03, "learning_rate": 4.99130233523025e-05, "loss": 1.0052, "step": 26657 }, { "epoch": 0.03, "learning_rate": 4.991301680579405e-05, "loss": 1.3352, "step": 26658 }, { "epoch": 0.03, "learning_rate": 4.991301025903967e-05, "loss": 1.2541, "step": 26659 }, { "epoch": 0.03, "learning_rate": 4.991300371203936e-05, "loss": 1.2058, "step": 26660 }, { "epoch": 0.03, "learning_rate": 4.991299716479312e-05, "loss": 1.0115, "step": 26661 }, { "epoch": 0.03, "learning_rate": 4.991299061730095e-05, "loss": 0.9362, "step": 26662 }, { "epoch": 0.03, "learning_rate": 4.991298406956284e-05, "loss": 0.834, "step": 26663 }, { "epoch": 0.03, "learning_rate": 4.9912977521578804e-05, "loss": 1.0199, "step": 26664 }, { "epoch": 0.03, "learning_rate": 4.991297097334884e-05, "loss": 1.1484, "step": 26665 }, { "epoch": 0.03, "learning_rate": 4.991296442487295e-05, "loss": 0.9857, "step": 26666 }, { "epoch": 0.03, "learning_rate": 4.9912957876151125e-05, "loss": 1.0808, "step": 26667 }, { "epoch": 0.03, "learning_rate": 4.991295132718336e-05, "loss": 1.1592, "step": 26668 }, { "epoch": 0.03, "learning_rate": 4.9912944777969684e-05, "loss": 0.9563, "step": 26669 }, { "epoch": 0.03, "learning_rate": 4.991293822851007e-05, "loss": 0.5057, "step": 26670 }, { "epoch": 0.03, "learning_rate": 4.991293167880452e-05, "loss": 1.091, "step": 26671 }, { "epoch": 0.03, "learning_rate": 4.991292512885305e-05, "loss": 0.9267, "step": 26672 }, { "epoch": 0.03, "learning_rate": 4.9912918578655644e-05, "loss": 1.0513, "step": 26673 }, { "epoch": 0.03, "learning_rate": 4.991291202821231e-05, "loss": 0.8995, "step": 26674 }, { "epoch": 0.03, "learning_rate": 4.9912905477523045e-05, "loss": 0.9566, "step": 26675 }, { "epoch": 0.03, "learning_rate": 4.991289892658785e-05, "loss": 0.8715, "step": 26676 }, { "epoch": 0.03, "learning_rate": 4.991289237540672e-05, "loss": 0.9222, "step": 26677 }, { "epoch": 0.03, "learning_rate": 4.991288582397967e-05, "loss": 0.9443, "step": 26678 }, { "epoch": 0.03, "learning_rate": 4.991287927230669e-05, "loss": 1.0669, "step": 26679 }, { "epoch": 0.03, "learning_rate": 4.991287272038777e-05, "loss": 1.1623, "step": 26680 }, { "epoch": 0.03, "learning_rate": 4.9912866168222934e-05, "loss": 1.1085, "step": 26681 }, { "epoch": 0.03, "learning_rate": 4.991285961581216e-05, "loss": 0.6031, "step": 26682 }, { "epoch": 0.03, "learning_rate": 4.991285306315546e-05, "loss": 0.8515, "step": 26683 }, { "epoch": 0.03, "learning_rate": 4.991284651025283e-05, "loss": 1.0193, "step": 26684 }, { "epoch": 0.03, "learning_rate": 4.9912839957104277e-05, "loss": 1.5152, "step": 26685 }, { "epoch": 0.03, "learning_rate": 4.991283340370978e-05, "loss": 1.2933, "step": 26686 }, { "epoch": 0.03, "learning_rate": 4.991282685006937e-05, "loss": 0.8884, "step": 26687 }, { "epoch": 0.03, "learning_rate": 4.9912820296183016e-05, "loss": 0.9091, "step": 26688 }, { "epoch": 0.03, "learning_rate": 4.9912813742050745e-05, "loss": 0.7794, "step": 26689 }, { "epoch": 0.03, "learning_rate": 4.9912807187672537e-05, "loss": 1.0308, "step": 26690 }, { "epoch": 0.03, "learning_rate": 4.9912800633048404e-05, "loss": 1.0187, "step": 26691 }, { "epoch": 0.03, "learning_rate": 4.991279407817835e-05, "loss": 0.8742, "step": 26692 }, { "epoch": 0.03, "learning_rate": 4.991278752306236e-05, "loss": 0.8315, "step": 26693 }, { "epoch": 0.03, "learning_rate": 4.991278096770044e-05, "loss": 1.3056, "step": 26694 }, { "epoch": 0.03, "learning_rate": 4.99127744120926e-05, "loss": 0.8068, "step": 26695 }, { "epoch": 0.03, "learning_rate": 4.991276785623882e-05, "loss": 0.7784, "step": 26696 }, { "epoch": 0.03, "learning_rate": 4.991276130013912e-05, "loss": 0.6242, "step": 26697 }, { "epoch": 0.03, "learning_rate": 4.991275474379349e-05, "loss": 0.9305, "step": 26698 }, { "epoch": 0.03, "learning_rate": 4.9912748187201924e-05, "loss": 1.1838, "step": 26699 }, { "epoch": 0.03, "learning_rate": 4.991274163036443e-05, "loss": 1.0683, "step": 26700 }, { "epoch": 0.03, "learning_rate": 4.991273507328102e-05, "loss": 0.8339, "step": 26701 }, { "epoch": 0.03, "learning_rate": 4.991272851595168e-05, "loss": 0.9989, "step": 26702 }, { "epoch": 0.03, "learning_rate": 4.991272195837641e-05, "loss": 1.0248, "step": 26703 }, { "epoch": 0.03, "learning_rate": 4.99127154005552e-05, "loss": 0.9769, "step": 26704 }, { "epoch": 0.03, "learning_rate": 4.9912708842488076e-05, "loss": 0.51, "step": 26705 }, { "epoch": 0.03, "learning_rate": 4.991270228417502e-05, "loss": 0.7487, "step": 26706 }, { "epoch": 0.03, "learning_rate": 4.991269572561604e-05, "loss": 1.1073, "step": 26707 }, { "epoch": 0.03, "learning_rate": 4.991268916681113e-05, "loss": 0.9207, "step": 26708 }, { "epoch": 0.03, "learning_rate": 4.9912682607760287e-05, "loss": 0.6842, "step": 26709 }, { "epoch": 0.03, "learning_rate": 4.991267604846353e-05, "loss": 1.3604, "step": 26710 }, { "epoch": 0.03, "learning_rate": 4.991266948892083e-05, "loss": 0.9541, "step": 26711 }, { "epoch": 0.03, "learning_rate": 4.991266292913221e-05, "loss": 0.9235, "step": 26712 }, { "epoch": 0.03, "learning_rate": 4.991265636909766e-05, "loss": 0.808, "step": 26713 }, { "epoch": 0.03, "learning_rate": 4.991264980881719e-05, "loss": 0.8906, "step": 26714 }, { "epoch": 0.03, "learning_rate": 4.991264324829078e-05, "loss": 0.9615, "step": 26715 }, { "epoch": 0.03, "learning_rate": 4.9912636687518455e-05, "loss": 0.9409, "step": 26716 }, { "epoch": 0.03, "learning_rate": 4.991263012650019e-05, "loss": 0.8495, "step": 26717 }, { "epoch": 0.03, "learning_rate": 4.991262356523601e-05, "loss": 1.1694, "step": 26718 }, { "epoch": 0.03, "learning_rate": 4.99126170037259e-05, "loss": 0.6641, "step": 26719 }, { "epoch": 0.03, "learning_rate": 4.991261044196987e-05, "loss": 0.8522, "step": 26720 }, { "epoch": 0.03, "learning_rate": 4.99126038799679e-05, "loss": 1.358, "step": 26721 }, { "epoch": 0.03, "learning_rate": 4.991259731772001e-05, "loss": 1.1143, "step": 26722 }, { "epoch": 0.03, "learning_rate": 4.991259075522619e-05, "loss": 1.0975, "step": 26723 }, { "epoch": 0.03, "learning_rate": 4.9912584192486445e-05, "loss": 1.1385, "step": 26724 }, { "epoch": 0.03, "learning_rate": 4.9912577629500776e-05, "loss": 1.267, "step": 26725 }, { "epoch": 0.03, "learning_rate": 4.9912571066269176e-05, "loss": 1.2471, "step": 26726 }, { "epoch": 0.03, "learning_rate": 4.991256450279166e-05, "loss": 0.7951, "step": 26727 }, { "epoch": 0.03, "learning_rate": 4.99125579390682e-05, "loss": 0.8791, "step": 26728 }, { "epoch": 0.03, "learning_rate": 4.991255137509882e-05, "loss": 0.9081, "step": 26729 }, { "epoch": 0.03, "learning_rate": 4.9912544810883526e-05, "loss": 0.9096, "step": 26730 }, { "epoch": 0.03, "learning_rate": 4.99125382464223e-05, "loss": 1.2326, "step": 26731 }, { "epoch": 0.03, "learning_rate": 4.991253168171515e-05, "loss": 1.0025, "step": 26732 }, { "epoch": 0.03, "learning_rate": 4.9912525116762064e-05, "loss": 1.1107, "step": 26733 }, { "epoch": 0.03, "learning_rate": 4.991251855156306e-05, "loss": 0.9335, "step": 26734 }, { "epoch": 0.03, "learning_rate": 4.9912511986118126e-05, "loss": 0.9886, "step": 26735 }, { "epoch": 0.03, "learning_rate": 4.991250542042727e-05, "loss": 0.8375, "step": 26736 }, { "epoch": 0.03, "learning_rate": 4.991249885449048e-05, "loss": 0.9776, "step": 26737 }, { "epoch": 0.03, "learning_rate": 4.9912492288307774e-05, "loss": 1.0044, "step": 26738 }, { "epoch": 0.03, "learning_rate": 4.991248572187913e-05, "loss": 1.0527, "step": 26739 }, { "epoch": 0.03, "learning_rate": 4.9912479155204574e-05, "loss": 1.5724, "step": 26740 }, { "epoch": 0.03, "learning_rate": 4.9912472588284084e-05, "loss": 1.2569, "step": 26741 }, { "epoch": 0.03, "learning_rate": 4.991246602111768e-05, "loss": 1.2253, "step": 26742 }, { "epoch": 0.03, "learning_rate": 4.991245945370534e-05, "loss": 1.0364, "step": 26743 }, { "epoch": 0.03, "learning_rate": 4.991245288604708e-05, "loss": 1.1396, "step": 26744 }, { "epoch": 0.03, "learning_rate": 4.991244631814289e-05, "loss": 0.9474, "step": 26745 }, { "epoch": 0.03, "learning_rate": 4.991243974999278e-05, "loss": 1.3108, "step": 26746 }, { "epoch": 0.03, "learning_rate": 4.991243318159674e-05, "loss": 1.1646, "step": 26747 }, { "epoch": 0.03, "learning_rate": 4.991242661295478e-05, "loss": 0.8901, "step": 26748 }, { "epoch": 0.03, "learning_rate": 4.991242004406689e-05, "loss": 0.9603, "step": 26749 }, { "epoch": 0.03, "learning_rate": 4.991241347493307e-05, "loss": 1.125, "step": 26750 }, { "epoch": 0.03, "learning_rate": 4.991240690555333e-05, "loss": 0.8842, "step": 26751 }, { "epoch": 0.03, "learning_rate": 4.991240033592767e-05, "loss": 0.2774, "step": 26752 }, { "epoch": 0.03, "learning_rate": 4.991239376605609e-05, "loss": 0.6267, "step": 26753 }, { "epoch": 0.03, "learning_rate": 4.991238719593857e-05, "loss": 0.9615, "step": 26754 }, { "epoch": 0.03, "learning_rate": 4.991238062557514e-05, "loss": 0.7919, "step": 26755 }, { "epoch": 0.03, "learning_rate": 4.991237405496577e-05, "loss": 0.9506, "step": 26756 }, { "epoch": 0.03, "learning_rate": 4.991236748411049e-05, "loss": 1.3704, "step": 26757 }, { "epoch": 0.03, "learning_rate": 4.991236091300928e-05, "loss": 0.5962, "step": 26758 }, { "epoch": 0.03, "learning_rate": 4.991235434166215e-05, "loss": 1.0961, "step": 26759 }, { "epoch": 0.03, "learning_rate": 4.991234777006909e-05, "loss": 0.9104, "step": 26760 }, { "epoch": 0.03, "learning_rate": 4.99123411982301e-05, "loss": 0.9005, "step": 26761 }, { "epoch": 0.03, "learning_rate": 4.99123346261452e-05, "loss": 0.9287, "step": 26762 }, { "epoch": 0.03, "learning_rate": 4.9912328053814364e-05, "loss": 1.0366, "step": 26763 }, { "epoch": 0.03, "learning_rate": 4.991232148123761e-05, "loss": 0.8067, "step": 26764 }, { "epoch": 0.03, "learning_rate": 4.991231490841493e-05, "loss": 0.8784, "step": 26765 }, { "epoch": 0.03, "learning_rate": 4.991230833534632e-05, "loss": 1.165, "step": 26766 }, { "epoch": 0.03, "learning_rate": 4.99123017620318e-05, "loss": 1.1018, "step": 26767 }, { "epoch": 0.03, "learning_rate": 4.9912295188471356e-05, "loss": 1.1175, "step": 26768 }, { "epoch": 0.03, "learning_rate": 4.991228861466498e-05, "loss": 1.15, "step": 26769 }, { "epoch": 0.03, "learning_rate": 4.991228204061268e-05, "loss": 1.0813, "step": 26770 }, { "epoch": 0.03, "learning_rate": 4.9912275466314465e-05, "loss": 1.2602, "step": 26771 }, { "epoch": 0.03, "learning_rate": 4.991226889177031e-05, "loss": 0.9275, "step": 26772 }, { "epoch": 0.03, "learning_rate": 4.991226231698024e-05, "loss": 0.8742, "step": 26773 }, { "epoch": 0.03, "learning_rate": 4.9912255741944256e-05, "loss": 0.7898, "step": 26774 }, { "epoch": 0.03, "learning_rate": 4.991224916666234e-05, "loss": 0.9416, "step": 26775 }, { "epoch": 0.03, "learning_rate": 4.99122425911345e-05, "loss": 1.0406, "step": 26776 }, { "epoch": 0.03, "learning_rate": 4.9912236015360744e-05, "loss": 0.7723, "step": 26777 }, { "epoch": 0.03, "learning_rate": 4.991222943934106e-05, "loss": 1.1298, "step": 26778 }, { "epoch": 0.03, "learning_rate": 4.991222286307545e-05, "loss": 0.8745, "step": 26779 }, { "epoch": 0.03, "learning_rate": 4.9912216286563914e-05, "loss": 0.7528, "step": 26780 }, { "epoch": 0.03, "learning_rate": 4.9912209709806465e-05, "loss": 0.7908, "step": 26781 }, { "epoch": 0.03, "learning_rate": 4.991220313280309e-05, "loss": 1.2125, "step": 26782 }, { "epoch": 0.03, "learning_rate": 4.9912196555553794e-05, "loss": 0.8367, "step": 26783 }, { "epoch": 0.03, "learning_rate": 4.9912189978058566e-05, "loss": 0.6942, "step": 26784 }, { "epoch": 0.03, "learning_rate": 4.991218340031743e-05, "loss": 0.8158, "step": 26785 }, { "epoch": 0.03, "learning_rate": 4.991217682233036e-05, "loss": 0.8163, "step": 26786 }, { "epoch": 0.03, "learning_rate": 4.991217024409737e-05, "loss": 0.7397, "step": 26787 }, { "epoch": 0.03, "learning_rate": 4.991216366561846e-05, "loss": 1.2894, "step": 26788 }, { "epoch": 0.03, "learning_rate": 4.991215708689362e-05, "loss": 0.9577, "step": 26789 }, { "epoch": 0.03, "learning_rate": 4.991215050792287e-05, "loss": 1.1177, "step": 26790 }, { "epoch": 0.03, "learning_rate": 4.991214392870619e-05, "loss": 1.2259, "step": 26791 }, { "epoch": 0.03, "learning_rate": 4.991213734924359e-05, "loss": 0.9205, "step": 26792 }, { "epoch": 0.03, "learning_rate": 4.9912130769535065e-05, "loss": 0.5936, "step": 26793 }, { "epoch": 0.03, "learning_rate": 4.991212418958062e-05, "loss": 0.3901, "step": 26794 }, { "epoch": 0.03, "learning_rate": 4.991211760938025e-05, "loss": 1.0342, "step": 26795 }, { "epoch": 0.03, "learning_rate": 4.9912111028933964e-05, "loss": 1.1426, "step": 26796 }, { "epoch": 0.03, "learning_rate": 4.9912104448241755e-05, "loss": 1.0071, "step": 26797 }, { "epoch": 0.03, "learning_rate": 4.991209786730362e-05, "loss": 0.8949, "step": 26798 }, { "epoch": 0.03, "learning_rate": 4.9912091286119566e-05, "loss": 0.6756, "step": 26799 }, { "epoch": 0.03, "learning_rate": 4.9912084704689585e-05, "loss": 1.0531, "step": 26800 }, { "epoch": 0.03, "learning_rate": 4.991207812301369e-05, "loss": 1.006, "step": 26801 }, { "epoch": 0.03, "learning_rate": 4.991207154109188e-05, "loss": 0.8581, "step": 26802 }, { "epoch": 0.03, "learning_rate": 4.991206495892413e-05, "loss": 0.9681, "step": 26803 }, { "epoch": 0.03, "learning_rate": 4.991205837651047e-05, "loss": 1.0809, "step": 26804 }, { "epoch": 0.03, "learning_rate": 4.991205179385089e-05, "loss": 1.1244, "step": 26805 }, { "epoch": 0.03, "learning_rate": 4.991204521094538e-05, "loss": 0.9817, "step": 26806 }, { "epoch": 0.03, "learning_rate": 4.9912038627793956e-05, "loss": 1.3168, "step": 26807 }, { "epoch": 0.03, "learning_rate": 4.991203204439661e-05, "loss": 1.3964, "step": 26808 }, { "epoch": 0.03, "learning_rate": 4.9912025460753344e-05, "loss": 1.0439, "step": 26809 }, { "epoch": 0.03, "learning_rate": 4.9912018876864155e-05, "loss": 1.1306, "step": 26810 }, { "epoch": 0.03, "learning_rate": 4.991201229272904e-05, "loss": 1.3033, "step": 26811 }, { "epoch": 0.03, "learning_rate": 4.991200570834801e-05, "loss": 1.3888, "step": 26812 }, { "epoch": 0.03, "learning_rate": 4.991199912372106e-05, "loss": 1.2217, "step": 26813 }, { "epoch": 0.03, "learning_rate": 4.9911992538848186e-05, "loss": 0.8826, "step": 26814 }, { "epoch": 0.03, "learning_rate": 4.9911985953729396e-05, "loss": 0.6732, "step": 26815 }, { "epoch": 0.03, "learning_rate": 4.991197936836468e-05, "loss": 0.93, "step": 26816 }, { "epoch": 0.03, "learning_rate": 4.9911972782754045e-05, "loss": 0.9521, "step": 26817 }, { "epoch": 0.03, "learning_rate": 4.991196619689749e-05, "loss": 0.9679, "step": 26818 }, { "epoch": 0.03, "learning_rate": 4.991195961079502e-05, "loss": 0.8592, "step": 26819 }, { "epoch": 0.03, "learning_rate": 4.991195302444662e-05, "loss": 1.4616, "step": 26820 }, { "epoch": 0.03, "learning_rate": 4.991194643785231e-05, "loss": 1.1019, "step": 26821 }, { "epoch": 0.03, "learning_rate": 4.991193985101207e-05, "loss": 1.008, "step": 26822 }, { "epoch": 0.03, "learning_rate": 4.9911933263925914e-05, "loss": 1.0432, "step": 26823 }, { "epoch": 0.03, "learning_rate": 4.9911926676593835e-05, "loss": 1.125, "step": 26824 }, { "epoch": 0.03, "learning_rate": 4.991192008901584e-05, "loss": 0.9843, "step": 26825 }, { "epoch": 0.03, "learning_rate": 4.9911913501191924e-05, "loss": 0.9426, "step": 26826 }, { "epoch": 0.03, "learning_rate": 4.9911906913122085e-05, "loss": 0.9161, "step": 26827 }, { "epoch": 0.03, "learning_rate": 4.991190032480633e-05, "loss": 1.0644, "step": 26828 }, { "epoch": 0.03, "learning_rate": 4.991189373624465e-05, "loss": 1.4092, "step": 26829 }, { "epoch": 0.03, "learning_rate": 4.991188714743706e-05, "loss": 1.0138, "step": 26830 }, { "epoch": 0.03, "learning_rate": 4.991188055838354e-05, "loss": 1.1918, "step": 26831 }, { "epoch": 0.03, "learning_rate": 4.991187396908411e-05, "loss": 1.2013, "step": 26832 }, { "epoch": 0.03, "learning_rate": 4.9911867379538755e-05, "loss": 1.0369, "step": 26833 }, { "epoch": 0.03, "learning_rate": 4.9911860789747475e-05, "loss": 1.2653, "step": 26834 }, { "epoch": 0.03, "learning_rate": 4.9911854199710284e-05, "loss": 1.2287, "step": 26835 }, { "epoch": 0.03, "learning_rate": 4.9911847609427175e-05, "loss": 0.9346, "step": 26836 }, { "epoch": 0.03, "learning_rate": 4.991184101889814e-05, "loss": 0.8314, "step": 26837 }, { "epoch": 0.03, "learning_rate": 4.991183442812319e-05, "loss": 0.8721, "step": 26838 }, { "epoch": 0.03, "learning_rate": 4.991182783710232e-05, "loss": 1.0569, "step": 26839 }, { "epoch": 0.03, "learning_rate": 4.9911821245835534e-05, "loss": 1.2035, "step": 26840 }, { "epoch": 0.03, "learning_rate": 4.991181465432282e-05, "loss": 1.5467, "step": 26841 }, { "epoch": 0.03, "learning_rate": 4.991180806256419e-05, "loss": 1.0387, "step": 26842 }, { "epoch": 0.03, "learning_rate": 4.991180147055965e-05, "loss": 0.8754, "step": 26843 }, { "epoch": 0.03, "learning_rate": 4.991179487830918e-05, "loss": 0.6551, "step": 26844 }, { "epoch": 0.03, "learning_rate": 4.99117882858128e-05, "loss": 0.9056, "step": 26845 }, { "epoch": 0.03, "learning_rate": 4.99117816930705e-05, "loss": 1.2566, "step": 26846 }, { "epoch": 0.03, "learning_rate": 4.991177510008227e-05, "loss": 1.1585, "step": 26847 }, { "epoch": 0.03, "learning_rate": 4.9911768506848135e-05, "loss": 1.1097, "step": 26848 }, { "epoch": 0.03, "learning_rate": 4.991176191336808e-05, "loss": 0.9204, "step": 26849 }, { "epoch": 0.03, "learning_rate": 4.991175531964211e-05, "loss": 0.854, "step": 26850 }, { "epoch": 0.03, "learning_rate": 4.9911748725670205e-05, "loss": 1.2184, "step": 26851 }, { "epoch": 0.03, "learning_rate": 4.991174213145239e-05, "loss": 1.29, "step": 26852 }, { "epoch": 0.03, "learning_rate": 4.991173553698866e-05, "loss": 1.1112, "step": 26853 }, { "epoch": 0.03, "learning_rate": 4.991172894227901e-05, "loss": 1.1708, "step": 26854 }, { "epoch": 0.03, "learning_rate": 4.9911722347323446e-05, "loss": 1.2395, "step": 26855 }, { "epoch": 0.03, "learning_rate": 4.9911715752121956e-05, "loss": 1.1501, "step": 26856 }, { "epoch": 0.03, "learning_rate": 4.9911709156674556e-05, "loss": 0.8496, "step": 26857 }, { "epoch": 0.03, "learning_rate": 4.991170256098123e-05, "loss": 0.9385, "step": 26858 }, { "epoch": 0.03, "learning_rate": 4.9911695965041996e-05, "loss": 1.0147, "step": 26859 }, { "epoch": 0.03, "learning_rate": 4.991168936885684e-05, "loss": 1.1267, "step": 26860 }, { "epoch": 0.03, "learning_rate": 4.991168277242576e-05, "loss": 1.0359, "step": 26861 }, { "epoch": 0.03, "learning_rate": 4.991167617574877e-05, "loss": 0.9257, "step": 26862 }, { "epoch": 0.03, "learning_rate": 4.991166957882586e-05, "loss": 0.8573, "step": 26863 }, { "epoch": 0.03, "learning_rate": 4.991166298165704e-05, "loss": 0.9707, "step": 26864 }, { "epoch": 0.03, "learning_rate": 4.991165638424229e-05, "loss": 1.1, "step": 26865 }, { "epoch": 0.03, "learning_rate": 4.991164978658163e-05, "loss": 0.95, "step": 26866 }, { "epoch": 0.03, "learning_rate": 4.9911643188675053e-05, "loss": 0.5659, "step": 26867 }, { "epoch": 0.03, "learning_rate": 4.991163659052255e-05, "loss": 1.2734, "step": 26868 }, { "epoch": 0.03, "learning_rate": 4.991162999212414e-05, "loss": 0.8616, "step": 26869 }, { "epoch": 0.03, "learning_rate": 4.9911623393479806e-05, "loss": 1.0021, "step": 26870 }, { "epoch": 0.03, "learning_rate": 4.9911616794589555e-05, "loss": 0.8867, "step": 26871 }, { "epoch": 0.03, "learning_rate": 4.9911610195453394e-05, "loss": 1.1057, "step": 26872 }, { "epoch": 0.03, "learning_rate": 4.991160359607131e-05, "loss": 0.9041, "step": 26873 }, { "epoch": 0.03, "learning_rate": 4.991159699644331e-05, "loss": 1.2288, "step": 26874 }, { "epoch": 0.03, "learning_rate": 4.99115903965694e-05, "loss": 0.9301, "step": 26875 }, { "epoch": 0.03, "learning_rate": 4.991158379644956e-05, "loss": 1.2943, "step": 26876 }, { "epoch": 0.03, "learning_rate": 4.991157719608381e-05, "loss": 1.1801, "step": 26877 }, { "epoch": 0.03, "learning_rate": 4.991157059547215e-05, "loss": 0.8251, "step": 26878 }, { "epoch": 0.03, "learning_rate": 4.9911563994614564e-05, "loss": 0.9208, "step": 26879 }, { "epoch": 0.03, "learning_rate": 4.9911557393511064e-05, "loss": 1.1103, "step": 26880 }, { "epoch": 0.03, "learning_rate": 4.991155079216165e-05, "loss": 0.972, "step": 26881 }, { "epoch": 0.03, "learning_rate": 4.9911544190566325e-05, "loss": 0.9369, "step": 26882 }, { "epoch": 0.03, "learning_rate": 4.991153758872508e-05, "loss": 0.7588, "step": 26883 }, { "epoch": 0.03, "learning_rate": 4.991153098663791e-05, "loss": 0.8079, "step": 26884 }, { "epoch": 0.03, "learning_rate": 4.991152438430483e-05, "loss": 0.98, "step": 26885 }, { "epoch": 0.03, "learning_rate": 4.991151778172584e-05, "loss": 0.7687, "step": 26886 }, { "epoch": 0.03, "learning_rate": 4.991151117890092e-05, "loss": 0.7565, "step": 26887 }, { "epoch": 0.03, "learning_rate": 4.991150457583009e-05, "loss": 1.194, "step": 26888 }, { "epoch": 0.03, "learning_rate": 4.9911497972513346e-05, "loss": 0.8899, "step": 26889 }, { "epoch": 0.03, "learning_rate": 4.991149136895069e-05, "loss": 1.291, "step": 26890 }, { "epoch": 0.03, "learning_rate": 4.9911484765142115e-05, "loss": 0.7324, "step": 26891 }, { "epoch": 0.03, "learning_rate": 4.991147816108762e-05, "loss": 0.883, "step": 26892 }, { "epoch": 0.03, "learning_rate": 4.991147155678721e-05, "loss": 1.1096, "step": 26893 }, { "epoch": 0.03, "learning_rate": 4.99114649522409e-05, "loss": 0.7795, "step": 26894 }, { "epoch": 0.03, "learning_rate": 4.9911458347448656e-05, "loss": 0.8717, "step": 26895 }, { "epoch": 0.03, "learning_rate": 4.9911451742410505e-05, "loss": 0.9188, "step": 26896 }, { "epoch": 0.03, "learning_rate": 4.9911445137126436e-05, "loss": 1.046, "step": 26897 }, { "epoch": 0.03, "learning_rate": 4.9911438531596457e-05, "loss": 0.8801, "step": 26898 }, { "epoch": 0.03, "learning_rate": 4.991143192582055e-05, "loss": 0.9049, "step": 26899 }, { "epoch": 0.03, "learning_rate": 4.991142531979874e-05, "loss": 1.2029, "step": 26900 }, { "epoch": 0.03, "learning_rate": 4.991141871353101e-05, "loss": 1.2144, "step": 26901 }, { "epoch": 0.03, "learning_rate": 4.9911412107017365e-05, "loss": 1.3744, "step": 26902 }, { "epoch": 0.03, "learning_rate": 4.99114055002578e-05, "loss": 1.3262, "step": 26903 }, { "epoch": 0.03, "learning_rate": 4.991139889325233e-05, "loss": 1.5156, "step": 26904 }, { "epoch": 0.03, "learning_rate": 4.991139228600094e-05, "loss": 1.0484, "step": 26905 }, { "epoch": 0.03, "learning_rate": 4.9911385678503644e-05, "loss": 1.0517, "step": 26906 }, { "epoch": 0.03, "learning_rate": 4.991137907076042e-05, "loss": 0.8222, "step": 26907 }, { "epoch": 0.03, "learning_rate": 4.991137246277129e-05, "loss": 1.0472, "step": 26908 }, { "epoch": 0.03, "learning_rate": 4.991136585453624e-05, "loss": 1.04, "step": 26909 }, { "epoch": 0.03, "learning_rate": 4.991135924605528e-05, "loss": 1.0255, "step": 26910 }, { "epoch": 0.03, "learning_rate": 4.99113526373284e-05, "loss": 0.9476, "step": 26911 }, { "epoch": 0.03, "learning_rate": 4.991134602835561e-05, "loss": 0.5468, "step": 26912 }, { "epoch": 0.03, "learning_rate": 4.9911339419136904e-05, "loss": 0.8107, "step": 26913 }, { "epoch": 0.03, "learning_rate": 4.991133280967229e-05, "loss": 0.9159, "step": 26914 }, { "epoch": 0.03, "learning_rate": 4.991132619996175e-05, "loss": 1.296, "step": 26915 }, { "epoch": 0.03, "learning_rate": 4.99113195900053e-05, "loss": 0.9748, "step": 26916 }, { "epoch": 0.03, "learning_rate": 4.991131297980294e-05, "loss": 1.0405, "step": 26917 }, { "epoch": 0.03, "learning_rate": 4.991130636935466e-05, "loss": 1.2153, "step": 26918 }, { "epoch": 0.03, "learning_rate": 4.991129975866048e-05, "loss": 1.0682, "step": 26919 }, { "epoch": 0.03, "learning_rate": 4.991129314772037e-05, "loss": 0.6975, "step": 26920 }, { "epoch": 0.03, "learning_rate": 4.991128653653435e-05, "loss": 0.7133, "step": 26921 }, { "epoch": 0.03, "learning_rate": 4.991127992510242e-05, "loss": 0.8922, "step": 26922 }, { "epoch": 0.03, "learning_rate": 4.9911273313424576e-05, "loss": 0.9832, "step": 26923 }, { "epoch": 0.03, "learning_rate": 4.991126670150081e-05, "loss": 0.8909, "step": 26924 }, { "epoch": 0.03, "learning_rate": 4.9911260089331146e-05, "loss": 0.8099, "step": 26925 }, { "epoch": 0.03, "learning_rate": 4.991125347691556e-05, "loss": 0.4222, "step": 26926 }, { "epoch": 0.03, "learning_rate": 4.9911246864254066e-05, "loss": 1.0042, "step": 26927 }, { "epoch": 0.03, "learning_rate": 4.9911240251346646e-05, "loss": 0.7435, "step": 26928 }, { "epoch": 0.03, "learning_rate": 4.991123363819332e-05, "loss": 1.5602, "step": 26929 }, { "epoch": 0.03, "learning_rate": 4.991122702479408e-05, "loss": 0.908, "step": 26930 }, { "epoch": 0.03, "learning_rate": 4.991122041114894e-05, "loss": 1.1267, "step": 26931 }, { "epoch": 0.03, "learning_rate": 4.991121379725787e-05, "loss": 1.353, "step": 26932 }, { "epoch": 0.03, "learning_rate": 4.991120718312089e-05, "loss": 1.1698, "step": 26933 }, { "epoch": 0.03, "learning_rate": 4.9911200568738007e-05, "loss": 1.0144, "step": 26934 }, { "epoch": 0.03, "learning_rate": 4.99111939541092e-05, "loss": 1.0006, "step": 26935 }, { "epoch": 0.03, "learning_rate": 4.9911187339234475e-05, "loss": 1.057, "step": 26936 }, { "epoch": 0.03, "learning_rate": 4.9911180724113846e-05, "loss": 1.0498, "step": 26937 }, { "epoch": 0.03, "learning_rate": 4.991117410874731e-05, "loss": 0.7731, "step": 26938 }, { "epoch": 0.03, "learning_rate": 4.991116749313486e-05, "loss": 1.0751, "step": 26939 }, { "epoch": 0.03, "learning_rate": 4.9911160877276484e-05, "loss": 1.0736, "step": 26940 }, { "epoch": 0.03, "learning_rate": 4.991115426117221e-05, "loss": 0.9844, "step": 26941 }, { "epoch": 0.03, "learning_rate": 4.991114764482202e-05, "loss": 0.8554, "step": 26942 }, { "epoch": 0.03, "learning_rate": 4.991114102822591e-05, "loss": 0.9061, "step": 26943 }, { "epoch": 0.03, "learning_rate": 4.9911134411383896e-05, "loss": 0.4937, "step": 26944 }, { "epoch": 0.03, "learning_rate": 4.991112779429596e-05, "loss": 0.5691, "step": 26945 }, { "epoch": 0.03, "learning_rate": 4.9911121176962125e-05, "loss": 0.1923, "step": 26946 }, { "epoch": 0.03, "learning_rate": 4.991111455938237e-05, "loss": 0.138, "step": 26947 }, { "epoch": 0.03, "learning_rate": 4.9911107941556704e-05, "loss": 0.1559, "step": 26948 }, { "epoch": 0.03, "learning_rate": 4.9911101323485135e-05, "loss": 0.1733, "step": 26949 }, { "epoch": 0.03, "learning_rate": 4.9911094705167635e-05, "loss": 0.4034, "step": 26950 }, { "epoch": 0.03, "learning_rate": 4.991108808660424e-05, "loss": 1.093, "step": 26951 }, { "epoch": 0.03, "learning_rate": 4.991108146779493e-05, "loss": 1.2736, "step": 26952 }, { "epoch": 0.03, "learning_rate": 4.9911074848739705e-05, "loss": 1.1731, "step": 26953 }, { "epoch": 0.03, "learning_rate": 4.991106822943857e-05, "loss": 0.9419, "step": 26954 }, { "epoch": 0.03, "learning_rate": 4.9911061609891515e-05, "loss": 0.4126, "step": 26955 }, { "epoch": 0.03, "learning_rate": 4.991105499009856e-05, "loss": 0.736, "step": 26956 }, { "epoch": 0.03, "learning_rate": 4.991104837005969e-05, "loss": 1.1068, "step": 26957 }, { "epoch": 0.03, "learning_rate": 4.991104174977491e-05, "loss": 0.9218, "step": 26958 }, { "epoch": 0.03, "learning_rate": 4.991103512924422e-05, "loss": 1.0234, "step": 26959 }, { "epoch": 0.03, "learning_rate": 4.991102850846761e-05, "loss": 0.6468, "step": 26960 }, { "epoch": 0.03, "learning_rate": 4.99110218874451e-05, "loss": 0.6965, "step": 26961 }, { "epoch": 0.03, "learning_rate": 4.9911015266176666e-05, "loss": 0.748, "step": 26962 }, { "epoch": 0.03, "learning_rate": 4.991100864466233e-05, "loss": 1.2039, "step": 26963 }, { "epoch": 0.03, "learning_rate": 4.991100202290208e-05, "loss": 1.1384, "step": 26964 }, { "epoch": 0.03, "learning_rate": 4.991099540089592e-05, "loss": 0.9328, "step": 26965 }, { "epoch": 0.03, "learning_rate": 4.991098877864385e-05, "loss": 0.7647, "step": 26966 }, { "epoch": 0.03, "learning_rate": 4.991098215614587e-05, "loss": 0.8638, "step": 26967 }, { "epoch": 0.03, "learning_rate": 4.9910975533401985e-05, "loss": 0.5231, "step": 26968 }, { "epoch": 0.03, "learning_rate": 4.991096891041218e-05, "loss": 0.4703, "step": 26969 }, { "epoch": 0.03, "learning_rate": 4.9910962287176465e-05, "loss": 0.4046, "step": 26970 }, { "epoch": 0.03, "learning_rate": 4.9910955663694835e-05, "loss": 0.482, "step": 26971 }, { "epoch": 0.03, "learning_rate": 4.99109490399673e-05, "loss": 0.6608, "step": 26972 }, { "epoch": 0.03, "learning_rate": 4.9910942415993864e-05, "loss": 0.9169, "step": 26973 }, { "epoch": 0.03, "learning_rate": 4.99109357917745e-05, "loss": 1.1318, "step": 26974 }, { "epoch": 0.03, "learning_rate": 4.991092916730924e-05, "loss": 0.9549, "step": 26975 }, { "epoch": 0.03, "learning_rate": 4.991092254259807e-05, "loss": 0.9387, "step": 26976 }, { "epoch": 0.03, "learning_rate": 4.9910915917640975e-05, "loss": 0.7964, "step": 26977 }, { "epoch": 0.03, "learning_rate": 4.9910909292437984e-05, "loss": 0.8403, "step": 26978 }, { "epoch": 0.03, "learning_rate": 4.9910902666989076e-05, "loss": 1.0187, "step": 26979 }, { "epoch": 0.03, "learning_rate": 4.9910896041294265e-05, "loss": 0.9564, "step": 26980 }, { "epoch": 0.03, "learning_rate": 4.991088941535354e-05, "loss": 1.2211, "step": 26981 }, { "epoch": 0.03, "learning_rate": 4.99108827891669e-05, "loss": 1.2985, "step": 26982 }, { "epoch": 0.03, "learning_rate": 4.991087616273436e-05, "loss": 1.3234, "step": 26983 }, { "epoch": 0.03, "learning_rate": 4.9910869536055906e-05, "loss": 1.011, "step": 26984 }, { "epoch": 0.03, "learning_rate": 4.991086290913154e-05, "loss": 0.9546, "step": 26985 }, { "epoch": 0.03, "learning_rate": 4.9910856281961266e-05, "loss": 1.0895, "step": 26986 }, { "epoch": 0.03, "learning_rate": 4.991084965454508e-05, "loss": 1.0329, "step": 26987 }, { "epoch": 0.03, "learning_rate": 4.991084302688299e-05, "loss": 1.0303, "step": 26988 }, { "epoch": 0.03, "learning_rate": 4.991083639897499e-05, "loss": 1.1332, "step": 26989 }, { "epoch": 0.03, "learning_rate": 4.991082977082108e-05, "loss": 0.9216, "step": 26990 }, { "epoch": 0.03, "learning_rate": 4.991082314242126e-05, "loss": 0.9576, "step": 26991 }, { "epoch": 0.03, "learning_rate": 4.9910816513775526e-05, "loss": 1.1569, "step": 26992 }, { "epoch": 0.03, "learning_rate": 4.991080988488389e-05, "loss": 0.9073, "step": 26993 }, { "epoch": 0.03, "learning_rate": 4.991080325574634e-05, "loss": 1.0934, "step": 26994 }, { "epoch": 0.03, "learning_rate": 4.9910796626362886e-05, "loss": 0.7477, "step": 26995 }, { "epoch": 0.03, "learning_rate": 4.991078999673352e-05, "loss": 0.4916, "step": 26996 }, { "epoch": 0.03, "learning_rate": 4.991078336685825e-05, "loss": 0.504, "step": 26997 }, { "epoch": 0.03, "learning_rate": 4.9910776736737065e-05, "loss": 0.4001, "step": 26998 }, { "epoch": 0.03, "learning_rate": 4.991077010636997e-05, "loss": 0.8864, "step": 26999 }, { "epoch": 0.03, "learning_rate": 4.9910763475756975e-05, "loss": 0.5926, "step": 27000 }, { "epoch": 0.03, "eval_loss": 1.06881582736969, "eval_runtime": 126.3852, "eval_samples_per_second": 10.959, "eval_steps_per_second": 5.483, "step": 27000 }, { "epoch": 0.03, "learning_rate": 4.991075684489806e-05, "loss": 0.6985, "step": 27001 }, { "epoch": 0.03, "learning_rate": 4.9910750213793243e-05, "loss": 0.5751, "step": 27002 }, { "epoch": 0.03, "learning_rate": 4.991074358244252e-05, "loss": 0.652, "step": 27003 }, { "epoch": 0.03, "learning_rate": 4.991073695084588e-05, "loss": 0.6291, "step": 27004 }, { "epoch": 0.03, "learning_rate": 4.991073031900334e-05, "loss": 0.4679, "step": 27005 }, { "epoch": 0.03, "learning_rate": 4.9910723686914887e-05, "loss": 0.6497, "step": 27006 }, { "epoch": 0.03, "learning_rate": 4.991071705458053e-05, "loss": 0.4919, "step": 27007 }, { "epoch": 0.03, "learning_rate": 4.991071042200026e-05, "loss": 0.466, "step": 27008 }, { "epoch": 0.03, "learning_rate": 4.991070378917409e-05, "loss": 0.4591, "step": 27009 }, { "epoch": 0.03, "learning_rate": 4.991069715610201e-05, "loss": 0.4455, "step": 27010 }, { "epoch": 0.03, "learning_rate": 4.9910690522784014e-05, "loss": 0.4484, "step": 27011 }, { "epoch": 0.03, "learning_rate": 4.991068388922012e-05, "loss": 0.4287, "step": 27012 }, { "epoch": 0.03, "learning_rate": 4.991067725541031e-05, "loss": 0.4423, "step": 27013 }, { "epoch": 0.03, "learning_rate": 4.991067062135459e-05, "loss": 0.4063, "step": 27014 }, { "epoch": 0.03, "learning_rate": 4.991066398705298e-05, "loss": 0.4004, "step": 27015 }, { "epoch": 0.03, "learning_rate": 4.991065735250544e-05, "loss": 0.4604, "step": 27016 }, { "epoch": 0.03, "learning_rate": 4.9910650717712e-05, "loss": 0.4911, "step": 27017 }, { "epoch": 0.03, "learning_rate": 4.991064408267266e-05, "loss": 0.5221, "step": 27018 }, { "epoch": 0.03, "learning_rate": 4.991063744738741e-05, "loss": 0.4918, "step": 27019 }, { "epoch": 0.03, "learning_rate": 4.991063081185625e-05, "loss": 0.4898, "step": 27020 }, { "epoch": 0.03, "learning_rate": 4.9910624176079185e-05, "loss": 0.4693, "step": 27021 }, { "epoch": 0.03, "learning_rate": 4.9910617540056216e-05, "loss": 0.48, "step": 27022 }, { "epoch": 0.03, "learning_rate": 4.991061090378733e-05, "loss": 0.5877, "step": 27023 }, { "epoch": 0.03, "learning_rate": 4.991060426727254e-05, "loss": 0.4786, "step": 27024 }, { "epoch": 0.03, "learning_rate": 4.9910597630511845e-05, "loss": 0.4195, "step": 27025 }, { "epoch": 0.03, "learning_rate": 4.991059099350525e-05, "loss": 0.3996, "step": 27026 }, { "epoch": 0.03, "learning_rate": 4.991058435625274e-05, "loss": 0.3994, "step": 27027 }, { "epoch": 0.03, "learning_rate": 4.991057771875433e-05, "loss": 0.4062, "step": 27028 }, { "epoch": 0.03, "learning_rate": 4.9910571081010004e-05, "loss": 0.4113, "step": 27029 }, { "epoch": 0.03, "learning_rate": 4.991056444301977e-05, "loss": 0.4673, "step": 27030 }, { "epoch": 0.03, "learning_rate": 4.9910557804783634e-05, "loss": 0.6835, "step": 27031 }, { "epoch": 0.03, "learning_rate": 4.991055116630159e-05, "loss": 0.5696, "step": 27032 }, { "epoch": 0.03, "learning_rate": 4.991054452757364e-05, "loss": 0.4912, "step": 27033 }, { "epoch": 0.03, "learning_rate": 4.99105378885998e-05, "loss": 0.5117, "step": 27034 }, { "epoch": 0.03, "learning_rate": 4.991053124938003e-05, "loss": 0.4814, "step": 27035 }, { "epoch": 0.03, "learning_rate": 4.991052460991437e-05, "loss": 0.5295, "step": 27036 }, { "epoch": 0.03, "learning_rate": 4.991051797020279e-05, "loss": 0.5073, "step": 27037 }, { "epoch": 0.03, "learning_rate": 4.991051133024531e-05, "loss": 0.5185, "step": 27038 }, { "epoch": 0.03, "learning_rate": 4.991050469004192e-05, "loss": 0.4338, "step": 27039 }, { "epoch": 0.03, "learning_rate": 4.991049804959264e-05, "loss": 0.4382, "step": 27040 }, { "epoch": 0.03, "learning_rate": 4.9910491408897435e-05, "loss": 0.452, "step": 27041 }, { "epoch": 0.03, "learning_rate": 4.9910484767956335e-05, "loss": 0.4877, "step": 27042 }, { "epoch": 0.03, "learning_rate": 4.9910478126769325e-05, "loss": 0.4815, "step": 27043 }, { "epoch": 0.03, "learning_rate": 4.9910471485336404e-05, "loss": 0.4188, "step": 27044 }, { "epoch": 0.03, "learning_rate": 4.9910464843657586e-05, "loss": 0.4334, "step": 27045 }, { "epoch": 0.03, "learning_rate": 4.991045820173286e-05, "loss": 0.4734, "step": 27046 }, { "epoch": 0.03, "learning_rate": 4.991045155956223e-05, "loss": 0.473, "step": 27047 }, { "epoch": 0.03, "learning_rate": 4.9910444917145696e-05, "loss": 0.445, "step": 27048 }, { "epoch": 0.03, "learning_rate": 4.991043827448325e-05, "loss": 0.4857, "step": 27049 }, { "epoch": 0.03, "learning_rate": 4.99104316315749e-05, "loss": 0.4434, "step": 27050 }, { "epoch": 0.03, "learning_rate": 4.9910424988420644e-05, "loss": 0.4397, "step": 27051 }, { "epoch": 0.03, "learning_rate": 4.9910418345020485e-05, "loss": 0.422, "step": 27052 }, { "epoch": 0.03, "learning_rate": 4.991041170137443e-05, "loss": 0.4954, "step": 27053 }, { "epoch": 0.03, "learning_rate": 4.991040505748246e-05, "loss": 0.5059, "step": 27054 }, { "epoch": 0.03, "learning_rate": 4.991039841334458e-05, "loss": 0.5205, "step": 27055 }, { "epoch": 0.03, "learning_rate": 4.991039176896081e-05, "loss": 0.4028, "step": 27056 }, { "epoch": 0.03, "learning_rate": 4.9910385124331116e-05, "loss": 0.4021, "step": 27057 }, { "epoch": 0.03, "learning_rate": 4.9910378479455535e-05, "loss": 0.4271, "step": 27058 }, { "epoch": 0.03, "learning_rate": 4.9910371834334036e-05, "loss": 0.3997, "step": 27059 }, { "epoch": 0.03, "learning_rate": 4.991036518896664e-05, "loss": 0.4413, "step": 27060 }, { "epoch": 0.03, "learning_rate": 4.9910358543353334e-05, "loss": 0.4646, "step": 27061 }, { "epoch": 0.03, "learning_rate": 4.991035189749413e-05, "loss": 0.4452, "step": 27062 }, { "epoch": 0.03, "learning_rate": 4.9910345251389016e-05, "loss": 0.4356, "step": 27063 }, { "epoch": 0.03, "learning_rate": 4.9910338605038e-05, "loss": 0.441, "step": 27064 }, { "epoch": 0.03, "learning_rate": 4.9910331958441084e-05, "loss": 0.4046, "step": 27065 }, { "epoch": 0.03, "learning_rate": 4.991032531159826e-05, "loss": 0.4549, "step": 27066 }, { "epoch": 0.03, "learning_rate": 4.991031866450953e-05, "loss": 0.4641, "step": 27067 }, { "epoch": 0.03, "learning_rate": 4.9910312017174896e-05, "loss": 0.5026, "step": 27068 }, { "epoch": 0.03, "learning_rate": 4.991030536959436e-05, "loss": 0.4674, "step": 27069 }, { "epoch": 0.03, "learning_rate": 4.991029872176791e-05, "loss": 0.5061, "step": 27070 }, { "epoch": 0.03, "learning_rate": 4.9910292073695567e-05, "loss": 0.474, "step": 27071 }, { "epoch": 0.03, "learning_rate": 4.9910285425377325e-05, "loss": 0.4524, "step": 27072 }, { "epoch": 0.03, "learning_rate": 4.9910278776813166e-05, "loss": 0.4787, "step": 27073 }, { "epoch": 0.03, "learning_rate": 4.991027212800311e-05, "loss": 0.4907, "step": 27074 }, { "epoch": 0.03, "learning_rate": 4.991026547894715e-05, "loss": 0.4861, "step": 27075 }, { "epoch": 0.03, "learning_rate": 4.9910258829645286e-05, "loss": 0.5151, "step": 27076 }, { "epoch": 0.03, "learning_rate": 4.991025218009752e-05, "loss": 0.4591, "step": 27077 }, { "epoch": 0.03, "learning_rate": 4.991024553030385e-05, "loss": 0.5556, "step": 27078 }, { "epoch": 0.03, "learning_rate": 4.991023888026427e-05, "loss": 0.5146, "step": 27079 }, { "epoch": 0.03, "learning_rate": 4.991023222997879e-05, "loss": 0.4401, "step": 27080 }, { "epoch": 0.03, "learning_rate": 4.991022557944741e-05, "loss": 0.4655, "step": 27081 }, { "epoch": 0.03, "learning_rate": 4.991021892867013e-05, "loss": 0.4523, "step": 27082 }, { "epoch": 0.03, "learning_rate": 4.991021227764694e-05, "loss": 0.4146, "step": 27083 }, { "epoch": 0.03, "learning_rate": 4.991020562637785e-05, "loss": 0.3843, "step": 27084 }, { "epoch": 0.03, "learning_rate": 4.991019897486285e-05, "loss": 0.4036, "step": 27085 }, { "epoch": 0.03, "learning_rate": 4.991019232310196e-05, "loss": 0.427, "step": 27086 }, { "epoch": 0.03, "learning_rate": 4.991018567109516e-05, "loss": 0.4155, "step": 27087 }, { "epoch": 0.03, "learning_rate": 4.991017901884245e-05, "loss": 0.4289, "step": 27088 }, { "epoch": 0.03, "learning_rate": 4.991017236634385e-05, "loss": 0.4104, "step": 27089 }, { "epoch": 0.03, "learning_rate": 4.9910165713599345e-05, "loss": 0.46, "step": 27090 }, { "epoch": 0.03, "learning_rate": 4.991015906060893e-05, "loss": 0.4813, "step": 27091 }, { "epoch": 0.03, "learning_rate": 4.991015240737261e-05, "loss": 0.4047, "step": 27092 }, { "epoch": 0.03, "learning_rate": 4.9910145753890404e-05, "loss": 0.4293, "step": 27093 }, { "epoch": 0.03, "learning_rate": 4.991013910016228e-05, "loss": 0.4493, "step": 27094 }, { "epoch": 0.03, "learning_rate": 4.991013244618826e-05, "loss": 0.4408, "step": 27095 }, { "epoch": 0.03, "learning_rate": 4.9910125791968335e-05, "loss": 0.4357, "step": 27096 }, { "epoch": 0.03, "learning_rate": 4.991011913750251e-05, "loss": 0.464, "step": 27097 }, { "epoch": 0.03, "learning_rate": 4.9910112482790783e-05, "loss": 0.3781, "step": 27098 }, { "epoch": 0.03, "learning_rate": 4.991010582783315e-05, "loss": 0.414, "step": 27099 }, { "epoch": 0.03, "learning_rate": 4.991009917262962e-05, "loss": 0.3884, "step": 27100 }, { "epoch": 0.03, "learning_rate": 4.9910092517180184e-05, "loss": 0.3994, "step": 27101 }, { "epoch": 0.03, "learning_rate": 4.991008586148485e-05, "loss": 0.4007, "step": 27102 }, { "epoch": 0.03, "learning_rate": 4.9910079205543615e-05, "loss": 0.4584, "step": 27103 }, { "epoch": 0.03, "learning_rate": 4.991007254935647e-05, "loss": 0.5652, "step": 27104 }, { "epoch": 0.03, "learning_rate": 4.991006589292343e-05, "loss": 0.7679, "step": 27105 }, { "epoch": 0.03, "learning_rate": 4.991005923624449e-05, "loss": 0.7162, "step": 27106 }, { "epoch": 0.03, "learning_rate": 4.991005257931964e-05, "loss": 0.5744, "step": 27107 }, { "epoch": 0.03, "learning_rate": 4.99100459221489e-05, "loss": 0.4677, "step": 27108 }, { "epoch": 0.03, "learning_rate": 4.991003926473225e-05, "loss": 0.4657, "step": 27109 }, { "epoch": 0.03, "learning_rate": 4.99100326070697e-05, "loss": 0.47, "step": 27110 }, { "epoch": 0.03, "learning_rate": 4.991002594916125e-05, "loss": 0.5322, "step": 27111 }, { "epoch": 0.03, "learning_rate": 4.9910019291006904e-05, "loss": 0.8757, "step": 27112 }, { "epoch": 0.03, "learning_rate": 4.991001263260665e-05, "loss": 0.8789, "step": 27113 }, { "epoch": 0.03, "learning_rate": 4.99100059739605e-05, "loss": 1.5043, "step": 27114 }, { "epoch": 0.03, "learning_rate": 4.990999931506843e-05, "loss": 0.7531, "step": 27115 }, { "epoch": 0.03, "learning_rate": 4.990999265593048e-05, "loss": 0.2777, "step": 27116 }, { "epoch": 0.03, "learning_rate": 4.990998599654663e-05, "loss": 0.2668, "step": 27117 }, { "epoch": 0.03, "learning_rate": 4.9909979336916866e-05, "loss": 0.1835, "step": 27118 }, { "epoch": 0.03, "learning_rate": 4.9909972677041205e-05, "loss": 0.2602, "step": 27119 }, { "epoch": 0.03, "learning_rate": 4.9909966016919654e-05, "loss": 1.0096, "step": 27120 }, { "epoch": 0.03, "learning_rate": 4.9909959356552186e-05, "loss": 1.2605, "step": 27121 }, { "epoch": 0.03, "learning_rate": 4.990995269593883e-05, "loss": 1.3213, "step": 27122 }, { "epoch": 0.03, "learning_rate": 4.990994603507957e-05, "loss": 1.0937, "step": 27123 }, { "epoch": 0.03, "learning_rate": 4.99099393739744e-05, "loss": 1.0715, "step": 27124 }, { "epoch": 0.03, "learning_rate": 4.990993271262334e-05, "loss": 1.3422, "step": 27125 }, { "epoch": 0.03, "learning_rate": 4.990992605102638e-05, "loss": 1.1623, "step": 27126 }, { "epoch": 0.03, "learning_rate": 4.990991938918351e-05, "loss": 1.7131, "step": 27127 }, { "epoch": 0.03, "learning_rate": 4.9909912727094756e-05, "loss": 1.3739, "step": 27128 }, { "epoch": 0.03, "learning_rate": 4.990990606476009e-05, "loss": 1.4769, "step": 27129 }, { "epoch": 0.03, "learning_rate": 4.990989940217953e-05, "loss": 1.6199, "step": 27130 }, { "epoch": 0.03, "learning_rate": 4.9909892739353057e-05, "loss": 1.4038, "step": 27131 }, { "epoch": 0.03, "learning_rate": 4.99098860762807e-05, "loss": 0.9505, "step": 27132 }, { "epoch": 0.03, "learning_rate": 4.9909879412962433e-05, "loss": 1.0532, "step": 27133 }, { "epoch": 0.03, "learning_rate": 4.9909872749398276e-05, "loss": 1.1387, "step": 27134 }, { "epoch": 0.03, "learning_rate": 4.990986608558821e-05, "loss": 0.9297, "step": 27135 }, { "epoch": 0.03, "learning_rate": 4.9909859421532244e-05, "loss": 0.7847, "step": 27136 }, { "epoch": 0.03, "learning_rate": 4.990985275723038e-05, "loss": 0.8098, "step": 27137 }, { "epoch": 0.03, "learning_rate": 4.990984609268262e-05, "loss": 1.1527, "step": 27138 }, { "epoch": 0.03, "learning_rate": 4.990983942788896e-05, "loss": 1.0261, "step": 27139 }, { "epoch": 0.03, "learning_rate": 4.99098327628494e-05, "loss": 0.9621, "step": 27140 }, { "epoch": 0.03, "learning_rate": 4.990982609756394e-05, "loss": 0.8721, "step": 27141 }, { "epoch": 0.03, "learning_rate": 4.9909819432032577e-05, "loss": 0.994, "step": 27142 }, { "epoch": 0.03, "learning_rate": 4.9909812766255326e-05, "loss": 0.9971, "step": 27143 }, { "epoch": 0.03, "learning_rate": 4.9909806100232164e-05, "loss": 1.0924, "step": 27144 }, { "epoch": 0.03, "learning_rate": 4.990979943396311e-05, "loss": 1.0448, "step": 27145 }, { "epoch": 0.03, "learning_rate": 4.990979276744815e-05, "loss": 1.0751, "step": 27146 }, { "epoch": 0.03, "learning_rate": 4.99097861006873e-05, "loss": 1.08, "step": 27147 }, { "epoch": 0.03, "learning_rate": 4.990977943368054e-05, "loss": 1.0588, "step": 27148 }, { "epoch": 0.03, "learning_rate": 4.990977276642789e-05, "loss": 0.6965, "step": 27149 }, { "epoch": 0.03, "learning_rate": 4.9909766098929337e-05, "loss": 1.1504, "step": 27150 }, { "epoch": 0.03, "learning_rate": 4.990975943118489e-05, "loss": 1.1032, "step": 27151 }, { "epoch": 0.03, "learning_rate": 4.990975276319454e-05, "loss": 1.0929, "step": 27152 }, { "epoch": 0.03, "learning_rate": 4.990974609495829e-05, "loss": 0.518, "step": 27153 }, { "epoch": 0.03, "learning_rate": 4.990973942647614e-05, "loss": 0.7915, "step": 27154 }, { "epoch": 0.03, "learning_rate": 4.99097327577481e-05, "loss": 0.7656, "step": 27155 }, { "epoch": 0.03, "learning_rate": 4.9909726088774154e-05, "loss": 1.322, "step": 27156 }, { "epoch": 0.03, "learning_rate": 4.990971941955431e-05, "loss": 0.9368, "step": 27157 }, { "epoch": 0.03, "learning_rate": 4.990971275008858e-05, "loss": 0.8154, "step": 27158 }, { "epoch": 0.03, "learning_rate": 4.990970608037694e-05, "loss": 1.1022, "step": 27159 }, { "epoch": 0.03, "learning_rate": 4.9909699410419404e-05, "loss": 0.8749, "step": 27160 }, { "epoch": 0.03, "learning_rate": 4.9909692740215966e-05, "loss": 0.7688, "step": 27161 }, { "epoch": 0.03, "learning_rate": 4.990968606976664e-05, "loss": 1.1618, "step": 27162 }, { "epoch": 0.03, "learning_rate": 4.9909679399071404e-05, "loss": 0.9059, "step": 27163 }, { "epoch": 0.03, "learning_rate": 4.990967272813028e-05, "loss": 0.9161, "step": 27164 }, { "epoch": 0.03, "learning_rate": 4.9909666056943255e-05, "loss": 1.1195, "step": 27165 }, { "epoch": 0.03, "learning_rate": 4.990965938551033e-05, "loss": 0.8816, "step": 27166 }, { "epoch": 0.03, "learning_rate": 4.990965271383151e-05, "loss": 0.7799, "step": 27167 }, { "epoch": 0.03, "learning_rate": 4.990964604190679e-05, "loss": 0.9019, "step": 27168 }, { "epoch": 0.03, "learning_rate": 4.990963936973618e-05, "loss": 1.2261, "step": 27169 }, { "epoch": 0.03, "learning_rate": 4.9909632697319665e-05, "loss": 1.3599, "step": 27170 }, { "epoch": 0.03, "learning_rate": 4.990962602465725e-05, "loss": 0.9132, "step": 27171 }, { "epoch": 0.03, "learning_rate": 4.990961935174895e-05, "loss": 0.9332, "step": 27172 }, { "epoch": 0.03, "learning_rate": 4.990961267859474e-05, "loss": 1.069, "step": 27173 }, { "epoch": 0.03, "learning_rate": 4.990960600519464e-05, "loss": 0.4079, "step": 27174 }, { "epoch": 0.03, "learning_rate": 4.990959933154864e-05, "loss": 0.2566, "step": 27175 }, { "epoch": 0.03, "learning_rate": 4.990959265765674e-05, "loss": 0.3453, "step": 27176 }, { "epoch": 0.03, "learning_rate": 4.990958598351895e-05, "loss": 0.1878, "step": 27177 }, { "epoch": 0.03, "learning_rate": 4.9909579309135255e-05, "loss": 0.2198, "step": 27178 }, { "epoch": 0.03, "learning_rate": 4.990957263450567e-05, "loss": 0.5531, "step": 27179 }, { "epoch": 0.03, "learning_rate": 4.990956595963018e-05, "loss": 0.9182, "step": 27180 }, { "epoch": 0.03, "learning_rate": 4.99095592845088e-05, "loss": 0.7606, "step": 27181 }, { "epoch": 0.03, "learning_rate": 4.9909552609141524e-05, "loss": 0.9266, "step": 27182 }, { "epoch": 0.03, "learning_rate": 4.990954593352835e-05, "loss": 0.9673, "step": 27183 }, { "epoch": 0.03, "learning_rate": 4.990953925766928e-05, "loss": 1.0877, "step": 27184 }, { "epoch": 0.03, "learning_rate": 4.990953258156431e-05, "loss": 1.4811, "step": 27185 }, { "epoch": 0.03, "learning_rate": 4.990952590521345e-05, "loss": 0.919, "step": 27186 }, { "epoch": 0.03, "learning_rate": 4.9909519228616685e-05, "loss": 0.9347, "step": 27187 }, { "epoch": 0.03, "learning_rate": 4.9909512551774026e-05, "loss": 0.6831, "step": 27188 }, { "epoch": 0.03, "learning_rate": 4.9909505874685484e-05, "loss": 0.609, "step": 27189 }, { "epoch": 0.03, "learning_rate": 4.990949919735103e-05, "loss": 0.7479, "step": 27190 }, { "epoch": 0.03, "learning_rate": 4.990949251977069e-05, "loss": 1.0012, "step": 27191 }, { "epoch": 0.03, "learning_rate": 4.990948584194445e-05, "loss": 0.8955, "step": 27192 }, { "epoch": 0.03, "learning_rate": 4.990947916387231e-05, "loss": 0.9165, "step": 27193 }, { "epoch": 0.03, "learning_rate": 4.9909472485554275e-05, "loss": 0.8733, "step": 27194 }, { "epoch": 0.03, "learning_rate": 4.9909465806990344e-05, "loss": 1.227, "step": 27195 }, { "epoch": 0.03, "learning_rate": 4.990945912818052e-05, "loss": 1.3044, "step": 27196 }, { "epoch": 0.03, "learning_rate": 4.99094524491248e-05, "loss": 1.2288, "step": 27197 }, { "epoch": 0.03, "learning_rate": 4.990944576982319e-05, "loss": 0.9239, "step": 27198 }, { "epoch": 0.03, "learning_rate": 4.990943909027568e-05, "loss": 1.2269, "step": 27199 }, { "epoch": 0.03, "learning_rate": 4.990943241048227e-05, "loss": 0.8428, "step": 27200 }, { "epoch": 0.03, "learning_rate": 4.990942573044297e-05, "loss": 1.1116, "step": 27201 }, { "epoch": 0.03, "learning_rate": 4.9909419050157766e-05, "loss": 0.7963, "step": 27202 }, { "epoch": 0.03, "learning_rate": 4.990941236962667e-05, "loss": 1.1804, "step": 27203 }, { "epoch": 0.03, "learning_rate": 4.990940568884969e-05, "loss": 0.9363, "step": 27204 }, { "epoch": 0.03, "learning_rate": 4.9909399007826806e-05, "loss": 1.0536, "step": 27205 }, { "epoch": 0.03, "learning_rate": 4.990939232655803e-05, "loss": 0.8306, "step": 27206 }, { "epoch": 0.03, "learning_rate": 4.990938564504336e-05, "loss": 1.0922, "step": 27207 }, { "epoch": 0.03, "learning_rate": 4.9909378963282785e-05, "loss": 1.0285, "step": 27208 }, { "epoch": 0.03, "learning_rate": 4.990937228127632e-05, "loss": 1.0117, "step": 27209 }, { "epoch": 0.03, "learning_rate": 4.990936559902397e-05, "loss": 0.9404, "step": 27210 }, { "epoch": 0.03, "learning_rate": 4.990935891652572e-05, "loss": 0.9749, "step": 27211 }, { "epoch": 0.03, "learning_rate": 4.990935223378157e-05, "loss": 0.7205, "step": 27212 }, { "epoch": 0.03, "learning_rate": 4.9909345550791525e-05, "loss": 1.0468, "step": 27213 }, { "epoch": 0.03, "learning_rate": 4.990933886755559e-05, "loss": 1.0374, "step": 27214 }, { "epoch": 0.03, "learning_rate": 4.990933218407375e-05, "loss": 1.1134, "step": 27215 }, { "epoch": 0.03, "learning_rate": 4.990932550034603e-05, "loss": 1.225, "step": 27216 }, { "epoch": 0.03, "learning_rate": 4.990931881637241e-05, "loss": 1.0937, "step": 27217 }, { "epoch": 0.03, "learning_rate": 4.9909312132152894e-05, "loss": 1.0351, "step": 27218 }, { "epoch": 0.03, "learning_rate": 4.9909305447687494e-05, "loss": 0.9798, "step": 27219 }, { "epoch": 0.03, "learning_rate": 4.9909298762976184e-05, "loss": 1.1342, "step": 27220 }, { "epoch": 0.03, "learning_rate": 4.990929207801899e-05, "loss": 1.1391, "step": 27221 }, { "epoch": 0.03, "learning_rate": 4.99092853928159e-05, "loss": 1.2792, "step": 27222 }, { "epoch": 0.03, "learning_rate": 4.990927870736691e-05, "loss": 0.8999, "step": 27223 }, { "epoch": 0.03, "learning_rate": 4.990927202167204e-05, "loss": 0.7395, "step": 27224 }, { "epoch": 0.03, "learning_rate": 4.9909265335731264e-05, "loss": 0.9834, "step": 27225 }, { "epoch": 0.03, "learning_rate": 4.99092586495446e-05, "loss": 0.7025, "step": 27226 }, { "epoch": 0.03, "learning_rate": 4.9909251963112036e-05, "loss": 1.3661, "step": 27227 }, { "epoch": 0.03, "learning_rate": 4.990924527643358e-05, "loss": 1.1217, "step": 27228 }, { "epoch": 0.03, "learning_rate": 4.990923858950924e-05, "loss": 0.9009, "step": 27229 }, { "epoch": 0.03, "learning_rate": 4.990923190233899e-05, "loss": 0.8427, "step": 27230 }, { "epoch": 0.03, "learning_rate": 4.990922521492286e-05, "loss": 0.9321, "step": 27231 }, { "epoch": 0.03, "learning_rate": 4.9909218527260834e-05, "loss": 1.1869, "step": 27232 }, { "epoch": 0.03, "learning_rate": 4.9909211839352916e-05, "loss": 1.1952, "step": 27233 }, { "epoch": 0.03, "learning_rate": 4.99092051511991e-05, "loss": 0.8818, "step": 27234 }, { "epoch": 0.03, "learning_rate": 4.9909198462799396e-05, "loss": 0.9442, "step": 27235 }, { "epoch": 0.03, "learning_rate": 4.9909191774153794e-05, "loss": 0.6818, "step": 27236 }, { "epoch": 0.03, "learning_rate": 4.99091850852623e-05, "loss": 1.3287, "step": 27237 }, { "epoch": 0.03, "learning_rate": 4.990917839612491e-05, "loss": 0.8901, "step": 27238 }, { "epoch": 0.03, "learning_rate": 4.990917170674164e-05, "loss": 1.1095, "step": 27239 }, { "epoch": 0.03, "learning_rate": 4.990916501711246e-05, "loss": 1.2713, "step": 27240 }, { "epoch": 0.03, "learning_rate": 4.99091583272374e-05, "loss": 1.3348, "step": 27241 }, { "epoch": 0.03, "learning_rate": 4.9909151637116444e-05, "loss": 1.0375, "step": 27242 }, { "epoch": 0.03, "learning_rate": 4.990914494674959e-05, "loss": 0.9162, "step": 27243 }, { "epoch": 0.03, "learning_rate": 4.9909138256136844e-05, "loss": 1.0462, "step": 27244 }, { "epoch": 0.03, "learning_rate": 4.990913156527821e-05, "loss": 0.859, "step": 27245 }, { "epoch": 0.03, "learning_rate": 4.9909124874173684e-05, "loss": 0.9271, "step": 27246 }, { "epoch": 0.03, "learning_rate": 4.990911818282327e-05, "loss": 1.2253, "step": 27247 }, { "epoch": 0.03, "learning_rate": 4.9909111491226956e-05, "loss": 1.1501, "step": 27248 }, { "epoch": 0.03, "learning_rate": 4.9909104799384746e-05, "loss": 0.989, "step": 27249 }, { "epoch": 0.03, "learning_rate": 4.9909098107296646e-05, "loss": 1.0255, "step": 27250 }, { "epoch": 0.03, "learning_rate": 4.990909141496266e-05, "loss": 1.0777, "step": 27251 }, { "epoch": 0.03, "learning_rate": 4.990908472238278e-05, "loss": 0.9465, "step": 27252 }, { "epoch": 0.03, "learning_rate": 4.990907802955701e-05, "loss": 1.8296, "step": 27253 }, { "epoch": 0.03, "learning_rate": 4.990907133648535e-05, "loss": 0.7213, "step": 27254 }, { "epoch": 0.03, "learning_rate": 4.990906464316779e-05, "loss": 0.6841, "step": 27255 }, { "epoch": 0.03, "learning_rate": 4.9909057949604336e-05, "loss": 0.6811, "step": 27256 }, { "epoch": 0.03, "learning_rate": 4.9909051255795e-05, "loss": 0.638, "step": 27257 }, { "epoch": 0.03, "learning_rate": 4.990904456173976e-05, "loss": 0.6217, "step": 27258 }, { "epoch": 0.03, "learning_rate": 4.990903786743865e-05, "loss": 0.5802, "step": 27259 }, { "epoch": 0.03, "learning_rate": 4.9909031172891626e-05, "loss": 0.5987, "step": 27260 }, { "epoch": 0.03, "learning_rate": 4.990902447809872e-05, "loss": 0.6173, "step": 27261 }, { "epoch": 0.03, "learning_rate": 4.990901778305992e-05, "loss": 0.6069, "step": 27262 }, { "epoch": 0.03, "learning_rate": 4.9909011087775235e-05, "loss": 0.606, "step": 27263 }, { "epoch": 0.03, "learning_rate": 4.990900439224465e-05, "loss": 0.6202, "step": 27264 }, { "epoch": 0.03, "learning_rate": 4.9908997696468183e-05, "loss": 0.5989, "step": 27265 }, { "epoch": 0.03, "learning_rate": 4.990899100044582e-05, "loss": 0.6124, "step": 27266 }, { "epoch": 0.03, "learning_rate": 4.9908984304177564e-05, "loss": 0.5295, "step": 27267 }, { "epoch": 0.03, "learning_rate": 4.990897760766342e-05, "loss": 0.6406, "step": 27268 }, { "epoch": 0.03, "learning_rate": 4.990897091090338e-05, "loss": 0.6193, "step": 27269 }, { "epoch": 0.03, "learning_rate": 4.990896421389746e-05, "loss": 0.6189, "step": 27270 }, { "epoch": 0.03, "learning_rate": 4.990895751664564e-05, "loss": 0.5185, "step": 27271 }, { "epoch": 0.03, "learning_rate": 4.9908950819147934e-05, "loss": 0.5183, "step": 27272 }, { "epoch": 0.03, "learning_rate": 4.990894412140433e-05, "loss": 0.5313, "step": 27273 }, { "epoch": 0.03, "learning_rate": 4.990893742341485e-05, "loss": 0.701, "step": 27274 }, { "epoch": 0.03, "learning_rate": 4.9908930725179466e-05, "loss": 0.9766, "step": 27275 }, { "epoch": 0.03, "learning_rate": 4.990892402669819e-05, "loss": 0.7653, "step": 27276 }, { "epoch": 0.03, "learning_rate": 4.9908917327971034e-05, "loss": 1.2227, "step": 27277 }, { "epoch": 0.03, "learning_rate": 4.9908910628997986e-05, "loss": 1.0102, "step": 27278 }, { "epoch": 0.03, "learning_rate": 4.990890392977904e-05, "loss": 1.1506, "step": 27279 }, { "epoch": 0.03, "learning_rate": 4.990889723031421e-05, "loss": 0.936, "step": 27280 }, { "epoch": 0.03, "learning_rate": 4.9908890530603487e-05, "loss": 1.2616, "step": 27281 }, { "epoch": 0.03, "learning_rate": 4.990888383064688e-05, "loss": 0.9683, "step": 27282 }, { "epoch": 0.03, "learning_rate": 4.990887713044438e-05, "loss": 0.7111, "step": 27283 }, { "epoch": 0.03, "learning_rate": 4.990887042999598e-05, "loss": 0.7616, "step": 27284 }, { "epoch": 0.03, "learning_rate": 4.99088637293017e-05, "loss": 1.1338, "step": 27285 }, { "epoch": 0.03, "learning_rate": 4.9908857028361524e-05, "loss": 0.9861, "step": 27286 }, { "epoch": 0.03, "learning_rate": 4.9908850327175464e-05, "loss": 1.1483, "step": 27287 }, { "epoch": 0.03, "learning_rate": 4.990884362574352e-05, "loss": 0.8348, "step": 27288 }, { "epoch": 0.03, "learning_rate": 4.990883692406567e-05, "loss": 1.123, "step": 27289 }, { "epoch": 0.03, "learning_rate": 4.990883022214195e-05, "loss": 0.9673, "step": 27290 }, { "epoch": 0.03, "learning_rate": 4.9908823519972325e-05, "loss": 1.0802, "step": 27291 }, { "epoch": 0.03, "learning_rate": 4.990881681755681e-05, "loss": 0.6471, "step": 27292 }, { "epoch": 0.03, "learning_rate": 4.9908810114895425e-05, "loss": 0.4516, "step": 27293 }, { "epoch": 0.03, "learning_rate": 4.990880341198813e-05, "loss": 1.072, "step": 27294 }, { "epoch": 0.03, "learning_rate": 4.9908796708834956e-05, "loss": 0.8767, "step": 27295 }, { "epoch": 0.03, "learning_rate": 4.990879000543589e-05, "loss": 0.9782, "step": 27296 }, { "epoch": 0.03, "learning_rate": 4.9908783301790926e-05, "loss": 1.2875, "step": 27297 }, { "epoch": 0.03, "learning_rate": 4.9908776597900086e-05, "loss": 1.2192, "step": 27298 }, { "epoch": 0.03, "learning_rate": 4.9908769893763356e-05, "loss": 0.7961, "step": 27299 }, { "epoch": 0.03, "learning_rate": 4.9908763189380735e-05, "loss": 1.0628, "step": 27300 }, { "epoch": 0.03, "learning_rate": 4.9908756484752224e-05, "loss": 1.1809, "step": 27301 }, { "epoch": 0.03, "learning_rate": 4.990874977987783e-05, "loss": 1.0669, "step": 27302 }, { "epoch": 0.03, "learning_rate": 4.990874307475754e-05, "loss": 1.0891, "step": 27303 }, { "epoch": 0.03, "learning_rate": 4.9908736369391364e-05, "loss": 0.9147, "step": 27304 }, { "epoch": 0.03, "learning_rate": 4.99087296637793e-05, "loss": 0.8192, "step": 27305 }, { "epoch": 0.03, "learning_rate": 4.9908722957921336e-05, "loss": 0.2421, "step": 27306 }, { "epoch": 0.03, "learning_rate": 4.99087162518175e-05, "loss": 1.3424, "step": 27307 }, { "epoch": 0.03, "learning_rate": 4.990870954546777e-05, "loss": 1.4184, "step": 27308 }, { "epoch": 0.03, "learning_rate": 4.990870283887215e-05, "loss": 0.6105, "step": 27309 }, { "epoch": 0.03, "learning_rate": 4.990869613203064e-05, "loss": 0.1075, "step": 27310 }, { "epoch": 0.03, "learning_rate": 4.9908689424943245e-05, "loss": 0.173, "step": 27311 }, { "epoch": 0.03, "learning_rate": 4.990868271760997e-05, "loss": 0.2078, "step": 27312 }, { "epoch": 0.03, "learning_rate": 4.9908676010030794e-05, "loss": 0.0899, "step": 27313 }, { "epoch": 0.03, "learning_rate": 4.990866930220574e-05, "loss": 0.169, "step": 27314 }, { "epoch": 0.03, "learning_rate": 4.990866259413479e-05, "loss": 0.4511, "step": 27315 }, { "epoch": 0.03, "learning_rate": 4.990865588581795e-05, "loss": 0.9065, "step": 27316 }, { "epoch": 0.03, "learning_rate": 4.990864917725523e-05, "loss": 1.0429, "step": 27317 }, { "epoch": 0.03, "learning_rate": 4.990864246844662e-05, "loss": 1.1346, "step": 27318 }, { "epoch": 0.03, "learning_rate": 4.9908635759392116e-05, "loss": 1.0786, "step": 27319 }, { "epoch": 0.03, "learning_rate": 4.990862905009174e-05, "loss": 1.1172, "step": 27320 }, { "epoch": 0.03, "learning_rate": 4.990862234054546e-05, "loss": 1.0875, "step": 27321 }, { "epoch": 0.03, "learning_rate": 4.9908615630753296e-05, "loss": 1.1038, "step": 27322 }, { "epoch": 0.03, "learning_rate": 4.990860892071525e-05, "loss": 0.9505, "step": 27323 }, { "epoch": 0.03, "learning_rate": 4.990860221043132e-05, "loss": 1.1706, "step": 27324 }, { "epoch": 0.03, "learning_rate": 4.9908595499901496e-05, "loss": 0.8188, "step": 27325 }, { "epoch": 0.03, "learning_rate": 4.990858878912579e-05, "loss": 1.129, "step": 27326 }, { "epoch": 0.03, "learning_rate": 4.990858207810419e-05, "loss": 0.7701, "step": 27327 }, { "epoch": 0.03, "learning_rate": 4.9908575366836705e-05, "loss": 0.8286, "step": 27328 }, { "epoch": 0.03, "learning_rate": 4.9908568655323334e-05, "loss": 0.8355, "step": 27329 }, { "epoch": 0.03, "learning_rate": 4.990856194356408e-05, "loss": 0.8331, "step": 27330 }, { "epoch": 0.03, "learning_rate": 4.9908555231558935e-05, "loss": 0.8105, "step": 27331 }, { "epoch": 0.03, "learning_rate": 4.99085485193079e-05, "loss": 0.9011, "step": 27332 }, { "epoch": 0.03, "learning_rate": 4.990854180681098e-05, "loss": 0.8298, "step": 27333 }, { "epoch": 0.03, "learning_rate": 4.990853509406818e-05, "loss": 1.1051, "step": 27334 }, { "epoch": 0.03, "learning_rate": 4.990852838107949e-05, "loss": 1.0444, "step": 27335 }, { "epoch": 0.03, "learning_rate": 4.990852166784491e-05, "loss": 1.1214, "step": 27336 }, { "epoch": 0.03, "learning_rate": 4.9908514954364446e-05, "loss": 0.7787, "step": 27337 }, { "epoch": 0.03, "learning_rate": 4.9908508240638096e-05, "loss": 0.9952, "step": 27338 }, { "epoch": 0.03, "learning_rate": 4.9908501526665863e-05, "loss": 0.6619, "step": 27339 }, { "epoch": 0.03, "learning_rate": 4.990849481244774e-05, "loss": 0.8137, "step": 27340 }, { "epoch": 0.03, "learning_rate": 4.9908488097983727e-05, "loss": 1.0868, "step": 27341 }, { "epoch": 0.03, "learning_rate": 4.990848138327383e-05, "loss": 0.8065, "step": 27342 }, { "epoch": 0.03, "learning_rate": 4.990847466831805e-05, "loss": 0.6516, "step": 27343 }, { "epoch": 0.03, "learning_rate": 4.990846795311639e-05, "loss": 1.1638, "step": 27344 }, { "epoch": 0.03, "learning_rate": 4.990846123766884e-05, "loss": 1.1088, "step": 27345 }, { "epoch": 0.03, "learning_rate": 4.990845452197539e-05, "loss": 0.8209, "step": 27346 }, { "epoch": 0.03, "learning_rate": 4.990844780603607e-05, "loss": 1.3272, "step": 27347 }, { "epoch": 0.03, "learning_rate": 4.990844108985086e-05, "loss": 1.08, "step": 27348 }, { "epoch": 0.03, "learning_rate": 4.9908434373419765e-05, "loss": 1.075, "step": 27349 }, { "epoch": 0.03, "learning_rate": 4.9908427656742786e-05, "loss": 1.1984, "step": 27350 }, { "epoch": 0.03, "learning_rate": 4.990842093981992e-05, "loss": 0.7415, "step": 27351 }, { "epoch": 0.03, "learning_rate": 4.9908414222651165e-05, "loss": 0.6888, "step": 27352 }, { "epoch": 0.03, "learning_rate": 4.990840750523653e-05, "loss": 0.9654, "step": 27353 }, { "epoch": 0.03, "learning_rate": 4.9908400787576e-05, "loss": 0.9118, "step": 27354 }, { "epoch": 0.03, "learning_rate": 4.990839406966959e-05, "loss": 0.8828, "step": 27355 }, { "epoch": 0.03, "learning_rate": 4.990838735151731e-05, "loss": 0.9993, "step": 27356 }, { "epoch": 0.03, "learning_rate": 4.9908380633119123e-05, "loss": 1.2514, "step": 27357 }, { "epoch": 0.03, "learning_rate": 4.9908373914475057e-05, "loss": 1.1863, "step": 27358 }, { "epoch": 0.03, "learning_rate": 4.990836719558511e-05, "loss": 1.7146, "step": 27359 }, { "epoch": 0.03, "learning_rate": 4.990836047644928e-05, "loss": 0.8938, "step": 27360 }, { "epoch": 0.03, "learning_rate": 4.990835375706756e-05, "loss": 0.834, "step": 27361 }, { "epoch": 0.03, "learning_rate": 4.9908347037439954e-05, "loss": 0.7464, "step": 27362 }, { "epoch": 0.03, "learning_rate": 4.990834031756647e-05, "loss": 0.9525, "step": 27363 }, { "epoch": 0.03, "learning_rate": 4.99083335974471e-05, "loss": 1.116, "step": 27364 }, { "epoch": 0.03, "learning_rate": 4.9908326877081836e-05, "loss": 0.9974, "step": 27365 }, { "epoch": 0.03, "learning_rate": 4.99083201564707e-05, "loss": 1.2265, "step": 27366 }, { "epoch": 0.03, "learning_rate": 4.9908313435613676e-05, "loss": 1.2869, "step": 27367 }, { "epoch": 0.03, "learning_rate": 4.990830671451076e-05, "loss": 1.0796, "step": 27368 }, { "epoch": 0.03, "learning_rate": 4.990829999316197e-05, "loss": 1.0934, "step": 27369 }, { "epoch": 0.03, "learning_rate": 4.9908293271567286e-05, "loss": 1.026, "step": 27370 }, { "epoch": 0.03, "learning_rate": 4.990828654972673e-05, "loss": 1.1199, "step": 27371 }, { "epoch": 0.03, "learning_rate": 4.990827982764028e-05, "loss": 1.1187, "step": 27372 }, { "epoch": 0.03, "learning_rate": 4.9908273105307944e-05, "loss": 1.5239, "step": 27373 }, { "epoch": 0.03, "learning_rate": 4.990826638272973e-05, "loss": 1.2044, "step": 27374 }, { "epoch": 0.03, "learning_rate": 4.9908259659905635e-05, "loss": 1.5089, "step": 27375 }, { "epoch": 0.03, "learning_rate": 4.9908252936835644e-05, "loss": 1.2654, "step": 27376 }, { "epoch": 0.03, "learning_rate": 4.9908246213519784e-05, "loss": 1.3399, "step": 27377 }, { "epoch": 0.03, "learning_rate": 4.9908239489958034e-05, "loss": 1.3356, "step": 27378 }, { "epoch": 0.03, "learning_rate": 4.99082327661504e-05, "loss": 1.3519, "step": 27379 }, { "epoch": 0.03, "learning_rate": 4.990822604209688e-05, "loss": 1.3109, "step": 27380 }, { "epoch": 0.03, "learning_rate": 4.9908219317797475e-05, "loss": 1.2395, "step": 27381 }, { "epoch": 0.03, "learning_rate": 4.990821259325219e-05, "loss": 1.0454, "step": 27382 }, { "epoch": 0.03, "learning_rate": 4.990820586846102e-05, "loss": 1.045, "step": 27383 }, { "epoch": 0.03, "learning_rate": 4.990819914342397e-05, "loss": 0.8534, "step": 27384 }, { "epoch": 0.03, "learning_rate": 4.990819241814104e-05, "loss": 0.9316, "step": 27385 }, { "epoch": 0.03, "learning_rate": 4.990818569261222e-05, "loss": 0.8912, "step": 27386 }, { "epoch": 0.03, "learning_rate": 4.990817896683752e-05, "loss": 1.1141, "step": 27387 }, { "epoch": 0.03, "learning_rate": 4.990817224081694e-05, "loss": 0.9962, "step": 27388 }, { "epoch": 0.03, "learning_rate": 4.990816551455047e-05, "loss": 0.9092, "step": 27389 }, { "epoch": 0.03, "learning_rate": 4.9908158788038115e-05, "loss": 1.4051, "step": 27390 }, { "epoch": 0.03, "learning_rate": 4.990815206127989e-05, "loss": 1.0029, "step": 27391 }, { "epoch": 0.03, "learning_rate": 4.990814533427577e-05, "loss": 1.1518, "step": 27392 }, { "epoch": 0.03, "learning_rate": 4.990813860702578e-05, "loss": 1.9848, "step": 27393 }, { "epoch": 0.03, "learning_rate": 4.99081318795299e-05, "loss": 1.7895, "step": 27394 }, { "epoch": 0.03, "learning_rate": 4.9908125151788134e-05, "loss": 0.9852, "step": 27395 }, { "epoch": 0.03, "learning_rate": 4.990811842380049e-05, "loss": 1.0671, "step": 27396 }, { "epoch": 0.03, "learning_rate": 4.990811169556696e-05, "loss": 1.1041, "step": 27397 }, { "epoch": 0.03, "learning_rate": 4.990810496708754e-05, "loss": 1.1199, "step": 27398 }, { "epoch": 0.03, "learning_rate": 4.990809823836225e-05, "loss": 1.0155, "step": 27399 }, { "epoch": 0.03, "learning_rate": 4.990809150939107e-05, "loss": 0.938, "step": 27400 }, { "epoch": 0.03, "learning_rate": 4.990808478017402e-05, "loss": 0.6957, "step": 27401 }, { "epoch": 0.03, "learning_rate": 4.990807805071108e-05, "loss": 0.8675, "step": 27402 }, { "epoch": 0.03, "learning_rate": 4.9908071321002256e-05, "loss": 0.9742, "step": 27403 }, { "epoch": 0.03, "learning_rate": 4.9908064591047555e-05, "loss": 0.8063, "step": 27404 }, { "epoch": 0.03, "learning_rate": 4.990805786084697e-05, "loss": 0.9868, "step": 27405 }, { "epoch": 0.03, "learning_rate": 4.99080511304005e-05, "loss": 0.9545, "step": 27406 }, { "epoch": 0.03, "learning_rate": 4.990804439970815e-05, "loss": 1.0049, "step": 27407 }, { "epoch": 0.03, "learning_rate": 4.990803766876992e-05, "loss": 0.6532, "step": 27408 }, { "epoch": 0.03, "learning_rate": 4.990803093758582e-05, "loss": 0.9871, "step": 27409 }, { "epoch": 0.03, "learning_rate": 4.9908024206155815e-05, "loss": 0.837, "step": 27410 }, { "epoch": 0.03, "learning_rate": 4.990801747447994e-05, "loss": 1.3217, "step": 27411 }, { "epoch": 0.03, "learning_rate": 4.990801074255819e-05, "loss": 0.8155, "step": 27412 }, { "epoch": 0.03, "learning_rate": 4.990800401039055e-05, "loss": 0.4876, "step": 27413 }, { "epoch": 0.03, "learning_rate": 4.9907997277977025e-05, "loss": 1.2493, "step": 27414 }, { "epoch": 0.03, "learning_rate": 4.9907990545317626e-05, "loss": 0.9459, "step": 27415 }, { "epoch": 0.03, "learning_rate": 4.990798381241234e-05, "loss": 1.1274, "step": 27416 }, { "epoch": 0.03, "learning_rate": 4.9907977079261184e-05, "loss": 1.0627, "step": 27417 }, { "epoch": 0.03, "learning_rate": 4.9907970345864134e-05, "loss": 0.8461, "step": 27418 }, { "epoch": 0.03, "learning_rate": 4.990796361222121e-05, "loss": 1.0114, "step": 27419 }, { "epoch": 0.03, "learning_rate": 4.9907956878332404e-05, "loss": 1.125, "step": 27420 }, { "epoch": 0.03, "learning_rate": 4.990795014419771e-05, "loss": 0.7903, "step": 27421 }, { "epoch": 0.03, "learning_rate": 4.990794340981715e-05, "loss": 0.5785, "step": 27422 }, { "epoch": 0.03, "learning_rate": 4.99079366751907e-05, "loss": 0.5387, "step": 27423 }, { "epoch": 0.03, "learning_rate": 4.990792994031837e-05, "loss": 0.3755, "step": 27424 }, { "epoch": 0.03, "learning_rate": 4.9907923205200155e-05, "loss": 0.797, "step": 27425 }, { "epoch": 0.03, "learning_rate": 4.990791646983607e-05, "loss": 0.567, "step": 27426 }, { "epoch": 0.03, "learning_rate": 4.99079097342261e-05, "loss": 0.4354, "step": 27427 }, { "epoch": 0.03, "learning_rate": 4.9907902998370246e-05, "loss": 0.5497, "step": 27428 }, { "epoch": 0.03, "learning_rate": 4.990789626226852e-05, "loss": 0.8607, "step": 27429 }, { "epoch": 0.03, "learning_rate": 4.99078895259209e-05, "loss": 0.9411, "step": 27430 }, { "epoch": 0.03, "learning_rate": 4.990788278932741e-05, "loss": 1.0779, "step": 27431 }, { "epoch": 0.03, "learning_rate": 4.990787605248804e-05, "loss": 1.0584, "step": 27432 }, { "epoch": 0.03, "learning_rate": 4.9907869315402786e-05, "loss": 0.886, "step": 27433 }, { "epoch": 0.03, "learning_rate": 4.9907862578071655e-05, "loss": 0.8585, "step": 27434 }, { "epoch": 0.03, "learning_rate": 4.9907855840494646e-05, "loss": 0.9671, "step": 27435 }, { "epoch": 0.03, "learning_rate": 4.990784910267175e-05, "loss": 1.1184, "step": 27436 }, { "epoch": 0.03, "learning_rate": 4.990784236460298e-05, "loss": 0.803, "step": 27437 }, { "epoch": 0.03, "learning_rate": 4.990783562628833e-05, "loss": 1.1492, "step": 27438 }, { "epoch": 0.03, "learning_rate": 4.99078288877278e-05, "loss": 0.9381, "step": 27439 }, { "epoch": 0.03, "learning_rate": 4.9907822148921386e-05, "loss": 1.2156, "step": 27440 }, { "epoch": 0.03, "learning_rate": 4.99078154098691e-05, "loss": 0.892, "step": 27441 }, { "epoch": 0.03, "learning_rate": 4.9907808670570924e-05, "loss": 0.7297, "step": 27442 }, { "epoch": 0.03, "learning_rate": 4.9907801931026875e-05, "loss": 1.1257, "step": 27443 }, { "epoch": 0.03, "learning_rate": 4.990779519123695e-05, "loss": 1.1545, "step": 27444 }, { "epoch": 0.03, "learning_rate": 4.990778845120114e-05, "loss": 0.9053, "step": 27445 }, { "epoch": 0.03, "learning_rate": 4.9907781710919453e-05, "loss": 0.7571, "step": 27446 }, { "epoch": 0.03, "learning_rate": 4.9907774970391884e-05, "loss": 0.789, "step": 27447 }, { "epoch": 0.03, "learning_rate": 4.9907768229618444e-05, "loss": 0.8883, "step": 27448 }, { "epoch": 0.03, "learning_rate": 4.990776148859912e-05, "loss": 1.0241, "step": 27449 }, { "epoch": 0.03, "learning_rate": 4.9907754747333914e-05, "loss": 1.134, "step": 27450 }, { "epoch": 0.03, "learning_rate": 4.990774800582283e-05, "loss": 1.1467, "step": 27451 }, { "epoch": 0.03, "learning_rate": 4.990774126406588e-05, "loss": 0.7086, "step": 27452 }, { "epoch": 0.03, "learning_rate": 4.9907734522063034e-05, "loss": 0.6248, "step": 27453 }, { "epoch": 0.03, "learning_rate": 4.990772777981432e-05, "loss": 1.2557, "step": 27454 }, { "epoch": 0.03, "learning_rate": 4.9907721037319723e-05, "loss": 1.7849, "step": 27455 }, { "epoch": 0.03, "learning_rate": 4.990771429457924e-05, "loss": 1.0018, "step": 27456 }, { "epoch": 0.03, "learning_rate": 4.990770755159289e-05, "loss": 0.7148, "step": 27457 }, { "epoch": 0.03, "learning_rate": 4.990770080836066e-05, "loss": 0.6617, "step": 27458 }, { "epoch": 0.03, "learning_rate": 4.990769406488255e-05, "loss": 0.89, "step": 27459 }, { "epoch": 0.03, "learning_rate": 4.990768732115856e-05, "loss": 0.8707, "step": 27460 }, { "epoch": 0.03, "learning_rate": 4.990768057718869e-05, "loss": 0.8775, "step": 27461 }, { "epoch": 0.03, "learning_rate": 4.990767383297295e-05, "loss": 0.584, "step": 27462 }, { "epoch": 0.03, "learning_rate": 4.9907667088511324e-05, "loss": 0.9088, "step": 27463 }, { "epoch": 0.03, "learning_rate": 4.990766034380382e-05, "loss": 0.9741, "step": 27464 }, { "epoch": 0.03, "learning_rate": 4.9907653598850445e-05, "loss": 1.0751, "step": 27465 }, { "epoch": 0.03, "learning_rate": 4.990764685365119e-05, "loss": 1.1836, "step": 27466 }, { "epoch": 0.03, "learning_rate": 4.990764010820605e-05, "loss": 0.8842, "step": 27467 }, { "epoch": 0.03, "learning_rate": 4.990763336251504e-05, "loss": 1.3296, "step": 27468 }, { "epoch": 0.03, "learning_rate": 4.990762661657815e-05, "loss": 1.2634, "step": 27469 }, { "epoch": 0.03, "learning_rate": 4.9907619870395385e-05, "loss": 0.9266, "step": 27470 }, { "epoch": 0.03, "learning_rate": 4.990761312396673e-05, "loss": 0.8711, "step": 27471 }, { "epoch": 0.03, "learning_rate": 4.990760637729222e-05, "loss": 1.0534, "step": 27472 }, { "epoch": 0.03, "learning_rate": 4.990759963037181e-05, "loss": 0.9023, "step": 27473 }, { "epoch": 0.03, "learning_rate": 4.990759288320554e-05, "loss": 0.7195, "step": 27474 }, { "epoch": 0.03, "learning_rate": 4.9907586135793386e-05, "loss": 0.9525, "step": 27475 }, { "epoch": 0.03, "learning_rate": 4.990757938813535e-05, "loss": 1.1232, "step": 27476 }, { "epoch": 0.03, "learning_rate": 4.9907572640231445e-05, "loss": 0.9577, "step": 27477 }, { "epoch": 0.03, "learning_rate": 4.9907565892081657e-05, "loss": 1.0836, "step": 27478 }, { "epoch": 0.03, "learning_rate": 4.9907559143686e-05, "loss": 1.0288, "step": 27479 }, { "epoch": 0.03, "learning_rate": 4.9907552395044456e-05, "loss": 0.9623, "step": 27480 }, { "epoch": 0.03, "learning_rate": 4.9907545646157037e-05, "loss": 0.9658, "step": 27481 }, { "epoch": 0.03, "learning_rate": 4.990753889702374e-05, "loss": 1.0748, "step": 27482 }, { "epoch": 0.03, "learning_rate": 4.9907532147644575e-05, "loss": 0.9827, "step": 27483 }, { "epoch": 0.03, "learning_rate": 4.9907525398019526e-05, "loss": 1.1319, "step": 27484 }, { "epoch": 0.03, "learning_rate": 4.990751864814861e-05, "loss": 1.0493, "step": 27485 }, { "epoch": 0.03, "learning_rate": 4.9907511898031804e-05, "loss": 0.6807, "step": 27486 }, { "epoch": 0.03, "learning_rate": 4.990750514766913e-05, "loss": 0.6604, "step": 27487 }, { "epoch": 0.03, "learning_rate": 4.9907498397060576e-05, "loss": 0.9694, "step": 27488 }, { "epoch": 0.03, "learning_rate": 4.990749164620614e-05, "loss": 1.0879, "step": 27489 }, { "epoch": 0.03, "learning_rate": 4.990748489510584e-05, "loss": 1.0444, "step": 27490 }, { "epoch": 0.03, "learning_rate": 4.9907478143759654e-05, "loss": 1.0801, "step": 27491 }, { "epoch": 0.03, "learning_rate": 4.9907471392167605e-05, "loss": 0.9199, "step": 27492 }, { "epoch": 0.03, "learning_rate": 4.990746464032967e-05, "loss": 0.982, "step": 27493 }, { "epoch": 0.03, "learning_rate": 4.9907457888245856e-05, "loss": 1.2102, "step": 27494 }, { "epoch": 0.03, "learning_rate": 4.990745113591617e-05, "loss": 0.7535, "step": 27495 }, { "epoch": 0.03, "learning_rate": 4.9907444383340606e-05, "loss": 0.8128, "step": 27496 }, { "epoch": 0.03, "learning_rate": 4.990743763051917e-05, "loss": 1.0657, "step": 27497 }, { "epoch": 0.03, "learning_rate": 4.9907430877451863e-05, "loss": 0.7447, "step": 27498 }, { "epoch": 0.03, "learning_rate": 4.990742412413867e-05, "loss": 0.979, "step": 27499 }, { "epoch": 0.03, "learning_rate": 4.990741737057961e-05, "loss": 0.972, "step": 27500 }, { "epoch": 0.03, "eval_loss": 1.0637550354003906, "eval_runtime": 124.9122, "eval_samples_per_second": 11.088, "eval_steps_per_second": 5.548, "step": 27500 }, { "epoch": 0.03, "learning_rate": 4.990741061677466e-05, "loss": 1.2999, "step": 27501 }, { "epoch": 0.03, "learning_rate": 4.990740386272385e-05, "loss": 1.151, "step": 27502 }, { "epoch": 0.03, "learning_rate": 4.990739710842716e-05, "loss": 1.0148, "step": 27503 }, { "epoch": 0.03, "learning_rate": 4.99073903538846e-05, "loss": 1.0193, "step": 27504 }, { "epoch": 0.03, "learning_rate": 4.9907383599096154e-05, "loss": 1.006, "step": 27505 }, { "epoch": 0.03, "learning_rate": 4.990737684406184e-05, "loss": 1.2982, "step": 27506 }, { "epoch": 0.03, "learning_rate": 4.9907370088781644e-05, "loss": 1.0302, "step": 27507 }, { "epoch": 0.03, "learning_rate": 4.9907363333255574e-05, "loss": 1.0527, "step": 27508 }, { "epoch": 0.03, "learning_rate": 4.9907356577483633e-05, "loss": 1.0209, "step": 27509 }, { "epoch": 0.03, "learning_rate": 4.990734982146582e-05, "loss": 1.0983, "step": 27510 }, { "epoch": 0.03, "learning_rate": 4.990734306520213e-05, "loss": 0.9588, "step": 27511 }, { "epoch": 0.03, "learning_rate": 4.9907336308692566e-05, "loss": 1.1109, "step": 27512 }, { "epoch": 0.03, "learning_rate": 4.990732955193712e-05, "loss": 0.9269, "step": 27513 }, { "epoch": 0.03, "learning_rate": 4.990732279493581e-05, "loss": 1.1877, "step": 27514 }, { "epoch": 0.03, "learning_rate": 4.990731603768861e-05, "loss": 0.8668, "step": 27515 }, { "epoch": 0.03, "learning_rate": 4.9907309280195545e-05, "loss": 0.911, "step": 27516 }, { "epoch": 0.03, "learning_rate": 4.990730252245661e-05, "loss": 0.9705, "step": 27517 }, { "epoch": 0.03, "learning_rate": 4.9907295764471794e-05, "loss": 0.9137, "step": 27518 }, { "epoch": 0.03, "learning_rate": 4.990728900624111e-05, "loss": 0.8024, "step": 27519 }, { "epoch": 0.03, "learning_rate": 4.990728224776455e-05, "loss": 1.2102, "step": 27520 }, { "epoch": 0.03, "learning_rate": 4.990727548904211e-05, "loss": 1.0246, "step": 27521 }, { "epoch": 0.03, "learning_rate": 4.99072687300738e-05, "loss": 1.0258, "step": 27522 }, { "epoch": 0.03, "learning_rate": 4.990726197085961e-05, "loss": 0.916, "step": 27523 }, { "epoch": 0.03, "learning_rate": 4.9907255211399555e-05, "loss": 0.861, "step": 27524 }, { "epoch": 0.03, "learning_rate": 4.990724845169362e-05, "loss": 0.895, "step": 27525 }, { "epoch": 0.03, "learning_rate": 4.990724169174181e-05, "loss": 1.0249, "step": 27526 }, { "epoch": 0.03, "learning_rate": 4.990723493154414e-05, "loss": 1.1805, "step": 27527 }, { "epoch": 0.03, "learning_rate": 4.9907228171100586e-05, "loss": 1.1585, "step": 27528 }, { "epoch": 0.03, "learning_rate": 4.990722141041115e-05, "loss": 0.7603, "step": 27529 }, { "epoch": 0.03, "learning_rate": 4.9907214649475855e-05, "loss": 1.0841, "step": 27530 }, { "epoch": 0.03, "learning_rate": 4.990720788829468e-05, "loss": 1.0684, "step": 27531 }, { "epoch": 0.03, "learning_rate": 4.990720112686763e-05, "loss": 0.978, "step": 27532 }, { "epoch": 0.03, "learning_rate": 4.990719436519471e-05, "loss": 1.0668, "step": 27533 }, { "epoch": 0.03, "learning_rate": 4.990718760327592e-05, "loss": 1.1842, "step": 27534 }, { "epoch": 0.03, "learning_rate": 4.9907180841111246e-05, "loss": 0.8993, "step": 27535 }, { "epoch": 0.03, "learning_rate": 4.99071740787007e-05, "loss": 1.0363, "step": 27536 }, { "epoch": 0.03, "learning_rate": 4.990716731604429e-05, "loss": 0.8786, "step": 27537 }, { "epoch": 0.03, "learning_rate": 4.9907160553142004e-05, "loss": 0.7507, "step": 27538 }, { "epoch": 0.03, "learning_rate": 4.9907153789993844e-05, "loss": 1.3899, "step": 27539 }, { "epoch": 0.03, "learning_rate": 4.9907147026599806e-05, "loss": 1.2364, "step": 27540 }, { "epoch": 0.03, "learning_rate": 4.99071402629599e-05, "loss": 1.3381, "step": 27541 }, { "epoch": 0.03, "learning_rate": 4.990713349907412e-05, "loss": 0.9828, "step": 27542 }, { "epoch": 0.03, "learning_rate": 4.9907126734942475e-05, "loss": 1.161, "step": 27543 }, { "epoch": 0.03, "learning_rate": 4.9907119970564944e-05, "loss": 1.0636, "step": 27544 }, { "epoch": 0.03, "learning_rate": 4.9907113205941544e-05, "loss": 1.0134, "step": 27545 }, { "epoch": 0.03, "learning_rate": 4.990710644107228e-05, "loss": 1.0167, "step": 27546 }, { "epoch": 0.03, "learning_rate": 4.990709967595714e-05, "loss": 1.0319, "step": 27547 }, { "epoch": 0.03, "learning_rate": 4.990709291059612e-05, "loss": 0.6867, "step": 27548 }, { "epoch": 0.03, "learning_rate": 4.9907086144989235e-05, "loss": 0.934, "step": 27549 }, { "epoch": 0.03, "learning_rate": 4.990707937913648e-05, "loss": 1.0217, "step": 27550 }, { "epoch": 0.03, "learning_rate": 4.9907072613037845e-05, "loss": 0.9361, "step": 27551 }, { "epoch": 0.03, "learning_rate": 4.990706584669335e-05, "loss": 1.1049, "step": 27552 }, { "epoch": 0.03, "learning_rate": 4.990705908010297e-05, "loss": 0.8648, "step": 27553 }, { "epoch": 0.03, "learning_rate": 4.9907052313266723e-05, "loss": 0.9345, "step": 27554 }, { "epoch": 0.03, "learning_rate": 4.99070455461846e-05, "loss": 1.0283, "step": 27555 }, { "epoch": 0.03, "learning_rate": 4.9907038778856606e-05, "loss": 1.1007, "step": 27556 }, { "epoch": 0.03, "learning_rate": 4.9907032011282746e-05, "loss": 0.9469, "step": 27557 }, { "epoch": 0.03, "learning_rate": 4.9907025243463016e-05, "loss": 0.9699, "step": 27558 }, { "epoch": 0.03, "learning_rate": 4.99070184753974e-05, "loss": 1.0613, "step": 27559 }, { "epoch": 0.03, "learning_rate": 4.9907011707085925e-05, "loss": 1.0942, "step": 27560 }, { "epoch": 0.03, "learning_rate": 4.990700493852858e-05, "loss": 1.0141, "step": 27561 }, { "epoch": 0.03, "learning_rate": 4.9906998169725355e-05, "loss": 1.2444, "step": 27562 }, { "epoch": 0.03, "learning_rate": 4.990699140067626e-05, "loss": 1.0579, "step": 27563 }, { "epoch": 0.03, "learning_rate": 4.99069846313813e-05, "loss": 1.2241, "step": 27564 }, { "epoch": 0.03, "learning_rate": 4.9906977861840464e-05, "loss": 0.9581, "step": 27565 }, { "epoch": 0.03, "learning_rate": 4.9906971092053754e-05, "loss": 0.987, "step": 27566 }, { "epoch": 0.03, "learning_rate": 4.9906964322021174e-05, "loss": 0.3809, "step": 27567 }, { "epoch": 0.03, "learning_rate": 4.990695755174273e-05, "loss": 0.7408, "step": 27568 }, { "epoch": 0.03, "learning_rate": 4.990695078121842e-05, "loss": 0.8478, "step": 27569 }, { "epoch": 0.03, "learning_rate": 4.990694401044822e-05, "loss": 1.1592, "step": 27570 }, { "epoch": 0.03, "learning_rate": 4.990693723943216e-05, "loss": 1.0349, "step": 27571 }, { "epoch": 0.03, "learning_rate": 4.990693046817023e-05, "loss": 0.994, "step": 27572 }, { "epoch": 0.03, "learning_rate": 4.9906923696662425e-05, "loss": 1.1201, "step": 27573 }, { "epoch": 0.03, "learning_rate": 4.9906916924908755e-05, "loss": 0.9709, "step": 27574 }, { "epoch": 0.03, "learning_rate": 4.9906910152909216e-05, "loss": 1.0456, "step": 27575 }, { "epoch": 0.03, "learning_rate": 4.990690338066379e-05, "loss": 0.9221, "step": 27576 }, { "epoch": 0.03, "learning_rate": 4.990689660817251e-05, "loss": 0.7766, "step": 27577 }, { "epoch": 0.03, "learning_rate": 4.990688983543536e-05, "loss": 0.9734, "step": 27578 }, { "epoch": 0.03, "learning_rate": 4.990688306245233e-05, "loss": 1.0414, "step": 27579 }, { "epoch": 0.03, "learning_rate": 4.9906876289223434e-05, "loss": 0.9512, "step": 27580 }, { "epoch": 0.03, "learning_rate": 4.990686951574867e-05, "loss": 1.0802, "step": 27581 }, { "epoch": 0.03, "learning_rate": 4.990686274202803e-05, "loss": 0.8409, "step": 27582 }, { "epoch": 0.03, "learning_rate": 4.9906855968061526e-05, "loss": 0.745, "step": 27583 }, { "epoch": 0.03, "learning_rate": 4.990684919384915e-05, "loss": 0.91, "step": 27584 }, { "epoch": 0.03, "learning_rate": 4.9906842419390904e-05, "loss": 1.1638, "step": 27585 }, { "epoch": 0.03, "learning_rate": 4.990683564468679e-05, "loss": 1.0809, "step": 27586 }, { "epoch": 0.03, "learning_rate": 4.9906828869736796e-05, "loss": 0.8113, "step": 27587 }, { "epoch": 0.03, "learning_rate": 4.9906822094540954e-05, "loss": 0.9303, "step": 27588 }, { "epoch": 0.03, "learning_rate": 4.990681531909922e-05, "loss": 0.8237, "step": 27589 }, { "epoch": 0.03, "learning_rate": 4.9906808543411626e-05, "loss": 1.0734, "step": 27590 }, { "epoch": 0.03, "learning_rate": 4.990680176747816e-05, "loss": 1.2248, "step": 27591 }, { "epoch": 0.03, "learning_rate": 4.990679499129883e-05, "loss": 0.8428, "step": 27592 }, { "epoch": 0.03, "learning_rate": 4.9906788214873626e-05, "loss": 0.8084, "step": 27593 }, { "epoch": 0.03, "learning_rate": 4.990678143820256e-05, "loss": 1.0489, "step": 27594 }, { "epoch": 0.03, "learning_rate": 4.990677466128562e-05, "loss": 0.8364, "step": 27595 }, { "epoch": 0.03, "learning_rate": 4.99067678841228e-05, "loss": 0.9677, "step": 27596 }, { "epoch": 0.03, "learning_rate": 4.990676110671412e-05, "loss": 1.3534, "step": 27597 }, { "epoch": 0.03, "learning_rate": 4.990675432905957e-05, "loss": 1.0235, "step": 27598 }, { "epoch": 0.03, "learning_rate": 4.990674755115916e-05, "loss": 0.866, "step": 27599 }, { "epoch": 0.03, "learning_rate": 4.990674077301287e-05, "loss": 0.7143, "step": 27600 }, { "epoch": 0.03, "learning_rate": 4.990673399462071e-05, "loss": 1.0331, "step": 27601 }, { "epoch": 0.03, "learning_rate": 4.990672721598268e-05, "loss": 0.9158, "step": 27602 }, { "epoch": 0.03, "learning_rate": 4.99067204370988e-05, "loss": 0.9673, "step": 27603 }, { "epoch": 0.03, "learning_rate": 4.990671365796903e-05, "loss": 0.624, "step": 27604 }, { "epoch": 0.03, "learning_rate": 4.99067068785934e-05, "loss": 0.7752, "step": 27605 }, { "epoch": 0.03, "learning_rate": 4.9906700098971905e-05, "loss": 0.9244, "step": 27606 }, { "epoch": 0.03, "learning_rate": 4.9906693319104533e-05, "loss": 0.9467, "step": 27607 }, { "epoch": 0.03, "learning_rate": 4.99066865389913e-05, "loss": 1.0462, "step": 27608 }, { "epoch": 0.03, "learning_rate": 4.9906679758632195e-05, "loss": 0.8384, "step": 27609 }, { "epoch": 0.03, "learning_rate": 4.990667297802723e-05, "loss": 0.9308, "step": 27610 }, { "epoch": 0.03, "learning_rate": 4.990666619717638e-05, "loss": 1.343, "step": 27611 }, { "epoch": 0.03, "learning_rate": 4.990665941607967e-05, "loss": 1.0103, "step": 27612 }, { "epoch": 0.03, "learning_rate": 4.990665263473709e-05, "loss": 1.0382, "step": 27613 }, { "epoch": 0.03, "learning_rate": 4.990664585314865e-05, "loss": 0.7143, "step": 27614 }, { "epoch": 0.03, "learning_rate": 4.990663907131434e-05, "loss": 0.678, "step": 27615 }, { "epoch": 0.03, "learning_rate": 4.9906632289234154e-05, "loss": 0.8631, "step": 27616 }, { "epoch": 0.03, "learning_rate": 4.990662550690811e-05, "loss": 0.8231, "step": 27617 }, { "epoch": 0.03, "learning_rate": 4.9906618724336196e-05, "loss": 0.7382, "step": 27618 }, { "epoch": 0.03, "learning_rate": 4.990661194151841e-05, "loss": 1.2207, "step": 27619 }, { "epoch": 0.03, "learning_rate": 4.990660515845476e-05, "loss": 1.0914, "step": 27620 }, { "epoch": 0.03, "learning_rate": 4.990659837514524e-05, "loss": 0.7356, "step": 27621 }, { "epoch": 0.03, "learning_rate": 4.9906591591589857e-05, "loss": 1.1634, "step": 27622 }, { "epoch": 0.03, "learning_rate": 4.99065848077886e-05, "loss": 0.9894, "step": 27623 }, { "epoch": 0.03, "learning_rate": 4.990657802374148e-05, "loss": 1.0268, "step": 27624 }, { "epoch": 0.03, "learning_rate": 4.99065712394485e-05, "loss": 0.6853, "step": 27625 }, { "epoch": 0.03, "learning_rate": 4.990656445490964e-05, "loss": 1.3608, "step": 27626 }, { "epoch": 0.03, "learning_rate": 4.990655767012492e-05, "loss": 0.9854, "step": 27627 }, { "epoch": 0.03, "learning_rate": 4.990655088509433e-05, "loss": 1.0742, "step": 27628 }, { "epoch": 0.03, "learning_rate": 4.990654409981787e-05, "loss": 0.9559, "step": 27629 }, { "epoch": 0.03, "learning_rate": 4.990653731429555e-05, "loss": 0.7125, "step": 27630 }, { "epoch": 0.03, "learning_rate": 4.990653052852736e-05, "loss": 0.6802, "step": 27631 }, { "epoch": 0.03, "learning_rate": 4.99065237425133e-05, "loss": 0.9675, "step": 27632 }, { "epoch": 0.03, "learning_rate": 4.990651695625338e-05, "loss": 1.1337, "step": 27633 }, { "epoch": 0.03, "learning_rate": 4.9906510169747586e-05, "loss": 1.0941, "step": 27634 }, { "epoch": 0.03, "learning_rate": 4.990650338299593e-05, "loss": 1.2268, "step": 27635 }, { "epoch": 0.03, "learning_rate": 4.9906496595998406e-05, "loss": 1.1015, "step": 27636 }, { "epoch": 0.03, "learning_rate": 4.990648980875502e-05, "loss": 0.9686, "step": 27637 }, { "epoch": 0.03, "learning_rate": 4.990648302126576e-05, "loss": 0.8732, "step": 27638 }, { "epoch": 0.03, "learning_rate": 4.9906476233530644e-05, "loss": 1.0634, "step": 27639 }, { "epoch": 0.03, "learning_rate": 4.990646944554965e-05, "loss": 1.1217, "step": 27640 }, { "epoch": 0.03, "learning_rate": 4.990646265732279e-05, "loss": 1.4184, "step": 27641 }, { "epoch": 0.03, "learning_rate": 4.9906455868850074e-05, "loss": 1.1893, "step": 27642 }, { "epoch": 0.03, "learning_rate": 4.990644908013149e-05, "loss": 0.8549, "step": 27643 }, { "epoch": 0.03, "learning_rate": 4.9906442291167036e-05, "loss": 0.8826, "step": 27644 }, { "epoch": 0.03, "learning_rate": 4.9906435501956715e-05, "loss": 1.3192, "step": 27645 }, { "epoch": 0.03, "learning_rate": 4.990642871250053e-05, "loss": 0.953, "step": 27646 }, { "epoch": 0.03, "learning_rate": 4.990642192279848e-05, "loss": 1.0286, "step": 27647 }, { "epoch": 0.03, "learning_rate": 4.990641513285057e-05, "loss": 1.2223, "step": 27648 }, { "epoch": 0.03, "learning_rate": 4.9906408342656786e-05, "loss": 1.1291, "step": 27649 }, { "epoch": 0.03, "learning_rate": 4.9906401552217136e-05, "loss": 0.645, "step": 27650 }, { "epoch": 0.03, "learning_rate": 4.990639476153162e-05, "loss": 0.2411, "step": 27651 }, { "epoch": 0.03, "learning_rate": 4.9906387970600246e-05, "loss": 0.2393, "step": 27652 }, { "epoch": 0.03, "learning_rate": 4.9906381179423e-05, "loss": 0.7807, "step": 27653 }, { "epoch": 0.03, "learning_rate": 4.990637438799989e-05, "loss": 0.9829, "step": 27654 }, { "epoch": 0.03, "learning_rate": 4.990636759633092e-05, "loss": 0.916, "step": 27655 }, { "epoch": 0.03, "learning_rate": 4.9906360804416075e-05, "loss": 1.0634, "step": 27656 }, { "epoch": 0.03, "learning_rate": 4.9906354012255375e-05, "loss": 1.2363, "step": 27657 }, { "epoch": 0.03, "learning_rate": 4.9906347219848806e-05, "loss": 1.0851, "step": 27658 }, { "epoch": 0.03, "learning_rate": 4.9906340427196374e-05, "loss": 0.4341, "step": 27659 }, { "epoch": 0.03, "learning_rate": 4.990633363429807e-05, "loss": 0.506, "step": 27660 }, { "epoch": 0.03, "learning_rate": 4.9906326841153906e-05, "loss": 0.8091, "step": 27661 }, { "epoch": 0.03, "learning_rate": 4.990632004776388e-05, "loss": 0.9059, "step": 27662 }, { "epoch": 0.03, "learning_rate": 4.9906313254127986e-05, "loss": 0.987, "step": 27663 }, { "epoch": 0.03, "learning_rate": 4.9906306460246224e-05, "loss": 1.2362, "step": 27664 }, { "epoch": 0.03, "learning_rate": 4.9906299666118606e-05, "loss": 1.0144, "step": 27665 }, { "epoch": 0.03, "learning_rate": 4.990629287174512e-05, "loss": 1.1037, "step": 27666 }, { "epoch": 0.03, "learning_rate": 4.990628607712576e-05, "loss": 1.3354, "step": 27667 }, { "epoch": 0.03, "learning_rate": 4.9906279282260546e-05, "loss": 1.3847, "step": 27668 }, { "epoch": 0.03, "learning_rate": 4.990627248714947e-05, "loss": 1.0799, "step": 27669 }, { "epoch": 0.03, "learning_rate": 4.990626569179252e-05, "loss": 1.0909, "step": 27670 }, { "epoch": 0.03, "learning_rate": 4.990625889618972e-05, "loss": 0.8836, "step": 27671 }, { "epoch": 0.03, "learning_rate": 4.990625210034104e-05, "loss": 1.0071, "step": 27672 }, { "epoch": 0.03, "learning_rate": 4.99062453042465e-05, "loss": 0.6878, "step": 27673 }, { "epoch": 0.03, "learning_rate": 4.99062385079061e-05, "loss": 0.7104, "step": 27674 }, { "epoch": 0.03, "learning_rate": 4.990623171131984e-05, "loss": 0.7282, "step": 27675 }, { "epoch": 0.03, "learning_rate": 4.990622491448771e-05, "loss": 0.9639, "step": 27676 }, { "epoch": 0.03, "learning_rate": 4.990621811740972e-05, "loss": 0.9342, "step": 27677 }, { "epoch": 0.03, "learning_rate": 4.990621132008586e-05, "loss": 0.894, "step": 27678 }, { "epoch": 0.03, "learning_rate": 4.9906204522516144e-05, "loss": 1.1951, "step": 27679 }, { "epoch": 0.03, "learning_rate": 4.990619772470056e-05, "loss": 0.8055, "step": 27680 }, { "epoch": 0.03, "learning_rate": 4.9906190926639115e-05, "loss": 0.8545, "step": 27681 }, { "epoch": 0.03, "learning_rate": 4.99061841283318e-05, "loss": 0.8146, "step": 27682 }, { "epoch": 0.03, "learning_rate": 4.990617732977863e-05, "loss": 1.1256, "step": 27683 }, { "epoch": 0.03, "learning_rate": 4.990617053097959e-05, "loss": 0.9315, "step": 27684 }, { "epoch": 0.03, "learning_rate": 4.9906163731934694e-05, "loss": 0.9664, "step": 27685 }, { "epoch": 0.03, "learning_rate": 4.9906156932643936e-05, "loss": 1.2161, "step": 27686 }, { "epoch": 0.03, "learning_rate": 4.990615013310731e-05, "loss": 0.9261, "step": 27687 }, { "epoch": 0.03, "learning_rate": 4.9906143333324816e-05, "loss": 1.0038, "step": 27688 }, { "epoch": 0.03, "learning_rate": 4.990613653329647e-05, "loss": 1.0052, "step": 27689 }, { "epoch": 0.03, "learning_rate": 4.9906129733022244e-05, "loss": 0.9924, "step": 27690 }, { "epoch": 0.03, "learning_rate": 4.990612293250217e-05, "loss": 0.7458, "step": 27691 }, { "epoch": 0.03, "learning_rate": 4.9906116131736226e-05, "loss": 0.8152, "step": 27692 }, { "epoch": 0.03, "learning_rate": 4.9906109330724425e-05, "loss": 0.7456, "step": 27693 }, { "epoch": 0.03, "learning_rate": 4.990610252946676e-05, "loss": 1.0483, "step": 27694 }, { "epoch": 0.03, "learning_rate": 4.990609572796323e-05, "loss": 0.9304, "step": 27695 }, { "epoch": 0.03, "learning_rate": 4.990608892621384e-05, "loss": 1.0806, "step": 27696 }, { "epoch": 0.03, "learning_rate": 4.990608212421859e-05, "loss": 0.7899, "step": 27697 }, { "epoch": 0.03, "learning_rate": 4.9906075321977476e-05, "loss": 1.106, "step": 27698 }, { "epoch": 0.03, "learning_rate": 4.990606851949049e-05, "loss": 1.2892, "step": 27699 }, { "epoch": 0.03, "learning_rate": 4.9906061716757653e-05, "loss": 1.0521, "step": 27700 }, { "epoch": 0.03, "learning_rate": 4.990605491377895e-05, "loss": 1.2741, "step": 27701 }, { "epoch": 0.03, "learning_rate": 4.990604811055438e-05, "loss": 1.3494, "step": 27702 }, { "epoch": 0.03, "learning_rate": 4.990604130708396e-05, "loss": 0.778, "step": 27703 }, { "epoch": 0.03, "learning_rate": 4.990603450336767e-05, "loss": 0.8735, "step": 27704 }, { "epoch": 0.03, "learning_rate": 4.990602769940552e-05, "loss": 1.032, "step": 27705 }, { "epoch": 0.03, "learning_rate": 4.990602089519751e-05, "loss": 0.958, "step": 27706 }, { "epoch": 0.03, "learning_rate": 4.9906014090743634e-05, "loss": 0.6649, "step": 27707 }, { "epoch": 0.03, "learning_rate": 4.99060072860439e-05, "loss": 0.8987, "step": 27708 }, { "epoch": 0.03, "learning_rate": 4.99060004810983e-05, "loss": 1.0549, "step": 27709 }, { "epoch": 0.03, "learning_rate": 4.990599367590684e-05, "loss": 1.139, "step": 27710 }, { "epoch": 0.03, "learning_rate": 4.9905986870469524e-05, "loss": 0.7844, "step": 27711 }, { "epoch": 0.03, "learning_rate": 4.9905980064786336e-05, "loss": 1.0607, "step": 27712 }, { "epoch": 0.03, "learning_rate": 4.99059732588573e-05, "loss": 1.0729, "step": 27713 }, { "epoch": 0.03, "learning_rate": 4.990596645268239e-05, "loss": 0.9109, "step": 27714 }, { "epoch": 0.03, "learning_rate": 4.990595964626163e-05, "loss": 0.978, "step": 27715 }, { "epoch": 0.03, "learning_rate": 4.9905952839595003e-05, "loss": 0.958, "step": 27716 }, { "epoch": 0.03, "learning_rate": 4.9905946032682514e-05, "loss": 1.0667, "step": 27717 }, { "epoch": 0.03, "learning_rate": 4.990593922552417e-05, "loss": 1.0727, "step": 27718 }, { "epoch": 0.03, "learning_rate": 4.990593241811996e-05, "loss": 0.8503, "step": 27719 }, { "epoch": 0.03, "learning_rate": 4.990592561046988e-05, "loss": 0.4109, "step": 27720 }, { "epoch": 0.03, "learning_rate": 4.990591880257396e-05, "loss": 0.6537, "step": 27721 }, { "epoch": 0.03, "learning_rate": 4.990591199443217e-05, "loss": 0.8554, "step": 27722 }, { "epoch": 0.03, "learning_rate": 4.990590518604451e-05, "loss": 0.8576, "step": 27723 }, { "epoch": 0.03, "learning_rate": 4.9905898377411e-05, "loss": 1.0251, "step": 27724 }, { "epoch": 0.03, "learning_rate": 4.9905891568531626e-05, "loss": 1.0885, "step": 27725 }, { "epoch": 0.03, "learning_rate": 4.990588475940639e-05, "loss": 0.7234, "step": 27726 }, { "epoch": 0.03, "learning_rate": 4.9905877950035294e-05, "loss": 0.6694, "step": 27727 }, { "epoch": 0.03, "learning_rate": 4.990587114041834e-05, "loss": 0.5955, "step": 27728 }, { "epoch": 0.03, "learning_rate": 4.990586433055553e-05, "loss": 0.852, "step": 27729 }, { "epoch": 0.03, "learning_rate": 4.990585752044685e-05, "loss": 0.9524, "step": 27730 }, { "epoch": 0.03, "learning_rate": 4.990585071009231e-05, "loss": 1.0313, "step": 27731 }, { "epoch": 0.03, "learning_rate": 4.9905843899491924e-05, "loss": 0.9356, "step": 27732 }, { "epoch": 0.03, "learning_rate": 4.9905837088645664e-05, "loss": 1.206, "step": 27733 }, { "epoch": 0.03, "learning_rate": 4.990583027755354e-05, "loss": 0.9267, "step": 27734 }, { "epoch": 0.03, "learning_rate": 4.9905823466215576e-05, "loss": 0.6736, "step": 27735 }, { "epoch": 0.03, "learning_rate": 4.990581665463174e-05, "loss": 0.8159, "step": 27736 }, { "epoch": 0.03, "learning_rate": 4.990580984280204e-05, "loss": 0.8483, "step": 27737 }, { "epoch": 0.03, "learning_rate": 4.990580303072649e-05, "loss": 1.1653, "step": 27738 }, { "epoch": 0.03, "learning_rate": 4.9905796218405075e-05, "loss": 1.0153, "step": 27739 }, { "epoch": 0.03, "learning_rate": 4.99057894058378e-05, "loss": 0.9638, "step": 27740 }, { "epoch": 0.03, "learning_rate": 4.990578259302466e-05, "loss": 1.0033, "step": 27741 }, { "epoch": 0.03, "learning_rate": 4.9905775779965675e-05, "loss": 1.0464, "step": 27742 }, { "epoch": 0.03, "learning_rate": 4.9905768966660825e-05, "loss": 0.8493, "step": 27743 }, { "epoch": 0.03, "learning_rate": 4.990576215311011e-05, "loss": 0.8214, "step": 27744 }, { "epoch": 0.03, "learning_rate": 4.990575533931354e-05, "loss": 1.1777, "step": 27745 }, { "epoch": 0.03, "learning_rate": 4.9905748525271114e-05, "loss": 0.6194, "step": 27746 }, { "epoch": 0.03, "learning_rate": 4.9905741710982825e-05, "loss": 0.9569, "step": 27747 }, { "epoch": 0.03, "learning_rate": 4.990573489644867e-05, "loss": 0.9122, "step": 27748 }, { "epoch": 0.03, "learning_rate": 4.990572808166867e-05, "loss": 1.0316, "step": 27749 }, { "epoch": 0.03, "learning_rate": 4.9905721266642804e-05, "loss": 1.0631, "step": 27750 }, { "epoch": 0.03, "learning_rate": 4.9905714451371075e-05, "loss": 0.8778, "step": 27751 }, { "epoch": 0.03, "learning_rate": 4.9905707635853496e-05, "loss": 0.9357, "step": 27752 }, { "epoch": 0.03, "learning_rate": 4.9905700820090055e-05, "loss": 0.816, "step": 27753 }, { "epoch": 0.03, "learning_rate": 4.9905694004080757e-05, "loss": 0.967, "step": 27754 }, { "epoch": 0.03, "learning_rate": 4.99056871878256e-05, "loss": 0.7972, "step": 27755 }, { "epoch": 0.03, "learning_rate": 4.9905680371324585e-05, "loss": 0.8806, "step": 27756 }, { "epoch": 0.03, "learning_rate": 4.9905673554577704e-05, "loss": 0.5902, "step": 27757 }, { "epoch": 0.03, "learning_rate": 4.990566673758497e-05, "loss": 1.0673, "step": 27758 }, { "epoch": 0.03, "learning_rate": 4.990565992034638e-05, "loss": 1.0391, "step": 27759 }, { "epoch": 0.03, "learning_rate": 4.990565310286193e-05, "loss": 1.0098, "step": 27760 }, { "epoch": 0.03, "learning_rate": 4.9905646285131623e-05, "loss": 0.9799, "step": 27761 }, { "epoch": 0.03, "learning_rate": 4.9905639467155454e-05, "loss": 0.6331, "step": 27762 }, { "epoch": 0.03, "learning_rate": 4.990563264893343e-05, "loss": 1.0631, "step": 27763 }, { "epoch": 0.03, "learning_rate": 4.990562583046555e-05, "loss": 1.0533, "step": 27764 }, { "epoch": 0.03, "learning_rate": 4.990561901175181e-05, "loss": 1.0353, "step": 27765 }, { "epoch": 0.03, "learning_rate": 4.990561219279221e-05, "loss": 0.9607, "step": 27766 }, { "epoch": 0.03, "learning_rate": 4.9905605373586754e-05, "loss": 0.9504, "step": 27767 }, { "epoch": 0.03, "learning_rate": 4.9905598554135446e-05, "loss": 0.7971, "step": 27768 }, { "epoch": 0.03, "learning_rate": 4.9905591734438275e-05, "loss": 1.0303, "step": 27769 }, { "epoch": 0.03, "learning_rate": 4.990558491449525e-05, "loss": 0.9365, "step": 27770 }, { "epoch": 0.03, "learning_rate": 4.990557809430636e-05, "loss": 0.9375, "step": 27771 }, { "epoch": 0.03, "learning_rate": 4.9905571273871616e-05, "loss": 0.5611, "step": 27772 }, { "epoch": 0.03, "learning_rate": 4.990556445319102e-05, "loss": 0.9402, "step": 27773 }, { "epoch": 0.03, "learning_rate": 4.990555763226455e-05, "loss": 1.1674, "step": 27774 }, { "epoch": 0.03, "learning_rate": 4.9905550811092244e-05, "loss": 1.1372, "step": 27775 }, { "epoch": 0.03, "learning_rate": 4.990554398967408e-05, "loss": 0.9021, "step": 27776 }, { "epoch": 0.03, "learning_rate": 4.990553716801004e-05, "loss": 0.9845, "step": 27777 }, { "epoch": 0.03, "learning_rate": 4.990553034610016e-05, "loss": 0.9703, "step": 27778 }, { "epoch": 0.03, "learning_rate": 4.9905523523944415e-05, "loss": 0.9554, "step": 27779 }, { "epoch": 0.03, "learning_rate": 4.9905516701542816e-05, "loss": 1.0286, "step": 27780 }, { "epoch": 0.03, "learning_rate": 4.990550987889536e-05, "loss": 0.9224, "step": 27781 }, { "epoch": 0.03, "learning_rate": 4.9905503056002044e-05, "loss": 0.8858, "step": 27782 }, { "epoch": 0.03, "learning_rate": 4.9905496232862884e-05, "loss": 0.7031, "step": 27783 }, { "epoch": 0.03, "learning_rate": 4.990548940947785e-05, "loss": 0.5892, "step": 27784 }, { "epoch": 0.03, "learning_rate": 4.9905482585846966e-05, "loss": 0.5881, "step": 27785 }, { "epoch": 0.03, "learning_rate": 4.9905475761970236e-05, "loss": 0.9092, "step": 27786 }, { "epoch": 0.03, "learning_rate": 4.990546893784763e-05, "loss": 0.8128, "step": 27787 }, { "epoch": 0.03, "learning_rate": 4.990546211347919e-05, "loss": 0.9252, "step": 27788 }, { "epoch": 0.03, "learning_rate": 4.9905455288864875e-05, "loss": 0.5737, "step": 27789 }, { "epoch": 0.03, "learning_rate": 4.990544846400471e-05, "loss": 0.6435, "step": 27790 }, { "epoch": 0.03, "learning_rate": 4.990544163889869e-05, "loss": 0.8783, "step": 27791 }, { "epoch": 0.03, "learning_rate": 4.990543481354681e-05, "loss": 0.8659, "step": 27792 }, { "epoch": 0.03, "learning_rate": 4.990542798794908e-05, "loss": 0.7595, "step": 27793 }, { "epoch": 0.03, "learning_rate": 4.990542116210549e-05, "loss": 0.4801, "step": 27794 }, { "epoch": 0.03, "learning_rate": 4.990541433601605e-05, "loss": 1.1763, "step": 27795 }, { "epoch": 0.03, "learning_rate": 4.990540750968075e-05, "loss": 0.7236, "step": 27796 }, { "epoch": 0.03, "learning_rate": 4.990540068309959e-05, "loss": 1.0081, "step": 27797 }, { "epoch": 0.03, "learning_rate": 4.9905393856272585e-05, "loss": 0.8558, "step": 27798 }, { "epoch": 0.03, "learning_rate": 4.9905387029199715e-05, "loss": 0.9825, "step": 27799 }, { "epoch": 0.03, "learning_rate": 4.990538020188099e-05, "loss": 1.1079, "step": 27800 }, { "epoch": 0.03, "learning_rate": 4.9905373374316414e-05, "loss": 0.8487, "step": 27801 }, { "epoch": 0.03, "learning_rate": 4.990536654650598e-05, "loss": 1.0253, "step": 27802 }, { "epoch": 0.03, "learning_rate": 4.990535971844969e-05, "loss": 1.0818, "step": 27803 }, { "epoch": 0.03, "learning_rate": 4.990535289014755e-05, "loss": 0.8961, "step": 27804 }, { "epoch": 0.03, "learning_rate": 4.9905346061599553e-05, "loss": 1.0148, "step": 27805 }, { "epoch": 0.03, "learning_rate": 4.99053392328057e-05, "loss": 1.0145, "step": 27806 }, { "epoch": 0.03, "learning_rate": 4.9905332403765995e-05, "loss": 1.0668, "step": 27807 }, { "epoch": 0.03, "learning_rate": 4.990532557448043e-05, "loss": 1.0569, "step": 27808 }, { "epoch": 0.03, "learning_rate": 4.990531874494901e-05, "loss": 0.9331, "step": 27809 }, { "epoch": 0.03, "learning_rate": 4.9905311915171735e-05, "loss": 0.7204, "step": 27810 }, { "epoch": 0.03, "learning_rate": 4.9905305085148616e-05, "loss": 0.8553, "step": 27811 }, { "epoch": 0.03, "learning_rate": 4.9905298254879626e-05, "loss": 1.2666, "step": 27812 }, { "epoch": 0.03, "learning_rate": 4.990529142436479e-05, "loss": 1.0179, "step": 27813 }, { "epoch": 0.03, "learning_rate": 4.99052845936041e-05, "loss": 1.0008, "step": 27814 }, { "epoch": 0.03, "learning_rate": 4.9905277762597554e-05, "loss": 1.0917, "step": 27815 }, { "epoch": 0.03, "learning_rate": 4.990527093134515e-05, "loss": 1.0309, "step": 27816 }, { "epoch": 0.03, "learning_rate": 4.9905264099846895e-05, "loss": 1.1319, "step": 27817 }, { "epoch": 0.03, "learning_rate": 4.990525726810279e-05, "loss": 0.9352, "step": 27818 }, { "epoch": 0.03, "learning_rate": 4.9905250436112824e-05, "loss": 0.8021, "step": 27819 }, { "epoch": 0.03, "learning_rate": 4.9905243603877004e-05, "loss": 1.0507, "step": 27820 }, { "epoch": 0.03, "learning_rate": 4.9905236771395334e-05, "loss": 0.9361, "step": 27821 }, { "epoch": 0.03, "learning_rate": 4.990522993866781e-05, "loss": 0.9819, "step": 27822 }, { "epoch": 0.03, "learning_rate": 4.990522310569443e-05, "loss": 1.1695, "step": 27823 }, { "epoch": 0.03, "learning_rate": 4.99052162724752e-05, "loss": 1.2755, "step": 27824 }, { "epoch": 0.03, "learning_rate": 4.990520943901011e-05, "loss": 1.1384, "step": 27825 }, { "epoch": 0.03, "learning_rate": 4.9905202605299166e-05, "loss": 1.0256, "step": 27826 }, { "epoch": 0.03, "learning_rate": 4.990519577134237e-05, "loss": 0.8643, "step": 27827 }, { "epoch": 0.03, "learning_rate": 4.990518893713972e-05, "loss": 0.5411, "step": 27828 }, { "epoch": 0.03, "learning_rate": 4.990518210269122e-05, "loss": 0.8263, "step": 27829 }, { "epoch": 0.03, "learning_rate": 4.990517526799687e-05, "loss": 0.758, "step": 27830 }, { "epoch": 0.03, "learning_rate": 4.990516843305666e-05, "loss": 1.0492, "step": 27831 }, { "epoch": 0.03, "learning_rate": 4.99051615978706e-05, "loss": 1.7941, "step": 27832 }, { "epoch": 0.03, "learning_rate": 4.9905154762438674e-05, "loss": 1.3182, "step": 27833 }, { "epoch": 0.03, "learning_rate": 4.990514792676091e-05, "loss": 1.244, "step": 27834 }, { "epoch": 0.03, "learning_rate": 4.9905141090837294e-05, "loss": 1.1169, "step": 27835 }, { "epoch": 0.03, "learning_rate": 4.990513425466782e-05, "loss": 1.0359, "step": 27836 }, { "epoch": 0.03, "learning_rate": 4.990512741825249e-05, "loss": 1.0787, "step": 27837 }, { "epoch": 0.03, "learning_rate": 4.9905120581591306e-05, "loss": 1.07, "step": 27838 }, { "epoch": 0.03, "learning_rate": 4.9905113744684276e-05, "loss": 0.9152, "step": 27839 }, { "epoch": 0.03, "learning_rate": 4.990510690753138e-05, "loss": 0.6232, "step": 27840 }, { "epoch": 0.03, "learning_rate": 4.990510007013265e-05, "loss": 0.8866, "step": 27841 }, { "epoch": 0.03, "learning_rate": 4.990509323248806e-05, "loss": 1.1758, "step": 27842 }, { "epoch": 0.03, "learning_rate": 4.990508639459761e-05, "loss": 1.0268, "step": 27843 }, { "epoch": 0.03, "learning_rate": 4.9905079556461324e-05, "loss": 1.01, "step": 27844 }, { "epoch": 0.03, "learning_rate": 4.990507271807917e-05, "loss": 0.8055, "step": 27845 }, { "epoch": 0.03, "learning_rate": 4.990506587945117e-05, "loss": 0.9076, "step": 27846 }, { "epoch": 0.03, "learning_rate": 4.9905059040577315e-05, "loss": 1.0533, "step": 27847 }, { "epoch": 0.03, "learning_rate": 4.990505220145761e-05, "loss": 0.8751, "step": 27848 }, { "epoch": 0.03, "learning_rate": 4.990504536209205e-05, "loss": 0.7527, "step": 27849 }, { "epoch": 0.03, "learning_rate": 4.990503852248064e-05, "loss": 0.9183, "step": 27850 }, { "epoch": 0.03, "learning_rate": 4.990503168262337e-05, "loss": 0.9372, "step": 27851 }, { "epoch": 0.03, "learning_rate": 4.9905024842520264e-05, "loss": 1.072, "step": 27852 }, { "epoch": 0.03, "learning_rate": 4.990501800217129e-05, "loss": 0.9203, "step": 27853 }, { "epoch": 0.03, "learning_rate": 4.9905011161576475e-05, "loss": 0.8439, "step": 27854 }, { "epoch": 0.03, "learning_rate": 4.9905004320735803e-05, "loss": 1.0147, "step": 27855 }, { "epoch": 0.03, "learning_rate": 4.990499747964928e-05, "loss": 0.9895, "step": 27856 }, { "epoch": 0.03, "learning_rate": 4.990499063831691e-05, "loss": 1.1987, "step": 27857 }, { "epoch": 0.03, "learning_rate": 4.990498379673869e-05, "loss": 0.9581, "step": 27858 }, { "epoch": 0.03, "learning_rate": 4.990497695491461e-05, "loss": 1.145, "step": 27859 }, { "epoch": 0.03, "learning_rate": 4.9904970112844686e-05, "loss": 1.3166, "step": 27860 }, { "epoch": 0.03, "learning_rate": 4.99049632705289e-05, "loss": 1.2875, "step": 27861 }, { "epoch": 0.03, "learning_rate": 4.9904956427967276e-05, "loss": 1.0196, "step": 27862 }, { "epoch": 0.03, "learning_rate": 4.9904949585159794e-05, "loss": 1.1823, "step": 27863 }, { "epoch": 0.03, "learning_rate": 4.990494274210646e-05, "loss": 0.9371, "step": 27864 }, { "epoch": 0.03, "learning_rate": 4.990493589880727e-05, "loss": 0.9609, "step": 27865 }, { "epoch": 0.03, "learning_rate": 4.990492905526224e-05, "loss": 1.189, "step": 27866 }, { "epoch": 0.03, "learning_rate": 4.9904922211471354e-05, "loss": 0.9733, "step": 27867 }, { "epoch": 0.03, "learning_rate": 4.990491536743462e-05, "loss": 0.9429, "step": 27868 }, { "epoch": 0.03, "learning_rate": 4.990490852315203e-05, "loss": 1.1624, "step": 27869 }, { "epoch": 0.03, "learning_rate": 4.990490167862359e-05, "loss": 1.4252, "step": 27870 }, { "epoch": 0.03, "learning_rate": 4.99048948338493e-05, "loss": 0.9018, "step": 27871 }, { "epoch": 0.03, "learning_rate": 4.9904887988829165e-05, "loss": 0.8046, "step": 27872 }, { "epoch": 0.03, "learning_rate": 4.990488114356318e-05, "loss": 1.3373, "step": 27873 }, { "epoch": 0.03, "learning_rate": 4.990487429805133e-05, "loss": 0.9253, "step": 27874 }, { "epoch": 0.03, "learning_rate": 4.990486745229364e-05, "loss": 0.5874, "step": 27875 }, { "epoch": 0.03, "learning_rate": 4.99048606062901e-05, "loss": 1.0182, "step": 27876 }, { "epoch": 0.03, "learning_rate": 4.990485376004071e-05, "loss": 0.9424, "step": 27877 }, { "epoch": 0.03, "learning_rate": 4.990484691354547e-05, "loss": 1.139, "step": 27878 }, { "epoch": 0.03, "learning_rate": 4.990484006680437e-05, "loss": 0.9923, "step": 27879 }, { "epoch": 0.03, "learning_rate": 4.990483321981743e-05, "loss": 1.2163, "step": 27880 }, { "epoch": 0.03, "learning_rate": 4.9904826372584635e-05, "loss": 1.0284, "step": 27881 }, { "epoch": 0.03, "learning_rate": 4.9904819525105996e-05, "loss": 1.0258, "step": 27882 }, { "epoch": 0.03, "learning_rate": 4.990481267738151e-05, "loss": 0.7791, "step": 27883 }, { "epoch": 0.03, "learning_rate": 4.990480582941116e-05, "loss": 0.9971, "step": 27884 }, { "epoch": 0.03, "learning_rate": 4.990479898119497e-05, "loss": 1.0735, "step": 27885 }, { "epoch": 0.03, "learning_rate": 4.9904792132732924e-05, "loss": 0.9736, "step": 27886 }, { "epoch": 0.03, "learning_rate": 4.990478528402504e-05, "loss": 1.0789, "step": 27887 }, { "epoch": 0.03, "learning_rate": 4.99047784350713e-05, "loss": 0.9628, "step": 27888 }, { "epoch": 0.03, "learning_rate": 4.990477158587171e-05, "loss": 0.6533, "step": 27889 }, { "epoch": 0.03, "learning_rate": 4.9904764736426265e-05, "loss": 0.7606, "step": 27890 }, { "epoch": 0.03, "learning_rate": 4.990475788673497e-05, "loss": 0.5276, "step": 27891 }, { "epoch": 0.03, "learning_rate": 4.990475103679784e-05, "loss": 0.8716, "step": 27892 }, { "epoch": 0.03, "learning_rate": 4.990474418661485e-05, "loss": 0.9351, "step": 27893 }, { "epoch": 0.03, "learning_rate": 4.990473733618601e-05, "loss": 0.8573, "step": 27894 }, { "epoch": 0.03, "learning_rate": 4.990473048551133e-05, "loss": 0.8419, "step": 27895 }, { "epoch": 0.03, "learning_rate": 4.99047236345908e-05, "loss": 0.4738, "step": 27896 }, { "epoch": 0.03, "learning_rate": 4.990471678342441e-05, "loss": 1.0962, "step": 27897 }, { "epoch": 0.03, "learning_rate": 4.9904709932012186e-05, "loss": 1.1059, "step": 27898 }, { "epoch": 0.03, "learning_rate": 4.99047030803541e-05, "loss": 1.1053, "step": 27899 }, { "epoch": 0.03, "learning_rate": 4.990469622845017e-05, "loss": 1.2107, "step": 27900 }, { "epoch": 0.03, "learning_rate": 4.99046893763004e-05, "loss": 0.9637, "step": 27901 }, { "epoch": 0.03, "learning_rate": 4.9904682523904766e-05, "loss": 0.5717, "step": 27902 }, { "epoch": 0.03, "learning_rate": 4.990467567126329e-05, "loss": 0.6492, "step": 27903 }, { "epoch": 0.03, "learning_rate": 4.990466881837597e-05, "loss": 1.0332, "step": 27904 }, { "epoch": 0.03, "learning_rate": 4.9904661965242795e-05, "loss": 1.0358, "step": 27905 }, { "epoch": 0.03, "learning_rate": 4.990465511186377e-05, "loss": 1.1549, "step": 27906 }, { "epoch": 0.03, "learning_rate": 4.99046482582389e-05, "loss": 1.0167, "step": 27907 }, { "epoch": 0.03, "learning_rate": 4.9904641404368185e-05, "loss": 1.3105, "step": 27908 }, { "epoch": 0.03, "learning_rate": 4.990463455025162e-05, "loss": 1.0382, "step": 27909 }, { "epoch": 0.03, "learning_rate": 4.9904627695889205e-05, "loss": 1.1233, "step": 27910 }, { "epoch": 0.03, "learning_rate": 4.990462084128095e-05, "loss": 0.7033, "step": 27911 }, { "epoch": 0.03, "learning_rate": 4.9904613986426834e-05, "loss": 0.9632, "step": 27912 }, { "epoch": 0.03, "learning_rate": 4.990460713132688e-05, "loss": 0.8804, "step": 27913 }, { "epoch": 0.03, "learning_rate": 4.990460027598107e-05, "loss": 0.9075, "step": 27914 }, { "epoch": 0.03, "learning_rate": 4.990459342038942e-05, "loss": 0.6978, "step": 27915 }, { "epoch": 0.03, "learning_rate": 4.9904586564551916e-05, "loss": 0.5195, "step": 27916 }, { "epoch": 0.03, "learning_rate": 4.990457970846857e-05, "loss": 0.9922, "step": 27917 }, { "epoch": 0.03, "learning_rate": 4.990457285213938e-05, "loss": 0.8882, "step": 27918 }, { "epoch": 0.03, "learning_rate": 4.9904565995564336e-05, "loss": 0.9277, "step": 27919 }, { "epoch": 0.03, "learning_rate": 4.9904559138743446e-05, "loss": 1.0428, "step": 27920 }, { "epoch": 0.03, "learning_rate": 4.9904552281676706e-05, "loss": 1.0743, "step": 27921 }, { "epoch": 0.03, "learning_rate": 4.9904545424364116e-05, "loss": 1.06, "step": 27922 }, { "epoch": 0.03, "learning_rate": 4.990453856680568e-05, "loss": 1.2203, "step": 27923 }, { "epoch": 0.03, "learning_rate": 4.9904531709001415e-05, "loss": 1.0635, "step": 27924 }, { "epoch": 0.03, "learning_rate": 4.990452485095128e-05, "loss": 0.8842, "step": 27925 }, { "epoch": 0.03, "learning_rate": 4.990451799265531e-05, "loss": 1.009, "step": 27926 }, { "epoch": 0.03, "learning_rate": 4.9904511134113484e-05, "loss": 1.065, "step": 27927 }, { "epoch": 0.03, "learning_rate": 4.9904504275325824e-05, "loss": 0.8137, "step": 27928 }, { "epoch": 0.03, "learning_rate": 4.990449741629231e-05, "loss": 0.9956, "step": 27929 }, { "epoch": 0.03, "learning_rate": 4.990449055701295e-05, "loss": 1.0317, "step": 27930 }, { "epoch": 0.03, "learning_rate": 4.990448369748774e-05, "loss": 0.9656, "step": 27931 }, { "epoch": 0.03, "learning_rate": 4.990447683771669e-05, "loss": 0.9636, "step": 27932 }, { "epoch": 0.03, "learning_rate": 4.9904469977699785e-05, "loss": 1.0347, "step": 27933 }, { "epoch": 0.03, "learning_rate": 4.990446311743704e-05, "loss": 0.9764, "step": 27934 }, { "epoch": 0.03, "learning_rate": 4.9904456256928447e-05, "loss": 1.0576, "step": 27935 }, { "epoch": 0.03, "learning_rate": 4.990444939617401e-05, "loss": 0.9052, "step": 27936 }, { "epoch": 0.03, "learning_rate": 4.990444253517372e-05, "loss": 1.1236, "step": 27937 }, { "epoch": 0.03, "learning_rate": 4.9904435673927594e-05, "loss": 0.8083, "step": 27938 }, { "epoch": 0.03, "learning_rate": 4.990442881243561e-05, "loss": 1.1179, "step": 27939 }, { "epoch": 0.03, "learning_rate": 4.9904421950697786e-05, "loss": 1.2141, "step": 27940 }, { "epoch": 0.03, "learning_rate": 4.9904415088714115e-05, "loss": 1.0086, "step": 27941 }, { "epoch": 0.03, "learning_rate": 4.99044082264846e-05, "loss": 1.0237, "step": 27942 }, { "epoch": 0.03, "learning_rate": 4.9904401364009243e-05, "loss": 1.1294, "step": 27943 }, { "epoch": 0.03, "learning_rate": 4.990439450128803e-05, "loss": 0.9394, "step": 27944 }, { "epoch": 0.03, "learning_rate": 4.990438763832098e-05, "loss": 0.8005, "step": 27945 }, { "epoch": 0.03, "learning_rate": 4.990438077510808e-05, "loss": 1.0421, "step": 27946 }, { "epoch": 0.03, "learning_rate": 4.990437391164934e-05, "loss": 1.0653, "step": 27947 }, { "epoch": 0.03, "learning_rate": 4.990436704794475e-05, "loss": 0.9845, "step": 27948 }, { "epoch": 0.03, "learning_rate": 4.990436018399431e-05, "loss": 0.8786, "step": 27949 }, { "epoch": 0.03, "learning_rate": 4.9904353319798036e-05, "loss": 0.9444, "step": 27950 }, { "epoch": 0.03, "learning_rate": 4.990434645535591e-05, "loss": 0.8474, "step": 27951 }, { "epoch": 0.03, "learning_rate": 4.990433959066794e-05, "loss": 0.8079, "step": 27952 }, { "epoch": 0.03, "learning_rate": 4.9904332725734126e-05, "loss": 0.8527, "step": 27953 }, { "epoch": 0.03, "learning_rate": 4.9904325860554464e-05, "loss": 1.0714, "step": 27954 }, { "epoch": 0.03, "learning_rate": 4.990431899512896e-05, "loss": 0.9427, "step": 27955 }, { "epoch": 0.03, "learning_rate": 4.990431212945761e-05, "loss": 1.0613, "step": 27956 }, { "epoch": 0.03, "learning_rate": 4.9904305263540406e-05, "loss": 1.2071, "step": 27957 }, { "epoch": 0.03, "learning_rate": 4.9904298397377366e-05, "loss": 0.8898, "step": 27958 }, { "epoch": 0.03, "learning_rate": 4.990429153096848e-05, "loss": 1.0221, "step": 27959 }, { "epoch": 0.03, "learning_rate": 4.990428466431375e-05, "loss": 0.9921, "step": 27960 }, { "epoch": 0.03, "learning_rate": 4.990427779741318e-05, "loss": 0.9794, "step": 27961 }, { "epoch": 0.03, "learning_rate": 4.990427093026676e-05, "loss": 0.8498, "step": 27962 }, { "epoch": 0.03, "learning_rate": 4.9904264062874494e-05, "loss": 0.2687, "step": 27963 }, { "epoch": 0.03, "learning_rate": 4.990425719523639e-05, "loss": 0.6226, "step": 27964 }, { "epoch": 0.03, "learning_rate": 4.990425032735243e-05, "loss": 0.7986, "step": 27965 }, { "epoch": 0.03, "learning_rate": 4.990424345922264e-05, "loss": 0.806, "step": 27966 }, { "epoch": 0.03, "learning_rate": 4.9904236590847e-05, "loss": 1.2528, "step": 27967 }, { "epoch": 0.03, "learning_rate": 4.990422972222552e-05, "loss": 1.0032, "step": 27968 }, { "epoch": 0.03, "learning_rate": 4.9904222853358185e-05, "loss": 0.9903, "step": 27969 }, { "epoch": 0.03, "learning_rate": 4.9904215984245004e-05, "loss": 1.0997, "step": 27970 }, { "epoch": 0.03, "learning_rate": 4.990420911488599e-05, "loss": 0.7579, "step": 27971 }, { "epoch": 0.03, "learning_rate": 4.9904202245281125e-05, "loss": 0.9817, "step": 27972 }, { "epoch": 0.03, "learning_rate": 4.990419537543043e-05, "loss": 1.1755, "step": 27973 }, { "epoch": 0.03, "learning_rate": 4.990418850533388e-05, "loss": 0.7276, "step": 27974 }, { "epoch": 0.03, "learning_rate": 4.990418163499149e-05, "loss": 0.9457, "step": 27975 }, { "epoch": 0.03, "learning_rate": 4.9904174764403255e-05, "loss": 0.6992, "step": 27976 }, { "epoch": 0.03, "learning_rate": 4.990416789356917e-05, "loss": 0.7609, "step": 27977 }, { "epoch": 0.03, "learning_rate": 4.990416102248925e-05, "loss": 1.111, "step": 27978 }, { "epoch": 0.03, "learning_rate": 4.990415415116348e-05, "loss": 1.0256, "step": 27979 }, { "epoch": 0.03, "learning_rate": 4.990414727959187e-05, "loss": 0.8097, "step": 27980 }, { "epoch": 0.03, "learning_rate": 4.9904140407774425e-05, "loss": 1.0942, "step": 27981 }, { "epoch": 0.03, "learning_rate": 4.990413353571113e-05, "loss": 0.7206, "step": 27982 }, { "epoch": 0.03, "learning_rate": 4.9904126663401984e-05, "loss": 0.8163, "step": 27983 }, { "epoch": 0.03, "learning_rate": 4.9904119790847e-05, "loss": 1.203, "step": 27984 }, { "epoch": 0.03, "learning_rate": 4.990411291804618e-05, "loss": 0.7933, "step": 27985 }, { "epoch": 0.03, "learning_rate": 4.990410604499951e-05, "loss": 1.026, "step": 27986 }, { "epoch": 0.03, "learning_rate": 4.9904099171707e-05, "loss": 1.2583, "step": 27987 }, { "epoch": 0.03, "learning_rate": 4.990409229816865e-05, "loss": 1.0807, "step": 27988 }, { "epoch": 0.03, "learning_rate": 4.990408542438445e-05, "loss": 0.9935, "step": 27989 }, { "epoch": 0.03, "learning_rate": 4.9904078550354415e-05, "loss": 0.8608, "step": 27990 }, { "epoch": 0.03, "learning_rate": 4.990407167607853e-05, "loss": 1.0443, "step": 27991 }, { "epoch": 0.03, "learning_rate": 4.990406480155681e-05, "loss": 1.0556, "step": 27992 }, { "epoch": 0.03, "learning_rate": 4.990405792678924e-05, "loss": 0.7044, "step": 27993 }, { "epoch": 0.03, "learning_rate": 4.990405105177583e-05, "loss": 0.8708, "step": 27994 }, { "epoch": 0.03, "learning_rate": 4.9904044176516586e-05, "loss": 1.0584, "step": 27995 }, { "epoch": 0.03, "learning_rate": 4.990403730101149e-05, "loss": 0.8515, "step": 27996 }, { "epoch": 0.03, "learning_rate": 4.990403042526056e-05, "loss": 1.2923, "step": 27997 }, { "epoch": 0.03, "learning_rate": 4.990402354926378e-05, "loss": 1.0697, "step": 27998 }, { "epoch": 0.03, "learning_rate": 4.990401667302116e-05, "loss": 1.2155, "step": 27999 }, { "epoch": 0.03, "learning_rate": 4.9904009796532694e-05, "loss": 1.2817, "step": 28000 }, { "epoch": 0.03, "eval_loss": 1.061301350593567, "eval_runtime": 129.3835, "eval_samples_per_second": 10.705, "eval_steps_per_second": 5.356, "step": 28000 }, { "epoch": 0.03, "learning_rate": 4.99040029197984e-05, "loss": 0.9983, "step": 28001 }, { "epoch": 0.03, "learning_rate": 4.9903996042818245e-05, "loss": 1.0991, "step": 28002 }, { "epoch": 0.03, "learning_rate": 4.990398916559226e-05, "loss": 1.1118, "step": 28003 }, { "epoch": 0.03, "learning_rate": 4.990398228812043e-05, "loss": 1.0434, "step": 28004 }, { "epoch": 0.03, "learning_rate": 4.9903975410402764e-05, "loss": 1.0248, "step": 28005 }, { "epoch": 0.03, "learning_rate": 4.990396853243925e-05, "loss": 1.0889, "step": 28006 }, { "epoch": 0.03, "learning_rate": 4.99039616542299e-05, "loss": 1.048, "step": 28007 }, { "epoch": 0.03, "learning_rate": 4.990395477577471e-05, "loss": 0.8462, "step": 28008 }, { "epoch": 0.03, "learning_rate": 4.9903947897073665e-05, "loss": 1.0136, "step": 28009 }, { "epoch": 0.03, "learning_rate": 4.990394101812679e-05, "loss": 0.7795, "step": 28010 }, { "epoch": 0.03, "learning_rate": 4.990393413893407e-05, "loss": 0.6053, "step": 28011 }, { "epoch": 0.03, "learning_rate": 4.990392725949551e-05, "loss": 1.0684, "step": 28012 }, { "epoch": 0.03, "learning_rate": 4.9903920379811115e-05, "loss": 0.8894, "step": 28013 }, { "epoch": 0.03, "learning_rate": 4.9903913499880875e-05, "loss": 1.1518, "step": 28014 }, { "epoch": 0.03, "learning_rate": 4.9903906619704786e-05, "loss": 0.9953, "step": 28015 }, { "epoch": 0.03, "learning_rate": 4.990389973928287e-05, "loss": 0.831, "step": 28016 }, { "epoch": 0.03, "learning_rate": 4.99038928586151e-05, "loss": 0.8108, "step": 28017 }, { "epoch": 0.03, "learning_rate": 4.9903885977701495e-05, "loss": 0.9599, "step": 28018 }, { "epoch": 0.03, "learning_rate": 4.990387909654205e-05, "loss": 0.8391, "step": 28019 }, { "epoch": 0.03, "learning_rate": 4.9903872215136765e-05, "loss": 0.8052, "step": 28020 }, { "epoch": 0.03, "learning_rate": 4.990386533348564e-05, "loss": 0.716, "step": 28021 }, { "epoch": 0.03, "learning_rate": 4.990385845158867e-05, "loss": 1.1299, "step": 28022 }, { "epoch": 0.03, "learning_rate": 4.990385156944586e-05, "loss": 1.4516, "step": 28023 }, { "epoch": 0.03, "learning_rate": 4.990384468705721e-05, "loss": 1.4178, "step": 28024 }, { "epoch": 0.03, "learning_rate": 4.990383780442273e-05, "loss": 1.4028, "step": 28025 }, { "epoch": 0.03, "learning_rate": 4.9903830921542394e-05, "loss": 1.387, "step": 28026 }, { "epoch": 0.03, "learning_rate": 4.990382403841623e-05, "loss": 1.3419, "step": 28027 }, { "epoch": 0.03, "learning_rate": 4.990381715504422e-05, "loss": 1.3017, "step": 28028 }, { "epoch": 0.03, "learning_rate": 4.990381027142637e-05, "loss": 1.3854, "step": 28029 }, { "epoch": 0.03, "learning_rate": 4.9903803387562686e-05, "loss": 1.3358, "step": 28030 }, { "epoch": 0.03, "learning_rate": 4.990379650345315e-05, "loss": 1.398, "step": 28031 }, { "epoch": 0.03, "learning_rate": 4.990378961909778e-05, "loss": 1.3703, "step": 28032 }, { "epoch": 0.03, "learning_rate": 4.990378273449658e-05, "loss": 1.4383, "step": 28033 }, { "epoch": 0.03, "learning_rate": 4.9903775849649533e-05, "loss": 1.4452, "step": 28034 }, { "epoch": 0.03, "learning_rate": 4.990376896455664e-05, "loss": 1.5407, "step": 28035 }, { "epoch": 0.03, "learning_rate": 4.990376207921792e-05, "loss": 1.4526, "step": 28036 }, { "epoch": 0.03, "learning_rate": 4.990375519363335e-05, "loss": 1.3148, "step": 28037 }, { "epoch": 0.03, "learning_rate": 4.990374830780294e-05, "loss": 1.3843, "step": 28038 }, { "epoch": 0.03, "learning_rate": 4.99037414217267e-05, "loss": 1.3411, "step": 28039 }, { "epoch": 0.03, "learning_rate": 4.9903734535404614e-05, "loss": 1.3977, "step": 28040 }, { "epoch": 0.03, "learning_rate": 4.990372764883669e-05, "loss": 1.4128, "step": 28041 }, { "epoch": 0.03, "learning_rate": 4.9903720762022934e-05, "loss": 1.395, "step": 28042 }, { "epoch": 0.03, "learning_rate": 4.9903713874963334e-05, "loss": 1.3484, "step": 28043 }, { "epoch": 0.03, "learning_rate": 4.99037069876579e-05, "loss": 1.3158, "step": 28044 }, { "epoch": 0.03, "learning_rate": 4.990370010010661e-05, "loss": 1.1074, "step": 28045 }, { "epoch": 0.03, "learning_rate": 4.9903693212309495e-05, "loss": 1.2379, "step": 28046 }, { "epoch": 0.03, "learning_rate": 4.990368632426654e-05, "loss": 0.7073, "step": 28047 }, { "epoch": 0.03, "learning_rate": 4.990367943597774e-05, "loss": 1.0106, "step": 28048 }, { "epoch": 0.03, "learning_rate": 4.9903672547443104e-05, "loss": 1.0081, "step": 28049 }, { "epoch": 0.03, "learning_rate": 4.990366565866264e-05, "loss": 1.245, "step": 28050 }, { "epoch": 0.03, "learning_rate": 4.990365876963633e-05, "loss": 1.0308, "step": 28051 }, { "epoch": 0.03, "learning_rate": 4.9903651880364175e-05, "loss": 1.2028, "step": 28052 }, { "epoch": 0.03, "learning_rate": 4.990364499084619e-05, "loss": 0.8116, "step": 28053 }, { "epoch": 0.03, "learning_rate": 4.990363810108236e-05, "loss": 1.0331, "step": 28054 }, { "epoch": 0.03, "learning_rate": 4.99036312110727e-05, "loss": 0.9464, "step": 28055 }, { "epoch": 0.03, "learning_rate": 4.990362432081719e-05, "loss": 1.2963, "step": 28056 }, { "epoch": 0.03, "learning_rate": 4.990361743031585e-05, "loss": 0.7556, "step": 28057 }, { "epoch": 0.03, "learning_rate": 4.9903610539568676e-05, "loss": 1.1131, "step": 28058 }, { "epoch": 0.03, "learning_rate": 4.990360364857566e-05, "loss": 0.8096, "step": 28059 }, { "epoch": 0.03, "learning_rate": 4.9903596757336803e-05, "loss": 1.0185, "step": 28060 }, { "epoch": 0.03, "learning_rate": 4.9903589865852106e-05, "loss": 0.8822, "step": 28061 }, { "epoch": 0.03, "learning_rate": 4.990358297412158e-05, "loss": 0.8732, "step": 28062 }, { "epoch": 0.03, "learning_rate": 4.990357608214521e-05, "loss": 1.0675, "step": 28063 }, { "epoch": 0.03, "learning_rate": 4.9903569189923004e-05, "loss": 1.0767, "step": 28064 }, { "epoch": 0.03, "learning_rate": 4.990356229745496e-05, "loss": 0.9652, "step": 28065 }, { "epoch": 0.03, "learning_rate": 4.9903555404741085e-05, "loss": 0.8328, "step": 28066 }, { "epoch": 0.03, "learning_rate": 4.9903548511781365e-05, "loss": 1.3185, "step": 28067 }, { "epoch": 0.03, "learning_rate": 4.990354161857581e-05, "loss": 1.6722, "step": 28068 }, { "epoch": 0.03, "learning_rate": 4.9903534725124415e-05, "loss": 1.5298, "step": 28069 }, { "epoch": 0.03, "learning_rate": 4.9903527831427187e-05, "loss": 1.1478, "step": 28070 }, { "epoch": 0.03, "learning_rate": 4.990352093748412e-05, "loss": 1.0807, "step": 28071 }, { "epoch": 0.03, "learning_rate": 4.990351404329522e-05, "loss": 1.1685, "step": 28072 }, { "epoch": 0.03, "learning_rate": 4.990350714886047e-05, "loss": 0.2383, "step": 28073 }, { "epoch": 0.03, "learning_rate": 4.9903500254179904e-05, "loss": 0.1214, "step": 28074 }, { "epoch": 0.03, "learning_rate": 4.990349335925348e-05, "loss": 0.0569, "step": 28075 }, { "epoch": 0.03, "learning_rate": 4.9903486464081236e-05, "loss": 0.1609, "step": 28076 }, { "epoch": 0.03, "learning_rate": 4.990347956866315e-05, "loss": 0.0924, "step": 28077 }, { "epoch": 0.03, "learning_rate": 4.990347267299922e-05, "loss": 0.2206, "step": 28078 }, { "epoch": 0.03, "learning_rate": 4.990346577708946e-05, "loss": 0.0928, "step": 28079 }, { "epoch": 0.03, "learning_rate": 4.990345888093386e-05, "loss": 0.0923, "step": 28080 }, { "epoch": 0.03, "learning_rate": 4.990345198453242e-05, "loss": 0.128, "step": 28081 }, { "epoch": 0.03, "learning_rate": 4.990344508788516e-05, "loss": 0.0848, "step": 28082 }, { "epoch": 0.03, "learning_rate": 4.990343819099205e-05, "loss": 0.1401, "step": 28083 }, { "epoch": 0.03, "learning_rate": 4.990343129385311e-05, "loss": 0.0911, "step": 28084 }, { "epoch": 0.03, "learning_rate": 4.9903424396468326e-05, "loss": 0.0883, "step": 28085 }, { "epoch": 0.03, "learning_rate": 4.9903417498837713e-05, "loss": 0.0866, "step": 28086 }, { "epoch": 0.03, "learning_rate": 4.990341060096126e-05, "loss": 0.0672, "step": 28087 }, { "epoch": 0.03, "learning_rate": 4.9903403702838967e-05, "loss": 0.0651, "step": 28088 }, { "epoch": 0.03, "learning_rate": 4.9903396804470846e-05, "loss": 0.1197, "step": 28089 }, { "epoch": 0.03, "learning_rate": 4.990338990585689e-05, "loss": 0.0862, "step": 28090 }, { "epoch": 0.03, "learning_rate": 4.9903383006997096e-05, "loss": 0.0857, "step": 28091 }, { "epoch": 0.03, "learning_rate": 4.990337610789147e-05, "loss": 0.1085, "step": 28092 }, { "epoch": 0.03, "learning_rate": 4.9903369208539995e-05, "loss": 0.0891, "step": 28093 }, { "epoch": 0.03, "learning_rate": 4.9903362308942694e-05, "loss": 0.1708, "step": 28094 }, { "epoch": 0.03, "learning_rate": 4.990335540909956e-05, "loss": 0.0792, "step": 28095 }, { "epoch": 0.03, "learning_rate": 4.9903348509010584e-05, "loss": 0.1123, "step": 28096 }, { "epoch": 0.03, "learning_rate": 4.990334160867578e-05, "loss": 0.0985, "step": 28097 }, { "epoch": 0.03, "learning_rate": 4.990333470809513e-05, "loss": 0.0868, "step": 28098 }, { "epoch": 0.03, "learning_rate": 4.9903327807268654e-05, "loss": 0.1019, "step": 28099 }, { "epoch": 0.03, "learning_rate": 4.990332090619634e-05, "loss": 0.1015, "step": 28100 }, { "epoch": 0.03, "learning_rate": 4.990331400487819e-05, "loss": 0.4918, "step": 28101 }, { "epoch": 0.03, "learning_rate": 4.990330710331421e-05, "loss": 0.1164, "step": 28102 }, { "epoch": 0.03, "learning_rate": 4.990330020150439e-05, "loss": 0.3043, "step": 28103 }, { "epoch": 0.03, "learning_rate": 4.9903293299448733e-05, "loss": 0.5307, "step": 28104 }, { "epoch": 0.03, "learning_rate": 4.990328639714724e-05, "loss": 0.2775, "step": 28105 }, { "epoch": 0.03, "learning_rate": 4.990327949459992e-05, "loss": 0.5126, "step": 28106 }, { "epoch": 0.03, "learning_rate": 4.990327259180676e-05, "loss": 0.2787, "step": 28107 }, { "epoch": 0.03, "learning_rate": 4.9903265688767766e-05, "loss": 0.0764, "step": 28108 }, { "epoch": 0.03, "learning_rate": 4.9903258785482945e-05, "loss": 0.1381, "step": 28109 }, { "epoch": 0.03, "learning_rate": 4.990325188195229e-05, "loss": 0.1171, "step": 28110 }, { "epoch": 0.03, "learning_rate": 4.9903244978175786e-05, "loss": 0.243, "step": 28111 }, { "epoch": 0.03, "learning_rate": 4.9903238074153456e-05, "loss": 0.2703, "step": 28112 }, { "epoch": 0.03, "learning_rate": 4.990323116988529e-05, "loss": 0.3748, "step": 28113 }, { "epoch": 0.03, "learning_rate": 4.990322426537129e-05, "loss": 0.8999, "step": 28114 }, { "epoch": 0.03, "learning_rate": 4.990321736061146e-05, "loss": 0.7932, "step": 28115 }, { "epoch": 0.03, "learning_rate": 4.990321045560579e-05, "loss": 1.0352, "step": 28116 }, { "epoch": 0.03, "learning_rate": 4.9903203550354286e-05, "loss": 1.2696, "step": 28117 }, { "epoch": 0.03, "learning_rate": 4.990319664485695e-05, "loss": 1.2741, "step": 28118 }, { "epoch": 0.03, "learning_rate": 4.9903189739113784e-05, "loss": 2.2469, "step": 28119 }, { "epoch": 0.03, "learning_rate": 4.9903182833124786e-05, "loss": 1.7212, "step": 28120 }, { "epoch": 0.03, "learning_rate": 4.9903175926889945e-05, "loss": 1.2671, "step": 28121 }, { "epoch": 0.03, "learning_rate": 4.9903169020409275e-05, "loss": 0.8349, "step": 28122 }, { "epoch": 0.03, "learning_rate": 4.9903162113682775e-05, "loss": 0.9352, "step": 28123 }, { "epoch": 0.03, "learning_rate": 4.990315520671043e-05, "loss": 0.2389, "step": 28124 }, { "epoch": 0.03, "learning_rate": 4.990314829949226e-05, "loss": 0.1606, "step": 28125 }, { "epoch": 0.03, "learning_rate": 4.990314139202826e-05, "loss": 0.2806, "step": 28126 }, { "epoch": 0.03, "learning_rate": 4.990313448431842e-05, "loss": 0.9006, "step": 28127 }, { "epoch": 0.03, "learning_rate": 4.990312757636275e-05, "loss": 0.1338, "step": 28128 }, { "epoch": 0.03, "learning_rate": 4.990312066816125e-05, "loss": 0.0675, "step": 28129 }, { "epoch": 0.03, "learning_rate": 4.990311375971391e-05, "loss": 0.0577, "step": 28130 }, { "epoch": 0.03, "learning_rate": 4.990310685102074e-05, "loss": 0.0589, "step": 28131 }, { "epoch": 0.03, "learning_rate": 4.990309994208173e-05, "loss": 0.0446, "step": 28132 }, { "epoch": 0.03, "learning_rate": 4.99030930328969e-05, "loss": 0.0436, "step": 28133 }, { "epoch": 0.03, "learning_rate": 4.990308612346623e-05, "loss": 0.0522, "step": 28134 }, { "epoch": 0.03, "learning_rate": 4.990307921378973e-05, "loss": 0.8274, "step": 28135 }, { "epoch": 0.03, "learning_rate": 4.9903072303867395e-05, "loss": 0.9632, "step": 28136 }, { "epoch": 0.03, "learning_rate": 4.990306539369923e-05, "loss": 0.9154, "step": 28137 }, { "epoch": 0.03, "learning_rate": 4.9903058483285224e-05, "loss": 0.966, "step": 28138 }, { "epoch": 0.03, "learning_rate": 4.9903051572625395e-05, "loss": 0.7533, "step": 28139 }, { "epoch": 0.03, "learning_rate": 4.9903044661719736e-05, "loss": 0.9982, "step": 28140 }, { "epoch": 0.03, "learning_rate": 4.9903037750568235e-05, "loss": 0.9029, "step": 28141 }, { "epoch": 0.03, "learning_rate": 4.9903030839170905e-05, "loss": 0.9644, "step": 28142 }, { "epoch": 0.03, "learning_rate": 4.9903023927527745e-05, "loss": 1.1409, "step": 28143 }, { "epoch": 0.03, "learning_rate": 4.9903017015638755e-05, "loss": 1.1985, "step": 28144 }, { "epoch": 0.03, "learning_rate": 4.990301010350393e-05, "loss": 1.0451, "step": 28145 }, { "epoch": 0.03, "learning_rate": 4.990300319112327e-05, "loss": 1.6279, "step": 28146 }, { "epoch": 0.03, "learning_rate": 4.9902996278496785e-05, "loss": 1.2381, "step": 28147 }, { "epoch": 0.03, "learning_rate": 4.990298936562446e-05, "loss": 1.0972, "step": 28148 }, { "epoch": 0.03, "learning_rate": 4.990298245250631e-05, "loss": 1.0395, "step": 28149 }, { "epoch": 0.03, "learning_rate": 4.9902975539142324e-05, "loss": 0.7105, "step": 28150 }, { "epoch": 0.03, "learning_rate": 4.99029686255325e-05, "loss": 1.5591, "step": 28151 }, { "epoch": 0.03, "learning_rate": 4.990296171167685e-05, "loss": 1.1703, "step": 28152 }, { "epoch": 0.03, "learning_rate": 4.990295479757537e-05, "loss": 1.2056, "step": 28153 }, { "epoch": 0.03, "learning_rate": 4.990294788322806e-05, "loss": 0.8667, "step": 28154 }, { "epoch": 0.03, "learning_rate": 4.990294096863492e-05, "loss": 0.8883, "step": 28155 }, { "epoch": 0.03, "learning_rate": 4.990293405379595e-05, "loss": 0.9588, "step": 28156 }, { "epoch": 0.03, "learning_rate": 4.990292713871114e-05, "loss": 1.1358, "step": 28157 }, { "epoch": 0.03, "learning_rate": 4.9902920223380506e-05, "loss": 1.3921, "step": 28158 }, { "epoch": 0.03, "learning_rate": 4.9902913307804036e-05, "loss": 1.0609, "step": 28159 }, { "epoch": 0.03, "learning_rate": 4.990290639198174e-05, "loss": 1.0441, "step": 28160 }, { "epoch": 0.03, "learning_rate": 4.990289947591361e-05, "loss": 1.1335, "step": 28161 }, { "epoch": 0.03, "learning_rate": 4.9902892559599646e-05, "loss": 1.1218, "step": 28162 }, { "epoch": 0.03, "learning_rate": 4.990288564303985e-05, "loss": 0.7981, "step": 28163 }, { "epoch": 0.03, "learning_rate": 4.990287872623424e-05, "loss": 1.0358, "step": 28164 }, { "epoch": 0.03, "learning_rate": 4.9902871809182786e-05, "loss": 1.1926, "step": 28165 }, { "epoch": 0.03, "learning_rate": 4.99028648918855e-05, "loss": 1.3177, "step": 28166 }, { "epoch": 0.03, "learning_rate": 4.990285797434239e-05, "loss": 0.8957, "step": 28167 }, { "epoch": 0.03, "learning_rate": 4.990285105655344e-05, "loss": 1.0052, "step": 28168 }, { "epoch": 0.03, "learning_rate": 4.9902844138518665e-05, "loss": 1.0644, "step": 28169 }, { "epoch": 0.03, "learning_rate": 4.990283722023806e-05, "loss": 0.6659, "step": 28170 }, { "epoch": 0.03, "learning_rate": 4.9902830301711626e-05, "loss": 0.9906, "step": 28171 }, { "epoch": 0.03, "learning_rate": 4.9902823382939356e-05, "loss": 1.0605, "step": 28172 }, { "epoch": 0.03, "learning_rate": 4.9902816463921256e-05, "loss": 0.9513, "step": 28173 }, { "epoch": 0.03, "learning_rate": 4.9902809544657334e-05, "loss": 0.7361, "step": 28174 }, { "epoch": 0.03, "learning_rate": 4.9902802625147576e-05, "loss": 0.9466, "step": 28175 }, { "epoch": 0.03, "learning_rate": 4.990279570539199e-05, "loss": 0.8939, "step": 28176 }, { "epoch": 0.03, "learning_rate": 4.990278878539058e-05, "loss": 1.091, "step": 28177 }, { "epoch": 0.03, "learning_rate": 4.9902781865143326e-05, "loss": 1.2918, "step": 28178 }, { "epoch": 0.03, "learning_rate": 4.990277494465026e-05, "loss": 0.8869, "step": 28179 }, { "epoch": 0.03, "learning_rate": 4.9902768023911346e-05, "loss": 0.4539, "step": 28180 }, { "epoch": 0.03, "learning_rate": 4.990276110292661e-05, "loss": 0.4692, "step": 28181 }, { "epoch": 0.03, "learning_rate": 4.990275418169604e-05, "loss": 0.4196, "step": 28182 }, { "epoch": 0.03, "learning_rate": 4.990274726021965e-05, "loss": 0.4764, "step": 28183 }, { "epoch": 0.03, "learning_rate": 4.990274033849743e-05, "loss": 0.3788, "step": 28184 }, { "epoch": 0.03, "learning_rate": 4.9902733416529376e-05, "loss": 0.3883, "step": 28185 }, { "epoch": 0.03, "learning_rate": 4.990272649431549e-05, "loss": 0.4327, "step": 28186 }, { "epoch": 0.03, "learning_rate": 4.9902719571855786e-05, "loss": 0.3982, "step": 28187 }, { "epoch": 0.03, "learning_rate": 4.990271264915024e-05, "loss": 0.4688, "step": 28188 }, { "epoch": 0.03, "learning_rate": 4.990270572619887e-05, "loss": 0.5443, "step": 28189 }, { "epoch": 0.03, "learning_rate": 4.990269880300167e-05, "loss": 0.3551, "step": 28190 }, { "epoch": 0.03, "learning_rate": 4.990269187955864e-05, "loss": 0.4007, "step": 28191 }, { "epoch": 0.03, "learning_rate": 4.9902684955869785e-05, "loss": 0.4227, "step": 28192 }, { "epoch": 0.03, "learning_rate": 4.99026780319351e-05, "loss": 0.3538, "step": 28193 }, { "epoch": 0.03, "learning_rate": 4.9902671107754585e-05, "loss": 0.3862, "step": 28194 }, { "epoch": 0.03, "learning_rate": 4.990266418332824e-05, "loss": 0.3946, "step": 28195 }, { "epoch": 0.03, "learning_rate": 4.9902657258656074e-05, "loss": 0.3744, "step": 28196 }, { "epoch": 0.03, "learning_rate": 4.990265033373807e-05, "loss": 0.4054, "step": 28197 }, { "epoch": 0.03, "learning_rate": 4.9902643408574246e-05, "loss": 0.3262, "step": 28198 }, { "epoch": 0.03, "learning_rate": 4.990263648316459e-05, "loss": 0.3724, "step": 28199 }, { "epoch": 0.03, "learning_rate": 4.99026295575091e-05, "loss": 0.8889, "step": 28200 }, { "epoch": 0.03, "learning_rate": 4.990262263160779e-05, "loss": 1.1527, "step": 28201 }, { "epoch": 0.03, "learning_rate": 4.990261570546064e-05, "loss": 1.1593, "step": 28202 }, { "epoch": 0.03, "learning_rate": 4.9902608779067675e-05, "loss": 0.6983, "step": 28203 }, { "epoch": 0.03, "learning_rate": 4.9902601852428874e-05, "loss": 1.044, "step": 28204 }, { "epoch": 0.03, "learning_rate": 4.990259492554425e-05, "loss": 1.0238, "step": 28205 }, { "epoch": 0.03, "learning_rate": 4.99025879984138e-05, "loss": 0.8343, "step": 28206 }, { "epoch": 0.03, "learning_rate": 4.9902581071037516e-05, "loss": 0.8743, "step": 28207 }, { "epoch": 0.03, "learning_rate": 4.9902574143415405e-05, "loss": 0.7489, "step": 28208 }, { "epoch": 0.03, "learning_rate": 4.990256721554747e-05, "loss": 0.9361, "step": 28209 }, { "epoch": 0.03, "learning_rate": 4.99025602874337e-05, "loss": 0.8569, "step": 28210 }, { "epoch": 0.03, "learning_rate": 4.990255335907411e-05, "loss": 0.7816, "step": 28211 }, { "epoch": 0.03, "learning_rate": 4.990254643046869e-05, "loss": 1.4015, "step": 28212 }, { "epoch": 0.03, "learning_rate": 4.990253950161744e-05, "loss": 0.7967, "step": 28213 }, { "epoch": 0.03, "learning_rate": 4.990253257252036e-05, "loss": 0.9797, "step": 28214 }, { "epoch": 0.03, "learning_rate": 4.990252564317747e-05, "loss": 1.334, "step": 28215 }, { "epoch": 0.03, "learning_rate": 4.9902518713588735e-05, "loss": 0.9531, "step": 28216 }, { "epoch": 0.03, "learning_rate": 4.990251178375418e-05, "loss": 0.9718, "step": 28217 }, { "epoch": 0.03, "learning_rate": 4.990250485367379e-05, "loss": 0.8477, "step": 28218 }, { "epoch": 0.03, "learning_rate": 4.9902497923347584e-05, "loss": 0.9986, "step": 28219 }, { "epoch": 0.03, "learning_rate": 4.990249099277555e-05, "loss": 1.2111, "step": 28220 }, { "epoch": 0.03, "learning_rate": 4.9902484061957683e-05, "loss": 1.1321, "step": 28221 }, { "epoch": 0.03, "learning_rate": 4.9902477130893996e-05, "loss": 1.0763, "step": 28222 }, { "epoch": 0.03, "learning_rate": 4.990247019958447e-05, "loss": 1.3244, "step": 28223 }, { "epoch": 0.03, "learning_rate": 4.990246326802913e-05, "loss": 1.1802, "step": 28224 }, { "epoch": 0.03, "learning_rate": 4.990245633622795e-05, "loss": 0.9435, "step": 28225 }, { "epoch": 0.03, "learning_rate": 4.990244940418096e-05, "loss": 0.8416, "step": 28226 }, { "epoch": 0.03, "learning_rate": 4.990244247188813e-05, "loss": 0.9066, "step": 28227 }, { "epoch": 0.03, "learning_rate": 4.9902435539349484e-05, "loss": 0.9976, "step": 28228 }, { "epoch": 0.03, "learning_rate": 4.990242860656501e-05, "loss": 0.9405, "step": 28229 }, { "epoch": 0.03, "learning_rate": 4.9902421673534703e-05, "loss": 1.0618, "step": 28230 }, { "epoch": 0.03, "learning_rate": 4.990241474025857e-05, "loss": 1.2564, "step": 28231 }, { "epoch": 0.03, "learning_rate": 4.990240780673662e-05, "loss": 0.4284, "step": 28232 }, { "epoch": 0.03, "learning_rate": 4.990240087296884e-05, "loss": 0.3401, "step": 28233 }, { "epoch": 0.03, "learning_rate": 4.9902393938955225e-05, "loss": 0.1685, "step": 28234 }, { "epoch": 0.03, "learning_rate": 4.990238700469579e-05, "loss": 0.1363, "step": 28235 }, { "epoch": 0.03, "learning_rate": 4.9902380070190534e-05, "loss": 0.1509, "step": 28236 }, { "epoch": 0.03, "learning_rate": 4.990237313543945e-05, "loss": 0.0958, "step": 28237 }, { "epoch": 0.03, "learning_rate": 4.990236620044253e-05, "loss": 0.2739, "step": 28238 }, { "epoch": 0.03, "learning_rate": 4.99023592651998e-05, "loss": 0.2117, "step": 28239 }, { "epoch": 0.03, "learning_rate": 4.9902352329711234e-05, "loss": 0.1476, "step": 28240 }, { "epoch": 0.03, "learning_rate": 4.9902345393976845e-05, "loss": 0.0874, "step": 28241 }, { "epoch": 0.03, "learning_rate": 4.990233845799663e-05, "loss": 0.0692, "step": 28242 }, { "epoch": 0.03, "learning_rate": 4.99023315217706e-05, "loss": 0.0705, "step": 28243 }, { "epoch": 0.03, "learning_rate": 4.9902324585298734e-05, "loss": 0.0752, "step": 28244 }, { "epoch": 0.03, "learning_rate": 4.990231764858104e-05, "loss": 0.101, "step": 28245 }, { "epoch": 0.03, "learning_rate": 4.990231071161753e-05, "loss": 0.2341, "step": 28246 }, { "epoch": 0.03, "learning_rate": 4.9902303774408186e-05, "loss": 0.8638, "step": 28247 }, { "epoch": 0.03, "learning_rate": 4.990229683695302e-05, "loss": 0.9512, "step": 28248 }, { "epoch": 0.03, "learning_rate": 4.990228989925203e-05, "loss": 1.1633, "step": 28249 }, { "epoch": 0.03, "learning_rate": 4.9902282961305216e-05, "loss": 1.0772, "step": 28250 }, { "epoch": 0.03, "learning_rate": 4.9902276023112574e-05, "loss": 1.0707, "step": 28251 }, { "epoch": 0.03, "learning_rate": 4.9902269084674116e-05, "loss": 1.0195, "step": 28252 }, { "epoch": 0.03, "learning_rate": 4.990226214598982e-05, "loss": 0.8853, "step": 28253 }, { "epoch": 0.03, "learning_rate": 4.990225520705971e-05, "loss": 1.1606, "step": 28254 }, { "epoch": 0.03, "learning_rate": 4.9902248267883774e-05, "loss": 0.9902, "step": 28255 }, { "epoch": 0.03, "learning_rate": 4.9902241328462006e-05, "loss": 0.8227, "step": 28256 }, { "epoch": 0.03, "learning_rate": 4.990223438879442e-05, "loss": 1.228, "step": 28257 }, { "epoch": 0.03, "learning_rate": 4.990222744888101e-05, "loss": 1.0418, "step": 28258 }, { "epoch": 0.03, "learning_rate": 4.9902220508721774e-05, "loss": 1.0187, "step": 28259 }, { "epoch": 0.03, "learning_rate": 4.9902213568316716e-05, "loss": 1.033, "step": 28260 }, { "epoch": 0.03, "learning_rate": 4.990220662766583e-05, "loss": 1.202, "step": 28261 }, { "epoch": 0.03, "learning_rate": 4.9902199686769125e-05, "loss": 1.0442, "step": 28262 }, { "epoch": 0.03, "learning_rate": 4.9902192745626586e-05, "loss": 1.0568, "step": 28263 }, { "epoch": 0.03, "learning_rate": 4.990218580423823e-05, "loss": 0.8532, "step": 28264 }, { "epoch": 0.03, "learning_rate": 4.9902178862604053e-05, "loss": 1.0368, "step": 28265 }, { "epoch": 0.03, "learning_rate": 4.9902171920724053e-05, "loss": 1.0165, "step": 28266 }, { "epoch": 0.03, "learning_rate": 4.990216497859822e-05, "loss": 0.973, "step": 28267 }, { "epoch": 0.03, "learning_rate": 4.9902158036226566e-05, "loss": 0.8017, "step": 28268 }, { "epoch": 0.03, "learning_rate": 4.99021510936091e-05, "loss": 1.253, "step": 28269 }, { "epoch": 0.03, "learning_rate": 4.9902144150745794e-05, "loss": 1.5565, "step": 28270 }, { "epoch": 0.03, "learning_rate": 4.9902137207636675e-05, "loss": 1.0083, "step": 28271 }, { "epoch": 0.03, "learning_rate": 4.9902130264281727e-05, "loss": 1.1158, "step": 28272 }, { "epoch": 0.03, "learning_rate": 4.990212332068096e-05, "loss": 0.8845, "step": 28273 }, { "epoch": 0.03, "learning_rate": 4.990211637683436e-05, "loss": 0.8166, "step": 28274 }, { "epoch": 0.03, "learning_rate": 4.990210943274195e-05, "loss": 0.8357, "step": 28275 }, { "epoch": 0.03, "learning_rate": 4.9902102488403714e-05, "loss": 0.9141, "step": 28276 }, { "epoch": 0.03, "learning_rate": 4.9902095543819646e-05, "loss": 0.9203, "step": 28277 }, { "epoch": 0.03, "learning_rate": 4.990208859898977e-05, "loss": 0.802, "step": 28278 }, { "epoch": 0.03, "learning_rate": 4.9902081653914056e-05, "loss": 1.1592, "step": 28279 }, { "epoch": 0.03, "learning_rate": 4.990207470859253e-05, "loss": 1.0773, "step": 28280 }, { "epoch": 0.03, "learning_rate": 4.9902067763025176e-05, "loss": 1.2081, "step": 28281 }, { "epoch": 0.03, "learning_rate": 4.9902060817211995e-05, "loss": 1.504, "step": 28282 }, { "epoch": 0.03, "learning_rate": 4.9902053871153e-05, "loss": 0.9278, "step": 28283 }, { "epoch": 0.03, "learning_rate": 4.990204692484818e-05, "loss": 1.0537, "step": 28284 }, { "epoch": 0.03, "learning_rate": 4.990203997829753e-05, "loss": 0.8877, "step": 28285 }, { "epoch": 0.03, "learning_rate": 4.9902033031501075e-05, "loss": 1.2112, "step": 28286 }, { "epoch": 0.03, "learning_rate": 4.990202608445878e-05, "loss": 1.1075, "step": 28287 }, { "epoch": 0.03, "learning_rate": 4.9902019137170667e-05, "loss": 1.0489, "step": 28288 }, { "epoch": 0.03, "learning_rate": 4.9902012189636735e-05, "loss": 0.9458, "step": 28289 }, { "epoch": 0.03, "learning_rate": 4.990200524185698e-05, "loss": 0.8962, "step": 28290 }, { "epoch": 0.03, "learning_rate": 4.9901998293831405e-05, "loss": 0.9982, "step": 28291 }, { "epoch": 0.03, "learning_rate": 4.9901991345560006e-05, "loss": 0.8113, "step": 28292 }, { "epoch": 0.03, "learning_rate": 4.9901984397042785e-05, "loss": 0.8342, "step": 28293 }, { "epoch": 0.03, "learning_rate": 4.990197744827974e-05, "loss": 0.6742, "step": 28294 }, { "epoch": 0.03, "learning_rate": 4.990197049927088e-05, "loss": 0.883, "step": 28295 }, { "epoch": 0.03, "learning_rate": 4.990196355001619e-05, "loss": 0.9611, "step": 28296 }, { "epoch": 0.03, "learning_rate": 4.990195660051568e-05, "loss": 1.0629, "step": 28297 }, { "epoch": 0.03, "learning_rate": 4.9901949650769353e-05, "loss": 0.8641, "step": 28298 }, { "epoch": 0.03, "learning_rate": 4.99019427007772e-05, "loss": 0.8821, "step": 28299 }, { "epoch": 0.03, "learning_rate": 4.9901935750539225e-05, "loss": 1.0841, "step": 28300 }, { "epoch": 0.03, "learning_rate": 4.990192880005543e-05, "loss": 0.6032, "step": 28301 }, { "epoch": 0.03, "learning_rate": 4.990192184932582e-05, "loss": 0.9826, "step": 28302 }, { "epoch": 0.03, "learning_rate": 4.990191489835038e-05, "loss": 0.7673, "step": 28303 }, { "epoch": 0.03, "learning_rate": 4.9901907947129124e-05, "loss": 0.8564, "step": 28304 }, { "epoch": 0.03, "learning_rate": 4.9901900995662046e-05, "loss": 1.152, "step": 28305 }, { "epoch": 0.03, "learning_rate": 4.990189404394914e-05, "loss": 0.9959, "step": 28306 }, { "epoch": 0.03, "learning_rate": 4.9901887091990416e-05, "loss": 0.5881, "step": 28307 }, { "epoch": 0.03, "learning_rate": 4.990188013978587e-05, "loss": 0.8403, "step": 28308 }, { "epoch": 0.03, "learning_rate": 4.9901873187335516e-05, "loss": 0.3552, "step": 28309 }, { "epoch": 0.03, "learning_rate": 4.9901866234639325e-05, "loss": 0.6944, "step": 28310 }, { "epoch": 0.03, "learning_rate": 4.9901859281697326e-05, "loss": 1.0386, "step": 28311 }, { "epoch": 0.03, "learning_rate": 4.9901852328509504e-05, "loss": 0.6666, "step": 28312 }, { "epoch": 0.03, "learning_rate": 4.990184537507585e-05, "loss": 0.2924, "step": 28313 }, { "epoch": 0.03, "learning_rate": 4.9901838421396385e-05, "loss": 0.7351, "step": 28314 }, { "epoch": 0.03, "learning_rate": 4.9901831467471096e-05, "loss": 0.6323, "step": 28315 }, { "epoch": 0.03, "learning_rate": 4.990182451329999e-05, "loss": 0.7889, "step": 28316 }, { "epoch": 0.03, "learning_rate": 4.990181755888306e-05, "loss": 1.1487, "step": 28317 }, { "epoch": 0.03, "learning_rate": 4.990181060422031e-05, "loss": 0.9646, "step": 28318 }, { "epoch": 0.03, "learning_rate": 4.9901803649311746e-05, "loss": 0.687, "step": 28319 }, { "epoch": 0.03, "learning_rate": 4.990179669415736e-05, "loss": 1.1109, "step": 28320 }, { "epoch": 0.03, "learning_rate": 4.9901789738757146e-05, "loss": 1.1146, "step": 28321 }, { "epoch": 0.03, "learning_rate": 4.990178278311112e-05, "loss": 1.0354, "step": 28322 }, { "epoch": 0.03, "learning_rate": 4.990177582721927e-05, "loss": 0.9813, "step": 28323 }, { "epoch": 0.03, "learning_rate": 4.9901768871081605e-05, "loss": 1.0834, "step": 28324 }, { "epoch": 0.03, "learning_rate": 4.990176191469812e-05, "loss": 0.9412, "step": 28325 }, { "epoch": 0.03, "learning_rate": 4.990175495806881e-05, "loss": 0.9777, "step": 28326 }, { "epoch": 0.03, "learning_rate": 4.990174800119368e-05, "loss": 0.7876, "step": 28327 }, { "epoch": 0.03, "learning_rate": 4.990174104407274e-05, "loss": 0.813, "step": 28328 }, { "epoch": 0.03, "learning_rate": 4.990173408670596e-05, "loss": 1.0251, "step": 28329 }, { "epoch": 0.03, "learning_rate": 4.990172712909338e-05, "loss": 1.1366, "step": 28330 }, { "epoch": 0.03, "learning_rate": 4.990172017123498e-05, "loss": 0.963, "step": 28331 }, { "epoch": 0.03, "learning_rate": 4.9901713213130755e-05, "loss": 0.8269, "step": 28332 }, { "epoch": 0.03, "learning_rate": 4.9901706254780714e-05, "loss": 1.1094, "step": 28333 }, { "epoch": 0.03, "learning_rate": 4.9901699296184844e-05, "loss": 0.8338, "step": 28334 }, { "epoch": 0.03, "learning_rate": 4.9901692337343165e-05, "loss": 1.0127, "step": 28335 }, { "epoch": 0.03, "learning_rate": 4.9901685378255664e-05, "loss": 1.3828, "step": 28336 }, { "epoch": 0.03, "learning_rate": 4.990167841892235e-05, "loss": 1.3905, "step": 28337 }, { "epoch": 0.03, "learning_rate": 4.99016714593432e-05, "loss": 1.0525, "step": 28338 }, { "epoch": 0.03, "learning_rate": 4.9901664499518245e-05, "loss": 1.2133, "step": 28339 }, { "epoch": 0.03, "learning_rate": 4.9901657539447474e-05, "loss": 0.7084, "step": 28340 }, { "epoch": 0.03, "learning_rate": 4.990165057913087e-05, "loss": 1.1137, "step": 28341 }, { "epoch": 0.03, "learning_rate": 4.9901643618568464e-05, "loss": 1.0293, "step": 28342 }, { "epoch": 0.03, "learning_rate": 4.990163665776023e-05, "loss": 1.179, "step": 28343 }, { "epoch": 0.03, "learning_rate": 4.9901629696706184e-05, "loss": 0.8268, "step": 28344 }, { "epoch": 0.03, "learning_rate": 4.9901622735406314e-05, "loss": 0.7043, "step": 28345 }, { "epoch": 0.03, "learning_rate": 4.990161577386062e-05, "loss": 0.381, "step": 28346 }, { "epoch": 0.03, "learning_rate": 4.990160881206912e-05, "loss": 0.967, "step": 28347 }, { "epoch": 0.03, "learning_rate": 4.9901601850031796e-05, "loss": 1.2824, "step": 28348 }, { "epoch": 0.03, "learning_rate": 4.9901594887748656e-05, "loss": 0.7894, "step": 28349 }, { "epoch": 0.03, "learning_rate": 4.9901587925219693e-05, "loss": 0.7901, "step": 28350 }, { "epoch": 0.03, "learning_rate": 4.9901580962444915e-05, "loss": 1.2485, "step": 28351 }, { "epoch": 0.03, "learning_rate": 4.990157399942432e-05, "loss": 1.0291, "step": 28352 }, { "epoch": 0.03, "learning_rate": 4.9901567036157905e-05, "loss": 0.8999, "step": 28353 }, { "epoch": 0.03, "learning_rate": 4.990156007264567e-05, "loss": 0.7966, "step": 28354 }, { "epoch": 0.03, "learning_rate": 4.9901553108887625e-05, "loss": 0.8517, "step": 28355 }, { "epoch": 0.03, "learning_rate": 4.990154614488376e-05, "loss": 1.205, "step": 28356 }, { "epoch": 0.03, "learning_rate": 4.9901539180634075e-05, "loss": 0.9917, "step": 28357 }, { "epoch": 0.03, "learning_rate": 4.9901532216138574e-05, "loss": 1.1575, "step": 28358 }, { "epoch": 0.03, "learning_rate": 4.990152525139725e-05, "loss": 0.9967, "step": 28359 }, { "epoch": 0.03, "learning_rate": 4.990151828641012e-05, "loss": 1.0684, "step": 28360 }, { "epoch": 0.03, "learning_rate": 4.990151132117716e-05, "loss": 1.2854, "step": 28361 }, { "epoch": 0.03, "learning_rate": 4.990150435569839e-05, "loss": 0.8797, "step": 28362 }, { "epoch": 0.03, "learning_rate": 4.990149738997381e-05, "loss": 0.9127, "step": 28363 }, { "epoch": 0.03, "learning_rate": 4.99014904240034e-05, "loss": 0.9272, "step": 28364 }, { "epoch": 0.03, "learning_rate": 4.990148345778718e-05, "loss": 1.0818, "step": 28365 }, { "epoch": 0.03, "learning_rate": 4.990147649132514e-05, "loss": 1.237, "step": 28366 }, { "epoch": 0.03, "learning_rate": 4.990146952461728e-05, "loss": 0.8212, "step": 28367 }, { "epoch": 0.03, "learning_rate": 4.990146255766361e-05, "loss": 0.8524, "step": 28368 }, { "epoch": 0.03, "learning_rate": 4.990145559046412e-05, "loss": 1.082, "step": 28369 }, { "epoch": 0.03, "learning_rate": 4.990144862301881e-05, "loss": 1.1686, "step": 28370 }, { "epoch": 0.03, "learning_rate": 4.990144165532769e-05, "loss": 0.9144, "step": 28371 }, { "epoch": 0.03, "learning_rate": 4.9901434687390746e-05, "loss": 0.884, "step": 28372 }, { "epoch": 0.03, "learning_rate": 4.9901427719207995e-05, "loss": 0.9381, "step": 28373 }, { "epoch": 0.03, "learning_rate": 4.990142075077942e-05, "loss": 0.9497, "step": 28374 }, { "epoch": 0.03, "learning_rate": 4.990141378210503e-05, "loss": 1.1184, "step": 28375 }, { "epoch": 0.03, "learning_rate": 4.9901406813184826e-05, "loss": 1.0486, "step": 28376 }, { "epoch": 0.03, "learning_rate": 4.9901399844018804e-05, "loss": 0.9403, "step": 28377 }, { "epoch": 0.03, "learning_rate": 4.990139287460697e-05, "loss": 1.0929, "step": 28378 }, { "epoch": 0.03, "learning_rate": 4.9901385904949315e-05, "loss": 0.9653, "step": 28379 }, { "epoch": 0.03, "learning_rate": 4.990137893504585e-05, "loss": 1.2322, "step": 28380 }, { "epoch": 0.03, "learning_rate": 4.990137196489656e-05, "loss": 1.1323, "step": 28381 }, { "epoch": 0.03, "learning_rate": 4.990136499450146e-05, "loss": 0.849, "step": 28382 }, { "epoch": 0.03, "learning_rate": 4.990135802386055e-05, "loss": 0.9672, "step": 28383 }, { "epoch": 0.03, "learning_rate": 4.990135105297381e-05, "loss": 0.8402, "step": 28384 }, { "epoch": 0.03, "learning_rate": 4.990134408184126e-05, "loss": 0.9123, "step": 28385 }, { "epoch": 0.03, "learning_rate": 4.99013371104629e-05, "loss": 1.0671, "step": 28386 }, { "epoch": 0.03, "learning_rate": 4.990133013883872e-05, "loss": 0.6706, "step": 28387 }, { "epoch": 0.03, "learning_rate": 4.9901323166968726e-05, "loss": 1.1225, "step": 28388 }, { "epoch": 0.03, "learning_rate": 4.9901316194852916e-05, "loss": 1.2613, "step": 28389 }, { "epoch": 0.03, "learning_rate": 4.990130922249129e-05, "loss": 1.1993, "step": 28390 }, { "epoch": 0.03, "learning_rate": 4.990130224988385e-05, "loss": 1.0665, "step": 28391 }, { "epoch": 0.03, "learning_rate": 4.990129527703059e-05, "loss": 0.9423, "step": 28392 }, { "epoch": 0.03, "learning_rate": 4.990128830393152e-05, "loss": 1.1464, "step": 28393 }, { "epoch": 0.03, "learning_rate": 4.9901281330586636e-05, "loss": 0.6218, "step": 28394 }, { "epoch": 0.03, "learning_rate": 4.990127435699593e-05, "loss": 1.0324, "step": 28395 }, { "epoch": 0.03, "learning_rate": 4.990126738315942e-05, "loss": 1.0962, "step": 28396 }, { "epoch": 0.03, "learning_rate": 4.990126040907709e-05, "loss": 0.9578, "step": 28397 }, { "epoch": 0.03, "learning_rate": 4.990125343474894e-05, "loss": 0.9774, "step": 28398 }, { "epoch": 0.03, "learning_rate": 4.9901246460174984e-05, "loss": 1.1583, "step": 28399 }, { "epoch": 0.03, "learning_rate": 4.990123948535521e-05, "loss": 1.0595, "step": 28400 }, { "epoch": 0.03, "learning_rate": 4.990123251028962e-05, "loss": 1.0672, "step": 28401 }, { "epoch": 0.03, "learning_rate": 4.9901225534978214e-05, "loss": 1.0685, "step": 28402 }, { "epoch": 0.03, "learning_rate": 4.990121855942099e-05, "loss": 0.9695, "step": 28403 }, { "epoch": 0.03, "learning_rate": 4.9901211583617965e-05, "loss": 1.1479, "step": 28404 }, { "epoch": 0.03, "learning_rate": 4.990120460756912e-05, "loss": 1.0288, "step": 28405 }, { "epoch": 0.03, "learning_rate": 4.990119763127445e-05, "loss": 1.0141, "step": 28406 }, { "epoch": 0.03, "learning_rate": 4.990119065473398e-05, "loss": 1.2768, "step": 28407 }, { "epoch": 0.03, "learning_rate": 4.990118367794769e-05, "loss": 1.0206, "step": 28408 }, { "epoch": 0.03, "learning_rate": 4.9901176700915587e-05, "loss": 0.8942, "step": 28409 }, { "epoch": 0.03, "learning_rate": 4.990116972363767e-05, "loss": 1.1095, "step": 28410 }, { "epoch": 0.03, "learning_rate": 4.990116274611394e-05, "loss": 1.043, "step": 28411 }, { "epoch": 0.03, "learning_rate": 4.99011557683444e-05, "loss": 1.0484, "step": 28412 }, { "epoch": 0.03, "learning_rate": 4.990114879032903e-05, "loss": 0.9186, "step": 28413 }, { "epoch": 0.03, "learning_rate": 4.990114181206786e-05, "loss": 0.8102, "step": 28414 }, { "epoch": 0.03, "learning_rate": 4.9901134833560883e-05, "loss": 0.9608, "step": 28415 }, { "epoch": 0.03, "learning_rate": 4.990112785480808e-05, "loss": 0.8926, "step": 28416 }, { "epoch": 0.03, "learning_rate": 4.990112087580947e-05, "loss": 1.1341, "step": 28417 }, { "epoch": 0.03, "learning_rate": 4.990111389656505e-05, "loss": 1.053, "step": 28418 }, { "epoch": 0.03, "learning_rate": 4.9901106917074805e-05, "loss": 0.9904, "step": 28419 }, { "epoch": 0.03, "learning_rate": 4.990109993733875e-05, "loss": 0.5957, "step": 28420 }, { "epoch": 0.03, "learning_rate": 4.990109295735689e-05, "loss": 1.2327, "step": 28421 }, { "epoch": 0.03, "learning_rate": 4.99010859771292e-05, "loss": 1.0083, "step": 28422 }, { "epoch": 0.03, "learning_rate": 4.9901078996655716e-05, "loss": 0.8457, "step": 28423 }, { "epoch": 0.03, "learning_rate": 4.990107201593641e-05, "loss": 0.9625, "step": 28424 }, { "epoch": 0.03, "learning_rate": 4.990106503497129e-05, "loss": 0.704, "step": 28425 }, { "epoch": 0.03, "learning_rate": 4.990105805376036e-05, "loss": 0.668, "step": 28426 }, { "epoch": 0.03, "learning_rate": 4.9901051072303615e-05, "loss": 0.7635, "step": 28427 }, { "epoch": 0.03, "learning_rate": 4.9901044090601066e-05, "loss": 1.2434, "step": 28428 }, { "epoch": 0.03, "learning_rate": 4.9901037108652694e-05, "loss": 1.0575, "step": 28429 }, { "epoch": 0.03, "learning_rate": 4.990103012645851e-05, "loss": 0.8439, "step": 28430 }, { "epoch": 0.03, "learning_rate": 4.9901023144018523e-05, "loss": 1.1202, "step": 28431 }, { "epoch": 0.03, "learning_rate": 4.990101616133271e-05, "loss": 1.0784, "step": 28432 }, { "epoch": 0.03, "learning_rate": 4.990100917840109e-05, "loss": 1.2409, "step": 28433 }, { "epoch": 0.03, "learning_rate": 4.990100219522367e-05, "loss": 0.9861, "step": 28434 }, { "epoch": 0.03, "learning_rate": 4.990099521180042e-05, "loss": 0.6059, "step": 28435 }, { "epoch": 0.03, "learning_rate": 4.990098822813136e-05, "loss": 0.7963, "step": 28436 }, { "epoch": 0.03, "learning_rate": 4.99009812442165e-05, "loss": 0.9145, "step": 28437 }, { "epoch": 0.03, "learning_rate": 4.990097426005582e-05, "loss": 1.3814, "step": 28438 }, { "epoch": 0.03, "learning_rate": 4.990096727564933e-05, "loss": 0.9499, "step": 28439 }, { "epoch": 0.03, "learning_rate": 4.990096029099702e-05, "loss": 0.7156, "step": 28440 }, { "epoch": 0.03, "learning_rate": 4.990095330609891e-05, "loss": 0.9307, "step": 28441 }, { "epoch": 0.03, "learning_rate": 4.990094632095499e-05, "loss": 1.083, "step": 28442 }, { "epoch": 0.03, "learning_rate": 4.990093933556524e-05, "loss": 0.6087, "step": 28443 }, { "epoch": 0.03, "learning_rate": 4.9900932349929696e-05, "loss": 0.931, "step": 28444 }, { "epoch": 0.03, "learning_rate": 4.9900925364048334e-05, "loss": 0.992, "step": 28445 }, { "epoch": 0.03, "learning_rate": 4.990091837792117e-05, "loss": 1.1452, "step": 28446 }, { "epoch": 0.03, "learning_rate": 4.9900911391548175e-05, "loss": 0.8142, "step": 28447 }, { "epoch": 0.03, "learning_rate": 4.9900904404929385e-05, "loss": 0.9016, "step": 28448 }, { "epoch": 0.03, "learning_rate": 4.990089741806478e-05, "loss": 0.913, "step": 28449 }, { "epoch": 0.03, "learning_rate": 4.990089043095436e-05, "loss": 0.6357, "step": 28450 }, { "epoch": 0.03, "learning_rate": 4.9900883443598136e-05, "loss": 0.8098, "step": 28451 }, { "epoch": 0.03, "learning_rate": 4.990087645599609e-05, "loss": 1.1466, "step": 28452 }, { "epoch": 0.03, "learning_rate": 4.990086946814824e-05, "loss": 0.9825, "step": 28453 }, { "epoch": 0.03, "learning_rate": 4.9900862480054586e-05, "loss": 0.997, "step": 28454 }, { "epoch": 0.03, "learning_rate": 4.9900855491715107e-05, "loss": 0.8618, "step": 28455 }, { "epoch": 0.03, "learning_rate": 4.9900848503129825e-05, "loss": 1.2629, "step": 28456 }, { "epoch": 0.03, "learning_rate": 4.9900841514298735e-05, "loss": 0.8893, "step": 28457 }, { "epoch": 0.03, "learning_rate": 4.990083452522183e-05, "loss": 0.3115, "step": 28458 }, { "epoch": 0.03, "learning_rate": 4.990082753589911e-05, "loss": 1.0696, "step": 28459 }, { "epoch": 0.03, "learning_rate": 4.990082054633058e-05, "loss": 1.4425, "step": 28460 }, { "epoch": 0.03, "learning_rate": 4.990081355651625e-05, "loss": 0.4861, "step": 28461 }, { "epoch": 0.03, "learning_rate": 4.990080656645611e-05, "loss": 0.2454, "step": 28462 }, { "epoch": 0.03, "learning_rate": 4.990079957615015e-05, "loss": 0.6129, "step": 28463 }, { "epoch": 0.03, "learning_rate": 4.990079258559838e-05, "loss": 0.6831, "step": 28464 }, { "epoch": 0.03, "learning_rate": 4.99007855948008e-05, "loss": 1.0326, "step": 28465 }, { "epoch": 0.03, "learning_rate": 4.990077860375742e-05, "loss": 1.0479, "step": 28466 }, { "epoch": 0.03, "learning_rate": 4.990077161246822e-05, "loss": 1.1652, "step": 28467 }, { "epoch": 0.03, "learning_rate": 4.990076462093321e-05, "loss": 0.8019, "step": 28468 }, { "epoch": 0.03, "learning_rate": 4.99007576291524e-05, "loss": 0.8394, "step": 28469 }, { "epoch": 0.03, "learning_rate": 4.9900750637125774e-05, "loss": 0.3869, "step": 28470 }, { "epoch": 0.03, "learning_rate": 4.990074364485334e-05, "loss": 1.0521, "step": 28471 }, { "epoch": 0.03, "learning_rate": 4.990073665233509e-05, "loss": 0.6032, "step": 28472 }, { "epoch": 0.03, "learning_rate": 4.990072965957103e-05, "loss": 1.002, "step": 28473 }, { "epoch": 0.03, "learning_rate": 4.9900722666561175e-05, "loss": 1.3056, "step": 28474 }, { "epoch": 0.03, "learning_rate": 4.9900715673305496e-05, "loss": 1.0339, "step": 28475 }, { "epoch": 0.03, "learning_rate": 4.9900708679804015e-05, "loss": 1.0378, "step": 28476 }, { "epoch": 0.03, "learning_rate": 4.9900701686056725e-05, "loss": 0.8763, "step": 28477 }, { "epoch": 0.03, "learning_rate": 4.990069469206362e-05, "loss": 0.3617, "step": 28478 }, { "epoch": 0.03, "learning_rate": 4.990068769782471e-05, "loss": 0.7455, "step": 28479 }, { "epoch": 0.03, "learning_rate": 4.990068070333999e-05, "loss": 0.8597, "step": 28480 }, { "epoch": 0.03, "learning_rate": 4.9900673708609455e-05, "loss": 0.6916, "step": 28481 }, { "epoch": 0.03, "learning_rate": 4.990066671363312e-05, "loss": 0.9329, "step": 28482 }, { "epoch": 0.03, "learning_rate": 4.990065971841097e-05, "loss": 1.4452, "step": 28483 }, { "epoch": 0.03, "learning_rate": 4.990065272294302e-05, "loss": 1.1547, "step": 28484 }, { "epoch": 0.03, "learning_rate": 4.990064572722925e-05, "loss": 1.1403, "step": 28485 }, { "epoch": 0.03, "learning_rate": 4.990063873126968e-05, "loss": 0.9198, "step": 28486 }, { "epoch": 0.03, "learning_rate": 4.99006317350643e-05, "loss": 1.2678, "step": 28487 }, { "epoch": 0.03, "learning_rate": 4.99006247386131e-05, "loss": 0.9227, "step": 28488 }, { "epoch": 0.03, "learning_rate": 4.9900617741916105e-05, "loss": 1.1111, "step": 28489 }, { "epoch": 0.03, "learning_rate": 4.99006107449733e-05, "loss": 0.7573, "step": 28490 }, { "epoch": 0.03, "learning_rate": 4.990060374778468e-05, "loss": 1.0136, "step": 28491 }, { "epoch": 0.03, "learning_rate": 4.990059675035026e-05, "loss": 0.8462, "step": 28492 }, { "epoch": 0.03, "learning_rate": 4.9900589752670026e-05, "loss": 1.2104, "step": 28493 }, { "epoch": 0.03, "learning_rate": 4.990058275474398e-05, "loss": 1.0044, "step": 28494 }, { "epoch": 0.03, "learning_rate": 4.990057575657214e-05, "loss": 0.2718, "step": 28495 }, { "epoch": 0.03, "learning_rate": 4.990056875815448e-05, "loss": 0.5345, "step": 28496 }, { "epoch": 0.03, "learning_rate": 4.9900561759491016e-05, "loss": 0.8885, "step": 28497 }, { "epoch": 0.03, "learning_rate": 4.9900554760581745e-05, "loss": 0.9176, "step": 28498 }, { "epoch": 0.03, "learning_rate": 4.9900547761426664e-05, "loss": 0.7901, "step": 28499 }, { "epoch": 0.03, "learning_rate": 4.990054076202577e-05, "loss": 0.8011, "step": 28500 }, { "epoch": 0.03, "eval_loss": 1.059694766998291, "eval_runtime": 126.0487, "eval_samples_per_second": 10.988, "eval_steps_per_second": 5.498, "step": 28500 }, { "epoch": 0.03, "learning_rate": 4.990053376237908e-05, "loss": 0.9402, "step": 28501 }, { "epoch": 0.03, "learning_rate": 4.9900526762486576e-05, "loss": 0.8917, "step": 28502 }, { "epoch": 0.03, "learning_rate": 4.9900519762348267e-05, "loss": 0.9597, "step": 28503 }, { "epoch": 0.03, "learning_rate": 4.990051276196415e-05, "loss": 0.7764, "step": 28504 }, { "epoch": 0.03, "learning_rate": 4.990050576133423e-05, "loss": 0.9763, "step": 28505 }, { "epoch": 0.03, "learning_rate": 4.990049876045849e-05, "loss": 0.9869, "step": 28506 }, { "epoch": 0.03, "learning_rate": 4.990049175933695e-05, "loss": 1.2697, "step": 28507 }, { "epoch": 0.03, "learning_rate": 4.99004847579696e-05, "loss": 0.7036, "step": 28508 }, { "epoch": 0.03, "learning_rate": 4.990047775635644e-05, "loss": 1.0836, "step": 28509 }, { "epoch": 0.03, "learning_rate": 4.9900470754497483e-05, "loss": 0.8644, "step": 28510 }, { "epoch": 0.03, "learning_rate": 4.990046375239272e-05, "loss": 1.0702, "step": 28511 }, { "epoch": 0.03, "learning_rate": 4.9900456750042146e-05, "loss": 0.9568, "step": 28512 }, { "epoch": 0.03, "learning_rate": 4.990044974744577e-05, "loss": 0.8329, "step": 28513 }, { "epoch": 0.03, "learning_rate": 4.990044274460357e-05, "loss": 1.0985, "step": 28514 }, { "epoch": 0.03, "learning_rate": 4.990043574151558e-05, "loss": 1.0644, "step": 28515 }, { "epoch": 0.03, "learning_rate": 4.9900428738181784e-05, "loss": 0.9356, "step": 28516 }, { "epoch": 0.03, "learning_rate": 4.990042173460216e-05, "loss": 0.974, "step": 28517 }, { "epoch": 0.03, "learning_rate": 4.990041473077676e-05, "loss": 1.0704, "step": 28518 }, { "epoch": 0.03, "learning_rate": 4.9900407726705534e-05, "loss": 0.9438, "step": 28519 }, { "epoch": 0.03, "learning_rate": 4.9900400722388506e-05, "loss": 0.822, "step": 28520 }, { "epoch": 0.03, "learning_rate": 4.990039371782567e-05, "loss": 0.8225, "step": 28521 }, { "epoch": 0.03, "learning_rate": 4.990038671301703e-05, "loss": 0.7054, "step": 28522 }, { "epoch": 0.03, "learning_rate": 4.990037970796259e-05, "loss": 0.8314, "step": 28523 }, { "epoch": 0.03, "learning_rate": 4.990037270266233e-05, "loss": 0.7516, "step": 28524 }, { "epoch": 0.03, "learning_rate": 4.990036569711627e-05, "loss": 0.9527, "step": 28525 }, { "epoch": 0.03, "learning_rate": 4.990035869132441e-05, "loss": 1.0401, "step": 28526 }, { "epoch": 0.03, "learning_rate": 4.990035168528674e-05, "loss": 0.9183, "step": 28527 }, { "epoch": 0.03, "learning_rate": 4.990034467900326e-05, "loss": 0.9367, "step": 28528 }, { "epoch": 0.03, "learning_rate": 4.990033767247399e-05, "loss": 1.0987, "step": 28529 }, { "epoch": 0.03, "learning_rate": 4.9900330665698897e-05, "loss": 1.0158, "step": 28530 }, { "epoch": 0.03, "learning_rate": 4.9900323658678004e-05, "loss": 1.0933, "step": 28531 }, { "epoch": 0.03, "learning_rate": 4.99003166514113e-05, "loss": 1.021, "step": 28532 }, { "epoch": 0.03, "learning_rate": 4.99003096438988e-05, "loss": 1.1417, "step": 28533 }, { "epoch": 0.03, "learning_rate": 4.990030263614049e-05, "loss": 1.0498, "step": 28534 }, { "epoch": 0.03, "learning_rate": 4.990029562813638e-05, "loss": 0.9506, "step": 28535 }, { "epoch": 0.03, "learning_rate": 4.990028861988646e-05, "loss": 1.035, "step": 28536 }, { "epoch": 0.03, "learning_rate": 4.990028161139073e-05, "loss": 1.0413, "step": 28537 }, { "epoch": 0.03, "learning_rate": 4.9900274602649206e-05, "loss": 0.7374, "step": 28538 }, { "epoch": 0.03, "learning_rate": 4.990026759366187e-05, "loss": 0.7415, "step": 28539 }, { "epoch": 0.03, "learning_rate": 4.990026058442872e-05, "loss": 0.987, "step": 28540 }, { "epoch": 0.03, "learning_rate": 4.990025357494978e-05, "loss": 1.2085, "step": 28541 }, { "epoch": 0.03, "learning_rate": 4.990024656522504e-05, "loss": 1.0309, "step": 28542 }, { "epoch": 0.03, "learning_rate": 4.990023955525448e-05, "loss": 0.5912, "step": 28543 }, { "epoch": 0.03, "learning_rate": 4.9900232545038123e-05, "loss": 0.8013, "step": 28544 }, { "epoch": 0.03, "learning_rate": 4.990022553457596e-05, "loss": 0.8033, "step": 28545 }, { "epoch": 0.03, "learning_rate": 4.9900218523867994e-05, "loss": 0.6054, "step": 28546 }, { "epoch": 0.03, "learning_rate": 4.990021151291422e-05, "loss": 0.362, "step": 28547 }, { "epoch": 0.03, "learning_rate": 4.990020450171464e-05, "loss": 0.5153, "step": 28548 }, { "epoch": 0.03, "learning_rate": 4.990019749026926e-05, "loss": 0.1314, "step": 28549 }, { "epoch": 0.03, "learning_rate": 4.990019047857808e-05, "loss": 0.2516, "step": 28550 }, { "epoch": 0.03, "learning_rate": 4.990018346664108e-05, "loss": 0.4292, "step": 28551 }, { "epoch": 0.03, "learning_rate": 4.990017645445829e-05, "loss": 0.248, "step": 28552 }, { "epoch": 0.03, "learning_rate": 4.9900169442029696e-05, "loss": 0.0982, "step": 28553 }, { "epoch": 0.03, "learning_rate": 4.990016242935529e-05, "loss": 0.1042, "step": 28554 }, { "epoch": 0.03, "learning_rate": 4.990015541643509e-05, "loss": 0.0644, "step": 28555 }, { "epoch": 0.03, "learning_rate": 4.990014840326908e-05, "loss": 0.3811, "step": 28556 }, { "epoch": 0.03, "learning_rate": 4.990014138985726e-05, "loss": 0.4002, "step": 28557 }, { "epoch": 0.03, "learning_rate": 4.990013437619965e-05, "loss": 0.287, "step": 28558 }, { "epoch": 0.03, "learning_rate": 4.9900127362296226e-05, "loss": 0.1507, "step": 28559 }, { "epoch": 0.03, "learning_rate": 4.990012034814701e-05, "loss": 0.2797, "step": 28560 }, { "epoch": 0.03, "learning_rate": 4.9900113333751977e-05, "loss": 0.0635, "step": 28561 }, { "epoch": 0.03, "learning_rate": 4.990010631911114e-05, "loss": 0.2525, "step": 28562 }, { "epoch": 0.03, "learning_rate": 4.990009930422451e-05, "loss": 0.2745, "step": 28563 }, { "epoch": 0.03, "learning_rate": 4.990009228909207e-05, "loss": 0.2307, "step": 28564 }, { "epoch": 0.03, "learning_rate": 4.990008527371383e-05, "loss": 0.1444, "step": 28565 }, { "epoch": 0.03, "learning_rate": 4.990007825808979e-05, "loss": 0.0858, "step": 28566 }, { "epoch": 0.03, "learning_rate": 4.9900071242219934e-05, "loss": 0.0412, "step": 28567 }, { "epoch": 0.03, "learning_rate": 4.990006422610429e-05, "loss": 0.7176, "step": 28568 }, { "epoch": 0.03, "learning_rate": 4.9900057209742836e-05, "loss": 0.6349, "step": 28569 }, { "epoch": 0.03, "learning_rate": 4.9900050193135584e-05, "loss": 0.5989, "step": 28570 }, { "epoch": 0.03, "learning_rate": 4.9900043176282516e-05, "loss": 0.6219, "step": 28571 }, { "epoch": 0.03, "learning_rate": 4.990003615918366e-05, "loss": 0.7263, "step": 28572 }, { "epoch": 0.03, "learning_rate": 4.9900029141838993e-05, "loss": 1.0618, "step": 28573 }, { "epoch": 0.03, "learning_rate": 4.9900022124248526e-05, "loss": 1.1109, "step": 28574 }, { "epoch": 0.03, "learning_rate": 4.9900015106412256e-05, "loss": 1.3556, "step": 28575 }, { "epoch": 0.03, "learning_rate": 4.990000808833018e-05, "loss": 1.0419, "step": 28576 }, { "epoch": 0.03, "learning_rate": 4.990000107000231e-05, "loss": 0.9642, "step": 28577 }, { "epoch": 0.03, "learning_rate": 4.989999405142863e-05, "loss": 1.1836, "step": 28578 }, { "epoch": 0.03, "learning_rate": 4.989998703260916e-05, "loss": 1.105, "step": 28579 }, { "epoch": 0.03, "learning_rate": 4.989998001354388e-05, "loss": 0.6727, "step": 28580 }, { "epoch": 0.03, "learning_rate": 4.989997299423279e-05, "loss": 0.8764, "step": 28581 }, { "epoch": 0.03, "learning_rate": 4.989996597467591e-05, "loss": 1.5111, "step": 28582 }, { "epoch": 0.03, "learning_rate": 4.989995895487322e-05, "loss": 0.8916, "step": 28583 }, { "epoch": 0.03, "learning_rate": 4.989995193482473e-05, "loss": 0.7059, "step": 28584 }, { "epoch": 0.03, "learning_rate": 4.989994491453044e-05, "loss": 0.8515, "step": 28585 }, { "epoch": 0.03, "learning_rate": 4.989993789399035e-05, "loss": 1.1137, "step": 28586 }, { "epoch": 0.03, "learning_rate": 4.989993087320445e-05, "loss": 0.9509, "step": 28587 }, { "epoch": 0.03, "learning_rate": 4.989992385217276e-05, "loss": 0.9518, "step": 28588 }, { "epoch": 0.03, "learning_rate": 4.989991683089526e-05, "loss": 0.5354, "step": 28589 }, { "epoch": 0.03, "learning_rate": 4.9899909809371956e-05, "loss": 1.2061, "step": 28590 }, { "epoch": 0.03, "learning_rate": 4.989990278760286e-05, "loss": 1.2859, "step": 28591 }, { "epoch": 0.03, "learning_rate": 4.9899895765587964e-05, "loss": 1.371, "step": 28592 }, { "epoch": 0.03, "learning_rate": 4.989988874332726e-05, "loss": 0.9958, "step": 28593 }, { "epoch": 0.03, "learning_rate": 4.989988172082075e-05, "loss": 0.5965, "step": 28594 }, { "epoch": 0.03, "learning_rate": 4.989987469806845e-05, "loss": 1.0089, "step": 28595 }, { "epoch": 0.03, "learning_rate": 4.989986767507035e-05, "loss": 0.9946, "step": 28596 }, { "epoch": 0.03, "learning_rate": 4.9899860651826436e-05, "loss": 1.074, "step": 28597 }, { "epoch": 0.03, "learning_rate": 4.989985362833672e-05, "loss": 1.4328, "step": 28598 }, { "epoch": 0.03, "learning_rate": 4.989984660460122e-05, "loss": 1.0511, "step": 28599 }, { "epoch": 0.03, "learning_rate": 4.989983958061991e-05, "loss": 0.8966, "step": 28600 }, { "epoch": 0.03, "learning_rate": 4.9899832556392796e-05, "loss": 1.0801, "step": 28601 }, { "epoch": 0.03, "learning_rate": 4.989982553191989e-05, "loss": 0.8664, "step": 28602 }, { "epoch": 0.03, "learning_rate": 4.9899818507201176e-05, "loss": 0.3566, "step": 28603 }, { "epoch": 0.03, "learning_rate": 4.9899811482236656e-05, "loss": 0.4538, "step": 28604 }, { "epoch": 0.03, "learning_rate": 4.989980445702635e-05, "loss": 1.1877, "step": 28605 }, { "epoch": 0.03, "learning_rate": 4.989979743157023e-05, "loss": 1.3525, "step": 28606 }, { "epoch": 0.03, "learning_rate": 4.989979040586832e-05, "loss": 0.6419, "step": 28607 }, { "epoch": 0.03, "learning_rate": 4.989978337992061e-05, "loss": 0.6325, "step": 28608 }, { "epoch": 0.03, "learning_rate": 4.989977635372709e-05, "loss": 0.575, "step": 28609 }, { "epoch": 0.03, "learning_rate": 4.989976932728778e-05, "loss": 0.56, "step": 28610 }, { "epoch": 0.03, "learning_rate": 4.9899762300602664e-05, "loss": 0.4272, "step": 28611 }, { "epoch": 0.03, "learning_rate": 4.989975527367175e-05, "loss": 0.9889, "step": 28612 }, { "epoch": 0.03, "learning_rate": 4.9899748246495035e-05, "loss": 0.9483, "step": 28613 }, { "epoch": 0.03, "learning_rate": 4.989974121907253e-05, "loss": 1.04, "step": 28614 }, { "epoch": 0.03, "learning_rate": 4.989973419140421e-05, "loss": 0.9122, "step": 28615 }, { "epoch": 0.03, "learning_rate": 4.98997271634901e-05, "loss": 1.1204, "step": 28616 }, { "epoch": 0.03, "learning_rate": 4.9899720135330184e-05, "loss": 0.8129, "step": 28617 }, { "epoch": 0.03, "learning_rate": 4.9899713106924474e-05, "loss": 1.0336, "step": 28618 }, { "epoch": 0.03, "learning_rate": 4.9899706078272955e-05, "loss": 1.1311, "step": 28619 }, { "epoch": 0.03, "learning_rate": 4.989969904937565e-05, "loss": 1.2843, "step": 28620 }, { "epoch": 0.03, "learning_rate": 4.989969202023254e-05, "loss": 0.8608, "step": 28621 }, { "epoch": 0.03, "learning_rate": 4.9899684990843626e-05, "loss": 1.4016, "step": 28622 }, { "epoch": 0.03, "learning_rate": 4.989967796120892e-05, "loss": 1.157, "step": 28623 }, { "epoch": 0.03, "learning_rate": 4.989967093132841e-05, "loss": 1.0876, "step": 28624 }, { "epoch": 0.03, "learning_rate": 4.98996639012021e-05, "loss": 1.2017, "step": 28625 }, { "epoch": 0.03, "learning_rate": 4.989965687083e-05, "loss": 1.1174, "step": 28626 }, { "epoch": 0.03, "learning_rate": 4.9899649840212096e-05, "loss": 1.2353, "step": 28627 }, { "epoch": 0.03, "learning_rate": 4.989964280934839e-05, "loss": 1.0438, "step": 28628 }, { "epoch": 0.03, "learning_rate": 4.9899635778238884e-05, "loss": 0.5082, "step": 28629 }, { "epoch": 0.03, "learning_rate": 4.989962874688359e-05, "loss": 0.8514, "step": 28630 }, { "epoch": 0.03, "learning_rate": 4.9899621715282485e-05, "loss": 1.0902, "step": 28631 }, { "epoch": 0.03, "learning_rate": 4.989961468343558e-05, "loss": 1.0801, "step": 28632 }, { "epoch": 0.03, "learning_rate": 4.989960765134288e-05, "loss": 0.8906, "step": 28633 }, { "epoch": 0.03, "learning_rate": 4.989960061900439e-05, "loss": 0.7168, "step": 28634 }, { "epoch": 0.03, "learning_rate": 4.989959358642009e-05, "loss": 0.6165, "step": 28635 }, { "epoch": 0.03, "learning_rate": 4.9899586553590005e-05, "loss": 0.7574, "step": 28636 }, { "epoch": 0.03, "learning_rate": 4.9899579520514114e-05, "loss": 1.0258, "step": 28637 }, { "epoch": 0.03, "learning_rate": 4.989957248719242e-05, "loss": 0.9461, "step": 28638 }, { "epoch": 0.03, "learning_rate": 4.989956545362494e-05, "loss": 0.8615, "step": 28639 }, { "epoch": 0.03, "learning_rate": 4.989955841981165e-05, "loss": 0.8364, "step": 28640 }, { "epoch": 0.03, "learning_rate": 4.989955138575257e-05, "loss": 0.215, "step": 28641 }, { "epoch": 0.03, "learning_rate": 4.989954435144769e-05, "loss": 0.3466, "step": 28642 }, { "epoch": 0.03, "learning_rate": 4.989953731689701e-05, "loss": 1.0194, "step": 28643 }, { "epoch": 0.03, "learning_rate": 4.9899530282100536e-05, "loss": 1.0064, "step": 28644 }, { "epoch": 0.03, "learning_rate": 4.989952324705826e-05, "loss": 1.6024, "step": 28645 }, { "epoch": 0.03, "learning_rate": 4.989951621177019e-05, "loss": 1.121, "step": 28646 }, { "epoch": 0.03, "learning_rate": 4.989950917623632e-05, "loss": 1.0352, "step": 28647 }, { "epoch": 0.03, "learning_rate": 4.9899502140456654e-05, "loss": 0.8574, "step": 28648 }, { "epoch": 0.03, "learning_rate": 4.989949510443119e-05, "loss": 1.0526, "step": 28649 }, { "epoch": 0.03, "learning_rate": 4.989948806815993e-05, "loss": 2.3448, "step": 28650 }, { "epoch": 0.03, "learning_rate": 4.989948103164287e-05, "loss": 0.9199, "step": 28651 }, { "epoch": 0.03, "learning_rate": 4.989947399488001e-05, "loss": 0.9277, "step": 28652 }, { "epoch": 0.03, "learning_rate": 4.989946695787137e-05, "loss": 1.2516, "step": 28653 }, { "epoch": 0.03, "learning_rate": 4.989945992061691e-05, "loss": 1.3325, "step": 28654 }, { "epoch": 0.03, "learning_rate": 4.989945288311667e-05, "loss": 1.1112, "step": 28655 }, { "epoch": 0.03, "learning_rate": 4.9899445845370616e-05, "loss": 1.0724, "step": 28656 }, { "epoch": 0.03, "learning_rate": 4.9899438807378784e-05, "loss": 1.0268, "step": 28657 }, { "epoch": 0.03, "learning_rate": 4.989943176914115e-05, "loss": 0.7917, "step": 28658 }, { "epoch": 0.03, "learning_rate": 4.9899424730657706e-05, "loss": 1.1986, "step": 28659 }, { "epoch": 0.03, "learning_rate": 4.989941769192848e-05, "loss": 1.1738, "step": 28660 }, { "epoch": 0.03, "learning_rate": 4.9899410652953446e-05, "loss": 0.916, "step": 28661 }, { "epoch": 0.03, "learning_rate": 4.989940361373263e-05, "loss": 0.8998, "step": 28662 }, { "epoch": 0.03, "learning_rate": 4.989939657426601e-05, "loss": 0.8212, "step": 28663 }, { "epoch": 0.03, "learning_rate": 4.989938953455359e-05, "loss": 0.8998, "step": 28664 }, { "epoch": 0.03, "learning_rate": 4.989938249459538e-05, "loss": 1.2043, "step": 28665 }, { "epoch": 0.03, "learning_rate": 4.989937545439137e-05, "loss": 1.2122, "step": 28666 }, { "epoch": 0.03, "learning_rate": 4.9899368413941564e-05, "loss": 0.9562, "step": 28667 }, { "epoch": 0.03, "learning_rate": 4.989936137324597e-05, "loss": 1.1825, "step": 28668 }, { "epoch": 0.03, "learning_rate": 4.989935433230457e-05, "loss": 1.4023, "step": 28669 }, { "epoch": 0.03, "learning_rate": 4.989934729111737e-05, "loss": 1.3786, "step": 28670 }, { "epoch": 0.03, "learning_rate": 4.989934024968438e-05, "loss": 1.295, "step": 28671 }, { "epoch": 0.03, "learning_rate": 4.98993332080056e-05, "loss": 1.1711, "step": 28672 }, { "epoch": 0.03, "learning_rate": 4.989932616608102e-05, "loss": 1.0478, "step": 28673 }, { "epoch": 0.03, "learning_rate": 4.989931912391065e-05, "loss": 0.9863, "step": 28674 }, { "epoch": 0.03, "learning_rate": 4.9899312081494474e-05, "loss": 1.2418, "step": 28675 }, { "epoch": 0.03, "learning_rate": 4.989930503883251e-05, "loss": 1.0336, "step": 28676 }, { "epoch": 0.03, "learning_rate": 4.989929799592474e-05, "loss": 1.6233, "step": 28677 }, { "epoch": 0.03, "learning_rate": 4.9899290952771185e-05, "loss": 1.2143, "step": 28678 }, { "epoch": 0.03, "learning_rate": 4.989928390937183e-05, "loss": 0.4854, "step": 28679 }, { "epoch": 0.03, "learning_rate": 4.9899276865726686e-05, "loss": 0.392, "step": 28680 }, { "epoch": 0.03, "learning_rate": 4.989926982183574e-05, "loss": 0.5483, "step": 28681 }, { "epoch": 0.03, "learning_rate": 4.9899262777699e-05, "loss": 0.9352, "step": 28682 }, { "epoch": 0.03, "learning_rate": 4.9899255733316465e-05, "loss": 0.9551, "step": 28683 }, { "epoch": 0.03, "learning_rate": 4.989924868868814e-05, "loss": 1.1013, "step": 28684 }, { "epoch": 0.03, "learning_rate": 4.989924164381401e-05, "loss": 0.8728, "step": 28685 }, { "epoch": 0.03, "learning_rate": 4.989923459869409e-05, "loss": 0.9696, "step": 28686 }, { "epoch": 0.03, "learning_rate": 4.9899227553328385e-05, "loss": 0.7358, "step": 28687 }, { "epoch": 0.03, "learning_rate": 4.989922050771687e-05, "loss": 1.0078, "step": 28688 }, { "epoch": 0.03, "learning_rate": 4.9899213461859575e-05, "loss": 1.0475, "step": 28689 }, { "epoch": 0.03, "learning_rate": 4.9899206415756474e-05, "loss": 0.9979, "step": 28690 }, { "epoch": 0.03, "learning_rate": 4.989919936940758e-05, "loss": 1.2319, "step": 28691 }, { "epoch": 0.03, "learning_rate": 4.989919232281289e-05, "loss": 1.2622, "step": 28692 }, { "epoch": 0.03, "learning_rate": 4.989918527597242e-05, "loss": 1.2747, "step": 28693 }, { "epoch": 0.03, "learning_rate": 4.989917822888614e-05, "loss": 1.1397, "step": 28694 }, { "epoch": 0.03, "learning_rate": 4.989917118155407e-05, "loss": 1.1862, "step": 28695 }, { "epoch": 0.03, "learning_rate": 4.9899164133976204e-05, "loss": 1.2062, "step": 28696 }, { "epoch": 0.03, "learning_rate": 4.9899157086152555e-05, "loss": 1.1237, "step": 28697 }, { "epoch": 0.03, "learning_rate": 4.98991500380831e-05, "loss": 0.841, "step": 28698 }, { "epoch": 0.03, "learning_rate": 4.989914298976786e-05, "loss": 1.1708, "step": 28699 }, { "epoch": 0.03, "learning_rate": 4.989913594120682e-05, "loss": 0.9753, "step": 28700 }, { "epoch": 0.03, "learning_rate": 4.989912889239998e-05, "loss": 1.0414, "step": 28701 }, { "epoch": 0.03, "learning_rate": 4.989912184334736e-05, "loss": 0.9227, "step": 28702 }, { "epoch": 0.03, "learning_rate": 4.989911479404894e-05, "loss": 1.1541, "step": 28703 }, { "epoch": 0.03, "learning_rate": 4.989910774450472e-05, "loss": 1.096, "step": 28704 }, { "epoch": 0.03, "learning_rate": 4.989910069471472e-05, "loss": 0.8285, "step": 28705 }, { "epoch": 0.03, "learning_rate": 4.9899093644678916e-05, "loss": 1.3583, "step": 28706 }, { "epoch": 0.03, "learning_rate": 4.989908659439733e-05, "loss": 1.4777, "step": 28707 }, { "epoch": 0.03, "learning_rate": 4.9899079543869936e-05, "loss": 1.2747, "step": 28708 }, { "epoch": 0.03, "learning_rate": 4.9899072493096755e-05, "loss": 1.1566, "step": 28709 }, { "epoch": 0.03, "learning_rate": 4.989906544207779e-05, "loss": 0.9429, "step": 28710 }, { "epoch": 0.03, "learning_rate": 4.9899058390813016e-05, "loss": 0.877, "step": 28711 }, { "epoch": 0.03, "learning_rate": 4.989905133930246e-05, "loss": 0.8357, "step": 28712 }, { "epoch": 0.03, "learning_rate": 4.98990442875461e-05, "loss": 1.1905, "step": 28713 }, { "epoch": 0.03, "learning_rate": 4.989903723554396e-05, "loss": 1.1011, "step": 28714 }, { "epoch": 0.03, "learning_rate": 4.9899030183296026e-05, "loss": 1.0383, "step": 28715 }, { "epoch": 0.03, "learning_rate": 4.989902313080229e-05, "loss": 1.2375, "step": 28716 }, { "epoch": 0.03, "learning_rate": 4.989901607806277e-05, "loss": 1.1616, "step": 28717 }, { "epoch": 0.03, "learning_rate": 4.9899009025077456e-05, "loss": 1.1348, "step": 28718 }, { "epoch": 0.03, "learning_rate": 4.989900197184634e-05, "loss": 1.0037, "step": 28719 }, { "epoch": 0.03, "learning_rate": 4.989899491836945e-05, "loss": 0.648, "step": 28720 }, { "epoch": 0.03, "learning_rate": 4.989898786464675e-05, "loss": 1.338, "step": 28721 }, { "epoch": 0.03, "learning_rate": 4.989898081067826e-05, "loss": 1.1413, "step": 28722 }, { "epoch": 0.03, "learning_rate": 4.989897375646399e-05, "loss": 1.0504, "step": 28723 }, { "epoch": 0.03, "learning_rate": 4.989896670200391e-05, "loss": 1.1597, "step": 28724 }, { "epoch": 0.03, "learning_rate": 4.989895964729805e-05, "loss": 1.1147, "step": 28725 }, { "epoch": 0.03, "learning_rate": 4.9898952592346396e-05, "loss": 1.2063, "step": 28726 }, { "epoch": 0.03, "learning_rate": 4.989894553714895e-05, "loss": 0.9647, "step": 28727 }, { "epoch": 0.03, "learning_rate": 4.9898938481705714e-05, "loss": 1.1278, "step": 28728 }, { "epoch": 0.03, "learning_rate": 4.989893142601668e-05, "loss": 0.3865, "step": 28729 }, { "epoch": 0.03, "learning_rate": 4.9898924370081864e-05, "loss": 0.858, "step": 28730 }, { "epoch": 0.03, "learning_rate": 4.989891731390125e-05, "loss": 1.045, "step": 28731 }, { "epoch": 0.03, "learning_rate": 4.989891025747484e-05, "loss": 0.9492, "step": 28732 }, { "epoch": 0.03, "learning_rate": 4.989890320080265e-05, "loss": 0.962, "step": 28733 }, { "epoch": 0.03, "learning_rate": 4.989889614388465e-05, "loss": 1.3121, "step": 28734 }, { "epoch": 0.03, "learning_rate": 4.989888908672088e-05, "loss": 1.5073, "step": 28735 }, { "epoch": 0.03, "learning_rate": 4.989888202931131e-05, "loss": 1.647, "step": 28736 }, { "epoch": 0.03, "learning_rate": 4.989887497165594e-05, "loss": 0.9382, "step": 28737 }, { "epoch": 0.03, "learning_rate": 4.989886791375479e-05, "loss": 1.0809, "step": 28738 }, { "epoch": 0.03, "learning_rate": 4.989886085560785e-05, "loss": 1.5419, "step": 28739 }, { "epoch": 0.03, "learning_rate": 4.9898853797215113e-05, "loss": 1.507, "step": 28740 }, { "epoch": 0.03, "learning_rate": 4.989884673857659e-05, "loss": 0.8779, "step": 28741 }, { "epoch": 0.03, "learning_rate": 4.989883967969227e-05, "loss": 0.5897, "step": 28742 }, { "epoch": 0.03, "learning_rate": 4.9898832620562166e-05, "loss": 0.5506, "step": 28743 }, { "epoch": 0.03, "learning_rate": 4.9898825561186256e-05, "loss": 0.5394, "step": 28744 }, { "epoch": 0.03, "learning_rate": 4.989881850156457e-05, "loss": 1.1847, "step": 28745 }, { "epoch": 0.03, "learning_rate": 4.989881144169709e-05, "loss": 1.1404, "step": 28746 }, { "epoch": 0.03, "learning_rate": 4.989880438158382e-05, "loss": 0.9296, "step": 28747 }, { "epoch": 0.03, "learning_rate": 4.989879732122475e-05, "loss": 0.8682, "step": 28748 }, { "epoch": 0.03, "learning_rate": 4.9898790260619905e-05, "loss": 1.2293, "step": 28749 }, { "epoch": 0.03, "learning_rate": 4.989878319976926e-05, "loss": 1.2358, "step": 28750 }, { "epoch": 0.03, "learning_rate": 4.989877613867283e-05, "loss": 1.027, "step": 28751 }, { "epoch": 0.03, "learning_rate": 4.989876907733061e-05, "loss": 0.8733, "step": 28752 }, { "epoch": 0.03, "learning_rate": 4.9898762015742595e-05, "loss": 0.8829, "step": 28753 }, { "epoch": 0.03, "learning_rate": 4.989875495390879e-05, "loss": 0.8916, "step": 28754 }, { "epoch": 0.03, "learning_rate": 4.98987478918292e-05, "loss": 0.8475, "step": 28755 }, { "epoch": 0.03, "learning_rate": 4.989874082950382e-05, "loss": 0.7215, "step": 28756 }, { "epoch": 0.03, "learning_rate": 4.989873376693264e-05, "loss": 0.9166, "step": 28757 }, { "epoch": 0.03, "learning_rate": 4.989872670411568e-05, "loss": 1.3159, "step": 28758 }, { "epoch": 0.03, "learning_rate": 4.989871964105293e-05, "loss": 1.1122, "step": 28759 }, { "epoch": 0.03, "learning_rate": 4.989871257774439e-05, "loss": 1.1782, "step": 28760 }, { "epoch": 0.03, "learning_rate": 4.9898705514190054e-05, "loss": 1.2824, "step": 28761 }, { "epoch": 0.03, "learning_rate": 4.9898698450389934e-05, "loss": 0.8737, "step": 28762 }, { "epoch": 0.03, "learning_rate": 4.989869138634402e-05, "loss": 1.5407, "step": 28763 }, { "epoch": 0.03, "learning_rate": 4.989868432205233e-05, "loss": 1.0472, "step": 28764 }, { "epoch": 0.03, "learning_rate": 4.989867725751483e-05, "loss": 0.9275, "step": 28765 }, { "epoch": 0.03, "learning_rate": 4.9898670192731554e-05, "loss": 1.0443, "step": 28766 }, { "epoch": 0.03, "learning_rate": 4.9898663127702484e-05, "loss": 0.9821, "step": 28767 }, { "epoch": 0.03, "learning_rate": 4.9898656062427626e-05, "loss": 1.0266, "step": 28768 }, { "epoch": 0.03, "learning_rate": 4.9898648996906986e-05, "loss": 1.3478, "step": 28769 }, { "epoch": 0.03, "learning_rate": 4.989864193114054e-05, "loss": 1.7217, "step": 28770 }, { "epoch": 0.03, "learning_rate": 4.9898634865128326e-05, "loss": 1.4757, "step": 28771 }, { "epoch": 0.03, "learning_rate": 4.989862779887031e-05, "loss": 0.886, "step": 28772 }, { "epoch": 0.03, "learning_rate": 4.989862073236651e-05, "loss": 0.6483, "step": 28773 }, { "epoch": 0.03, "learning_rate": 4.9898613665616914e-05, "loss": 0.7847, "step": 28774 }, { "epoch": 0.03, "learning_rate": 4.989860659862154e-05, "loss": 0.9859, "step": 28775 }, { "epoch": 0.03, "learning_rate": 4.9898599531380374e-05, "loss": 0.8349, "step": 28776 }, { "epoch": 0.03, "learning_rate": 4.989859246389341e-05, "loss": 1.1458, "step": 28777 }, { "epoch": 0.03, "learning_rate": 4.9898585396160666e-05, "loss": 0.9526, "step": 28778 }, { "epoch": 0.03, "learning_rate": 4.989857832818214e-05, "loss": 0.8921, "step": 28779 }, { "epoch": 0.03, "learning_rate": 4.989857125995782e-05, "loss": 0.6175, "step": 28780 }, { "epoch": 0.03, "learning_rate": 4.989856419148771e-05, "loss": 0.7263, "step": 28781 }, { "epoch": 0.03, "learning_rate": 4.989855712277181e-05, "loss": 1.053, "step": 28782 }, { "epoch": 0.03, "learning_rate": 4.989855005381012e-05, "loss": 0.8914, "step": 28783 }, { "epoch": 0.03, "learning_rate": 4.9898542984602656e-05, "loss": 0.9494, "step": 28784 }, { "epoch": 0.03, "learning_rate": 4.989853591514939e-05, "loss": 1.071, "step": 28785 }, { "epoch": 0.03, "learning_rate": 4.9898528845450344e-05, "loss": 1.0856, "step": 28786 }, { "epoch": 0.03, "learning_rate": 4.98985217755055e-05, "loss": 0.908, "step": 28787 }, { "epoch": 0.03, "learning_rate": 4.989851470531488e-05, "loss": 0.9302, "step": 28788 }, { "epoch": 0.03, "learning_rate": 4.9898507634878464e-05, "loss": 1.0564, "step": 28789 }, { "epoch": 0.03, "learning_rate": 4.989850056419627e-05, "loss": 1.0, "step": 28790 }, { "epoch": 0.03, "learning_rate": 4.9898493493268286e-05, "loss": 0.8991, "step": 28791 }, { "epoch": 0.03, "learning_rate": 4.989848642209451e-05, "loss": 0.8118, "step": 28792 }, { "epoch": 0.03, "learning_rate": 4.989847935067494e-05, "loss": 0.9091, "step": 28793 }, { "epoch": 0.03, "learning_rate": 4.98984722790096e-05, "loss": 1.3065, "step": 28794 }, { "epoch": 0.03, "learning_rate": 4.989846520709846e-05, "loss": 0.9606, "step": 28795 }, { "epoch": 0.03, "learning_rate": 4.989845813494154e-05, "loss": 1.0161, "step": 28796 }, { "epoch": 0.03, "learning_rate": 4.9898451062538824e-05, "loss": 1.1581, "step": 28797 }, { "epoch": 0.03, "learning_rate": 4.9898443989890327e-05, "loss": 1.1614, "step": 28798 }, { "epoch": 0.03, "learning_rate": 4.989843691699604e-05, "loss": 0.9223, "step": 28799 }, { "epoch": 0.03, "learning_rate": 4.989842984385597e-05, "loss": 0.9396, "step": 28800 }, { "epoch": 0.03, "learning_rate": 4.989842277047011e-05, "loss": 0.8434, "step": 28801 }, { "epoch": 0.03, "learning_rate": 4.9898415696838465e-05, "loss": 0.9033, "step": 28802 }, { "epoch": 0.03, "learning_rate": 4.989840862296103e-05, "loss": 0.998, "step": 28803 }, { "epoch": 0.03, "learning_rate": 4.9898401548837816e-05, "loss": 0.5246, "step": 28804 }, { "epoch": 0.03, "learning_rate": 4.989839447446881e-05, "loss": 0.834, "step": 28805 }, { "epoch": 0.03, "learning_rate": 4.989838739985402e-05, "loss": 1.177, "step": 28806 }, { "epoch": 0.03, "learning_rate": 4.9898380324993445e-05, "loss": 1.0343, "step": 28807 }, { "epoch": 0.03, "learning_rate": 4.9898373249887075e-05, "loss": 0.9242, "step": 28808 }, { "epoch": 0.03, "learning_rate": 4.989836617453493e-05, "loss": 1.132, "step": 28809 }, { "epoch": 0.03, "learning_rate": 4.989835909893698e-05, "loss": 0.9288, "step": 28810 }, { "epoch": 0.03, "learning_rate": 4.989835202309327e-05, "loss": 1.4192, "step": 28811 }, { "epoch": 0.03, "learning_rate": 4.9898344947003764e-05, "loss": 0.874, "step": 28812 }, { "epoch": 0.03, "learning_rate": 4.9898337870668464e-05, "loss": 1.0229, "step": 28813 }, { "epoch": 0.03, "learning_rate": 4.989833079408738e-05, "loss": 1.0179, "step": 28814 }, { "epoch": 0.03, "learning_rate": 4.989832371726052e-05, "loss": 1.1, "step": 28815 }, { "epoch": 0.03, "learning_rate": 4.989831664018787e-05, "loss": 0.829, "step": 28816 }, { "epoch": 0.03, "learning_rate": 4.989830956286943e-05, "loss": 1.0219, "step": 28817 }, { "epoch": 0.03, "learning_rate": 4.9898302485305205e-05, "loss": 0.4041, "step": 28818 }, { "epoch": 0.03, "learning_rate": 4.9898295407495194e-05, "loss": 0.6985, "step": 28819 }, { "epoch": 0.03, "learning_rate": 4.98982883294394e-05, "loss": 0.8285, "step": 28820 }, { "epoch": 0.03, "learning_rate": 4.989828125113782e-05, "loss": 1.019, "step": 28821 }, { "epoch": 0.03, "learning_rate": 4.9898274172590456e-05, "loss": 1.0239, "step": 28822 }, { "epoch": 0.03, "learning_rate": 4.9898267093797304e-05, "loss": 1.0249, "step": 28823 }, { "epoch": 0.03, "learning_rate": 4.989826001475837e-05, "loss": 1.2123, "step": 28824 }, { "epoch": 0.03, "learning_rate": 4.989825293547364e-05, "loss": 1.2797, "step": 28825 }, { "epoch": 0.03, "learning_rate": 4.989824585594314e-05, "loss": 1.1551, "step": 28826 }, { "epoch": 0.03, "learning_rate": 4.989823877616685e-05, "loss": 1.0152, "step": 28827 }, { "epoch": 0.03, "learning_rate": 4.989823169614477e-05, "loss": 0.6396, "step": 28828 }, { "epoch": 0.03, "learning_rate": 4.989822461587691e-05, "loss": 1.0515, "step": 28829 }, { "epoch": 0.03, "learning_rate": 4.989821753536327e-05, "loss": 0.978, "step": 28830 }, { "epoch": 0.03, "learning_rate": 4.989821045460383e-05, "loss": 0.6876, "step": 28831 }, { "epoch": 0.03, "learning_rate": 4.989820337359862e-05, "loss": 0.9216, "step": 28832 }, { "epoch": 0.03, "learning_rate": 4.989819629234762e-05, "loss": 0.9109, "step": 28833 }, { "epoch": 0.03, "learning_rate": 4.989818921085084e-05, "loss": 0.6072, "step": 28834 }, { "epoch": 0.03, "learning_rate": 4.989818212910827e-05, "loss": 1.0917, "step": 28835 }, { "epoch": 0.03, "learning_rate": 4.9898175047119915e-05, "loss": 1.0272, "step": 28836 }, { "epoch": 0.03, "learning_rate": 4.989816796488578e-05, "loss": 0.8012, "step": 28837 }, { "epoch": 0.03, "learning_rate": 4.9898160882405856e-05, "loss": 0.9323, "step": 28838 }, { "epoch": 0.03, "learning_rate": 4.9898153799680154e-05, "loss": 1.0394, "step": 28839 }, { "epoch": 0.03, "learning_rate": 4.989814671670866e-05, "loss": 1.4442, "step": 28840 }, { "epoch": 0.03, "learning_rate": 4.989813963349139e-05, "loss": 0.9732, "step": 28841 }, { "epoch": 0.03, "learning_rate": 4.989813255002833e-05, "loss": 0.3551, "step": 28842 }, { "epoch": 0.03, "learning_rate": 4.989812546631949e-05, "loss": 0.3674, "step": 28843 }, { "epoch": 0.03, "learning_rate": 4.989811838236487e-05, "loss": 0.3411, "step": 28844 }, { "epoch": 0.03, "learning_rate": 4.989811129816446e-05, "loss": 0.4558, "step": 28845 }, { "epoch": 0.03, "learning_rate": 4.9898104213718264e-05, "loss": 0.2535, "step": 28846 }, { "epoch": 0.03, "learning_rate": 4.989809712902629e-05, "loss": 0.1645, "step": 28847 }, { "epoch": 0.03, "learning_rate": 4.989809004408853e-05, "loss": 0.0995, "step": 28848 }, { "epoch": 0.03, "learning_rate": 4.9898082958904986e-05, "loss": 0.072, "step": 28849 }, { "epoch": 0.03, "learning_rate": 4.9898075873475657e-05, "loss": 0.0719, "step": 28850 }, { "epoch": 0.03, "learning_rate": 4.989806878780055e-05, "loss": 0.0329, "step": 28851 }, { "epoch": 0.03, "learning_rate": 4.989806170187966e-05, "loss": 0.0611, "step": 28852 }, { "epoch": 0.03, "learning_rate": 4.989805461571298e-05, "loss": 0.0598, "step": 28853 }, { "epoch": 0.03, "learning_rate": 4.989804752930052e-05, "loss": 0.0613, "step": 28854 }, { "epoch": 0.03, "learning_rate": 4.989804044264228e-05, "loss": 0.198, "step": 28855 }, { "epoch": 0.03, "learning_rate": 4.989803335573825e-05, "loss": 0.0957, "step": 28856 }, { "epoch": 0.03, "learning_rate": 4.989802626858844e-05, "loss": 0.049, "step": 28857 }, { "epoch": 0.03, "learning_rate": 4.989801918119286e-05, "loss": 0.0785, "step": 28858 }, { "epoch": 0.03, "learning_rate": 4.9898012093551484e-05, "loss": 1.0949, "step": 28859 }, { "epoch": 0.03, "learning_rate": 4.989800500566432e-05, "loss": 1.0429, "step": 28860 }, { "epoch": 0.03, "learning_rate": 4.989799791753138e-05, "loss": 1.2065, "step": 28861 }, { "epoch": 0.03, "learning_rate": 4.9897990829152666e-05, "loss": 1.178, "step": 28862 }, { "epoch": 0.03, "learning_rate": 4.989798374052816e-05, "loss": 0.927, "step": 28863 }, { "epoch": 0.03, "learning_rate": 4.9897976651657876e-05, "loss": 1.1034, "step": 28864 }, { "epoch": 0.03, "learning_rate": 4.9897969562541805e-05, "loss": 0.6125, "step": 28865 }, { "epoch": 0.03, "learning_rate": 4.989796247317995e-05, "loss": 0.9577, "step": 28866 }, { "epoch": 0.03, "learning_rate": 4.989795538357232e-05, "loss": 0.7278, "step": 28867 }, { "epoch": 0.03, "learning_rate": 4.98979482937189e-05, "loss": 0.3926, "step": 28868 }, { "epoch": 0.03, "learning_rate": 4.98979412036197e-05, "loss": 0.3831, "step": 28869 }, { "epoch": 0.03, "learning_rate": 4.989793411327473e-05, "loss": 0.9334, "step": 28870 }, { "epoch": 0.03, "learning_rate": 4.9897927022683965e-05, "loss": 1.0732, "step": 28871 }, { "epoch": 0.03, "learning_rate": 4.989791993184742e-05, "loss": 0.9936, "step": 28872 }, { "epoch": 0.03, "learning_rate": 4.9897912840765095e-05, "loss": 1.0296, "step": 28873 }, { "epoch": 0.03, "learning_rate": 4.9897905749436994e-05, "loss": 0.9023, "step": 28874 }, { "epoch": 0.03, "learning_rate": 4.9897898657863104e-05, "loss": 1.0704, "step": 28875 }, { "epoch": 0.03, "learning_rate": 4.989789156604343e-05, "loss": 1.033, "step": 28876 }, { "epoch": 0.03, "learning_rate": 4.989788447397798e-05, "loss": 0.9554, "step": 28877 }, { "epoch": 0.03, "learning_rate": 4.989787738166674e-05, "loss": 1.0071, "step": 28878 }, { "epoch": 0.03, "learning_rate": 4.989787028910973e-05, "loss": 1.0048, "step": 28879 }, { "epoch": 0.03, "learning_rate": 4.989786319630694e-05, "loss": 0.5843, "step": 28880 }, { "epoch": 0.03, "learning_rate": 4.989785610325836e-05, "loss": 0.3442, "step": 28881 }, { "epoch": 0.03, "learning_rate": 4.9897849009964004e-05, "loss": 0.4914, "step": 28882 }, { "epoch": 0.03, "learning_rate": 4.989784191642386e-05, "loss": 0.3567, "step": 28883 }, { "epoch": 0.03, "learning_rate": 4.9897834822637946e-05, "loss": 0.3282, "step": 28884 }, { "epoch": 0.03, "learning_rate": 4.989782772860624e-05, "loss": 0.307, "step": 28885 }, { "epoch": 0.03, "learning_rate": 4.989782063432876e-05, "loss": 0.369, "step": 28886 }, { "epoch": 0.03, "learning_rate": 4.98978135398055e-05, "loss": 0.5162, "step": 28887 }, { "epoch": 0.03, "learning_rate": 4.9897806445036455e-05, "loss": 1.1952, "step": 28888 }, { "epoch": 0.03, "learning_rate": 4.989779935002163e-05, "loss": 1.0199, "step": 28889 }, { "epoch": 0.03, "learning_rate": 4.989779225476103e-05, "loss": 0.9132, "step": 28890 }, { "epoch": 0.03, "learning_rate": 4.989778515925464e-05, "loss": 1.1699, "step": 28891 }, { "epoch": 0.03, "learning_rate": 4.9897778063502475e-05, "loss": 0.8231, "step": 28892 }, { "epoch": 0.03, "learning_rate": 4.989777096750453e-05, "loss": 1.1207, "step": 28893 }, { "epoch": 0.03, "learning_rate": 4.989776387126081e-05, "loss": 1.0462, "step": 28894 }, { "epoch": 0.03, "learning_rate": 4.98977567747713e-05, "loss": 0.434, "step": 28895 }, { "epoch": 0.03, "learning_rate": 4.989774967803601e-05, "loss": 0.2738, "step": 28896 }, { "epoch": 0.03, "learning_rate": 4.989774258105495e-05, "loss": 0.8974, "step": 28897 }, { "epoch": 0.03, "learning_rate": 4.9897735483828104e-05, "loss": 1.1216, "step": 28898 }, { "epoch": 0.03, "learning_rate": 4.989772838635547e-05, "loss": 1.1644, "step": 28899 }, { "epoch": 0.03, "learning_rate": 4.989772128863707e-05, "loss": 1.8721, "step": 28900 }, { "epoch": 0.03, "learning_rate": 4.9897714190672876e-05, "loss": 1.4062, "step": 28901 }, { "epoch": 0.03, "learning_rate": 4.989770709246292e-05, "loss": 1.3411, "step": 28902 }, { "epoch": 0.03, "learning_rate": 4.989769999400717e-05, "loss": 0.4413, "step": 28903 }, { "epoch": 0.03, "learning_rate": 4.989769289530564e-05, "loss": 1.0114, "step": 28904 }, { "epoch": 0.03, "learning_rate": 4.989768579635834e-05, "loss": 1.4632, "step": 28905 }, { "epoch": 0.03, "learning_rate": 4.989767869716525e-05, "loss": 0.9671, "step": 28906 }, { "epoch": 0.03, "learning_rate": 4.989767159772639e-05, "loss": 0.9991, "step": 28907 }, { "epoch": 0.03, "learning_rate": 4.989766449804175e-05, "loss": 0.0565, "step": 28908 }, { "epoch": 0.03, "learning_rate": 4.9897657398111324e-05, "loss": 0.1174, "step": 28909 }, { "epoch": 0.03, "learning_rate": 4.989765029793512e-05, "loss": 0.0573, "step": 28910 }, { "epoch": 0.03, "learning_rate": 4.989764319751315e-05, "loss": 0.793, "step": 28911 }, { "epoch": 0.03, "learning_rate": 4.9897636096845376e-05, "loss": 0.9251, "step": 28912 }, { "epoch": 0.03, "learning_rate": 4.9897628995931844e-05, "loss": 1.1071, "step": 28913 }, { "epoch": 0.03, "learning_rate": 4.989762189477253e-05, "loss": 1.2617, "step": 28914 }, { "epoch": 0.03, "learning_rate": 4.989761479336743e-05, "loss": 0.634, "step": 28915 }, { "epoch": 0.03, "learning_rate": 4.9897607691716555e-05, "loss": 0.9604, "step": 28916 }, { "epoch": 0.03, "learning_rate": 4.98976005898199e-05, "loss": 0.9846, "step": 28917 }, { "epoch": 0.03, "learning_rate": 4.989759348767747e-05, "loss": 1.0023, "step": 28918 }, { "epoch": 0.03, "learning_rate": 4.9897586385289257e-05, "loss": 0.9485, "step": 28919 }, { "epoch": 0.03, "learning_rate": 4.989757928265527e-05, "loss": 1.1136, "step": 28920 }, { "epoch": 0.03, "learning_rate": 4.98975721797755e-05, "loss": 0.798, "step": 28921 }, { "epoch": 0.03, "learning_rate": 4.9897565076649955e-05, "loss": 0.9756, "step": 28922 }, { "epoch": 0.03, "learning_rate": 4.989755797327863e-05, "loss": 1.1705, "step": 28923 }, { "epoch": 0.03, "learning_rate": 4.9897550869661525e-05, "loss": 0.9846, "step": 28924 }, { "epoch": 0.03, "learning_rate": 4.989754376579865e-05, "loss": 0.6421, "step": 28925 }, { "epoch": 0.03, "learning_rate": 4.9897536661689995e-05, "loss": 1.0426, "step": 28926 }, { "epoch": 0.03, "learning_rate": 4.989752955733556e-05, "loss": 1.1374, "step": 28927 }, { "epoch": 0.03, "learning_rate": 4.989752245273534e-05, "loss": 0.9853, "step": 28928 }, { "epoch": 0.03, "learning_rate": 4.989751534788934e-05, "loss": 0.7698, "step": 28929 }, { "epoch": 0.03, "learning_rate": 4.989750824279757e-05, "loss": 0.9847, "step": 28930 }, { "epoch": 0.03, "learning_rate": 4.989750113746002e-05, "loss": 1.0451, "step": 28931 }, { "epoch": 0.03, "learning_rate": 4.98974940318767e-05, "loss": 1.205, "step": 28932 }, { "epoch": 0.03, "learning_rate": 4.98974869260476e-05, "loss": 0.7799, "step": 28933 }, { "epoch": 0.03, "learning_rate": 4.9897479819972715e-05, "loss": 1.0969, "step": 28934 }, { "epoch": 0.03, "learning_rate": 4.989747271365206e-05, "loss": 1.1205, "step": 28935 }, { "epoch": 0.03, "learning_rate": 4.989746560708562e-05, "loss": 1.1791, "step": 28936 }, { "epoch": 0.03, "learning_rate": 4.9897458500273406e-05, "loss": 0.8873, "step": 28937 }, { "epoch": 0.03, "learning_rate": 4.989745139321542e-05, "loss": 1.1595, "step": 28938 }, { "epoch": 0.03, "learning_rate": 4.9897444285911656e-05, "loss": 0.8781, "step": 28939 }, { "epoch": 0.03, "learning_rate": 4.989743717836211e-05, "loss": 1.0265, "step": 28940 }, { "epoch": 0.03, "learning_rate": 4.9897430070566786e-05, "loss": 1.5526, "step": 28941 }, { "epoch": 0.03, "learning_rate": 4.989742296252569e-05, "loss": 0.9612, "step": 28942 }, { "epoch": 0.03, "learning_rate": 4.9897415854238814e-05, "loss": 0.8811, "step": 28943 }, { "epoch": 0.03, "learning_rate": 4.9897408745706166e-05, "loss": 1.1159, "step": 28944 }, { "epoch": 0.03, "learning_rate": 4.989740163692773e-05, "loss": 0.9269, "step": 28945 }, { "epoch": 0.03, "learning_rate": 4.989739452790353e-05, "loss": 1.131, "step": 28946 }, { "epoch": 0.03, "learning_rate": 4.9897387418633544e-05, "loss": 0.9195, "step": 28947 }, { "epoch": 0.03, "learning_rate": 4.989738030911779e-05, "loss": 0.8737, "step": 28948 }, { "epoch": 0.03, "learning_rate": 4.989737319935625e-05, "loss": 0.6232, "step": 28949 }, { "epoch": 0.03, "learning_rate": 4.989736608934894e-05, "loss": 1.0677, "step": 28950 }, { "epoch": 0.03, "learning_rate": 4.989735897909585e-05, "loss": 1.4946, "step": 28951 }, { "epoch": 0.03, "learning_rate": 4.9897351868596983e-05, "loss": 1.1452, "step": 28952 }, { "epoch": 0.03, "learning_rate": 4.9897344757852346e-05, "loss": 0.9629, "step": 28953 }, { "epoch": 0.03, "learning_rate": 4.9897337646861933e-05, "loss": 1.0514, "step": 28954 }, { "epoch": 0.03, "learning_rate": 4.989733053562574e-05, "loss": 1.3857, "step": 28955 }, { "epoch": 0.03, "learning_rate": 4.989732342414377e-05, "loss": 1.0388, "step": 28956 }, { "epoch": 0.03, "learning_rate": 4.989731631241603e-05, "loss": 0.9176, "step": 28957 }, { "epoch": 0.03, "learning_rate": 4.989730920044251e-05, "loss": 0.8916, "step": 28958 }, { "epoch": 0.03, "learning_rate": 4.989730208822321e-05, "loss": 1.6187, "step": 28959 }, { "epoch": 0.03, "learning_rate": 4.989729497575815e-05, "loss": 1.1341, "step": 28960 }, { "epoch": 0.03, "learning_rate": 4.989728786304729e-05, "loss": 1.3322, "step": 28961 }, { "epoch": 0.03, "learning_rate": 4.9897280750090676e-05, "loss": 1.1932, "step": 28962 }, { "epoch": 0.03, "learning_rate": 4.9897273636888275e-05, "loss": 1.1081, "step": 28963 }, { "epoch": 0.03, "learning_rate": 4.98972665234401e-05, "loss": 1.035, "step": 28964 }, { "epoch": 0.03, "learning_rate": 4.989725940974615e-05, "loss": 0.8887, "step": 28965 }, { "epoch": 0.03, "learning_rate": 4.989725229580643e-05, "loss": 0.6033, "step": 28966 }, { "epoch": 0.03, "learning_rate": 4.989724518162093e-05, "loss": 1.0705, "step": 28967 }, { "epoch": 0.03, "learning_rate": 4.989723806718965e-05, "loss": 1.3072, "step": 28968 }, { "epoch": 0.03, "learning_rate": 4.989723095251261e-05, "loss": 0.5436, "step": 28969 }, { "epoch": 0.03, "learning_rate": 4.989722383758979e-05, "loss": 0.637, "step": 28970 }, { "epoch": 0.03, "learning_rate": 4.9897216722421185e-05, "loss": 0.6695, "step": 28971 }, { "epoch": 0.03, "learning_rate": 4.9897209607006814e-05, "loss": 1.0799, "step": 28972 }, { "epoch": 0.03, "learning_rate": 4.989720249134666e-05, "loss": 1.1542, "step": 28973 }, { "epoch": 0.03, "learning_rate": 4.989719537544074e-05, "loss": 1.1452, "step": 28974 }, { "epoch": 0.03, "learning_rate": 4.989718825928904e-05, "loss": 0.9649, "step": 28975 }, { "epoch": 0.03, "learning_rate": 4.989718114289157e-05, "loss": 1.1745, "step": 28976 }, { "epoch": 0.03, "learning_rate": 4.989717402624832e-05, "loss": 1.3928, "step": 28977 }, { "epoch": 0.03, "learning_rate": 4.9897166909359306e-05, "loss": 1.8586, "step": 28978 }, { "epoch": 0.03, "learning_rate": 4.989715979222451e-05, "loss": 1.354, "step": 28979 }, { "epoch": 0.03, "learning_rate": 4.989715267484394e-05, "loss": 1.279, "step": 28980 }, { "epoch": 0.03, "learning_rate": 4.989714555721759e-05, "loss": 1.0072, "step": 28981 }, { "epoch": 0.03, "learning_rate": 4.989713843934548e-05, "loss": 0.8671, "step": 28982 }, { "epoch": 0.03, "learning_rate": 4.989713132122759e-05, "loss": 1.1895, "step": 28983 }, { "epoch": 0.03, "learning_rate": 4.9897124202863916e-05, "loss": 1.0972, "step": 28984 }, { "epoch": 0.03, "learning_rate": 4.9897117084254476e-05, "loss": 1.119, "step": 28985 }, { "epoch": 0.03, "learning_rate": 4.989710996539926e-05, "loss": 1.0334, "step": 28986 }, { "epoch": 0.03, "learning_rate": 4.989710284629827e-05, "loss": 1.1169, "step": 28987 }, { "epoch": 0.03, "learning_rate": 4.989709572695151e-05, "loss": 1.0835, "step": 28988 }, { "epoch": 0.03, "learning_rate": 4.989708860735898e-05, "loss": 1.0788, "step": 28989 }, { "epoch": 0.03, "learning_rate": 4.989708148752067e-05, "loss": 1.0233, "step": 28990 }, { "epoch": 0.03, "learning_rate": 4.9897074367436586e-05, "loss": 0.7061, "step": 28991 }, { "epoch": 0.03, "learning_rate": 4.9897067247106734e-05, "loss": 0.7283, "step": 28992 }, { "epoch": 0.03, "learning_rate": 4.98970601265311e-05, "loss": 1.3084, "step": 28993 }, { "epoch": 0.03, "learning_rate": 4.98970530057097e-05, "loss": 0.9021, "step": 28994 }, { "epoch": 0.03, "learning_rate": 4.989704588464252e-05, "loss": 0.9384, "step": 28995 }, { "epoch": 0.03, "learning_rate": 4.9897038763329576e-05, "loss": 1.0217, "step": 28996 }, { "epoch": 0.03, "learning_rate": 4.9897031641770855e-05, "loss": 0.9787, "step": 28997 }, { "epoch": 0.03, "learning_rate": 4.989702451996636e-05, "loss": 1.0799, "step": 28998 }, { "epoch": 0.03, "learning_rate": 4.989701739791609e-05, "loss": 0.9109, "step": 28999 }, { "epoch": 0.03, "learning_rate": 4.989701027562005e-05, "loss": 1.0983, "step": 29000 }, { "epoch": 0.03, "eval_loss": 1.0596212148666382, "eval_runtime": 126.066, "eval_samples_per_second": 10.986, "eval_steps_per_second": 5.497, "step": 29000 }, { "epoch": 0.03, "learning_rate": 4.9897003153078235e-05, "loss": 1.2042, "step": 29001 }, { "epoch": 0.03, "learning_rate": 4.989699603029065e-05, "loss": 0.9688, "step": 29002 }, { "epoch": 0.03, "learning_rate": 4.989698890725729e-05, "loss": 0.929, "step": 29003 }, { "epoch": 0.03, "learning_rate": 4.9896981783978156e-05, "loss": 0.8213, "step": 29004 }, { "epoch": 0.03, "learning_rate": 4.9896974660453255e-05, "loss": 0.849, "step": 29005 }, { "epoch": 0.03, "learning_rate": 4.989696753668258e-05, "loss": 0.8995, "step": 29006 }, { "epoch": 0.03, "learning_rate": 4.9896960412666135e-05, "loss": 1.3146, "step": 29007 }, { "epoch": 0.03, "learning_rate": 4.9896953288403915e-05, "loss": 0.8484, "step": 29008 }, { "epoch": 0.03, "learning_rate": 4.989694616389592e-05, "loss": 1.1348, "step": 29009 }, { "epoch": 0.03, "learning_rate": 4.989693903914215e-05, "loss": 1.0171, "step": 29010 }, { "epoch": 0.03, "learning_rate": 4.9896931914142614e-05, "loss": 1.271, "step": 29011 }, { "epoch": 0.03, "learning_rate": 4.98969247888973e-05, "loss": 1.4952, "step": 29012 }, { "epoch": 0.03, "learning_rate": 4.9896917663406226e-05, "loss": 1.0631, "step": 29013 }, { "epoch": 0.03, "learning_rate": 4.989691053766937e-05, "loss": 1.0919, "step": 29014 }, { "epoch": 0.03, "learning_rate": 4.9896903411686745e-05, "loss": 1.0492, "step": 29015 }, { "epoch": 0.03, "learning_rate": 4.989689628545835e-05, "loss": 1.1008, "step": 29016 }, { "epoch": 0.03, "learning_rate": 4.9896889158984184e-05, "loss": 0.6918, "step": 29017 }, { "epoch": 0.03, "learning_rate": 4.989688203226424e-05, "loss": 0.7526, "step": 29018 }, { "epoch": 0.03, "learning_rate": 4.989687490529853e-05, "loss": 1.0927, "step": 29019 }, { "epoch": 0.03, "learning_rate": 4.989686777808704e-05, "loss": 0.9819, "step": 29020 }, { "epoch": 0.03, "learning_rate": 4.9896860650629795e-05, "loss": 1.0669, "step": 29021 }, { "epoch": 0.03, "learning_rate": 4.9896853522926766e-05, "loss": 0.9665, "step": 29022 }, { "epoch": 0.03, "learning_rate": 4.989684639497797e-05, "loss": 1.0373, "step": 29023 }, { "epoch": 0.03, "learning_rate": 4.98968392667834e-05, "loss": 1.1351, "step": 29024 }, { "epoch": 0.03, "learning_rate": 4.9896832138343066e-05, "loss": 1.082, "step": 29025 }, { "epoch": 0.03, "learning_rate": 4.989682500965695e-05, "loss": 1.2862, "step": 29026 }, { "epoch": 0.03, "learning_rate": 4.9896817880725065e-05, "loss": 0.8061, "step": 29027 }, { "epoch": 0.03, "learning_rate": 4.989681075154742e-05, "loss": 1.0785, "step": 29028 }, { "epoch": 0.03, "learning_rate": 4.9896803622124e-05, "loss": 1.1208, "step": 29029 }, { "epoch": 0.03, "learning_rate": 4.98967964924548e-05, "loss": 1.0553, "step": 29030 }, { "epoch": 0.03, "learning_rate": 4.9896789362539835e-05, "loss": 0.8657, "step": 29031 }, { "epoch": 0.03, "learning_rate": 4.98967822323791e-05, "loss": 0.6038, "step": 29032 }, { "epoch": 0.03, "learning_rate": 4.9896775101972594e-05, "loss": 0.9803, "step": 29033 }, { "epoch": 0.03, "learning_rate": 4.989676797132032e-05, "loss": 0.9201, "step": 29034 }, { "epoch": 0.03, "learning_rate": 4.989676084042227e-05, "loss": 0.8263, "step": 29035 }, { "epoch": 0.03, "learning_rate": 4.989675370927845e-05, "loss": 1.341, "step": 29036 }, { "epoch": 0.03, "learning_rate": 4.989674657788887e-05, "loss": 0.8832, "step": 29037 }, { "epoch": 0.03, "learning_rate": 4.9896739446253514e-05, "loss": 1.0133, "step": 29038 }, { "epoch": 0.03, "learning_rate": 4.989673231437238e-05, "loss": 1.0844, "step": 29039 }, { "epoch": 0.03, "learning_rate": 4.989672518224548e-05, "loss": 0.9587, "step": 29040 }, { "epoch": 0.03, "learning_rate": 4.989671804987282e-05, "loss": 1.1303, "step": 29041 }, { "epoch": 0.03, "learning_rate": 4.9896710917254384e-05, "loss": 0.9136, "step": 29042 }, { "epoch": 0.03, "learning_rate": 4.989670378439017e-05, "loss": 0.6165, "step": 29043 }, { "epoch": 0.03, "learning_rate": 4.989669665128019e-05, "loss": 0.6482, "step": 29044 }, { "epoch": 0.03, "learning_rate": 4.989668951792446e-05, "loss": 0.9987, "step": 29045 }, { "epoch": 0.03, "learning_rate": 4.9896682384322935e-05, "loss": 0.9547, "step": 29046 }, { "epoch": 0.03, "learning_rate": 4.989667525047565e-05, "loss": 1.2747, "step": 29047 }, { "epoch": 0.03, "learning_rate": 4.98966681163826e-05, "loss": 1.3389, "step": 29048 }, { "epoch": 0.03, "learning_rate": 4.989666098204377e-05, "loss": 1.036, "step": 29049 }, { "epoch": 0.03, "learning_rate": 4.989665384745917e-05, "loss": 1.073, "step": 29050 }, { "epoch": 0.03, "learning_rate": 4.9896646712628814e-05, "loss": 0.926, "step": 29051 }, { "epoch": 0.03, "learning_rate": 4.989663957755268e-05, "loss": 0.8147, "step": 29052 }, { "epoch": 0.03, "learning_rate": 4.989663244223078e-05, "loss": 0.9254, "step": 29053 }, { "epoch": 0.03, "learning_rate": 4.98966253066631e-05, "loss": 0.8017, "step": 29054 }, { "epoch": 0.03, "learning_rate": 4.989661817084967e-05, "loss": 0.6406, "step": 29055 }, { "epoch": 0.03, "learning_rate": 4.989661103479046e-05, "loss": 1.5227, "step": 29056 }, { "epoch": 0.03, "learning_rate": 4.989660389848548e-05, "loss": 0.8772, "step": 29057 }, { "epoch": 0.03, "learning_rate": 4.9896596761934734e-05, "loss": 0.354, "step": 29058 }, { "epoch": 0.03, "learning_rate": 4.989658962513822e-05, "loss": 0.7528, "step": 29059 }, { "epoch": 0.03, "learning_rate": 4.9896582488095936e-05, "loss": 1.3361, "step": 29060 }, { "epoch": 0.03, "learning_rate": 4.989657535080788e-05, "loss": 1.1203, "step": 29061 }, { "epoch": 0.03, "learning_rate": 4.9896568213274065e-05, "loss": 1.0425, "step": 29062 }, { "epoch": 0.03, "learning_rate": 4.9896561075494474e-05, "loss": 1.0864, "step": 29063 }, { "epoch": 0.03, "learning_rate": 4.9896553937469114e-05, "loss": 1.0477, "step": 29064 }, { "epoch": 0.03, "learning_rate": 4.989654679919799e-05, "loss": 0.8958, "step": 29065 }, { "epoch": 0.03, "learning_rate": 4.98965396606811e-05, "loss": 0.3584, "step": 29066 }, { "epoch": 0.03, "learning_rate": 4.989653252191844e-05, "loss": 0.3127, "step": 29067 }, { "epoch": 0.03, "learning_rate": 4.9896525382910006e-05, "loss": 0.6144, "step": 29068 }, { "epoch": 0.03, "learning_rate": 4.9896518243655805e-05, "loss": 1.0873, "step": 29069 }, { "epoch": 0.03, "learning_rate": 4.989651110415584e-05, "loss": 1.0176, "step": 29070 }, { "epoch": 0.03, "learning_rate": 4.9896503964410105e-05, "loss": 1.0428, "step": 29071 }, { "epoch": 0.03, "learning_rate": 4.9896496824418606e-05, "loss": 1.0524, "step": 29072 }, { "epoch": 0.03, "learning_rate": 4.989648968418134e-05, "loss": 0.8031, "step": 29073 }, { "epoch": 0.03, "learning_rate": 4.9896482543698295e-05, "loss": 0.8666, "step": 29074 }, { "epoch": 0.03, "learning_rate": 4.989647540296949e-05, "loss": 1.1541, "step": 29075 }, { "epoch": 0.03, "learning_rate": 4.989646826199492e-05, "loss": 1.0675, "step": 29076 }, { "epoch": 0.03, "learning_rate": 4.989646112077458e-05, "loss": 0.8997, "step": 29077 }, { "epoch": 0.03, "learning_rate": 4.989645397930847e-05, "loss": 0.8984, "step": 29078 }, { "epoch": 0.03, "learning_rate": 4.98964468375966e-05, "loss": 1.1646, "step": 29079 }, { "epoch": 0.03, "learning_rate": 4.989643969563895e-05, "loss": 0.9267, "step": 29080 }, { "epoch": 0.03, "learning_rate": 4.9896432553435546e-05, "loss": 0.7993, "step": 29081 }, { "epoch": 0.03, "learning_rate": 4.989642541098637e-05, "loss": 1.1879, "step": 29082 }, { "epoch": 0.03, "learning_rate": 4.9896418268291426e-05, "loss": 0.6457, "step": 29083 }, { "epoch": 0.03, "learning_rate": 4.9896411125350714e-05, "loss": 0.3846, "step": 29084 }, { "epoch": 0.03, "learning_rate": 4.989640398216424e-05, "loss": 0.7882, "step": 29085 }, { "epoch": 0.03, "learning_rate": 4.9896396838732e-05, "loss": 0.8765, "step": 29086 }, { "epoch": 0.03, "learning_rate": 4.989638969505399e-05, "loss": 0.9096, "step": 29087 }, { "epoch": 0.03, "learning_rate": 4.989638255113021e-05, "loss": 0.8133, "step": 29088 }, { "epoch": 0.03, "learning_rate": 4.989637540696066e-05, "loss": 0.8176, "step": 29089 }, { "epoch": 0.03, "learning_rate": 4.9896368262545355e-05, "loss": 1.0449, "step": 29090 }, { "epoch": 0.03, "learning_rate": 4.9896361117884285e-05, "loss": 0.9024, "step": 29091 }, { "epoch": 0.03, "learning_rate": 4.989635397297744e-05, "loss": 0.9412, "step": 29092 }, { "epoch": 0.03, "learning_rate": 4.9896346827824826e-05, "loss": 0.9131, "step": 29093 }, { "epoch": 0.03, "learning_rate": 4.989633968242645e-05, "loss": 1.0466, "step": 29094 }, { "epoch": 0.03, "learning_rate": 4.989633253678231e-05, "loss": 0.9202, "step": 29095 }, { "epoch": 0.03, "learning_rate": 4.9896325390892404e-05, "loss": 1.0047, "step": 29096 }, { "epoch": 0.03, "learning_rate": 4.989631824475673e-05, "loss": 0.8466, "step": 29097 }, { "epoch": 0.03, "learning_rate": 4.989631109837529e-05, "loss": 0.867, "step": 29098 }, { "epoch": 0.03, "learning_rate": 4.989630395174808e-05, "loss": 0.5349, "step": 29099 }, { "epoch": 0.03, "learning_rate": 4.989629680487511e-05, "loss": 0.7237, "step": 29100 }, { "epoch": 0.03, "learning_rate": 4.9896289657756376e-05, "loss": 0.7533, "step": 29101 }, { "epoch": 0.03, "learning_rate": 4.9896282510391875e-05, "loss": 0.4468, "step": 29102 }, { "epoch": 0.03, "learning_rate": 4.98962753627816e-05, "loss": 1.0564, "step": 29103 }, { "epoch": 0.03, "learning_rate": 4.989626821492557e-05, "loss": 0.884, "step": 29104 }, { "epoch": 0.03, "learning_rate": 4.989626106682377e-05, "loss": 1.0782, "step": 29105 }, { "epoch": 0.03, "learning_rate": 4.98962539184762e-05, "loss": 1.3443, "step": 29106 }, { "epoch": 0.03, "learning_rate": 4.989624676988287e-05, "loss": 1.4795, "step": 29107 }, { "epoch": 0.03, "learning_rate": 4.9896239621043776e-05, "loss": 0.9764, "step": 29108 }, { "epoch": 0.03, "learning_rate": 4.989623247195892e-05, "loss": 1.1921, "step": 29109 }, { "epoch": 0.03, "learning_rate": 4.989622532262829e-05, "loss": 1.174, "step": 29110 }, { "epoch": 0.03, "learning_rate": 4.98962181730519e-05, "loss": 1.2818, "step": 29111 }, { "epoch": 0.03, "learning_rate": 4.9896211023229745e-05, "loss": 1.3059, "step": 29112 }, { "epoch": 0.03, "learning_rate": 4.989620387316182e-05, "loss": 0.9999, "step": 29113 }, { "epoch": 0.03, "learning_rate": 4.989619672284814e-05, "loss": 0.9788, "step": 29114 }, { "epoch": 0.03, "learning_rate": 4.9896189572288685e-05, "loss": 1.0644, "step": 29115 }, { "epoch": 0.03, "learning_rate": 4.9896182421483475e-05, "loss": 0.7423, "step": 29116 }, { "epoch": 0.03, "learning_rate": 4.989617527043249e-05, "loss": 0.901, "step": 29117 }, { "epoch": 0.03, "learning_rate": 4.989616811913575e-05, "loss": 1.496, "step": 29118 }, { "epoch": 0.03, "learning_rate": 4.989616096759324e-05, "loss": 0.9832, "step": 29119 }, { "epoch": 0.03, "learning_rate": 4.989615381580497e-05, "loss": 1.0821, "step": 29120 }, { "epoch": 0.03, "learning_rate": 4.989614666377093e-05, "loss": 1.0936, "step": 29121 }, { "epoch": 0.03, "learning_rate": 4.989613951149112e-05, "loss": 1.0122, "step": 29122 }, { "epoch": 0.03, "learning_rate": 4.989613235896556e-05, "loss": 1.0169, "step": 29123 }, { "epoch": 0.03, "learning_rate": 4.989612520619423e-05, "loss": 1.4078, "step": 29124 }, { "epoch": 0.03, "learning_rate": 4.9896118053177136e-05, "loss": 1.3354, "step": 29125 }, { "epoch": 0.03, "learning_rate": 4.9896110899914284e-05, "loss": 0.3098, "step": 29126 }, { "epoch": 0.03, "learning_rate": 4.9896103746405656e-05, "loss": 0.3136, "step": 29127 }, { "epoch": 0.03, "learning_rate": 4.9896096592651275e-05, "loss": 4.1156, "step": 29128 }, { "epoch": 0.03, "learning_rate": 4.9896089438651124e-05, "loss": 1.9682, "step": 29129 }, { "epoch": 0.03, "learning_rate": 4.989608228440521e-05, "loss": 1.1782, "step": 29130 }, { "epoch": 0.03, "learning_rate": 4.989607512991353e-05, "loss": 0.883, "step": 29131 }, { "epoch": 0.03, "learning_rate": 4.989606797517609e-05, "loss": 0.8494, "step": 29132 }, { "epoch": 0.03, "learning_rate": 4.989606082019289e-05, "loss": 1.2184, "step": 29133 }, { "epoch": 0.03, "learning_rate": 4.9896053664963916e-05, "loss": 0.9239, "step": 29134 }, { "epoch": 0.03, "learning_rate": 4.989604650948919e-05, "loss": 0.4045, "step": 29135 }, { "epoch": 0.03, "learning_rate": 4.9896039353768696e-05, "loss": 0.1424, "step": 29136 }, { "epoch": 0.03, "learning_rate": 4.989603219780243e-05, "loss": 0.8851, "step": 29137 }, { "epoch": 0.03, "learning_rate": 4.9896025041590415e-05, "loss": 1.1703, "step": 29138 }, { "epoch": 0.03, "learning_rate": 4.9896017885132636e-05, "loss": 0.848, "step": 29139 }, { "epoch": 0.03, "learning_rate": 4.989601072842909e-05, "loss": 1.1787, "step": 29140 }, { "epoch": 0.03, "learning_rate": 4.989600357147978e-05, "loss": 0.9603, "step": 29141 }, { "epoch": 0.03, "learning_rate": 4.989599641428471e-05, "loss": 0.6911, "step": 29142 }, { "epoch": 0.03, "learning_rate": 4.9895989256843864e-05, "loss": 0.1579, "step": 29143 }, { "epoch": 0.03, "learning_rate": 4.989598209915727e-05, "loss": 0.1521, "step": 29144 }, { "epoch": 0.03, "learning_rate": 4.989597494122491e-05, "loss": 0.3108, "step": 29145 }, { "epoch": 0.03, "learning_rate": 4.9895967783046785e-05, "loss": 0.9406, "step": 29146 }, { "epoch": 0.03, "learning_rate": 4.989596062462291e-05, "loss": 0.8081, "step": 29147 }, { "epoch": 0.03, "learning_rate": 4.9895953465953254e-05, "loss": 1.0443, "step": 29148 }, { "epoch": 0.03, "learning_rate": 4.9895946307037846e-05, "loss": 1.1583, "step": 29149 }, { "epoch": 0.03, "learning_rate": 4.989593914787667e-05, "loss": 1.0134, "step": 29150 }, { "epoch": 0.03, "learning_rate": 4.989593198846973e-05, "loss": 1.0383, "step": 29151 }, { "epoch": 0.03, "learning_rate": 4.989592482881704e-05, "loss": 0.8247, "step": 29152 }, { "epoch": 0.03, "learning_rate": 4.989591766891858e-05, "loss": 1.0257, "step": 29153 }, { "epoch": 0.03, "learning_rate": 4.9895910508774357e-05, "loss": 1.0188, "step": 29154 }, { "epoch": 0.03, "learning_rate": 4.989590334838438e-05, "loss": 1.0004, "step": 29155 }, { "epoch": 0.03, "learning_rate": 4.9895896187748634e-05, "loss": 1.0485, "step": 29156 }, { "epoch": 0.03, "learning_rate": 4.989588902686713e-05, "loss": 1.0402, "step": 29157 }, { "epoch": 0.03, "learning_rate": 4.989588186573986e-05, "loss": 1.0171, "step": 29158 }, { "epoch": 0.03, "learning_rate": 4.989587470436683e-05, "loss": 1.0026, "step": 29159 }, { "epoch": 0.03, "learning_rate": 4.989586754274803e-05, "loss": 1.2145, "step": 29160 }, { "epoch": 0.03, "learning_rate": 4.9895860380883484e-05, "loss": 1.0712, "step": 29161 }, { "epoch": 0.03, "learning_rate": 4.989585321877317e-05, "loss": 1.0376, "step": 29162 }, { "epoch": 0.03, "learning_rate": 4.9895846056417094e-05, "loss": 0.9398, "step": 29163 }, { "epoch": 0.03, "learning_rate": 4.9895838893815256e-05, "loss": 0.8942, "step": 29164 }, { "epoch": 0.03, "learning_rate": 4.989583173096766e-05, "loss": 2.1347, "step": 29165 }, { "epoch": 0.03, "learning_rate": 4.9895824567874296e-05, "loss": 1.1978, "step": 29166 }, { "epoch": 0.03, "learning_rate": 4.9895817404535174e-05, "loss": 0.7928, "step": 29167 }, { "epoch": 0.03, "learning_rate": 4.989581024095029e-05, "loss": 1.214, "step": 29168 }, { "epoch": 0.03, "learning_rate": 4.989580307711965e-05, "loss": 1.015, "step": 29169 }, { "epoch": 0.03, "learning_rate": 4.9895795913043245e-05, "loss": 0.76, "step": 29170 }, { "epoch": 0.03, "learning_rate": 4.9895788748721084e-05, "loss": 1.1821, "step": 29171 }, { "epoch": 0.03, "learning_rate": 4.9895781584153154e-05, "loss": 0.4537, "step": 29172 }, { "epoch": 0.03, "learning_rate": 4.989577441933947e-05, "loss": 0.8343, "step": 29173 }, { "epoch": 0.03, "learning_rate": 4.989576725428002e-05, "loss": 1.0043, "step": 29174 }, { "epoch": 0.03, "learning_rate": 4.9895760088974816e-05, "loss": 0.9862, "step": 29175 }, { "epoch": 0.03, "learning_rate": 4.989575292342385e-05, "loss": 1.0313, "step": 29176 }, { "epoch": 0.03, "learning_rate": 4.989574575762712e-05, "loss": 0.8568, "step": 29177 }, { "epoch": 0.03, "learning_rate": 4.989573859158464e-05, "loss": 1.2798, "step": 29178 }, { "epoch": 0.03, "learning_rate": 4.989573142529639e-05, "loss": 0.8715, "step": 29179 }, { "epoch": 0.03, "learning_rate": 4.9895724258762375e-05, "loss": 0.8769, "step": 29180 }, { "epoch": 0.03, "learning_rate": 4.9895717091982605e-05, "loss": 0.8463, "step": 29181 }, { "epoch": 0.03, "learning_rate": 4.989570992495708e-05, "loss": 1.1206, "step": 29182 }, { "epoch": 0.03, "learning_rate": 4.9895702757685794e-05, "loss": 1.2376, "step": 29183 }, { "epoch": 0.03, "learning_rate": 4.9895695590168746e-05, "loss": 1.3365, "step": 29184 }, { "epoch": 0.03, "learning_rate": 4.989568842240594e-05, "loss": 0.9084, "step": 29185 }, { "epoch": 0.03, "learning_rate": 4.9895681254397366e-05, "loss": 0.9741, "step": 29186 }, { "epoch": 0.03, "learning_rate": 4.989567408614304e-05, "loss": 0.9876, "step": 29187 }, { "epoch": 0.03, "learning_rate": 4.989566691764295e-05, "loss": 0.7732, "step": 29188 }, { "epoch": 0.03, "learning_rate": 4.9895659748897104e-05, "loss": 0.7497, "step": 29189 }, { "epoch": 0.03, "learning_rate": 4.98956525799055e-05, "loss": 0.9963, "step": 29190 }, { "epoch": 0.03, "learning_rate": 4.9895645410668136e-05, "loss": 0.9801, "step": 29191 }, { "epoch": 0.03, "learning_rate": 4.9895638241185e-05, "loss": 0.9852, "step": 29192 }, { "epoch": 0.03, "learning_rate": 4.989563107145613e-05, "loss": 1.5967, "step": 29193 }, { "epoch": 0.03, "learning_rate": 4.989562390148148e-05, "loss": 1.0507, "step": 29194 }, { "epoch": 0.03, "learning_rate": 4.989561673126108e-05, "loss": 0.865, "step": 29195 }, { "epoch": 0.03, "learning_rate": 4.989560956079492e-05, "loss": 0.9105, "step": 29196 }, { "epoch": 0.03, "learning_rate": 4.9895602390082995e-05, "loss": 0.9884, "step": 29197 }, { "epoch": 0.03, "learning_rate": 4.989559521912532e-05, "loss": 0.8667, "step": 29198 }, { "epoch": 0.03, "learning_rate": 4.989558804792188e-05, "loss": 0.8648, "step": 29199 }, { "epoch": 0.03, "learning_rate": 4.989558087647268e-05, "loss": 0.9377, "step": 29200 }, { "epoch": 0.03, "learning_rate": 4.9895573704777724e-05, "loss": 1.1753, "step": 29201 }, { "epoch": 0.03, "learning_rate": 4.989556653283701e-05, "loss": 0.8946, "step": 29202 }, { "epoch": 0.03, "learning_rate": 4.989555936065054e-05, "loss": 1.2819, "step": 29203 }, { "epoch": 0.03, "learning_rate": 4.989555218821831e-05, "loss": 0.7089, "step": 29204 }, { "epoch": 0.03, "learning_rate": 4.989554501554032e-05, "loss": 0.7565, "step": 29205 }, { "epoch": 0.03, "learning_rate": 4.989553784261657e-05, "loss": 1.1422, "step": 29206 }, { "epoch": 0.03, "learning_rate": 4.989553066944707e-05, "loss": 1.0805, "step": 29207 }, { "epoch": 0.03, "learning_rate": 4.98955234960318e-05, "loss": 1.0336, "step": 29208 }, { "epoch": 0.03, "learning_rate": 4.989551632237078e-05, "loss": 0.8886, "step": 29209 }, { "epoch": 0.03, "learning_rate": 4.989550914846399e-05, "loss": 0.8117, "step": 29210 }, { "epoch": 0.03, "learning_rate": 4.989550197431145e-05, "loss": 1.1089, "step": 29211 }, { "epoch": 0.03, "learning_rate": 4.989549479991316e-05, "loss": 1.0928, "step": 29212 }, { "epoch": 0.03, "learning_rate": 4.9895487625269105e-05, "loss": 1.1038, "step": 29213 }, { "epoch": 0.03, "learning_rate": 4.989548045037929e-05, "loss": 0.8486, "step": 29214 }, { "epoch": 0.03, "learning_rate": 4.989547327524372e-05, "loss": 1.0194, "step": 29215 }, { "epoch": 0.03, "learning_rate": 4.98954660998624e-05, "loss": 0.9102, "step": 29216 }, { "epoch": 0.03, "learning_rate": 4.989545892423531e-05, "loss": 0.9763, "step": 29217 }, { "epoch": 0.03, "learning_rate": 4.989545174836247e-05, "loss": 0.9738, "step": 29218 }, { "epoch": 0.03, "learning_rate": 4.989544457224387e-05, "loss": 1.128, "step": 29219 }, { "epoch": 0.03, "learning_rate": 4.9895437395879514e-05, "loss": 1.0363, "step": 29220 }, { "epoch": 0.03, "learning_rate": 4.98954302192694e-05, "loss": 1.3288, "step": 29221 }, { "epoch": 0.03, "learning_rate": 4.989542304241353e-05, "loss": 1.1105, "step": 29222 }, { "epoch": 0.03, "learning_rate": 4.98954158653119e-05, "loss": 0.7951, "step": 29223 }, { "epoch": 0.03, "learning_rate": 4.9895408687964515e-05, "loss": 1.1573, "step": 29224 }, { "epoch": 0.03, "learning_rate": 4.989540151037137e-05, "loss": 0.7594, "step": 29225 }, { "epoch": 0.03, "learning_rate": 4.989539433253247e-05, "loss": 1.0222, "step": 29226 }, { "epoch": 0.03, "learning_rate": 4.989538715444781e-05, "loss": 0.7839, "step": 29227 }, { "epoch": 0.03, "learning_rate": 4.989537997611739e-05, "loss": 0.7927, "step": 29228 }, { "epoch": 0.03, "learning_rate": 4.9895372797541226e-05, "loss": 1.049, "step": 29229 }, { "epoch": 0.03, "learning_rate": 4.9895365618719303e-05, "loss": 0.9789, "step": 29230 }, { "epoch": 0.03, "learning_rate": 4.989535843965162e-05, "loss": 0.6173, "step": 29231 }, { "epoch": 0.03, "learning_rate": 4.9895351260338174e-05, "loss": 0.8645, "step": 29232 }, { "epoch": 0.03, "learning_rate": 4.989534408077898e-05, "loss": 0.7539, "step": 29233 }, { "epoch": 0.03, "learning_rate": 4.989533690097403e-05, "loss": 0.5883, "step": 29234 }, { "epoch": 0.03, "learning_rate": 4.9895329720923324e-05, "loss": 1.173, "step": 29235 }, { "epoch": 0.03, "learning_rate": 4.989532254062685e-05, "loss": 0.8586, "step": 29236 }, { "epoch": 0.03, "learning_rate": 4.989531536008464e-05, "loss": 0.8761, "step": 29237 }, { "epoch": 0.03, "learning_rate": 4.989530817929666e-05, "loss": 0.437, "step": 29238 }, { "epoch": 0.03, "learning_rate": 4.989530099826293e-05, "loss": 0.3464, "step": 29239 }, { "epoch": 0.03, "learning_rate": 4.9895293816983434e-05, "loss": 0.2199, "step": 29240 }, { "epoch": 0.03, "learning_rate": 4.9895286635458196e-05, "loss": 0.1708, "step": 29241 }, { "epoch": 0.03, "learning_rate": 4.989527945368719e-05, "loss": 0.1026, "step": 29242 }, { "epoch": 0.03, "learning_rate": 4.9895272271670436e-05, "loss": 0.1849, "step": 29243 }, { "epoch": 0.03, "learning_rate": 4.989526508940792e-05, "loss": 0.9639, "step": 29244 }, { "epoch": 0.03, "learning_rate": 4.989525790689965e-05, "loss": 0.7573, "step": 29245 }, { "epoch": 0.03, "learning_rate": 4.9895250724145624e-05, "loss": 0.8215, "step": 29246 }, { "epoch": 0.03, "learning_rate": 4.989524354114585e-05, "loss": 0.8524, "step": 29247 }, { "epoch": 0.03, "learning_rate": 4.989523635790032e-05, "loss": 1.6066, "step": 29248 }, { "epoch": 0.03, "learning_rate": 4.989522917440903e-05, "loss": 1.6906, "step": 29249 }, { "epoch": 0.03, "learning_rate": 4.989522199067198e-05, "loss": 0.4315, "step": 29250 }, { "epoch": 0.03, "learning_rate": 4.989521480668918e-05, "loss": 0.177, "step": 29251 }, { "epoch": 0.03, "learning_rate": 4.989520762246063e-05, "loss": 0.119, "step": 29252 }, { "epoch": 0.03, "learning_rate": 4.989520043798632e-05, "loss": 0.0758, "step": 29253 }, { "epoch": 0.03, "learning_rate": 4.989519325326626e-05, "loss": 0.0594, "step": 29254 }, { "epoch": 0.03, "learning_rate": 4.989518606830044e-05, "loss": 0.0302, "step": 29255 }, { "epoch": 0.03, "learning_rate": 4.989517888308886e-05, "loss": 0.4339, "step": 29256 }, { "epoch": 0.03, "learning_rate": 4.989517169763153e-05, "loss": 0.9214, "step": 29257 }, { "epoch": 0.03, "learning_rate": 4.989516451192845e-05, "loss": 1.5124, "step": 29258 }, { "epoch": 0.03, "learning_rate": 4.989515732597961e-05, "loss": 1.1203, "step": 29259 }, { "epoch": 0.03, "learning_rate": 4.989515013978502e-05, "loss": 1.3431, "step": 29260 }, { "epoch": 0.03, "learning_rate": 4.9895142953344673e-05, "loss": 1.1845, "step": 29261 }, { "epoch": 0.03, "learning_rate": 4.9895135766658565e-05, "loss": 0.9739, "step": 29262 }, { "epoch": 0.03, "learning_rate": 4.9895128579726716e-05, "loss": 1.1506, "step": 29263 }, { "epoch": 0.03, "learning_rate": 4.9895121392549105e-05, "loss": 0.7151, "step": 29264 }, { "epoch": 0.03, "learning_rate": 4.989511420512574e-05, "loss": 0.7792, "step": 29265 }, { "epoch": 0.03, "learning_rate": 4.989510701745662e-05, "loss": 0.752, "step": 29266 }, { "epoch": 0.03, "learning_rate": 4.9895099829541744e-05, "loss": 0.7978, "step": 29267 }, { "epoch": 0.03, "learning_rate": 4.989509264138113e-05, "loss": 0.205, "step": 29268 }, { "epoch": 0.03, "learning_rate": 4.989508545297474e-05, "loss": 0.1651, "step": 29269 }, { "epoch": 0.03, "learning_rate": 4.989507826432261e-05, "loss": 0.1175, "step": 29270 }, { "epoch": 0.03, "learning_rate": 4.989507107542472e-05, "loss": 0.1591, "step": 29271 }, { "epoch": 0.03, "learning_rate": 4.989506388628108e-05, "loss": 0.6633, "step": 29272 }, { "epoch": 0.03, "learning_rate": 4.9895056696891684e-05, "loss": 1.0926, "step": 29273 }, { "epoch": 0.03, "learning_rate": 4.989504950725654e-05, "loss": 0.9763, "step": 29274 }, { "epoch": 0.03, "learning_rate": 4.989504231737564e-05, "loss": 1.3217, "step": 29275 }, { "epoch": 0.03, "learning_rate": 4.989503512724898e-05, "loss": 1.1074, "step": 29276 }, { "epoch": 0.03, "learning_rate": 4.989502793687657e-05, "loss": 0.849, "step": 29277 }, { "epoch": 0.03, "learning_rate": 4.989502074625841e-05, "loss": 1.3484, "step": 29278 }, { "epoch": 0.03, "learning_rate": 4.98950135553945e-05, "loss": 1.0595, "step": 29279 }, { "epoch": 0.03, "learning_rate": 4.989500636428483e-05, "loss": 1.0959, "step": 29280 }, { "epoch": 0.03, "learning_rate": 4.9894999172929406e-05, "loss": 1.2172, "step": 29281 }, { "epoch": 0.03, "learning_rate": 4.989499198132823e-05, "loss": 1.1578, "step": 29282 }, { "epoch": 0.03, "learning_rate": 4.9894984789481306e-05, "loss": 1.1875, "step": 29283 }, { "epoch": 0.03, "learning_rate": 4.989497759738863e-05, "loss": 0.8455, "step": 29284 }, { "epoch": 0.03, "learning_rate": 4.9894970405050194e-05, "loss": 1.1169, "step": 29285 }, { "epoch": 0.03, "learning_rate": 4.989496321246601e-05, "loss": 0.8254, "step": 29286 }, { "epoch": 0.03, "learning_rate": 4.9894956019636076e-05, "loss": 0.9896, "step": 29287 }, { "epoch": 0.03, "learning_rate": 4.9894948826560385e-05, "loss": 1.0347, "step": 29288 }, { "epoch": 0.03, "learning_rate": 4.989494163323894e-05, "loss": 0.7141, "step": 29289 }, { "epoch": 0.03, "learning_rate": 4.9894934439671745e-05, "loss": 1.2202, "step": 29290 }, { "epoch": 0.03, "learning_rate": 4.9894927245858804e-05, "loss": 0.9508, "step": 29291 }, { "epoch": 0.03, "learning_rate": 4.98949200518001e-05, "loss": 0.8317, "step": 29292 }, { "epoch": 0.03, "learning_rate": 4.989491285749565e-05, "loss": 0.8957, "step": 29293 }, { "epoch": 0.03, "learning_rate": 4.989490566294544e-05, "loss": 0.6932, "step": 29294 }, { "epoch": 0.03, "learning_rate": 4.989489846814949e-05, "loss": 1.2097, "step": 29295 }, { "epoch": 0.03, "learning_rate": 4.989489127310778e-05, "loss": 1.155, "step": 29296 }, { "epoch": 0.03, "learning_rate": 4.9894884077820324e-05, "loss": 0.8829, "step": 29297 }, { "epoch": 0.03, "learning_rate": 4.989487688228711e-05, "loss": 1.0818, "step": 29298 }, { "epoch": 0.03, "learning_rate": 4.989486968650815e-05, "loss": 1.4986, "step": 29299 }, { "epoch": 0.03, "learning_rate": 4.989486249048343e-05, "loss": 0.9708, "step": 29300 }, { "epoch": 0.03, "learning_rate": 4.989485529421296e-05, "loss": 1.179, "step": 29301 }, { "epoch": 0.03, "learning_rate": 4.989484809769675e-05, "loss": 1.2555, "step": 29302 }, { "epoch": 0.03, "learning_rate": 4.989484090093478e-05, "loss": 0.3215, "step": 29303 }, { "epoch": 0.03, "learning_rate": 4.9894833703927055e-05, "loss": 0.9909, "step": 29304 }, { "epoch": 0.03, "learning_rate": 4.989482650667359e-05, "loss": 0.7572, "step": 29305 }, { "epoch": 0.03, "learning_rate": 4.9894819309174365e-05, "loss": 0.9826, "step": 29306 }, { "epoch": 0.03, "learning_rate": 4.989481211142939e-05, "loss": 0.9851, "step": 29307 }, { "epoch": 0.03, "learning_rate": 4.989480491343866e-05, "loss": 1.0092, "step": 29308 }, { "epoch": 0.03, "learning_rate": 4.989479771520219e-05, "loss": 1.1625, "step": 29309 }, { "epoch": 0.03, "learning_rate": 4.9894790516719955e-05, "loss": 0.9372, "step": 29310 }, { "epoch": 0.03, "learning_rate": 4.9894783317991986e-05, "loss": 0.9822, "step": 29311 }, { "epoch": 0.03, "learning_rate": 4.9894776119018256e-05, "loss": 1.0692, "step": 29312 }, { "epoch": 0.03, "learning_rate": 4.989476891979877e-05, "loss": 1.051, "step": 29313 }, { "epoch": 0.03, "learning_rate": 4.9894761720333544e-05, "loss": 0.9704, "step": 29314 }, { "epoch": 0.03, "learning_rate": 4.989475452062256e-05, "loss": 1.0557, "step": 29315 }, { "epoch": 0.03, "learning_rate": 4.9894747320665834e-05, "loss": 0.7574, "step": 29316 }, { "epoch": 0.03, "learning_rate": 4.989474012046335e-05, "loss": 1.152, "step": 29317 }, { "epoch": 0.03, "learning_rate": 4.989473292001512e-05, "loss": 1.0923, "step": 29318 }, { "epoch": 0.03, "learning_rate": 4.989472571932113e-05, "loss": 1.1042, "step": 29319 }, { "epoch": 0.03, "learning_rate": 4.9894718518381404e-05, "loss": 0.6948, "step": 29320 }, { "epoch": 0.03, "learning_rate": 4.989471131719592e-05, "loss": 0.9491, "step": 29321 }, { "epoch": 0.03, "learning_rate": 4.9894704115764684e-05, "loss": 0.9543, "step": 29322 }, { "epoch": 0.03, "learning_rate": 4.9894696914087705e-05, "loss": 0.7987, "step": 29323 }, { "epoch": 0.03, "learning_rate": 4.9894689712164965e-05, "loss": 1.0699, "step": 29324 }, { "epoch": 0.03, "learning_rate": 4.9894682509996484e-05, "loss": 0.9408, "step": 29325 }, { "epoch": 0.03, "learning_rate": 4.989467530758225e-05, "loss": 1.0876, "step": 29326 }, { "epoch": 0.03, "learning_rate": 4.989466810492227e-05, "loss": 1.0953, "step": 29327 }, { "epoch": 0.03, "learning_rate": 4.989466090201654e-05, "loss": 0.7599, "step": 29328 }, { "epoch": 0.03, "learning_rate": 4.989465369886506e-05, "loss": 0.4469, "step": 29329 }, { "epoch": 0.03, "learning_rate": 4.9894646495467824e-05, "loss": 0.806, "step": 29330 }, { "epoch": 0.03, "learning_rate": 4.989463929182484e-05, "loss": 0.6327, "step": 29331 }, { "epoch": 0.03, "learning_rate": 4.989463208793611e-05, "loss": 0.9187, "step": 29332 }, { "epoch": 0.03, "learning_rate": 4.989462488380163e-05, "loss": 1.053, "step": 29333 }, { "epoch": 0.03, "learning_rate": 4.9894617679421405e-05, "loss": 0.8819, "step": 29334 }, { "epoch": 0.03, "learning_rate": 4.989461047479543e-05, "loss": 0.98, "step": 29335 }, { "epoch": 0.03, "learning_rate": 4.98946032699237e-05, "loss": 1.0498, "step": 29336 }, { "epoch": 0.03, "learning_rate": 4.9894596064806224e-05, "loss": 0.8612, "step": 29337 }, { "epoch": 0.03, "learning_rate": 4.9894588859443e-05, "loss": 1.035, "step": 29338 }, { "epoch": 0.03, "learning_rate": 4.9894581653834025e-05, "loss": 0.96, "step": 29339 }, { "epoch": 0.03, "learning_rate": 4.9894574447979304e-05, "loss": 0.8642, "step": 29340 }, { "epoch": 0.03, "learning_rate": 4.989456724187883e-05, "loss": 0.7683, "step": 29341 }, { "epoch": 0.03, "learning_rate": 4.989456003553261e-05, "loss": 0.7102, "step": 29342 }, { "epoch": 0.03, "learning_rate": 4.9894552828940646e-05, "loss": 0.8897, "step": 29343 }, { "epoch": 0.03, "learning_rate": 4.9894545622102926e-05, "loss": 0.9385, "step": 29344 }, { "epoch": 0.03, "learning_rate": 4.989453841501946e-05, "loss": 1.0232, "step": 29345 }, { "epoch": 0.03, "learning_rate": 4.989453120769024e-05, "loss": 1.014, "step": 29346 }, { "epoch": 0.03, "learning_rate": 4.9894524000115285e-05, "loss": 0.8275, "step": 29347 }, { "epoch": 0.03, "learning_rate": 4.989451679229457e-05, "loss": 0.9315, "step": 29348 }, { "epoch": 0.03, "learning_rate": 4.9894509584228113e-05, "loss": 1.2444, "step": 29349 }, { "epoch": 0.03, "learning_rate": 4.9894502375915906e-05, "loss": 1.0169, "step": 29350 }, { "epoch": 0.03, "learning_rate": 4.989449516735794e-05, "loss": 0.8404, "step": 29351 }, { "epoch": 0.03, "learning_rate": 4.9894487958554246e-05, "loss": 0.6489, "step": 29352 }, { "epoch": 0.03, "learning_rate": 4.9894480749504794e-05, "loss": 1.3753, "step": 29353 }, { "epoch": 0.03, "learning_rate": 4.9894473540209595e-05, "loss": 0.8628, "step": 29354 }, { "epoch": 0.03, "learning_rate": 4.989446633066865e-05, "loss": 0.9753, "step": 29355 }, { "epoch": 0.03, "learning_rate": 4.989445912088195e-05, "loss": 1.0818, "step": 29356 }, { "epoch": 0.03, "learning_rate": 4.9894451910849515e-05, "loss": 1.1394, "step": 29357 }, { "epoch": 0.03, "learning_rate": 4.989444470057132e-05, "loss": 0.8914, "step": 29358 }, { "epoch": 0.03, "learning_rate": 4.9894437490047384e-05, "loss": 1.003, "step": 29359 }, { "epoch": 0.03, "learning_rate": 4.98944302792777e-05, "loss": 0.7021, "step": 29360 }, { "epoch": 0.03, "learning_rate": 4.9894423068262274e-05, "loss": 0.8251, "step": 29361 }, { "epoch": 0.03, "learning_rate": 4.989441585700109e-05, "loss": 0.4326, "step": 29362 }, { "epoch": 0.03, "learning_rate": 4.9894408645494166e-05, "loss": 0.3051, "step": 29363 }, { "epoch": 0.03, "learning_rate": 4.989440143374149e-05, "loss": 0.2911, "step": 29364 }, { "epoch": 0.03, "learning_rate": 4.989439422174307e-05, "loss": 0.2458, "step": 29365 }, { "epoch": 0.03, "learning_rate": 4.9894387009498904e-05, "loss": 0.2318, "step": 29366 }, { "epoch": 0.03, "learning_rate": 4.9894379797008994e-05, "loss": 0.351, "step": 29367 }, { "epoch": 0.03, "learning_rate": 4.989437258427333e-05, "loss": 0.2303, "step": 29368 }, { "epoch": 0.03, "learning_rate": 4.9894365371291924e-05, "loss": 0.2579, "step": 29369 }, { "epoch": 0.03, "learning_rate": 4.989435815806477e-05, "loss": 0.2376, "step": 29370 }, { "epoch": 0.03, "learning_rate": 4.989435094459187e-05, "loss": 0.2198, "step": 29371 }, { "epoch": 0.03, "learning_rate": 4.989434373087322e-05, "loss": 1.1669, "step": 29372 }, { "epoch": 0.03, "learning_rate": 4.989433651690882e-05, "loss": 1.2567, "step": 29373 }, { "epoch": 0.03, "learning_rate": 4.989432930269868e-05, "loss": 0.7841, "step": 29374 }, { "epoch": 0.03, "learning_rate": 4.9894322088242796e-05, "loss": 1.3161, "step": 29375 }, { "epoch": 0.03, "learning_rate": 4.989431487354117e-05, "loss": 0.882, "step": 29376 }, { "epoch": 0.03, "learning_rate": 4.9894307658593785e-05, "loss": 0.739, "step": 29377 }, { "epoch": 0.03, "learning_rate": 4.989430044340067e-05, "loss": 0.9347, "step": 29378 }, { "epoch": 0.03, "learning_rate": 4.989429322796179e-05, "loss": 1.0013, "step": 29379 }, { "epoch": 0.03, "learning_rate": 4.989428601227717e-05, "loss": 1.3179, "step": 29380 }, { "epoch": 0.03, "learning_rate": 4.989427879634681e-05, "loss": 0.9084, "step": 29381 }, { "epoch": 0.03, "learning_rate": 4.9894271580170706e-05, "loss": 1.1538, "step": 29382 }, { "epoch": 0.03, "learning_rate": 4.989426436374884e-05, "loss": 1.0696, "step": 29383 }, { "epoch": 0.03, "learning_rate": 4.9894257147081244e-05, "loss": 1.0778, "step": 29384 }, { "epoch": 0.03, "learning_rate": 4.9894249930167904e-05, "loss": 0.8512, "step": 29385 }, { "epoch": 0.03, "learning_rate": 4.989424271300881e-05, "loss": 1.043, "step": 29386 }, { "epoch": 0.03, "learning_rate": 4.989423549560397e-05, "loss": 0.8696, "step": 29387 }, { "epoch": 0.03, "learning_rate": 4.989422827795339e-05, "loss": 0.7935, "step": 29388 }, { "epoch": 0.03, "learning_rate": 4.989422106005706e-05, "loss": 0.5885, "step": 29389 }, { "epoch": 0.03, "learning_rate": 4.989421384191499e-05, "loss": 0.9029, "step": 29390 }, { "epoch": 0.03, "learning_rate": 4.989420662352717e-05, "loss": 0.6442, "step": 29391 }, { "epoch": 0.03, "learning_rate": 4.9894199404893605e-05, "loss": 0.5084, "step": 29392 }, { "epoch": 0.03, "learning_rate": 4.98941921860143e-05, "loss": 0.4674, "step": 29393 }, { "epoch": 0.03, "learning_rate": 4.9894184966889245e-05, "loss": 0.6737, "step": 29394 }, { "epoch": 0.03, "learning_rate": 4.9894177747518446e-05, "loss": 0.9451, "step": 29395 }, { "epoch": 0.03, "learning_rate": 4.98941705279019e-05, "loss": 1.289, "step": 29396 }, { "epoch": 0.03, "learning_rate": 4.989416330803962e-05, "loss": 0.9939, "step": 29397 }, { "epoch": 0.03, "learning_rate": 4.989415608793158e-05, "loss": 0.8534, "step": 29398 }, { "epoch": 0.03, "learning_rate": 4.9894148867577804e-05, "loss": 1.1014, "step": 29399 }, { "epoch": 0.03, "learning_rate": 4.989414164697828e-05, "loss": 0.7335, "step": 29400 }, { "epoch": 0.03, "learning_rate": 4.989413442613301e-05, "loss": 0.9387, "step": 29401 }, { "epoch": 0.03, "learning_rate": 4.9894127205042005e-05, "loss": 0.4847, "step": 29402 }, { "epoch": 0.03, "learning_rate": 4.989411998370525e-05, "loss": 1.0982, "step": 29403 }, { "epoch": 0.03, "learning_rate": 4.989411276212275e-05, "loss": 0.8505, "step": 29404 }, { "epoch": 0.03, "learning_rate": 4.989410554029451e-05, "loss": 1.1021, "step": 29405 }, { "epoch": 0.03, "learning_rate": 4.9894098318220515e-05, "loss": 1.0102, "step": 29406 }, { "epoch": 0.03, "learning_rate": 4.989409109590079e-05, "loss": 1.0886, "step": 29407 }, { "epoch": 0.03, "learning_rate": 4.989408387333531e-05, "loss": 1.1411, "step": 29408 }, { "epoch": 0.03, "learning_rate": 4.989407665052409e-05, "loss": 1.0579, "step": 29409 }, { "epoch": 0.03, "learning_rate": 4.9894069427467126e-05, "loss": 0.9267, "step": 29410 }, { "epoch": 0.03, "learning_rate": 4.989406220416443e-05, "loss": 0.9765, "step": 29411 }, { "epoch": 0.03, "learning_rate": 4.9894054980615974e-05, "loss": 0.5028, "step": 29412 }, { "epoch": 0.03, "learning_rate": 4.989404775682178e-05, "loss": 0.7599, "step": 29413 }, { "epoch": 0.03, "learning_rate": 4.9894040532781836e-05, "loss": 0.8948, "step": 29414 }, { "epoch": 0.03, "learning_rate": 4.989403330849616e-05, "loss": 1.1558, "step": 29415 }, { "epoch": 0.03, "learning_rate": 4.9894026083964734e-05, "loss": 0.6147, "step": 29416 }, { "epoch": 0.03, "learning_rate": 4.989401885918756e-05, "loss": 1.056, "step": 29417 }, { "epoch": 0.03, "learning_rate": 4.9894011634164654e-05, "loss": 1.095, "step": 29418 }, { "epoch": 0.03, "learning_rate": 4.9894004408896e-05, "loss": 0.8667, "step": 29419 }, { "epoch": 0.03, "learning_rate": 4.98939971833816e-05, "loss": 0.929, "step": 29420 }, { "epoch": 0.03, "learning_rate": 4.9893989957621464e-05, "loss": 0.8799, "step": 29421 }, { "epoch": 0.03, "learning_rate": 4.989398273161557e-05, "loss": 1.1629, "step": 29422 }, { "epoch": 0.03, "learning_rate": 4.989397550536395e-05, "loss": 0.8678, "step": 29423 }, { "epoch": 0.03, "learning_rate": 4.989396827886658e-05, "loss": 0.9394, "step": 29424 }, { "epoch": 0.03, "learning_rate": 4.9893961052123474e-05, "loss": 0.9161, "step": 29425 }, { "epoch": 0.03, "learning_rate": 4.9893953825134616e-05, "loss": 1.0829, "step": 29426 }, { "epoch": 0.03, "learning_rate": 4.989394659790002e-05, "loss": 0.9521, "step": 29427 }, { "epoch": 0.03, "learning_rate": 4.989393937041967e-05, "loss": 0.8078, "step": 29428 }, { "epoch": 0.03, "learning_rate": 4.98939321426936e-05, "loss": 1.1222, "step": 29429 }, { "epoch": 0.03, "learning_rate": 4.989392491472177e-05, "loss": 1.047, "step": 29430 }, { "epoch": 0.03, "learning_rate": 4.98939176865042e-05, "loss": 0.9694, "step": 29431 }, { "epoch": 0.03, "learning_rate": 4.9893910458040894e-05, "loss": 0.7196, "step": 29432 }, { "epoch": 0.03, "learning_rate": 4.989390322933184e-05, "loss": 0.6405, "step": 29433 }, { "epoch": 0.03, "learning_rate": 4.989389600037705e-05, "loss": 0.651, "step": 29434 }, { "epoch": 0.03, "learning_rate": 4.989388877117651e-05, "loss": 0.4802, "step": 29435 }, { "epoch": 0.03, "learning_rate": 4.989388154173023e-05, "loss": 0.5764, "step": 29436 }, { "epoch": 0.03, "learning_rate": 4.989387431203821e-05, "loss": 0.5383, "step": 29437 }, { "epoch": 0.03, "learning_rate": 4.9893867082100454e-05, "loss": 0.6518, "step": 29438 }, { "epoch": 0.03, "learning_rate": 4.989385985191695e-05, "loss": 0.7855, "step": 29439 }, { "epoch": 0.03, "learning_rate": 4.98938526214877e-05, "loss": 0.9099, "step": 29440 }, { "epoch": 0.03, "learning_rate": 4.989384539081272e-05, "loss": 1.2821, "step": 29441 }, { "epoch": 0.03, "learning_rate": 4.989383815989199e-05, "loss": 0.9639, "step": 29442 }, { "epoch": 0.03, "learning_rate": 4.9893830928725516e-05, "loss": 1.1148, "step": 29443 }, { "epoch": 0.03, "learning_rate": 4.9893823697313304e-05, "loss": 0.9673, "step": 29444 }, { "epoch": 0.03, "learning_rate": 4.989381646565535e-05, "loss": 0.5411, "step": 29445 }, { "epoch": 0.03, "learning_rate": 4.989380923375166e-05, "loss": 1.6845, "step": 29446 }, { "epoch": 0.03, "learning_rate": 4.9893802001602226e-05, "loss": 1.2677, "step": 29447 }, { "epoch": 0.03, "learning_rate": 4.989379476920705e-05, "loss": 0.7428, "step": 29448 }, { "epoch": 0.03, "learning_rate": 4.989378753656613e-05, "loss": 0.7558, "step": 29449 }, { "epoch": 0.03, "learning_rate": 4.989378030367947e-05, "loss": 0.4081, "step": 29450 }, { "epoch": 0.03, "learning_rate": 4.989377307054707e-05, "loss": 0.9302, "step": 29451 }, { "epoch": 0.03, "learning_rate": 4.9893765837168934e-05, "loss": 0.9861, "step": 29452 }, { "epoch": 0.03, "learning_rate": 4.989375860354505e-05, "loss": 1.0523, "step": 29453 }, { "epoch": 0.03, "learning_rate": 4.9893751369675433e-05, "loss": 0.6775, "step": 29454 }, { "epoch": 0.03, "learning_rate": 4.9893744135560075e-05, "loss": 0.5099, "step": 29455 }, { "epoch": 0.03, "learning_rate": 4.989373690119896e-05, "loss": 1.2656, "step": 29456 }, { "epoch": 0.03, "learning_rate": 4.9893729666592126e-05, "loss": 0.5586, "step": 29457 }, { "epoch": 0.03, "learning_rate": 4.9893722431739544e-05, "loss": 0.9869, "step": 29458 }, { "epoch": 0.03, "learning_rate": 4.9893715196641213e-05, "loss": 0.9622, "step": 29459 }, { "epoch": 0.03, "learning_rate": 4.9893707961297155e-05, "loss": 1.0984, "step": 29460 }, { "epoch": 0.03, "learning_rate": 4.989370072570735e-05, "loss": 1.3423, "step": 29461 }, { "epoch": 0.03, "learning_rate": 4.98936934898718e-05, "loss": 1.3087, "step": 29462 }, { "epoch": 0.03, "learning_rate": 4.989368625379052e-05, "loss": 0.9516, "step": 29463 }, { "epoch": 0.03, "learning_rate": 4.989367901746349e-05, "loss": 0.7328, "step": 29464 }, { "epoch": 0.03, "learning_rate": 4.9893671780890727e-05, "loss": 0.6059, "step": 29465 }, { "epoch": 0.03, "learning_rate": 4.989366454407223e-05, "loss": 1.2643, "step": 29466 }, { "epoch": 0.03, "learning_rate": 4.989365730700798e-05, "loss": 1.1962, "step": 29467 }, { "epoch": 0.03, "learning_rate": 4.9893650069697994e-05, "loss": 0.9147, "step": 29468 }, { "epoch": 0.03, "learning_rate": 4.989364283214227e-05, "loss": 1.1372, "step": 29469 }, { "epoch": 0.03, "learning_rate": 4.989363559434081e-05, "loss": 0.9389, "step": 29470 }, { "epoch": 0.03, "learning_rate": 4.989362835629361e-05, "loss": 0.8516, "step": 29471 }, { "epoch": 0.03, "learning_rate": 4.9893621118000665e-05, "loss": 0.9088, "step": 29472 }, { "epoch": 0.03, "learning_rate": 4.9893613879461985e-05, "loss": 0.9734, "step": 29473 }, { "epoch": 0.03, "learning_rate": 4.9893606640677564e-05, "loss": 0.8195, "step": 29474 }, { "epoch": 0.03, "learning_rate": 4.989359940164741e-05, "loss": 1.0494, "step": 29475 }, { "epoch": 0.03, "learning_rate": 4.9893592162371504e-05, "loss": 1.0731, "step": 29476 }, { "epoch": 0.03, "learning_rate": 4.9893584922849866e-05, "loss": 1.0533, "step": 29477 }, { "epoch": 0.03, "learning_rate": 4.989357768308249e-05, "loss": 0.932, "step": 29478 }, { "epoch": 0.03, "learning_rate": 4.989357044306938e-05, "loss": 0.7419, "step": 29479 }, { "epoch": 0.03, "learning_rate": 4.9893563202810525e-05, "loss": 0.8148, "step": 29480 }, { "epoch": 0.03, "learning_rate": 4.989355596230593e-05, "loss": 1.437, "step": 29481 }, { "epoch": 0.03, "learning_rate": 4.989354872155559e-05, "loss": 1.2238, "step": 29482 }, { "epoch": 0.03, "learning_rate": 4.9893541480559526e-05, "loss": 0.8273, "step": 29483 }, { "epoch": 0.03, "learning_rate": 4.989353423931772e-05, "loss": 0.7372, "step": 29484 }, { "epoch": 0.03, "learning_rate": 4.989352699783017e-05, "loss": 1.0574, "step": 29485 }, { "epoch": 0.03, "learning_rate": 4.9893519756096884e-05, "loss": 1.2833, "step": 29486 }, { "epoch": 0.03, "learning_rate": 4.9893512514117854e-05, "loss": 1.1901, "step": 29487 }, { "epoch": 0.03, "learning_rate": 4.9893505271893097e-05, "loss": 1.1349, "step": 29488 }, { "epoch": 0.03, "learning_rate": 4.98934980294226e-05, "loss": 0.4467, "step": 29489 }, { "epoch": 0.03, "learning_rate": 4.989349078670635e-05, "loss": 0.9104, "step": 29490 }, { "epoch": 0.03, "learning_rate": 4.989348354374438e-05, "loss": 1.0715, "step": 29491 }, { "epoch": 0.03, "learning_rate": 4.989347630053667e-05, "loss": 1.1724, "step": 29492 }, { "epoch": 0.03, "learning_rate": 4.989346905708321e-05, "loss": 1.1202, "step": 29493 }, { "epoch": 0.03, "learning_rate": 4.989346181338402e-05, "loss": 1.0046, "step": 29494 }, { "epoch": 0.03, "learning_rate": 4.989345456943909e-05, "loss": 1.161, "step": 29495 }, { "epoch": 0.03, "learning_rate": 4.989344732524843e-05, "loss": 0.7851, "step": 29496 }, { "epoch": 0.03, "learning_rate": 4.989344008081203e-05, "loss": 0.9665, "step": 29497 }, { "epoch": 0.03, "learning_rate": 4.9893432836129885e-05, "loss": 0.9875, "step": 29498 }, { "epoch": 0.03, "learning_rate": 4.9893425591202e-05, "loss": 1.5392, "step": 29499 }, { "epoch": 0.03, "learning_rate": 4.9893418346028383e-05, "loss": 1.1259, "step": 29500 }, { "epoch": 0.03, "eval_loss": 1.0635902881622314, "eval_runtime": 130.3328, "eval_samples_per_second": 10.627, "eval_steps_per_second": 5.317, "step": 29500 }, { "epoch": 0.03, "learning_rate": 4.989341110060903e-05, "loss": 1.2886, "step": 29501 }, { "epoch": 0.03, "learning_rate": 4.9893403854943944e-05, "loss": 1.2668, "step": 29502 }, { "epoch": 0.03, "learning_rate": 4.9893396609033116e-05, "loss": 1.3232, "step": 29503 }, { "epoch": 0.03, "learning_rate": 4.9893389362876554e-05, "loss": 0.9349, "step": 29504 }, { "epoch": 0.03, "learning_rate": 4.989338211647425e-05, "loss": 1.0031, "step": 29505 }, { "epoch": 0.03, "learning_rate": 4.989337486982621e-05, "loss": 1.0684, "step": 29506 }, { "epoch": 0.03, "learning_rate": 4.989336762293243e-05, "loss": 1.2405, "step": 29507 }, { "epoch": 0.03, "learning_rate": 4.989336037579292e-05, "loss": 0.9994, "step": 29508 }, { "epoch": 0.03, "learning_rate": 4.989335312840767e-05, "loss": 1.1054, "step": 29509 }, { "epoch": 0.03, "learning_rate": 4.989334588077669e-05, "loss": 0.9814, "step": 29510 }, { "epoch": 0.03, "learning_rate": 4.989333863289996e-05, "loss": 0.9013, "step": 29511 }, { "epoch": 0.03, "learning_rate": 4.98933313847775e-05, "loss": 0.344, "step": 29512 }, { "epoch": 0.03, "learning_rate": 4.98933241364093e-05, "loss": 0.7135, "step": 29513 }, { "epoch": 0.03, "learning_rate": 4.9893316887795374e-05, "loss": 0.8927, "step": 29514 }, { "epoch": 0.03, "learning_rate": 4.9893309638935706e-05, "loss": 0.993, "step": 29515 }, { "epoch": 0.03, "learning_rate": 4.9893302389830296e-05, "loss": 1.1201, "step": 29516 }, { "epoch": 0.03, "learning_rate": 4.989329514047916e-05, "loss": 1.2187, "step": 29517 }, { "epoch": 0.03, "learning_rate": 4.989328789088228e-05, "loss": 1.1157, "step": 29518 }, { "epoch": 0.03, "learning_rate": 4.989328064103967e-05, "loss": 1.2772, "step": 29519 }, { "epoch": 0.03, "learning_rate": 4.989327339095132e-05, "loss": 0.8622, "step": 29520 }, { "epoch": 0.03, "learning_rate": 4.989326614061723e-05, "loss": 0.9657, "step": 29521 }, { "epoch": 0.03, "learning_rate": 4.989325889003741e-05, "loss": 1.0221, "step": 29522 }, { "epoch": 0.03, "learning_rate": 4.989325163921186e-05, "loss": 1.0889, "step": 29523 }, { "epoch": 0.03, "learning_rate": 4.989324438814056e-05, "loss": 1.3228, "step": 29524 }, { "epoch": 0.03, "learning_rate": 4.9893237136823536e-05, "loss": 0.7939, "step": 29525 }, { "epoch": 0.03, "learning_rate": 4.9893229885260775e-05, "loss": 0.98, "step": 29526 }, { "epoch": 0.03, "learning_rate": 4.9893222633452274e-05, "loss": 1.2033, "step": 29527 }, { "epoch": 0.03, "learning_rate": 4.989321538139804e-05, "loss": 1.0589, "step": 29528 }, { "epoch": 0.03, "learning_rate": 4.989320812909807e-05, "loss": 1.048, "step": 29529 }, { "epoch": 0.03, "learning_rate": 4.989320087655236e-05, "loss": 0.9237, "step": 29530 }, { "epoch": 0.03, "learning_rate": 4.989319362376092e-05, "loss": 1.0848, "step": 29531 }, { "epoch": 0.03, "learning_rate": 4.989318637072374e-05, "loss": 0.995, "step": 29532 }, { "epoch": 0.03, "learning_rate": 4.9893179117440835e-05, "loss": 1.0574, "step": 29533 }, { "epoch": 0.03, "learning_rate": 4.989317186391219e-05, "loss": 1.1163, "step": 29534 }, { "epoch": 0.03, "learning_rate": 4.989316461013781e-05, "loss": 0.9879, "step": 29535 }, { "epoch": 0.03, "learning_rate": 4.989315735611769e-05, "loss": 1.0952, "step": 29536 }, { "epoch": 0.03, "learning_rate": 4.989315010185185e-05, "loss": 1.2547, "step": 29537 }, { "epoch": 0.03, "learning_rate": 4.989314284734026e-05, "loss": 0.833, "step": 29538 }, { "epoch": 0.03, "learning_rate": 4.989313559258294e-05, "loss": 0.9863, "step": 29539 }, { "epoch": 0.03, "learning_rate": 4.989312833757989e-05, "loss": 0.7732, "step": 29540 }, { "epoch": 0.03, "learning_rate": 4.98931210823311e-05, "loss": 0.9732, "step": 29541 }, { "epoch": 0.03, "learning_rate": 4.989311382683658e-05, "loss": 0.9168, "step": 29542 }, { "epoch": 0.03, "learning_rate": 4.989310657109632e-05, "loss": 0.8702, "step": 29543 }, { "epoch": 0.03, "learning_rate": 4.9893099315110326e-05, "loss": 0.7823, "step": 29544 }, { "epoch": 0.03, "learning_rate": 4.9893092058878605e-05, "loss": 0.5991, "step": 29545 }, { "epoch": 0.03, "learning_rate": 4.989308480240115e-05, "loss": 0.8633, "step": 29546 }, { "epoch": 0.03, "learning_rate": 4.9893077545677946e-05, "loss": 1.0289, "step": 29547 }, { "epoch": 0.03, "learning_rate": 4.989307028870902e-05, "loss": 1.2133, "step": 29548 }, { "epoch": 0.03, "learning_rate": 4.9893063031494356e-05, "loss": 0.8713, "step": 29549 }, { "epoch": 0.03, "learning_rate": 4.989305577403396e-05, "loss": 0.7791, "step": 29550 }, { "epoch": 0.03, "learning_rate": 4.9893048516327835e-05, "loss": 1.1244, "step": 29551 }, { "epoch": 0.03, "learning_rate": 4.989304125837597e-05, "loss": 0.9407, "step": 29552 }, { "epoch": 0.03, "learning_rate": 4.9893034000178376e-05, "loss": 0.7906, "step": 29553 }, { "epoch": 0.03, "learning_rate": 4.989302674173505e-05, "loss": 0.8822, "step": 29554 }, { "epoch": 0.03, "learning_rate": 4.989301948304598e-05, "loss": 1.1319, "step": 29555 }, { "epoch": 0.03, "learning_rate": 4.9893012224111194e-05, "loss": 0.8197, "step": 29556 }, { "epoch": 0.03, "learning_rate": 4.989300496493066e-05, "loss": 0.4789, "step": 29557 }, { "epoch": 0.03, "learning_rate": 4.98929977055044e-05, "loss": 0.7079, "step": 29558 }, { "epoch": 0.03, "learning_rate": 4.98929904458324e-05, "loss": 0.9878, "step": 29559 }, { "epoch": 0.03, "learning_rate": 4.9892983185914664e-05, "loss": 0.9427, "step": 29560 }, { "epoch": 0.03, "learning_rate": 4.9892975925751205e-05, "loss": 0.9449, "step": 29561 }, { "epoch": 0.03, "learning_rate": 4.989296866534201e-05, "loss": 0.9034, "step": 29562 }, { "epoch": 0.03, "learning_rate": 4.989296140468708e-05, "loss": 1.5288, "step": 29563 }, { "epoch": 0.03, "learning_rate": 4.989295414378642e-05, "loss": 1.3956, "step": 29564 }, { "epoch": 0.03, "learning_rate": 4.989294688264002e-05, "loss": 0.5595, "step": 29565 }, { "epoch": 0.03, "learning_rate": 4.98929396212479e-05, "loss": 2.1128, "step": 29566 }, { "epoch": 0.03, "learning_rate": 4.989293235961004e-05, "loss": 1.4597, "step": 29567 }, { "epoch": 0.03, "learning_rate": 4.9892925097726446e-05, "loss": 0.8417, "step": 29568 }, { "epoch": 0.03, "learning_rate": 4.989291783559712e-05, "loss": 0.423, "step": 29569 }, { "epoch": 0.03, "learning_rate": 4.9892910573222065e-05, "loss": 0.4306, "step": 29570 }, { "epoch": 0.03, "learning_rate": 4.989290331060128e-05, "loss": 1.1656, "step": 29571 }, { "epoch": 0.03, "learning_rate": 4.989289604773475e-05, "loss": 2.3191, "step": 29572 }, { "epoch": 0.03, "learning_rate": 4.989288878462251e-05, "loss": 2.3073, "step": 29573 }, { "epoch": 0.03, "learning_rate": 4.989288152126451e-05, "loss": 2.3022, "step": 29574 }, { "epoch": 0.03, "learning_rate": 4.9892874257660796e-05, "loss": 2.3115, "step": 29575 }, { "epoch": 0.03, "learning_rate": 4.989286699381135e-05, "loss": 2.3062, "step": 29576 }, { "epoch": 0.03, "learning_rate": 4.9892859729716166e-05, "loss": 2.296, "step": 29577 }, { "epoch": 0.03, "learning_rate": 4.989285246537525e-05, "loss": 2.2944, "step": 29578 }, { "epoch": 0.03, "learning_rate": 4.9892845200788605e-05, "loss": 2.2922, "step": 29579 }, { "epoch": 0.03, "learning_rate": 4.989283793595623e-05, "loss": 2.2831, "step": 29580 }, { "epoch": 0.03, "learning_rate": 4.989283067087812e-05, "loss": 2.2653, "step": 29581 }, { "epoch": 0.03, "learning_rate": 4.989282340555428e-05, "loss": 2.2366, "step": 29582 }, { "epoch": 0.03, "learning_rate": 4.989281613998471e-05, "loss": 2.2177, "step": 29583 }, { "epoch": 0.03, "learning_rate": 4.98928088741694e-05, "loss": 2.1955, "step": 29584 }, { "epoch": 0.03, "learning_rate": 4.989280160810837e-05, "loss": 2.1125, "step": 29585 }, { "epoch": 0.03, "learning_rate": 4.989279434180161e-05, "loss": 1.15, "step": 29586 }, { "epoch": 0.03, "learning_rate": 4.9892787075249115e-05, "loss": 0.1784, "step": 29587 }, { "epoch": 0.03, "learning_rate": 4.989277980845088e-05, "loss": 0.0015, "step": 29588 }, { "epoch": 0.03, "learning_rate": 4.989277254140692e-05, "loss": 0.0011, "step": 29589 }, { "epoch": 0.03, "learning_rate": 4.989276527411724e-05, "loss": 0.0037, "step": 29590 }, { "epoch": 0.03, "learning_rate": 4.9892758006581815e-05, "loss": 0.0006, "step": 29591 }, { "epoch": 0.03, "learning_rate": 4.9892750738800664e-05, "loss": 0.7902, "step": 29592 }, { "epoch": 0.03, "learning_rate": 4.9892743470773785e-05, "loss": 1.4945, "step": 29593 }, { "epoch": 0.03, "learning_rate": 4.9892736202501166e-05, "loss": 1.5566, "step": 29594 }, { "epoch": 0.03, "learning_rate": 4.9892728933982825e-05, "loss": 0.6584, "step": 29595 }, { "epoch": 0.03, "learning_rate": 4.989272166521875e-05, "loss": 1.2504, "step": 29596 }, { "epoch": 0.03, "learning_rate": 4.989271439620895e-05, "loss": 1.072, "step": 29597 }, { "epoch": 0.03, "learning_rate": 4.989270712695341e-05, "loss": 0.9854, "step": 29598 }, { "epoch": 0.03, "learning_rate": 4.9892699857452145e-05, "loss": 0.9972, "step": 29599 }, { "epoch": 0.03, "learning_rate": 4.9892692587705146e-05, "loss": 0.9128, "step": 29600 }, { "epoch": 0.03, "learning_rate": 4.9892685317712426e-05, "loss": 1.0996, "step": 29601 }, { "epoch": 0.03, "learning_rate": 4.989267804747397e-05, "loss": 0.8003, "step": 29602 }, { "epoch": 0.03, "learning_rate": 4.989267077698978e-05, "loss": 1.0765, "step": 29603 }, { "epoch": 0.03, "learning_rate": 4.9892663506259866e-05, "loss": 0.962, "step": 29604 }, { "epoch": 0.03, "learning_rate": 4.989265623528422e-05, "loss": 1.5177, "step": 29605 }, { "epoch": 0.03, "learning_rate": 4.989264896406285e-05, "loss": 1.6021, "step": 29606 }, { "epoch": 0.03, "learning_rate": 4.9892641692595744e-05, "loss": 0.8687, "step": 29607 }, { "epoch": 0.03, "learning_rate": 4.989263442088291e-05, "loss": 1.244, "step": 29608 }, { "epoch": 0.03, "learning_rate": 4.989262714892434e-05, "loss": 0.8656, "step": 29609 }, { "epoch": 0.03, "learning_rate": 4.9892619876720045e-05, "loss": 0.8696, "step": 29610 }, { "epoch": 0.03, "learning_rate": 4.989261260427003e-05, "loss": 0.7742, "step": 29611 }, { "epoch": 0.03, "learning_rate": 4.989260533157427e-05, "loss": 1.1124, "step": 29612 }, { "epoch": 0.03, "learning_rate": 4.989259805863279e-05, "loss": 1.1395, "step": 29613 }, { "epoch": 0.03, "learning_rate": 4.989259078544558e-05, "loss": 1.173, "step": 29614 }, { "epoch": 0.03, "learning_rate": 4.989258351201264e-05, "loss": 0.8595, "step": 29615 }, { "epoch": 0.03, "learning_rate": 4.989257623833397e-05, "loss": 0.5478, "step": 29616 }, { "epoch": 0.03, "learning_rate": 4.989256896440957e-05, "loss": 0.6014, "step": 29617 }, { "epoch": 0.03, "learning_rate": 4.989256169023945e-05, "loss": 0.5618, "step": 29618 }, { "epoch": 0.03, "learning_rate": 4.989255441582359e-05, "loss": 0.46, "step": 29619 }, { "epoch": 0.03, "learning_rate": 4.9892547141162004e-05, "loss": 0.5252, "step": 29620 }, { "epoch": 0.03, "learning_rate": 4.989253986625469e-05, "loss": 0.4168, "step": 29621 }, { "epoch": 0.03, "learning_rate": 4.989253259110165e-05, "loss": 0.3783, "step": 29622 }, { "epoch": 0.03, "learning_rate": 4.989252531570287e-05, "loss": 0.4579, "step": 29623 }, { "epoch": 0.03, "learning_rate": 4.9892518040058376e-05, "loss": 0.3815, "step": 29624 }, { "epoch": 0.03, "learning_rate": 4.9892510764168145e-05, "loss": 0.4469, "step": 29625 }, { "epoch": 0.03, "learning_rate": 4.9892503488032194e-05, "loss": 0.4938, "step": 29626 }, { "epoch": 0.03, "learning_rate": 4.989249621165051e-05, "loss": 1.121, "step": 29627 }, { "epoch": 0.03, "learning_rate": 4.989248893502309e-05, "loss": 1.0638, "step": 29628 }, { "epoch": 0.03, "learning_rate": 4.9892481658149945e-05, "loss": 1.3473, "step": 29629 }, { "epoch": 0.03, "learning_rate": 4.989247438103108e-05, "loss": 0.8908, "step": 29630 }, { "epoch": 0.03, "learning_rate": 4.9892467103666486e-05, "loss": 1.2684, "step": 29631 }, { "epoch": 0.03, "learning_rate": 4.989245982605616e-05, "loss": 0.9739, "step": 29632 }, { "epoch": 0.03, "learning_rate": 4.989245254820011e-05, "loss": 0.903, "step": 29633 }, { "epoch": 0.03, "learning_rate": 4.989244527009833e-05, "loss": 1.1687, "step": 29634 }, { "epoch": 0.03, "learning_rate": 4.9892437991750815e-05, "loss": 1.2772, "step": 29635 }, { "epoch": 0.03, "learning_rate": 4.989243071315758e-05, "loss": 1.2203, "step": 29636 }, { "epoch": 0.03, "learning_rate": 4.989242343431862e-05, "loss": 1.042, "step": 29637 }, { "epoch": 0.03, "learning_rate": 4.9892416155233926e-05, "loss": 1.1775, "step": 29638 }, { "epoch": 0.03, "learning_rate": 4.989240887590351e-05, "loss": 1.2283, "step": 29639 }, { "epoch": 0.03, "learning_rate": 4.989240159632736e-05, "loss": 1.2469, "step": 29640 }, { "epoch": 0.03, "learning_rate": 4.989239431650549e-05, "loss": 1.0879, "step": 29641 }, { "epoch": 0.03, "learning_rate": 4.989238703643788e-05, "loss": 0.7164, "step": 29642 }, { "epoch": 0.03, "learning_rate": 4.989237975612456e-05, "loss": 0.4945, "step": 29643 }, { "epoch": 0.03, "learning_rate": 4.98923724755655e-05, "loss": 0.9401, "step": 29644 }, { "epoch": 0.03, "learning_rate": 4.989236519476072e-05, "loss": 0.7985, "step": 29645 }, { "epoch": 0.03, "learning_rate": 4.989235791371021e-05, "loss": 1.3772, "step": 29646 }, { "epoch": 0.03, "learning_rate": 4.9892350632413975e-05, "loss": 1.0971, "step": 29647 }, { "epoch": 0.03, "learning_rate": 4.9892343350872015e-05, "loss": 1.0775, "step": 29648 }, { "epoch": 0.03, "learning_rate": 4.989233606908432e-05, "loss": 0.4205, "step": 29649 }, { "epoch": 0.03, "learning_rate": 4.9892328787050905e-05, "loss": 1.2014, "step": 29650 }, { "epoch": 0.03, "learning_rate": 4.989232150477177e-05, "loss": 1.2737, "step": 29651 }, { "epoch": 0.03, "learning_rate": 4.989231422224689e-05, "loss": 1.0253, "step": 29652 }, { "epoch": 0.03, "learning_rate": 4.98923069394763e-05, "loss": 1.1168, "step": 29653 }, { "epoch": 0.03, "learning_rate": 4.989229965645998e-05, "loss": 0.8531, "step": 29654 }, { "epoch": 0.03, "learning_rate": 4.989229237319793e-05, "loss": 0.874, "step": 29655 }, { "epoch": 0.03, "learning_rate": 4.989228508969016e-05, "loss": 0.1529, "step": 29656 }, { "epoch": 0.03, "learning_rate": 4.989227780593666e-05, "loss": 0.2259, "step": 29657 }, { "epoch": 0.03, "learning_rate": 4.989227052193743e-05, "loss": 0.8595, "step": 29658 }, { "epoch": 0.03, "learning_rate": 4.989226323769248e-05, "loss": 0.8849, "step": 29659 }, { "epoch": 0.03, "learning_rate": 4.989225595320179e-05, "loss": 0.9432, "step": 29660 }, { "epoch": 0.03, "learning_rate": 4.989224866846539e-05, "loss": 0.8398, "step": 29661 }, { "epoch": 0.03, "learning_rate": 4.9892241383483266e-05, "loss": 0.9027, "step": 29662 }, { "epoch": 0.03, "learning_rate": 4.989223409825541e-05, "loss": 0.6957, "step": 29663 }, { "epoch": 0.03, "learning_rate": 4.989222681278183e-05, "loss": 0.809, "step": 29664 }, { "epoch": 0.03, "learning_rate": 4.989221952706252e-05, "loss": 1.0714, "step": 29665 }, { "epoch": 0.03, "learning_rate": 4.989221224109749e-05, "loss": 1.0642, "step": 29666 }, { "epoch": 0.03, "learning_rate": 4.989220495488673e-05, "loss": 1.0366, "step": 29667 }, { "epoch": 0.03, "learning_rate": 4.989219766843024e-05, "loss": 1.2124, "step": 29668 }, { "epoch": 0.03, "learning_rate": 4.989219038172803e-05, "loss": 1.3846, "step": 29669 }, { "epoch": 0.03, "learning_rate": 4.98921830947801e-05, "loss": 1.0565, "step": 29670 }, { "epoch": 0.03, "learning_rate": 4.989217580758644e-05, "loss": 1.0992, "step": 29671 }, { "epoch": 0.03, "learning_rate": 4.989216852014706e-05, "loss": 1.0438, "step": 29672 }, { "epoch": 0.03, "learning_rate": 4.989216123246195e-05, "loss": 0.7611, "step": 29673 }, { "epoch": 0.03, "learning_rate": 4.9892153944531116e-05, "loss": 1.1699, "step": 29674 }, { "epoch": 0.03, "learning_rate": 4.989214665635456e-05, "loss": 0.9703, "step": 29675 }, { "epoch": 0.03, "learning_rate": 4.989213936793227e-05, "loss": 1.1886, "step": 29676 }, { "epoch": 0.03, "learning_rate": 4.989213207926426e-05, "loss": 0.661, "step": 29677 }, { "epoch": 0.03, "learning_rate": 4.989212479035053e-05, "loss": 1.1065, "step": 29678 }, { "epoch": 0.03, "learning_rate": 4.989211750119107e-05, "loss": 0.9948, "step": 29679 }, { "epoch": 0.03, "learning_rate": 4.9892110211785894e-05, "loss": 0.9044, "step": 29680 }, { "epoch": 0.03, "learning_rate": 4.989210292213499e-05, "loss": 0.9745, "step": 29681 }, { "epoch": 0.03, "learning_rate": 4.989209563223836e-05, "loss": 0.8212, "step": 29682 }, { "epoch": 0.03, "learning_rate": 4.9892088342096e-05, "loss": 0.8004, "step": 29683 }, { "epoch": 0.03, "learning_rate": 4.989208105170792e-05, "loss": 1.0106, "step": 29684 }, { "epoch": 0.03, "learning_rate": 4.9892073761074115e-05, "loss": 0.5492, "step": 29685 }, { "epoch": 0.03, "learning_rate": 4.9892066470194586e-05, "loss": 0.911, "step": 29686 }, { "epoch": 0.03, "learning_rate": 4.989205917906934e-05, "loss": 0.9088, "step": 29687 }, { "epoch": 0.03, "learning_rate": 4.989205188769836e-05, "loss": 0.8455, "step": 29688 }, { "epoch": 0.03, "learning_rate": 4.989204459608166e-05, "loss": 1.0994, "step": 29689 }, { "epoch": 0.03, "learning_rate": 4.989203730421924e-05, "loss": 1.1868, "step": 29690 }, { "epoch": 0.03, "learning_rate": 4.989203001211109e-05, "loss": 1.067, "step": 29691 }, { "epoch": 0.03, "learning_rate": 4.989202271975722e-05, "loss": 1.1097, "step": 29692 }, { "epoch": 0.03, "learning_rate": 4.989201542715762e-05, "loss": 0.8404, "step": 29693 }, { "epoch": 0.03, "learning_rate": 4.9892008134312305e-05, "loss": 0.8055, "step": 29694 }, { "epoch": 0.03, "learning_rate": 4.989200084122127e-05, "loss": 1.1603, "step": 29695 }, { "epoch": 0.03, "learning_rate": 4.989199354788451e-05, "loss": 1.2454, "step": 29696 }, { "epoch": 0.03, "learning_rate": 4.9891986254302016e-05, "loss": 0.9969, "step": 29697 }, { "epoch": 0.03, "learning_rate": 4.989197896047381e-05, "loss": 0.7465, "step": 29698 }, { "epoch": 0.03, "learning_rate": 4.9891971666399875e-05, "loss": 0.4535, "step": 29699 }, { "epoch": 0.03, "learning_rate": 4.989196437208021e-05, "loss": 1.0462, "step": 29700 }, { "epoch": 0.03, "learning_rate": 4.9891957077514836e-05, "loss": 1.1903, "step": 29701 }, { "epoch": 0.03, "learning_rate": 4.9891949782703726e-05, "loss": 1.1135, "step": 29702 }, { "epoch": 0.03, "learning_rate": 4.98919424876469e-05, "loss": 1.1866, "step": 29703 }, { "epoch": 0.03, "learning_rate": 4.9891935192344355e-05, "loss": 0.9945, "step": 29704 }, { "epoch": 0.03, "learning_rate": 4.989192789679609e-05, "loss": 0.8417, "step": 29705 }, { "epoch": 0.03, "learning_rate": 4.9891920601002094e-05, "loss": 1.0148, "step": 29706 }, { "epoch": 0.03, "learning_rate": 4.989191330496238e-05, "loss": 0.8567, "step": 29707 }, { "epoch": 0.03, "learning_rate": 4.989190600867694e-05, "loss": 0.8136, "step": 29708 }, { "epoch": 0.03, "learning_rate": 4.989189871214578e-05, "loss": 0.8453, "step": 29709 }, { "epoch": 0.03, "learning_rate": 4.9891891415368895e-05, "loss": 1.0365, "step": 29710 }, { "epoch": 0.03, "learning_rate": 4.989188411834629e-05, "loss": 1.1308, "step": 29711 }, { "epoch": 0.03, "learning_rate": 4.989187682107796e-05, "loss": 1.4532, "step": 29712 }, { "epoch": 0.03, "learning_rate": 4.989186952356391e-05, "loss": 0.9661, "step": 29713 }, { "epoch": 0.03, "learning_rate": 4.9891862225804135e-05, "loss": 0.6396, "step": 29714 }, { "epoch": 0.03, "learning_rate": 4.989185492779864e-05, "loss": 0.854, "step": 29715 }, { "epoch": 0.03, "learning_rate": 4.989184762954743e-05, "loss": 0.9681, "step": 29716 }, { "epoch": 0.03, "learning_rate": 4.9891840331050486e-05, "loss": 0.8403, "step": 29717 }, { "epoch": 0.03, "learning_rate": 4.9891833032307835e-05, "loss": 0.9704, "step": 29718 }, { "epoch": 0.03, "learning_rate": 4.9891825733319456e-05, "loss": 1.1809, "step": 29719 }, { "epoch": 0.03, "learning_rate": 4.989181843408535e-05, "loss": 0.9255, "step": 29720 }, { "epoch": 0.03, "learning_rate": 4.989181113460552e-05, "loss": 1.1357, "step": 29721 }, { "epoch": 0.03, "learning_rate": 4.989180383487997e-05, "loss": 1.1232, "step": 29722 }, { "epoch": 0.03, "learning_rate": 4.989179653490871e-05, "loss": 0.8862, "step": 29723 }, { "epoch": 0.03, "learning_rate": 4.989178923469172e-05, "loss": 0.681, "step": 29724 }, { "epoch": 0.03, "learning_rate": 4.989178193422901e-05, "loss": 1.0773, "step": 29725 }, { "epoch": 0.03, "learning_rate": 4.9891774633520584e-05, "loss": 1.0559, "step": 29726 }, { "epoch": 0.03, "learning_rate": 4.989176733256643e-05, "loss": 0.9681, "step": 29727 }, { "epoch": 0.03, "learning_rate": 4.989176003136655e-05, "loss": 0.8999, "step": 29728 }, { "epoch": 0.03, "learning_rate": 4.989175272992096e-05, "loss": 1.0365, "step": 29729 }, { "epoch": 0.03, "learning_rate": 4.989174542822964e-05, "loss": 1.037, "step": 29730 }, { "epoch": 0.03, "learning_rate": 4.989173812629261e-05, "loss": 1.2131, "step": 29731 }, { "epoch": 0.03, "learning_rate": 4.989173082410985e-05, "loss": 0.8438, "step": 29732 }, { "epoch": 0.03, "learning_rate": 4.989172352168138e-05, "loss": 0.9713, "step": 29733 }, { "epoch": 0.03, "learning_rate": 4.989171621900718e-05, "loss": 1.2086, "step": 29734 }, { "epoch": 0.03, "learning_rate": 4.989170891608725e-05, "loss": 1.1198, "step": 29735 }, { "epoch": 0.03, "learning_rate": 4.989170161292162e-05, "loss": 1.0239, "step": 29736 }, { "epoch": 0.03, "learning_rate": 4.989169430951026e-05, "loss": 0.8525, "step": 29737 }, { "epoch": 0.03, "learning_rate": 4.989168700585318e-05, "loss": 0.9392, "step": 29738 }, { "epoch": 0.03, "learning_rate": 4.989167970195038e-05, "loss": 1.047, "step": 29739 }, { "epoch": 0.03, "learning_rate": 4.989167239780186e-05, "loss": 0.8601, "step": 29740 }, { "epoch": 0.03, "learning_rate": 4.989166509340762e-05, "loss": 1.0527, "step": 29741 }, { "epoch": 0.03, "learning_rate": 4.989165778876766e-05, "loss": 0.9109, "step": 29742 }, { "epoch": 0.03, "learning_rate": 4.989165048388198e-05, "loss": 0.59, "step": 29743 }, { "epoch": 0.03, "learning_rate": 4.989164317875058e-05, "loss": 0.9763, "step": 29744 }, { "epoch": 0.03, "learning_rate": 4.989163587337346e-05, "loss": 1.3292, "step": 29745 }, { "epoch": 0.03, "learning_rate": 4.989162856775062e-05, "loss": 1.0201, "step": 29746 }, { "epoch": 0.03, "learning_rate": 4.989162126188206e-05, "loss": 0.6978, "step": 29747 }, { "epoch": 0.03, "learning_rate": 4.989161395576778e-05, "loss": 0.788, "step": 29748 }, { "epoch": 0.03, "learning_rate": 4.9891606649407785e-05, "loss": 1.0291, "step": 29749 }, { "epoch": 0.03, "learning_rate": 4.9891599342802066e-05, "loss": 1.0938, "step": 29750 }, { "epoch": 0.03, "learning_rate": 4.9891592035950626e-05, "loss": 1.0706, "step": 29751 }, { "epoch": 0.03, "learning_rate": 4.989158472885347e-05, "loss": 1.0645, "step": 29752 }, { "epoch": 0.03, "learning_rate": 4.98915774215106e-05, "loss": 1.081, "step": 29753 }, { "epoch": 0.03, "learning_rate": 4.9891570113922e-05, "loss": 1.4387, "step": 29754 }, { "epoch": 0.03, "learning_rate": 4.989156280608769e-05, "loss": 0.8076, "step": 29755 }, { "epoch": 0.03, "learning_rate": 4.989155549800765e-05, "loss": 1.0695, "step": 29756 }, { "epoch": 0.03, "learning_rate": 4.98915481896819e-05, "loss": 0.9641, "step": 29757 }, { "epoch": 0.03, "learning_rate": 4.9891540881110436e-05, "loss": 0.5471, "step": 29758 }, { "epoch": 0.03, "learning_rate": 4.989153357229324e-05, "loss": 1.2235, "step": 29759 }, { "epoch": 0.03, "learning_rate": 4.9891526263230335e-05, "loss": 1.0957, "step": 29760 }, { "epoch": 0.03, "learning_rate": 4.9891518953921706e-05, "loss": 0.9794, "step": 29761 }, { "epoch": 0.03, "learning_rate": 4.9891511644367364e-05, "loss": 0.8785, "step": 29762 }, { "epoch": 0.03, "learning_rate": 4.989150433456729e-05, "loss": 0.4749, "step": 29763 }, { "epoch": 0.03, "learning_rate": 4.9891497024521516e-05, "loss": 0.9446, "step": 29764 }, { "epoch": 0.03, "learning_rate": 4.989148971423001e-05, "loss": 0.952, "step": 29765 }, { "epoch": 0.03, "learning_rate": 4.989148240369279e-05, "loss": 0.9412, "step": 29766 }, { "epoch": 0.03, "learning_rate": 4.989147509290986e-05, "loss": 0.8863, "step": 29767 }, { "epoch": 0.03, "learning_rate": 4.9891467781881204e-05, "loss": 0.9866, "step": 29768 }, { "epoch": 0.03, "learning_rate": 4.989146047060683e-05, "loss": 2.0329, "step": 29769 }, { "epoch": 0.03, "learning_rate": 4.989145315908673e-05, "loss": 1.9651, "step": 29770 }, { "epoch": 0.03, "learning_rate": 4.989144584732093e-05, "loss": 1.3415, "step": 29771 }, { "epoch": 0.03, "learning_rate": 4.98914385353094e-05, "loss": 1.1383, "step": 29772 }, { "epoch": 0.03, "learning_rate": 4.9891431223052154e-05, "loss": 0.9418, "step": 29773 }, { "epoch": 0.03, "learning_rate": 4.989142391054919e-05, "loss": 1.0114, "step": 29774 }, { "epoch": 0.03, "learning_rate": 4.989141659780051e-05, "loss": 0.9288, "step": 29775 }, { "epoch": 0.03, "learning_rate": 4.9891409284806114e-05, "loss": 0.929, "step": 29776 }, { "epoch": 0.03, "learning_rate": 4.9891401971566e-05, "loss": 0.945, "step": 29777 }, { "epoch": 0.03, "learning_rate": 4.989139465808016e-05, "loss": 1.0671, "step": 29778 }, { "epoch": 0.03, "learning_rate": 4.989138734434861e-05, "loss": 0.8631, "step": 29779 }, { "epoch": 0.03, "learning_rate": 4.989138003037134e-05, "loss": 1.0465, "step": 29780 }, { "epoch": 0.03, "learning_rate": 4.989137271614836e-05, "loss": 1.1379, "step": 29781 }, { "epoch": 0.03, "learning_rate": 4.989136540167966e-05, "loss": 0.6224, "step": 29782 }, { "epoch": 0.03, "learning_rate": 4.989135808696523e-05, "loss": 1.0469, "step": 29783 }, { "epoch": 0.03, "learning_rate": 4.98913507720051e-05, "loss": 0.8027, "step": 29784 }, { "epoch": 0.03, "learning_rate": 4.989134345679924e-05, "loss": 0.9904, "step": 29785 }, { "epoch": 0.03, "learning_rate": 4.989133614134768e-05, "loss": 0.9246, "step": 29786 }, { "epoch": 0.03, "learning_rate": 4.989132882565039e-05, "loss": 1.187, "step": 29787 }, { "epoch": 0.03, "learning_rate": 4.989132150970739e-05, "loss": 1.1243, "step": 29788 }, { "epoch": 0.03, "learning_rate": 4.989131419351867e-05, "loss": 1.0792, "step": 29789 }, { "epoch": 0.03, "learning_rate": 4.989130687708423e-05, "loss": 0.8464, "step": 29790 }, { "epoch": 0.03, "learning_rate": 4.989129956040408e-05, "loss": 0.9706, "step": 29791 }, { "epoch": 0.03, "learning_rate": 4.9891292243478204e-05, "loss": 0.8199, "step": 29792 }, { "epoch": 0.03, "learning_rate": 4.989128492630662e-05, "loss": 0.9078, "step": 29793 }, { "epoch": 0.03, "learning_rate": 4.989127760888932e-05, "loss": 1.1443, "step": 29794 }, { "epoch": 0.03, "learning_rate": 4.98912702912263e-05, "loss": 0.9329, "step": 29795 }, { "epoch": 0.03, "learning_rate": 4.989126297331757e-05, "loss": 0.7379, "step": 29796 }, { "epoch": 0.03, "learning_rate": 4.989125565516312e-05, "loss": 0.6958, "step": 29797 }, { "epoch": 0.03, "learning_rate": 4.9891248336762945e-05, "loss": 0.9879, "step": 29798 }, { "epoch": 0.03, "learning_rate": 4.9891241018117066e-05, "loss": 0.8614, "step": 29799 }, { "epoch": 0.03, "learning_rate": 4.989123369922547e-05, "loss": 0.7644, "step": 29800 }, { "epoch": 0.03, "learning_rate": 4.989122638008815e-05, "loss": 0.3576, "step": 29801 }, { "epoch": 0.03, "learning_rate": 4.9891219060705116e-05, "loss": 0.6654, "step": 29802 }, { "epoch": 0.03, "learning_rate": 4.9891211741076374e-05, "loss": 0.6553, "step": 29803 }, { "epoch": 0.03, "learning_rate": 4.989120442120192e-05, "loss": 0.8626, "step": 29804 }, { "epoch": 0.03, "learning_rate": 4.989119710108175e-05, "loss": 0.844, "step": 29805 }, { "epoch": 0.03, "learning_rate": 4.989118978071585e-05, "loss": 1.1689, "step": 29806 }, { "epoch": 0.03, "learning_rate": 4.989118246010424e-05, "loss": 0.9637, "step": 29807 }, { "epoch": 0.03, "learning_rate": 4.989117513924692e-05, "loss": 0.9207, "step": 29808 }, { "epoch": 0.03, "learning_rate": 4.989116781814388e-05, "loss": 1.0986, "step": 29809 }, { "epoch": 0.03, "learning_rate": 4.989116049679513e-05, "loss": 0.828, "step": 29810 }, { "epoch": 0.03, "learning_rate": 4.989115317520065e-05, "loss": 0.4433, "step": 29811 }, { "epoch": 0.03, "learning_rate": 4.989114585336048e-05, "loss": 0.4017, "step": 29812 }, { "epoch": 0.03, "learning_rate": 4.989113853127458e-05, "loss": 0.3764, "step": 29813 }, { "epoch": 0.03, "learning_rate": 4.989113120894296e-05, "loss": 0.4369, "step": 29814 }, { "epoch": 0.03, "learning_rate": 4.9891123886365644e-05, "loss": 0.3908, "step": 29815 }, { "epoch": 0.03, "learning_rate": 4.989111656354259e-05, "loss": 0.2859, "step": 29816 }, { "epoch": 0.03, "learning_rate": 4.9891109240473845e-05, "loss": 0.3888, "step": 29817 }, { "epoch": 0.03, "learning_rate": 4.9891101917159364e-05, "loss": 0.3302, "step": 29818 }, { "epoch": 0.03, "learning_rate": 4.989109459359918e-05, "loss": 0.2882, "step": 29819 }, { "epoch": 0.03, "learning_rate": 4.989108726979328e-05, "loss": 0.257, "step": 29820 }, { "epoch": 0.03, "learning_rate": 4.989107994574167e-05, "loss": 0.1577, "step": 29821 }, { "epoch": 0.03, "learning_rate": 4.989107262144434e-05, "loss": 0.2615, "step": 29822 }, { "epoch": 0.03, "learning_rate": 4.9891065296901296e-05, "loss": 0.3558, "step": 29823 }, { "epoch": 0.03, "learning_rate": 4.9891057972112544e-05, "loss": 0.877, "step": 29824 }, { "epoch": 0.03, "learning_rate": 4.989105064707807e-05, "loss": 0.8381, "step": 29825 }, { "epoch": 0.03, "learning_rate": 4.9891043321797884e-05, "loss": 0.8544, "step": 29826 }, { "epoch": 0.03, "learning_rate": 4.9891035996271983e-05, "loss": 0.903, "step": 29827 }, { "epoch": 0.03, "learning_rate": 4.9891028670500375e-05, "loss": 0.5066, "step": 29828 }, { "epoch": 0.03, "learning_rate": 4.9891021344483046e-05, "loss": 0.6762, "step": 29829 }, { "epoch": 0.03, "learning_rate": 4.989101401822e-05, "loss": 0.8143, "step": 29830 }, { "epoch": 0.03, "learning_rate": 4.989100669171125e-05, "loss": 0.6154, "step": 29831 }, { "epoch": 0.03, "learning_rate": 4.989099936495679e-05, "loss": 0.5266, "step": 29832 }, { "epoch": 0.03, "learning_rate": 4.989099203795661e-05, "loss": 0.9403, "step": 29833 }, { "epoch": 0.03, "learning_rate": 4.989098471071071e-05, "loss": 0.6278, "step": 29834 }, { "epoch": 0.03, "learning_rate": 4.989097738321911e-05, "loss": 0.5365, "step": 29835 }, { "epoch": 0.03, "learning_rate": 4.989097005548178e-05, "loss": 1.166, "step": 29836 }, { "epoch": 0.03, "learning_rate": 4.9890962727498746e-05, "loss": 1.0713, "step": 29837 }, { "epoch": 0.03, "learning_rate": 4.9890955399270004e-05, "loss": 1.037, "step": 29838 }, { "epoch": 0.03, "learning_rate": 4.989094807079554e-05, "loss": 1.0923, "step": 29839 }, { "epoch": 0.03, "learning_rate": 4.989094074207536e-05, "loss": 0.8037, "step": 29840 }, { "epoch": 0.03, "learning_rate": 4.9890933413109485e-05, "loss": 0.6719, "step": 29841 }, { "epoch": 0.03, "learning_rate": 4.989092608389788e-05, "loss": 1.012, "step": 29842 }, { "epoch": 0.03, "learning_rate": 4.9890918754440574e-05, "loss": 1.1544, "step": 29843 }, { "epoch": 0.03, "learning_rate": 4.989091142473755e-05, "loss": 0.2698, "step": 29844 }, { "epoch": 0.03, "learning_rate": 4.989090409478881e-05, "loss": 0.1179, "step": 29845 }, { "epoch": 0.03, "learning_rate": 4.9890896764594364e-05, "loss": 0.8736, "step": 29846 }, { "epoch": 0.03, "learning_rate": 4.98908894341542e-05, "loss": 1.1104, "step": 29847 }, { "epoch": 0.03, "learning_rate": 4.989088210346833e-05, "loss": 0.9849, "step": 29848 }, { "epoch": 0.03, "learning_rate": 4.989087477253674e-05, "loss": 1.4127, "step": 29849 }, { "epoch": 0.03, "learning_rate": 4.989086744135944e-05, "loss": 1.1487, "step": 29850 }, { "epoch": 0.03, "learning_rate": 4.989086010993643e-05, "loss": 0.8982, "step": 29851 }, { "epoch": 0.03, "learning_rate": 4.989085277826771e-05, "loss": 0.8402, "step": 29852 }, { "epoch": 0.03, "learning_rate": 4.989084544635327e-05, "loss": 1.1107, "step": 29853 }, { "epoch": 0.03, "learning_rate": 4.989083811419313e-05, "loss": 0.9165, "step": 29854 }, { "epoch": 0.03, "learning_rate": 4.989083078178727e-05, "loss": 1.0202, "step": 29855 }, { "epoch": 0.03, "learning_rate": 4.9890823449135706e-05, "loss": 0.9647, "step": 29856 }, { "epoch": 0.03, "learning_rate": 4.9890816116238423e-05, "loss": 1.1339, "step": 29857 }, { "epoch": 0.03, "learning_rate": 4.989080878309542e-05, "loss": 0.9164, "step": 29858 }, { "epoch": 0.03, "learning_rate": 4.989080144970672e-05, "loss": 0.9995, "step": 29859 }, { "epoch": 0.03, "learning_rate": 4.98907941160723e-05, "loss": 0.8818, "step": 29860 }, { "epoch": 0.03, "learning_rate": 4.989078678219218e-05, "loss": 1.1307, "step": 29861 }, { "epoch": 0.03, "learning_rate": 4.989077944806634e-05, "loss": 1.0852, "step": 29862 }, { "epoch": 0.03, "learning_rate": 4.989077211369478e-05, "loss": 1.1966, "step": 29863 }, { "epoch": 0.03, "learning_rate": 4.989076477907753e-05, "loss": 0.9594, "step": 29864 }, { "epoch": 0.03, "learning_rate": 4.989075744421455e-05, "loss": 1.0555, "step": 29865 }, { "epoch": 0.03, "learning_rate": 4.989075010910586e-05, "loss": 0.999, "step": 29866 }, { "epoch": 0.03, "learning_rate": 4.989074277375147e-05, "loss": 0.7432, "step": 29867 }, { "epoch": 0.03, "learning_rate": 4.989073543815136e-05, "loss": 1.222, "step": 29868 }, { "epoch": 0.03, "learning_rate": 4.989072810230555e-05, "loss": 0.9464, "step": 29869 }, { "epoch": 0.03, "learning_rate": 4.989072076621402e-05, "loss": 1.0938, "step": 29870 }, { "epoch": 0.03, "learning_rate": 4.989071342987678e-05, "loss": 0.9997, "step": 29871 }, { "epoch": 0.03, "learning_rate": 4.989070609329383e-05, "loss": 1.0004, "step": 29872 }, { "epoch": 0.03, "learning_rate": 4.989069875646517e-05, "loss": 0.7596, "step": 29873 }, { "epoch": 0.03, "learning_rate": 4.98906914193908e-05, "loss": 0.8992, "step": 29874 }, { "epoch": 0.03, "learning_rate": 4.989068408207072e-05, "loss": 1.2514, "step": 29875 }, { "epoch": 0.03, "learning_rate": 4.9890676744504936e-05, "loss": 1.0526, "step": 29876 }, { "epoch": 0.03, "learning_rate": 4.989066940669343e-05, "loss": 0.9493, "step": 29877 }, { "epoch": 0.03, "learning_rate": 4.989066206863622e-05, "loss": 0.9147, "step": 29878 }, { "epoch": 0.03, "learning_rate": 4.9890654730333295e-05, "loss": 1.3534, "step": 29879 }, { "epoch": 0.03, "learning_rate": 4.989064739178467e-05, "loss": 0.8961, "step": 29880 }, { "epoch": 0.03, "learning_rate": 4.989064005299032e-05, "loss": 1.0245, "step": 29881 }, { "epoch": 0.03, "learning_rate": 4.989063271395027e-05, "loss": 0.9607, "step": 29882 }, { "epoch": 0.03, "learning_rate": 4.9890625374664514e-05, "loss": 0.8638, "step": 29883 }, { "epoch": 0.03, "learning_rate": 4.989061803513304e-05, "loss": 1.1458, "step": 29884 }, { "epoch": 0.03, "learning_rate": 4.989061069535586e-05, "loss": 1.1629, "step": 29885 }, { "epoch": 0.03, "learning_rate": 4.9890603355332965e-05, "loss": 1.1175, "step": 29886 }, { "epoch": 0.03, "learning_rate": 4.9890596015064375e-05, "loss": 0.9372, "step": 29887 }, { "epoch": 0.03, "learning_rate": 4.9890588674550056e-05, "loss": 1.0803, "step": 29888 }, { "epoch": 0.03, "learning_rate": 4.9890581333790044e-05, "loss": 0.6242, "step": 29889 }, { "epoch": 0.03, "learning_rate": 4.989057399278432e-05, "loss": 0.335, "step": 29890 }, { "epoch": 0.03, "learning_rate": 4.989056665153288e-05, "loss": 0.9443, "step": 29891 }, { "epoch": 0.03, "learning_rate": 4.989055931003573e-05, "loss": 1.096, "step": 29892 }, { "epoch": 0.03, "learning_rate": 4.9890551968292875e-05, "loss": 1.0462, "step": 29893 }, { "epoch": 0.03, "learning_rate": 4.9890544626304306e-05, "loss": 1.0887, "step": 29894 }, { "epoch": 0.03, "learning_rate": 4.9890537284070037e-05, "loss": 0.9552, "step": 29895 }, { "epoch": 0.03, "learning_rate": 4.989052994159005e-05, "loss": 0.9143, "step": 29896 }, { "epoch": 0.03, "learning_rate": 4.989052259886436e-05, "loss": 1.0504, "step": 29897 }, { "epoch": 0.03, "learning_rate": 4.9890515255892964e-05, "loss": 1.1351, "step": 29898 }, { "epoch": 0.03, "learning_rate": 4.989050791267585e-05, "loss": 1.5387, "step": 29899 }, { "epoch": 0.03, "learning_rate": 4.989050056921304e-05, "loss": 0.9371, "step": 29900 }, { "epoch": 0.03, "learning_rate": 4.989049322550451e-05, "loss": 1.3925, "step": 29901 }, { "epoch": 0.03, "learning_rate": 4.989048588155028e-05, "loss": 1.0329, "step": 29902 }, { "epoch": 0.03, "learning_rate": 4.9890478537350335e-05, "loss": 0.8854, "step": 29903 }, { "epoch": 0.03, "learning_rate": 4.989047119290468e-05, "loss": 1.4424, "step": 29904 }, { "epoch": 0.03, "learning_rate": 4.989046384821332e-05, "loss": 0.6747, "step": 29905 }, { "epoch": 0.03, "learning_rate": 4.989045650327625e-05, "loss": 1.1638, "step": 29906 }, { "epoch": 0.03, "learning_rate": 4.9890449158093475e-05, "loss": 1.1735, "step": 29907 }, { "epoch": 0.03, "learning_rate": 4.9890441812665e-05, "loss": 0.6172, "step": 29908 }, { "epoch": 0.03, "learning_rate": 4.9890434466990804e-05, "loss": 1.0204, "step": 29909 }, { "epoch": 0.03, "learning_rate": 4.9890427121070905e-05, "loss": 0.8543, "step": 29910 }, { "epoch": 0.03, "learning_rate": 4.98904197749053e-05, "loss": 0.8891, "step": 29911 }, { "epoch": 0.03, "learning_rate": 4.9890412428493984e-05, "loss": 0.9005, "step": 29912 }, { "epoch": 0.03, "learning_rate": 4.9890405081836955e-05, "loss": 0.9355, "step": 29913 }, { "epoch": 0.03, "learning_rate": 4.9890397734934226e-05, "loss": 1.0309, "step": 29914 }, { "epoch": 0.03, "learning_rate": 4.989039038778579e-05, "loss": 0.845, "step": 29915 }, { "epoch": 0.03, "learning_rate": 4.989038304039164e-05, "loss": 1.028, "step": 29916 }, { "epoch": 0.03, "learning_rate": 4.989037569275179e-05, "loss": 0.871, "step": 29917 }, { "epoch": 0.03, "learning_rate": 4.989036834486624e-05, "loss": 0.9336, "step": 29918 }, { "epoch": 0.03, "learning_rate": 4.9890360996734965e-05, "loss": 0.5238, "step": 29919 }, { "epoch": 0.03, "learning_rate": 4.989035364835799e-05, "loss": 0.7927, "step": 29920 }, { "epoch": 0.03, "learning_rate": 4.989034629973531e-05, "loss": 0.8324, "step": 29921 }, { "epoch": 0.03, "learning_rate": 4.9890338950866925e-05, "loss": 1.119, "step": 29922 }, { "epoch": 0.03, "learning_rate": 4.989033160175283e-05, "loss": 0.9068, "step": 29923 }, { "epoch": 0.03, "learning_rate": 4.989032425239303e-05, "loss": 0.7505, "step": 29924 }, { "epoch": 0.03, "learning_rate": 4.989031690278752e-05, "loss": 0.8844, "step": 29925 }, { "epoch": 0.03, "learning_rate": 4.98903095529363e-05, "loss": 0.9936, "step": 29926 }, { "epoch": 0.03, "learning_rate": 4.989030220283938e-05, "loss": 1.1153, "step": 29927 }, { "epoch": 0.03, "learning_rate": 4.9890294852496746e-05, "loss": 0.8911, "step": 29928 }, { "epoch": 0.03, "learning_rate": 4.9890287501908414e-05, "loss": 1.2275, "step": 29929 }, { "epoch": 0.03, "learning_rate": 4.989028015107437e-05, "loss": 0.9951, "step": 29930 }, { "epoch": 0.03, "learning_rate": 4.989027279999462e-05, "loss": 0.865, "step": 29931 }, { "epoch": 0.03, "learning_rate": 4.9890265448669175e-05, "loss": 0.9361, "step": 29932 }, { "epoch": 0.03, "learning_rate": 4.9890258097098007e-05, "loss": 1.0378, "step": 29933 }, { "epoch": 0.03, "learning_rate": 4.9890250745281145e-05, "loss": 0.5105, "step": 29934 }, { "epoch": 0.03, "learning_rate": 4.989024339321857e-05, "loss": 0.3743, "step": 29935 }, { "epoch": 0.03, "learning_rate": 4.9890236040910286e-05, "loss": 0.3252, "step": 29936 }, { "epoch": 0.03, "learning_rate": 4.98902286883563e-05, "loss": 0.1323, "step": 29937 }, { "epoch": 0.03, "learning_rate": 4.989022133555661e-05, "loss": 0.2022, "step": 29938 }, { "epoch": 0.03, "learning_rate": 4.989021398251121e-05, "loss": 0.1352, "step": 29939 }, { "epoch": 0.03, "learning_rate": 4.9890206629220114e-05, "loss": 0.1152, "step": 29940 }, { "epoch": 0.03, "learning_rate": 4.98901992756833e-05, "loss": 0.5077, "step": 29941 }, { "epoch": 0.03, "learning_rate": 4.9890191921900794e-05, "loss": 0.4188, "step": 29942 }, { "epoch": 0.03, "learning_rate": 4.989018456787258e-05, "loss": 0.8156, "step": 29943 }, { "epoch": 0.03, "learning_rate": 4.989017721359865e-05, "loss": 1.4582, "step": 29944 }, { "epoch": 0.03, "learning_rate": 4.989016985907902e-05, "loss": 0.8036, "step": 29945 }, { "epoch": 0.03, "learning_rate": 4.989016250431369e-05, "loss": 1.075, "step": 29946 }, { "epoch": 0.03, "learning_rate": 4.9890155149302644e-05, "loss": 0.7815, "step": 29947 }, { "epoch": 0.03, "learning_rate": 4.98901477940459e-05, "loss": 1.0642, "step": 29948 }, { "epoch": 0.03, "learning_rate": 4.9890140438543456e-05, "loss": 0.2043, "step": 29949 }, { "epoch": 0.03, "learning_rate": 4.98901330827953e-05, "loss": 0.1035, "step": 29950 }, { "epoch": 0.03, "learning_rate": 4.989012572680143e-05, "loss": 0.1032, "step": 29951 }, { "epoch": 0.03, "learning_rate": 4.9890118370561866e-05, "loss": 0.1123, "step": 29952 }, { "epoch": 0.03, "learning_rate": 4.98901110140766e-05, "loss": 0.095, "step": 29953 }, { "epoch": 0.03, "learning_rate": 4.9890103657345624e-05, "loss": 0.0672, "step": 29954 }, { "epoch": 0.03, "learning_rate": 4.989009630036895e-05, "loss": 0.0599, "step": 29955 }, { "epoch": 0.03, "learning_rate": 4.989008894314656e-05, "loss": 0.0678, "step": 29956 }, { "epoch": 0.03, "learning_rate": 4.989008158567847e-05, "loss": 0.0629, "step": 29957 }, { "epoch": 0.03, "learning_rate": 4.989007422796468e-05, "loss": 0.0768, "step": 29958 }, { "epoch": 0.03, "learning_rate": 4.9890066870005184e-05, "loss": 0.0735, "step": 29959 }, { "epoch": 0.03, "learning_rate": 4.989005951179999e-05, "loss": 0.0684, "step": 29960 }, { "epoch": 0.03, "learning_rate": 4.9890052153349075e-05, "loss": 0.0601, "step": 29961 }, { "epoch": 0.03, "learning_rate": 4.989004479465247e-05, "loss": 0.0544, "step": 29962 }, { "epoch": 0.03, "learning_rate": 4.9890037435710156e-05, "loss": 0.0517, "step": 29963 }, { "epoch": 0.03, "learning_rate": 4.989003007652214e-05, "loss": 0.8741, "step": 29964 }, { "epoch": 0.03, "learning_rate": 4.989002271708842e-05, "loss": 1.17, "step": 29965 }, { "epoch": 0.03, "learning_rate": 4.9890015357408994e-05, "loss": 1.0292, "step": 29966 }, { "epoch": 0.03, "learning_rate": 4.9890007997483865e-05, "loss": 0.8326, "step": 29967 }, { "epoch": 0.03, "learning_rate": 4.9890000637313036e-05, "loss": 0.9014, "step": 29968 }, { "epoch": 0.03, "learning_rate": 4.98899932768965e-05, "loss": 1.0025, "step": 29969 }, { "epoch": 0.03, "learning_rate": 4.9889985916234256e-05, "loss": 0.858, "step": 29970 }, { "epoch": 0.03, "learning_rate": 4.988997855532631e-05, "loss": 1.3052, "step": 29971 }, { "epoch": 0.03, "learning_rate": 4.9889971194172666e-05, "loss": 0.9992, "step": 29972 }, { "epoch": 0.03, "learning_rate": 4.988996383277332e-05, "loss": 1.094, "step": 29973 }, { "epoch": 0.03, "learning_rate": 4.988995647112826e-05, "loss": 0.9455, "step": 29974 }, { "epoch": 0.03, "learning_rate": 4.988994910923751e-05, "loss": 1.3309, "step": 29975 }, { "epoch": 0.03, "learning_rate": 4.988994174710105e-05, "loss": 0.8915, "step": 29976 }, { "epoch": 0.03, "learning_rate": 4.9889934384718886e-05, "loss": 1.1632, "step": 29977 }, { "epoch": 0.03, "learning_rate": 4.988992702209102e-05, "loss": 0.9303, "step": 29978 }, { "epoch": 0.03, "learning_rate": 4.9889919659217455e-05, "loss": 0.6541, "step": 29979 }, { "epoch": 0.03, "learning_rate": 4.9889912296098186e-05, "loss": 0.9076, "step": 29980 }, { "epoch": 0.03, "learning_rate": 4.988990493273321e-05, "loss": 1.3652, "step": 29981 }, { "epoch": 0.03, "learning_rate": 4.988989756912253e-05, "loss": 0.9591, "step": 29982 }, { "epoch": 0.03, "learning_rate": 4.9889890205266154e-05, "loss": 0.9511, "step": 29983 }, { "epoch": 0.03, "learning_rate": 4.988988284116407e-05, "loss": 1.3999, "step": 29984 }, { "epoch": 0.03, "learning_rate": 4.988987547681629e-05, "loss": 1.8046, "step": 29985 }, { "epoch": 0.03, "learning_rate": 4.98898681122228e-05, "loss": 1.4365, "step": 29986 }, { "epoch": 0.03, "learning_rate": 4.9889860747383616e-05, "loss": 0.665, "step": 29987 }, { "epoch": 0.03, "learning_rate": 4.988985338229872e-05, "loss": 0.7023, "step": 29988 }, { "epoch": 0.03, "learning_rate": 4.9889846016968134e-05, "loss": 0.7238, "step": 29989 }, { "epoch": 0.03, "learning_rate": 4.988983865139184e-05, "loss": 1.0411, "step": 29990 }, { "epoch": 0.03, "learning_rate": 4.988983128556984e-05, "loss": 1.1723, "step": 29991 }, { "epoch": 0.03, "learning_rate": 4.9889823919502146e-05, "loss": 3.1638, "step": 29992 }, { "epoch": 0.03, "learning_rate": 4.988981655318874e-05, "loss": 5.9895, "step": 29993 }, { "epoch": 0.03, "learning_rate": 4.988980918662964e-05, "loss": 5.9314, "step": 29994 }, { "epoch": 0.03, "learning_rate": 4.9889801819824833e-05, "loss": 5.1085, "step": 29995 }, { "epoch": 0.03, "learning_rate": 4.988979445277433e-05, "loss": 1.0203, "step": 29996 }, { "epoch": 0.03, "learning_rate": 4.988978708547812e-05, "loss": 0.6293, "step": 29997 }, { "epoch": 0.03, "learning_rate": 4.9889779717936216e-05, "loss": 1.1096, "step": 29998 }, { "epoch": 0.03, "learning_rate": 4.98897723501486e-05, "loss": 1.2676, "step": 29999 }, { "epoch": 0.03, "learning_rate": 4.9889764982115295e-05, "loss": 0.9795, "step": 30000 }, { "epoch": 0.03, "eval_loss": 1.067270040512085, "eval_runtime": 125.8764, "eval_samples_per_second": 11.003, "eval_steps_per_second": 5.505, "step": 30000 }, { "epoch": 0.03, "learning_rate": 4.988975761383628e-05, "loss": 0.8561, "step": 30001 }, { "epoch": 0.03, "learning_rate": 4.9889750245311564e-05, "loss": 0.7536, "step": 30002 }, { "epoch": 0.03, "learning_rate": 4.988974287654115e-05, "loss": 0.9649, "step": 30003 }, { "epoch": 0.03, "learning_rate": 4.988973550752504e-05, "loss": 0.6546, "step": 30004 }, { "epoch": 0.03, "learning_rate": 4.9889728138263215e-05, "loss": 0.9403, "step": 30005 }, { "epoch": 0.03, "learning_rate": 4.98897207687557e-05, "loss": 0.992, "step": 30006 }, { "epoch": 0.03, "learning_rate": 4.988971339900248e-05, "loss": 1.0207, "step": 30007 }, { "epoch": 0.03, "learning_rate": 4.988970602900356e-05, "loss": 1.3156, "step": 30008 }, { "epoch": 0.03, "learning_rate": 4.988969865875894e-05, "loss": 1.1153, "step": 30009 }, { "epoch": 0.03, "learning_rate": 4.988969128826862e-05, "loss": 0.8449, "step": 30010 }, { "epoch": 0.03, "learning_rate": 4.9889683917532595e-05, "loss": 0.9724, "step": 30011 }, { "epoch": 0.03, "learning_rate": 4.988967654655088e-05, "loss": 1.0259, "step": 30012 }, { "epoch": 0.03, "learning_rate": 4.988966917532345e-05, "loss": 0.9029, "step": 30013 }, { "epoch": 0.03, "learning_rate": 4.988966180385033e-05, "loss": 1.072, "step": 30014 }, { "epoch": 0.03, "learning_rate": 4.98896544321315e-05, "loss": 1.3351, "step": 30015 }, { "epoch": 0.03, "learning_rate": 4.9889647060166986e-05, "loss": 1.1017, "step": 30016 }, { "epoch": 0.03, "learning_rate": 4.9889639687956756e-05, "loss": 1.0928, "step": 30017 }, { "epoch": 0.03, "learning_rate": 4.988963231550083e-05, "loss": 0.9602, "step": 30018 }, { "epoch": 0.03, "learning_rate": 4.988962494279922e-05, "loss": 0.928, "step": 30019 }, { "epoch": 0.03, "learning_rate": 4.988961756985189e-05, "loss": 1.2747, "step": 30020 }, { "epoch": 0.03, "learning_rate": 4.988961019665886e-05, "loss": 0.977, "step": 30021 }, { "epoch": 0.03, "learning_rate": 4.988960282322015e-05, "loss": 1.3526, "step": 30022 }, { "epoch": 0.03, "learning_rate": 4.988959544953572e-05, "loss": 1.0309, "step": 30023 }, { "epoch": 0.03, "learning_rate": 4.9889588075605596e-05, "loss": 0.9657, "step": 30024 }, { "epoch": 0.03, "learning_rate": 4.988958070142978e-05, "loss": 1.1161, "step": 30025 }, { "epoch": 0.03, "learning_rate": 4.988957332700826e-05, "loss": 1.197, "step": 30026 }, { "epoch": 0.03, "learning_rate": 4.988956595234104e-05, "loss": 1.6176, "step": 30027 }, { "epoch": 0.03, "learning_rate": 4.988955857742812e-05, "loss": 0.8076, "step": 30028 }, { "epoch": 0.03, "learning_rate": 4.98895512022695e-05, "loss": 1.0706, "step": 30029 }, { "epoch": 0.03, "learning_rate": 4.988954382686518e-05, "loss": 0.9241, "step": 30030 }, { "epoch": 0.03, "learning_rate": 4.988953645121517e-05, "loss": 0.7848, "step": 30031 }, { "epoch": 0.03, "learning_rate": 4.9889529075319454e-05, "loss": 0.968, "step": 30032 }, { "epoch": 0.03, "learning_rate": 4.9889521699178035e-05, "loss": 0.9676, "step": 30033 }, { "epoch": 0.03, "learning_rate": 4.988951432279092e-05, "loss": 0.635, "step": 30034 }, { "epoch": 0.03, "learning_rate": 4.98895069461581e-05, "loss": 0.6899, "step": 30035 }, { "epoch": 0.03, "learning_rate": 4.9889499569279597e-05, "loss": 1.1009, "step": 30036 }, { "epoch": 0.03, "learning_rate": 4.988949219215538e-05, "loss": 1.3362, "step": 30037 }, { "epoch": 0.03, "learning_rate": 4.988948481478548e-05, "loss": 0.9635, "step": 30038 }, { "epoch": 0.03, "learning_rate": 4.988947743716987e-05, "loss": 0.8819, "step": 30039 }, { "epoch": 0.03, "learning_rate": 4.9889470059308565e-05, "loss": 0.8034, "step": 30040 }, { "epoch": 0.03, "learning_rate": 4.988946268120156e-05, "loss": 0.9142, "step": 30041 }, { "epoch": 0.03, "learning_rate": 4.988945530284886e-05, "loss": 1.2261, "step": 30042 }, { "epoch": 0.03, "learning_rate": 4.9889447924250455e-05, "loss": 1.7407, "step": 30043 }, { "epoch": 0.03, "learning_rate": 4.9889440545406365e-05, "loss": 1.4434, "step": 30044 }, { "epoch": 0.03, "learning_rate": 4.988943316631656e-05, "loss": 1.3505, "step": 30045 }, { "epoch": 0.03, "learning_rate": 4.988942578698107e-05, "loss": 0.8519, "step": 30046 }, { "epoch": 0.03, "learning_rate": 4.9889418407399876e-05, "loss": 1.3265, "step": 30047 }, { "epoch": 0.03, "learning_rate": 4.988941102757298e-05, "loss": 1.1079, "step": 30048 }, { "epoch": 0.03, "learning_rate": 4.9889403647500396e-05, "loss": 1.2416, "step": 30049 }, { "epoch": 0.03, "learning_rate": 4.988939626718211e-05, "loss": 1.285, "step": 30050 }, { "epoch": 0.03, "learning_rate": 4.988938888661813e-05, "loss": 0.5974, "step": 30051 }, { "epoch": 0.03, "learning_rate": 4.988938150580845e-05, "loss": 0.1412, "step": 30052 }, { "epoch": 0.03, "learning_rate": 4.9889374124753065e-05, "loss": 0.8011, "step": 30053 }, { "epoch": 0.03, "learning_rate": 4.9889366743451996e-05, "loss": 1.0467, "step": 30054 }, { "epoch": 0.03, "learning_rate": 4.988935936190522e-05, "loss": 0.794, "step": 30055 }, { "epoch": 0.03, "learning_rate": 4.988935198011275e-05, "loss": 1.0841, "step": 30056 }, { "epoch": 0.03, "learning_rate": 4.988934459807458e-05, "loss": 0.871, "step": 30057 }, { "epoch": 0.03, "learning_rate": 4.988933721579071e-05, "loss": 1.0036, "step": 30058 }, { "epoch": 0.03, "learning_rate": 4.988932983326116e-05, "loss": 0.9519, "step": 30059 }, { "epoch": 0.03, "learning_rate": 4.988932245048589e-05, "loss": 0.9271, "step": 30060 }, { "epoch": 0.03, "learning_rate": 4.988931506746494e-05, "loss": 0.7764, "step": 30061 }, { "epoch": 0.03, "learning_rate": 4.9889307684198286e-05, "loss": 0.9126, "step": 30062 }, { "epoch": 0.03, "learning_rate": 4.988930030068594e-05, "loss": 1.0509, "step": 30063 }, { "epoch": 0.03, "learning_rate": 4.9889292916927885e-05, "loss": 0.8942, "step": 30064 }, { "epoch": 0.03, "learning_rate": 4.9889285532924144e-05, "loss": 1.6981, "step": 30065 }, { "epoch": 0.03, "learning_rate": 4.988927814867471e-05, "loss": 0.997, "step": 30066 }, { "epoch": 0.03, "learning_rate": 4.9889270764179574e-05, "loss": 0.942, "step": 30067 }, { "epoch": 0.03, "learning_rate": 4.988926337943874e-05, "loss": 0.7465, "step": 30068 }, { "epoch": 0.03, "learning_rate": 4.988925599445221e-05, "loss": 0.983, "step": 30069 }, { "epoch": 0.03, "learning_rate": 4.9889248609219985e-05, "loss": 0.9202, "step": 30070 }, { "epoch": 0.03, "learning_rate": 4.988924122374207e-05, "loss": 1.1255, "step": 30071 }, { "epoch": 0.03, "learning_rate": 4.9889233838018443e-05, "loss": 0.8071, "step": 30072 }, { "epoch": 0.03, "learning_rate": 4.988922645204913e-05, "loss": 1.0476, "step": 30073 }, { "epoch": 0.03, "learning_rate": 4.988921906583413e-05, "loss": 1.4412, "step": 30074 }, { "epoch": 0.03, "learning_rate": 4.988921167937342e-05, "loss": 0.7119, "step": 30075 }, { "epoch": 0.03, "learning_rate": 4.988920429266702e-05, "loss": 0.7946, "step": 30076 }, { "epoch": 0.03, "learning_rate": 4.988919690571492e-05, "loss": 0.7659, "step": 30077 }, { "epoch": 0.03, "learning_rate": 4.9889189518517134e-05, "loss": 1.1851, "step": 30078 }, { "epoch": 0.03, "learning_rate": 4.988918213107364e-05, "loss": 1.2242, "step": 30079 }, { "epoch": 0.03, "learning_rate": 4.988917474338446e-05, "loss": 0.8172, "step": 30080 }, { "epoch": 0.03, "learning_rate": 4.988916735544958e-05, "loss": 0.5037, "step": 30081 }, { "epoch": 0.03, "learning_rate": 4.988915996726901e-05, "loss": 0.6559, "step": 30082 }, { "epoch": 0.03, "learning_rate": 4.988915257884274e-05, "loss": 0.6095, "step": 30083 }, { "epoch": 0.03, "learning_rate": 4.9889145190170775e-05, "loss": 0.5905, "step": 30084 }, { "epoch": 0.03, "learning_rate": 4.988913780125312e-05, "loss": 0.6398, "step": 30085 }, { "epoch": 0.03, "learning_rate": 4.988913041208976e-05, "loss": 0.6047, "step": 30086 }, { "epoch": 0.03, "learning_rate": 4.988912302268071e-05, "loss": 0.7, "step": 30087 }, { "epoch": 0.03, "learning_rate": 4.9889115633025964e-05, "loss": 0.5251, "step": 30088 }, { "epoch": 0.03, "learning_rate": 4.9889108243125525e-05, "loss": 0.4846, "step": 30089 }, { "epoch": 0.03, "learning_rate": 4.988910085297939e-05, "loss": 0.9276, "step": 30090 }, { "epoch": 0.03, "learning_rate": 4.9889093462587566e-05, "loss": 1.2737, "step": 30091 }, { "epoch": 0.03, "learning_rate": 4.988908607195004e-05, "loss": 1.2495, "step": 30092 }, { "epoch": 0.03, "learning_rate": 4.988907868106682e-05, "loss": 1.2025, "step": 30093 }, { "epoch": 0.03, "learning_rate": 4.9889071289937904e-05, "loss": 1.408, "step": 30094 }, { "epoch": 0.03, "learning_rate": 4.98890638985633e-05, "loss": 1.1675, "step": 30095 }, { "epoch": 0.03, "learning_rate": 4.9889056506943e-05, "loss": 1.1656, "step": 30096 }, { "epoch": 0.03, "learning_rate": 4.9889049115077005e-05, "loss": 1.404, "step": 30097 }, { "epoch": 0.03, "learning_rate": 4.9889041722965316e-05, "loss": 1.0117, "step": 30098 }, { "epoch": 0.03, "learning_rate": 4.988903433060793e-05, "loss": 1.0372, "step": 30099 }, { "epoch": 0.03, "learning_rate": 4.988902693800485e-05, "loss": 1.3365, "step": 30100 }, { "epoch": 0.03, "learning_rate": 4.988901954515608e-05, "loss": 1.1179, "step": 30101 }, { "epoch": 0.03, "learning_rate": 4.988901215206161e-05, "loss": 0.9046, "step": 30102 }, { "epoch": 0.03, "learning_rate": 4.9889004758721455e-05, "loss": 0.6692, "step": 30103 }, { "epoch": 0.03, "learning_rate": 4.9888997365135596e-05, "loss": 0.8716, "step": 30104 }, { "epoch": 0.03, "learning_rate": 4.988898997130405e-05, "loss": 1.0247, "step": 30105 }, { "epoch": 0.03, "learning_rate": 4.9888982577226804e-05, "loss": 1.0656, "step": 30106 }, { "epoch": 0.03, "learning_rate": 4.988897518290387e-05, "loss": 1.2995, "step": 30107 }, { "epoch": 0.03, "learning_rate": 4.988896778833525e-05, "loss": 0.8465, "step": 30108 }, { "epoch": 0.03, "learning_rate": 4.988896039352092e-05, "loss": 0.8802, "step": 30109 }, { "epoch": 0.03, "learning_rate": 4.9888952998460915e-05, "loss": 0.6234, "step": 30110 }, { "epoch": 0.03, "learning_rate": 4.98889456031552e-05, "loss": 0.8553, "step": 30111 }, { "epoch": 0.03, "learning_rate": 4.9888938207603804e-05, "loss": 0.7358, "step": 30112 }, { "epoch": 0.03, "learning_rate": 4.988893081180671e-05, "loss": 0.7124, "step": 30113 }, { "epoch": 0.03, "learning_rate": 4.988892341576392e-05, "loss": 1.1041, "step": 30114 }, { "epoch": 0.03, "learning_rate": 4.988891601947544e-05, "loss": 0.8704, "step": 30115 }, { "epoch": 0.03, "learning_rate": 4.9888908622941265e-05, "loss": 0.892, "step": 30116 }, { "epoch": 0.03, "learning_rate": 4.98889012261614e-05, "loss": 1.0348, "step": 30117 }, { "epoch": 0.03, "learning_rate": 4.988889382913584e-05, "loss": 1.0378, "step": 30118 }, { "epoch": 0.03, "learning_rate": 4.988888643186459e-05, "loss": 1.0489, "step": 30119 }, { "epoch": 0.03, "learning_rate": 4.9888879034347645e-05, "loss": 1.3287, "step": 30120 }, { "epoch": 0.03, "learning_rate": 4.988887163658501e-05, "loss": 0.9195, "step": 30121 }, { "epoch": 0.03, "learning_rate": 4.988886423857668e-05, "loss": 1.4397, "step": 30122 }, { "epoch": 0.03, "learning_rate": 4.988885684032266e-05, "loss": 1.2695, "step": 30123 }, { "epoch": 0.03, "learning_rate": 4.9888849441822946e-05, "loss": 0.6715, "step": 30124 }, { "epoch": 0.03, "learning_rate": 4.988884204307754e-05, "loss": 1.033, "step": 30125 }, { "epoch": 0.03, "learning_rate": 4.9888834644086444e-05, "loss": 1.0234, "step": 30126 }, { "epoch": 0.03, "learning_rate": 4.988882724484965e-05, "loss": 1.1189, "step": 30127 }, { "epoch": 0.03, "learning_rate": 4.988881984536717e-05, "loss": 1.0796, "step": 30128 }, { "epoch": 0.03, "learning_rate": 4.988881244563899e-05, "loss": 0.9846, "step": 30129 }, { "epoch": 0.03, "learning_rate": 4.988880504566513e-05, "loss": 0.564, "step": 30130 }, { "epoch": 0.03, "learning_rate": 4.988879764544557e-05, "loss": 0.5768, "step": 30131 }, { "epoch": 0.03, "learning_rate": 4.988879024498032e-05, "loss": 0.5614, "step": 30132 }, { "epoch": 0.03, "learning_rate": 4.988878284426938e-05, "loss": 0.6401, "step": 30133 }, { "epoch": 0.03, "learning_rate": 4.9888775443312746e-05, "loss": 0.8961, "step": 30134 }, { "epoch": 0.03, "learning_rate": 4.988876804211042e-05, "loss": 1.0739, "step": 30135 }, { "epoch": 0.03, "learning_rate": 4.98887606406624e-05, "loss": 0.8521, "step": 30136 }, { "epoch": 0.03, "learning_rate": 4.9888753238968696e-05, "loss": 0.6555, "step": 30137 }, { "epoch": 0.03, "learning_rate": 4.98887458370293e-05, "loss": 1.0566, "step": 30138 }, { "epoch": 0.03, "learning_rate": 4.9888738434844204e-05, "loss": 0.9956, "step": 30139 }, { "epoch": 0.03, "learning_rate": 4.988873103241343e-05, "loss": 1.115, "step": 30140 }, { "epoch": 0.03, "learning_rate": 4.9888723629736957e-05, "loss": 1.0433, "step": 30141 }, { "epoch": 0.03, "learning_rate": 4.9888716226814796e-05, "loss": 1.0275, "step": 30142 }, { "epoch": 0.03, "learning_rate": 4.988870882364695e-05, "loss": 0.9016, "step": 30143 }, { "epoch": 0.03, "learning_rate": 4.98887014202334e-05, "loss": 1.2042, "step": 30144 }, { "epoch": 0.03, "learning_rate": 4.9888694016574164e-05, "loss": 1.0766, "step": 30145 }, { "epoch": 0.03, "learning_rate": 4.9888686612669235e-05, "loss": 1.0458, "step": 30146 }, { "epoch": 0.03, "learning_rate": 4.988867920851862e-05, "loss": 0.9327, "step": 30147 }, { "epoch": 0.03, "learning_rate": 4.988867180412231e-05, "loss": 1.0657, "step": 30148 }, { "epoch": 0.03, "learning_rate": 4.988866439948031e-05, "loss": 1.3534, "step": 30149 }, { "epoch": 0.03, "learning_rate": 4.988865699459263e-05, "loss": 1.3681, "step": 30150 }, { "epoch": 0.03, "learning_rate": 4.988864958945924e-05, "loss": 0.9059, "step": 30151 }, { "epoch": 0.03, "learning_rate": 4.988864218408018e-05, "loss": 0.777, "step": 30152 }, { "epoch": 0.03, "learning_rate": 4.988863477845542e-05, "loss": 1.0503, "step": 30153 }, { "epoch": 0.03, "learning_rate": 4.9888627372584975e-05, "loss": 1.2866, "step": 30154 }, { "epoch": 0.03, "learning_rate": 4.988861996646883e-05, "loss": 1.1146, "step": 30155 }, { "epoch": 0.03, "learning_rate": 4.988861256010701e-05, "loss": 0.9585, "step": 30156 }, { "epoch": 0.03, "learning_rate": 4.988860515349948e-05, "loss": 0.7846, "step": 30157 }, { "epoch": 0.03, "learning_rate": 4.9888597746646276e-05, "loss": 0.8215, "step": 30158 }, { "epoch": 0.03, "learning_rate": 4.988859033954738e-05, "loss": 0.9612, "step": 30159 }, { "epoch": 0.03, "learning_rate": 4.988858293220279e-05, "loss": 1.2376, "step": 30160 }, { "epoch": 0.03, "learning_rate": 4.988857552461251e-05, "loss": 0.7119, "step": 30161 }, { "epoch": 0.03, "learning_rate": 4.9888568116776544e-05, "loss": 1.0639, "step": 30162 }, { "epoch": 0.03, "learning_rate": 4.988856070869489e-05, "loss": 1.3227, "step": 30163 }, { "epoch": 0.03, "learning_rate": 4.988855330036754e-05, "loss": 0.8215, "step": 30164 }, { "epoch": 0.03, "learning_rate": 4.98885458917945e-05, "loss": 0.6373, "step": 30165 }, { "epoch": 0.03, "learning_rate": 4.988853848297578e-05, "loss": 0.5602, "step": 30166 }, { "epoch": 0.03, "learning_rate": 4.988853107391136e-05, "loss": 0.4477, "step": 30167 }, { "epoch": 0.03, "learning_rate": 4.988852366460126e-05, "loss": 0.5147, "step": 30168 }, { "epoch": 0.03, "learning_rate": 4.988851625504547e-05, "loss": 0.5639, "step": 30169 }, { "epoch": 0.03, "learning_rate": 4.988850884524399e-05, "loss": 0.348, "step": 30170 }, { "epoch": 0.03, "learning_rate": 4.988850143519682e-05, "loss": 0.831, "step": 30171 }, { "epoch": 0.03, "learning_rate": 4.988849402490397e-05, "loss": 1.08, "step": 30172 }, { "epoch": 0.03, "learning_rate": 4.9888486614365414e-05, "loss": 1.1954, "step": 30173 }, { "epoch": 0.03, "learning_rate": 4.988847920358118e-05, "loss": 1.1147, "step": 30174 }, { "epoch": 0.03, "learning_rate": 4.9888471792551255e-05, "loss": 0.9505, "step": 30175 }, { "epoch": 0.03, "learning_rate": 4.988846438127565e-05, "loss": 1.1232, "step": 30176 }, { "epoch": 0.03, "learning_rate": 4.9888456969754335e-05, "loss": 0.8701, "step": 30177 }, { "epoch": 0.03, "learning_rate": 4.988844955798735e-05, "loss": 0.9474, "step": 30178 }, { "epoch": 0.03, "learning_rate": 4.988844214597467e-05, "loss": 0.9673, "step": 30179 }, { "epoch": 0.03, "learning_rate": 4.9888434733716305e-05, "loss": 0.8828, "step": 30180 }, { "epoch": 0.03, "learning_rate": 4.988842732121225e-05, "loss": 0.8849, "step": 30181 }, { "epoch": 0.03, "learning_rate": 4.988841990846251e-05, "loss": 1.1111, "step": 30182 }, { "epoch": 0.03, "learning_rate": 4.988841249546708e-05, "loss": 1.004, "step": 30183 }, { "epoch": 0.03, "learning_rate": 4.988840508222596e-05, "loss": 1.0423, "step": 30184 }, { "epoch": 0.03, "learning_rate": 4.988839766873915e-05, "loss": 0.8431, "step": 30185 }, { "epoch": 0.03, "learning_rate": 4.9888390255006656e-05, "loss": 0.946, "step": 30186 }, { "epoch": 0.03, "learning_rate": 4.988838284102848e-05, "loss": 0.79, "step": 30187 }, { "epoch": 0.03, "learning_rate": 4.9888375426804615e-05, "loss": 0.7173, "step": 30188 }, { "epoch": 0.03, "learning_rate": 4.988836801233505e-05, "loss": 1.0165, "step": 30189 }, { "epoch": 0.03, "learning_rate": 4.988836059761981e-05, "loss": 1.2401, "step": 30190 }, { "epoch": 0.03, "learning_rate": 4.9888353182658874e-05, "loss": 0.8419, "step": 30191 }, { "epoch": 0.03, "learning_rate": 4.988834576745226e-05, "loss": 0.8808, "step": 30192 }, { "epoch": 0.03, "learning_rate": 4.9888338351999956e-05, "loss": 0.8797, "step": 30193 }, { "epoch": 0.03, "learning_rate": 4.988833093630195e-05, "loss": 1.0948, "step": 30194 }, { "epoch": 0.03, "learning_rate": 4.988832352035828e-05, "loss": 1.037, "step": 30195 }, { "epoch": 0.03, "learning_rate": 4.988831610416891e-05, "loss": 0.9667, "step": 30196 }, { "epoch": 0.03, "learning_rate": 4.9888308687733856e-05, "loss": 1.2562, "step": 30197 }, { "epoch": 0.03, "learning_rate": 4.988830127105311e-05, "loss": 1.1892, "step": 30198 }, { "epoch": 0.03, "learning_rate": 4.988829385412668e-05, "loss": 1.059, "step": 30199 }, { "epoch": 0.03, "learning_rate": 4.9888286436954566e-05, "loss": 0.6695, "step": 30200 }, { "epoch": 0.03, "learning_rate": 4.9888279019536765e-05, "loss": 0.7663, "step": 30201 }, { "epoch": 0.03, "learning_rate": 4.9888271601873276e-05, "loss": 0.6876, "step": 30202 }, { "epoch": 0.03, "learning_rate": 4.98882641839641e-05, "loss": 0.9289, "step": 30203 }, { "epoch": 0.03, "learning_rate": 4.988825676580924e-05, "loss": 1.1614, "step": 30204 }, { "epoch": 0.03, "learning_rate": 4.988824934740869e-05, "loss": 0.1559, "step": 30205 }, { "epoch": 0.03, "learning_rate": 4.988824192876246e-05, "loss": 0.5756, "step": 30206 }, { "epoch": 0.03, "learning_rate": 4.9888234509870535e-05, "loss": 1.3487, "step": 30207 }, { "epoch": 0.03, "learning_rate": 4.988822709073293e-05, "loss": 1.3009, "step": 30208 }, { "epoch": 0.03, "learning_rate": 4.988821967134964e-05, "loss": 1.135, "step": 30209 }, { "epoch": 0.03, "learning_rate": 4.9888212251720656e-05, "loss": 0.7026, "step": 30210 }, { "epoch": 0.03, "learning_rate": 4.9888204831846e-05, "loss": 1.128, "step": 30211 }, { "epoch": 0.03, "learning_rate": 4.9888197411725646e-05, "loss": 1.335, "step": 30212 }, { "epoch": 0.03, "learning_rate": 4.9888189991359614e-05, "loss": 0.7835, "step": 30213 }, { "epoch": 0.03, "learning_rate": 4.988818257074789e-05, "loss": 1.2684, "step": 30214 }, { "epoch": 0.03, "learning_rate": 4.988817514989048e-05, "loss": 0.8821, "step": 30215 }, { "epoch": 0.03, "learning_rate": 4.988816772878738e-05, "loss": 0.8863, "step": 30216 }, { "epoch": 0.03, "learning_rate": 4.988816030743861e-05, "loss": 0.783, "step": 30217 }, { "epoch": 0.03, "learning_rate": 4.988815288584414e-05, "loss": 0.9566, "step": 30218 }, { "epoch": 0.03, "learning_rate": 4.9888145464003995e-05, "loss": 0.742, "step": 30219 }, { "epoch": 0.03, "learning_rate": 4.988813804191816e-05, "loss": 1.0893, "step": 30220 }, { "epoch": 0.03, "learning_rate": 4.988813061958664e-05, "loss": 1.2597, "step": 30221 }, { "epoch": 0.03, "learning_rate": 4.988812319700943e-05, "loss": 0.8772, "step": 30222 }, { "epoch": 0.03, "learning_rate": 4.988811577418654e-05, "loss": 0.3902, "step": 30223 }, { "epoch": 0.03, "learning_rate": 4.988810835111797e-05, "loss": 1.0076, "step": 30224 }, { "epoch": 0.03, "learning_rate": 4.9888100927803706e-05, "loss": 1.2914, "step": 30225 }, { "epoch": 0.03, "learning_rate": 4.9888093504243764e-05, "loss": 1.1535, "step": 30226 }, { "epoch": 0.03, "learning_rate": 4.9888086080438134e-05, "loss": 0.8601, "step": 30227 }, { "epoch": 0.03, "learning_rate": 4.988807865638682e-05, "loss": 0.5766, "step": 30228 }, { "epoch": 0.03, "learning_rate": 4.9888071232089814e-05, "loss": 0.7543, "step": 30229 }, { "epoch": 0.03, "learning_rate": 4.988806380754714e-05, "loss": 0.5239, "step": 30230 }, { "epoch": 0.03, "learning_rate": 4.988805638275877e-05, "loss": 0.5673, "step": 30231 }, { "epoch": 0.03, "learning_rate": 4.9888048957724716e-05, "loss": 0.8708, "step": 30232 }, { "epoch": 0.03, "learning_rate": 4.988804153244498e-05, "loss": 1.587, "step": 30233 }, { "epoch": 0.03, "learning_rate": 4.988803410691956e-05, "loss": 0.7112, "step": 30234 }, { "epoch": 0.03, "learning_rate": 4.988802668114845e-05, "loss": 1.0927, "step": 30235 }, { "epoch": 0.03, "learning_rate": 4.988801925513166e-05, "loss": 1.027, "step": 30236 }, { "epoch": 0.03, "learning_rate": 4.9888011828869184e-05, "loss": 0.9851, "step": 30237 }, { "epoch": 0.03, "learning_rate": 4.988800440236103e-05, "loss": 1.024, "step": 30238 }, { "epoch": 0.03, "learning_rate": 4.9887996975607185e-05, "loss": 0.8952, "step": 30239 }, { "epoch": 0.03, "learning_rate": 4.988798954860766e-05, "loss": 1.168, "step": 30240 }, { "epoch": 0.03, "learning_rate": 4.988798212136245e-05, "loss": 1.1928, "step": 30241 }, { "epoch": 0.03, "learning_rate": 4.9887974693871565e-05, "loss": 1.0965, "step": 30242 }, { "epoch": 0.03, "learning_rate": 4.988796726613499e-05, "loss": 0.8589, "step": 30243 }, { "epoch": 0.03, "learning_rate": 4.988795983815273e-05, "loss": 0.9743, "step": 30244 }, { "epoch": 0.03, "learning_rate": 4.988795240992479e-05, "loss": 1.1529, "step": 30245 }, { "epoch": 0.03, "learning_rate": 4.9887944981451154e-05, "loss": 1.2218, "step": 30246 }, { "epoch": 0.03, "learning_rate": 4.9887937552731846e-05, "loss": 0.8461, "step": 30247 }, { "epoch": 0.03, "learning_rate": 4.988793012376686e-05, "loss": 0.8287, "step": 30248 }, { "epoch": 0.03, "learning_rate": 4.9887922694556175e-05, "loss": 1.0073, "step": 30249 }, { "epoch": 0.03, "learning_rate": 4.988791526509981e-05, "loss": 0.8315, "step": 30250 }, { "epoch": 0.03, "learning_rate": 4.988790783539777e-05, "loss": 0.9716, "step": 30251 }, { "epoch": 0.03, "learning_rate": 4.988790040545005e-05, "loss": 1.0512, "step": 30252 }, { "epoch": 0.03, "learning_rate": 4.9887892975256644e-05, "loss": 0.9775, "step": 30253 }, { "epoch": 0.03, "learning_rate": 4.9887885544817554e-05, "loss": 0.744, "step": 30254 }, { "epoch": 0.03, "learning_rate": 4.988787811413278e-05, "loss": 1.4164, "step": 30255 }, { "epoch": 0.03, "learning_rate": 4.988787068320232e-05, "loss": 1.6704, "step": 30256 }, { "epoch": 0.03, "learning_rate": 4.988786325202618e-05, "loss": 1.2049, "step": 30257 }, { "epoch": 0.03, "learning_rate": 4.9887855820604365e-05, "loss": 1.2972, "step": 30258 }, { "epoch": 0.03, "learning_rate": 4.988784838893686e-05, "loss": 0.9573, "step": 30259 }, { "epoch": 0.03, "learning_rate": 4.9887840957023675e-05, "loss": 0.9949, "step": 30260 }, { "epoch": 0.03, "learning_rate": 4.98878335248648e-05, "loss": 1.1524, "step": 30261 }, { "epoch": 0.03, "learning_rate": 4.988782609246026e-05, "loss": 0.9001, "step": 30262 }, { "epoch": 0.03, "learning_rate": 4.988781865981002e-05, "loss": 1.1154, "step": 30263 }, { "epoch": 0.03, "learning_rate": 4.9887811226914106e-05, "loss": 1.31, "step": 30264 }, { "epoch": 0.03, "learning_rate": 4.9887803793772514e-05, "loss": 1.0048, "step": 30265 }, { "epoch": 0.03, "learning_rate": 4.9887796360385234e-05, "loss": 1.419, "step": 30266 }, { "epoch": 0.03, "learning_rate": 4.9887788926752274e-05, "loss": 1.0197, "step": 30267 }, { "epoch": 0.03, "learning_rate": 4.9887781492873634e-05, "loss": 1.3006, "step": 30268 }, { "epoch": 0.03, "learning_rate": 4.9887774058749307e-05, "loss": 1.4185, "step": 30269 }, { "epoch": 0.03, "learning_rate": 4.988776662437929e-05, "loss": 1.0588, "step": 30270 }, { "epoch": 0.03, "learning_rate": 4.988775918976361e-05, "loss": 0.9786, "step": 30271 }, { "epoch": 0.03, "learning_rate": 4.9887751754902244e-05, "loss": 0.867, "step": 30272 }, { "epoch": 0.03, "learning_rate": 4.9887744319795196e-05, "loss": 1.1618, "step": 30273 }, { "epoch": 0.03, "learning_rate": 4.988773688444246e-05, "loss": 1.2939, "step": 30274 }, { "epoch": 0.03, "learning_rate": 4.988772944884405e-05, "loss": 0.9406, "step": 30275 }, { "epoch": 0.03, "learning_rate": 4.988772201299996e-05, "loss": 0.8931, "step": 30276 }, { "epoch": 0.03, "learning_rate": 4.988771457691018e-05, "loss": 1.3811, "step": 30277 }, { "epoch": 0.03, "learning_rate": 4.9887707140574726e-05, "loss": 1.0532, "step": 30278 }, { "epoch": 0.03, "learning_rate": 4.988769970399358e-05, "loss": 0.9081, "step": 30279 }, { "epoch": 0.03, "learning_rate": 4.9887692267166766e-05, "loss": 0.3936, "step": 30280 }, { "epoch": 0.03, "learning_rate": 4.988768483009427e-05, "loss": 0.2737, "step": 30281 }, { "epoch": 0.03, "learning_rate": 4.9887677392776087e-05, "loss": 0.7237, "step": 30282 }, { "epoch": 0.03, "learning_rate": 4.9887669955212236e-05, "loss": 1.9378, "step": 30283 }, { "epoch": 0.03, "learning_rate": 4.9887662517402686e-05, "loss": 0.97, "step": 30284 }, { "epoch": 0.03, "learning_rate": 4.9887655079347475e-05, "loss": 1.0471, "step": 30285 }, { "epoch": 0.03, "learning_rate": 4.988764764104657e-05, "loss": 1.0743, "step": 30286 }, { "epoch": 0.03, "learning_rate": 4.9887640202499986e-05, "loss": 1.0297, "step": 30287 }, { "epoch": 0.03, "learning_rate": 4.988763276370773e-05, "loss": 0.964, "step": 30288 }, { "epoch": 0.03, "learning_rate": 4.9887625324669784e-05, "loss": 0.9504, "step": 30289 }, { "epoch": 0.03, "learning_rate": 4.9887617885386165e-05, "loss": 1.1814, "step": 30290 }, { "epoch": 0.03, "learning_rate": 4.988761044585686e-05, "loss": 0.9997, "step": 30291 }, { "epoch": 0.03, "learning_rate": 4.9887603006081874e-05, "loss": 0.9505, "step": 30292 }, { "epoch": 0.03, "learning_rate": 4.9887595566061215e-05, "loss": 0.81, "step": 30293 }, { "epoch": 0.03, "learning_rate": 4.988758812579487e-05, "loss": 0.7603, "step": 30294 }, { "epoch": 0.03, "learning_rate": 4.988758068528285e-05, "loss": 0.9263, "step": 30295 }, { "epoch": 0.03, "learning_rate": 4.988757324452514e-05, "loss": 0.5048, "step": 30296 }, { "epoch": 0.03, "learning_rate": 4.9887565803521763e-05, "loss": 0.7699, "step": 30297 }, { "epoch": 0.03, "learning_rate": 4.98875583622727e-05, "loss": 0.9971, "step": 30298 }, { "epoch": 0.03, "learning_rate": 4.9887550920777956e-05, "loss": 1.2109, "step": 30299 }, { "epoch": 0.03, "learning_rate": 4.988754347903755e-05, "loss": 0.9692, "step": 30300 }, { "epoch": 0.03, "learning_rate": 4.988753603705144e-05, "loss": 1.2812, "step": 30301 }, { "epoch": 0.03, "learning_rate": 4.988752859481967e-05, "loss": 1.3916, "step": 30302 }, { "epoch": 0.03, "learning_rate": 4.9887521152342207e-05, "loss": 0.8942, "step": 30303 }, { "epoch": 0.03, "learning_rate": 4.988751370961907e-05, "loss": 1.3513, "step": 30304 }, { "epoch": 0.03, "learning_rate": 4.988750626665025e-05, "loss": 0.8628, "step": 30305 }, { "epoch": 0.03, "learning_rate": 4.988749882343576e-05, "loss": 1.2382, "step": 30306 }, { "epoch": 0.03, "learning_rate": 4.988749137997558e-05, "loss": 1.1532, "step": 30307 }, { "epoch": 0.03, "learning_rate": 4.988748393626973e-05, "loss": 0.9638, "step": 30308 }, { "epoch": 0.03, "learning_rate": 4.98874764923182e-05, "loss": 1.3169, "step": 30309 }, { "epoch": 0.03, "learning_rate": 4.9887469048120994e-05, "loss": 1.0043, "step": 30310 }, { "epoch": 0.03, "learning_rate": 4.98874616036781e-05, "loss": 0.8142, "step": 30311 }, { "epoch": 0.03, "learning_rate": 4.9887454158989535e-05, "loss": 1.01, "step": 30312 }, { "epoch": 0.03, "learning_rate": 4.9887446714055285e-05, "loss": 1.4629, "step": 30313 }, { "epoch": 0.03, "learning_rate": 4.988743926887537e-05, "loss": 0.9614, "step": 30314 }, { "epoch": 0.03, "learning_rate": 4.988743182344976e-05, "loss": 1.092, "step": 30315 }, { "epoch": 0.03, "learning_rate": 4.988742437777848e-05, "loss": 0.9066, "step": 30316 }, { "epoch": 0.03, "learning_rate": 4.988741693186152e-05, "loss": 1.1068, "step": 30317 }, { "epoch": 0.03, "learning_rate": 4.988740948569889e-05, "loss": 1.0629, "step": 30318 }, { "epoch": 0.03, "learning_rate": 4.988740203929057e-05, "loss": 0.796, "step": 30319 }, { "epoch": 0.03, "learning_rate": 4.988739459263658e-05, "loss": 0.9685, "step": 30320 }, { "epoch": 0.03, "learning_rate": 4.988738714573691e-05, "loss": 0.3947, "step": 30321 }, { "epoch": 0.03, "learning_rate": 4.9887379698591564e-05, "loss": 0.2907, "step": 30322 }, { "epoch": 0.03, "learning_rate": 4.988737225120054e-05, "loss": 1.1765, "step": 30323 }, { "epoch": 0.03, "learning_rate": 4.988736480356383e-05, "loss": 1.0742, "step": 30324 }, { "epoch": 0.03, "learning_rate": 4.988735735568145e-05, "loss": 1.1261, "step": 30325 }, { "epoch": 0.03, "learning_rate": 4.988734990755339e-05, "loss": 1.1172, "step": 30326 }, { "epoch": 0.03, "learning_rate": 4.988734245917965e-05, "loss": 1.0789, "step": 30327 }, { "epoch": 0.03, "learning_rate": 4.988733501056024e-05, "loss": 1.1008, "step": 30328 }, { "epoch": 0.03, "learning_rate": 4.988732756169515e-05, "loss": 1.0198, "step": 30329 }, { "epoch": 0.03, "learning_rate": 4.988732011258438e-05, "loss": 1.0322, "step": 30330 }, { "epoch": 0.03, "learning_rate": 4.9887312663227935e-05, "loss": 0.8964, "step": 30331 }, { "epoch": 0.03, "learning_rate": 4.9887305213625815e-05, "loss": 0.6941, "step": 30332 }, { "epoch": 0.03, "learning_rate": 4.9887297763778016e-05, "loss": 1.0092, "step": 30333 }, { "epoch": 0.03, "learning_rate": 4.988729031368453e-05, "loss": 0.958, "step": 30334 }, { "epoch": 0.03, "learning_rate": 4.988728286334538e-05, "loss": 1.2494, "step": 30335 }, { "epoch": 0.03, "learning_rate": 4.988727541276055e-05, "loss": 1.1272, "step": 30336 }, { "epoch": 0.03, "learning_rate": 4.988726796193005e-05, "loss": 1.1704, "step": 30337 }, { "epoch": 0.03, "learning_rate": 4.988726051085386e-05, "loss": 0.9745, "step": 30338 }, { "epoch": 0.03, "learning_rate": 4.9887253059532e-05, "loss": 1.1959, "step": 30339 }, { "epoch": 0.03, "learning_rate": 4.9887245607964466e-05, "loss": 1.0604, "step": 30340 }, { "epoch": 0.03, "learning_rate": 4.988723815615125e-05, "loss": 0.9175, "step": 30341 }, { "epoch": 0.03, "learning_rate": 4.9887230704092364e-05, "loss": 1.0695, "step": 30342 }, { "epoch": 0.03, "learning_rate": 4.9887223251787796e-05, "loss": 1.019, "step": 30343 }, { "epoch": 0.03, "learning_rate": 4.9887215799237555e-05, "loss": 1.1072, "step": 30344 }, { "epoch": 0.03, "learning_rate": 4.988720834644164e-05, "loss": 1.0298, "step": 30345 }, { "epoch": 0.03, "learning_rate": 4.9887200893400045e-05, "loss": 0.8788, "step": 30346 }, { "epoch": 0.03, "learning_rate": 4.9887193440112776e-05, "loss": 0.7512, "step": 30347 }, { "epoch": 0.03, "learning_rate": 4.9887185986579834e-05, "loss": 0.7697, "step": 30348 }, { "epoch": 0.03, "learning_rate": 4.988717853280121e-05, "loss": 1.3264, "step": 30349 }, { "epoch": 0.03, "learning_rate": 4.988717107877691e-05, "loss": 0.8916, "step": 30350 }, { "epoch": 0.03, "learning_rate": 4.988716362450694e-05, "loss": 1.1013, "step": 30351 }, { "epoch": 0.03, "learning_rate": 4.988715616999129e-05, "loss": 1.1604, "step": 30352 }, { "epoch": 0.03, "learning_rate": 4.988714871522997e-05, "loss": 1.1834, "step": 30353 }, { "epoch": 0.03, "learning_rate": 4.9887141260222966e-05, "loss": 0.7964, "step": 30354 }, { "epoch": 0.03, "learning_rate": 4.988713380497029e-05, "loss": 0.9598, "step": 30355 }, { "epoch": 0.03, "learning_rate": 4.988712634947195e-05, "loss": 1.1074, "step": 30356 }, { "epoch": 0.03, "learning_rate": 4.988711889372792e-05, "loss": 1.3051, "step": 30357 }, { "epoch": 0.03, "learning_rate": 4.988711143773822e-05, "loss": 0.7754, "step": 30358 }, { "epoch": 0.03, "learning_rate": 4.988710398150284e-05, "loss": 0.861, "step": 30359 }, { "epoch": 0.03, "learning_rate": 4.988709652502179e-05, "loss": 1.189, "step": 30360 }, { "epoch": 0.03, "learning_rate": 4.988708906829507e-05, "loss": 1.0141, "step": 30361 }, { "epoch": 0.03, "learning_rate": 4.9887081611322665e-05, "loss": 0.897, "step": 30362 }, { "epoch": 0.03, "learning_rate": 4.988707415410459e-05, "loss": 0.8908, "step": 30363 }, { "epoch": 0.03, "learning_rate": 4.988706669664084e-05, "loss": 1.6998, "step": 30364 }, { "epoch": 0.03, "learning_rate": 4.988705923893141e-05, "loss": 1.6928, "step": 30365 }, { "epoch": 0.03, "learning_rate": 4.9887051780976315e-05, "loss": 1.213, "step": 30366 }, { "epoch": 0.03, "learning_rate": 4.9887044322775544e-05, "loss": 1.1294, "step": 30367 }, { "epoch": 0.03, "learning_rate": 4.988703686432909e-05, "loss": 0.9543, "step": 30368 }, { "epoch": 0.03, "learning_rate": 4.9887029405636975e-05, "loss": 1.3386, "step": 30369 }, { "epoch": 0.03, "learning_rate": 4.9887021946699177e-05, "loss": 0.9312, "step": 30370 }, { "epoch": 0.03, "learning_rate": 4.9887014487515705e-05, "loss": 0.9858, "step": 30371 }, { "epoch": 0.03, "learning_rate": 4.988700702808656e-05, "loss": 1.2342, "step": 30372 }, { "epoch": 0.03, "learning_rate": 4.988699956841174e-05, "loss": 0.9419, "step": 30373 }, { "epoch": 0.03, "learning_rate": 4.988699210849125e-05, "loss": 1.0845, "step": 30374 }, { "epoch": 0.03, "learning_rate": 4.9886984648325084e-05, "loss": 1.1061, "step": 30375 }, { "epoch": 0.03, "learning_rate": 4.9886977187913245e-05, "loss": 1.157, "step": 30376 }, { "epoch": 0.03, "learning_rate": 4.9886969727255726e-05, "loss": 1.0333, "step": 30377 }, { "epoch": 0.03, "learning_rate": 4.988696226635254e-05, "loss": 0.8983, "step": 30378 }, { "epoch": 0.03, "learning_rate": 4.988695480520368e-05, "loss": 1.323, "step": 30379 }, { "epoch": 0.03, "learning_rate": 4.988694734380914e-05, "loss": 1.1321, "step": 30380 }, { "epoch": 0.03, "learning_rate": 4.9886939882168936e-05, "loss": 1.1689, "step": 30381 }, { "epoch": 0.03, "learning_rate": 4.988693242028306e-05, "loss": 0.5773, "step": 30382 }, { "epoch": 0.03, "learning_rate": 4.9886924958151504e-05, "loss": 0.629, "step": 30383 }, { "epoch": 0.03, "learning_rate": 4.988691749577427e-05, "loss": 0.8304, "step": 30384 }, { "epoch": 0.03, "learning_rate": 4.988691003315137e-05, "loss": 1.2591, "step": 30385 }, { "epoch": 0.03, "learning_rate": 4.98869025702828e-05, "loss": 0.7578, "step": 30386 }, { "epoch": 0.03, "learning_rate": 4.988689510716855e-05, "loss": 1.0838, "step": 30387 }, { "epoch": 0.03, "learning_rate": 4.9886887643808624e-05, "loss": 1.2589, "step": 30388 }, { "epoch": 0.03, "learning_rate": 4.988688018020303e-05, "loss": 1.2296, "step": 30389 }, { "epoch": 0.03, "learning_rate": 4.988687271635177e-05, "loss": 1.0462, "step": 30390 }, { "epoch": 0.03, "learning_rate": 4.988686525225483e-05, "loss": 1.3683, "step": 30391 }, { "epoch": 0.03, "learning_rate": 4.988685778791222e-05, "loss": 0.8464, "step": 30392 }, { "epoch": 0.03, "learning_rate": 4.988685032332393e-05, "loss": 1.131, "step": 30393 }, { "epoch": 0.03, "learning_rate": 4.988684285848998e-05, "loss": 1.151, "step": 30394 }, { "epoch": 0.03, "learning_rate": 4.988683539341035e-05, "loss": 0.9147, "step": 30395 }, { "epoch": 0.03, "learning_rate": 4.988682792808505e-05, "loss": 0.7759, "step": 30396 }, { "epoch": 0.03, "learning_rate": 4.988682046251407e-05, "loss": 0.798, "step": 30397 }, { "epoch": 0.03, "learning_rate": 4.988681299669743e-05, "loss": 1.1345, "step": 30398 }, { "epoch": 0.03, "learning_rate": 4.9886805530635105e-05, "loss": 1.1985, "step": 30399 }, { "epoch": 0.03, "learning_rate": 4.988679806432712e-05, "loss": 0.8181, "step": 30400 }, { "epoch": 0.03, "learning_rate": 4.9886790597773456e-05, "loss": 0.7605, "step": 30401 }, { "epoch": 0.03, "learning_rate": 4.988678313097412e-05, "loss": 0.8699, "step": 30402 }, { "epoch": 0.03, "learning_rate": 4.988677566392912e-05, "loss": 0.8725, "step": 30403 }, { "epoch": 0.03, "learning_rate": 4.988676819663843e-05, "loss": 0.85, "step": 30404 }, { "epoch": 0.03, "learning_rate": 4.988676072910209e-05, "loss": 0.9312, "step": 30405 }, { "epoch": 0.03, "learning_rate": 4.988675326132007e-05, "loss": 0.7508, "step": 30406 }, { "epoch": 0.03, "learning_rate": 4.988674579329238e-05, "loss": 1.189, "step": 30407 }, { "epoch": 0.03, "learning_rate": 4.988673832501901e-05, "loss": 1.0089, "step": 30408 }, { "epoch": 0.03, "learning_rate": 4.988673085649998e-05, "loss": 1.2847, "step": 30409 }, { "epoch": 0.03, "learning_rate": 4.988672338773527e-05, "loss": 0.8953, "step": 30410 }, { "epoch": 0.03, "learning_rate": 4.988671591872489e-05, "loss": 0.8516, "step": 30411 }, { "epoch": 0.03, "learning_rate": 4.9886708449468845e-05, "loss": 1.048, "step": 30412 }, { "epoch": 0.03, "learning_rate": 4.988670097996712e-05, "loss": 0.6563, "step": 30413 }, { "epoch": 0.03, "learning_rate": 4.988669351021973e-05, "loss": 0.9653, "step": 30414 }, { "epoch": 0.03, "learning_rate": 4.988668604022667e-05, "loss": 0.9193, "step": 30415 }, { "epoch": 0.03, "learning_rate": 4.988667856998793e-05, "loss": 0.9939, "step": 30416 }, { "epoch": 0.03, "learning_rate": 4.9886671099503534e-05, "loss": 1.0528, "step": 30417 }, { "epoch": 0.03, "learning_rate": 4.988666362877346e-05, "loss": 1.3531, "step": 30418 }, { "epoch": 0.03, "learning_rate": 4.988665615779771e-05, "loss": 0.8761, "step": 30419 }, { "epoch": 0.03, "learning_rate": 4.98866486865763e-05, "loss": 0.9057, "step": 30420 }, { "epoch": 0.03, "learning_rate": 4.988664121510921e-05, "loss": 0.9434, "step": 30421 }, { "epoch": 0.03, "learning_rate": 4.988663374339645e-05, "loss": 0.7422, "step": 30422 }, { "epoch": 0.03, "learning_rate": 4.9886626271438026e-05, "loss": 1.0685, "step": 30423 }, { "epoch": 0.03, "learning_rate": 4.988661879923393e-05, "loss": 1.315, "step": 30424 }, { "epoch": 0.03, "learning_rate": 4.988661132678416e-05, "loss": 0.9088, "step": 30425 }, { "epoch": 0.03, "learning_rate": 4.988660385408872e-05, "loss": 1.104, "step": 30426 }, { "epoch": 0.03, "learning_rate": 4.9886596381147615e-05, "loss": 1.1951, "step": 30427 }, { "epoch": 0.03, "learning_rate": 4.988658890796084e-05, "loss": 0.816, "step": 30428 }, { "epoch": 0.03, "learning_rate": 4.988658143452839e-05, "loss": 0.9876, "step": 30429 }, { "epoch": 0.03, "learning_rate": 4.9886573960850266e-05, "loss": 0.9168, "step": 30430 }, { "epoch": 0.03, "learning_rate": 4.988656648692648e-05, "loss": 0.9905, "step": 30431 }, { "epoch": 0.03, "learning_rate": 4.9886559012757027e-05, "loss": 1.182, "step": 30432 }, { "epoch": 0.03, "learning_rate": 4.98865515383419e-05, "loss": 0.9935, "step": 30433 }, { "epoch": 0.03, "learning_rate": 4.988654406368111e-05, "loss": 0.8657, "step": 30434 }, { "epoch": 0.03, "learning_rate": 4.988653658877464e-05, "loss": 0.7408, "step": 30435 }, { "epoch": 0.03, "learning_rate": 4.988652911362251e-05, "loss": 0.9621, "step": 30436 }, { "epoch": 0.03, "learning_rate": 4.98865216382247e-05, "loss": 0.9885, "step": 30437 }, { "epoch": 0.03, "learning_rate": 4.988651416258122e-05, "loss": 1.0256, "step": 30438 }, { "epoch": 0.03, "learning_rate": 4.988650668669208e-05, "loss": 0.9647, "step": 30439 }, { "epoch": 0.03, "learning_rate": 4.988649921055727e-05, "loss": 0.8234, "step": 30440 }, { "epoch": 0.03, "learning_rate": 4.988649173417679e-05, "loss": 0.5206, "step": 30441 }, { "epoch": 0.03, "learning_rate": 4.9886484257550634e-05, "loss": 0.766, "step": 30442 }, { "epoch": 0.03, "learning_rate": 4.988647678067882e-05, "loss": 0.4214, "step": 30443 }, { "epoch": 0.03, "learning_rate": 4.9886469303561334e-05, "loss": 0.8505, "step": 30444 }, { "epoch": 0.03, "learning_rate": 4.988646182619817e-05, "loss": 1.0268, "step": 30445 }, { "epoch": 0.03, "learning_rate": 4.9886454348589354e-05, "loss": 0.8361, "step": 30446 }, { "epoch": 0.03, "learning_rate": 4.988644687073486e-05, "loss": 1.1138, "step": 30447 }, { "epoch": 0.03, "learning_rate": 4.98864393926347e-05, "loss": 0.8205, "step": 30448 }, { "epoch": 0.03, "learning_rate": 4.988643191428887e-05, "loss": 0.9563, "step": 30449 }, { "epoch": 0.03, "learning_rate": 4.9886424435697366e-05, "loss": 0.574, "step": 30450 }, { "epoch": 0.03, "learning_rate": 4.9886416956860196e-05, "loss": 0.9863, "step": 30451 }, { "epoch": 0.03, "learning_rate": 4.9886409477777366e-05, "loss": 0.7047, "step": 30452 }, { "epoch": 0.03, "learning_rate": 4.988640199844886e-05, "loss": 0.7565, "step": 30453 }, { "epoch": 0.03, "learning_rate": 4.988639451887469e-05, "loss": 0.9902, "step": 30454 }, { "epoch": 0.03, "learning_rate": 4.988638703905485e-05, "loss": 1.0765, "step": 30455 }, { "epoch": 0.03, "learning_rate": 4.988637955898934e-05, "loss": 1.0259, "step": 30456 }, { "epoch": 0.03, "learning_rate": 4.988637207867817e-05, "loss": 0.8864, "step": 30457 }, { "epoch": 0.03, "learning_rate": 4.9886364598121323e-05, "loss": 1.0341, "step": 30458 }, { "epoch": 0.03, "learning_rate": 4.988635711731881e-05, "loss": 0.9685, "step": 30459 }, { "epoch": 0.03, "learning_rate": 4.9886349636270636e-05, "loss": 0.9665, "step": 30460 }, { "epoch": 0.03, "learning_rate": 4.9886342154976785e-05, "loss": 1.1273, "step": 30461 }, { "epoch": 0.03, "learning_rate": 4.9886334673437275e-05, "loss": 1.0754, "step": 30462 }, { "epoch": 0.03, "learning_rate": 4.988632719165209e-05, "loss": 0.7055, "step": 30463 }, { "epoch": 0.03, "learning_rate": 4.988631970962124e-05, "loss": 0.9298, "step": 30464 }, { "epoch": 0.03, "learning_rate": 4.9886312227344724e-05, "loss": 1.1543, "step": 30465 }, { "epoch": 0.03, "learning_rate": 4.988630474482254e-05, "loss": 0.8627, "step": 30466 }, { "epoch": 0.03, "learning_rate": 4.9886297262054696e-05, "loss": 0.8864, "step": 30467 }, { "epoch": 0.03, "learning_rate": 4.988628977904117e-05, "loss": 0.961, "step": 30468 }, { "epoch": 0.03, "learning_rate": 4.9886282295781995e-05, "loss": 1.0342, "step": 30469 }, { "epoch": 0.03, "learning_rate": 4.9886274812277145e-05, "loss": 1.2027, "step": 30470 }, { "epoch": 0.03, "learning_rate": 4.988626732852663e-05, "loss": 0.8824, "step": 30471 }, { "epoch": 0.03, "learning_rate": 4.988625984453044e-05, "loss": 0.9542, "step": 30472 }, { "epoch": 0.03, "learning_rate": 4.988625236028859e-05, "loss": 1.0521, "step": 30473 }, { "epoch": 0.03, "learning_rate": 4.988624487580107e-05, "loss": 0.9454, "step": 30474 }, { "epoch": 0.03, "learning_rate": 4.9886237391067886e-05, "loss": 0.8952, "step": 30475 }, { "epoch": 0.03, "learning_rate": 4.9886229906089036e-05, "loss": 0.8995, "step": 30476 }, { "epoch": 0.03, "learning_rate": 4.988622242086451e-05, "loss": 0.4351, "step": 30477 }, { "epoch": 0.03, "learning_rate": 4.9886214935394335e-05, "loss": 0.2577, "step": 30478 }, { "epoch": 0.03, "learning_rate": 4.9886207449678485e-05, "loss": 0.6427, "step": 30479 }, { "epoch": 0.03, "learning_rate": 4.9886199963716975e-05, "loss": 1.2303, "step": 30480 }, { "epoch": 0.03, "learning_rate": 4.9886192477509784e-05, "loss": 0.9121, "step": 30481 }, { "epoch": 0.03, "learning_rate": 4.988618499105694e-05, "loss": 1.165, "step": 30482 }, { "epoch": 0.03, "learning_rate": 4.9886177504358424e-05, "loss": 1.0825, "step": 30483 }, { "epoch": 0.03, "learning_rate": 4.988617001741425e-05, "loss": 0.9241, "step": 30484 }, { "epoch": 0.03, "learning_rate": 4.98861625302244e-05, "loss": 1.1902, "step": 30485 }, { "epoch": 0.03, "learning_rate": 4.988615504278889e-05, "loss": 0.9957, "step": 30486 }, { "epoch": 0.03, "learning_rate": 4.988614755510771e-05, "loss": 0.9244, "step": 30487 }, { "epoch": 0.03, "learning_rate": 4.988614006718087e-05, "loss": 1.1216, "step": 30488 }, { "epoch": 0.03, "learning_rate": 4.988613257900836e-05, "loss": 1.0553, "step": 30489 }, { "epoch": 0.03, "learning_rate": 4.988612509059019e-05, "loss": 1.2569, "step": 30490 }, { "epoch": 0.03, "learning_rate": 4.988611760192635e-05, "loss": 1.0344, "step": 30491 }, { "epoch": 0.03, "learning_rate": 4.9886110113016846e-05, "loss": 0.7129, "step": 30492 }, { "epoch": 0.03, "learning_rate": 4.9886102623861675e-05, "loss": 1.0227, "step": 30493 }, { "epoch": 0.03, "learning_rate": 4.988609513446084e-05, "loss": 1.1031, "step": 30494 }, { "epoch": 0.03, "learning_rate": 4.988608764481434e-05, "loss": 0.9299, "step": 30495 }, { "epoch": 0.03, "learning_rate": 4.988608015492218e-05, "loss": 0.8762, "step": 30496 }, { "epoch": 0.03, "learning_rate": 4.9886072664784356e-05, "loss": 1.1619, "step": 30497 }, { "epoch": 0.03, "learning_rate": 4.988606517440086e-05, "loss": 0.8972, "step": 30498 }, { "epoch": 0.03, "learning_rate": 4.98860576837717e-05, "loss": 0.9351, "step": 30499 }, { "epoch": 0.03, "learning_rate": 4.988605019289688e-05, "loss": 0.9589, "step": 30500 }, { "epoch": 0.03, "eval_loss": 1.062410831451416, "eval_runtime": 127.2932, "eval_samples_per_second": 10.88, "eval_steps_per_second": 5.444, "step": 30500 }, { "epoch": 0.03, "learning_rate": 4.988604270177639e-05, "loss": 0.6742, "step": 30501 }, { "epoch": 0.03, "learning_rate": 4.9886035210410234e-05, "loss": 0.9181, "step": 30502 }, { "epoch": 0.03, "learning_rate": 4.9886027718798424e-05, "loss": 0.9277, "step": 30503 }, { "epoch": 0.03, "learning_rate": 4.988602022694094e-05, "loss": 1.1456, "step": 30504 }, { "epoch": 0.03, "learning_rate": 4.98860127348378e-05, "loss": 0.303, "step": 30505 }, { "epoch": 0.03, "learning_rate": 4.988600524248898e-05, "loss": 1.084, "step": 30506 }, { "epoch": 0.03, "learning_rate": 4.9885997749894506e-05, "loss": 1.4629, "step": 30507 }, { "epoch": 0.03, "learning_rate": 4.988599025705437e-05, "loss": 0.589, "step": 30508 }, { "epoch": 0.03, "learning_rate": 4.988598276396857e-05, "loss": 0.7545, "step": 30509 }, { "epoch": 0.03, "learning_rate": 4.9885975270637104e-05, "loss": 0.9913, "step": 30510 }, { "epoch": 0.03, "learning_rate": 4.9885967777059975e-05, "loss": 0.9444, "step": 30511 }, { "epoch": 0.03, "learning_rate": 4.988596028323718e-05, "loss": 0.8748, "step": 30512 }, { "epoch": 0.03, "learning_rate": 4.988595278916872e-05, "loss": 0.9296, "step": 30513 }, { "epoch": 0.03, "learning_rate": 4.9885945294854594e-05, "loss": 1.1932, "step": 30514 }, { "epoch": 0.03, "learning_rate": 4.988593780029481e-05, "loss": 0.9905, "step": 30515 }, { "epoch": 0.03, "learning_rate": 4.988593030548937e-05, "loss": 0.4419, "step": 30516 }, { "epoch": 0.03, "learning_rate": 4.9885922810438254e-05, "loss": 0.7889, "step": 30517 }, { "epoch": 0.03, "learning_rate": 4.9885915315141485e-05, "loss": 1.2019, "step": 30518 }, { "epoch": 0.03, "learning_rate": 4.988590781959904e-05, "loss": 1.063, "step": 30519 }, { "epoch": 0.03, "learning_rate": 4.988590032381094e-05, "loss": 0.874, "step": 30520 }, { "epoch": 0.03, "learning_rate": 4.988589282777718e-05, "loss": 1.0266, "step": 30521 }, { "epoch": 0.03, "learning_rate": 4.988588533149775e-05, "loss": 1.2642, "step": 30522 }, { "epoch": 0.03, "learning_rate": 4.988587783497266e-05, "loss": 1.3523, "step": 30523 }, { "epoch": 0.03, "learning_rate": 4.9885870338201907e-05, "loss": 1.1599, "step": 30524 }, { "epoch": 0.03, "learning_rate": 4.988586284118549e-05, "loss": 0.9617, "step": 30525 }, { "epoch": 0.03, "learning_rate": 4.9885855343923417e-05, "loss": 1.3953, "step": 30526 }, { "epoch": 0.03, "learning_rate": 4.988584784641567e-05, "loss": 1.348, "step": 30527 }, { "epoch": 0.03, "learning_rate": 4.988584034866227e-05, "loss": 1.0633, "step": 30528 }, { "epoch": 0.03, "learning_rate": 4.98858328506632e-05, "loss": 0.7868, "step": 30529 }, { "epoch": 0.03, "learning_rate": 4.988582535241847e-05, "loss": 1.1971, "step": 30530 }, { "epoch": 0.03, "learning_rate": 4.988581785392808e-05, "loss": 0.6082, "step": 30531 }, { "epoch": 0.03, "learning_rate": 4.988581035519203e-05, "loss": 0.9953, "step": 30532 }, { "epoch": 0.03, "learning_rate": 4.988580285621031e-05, "loss": 1.0583, "step": 30533 }, { "epoch": 0.03, "learning_rate": 4.988579535698293e-05, "loss": 1.1714, "step": 30534 }, { "epoch": 0.03, "learning_rate": 4.988578785750989e-05, "loss": 1.1749, "step": 30535 }, { "epoch": 0.03, "learning_rate": 4.988578035779119e-05, "loss": 0.9516, "step": 30536 }, { "epoch": 0.03, "learning_rate": 4.988577285782682e-05, "loss": 1.0658, "step": 30537 }, { "epoch": 0.03, "learning_rate": 4.9885765357616796e-05, "loss": 1.2269, "step": 30538 }, { "epoch": 0.03, "learning_rate": 4.988575785716111e-05, "loss": 1.1412, "step": 30539 }, { "epoch": 0.03, "learning_rate": 4.988575035645976e-05, "loss": 1.0141, "step": 30540 }, { "epoch": 0.03, "learning_rate": 4.9885742855512754e-05, "loss": 1.0834, "step": 30541 }, { "epoch": 0.03, "learning_rate": 4.9885735354320074e-05, "loss": 0.6472, "step": 30542 }, { "epoch": 0.03, "learning_rate": 4.988572785288174e-05, "loss": 1.139, "step": 30543 }, { "epoch": 0.03, "learning_rate": 4.9885720351197746e-05, "loss": 1.0964, "step": 30544 }, { "epoch": 0.03, "learning_rate": 4.9885712849268086e-05, "loss": 1.0609, "step": 30545 }, { "epoch": 0.03, "learning_rate": 4.988570534709277e-05, "loss": 1.0681, "step": 30546 }, { "epoch": 0.03, "learning_rate": 4.988569784467179e-05, "loss": 1.0075, "step": 30547 }, { "epoch": 0.03, "learning_rate": 4.988569034200515e-05, "loss": 0.8369, "step": 30548 }, { "epoch": 0.03, "learning_rate": 4.9885682839092854e-05, "loss": 0.9221, "step": 30549 }, { "epoch": 0.03, "learning_rate": 4.988567533593489e-05, "loss": 1.0704, "step": 30550 }, { "epoch": 0.03, "learning_rate": 4.988566783253126e-05, "loss": 0.7737, "step": 30551 }, { "epoch": 0.03, "learning_rate": 4.9885660328881975e-05, "loss": 1.1384, "step": 30552 }, { "epoch": 0.03, "learning_rate": 4.9885652824987036e-05, "loss": 1.0507, "step": 30553 }, { "epoch": 0.03, "learning_rate": 4.988564532084643e-05, "loss": 0.9285, "step": 30554 }, { "epoch": 0.03, "learning_rate": 4.9885637816460164e-05, "loss": 0.833, "step": 30555 }, { "epoch": 0.03, "learning_rate": 4.988563031182824e-05, "loss": 0.8649, "step": 30556 }, { "epoch": 0.03, "learning_rate": 4.9885622806950646e-05, "loss": 1.0348, "step": 30557 }, { "epoch": 0.03, "learning_rate": 4.98856153018274e-05, "loss": 1.1943, "step": 30558 }, { "epoch": 0.03, "learning_rate": 4.988560779645849e-05, "loss": 0.8494, "step": 30559 }, { "epoch": 0.03, "learning_rate": 4.988560029084392e-05, "loss": 1.0628, "step": 30560 }, { "epoch": 0.03, "learning_rate": 4.988559278498369e-05, "loss": 0.7865, "step": 30561 }, { "epoch": 0.03, "learning_rate": 4.98855852788778e-05, "loss": 1.2546, "step": 30562 }, { "epoch": 0.03, "learning_rate": 4.9885577772526255e-05, "loss": 1.2787, "step": 30563 }, { "epoch": 0.03, "learning_rate": 4.988557026592905e-05, "loss": 1.111, "step": 30564 }, { "epoch": 0.03, "learning_rate": 4.988556275908618e-05, "loss": 1.0025, "step": 30565 }, { "epoch": 0.03, "learning_rate": 4.9885555251997654e-05, "loss": 1.1355, "step": 30566 }, { "epoch": 0.03, "learning_rate": 4.988554774466346e-05, "loss": 1.0609, "step": 30567 }, { "epoch": 0.03, "learning_rate": 4.988554023708362e-05, "loss": 0.928, "step": 30568 }, { "epoch": 0.03, "learning_rate": 4.988553272925811e-05, "loss": 0.7891, "step": 30569 }, { "epoch": 0.03, "learning_rate": 4.9885525221186944e-05, "loss": 0.6626, "step": 30570 }, { "epoch": 0.03, "learning_rate": 4.988551771287011e-05, "loss": 0.5382, "step": 30571 }, { "epoch": 0.03, "learning_rate": 4.988551020430763e-05, "loss": 0.7892, "step": 30572 }, { "epoch": 0.03, "learning_rate": 4.9885502695499486e-05, "loss": 0.7419, "step": 30573 }, { "epoch": 0.03, "learning_rate": 4.988549518644568e-05, "loss": 0.8191, "step": 30574 }, { "epoch": 0.03, "learning_rate": 4.988548767714621e-05, "loss": 0.7508, "step": 30575 }, { "epoch": 0.03, "learning_rate": 4.9885480167601095e-05, "loss": 0.8749, "step": 30576 }, { "epoch": 0.03, "learning_rate": 4.988547265781032e-05, "loss": 1.1206, "step": 30577 }, { "epoch": 0.03, "learning_rate": 4.9885465147773876e-05, "loss": 0.9604, "step": 30578 }, { "epoch": 0.03, "learning_rate": 4.988545763749177e-05, "loss": 1.2183, "step": 30579 }, { "epoch": 0.03, "learning_rate": 4.988545012696402e-05, "loss": 1.2267, "step": 30580 }, { "epoch": 0.03, "learning_rate": 4.98854426161906e-05, "loss": 0.9436, "step": 30581 }, { "epoch": 0.03, "learning_rate": 4.988543510517153e-05, "loss": 0.7906, "step": 30582 }, { "epoch": 0.03, "learning_rate": 4.9885427593906796e-05, "loss": 1.1821, "step": 30583 }, { "epoch": 0.03, "learning_rate": 4.988542008239641e-05, "loss": 1.0951, "step": 30584 }, { "epoch": 0.03, "learning_rate": 4.988541257064036e-05, "loss": 1.1501, "step": 30585 }, { "epoch": 0.03, "learning_rate": 4.988540505863864e-05, "loss": 1.2809, "step": 30586 }, { "epoch": 0.03, "learning_rate": 4.9885397546391275e-05, "loss": 1.3834, "step": 30587 }, { "epoch": 0.03, "learning_rate": 4.9885390033898246e-05, "loss": 1.0349, "step": 30588 }, { "epoch": 0.03, "learning_rate": 4.988538252115957e-05, "loss": 0.9283, "step": 30589 }, { "epoch": 0.03, "learning_rate": 4.988537500817523e-05, "loss": 0.8912, "step": 30590 }, { "epoch": 0.03, "learning_rate": 4.988536749494523e-05, "loss": 0.5636, "step": 30591 }, { "epoch": 0.03, "learning_rate": 4.988535998146957e-05, "loss": 1.1915, "step": 30592 }, { "epoch": 0.03, "learning_rate": 4.988535246774826e-05, "loss": 1.1891, "step": 30593 }, { "epoch": 0.03, "learning_rate": 4.9885344953781286e-05, "loss": 0.884, "step": 30594 }, { "epoch": 0.03, "learning_rate": 4.988533743956866e-05, "loss": 1.0431, "step": 30595 }, { "epoch": 0.03, "learning_rate": 4.988532992511037e-05, "loss": 1.1533, "step": 30596 }, { "epoch": 0.03, "learning_rate": 4.988532241040642e-05, "loss": 1.4342, "step": 30597 }, { "epoch": 0.03, "learning_rate": 4.9885314895456825e-05, "loss": 1.4455, "step": 30598 }, { "epoch": 0.03, "learning_rate": 4.988530738026156e-05, "loss": 1.1039, "step": 30599 }, { "epoch": 0.03, "learning_rate": 4.9885299864820645e-05, "loss": 0.6407, "step": 30600 }, { "epoch": 0.03, "learning_rate": 4.988529234913407e-05, "loss": 0.6461, "step": 30601 }, { "epoch": 0.03, "learning_rate": 4.988528483320184e-05, "loss": 0.874, "step": 30602 }, { "epoch": 0.03, "learning_rate": 4.9885277317023946e-05, "loss": 1.035, "step": 30603 }, { "epoch": 0.03, "learning_rate": 4.988526980060041e-05, "loss": 1.0288, "step": 30604 }, { "epoch": 0.03, "learning_rate": 4.98852622839312e-05, "loss": 0.8685, "step": 30605 }, { "epoch": 0.03, "learning_rate": 4.988525476701634e-05, "loss": 1.1442, "step": 30606 }, { "epoch": 0.03, "learning_rate": 4.9885247249855824e-05, "loss": 0.8064, "step": 30607 }, { "epoch": 0.03, "learning_rate": 4.988523973244966e-05, "loss": 1.2525, "step": 30608 }, { "epoch": 0.03, "learning_rate": 4.988523221479783e-05, "loss": 1.1239, "step": 30609 }, { "epoch": 0.03, "learning_rate": 4.988522469690035e-05, "loss": 0.8973, "step": 30610 }, { "epoch": 0.03, "learning_rate": 4.9885217178757204e-05, "loss": 1.0113, "step": 30611 }, { "epoch": 0.03, "learning_rate": 4.9885209660368406e-05, "loss": 1.0434, "step": 30612 }, { "epoch": 0.03, "learning_rate": 4.988520214173394e-05, "loss": 0.8621, "step": 30613 }, { "epoch": 0.03, "learning_rate": 4.988519462285384e-05, "loss": 1.0373, "step": 30614 }, { "epoch": 0.03, "learning_rate": 4.988518710372807e-05, "loss": 1.1877, "step": 30615 }, { "epoch": 0.03, "learning_rate": 4.9885179584356643e-05, "loss": 1.2804, "step": 30616 }, { "epoch": 0.03, "learning_rate": 4.9885172064739566e-05, "loss": 1.2374, "step": 30617 }, { "epoch": 0.03, "learning_rate": 4.988516454487683e-05, "loss": 1.5123, "step": 30618 }, { "epoch": 0.03, "learning_rate": 4.988515702476844e-05, "loss": 1.0186, "step": 30619 }, { "epoch": 0.03, "learning_rate": 4.98851495044144e-05, "loss": 1.0073, "step": 30620 }, { "epoch": 0.03, "learning_rate": 4.9885141983814694e-05, "loss": 1.1017, "step": 30621 }, { "epoch": 0.03, "learning_rate": 4.988513446296933e-05, "loss": 0.8871, "step": 30622 }, { "epoch": 0.03, "learning_rate": 4.9885126941878315e-05, "loss": 0.9008, "step": 30623 }, { "epoch": 0.03, "learning_rate": 4.9885119420541646e-05, "loss": 0.9932, "step": 30624 }, { "epoch": 0.03, "learning_rate": 4.9885111898959324e-05, "loss": 1.1745, "step": 30625 }, { "epoch": 0.03, "learning_rate": 4.988510437713135e-05, "loss": 0.605, "step": 30626 }, { "epoch": 0.03, "learning_rate": 4.9885096855057714e-05, "loss": 0.8405, "step": 30627 }, { "epoch": 0.03, "learning_rate": 4.988508933273842e-05, "loss": 0.9884, "step": 30628 }, { "epoch": 0.03, "learning_rate": 4.988508181017347e-05, "loss": 0.6843, "step": 30629 }, { "epoch": 0.03, "learning_rate": 4.9885074287362876e-05, "loss": 0.8055, "step": 30630 }, { "epoch": 0.03, "learning_rate": 4.988506676430662e-05, "loss": 1.1032, "step": 30631 }, { "epoch": 0.03, "learning_rate": 4.988505924100471e-05, "loss": 0.8871, "step": 30632 }, { "epoch": 0.03, "learning_rate": 4.988505171745714e-05, "loss": 1.0735, "step": 30633 }, { "epoch": 0.03, "learning_rate": 4.988504419366392e-05, "loss": 0.9792, "step": 30634 }, { "epoch": 0.03, "learning_rate": 4.9885036669625054e-05, "loss": 0.8541, "step": 30635 }, { "epoch": 0.03, "learning_rate": 4.988502914534052e-05, "loss": 0.8583, "step": 30636 }, { "epoch": 0.03, "learning_rate": 4.9885021620810334e-05, "loss": 1.2462, "step": 30637 }, { "epoch": 0.03, "learning_rate": 4.98850140960345e-05, "loss": 1.3682, "step": 30638 }, { "epoch": 0.03, "learning_rate": 4.9885006571013e-05, "loss": 1.0604, "step": 30639 }, { "epoch": 0.03, "learning_rate": 4.988499904574586e-05, "loss": 0.7721, "step": 30640 }, { "epoch": 0.03, "learning_rate": 4.988499152023306e-05, "loss": 1.0694, "step": 30641 }, { "epoch": 0.03, "learning_rate": 4.98849839944746e-05, "loss": 0.9346, "step": 30642 }, { "epoch": 0.03, "learning_rate": 4.988497646847049e-05, "loss": 0.7156, "step": 30643 }, { "epoch": 0.03, "learning_rate": 4.9884968942220735e-05, "loss": 0.7742, "step": 30644 }, { "epoch": 0.03, "learning_rate": 4.988496141572531e-05, "loss": 0.8534, "step": 30645 }, { "epoch": 0.03, "learning_rate": 4.9884953888984245e-05, "loss": 1.0803, "step": 30646 }, { "epoch": 0.03, "learning_rate": 4.988494636199752e-05, "loss": 0.8803, "step": 30647 }, { "epoch": 0.03, "learning_rate": 4.988493883476514e-05, "loss": 0.7285, "step": 30648 }, { "epoch": 0.03, "learning_rate": 4.9884931307287106e-05, "loss": 0.9309, "step": 30649 }, { "epoch": 0.03, "learning_rate": 4.988492377956342e-05, "loss": 1.2221, "step": 30650 }, { "epoch": 0.03, "learning_rate": 4.9884916251594086e-05, "loss": 1.3587, "step": 30651 }, { "epoch": 0.03, "learning_rate": 4.988490872337909e-05, "loss": 1.4991, "step": 30652 }, { "epoch": 0.03, "learning_rate": 4.9884901194918446e-05, "loss": 1.0185, "step": 30653 }, { "epoch": 0.03, "learning_rate": 4.988489366621214e-05, "loss": 0.8844, "step": 30654 }, { "epoch": 0.03, "learning_rate": 4.988488613726019e-05, "loss": 0.8778, "step": 30655 }, { "epoch": 0.03, "learning_rate": 4.9884878608062585e-05, "loss": 1.3571, "step": 30656 }, { "epoch": 0.03, "learning_rate": 4.988487107861933e-05, "loss": 0.9396, "step": 30657 }, { "epoch": 0.03, "learning_rate": 4.9884863548930414e-05, "loss": 1.2973, "step": 30658 }, { "epoch": 0.03, "learning_rate": 4.988485601899585e-05, "loss": 1.6291, "step": 30659 }, { "epoch": 0.03, "learning_rate": 4.988484848881563e-05, "loss": 1.8122, "step": 30660 }, { "epoch": 0.03, "learning_rate": 4.988484095838977e-05, "loss": 1.6608, "step": 30661 }, { "epoch": 0.03, "learning_rate": 4.988483342771824e-05, "loss": 1.3499, "step": 30662 }, { "epoch": 0.03, "learning_rate": 4.988482589680107e-05, "loss": 1.5528, "step": 30663 }, { "epoch": 0.03, "learning_rate": 4.988481836563824e-05, "loss": 1.2257, "step": 30664 }, { "epoch": 0.03, "learning_rate": 4.9884810834229754e-05, "loss": 1.1688, "step": 30665 }, { "epoch": 0.03, "learning_rate": 4.988480330257562e-05, "loss": 0.9462, "step": 30666 }, { "epoch": 0.03, "learning_rate": 4.988479577067584e-05, "loss": 1.0802, "step": 30667 }, { "epoch": 0.03, "learning_rate": 4.988478823853041e-05, "loss": 1.1848, "step": 30668 }, { "epoch": 0.03, "learning_rate": 4.988478070613931e-05, "loss": 1.2646, "step": 30669 }, { "epoch": 0.03, "learning_rate": 4.9884773173502576e-05, "loss": 0.8181, "step": 30670 }, { "epoch": 0.03, "learning_rate": 4.988476564062018e-05, "loss": 1.0545, "step": 30671 }, { "epoch": 0.03, "learning_rate": 4.988475810749213e-05, "loss": 1.0717, "step": 30672 }, { "epoch": 0.03, "learning_rate": 4.988475057411844e-05, "loss": 0.9111, "step": 30673 }, { "epoch": 0.03, "learning_rate": 4.9884743040499096e-05, "loss": 1.5837, "step": 30674 }, { "epoch": 0.03, "learning_rate": 4.988473550663409e-05, "loss": 1.2105, "step": 30675 }, { "epoch": 0.03, "learning_rate": 4.988472797252344e-05, "loss": 0.9293, "step": 30676 }, { "epoch": 0.03, "learning_rate": 4.988472043816713e-05, "loss": 1.0433, "step": 30677 }, { "epoch": 0.03, "learning_rate": 4.988471290356518e-05, "loss": 0.9289, "step": 30678 }, { "epoch": 0.03, "learning_rate": 4.988470536871757e-05, "loss": 0.9399, "step": 30679 }, { "epoch": 0.03, "learning_rate": 4.988469783362432e-05, "loss": 0.8584, "step": 30680 }, { "epoch": 0.03, "learning_rate": 4.988469029828541e-05, "loss": 1.0891, "step": 30681 }, { "epoch": 0.03, "learning_rate": 4.988468276270085e-05, "loss": 0.9515, "step": 30682 }, { "epoch": 0.03, "learning_rate": 4.988467522687064e-05, "loss": 1.1136, "step": 30683 }, { "epoch": 0.03, "learning_rate": 4.9884667690794776e-05, "loss": 0.8503, "step": 30684 }, { "epoch": 0.03, "learning_rate": 4.988466015447326e-05, "loss": 0.1559, "step": 30685 }, { "epoch": 0.03, "learning_rate": 4.988465261790609e-05, "loss": 1.1242, "step": 30686 }, { "epoch": 0.03, "learning_rate": 4.988464508109328e-05, "loss": 1.0728, "step": 30687 }, { "epoch": 0.03, "learning_rate": 4.988463754403481e-05, "loss": 0.8833, "step": 30688 }, { "epoch": 0.03, "learning_rate": 4.9884630006730695e-05, "loss": 0.4325, "step": 30689 }, { "epoch": 0.03, "learning_rate": 4.988462246918093e-05, "loss": 0.6094, "step": 30690 }, { "epoch": 0.03, "learning_rate": 4.988461493138551e-05, "loss": 0.627, "step": 30691 }, { "epoch": 0.03, "learning_rate": 4.988460739334444e-05, "loss": 0.6011, "step": 30692 }, { "epoch": 0.03, "learning_rate": 4.988459985505772e-05, "loss": 0.531, "step": 30693 }, { "epoch": 0.03, "learning_rate": 4.9884592316525355e-05, "loss": 0.4345, "step": 30694 }, { "epoch": 0.03, "learning_rate": 4.9884584777747336e-05, "loss": 0.8148, "step": 30695 }, { "epoch": 0.03, "learning_rate": 4.988457723872367e-05, "loss": 1.0006, "step": 30696 }, { "epoch": 0.03, "learning_rate": 4.988456969945435e-05, "loss": 1.0489, "step": 30697 }, { "epoch": 0.03, "learning_rate": 4.9884562159939386e-05, "loss": 1.4584, "step": 30698 }, { "epoch": 0.03, "learning_rate": 4.988455462017876e-05, "loss": 1.4354, "step": 30699 }, { "epoch": 0.03, "learning_rate": 4.98845470801725e-05, "loss": 0.8983, "step": 30700 }, { "epoch": 0.03, "learning_rate": 4.988453953992057e-05, "loss": 1.4133, "step": 30701 }, { "epoch": 0.03, "learning_rate": 4.988453199942301e-05, "loss": 1.4515, "step": 30702 }, { "epoch": 0.03, "learning_rate": 4.988452445867979e-05, "loss": 1.0514, "step": 30703 }, { "epoch": 0.03, "learning_rate": 4.988451691769092e-05, "loss": 0.9635, "step": 30704 }, { "epoch": 0.03, "learning_rate": 4.9884509376456405e-05, "loss": 1.3388, "step": 30705 }, { "epoch": 0.03, "learning_rate": 4.9884501834976235e-05, "loss": 1.0488, "step": 30706 }, { "epoch": 0.03, "learning_rate": 4.988449429325042e-05, "loss": 0.9885, "step": 30707 }, { "epoch": 0.03, "learning_rate": 4.988448675127896e-05, "loss": 1.0295, "step": 30708 }, { "epoch": 0.03, "learning_rate": 4.988447920906185e-05, "loss": 1.2971, "step": 30709 }, { "epoch": 0.03, "learning_rate": 4.988447166659908e-05, "loss": 1.0716, "step": 30710 }, { "epoch": 0.03, "learning_rate": 4.988446412389067e-05, "loss": 0.8338, "step": 30711 }, { "epoch": 0.03, "learning_rate": 4.988445658093661e-05, "loss": 0.8478, "step": 30712 }, { "epoch": 0.03, "learning_rate": 4.9884449037736904e-05, "loss": 0.7238, "step": 30713 }, { "epoch": 0.03, "learning_rate": 4.9884441494291536e-05, "loss": 0.9706, "step": 30714 }, { "epoch": 0.03, "learning_rate": 4.9884433950600536e-05, "loss": 1.1998, "step": 30715 }, { "epoch": 0.03, "learning_rate": 4.9884426406663876e-05, "loss": 0.8999, "step": 30716 }, { "epoch": 0.03, "learning_rate": 4.988441886248158e-05, "loss": 1.2387, "step": 30717 }, { "epoch": 0.03, "learning_rate": 4.988441131805362e-05, "loss": 1.0482, "step": 30718 }, { "epoch": 0.03, "learning_rate": 4.988440377338002e-05, "loss": 0.9968, "step": 30719 }, { "epoch": 0.03, "learning_rate": 4.9884396228460766e-05, "loss": 1.0392, "step": 30720 }, { "epoch": 0.03, "learning_rate": 4.988438868329587e-05, "loss": 1.1175, "step": 30721 }, { "epoch": 0.03, "learning_rate": 4.988438113788533e-05, "loss": 1.0123, "step": 30722 }, { "epoch": 0.03, "learning_rate": 4.9884373592229125e-05, "loss": 1.0363, "step": 30723 }, { "epoch": 0.03, "learning_rate": 4.988436604632729e-05, "loss": 1.1278, "step": 30724 }, { "epoch": 0.03, "learning_rate": 4.98843585001798e-05, "loss": 0.542, "step": 30725 }, { "epoch": 0.03, "learning_rate": 4.9884350953786654e-05, "loss": 0.576, "step": 30726 }, { "epoch": 0.03, "learning_rate": 4.988434340714787e-05, "loss": 0.3973, "step": 30727 }, { "epoch": 0.03, "learning_rate": 4.9884335860263434e-05, "loss": 0.3287, "step": 30728 }, { "epoch": 0.03, "learning_rate": 4.988432831313335e-05, "loss": 1.2414, "step": 30729 }, { "epoch": 0.03, "learning_rate": 4.988432076575763e-05, "loss": 1.4549, "step": 30730 }, { "epoch": 0.03, "learning_rate": 4.988431321813625e-05, "loss": 1.2731, "step": 30731 }, { "epoch": 0.03, "learning_rate": 4.9884305670269226e-05, "loss": 1.0209, "step": 30732 }, { "epoch": 0.03, "learning_rate": 4.988429812215656e-05, "loss": 0.8647, "step": 30733 }, { "epoch": 0.03, "learning_rate": 4.988429057379824e-05, "loss": 0.6275, "step": 30734 }, { "epoch": 0.03, "learning_rate": 4.988428302519427e-05, "loss": 0.8931, "step": 30735 }, { "epoch": 0.03, "learning_rate": 4.9884275476344656e-05, "loss": 0.8454, "step": 30736 }, { "epoch": 0.03, "learning_rate": 4.9884267927249396e-05, "loss": 1.054, "step": 30737 }, { "epoch": 0.03, "learning_rate": 4.988426037790849e-05, "loss": 0.8033, "step": 30738 }, { "epoch": 0.03, "learning_rate": 4.988425282832193e-05, "loss": 0.7815, "step": 30739 }, { "epoch": 0.03, "learning_rate": 4.988424527848973e-05, "loss": 1.2287, "step": 30740 }, { "epoch": 0.03, "learning_rate": 4.9884237728411886e-05, "loss": 0.895, "step": 30741 }, { "epoch": 0.03, "learning_rate": 4.988423017808839e-05, "loss": 1.2197, "step": 30742 }, { "epoch": 0.03, "learning_rate": 4.988422262751925e-05, "loss": 1.1645, "step": 30743 }, { "epoch": 0.03, "learning_rate": 4.988421507670446e-05, "loss": 0.9192, "step": 30744 }, { "epoch": 0.03, "learning_rate": 4.988420752564403e-05, "loss": 1.1278, "step": 30745 }, { "epoch": 0.03, "learning_rate": 4.988419997433795e-05, "loss": 0.894, "step": 30746 }, { "epoch": 0.03, "learning_rate": 4.988419242278622e-05, "loss": 0.8949, "step": 30747 }, { "epoch": 0.03, "learning_rate": 4.988418487098885e-05, "loss": 1.2168, "step": 30748 }, { "epoch": 0.03, "learning_rate": 4.988417731894582e-05, "loss": 1.4005, "step": 30749 }, { "epoch": 0.03, "learning_rate": 4.9884169766657165e-05, "loss": 0.9872, "step": 30750 }, { "epoch": 0.03, "learning_rate": 4.988416221412285e-05, "loss": 1.2191, "step": 30751 }, { "epoch": 0.03, "learning_rate": 4.988415466134289e-05, "loss": 1.0413, "step": 30752 }, { "epoch": 0.03, "learning_rate": 4.9884147108317293e-05, "loss": 0.9987, "step": 30753 }, { "epoch": 0.03, "learning_rate": 4.988413955504604e-05, "loss": 1.1717, "step": 30754 }, { "epoch": 0.03, "learning_rate": 4.9884132001529145e-05, "loss": 1.0951, "step": 30755 }, { "epoch": 0.03, "learning_rate": 4.98841244477666e-05, "loss": 1.2327, "step": 30756 }, { "epoch": 0.03, "learning_rate": 4.988411689375841e-05, "loss": 0.7399, "step": 30757 }, { "epoch": 0.03, "learning_rate": 4.988410933950458e-05, "loss": 0.8827, "step": 30758 }, { "epoch": 0.03, "learning_rate": 4.988410178500511e-05, "loss": 1.0662, "step": 30759 }, { "epoch": 0.03, "learning_rate": 4.988409423025998e-05, "loss": 1.015, "step": 30760 }, { "epoch": 0.03, "learning_rate": 4.9884086675269215e-05, "loss": 0.9964, "step": 30761 }, { "epoch": 0.03, "learning_rate": 4.98840791200328e-05, "loss": 1.0911, "step": 30762 }, { "epoch": 0.03, "learning_rate": 4.988407156455075e-05, "loss": 0.9063, "step": 30763 }, { "epoch": 0.03, "learning_rate": 4.988406400882304e-05, "loss": 1.0577, "step": 30764 }, { "epoch": 0.03, "learning_rate": 4.988405645284969e-05, "loss": 0.5652, "step": 30765 }, { "epoch": 0.03, "learning_rate": 4.98840488966307e-05, "loss": 0.2647, "step": 30766 }, { "epoch": 0.03, "learning_rate": 4.9884041340166064e-05, "loss": 0.1963, "step": 30767 }, { "epoch": 0.03, "learning_rate": 4.988403378345578e-05, "loss": 0.1829, "step": 30768 }, { "epoch": 0.03, "learning_rate": 4.988402622649985e-05, "loss": 0.2239, "step": 30769 }, { "epoch": 0.03, "learning_rate": 4.9884018669298284e-05, "loss": 0.178, "step": 30770 }, { "epoch": 0.03, "learning_rate": 4.988401111185106e-05, "loss": 0.147, "step": 30771 }, { "epoch": 0.03, "learning_rate": 4.98840035541582e-05, "loss": 0.1481, "step": 30772 }, { "epoch": 0.03, "learning_rate": 4.988399599621969e-05, "loss": 0.1164, "step": 30773 }, { "epoch": 0.03, "learning_rate": 4.9883988438035546e-05, "loss": 0.1026, "step": 30774 }, { "epoch": 0.03, "learning_rate": 4.988398087960575e-05, "loss": 0.1238, "step": 30775 }, { "epoch": 0.03, "learning_rate": 4.988397332093031e-05, "loss": 0.4293, "step": 30776 }, { "epoch": 0.03, "learning_rate": 4.988396576200923e-05, "loss": 0.7289, "step": 30777 }, { "epoch": 0.03, "learning_rate": 4.988395820284251e-05, "loss": 0.5855, "step": 30778 }, { "epoch": 0.03, "learning_rate": 4.9883950643430135e-05, "loss": 0.5322, "step": 30779 }, { "epoch": 0.03, "learning_rate": 4.9883943083772124e-05, "loss": 0.462, "step": 30780 }, { "epoch": 0.03, "learning_rate": 4.9883935523868466e-05, "loss": 0.3822, "step": 30781 }, { "epoch": 0.03, "learning_rate": 4.9883927963719155e-05, "loss": 0.3764, "step": 30782 }, { "epoch": 0.03, "learning_rate": 4.988392040332421e-05, "loss": 0.2117, "step": 30783 }, { "epoch": 0.03, "learning_rate": 4.988391284268362e-05, "loss": 0.152, "step": 30784 }, { "epoch": 0.03, "learning_rate": 4.988390528179739e-05, "loss": 0.124, "step": 30785 }, { "epoch": 0.03, "learning_rate": 4.9883897720665515e-05, "loss": 0.0983, "step": 30786 }, { "epoch": 0.03, "learning_rate": 4.988389015928799e-05, "loss": 0.071, "step": 30787 }, { "epoch": 0.03, "learning_rate": 4.988388259766483e-05, "loss": 0.2442, "step": 30788 }, { "epoch": 0.03, "learning_rate": 4.988387503579603e-05, "loss": 0.8912, "step": 30789 }, { "epoch": 0.03, "learning_rate": 4.988386747368158e-05, "loss": 0.9765, "step": 30790 }, { "epoch": 0.03, "learning_rate": 4.988385991132149e-05, "loss": 0.9077, "step": 30791 }, { "epoch": 0.03, "learning_rate": 4.9883852348715754e-05, "loss": 0.939, "step": 30792 }, { "epoch": 0.03, "learning_rate": 4.9883844785864374e-05, "loss": 1.0323, "step": 30793 }, { "epoch": 0.03, "learning_rate": 4.9883837222767354e-05, "loss": 0.9328, "step": 30794 }, { "epoch": 0.03, "learning_rate": 4.988382965942469e-05, "loss": 1.1524, "step": 30795 }, { "epoch": 0.03, "learning_rate": 4.988382209583638e-05, "loss": 0.9159, "step": 30796 }, { "epoch": 0.03, "learning_rate": 4.988381453200244e-05, "loss": 0.9552, "step": 30797 }, { "epoch": 0.03, "learning_rate": 4.988380696792284e-05, "loss": 1.0919, "step": 30798 }, { "epoch": 0.03, "learning_rate": 4.988379940359761e-05, "loss": 1.0775, "step": 30799 }, { "epoch": 0.03, "learning_rate": 4.988379183902673e-05, "loss": 0.9526, "step": 30800 }, { "epoch": 0.03, "learning_rate": 4.9883784274210214e-05, "loss": 1.1698, "step": 30801 }, { "epoch": 0.03, "learning_rate": 4.988377670914806e-05, "loss": 0.9151, "step": 30802 }, { "epoch": 0.03, "learning_rate": 4.9883769143840255e-05, "loss": 0.777, "step": 30803 }, { "epoch": 0.03, "learning_rate": 4.98837615782868e-05, "loss": 0.6588, "step": 30804 }, { "epoch": 0.03, "learning_rate": 4.988375401248772e-05, "loss": 1.2157, "step": 30805 }, { "epoch": 0.03, "learning_rate": 4.988374644644299e-05, "loss": 1.0752, "step": 30806 }, { "epoch": 0.03, "learning_rate": 4.988373888015262e-05, "loss": 1.0536, "step": 30807 }, { "epoch": 0.03, "learning_rate": 4.988373131361661e-05, "loss": 1.1489, "step": 30808 }, { "epoch": 0.03, "learning_rate": 4.9883723746834954e-05, "loss": 1.0571, "step": 30809 }, { "epoch": 0.03, "learning_rate": 4.988371617980766e-05, "loss": 0.8992, "step": 30810 }, { "epoch": 0.03, "learning_rate": 4.988370861253472e-05, "loss": 1.0226, "step": 30811 }, { "epoch": 0.03, "learning_rate": 4.988370104501614e-05, "loss": 0.7734, "step": 30812 }, { "epoch": 0.03, "learning_rate": 4.9883693477251924e-05, "loss": 0.9067, "step": 30813 }, { "epoch": 0.03, "learning_rate": 4.988368590924206e-05, "loss": 0.8182, "step": 30814 }, { "epoch": 0.03, "learning_rate": 4.9883678340986555e-05, "loss": 0.3483, "step": 30815 }, { "epoch": 0.03, "learning_rate": 4.988367077248541e-05, "loss": 0.594, "step": 30816 }, { "epoch": 0.03, "learning_rate": 4.988366320373863e-05, "loss": 1.1313, "step": 30817 }, { "epoch": 0.03, "learning_rate": 4.9883655634746204e-05, "loss": 1.1897, "step": 30818 }, { "epoch": 0.03, "learning_rate": 4.9883648065508135e-05, "loss": 0.977, "step": 30819 }, { "epoch": 0.03, "learning_rate": 4.9883640496024426e-05, "loss": 1.0866, "step": 30820 }, { "epoch": 0.03, "learning_rate": 4.988363292629508e-05, "loss": 1.1767, "step": 30821 }, { "epoch": 0.03, "learning_rate": 4.988362535632008e-05, "loss": 1.0421, "step": 30822 }, { "epoch": 0.03, "learning_rate": 4.9883617786099455e-05, "loss": 0.9359, "step": 30823 }, { "epoch": 0.03, "learning_rate": 4.988361021563319e-05, "loss": 0.7084, "step": 30824 }, { "epoch": 0.03, "learning_rate": 4.9883602644921274e-05, "loss": 1.358, "step": 30825 }, { "epoch": 0.03, "learning_rate": 4.988359507396372e-05, "loss": 1.149, "step": 30826 }, { "epoch": 0.03, "learning_rate": 4.988358750276053e-05, "loss": 1.0756, "step": 30827 }, { "epoch": 0.03, "learning_rate": 4.9883579931311696e-05, "loss": 1.1956, "step": 30828 }, { "epoch": 0.03, "learning_rate": 4.9883572359617224e-05, "loss": 1.5688, "step": 30829 }, { "epoch": 0.03, "learning_rate": 4.988356478767711e-05, "loss": 0.8604, "step": 30830 }, { "epoch": 0.03, "learning_rate": 4.988355721549136e-05, "loss": 0.8707, "step": 30831 }, { "epoch": 0.03, "learning_rate": 4.9883549643059965e-05, "loss": 0.9937, "step": 30832 }, { "epoch": 0.03, "learning_rate": 4.9883542070382935e-05, "loss": 0.8325, "step": 30833 }, { "epoch": 0.03, "learning_rate": 4.988353449746026e-05, "loss": 1.004, "step": 30834 }, { "epoch": 0.03, "learning_rate": 4.988352692429194e-05, "loss": 1.0539, "step": 30835 }, { "epoch": 0.03, "learning_rate": 4.9883519350877994e-05, "loss": 0.98, "step": 30836 }, { "epoch": 0.03, "learning_rate": 4.9883511777218406e-05, "loss": 1.102, "step": 30837 }, { "epoch": 0.03, "learning_rate": 4.988350420331317e-05, "loss": 0.8767, "step": 30838 }, { "epoch": 0.03, "learning_rate": 4.9883496629162304e-05, "loss": 0.9658, "step": 30839 }, { "epoch": 0.03, "learning_rate": 4.988348905476579e-05, "loss": 0.989, "step": 30840 }, { "epoch": 0.03, "learning_rate": 4.988348148012364e-05, "loss": 0.9126, "step": 30841 }, { "epoch": 0.03, "learning_rate": 4.988347390523586e-05, "loss": 0.8621, "step": 30842 }, { "epoch": 0.03, "learning_rate": 4.9883466330102426e-05, "loss": 0.9279, "step": 30843 }, { "epoch": 0.03, "learning_rate": 4.9883458754723354e-05, "loss": 0.8736, "step": 30844 }, { "epoch": 0.03, "learning_rate": 4.988345117909865e-05, "loss": 0.9437, "step": 30845 }, { "epoch": 0.03, "learning_rate": 4.9883443603228305e-05, "loss": 1.0656, "step": 30846 }, { "epoch": 0.03, "learning_rate": 4.988343602711232e-05, "loss": 0.9205, "step": 30847 }, { "epoch": 0.03, "learning_rate": 4.9883428450750705e-05, "loss": 1.1868, "step": 30848 }, { "epoch": 0.03, "learning_rate": 4.9883420874143435e-05, "loss": 1.2871, "step": 30849 }, { "epoch": 0.03, "learning_rate": 4.988341329729054e-05, "loss": 1.2622, "step": 30850 }, { "epoch": 0.03, "learning_rate": 4.9883405720192e-05, "loss": 1.1967, "step": 30851 }, { "epoch": 0.03, "learning_rate": 4.988339814284782e-05, "loss": 1.1167, "step": 30852 }, { "epoch": 0.03, "learning_rate": 4.988339056525801e-05, "loss": 0.9046, "step": 30853 }, { "epoch": 0.03, "learning_rate": 4.9883382987422554e-05, "loss": 0.6961, "step": 30854 }, { "epoch": 0.03, "learning_rate": 4.988337540934146e-05, "loss": 1.0732, "step": 30855 }, { "epoch": 0.03, "learning_rate": 4.9883367831014735e-05, "loss": 1.0166, "step": 30856 }, { "epoch": 0.03, "learning_rate": 4.988336025244236e-05, "loss": 0.8903, "step": 30857 }, { "epoch": 0.03, "learning_rate": 4.988335267362436e-05, "loss": 0.9497, "step": 30858 }, { "epoch": 0.03, "learning_rate": 4.988334509456071e-05, "loss": 0.7152, "step": 30859 }, { "epoch": 0.03, "learning_rate": 4.988333751525143e-05, "loss": 1.1104, "step": 30860 }, { "epoch": 0.03, "learning_rate": 4.988332993569651e-05, "loss": 0.983, "step": 30861 }, { "epoch": 0.03, "learning_rate": 4.9883322355895954e-05, "loss": 0.9386, "step": 30862 }, { "epoch": 0.03, "learning_rate": 4.988331477584976e-05, "loss": 1.1622, "step": 30863 }, { "epoch": 0.03, "learning_rate": 4.988330719555793e-05, "loss": 0.8267, "step": 30864 }, { "epoch": 0.03, "learning_rate": 4.988329961502045e-05, "loss": 0.4029, "step": 30865 }, { "epoch": 0.03, "learning_rate": 4.9883292034237345e-05, "loss": 0.7313, "step": 30866 }, { "epoch": 0.03, "learning_rate": 4.9883284453208604e-05, "loss": 1.3283, "step": 30867 }, { "epoch": 0.03, "learning_rate": 4.9883276871934217e-05, "loss": 0.8308, "step": 30868 }, { "epoch": 0.03, "learning_rate": 4.9883269290414197e-05, "loss": 0.6896, "step": 30869 }, { "epoch": 0.03, "learning_rate": 4.988326170864854e-05, "loss": 0.9304, "step": 30870 }, { "epoch": 0.03, "learning_rate": 4.9883254126637245e-05, "loss": 0.9377, "step": 30871 }, { "epoch": 0.03, "learning_rate": 4.988324654438031e-05, "loss": 0.8895, "step": 30872 }, { "epoch": 0.03, "learning_rate": 4.988323896187774e-05, "loss": 0.9685, "step": 30873 }, { "epoch": 0.03, "learning_rate": 4.9883231379129544e-05, "loss": 1.1858, "step": 30874 }, { "epoch": 0.03, "learning_rate": 4.98832237961357e-05, "loss": 0.8138, "step": 30875 }, { "epoch": 0.03, "learning_rate": 4.988321621289622e-05, "loss": 0.8471, "step": 30876 }, { "epoch": 0.03, "learning_rate": 4.988320862941111e-05, "loss": 1.1149, "step": 30877 }, { "epoch": 0.03, "learning_rate": 4.988320104568035e-05, "loss": 1.0742, "step": 30878 }, { "epoch": 0.03, "learning_rate": 4.988319346170396e-05, "loss": 1.1216, "step": 30879 }, { "epoch": 0.03, "learning_rate": 4.988318587748194e-05, "loss": 0.7568, "step": 30880 }, { "epoch": 0.03, "learning_rate": 4.9883178293014274e-05, "loss": 1.0571, "step": 30881 }, { "epoch": 0.03, "learning_rate": 4.988317070830099e-05, "loss": 0.9979, "step": 30882 }, { "epoch": 0.03, "learning_rate": 4.988316312334205e-05, "loss": 1.0705, "step": 30883 }, { "epoch": 0.03, "learning_rate": 4.988315553813748e-05, "loss": 1.4355, "step": 30884 }, { "epoch": 0.03, "learning_rate": 4.988314795268727e-05, "loss": 1.0151, "step": 30885 }, { "epoch": 0.03, "learning_rate": 4.988314036699143e-05, "loss": 1.1235, "step": 30886 }, { "epoch": 0.03, "learning_rate": 4.988313278104996e-05, "loss": 1.0384, "step": 30887 }, { "epoch": 0.03, "learning_rate": 4.988312519486285e-05, "loss": 1.0041, "step": 30888 }, { "epoch": 0.03, "learning_rate": 4.98831176084301e-05, "loss": 1.1394, "step": 30889 }, { "epoch": 0.03, "learning_rate": 4.988311002175171e-05, "loss": 1.0498, "step": 30890 }, { "epoch": 0.03, "learning_rate": 4.9883102434827686e-05, "loss": 0.9868, "step": 30891 }, { "epoch": 0.03, "learning_rate": 4.988309484765803e-05, "loss": 1.248, "step": 30892 }, { "epoch": 0.03, "learning_rate": 4.988308726024274e-05, "loss": 1.1788, "step": 30893 }, { "epoch": 0.03, "learning_rate": 4.988307967258181e-05, "loss": 1.0357, "step": 30894 }, { "epoch": 0.03, "learning_rate": 4.9883072084675255e-05, "loss": 0.8957, "step": 30895 }, { "epoch": 0.03, "learning_rate": 4.9883064496523055e-05, "loss": 1.2091, "step": 30896 }, { "epoch": 0.03, "learning_rate": 4.988305690812522e-05, "loss": 1.1129, "step": 30897 }, { "epoch": 0.03, "learning_rate": 4.9883049319481757e-05, "loss": 0.8051, "step": 30898 }, { "epoch": 0.03, "learning_rate": 4.988304173059266e-05, "loss": 1.0976, "step": 30899 }, { "epoch": 0.03, "learning_rate": 4.988303414145792e-05, "loss": 0.9567, "step": 30900 }, { "epoch": 0.03, "learning_rate": 4.988302655207755e-05, "loss": 1.2036, "step": 30901 }, { "epoch": 0.03, "learning_rate": 4.9883018962451546e-05, "loss": 1.0303, "step": 30902 }, { "epoch": 0.03, "learning_rate": 4.9883011372579903e-05, "loss": 0.7258, "step": 30903 }, { "epoch": 0.03, "learning_rate": 4.988300378246262e-05, "loss": 0.6816, "step": 30904 }, { "epoch": 0.03, "learning_rate": 4.988299619209972e-05, "loss": 0.7713, "step": 30905 }, { "epoch": 0.03, "learning_rate": 4.988298860149117e-05, "loss": 1.0189, "step": 30906 }, { "epoch": 0.03, "learning_rate": 4.9882981010636983e-05, "loss": 0.8823, "step": 30907 }, { "epoch": 0.03, "learning_rate": 4.988297341953718e-05, "loss": 1.0655, "step": 30908 }, { "epoch": 0.03, "learning_rate": 4.988296582819173e-05, "loss": 1.0257, "step": 30909 }, { "epoch": 0.03, "learning_rate": 4.988295823660065e-05, "loss": 0.8649, "step": 30910 }, { "epoch": 0.03, "learning_rate": 4.988295064476393e-05, "loss": 0.9377, "step": 30911 }, { "epoch": 0.03, "learning_rate": 4.988294305268159e-05, "loss": 0.4873, "step": 30912 }, { "epoch": 0.03, "learning_rate": 4.9882935460353596e-05, "loss": 2.4373, "step": 30913 }, { "epoch": 0.03, "learning_rate": 4.988292786777998e-05, "loss": 2.1283, "step": 30914 }, { "epoch": 0.03, "learning_rate": 4.988292027496073e-05, "loss": 1.1733, "step": 30915 }, { "epoch": 0.03, "learning_rate": 4.988291268189585e-05, "loss": 1.0675, "step": 30916 }, { "epoch": 0.03, "learning_rate": 4.9882905088585325e-05, "loss": 1.4456, "step": 30917 }, { "epoch": 0.03, "learning_rate": 4.9882897495029177e-05, "loss": 1.0112, "step": 30918 }, { "epoch": 0.03, "learning_rate": 4.988288990122739e-05, "loss": 1.0282, "step": 30919 }, { "epoch": 0.03, "learning_rate": 4.988288230717997e-05, "loss": 1.4935, "step": 30920 }, { "epoch": 0.03, "learning_rate": 4.9882874712886915e-05, "loss": 1.2815, "step": 30921 }, { "epoch": 0.03, "learning_rate": 4.988286711834823e-05, "loss": 1.1714, "step": 30922 }, { "epoch": 0.03, "learning_rate": 4.988285952356391e-05, "loss": 1.1801, "step": 30923 }, { "epoch": 0.03, "learning_rate": 4.988285192853396e-05, "loss": 1.3389, "step": 30924 }, { "epoch": 0.03, "learning_rate": 4.9882844333258374e-05, "loss": 1.2132, "step": 30925 }, { "epoch": 0.03, "learning_rate": 4.988283673773716e-05, "loss": 1.2709, "step": 30926 }, { "epoch": 0.03, "learning_rate": 4.988282914197031e-05, "loss": 1.027, "step": 30927 }, { "epoch": 0.03, "learning_rate": 4.9882821545957824e-05, "loss": 1.1039, "step": 30928 }, { "epoch": 0.03, "learning_rate": 4.98828139496997e-05, "loss": 1.0166, "step": 30929 }, { "epoch": 0.03, "learning_rate": 4.9882806353195953e-05, "loss": 0.895, "step": 30930 }, { "epoch": 0.03, "learning_rate": 4.988279875644657e-05, "loss": 0.4044, "step": 30931 }, { "epoch": 0.03, "learning_rate": 4.988279115945156e-05, "loss": 0.5636, "step": 30932 }, { "epoch": 0.03, "learning_rate": 4.988278356221091e-05, "loss": 0.9179, "step": 30933 }, { "epoch": 0.03, "learning_rate": 4.988277596472464e-05, "loss": 1.3104, "step": 30934 }, { "epoch": 0.03, "learning_rate": 4.988276836699273e-05, "loss": 1.2106, "step": 30935 }, { "epoch": 0.03, "learning_rate": 4.988276076901518e-05, "loss": 1.2803, "step": 30936 }, { "epoch": 0.03, "learning_rate": 4.9882753170792004e-05, "loss": 0.6509, "step": 30937 }, { "epoch": 0.03, "learning_rate": 4.98827455723232e-05, "loss": 0.9695, "step": 30938 }, { "epoch": 0.03, "learning_rate": 4.988273797360876e-05, "loss": 1.0386, "step": 30939 }, { "epoch": 0.03, "learning_rate": 4.988273037464868e-05, "loss": 1.0993, "step": 30940 }, { "epoch": 0.03, "learning_rate": 4.988272277544298e-05, "loss": 1.1072, "step": 30941 }, { "epoch": 0.03, "learning_rate": 4.9882715175991644e-05, "loss": 1.0058, "step": 30942 }, { "epoch": 0.03, "learning_rate": 4.9882707576294677e-05, "loss": 1.0063, "step": 30943 }, { "epoch": 0.03, "learning_rate": 4.988269997635208e-05, "loss": 0.5996, "step": 30944 }, { "epoch": 0.03, "learning_rate": 4.9882692376163856e-05, "loss": 1.251, "step": 30945 }, { "epoch": 0.03, "learning_rate": 4.988268477572999e-05, "loss": 1.4792, "step": 30946 }, { "epoch": 0.03, "learning_rate": 4.98826771750505e-05, "loss": 0.9442, "step": 30947 }, { "epoch": 0.03, "learning_rate": 4.9882669574125374e-05, "loss": 0.9332, "step": 30948 }, { "epoch": 0.03, "learning_rate": 4.9882661972954616e-05, "loss": 0.6913, "step": 30949 }, { "epoch": 0.03, "learning_rate": 4.9882654371538226e-05, "loss": 0.9789, "step": 30950 }, { "epoch": 0.03, "learning_rate": 4.9882646769876216e-05, "loss": 0.8085, "step": 30951 }, { "epoch": 0.03, "learning_rate": 4.988263916796857e-05, "loss": 1.0667, "step": 30952 }, { "epoch": 0.03, "learning_rate": 4.9882631565815285e-05, "loss": 1.8214, "step": 30953 }, { "epoch": 0.03, "learning_rate": 4.988262396341638e-05, "loss": 1.1644, "step": 30954 }, { "epoch": 0.03, "learning_rate": 4.988261636077184e-05, "loss": 1.1427, "step": 30955 }, { "epoch": 0.03, "learning_rate": 4.988260875788166e-05, "loss": 1.3241, "step": 30956 }, { "epoch": 0.03, "learning_rate": 4.9882601154745864e-05, "loss": 0.9922, "step": 30957 }, { "epoch": 0.03, "learning_rate": 4.988259355136443e-05, "loss": 0.9675, "step": 30958 }, { "epoch": 0.03, "learning_rate": 4.988258594773737e-05, "loss": 0.9938, "step": 30959 }, { "epoch": 0.03, "learning_rate": 4.988257834386467e-05, "loss": 1.2241, "step": 30960 }, { "epoch": 0.03, "learning_rate": 4.9882570739746345e-05, "loss": 1.0492, "step": 30961 }, { "epoch": 0.03, "learning_rate": 4.9882563135382395e-05, "loss": 0.6426, "step": 30962 }, { "epoch": 0.03, "learning_rate": 4.9882555530772806e-05, "loss": 0.855, "step": 30963 }, { "epoch": 0.03, "learning_rate": 4.988254792591759e-05, "loss": 0.9148, "step": 30964 }, { "epoch": 0.03, "learning_rate": 4.988254032081675e-05, "loss": 0.972, "step": 30965 }, { "epoch": 0.03, "learning_rate": 4.9882532715470275e-05, "loss": 0.9347, "step": 30966 }, { "epoch": 0.03, "learning_rate": 4.988252510987817e-05, "loss": 1.3056, "step": 30967 }, { "epoch": 0.03, "learning_rate": 4.9882517504040435e-05, "loss": 1.072, "step": 30968 }, { "epoch": 0.03, "learning_rate": 4.988250989795707e-05, "loss": 1.0219, "step": 30969 }, { "epoch": 0.03, "learning_rate": 4.988250229162808e-05, "loss": 0.9766, "step": 30970 }, { "epoch": 0.03, "learning_rate": 4.988249468505345e-05, "loss": 1.0861, "step": 30971 }, { "epoch": 0.03, "learning_rate": 4.9882487078233196e-05, "loss": 1.0784, "step": 30972 }, { "epoch": 0.03, "learning_rate": 4.988247947116732e-05, "loss": 1.4828, "step": 30973 }, { "epoch": 0.03, "learning_rate": 4.9882471863855804e-05, "loss": 1.1959, "step": 30974 }, { "epoch": 0.03, "learning_rate": 4.988246425629867e-05, "loss": 1.0805, "step": 30975 }, { "epoch": 0.03, "learning_rate": 4.9882456648495894e-05, "loss": 0.6388, "step": 30976 }, { "epoch": 0.03, "learning_rate": 4.98824490404475e-05, "loss": 0.9993, "step": 30977 }, { "epoch": 0.03, "learning_rate": 4.988244143215347e-05, "loss": 0.8818, "step": 30978 }, { "epoch": 0.03, "learning_rate": 4.988243382361381e-05, "loss": 1.5963, "step": 30979 }, { "epoch": 0.03, "learning_rate": 4.988242621482853e-05, "loss": 1.1106, "step": 30980 }, { "epoch": 0.03, "learning_rate": 4.988241860579761e-05, "loss": 1.1211, "step": 30981 }, { "epoch": 0.03, "learning_rate": 4.9882410996521065e-05, "loss": 0.7224, "step": 30982 }, { "epoch": 0.03, "learning_rate": 4.9882403386998895e-05, "loss": 0.2853, "step": 30983 }, { "epoch": 0.03, "learning_rate": 4.988239577723109e-05, "loss": 0.1742, "step": 30984 }, { "epoch": 0.03, "learning_rate": 4.988238816721766e-05, "loss": 0.1263, "step": 30985 }, { "epoch": 0.03, "learning_rate": 4.988238055695861e-05, "loss": 0.036, "step": 30986 }, { "epoch": 0.03, "learning_rate": 4.988237294645391e-05, "loss": 0.0554, "step": 30987 }, { "epoch": 0.03, "learning_rate": 4.9882365335703605e-05, "loss": 0.2686, "step": 30988 }, { "epoch": 0.03, "learning_rate": 4.9882357724707665e-05, "loss": 1.0851, "step": 30989 }, { "epoch": 0.03, "learning_rate": 4.988235011346609e-05, "loss": 0.6612, "step": 30990 }, { "epoch": 0.03, "learning_rate": 4.988234250197889e-05, "loss": 0.8276, "step": 30991 }, { "epoch": 0.03, "learning_rate": 4.988233489024606e-05, "loss": 0.8289, "step": 30992 }, { "epoch": 0.03, "learning_rate": 4.988232727826761e-05, "loss": 1.2702, "step": 30993 }, { "epoch": 0.03, "learning_rate": 4.988231966604353e-05, "loss": 0.7934, "step": 30994 }, { "epoch": 0.03, "learning_rate": 4.988231205357381e-05, "loss": 0.7554, "step": 30995 }, { "epoch": 0.03, "learning_rate": 4.9882304440858476e-05, "loss": 1.058, "step": 30996 }, { "epoch": 0.03, "learning_rate": 4.9882296827897515e-05, "loss": 1.2017, "step": 30997 }, { "epoch": 0.03, "learning_rate": 4.9882289214690914e-05, "loss": 0.8772, "step": 30998 }, { "epoch": 0.03, "learning_rate": 4.9882281601238694e-05, "loss": 1.1035, "step": 30999 }, { "epoch": 0.03, "learning_rate": 4.988227398754085e-05, "loss": 0.9526, "step": 31000 }, { "epoch": 0.03, "eval_loss": 1.0707390308380127, "eval_runtime": 126.0963, "eval_samples_per_second": 10.984, "eval_steps_per_second": 5.496, "step": 31000 }, { "epoch": 0.03, "learning_rate": 4.9882266373597375e-05, "loss": 0.7795, "step": 31001 }, { "epoch": 0.03, "learning_rate": 4.988225875940827e-05, "loss": 1.2093, "step": 31002 }, { "epoch": 0.03, "learning_rate": 4.988225114497353e-05, "loss": 0.979, "step": 31003 }, { "epoch": 0.03, "learning_rate": 4.988224353029318e-05, "loss": 0.9168, "step": 31004 }, { "epoch": 0.03, "learning_rate": 4.98822359153672e-05, "loss": 1.0257, "step": 31005 }, { "epoch": 0.03, "learning_rate": 4.988222830019559e-05, "loss": 1.0683, "step": 31006 }, { "epoch": 0.03, "learning_rate": 4.988222068477835e-05, "loss": 1.2736, "step": 31007 }, { "epoch": 0.03, "learning_rate": 4.988221306911548e-05, "loss": 0.876, "step": 31008 }, { "epoch": 0.03, "learning_rate": 4.988220545320699e-05, "loss": 1.008, "step": 31009 }, { "epoch": 0.03, "learning_rate": 4.988219783705287e-05, "loss": 0.9431, "step": 31010 }, { "epoch": 0.03, "learning_rate": 4.988219022065313e-05, "loss": 0.8674, "step": 31011 }, { "epoch": 0.03, "learning_rate": 4.9882182604007755e-05, "loss": 0.6015, "step": 31012 }, { "epoch": 0.03, "learning_rate": 4.988217498711676e-05, "loss": 1.0582, "step": 31013 }, { "epoch": 0.03, "learning_rate": 4.988216736998014e-05, "loss": 1.0658, "step": 31014 }, { "epoch": 0.03, "learning_rate": 4.9882159752597884e-05, "loss": 0.994, "step": 31015 }, { "epoch": 0.03, "learning_rate": 4.988215213497001e-05, "loss": 1.0014, "step": 31016 }, { "epoch": 0.03, "learning_rate": 4.9882144517096506e-05, "loss": 0.8718, "step": 31017 }, { "epoch": 0.03, "learning_rate": 4.988213689897738e-05, "loss": 1.1802, "step": 31018 }, { "epoch": 0.03, "learning_rate": 4.9882129280612624e-05, "loss": 0.9197, "step": 31019 }, { "epoch": 0.03, "learning_rate": 4.9882121662002244e-05, "loss": 0.8151, "step": 31020 }, { "epoch": 0.03, "learning_rate": 4.988211404314623e-05, "loss": 1.0026, "step": 31021 }, { "epoch": 0.03, "learning_rate": 4.9882106424044607e-05, "loss": 1.0432, "step": 31022 }, { "epoch": 0.03, "learning_rate": 4.988209880469734e-05, "loss": 0.9839, "step": 31023 }, { "epoch": 0.03, "learning_rate": 4.9882091185104465e-05, "loss": 0.9435, "step": 31024 }, { "epoch": 0.03, "learning_rate": 4.9882083565265955e-05, "loss": 1.1737, "step": 31025 }, { "epoch": 0.03, "learning_rate": 4.988207594518182e-05, "loss": 1.0965, "step": 31026 }, { "epoch": 0.03, "learning_rate": 4.988206832485206e-05, "loss": 1.1231, "step": 31027 }, { "epoch": 0.03, "learning_rate": 4.9882060704276676e-05, "loss": 0.9103, "step": 31028 }, { "epoch": 0.03, "learning_rate": 4.988205308345566e-05, "loss": 0.9904, "step": 31029 }, { "epoch": 0.03, "learning_rate": 4.988204546238903e-05, "loss": 1.0698, "step": 31030 }, { "epoch": 0.03, "learning_rate": 4.988203784107677e-05, "loss": 0.6189, "step": 31031 }, { "epoch": 0.03, "learning_rate": 4.9882030219518885e-05, "loss": 0.8418, "step": 31032 }, { "epoch": 0.03, "learning_rate": 4.9882022597715374e-05, "loss": 0.3539, "step": 31033 }, { "epoch": 0.03, "learning_rate": 4.9882014975666236e-05, "loss": 0.9451, "step": 31034 }, { "epoch": 0.03, "learning_rate": 4.988200735337147e-05, "loss": 1.2266, "step": 31035 }, { "epoch": 0.03, "learning_rate": 4.988199973083109e-05, "loss": 1.4084, "step": 31036 }, { "epoch": 0.03, "learning_rate": 4.988199210804508e-05, "loss": 1.0336, "step": 31037 }, { "epoch": 0.03, "learning_rate": 4.988198448501345e-05, "loss": 1.37, "step": 31038 }, { "epoch": 0.03, "learning_rate": 4.988197686173619e-05, "loss": 1.0472, "step": 31039 }, { "epoch": 0.03, "learning_rate": 4.988196923821331e-05, "loss": 1.3074, "step": 31040 }, { "epoch": 0.03, "learning_rate": 4.98819616144448e-05, "loss": 1.2019, "step": 31041 }, { "epoch": 0.03, "learning_rate": 4.988195399043067e-05, "loss": 0.778, "step": 31042 }, { "epoch": 0.03, "learning_rate": 4.988194636617092e-05, "loss": 1.1895, "step": 31043 }, { "epoch": 0.03, "learning_rate": 4.9881938741665537e-05, "loss": 1.0543, "step": 31044 }, { "epoch": 0.03, "learning_rate": 4.9881931116914534e-05, "loss": 1.1415, "step": 31045 }, { "epoch": 0.03, "learning_rate": 4.9881923491917905e-05, "loss": 1.052, "step": 31046 }, { "epoch": 0.03, "learning_rate": 4.988191586667566e-05, "loss": 0.9192, "step": 31047 }, { "epoch": 0.03, "learning_rate": 4.988190824118778e-05, "loss": 1.1586, "step": 31048 }, { "epoch": 0.03, "learning_rate": 4.988190061545428e-05, "loss": 1.052, "step": 31049 }, { "epoch": 0.03, "learning_rate": 4.988189298947516e-05, "loss": 0.8031, "step": 31050 }, { "epoch": 0.03, "learning_rate": 4.988188536325041e-05, "loss": 0.9432, "step": 31051 }, { "epoch": 0.03, "learning_rate": 4.988187773678003e-05, "loss": 1.044, "step": 31052 }, { "epoch": 0.03, "learning_rate": 4.988187011006404e-05, "loss": 1.0176, "step": 31053 }, { "epoch": 0.03, "learning_rate": 4.988186248310243e-05, "loss": 0.9243, "step": 31054 }, { "epoch": 0.03, "learning_rate": 4.988185485589519e-05, "loss": 1.023, "step": 31055 }, { "epoch": 0.03, "learning_rate": 4.988184722844233e-05, "loss": 1.1306, "step": 31056 }, { "epoch": 0.03, "learning_rate": 4.988183960074384e-05, "loss": 0.8476, "step": 31057 }, { "epoch": 0.03, "learning_rate": 4.9881831972799734e-05, "loss": 0.9711, "step": 31058 }, { "epoch": 0.03, "learning_rate": 4.988182434461e-05, "loss": 0.8711, "step": 31059 }, { "epoch": 0.03, "learning_rate": 4.988181671617464e-05, "loss": 1.1508, "step": 31060 }, { "epoch": 0.03, "learning_rate": 4.988180908749367e-05, "loss": 1.3703, "step": 31061 }, { "epoch": 0.03, "learning_rate": 4.9881801458567066e-05, "loss": 1.0606, "step": 31062 }, { "epoch": 0.03, "learning_rate": 4.988179382939484e-05, "loss": 0.8126, "step": 31063 }, { "epoch": 0.03, "learning_rate": 4.9881786199977e-05, "loss": 1.1637, "step": 31064 }, { "epoch": 0.03, "learning_rate": 4.988177857031353e-05, "loss": 0.9856, "step": 31065 }, { "epoch": 0.03, "learning_rate": 4.9881770940404444e-05, "loss": 1.1652, "step": 31066 }, { "epoch": 0.03, "learning_rate": 4.988176331024973e-05, "loss": 0.8629, "step": 31067 }, { "epoch": 0.03, "learning_rate": 4.98817556798494e-05, "loss": 1.0797, "step": 31068 }, { "epoch": 0.03, "learning_rate": 4.988174804920344e-05, "loss": 1.0835, "step": 31069 }, { "epoch": 0.03, "learning_rate": 4.988174041831186e-05, "loss": 0.9097, "step": 31070 }, { "epoch": 0.03, "learning_rate": 4.9881732787174655e-05, "loss": 1.2518, "step": 31071 }, { "epoch": 0.03, "learning_rate": 4.988172515579184e-05, "loss": 0.9548, "step": 31072 }, { "epoch": 0.03, "learning_rate": 4.988171752416338e-05, "loss": 0.7177, "step": 31073 }, { "epoch": 0.03, "learning_rate": 4.988170989228932e-05, "loss": 0.8398, "step": 31074 }, { "epoch": 0.03, "learning_rate": 4.988170226016963e-05, "loss": 1.6581, "step": 31075 }, { "epoch": 0.03, "learning_rate": 4.988169462780432e-05, "loss": 1.3204, "step": 31076 }, { "epoch": 0.03, "learning_rate": 4.988168699519339e-05, "loss": 0.9408, "step": 31077 }, { "epoch": 0.03, "learning_rate": 4.988167936233684e-05, "loss": 0.8958, "step": 31078 }, { "epoch": 0.03, "learning_rate": 4.988167172923466e-05, "loss": 1.0068, "step": 31079 }, { "epoch": 0.03, "learning_rate": 4.988166409588686e-05, "loss": 0.9276, "step": 31080 }, { "epoch": 0.03, "learning_rate": 4.9881656462293446e-05, "loss": 1.1198, "step": 31081 }, { "epoch": 0.03, "learning_rate": 4.988164882845441e-05, "loss": 1.3312, "step": 31082 }, { "epoch": 0.03, "learning_rate": 4.988164119436974e-05, "loss": 1.5236, "step": 31083 }, { "epoch": 0.03, "learning_rate": 4.9881633560039466e-05, "loss": 1.5965, "step": 31084 }, { "epoch": 0.03, "learning_rate": 4.988162592546356e-05, "loss": 2.0023, "step": 31085 }, { "epoch": 0.03, "learning_rate": 4.988161829064204e-05, "loss": 1.9275, "step": 31086 }, { "epoch": 0.03, "learning_rate": 4.988161065557489e-05, "loss": 1.0676, "step": 31087 }, { "epoch": 0.03, "learning_rate": 4.988160302026213e-05, "loss": 0.9395, "step": 31088 }, { "epoch": 0.03, "learning_rate": 4.9881595384703743e-05, "loss": 1.0419, "step": 31089 }, { "epoch": 0.03, "learning_rate": 4.988158774889974e-05, "loss": 0.9633, "step": 31090 }, { "epoch": 0.03, "learning_rate": 4.988158011285011e-05, "loss": 1.0848, "step": 31091 }, { "epoch": 0.03, "learning_rate": 4.988157247655486e-05, "loss": 0.7604, "step": 31092 }, { "epoch": 0.03, "learning_rate": 4.9881564840013994e-05, "loss": 0.893, "step": 31093 }, { "epoch": 0.03, "learning_rate": 4.9881557203227506e-05, "loss": 0.5856, "step": 31094 }, { "epoch": 0.03, "learning_rate": 4.988154956619539e-05, "loss": 0.797, "step": 31095 }, { "epoch": 0.03, "learning_rate": 4.9881541928917674e-05, "loss": 1.2096, "step": 31096 }, { "epoch": 0.03, "learning_rate": 4.988153429139432e-05, "loss": 0.8775, "step": 31097 }, { "epoch": 0.03, "learning_rate": 4.988152665362535e-05, "loss": 1.0341, "step": 31098 }, { "epoch": 0.03, "learning_rate": 4.988151901561077e-05, "loss": 0.4907, "step": 31099 }, { "epoch": 0.03, "learning_rate": 4.988151137735055e-05, "loss": 0.9136, "step": 31100 }, { "epoch": 0.03, "learning_rate": 4.9881503738844724e-05, "loss": 0.9816, "step": 31101 }, { "epoch": 0.03, "learning_rate": 4.988149610009328e-05, "loss": 0.6617, "step": 31102 }, { "epoch": 0.03, "learning_rate": 4.9881488461096216e-05, "loss": 0.6599, "step": 31103 }, { "epoch": 0.03, "learning_rate": 4.988148082185353e-05, "loss": 0.9855, "step": 31104 }, { "epoch": 0.03, "learning_rate": 4.9881473182365224e-05, "loss": 0.4911, "step": 31105 }, { "epoch": 0.03, "learning_rate": 4.988146554263129e-05, "loss": 0.9109, "step": 31106 }, { "epoch": 0.03, "learning_rate": 4.988145790265175e-05, "loss": 1.0325, "step": 31107 }, { "epoch": 0.03, "learning_rate": 4.9881450262426586e-05, "loss": 1.1878, "step": 31108 }, { "epoch": 0.03, "learning_rate": 4.9881442621955796e-05, "loss": 0.8843, "step": 31109 }, { "epoch": 0.03, "learning_rate": 4.9881434981239395e-05, "loss": 1.089, "step": 31110 }, { "epoch": 0.03, "learning_rate": 4.9881427340277374e-05, "loss": 1.1829, "step": 31111 }, { "epoch": 0.03, "learning_rate": 4.9881419699069734e-05, "loss": 0.8589, "step": 31112 }, { "epoch": 0.03, "learning_rate": 4.9881412057616475e-05, "loss": 1.1502, "step": 31113 }, { "epoch": 0.03, "learning_rate": 4.9881404415917596e-05, "loss": 0.9139, "step": 31114 }, { "epoch": 0.03, "learning_rate": 4.98813967739731e-05, "loss": 0.8628, "step": 31115 }, { "epoch": 0.03, "learning_rate": 4.988138913178299e-05, "loss": 1.1171, "step": 31116 }, { "epoch": 0.03, "learning_rate": 4.9881381489347245e-05, "loss": 0.8848, "step": 31117 }, { "epoch": 0.03, "learning_rate": 4.98813738466659e-05, "loss": 0.4539, "step": 31118 }, { "epoch": 0.03, "learning_rate": 4.988136620373892e-05, "loss": 1.135, "step": 31119 }, { "epoch": 0.03, "learning_rate": 4.9881358560566335e-05, "loss": 1.098, "step": 31120 }, { "epoch": 0.03, "learning_rate": 4.988135091714813e-05, "loss": 0.9795, "step": 31121 }, { "epoch": 0.03, "learning_rate": 4.9881343273484296e-05, "loss": 1.0612, "step": 31122 }, { "epoch": 0.03, "learning_rate": 4.988133562957485e-05, "loss": 0.6399, "step": 31123 }, { "epoch": 0.03, "learning_rate": 4.988132798541979e-05, "loss": 0.6805, "step": 31124 }, { "epoch": 0.03, "learning_rate": 4.988132034101911e-05, "loss": 0.2489, "step": 31125 }, { "epoch": 0.03, "learning_rate": 4.988131269637281e-05, "loss": 0.3996, "step": 31126 }, { "epoch": 0.03, "learning_rate": 4.98813050514809e-05, "loss": 0.518, "step": 31127 }, { "epoch": 0.03, "learning_rate": 4.9881297406343367e-05, "loss": 0.5071, "step": 31128 }, { "epoch": 0.03, "learning_rate": 4.9881289760960206e-05, "loss": 0.9123, "step": 31129 }, { "epoch": 0.03, "learning_rate": 4.988128211533144e-05, "loss": 0.8301, "step": 31130 }, { "epoch": 0.03, "learning_rate": 4.988127446945705e-05, "loss": 0.9191, "step": 31131 }, { "epoch": 0.03, "learning_rate": 4.9881266823337045e-05, "loss": 1.0048, "step": 31132 }, { "epoch": 0.03, "learning_rate": 4.988125917697143e-05, "loss": 0.9482, "step": 31133 }, { "epoch": 0.03, "learning_rate": 4.988125153036019e-05, "loss": 1.145, "step": 31134 }, { "epoch": 0.03, "learning_rate": 4.988124388350333e-05, "loss": 1.0047, "step": 31135 }, { "epoch": 0.03, "learning_rate": 4.9881236236400856e-05, "loss": 0.9471, "step": 31136 }, { "epoch": 0.03, "learning_rate": 4.988122858905277e-05, "loss": 1.1327, "step": 31137 }, { "epoch": 0.03, "learning_rate": 4.988122094145906e-05, "loss": 0.8633, "step": 31138 }, { "epoch": 0.03, "learning_rate": 4.988121329361973e-05, "loss": 0.9268, "step": 31139 }, { "epoch": 0.03, "learning_rate": 4.988120564553479e-05, "loss": 1.0376, "step": 31140 }, { "epoch": 0.03, "learning_rate": 4.9881197997204236e-05, "loss": 1.1929, "step": 31141 }, { "epoch": 0.03, "learning_rate": 4.988119034862806e-05, "loss": 1.0171, "step": 31142 }, { "epoch": 0.03, "learning_rate": 4.988118269980627e-05, "loss": 1.0079, "step": 31143 }, { "epoch": 0.03, "learning_rate": 4.988117505073886e-05, "loss": 0.9677, "step": 31144 }, { "epoch": 0.03, "learning_rate": 4.9881167401425837e-05, "loss": 1.2103, "step": 31145 }, { "epoch": 0.03, "learning_rate": 4.9881159751867204e-05, "loss": 1.1452, "step": 31146 }, { "epoch": 0.03, "learning_rate": 4.988115210206294e-05, "loss": 1.0002, "step": 31147 }, { "epoch": 0.03, "learning_rate": 4.988114445201307e-05, "loss": 0.9408, "step": 31148 }, { "epoch": 0.03, "learning_rate": 4.988113680171758e-05, "loss": 1.1108, "step": 31149 }, { "epoch": 0.03, "learning_rate": 4.9881129151176474e-05, "loss": 1.169, "step": 31150 }, { "epoch": 0.03, "learning_rate": 4.988112150038975e-05, "loss": 0.6471, "step": 31151 }, { "epoch": 0.03, "learning_rate": 4.988111384935741e-05, "loss": 0.9441, "step": 31152 }, { "epoch": 0.03, "learning_rate": 4.988110619807946e-05, "loss": 0.7557, "step": 31153 }, { "epoch": 0.03, "learning_rate": 4.988109854655589e-05, "loss": 1.2642, "step": 31154 }, { "epoch": 0.03, "learning_rate": 4.9881090894786705e-05, "loss": 1.0488, "step": 31155 }, { "epoch": 0.03, "learning_rate": 4.98810832427719e-05, "loss": 0.9651, "step": 31156 }, { "epoch": 0.03, "learning_rate": 4.988107559051148e-05, "loss": 0.8601, "step": 31157 }, { "epoch": 0.03, "learning_rate": 4.9881067938005454e-05, "loss": 1.0281, "step": 31158 }, { "epoch": 0.03, "learning_rate": 4.988106028525381e-05, "loss": 1.2271, "step": 31159 }, { "epoch": 0.03, "learning_rate": 4.9881052632256544e-05, "loss": 1.0188, "step": 31160 }, { "epoch": 0.03, "learning_rate": 4.988104497901367e-05, "loss": 0.6781, "step": 31161 }, { "epoch": 0.03, "learning_rate": 4.988103732552518e-05, "loss": 1.0964, "step": 31162 }, { "epoch": 0.03, "learning_rate": 4.988102967179107e-05, "loss": 0.5624, "step": 31163 }, { "epoch": 0.03, "learning_rate": 4.988102201781134e-05, "loss": 0.2672, "step": 31164 }, { "epoch": 0.03, "learning_rate": 4.988101436358601e-05, "loss": 0.7303, "step": 31165 }, { "epoch": 0.03, "learning_rate": 4.9881006709115054e-05, "loss": 0.976, "step": 31166 }, { "epoch": 0.03, "learning_rate": 4.988099905439848e-05, "loss": 0.9867, "step": 31167 }, { "epoch": 0.03, "learning_rate": 4.9880991399436304e-05, "loss": 1.0032, "step": 31168 }, { "epoch": 0.03, "learning_rate": 4.98809837442285e-05, "loss": 1.0153, "step": 31169 }, { "epoch": 0.03, "learning_rate": 4.98809760887751e-05, "loss": 0.9615, "step": 31170 }, { "epoch": 0.03, "learning_rate": 4.988096843307607e-05, "loss": 1.0578, "step": 31171 }, { "epoch": 0.03, "learning_rate": 4.988096077713143e-05, "loss": 1.0015, "step": 31172 }, { "epoch": 0.03, "learning_rate": 4.988095312094117e-05, "loss": 0.6541, "step": 31173 }, { "epoch": 0.03, "learning_rate": 4.98809454645053e-05, "loss": 1.0696, "step": 31174 }, { "epoch": 0.03, "learning_rate": 4.988093780782382e-05, "loss": 0.8195, "step": 31175 }, { "epoch": 0.03, "learning_rate": 4.988093015089672e-05, "loss": 0.9658, "step": 31176 }, { "epoch": 0.03, "learning_rate": 4.988092249372401e-05, "loss": 0.9435, "step": 31177 }, { "epoch": 0.03, "learning_rate": 4.988091483630568e-05, "loss": 1.1178, "step": 31178 }, { "epoch": 0.03, "learning_rate": 4.988090717864175e-05, "loss": 0.8645, "step": 31179 }, { "epoch": 0.03, "learning_rate": 4.988089952073219e-05, "loss": 1.0417, "step": 31180 }, { "epoch": 0.03, "learning_rate": 4.9880891862577026e-05, "loss": 1.0651, "step": 31181 }, { "epoch": 0.03, "learning_rate": 4.9880884204176237e-05, "loss": 1.1419, "step": 31182 }, { "epoch": 0.03, "learning_rate": 4.988087654552985e-05, "loss": 0.9721, "step": 31183 }, { "epoch": 0.03, "learning_rate": 4.988086888663784e-05, "loss": 0.9936, "step": 31184 }, { "epoch": 0.03, "learning_rate": 4.988086122750022e-05, "loss": 1.0207, "step": 31185 }, { "epoch": 0.03, "learning_rate": 4.988085356811698e-05, "loss": 0.9599, "step": 31186 }, { "epoch": 0.03, "learning_rate": 4.988084590848813e-05, "loss": 1.019, "step": 31187 }, { "epoch": 0.03, "learning_rate": 4.988083824861367e-05, "loss": 0.8259, "step": 31188 }, { "epoch": 0.03, "learning_rate": 4.9880830588493596e-05, "loss": 0.6954, "step": 31189 }, { "epoch": 0.03, "learning_rate": 4.9880822928127914e-05, "loss": 1.0704, "step": 31190 }, { "epoch": 0.03, "learning_rate": 4.988081526751661e-05, "loss": 0.2421, "step": 31191 }, { "epoch": 0.03, "learning_rate": 4.98808076066597e-05, "loss": 0.1939, "step": 31192 }, { "epoch": 0.03, "learning_rate": 4.9880799945557166e-05, "loss": 0.1959, "step": 31193 }, { "epoch": 0.03, "learning_rate": 4.9880792284209034e-05, "loss": 0.2206, "step": 31194 }, { "epoch": 0.03, "learning_rate": 4.9880784622615276e-05, "loss": 0.2204, "step": 31195 }, { "epoch": 0.03, "learning_rate": 4.988077696077591e-05, "loss": 0.118, "step": 31196 }, { "epoch": 0.03, "learning_rate": 4.988076929869093e-05, "loss": 0.1738, "step": 31197 }, { "epoch": 0.03, "learning_rate": 4.988076163636035e-05, "loss": 0.2748, "step": 31198 }, { "epoch": 0.03, "learning_rate": 4.988075397378415e-05, "loss": 0.2908, "step": 31199 }, { "epoch": 0.03, "learning_rate": 4.988074631096233e-05, "loss": 1.133, "step": 31200 }, { "epoch": 0.03, "learning_rate": 4.98807386478949e-05, "loss": 1.1462, "step": 31201 }, { "epoch": 0.03, "learning_rate": 4.988073098458187e-05, "loss": 1.2061, "step": 31202 }, { "epoch": 0.03, "learning_rate": 4.988072332102322e-05, "loss": 0.4324, "step": 31203 }, { "epoch": 0.03, "learning_rate": 4.988071565721895e-05, "loss": 0.4541, "step": 31204 }, { "epoch": 0.03, "learning_rate": 4.988070799316907e-05, "loss": 0.6359, "step": 31205 }, { "epoch": 0.03, "learning_rate": 4.9880700328873584e-05, "loss": 1.0166, "step": 31206 }, { "epoch": 0.03, "learning_rate": 4.98806926643325e-05, "loss": 1.4289, "step": 31207 }, { "epoch": 0.03, "learning_rate": 4.988068499954578e-05, "loss": 0.9895, "step": 31208 }, { "epoch": 0.03, "learning_rate": 4.988067733451346e-05, "loss": 1.0679, "step": 31209 }, { "epoch": 0.03, "learning_rate": 4.988066966923553e-05, "loss": 1.2157, "step": 31210 }, { "epoch": 0.03, "learning_rate": 4.9880662003711984e-05, "loss": 1.1808, "step": 31211 }, { "epoch": 0.03, "learning_rate": 4.988065433794282e-05, "loss": 0.9076, "step": 31212 }, { "epoch": 0.03, "learning_rate": 4.988064667192805e-05, "loss": 0.9739, "step": 31213 }, { "epoch": 0.03, "learning_rate": 4.988063900566768e-05, "loss": 1.1349, "step": 31214 }, { "epoch": 0.03, "learning_rate": 4.9880631339161685e-05, "loss": 0.9991, "step": 31215 }, { "epoch": 0.03, "learning_rate": 4.9880623672410086e-05, "loss": 1.1511, "step": 31216 }, { "epoch": 0.03, "learning_rate": 4.9880616005412874e-05, "loss": 0.5119, "step": 31217 }, { "epoch": 0.03, "learning_rate": 4.988060833817006e-05, "loss": 0.792, "step": 31218 }, { "epoch": 0.03, "learning_rate": 4.988060067068162e-05, "loss": 0.8691, "step": 31219 }, { "epoch": 0.03, "learning_rate": 4.988059300294757e-05, "loss": 1.0, "step": 31220 }, { "epoch": 0.03, "learning_rate": 4.9880585334967924e-05, "loss": 1.0261, "step": 31221 }, { "epoch": 0.03, "learning_rate": 4.988057766674266e-05, "loss": 1.1777, "step": 31222 }, { "epoch": 0.03, "learning_rate": 4.988056999827178e-05, "loss": 0.5877, "step": 31223 }, { "epoch": 0.03, "learning_rate": 4.988056232955529e-05, "loss": 1.0576, "step": 31224 }, { "epoch": 0.03, "learning_rate": 4.9880554660593195e-05, "loss": 1.0662, "step": 31225 }, { "epoch": 0.03, "learning_rate": 4.988054699138548e-05, "loss": 0.9929, "step": 31226 }, { "epoch": 0.03, "learning_rate": 4.988053932193216e-05, "loss": 0.7149, "step": 31227 }, { "epoch": 0.03, "learning_rate": 4.9880531652233234e-05, "loss": 1.0287, "step": 31228 }, { "epoch": 0.03, "learning_rate": 4.9880523982288694e-05, "loss": 1.0477, "step": 31229 }, { "epoch": 0.03, "learning_rate": 4.988051631209855e-05, "loss": 1.1147, "step": 31230 }, { "epoch": 0.03, "learning_rate": 4.9880508641662796e-05, "loss": 0.8166, "step": 31231 }, { "epoch": 0.03, "learning_rate": 4.9880500970981424e-05, "loss": 1.028, "step": 31232 }, { "epoch": 0.03, "learning_rate": 4.988049330005444e-05, "loss": 0.6914, "step": 31233 }, { "epoch": 0.03, "learning_rate": 4.988048562888186e-05, "loss": 0.9805, "step": 31234 }, { "epoch": 0.03, "learning_rate": 4.9880477957463656e-05, "loss": 0.906, "step": 31235 }, { "epoch": 0.03, "learning_rate": 4.988047028579985e-05, "loss": 0.8714, "step": 31236 }, { "epoch": 0.03, "learning_rate": 4.9880462613890436e-05, "loss": 0.9737, "step": 31237 }, { "epoch": 0.03, "learning_rate": 4.988045494173541e-05, "loss": 0.9573, "step": 31238 }, { "epoch": 0.03, "learning_rate": 4.988044726933477e-05, "loss": 1.0233, "step": 31239 }, { "epoch": 0.03, "learning_rate": 4.988043959668853e-05, "loss": 0.7674, "step": 31240 }, { "epoch": 0.03, "learning_rate": 4.988043192379667e-05, "loss": 0.1684, "step": 31241 }, { "epoch": 0.03, "learning_rate": 4.988042425065921e-05, "loss": 1.0099, "step": 31242 }, { "epoch": 0.03, "learning_rate": 4.988041657727614e-05, "loss": 1.1556, "step": 31243 }, { "epoch": 0.03, "learning_rate": 4.988040890364746e-05, "loss": 1.0267, "step": 31244 }, { "epoch": 0.03, "learning_rate": 4.9880401229773164e-05, "loss": 0.6688, "step": 31245 }, { "epoch": 0.03, "learning_rate": 4.9880393555653266e-05, "loss": 0.5381, "step": 31246 }, { "epoch": 0.03, "learning_rate": 4.988038588128776e-05, "loss": 0.5451, "step": 31247 }, { "epoch": 0.03, "learning_rate": 4.988037820667664e-05, "loss": 0.987, "step": 31248 }, { "epoch": 0.03, "learning_rate": 4.988037053181992e-05, "loss": 1.3244, "step": 31249 }, { "epoch": 0.03, "learning_rate": 4.9880362856717584e-05, "loss": 1.0293, "step": 31250 }, { "epoch": 0.03, "learning_rate": 4.9880355181369644e-05, "loss": 0.9325, "step": 31251 }, { "epoch": 0.03, "learning_rate": 4.988034750577609e-05, "loss": 0.9609, "step": 31252 }, { "epoch": 0.03, "learning_rate": 4.9880339829936926e-05, "loss": 0.9002, "step": 31253 }, { "epoch": 0.03, "learning_rate": 4.988033215385216e-05, "loss": 0.9267, "step": 31254 }, { "epoch": 0.03, "learning_rate": 4.9880324477521786e-05, "loss": 0.8006, "step": 31255 }, { "epoch": 0.03, "learning_rate": 4.9880316800945805e-05, "loss": 0.7568, "step": 31256 }, { "epoch": 0.03, "learning_rate": 4.988030912412422e-05, "loss": 0.7206, "step": 31257 }, { "epoch": 0.03, "learning_rate": 4.988030144705702e-05, "loss": 0.67, "step": 31258 }, { "epoch": 0.03, "learning_rate": 4.9880293769744205e-05, "loss": 0.6345, "step": 31259 }, { "epoch": 0.03, "learning_rate": 4.988028609218579e-05, "loss": 0.6068, "step": 31260 }, { "epoch": 0.03, "learning_rate": 4.988027841438177e-05, "loss": 0.588, "step": 31261 }, { "epoch": 0.03, "learning_rate": 4.988027073633214e-05, "loss": 0.5887, "step": 31262 }, { "epoch": 0.03, "learning_rate": 4.9880263058036905e-05, "loss": 0.5562, "step": 31263 }, { "epoch": 0.03, "learning_rate": 4.988025537949606e-05, "loss": 0.5501, "step": 31264 }, { "epoch": 0.03, "learning_rate": 4.9880247700709604e-05, "loss": 0.5397, "step": 31265 }, { "epoch": 0.03, "learning_rate": 4.9880240021677545e-05, "loss": 0.6111, "step": 31266 }, { "epoch": 0.03, "learning_rate": 4.988023234239988e-05, "loss": 0.5958, "step": 31267 }, { "epoch": 0.03, "learning_rate": 4.98802246628766e-05, "loss": 0.5405, "step": 31268 }, { "epoch": 0.03, "learning_rate": 4.988021698310772e-05, "loss": 0.5337, "step": 31269 }, { "epoch": 0.03, "learning_rate": 4.988020930309323e-05, "loss": 0.5649, "step": 31270 }, { "epoch": 0.03, "learning_rate": 4.988020162283313e-05, "loss": 0.5318, "step": 31271 }, { "epoch": 0.03, "learning_rate": 4.988019394232744e-05, "loss": 0.5023, "step": 31272 }, { "epoch": 0.03, "learning_rate": 4.9880186261576126e-05, "loss": 0.5171, "step": 31273 }, { "epoch": 0.03, "learning_rate": 4.988017858057921e-05, "loss": 0.5374, "step": 31274 }, { "epoch": 0.03, "learning_rate": 4.9880170899336685e-05, "loss": 0.5112, "step": 31275 }, { "epoch": 0.03, "learning_rate": 4.9880163217848556e-05, "loss": 0.4862, "step": 31276 }, { "epoch": 0.03, "learning_rate": 4.988015553611483e-05, "loss": 0.4903, "step": 31277 }, { "epoch": 0.03, "learning_rate": 4.988014785413549e-05, "loss": 0.4757, "step": 31278 }, { "epoch": 0.03, "learning_rate": 4.9880140171910534e-05, "loss": 0.4871, "step": 31279 }, { "epoch": 0.03, "learning_rate": 4.988013248943998e-05, "loss": 0.5023, "step": 31280 }, { "epoch": 0.03, "learning_rate": 4.988012480672382e-05, "loss": 0.4761, "step": 31281 }, { "epoch": 0.03, "learning_rate": 4.988011712376205e-05, "loss": 0.4745, "step": 31282 }, { "epoch": 0.03, "learning_rate": 4.988010944055468e-05, "loss": 0.4805, "step": 31283 }, { "epoch": 0.03, "learning_rate": 4.98801017571017e-05, "loss": 0.4549, "step": 31284 }, { "epoch": 0.03, "learning_rate": 4.988009407340312e-05, "loss": 0.4824, "step": 31285 }, { "epoch": 0.03, "learning_rate": 4.9880086389458924e-05, "loss": 0.4908, "step": 31286 }, { "epoch": 0.03, "learning_rate": 4.9880078705269126e-05, "loss": 0.4913, "step": 31287 }, { "epoch": 0.03, "learning_rate": 4.988007102083373e-05, "loss": 0.4724, "step": 31288 }, { "epoch": 0.03, "learning_rate": 4.988006333615272e-05, "loss": 0.4766, "step": 31289 }, { "epoch": 0.03, "learning_rate": 4.988005565122611e-05, "loss": 0.4756, "step": 31290 }, { "epoch": 0.03, "learning_rate": 4.988004796605389e-05, "loss": 0.4859, "step": 31291 }, { "epoch": 0.03, "learning_rate": 4.988004028063606e-05, "loss": 0.4854, "step": 31292 }, { "epoch": 0.03, "learning_rate": 4.9880032594972637e-05, "loss": 0.4822, "step": 31293 }, { "epoch": 0.03, "learning_rate": 4.98800249090636e-05, "loss": 0.4801, "step": 31294 }, { "epoch": 0.03, "learning_rate": 4.988001722290896e-05, "loss": 0.4689, "step": 31295 }, { "epoch": 0.03, "learning_rate": 4.988000953650872e-05, "loss": 0.4696, "step": 31296 }, { "epoch": 0.03, "learning_rate": 4.988000184986287e-05, "loss": 0.4898, "step": 31297 }, { "epoch": 0.03, "learning_rate": 4.987999416297141e-05, "loss": 0.4761, "step": 31298 }, { "epoch": 0.03, "learning_rate": 4.987998647583435e-05, "loss": 0.4554, "step": 31299 }, { "epoch": 0.03, "learning_rate": 4.9879978788451684e-05, "loss": 0.4613, "step": 31300 }, { "epoch": 0.03, "learning_rate": 4.987997110082342e-05, "loss": 0.4496, "step": 31301 }, { "epoch": 0.03, "learning_rate": 4.987996341294955e-05, "loss": 0.4686, "step": 31302 }, { "epoch": 0.03, "learning_rate": 4.9879955724830075e-05, "loss": 0.4624, "step": 31303 }, { "epoch": 0.03, "learning_rate": 4.987994803646499e-05, "loss": 0.4656, "step": 31304 }, { "epoch": 0.03, "learning_rate": 4.9879940347854306e-05, "loss": 0.4751, "step": 31305 }, { "epoch": 0.03, "learning_rate": 4.9879932658998014e-05, "loss": 0.4573, "step": 31306 }, { "epoch": 0.03, "learning_rate": 4.987992496989612e-05, "loss": 0.4659, "step": 31307 }, { "epoch": 0.03, "learning_rate": 4.987991728054862e-05, "loss": 0.4605, "step": 31308 }, { "epoch": 0.03, "learning_rate": 4.9879909590955515e-05, "loss": 0.4565, "step": 31309 }, { "epoch": 0.03, "learning_rate": 4.9879901901116813e-05, "loss": 0.4605, "step": 31310 }, { "epoch": 0.03, "learning_rate": 4.98798942110325e-05, "loss": 0.448, "step": 31311 }, { "epoch": 0.03, "learning_rate": 4.9879886520702586e-05, "loss": 0.4585, "step": 31312 }, { "epoch": 0.03, "learning_rate": 4.987987883012706e-05, "loss": 0.456, "step": 31313 }, { "epoch": 0.03, "learning_rate": 4.987987113930594e-05, "loss": 0.458, "step": 31314 }, { "epoch": 0.03, "learning_rate": 4.987986344823922e-05, "loss": 0.4532, "step": 31315 }, { "epoch": 0.03, "learning_rate": 4.9879855756926884e-05, "loss": 0.4632, "step": 31316 }, { "epoch": 0.03, "learning_rate": 4.9879848065368956e-05, "loss": 0.4532, "step": 31317 }, { "epoch": 0.03, "learning_rate": 4.9879840373565415e-05, "loss": 0.453, "step": 31318 }, { "epoch": 0.03, "learning_rate": 4.987983268151628e-05, "loss": 0.4619, "step": 31319 }, { "epoch": 0.03, "learning_rate": 4.987982498922154e-05, "loss": 0.4454, "step": 31320 }, { "epoch": 0.03, "learning_rate": 4.9879817296681194e-05, "loss": 0.4677, "step": 31321 }, { "epoch": 0.03, "learning_rate": 4.9879809603895244e-05, "loss": 0.4514, "step": 31322 }, { "epoch": 0.03, "learning_rate": 4.987980191086369e-05, "loss": 0.4483, "step": 31323 }, { "epoch": 0.03, "learning_rate": 4.9879794217586535e-05, "loss": 0.4437, "step": 31324 }, { "epoch": 0.03, "learning_rate": 4.9879786524063775e-05, "loss": 0.4434, "step": 31325 }, { "epoch": 0.03, "learning_rate": 4.987977883029542e-05, "loss": 0.4409, "step": 31326 }, { "epoch": 0.03, "learning_rate": 4.987977113628145e-05, "loss": 0.4442, "step": 31327 }, { "epoch": 0.03, "learning_rate": 4.987976344202189e-05, "loss": 0.4423, "step": 31328 }, { "epoch": 0.03, "learning_rate": 4.987975574751672e-05, "loss": 0.4465, "step": 31329 }, { "epoch": 0.03, "learning_rate": 4.987974805276595e-05, "loss": 0.437, "step": 31330 }, { "epoch": 0.03, "learning_rate": 4.987974035776958e-05, "loss": 0.4347, "step": 31331 }, { "epoch": 0.03, "learning_rate": 4.98797326625276e-05, "loss": 0.4386, "step": 31332 }, { "epoch": 0.03, "learning_rate": 4.987972496704002e-05, "loss": 0.4444, "step": 31333 }, { "epoch": 0.03, "learning_rate": 4.9879717271306844e-05, "loss": 0.4403, "step": 31334 }, { "epoch": 0.03, "learning_rate": 4.987970957532806e-05, "loss": 0.4377, "step": 31335 }, { "epoch": 0.03, "learning_rate": 4.987970187910368e-05, "loss": 0.4471, "step": 31336 }, { "epoch": 0.03, "learning_rate": 4.987969418263369e-05, "loss": 0.4293, "step": 31337 }, { "epoch": 0.03, "learning_rate": 4.98796864859181e-05, "loss": 0.4498, "step": 31338 }, { "epoch": 0.03, "learning_rate": 4.9879678788956915e-05, "loss": 0.4412, "step": 31339 }, { "epoch": 0.03, "learning_rate": 4.9879671091750125e-05, "loss": 0.4269, "step": 31340 }, { "epoch": 0.03, "learning_rate": 4.987966339429773e-05, "loss": 0.4225, "step": 31341 }, { "epoch": 0.03, "learning_rate": 4.987965569659974e-05, "loss": 0.4468, "step": 31342 }, { "epoch": 0.03, "learning_rate": 4.9879647998656146e-05, "loss": 0.4182, "step": 31343 }, { "epoch": 0.03, "learning_rate": 4.9879640300466946e-05, "loss": 0.432, "step": 31344 }, { "epoch": 0.03, "learning_rate": 4.987963260203215e-05, "loss": 0.4348, "step": 31345 }, { "epoch": 0.03, "learning_rate": 4.987962490335174e-05, "loss": 0.4464, "step": 31346 }, { "epoch": 0.03, "learning_rate": 4.9879617204425746e-05, "loss": 0.4143, "step": 31347 }, { "epoch": 0.03, "learning_rate": 4.9879609505254144e-05, "loss": 0.4289, "step": 31348 }, { "epoch": 0.03, "learning_rate": 4.987960180583694e-05, "loss": 0.4219, "step": 31349 }, { "epoch": 0.03, "learning_rate": 4.987959410617413e-05, "loss": 0.4398, "step": 31350 }, { "epoch": 0.03, "learning_rate": 4.987958640626573e-05, "loss": 0.4067, "step": 31351 }, { "epoch": 0.03, "learning_rate": 4.9879578706111726e-05, "loss": 0.4154, "step": 31352 }, { "epoch": 0.03, "learning_rate": 4.9879571005712116e-05, "loss": 0.4181, "step": 31353 }, { "epoch": 0.03, "learning_rate": 4.987956330506691e-05, "loss": 0.4394, "step": 31354 }, { "epoch": 0.03, "learning_rate": 4.98795556041761e-05, "loss": 0.4179, "step": 31355 }, { "epoch": 0.03, "learning_rate": 4.987954790303969e-05, "loss": 0.4269, "step": 31356 }, { "epoch": 0.03, "learning_rate": 4.987954020165768e-05, "loss": 0.4202, "step": 31357 }, { "epoch": 0.03, "learning_rate": 4.987953250003008e-05, "loss": 0.4203, "step": 31358 }, { "epoch": 0.03, "learning_rate": 4.987952479815687e-05, "loss": 0.4249, "step": 31359 }, { "epoch": 0.03, "learning_rate": 4.987951709603806e-05, "loss": 0.4199, "step": 31360 }, { "epoch": 0.03, "learning_rate": 4.987950939367365e-05, "loss": 0.4352, "step": 31361 }, { "epoch": 0.03, "learning_rate": 4.987950169106363e-05, "loss": 0.4265, "step": 31362 }, { "epoch": 0.03, "learning_rate": 4.987949398820803e-05, "loss": 0.4129, "step": 31363 }, { "epoch": 0.03, "learning_rate": 4.987948628510682e-05, "loss": 0.4278, "step": 31364 }, { "epoch": 0.03, "learning_rate": 4.987947858176001e-05, "loss": 0.6878, "step": 31365 }, { "epoch": 0.03, "learning_rate": 4.98794708781676e-05, "loss": 1.3427, "step": 31366 }, { "epoch": 0.03, "learning_rate": 4.9879463174329594e-05, "loss": 1.7052, "step": 31367 }, { "epoch": 0.03, "learning_rate": 4.9879455470245986e-05, "loss": 1.1452, "step": 31368 }, { "epoch": 0.03, "learning_rate": 4.9879447765916773e-05, "loss": 1.1242, "step": 31369 }, { "epoch": 0.03, "learning_rate": 4.987944006134197e-05, "loss": 1.1614, "step": 31370 }, { "epoch": 0.03, "learning_rate": 4.987943235652156e-05, "loss": 0.7276, "step": 31371 }, { "epoch": 0.03, "learning_rate": 4.987942465145556e-05, "loss": 0.6911, "step": 31372 }, { "epoch": 0.03, "learning_rate": 4.987941694614395e-05, "loss": 1.0849, "step": 31373 }, { "epoch": 0.03, "learning_rate": 4.987940924058675e-05, "loss": 1.21, "step": 31374 }, { "epoch": 0.03, "learning_rate": 4.9879401534783946e-05, "loss": 1.1928, "step": 31375 }, { "epoch": 0.03, "learning_rate": 4.987939382873554e-05, "loss": 0.7919, "step": 31376 }, { "epoch": 0.03, "learning_rate": 4.987938612244154e-05, "loss": 0.9625, "step": 31377 }, { "epoch": 0.03, "learning_rate": 4.987937841590195e-05, "loss": 0.8772, "step": 31378 }, { "epoch": 0.03, "learning_rate": 4.9879370709116746e-05, "loss": 0.8403, "step": 31379 }, { "epoch": 0.03, "learning_rate": 4.987936300208595e-05, "loss": 0.8169, "step": 31380 }, { "epoch": 0.03, "learning_rate": 4.987935529480955e-05, "loss": 1.2746, "step": 31381 }, { "epoch": 0.03, "learning_rate": 4.987934758728756e-05, "loss": 1.1218, "step": 31382 }, { "epoch": 0.03, "learning_rate": 4.987933987951996e-05, "loss": 1.0376, "step": 31383 }, { "epoch": 0.03, "learning_rate": 4.987933217150677e-05, "loss": 0.9946, "step": 31384 }, { "epoch": 0.03, "learning_rate": 4.987932446324799e-05, "loss": 0.7909, "step": 31385 }, { "epoch": 0.03, "learning_rate": 4.98793167547436e-05, "loss": 1.196, "step": 31386 }, { "epoch": 0.03, "learning_rate": 4.987930904599362e-05, "loss": 0.2917, "step": 31387 }, { "epoch": 0.03, "learning_rate": 4.987930133699803e-05, "loss": 1.0851, "step": 31388 }, { "epoch": 0.03, "learning_rate": 4.987929362775685e-05, "loss": 0.9662, "step": 31389 }, { "epoch": 0.03, "learning_rate": 4.9879285918270066e-05, "loss": 1.0555, "step": 31390 }, { "epoch": 0.03, "learning_rate": 4.98792782085377e-05, "loss": 0.989, "step": 31391 }, { "epoch": 0.03, "learning_rate": 4.987927049855972e-05, "loss": 1.2602, "step": 31392 }, { "epoch": 0.03, "learning_rate": 4.9879262788336144e-05, "loss": 0.993, "step": 31393 }, { "epoch": 0.03, "learning_rate": 4.987925507786697e-05, "loss": 0.692, "step": 31394 }, { "epoch": 0.03, "learning_rate": 4.98792473671522e-05, "loss": 1.1419, "step": 31395 }, { "epoch": 0.03, "learning_rate": 4.987923965619184e-05, "loss": 1.0456, "step": 31396 }, { "epoch": 0.03, "learning_rate": 4.987923194498588e-05, "loss": 0.8825, "step": 31397 }, { "epoch": 0.03, "learning_rate": 4.9879224233534316e-05, "loss": 0.8381, "step": 31398 }, { "epoch": 0.03, "learning_rate": 4.987921652183716e-05, "loss": 1.0332, "step": 31399 }, { "epoch": 0.03, "learning_rate": 4.9879208809894405e-05, "loss": 1.0685, "step": 31400 }, { "epoch": 0.03, "learning_rate": 4.9879201097706055e-05, "loss": 0.8966, "step": 31401 }, { "epoch": 0.03, "learning_rate": 4.9879193385272105e-05, "loss": 0.553, "step": 31402 }, { "epoch": 0.03, "learning_rate": 4.987918567259256e-05, "loss": 1.2182, "step": 31403 }, { "epoch": 0.03, "learning_rate": 4.9879177959667417e-05, "loss": 1.6622, "step": 31404 }, { "epoch": 0.03, "learning_rate": 4.987917024649668e-05, "loss": 0.9646, "step": 31405 }, { "epoch": 0.03, "learning_rate": 4.9879162533080346e-05, "loss": 1.0479, "step": 31406 }, { "epoch": 0.03, "learning_rate": 4.987915481941841e-05, "loss": 1.0013, "step": 31407 }, { "epoch": 0.03, "learning_rate": 4.987914710551088e-05, "loss": 1.0383, "step": 31408 }, { "epoch": 0.03, "learning_rate": 4.987913939135775e-05, "loss": 0.9929, "step": 31409 }, { "epoch": 0.03, "learning_rate": 4.987913167695904e-05, "loss": 1.7615, "step": 31410 }, { "epoch": 0.03, "learning_rate": 4.987912396231472e-05, "loss": 1.2479, "step": 31411 }, { "epoch": 0.03, "learning_rate": 4.987911624742481e-05, "loss": 0.9222, "step": 31412 }, { "epoch": 0.03, "learning_rate": 4.987910853228929e-05, "loss": 1.0115, "step": 31413 }, { "epoch": 0.03, "learning_rate": 4.987910081690819e-05, "loss": 1.0255, "step": 31414 }, { "epoch": 0.03, "learning_rate": 4.987909310128148e-05, "loss": 1.0233, "step": 31415 }, { "epoch": 0.03, "learning_rate": 4.987908538540919e-05, "loss": 0.7692, "step": 31416 }, { "epoch": 0.03, "learning_rate": 4.987907766929129e-05, "loss": 0.9952, "step": 31417 }, { "epoch": 0.03, "learning_rate": 4.9879069952927804e-05, "loss": 0.9864, "step": 31418 }, { "epoch": 0.03, "learning_rate": 4.987906223631872e-05, "loss": 1.1943, "step": 31419 }, { "epoch": 0.03, "learning_rate": 4.987905451946404e-05, "loss": 0.9697, "step": 31420 }, { "epoch": 0.03, "learning_rate": 4.987904680236376e-05, "loss": 1.065, "step": 31421 }, { "epoch": 0.03, "learning_rate": 4.9879039085017884e-05, "loss": 1.1168, "step": 31422 }, { "epoch": 0.03, "learning_rate": 4.987903136742642e-05, "loss": 0.9716, "step": 31423 }, { "epoch": 0.03, "learning_rate": 4.9879023649589355e-05, "loss": 1.0178, "step": 31424 }, { "epoch": 0.03, "learning_rate": 4.98790159315067e-05, "loss": 1.0812, "step": 31425 }, { "epoch": 0.03, "learning_rate": 4.987900821317844e-05, "loss": 1.1416, "step": 31426 }, { "epoch": 0.03, "learning_rate": 4.9879000494604596e-05, "loss": 0.9744, "step": 31427 }, { "epoch": 0.03, "learning_rate": 4.987899277578515e-05, "loss": 1.0919, "step": 31428 }, { "epoch": 0.03, "learning_rate": 4.987898505672011e-05, "loss": 0.8761, "step": 31429 }, { "epoch": 0.03, "learning_rate": 4.987897733740948e-05, "loss": 1.2288, "step": 31430 }, { "epoch": 0.03, "learning_rate": 4.987896961785325e-05, "loss": 0.6179, "step": 31431 }, { "epoch": 0.03, "learning_rate": 4.987896189805143e-05, "loss": 1.1744, "step": 31432 }, { "epoch": 0.03, "learning_rate": 4.987895417800401e-05, "loss": 0.9555, "step": 31433 }, { "epoch": 0.03, "learning_rate": 4.9878946457711004e-05, "loss": 1.2816, "step": 31434 }, { "epoch": 0.03, "learning_rate": 4.987893873717239e-05, "loss": 1.0224, "step": 31435 }, { "epoch": 0.03, "learning_rate": 4.9878931016388196e-05, "loss": 0.7763, "step": 31436 }, { "epoch": 0.03, "learning_rate": 4.98789232953584e-05, "loss": 0.625, "step": 31437 }, { "epoch": 0.03, "learning_rate": 4.9878915574083006e-05, "loss": 1.1646, "step": 31438 }, { "epoch": 0.03, "learning_rate": 4.9878907852562026e-05, "loss": 0.7406, "step": 31439 }, { "epoch": 0.03, "learning_rate": 4.987890013079545e-05, "loss": 1.1652, "step": 31440 }, { "epoch": 0.03, "learning_rate": 4.987889240878327e-05, "loss": 1.2209, "step": 31441 }, { "epoch": 0.03, "learning_rate": 4.987888468652551e-05, "loss": 1.4779, "step": 31442 }, { "epoch": 0.03, "learning_rate": 4.9878876964022145e-05, "loss": 0.99, "step": 31443 }, { "epoch": 0.03, "learning_rate": 4.98788692412732e-05, "loss": 0.9853, "step": 31444 }, { "epoch": 0.03, "learning_rate": 4.987886151827865e-05, "loss": 2.0161, "step": 31445 }, { "epoch": 0.03, "learning_rate": 4.9878853795038513e-05, "loss": 1.0006, "step": 31446 }, { "epoch": 0.03, "learning_rate": 4.9878846071552776e-05, "loss": 0.882, "step": 31447 }, { "epoch": 0.03, "learning_rate": 4.987883834782145e-05, "loss": 1.256, "step": 31448 }, { "epoch": 0.03, "learning_rate": 4.987883062384453e-05, "loss": 0.8581, "step": 31449 }, { "epoch": 0.03, "learning_rate": 4.987882289962201e-05, "loss": 1.1494, "step": 31450 }, { "epoch": 0.03, "learning_rate": 4.9878815175153906e-05, "loss": 0.8532, "step": 31451 }, { "epoch": 0.03, "learning_rate": 4.98788074504402e-05, "loss": 0.9982, "step": 31452 }, { "epoch": 0.03, "learning_rate": 4.9878799725480905e-05, "loss": 0.8695, "step": 31453 }, { "epoch": 0.03, "learning_rate": 4.987879200027602e-05, "loss": 0.8604, "step": 31454 }, { "epoch": 0.03, "learning_rate": 4.987878427482554e-05, "loss": 0.8363, "step": 31455 }, { "epoch": 0.03, "learning_rate": 4.987877654912947e-05, "loss": 0.8055, "step": 31456 }, { "epoch": 0.03, "learning_rate": 4.9878768823187804e-05, "loss": 0.9307, "step": 31457 }, { "epoch": 0.03, "learning_rate": 4.987876109700055e-05, "loss": 0.9555, "step": 31458 }, { "epoch": 0.03, "learning_rate": 4.98787533705677e-05, "loss": 1.486, "step": 31459 }, { "epoch": 0.03, "learning_rate": 4.987874564388925e-05, "loss": 1.9775, "step": 31460 }, { "epoch": 0.03, "learning_rate": 4.9878737916965216e-05, "loss": 1.8424, "step": 31461 }, { "epoch": 0.03, "learning_rate": 4.987873018979558e-05, "loss": 1.2312, "step": 31462 }, { "epoch": 0.03, "learning_rate": 4.9878722462380365e-05, "loss": 1.0083, "step": 31463 }, { "epoch": 0.03, "learning_rate": 4.9878714734719555e-05, "loss": 0.8371, "step": 31464 }, { "epoch": 0.03, "learning_rate": 4.987870700681315e-05, "loss": 0.8216, "step": 31465 }, { "epoch": 0.03, "learning_rate": 4.987869927866116e-05, "loss": 0.8231, "step": 31466 }, { "epoch": 0.03, "learning_rate": 4.987869155026356e-05, "loss": 1.2153, "step": 31467 }, { "epoch": 0.03, "learning_rate": 4.987868382162038e-05, "loss": 1.1343, "step": 31468 }, { "epoch": 0.03, "learning_rate": 4.987867609273161e-05, "loss": 1.2053, "step": 31469 }, { "epoch": 0.03, "learning_rate": 4.987866836359725e-05, "loss": 1.0998, "step": 31470 }, { "epoch": 0.03, "learning_rate": 4.9878660634217286e-05, "loss": 1.0523, "step": 31471 }, { "epoch": 0.03, "learning_rate": 4.987865290459174e-05, "loss": 1.1063, "step": 31472 }, { "epoch": 0.03, "learning_rate": 4.987864517472059e-05, "loss": 1.0122, "step": 31473 }, { "epoch": 0.03, "learning_rate": 4.987863744460387e-05, "loss": 1.0835, "step": 31474 }, { "epoch": 0.03, "learning_rate": 4.9878629714241545e-05, "loss": 1.0043, "step": 31475 }, { "epoch": 0.03, "learning_rate": 4.987862198363363e-05, "loss": 0.9682, "step": 31476 }, { "epoch": 0.03, "learning_rate": 4.987861425278012e-05, "loss": 1.1624, "step": 31477 }, { "epoch": 0.03, "learning_rate": 4.987860652168102e-05, "loss": 1.1127, "step": 31478 }, { "epoch": 0.03, "learning_rate": 4.987859879033634e-05, "loss": 1.5615, "step": 31479 }, { "epoch": 0.03, "learning_rate": 4.987859105874606e-05, "loss": 1.3652, "step": 31480 }, { "epoch": 0.03, "learning_rate": 4.987858332691019e-05, "loss": 1.4139, "step": 31481 }, { "epoch": 0.03, "learning_rate": 4.987857559482873e-05, "loss": 1.3209, "step": 31482 }, { "epoch": 0.03, "learning_rate": 4.9878567862501677e-05, "loss": 1.0365, "step": 31483 }, { "epoch": 0.03, "learning_rate": 4.987856012992904e-05, "loss": 0.9814, "step": 31484 }, { "epoch": 0.03, "learning_rate": 4.98785523971108e-05, "loss": 1.0244, "step": 31485 }, { "epoch": 0.03, "learning_rate": 4.987854466404698e-05, "loss": 0.8571, "step": 31486 }, { "epoch": 0.03, "learning_rate": 4.987853693073756e-05, "loss": 1.2056, "step": 31487 }, { "epoch": 0.03, "learning_rate": 4.987852919718256e-05, "loss": 0.946, "step": 31488 }, { "epoch": 0.03, "learning_rate": 4.987852146338197e-05, "loss": 0.8998, "step": 31489 }, { "epoch": 0.03, "learning_rate": 4.9878513729335784e-05, "loss": 1.0761, "step": 31490 }, { "epoch": 0.03, "learning_rate": 4.987850599504401e-05, "loss": 0.4236, "step": 31491 }, { "epoch": 0.03, "learning_rate": 4.987849826050664e-05, "loss": 1.1291, "step": 31492 }, { "epoch": 0.03, "learning_rate": 4.9878490525723683e-05, "loss": 0.8313, "step": 31493 }, { "epoch": 0.03, "learning_rate": 4.9878482790695144e-05, "loss": 0.9125, "step": 31494 }, { "epoch": 0.03, "learning_rate": 4.9878475055421005e-05, "loss": 1.2935, "step": 31495 }, { "epoch": 0.03, "learning_rate": 4.987846731990128e-05, "loss": 1.1113, "step": 31496 }, { "epoch": 0.03, "learning_rate": 4.987845958413596e-05, "loss": 1.0944, "step": 31497 }, { "epoch": 0.03, "learning_rate": 4.9878451848125064e-05, "loss": 1.0799, "step": 31498 }, { "epoch": 0.03, "learning_rate": 4.987844411186856e-05, "loss": 0.7677, "step": 31499 }, { "epoch": 0.03, "learning_rate": 4.9878436375366484e-05, "loss": 0.7831, "step": 31500 }, { "epoch": 0.03, "eval_loss": 1.06670081615448, "eval_runtime": 127.8487, "eval_samples_per_second": 10.833, "eval_steps_per_second": 5.42, "step": 31500 }, { "epoch": 0.03, "learning_rate": 4.9878428638618813e-05, "loss": 0.9292, "step": 31501 }, { "epoch": 0.03, "learning_rate": 4.9878420901625544e-05, "loss": 1.1426, "step": 31502 }, { "epoch": 0.03, "learning_rate": 4.9878413164386695e-05, "loss": 0.9489, "step": 31503 }, { "epoch": 0.03, "learning_rate": 4.9878405426902254e-05, "loss": 0.8089, "step": 31504 }, { "epoch": 0.03, "learning_rate": 4.987839768917222e-05, "loss": 1.0579, "step": 31505 }, { "epoch": 0.03, "learning_rate": 4.9878389951196604e-05, "loss": 0.9912, "step": 31506 }, { "epoch": 0.03, "learning_rate": 4.9878382212975393e-05, "loss": 0.999, "step": 31507 }, { "epoch": 0.03, "learning_rate": 4.98783744745086e-05, "loss": 1.3683, "step": 31508 }, { "epoch": 0.03, "learning_rate": 4.987836673579621e-05, "loss": 1.0218, "step": 31509 }, { "epoch": 0.03, "learning_rate": 4.987835899683824e-05, "loss": 0.7452, "step": 31510 }, { "epoch": 0.03, "learning_rate": 4.987835125763468e-05, "loss": 0.9736, "step": 31511 }, { "epoch": 0.03, "learning_rate": 4.987834351818552e-05, "loss": 0.952, "step": 31512 }, { "epoch": 0.03, "learning_rate": 4.987833577849078e-05, "loss": 1.2411, "step": 31513 }, { "epoch": 0.03, "learning_rate": 4.987832803855045e-05, "loss": 1.0413, "step": 31514 }, { "epoch": 0.03, "learning_rate": 4.987832029836453e-05, "loss": 0.9205, "step": 31515 }, { "epoch": 0.03, "learning_rate": 4.9878312557933026e-05, "loss": 0.8793, "step": 31516 }, { "epoch": 0.03, "learning_rate": 4.987830481725594e-05, "loss": 0.8732, "step": 31517 }, { "epoch": 0.03, "learning_rate": 4.987829707633326e-05, "loss": 0.8929, "step": 31518 }, { "epoch": 0.03, "learning_rate": 4.9878289335164984e-05, "loss": 0.8626, "step": 31519 }, { "epoch": 0.03, "learning_rate": 4.987828159375113e-05, "loss": 1.1934, "step": 31520 }, { "epoch": 0.03, "learning_rate": 4.9878273852091685e-05, "loss": 0.9111, "step": 31521 }, { "epoch": 0.03, "learning_rate": 4.987826611018665e-05, "loss": 0.7997, "step": 31522 }, { "epoch": 0.03, "learning_rate": 4.9878258368036024e-05, "loss": 1.0889, "step": 31523 }, { "epoch": 0.03, "learning_rate": 4.9878250625639815e-05, "loss": 1.0304, "step": 31524 }, { "epoch": 0.03, "learning_rate": 4.9878242882998015e-05, "loss": 1.1372, "step": 31525 }, { "epoch": 0.03, "learning_rate": 4.9878235140110636e-05, "loss": 1.0453, "step": 31526 }, { "epoch": 0.03, "learning_rate": 4.9878227396977664e-05, "loss": 1.1428, "step": 31527 }, { "epoch": 0.03, "learning_rate": 4.98782196535991e-05, "loss": 0.6646, "step": 31528 }, { "epoch": 0.03, "learning_rate": 4.987821190997496e-05, "loss": 0.9391, "step": 31529 }, { "epoch": 0.03, "learning_rate": 4.9878204166105225e-05, "loss": 0.95, "step": 31530 }, { "epoch": 0.03, "learning_rate": 4.9878196421989905e-05, "loss": 1.034, "step": 31531 }, { "epoch": 0.03, "learning_rate": 4.9878188677628994e-05, "loss": 1.0388, "step": 31532 }, { "epoch": 0.03, "learning_rate": 4.98781809330225e-05, "loss": 1.0364, "step": 31533 }, { "epoch": 0.03, "learning_rate": 4.987817318817042e-05, "loss": 1.1672, "step": 31534 }, { "epoch": 0.03, "learning_rate": 4.987816544307275e-05, "loss": 0.891, "step": 31535 }, { "epoch": 0.03, "learning_rate": 4.9878157697729494e-05, "loss": 1.0793, "step": 31536 }, { "epoch": 0.03, "learning_rate": 4.987814995214065e-05, "loss": 0.8345, "step": 31537 }, { "epoch": 0.03, "learning_rate": 4.987814220630622e-05, "loss": 0.7572, "step": 31538 }, { "epoch": 0.03, "learning_rate": 4.987813446022621e-05, "loss": 0.8625, "step": 31539 }, { "epoch": 0.03, "learning_rate": 4.987812671390061e-05, "loss": 1.0572, "step": 31540 }, { "epoch": 0.03, "learning_rate": 4.9878118967329415e-05, "loss": 1.2636, "step": 31541 }, { "epoch": 0.03, "learning_rate": 4.9878111220512644e-05, "loss": 1.0138, "step": 31542 }, { "epoch": 0.03, "learning_rate": 4.987810347345028e-05, "loss": 0.7418, "step": 31543 }, { "epoch": 0.03, "learning_rate": 4.987809572614234e-05, "loss": 0.7776, "step": 31544 }, { "epoch": 0.03, "learning_rate": 4.98780879785888e-05, "loss": 1.0206, "step": 31545 }, { "epoch": 0.03, "learning_rate": 4.987808023078968e-05, "loss": 1.1416, "step": 31546 }, { "epoch": 0.03, "learning_rate": 4.987807248274498e-05, "loss": 1.0173, "step": 31547 }, { "epoch": 0.03, "learning_rate": 4.987806473445469e-05, "loss": 0.9082, "step": 31548 }, { "epoch": 0.03, "learning_rate": 4.987805698591881e-05, "loss": 0.9434, "step": 31549 }, { "epoch": 0.03, "learning_rate": 4.987804923713735e-05, "loss": 1.1064, "step": 31550 }, { "epoch": 0.03, "learning_rate": 4.98780414881103e-05, "loss": 0.909, "step": 31551 }, { "epoch": 0.03, "learning_rate": 4.987803373883767e-05, "loss": 0.9012, "step": 31552 }, { "epoch": 0.03, "learning_rate": 4.9878025989319454e-05, "loss": 1.1068, "step": 31553 }, { "epoch": 0.03, "learning_rate": 4.9878018239555644e-05, "loss": 0.9233, "step": 31554 }, { "epoch": 0.03, "learning_rate": 4.9878010489546256e-05, "loss": 0.9125, "step": 31555 }, { "epoch": 0.03, "learning_rate": 4.987800273929128e-05, "loss": 1.7135, "step": 31556 }, { "epoch": 0.03, "learning_rate": 4.9877994988790724e-05, "loss": 0.9676, "step": 31557 }, { "epoch": 0.03, "learning_rate": 4.987798723804458e-05, "loss": 0.9832, "step": 31558 }, { "epoch": 0.03, "learning_rate": 4.987797948705285e-05, "loss": 0.5851, "step": 31559 }, { "epoch": 0.03, "learning_rate": 4.987797173581553e-05, "loss": 0.4362, "step": 31560 }, { "epoch": 0.03, "learning_rate": 4.987796398433263e-05, "loss": 0.9118, "step": 31561 }, { "epoch": 0.03, "learning_rate": 4.987795623260415e-05, "loss": 1.1061, "step": 31562 }, { "epoch": 0.03, "learning_rate": 4.987794848063008e-05, "loss": 1.1071, "step": 31563 }, { "epoch": 0.03, "learning_rate": 4.987794072841043e-05, "loss": 0.8589, "step": 31564 }, { "epoch": 0.03, "learning_rate": 4.987793297594519e-05, "loss": 0.8587, "step": 31565 }, { "epoch": 0.03, "learning_rate": 4.987792522323437e-05, "loss": 0.8679, "step": 31566 }, { "epoch": 0.03, "learning_rate": 4.9877917470277956e-05, "loss": 1.131, "step": 31567 }, { "epoch": 0.03, "learning_rate": 4.9877909717075966e-05, "loss": 0.6241, "step": 31568 }, { "epoch": 0.03, "learning_rate": 4.9877901963628396e-05, "loss": 0.8259, "step": 31569 }, { "epoch": 0.03, "learning_rate": 4.987789420993523e-05, "loss": 0.9802, "step": 31570 }, { "epoch": 0.03, "learning_rate": 4.987788645599649e-05, "loss": 1.5742, "step": 31571 }, { "epoch": 0.03, "learning_rate": 4.9877878701812156e-05, "loss": 1.6131, "step": 31572 }, { "epoch": 0.03, "learning_rate": 4.9877870947382245e-05, "loss": 1.584, "step": 31573 }, { "epoch": 0.03, "learning_rate": 4.9877863192706756e-05, "loss": 1.5155, "step": 31574 }, { "epoch": 0.03, "learning_rate": 4.987785543778567e-05, "loss": 1.4825, "step": 31575 }, { "epoch": 0.03, "learning_rate": 4.987784768261901e-05, "loss": 1.5186, "step": 31576 }, { "epoch": 0.03, "learning_rate": 4.987783992720676e-05, "loss": 1.4775, "step": 31577 }, { "epoch": 0.03, "learning_rate": 4.987783217154893e-05, "loss": 1.4539, "step": 31578 }, { "epoch": 0.03, "learning_rate": 4.9877824415645516e-05, "loss": 1.4228, "step": 31579 }, { "epoch": 0.03, "learning_rate": 4.9877816659496515e-05, "loss": 1.4172, "step": 31580 }, { "epoch": 0.03, "learning_rate": 4.987780890310194e-05, "loss": 1.3869, "step": 31581 }, { "epoch": 0.03, "learning_rate": 4.987780114646177e-05, "loss": 1.3251, "step": 31582 }, { "epoch": 0.03, "learning_rate": 4.9877793389576026e-05, "loss": 1.3677, "step": 31583 }, { "epoch": 0.03, "learning_rate": 4.9877785632444696e-05, "loss": 1.3353, "step": 31584 }, { "epoch": 0.03, "learning_rate": 4.987777787506778e-05, "loss": 1.2822, "step": 31585 }, { "epoch": 0.03, "learning_rate": 4.987777011744529e-05, "loss": 1.2558, "step": 31586 }, { "epoch": 0.03, "learning_rate": 4.9877762359577204e-05, "loss": 1.1929, "step": 31587 }, { "epoch": 0.03, "learning_rate": 4.987775460146355e-05, "loss": 1.2535, "step": 31588 }, { "epoch": 0.03, "learning_rate": 4.98777468431043e-05, "loss": 1.2171, "step": 31589 }, { "epoch": 0.03, "learning_rate": 4.987773908449948e-05, "loss": 1.1873, "step": 31590 }, { "epoch": 0.03, "learning_rate": 4.9877731325649064e-05, "loss": 1.1546, "step": 31591 }, { "epoch": 0.03, "learning_rate": 4.987772356655307e-05, "loss": 1.1065, "step": 31592 }, { "epoch": 0.03, "learning_rate": 4.98777158072115e-05, "loss": 1.1368, "step": 31593 }, { "epoch": 0.03, "learning_rate": 4.987770804762434e-05, "loss": 1.1545, "step": 31594 }, { "epoch": 0.03, "learning_rate": 4.98777002877916e-05, "loss": 1.0863, "step": 31595 }, { "epoch": 0.03, "learning_rate": 4.987769252771328e-05, "loss": 1.0981, "step": 31596 }, { "epoch": 0.03, "learning_rate": 4.9877684767389375e-05, "loss": 1.0569, "step": 31597 }, { "epoch": 0.03, "learning_rate": 4.987767700681989e-05, "loss": 1.0647, "step": 31598 }, { "epoch": 0.03, "learning_rate": 4.987766924600482e-05, "loss": 1.1359, "step": 31599 }, { "epoch": 0.03, "learning_rate": 4.987766148494417e-05, "loss": 1.1418, "step": 31600 }, { "epoch": 0.03, "learning_rate": 4.987765372363794e-05, "loss": 1.0434, "step": 31601 }, { "epoch": 0.03, "learning_rate": 4.9877645962086125e-05, "loss": 1.2659, "step": 31602 }, { "epoch": 0.03, "learning_rate": 4.9877638200288735e-05, "loss": 1.1641, "step": 31603 }, { "epoch": 0.03, "learning_rate": 4.987763043824576e-05, "loss": 0.8731, "step": 31604 }, { "epoch": 0.03, "learning_rate": 4.98776226759572e-05, "loss": 1.0412, "step": 31605 }, { "epoch": 0.03, "learning_rate": 4.987761491342307e-05, "loss": 1.1791, "step": 31606 }, { "epoch": 0.03, "learning_rate": 4.9877607150643343e-05, "loss": 1.1265, "step": 31607 }, { "epoch": 0.03, "learning_rate": 4.987759938761804e-05, "loss": 0.8375, "step": 31608 }, { "epoch": 0.03, "learning_rate": 4.987759162434716e-05, "loss": 1.0053, "step": 31609 }, { "epoch": 0.03, "learning_rate": 4.987758386083069e-05, "loss": 0.7843, "step": 31610 }, { "epoch": 0.03, "learning_rate": 4.9877576097068654e-05, "loss": 1.1921, "step": 31611 }, { "epoch": 0.03, "learning_rate": 4.987756833306102e-05, "loss": 0.929, "step": 31612 }, { "epoch": 0.03, "learning_rate": 4.987756056880782e-05, "loss": 0.9057, "step": 31613 }, { "epoch": 0.03, "learning_rate": 4.987755280430903e-05, "loss": 1.087, "step": 31614 }, { "epoch": 0.03, "learning_rate": 4.9877545039564666e-05, "loss": 0.7259, "step": 31615 }, { "epoch": 0.03, "learning_rate": 4.9877537274574714e-05, "loss": 1.0172, "step": 31616 }, { "epoch": 0.03, "learning_rate": 4.9877529509339184e-05, "loss": 1.1857, "step": 31617 }, { "epoch": 0.03, "learning_rate": 4.9877521743858075e-05, "loss": 0.8986, "step": 31618 }, { "epoch": 0.03, "learning_rate": 4.987751397813139e-05, "loss": 1.5541, "step": 31619 }, { "epoch": 0.03, "learning_rate": 4.9877506212159115e-05, "loss": 0.7627, "step": 31620 }, { "epoch": 0.03, "learning_rate": 4.9877498445941264e-05, "loss": 0.8395, "step": 31621 }, { "epoch": 0.03, "learning_rate": 4.9877490679477834e-05, "loss": 1.1617, "step": 31622 }, { "epoch": 0.03, "learning_rate": 4.987748291276882e-05, "loss": 1.142, "step": 31623 }, { "epoch": 0.03, "learning_rate": 4.987747514581423e-05, "loss": 1.1582, "step": 31624 }, { "epoch": 0.03, "learning_rate": 4.987746737861406e-05, "loss": 1.1773, "step": 31625 }, { "epoch": 0.03, "learning_rate": 4.987745961116831e-05, "loss": 0.9595, "step": 31626 }, { "epoch": 0.03, "learning_rate": 4.987745184347697e-05, "loss": 0.7063, "step": 31627 }, { "epoch": 0.03, "learning_rate": 4.987744407554007e-05, "loss": 0.2207, "step": 31628 }, { "epoch": 0.03, "learning_rate": 4.987743630735757e-05, "loss": 0.3346, "step": 31629 }, { "epoch": 0.03, "learning_rate": 4.9877428538929505e-05, "loss": 1.0273, "step": 31630 }, { "epoch": 0.03, "learning_rate": 4.9877420770255854e-05, "loss": 1.1559, "step": 31631 }, { "epoch": 0.03, "learning_rate": 4.9877413001336625e-05, "loss": 1.4585, "step": 31632 }, { "epoch": 0.03, "learning_rate": 4.987740523217182e-05, "loss": 0.9491, "step": 31633 }, { "epoch": 0.03, "learning_rate": 4.987739746276143e-05, "loss": 0.9616, "step": 31634 }, { "epoch": 0.03, "learning_rate": 4.987738969310547e-05, "loss": 1.0613, "step": 31635 }, { "epoch": 0.03, "learning_rate": 4.987738192320392e-05, "loss": 1.8023, "step": 31636 }, { "epoch": 0.03, "learning_rate": 4.98773741530568e-05, "loss": 1.8394, "step": 31637 }, { "epoch": 0.03, "learning_rate": 4.987736638266409e-05, "loss": 0.8982, "step": 31638 }, { "epoch": 0.03, "learning_rate": 4.987735861202581e-05, "loss": 0.889, "step": 31639 }, { "epoch": 0.03, "learning_rate": 4.987735084114194e-05, "loss": 0.8388, "step": 31640 }, { "epoch": 0.03, "learning_rate": 4.9877343070012506e-05, "loss": 1.3606, "step": 31641 }, { "epoch": 0.03, "learning_rate": 4.987733529863749e-05, "loss": 1.4235, "step": 31642 }, { "epoch": 0.03, "learning_rate": 4.987732752701689e-05, "loss": 1.2907, "step": 31643 }, { "epoch": 0.03, "learning_rate": 4.987731975515072e-05, "loss": 1.2511, "step": 31644 }, { "epoch": 0.03, "learning_rate": 4.987731198303897e-05, "loss": 0.6254, "step": 31645 }, { "epoch": 0.03, "learning_rate": 4.987730421068163e-05, "loss": 0.7027, "step": 31646 }, { "epoch": 0.03, "learning_rate": 4.9877296438078725e-05, "loss": 0.3254, "step": 31647 }, { "epoch": 0.03, "learning_rate": 4.987728866523023e-05, "loss": 0.6849, "step": 31648 }, { "epoch": 0.03, "learning_rate": 4.987728089213617e-05, "loss": 1.3699, "step": 31649 }, { "epoch": 0.03, "learning_rate": 4.987727311879652e-05, "loss": 1.284, "step": 31650 }, { "epoch": 0.03, "learning_rate": 4.98772653452113e-05, "loss": 1.3115, "step": 31651 }, { "epoch": 0.03, "learning_rate": 4.98772575713805e-05, "loss": 1.1954, "step": 31652 }, { "epoch": 0.03, "learning_rate": 4.9877249797304115e-05, "loss": 1.2254, "step": 31653 }, { "epoch": 0.03, "learning_rate": 4.9877242022982164e-05, "loss": 1.3375, "step": 31654 }, { "epoch": 0.03, "learning_rate": 4.987723424841463e-05, "loss": 1.0722, "step": 31655 }, { "epoch": 0.03, "learning_rate": 4.9877226473601513e-05, "loss": 0.7539, "step": 31656 }, { "epoch": 0.03, "learning_rate": 4.987721869854283e-05, "loss": 0.9471, "step": 31657 }, { "epoch": 0.03, "learning_rate": 4.987721092323856e-05, "loss": 1.0756, "step": 31658 }, { "epoch": 0.03, "learning_rate": 4.987720314768872e-05, "loss": 0.7231, "step": 31659 }, { "epoch": 0.03, "learning_rate": 4.9877195371893296e-05, "loss": 0.9221, "step": 31660 }, { "epoch": 0.03, "learning_rate": 4.9877187595852295e-05, "loss": 0.6837, "step": 31661 }, { "epoch": 0.03, "learning_rate": 4.987717981956572e-05, "loss": 0.9976, "step": 31662 }, { "epoch": 0.03, "learning_rate": 4.987717204303357e-05, "loss": 1.1344, "step": 31663 }, { "epoch": 0.03, "learning_rate": 4.987716426625584e-05, "loss": 1.1505, "step": 31664 }, { "epoch": 0.03, "learning_rate": 4.987715648923253e-05, "loss": 0.9606, "step": 31665 }, { "epoch": 0.03, "learning_rate": 4.987714871196365e-05, "loss": 0.9186, "step": 31666 }, { "epoch": 0.03, "learning_rate": 4.987714093444919e-05, "loss": 0.9501, "step": 31667 }, { "epoch": 0.03, "learning_rate": 4.9877133156689156e-05, "loss": 1.0612, "step": 31668 }, { "epoch": 0.03, "learning_rate": 4.987712537868354e-05, "loss": 1.0047, "step": 31669 }, { "epoch": 0.03, "learning_rate": 4.987711760043235e-05, "loss": 1.2236, "step": 31670 }, { "epoch": 0.03, "learning_rate": 4.9877109821935587e-05, "loss": 0.9869, "step": 31671 }, { "epoch": 0.03, "learning_rate": 4.987710204319325e-05, "loss": 1.1288, "step": 31672 }, { "epoch": 0.03, "learning_rate": 4.9877094264205325e-05, "loss": 1.0172, "step": 31673 }, { "epoch": 0.03, "learning_rate": 4.987708648497183e-05, "loss": 1.0775, "step": 31674 }, { "epoch": 0.03, "learning_rate": 4.9877078705492756e-05, "loss": 0.9275, "step": 31675 }, { "epoch": 0.03, "learning_rate": 4.987707092576811e-05, "loss": 1.2585, "step": 31676 }, { "epoch": 0.03, "learning_rate": 4.9877063145797887e-05, "loss": 0.6953, "step": 31677 }, { "epoch": 0.03, "learning_rate": 4.987705536558209e-05, "loss": 0.9898, "step": 31678 }, { "epoch": 0.03, "learning_rate": 4.9877047585120716e-05, "loss": 0.783, "step": 31679 }, { "epoch": 0.03, "learning_rate": 4.987703980441376e-05, "loss": 0.8523, "step": 31680 }, { "epoch": 0.03, "learning_rate": 4.987703202346124e-05, "loss": 1.1208, "step": 31681 }, { "epoch": 0.03, "learning_rate": 4.9877024242263135e-05, "loss": 1.3804, "step": 31682 }, { "epoch": 0.03, "learning_rate": 4.987701646081946e-05, "loss": 1.1707, "step": 31683 }, { "epoch": 0.03, "learning_rate": 4.987700867913021e-05, "loss": 1.0394, "step": 31684 }, { "epoch": 0.03, "learning_rate": 4.987700089719538e-05, "loss": 1.1449, "step": 31685 }, { "epoch": 0.03, "learning_rate": 4.987699311501498e-05, "loss": 0.4049, "step": 31686 }, { "epoch": 0.03, "learning_rate": 4.9876985332589e-05, "loss": 0.3557, "step": 31687 }, { "epoch": 0.03, "learning_rate": 4.9876977549917446e-05, "loss": 0.4335, "step": 31688 }, { "epoch": 0.03, "learning_rate": 4.987696976700031e-05, "loss": 0.3495, "step": 31689 }, { "epoch": 0.03, "learning_rate": 4.987696198383761e-05, "loss": 0.2413, "step": 31690 }, { "epoch": 0.03, "learning_rate": 4.987695420042934e-05, "loss": 0.4409, "step": 31691 }, { "epoch": 0.03, "learning_rate": 4.987694641677548e-05, "loss": 0.5558, "step": 31692 }, { "epoch": 0.03, "learning_rate": 4.9876938632876056e-05, "loss": 1.0945, "step": 31693 }, { "epoch": 0.03, "learning_rate": 4.987693084873105e-05, "loss": 0.9483, "step": 31694 }, { "epoch": 0.03, "learning_rate": 4.987692306434047e-05, "loss": 1.0021, "step": 31695 }, { "epoch": 0.03, "learning_rate": 4.9876915279704324e-05, "loss": 1.0685, "step": 31696 }, { "epoch": 0.03, "learning_rate": 4.98769074948226e-05, "loss": 1.1254, "step": 31697 }, { "epoch": 0.03, "learning_rate": 4.98768997096953e-05, "loss": 0.8902, "step": 31698 }, { "epoch": 0.03, "learning_rate": 4.987689192432242e-05, "loss": 0.9064, "step": 31699 }, { "epoch": 0.03, "learning_rate": 4.987688413870397e-05, "loss": 0.8305, "step": 31700 }, { "epoch": 0.03, "learning_rate": 4.9876876352839954e-05, "loss": 0.8932, "step": 31701 }, { "epoch": 0.03, "learning_rate": 4.987686856673035e-05, "loss": 0.9019, "step": 31702 }, { "epoch": 0.03, "learning_rate": 4.987686078037518e-05, "loss": 0.3333, "step": 31703 }, { "epoch": 0.03, "learning_rate": 4.987685299377444e-05, "loss": 0.8579, "step": 31704 }, { "epoch": 0.03, "learning_rate": 4.9876845206928116e-05, "loss": 0.3433, "step": 31705 }, { "epoch": 0.03, "learning_rate": 4.987683741983623e-05, "loss": 0.957, "step": 31706 }, { "epoch": 0.03, "learning_rate": 4.987682963249876e-05, "loss": 0.6483, "step": 31707 }, { "epoch": 0.03, "learning_rate": 4.9876821844915725e-05, "loss": 1.2886, "step": 31708 }, { "epoch": 0.03, "learning_rate": 4.987681405708711e-05, "loss": 0.847, "step": 31709 }, { "epoch": 0.03, "learning_rate": 4.987680626901292e-05, "loss": 0.7357, "step": 31710 }, { "epoch": 0.03, "learning_rate": 4.987679848069317e-05, "loss": 0.816, "step": 31711 }, { "epoch": 0.03, "learning_rate": 4.987679069212783e-05, "loss": 0.8368, "step": 31712 }, { "epoch": 0.03, "learning_rate": 4.987678290331692e-05, "loss": 0.7033, "step": 31713 }, { "epoch": 0.03, "learning_rate": 4.987677511426044e-05, "loss": 1.0977, "step": 31714 }, { "epoch": 0.03, "learning_rate": 4.987676732495839e-05, "loss": 1.2245, "step": 31715 }, { "epoch": 0.03, "learning_rate": 4.9876759535410764e-05, "loss": 1.4229, "step": 31716 }, { "epoch": 0.03, "learning_rate": 4.9876751745617566e-05, "loss": 1.1627, "step": 31717 }, { "epoch": 0.03, "learning_rate": 4.98767439555788e-05, "loss": 1.2098, "step": 31718 }, { "epoch": 0.03, "learning_rate": 4.987673616529445e-05, "loss": 1.2843, "step": 31719 }, { "epoch": 0.03, "learning_rate": 4.9876728374764537e-05, "loss": 1.4804, "step": 31720 }, { "epoch": 0.03, "learning_rate": 4.9876720583989045e-05, "loss": 0.7682, "step": 31721 }, { "epoch": 0.03, "learning_rate": 4.987671279296798e-05, "loss": 0.641, "step": 31722 }, { "epoch": 0.03, "learning_rate": 4.9876705001701354e-05, "loss": 0.9784, "step": 31723 }, { "epoch": 0.03, "learning_rate": 4.987669721018915e-05, "loss": 0.9284, "step": 31724 }, { "epoch": 0.03, "learning_rate": 4.987668941843137e-05, "loss": 0.6791, "step": 31725 }, { "epoch": 0.03, "learning_rate": 4.987668162642801e-05, "loss": 1.0474, "step": 31726 }, { "epoch": 0.03, "learning_rate": 4.987667383417909e-05, "loss": 1.1024, "step": 31727 }, { "epoch": 0.03, "learning_rate": 4.9876666041684595e-05, "loss": 0.7601, "step": 31728 }, { "epoch": 0.03, "learning_rate": 4.987665824894453e-05, "loss": 0.3253, "step": 31729 }, { "epoch": 0.03, "learning_rate": 4.987665045595889e-05, "loss": 1.1772, "step": 31730 }, { "epoch": 0.03, "learning_rate": 4.9876642662727676e-05, "loss": 0.858, "step": 31731 }, { "epoch": 0.03, "learning_rate": 4.98766348692509e-05, "loss": 0.9403, "step": 31732 }, { "epoch": 0.03, "learning_rate": 4.987662707552854e-05, "loss": 0.9707, "step": 31733 }, { "epoch": 0.03, "learning_rate": 4.9876619281560624e-05, "loss": 0.9898, "step": 31734 }, { "epoch": 0.03, "learning_rate": 4.987661148734713e-05, "loss": 0.3916, "step": 31735 }, { "epoch": 0.03, "learning_rate": 4.987660369288805e-05, "loss": 0.2784, "step": 31736 }, { "epoch": 0.03, "learning_rate": 4.9876595898183406e-05, "loss": 0.2993, "step": 31737 }, { "epoch": 0.03, "learning_rate": 4.98765881032332e-05, "loss": 0.2481, "step": 31738 }, { "epoch": 0.03, "learning_rate": 4.987658030803742e-05, "loss": 0.1896, "step": 31739 }, { "epoch": 0.03, "learning_rate": 4.987657251259606e-05, "loss": 0.2071, "step": 31740 }, { "epoch": 0.03, "learning_rate": 4.987656471690914e-05, "loss": 0.2091, "step": 31741 }, { "epoch": 0.03, "learning_rate": 4.987655692097665e-05, "loss": 0.4274, "step": 31742 }, { "epoch": 0.03, "learning_rate": 4.987654912479858e-05, "loss": 1.7362, "step": 31743 }, { "epoch": 0.03, "learning_rate": 4.9876541328374936e-05, "loss": 1.0943, "step": 31744 }, { "epoch": 0.03, "learning_rate": 4.9876533531705735e-05, "loss": 0.8546, "step": 31745 }, { "epoch": 0.03, "learning_rate": 4.9876525734790956e-05, "loss": 0.8424, "step": 31746 }, { "epoch": 0.03, "learning_rate": 4.9876517937630604e-05, "loss": 0.4817, "step": 31747 }, { "epoch": 0.03, "learning_rate": 4.987651014022469e-05, "loss": 1.0888, "step": 31748 }, { "epoch": 0.03, "learning_rate": 4.9876502342573193e-05, "loss": 1.0747, "step": 31749 }, { "epoch": 0.03, "learning_rate": 4.9876494544676134e-05, "loss": 1.1098, "step": 31750 }, { "epoch": 0.03, "learning_rate": 4.98764867465335e-05, "loss": 1.1667, "step": 31751 }, { "epoch": 0.03, "learning_rate": 4.98764789481453e-05, "loss": 0.5702, "step": 31752 }, { "epoch": 0.03, "learning_rate": 4.987647114951153e-05, "loss": 0.4331, "step": 31753 }, { "epoch": 0.03, "learning_rate": 4.987646335063219e-05, "loss": 0.8206, "step": 31754 }, { "epoch": 0.03, "learning_rate": 4.9876455551507285e-05, "loss": 0.9976, "step": 31755 }, { "epoch": 0.03, "learning_rate": 4.98764477521368e-05, "loss": 1.0751, "step": 31756 }, { "epoch": 0.03, "learning_rate": 4.9876439952520746e-05, "loss": 0.8512, "step": 31757 }, { "epoch": 0.03, "learning_rate": 4.9876432152659125e-05, "loss": 0.9071, "step": 31758 }, { "epoch": 0.03, "learning_rate": 4.987642435255194e-05, "loss": 0.8068, "step": 31759 }, { "epoch": 0.03, "learning_rate": 4.987641655219918e-05, "loss": 1.1453, "step": 31760 }, { "epoch": 0.03, "learning_rate": 4.9876408751600846e-05, "loss": 1.1479, "step": 31761 }, { "epoch": 0.03, "learning_rate": 4.987640095075695e-05, "loss": 0.5931, "step": 31762 }, { "epoch": 0.03, "learning_rate": 4.9876393149667486e-05, "loss": 0.4754, "step": 31763 }, { "epoch": 0.03, "learning_rate": 4.987638534833244e-05, "loss": 0.4786, "step": 31764 }, { "epoch": 0.03, "learning_rate": 4.987637754675184e-05, "loss": 0.4964, "step": 31765 }, { "epoch": 0.03, "learning_rate": 4.9876369744925664e-05, "loss": 0.42, "step": 31766 }, { "epoch": 0.03, "learning_rate": 4.9876361942853925e-05, "loss": 0.3279, "step": 31767 }, { "epoch": 0.03, "learning_rate": 4.9876354140536606e-05, "loss": 0.3479, "step": 31768 }, { "epoch": 0.03, "learning_rate": 4.987634633797372e-05, "loss": 0.3595, "step": 31769 }, { "epoch": 0.03, "learning_rate": 4.987633853516527e-05, "loss": 0.4791, "step": 31770 }, { "epoch": 0.03, "learning_rate": 4.9876330732111255e-05, "loss": 0.5427, "step": 31771 }, { "epoch": 0.03, "learning_rate": 4.9876322928811656e-05, "loss": 0.351, "step": 31772 }, { "epoch": 0.03, "learning_rate": 4.9876315125266506e-05, "loss": 0.3722, "step": 31773 }, { "epoch": 0.03, "learning_rate": 4.987630732147578e-05, "loss": 0.4441, "step": 31774 }, { "epoch": 0.03, "learning_rate": 4.9876299517439484e-05, "loss": 0.3322, "step": 31775 }, { "epoch": 0.03, "learning_rate": 4.987629171315762e-05, "loss": 0.3475, "step": 31776 }, { "epoch": 0.03, "learning_rate": 4.9876283908630195e-05, "loss": 0.3592, "step": 31777 }, { "epoch": 0.03, "learning_rate": 4.9876276103857186e-05, "loss": 0.3817, "step": 31778 }, { "epoch": 0.03, "learning_rate": 4.9876268298838626e-05, "loss": 0.2885, "step": 31779 }, { "epoch": 0.03, "learning_rate": 4.987626049357449e-05, "loss": 0.3136, "step": 31780 }, { "epoch": 0.03, "learning_rate": 4.987625268806478e-05, "loss": 0.3134, "step": 31781 }, { "epoch": 0.03, "learning_rate": 4.987624488230952e-05, "loss": 0.6942, "step": 31782 }, { "epoch": 0.03, "learning_rate": 4.987623707630867e-05, "loss": 0.7933, "step": 31783 }, { "epoch": 0.03, "learning_rate": 4.987622927006227e-05, "loss": 1.1611, "step": 31784 }, { "epoch": 0.03, "learning_rate": 4.9876221463570296e-05, "loss": 0.7482, "step": 31785 }, { "epoch": 0.03, "learning_rate": 4.9876213656832754e-05, "loss": 0.8452, "step": 31786 }, { "epoch": 0.03, "learning_rate": 4.9876205849849646e-05, "loss": 1.1772, "step": 31787 }, { "epoch": 0.03, "learning_rate": 4.9876198042620966e-05, "loss": 1.0983, "step": 31788 }, { "epoch": 0.03, "learning_rate": 4.987619023514672e-05, "loss": 1.3966, "step": 31789 }, { "epoch": 0.03, "learning_rate": 4.987618242742691e-05, "loss": 1.4792, "step": 31790 }, { "epoch": 0.03, "learning_rate": 4.987617461946154e-05, "loss": 1.2043, "step": 31791 }, { "epoch": 0.03, "learning_rate": 4.987616681125059e-05, "loss": 3.2489, "step": 31792 }, { "epoch": 0.03, "learning_rate": 4.9876159002794074e-05, "loss": 1.5124, "step": 31793 }, { "epoch": 0.03, "learning_rate": 4.9876151194092e-05, "loss": 1.1884, "step": 31794 }, { "epoch": 0.03, "learning_rate": 4.9876143385144356e-05, "loss": 0.7855, "step": 31795 }, { "epoch": 0.03, "learning_rate": 4.9876135575951136e-05, "loss": 0.3306, "step": 31796 }, { "epoch": 0.03, "learning_rate": 4.987612776651236e-05, "loss": 0.8425, "step": 31797 }, { "epoch": 0.03, "learning_rate": 4.987611995682801e-05, "loss": 1.0788, "step": 31798 }, { "epoch": 0.03, "learning_rate": 4.98761121468981e-05, "loss": 1.1452, "step": 31799 }, { "epoch": 0.03, "learning_rate": 4.987610433672262e-05, "loss": 0.8731, "step": 31800 }, { "epoch": 0.03, "learning_rate": 4.9876096526301575e-05, "loss": 0.3927, "step": 31801 }, { "epoch": 0.03, "learning_rate": 4.987608871563496e-05, "loss": 0.1377, "step": 31802 }, { "epoch": 0.03, "learning_rate": 4.9876080904722776e-05, "loss": 0.2839, "step": 31803 }, { "epoch": 0.03, "learning_rate": 4.9876073093565036e-05, "loss": 0.2665, "step": 31804 }, { "epoch": 0.03, "learning_rate": 4.9876065282161724e-05, "loss": 0.7329, "step": 31805 }, { "epoch": 0.03, "learning_rate": 4.987605747051285e-05, "loss": 0.75, "step": 31806 }, { "epoch": 0.03, "learning_rate": 4.9876049658618405e-05, "loss": 1.3929, "step": 31807 }, { "epoch": 0.03, "learning_rate": 4.987604184647839e-05, "loss": 1.36, "step": 31808 }, { "epoch": 0.03, "learning_rate": 4.987603403409282e-05, "loss": 0.9154, "step": 31809 }, { "epoch": 0.03, "learning_rate": 4.987602622146168e-05, "loss": 0.7824, "step": 31810 }, { "epoch": 0.03, "learning_rate": 4.987601840858498e-05, "loss": 0.8329, "step": 31811 }, { "epoch": 0.03, "learning_rate": 4.98760105954627e-05, "loss": 0.9107, "step": 31812 }, { "epoch": 0.03, "learning_rate": 4.987600278209487e-05, "loss": 0.7013, "step": 31813 }, { "epoch": 0.03, "learning_rate": 4.9875994968481465e-05, "loss": 1.2348, "step": 31814 }, { "epoch": 0.03, "learning_rate": 4.9875987154622496e-05, "loss": 1.1806, "step": 31815 }, { "epoch": 0.03, "learning_rate": 4.987597934051796e-05, "loss": 1.0009, "step": 31816 }, { "epoch": 0.03, "learning_rate": 4.9875971526167863e-05, "loss": 1.1363, "step": 31817 }, { "epoch": 0.03, "learning_rate": 4.9875963711572206e-05, "loss": 0.8128, "step": 31818 }, { "epoch": 0.03, "learning_rate": 4.987595589673098e-05, "loss": 0.7929, "step": 31819 }, { "epoch": 0.03, "learning_rate": 4.987594808164418e-05, "loss": 0.9732, "step": 31820 }, { "epoch": 0.03, "learning_rate": 4.987594026631182e-05, "loss": 1.1977, "step": 31821 }, { "epoch": 0.03, "learning_rate": 4.98759324507339e-05, "loss": 4.0099, "step": 31822 }, { "epoch": 0.03, "learning_rate": 4.987592463491041e-05, "loss": 1.0616, "step": 31823 }, { "epoch": 0.03, "learning_rate": 4.987591681884136e-05, "loss": 1.0005, "step": 31824 }, { "epoch": 0.03, "learning_rate": 4.9875909002526746e-05, "loss": 0.5469, "step": 31825 }, { "epoch": 0.03, "learning_rate": 4.9875901185966555e-05, "loss": 1.056, "step": 31826 }, { "epoch": 0.03, "learning_rate": 4.9875893369160806e-05, "loss": 0.1355, "step": 31827 }, { "epoch": 0.03, "learning_rate": 4.9875885552109506e-05, "loss": 0.0795, "step": 31828 }, { "epoch": 0.03, "learning_rate": 4.987587773481263e-05, "loss": 0.0818, "step": 31829 }, { "epoch": 0.03, "learning_rate": 4.987586991727019e-05, "loss": 0.0801, "step": 31830 }, { "epoch": 0.03, "learning_rate": 4.987586209948219e-05, "loss": 0.0864, "step": 31831 }, { "epoch": 0.03, "learning_rate": 4.987585428144862e-05, "loss": 0.0468, "step": 31832 }, { "epoch": 0.03, "learning_rate": 4.987584646316948e-05, "loss": 0.0412, "step": 31833 }, { "epoch": 0.03, "learning_rate": 4.987583864464479e-05, "loss": 0.0527, "step": 31834 }, { "epoch": 0.03, "learning_rate": 4.9875830825874535e-05, "loss": 0.0366, "step": 31835 }, { "epoch": 0.03, "learning_rate": 4.9875823006858714e-05, "loss": 0.1616, "step": 31836 }, { "epoch": 0.03, "learning_rate": 4.987581518759733e-05, "loss": 0.2796, "step": 31837 }, { "epoch": 0.03, "learning_rate": 4.987580736809038e-05, "loss": 0.0781, "step": 31838 }, { "epoch": 0.03, "learning_rate": 4.987579954833786e-05, "loss": 0.0846, "step": 31839 }, { "epoch": 0.03, "learning_rate": 4.987579172833979e-05, "loss": 0.6318, "step": 31840 }, { "epoch": 0.03, "learning_rate": 4.987578390809615e-05, "loss": 0.8273, "step": 31841 }, { "epoch": 0.03, "learning_rate": 4.9875776087606946e-05, "loss": 1.1391, "step": 31842 }, { "epoch": 0.03, "learning_rate": 4.987576826687218e-05, "loss": 0.8729, "step": 31843 }, { "epoch": 0.03, "learning_rate": 4.987576044589185e-05, "loss": 1.1101, "step": 31844 }, { "epoch": 0.03, "learning_rate": 4.987575262466596e-05, "loss": 0.9679, "step": 31845 }, { "epoch": 0.03, "learning_rate": 4.98757448031945e-05, "loss": 1.2256, "step": 31846 }, { "epoch": 0.03, "learning_rate": 4.9875736981477485e-05, "loss": 1.2033, "step": 31847 }, { "epoch": 0.03, "learning_rate": 4.9875729159514905e-05, "loss": 1.0598, "step": 31848 }, { "epoch": 0.03, "learning_rate": 4.987572133730676e-05, "loss": 1.0714, "step": 31849 }, { "epoch": 0.03, "learning_rate": 4.987571351485305e-05, "loss": 0.8735, "step": 31850 }, { "epoch": 0.03, "learning_rate": 4.987570569215379e-05, "loss": 1.2612, "step": 31851 }, { "epoch": 0.03, "learning_rate": 4.9875697869208954e-05, "loss": 3.1046, "step": 31852 }, { "epoch": 0.03, "learning_rate": 4.987569004601856e-05, "loss": 3.0212, "step": 31853 }, { "epoch": 0.03, "learning_rate": 4.9875682222582605e-05, "loss": 3.1012, "step": 31854 }, { "epoch": 0.03, "learning_rate": 4.987567439890108e-05, "loss": 3.0536, "step": 31855 }, { "epoch": 0.03, "learning_rate": 4.987566657497401e-05, "loss": 3.0911, "step": 31856 }, { "epoch": 0.03, "learning_rate": 4.9875658750801364e-05, "loss": 3.0392, "step": 31857 }, { "epoch": 0.03, "learning_rate": 4.9875650926383154e-05, "loss": 3.0407, "step": 31858 }, { "epoch": 0.03, "learning_rate": 4.987564310171939e-05, "loss": 3.063, "step": 31859 }, { "epoch": 0.03, "learning_rate": 4.9875635276810065e-05, "loss": 3.0566, "step": 31860 }, { "epoch": 0.03, "learning_rate": 4.9875627451655174e-05, "loss": 3.0408, "step": 31861 }, { "epoch": 0.03, "learning_rate": 4.9875619626254724e-05, "loss": 1.5149, "step": 31862 }, { "epoch": 0.03, "learning_rate": 4.9875611800608715e-05, "loss": 1.7108, "step": 31863 }, { "epoch": 0.03, "learning_rate": 4.9875603974717135e-05, "loss": 1.1433, "step": 31864 }, { "epoch": 0.03, "learning_rate": 4.987559614858e-05, "loss": 0.635, "step": 31865 }, { "epoch": 0.03, "learning_rate": 4.98755883221973e-05, "loss": 1.0935, "step": 31866 }, { "epoch": 0.03, "learning_rate": 4.987558049556904e-05, "loss": 1.0038, "step": 31867 }, { "epoch": 0.03, "learning_rate": 4.987557266869522e-05, "loss": 0.9091, "step": 31868 }, { "epoch": 0.03, "learning_rate": 4.987556484157584e-05, "loss": 1.0211, "step": 31869 }, { "epoch": 0.03, "learning_rate": 4.987555701421089e-05, "loss": 0.9729, "step": 31870 }, { "epoch": 0.03, "learning_rate": 4.987554918660039e-05, "loss": 1.053, "step": 31871 }, { "epoch": 0.03, "learning_rate": 4.9875541358744324e-05, "loss": 1.1989, "step": 31872 }, { "epoch": 0.03, "learning_rate": 4.9875533530642706e-05, "loss": 1.1059, "step": 31873 }, { "epoch": 0.03, "learning_rate": 4.987552570229551e-05, "loss": 0.9189, "step": 31874 }, { "epoch": 0.03, "learning_rate": 4.987551787370277e-05, "loss": 1.1019, "step": 31875 }, { "epoch": 0.03, "learning_rate": 4.9875510044864456e-05, "loss": 1.0345, "step": 31876 }, { "epoch": 0.03, "learning_rate": 4.987550221578059e-05, "loss": 1.0127, "step": 31877 }, { "epoch": 0.03, "learning_rate": 4.987549438645116e-05, "loss": 1.0473, "step": 31878 }, { "epoch": 0.03, "learning_rate": 4.987548655687617e-05, "loss": 1.1009, "step": 31879 }, { "epoch": 0.03, "learning_rate": 4.987547872705562e-05, "loss": 1.1956, "step": 31880 }, { "epoch": 0.03, "learning_rate": 4.987547089698952e-05, "loss": 0.9584, "step": 31881 }, { "epoch": 0.03, "learning_rate": 4.987546306667784e-05, "loss": 0.8262, "step": 31882 }, { "epoch": 0.03, "learning_rate": 4.9875455236120616e-05, "loss": 0.9753, "step": 31883 }, { "epoch": 0.03, "learning_rate": 4.9875447405317824e-05, "loss": 0.7293, "step": 31884 }, { "epoch": 0.03, "learning_rate": 4.987543957426948e-05, "loss": 1.0467, "step": 31885 }, { "epoch": 0.03, "learning_rate": 4.9875431742975565e-05, "loss": 1.0311, "step": 31886 }, { "epoch": 0.03, "learning_rate": 4.987542391143609e-05, "loss": 0.9075, "step": 31887 }, { "epoch": 0.03, "learning_rate": 4.9875416079651066e-05, "loss": 0.7952, "step": 31888 }, { "epoch": 0.03, "learning_rate": 4.987540824762048e-05, "loss": 1.0468, "step": 31889 }, { "epoch": 0.03, "learning_rate": 4.9875400415344334e-05, "loss": 1.0539, "step": 31890 }, { "epoch": 0.03, "learning_rate": 4.987539258282262e-05, "loss": 1.1983, "step": 31891 }, { "epoch": 0.03, "learning_rate": 4.9875384750055355e-05, "loss": 1.2786, "step": 31892 }, { "epoch": 0.03, "learning_rate": 4.9875376917042526e-05, "loss": 1.0487, "step": 31893 }, { "epoch": 0.03, "learning_rate": 4.987536908378414e-05, "loss": 1.1543, "step": 31894 }, { "epoch": 0.03, "learning_rate": 4.98753612502802e-05, "loss": 0.9508, "step": 31895 }, { "epoch": 0.03, "learning_rate": 4.987535341653069e-05, "loss": 0.9337, "step": 31896 }, { "epoch": 0.03, "learning_rate": 4.987534558253562e-05, "loss": 0.9819, "step": 31897 }, { "epoch": 0.03, "learning_rate": 4.9875337748295e-05, "loss": 0.7244, "step": 31898 }, { "epoch": 0.03, "learning_rate": 4.987532991380883e-05, "loss": 1.1801, "step": 31899 }, { "epoch": 0.03, "learning_rate": 4.9875322079077085e-05, "loss": 0.9528, "step": 31900 }, { "epoch": 0.03, "learning_rate": 4.987531424409979e-05, "loss": 1.0706, "step": 31901 }, { "epoch": 0.03, "learning_rate": 4.987530640887693e-05, "loss": 1.0384, "step": 31902 }, { "epoch": 0.03, "learning_rate": 4.9875298573408516e-05, "loss": 1.0503, "step": 31903 }, { "epoch": 0.03, "learning_rate": 4.987529073769454e-05, "loss": 1.0293, "step": 31904 }, { "epoch": 0.03, "learning_rate": 4.987528290173501e-05, "loss": 0.6786, "step": 31905 }, { "epoch": 0.03, "learning_rate": 4.9875275065529916e-05, "loss": 0.6744, "step": 31906 }, { "epoch": 0.03, "learning_rate": 4.987526722907927e-05, "loss": 0.3211, "step": 31907 }, { "epoch": 0.03, "learning_rate": 4.987525939238307e-05, "loss": 0.9709, "step": 31908 }, { "epoch": 0.03, "learning_rate": 4.98752515554413e-05, "loss": 0.8232, "step": 31909 }, { "epoch": 0.03, "learning_rate": 4.987524371825397e-05, "loss": 1.0957, "step": 31910 }, { "epoch": 0.03, "learning_rate": 4.9875235880821094e-05, "loss": 1.2013, "step": 31911 }, { "epoch": 0.03, "learning_rate": 4.987522804314265e-05, "loss": 1.1761, "step": 31912 }, { "epoch": 0.03, "learning_rate": 4.9875220205218656e-05, "loss": 1.115, "step": 31913 }, { "epoch": 0.03, "learning_rate": 4.987521236704911e-05, "loss": 0.966, "step": 31914 }, { "epoch": 0.03, "learning_rate": 4.987520452863399e-05, "loss": 1.0529, "step": 31915 }, { "epoch": 0.03, "learning_rate": 4.987519668997332e-05, "loss": 0.9583, "step": 31916 }, { "epoch": 0.03, "learning_rate": 4.9875188851067094e-05, "loss": 1.0661, "step": 31917 }, { "epoch": 0.03, "learning_rate": 4.987518101191531e-05, "loss": 0.954, "step": 31918 }, { "epoch": 0.03, "learning_rate": 4.987517317251797e-05, "loss": 0.9209, "step": 31919 }, { "epoch": 0.03, "learning_rate": 4.9875165332875075e-05, "loss": 0.8326, "step": 31920 }, { "epoch": 0.03, "learning_rate": 4.9875157492986615e-05, "loss": 0.9026, "step": 31921 }, { "epoch": 0.03, "learning_rate": 4.98751496528526e-05, "loss": 1.3498, "step": 31922 }, { "epoch": 0.03, "learning_rate": 4.987514181247303e-05, "loss": 0.8136, "step": 31923 }, { "epoch": 0.03, "learning_rate": 4.9875133971847904e-05, "loss": 0.9208, "step": 31924 }, { "epoch": 0.03, "learning_rate": 4.9875126130977224e-05, "loss": 1.0414, "step": 31925 }, { "epoch": 0.03, "learning_rate": 4.987511828986098e-05, "loss": 0.9674, "step": 31926 }, { "epoch": 0.03, "learning_rate": 4.987511044849918e-05, "loss": 1.0088, "step": 31927 }, { "epoch": 0.03, "learning_rate": 4.9875102606891834e-05, "loss": 1.02, "step": 31928 }, { "epoch": 0.03, "learning_rate": 4.987509476503892e-05, "loss": 1.4472, "step": 31929 }, { "epoch": 0.03, "learning_rate": 4.987508692294045e-05, "loss": 0.6673, "step": 31930 }, { "epoch": 0.03, "learning_rate": 4.9875079080596434e-05, "loss": 0.8934, "step": 31931 }, { "epoch": 0.03, "learning_rate": 4.9875071238006846e-05, "loss": 1.029, "step": 31932 }, { "epoch": 0.03, "learning_rate": 4.9875063395171714e-05, "loss": 1.2148, "step": 31933 }, { "epoch": 0.03, "learning_rate": 4.987505555209102e-05, "loss": 1.2076, "step": 31934 }, { "epoch": 0.03, "learning_rate": 4.9875047708764774e-05, "loss": 1.2639, "step": 31935 }, { "epoch": 0.03, "learning_rate": 4.987503986519297e-05, "loss": 2.7615, "step": 31936 }, { "epoch": 0.03, "learning_rate": 4.987503202137561e-05, "loss": 1.019, "step": 31937 }, { "epoch": 0.03, "learning_rate": 4.987502417731269e-05, "loss": 0.6664, "step": 31938 }, { "epoch": 0.03, "learning_rate": 4.987501633300422e-05, "loss": 0.6963, "step": 31939 }, { "epoch": 0.03, "learning_rate": 4.987500848845019e-05, "loss": 1.169, "step": 31940 }, { "epoch": 0.03, "learning_rate": 4.987500064365061e-05, "loss": 0.7437, "step": 31941 }, { "epoch": 0.03, "learning_rate": 4.9874992798605476e-05, "loss": 1.224, "step": 31942 }, { "epoch": 0.03, "learning_rate": 4.987498495331478e-05, "loss": 1.132, "step": 31943 }, { "epoch": 0.03, "learning_rate": 4.987497710777853e-05, "loss": 1.0177, "step": 31944 }, { "epoch": 0.03, "learning_rate": 4.987496926199673e-05, "loss": 1.2099, "step": 31945 }, { "epoch": 0.03, "learning_rate": 4.987496141596937e-05, "loss": 0.85, "step": 31946 }, { "epoch": 0.03, "learning_rate": 4.987495356969645e-05, "loss": 0.9207, "step": 31947 }, { "epoch": 0.03, "learning_rate": 4.9874945723177976e-05, "loss": 0.9421, "step": 31948 }, { "epoch": 0.03, "learning_rate": 4.987493787641396e-05, "loss": 0.9029, "step": 31949 }, { "epoch": 0.03, "learning_rate": 4.987493002940438e-05, "loss": 0.98, "step": 31950 }, { "epoch": 0.03, "learning_rate": 4.9874922182149244e-05, "loss": 1.0686, "step": 31951 }, { "epoch": 0.03, "learning_rate": 4.987491433464855e-05, "loss": 0.7382, "step": 31952 }, { "epoch": 0.03, "learning_rate": 4.987490648690231e-05, "loss": 1.0467, "step": 31953 }, { "epoch": 0.03, "learning_rate": 4.987489863891051e-05, "loss": 1.1727, "step": 31954 }, { "epoch": 0.03, "learning_rate": 4.9874890790673156e-05, "loss": 0.5697, "step": 31955 }, { "epoch": 0.03, "learning_rate": 4.987488294219025e-05, "loss": 0.4963, "step": 31956 }, { "epoch": 0.03, "learning_rate": 4.987487509346179e-05, "loss": 0.7733, "step": 31957 }, { "epoch": 0.03, "learning_rate": 4.9874867244487775e-05, "loss": 0.9277, "step": 31958 }, { "epoch": 0.03, "learning_rate": 4.98748593952682e-05, "loss": 1.258, "step": 31959 }, { "epoch": 0.03, "learning_rate": 4.987485154580308e-05, "loss": 1.753, "step": 31960 }, { "epoch": 0.03, "learning_rate": 4.98748436960924e-05, "loss": 1.6512, "step": 31961 }, { "epoch": 0.03, "learning_rate": 4.987483584613617e-05, "loss": 0.4306, "step": 31962 }, { "epoch": 0.03, "learning_rate": 4.987482799593438e-05, "loss": 1.2749, "step": 31963 }, { "epoch": 0.03, "learning_rate": 4.987482014548704e-05, "loss": 1.0235, "step": 31964 }, { "epoch": 0.03, "learning_rate": 4.987481229479415e-05, "loss": 0.7891, "step": 31965 }, { "epoch": 0.03, "learning_rate": 4.9874804443855696e-05, "loss": 0.7628, "step": 31966 }, { "epoch": 0.03, "learning_rate": 4.98747965926717e-05, "loss": 1.0068, "step": 31967 }, { "epoch": 0.03, "learning_rate": 4.987478874124214e-05, "loss": 0.9285, "step": 31968 }, { "epoch": 0.03, "learning_rate": 4.987478088956703e-05, "loss": 1.2886, "step": 31969 }, { "epoch": 0.03, "learning_rate": 4.987477303764637e-05, "loss": 1.0323, "step": 31970 }, { "epoch": 0.03, "learning_rate": 4.9874765185480154e-05, "loss": 0.8317, "step": 31971 }, { "epoch": 0.03, "learning_rate": 4.9874757333068386e-05, "loss": 1.1297, "step": 31972 }, { "epoch": 0.03, "learning_rate": 4.9874749480411066e-05, "loss": 1.2266, "step": 31973 }, { "epoch": 0.03, "learning_rate": 4.9874741627508195e-05, "loss": 0.6333, "step": 31974 }, { "epoch": 0.03, "learning_rate": 4.9874733774359766e-05, "loss": 0.5871, "step": 31975 }, { "epoch": 0.03, "learning_rate": 4.9874725920965785e-05, "loss": 1.0141, "step": 31976 }, { "epoch": 0.03, "learning_rate": 4.987471806732625e-05, "loss": 1.096, "step": 31977 }, { "epoch": 0.03, "learning_rate": 4.987471021344117e-05, "loss": 0.916, "step": 31978 }, { "epoch": 0.03, "learning_rate": 4.9874702359310533e-05, "loss": 1.3655, "step": 31979 }, { "epoch": 0.03, "learning_rate": 4.987469450493434e-05, "loss": 0.8321, "step": 31980 }, { "epoch": 0.03, "learning_rate": 4.9874686650312595e-05, "loss": 1.0193, "step": 31981 }, { "epoch": 0.03, "learning_rate": 4.98746787954453e-05, "loss": 0.7254, "step": 31982 }, { "epoch": 0.03, "learning_rate": 4.987467094033245e-05, "loss": 0.6001, "step": 31983 }, { "epoch": 0.03, "learning_rate": 4.987466308497405e-05, "loss": 0.6294, "step": 31984 }, { "epoch": 0.03, "learning_rate": 4.98746552293701e-05, "loss": 0.6862, "step": 31985 }, { "epoch": 0.03, "learning_rate": 4.987464737352059e-05, "loss": 0.6575, "step": 31986 }, { "epoch": 0.03, "learning_rate": 4.987463951742554e-05, "loss": 0.7447, "step": 31987 }, { "epoch": 0.03, "learning_rate": 4.9874631661084926e-05, "loss": 0.8369, "step": 31988 }, { "epoch": 0.03, "learning_rate": 4.987462380449876e-05, "loss": 0.7617, "step": 31989 }, { "epoch": 0.03, "learning_rate": 4.987461594766706e-05, "loss": 0.9034, "step": 31990 }, { "epoch": 0.03, "learning_rate": 4.9874608090589795e-05, "loss": 1.0087, "step": 31991 }, { "epoch": 0.03, "learning_rate": 4.987460023326698e-05, "loss": 0.9426, "step": 31992 }, { "epoch": 0.03, "learning_rate": 4.987459237569861e-05, "loss": 1.3042, "step": 31993 }, { "epoch": 0.03, "learning_rate": 4.9874584517884694e-05, "loss": 1.0579, "step": 31994 }, { "epoch": 0.03, "learning_rate": 4.987457665982522e-05, "loss": 0.7256, "step": 31995 }, { "epoch": 0.03, "learning_rate": 4.987456880152019e-05, "loss": 0.5277, "step": 31996 }, { "epoch": 0.03, "learning_rate": 4.987456094296963e-05, "loss": 0.9389, "step": 31997 }, { "epoch": 0.03, "learning_rate": 4.98745530841735e-05, "loss": 0.959, "step": 31998 }, { "epoch": 0.03, "learning_rate": 4.9874545225131825e-05, "loss": 1.1667, "step": 31999 }, { "epoch": 0.03, "learning_rate": 4.9874537365844595e-05, "loss": 1.1114, "step": 32000 }, { "epoch": 0.03, "eval_loss": 1.0753909349441528, "eval_runtime": 126.3526, "eval_samples_per_second": 10.961, "eval_steps_per_second": 5.485, "step": 32000 }, { "epoch": 0.03, "learning_rate": 4.987452950631182e-05, "loss": 1.2136, "step": 32001 }, { "epoch": 0.03, "learning_rate": 4.9874521646533495e-05, "loss": 1.2355, "step": 32002 }, { "epoch": 0.03, "learning_rate": 4.987451378650961e-05, "loss": 1.2057, "step": 32003 }, { "epoch": 0.03, "learning_rate": 4.987450592624019e-05, "loss": 1.0877, "step": 32004 }, { "epoch": 0.03, "learning_rate": 4.98744980657252e-05, "loss": 0.8758, "step": 32005 }, { "epoch": 0.03, "learning_rate": 4.987449020496468e-05, "loss": 0.877, "step": 32006 }, { "epoch": 0.03, "learning_rate": 4.9874482343958595e-05, "loss": 1.0007, "step": 32007 }, { "epoch": 0.03, "learning_rate": 4.987447448270696e-05, "loss": 0.8432, "step": 32008 }, { "epoch": 0.03, "learning_rate": 4.987446662120978e-05, "loss": 0.9247, "step": 32009 }, { "epoch": 0.03, "learning_rate": 4.9874458759467044e-05, "loss": 1.1426, "step": 32010 }, { "epoch": 0.03, "learning_rate": 4.987445089747876e-05, "loss": 1.1699, "step": 32011 }, { "epoch": 0.03, "learning_rate": 4.9874443035244935e-05, "loss": 0.8658, "step": 32012 }, { "epoch": 0.03, "learning_rate": 4.987443517276556e-05, "loss": 1.0073, "step": 32013 }, { "epoch": 0.03, "learning_rate": 4.9874427310040613e-05, "loss": 1.0153, "step": 32014 }, { "epoch": 0.03, "learning_rate": 4.987441944707014e-05, "loss": 0.8991, "step": 32015 }, { "epoch": 0.03, "learning_rate": 4.9874411583854106e-05, "loss": 0.9898, "step": 32016 }, { "epoch": 0.03, "learning_rate": 4.987440372039253e-05, "loss": 0.8066, "step": 32017 }, { "epoch": 0.03, "learning_rate": 4.9874395856685394e-05, "loss": 1.2558, "step": 32018 }, { "epoch": 0.03, "learning_rate": 4.9874387992732707e-05, "loss": 1.0333, "step": 32019 }, { "epoch": 0.03, "learning_rate": 4.987438012853448e-05, "loss": 0.8304, "step": 32020 }, { "epoch": 0.03, "learning_rate": 4.98743722640907e-05, "loss": 1.0565, "step": 32021 }, { "epoch": 0.03, "learning_rate": 4.987436439940138e-05, "loss": 0.794, "step": 32022 }, { "epoch": 0.03, "learning_rate": 4.987435653446649e-05, "loss": 0.9759, "step": 32023 }, { "epoch": 0.03, "learning_rate": 4.987434866928606e-05, "loss": 0.7237, "step": 32024 }, { "epoch": 0.03, "learning_rate": 4.9874340803860085e-05, "loss": 1.0472, "step": 32025 }, { "epoch": 0.03, "learning_rate": 4.987433293818856e-05, "loss": 0.9337, "step": 32026 }, { "epoch": 0.03, "learning_rate": 4.9874325072271487e-05, "loss": 0.9919, "step": 32027 }, { "epoch": 0.03, "learning_rate": 4.987431720610887e-05, "loss": 0.7039, "step": 32028 }, { "epoch": 0.03, "learning_rate": 4.987430933970069e-05, "loss": 1.026, "step": 32029 }, { "epoch": 0.03, "learning_rate": 4.987430147304697e-05, "loss": 0.8702, "step": 32030 }, { "epoch": 0.03, "learning_rate": 4.9874293606147706e-05, "loss": 1.022, "step": 32031 }, { "epoch": 0.03, "learning_rate": 4.9874285739002884e-05, "loss": 1.2144, "step": 32032 }, { "epoch": 0.03, "learning_rate": 4.987427787161252e-05, "loss": 0.8762, "step": 32033 }, { "epoch": 0.03, "learning_rate": 4.9874270003976605e-05, "loss": 1.0264, "step": 32034 }, { "epoch": 0.03, "learning_rate": 4.987426213609514e-05, "loss": 1.1591, "step": 32035 }, { "epoch": 0.03, "learning_rate": 4.987425426796813e-05, "loss": 0.6836, "step": 32036 }, { "epoch": 0.03, "learning_rate": 4.987424639959557e-05, "loss": 0.7129, "step": 32037 }, { "epoch": 0.03, "learning_rate": 4.987423853097746e-05, "loss": 0.9645, "step": 32038 }, { "epoch": 0.03, "learning_rate": 4.987423066211381e-05, "loss": 1.0317, "step": 32039 }, { "epoch": 0.03, "learning_rate": 4.9874222793004596e-05, "loss": 1.0358, "step": 32040 }, { "epoch": 0.03, "learning_rate": 4.987421492364985e-05, "loss": 0.5857, "step": 32041 }, { "epoch": 0.03, "learning_rate": 4.987420705404955e-05, "loss": 0.7896, "step": 32042 }, { "epoch": 0.03, "learning_rate": 4.987419918420371e-05, "loss": 0.6572, "step": 32043 }, { "epoch": 0.03, "learning_rate": 4.987419131411231e-05, "loss": 1.0858, "step": 32044 }, { "epoch": 0.03, "learning_rate": 4.987418344377537e-05, "loss": 0.9656, "step": 32045 }, { "epoch": 0.03, "learning_rate": 4.9874175573192877e-05, "loss": 1.0998, "step": 32046 }, { "epoch": 0.03, "learning_rate": 4.987416770236484e-05, "loss": 0.7939, "step": 32047 }, { "epoch": 0.03, "learning_rate": 4.987415983129126e-05, "loss": 0.5796, "step": 32048 }, { "epoch": 0.03, "learning_rate": 4.987415195997213e-05, "loss": 0.9221, "step": 32049 }, { "epoch": 0.03, "learning_rate": 4.987414408840745e-05, "loss": 1.0348, "step": 32050 }, { "epoch": 0.03, "learning_rate": 4.9874136216597223e-05, "loss": 1.0743, "step": 32051 }, { "epoch": 0.03, "learning_rate": 4.9874128344541447e-05, "loss": 1.2114, "step": 32052 }, { "epoch": 0.03, "learning_rate": 4.9874120472240125e-05, "loss": 1.6453, "step": 32053 }, { "epoch": 0.03, "learning_rate": 4.9874112599693266e-05, "loss": 1.1443, "step": 32054 }, { "epoch": 0.03, "learning_rate": 4.987410472690085e-05, "loss": 0.6813, "step": 32055 }, { "epoch": 0.03, "learning_rate": 4.987409685386289e-05, "loss": 0.8601, "step": 32056 }, { "epoch": 0.03, "learning_rate": 4.987408898057938e-05, "loss": 0.8051, "step": 32057 }, { "epoch": 0.03, "learning_rate": 4.987408110705033e-05, "loss": 0.8645, "step": 32058 }, { "epoch": 0.03, "learning_rate": 4.987407323327573e-05, "loss": 0.7239, "step": 32059 }, { "epoch": 0.03, "learning_rate": 4.987406535925558e-05, "loss": 0.6302, "step": 32060 }, { "epoch": 0.03, "learning_rate": 4.987405748498989e-05, "loss": 0.7608, "step": 32061 }, { "epoch": 0.03, "learning_rate": 4.987404961047866e-05, "loss": 0.5376, "step": 32062 }, { "epoch": 0.03, "learning_rate": 4.9874041735721864e-05, "loss": 0.9317, "step": 32063 }, { "epoch": 0.03, "learning_rate": 4.987403386071954e-05, "loss": 0.9353, "step": 32064 }, { "epoch": 0.03, "learning_rate": 4.987402598547166e-05, "loss": 0.75, "step": 32065 }, { "epoch": 0.03, "learning_rate": 4.9874018109978236e-05, "loss": 1.1904, "step": 32066 }, { "epoch": 0.03, "learning_rate": 4.987401023423927e-05, "loss": 1.3508, "step": 32067 }, { "epoch": 0.03, "learning_rate": 4.9874002358254756e-05, "loss": 0.7214, "step": 32068 }, { "epoch": 0.03, "learning_rate": 4.98739944820247e-05, "loss": 0.9986, "step": 32069 }, { "epoch": 0.03, "learning_rate": 4.98739866055491e-05, "loss": 0.9832, "step": 32070 }, { "epoch": 0.03, "learning_rate": 4.9873978728827945e-05, "loss": 1.0319, "step": 32071 }, { "epoch": 0.03, "learning_rate": 4.987397085186125e-05, "loss": 0.767, "step": 32072 }, { "epoch": 0.03, "learning_rate": 4.9873962974649005e-05, "loss": 1.0746, "step": 32073 }, { "epoch": 0.03, "learning_rate": 4.987395509719122e-05, "loss": 1.2047, "step": 32074 }, { "epoch": 0.03, "learning_rate": 4.987394721948788e-05, "loss": 0.8954, "step": 32075 }, { "epoch": 0.03, "learning_rate": 4.987393934153901e-05, "loss": 0.761, "step": 32076 }, { "epoch": 0.03, "learning_rate": 4.987393146334459e-05, "loss": 0.9105, "step": 32077 }, { "epoch": 0.03, "learning_rate": 4.987392358490462e-05, "loss": 1.0795, "step": 32078 }, { "epoch": 0.03, "learning_rate": 4.987391570621911e-05, "loss": 1.642, "step": 32079 }, { "epoch": 0.03, "learning_rate": 4.987390782728805e-05, "loss": 1.1692, "step": 32080 }, { "epoch": 0.03, "learning_rate": 4.987389994811145e-05, "loss": 0.9686, "step": 32081 }, { "epoch": 0.03, "learning_rate": 4.98738920686893e-05, "loss": 0.9879, "step": 32082 }, { "epoch": 0.03, "learning_rate": 4.9873884189021616e-05, "loss": 1.2706, "step": 32083 }, { "epoch": 0.03, "learning_rate": 4.9873876309108375e-05, "loss": 0.627, "step": 32084 }, { "epoch": 0.03, "learning_rate": 4.9873868428949597e-05, "loss": 1.0496, "step": 32085 }, { "epoch": 0.03, "learning_rate": 4.987386054854527e-05, "loss": 1.2311, "step": 32086 }, { "epoch": 0.03, "learning_rate": 4.9873852667895405e-05, "loss": 0.7827, "step": 32087 }, { "epoch": 0.03, "learning_rate": 4.987384478699999e-05, "loss": 0.7382, "step": 32088 }, { "epoch": 0.03, "learning_rate": 4.9873836905859035e-05, "loss": 0.913, "step": 32089 }, { "epoch": 0.03, "learning_rate": 4.9873829024472526e-05, "loss": 2.3012, "step": 32090 }, { "epoch": 0.03, "learning_rate": 4.9873821142840486e-05, "loss": 2.4916, "step": 32091 }, { "epoch": 0.03, "learning_rate": 4.98738132609629e-05, "loss": 2.109, "step": 32092 }, { "epoch": 0.03, "learning_rate": 4.9873805378839765e-05, "loss": 1.0196, "step": 32093 }, { "epoch": 0.03, "learning_rate": 4.987379749647109e-05, "loss": 1.1358, "step": 32094 }, { "epoch": 0.03, "learning_rate": 4.987378961385687e-05, "loss": 1.1044, "step": 32095 }, { "epoch": 0.03, "learning_rate": 4.98737817309971e-05, "loss": 0.842, "step": 32096 }, { "epoch": 0.03, "learning_rate": 4.98737738478918e-05, "loss": 0.835, "step": 32097 }, { "epoch": 0.03, "learning_rate": 4.9873765964540955e-05, "loss": 0.9828, "step": 32098 }, { "epoch": 0.03, "learning_rate": 4.987375808094456e-05, "loss": 1.2344, "step": 32099 }, { "epoch": 0.03, "learning_rate": 4.987375019710262e-05, "loss": 1.125, "step": 32100 }, { "epoch": 0.03, "learning_rate": 4.9873742313015135e-05, "loss": 1.1052, "step": 32101 }, { "epoch": 0.03, "learning_rate": 4.987373442868212e-05, "loss": 1.1208, "step": 32102 }, { "epoch": 0.03, "learning_rate": 4.987372654410355e-05, "loss": 0.9908, "step": 32103 }, { "epoch": 0.03, "learning_rate": 4.987371865927945e-05, "loss": 0.8829, "step": 32104 }, { "epoch": 0.03, "learning_rate": 4.9873710774209795e-05, "loss": 1.0347, "step": 32105 }, { "epoch": 0.03, "learning_rate": 4.98737028888946e-05, "loss": 1.0009, "step": 32106 }, { "epoch": 0.03, "learning_rate": 4.987369500333386e-05, "loss": 1.1277, "step": 32107 }, { "epoch": 0.03, "learning_rate": 4.987368711752758e-05, "loss": 1.1338, "step": 32108 }, { "epoch": 0.03, "learning_rate": 4.9873679231475753e-05, "loss": 1.3359, "step": 32109 }, { "epoch": 0.03, "learning_rate": 4.987367134517839e-05, "loss": 1.1322, "step": 32110 }, { "epoch": 0.03, "learning_rate": 4.9873663458635485e-05, "loss": 1.1468, "step": 32111 }, { "epoch": 0.03, "learning_rate": 4.9873655571847034e-05, "loss": 1.0133, "step": 32112 }, { "epoch": 0.03, "learning_rate": 4.987364768481304e-05, "loss": 0.7846, "step": 32113 }, { "epoch": 0.03, "learning_rate": 4.9873639797533504e-05, "loss": 0.8665, "step": 32114 }, { "epoch": 0.03, "learning_rate": 4.987363191000843e-05, "loss": 0.7232, "step": 32115 }, { "epoch": 0.03, "learning_rate": 4.987362402223781e-05, "loss": 0.726, "step": 32116 }, { "epoch": 0.03, "learning_rate": 4.9873616134221654e-05, "loss": 0.7718, "step": 32117 }, { "epoch": 0.03, "learning_rate": 4.987360824595995e-05, "loss": 0.7851, "step": 32118 }, { "epoch": 0.03, "learning_rate": 4.9873600357452704e-05, "loss": 0.6762, "step": 32119 }, { "epoch": 0.03, "learning_rate": 4.987359246869992e-05, "loss": 0.2888, "step": 32120 }, { "epoch": 0.03, "learning_rate": 4.9873584579701596e-05, "loss": 0.2416, "step": 32121 }, { "epoch": 0.03, "learning_rate": 4.9873576690457725e-05, "loss": 0.2012, "step": 32122 }, { "epoch": 0.03, "learning_rate": 4.987356880096831e-05, "loss": 0.4416, "step": 32123 }, { "epoch": 0.03, "learning_rate": 4.987356091123336e-05, "loss": 0.9283, "step": 32124 }, { "epoch": 0.03, "learning_rate": 4.987355302125287e-05, "loss": 0.9956, "step": 32125 }, { "epoch": 0.03, "learning_rate": 4.9873545131026835e-05, "loss": 1.0622, "step": 32126 }, { "epoch": 0.03, "learning_rate": 4.9873537240555254e-05, "loss": 0.8604, "step": 32127 }, { "epoch": 0.03, "learning_rate": 4.987352934983814e-05, "loss": 1.3243, "step": 32128 }, { "epoch": 0.03, "learning_rate": 4.9873521458875485e-05, "loss": 1.7325, "step": 32129 }, { "epoch": 0.03, "learning_rate": 4.987351356766728e-05, "loss": 1.6223, "step": 32130 }, { "epoch": 0.03, "learning_rate": 4.987350567621355e-05, "loss": 1.6283, "step": 32131 }, { "epoch": 0.03, "learning_rate": 4.987349778451427e-05, "loss": 1.0826, "step": 32132 }, { "epoch": 0.03, "learning_rate": 4.9873489892569445e-05, "loss": 1.12, "step": 32133 }, { "epoch": 0.03, "learning_rate": 4.9873482000379085e-05, "loss": 0.687, "step": 32134 }, { "epoch": 0.03, "learning_rate": 4.987347410794319e-05, "loss": 0.9835, "step": 32135 }, { "epoch": 0.03, "learning_rate": 4.9873466215261744e-05, "loss": 1.0354, "step": 32136 }, { "epoch": 0.03, "learning_rate": 4.987345832233476e-05, "loss": 1.08, "step": 32137 }, { "epoch": 0.03, "learning_rate": 4.987345042916224e-05, "loss": 0.9266, "step": 32138 }, { "epoch": 0.03, "learning_rate": 4.9873442535744176e-05, "loss": 0.9142, "step": 32139 }, { "epoch": 0.03, "learning_rate": 4.987343464208057e-05, "loss": 0.8697, "step": 32140 }, { "epoch": 0.03, "learning_rate": 4.987342674817142e-05, "loss": 1.0542, "step": 32141 }, { "epoch": 0.03, "learning_rate": 4.987341885401674e-05, "loss": 1.0708, "step": 32142 }, { "epoch": 0.03, "learning_rate": 4.987341095961652e-05, "loss": 0.7928, "step": 32143 }, { "epoch": 0.03, "learning_rate": 4.987340306497076e-05, "loss": 0.9439, "step": 32144 }, { "epoch": 0.03, "learning_rate": 4.9873395170079454e-05, "loss": 0.7858, "step": 32145 }, { "epoch": 0.03, "learning_rate": 4.987338727494261e-05, "loss": 0.9896, "step": 32146 }, { "epoch": 0.03, "learning_rate": 4.9873379379560226e-05, "loss": 0.8663, "step": 32147 }, { "epoch": 0.03, "learning_rate": 4.987337148393231e-05, "loss": 0.6797, "step": 32148 }, { "epoch": 0.03, "learning_rate": 4.9873363588058846e-05, "loss": 1.0168, "step": 32149 }, { "epoch": 0.03, "learning_rate": 4.987335569193985e-05, "loss": 1.0112, "step": 32150 }, { "epoch": 0.03, "learning_rate": 4.987334779557531e-05, "loss": 0.621, "step": 32151 }, { "epoch": 0.03, "learning_rate": 4.9873339898965224e-05, "loss": 1.0882, "step": 32152 }, { "epoch": 0.03, "learning_rate": 4.987333200210961e-05, "loss": 0.469, "step": 32153 }, { "epoch": 0.03, "learning_rate": 4.987332410500845e-05, "loss": 0.9131, "step": 32154 }, { "epoch": 0.03, "learning_rate": 4.9873316207661755e-05, "loss": 0.9622, "step": 32155 }, { "epoch": 0.03, "learning_rate": 4.987330831006952e-05, "loss": 1.1395, "step": 32156 }, { "epoch": 0.03, "learning_rate": 4.987330041223175e-05, "loss": 0.7577, "step": 32157 }, { "epoch": 0.03, "learning_rate": 4.987329251414844e-05, "loss": 0.9235, "step": 32158 }, { "epoch": 0.03, "learning_rate": 4.987328461581958e-05, "loss": 1.1154, "step": 32159 }, { "epoch": 0.03, "learning_rate": 4.9873276717245196e-05, "loss": 0.9374, "step": 32160 }, { "epoch": 0.03, "learning_rate": 4.9873268818425265e-05, "loss": 0.9297, "step": 32161 }, { "epoch": 0.03, "learning_rate": 4.9873260919359796e-05, "loss": 1.1365, "step": 32162 }, { "epoch": 0.03, "learning_rate": 4.9873253020048796e-05, "loss": 1.0673, "step": 32163 }, { "epoch": 0.03, "learning_rate": 4.987324512049225e-05, "loss": 1.0278, "step": 32164 }, { "epoch": 0.03, "learning_rate": 4.987323722069017e-05, "loss": 0.8769, "step": 32165 }, { "epoch": 0.03, "learning_rate": 4.987322932064255e-05, "loss": 0.9991, "step": 32166 }, { "epoch": 0.03, "learning_rate": 4.987322142034939e-05, "loss": 1.1742, "step": 32167 }, { "epoch": 0.03, "learning_rate": 4.9873213519810694e-05, "loss": 1.2638, "step": 32168 }, { "epoch": 0.03, "learning_rate": 4.9873205619026466e-05, "loss": 1.0637, "step": 32169 }, { "epoch": 0.03, "learning_rate": 4.987319771799669e-05, "loss": 0.957, "step": 32170 }, { "epoch": 0.03, "learning_rate": 4.987318981672138e-05, "loss": 1.0365, "step": 32171 }, { "epoch": 0.03, "learning_rate": 4.9873181915200536e-05, "loss": 0.8617, "step": 32172 }, { "epoch": 0.03, "learning_rate": 4.987317401343415e-05, "loss": 0.6602, "step": 32173 }, { "epoch": 0.03, "learning_rate": 4.987316611142223e-05, "loss": 1.0235, "step": 32174 }, { "epoch": 0.03, "learning_rate": 4.987315820916477e-05, "loss": 0.985, "step": 32175 }, { "epoch": 0.03, "learning_rate": 4.987315030666178e-05, "loss": 1.0612, "step": 32176 }, { "epoch": 0.03, "learning_rate": 4.987314240391324e-05, "loss": 0.9585, "step": 32177 }, { "epoch": 0.03, "learning_rate": 4.9873134500919166e-05, "loss": 0.9039, "step": 32178 }, { "epoch": 0.03, "learning_rate": 4.987312659767956e-05, "loss": 0.9694, "step": 32179 }, { "epoch": 0.03, "learning_rate": 4.9873118694194415e-05, "loss": 0.9437, "step": 32180 }, { "epoch": 0.03, "learning_rate": 4.9873110790463726e-05, "loss": 1.0862, "step": 32181 }, { "epoch": 0.03, "learning_rate": 4.987310288648751e-05, "loss": 1.0017, "step": 32182 }, { "epoch": 0.03, "learning_rate": 4.9873094982265754e-05, "loss": 0.9497, "step": 32183 }, { "epoch": 0.03, "learning_rate": 4.9873087077798465e-05, "loss": 1.0259, "step": 32184 }, { "epoch": 0.03, "learning_rate": 4.987307917308564e-05, "loss": 1.1924, "step": 32185 }, { "epoch": 0.03, "learning_rate": 4.987307126812727e-05, "loss": 1.2114, "step": 32186 }, { "epoch": 0.03, "learning_rate": 4.987306336292337e-05, "loss": 1.3263, "step": 32187 }, { "epoch": 0.03, "learning_rate": 4.987305545747393e-05, "loss": 0.8753, "step": 32188 }, { "epoch": 0.03, "learning_rate": 4.987304755177895e-05, "loss": 0.8218, "step": 32189 }, { "epoch": 0.03, "learning_rate": 4.9873039645838446e-05, "loss": 0.7118, "step": 32190 }, { "epoch": 0.03, "learning_rate": 4.98730317396524e-05, "loss": 0.5357, "step": 32191 }, { "epoch": 0.03, "learning_rate": 4.987302383322081e-05, "loss": 0.8419, "step": 32192 }, { "epoch": 0.03, "learning_rate": 4.9873015926543696e-05, "loss": 1.1089, "step": 32193 }, { "epoch": 0.03, "learning_rate": 4.9873008019621034e-05, "loss": 1.0913, "step": 32194 }, { "epoch": 0.03, "learning_rate": 4.987300011245285e-05, "loss": 1.0098, "step": 32195 }, { "epoch": 0.03, "learning_rate": 4.9872992205039124e-05, "loss": 0.9657, "step": 32196 }, { "epoch": 0.03, "learning_rate": 4.9872984297379855e-05, "loss": 1.0139, "step": 32197 }, { "epoch": 0.03, "learning_rate": 4.9872976389475056e-05, "loss": 1.048, "step": 32198 }, { "epoch": 0.03, "learning_rate": 4.9872968481324725e-05, "loss": 1.0603, "step": 32199 }, { "epoch": 0.03, "learning_rate": 4.9872960572928856e-05, "loss": 0.8554, "step": 32200 }, { "epoch": 0.03, "learning_rate": 4.9872952664287456e-05, "loss": 1.222, "step": 32201 }, { "epoch": 0.03, "learning_rate": 4.9872944755400505e-05, "loss": 1.0251, "step": 32202 }, { "epoch": 0.03, "learning_rate": 4.9872936846268036e-05, "loss": 1.1161, "step": 32203 }, { "epoch": 0.03, "learning_rate": 4.987292893689003e-05, "loss": 1.1518, "step": 32204 }, { "epoch": 0.03, "learning_rate": 4.9872921027266484e-05, "loss": 0.7286, "step": 32205 }, { "epoch": 0.03, "learning_rate": 4.98729131173974e-05, "loss": 0.3899, "step": 32206 }, { "epoch": 0.03, "learning_rate": 4.987290520728278e-05, "loss": 1.0026, "step": 32207 }, { "epoch": 0.03, "learning_rate": 4.9872897296922636e-05, "loss": 0.9712, "step": 32208 }, { "epoch": 0.03, "learning_rate": 4.987288938631695e-05, "loss": 0.7481, "step": 32209 }, { "epoch": 0.03, "learning_rate": 4.9872881475465725e-05, "loss": 0.8154, "step": 32210 }, { "epoch": 0.03, "learning_rate": 4.987287356436898e-05, "loss": 1.4481, "step": 32211 }, { "epoch": 0.03, "learning_rate": 4.987286565302669e-05, "loss": 1.0816, "step": 32212 }, { "epoch": 0.03, "learning_rate": 4.987285774143886e-05, "loss": 0.987, "step": 32213 }, { "epoch": 0.03, "learning_rate": 4.9872849829605504e-05, "loss": 0.9902, "step": 32214 }, { "epoch": 0.03, "learning_rate": 4.9872841917526614e-05, "loss": 1.0435, "step": 32215 }, { "epoch": 0.03, "learning_rate": 4.987283400520219e-05, "loss": 1.0892, "step": 32216 }, { "epoch": 0.03, "learning_rate": 4.987282609263223e-05, "loss": 1.6691, "step": 32217 }, { "epoch": 0.03, "learning_rate": 4.987281817981674e-05, "loss": 0.781, "step": 32218 }, { "epoch": 0.03, "learning_rate": 4.987281026675571e-05, "loss": 0.739, "step": 32219 }, { "epoch": 0.03, "learning_rate": 4.987280235344914e-05, "loss": 0.9163, "step": 32220 }, { "epoch": 0.03, "learning_rate": 4.9872794439897045e-05, "loss": 0.6466, "step": 32221 }, { "epoch": 0.03, "learning_rate": 4.987278652609942e-05, "loss": 0.4372, "step": 32222 }, { "epoch": 0.03, "learning_rate": 4.987277861205626e-05, "loss": 0.922, "step": 32223 }, { "epoch": 0.03, "learning_rate": 4.9872770697767554e-05, "loss": 0.9833, "step": 32224 }, { "epoch": 0.03, "learning_rate": 4.987276278323333e-05, "loss": 0.8583, "step": 32225 }, { "epoch": 0.03, "learning_rate": 4.9872754868453566e-05, "loss": 0.8442, "step": 32226 }, { "epoch": 0.03, "learning_rate": 4.987274695342827e-05, "loss": 0.9774, "step": 32227 }, { "epoch": 0.03, "learning_rate": 4.987273903815743e-05, "loss": 0.6232, "step": 32228 }, { "epoch": 0.03, "learning_rate": 4.987273112264107e-05, "loss": 1.2125, "step": 32229 }, { "epoch": 0.03, "learning_rate": 4.987272320687918e-05, "loss": 1.9775, "step": 32230 }, { "epoch": 0.03, "learning_rate": 4.9872715290871746e-05, "loss": 2.0903, "step": 32231 }, { "epoch": 0.03, "learning_rate": 4.987270737461878e-05, "loss": 1.348, "step": 32232 }, { "epoch": 0.03, "learning_rate": 4.987269945812029e-05, "loss": 0.6764, "step": 32233 }, { "epoch": 0.03, "learning_rate": 4.987269154137626e-05, "loss": 1.1157, "step": 32234 }, { "epoch": 0.03, "learning_rate": 4.987268362438669e-05, "loss": 1.1064, "step": 32235 }, { "epoch": 0.03, "learning_rate": 4.987267570715161e-05, "loss": 1.077, "step": 32236 }, { "epoch": 0.03, "learning_rate": 4.987266778967098e-05, "loss": 0.9358, "step": 32237 }, { "epoch": 0.03, "learning_rate": 4.987265987194482e-05, "loss": 0.8005, "step": 32238 }, { "epoch": 0.03, "learning_rate": 4.987265195397313e-05, "loss": 0.9895, "step": 32239 }, { "epoch": 0.03, "learning_rate": 4.98726440357559e-05, "loss": 0.9428, "step": 32240 }, { "epoch": 0.03, "learning_rate": 4.987263611729315e-05, "loss": 1.0787, "step": 32241 }, { "epoch": 0.03, "learning_rate": 4.987262819858486e-05, "loss": 1.0498, "step": 32242 }, { "epoch": 0.03, "learning_rate": 4.9872620279631037e-05, "loss": 0.9515, "step": 32243 }, { "epoch": 0.03, "learning_rate": 4.987261236043169e-05, "loss": 0.9531, "step": 32244 }, { "epoch": 0.03, "learning_rate": 4.9872604440986804e-05, "loss": 0.8553, "step": 32245 }, { "epoch": 0.03, "learning_rate": 4.987259652129639e-05, "loss": 0.3532, "step": 32246 }, { "epoch": 0.03, "learning_rate": 4.9872588601360434e-05, "loss": 0.6645, "step": 32247 }, { "epoch": 0.03, "learning_rate": 4.987258068117896e-05, "loss": 0.673, "step": 32248 }, { "epoch": 0.03, "learning_rate": 4.9872572760751946e-05, "loss": 1.029, "step": 32249 }, { "epoch": 0.03, "learning_rate": 4.9872564840079405e-05, "loss": 0.7275, "step": 32250 }, { "epoch": 0.03, "learning_rate": 4.987255691916133e-05, "loss": 0.7414, "step": 32251 }, { "epoch": 0.03, "learning_rate": 4.9872548997997724e-05, "loss": 0.4638, "step": 32252 }, { "epoch": 0.03, "learning_rate": 4.987254107658858e-05, "loss": 0.6007, "step": 32253 }, { "epoch": 0.03, "learning_rate": 4.987253315493392e-05, "loss": 0.5074, "step": 32254 }, { "epoch": 0.03, "learning_rate": 4.9872525233033715e-05, "loss": 0.4355, "step": 32255 }, { "epoch": 0.03, "learning_rate": 4.987251731088799e-05, "loss": 0.7918, "step": 32256 }, { "epoch": 0.03, "learning_rate": 4.987250938849672e-05, "loss": 1.6196, "step": 32257 }, { "epoch": 0.03, "learning_rate": 4.987250146585993e-05, "loss": 0.8896, "step": 32258 }, { "epoch": 0.03, "learning_rate": 4.987249354297761e-05, "loss": 0.6936, "step": 32259 }, { "epoch": 0.03, "learning_rate": 4.987248561984976e-05, "loss": 0.8433, "step": 32260 }, { "epoch": 0.03, "learning_rate": 4.987247769647637e-05, "loss": 0.7155, "step": 32261 }, { "epoch": 0.03, "learning_rate": 4.9872469772857455e-05, "loss": 1.1458, "step": 32262 }, { "epoch": 0.03, "learning_rate": 4.9872461848993016e-05, "loss": 0.9279, "step": 32263 }, { "epoch": 0.03, "learning_rate": 4.987245392488303e-05, "loss": 1.0874, "step": 32264 }, { "epoch": 0.03, "learning_rate": 4.987244600052753e-05, "loss": 1.0139, "step": 32265 }, { "epoch": 0.03, "learning_rate": 4.9872438075926486e-05, "loss": 1.1876, "step": 32266 }, { "epoch": 0.03, "learning_rate": 4.9872430151079916e-05, "loss": 0.5851, "step": 32267 }, { "epoch": 0.03, "learning_rate": 4.987242222598782e-05, "loss": 0.8631, "step": 32268 }, { "epoch": 0.03, "learning_rate": 4.9872414300650196e-05, "loss": 1.2758, "step": 32269 }, { "epoch": 0.03, "learning_rate": 4.987240637506704e-05, "loss": 1.162, "step": 32270 }, { "epoch": 0.03, "learning_rate": 4.987239844923836e-05, "loss": 1.1979, "step": 32271 }, { "epoch": 0.03, "learning_rate": 4.987239052316414e-05, "loss": 0.9473, "step": 32272 }, { "epoch": 0.03, "learning_rate": 4.987238259684439e-05, "loss": 1.1872, "step": 32273 }, { "epoch": 0.03, "learning_rate": 4.987237467027911e-05, "loss": 0.8769, "step": 32274 }, { "epoch": 0.03, "learning_rate": 4.98723667434683e-05, "loss": 0.9966, "step": 32275 }, { "epoch": 0.03, "learning_rate": 4.9872358816411966e-05, "loss": 0.9936, "step": 32276 }, { "epoch": 0.03, "learning_rate": 4.9872350889110105e-05, "loss": 1.1498, "step": 32277 }, { "epoch": 0.03, "learning_rate": 4.987234296156271e-05, "loss": 0.6918, "step": 32278 }, { "epoch": 0.03, "learning_rate": 4.987233503376978e-05, "loss": 1.1686, "step": 32279 }, { "epoch": 0.03, "learning_rate": 4.987232710573133e-05, "loss": 1.0333, "step": 32280 }, { "epoch": 0.03, "learning_rate": 4.987231917744734e-05, "loss": 0.989, "step": 32281 }, { "epoch": 0.03, "learning_rate": 4.9872311248917833e-05, "loss": 1.0477, "step": 32282 }, { "epoch": 0.03, "learning_rate": 4.98723033201428e-05, "loss": 1.1021, "step": 32283 }, { "epoch": 0.03, "learning_rate": 4.987229539112223e-05, "loss": 0.7447, "step": 32284 }, { "epoch": 0.03, "learning_rate": 4.9872287461856124e-05, "loss": 1.1506, "step": 32285 }, { "epoch": 0.03, "learning_rate": 4.98722795323445e-05, "loss": 1.0141, "step": 32286 }, { "epoch": 0.03, "learning_rate": 4.987227160258735e-05, "loss": 1.0178, "step": 32287 }, { "epoch": 0.03, "learning_rate": 4.987226367258466e-05, "loss": 0.7663, "step": 32288 }, { "epoch": 0.03, "learning_rate": 4.9872255742336447e-05, "loss": 0.8601, "step": 32289 }, { "epoch": 0.03, "learning_rate": 4.987224781184271e-05, "loss": 1.008, "step": 32290 }, { "epoch": 0.03, "learning_rate": 4.9872239881103445e-05, "loss": 1.0658, "step": 32291 }, { "epoch": 0.03, "learning_rate": 4.9872231950118644e-05, "loss": 0.7518, "step": 32292 }, { "epoch": 0.03, "learning_rate": 4.987222401888831e-05, "loss": 1.0048, "step": 32293 }, { "epoch": 0.03, "learning_rate": 4.987221608741246e-05, "loss": 0.8833, "step": 32294 }, { "epoch": 0.03, "learning_rate": 4.987220815569108e-05, "loss": 1.1924, "step": 32295 }, { "epoch": 0.03, "learning_rate": 4.9872200223724176e-05, "loss": 0.7981, "step": 32296 }, { "epoch": 0.03, "learning_rate": 4.987219229151173e-05, "loss": 1.1354, "step": 32297 }, { "epoch": 0.03, "learning_rate": 4.987218435905377e-05, "loss": 1.507, "step": 32298 }, { "epoch": 0.03, "learning_rate": 4.9872176426350274e-05, "loss": 0.1541, "step": 32299 }, { "epoch": 0.03, "learning_rate": 4.987216849340125e-05, "loss": 0.1656, "step": 32300 }, { "epoch": 0.03, "learning_rate": 4.9872160560206704e-05, "loss": 0.1489, "step": 32301 }, { "epoch": 0.03, "learning_rate": 4.987215262676663e-05, "loss": 0.098, "step": 32302 }, { "epoch": 0.03, "learning_rate": 4.987214469308102e-05, "loss": 0.0884, "step": 32303 }, { "epoch": 0.03, "learning_rate": 4.9872136759149896e-05, "loss": 0.0681, "step": 32304 }, { "epoch": 0.03, "learning_rate": 4.9872128824973244e-05, "loss": 0.8575, "step": 32305 }, { "epoch": 0.03, "learning_rate": 4.9872120890551055e-05, "loss": 1.4759, "step": 32306 }, { "epoch": 0.03, "learning_rate": 4.987211295588334e-05, "loss": 1.2485, "step": 32307 }, { "epoch": 0.03, "learning_rate": 4.98721050209701e-05, "loss": 1.0156, "step": 32308 }, { "epoch": 0.03, "learning_rate": 4.9872097085811334e-05, "loss": 1.0028, "step": 32309 }, { "epoch": 0.03, "learning_rate": 4.987208915040704e-05, "loss": 0.9394, "step": 32310 }, { "epoch": 0.03, "learning_rate": 4.987208121475722e-05, "loss": 0.8904, "step": 32311 }, { "epoch": 0.03, "learning_rate": 4.9872073278861873e-05, "loss": 0.8638, "step": 32312 }, { "epoch": 0.03, "learning_rate": 4.9872065342721e-05, "loss": 0.9758, "step": 32313 }, { "epoch": 0.03, "learning_rate": 4.98720574063346e-05, "loss": 0.8389, "step": 32314 }, { "epoch": 0.03, "learning_rate": 4.987204946970267e-05, "loss": 1.0429, "step": 32315 }, { "epoch": 0.03, "learning_rate": 4.987204153282522e-05, "loss": 0.9808, "step": 32316 }, { "epoch": 0.03, "learning_rate": 4.987203359570224e-05, "loss": 0.989, "step": 32317 }, { "epoch": 0.03, "learning_rate": 4.987202565833374e-05, "loss": 0.7948, "step": 32318 }, { "epoch": 0.03, "learning_rate": 4.987201772071971e-05, "loss": 1.0004, "step": 32319 }, { "epoch": 0.03, "learning_rate": 4.987200978286015e-05, "loss": 1.1524, "step": 32320 }, { "epoch": 0.03, "learning_rate": 4.987200184475507e-05, "loss": 1.2241, "step": 32321 }, { "epoch": 0.03, "learning_rate": 4.9871993906404455e-05, "loss": 1.2511, "step": 32322 }, { "epoch": 0.03, "learning_rate": 4.9871985967808324e-05, "loss": 1.7734, "step": 32323 }, { "epoch": 0.03, "learning_rate": 4.987197802896666e-05, "loss": 1.8421, "step": 32324 }, { "epoch": 0.03, "learning_rate": 4.987197008987947e-05, "loss": 1.3899, "step": 32325 }, { "epoch": 0.03, "learning_rate": 4.987196215054676e-05, "loss": 1.0676, "step": 32326 }, { "epoch": 0.03, "learning_rate": 4.9871954210968525e-05, "loss": 1.0439, "step": 32327 }, { "epoch": 0.03, "learning_rate": 4.9871946271144766e-05, "loss": 1.4981, "step": 32328 }, { "epoch": 0.03, "learning_rate": 4.9871938331075475e-05, "loss": 1.6174, "step": 32329 }, { "epoch": 0.03, "learning_rate": 4.9871930390760654e-05, "loss": 1.143, "step": 32330 }, { "epoch": 0.03, "learning_rate": 4.987192245020032e-05, "loss": 1.0296, "step": 32331 }, { "epoch": 0.03, "learning_rate": 4.987191450939446e-05, "loss": 0.924, "step": 32332 }, { "epoch": 0.03, "learning_rate": 4.9871906568343064e-05, "loss": 1.2836, "step": 32333 }, { "epoch": 0.03, "learning_rate": 4.987189862704615e-05, "loss": 1.3367, "step": 32334 }, { "epoch": 0.03, "learning_rate": 4.987189068550371e-05, "loss": 1.2626, "step": 32335 }, { "epoch": 0.03, "learning_rate": 4.987188274371575e-05, "loss": 1.1438, "step": 32336 }, { "epoch": 0.03, "learning_rate": 4.987187480168226e-05, "loss": 0.5562, "step": 32337 }, { "epoch": 0.03, "learning_rate": 4.9871866859403246e-05, "loss": 0.5784, "step": 32338 }, { "epoch": 0.03, "learning_rate": 4.987185891687871e-05, "loss": 0.8307, "step": 32339 }, { "epoch": 0.03, "learning_rate": 4.987185097410865e-05, "loss": 1.0153, "step": 32340 }, { "epoch": 0.03, "learning_rate": 4.987184303109306e-05, "loss": 1.2075, "step": 32341 }, { "epoch": 0.03, "learning_rate": 4.9871835087831945e-05, "loss": 1.0312, "step": 32342 }, { "epoch": 0.03, "learning_rate": 4.987182714432531e-05, "loss": 0.8384, "step": 32343 }, { "epoch": 0.03, "learning_rate": 4.987181920057316e-05, "loss": 0.6897, "step": 32344 }, { "epoch": 0.03, "learning_rate": 4.987181125657547e-05, "loss": 0.5851, "step": 32345 }, { "epoch": 0.03, "learning_rate": 4.987180331233226e-05, "loss": 0.9492, "step": 32346 }, { "epoch": 0.03, "learning_rate": 4.9871795367843524e-05, "loss": 1.1692, "step": 32347 }, { "epoch": 0.03, "learning_rate": 4.987178742310927e-05, "loss": 0.8737, "step": 32348 }, { "epoch": 0.03, "learning_rate": 4.9871779478129494e-05, "loss": 1.014, "step": 32349 }, { "epoch": 0.03, "learning_rate": 4.9871771532904185e-05, "loss": 1.108, "step": 32350 }, { "epoch": 0.03, "learning_rate": 4.987176358743336e-05, "loss": 1.0333, "step": 32351 }, { "epoch": 0.03, "learning_rate": 4.987175564171701e-05, "loss": 0.9596, "step": 32352 }, { "epoch": 0.03, "learning_rate": 4.987174769575514e-05, "loss": 0.9578, "step": 32353 }, { "epoch": 0.03, "learning_rate": 4.987173974954774e-05, "loss": 0.7098, "step": 32354 }, { "epoch": 0.03, "learning_rate": 4.9871731803094825e-05, "loss": 1.264, "step": 32355 }, { "epoch": 0.03, "learning_rate": 4.9871723856396377e-05, "loss": 1.0822, "step": 32356 }, { "epoch": 0.03, "learning_rate": 4.987171590945241e-05, "loss": 1.1312, "step": 32357 }, { "epoch": 0.03, "learning_rate": 4.9871707962262914e-05, "loss": 1.2645, "step": 32358 }, { "epoch": 0.03, "learning_rate": 4.987170001482791e-05, "loss": 0.9485, "step": 32359 }, { "epoch": 0.03, "learning_rate": 4.987169206714737e-05, "loss": 0.7995, "step": 32360 }, { "epoch": 0.03, "learning_rate": 4.9871684119221305e-05, "loss": 1.227, "step": 32361 }, { "epoch": 0.03, "learning_rate": 4.987167617104973e-05, "loss": 0.8371, "step": 32362 }, { "epoch": 0.03, "learning_rate": 4.987166822263262e-05, "loss": 1.5139, "step": 32363 }, { "epoch": 0.03, "learning_rate": 4.987166027397e-05, "loss": 1.9961, "step": 32364 }, { "epoch": 0.03, "learning_rate": 4.9871652325061844e-05, "loss": 2.1983, "step": 32365 }, { "epoch": 0.03, "learning_rate": 4.987164437590817e-05, "loss": 2.116, "step": 32366 }, { "epoch": 0.03, "learning_rate": 4.987163642650898e-05, "loss": 1.9973, "step": 32367 }, { "epoch": 0.03, "learning_rate": 4.9871628476864264e-05, "loss": 2.0621, "step": 32368 }, { "epoch": 0.03, "learning_rate": 4.987162052697402e-05, "loss": 2.1445, "step": 32369 }, { "epoch": 0.03, "learning_rate": 4.9871612576838265e-05, "loss": 1.8926, "step": 32370 }, { "epoch": 0.03, "learning_rate": 4.987160462645698e-05, "loss": 2.0815, "step": 32371 }, { "epoch": 0.03, "learning_rate": 4.987159667583017e-05, "loss": 1.9003, "step": 32372 }, { "epoch": 0.03, "learning_rate": 4.987158872495784e-05, "loss": 1.6653, "step": 32373 }, { "epoch": 0.03, "learning_rate": 4.987158077384e-05, "loss": 1.3378, "step": 32374 }, { "epoch": 0.03, "learning_rate": 4.9871572822476624e-05, "loss": 0.8218, "step": 32375 }, { "epoch": 0.03, "learning_rate": 4.987156487086773e-05, "loss": 0.9538, "step": 32376 }, { "epoch": 0.03, "learning_rate": 4.987155691901332e-05, "loss": 1.0719, "step": 32377 }, { "epoch": 0.03, "learning_rate": 4.987154896691338e-05, "loss": 1.1583, "step": 32378 }, { "epoch": 0.03, "learning_rate": 4.987154101456793e-05, "loss": 1.1669, "step": 32379 }, { "epoch": 0.03, "learning_rate": 4.987153306197695e-05, "loss": 1.0038, "step": 32380 }, { "epoch": 0.03, "learning_rate": 4.9871525109140446e-05, "loss": 0.8791, "step": 32381 }, { "epoch": 0.03, "learning_rate": 4.9871517156058426e-05, "loss": 1.1637, "step": 32382 }, { "epoch": 0.03, "learning_rate": 4.987150920273088e-05, "loss": 1.1116, "step": 32383 }, { "epoch": 0.03, "learning_rate": 4.987150124915781e-05, "loss": 0.9417, "step": 32384 }, { "epoch": 0.03, "learning_rate": 4.987149329533923e-05, "loss": 1.144, "step": 32385 }, { "epoch": 0.03, "learning_rate": 4.987148534127513e-05, "loss": 1.3253, "step": 32386 }, { "epoch": 0.03, "learning_rate": 4.987147738696551e-05, "loss": 1.043, "step": 32387 }, { "epoch": 0.03, "learning_rate": 4.9871469432410354e-05, "loss": 1.1765, "step": 32388 }, { "epoch": 0.03, "learning_rate": 4.987146147760969e-05, "loss": 1.188, "step": 32389 }, { "epoch": 0.03, "learning_rate": 4.98714535225635e-05, "loss": 1.0625, "step": 32390 }, { "epoch": 0.03, "learning_rate": 4.987144556727179e-05, "loss": 0.6026, "step": 32391 }, { "epoch": 0.03, "learning_rate": 4.987143761173456e-05, "loss": 0.6859, "step": 32392 }, { "epoch": 0.03, "learning_rate": 4.987142965595182e-05, "loss": 1.0853, "step": 32393 }, { "epoch": 0.03, "learning_rate": 4.987142169992355e-05, "loss": 1.6556, "step": 32394 }, { "epoch": 0.03, "learning_rate": 4.987141374364975e-05, "loss": 0.8779, "step": 32395 }, { "epoch": 0.03, "learning_rate": 4.9871405787130444e-05, "loss": 0.9329, "step": 32396 }, { "epoch": 0.03, "learning_rate": 4.987139783036562e-05, "loss": 0.9895, "step": 32397 }, { "epoch": 0.03, "learning_rate": 4.987138987335527e-05, "loss": 0.6712, "step": 32398 }, { "epoch": 0.03, "learning_rate": 4.987138191609939e-05, "loss": 0.8732, "step": 32399 }, { "epoch": 0.03, "learning_rate": 4.9871373958598006e-05, "loss": 1.1098, "step": 32400 }, { "epoch": 0.03, "learning_rate": 4.98713660008511e-05, "loss": 1.0436, "step": 32401 }, { "epoch": 0.03, "learning_rate": 4.987135804285867e-05, "loss": 1.1275, "step": 32402 }, { "epoch": 0.03, "learning_rate": 4.9871350084620714e-05, "loss": 0.5934, "step": 32403 }, { "epoch": 0.03, "learning_rate": 4.987134212613725e-05, "loss": 0.917, "step": 32404 }, { "epoch": 0.03, "learning_rate": 4.987133416740826e-05, "loss": 0.7873, "step": 32405 }, { "epoch": 0.03, "learning_rate": 4.987132620843376e-05, "loss": 1.0199, "step": 32406 }, { "epoch": 0.03, "learning_rate": 4.987131824921373e-05, "loss": 0.9801, "step": 32407 }, { "epoch": 0.03, "learning_rate": 4.9871310289748184e-05, "loss": 1.1615, "step": 32408 }, { "epoch": 0.03, "learning_rate": 4.9871302330037115e-05, "loss": 0.5716, "step": 32409 }, { "epoch": 0.03, "learning_rate": 4.9871294370080534e-05, "loss": 0.756, "step": 32410 }, { "epoch": 0.03, "learning_rate": 4.9871286409878437e-05, "loss": 1.0281, "step": 32411 }, { "epoch": 0.03, "learning_rate": 4.9871278449430814e-05, "loss": 0.9306, "step": 32412 }, { "epoch": 0.03, "learning_rate": 4.9871270488737674e-05, "loss": 1.134, "step": 32413 }, { "epoch": 0.03, "learning_rate": 4.987126252779901e-05, "loss": 1.1205, "step": 32414 }, { "epoch": 0.03, "learning_rate": 4.987125456661484e-05, "loss": 1.1233, "step": 32415 }, { "epoch": 0.03, "learning_rate": 4.987124660518514e-05, "loss": 0.9104, "step": 32416 }, { "epoch": 0.03, "learning_rate": 4.987123864350993e-05, "loss": 0.9661, "step": 32417 }, { "epoch": 0.03, "learning_rate": 4.98712306815892e-05, "loss": 0.9432, "step": 32418 }, { "epoch": 0.03, "learning_rate": 4.9871222719422946e-05, "loss": 0.9023, "step": 32419 }, { "epoch": 0.03, "learning_rate": 4.9871214757011176e-05, "loss": 1.2718, "step": 32420 }, { "epoch": 0.03, "learning_rate": 4.987120679435389e-05, "loss": 0.9453, "step": 32421 }, { "epoch": 0.03, "learning_rate": 4.9871198831451084e-05, "loss": 1.2669, "step": 32422 }, { "epoch": 0.03, "learning_rate": 4.987119086830276e-05, "loss": 1.1683, "step": 32423 }, { "epoch": 0.03, "learning_rate": 4.987118290490892e-05, "loss": 0.941, "step": 32424 }, { "epoch": 0.03, "learning_rate": 4.987117494126956e-05, "loss": 0.9729, "step": 32425 }, { "epoch": 0.03, "learning_rate": 4.987116697738468e-05, "loss": 0.7289, "step": 32426 }, { "epoch": 0.03, "learning_rate": 4.987115901325429e-05, "loss": 1.3169, "step": 32427 }, { "epoch": 0.03, "learning_rate": 4.987115104887837e-05, "loss": 0.993, "step": 32428 }, { "epoch": 0.03, "learning_rate": 4.9871143084256944e-05, "loss": 0.9012, "step": 32429 }, { "epoch": 0.03, "learning_rate": 4.987113511939e-05, "loss": 1.0473, "step": 32430 }, { "epoch": 0.03, "learning_rate": 4.987112715427753e-05, "loss": 0.5538, "step": 32431 }, { "epoch": 0.03, "learning_rate": 4.9871119188919555e-05, "loss": 1.0485, "step": 32432 }, { "epoch": 0.03, "learning_rate": 4.987111122331605e-05, "loss": 0.9065, "step": 32433 }, { "epoch": 0.03, "learning_rate": 4.987110325746704e-05, "loss": 0.9184, "step": 32434 }, { "epoch": 0.03, "learning_rate": 4.987109529137251e-05, "loss": 0.3177, "step": 32435 }, { "epoch": 0.03, "learning_rate": 4.987108732503245e-05, "loss": 0.2342, "step": 32436 }, { "epoch": 0.03, "learning_rate": 4.987107935844689e-05, "loss": 0.1621, "step": 32437 }, { "epoch": 0.03, "learning_rate": 4.98710713916158e-05, "loss": 0.1623, "step": 32438 }, { "epoch": 0.03, "learning_rate": 4.98710634245392e-05, "loss": 0.2278, "step": 32439 }, { "epoch": 0.03, "learning_rate": 4.987105545721708e-05, "loss": 0.343, "step": 32440 }, { "epoch": 0.03, "learning_rate": 4.987104748964945e-05, "loss": 0.6608, "step": 32441 }, { "epoch": 0.03, "learning_rate": 4.98710395218363e-05, "loss": 0.2037, "step": 32442 }, { "epoch": 0.03, "learning_rate": 4.9871031553777634e-05, "loss": 0.288, "step": 32443 }, { "epoch": 0.03, "learning_rate": 4.9871023585473456e-05, "loss": 0.2996, "step": 32444 }, { "epoch": 0.03, "learning_rate": 4.9871015616923754e-05, "loss": 0.2001, "step": 32445 }, { "epoch": 0.03, "learning_rate": 4.987100764812854e-05, "loss": 0.1922, "step": 32446 }, { "epoch": 0.03, "learning_rate": 4.98709996790878e-05, "loss": 0.3918, "step": 32447 }, { "epoch": 0.03, "learning_rate": 4.9870991709801554e-05, "loss": 1.2726, "step": 32448 }, { "epoch": 0.03, "learning_rate": 4.9870983740269795e-05, "loss": 1.1299, "step": 32449 }, { "epoch": 0.03, "learning_rate": 4.987097577049251e-05, "loss": 1.1291, "step": 32450 }, { "epoch": 0.03, "learning_rate": 4.987096780046972e-05, "loss": 0.5883, "step": 32451 }, { "epoch": 0.03, "learning_rate": 4.9870959830201405e-05, "loss": 0.87, "step": 32452 }, { "epoch": 0.03, "learning_rate": 4.987095185968758e-05, "loss": 1.0524, "step": 32453 }, { "epoch": 0.03, "learning_rate": 4.987094388892824e-05, "loss": 0.885, "step": 32454 }, { "epoch": 0.03, "learning_rate": 4.987093591792338e-05, "loss": 1.017, "step": 32455 }, { "epoch": 0.03, "learning_rate": 4.987092794667301e-05, "loss": 0.9161, "step": 32456 }, { "epoch": 0.03, "learning_rate": 4.987091997517712e-05, "loss": 1.0234, "step": 32457 }, { "epoch": 0.03, "learning_rate": 4.987091200343572e-05, "loss": 0.832, "step": 32458 }, { "epoch": 0.03, "learning_rate": 4.9870904031448797e-05, "loss": 1.2359, "step": 32459 }, { "epoch": 0.03, "learning_rate": 4.9870896059216364e-05, "loss": 1.3251, "step": 32460 }, { "epoch": 0.03, "learning_rate": 4.9870888086738414e-05, "loss": 1.4955, "step": 32461 }, { "epoch": 0.03, "learning_rate": 4.9870880114014954e-05, "loss": 1.0375, "step": 32462 }, { "epoch": 0.03, "learning_rate": 4.9870872141045975e-05, "loss": 1.1083, "step": 32463 }, { "epoch": 0.03, "learning_rate": 4.987086416783148e-05, "loss": 1.0148, "step": 32464 }, { "epoch": 0.03, "learning_rate": 4.987085619437147e-05, "loss": 0.6818, "step": 32465 }, { "epoch": 0.03, "learning_rate": 4.987084822066594e-05, "loss": 0.9898, "step": 32466 }, { "epoch": 0.03, "learning_rate": 4.9870840246714914e-05, "loss": 0.9855, "step": 32467 }, { "epoch": 0.03, "learning_rate": 4.987083227251836e-05, "loss": 1.038, "step": 32468 }, { "epoch": 0.03, "learning_rate": 4.987082429807629e-05, "loss": 1.0667, "step": 32469 }, { "epoch": 0.03, "learning_rate": 4.9870816323388716e-05, "loss": 0.93, "step": 32470 }, { "epoch": 0.03, "learning_rate": 4.9870808348455625e-05, "loss": 1.0743, "step": 32471 }, { "epoch": 0.03, "learning_rate": 4.987080037327701e-05, "loss": 0.7369, "step": 32472 }, { "epoch": 0.03, "learning_rate": 4.987079239785288e-05, "loss": 0.6631, "step": 32473 }, { "epoch": 0.03, "learning_rate": 4.987078442218325e-05, "loss": 1.0776, "step": 32474 }, { "epoch": 0.03, "learning_rate": 4.9870776446268095e-05, "loss": 1.2047, "step": 32475 }, { "epoch": 0.03, "learning_rate": 4.987076847010743e-05, "loss": 0.9493, "step": 32476 }, { "epoch": 0.03, "learning_rate": 4.987076049370126e-05, "loss": 1.1533, "step": 32477 }, { "epoch": 0.03, "learning_rate": 4.987075251704957e-05, "loss": 1.0636, "step": 32478 }, { "epoch": 0.03, "learning_rate": 4.987074454015236e-05, "loss": 0.8898, "step": 32479 }, { "epoch": 0.03, "learning_rate": 4.987073656300964e-05, "loss": 0.7711, "step": 32480 }, { "epoch": 0.03, "learning_rate": 4.987072858562141e-05, "loss": 1.0587, "step": 32481 }, { "epoch": 0.03, "learning_rate": 4.9870720607987664e-05, "loss": 1.0777, "step": 32482 }, { "epoch": 0.03, "learning_rate": 4.98707126301084e-05, "loss": 0.9302, "step": 32483 }, { "epoch": 0.03, "learning_rate": 4.987070465198363e-05, "loss": 0.5962, "step": 32484 }, { "epoch": 0.03, "learning_rate": 4.9870696673613345e-05, "loss": 0.435, "step": 32485 }, { "epoch": 0.03, "learning_rate": 4.9870688694997544e-05, "loss": 0.7689, "step": 32486 }, { "epoch": 0.03, "learning_rate": 4.987068071613623e-05, "loss": 0.5232, "step": 32487 }, { "epoch": 0.03, "learning_rate": 4.987067273702941e-05, "loss": 0.4091, "step": 32488 }, { "epoch": 0.03, "learning_rate": 4.9870664757677075e-05, "loss": 0.4924, "step": 32489 }, { "epoch": 0.03, "learning_rate": 4.987065677807923e-05, "loss": 0.4981, "step": 32490 }, { "epoch": 0.03, "learning_rate": 4.987064879823586e-05, "loss": 0.3439, "step": 32491 }, { "epoch": 0.03, "learning_rate": 4.987064081814699e-05, "loss": 0.3957, "step": 32492 }, { "epoch": 0.03, "learning_rate": 4.98706328378126e-05, "loss": 0.987, "step": 32493 }, { "epoch": 0.03, "learning_rate": 4.98706248572327e-05, "loss": 1.0982, "step": 32494 }, { "epoch": 0.03, "learning_rate": 4.9870616876407284e-05, "loss": 0.975, "step": 32495 }, { "epoch": 0.03, "learning_rate": 4.987060889533636e-05, "loss": 0.7182, "step": 32496 }, { "epoch": 0.03, "learning_rate": 4.987060091401993e-05, "loss": 0.8848, "step": 32497 }, { "epoch": 0.03, "learning_rate": 4.9870592932457976e-05, "loss": 0.8316, "step": 32498 }, { "epoch": 0.03, "learning_rate": 4.9870584950650514e-05, "loss": 0.8624, "step": 32499 }, { "epoch": 0.03, "learning_rate": 4.987057696859754e-05, "loss": 1.244, "step": 32500 }, { "epoch": 0.03, "eval_loss": 1.0792115926742554, "eval_runtime": 126.9385, "eval_samples_per_second": 10.911, "eval_steps_per_second": 5.459, "step": 32500 }, { "epoch": 0.03, "learning_rate": 4.987056898629906e-05, "loss": 0.9262, "step": 32501 }, { "epoch": 0.03, "learning_rate": 4.987056100375506e-05, "loss": 1.0586, "step": 32502 }, { "epoch": 0.03, "learning_rate": 4.987055302096556e-05, "loss": 0.9858, "step": 32503 }, { "epoch": 0.03, "learning_rate": 4.987054503793054e-05, "loss": 0.9738, "step": 32504 }, { "epoch": 0.03, "learning_rate": 4.987053705465e-05, "loss": 1.1497, "step": 32505 }, { "epoch": 0.03, "learning_rate": 4.987052907112396e-05, "loss": 1.0458, "step": 32506 }, { "epoch": 0.03, "learning_rate": 4.987052108735241e-05, "loss": 0.6923, "step": 32507 }, { "epoch": 0.03, "learning_rate": 4.987051310333534e-05, "loss": 0.5825, "step": 32508 }, { "epoch": 0.03, "learning_rate": 4.9870505119072764e-05, "loss": 1.2464, "step": 32509 }, { "epoch": 0.03, "learning_rate": 4.987049713456468e-05, "loss": 0.9263, "step": 32510 }, { "epoch": 0.03, "learning_rate": 4.987048914981108e-05, "loss": 0.8394, "step": 32511 }, { "epoch": 0.03, "learning_rate": 4.987048116481197e-05, "loss": 1.0763, "step": 32512 }, { "epoch": 0.03, "learning_rate": 4.987047317956735e-05, "loss": 1.3144, "step": 32513 }, { "epoch": 0.03, "learning_rate": 4.987046519407722e-05, "loss": 1.6761, "step": 32514 }, { "epoch": 0.03, "learning_rate": 4.9870457208341576e-05, "loss": 0.7291, "step": 32515 }, { "epoch": 0.03, "learning_rate": 4.987044922236042e-05, "loss": 0.8246, "step": 32516 }, { "epoch": 0.03, "learning_rate": 4.987044123613376e-05, "loss": 0.8682, "step": 32517 }, { "epoch": 0.03, "learning_rate": 4.9870433249661585e-05, "loss": 1.1686, "step": 32518 }, { "epoch": 0.03, "learning_rate": 4.98704252629439e-05, "loss": 1.0361, "step": 32519 }, { "epoch": 0.03, "learning_rate": 4.9870417275980704e-05, "loss": 0.9135, "step": 32520 }, { "epoch": 0.03, "learning_rate": 4.9870409288772005e-05, "loss": 1.3881, "step": 32521 }, { "epoch": 0.03, "learning_rate": 4.987040130131778e-05, "loss": 0.9723, "step": 32522 }, { "epoch": 0.03, "learning_rate": 4.987039331361806e-05, "loss": 0.9218, "step": 32523 }, { "epoch": 0.03, "learning_rate": 4.987038532567282e-05, "loss": 0.8795, "step": 32524 }, { "epoch": 0.03, "learning_rate": 4.987037733748208e-05, "loss": 1.0293, "step": 32525 }, { "epoch": 0.03, "learning_rate": 4.987036934904582e-05, "loss": 0.9669, "step": 32526 }, { "epoch": 0.03, "learning_rate": 4.987036136036406e-05, "loss": 0.9861, "step": 32527 }, { "epoch": 0.03, "learning_rate": 4.987035337143678e-05, "loss": 1.135, "step": 32528 }, { "epoch": 0.03, "learning_rate": 4.9870345382263996e-05, "loss": 1.1283, "step": 32529 }, { "epoch": 0.03, "learning_rate": 4.98703373928457e-05, "loss": 1.0967, "step": 32530 }, { "epoch": 0.03, "learning_rate": 4.9870329403181905e-05, "loss": 1.3808, "step": 32531 }, { "epoch": 0.03, "learning_rate": 4.9870321413272586e-05, "loss": 1.414, "step": 32532 }, { "epoch": 0.03, "learning_rate": 4.987031342311776e-05, "loss": 1.3275, "step": 32533 }, { "epoch": 0.03, "learning_rate": 4.9870305432717436e-05, "loss": 1.0497, "step": 32534 }, { "epoch": 0.03, "learning_rate": 4.987029744207159e-05, "loss": 0.688, "step": 32535 }, { "epoch": 0.03, "learning_rate": 4.9870289451180244e-05, "loss": 1.3293, "step": 32536 }, { "epoch": 0.03, "learning_rate": 4.9870281460043384e-05, "loss": 0.9543, "step": 32537 }, { "epoch": 0.03, "learning_rate": 4.9870273468661014e-05, "loss": 1.0291, "step": 32538 }, { "epoch": 0.03, "learning_rate": 4.987026547703314e-05, "loss": 1.1207, "step": 32539 }, { "epoch": 0.03, "learning_rate": 4.9870257485159755e-05, "loss": 0.9516, "step": 32540 }, { "epoch": 0.03, "learning_rate": 4.987024949304086e-05, "loss": 0.6811, "step": 32541 }, { "epoch": 0.03, "learning_rate": 4.987024150067646e-05, "loss": 0.7636, "step": 32542 }, { "epoch": 0.03, "learning_rate": 4.987023350806654e-05, "loss": 1.0012, "step": 32543 }, { "epoch": 0.03, "learning_rate": 4.987022551521112e-05, "loss": 1.3992, "step": 32544 }, { "epoch": 0.03, "learning_rate": 4.9870217522110196e-05, "loss": 0.877, "step": 32545 }, { "epoch": 0.03, "learning_rate": 4.987020952876376e-05, "loss": 1.0162, "step": 32546 }, { "epoch": 0.03, "learning_rate": 4.987020153517181e-05, "loss": 0.8529, "step": 32547 }, { "epoch": 0.03, "learning_rate": 4.987019354133436e-05, "loss": 0.9245, "step": 32548 }, { "epoch": 0.03, "learning_rate": 4.98701855472514e-05, "loss": 1.2348, "step": 32549 }, { "epoch": 0.03, "learning_rate": 4.9870177552922934e-05, "loss": 0.7148, "step": 32550 }, { "epoch": 0.03, "learning_rate": 4.987016955834896e-05, "loss": 1.3742, "step": 32551 }, { "epoch": 0.03, "learning_rate": 4.987016156352947e-05, "loss": 1.9925, "step": 32552 }, { "epoch": 0.03, "learning_rate": 4.9870153568464486e-05, "loss": 1.1859, "step": 32553 }, { "epoch": 0.03, "learning_rate": 4.9870145573153983e-05, "loss": 1.0893, "step": 32554 }, { "epoch": 0.03, "learning_rate": 4.987013757759798e-05, "loss": 0.8206, "step": 32555 }, { "epoch": 0.03, "learning_rate": 4.987012958179645e-05, "loss": 1.0804, "step": 32556 }, { "epoch": 0.03, "learning_rate": 4.987012158574943e-05, "loss": 1.2362, "step": 32557 }, { "epoch": 0.03, "learning_rate": 4.98701135894569e-05, "loss": 1.3295, "step": 32558 }, { "epoch": 0.03, "learning_rate": 4.987010559291887e-05, "loss": 0.9639, "step": 32559 }, { "epoch": 0.03, "learning_rate": 4.9870097596135324e-05, "loss": 0.6525, "step": 32560 }, { "epoch": 0.03, "learning_rate": 4.9870089599106274e-05, "loss": 0.3259, "step": 32561 }, { "epoch": 0.03, "learning_rate": 4.9870081601831705e-05, "loss": 0.2271, "step": 32562 }, { "epoch": 0.03, "learning_rate": 4.9870073604311646e-05, "loss": 0.2366, "step": 32563 }, { "epoch": 0.03, "learning_rate": 4.987006560654608e-05, "loss": 0.6203, "step": 32564 }, { "epoch": 0.03, "learning_rate": 4.987005760853499e-05, "loss": 0.982, "step": 32565 }, { "epoch": 0.03, "learning_rate": 4.9870049610278405e-05, "loss": 0.9879, "step": 32566 }, { "epoch": 0.03, "learning_rate": 4.9870041611776316e-05, "loss": 0.8842, "step": 32567 }, { "epoch": 0.03, "learning_rate": 4.987003361302872e-05, "loss": 0.9936, "step": 32568 }, { "epoch": 0.03, "learning_rate": 4.987002561403561e-05, "loss": 1.0967, "step": 32569 }, { "epoch": 0.03, "learning_rate": 4.9870017614797e-05, "loss": 1.0041, "step": 32570 }, { "epoch": 0.03, "learning_rate": 4.987000961531288e-05, "loss": 1.0296, "step": 32571 }, { "epoch": 0.03, "learning_rate": 4.987000161558326e-05, "loss": 1.1526, "step": 32572 }, { "epoch": 0.03, "learning_rate": 4.9869993615608126e-05, "loss": 1.219, "step": 32573 }, { "epoch": 0.03, "learning_rate": 4.986998561538749e-05, "loss": 1.0665, "step": 32574 }, { "epoch": 0.03, "learning_rate": 4.986997761492135e-05, "loss": 0.7738, "step": 32575 }, { "epoch": 0.03, "learning_rate": 4.9869969614209696e-05, "loss": 0.7791, "step": 32576 }, { "epoch": 0.03, "learning_rate": 4.986996161325254e-05, "loss": 0.913, "step": 32577 }, { "epoch": 0.03, "learning_rate": 4.986995361204988e-05, "loss": 1.0539, "step": 32578 }, { "epoch": 0.03, "learning_rate": 4.986994561060171e-05, "loss": 0.9641, "step": 32579 }, { "epoch": 0.03, "learning_rate": 4.986993760890804e-05, "loss": 0.9887, "step": 32580 }, { "epoch": 0.03, "learning_rate": 4.986992960696886e-05, "loss": 0.973, "step": 32581 }, { "epoch": 0.03, "learning_rate": 4.986992160478418e-05, "loss": 1.0166, "step": 32582 }, { "epoch": 0.03, "learning_rate": 4.986991360235399e-05, "loss": 0.5874, "step": 32583 }, { "epoch": 0.03, "learning_rate": 4.98699055996783e-05, "loss": 1.0478, "step": 32584 }, { "epoch": 0.03, "learning_rate": 4.98698975967571e-05, "loss": 0.9003, "step": 32585 }, { "epoch": 0.03, "learning_rate": 4.9869889593590395e-05, "loss": 1.1639, "step": 32586 }, { "epoch": 0.03, "learning_rate": 4.9869881590178185e-05, "loss": 1.256, "step": 32587 }, { "epoch": 0.03, "learning_rate": 4.986987358652048e-05, "loss": 1.3066, "step": 32588 }, { "epoch": 0.03, "learning_rate": 4.986986558261725e-05, "loss": 0.8196, "step": 32589 }, { "epoch": 0.03, "learning_rate": 4.9869857578468525e-05, "loss": 1.2007, "step": 32590 }, { "epoch": 0.03, "learning_rate": 4.98698495740743e-05, "loss": 1.0245, "step": 32591 }, { "epoch": 0.03, "learning_rate": 4.986984156943457e-05, "loss": 0.8127, "step": 32592 }, { "epoch": 0.03, "learning_rate": 4.986983356454933e-05, "loss": 1.0742, "step": 32593 }, { "epoch": 0.03, "learning_rate": 4.986982555941859e-05, "loss": 0.8032, "step": 32594 }, { "epoch": 0.03, "learning_rate": 4.986981755404234e-05, "loss": 0.8665, "step": 32595 }, { "epoch": 0.03, "learning_rate": 4.986980954842059e-05, "loss": 0.9234, "step": 32596 }, { "epoch": 0.03, "learning_rate": 4.986980154255333e-05, "loss": 1.0193, "step": 32597 }, { "epoch": 0.03, "learning_rate": 4.986979353644057e-05, "loss": 1.1294, "step": 32598 }, { "epoch": 0.03, "learning_rate": 4.986978553008231e-05, "loss": 0.9738, "step": 32599 }, { "epoch": 0.03, "learning_rate": 4.9869777523478544e-05, "loss": 1.1541, "step": 32600 }, { "epoch": 0.03, "learning_rate": 4.986976951662927e-05, "loss": 1.0368, "step": 32601 }, { "epoch": 0.03, "learning_rate": 4.9869761509534494e-05, "loss": 1.0119, "step": 32602 }, { "epoch": 0.03, "learning_rate": 4.986975350219422e-05, "loss": 0.7952, "step": 32603 }, { "epoch": 0.03, "learning_rate": 4.9869745494608434e-05, "loss": 0.9839, "step": 32604 }, { "epoch": 0.03, "learning_rate": 4.9869737486777145e-05, "loss": 0.9227, "step": 32605 }, { "epoch": 0.03, "learning_rate": 4.986972947870035e-05, "loss": 1.0534, "step": 32606 }, { "epoch": 0.03, "learning_rate": 4.986972147037806e-05, "loss": 1.3039, "step": 32607 }, { "epoch": 0.03, "learning_rate": 4.986971346181026e-05, "loss": 1.5221, "step": 32608 }, { "epoch": 0.03, "learning_rate": 4.986970545299696e-05, "loss": 1.16, "step": 32609 }, { "epoch": 0.03, "learning_rate": 4.986969744393816e-05, "loss": 1.1248, "step": 32610 }, { "epoch": 0.03, "learning_rate": 4.986968943463385e-05, "loss": 1.387, "step": 32611 }, { "epoch": 0.03, "learning_rate": 4.986968142508404e-05, "loss": 1.2933, "step": 32612 }, { "epoch": 0.03, "learning_rate": 4.986967341528873e-05, "loss": 0.8602, "step": 32613 }, { "epoch": 0.03, "learning_rate": 4.986966540524791e-05, "loss": 1.1479, "step": 32614 }, { "epoch": 0.03, "learning_rate": 4.9869657394961594e-05, "loss": 0.6923, "step": 32615 }, { "epoch": 0.03, "learning_rate": 4.9869649384429774e-05, "loss": 0.9425, "step": 32616 }, { "epoch": 0.03, "learning_rate": 4.986964137365245e-05, "loss": 0.8861, "step": 32617 }, { "epoch": 0.03, "learning_rate": 4.986963336262962e-05, "loss": 0.9323, "step": 32618 }, { "epoch": 0.03, "learning_rate": 4.986962535136129e-05, "loss": 1.0681, "step": 32619 }, { "epoch": 0.03, "learning_rate": 4.9869617339847454e-05, "loss": 0.8883, "step": 32620 }, { "epoch": 0.03, "learning_rate": 4.986960932808812e-05, "loss": 1.028, "step": 32621 }, { "epoch": 0.03, "learning_rate": 4.9869601316083284e-05, "loss": 0.8639, "step": 32622 }, { "epoch": 0.03, "learning_rate": 4.986959330383295e-05, "loss": 1.3352, "step": 32623 }, { "epoch": 0.03, "learning_rate": 4.9869585291337104e-05, "loss": 0.9881, "step": 32624 }, { "epoch": 0.03, "learning_rate": 4.986957727859576e-05, "loss": 1.0773, "step": 32625 }, { "epoch": 0.03, "learning_rate": 4.9869569265608915e-05, "loss": 1.25, "step": 32626 }, { "epoch": 0.03, "learning_rate": 4.986956125237657e-05, "loss": 1.2224, "step": 32627 }, { "epoch": 0.03, "learning_rate": 4.986955323889872e-05, "loss": 1.2481, "step": 32628 }, { "epoch": 0.03, "learning_rate": 4.986954522517537e-05, "loss": 1.0396, "step": 32629 }, { "epoch": 0.03, "learning_rate": 4.9869537211206516e-05, "loss": 0.6566, "step": 32630 }, { "epoch": 0.03, "learning_rate": 4.986952919699216e-05, "loss": 0.9755, "step": 32631 }, { "epoch": 0.03, "learning_rate": 4.9869521182532305e-05, "loss": 0.9495, "step": 32632 }, { "epoch": 0.03, "learning_rate": 4.986951316782695e-05, "loss": 0.5497, "step": 32633 }, { "epoch": 0.03, "learning_rate": 4.986950515287609e-05, "loss": 0.3568, "step": 32634 }, { "epoch": 0.03, "learning_rate": 4.986949713767973e-05, "loss": 0.3079, "step": 32635 }, { "epoch": 0.03, "learning_rate": 4.986948912223788e-05, "loss": 0.3223, "step": 32636 }, { "epoch": 0.03, "learning_rate": 4.986948110655051e-05, "loss": 0.274, "step": 32637 }, { "epoch": 0.03, "learning_rate": 4.9869473090617645e-05, "loss": 0.2427, "step": 32638 }, { "epoch": 0.03, "learning_rate": 4.986946507443929e-05, "loss": 0.2892, "step": 32639 }, { "epoch": 0.03, "learning_rate": 4.9869457058015424e-05, "loss": 0.233, "step": 32640 }, { "epoch": 0.03, "learning_rate": 4.986944904134606e-05, "loss": 0.26, "step": 32641 }, { "epoch": 0.03, "learning_rate": 4.986944102443119e-05, "loss": 0.2979, "step": 32642 }, { "epoch": 0.03, "learning_rate": 4.986943300727082e-05, "loss": 0.2401, "step": 32643 }, { "epoch": 0.03, "learning_rate": 4.9869424989864955e-05, "loss": 0.2478, "step": 32644 }, { "epoch": 0.03, "learning_rate": 4.986941697221359e-05, "loss": 0.2392, "step": 32645 }, { "epoch": 0.03, "learning_rate": 4.986940895431672e-05, "loss": 0.2948, "step": 32646 }, { "epoch": 0.03, "learning_rate": 4.986940093617435e-05, "loss": 0.3184, "step": 32647 }, { "epoch": 0.03, "learning_rate": 4.986939291778648e-05, "loss": 0.285, "step": 32648 }, { "epoch": 0.03, "learning_rate": 4.986938489915312e-05, "loss": 0.2262, "step": 32649 }, { "epoch": 0.03, "learning_rate": 4.986937688027424e-05, "loss": 0.2522, "step": 32650 }, { "epoch": 0.03, "learning_rate": 4.9869368861149876e-05, "loss": 0.5233, "step": 32651 }, { "epoch": 0.03, "learning_rate": 4.9869360841780006e-05, "loss": 0.7757, "step": 32652 }, { "epoch": 0.03, "learning_rate": 4.986935282216464e-05, "loss": 0.9077, "step": 32653 }, { "epoch": 0.03, "learning_rate": 4.986934480230376e-05, "loss": 0.8529, "step": 32654 }, { "epoch": 0.03, "learning_rate": 4.98693367821974e-05, "loss": 1.1712, "step": 32655 }, { "epoch": 0.03, "learning_rate": 4.986932876184553e-05, "loss": 0.6823, "step": 32656 }, { "epoch": 0.03, "learning_rate": 4.986932074124816e-05, "loss": 0.6295, "step": 32657 }, { "epoch": 0.03, "learning_rate": 4.986931272040529e-05, "loss": 1.1555, "step": 32658 }, { "epoch": 0.03, "learning_rate": 4.986930469931693e-05, "loss": 1.103, "step": 32659 }, { "epoch": 0.03, "learning_rate": 4.986929667798307e-05, "loss": 0.9541, "step": 32660 }, { "epoch": 0.03, "learning_rate": 4.9869288656403704e-05, "loss": 0.9156, "step": 32661 }, { "epoch": 0.03, "learning_rate": 4.9869280634578835e-05, "loss": 1.0801, "step": 32662 }, { "epoch": 0.03, "learning_rate": 4.9869272612508474e-05, "loss": 1.1363, "step": 32663 }, { "epoch": 0.03, "learning_rate": 4.986926459019261e-05, "loss": 0.8403, "step": 32664 }, { "epoch": 0.03, "learning_rate": 4.986925656763125e-05, "loss": 0.892, "step": 32665 }, { "epoch": 0.03, "learning_rate": 4.986924854482439e-05, "loss": 0.9242, "step": 32666 }, { "epoch": 0.03, "learning_rate": 4.9869240521772034e-05, "loss": 1.2257, "step": 32667 }, { "epoch": 0.03, "learning_rate": 4.9869232498474174e-05, "loss": 0.9471, "step": 32668 }, { "epoch": 0.03, "learning_rate": 4.986922447493082e-05, "loss": 0.9777, "step": 32669 }, { "epoch": 0.03, "learning_rate": 4.986921645114196e-05, "loss": 0.9246, "step": 32670 }, { "epoch": 0.03, "learning_rate": 4.986920842710761e-05, "loss": 1.1279, "step": 32671 }, { "epoch": 0.03, "learning_rate": 4.986920040282776e-05, "loss": 1.1623, "step": 32672 }, { "epoch": 0.03, "learning_rate": 4.986919237830241e-05, "loss": 1.264, "step": 32673 }, { "epoch": 0.03, "learning_rate": 4.986918435353156e-05, "loss": 1.0087, "step": 32674 }, { "epoch": 0.03, "learning_rate": 4.986917632851522e-05, "loss": 1.1502, "step": 32675 }, { "epoch": 0.03, "learning_rate": 4.986916830325338e-05, "loss": 1.2548, "step": 32676 }, { "epoch": 0.03, "learning_rate": 4.986916027774603e-05, "loss": 1.0537, "step": 32677 }, { "epoch": 0.03, "learning_rate": 4.986915225199319e-05, "loss": 0.9079, "step": 32678 }, { "epoch": 0.03, "learning_rate": 4.9869144225994855e-05, "loss": 0.8452, "step": 32679 }, { "epoch": 0.03, "learning_rate": 4.986913619975102e-05, "loss": 0.7205, "step": 32680 }, { "epoch": 0.03, "learning_rate": 4.9869128173261684e-05, "loss": 1.2636, "step": 32681 }, { "epoch": 0.03, "learning_rate": 4.986912014652686e-05, "loss": 0.9043, "step": 32682 }, { "epoch": 0.03, "learning_rate": 4.986911211954653e-05, "loss": 1.1262, "step": 32683 }, { "epoch": 0.03, "learning_rate": 4.986910409232071e-05, "loss": 0.7153, "step": 32684 }, { "epoch": 0.03, "learning_rate": 4.986909606484939e-05, "loss": 0.8261, "step": 32685 }, { "epoch": 0.03, "learning_rate": 4.986908803713256e-05, "loss": 0.7673, "step": 32686 }, { "epoch": 0.03, "learning_rate": 4.986908000917024e-05, "loss": 0.9044, "step": 32687 }, { "epoch": 0.03, "learning_rate": 4.9869071980962437e-05, "loss": 0.7701, "step": 32688 }, { "epoch": 0.03, "learning_rate": 4.986906395250912e-05, "loss": 0.8163, "step": 32689 }, { "epoch": 0.03, "learning_rate": 4.9869055923810314e-05, "loss": 1.1761, "step": 32690 }, { "epoch": 0.03, "learning_rate": 4.9869047894866004e-05, "loss": 1.1587, "step": 32691 }, { "epoch": 0.03, "learning_rate": 4.98690398656762e-05, "loss": 0.8697, "step": 32692 }, { "epoch": 0.03, "learning_rate": 4.986903183624091e-05, "loss": 0.6402, "step": 32693 }, { "epoch": 0.03, "learning_rate": 4.986902380656011e-05, "loss": 1.0906, "step": 32694 }, { "epoch": 0.03, "learning_rate": 4.9869015776633824e-05, "loss": 1.1974, "step": 32695 }, { "epoch": 0.03, "learning_rate": 4.986900774646204e-05, "loss": 1.5868, "step": 32696 }, { "epoch": 0.03, "learning_rate": 4.9868999716044754e-05, "loss": 1.1387, "step": 32697 }, { "epoch": 0.03, "learning_rate": 4.986899168538197e-05, "loss": 1.028, "step": 32698 }, { "epoch": 0.03, "learning_rate": 4.9868983654473695e-05, "loss": 0.8013, "step": 32699 }, { "epoch": 0.03, "learning_rate": 4.986897562331992e-05, "loss": 0.992, "step": 32700 }, { "epoch": 0.03, "learning_rate": 4.986896759192066e-05, "loss": 1.1099, "step": 32701 }, { "epoch": 0.03, "learning_rate": 4.986895956027589e-05, "loss": 0.9473, "step": 32702 }, { "epoch": 0.03, "learning_rate": 4.986895152838563e-05, "loss": 0.9935, "step": 32703 }, { "epoch": 0.03, "learning_rate": 4.9868943496249875e-05, "loss": 1.0243, "step": 32704 }, { "epoch": 0.03, "learning_rate": 4.9868935463868624e-05, "loss": 1.0594, "step": 32705 }, { "epoch": 0.03, "learning_rate": 4.9868927431241875e-05, "loss": 0.8347, "step": 32706 }, { "epoch": 0.03, "learning_rate": 4.986891939836963e-05, "loss": 1.0677, "step": 32707 }, { "epoch": 0.03, "learning_rate": 4.986891136525189e-05, "loss": 0.8836, "step": 32708 }, { "epoch": 0.03, "learning_rate": 4.986890333188866e-05, "loss": 1.0941, "step": 32709 }, { "epoch": 0.03, "learning_rate": 4.986889529827993e-05, "loss": 1.3279, "step": 32710 }, { "epoch": 0.03, "learning_rate": 4.98688872644257e-05, "loss": 1.008, "step": 32711 }, { "epoch": 0.03, "learning_rate": 4.986887923032598e-05, "loss": 1.1716, "step": 32712 }, { "epoch": 0.03, "learning_rate": 4.986887119598077e-05, "loss": 0.7559, "step": 32713 }, { "epoch": 0.03, "learning_rate": 4.986886316139006e-05, "loss": 0.7059, "step": 32714 }, { "epoch": 0.03, "learning_rate": 4.986885512655385e-05, "loss": 0.4952, "step": 32715 }, { "epoch": 0.03, "learning_rate": 4.9868847091472146e-05, "loss": 0.565, "step": 32716 }, { "epoch": 0.03, "learning_rate": 4.9868839056144955e-05, "loss": 0.5112, "step": 32717 }, { "epoch": 0.03, "learning_rate": 4.986883102057227e-05, "loss": 0.4833, "step": 32718 }, { "epoch": 0.03, "learning_rate": 4.986882298475408e-05, "loss": 0.3337, "step": 32719 }, { "epoch": 0.03, "learning_rate": 4.9868814948690404e-05, "loss": 1.2476, "step": 32720 }, { "epoch": 0.03, "learning_rate": 4.986880691238123e-05, "loss": 0.8517, "step": 32721 }, { "epoch": 0.03, "learning_rate": 4.9868798875826566e-05, "loss": 0.8925, "step": 32722 }, { "epoch": 0.03, "learning_rate": 4.9868790839026404e-05, "loss": 1.1276, "step": 32723 }, { "epoch": 0.03, "learning_rate": 4.986878280198075e-05, "loss": 1.0944, "step": 32724 }, { "epoch": 0.03, "learning_rate": 4.9868774764689596e-05, "loss": 1.052, "step": 32725 }, { "epoch": 0.03, "learning_rate": 4.986876672715296e-05, "loss": 1.0775, "step": 32726 }, { "epoch": 0.03, "learning_rate": 4.986875868937081e-05, "loss": 1.1323, "step": 32727 }, { "epoch": 0.03, "learning_rate": 4.9868750651343186e-05, "loss": 0.7434, "step": 32728 }, { "epoch": 0.03, "learning_rate": 4.9868742613070054e-05, "loss": 0.9964, "step": 32729 }, { "epoch": 0.03, "learning_rate": 4.986873457455143e-05, "loss": 0.661, "step": 32730 }, { "epoch": 0.03, "learning_rate": 4.9868726535787327e-05, "loss": 0.9515, "step": 32731 }, { "epoch": 0.03, "learning_rate": 4.9868718496777724e-05, "loss": 0.9468, "step": 32732 }, { "epoch": 0.03, "learning_rate": 4.986871045752262e-05, "loss": 1.0025, "step": 32733 }, { "epoch": 0.03, "learning_rate": 4.9868702418022026e-05, "loss": 1.0426, "step": 32734 }, { "epoch": 0.03, "learning_rate": 4.986869437827594e-05, "loss": 0.8334, "step": 32735 }, { "epoch": 0.03, "learning_rate": 4.986868633828435e-05, "loss": 0.8879, "step": 32736 }, { "epoch": 0.03, "learning_rate": 4.9868678298047285e-05, "loss": 1.6829, "step": 32737 }, { "epoch": 0.03, "learning_rate": 4.986867025756472e-05, "loss": 1.1365, "step": 32738 }, { "epoch": 0.03, "learning_rate": 4.986866221683666e-05, "loss": 1.0381, "step": 32739 }, { "epoch": 0.03, "learning_rate": 4.986865417586311e-05, "loss": 0.7808, "step": 32740 }, { "epoch": 0.03, "learning_rate": 4.986864613464406e-05, "loss": 0.996, "step": 32741 }, { "epoch": 0.03, "learning_rate": 4.986863809317952e-05, "loss": 0.775, "step": 32742 }, { "epoch": 0.03, "learning_rate": 4.986863005146949e-05, "loss": 0.9114, "step": 32743 }, { "epoch": 0.03, "learning_rate": 4.986862200951397e-05, "loss": 0.7647, "step": 32744 }, { "epoch": 0.03, "learning_rate": 4.986861396731295e-05, "loss": 0.744, "step": 32745 }, { "epoch": 0.03, "learning_rate": 4.986860592486644e-05, "loss": 0.5315, "step": 32746 }, { "epoch": 0.03, "learning_rate": 4.986859788217444e-05, "loss": 1.1544, "step": 32747 }, { "epoch": 0.03, "learning_rate": 4.9868589839236946e-05, "loss": 1.0075, "step": 32748 }, { "epoch": 0.03, "learning_rate": 4.986858179605396e-05, "loss": 0.963, "step": 32749 }, { "epoch": 0.03, "learning_rate": 4.986857375262548e-05, "loss": 0.7979, "step": 32750 }, { "epoch": 0.03, "learning_rate": 4.986856570895151e-05, "loss": 0.9047, "step": 32751 }, { "epoch": 0.03, "learning_rate": 4.986855766503205e-05, "loss": 1.1423, "step": 32752 }, { "epoch": 0.03, "learning_rate": 4.98685496208671e-05, "loss": 1.0281, "step": 32753 }, { "epoch": 0.03, "learning_rate": 4.986854157645665e-05, "loss": 0.9408, "step": 32754 }, { "epoch": 0.03, "learning_rate": 4.9868533531800714e-05, "loss": 0.8022, "step": 32755 }, { "epoch": 0.03, "learning_rate": 4.9868525486899285e-05, "loss": 1.1286, "step": 32756 }, { "epoch": 0.03, "learning_rate": 4.986851744175236e-05, "loss": 0.8981, "step": 32757 }, { "epoch": 0.03, "learning_rate": 4.986850939635995e-05, "loss": 0.9574, "step": 32758 }, { "epoch": 0.03, "learning_rate": 4.986850135072205e-05, "loss": 0.8572, "step": 32759 }, { "epoch": 0.03, "learning_rate": 4.986849330483865e-05, "loss": 0.8855, "step": 32760 }, { "epoch": 0.03, "learning_rate": 4.9868485258709765e-05, "loss": 0.834, "step": 32761 }, { "epoch": 0.03, "learning_rate": 4.986847721233539e-05, "loss": 1.2317, "step": 32762 }, { "epoch": 0.03, "learning_rate": 4.986846916571552e-05, "loss": 0.7281, "step": 32763 }, { "epoch": 0.03, "learning_rate": 4.986846111885016e-05, "loss": 0.9301, "step": 32764 }, { "epoch": 0.03, "learning_rate": 4.986845307173931e-05, "loss": 0.8766, "step": 32765 }, { "epoch": 0.03, "learning_rate": 4.986844502438297e-05, "loss": 0.9497, "step": 32766 }, { "epoch": 0.03, "learning_rate": 4.986843697678113e-05, "loss": 0.8874, "step": 32767 }, { "epoch": 0.03, "learning_rate": 4.986842892893381e-05, "loss": 0.9153, "step": 32768 }, { "epoch": 0.03, "learning_rate": 4.9868420880841e-05, "loss": 0.8775, "step": 32769 }, { "epoch": 0.03, "learning_rate": 4.98684128325027e-05, "loss": 0.845, "step": 32770 }, { "epoch": 0.03, "learning_rate": 4.98684047839189e-05, "loss": 1.1355, "step": 32771 }, { "epoch": 0.03, "learning_rate": 4.9868396735089616e-05, "loss": 1.0978, "step": 32772 }, { "epoch": 0.03, "learning_rate": 4.9868388686014836e-05, "loss": 1.069, "step": 32773 }, { "epoch": 0.03, "learning_rate": 4.986838063669457e-05, "loss": 0.9329, "step": 32774 }, { "epoch": 0.03, "learning_rate": 4.986837258712882e-05, "loss": 1.0344, "step": 32775 }, { "epoch": 0.03, "learning_rate": 4.9868364537317566e-05, "loss": 1.0649, "step": 32776 }, { "epoch": 0.03, "learning_rate": 4.986835648726084e-05, "loss": 0.8957, "step": 32777 }, { "epoch": 0.03, "learning_rate": 4.986834843695861e-05, "loss": 1.1022, "step": 32778 }, { "epoch": 0.03, "learning_rate": 4.9868340386410895e-05, "loss": 1.4241, "step": 32779 }, { "epoch": 0.03, "learning_rate": 4.986833233561768e-05, "loss": 0.8848, "step": 32780 }, { "epoch": 0.03, "learning_rate": 4.986832428457899e-05, "loss": 1.007, "step": 32781 }, { "epoch": 0.03, "learning_rate": 4.986831623329481e-05, "loss": 1.1292, "step": 32782 }, { "epoch": 0.03, "learning_rate": 4.986830818176513e-05, "loss": 0.8276, "step": 32783 }, { "epoch": 0.03, "learning_rate": 4.9868300129989964e-05, "loss": 1.3835, "step": 32784 }, { "epoch": 0.03, "learning_rate": 4.986829207796931e-05, "loss": 0.8078, "step": 32785 }, { "epoch": 0.03, "learning_rate": 4.986828402570317e-05, "loss": 0.785, "step": 32786 }, { "epoch": 0.03, "learning_rate": 4.9868275973191535e-05, "loss": 0.8556, "step": 32787 }, { "epoch": 0.03, "learning_rate": 4.986826792043442e-05, "loss": 0.9444, "step": 32788 }, { "epoch": 0.03, "learning_rate": 4.986825986743181e-05, "loss": 1.2487, "step": 32789 }, { "epoch": 0.03, "learning_rate": 4.9868251814183706e-05, "loss": 1.3865, "step": 32790 }, { "epoch": 0.03, "learning_rate": 4.9868243760690124e-05, "loss": 1.1568, "step": 32791 }, { "epoch": 0.03, "learning_rate": 4.9868235706951045e-05, "loss": 1.167, "step": 32792 }, { "epoch": 0.03, "learning_rate": 4.986822765296648e-05, "loss": 0.9974, "step": 32793 }, { "epoch": 0.03, "learning_rate": 4.986821959873642e-05, "loss": 0.9881, "step": 32794 }, { "epoch": 0.03, "learning_rate": 4.986821154426088e-05, "loss": 0.8252, "step": 32795 }, { "epoch": 0.03, "learning_rate": 4.986820348953985e-05, "loss": 0.6458, "step": 32796 }, { "epoch": 0.03, "learning_rate": 4.986819543457333e-05, "loss": 0.3983, "step": 32797 }, { "epoch": 0.03, "learning_rate": 4.986818737936132e-05, "loss": 0.5553, "step": 32798 }, { "epoch": 0.03, "learning_rate": 4.986817932390383e-05, "loss": 0.922, "step": 32799 }, { "epoch": 0.03, "learning_rate": 4.986817126820084e-05, "loss": 0.6957, "step": 32800 }, { "epoch": 0.03, "learning_rate": 4.986816321225237e-05, "loss": 1.4219, "step": 32801 }, { "epoch": 0.03, "learning_rate": 4.98681551560584e-05, "loss": 0.9473, "step": 32802 }, { "epoch": 0.03, "learning_rate": 4.9868147099618955e-05, "loss": 0.8883, "step": 32803 }, { "epoch": 0.03, "learning_rate": 4.9868139042934024e-05, "loss": 0.9685, "step": 32804 }, { "epoch": 0.03, "learning_rate": 4.98681309860036e-05, "loss": 0.8416, "step": 32805 }, { "epoch": 0.03, "learning_rate": 4.9868122928827685e-05, "loss": 0.9497, "step": 32806 }, { "epoch": 0.03, "learning_rate": 4.986811487140629e-05, "loss": 0.9441, "step": 32807 }, { "epoch": 0.03, "learning_rate": 4.98681068137394e-05, "loss": 1.9508, "step": 32808 }, { "epoch": 0.03, "learning_rate": 4.986809875582703e-05, "loss": 1.1203, "step": 32809 }, { "epoch": 0.03, "learning_rate": 4.986809069766917e-05, "loss": 1.2226, "step": 32810 }, { "epoch": 0.03, "learning_rate": 4.986808263926581e-05, "loss": 1.0671, "step": 32811 }, { "epoch": 0.03, "learning_rate": 4.9868074580616975e-05, "loss": 0.9469, "step": 32812 }, { "epoch": 0.03, "learning_rate": 4.986806652172266e-05, "loss": 1.0367, "step": 32813 }, { "epoch": 0.03, "learning_rate": 4.986805846258284e-05, "loss": 1.2934, "step": 32814 }, { "epoch": 0.03, "learning_rate": 4.9868050403197543e-05, "loss": 1.0082, "step": 32815 }, { "epoch": 0.03, "learning_rate": 4.9868042343566754e-05, "loss": 1.0065, "step": 32816 }, { "epoch": 0.03, "learning_rate": 4.986803428369049e-05, "loss": 1.2302, "step": 32817 }, { "epoch": 0.03, "learning_rate": 4.986802622356873e-05, "loss": 1.7011, "step": 32818 }, { "epoch": 0.03, "learning_rate": 4.9868018163201484e-05, "loss": 0.9184, "step": 32819 }, { "epoch": 0.03, "learning_rate": 4.986801010258876e-05, "loss": 0.8603, "step": 32820 }, { "epoch": 0.03, "learning_rate": 4.986800204173053e-05, "loss": 0.9228, "step": 32821 }, { "epoch": 0.03, "learning_rate": 4.986799398062683e-05, "loss": 1.4186, "step": 32822 }, { "epoch": 0.03, "learning_rate": 4.986798591927764e-05, "loss": 1.1009, "step": 32823 }, { "epoch": 0.03, "learning_rate": 4.9867977857682965e-05, "loss": 0.7349, "step": 32824 }, { "epoch": 0.03, "learning_rate": 4.98679697958428e-05, "loss": 1.1221, "step": 32825 }, { "epoch": 0.03, "learning_rate": 4.986796173375715e-05, "loss": 0.8156, "step": 32826 }, { "epoch": 0.03, "learning_rate": 4.986795367142602e-05, "loss": 0.7563, "step": 32827 }, { "epoch": 0.03, "learning_rate": 4.986794560884939e-05, "loss": 0.6787, "step": 32828 }, { "epoch": 0.03, "learning_rate": 4.9867937546027286e-05, "loss": 0.5335, "step": 32829 }, { "epoch": 0.03, "learning_rate": 4.98679294829597e-05, "loss": 0.5028, "step": 32830 }, { "epoch": 0.03, "learning_rate": 4.986792141964661e-05, "loss": 0.505, "step": 32831 }, { "epoch": 0.03, "learning_rate": 4.986791335608805e-05, "loss": 0.3059, "step": 32832 }, { "epoch": 0.03, "learning_rate": 4.9867905292284003e-05, "loss": 0.3066, "step": 32833 }, { "epoch": 0.03, "learning_rate": 4.986789722823446e-05, "loss": 0.6744, "step": 32834 }, { "epoch": 0.03, "learning_rate": 4.986788916393944e-05, "loss": 0.7872, "step": 32835 }, { "epoch": 0.03, "learning_rate": 4.9867881099398936e-05, "loss": 0.805, "step": 32836 }, { "epoch": 0.03, "learning_rate": 4.986787303461294e-05, "loss": 1.2698, "step": 32837 }, { "epoch": 0.03, "learning_rate": 4.986786496958147e-05, "loss": 0.7927, "step": 32838 }, { "epoch": 0.03, "learning_rate": 4.986785690430451e-05, "loss": 1.5429, "step": 32839 }, { "epoch": 0.03, "learning_rate": 4.986784883878206e-05, "loss": 1.4712, "step": 32840 }, { "epoch": 0.03, "learning_rate": 4.986784077301413e-05, "loss": 1.4562, "step": 32841 }, { "epoch": 0.03, "learning_rate": 4.986783270700071e-05, "loss": 1.6453, "step": 32842 }, { "epoch": 0.03, "learning_rate": 4.986782464074181e-05, "loss": 1.4062, "step": 32843 }, { "epoch": 0.03, "learning_rate": 4.986781657423742e-05, "loss": 0.9093, "step": 32844 }, { "epoch": 0.03, "learning_rate": 4.986780850748756e-05, "loss": 0.9786, "step": 32845 }, { "epoch": 0.03, "learning_rate": 4.98678004404922e-05, "loss": 0.9596, "step": 32846 }, { "epoch": 0.03, "learning_rate": 4.986779237325135e-05, "loss": 0.9841, "step": 32847 }, { "epoch": 0.03, "learning_rate": 4.986778430576503e-05, "loss": 0.7686, "step": 32848 }, { "epoch": 0.03, "learning_rate": 4.9867776238033224e-05, "loss": 0.8392, "step": 32849 }, { "epoch": 0.03, "learning_rate": 4.9867768170055935e-05, "loss": 1.1617, "step": 32850 }, { "epoch": 0.03, "learning_rate": 4.9867760101833155e-05, "loss": 0.9683, "step": 32851 }, { "epoch": 0.03, "learning_rate": 4.98677520333649e-05, "loss": 0.71, "step": 32852 }, { "epoch": 0.03, "learning_rate": 4.986774396465115e-05, "loss": 1.1089, "step": 32853 }, { "epoch": 0.03, "learning_rate": 4.9867735895691927e-05, "loss": 0.8703, "step": 32854 }, { "epoch": 0.03, "learning_rate": 4.986772782648721e-05, "loss": 0.6506, "step": 32855 }, { "epoch": 0.03, "learning_rate": 4.9867719757037014e-05, "loss": 0.7203, "step": 32856 }, { "epoch": 0.03, "learning_rate": 4.986771168734133e-05, "loss": 0.6353, "step": 32857 }, { "epoch": 0.03, "learning_rate": 4.986770361740017e-05, "loss": 0.7806, "step": 32858 }, { "epoch": 0.03, "learning_rate": 4.986769554721352e-05, "loss": 0.8945, "step": 32859 }, { "epoch": 0.03, "learning_rate": 4.9867687476781397e-05, "loss": 0.5753, "step": 32860 }, { "epoch": 0.03, "learning_rate": 4.9867679406103786e-05, "loss": 1.1495, "step": 32861 }, { "epoch": 0.03, "learning_rate": 4.9867671335180686e-05, "loss": 0.9167, "step": 32862 }, { "epoch": 0.03, "learning_rate": 4.98676632640121e-05, "loss": 0.7187, "step": 32863 }, { "epoch": 0.03, "learning_rate": 4.986765519259804e-05, "loss": 1.2179, "step": 32864 }, { "epoch": 0.03, "learning_rate": 4.9867647120938495e-05, "loss": 0.9852, "step": 32865 }, { "epoch": 0.03, "learning_rate": 4.986763904903347e-05, "loss": 1.104, "step": 32866 }, { "epoch": 0.03, "learning_rate": 4.986763097688295e-05, "loss": 1.0321, "step": 32867 }, { "epoch": 0.03, "learning_rate": 4.986762290448696e-05, "loss": 1.2008, "step": 32868 }, { "epoch": 0.03, "learning_rate": 4.9867614831845486e-05, "loss": 0.8316, "step": 32869 }, { "epoch": 0.03, "learning_rate": 4.9867606758958515e-05, "loss": 0.3834, "step": 32870 }, { "epoch": 0.03, "learning_rate": 4.9867598685826075e-05, "loss": 0.3703, "step": 32871 }, { "epoch": 0.03, "learning_rate": 4.986759061244816e-05, "loss": 0.3433, "step": 32872 }, { "epoch": 0.03, "learning_rate": 4.986758253882475e-05, "loss": 0.3707, "step": 32873 }, { "epoch": 0.03, "learning_rate": 4.986757446495586e-05, "loss": 0.3672, "step": 32874 }, { "epoch": 0.03, "learning_rate": 4.9867566390841484e-05, "loss": 0.4027, "step": 32875 }, { "epoch": 0.03, "learning_rate": 4.986755831648163e-05, "loss": 0.7761, "step": 32876 }, { "epoch": 0.03, "learning_rate": 4.9867550241876296e-05, "loss": 0.9081, "step": 32877 }, { "epoch": 0.03, "learning_rate": 4.9867542167025476e-05, "loss": 0.9736, "step": 32878 }, { "epoch": 0.03, "learning_rate": 4.986753409192918e-05, "loss": 1.2494, "step": 32879 }, { "epoch": 0.03, "learning_rate": 4.986752601658739e-05, "loss": 1.0593, "step": 32880 }, { "epoch": 0.03, "learning_rate": 4.986751794100013e-05, "loss": 1.3639, "step": 32881 }, { "epoch": 0.03, "learning_rate": 4.986750986516739e-05, "loss": 0.9379, "step": 32882 }, { "epoch": 0.03, "learning_rate": 4.986750178908916e-05, "loss": 1.0351, "step": 32883 }, { "epoch": 0.03, "learning_rate": 4.9867493712765456e-05, "loss": 1.0376, "step": 32884 }, { "epoch": 0.03, "learning_rate": 4.9867485636196264e-05, "loss": 1.0915, "step": 32885 }, { "epoch": 0.03, "learning_rate": 4.9867477559381595e-05, "loss": 1.004, "step": 32886 }, { "epoch": 0.03, "learning_rate": 4.9867469482321436e-05, "loss": 1.1297, "step": 32887 }, { "epoch": 0.03, "learning_rate": 4.986746140501581e-05, "loss": 1.1506, "step": 32888 }, { "epoch": 0.03, "learning_rate": 4.9867453327464694e-05, "loss": 1.1691, "step": 32889 }, { "epoch": 0.03, "learning_rate": 4.98674452496681e-05, "loss": 0.9571, "step": 32890 }, { "epoch": 0.03, "learning_rate": 4.9867437171626017e-05, "loss": 1.1372, "step": 32891 }, { "epoch": 0.03, "learning_rate": 4.9867429093338466e-05, "loss": 1.0509, "step": 32892 }, { "epoch": 0.03, "learning_rate": 4.9867421014805425e-05, "loss": 1.9297, "step": 32893 }, { "epoch": 0.03, "learning_rate": 4.986741293602691e-05, "loss": 2.0199, "step": 32894 }, { "epoch": 0.03, "learning_rate": 4.9867404857002905e-05, "loss": 1.8944, "step": 32895 }, { "epoch": 0.03, "learning_rate": 4.9867396777733426e-05, "loss": 0.9556, "step": 32896 }, { "epoch": 0.03, "learning_rate": 4.9867388698218464e-05, "loss": 1.0078, "step": 32897 }, { "epoch": 0.03, "learning_rate": 4.9867380618458025e-05, "loss": 1.0279, "step": 32898 }, { "epoch": 0.03, "learning_rate": 4.986737253845211e-05, "loss": 0.9194, "step": 32899 }, { "epoch": 0.03, "learning_rate": 4.98673644582007e-05, "loss": 0.5707, "step": 32900 }, { "epoch": 0.03, "learning_rate": 4.9867356377703825e-05, "loss": 0.6351, "step": 32901 }, { "epoch": 0.03, "learning_rate": 4.986734829696146e-05, "loss": 1.0533, "step": 32902 }, { "epoch": 0.03, "learning_rate": 4.986734021597362e-05, "loss": 1.0529, "step": 32903 }, { "epoch": 0.03, "learning_rate": 4.98673321347403e-05, "loss": 1.0909, "step": 32904 }, { "epoch": 0.03, "learning_rate": 4.98673240532615e-05, "loss": 1.2387, "step": 32905 }, { "epoch": 0.03, "learning_rate": 4.986731597153721e-05, "loss": 0.964, "step": 32906 }, { "epoch": 0.03, "learning_rate": 4.9867307889567454e-05, "loss": 1.1236, "step": 32907 }, { "epoch": 0.03, "learning_rate": 4.986729980735222e-05, "loss": 1.1281, "step": 32908 }, { "epoch": 0.03, "learning_rate": 4.98672917248915e-05, "loss": 1.2631, "step": 32909 }, { "epoch": 0.03, "learning_rate": 4.9867283642185296e-05, "loss": 1.0974, "step": 32910 }, { "epoch": 0.03, "learning_rate": 4.9867275559233616e-05, "loss": 1.1746, "step": 32911 }, { "epoch": 0.03, "learning_rate": 4.986726747603646e-05, "loss": 0.8562, "step": 32912 }, { "epoch": 0.03, "learning_rate": 4.9867259392593825e-05, "loss": 1.1795, "step": 32913 }, { "epoch": 0.03, "learning_rate": 4.986725130890571e-05, "loss": 1.1407, "step": 32914 }, { "epoch": 0.03, "learning_rate": 4.9867243224972114e-05, "loss": 0.9754, "step": 32915 }, { "epoch": 0.03, "learning_rate": 4.986723514079304e-05, "loss": 0.6984, "step": 32916 }, { "epoch": 0.03, "learning_rate": 4.986722705636848e-05, "loss": 1.4947, "step": 32917 }, { "epoch": 0.03, "learning_rate": 4.9867218971698455e-05, "loss": 1.3181, "step": 32918 }, { "epoch": 0.03, "learning_rate": 4.9867210886782946e-05, "loss": 0.7297, "step": 32919 }, { "epoch": 0.03, "learning_rate": 4.9867202801621954e-05, "loss": 1.2042, "step": 32920 }, { "epoch": 0.03, "learning_rate": 4.9867194716215484e-05, "loss": 0.9698, "step": 32921 }, { "epoch": 0.03, "learning_rate": 4.986718663056354e-05, "loss": 0.761, "step": 32922 }, { "epoch": 0.03, "learning_rate": 4.9867178544666114e-05, "loss": 1.1502, "step": 32923 }, { "epoch": 0.03, "learning_rate": 4.9867170458523214e-05, "loss": 1.2023, "step": 32924 }, { "epoch": 0.03, "learning_rate": 4.986716237213484e-05, "loss": 0.9473, "step": 32925 }, { "epoch": 0.03, "learning_rate": 4.986715428550097e-05, "loss": 0.6452, "step": 32926 }, { "epoch": 0.03, "learning_rate": 4.986714619862164e-05, "loss": 0.6938, "step": 32927 }, { "epoch": 0.03, "learning_rate": 4.9867138111496823e-05, "loss": 0.5521, "step": 32928 }, { "epoch": 0.03, "learning_rate": 4.986713002412653e-05, "loss": 0.8039, "step": 32929 }, { "epoch": 0.03, "learning_rate": 4.9867121936510756e-05, "loss": 0.6341, "step": 32930 }, { "epoch": 0.03, "learning_rate": 4.9867113848649504e-05, "loss": 0.7803, "step": 32931 }, { "epoch": 0.03, "learning_rate": 4.986710576054278e-05, "loss": 0.4771, "step": 32932 }, { "epoch": 0.03, "learning_rate": 4.9867097672190575e-05, "loss": 1.0048, "step": 32933 }, { "epoch": 0.03, "learning_rate": 4.98670895835929e-05, "loss": 1.273, "step": 32934 }, { "epoch": 0.03, "learning_rate": 4.986708149474973e-05, "loss": 0.9776, "step": 32935 }, { "epoch": 0.03, "learning_rate": 4.98670734056611e-05, "loss": 1.0554, "step": 32936 }, { "epoch": 0.03, "learning_rate": 4.986706531632699e-05, "loss": 1.0436, "step": 32937 }, { "epoch": 0.03, "learning_rate": 4.98670572267474e-05, "loss": 1.102, "step": 32938 }, { "epoch": 0.03, "learning_rate": 4.986704913692233e-05, "loss": 0.8194, "step": 32939 }, { "epoch": 0.03, "learning_rate": 4.986704104685178e-05, "loss": 1.0149, "step": 32940 }, { "epoch": 0.03, "learning_rate": 4.9867032956535764e-05, "loss": 0.9471, "step": 32941 }, { "epoch": 0.03, "learning_rate": 4.9867024865974265e-05, "loss": 0.8109, "step": 32942 }, { "epoch": 0.03, "learning_rate": 4.986701677516728e-05, "loss": 1.0169, "step": 32943 }, { "epoch": 0.03, "learning_rate": 4.986700868411484e-05, "loss": 1.2865, "step": 32944 }, { "epoch": 0.03, "learning_rate": 4.986700059281691e-05, "loss": 1.3324, "step": 32945 }, { "epoch": 0.03, "learning_rate": 4.98669925012735e-05, "loss": 2.3429, "step": 32946 }, { "epoch": 0.03, "learning_rate": 4.986698440948462e-05, "loss": 2.5049, "step": 32947 }, { "epoch": 0.03, "learning_rate": 4.9866976317450266e-05, "loss": 0.9527, "step": 32948 }, { "epoch": 0.03, "learning_rate": 4.986696822517043e-05, "loss": 0.8079, "step": 32949 }, { "epoch": 0.03, "learning_rate": 4.986696013264512e-05, "loss": 1.1365, "step": 32950 }, { "epoch": 0.03, "learning_rate": 4.986695203987433e-05, "loss": 1.2244, "step": 32951 }, { "epoch": 0.03, "learning_rate": 4.986694394685807e-05, "loss": 0.728, "step": 32952 }, { "epoch": 0.03, "learning_rate": 4.986693585359633e-05, "loss": 1.3202, "step": 32953 }, { "epoch": 0.03, "learning_rate": 4.986692776008912e-05, "loss": 0.9238, "step": 32954 }, { "epoch": 0.03, "learning_rate": 4.986691966633642e-05, "loss": 0.9817, "step": 32955 }, { "epoch": 0.03, "learning_rate": 4.986691157233826e-05, "loss": 0.4226, "step": 32956 }, { "epoch": 0.03, "learning_rate": 4.986690347809462e-05, "loss": 0.5311, "step": 32957 }, { "epoch": 0.03, "learning_rate": 4.9866895383605496e-05, "loss": 0.8795, "step": 32958 }, { "epoch": 0.03, "learning_rate": 4.98668872888709e-05, "loss": 0.7843, "step": 32959 }, { "epoch": 0.03, "learning_rate": 4.9866879193890833e-05, "loss": 0.8699, "step": 32960 }, { "epoch": 0.03, "learning_rate": 4.9866871098665294e-05, "loss": 0.9423, "step": 32961 }, { "epoch": 0.03, "learning_rate": 4.986686300319428e-05, "loss": 0.8998, "step": 32962 }, { "epoch": 0.03, "learning_rate": 4.986685490747778e-05, "loss": 1.0219, "step": 32963 }, { "epoch": 0.03, "learning_rate": 4.986684681151581e-05, "loss": 0.8769, "step": 32964 }, { "epoch": 0.03, "learning_rate": 4.986683871530836e-05, "loss": 0.7328, "step": 32965 }, { "epoch": 0.03, "learning_rate": 4.986683061885544e-05, "loss": 1.3538, "step": 32966 }, { "epoch": 0.03, "learning_rate": 4.9866822522157054e-05, "loss": 0.8418, "step": 32967 }, { "epoch": 0.03, "learning_rate": 4.986681442521318e-05, "loss": 0.9946, "step": 32968 }, { "epoch": 0.03, "learning_rate": 4.986680632802384e-05, "loss": 1.0181, "step": 32969 }, { "epoch": 0.03, "learning_rate": 4.9866798230589015e-05, "loss": 1.2978, "step": 32970 }, { "epoch": 0.03, "learning_rate": 4.986679013290873e-05, "loss": 0.8339, "step": 32971 }, { "epoch": 0.03, "learning_rate": 4.986678203498296e-05, "loss": 1.2131, "step": 32972 }, { "epoch": 0.03, "learning_rate": 4.986677393681172e-05, "loss": 0.9509, "step": 32973 }, { "epoch": 0.03, "learning_rate": 4.9866765838395e-05, "loss": 1.1187, "step": 32974 }, { "epoch": 0.03, "learning_rate": 4.98667577397328e-05, "loss": 1.1068, "step": 32975 }, { "epoch": 0.03, "learning_rate": 4.986674964082514e-05, "loss": 1.1811, "step": 32976 }, { "epoch": 0.03, "learning_rate": 4.9866741541672e-05, "loss": 1.3917, "step": 32977 }, { "epoch": 0.03, "learning_rate": 4.986673344227339e-05, "loss": 0.4617, "step": 32978 }, { "epoch": 0.03, "learning_rate": 4.98667253426293e-05, "loss": 0.4081, "step": 32979 }, { "epoch": 0.03, "learning_rate": 4.986671724273975e-05, "loss": 0.3844, "step": 32980 }, { "epoch": 0.03, "learning_rate": 4.986670914260471e-05, "loss": 0.3767, "step": 32981 }, { "epoch": 0.03, "learning_rate": 4.9866701042224204e-05, "loss": 0.3815, "step": 32982 }, { "epoch": 0.03, "learning_rate": 4.9866692941598226e-05, "loss": 0.3987, "step": 32983 }, { "epoch": 0.03, "learning_rate": 4.986668484072677e-05, "loss": 0.3674, "step": 32984 }, { "epoch": 0.03, "learning_rate": 4.986667673960984e-05, "loss": 0.347, "step": 32985 }, { "epoch": 0.03, "learning_rate": 4.986666863824744e-05, "loss": 0.3379, "step": 32986 }, { "epoch": 0.03, "learning_rate": 4.9866660536639566e-05, "loss": 0.3167, "step": 32987 }, { "epoch": 0.03, "learning_rate": 4.986665243478621e-05, "loss": 0.3468, "step": 32988 }, { "epoch": 0.03, "learning_rate": 4.98666443326874e-05, "loss": 0.3668, "step": 32989 }, { "epoch": 0.03, "learning_rate": 4.98666362303431e-05, "loss": 0.4079, "step": 32990 }, { "epoch": 0.03, "learning_rate": 4.986662812775333e-05, "loss": 0.3259, "step": 32991 }, { "epoch": 0.03, "learning_rate": 4.9866620024918086e-05, "loss": 0.3237, "step": 32992 }, { "epoch": 0.03, "learning_rate": 4.986661192183738e-05, "loss": 0.3472, "step": 32993 }, { "epoch": 0.03, "learning_rate": 4.9866603818511196e-05, "loss": 0.3247, "step": 32994 }, { "epoch": 0.03, "learning_rate": 4.9866595714939536e-05, "loss": 0.3752, "step": 32995 }, { "epoch": 0.03, "learning_rate": 4.98665876111224e-05, "loss": 0.3747, "step": 32996 }, { "epoch": 0.03, "learning_rate": 4.98665795070598e-05, "loss": 0.3929, "step": 32997 }, { "epoch": 0.03, "learning_rate": 4.986657140275172e-05, "loss": 1.0142, "step": 32998 }, { "epoch": 0.03, "learning_rate": 4.986656329819818e-05, "loss": 1.1707, "step": 32999 }, { "epoch": 0.03, "learning_rate": 4.9866555193399155e-05, "loss": 0.8089, "step": 33000 }, { "epoch": 0.03, "eval_loss": 1.1077449321746826, "eval_runtime": 128.1069, "eval_samples_per_second": 10.811, "eval_steps_per_second": 5.41, "step": 33000 }, { "epoch": 0.03, "learning_rate": 4.986654708835467e-05, "loss": 0.7104, "step": 33001 }, { "epoch": 0.03, "learning_rate": 4.9866538983064695e-05, "loss": 0.8027, "step": 33002 }, { "epoch": 0.03, "learning_rate": 4.986653087752926e-05, "loss": 0.7417, "step": 33003 }, { "epoch": 0.03, "learning_rate": 4.986652277174835e-05, "loss": 0.475, "step": 33004 }, { "epoch": 0.03, "learning_rate": 4.986651466572197e-05, "loss": 0.9183, "step": 33005 }, { "epoch": 0.03, "learning_rate": 4.986650655945012e-05, "loss": 0.7274, "step": 33006 }, { "epoch": 0.03, "learning_rate": 4.98664984529328e-05, "loss": 1.03, "step": 33007 }, { "epoch": 0.03, "learning_rate": 4.9866490346170006e-05, "loss": 1.1065, "step": 33008 }, { "epoch": 0.03, "learning_rate": 4.986648223916173e-05, "loss": 1.1011, "step": 33009 }, { "epoch": 0.03, "learning_rate": 4.986647413190799e-05, "loss": 1.2691, "step": 33010 }, { "epoch": 0.03, "learning_rate": 4.986646602440879e-05, "loss": 0.8662, "step": 33011 }, { "epoch": 0.03, "learning_rate": 4.98664579166641e-05, "loss": 1.0739, "step": 33012 }, { "epoch": 0.03, "learning_rate": 4.986644980867396e-05, "loss": 1.6486, "step": 33013 }, { "epoch": 0.03, "learning_rate": 4.9866441700438324e-05, "loss": 0.9896, "step": 33014 }, { "epoch": 0.03, "learning_rate": 4.986643359195723e-05, "loss": 1.2565, "step": 33015 }, { "epoch": 0.03, "learning_rate": 4.986642548323067e-05, "loss": 0.9514, "step": 33016 }, { "epoch": 0.03, "learning_rate": 4.9866417374258635e-05, "loss": 1.1112, "step": 33017 }, { "epoch": 0.03, "learning_rate": 4.986640926504113e-05, "loss": 1.0109, "step": 33018 }, { "epoch": 0.03, "learning_rate": 4.986640115557814e-05, "loss": 0.6881, "step": 33019 }, { "epoch": 0.03, "learning_rate": 4.9866393045869695e-05, "loss": 0.9575, "step": 33020 }, { "epoch": 0.03, "learning_rate": 4.986638493591578e-05, "loss": 0.9427, "step": 33021 }, { "epoch": 0.03, "learning_rate": 4.986637682571639e-05, "loss": 1.9894, "step": 33022 }, { "epoch": 0.03, "learning_rate": 4.986636871527153e-05, "loss": 3.3986, "step": 33023 }, { "epoch": 0.03, "learning_rate": 4.98663606045812e-05, "loss": 0.8276, "step": 33024 }, { "epoch": 0.03, "learning_rate": 4.9866352493645404e-05, "loss": 0.756, "step": 33025 }, { "epoch": 0.03, "learning_rate": 4.986634438246413e-05, "loss": 0.7374, "step": 33026 }, { "epoch": 0.03, "learning_rate": 4.9866336271037396e-05, "loss": 0.7242, "step": 33027 }, { "epoch": 0.03, "learning_rate": 4.986632815936518e-05, "loss": 0.7394, "step": 33028 }, { "epoch": 0.03, "learning_rate": 4.98663200474475e-05, "loss": 0.7326, "step": 33029 }, { "epoch": 0.03, "learning_rate": 4.986631193528435e-05, "loss": 0.7098, "step": 33030 }, { "epoch": 0.03, "learning_rate": 4.9866303822875726e-05, "loss": 0.7133, "step": 33031 }, { "epoch": 0.03, "learning_rate": 4.986629571022164e-05, "loss": 0.6875, "step": 33032 }, { "epoch": 0.03, "learning_rate": 4.986628759732208e-05, "loss": 0.8125, "step": 33033 }, { "epoch": 0.03, "learning_rate": 4.986627948417705e-05, "loss": 0.6954, "step": 33034 }, { "epoch": 0.03, "learning_rate": 4.986627137078655e-05, "loss": 1.0913, "step": 33035 }, { "epoch": 0.03, "learning_rate": 4.986626325715058e-05, "loss": 1.449, "step": 33036 }, { "epoch": 0.03, "learning_rate": 4.986625514326915e-05, "loss": 1.0696, "step": 33037 }, { "epoch": 0.03, "learning_rate": 4.9866247029142245e-05, "loss": 1.1017, "step": 33038 }, { "epoch": 0.03, "learning_rate": 4.986623891476987e-05, "loss": 1.2841, "step": 33039 }, { "epoch": 0.03, "learning_rate": 4.986623080015203e-05, "loss": 0.8803, "step": 33040 }, { "epoch": 0.03, "learning_rate": 4.986622268528871e-05, "loss": 0.8335, "step": 33041 }, { "epoch": 0.03, "learning_rate": 4.986621457017992e-05, "loss": 0.8166, "step": 33042 }, { "epoch": 0.03, "learning_rate": 4.986620645482568e-05, "loss": 1.6526, "step": 33043 }, { "epoch": 0.03, "learning_rate": 4.9866198339225957e-05, "loss": 1.9394, "step": 33044 }, { "epoch": 0.03, "learning_rate": 4.986619022338077e-05, "loss": 0.8341, "step": 33045 }, { "epoch": 0.03, "learning_rate": 4.986618210729012e-05, "loss": 1.001, "step": 33046 }, { "epoch": 0.03, "learning_rate": 4.986617399095399e-05, "loss": 1.0913, "step": 33047 }, { "epoch": 0.03, "learning_rate": 4.98661658743724e-05, "loss": 1.2703, "step": 33048 }, { "epoch": 0.03, "learning_rate": 4.9866157757545335e-05, "loss": 1.1661, "step": 33049 }, { "epoch": 0.03, "learning_rate": 4.9866149640472803e-05, "loss": 1.2495, "step": 33050 }, { "epoch": 0.03, "learning_rate": 4.986614152315481e-05, "loss": 1.0336, "step": 33051 }, { "epoch": 0.03, "learning_rate": 4.986613340559134e-05, "loss": 0.8475, "step": 33052 }, { "epoch": 0.03, "learning_rate": 4.986612528778241e-05, "loss": 1.0408, "step": 33053 }, { "epoch": 0.03, "learning_rate": 4.9866117169728e-05, "loss": 1.2201, "step": 33054 }, { "epoch": 0.03, "learning_rate": 4.986610905142813e-05, "loss": 1.1473, "step": 33055 }, { "epoch": 0.03, "learning_rate": 4.98661009328828e-05, "loss": 1.1419, "step": 33056 }, { "epoch": 0.03, "learning_rate": 4.986609281409199e-05, "loss": 1.2108, "step": 33057 }, { "epoch": 0.03, "learning_rate": 4.986608469505572e-05, "loss": 0.9765, "step": 33058 }, { "epoch": 0.03, "learning_rate": 4.986607657577398e-05, "loss": 1.0187, "step": 33059 }, { "epoch": 0.03, "learning_rate": 4.986606845624677e-05, "loss": 1.1752, "step": 33060 }, { "epoch": 0.03, "learning_rate": 4.98660603364741e-05, "loss": 0.7546, "step": 33061 }, { "epoch": 0.03, "learning_rate": 4.986605221645595e-05, "loss": 1.0998, "step": 33062 }, { "epoch": 0.03, "learning_rate": 4.986604409619234e-05, "loss": 0.8017, "step": 33063 }, { "epoch": 0.03, "learning_rate": 4.986603597568327e-05, "loss": 0.8413, "step": 33064 }, { "epoch": 0.03, "learning_rate": 4.986602785492872e-05, "loss": 1.0091, "step": 33065 }, { "epoch": 0.03, "learning_rate": 4.986601973392871e-05, "loss": 1.1373, "step": 33066 }, { "epoch": 0.03, "learning_rate": 4.986601161268323e-05, "loss": 0.9977, "step": 33067 }, { "epoch": 0.03, "learning_rate": 4.986600349119229e-05, "loss": 1.3586, "step": 33068 }, { "epoch": 0.03, "learning_rate": 4.9865995369455876e-05, "loss": 1.4936, "step": 33069 }, { "epoch": 0.03, "learning_rate": 4.9865987247474e-05, "loss": 1.3752, "step": 33070 }, { "epoch": 0.03, "learning_rate": 4.986597912524665e-05, "loss": 1.214, "step": 33071 }, { "epoch": 0.03, "learning_rate": 4.9865971002773844e-05, "loss": 1.0869, "step": 33072 }, { "epoch": 0.03, "learning_rate": 4.986596288005556e-05, "loss": 0.9108, "step": 33073 }, { "epoch": 0.03, "learning_rate": 4.986595475709182e-05, "loss": 0.7096, "step": 33074 }, { "epoch": 0.03, "learning_rate": 4.986594663388261e-05, "loss": 0.9052, "step": 33075 }, { "epoch": 0.03, "learning_rate": 4.986593851042793e-05, "loss": 0.6274, "step": 33076 }, { "epoch": 0.03, "learning_rate": 4.9865930386727795e-05, "loss": 0.7202, "step": 33077 }, { "epoch": 0.03, "learning_rate": 4.986592226278218e-05, "loss": 0.9891, "step": 33078 }, { "epoch": 0.03, "learning_rate": 4.9865914138591115e-05, "loss": 0.9417, "step": 33079 }, { "epoch": 0.03, "learning_rate": 4.9865906014154574e-05, "loss": 0.969, "step": 33080 }, { "epoch": 0.03, "learning_rate": 4.986589788947257e-05, "loss": 0.9762, "step": 33081 }, { "epoch": 0.03, "learning_rate": 4.9865889764545094e-05, "loss": 1.1048, "step": 33082 }, { "epoch": 0.03, "learning_rate": 4.9865881639372156e-05, "loss": 0.9153, "step": 33083 }, { "epoch": 0.03, "learning_rate": 4.9865873513953754e-05, "loss": 1.3031, "step": 33084 }, { "epoch": 0.03, "learning_rate": 4.986586538828989e-05, "loss": 1.3788, "step": 33085 }, { "epoch": 0.03, "learning_rate": 4.986585726238055e-05, "loss": 1.2371, "step": 33086 }, { "epoch": 0.03, "learning_rate": 4.986584913622576e-05, "loss": 1.3113, "step": 33087 }, { "epoch": 0.03, "learning_rate": 4.98658410098255e-05, "loss": 1.4019, "step": 33088 }, { "epoch": 0.03, "learning_rate": 4.986583288317977e-05, "loss": 1.1974, "step": 33089 }, { "epoch": 0.03, "learning_rate": 4.986582475628857e-05, "loss": 1.318, "step": 33090 }, { "epoch": 0.03, "learning_rate": 4.986581662915192e-05, "loss": 1.3099, "step": 33091 }, { "epoch": 0.03, "learning_rate": 4.986580850176979e-05, "loss": 1.112, "step": 33092 }, { "epoch": 0.03, "learning_rate": 4.9865800374142204e-05, "loss": 0.9023, "step": 33093 }, { "epoch": 0.03, "learning_rate": 4.986579224626915e-05, "loss": 0.9732, "step": 33094 }, { "epoch": 0.03, "learning_rate": 4.986578411815063e-05, "loss": 0.996, "step": 33095 }, { "epoch": 0.03, "learning_rate": 4.9865775989786653e-05, "loss": 1.0636, "step": 33096 }, { "epoch": 0.03, "learning_rate": 4.98657678611772e-05, "loss": 1.2013, "step": 33097 }, { "epoch": 0.03, "learning_rate": 4.986575973232229e-05, "loss": 1.0979, "step": 33098 }, { "epoch": 0.03, "learning_rate": 4.986575160322192e-05, "loss": 0.9918, "step": 33099 }, { "epoch": 0.03, "learning_rate": 4.986574347387608e-05, "loss": 1.0657, "step": 33100 }, { "epoch": 0.03, "learning_rate": 4.9865735344284776e-05, "loss": 0.7863, "step": 33101 }, { "epoch": 0.03, "learning_rate": 4.9865727214448013e-05, "loss": 1.137, "step": 33102 }, { "epoch": 0.03, "learning_rate": 4.9865719084365774e-05, "loss": 1.0307, "step": 33103 }, { "epoch": 0.03, "learning_rate": 4.986571095403808e-05, "loss": 1.0934, "step": 33104 }, { "epoch": 0.03, "learning_rate": 4.9865702823464924e-05, "loss": 1.1962, "step": 33105 }, { "epoch": 0.03, "learning_rate": 4.9865694692646295e-05, "loss": 1.1201, "step": 33106 }, { "epoch": 0.03, "learning_rate": 4.9865686561582215e-05, "loss": 1.0853, "step": 33107 }, { "epoch": 0.03, "learning_rate": 4.986567843027266e-05, "loss": 1.1154, "step": 33108 }, { "epoch": 0.03, "learning_rate": 4.9865670298717645e-05, "loss": 0.7638, "step": 33109 }, { "epoch": 0.03, "learning_rate": 4.986566216691717e-05, "loss": 1.0088, "step": 33110 }, { "epoch": 0.03, "learning_rate": 4.986565403487123e-05, "loss": 0.9251, "step": 33111 }, { "epoch": 0.03, "learning_rate": 4.9865645902579825e-05, "loss": 0.877, "step": 33112 }, { "epoch": 0.03, "learning_rate": 4.986563777004296e-05, "loss": 0.831, "step": 33113 }, { "epoch": 0.03, "learning_rate": 4.986562963726062e-05, "loss": 0.9163, "step": 33114 }, { "epoch": 0.03, "learning_rate": 4.9865621504232835e-05, "loss": 0.9736, "step": 33115 }, { "epoch": 0.03, "learning_rate": 4.986561337095958e-05, "loss": 0.6199, "step": 33116 }, { "epoch": 0.03, "learning_rate": 4.986560523744086e-05, "loss": 1.1655, "step": 33117 }, { "epoch": 0.03, "learning_rate": 4.9865597103676674e-05, "loss": 0.7523, "step": 33118 }, { "epoch": 0.03, "learning_rate": 4.9865588969667034e-05, "loss": 0.9462, "step": 33119 }, { "epoch": 0.03, "learning_rate": 4.986558083541193e-05, "loss": 0.9223, "step": 33120 }, { "epoch": 0.03, "learning_rate": 4.986557270091136e-05, "loss": 1.1487, "step": 33121 }, { "epoch": 0.03, "learning_rate": 4.986556456616533e-05, "loss": 1.2566, "step": 33122 }, { "epoch": 0.03, "learning_rate": 4.986555643117383e-05, "loss": 1.2188, "step": 33123 }, { "epoch": 0.03, "learning_rate": 4.9865548295936876e-05, "loss": 0.9495, "step": 33124 }, { "epoch": 0.03, "learning_rate": 4.9865540160454456e-05, "loss": 1.3351, "step": 33125 }, { "epoch": 0.03, "learning_rate": 4.986553202472657e-05, "loss": 0.9941, "step": 33126 }, { "epoch": 0.03, "learning_rate": 4.986552388875323e-05, "loss": 0.9929, "step": 33127 }, { "epoch": 0.03, "learning_rate": 4.986551575253443e-05, "loss": 0.9276, "step": 33128 }, { "epoch": 0.03, "learning_rate": 4.986550761607016e-05, "loss": 1.231, "step": 33129 }, { "epoch": 0.03, "learning_rate": 4.986549947936043e-05, "loss": 1.0074, "step": 33130 }, { "epoch": 0.03, "learning_rate": 4.986549134240524e-05, "loss": 0.8524, "step": 33131 }, { "epoch": 0.03, "learning_rate": 4.9865483205204586e-05, "loss": 0.8722, "step": 33132 }, { "epoch": 0.03, "learning_rate": 4.986547506775847e-05, "loss": 0.7019, "step": 33133 }, { "epoch": 0.03, "learning_rate": 4.9865466930066895e-05, "loss": 0.7256, "step": 33134 }, { "epoch": 0.03, "learning_rate": 4.986545879212986e-05, "loss": 0.9898, "step": 33135 }, { "epoch": 0.03, "learning_rate": 4.9865450653947364e-05, "loss": 1.3975, "step": 33136 }, { "epoch": 0.03, "learning_rate": 4.9865442515519404e-05, "loss": 0.775, "step": 33137 }, { "epoch": 0.03, "learning_rate": 4.9865434376845987e-05, "loss": 0.9602, "step": 33138 }, { "epoch": 0.03, "learning_rate": 4.9865426237927106e-05, "loss": 0.9798, "step": 33139 }, { "epoch": 0.03, "learning_rate": 4.986541809876276e-05, "loss": 0.8368, "step": 33140 }, { "epoch": 0.03, "learning_rate": 4.986540995935296e-05, "loss": 0.6477, "step": 33141 }, { "epoch": 0.03, "learning_rate": 4.986540181969769e-05, "loss": 0.5584, "step": 33142 }, { "epoch": 0.03, "learning_rate": 4.986539367979697e-05, "loss": 0.8834, "step": 33143 }, { "epoch": 0.03, "learning_rate": 4.986538553965078e-05, "loss": 0.8545, "step": 33144 }, { "epoch": 0.03, "learning_rate": 4.986537739925914e-05, "loss": 0.7762, "step": 33145 }, { "epoch": 0.03, "learning_rate": 4.986536925862203e-05, "loss": 1.3972, "step": 33146 }, { "epoch": 0.03, "learning_rate": 4.986536111773946e-05, "loss": 1.3655, "step": 33147 }, { "epoch": 0.03, "learning_rate": 4.986535297661143e-05, "loss": 0.8172, "step": 33148 }, { "epoch": 0.03, "learning_rate": 4.986534483523795e-05, "loss": 0.5255, "step": 33149 }, { "epoch": 0.03, "learning_rate": 4.9865336693619e-05, "loss": 0.8313, "step": 33150 }, { "epoch": 0.03, "learning_rate": 4.9865328551754594e-05, "loss": 1.1157, "step": 33151 }, { "epoch": 0.03, "learning_rate": 4.986532040964472e-05, "loss": 1.0486, "step": 33152 }, { "epoch": 0.03, "learning_rate": 4.9865312267289396e-05, "loss": 0.7746, "step": 33153 }, { "epoch": 0.03, "learning_rate": 4.9865304124688606e-05, "loss": 0.8486, "step": 33154 }, { "epoch": 0.03, "learning_rate": 4.986529598184236e-05, "loss": 1.0784, "step": 33155 }, { "epoch": 0.03, "learning_rate": 4.986528783875065e-05, "loss": 1.3022, "step": 33156 }, { "epoch": 0.03, "learning_rate": 4.986527969541348e-05, "loss": 0.8704, "step": 33157 }, { "epoch": 0.03, "learning_rate": 4.986527155183086e-05, "loss": 1.0591, "step": 33158 }, { "epoch": 0.03, "learning_rate": 4.986526340800277e-05, "loss": 1.0008, "step": 33159 }, { "epoch": 0.03, "learning_rate": 4.986525526392922e-05, "loss": 1.1886, "step": 33160 }, { "epoch": 0.03, "learning_rate": 4.986524711961022e-05, "loss": 0.9948, "step": 33161 }, { "epoch": 0.03, "learning_rate": 4.986523897504576e-05, "loss": 0.9535, "step": 33162 }, { "epoch": 0.03, "learning_rate": 4.986523083023583e-05, "loss": 0.9846, "step": 33163 }, { "epoch": 0.03, "learning_rate": 4.986522268518045e-05, "loss": 0.9877, "step": 33164 }, { "epoch": 0.03, "learning_rate": 4.986521453987961e-05, "loss": 1.1791, "step": 33165 }, { "epoch": 0.03, "learning_rate": 4.986520639433331e-05, "loss": 0.7762, "step": 33166 }, { "epoch": 0.03, "learning_rate": 4.986519824854154e-05, "loss": 1.0274, "step": 33167 }, { "epoch": 0.03, "learning_rate": 4.986519010250433e-05, "loss": 1.1417, "step": 33168 }, { "epoch": 0.03, "learning_rate": 4.986518195622165e-05, "loss": 0.8603, "step": 33169 }, { "epoch": 0.03, "learning_rate": 4.986517380969352e-05, "loss": 0.4588, "step": 33170 }, { "epoch": 0.03, "learning_rate": 4.9865165662919924e-05, "loss": 0.8191, "step": 33171 }, { "epoch": 0.03, "learning_rate": 4.9865157515900874e-05, "loss": 0.6148, "step": 33172 }, { "epoch": 0.03, "learning_rate": 4.986514936863636e-05, "loss": 1.1422, "step": 33173 }, { "epoch": 0.03, "learning_rate": 4.986514122112639e-05, "loss": 0.7298, "step": 33174 }, { "epoch": 0.03, "learning_rate": 4.9865133073370965e-05, "loss": 0.7888, "step": 33175 }, { "epoch": 0.03, "learning_rate": 4.9865124925370075e-05, "loss": 0.8774, "step": 33176 }, { "epoch": 0.03, "learning_rate": 4.9865116777123735e-05, "loss": 0.8949, "step": 33177 }, { "epoch": 0.03, "learning_rate": 4.986510862863193e-05, "loss": 0.9649, "step": 33178 }, { "epoch": 0.03, "learning_rate": 4.986510047989468e-05, "loss": 1.0939, "step": 33179 }, { "epoch": 0.03, "learning_rate": 4.986509233091196e-05, "loss": 0.8991, "step": 33180 }, { "epoch": 0.03, "learning_rate": 4.9865084181683784e-05, "loss": 1.1394, "step": 33181 }, { "epoch": 0.03, "learning_rate": 4.986507603221015e-05, "loss": 1.3993, "step": 33182 }, { "epoch": 0.03, "learning_rate": 4.986506788249106e-05, "loss": 0.9681, "step": 33183 }, { "epoch": 0.03, "learning_rate": 4.986505973252651e-05, "loss": 0.6848, "step": 33184 }, { "epoch": 0.03, "learning_rate": 4.9865051582316514e-05, "loss": 0.4539, "step": 33185 }, { "epoch": 0.03, "learning_rate": 4.9865043431861045e-05, "loss": 0.8712, "step": 33186 }, { "epoch": 0.03, "learning_rate": 4.9865035281160126e-05, "loss": 0.9651, "step": 33187 }, { "epoch": 0.03, "learning_rate": 4.986502713021375e-05, "loss": 1.1486, "step": 33188 }, { "epoch": 0.03, "learning_rate": 4.986501897902192e-05, "loss": 1.1465, "step": 33189 }, { "epoch": 0.03, "learning_rate": 4.986501082758462e-05, "loss": 0.7702, "step": 33190 }, { "epoch": 0.03, "learning_rate": 4.986500267590188e-05, "loss": 1.0391, "step": 33191 }, { "epoch": 0.03, "learning_rate": 4.9864994523973675e-05, "loss": 1.1194, "step": 33192 }, { "epoch": 0.03, "learning_rate": 4.9864986371800016e-05, "loss": 0.559, "step": 33193 }, { "epoch": 0.03, "learning_rate": 4.98649782193809e-05, "loss": 0.8596, "step": 33194 }, { "epoch": 0.03, "learning_rate": 4.986497006671632e-05, "loss": 1.1285, "step": 33195 }, { "epoch": 0.03, "learning_rate": 4.9864961913806294e-05, "loss": 0.8833, "step": 33196 }, { "epoch": 0.03, "learning_rate": 4.986495376065081e-05, "loss": 0.8909, "step": 33197 }, { "epoch": 0.03, "learning_rate": 4.986494560724986e-05, "loss": 1.0831, "step": 33198 }, { "epoch": 0.03, "learning_rate": 4.986493745360347e-05, "loss": 1.073, "step": 33199 }, { "epoch": 0.03, "learning_rate": 4.986492929971161e-05, "loss": 1.0871, "step": 33200 }, { "epoch": 0.03, "learning_rate": 4.9864921145574296e-05, "loss": 1.0656, "step": 33201 }, { "epoch": 0.03, "learning_rate": 4.986491299119153e-05, "loss": 0.869, "step": 33202 }, { "epoch": 0.03, "learning_rate": 4.9864904836563303e-05, "loss": 1.1164, "step": 33203 }, { "epoch": 0.03, "learning_rate": 4.986489668168962e-05, "loss": 0.9311, "step": 33204 }, { "epoch": 0.03, "learning_rate": 4.9864888526570484e-05, "loss": 0.9337, "step": 33205 }, { "epoch": 0.03, "learning_rate": 4.98648803712059e-05, "loss": 0.9319, "step": 33206 }, { "epoch": 0.03, "learning_rate": 4.9864872215595846e-05, "loss": 0.8306, "step": 33207 }, { "epoch": 0.03, "learning_rate": 4.9864864059740345e-05, "loss": 0.9204, "step": 33208 }, { "epoch": 0.03, "learning_rate": 4.986485590363939e-05, "loss": 0.8528, "step": 33209 }, { "epoch": 0.03, "learning_rate": 4.9864847747292974e-05, "loss": 1.3852, "step": 33210 }, { "epoch": 0.03, "learning_rate": 4.986483959070111e-05, "loss": 0.8442, "step": 33211 }, { "epoch": 0.03, "learning_rate": 4.9864831433863776e-05, "loss": 0.7281, "step": 33212 }, { "epoch": 0.03, "learning_rate": 4.9864823276781006e-05, "loss": 1.1583, "step": 33213 }, { "epoch": 0.03, "learning_rate": 4.9864815119452766e-05, "loss": 1.1781, "step": 33214 }, { "epoch": 0.03, "learning_rate": 4.986480696187908e-05, "loss": 0.7323, "step": 33215 }, { "epoch": 0.03, "learning_rate": 4.9864798804059935e-05, "loss": 1.0798, "step": 33216 }, { "epoch": 0.03, "learning_rate": 4.986479064599534e-05, "loss": 1.0151, "step": 33217 }, { "epoch": 0.03, "learning_rate": 4.986478248768528e-05, "loss": 1.2913, "step": 33218 }, { "epoch": 0.03, "learning_rate": 4.9864774329129776e-05, "loss": 1.2798, "step": 33219 }, { "epoch": 0.03, "learning_rate": 4.9864766170328816e-05, "loss": 0.8936, "step": 33220 }, { "epoch": 0.03, "learning_rate": 4.98647580112824e-05, "loss": 0.5811, "step": 33221 }, { "epoch": 0.03, "learning_rate": 4.986474985199052e-05, "loss": 0.6489, "step": 33222 }, { "epoch": 0.03, "learning_rate": 4.98647416924532e-05, "loss": 0.988, "step": 33223 }, { "epoch": 0.03, "learning_rate": 4.986473353267042e-05, "loss": 0.6386, "step": 33224 }, { "epoch": 0.03, "learning_rate": 4.986472537264218e-05, "loss": 0.7838, "step": 33225 }, { "epoch": 0.03, "learning_rate": 4.98647172123685e-05, "loss": 0.8724, "step": 33226 }, { "epoch": 0.03, "learning_rate": 4.9864709051849355e-05, "loss": 0.7571, "step": 33227 }, { "epoch": 0.03, "learning_rate": 4.9864700891084756e-05, "loss": 0.9087, "step": 33228 }, { "epoch": 0.03, "learning_rate": 4.98646927300747e-05, "loss": 0.8928, "step": 33229 }, { "epoch": 0.03, "learning_rate": 4.98646845688192e-05, "loss": 0.9023, "step": 33230 }, { "epoch": 0.03, "learning_rate": 4.986467640731825e-05, "loss": 1.2251, "step": 33231 }, { "epoch": 0.03, "learning_rate": 4.9864668245571836e-05, "loss": 0.9686, "step": 33232 }, { "epoch": 0.03, "learning_rate": 4.986466008357997e-05, "loss": 1.1483, "step": 33233 }, { "epoch": 0.03, "learning_rate": 4.9864651921342656e-05, "loss": 0.4775, "step": 33234 }, { "epoch": 0.03, "learning_rate": 4.9864643758859875e-05, "loss": 0.7181, "step": 33235 }, { "epoch": 0.03, "learning_rate": 4.986463559613165e-05, "loss": 0.9432, "step": 33236 }, { "epoch": 0.03, "learning_rate": 4.9864627433157976e-05, "loss": 1.0283, "step": 33237 }, { "epoch": 0.03, "learning_rate": 4.9864619269938845e-05, "loss": 0.9469, "step": 33238 }, { "epoch": 0.03, "learning_rate": 4.986461110647426e-05, "loss": 0.9334, "step": 33239 }, { "epoch": 0.03, "learning_rate": 4.986460294276422e-05, "loss": 0.9711, "step": 33240 }, { "epoch": 0.03, "learning_rate": 4.986459477880873e-05, "loss": 0.9848, "step": 33241 }, { "epoch": 0.03, "learning_rate": 4.986458661460779e-05, "loss": 0.908, "step": 33242 }, { "epoch": 0.03, "learning_rate": 4.986457845016139e-05, "loss": 1.0281, "step": 33243 }, { "epoch": 0.03, "learning_rate": 4.9864570285469546e-05, "loss": 0.7234, "step": 33244 }, { "epoch": 0.03, "learning_rate": 4.9864562120532246e-05, "loss": 0.9693, "step": 33245 }, { "epoch": 0.03, "learning_rate": 4.986455395534949e-05, "loss": 0.9946, "step": 33246 }, { "epoch": 0.03, "learning_rate": 4.986454578992128e-05, "loss": 1.405, "step": 33247 }, { "epoch": 0.03, "learning_rate": 4.986453762424763e-05, "loss": 1.0591, "step": 33248 }, { "epoch": 0.03, "learning_rate": 4.9864529458328514e-05, "loss": 1.2111, "step": 33249 }, { "epoch": 0.03, "learning_rate": 4.986452129216396e-05, "loss": 0.5647, "step": 33250 }, { "epoch": 0.03, "learning_rate": 4.986451312575394e-05, "loss": 0.4444, "step": 33251 }, { "epoch": 0.03, "learning_rate": 4.986450495909848e-05, "loss": 1.0604, "step": 33252 }, { "epoch": 0.03, "learning_rate": 4.986449679219756e-05, "loss": 1.2643, "step": 33253 }, { "epoch": 0.03, "learning_rate": 4.986448862505119e-05, "loss": 0.5515, "step": 33254 }, { "epoch": 0.03, "learning_rate": 4.9864480457659364e-05, "loss": 1.1929, "step": 33255 }, { "epoch": 0.03, "learning_rate": 4.986447229002209e-05, "loss": 1.2093, "step": 33256 }, { "epoch": 0.03, "learning_rate": 4.9864464122139364e-05, "loss": 0.7555, "step": 33257 }, { "epoch": 0.03, "learning_rate": 4.986445595401118e-05, "loss": 0.3241, "step": 33258 }, { "epoch": 0.03, "learning_rate": 4.986444778563756e-05, "loss": 0.3535, "step": 33259 }, { "epoch": 0.03, "learning_rate": 4.986443961701848e-05, "loss": 0.2637, "step": 33260 }, { "epoch": 0.03, "learning_rate": 4.9864431448153946e-05, "loss": 0.2261, "step": 33261 }, { "epoch": 0.03, "learning_rate": 4.9864423279043965e-05, "loss": 0.2062, "step": 33262 }, { "epoch": 0.03, "learning_rate": 4.9864415109688535e-05, "loss": 0.2562, "step": 33263 }, { "epoch": 0.03, "learning_rate": 4.986440694008765e-05, "loss": 0.1917, "step": 33264 }, { "epoch": 0.03, "learning_rate": 4.986439877024132e-05, "loss": 0.2352, "step": 33265 }, { "epoch": 0.03, "learning_rate": 4.9864390600149524e-05, "loss": 0.8907, "step": 33266 }, { "epoch": 0.03, "learning_rate": 4.986438242981229e-05, "loss": 0.7324, "step": 33267 }, { "epoch": 0.03, "learning_rate": 4.98643742592296e-05, "loss": 0.994, "step": 33268 }, { "epoch": 0.03, "learning_rate": 4.9864366088401464e-05, "loss": 1.0016, "step": 33269 }, { "epoch": 0.03, "learning_rate": 4.986435791732788e-05, "loss": 0.8314, "step": 33270 }, { "epoch": 0.03, "learning_rate": 4.9864349746008836e-05, "loss": 0.7881, "step": 33271 }, { "epoch": 0.03, "learning_rate": 4.986434157444434e-05, "loss": 1.4851, "step": 33272 }, { "epoch": 0.03, "learning_rate": 4.986433340263441e-05, "loss": 1.4032, "step": 33273 }, { "epoch": 0.03, "learning_rate": 4.9864325230579015e-05, "loss": 1.0758, "step": 33274 }, { "epoch": 0.03, "learning_rate": 4.9864317058278174e-05, "loss": 0.9204, "step": 33275 }, { "epoch": 0.03, "learning_rate": 4.986430888573189e-05, "loss": 0.9876, "step": 33276 }, { "epoch": 0.03, "learning_rate": 4.986430071294015e-05, "loss": 1.2784, "step": 33277 }, { "epoch": 0.03, "learning_rate": 4.9864292539902956e-05, "loss": 1.1692, "step": 33278 }, { "epoch": 0.03, "learning_rate": 4.9864284366620314e-05, "loss": 1.1482, "step": 33279 }, { "epoch": 0.03, "learning_rate": 4.986427619309223e-05, "loss": 1.2649, "step": 33280 }, { "epoch": 0.03, "learning_rate": 4.986426801931868e-05, "loss": 1.1967, "step": 33281 }, { "epoch": 0.03, "learning_rate": 4.98642598452997e-05, "loss": 1.1917, "step": 33282 }, { "epoch": 0.03, "learning_rate": 4.986425167103526e-05, "loss": 1.0152, "step": 33283 }, { "epoch": 0.03, "learning_rate": 4.986424349652537e-05, "loss": 1.2074, "step": 33284 }, { "epoch": 0.03, "learning_rate": 4.9864235321770034e-05, "loss": 1.0559, "step": 33285 }, { "epoch": 0.03, "learning_rate": 4.986422714676925e-05, "loss": 0.9479, "step": 33286 }, { "epoch": 0.03, "learning_rate": 4.986421897152301e-05, "loss": 1.3159, "step": 33287 }, { "epoch": 0.03, "learning_rate": 4.986421079603133e-05, "loss": 1.0437, "step": 33288 }, { "epoch": 0.03, "learning_rate": 4.9864202620294196e-05, "loss": 0.899, "step": 33289 }, { "epoch": 0.03, "learning_rate": 4.9864194444311606e-05, "loss": 1.1748, "step": 33290 }, { "epoch": 0.03, "learning_rate": 4.986418626808358e-05, "loss": 1.0876, "step": 33291 }, { "epoch": 0.03, "learning_rate": 4.9864178091610106e-05, "loss": 1.0943, "step": 33292 }, { "epoch": 0.03, "learning_rate": 4.986416991489117e-05, "loss": 1.0308, "step": 33293 }, { "epoch": 0.03, "learning_rate": 4.98641617379268e-05, "loss": 1.0171, "step": 33294 }, { "epoch": 0.03, "learning_rate": 4.986415356071697e-05, "loss": 1.0481, "step": 33295 }, { "epoch": 0.03, "learning_rate": 4.986414538326169e-05, "loss": 1.0538, "step": 33296 }, { "epoch": 0.03, "learning_rate": 4.986413720556097e-05, "loss": 0.9384, "step": 33297 }, { "epoch": 0.03, "learning_rate": 4.98641290276148e-05, "loss": 0.5404, "step": 33298 }, { "epoch": 0.03, "learning_rate": 4.9864120849423177e-05, "loss": 0.7994, "step": 33299 }, { "epoch": 0.03, "learning_rate": 4.986411267098611e-05, "loss": 0.7401, "step": 33300 }, { "epoch": 0.03, "learning_rate": 4.98641044923036e-05, "loss": 1.0383, "step": 33301 }, { "epoch": 0.03, "learning_rate": 4.986409631337563e-05, "loss": 0.531, "step": 33302 }, { "epoch": 0.03, "learning_rate": 4.9864088134202216e-05, "loss": 0.9411, "step": 33303 }, { "epoch": 0.03, "learning_rate": 4.986407995478336e-05, "loss": 1.3836, "step": 33304 }, { "epoch": 0.03, "learning_rate": 4.9864071775119054e-05, "loss": 1.0159, "step": 33305 }, { "epoch": 0.03, "learning_rate": 4.98640635952093e-05, "loss": 1.0094, "step": 33306 }, { "epoch": 0.03, "learning_rate": 4.98640554150541e-05, "loss": 0.8566, "step": 33307 }, { "epoch": 0.03, "learning_rate": 4.986404723465345e-05, "loss": 0.7517, "step": 33308 }, { "epoch": 0.03, "learning_rate": 4.986403905400735e-05, "loss": 0.688, "step": 33309 }, { "epoch": 0.03, "learning_rate": 4.986403087311581e-05, "loss": 0.4377, "step": 33310 }, { "epoch": 0.03, "learning_rate": 4.986402269197882e-05, "loss": 0.4737, "step": 33311 }, { "epoch": 0.03, "learning_rate": 4.986401451059638e-05, "loss": 0.675, "step": 33312 }, { "epoch": 0.03, "learning_rate": 4.986400632896849e-05, "loss": 1.0879, "step": 33313 }, { "epoch": 0.03, "learning_rate": 4.9863998147095156e-05, "loss": 1.1017, "step": 33314 }, { "epoch": 0.03, "learning_rate": 4.9863989964976375e-05, "loss": 1.2489, "step": 33315 }, { "epoch": 0.03, "learning_rate": 4.986398178261215e-05, "loss": 1.0957, "step": 33316 }, { "epoch": 0.03, "learning_rate": 4.986397360000248e-05, "loss": 0.6815, "step": 33317 }, { "epoch": 0.03, "learning_rate": 4.986396541714736e-05, "loss": 0.751, "step": 33318 }, { "epoch": 0.03, "learning_rate": 4.9863957234046795e-05, "loss": 0.2276, "step": 33319 }, { "epoch": 0.03, "learning_rate": 4.986394905070077e-05, "loss": 0.2688, "step": 33320 }, { "epoch": 0.03, "learning_rate": 4.9863940867109314e-05, "loss": 0.1528, "step": 33321 }, { "epoch": 0.03, "learning_rate": 4.986393268327241e-05, "loss": 0.2781, "step": 33322 }, { "epoch": 0.03, "learning_rate": 4.986392449919006e-05, "loss": 0.2825, "step": 33323 }, { "epoch": 0.03, "learning_rate": 4.9863916314862264e-05, "loss": 0.2995, "step": 33324 }, { "epoch": 0.03, "learning_rate": 4.986390813028902e-05, "loss": 0.5413, "step": 33325 }, { "epoch": 0.03, "learning_rate": 4.9863899945470325e-05, "loss": 1.2495, "step": 33326 }, { "epoch": 0.03, "learning_rate": 4.986389176040619e-05, "loss": 1.3056, "step": 33327 }, { "epoch": 0.03, "learning_rate": 4.9863883575096606e-05, "loss": 1.4484, "step": 33328 }, { "epoch": 0.03, "learning_rate": 4.986387538954158e-05, "loss": 0.6052, "step": 33329 }, { "epoch": 0.03, "learning_rate": 4.986386720374111e-05, "loss": 0.7994, "step": 33330 }, { "epoch": 0.03, "learning_rate": 4.986385901769519e-05, "loss": 0.437, "step": 33331 }, { "epoch": 0.03, "learning_rate": 4.9863850831403826e-05, "loss": 0.4779, "step": 33332 }, { "epoch": 0.03, "learning_rate": 4.9863842644867016e-05, "loss": 0.6643, "step": 33333 }, { "epoch": 0.03, "learning_rate": 4.986383445808476e-05, "loss": 1.0035, "step": 33334 }, { "epoch": 0.03, "learning_rate": 4.9863826271057055e-05, "loss": 0.9534, "step": 33335 }, { "epoch": 0.03, "learning_rate": 4.986381808378391e-05, "loss": 0.8191, "step": 33336 }, { "epoch": 0.03, "learning_rate": 4.986380989626532e-05, "loss": 0.6711, "step": 33337 }, { "epoch": 0.03, "learning_rate": 4.986380170850128e-05, "loss": 0.4215, "step": 33338 }, { "epoch": 0.03, "learning_rate": 4.98637935204918e-05, "loss": 1.0389, "step": 33339 }, { "epoch": 0.03, "learning_rate": 4.986378533223688e-05, "loss": 1.3317, "step": 33340 }, { "epoch": 0.03, "learning_rate": 4.98637771437365e-05, "loss": 1.2977, "step": 33341 }, { "epoch": 0.03, "learning_rate": 4.9863768954990686e-05, "loss": 1.4446, "step": 33342 }, { "epoch": 0.03, "learning_rate": 4.986376076599943e-05, "loss": 1.2198, "step": 33343 }, { "epoch": 0.03, "learning_rate": 4.986375257676272e-05, "loss": 0.352, "step": 33344 }, { "epoch": 0.03, "learning_rate": 4.9863744387280576e-05, "loss": 0.5044, "step": 33345 }, { "epoch": 0.03, "learning_rate": 4.986373619755298e-05, "loss": 0.3465, "step": 33346 }, { "epoch": 0.03, "learning_rate": 4.986372800757994e-05, "loss": 0.424, "step": 33347 }, { "epoch": 0.03, "learning_rate": 4.986371981736146e-05, "loss": 0.4246, "step": 33348 }, { "epoch": 0.03, "learning_rate": 4.986371162689753e-05, "loss": 1.3911, "step": 33349 }, { "epoch": 0.03, "learning_rate": 4.986370343618816e-05, "loss": 0.9933, "step": 33350 }, { "epoch": 0.03, "learning_rate": 4.9863695245233346e-05, "loss": 1.2561, "step": 33351 }, { "epoch": 0.03, "learning_rate": 4.9863687054033085e-05, "loss": 1.129, "step": 33352 }, { "epoch": 0.03, "learning_rate": 4.986367886258738e-05, "loss": 1.0901, "step": 33353 }, { "epoch": 0.03, "learning_rate": 4.986367067089624e-05, "loss": 0.9852, "step": 33354 }, { "epoch": 0.03, "learning_rate": 4.986366247895965e-05, "loss": 0.8538, "step": 33355 }, { "epoch": 0.03, "learning_rate": 4.9863654286777616e-05, "loss": 1.0462, "step": 33356 }, { "epoch": 0.03, "learning_rate": 4.986364609435013e-05, "loss": 0.993, "step": 33357 }, { "epoch": 0.03, "learning_rate": 4.986363790167722e-05, "loss": 1.0222, "step": 33358 }, { "epoch": 0.03, "learning_rate": 4.986362970875885e-05, "loss": 1.2745, "step": 33359 }, { "epoch": 0.03, "learning_rate": 4.986362151559505e-05, "loss": 1.2363, "step": 33360 }, { "epoch": 0.03, "learning_rate": 4.98636133221858e-05, "loss": 0.9082, "step": 33361 }, { "epoch": 0.03, "learning_rate": 4.9863605128531105e-05, "loss": 0.9747, "step": 33362 }, { "epoch": 0.03, "learning_rate": 4.986359693463097e-05, "loss": 0.659, "step": 33363 }, { "epoch": 0.03, "learning_rate": 4.9863588740485386e-05, "loss": 0.4724, "step": 33364 }, { "epoch": 0.03, "learning_rate": 4.9863580546094366e-05, "loss": 0.4628, "step": 33365 }, { "epoch": 0.03, "learning_rate": 4.98635723514579e-05, "loss": 0.8361, "step": 33366 }, { "epoch": 0.03, "learning_rate": 4.9863564156575995e-05, "loss": 0.9667, "step": 33367 }, { "epoch": 0.03, "learning_rate": 4.9863555961448645e-05, "loss": 1.0612, "step": 33368 }, { "epoch": 0.03, "learning_rate": 4.9863547766075845e-05, "loss": 0.8664, "step": 33369 }, { "epoch": 0.03, "learning_rate": 4.9863539570457616e-05, "loss": 1.1217, "step": 33370 }, { "epoch": 0.03, "learning_rate": 4.986353137459394e-05, "loss": 0.9522, "step": 33371 }, { "epoch": 0.03, "learning_rate": 4.9863523178484815e-05, "loss": 1.0418, "step": 33372 }, { "epoch": 0.03, "learning_rate": 4.986351498213026e-05, "loss": 0.6988, "step": 33373 }, { "epoch": 0.03, "learning_rate": 4.986350678553025e-05, "loss": 0.8279, "step": 33374 }, { "epoch": 0.03, "learning_rate": 4.9863498588684804e-05, "loss": 1.2838, "step": 33375 }, { "epoch": 0.03, "learning_rate": 4.9863490391593917e-05, "loss": 0.5169, "step": 33376 }, { "epoch": 0.03, "learning_rate": 4.986348219425758e-05, "loss": 1.1723, "step": 33377 }, { "epoch": 0.03, "learning_rate": 4.986347399667581e-05, "loss": 0.9671, "step": 33378 }, { "epoch": 0.03, "learning_rate": 4.9863465798848596e-05, "loss": 1.2153, "step": 33379 }, { "epoch": 0.03, "learning_rate": 4.986345760077594e-05, "loss": 1.1858, "step": 33380 }, { "epoch": 0.03, "learning_rate": 4.986344940245785e-05, "loss": 0.818, "step": 33381 }, { "epoch": 0.03, "learning_rate": 4.986344120389431e-05, "loss": 0.3974, "step": 33382 }, { "epoch": 0.03, "learning_rate": 4.9863433005085326e-05, "loss": 1.079, "step": 33383 }, { "epoch": 0.03, "learning_rate": 4.98634248060309e-05, "loss": 1.0841, "step": 33384 }, { "epoch": 0.03, "learning_rate": 4.986341660673104e-05, "loss": 0.9409, "step": 33385 }, { "epoch": 0.03, "learning_rate": 4.986340840718574e-05, "loss": 0.7624, "step": 33386 }, { "epoch": 0.03, "learning_rate": 4.9863400207394995e-05, "loss": 0.6385, "step": 33387 }, { "epoch": 0.03, "learning_rate": 4.9863392007358805e-05, "loss": 0.9201, "step": 33388 }, { "epoch": 0.03, "learning_rate": 4.986338380707718e-05, "loss": 0.8601, "step": 33389 }, { "epoch": 0.03, "learning_rate": 4.986337560655011e-05, "loss": 0.8799, "step": 33390 }, { "epoch": 0.03, "learning_rate": 4.986336740577761e-05, "loss": 1.323, "step": 33391 }, { "epoch": 0.03, "learning_rate": 4.986335920475965e-05, "loss": 0.9182, "step": 33392 }, { "epoch": 0.03, "learning_rate": 4.986335100349626e-05, "loss": 0.9512, "step": 33393 }, { "epoch": 0.03, "learning_rate": 4.986334280198743e-05, "loss": 1.08, "step": 33394 }, { "epoch": 0.03, "learning_rate": 4.986333460023316e-05, "loss": 0.8034, "step": 33395 }, { "epoch": 0.03, "learning_rate": 4.9863326398233454e-05, "loss": 0.8425, "step": 33396 }, { "epoch": 0.03, "learning_rate": 4.98633181959883e-05, "loss": 0.6959, "step": 33397 }, { "epoch": 0.03, "learning_rate": 4.986330999349771e-05, "loss": 0.8993, "step": 33398 }, { "epoch": 0.03, "learning_rate": 4.986330179076167e-05, "loss": 0.8056, "step": 33399 }, { "epoch": 0.03, "learning_rate": 4.98632935877802e-05, "loss": 1.0296, "step": 33400 }, { "epoch": 0.03, "learning_rate": 4.986328538455329e-05, "loss": 1.1833, "step": 33401 }, { "epoch": 0.03, "learning_rate": 4.9863277181080944e-05, "loss": 1.1059, "step": 33402 }, { "epoch": 0.03, "learning_rate": 4.9863268977363156e-05, "loss": 1.2165, "step": 33403 }, { "epoch": 0.03, "learning_rate": 4.9863260773399924e-05, "loss": 0.7888, "step": 33404 }, { "epoch": 0.03, "learning_rate": 4.986325256919125e-05, "loss": 1.0692, "step": 33405 }, { "epoch": 0.03, "learning_rate": 4.986324436473714e-05, "loss": 1.0684, "step": 33406 }, { "epoch": 0.03, "learning_rate": 4.9863236160037584e-05, "loss": 1.6892, "step": 33407 }, { "epoch": 0.03, "learning_rate": 4.98632279550926e-05, "loss": 0.9303, "step": 33408 }, { "epoch": 0.03, "learning_rate": 4.986321974990217e-05, "loss": 0.8978, "step": 33409 }, { "epoch": 0.03, "learning_rate": 4.9863211544466304e-05, "loss": 0.8035, "step": 33410 }, { "epoch": 0.03, "learning_rate": 4.9863203338785e-05, "loss": 1.0848, "step": 33411 }, { "epoch": 0.03, "learning_rate": 4.9863195132858256e-05, "loss": 1.2578, "step": 33412 }, { "epoch": 0.03, "learning_rate": 4.9863186926686064e-05, "loss": 0.8859, "step": 33413 }, { "epoch": 0.03, "learning_rate": 4.986317872026844e-05, "loss": 0.993, "step": 33414 }, { "epoch": 0.03, "learning_rate": 4.986317051360538e-05, "loss": 0.874, "step": 33415 }, { "epoch": 0.03, "learning_rate": 4.986316230669688e-05, "loss": 0.9763, "step": 33416 }, { "epoch": 0.03, "learning_rate": 4.986315409954294e-05, "loss": 1.3388, "step": 33417 }, { "epoch": 0.03, "learning_rate": 4.986314589214356e-05, "loss": 1.3318, "step": 33418 }, { "epoch": 0.03, "learning_rate": 4.986313768449874e-05, "loss": 0.8742, "step": 33419 }, { "epoch": 0.03, "learning_rate": 4.9863129476608485e-05, "loss": 0.7141, "step": 33420 }, { "epoch": 0.03, "learning_rate": 4.986312126847279e-05, "loss": 1.1144, "step": 33421 }, { "epoch": 0.03, "learning_rate": 4.986311306009166e-05, "loss": 1.2839, "step": 33422 }, { "epoch": 0.03, "learning_rate": 4.986310485146509e-05, "loss": 1.4322, "step": 33423 }, { "epoch": 0.03, "learning_rate": 4.9863096642593085e-05, "loss": 1.0171, "step": 33424 }, { "epoch": 0.03, "learning_rate": 4.986308843347564e-05, "loss": 0.9977, "step": 33425 }, { "epoch": 0.03, "learning_rate": 4.9863080224112753e-05, "loss": 1.0725, "step": 33426 }, { "epoch": 0.03, "learning_rate": 4.9863072014504433e-05, "loss": 1.6619, "step": 33427 }, { "epoch": 0.03, "learning_rate": 4.986306380465067e-05, "loss": 0.915, "step": 33428 }, { "epoch": 0.03, "learning_rate": 4.986305559455147e-05, "loss": 1.1427, "step": 33429 }, { "epoch": 0.03, "learning_rate": 4.9863047384206835e-05, "loss": 0.8539, "step": 33430 }, { "epoch": 0.03, "learning_rate": 4.9863039173616763e-05, "loss": 1.135, "step": 33431 }, { "epoch": 0.03, "learning_rate": 4.9863030962781255e-05, "loss": 0.8316, "step": 33432 }, { "epoch": 0.03, "learning_rate": 4.9863022751700304e-05, "loss": 0.7879, "step": 33433 }, { "epoch": 0.03, "learning_rate": 4.9863014540373923e-05, "loss": 0.5543, "step": 33434 }, { "epoch": 0.03, "learning_rate": 4.98630063288021e-05, "loss": 0.9223, "step": 33435 }, { "epoch": 0.03, "learning_rate": 4.986299811698484e-05, "loss": 0.9571, "step": 33436 }, { "epoch": 0.03, "learning_rate": 4.9862989904922144e-05, "loss": 0.4349, "step": 33437 }, { "epoch": 0.03, "learning_rate": 4.986298169261401e-05, "loss": 0.502, "step": 33438 }, { "epoch": 0.03, "learning_rate": 4.9862973480060435e-05, "loss": 0.8122, "step": 33439 }, { "epoch": 0.03, "learning_rate": 4.986296526726143e-05, "loss": 1.1038, "step": 33440 }, { "epoch": 0.03, "learning_rate": 4.986295705421698e-05, "loss": 1.0493, "step": 33441 }, { "epoch": 0.03, "learning_rate": 4.986294884092711e-05, "loss": 0.8828, "step": 33442 }, { "epoch": 0.03, "learning_rate": 4.986294062739179e-05, "loss": 1.1571, "step": 33443 }, { "epoch": 0.03, "learning_rate": 4.986293241361104e-05, "loss": 1.1052, "step": 33444 }, { "epoch": 0.03, "learning_rate": 4.986292419958485e-05, "loss": 0.9533, "step": 33445 }, { "epoch": 0.03, "learning_rate": 4.986291598531322e-05, "loss": 0.7456, "step": 33446 }, { "epoch": 0.03, "learning_rate": 4.9862907770796166e-05, "loss": 0.621, "step": 33447 }, { "epoch": 0.03, "learning_rate": 4.986289955603366e-05, "loss": 0.9741, "step": 33448 }, { "epoch": 0.03, "learning_rate": 4.986289134102573e-05, "loss": 1.5571, "step": 33449 }, { "epoch": 0.03, "learning_rate": 4.986288312577236e-05, "loss": 1.0426, "step": 33450 }, { "epoch": 0.03, "learning_rate": 4.986287491027355e-05, "loss": 0.7291, "step": 33451 }, { "epoch": 0.03, "learning_rate": 4.986286669452931e-05, "loss": 0.7901, "step": 33452 }, { "epoch": 0.03, "learning_rate": 4.986285847853963e-05, "loss": 1.2969, "step": 33453 }, { "epoch": 0.03, "learning_rate": 4.986285026230452e-05, "loss": 1.3182, "step": 33454 }, { "epoch": 0.03, "learning_rate": 4.986284204582396e-05, "loss": 1.265, "step": 33455 }, { "epoch": 0.03, "learning_rate": 4.986283382909799e-05, "loss": 1.2099, "step": 33456 }, { "epoch": 0.03, "learning_rate": 4.986282561212656e-05, "loss": 1.2938, "step": 33457 }, { "epoch": 0.03, "learning_rate": 4.986281739490971e-05, "loss": 1.2396, "step": 33458 }, { "epoch": 0.03, "learning_rate": 4.986280917744742e-05, "loss": 0.5777, "step": 33459 }, { "epoch": 0.03, "learning_rate": 4.986280095973969e-05, "loss": 1.2069, "step": 33460 }, { "epoch": 0.03, "learning_rate": 4.986279274178653e-05, "loss": 1.4246, "step": 33461 }, { "epoch": 0.03, "learning_rate": 4.986278452358794e-05, "loss": 1.3444, "step": 33462 }, { "epoch": 0.03, "learning_rate": 4.9862776305143905e-05, "loss": 1.1258, "step": 33463 }, { "epoch": 0.03, "learning_rate": 4.9862768086454436e-05, "loss": 1.3033, "step": 33464 }, { "epoch": 0.03, "learning_rate": 4.986275986751954e-05, "loss": 0.9861, "step": 33465 }, { "epoch": 0.03, "learning_rate": 4.98627516483392e-05, "loss": 0.9587, "step": 33466 }, { "epoch": 0.03, "learning_rate": 4.986274342891343e-05, "loss": 1.0021, "step": 33467 }, { "epoch": 0.03, "learning_rate": 4.9862735209242226e-05, "loss": 0.9748, "step": 33468 }, { "epoch": 0.03, "learning_rate": 4.986272698932559e-05, "loss": 0.9272, "step": 33469 }, { "epoch": 0.03, "learning_rate": 4.9862718769163516e-05, "loss": 0.736, "step": 33470 }, { "epoch": 0.03, "learning_rate": 4.986271054875601e-05, "loss": 1.1809, "step": 33471 }, { "epoch": 0.03, "learning_rate": 4.986270232810307e-05, "loss": 1.337, "step": 33472 }, { "epoch": 0.03, "learning_rate": 4.986269410720469e-05, "loss": 0.9071, "step": 33473 }, { "epoch": 0.03, "learning_rate": 4.986268588606088e-05, "loss": 1.0152, "step": 33474 }, { "epoch": 0.03, "learning_rate": 4.986267766467164e-05, "loss": 1.0829, "step": 33475 }, { "epoch": 0.03, "learning_rate": 4.986266944303696e-05, "loss": 0.8169, "step": 33476 }, { "epoch": 0.03, "learning_rate": 4.9862661221156846e-05, "loss": 1.0335, "step": 33477 }, { "epoch": 0.03, "learning_rate": 4.9862652999031304e-05, "loss": 1.1123, "step": 33478 }, { "epoch": 0.03, "learning_rate": 4.9862644776660325e-05, "loss": 0.6938, "step": 33479 }, { "epoch": 0.03, "learning_rate": 4.9862636554043916e-05, "loss": 1.1105, "step": 33480 }, { "epoch": 0.03, "learning_rate": 4.986262833118207e-05, "loss": 0.906, "step": 33481 }, { "epoch": 0.03, "learning_rate": 4.9862620108074784e-05, "loss": 1.0405, "step": 33482 }, { "epoch": 0.03, "learning_rate": 4.9862611884722074e-05, "loss": 1.1375, "step": 33483 }, { "epoch": 0.03, "learning_rate": 4.986260366112393e-05, "loss": 0.6894, "step": 33484 }, { "epoch": 0.03, "learning_rate": 4.986259543728035e-05, "loss": 0.974, "step": 33485 }, { "epoch": 0.03, "learning_rate": 4.986258721319134e-05, "loss": 1.0184, "step": 33486 }, { "epoch": 0.03, "learning_rate": 4.98625789888569e-05, "loss": 1.1276, "step": 33487 }, { "epoch": 0.03, "learning_rate": 4.986257076427702e-05, "loss": 0.9802, "step": 33488 }, { "epoch": 0.03, "learning_rate": 4.9862562539451705e-05, "loss": 0.9898, "step": 33489 }, { "epoch": 0.03, "learning_rate": 4.986255431438096e-05, "loss": 1.0122, "step": 33490 }, { "epoch": 0.03, "learning_rate": 4.986254608906479e-05, "loss": 1.4325, "step": 33491 }, { "epoch": 0.03, "learning_rate": 4.986253786350318e-05, "loss": 1.2853, "step": 33492 }, { "epoch": 0.03, "learning_rate": 4.986252963769614e-05, "loss": 1.1527, "step": 33493 }, { "epoch": 0.03, "learning_rate": 4.9862521411643663e-05, "loss": 1.145, "step": 33494 }, { "epoch": 0.03, "learning_rate": 4.986251318534576e-05, "loss": 1.0031, "step": 33495 }, { "epoch": 0.03, "learning_rate": 4.986250495880242e-05, "loss": 0.7876, "step": 33496 }, { "epoch": 0.03, "learning_rate": 4.9862496732013654e-05, "loss": 1.0562, "step": 33497 }, { "epoch": 0.03, "learning_rate": 4.986248850497945e-05, "loss": 1.0032, "step": 33498 }, { "epoch": 0.03, "learning_rate": 4.986248027769981e-05, "loss": 0.8443, "step": 33499 }, { "epoch": 0.03, "learning_rate": 4.9862472050174745e-05, "loss": 0.3432, "step": 33500 }, { "epoch": 0.03, "eval_loss": 1.0758426189422607, "eval_runtime": 125.6818, "eval_samples_per_second": 11.02, "eval_steps_per_second": 5.514, "step": 33500 }, { "epoch": 0.03, "learning_rate": 4.986246382240425e-05, "loss": 0.7522, "step": 33501 }, { "epoch": 0.03, "learning_rate": 4.986245559438832e-05, "loss": 1.1444, "step": 33502 }, { "epoch": 0.03, "learning_rate": 4.9862447366126956e-05, "loss": 1.1291, "step": 33503 }, { "epoch": 0.03, "learning_rate": 4.9862439137620166e-05, "loss": 1.1523, "step": 33504 }, { "epoch": 0.03, "learning_rate": 4.986243090886794e-05, "loss": 1.1179, "step": 33505 }, { "epoch": 0.03, "learning_rate": 4.986242267987028e-05, "loss": 1.1741, "step": 33506 }, { "epoch": 0.03, "learning_rate": 4.986241445062719e-05, "loss": 1.1039, "step": 33507 }, { "epoch": 0.03, "learning_rate": 4.986240622113868e-05, "loss": 1.1525, "step": 33508 }, { "epoch": 0.03, "learning_rate": 4.986239799140473e-05, "loss": 1.0584, "step": 33509 }, { "epoch": 0.03, "learning_rate": 4.986238976142534e-05, "loss": 1.0825, "step": 33510 }, { "epoch": 0.03, "learning_rate": 4.986238153120053e-05, "loss": 1.3499, "step": 33511 }, { "epoch": 0.03, "learning_rate": 4.986237330073029e-05, "loss": 1.0896, "step": 33512 }, { "epoch": 0.03, "learning_rate": 4.9862365070014615e-05, "loss": 1.0143, "step": 33513 }, { "epoch": 0.03, "learning_rate": 4.9862356839053516e-05, "loss": 0.9727, "step": 33514 }, { "epoch": 0.03, "learning_rate": 4.9862348607846974e-05, "loss": 0.8683, "step": 33515 }, { "epoch": 0.03, "learning_rate": 4.986234037639501e-05, "loss": 1.1061, "step": 33516 }, { "epoch": 0.03, "learning_rate": 4.986233214469761e-05, "loss": 1.0572, "step": 33517 }, { "epoch": 0.03, "learning_rate": 4.9862323912754784e-05, "loss": 0.8339, "step": 33518 }, { "epoch": 0.03, "learning_rate": 4.986231568056653e-05, "loss": 0.9006, "step": 33519 }, { "epoch": 0.03, "learning_rate": 4.986230744813284e-05, "loss": 0.9088, "step": 33520 }, { "epoch": 0.03, "learning_rate": 4.986229921545372e-05, "loss": 1.1302, "step": 33521 }, { "epoch": 0.03, "learning_rate": 4.9862290982529174e-05, "loss": 1.5188, "step": 33522 }, { "epoch": 0.03, "learning_rate": 4.9862282749359196e-05, "loss": 0.9108, "step": 33523 }, { "epoch": 0.03, "learning_rate": 4.986227451594379e-05, "loss": 1.0158, "step": 33524 }, { "epoch": 0.03, "learning_rate": 4.986226628228295e-05, "loss": 1.0775, "step": 33525 }, { "epoch": 0.03, "learning_rate": 4.9862258048376685e-05, "loss": 1.3584, "step": 33526 }, { "epoch": 0.03, "learning_rate": 4.986224981422498e-05, "loss": 1.2812, "step": 33527 }, { "epoch": 0.03, "learning_rate": 4.9862241579827856e-05, "loss": 0.9884, "step": 33528 }, { "epoch": 0.03, "learning_rate": 4.98622333451853e-05, "loss": 0.7482, "step": 33529 }, { "epoch": 0.03, "learning_rate": 4.986222511029731e-05, "loss": 0.8926, "step": 33530 }, { "epoch": 0.03, "learning_rate": 4.9862216875163896e-05, "loss": 1.2694, "step": 33531 }, { "epoch": 0.03, "learning_rate": 4.986220863978506e-05, "loss": 1.1749, "step": 33532 }, { "epoch": 0.03, "learning_rate": 4.986220040416078e-05, "loss": 1.0435, "step": 33533 }, { "epoch": 0.03, "learning_rate": 4.986219216829108e-05, "loss": 0.8246, "step": 33534 }, { "epoch": 0.03, "learning_rate": 4.9862183932175946e-05, "loss": 1.0531, "step": 33535 }, { "epoch": 0.03, "learning_rate": 4.9862175695815384e-05, "loss": 0.8352, "step": 33536 }, { "epoch": 0.03, "learning_rate": 4.986216745920939e-05, "loss": 1.0182, "step": 33537 }, { "epoch": 0.03, "learning_rate": 4.986215922235797e-05, "loss": 1.0589, "step": 33538 }, { "epoch": 0.03, "learning_rate": 4.986215098526112e-05, "loss": 1.0467, "step": 33539 }, { "epoch": 0.03, "learning_rate": 4.986214274791885e-05, "loss": 1.1889, "step": 33540 }, { "epoch": 0.03, "learning_rate": 4.9862134510331135e-05, "loss": 1.0346, "step": 33541 }, { "epoch": 0.03, "learning_rate": 4.986212627249801e-05, "loss": 1.0149, "step": 33542 }, { "epoch": 0.03, "learning_rate": 4.986211803441945e-05, "loss": 0.8937, "step": 33543 }, { "epoch": 0.03, "learning_rate": 4.986210979609546e-05, "loss": 1.255, "step": 33544 }, { "epoch": 0.03, "learning_rate": 4.986210155752604e-05, "loss": 1.2988, "step": 33545 }, { "epoch": 0.03, "learning_rate": 4.986209331871119e-05, "loss": 1.1519, "step": 33546 }, { "epoch": 0.03, "learning_rate": 4.986208507965091e-05, "loss": 0.9646, "step": 33547 }, { "epoch": 0.03, "learning_rate": 4.986207684034521e-05, "loss": 0.9045, "step": 33548 }, { "epoch": 0.03, "learning_rate": 4.986206860079408e-05, "loss": 1.1882, "step": 33549 }, { "epoch": 0.03, "learning_rate": 4.986206036099752e-05, "loss": 0.7448, "step": 33550 }, { "epoch": 0.03, "learning_rate": 4.986205212095553e-05, "loss": 0.9056, "step": 33551 }, { "epoch": 0.03, "learning_rate": 4.986204388066812e-05, "loss": 0.9749, "step": 33552 }, { "epoch": 0.03, "learning_rate": 4.986203564013528e-05, "loss": 0.8442, "step": 33553 }, { "epoch": 0.03, "learning_rate": 4.9862027399357005e-05, "loss": 0.8127, "step": 33554 }, { "epoch": 0.03, "learning_rate": 4.9862019158333304e-05, "loss": 0.9743, "step": 33555 }, { "epoch": 0.03, "learning_rate": 4.986201091706419e-05, "loss": 0.8844, "step": 33556 }, { "epoch": 0.03, "learning_rate": 4.986200267554963e-05, "loss": 1.0277, "step": 33557 }, { "epoch": 0.03, "learning_rate": 4.986199443378965e-05, "loss": 1.0833, "step": 33558 }, { "epoch": 0.03, "learning_rate": 4.986198619178425e-05, "loss": 0.9822, "step": 33559 }, { "epoch": 0.03, "learning_rate": 4.986197794953341e-05, "loss": 0.6217, "step": 33560 }, { "epoch": 0.03, "learning_rate": 4.986196970703715e-05, "loss": 1.5138, "step": 33561 }, { "epoch": 0.03, "learning_rate": 4.986196146429546e-05, "loss": 0.9536, "step": 33562 }, { "epoch": 0.03, "learning_rate": 4.986195322130835e-05, "loss": 0.7199, "step": 33563 }, { "epoch": 0.03, "learning_rate": 4.9861944978075813e-05, "loss": 0.9798, "step": 33564 }, { "epoch": 0.03, "learning_rate": 4.9861936734597844e-05, "loss": 1.1139, "step": 33565 }, { "epoch": 0.03, "learning_rate": 4.9861928490874446e-05, "loss": 1.0247, "step": 33566 }, { "epoch": 0.03, "learning_rate": 4.986192024690563e-05, "loss": 0.9689, "step": 33567 }, { "epoch": 0.03, "learning_rate": 4.986191200269138e-05, "loss": 0.8621, "step": 33568 }, { "epoch": 0.03, "learning_rate": 4.9861903758231706e-05, "loss": 0.9795, "step": 33569 }, { "epoch": 0.03, "learning_rate": 4.986189551352661e-05, "loss": 1.3599, "step": 33570 }, { "epoch": 0.03, "learning_rate": 4.9861887268576084e-05, "loss": 1.2815, "step": 33571 }, { "epoch": 0.03, "learning_rate": 4.986187902338013e-05, "loss": 1.0573, "step": 33572 }, { "epoch": 0.03, "learning_rate": 4.986187077793876e-05, "loss": 1.1127, "step": 33573 }, { "epoch": 0.03, "learning_rate": 4.986186253225195e-05, "loss": 0.8499, "step": 33574 }, { "epoch": 0.03, "learning_rate": 4.986185428631972e-05, "loss": 0.946, "step": 33575 }, { "epoch": 0.03, "learning_rate": 4.986184604014207e-05, "loss": 0.6648, "step": 33576 }, { "epoch": 0.03, "learning_rate": 4.986183779371899e-05, "loss": 0.6509, "step": 33577 }, { "epoch": 0.03, "learning_rate": 4.986182954705048e-05, "loss": 0.8923, "step": 33578 }, { "epoch": 0.03, "learning_rate": 4.9861821300136545e-05, "loss": 0.8871, "step": 33579 }, { "epoch": 0.03, "learning_rate": 4.986181305297719e-05, "loss": 1.0712, "step": 33580 }, { "epoch": 0.03, "learning_rate": 4.986180480557241e-05, "loss": 1.0254, "step": 33581 }, { "epoch": 0.03, "learning_rate": 4.9861796557922205e-05, "loss": 1.2745, "step": 33582 }, { "epoch": 0.03, "learning_rate": 4.986178831002657e-05, "loss": 1.1973, "step": 33583 }, { "epoch": 0.03, "learning_rate": 4.986178006188551e-05, "loss": 1.1043, "step": 33584 }, { "epoch": 0.03, "learning_rate": 4.9861771813499026e-05, "loss": 0.8993, "step": 33585 }, { "epoch": 0.03, "learning_rate": 4.986176356486712e-05, "loss": 0.7855, "step": 33586 }, { "epoch": 0.03, "learning_rate": 4.986175531598979e-05, "loss": 0.9223, "step": 33587 }, { "epoch": 0.03, "learning_rate": 4.986174706686702e-05, "loss": 0.8804, "step": 33588 }, { "epoch": 0.03, "learning_rate": 4.986173881749885e-05, "loss": 0.9589, "step": 33589 }, { "epoch": 0.03, "learning_rate": 4.986173056788524e-05, "loss": 1.0757, "step": 33590 }, { "epoch": 0.03, "learning_rate": 4.98617223180262e-05, "loss": 1.0918, "step": 33591 }, { "epoch": 0.03, "learning_rate": 4.986171406792175e-05, "loss": 1.1094, "step": 33592 }, { "epoch": 0.03, "learning_rate": 4.986170581757187e-05, "loss": 0.906, "step": 33593 }, { "epoch": 0.03, "learning_rate": 4.986169756697657e-05, "loss": 0.907, "step": 33594 }, { "epoch": 0.03, "learning_rate": 4.986168931613584e-05, "loss": 1.0866, "step": 33595 }, { "epoch": 0.03, "learning_rate": 4.986168106504968e-05, "loss": 0.9455, "step": 33596 }, { "epoch": 0.03, "learning_rate": 4.9861672813718105e-05, "loss": 1.0826, "step": 33597 }, { "epoch": 0.03, "learning_rate": 4.9861664562141105e-05, "loss": 0.7657, "step": 33598 }, { "epoch": 0.03, "learning_rate": 4.986165631031868e-05, "loss": 0.7774, "step": 33599 }, { "epoch": 0.03, "learning_rate": 4.986164805825083e-05, "loss": 0.777, "step": 33600 }, { "epoch": 0.03, "learning_rate": 4.986163980593756e-05, "loss": 1.0227, "step": 33601 }, { "epoch": 0.03, "learning_rate": 4.9861631553378864e-05, "loss": 1.199, "step": 33602 }, { "epoch": 0.03, "learning_rate": 4.9861623300574744e-05, "loss": 0.964, "step": 33603 }, { "epoch": 0.03, "learning_rate": 4.98616150475252e-05, "loss": 1.1142, "step": 33604 }, { "epoch": 0.03, "learning_rate": 4.9861606794230235e-05, "loss": 0.9025, "step": 33605 }, { "epoch": 0.03, "learning_rate": 4.9861598540689846e-05, "loss": 0.9811, "step": 33606 }, { "epoch": 0.03, "learning_rate": 4.9861590286904035e-05, "loss": 0.9459, "step": 33607 }, { "epoch": 0.03, "learning_rate": 4.98615820328728e-05, "loss": 0.9288, "step": 33608 }, { "epoch": 0.03, "learning_rate": 4.986157377859614e-05, "loss": 0.514, "step": 33609 }, { "epoch": 0.03, "learning_rate": 4.986156552407406e-05, "loss": 0.3054, "step": 33610 }, { "epoch": 0.03, "learning_rate": 4.986155726930655e-05, "loss": 0.3666, "step": 33611 }, { "epoch": 0.03, "learning_rate": 4.9861549014293624e-05, "loss": 0.397, "step": 33612 }, { "epoch": 0.03, "learning_rate": 4.986154075903528e-05, "loss": 0.3303, "step": 33613 }, { "epoch": 0.03, "learning_rate": 4.98615325035315e-05, "loss": 0.3109, "step": 33614 }, { "epoch": 0.03, "learning_rate": 4.986152424778231e-05, "loss": 0.3327, "step": 33615 }, { "epoch": 0.03, "learning_rate": 4.986151599178769e-05, "loss": 0.8334, "step": 33616 }, { "epoch": 0.03, "learning_rate": 4.9861507735547646e-05, "loss": 0.6946, "step": 33617 }, { "epoch": 0.03, "learning_rate": 4.986149947906218e-05, "loss": 0.9979, "step": 33618 }, { "epoch": 0.03, "learning_rate": 4.9861491222331294e-05, "loss": 0.794, "step": 33619 }, { "epoch": 0.03, "learning_rate": 4.986148296535499e-05, "loss": 0.8992, "step": 33620 }, { "epoch": 0.03, "learning_rate": 4.986147470813326e-05, "loss": 2.0438, "step": 33621 }, { "epoch": 0.03, "learning_rate": 4.9861466450666106e-05, "loss": 0.7005, "step": 33622 }, { "epoch": 0.03, "learning_rate": 4.986145819295354e-05, "loss": 0.9095, "step": 33623 }, { "epoch": 0.03, "learning_rate": 4.986144993499554e-05, "loss": 1.2311, "step": 33624 }, { "epoch": 0.03, "learning_rate": 4.986144167679212e-05, "loss": 1.2647, "step": 33625 }, { "epoch": 0.03, "learning_rate": 4.9861433418343285e-05, "loss": 0.9682, "step": 33626 }, { "epoch": 0.03, "learning_rate": 4.986142515964902e-05, "loss": 0.7254, "step": 33627 }, { "epoch": 0.03, "learning_rate": 4.9861416900709344e-05, "loss": 1.2507, "step": 33628 }, { "epoch": 0.03, "learning_rate": 4.986140864152424e-05, "loss": 1.2191, "step": 33629 }, { "epoch": 0.03, "learning_rate": 4.986140038209371e-05, "loss": 1.0886, "step": 33630 }, { "epoch": 0.03, "learning_rate": 4.986139212241777e-05, "loss": 1.0491, "step": 33631 }, { "epoch": 0.03, "learning_rate": 4.9861383862496405e-05, "loss": 0.8773, "step": 33632 }, { "epoch": 0.03, "learning_rate": 4.9861375602329616e-05, "loss": 0.368, "step": 33633 }, { "epoch": 0.03, "learning_rate": 4.986136734191741e-05, "loss": 0.3699, "step": 33634 }, { "epoch": 0.03, "learning_rate": 4.986135908125977e-05, "loss": 0.224, "step": 33635 }, { "epoch": 0.03, "learning_rate": 4.986135082035672e-05, "loss": 0.3676, "step": 33636 }, { "epoch": 0.03, "learning_rate": 4.986134255920825e-05, "loss": 0.1872, "step": 33637 }, { "epoch": 0.03, "learning_rate": 4.986133429781436e-05, "loss": 0.1621, "step": 33638 }, { "epoch": 0.03, "learning_rate": 4.986132603617505e-05, "loss": 0.1353, "step": 33639 }, { "epoch": 0.03, "learning_rate": 4.9861317774290316e-05, "loss": 0.1106, "step": 33640 }, { "epoch": 0.03, "learning_rate": 4.986130951216016e-05, "loss": 0.1834, "step": 33641 }, { "epoch": 0.03, "learning_rate": 4.986130124978459e-05, "loss": 0.1416, "step": 33642 }, { "epoch": 0.03, "learning_rate": 4.9861292987163586e-05, "loss": 0.121, "step": 33643 }, { "epoch": 0.03, "learning_rate": 4.9861284724297175e-05, "loss": 0.0829, "step": 33644 }, { "epoch": 0.03, "learning_rate": 4.986127646118534e-05, "loss": 0.0879, "step": 33645 }, { "epoch": 0.03, "learning_rate": 4.9861268197828084e-05, "loss": 0.0852, "step": 33646 }, { "epoch": 0.03, "learning_rate": 4.986125993422541e-05, "loss": 0.0658, "step": 33647 }, { "epoch": 0.03, "learning_rate": 4.986125167037732e-05, "loss": 0.2018, "step": 33648 }, { "epoch": 0.03, "learning_rate": 4.98612434062838e-05, "loss": 0.4377, "step": 33649 }, { "epoch": 0.03, "learning_rate": 4.9861235141944865e-05, "loss": 0.4105, "step": 33650 }, { "epoch": 0.03, "learning_rate": 4.986122687736051e-05, "loss": 0.3639, "step": 33651 }, { "epoch": 0.03, "learning_rate": 4.9861218612530736e-05, "loss": 0.4641, "step": 33652 }, { "epoch": 0.03, "learning_rate": 4.986121034745555e-05, "loss": 0.6518, "step": 33653 }, { "epoch": 0.03, "learning_rate": 4.986120208213494e-05, "loss": 0.6357, "step": 33654 }, { "epoch": 0.03, "learning_rate": 4.98611938165689e-05, "loss": 0.8183, "step": 33655 }, { "epoch": 0.03, "learning_rate": 4.986118555075745e-05, "loss": 0.8038, "step": 33656 }, { "epoch": 0.03, "learning_rate": 4.986117728470058e-05, "loss": 0.9412, "step": 33657 }, { "epoch": 0.03, "learning_rate": 4.986116901839829e-05, "loss": 0.7992, "step": 33658 }, { "epoch": 0.03, "learning_rate": 4.986116075185058e-05, "loss": 0.9032, "step": 33659 }, { "epoch": 0.03, "learning_rate": 4.9861152485057454e-05, "loss": 1.1367, "step": 33660 }, { "epoch": 0.03, "learning_rate": 4.9861144218018904e-05, "loss": 0.8386, "step": 33661 }, { "epoch": 0.03, "learning_rate": 4.9861135950734945e-05, "loss": 1.0348, "step": 33662 }, { "epoch": 0.03, "learning_rate": 4.9861127683205564e-05, "loss": 1.2024, "step": 33663 }, { "epoch": 0.03, "learning_rate": 4.986111941543076e-05, "loss": 1.1814, "step": 33664 }, { "epoch": 0.03, "learning_rate": 4.986111114741054e-05, "loss": 0.8639, "step": 33665 }, { "epoch": 0.03, "learning_rate": 4.98611028791449e-05, "loss": 0.952, "step": 33666 }, { "epoch": 0.03, "learning_rate": 4.986109461063384e-05, "loss": 1.3304, "step": 33667 }, { "epoch": 0.03, "learning_rate": 4.986108634187737e-05, "loss": 1.0978, "step": 33668 }, { "epoch": 0.03, "learning_rate": 4.986107807287548e-05, "loss": 1.1427, "step": 33669 }, { "epoch": 0.03, "learning_rate": 4.9861069803628165e-05, "loss": 0.9461, "step": 33670 }, { "epoch": 0.03, "learning_rate": 4.9861061534135435e-05, "loss": 0.904, "step": 33671 }, { "epoch": 0.03, "learning_rate": 4.986105326439728e-05, "loss": 0.7785, "step": 33672 }, { "epoch": 0.03, "learning_rate": 4.986104499441372e-05, "loss": 0.8659, "step": 33673 }, { "epoch": 0.03, "learning_rate": 4.986103672418474e-05, "loss": 1.2259, "step": 33674 }, { "epoch": 0.03, "learning_rate": 4.986102845371034e-05, "loss": 0.9233, "step": 33675 }, { "epoch": 0.03, "learning_rate": 4.9861020182990515e-05, "loss": 0.9296, "step": 33676 }, { "epoch": 0.03, "learning_rate": 4.9861011912025276e-05, "loss": 0.9835, "step": 33677 }, { "epoch": 0.03, "learning_rate": 4.986100364081463e-05, "loss": 1.0317, "step": 33678 }, { "epoch": 0.03, "learning_rate": 4.986099536935855e-05, "loss": 1.1543, "step": 33679 }, { "epoch": 0.03, "learning_rate": 4.986098709765707e-05, "loss": 1.0584, "step": 33680 }, { "epoch": 0.03, "learning_rate": 4.986097882571016e-05, "loss": 0.968, "step": 33681 }, { "epoch": 0.03, "learning_rate": 4.986097055351784e-05, "loss": 1.0692, "step": 33682 }, { "epoch": 0.03, "learning_rate": 4.98609622810801e-05, "loss": 1.1004, "step": 33683 }, { "epoch": 0.03, "learning_rate": 4.9860954008396946e-05, "loss": 0.9018, "step": 33684 }, { "epoch": 0.03, "learning_rate": 4.986094573546837e-05, "loss": 1.028, "step": 33685 }, { "epoch": 0.03, "learning_rate": 4.9860937462294384e-05, "loss": 0.709, "step": 33686 }, { "epoch": 0.03, "learning_rate": 4.986092918887497e-05, "loss": 1.0071, "step": 33687 }, { "epoch": 0.03, "learning_rate": 4.986092091521015e-05, "loss": 0.9479, "step": 33688 }, { "epoch": 0.03, "learning_rate": 4.986091264129991e-05, "loss": 1.0955, "step": 33689 }, { "epoch": 0.03, "learning_rate": 4.9860904367144256e-05, "loss": 1.2036, "step": 33690 }, { "epoch": 0.03, "learning_rate": 4.986089609274318e-05, "loss": 1.0052, "step": 33691 }, { "epoch": 0.03, "learning_rate": 4.986088781809669e-05, "loss": 1.0163, "step": 33692 }, { "epoch": 0.03, "learning_rate": 4.986087954320479e-05, "loss": 1.4665, "step": 33693 }, { "epoch": 0.03, "learning_rate": 4.9860871268067465e-05, "loss": 0.6604, "step": 33694 }, { "epoch": 0.03, "learning_rate": 4.986086299268473e-05, "loss": 0.4471, "step": 33695 }, { "epoch": 0.03, "learning_rate": 4.986085471705658e-05, "loss": 0.5022, "step": 33696 }, { "epoch": 0.03, "learning_rate": 4.986084644118301e-05, "loss": 0.557, "step": 33697 }, { "epoch": 0.03, "learning_rate": 4.986083816506403e-05, "loss": 1.0043, "step": 33698 }, { "epoch": 0.03, "learning_rate": 4.9860829888699624e-05, "loss": 1.0392, "step": 33699 }, { "epoch": 0.03, "learning_rate": 4.9860821612089805e-05, "loss": 0.6509, "step": 33700 }, { "epoch": 0.03, "learning_rate": 4.986081333523457e-05, "loss": 1.024, "step": 33701 }, { "epoch": 0.03, "learning_rate": 4.9860805058133934e-05, "loss": 0.8192, "step": 33702 }, { "epoch": 0.03, "learning_rate": 4.986079678078787e-05, "loss": 1.1838, "step": 33703 }, { "epoch": 0.03, "learning_rate": 4.9860788503196384e-05, "loss": 1.1246, "step": 33704 }, { "epoch": 0.03, "learning_rate": 4.986078022535949e-05, "loss": 1.246, "step": 33705 }, { "epoch": 0.03, "learning_rate": 4.9860771947277185e-05, "loss": 1.2903, "step": 33706 }, { "epoch": 0.03, "learning_rate": 4.986076366894946e-05, "loss": 0.6125, "step": 33707 }, { "epoch": 0.03, "learning_rate": 4.986075539037632e-05, "loss": 1.0932, "step": 33708 }, { "epoch": 0.03, "learning_rate": 4.9860747111557773e-05, "loss": 0.9859, "step": 33709 }, { "epoch": 0.03, "learning_rate": 4.98607388324938e-05, "loss": 1.0451, "step": 33710 }, { "epoch": 0.03, "learning_rate": 4.986073055318442e-05, "loss": 0.9007, "step": 33711 }, { "epoch": 0.03, "learning_rate": 4.9860722273629625e-05, "loss": 0.9443, "step": 33712 }, { "epoch": 0.03, "learning_rate": 4.986071399382941e-05, "loss": 1.0697, "step": 33713 }, { "epoch": 0.03, "learning_rate": 4.9860705713783784e-05, "loss": 1.0372, "step": 33714 }, { "epoch": 0.03, "learning_rate": 4.986069743349274e-05, "loss": 1.0855, "step": 33715 }, { "epoch": 0.03, "learning_rate": 4.9860689152956286e-05, "loss": 0.661, "step": 33716 }, { "epoch": 0.03, "learning_rate": 4.986068087217442e-05, "loss": 1.1058, "step": 33717 }, { "epoch": 0.03, "learning_rate": 4.986067259114713e-05, "loss": 0.8275, "step": 33718 }, { "epoch": 0.03, "learning_rate": 4.986066430987444e-05, "loss": 0.1493, "step": 33719 }, { "epoch": 0.03, "learning_rate": 4.986065602835632e-05, "loss": 0.0587, "step": 33720 }, { "epoch": 0.03, "learning_rate": 4.98606477465928e-05, "loss": 0.1474, "step": 33721 }, { "epoch": 0.03, "learning_rate": 4.986063946458386e-05, "loss": 0.992, "step": 33722 }, { "epoch": 0.03, "learning_rate": 4.986063118232951e-05, "loss": 1.3843, "step": 33723 }, { "epoch": 0.03, "learning_rate": 4.986062289982973e-05, "loss": 1.6819, "step": 33724 }, { "epoch": 0.03, "learning_rate": 4.9860614617084556e-05, "loss": 1.3809, "step": 33725 }, { "epoch": 0.03, "learning_rate": 4.986060633409396e-05, "loss": 1.0346, "step": 33726 }, { "epoch": 0.03, "learning_rate": 4.9860598050857955e-05, "loss": 0.8585, "step": 33727 }, { "epoch": 0.03, "learning_rate": 4.986058976737653e-05, "loss": 0.9606, "step": 33728 }, { "epoch": 0.03, "learning_rate": 4.98605814836497e-05, "loss": 0.5911, "step": 33729 }, { "epoch": 0.03, "learning_rate": 4.9860573199677454e-05, "loss": 0.9488, "step": 33730 }, { "epoch": 0.03, "learning_rate": 4.9860564915459796e-05, "loss": 0.7271, "step": 33731 }, { "epoch": 0.03, "learning_rate": 4.986055663099672e-05, "loss": 0.5435, "step": 33732 }, { "epoch": 0.03, "learning_rate": 4.986054834628823e-05, "loss": 0.7804, "step": 33733 }, { "epoch": 0.03, "learning_rate": 4.986054006133433e-05, "loss": 0.4701, "step": 33734 }, { "epoch": 0.03, "learning_rate": 4.986053177613502e-05, "loss": 1.1293, "step": 33735 }, { "epoch": 0.03, "learning_rate": 4.986052349069029e-05, "loss": 1.0277, "step": 33736 }, { "epoch": 0.03, "learning_rate": 4.9860515205000156e-05, "loss": 0.9461, "step": 33737 }, { "epoch": 0.03, "learning_rate": 4.9860506919064607e-05, "loss": 1.0253, "step": 33738 }, { "epoch": 0.03, "learning_rate": 4.986049863288365e-05, "loss": 0.9401, "step": 33739 }, { "epoch": 0.03, "learning_rate": 4.9860490346457266e-05, "loss": 1.2282, "step": 33740 }, { "epoch": 0.03, "learning_rate": 4.986048205978548e-05, "loss": 1.3111, "step": 33741 }, { "epoch": 0.03, "learning_rate": 4.986047377286829e-05, "loss": 0.8451, "step": 33742 }, { "epoch": 0.03, "learning_rate": 4.986046548570568e-05, "loss": 0.959, "step": 33743 }, { "epoch": 0.03, "learning_rate": 4.986045719829765e-05, "loss": 0.8281, "step": 33744 }, { "epoch": 0.03, "learning_rate": 4.986044891064421e-05, "loss": 0.7696, "step": 33745 }, { "epoch": 0.03, "learning_rate": 4.986044062274536e-05, "loss": 0.7486, "step": 33746 }, { "epoch": 0.03, "learning_rate": 4.98604323346011e-05, "loss": 0.5401, "step": 33747 }, { "epoch": 0.03, "learning_rate": 4.986042404621144e-05, "loss": 0.9663, "step": 33748 }, { "epoch": 0.03, "learning_rate": 4.986041575757635e-05, "loss": 0.987, "step": 33749 }, { "epoch": 0.03, "learning_rate": 4.9860407468695856e-05, "loss": 0.9324, "step": 33750 }, { "epoch": 0.03, "learning_rate": 4.986039917956995e-05, "loss": 0.9915, "step": 33751 }, { "epoch": 0.03, "learning_rate": 4.986039089019864e-05, "loss": 1.1588, "step": 33752 }, { "epoch": 0.03, "learning_rate": 4.986038260058191e-05, "loss": 0.8544, "step": 33753 }, { "epoch": 0.03, "learning_rate": 4.986037431071977e-05, "loss": 0.9916, "step": 33754 }, { "epoch": 0.03, "learning_rate": 4.986036602061222e-05, "loss": 0.784, "step": 33755 }, { "epoch": 0.03, "learning_rate": 4.986035773025926e-05, "loss": 0.4788, "step": 33756 }, { "epoch": 0.03, "learning_rate": 4.986034943966089e-05, "loss": 0.4185, "step": 33757 }, { "epoch": 0.03, "learning_rate": 4.98603411488171e-05, "loss": 0.5968, "step": 33758 }, { "epoch": 0.03, "learning_rate": 4.986033285772791e-05, "loss": 0.7569, "step": 33759 }, { "epoch": 0.03, "learning_rate": 4.98603245663933e-05, "loss": 0.8722, "step": 33760 }, { "epoch": 0.03, "learning_rate": 4.986031627481329e-05, "loss": 0.91, "step": 33761 }, { "epoch": 0.03, "learning_rate": 4.986030798298786e-05, "loss": 1.1148, "step": 33762 }, { "epoch": 0.03, "learning_rate": 4.986029969091702e-05, "loss": 1.3528, "step": 33763 }, { "epoch": 0.03, "learning_rate": 4.986029139860078e-05, "loss": 1.0234, "step": 33764 }, { "epoch": 0.03, "learning_rate": 4.986028310603912e-05, "loss": 0.3014, "step": 33765 }, { "epoch": 0.03, "learning_rate": 4.986027481323205e-05, "loss": 0.0812, "step": 33766 }, { "epoch": 0.03, "learning_rate": 4.986026652017958e-05, "loss": 0.1336, "step": 33767 }, { "epoch": 0.03, "learning_rate": 4.986025822688169e-05, "loss": 0.128, "step": 33768 }, { "epoch": 0.03, "learning_rate": 4.986024993333839e-05, "loss": 0.0835, "step": 33769 }, { "epoch": 0.03, "learning_rate": 4.986024163954968e-05, "loss": 0.0627, "step": 33770 }, { "epoch": 0.03, "learning_rate": 4.9860233345515565e-05, "loss": 0.0518, "step": 33771 }, { "epoch": 0.03, "learning_rate": 4.986022505123604e-05, "loss": 1.0668, "step": 33772 }, { "epoch": 0.03, "learning_rate": 4.98602167567111e-05, "loss": 0.8585, "step": 33773 }, { "epoch": 0.03, "learning_rate": 4.986020846194076e-05, "loss": 0.9693, "step": 33774 }, { "epoch": 0.03, "learning_rate": 4.9860200166925e-05, "loss": 0.7862, "step": 33775 }, { "epoch": 0.03, "learning_rate": 4.986019187166383e-05, "loss": 0.9602, "step": 33776 }, { "epoch": 0.03, "learning_rate": 4.986018357615726e-05, "loss": 0.9935, "step": 33777 }, { "epoch": 0.03, "learning_rate": 4.9860175280405276e-05, "loss": 1.0174, "step": 33778 }, { "epoch": 0.03, "learning_rate": 4.986016698440788e-05, "loss": 1.1619, "step": 33779 }, { "epoch": 0.03, "learning_rate": 4.986015868816508e-05, "loss": 1.1411, "step": 33780 }, { "epoch": 0.03, "learning_rate": 4.986015039167687e-05, "loss": 1.172, "step": 33781 }, { "epoch": 0.03, "learning_rate": 4.986014209494325e-05, "loss": 0.9112, "step": 33782 }, { "epoch": 0.03, "learning_rate": 4.986013379796422e-05, "loss": 1.3467, "step": 33783 }, { "epoch": 0.03, "learning_rate": 4.986012550073978e-05, "loss": 1.183, "step": 33784 }, { "epoch": 0.03, "learning_rate": 4.986011720326994e-05, "loss": 1.4017, "step": 33785 }, { "epoch": 0.03, "learning_rate": 4.986010890555468e-05, "loss": 0.2349, "step": 33786 }, { "epoch": 0.03, "learning_rate": 4.986010060759402e-05, "loss": 0.9046, "step": 33787 }, { "epoch": 0.03, "learning_rate": 4.986009230938794e-05, "loss": 1.0847, "step": 33788 }, { "epoch": 0.03, "learning_rate": 4.986008401093646e-05, "loss": 1.1162, "step": 33789 }, { "epoch": 0.03, "learning_rate": 4.986007571223957e-05, "loss": 1.1838, "step": 33790 }, { "epoch": 0.03, "learning_rate": 4.9860067413297276e-05, "loss": 1.1451, "step": 33791 }, { "epoch": 0.03, "learning_rate": 4.986005911410957e-05, "loss": 1.0351, "step": 33792 }, { "epoch": 0.03, "learning_rate": 4.986005081467646e-05, "loss": 0.982, "step": 33793 }, { "epoch": 0.03, "learning_rate": 4.986004251499794e-05, "loss": 0.6071, "step": 33794 }, { "epoch": 0.03, "learning_rate": 4.986003421507401e-05, "loss": 1.2957, "step": 33795 }, { "epoch": 0.03, "learning_rate": 4.9860025914904676e-05, "loss": 1.2825, "step": 33796 }, { "epoch": 0.03, "learning_rate": 4.986001761448993e-05, "loss": 0.8359, "step": 33797 }, { "epoch": 0.03, "learning_rate": 4.986000931382978e-05, "loss": 1.0188, "step": 33798 }, { "epoch": 0.03, "learning_rate": 4.9860001012924215e-05, "loss": 0.7496, "step": 33799 }, { "epoch": 0.03, "learning_rate": 4.985999271177325e-05, "loss": 0.8832, "step": 33800 }, { "epoch": 0.03, "learning_rate": 4.9859984410376875e-05, "loss": 1.0659, "step": 33801 }, { "epoch": 0.03, "learning_rate": 4.98599761087351e-05, "loss": 1.2253, "step": 33802 }, { "epoch": 0.03, "learning_rate": 4.9859967806847905e-05, "loss": 1.702, "step": 33803 }, { "epoch": 0.03, "learning_rate": 4.985995950471531e-05, "loss": 1.4522, "step": 33804 }, { "epoch": 0.03, "learning_rate": 4.9859951202337306e-05, "loss": 1.1522, "step": 33805 }, { "epoch": 0.03, "learning_rate": 4.98599428997139e-05, "loss": 1.34, "step": 33806 }, { "epoch": 0.03, "learning_rate": 4.985993459684508e-05, "loss": 1.2799, "step": 33807 }, { "epoch": 0.03, "learning_rate": 4.985992629373085e-05, "loss": 0.8788, "step": 33808 }, { "epoch": 0.03, "learning_rate": 4.9859917990371224e-05, "loss": 0.7812, "step": 33809 }, { "epoch": 0.03, "learning_rate": 4.985990968676619e-05, "loss": 0.8997, "step": 33810 }, { "epoch": 0.03, "learning_rate": 4.985990138291574e-05, "loss": 0.9491, "step": 33811 }, { "epoch": 0.03, "learning_rate": 4.98598930788199e-05, "loss": 0.8777, "step": 33812 }, { "epoch": 0.03, "learning_rate": 4.9859884774478636e-05, "loss": 1.0163, "step": 33813 }, { "epoch": 0.03, "learning_rate": 4.9859876469891976e-05, "loss": 1.0322, "step": 33814 }, { "epoch": 0.03, "learning_rate": 4.985986816505991e-05, "loss": 0.8996, "step": 33815 }, { "epoch": 0.03, "learning_rate": 4.985985985998243e-05, "loss": 1.084, "step": 33816 }, { "epoch": 0.03, "learning_rate": 4.9859851554659555e-05, "loss": 1.078, "step": 33817 }, { "epoch": 0.03, "learning_rate": 4.9859843249091266e-05, "loss": 1.1194, "step": 33818 }, { "epoch": 0.03, "learning_rate": 4.9859834943277574e-05, "loss": 0.9613, "step": 33819 }, { "epoch": 0.03, "learning_rate": 4.985982663721847e-05, "loss": 0.9705, "step": 33820 }, { "epoch": 0.03, "learning_rate": 4.985981833091396e-05, "loss": 1.2384, "step": 33821 }, { "epoch": 0.03, "learning_rate": 4.9859810024364056e-05, "loss": 1.1025, "step": 33822 }, { "epoch": 0.03, "learning_rate": 4.985980171756874e-05, "loss": 1.1178, "step": 33823 }, { "epoch": 0.03, "learning_rate": 4.9859793410528024e-05, "loss": 1.2171, "step": 33824 }, { "epoch": 0.03, "learning_rate": 4.985978510324189e-05, "loss": 1.4061, "step": 33825 }, { "epoch": 0.03, "learning_rate": 4.985977679571036e-05, "loss": 1.3367, "step": 33826 }, { "epoch": 0.03, "learning_rate": 4.985976848793342e-05, "loss": 1.0902, "step": 33827 }, { "epoch": 0.03, "learning_rate": 4.985976017991108e-05, "loss": 1.1648, "step": 33828 }, { "epoch": 0.03, "learning_rate": 4.985975187164333e-05, "loss": 0.9861, "step": 33829 }, { "epoch": 0.03, "learning_rate": 4.985974356313018e-05, "loss": 1.1153, "step": 33830 }, { "epoch": 0.03, "learning_rate": 4.985973525437162e-05, "loss": 1.094, "step": 33831 }, { "epoch": 0.03, "learning_rate": 4.9859726945367656e-05, "loss": 0.9598, "step": 33832 }, { "epoch": 0.03, "learning_rate": 4.985971863611829e-05, "loss": 1.0594, "step": 33833 }, { "epoch": 0.03, "learning_rate": 4.985971032662352e-05, "loss": 0.936, "step": 33834 }, { "epoch": 0.03, "learning_rate": 4.985970201688335e-05, "loss": 0.7627, "step": 33835 }, { "epoch": 0.03, "learning_rate": 4.9859693706897764e-05, "loss": 0.884, "step": 33836 }, { "epoch": 0.03, "learning_rate": 4.9859685396666775e-05, "loss": 0.6189, "step": 33837 }, { "epoch": 0.03, "learning_rate": 4.985967708619039e-05, "loss": 0.7422, "step": 33838 }, { "epoch": 0.03, "learning_rate": 4.9859668775468596e-05, "loss": 0.4087, "step": 33839 }, { "epoch": 0.03, "learning_rate": 4.98596604645014e-05, "loss": 1.3426, "step": 33840 }, { "epoch": 0.03, "learning_rate": 4.9859652153288794e-05, "loss": 1.0961, "step": 33841 }, { "epoch": 0.03, "learning_rate": 4.985964384183079e-05, "loss": 0.9081, "step": 33842 }, { "epoch": 0.03, "learning_rate": 4.985963553012738e-05, "loss": 1.0201, "step": 33843 }, { "epoch": 0.03, "learning_rate": 4.985962721817857e-05, "loss": 1.1648, "step": 33844 }, { "epoch": 0.03, "learning_rate": 4.985961890598435e-05, "loss": 0.7337, "step": 33845 }, { "epoch": 0.03, "learning_rate": 4.9859610593544724e-05, "loss": 0.8561, "step": 33846 }, { "epoch": 0.03, "learning_rate": 4.9859602280859705e-05, "loss": 1.0508, "step": 33847 }, { "epoch": 0.03, "learning_rate": 4.9859593967929276e-05, "loss": 0.8947, "step": 33848 }, { "epoch": 0.03, "learning_rate": 4.9859585654753444e-05, "loss": 1.1932, "step": 33849 }, { "epoch": 0.03, "learning_rate": 4.985957734133221e-05, "loss": 0.8381, "step": 33850 }, { "epoch": 0.03, "learning_rate": 4.985956902766557e-05, "loss": 1.1564, "step": 33851 }, { "epoch": 0.03, "learning_rate": 4.985956071375353e-05, "loss": 1.4844, "step": 33852 }, { "epoch": 0.03, "learning_rate": 4.985955239959608e-05, "loss": 1.0528, "step": 33853 }, { "epoch": 0.03, "learning_rate": 4.985954408519323e-05, "loss": 1.0026, "step": 33854 }, { "epoch": 0.03, "learning_rate": 4.9859535770544986e-05, "loss": 1.2354, "step": 33855 }, { "epoch": 0.03, "learning_rate": 4.985952745565133e-05, "loss": 1.1073, "step": 33856 }, { "epoch": 0.03, "learning_rate": 4.985951914051228e-05, "loss": 0.8309, "step": 33857 }, { "epoch": 0.03, "learning_rate": 4.9859510825127815e-05, "loss": 0.7578, "step": 33858 }, { "epoch": 0.03, "learning_rate": 4.985950250949796e-05, "loss": 0.9832, "step": 33859 }, { "epoch": 0.03, "learning_rate": 4.985949419362269e-05, "loss": 0.721, "step": 33860 }, { "epoch": 0.03, "learning_rate": 4.985948587750202e-05, "loss": 0.7769, "step": 33861 }, { "epoch": 0.03, "learning_rate": 4.985947756113595e-05, "loss": 0.7705, "step": 33862 }, { "epoch": 0.03, "learning_rate": 4.985946924452448e-05, "loss": 0.8211, "step": 33863 }, { "epoch": 0.03, "learning_rate": 4.9859460927667614e-05, "loss": 0.7427, "step": 33864 }, { "epoch": 0.03, "learning_rate": 4.985945261056533e-05, "loss": 1.0492, "step": 33865 }, { "epoch": 0.03, "learning_rate": 4.985944429321765e-05, "loss": 0.5871, "step": 33866 }, { "epoch": 0.03, "learning_rate": 4.985943597562458e-05, "loss": 0.9724, "step": 33867 }, { "epoch": 0.03, "learning_rate": 4.9859427657786095e-05, "loss": 0.9327, "step": 33868 }, { "epoch": 0.03, "learning_rate": 4.9859419339702216e-05, "loss": 1.1056, "step": 33869 }, { "epoch": 0.03, "learning_rate": 4.985941102137293e-05, "loss": 0.9406, "step": 33870 }, { "epoch": 0.03, "learning_rate": 4.9859402702798244e-05, "loss": 0.9978, "step": 33871 }, { "epoch": 0.03, "learning_rate": 4.985939438397815e-05, "loss": 1.2994, "step": 33872 }, { "epoch": 0.03, "learning_rate": 4.985938606491266e-05, "loss": 0.7051, "step": 33873 }, { "epoch": 0.03, "learning_rate": 4.985937774560177e-05, "loss": 0.8009, "step": 33874 }, { "epoch": 0.03, "learning_rate": 4.985936942604548e-05, "loss": 0.9114, "step": 33875 }, { "epoch": 0.03, "learning_rate": 4.985936110624378e-05, "loss": 1.0395, "step": 33876 }, { "epoch": 0.03, "learning_rate": 4.985935278619669e-05, "loss": 1.5542, "step": 33877 }, { "epoch": 0.03, "learning_rate": 4.98593444659042e-05, "loss": 1.1879, "step": 33878 }, { "epoch": 0.03, "learning_rate": 4.98593361453663e-05, "loss": 1.033, "step": 33879 }, { "epoch": 0.03, "learning_rate": 4.9859327824583e-05, "loss": 1.0282, "step": 33880 }, { "epoch": 0.03, "learning_rate": 4.98593195035543e-05, "loss": 1.1073, "step": 33881 }, { "epoch": 0.03, "learning_rate": 4.98593111822802e-05, "loss": 1.0908, "step": 33882 }, { "epoch": 0.03, "learning_rate": 4.9859302860760706e-05, "loss": 1.0893, "step": 33883 }, { "epoch": 0.03, "learning_rate": 4.9859294538995805e-05, "loss": 1.0371, "step": 33884 }, { "epoch": 0.03, "learning_rate": 4.98592862169855e-05, "loss": 0.9711, "step": 33885 }, { "epoch": 0.03, "learning_rate": 4.98592778947298e-05, "loss": 0.9939, "step": 33886 }, { "epoch": 0.03, "learning_rate": 4.9859269572228694e-05, "loss": 0.7634, "step": 33887 }, { "epoch": 0.03, "learning_rate": 4.9859261249482196e-05, "loss": 0.8912, "step": 33888 }, { "epoch": 0.03, "learning_rate": 4.98592529264903e-05, "loss": 0.9574, "step": 33889 }, { "epoch": 0.03, "learning_rate": 4.9859244603252995e-05, "loss": 0.77, "step": 33890 }, { "epoch": 0.03, "learning_rate": 4.98592362797703e-05, "loss": 0.954, "step": 33891 }, { "epoch": 0.03, "learning_rate": 4.985922795604219e-05, "loss": 1.0971, "step": 33892 }, { "epoch": 0.03, "learning_rate": 4.985921963206869e-05, "loss": 0.9333, "step": 33893 }, { "epoch": 0.03, "learning_rate": 4.985921130784978e-05, "loss": 0.8151, "step": 33894 }, { "epoch": 0.03, "learning_rate": 4.985920298338548e-05, "loss": 0.9034, "step": 33895 }, { "epoch": 0.03, "learning_rate": 4.985919465867579e-05, "loss": 1.2069, "step": 33896 }, { "epoch": 0.03, "learning_rate": 4.985918633372069e-05, "loss": 1.0328, "step": 33897 }, { "epoch": 0.03, "learning_rate": 4.985917800852019e-05, "loss": 1.1812, "step": 33898 }, { "epoch": 0.03, "learning_rate": 4.9859169683074284e-05, "loss": 1.1403, "step": 33899 }, { "epoch": 0.03, "learning_rate": 4.9859161357382986e-05, "loss": 0.9704, "step": 33900 }, { "epoch": 0.03, "learning_rate": 4.985915303144629e-05, "loss": 0.6214, "step": 33901 }, { "epoch": 0.03, "learning_rate": 4.9859144705264195e-05, "loss": 0.6818, "step": 33902 }, { "epoch": 0.03, "learning_rate": 4.98591363788367e-05, "loss": 1.3086, "step": 33903 }, { "epoch": 0.03, "learning_rate": 4.985912805216381e-05, "loss": 0.7616, "step": 33904 }, { "epoch": 0.03, "learning_rate": 4.9859119725245504e-05, "loss": 1.4608, "step": 33905 }, { "epoch": 0.03, "learning_rate": 4.985911139808182e-05, "loss": 0.9855, "step": 33906 }, { "epoch": 0.03, "learning_rate": 4.985910307067273e-05, "loss": 0.7321, "step": 33907 }, { "epoch": 0.03, "learning_rate": 4.9859094743018245e-05, "loss": 0.9514, "step": 33908 }, { "epoch": 0.03, "learning_rate": 4.985908641511835e-05, "loss": 0.957, "step": 33909 }, { "epoch": 0.03, "learning_rate": 4.9859078086973063e-05, "loss": 0.9312, "step": 33910 }, { "epoch": 0.03, "learning_rate": 4.985906975858238e-05, "loss": 0.8321, "step": 33911 }, { "epoch": 0.03, "learning_rate": 4.98590614299463e-05, "loss": 0.7319, "step": 33912 }, { "epoch": 0.03, "learning_rate": 4.985905310106482e-05, "loss": 0.9653, "step": 33913 }, { "epoch": 0.03, "learning_rate": 4.985904477193793e-05, "loss": 1.1554, "step": 33914 }, { "epoch": 0.03, "learning_rate": 4.985903644256566e-05, "loss": 0.9241, "step": 33915 }, { "epoch": 0.03, "learning_rate": 4.985902811294799e-05, "loss": 1.1628, "step": 33916 }, { "epoch": 0.03, "learning_rate": 4.985901978308492e-05, "loss": 1.0895, "step": 33917 }, { "epoch": 0.03, "learning_rate": 4.9859011452976444e-05, "loss": 0.9313, "step": 33918 }, { "epoch": 0.03, "learning_rate": 4.9859003122622574e-05, "loss": 1.1098, "step": 33919 }, { "epoch": 0.03, "learning_rate": 4.985899479202331e-05, "loss": 0.9051, "step": 33920 }, { "epoch": 0.03, "learning_rate": 4.985898646117865e-05, "loss": 0.4782, "step": 33921 }, { "epoch": 0.03, "learning_rate": 4.9858978130088585e-05, "loss": 0.8414, "step": 33922 }, { "epoch": 0.03, "learning_rate": 4.9858969798753126e-05, "loss": 1.0064, "step": 33923 }, { "epoch": 0.03, "learning_rate": 4.985896146717227e-05, "loss": 0.7264, "step": 33924 }, { "epoch": 0.03, "learning_rate": 4.985895313534602e-05, "loss": 0.8513, "step": 33925 }, { "epoch": 0.03, "learning_rate": 4.985894480327436e-05, "loss": 0.9792, "step": 33926 }, { "epoch": 0.03, "learning_rate": 4.9858936470957315e-05, "loss": 0.9771, "step": 33927 }, { "epoch": 0.03, "learning_rate": 4.985892813839488e-05, "loss": 0.4654, "step": 33928 }, { "epoch": 0.03, "learning_rate": 4.985891980558703e-05, "loss": 0.6709, "step": 33929 }, { "epoch": 0.03, "learning_rate": 4.98589114725338e-05, "loss": 1.1448, "step": 33930 }, { "epoch": 0.03, "learning_rate": 4.985890313923516e-05, "loss": 1.0911, "step": 33931 }, { "epoch": 0.03, "learning_rate": 4.9858894805691134e-05, "loss": 0.8967, "step": 33932 }, { "epoch": 0.03, "learning_rate": 4.9858886471901706e-05, "loss": 1.1825, "step": 33933 }, { "epoch": 0.03, "learning_rate": 4.985887813786688e-05, "loss": 1.3135, "step": 33934 }, { "epoch": 0.03, "learning_rate": 4.985886980358666e-05, "loss": 0.9031, "step": 33935 }, { "epoch": 0.03, "learning_rate": 4.985886146906105e-05, "loss": 0.9792, "step": 33936 }, { "epoch": 0.03, "learning_rate": 4.9858853134290036e-05, "loss": 1.2157, "step": 33937 }, { "epoch": 0.03, "learning_rate": 4.985884479927362e-05, "loss": 1.0864, "step": 33938 }, { "epoch": 0.03, "learning_rate": 4.985883646401182e-05, "loss": 0.8543, "step": 33939 }, { "epoch": 0.03, "learning_rate": 4.985882812850462e-05, "loss": 1.0935, "step": 33940 }, { "epoch": 0.03, "learning_rate": 4.985881979275202e-05, "loss": 0.9041, "step": 33941 }, { "epoch": 0.03, "learning_rate": 4.985881145675403e-05, "loss": 0.8985, "step": 33942 }, { "epoch": 0.03, "learning_rate": 4.985880312051064e-05, "loss": 1.0251, "step": 33943 }, { "epoch": 0.03, "learning_rate": 4.9858794784021856e-05, "loss": 0.5386, "step": 33944 }, { "epoch": 0.03, "learning_rate": 4.985878644728768e-05, "loss": 1.267, "step": 33945 }, { "epoch": 0.03, "learning_rate": 4.98587781103081e-05, "loss": 1.1633, "step": 33946 }, { "epoch": 0.03, "learning_rate": 4.9858769773083135e-05, "loss": 1.142, "step": 33947 }, { "epoch": 0.03, "learning_rate": 4.9858761435612766e-05, "loss": 0.8097, "step": 33948 }, { "epoch": 0.03, "learning_rate": 4.985875309789701e-05, "loss": 0.661, "step": 33949 }, { "epoch": 0.03, "learning_rate": 4.9858744759935854e-05, "loss": 0.5319, "step": 33950 }, { "epoch": 0.03, "learning_rate": 4.9858736421729305e-05, "loss": 0.388, "step": 33951 }, { "epoch": 0.03, "learning_rate": 4.985872808327735e-05, "loss": 0.409, "step": 33952 }, { "epoch": 0.03, "learning_rate": 4.985871974458002e-05, "loss": 0.3321, "step": 33953 }, { "epoch": 0.03, "learning_rate": 4.985871140563728e-05, "loss": 0.2946, "step": 33954 }, { "epoch": 0.03, "learning_rate": 4.985870306644915e-05, "loss": 0.2411, "step": 33955 }, { "epoch": 0.03, "learning_rate": 4.9858694727015624e-05, "loss": 0.2682, "step": 33956 }, { "epoch": 0.03, "learning_rate": 4.985868638733671e-05, "loss": 0.2444, "step": 33957 }, { "epoch": 0.03, "learning_rate": 4.9858678047412395e-05, "loss": 0.1786, "step": 33958 }, { "epoch": 0.03, "learning_rate": 4.985866970724268e-05, "loss": 0.2111, "step": 33959 }, { "epoch": 0.03, "learning_rate": 4.985866136682758e-05, "loss": 0.9598, "step": 33960 }, { "epoch": 0.03, "learning_rate": 4.9858653026167086e-05, "loss": 1.1848, "step": 33961 }, { "epoch": 0.03, "learning_rate": 4.985864468526119e-05, "loss": 1.0258, "step": 33962 }, { "epoch": 0.03, "learning_rate": 4.985863634410991e-05, "loss": 1.3284, "step": 33963 }, { "epoch": 0.03, "learning_rate": 4.985862800271322e-05, "loss": 1.5762, "step": 33964 }, { "epoch": 0.03, "learning_rate": 4.985861966107116e-05, "loss": 0.866, "step": 33965 }, { "epoch": 0.03, "learning_rate": 4.985861131918369e-05, "loss": 1.4419, "step": 33966 }, { "epoch": 0.03, "learning_rate": 4.985860297705083e-05, "loss": 1.0207, "step": 33967 }, { "epoch": 0.03, "learning_rate": 4.985859463467257e-05, "loss": 0.9778, "step": 33968 }, { "epoch": 0.03, "learning_rate": 4.985858629204892e-05, "loss": 0.5675, "step": 33969 }, { "epoch": 0.03, "learning_rate": 4.9858577949179884e-05, "loss": 1.0873, "step": 33970 }, { "epoch": 0.03, "learning_rate": 4.9858569606065456e-05, "loss": 0.9447, "step": 33971 }, { "epoch": 0.03, "learning_rate": 4.9858561262705626e-05, "loss": 1.1241, "step": 33972 }, { "epoch": 0.03, "learning_rate": 4.985855291910041e-05, "loss": 1.2523, "step": 33973 }, { "epoch": 0.03, "learning_rate": 4.9858544575249786e-05, "loss": 1.0991, "step": 33974 }, { "epoch": 0.03, "learning_rate": 4.985853623115378e-05, "loss": 1.0947, "step": 33975 }, { "epoch": 0.03, "learning_rate": 4.985852788681238e-05, "loss": 1.1238, "step": 33976 }, { "epoch": 0.03, "learning_rate": 4.985851954222559e-05, "loss": 0.7725, "step": 33977 }, { "epoch": 0.03, "learning_rate": 4.98585111973934e-05, "loss": 0.9548, "step": 33978 }, { "epoch": 0.03, "learning_rate": 4.9858502852315827e-05, "loss": 0.8732, "step": 33979 }, { "epoch": 0.03, "learning_rate": 4.9858494506992845e-05, "loss": 1.0026, "step": 33980 }, { "epoch": 0.03, "learning_rate": 4.985848616142449e-05, "loss": 1.1166, "step": 33981 }, { "epoch": 0.03, "learning_rate": 4.985847781561073e-05, "loss": 1.0227, "step": 33982 }, { "epoch": 0.03, "learning_rate": 4.985846946955158e-05, "loss": 1.0714, "step": 33983 }, { "epoch": 0.03, "learning_rate": 4.985846112324704e-05, "loss": 0.9952, "step": 33984 }, { "epoch": 0.03, "learning_rate": 4.9858452776697106e-05, "loss": 0.9611, "step": 33985 }, { "epoch": 0.03, "learning_rate": 4.985844442990178e-05, "loss": 1.0078, "step": 33986 }, { "epoch": 0.03, "learning_rate": 4.985843608286106e-05, "loss": 0.9117, "step": 33987 }, { "epoch": 0.03, "learning_rate": 4.9858427735574956e-05, "loss": 1.0656, "step": 33988 }, { "epoch": 0.03, "learning_rate": 4.9858419388043454e-05, "loss": 0.5659, "step": 33989 }, { "epoch": 0.03, "learning_rate": 4.9858411040266564e-05, "loss": 0.643, "step": 33990 }, { "epoch": 0.03, "learning_rate": 4.985840269224427e-05, "loss": 0.6672, "step": 33991 }, { "epoch": 0.03, "learning_rate": 4.98583943439766e-05, "loss": 0.9634, "step": 33992 }, { "epoch": 0.03, "learning_rate": 4.985838599546353e-05, "loss": 0.949, "step": 33993 }, { "epoch": 0.03, "learning_rate": 4.985837764670507e-05, "loss": 0.9597, "step": 33994 }, { "epoch": 0.03, "learning_rate": 4.9858369297701214e-05, "loss": 0.8844, "step": 33995 }, { "epoch": 0.03, "learning_rate": 4.985836094845197e-05, "loss": 1.3336, "step": 33996 }, { "epoch": 0.03, "learning_rate": 4.985835259895734e-05, "loss": 1.492, "step": 33997 }, { "epoch": 0.03, "learning_rate": 4.985834424921732e-05, "loss": 1.0724, "step": 33998 }, { "epoch": 0.03, "learning_rate": 4.9858335899231894e-05, "loss": 0.9564, "step": 33999 }, { "epoch": 0.03, "learning_rate": 4.985832754900109e-05, "loss": 0.8507, "step": 34000 }, { "epoch": 0.03, "eval_loss": 1.0717129707336426, "eval_runtime": 126.1661, "eval_samples_per_second": 10.978, "eval_steps_per_second": 5.493, "step": 34000 }, { "epoch": 0.03, "learning_rate": 4.985831919852489e-05, "loss": 1.2421, "step": 34001 }, { "epoch": 0.03, "learning_rate": 4.9858310847803304e-05, "loss": 1.1952, "step": 34002 }, { "epoch": 0.03, "learning_rate": 4.985830249683632e-05, "loss": 0.5941, "step": 34003 }, { "epoch": 0.03, "learning_rate": 4.985829414562395e-05, "loss": 1.0543, "step": 34004 }, { "epoch": 0.03, "learning_rate": 4.985828579416619e-05, "loss": 1.0804, "step": 34005 }, { "epoch": 0.03, "learning_rate": 4.9858277442463036e-05, "loss": 1.046, "step": 34006 }, { "epoch": 0.03, "learning_rate": 4.9858269090514495e-05, "loss": 1.0272, "step": 34007 }, { "epoch": 0.03, "learning_rate": 4.985826073832056e-05, "loss": 1.2933, "step": 34008 }, { "epoch": 0.03, "learning_rate": 4.985825238588124e-05, "loss": 0.9753, "step": 34009 }, { "epoch": 0.03, "learning_rate": 4.985824403319652e-05, "loss": 1.1986, "step": 34010 }, { "epoch": 0.03, "learning_rate": 4.9858235680266416e-05, "loss": 1.0474, "step": 34011 }, { "epoch": 0.03, "learning_rate": 4.985822732709093e-05, "loss": 1.0874, "step": 34012 }, { "epoch": 0.03, "learning_rate": 4.9858218973670044e-05, "loss": 0.8016, "step": 34013 }, { "epoch": 0.03, "learning_rate": 4.9858210620003774e-05, "loss": 1.2613, "step": 34014 }, { "epoch": 0.03, "learning_rate": 4.985820226609211e-05, "loss": 0.925, "step": 34015 }, { "epoch": 0.03, "learning_rate": 4.9858193911935055e-05, "loss": 1.1572, "step": 34016 }, { "epoch": 0.03, "learning_rate": 4.985818555753261e-05, "loss": 1.245, "step": 34017 }, { "epoch": 0.03, "learning_rate": 4.985817720288478e-05, "loss": 1.166, "step": 34018 }, { "epoch": 0.03, "learning_rate": 4.985816884799156e-05, "loss": 0.9613, "step": 34019 }, { "epoch": 0.03, "learning_rate": 4.985816049285295e-05, "loss": 1.0077, "step": 34020 }, { "epoch": 0.03, "learning_rate": 4.9858152137468946e-05, "loss": 0.9783, "step": 34021 }, { "epoch": 0.03, "learning_rate": 4.985814378183956e-05, "loss": 1.0221, "step": 34022 }, { "epoch": 0.03, "learning_rate": 4.9858135425964784e-05, "loss": 1.2304, "step": 34023 }, { "epoch": 0.03, "learning_rate": 4.985812706984462e-05, "loss": 0.8828, "step": 34024 }, { "epoch": 0.03, "learning_rate": 4.9858118713479066e-05, "loss": 0.9252, "step": 34025 }, { "epoch": 0.03, "learning_rate": 4.985811035686812e-05, "loss": 1.2788, "step": 34026 }, { "epoch": 0.03, "learning_rate": 4.985810200001178e-05, "loss": 0.8731, "step": 34027 }, { "epoch": 0.03, "learning_rate": 4.985809364291006e-05, "loss": 1.0749, "step": 34028 }, { "epoch": 0.03, "learning_rate": 4.985808528556295e-05, "loss": 1.1176, "step": 34029 }, { "epoch": 0.03, "learning_rate": 4.9858076927970446e-05, "loss": 1.0732, "step": 34030 }, { "epoch": 0.03, "learning_rate": 4.985806857013256e-05, "loss": 0.8841, "step": 34031 }, { "epoch": 0.03, "learning_rate": 4.9858060212049284e-05, "loss": 1.068, "step": 34032 }, { "epoch": 0.03, "learning_rate": 4.985805185372062e-05, "loss": 0.9875, "step": 34033 }, { "epoch": 0.03, "learning_rate": 4.985804349514657e-05, "loss": 0.8731, "step": 34034 }, { "epoch": 0.03, "learning_rate": 4.9858035136327126e-05, "loss": 0.578, "step": 34035 }, { "epoch": 0.03, "learning_rate": 4.9858026777262295e-05, "loss": 0.7443, "step": 34036 }, { "epoch": 0.03, "learning_rate": 4.985801841795208e-05, "loss": 0.8692, "step": 34037 }, { "epoch": 0.03, "learning_rate": 4.9858010058396474e-05, "loss": 0.9967, "step": 34038 }, { "epoch": 0.03, "learning_rate": 4.985800169859548e-05, "loss": 1.1445, "step": 34039 }, { "epoch": 0.03, "learning_rate": 4.9857993338549105e-05, "loss": 1.0685, "step": 34040 }, { "epoch": 0.03, "learning_rate": 4.985798497825733e-05, "loss": 0.9293, "step": 34041 }, { "epoch": 0.03, "learning_rate": 4.985797661772017e-05, "loss": 0.8336, "step": 34042 }, { "epoch": 0.03, "learning_rate": 4.9857968256937634e-05, "loss": 0.6361, "step": 34043 }, { "epoch": 0.03, "learning_rate": 4.98579598959097e-05, "loss": 0.3616, "step": 34044 }, { "epoch": 0.03, "learning_rate": 4.9857951534636383e-05, "loss": 0.39, "step": 34045 }, { "epoch": 0.03, "learning_rate": 4.985794317311768e-05, "loss": 1.0298, "step": 34046 }, { "epoch": 0.03, "learning_rate": 4.9857934811353584e-05, "loss": 1.0045, "step": 34047 }, { "epoch": 0.03, "learning_rate": 4.985792644934411e-05, "loss": 0.9766, "step": 34048 }, { "epoch": 0.03, "learning_rate": 4.985791808708924e-05, "loss": 0.8153, "step": 34049 }, { "epoch": 0.03, "learning_rate": 4.985790972458899e-05, "loss": 0.7792, "step": 34050 }, { "epoch": 0.03, "learning_rate": 4.985790136184335e-05, "loss": 1.2586, "step": 34051 }, { "epoch": 0.03, "learning_rate": 4.985789299885232e-05, "loss": 0.9309, "step": 34052 }, { "epoch": 0.03, "learning_rate": 4.98578846356159e-05, "loss": 1.3097, "step": 34053 }, { "epoch": 0.03, "learning_rate": 4.9857876272134105e-05, "loss": 1.0393, "step": 34054 }, { "epoch": 0.03, "learning_rate": 4.985786790840692e-05, "loss": 1.2942, "step": 34055 }, { "epoch": 0.03, "learning_rate": 4.985785954443435e-05, "loss": 1.1817, "step": 34056 }, { "epoch": 0.03, "learning_rate": 4.9857851180216385e-05, "loss": 0.9757, "step": 34057 }, { "epoch": 0.03, "learning_rate": 4.985784281575304e-05, "loss": 1.0618, "step": 34058 }, { "epoch": 0.03, "learning_rate": 4.985783445104431e-05, "loss": 1.2726, "step": 34059 }, { "epoch": 0.03, "learning_rate": 4.985782608609019e-05, "loss": 1.2011, "step": 34060 }, { "epoch": 0.03, "learning_rate": 4.985781772089069e-05, "loss": 0.967, "step": 34061 }, { "epoch": 0.03, "learning_rate": 4.98578093554458e-05, "loss": 1.0221, "step": 34062 }, { "epoch": 0.03, "learning_rate": 4.985780098975553e-05, "loss": 1.0006, "step": 34063 }, { "epoch": 0.03, "learning_rate": 4.9857792623819864e-05, "loss": 0.8386, "step": 34064 }, { "epoch": 0.03, "learning_rate": 4.985778425763882e-05, "loss": 0.877, "step": 34065 }, { "epoch": 0.03, "learning_rate": 4.985777589121239e-05, "loss": 1.0558, "step": 34066 }, { "epoch": 0.03, "learning_rate": 4.985776752454057e-05, "loss": 1.2105, "step": 34067 }, { "epoch": 0.03, "learning_rate": 4.985775915762336e-05, "loss": 1.256, "step": 34068 }, { "epoch": 0.03, "learning_rate": 4.985775079046078e-05, "loss": 1.1192, "step": 34069 }, { "epoch": 0.03, "learning_rate": 4.98577424230528e-05, "loss": 0.9003, "step": 34070 }, { "epoch": 0.03, "learning_rate": 4.985773405539944e-05, "loss": 0.5294, "step": 34071 }, { "epoch": 0.03, "learning_rate": 4.98577256875007e-05, "loss": 0.4211, "step": 34072 }, { "epoch": 0.03, "learning_rate": 4.985771731935657e-05, "loss": 0.4752, "step": 34073 }, { "epoch": 0.03, "learning_rate": 4.985770895096705e-05, "loss": 1.1204, "step": 34074 }, { "epoch": 0.03, "learning_rate": 4.985770058233216e-05, "loss": 1.0566, "step": 34075 }, { "epoch": 0.03, "learning_rate": 4.9857692213451876e-05, "loss": 1.0663, "step": 34076 }, { "epoch": 0.03, "learning_rate": 4.985768384432621e-05, "loss": 1.2969, "step": 34077 }, { "epoch": 0.03, "learning_rate": 4.985767547495516e-05, "loss": 1.3024, "step": 34078 }, { "epoch": 0.03, "learning_rate": 4.985766710533871e-05, "loss": 1.0548, "step": 34079 }, { "epoch": 0.03, "learning_rate": 4.98576587354769e-05, "loss": 1.0848, "step": 34080 }, { "epoch": 0.03, "learning_rate": 4.985765036536969e-05, "loss": 1.159, "step": 34081 }, { "epoch": 0.03, "learning_rate": 4.98576419950171e-05, "loss": 1.3518, "step": 34082 }, { "epoch": 0.03, "learning_rate": 4.985763362441912e-05, "loss": 0.7376, "step": 34083 }, { "epoch": 0.03, "learning_rate": 4.985762525357577e-05, "loss": 0.6289, "step": 34084 }, { "epoch": 0.03, "learning_rate": 4.985761688248702e-05, "loss": 0.8696, "step": 34085 }, { "epoch": 0.03, "learning_rate": 4.98576085111529e-05, "loss": 1.3835, "step": 34086 }, { "epoch": 0.03, "learning_rate": 4.9857600139573386e-05, "loss": 0.7254, "step": 34087 }, { "epoch": 0.03, "learning_rate": 4.985759176774849e-05, "loss": 0.9754, "step": 34088 }, { "epoch": 0.03, "learning_rate": 4.9857583395678216e-05, "loss": 0.9852, "step": 34089 }, { "epoch": 0.03, "learning_rate": 4.985757502336256e-05, "loss": 1.0626, "step": 34090 }, { "epoch": 0.03, "learning_rate": 4.985756665080151e-05, "loss": 1.0363, "step": 34091 }, { "epoch": 0.03, "learning_rate": 4.985755827799508e-05, "loss": 1.0363, "step": 34092 }, { "epoch": 0.03, "learning_rate": 4.985754990494327e-05, "loss": 0.9619, "step": 34093 }, { "epoch": 0.03, "learning_rate": 4.985754153164607e-05, "loss": 0.9903, "step": 34094 }, { "epoch": 0.03, "learning_rate": 4.9857533158103496e-05, "loss": 1.1122, "step": 34095 }, { "epoch": 0.03, "learning_rate": 4.985752478431553e-05, "loss": 1.3097, "step": 34096 }, { "epoch": 0.03, "learning_rate": 4.985751641028219e-05, "loss": 1.2131, "step": 34097 }, { "epoch": 0.03, "learning_rate": 4.9857508036003464e-05, "loss": 1.0639, "step": 34098 }, { "epoch": 0.03, "learning_rate": 4.9857499661479354e-05, "loss": 1.123, "step": 34099 }, { "epoch": 0.03, "learning_rate": 4.985749128670986e-05, "loss": 1.2371, "step": 34100 }, { "epoch": 0.03, "learning_rate": 4.985748291169499e-05, "loss": 0.9314, "step": 34101 }, { "epoch": 0.03, "learning_rate": 4.985747453643473e-05, "loss": 1.0645, "step": 34102 }, { "epoch": 0.03, "learning_rate": 4.985746616092909e-05, "loss": 1.198, "step": 34103 }, { "epoch": 0.03, "learning_rate": 4.9857457785178065e-05, "loss": 1.2635, "step": 34104 }, { "epoch": 0.03, "learning_rate": 4.985744940918166e-05, "loss": 0.4297, "step": 34105 }, { "epoch": 0.03, "learning_rate": 4.985744103293987e-05, "loss": 0.3384, "step": 34106 }, { "epoch": 0.03, "learning_rate": 4.9857432656452705e-05, "loss": 0.766, "step": 34107 }, { "epoch": 0.03, "learning_rate": 4.9857424279720156e-05, "loss": 1.0877, "step": 34108 }, { "epoch": 0.03, "learning_rate": 4.985741590274222e-05, "loss": 1.1859, "step": 34109 }, { "epoch": 0.03, "learning_rate": 4.98574075255189e-05, "loss": 1.2097, "step": 34110 }, { "epoch": 0.03, "learning_rate": 4.9857399148050204e-05, "loss": 1.1592, "step": 34111 }, { "epoch": 0.03, "learning_rate": 4.985739077033613e-05, "loss": 1.091, "step": 34112 }, { "epoch": 0.03, "learning_rate": 4.985738239237666e-05, "loss": 1.1913, "step": 34113 }, { "epoch": 0.03, "learning_rate": 4.985737401417182e-05, "loss": 1.0178, "step": 34114 }, { "epoch": 0.03, "learning_rate": 4.98573656357216e-05, "loss": 1.0277, "step": 34115 }, { "epoch": 0.03, "learning_rate": 4.985735725702599e-05, "loss": 0.7696, "step": 34116 }, { "epoch": 0.03, "learning_rate": 4.985734887808501e-05, "loss": 0.8873, "step": 34117 }, { "epoch": 0.03, "learning_rate": 4.985734049889864e-05, "loss": 1.2984, "step": 34118 }, { "epoch": 0.03, "learning_rate": 4.9857332119466884e-05, "loss": 0.8285, "step": 34119 }, { "epoch": 0.03, "learning_rate": 4.985732373978976e-05, "loss": 0.9298, "step": 34120 }, { "epoch": 0.03, "learning_rate": 4.985731535986725e-05, "loss": 1.191, "step": 34121 }, { "epoch": 0.03, "learning_rate": 4.985730697969935e-05, "loss": 1.1721, "step": 34122 }, { "epoch": 0.03, "learning_rate": 4.9857298599286085e-05, "loss": 1.4129, "step": 34123 }, { "epoch": 0.03, "learning_rate": 4.985729021862743e-05, "loss": 1.312, "step": 34124 }, { "epoch": 0.03, "learning_rate": 4.9857281837723394e-05, "loss": 1.1403, "step": 34125 }, { "epoch": 0.03, "learning_rate": 4.9857273456573975e-05, "loss": 1.0886, "step": 34126 }, { "epoch": 0.03, "learning_rate": 4.985726507517918e-05, "loss": 0.8522, "step": 34127 }, { "epoch": 0.03, "learning_rate": 4.9857256693539e-05, "loss": 1.3873, "step": 34128 }, { "epoch": 0.03, "learning_rate": 4.9857248311653446e-05, "loss": 0.9957, "step": 34129 }, { "epoch": 0.03, "learning_rate": 4.985723992952251e-05, "loss": 0.7155, "step": 34130 }, { "epoch": 0.03, "learning_rate": 4.9857231547146197e-05, "loss": 1.1852, "step": 34131 }, { "epoch": 0.03, "learning_rate": 4.98572231645245e-05, "loss": 0.9499, "step": 34132 }, { "epoch": 0.03, "learning_rate": 4.985721478165742e-05, "loss": 1.1504, "step": 34133 }, { "epoch": 0.03, "learning_rate": 4.9857206398544964e-05, "loss": 1.0706, "step": 34134 }, { "epoch": 0.03, "learning_rate": 4.985719801518712e-05, "loss": 0.9297, "step": 34135 }, { "epoch": 0.03, "learning_rate": 4.9857189631583914e-05, "loss": 0.7788, "step": 34136 }, { "epoch": 0.03, "learning_rate": 4.9857181247735306e-05, "loss": 0.4967, "step": 34137 }, { "epoch": 0.03, "learning_rate": 4.985717286364133e-05, "loss": 0.4504, "step": 34138 }, { "epoch": 0.03, "learning_rate": 4.985716447930198e-05, "loss": 0.8724, "step": 34139 }, { "epoch": 0.03, "learning_rate": 4.9857156094717244e-05, "loss": 0.7781, "step": 34140 }, { "epoch": 0.03, "learning_rate": 4.985714770988713e-05, "loss": 1.0526, "step": 34141 }, { "epoch": 0.03, "learning_rate": 4.985713932481164e-05, "loss": 1.1431, "step": 34142 }, { "epoch": 0.03, "learning_rate": 4.985713093949076e-05, "loss": 1.0364, "step": 34143 }, { "epoch": 0.03, "learning_rate": 4.985712255392451e-05, "loss": 1.0444, "step": 34144 }, { "epoch": 0.03, "learning_rate": 4.985711416811288e-05, "loss": 1.2504, "step": 34145 }, { "epoch": 0.03, "learning_rate": 4.9857105782055865e-05, "loss": 1.3723, "step": 34146 }, { "epoch": 0.03, "learning_rate": 4.9857097395753484e-05, "loss": 0.9566, "step": 34147 }, { "epoch": 0.03, "learning_rate": 4.9857089009205714e-05, "loss": 1.0383, "step": 34148 }, { "epoch": 0.03, "learning_rate": 4.985708062241257e-05, "loss": 1.083, "step": 34149 }, { "epoch": 0.03, "learning_rate": 4.9857072235374035e-05, "loss": 1.1264, "step": 34150 }, { "epoch": 0.03, "learning_rate": 4.985706384809013e-05, "loss": 1.1129, "step": 34151 }, { "epoch": 0.03, "learning_rate": 4.9857055460560854e-05, "loss": 0.759, "step": 34152 }, { "epoch": 0.03, "learning_rate": 4.985704707278619e-05, "loss": 1.1998, "step": 34153 }, { "epoch": 0.03, "learning_rate": 4.985703868476615e-05, "loss": 1.1012, "step": 34154 }, { "epoch": 0.03, "learning_rate": 4.985703029650074e-05, "loss": 1.0915, "step": 34155 }, { "epoch": 0.03, "learning_rate": 4.985702190798994e-05, "loss": 0.8362, "step": 34156 }, { "epoch": 0.03, "learning_rate": 4.985701351923377e-05, "loss": 1.4004, "step": 34157 }, { "epoch": 0.03, "learning_rate": 4.985700513023222e-05, "loss": 1.291, "step": 34158 }, { "epoch": 0.03, "learning_rate": 4.9856996740985286e-05, "loss": 0.7945, "step": 34159 }, { "epoch": 0.03, "learning_rate": 4.9856988351492985e-05, "loss": 1.0423, "step": 34160 }, { "epoch": 0.03, "learning_rate": 4.9856979961755296e-05, "loss": 0.8977, "step": 34161 }, { "epoch": 0.03, "learning_rate": 4.985697157177224e-05, "loss": 0.8891, "step": 34162 }, { "epoch": 0.03, "learning_rate": 4.985696318154379e-05, "loss": 0.663, "step": 34163 }, { "epoch": 0.03, "learning_rate": 4.9856954791069976e-05, "loss": 1.3667, "step": 34164 }, { "epoch": 0.03, "learning_rate": 4.985694640035078e-05, "loss": 1.1049, "step": 34165 }, { "epoch": 0.03, "learning_rate": 4.9856938009386214e-05, "loss": 1.0675, "step": 34166 }, { "epoch": 0.03, "learning_rate": 4.985692961817626e-05, "loss": 0.8174, "step": 34167 }, { "epoch": 0.03, "learning_rate": 4.985692122672093e-05, "loss": 0.8236, "step": 34168 }, { "epoch": 0.03, "learning_rate": 4.985691283502023e-05, "loss": 1.1985, "step": 34169 }, { "epoch": 0.03, "learning_rate": 4.985690444307415e-05, "loss": 0.5289, "step": 34170 }, { "epoch": 0.03, "learning_rate": 4.9856896050882695e-05, "loss": 0.8096, "step": 34171 }, { "epoch": 0.03, "learning_rate": 4.9856887658445864e-05, "loss": 1.0001, "step": 34172 }, { "epoch": 0.03, "learning_rate": 4.9856879265763644e-05, "loss": 0.9507, "step": 34173 }, { "epoch": 0.03, "learning_rate": 4.9856870872836056e-05, "loss": 0.7616, "step": 34174 }, { "epoch": 0.03, "learning_rate": 4.98568624796631e-05, "loss": 0.3165, "step": 34175 }, { "epoch": 0.03, "learning_rate": 4.985685408624476e-05, "loss": 0.3665, "step": 34176 }, { "epoch": 0.03, "learning_rate": 4.9856845692581046e-05, "loss": 0.8369, "step": 34177 }, { "epoch": 0.03, "learning_rate": 4.985683729867194e-05, "loss": 0.9192, "step": 34178 }, { "epoch": 0.03, "learning_rate": 4.985682890451748e-05, "loss": 0.5573, "step": 34179 }, { "epoch": 0.03, "learning_rate": 4.985682051011763e-05, "loss": 0.7076, "step": 34180 }, { "epoch": 0.03, "learning_rate": 4.985681211547241e-05, "loss": 0.9464, "step": 34181 }, { "epoch": 0.03, "learning_rate": 4.985680372058181e-05, "loss": 0.8228, "step": 34182 }, { "epoch": 0.03, "learning_rate": 4.985679532544584e-05, "loss": 0.9316, "step": 34183 }, { "epoch": 0.03, "learning_rate": 4.985678693006449e-05, "loss": 1.3122, "step": 34184 }, { "epoch": 0.03, "learning_rate": 4.985677853443776e-05, "loss": 0.9066, "step": 34185 }, { "epoch": 0.03, "learning_rate": 4.9856770138565664e-05, "loss": 1.1784, "step": 34186 }, { "epoch": 0.03, "learning_rate": 4.985676174244819e-05, "loss": 1.004, "step": 34187 }, { "epoch": 0.03, "learning_rate": 4.985675334608534e-05, "loss": 0.9785, "step": 34188 }, { "epoch": 0.03, "learning_rate": 4.985674494947711e-05, "loss": 1.1564, "step": 34189 }, { "epoch": 0.03, "learning_rate": 4.9856736552623504e-05, "loss": 0.8512, "step": 34190 }, { "epoch": 0.03, "learning_rate": 4.9856728155524536e-05, "loss": 1.2155, "step": 34191 }, { "epoch": 0.03, "learning_rate": 4.985671975818018e-05, "loss": 1.1769, "step": 34192 }, { "epoch": 0.03, "learning_rate": 4.985671136059045e-05, "loss": 1.1814, "step": 34193 }, { "epoch": 0.03, "learning_rate": 4.9856702962755346e-05, "loss": 1.1414, "step": 34194 }, { "epoch": 0.03, "learning_rate": 4.9856694564674863e-05, "loss": 1.2505, "step": 34195 }, { "epoch": 0.03, "learning_rate": 4.985668616634902e-05, "loss": 1.1978, "step": 34196 }, { "epoch": 0.03, "learning_rate": 4.985667776777779e-05, "loss": 1.0414, "step": 34197 }, { "epoch": 0.03, "learning_rate": 4.985666936896119e-05, "loss": 0.6639, "step": 34198 }, { "epoch": 0.03, "learning_rate": 4.9856660969899214e-05, "loss": 0.4471, "step": 34199 }, { "epoch": 0.03, "learning_rate": 4.985665257059186e-05, "loss": 0.7835, "step": 34200 }, { "epoch": 0.03, "learning_rate": 4.9856644171039134e-05, "loss": 0.874, "step": 34201 }, { "epoch": 0.03, "learning_rate": 4.985663577124103e-05, "loss": 1.0678, "step": 34202 }, { "epoch": 0.03, "learning_rate": 4.985662737119756e-05, "loss": 1.1088, "step": 34203 }, { "epoch": 0.03, "learning_rate": 4.985661897090871e-05, "loss": 1.064, "step": 34204 }, { "epoch": 0.03, "learning_rate": 4.98566105703745e-05, "loss": 0.8118, "step": 34205 }, { "epoch": 0.03, "learning_rate": 4.9856602169594893e-05, "loss": 0.9566, "step": 34206 }, { "epoch": 0.03, "learning_rate": 4.985659376856993e-05, "loss": 1.026, "step": 34207 }, { "epoch": 0.03, "learning_rate": 4.985658536729958e-05, "loss": 0.5102, "step": 34208 }, { "epoch": 0.03, "learning_rate": 4.9856576965783864e-05, "loss": 1.0517, "step": 34209 }, { "epoch": 0.03, "learning_rate": 4.985656856402278e-05, "loss": 0.9208, "step": 34210 }, { "epoch": 0.03, "learning_rate": 4.9856560162016305e-05, "loss": 0.9962, "step": 34211 }, { "epoch": 0.03, "learning_rate": 4.985655175976447e-05, "loss": 1.1236, "step": 34212 }, { "epoch": 0.03, "learning_rate": 4.985654335726726e-05, "loss": 0.9571, "step": 34213 }, { "epoch": 0.03, "learning_rate": 4.985653495452467e-05, "loss": 1.1326, "step": 34214 }, { "epoch": 0.03, "learning_rate": 4.985652655153672e-05, "loss": 1.2767, "step": 34215 }, { "epoch": 0.03, "learning_rate": 4.985651814830339e-05, "loss": 1.0918, "step": 34216 }, { "epoch": 0.03, "learning_rate": 4.9856509744824684e-05, "loss": 0.8839, "step": 34217 }, { "epoch": 0.03, "learning_rate": 4.9856501341100604e-05, "loss": 0.9258, "step": 34218 }, { "epoch": 0.03, "learning_rate": 4.985649293713115e-05, "loss": 0.8017, "step": 34219 }, { "epoch": 0.03, "learning_rate": 4.9856484532916324e-05, "loss": 0.9739, "step": 34220 }, { "epoch": 0.03, "learning_rate": 4.985647612845613e-05, "loss": 0.9212, "step": 34221 }, { "epoch": 0.03, "learning_rate": 4.9856467723750564e-05, "loss": 1.295, "step": 34222 }, { "epoch": 0.03, "learning_rate": 4.985645931879963e-05, "loss": 1.0226, "step": 34223 }, { "epoch": 0.03, "learning_rate": 4.9856450913603316e-05, "loss": 0.9896, "step": 34224 }, { "epoch": 0.03, "learning_rate": 4.985644250816163e-05, "loss": 0.933, "step": 34225 }, { "epoch": 0.03, "learning_rate": 4.985643410247457e-05, "loss": 1.1689, "step": 34226 }, { "epoch": 0.03, "learning_rate": 4.9856425696542136e-05, "loss": 1.3671, "step": 34227 }, { "epoch": 0.03, "learning_rate": 4.985641729036433e-05, "loss": 1.4373, "step": 34228 }, { "epoch": 0.03, "learning_rate": 4.9856408883941155e-05, "loss": 1.5735, "step": 34229 }, { "epoch": 0.03, "learning_rate": 4.985640047727261e-05, "loss": 1.0086, "step": 34230 }, { "epoch": 0.03, "learning_rate": 4.9856392070358694e-05, "loss": 1.1646, "step": 34231 }, { "epoch": 0.03, "learning_rate": 4.98563836631994e-05, "loss": 0.8673, "step": 34232 }, { "epoch": 0.03, "learning_rate": 4.985637525579474e-05, "loss": 1.1534, "step": 34233 }, { "epoch": 0.03, "learning_rate": 4.985636684814471e-05, "loss": 1.0859, "step": 34234 }, { "epoch": 0.03, "learning_rate": 4.98563584402493e-05, "loss": 0.8189, "step": 34235 }, { "epoch": 0.03, "learning_rate": 4.985635003210852e-05, "loss": 1.0769, "step": 34236 }, { "epoch": 0.03, "learning_rate": 4.985634162372237e-05, "loss": 0.9307, "step": 34237 }, { "epoch": 0.03, "learning_rate": 4.985633321509085e-05, "loss": 0.6439, "step": 34238 }, { "epoch": 0.03, "learning_rate": 4.985632480621396e-05, "loss": 0.6983, "step": 34239 }, { "epoch": 0.03, "learning_rate": 4.98563163970917e-05, "loss": 0.6137, "step": 34240 }, { "epoch": 0.03, "learning_rate": 4.9856307987724074e-05, "loss": 1.0349, "step": 34241 }, { "epoch": 0.03, "learning_rate": 4.985629957811107e-05, "loss": 0.9781, "step": 34242 }, { "epoch": 0.03, "learning_rate": 4.985629116825269e-05, "loss": 1.1246, "step": 34243 }, { "epoch": 0.03, "learning_rate": 4.9856282758148945e-05, "loss": 0.8216, "step": 34244 }, { "epoch": 0.03, "learning_rate": 4.985627434779983e-05, "loss": 0.8708, "step": 34245 }, { "epoch": 0.03, "learning_rate": 4.985626593720534e-05, "loss": 1.1545, "step": 34246 }, { "epoch": 0.03, "learning_rate": 4.985625752636548e-05, "loss": 1.0479, "step": 34247 }, { "epoch": 0.03, "learning_rate": 4.985624911528025e-05, "loss": 0.9403, "step": 34248 }, { "epoch": 0.03, "learning_rate": 4.985624070394965e-05, "loss": 0.9316, "step": 34249 }, { "epoch": 0.03, "learning_rate": 4.985623229237368e-05, "loss": 0.9982, "step": 34250 }, { "epoch": 0.03, "learning_rate": 4.985622388055235e-05, "loss": 0.8623, "step": 34251 }, { "epoch": 0.03, "learning_rate": 4.9856215468485636e-05, "loss": 1.1968, "step": 34252 }, { "epoch": 0.03, "learning_rate": 4.9856207056173556e-05, "loss": 1.0425, "step": 34253 }, { "epoch": 0.03, "learning_rate": 4.98561986436161e-05, "loss": 0.911, "step": 34254 }, { "epoch": 0.03, "learning_rate": 4.9856190230813285e-05, "loss": 1.2585, "step": 34255 }, { "epoch": 0.03, "learning_rate": 4.98561818177651e-05, "loss": 0.9833, "step": 34256 }, { "epoch": 0.03, "learning_rate": 4.985617340447154e-05, "loss": 1.0908, "step": 34257 }, { "epoch": 0.03, "learning_rate": 4.985616499093261e-05, "loss": 0.4708, "step": 34258 }, { "epoch": 0.03, "learning_rate": 4.9856156577148306e-05, "loss": 1.5452, "step": 34259 }, { "epoch": 0.03, "learning_rate": 4.985614816311864e-05, "loss": 1.3441, "step": 34260 }, { "epoch": 0.03, "learning_rate": 4.98561397488436e-05, "loss": 1.1102, "step": 34261 }, { "epoch": 0.03, "learning_rate": 4.9856131334323196e-05, "loss": 1.3881, "step": 34262 }, { "epoch": 0.03, "learning_rate": 4.9856122919557425e-05, "loss": 1.0704, "step": 34263 }, { "epoch": 0.03, "learning_rate": 4.985611450454628e-05, "loss": 1.3388, "step": 34264 }, { "epoch": 0.03, "learning_rate": 4.985610608928976e-05, "loss": 1.1653, "step": 34265 }, { "epoch": 0.03, "learning_rate": 4.985609767378788e-05, "loss": 1.3242, "step": 34266 }, { "epoch": 0.03, "learning_rate": 4.985608925804063e-05, "loss": 1.4148, "step": 34267 }, { "epoch": 0.03, "learning_rate": 4.9856080842048e-05, "loss": 1.2191, "step": 34268 }, { "epoch": 0.03, "learning_rate": 4.985607242581002e-05, "loss": 1.1026, "step": 34269 }, { "epoch": 0.03, "learning_rate": 4.985606400932665e-05, "loss": 1.3693, "step": 34270 }, { "epoch": 0.03, "learning_rate": 4.9856055592597927e-05, "loss": 1.3435, "step": 34271 }, { "epoch": 0.03, "learning_rate": 4.985604717562383e-05, "loss": 1.2944, "step": 34272 }, { "epoch": 0.03, "learning_rate": 4.985603875840437e-05, "loss": 1.6247, "step": 34273 }, { "epoch": 0.03, "learning_rate": 4.9856030340939544e-05, "loss": 1.8307, "step": 34274 }, { "epoch": 0.03, "learning_rate": 4.985602192322934e-05, "loss": 1.1555, "step": 34275 }, { "epoch": 0.03, "learning_rate": 4.985601350527377e-05, "loss": 0.8411, "step": 34276 }, { "epoch": 0.03, "learning_rate": 4.985600508707283e-05, "loss": 0.9162, "step": 34277 }, { "epoch": 0.03, "learning_rate": 4.9855996668626525e-05, "loss": 0.9209, "step": 34278 }, { "epoch": 0.03, "learning_rate": 4.985598824993485e-05, "loss": 0.5336, "step": 34279 }, { "epoch": 0.03, "learning_rate": 4.9855979830997815e-05, "loss": 1.0009, "step": 34280 }, { "epoch": 0.03, "learning_rate": 4.9855971411815404e-05, "loss": 1.2116, "step": 34281 }, { "epoch": 0.03, "learning_rate": 4.985596299238763e-05, "loss": 0.9517, "step": 34282 }, { "epoch": 0.03, "learning_rate": 4.985595457271448e-05, "loss": 0.9045, "step": 34283 }, { "epoch": 0.03, "learning_rate": 4.985594615279597e-05, "loss": 0.6312, "step": 34284 }, { "epoch": 0.03, "learning_rate": 4.9855937732632094e-05, "loss": 0.8536, "step": 34285 }, { "epoch": 0.03, "learning_rate": 4.985592931222285e-05, "loss": 0.8172, "step": 34286 }, { "epoch": 0.03, "learning_rate": 4.985592089156824e-05, "loss": 0.4302, "step": 34287 }, { "epoch": 0.03, "learning_rate": 4.9855912470668254e-05, "loss": 0.1916, "step": 34288 }, { "epoch": 0.03, "learning_rate": 4.985590404952291e-05, "loss": 0.7564, "step": 34289 }, { "epoch": 0.03, "learning_rate": 4.9855895628132194e-05, "loss": 0.3074, "step": 34290 }, { "epoch": 0.03, "learning_rate": 4.985588720649611e-05, "loss": 0.8341, "step": 34291 }, { "epoch": 0.03, "learning_rate": 4.985587878461466e-05, "loss": 1.4017, "step": 34292 }, { "epoch": 0.03, "learning_rate": 4.985587036248785e-05, "loss": 0.6299, "step": 34293 }, { "epoch": 0.03, "learning_rate": 4.9855861940115666e-05, "loss": 0.3603, "step": 34294 }, { "epoch": 0.03, "learning_rate": 4.9855853517498116e-05, "loss": 0.3237, "step": 34295 }, { "epoch": 0.03, "learning_rate": 4.98558450946352e-05, "loss": 0.8422, "step": 34296 }, { "epoch": 0.03, "learning_rate": 4.9855836671526924e-05, "loss": 1.0106, "step": 34297 }, { "epoch": 0.03, "learning_rate": 4.9855828248173275e-05, "loss": 1.1143, "step": 34298 }, { "epoch": 0.03, "learning_rate": 4.985581982457426e-05, "loss": 1.0248, "step": 34299 }, { "epoch": 0.03, "learning_rate": 4.985581140072988e-05, "loss": 0.9387, "step": 34300 }, { "epoch": 0.03, "learning_rate": 4.985580297664013e-05, "loss": 0.728, "step": 34301 }, { "epoch": 0.03, "learning_rate": 4.985579455230502e-05, "loss": 0.7976, "step": 34302 }, { "epoch": 0.03, "learning_rate": 4.9855786127724545e-05, "loss": 1.0563, "step": 34303 }, { "epoch": 0.03, "learning_rate": 4.98557777028987e-05, "loss": 1.0928, "step": 34304 }, { "epoch": 0.03, "learning_rate": 4.985576927782749e-05, "loss": 1.0562, "step": 34305 }, { "epoch": 0.03, "learning_rate": 4.9855760852510915e-05, "loss": 1.0957, "step": 34306 }, { "epoch": 0.03, "learning_rate": 4.9855752426948976e-05, "loss": 0.6408, "step": 34307 }, { "epoch": 0.03, "learning_rate": 4.985574400114167e-05, "loss": 0.7327, "step": 34308 }, { "epoch": 0.03, "learning_rate": 4.9855735575088995e-05, "loss": 0.8643, "step": 34309 }, { "epoch": 0.03, "learning_rate": 4.985572714879095e-05, "loss": 1.6271, "step": 34310 }, { "epoch": 0.03, "learning_rate": 4.985571872224756e-05, "loss": 1.1626, "step": 34311 }, { "epoch": 0.03, "learning_rate": 4.985571029545879e-05, "loss": 0.7844, "step": 34312 }, { "epoch": 0.03, "learning_rate": 4.9855701868424656e-05, "loss": 0.9146, "step": 34313 }, { "epoch": 0.03, "learning_rate": 4.985569344114516e-05, "loss": 0.8226, "step": 34314 }, { "epoch": 0.03, "learning_rate": 4.985568501362029e-05, "loss": 1.0037, "step": 34315 }, { "epoch": 0.03, "learning_rate": 4.9855676585850066e-05, "loss": 0.835, "step": 34316 }, { "epoch": 0.03, "learning_rate": 4.9855668157834476e-05, "loss": 0.3416, "step": 34317 }, { "epoch": 0.03, "learning_rate": 4.985565972957352e-05, "loss": 0.7585, "step": 34318 }, { "epoch": 0.03, "learning_rate": 4.98556513010672e-05, "loss": 1.0998, "step": 34319 }, { "epoch": 0.03, "learning_rate": 4.985564287231551e-05, "loss": 1.0285, "step": 34320 }, { "epoch": 0.03, "learning_rate": 4.9855634443318454e-05, "loss": 0.9299, "step": 34321 }, { "epoch": 0.03, "learning_rate": 4.985562601407604e-05, "loss": 1.0964, "step": 34322 }, { "epoch": 0.03, "learning_rate": 4.985561758458826e-05, "loss": 0.8882, "step": 34323 }, { "epoch": 0.03, "learning_rate": 4.985560915485512e-05, "loss": 0.9216, "step": 34324 }, { "epoch": 0.03, "learning_rate": 4.985560072487662e-05, "loss": 1.3295, "step": 34325 }, { "epoch": 0.03, "learning_rate": 4.985559229465274e-05, "loss": 0.8855, "step": 34326 }, { "epoch": 0.03, "learning_rate": 4.9855583864183505e-05, "loss": 0.8746, "step": 34327 }, { "epoch": 0.03, "learning_rate": 4.98555754334689e-05, "loss": 1.1478, "step": 34328 }, { "epoch": 0.03, "learning_rate": 4.985556700250894e-05, "loss": 0.6992, "step": 34329 }, { "epoch": 0.03, "learning_rate": 4.985555857130362e-05, "loss": 0.9328, "step": 34330 }, { "epoch": 0.03, "learning_rate": 4.9855550139852926e-05, "loss": 0.8405, "step": 34331 }, { "epoch": 0.03, "learning_rate": 4.985554170815687e-05, "loss": 0.5517, "step": 34332 }, { "epoch": 0.03, "learning_rate": 4.985553327621545e-05, "loss": 1.3978, "step": 34333 }, { "epoch": 0.03, "learning_rate": 4.985552484402868e-05, "loss": 0.9248, "step": 34334 }, { "epoch": 0.03, "learning_rate": 4.985551641159653e-05, "loss": 0.9484, "step": 34335 }, { "epoch": 0.03, "learning_rate": 4.985550797891902e-05, "loss": 0.7576, "step": 34336 }, { "epoch": 0.03, "learning_rate": 4.985549954599615e-05, "loss": 1.1318, "step": 34337 }, { "epoch": 0.03, "learning_rate": 4.9855491112827925e-05, "loss": 0.7991, "step": 34338 }, { "epoch": 0.03, "learning_rate": 4.9855482679414324e-05, "loss": 1.0941, "step": 34339 }, { "epoch": 0.03, "learning_rate": 4.985547424575536e-05, "loss": 1.0493, "step": 34340 }, { "epoch": 0.03, "learning_rate": 4.985546581185104e-05, "loss": 0.9269, "step": 34341 }, { "epoch": 0.03, "learning_rate": 4.9855457377701356e-05, "loss": 1.0096, "step": 34342 }, { "epoch": 0.03, "learning_rate": 4.9855448943306314e-05, "loss": 1.0721, "step": 34343 }, { "epoch": 0.03, "learning_rate": 4.9855440508665904e-05, "loss": 0.8061, "step": 34344 }, { "epoch": 0.03, "learning_rate": 4.9855432073780125e-05, "loss": 0.813, "step": 34345 }, { "epoch": 0.03, "learning_rate": 4.9855423638649e-05, "loss": 0.804, "step": 34346 }, { "epoch": 0.03, "learning_rate": 4.9855415203272496e-05, "loss": 1.4011, "step": 34347 }, { "epoch": 0.03, "learning_rate": 4.985540676765064e-05, "loss": 1.5175, "step": 34348 }, { "epoch": 0.03, "learning_rate": 4.985539833178342e-05, "loss": 0.9888, "step": 34349 }, { "epoch": 0.03, "learning_rate": 4.985538989567083e-05, "loss": 1.0581, "step": 34350 }, { "epoch": 0.03, "learning_rate": 4.985538145931289e-05, "loss": 1.1492, "step": 34351 }, { "epoch": 0.03, "learning_rate": 4.985537302270959e-05, "loss": 1.1928, "step": 34352 }, { "epoch": 0.03, "learning_rate": 4.985536458586092e-05, "loss": 0.6623, "step": 34353 }, { "epoch": 0.03, "learning_rate": 4.985535614876689e-05, "loss": 1.0827, "step": 34354 }, { "epoch": 0.03, "learning_rate": 4.985534771142749e-05, "loss": 0.6672, "step": 34355 }, { "epoch": 0.03, "learning_rate": 4.985533927384274e-05, "loss": 1.0316, "step": 34356 }, { "epoch": 0.03, "learning_rate": 4.985533083601263e-05, "loss": 0.9444, "step": 34357 }, { "epoch": 0.03, "learning_rate": 4.985532239793715e-05, "loss": 0.8993, "step": 34358 }, { "epoch": 0.03, "learning_rate": 4.985531395961631e-05, "loss": 0.9944, "step": 34359 }, { "epoch": 0.03, "learning_rate": 4.985530552105011e-05, "loss": 0.6894, "step": 34360 }, { "epoch": 0.03, "learning_rate": 4.985529708223855e-05, "loss": 0.5894, "step": 34361 }, { "epoch": 0.03, "learning_rate": 4.985528864318163e-05, "loss": 0.8135, "step": 34362 }, { "epoch": 0.03, "learning_rate": 4.985528020387935e-05, "loss": 0.6002, "step": 34363 }, { "epoch": 0.03, "learning_rate": 4.985527176433171e-05, "loss": 0.2754, "step": 34364 }, { "epoch": 0.03, "learning_rate": 4.9855263324538704e-05, "loss": 0.5544, "step": 34365 }, { "epoch": 0.03, "learning_rate": 4.985525488450035e-05, "loss": 0.5523, "step": 34366 }, { "epoch": 0.03, "learning_rate": 4.985524644421662e-05, "loss": 1.1401, "step": 34367 }, { "epoch": 0.03, "learning_rate": 4.9855238003687535e-05, "loss": 1.0845, "step": 34368 }, { "epoch": 0.03, "learning_rate": 4.985522956291309e-05, "loss": 1.5092, "step": 34369 }, { "epoch": 0.03, "learning_rate": 4.985522112189328e-05, "loss": 0.8884, "step": 34370 }, { "epoch": 0.03, "learning_rate": 4.985521268062812e-05, "loss": 1.0539, "step": 34371 }, { "epoch": 0.03, "learning_rate": 4.985520423911759e-05, "loss": 0.9563, "step": 34372 }, { "epoch": 0.03, "learning_rate": 4.9855195797361705e-05, "loss": 0.941, "step": 34373 }, { "epoch": 0.03, "learning_rate": 4.9855187355360457e-05, "loss": 1.2124, "step": 34374 }, { "epoch": 0.03, "learning_rate": 4.985517891311385e-05, "loss": 1.1158, "step": 34375 }, { "epoch": 0.03, "learning_rate": 4.985517047062188e-05, "loss": 1.053, "step": 34376 }, { "epoch": 0.03, "learning_rate": 4.985516202788456e-05, "loss": 1.1563, "step": 34377 }, { "epoch": 0.03, "learning_rate": 4.985515358490187e-05, "loss": 1.1935, "step": 34378 }, { "epoch": 0.03, "learning_rate": 4.985514514167383e-05, "loss": 1.0544, "step": 34379 }, { "epoch": 0.03, "learning_rate": 4.9855136698200424e-05, "loss": 1.1916, "step": 34380 }, { "epoch": 0.03, "learning_rate": 4.985512825448166e-05, "loss": 1.3111, "step": 34381 }, { "epoch": 0.03, "learning_rate": 4.9855119810517535e-05, "loss": 1.2643, "step": 34382 }, { "epoch": 0.03, "learning_rate": 4.985511136630805e-05, "loss": 1.0434, "step": 34383 }, { "epoch": 0.03, "learning_rate": 4.9855102921853206e-05, "loss": 0.9914, "step": 34384 }, { "epoch": 0.03, "learning_rate": 4.9855094477153005e-05, "loss": 1.0398, "step": 34385 }, { "epoch": 0.03, "learning_rate": 4.985508603220745e-05, "loss": 1.2846, "step": 34386 }, { "epoch": 0.03, "learning_rate": 4.985507758701653e-05, "loss": 0.9126, "step": 34387 }, { "epoch": 0.03, "learning_rate": 4.9855069141580254e-05, "loss": 0.941, "step": 34388 }, { "epoch": 0.03, "learning_rate": 4.9855060695898614e-05, "loss": 1.1884, "step": 34389 }, { "epoch": 0.03, "learning_rate": 4.9855052249971625e-05, "loss": 1.126, "step": 34390 }, { "epoch": 0.03, "learning_rate": 4.985504380379926e-05, "loss": 1.3957, "step": 34391 }, { "epoch": 0.03, "learning_rate": 4.985503535738155e-05, "loss": 1.042, "step": 34392 }, { "epoch": 0.03, "learning_rate": 4.985502691071848e-05, "loss": 1.1398, "step": 34393 }, { "epoch": 0.03, "learning_rate": 4.985501846381005e-05, "loss": 1.1774, "step": 34394 }, { "epoch": 0.03, "learning_rate": 4.985501001665627e-05, "loss": 1.4098, "step": 34395 }, { "epoch": 0.03, "learning_rate": 4.985500156925712e-05, "loss": 0.8969, "step": 34396 }, { "epoch": 0.03, "learning_rate": 4.985499312161262e-05, "loss": 0.9181, "step": 34397 }, { "epoch": 0.03, "learning_rate": 4.9854984673722755e-05, "loss": 0.9174, "step": 34398 }, { "epoch": 0.03, "learning_rate": 4.985497622558754e-05, "loss": 0.814, "step": 34399 }, { "epoch": 0.03, "learning_rate": 4.985496777720696e-05, "loss": 1.3122, "step": 34400 }, { "epoch": 0.03, "learning_rate": 4.985495932858102e-05, "loss": 1.3095, "step": 34401 }, { "epoch": 0.03, "learning_rate": 4.985495087970973e-05, "loss": 1.3991, "step": 34402 }, { "epoch": 0.03, "learning_rate": 4.985494243059308e-05, "loss": 1.1026, "step": 34403 }, { "epoch": 0.03, "learning_rate": 4.985493398123107e-05, "loss": 1.265, "step": 34404 }, { "epoch": 0.03, "learning_rate": 4.98549255316237e-05, "loss": 0.7274, "step": 34405 }, { "epoch": 0.03, "learning_rate": 4.985491708177098e-05, "loss": 0.8237, "step": 34406 }, { "epoch": 0.03, "learning_rate": 4.98549086316729e-05, "loss": 0.619, "step": 34407 }, { "epoch": 0.03, "learning_rate": 4.985490018132946e-05, "loss": 1.1602, "step": 34408 }, { "epoch": 0.03, "learning_rate": 4.985489173074067e-05, "loss": 0.9617, "step": 34409 }, { "epoch": 0.03, "learning_rate": 4.985488327990652e-05, "loss": 0.8076, "step": 34410 }, { "epoch": 0.03, "learning_rate": 4.985487482882701e-05, "loss": 0.7022, "step": 34411 }, { "epoch": 0.03, "learning_rate": 4.9854866377502144e-05, "loss": 1.0199, "step": 34412 }, { "epoch": 0.03, "learning_rate": 4.9854857925931917e-05, "loss": 1.1818, "step": 34413 }, { "epoch": 0.03, "learning_rate": 4.985484947411634e-05, "loss": 0.9537, "step": 34414 }, { "epoch": 0.03, "learning_rate": 4.9854841022055404e-05, "loss": 0.791, "step": 34415 }, { "epoch": 0.03, "learning_rate": 4.985483256974911e-05, "loss": 0.618, "step": 34416 }, { "epoch": 0.03, "learning_rate": 4.9854824117197465e-05, "loss": 0.909, "step": 34417 }, { "epoch": 0.03, "learning_rate": 4.9854815664400464e-05, "loss": 1.1024, "step": 34418 }, { "epoch": 0.03, "learning_rate": 4.98548072113581e-05, "loss": 1.1297, "step": 34419 }, { "epoch": 0.03, "learning_rate": 4.985479875807038e-05, "loss": 1.1283, "step": 34420 }, { "epoch": 0.03, "learning_rate": 4.985479030453731e-05, "loss": 0.9951, "step": 34421 }, { "epoch": 0.03, "learning_rate": 4.985478185075888e-05, "loss": 1.303, "step": 34422 }, { "epoch": 0.03, "learning_rate": 4.9854773396735096e-05, "loss": 0.8492, "step": 34423 }, { "epoch": 0.03, "learning_rate": 4.985476494246595e-05, "loss": 0.8708, "step": 34424 }, { "epoch": 0.03, "learning_rate": 4.985475648795146e-05, "loss": 1.1991, "step": 34425 }, { "epoch": 0.03, "learning_rate": 4.98547480331916e-05, "loss": 1.0628, "step": 34426 }, { "epoch": 0.03, "learning_rate": 4.985473957818639e-05, "loss": 1.0638, "step": 34427 }, { "epoch": 0.03, "learning_rate": 4.9854731122935835e-05, "loss": 0.9732, "step": 34428 }, { "epoch": 0.03, "learning_rate": 4.9854722667439915e-05, "loss": 0.7582, "step": 34429 }, { "epoch": 0.03, "learning_rate": 4.985471421169864e-05, "loss": 0.876, "step": 34430 }, { "epoch": 0.03, "learning_rate": 4.9854705755712004e-05, "loss": 1.0918, "step": 34431 }, { "epoch": 0.03, "learning_rate": 4.9854697299480026e-05, "loss": 0.9135, "step": 34432 }, { "epoch": 0.03, "learning_rate": 4.9854688843002686e-05, "loss": 1.0965, "step": 34433 }, { "epoch": 0.03, "learning_rate": 4.985468038627999e-05, "loss": 1.1302, "step": 34434 }, { "epoch": 0.03, "learning_rate": 4.9854671929311935e-05, "loss": 0.6638, "step": 34435 }, { "epoch": 0.03, "learning_rate": 4.985466347209854e-05, "loss": 0.7862, "step": 34436 }, { "epoch": 0.03, "learning_rate": 4.985465501463977e-05, "loss": 0.717, "step": 34437 }, { "epoch": 0.03, "learning_rate": 4.9854646556935657e-05, "loss": 0.9856, "step": 34438 }, { "epoch": 0.03, "learning_rate": 4.9854638098986194e-05, "loss": 0.9981, "step": 34439 }, { "epoch": 0.03, "learning_rate": 4.985462964079137e-05, "loss": 1.201, "step": 34440 }, { "epoch": 0.03, "learning_rate": 4.985462118235119e-05, "loss": 0.9518, "step": 34441 }, { "epoch": 0.03, "learning_rate": 4.985461272366566e-05, "loss": 0.7803, "step": 34442 }, { "epoch": 0.03, "learning_rate": 4.9854604264734774e-05, "loss": 1.148, "step": 34443 }, { "epoch": 0.03, "learning_rate": 4.9854595805558536e-05, "loss": 1.0653, "step": 34444 }, { "epoch": 0.03, "learning_rate": 4.985458734613694e-05, "loss": 0.9986, "step": 34445 }, { "epoch": 0.03, "learning_rate": 4.985457888646999e-05, "loss": 0.6619, "step": 34446 }, { "epoch": 0.03, "learning_rate": 4.985457042655769e-05, "loss": 0.9033, "step": 34447 }, { "epoch": 0.03, "learning_rate": 4.9854561966400024e-05, "loss": 0.7544, "step": 34448 }, { "epoch": 0.03, "learning_rate": 4.985455350599702e-05, "loss": 0.9976, "step": 34449 }, { "epoch": 0.03, "learning_rate": 4.985454504534866e-05, "loss": 1.0565, "step": 34450 }, { "epoch": 0.03, "learning_rate": 4.9854536584454936e-05, "loss": 1.3199, "step": 34451 }, { "epoch": 0.03, "learning_rate": 4.9854528123315866e-05, "loss": 0.6467, "step": 34452 }, { "epoch": 0.03, "learning_rate": 4.985451966193144e-05, "loss": 0.5988, "step": 34453 }, { "epoch": 0.03, "learning_rate": 4.985451120030167e-05, "loss": 1.1813, "step": 34454 }, { "epoch": 0.03, "learning_rate": 4.985450273842653e-05, "loss": 0.8955, "step": 34455 }, { "epoch": 0.03, "learning_rate": 4.985449427630605e-05, "loss": 0.893, "step": 34456 }, { "epoch": 0.03, "learning_rate": 4.985448581394021e-05, "loss": 0.8624, "step": 34457 }, { "epoch": 0.03, "learning_rate": 4.985447735132902e-05, "loss": 1.3402, "step": 34458 }, { "epoch": 0.03, "learning_rate": 4.985446888847248e-05, "loss": 0.8585, "step": 34459 }, { "epoch": 0.03, "learning_rate": 4.985446042537059e-05, "loss": 1.048, "step": 34460 }, { "epoch": 0.03, "learning_rate": 4.985445196202333e-05, "loss": 0.9036, "step": 34461 }, { "epoch": 0.03, "learning_rate": 4.985444349843073e-05, "loss": 1.3569, "step": 34462 }, { "epoch": 0.03, "learning_rate": 4.9854435034592775e-05, "loss": 1.221, "step": 34463 }, { "epoch": 0.03, "learning_rate": 4.985442657050947e-05, "loss": 1.0226, "step": 34464 }, { "epoch": 0.03, "learning_rate": 4.985441810618081e-05, "loss": 0.9111, "step": 34465 }, { "epoch": 0.03, "learning_rate": 4.98544096416068e-05, "loss": 0.9322, "step": 34466 }, { "epoch": 0.03, "learning_rate": 4.9854401176787436e-05, "loss": 1.0081, "step": 34467 }, { "epoch": 0.03, "learning_rate": 4.985439271172272e-05, "loss": 1.0926, "step": 34468 }, { "epoch": 0.03, "learning_rate": 4.9854384246412656e-05, "loss": 0.8183, "step": 34469 }, { "epoch": 0.03, "learning_rate": 4.985437578085723e-05, "loss": 0.4543, "step": 34470 }, { "epoch": 0.03, "learning_rate": 4.985436731505646e-05, "loss": 0.7401, "step": 34471 }, { "epoch": 0.03, "learning_rate": 4.985435884901034e-05, "loss": 1.1046, "step": 34472 }, { "epoch": 0.03, "learning_rate": 4.985435038271886e-05, "loss": 1.1276, "step": 34473 }, { "epoch": 0.03, "learning_rate": 4.9854341916182034e-05, "loss": 1.1131, "step": 34474 }, { "epoch": 0.03, "learning_rate": 4.985433344939986e-05, "loss": 0.7169, "step": 34475 }, { "epoch": 0.03, "learning_rate": 4.9854324982372326e-05, "loss": 0.5857, "step": 34476 }, { "epoch": 0.03, "learning_rate": 4.985431651509944e-05, "loss": 0.9316, "step": 34477 }, { "epoch": 0.03, "learning_rate": 4.985430804758121e-05, "loss": 1.2391, "step": 34478 }, { "epoch": 0.03, "learning_rate": 4.985429957981763e-05, "loss": 0.9589, "step": 34479 }, { "epoch": 0.03, "learning_rate": 4.985429111180869e-05, "loss": 0.9729, "step": 34480 }, { "epoch": 0.03, "learning_rate": 4.985428264355441e-05, "loss": 0.7087, "step": 34481 }, { "epoch": 0.03, "learning_rate": 4.985427417505477e-05, "loss": 0.9523, "step": 34482 }, { "epoch": 0.03, "learning_rate": 4.985426570630978e-05, "loss": 1.7054, "step": 34483 }, { "epoch": 0.03, "learning_rate": 4.985425723731944e-05, "loss": 0.7755, "step": 34484 }, { "epoch": 0.03, "learning_rate": 4.985424876808375e-05, "loss": 1.0261, "step": 34485 }, { "epoch": 0.03, "learning_rate": 4.985424029860271e-05, "loss": 0.7073, "step": 34486 }, { "epoch": 0.03, "learning_rate": 4.985423182887632e-05, "loss": 0.9958, "step": 34487 }, { "epoch": 0.03, "learning_rate": 4.985422335890458e-05, "loss": 1.1362, "step": 34488 }, { "epoch": 0.03, "learning_rate": 4.9854214888687486e-05, "loss": 0.9782, "step": 34489 }, { "epoch": 0.03, "learning_rate": 4.985420641822504e-05, "loss": 1.3148, "step": 34490 }, { "epoch": 0.03, "learning_rate": 4.985419794751725e-05, "loss": 1.004, "step": 34491 }, { "epoch": 0.03, "learning_rate": 4.9854189476564105e-05, "loss": 0.7481, "step": 34492 }, { "epoch": 0.03, "learning_rate": 4.985418100536561e-05, "loss": 0.9865, "step": 34493 }, { "epoch": 0.03, "learning_rate": 4.9854172533921775e-05, "loss": 1.1892, "step": 34494 }, { "epoch": 0.03, "learning_rate": 4.985416406223258e-05, "loss": 1.218, "step": 34495 }, { "epoch": 0.03, "learning_rate": 4.985415559029804e-05, "loss": 1.1642, "step": 34496 }, { "epoch": 0.03, "learning_rate": 4.9854147118118144e-05, "loss": 1.4681, "step": 34497 }, { "epoch": 0.03, "learning_rate": 4.98541386456929e-05, "loss": 1.2973, "step": 34498 }, { "epoch": 0.03, "learning_rate": 4.9854130173022305e-05, "loss": 0.5564, "step": 34499 }, { "epoch": 0.03, "learning_rate": 4.9854121700106373e-05, "loss": 0.9956, "step": 34500 }, { "epoch": 0.03, "eval_loss": 1.070044755935669, "eval_runtime": 128.0614, "eval_samples_per_second": 10.815, "eval_steps_per_second": 5.411, "step": 34500 }, { "epoch": 0.03, "learning_rate": 4.985411322694507e-05, "loss": 0.9002, "step": 34501 }, { "epoch": 0.03, "learning_rate": 4.985410475353843e-05, "loss": 1.0467, "step": 34502 }, { "epoch": 0.03, "learning_rate": 4.985409627988645e-05, "loss": 0.806, "step": 34503 }, { "epoch": 0.03, "learning_rate": 4.985408780598911e-05, "loss": 0.9523, "step": 34504 }, { "epoch": 0.03, "learning_rate": 4.985407933184642e-05, "loss": 1.0966, "step": 34505 }, { "epoch": 0.03, "learning_rate": 4.9854070857458383e-05, "loss": 1.0425, "step": 34506 }, { "epoch": 0.03, "learning_rate": 4.9854062382824994e-05, "loss": 1.2458, "step": 34507 }, { "epoch": 0.03, "learning_rate": 4.9854053907946264e-05, "loss": 1.0693, "step": 34508 }, { "epoch": 0.03, "learning_rate": 4.9854045432822185e-05, "loss": 0.8901, "step": 34509 }, { "epoch": 0.03, "learning_rate": 4.985403695745275e-05, "loss": 0.5852, "step": 34510 }, { "epoch": 0.03, "learning_rate": 4.985402848183797e-05, "loss": 1.2944, "step": 34511 }, { "epoch": 0.03, "learning_rate": 4.985402000597784e-05, "loss": 0.6807, "step": 34512 }, { "epoch": 0.03, "learning_rate": 4.985401152987236e-05, "loss": 0.368, "step": 34513 }, { "epoch": 0.03, "learning_rate": 4.9854003053521534e-05, "loss": 0.8308, "step": 34514 }, { "epoch": 0.03, "learning_rate": 4.9853994576925366e-05, "loss": 1.1826, "step": 34515 }, { "epoch": 0.03, "learning_rate": 4.9853986100083843e-05, "loss": 1.1465, "step": 34516 }, { "epoch": 0.03, "learning_rate": 4.985397762299697e-05, "loss": 0.9345, "step": 34517 }, { "epoch": 0.03, "learning_rate": 4.985396914566476e-05, "loss": 1.1336, "step": 34518 }, { "epoch": 0.03, "learning_rate": 4.9853960668087186e-05, "loss": 1.0758, "step": 34519 }, { "epoch": 0.03, "learning_rate": 4.985395219026428e-05, "loss": 1.0667, "step": 34520 }, { "epoch": 0.03, "learning_rate": 4.9853943712196013e-05, "loss": 1.264, "step": 34521 }, { "epoch": 0.03, "learning_rate": 4.98539352338824e-05, "loss": 0.8498, "step": 34522 }, { "epoch": 0.03, "learning_rate": 4.985392675532345e-05, "loss": 1.0258, "step": 34523 }, { "epoch": 0.03, "learning_rate": 4.985391827651914e-05, "loss": 1.319, "step": 34524 }, { "epoch": 0.03, "learning_rate": 4.985390979746949e-05, "loss": 0.6629, "step": 34525 }, { "epoch": 0.03, "learning_rate": 4.985390131817449e-05, "loss": 1.2153, "step": 34526 }, { "epoch": 0.03, "learning_rate": 4.985389283863415e-05, "loss": 1.1488, "step": 34527 }, { "epoch": 0.03, "learning_rate": 4.985388435884846e-05, "loss": 1.4977, "step": 34528 }, { "epoch": 0.03, "learning_rate": 4.985387587881741e-05, "loss": 1.187, "step": 34529 }, { "epoch": 0.03, "learning_rate": 4.9853867398541026e-05, "loss": 1.0183, "step": 34530 }, { "epoch": 0.03, "learning_rate": 4.9853858918019295e-05, "loss": 1.2348, "step": 34531 }, { "epoch": 0.03, "learning_rate": 4.9853850437252214e-05, "loss": 1.1398, "step": 34532 }, { "epoch": 0.03, "learning_rate": 4.9853841956239786e-05, "loss": 0.9759, "step": 34533 }, { "epoch": 0.03, "learning_rate": 4.985383347498201e-05, "loss": 1.4542, "step": 34534 }, { "epoch": 0.03, "learning_rate": 4.9853824993478885e-05, "loss": 1.2052, "step": 34535 }, { "epoch": 0.03, "learning_rate": 4.985381651173042e-05, "loss": 1.1986, "step": 34536 }, { "epoch": 0.03, "learning_rate": 4.9853808029736605e-05, "loss": 1.0271, "step": 34537 }, { "epoch": 0.03, "learning_rate": 4.985379954749745e-05, "loss": 1.5995, "step": 34538 }, { "epoch": 0.03, "learning_rate": 4.985379106501294e-05, "loss": 0.966, "step": 34539 }, { "epoch": 0.03, "learning_rate": 4.985378258228309e-05, "loss": 0.9658, "step": 34540 }, { "epoch": 0.03, "learning_rate": 4.985377409930789e-05, "loss": 0.8953, "step": 34541 }, { "epoch": 0.03, "learning_rate": 4.9853765616087345e-05, "loss": 1.0207, "step": 34542 }, { "epoch": 0.03, "learning_rate": 4.985375713262145e-05, "loss": 0.9769, "step": 34543 }, { "epoch": 0.03, "learning_rate": 4.985374864891022e-05, "loss": 0.8908, "step": 34544 }, { "epoch": 0.03, "learning_rate": 4.985374016495363e-05, "loss": 1.0936, "step": 34545 }, { "epoch": 0.03, "learning_rate": 4.9853731680751704e-05, "loss": 1.1127, "step": 34546 }, { "epoch": 0.03, "learning_rate": 4.9853723196304435e-05, "loss": 0.7548, "step": 34547 }, { "epoch": 0.03, "learning_rate": 4.985371471161181e-05, "loss": 0.9137, "step": 34548 }, { "epoch": 0.03, "learning_rate": 4.9853706226673846e-05, "loss": 0.8141, "step": 34549 }, { "epoch": 0.03, "learning_rate": 4.985369774149054e-05, "loss": 1.2328, "step": 34550 }, { "epoch": 0.03, "learning_rate": 4.9853689256061885e-05, "loss": 1.1128, "step": 34551 }, { "epoch": 0.03, "learning_rate": 4.985368077038788e-05, "loss": 1.0795, "step": 34552 }, { "epoch": 0.03, "learning_rate": 4.985367228446854e-05, "loss": 0.8949, "step": 34553 }, { "epoch": 0.03, "learning_rate": 4.985366379830385e-05, "loss": 0.8482, "step": 34554 }, { "epoch": 0.03, "learning_rate": 4.985365531189382e-05, "loss": 0.9408, "step": 34555 }, { "epoch": 0.03, "learning_rate": 4.985364682523843e-05, "loss": 0.9731, "step": 34556 }, { "epoch": 0.03, "learning_rate": 4.985363833833771e-05, "loss": 0.9034, "step": 34557 }, { "epoch": 0.03, "learning_rate": 4.985362985119164e-05, "loss": 0.9098, "step": 34558 }, { "epoch": 0.03, "learning_rate": 4.985362136380023e-05, "loss": 0.9689, "step": 34559 }, { "epoch": 0.03, "learning_rate": 4.985361287616347e-05, "loss": 1.0274, "step": 34560 }, { "epoch": 0.03, "learning_rate": 4.9853604388281364e-05, "loss": 0.918, "step": 34561 }, { "epoch": 0.03, "learning_rate": 4.985359590015392e-05, "loss": 0.956, "step": 34562 }, { "epoch": 0.03, "learning_rate": 4.9853587411781135e-05, "loss": 0.8151, "step": 34563 }, { "epoch": 0.03, "learning_rate": 4.9853578923162994e-05, "loss": 0.9388, "step": 34564 }, { "epoch": 0.03, "learning_rate": 4.985357043429951e-05, "loss": 1.1012, "step": 34565 }, { "epoch": 0.03, "learning_rate": 4.985356194519069e-05, "loss": 0.8647, "step": 34566 }, { "epoch": 0.03, "learning_rate": 4.985355345583652e-05, "loss": 0.7398, "step": 34567 }, { "epoch": 0.03, "learning_rate": 4.985354496623701e-05, "loss": 1.0564, "step": 34568 }, { "epoch": 0.03, "learning_rate": 4.985353647639216e-05, "loss": 0.8831, "step": 34569 }, { "epoch": 0.03, "learning_rate": 4.985352798630196e-05, "loss": 1.0952, "step": 34570 }, { "epoch": 0.03, "learning_rate": 4.985351949596642e-05, "loss": 0.7677, "step": 34571 }, { "epoch": 0.03, "learning_rate": 4.9853511005385533e-05, "loss": 0.5903, "step": 34572 }, { "epoch": 0.03, "learning_rate": 4.98535025145593e-05, "loss": 0.9774, "step": 34573 }, { "epoch": 0.03, "learning_rate": 4.985349402348773e-05, "loss": 0.7482, "step": 34574 }, { "epoch": 0.03, "learning_rate": 4.9853485532170816e-05, "loss": 0.9648, "step": 34575 }, { "epoch": 0.03, "learning_rate": 4.985347704060856e-05, "loss": 0.8624, "step": 34576 }, { "epoch": 0.03, "learning_rate": 4.985346854880096e-05, "loss": 1.1364, "step": 34577 }, { "epoch": 0.03, "learning_rate": 4.985346005674801e-05, "loss": 1.2188, "step": 34578 }, { "epoch": 0.03, "learning_rate": 4.9853451564449724e-05, "loss": 0.9729, "step": 34579 }, { "epoch": 0.03, "learning_rate": 4.9853443071906094e-05, "loss": 0.869, "step": 34580 }, { "epoch": 0.03, "learning_rate": 4.985343457911712e-05, "loss": 1.0993, "step": 34581 }, { "epoch": 0.03, "learning_rate": 4.985342608608281e-05, "loss": 0.9205, "step": 34582 }, { "epoch": 0.03, "learning_rate": 4.9853417592803156e-05, "loss": 0.8578, "step": 34583 }, { "epoch": 0.03, "learning_rate": 4.985340909927815e-05, "loss": 0.0849, "step": 34584 }, { "epoch": 0.03, "learning_rate": 4.9853400605507804e-05, "loss": 0.0617, "step": 34585 }, { "epoch": 0.03, "learning_rate": 4.985339211149212e-05, "loss": 0.1339, "step": 34586 }, { "epoch": 0.03, "learning_rate": 4.98533836172311e-05, "loss": 0.139, "step": 34587 }, { "epoch": 0.03, "learning_rate": 4.9853375122724724e-05, "loss": 0.0544, "step": 34588 }, { "epoch": 0.03, "learning_rate": 4.9853366627973015e-05, "loss": 0.7058, "step": 34589 }, { "epoch": 0.03, "learning_rate": 4.985335813297596e-05, "loss": 0.9635, "step": 34590 }, { "epoch": 0.03, "learning_rate": 4.9853349637733566e-05, "loss": 0.8457, "step": 34591 }, { "epoch": 0.03, "learning_rate": 4.9853341142245825e-05, "loss": 1.3173, "step": 34592 }, { "epoch": 0.03, "learning_rate": 4.985333264651275e-05, "loss": 0.6016, "step": 34593 }, { "epoch": 0.03, "learning_rate": 4.985332415053433e-05, "loss": 0.5378, "step": 34594 }, { "epoch": 0.03, "learning_rate": 4.985331565431056e-05, "loss": 0.4166, "step": 34595 }, { "epoch": 0.03, "learning_rate": 4.9853307157841464e-05, "loss": 0.8026, "step": 34596 }, { "epoch": 0.03, "learning_rate": 4.985329866112702e-05, "loss": 0.4371, "step": 34597 }, { "epoch": 0.03, "learning_rate": 4.9853290164167235e-05, "loss": 0.2057, "step": 34598 }, { "epoch": 0.03, "learning_rate": 4.9853281666962105e-05, "loss": 0.1412, "step": 34599 }, { "epoch": 0.03, "learning_rate": 4.985327316951164e-05, "loss": 0.2341, "step": 34600 }, { "epoch": 0.03, "learning_rate": 4.985326467181583e-05, "loss": 0.4327, "step": 34601 }, { "epoch": 0.03, "learning_rate": 4.9853256173874674e-05, "loss": 0.2939, "step": 34602 }, { "epoch": 0.03, "learning_rate": 4.9853247675688185e-05, "loss": 0.6735, "step": 34603 }, { "epoch": 0.03, "learning_rate": 4.985323917725635e-05, "loss": 0.5166, "step": 34604 }, { "epoch": 0.03, "learning_rate": 4.9853230678579176e-05, "loss": 0.7071, "step": 34605 }, { "epoch": 0.03, "learning_rate": 4.985322217965667e-05, "loss": 0.3008, "step": 34606 }, { "epoch": 0.03, "learning_rate": 4.985321368048881e-05, "loss": 0.4482, "step": 34607 }, { "epoch": 0.03, "learning_rate": 4.985320518107561e-05, "loss": 0.5374, "step": 34608 }, { "epoch": 0.03, "learning_rate": 4.985319668141708e-05, "loss": 0.3154, "step": 34609 }, { "epoch": 0.03, "learning_rate": 4.985318818151321e-05, "loss": 1.0703, "step": 34610 }, { "epoch": 0.03, "learning_rate": 4.985317968136399e-05, "loss": 1.2456, "step": 34611 }, { "epoch": 0.03, "learning_rate": 4.9853171180969436e-05, "loss": 0.9171, "step": 34612 }, { "epoch": 0.03, "learning_rate": 4.985316268032954e-05, "loss": 0.9166, "step": 34613 }, { "epoch": 0.03, "learning_rate": 4.98531541794443e-05, "loss": 0.8101, "step": 34614 }, { "epoch": 0.03, "learning_rate": 4.985314567831373e-05, "loss": 1.1228, "step": 34615 }, { "epoch": 0.03, "learning_rate": 4.985313717693782e-05, "loss": 0.921, "step": 34616 }, { "epoch": 0.03, "learning_rate": 4.985312867531656e-05, "loss": 1.251, "step": 34617 }, { "epoch": 0.03, "learning_rate": 4.9853120173449966e-05, "loss": 1.3111, "step": 34618 }, { "epoch": 0.03, "learning_rate": 4.9853111671338035e-05, "loss": 0.9955, "step": 34619 }, { "epoch": 0.03, "learning_rate": 4.9853103168980756e-05, "loss": 1.1905, "step": 34620 }, { "epoch": 0.03, "learning_rate": 4.985309466637814e-05, "loss": 0.9228, "step": 34621 }, { "epoch": 0.03, "learning_rate": 4.9853086163530194e-05, "loss": 1.2264, "step": 34622 }, { "epoch": 0.03, "learning_rate": 4.9853077660436904e-05, "loss": 1.607, "step": 34623 }, { "epoch": 0.03, "learning_rate": 4.985306915709827e-05, "loss": 1.1561, "step": 34624 }, { "epoch": 0.03, "learning_rate": 4.985306065351431e-05, "loss": 0.6487, "step": 34625 }, { "epoch": 0.03, "learning_rate": 4.9853052149685e-05, "loss": 0.8514, "step": 34626 }, { "epoch": 0.03, "learning_rate": 4.9853043645610345e-05, "loss": 0.8816, "step": 34627 }, { "epoch": 0.03, "learning_rate": 4.985303514129036e-05, "loss": 0.8512, "step": 34628 }, { "epoch": 0.03, "learning_rate": 4.985302663672504e-05, "loss": 0.829, "step": 34629 }, { "epoch": 0.03, "learning_rate": 4.985301813191438e-05, "loss": 1.0589, "step": 34630 }, { "epoch": 0.03, "learning_rate": 4.985300962685837e-05, "loss": 0.5872, "step": 34631 }, { "epoch": 0.03, "learning_rate": 4.985300112155703e-05, "loss": 0.9229, "step": 34632 }, { "epoch": 0.03, "learning_rate": 4.985299261601035e-05, "loss": 1.1925, "step": 34633 }, { "epoch": 0.03, "learning_rate": 4.9852984110218334e-05, "loss": 0.9117, "step": 34634 }, { "epoch": 0.03, "learning_rate": 4.985297560418098e-05, "loss": 0.572, "step": 34635 }, { "epoch": 0.03, "learning_rate": 4.985296709789829e-05, "loss": 1.1107, "step": 34636 }, { "epoch": 0.03, "learning_rate": 4.985295859137026e-05, "loss": 1.0939, "step": 34637 }, { "epoch": 0.03, "learning_rate": 4.985295008459689e-05, "loss": 1.1331, "step": 34638 }, { "epoch": 0.03, "learning_rate": 4.9852941577578184e-05, "loss": 1.0943, "step": 34639 }, { "epoch": 0.03, "learning_rate": 4.985293307031413e-05, "loss": 0.4492, "step": 34640 }, { "epoch": 0.03, "learning_rate": 4.985292456280475e-05, "loss": 0.4913, "step": 34641 }, { "epoch": 0.03, "learning_rate": 4.9852916055050035e-05, "loss": 1.2316, "step": 34642 }, { "epoch": 0.03, "learning_rate": 4.985290754704998e-05, "loss": 1.1206, "step": 34643 }, { "epoch": 0.03, "learning_rate": 4.985289903880458e-05, "loss": 0.8769, "step": 34644 }, { "epoch": 0.03, "learning_rate": 4.985289053031385e-05, "loss": 1.2616, "step": 34645 }, { "epoch": 0.03, "learning_rate": 4.985288202157779e-05, "loss": 0.9927, "step": 34646 }, { "epoch": 0.03, "learning_rate": 4.985287351259638e-05, "loss": 0.8877, "step": 34647 }, { "epoch": 0.03, "learning_rate": 4.985286500336963e-05, "loss": 0.8737, "step": 34648 }, { "epoch": 0.03, "learning_rate": 4.985285649389755e-05, "loss": 0.8282, "step": 34649 }, { "epoch": 0.03, "learning_rate": 4.985284798418013e-05, "loss": 0.7544, "step": 34650 }, { "epoch": 0.03, "learning_rate": 4.985283947421738e-05, "loss": 0.7669, "step": 34651 }, { "epoch": 0.03, "learning_rate": 4.985283096400929e-05, "loss": 0.8811, "step": 34652 }, { "epoch": 0.03, "learning_rate": 4.985282245355586e-05, "loss": 1.6391, "step": 34653 }, { "epoch": 0.03, "learning_rate": 4.9852813942857094e-05, "loss": 1.6844, "step": 34654 }, { "epoch": 0.03, "learning_rate": 4.9852805431912994e-05, "loss": 1.0814, "step": 34655 }, { "epoch": 0.03, "learning_rate": 4.9852796920723554e-05, "loss": 1.0355, "step": 34656 }, { "epoch": 0.03, "learning_rate": 4.985278840928878e-05, "loss": 0.718, "step": 34657 }, { "epoch": 0.03, "learning_rate": 4.985277989760867e-05, "loss": 0.677, "step": 34658 }, { "epoch": 0.03, "learning_rate": 4.985277138568323e-05, "loss": 1.1842, "step": 34659 }, { "epoch": 0.03, "learning_rate": 4.9852762873512445e-05, "loss": 1.1563, "step": 34660 }, { "epoch": 0.03, "learning_rate": 4.9852754361096324e-05, "loss": 1.064, "step": 34661 }, { "epoch": 0.03, "learning_rate": 4.985274584843487e-05, "loss": 0.9928, "step": 34662 }, { "epoch": 0.03, "learning_rate": 4.985273733552808e-05, "loss": 0.844, "step": 34663 }, { "epoch": 0.03, "learning_rate": 4.9852728822375955e-05, "loss": 0.896, "step": 34664 }, { "epoch": 0.03, "learning_rate": 4.9852720308978496e-05, "loss": 1.0529, "step": 34665 }, { "epoch": 0.03, "learning_rate": 4.9852711795335695e-05, "loss": 0.6596, "step": 34666 }, { "epoch": 0.03, "learning_rate": 4.985270328144757e-05, "loss": 0.7632, "step": 34667 }, { "epoch": 0.03, "learning_rate": 4.98526947673141e-05, "loss": 0.4963, "step": 34668 }, { "epoch": 0.03, "learning_rate": 4.9852686252935286e-05, "loss": 0.5454, "step": 34669 }, { "epoch": 0.03, "learning_rate": 4.985267773831115e-05, "loss": 0.9525, "step": 34670 }, { "epoch": 0.03, "learning_rate": 4.985266922344168e-05, "loss": 1.1025, "step": 34671 }, { "epoch": 0.03, "learning_rate": 4.985266070832687e-05, "loss": 0.593, "step": 34672 }, { "epoch": 0.03, "learning_rate": 4.985265219296672e-05, "loss": 0.8715, "step": 34673 }, { "epoch": 0.03, "learning_rate": 4.9852643677361245e-05, "loss": 1.0766, "step": 34674 }, { "epoch": 0.03, "learning_rate": 4.985263516151043e-05, "loss": 1.0143, "step": 34675 }, { "epoch": 0.03, "learning_rate": 4.9852626645414287e-05, "loss": 1.2449, "step": 34676 }, { "epoch": 0.03, "learning_rate": 4.9852618129072805e-05, "loss": 0.9316, "step": 34677 }, { "epoch": 0.03, "learning_rate": 4.985260961248598e-05, "loss": 1.14, "step": 34678 }, { "epoch": 0.03, "learning_rate": 4.985260109565383e-05, "loss": 1.0306, "step": 34679 }, { "epoch": 0.03, "learning_rate": 4.985259257857635e-05, "loss": 0.8377, "step": 34680 }, { "epoch": 0.03, "learning_rate": 4.985258406125353e-05, "loss": 0.5928, "step": 34681 }, { "epoch": 0.03, "learning_rate": 4.985257554368537e-05, "loss": 1.1061, "step": 34682 }, { "epoch": 0.03, "learning_rate": 4.9852567025871884e-05, "loss": 0.8376, "step": 34683 }, { "epoch": 0.03, "learning_rate": 4.985255850781306e-05, "loss": 1.5607, "step": 34684 }, { "epoch": 0.03, "learning_rate": 4.98525499895089e-05, "loss": 1.0265, "step": 34685 }, { "epoch": 0.03, "learning_rate": 4.9852541470959415e-05, "loss": 1.1198, "step": 34686 }, { "epoch": 0.03, "learning_rate": 4.985253295216459e-05, "loss": 0.885, "step": 34687 }, { "epoch": 0.03, "learning_rate": 4.9852524433124426e-05, "loss": 1.0302, "step": 34688 }, { "epoch": 0.03, "learning_rate": 4.9852515913838936e-05, "loss": 1.0395, "step": 34689 }, { "epoch": 0.03, "learning_rate": 4.985250739430811e-05, "loss": 0.8733, "step": 34690 }, { "epoch": 0.03, "learning_rate": 4.985249887453195e-05, "loss": 0.8179, "step": 34691 }, { "epoch": 0.03, "learning_rate": 4.985249035451046e-05, "loss": 0.9458, "step": 34692 }, { "epoch": 0.03, "learning_rate": 4.985248183424363e-05, "loss": 1.7071, "step": 34693 }, { "epoch": 0.03, "learning_rate": 4.985247331373147e-05, "loss": 1.3194, "step": 34694 }, { "epoch": 0.03, "learning_rate": 4.985246479297398e-05, "loss": 1.3808, "step": 34695 }, { "epoch": 0.03, "learning_rate": 4.9852456271971157e-05, "loss": 0.988, "step": 34696 }, { "epoch": 0.03, "learning_rate": 4.9852447750722996e-05, "loss": 1.0318, "step": 34697 }, { "epoch": 0.03, "learning_rate": 4.98524392292295e-05, "loss": 1.1555, "step": 34698 }, { "epoch": 0.03, "learning_rate": 4.9852430707490686e-05, "loss": 0.8348, "step": 34699 }, { "epoch": 0.03, "learning_rate": 4.985242218550652e-05, "loss": 0.866, "step": 34700 }, { "epoch": 0.03, "learning_rate": 4.985241366327703e-05, "loss": 0.9565, "step": 34701 }, { "epoch": 0.03, "learning_rate": 4.985240514080221e-05, "loss": 1.0578, "step": 34702 }, { "epoch": 0.03, "learning_rate": 4.9852396618082055e-05, "loss": 0.8996, "step": 34703 }, { "epoch": 0.03, "learning_rate": 4.985238809511657e-05, "loss": 0.9334, "step": 34704 }, { "epoch": 0.03, "learning_rate": 4.985237957190575e-05, "loss": 1.0315, "step": 34705 }, { "epoch": 0.03, "learning_rate": 4.98523710484496e-05, "loss": 1.0671, "step": 34706 }, { "epoch": 0.03, "learning_rate": 4.985236252474812e-05, "loss": 1.1546, "step": 34707 }, { "epoch": 0.03, "learning_rate": 4.98523540008013e-05, "loss": 1.0382, "step": 34708 }, { "epoch": 0.03, "learning_rate": 4.985234547660915e-05, "loss": 0.956, "step": 34709 }, { "epoch": 0.03, "learning_rate": 4.985233695217167e-05, "loss": 1.1609, "step": 34710 }, { "epoch": 0.03, "learning_rate": 4.985232842748886e-05, "loss": 0.9149, "step": 34711 }, { "epoch": 0.03, "learning_rate": 4.9852319902560716e-05, "loss": 0.9112, "step": 34712 }, { "epoch": 0.03, "learning_rate": 4.985231137738724e-05, "loss": 1.2342, "step": 34713 }, { "epoch": 0.03, "learning_rate": 4.9852302851968435e-05, "loss": 0.8864, "step": 34714 }, { "epoch": 0.03, "learning_rate": 4.98522943263043e-05, "loss": 1.1842, "step": 34715 }, { "epoch": 0.03, "learning_rate": 4.9852285800394835e-05, "loss": 0.9632, "step": 34716 }, { "epoch": 0.03, "learning_rate": 4.9852277274240036e-05, "loss": 0.6628, "step": 34717 }, { "epoch": 0.03, "learning_rate": 4.98522687478399e-05, "loss": 1.0577, "step": 34718 }, { "epoch": 0.03, "learning_rate": 4.9852260221194436e-05, "loss": 0.8766, "step": 34719 }, { "epoch": 0.03, "learning_rate": 4.985225169430364e-05, "loss": 0.8547, "step": 34720 }, { "epoch": 0.03, "learning_rate": 4.9852243167167523e-05, "loss": 0.8861, "step": 34721 }, { "epoch": 0.03, "learning_rate": 4.9852234639786065e-05, "loss": 1.1018, "step": 34722 }, { "epoch": 0.03, "learning_rate": 4.985222611215928e-05, "loss": 0.6773, "step": 34723 }, { "epoch": 0.03, "learning_rate": 4.985221758428716e-05, "loss": 0.8537, "step": 34724 }, { "epoch": 0.03, "learning_rate": 4.985220905616971e-05, "loss": 1.0186, "step": 34725 }, { "epoch": 0.03, "learning_rate": 4.9852200527806934e-05, "loss": 1.0239, "step": 34726 }, { "epoch": 0.03, "learning_rate": 4.985219199919883e-05, "loss": 0.6506, "step": 34727 }, { "epoch": 0.03, "learning_rate": 4.9852183470345385e-05, "loss": 0.6663, "step": 34728 }, { "epoch": 0.03, "learning_rate": 4.985217494124662e-05, "loss": 0.658, "step": 34729 }, { "epoch": 0.03, "learning_rate": 4.985216641190252e-05, "loss": 0.5966, "step": 34730 }, { "epoch": 0.03, "learning_rate": 4.985215788231309e-05, "loss": 0.5591, "step": 34731 }, { "epoch": 0.03, "learning_rate": 4.9852149352478325e-05, "loss": 0.5731, "step": 34732 }, { "epoch": 0.03, "learning_rate": 4.985214082239824e-05, "loss": 0.6202, "step": 34733 }, { "epoch": 0.03, "learning_rate": 4.9852132292072814e-05, "loss": 1.2049, "step": 34734 }, { "epoch": 0.03, "learning_rate": 4.985212376150207e-05, "loss": 1.111, "step": 34735 }, { "epoch": 0.03, "learning_rate": 4.985211523068599e-05, "loss": 1.1131, "step": 34736 }, { "epoch": 0.03, "learning_rate": 4.985210669962458e-05, "loss": 0.8994, "step": 34737 }, { "epoch": 0.03, "learning_rate": 4.9852098168317845e-05, "loss": 1.1207, "step": 34738 }, { "epoch": 0.03, "learning_rate": 4.985208963676577e-05, "loss": 0.9284, "step": 34739 }, { "epoch": 0.03, "learning_rate": 4.9852081104968387e-05, "loss": 0.4906, "step": 34740 }, { "epoch": 0.03, "learning_rate": 4.985207257292566e-05, "loss": 0.7293, "step": 34741 }, { "epoch": 0.03, "learning_rate": 4.98520640406376e-05, "loss": 1.0558, "step": 34742 }, { "epoch": 0.03, "learning_rate": 4.985205550810421e-05, "loss": 0.5153, "step": 34743 }, { "epoch": 0.03, "learning_rate": 4.985204697532551e-05, "loss": 0.6357, "step": 34744 }, { "epoch": 0.03, "learning_rate": 4.985203844230146e-05, "loss": 0.705, "step": 34745 }, { "epoch": 0.03, "learning_rate": 4.985202990903209e-05, "loss": 0.6337, "step": 34746 }, { "epoch": 0.03, "learning_rate": 4.985202137551739e-05, "loss": 1.3914, "step": 34747 }, { "epoch": 0.03, "learning_rate": 4.985201284175736e-05, "loss": 1.2433, "step": 34748 }, { "epoch": 0.03, "learning_rate": 4.985200430775201e-05, "loss": 1.3301, "step": 34749 }, { "epoch": 0.03, "learning_rate": 4.9851995773501315e-05, "loss": 1.0067, "step": 34750 }, { "epoch": 0.03, "learning_rate": 4.98519872390053e-05, "loss": 0.9287, "step": 34751 }, { "epoch": 0.03, "learning_rate": 4.9851978704263955e-05, "loss": 0.7264, "step": 34752 }, { "epoch": 0.03, "learning_rate": 4.985197016927729e-05, "loss": 1.1368, "step": 34753 }, { "epoch": 0.03, "learning_rate": 4.985196163404529e-05, "loss": 0.9996, "step": 34754 }, { "epoch": 0.03, "learning_rate": 4.985195309856796e-05, "loss": 0.9146, "step": 34755 }, { "epoch": 0.03, "learning_rate": 4.985194456284531e-05, "loss": 1.0284, "step": 34756 }, { "epoch": 0.03, "learning_rate": 4.985193602687732e-05, "loss": 1.0017, "step": 34757 }, { "epoch": 0.03, "learning_rate": 4.985192749066401e-05, "loss": 1.0083, "step": 34758 }, { "epoch": 0.03, "learning_rate": 4.985191895420538e-05, "loss": 1.1071, "step": 34759 }, { "epoch": 0.03, "learning_rate": 4.985191041750141e-05, "loss": 1.1126, "step": 34760 }, { "epoch": 0.03, "learning_rate": 4.9851901880552113e-05, "loss": 0.7936, "step": 34761 }, { "epoch": 0.03, "learning_rate": 4.9851893343357494e-05, "loss": 1.0472, "step": 34762 }, { "epoch": 0.03, "learning_rate": 4.985188480591755e-05, "loss": 0.8933, "step": 34763 }, { "epoch": 0.03, "learning_rate": 4.9851876268232265e-05, "loss": 1.115, "step": 34764 }, { "epoch": 0.03, "learning_rate": 4.985186773030166e-05, "loss": 1.1586, "step": 34765 }, { "epoch": 0.03, "learning_rate": 4.985185919212574e-05, "loss": 1.1961, "step": 34766 }, { "epoch": 0.03, "learning_rate": 4.985185065370448e-05, "loss": 1.0175, "step": 34767 }, { "epoch": 0.03, "learning_rate": 4.9851842115037893e-05, "loss": 1.0945, "step": 34768 }, { "epoch": 0.03, "learning_rate": 4.9851833576125986e-05, "loss": 0.9411, "step": 34769 }, { "epoch": 0.03, "learning_rate": 4.9851825036968744e-05, "loss": 0.9475, "step": 34770 }, { "epoch": 0.03, "learning_rate": 4.985181649756618e-05, "loss": 1.1305, "step": 34771 }, { "epoch": 0.03, "learning_rate": 4.985180795791828e-05, "loss": 0.6549, "step": 34772 }, { "epoch": 0.03, "learning_rate": 4.985179941802507e-05, "loss": 0.7362, "step": 34773 }, { "epoch": 0.03, "learning_rate": 4.9851790877886524e-05, "loss": 0.8679, "step": 34774 }, { "epoch": 0.03, "learning_rate": 4.985178233750265e-05, "loss": 0.6782, "step": 34775 }, { "epoch": 0.03, "learning_rate": 4.985177379687346e-05, "loss": 0.88, "step": 34776 }, { "epoch": 0.03, "learning_rate": 4.985176525599893e-05, "loss": 1.2806, "step": 34777 }, { "epoch": 0.03, "learning_rate": 4.9851756714879086e-05, "loss": 1.136, "step": 34778 }, { "epoch": 0.03, "learning_rate": 4.98517481735139e-05, "loss": 1.0058, "step": 34779 }, { "epoch": 0.03, "learning_rate": 4.9851739631903407e-05, "loss": 1.0606, "step": 34780 }, { "epoch": 0.03, "learning_rate": 4.985173109004758e-05, "loss": 1.1788, "step": 34781 }, { "epoch": 0.03, "learning_rate": 4.985172254794642e-05, "loss": 1.0498, "step": 34782 }, { "epoch": 0.03, "learning_rate": 4.985171400559994e-05, "loss": 0.9889, "step": 34783 }, { "epoch": 0.03, "learning_rate": 4.9851705463008134e-05, "loss": 0.8664, "step": 34784 }, { "epoch": 0.03, "learning_rate": 4.985169692017101e-05, "loss": 1.1718, "step": 34785 }, { "epoch": 0.03, "learning_rate": 4.9851688377088554e-05, "loss": 1.0763, "step": 34786 }, { "epoch": 0.03, "learning_rate": 4.9851679833760766e-05, "loss": 1.133, "step": 34787 }, { "epoch": 0.03, "learning_rate": 4.985167129018766e-05, "loss": 1.164, "step": 34788 }, { "epoch": 0.03, "learning_rate": 4.985166274636923e-05, "loss": 1.1523, "step": 34789 }, { "epoch": 0.03, "learning_rate": 4.9851654202305474e-05, "loss": 1.1769, "step": 34790 }, { "epoch": 0.03, "learning_rate": 4.9851645657996395e-05, "loss": 1.2291, "step": 34791 }, { "epoch": 0.03, "learning_rate": 4.985163711344199e-05, "loss": 1.3421, "step": 34792 }, { "epoch": 0.03, "learning_rate": 4.985162856864225e-05, "loss": 0.7721, "step": 34793 }, { "epoch": 0.03, "learning_rate": 4.9851620023597196e-05, "loss": 1.0975, "step": 34794 }, { "epoch": 0.03, "learning_rate": 4.985161147830681e-05, "loss": 1.2812, "step": 34795 }, { "epoch": 0.03, "learning_rate": 4.985160293277111e-05, "loss": 0.9846, "step": 34796 }, { "epoch": 0.03, "learning_rate": 4.9851594386990074e-05, "loss": 1.1827, "step": 34797 }, { "epoch": 0.03, "learning_rate": 4.985158584096372e-05, "loss": 1.0064, "step": 34798 }, { "epoch": 0.03, "learning_rate": 4.9851577294692045e-05, "loss": 1.2609, "step": 34799 }, { "epoch": 0.03, "learning_rate": 4.9851568748175035e-05, "loss": 1.0856, "step": 34800 }, { "epoch": 0.03, "learning_rate": 4.9851560201412705e-05, "loss": 1.105, "step": 34801 }, { "epoch": 0.03, "learning_rate": 4.985155165440506e-05, "loss": 0.4654, "step": 34802 }, { "epoch": 0.03, "learning_rate": 4.985154310715208e-05, "loss": 0.4823, "step": 34803 }, { "epoch": 0.03, "learning_rate": 4.985153455965378e-05, "loss": 0.5348, "step": 34804 }, { "epoch": 0.03, "learning_rate": 4.985152601191016e-05, "loss": 0.4409, "step": 34805 }, { "epoch": 0.03, "learning_rate": 4.985151746392121e-05, "loss": 0.4025, "step": 34806 }, { "epoch": 0.03, "learning_rate": 4.985150891568694e-05, "loss": 0.2957, "step": 34807 }, { "epoch": 0.03, "learning_rate": 4.985150036720734e-05, "loss": 0.6545, "step": 34808 }, { "epoch": 0.03, "learning_rate": 4.985149181848242e-05, "loss": 0.7997, "step": 34809 }, { "epoch": 0.03, "learning_rate": 4.985148326951218e-05, "loss": 1.0515, "step": 34810 }, { "epoch": 0.03, "learning_rate": 4.985147472029661e-05, "loss": 0.6765, "step": 34811 }, { "epoch": 0.03, "learning_rate": 4.9851466170835726e-05, "loss": 1.0246, "step": 34812 }, { "epoch": 0.03, "learning_rate": 4.9851457621129515e-05, "loss": 0.988, "step": 34813 }, { "epoch": 0.03, "learning_rate": 4.9851449071177984e-05, "loss": 0.9399, "step": 34814 }, { "epoch": 0.03, "learning_rate": 4.985144052098112e-05, "loss": 1.2341, "step": 34815 }, { "epoch": 0.03, "learning_rate": 4.9851431970538936e-05, "loss": 1.1837, "step": 34816 }, { "epoch": 0.03, "learning_rate": 4.9851423419851435e-05, "loss": 1.0106, "step": 34817 }, { "epoch": 0.03, "learning_rate": 4.9851414868918605e-05, "loss": 1.0361, "step": 34818 }, { "epoch": 0.03, "learning_rate": 4.9851406317740454e-05, "loss": 1.0299, "step": 34819 }, { "epoch": 0.03, "learning_rate": 4.985139776631699e-05, "loss": 0.8997, "step": 34820 }, { "epoch": 0.03, "learning_rate": 4.985138921464819e-05, "loss": 0.8869, "step": 34821 }, { "epoch": 0.03, "learning_rate": 4.985138066273407e-05, "loss": 0.6425, "step": 34822 }, { "epoch": 0.03, "learning_rate": 4.9851372110574634e-05, "loss": 1.0946, "step": 34823 }, { "epoch": 0.03, "learning_rate": 4.985136355816987e-05, "loss": 1.5798, "step": 34824 }, { "epoch": 0.03, "learning_rate": 4.985135500551979e-05, "loss": 1.4891, "step": 34825 }, { "epoch": 0.03, "learning_rate": 4.985134645262438e-05, "loss": 1.1264, "step": 34826 }, { "epoch": 0.03, "learning_rate": 4.985133789948365e-05, "loss": 1.3564, "step": 34827 }, { "epoch": 0.03, "learning_rate": 4.98513293460976e-05, "loss": 1.2601, "step": 34828 }, { "epoch": 0.03, "learning_rate": 4.985132079246623e-05, "loss": 1.0127, "step": 34829 }, { "epoch": 0.03, "learning_rate": 4.985131223858954e-05, "loss": 1.068, "step": 34830 }, { "epoch": 0.03, "learning_rate": 4.985130368446752e-05, "loss": 0.8386, "step": 34831 }, { "epoch": 0.03, "learning_rate": 4.985129513010018e-05, "loss": 1.0739, "step": 34832 }, { "epoch": 0.03, "learning_rate": 4.985128657548753e-05, "loss": 1.3338, "step": 34833 }, { "epoch": 0.03, "learning_rate": 4.985127802062954e-05, "loss": 1.1487, "step": 34834 }, { "epoch": 0.03, "learning_rate": 4.985126946552624e-05, "loss": 1.0874, "step": 34835 }, { "epoch": 0.03, "learning_rate": 4.985126091017762e-05, "loss": 1.0312, "step": 34836 }, { "epoch": 0.03, "learning_rate": 4.9851252354583676e-05, "loss": 1.1719, "step": 34837 }, { "epoch": 0.03, "learning_rate": 4.985124379874441e-05, "loss": 0.8483, "step": 34838 }, { "epoch": 0.03, "learning_rate": 4.9851235242659825e-05, "loss": 1.0129, "step": 34839 }, { "epoch": 0.03, "learning_rate": 4.985122668632992e-05, "loss": 0.9348, "step": 34840 }, { "epoch": 0.03, "learning_rate": 4.98512181297547e-05, "loss": 1.0997, "step": 34841 }, { "epoch": 0.03, "learning_rate": 4.985120957293414e-05, "loss": 1.149, "step": 34842 }, { "epoch": 0.03, "learning_rate": 4.985120101586828e-05, "loss": 0.9304, "step": 34843 }, { "epoch": 0.03, "learning_rate": 4.985119245855708e-05, "loss": 1.1397, "step": 34844 }, { "epoch": 0.03, "learning_rate": 4.9851183901000576e-05, "loss": 0.8995, "step": 34845 }, { "epoch": 0.03, "learning_rate": 4.985117534319874e-05, "loss": 0.8667, "step": 34846 }, { "epoch": 0.03, "learning_rate": 4.9851166785151595e-05, "loss": 0.9021, "step": 34847 }, { "epoch": 0.03, "learning_rate": 4.9851158226859126e-05, "loss": 1.204, "step": 34848 }, { "epoch": 0.03, "learning_rate": 4.985114966832133e-05, "loss": 1.2776, "step": 34849 }, { "epoch": 0.03, "learning_rate": 4.9851141109538226e-05, "loss": 0.9706, "step": 34850 }, { "epoch": 0.03, "learning_rate": 4.9851132550509794e-05, "loss": 1.1576, "step": 34851 }, { "epoch": 0.03, "learning_rate": 4.9851123991236035e-05, "loss": 1.0305, "step": 34852 }, { "epoch": 0.03, "learning_rate": 4.985111543171697e-05, "loss": 0.9365, "step": 34853 }, { "epoch": 0.03, "learning_rate": 4.985110687195258e-05, "loss": 1.1005, "step": 34854 }, { "epoch": 0.03, "learning_rate": 4.985109831194287e-05, "loss": 1.0361, "step": 34855 }, { "epoch": 0.03, "learning_rate": 4.985108975168783e-05, "loss": 1.033, "step": 34856 }, { "epoch": 0.03, "learning_rate": 4.985108119118749e-05, "loss": 1.1448, "step": 34857 }, { "epoch": 0.03, "learning_rate": 4.985107263044182e-05, "loss": 1.001, "step": 34858 }, { "epoch": 0.03, "learning_rate": 4.985106406945082e-05, "loss": 0.8395, "step": 34859 }, { "epoch": 0.03, "learning_rate": 4.985105550821452e-05, "loss": 0.8597, "step": 34860 }, { "epoch": 0.03, "learning_rate": 4.985104694673289e-05, "loss": 0.9253, "step": 34861 }, { "epoch": 0.03, "learning_rate": 4.985103838500595e-05, "loss": 0.9248, "step": 34862 }, { "epoch": 0.03, "learning_rate": 4.9851029823033685e-05, "loss": 0.9853, "step": 34863 }, { "epoch": 0.03, "learning_rate": 4.98510212608161e-05, "loss": 0.9644, "step": 34864 }, { "epoch": 0.03, "learning_rate": 4.98510126983532e-05, "loss": 1.1919, "step": 34865 }, { "epoch": 0.03, "learning_rate": 4.985100413564497e-05, "loss": 1.2099, "step": 34866 }, { "epoch": 0.03, "learning_rate": 4.985099557269144e-05, "loss": 0.977, "step": 34867 }, { "epoch": 0.03, "learning_rate": 4.9850987009492575e-05, "loss": 0.8892, "step": 34868 }, { "epoch": 0.03, "learning_rate": 4.9850978446048405e-05, "loss": 1.0147, "step": 34869 }, { "epoch": 0.03, "learning_rate": 4.985096988235891e-05, "loss": 1.0968, "step": 34870 }, { "epoch": 0.03, "learning_rate": 4.9850961318424094e-05, "loss": 0.6623, "step": 34871 }, { "epoch": 0.03, "learning_rate": 4.985095275424396e-05, "loss": 0.5332, "step": 34872 }, { "epoch": 0.03, "learning_rate": 4.985094418981852e-05, "loss": 1.1229, "step": 34873 }, { "epoch": 0.03, "learning_rate": 4.9850935625147744e-05, "loss": 1.1852, "step": 34874 }, { "epoch": 0.03, "learning_rate": 4.9850927060231654e-05, "loss": 1.0084, "step": 34875 }, { "epoch": 0.03, "learning_rate": 4.9850918495070256e-05, "loss": 1.295, "step": 34876 }, { "epoch": 0.03, "learning_rate": 4.985090992966354e-05, "loss": 1.4525, "step": 34877 }, { "epoch": 0.03, "learning_rate": 4.985090136401149e-05, "loss": 1.0423, "step": 34878 }, { "epoch": 0.03, "learning_rate": 4.985089279811414e-05, "loss": 1.1942, "step": 34879 }, { "epoch": 0.03, "learning_rate": 4.985088423197146e-05, "loss": 0.7309, "step": 34880 }, { "epoch": 0.03, "learning_rate": 4.985087566558347e-05, "loss": 0.8905, "step": 34881 }, { "epoch": 0.03, "learning_rate": 4.985086709895016e-05, "loss": 0.9756, "step": 34882 }, { "epoch": 0.03, "learning_rate": 4.985085853207154e-05, "loss": 0.5623, "step": 34883 }, { "epoch": 0.03, "learning_rate": 4.985084996494759e-05, "loss": 0.432, "step": 34884 }, { "epoch": 0.03, "learning_rate": 4.985084139757833e-05, "loss": 1.103, "step": 34885 }, { "epoch": 0.03, "learning_rate": 4.985083282996376e-05, "loss": 1.1797, "step": 34886 }, { "epoch": 0.03, "learning_rate": 4.9850824262103855e-05, "loss": 1.31, "step": 34887 }, { "epoch": 0.03, "learning_rate": 4.9850815693998646e-05, "loss": 0.8807, "step": 34888 }, { "epoch": 0.03, "learning_rate": 4.985080712564812e-05, "loss": 0.9781, "step": 34889 }, { "epoch": 0.03, "learning_rate": 4.9850798557052277e-05, "loss": 1.2669, "step": 34890 }, { "epoch": 0.03, "learning_rate": 4.985078998821111e-05, "loss": 1.0026, "step": 34891 }, { "epoch": 0.03, "learning_rate": 4.985078141912464e-05, "loss": 1.3281, "step": 34892 }, { "epoch": 0.03, "learning_rate": 4.9850772849792835e-05, "loss": 1.225, "step": 34893 }, { "epoch": 0.03, "learning_rate": 4.9850764280215726e-05, "loss": 0.3505, "step": 34894 }, { "epoch": 0.03, "learning_rate": 4.98507557103933e-05, "loss": 0.9193, "step": 34895 }, { "epoch": 0.03, "learning_rate": 4.985074714032556e-05, "loss": 1.1438, "step": 34896 }, { "epoch": 0.03, "learning_rate": 4.98507385700125e-05, "loss": 1.195, "step": 34897 }, { "epoch": 0.03, "learning_rate": 4.985072999945413e-05, "loss": 1.3047, "step": 34898 }, { "epoch": 0.03, "learning_rate": 4.985072142865043e-05, "loss": 1.1137, "step": 34899 }, { "epoch": 0.03, "learning_rate": 4.9850712857601425e-05, "loss": 1.1797, "step": 34900 }, { "epoch": 0.03, "learning_rate": 4.98507042863071e-05, "loss": 1.086, "step": 34901 }, { "epoch": 0.03, "learning_rate": 4.985069571476746e-05, "loss": 0.9847, "step": 34902 }, { "epoch": 0.03, "learning_rate": 4.9850687142982514e-05, "loss": 0.6972, "step": 34903 }, { "epoch": 0.03, "learning_rate": 4.9850678570952235e-05, "loss": 0.8975, "step": 34904 }, { "epoch": 0.03, "learning_rate": 4.9850669998676655e-05, "loss": 0.8192, "step": 34905 }, { "epoch": 0.03, "learning_rate": 4.9850661426155754e-05, "loss": 1.0229, "step": 34906 }, { "epoch": 0.03, "learning_rate": 4.985065285338953e-05, "loss": 0.7504, "step": 34907 }, { "epoch": 0.03, "learning_rate": 4.9850644280378e-05, "loss": 0.4522, "step": 34908 }, { "epoch": 0.03, "learning_rate": 4.985063570712116e-05, "loss": 0.4744, "step": 34909 }, { "epoch": 0.03, "learning_rate": 4.9850627133619e-05, "loss": 0.6788, "step": 34910 }, { "epoch": 0.03, "learning_rate": 4.985061855987152e-05, "loss": 0.8266, "step": 34911 }, { "epoch": 0.03, "learning_rate": 4.985060998587873e-05, "loss": 0.6279, "step": 34912 }, { "epoch": 0.03, "learning_rate": 4.985060141164062e-05, "loss": 0.6941, "step": 34913 }, { "epoch": 0.03, "learning_rate": 4.9850592837157205e-05, "loss": 1.2494, "step": 34914 }, { "epoch": 0.03, "learning_rate": 4.985058426242847e-05, "loss": 1.0798, "step": 34915 }, { "epoch": 0.03, "learning_rate": 4.985057568745441e-05, "loss": 1.0183, "step": 34916 }, { "epoch": 0.03, "learning_rate": 4.985056711223506e-05, "loss": 1.0499, "step": 34917 }, { "epoch": 0.03, "learning_rate": 4.9850558536770374e-05, "loss": 1.0155, "step": 34918 }, { "epoch": 0.03, "learning_rate": 4.985054996106038e-05, "loss": 1.4859, "step": 34919 }, { "epoch": 0.03, "learning_rate": 4.985054138510507e-05, "loss": 1.0159, "step": 34920 }, { "epoch": 0.03, "learning_rate": 4.985053280890446e-05, "loss": 1.2786, "step": 34921 }, { "epoch": 0.03, "learning_rate": 4.9850524232458515e-05, "loss": 1.3723, "step": 34922 }, { "epoch": 0.03, "learning_rate": 4.985051565576727e-05, "loss": 1.004, "step": 34923 }, { "epoch": 0.03, "learning_rate": 4.985050707883071e-05, "loss": 0.7014, "step": 34924 }, { "epoch": 0.03, "learning_rate": 4.9850498501648833e-05, "loss": 0.3208, "step": 34925 }, { "epoch": 0.03, "learning_rate": 4.985048992422165e-05, "loss": 0.676, "step": 34926 }, { "epoch": 0.03, "learning_rate": 4.985048134654914e-05, "loss": 0.9864, "step": 34927 }, { "epoch": 0.03, "learning_rate": 4.9850472768631323e-05, "loss": 1.1744, "step": 34928 }, { "epoch": 0.03, "learning_rate": 4.9850464190468194e-05, "loss": 0.8535, "step": 34929 }, { "epoch": 0.03, "learning_rate": 4.985045561205975e-05, "loss": 0.8951, "step": 34930 }, { "epoch": 0.03, "learning_rate": 4.985044703340599e-05, "loss": 1.0114, "step": 34931 }, { "epoch": 0.03, "learning_rate": 4.985043845450692e-05, "loss": 1.0993, "step": 34932 }, { "epoch": 0.03, "learning_rate": 4.985042987536254e-05, "loss": 1.1287, "step": 34933 }, { "epoch": 0.03, "learning_rate": 4.9850421295972845e-05, "loss": 1.3724, "step": 34934 }, { "epoch": 0.03, "learning_rate": 4.9850412716337836e-05, "loss": 1.2491, "step": 34935 }, { "epoch": 0.03, "learning_rate": 4.985040413645751e-05, "loss": 0.9716, "step": 34936 }, { "epoch": 0.03, "learning_rate": 4.985039555633188e-05, "loss": 1.2134, "step": 34937 }, { "epoch": 0.03, "learning_rate": 4.985038697596093e-05, "loss": 1.1063, "step": 34938 }, { "epoch": 0.03, "learning_rate": 4.985037839534468e-05, "loss": 1.1697, "step": 34939 }, { "epoch": 0.03, "learning_rate": 4.98503698144831e-05, "loss": 1.1417, "step": 34940 }, { "epoch": 0.03, "learning_rate": 4.985036123337622e-05, "loss": 0.7838, "step": 34941 }, { "epoch": 0.03, "learning_rate": 4.9850352652024016e-05, "loss": 1.0432, "step": 34942 }, { "epoch": 0.03, "learning_rate": 4.9850344070426514e-05, "loss": 1.0318, "step": 34943 }, { "epoch": 0.03, "learning_rate": 4.985033548858369e-05, "loss": 0.9542, "step": 34944 }, { "epoch": 0.03, "learning_rate": 4.985032690649556e-05, "loss": 1.0233, "step": 34945 }, { "epoch": 0.03, "learning_rate": 4.985031832416211e-05, "loss": 1.1256, "step": 34946 }, { "epoch": 0.03, "learning_rate": 4.985030974158335e-05, "loss": 1.2297, "step": 34947 }, { "epoch": 0.03, "learning_rate": 4.9850301158759286e-05, "loss": 1.1273, "step": 34948 }, { "epoch": 0.03, "learning_rate": 4.9850292575689904e-05, "loss": 0.9187, "step": 34949 }, { "epoch": 0.03, "learning_rate": 4.9850283992375215e-05, "loss": 0.8696, "step": 34950 }, { "epoch": 0.03, "learning_rate": 4.9850275408815205e-05, "loss": 0.8027, "step": 34951 }, { "epoch": 0.03, "learning_rate": 4.9850266825009894e-05, "loss": 1.0676, "step": 34952 }, { "epoch": 0.03, "learning_rate": 4.985025824095927e-05, "loss": 1.3017, "step": 34953 }, { "epoch": 0.03, "learning_rate": 4.985024965666333e-05, "loss": 1.2091, "step": 34954 }, { "epoch": 0.03, "learning_rate": 4.985024107212208e-05, "loss": 1.1124, "step": 34955 }, { "epoch": 0.03, "learning_rate": 4.985023248733553e-05, "loss": 1.156, "step": 34956 }, { "epoch": 0.03, "learning_rate": 4.985022390230365e-05, "loss": 1.3985, "step": 34957 }, { "epoch": 0.03, "learning_rate": 4.985021531702647e-05, "loss": 1.1125, "step": 34958 }, { "epoch": 0.03, "learning_rate": 4.985020673150398e-05, "loss": 0.951, "step": 34959 }, { "epoch": 0.03, "learning_rate": 4.985019814573617e-05, "loss": 1.0205, "step": 34960 }, { "epoch": 0.03, "learning_rate": 4.985018955972306e-05, "loss": 1.1025, "step": 34961 }, { "epoch": 0.03, "learning_rate": 4.9850180973464634e-05, "loss": 0.8249, "step": 34962 }, { "epoch": 0.03, "learning_rate": 4.98501723869609e-05, "loss": 0.8069, "step": 34963 }, { "epoch": 0.03, "learning_rate": 4.985016380021185e-05, "loss": 1.0272, "step": 34964 }, { "epoch": 0.03, "learning_rate": 4.9850155213217494e-05, "loss": 1.4139, "step": 34965 }, { "epoch": 0.03, "learning_rate": 4.985014662597783e-05, "loss": 0.816, "step": 34966 }, { "epoch": 0.03, "learning_rate": 4.985013803849285e-05, "loss": 0.8349, "step": 34967 }, { "epoch": 0.03, "learning_rate": 4.985012945076257e-05, "loss": 1.1302, "step": 34968 }, { "epoch": 0.03, "learning_rate": 4.985012086278697e-05, "loss": 1.1969, "step": 34969 }, { "epoch": 0.03, "learning_rate": 4.985011227456606e-05, "loss": 0.9089, "step": 34970 }, { "epoch": 0.03, "learning_rate": 4.9850103686099846e-05, "loss": 1.3359, "step": 34971 }, { "epoch": 0.03, "learning_rate": 4.985009509738832e-05, "loss": 1.2993, "step": 34972 }, { "epoch": 0.03, "learning_rate": 4.985008650843148e-05, "loss": 1.0437, "step": 34973 }, { "epoch": 0.03, "learning_rate": 4.985007791922934e-05, "loss": 1.0615, "step": 34974 }, { "epoch": 0.03, "learning_rate": 4.9850069329781884e-05, "loss": 1.4714, "step": 34975 }, { "epoch": 0.03, "learning_rate": 4.9850060740089125e-05, "loss": 1.5359, "step": 34976 }, { "epoch": 0.03, "learning_rate": 4.985005215015105e-05, "loss": 1.4002, "step": 34977 }, { "epoch": 0.03, "learning_rate": 4.9850043559967664e-05, "loss": 1.4875, "step": 34978 }, { "epoch": 0.03, "learning_rate": 4.9850034969538975e-05, "loss": 1.4917, "step": 34979 }, { "epoch": 0.03, "learning_rate": 4.985002637886497e-05, "loss": 0.9054, "step": 34980 }, { "epoch": 0.03, "learning_rate": 4.9850017787945655e-05, "loss": 0.1122, "step": 34981 }, { "epoch": 0.03, "learning_rate": 4.9850009196781044e-05, "loss": 0.0933, "step": 34982 }, { "epoch": 0.03, "learning_rate": 4.985000060537112e-05, "loss": 0.1112, "step": 34983 }, { "epoch": 0.03, "learning_rate": 4.984999201371587e-05, "loss": 0.0775, "step": 34984 }, { "epoch": 0.03, "learning_rate": 4.984998342181533e-05, "loss": 0.0788, "step": 34985 }, { "epoch": 0.03, "learning_rate": 4.984997482966948e-05, "loss": 0.0343, "step": 34986 }, { "epoch": 0.03, "learning_rate": 4.9849966237278315e-05, "loss": 0.694, "step": 34987 }, { "epoch": 0.03, "learning_rate": 4.9849957644641846e-05, "loss": 0.8573, "step": 34988 }, { "epoch": 0.03, "learning_rate": 4.984994905176007e-05, "loss": 0.9291, "step": 34989 }, { "epoch": 0.03, "learning_rate": 4.9849940458632984e-05, "loss": 1.3698, "step": 34990 }, { "epoch": 0.03, "learning_rate": 4.9849931865260585e-05, "loss": 1.006, "step": 34991 }, { "epoch": 0.03, "learning_rate": 4.9849923271642885e-05, "loss": 1.2665, "step": 34992 }, { "epoch": 0.03, "learning_rate": 4.984991467777987e-05, "loss": 1.0887, "step": 34993 }, { "epoch": 0.03, "learning_rate": 4.984990608367155e-05, "loss": 1.1577, "step": 34994 }, { "epoch": 0.03, "learning_rate": 4.984989748931792e-05, "loss": 0.5345, "step": 34995 }, { "epoch": 0.03, "learning_rate": 4.984988889471899e-05, "loss": 0.9198, "step": 34996 }, { "epoch": 0.03, "learning_rate": 4.9849880299874746e-05, "loss": 0.6681, "step": 34997 }, { "epoch": 0.03, "learning_rate": 4.9849871704785195e-05, "loss": 1.0401, "step": 34998 }, { "epoch": 0.03, "learning_rate": 4.984986310945034e-05, "loss": 0.9903, "step": 34999 }, { "epoch": 0.04, "learning_rate": 4.9849854513870175e-05, "loss": 1.0312, "step": 35000 }, { "epoch": 0.04, "eval_loss": 1.0894055366516113, "eval_runtime": 125.4337, "eval_samples_per_second": 11.042, "eval_steps_per_second": 5.525, "step": 35000 }, { "epoch": 0.04, "learning_rate": 4.98498459180447e-05, "loss": 1.0623, "step": 35001 }, { "epoch": 0.04, "learning_rate": 4.9849837321973926e-05, "loss": 0.8734, "step": 35002 }, { "epoch": 0.04, "learning_rate": 4.984982872565783e-05, "loss": 1.1336, "step": 35003 }, { "epoch": 0.04, "learning_rate": 4.984982012909644e-05, "loss": 1.0945, "step": 35004 }, { "epoch": 0.04, "learning_rate": 4.984981153228974e-05, "loss": 1.1202, "step": 35005 }, { "epoch": 0.04, "learning_rate": 4.984980293523773e-05, "loss": 1.0757, "step": 35006 }, { "epoch": 0.04, "learning_rate": 4.9849794337940425e-05, "loss": 1.118, "step": 35007 }, { "epoch": 0.04, "learning_rate": 4.9849785740397805e-05, "loss": 1.0745, "step": 35008 }, { "epoch": 0.04, "learning_rate": 4.984977714260988e-05, "loss": 0.6937, "step": 35009 }, { "epoch": 0.04, "learning_rate": 4.9849768544576636e-05, "loss": 1.4132, "step": 35010 }, { "epoch": 0.04, "learning_rate": 4.98497599462981e-05, "loss": 0.9892, "step": 35011 }, { "epoch": 0.04, "learning_rate": 4.9849751347774257e-05, "loss": 0.8812, "step": 35012 }, { "epoch": 0.04, "learning_rate": 4.98497427490051e-05, "loss": 0.824, "step": 35013 }, { "epoch": 0.04, "learning_rate": 4.984973414999064e-05, "loss": 0.4144, "step": 35014 }, { "epoch": 0.04, "learning_rate": 4.9849725550730875e-05, "loss": 0.9108, "step": 35015 }, { "epoch": 0.04, "learning_rate": 4.98497169512258e-05, "loss": 1.2984, "step": 35016 }, { "epoch": 0.04, "learning_rate": 4.984970835147543e-05, "loss": 0.9801, "step": 35017 }, { "epoch": 0.04, "learning_rate": 4.984969975147974e-05, "loss": 0.7796, "step": 35018 }, { "epoch": 0.04, "learning_rate": 4.9849691151238756e-05, "loss": 0.9012, "step": 35019 }, { "epoch": 0.04, "learning_rate": 4.9849682550752466e-05, "loss": 0.9529, "step": 35020 }, { "epoch": 0.04, "learning_rate": 4.984967395002086e-05, "loss": 1.3308, "step": 35021 }, { "epoch": 0.04, "learning_rate": 4.984966534904395e-05, "loss": 1.3805, "step": 35022 }, { "epoch": 0.04, "learning_rate": 4.9849656747821744e-05, "loss": 1.3476, "step": 35023 }, { "epoch": 0.04, "learning_rate": 4.9849648146354224e-05, "loss": 1.1513, "step": 35024 }, { "epoch": 0.04, "learning_rate": 4.9849639544641404e-05, "loss": 1.1367, "step": 35025 }, { "epoch": 0.04, "learning_rate": 4.9849630942683275e-05, "loss": 0.9496, "step": 35026 }, { "epoch": 0.04, "learning_rate": 4.9849622340479846e-05, "loss": 1.1523, "step": 35027 }, { "epoch": 0.04, "learning_rate": 4.984961373803111e-05, "loss": 0.6997, "step": 35028 }, { "epoch": 0.04, "learning_rate": 4.9849605135337066e-05, "loss": 1.188, "step": 35029 }, { "epoch": 0.04, "learning_rate": 4.984959653239772e-05, "loss": 0.5121, "step": 35030 }, { "epoch": 0.04, "learning_rate": 4.984958792921307e-05, "loss": 0.6967, "step": 35031 }, { "epoch": 0.04, "learning_rate": 4.9849579325783116e-05, "loss": 0.9343, "step": 35032 }, { "epoch": 0.04, "learning_rate": 4.9849570722107855e-05, "loss": 0.9797, "step": 35033 }, { "epoch": 0.04, "learning_rate": 4.984956211818729e-05, "loss": 1.1391, "step": 35034 }, { "epoch": 0.04, "learning_rate": 4.984955351402142e-05, "loss": 0.8936, "step": 35035 }, { "epoch": 0.04, "learning_rate": 4.984954490961025e-05, "loss": 1.013, "step": 35036 }, { "epoch": 0.04, "learning_rate": 4.984953630495377e-05, "loss": 0.7365, "step": 35037 }, { "epoch": 0.04, "learning_rate": 4.984952770005199e-05, "loss": 1.008, "step": 35038 }, { "epoch": 0.04, "learning_rate": 4.9849519094904895e-05, "loss": 1.2795, "step": 35039 }, { "epoch": 0.04, "learning_rate": 4.984951048951251e-05, "loss": 0.9438, "step": 35040 }, { "epoch": 0.04, "learning_rate": 4.9849501883874815e-05, "loss": 0.9881, "step": 35041 }, { "epoch": 0.04, "learning_rate": 4.9849493277991814e-05, "loss": 1.2154, "step": 35042 }, { "epoch": 0.04, "learning_rate": 4.984948467186351e-05, "loss": 1.1271, "step": 35043 }, { "epoch": 0.04, "learning_rate": 4.984947606548991e-05, "loss": 0.5867, "step": 35044 }, { "epoch": 0.04, "learning_rate": 4.9849467458871e-05, "loss": 0.3893, "step": 35045 }, { "epoch": 0.04, "learning_rate": 4.984945885200679e-05, "loss": 0.3878, "step": 35046 }, { "epoch": 0.04, "learning_rate": 4.984945024489727e-05, "loss": 0.7078, "step": 35047 }, { "epoch": 0.04, "learning_rate": 4.984944163754245e-05, "loss": 0.9788, "step": 35048 }, { "epoch": 0.04, "learning_rate": 4.9849433029942324e-05, "loss": 1.2769, "step": 35049 }, { "epoch": 0.04, "learning_rate": 4.9849424422096904e-05, "loss": 0.7183, "step": 35050 }, { "epoch": 0.04, "learning_rate": 4.9849415814006176e-05, "loss": 1.2445, "step": 35051 }, { "epoch": 0.04, "learning_rate": 4.984940720567015e-05, "loss": 0.9923, "step": 35052 }, { "epoch": 0.04, "learning_rate": 4.984939859708882e-05, "loss": 1.0441, "step": 35053 }, { "epoch": 0.04, "learning_rate": 4.984938998826217e-05, "loss": 1.1863, "step": 35054 }, { "epoch": 0.04, "learning_rate": 4.984938137919023e-05, "loss": 0.8559, "step": 35055 }, { "epoch": 0.04, "learning_rate": 4.984937276987299e-05, "loss": 0.7301, "step": 35056 }, { "epoch": 0.04, "learning_rate": 4.9849364160310444e-05, "loss": 0.6529, "step": 35057 }, { "epoch": 0.04, "learning_rate": 4.98493555505026e-05, "loss": 1.0929, "step": 35058 }, { "epoch": 0.04, "learning_rate": 4.984934694044945e-05, "loss": 1.142, "step": 35059 }, { "epoch": 0.04, "learning_rate": 4.9849338330150994e-05, "loss": 1.2402, "step": 35060 }, { "epoch": 0.04, "learning_rate": 4.984932971960724e-05, "loss": 0.9388, "step": 35061 }, { "epoch": 0.04, "learning_rate": 4.984932110881819e-05, "loss": 1.0686, "step": 35062 }, { "epoch": 0.04, "learning_rate": 4.984931249778383e-05, "loss": 0.919, "step": 35063 }, { "epoch": 0.04, "learning_rate": 4.9849303886504166e-05, "loss": 0.891, "step": 35064 }, { "epoch": 0.04, "learning_rate": 4.98492952749792e-05, "loss": 1.0403, "step": 35065 }, { "epoch": 0.04, "learning_rate": 4.984928666320894e-05, "loss": 1.0117, "step": 35066 }, { "epoch": 0.04, "learning_rate": 4.9849278051193374e-05, "loss": 1.5156, "step": 35067 }, { "epoch": 0.04, "learning_rate": 4.984926943893251e-05, "loss": 1.3507, "step": 35068 }, { "epoch": 0.04, "learning_rate": 4.984926082642634e-05, "loss": 1.0346, "step": 35069 }, { "epoch": 0.04, "learning_rate": 4.984925221367487e-05, "loss": 0.9328, "step": 35070 }, { "epoch": 0.04, "learning_rate": 4.98492436006781e-05, "loss": 0.9392, "step": 35071 }, { "epoch": 0.04, "learning_rate": 4.984923498743602e-05, "loss": 0.8452, "step": 35072 }, { "epoch": 0.04, "learning_rate": 4.9849226373948656e-05, "loss": 0.4713, "step": 35073 }, { "epoch": 0.04, "learning_rate": 4.984921776021598e-05, "loss": 0.9736, "step": 35074 }, { "epoch": 0.04, "learning_rate": 4.9849209146238e-05, "loss": 1.042, "step": 35075 }, { "epoch": 0.04, "learning_rate": 4.984920053201472e-05, "loss": 0.4969, "step": 35076 }, { "epoch": 0.04, "learning_rate": 4.984919191754614e-05, "loss": 0.9774, "step": 35077 }, { "epoch": 0.04, "learning_rate": 4.984918330283227e-05, "loss": 1.1382, "step": 35078 }, { "epoch": 0.04, "learning_rate": 4.984917468787309e-05, "loss": 1.1217, "step": 35079 }, { "epoch": 0.04, "learning_rate": 4.984916607266861e-05, "loss": 1.0447, "step": 35080 }, { "epoch": 0.04, "learning_rate": 4.984915745721883e-05, "loss": 1.1908, "step": 35081 }, { "epoch": 0.04, "learning_rate": 4.9849148841523745e-05, "loss": 1.1169, "step": 35082 }, { "epoch": 0.04, "learning_rate": 4.9849140225583366e-05, "loss": 0.9859, "step": 35083 }, { "epoch": 0.04, "learning_rate": 4.984913160939768e-05, "loss": 1.1006, "step": 35084 }, { "epoch": 0.04, "learning_rate": 4.98491229929667e-05, "loss": 0.9824, "step": 35085 }, { "epoch": 0.04, "learning_rate": 4.984911437629042e-05, "loss": 1.0418, "step": 35086 }, { "epoch": 0.04, "learning_rate": 4.984910575936884e-05, "loss": 1.4116, "step": 35087 }, { "epoch": 0.04, "learning_rate": 4.984909714220196e-05, "loss": 1.0648, "step": 35088 }, { "epoch": 0.04, "learning_rate": 4.9849088524789777e-05, "loss": 0.812, "step": 35089 }, { "epoch": 0.04, "learning_rate": 4.984907990713229e-05, "loss": 0.5262, "step": 35090 }, { "epoch": 0.04, "learning_rate": 4.9849071289229514e-05, "loss": 0.6011, "step": 35091 }, { "epoch": 0.04, "learning_rate": 4.9849062671081435e-05, "loss": 0.8062, "step": 35092 }, { "epoch": 0.04, "learning_rate": 4.9849054052688056e-05, "loss": 0.8015, "step": 35093 }, { "epoch": 0.04, "learning_rate": 4.984904543404938e-05, "loss": 1.001, "step": 35094 }, { "epoch": 0.04, "learning_rate": 4.9849036815165394e-05, "loss": 0.5645, "step": 35095 }, { "epoch": 0.04, "learning_rate": 4.984902819603612e-05, "loss": 0.8296, "step": 35096 }, { "epoch": 0.04, "learning_rate": 4.9849019576661544e-05, "loss": 0.7638, "step": 35097 }, { "epoch": 0.04, "learning_rate": 4.984901095704167e-05, "loss": 0.5477, "step": 35098 }, { "epoch": 0.04, "learning_rate": 4.984900233717649e-05, "loss": 1.4543, "step": 35099 }, { "epoch": 0.04, "learning_rate": 4.984899371706602e-05, "loss": 1.0837, "step": 35100 }, { "epoch": 0.04, "learning_rate": 4.9848985096710247e-05, "loss": 0.864, "step": 35101 }, { "epoch": 0.04, "learning_rate": 4.9848976476109174e-05, "loss": 1.4664, "step": 35102 }, { "epoch": 0.04, "learning_rate": 4.984896785526281e-05, "loss": 1.0748, "step": 35103 }, { "epoch": 0.04, "learning_rate": 4.984895923417114e-05, "loss": 0.773, "step": 35104 }, { "epoch": 0.04, "learning_rate": 4.984895061283417e-05, "loss": 0.8841, "step": 35105 }, { "epoch": 0.04, "learning_rate": 4.984894199125191e-05, "loss": 1.0091, "step": 35106 }, { "epoch": 0.04, "learning_rate": 4.9848933369424347e-05, "loss": 1.4164, "step": 35107 }, { "epoch": 0.04, "learning_rate": 4.9848924747351476e-05, "loss": 1.1177, "step": 35108 }, { "epoch": 0.04, "learning_rate": 4.984891612503332e-05, "loss": 0.9043, "step": 35109 }, { "epoch": 0.04, "learning_rate": 4.984890750246987e-05, "loss": 1.0209, "step": 35110 }, { "epoch": 0.04, "learning_rate": 4.984889887966111e-05, "loss": 0.6882, "step": 35111 }, { "epoch": 0.04, "learning_rate": 4.9848890256607055e-05, "loss": 0.9862, "step": 35112 }, { "epoch": 0.04, "learning_rate": 4.98488816333077e-05, "loss": 0.9784, "step": 35113 }, { "epoch": 0.04, "learning_rate": 4.9848873009763053e-05, "loss": 1.0302, "step": 35114 }, { "epoch": 0.04, "learning_rate": 4.9848864385973105e-05, "loss": 0.7577, "step": 35115 }, { "epoch": 0.04, "learning_rate": 4.984885576193786e-05, "loss": 0.6143, "step": 35116 }, { "epoch": 0.04, "learning_rate": 4.9848847137657326e-05, "loss": 0.8948, "step": 35117 }, { "epoch": 0.04, "learning_rate": 4.984883851313148e-05, "loss": 1.3249, "step": 35118 }, { "epoch": 0.04, "learning_rate": 4.984882988836035e-05, "loss": 1.8569, "step": 35119 }, { "epoch": 0.04, "learning_rate": 4.984882126334391e-05, "loss": 1.5735, "step": 35120 }, { "epoch": 0.04, "learning_rate": 4.9848812638082187e-05, "loss": 1.6746, "step": 35121 }, { "epoch": 0.04, "learning_rate": 4.984880401257515e-05, "loss": 0.9175, "step": 35122 }, { "epoch": 0.04, "learning_rate": 4.984879538682283e-05, "loss": 1.8466, "step": 35123 }, { "epoch": 0.04, "learning_rate": 4.984878676082521e-05, "loss": 1.6194, "step": 35124 }, { "epoch": 0.04, "learning_rate": 4.984877813458229e-05, "loss": 1.7118, "step": 35125 }, { "epoch": 0.04, "learning_rate": 4.984876950809408e-05, "loss": 1.3658, "step": 35126 }, { "epoch": 0.04, "learning_rate": 4.984876088136057e-05, "loss": 1.0392, "step": 35127 }, { "epoch": 0.04, "learning_rate": 4.984875225438176e-05, "loss": 0.9247, "step": 35128 }, { "epoch": 0.04, "learning_rate": 4.9848743627157655e-05, "loss": 0.884, "step": 35129 }, { "epoch": 0.04, "learning_rate": 4.984873499968825e-05, "loss": 0.4157, "step": 35130 }, { "epoch": 0.04, "learning_rate": 4.984872637197356e-05, "loss": 0.789, "step": 35131 }, { "epoch": 0.04, "learning_rate": 4.9848717744013564e-05, "loss": 1.0047, "step": 35132 }, { "epoch": 0.04, "learning_rate": 4.9848709115808276e-05, "loss": 0.9043, "step": 35133 }, { "epoch": 0.04, "learning_rate": 4.984870048735769e-05, "loss": 0.9412, "step": 35134 }, { "epoch": 0.04, "learning_rate": 4.984869185866181e-05, "loss": 0.9589, "step": 35135 }, { "epoch": 0.04, "learning_rate": 4.984868322972063e-05, "loss": 0.9689, "step": 35136 }, { "epoch": 0.04, "learning_rate": 4.984867460053416e-05, "loss": 1.1717, "step": 35137 }, { "epoch": 0.04, "learning_rate": 4.984866597110239e-05, "loss": 1.0968, "step": 35138 }, { "epoch": 0.04, "learning_rate": 4.984865734142533e-05, "loss": 0.9518, "step": 35139 }, { "epoch": 0.04, "learning_rate": 4.984864871150297e-05, "loss": 0.5801, "step": 35140 }, { "epoch": 0.04, "learning_rate": 4.984864008133532e-05, "loss": 1.1701, "step": 35141 }, { "epoch": 0.04, "learning_rate": 4.984863145092237e-05, "loss": 1.1502, "step": 35142 }, { "epoch": 0.04, "learning_rate": 4.984862282026413e-05, "loss": 1.0741, "step": 35143 }, { "epoch": 0.04, "learning_rate": 4.984861418936058e-05, "loss": 1.0057, "step": 35144 }, { "epoch": 0.04, "learning_rate": 4.984860555821175e-05, "loss": 0.4241, "step": 35145 }, { "epoch": 0.04, "learning_rate": 4.9848596926817615e-05, "loss": 0.8475, "step": 35146 }, { "epoch": 0.04, "learning_rate": 4.984858829517819e-05, "loss": 0.952, "step": 35147 }, { "epoch": 0.04, "learning_rate": 4.984857966329347e-05, "loss": 1.0483, "step": 35148 }, { "epoch": 0.04, "learning_rate": 4.984857103116346e-05, "loss": 0.9244, "step": 35149 }, { "epoch": 0.04, "learning_rate": 4.9848562398788145e-05, "loss": 1.0326, "step": 35150 }, { "epoch": 0.04, "learning_rate": 4.9848553766167546e-05, "loss": 0.9544, "step": 35151 }, { "epoch": 0.04, "learning_rate": 4.984854513330165e-05, "loss": 1.0788, "step": 35152 }, { "epoch": 0.04, "learning_rate": 4.984853650019046e-05, "loss": 1.0107, "step": 35153 }, { "epoch": 0.04, "learning_rate": 4.984852786683397e-05, "loss": 1.0957, "step": 35154 }, { "epoch": 0.04, "learning_rate": 4.984851923323218e-05, "loss": 1.2412, "step": 35155 }, { "epoch": 0.04, "learning_rate": 4.9848510599385114e-05, "loss": 0.9693, "step": 35156 }, { "epoch": 0.04, "learning_rate": 4.9848501965292745e-05, "loss": 1.2893, "step": 35157 }, { "epoch": 0.04, "learning_rate": 4.984849333095508e-05, "loss": 1.0818, "step": 35158 }, { "epoch": 0.04, "learning_rate": 4.9848484696372124e-05, "loss": 1.0476, "step": 35159 }, { "epoch": 0.04, "learning_rate": 4.984847606154387e-05, "loss": 1.0029, "step": 35160 }, { "epoch": 0.04, "learning_rate": 4.984846742647033e-05, "loss": 1.9273, "step": 35161 }, { "epoch": 0.04, "learning_rate": 4.984845879115149e-05, "loss": 1.8755, "step": 35162 }, { "epoch": 0.04, "learning_rate": 4.9848450155587354e-05, "loss": 1.5557, "step": 35163 }, { "epoch": 0.04, "learning_rate": 4.9848441519777934e-05, "loss": 1.6269, "step": 35164 }, { "epoch": 0.04, "learning_rate": 4.984843288372322e-05, "loss": 1.0303, "step": 35165 }, { "epoch": 0.04, "learning_rate": 4.9848424247423204e-05, "loss": 1.173, "step": 35166 }, { "epoch": 0.04, "learning_rate": 4.9848415610877895e-05, "loss": 1.0805, "step": 35167 }, { "epoch": 0.04, "learning_rate": 4.9848406974087305e-05, "loss": 0.8099, "step": 35168 }, { "epoch": 0.04, "learning_rate": 4.984839833705141e-05, "loss": 1.2668, "step": 35169 }, { "epoch": 0.04, "learning_rate": 4.9848389699770223e-05, "loss": 1.4652, "step": 35170 }, { "epoch": 0.04, "learning_rate": 4.984838106224375e-05, "loss": 1.0246, "step": 35171 }, { "epoch": 0.04, "learning_rate": 4.984837242447198e-05, "loss": 1.1662, "step": 35172 }, { "epoch": 0.04, "learning_rate": 4.984836378645492e-05, "loss": 1.1885, "step": 35173 }, { "epoch": 0.04, "learning_rate": 4.984835514819257e-05, "loss": 0.8481, "step": 35174 }, { "epoch": 0.04, "learning_rate": 4.984834650968492e-05, "loss": 0.9474, "step": 35175 }, { "epoch": 0.04, "learning_rate": 4.9848337870931985e-05, "loss": 0.8052, "step": 35176 }, { "epoch": 0.04, "learning_rate": 4.984832923193374e-05, "loss": 0.7693, "step": 35177 }, { "epoch": 0.04, "learning_rate": 4.984832059269022e-05, "loss": 0.8429, "step": 35178 }, { "epoch": 0.04, "learning_rate": 4.984831195320141e-05, "loss": 1.015, "step": 35179 }, { "epoch": 0.04, "learning_rate": 4.98483033134673e-05, "loss": 0.9952, "step": 35180 }, { "epoch": 0.04, "learning_rate": 4.98482946734879e-05, "loss": 0.9698, "step": 35181 }, { "epoch": 0.04, "learning_rate": 4.984828603326321e-05, "loss": 0.7457, "step": 35182 }, { "epoch": 0.04, "learning_rate": 4.9848277392793225e-05, "loss": 0.8136, "step": 35183 }, { "epoch": 0.04, "learning_rate": 4.9848268752077946e-05, "loss": 0.7728, "step": 35184 }, { "epoch": 0.04, "learning_rate": 4.9848260111117386e-05, "loss": 1.0528, "step": 35185 }, { "epoch": 0.04, "learning_rate": 4.9848251469911525e-05, "loss": 1.1624, "step": 35186 }, { "epoch": 0.04, "learning_rate": 4.984824282846037e-05, "loss": 0.9558, "step": 35187 }, { "epoch": 0.04, "learning_rate": 4.984823418676393e-05, "loss": 0.9299, "step": 35188 }, { "epoch": 0.04, "learning_rate": 4.98482255448222e-05, "loss": 0.3303, "step": 35189 }, { "epoch": 0.04, "learning_rate": 4.984821690263517e-05, "loss": 0.845, "step": 35190 }, { "epoch": 0.04, "learning_rate": 4.984820826020285e-05, "loss": 0.5868, "step": 35191 }, { "epoch": 0.04, "learning_rate": 4.984819961752525e-05, "loss": 0.9096, "step": 35192 }, { "epoch": 0.04, "learning_rate": 4.984819097460236e-05, "loss": 1.0104, "step": 35193 }, { "epoch": 0.04, "learning_rate": 4.9848182331434166e-05, "loss": 0.7899, "step": 35194 }, { "epoch": 0.04, "learning_rate": 4.9848173688020686e-05, "loss": 1.0698, "step": 35195 }, { "epoch": 0.04, "learning_rate": 4.984816504436191e-05, "loss": 0.9921, "step": 35196 }, { "epoch": 0.04, "learning_rate": 4.984815640045786e-05, "loss": 1.1188, "step": 35197 }, { "epoch": 0.04, "learning_rate": 4.9848147756308505e-05, "loss": 1.102, "step": 35198 }, { "epoch": 0.04, "learning_rate": 4.9848139111913864e-05, "loss": 1.0997, "step": 35199 }, { "epoch": 0.04, "learning_rate": 4.9848130467273935e-05, "loss": 1.0516, "step": 35200 }, { "epoch": 0.04, "learning_rate": 4.984812182238872e-05, "loss": 0.8551, "step": 35201 }, { "epoch": 0.04, "learning_rate": 4.98481131772582e-05, "loss": 0.9198, "step": 35202 }, { "epoch": 0.04, "learning_rate": 4.98481045318824e-05, "loss": 0.9318, "step": 35203 }, { "epoch": 0.04, "learning_rate": 4.98480958862613e-05, "loss": 0.9444, "step": 35204 }, { "epoch": 0.04, "learning_rate": 4.984808724039492e-05, "loss": 1.0138, "step": 35205 }, { "epoch": 0.04, "learning_rate": 4.984807859428326e-05, "loss": 0.9552, "step": 35206 }, { "epoch": 0.04, "learning_rate": 4.984806994792629e-05, "loss": 0.8876, "step": 35207 }, { "epoch": 0.04, "learning_rate": 4.984806130132404e-05, "loss": 0.8737, "step": 35208 }, { "epoch": 0.04, "learning_rate": 4.98480526544765e-05, "loss": 1.104, "step": 35209 }, { "epoch": 0.04, "learning_rate": 4.984804400738366e-05, "loss": 1.079, "step": 35210 }, { "epoch": 0.04, "learning_rate": 4.9848035360045554e-05, "loss": 0.9533, "step": 35211 }, { "epoch": 0.04, "learning_rate": 4.984802671246214e-05, "loss": 0.8287, "step": 35212 }, { "epoch": 0.04, "learning_rate": 4.984801806463344e-05, "loss": 0.82, "step": 35213 }, { "epoch": 0.04, "learning_rate": 4.984800941655945e-05, "loss": 0.9085, "step": 35214 }, { "epoch": 0.04, "learning_rate": 4.9848000768240175e-05, "loss": 1.2352, "step": 35215 }, { "epoch": 0.04, "learning_rate": 4.984799211967561e-05, "loss": 0.9492, "step": 35216 }, { "epoch": 0.04, "learning_rate": 4.984798347086575e-05, "loss": 1.1153, "step": 35217 }, { "epoch": 0.04, "learning_rate": 4.984797482181062e-05, "loss": 1.0779, "step": 35218 }, { "epoch": 0.04, "learning_rate": 4.984796617251019e-05, "loss": 0.8904, "step": 35219 }, { "epoch": 0.04, "learning_rate": 4.984795752296446e-05, "loss": 0.7934, "step": 35220 }, { "epoch": 0.04, "learning_rate": 4.9847948873173454e-05, "loss": 1.0323, "step": 35221 }, { "epoch": 0.04, "learning_rate": 4.984794022313716e-05, "loss": 1.1143, "step": 35222 }, { "epoch": 0.04, "learning_rate": 4.9847931572855564e-05, "loss": 0.7686, "step": 35223 }, { "epoch": 0.04, "learning_rate": 4.9847922922328695e-05, "loss": 0.3302, "step": 35224 }, { "epoch": 0.04, "learning_rate": 4.9847914271556526e-05, "loss": 0.1488, "step": 35225 }, { "epoch": 0.04, "learning_rate": 4.984790562053908e-05, "loss": 0.0552, "step": 35226 }, { "epoch": 0.04, "learning_rate": 4.984789696927634e-05, "loss": 0.5229, "step": 35227 }, { "epoch": 0.04, "learning_rate": 4.9847888317768315e-05, "loss": 0.9628, "step": 35228 }, { "epoch": 0.04, "learning_rate": 4.9847879666014997e-05, "loss": 0.9744, "step": 35229 }, { "epoch": 0.04, "learning_rate": 4.98478710140164e-05, "loss": 0.7234, "step": 35230 }, { "epoch": 0.04, "learning_rate": 4.9847862361772505e-05, "loss": 0.8032, "step": 35231 }, { "epoch": 0.04, "learning_rate": 4.984785370928333e-05, "loss": 0.9162, "step": 35232 }, { "epoch": 0.04, "learning_rate": 4.9847845056548865e-05, "loss": 0.7407, "step": 35233 }, { "epoch": 0.04, "learning_rate": 4.984783640356911e-05, "loss": 1.4039, "step": 35234 }, { "epoch": 0.04, "learning_rate": 4.984782775034407e-05, "loss": 1.2139, "step": 35235 }, { "epoch": 0.04, "learning_rate": 4.984781909687374e-05, "loss": 0.9048, "step": 35236 }, { "epoch": 0.04, "learning_rate": 4.9847810443158126e-05, "loss": 0.9498, "step": 35237 }, { "epoch": 0.04, "learning_rate": 4.984780178919722e-05, "loss": 1.1104, "step": 35238 }, { "epoch": 0.04, "learning_rate": 4.984779313499104e-05, "loss": 1.0529, "step": 35239 }, { "epoch": 0.04, "learning_rate": 4.984778448053956e-05, "loss": 0.7387, "step": 35240 }, { "epoch": 0.04, "learning_rate": 4.9847775825842804e-05, "loss": 1.0026, "step": 35241 }, { "epoch": 0.04, "learning_rate": 4.984776717090075e-05, "loss": 1.3559, "step": 35242 }, { "epoch": 0.04, "learning_rate": 4.9847758515713414e-05, "loss": 0.9414, "step": 35243 }, { "epoch": 0.04, "learning_rate": 4.984774986028079e-05, "loss": 0.8637, "step": 35244 }, { "epoch": 0.04, "learning_rate": 4.984774120460288e-05, "loss": 1.0638, "step": 35245 }, { "epoch": 0.04, "learning_rate": 4.984773254867968e-05, "loss": 1.0209, "step": 35246 }, { "epoch": 0.04, "learning_rate": 4.98477238925112e-05, "loss": 1.664, "step": 35247 }, { "epoch": 0.04, "learning_rate": 4.984771523609744e-05, "loss": 1.1672, "step": 35248 }, { "epoch": 0.04, "learning_rate": 4.9847706579438383e-05, "loss": 1.25, "step": 35249 }, { "epoch": 0.04, "learning_rate": 4.984769792253404e-05, "loss": 0.9078, "step": 35250 }, { "epoch": 0.04, "learning_rate": 4.984768926538441e-05, "loss": 1.244, "step": 35251 }, { "epoch": 0.04, "learning_rate": 4.98476806079895e-05, "loss": 0.9359, "step": 35252 }, { "epoch": 0.04, "learning_rate": 4.9847671950349304e-05, "loss": 0.5297, "step": 35253 }, { "epoch": 0.04, "learning_rate": 4.984766329246382e-05, "loss": 0.9301, "step": 35254 }, { "epoch": 0.04, "learning_rate": 4.9847654634333055e-05, "loss": 0.9451, "step": 35255 }, { "epoch": 0.04, "learning_rate": 4.984764597595699e-05, "loss": 0.9379, "step": 35256 }, { "epoch": 0.04, "learning_rate": 4.984763731733565e-05, "loss": 0.8832, "step": 35257 }, { "epoch": 0.04, "learning_rate": 4.9847628658469024e-05, "loss": 1.3297, "step": 35258 }, { "epoch": 0.04, "learning_rate": 4.984761999935712e-05, "loss": 1.1195, "step": 35259 }, { "epoch": 0.04, "learning_rate": 4.9847611339999916e-05, "loss": 0.9588, "step": 35260 }, { "epoch": 0.04, "learning_rate": 4.984760268039743e-05, "loss": 0.8528, "step": 35261 }, { "epoch": 0.04, "learning_rate": 4.9847594020549674e-05, "loss": 1.1645, "step": 35262 }, { "epoch": 0.04, "learning_rate": 4.984758536045662e-05, "loss": 1.1353, "step": 35263 }, { "epoch": 0.04, "learning_rate": 4.984757670011828e-05, "loss": 0.9786, "step": 35264 }, { "epoch": 0.04, "learning_rate": 4.984756803953466e-05, "loss": 1.0999, "step": 35265 }, { "epoch": 0.04, "learning_rate": 4.9847559378705756e-05, "loss": 0.8423, "step": 35266 }, { "epoch": 0.04, "learning_rate": 4.9847550717631565e-05, "loss": 0.8518, "step": 35267 }, { "epoch": 0.04, "learning_rate": 4.984754205631209e-05, "loss": 1.1776, "step": 35268 }, { "epoch": 0.04, "learning_rate": 4.984753339474733e-05, "loss": 0.7055, "step": 35269 }, { "epoch": 0.04, "learning_rate": 4.9847524732937276e-05, "loss": 1.0613, "step": 35270 }, { "epoch": 0.04, "learning_rate": 4.984751607088195e-05, "loss": 0.6846, "step": 35271 }, { "epoch": 0.04, "learning_rate": 4.984750740858134e-05, "loss": 0.9037, "step": 35272 }, { "epoch": 0.04, "learning_rate": 4.9847498746035444e-05, "loss": 1.0289, "step": 35273 }, { "epoch": 0.04, "learning_rate": 4.984749008324426e-05, "loss": 0.7932, "step": 35274 }, { "epoch": 0.04, "learning_rate": 4.9847481420207795e-05, "loss": 0.7977, "step": 35275 }, { "epoch": 0.04, "learning_rate": 4.9847472756926053e-05, "loss": 0.8965, "step": 35276 }, { "epoch": 0.04, "learning_rate": 4.984746409339902e-05, "loss": 1.0726, "step": 35277 }, { "epoch": 0.04, "learning_rate": 4.98474554296267e-05, "loss": 1.0342, "step": 35278 }, { "epoch": 0.04, "learning_rate": 4.98474467656091e-05, "loss": 1.2878, "step": 35279 }, { "epoch": 0.04, "learning_rate": 4.984743810134621e-05, "loss": 0.9038, "step": 35280 }, { "epoch": 0.04, "learning_rate": 4.9847429436838045e-05, "loss": 1.1228, "step": 35281 }, { "epoch": 0.04, "learning_rate": 4.9847420772084593e-05, "loss": 1.1851, "step": 35282 }, { "epoch": 0.04, "learning_rate": 4.984741210708587e-05, "loss": 0.991, "step": 35283 }, { "epoch": 0.04, "learning_rate": 4.984740344184184e-05, "loss": 0.8864, "step": 35284 }, { "epoch": 0.04, "learning_rate": 4.9847394776352544e-05, "loss": 1.1357, "step": 35285 }, { "epoch": 0.04, "learning_rate": 4.9847386110617964e-05, "loss": 1.1755, "step": 35286 }, { "epoch": 0.04, "learning_rate": 4.98473774446381e-05, "loss": 0.8697, "step": 35287 }, { "epoch": 0.04, "learning_rate": 4.984736877841295e-05, "loss": 1.199, "step": 35288 }, { "epoch": 0.04, "learning_rate": 4.984736011194252e-05, "loss": 1.168, "step": 35289 }, { "epoch": 0.04, "learning_rate": 4.984735144522681e-05, "loss": 0.5979, "step": 35290 }, { "epoch": 0.04, "learning_rate": 4.9847342778265806e-05, "loss": 0.6703, "step": 35291 }, { "epoch": 0.04, "learning_rate": 4.984733411105953e-05, "loss": 0.7635, "step": 35292 }, { "epoch": 0.04, "learning_rate": 4.984732544360797e-05, "loss": 1.0139, "step": 35293 }, { "epoch": 0.04, "learning_rate": 4.9847316775911125e-05, "loss": 0.9747, "step": 35294 }, { "epoch": 0.04, "learning_rate": 4.9847308107969e-05, "loss": 0.9673, "step": 35295 }, { "epoch": 0.04, "learning_rate": 4.984729943978159e-05, "loss": 1.092, "step": 35296 }, { "epoch": 0.04, "learning_rate": 4.98472907713489e-05, "loss": 0.8167, "step": 35297 }, { "epoch": 0.04, "learning_rate": 4.984728210267092e-05, "loss": 0.8897, "step": 35298 }, { "epoch": 0.04, "learning_rate": 4.984727343374767e-05, "loss": 0.8506, "step": 35299 }, { "epoch": 0.04, "learning_rate": 4.9847264764579136e-05, "loss": 1.0463, "step": 35300 }, { "epoch": 0.04, "learning_rate": 4.9847256095165316e-05, "loss": 1.1365, "step": 35301 }, { "epoch": 0.04, "learning_rate": 4.9847247425506216e-05, "loss": 0.7955, "step": 35302 }, { "epoch": 0.04, "learning_rate": 4.9847238755601836e-05, "loss": 0.852, "step": 35303 }, { "epoch": 0.04, "learning_rate": 4.984723008545218e-05, "loss": 0.9365, "step": 35304 }, { "epoch": 0.04, "learning_rate": 4.9847221415057234e-05, "loss": 1.0777, "step": 35305 }, { "epoch": 0.04, "learning_rate": 4.9847212744417005e-05, "loss": 1.3479, "step": 35306 }, { "epoch": 0.04, "learning_rate": 4.98472040735315e-05, "loss": 1.1182, "step": 35307 }, { "epoch": 0.04, "learning_rate": 4.9847195402400714e-05, "loss": 0.6509, "step": 35308 }, { "epoch": 0.04, "learning_rate": 4.9847186731024644e-05, "loss": 0.8631, "step": 35309 }, { "epoch": 0.04, "learning_rate": 4.98471780594033e-05, "loss": 1.006, "step": 35310 }, { "epoch": 0.04, "learning_rate": 4.984716938753666e-05, "loss": 0.9823, "step": 35311 }, { "epoch": 0.04, "learning_rate": 4.984716071542476e-05, "loss": 0.9474, "step": 35312 }, { "epoch": 0.04, "learning_rate": 4.984715204306756e-05, "loss": 0.9207, "step": 35313 }, { "epoch": 0.04, "learning_rate": 4.984714337046509e-05, "loss": 0.9666, "step": 35314 }, { "epoch": 0.04, "learning_rate": 4.984713469761734e-05, "loss": 1.0452, "step": 35315 }, { "epoch": 0.04, "learning_rate": 4.9847126024524304e-05, "loss": 1.3953, "step": 35316 }, { "epoch": 0.04, "learning_rate": 4.984711735118599e-05, "loss": 1.2963, "step": 35317 }, { "epoch": 0.04, "learning_rate": 4.984710867760239e-05, "loss": 1.0749, "step": 35318 }, { "epoch": 0.04, "learning_rate": 4.9847100003773524e-05, "loss": 0.808, "step": 35319 }, { "epoch": 0.04, "learning_rate": 4.984709132969937e-05, "loss": 1.1279, "step": 35320 }, { "epoch": 0.04, "learning_rate": 4.984708265537993e-05, "loss": 1.0747, "step": 35321 }, { "epoch": 0.04, "learning_rate": 4.984707398081522e-05, "loss": 0.8571, "step": 35322 }, { "epoch": 0.04, "learning_rate": 4.9847065306005225e-05, "loss": 0.9806, "step": 35323 }, { "epoch": 0.04, "learning_rate": 4.984705663094995e-05, "loss": 0.6652, "step": 35324 }, { "epoch": 0.04, "learning_rate": 4.984704795564939e-05, "loss": 0.9588, "step": 35325 }, { "epoch": 0.04, "learning_rate": 4.984703928010356e-05, "loss": 1.0593, "step": 35326 }, { "epoch": 0.04, "learning_rate": 4.984703060431245e-05, "loss": 0.864, "step": 35327 }, { "epoch": 0.04, "learning_rate": 4.984702192827606e-05, "loss": 0.6823, "step": 35328 }, { "epoch": 0.04, "learning_rate": 4.984701325199439e-05, "loss": 0.479, "step": 35329 }, { "epoch": 0.04, "learning_rate": 4.984700457546744e-05, "loss": 1.0757, "step": 35330 }, { "epoch": 0.04, "learning_rate": 4.9846995898695206e-05, "loss": 1.1516, "step": 35331 }, { "epoch": 0.04, "learning_rate": 4.98469872216777e-05, "loss": 1.1057, "step": 35332 }, { "epoch": 0.04, "learning_rate": 4.9846978544414915e-05, "loss": 1.171, "step": 35333 }, { "epoch": 0.04, "learning_rate": 4.984696986690685e-05, "loss": 1.1742, "step": 35334 }, { "epoch": 0.04, "learning_rate": 4.9846961189153495e-05, "loss": 0.9171, "step": 35335 }, { "epoch": 0.04, "learning_rate": 4.9846952511154874e-05, "loss": 1.0105, "step": 35336 }, { "epoch": 0.04, "learning_rate": 4.984694383291097e-05, "loss": 1.1043, "step": 35337 }, { "epoch": 0.04, "learning_rate": 4.984693515442179e-05, "loss": 0.3692, "step": 35338 }, { "epoch": 0.04, "learning_rate": 4.984692647568733e-05, "loss": 1.9987, "step": 35339 }, { "epoch": 0.04, "learning_rate": 4.9846917796707595e-05, "loss": 1.3376, "step": 35340 }, { "epoch": 0.04, "learning_rate": 4.984690911748258e-05, "loss": 0.8989, "step": 35341 }, { "epoch": 0.04, "learning_rate": 4.9846900438012276e-05, "loss": 0.848, "step": 35342 }, { "epoch": 0.04, "learning_rate": 4.984689175829671e-05, "loss": 0.7614, "step": 35343 }, { "epoch": 0.04, "learning_rate": 4.984688307833586e-05, "loss": 1.1036, "step": 35344 }, { "epoch": 0.04, "learning_rate": 4.9846874398129725e-05, "loss": 1.086, "step": 35345 }, { "epoch": 0.04, "learning_rate": 4.984686571767832e-05, "loss": 1.011, "step": 35346 }, { "epoch": 0.04, "learning_rate": 4.984685703698163e-05, "loss": 1.0425, "step": 35347 }, { "epoch": 0.04, "learning_rate": 4.984684835603968e-05, "loss": 0.8912, "step": 35348 }, { "epoch": 0.04, "learning_rate": 4.984683967485243e-05, "loss": 0.784, "step": 35349 }, { "epoch": 0.04, "learning_rate": 4.984683099341991e-05, "loss": 0.9387, "step": 35350 }, { "epoch": 0.04, "learning_rate": 4.984682231174212e-05, "loss": 0.7556, "step": 35351 }, { "epoch": 0.04, "learning_rate": 4.9846813629819047e-05, "loss": 0.5313, "step": 35352 }, { "epoch": 0.04, "learning_rate": 4.98468049476507e-05, "loss": 0.8471, "step": 35353 }, { "epoch": 0.04, "learning_rate": 4.984679626523707e-05, "loss": 0.829, "step": 35354 }, { "epoch": 0.04, "learning_rate": 4.984678758257817e-05, "loss": 0.9784, "step": 35355 }, { "epoch": 0.04, "learning_rate": 4.984677889967398e-05, "loss": 1.0198, "step": 35356 }, { "epoch": 0.04, "learning_rate": 4.984677021652453e-05, "loss": 0.9779, "step": 35357 }, { "epoch": 0.04, "learning_rate": 4.984676153312979e-05, "loss": 1.1738, "step": 35358 }, { "epoch": 0.04, "learning_rate": 4.9846752849489784e-05, "loss": 1.0099, "step": 35359 }, { "epoch": 0.04, "learning_rate": 4.9846744165604494e-05, "loss": 0.9488, "step": 35360 }, { "epoch": 0.04, "learning_rate": 4.9846735481473925e-05, "loss": 0.6395, "step": 35361 }, { "epoch": 0.04, "learning_rate": 4.984672679709809e-05, "loss": 0.9952, "step": 35362 }, { "epoch": 0.04, "learning_rate": 4.984671811247698e-05, "loss": 1.4528, "step": 35363 }, { "epoch": 0.04, "learning_rate": 4.984670942761058e-05, "loss": 1.8283, "step": 35364 }, { "epoch": 0.04, "learning_rate": 4.984670074249891e-05, "loss": 1.6398, "step": 35365 }, { "epoch": 0.04, "learning_rate": 4.9846692057141964e-05, "loss": 1.0044, "step": 35366 }, { "epoch": 0.04, "learning_rate": 4.984668337153974e-05, "loss": 0.9758, "step": 35367 }, { "epoch": 0.04, "learning_rate": 4.984667468569224e-05, "loss": 1.0854, "step": 35368 }, { "epoch": 0.04, "learning_rate": 4.984666599959946e-05, "loss": 1.1148, "step": 35369 }, { "epoch": 0.04, "learning_rate": 4.9846657313261414e-05, "loss": 1.1253, "step": 35370 }, { "epoch": 0.04, "learning_rate": 4.984664862667809e-05, "loss": 1.0509, "step": 35371 }, { "epoch": 0.04, "learning_rate": 4.984663993984949e-05, "loss": 1.1549, "step": 35372 }, { "epoch": 0.04, "learning_rate": 4.9846631252775614e-05, "loss": 0.8335, "step": 35373 }, { "epoch": 0.04, "learning_rate": 4.984662256545646e-05, "loss": 1.1395, "step": 35374 }, { "epoch": 0.04, "learning_rate": 4.984661387789203e-05, "loss": 0.8623, "step": 35375 }, { "epoch": 0.04, "learning_rate": 4.984660519008233e-05, "loss": 1.0867, "step": 35376 }, { "epoch": 0.04, "learning_rate": 4.984659650202735e-05, "loss": 1.2312, "step": 35377 }, { "epoch": 0.04, "learning_rate": 4.98465878137271e-05, "loss": 1.1237, "step": 35378 }, { "epoch": 0.04, "learning_rate": 4.984657912518157e-05, "loss": 0.887, "step": 35379 }, { "epoch": 0.04, "learning_rate": 4.984657043639076e-05, "loss": 1.3979, "step": 35380 }, { "epoch": 0.04, "learning_rate": 4.9846561747354684e-05, "loss": 0.9779, "step": 35381 }, { "epoch": 0.04, "learning_rate": 4.9846553058073334e-05, "loss": 1.0117, "step": 35382 }, { "epoch": 0.04, "learning_rate": 4.9846544368546703e-05, "loss": 1.0942, "step": 35383 }, { "epoch": 0.04, "learning_rate": 4.98465356787748e-05, "loss": 0.5182, "step": 35384 }, { "epoch": 0.04, "learning_rate": 4.984652698875762e-05, "loss": 0.9672, "step": 35385 }, { "epoch": 0.04, "learning_rate": 4.984651829849517e-05, "loss": 0.9896, "step": 35386 }, { "epoch": 0.04, "learning_rate": 4.9846509607987445e-05, "loss": 0.9358, "step": 35387 }, { "epoch": 0.04, "learning_rate": 4.9846500917234446e-05, "loss": 0.9495, "step": 35388 }, { "epoch": 0.04, "learning_rate": 4.9846492226236174e-05, "loss": 1.1464, "step": 35389 }, { "epoch": 0.04, "learning_rate": 4.984648353499263e-05, "loss": 1.0408, "step": 35390 }, { "epoch": 0.04, "learning_rate": 4.98464748435038e-05, "loss": 1.0577, "step": 35391 }, { "epoch": 0.04, "learning_rate": 4.984646615176971e-05, "loss": 1.0566, "step": 35392 }, { "epoch": 0.04, "learning_rate": 4.9846457459790335e-05, "loss": 1.1429, "step": 35393 }, { "epoch": 0.04, "learning_rate": 4.984644876756569e-05, "loss": 1.1724, "step": 35394 }, { "epoch": 0.04, "learning_rate": 4.984644007509577e-05, "loss": 0.9468, "step": 35395 }, { "epoch": 0.04, "learning_rate": 4.984643138238058e-05, "loss": 0.7389, "step": 35396 }, { "epoch": 0.04, "learning_rate": 4.984642268942011e-05, "loss": 0.5397, "step": 35397 }, { "epoch": 0.04, "learning_rate": 4.984641399621438e-05, "loss": 0.5567, "step": 35398 }, { "epoch": 0.04, "learning_rate": 4.9846405302763367e-05, "loss": 0.5169, "step": 35399 }, { "epoch": 0.04, "learning_rate": 4.9846396609067084e-05, "loss": 0.5509, "step": 35400 }, { "epoch": 0.04, "learning_rate": 4.984638791512553e-05, "loss": 0.5026, "step": 35401 }, { "epoch": 0.04, "learning_rate": 4.984637922093869e-05, "loss": 0.503, "step": 35402 }, { "epoch": 0.04, "learning_rate": 4.984637052650659e-05, "loss": 0.665, "step": 35403 }, { "epoch": 0.04, "learning_rate": 4.984636183182921e-05, "loss": 1.1975, "step": 35404 }, { "epoch": 0.04, "learning_rate": 4.984635313690657e-05, "loss": 1.115, "step": 35405 }, { "epoch": 0.04, "learning_rate": 4.9846344441738644e-05, "loss": 0.7975, "step": 35406 }, { "epoch": 0.04, "learning_rate": 4.9846335746325446e-05, "loss": 1.054, "step": 35407 }, { "epoch": 0.04, "learning_rate": 4.9846327050666975e-05, "loss": 0.9657, "step": 35408 }, { "epoch": 0.04, "learning_rate": 4.9846318354763236e-05, "loss": 1.3256, "step": 35409 }, { "epoch": 0.04, "learning_rate": 4.984630965861422e-05, "loss": 1.1433, "step": 35410 }, { "epoch": 0.04, "learning_rate": 4.984630096221994e-05, "loss": 0.8217, "step": 35411 }, { "epoch": 0.04, "learning_rate": 4.9846292265580386e-05, "loss": 1.2064, "step": 35412 }, { "epoch": 0.04, "learning_rate": 4.9846283568695553e-05, "loss": 1.1215, "step": 35413 }, { "epoch": 0.04, "learning_rate": 4.9846274871565454e-05, "loss": 1.5455, "step": 35414 }, { "epoch": 0.04, "learning_rate": 4.984626617419008e-05, "loss": 0.7606, "step": 35415 }, { "epoch": 0.04, "learning_rate": 4.9846257476569433e-05, "loss": 0.7441, "step": 35416 }, { "epoch": 0.04, "learning_rate": 4.984624877870352e-05, "loss": 0.7863, "step": 35417 }, { "epoch": 0.04, "learning_rate": 4.984624008059233e-05, "loss": 0.9624, "step": 35418 }, { "epoch": 0.04, "learning_rate": 4.984623138223587e-05, "loss": 1.1976, "step": 35419 }, { "epoch": 0.04, "learning_rate": 4.9846222683634136e-05, "loss": 1.094, "step": 35420 }, { "epoch": 0.04, "learning_rate": 4.9846213984787135e-05, "loss": 0.6869, "step": 35421 }, { "epoch": 0.04, "learning_rate": 4.984620528569486e-05, "loss": 0.8052, "step": 35422 }, { "epoch": 0.04, "learning_rate": 4.984619658635732e-05, "loss": 1.751, "step": 35423 }, { "epoch": 0.04, "learning_rate": 4.9846187886774495e-05, "loss": 1.2522, "step": 35424 }, { "epoch": 0.04, "learning_rate": 4.9846179186946405e-05, "loss": 1.0573, "step": 35425 }, { "epoch": 0.04, "learning_rate": 4.984617048687305e-05, "loss": 0.9347, "step": 35426 }, { "epoch": 0.04, "learning_rate": 4.9846161786554426e-05, "loss": 1.1104, "step": 35427 }, { "epoch": 0.04, "learning_rate": 4.984615308599052e-05, "loss": 1.0164, "step": 35428 }, { "epoch": 0.04, "learning_rate": 4.9846144385181345e-05, "loss": 1.0149, "step": 35429 }, { "epoch": 0.04, "learning_rate": 4.98461356841269e-05, "loss": 0.8992, "step": 35430 }, { "epoch": 0.04, "learning_rate": 4.984612698282719e-05, "loss": 1.0848, "step": 35431 }, { "epoch": 0.04, "learning_rate": 4.984611828128221e-05, "loss": 1.0297, "step": 35432 }, { "epoch": 0.04, "learning_rate": 4.9846109579491955e-05, "loss": 1.0195, "step": 35433 }, { "epoch": 0.04, "learning_rate": 4.9846100877456436e-05, "loss": 1.191, "step": 35434 }, { "epoch": 0.04, "learning_rate": 4.984609217517564e-05, "loss": 0.8629, "step": 35435 }, { "epoch": 0.04, "learning_rate": 4.984608347264957e-05, "loss": 0.834, "step": 35436 }, { "epoch": 0.04, "learning_rate": 4.984607476987824e-05, "loss": 1.0474, "step": 35437 }, { "epoch": 0.04, "learning_rate": 4.9846066066861634e-05, "loss": 0.7199, "step": 35438 }, { "epoch": 0.04, "learning_rate": 4.984605736359976e-05, "loss": 0.7822, "step": 35439 }, { "epoch": 0.04, "learning_rate": 4.9846048660092615e-05, "loss": 1.0809, "step": 35440 }, { "epoch": 0.04, "learning_rate": 4.984603995634021e-05, "loss": 1.2252, "step": 35441 }, { "epoch": 0.04, "learning_rate": 4.984603125234252e-05, "loss": 1.1013, "step": 35442 }, { "epoch": 0.04, "learning_rate": 4.984602254809957e-05, "loss": 1.1287, "step": 35443 }, { "epoch": 0.04, "learning_rate": 4.9846013843611347e-05, "loss": 1.203, "step": 35444 }, { "epoch": 0.04, "learning_rate": 4.984600513887785e-05, "loss": 1.2037, "step": 35445 }, { "epoch": 0.04, "learning_rate": 4.984599643389909e-05, "loss": 1.0088, "step": 35446 }, { "epoch": 0.04, "learning_rate": 4.9845987728675056e-05, "loss": 1.2701, "step": 35447 }, { "epoch": 0.04, "learning_rate": 4.984597902320576e-05, "loss": 1.1544, "step": 35448 }, { "epoch": 0.04, "learning_rate": 4.98459703174912e-05, "loss": 1.0155, "step": 35449 }, { "epoch": 0.04, "learning_rate": 4.984596161153136e-05, "loss": 1.1083, "step": 35450 }, { "epoch": 0.04, "learning_rate": 4.984595290532625e-05, "loss": 1.0148, "step": 35451 }, { "epoch": 0.04, "learning_rate": 4.984594419887587e-05, "loss": 0.6365, "step": 35452 }, { "epoch": 0.04, "learning_rate": 4.984593549218023e-05, "loss": 1.109, "step": 35453 }, { "epoch": 0.04, "learning_rate": 4.984592678523931e-05, "loss": 0.9471, "step": 35454 }, { "epoch": 0.04, "learning_rate": 4.9845918078053137e-05, "loss": 0.4936, "step": 35455 }, { "epoch": 0.04, "learning_rate": 4.984590937062169e-05, "loss": 0.7972, "step": 35456 }, { "epoch": 0.04, "learning_rate": 4.984590066294497e-05, "loss": 1.0959, "step": 35457 }, { "epoch": 0.04, "learning_rate": 4.9845891955022986e-05, "loss": 0.8842, "step": 35458 }, { "epoch": 0.04, "learning_rate": 4.984588324685573e-05, "loss": 0.8078, "step": 35459 }, { "epoch": 0.04, "learning_rate": 4.9845874538443204e-05, "loss": 1.1154, "step": 35460 }, { "epoch": 0.04, "learning_rate": 4.984586582978542e-05, "loss": 0.8691, "step": 35461 }, { "epoch": 0.04, "learning_rate": 4.984585712088236e-05, "loss": 0.8443, "step": 35462 }, { "epoch": 0.04, "learning_rate": 4.984584841173403e-05, "loss": 1.0395, "step": 35463 }, { "epoch": 0.04, "learning_rate": 4.984583970234044e-05, "loss": 0.921, "step": 35464 }, { "epoch": 0.04, "learning_rate": 4.984583099270158e-05, "loss": 1.181, "step": 35465 }, { "epoch": 0.04, "learning_rate": 4.9845822282817454e-05, "loss": 0.829, "step": 35466 }, { "epoch": 0.04, "learning_rate": 4.984581357268805e-05, "loss": 0.8694, "step": 35467 }, { "epoch": 0.04, "learning_rate": 4.984580486231339e-05, "loss": 0.817, "step": 35468 }, { "epoch": 0.04, "learning_rate": 4.9845796151693456e-05, "loss": 1.0431, "step": 35469 }, { "epoch": 0.04, "learning_rate": 4.984578744082826e-05, "loss": 1.2206, "step": 35470 }, { "epoch": 0.04, "learning_rate": 4.9845778729717796e-05, "loss": 0.9296, "step": 35471 }, { "epoch": 0.04, "learning_rate": 4.9845770018362056e-05, "loss": 1.1075, "step": 35472 }, { "epoch": 0.04, "learning_rate": 4.984576130676106e-05, "loss": 0.8346, "step": 35473 }, { "epoch": 0.04, "learning_rate": 4.9845752594914795e-05, "loss": 0.8243, "step": 35474 }, { "epoch": 0.04, "learning_rate": 4.984574388282326e-05, "loss": 0.2984, "step": 35475 }, { "epoch": 0.04, "learning_rate": 4.984573517048646e-05, "loss": 0.0629, "step": 35476 }, { "epoch": 0.04, "learning_rate": 4.98457264579044e-05, "loss": 0.0291, "step": 35477 }, { "epoch": 0.04, "learning_rate": 4.9845717745077065e-05, "loss": 0.0264, "step": 35478 }, { "epoch": 0.04, "learning_rate": 4.984570903200446e-05, "loss": 0.0841, "step": 35479 }, { "epoch": 0.04, "learning_rate": 4.9845700318686595e-05, "loss": 0.2923, "step": 35480 }, { "epoch": 0.04, "learning_rate": 4.9845691605123466e-05, "loss": 1.3376, "step": 35481 }, { "epoch": 0.04, "learning_rate": 4.9845682891315064e-05, "loss": 1.1168, "step": 35482 }, { "epoch": 0.04, "learning_rate": 4.98456741772614e-05, "loss": 1.088, "step": 35483 }, { "epoch": 0.04, "learning_rate": 4.984566546296247e-05, "loss": 1.1247, "step": 35484 }, { "epoch": 0.04, "learning_rate": 4.984565674841828e-05, "loss": 1.1119, "step": 35485 }, { "epoch": 0.04, "learning_rate": 4.9845648033628815e-05, "loss": 1.4529, "step": 35486 }, { "epoch": 0.04, "learning_rate": 4.9845639318594085e-05, "loss": 0.9286, "step": 35487 }, { "epoch": 0.04, "learning_rate": 4.984563060331409e-05, "loss": 0.9182, "step": 35488 }, { "epoch": 0.04, "learning_rate": 4.984562188778883e-05, "loss": 0.9062, "step": 35489 }, { "epoch": 0.04, "learning_rate": 4.984561317201831e-05, "loss": 1.138, "step": 35490 }, { "epoch": 0.04, "learning_rate": 4.984560445600251e-05, "loss": 0.8796, "step": 35491 }, { "epoch": 0.04, "learning_rate": 4.9845595739741455e-05, "loss": 0.6116, "step": 35492 }, { "epoch": 0.04, "learning_rate": 4.984558702323514e-05, "loss": 1.2383, "step": 35493 }, { "epoch": 0.04, "learning_rate": 4.984557830648355e-05, "loss": 1.1968, "step": 35494 }, { "epoch": 0.04, "learning_rate": 4.98455695894867e-05, "loss": 0.9655, "step": 35495 }, { "epoch": 0.04, "learning_rate": 4.984556087224459e-05, "loss": 1.207, "step": 35496 }, { "epoch": 0.04, "learning_rate": 4.98455521547572e-05, "loss": 1.1957, "step": 35497 }, { "epoch": 0.04, "learning_rate": 4.9845543437024555e-05, "loss": 1.1315, "step": 35498 }, { "epoch": 0.04, "learning_rate": 4.984553471904665e-05, "loss": 1.2101, "step": 35499 }, { "epoch": 0.04, "learning_rate": 4.984552600082347e-05, "loss": 0.9307, "step": 35500 }, { "epoch": 0.04, "eval_loss": 1.074920654296875, "eval_runtime": 128.552, "eval_samples_per_second": 10.774, "eval_steps_per_second": 5.391, "step": 35500 }, { "epoch": 0.04, "learning_rate": 4.9845517282355034e-05, "loss": 1.0622, "step": 35501 }, { "epoch": 0.04, "learning_rate": 4.984550856364133e-05, "loss": 1.0732, "step": 35502 }, { "epoch": 0.04, "learning_rate": 4.984549984468236e-05, "loss": 0.9974, "step": 35503 }, { "epoch": 0.04, "learning_rate": 4.984549112547813e-05, "loss": 0.8338, "step": 35504 }, { "epoch": 0.04, "learning_rate": 4.984548240602863e-05, "loss": 0.379, "step": 35505 }, { "epoch": 0.04, "learning_rate": 4.9845473686333865e-05, "loss": 0.7419, "step": 35506 }, { "epoch": 0.04, "learning_rate": 4.984546496639384e-05, "loss": 1.0488, "step": 35507 }, { "epoch": 0.04, "learning_rate": 4.984545624620855e-05, "loss": 1.0695, "step": 35508 }, { "epoch": 0.04, "learning_rate": 4.9845447525778e-05, "loss": 0.882, "step": 35509 }, { "epoch": 0.04, "learning_rate": 4.984543880510219e-05, "loss": 1.0561, "step": 35510 }, { "epoch": 0.04, "learning_rate": 4.98454300841811e-05, "loss": 0.9131, "step": 35511 }, { "epoch": 0.04, "learning_rate": 4.984542136301475e-05, "loss": 0.6939, "step": 35512 }, { "epoch": 0.04, "learning_rate": 4.984541264160315e-05, "loss": 0.9903, "step": 35513 }, { "epoch": 0.04, "learning_rate": 4.984540391994628e-05, "loss": 1.1215, "step": 35514 }, { "epoch": 0.04, "learning_rate": 4.9845395198044145e-05, "loss": 1.2417, "step": 35515 }, { "epoch": 0.04, "learning_rate": 4.984538647589675e-05, "loss": 0.951, "step": 35516 }, { "epoch": 0.04, "learning_rate": 4.9845377753504083e-05, "loss": 0.9214, "step": 35517 }, { "epoch": 0.04, "learning_rate": 4.984536903086616e-05, "loss": 0.8145, "step": 35518 }, { "epoch": 0.04, "learning_rate": 4.9845360307982975e-05, "loss": 0.7352, "step": 35519 }, { "epoch": 0.04, "learning_rate": 4.984535158485453e-05, "loss": 0.8548, "step": 35520 }, { "epoch": 0.04, "learning_rate": 4.984534286148081e-05, "loss": 0.8834, "step": 35521 }, { "epoch": 0.04, "learning_rate": 4.984533413786183e-05, "loss": 1.0662, "step": 35522 }, { "epoch": 0.04, "learning_rate": 4.9845325413997596e-05, "loss": 0.9115, "step": 35523 }, { "epoch": 0.04, "learning_rate": 4.9845316689888094e-05, "loss": 0.4238, "step": 35524 }, { "epoch": 0.04, "learning_rate": 4.9845307965533325e-05, "loss": 0.8906, "step": 35525 }, { "epoch": 0.04, "learning_rate": 4.984529924093331e-05, "loss": 0.869, "step": 35526 }, { "epoch": 0.04, "learning_rate": 4.9845290516088015e-05, "loss": 0.841, "step": 35527 }, { "epoch": 0.04, "learning_rate": 4.9845281790997466e-05, "loss": 0.4218, "step": 35528 }, { "epoch": 0.04, "learning_rate": 4.984527306566166e-05, "loss": 0.6233, "step": 35529 }, { "epoch": 0.04, "learning_rate": 4.9845264340080574e-05, "loss": 1.0356, "step": 35530 }, { "epoch": 0.04, "learning_rate": 4.9845255614254245e-05, "loss": 1.0757, "step": 35531 }, { "epoch": 0.04, "learning_rate": 4.984524688818264e-05, "loss": 0.9746, "step": 35532 }, { "epoch": 0.04, "learning_rate": 4.984523816186578e-05, "loss": 0.9402, "step": 35533 }, { "epoch": 0.04, "learning_rate": 4.984522943530366e-05, "loss": 0.9099, "step": 35534 }, { "epoch": 0.04, "learning_rate": 4.9845220708496273e-05, "loss": 0.9519, "step": 35535 }, { "epoch": 0.04, "learning_rate": 4.984521198144363e-05, "loss": 1.1088, "step": 35536 }, { "epoch": 0.04, "learning_rate": 4.984520325414572e-05, "loss": 1.0734, "step": 35537 }, { "epoch": 0.04, "learning_rate": 4.984519452660256e-05, "loss": 0.7664, "step": 35538 }, { "epoch": 0.04, "learning_rate": 4.9845185798814134e-05, "loss": 0.8211, "step": 35539 }, { "epoch": 0.04, "learning_rate": 4.984517707078044e-05, "loss": 1.0537, "step": 35540 }, { "epoch": 0.04, "learning_rate": 4.9845168342501487e-05, "loss": 1.1324, "step": 35541 }, { "epoch": 0.04, "learning_rate": 4.984515961397727e-05, "loss": 1.0452, "step": 35542 }, { "epoch": 0.04, "learning_rate": 4.98451508852078e-05, "loss": 1.1262, "step": 35543 }, { "epoch": 0.04, "learning_rate": 4.984514215619306e-05, "loss": 0.8965, "step": 35544 }, { "epoch": 0.04, "learning_rate": 4.9845133426933064e-05, "loss": 0.9886, "step": 35545 }, { "epoch": 0.04, "learning_rate": 4.9845124697427813e-05, "loss": 1.2205, "step": 35546 }, { "epoch": 0.04, "learning_rate": 4.984511596767729e-05, "loss": 1.0511, "step": 35547 }, { "epoch": 0.04, "learning_rate": 4.984510723768152e-05, "loss": 1.0725, "step": 35548 }, { "epoch": 0.04, "learning_rate": 4.984509850744048e-05, "loss": 0.8924, "step": 35549 }, { "epoch": 0.04, "learning_rate": 4.9845089776954176e-05, "loss": 1.041, "step": 35550 }, { "epoch": 0.04, "learning_rate": 4.9845081046222625e-05, "loss": 0.9047, "step": 35551 }, { "epoch": 0.04, "learning_rate": 4.984507231524581e-05, "loss": 0.9395, "step": 35552 }, { "epoch": 0.04, "learning_rate": 4.984506358402373e-05, "loss": 1.0754, "step": 35553 }, { "epoch": 0.04, "learning_rate": 4.984505485255639e-05, "loss": 1.076, "step": 35554 }, { "epoch": 0.04, "learning_rate": 4.984504612084379e-05, "loss": 0.8581, "step": 35555 }, { "epoch": 0.04, "learning_rate": 4.984503738888593e-05, "loss": 0.9374, "step": 35556 }, { "epoch": 0.04, "learning_rate": 4.9845028656682816e-05, "loss": 0.8387, "step": 35557 }, { "epoch": 0.04, "learning_rate": 4.984501992423444e-05, "loss": 0.8342, "step": 35558 }, { "epoch": 0.04, "learning_rate": 4.9845011191540805e-05, "loss": 0.7948, "step": 35559 }, { "epoch": 0.04, "learning_rate": 4.98450024586019e-05, "loss": 1.2674, "step": 35560 }, { "epoch": 0.04, "learning_rate": 4.984499372541774e-05, "loss": 1.349, "step": 35561 }, { "epoch": 0.04, "learning_rate": 4.984498499198833e-05, "loss": 0.8993, "step": 35562 }, { "epoch": 0.04, "learning_rate": 4.984497625831366e-05, "loss": 0.7697, "step": 35563 }, { "epoch": 0.04, "learning_rate": 4.9844967524393725e-05, "loss": 1.0755, "step": 35564 }, { "epoch": 0.04, "learning_rate": 4.984495879022853e-05, "loss": 1.2239, "step": 35565 }, { "epoch": 0.04, "learning_rate": 4.984495005581808e-05, "loss": 1.3209, "step": 35566 }, { "epoch": 0.04, "learning_rate": 4.984494132116237e-05, "loss": 0.8872, "step": 35567 }, { "epoch": 0.04, "learning_rate": 4.98449325862614e-05, "loss": 0.9954, "step": 35568 }, { "epoch": 0.04, "learning_rate": 4.984492385111517e-05, "loss": 1.1882, "step": 35569 }, { "epoch": 0.04, "learning_rate": 4.984491511572368e-05, "loss": 0.8167, "step": 35570 }, { "epoch": 0.04, "learning_rate": 4.9844906380086944e-05, "loss": 0.7586, "step": 35571 }, { "epoch": 0.04, "learning_rate": 4.984489764420494e-05, "loss": 1.0209, "step": 35572 }, { "epoch": 0.04, "learning_rate": 4.984488890807767e-05, "loss": 0.9987, "step": 35573 }, { "epoch": 0.04, "learning_rate": 4.984488017170515e-05, "loss": 0.8803, "step": 35574 }, { "epoch": 0.04, "learning_rate": 4.984487143508737e-05, "loss": 0.9528, "step": 35575 }, { "epoch": 0.04, "learning_rate": 4.984486269822434e-05, "loss": 0.7082, "step": 35576 }, { "epoch": 0.04, "learning_rate": 4.9844853961116047e-05, "loss": 0.8565, "step": 35577 }, { "epoch": 0.04, "learning_rate": 4.9844845223762493e-05, "loss": 0.7315, "step": 35578 }, { "epoch": 0.04, "learning_rate": 4.984483648616368e-05, "loss": 0.9103, "step": 35579 }, { "epoch": 0.04, "learning_rate": 4.9844827748319614e-05, "loss": 0.9051, "step": 35580 }, { "epoch": 0.04, "learning_rate": 4.984481901023029e-05, "loss": 0.8553, "step": 35581 }, { "epoch": 0.04, "learning_rate": 4.98448102718957e-05, "loss": 1.0612, "step": 35582 }, { "epoch": 0.04, "learning_rate": 4.984480153331586e-05, "loss": 0.9809, "step": 35583 }, { "epoch": 0.04, "learning_rate": 4.984479279449076e-05, "loss": 0.8094, "step": 35584 }, { "epoch": 0.04, "learning_rate": 4.984478405542041e-05, "loss": 0.9999, "step": 35585 }, { "epoch": 0.04, "learning_rate": 4.9844775316104794e-05, "loss": 0.8881, "step": 35586 }, { "epoch": 0.04, "learning_rate": 4.984476657654392e-05, "loss": 1.0075, "step": 35587 }, { "epoch": 0.04, "learning_rate": 4.98447578367378e-05, "loss": 0.8136, "step": 35588 }, { "epoch": 0.04, "learning_rate": 4.9844749096686414e-05, "loss": 0.5092, "step": 35589 }, { "epoch": 0.04, "learning_rate": 4.984474035638977e-05, "loss": 0.2087, "step": 35590 }, { "epoch": 0.04, "learning_rate": 4.984473161584787e-05, "loss": 0.3954, "step": 35591 }, { "epoch": 0.04, "learning_rate": 4.9844722875060726e-05, "loss": 0.9006, "step": 35592 }, { "epoch": 0.04, "learning_rate": 4.9844714134028306e-05, "loss": 0.9634, "step": 35593 }, { "epoch": 0.04, "learning_rate": 4.984470539275064e-05, "loss": 1.1577, "step": 35594 }, { "epoch": 0.04, "learning_rate": 4.984469665122772e-05, "loss": 0.9211, "step": 35595 }, { "epoch": 0.04, "learning_rate": 4.984468790945954e-05, "loss": 0.9429, "step": 35596 }, { "epoch": 0.04, "learning_rate": 4.98446791674461e-05, "loss": 1.3022, "step": 35597 }, { "epoch": 0.04, "learning_rate": 4.9844670425187404e-05, "loss": 0.9114, "step": 35598 }, { "epoch": 0.04, "learning_rate": 4.984466168268346e-05, "loss": 0.9705, "step": 35599 }, { "epoch": 0.04, "learning_rate": 4.9844652939934256e-05, "loss": 1.1915, "step": 35600 }, { "epoch": 0.04, "learning_rate": 4.9844644196939796e-05, "loss": 0.7148, "step": 35601 }, { "epoch": 0.04, "learning_rate": 4.9844635453700075e-05, "loss": 0.6594, "step": 35602 }, { "epoch": 0.04, "learning_rate": 4.98446267102151e-05, "loss": 0.5457, "step": 35603 }, { "epoch": 0.04, "learning_rate": 4.9844617966484874e-05, "loss": 0.8753, "step": 35604 }, { "epoch": 0.04, "learning_rate": 4.984460922250939e-05, "loss": 0.9331, "step": 35605 }, { "epoch": 0.04, "learning_rate": 4.984460047828865e-05, "loss": 0.9237, "step": 35606 }, { "epoch": 0.04, "learning_rate": 4.984459173382265e-05, "loss": 0.9059, "step": 35607 }, { "epoch": 0.04, "learning_rate": 4.9844582989111406e-05, "loss": 1.0297, "step": 35608 }, { "epoch": 0.04, "learning_rate": 4.98445742441549e-05, "loss": 1.2184, "step": 35609 }, { "epoch": 0.04, "learning_rate": 4.984456549895313e-05, "loss": 1.1075, "step": 35610 }, { "epoch": 0.04, "learning_rate": 4.9844556753506124e-05, "loss": 0.9655, "step": 35611 }, { "epoch": 0.04, "learning_rate": 4.984454800781385e-05, "loss": 0.9831, "step": 35612 }, { "epoch": 0.04, "learning_rate": 4.984453926187632e-05, "loss": 0.7876, "step": 35613 }, { "epoch": 0.04, "learning_rate": 4.984453051569354e-05, "loss": 0.4685, "step": 35614 }, { "epoch": 0.04, "learning_rate": 4.984452176926551e-05, "loss": 0.3275, "step": 35615 }, { "epoch": 0.04, "learning_rate": 4.984451302259222e-05, "loss": 0.8144, "step": 35616 }, { "epoch": 0.04, "learning_rate": 4.9844504275673674e-05, "loss": 0.9203, "step": 35617 }, { "epoch": 0.04, "learning_rate": 4.984449552850987e-05, "loss": 1.0247, "step": 35618 }, { "epoch": 0.04, "learning_rate": 4.984448678110082e-05, "loss": 0.6517, "step": 35619 }, { "epoch": 0.04, "learning_rate": 4.984447803344652e-05, "loss": 0.7768, "step": 35620 }, { "epoch": 0.04, "learning_rate": 4.984446928554695e-05, "loss": 1.3702, "step": 35621 }, { "epoch": 0.04, "learning_rate": 4.984446053740214e-05, "loss": 1.2153, "step": 35622 }, { "epoch": 0.04, "learning_rate": 4.984445178901207e-05, "loss": 1.3629, "step": 35623 }, { "epoch": 0.04, "learning_rate": 4.9844443040376744e-05, "loss": 1.3987, "step": 35624 }, { "epoch": 0.04, "learning_rate": 4.984443429149617e-05, "loss": 1.1554, "step": 35625 }, { "epoch": 0.04, "learning_rate": 4.984442554237033e-05, "loss": 1.2544, "step": 35626 }, { "epoch": 0.04, "learning_rate": 4.984441679299925e-05, "loss": 0.8952, "step": 35627 }, { "epoch": 0.04, "learning_rate": 4.984440804338291e-05, "loss": 1.1739, "step": 35628 }, { "epoch": 0.04, "learning_rate": 4.9844399293521315e-05, "loss": 0.9961, "step": 35629 }, { "epoch": 0.04, "learning_rate": 4.984439054341447e-05, "loss": 0.6917, "step": 35630 }, { "epoch": 0.04, "learning_rate": 4.984438179306237e-05, "loss": 0.97, "step": 35631 }, { "epoch": 0.04, "learning_rate": 4.984437304246502e-05, "loss": 0.9295, "step": 35632 }, { "epoch": 0.04, "learning_rate": 4.9844364291622414e-05, "loss": 0.768, "step": 35633 }, { "epoch": 0.04, "learning_rate": 4.9844355540534547e-05, "loss": 1.0434, "step": 35634 }, { "epoch": 0.04, "learning_rate": 4.984434678920144e-05, "loss": 1.0461, "step": 35635 }, { "epoch": 0.04, "learning_rate": 4.984433803762307e-05, "loss": 1.0325, "step": 35636 }, { "epoch": 0.04, "learning_rate": 4.984432928579946e-05, "loss": 1.0239, "step": 35637 }, { "epoch": 0.04, "learning_rate": 4.9844320533730586e-05, "loss": 0.7943, "step": 35638 }, { "epoch": 0.04, "learning_rate": 4.9844311781416466e-05, "loss": 1.0785, "step": 35639 }, { "epoch": 0.04, "learning_rate": 4.9844303028857086e-05, "loss": 0.8401, "step": 35640 }, { "epoch": 0.04, "learning_rate": 4.984429427605246e-05, "loss": 1.0434, "step": 35641 }, { "epoch": 0.04, "learning_rate": 4.984428552300258e-05, "loss": 0.9105, "step": 35642 }, { "epoch": 0.04, "learning_rate": 4.9844276769707446e-05, "loss": 1.0619, "step": 35643 }, { "epoch": 0.04, "learning_rate": 4.984426801616706e-05, "loss": 1.3015, "step": 35644 }, { "epoch": 0.04, "learning_rate": 4.9844259262381426e-05, "loss": 1.0374, "step": 35645 }, { "epoch": 0.04, "learning_rate": 4.984425050835054e-05, "loss": 0.5783, "step": 35646 }, { "epoch": 0.04, "learning_rate": 4.984424175407439e-05, "loss": 0.7916, "step": 35647 }, { "epoch": 0.04, "learning_rate": 4.9844232999553e-05, "loss": 1.0366, "step": 35648 }, { "epoch": 0.04, "learning_rate": 4.9844224244786354e-05, "loss": 0.985, "step": 35649 }, { "epoch": 0.04, "learning_rate": 4.984421548977446e-05, "loss": 1.0898, "step": 35650 }, { "epoch": 0.04, "learning_rate": 4.9844206734517315e-05, "loss": 1.0672, "step": 35651 }, { "epoch": 0.04, "learning_rate": 4.984419797901491e-05, "loss": 0.8315, "step": 35652 }, { "epoch": 0.04, "learning_rate": 4.9844189223267256e-05, "loss": 0.3422, "step": 35653 }, { "epoch": 0.04, "learning_rate": 4.9844180467274356e-05, "loss": 0.1713, "step": 35654 }, { "epoch": 0.04, "learning_rate": 4.9844171711036204e-05, "loss": 0.159, "step": 35655 }, { "epoch": 0.04, "learning_rate": 4.98441629545528e-05, "loss": 0.1263, "step": 35656 }, { "epoch": 0.04, "learning_rate": 4.9844154197824145e-05, "loss": 0.3435, "step": 35657 }, { "epoch": 0.04, "learning_rate": 4.984414544085023e-05, "loss": 1.0127, "step": 35658 }, { "epoch": 0.04, "learning_rate": 4.9844136683631074e-05, "loss": 0.3764, "step": 35659 }, { "epoch": 0.04, "learning_rate": 4.984412792616667e-05, "loss": 0.4668, "step": 35660 }, { "epoch": 0.04, "learning_rate": 4.9844119168457e-05, "loss": 0.9383, "step": 35661 }, { "epoch": 0.04, "learning_rate": 4.98441104105021e-05, "loss": 0.5878, "step": 35662 }, { "epoch": 0.04, "learning_rate": 4.984410165230193e-05, "loss": 0.1589, "step": 35663 }, { "epoch": 0.04, "learning_rate": 4.9844092893856526e-05, "loss": 0.7427, "step": 35664 }, { "epoch": 0.04, "learning_rate": 4.984408413516586e-05, "loss": 0.4986, "step": 35665 }, { "epoch": 0.04, "learning_rate": 4.984407537622995e-05, "loss": 1.1592, "step": 35666 }, { "epoch": 0.04, "learning_rate": 4.984406661704879e-05, "loss": 1.3337, "step": 35667 }, { "epoch": 0.04, "learning_rate": 4.984405785762237e-05, "loss": 0.9085, "step": 35668 }, { "epoch": 0.04, "learning_rate": 4.9844049097950714e-05, "loss": 0.8542, "step": 35669 }, { "epoch": 0.04, "learning_rate": 4.98440403380338e-05, "loss": 0.9674, "step": 35670 }, { "epoch": 0.04, "learning_rate": 4.984403157787164e-05, "loss": 0.9193, "step": 35671 }, { "epoch": 0.04, "learning_rate": 4.9844022817464226e-05, "loss": 0.8836, "step": 35672 }, { "epoch": 0.04, "learning_rate": 4.984401405681156e-05, "loss": 1.0496, "step": 35673 }, { "epoch": 0.04, "learning_rate": 4.9844005295913657e-05, "loss": 1.1534, "step": 35674 }, { "epoch": 0.04, "learning_rate": 4.9843996534770485e-05, "loss": 0.9335, "step": 35675 }, { "epoch": 0.04, "learning_rate": 4.984398777338208e-05, "loss": 0.9213, "step": 35676 }, { "epoch": 0.04, "learning_rate": 4.984397901174842e-05, "loss": 1.1905, "step": 35677 }, { "epoch": 0.04, "learning_rate": 4.984397024986951e-05, "loss": 1.3698, "step": 35678 }, { "epoch": 0.04, "learning_rate": 4.984396148774535e-05, "loss": 1.7347, "step": 35679 }, { "epoch": 0.04, "learning_rate": 4.9843952725375944e-05, "loss": 1.4196, "step": 35680 }, { "epoch": 0.04, "learning_rate": 4.984394396276129e-05, "loss": 1.2891, "step": 35681 }, { "epoch": 0.04, "learning_rate": 4.984393519990138e-05, "loss": 1.4047, "step": 35682 }, { "epoch": 0.04, "learning_rate": 4.9843926436796226e-05, "loss": 1.4042, "step": 35683 }, { "epoch": 0.04, "learning_rate": 4.984391767344582e-05, "loss": 1.2641, "step": 35684 }, { "epoch": 0.04, "learning_rate": 4.9843908909850165e-05, "loss": 1.2154, "step": 35685 }, { "epoch": 0.04, "learning_rate": 4.984390014600927e-05, "loss": 1.2608, "step": 35686 }, { "epoch": 0.04, "learning_rate": 4.984389138192312e-05, "loss": 1.3248, "step": 35687 }, { "epoch": 0.04, "learning_rate": 4.984388261759172e-05, "loss": 1.2879, "step": 35688 }, { "epoch": 0.04, "learning_rate": 4.984387385301508e-05, "loss": 1.3666, "step": 35689 }, { "epoch": 0.04, "learning_rate": 4.984386508819318e-05, "loss": 1.3216, "step": 35690 }, { "epoch": 0.04, "learning_rate": 4.984385632312604e-05, "loss": 1.2325, "step": 35691 }, { "epoch": 0.04, "learning_rate": 4.984384755781365e-05, "loss": 1.3281, "step": 35692 }, { "epoch": 0.04, "learning_rate": 4.984383879225601e-05, "loss": 1.1993, "step": 35693 }, { "epoch": 0.04, "learning_rate": 4.984383002645312e-05, "loss": 1.3123, "step": 35694 }, { "epoch": 0.04, "learning_rate": 4.984382126040499e-05, "loss": 1.3348, "step": 35695 }, { "epoch": 0.04, "learning_rate": 4.9843812494111606e-05, "loss": 1.2534, "step": 35696 }, { "epoch": 0.04, "learning_rate": 4.984380372757298e-05, "loss": 1.242, "step": 35697 }, { "epoch": 0.04, "learning_rate": 4.98437949607891e-05, "loss": 1.335, "step": 35698 }, { "epoch": 0.04, "learning_rate": 4.984378619375998e-05, "loss": 1.078, "step": 35699 }, { "epoch": 0.04, "learning_rate": 4.9843777426485604e-05, "loss": 0.9686, "step": 35700 }, { "epoch": 0.04, "learning_rate": 4.984376865896598e-05, "loss": 0.7852, "step": 35701 }, { "epoch": 0.04, "learning_rate": 4.9843759891201114e-05, "loss": 0.9842, "step": 35702 }, { "epoch": 0.04, "learning_rate": 4.9843751123191006e-05, "loss": 1.2287, "step": 35703 }, { "epoch": 0.04, "learning_rate": 4.984374235493564e-05, "loss": 1.1586, "step": 35704 }, { "epoch": 0.04, "learning_rate": 4.984373358643504e-05, "loss": 0.9486, "step": 35705 }, { "epoch": 0.04, "learning_rate": 4.984372481768918e-05, "loss": 1.3406, "step": 35706 }, { "epoch": 0.04, "learning_rate": 4.984371604869808e-05, "loss": 0.9539, "step": 35707 }, { "epoch": 0.04, "learning_rate": 4.9843707279461734e-05, "loss": 1.0154, "step": 35708 }, { "epoch": 0.04, "learning_rate": 4.984369850998013e-05, "loss": 1.2615, "step": 35709 }, { "epoch": 0.04, "learning_rate": 4.98436897402533e-05, "loss": 0.9997, "step": 35710 }, { "epoch": 0.04, "learning_rate": 4.984368097028121e-05, "loss": 0.8818, "step": 35711 }, { "epoch": 0.04, "learning_rate": 4.984367220006387e-05, "loss": 0.6707, "step": 35712 }, { "epoch": 0.04, "learning_rate": 4.98436634296013e-05, "loss": 0.4047, "step": 35713 }, { "epoch": 0.04, "learning_rate": 4.984365465889347e-05, "loss": 0.42, "step": 35714 }, { "epoch": 0.04, "learning_rate": 4.984364588794039e-05, "loss": 1.534, "step": 35715 }, { "epoch": 0.04, "learning_rate": 4.984363711674208e-05, "loss": 1.4167, "step": 35716 }, { "epoch": 0.04, "learning_rate": 4.984362834529851e-05, "loss": 1.1574, "step": 35717 }, { "epoch": 0.04, "learning_rate": 4.98436195736097e-05, "loss": 0.6924, "step": 35718 }, { "epoch": 0.04, "learning_rate": 4.9843610801675646e-05, "loss": 0.9367, "step": 35719 }, { "epoch": 0.04, "learning_rate": 4.984360202949635e-05, "loss": 0.8826, "step": 35720 }, { "epoch": 0.04, "learning_rate": 4.9843593257071795e-05, "loss": 1.2747, "step": 35721 }, { "epoch": 0.04, "learning_rate": 4.984358448440201e-05, "loss": 1.2626, "step": 35722 }, { "epoch": 0.04, "learning_rate": 4.9843575711486964e-05, "loss": 1.3765, "step": 35723 }, { "epoch": 0.04, "learning_rate": 4.984356693832668e-05, "loss": 1.0162, "step": 35724 }, { "epoch": 0.04, "learning_rate": 4.9843558164921154e-05, "loss": 0.471, "step": 35725 }, { "epoch": 0.04, "learning_rate": 4.984354939127038e-05, "loss": 0.1131, "step": 35726 }, { "epoch": 0.04, "learning_rate": 4.9843540617374364e-05, "loss": 0.0706, "step": 35727 }, { "epoch": 0.04, "learning_rate": 4.984353184323309e-05, "loss": 0.0601, "step": 35728 }, { "epoch": 0.04, "learning_rate": 4.984352306884659e-05, "loss": 0.057, "step": 35729 }, { "epoch": 0.04, "learning_rate": 4.984351429421483e-05, "loss": 0.0385, "step": 35730 }, { "epoch": 0.04, "learning_rate": 4.984350551933784e-05, "loss": 0.0489, "step": 35731 }, { "epoch": 0.04, "learning_rate": 4.9843496744215596e-05, "loss": 0.075, "step": 35732 }, { "epoch": 0.04, "learning_rate": 4.9843487968848105e-05, "loss": 0.7287, "step": 35733 }, { "epoch": 0.04, "learning_rate": 4.9843479193235375e-05, "loss": 0.8929, "step": 35734 }, { "epoch": 0.04, "learning_rate": 4.9843470417377405e-05, "loss": 1.1149, "step": 35735 }, { "epoch": 0.04, "learning_rate": 4.984346164127418e-05, "loss": 0.8072, "step": 35736 }, { "epoch": 0.04, "learning_rate": 4.984345286492571e-05, "loss": 1.1397, "step": 35737 }, { "epoch": 0.04, "learning_rate": 4.9843444088332e-05, "loss": 0.9059, "step": 35738 }, { "epoch": 0.04, "learning_rate": 4.984343531149305e-05, "loss": 0.6744, "step": 35739 }, { "epoch": 0.04, "learning_rate": 4.984342653440886e-05, "loss": 1.006, "step": 35740 }, { "epoch": 0.04, "learning_rate": 4.9843417757079415e-05, "loss": 0.8156, "step": 35741 }, { "epoch": 0.04, "learning_rate": 4.984340897950473e-05, "loss": 1.0036, "step": 35742 }, { "epoch": 0.04, "learning_rate": 4.9843400201684804e-05, "loss": 0.6912, "step": 35743 }, { "epoch": 0.04, "learning_rate": 4.9843391423619635e-05, "loss": 0.7325, "step": 35744 }, { "epoch": 0.04, "learning_rate": 4.9843382645309214e-05, "loss": 1.281, "step": 35745 }, { "epoch": 0.04, "learning_rate": 4.984337386675356e-05, "loss": 1.089, "step": 35746 }, { "epoch": 0.04, "learning_rate": 4.984336508795265e-05, "loss": 1.1541, "step": 35747 }, { "epoch": 0.04, "learning_rate": 4.984335630890651e-05, "loss": 1.2013, "step": 35748 }, { "epoch": 0.04, "learning_rate": 4.984334752961512e-05, "loss": 1.1518, "step": 35749 }, { "epoch": 0.04, "learning_rate": 4.9843338750078495e-05, "loss": 1.2795, "step": 35750 }, { "epoch": 0.04, "learning_rate": 4.9843329970296614e-05, "loss": 1.0765, "step": 35751 }, { "epoch": 0.04, "learning_rate": 4.98433211902695e-05, "loss": 1.0017, "step": 35752 }, { "epoch": 0.04, "learning_rate": 4.984331240999715e-05, "loss": 1.2948, "step": 35753 }, { "epoch": 0.04, "learning_rate": 4.984330362947954e-05, "loss": 1.3663, "step": 35754 }, { "epoch": 0.04, "learning_rate": 4.9843294848716694e-05, "loss": 1.2718, "step": 35755 }, { "epoch": 0.04, "learning_rate": 4.984328606770861e-05, "loss": 1.0776, "step": 35756 }, { "epoch": 0.04, "learning_rate": 4.984327728645528e-05, "loss": 1.0094, "step": 35757 }, { "epoch": 0.04, "learning_rate": 4.984326850495671e-05, "loss": 0.7132, "step": 35758 }, { "epoch": 0.04, "learning_rate": 4.984325972321289e-05, "loss": 1.1757, "step": 35759 }, { "epoch": 0.04, "learning_rate": 4.984325094122383e-05, "loss": 1.3481, "step": 35760 }, { "epoch": 0.04, "learning_rate": 4.984324215898953e-05, "loss": 2.5868, "step": 35761 }, { "epoch": 0.04, "learning_rate": 4.984323337650999e-05, "loss": 2.6195, "step": 35762 }, { "epoch": 0.04, "learning_rate": 4.984322459378521e-05, "loss": 1.415, "step": 35763 }, { "epoch": 0.04, "learning_rate": 4.9843215810815186e-05, "loss": 0.9975, "step": 35764 }, { "epoch": 0.04, "learning_rate": 4.984320702759992e-05, "loss": 0.8945, "step": 35765 }, { "epoch": 0.04, "learning_rate": 4.984319824413941e-05, "loss": 0.8188, "step": 35766 }, { "epoch": 0.04, "learning_rate": 4.984318946043366e-05, "loss": 0.5958, "step": 35767 }, { "epoch": 0.04, "learning_rate": 4.984318067648267e-05, "loss": 1.0995, "step": 35768 }, { "epoch": 0.04, "learning_rate": 4.984317189228643e-05, "loss": 1.2327, "step": 35769 }, { "epoch": 0.04, "learning_rate": 4.9843163107844956e-05, "loss": 1.0035, "step": 35770 }, { "epoch": 0.04, "learning_rate": 4.9843154323158246e-05, "loss": 0.5474, "step": 35771 }, { "epoch": 0.04, "learning_rate": 4.984314553822629e-05, "loss": 0.9737, "step": 35772 }, { "epoch": 0.04, "learning_rate": 4.9843136753049094e-05, "loss": 0.6853, "step": 35773 }, { "epoch": 0.04, "learning_rate": 4.984312796762666e-05, "loss": 0.8506, "step": 35774 }, { "epoch": 0.04, "learning_rate": 4.9843119181958977e-05, "loss": 0.7606, "step": 35775 }, { "epoch": 0.04, "learning_rate": 4.9843110396046055e-05, "loss": 0.9043, "step": 35776 }, { "epoch": 0.04, "learning_rate": 4.984310160988789e-05, "loss": 1.0766, "step": 35777 }, { "epoch": 0.04, "learning_rate": 4.984309282348449e-05, "loss": 0.9974, "step": 35778 }, { "epoch": 0.04, "learning_rate": 4.984308403683585e-05, "loss": 1.0074, "step": 35779 }, { "epoch": 0.04, "learning_rate": 4.984307524994197e-05, "loss": 1.0311, "step": 35780 }, { "epoch": 0.04, "learning_rate": 4.984306646280285e-05, "loss": 1.0388, "step": 35781 }, { "epoch": 0.04, "learning_rate": 4.984305767541848e-05, "loss": 1.0477, "step": 35782 }, { "epoch": 0.04, "learning_rate": 4.984304888778888e-05, "loss": 0.9412, "step": 35783 }, { "epoch": 0.04, "learning_rate": 4.9843040099914034e-05, "loss": 0.7505, "step": 35784 }, { "epoch": 0.04, "learning_rate": 4.9843031311793955e-05, "loss": 0.9229, "step": 35785 }, { "epoch": 0.04, "learning_rate": 4.984302252342863e-05, "loss": 1.1431, "step": 35786 }, { "epoch": 0.04, "learning_rate": 4.984301373481807e-05, "loss": 1.0372, "step": 35787 }, { "epoch": 0.04, "learning_rate": 4.984300494596226e-05, "loss": 0.9026, "step": 35788 }, { "epoch": 0.04, "learning_rate": 4.9842996156861225e-05, "loss": 0.9564, "step": 35789 }, { "epoch": 0.04, "learning_rate": 4.984298736751494e-05, "loss": 1.1121, "step": 35790 }, { "epoch": 0.04, "learning_rate": 4.9842978577923414e-05, "loss": 0.8662, "step": 35791 }, { "epoch": 0.04, "learning_rate": 4.9842969788086656e-05, "loss": 0.645, "step": 35792 }, { "epoch": 0.04, "learning_rate": 4.984296099800465e-05, "loss": 0.9325, "step": 35793 }, { "epoch": 0.04, "learning_rate": 4.9842952207677414e-05, "loss": 0.7765, "step": 35794 }, { "epoch": 0.04, "learning_rate": 4.984294341710494e-05, "loss": 1.3199, "step": 35795 }, { "epoch": 0.04, "learning_rate": 4.984293462628722e-05, "loss": 0.8793, "step": 35796 }, { "epoch": 0.04, "learning_rate": 4.9842925835224256e-05, "loss": 0.864, "step": 35797 }, { "epoch": 0.04, "learning_rate": 4.984291704391606e-05, "loss": 1.0938, "step": 35798 }, { "epoch": 0.04, "learning_rate": 4.984290825236263e-05, "loss": 0.9801, "step": 35799 }, { "epoch": 0.04, "learning_rate": 4.9842899460563954e-05, "loss": 1.2037, "step": 35800 }, { "epoch": 0.04, "learning_rate": 4.9842890668520045e-05, "loss": 0.9513, "step": 35801 }, { "epoch": 0.04, "learning_rate": 4.9842881876230896e-05, "loss": 1.1576, "step": 35802 }, { "epoch": 0.04, "learning_rate": 4.984287308369651e-05, "loss": 1.0167, "step": 35803 }, { "epoch": 0.04, "learning_rate": 4.984286429091687e-05, "loss": 0.9971, "step": 35804 }, { "epoch": 0.04, "learning_rate": 4.984285549789201e-05, "loss": 0.7449, "step": 35805 }, { "epoch": 0.04, "learning_rate": 4.98428467046219e-05, "loss": 0.6653, "step": 35806 }, { "epoch": 0.04, "learning_rate": 4.984283791110656e-05, "loss": 0.8425, "step": 35807 }, { "epoch": 0.04, "learning_rate": 4.984282911734598e-05, "loss": 1.1481, "step": 35808 }, { "epoch": 0.04, "learning_rate": 4.984282032334017e-05, "loss": 0.9919, "step": 35809 }, { "epoch": 0.04, "learning_rate": 4.984281152908911e-05, "loss": 1.9367, "step": 35810 }, { "epoch": 0.04, "learning_rate": 4.984280273459282e-05, "loss": 1.664, "step": 35811 }, { "epoch": 0.04, "learning_rate": 4.984279393985128e-05, "loss": 1.7057, "step": 35812 }, { "epoch": 0.04, "learning_rate": 4.984278514486451e-05, "loss": 1.8662, "step": 35813 }, { "epoch": 0.04, "learning_rate": 4.984277634963251e-05, "loss": 1.7142, "step": 35814 }, { "epoch": 0.04, "learning_rate": 4.9842767554155266e-05, "loss": 1.4657, "step": 35815 }, { "epoch": 0.04, "learning_rate": 4.984275875843278e-05, "loss": 1.1538, "step": 35816 }, { "epoch": 0.04, "learning_rate": 4.984274996246506e-05, "loss": 0.9053, "step": 35817 }, { "epoch": 0.04, "learning_rate": 4.984274116625211e-05, "loss": 1.4847, "step": 35818 }, { "epoch": 0.04, "learning_rate": 4.984273236979392e-05, "loss": 1.0648, "step": 35819 }, { "epoch": 0.04, "learning_rate": 4.984272357309049e-05, "loss": 0.9422, "step": 35820 }, { "epoch": 0.04, "learning_rate": 4.984271477614182e-05, "loss": 1.0947, "step": 35821 }, { "epoch": 0.04, "learning_rate": 4.984270597894792e-05, "loss": 0.6728, "step": 35822 }, { "epoch": 0.04, "learning_rate": 4.984269718150878e-05, "loss": 1.0015, "step": 35823 }, { "epoch": 0.04, "learning_rate": 4.98426883838244e-05, "loss": 0.964, "step": 35824 }, { "epoch": 0.04, "learning_rate": 4.984267958589479e-05, "loss": 0.7428, "step": 35825 }, { "epoch": 0.04, "learning_rate": 4.9842670787719934e-05, "loss": 1.3861, "step": 35826 }, { "epoch": 0.04, "learning_rate": 4.984266198929985e-05, "loss": 1.0199, "step": 35827 }, { "epoch": 0.04, "learning_rate": 4.984265319063453e-05, "loss": 1.107, "step": 35828 }, { "epoch": 0.04, "learning_rate": 4.984264439172397e-05, "loss": 1.2811, "step": 35829 }, { "epoch": 0.04, "learning_rate": 4.9842635592568175e-05, "loss": 1.1253, "step": 35830 }, { "epoch": 0.04, "learning_rate": 4.984262679316715e-05, "loss": 1.1324, "step": 35831 }, { "epoch": 0.04, "learning_rate": 4.984261799352088e-05, "loss": 0.856, "step": 35832 }, { "epoch": 0.04, "learning_rate": 4.9842609193629374e-05, "loss": 0.7683, "step": 35833 }, { "epoch": 0.04, "learning_rate": 4.984260039349264e-05, "loss": 1.1629, "step": 35834 }, { "epoch": 0.04, "learning_rate": 4.984259159311067e-05, "loss": 0.802, "step": 35835 }, { "epoch": 0.04, "learning_rate": 4.984258279248345e-05, "loss": 1.1234, "step": 35836 }, { "epoch": 0.04, "learning_rate": 4.984257399161101e-05, "loss": 1.4686, "step": 35837 }, { "epoch": 0.04, "learning_rate": 4.9842565190493326e-05, "loss": 1.3127, "step": 35838 }, { "epoch": 0.04, "learning_rate": 4.9842556389130415e-05, "loss": 1.1848, "step": 35839 }, { "epoch": 0.04, "learning_rate": 4.984254758752226e-05, "loss": 0.9607, "step": 35840 }, { "epoch": 0.04, "learning_rate": 4.984253878566887e-05, "loss": 0.9921, "step": 35841 }, { "epoch": 0.04, "learning_rate": 4.984252998357025e-05, "loss": 0.9933, "step": 35842 }, { "epoch": 0.04, "learning_rate": 4.98425211812264e-05, "loss": 0.985, "step": 35843 }, { "epoch": 0.04, "learning_rate": 4.984251237863731e-05, "loss": 0.7496, "step": 35844 }, { "epoch": 0.04, "learning_rate": 4.984250357580298e-05, "loss": 1.0202, "step": 35845 }, { "epoch": 0.04, "learning_rate": 4.984249477272342e-05, "loss": 0.8187, "step": 35846 }, { "epoch": 0.04, "learning_rate": 4.984248596939862e-05, "loss": 1.3175, "step": 35847 }, { "epoch": 0.04, "learning_rate": 4.984247716582859e-05, "loss": 0.9956, "step": 35848 }, { "epoch": 0.04, "learning_rate": 4.9842468362013325e-05, "loss": 0.9331, "step": 35849 }, { "epoch": 0.04, "learning_rate": 4.984245955795283e-05, "loss": 1.2494, "step": 35850 }, { "epoch": 0.04, "learning_rate": 4.9842450753647096e-05, "loss": 1.108, "step": 35851 }, { "epoch": 0.04, "learning_rate": 4.984244194909613e-05, "loss": 1.3438, "step": 35852 }, { "epoch": 0.04, "learning_rate": 4.984243314429993e-05, "loss": 1.1418, "step": 35853 }, { "epoch": 0.04, "learning_rate": 4.984242433925849e-05, "loss": 1.0287, "step": 35854 }, { "epoch": 0.04, "learning_rate": 4.984241553397182e-05, "loss": 1.4552, "step": 35855 }, { "epoch": 0.04, "learning_rate": 4.984240672843992e-05, "loss": 0.7341, "step": 35856 }, { "epoch": 0.04, "learning_rate": 4.9842397922662785e-05, "loss": 1.0553, "step": 35857 }, { "epoch": 0.04, "learning_rate": 4.984238911664041e-05, "loss": 0.9907, "step": 35858 }, { "epoch": 0.04, "learning_rate": 4.98423803103728e-05, "loss": 1.0518, "step": 35859 }, { "epoch": 0.04, "learning_rate": 4.984237150385997e-05, "loss": 0.9776, "step": 35860 }, { "epoch": 0.04, "learning_rate": 4.984236269710189e-05, "loss": 1.4611, "step": 35861 }, { "epoch": 0.04, "learning_rate": 4.984235389009859e-05, "loss": 1.0435, "step": 35862 }, { "epoch": 0.04, "learning_rate": 4.984234508285005e-05, "loss": 0.9104, "step": 35863 }, { "epoch": 0.04, "learning_rate": 4.984233627535628e-05, "loss": 0.4088, "step": 35864 }, { "epoch": 0.04, "learning_rate": 4.984232746761728e-05, "loss": 0.2331, "step": 35865 }, { "epoch": 0.04, "learning_rate": 4.9842318659633045e-05, "loss": 0.738, "step": 35866 }, { "epoch": 0.04, "learning_rate": 4.9842309851403576e-05, "loss": 1.4269, "step": 35867 }, { "epoch": 0.04, "learning_rate": 4.9842301042928875e-05, "loss": 1.4492, "step": 35868 }, { "epoch": 0.04, "learning_rate": 4.9842292234208933e-05, "loss": 1.4628, "step": 35869 }, { "epoch": 0.04, "learning_rate": 4.9842283425243766e-05, "loss": 1.4231, "step": 35870 }, { "epoch": 0.04, "learning_rate": 4.9842274616033365e-05, "loss": 1.3268, "step": 35871 }, { "epoch": 0.04, "learning_rate": 4.984226580657774e-05, "loss": 1.3977, "step": 35872 }, { "epoch": 0.04, "learning_rate": 4.9842256996876865e-05, "loss": 1.1331, "step": 35873 }, { "epoch": 0.04, "learning_rate": 4.984224818693077e-05, "loss": 0.874, "step": 35874 }, { "epoch": 0.04, "learning_rate": 4.984223937673944e-05, "loss": 1.319, "step": 35875 }, { "epoch": 0.04, "learning_rate": 4.9842230566302875e-05, "loss": 1.0888, "step": 35876 }, { "epoch": 0.04, "learning_rate": 4.984222175562108e-05, "loss": 0.9834, "step": 35877 }, { "epoch": 0.04, "learning_rate": 4.984221294469406e-05, "loss": 0.8755, "step": 35878 }, { "epoch": 0.04, "learning_rate": 4.9842204133521794e-05, "loss": 0.9303, "step": 35879 }, { "epoch": 0.04, "learning_rate": 4.984219532210431e-05, "loss": 1.0206, "step": 35880 }, { "epoch": 0.04, "learning_rate": 4.9842186510441594e-05, "loss": 0.873, "step": 35881 }, { "epoch": 0.04, "learning_rate": 4.984217769853363e-05, "loss": 1.0413, "step": 35882 }, { "epoch": 0.04, "learning_rate": 4.984216888638045e-05, "loss": 1.5246, "step": 35883 }, { "epoch": 0.04, "learning_rate": 4.984216007398204e-05, "loss": 0.8918, "step": 35884 }, { "epoch": 0.04, "learning_rate": 4.984215126133839e-05, "loss": 0.9504, "step": 35885 }, { "epoch": 0.04, "learning_rate": 4.984214244844951e-05, "loss": 0.1445, "step": 35886 }, { "epoch": 0.04, "learning_rate": 4.98421336353154e-05, "loss": 0.1599, "step": 35887 }, { "epoch": 0.04, "learning_rate": 4.984212482193605e-05, "loss": 0.1363, "step": 35888 }, { "epoch": 0.04, "learning_rate": 4.9842116008311486e-05, "loss": 0.1803, "step": 35889 }, { "epoch": 0.04, "learning_rate": 4.9842107194441686e-05, "loss": 0.2928, "step": 35890 }, { "epoch": 0.04, "learning_rate": 4.984209838032665e-05, "loss": 0.1829, "step": 35891 }, { "epoch": 0.04, "learning_rate": 4.984208956596639e-05, "loss": 0.1328, "step": 35892 }, { "epoch": 0.04, "learning_rate": 4.984208075136089e-05, "loss": 0.2917, "step": 35893 }, { "epoch": 0.04, "learning_rate": 4.984207193651016e-05, "loss": 1.0335, "step": 35894 }, { "epoch": 0.04, "learning_rate": 4.9842063121414205e-05, "loss": 0.9299, "step": 35895 }, { "epoch": 0.04, "learning_rate": 4.984205430607302e-05, "loss": 0.9394, "step": 35896 }, { "epoch": 0.04, "learning_rate": 4.98420454904866e-05, "loss": 0.9896, "step": 35897 }, { "epoch": 0.04, "learning_rate": 4.984203667465496e-05, "loss": 1.0243, "step": 35898 }, { "epoch": 0.04, "learning_rate": 4.984202785857808e-05, "loss": 1.1984, "step": 35899 }, { "epoch": 0.04, "learning_rate": 4.9842019042255973e-05, "loss": 0.9209, "step": 35900 }, { "epoch": 0.04, "learning_rate": 4.984201022568864e-05, "loss": 0.9694, "step": 35901 }, { "epoch": 0.04, "learning_rate": 4.984200140887607e-05, "loss": 0.9917, "step": 35902 }, { "epoch": 0.04, "learning_rate": 4.9841992591818274e-05, "loss": 0.6432, "step": 35903 }, { "epoch": 0.04, "learning_rate": 4.9841983774515246e-05, "loss": 1.0637, "step": 35904 }, { "epoch": 0.04, "learning_rate": 4.9841974956967e-05, "loss": 1.0296, "step": 35905 }, { "epoch": 0.04, "learning_rate": 4.9841966139173505e-05, "loss": 1.1093, "step": 35906 }, { "epoch": 0.04, "learning_rate": 4.984195732113479e-05, "loss": 1.0239, "step": 35907 }, { "epoch": 0.04, "learning_rate": 4.9841948502850846e-05, "loss": 1.014, "step": 35908 }, { "epoch": 0.04, "learning_rate": 4.9841939684321673e-05, "loss": 1.0401, "step": 35909 }, { "epoch": 0.04, "learning_rate": 4.984193086554727e-05, "loss": 1.068, "step": 35910 }, { "epoch": 0.04, "learning_rate": 4.984192204652764e-05, "loss": 0.8777, "step": 35911 }, { "epoch": 0.04, "learning_rate": 4.984191322726278e-05, "loss": 1.1457, "step": 35912 }, { "epoch": 0.04, "learning_rate": 4.9841904407752695e-05, "loss": 0.9329, "step": 35913 }, { "epoch": 0.04, "learning_rate": 4.984189558799737e-05, "loss": 0.8354, "step": 35914 }, { "epoch": 0.04, "learning_rate": 4.984188676799683e-05, "loss": 0.748, "step": 35915 }, { "epoch": 0.04, "learning_rate": 4.9841877947751045e-05, "loss": 1.0405, "step": 35916 }, { "epoch": 0.04, "learning_rate": 4.984186912726004e-05, "loss": 1.3368, "step": 35917 }, { "epoch": 0.04, "learning_rate": 4.9841860306523814e-05, "loss": 0.7639, "step": 35918 }, { "epoch": 0.04, "learning_rate": 4.984185148554236e-05, "loss": 0.8809, "step": 35919 }, { "epoch": 0.04, "learning_rate": 4.9841842664315665e-05, "loss": 0.855, "step": 35920 }, { "epoch": 0.04, "learning_rate": 4.9841833842843744e-05, "loss": 0.6448, "step": 35921 }, { "epoch": 0.04, "learning_rate": 4.98418250211266e-05, "loss": 0.6011, "step": 35922 }, { "epoch": 0.04, "learning_rate": 4.9841816199164224e-05, "loss": 0.4623, "step": 35923 }, { "epoch": 0.04, "learning_rate": 4.984180737695663e-05, "loss": 0.7507, "step": 35924 }, { "epoch": 0.04, "learning_rate": 4.984179855450379e-05, "loss": 0.6548, "step": 35925 }, { "epoch": 0.04, "learning_rate": 4.984178973180574e-05, "loss": 0.7536, "step": 35926 }, { "epoch": 0.04, "learning_rate": 4.9841780908862456e-05, "loss": 0.9634, "step": 35927 }, { "epoch": 0.04, "learning_rate": 4.9841772085673945e-05, "loss": 1.1703, "step": 35928 }, { "epoch": 0.04, "learning_rate": 4.98417632622402e-05, "loss": 0.976, "step": 35929 }, { "epoch": 0.04, "learning_rate": 4.984175443856123e-05, "loss": 1.076, "step": 35930 }, { "epoch": 0.04, "learning_rate": 4.9841745614637035e-05, "loss": 1.0893, "step": 35931 }, { "epoch": 0.04, "learning_rate": 4.984173679046762e-05, "loss": 1.0244, "step": 35932 }, { "epoch": 0.04, "learning_rate": 4.984172796605296e-05, "loss": 0.9607, "step": 35933 }, { "epoch": 0.04, "learning_rate": 4.9841719141393095e-05, "loss": 0.9838, "step": 35934 }, { "epoch": 0.04, "learning_rate": 4.984171031648799e-05, "loss": 0.8608, "step": 35935 }, { "epoch": 0.04, "learning_rate": 4.984170149133765e-05, "loss": 1.1162, "step": 35936 }, { "epoch": 0.04, "learning_rate": 4.98416926659421e-05, "loss": 1.2094, "step": 35937 }, { "epoch": 0.04, "learning_rate": 4.984168384030132e-05, "loss": 1.5595, "step": 35938 }, { "epoch": 0.04, "learning_rate": 4.984167501441531e-05, "loss": 1.677, "step": 35939 }, { "epoch": 0.04, "learning_rate": 4.984166618828407e-05, "loss": 0.8705, "step": 35940 }, { "epoch": 0.04, "learning_rate": 4.98416573619076e-05, "loss": 1.1032, "step": 35941 }, { "epoch": 0.04, "learning_rate": 4.984164853528592e-05, "loss": 0.9135, "step": 35942 }, { "epoch": 0.04, "learning_rate": 4.9841639708419e-05, "loss": 1.0587, "step": 35943 }, { "epoch": 0.04, "learning_rate": 4.984163088130686e-05, "loss": 1.2486, "step": 35944 }, { "epoch": 0.04, "learning_rate": 4.984162205394949e-05, "loss": 0.7723, "step": 35945 }, { "epoch": 0.04, "learning_rate": 4.98416132263469e-05, "loss": 1.0054, "step": 35946 }, { "epoch": 0.04, "learning_rate": 4.9841604398499076e-05, "loss": 0.5999, "step": 35947 }, { "epoch": 0.04, "learning_rate": 4.984159557040603e-05, "loss": 0.7698, "step": 35948 }, { "epoch": 0.04, "learning_rate": 4.984158674206776e-05, "loss": 0.7885, "step": 35949 }, { "epoch": 0.04, "learning_rate": 4.984157791348426e-05, "loss": 0.9145, "step": 35950 }, { "epoch": 0.04, "learning_rate": 4.984156908465554e-05, "loss": 1.3103, "step": 35951 }, { "epoch": 0.04, "learning_rate": 4.9841560255581585e-05, "loss": 1.1283, "step": 35952 }, { "epoch": 0.04, "learning_rate": 4.984155142626241e-05, "loss": 0.934, "step": 35953 }, { "epoch": 0.04, "learning_rate": 4.984154259669801e-05, "loss": 0.9433, "step": 35954 }, { "epoch": 0.04, "learning_rate": 4.9841533766888395e-05, "loss": 1.128, "step": 35955 }, { "epoch": 0.04, "learning_rate": 4.9841524936833537e-05, "loss": 1.1882, "step": 35956 }, { "epoch": 0.04, "learning_rate": 4.984151610653347e-05, "loss": 0.8243, "step": 35957 }, { "epoch": 0.04, "learning_rate": 4.984150727598816e-05, "loss": 1.0738, "step": 35958 }, { "epoch": 0.04, "learning_rate": 4.9841498445197646e-05, "loss": 0.9336, "step": 35959 }, { "epoch": 0.04, "learning_rate": 4.9841489614161896e-05, "loss": 1.2405, "step": 35960 }, { "epoch": 0.04, "learning_rate": 4.984148078288092e-05, "loss": 1.0776, "step": 35961 }, { "epoch": 0.04, "learning_rate": 4.9841471951354726e-05, "loss": 1.0468, "step": 35962 }, { "epoch": 0.04, "learning_rate": 4.98414631195833e-05, "loss": 0.8313, "step": 35963 }, { "epoch": 0.04, "learning_rate": 4.984145428756665e-05, "loss": 1.0999, "step": 35964 }, { "epoch": 0.04, "learning_rate": 4.984144545530478e-05, "loss": 0.847, "step": 35965 }, { "epoch": 0.04, "learning_rate": 4.984143662279768e-05, "loss": 0.6584, "step": 35966 }, { "epoch": 0.04, "learning_rate": 4.984142779004536e-05, "loss": 1.0656, "step": 35967 }, { "epoch": 0.04, "learning_rate": 4.984141895704781e-05, "loss": 1.7565, "step": 35968 }, { "epoch": 0.04, "learning_rate": 4.984141012380504e-05, "loss": 1.947, "step": 35969 }, { "epoch": 0.04, "learning_rate": 4.984140129031705e-05, "loss": 2.0046, "step": 35970 }, { "epoch": 0.04, "learning_rate": 4.984139245658384e-05, "loss": 1.8425, "step": 35971 }, { "epoch": 0.04, "learning_rate": 4.9841383622605395e-05, "loss": 1.564, "step": 35972 }, { "epoch": 0.04, "learning_rate": 4.984137478838173e-05, "loss": 0.5267, "step": 35973 }, { "epoch": 0.04, "learning_rate": 4.984136595391284e-05, "loss": 0.6167, "step": 35974 }, { "epoch": 0.04, "learning_rate": 4.9841357119198734e-05, "loss": 1.3876, "step": 35975 }, { "epoch": 0.04, "learning_rate": 4.98413482842394e-05, "loss": 1.4475, "step": 35976 }, { "epoch": 0.04, "learning_rate": 4.984133944903484e-05, "loss": 0.7751, "step": 35977 }, { "epoch": 0.04, "learning_rate": 4.984133061358506e-05, "loss": 0.9983, "step": 35978 }, { "epoch": 0.04, "learning_rate": 4.984132177789006e-05, "loss": 0.9912, "step": 35979 }, { "epoch": 0.04, "learning_rate": 4.984131294194983e-05, "loss": 1.0256, "step": 35980 }, { "epoch": 0.04, "learning_rate": 4.984130410576438e-05, "loss": 1.1346, "step": 35981 }, { "epoch": 0.04, "learning_rate": 4.984129526933371e-05, "loss": 0.5768, "step": 35982 }, { "epoch": 0.04, "learning_rate": 4.984128643265781e-05, "loss": 0.5961, "step": 35983 }, { "epoch": 0.04, "learning_rate": 4.9841277595736694e-05, "loss": 0.412, "step": 35984 }, { "epoch": 0.04, "learning_rate": 4.984126875857035e-05, "loss": 0.4683, "step": 35985 }, { "epoch": 0.04, "learning_rate": 4.984125992115879e-05, "loss": 0.313, "step": 35986 }, { "epoch": 0.04, "learning_rate": 4.9841251083502e-05, "loss": 0.3498, "step": 35987 }, { "epoch": 0.04, "learning_rate": 4.9841242245599996e-05, "loss": 0.3361, "step": 35988 }, { "epoch": 0.04, "learning_rate": 4.984123340745276e-05, "loss": 0.4745, "step": 35989 }, { "epoch": 0.04, "learning_rate": 4.984122456906031e-05, "loss": 0.4374, "step": 35990 }, { "epoch": 0.04, "learning_rate": 4.984121573042263e-05, "loss": 0.2949, "step": 35991 }, { "epoch": 0.04, "learning_rate": 4.984120689153974e-05, "loss": 0.5101, "step": 35992 }, { "epoch": 0.04, "learning_rate": 4.984119805241162e-05, "loss": 1.1441, "step": 35993 }, { "epoch": 0.04, "learning_rate": 4.9841189213038276e-05, "loss": 1.0594, "step": 35994 }, { "epoch": 0.04, "learning_rate": 4.984118037341972e-05, "loss": 1.1236, "step": 35995 }, { "epoch": 0.04, "learning_rate": 4.9841171533555934e-05, "loss": 0.6291, "step": 35996 }, { "epoch": 0.04, "learning_rate": 4.9841162693446916e-05, "loss": 0.5097, "step": 35997 }, { "epoch": 0.04, "learning_rate": 4.984115385309269e-05, "loss": 0.6067, "step": 35998 }, { "epoch": 0.04, "learning_rate": 4.9841145012493244e-05, "loss": 0.5333, "step": 35999 }, { "epoch": 0.04, "learning_rate": 4.9841136171648575e-05, "loss": 0.6414, "step": 36000 }, { "epoch": 0.04, "eval_loss": 1.0994189977645874, "eval_runtime": 127.0969, "eval_samples_per_second": 10.897, "eval_steps_per_second": 5.453, "step": 36000 }, { "epoch": 0.04, "learning_rate": 4.984112733055869e-05, "loss": 0.6357, "step": 36001 }, { "epoch": 0.04, "learning_rate": 4.984111848922357e-05, "loss": 0.9692, "step": 36002 }, { "epoch": 0.04, "learning_rate": 4.984110964764324e-05, "loss": 1.0329, "step": 36003 }, { "epoch": 0.04, "learning_rate": 4.9841100805817686e-05, "loss": 0.863, "step": 36004 }, { "epoch": 0.04, "learning_rate": 4.984109196374691e-05, "loss": 1.1844, "step": 36005 }, { "epoch": 0.04, "learning_rate": 4.984108312143091e-05, "loss": 1.0277, "step": 36006 }, { "epoch": 0.04, "learning_rate": 4.9841074278869696e-05, "loss": 1.3178, "step": 36007 }, { "epoch": 0.04, "learning_rate": 4.984106543606326e-05, "loss": 1.1538, "step": 36008 }, { "epoch": 0.04, "learning_rate": 4.98410565930116e-05, "loss": 0.6218, "step": 36009 }, { "epoch": 0.04, "learning_rate": 4.984104774971472e-05, "loss": 0.9653, "step": 36010 }, { "epoch": 0.04, "learning_rate": 4.984103890617262e-05, "loss": 1.2337, "step": 36011 }, { "epoch": 0.04, "learning_rate": 4.98410300623853e-05, "loss": 0.9217, "step": 36012 }, { "epoch": 0.04, "learning_rate": 4.9841021218352765e-05, "loss": 1.234, "step": 36013 }, { "epoch": 0.04, "learning_rate": 4.9841012374075003e-05, "loss": 0.7716, "step": 36014 }, { "epoch": 0.04, "learning_rate": 4.984100352955202e-05, "loss": 0.9316, "step": 36015 }, { "epoch": 0.04, "learning_rate": 4.984099468478382e-05, "loss": 0.8498, "step": 36016 }, { "epoch": 0.04, "learning_rate": 4.9840985839770404e-05, "loss": 0.7091, "step": 36017 }, { "epoch": 0.04, "learning_rate": 4.9840976994511765e-05, "loss": 1.1869, "step": 36018 }, { "epoch": 0.04, "learning_rate": 4.984096814900791e-05, "loss": 0.8455, "step": 36019 }, { "epoch": 0.04, "learning_rate": 4.984095930325883e-05, "loss": 0.898, "step": 36020 }, { "epoch": 0.04, "learning_rate": 4.984095045726453e-05, "loss": 0.9959, "step": 36021 }, { "epoch": 0.04, "learning_rate": 4.9840941611025016e-05, "loss": 0.9665, "step": 36022 }, { "epoch": 0.04, "learning_rate": 4.984093276454028e-05, "loss": 0.8556, "step": 36023 }, { "epoch": 0.04, "learning_rate": 4.9840923917810324e-05, "loss": 1.0557, "step": 36024 }, { "epoch": 0.04, "learning_rate": 4.984091507083515e-05, "loss": 1.124, "step": 36025 }, { "epoch": 0.04, "learning_rate": 4.9840906223614756e-05, "loss": 1.0641, "step": 36026 }, { "epoch": 0.04, "learning_rate": 4.984089737614914e-05, "loss": 0.8255, "step": 36027 }, { "epoch": 0.04, "learning_rate": 4.984088852843831e-05, "loss": 0.9285, "step": 36028 }, { "epoch": 0.04, "learning_rate": 4.984087968048226e-05, "loss": 0.9995, "step": 36029 }, { "epoch": 0.04, "learning_rate": 4.984087083228099e-05, "loss": 0.8917, "step": 36030 }, { "epoch": 0.04, "learning_rate": 4.984086198383451e-05, "loss": 0.6413, "step": 36031 }, { "epoch": 0.04, "learning_rate": 4.9840853135142796e-05, "loss": 0.49, "step": 36032 }, { "epoch": 0.04, "learning_rate": 4.984084428620587e-05, "loss": 0.5568, "step": 36033 }, { "epoch": 0.04, "learning_rate": 4.984083543702373e-05, "loss": 0.9611, "step": 36034 }, { "epoch": 0.04, "learning_rate": 4.984082658759637e-05, "loss": 1.3466, "step": 36035 }, { "epoch": 0.04, "learning_rate": 4.984081773792379e-05, "loss": 1.2394, "step": 36036 }, { "epoch": 0.04, "learning_rate": 4.9840808888006e-05, "loss": 0.9979, "step": 36037 }, { "epoch": 0.04, "learning_rate": 4.984080003784298e-05, "loss": 0.7268, "step": 36038 }, { "epoch": 0.04, "learning_rate": 4.9840791187434745e-05, "loss": 0.4704, "step": 36039 }, { "epoch": 0.04, "learning_rate": 4.9840782336781285e-05, "loss": 0.768, "step": 36040 }, { "epoch": 0.04, "learning_rate": 4.984077348588263e-05, "loss": 1.8639, "step": 36041 }, { "epoch": 0.04, "learning_rate": 4.9840764634738736e-05, "loss": 1.4552, "step": 36042 }, { "epoch": 0.04, "learning_rate": 4.984075578334964e-05, "loss": 1.4004, "step": 36043 }, { "epoch": 0.04, "learning_rate": 4.984074693171531e-05, "loss": 0.6935, "step": 36044 }, { "epoch": 0.04, "learning_rate": 4.984073807983577e-05, "loss": 0.9069, "step": 36045 }, { "epoch": 0.04, "learning_rate": 4.9840729227711016e-05, "loss": 1.0758, "step": 36046 }, { "epoch": 0.04, "learning_rate": 4.984072037534104e-05, "loss": 0.738, "step": 36047 }, { "epoch": 0.04, "learning_rate": 4.9840711522725853e-05, "loss": 1.1073, "step": 36048 }, { "epoch": 0.04, "learning_rate": 4.9840702669865446e-05, "loss": 1.2305, "step": 36049 }, { "epoch": 0.04, "learning_rate": 4.984069381675982e-05, "loss": 1.3163, "step": 36050 }, { "epoch": 0.04, "learning_rate": 4.984068496340898e-05, "loss": 1.3099, "step": 36051 }, { "epoch": 0.04, "learning_rate": 4.984067610981292e-05, "loss": 0.8676, "step": 36052 }, { "epoch": 0.04, "learning_rate": 4.9840667255971644e-05, "loss": 1.2094, "step": 36053 }, { "epoch": 0.04, "learning_rate": 4.984065840188515e-05, "loss": 0.8381, "step": 36054 }, { "epoch": 0.04, "learning_rate": 4.984064954755344e-05, "loss": 0.3978, "step": 36055 }, { "epoch": 0.04, "learning_rate": 4.984064069297652e-05, "loss": 0.2875, "step": 36056 }, { "epoch": 0.04, "learning_rate": 4.9840631838154385e-05, "loss": 0.7138, "step": 36057 }, { "epoch": 0.04, "learning_rate": 4.9840622983087024e-05, "loss": 0.4342, "step": 36058 }, { "epoch": 0.04, "learning_rate": 4.984061412777445e-05, "loss": 0.1264, "step": 36059 }, { "epoch": 0.04, "learning_rate": 4.9840605272216656e-05, "loss": 0.903, "step": 36060 }, { "epoch": 0.04, "learning_rate": 4.9840596416413656e-05, "loss": 1.0193, "step": 36061 }, { "epoch": 0.04, "learning_rate": 4.984058756036544e-05, "loss": 0.9833, "step": 36062 }, { "epoch": 0.04, "learning_rate": 4.984057870407199e-05, "loss": 1.0391, "step": 36063 }, { "epoch": 0.04, "learning_rate": 4.984056984753335e-05, "loss": 1.2619, "step": 36064 }, { "epoch": 0.04, "learning_rate": 4.984056099074948e-05, "loss": 0.7886, "step": 36065 }, { "epoch": 0.04, "learning_rate": 4.9840552133720394e-05, "loss": 0.9808, "step": 36066 }, { "epoch": 0.04, "learning_rate": 4.984054327644609e-05, "loss": 0.4059, "step": 36067 }, { "epoch": 0.04, "learning_rate": 4.9840534418926576e-05, "loss": 0.8065, "step": 36068 }, { "epoch": 0.04, "learning_rate": 4.984052556116184e-05, "loss": 1.2725, "step": 36069 }, { "epoch": 0.04, "learning_rate": 4.98405167031519e-05, "loss": 1.1076, "step": 36070 }, { "epoch": 0.04, "learning_rate": 4.984050784489674e-05, "loss": 1.0758, "step": 36071 }, { "epoch": 0.04, "learning_rate": 4.984049898639637e-05, "loss": 0.8058, "step": 36072 }, { "epoch": 0.04, "learning_rate": 4.9840490127650776e-05, "loss": 0.9706, "step": 36073 }, { "epoch": 0.04, "learning_rate": 4.984048126865997e-05, "loss": 0.8857, "step": 36074 }, { "epoch": 0.04, "learning_rate": 4.984047240942395e-05, "loss": 1.132, "step": 36075 }, { "epoch": 0.04, "learning_rate": 4.984046354994272e-05, "loss": 1.2272, "step": 36076 }, { "epoch": 0.04, "learning_rate": 4.984045469021627e-05, "loss": 0.9162, "step": 36077 }, { "epoch": 0.04, "learning_rate": 4.9840445830244606e-05, "loss": 0.9859, "step": 36078 }, { "epoch": 0.04, "learning_rate": 4.9840436970027725e-05, "loss": 1.0873, "step": 36079 }, { "epoch": 0.04, "learning_rate": 4.984042810956563e-05, "loss": 1.2033, "step": 36080 }, { "epoch": 0.04, "learning_rate": 4.984041924885833e-05, "loss": 0.9602, "step": 36081 }, { "epoch": 0.04, "learning_rate": 4.9840410387905805e-05, "loss": 1.0285, "step": 36082 }, { "epoch": 0.04, "learning_rate": 4.984040152670807e-05, "loss": 0.9018, "step": 36083 }, { "epoch": 0.04, "learning_rate": 4.984039266526512e-05, "loss": 0.9401, "step": 36084 }, { "epoch": 0.04, "learning_rate": 4.9840383803576964e-05, "loss": 1.018, "step": 36085 }, { "epoch": 0.04, "learning_rate": 4.984037494164358e-05, "loss": 0.8266, "step": 36086 }, { "epoch": 0.04, "learning_rate": 4.9840366079465e-05, "loss": 0.8032, "step": 36087 }, { "epoch": 0.04, "learning_rate": 4.984035721704119e-05, "loss": 0.7684, "step": 36088 }, { "epoch": 0.04, "learning_rate": 4.984034835437217e-05, "loss": 0.9346, "step": 36089 }, { "epoch": 0.04, "learning_rate": 4.984033949145794e-05, "loss": 0.903, "step": 36090 }, { "epoch": 0.04, "learning_rate": 4.984033062829849e-05, "loss": 0.9084, "step": 36091 }, { "epoch": 0.04, "learning_rate": 4.984032176489384e-05, "loss": 0.8013, "step": 36092 }, { "epoch": 0.04, "learning_rate": 4.984031290124397e-05, "loss": 1.2265, "step": 36093 }, { "epoch": 0.04, "learning_rate": 4.984030403734888e-05, "loss": 0.874, "step": 36094 }, { "epoch": 0.04, "learning_rate": 4.9840295173208586e-05, "loss": 1.26, "step": 36095 }, { "epoch": 0.04, "learning_rate": 4.9840286308823076e-05, "loss": 0.7702, "step": 36096 }, { "epoch": 0.04, "learning_rate": 4.984027744419235e-05, "loss": 0.4496, "step": 36097 }, { "epoch": 0.04, "learning_rate": 4.984026857931642e-05, "loss": 0.9595, "step": 36098 }, { "epoch": 0.04, "learning_rate": 4.984025971419527e-05, "loss": 1.0887, "step": 36099 }, { "epoch": 0.04, "learning_rate": 4.984025084882891e-05, "loss": 0.8328, "step": 36100 }, { "epoch": 0.04, "learning_rate": 4.984024198321734e-05, "loss": 0.5845, "step": 36101 }, { "epoch": 0.04, "learning_rate": 4.984023311736055e-05, "loss": 1.1231, "step": 36102 }, { "epoch": 0.04, "learning_rate": 4.984022425125855e-05, "loss": 1.039, "step": 36103 }, { "epoch": 0.04, "learning_rate": 4.984021538491134e-05, "loss": 1.3474, "step": 36104 }, { "epoch": 0.04, "learning_rate": 4.984020651831892e-05, "loss": 0.9911, "step": 36105 }, { "epoch": 0.04, "learning_rate": 4.9840197651481286e-05, "loss": 1.247, "step": 36106 }, { "epoch": 0.04, "learning_rate": 4.984018878439844e-05, "loss": 1.1287, "step": 36107 }, { "epoch": 0.04, "learning_rate": 4.984017991707038e-05, "loss": 1.1348, "step": 36108 }, { "epoch": 0.04, "learning_rate": 4.984017104949711e-05, "loss": 1.0161, "step": 36109 }, { "epoch": 0.04, "learning_rate": 4.984016218167863e-05, "loss": 1.1611, "step": 36110 }, { "epoch": 0.04, "learning_rate": 4.9840153313614934e-05, "loss": 0.9976, "step": 36111 }, { "epoch": 0.04, "learning_rate": 4.9840144445306035e-05, "loss": 1.1252, "step": 36112 }, { "epoch": 0.04, "learning_rate": 4.9840135576751916e-05, "loss": 1.1123, "step": 36113 }, { "epoch": 0.04, "learning_rate": 4.984012670795259e-05, "loss": 1.0918, "step": 36114 }, { "epoch": 0.04, "learning_rate": 4.984011783890805e-05, "loss": 0.9481, "step": 36115 }, { "epoch": 0.04, "learning_rate": 4.9840108969618306e-05, "loss": 1.0242, "step": 36116 }, { "epoch": 0.04, "learning_rate": 4.9840100100083344e-05, "loss": 1.0893, "step": 36117 }, { "epoch": 0.04, "learning_rate": 4.984009123030316e-05, "loss": 1.0855, "step": 36118 }, { "epoch": 0.04, "learning_rate": 4.984008236027778e-05, "loss": 0.8959, "step": 36119 }, { "epoch": 0.04, "learning_rate": 4.984007349000719e-05, "loss": 0.8306, "step": 36120 }, { "epoch": 0.04, "learning_rate": 4.984006461949139e-05, "loss": 0.7968, "step": 36121 }, { "epoch": 0.04, "learning_rate": 4.984005574873038e-05, "loss": 0.8668, "step": 36122 }, { "epoch": 0.04, "learning_rate": 4.9840046877724146e-05, "loss": 0.6239, "step": 36123 }, { "epoch": 0.04, "learning_rate": 4.984003800647271e-05, "loss": 0.7585, "step": 36124 }, { "epoch": 0.04, "learning_rate": 4.9840029134976065e-05, "loss": 1.0639, "step": 36125 }, { "epoch": 0.04, "learning_rate": 4.98400202632342e-05, "loss": 1.1409, "step": 36126 }, { "epoch": 0.04, "learning_rate": 4.9840011391247135e-05, "loss": 0.9677, "step": 36127 }, { "epoch": 0.04, "learning_rate": 4.984000251901486e-05, "loss": 1.9685, "step": 36128 }, { "epoch": 0.04, "learning_rate": 4.9839993646537374e-05, "loss": 1.2319, "step": 36129 }, { "epoch": 0.04, "learning_rate": 4.9839984773814674e-05, "loss": 0.4266, "step": 36130 }, { "epoch": 0.04, "learning_rate": 4.983997590084677e-05, "loss": 0.3192, "step": 36131 }, { "epoch": 0.04, "learning_rate": 4.983996702763365e-05, "loss": 0.2244, "step": 36132 }, { "epoch": 0.04, "learning_rate": 4.983995815417533e-05, "loss": 0.1375, "step": 36133 }, { "epoch": 0.04, "learning_rate": 4.983994928047179e-05, "loss": 0.1324, "step": 36134 }, { "epoch": 0.04, "learning_rate": 4.983994040652305e-05, "loss": 0.5312, "step": 36135 }, { "epoch": 0.04, "learning_rate": 4.9839931532329086e-05, "loss": 0.2103, "step": 36136 }, { "epoch": 0.04, "learning_rate": 4.9839922657889925e-05, "loss": 0.268, "step": 36137 }, { "epoch": 0.04, "learning_rate": 4.983991378320555e-05, "loss": 0.1435, "step": 36138 }, { "epoch": 0.04, "learning_rate": 4.983990490827597e-05, "loss": 0.2541, "step": 36139 }, { "epoch": 0.04, "learning_rate": 4.983989603310117e-05, "loss": 0.2068, "step": 36140 }, { "epoch": 0.04, "learning_rate": 4.9839887157681175e-05, "loss": 0.1039, "step": 36141 }, { "epoch": 0.04, "learning_rate": 4.9839878282015964e-05, "loss": 0.2614, "step": 36142 }, { "epoch": 0.04, "learning_rate": 4.983986940610554e-05, "loss": 0.1407, "step": 36143 }, { "epoch": 0.04, "learning_rate": 4.9839860529949925e-05, "loss": 0.1547, "step": 36144 }, { "epoch": 0.04, "learning_rate": 4.983985165354908e-05, "loss": 0.1354, "step": 36145 }, { "epoch": 0.04, "learning_rate": 4.983984277690304e-05, "loss": 0.1374, "step": 36146 }, { "epoch": 0.04, "learning_rate": 4.983983390001179e-05, "loss": 0.6704, "step": 36147 }, { "epoch": 0.04, "learning_rate": 4.983982502287532e-05, "loss": 1.2066, "step": 36148 }, { "epoch": 0.04, "learning_rate": 4.983981614549366e-05, "loss": 0.8632, "step": 36149 }, { "epoch": 0.04, "learning_rate": 4.983980726786678e-05, "loss": 1.1672, "step": 36150 }, { "epoch": 0.04, "learning_rate": 4.983979838999469e-05, "loss": 1.1825, "step": 36151 }, { "epoch": 0.04, "learning_rate": 4.98397895118774e-05, "loss": 1.1686, "step": 36152 }, { "epoch": 0.04, "learning_rate": 4.9839780633514897e-05, "loss": 1.2156, "step": 36153 }, { "epoch": 0.04, "learning_rate": 4.983977175490718e-05, "loss": 1.1535, "step": 36154 }, { "epoch": 0.04, "learning_rate": 4.983976287605427e-05, "loss": 0.9208, "step": 36155 }, { "epoch": 0.04, "learning_rate": 4.983975399695614e-05, "loss": 0.9759, "step": 36156 }, { "epoch": 0.04, "learning_rate": 4.9839745117612815e-05, "loss": 1.2419, "step": 36157 }, { "epoch": 0.04, "learning_rate": 4.983973623802427e-05, "loss": 0.9781, "step": 36158 }, { "epoch": 0.04, "learning_rate": 4.9839727358190525e-05, "loss": 0.8762, "step": 36159 }, { "epoch": 0.04, "learning_rate": 4.983971847811157e-05, "loss": 0.6648, "step": 36160 }, { "epoch": 0.04, "learning_rate": 4.983970959778741e-05, "loss": 0.9013, "step": 36161 }, { "epoch": 0.04, "learning_rate": 4.9839700717218036e-05, "loss": 0.9752, "step": 36162 }, { "epoch": 0.04, "learning_rate": 4.9839691836403455e-05, "loss": 1.05, "step": 36163 }, { "epoch": 0.04, "learning_rate": 4.983968295534368e-05, "loss": 0.8705, "step": 36164 }, { "epoch": 0.04, "learning_rate": 4.983967407403869e-05, "loss": 1.1399, "step": 36165 }, { "epoch": 0.04, "learning_rate": 4.983966519248849e-05, "loss": 1.0621, "step": 36166 }, { "epoch": 0.04, "learning_rate": 4.9839656310693084e-05, "loss": 1.0423, "step": 36167 }, { "epoch": 0.04, "learning_rate": 4.983964742865247e-05, "loss": 0.9834, "step": 36168 }, { "epoch": 0.04, "learning_rate": 4.9839638546366656e-05, "loss": 1.2641, "step": 36169 }, { "epoch": 0.04, "learning_rate": 4.9839629663835634e-05, "loss": 1.0447, "step": 36170 }, { "epoch": 0.04, "learning_rate": 4.9839620781059405e-05, "loss": 1.2176, "step": 36171 }, { "epoch": 0.04, "learning_rate": 4.983961189803797e-05, "loss": 0.7322, "step": 36172 }, { "epoch": 0.04, "learning_rate": 4.983960301477133e-05, "loss": 0.7928, "step": 36173 }, { "epoch": 0.04, "learning_rate": 4.9839594131259485e-05, "loss": 1.1723, "step": 36174 }, { "epoch": 0.04, "learning_rate": 4.9839585247502426e-05, "loss": 1.0021, "step": 36175 }, { "epoch": 0.04, "learning_rate": 4.983957636350016e-05, "loss": 1.071, "step": 36176 }, { "epoch": 0.04, "learning_rate": 4.9839567479252705e-05, "loss": 1.0746, "step": 36177 }, { "epoch": 0.04, "learning_rate": 4.983955859476003e-05, "loss": 0.7886, "step": 36178 }, { "epoch": 0.04, "learning_rate": 4.9839549710022146e-05, "loss": 0.9904, "step": 36179 }, { "epoch": 0.04, "learning_rate": 4.9839540825039065e-05, "loss": 0.9323, "step": 36180 }, { "epoch": 0.04, "learning_rate": 4.983953193981078e-05, "loss": 1.3339, "step": 36181 }, { "epoch": 0.04, "learning_rate": 4.9839523054337285e-05, "loss": 0.7238, "step": 36182 }, { "epoch": 0.04, "learning_rate": 4.9839514168618586e-05, "loss": 1.0126, "step": 36183 }, { "epoch": 0.04, "learning_rate": 4.983950528265468e-05, "loss": 0.814, "step": 36184 }, { "epoch": 0.04, "learning_rate": 4.983949639644557e-05, "loss": 0.9338, "step": 36185 }, { "epoch": 0.04, "learning_rate": 4.9839487509991254e-05, "loss": 2.0373, "step": 36186 }, { "epoch": 0.04, "learning_rate": 4.983947862329174e-05, "loss": 1.7628, "step": 36187 }, { "epoch": 0.04, "learning_rate": 4.983946973634701e-05, "loss": 1.0126, "step": 36188 }, { "epoch": 0.04, "learning_rate": 4.983946084915708e-05, "loss": 0.7795, "step": 36189 }, { "epoch": 0.04, "learning_rate": 4.983945196172195e-05, "loss": 0.7862, "step": 36190 }, { "epoch": 0.04, "learning_rate": 4.983944307404162e-05, "loss": 0.9197, "step": 36191 }, { "epoch": 0.04, "learning_rate": 4.983943418611607e-05, "loss": 1.0472, "step": 36192 }, { "epoch": 0.04, "learning_rate": 4.983942529794532e-05, "loss": 1.2454, "step": 36193 }, { "epoch": 0.04, "learning_rate": 4.983941640952937e-05, "loss": 1.0829, "step": 36194 }, { "epoch": 0.04, "learning_rate": 4.983940752086821e-05, "loss": 1.0528, "step": 36195 }, { "epoch": 0.04, "learning_rate": 4.983939863196185e-05, "loss": 0.9795, "step": 36196 }, { "epoch": 0.04, "learning_rate": 4.9839389742810286e-05, "loss": 0.8911, "step": 36197 }, { "epoch": 0.04, "learning_rate": 4.983938085341352e-05, "loss": 1.1514, "step": 36198 }, { "epoch": 0.04, "learning_rate": 4.9839371963771546e-05, "loss": 1.1321, "step": 36199 }, { "epoch": 0.04, "learning_rate": 4.983936307388437e-05, "loss": 1.0051, "step": 36200 }, { "epoch": 0.04, "learning_rate": 4.9839354183752e-05, "loss": 1.052, "step": 36201 }, { "epoch": 0.04, "learning_rate": 4.983934529337441e-05, "loss": 0.9704, "step": 36202 }, { "epoch": 0.04, "learning_rate": 4.9839336402751624e-05, "loss": 0.7938, "step": 36203 }, { "epoch": 0.04, "learning_rate": 4.983932751188363e-05, "loss": 0.7989, "step": 36204 }, { "epoch": 0.04, "learning_rate": 4.9839318620770435e-05, "loss": 0.8196, "step": 36205 }, { "epoch": 0.04, "learning_rate": 4.983930972941204e-05, "loss": 0.9134, "step": 36206 }, { "epoch": 0.04, "learning_rate": 4.983930083780844e-05, "loss": 0.8086, "step": 36207 }, { "epoch": 0.04, "learning_rate": 4.983929194595964e-05, "loss": 1.1251, "step": 36208 }, { "epoch": 0.04, "learning_rate": 4.983928305386563e-05, "loss": 1.3068, "step": 36209 }, { "epoch": 0.04, "learning_rate": 4.983927416152642e-05, "loss": 0.9173, "step": 36210 }, { "epoch": 0.04, "learning_rate": 4.983926526894201e-05, "loss": 1.0446, "step": 36211 }, { "epoch": 0.04, "learning_rate": 4.98392563761124e-05, "loss": 0.911, "step": 36212 }, { "epoch": 0.04, "learning_rate": 4.983924748303758e-05, "loss": 1.0577, "step": 36213 }, { "epoch": 0.04, "learning_rate": 4.983923858971756e-05, "loss": 0.7771, "step": 36214 }, { "epoch": 0.04, "learning_rate": 4.983922969615233e-05, "loss": 0.7869, "step": 36215 }, { "epoch": 0.04, "learning_rate": 4.9839220802341914e-05, "loss": 0.5327, "step": 36216 }, { "epoch": 0.04, "learning_rate": 4.983921190828628e-05, "loss": 0.866, "step": 36217 }, { "epoch": 0.04, "learning_rate": 4.983920301398545e-05, "loss": 0.756, "step": 36218 }, { "epoch": 0.04, "learning_rate": 4.9839194119439424e-05, "loss": 0.7593, "step": 36219 }, { "epoch": 0.04, "learning_rate": 4.983918522464819e-05, "loss": 0.8556, "step": 36220 }, { "epoch": 0.04, "learning_rate": 4.9839176329611755e-05, "loss": 1.262, "step": 36221 }, { "epoch": 0.04, "learning_rate": 4.9839167434330115e-05, "loss": 0.6669, "step": 36222 }, { "epoch": 0.04, "learning_rate": 4.9839158538803276e-05, "loss": 0.6324, "step": 36223 }, { "epoch": 0.04, "learning_rate": 4.983914964303124e-05, "loss": 0.9579, "step": 36224 }, { "epoch": 0.04, "learning_rate": 4.983914074701399e-05, "loss": 0.6922, "step": 36225 }, { "epoch": 0.04, "learning_rate": 4.9839131850751544e-05, "loss": 0.6631, "step": 36226 }, { "epoch": 0.04, "learning_rate": 4.98391229542439e-05, "loss": 0.7537, "step": 36227 }, { "epoch": 0.04, "learning_rate": 4.9839114057491054e-05, "loss": 0.771, "step": 36228 }, { "epoch": 0.04, "learning_rate": 4.9839105160493007e-05, "loss": 0.8876, "step": 36229 }, { "epoch": 0.04, "learning_rate": 4.983909626324976e-05, "loss": 0.7129, "step": 36230 }, { "epoch": 0.04, "learning_rate": 4.983908736576131e-05, "loss": 0.9229, "step": 36231 }, { "epoch": 0.04, "learning_rate": 4.983907846802766e-05, "loss": 1.1804, "step": 36232 }, { "epoch": 0.04, "learning_rate": 4.98390695700488e-05, "loss": 1.1319, "step": 36233 }, { "epoch": 0.04, "learning_rate": 4.9839060671824744e-05, "loss": 0.9919, "step": 36234 }, { "epoch": 0.04, "learning_rate": 4.9839051773355495e-05, "loss": 1.1085, "step": 36235 }, { "epoch": 0.04, "learning_rate": 4.983904287464104e-05, "loss": 1.011, "step": 36236 }, { "epoch": 0.04, "learning_rate": 4.983903397568138e-05, "loss": 1.1694, "step": 36237 }, { "epoch": 0.04, "learning_rate": 4.983902507647653e-05, "loss": 0.82, "step": 36238 }, { "epoch": 0.04, "learning_rate": 4.9839016177026476e-05, "loss": 1.0341, "step": 36239 }, { "epoch": 0.04, "learning_rate": 4.983900727733122e-05, "loss": 1.1019, "step": 36240 }, { "epoch": 0.04, "learning_rate": 4.9838998377390755e-05, "loss": 1.6458, "step": 36241 }, { "epoch": 0.04, "learning_rate": 4.98389894772051e-05, "loss": 1.0425, "step": 36242 }, { "epoch": 0.04, "learning_rate": 4.983898057677424e-05, "loss": 1.2461, "step": 36243 }, { "epoch": 0.04, "learning_rate": 4.983897167609819e-05, "loss": 1.036, "step": 36244 }, { "epoch": 0.04, "learning_rate": 4.983896277517693e-05, "loss": 0.6491, "step": 36245 }, { "epoch": 0.04, "learning_rate": 4.983895387401047e-05, "loss": 0.8753, "step": 36246 }, { "epoch": 0.04, "learning_rate": 4.983894497259882e-05, "loss": 1.008, "step": 36247 }, { "epoch": 0.04, "learning_rate": 4.9838936070941964e-05, "loss": 0.582, "step": 36248 }, { "epoch": 0.04, "learning_rate": 4.983892716903991e-05, "loss": 0.8545, "step": 36249 }, { "epoch": 0.04, "learning_rate": 4.983891826689265e-05, "loss": 1.2027, "step": 36250 }, { "epoch": 0.04, "learning_rate": 4.98389093645002e-05, "loss": 0.8049, "step": 36251 }, { "epoch": 0.04, "learning_rate": 4.9838900461862544e-05, "loss": 0.9199, "step": 36252 }, { "epoch": 0.04, "learning_rate": 4.983889155897969e-05, "loss": 0.9855, "step": 36253 }, { "epoch": 0.04, "learning_rate": 4.983888265585164e-05, "loss": 1.1091, "step": 36254 }, { "epoch": 0.04, "learning_rate": 4.9838873752478396e-05, "loss": 0.9902, "step": 36255 }, { "epoch": 0.04, "learning_rate": 4.9838864848859946e-05, "loss": 1.0596, "step": 36256 }, { "epoch": 0.04, "learning_rate": 4.98388559449963e-05, "loss": 1.6448, "step": 36257 }, { "epoch": 0.04, "learning_rate": 4.983884704088745e-05, "loss": 1.1522, "step": 36258 }, { "epoch": 0.04, "learning_rate": 4.983883813653341e-05, "loss": 1.0655, "step": 36259 }, { "epoch": 0.04, "learning_rate": 4.9838829231934156e-05, "loss": 1.077, "step": 36260 }, { "epoch": 0.04, "learning_rate": 4.983882032708972e-05, "loss": 0.8504, "step": 36261 }, { "epoch": 0.04, "learning_rate": 4.983881142200007e-05, "loss": 1.063, "step": 36262 }, { "epoch": 0.04, "learning_rate": 4.983880251666524e-05, "loss": 1.0273, "step": 36263 }, { "epoch": 0.04, "learning_rate": 4.98387936110852e-05, "loss": 0.7588, "step": 36264 }, { "epoch": 0.04, "learning_rate": 4.9838784705259965e-05, "loss": 1.6055, "step": 36265 }, { "epoch": 0.04, "learning_rate": 4.983877579918953e-05, "loss": 0.9031, "step": 36266 }, { "epoch": 0.04, "learning_rate": 4.98387668928739e-05, "loss": 1.0228, "step": 36267 }, { "epoch": 0.04, "learning_rate": 4.9838757986313065e-05, "loss": 1.1927, "step": 36268 }, { "epoch": 0.04, "learning_rate": 4.983874907950704e-05, "loss": 0.9036, "step": 36269 }, { "epoch": 0.04, "learning_rate": 4.983874017245582e-05, "loss": 1.3422, "step": 36270 }, { "epoch": 0.04, "learning_rate": 4.9838731265159394e-05, "loss": 0.9185, "step": 36271 }, { "epoch": 0.04, "learning_rate": 4.983872235761777e-05, "loss": 0.9251, "step": 36272 }, { "epoch": 0.04, "learning_rate": 4.983871344983096e-05, "loss": 1.1865, "step": 36273 }, { "epoch": 0.04, "learning_rate": 4.9838704541798945e-05, "loss": 0.7768, "step": 36274 }, { "epoch": 0.04, "learning_rate": 4.983869563352173e-05, "loss": 1.0039, "step": 36275 }, { "epoch": 0.04, "learning_rate": 4.983868672499932e-05, "loss": 0.9592, "step": 36276 }, { "epoch": 0.04, "learning_rate": 4.983867781623172e-05, "loss": 1.3468, "step": 36277 }, { "epoch": 0.04, "learning_rate": 4.9838668907218914e-05, "loss": 0.5559, "step": 36278 }, { "epoch": 0.04, "learning_rate": 4.983865999796091e-05, "loss": 0.8386, "step": 36279 }, { "epoch": 0.04, "learning_rate": 4.9838651088457725e-05, "loss": 1.0412, "step": 36280 }, { "epoch": 0.04, "learning_rate": 4.9838642178709324e-05, "loss": 0.8396, "step": 36281 }, { "epoch": 0.04, "learning_rate": 4.983863326871574e-05, "loss": 0.9549, "step": 36282 }, { "epoch": 0.04, "learning_rate": 4.983862435847695e-05, "loss": 0.9385, "step": 36283 }, { "epoch": 0.04, "learning_rate": 4.983861544799296e-05, "loss": 0.9189, "step": 36284 }, { "epoch": 0.04, "learning_rate": 4.983860653726379e-05, "loss": 0.9914, "step": 36285 }, { "epoch": 0.04, "learning_rate": 4.983859762628941e-05, "loss": 1.1612, "step": 36286 }, { "epoch": 0.04, "learning_rate": 4.9838588715069844e-05, "loss": 1.3049, "step": 36287 }, { "epoch": 0.04, "learning_rate": 4.983857980360507e-05, "loss": 0.8165, "step": 36288 }, { "epoch": 0.04, "learning_rate": 4.983857089189511e-05, "loss": 1.2028, "step": 36289 }, { "epoch": 0.04, "learning_rate": 4.983856197993995e-05, "loss": 1.2213, "step": 36290 }, { "epoch": 0.04, "learning_rate": 4.9838553067739595e-05, "loss": 1.2799, "step": 36291 }, { "epoch": 0.04, "learning_rate": 4.9838544155294046e-05, "loss": 1.0248, "step": 36292 }, { "epoch": 0.04, "learning_rate": 4.9838535242603304e-05, "loss": 1.1953, "step": 36293 }, { "epoch": 0.04, "learning_rate": 4.9838526329667356e-05, "loss": 1.1197, "step": 36294 }, { "epoch": 0.04, "learning_rate": 4.9838517416486216e-05, "loss": 0.8921, "step": 36295 }, { "epoch": 0.04, "learning_rate": 4.983850850305989e-05, "loss": 0.8819, "step": 36296 }, { "epoch": 0.04, "learning_rate": 4.983849958938835e-05, "loss": 1.1251, "step": 36297 }, { "epoch": 0.04, "learning_rate": 4.9838490675471635e-05, "loss": 1.0817, "step": 36298 }, { "epoch": 0.04, "learning_rate": 4.983848176130971e-05, "loss": 0.7969, "step": 36299 }, { "epoch": 0.04, "learning_rate": 4.98384728469026e-05, "loss": 1.1206, "step": 36300 }, { "epoch": 0.04, "learning_rate": 4.983846393225029e-05, "loss": 1.3017, "step": 36301 }, { "epoch": 0.04, "learning_rate": 4.983845501735279e-05, "loss": 1.1686, "step": 36302 }, { "epoch": 0.04, "learning_rate": 4.983844610221009e-05, "loss": 1.0222, "step": 36303 }, { "epoch": 0.04, "learning_rate": 4.98384371868222e-05, "loss": 1.4083, "step": 36304 }, { "epoch": 0.04, "learning_rate": 4.983842827118911e-05, "loss": 0.8972, "step": 36305 }, { "epoch": 0.04, "learning_rate": 4.9838419355310825e-05, "loss": 1.0247, "step": 36306 }, { "epoch": 0.04, "learning_rate": 4.983841043918736e-05, "loss": 1.0061, "step": 36307 }, { "epoch": 0.04, "learning_rate": 4.9838401522818676e-05, "loss": 0.9431, "step": 36308 }, { "epoch": 0.04, "learning_rate": 4.983839260620482e-05, "loss": 0.9829, "step": 36309 }, { "epoch": 0.04, "learning_rate": 4.983838368934576e-05, "loss": 0.6701, "step": 36310 }, { "epoch": 0.04, "learning_rate": 4.9838374772241506e-05, "loss": 1.2546, "step": 36311 }, { "epoch": 0.04, "learning_rate": 4.9838365854892056e-05, "loss": 1.101, "step": 36312 }, { "epoch": 0.04, "learning_rate": 4.983835693729741e-05, "loss": 0.9267, "step": 36313 }, { "epoch": 0.04, "learning_rate": 4.983834801945758e-05, "loss": 0.9814, "step": 36314 }, { "epoch": 0.04, "learning_rate": 4.983833910137256e-05, "loss": 0.7914, "step": 36315 }, { "epoch": 0.04, "learning_rate": 4.983833018304233e-05, "loss": 0.8881, "step": 36316 }, { "epoch": 0.04, "learning_rate": 4.983832126446691e-05, "loss": 1.0799, "step": 36317 }, { "epoch": 0.04, "learning_rate": 4.983831234564631e-05, "loss": 1.0318, "step": 36318 }, { "epoch": 0.04, "learning_rate": 4.98383034265805e-05, "loss": 1.089, "step": 36319 }, { "epoch": 0.04, "learning_rate": 4.98382945072695e-05, "loss": 1.053, "step": 36320 }, { "epoch": 0.04, "learning_rate": 4.9838285587713315e-05, "loss": 0.9615, "step": 36321 }, { "epoch": 0.04, "learning_rate": 4.9838276667911935e-05, "loss": 1.0193, "step": 36322 }, { "epoch": 0.04, "learning_rate": 4.983826774786536e-05, "loss": 1.1666, "step": 36323 }, { "epoch": 0.04, "learning_rate": 4.983825882757359e-05, "loss": 0.9801, "step": 36324 }, { "epoch": 0.04, "learning_rate": 4.983824990703663e-05, "loss": 1.7853, "step": 36325 }, { "epoch": 0.04, "learning_rate": 4.9838240986254484e-05, "loss": 1.2707, "step": 36326 }, { "epoch": 0.04, "learning_rate": 4.983823206522713e-05, "loss": 0.8186, "step": 36327 }, { "epoch": 0.04, "learning_rate": 4.98382231439546e-05, "loss": 0.7131, "step": 36328 }, { "epoch": 0.04, "learning_rate": 4.983821422243686e-05, "loss": 1.1457, "step": 36329 }, { "epoch": 0.04, "learning_rate": 4.983820530067394e-05, "loss": 1.2189, "step": 36330 }, { "epoch": 0.04, "learning_rate": 4.9838196378665824e-05, "loss": 1.0234, "step": 36331 }, { "epoch": 0.04, "learning_rate": 4.9838187456412514e-05, "loss": 1.3231, "step": 36332 }, { "epoch": 0.04, "learning_rate": 4.983817853391402e-05, "loss": 1.2075, "step": 36333 }, { "epoch": 0.04, "learning_rate": 4.983816961117033e-05, "loss": 1.301, "step": 36334 }, { "epoch": 0.04, "learning_rate": 4.983816068818144e-05, "loss": 1.1963, "step": 36335 }, { "epoch": 0.04, "learning_rate": 4.983815176494736e-05, "loss": 1.1028, "step": 36336 }, { "epoch": 0.04, "learning_rate": 4.983814284146809e-05, "loss": 0.6045, "step": 36337 }, { "epoch": 0.04, "learning_rate": 4.9838133917743634e-05, "loss": 0.6063, "step": 36338 }, { "epoch": 0.04, "learning_rate": 4.9838124993773984e-05, "loss": 1.2274, "step": 36339 }, { "epoch": 0.04, "learning_rate": 4.983811606955914e-05, "loss": 1.1496, "step": 36340 }, { "epoch": 0.04, "learning_rate": 4.983810714509911e-05, "loss": 1.3307, "step": 36341 }, { "epoch": 0.04, "learning_rate": 4.983809822039388e-05, "loss": 0.9699, "step": 36342 }, { "epoch": 0.04, "learning_rate": 4.983808929544347e-05, "loss": 0.8666, "step": 36343 }, { "epoch": 0.04, "learning_rate": 4.9838080370247864e-05, "loss": 1.3348, "step": 36344 }, { "epoch": 0.04, "learning_rate": 4.983807144480706e-05, "loss": 0.97, "step": 36345 }, { "epoch": 0.04, "learning_rate": 4.983806251912107e-05, "loss": 1.6129, "step": 36346 }, { "epoch": 0.04, "learning_rate": 4.983805359318989e-05, "loss": 1.3962, "step": 36347 }, { "epoch": 0.04, "learning_rate": 4.9838044667013516e-05, "loss": 1.0557, "step": 36348 }, { "epoch": 0.04, "learning_rate": 4.983803574059196e-05, "loss": 0.4388, "step": 36349 }, { "epoch": 0.04, "learning_rate": 4.9838026813925205e-05, "loss": 0.8511, "step": 36350 }, { "epoch": 0.04, "learning_rate": 4.983801788701326e-05, "loss": 0.7897, "step": 36351 }, { "epoch": 0.04, "learning_rate": 4.983800895985613e-05, "loss": 0.868, "step": 36352 }, { "epoch": 0.04, "learning_rate": 4.98380000324538e-05, "loss": 0.8487, "step": 36353 }, { "epoch": 0.04, "learning_rate": 4.9837991104806295e-05, "loss": 1.1068, "step": 36354 }, { "epoch": 0.04, "learning_rate": 4.983798217691359e-05, "loss": 1.1727, "step": 36355 }, { "epoch": 0.04, "learning_rate": 4.9837973248775697e-05, "loss": 0.6174, "step": 36356 }, { "epoch": 0.04, "learning_rate": 4.9837964320392605e-05, "loss": 0.9871, "step": 36357 }, { "epoch": 0.04, "learning_rate": 4.9837955391764335e-05, "loss": 0.9151, "step": 36358 }, { "epoch": 0.04, "learning_rate": 4.9837946462890866e-05, "loss": 1.1448, "step": 36359 }, { "epoch": 0.04, "learning_rate": 4.983793753377222e-05, "loss": 0.7535, "step": 36360 }, { "epoch": 0.04, "learning_rate": 4.9837928604408365e-05, "loss": 1.0201, "step": 36361 }, { "epoch": 0.04, "learning_rate": 4.983791967479934e-05, "loss": 1.0167, "step": 36362 }, { "epoch": 0.04, "learning_rate": 4.9837910744945114e-05, "loss": 0.8897, "step": 36363 }, { "epoch": 0.04, "learning_rate": 4.98379018148457e-05, "loss": 0.6107, "step": 36364 }, { "epoch": 0.04, "learning_rate": 4.98378928845011e-05, "loss": 0.4304, "step": 36365 }, { "epoch": 0.04, "learning_rate": 4.983788395391131e-05, "loss": 0.6664, "step": 36366 }, { "epoch": 0.04, "learning_rate": 4.983787502307633e-05, "loss": 0.9567, "step": 36367 }, { "epoch": 0.04, "learning_rate": 4.983786609199616e-05, "loss": 1.0594, "step": 36368 }, { "epoch": 0.04, "learning_rate": 4.98378571606708e-05, "loss": 1.206, "step": 36369 }, { "epoch": 0.04, "learning_rate": 4.983784822910026e-05, "loss": 1.1449, "step": 36370 }, { "epoch": 0.04, "learning_rate": 4.983783929728452e-05, "loss": 1.1474, "step": 36371 }, { "epoch": 0.04, "learning_rate": 4.98378303652236e-05, "loss": 1.2334, "step": 36372 }, { "epoch": 0.04, "learning_rate": 4.9837821432917486e-05, "loss": 0.8691, "step": 36373 }, { "epoch": 0.04, "learning_rate": 4.983781250036618e-05, "loss": 1.1021, "step": 36374 }, { "epoch": 0.04, "learning_rate": 4.9837803567569694e-05, "loss": 0.8225, "step": 36375 }, { "epoch": 0.04, "learning_rate": 4.983779463452801e-05, "loss": 1.1686, "step": 36376 }, { "epoch": 0.04, "learning_rate": 4.9837785701241147e-05, "loss": 1.0152, "step": 36377 }, { "epoch": 0.04, "learning_rate": 4.983777676770909e-05, "loss": 1.1812, "step": 36378 }, { "epoch": 0.04, "learning_rate": 4.983776783393185e-05, "loss": 1.2314, "step": 36379 }, { "epoch": 0.04, "learning_rate": 4.9837758899909424e-05, "loss": 1.1869, "step": 36380 }, { "epoch": 0.04, "learning_rate": 4.98377499656418e-05, "loss": 0.9615, "step": 36381 }, { "epoch": 0.04, "learning_rate": 4.9837741031129e-05, "loss": 1.1581, "step": 36382 }, { "epoch": 0.04, "learning_rate": 4.9837732096371004e-05, "loss": 1.0246, "step": 36383 }, { "epoch": 0.04, "learning_rate": 4.983772316136782e-05, "loss": 1.3187, "step": 36384 }, { "epoch": 0.04, "learning_rate": 4.9837714226119454e-05, "loss": 1.2377, "step": 36385 }, { "epoch": 0.04, "learning_rate": 4.9837705290625894e-05, "loss": 1.0013, "step": 36386 }, { "epoch": 0.04, "learning_rate": 4.9837696354887155e-05, "loss": 1.2781, "step": 36387 }, { "epoch": 0.04, "learning_rate": 4.9837687418903224e-05, "loss": 0.8476, "step": 36388 }, { "epoch": 0.04, "learning_rate": 4.983767848267411e-05, "loss": 0.7404, "step": 36389 }, { "epoch": 0.04, "learning_rate": 4.98376695461998e-05, "loss": 0.8634, "step": 36390 }, { "epoch": 0.04, "learning_rate": 4.983766060948031e-05, "loss": 1.0988, "step": 36391 }, { "epoch": 0.04, "learning_rate": 4.983765167251563e-05, "loss": 1.3475, "step": 36392 }, { "epoch": 0.04, "learning_rate": 4.9837642735305765e-05, "loss": 1.2042, "step": 36393 }, { "epoch": 0.04, "learning_rate": 4.9837633797850707e-05, "loss": 0.8341, "step": 36394 }, { "epoch": 0.04, "learning_rate": 4.983762486015047e-05, "loss": 1.2152, "step": 36395 }, { "epoch": 0.04, "learning_rate": 4.983761592220504e-05, "loss": 0.7966, "step": 36396 }, { "epoch": 0.04, "learning_rate": 4.9837606984014426e-05, "loss": 0.3903, "step": 36397 }, { "epoch": 0.04, "learning_rate": 4.983759804557863e-05, "loss": 0.4893, "step": 36398 }, { "epoch": 0.04, "learning_rate": 4.983758910689765e-05, "loss": 0.8536, "step": 36399 }, { "epoch": 0.04, "learning_rate": 4.9837580167971476e-05, "loss": 1.0659, "step": 36400 }, { "epoch": 0.04, "learning_rate": 4.983757122880012e-05, "loss": 0.5378, "step": 36401 }, { "epoch": 0.04, "learning_rate": 4.9837562289383577e-05, "loss": 1.0712, "step": 36402 }, { "epoch": 0.04, "learning_rate": 4.983755334972184e-05, "loss": 1.2348, "step": 36403 }, { "epoch": 0.04, "learning_rate": 4.983754440981493e-05, "loss": 0.8523, "step": 36404 }, { "epoch": 0.04, "learning_rate": 4.983753546966283e-05, "loss": 0.6551, "step": 36405 }, { "epoch": 0.04, "learning_rate": 4.9837526529265545e-05, "loss": 0.9439, "step": 36406 }, { "epoch": 0.04, "learning_rate": 4.983751758862307e-05, "loss": 1.2621, "step": 36407 }, { "epoch": 0.04, "learning_rate": 4.983750864773541e-05, "loss": 1.1314, "step": 36408 }, { "epoch": 0.04, "learning_rate": 4.9837499706602565e-05, "loss": 1.015, "step": 36409 }, { "epoch": 0.04, "learning_rate": 4.983749076522454e-05, "loss": 0.9599, "step": 36410 }, { "epoch": 0.04, "learning_rate": 4.9837481823601326e-05, "loss": 0.9688, "step": 36411 }, { "epoch": 0.04, "learning_rate": 4.983747288173293e-05, "loss": 1.2966, "step": 36412 }, { "epoch": 0.04, "learning_rate": 4.9837463939619344e-05, "loss": 0.9723, "step": 36413 }, { "epoch": 0.04, "learning_rate": 4.983745499726057e-05, "loss": 1.5354, "step": 36414 }, { "epoch": 0.04, "learning_rate": 4.9837446054656614e-05, "loss": 0.9238, "step": 36415 }, { "epoch": 0.04, "learning_rate": 4.983743711180748e-05, "loss": 1.1238, "step": 36416 }, { "epoch": 0.04, "learning_rate": 4.983742816871315e-05, "loss": 0.5705, "step": 36417 }, { "epoch": 0.04, "learning_rate": 4.983741922537365e-05, "loss": 0.9367, "step": 36418 }, { "epoch": 0.04, "learning_rate": 4.9837410281788955e-05, "loss": 1.14, "step": 36419 }, { "epoch": 0.04, "learning_rate": 4.983740133795908e-05, "loss": 0.6573, "step": 36420 }, { "epoch": 0.04, "learning_rate": 4.983739239388402e-05, "loss": 0.9054, "step": 36421 }, { "epoch": 0.04, "learning_rate": 4.9837383449563765e-05, "loss": 0.9611, "step": 36422 }, { "epoch": 0.04, "learning_rate": 4.983737450499834e-05, "loss": 0.5259, "step": 36423 }, { "epoch": 0.04, "learning_rate": 4.983736556018773e-05, "loss": 0.5966, "step": 36424 }, { "epoch": 0.04, "learning_rate": 4.9837356615131926e-05, "loss": 0.5382, "step": 36425 }, { "epoch": 0.04, "learning_rate": 4.9837347669830944e-05, "loss": 0.2278, "step": 36426 }, { "epoch": 0.04, "learning_rate": 4.983733872428478e-05, "loss": 0.1921, "step": 36427 }, { "epoch": 0.04, "learning_rate": 4.983732977849343e-05, "loss": 0.2317, "step": 36428 }, { "epoch": 0.04, "learning_rate": 4.98373208324569e-05, "loss": 0.3406, "step": 36429 }, { "epoch": 0.04, "learning_rate": 4.983731188617518e-05, "loss": 0.4127, "step": 36430 }, { "epoch": 0.04, "learning_rate": 4.983730293964828e-05, "loss": 1.0662, "step": 36431 }, { "epoch": 0.04, "learning_rate": 4.98372939928762e-05, "loss": 1.645, "step": 36432 }, { "epoch": 0.04, "learning_rate": 4.983728504585893e-05, "loss": 1.646, "step": 36433 }, { "epoch": 0.04, "learning_rate": 4.983727609859648e-05, "loss": 1.601, "step": 36434 }, { "epoch": 0.04, "learning_rate": 4.9837267151088846e-05, "loss": 1.3013, "step": 36435 }, { "epoch": 0.04, "learning_rate": 4.983725820333603e-05, "loss": 1.3478, "step": 36436 }, { "epoch": 0.04, "learning_rate": 4.983724925533803e-05, "loss": 1.2019, "step": 36437 }, { "epoch": 0.04, "learning_rate": 4.9837240307094845e-05, "loss": 0.9599, "step": 36438 }, { "epoch": 0.04, "learning_rate": 4.9837231358606485e-05, "loss": 1.0109, "step": 36439 }, { "epoch": 0.04, "learning_rate": 4.983722240987293e-05, "loss": 0.796, "step": 36440 }, { "epoch": 0.04, "learning_rate": 4.983721346089421e-05, "loss": 1.2218, "step": 36441 }, { "epoch": 0.04, "learning_rate": 4.983720451167029e-05, "loss": 1.2724, "step": 36442 }, { "epoch": 0.04, "learning_rate": 4.98371955622012e-05, "loss": 1.2245, "step": 36443 }, { "epoch": 0.04, "learning_rate": 4.983718661248692e-05, "loss": 1.6282, "step": 36444 }, { "epoch": 0.04, "learning_rate": 4.983717766252746e-05, "loss": 1.6637, "step": 36445 }, { "epoch": 0.04, "learning_rate": 4.9837168712322816e-05, "loss": 1.2517, "step": 36446 }, { "epoch": 0.04, "learning_rate": 4.9837159761873e-05, "loss": 1.1001, "step": 36447 }, { "epoch": 0.04, "learning_rate": 4.983715081117799e-05, "loss": 1.3622, "step": 36448 }, { "epoch": 0.04, "learning_rate": 4.98371418602378e-05, "loss": 0.8839, "step": 36449 }, { "epoch": 0.04, "learning_rate": 4.9837132909052434e-05, "loss": 0.824, "step": 36450 }, { "epoch": 0.04, "learning_rate": 4.983712395762188e-05, "loss": 0.7723, "step": 36451 }, { "epoch": 0.04, "learning_rate": 4.983711500594614e-05, "loss": 0.8147, "step": 36452 }, { "epoch": 0.04, "learning_rate": 4.983710605402523e-05, "loss": 1.0763, "step": 36453 }, { "epoch": 0.04, "learning_rate": 4.983709710185914e-05, "loss": 1.1142, "step": 36454 }, { "epoch": 0.04, "learning_rate": 4.983708814944786e-05, "loss": 0.9714, "step": 36455 }, { "epoch": 0.04, "learning_rate": 4.98370791967914e-05, "loss": 1.142, "step": 36456 }, { "epoch": 0.04, "learning_rate": 4.983707024388976e-05, "loss": 0.5532, "step": 36457 }, { "epoch": 0.04, "learning_rate": 4.9837061290742945e-05, "loss": 1.1034, "step": 36458 }, { "epoch": 0.04, "learning_rate": 4.983705233735094e-05, "loss": 0.9163, "step": 36459 }, { "epoch": 0.04, "learning_rate": 4.9837043383713753e-05, "loss": 0.756, "step": 36460 }, { "epoch": 0.04, "learning_rate": 4.98370344298314e-05, "loss": 1.165, "step": 36461 }, { "epoch": 0.04, "learning_rate": 4.9837025475703846e-05, "loss": 1.0016, "step": 36462 }, { "epoch": 0.04, "learning_rate": 4.983701652133113e-05, "loss": 1.2015, "step": 36463 }, { "epoch": 0.04, "learning_rate": 4.9837007566713224e-05, "loss": 0.7688, "step": 36464 }, { "epoch": 0.04, "learning_rate": 4.9836998611850135e-05, "loss": 1.1314, "step": 36465 }, { "epoch": 0.04, "learning_rate": 4.983698965674187e-05, "loss": 1.0174, "step": 36466 }, { "epoch": 0.04, "learning_rate": 4.9836980701388427e-05, "loss": 1.095, "step": 36467 }, { "epoch": 0.04, "learning_rate": 4.9836971745789794e-05, "loss": 1.0393, "step": 36468 }, { "epoch": 0.04, "learning_rate": 4.983696278994598e-05, "loss": 0.7831, "step": 36469 }, { "epoch": 0.04, "learning_rate": 4.9836953833857006e-05, "loss": 0.7903, "step": 36470 }, { "epoch": 0.04, "learning_rate": 4.983694487752284e-05, "loss": 1.4346, "step": 36471 }, { "epoch": 0.04, "learning_rate": 4.983693592094348e-05, "loss": 0.8633, "step": 36472 }, { "epoch": 0.04, "learning_rate": 4.983692696411896e-05, "loss": 0.8533, "step": 36473 }, { "epoch": 0.04, "learning_rate": 4.983691800704925e-05, "loss": 1.0254, "step": 36474 }, { "epoch": 0.04, "learning_rate": 4.983690904973437e-05, "loss": 1.3079, "step": 36475 }, { "epoch": 0.04, "learning_rate": 4.98369000921743e-05, "loss": 1.1198, "step": 36476 }, { "epoch": 0.04, "learning_rate": 4.983689113436906e-05, "loss": 0.9986, "step": 36477 }, { "epoch": 0.04, "learning_rate": 4.983688217631863e-05, "loss": 0.9721, "step": 36478 }, { "epoch": 0.04, "learning_rate": 4.9836873218023025e-05, "loss": 1.1015, "step": 36479 }, { "epoch": 0.04, "learning_rate": 4.9836864259482254e-05, "loss": 1.0201, "step": 36480 }, { "epoch": 0.04, "learning_rate": 4.9836855300696284e-05, "loss": 0.9897, "step": 36481 }, { "epoch": 0.04, "learning_rate": 4.983684634166514e-05, "loss": 0.9728, "step": 36482 }, { "epoch": 0.04, "learning_rate": 4.983683738238882e-05, "loss": 1.0963, "step": 36483 }, { "epoch": 0.04, "learning_rate": 4.9836828422867335e-05, "loss": 0.8765, "step": 36484 }, { "epoch": 0.04, "learning_rate": 4.9836819463100657e-05, "loss": 0.9878, "step": 36485 }, { "epoch": 0.04, "learning_rate": 4.983681050308879e-05, "loss": 0.828, "step": 36486 }, { "epoch": 0.04, "learning_rate": 4.9836801542831764e-05, "loss": 1.0672, "step": 36487 }, { "epoch": 0.04, "learning_rate": 4.983679258232955e-05, "loss": 1.7245, "step": 36488 }, { "epoch": 0.04, "learning_rate": 4.983678362158216e-05, "loss": 2.4195, "step": 36489 }, { "epoch": 0.04, "learning_rate": 4.983677466058959e-05, "loss": 2.2616, "step": 36490 }, { "epoch": 0.04, "learning_rate": 4.9836765699351846e-05, "loss": 1.5746, "step": 36491 }, { "epoch": 0.04, "learning_rate": 4.9836756737868916e-05, "loss": 1.5618, "step": 36492 }, { "epoch": 0.04, "learning_rate": 4.9836747776140815e-05, "loss": 1.708, "step": 36493 }, { "epoch": 0.04, "learning_rate": 4.9836738814167534e-05, "loss": 1.379, "step": 36494 }, { "epoch": 0.04, "learning_rate": 4.983672985194907e-05, "loss": 1.0986, "step": 36495 }, { "epoch": 0.04, "learning_rate": 4.983672088948544e-05, "loss": 0.7218, "step": 36496 }, { "epoch": 0.04, "learning_rate": 4.983671192677663e-05, "loss": 1.1069, "step": 36497 }, { "epoch": 0.04, "learning_rate": 4.983670296382263e-05, "loss": 1.2957, "step": 36498 }, { "epoch": 0.04, "learning_rate": 4.9836694000623464e-05, "loss": 1.4501, "step": 36499 }, { "epoch": 0.04, "learning_rate": 4.983668503717912e-05, "loss": 1.1574, "step": 36500 }, { "epoch": 0.04, "eval_loss": 1.0801715850830078, "eval_runtime": 126.5217, "eval_samples_per_second": 10.947, "eval_steps_per_second": 5.477, "step": 36500 }, { "epoch": 0.04, "learning_rate": 4.983667607348959e-05, "loss": 1.1581, "step": 36501 }, { "epoch": 0.04, "learning_rate": 4.983666710955489e-05, "loss": 0.7902, "step": 36502 }, { "epoch": 0.04, "learning_rate": 4.983665814537501e-05, "loss": 0.7313, "step": 36503 }, { "epoch": 0.04, "learning_rate": 4.983664918094996e-05, "loss": 0.9981, "step": 36504 }, { "epoch": 0.04, "learning_rate": 4.983664021627973e-05, "loss": 0.5329, "step": 36505 }, { "epoch": 0.04, "learning_rate": 4.983663125136432e-05, "loss": 0.6306, "step": 36506 }, { "epoch": 0.04, "learning_rate": 4.983662228620373e-05, "loss": 0.5505, "step": 36507 }, { "epoch": 0.04, "learning_rate": 4.9836613320797966e-05, "loss": 1.2046, "step": 36508 }, { "epoch": 0.04, "learning_rate": 4.983660435514703e-05, "loss": 1.0974, "step": 36509 }, { "epoch": 0.04, "learning_rate": 4.983659538925091e-05, "loss": 0.8238, "step": 36510 }, { "epoch": 0.04, "learning_rate": 4.983658642310962e-05, "loss": 1.0088, "step": 36511 }, { "epoch": 0.04, "learning_rate": 4.9836577456723154e-05, "loss": 1.1824, "step": 36512 }, { "epoch": 0.04, "learning_rate": 4.9836568490091504e-05, "loss": 1.3212, "step": 36513 }, { "epoch": 0.04, "learning_rate": 4.983655952321469e-05, "loss": 1.2092, "step": 36514 }, { "epoch": 0.04, "learning_rate": 4.983655055609269e-05, "loss": 0.8786, "step": 36515 }, { "epoch": 0.04, "learning_rate": 4.9836541588725515e-05, "loss": 0.6624, "step": 36516 }, { "epoch": 0.04, "learning_rate": 4.983653262111317e-05, "loss": 0.851, "step": 36517 }, { "epoch": 0.04, "learning_rate": 4.983652365325564e-05, "loss": 1.3885, "step": 36518 }, { "epoch": 0.04, "learning_rate": 4.9836514685152945e-05, "loss": 1.282, "step": 36519 }, { "epoch": 0.04, "learning_rate": 4.983650571680507e-05, "loss": 1.0518, "step": 36520 }, { "epoch": 0.04, "learning_rate": 4.983649674821201e-05, "loss": 0.7823, "step": 36521 }, { "epoch": 0.04, "learning_rate": 4.9836487779373786e-05, "loss": 0.5314, "step": 36522 }, { "epoch": 0.04, "learning_rate": 4.983647881029039e-05, "loss": 0.5989, "step": 36523 }, { "epoch": 0.04, "learning_rate": 4.9836469840961806e-05, "loss": 1.0711, "step": 36524 }, { "epoch": 0.04, "learning_rate": 4.983646087138806e-05, "loss": 1.4122, "step": 36525 }, { "epoch": 0.04, "learning_rate": 4.983645190156913e-05, "loss": 0.9583, "step": 36526 }, { "epoch": 0.04, "learning_rate": 4.983644293150503e-05, "loss": 0.7795, "step": 36527 }, { "epoch": 0.04, "learning_rate": 4.983643396119575e-05, "loss": 1.0108, "step": 36528 }, { "epoch": 0.04, "learning_rate": 4.983642499064129e-05, "loss": 0.87, "step": 36529 }, { "epoch": 0.04, "learning_rate": 4.983641601984167e-05, "loss": 1.055, "step": 36530 }, { "epoch": 0.04, "learning_rate": 4.983640704879687e-05, "loss": 1.1491, "step": 36531 }, { "epoch": 0.04, "learning_rate": 4.983639807750689e-05, "loss": 0.6437, "step": 36532 }, { "epoch": 0.04, "learning_rate": 4.983638910597174e-05, "loss": 1.0763, "step": 36533 }, { "epoch": 0.04, "learning_rate": 4.983638013419142e-05, "loss": 1.0292, "step": 36534 }, { "epoch": 0.04, "learning_rate": 4.983637116216592e-05, "loss": 1.0523, "step": 36535 }, { "epoch": 0.04, "learning_rate": 4.9836362189895245e-05, "loss": 0.9211, "step": 36536 }, { "epoch": 0.04, "learning_rate": 4.98363532173794e-05, "loss": 0.9905, "step": 36537 }, { "epoch": 0.04, "learning_rate": 4.9836344244618375e-05, "loss": 0.9577, "step": 36538 }, { "epoch": 0.04, "learning_rate": 4.983633527161218e-05, "loss": 0.8783, "step": 36539 }, { "epoch": 0.04, "learning_rate": 4.983632629836082e-05, "loss": 1.3294, "step": 36540 }, { "epoch": 0.04, "learning_rate": 4.983631732486427e-05, "loss": 0.9756, "step": 36541 }, { "epoch": 0.04, "learning_rate": 4.983630835112255e-05, "loss": 1.099, "step": 36542 }, { "epoch": 0.04, "learning_rate": 4.9836299377135667e-05, "loss": 1.1711, "step": 36543 }, { "epoch": 0.04, "learning_rate": 4.98362904029036e-05, "loss": 0.8608, "step": 36544 }, { "epoch": 0.04, "learning_rate": 4.983628142842636e-05, "loss": 1.1201, "step": 36545 }, { "epoch": 0.04, "learning_rate": 4.9836272453703955e-05, "loss": 1.1674, "step": 36546 }, { "epoch": 0.04, "learning_rate": 4.983626347873637e-05, "loss": 0.8879, "step": 36547 }, { "epoch": 0.04, "learning_rate": 4.983625450352361e-05, "loss": 1.0956, "step": 36548 }, { "epoch": 0.04, "learning_rate": 4.983624552806568e-05, "loss": 1.0042, "step": 36549 }, { "epoch": 0.04, "learning_rate": 4.983623655236258e-05, "loss": 1.1405, "step": 36550 }, { "epoch": 0.04, "learning_rate": 4.98362275764143e-05, "loss": 0.8372, "step": 36551 }, { "epoch": 0.04, "learning_rate": 4.983621860022085e-05, "loss": 1.0878, "step": 36552 }, { "epoch": 0.04, "learning_rate": 4.9836209623782235e-05, "loss": 0.9614, "step": 36553 }, { "epoch": 0.04, "learning_rate": 4.983620064709844e-05, "loss": 0.7617, "step": 36554 }, { "epoch": 0.04, "learning_rate": 4.983619167016948e-05, "loss": 0.4594, "step": 36555 }, { "epoch": 0.04, "learning_rate": 4.983618269299534e-05, "loss": 0.7125, "step": 36556 }, { "epoch": 0.04, "learning_rate": 4.9836173715576024e-05, "loss": 1.2592, "step": 36557 }, { "epoch": 0.04, "learning_rate": 4.983616473791154e-05, "loss": 0.8533, "step": 36558 }, { "epoch": 0.04, "learning_rate": 4.983615576000188e-05, "loss": 0.7256, "step": 36559 }, { "epoch": 0.04, "learning_rate": 4.983614678184706e-05, "loss": 0.6465, "step": 36560 }, { "epoch": 0.04, "learning_rate": 4.983613780344706e-05, "loss": 0.5999, "step": 36561 }, { "epoch": 0.04, "learning_rate": 4.983612882480189e-05, "loss": 0.8945, "step": 36562 }, { "epoch": 0.04, "learning_rate": 4.9836119845911544e-05, "loss": 1.1469, "step": 36563 }, { "epoch": 0.04, "learning_rate": 4.9836110866776034e-05, "loss": 1.3692, "step": 36564 }, { "epoch": 0.04, "learning_rate": 4.983610188739534e-05, "loss": 0.4626, "step": 36565 }, { "epoch": 0.04, "learning_rate": 4.9836092907769486e-05, "loss": 0.3686, "step": 36566 }, { "epoch": 0.04, "learning_rate": 4.983608392789846e-05, "loss": 0.3258, "step": 36567 }, { "epoch": 0.04, "learning_rate": 4.983607494778225e-05, "loss": 1.0006, "step": 36568 }, { "epoch": 0.04, "learning_rate": 4.9836065967420886e-05, "loss": 0.5486, "step": 36569 }, { "epoch": 0.04, "learning_rate": 4.983605698681434e-05, "loss": 1.0318, "step": 36570 }, { "epoch": 0.04, "learning_rate": 4.9836048005962624e-05, "loss": 0.9734, "step": 36571 }, { "epoch": 0.04, "learning_rate": 4.9836039024865745e-05, "loss": 0.7297, "step": 36572 }, { "epoch": 0.04, "learning_rate": 4.983603004352369e-05, "loss": 1.1062, "step": 36573 }, { "epoch": 0.04, "learning_rate": 4.983602106193646e-05, "loss": 0.8046, "step": 36574 }, { "epoch": 0.04, "learning_rate": 4.983601208010406e-05, "loss": 0.9771, "step": 36575 }, { "epoch": 0.04, "learning_rate": 4.983600309802649e-05, "loss": 0.9632, "step": 36576 }, { "epoch": 0.04, "learning_rate": 4.983599411570376e-05, "loss": 0.7491, "step": 36577 }, { "epoch": 0.04, "learning_rate": 4.983598513313584e-05, "loss": 1.4442, "step": 36578 }, { "epoch": 0.04, "learning_rate": 4.9835976150322764e-05, "loss": 1.2506, "step": 36579 }, { "epoch": 0.04, "learning_rate": 4.9835967167264516e-05, "loss": 1.0025, "step": 36580 }, { "epoch": 0.04, "learning_rate": 4.9835958183961096e-05, "loss": 1.2585, "step": 36581 }, { "epoch": 0.04, "learning_rate": 4.9835949200412504e-05, "loss": 1.1767, "step": 36582 }, { "epoch": 0.04, "learning_rate": 4.983594021661874e-05, "loss": 1.0739, "step": 36583 }, { "epoch": 0.04, "learning_rate": 4.983593123257981e-05, "loss": 1.0516, "step": 36584 }, { "epoch": 0.04, "learning_rate": 4.983592224829571e-05, "loss": 0.9998, "step": 36585 }, { "epoch": 0.04, "learning_rate": 4.983591326376644e-05, "loss": 1.1904, "step": 36586 }, { "epoch": 0.04, "learning_rate": 4.9835904278992005e-05, "loss": 0.9756, "step": 36587 }, { "epoch": 0.04, "learning_rate": 4.983589529397239e-05, "loss": 1.3235, "step": 36588 }, { "epoch": 0.04, "learning_rate": 4.983588630870761e-05, "loss": 1.221, "step": 36589 }, { "epoch": 0.04, "learning_rate": 4.9835877323197664e-05, "loss": 0.8106, "step": 36590 }, { "epoch": 0.04, "learning_rate": 4.983586833744255e-05, "loss": 1.0327, "step": 36591 }, { "epoch": 0.04, "learning_rate": 4.983585935144226e-05, "loss": 1.2364, "step": 36592 }, { "epoch": 0.04, "learning_rate": 4.983585036519681e-05, "loss": 1.0057, "step": 36593 }, { "epoch": 0.04, "learning_rate": 4.983584137870618e-05, "loss": 0.9911, "step": 36594 }, { "epoch": 0.04, "learning_rate": 4.983583239197039e-05, "loss": 1.1683, "step": 36595 }, { "epoch": 0.04, "learning_rate": 4.983582340498942e-05, "loss": 1.255, "step": 36596 }, { "epoch": 0.04, "learning_rate": 4.98358144177633e-05, "loss": 1.1747, "step": 36597 }, { "epoch": 0.04, "learning_rate": 4.9835805430291994e-05, "loss": 0.7462, "step": 36598 }, { "epoch": 0.04, "learning_rate": 4.9835796442575525e-05, "loss": 0.898, "step": 36599 }, { "epoch": 0.04, "learning_rate": 4.983578745461389e-05, "loss": 0.9625, "step": 36600 }, { "epoch": 0.04, "learning_rate": 4.983577846640708e-05, "loss": 0.8622, "step": 36601 }, { "epoch": 0.04, "learning_rate": 4.983576947795511e-05, "loss": 1.0915, "step": 36602 }, { "epoch": 0.04, "learning_rate": 4.9835760489257966e-05, "loss": 0.2939, "step": 36603 }, { "epoch": 0.04, "learning_rate": 4.983575150031565e-05, "loss": 0.2163, "step": 36604 }, { "epoch": 0.04, "learning_rate": 4.983574251112818e-05, "loss": 0.1621, "step": 36605 }, { "epoch": 0.04, "learning_rate": 4.9835733521695525e-05, "loss": 0.1711, "step": 36606 }, { "epoch": 0.04, "learning_rate": 4.9835724532017715e-05, "loss": 0.1762, "step": 36607 }, { "epoch": 0.04, "learning_rate": 4.983571554209473e-05, "loss": 0.1318, "step": 36608 }, { "epoch": 0.04, "learning_rate": 4.9835706551926585e-05, "loss": 0.1487, "step": 36609 }, { "epoch": 0.04, "learning_rate": 4.9835697561513265e-05, "loss": 0.1226, "step": 36610 }, { "epoch": 0.04, "learning_rate": 4.983568857085478e-05, "loss": 0.0732, "step": 36611 }, { "epoch": 0.04, "learning_rate": 4.983567957995113e-05, "loss": 0.1173, "step": 36612 }, { "epoch": 0.04, "learning_rate": 4.983567058880231e-05, "loss": 0.1369, "step": 36613 }, { "epoch": 0.04, "learning_rate": 4.983566159740832e-05, "loss": 0.142, "step": 36614 }, { "epoch": 0.04, "learning_rate": 4.983565260576917e-05, "loss": 0.1397, "step": 36615 }, { "epoch": 0.04, "learning_rate": 4.983564361388485e-05, "loss": 0.1351, "step": 36616 }, { "epoch": 0.04, "learning_rate": 4.9835634621755354e-05, "loss": 0.1188, "step": 36617 }, { "epoch": 0.04, "learning_rate": 4.98356256293807e-05, "loss": 0.1446, "step": 36618 }, { "epoch": 0.04, "learning_rate": 4.983561663676088e-05, "loss": 0.1566, "step": 36619 }, { "epoch": 0.04, "learning_rate": 4.9835607643895887e-05, "loss": 0.094, "step": 36620 }, { "epoch": 0.04, "learning_rate": 4.9835598650785736e-05, "loss": 0.0982, "step": 36621 }, { "epoch": 0.04, "learning_rate": 4.983558965743041e-05, "loss": 0.1192, "step": 36622 }, { "epoch": 0.04, "learning_rate": 4.9835580663829925e-05, "loss": 0.1242, "step": 36623 }, { "epoch": 0.04, "learning_rate": 4.983557166998427e-05, "loss": 0.1149, "step": 36624 }, { "epoch": 0.04, "learning_rate": 4.983556267589345e-05, "loss": 0.1604, "step": 36625 }, { "epoch": 0.04, "learning_rate": 4.983555368155746e-05, "loss": 0.0998, "step": 36626 }, { "epoch": 0.04, "learning_rate": 4.983554468697631e-05, "loss": 0.1117, "step": 36627 }, { "epoch": 0.04, "learning_rate": 4.983553569214999e-05, "loss": 0.1402, "step": 36628 }, { "epoch": 0.04, "learning_rate": 4.98355266970785e-05, "loss": 0.1217, "step": 36629 }, { "epoch": 0.04, "learning_rate": 4.9835517701761844e-05, "loss": 0.103, "step": 36630 }, { "epoch": 0.04, "learning_rate": 4.983550870620003e-05, "loss": 0.1668, "step": 36631 }, { "epoch": 0.04, "learning_rate": 4.983549971039305e-05, "loss": 0.1757, "step": 36632 }, { "epoch": 0.04, "learning_rate": 4.9835490714340906e-05, "loss": 0.2533, "step": 36633 }, { "epoch": 0.04, "learning_rate": 4.983548171804359e-05, "loss": 1.0659, "step": 36634 }, { "epoch": 0.04, "learning_rate": 4.983547272150111e-05, "loss": 0.9119, "step": 36635 }, { "epoch": 0.04, "learning_rate": 4.983546372471346e-05, "loss": 0.9229, "step": 36636 }, { "epoch": 0.04, "learning_rate": 4.9835454727680653e-05, "loss": 1.197, "step": 36637 }, { "epoch": 0.04, "learning_rate": 4.9835445730402674e-05, "loss": 0.9739, "step": 36638 }, { "epoch": 0.04, "learning_rate": 4.9835436732879536e-05, "loss": 1.2517, "step": 36639 }, { "epoch": 0.04, "learning_rate": 4.9835427735111226e-05, "loss": 1.0737, "step": 36640 }, { "epoch": 0.04, "learning_rate": 4.9835418737097764e-05, "loss": 1.0102, "step": 36641 }, { "epoch": 0.04, "learning_rate": 4.983540973883913e-05, "loss": 0.8781, "step": 36642 }, { "epoch": 0.04, "learning_rate": 4.9835400740335325e-05, "loss": 0.8746, "step": 36643 }, { "epoch": 0.04, "learning_rate": 4.983539174158637e-05, "loss": 0.7693, "step": 36644 }, { "epoch": 0.04, "learning_rate": 4.983538274259224e-05, "loss": 1.1705, "step": 36645 }, { "epoch": 0.04, "learning_rate": 4.983537374335294e-05, "loss": 1.1494, "step": 36646 }, { "epoch": 0.04, "learning_rate": 4.9835364743868484e-05, "loss": 0.9667, "step": 36647 }, { "epoch": 0.04, "learning_rate": 4.983535574413886e-05, "loss": 0.7656, "step": 36648 }, { "epoch": 0.04, "learning_rate": 4.983534674416408e-05, "loss": 1.0816, "step": 36649 }, { "epoch": 0.04, "learning_rate": 4.9835337743944126e-05, "loss": 1.1325, "step": 36650 }, { "epoch": 0.04, "learning_rate": 4.983532874347901e-05, "loss": 1.318, "step": 36651 }, { "epoch": 0.04, "learning_rate": 4.983531974276874e-05, "loss": 1.1953, "step": 36652 }, { "epoch": 0.04, "learning_rate": 4.983531074181329e-05, "loss": 1.1058, "step": 36653 }, { "epoch": 0.04, "learning_rate": 4.9835301740612686e-05, "loss": 0.7405, "step": 36654 }, { "epoch": 0.04, "learning_rate": 4.983529273916692e-05, "loss": 0.8232, "step": 36655 }, { "epoch": 0.04, "learning_rate": 4.983528373747599e-05, "loss": 1.0461, "step": 36656 }, { "epoch": 0.04, "learning_rate": 4.9835274735539894e-05, "loss": 0.9282, "step": 36657 }, { "epoch": 0.04, "learning_rate": 4.983526573335863e-05, "loss": 1.0718, "step": 36658 }, { "epoch": 0.04, "learning_rate": 4.9835256730932215e-05, "loss": 1.1952, "step": 36659 }, { "epoch": 0.04, "learning_rate": 4.983524772826063e-05, "loss": 0.9948, "step": 36660 }, { "epoch": 0.04, "learning_rate": 4.9835238725343875e-05, "loss": 0.9008, "step": 36661 }, { "epoch": 0.04, "learning_rate": 4.9835229722181964e-05, "loss": 1.1112, "step": 36662 }, { "epoch": 0.04, "learning_rate": 4.983522071877489e-05, "loss": 0.8711, "step": 36663 }, { "epoch": 0.04, "learning_rate": 4.983521171512265e-05, "loss": 0.968, "step": 36664 }, { "epoch": 0.04, "learning_rate": 4.983520271122525e-05, "loss": 1.0797, "step": 36665 }, { "epoch": 0.04, "learning_rate": 4.983519370708268e-05, "loss": 1.3646, "step": 36666 }, { "epoch": 0.04, "learning_rate": 4.983518470269496e-05, "loss": 0.5502, "step": 36667 }, { "epoch": 0.04, "learning_rate": 4.9835175698062076e-05, "loss": 0.9814, "step": 36668 }, { "epoch": 0.04, "learning_rate": 4.983516669318402e-05, "loss": 0.9783, "step": 36669 }, { "epoch": 0.04, "learning_rate": 4.983515768806081e-05, "loss": 1.0796, "step": 36670 }, { "epoch": 0.04, "learning_rate": 4.983514868269243e-05, "loss": 0.8082, "step": 36671 }, { "epoch": 0.04, "learning_rate": 4.98351396770789e-05, "loss": 0.8245, "step": 36672 }, { "epoch": 0.04, "learning_rate": 4.9835130671220195e-05, "loss": 1.0604, "step": 36673 }, { "epoch": 0.04, "learning_rate": 4.9835121665116336e-05, "loss": 0.8059, "step": 36674 }, { "epoch": 0.04, "learning_rate": 4.983511265876731e-05, "loss": 0.5609, "step": 36675 }, { "epoch": 0.04, "learning_rate": 4.9835103652173134e-05, "loss": 0.5552, "step": 36676 }, { "epoch": 0.04, "learning_rate": 4.983509464533378e-05, "loss": 1.3113, "step": 36677 }, { "epoch": 0.04, "learning_rate": 4.983508563824927e-05, "loss": 1.4318, "step": 36678 }, { "epoch": 0.04, "learning_rate": 4.983507663091961e-05, "loss": 1.0463, "step": 36679 }, { "epoch": 0.04, "learning_rate": 4.9835067623344776e-05, "loss": 0.9647, "step": 36680 }, { "epoch": 0.04, "learning_rate": 4.983505861552478e-05, "loss": 0.7846, "step": 36681 }, { "epoch": 0.04, "learning_rate": 4.9835049607459626e-05, "loss": 0.7959, "step": 36682 }, { "epoch": 0.04, "learning_rate": 4.9835040599149313e-05, "loss": 0.6848, "step": 36683 }, { "epoch": 0.04, "learning_rate": 4.983503159059384e-05, "loss": 0.7586, "step": 36684 }, { "epoch": 0.04, "learning_rate": 4.98350225817932e-05, "loss": 1.0414, "step": 36685 }, { "epoch": 0.04, "learning_rate": 4.9835013572747404e-05, "loss": 1.1165, "step": 36686 }, { "epoch": 0.04, "learning_rate": 4.9835004563456444e-05, "loss": 0.8255, "step": 36687 }, { "epoch": 0.04, "learning_rate": 4.9834995553920326e-05, "loss": 0.9242, "step": 36688 }, { "epoch": 0.04, "learning_rate": 4.983498654413905e-05, "loss": 0.6501, "step": 36689 }, { "epoch": 0.04, "learning_rate": 4.983497753411261e-05, "loss": 1.0977, "step": 36690 }, { "epoch": 0.04, "learning_rate": 4.9834968523841006e-05, "loss": 0.9076, "step": 36691 }, { "epoch": 0.04, "learning_rate": 4.983495951332424e-05, "loss": 1.1035, "step": 36692 }, { "epoch": 0.04, "learning_rate": 4.983495050256233e-05, "loss": 0.8694, "step": 36693 }, { "epoch": 0.04, "learning_rate": 4.983494149155524e-05, "loss": 0.9792, "step": 36694 }, { "epoch": 0.04, "learning_rate": 4.9834932480303e-05, "loss": 0.9852, "step": 36695 }, { "epoch": 0.04, "learning_rate": 4.98349234688056e-05, "loss": 1.1988, "step": 36696 }, { "epoch": 0.04, "learning_rate": 4.983491445706304e-05, "loss": 1.1708, "step": 36697 }, { "epoch": 0.04, "learning_rate": 4.983490544507532e-05, "loss": 1.2852, "step": 36698 }, { "epoch": 0.04, "learning_rate": 4.983489643284244e-05, "loss": 1.1419, "step": 36699 }, { "epoch": 0.04, "learning_rate": 4.98348874203644e-05, "loss": 1.1139, "step": 36700 }, { "epoch": 0.04, "learning_rate": 4.983487840764119e-05, "loss": 1.1473, "step": 36701 }, { "epoch": 0.04, "learning_rate": 4.9834869394672835e-05, "loss": 1.074, "step": 36702 }, { "epoch": 0.04, "learning_rate": 4.983486038145931e-05, "loss": 0.9082, "step": 36703 }, { "epoch": 0.04, "learning_rate": 4.983485136800064e-05, "loss": 0.9913, "step": 36704 }, { "epoch": 0.04, "learning_rate": 4.98348423542968e-05, "loss": 1.2458, "step": 36705 }, { "epoch": 0.04, "learning_rate": 4.983483334034781e-05, "loss": 1.1115, "step": 36706 }, { "epoch": 0.04, "learning_rate": 4.983482432615365e-05, "loss": 1.083, "step": 36707 }, { "epoch": 0.04, "learning_rate": 4.983481531171433e-05, "loss": 1.0694, "step": 36708 }, { "epoch": 0.04, "learning_rate": 4.983480629702986e-05, "loss": 1.1429, "step": 36709 }, { "epoch": 0.04, "learning_rate": 4.9834797282100235e-05, "loss": 1.0356, "step": 36710 }, { "epoch": 0.04, "learning_rate": 4.983478826692544e-05, "loss": 1.1572, "step": 36711 }, { "epoch": 0.04, "learning_rate": 4.983477925150549e-05, "loss": 0.7691, "step": 36712 }, { "epoch": 0.04, "learning_rate": 4.9834770235840384e-05, "loss": 0.9285, "step": 36713 }, { "epoch": 0.04, "learning_rate": 4.983476121993011e-05, "loss": 1.139, "step": 36714 }, { "epoch": 0.04, "learning_rate": 4.9834752203774695e-05, "loss": 1.2964, "step": 36715 }, { "epoch": 0.04, "learning_rate": 4.983474318737411e-05, "loss": 1.0772, "step": 36716 }, { "epoch": 0.04, "learning_rate": 4.983473417072837e-05, "loss": 1.051, "step": 36717 }, { "epoch": 0.04, "learning_rate": 4.983472515383747e-05, "loss": 1.0587, "step": 36718 }, { "epoch": 0.04, "learning_rate": 4.983471613670142e-05, "loss": 0.8612, "step": 36719 }, { "epoch": 0.04, "learning_rate": 4.98347071193202e-05, "loss": 0.9058, "step": 36720 }, { "epoch": 0.04, "learning_rate": 4.983469810169383e-05, "loss": 0.8373, "step": 36721 }, { "epoch": 0.04, "learning_rate": 4.9834689083822295e-05, "loss": 0.8507, "step": 36722 }, { "epoch": 0.04, "learning_rate": 4.983468006570561e-05, "loss": 0.9668, "step": 36723 }, { "epoch": 0.04, "learning_rate": 4.983467104734376e-05, "loss": 0.8768, "step": 36724 }, { "epoch": 0.04, "learning_rate": 4.983466202873677e-05, "loss": 1.0059, "step": 36725 }, { "epoch": 0.04, "learning_rate": 4.98346530098846e-05, "loss": 0.9779, "step": 36726 }, { "epoch": 0.04, "learning_rate": 4.983464399078729e-05, "loss": 1.1633, "step": 36727 }, { "epoch": 0.04, "learning_rate": 4.983463497144481e-05, "loss": 1.1685, "step": 36728 }, { "epoch": 0.04, "learning_rate": 4.9834625951857185e-05, "loss": 1.1379, "step": 36729 }, { "epoch": 0.04, "learning_rate": 4.983461693202439e-05, "loss": 1.0121, "step": 36730 }, { "epoch": 0.04, "learning_rate": 4.9834607911946454e-05, "loss": 1.3027, "step": 36731 }, { "epoch": 0.04, "learning_rate": 4.9834598891623343e-05, "loss": 1.0348, "step": 36732 }, { "epoch": 0.04, "learning_rate": 4.983458987105509e-05, "loss": 0.8856, "step": 36733 }, { "epoch": 0.04, "learning_rate": 4.9834580850241675e-05, "loss": 1.0344, "step": 36734 }, { "epoch": 0.04, "learning_rate": 4.98345718291831e-05, "loss": 1.0276, "step": 36735 }, { "epoch": 0.04, "learning_rate": 4.983456280787937e-05, "loss": 0.8174, "step": 36736 }, { "epoch": 0.04, "learning_rate": 4.983455378633049e-05, "loss": 0.6851, "step": 36737 }, { "epoch": 0.04, "learning_rate": 4.983454476453645e-05, "loss": 0.9232, "step": 36738 }, { "epoch": 0.04, "learning_rate": 4.983453574249725e-05, "loss": 0.9866, "step": 36739 }, { "epoch": 0.04, "learning_rate": 4.98345267202129e-05, "loss": 1.1603, "step": 36740 }, { "epoch": 0.04, "learning_rate": 4.983451769768339e-05, "loss": 0.9041, "step": 36741 }, { "epoch": 0.04, "learning_rate": 4.9834508674908724e-05, "loss": 0.9886, "step": 36742 }, { "epoch": 0.04, "learning_rate": 4.9834499651888905e-05, "loss": 1.3068, "step": 36743 }, { "epoch": 0.04, "learning_rate": 4.983449062862393e-05, "loss": 1.3399, "step": 36744 }, { "epoch": 0.04, "learning_rate": 4.98344816051138e-05, "loss": 1.4593, "step": 36745 }, { "epoch": 0.04, "learning_rate": 4.983447258135851e-05, "loss": 0.9628, "step": 36746 }, { "epoch": 0.04, "learning_rate": 4.983446355735807e-05, "loss": 1.1232, "step": 36747 }, { "epoch": 0.04, "learning_rate": 4.983445453311247e-05, "loss": 1.0046, "step": 36748 }, { "epoch": 0.04, "learning_rate": 4.983444550862172e-05, "loss": 1.1928, "step": 36749 }, { "epoch": 0.04, "learning_rate": 4.98344364838858e-05, "loss": 1.1138, "step": 36750 }, { "epoch": 0.04, "learning_rate": 4.9834427458904743e-05, "loss": 0.5842, "step": 36751 }, { "epoch": 0.04, "learning_rate": 4.983441843367852e-05, "loss": 1.3033, "step": 36752 }, { "epoch": 0.04, "learning_rate": 4.983440940820715e-05, "loss": 0.8558, "step": 36753 }, { "epoch": 0.04, "learning_rate": 4.983440038249062e-05, "loss": 0.8939, "step": 36754 }, { "epoch": 0.04, "learning_rate": 4.983439135652894e-05, "loss": 0.7753, "step": 36755 }, { "epoch": 0.04, "learning_rate": 4.98343823303221e-05, "loss": 1.3038, "step": 36756 }, { "epoch": 0.04, "learning_rate": 4.9834373303870106e-05, "loss": 1.045, "step": 36757 }, { "epoch": 0.04, "learning_rate": 4.9834364277172965e-05, "loss": 0.8739, "step": 36758 }, { "epoch": 0.04, "learning_rate": 4.983435525023066e-05, "loss": 0.7775, "step": 36759 }, { "epoch": 0.04, "learning_rate": 4.983434622304321e-05, "loss": 1.0601, "step": 36760 }, { "epoch": 0.04, "learning_rate": 4.983433719561059e-05, "loss": 0.8737, "step": 36761 }, { "epoch": 0.04, "learning_rate": 4.983432816793283e-05, "loss": 0.952, "step": 36762 }, { "epoch": 0.04, "learning_rate": 4.983431914000991e-05, "loss": 0.2886, "step": 36763 }, { "epoch": 0.04, "learning_rate": 4.983431011184184e-05, "loss": 0.0911, "step": 36764 }, { "epoch": 0.04, "learning_rate": 4.983430108342861e-05, "loss": 0.7786, "step": 36765 }, { "epoch": 0.04, "learning_rate": 4.983429205477024e-05, "loss": 1.0432, "step": 36766 }, { "epoch": 0.04, "learning_rate": 4.9834283025866704e-05, "loss": 0.9589, "step": 36767 }, { "epoch": 0.04, "learning_rate": 4.983427399671802e-05, "loss": 1.1058, "step": 36768 }, { "epoch": 0.04, "learning_rate": 4.9834264967324175e-05, "loss": 1.0817, "step": 36769 }, { "epoch": 0.04, "learning_rate": 4.983425593768518e-05, "loss": 0.8909, "step": 36770 }, { "epoch": 0.04, "learning_rate": 4.983424690780103e-05, "loss": 1.3273, "step": 36771 }, { "epoch": 0.04, "learning_rate": 4.9834237877671734e-05, "loss": 1.2326, "step": 36772 }, { "epoch": 0.04, "learning_rate": 4.9834228847297284e-05, "loss": 0.8338, "step": 36773 }, { "epoch": 0.04, "learning_rate": 4.9834219816677675e-05, "loss": 0.899, "step": 36774 }, { "epoch": 0.04, "learning_rate": 4.983421078581292e-05, "loss": 1.1268, "step": 36775 }, { "epoch": 0.04, "learning_rate": 4.9834201754703e-05, "loss": 0.979, "step": 36776 }, { "epoch": 0.04, "learning_rate": 4.983419272334794e-05, "loss": 0.8565, "step": 36777 }, { "epoch": 0.04, "learning_rate": 4.983418369174772e-05, "loss": 0.8901, "step": 36778 }, { "epoch": 0.04, "learning_rate": 4.9834174659902356e-05, "loss": 0.9932, "step": 36779 }, { "epoch": 0.04, "learning_rate": 4.983416562781183e-05, "loss": 0.8369, "step": 36780 }, { "epoch": 0.04, "learning_rate": 4.9834156595476154e-05, "loss": 0.9724, "step": 36781 }, { "epoch": 0.04, "learning_rate": 4.9834147562895325e-05, "loss": 1.4079, "step": 36782 }, { "epoch": 0.04, "learning_rate": 4.9834138530069344e-05, "loss": 1.0093, "step": 36783 }, { "epoch": 0.04, "learning_rate": 4.983412949699822e-05, "loss": 1.0201, "step": 36784 }, { "epoch": 0.04, "learning_rate": 4.9834120463681935e-05, "loss": 1.2304, "step": 36785 }, { "epoch": 0.04, "learning_rate": 4.983411143012049e-05, "loss": 0.9665, "step": 36786 }, { "epoch": 0.04, "learning_rate": 4.9834102396313905e-05, "loss": 1.4841, "step": 36787 }, { "epoch": 0.04, "learning_rate": 4.9834093362262167e-05, "loss": 1.146, "step": 36788 }, { "epoch": 0.04, "learning_rate": 4.983408432796528e-05, "loss": 0.9318, "step": 36789 }, { "epoch": 0.04, "learning_rate": 4.9834075293423234e-05, "loss": 1.0801, "step": 36790 }, { "epoch": 0.04, "learning_rate": 4.983406625863604e-05, "loss": 1.1343, "step": 36791 }, { "epoch": 0.04, "learning_rate": 4.9834057223603695e-05, "loss": 1.0616, "step": 36792 }, { "epoch": 0.04, "learning_rate": 4.98340481883262e-05, "loss": 1.2639, "step": 36793 }, { "epoch": 0.04, "learning_rate": 4.983403915280355e-05, "loss": 1.0968, "step": 36794 }, { "epoch": 0.04, "learning_rate": 4.983403011703575e-05, "loss": 0.9372, "step": 36795 }, { "epoch": 0.04, "learning_rate": 4.9834021081022796e-05, "loss": 1.0039, "step": 36796 }, { "epoch": 0.04, "learning_rate": 4.98340120447647e-05, "loss": 0.3631, "step": 36797 }, { "epoch": 0.04, "learning_rate": 4.9834003008261443e-05, "loss": 0.3807, "step": 36798 }, { "epoch": 0.04, "learning_rate": 4.9833993971513036e-05, "loss": 0.9153, "step": 36799 }, { "epoch": 0.04, "learning_rate": 4.9833984934519484e-05, "loss": 1.0009, "step": 36800 }, { "epoch": 0.04, "learning_rate": 4.983397589728078e-05, "loss": 0.9898, "step": 36801 }, { "epoch": 0.04, "learning_rate": 4.9833966859796925e-05, "loss": 1.1514, "step": 36802 }, { "epoch": 0.04, "learning_rate": 4.9833957822067925e-05, "loss": 1.0338, "step": 36803 }, { "epoch": 0.04, "learning_rate": 4.9833948784093766e-05, "loss": 1.3844, "step": 36804 }, { "epoch": 0.04, "learning_rate": 4.9833939745874456e-05, "loss": 1.1031, "step": 36805 }, { "epoch": 0.04, "learning_rate": 4.983393070741e-05, "loss": 1.0982, "step": 36806 }, { "epoch": 0.04, "learning_rate": 4.9833921668700394e-05, "loss": 1.0535, "step": 36807 }, { "epoch": 0.04, "learning_rate": 4.9833912629745635e-05, "loss": 1.204, "step": 36808 }, { "epoch": 0.04, "learning_rate": 4.983390359054573e-05, "loss": 1.0828, "step": 36809 }, { "epoch": 0.04, "learning_rate": 4.983389455110068e-05, "loss": 1.0073, "step": 36810 }, { "epoch": 0.04, "learning_rate": 4.983388551141047e-05, "loss": 1.033, "step": 36811 }, { "epoch": 0.04, "learning_rate": 4.983387647147512e-05, "loss": 1.4636, "step": 36812 }, { "epoch": 0.04, "learning_rate": 4.9833867431294615e-05, "loss": 1.0949, "step": 36813 }, { "epoch": 0.04, "learning_rate": 4.983385839086895e-05, "loss": 1.151, "step": 36814 }, { "epoch": 0.04, "learning_rate": 4.9833849350198147e-05, "loss": 1.1209, "step": 36815 }, { "epoch": 0.04, "learning_rate": 4.9833840309282195e-05, "loss": 1.1683, "step": 36816 }, { "epoch": 0.04, "learning_rate": 4.983383126812109e-05, "loss": 0.6928, "step": 36817 }, { "epoch": 0.04, "learning_rate": 4.9833822226714844e-05, "loss": 0.7399, "step": 36818 }, { "epoch": 0.04, "learning_rate": 4.983381318506344e-05, "loss": 1.0921, "step": 36819 }, { "epoch": 0.04, "learning_rate": 4.983380414316689e-05, "loss": 0.9695, "step": 36820 }, { "epoch": 0.04, "learning_rate": 4.9833795101025196e-05, "loss": 1.1091, "step": 36821 }, { "epoch": 0.04, "learning_rate": 4.983378605863834e-05, "loss": 1.1981, "step": 36822 }, { "epoch": 0.04, "learning_rate": 4.983377701600635e-05, "loss": 0.8559, "step": 36823 }, { "epoch": 0.04, "learning_rate": 4.983376797312921e-05, "loss": 1.0491, "step": 36824 }, { "epoch": 0.04, "learning_rate": 4.9833758930006915e-05, "loss": 1.1326, "step": 36825 }, { "epoch": 0.04, "learning_rate": 4.983374988663947e-05, "loss": 0.9639, "step": 36826 }, { "epoch": 0.04, "learning_rate": 4.9833740843026874e-05, "loss": 1.2676, "step": 36827 }, { "epoch": 0.04, "learning_rate": 4.983373179916914e-05, "loss": 0.9993, "step": 36828 }, { "epoch": 0.04, "learning_rate": 4.983372275506626e-05, "loss": 0.6418, "step": 36829 }, { "epoch": 0.04, "learning_rate": 4.983371371071822e-05, "loss": 1.2203, "step": 36830 }, { "epoch": 0.04, "learning_rate": 4.9833704666125034e-05, "loss": 0.9394, "step": 36831 }, { "epoch": 0.04, "learning_rate": 4.983369562128671e-05, "loss": 1.1914, "step": 36832 }, { "epoch": 0.04, "learning_rate": 4.9833686576203234e-05, "loss": 1.031, "step": 36833 }, { "epoch": 0.04, "learning_rate": 4.983367753087461e-05, "loss": 1.0077, "step": 36834 }, { "epoch": 0.04, "learning_rate": 4.9833668485300835e-05, "loss": 1.0008, "step": 36835 }, { "epoch": 0.04, "learning_rate": 4.983365943948191e-05, "loss": 0.9195, "step": 36836 }, { "epoch": 0.04, "learning_rate": 4.983365039341784e-05, "loss": 1.0188, "step": 36837 }, { "epoch": 0.04, "learning_rate": 4.983364134710863e-05, "loss": 0.9124, "step": 36838 }, { "epoch": 0.04, "learning_rate": 4.983363230055427e-05, "loss": 0.8233, "step": 36839 }, { "epoch": 0.04, "learning_rate": 4.983362325375476e-05, "loss": 0.9263, "step": 36840 }, { "epoch": 0.04, "learning_rate": 4.98336142067101e-05, "loss": 1.1354, "step": 36841 }, { "epoch": 0.04, "learning_rate": 4.983360515942029e-05, "loss": 1.0681, "step": 36842 }, { "epoch": 0.04, "learning_rate": 4.983359611188535e-05, "loss": 1.1032, "step": 36843 }, { "epoch": 0.04, "learning_rate": 4.983358706410525e-05, "loss": 1.4875, "step": 36844 }, { "epoch": 0.04, "learning_rate": 4.983357801608e-05, "loss": 1.8018, "step": 36845 }, { "epoch": 0.04, "learning_rate": 4.983356896780961e-05, "loss": 0.6858, "step": 36846 }, { "epoch": 0.04, "learning_rate": 4.983355991929407e-05, "loss": 1.0381, "step": 36847 }, { "epoch": 0.04, "learning_rate": 4.983355087053339e-05, "loss": 1.0104, "step": 36848 }, { "epoch": 0.04, "learning_rate": 4.983354182152756e-05, "loss": 0.9331, "step": 36849 }, { "epoch": 0.04, "learning_rate": 4.983353277227658e-05, "loss": 0.6825, "step": 36850 }, { "epoch": 0.04, "learning_rate": 4.9833523722780464e-05, "loss": 1.1063, "step": 36851 }, { "epoch": 0.04, "learning_rate": 4.9833514673039195e-05, "loss": 4.6928, "step": 36852 }, { "epoch": 0.04, "learning_rate": 4.983350562305278e-05, "loss": 5.9102, "step": 36853 }, { "epoch": 0.04, "learning_rate": 4.9833496572821215e-05, "loss": 5.8578, "step": 36854 }, { "epoch": 0.04, "learning_rate": 4.983348752234451e-05, "loss": 3.2318, "step": 36855 }, { "epoch": 0.04, "learning_rate": 4.9833478471622656e-05, "loss": 1.2916, "step": 36856 }, { "epoch": 0.04, "learning_rate": 4.983346942065566e-05, "loss": 1.0646, "step": 36857 }, { "epoch": 0.04, "learning_rate": 4.983346036944352e-05, "loss": 0.6109, "step": 36858 }, { "epoch": 0.04, "learning_rate": 4.983345131798623e-05, "loss": 0.9308, "step": 36859 }, { "epoch": 0.04, "learning_rate": 4.983344226628379e-05, "loss": 1.3508, "step": 36860 }, { "epoch": 0.04, "learning_rate": 4.983343321433621e-05, "loss": 0.9147, "step": 36861 }, { "epoch": 0.04, "learning_rate": 4.9833424162143494e-05, "loss": 0.8364, "step": 36862 }, { "epoch": 0.04, "learning_rate": 4.983341510970562e-05, "loss": 0.9622, "step": 36863 }, { "epoch": 0.04, "learning_rate": 4.98334060570226e-05, "loss": 1.1907, "step": 36864 }, { "epoch": 0.04, "learning_rate": 4.9833397004094444e-05, "loss": 1.0463, "step": 36865 }, { "epoch": 0.04, "learning_rate": 4.983338795092114e-05, "loss": 1.2072, "step": 36866 }, { "epoch": 0.04, "learning_rate": 4.983337889750269e-05, "loss": 0.9537, "step": 36867 }, { "epoch": 0.04, "learning_rate": 4.9833369843839096e-05, "loss": 0.867, "step": 36868 }, { "epoch": 0.04, "learning_rate": 4.9833360789930364e-05, "loss": 1.0726, "step": 36869 }, { "epoch": 0.04, "learning_rate": 4.983335173577648e-05, "loss": 1.4943, "step": 36870 }, { "epoch": 0.04, "learning_rate": 4.983334268137745e-05, "loss": 1.1519, "step": 36871 }, { "epoch": 0.04, "learning_rate": 4.983333362673328e-05, "loss": 1.2276, "step": 36872 }, { "epoch": 0.04, "learning_rate": 4.9833324571843966e-05, "loss": 0.9812, "step": 36873 }, { "epoch": 0.04, "learning_rate": 4.98333155167095e-05, "loss": 0.6455, "step": 36874 }, { "epoch": 0.04, "learning_rate": 4.98333064613299e-05, "loss": 1.1467, "step": 36875 }, { "epoch": 0.04, "learning_rate": 4.983329740570515e-05, "loss": 1.1002, "step": 36876 }, { "epoch": 0.04, "learning_rate": 4.983328834983526e-05, "loss": 1.1585, "step": 36877 }, { "epoch": 0.04, "learning_rate": 4.9833279293720227e-05, "loss": 1.4414, "step": 36878 }, { "epoch": 0.04, "learning_rate": 4.983327023736004e-05, "loss": 1.0959, "step": 36879 }, { "epoch": 0.04, "learning_rate": 4.983326118075472e-05, "loss": 1.1214, "step": 36880 }, { "epoch": 0.04, "learning_rate": 4.9833252123904255e-05, "loss": 1.2223, "step": 36881 }, { "epoch": 0.04, "learning_rate": 4.9833243066808646e-05, "loss": 1.3191, "step": 36882 }, { "epoch": 0.04, "learning_rate": 4.9833234009467886e-05, "loss": 1.0163, "step": 36883 }, { "epoch": 0.04, "learning_rate": 4.9833224951881994e-05, "loss": 0.8561, "step": 36884 }, { "epoch": 0.04, "learning_rate": 4.983321589405096e-05, "loss": 0.7964, "step": 36885 }, { "epoch": 0.04, "learning_rate": 4.9833206835974775e-05, "loss": 0.8973, "step": 36886 }, { "epoch": 0.04, "learning_rate": 4.983319777765345e-05, "loss": 0.5044, "step": 36887 }, { "epoch": 0.04, "learning_rate": 4.9833188719086984e-05, "loss": 1.0611, "step": 36888 }, { "epoch": 0.04, "learning_rate": 4.983317966027537e-05, "loss": 1.3026, "step": 36889 }, { "epoch": 0.04, "learning_rate": 4.983317060121861e-05, "loss": 1.4119, "step": 36890 }, { "epoch": 0.04, "learning_rate": 4.983316154191672e-05, "loss": 1.0772, "step": 36891 }, { "epoch": 0.04, "learning_rate": 4.983315248236968e-05, "loss": 1.1458, "step": 36892 }, { "epoch": 0.04, "learning_rate": 4.98331434225775e-05, "loss": 1.1477, "step": 36893 }, { "epoch": 0.04, "learning_rate": 4.983313436254018e-05, "loss": 0.8542, "step": 36894 }, { "epoch": 0.04, "learning_rate": 4.983312530225771e-05, "loss": 1.92, "step": 36895 }, { "epoch": 0.04, "learning_rate": 4.9833116241730104e-05, "loss": 1.778, "step": 36896 }, { "epoch": 0.04, "learning_rate": 4.983310718095735e-05, "loss": 1.0704, "step": 36897 }, { "epoch": 0.04, "learning_rate": 4.9833098119939456e-05, "loss": 0.7534, "step": 36898 }, { "epoch": 0.04, "learning_rate": 4.983308905867643e-05, "loss": 0.7985, "step": 36899 }, { "epoch": 0.04, "learning_rate": 4.9833079997168255e-05, "loss": 0.663, "step": 36900 }, { "epoch": 0.04, "learning_rate": 4.983307093541494e-05, "loss": 0.6427, "step": 36901 }, { "epoch": 0.04, "learning_rate": 4.9833061873416474e-05, "loss": 0.5918, "step": 36902 }, { "epoch": 0.04, "learning_rate": 4.983305281117288e-05, "loss": 0.5693, "step": 36903 }, { "epoch": 0.04, "learning_rate": 4.983304374868414e-05, "loss": 0.5037, "step": 36904 }, { "epoch": 0.04, "learning_rate": 4.9833034685950254e-05, "loss": 1.0081, "step": 36905 }, { "epoch": 0.04, "learning_rate": 4.983302562297123e-05, "loss": 1.3745, "step": 36906 }, { "epoch": 0.04, "learning_rate": 4.983301655974706e-05, "loss": 0.9751, "step": 36907 }, { "epoch": 0.04, "learning_rate": 4.983300749627775e-05, "loss": 1.1809, "step": 36908 }, { "epoch": 0.04, "learning_rate": 4.9832998432563304e-05, "loss": 1.0226, "step": 36909 }, { "epoch": 0.04, "learning_rate": 4.983298936860372e-05, "loss": 1.1364, "step": 36910 }, { "epoch": 0.04, "learning_rate": 4.983298030439899e-05, "loss": 1.1925, "step": 36911 }, { "epoch": 0.04, "learning_rate": 4.983297123994912e-05, "loss": 1.3589, "step": 36912 }, { "epoch": 0.04, "learning_rate": 4.98329621752541e-05, "loss": 1.0044, "step": 36913 }, { "epoch": 0.04, "learning_rate": 4.983295311031396e-05, "loss": 0.6888, "step": 36914 }, { "epoch": 0.04, "learning_rate": 4.9832944045128664e-05, "loss": 1.6713, "step": 36915 }, { "epoch": 0.04, "learning_rate": 4.983293497969823e-05, "loss": 1.2617, "step": 36916 }, { "epoch": 0.04, "learning_rate": 4.983292591402266e-05, "loss": 0.9379, "step": 36917 }, { "epoch": 0.04, "learning_rate": 4.983291684810195e-05, "loss": 0.8377, "step": 36918 }, { "epoch": 0.04, "learning_rate": 4.9832907781936095e-05, "loss": 0.8497, "step": 36919 }, { "epoch": 0.04, "learning_rate": 4.98328987155251e-05, "loss": 0.6882, "step": 36920 }, { "epoch": 0.04, "learning_rate": 4.983288964886898e-05, "loss": 0.8182, "step": 36921 }, { "epoch": 0.04, "learning_rate": 4.98328805819677e-05, "loss": 0.9593, "step": 36922 }, { "epoch": 0.04, "learning_rate": 4.983287151482129e-05, "loss": 1.0801, "step": 36923 }, { "epoch": 0.04, "learning_rate": 4.9832862447429736e-05, "loss": 1.3052, "step": 36924 }, { "epoch": 0.04, "learning_rate": 4.9832853379793046e-05, "loss": 0.9283, "step": 36925 }, { "epoch": 0.04, "learning_rate": 4.983284431191121e-05, "loss": 1.0073, "step": 36926 }, { "epoch": 0.04, "learning_rate": 4.9832835243784245e-05, "loss": 0.9043, "step": 36927 }, { "epoch": 0.04, "learning_rate": 4.9832826175412134e-05, "loss": 0.8761, "step": 36928 }, { "epoch": 0.04, "learning_rate": 4.9832817106794885e-05, "loss": 0.7138, "step": 36929 }, { "epoch": 0.04, "learning_rate": 4.98328080379325e-05, "loss": 1.1774, "step": 36930 }, { "epoch": 0.04, "learning_rate": 4.9832798968824965e-05, "loss": 1.4419, "step": 36931 }, { "epoch": 0.04, "learning_rate": 4.98327898994723e-05, "loss": 1.2212, "step": 36932 }, { "epoch": 0.04, "learning_rate": 4.9832780829874506e-05, "loss": 0.7972, "step": 36933 }, { "epoch": 0.04, "learning_rate": 4.983277176003156e-05, "loss": 1.0358, "step": 36934 }, { "epoch": 0.04, "learning_rate": 4.9832762689943475e-05, "loss": 1.2322, "step": 36935 }, { "epoch": 0.04, "learning_rate": 4.983275361961026e-05, "loss": 1.1425, "step": 36936 }, { "epoch": 0.04, "learning_rate": 4.98327445490319e-05, "loss": 1.0652, "step": 36937 }, { "epoch": 0.04, "learning_rate": 4.98327354782084e-05, "loss": 1.1425, "step": 36938 }, { "epoch": 0.04, "learning_rate": 4.983272640713976e-05, "loss": 0.7006, "step": 36939 }, { "epoch": 0.04, "learning_rate": 4.983271733582599e-05, "loss": 0.8883, "step": 36940 }, { "epoch": 0.04, "learning_rate": 4.983270826426708e-05, "loss": 0.8184, "step": 36941 }, { "epoch": 0.04, "learning_rate": 4.983269919246303e-05, "loss": 0.9861, "step": 36942 }, { "epoch": 0.04, "learning_rate": 4.983269012041384e-05, "loss": 0.9146, "step": 36943 }, { "epoch": 0.04, "learning_rate": 4.983268104811951e-05, "loss": 0.9482, "step": 36944 }, { "epoch": 0.04, "learning_rate": 4.9832671975580045e-05, "loss": 1.0169, "step": 36945 }, { "epoch": 0.04, "learning_rate": 4.983266290279545e-05, "loss": 0.9256, "step": 36946 }, { "epoch": 0.04, "learning_rate": 4.9832653829765706e-05, "loss": 0.8774, "step": 36947 }, { "epoch": 0.04, "learning_rate": 4.983264475649083e-05, "loss": 1.2414, "step": 36948 }, { "epoch": 0.04, "learning_rate": 4.983263568297082e-05, "loss": 1.1182, "step": 36949 }, { "epoch": 0.04, "learning_rate": 4.983262660920567e-05, "loss": 0.9541, "step": 36950 }, { "epoch": 0.04, "learning_rate": 4.9832617535195384e-05, "loss": 0.9535, "step": 36951 }, { "epoch": 0.04, "learning_rate": 4.983260846093996e-05, "loss": 1.0735, "step": 36952 }, { "epoch": 0.04, "learning_rate": 4.9832599386439394e-05, "loss": 1.4105, "step": 36953 }, { "epoch": 0.04, "learning_rate": 4.983259031169369e-05, "loss": 1.0232, "step": 36954 }, { "epoch": 0.04, "learning_rate": 4.983258123670286e-05, "loss": 0.8404, "step": 36955 }, { "epoch": 0.04, "learning_rate": 4.983257216146689e-05, "loss": 0.8381, "step": 36956 }, { "epoch": 0.04, "learning_rate": 4.983256308598577e-05, "loss": 1.0837, "step": 36957 }, { "epoch": 0.04, "learning_rate": 4.983255401025953e-05, "loss": 1.1877, "step": 36958 }, { "epoch": 0.04, "learning_rate": 4.983254493428815e-05, "loss": 1.0531, "step": 36959 }, { "epoch": 0.04, "learning_rate": 4.9832535858071626e-05, "loss": 1.0451, "step": 36960 }, { "epoch": 0.04, "learning_rate": 4.9832526781609964e-05, "loss": 1.2923, "step": 36961 }, { "epoch": 0.04, "learning_rate": 4.983251770490318e-05, "loss": 0.9204, "step": 36962 }, { "epoch": 0.04, "learning_rate": 4.983250862795125e-05, "loss": 0.6064, "step": 36963 }, { "epoch": 0.04, "learning_rate": 4.983249955075418e-05, "loss": 1.2597, "step": 36964 }, { "epoch": 0.04, "learning_rate": 4.983249047331199e-05, "loss": 1.1686, "step": 36965 }, { "epoch": 0.04, "learning_rate": 4.983248139562465e-05, "loss": 1.2909, "step": 36966 }, { "epoch": 0.04, "learning_rate": 4.983247231769218e-05, "loss": 1.306, "step": 36967 }, { "epoch": 0.04, "learning_rate": 4.983246323951457e-05, "loss": 1.1465, "step": 36968 }, { "epoch": 0.04, "learning_rate": 4.983245416109183e-05, "loss": 1.083, "step": 36969 }, { "epoch": 0.04, "learning_rate": 4.983244508242395e-05, "loss": 0.9662, "step": 36970 }, { "epoch": 0.04, "learning_rate": 4.983243600351094e-05, "loss": 0.706, "step": 36971 }, { "epoch": 0.04, "learning_rate": 4.983242692435279e-05, "loss": 0.8063, "step": 36972 }, { "epoch": 0.04, "learning_rate": 4.9832417844949504e-05, "loss": 0.7978, "step": 36973 }, { "epoch": 0.04, "learning_rate": 4.983240876530109e-05, "loss": 1.019, "step": 36974 }, { "epoch": 0.04, "learning_rate": 4.9832399685407526e-05, "loss": 1.0451, "step": 36975 }, { "epoch": 0.04, "learning_rate": 4.983239060526884e-05, "loss": 1.0261, "step": 36976 }, { "epoch": 0.04, "learning_rate": 4.9832381524885015e-05, "loss": 1.1023, "step": 36977 }, { "epoch": 0.04, "learning_rate": 4.983237244425606e-05, "loss": 0.9134, "step": 36978 }, { "epoch": 0.04, "learning_rate": 4.983236336338196e-05, "loss": 0.9541, "step": 36979 }, { "epoch": 0.04, "learning_rate": 4.9832354282262733e-05, "loss": 1.0336, "step": 36980 }, { "epoch": 0.04, "learning_rate": 4.983234520089837e-05, "loss": 0.8777, "step": 36981 }, { "epoch": 0.04, "learning_rate": 4.983233611928887e-05, "loss": 0.8228, "step": 36982 }, { "epoch": 0.04, "learning_rate": 4.983232703743424e-05, "loss": 1.0182, "step": 36983 }, { "epoch": 0.04, "learning_rate": 4.983231795533447e-05, "loss": 1.0287, "step": 36984 }, { "epoch": 0.04, "learning_rate": 4.9832308872989576e-05, "loss": 0.8631, "step": 36985 }, { "epoch": 0.04, "learning_rate": 4.9832299790399536e-05, "loss": 0.8082, "step": 36986 }, { "epoch": 0.04, "learning_rate": 4.9832290707564365e-05, "loss": 1.0037, "step": 36987 }, { "epoch": 0.04, "learning_rate": 4.983228162448407e-05, "loss": 0.9391, "step": 36988 }, { "epoch": 0.04, "learning_rate": 4.9832272541158634e-05, "loss": 0.8964, "step": 36989 }, { "epoch": 0.04, "learning_rate": 4.9832263457588055e-05, "loss": 1.101, "step": 36990 }, { "epoch": 0.04, "learning_rate": 4.983225437377236e-05, "loss": 1.066, "step": 36991 }, { "epoch": 0.04, "learning_rate": 4.983224528971152e-05, "loss": 0.9306, "step": 36992 }, { "epoch": 0.04, "learning_rate": 4.983223620540555e-05, "loss": 0.7544, "step": 36993 }, { "epoch": 0.04, "learning_rate": 4.9832227120854445e-05, "loss": 1.0045, "step": 36994 }, { "epoch": 0.04, "learning_rate": 4.983221803605821e-05, "loss": 0.7731, "step": 36995 }, { "epoch": 0.04, "learning_rate": 4.9832208951016835e-05, "loss": 1.1674, "step": 36996 }, { "epoch": 0.04, "learning_rate": 4.9832199865730337e-05, "loss": 0.8587, "step": 36997 }, { "epoch": 0.04, "learning_rate": 4.983219078019869e-05, "loss": 0.9451, "step": 36998 }, { "epoch": 0.04, "learning_rate": 4.9832181694421925e-05, "loss": 1.1191, "step": 36999 }, { "epoch": 0.04, "learning_rate": 4.9832172608400026e-05, "loss": 0.9969, "step": 37000 }, { "epoch": 0.04, "eval_loss": 1.0708471536636353, "eval_runtime": 125.1415, "eval_samples_per_second": 11.067, "eval_steps_per_second": 5.538, "step": 37000 }, { "epoch": 0.04, "learning_rate": 4.983216352213299e-05, "loss": 0.9737, "step": 37001 }, { "epoch": 0.04, "learning_rate": 4.983215443562082e-05, "loss": 0.9251, "step": 37002 }, { "epoch": 0.04, "learning_rate": 4.983214534886352e-05, "loss": 1.1195, "step": 37003 }, { "epoch": 0.04, "learning_rate": 4.983213626186108e-05, "loss": 0.9834, "step": 37004 }, { "epoch": 0.04, "learning_rate": 4.983212717461352e-05, "loss": 0.7544, "step": 37005 }, { "epoch": 0.04, "learning_rate": 4.9832118087120824e-05, "loss": 0.797, "step": 37006 }, { "epoch": 0.04, "learning_rate": 4.983210899938299e-05, "loss": 0.9263, "step": 37007 }, { "epoch": 0.04, "learning_rate": 4.983209991140003e-05, "loss": 0.8319, "step": 37008 }, { "epoch": 0.04, "learning_rate": 4.983209082317194e-05, "loss": 1.1437, "step": 37009 }, { "epoch": 0.04, "learning_rate": 4.983208173469871e-05, "loss": 0.979, "step": 37010 }, { "epoch": 0.04, "learning_rate": 4.983207264598036e-05, "loss": 0.94, "step": 37011 }, { "epoch": 0.04, "learning_rate": 4.983206355701686e-05, "loss": 1.0042, "step": 37012 }, { "epoch": 0.04, "learning_rate": 4.983205446780824e-05, "loss": 0.7373, "step": 37013 }, { "epoch": 0.04, "learning_rate": 4.983204537835449e-05, "loss": 1.4939, "step": 37014 }, { "epoch": 0.04, "learning_rate": 4.9832036288655606e-05, "loss": 0.854, "step": 37015 }, { "epoch": 0.04, "learning_rate": 4.983202719871159e-05, "loss": 1.0409, "step": 37016 }, { "epoch": 0.04, "learning_rate": 4.9832018108522447e-05, "loss": 1.3098, "step": 37017 }, { "epoch": 0.04, "learning_rate": 4.983200901808816e-05, "loss": 1.0637, "step": 37018 }, { "epoch": 0.04, "learning_rate": 4.9831999927408755e-05, "loss": 0.5199, "step": 37019 }, { "epoch": 0.04, "learning_rate": 4.9831990836484215e-05, "loss": 0.509, "step": 37020 }, { "epoch": 0.04, "learning_rate": 4.9831981745314544e-05, "loss": 0.9498, "step": 37021 }, { "epoch": 0.04, "learning_rate": 4.983197265389974e-05, "loss": 0.9969, "step": 37022 }, { "epoch": 0.04, "learning_rate": 4.983196356223981e-05, "loss": 1.0809, "step": 37023 }, { "epoch": 0.04, "learning_rate": 4.983195447033474e-05, "loss": 1.1254, "step": 37024 }, { "epoch": 0.04, "learning_rate": 4.9831945378184554e-05, "loss": 1.185, "step": 37025 }, { "epoch": 0.04, "learning_rate": 4.9831936285789226e-05, "loss": 1.4095, "step": 37026 }, { "epoch": 0.04, "learning_rate": 4.9831927193148774e-05, "loss": 1.1762, "step": 37027 }, { "epoch": 0.04, "learning_rate": 4.9831918100263184e-05, "loss": 0.9402, "step": 37028 }, { "epoch": 0.04, "learning_rate": 4.983190900713247e-05, "loss": 0.99, "step": 37029 }, { "epoch": 0.04, "learning_rate": 4.983189991375662e-05, "loss": 1.0937, "step": 37030 }, { "epoch": 0.04, "learning_rate": 4.983189082013565e-05, "loss": 1.0125, "step": 37031 }, { "epoch": 0.04, "learning_rate": 4.983188172626955e-05, "loss": 0.8702, "step": 37032 }, { "epoch": 0.04, "learning_rate": 4.98318726321583e-05, "loss": 0.8111, "step": 37033 }, { "epoch": 0.04, "learning_rate": 4.9831863537801945e-05, "loss": 1.0178, "step": 37034 }, { "epoch": 0.04, "learning_rate": 4.983185444320044e-05, "loss": 0.7566, "step": 37035 }, { "epoch": 0.04, "learning_rate": 4.983184534835382e-05, "loss": 0.8191, "step": 37036 }, { "epoch": 0.04, "learning_rate": 4.983183625326206e-05, "loss": 1.0536, "step": 37037 }, { "epoch": 0.04, "learning_rate": 4.983182715792518e-05, "loss": 0.9071, "step": 37038 }, { "epoch": 0.04, "learning_rate": 4.983181806234317e-05, "loss": 1.1715, "step": 37039 }, { "epoch": 0.04, "learning_rate": 4.9831808966516026e-05, "loss": 0.7357, "step": 37040 }, { "epoch": 0.04, "learning_rate": 4.9831799870443756e-05, "loss": 0.5577, "step": 37041 }, { "epoch": 0.04, "learning_rate": 4.9831790774126354e-05, "loss": 0.602, "step": 37042 }, { "epoch": 0.04, "learning_rate": 4.983178167756383e-05, "loss": 0.8586, "step": 37043 }, { "epoch": 0.04, "learning_rate": 4.983177258075617e-05, "loss": 0.7452, "step": 37044 }, { "epoch": 0.04, "learning_rate": 4.983176348370338e-05, "loss": 0.5757, "step": 37045 }, { "epoch": 0.04, "learning_rate": 4.983175438640547e-05, "loss": 1.0904, "step": 37046 }, { "epoch": 0.04, "learning_rate": 4.983174528886242e-05, "loss": 1.215, "step": 37047 }, { "epoch": 0.04, "learning_rate": 4.983173619107425e-05, "loss": 0.9451, "step": 37048 }, { "epoch": 0.04, "learning_rate": 4.9831727093040945e-05, "loss": 1.0658, "step": 37049 }, { "epoch": 0.04, "learning_rate": 4.983171799476252e-05, "loss": 0.9416, "step": 37050 }, { "epoch": 0.04, "learning_rate": 4.983170889623896e-05, "loss": 0.8319, "step": 37051 }, { "epoch": 0.04, "learning_rate": 4.9831699797470275e-05, "loss": 0.9028, "step": 37052 }, { "epoch": 0.04, "learning_rate": 4.9831690698456455e-05, "loss": 1.1375, "step": 37053 }, { "epoch": 0.04, "learning_rate": 4.983168159919752e-05, "loss": 1.2942, "step": 37054 }, { "epoch": 0.04, "learning_rate": 4.983167249969345e-05, "loss": 1.0182, "step": 37055 }, { "epoch": 0.04, "learning_rate": 4.983166339994425e-05, "loss": 1.2371, "step": 37056 }, { "epoch": 0.04, "learning_rate": 4.9831654299949925e-05, "loss": 1.449, "step": 37057 }, { "epoch": 0.04, "learning_rate": 4.983164519971047e-05, "loss": 1.0411, "step": 37058 }, { "epoch": 0.04, "learning_rate": 4.983163609922589e-05, "loss": 1.0434, "step": 37059 }, { "epoch": 0.04, "learning_rate": 4.9831626998496184e-05, "loss": 0.8422, "step": 37060 }, { "epoch": 0.04, "learning_rate": 4.983161789752134e-05, "loss": 0.5488, "step": 37061 }, { "epoch": 0.04, "learning_rate": 4.983160879630138e-05, "loss": 0.3564, "step": 37062 }, { "epoch": 0.04, "learning_rate": 4.983159969483629e-05, "loss": 0.1234, "step": 37063 }, { "epoch": 0.04, "learning_rate": 4.983159059312608e-05, "loss": 0.7602, "step": 37064 }, { "epoch": 0.04, "learning_rate": 4.983158149117073e-05, "loss": 0.7228, "step": 37065 }, { "epoch": 0.04, "learning_rate": 4.983157238897026e-05, "loss": 0.8569, "step": 37066 }, { "epoch": 0.04, "learning_rate": 4.983156328652466e-05, "loss": 1.1943, "step": 37067 }, { "epoch": 0.04, "learning_rate": 4.983155418383394e-05, "loss": 0.6971, "step": 37068 }, { "epoch": 0.04, "learning_rate": 4.9831545080898086e-05, "loss": 1.0632, "step": 37069 }, { "epoch": 0.04, "learning_rate": 4.98315359777171e-05, "loss": 0.8471, "step": 37070 }, { "epoch": 0.04, "learning_rate": 4.9831526874291e-05, "loss": 1.4911, "step": 37071 }, { "epoch": 0.04, "learning_rate": 4.983151777061977e-05, "loss": 1.2158, "step": 37072 }, { "epoch": 0.04, "learning_rate": 4.9831508666703415e-05, "loss": 1.0548, "step": 37073 }, { "epoch": 0.04, "learning_rate": 4.9831499562541925e-05, "loss": 1.1568, "step": 37074 }, { "epoch": 0.04, "learning_rate": 4.983149045813532e-05, "loss": 0.9809, "step": 37075 }, { "epoch": 0.04, "learning_rate": 4.9831481353483586e-05, "loss": 1.1139, "step": 37076 }, { "epoch": 0.04, "learning_rate": 4.9831472248586723e-05, "loss": 0.6102, "step": 37077 }, { "epoch": 0.04, "learning_rate": 4.983146314344474e-05, "loss": 0.6285, "step": 37078 }, { "epoch": 0.04, "learning_rate": 4.9831454038057624e-05, "loss": 0.4838, "step": 37079 }, { "epoch": 0.04, "learning_rate": 4.983144493242539e-05, "loss": 0.7148, "step": 37080 }, { "epoch": 0.04, "learning_rate": 4.9831435826548026e-05, "loss": 0.7937, "step": 37081 }, { "epoch": 0.04, "learning_rate": 4.983142672042553e-05, "loss": 0.6947, "step": 37082 }, { "epoch": 0.04, "learning_rate": 4.9831417614057916e-05, "loss": 1.1308, "step": 37083 }, { "epoch": 0.04, "learning_rate": 4.9831408507445174e-05, "loss": 0.9788, "step": 37084 }, { "epoch": 0.04, "learning_rate": 4.9831399400587315e-05, "loss": 1.1711, "step": 37085 }, { "epoch": 0.04, "learning_rate": 4.9831390293484324e-05, "loss": 1.0112, "step": 37086 }, { "epoch": 0.04, "learning_rate": 4.983138118613621e-05, "loss": 1.0968, "step": 37087 }, { "epoch": 0.04, "learning_rate": 4.983137207854297e-05, "loss": 1.1584, "step": 37088 }, { "epoch": 0.04, "learning_rate": 4.9831362970704604e-05, "loss": 1.7219, "step": 37089 }, { "epoch": 0.04, "learning_rate": 4.983135386262111e-05, "loss": 1.6138, "step": 37090 }, { "epoch": 0.04, "learning_rate": 4.9831344754292494e-05, "loss": 1.2035, "step": 37091 }, { "epoch": 0.04, "learning_rate": 4.9831335645718756e-05, "loss": 1.1442, "step": 37092 }, { "epoch": 0.04, "learning_rate": 4.9831326536899894e-05, "loss": 1.0054, "step": 37093 }, { "epoch": 0.04, "learning_rate": 4.983131742783591e-05, "loss": 0.9216, "step": 37094 }, { "epoch": 0.04, "learning_rate": 4.9831308318526794e-05, "loss": 0.9446, "step": 37095 }, { "epoch": 0.04, "learning_rate": 4.983129920897256e-05, "loss": 0.9727, "step": 37096 }, { "epoch": 0.04, "learning_rate": 4.98312900991732e-05, "loss": 0.6697, "step": 37097 }, { "epoch": 0.04, "learning_rate": 4.983128098912871e-05, "loss": 0.1486, "step": 37098 }, { "epoch": 0.04, "learning_rate": 4.98312718788391e-05, "loss": 0.1014, "step": 37099 }, { "epoch": 0.04, "learning_rate": 4.983126276830437e-05, "loss": 0.1094, "step": 37100 }, { "epoch": 0.04, "learning_rate": 4.9831253657524514e-05, "loss": 0.1058, "step": 37101 }, { "epoch": 0.04, "learning_rate": 4.983124454649953e-05, "loss": 0.1476, "step": 37102 }, { "epoch": 0.04, "learning_rate": 4.983123543522943e-05, "loss": 0.7706, "step": 37103 }, { "epoch": 0.04, "learning_rate": 4.98312263237142e-05, "loss": 1.3213, "step": 37104 }, { "epoch": 0.04, "learning_rate": 4.983121721195385e-05, "loss": 0.886, "step": 37105 }, { "epoch": 0.04, "learning_rate": 4.983120809994838e-05, "loss": 1.0118, "step": 37106 }, { "epoch": 0.04, "learning_rate": 4.983119898769778e-05, "loss": 1.0911, "step": 37107 }, { "epoch": 0.04, "learning_rate": 4.9831189875202066e-05, "loss": 0.7217, "step": 37108 }, { "epoch": 0.04, "learning_rate": 4.983118076246122e-05, "loss": 1.1211, "step": 37109 }, { "epoch": 0.04, "learning_rate": 4.983117164947525e-05, "loss": 1.1343, "step": 37110 }, { "epoch": 0.04, "learning_rate": 4.9831162536244166e-05, "loss": 1.0611, "step": 37111 }, { "epoch": 0.04, "learning_rate": 4.983115342276795e-05, "loss": 0.7942, "step": 37112 }, { "epoch": 0.04, "learning_rate": 4.983114430904662e-05, "loss": 1.0665, "step": 37113 }, { "epoch": 0.04, "learning_rate": 4.983113519508016e-05, "loss": 1.018, "step": 37114 }, { "epoch": 0.04, "learning_rate": 4.983112608086858e-05, "loss": 1.096, "step": 37115 }, { "epoch": 0.04, "learning_rate": 4.983111696641188e-05, "loss": 0.9788, "step": 37116 }, { "epoch": 0.04, "learning_rate": 4.983110785171006e-05, "loss": 1.5928, "step": 37117 }, { "epoch": 0.04, "learning_rate": 4.983109873676311e-05, "loss": 0.7687, "step": 37118 }, { "epoch": 0.04, "learning_rate": 4.983108962157103e-05, "loss": 1.072, "step": 37119 }, { "epoch": 0.04, "learning_rate": 4.9831080506133855e-05, "loss": 1.1903, "step": 37120 }, { "epoch": 0.04, "learning_rate": 4.983107139045153e-05, "loss": 1.0856, "step": 37121 }, { "epoch": 0.04, "learning_rate": 4.9831062274524104e-05, "loss": 1.7782, "step": 37122 }, { "epoch": 0.04, "learning_rate": 4.9831053158351545e-05, "loss": 1.4931, "step": 37123 }, { "epoch": 0.04, "learning_rate": 4.9831044041933875e-05, "loss": 1.3828, "step": 37124 }, { "epoch": 0.04, "learning_rate": 4.983103492527107e-05, "loss": 1.0567, "step": 37125 }, { "epoch": 0.04, "learning_rate": 4.983102580836315e-05, "loss": 0.9978, "step": 37126 }, { "epoch": 0.04, "learning_rate": 4.983101669121011e-05, "loss": 0.9263, "step": 37127 }, { "epoch": 0.04, "learning_rate": 4.983100757381194e-05, "loss": 0.8946, "step": 37128 }, { "epoch": 0.04, "learning_rate": 4.983099845616865e-05, "loss": 1.2699, "step": 37129 }, { "epoch": 0.04, "learning_rate": 4.983098933828025e-05, "loss": 1.3854, "step": 37130 }, { "epoch": 0.04, "learning_rate": 4.9830980220146726e-05, "loss": 1.3102, "step": 37131 }, { "epoch": 0.04, "learning_rate": 4.983097110176807e-05, "loss": 1.2847, "step": 37132 }, { "epoch": 0.04, "learning_rate": 4.983096198314431e-05, "loss": 0.9375, "step": 37133 }, { "epoch": 0.04, "learning_rate": 4.9830952864275414e-05, "loss": 1.0293, "step": 37134 }, { "epoch": 0.04, "learning_rate": 4.983094374516141e-05, "loss": 1.0099, "step": 37135 }, { "epoch": 0.04, "learning_rate": 4.983093462580227e-05, "loss": 0.8977, "step": 37136 }, { "epoch": 0.04, "learning_rate": 4.983092550619802e-05, "loss": 1.1213, "step": 37137 }, { "epoch": 0.04, "learning_rate": 4.983091638634865e-05, "loss": 0.3391, "step": 37138 }, { "epoch": 0.04, "learning_rate": 4.983090726625416e-05, "loss": 1.1347, "step": 37139 }, { "epoch": 0.04, "learning_rate": 4.983089814591454e-05, "loss": 0.9314, "step": 37140 }, { "epoch": 0.04, "learning_rate": 4.983088902532981e-05, "loss": 1.0865, "step": 37141 }, { "epoch": 0.04, "learning_rate": 4.983087990449996e-05, "loss": 1.0551, "step": 37142 }, { "epoch": 0.04, "learning_rate": 4.983087078342499e-05, "loss": 0.7133, "step": 37143 }, { "epoch": 0.04, "learning_rate": 4.983086166210489e-05, "loss": 1.228, "step": 37144 }, { "epoch": 0.04, "learning_rate": 4.983085254053967e-05, "loss": 0.8772, "step": 37145 }, { "epoch": 0.04, "learning_rate": 4.983084341872934e-05, "loss": 1.0035, "step": 37146 }, { "epoch": 0.04, "learning_rate": 4.983083429667389e-05, "loss": 1.0922, "step": 37147 }, { "epoch": 0.04, "learning_rate": 4.983082517437332e-05, "loss": 0.9889, "step": 37148 }, { "epoch": 0.04, "learning_rate": 4.9830816051827636e-05, "loss": 1.1327, "step": 37149 }, { "epoch": 0.04, "learning_rate": 4.983080692903682e-05, "loss": 0.969, "step": 37150 }, { "epoch": 0.04, "learning_rate": 4.983079780600089e-05, "loss": 0.9858, "step": 37151 }, { "epoch": 0.04, "learning_rate": 4.983078868271984e-05, "loss": 1.025, "step": 37152 }, { "epoch": 0.04, "learning_rate": 4.9830779559193676e-05, "loss": 1.0544, "step": 37153 }, { "epoch": 0.04, "learning_rate": 4.9830770435422384e-05, "loss": 1.3377, "step": 37154 }, { "epoch": 0.04, "learning_rate": 4.983076131140598e-05, "loss": 0.9999, "step": 37155 }, { "epoch": 0.04, "learning_rate": 4.983075218714446e-05, "loss": 1.7338, "step": 37156 }, { "epoch": 0.04, "learning_rate": 4.983074306263782e-05, "loss": 1.3804, "step": 37157 }, { "epoch": 0.04, "learning_rate": 4.9830733937886054e-05, "loss": 0.9534, "step": 37158 }, { "epoch": 0.04, "learning_rate": 4.9830724812889174e-05, "loss": 0.8779, "step": 37159 }, { "epoch": 0.04, "learning_rate": 4.9830715687647176e-05, "loss": 0.7322, "step": 37160 }, { "epoch": 0.04, "learning_rate": 4.983070656216006e-05, "loss": 1.4618, "step": 37161 }, { "epoch": 0.04, "learning_rate": 4.983069743642782e-05, "loss": 0.8874, "step": 37162 }, { "epoch": 0.04, "learning_rate": 4.983068831045047e-05, "loss": 0.8696, "step": 37163 }, { "epoch": 0.04, "learning_rate": 4.983067918422799e-05, "loss": 0.8591, "step": 37164 }, { "epoch": 0.04, "learning_rate": 4.9830670057760405e-05, "loss": 1.1636, "step": 37165 }, { "epoch": 0.04, "learning_rate": 4.983066093104769e-05, "loss": 1.0825, "step": 37166 }, { "epoch": 0.04, "learning_rate": 4.9830651804089864e-05, "loss": 0.9028, "step": 37167 }, { "epoch": 0.04, "learning_rate": 4.983064267688693e-05, "loss": 1.2529, "step": 37168 }, { "epoch": 0.04, "learning_rate": 4.9830633549438865e-05, "loss": 0.8118, "step": 37169 }, { "epoch": 0.04, "learning_rate": 4.983062442174569e-05, "loss": 0.7312, "step": 37170 }, { "epoch": 0.04, "learning_rate": 4.983061529380739e-05, "loss": 0.9203, "step": 37171 }, { "epoch": 0.04, "learning_rate": 4.983060616562398e-05, "loss": 0.9994, "step": 37172 }, { "epoch": 0.04, "learning_rate": 4.983059703719545e-05, "loss": 1.006, "step": 37173 }, { "epoch": 0.04, "learning_rate": 4.9830587908521794e-05, "loss": 2.2062, "step": 37174 }, { "epoch": 0.04, "learning_rate": 4.983057877960303e-05, "loss": 1.5178, "step": 37175 }, { "epoch": 0.04, "learning_rate": 4.9830569650439154e-05, "loss": 1.0704, "step": 37176 }, { "epoch": 0.04, "learning_rate": 4.983056052103015e-05, "loss": 1.0759, "step": 37177 }, { "epoch": 0.04, "learning_rate": 4.9830551391376035e-05, "loss": 0.8348, "step": 37178 }, { "epoch": 0.04, "learning_rate": 4.9830542261476806e-05, "loss": 0.6089, "step": 37179 }, { "epoch": 0.04, "learning_rate": 4.983053313133246e-05, "loss": 0.9969, "step": 37180 }, { "epoch": 0.04, "learning_rate": 4.9830524000942994e-05, "loss": 0.8839, "step": 37181 }, { "epoch": 0.04, "learning_rate": 4.983051487030841e-05, "loss": 0.8493, "step": 37182 }, { "epoch": 0.04, "learning_rate": 4.983050573942871e-05, "loss": 1.1578, "step": 37183 }, { "epoch": 0.04, "learning_rate": 4.9830496608303886e-05, "loss": 1.3786, "step": 37184 }, { "epoch": 0.04, "learning_rate": 4.9830487476933964e-05, "loss": 1.5479, "step": 37185 }, { "epoch": 0.04, "learning_rate": 4.983047834531892e-05, "loss": 1.3355, "step": 37186 }, { "epoch": 0.04, "learning_rate": 4.983046921345875e-05, "loss": 0.9387, "step": 37187 }, { "epoch": 0.04, "learning_rate": 4.983046008135347e-05, "loss": 1.2502, "step": 37188 }, { "epoch": 0.04, "learning_rate": 4.9830450949003076e-05, "loss": 1.2744, "step": 37189 }, { "epoch": 0.04, "learning_rate": 4.983044181640757e-05, "loss": 1.1792, "step": 37190 }, { "epoch": 0.04, "learning_rate": 4.983043268356694e-05, "loss": 1.1053, "step": 37191 }, { "epoch": 0.04, "learning_rate": 4.9830423550481196e-05, "loss": 1.372, "step": 37192 }, { "epoch": 0.04, "learning_rate": 4.983041441715034e-05, "loss": 0.9316, "step": 37193 }, { "epoch": 0.04, "learning_rate": 4.983040528357437e-05, "loss": 1.0529, "step": 37194 }, { "epoch": 0.04, "learning_rate": 4.9830396149753276e-05, "loss": 0.9536, "step": 37195 }, { "epoch": 0.04, "learning_rate": 4.983038701568707e-05, "loss": 0.8556, "step": 37196 }, { "epoch": 0.04, "learning_rate": 4.9830377881375756e-05, "loss": 0.9784, "step": 37197 }, { "epoch": 0.04, "learning_rate": 4.983036874681932e-05, "loss": 0.9089, "step": 37198 }, { "epoch": 0.04, "learning_rate": 4.983035961201777e-05, "loss": 1.1029, "step": 37199 }, { "epoch": 0.04, "learning_rate": 4.983035047697111e-05, "loss": 0.4404, "step": 37200 }, { "epoch": 0.04, "learning_rate": 4.983034134167933e-05, "loss": 0.2576, "step": 37201 }, { "epoch": 0.04, "learning_rate": 4.9830332206142446e-05, "loss": 0.9426, "step": 37202 }, { "epoch": 0.04, "learning_rate": 4.983032307036044e-05, "loss": 0.8975, "step": 37203 }, { "epoch": 0.04, "learning_rate": 4.983031393433332e-05, "loss": 1.0942, "step": 37204 }, { "epoch": 0.04, "learning_rate": 4.983030479806108e-05, "loss": 0.8694, "step": 37205 }, { "epoch": 0.04, "learning_rate": 4.9830295661543726e-05, "loss": 0.9178, "step": 37206 }, { "epoch": 0.04, "learning_rate": 4.9830286524781266e-05, "loss": 0.9669, "step": 37207 }, { "epoch": 0.04, "learning_rate": 4.983027738777368e-05, "loss": 1.4771, "step": 37208 }, { "epoch": 0.04, "learning_rate": 4.9830268250520994e-05, "loss": 1.108, "step": 37209 }, { "epoch": 0.04, "learning_rate": 4.983025911302318e-05, "loss": 0.8844, "step": 37210 }, { "epoch": 0.04, "learning_rate": 4.9830249975280264e-05, "loss": 1.2477, "step": 37211 }, { "epoch": 0.04, "learning_rate": 4.983024083729223e-05, "loss": 1.6851, "step": 37212 }, { "epoch": 0.04, "learning_rate": 4.9830231699059084e-05, "loss": 1.2402, "step": 37213 }, { "epoch": 0.04, "learning_rate": 4.983022256058082e-05, "loss": 0.7785, "step": 37214 }, { "epoch": 0.04, "learning_rate": 4.983021342185744e-05, "loss": 1.0528, "step": 37215 }, { "epoch": 0.04, "learning_rate": 4.983020428288896e-05, "loss": 1.101, "step": 37216 }, { "epoch": 0.04, "learning_rate": 4.983019514367535e-05, "loss": 1.1284, "step": 37217 }, { "epoch": 0.04, "learning_rate": 4.9830186004216636e-05, "loss": 1.0795, "step": 37218 }, { "epoch": 0.04, "learning_rate": 4.983017686451281e-05, "loss": 1.0727, "step": 37219 }, { "epoch": 0.04, "learning_rate": 4.983016772456387e-05, "loss": 1.2518, "step": 37220 }, { "epoch": 0.04, "learning_rate": 4.9830158584369814e-05, "loss": 1.0589, "step": 37221 }, { "epoch": 0.04, "learning_rate": 4.983014944393065e-05, "loss": 1.0621, "step": 37222 }, { "epoch": 0.04, "learning_rate": 4.983014030324637e-05, "loss": 1.0302, "step": 37223 }, { "epoch": 0.04, "learning_rate": 4.983013116231697e-05, "loss": 1.0525, "step": 37224 }, { "epoch": 0.04, "learning_rate": 4.9830122021142474e-05, "loss": 0.9058, "step": 37225 }, { "epoch": 0.04, "learning_rate": 4.983011287972286e-05, "loss": 1.0538, "step": 37226 }, { "epoch": 0.04, "learning_rate": 4.983010373805812e-05, "loss": 1.0014, "step": 37227 }, { "epoch": 0.04, "learning_rate": 4.983009459614828e-05, "loss": 0.7622, "step": 37228 }, { "epoch": 0.04, "learning_rate": 4.9830085453993324e-05, "loss": 0.514, "step": 37229 }, { "epoch": 0.04, "learning_rate": 4.983007631159326e-05, "loss": 0.8012, "step": 37230 }, { "epoch": 0.04, "learning_rate": 4.983006716894808e-05, "loss": 0.9157, "step": 37231 }, { "epoch": 0.04, "learning_rate": 4.983005802605779e-05, "loss": 1.0883, "step": 37232 }, { "epoch": 0.04, "learning_rate": 4.9830048882922384e-05, "loss": 1.037, "step": 37233 }, { "epoch": 0.04, "learning_rate": 4.983003973954188e-05, "loss": 0.9878, "step": 37234 }, { "epoch": 0.04, "learning_rate": 4.983003059591625e-05, "loss": 1.2869, "step": 37235 }, { "epoch": 0.04, "learning_rate": 4.983002145204551e-05, "loss": 1.3243, "step": 37236 }, { "epoch": 0.04, "learning_rate": 4.983001230792966e-05, "loss": 1.076, "step": 37237 }, { "epoch": 0.04, "learning_rate": 4.98300031635687e-05, "loss": 1.2077, "step": 37238 }, { "epoch": 0.04, "learning_rate": 4.982999401896263e-05, "loss": 0.9647, "step": 37239 }, { "epoch": 0.04, "learning_rate": 4.9829984874111444e-05, "loss": 0.4348, "step": 37240 }, { "epoch": 0.04, "learning_rate": 4.9829975729015156e-05, "loss": 0.8925, "step": 37241 }, { "epoch": 0.04, "learning_rate": 4.982996658367375e-05, "loss": 1.0116, "step": 37242 }, { "epoch": 0.04, "learning_rate": 4.982995743808723e-05, "loss": 0.9373, "step": 37243 }, { "epoch": 0.04, "learning_rate": 4.98299482922556e-05, "loss": 1.0082, "step": 37244 }, { "epoch": 0.04, "learning_rate": 4.982993914617886e-05, "loss": 0.9786, "step": 37245 }, { "epoch": 0.04, "learning_rate": 4.982992999985702e-05, "loss": 0.8296, "step": 37246 }, { "epoch": 0.04, "learning_rate": 4.982992085329006e-05, "loss": 0.9245, "step": 37247 }, { "epoch": 0.04, "learning_rate": 4.9829911706477986e-05, "loss": 0.9429, "step": 37248 }, { "epoch": 0.04, "learning_rate": 4.9829902559420803e-05, "loss": 1.1214, "step": 37249 }, { "epoch": 0.04, "learning_rate": 4.982989341211852e-05, "loss": 1.2272, "step": 37250 }, { "epoch": 0.04, "learning_rate": 4.982988426457111e-05, "loss": 1.1335, "step": 37251 }, { "epoch": 0.04, "learning_rate": 4.98298751167786e-05, "loss": 0.9392, "step": 37252 }, { "epoch": 0.04, "learning_rate": 4.982986596874098e-05, "loss": 1.0155, "step": 37253 }, { "epoch": 0.04, "learning_rate": 4.982985682045825e-05, "loss": 1.2835, "step": 37254 }, { "epoch": 0.04, "learning_rate": 4.982984767193041e-05, "loss": 1.077, "step": 37255 }, { "epoch": 0.04, "learning_rate": 4.9829838523157465e-05, "loss": 1.2109, "step": 37256 }, { "epoch": 0.04, "learning_rate": 4.98298293741394e-05, "loss": 1.1714, "step": 37257 }, { "epoch": 0.04, "learning_rate": 4.982982022487623e-05, "loss": 1.1337, "step": 37258 }, { "epoch": 0.04, "learning_rate": 4.982981107536795e-05, "loss": 1.0369, "step": 37259 }, { "epoch": 0.04, "learning_rate": 4.982980192561456e-05, "loss": 1.4234, "step": 37260 }, { "epoch": 0.04, "learning_rate": 4.982979277561606e-05, "loss": 1.0189, "step": 37261 }, { "epoch": 0.04, "learning_rate": 4.982978362537246e-05, "loss": 0.8157, "step": 37262 }, { "epoch": 0.04, "learning_rate": 4.982977447488374e-05, "loss": 1.0641, "step": 37263 }, { "epoch": 0.04, "learning_rate": 4.982976532414991e-05, "loss": 1.0226, "step": 37264 }, { "epoch": 0.04, "learning_rate": 4.982975617317098e-05, "loss": 0.8365, "step": 37265 }, { "epoch": 0.04, "learning_rate": 4.982974702194694e-05, "loss": 0.6841, "step": 37266 }, { "epoch": 0.04, "learning_rate": 4.9829737870477785e-05, "loss": 1.0007, "step": 37267 }, { "epoch": 0.04, "learning_rate": 4.982972871876353e-05, "loss": 0.8509, "step": 37268 }, { "epoch": 0.04, "learning_rate": 4.982971956680416e-05, "loss": 0.8774, "step": 37269 }, { "epoch": 0.04, "learning_rate": 4.982971041459967e-05, "loss": 0.8619, "step": 37270 }, { "epoch": 0.04, "learning_rate": 4.982970126215009e-05, "loss": 0.9694, "step": 37271 }, { "epoch": 0.04, "learning_rate": 4.982969210945539e-05, "loss": 1.9738, "step": 37272 }, { "epoch": 0.04, "learning_rate": 4.982968295651559e-05, "loss": 1.1949, "step": 37273 }, { "epoch": 0.04, "learning_rate": 4.9829673803330677e-05, "loss": 1.0747, "step": 37274 }, { "epoch": 0.04, "learning_rate": 4.982966464990066e-05, "loss": 1.0532, "step": 37275 }, { "epoch": 0.04, "learning_rate": 4.982965549622553e-05, "loss": 1.1967, "step": 37276 }, { "epoch": 0.04, "learning_rate": 4.982964634230529e-05, "loss": 1.1589, "step": 37277 }, { "epoch": 0.04, "learning_rate": 4.982963718813995e-05, "loss": 0.9855, "step": 37278 }, { "epoch": 0.04, "learning_rate": 4.98296280337295e-05, "loss": 2.5648, "step": 37279 }, { "epoch": 0.04, "learning_rate": 4.9829618879073944e-05, "loss": 1.0919, "step": 37280 }, { "epoch": 0.04, "learning_rate": 4.982960972417328e-05, "loss": 0.8846, "step": 37281 }, { "epoch": 0.04, "learning_rate": 4.98296005690275e-05, "loss": 0.909, "step": 37282 }, { "epoch": 0.04, "learning_rate": 4.9829591413636624e-05, "loss": 1.0312, "step": 37283 }, { "epoch": 0.04, "learning_rate": 4.982958225800064e-05, "loss": 0.9721, "step": 37284 }, { "epoch": 0.04, "learning_rate": 4.982957310211954e-05, "loss": 0.9649, "step": 37285 }, { "epoch": 0.04, "learning_rate": 4.982956394599334e-05, "loss": 0.8168, "step": 37286 }, { "epoch": 0.04, "learning_rate": 4.9829554789622025e-05, "loss": 0.907, "step": 37287 }, { "epoch": 0.04, "learning_rate": 4.982954563300561e-05, "loss": 0.9205, "step": 37288 }, { "epoch": 0.04, "learning_rate": 4.982953647614409e-05, "loss": 1.0218, "step": 37289 }, { "epoch": 0.04, "learning_rate": 4.9829527319037455e-05, "loss": 1.1242, "step": 37290 }, { "epoch": 0.04, "learning_rate": 4.982951816168572e-05, "loss": 1.0537, "step": 37291 }, { "epoch": 0.04, "learning_rate": 4.982950900408888e-05, "loss": 0.7965, "step": 37292 }, { "epoch": 0.04, "learning_rate": 4.982949984624693e-05, "loss": 1.0696, "step": 37293 }, { "epoch": 0.04, "learning_rate": 4.982949068815988e-05, "loss": 1.0365, "step": 37294 }, { "epoch": 0.04, "learning_rate": 4.982948152982771e-05, "loss": 0.9073, "step": 37295 }, { "epoch": 0.04, "learning_rate": 4.982947237125044e-05, "loss": 0.9954, "step": 37296 }, { "epoch": 0.04, "learning_rate": 4.982946321242807e-05, "loss": 0.8947, "step": 37297 }, { "epoch": 0.04, "learning_rate": 4.9829454053360584e-05, "loss": 1.082, "step": 37298 }, { "epoch": 0.04, "learning_rate": 4.9829444894048e-05, "loss": 1.3861, "step": 37299 }, { "epoch": 0.04, "learning_rate": 4.982943573449031e-05, "loss": 0.7791, "step": 37300 }, { "epoch": 0.04, "learning_rate": 4.982942657468751e-05, "loss": 0.616, "step": 37301 }, { "epoch": 0.04, "learning_rate": 4.98294174146396e-05, "loss": 0.5533, "step": 37302 }, { "epoch": 0.04, "learning_rate": 4.98294082543466e-05, "loss": 0.673, "step": 37303 }, { "epoch": 0.04, "learning_rate": 4.9829399093808484e-05, "loss": 0.6749, "step": 37304 }, { "epoch": 0.04, "learning_rate": 4.982938993302526e-05, "loss": 0.8762, "step": 37305 }, { "epoch": 0.04, "learning_rate": 4.9829380771996934e-05, "loss": 0.8843, "step": 37306 }, { "epoch": 0.04, "learning_rate": 4.982937161072351e-05, "loss": 0.8328, "step": 37307 }, { "epoch": 0.04, "learning_rate": 4.9829362449204974e-05, "loss": 0.8002, "step": 37308 }, { "epoch": 0.04, "learning_rate": 4.9829353287441334e-05, "loss": 1.073, "step": 37309 }, { "epoch": 0.04, "learning_rate": 4.9829344125432584e-05, "loss": 1.0972, "step": 37310 }, { "epoch": 0.04, "learning_rate": 4.9829334963178736e-05, "loss": 1.1166, "step": 37311 }, { "epoch": 0.04, "learning_rate": 4.982932580067978e-05, "loss": 1.0947, "step": 37312 }, { "epoch": 0.04, "learning_rate": 4.982931663793572e-05, "loss": 1.0949, "step": 37313 }, { "epoch": 0.04, "learning_rate": 4.982930747494656e-05, "loss": 0.9524, "step": 37314 }, { "epoch": 0.04, "learning_rate": 4.9829298311712294e-05, "loss": 1.0455, "step": 37315 }, { "epoch": 0.04, "learning_rate": 4.982928914823292e-05, "loss": 0.2604, "step": 37316 }, { "epoch": 0.04, "learning_rate": 4.982927998450844e-05, "loss": 1.0347, "step": 37317 }, { "epoch": 0.04, "learning_rate": 4.982927082053886e-05, "loss": 0.9017, "step": 37318 }, { "epoch": 0.04, "learning_rate": 4.982926165632417e-05, "loss": 0.5978, "step": 37319 }, { "epoch": 0.04, "learning_rate": 4.9829252491864385e-05, "loss": 0.5941, "step": 37320 }, { "epoch": 0.04, "learning_rate": 4.9829243327159494e-05, "loss": 0.782, "step": 37321 }, { "epoch": 0.04, "learning_rate": 4.982923416220949e-05, "loss": 0.9892, "step": 37322 }, { "epoch": 0.04, "learning_rate": 4.982922499701439e-05, "loss": 1.1659, "step": 37323 }, { "epoch": 0.04, "learning_rate": 4.982921583157419e-05, "loss": 1.0593, "step": 37324 }, { "epoch": 0.04, "learning_rate": 4.982920666588888e-05, "loss": 0.8747, "step": 37325 }, { "epoch": 0.04, "learning_rate": 4.982919749995847e-05, "loss": 0.6159, "step": 37326 }, { "epoch": 0.04, "learning_rate": 4.982918833378295e-05, "loss": 1.0445, "step": 37327 }, { "epoch": 0.04, "learning_rate": 4.9829179167362336e-05, "loss": 0.8682, "step": 37328 }, { "epoch": 0.04, "learning_rate": 4.982917000069661e-05, "loss": 0.8788, "step": 37329 }, { "epoch": 0.04, "learning_rate": 4.982916083378579e-05, "loss": 0.9568, "step": 37330 }, { "epoch": 0.04, "learning_rate": 4.982915166662986e-05, "loss": 0.7741, "step": 37331 }, { "epoch": 0.04, "learning_rate": 4.9829142499228834e-05, "loss": 1.2552, "step": 37332 }, { "epoch": 0.04, "learning_rate": 4.98291333315827e-05, "loss": 1.221, "step": 37333 }, { "epoch": 0.04, "learning_rate": 4.982912416369146e-05, "loss": 1.1357, "step": 37334 }, { "epoch": 0.04, "learning_rate": 4.982911499555512e-05, "loss": 0.8158, "step": 37335 }, { "epoch": 0.04, "learning_rate": 4.9829105827173685e-05, "loss": 0.6852, "step": 37336 }, { "epoch": 0.04, "learning_rate": 4.982909665854713e-05, "loss": 0.9538, "step": 37337 }, { "epoch": 0.04, "learning_rate": 4.982908748967549e-05, "loss": 1.119, "step": 37338 }, { "epoch": 0.04, "learning_rate": 4.9829078320558745e-05, "loss": 1.0769, "step": 37339 }, { "epoch": 0.04, "learning_rate": 4.9829069151196894e-05, "loss": 1.1174, "step": 37340 }, { "epoch": 0.04, "learning_rate": 4.982905998158994e-05, "loss": 0.8614, "step": 37341 }, { "epoch": 0.04, "learning_rate": 4.982905081173789e-05, "loss": 0.8377, "step": 37342 }, { "epoch": 0.04, "learning_rate": 4.982904164164073e-05, "loss": 0.9851, "step": 37343 }, { "epoch": 0.04, "learning_rate": 4.9829032471298464e-05, "loss": 1.5006, "step": 37344 }, { "epoch": 0.04, "learning_rate": 4.9829023300711106e-05, "loss": 1.2153, "step": 37345 }, { "epoch": 0.04, "learning_rate": 4.982901412987865e-05, "loss": 1.0373, "step": 37346 }, { "epoch": 0.04, "learning_rate": 4.9829004958801084e-05, "loss": 1.1634, "step": 37347 }, { "epoch": 0.04, "learning_rate": 4.982899578747842e-05, "loss": 0.9354, "step": 37348 }, { "epoch": 0.04, "learning_rate": 4.982898661591066e-05, "loss": 0.9677, "step": 37349 }, { "epoch": 0.04, "learning_rate": 4.9828977444097786e-05, "loss": 1.0071, "step": 37350 }, { "epoch": 0.04, "learning_rate": 4.9828968272039816e-05, "loss": 1.0833, "step": 37351 }, { "epoch": 0.04, "learning_rate": 4.982895909973674e-05, "loss": 0.8937, "step": 37352 }, { "epoch": 0.04, "learning_rate": 4.982894992718857e-05, "loss": 0.417, "step": 37353 }, { "epoch": 0.04, "learning_rate": 4.98289407543953e-05, "loss": 0.9609, "step": 37354 }, { "epoch": 0.04, "learning_rate": 4.982893158135693e-05, "loss": 0.9911, "step": 37355 }, { "epoch": 0.04, "learning_rate": 4.982892240807345e-05, "loss": 1.1674, "step": 37356 }, { "epoch": 0.04, "learning_rate": 4.982891323454488e-05, "loss": 0.7699, "step": 37357 }, { "epoch": 0.04, "learning_rate": 4.982890406077121e-05, "loss": 0.7248, "step": 37358 }, { "epoch": 0.04, "learning_rate": 4.982889488675243e-05, "loss": 1.0522, "step": 37359 }, { "epoch": 0.04, "learning_rate": 4.982888571248856e-05, "loss": 1.0011, "step": 37360 }, { "epoch": 0.04, "learning_rate": 4.982887653797958e-05, "loss": 1.1132, "step": 37361 }, { "epoch": 0.04, "learning_rate": 4.9828867363225505e-05, "loss": 1.1456, "step": 37362 }, { "epoch": 0.04, "learning_rate": 4.9828858188226325e-05, "loss": 1.199, "step": 37363 }, { "epoch": 0.04, "learning_rate": 4.982884901298205e-05, "loss": 1.0583, "step": 37364 }, { "epoch": 0.04, "learning_rate": 4.982883983749267e-05, "loss": 0.8987, "step": 37365 }, { "epoch": 0.04, "learning_rate": 4.9828830661758193e-05, "loss": 1.3758, "step": 37366 }, { "epoch": 0.04, "learning_rate": 4.9828821485778624e-05, "loss": 1.0459, "step": 37367 }, { "epoch": 0.04, "learning_rate": 4.982881230955395e-05, "loss": 1.116, "step": 37368 }, { "epoch": 0.04, "learning_rate": 4.982880313308417e-05, "loss": 1.1657, "step": 37369 }, { "epoch": 0.04, "learning_rate": 4.98287939563693e-05, "loss": 0.8983, "step": 37370 }, { "epoch": 0.04, "learning_rate": 4.982878477940932e-05, "loss": 0.6178, "step": 37371 }, { "epoch": 0.04, "learning_rate": 4.982877560220425e-05, "loss": 1.0327, "step": 37372 }, { "epoch": 0.04, "learning_rate": 4.9828766424754084e-05, "loss": 1.1051, "step": 37373 }, { "epoch": 0.04, "learning_rate": 4.982875724705881e-05, "loss": 0.9129, "step": 37374 }, { "epoch": 0.04, "learning_rate": 4.9828748069118435e-05, "loss": 0.9454, "step": 37375 }, { "epoch": 0.04, "learning_rate": 4.982873889093297e-05, "loss": 1.2836, "step": 37376 }, { "epoch": 0.04, "learning_rate": 4.9828729712502396e-05, "loss": 1.2955, "step": 37377 }, { "epoch": 0.04, "learning_rate": 4.9828720533826734e-05, "loss": 1.0645, "step": 37378 }, { "epoch": 0.04, "learning_rate": 4.982871135490597e-05, "loss": 1.1577, "step": 37379 }, { "epoch": 0.04, "learning_rate": 4.9828702175740104e-05, "loss": 0.9552, "step": 37380 }, { "epoch": 0.04, "learning_rate": 4.982869299632914e-05, "loss": 1.3038, "step": 37381 }, { "epoch": 0.04, "learning_rate": 4.982868381667308e-05, "loss": 1.3183, "step": 37382 }, { "epoch": 0.04, "learning_rate": 4.982867463677192e-05, "loss": 1.1463, "step": 37383 }, { "epoch": 0.04, "learning_rate": 4.982866545662566e-05, "loss": 1.2771, "step": 37384 }, { "epoch": 0.04, "learning_rate": 4.982865627623431e-05, "loss": 0.603, "step": 37385 }, { "epoch": 0.04, "learning_rate": 4.9828647095597854e-05, "loss": 0.409, "step": 37386 }, { "epoch": 0.04, "learning_rate": 4.98286379147163e-05, "loss": 0.9849, "step": 37387 }, { "epoch": 0.04, "learning_rate": 4.982862873358965e-05, "loss": 1.1076, "step": 37388 }, { "epoch": 0.04, "learning_rate": 4.98286195522179e-05, "loss": 1.2038, "step": 37389 }, { "epoch": 0.04, "learning_rate": 4.982861037060106e-05, "loss": 0.7608, "step": 37390 }, { "epoch": 0.04, "learning_rate": 4.982860118873911e-05, "loss": 1.1362, "step": 37391 }, { "epoch": 0.04, "learning_rate": 4.9828592006632076e-05, "loss": 1.3803, "step": 37392 }, { "epoch": 0.04, "learning_rate": 4.982858282427993e-05, "loss": 0.9596, "step": 37393 }, { "epoch": 0.04, "learning_rate": 4.98285736416827e-05, "loss": 0.7998, "step": 37394 }, { "epoch": 0.04, "learning_rate": 4.9828564458840364e-05, "loss": 1.0349, "step": 37395 }, { "epoch": 0.04, "learning_rate": 4.982855527575294e-05, "loss": 1.0931, "step": 37396 }, { "epoch": 0.04, "learning_rate": 4.9828546092420414e-05, "loss": 1.2423, "step": 37397 }, { "epoch": 0.04, "learning_rate": 4.9828536908842796e-05, "loss": 1.1727, "step": 37398 }, { "epoch": 0.04, "learning_rate": 4.982852772502007e-05, "loss": 0.6069, "step": 37399 }, { "epoch": 0.04, "learning_rate": 4.9828518540952254e-05, "loss": 0.6873, "step": 37400 }, { "epoch": 0.04, "learning_rate": 4.982850935663934e-05, "loss": 1.0047, "step": 37401 }, { "epoch": 0.04, "learning_rate": 4.982850017208133e-05, "loss": 1.2878, "step": 37402 }, { "epoch": 0.04, "learning_rate": 4.9828490987278224e-05, "loss": 1.0193, "step": 37403 }, { "epoch": 0.04, "learning_rate": 4.982848180223002e-05, "loss": 0.9599, "step": 37404 }, { "epoch": 0.04, "learning_rate": 4.982847261693672e-05, "loss": 1.0089, "step": 37405 }, { "epoch": 0.04, "learning_rate": 4.982846343139832e-05, "loss": 1.2323, "step": 37406 }, { "epoch": 0.04, "learning_rate": 4.982845424561483e-05, "loss": 0.9148, "step": 37407 }, { "epoch": 0.04, "learning_rate": 4.982844505958625e-05, "loss": 0.379, "step": 37408 }, { "epoch": 0.04, "learning_rate": 4.982843587331256e-05, "loss": 0.8886, "step": 37409 }, { "epoch": 0.04, "learning_rate": 4.9828426686793785e-05, "loss": 1.2476, "step": 37410 }, { "epoch": 0.04, "learning_rate": 4.9828417500029914e-05, "loss": 0.9793, "step": 37411 }, { "epoch": 0.04, "learning_rate": 4.982840831302094e-05, "loss": 0.7201, "step": 37412 }, { "epoch": 0.04, "learning_rate": 4.9828399125766865e-05, "loss": 1.0182, "step": 37413 }, { "epoch": 0.04, "learning_rate": 4.982838993826771e-05, "loss": 1.0478, "step": 37414 }, { "epoch": 0.04, "learning_rate": 4.9828380750523454e-05, "loss": 0.8948, "step": 37415 }, { "epoch": 0.04, "learning_rate": 4.9828371562534095e-05, "loss": 0.8877, "step": 37416 }, { "epoch": 0.04, "learning_rate": 4.982836237429965e-05, "loss": 0.9808, "step": 37417 }, { "epoch": 0.04, "learning_rate": 4.9828353185820106e-05, "loss": 0.8657, "step": 37418 }, { "epoch": 0.04, "learning_rate": 4.9828343997095476e-05, "loss": 0.9924, "step": 37419 }, { "epoch": 0.04, "learning_rate": 4.9828334808125734e-05, "loss": 1.0091, "step": 37420 }, { "epoch": 0.04, "learning_rate": 4.982832561891091e-05, "loss": 0.7487, "step": 37421 }, { "epoch": 0.04, "learning_rate": 4.9828316429450985e-05, "loss": 1.2267, "step": 37422 }, { "epoch": 0.04, "learning_rate": 4.982830723974597e-05, "loss": 1.2121, "step": 37423 }, { "epoch": 0.04, "learning_rate": 4.982829804979586e-05, "loss": 1.1644, "step": 37424 }, { "epoch": 0.04, "learning_rate": 4.9828288859600645e-05, "loss": 0.877, "step": 37425 }, { "epoch": 0.04, "learning_rate": 4.9828279669160346e-05, "loss": 0.9697, "step": 37426 }, { "epoch": 0.04, "learning_rate": 4.9828270478474956e-05, "loss": 1.0626, "step": 37427 }, { "epoch": 0.04, "learning_rate": 4.982826128754446e-05, "loss": 0.9484, "step": 37428 }, { "epoch": 0.04, "learning_rate": 4.9828252096368884e-05, "loss": 0.7933, "step": 37429 }, { "epoch": 0.04, "learning_rate": 4.98282429049482e-05, "loss": 0.8295, "step": 37430 }, { "epoch": 0.04, "learning_rate": 4.9828233713282435e-05, "loss": 0.9083, "step": 37431 }, { "epoch": 0.04, "learning_rate": 4.982822452137157e-05, "loss": 1.0233, "step": 37432 }, { "epoch": 0.04, "learning_rate": 4.982821532921561e-05, "loss": 0.9121, "step": 37433 }, { "epoch": 0.04, "learning_rate": 4.982820613681455e-05, "loss": 1.0966, "step": 37434 }, { "epoch": 0.04, "learning_rate": 4.982819694416841e-05, "loss": 1.1314, "step": 37435 }, { "epoch": 0.04, "learning_rate": 4.982818775127717e-05, "loss": 0.9794, "step": 37436 }, { "epoch": 0.04, "learning_rate": 4.982817855814084e-05, "loss": 1.3281, "step": 37437 }, { "epoch": 0.04, "learning_rate": 4.982816936475941e-05, "loss": 2.2194, "step": 37438 }, { "epoch": 0.04, "learning_rate": 4.9828160171132886e-05, "loss": 4.7924, "step": 37439 }, { "epoch": 0.04, "learning_rate": 4.9828150977261276e-05, "loss": 4.7663, "step": 37440 }, { "epoch": 0.04, "learning_rate": 4.982814178314457e-05, "loss": 2.8603, "step": 37441 }, { "epoch": 0.04, "learning_rate": 4.982813258878277e-05, "loss": 1.088, "step": 37442 }, { "epoch": 0.04, "learning_rate": 4.982812339417588e-05, "loss": 0.9302, "step": 37443 }, { "epoch": 0.04, "learning_rate": 4.982811419932389e-05, "loss": 0.9564, "step": 37444 }, { "epoch": 0.04, "learning_rate": 4.9828105004226815e-05, "loss": 0.8171, "step": 37445 }, { "epoch": 0.04, "learning_rate": 4.982809580888465e-05, "loss": 0.6664, "step": 37446 }, { "epoch": 0.04, "learning_rate": 4.9828086613297383e-05, "loss": 1.2954, "step": 37447 }, { "epoch": 0.04, "learning_rate": 4.982807741746503e-05, "loss": 0.9012, "step": 37448 }, { "epoch": 0.04, "learning_rate": 4.982806822138758e-05, "loss": 1.0894, "step": 37449 }, { "epoch": 0.04, "learning_rate": 4.9828059025065046e-05, "loss": 1.0658, "step": 37450 }, { "epoch": 0.04, "learning_rate": 4.982804982849741e-05, "loss": 0.9781, "step": 37451 }, { "epoch": 0.04, "learning_rate": 4.9828040631684694e-05, "loss": 1.0285, "step": 37452 }, { "epoch": 0.04, "learning_rate": 4.982803143462688e-05, "loss": 1.1707, "step": 37453 }, { "epoch": 0.04, "learning_rate": 4.9828022237323966e-05, "loss": 1.3529, "step": 37454 }, { "epoch": 0.04, "learning_rate": 4.982801303977597e-05, "loss": 1.1045, "step": 37455 }, { "epoch": 0.04, "learning_rate": 4.9828003841982875e-05, "loss": 0.9579, "step": 37456 }, { "epoch": 0.04, "learning_rate": 4.982799464394469e-05, "loss": 1.0527, "step": 37457 }, { "epoch": 0.04, "learning_rate": 4.9827985445661415e-05, "loss": 1.0128, "step": 37458 }, { "epoch": 0.04, "learning_rate": 4.9827976247133056e-05, "loss": 1.0753, "step": 37459 }, { "epoch": 0.04, "learning_rate": 4.98279670483596e-05, "loss": 1.1842, "step": 37460 }, { "epoch": 0.04, "learning_rate": 4.982795784934105e-05, "loss": 0.5967, "step": 37461 }, { "epoch": 0.04, "learning_rate": 4.982794865007741e-05, "loss": 0.4156, "step": 37462 }, { "epoch": 0.04, "learning_rate": 4.9827939450568684e-05, "loss": 0.2667, "step": 37463 }, { "epoch": 0.04, "learning_rate": 4.982793025081486e-05, "loss": 0.7829, "step": 37464 }, { "epoch": 0.04, "learning_rate": 4.982792105081595e-05, "loss": 0.9547, "step": 37465 }, { "epoch": 0.04, "learning_rate": 4.982791185057195e-05, "loss": 1.0089, "step": 37466 }, { "epoch": 0.04, "learning_rate": 4.982790265008285e-05, "loss": 1.4048, "step": 37467 }, { "epoch": 0.04, "learning_rate": 4.9827893449348666e-05, "loss": 1.1296, "step": 37468 }, { "epoch": 0.04, "learning_rate": 4.9827884248369393e-05, "loss": 0.746, "step": 37469 }, { "epoch": 0.04, "learning_rate": 4.982787504714503e-05, "loss": 1.0201, "step": 37470 }, { "epoch": 0.04, "learning_rate": 4.982786584567557e-05, "loss": 1.1548, "step": 37471 }, { "epoch": 0.04, "learning_rate": 4.9827856643961024e-05, "loss": 1.0498, "step": 37472 }, { "epoch": 0.04, "learning_rate": 4.982784744200139e-05, "loss": 1.0413, "step": 37473 }, { "epoch": 0.04, "learning_rate": 4.982783823979666e-05, "loss": 0.9483, "step": 37474 }, { "epoch": 0.04, "learning_rate": 4.982782903734685e-05, "loss": 1.2854, "step": 37475 }, { "epoch": 0.04, "learning_rate": 4.9827819834651944e-05, "loss": 1.0285, "step": 37476 }, { "epoch": 0.04, "learning_rate": 4.9827810631711946e-05, "loss": 0.9625, "step": 37477 }, { "epoch": 0.04, "learning_rate": 4.9827801428526864e-05, "loss": 1.1914, "step": 37478 }, { "epoch": 0.04, "learning_rate": 4.982779222509669e-05, "loss": 1.1203, "step": 37479 }, { "epoch": 0.04, "learning_rate": 4.982778302142143e-05, "loss": 1.2318, "step": 37480 }, { "epoch": 0.04, "learning_rate": 4.982777381750107e-05, "loss": 0.7368, "step": 37481 }, { "epoch": 0.04, "learning_rate": 4.9827764613335635e-05, "loss": 1.1756, "step": 37482 }, { "epoch": 0.04, "learning_rate": 4.9827755408925106e-05, "loss": 1.015, "step": 37483 }, { "epoch": 0.04, "learning_rate": 4.982774620426948e-05, "loss": 0.9292, "step": 37484 }, { "epoch": 0.04, "learning_rate": 4.982773699936877e-05, "loss": 1.0634, "step": 37485 }, { "epoch": 0.04, "learning_rate": 4.982772779422297e-05, "loss": 0.7604, "step": 37486 }, { "epoch": 0.04, "learning_rate": 4.9827718588832084e-05, "loss": 0.9644, "step": 37487 }, { "epoch": 0.04, "learning_rate": 4.98277093831961e-05, "loss": 0.7888, "step": 37488 }, { "epoch": 0.04, "learning_rate": 4.982770017731504e-05, "loss": 1.0689, "step": 37489 }, { "epoch": 0.04, "learning_rate": 4.9827690971188885e-05, "loss": 0.9213, "step": 37490 }, { "epoch": 0.04, "learning_rate": 4.9827681764817644e-05, "loss": 1.1504, "step": 37491 }, { "epoch": 0.04, "learning_rate": 4.982767255820131e-05, "loss": 1.0361, "step": 37492 }, { "epoch": 0.04, "learning_rate": 4.982766335133989e-05, "loss": 1.0419, "step": 37493 }, { "epoch": 0.04, "learning_rate": 4.9827654144233385e-05, "loss": 0.8787, "step": 37494 }, { "epoch": 0.04, "learning_rate": 4.982764493688179e-05, "loss": 0.8551, "step": 37495 }, { "epoch": 0.04, "learning_rate": 4.982763572928511e-05, "loss": 1.1008, "step": 37496 }, { "epoch": 0.04, "learning_rate": 4.982762652144333e-05, "loss": 0.6015, "step": 37497 }, { "epoch": 0.04, "learning_rate": 4.982761731335648e-05, "loss": 0.7811, "step": 37498 }, { "epoch": 0.04, "learning_rate": 4.982760810502453e-05, "loss": 0.8312, "step": 37499 }, { "epoch": 0.04, "learning_rate": 4.982759889644749e-05, "loss": 0.8835, "step": 37500 }, { "epoch": 0.04, "eval_loss": 1.0715769529342651, "eval_runtime": 128.7146, "eval_samples_per_second": 10.76, "eval_steps_per_second": 5.384, "step": 37500 }, { "epoch": 0.04, "learning_rate": 4.982758968762537e-05, "loss": 0.9383, "step": 37501 }, { "epoch": 0.04, "learning_rate": 4.9827580478558154e-05, "loss": 0.9524, "step": 37502 }, { "epoch": 0.04, "learning_rate": 4.9827571269245866e-05, "loss": 0.899, "step": 37503 }, { "epoch": 0.04, "learning_rate": 4.9827562059688474e-05, "loss": 1.2275, "step": 37504 }, { "epoch": 0.04, "learning_rate": 4.9827552849886004e-05, "loss": 1.1184, "step": 37505 }, { "epoch": 0.04, "learning_rate": 4.982754363983845e-05, "loss": 0.8637, "step": 37506 }, { "epoch": 0.04, "learning_rate": 4.98275344295458e-05, "loss": 0.9897, "step": 37507 }, { "epoch": 0.04, "learning_rate": 4.9827525219008065e-05, "loss": 0.9372, "step": 37508 }, { "epoch": 0.04, "learning_rate": 4.982751600822525e-05, "loss": 1.4189, "step": 37509 }, { "epoch": 0.04, "learning_rate": 4.9827506797197344e-05, "loss": 1.1145, "step": 37510 }, { "epoch": 0.04, "learning_rate": 4.9827497585924344e-05, "loss": 0.9853, "step": 37511 }, { "epoch": 0.04, "learning_rate": 4.9827488374406266e-05, "loss": 1.1735, "step": 37512 }, { "epoch": 0.04, "learning_rate": 4.98274791626431e-05, "loss": 1.1175, "step": 37513 }, { "epoch": 0.04, "learning_rate": 4.982746995063485e-05, "loss": 1.1464, "step": 37514 }, { "epoch": 0.04, "learning_rate": 4.9827460738381504e-05, "loss": 1.2004, "step": 37515 }, { "epoch": 0.04, "learning_rate": 4.9827451525883085e-05, "loss": 1.0329, "step": 37516 }, { "epoch": 0.04, "learning_rate": 4.9827442313139575e-05, "loss": 1.1149, "step": 37517 }, { "epoch": 0.04, "learning_rate": 4.9827433100150974e-05, "loss": 0.8091, "step": 37518 }, { "epoch": 0.04, "learning_rate": 4.982742388691729e-05, "loss": 0.6117, "step": 37519 }, { "epoch": 0.04, "learning_rate": 4.982741467343852e-05, "loss": 0.8948, "step": 37520 }, { "epoch": 0.04, "learning_rate": 4.982740545971467e-05, "loss": 1.0885, "step": 37521 }, { "epoch": 0.04, "learning_rate": 4.9827396245745724e-05, "loss": 0.9918, "step": 37522 }, { "epoch": 0.04, "learning_rate": 4.98273870315317e-05, "loss": 1.0615, "step": 37523 }, { "epoch": 0.04, "learning_rate": 4.9827377817072586e-05, "loss": 0.8391, "step": 37524 }, { "epoch": 0.04, "learning_rate": 4.9827368602368384e-05, "loss": 0.947, "step": 37525 }, { "epoch": 0.04, "learning_rate": 4.9827359387419105e-05, "loss": 1.3187, "step": 37526 }, { "epoch": 0.04, "learning_rate": 4.9827350172224736e-05, "loss": 1.3164, "step": 37527 }, { "epoch": 0.04, "learning_rate": 4.982734095678528e-05, "loss": 0.8011, "step": 37528 }, { "epoch": 0.04, "learning_rate": 4.9827331741100745e-05, "loss": 0.922, "step": 37529 }, { "epoch": 0.04, "learning_rate": 4.9827322525171124e-05, "loss": 1.0546, "step": 37530 }, { "epoch": 0.04, "learning_rate": 4.982731330899641e-05, "loss": 0.7481, "step": 37531 }, { "epoch": 0.04, "learning_rate": 4.982730409257662e-05, "loss": 0.9419, "step": 37532 }, { "epoch": 0.04, "learning_rate": 4.982729487591174e-05, "loss": 0.9485, "step": 37533 }, { "epoch": 0.04, "learning_rate": 4.982728565900178e-05, "loss": 0.8413, "step": 37534 }, { "epoch": 0.04, "learning_rate": 4.982727644184674e-05, "loss": 0.8795, "step": 37535 }, { "epoch": 0.04, "learning_rate": 4.982726722444661e-05, "loss": 1.2722, "step": 37536 }, { "epoch": 0.04, "learning_rate": 4.982725800680139e-05, "loss": 1.1379, "step": 37537 }, { "epoch": 0.04, "learning_rate": 4.9827248788911086e-05, "loss": 1.1052, "step": 37538 }, { "epoch": 0.04, "learning_rate": 4.982723957077571e-05, "loss": 1.2486, "step": 37539 }, { "epoch": 0.04, "learning_rate": 4.9827230352395235e-05, "loss": 1.8608, "step": 37540 }, { "epoch": 0.04, "learning_rate": 4.9827221133769684e-05, "loss": 1.0146, "step": 37541 }, { "epoch": 0.04, "learning_rate": 4.982721191489905e-05, "loss": 0.8628, "step": 37542 }, { "epoch": 0.04, "learning_rate": 4.9827202695783324e-05, "loss": 1.0994, "step": 37543 }, { "epoch": 0.04, "learning_rate": 4.982719347642253e-05, "loss": 1.05, "step": 37544 }, { "epoch": 0.04, "learning_rate": 4.982718425681664e-05, "loss": 0.8406, "step": 37545 }, { "epoch": 0.04, "learning_rate": 4.9827175036965664e-05, "loss": 0.7722, "step": 37546 }, { "epoch": 0.04, "learning_rate": 4.982716581686962e-05, "loss": 0.7984, "step": 37547 }, { "epoch": 0.04, "learning_rate": 4.982715659652848e-05, "loss": 0.9945, "step": 37548 }, { "epoch": 0.04, "learning_rate": 4.982714737594226e-05, "loss": 0.9952, "step": 37549 }, { "epoch": 0.04, "learning_rate": 4.982713815511095e-05, "loss": 1.0038, "step": 37550 }, { "epoch": 0.04, "learning_rate": 4.9827128934034574e-05, "loss": 0.9206, "step": 37551 }, { "epoch": 0.04, "learning_rate": 4.98271197127131e-05, "loss": 0.7225, "step": 37552 }, { "epoch": 0.04, "learning_rate": 4.982711049114654e-05, "loss": 1.2702, "step": 37553 }, { "epoch": 0.04, "learning_rate": 4.9827101269334914e-05, "loss": 1.0824, "step": 37554 }, { "epoch": 0.04, "learning_rate": 4.9827092047278195e-05, "loss": 0.6195, "step": 37555 }, { "epoch": 0.04, "learning_rate": 4.982708282497639e-05, "loss": 0.8551, "step": 37556 }, { "epoch": 0.04, "learning_rate": 4.982707360242951e-05, "loss": 0.512, "step": 37557 }, { "epoch": 0.04, "learning_rate": 4.982706437963754e-05, "loss": 0.7669, "step": 37558 }, { "epoch": 0.04, "learning_rate": 4.98270551566005e-05, "loss": 1.1311, "step": 37559 }, { "epoch": 0.04, "learning_rate": 4.982704593331837e-05, "loss": 0.951, "step": 37560 }, { "epoch": 0.04, "learning_rate": 4.982703670979116e-05, "loss": 0.8504, "step": 37561 }, { "epoch": 0.04, "learning_rate": 4.982702748601886e-05, "loss": 1.3428, "step": 37562 }, { "epoch": 0.04, "learning_rate": 4.982701826200149e-05, "loss": 1.0657, "step": 37563 }, { "epoch": 0.04, "learning_rate": 4.982700903773904e-05, "loss": 0.83, "step": 37564 }, { "epoch": 0.04, "learning_rate": 4.982699981323149e-05, "loss": 0.9078, "step": 37565 }, { "epoch": 0.04, "learning_rate": 4.982699058847887e-05, "loss": 1.1336, "step": 37566 }, { "epoch": 0.04, "learning_rate": 4.9826981363481174e-05, "loss": 0.5179, "step": 37567 }, { "epoch": 0.04, "learning_rate": 4.982697213823838e-05, "loss": 0.5586, "step": 37568 }, { "epoch": 0.04, "learning_rate": 4.982696291275052e-05, "loss": 1.2557, "step": 37569 }, { "epoch": 0.04, "learning_rate": 4.9826953687017575e-05, "loss": 1.2038, "step": 37570 }, { "epoch": 0.04, "learning_rate": 4.982694446103955e-05, "loss": 1.3674, "step": 37571 }, { "epoch": 0.04, "learning_rate": 4.982693523481644e-05, "loss": 0.8918, "step": 37572 }, { "epoch": 0.04, "learning_rate": 4.982692600834825e-05, "loss": 1.249, "step": 37573 }, { "epoch": 0.04, "learning_rate": 4.9826916781634976e-05, "loss": 1.6282, "step": 37574 }, { "epoch": 0.04, "learning_rate": 4.982690755467662e-05, "loss": 1.0107, "step": 37575 }, { "epoch": 0.04, "learning_rate": 4.9826898327473193e-05, "loss": 1.1451, "step": 37576 }, { "epoch": 0.04, "learning_rate": 4.9826889100024676e-05, "loss": 1.0426, "step": 37577 }, { "epoch": 0.04, "learning_rate": 4.9826879872331075e-05, "loss": 1.0873, "step": 37578 }, { "epoch": 0.04, "learning_rate": 4.982687064439241e-05, "loss": 1.0867, "step": 37579 }, { "epoch": 0.04, "learning_rate": 4.982686141620865e-05, "loss": 0.8743, "step": 37580 }, { "epoch": 0.04, "learning_rate": 4.9826852187779816e-05, "loss": 0.9376, "step": 37581 }, { "epoch": 0.04, "learning_rate": 4.98268429591059e-05, "loss": 0.6439, "step": 37582 }, { "epoch": 0.04, "learning_rate": 4.9826833730186906e-05, "loss": 0.8197, "step": 37583 }, { "epoch": 0.04, "learning_rate": 4.982682450102283e-05, "loss": 1.2205, "step": 37584 }, { "epoch": 0.04, "learning_rate": 4.9826815271613674e-05, "loss": 1.1206, "step": 37585 }, { "epoch": 0.04, "learning_rate": 4.982680604195944e-05, "loss": 1.0552, "step": 37586 }, { "epoch": 0.04, "learning_rate": 4.982679681206012e-05, "loss": 0.848, "step": 37587 }, { "epoch": 0.04, "learning_rate": 4.982678758191573e-05, "loss": 1.7233, "step": 37588 }, { "epoch": 0.04, "learning_rate": 4.982677835152626e-05, "loss": 1.2058, "step": 37589 }, { "epoch": 0.04, "learning_rate": 4.9826769120891704e-05, "loss": 1.2495, "step": 37590 }, { "epoch": 0.04, "learning_rate": 4.9826759890012066e-05, "loss": 1.2395, "step": 37591 }, { "epoch": 0.04, "learning_rate": 4.982675065888736e-05, "loss": 0.8732, "step": 37592 }, { "epoch": 0.04, "learning_rate": 4.9826741427517567e-05, "loss": 0.9617, "step": 37593 }, { "epoch": 0.04, "learning_rate": 4.982673219590269e-05, "loss": 1.2338, "step": 37594 }, { "epoch": 0.04, "learning_rate": 4.982672296404274e-05, "loss": 0.6072, "step": 37595 }, { "epoch": 0.04, "learning_rate": 4.9826713731937715e-05, "loss": 0.9944, "step": 37596 }, { "epoch": 0.04, "learning_rate": 4.982670449958761e-05, "loss": 1.2119, "step": 37597 }, { "epoch": 0.04, "learning_rate": 4.9826695266992417e-05, "loss": 1.0546, "step": 37598 }, { "epoch": 0.04, "learning_rate": 4.982668603415215e-05, "loss": 0.5288, "step": 37599 }, { "epoch": 0.04, "learning_rate": 4.982667680106681e-05, "loss": 0.7271, "step": 37600 }, { "epoch": 0.04, "learning_rate": 4.982666756773639e-05, "loss": 1.155, "step": 37601 }, { "epoch": 0.04, "learning_rate": 4.982665833416088e-05, "loss": 1.2068, "step": 37602 }, { "epoch": 0.04, "learning_rate": 4.9826649100340305e-05, "loss": 1.0057, "step": 37603 }, { "epoch": 0.04, "learning_rate": 4.982663986627465e-05, "loss": 1.1644, "step": 37604 }, { "epoch": 0.04, "learning_rate": 4.982663063196391e-05, "loss": 2.231, "step": 37605 }, { "epoch": 0.04, "learning_rate": 4.98266213974081e-05, "loss": 1.8357, "step": 37606 }, { "epoch": 0.04, "learning_rate": 4.9826612162607206e-05, "loss": 0.6903, "step": 37607 }, { "epoch": 0.04, "learning_rate": 4.982660292756124e-05, "loss": 1.0285, "step": 37608 }, { "epoch": 0.04, "learning_rate": 4.982659369227019e-05, "loss": 1.0583, "step": 37609 }, { "epoch": 0.04, "learning_rate": 4.982658445673407e-05, "loss": 1.2284, "step": 37610 }, { "epoch": 0.04, "learning_rate": 4.982657522095287e-05, "loss": 1.0853, "step": 37611 }, { "epoch": 0.04, "learning_rate": 4.982656598492658e-05, "loss": 0.8913, "step": 37612 }, { "epoch": 0.04, "learning_rate": 4.982655674865523e-05, "loss": 1.2321, "step": 37613 }, { "epoch": 0.04, "learning_rate": 4.982654751213879e-05, "loss": 0.9982, "step": 37614 }, { "epoch": 0.04, "learning_rate": 4.9826538275377274e-05, "loss": 0.8917, "step": 37615 }, { "epoch": 0.04, "learning_rate": 4.982652903837069e-05, "loss": 0.8834, "step": 37616 }, { "epoch": 0.04, "learning_rate": 4.982651980111902e-05, "loss": 1.0217, "step": 37617 }, { "epoch": 0.04, "learning_rate": 4.982651056362228e-05, "loss": 0.8874, "step": 37618 }, { "epoch": 0.04, "learning_rate": 4.982650132588046e-05, "loss": 1.0315, "step": 37619 }, { "epoch": 0.04, "learning_rate": 4.982649208789356e-05, "loss": 1.0172, "step": 37620 }, { "epoch": 0.04, "learning_rate": 4.9826482849661594e-05, "loss": 0.8903, "step": 37621 }, { "epoch": 0.04, "learning_rate": 4.9826473611184545e-05, "loss": 1.1619, "step": 37622 }, { "epoch": 0.04, "learning_rate": 4.982646437246241e-05, "loss": 1.2993, "step": 37623 }, { "epoch": 0.04, "learning_rate": 4.982645513349521e-05, "loss": 1.0269, "step": 37624 }, { "epoch": 0.04, "learning_rate": 4.9826445894282936e-05, "loss": 0.9336, "step": 37625 }, { "epoch": 0.04, "learning_rate": 4.982643665482558e-05, "loss": 0.5989, "step": 37626 }, { "epoch": 0.04, "learning_rate": 4.9826427415123145e-05, "loss": 0.5602, "step": 37627 }, { "epoch": 0.04, "learning_rate": 4.982641817517564e-05, "loss": 0.3217, "step": 37628 }, { "epoch": 0.04, "learning_rate": 4.982640893498306e-05, "loss": 1.0108, "step": 37629 }, { "epoch": 0.04, "learning_rate": 4.98263996945454e-05, "loss": 0.8134, "step": 37630 }, { "epoch": 0.04, "learning_rate": 4.9826390453862656e-05, "loss": 0.4018, "step": 37631 }, { "epoch": 0.04, "learning_rate": 4.982638121293485e-05, "loss": 0.3629, "step": 37632 }, { "epoch": 0.04, "learning_rate": 4.9826371971761966e-05, "loss": 0.3606, "step": 37633 }, { "epoch": 0.04, "learning_rate": 4.9826362730344e-05, "loss": 0.9593, "step": 37634 }, { "epoch": 0.04, "learning_rate": 4.982635348868097e-05, "loss": 0.9633, "step": 37635 }, { "epoch": 0.04, "learning_rate": 4.982634424677285e-05, "loss": 0.9922, "step": 37636 }, { "epoch": 0.04, "learning_rate": 4.982633500461967e-05, "loss": 1.0503, "step": 37637 }, { "epoch": 0.04, "learning_rate": 4.9826325762221404e-05, "loss": 1.1776, "step": 37638 }, { "epoch": 0.04, "learning_rate": 4.9826316519578065e-05, "loss": 1.1534, "step": 37639 }, { "epoch": 0.04, "learning_rate": 4.9826307276689655e-05, "loss": 1.0774, "step": 37640 }, { "epoch": 0.04, "learning_rate": 4.982629803355617e-05, "loss": 1.2555, "step": 37641 }, { "epoch": 0.04, "learning_rate": 4.9826288790177604e-05, "loss": 0.9195, "step": 37642 }, { "epoch": 0.04, "learning_rate": 4.982627954655397e-05, "loss": 1.1163, "step": 37643 }, { "epoch": 0.04, "learning_rate": 4.982627030268526e-05, "loss": 1.1512, "step": 37644 }, { "epoch": 0.04, "learning_rate": 4.982626105857147e-05, "loss": 0.9096, "step": 37645 }, { "epoch": 0.04, "learning_rate": 4.982625181421261e-05, "loss": 1.0354, "step": 37646 }, { "epoch": 0.04, "learning_rate": 4.982624256960868e-05, "loss": 0.7497, "step": 37647 }, { "epoch": 0.04, "learning_rate": 4.982623332475967e-05, "loss": 1.0094, "step": 37648 }, { "epoch": 0.04, "learning_rate": 4.982622407966558e-05, "loss": 0.8887, "step": 37649 }, { "epoch": 0.04, "learning_rate": 4.982621483432643e-05, "loss": 1.1151, "step": 37650 }, { "epoch": 0.04, "learning_rate": 4.98262055887422e-05, "loss": 1.0481, "step": 37651 }, { "epoch": 0.04, "learning_rate": 4.982619634291289e-05, "loss": 1.157, "step": 37652 }, { "epoch": 0.04, "learning_rate": 4.982618709683852e-05, "loss": 1.0531, "step": 37653 }, { "epoch": 0.04, "learning_rate": 4.982617785051906e-05, "loss": 1.109, "step": 37654 }, { "epoch": 0.04, "learning_rate": 4.9826168603954535e-05, "loss": 1.2945, "step": 37655 }, { "epoch": 0.04, "learning_rate": 4.982615935714494e-05, "loss": 0.7544, "step": 37656 }, { "epoch": 0.04, "learning_rate": 4.9826150110090265e-05, "loss": 0.4126, "step": 37657 }, { "epoch": 0.04, "learning_rate": 4.9826140862790515e-05, "loss": 1.1122, "step": 37658 }, { "epoch": 0.04, "learning_rate": 4.98261316152457e-05, "loss": 0.9373, "step": 37659 }, { "epoch": 0.04, "learning_rate": 4.982612236745581e-05, "loss": 0.6127, "step": 37660 }, { "epoch": 0.04, "learning_rate": 4.982611311942084e-05, "loss": 0.8029, "step": 37661 }, { "epoch": 0.04, "learning_rate": 4.982610387114081e-05, "loss": 0.8212, "step": 37662 }, { "epoch": 0.04, "learning_rate": 4.982609462261569e-05, "loss": 0.7931, "step": 37663 }, { "epoch": 0.04, "learning_rate": 4.982608537384551e-05, "loss": 0.8193, "step": 37664 }, { "epoch": 0.04, "learning_rate": 4.9826076124830256e-05, "loss": 1.1159, "step": 37665 }, { "epoch": 0.04, "learning_rate": 4.982606687556993e-05, "loss": 1.0834, "step": 37666 }, { "epoch": 0.04, "learning_rate": 4.9826057626064524e-05, "loss": 1.2108, "step": 37667 }, { "epoch": 0.04, "learning_rate": 4.982604837631405e-05, "loss": 1.1416, "step": 37668 }, { "epoch": 0.04, "learning_rate": 4.9826039126318505e-05, "loss": 0.7328, "step": 37669 }, { "epoch": 0.04, "learning_rate": 4.982602987607788e-05, "loss": 0.3384, "step": 37670 }, { "epoch": 0.04, "learning_rate": 4.98260206255922e-05, "loss": 0.9686, "step": 37671 }, { "epoch": 0.04, "learning_rate": 4.9826011374861435e-05, "loss": 0.9452, "step": 37672 }, { "epoch": 0.04, "learning_rate": 4.9826002123885595e-05, "loss": 0.7548, "step": 37673 }, { "epoch": 0.04, "learning_rate": 4.98259928726647e-05, "loss": 0.6547, "step": 37674 }, { "epoch": 0.04, "learning_rate": 4.982598362119871e-05, "loss": 1.1219, "step": 37675 }, { "epoch": 0.04, "learning_rate": 4.9825974369487667e-05, "loss": 0.9821, "step": 37676 }, { "epoch": 0.04, "learning_rate": 4.9825965117531545e-05, "loss": 0.5288, "step": 37677 }, { "epoch": 0.04, "learning_rate": 4.9825955865330354e-05, "loss": 0.9004, "step": 37678 }, { "epoch": 0.04, "learning_rate": 4.982594661288409e-05, "loss": 0.9357, "step": 37679 }, { "epoch": 0.04, "learning_rate": 4.982593736019275e-05, "loss": 1.1883, "step": 37680 }, { "epoch": 0.04, "learning_rate": 4.982592810725635e-05, "loss": 0.9168, "step": 37681 }, { "epoch": 0.04, "learning_rate": 4.982591885407487e-05, "loss": 1.1395, "step": 37682 }, { "epoch": 0.04, "learning_rate": 4.982590960064833e-05, "loss": 1.4296, "step": 37683 }, { "epoch": 0.04, "learning_rate": 4.9825900346976706e-05, "loss": 1.1866, "step": 37684 }, { "epoch": 0.04, "learning_rate": 4.982589109306001e-05, "loss": 0.9594, "step": 37685 }, { "epoch": 0.04, "learning_rate": 4.982588183889825e-05, "loss": 0.7541, "step": 37686 }, { "epoch": 0.04, "learning_rate": 4.9825872584491416e-05, "loss": 1.1676, "step": 37687 }, { "epoch": 0.04, "learning_rate": 4.9825863329839514e-05, "loss": 0.9706, "step": 37688 }, { "epoch": 0.04, "learning_rate": 4.982585407494255e-05, "loss": 0.6147, "step": 37689 }, { "epoch": 0.04, "learning_rate": 4.9825844819800506e-05, "loss": 1.2675, "step": 37690 }, { "epoch": 0.04, "learning_rate": 4.982583556441339e-05, "loss": 1.0199, "step": 37691 }, { "epoch": 0.04, "learning_rate": 4.982582630878121e-05, "loss": 1.1088, "step": 37692 }, { "epoch": 0.04, "learning_rate": 4.9825817052903956e-05, "loss": 1.0665, "step": 37693 }, { "epoch": 0.04, "learning_rate": 4.982580779678163e-05, "loss": 1.0767, "step": 37694 }, { "epoch": 0.04, "learning_rate": 4.982579854041424e-05, "loss": 1.0137, "step": 37695 }, { "epoch": 0.04, "learning_rate": 4.9825789283801774e-05, "loss": 0.8173, "step": 37696 }, { "epoch": 0.04, "learning_rate": 4.982578002694424e-05, "loss": 0.8918, "step": 37697 }, { "epoch": 0.04, "learning_rate": 4.982577076984164e-05, "loss": 1.3094, "step": 37698 }, { "epoch": 0.04, "learning_rate": 4.9825761512493964e-05, "loss": 0.9356, "step": 37699 }, { "epoch": 0.04, "learning_rate": 4.9825752254901225e-05, "loss": 1.0097, "step": 37700 }, { "epoch": 0.04, "learning_rate": 4.9825742997063415e-05, "loss": 1.3157, "step": 37701 }, { "epoch": 0.04, "learning_rate": 4.982573373898053e-05, "loss": 1.1561, "step": 37702 }, { "epoch": 0.04, "learning_rate": 4.982572448065258e-05, "loss": 0.8695, "step": 37703 }, { "epoch": 0.04, "learning_rate": 4.982571522207956e-05, "loss": 1.0331, "step": 37704 }, { "epoch": 0.04, "learning_rate": 4.982570596326147e-05, "loss": 0.8174, "step": 37705 }, { "epoch": 0.04, "learning_rate": 4.982569670419832e-05, "loss": 1.0476, "step": 37706 }, { "epoch": 0.04, "learning_rate": 4.982568744489009e-05, "loss": 1.0244, "step": 37707 }, { "epoch": 0.04, "learning_rate": 4.98256781853368e-05, "loss": 1.101, "step": 37708 }, { "epoch": 0.04, "learning_rate": 4.982566892553844e-05, "loss": 1.1129, "step": 37709 }, { "epoch": 0.04, "learning_rate": 4.9825659665495004e-05, "loss": 1.0222, "step": 37710 }, { "epoch": 0.04, "learning_rate": 4.9825650405206504e-05, "loss": 1.0425, "step": 37711 }, { "epoch": 0.04, "learning_rate": 4.982564114467294e-05, "loss": 1.0513, "step": 37712 }, { "epoch": 0.04, "learning_rate": 4.98256318838943e-05, "loss": 1.0936, "step": 37713 }, { "epoch": 0.04, "learning_rate": 4.98256226228706e-05, "loss": 1.0382, "step": 37714 }, { "epoch": 0.04, "learning_rate": 4.982561336160182e-05, "loss": 1.0504, "step": 37715 }, { "epoch": 0.04, "learning_rate": 4.9825604100087985e-05, "loss": 0.9562, "step": 37716 }, { "epoch": 0.04, "learning_rate": 4.982559483832908e-05, "loss": 1.0506, "step": 37717 }, { "epoch": 0.04, "learning_rate": 4.98255855763251e-05, "loss": 0.9885, "step": 37718 }, { "epoch": 0.04, "learning_rate": 4.9825576314076057e-05, "loss": 0.8234, "step": 37719 }, { "epoch": 0.04, "learning_rate": 4.9825567051581944e-05, "loss": 0.7615, "step": 37720 }, { "epoch": 0.04, "learning_rate": 4.982555778884277e-05, "loss": 0.8885, "step": 37721 }, { "epoch": 0.04, "learning_rate": 4.982554852585852e-05, "loss": 1.0444, "step": 37722 }, { "epoch": 0.04, "learning_rate": 4.9825539262629205e-05, "loss": 0.9999, "step": 37723 }, { "epoch": 0.04, "learning_rate": 4.9825529999154825e-05, "loss": 0.7871, "step": 37724 }, { "epoch": 0.04, "learning_rate": 4.9825520735435375e-05, "loss": 0.9044, "step": 37725 }, { "epoch": 0.04, "learning_rate": 4.982551147147086e-05, "loss": 0.8813, "step": 37726 }, { "epoch": 0.04, "learning_rate": 4.9825502207261276e-05, "loss": 1.2009, "step": 37727 }, { "epoch": 0.04, "learning_rate": 4.982549294280663e-05, "loss": 1.4119, "step": 37728 }, { "epoch": 0.04, "learning_rate": 4.982548367810691e-05, "loss": 1.0609, "step": 37729 }, { "epoch": 0.04, "learning_rate": 4.9825474413162134e-05, "loss": 1.4825, "step": 37730 }, { "epoch": 0.04, "learning_rate": 4.982546514797228e-05, "loss": 1.1955, "step": 37731 }, { "epoch": 0.04, "learning_rate": 4.9825455882537366e-05, "loss": 1.2237, "step": 37732 }, { "epoch": 0.04, "learning_rate": 4.982544661685739e-05, "loss": 1.0947, "step": 37733 }, { "epoch": 0.04, "learning_rate": 4.982543735093234e-05, "loss": 1.1269, "step": 37734 }, { "epoch": 0.04, "learning_rate": 4.982542808476222e-05, "loss": 0.9634, "step": 37735 }, { "epoch": 0.04, "learning_rate": 4.982541881834703e-05, "loss": 0.9413, "step": 37736 }, { "epoch": 0.04, "learning_rate": 4.982540955168679e-05, "loss": 0.8842, "step": 37737 }, { "epoch": 0.04, "learning_rate": 4.9825400284781475e-05, "loss": 1.0507, "step": 37738 }, { "epoch": 0.04, "learning_rate": 4.9825391017631094e-05, "loss": 0.475, "step": 37739 }, { "epoch": 0.04, "learning_rate": 4.9825381750235656e-05, "loss": 0.2822, "step": 37740 }, { "epoch": 0.04, "learning_rate": 4.982537248259515e-05, "loss": 0.2194, "step": 37741 }, { "epoch": 0.04, "learning_rate": 4.982536321470956e-05, "loss": 0.237, "step": 37742 }, { "epoch": 0.04, "learning_rate": 4.9825353946578926e-05, "loss": 0.1822, "step": 37743 }, { "epoch": 0.04, "learning_rate": 4.982534467820322e-05, "loss": 0.1458, "step": 37744 }, { "epoch": 0.04, "learning_rate": 4.9825335409582444e-05, "loss": 0.916, "step": 37745 }, { "epoch": 0.04, "learning_rate": 4.982532614071661e-05, "loss": 0.8612, "step": 37746 }, { "epoch": 0.04, "learning_rate": 4.98253168716057e-05, "loss": 0.4972, "step": 37747 }, { "epoch": 0.04, "learning_rate": 4.982530760224974e-05, "loss": 0.6401, "step": 37748 }, { "epoch": 0.04, "learning_rate": 4.9825298332648704e-05, "loss": 0.3499, "step": 37749 }, { "epoch": 0.04, "learning_rate": 4.98252890628026e-05, "loss": 0.2411, "step": 37750 }, { "epoch": 0.04, "learning_rate": 4.9825279792711445e-05, "loss": 0.313, "step": 37751 }, { "epoch": 0.04, "learning_rate": 4.982527052237521e-05, "loss": 0.3636, "step": 37752 }, { "epoch": 0.04, "learning_rate": 4.982526125179392e-05, "loss": 0.5383, "step": 37753 }, { "epoch": 0.04, "learning_rate": 4.982525198096757e-05, "loss": 0.5007, "step": 37754 }, { "epoch": 0.04, "learning_rate": 4.9825242709896145e-05, "loss": 0.3702, "step": 37755 }, { "epoch": 0.04, "learning_rate": 4.982523343857966e-05, "loss": 0.2086, "step": 37756 }, { "epoch": 0.04, "learning_rate": 4.982522416701812e-05, "loss": 0.2223, "step": 37757 }, { "epoch": 0.04, "learning_rate": 4.98252148952115e-05, "loss": 0.172, "step": 37758 }, { "epoch": 0.04, "learning_rate": 4.982520562315982e-05, "loss": 0.243, "step": 37759 }, { "epoch": 0.04, "learning_rate": 4.982519635086308e-05, "loss": 0.2865, "step": 37760 }, { "epoch": 0.04, "learning_rate": 4.982518707832127e-05, "loss": 0.1401, "step": 37761 }, { "epoch": 0.04, "learning_rate": 4.982517780553441e-05, "loss": 0.1747, "step": 37762 }, { "epoch": 0.04, "learning_rate": 4.982516853250247e-05, "loss": 0.1912, "step": 37763 }, { "epoch": 0.04, "learning_rate": 4.9825159259225475e-05, "loss": 1.4096, "step": 37764 }, { "epoch": 0.04, "learning_rate": 4.9825149985703414e-05, "loss": 0.992, "step": 37765 }, { "epoch": 0.04, "learning_rate": 4.982514071193629e-05, "loss": 0.9733, "step": 37766 }, { "epoch": 0.04, "learning_rate": 4.98251314379241e-05, "loss": 0.8868, "step": 37767 }, { "epoch": 0.04, "learning_rate": 4.982512216366685e-05, "loss": 1.1001, "step": 37768 }, { "epoch": 0.04, "learning_rate": 4.982511288916454e-05, "loss": 1.2865, "step": 37769 }, { "epoch": 0.04, "learning_rate": 4.982510361441716e-05, "loss": 1.2552, "step": 37770 }, { "epoch": 0.04, "learning_rate": 4.9825094339424725e-05, "loss": 1.2305, "step": 37771 }, { "epoch": 0.04, "learning_rate": 4.982508506418722e-05, "loss": 1.0047, "step": 37772 }, { "epoch": 0.04, "learning_rate": 4.9825075788704656e-05, "loss": 1.1827, "step": 37773 }, { "epoch": 0.04, "learning_rate": 4.982506651297702e-05, "loss": 1.0624, "step": 37774 }, { "epoch": 0.04, "learning_rate": 4.982505723700433e-05, "loss": 1.1467, "step": 37775 }, { "epoch": 0.04, "learning_rate": 4.9825047960786585e-05, "loss": 0.8524, "step": 37776 }, { "epoch": 0.04, "learning_rate": 4.982503868432377e-05, "loss": 1.2205, "step": 37777 }, { "epoch": 0.04, "learning_rate": 4.9825029407615894e-05, "loss": 1.1961, "step": 37778 }, { "epoch": 0.04, "learning_rate": 4.982502013066295e-05, "loss": 1.1125, "step": 37779 }, { "epoch": 0.04, "learning_rate": 4.982501085346495e-05, "loss": 1.2402, "step": 37780 }, { "epoch": 0.04, "learning_rate": 4.982500157602189e-05, "loss": 1.2266, "step": 37781 }, { "epoch": 0.04, "learning_rate": 4.982499229833376e-05, "loss": 1.2703, "step": 37782 }, { "epoch": 0.04, "learning_rate": 4.982498302040057e-05, "loss": 0.8016, "step": 37783 }, { "epoch": 0.04, "learning_rate": 4.982497374222233e-05, "loss": 0.8698, "step": 37784 }, { "epoch": 0.04, "learning_rate": 4.9824964463799016e-05, "loss": 1.0327, "step": 37785 }, { "epoch": 0.04, "learning_rate": 4.982495518513064e-05, "loss": 0.9866, "step": 37786 }, { "epoch": 0.04, "learning_rate": 4.982494590621721e-05, "loss": 1.1241, "step": 37787 }, { "epoch": 0.04, "learning_rate": 4.982493662705871e-05, "loss": 0.985, "step": 37788 }, { "epoch": 0.04, "learning_rate": 4.982492734765515e-05, "loss": 0.8905, "step": 37789 }, { "epoch": 0.04, "learning_rate": 4.9824918068006534e-05, "loss": 1.1972, "step": 37790 }, { "epoch": 0.04, "learning_rate": 4.9824908788112854e-05, "loss": 1.2648, "step": 37791 }, { "epoch": 0.04, "learning_rate": 4.982489950797412e-05, "loss": 1.2568, "step": 37792 }, { "epoch": 0.04, "learning_rate": 4.9824890227590317e-05, "loss": 1.4227, "step": 37793 }, { "epoch": 0.04, "learning_rate": 4.982488094696145e-05, "loss": 1.1169, "step": 37794 }, { "epoch": 0.04, "learning_rate": 4.982487166608753e-05, "loss": 0.9101, "step": 37795 }, { "epoch": 0.04, "learning_rate": 4.982486238496855e-05, "loss": 0.9534, "step": 37796 }, { "epoch": 0.04, "learning_rate": 4.98248531036045e-05, "loss": 0.8965, "step": 37797 }, { "epoch": 0.04, "learning_rate": 4.98248438219954e-05, "loss": 1.0249, "step": 37798 }, { "epoch": 0.04, "learning_rate": 4.982483454014123e-05, "loss": 0.921, "step": 37799 }, { "epoch": 0.04, "learning_rate": 4.982482525804201e-05, "loss": 0.9748, "step": 37800 }, { "epoch": 0.04, "learning_rate": 4.9824815975697725e-05, "loss": 1.0824, "step": 37801 }, { "epoch": 0.04, "learning_rate": 4.982480669310837e-05, "loss": 0.9552, "step": 37802 }, { "epoch": 0.04, "learning_rate": 4.982479741027397e-05, "loss": 1.1658, "step": 37803 }, { "epoch": 0.04, "learning_rate": 4.9824788127194504e-05, "loss": 0.9076, "step": 37804 }, { "epoch": 0.04, "learning_rate": 4.982477884386998e-05, "loss": 0.8755, "step": 37805 }, { "epoch": 0.04, "learning_rate": 4.9824769560300395e-05, "loss": 0.8408, "step": 37806 }, { "epoch": 0.04, "learning_rate": 4.982476027648575e-05, "loss": 0.9859, "step": 37807 }, { "epoch": 0.04, "learning_rate": 4.9824750992426045e-05, "loss": 1.2804, "step": 37808 }, { "epoch": 0.04, "learning_rate": 4.982474170812128e-05, "loss": 1.0153, "step": 37809 }, { "epoch": 0.04, "learning_rate": 4.9824732423571455e-05, "loss": 0.9464, "step": 37810 }, { "epoch": 0.04, "learning_rate": 4.982472313877658e-05, "loss": 1.1302, "step": 37811 }, { "epoch": 0.04, "learning_rate": 4.982471385373663e-05, "loss": 1.4169, "step": 37812 }, { "epoch": 0.04, "learning_rate": 4.982470456845163e-05, "loss": 1.1861, "step": 37813 }, { "epoch": 0.04, "learning_rate": 4.982469528292157e-05, "loss": 1.1452, "step": 37814 }, { "epoch": 0.04, "learning_rate": 4.9824685997146454e-05, "loss": 1.1937, "step": 37815 }, { "epoch": 0.04, "learning_rate": 4.982467671112627e-05, "loss": 0.7352, "step": 37816 }, { "epoch": 0.04, "learning_rate": 4.982466742486104e-05, "loss": 1.2377, "step": 37817 }, { "epoch": 0.04, "learning_rate": 4.9824658138350746e-05, "loss": 1.1742, "step": 37818 }, { "epoch": 0.04, "learning_rate": 4.982464885159539e-05, "loss": 1.1243, "step": 37819 }, { "epoch": 0.04, "learning_rate": 4.982463956459498e-05, "loss": 1.1449, "step": 37820 }, { "epoch": 0.04, "learning_rate": 4.9824630277349506e-05, "loss": 1.3624, "step": 37821 }, { "epoch": 0.04, "learning_rate": 4.982462098985898e-05, "loss": 0.7169, "step": 37822 }, { "epoch": 0.04, "learning_rate": 4.982461170212339e-05, "loss": 0.9842, "step": 37823 }, { "epoch": 0.04, "learning_rate": 4.982460241414275e-05, "loss": 0.9507, "step": 37824 }, { "epoch": 0.04, "learning_rate": 4.982459312591704e-05, "loss": 0.9085, "step": 37825 }, { "epoch": 0.04, "learning_rate": 4.982458383744628e-05, "loss": 0.9287, "step": 37826 }, { "epoch": 0.04, "learning_rate": 4.982457454873046e-05, "loss": 0.9323, "step": 37827 }, { "epoch": 0.04, "learning_rate": 4.9824565259769584e-05, "loss": 0.7083, "step": 37828 }, { "epoch": 0.04, "learning_rate": 4.9824555970563644e-05, "loss": 1.0151, "step": 37829 }, { "epoch": 0.04, "learning_rate": 4.982454668111266e-05, "loss": 0.9819, "step": 37830 }, { "epoch": 0.04, "learning_rate": 4.9824537391416607e-05, "loss": 1.153, "step": 37831 }, { "epoch": 0.04, "learning_rate": 4.9824528101475496e-05, "loss": 1.1613, "step": 37832 }, { "epoch": 0.04, "learning_rate": 4.982451881128934e-05, "loss": 1.0539, "step": 37833 }, { "epoch": 0.04, "learning_rate": 4.982450952085812e-05, "loss": 1.2756, "step": 37834 }, { "epoch": 0.04, "learning_rate": 4.982450023018184e-05, "loss": 1.1452, "step": 37835 }, { "epoch": 0.04, "learning_rate": 4.9824490939260507e-05, "loss": 0.3709, "step": 37836 }, { "epoch": 0.04, "learning_rate": 4.982448164809411e-05, "loss": 0.3397, "step": 37837 }, { "epoch": 0.04, "learning_rate": 4.982447235668266e-05, "loss": 0.5466, "step": 37838 }, { "epoch": 0.04, "learning_rate": 4.982446306502615e-05, "loss": 0.3393, "step": 37839 }, { "epoch": 0.04, "learning_rate": 4.982445377312459e-05, "loss": 0.9014, "step": 37840 }, { "epoch": 0.04, "learning_rate": 4.982444448097797e-05, "loss": 0.9129, "step": 37841 }, { "epoch": 0.04, "learning_rate": 4.98244351885863e-05, "loss": 0.4141, "step": 37842 }, { "epoch": 0.04, "learning_rate": 4.9824425895949566e-05, "loss": 0.306, "step": 37843 }, { "epoch": 0.04, "learning_rate": 4.982441660306778e-05, "loss": 0.6596, "step": 37844 }, { "epoch": 0.04, "learning_rate": 4.982440730994093e-05, "loss": 1.3893, "step": 37845 }, { "epoch": 0.04, "learning_rate": 4.982439801656903e-05, "loss": 1.1021, "step": 37846 }, { "epoch": 0.04, "learning_rate": 4.982438872295208e-05, "loss": 0.8617, "step": 37847 }, { "epoch": 0.04, "learning_rate": 4.9824379429090065e-05, "loss": 1.0718, "step": 37848 }, { "epoch": 0.04, "learning_rate": 4.9824370134983e-05, "loss": 0.7606, "step": 37849 }, { "epoch": 0.04, "learning_rate": 4.9824360840630877e-05, "loss": 0.9479, "step": 37850 }, { "epoch": 0.04, "learning_rate": 4.98243515460337e-05, "loss": 1.0252, "step": 37851 }, { "epoch": 0.04, "learning_rate": 4.982434225119146e-05, "loss": 1.1432, "step": 37852 }, { "epoch": 0.04, "learning_rate": 4.982433295610417e-05, "loss": 0.9762, "step": 37853 }, { "epoch": 0.04, "learning_rate": 4.982432366077182e-05, "loss": 1.1489, "step": 37854 }, { "epoch": 0.04, "learning_rate": 4.982431436519442e-05, "loss": 1.0301, "step": 37855 }, { "epoch": 0.04, "learning_rate": 4.982430506937197e-05, "loss": 0.9778, "step": 37856 }, { "epoch": 0.04, "learning_rate": 4.982429577330445e-05, "loss": 0.8897, "step": 37857 }, { "epoch": 0.04, "learning_rate": 4.982428647699189e-05, "loss": 1.1369, "step": 37858 }, { "epoch": 0.04, "learning_rate": 4.982427718043427e-05, "loss": 0.7642, "step": 37859 }, { "epoch": 0.04, "learning_rate": 4.9824267883631594e-05, "loss": 1.1878, "step": 37860 }, { "epoch": 0.04, "learning_rate": 4.9824258586583864e-05, "loss": 0.9003, "step": 37861 }, { "epoch": 0.04, "learning_rate": 4.982424928929108e-05, "loss": 1.0892, "step": 37862 }, { "epoch": 0.04, "learning_rate": 4.982423999175324e-05, "loss": 0.8865, "step": 37863 }, { "epoch": 0.04, "learning_rate": 4.9824230693970345e-05, "loss": 0.7417, "step": 37864 }, { "epoch": 0.04, "learning_rate": 4.9824221395942394e-05, "loss": 0.8427, "step": 37865 }, { "epoch": 0.04, "learning_rate": 4.9824212097669386e-05, "loss": 1.3075, "step": 37866 }, { "epoch": 0.04, "learning_rate": 4.982420279915133e-05, "loss": 0.766, "step": 37867 }, { "epoch": 0.04, "learning_rate": 4.982419350038822e-05, "loss": 0.984, "step": 37868 }, { "epoch": 0.04, "learning_rate": 4.9824184201380056e-05, "loss": 1.0708, "step": 37869 }, { "epoch": 0.04, "learning_rate": 4.982417490212684e-05, "loss": 0.8843, "step": 37870 }, { "epoch": 0.04, "learning_rate": 4.982416560262857e-05, "loss": 0.9662, "step": 37871 }, { "epoch": 0.04, "learning_rate": 4.9824156302885236e-05, "loss": 0.8243, "step": 37872 }, { "epoch": 0.04, "learning_rate": 4.982414700289686e-05, "loss": 1.1143, "step": 37873 }, { "epoch": 0.04, "learning_rate": 4.982413770266342e-05, "loss": 1.1289, "step": 37874 }, { "epoch": 0.04, "learning_rate": 4.982412840218494e-05, "loss": 0.6466, "step": 37875 }, { "epoch": 0.04, "learning_rate": 4.9824119101461396e-05, "loss": 0.9889, "step": 37876 }, { "epoch": 0.04, "learning_rate": 4.9824109800492804e-05, "loss": 0.8796, "step": 37877 }, { "epoch": 0.04, "learning_rate": 4.9824100499279156e-05, "loss": 0.9545, "step": 37878 }, { "epoch": 0.04, "learning_rate": 4.982409119782046e-05, "loss": 1.0508, "step": 37879 }, { "epoch": 0.04, "learning_rate": 4.98240818961167e-05, "loss": 0.941, "step": 37880 }, { "epoch": 0.04, "learning_rate": 4.98240725941679e-05, "loss": 1.1045, "step": 37881 }, { "epoch": 0.04, "learning_rate": 4.982406329197404e-05, "loss": 1.2596, "step": 37882 }, { "epoch": 0.04, "learning_rate": 4.982405398953513e-05, "loss": 0.9948, "step": 37883 }, { "epoch": 0.04, "learning_rate": 4.9824044686851166e-05, "loss": 1.1863, "step": 37884 }, { "epoch": 0.04, "learning_rate": 4.9824035383922154e-05, "loss": 1.3526, "step": 37885 }, { "epoch": 0.04, "learning_rate": 4.9824026080748085e-05, "loss": 1.455, "step": 37886 }, { "epoch": 0.04, "learning_rate": 4.9824016777328966e-05, "loss": 1.0095, "step": 37887 }, { "epoch": 0.04, "learning_rate": 4.982400747366479e-05, "loss": 0.8219, "step": 37888 }, { "epoch": 0.04, "learning_rate": 4.982399816975556e-05, "loss": 1.1278, "step": 37889 }, { "epoch": 0.04, "learning_rate": 4.982398886560129e-05, "loss": 0.8639, "step": 37890 }, { "epoch": 0.04, "learning_rate": 4.9823979561201956e-05, "loss": 0.7878, "step": 37891 }, { "epoch": 0.04, "learning_rate": 4.982397025655757e-05, "loss": 1.1648, "step": 37892 }, { "epoch": 0.04, "learning_rate": 4.982396095166814e-05, "loss": 0.8872, "step": 37893 }, { "epoch": 0.04, "learning_rate": 4.982395164653366e-05, "loss": 0.8958, "step": 37894 }, { "epoch": 0.04, "learning_rate": 4.9823942341154125e-05, "loss": 0.9464, "step": 37895 }, { "epoch": 0.04, "learning_rate": 4.9823933035529535e-05, "loss": 0.8305, "step": 37896 }, { "epoch": 0.04, "learning_rate": 4.9823923729659895e-05, "loss": 0.6779, "step": 37897 }, { "epoch": 0.04, "learning_rate": 4.9823914423545206e-05, "loss": 1.0682, "step": 37898 }, { "epoch": 0.04, "learning_rate": 4.9823905117185466e-05, "loss": 1.165, "step": 37899 }, { "epoch": 0.04, "learning_rate": 4.982389581058067e-05, "loss": 1.0467, "step": 37900 }, { "epoch": 0.04, "learning_rate": 4.982388650373083e-05, "loss": 0.9446, "step": 37901 }, { "epoch": 0.04, "learning_rate": 4.9823877196635934e-05, "loss": 0.7909, "step": 37902 }, { "epoch": 0.04, "learning_rate": 4.982386788929598e-05, "loss": 1.385, "step": 37903 }, { "epoch": 0.04, "learning_rate": 4.982385858171099e-05, "loss": 1.0452, "step": 37904 }, { "epoch": 0.04, "learning_rate": 4.982384927388094e-05, "loss": 1.05, "step": 37905 }, { "epoch": 0.04, "learning_rate": 4.982383996580584e-05, "loss": 0.9792, "step": 37906 }, { "epoch": 0.04, "learning_rate": 4.9823830657485694e-05, "loss": 0.7622, "step": 37907 }, { "epoch": 0.04, "learning_rate": 4.98238213489205e-05, "loss": 0.9241, "step": 37908 }, { "epoch": 0.04, "learning_rate": 4.982381204011025e-05, "loss": 1.2248, "step": 37909 }, { "epoch": 0.04, "learning_rate": 4.982380273105495e-05, "loss": 0.7885, "step": 37910 }, { "epoch": 0.04, "learning_rate": 4.98237934217546e-05, "loss": 1.0127, "step": 37911 }, { "epoch": 0.04, "learning_rate": 4.9823784112209204e-05, "loss": 1.1456, "step": 37912 }, { "epoch": 0.04, "learning_rate": 4.982377480241876e-05, "loss": 1.2071, "step": 37913 }, { "epoch": 0.04, "learning_rate": 4.9823765492383254e-05, "loss": 0.7298, "step": 37914 }, { "epoch": 0.04, "learning_rate": 4.982375618210271e-05, "loss": 0.7436, "step": 37915 }, { "epoch": 0.04, "learning_rate": 4.9823746871577104e-05, "loss": 1.2327, "step": 37916 }, { "epoch": 0.04, "learning_rate": 4.9823737560806464e-05, "loss": 0.8231, "step": 37917 }, { "epoch": 0.04, "learning_rate": 4.982372824979076e-05, "loss": 1.4733, "step": 37918 }, { "epoch": 0.04, "learning_rate": 4.982371893853002e-05, "loss": 0.7967, "step": 37919 }, { "epoch": 0.04, "learning_rate": 4.9823709627024216e-05, "loss": 1.3101, "step": 37920 }, { "epoch": 0.04, "learning_rate": 4.982370031527337e-05, "loss": 0.8584, "step": 37921 }, { "epoch": 0.04, "learning_rate": 4.982369100327748e-05, "loss": 1.1106, "step": 37922 }, { "epoch": 0.04, "learning_rate": 4.982368169103653e-05, "loss": 0.9103, "step": 37923 }, { "epoch": 0.04, "learning_rate": 4.982367237855054e-05, "loss": 1.0471, "step": 37924 }, { "epoch": 0.04, "learning_rate": 4.98236630658195e-05, "loss": 0.8692, "step": 37925 }, { "epoch": 0.04, "learning_rate": 4.982365375284341e-05, "loss": 1.2036, "step": 37926 }, { "epoch": 0.04, "learning_rate": 4.982364443962227e-05, "loss": 0.4632, "step": 37927 }, { "epoch": 0.04, "learning_rate": 4.982363512615609e-05, "loss": 0.2654, "step": 37928 }, { "epoch": 0.04, "learning_rate": 4.9823625812444855e-05, "loss": 0.1399, "step": 37929 }, { "epoch": 0.04, "learning_rate": 4.9823616498488565e-05, "loss": 0.1971, "step": 37930 }, { "epoch": 0.04, "learning_rate": 4.982360718428724e-05, "loss": 0.1327, "step": 37931 }, { "epoch": 0.04, "learning_rate": 4.9823597869840855e-05, "loss": 0.0909, "step": 37932 }, { "epoch": 0.04, "learning_rate": 4.982358855514943e-05, "loss": 0.1521, "step": 37933 }, { "epoch": 0.04, "learning_rate": 4.9823579240212944e-05, "loss": 0.2204, "step": 37934 }, { "epoch": 0.04, "learning_rate": 4.9823569925031424e-05, "loss": 1.0943, "step": 37935 }, { "epoch": 0.04, "learning_rate": 4.982356060960485e-05, "loss": 0.6719, "step": 37936 }, { "epoch": 0.04, "learning_rate": 4.982355129393323e-05, "loss": 0.9924, "step": 37937 }, { "epoch": 0.04, "learning_rate": 4.9823541978016564e-05, "loss": 0.9906, "step": 37938 }, { "epoch": 0.04, "learning_rate": 4.9823532661854844e-05, "loss": 1.1826, "step": 37939 }, { "epoch": 0.04, "learning_rate": 4.982352334544809e-05, "loss": 0.8392, "step": 37940 }, { "epoch": 0.04, "learning_rate": 4.9823514028796274e-05, "loss": 0.9719, "step": 37941 }, { "epoch": 0.04, "learning_rate": 4.982350471189942e-05, "loss": 0.8721, "step": 37942 }, { "epoch": 0.04, "learning_rate": 4.982349539475751e-05, "loss": 1.2905, "step": 37943 }, { "epoch": 0.04, "learning_rate": 4.982348607737056e-05, "loss": 1.2568, "step": 37944 }, { "epoch": 0.04, "learning_rate": 4.982347675973856e-05, "loss": 0.9812, "step": 37945 }, { "epoch": 0.04, "learning_rate": 4.982346744186152e-05, "loss": 0.9766, "step": 37946 }, { "epoch": 0.04, "learning_rate": 4.9823458123739425e-05, "loss": 0.9826, "step": 37947 }, { "epoch": 0.04, "learning_rate": 4.982344880537229e-05, "loss": 0.7436, "step": 37948 }, { "epoch": 0.04, "learning_rate": 4.98234394867601e-05, "loss": 1.0601, "step": 37949 }, { "epoch": 0.04, "learning_rate": 4.9823430167902866e-05, "loss": 0.9752, "step": 37950 }, { "epoch": 0.04, "learning_rate": 4.982342084880059e-05, "loss": 0.8136, "step": 37951 }, { "epoch": 0.04, "learning_rate": 4.982341152945327e-05, "loss": 0.95, "step": 37952 }, { "epoch": 0.04, "learning_rate": 4.982340220986089e-05, "loss": 1.1965, "step": 37953 }, { "epoch": 0.04, "learning_rate": 4.982339289002348e-05, "loss": 0.8611, "step": 37954 }, { "epoch": 0.04, "learning_rate": 4.9823383569941014e-05, "loss": 1.0006, "step": 37955 }, { "epoch": 0.04, "learning_rate": 4.9823374249613505e-05, "loss": 1.3582, "step": 37956 }, { "epoch": 0.04, "learning_rate": 4.982336492904095e-05, "loss": 1.1068, "step": 37957 }, { "epoch": 0.04, "learning_rate": 4.982335560822334e-05, "loss": 1.0497, "step": 37958 }, { "epoch": 0.04, "learning_rate": 4.98233462871607e-05, "loss": 1.0934, "step": 37959 }, { "epoch": 0.04, "learning_rate": 4.982333696585301e-05, "loss": 1.0228, "step": 37960 }, { "epoch": 0.04, "learning_rate": 4.982332764430027e-05, "loss": 0.83, "step": 37961 }, { "epoch": 0.04, "learning_rate": 4.9823318322502486e-05, "loss": 0.8178, "step": 37962 }, { "epoch": 0.04, "learning_rate": 4.9823309000459653e-05, "loss": 0.9877, "step": 37963 }, { "epoch": 0.04, "learning_rate": 4.9823299678171784e-05, "loss": 0.8448, "step": 37964 }, { "epoch": 0.04, "learning_rate": 4.9823290355638865e-05, "loss": 0.7719, "step": 37965 }, { "epoch": 0.04, "learning_rate": 4.98232810328609e-05, "loss": 1.0936, "step": 37966 }, { "epoch": 0.04, "learning_rate": 4.982327170983789e-05, "loss": 0.9873, "step": 37967 }, { "epoch": 0.04, "learning_rate": 4.9823262386569836e-05, "loss": 0.7429, "step": 37968 }, { "epoch": 0.04, "learning_rate": 4.9823253063056744e-05, "loss": 1.1295, "step": 37969 }, { "epoch": 0.04, "learning_rate": 4.9823243739298595e-05, "loss": 1.2669, "step": 37970 }, { "epoch": 0.04, "learning_rate": 4.98232344152954e-05, "loss": 1.0981, "step": 37971 }, { "epoch": 0.04, "learning_rate": 4.9823225091047174e-05, "loss": 1.1248, "step": 37972 }, { "epoch": 0.04, "learning_rate": 4.9823215766553896e-05, "loss": 1.2389, "step": 37973 }, { "epoch": 0.04, "learning_rate": 4.9823206441815574e-05, "loss": 1.027, "step": 37974 }, { "epoch": 0.04, "learning_rate": 4.982319711683221e-05, "loss": 0.9174, "step": 37975 }, { "epoch": 0.04, "learning_rate": 4.98231877916038e-05, "loss": 0.8941, "step": 37976 }, { "epoch": 0.04, "learning_rate": 4.982317846613035e-05, "loss": 0.9427, "step": 37977 }, { "epoch": 0.04, "learning_rate": 4.9823169140411855e-05, "loss": 1.085, "step": 37978 }, { "epoch": 0.04, "learning_rate": 4.9823159814448303e-05, "loss": 1.056, "step": 37979 }, { "epoch": 0.04, "learning_rate": 4.982315048823972e-05, "loss": 1.0494, "step": 37980 }, { "epoch": 0.04, "learning_rate": 4.982314116178609e-05, "loss": 1.1379, "step": 37981 }, { "epoch": 0.04, "learning_rate": 4.9823131835087423e-05, "loss": 1.3106, "step": 37982 }, { "epoch": 0.04, "learning_rate": 4.9823122508143706e-05, "loss": 1.2157, "step": 37983 }, { "epoch": 0.04, "learning_rate": 4.982311318095494e-05, "loss": 1.117, "step": 37984 }, { "epoch": 0.04, "learning_rate": 4.982310385352114e-05, "loss": 1.0977, "step": 37985 }, { "epoch": 0.04, "learning_rate": 4.982309452584229e-05, "loss": 1.1357, "step": 37986 }, { "epoch": 0.04, "learning_rate": 4.982308519791841e-05, "loss": 1.1279, "step": 37987 }, { "epoch": 0.04, "learning_rate": 4.982307586974947e-05, "loss": 1.1939, "step": 37988 }, { "epoch": 0.04, "learning_rate": 4.982306654133549e-05, "loss": 1.0487, "step": 37989 }, { "epoch": 0.04, "learning_rate": 4.9823057212676485e-05, "loss": 1.006, "step": 37990 }, { "epoch": 0.04, "learning_rate": 4.982304788377242e-05, "loss": 1.0055, "step": 37991 }, { "epoch": 0.04, "learning_rate": 4.9823038554623314e-05, "loss": 1.4001, "step": 37992 }, { "epoch": 0.04, "learning_rate": 4.982302922522917e-05, "loss": 0.9525, "step": 37993 }, { "epoch": 0.04, "learning_rate": 4.9823019895589984e-05, "loss": 0.9925, "step": 37994 }, { "epoch": 0.04, "learning_rate": 4.9823010565705754e-05, "loss": 0.9272, "step": 37995 }, { "epoch": 0.04, "learning_rate": 4.9823001235576475e-05, "loss": 0.5647, "step": 37996 }, { "epoch": 0.04, "learning_rate": 4.982299190520216e-05, "loss": 0.2192, "step": 37997 }, { "epoch": 0.04, "learning_rate": 4.98229825745828e-05, "loss": 0.1178, "step": 37998 }, { "epoch": 0.04, "learning_rate": 4.982297324371841e-05, "loss": 0.1188, "step": 37999 }, { "epoch": 0.04, "learning_rate": 4.9822963912608963e-05, "loss": 0.1006, "step": 38000 }, { "epoch": 0.04, "eval_loss": 1.0789852142333984, "eval_runtime": 126.8511, "eval_samples_per_second": 10.918, "eval_steps_per_second": 5.463, "step": 38000 }, { "epoch": 0.04, "learning_rate": 4.9822954581254474e-05, "loss": 0.12, "step": 38001 }, { "epoch": 0.04, "learning_rate": 4.982294524965495e-05, "loss": 0.0891, "step": 38002 }, { "epoch": 0.04, "learning_rate": 4.9822935917810386e-05, "loss": 0.1732, "step": 38003 }, { "epoch": 0.04, "learning_rate": 4.9822926585720774e-05, "loss": 0.0763, "step": 38004 }, { "epoch": 0.04, "learning_rate": 4.9822917253386125e-05, "loss": 0.1188, "step": 38005 }, { "epoch": 0.04, "learning_rate": 4.982290792080644e-05, "loss": 0.0688, "step": 38006 }, { "epoch": 0.04, "learning_rate": 4.9822898587981705e-05, "loss": 0.098, "step": 38007 }, { "epoch": 0.04, "learning_rate": 4.982288925491193e-05, "loss": 0.1234, "step": 38008 }, { "epoch": 0.04, "learning_rate": 4.982287992159712e-05, "loss": 0.0912, "step": 38009 }, { "epoch": 0.04, "learning_rate": 4.982287058803726e-05, "loss": 0.0985, "step": 38010 }, { "epoch": 0.04, "learning_rate": 4.9822861254232365e-05, "loss": 0.2144, "step": 38011 }, { "epoch": 0.04, "learning_rate": 4.982285192018242e-05, "loss": 0.1565, "step": 38012 }, { "epoch": 0.04, "learning_rate": 4.9822842585887446e-05, "loss": 0.1193, "step": 38013 }, { "epoch": 0.04, "learning_rate": 4.982283325134743e-05, "loss": 0.0553, "step": 38014 }, { "epoch": 0.04, "learning_rate": 4.982282391656237e-05, "loss": 0.0748, "step": 38015 }, { "epoch": 0.04, "learning_rate": 4.982281458153227e-05, "loss": 0.0539, "step": 38016 }, { "epoch": 0.04, "learning_rate": 4.982280524625713e-05, "loss": 0.0743, "step": 38017 }, { "epoch": 0.04, "learning_rate": 4.9822795910736946e-05, "loss": 0.051, "step": 38018 }, { "epoch": 0.04, "learning_rate": 4.9822786574971726e-05, "loss": 0.055, "step": 38019 }, { "epoch": 0.04, "learning_rate": 4.982277723896146e-05, "loss": 0.1481, "step": 38020 }, { "epoch": 0.04, "learning_rate": 4.982276790270617e-05, "loss": 0.1237, "step": 38021 }, { "epoch": 0.04, "learning_rate": 4.9822758566205826e-05, "loss": 0.1172, "step": 38022 }, { "epoch": 0.04, "learning_rate": 4.982274922946044e-05, "loss": 0.2439, "step": 38023 }, { "epoch": 0.04, "learning_rate": 4.982273989247003e-05, "loss": 0.2164, "step": 38024 }, { "epoch": 0.04, "learning_rate": 4.9822730555234564e-05, "loss": 0.1949, "step": 38025 }, { "epoch": 0.04, "learning_rate": 4.982272121775406e-05, "loss": 0.1243, "step": 38026 }, { "epoch": 0.04, "learning_rate": 4.982271188002853e-05, "loss": 0.4218, "step": 38027 }, { "epoch": 0.04, "learning_rate": 4.982270254205795e-05, "loss": 0.5473, "step": 38028 }, { "epoch": 0.04, "learning_rate": 4.982269320384233e-05, "loss": 0.108, "step": 38029 }, { "epoch": 0.04, "learning_rate": 4.9822683865381666e-05, "loss": 0.111, "step": 38030 }, { "epoch": 0.04, "learning_rate": 4.9822674526675975e-05, "loss": 0.06, "step": 38031 }, { "epoch": 0.04, "learning_rate": 4.982266518772524e-05, "loss": 0.099, "step": 38032 }, { "epoch": 0.04, "learning_rate": 4.982265584852947e-05, "loss": 0.0797, "step": 38033 }, { "epoch": 0.04, "learning_rate": 4.9822646509088654e-05, "loss": 0.0541, "step": 38034 }, { "epoch": 0.04, "learning_rate": 4.982263716940281e-05, "loss": 0.3332, "step": 38035 }, { "epoch": 0.04, "learning_rate": 4.9822627829471915e-05, "loss": 0.8138, "step": 38036 }, { "epoch": 0.04, "learning_rate": 4.982261848929599e-05, "loss": 0.9497, "step": 38037 }, { "epoch": 0.04, "learning_rate": 4.982260914887502e-05, "loss": 0.8052, "step": 38038 }, { "epoch": 0.04, "learning_rate": 4.982259980820901e-05, "loss": 0.82, "step": 38039 }, { "epoch": 0.04, "learning_rate": 4.982259046729797e-05, "loss": 0.8285, "step": 38040 }, { "epoch": 0.04, "learning_rate": 4.982258112614189e-05, "loss": 1.3069, "step": 38041 }, { "epoch": 0.04, "learning_rate": 4.982257178474077e-05, "loss": 1.3065, "step": 38042 }, { "epoch": 0.04, "learning_rate": 4.982256244309461e-05, "loss": 1.0987, "step": 38043 }, { "epoch": 0.04, "learning_rate": 4.982255310120342e-05, "loss": 1.144, "step": 38044 }, { "epoch": 0.04, "learning_rate": 4.9822543759067184e-05, "loss": 1.0826, "step": 38045 }, { "epoch": 0.04, "learning_rate": 4.982253441668592e-05, "loss": 1.4089, "step": 38046 }, { "epoch": 0.04, "learning_rate": 4.982252507405961e-05, "loss": 1.3401, "step": 38047 }, { "epoch": 0.04, "learning_rate": 4.9822515731188256e-05, "loss": 1.1262, "step": 38048 }, { "epoch": 0.04, "learning_rate": 4.9822506388071874e-05, "loss": 1.0422, "step": 38049 }, { "epoch": 0.04, "learning_rate": 4.9822497044710456e-05, "loss": 1.1504, "step": 38050 }, { "epoch": 0.04, "learning_rate": 4.9822487701104e-05, "loss": 0.8354, "step": 38051 }, { "epoch": 0.04, "learning_rate": 4.98224783572525e-05, "loss": 1.3611, "step": 38052 }, { "epoch": 0.04, "learning_rate": 4.982246901315597e-05, "loss": 0.9502, "step": 38053 }, { "epoch": 0.04, "learning_rate": 4.9822459668814404e-05, "loss": 1.0657, "step": 38054 }, { "epoch": 0.04, "learning_rate": 4.9822450324227796e-05, "loss": 0.8627, "step": 38055 }, { "epoch": 0.04, "learning_rate": 4.982244097939616e-05, "loss": 0.9574, "step": 38056 }, { "epoch": 0.04, "learning_rate": 4.982243163431948e-05, "loss": 1.1658, "step": 38057 }, { "epoch": 0.04, "learning_rate": 4.982242228899776e-05, "loss": 1.1974, "step": 38058 }, { "epoch": 0.04, "learning_rate": 4.982241294343101e-05, "loss": 0.8426, "step": 38059 }, { "epoch": 0.04, "learning_rate": 4.982240359761922e-05, "loss": 1.0038, "step": 38060 }, { "epoch": 0.04, "learning_rate": 4.982239425156239e-05, "loss": 1.0122, "step": 38061 }, { "epoch": 0.04, "learning_rate": 4.9822384905260535e-05, "loss": 1.026, "step": 38062 }, { "epoch": 0.04, "learning_rate": 4.982237555871363e-05, "loss": 1.1412, "step": 38063 }, { "epoch": 0.04, "learning_rate": 4.98223662119217e-05, "loss": 0.9071, "step": 38064 }, { "epoch": 0.04, "learning_rate": 4.9822356864884735e-05, "loss": 0.9825, "step": 38065 }, { "epoch": 0.04, "learning_rate": 4.9822347517602726e-05, "loss": 1.2179, "step": 38066 }, { "epoch": 0.04, "learning_rate": 4.982233817007568e-05, "loss": 1.4152, "step": 38067 }, { "epoch": 0.04, "learning_rate": 4.982232882230361e-05, "loss": 0.8876, "step": 38068 }, { "epoch": 0.04, "learning_rate": 4.9822319474286494e-05, "loss": 1.0453, "step": 38069 }, { "epoch": 0.04, "learning_rate": 4.982231012602435e-05, "loss": 0.9065, "step": 38070 }, { "epoch": 0.04, "learning_rate": 4.982230077751716e-05, "loss": 1.0494, "step": 38071 }, { "epoch": 0.04, "learning_rate": 4.982229142876494e-05, "loss": 0.9769, "step": 38072 }, { "epoch": 0.04, "learning_rate": 4.982228207976769e-05, "loss": 1.093, "step": 38073 }, { "epoch": 0.04, "learning_rate": 4.9822272730525404e-05, "loss": 1.0402, "step": 38074 }, { "epoch": 0.04, "learning_rate": 4.982226338103808e-05, "loss": 1.1817, "step": 38075 }, { "epoch": 0.04, "learning_rate": 4.982225403130572e-05, "loss": 0.9066, "step": 38076 }, { "epoch": 0.04, "learning_rate": 4.982224468132833e-05, "loss": 1.0534, "step": 38077 }, { "epoch": 0.04, "learning_rate": 4.98222353311059e-05, "loss": 0.9999, "step": 38078 }, { "epoch": 0.04, "learning_rate": 4.9822225980638436e-05, "loss": 1.0229, "step": 38079 }, { "epoch": 0.04, "learning_rate": 4.982221662992594e-05, "loss": 1.1281, "step": 38080 }, { "epoch": 0.04, "learning_rate": 4.982220727896841e-05, "loss": 1.0489, "step": 38081 }, { "epoch": 0.04, "learning_rate": 4.9822197927765843e-05, "loss": 1.0261, "step": 38082 }, { "epoch": 0.04, "learning_rate": 4.982218857631824e-05, "loss": 1.0254, "step": 38083 }, { "epoch": 0.04, "learning_rate": 4.982217922462561e-05, "loss": 1.3157, "step": 38084 }, { "epoch": 0.04, "learning_rate": 4.982216987268794e-05, "loss": 0.9929, "step": 38085 }, { "epoch": 0.04, "learning_rate": 4.982216052050523e-05, "loss": 1.084, "step": 38086 }, { "epoch": 0.04, "learning_rate": 4.98221511680775e-05, "loss": 1.3969, "step": 38087 }, { "epoch": 0.04, "learning_rate": 4.982214181540473e-05, "loss": 0.9169, "step": 38088 }, { "epoch": 0.04, "learning_rate": 4.982213246248693e-05, "loss": 1.1104, "step": 38089 }, { "epoch": 0.04, "learning_rate": 4.9822123109324084e-05, "loss": 1.0561, "step": 38090 }, { "epoch": 0.04, "learning_rate": 4.982211375591621e-05, "loss": 1.0665, "step": 38091 }, { "epoch": 0.04, "learning_rate": 4.982210440226331e-05, "loss": 1.2545, "step": 38092 }, { "epoch": 0.04, "learning_rate": 4.982209504836537e-05, "loss": 1.0088, "step": 38093 }, { "epoch": 0.04, "learning_rate": 4.98220856942224e-05, "loss": 0.818, "step": 38094 }, { "epoch": 0.04, "learning_rate": 4.982207633983439e-05, "loss": 0.5331, "step": 38095 }, { "epoch": 0.04, "learning_rate": 4.9822066985201354e-05, "loss": 1.3049, "step": 38096 }, { "epoch": 0.04, "learning_rate": 4.982205763032328e-05, "loss": 0.9678, "step": 38097 }, { "epoch": 0.04, "learning_rate": 4.982204827520018e-05, "loss": 0.9624, "step": 38098 }, { "epoch": 0.04, "learning_rate": 4.982203891983205e-05, "loss": 1.0975, "step": 38099 }, { "epoch": 0.04, "learning_rate": 4.9822029564218876e-05, "loss": 1.1626, "step": 38100 }, { "epoch": 0.04, "learning_rate": 4.982202020836068e-05, "loss": 0.8275, "step": 38101 }, { "epoch": 0.04, "learning_rate": 4.982201085225744e-05, "loss": 0.5216, "step": 38102 }, { "epoch": 0.04, "learning_rate": 4.9822001495909176e-05, "loss": 1.1174, "step": 38103 }, { "epoch": 0.04, "learning_rate": 4.9821992139315876e-05, "loss": 1.039, "step": 38104 }, { "epoch": 0.04, "learning_rate": 4.982198278247754e-05, "loss": 1.1559, "step": 38105 }, { "epoch": 0.04, "learning_rate": 4.9821973425394174e-05, "loss": 1.1449, "step": 38106 }, { "epoch": 0.04, "learning_rate": 4.9821964068065785e-05, "loss": 0.9628, "step": 38107 }, { "epoch": 0.04, "learning_rate": 4.982195471049236e-05, "loss": 1.15, "step": 38108 }, { "epoch": 0.04, "learning_rate": 4.982194535267389e-05, "loss": 0.9535, "step": 38109 }, { "epoch": 0.04, "learning_rate": 4.98219359946104e-05, "loss": 1.1012, "step": 38110 }, { "epoch": 0.04, "learning_rate": 4.9821926636301886e-05, "loss": 1.005, "step": 38111 }, { "epoch": 0.04, "learning_rate": 4.982191727774833e-05, "loss": 0.9317, "step": 38112 }, { "epoch": 0.04, "learning_rate": 4.982190791894974e-05, "loss": 1.1276, "step": 38113 }, { "epoch": 0.04, "learning_rate": 4.9821898559906125e-05, "loss": 0.9433, "step": 38114 }, { "epoch": 0.04, "learning_rate": 4.982188920061748e-05, "loss": 0.8623, "step": 38115 }, { "epoch": 0.04, "learning_rate": 4.9821879841083795e-05, "loss": 0.902, "step": 38116 }, { "epoch": 0.04, "learning_rate": 4.982187048130509e-05, "loss": 1.5502, "step": 38117 }, { "epoch": 0.04, "learning_rate": 4.982186112128135e-05, "loss": 1.7813, "step": 38118 }, { "epoch": 0.04, "learning_rate": 4.9821851761012575e-05, "loss": 1.9267, "step": 38119 }, { "epoch": 0.04, "learning_rate": 4.9821842400498766e-05, "loss": 1.3723, "step": 38120 }, { "epoch": 0.04, "learning_rate": 4.982183303973993e-05, "loss": 1.5272, "step": 38121 }, { "epoch": 0.04, "learning_rate": 4.982182367873607e-05, "loss": 0.7117, "step": 38122 }, { "epoch": 0.04, "learning_rate": 4.982181431748717e-05, "loss": 1.0498, "step": 38123 }, { "epoch": 0.04, "learning_rate": 4.982180495599325e-05, "loss": 1.1127, "step": 38124 }, { "epoch": 0.04, "learning_rate": 4.9821795594254285e-05, "loss": 1.1161, "step": 38125 }, { "epoch": 0.04, "learning_rate": 4.98217862322703e-05, "loss": 1.2453, "step": 38126 }, { "epoch": 0.04, "learning_rate": 4.982177687004128e-05, "loss": 1.2939, "step": 38127 }, { "epoch": 0.04, "learning_rate": 4.982176750756723e-05, "loss": 0.8829, "step": 38128 }, { "epoch": 0.04, "learning_rate": 4.982175814484816e-05, "loss": 1.0575, "step": 38129 }, { "epoch": 0.04, "learning_rate": 4.982174878188405e-05, "loss": 1.0802, "step": 38130 }, { "epoch": 0.04, "learning_rate": 4.982173941867491e-05, "loss": 0.945, "step": 38131 }, { "epoch": 0.04, "learning_rate": 4.982173005522075e-05, "loss": 0.8847, "step": 38132 }, { "epoch": 0.04, "learning_rate": 4.9821720691521546e-05, "loss": 0.9847, "step": 38133 }, { "epoch": 0.04, "learning_rate": 4.982171132757732e-05, "loss": 1.1142, "step": 38134 }, { "epoch": 0.04, "learning_rate": 4.982170196338807e-05, "loss": 1.0618, "step": 38135 }, { "epoch": 0.04, "learning_rate": 4.9821692598953784e-05, "loss": 0.8979, "step": 38136 }, { "epoch": 0.04, "learning_rate": 4.9821683234274464e-05, "loss": 0.7951, "step": 38137 }, { "epoch": 0.04, "learning_rate": 4.982167386935012e-05, "loss": 1.0714, "step": 38138 }, { "epoch": 0.04, "learning_rate": 4.982166450418075e-05, "loss": 0.9983, "step": 38139 }, { "epoch": 0.04, "learning_rate": 4.9821655138766346e-05, "loss": 1.155, "step": 38140 }, { "epoch": 0.04, "learning_rate": 4.9821645773106914e-05, "loss": 0.7751, "step": 38141 }, { "epoch": 0.04, "learning_rate": 4.982163640720245e-05, "loss": 0.8457, "step": 38142 }, { "epoch": 0.04, "learning_rate": 4.982162704105297e-05, "loss": 0.9632, "step": 38143 }, { "epoch": 0.04, "learning_rate": 4.9821617674658447e-05, "loss": 0.9474, "step": 38144 }, { "epoch": 0.04, "learning_rate": 4.982160830801891e-05, "loss": 0.9177, "step": 38145 }, { "epoch": 0.04, "learning_rate": 4.9821598941134335e-05, "loss": 0.9551, "step": 38146 }, { "epoch": 0.04, "learning_rate": 4.982158957400473e-05, "loss": 1.0378, "step": 38147 }, { "epoch": 0.04, "learning_rate": 4.9821580206630106e-05, "loss": 1.056, "step": 38148 }, { "epoch": 0.04, "learning_rate": 4.982157083901044e-05, "loss": 0.9823, "step": 38149 }, { "epoch": 0.04, "learning_rate": 4.982156147114576e-05, "loss": 0.7644, "step": 38150 }, { "epoch": 0.04, "learning_rate": 4.982155210303604e-05, "loss": 0.9333, "step": 38151 }, { "epoch": 0.04, "learning_rate": 4.9821542734681297e-05, "loss": 0.5851, "step": 38152 }, { "epoch": 0.04, "learning_rate": 4.982153336608153e-05, "loss": 1.0811, "step": 38153 }, { "epoch": 0.04, "learning_rate": 4.982152399723673e-05, "loss": 1.0723, "step": 38154 }, { "epoch": 0.04, "learning_rate": 4.98215146281469e-05, "loss": 3.9783, "step": 38155 }, { "epoch": 0.04, "learning_rate": 4.982150525881205e-05, "loss": 3.2295, "step": 38156 }, { "epoch": 0.04, "learning_rate": 4.9821495889232166e-05, "loss": 2.426, "step": 38157 }, { "epoch": 0.04, "learning_rate": 4.9821486519407257e-05, "loss": 2.6046, "step": 38158 }, { "epoch": 0.04, "learning_rate": 4.9821477149337324e-05, "loss": 3.1165, "step": 38159 }, { "epoch": 0.04, "learning_rate": 4.982146777902236e-05, "loss": 3.1365, "step": 38160 }, { "epoch": 0.04, "learning_rate": 4.9821458408462376e-05, "loss": 3.9309, "step": 38161 }, { "epoch": 0.04, "learning_rate": 4.9821449037657355e-05, "loss": 3.2747, "step": 38162 }, { "epoch": 0.04, "learning_rate": 4.982143966660731e-05, "loss": 3.3296, "step": 38163 }, { "epoch": 0.04, "learning_rate": 4.982143029531224e-05, "loss": 2.5927, "step": 38164 }, { "epoch": 0.04, "learning_rate": 4.9821420923772145e-05, "loss": 3.1314, "step": 38165 }, { "epoch": 0.04, "learning_rate": 4.982141155198702e-05, "loss": 2.8503, "step": 38166 }, { "epoch": 0.04, "learning_rate": 4.9821402179956875e-05, "loss": 3.0498, "step": 38167 }, { "epoch": 0.04, "learning_rate": 4.9821392807681696e-05, "loss": 2.6347, "step": 38168 }, { "epoch": 0.04, "learning_rate": 4.982138343516149e-05, "loss": 3.356, "step": 38169 }, { "epoch": 0.04, "learning_rate": 4.982137406239626e-05, "loss": 3.0016, "step": 38170 }, { "epoch": 0.04, "learning_rate": 4.9821364689386006e-05, "loss": 2.3588, "step": 38171 }, { "epoch": 0.04, "learning_rate": 4.982135531613072e-05, "loss": 2.7685, "step": 38172 }, { "epoch": 0.04, "learning_rate": 4.9821345942630414e-05, "loss": 3.4255, "step": 38173 }, { "epoch": 0.04, "learning_rate": 4.982133656888508e-05, "loss": 2.5384, "step": 38174 }, { "epoch": 0.04, "learning_rate": 4.982132719489473e-05, "loss": 5.8486, "step": 38175 }, { "epoch": 0.04, "learning_rate": 4.982131782065934e-05, "loss": 5.9549, "step": 38176 }, { "epoch": 0.04, "learning_rate": 4.9821308446178934e-05, "loss": 5.9054, "step": 38177 }, { "epoch": 0.04, "learning_rate": 4.98212990714535e-05, "loss": 5.8674, "step": 38178 }, { "epoch": 0.04, "learning_rate": 4.982128969648303e-05, "loss": 5.8574, "step": 38179 }, { "epoch": 0.04, "learning_rate": 4.982128032126755e-05, "loss": 5.7708, "step": 38180 }, { "epoch": 0.04, "learning_rate": 4.982127094580703e-05, "loss": 5.8301, "step": 38181 }, { "epoch": 0.04, "learning_rate": 4.98212615701015e-05, "loss": 5.8647, "step": 38182 }, { "epoch": 0.04, "learning_rate": 4.9821252194150935e-05, "loss": 5.8378, "step": 38183 }, { "epoch": 0.04, "learning_rate": 4.982124281795535e-05, "loss": 5.8297, "step": 38184 }, { "epoch": 0.04, "learning_rate": 4.982123344151474e-05, "loss": 5.851, "step": 38185 }, { "epoch": 0.04, "learning_rate": 4.9821224064829106e-05, "loss": 5.7983, "step": 38186 }, { "epoch": 0.04, "learning_rate": 4.982121468789844e-05, "loss": 5.7924, "step": 38187 }, { "epoch": 0.04, "learning_rate": 4.982120531072276e-05, "loss": 5.7823, "step": 38188 }, { "epoch": 0.04, "learning_rate": 4.982119593330204e-05, "loss": 5.8203, "step": 38189 }, { "epoch": 0.04, "learning_rate": 4.982118655563631e-05, "loss": 5.4615, "step": 38190 }, { "epoch": 0.04, "learning_rate": 4.982117717772555e-05, "loss": 0.5152, "step": 38191 }, { "epoch": 0.04, "learning_rate": 4.982116779956977e-05, "loss": 0.5855, "step": 38192 }, { "epoch": 0.04, "learning_rate": 4.982115842116897e-05, "loss": 1.1657, "step": 38193 }, { "epoch": 0.04, "learning_rate": 4.982114904252313e-05, "loss": 1.0176, "step": 38194 }, { "epoch": 0.04, "learning_rate": 4.982113966363228e-05, "loss": 0.8379, "step": 38195 }, { "epoch": 0.04, "learning_rate": 4.98211302844964e-05, "loss": 1.3808, "step": 38196 }, { "epoch": 0.04, "learning_rate": 4.982112090511549e-05, "loss": 1.151, "step": 38197 }, { "epoch": 0.04, "learning_rate": 4.982111152548957e-05, "loss": 0.8669, "step": 38198 }, { "epoch": 0.04, "learning_rate": 4.982110214561862e-05, "loss": 0.7616, "step": 38199 }, { "epoch": 0.04, "learning_rate": 4.9821092765502646e-05, "loss": 1.221, "step": 38200 }, { "epoch": 0.04, "learning_rate": 4.9821083385141656e-05, "loss": 0.9557, "step": 38201 }, { "epoch": 0.04, "learning_rate": 4.982107400453563e-05, "loss": 1.1799, "step": 38202 }, { "epoch": 0.04, "learning_rate": 4.982106462368459e-05, "loss": 0.9695, "step": 38203 }, { "epoch": 0.04, "learning_rate": 4.982105524258852e-05, "loss": 1.0699, "step": 38204 }, { "epoch": 0.04, "learning_rate": 4.9821045861247427e-05, "loss": 0.8565, "step": 38205 }, { "epoch": 0.04, "learning_rate": 4.982103647966132e-05, "loss": 1.4334, "step": 38206 }, { "epoch": 0.04, "learning_rate": 4.982102709783019e-05, "loss": 1.2557, "step": 38207 }, { "epoch": 0.04, "learning_rate": 4.9821017715754024e-05, "loss": 1.1115, "step": 38208 }, { "epoch": 0.04, "learning_rate": 4.982100833343285e-05, "loss": 0.9721, "step": 38209 }, { "epoch": 0.04, "learning_rate": 4.982099895086664e-05, "loss": 1.1341, "step": 38210 }, { "epoch": 0.04, "learning_rate": 4.982098956805542e-05, "loss": 1.167, "step": 38211 }, { "epoch": 0.04, "learning_rate": 4.982098018499917e-05, "loss": 0.9263, "step": 38212 }, { "epoch": 0.04, "learning_rate": 4.98209708016979e-05, "loss": 0.9583, "step": 38213 }, { "epoch": 0.04, "learning_rate": 4.9820961418151614e-05, "loss": 0.8236, "step": 38214 }, { "epoch": 0.04, "learning_rate": 4.9820952034360296e-05, "loss": 1.3324, "step": 38215 }, { "epoch": 0.04, "learning_rate": 4.982094265032396e-05, "loss": 1.211, "step": 38216 }, { "epoch": 0.04, "learning_rate": 4.98209332660426e-05, "loss": 1.0093, "step": 38217 }, { "epoch": 0.04, "learning_rate": 4.982092388151623e-05, "loss": 1.0561, "step": 38218 }, { "epoch": 0.04, "learning_rate": 4.982091449674482e-05, "loss": 1.1684, "step": 38219 }, { "epoch": 0.04, "learning_rate": 4.98209051117284e-05, "loss": 0.8693, "step": 38220 }, { "epoch": 0.04, "learning_rate": 4.982089572646695e-05, "loss": 0.912, "step": 38221 }, { "epoch": 0.04, "learning_rate": 4.9820886340960486e-05, "loss": 0.8524, "step": 38222 }, { "epoch": 0.04, "learning_rate": 4.9820876955209e-05, "loss": 0.9464, "step": 38223 }, { "epoch": 0.04, "learning_rate": 4.9820867569212496e-05, "loss": 0.6016, "step": 38224 }, { "epoch": 0.04, "learning_rate": 4.982085818297096e-05, "loss": 0.3749, "step": 38225 }, { "epoch": 0.04, "learning_rate": 4.982084879648441e-05, "loss": 0.9142, "step": 38226 }, { "epoch": 0.04, "learning_rate": 4.982083940975284e-05, "loss": 1.1594, "step": 38227 }, { "epoch": 0.04, "learning_rate": 4.9820830022776246e-05, "loss": 1.282, "step": 38228 }, { "epoch": 0.04, "learning_rate": 4.9820820635554635e-05, "loss": 0.9402, "step": 38229 }, { "epoch": 0.04, "learning_rate": 4.9820811248087994e-05, "loss": 0.7548, "step": 38230 }, { "epoch": 0.04, "learning_rate": 4.9820801860376344e-05, "loss": 0.6946, "step": 38231 }, { "epoch": 0.04, "learning_rate": 4.982079247241967e-05, "loss": 1.0818, "step": 38232 }, { "epoch": 0.04, "learning_rate": 4.982078308421797e-05, "loss": 1.0565, "step": 38233 }, { "epoch": 0.04, "learning_rate": 4.982077369577126e-05, "loss": 1.3118, "step": 38234 }, { "epoch": 0.04, "learning_rate": 4.982076430707953e-05, "loss": 0.9632, "step": 38235 }, { "epoch": 0.04, "learning_rate": 4.982075491814276e-05, "loss": 0.9824, "step": 38236 }, { "epoch": 0.04, "learning_rate": 4.982074552896099e-05, "loss": 1.3887, "step": 38237 }, { "epoch": 0.04, "learning_rate": 4.98207361395342e-05, "loss": 1.0321, "step": 38238 }, { "epoch": 0.04, "learning_rate": 4.982072674986238e-05, "loss": 0.9153, "step": 38239 }, { "epoch": 0.04, "learning_rate": 4.982071735994555e-05, "loss": 0.7762, "step": 38240 }, { "epoch": 0.04, "learning_rate": 4.982070796978369e-05, "loss": 1.0465, "step": 38241 }, { "epoch": 0.04, "learning_rate": 4.9820698579376814e-05, "loss": 0.5702, "step": 38242 }, { "epoch": 0.04, "learning_rate": 4.9820689188724916e-05, "loss": 0.4105, "step": 38243 }, { "epoch": 0.04, "learning_rate": 4.982067979782801e-05, "loss": 0.2697, "step": 38244 }, { "epoch": 0.04, "learning_rate": 4.982067040668607e-05, "loss": 0.4437, "step": 38245 }, { "epoch": 0.04, "learning_rate": 4.982066101529912e-05, "loss": 0.8899, "step": 38246 }, { "epoch": 0.04, "learning_rate": 4.982065162366715e-05, "loss": 0.9983, "step": 38247 }, { "epoch": 0.04, "learning_rate": 4.9820642231790156e-05, "loss": 0.9695, "step": 38248 }, { "epoch": 0.04, "learning_rate": 4.982063283966815e-05, "loss": 0.6302, "step": 38249 }, { "epoch": 0.04, "learning_rate": 4.9820623447301116e-05, "loss": 0.7445, "step": 38250 }, { "epoch": 0.04, "learning_rate": 4.9820614054689076e-05, "loss": 0.8329, "step": 38251 }, { "epoch": 0.04, "learning_rate": 4.9820604661832005e-05, "loss": 1.5827, "step": 38252 }, { "epoch": 0.04, "learning_rate": 4.982059526872992e-05, "loss": 2.1867, "step": 38253 }, { "epoch": 0.04, "learning_rate": 4.982058587538282e-05, "loss": 1.9641, "step": 38254 }, { "epoch": 0.04, "learning_rate": 4.98205764817907e-05, "loss": 2.2447, "step": 38255 }, { "epoch": 0.04, "learning_rate": 4.982056708795356e-05, "loss": 1.9846, "step": 38256 }, { "epoch": 0.04, "learning_rate": 4.98205576938714e-05, "loss": 1.9337, "step": 38257 }, { "epoch": 0.04, "learning_rate": 4.982054829954422e-05, "loss": 1.9961, "step": 38258 }, { "epoch": 0.04, "learning_rate": 4.982053890497203e-05, "loss": 1.9138, "step": 38259 }, { "epoch": 0.04, "learning_rate": 4.9820529510154826e-05, "loss": 2.1728, "step": 38260 }, { "epoch": 0.04, "learning_rate": 4.98205201150926e-05, "loss": 2.021, "step": 38261 }, { "epoch": 0.04, "learning_rate": 4.9820510719785345e-05, "loss": 1.9859, "step": 38262 }, { "epoch": 0.04, "learning_rate": 4.9820501324233083e-05, "loss": 2.0462, "step": 38263 }, { "epoch": 0.04, "learning_rate": 4.98204919284358e-05, "loss": 1.9, "step": 38264 }, { "epoch": 0.04, "learning_rate": 4.98204825323935e-05, "loss": 2.321, "step": 38265 }, { "epoch": 0.04, "learning_rate": 4.982047313610618e-05, "loss": 1.9908, "step": 38266 }, { "epoch": 0.04, "learning_rate": 4.982046373957385e-05, "loss": 1.8927, "step": 38267 }, { "epoch": 0.04, "learning_rate": 4.98204543427965e-05, "loss": 1.9252, "step": 38268 }, { "epoch": 0.04, "learning_rate": 4.982044494577413e-05, "loss": 1.941, "step": 38269 }, { "epoch": 0.04, "learning_rate": 4.9820435548506746e-05, "loss": 1.7562, "step": 38270 }, { "epoch": 0.04, "learning_rate": 4.9820426150994335e-05, "loss": 1.7425, "step": 38271 }, { "epoch": 0.04, "learning_rate": 4.982041675323692e-05, "loss": 1.4208, "step": 38272 }, { "epoch": 0.04, "learning_rate": 4.9820407355234485e-05, "loss": 1.2827, "step": 38273 }, { "epoch": 0.04, "learning_rate": 4.982039795698703e-05, "loss": 1.1166, "step": 38274 }, { "epoch": 0.04, "learning_rate": 4.982038855849457e-05, "loss": 1.1381, "step": 38275 }, { "epoch": 0.04, "learning_rate": 4.982037915975708e-05, "loss": 1.0243, "step": 38276 }, { "epoch": 0.04, "learning_rate": 4.982036976077458e-05, "loss": 0.846, "step": 38277 }, { "epoch": 0.04, "learning_rate": 4.982036036154706e-05, "loss": 0.8511, "step": 38278 }, { "epoch": 0.04, "learning_rate": 4.982035096207453e-05, "loss": 1.0444, "step": 38279 }, { "epoch": 0.04, "learning_rate": 4.982034156235698e-05, "loss": 1.3235, "step": 38280 }, { "epoch": 0.04, "learning_rate": 4.982033216239441e-05, "loss": 0.8964, "step": 38281 }, { "epoch": 0.04, "learning_rate": 4.982032276218683e-05, "loss": 0.8448, "step": 38282 }, { "epoch": 0.04, "learning_rate": 4.9820313361734226e-05, "loss": 0.9642, "step": 38283 }, { "epoch": 0.04, "learning_rate": 4.982030396103662e-05, "loss": 1.1826, "step": 38284 }, { "epoch": 0.04, "learning_rate": 4.982029456009399e-05, "loss": 1.1928, "step": 38285 }, { "epoch": 0.04, "learning_rate": 4.982028515890634e-05, "loss": 1.1797, "step": 38286 }, { "epoch": 0.04, "learning_rate": 4.982027575747369e-05, "loss": 0.8409, "step": 38287 }, { "epoch": 0.04, "learning_rate": 4.982026635579601e-05, "loss": 0.9058, "step": 38288 }, { "epoch": 0.04, "learning_rate": 4.982025695387332e-05, "loss": 1.2135, "step": 38289 }, { "epoch": 0.04, "learning_rate": 4.982024755170561e-05, "loss": 1.2771, "step": 38290 }, { "epoch": 0.04, "learning_rate": 4.9820238149292896e-05, "loss": 0.989, "step": 38291 }, { "epoch": 0.04, "learning_rate": 4.982022874663515e-05, "loss": 1.1629, "step": 38292 }, { "epoch": 0.04, "learning_rate": 4.982021934373241e-05, "loss": 0.6651, "step": 38293 }, { "epoch": 0.04, "learning_rate": 4.9820209940584645e-05, "loss": 0.8528, "step": 38294 }, { "epoch": 0.04, "learning_rate": 4.982020053719186e-05, "loss": 2.1207, "step": 38295 }, { "epoch": 0.04, "learning_rate": 4.982019113355407e-05, "loss": 0.9249, "step": 38296 }, { "epoch": 0.04, "learning_rate": 4.982018172967127e-05, "loss": 0.7325, "step": 38297 }, { "epoch": 0.04, "learning_rate": 4.9820172325543443e-05, "loss": 0.9436, "step": 38298 }, { "epoch": 0.04, "learning_rate": 4.98201629211706e-05, "loss": 1.0186, "step": 38299 }, { "epoch": 0.04, "learning_rate": 4.982015351655275e-05, "loss": 0.7706, "step": 38300 }, { "epoch": 0.04, "learning_rate": 4.982014411168989e-05, "loss": 1.0349, "step": 38301 }, { "epoch": 0.04, "learning_rate": 4.9820134706582004e-05, "loss": 1.134, "step": 38302 }, { "epoch": 0.04, "learning_rate": 4.982012530122912e-05, "loss": 0.9498, "step": 38303 }, { "epoch": 0.04, "learning_rate": 4.98201158956312e-05, "loss": 1.0202, "step": 38304 }, { "epoch": 0.04, "learning_rate": 4.982010648978829e-05, "loss": 1.5474, "step": 38305 }, { "epoch": 0.04, "learning_rate": 4.982009708370035e-05, "loss": 1.2205, "step": 38306 }, { "epoch": 0.04, "learning_rate": 4.98200876773674e-05, "loss": 1.0119, "step": 38307 }, { "epoch": 0.04, "learning_rate": 4.982007827078944e-05, "loss": 1.072, "step": 38308 }, { "epoch": 0.04, "learning_rate": 4.9820068863966464e-05, "loss": 0.9588, "step": 38309 }, { "epoch": 0.04, "learning_rate": 4.982005945689848e-05, "loss": 1.1203, "step": 38310 }, { "epoch": 0.04, "learning_rate": 4.9820050049585476e-05, "loss": 0.9451, "step": 38311 }, { "epoch": 0.04, "learning_rate": 4.9820040642027464e-05, "loss": 0.9936, "step": 38312 }, { "epoch": 0.04, "learning_rate": 4.9820031234224436e-05, "loss": 1.0132, "step": 38313 }, { "epoch": 0.04, "learning_rate": 4.982002182617639e-05, "loss": 0.7905, "step": 38314 }, { "epoch": 0.04, "learning_rate": 4.9820012417883346e-05, "loss": 0.7191, "step": 38315 }, { "epoch": 0.04, "learning_rate": 4.9820003009345276e-05, "loss": 0.6211, "step": 38316 }, { "epoch": 0.04, "learning_rate": 4.98199936005622e-05, "loss": 0.6226, "step": 38317 }, { "epoch": 0.04, "learning_rate": 4.981998419153411e-05, "loss": 0.8518, "step": 38318 }, { "epoch": 0.04, "learning_rate": 4.9819974782261005e-05, "loss": 1.266, "step": 38319 }, { "epoch": 0.04, "learning_rate": 4.981996537274289e-05, "loss": 1.1437, "step": 38320 }, { "epoch": 0.04, "learning_rate": 4.981995596297976e-05, "loss": 0.8909, "step": 38321 }, { "epoch": 0.04, "learning_rate": 4.981994655297162e-05, "loss": 1.1259, "step": 38322 }, { "epoch": 0.04, "learning_rate": 4.981993714271847e-05, "loss": 1.2279, "step": 38323 }, { "epoch": 0.04, "learning_rate": 4.981992773222031e-05, "loss": 0.8839, "step": 38324 }, { "epoch": 0.04, "learning_rate": 4.981991832147713e-05, "loss": 0.9806, "step": 38325 }, { "epoch": 0.04, "learning_rate": 4.9819908910488945e-05, "loss": 1.2215, "step": 38326 }, { "epoch": 0.04, "learning_rate": 4.981989949925574e-05, "loss": 1.2816, "step": 38327 }, { "epoch": 0.04, "learning_rate": 4.981989008777753e-05, "loss": 0.96, "step": 38328 }, { "epoch": 0.04, "learning_rate": 4.981988067605431e-05, "loss": 1.1456, "step": 38329 }, { "epoch": 0.04, "learning_rate": 4.981987126408607e-05, "loss": 1.0151, "step": 38330 }, { "epoch": 0.04, "learning_rate": 4.9819861851872826e-05, "loss": 0.9391, "step": 38331 }, { "epoch": 0.04, "learning_rate": 4.981985243941457e-05, "loss": 1.1205, "step": 38332 }, { "epoch": 0.04, "learning_rate": 4.9819843026711304e-05, "loss": 1.0132, "step": 38333 }, { "epoch": 0.04, "learning_rate": 4.9819833613763025e-05, "loss": 1.0287, "step": 38334 }, { "epoch": 0.04, "learning_rate": 4.981982420056974e-05, "loss": 0.8527, "step": 38335 }, { "epoch": 0.04, "learning_rate": 4.981981478713143e-05, "loss": 0.894, "step": 38336 }, { "epoch": 0.04, "learning_rate": 4.9819805373448126e-05, "loss": 0.823, "step": 38337 }, { "epoch": 0.04, "learning_rate": 4.98197959595198e-05, "loss": 0.7801, "step": 38338 }, { "epoch": 0.04, "learning_rate": 4.981978654534647e-05, "loss": 1.1055, "step": 38339 }, { "epoch": 0.04, "learning_rate": 4.981977713092812e-05, "loss": 1.0594, "step": 38340 }, { "epoch": 0.04, "learning_rate": 4.981976771626478e-05, "loss": 1.1607, "step": 38341 }, { "epoch": 0.04, "learning_rate": 4.981975830135641e-05, "loss": 0.8448, "step": 38342 }, { "epoch": 0.04, "learning_rate": 4.9819748886203035e-05, "loss": 1.0283, "step": 38343 }, { "epoch": 0.04, "learning_rate": 4.9819739470804656e-05, "loss": 1.1144, "step": 38344 }, { "epoch": 0.04, "learning_rate": 4.981973005516126e-05, "loss": 0.9208, "step": 38345 }, { "epoch": 0.04, "learning_rate": 4.981972063927286e-05, "loss": 0.8463, "step": 38346 }, { "epoch": 0.04, "learning_rate": 4.981971122313944e-05, "loss": 0.8361, "step": 38347 }, { "epoch": 0.04, "learning_rate": 4.981970180676102e-05, "loss": 0.6333, "step": 38348 }, { "epoch": 0.04, "learning_rate": 4.981969239013759e-05, "loss": 0.7243, "step": 38349 }, { "epoch": 0.04, "learning_rate": 4.9819682973269145e-05, "loss": 1.3726, "step": 38350 }, { "epoch": 0.04, "learning_rate": 4.9819673556155693e-05, "loss": 1.1268, "step": 38351 }, { "epoch": 0.04, "learning_rate": 4.9819664138797226e-05, "loss": 0.7949, "step": 38352 }, { "epoch": 0.04, "learning_rate": 4.981965472119376e-05, "loss": 1.0546, "step": 38353 }, { "epoch": 0.04, "learning_rate": 4.981964530334528e-05, "loss": 1.2077, "step": 38354 }, { "epoch": 0.04, "learning_rate": 4.9819635885251794e-05, "loss": 0.8945, "step": 38355 }, { "epoch": 0.04, "learning_rate": 4.9819626466913296e-05, "loss": 0.2313, "step": 38356 }, { "epoch": 0.04, "learning_rate": 4.981961704832979e-05, "loss": 0.8967, "step": 38357 }, { "epoch": 0.04, "learning_rate": 4.981960762950127e-05, "loss": 1.8335, "step": 38358 }, { "epoch": 0.04, "learning_rate": 4.9819598210427746e-05, "loss": 2.5156, "step": 38359 }, { "epoch": 0.04, "learning_rate": 4.981958879110922e-05, "loss": 1.4753, "step": 38360 }, { "epoch": 0.04, "learning_rate": 4.981957937154568e-05, "loss": 0.9679, "step": 38361 }, { "epoch": 0.04, "learning_rate": 4.9819569951737124e-05, "loss": 1.5134, "step": 38362 }, { "epoch": 0.04, "learning_rate": 4.981956053168357e-05, "loss": 1.4071, "step": 38363 }, { "epoch": 0.04, "learning_rate": 4.9819551111385e-05, "loss": 1.2637, "step": 38364 }, { "epoch": 0.04, "learning_rate": 4.9819541690841425e-05, "loss": 1.3477, "step": 38365 }, { "epoch": 0.04, "learning_rate": 4.981953227005284e-05, "loss": 0.91, "step": 38366 }, { "epoch": 0.04, "learning_rate": 4.981952284901925e-05, "loss": 1.0328, "step": 38367 }, { "epoch": 0.04, "learning_rate": 4.9819513427740656e-05, "loss": 0.7044, "step": 38368 }, { "epoch": 0.04, "learning_rate": 4.981950400621705e-05, "loss": 1.2595, "step": 38369 }, { "epoch": 0.04, "learning_rate": 4.9819494584448434e-05, "loss": 0.5962, "step": 38370 }, { "epoch": 0.04, "learning_rate": 4.9819485162434816e-05, "loss": 0.9166, "step": 38371 }, { "epoch": 0.04, "learning_rate": 4.981947574017619e-05, "loss": 0.7567, "step": 38372 }, { "epoch": 0.04, "learning_rate": 4.9819466317672544e-05, "loss": 0.5856, "step": 38373 }, { "epoch": 0.04, "learning_rate": 4.9819456894923905e-05, "loss": 1.0513, "step": 38374 }, { "epoch": 0.04, "learning_rate": 4.9819447471930256e-05, "loss": 1.0143, "step": 38375 }, { "epoch": 0.04, "learning_rate": 4.98194380486916e-05, "loss": 1.0559, "step": 38376 }, { "epoch": 0.04, "learning_rate": 4.981942862520793e-05, "loss": 1.1043, "step": 38377 }, { "epoch": 0.04, "learning_rate": 4.981941920147925e-05, "loss": 0.9608, "step": 38378 }, { "epoch": 0.04, "learning_rate": 4.981940977750558e-05, "loss": 0.9287, "step": 38379 }, { "epoch": 0.04, "learning_rate": 4.981940035328689e-05, "loss": 0.5108, "step": 38380 }, { "epoch": 0.04, "learning_rate": 4.9819390928823206e-05, "loss": 0.9565, "step": 38381 }, { "epoch": 0.04, "learning_rate": 4.9819381504114505e-05, "loss": 1.0739, "step": 38382 }, { "epoch": 0.04, "learning_rate": 4.9819372079160795e-05, "loss": 1.0763, "step": 38383 }, { "epoch": 0.04, "learning_rate": 4.981936265396209e-05, "loss": 1.0349, "step": 38384 }, { "epoch": 0.04, "learning_rate": 4.9819353228518374e-05, "loss": 0.8668, "step": 38385 }, { "epoch": 0.04, "learning_rate": 4.981934380282964e-05, "loss": 0.8446, "step": 38386 }, { "epoch": 0.04, "learning_rate": 4.9819334376895915e-05, "loss": 0.8885, "step": 38387 }, { "epoch": 0.04, "learning_rate": 4.981932495071718e-05, "loss": 1.413, "step": 38388 }, { "epoch": 0.04, "learning_rate": 4.981931552429344e-05, "loss": 1.0934, "step": 38389 }, { "epoch": 0.04, "learning_rate": 4.981930609762468e-05, "loss": 0.8893, "step": 38390 }, { "epoch": 0.04, "learning_rate": 4.981929667071094e-05, "loss": 0.6819, "step": 38391 }, { "epoch": 0.04, "learning_rate": 4.981928724355218e-05, "loss": 0.6188, "step": 38392 }, { "epoch": 0.04, "learning_rate": 4.9819277816148415e-05, "loss": 0.3937, "step": 38393 }, { "epoch": 0.04, "learning_rate": 4.981926838849964e-05, "loss": 0.3549, "step": 38394 }, { "epoch": 0.04, "learning_rate": 4.981925896060587e-05, "loss": 0.5067, "step": 38395 }, { "epoch": 0.04, "learning_rate": 4.981924953246708e-05, "loss": 0.3479, "step": 38396 }, { "epoch": 0.04, "learning_rate": 4.9819240104083295e-05, "loss": 0.4608, "step": 38397 }, { "epoch": 0.04, "learning_rate": 4.981923067545451e-05, "loss": 0.5041, "step": 38398 }, { "epoch": 0.04, "learning_rate": 4.981922124658071e-05, "loss": 0.3056, "step": 38399 }, { "epoch": 0.04, "learning_rate": 4.981921181746191e-05, "loss": 0.5782, "step": 38400 }, { "epoch": 0.04, "learning_rate": 4.981920238809811e-05, "loss": 1.2711, "step": 38401 }, { "epoch": 0.04, "learning_rate": 4.9819192958489294e-05, "loss": 0.9225, "step": 38402 }, { "epoch": 0.04, "learning_rate": 4.981918352863548e-05, "loss": 0.8207, "step": 38403 }, { "epoch": 0.04, "learning_rate": 4.981917409853666e-05, "loss": 0.9274, "step": 38404 }, { "epoch": 0.04, "learning_rate": 4.981916466819283e-05, "loss": 0.91, "step": 38405 }, { "epoch": 0.04, "learning_rate": 4.9819155237604006e-05, "loss": 0.8545, "step": 38406 }, { "epoch": 0.04, "learning_rate": 4.981914580677017e-05, "loss": 1.2749, "step": 38407 }, { "epoch": 0.04, "learning_rate": 4.981913637569133e-05, "loss": 0.7851, "step": 38408 }, { "epoch": 0.04, "learning_rate": 4.981912694436749e-05, "loss": 0.855, "step": 38409 }, { "epoch": 0.04, "learning_rate": 4.981911751279865e-05, "loss": 0.9687, "step": 38410 }, { "epoch": 0.04, "learning_rate": 4.98191080809848e-05, "loss": 1.0177, "step": 38411 }, { "epoch": 0.04, "learning_rate": 4.981909864892595e-05, "loss": 0.9417, "step": 38412 }, { "epoch": 0.04, "learning_rate": 4.981908921662209e-05, "loss": 0.9325, "step": 38413 }, { "epoch": 0.04, "learning_rate": 4.9819079784073234e-05, "loss": 0.8303, "step": 38414 }, { "epoch": 0.04, "learning_rate": 4.981907035127937e-05, "loss": 0.9853, "step": 38415 }, { "epoch": 0.04, "learning_rate": 4.98190609182405e-05, "loss": 1.1194, "step": 38416 }, { "epoch": 0.04, "learning_rate": 4.981905148495664e-05, "loss": 1.301, "step": 38417 }, { "epoch": 0.04, "learning_rate": 4.9819042051427755e-05, "loss": 1.0164, "step": 38418 }, { "epoch": 0.04, "learning_rate": 4.9819032617653883e-05, "loss": 0.547, "step": 38419 }, { "epoch": 0.04, "learning_rate": 4.9819023183635e-05, "loss": 0.9329, "step": 38420 }, { "epoch": 0.04, "learning_rate": 4.981901374937113e-05, "loss": 0.9355, "step": 38421 }, { "epoch": 0.04, "learning_rate": 4.981900431486224e-05, "loss": 0.9196, "step": 38422 }, { "epoch": 0.04, "learning_rate": 4.981899488010835e-05, "loss": 1.1356, "step": 38423 }, { "epoch": 0.04, "learning_rate": 4.9818985445109455e-05, "loss": 1.003, "step": 38424 }, { "epoch": 0.04, "learning_rate": 4.981897600986556e-05, "loss": 1.0198, "step": 38425 }, { "epoch": 0.04, "learning_rate": 4.9818966574376666e-05, "loss": 1.3223, "step": 38426 }, { "epoch": 0.04, "learning_rate": 4.9818957138642774e-05, "loss": 1.3583, "step": 38427 }, { "epoch": 0.04, "learning_rate": 4.981894770266387e-05, "loss": 1.812, "step": 38428 }, { "epoch": 0.04, "learning_rate": 4.981893826643997e-05, "loss": 1.952, "step": 38429 }, { "epoch": 0.04, "learning_rate": 4.981892882997106e-05, "loss": 1.8678, "step": 38430 }, { "epoch": 0.04, "learning_rate": 4.981891939325716e-05, "loss": 1.831, "step": 38431 }, { "epoch": 0.04, "learning_rate": 4.981890995629825e-05, "loss": 1.6897, "step": 38432 }, { "epoch": 0.04, "learning_rate": 4.9818900519094335e-05, "loss": 1.3581, "step": 38433 }, { "epoch": 0.04, "learning_rate": 4.9818891081645425e-05, "loss": 1.0568, "step": 38434 }, { "epoch": 0.04, "learning_rate": 4.981888164395151e-05, "loss": 0.932, "step": 38435 }, { "epoch": 0.04, "learning_rate": 4.98188722060126e-05, "loss": 0.6176, "step": 38436 }, { "epoch": 0.04, "learning_rate": 4.981886276782868e-05, "loss": 0.5352, "step": 38437 }, { "epoch": 0.04, "learning_rate": 4.981885332939976e-05, "loss": 0.5742, "step": 38438 }, { "epoch": 0.04, "learning_rate": 4.981884389072584e-05, "loss": 0.5076, "step": 38439 }, { "epoch": 0.04, "learning_rate": 4.9818834451806925e-05, "loss": 0.2194, "step": 38440 }, { "epoch": 0.04, "learning_rate": 4.9818825012643e-05, "loss": 2.0516, "step": 38441 }, { "epoch": 0.04, "learning_rate": 4.981881557323408e-05, "loss": 1.6366, "step": 38442 }, { "epoch": 0.04, "learning_rate": 4.981880613358015e-05, "loss": 1.2064, "step": 38443 }, { "epoch": 0.04, "learning_rate": 4.981879669368123e-05, "loss": 1.3509, "step": 38444 }, { "epoch": 0.04, "learning_rate": 4.98187872535373e-05, "loss": 0.9231, "step": 38445 }, { "epoch": 0.04, "learning_rate": 4.981877781314838e-05, "loss": 1.0718, "step": 38446 }, { "epoch": 0.04, "learning_rate": 4.9818768372514455e-05, "loss": 1.4016, "step": 38447 }, { "epoch": 0.04, "learning_rate": 4.981875893163553e-05, "loss": 1.4218, "step": 38448 }, { "epoch": 0.04, "learning_rate": 4.98187494905116e-05, "loss": 1.2848, "step": 38449 }, { "epoch": 0.04, "learning_rate": 4.981874004914268e-05, "loss": 1.0948, "step": 38450 }, { "epoch": 0.04, "learning_rate": 4.9818730607528744e-05, "loss": 0.9559, "step": 38451 }, { "epoch": 0.04, "learning_rate": 4.981872116566982e-05, "loss": 0.2871, "step": 38452 }, { "epoch": 0.04, "learning_rate": 4.981871172356589e-05, "loss": 0.7197, "step": 38453 }, { "epoch": 0.04, "learning_rate": 4.981870228121697e-05, "loss": 1.2391, "step": 38454 }, { "epoch": 0.04, "learning_rate": 4.981869283862304e-05, "loss": 0.7527, "step": 38455 }, { "epoch": 0.04, "learning_rate": 4.981868339578411e-05, "loss": 0.8071, "step": 38456 }, { "epoch": 0.04, "learning_rate": 4.981867395270018e-05, "loss": 0.7371, "step": 38457 }, { "epoch": 0.04, "learning_rate": 4.981866450937126e-05, "loss": 0.6113, "step": 38458 }, { "epoch": 0.04, "learning_rate": 4.981865506579734e-05, "loss": 0.3383, "step": 38459 }, { "epoch": 0.04, "learning_rate": 4.981864562197841e-05, "loss": 0.4462, "step": 38460 }, { "epoch": 0.04, "learning_rate": 4.981863617791449e-05, "loss": 0.3533, "step": 38461 }, { "epoch": 0.04, "learning_rate": 4.981862673360557e-05, "loss": 0.4039, "step": 38462 }, { "epoch": 0.04, "learning_rate": 4.981861728905164e-05, "loss": 0.3231, "step": 38463 }, { "epoch": 0.04, "learning_rate": 4.9818607844252724e-05, "loss": 0.4778, "step": 38464 }, { "epoch": 0.04, "learning_rate": 4.98185983992088e-05, "loss": 0.3968, "step": 38465 }, { "epoch": 0.04, "learning_rate": 4.9818588953919884e-05, "loss": 0.3479, "step": 38466 }, { "epoch": 0.04, "learning_rate": 4.981857950838596e-05, "loss": 1.0353, "step": 38467 }, { "epoch": 0.04, "learning_rate": 4.9818570062607054e-05, "loss": 1.3072, "step": 38468 }, { "epoch": 0.04, "learning_rate": 4.9818560616583134e-05, "loss": 1.2412, "step": 38469 }, { "epoch": 0.04, "learning_rate": 4.9818551170314226e-05, "loss": 1.0858, "step": 38470 }, { "epoch": 0.04, "learning_rate": 4.9818541723800316e-05, "loss": 0.9076, "step": 38471 }, { "epoch": 0.04, "learning_rate": 4.981853227704141e-05, "loss": 1.1609, "step": 38472 }, { "epoch": 0.04, "learning_rate": 4.98185228300375e-05, "loss": 0.735, "step": 38473 }, { "epoch": 0.04, "learning_rate": 4.9818513382788595e-05, "loss": 0.9604, "step": 38474 }, { "epoch": 0.04, "learning_rate": 4.9818503935294694e-05, "loss": 1.2164, "step": 38475 }, { "epoch": 0.04, "learning_rate": 4.981849448755579e-05, "loss": 0.9835, "step": 38476 }, { "epoch": 0.04, "learning_rate": 4.981848503957189e-05, "loss": 0.8852, "step": 38477 }, { "epoch": 0.04, "learning_rate": 4.9818475591343004e-05, "loss": 1.0263, "step": 38478 }, { "epoch": 0.04, "learning_rate": 4.98184661428691e-05, "loss": 0.9205, "step": 38479 }, { "epoch": 0.04, "learning_rate": 4.981845669415022e-05, "loss": 1.021, "step": 38480 }, { "epoch": 0.04, "learning_rate": 4.981844724518633e-05, "loss": 1.0312, "step": 38481 }, { "epoch": 0.04, "learning_rate": 4.981843779597744e-05, "loss": 0.7331, "step": 38482 }, { "epoch": 0.04, "learning_rate": 4.981842834652356e-05, "loss": 0.9252, "step": 38483 }, { "epoch": 0.04, "learning_rate": 4.981841889682468e-05, "loss": 1.1737, "step": 38484 }, { "epoch": 0.04, "learning_rate": 4.9818409446880795e-05, "loss": 1.1676, "step": 38485 }, { "epoch": 0.04, "learning_rate": 4.981839999669193e-05, "loss": 1.4161, "step": 38486 }, { "epoch": 0.04, "learning_rate": 4.9818390546258056e-05, "loss": 0.8829, "step": 38487 }, { "epoch": 0.04, "learning_rate": 4.981838109557919e-05, "loss": 1.1827, "step": 38488 }, { "epoch": 0.04, "learning_rate": 4.981837164465533e-05, "loss": 0.9142, "step": 38489 }, { "epoch": 0.04, "learning_rate": 4.981836219348647e-05, "loss": 0.7533, "step": 38490 }, { "epoch": 0.04, "learning_rate": 4.981835274207261e-05, "loss": 0.8827, "step": 38491 }, { "epoch": 0.04, "learning_rate": 4.981834329041376e-05, "loss": 0.9854, "step": 38492 }, { "epoch": 0.04, "learning_rate": 4.981833383850991e-05, "loss": 1.1419, "step": 38493 }, { "epoch": 0.04, "learning_rate": 4.9818324386361064e-05, "loss": 0.8786, "step": 38494 }, { "epoch": 0.04, "learning_rate": 4.981831493396723e-05, "loss": 0.1386, "step": 38495 }, { "epoch": 0.04, "learning_rate": 4.981830548132839e-05, "loss": 0.7674, "step": 38496 }, { "epoch": 0.04, "learning_rate": 4.981829602844456e-05, "loss": 1.2672, "step": 38497 }, { "epoch": 0.04, "learning_rate": 4.981828657531573e-05, "loss": 1.1475, "step": 38498 }, { "epoch": 0.04, "learning_rate": 4.9818277121941906e-05, "loss": 0.9492, "step": 38499 }, { "epoch": 0.04, "learning_rate": 4.981826766832308e-05, "loss": 0.8442, "step": 38500 }, { "epoch": 0.04, "eval_loss": 1.11681067943573, "eval_runtime": 126.5635, "eval_samples_per_second": 10.943, "eval_steps_per_second": 5.476, "step": 38500 }, { "epoch": 0.04, "learning_rate": 4.981825821445927e-05, "loss": 0.9903, "step": 38501 }, { "epoch": 0.04, "learning_rate": 4.981824876035045e-05, "loss": 0.8488, "step": 38502 }, { "epoch": 0.04, "learning_rate": 4.981823930599665e-05, "loss": 0.8859, "step": 38503 }, { "epoch": 0.04, "learning_rate": 4.9818229851397846e-05, "loss": 0.9386, "step": 38504 }, { "epoch": 0.04, "learning_rate": 4.981822039655405e-05, "loss": 0.8622, "step": 38505 }, { "epoch": 0.04, "learning_rate": 4.981821094146526e-05, "loss": 0.9196, "step": 38506 }, { "epoch": 0.04, "learning_rate": 4.981820148613147e-05, "loss": 1.1095, "step": 38507 }, { "epoch": 0.04, "learning_rate": 4.981819203055269e-05, "loss": 1.112, "step": 38508 }, { "epoch": 0.04, "learning_rate": 4.981818257472891e-05, "loss": 1.1259, "step": 38509 }, { "epoch": 0.04, "learning_rate": 4.9818173118660144e-05, "loss": 0.9027, "step": 38510 }, { "epoch": 0.04, "learning_rate": 4.981816366234638e-05, "loss": 0.904, "step": 38511 }, { "epoch": 0.04, "learning_rate": 4.981815420578762e-05, "loss": 0.949, "step": 38512 }, { "epoch": 0.04, "learning_rate": 4.9818144748983866e-05, "loss": 0.4521, "step": 38513 }, { "epoch": 0.04, "learning_rate": 4.981813529193512e-05, "loss": 0.8411, "step": 38514 }, { "epoch": 0.04, "learning_rate": 4.981812583464137e-05, "loss": 1.2238, "step": 38515 }, { "epoch": 0.04, "learning_rate": 4.981811637710264e-05, "loss": 1.275, "step": 38516 }, { "epoch": 0.04, "learning_rate": 4.981810691931891e-05, "loss": 0.9603, "step": 38517 }, { "epoch": 0.04, "learning_rate": 4.981809746129018e-05, "loss": 0.9464, "step": 38518 }, { "epoch": 0.04, "learning_rate": 4.9818088003016464e-05, "loss": 0.9716, "step": 38519 }, { "epoch": 0.04, "learning_rate": 4.981807854449775e-05, "loss": 0.9918, "step": 38520 }, { "epoch": 0.04, "learning_rate": 4.981806908573405e-05, "loss": 1.133, "step": 38521 }, { "epoch": 0.04, "learning_rate": 4.981805962672535e-05, "loss": 1.1406, "step": 38522 }, { "epoch": 0.04, "learning_rate": 4.9818050167471654e-05, "loss": 1.0353, "step": 38523 }, { "epoch": 0.04, "learning_rate": 4.9818040707972966e-05, "loss": 0.8461, "step": 38524 }, { "epoch": 0.04, "learning_rate": 4.981803124822929e-05, "loss": 0.988, "step": 38525 }, { "epoch": 0.04, "learning_rate": 4.981802178824062e-05, "loss": 1.2159, "step": 38526 }, { "epoch": 0.04, "learning_rate": 4.981801232800696e-05, "loss": 0.7613, "step": 38527 }, { "epoch": 0.04, "learning_rate": 4.9818002867528294e-05, "loss": 1.032, "step": 38528 }, { "epoch": 0.04, "learning_rate": 4.981799340680464e-05, "loss": 1.0049, "step": 38529 }, { "epoch": 0.04, "learning_rate": 4.9817983945836e-05, "loss": 0.8946, "step": 38530 }, { "epoch": 0.04, "learning_rate": 4.9817974484622364e-05, "loss": 1.1809, "step": 38531 }, { "epoch": 0.04, "learning_rate": 4.9817965023163736e-05, "loss": 1.0515, "step": 38532 }, { "epoch": 0.04, "learning_rate": 4.981795556146012e-05, "loss": 1.271, "step": 38533 }, { "epoch": 0.04, "learning_rate": 4.9817946099511506e-05, "loss": 1.1589, "step": 38534 }, { "epoch": 0.04, "learning_rate": 4.98179366373179e-05, "loss": 1.2779, "step": 38535 }, { "epoch": 0.04, "learning_rate": 4.98179271748793e-05, "loss": 1.2558, "step": 38536 }, { "epoch": 0.04, "learning_rate": 4.9817917712195706e-05, "loss": 1.2964, "step": 38537 }, { "epoch": 0.04, "learning_rate": 4.981790824926712e-05, "loss": 1.4458, "step": 38538 }, { "epoch": 0.04, "learning_rate": 4.9817898786093545e-05, "loss": 1.1785, "step": 38539 }, { "epoch": 0.04, "learning_rate": 4.9817889322674985e-05, "loss": 1.158, "step": 38540 }, { "epoch": 0.04, "learning_rate": 4.981787985901142e-05, "loss": 0.8979, "step": 38541 }, { "epoch": 0.04, "learning_rate": 4.981787039510287e-05, "loss": 0.6473, "step": 38542 }, { "epoch": 0.04, "learning_rate": 4.9817860930949335e-05, "loss": 0.7189, "step": 38543 }, { "epoch": 0.04, "learning_rate": 4.9817851466550805e-05, "loss": 1.0136, "step": 38544 }, { "epoch": 0.04, "learning_rate": 4.981784200190728e-05, "loss": 1.2871, "step": 38545 }, { "epoch": 0.04, "learning_rate": 4.981783253701876e-05, "loss": 1.3234, "step": 38546 }, { "epoch": 0.04, "learning_rate": 4.981782307188525e-05, "loss": 0.9848, "step": 38547 }, { "epoch": 0.04, "learning_rate": 4.981781360650676e-05, "loss": 0.6009, "step": 38548 }, { "epoch": 0.04, "learning_rate": 4.981780414088327e-05, "loss": 0.6753, "step": 38549 }, { "epoch": 0.04, "learning_rate": 4.9817794675014785e-05, "loss": 0.3237, "step": 38550 }, { "epoch": 0.04, "learning_rate": 4.981778520890131e-05, "loss": 0.9412, "step": 38551 }, { "epoch": 0.04, "learning_rate": 4.9817775742542855e-05, "loss": 0.7891, "step": 38552 }, { "epoch": 0.04, "learning_rate": 4.9817766275939403e-05, "loss": 0.8832, "step": 38553 }, { "epoch": 0.04, "learning_rate": 4.981775680909096e-05, "loss": 1.1064, "step": 38554 }, { "epoch": 0.04, "learning_rate": 4.9817747341997526e-05, "loss": 0.8338, "step": 38555 }, { "epoch": 0.04, "learning_rate": 4.981773787465911e-05, "loss": 0.8247, "step": 38556 }, { "epoch": 0.04, "learning_rate": 4.981772840707569e-05, "loss": 1.0508, "step": 38557 }, { "epoch": 0.04, "learning_rate": 4.981771893924729e-05, "loss": 1.2707, "step": 38558 }, { "epoch": 0.04, "learning_rate": 4.9817709471173895e-05, "loss": 1.1821, "step": 38559 }, { "epoch": 0.04, "learning_rate": 4.981770000285551e-05, "loss": 1.0856, "step": 38560 }, { "epoch": 0.04, "learning_rate": 4.9817690534292135e-05, "loss": 0.5991, "step": 38561 }, { "epoch": 0.04, "learning_rate": 4.9817681065483775e-05, "loss": 1.1235, "step": 38562 }, { "epoch": 0.04, "learning_rate": 4.981767159643042e-05, "loss": 1.1102, "step": 38563 }, { "epoch": 0.04, "learning_rate": 4.981766212713208e-05, "loss": 1.156, "step": 38564 }, { "epoch": 0.04, "learning_rate": 4.981765265758875e-05, "loss": 0.9242, "step": 38565 }, { "epoch": 0.04, "learning_rate": 4.9817643187800425e-05, "loss": 0.2639, "step": 38566 }, { "epoch": 0.04, "learning_rate": 4.981763371776711e-05, "loss": 0.7855, "step": 38567 }, { "epoch": 0.04, "learning_rate": 4.981762424748881e-05, "loss": 0.8915, "step": 38568 }, { "epoch": 0.04, "learning_rate": 4.981761477696553e-05, "loss": 0.9768, "step": 38569 }, { "epoch": 0.04, "learning_rate": 4.981760530619724e-05, "loss": 0.6877, "step": 38570 }, { "epoch": 0.04, "learning_rate": 4.981759583518398e-05, "loss": 1.0143, "step": 38571 }, { "epoch": 0.04, "learning_rate": 4.981758636392573e-05, "loss": 0.8901, "step": 38572 }, { "epoch": 0.04, "learning_rate": 4.981757689242248e-05, "loss": 1.1905, "step": 38573 }, { "epoch": 0.04, "learning_rate": 4.981756742067425e-05, "loss": 0.7455, "step": 38574 }, { "epoch": 0.04, "learning_rate": 4.9817557948681026e-05, "loss": 1.2296, "step": 38575 }, { "epoch": 0.04, "learning_rate": 4.981754847644282e-05, "loss": 1.1804, "step": 38576 }, { "epoch": 0.04, "learning_rate": 4.981753900395961e-05, "loss": 0.914, "step": 38577 }, { "epoch": 0.04, "learning_rate": 4.981752953123143e-05, "loss": 0.9731, "step": 38578 }, { "epoch": 0.04, "learning_rate": 4.981752005825825e-05, "loss": 0.927, "step": 38579 }, { "epoch": 0.04, "learning_rate": 4.981751058504009e-05, "loss": 0.92, "step": 38580 }, { "epoch": 0.04, "learning_rate": 4.981750111157694e-05, "loss": 1.158, "step": 38581 }, { "epoch": 0.04, "learning_rate": 4.98174916378688e-05, "loss": 1.0457, "step": 38582 }, { "epoch": 0.04, "learning_rate": 4.9817482163915674e-05, "loss": 1.0584, "step": 38583 }, { "epoch": 0.04, "learning_rate": 4.981747268971756e-05, "loss": 0.9534, "step": 38584 }, { "epoch": 0.04, "learning_rate": 4.9817463215274454e-05, "loss": 1.0026, "step": 38585 }, { "epoch": 0.04, "learning_rate": 4.9817453740586364e-05, "loss": 1.1582, "step": 38586 }, { "epoch": 0.04, "learning_rate": 4.981744426565329e-05, "loss": 0.9715, "step": 38587 }, { "epoch": 0.04, "learning_rate": 4.9817434790475224e-05, "loss": 1.1092, "step": 38588 }, { "epoch": 0.04, "learning_rate": 4.981742531505217e-05, "loss": 1.1128, "step": 38589 }, { "epoch": 0.04, "learning_rate": 4.981741583938413e-05, "loss": 1.1797, "step": 38590 }, { "epoch": 0.04, "learning_rate": 4.9817406363471106e-05, "loss": 0.8735, "step": 38591 }, { "epoch": 0.04, "learning_rate": 4.9817396887313095e-05, "loss": 0.9589, "step": 38592 }, { "epoch": 0.04, "learning_rate": 4.9817387410910095e-05, "loss": 0.9594, "step": 38593 }, { "epoch": 0.04, "learning_rate": 4.981737793426211e-05, "loss": 0.7385, "step": 38594 }, { "epoch": 0.04, "learning_rate": 4.981736845736913e-05, "loss": 0.7655, "step": 38595 }, { "epoch": 0.04, "learning_rate": 4.981735898023117e-05, "loss": 0.952, "step": 38596 }, { "epoch": 0.04, "learning_rate": 4.9817349502848227e-05, "loss": 0.8738, "step": 38597 }, { "epoch": 0.04, "learning_rate": 4.9817340025220295e-05, "loss": 0.7058, "step": 38598 }, { "epoch": 0.04, "learning_rate": 4.9817330547347375e-05, "loss": 1.1552, "step": 38599 }, { "epoch": 0.04, "learning_rate": 4.9817321069229465e-05, "loss": 0.5237, "step": 38600 }, { "epoch": 0.04, "learning_rate": 4.981731159086658e-05, "loss": 0.7605, "step": 38601 }, { "epoch": 0.04, "learning_rate": 4.981730211225869e-05, "loss": 1.1657, "step": 38602 }, { "epoch": 0.04, "learning_rate": 4.981729263340583e-05, "loss": 1.2891, "step": 38603 }, { "epoch": 0.04, "learning_rate": 4.981728315430798e-05, "loss": 1.045, "step": 38604 }, { "epoch": 0.04, "learning_rate": 4.981727367496515e-05, "loss": 1.187, "step": 38605 }, { "epoch": 0.04, "learning_rate": 4.981726419537733e-05, "loss": 0.927, "step": 38606 }, { "epoch": 0.04, "learning_rate": 4.981725471554452e-05, "loss": 0.8706, "step": 38607 }, { "epoch": 0.04, "learning_rate": 4.981724523546673e-05, "loss": 1.1098, "step": 38608 }, { "epoch": 0.04, "learning_rate": 4.981723575514395e-05, "loss": 1.4504, "step": 38609 }, { "epoch": 0.04, "learning_rate": 4.981722627457619e-05, "loss": 1.2554, "step": 38610 }, { "epoch": 0.04, "learning_rate": 4.9817216793763443e-05, "loss": 1.1427, "step": 38611 }, { "epoch": 0.04, "learning_rate": 4.9817207312705706e-05, "loss": 0.7877, "step": 38612 }, { "epoch": 0.04, "learning_rate": 4.981719783140299e-05, "loss": 0.8211, "step": 38613 }, { "epoch": 0.04, "learning_rate": 4.981718834985528e-05, "loss": 1.1989, "step": 38614 }, { "epoch": 0.04, "learning_rate": 4.9817178868062595e-05, "loss": 1.1931, "step": 38615 }, { "epoch": 0.04, "learning_rate": 4.981716938602493e-05, "loss": 1.3992, "step": 38616 }, { "epoch": 0.04, "learning_rate": 4.9817159903742266e-05, "loss": 1.0418, "step": 38617 }, { "epoch": 0.04, "learning_rate": 4.981715042121463e-05, "loss": 1.0445, "step": 38618 }, { "epoch": 0.04, "learning_rate": 4.9817140938441996e-05, "loss": 1.2045, "step": 38619 }, { "epoch": 0.04, "learning_rate": 4.981713145542439e-05, "loss": 1.4177, "step": 38620 }, { "epoch": 0.04, "learning_rate": 4.9817121972161796e-05, "loss": 1.0583, "step": 38621 }, { "epoch": 0.04, "learning_rate": 4.9817112488654216e-05, "loss": 1.0966, "step": 38622 }, { "epoch": 0.04, "learning_rate": 4.9817103004901654e-05, "loss": 0.8859, "step": 38623 }, { "epoch": 0.04, "learning_rate": 4.98170935209041e-05, "loss": 0.3799, "step": 38624 }, { "epoch": 0.04, "learning_rate": 4.9817084036661575e-05, "loss": 0.5734, "step": 38625 }, { "epoch": 0.04, "learning_rate": 4.981707455217406e-05, "loss": 0.9611, "step": 38626 }, { "epoch": 0.04, "learning_rate": 4.981706506744156e-05, "loss": 0.8345, "step": 38627 }, { "epoch": 0.04, "learning_rate": 4.981705558246408e-05, "loss": 0.8553, "step": 38628 }, { "epoch": 0.04, "learning_rate": 4.981704609724162e-05, "loss": 0.9553, "step": 38629 }, { "epoch": 0.04, "learning_rate": 4.981703661177416e-05, "loss": 0.8642, "step": 38630 }, { "epoch": 0.04, "learning_rate": 4.981702712606173e-05, "loss": 1.0516, "step": 38631 }, { "epoch": 0.04, "learning_rate": 4.981701764010431e-05, "loss": 1.0887, "step": 38632 }, { "epoch": 0.04, "learning_rate": 4.981700815390192e-05, "loss": 0.9784, "step": 38633 }, { "epoch": 0.04, "learning_rate": 4.981699866745453e-05, "loss": 0.6115, "step": 38634 }, { "epoch": 0.04, "learning_rate": 4.981698918076217e-05, "loss": 1.0661, "step": 38635 }, { "epoch": 0.04, "learning_rate": 4.981697969382482e-05, "loss": 1.2892, "step": 38636 }, { "epoch": 0.04, "learning_rate": 4.981697020664249e-05, "loss": 1.1018, "step": 38637 }, { "epoch": 0.04, "learning_rate": 4.981696071921518e-05, "loss": 1.0108, "step": 38638 }, { "epoch": 0.04, "learning_rate": 4.981695123154289e-05, "loss": 1.091, "step": 38639 }, { "epoch": 0.04, "learning_rate": 4.9816941743625606e-05, "loss": 1.3134, "step": 38640 }, { "epoch": 0.04, "learning_rate": 4.981693225546334e-05, "loss": 1.3431, "step": 38641 }, { "epoch": 0.04, "learning_rate": 4.9816922767056104e-05, "loss": 0.6938, "step": 38642 }, { "epoch": 0.04, "learning_rate": 4.9816913278403876e-05, "loss": 1.5029, "step": 38643 }, { "epoch": 0.04, "learning_rate": 4.9816903789506666e-05, "loss": 1.2726, "step": 38644 }, { "epoch": 0.04, "learning_rate": 4.981689430036448e-05, "loss": 1.2996, "step": 38645 }, { "epoch": 0.04, "learning_rate": 4.981688481097731e-05, "loss": 0.845, "step": 38646 }, { "epoch": 0.04, "learning_rate": 4.9816875321345156e-05, "loss": 0.9617, "step": 38647 }, { "epoch": 0.04, "learning_rate": 4.981686583146802e-05, "loss": 0.9718, "step": 38648 }, { "epoch": 0.04, "learning_rate": 4.981685634134591e-05, "loss": 0.8916, "step": 38649 }, { "epoch": 0.04, "learning_rate": 4.9816846850978806e-05, "loss": 0.9379, "step": 38650 }, { "epoch": 0.04, "learning_rate": 4.9816837360366734e-05, "loss": 1.0016, "step": 38651 }, { "epoch": 0.04, "learning_rate": 4.981682786950967e-05, "loss": 1.144, "step": 38652 }, { "epoch": 0.04, "learning_rate": 4.981681837840763e-05, "loss": 0.9776, "step": 38653 }, { "epoch": 0.04, "learning_rate": 4.981680888706061e-05, "loss": 1.3057, "step": 38654 }, { "epoch": 0.04, "learning_rate": 4.98167993954686e-05, "loss": 1.6263, "step": 38655 }, { "epoch": 0.04, "learning_rate": 4.981678990363161e-05, "loss": 1.2968, "step": 38656 }, { "epoch": 0.04, "learning_rate": 4.9816780411549645e-05, "loss": 0.8111, "step": 38657 }, { "epoch": 0.04, "learning_rate": 4.98167709192227e-05, "loss": 1.0673, "step": 38658 }, { "epoch": 0.04, "learning_rate": 4.9816761426650774e-05, "loss": 0.8924, "step": 38659 }, { "epoch": 0.04, "learning_rate": 4.981675193383387e-05, "loss": 0.7548, "step": 38660 }, { "epoch": 0.04, "learning_rate": 4.981674244077198e-05, "loss": 0.8485, "step": 38661 }, { "epoch": 0.04, "learning_rate": 4.98167329474651e-05, "loss": 0.8163, "step": 38662 }, { "epoch": 0.04, "learning_rate": 4.9816723453913264e-05, "loss": 0.8431, "step": 38663 }, { "epoch": 0.04, "learning_rate": 4.981671396011643e-05, "loss": 0.9805, "step": 38664 }, { "epoch": 0.04, "learning_rate": 4.981670446607461e-05, "loss": 1.0454, "step": 38665 }, { "epoch": 0.04, "learning_rate": 4.9816694971787826e-05, "loss": 1.1272, "step": 38666 }, { "epoch": 0.04, "learning_rate": 4.981668547725606e-05, "loss": 0.9254, "step": 38667 }, { "epoch": 0.04, "learning_rate": 4.981667598247931e-05, "loss": 0.8338, "step": 38668 }, { "epoch": 0.04, "learning_rate": 4.981666648745758e-05, "loss": 0.7451, "step": 38669 }, { "epoch": 0.04, "learning_rate": 4.981665699219087e-05, "loss": 0.8548, "step": 38670 }, { "epoch": 0.04, "learning_rate": 4.9816647496679184e-05, "loss": 0.9996, "step": 38671 }, { "epoch": 0.04, "learning_rate": 4.981663800092251e-05, "loss": 0.9367, "step": 38672 }, { "epoch": 0.04, "learning_rate": 4.981662850492086e-05, "loss": 0.9548, "step": 38673 }, { "epoch": 0.04, "learning_rate": 4.981661900867424e-05, "loss": 1.189, "step": 38674 }, { "epoch": 0.04, "learning_rate": 4.9816609512182634e-05, "loss": 0.9223, "step": 38675 }, { "epoch": 0.04, "learning_rate": 4.9816600015446046e-05, "loss": 1.0735, "step": 38676 }, { "epoch": 0.04, "learning_rate": 4.981659051846448e-05, "loss": 1.1271, "step": 38677 }, { "epoch": 0.04, "learning_rate": 4.981658102123794e-05, "loss": 0.9198, "step": 38678 }, { "epoch": 0.04, "learning_rate": 4.981657152376642e-05, "loss": 0.8597, "step": 38679 }, { "epoch": 0.04, "learning_rate": 4.981656202604992e-05, "loss": 1.1156, "step": 38680 }, { "epoch": 0.04, "learning_rate": 4.981655252808844e-05, "loss": 0.9388, "step": 38681 }, { "epoch": 0.04, "learning_rate": 4.981654302988198e-05, "loss": 1.2234, "step": 38682 }, { "epoch": 0.04, "learning_rate": 4.981653353143054e-05, "loss": 1.01, "step": 38683 }, { "epoch": 0.04, "learning_rate": 4.981652403273413e-05, "loss": 1.1347, "step": 38684 }, { "epoch": 0.04, "learning_rate": 4.9816514533792735e-05, "loss": 1.2967, "step": 38685 }, { "epoch": 0.04, "learning_rate": 4.9816505034606365e-05, "loss": 0.9802, "step": 38686 }, { "epoch": 0.04, "learning_rate": 4.981649553517501e-05, "loss": 1.0123, "step": 38687 }, { "epoch": 0.04, "learning_rate": 4.981648603549869e-05, "loss": 1.065, "step": 38688 }, { "epoch": 0.04, "learning_rate": 4.981647653557738e-05, "loss": 1.0703, "step": 38689 }, { "epoch": 0.04, "learning_rate": 4.98164670354111e-05, "loss": 1.0233, "step": 38690 }, { "epoch": 0.04, "learning_rate": 4.9816457534999835e-05, "loss": 0.9699, "step": 38691 }, { "epoch": 0.04, "learning_rate": 4.98164480343436e-05, "loss": 0.9528, "step": 38692 }, { "epoch": 0.04, "learning_rate": 4.981643853344239e-05, "loss": 1.1872, "step": 38693 }, { "epoch": 0.04, "learning_rate": 4.98164290322962e-05, "loss": 1.1629, "step": 38694 }, { "epoch": 0.04, "learning_rate": 4.981641953090502e-05, "loss": 0.9527, "step": 38695 }, { "epoch": 0.04, "learning_rate": 4.981641002926888e-05, "loss": 1.1046, "step": 38696 }, { "epoch": 0.04, "learning_rate": 4.981640052738775e-05, "loss": 1.2397, "step": 38697 }, { "epoch": 0.04, "learning_rate": 4.981639102526164e-05, "loss": 0.8335, "step": 38698 }, { "epoch": 0.04, "learning_rate": 4.981638152289056e-05, "loss": 0.9702, "step": 38699 }, { "epoch": 0.04, "learning_rate": 4.9816372020274516e-05, "loss": 1.1826, "step": 38700 }, { "epoch": 0.04, "learning_rate": 4.981636251741349e-05, "loss": 1.0416, "step": 38701 }, { "epoch": 0.04, "learning_rate": 4.981635301430747e-05, "loss": 1.1855, "step": 38702 }, { "epoch": 0.04, "learning_rate": 4.981634351095649e-05, "loss": 1.005, "step": 38703 }, { "epoch": 0.04, "learning_rate": 4.981633400736052e-05, "loss": 1.1964, "step": 38704 }, { "epoch": 0.04, "learning_rate": 4.981632450351959e-05, "loss": 1.2172, "step": 38705 }, { "epoch": 0.04, "learning_rate": 4.981631499943367e-05, "loss": 1.2207, "step": 38706 }, { "epoch": 0.04, "learning_rate": 4.981630549510278e-05, "loss": 0.9199, "step": 38707 }, { "epoch": 0.04, "learning_rate": 4.9816295990526914e-05, "loss": 1.0676, "step": 38708 }, { "epoch": 0.04, "learning_rate": 4.981628648570607e-05, "loss": 1.0762, "step": 38709 }, { "epoch": 0.04, "learning_rate": 4.981627698064025e-05, "loss": 1.2543, "step": 38710 }, { "epoch": 0.04, "learning_rate": 4.9816267475329454e-05, "loss": 1.1425, "step": 38711 }, { "epoch": 0.04, "learning_rate": 4.981625796977368e-05, "loss": 1.0488, "step": 38712 }, { "epoch": 0.04, "learning_rate": 4.981624846397294e-05, "loss": 0.2537, "step": 38713 }, { "epoch": 0.04, "learning_rate": 4.9816238957927215e-05, "loss": 0.8525, "step": 38714 }, { "epoch": 0.04, "learning_rate": 4.981622945163652e-05, "loss": 0.856, "step": 38715 }, { "epoch": 0.04, "learning_rate": 4.981621994510085e-05, "loss": 0.9352, "step": 38716 }, { "epoch": 0.04, "learning_rate": 4.98162104383202e-05, "loss": 1.1838, "step": 38717 }, { "epoch": 0.04, "learning_rate": 4.981620093129458e-05, "loss": 0.8104, "step": 38718 }, { "epoch": 0.04, "learning_rate": 4.981619142402397e-05, "loss": 0.9906, "step": 38719 }, { "epoch": 0.04, "learning_rate": 4.98161819165084e-05, "loss": 1.2507, "step": 38720 }, { "epoch": 0.04, "learning_rate": 4.9816172408747854e-05, "loss": 0.8479, "step": 38721 }, { "epoch": 0.04, "learning_rate": 4.981616290074233e-05, "loss": 1.166, "step": 38722 }, { "epoch": 0.04, "learning_rate": 4.981615339249183e-05, "loss": 0.5848, "step": 38723 }, { "epoch": 0.04, "learning_rate": 4.981614388399636e-05, "loss": 0.9485, "step": 38724 }, { "epoch": 0.04, "learning_rate": 4.981613437525591e-05, "loss": 0.9201, "step": 38725 }, { "epoch": 0.04, "learning_rate": 4.981612486627049e-05, "loss": 0.9666, "step": 38726 }, { "epoch": 0.04, "learning_rate": 4.9816115357040095e-05, "loss": 0.9216, "step": 38727 }, { "epoch": 0.04, "learning_rate": 4.9816105847564734e-05, "loss": 0.9109, "step": 38728 }, { "epoch": 0.04, "learning_rate": 4.9816096337844385e-05, "loss": 0.7993, "step": 38729 }, { "epoch": 0.04, "learning_rate": 4.9816086827879066e-05, "loss": 0.6785, "step": 38730 }, { "epoch": 0.04, "learning_rate": 4.981607731766877e-05, "loss": 0.9905, "step": 38731 }, { "epoch": 0.04, "learning_rate": 4.9816067807213504e-05, "loss": 1.0374, "step": 38732 }, { "epoch": 0.04, "learning_rate": 4.9816058296513266e-05, "loss": 0.9712, "step": 38733 }, { "epoch": 0.04, "learning_rate": 4.981604878556806e-05, "loss": 1.1934, "step": 38734 }, { "epoch": 0.04, "learning_rate": 4.981603927437787e-05, "loss": 1.2141, "step": 38735 }, { "epoch": 0.04, "learning_rate": 4.9816029762942706e-05, "loss": 1.064, "step": 38736 }, { "epoch": 0.04, "learning_rate": 4.981602025126257e-05, "loss": 0.9321, "step": 38737 }, { "epoch": 0.04, "learning_rate": 4.981601073933747e-05, "loss": 1.0492, "step": 38738 }, { "epoch": 0.04, "learning_rate": 4.981600122716739e-05, "loss": 1.1333, "step": 38739 }, { "epoch": 0.04, "learning_rate": 4.981599171475233e-05, "loss": 1.2262, "step": 38740 }, { "epoch": 0.04, "learning_rate": 4.9815982202092306e-05, "loss": 1.0315, "step": 38741 }, { "epoch": 0.04, "learning_rate": 4.981597268918731e-05, "loss": 1.1986, "step": 38742 }, { "epoch": 0.04, "learning_rate": 4.981596317603734e-05, "loss": 1.296, "step": 38743 }, { "epoch": 0.04, "learning_rate": 4.981595366264239e-05, "loss": 1.1183, "step": 38744 }, { "epoch": 0.04, "learning_rate": 4.981594414900247e-05, "loss": 0.7375, "step": 38745 }, { "epoch": 0.04, "learning_rate": 4.981593463511758e-05, "loss": 0.8656, "step": 38746 }, { "epoch": 0.04, "learning_rate": 4.981592512098772e-05, "loss": 1.007, "step": 38747 }, { "epoch": 0.04, "learning_rate": 4.9815915606612884e-05, "loss": 1.0768, "step": 38748 }, { "epoch": 0.04, "learning_rate": 4.981590609199308e-05, "loss": 0.8513, "step": 38749 }, { "epoch": 0.04, "learning_rate": 4.98158965771283e-05, "loss": 0.975, "step": 38750 }, { "epoch": 0.04, "learning_rate": 4.981588706201855e-05, "loss": 0.9673, "step": 38751 }, { "epoch": 0.04, "learning_rate": 4.9815877546663825e-05, "loss": 0.9905, "step": 38752 }, { "epoch": 0.04, "learning_rate": 4.9815868031064125e-05, "loss": 1.0837, "step": 38753 }, { "epoch": 0.04, "learning_rate": 4.981585851521946e-05, "loss": 0.8079, "step": 38754 }, { "epoch": 0.04, "learning_rate": 4.9815848999129825e-05, "loss": 0.2223, "step": 38755 }, { "epoch": 0.04, "learning_rate": 4.981583948279521e-05, "loss": 0.3054, "step": 38756 }, { "epoch": 0.04, "learning_rate": 4.981582996621563e-05, "loss": 0.5691, "step": 38757 }, { "epoch": 0.04, "learning_rate": 4.981582044939108e-05, "loss": 1.0323, "step": 38758 }, { "epoch": 0.04, "learning_rate": 4.981581093232156e-05, "loss": 0.9665, "step": 38759 }, { "epoch": 0.04, "learning_rate": 4.981580141500706e-05, "loss": 0.8187, "step": 38760 }, { "epoch": 0.04, "learning_rate": 4.9815791897447596e-05, "loss": 1.0539, "step": 38761 }, { "epoch": 0.04, "learning_rate": 4.981578237964316e-05, "loss": 0.9874, "step": 38762 }, { "epoch": 0.04, "learning_rate": 4.9815772861593744e-05, "loss": 1.1652, "step": 38763 }, { "epoch": 0.04, "learning_rate": 4.9815763343299374e-05, "loss": 1.4567, "step": 38764 }, { "epoch": 0.04, "learning_rate": 4.9815753824760016e-05, "loss": 1.8322, "step": 38765 }, { "epoch": 0.04, "learning_rate": 4.9815744305975695e-05, "loss": 1.1357, "step": 38766 }, { "epoch": 0.04, "learning_rate": 4.9815734786946406e-05, "loss": 1.0296, "step": 38767 }, { "epoch": 0.04, "learning_rate": 4.981572526767214e-05, "loss": 0.9136, "step": 38768 }, { "epoch": 0.04, "learning_rate": 4.9815715748152915e-05, "loss": 1.0399, "step": 38769 }, { "epoch": 0.04, "learning_rate": 4.98157062283887e-05, "loss": 0.8444, "step": 38770 }, { "epoch": 0.04, "learning_rate": 4.9815696708379536e-05, "loss": 0.9544, "step": 38771 }, { "epoch": 0.04, "learning_rate": 4.981568718812539e-05, "loss": 0.8772, "step": 38772 }, { "epoch": 0.04, "learning_rate": 4.981567766762628e-05, "loss": 0.8287, "step": 38773 }, { "epoch": 0.04, "learning_rate": 4.981566814688219e-05, "loss": 1.0904, "step": 38774 }, { "epoch": 0.04, "learning_rate": 4.981565862589314e-05, "loss": 0.9208, "step": 38775 }, { "epoch": 0.04, "learning_rate": 4.9815649104659125e-05, "loss": 0.6437, "step": 38776 }, { "epoch": 0.04, "learning_rate": 4.981563958318013e-05, "loss": 0.8286, "step": 38777 }, { "epoch": 0.04, "learning_rate": 4.9815630061456164e-05, "loss": 0.9481, "step": 38778 }, { "epoch": 0.04, "learning_rate": 4.981562053948724e-05, "loss": 1.0182, "step": 38779 }, { "epoch": 0.04, "learning_rate": 4.9815611017273334e-05, "loss": 1.1951, "step": 38780 }, { "epoch": 0.04, "learning_rate": 4.981560149481447e-05, "loss": 1.1186, "step": 38781 }, { "epoch": 0.04, "learning_rate": 4.981559197211062e-05, "loss": 1.2194, "step": 38782 }, { "epoch": 0.04, "learning_rate": 4.981558244916182e-05, "loss": 0.5216, "step": 38783 }, { "epoch": 0.04, "learning_rate": 4.981557292596804e-05, "loss": 0.9674, "step": 38784 }, { "epoch": 0.04, "learning_rate": 4.98155634025293e-05, "loss": 0.8322, "step": 38785 }, { "epoch": 0.04, "learning_rate": 4.9815553878845576e-05, "loss": 0.7888, "step": 38786 }, { "epoch": 0.04, "learning_rate": 4.98155443549169e-05, "loss": 0.5348, "step": 38787 }, { "epoch": 0.04, "learning_rate": 4.981553483074324e-05, "loss": 0.4246, "step": 38788 }, { "epoch": 0.04, "learning_rate": 4.9815525306324627e-05, "loss": 0.6317, "step": 38789 }, { "epoch": 0.04, "learning_rate": 4.981551578166104e-05, "loss": 0.9471, "step": 38790 }, { "epoch": 0.04, "learning_rate": 4.981550625675248e-05, "loss": 0.6664, "step": 38791 }, { "epoch": 0.04, "learning_rate": 4.981549673159895e-05, "loss": 1.1274, "step": 38792 }, { "epoch": 0.04, "learning_rate": 4.981548720620046e-05, "loss": 1.117, "step": 38793 }, { "epoch": 0.04, "learning_rate": 4.9815477680557e-05, "loss": 1.0252, "step": 38794 }, { "epoch": 0.04, "learning_rate": 4.9815468154668565e-05, "loss": 0.9435, "step": 38795 }, { "epoch": 0.04, "learning_rate": 4.981545862853517e-05, "loss": 1.0388, "step": 38796 }, { "epoch": 0.04, "learning_rate": 4.9815449102156807e-05, "loss": 1.1585, "step": 38797 }, { "epoch": 0.04, "learning_rate": 4.9815439575533474e-05, "loss": 1.1186, "step": 38798 }, { "epoch": 0.04, "learning_rate": 4.981543004866517e-05, "loss": 1.19, "step": 38799 }, { "epoch": 0.04, "learning_rate": 4.98154205215519e-05, "loss": 0.6933, "step": 38800 }, { "epoch": 0.04, "learning_rate": 4.9815410994193665e-05, "loss": 0.7491, "step": 38801 }, { "epoch": 0.04, "learning_rate": 4.9815401466590464e-05, "loss": 0.7396, "step": 38802 }, { "epoch": 0.04, "learning_rate": 4.9815391938742295e-05, "loss": 0.3466, "step": 38803 }, { "epoch": 0.04, "learning_rate": 4.981538241064916e-05, "loss": 0.3558, "step": 38804 }, { "epoch": 0.04, "learning_rate": 4.9815372882311064e-05, "loss": 0.5738, "step": 38805 }, { "epoch": 0.04, "learning_rate": 4.981536335372798e-05, "loss": 1.2387, "step": 38806 }, { "epoch": 0.04, "learning_rate": 4.9815353824899945e-05, "loss": 1.5099, "step": 38807 }, { "epoch": 0.04, "learning_rate": 4.9815344295826946e-05, "loss": 1.0678, "step": 38808 }, { "epoch": 0.04, "learning_rate": 4.981533476650898e-05, "loss": 1.0844, "step": 38809 }, { "epoch": 0.04, "learning_rate": 4.9815325236946034e-05, "loss": 0.8271, "step": 38810 }, { "epoch": 0.04, "learning_rate": 4.9815315707138136e-05, "loss": 1.0233, "step": 38811 }, { "epoch": 0.04, "learning_rate": 4.981530617708526e-05, "loss": 0.8519, "step": 38812 }, { "epoch": 0.04, "learning_rate": 4.9815296646787426e-05, "loss": 0.8983, "step": 38813 }, { "epoch": 0.04, "learning_rate": 4.981528711624462e-05, "loss": 0.98, "step": 38814 }, { "epoch": 0.04, "learning_rate": 4.9815277585456855e-05, "loss": 0.8575, "step": 38815 }, { "epoch": 0.04, "learning_rate": 4.9815268054424126e-05, "loss": 0.9987, "step": 38816 }, { "epoch": 0.04, "learning_rate": 4.981525852314642e-05, "loss": 0.9513, "step": 38817 }, { "epoch": 0.04, "learning_rate": 4.981524899162375e-05, "loss": 0.6333, "step": 38818 }, { "epoch": 0.04, "learning_rate": 4.981523945985612e-05, "loss": 0.7172, "step": 38819 }, { "epoch": 0.04, "learning_rate": 4.9815229927843526e-05, "loss": 0.9969, "step": 38820 }, { "epoch": 0.04, "learning_rate": 4.981522039558597e-05, "loss": 0.4864, "step": 38821 }, { "epoch": 0.04, "learning_rate": 4.981521086308344e-05, "loss": 0.7031, "step": 38822 }, { "epoch": 0.04, "learning_rate": 4.9815201330335944e-05, "loss": 0.649, "step": 38823 }, { "epoch": 0.04, "learning_rate": 4.9815191797343486e-05, "loss": 1.1624, "step": 38824 }, { "epoch": 0.04, "learning_rate": 4.981518226410606e-05, "loss": 0.9243, "step": 38825 }, { "epoch": 0.04, "learning_rate": 4.981517273062367e-05, "loss": 1.0494, "step": 38826 }, { "epoch": 0.04, "learning_rate": 4.981516319689632e-05, "loss": 1.0783, "step": 38827 }, { "epoch": 0.04, "learning_rate": 4.9815153662923995e-05, "loss": 1.1445, "step": 38828 }, { "epoch": 0.04, "learning_rate": 4.9815144128706714e-05, "loss": 0.8494, "step": 38829 }, { "epoch": 0.04, "learning_rate": 4.9815134594244465e-05, "loss": 0.9301, "step": 38830 }, { "epoch": 0.04, "learning_rate": 4.981512505953725e-05, "loss": 0.3366, "step": 38831 }, { "epoch": 0.04, "learning_rate": 4.981511552458508e-05, "loss": 0.4267, "step": 38832 }, { "epoch": 0.04, "learning_rate": 4.981510598938793e-05, "loss": 0.2562, "step": 38833 }, { "epoch": 0.04, "learning_rate": 4.981509645394583e-05, "loss": 0.3249, "step": 38834 }, { "epoch": 0.04, "learning_rate": 4.981508691825876e-05, "loss": 0.6466, "step": 38835 }, { "epoch": 0.04, "learning_rate": 4.9815077382326726e-05, "loss": 1.0803, "step": 38836 }, { "epoch": 0.04, "learning_rate": 4.981506784614973e-05, "loss": 1.2746, "step": 38837 }, { "epoch": 0.04, "learning_rate": 4.981505830972777e-05, "loss": 1.1826, "step": 38838 }, { "epoch": 0.04, "learning_rate": 4.981504877306084e-05, "loss": 0.7479, "step": 38839 }, { "epoch": 0.04, "learning_rate": 4.9815039236148954e-05, "loss": 0.8456, "step": 38840 }, { "epoch": 0.04, "learning_rate": 4.98150296989921e-05, "loss": 1.1616, "step": 38841 }, { "epoch": 0.04, "learning_rate": 4.981502016159028e-05, "loss": 1.0679, "step": 38842 }, { "epoch": 0.04, "learning_rate": 4.98150106239435e-05, "loss": 0.9672, "step": 38843 }, { "epoch": 0.04, "learning_rate": 4.981500108605176e-05, "loss": 0.7996, "step": 38844 }, { "epoch": 0.04, "learning_rate": 4.981499154791505e-05, "loss": 1.1838, "step": 38845 }, { "epoch": 0.04, "learning_rate": 4.9814982009533384e-05, "loss": 1.0481, "step": 38846 }, { "epoch": 0.04, "learning_rate": 4.9814972470906754e-05, "loss": 1.0289, "step": 38847 }, { "epoch": 0.04, "learning_rate": 4.9814962932035156e-05, "loss": 0.9079, "step": 38848 }, { "epoch": 0.04, "learning_rate": 4.9814953392918596e-05, "loss": 1.0222, "step": 38849 }, { "epoch": 0.04, "learning_rate": 4.9814943853557074e-05, "loss": 1.5282, "step": 38850 }, { "epoch": 0.04, "learning_rate": 4.981493431395059e-05, "loss": 0.7955, "step": 38851 }, { "epoch": 0.04, "learning_rate": 4.981492477409915e-05, "loss": 0.2277, "step": 38852 }, { "epoch": 0.04, "learning_rate": 4.9814915234002736e-05, "loss": 0.2353, "step": 38853 }, { "epoch": 0.04, "learning_rate": 4.981490569366136e-05, "loss": 0.2032, "step": 38854 }, { "epoch": 0.04, "learning_rate": 4.981489615307503e-05, "loss": 0.2942, "step": 38855 }, { "epoch": 0.04, "learning_rate": 4.9814886612243734e-05, "loss": 0.1924, "step": 38856 }, { "epoch": 0.04, "learning_rate": 4.981487707116748e-05, "loss": 0.7582, "step": 38857 }, { "epoch": 0.04, "learning_rate": 4.9814867529846254e-05, "loss": 1.0894, "step": 38858 }, { "epoch": 0.04, "learning_rate": 4.9814857988280074e-05, "loss": 1.096, "step": 38859 }, { "epoch": 0.04, "learning_rate": 4.981484844646893e-05, "loss": 0.7961, "step": 38860 }, { "epoch": 0.04, "learning_rate": 4.981483890441282e-05, "loss": 0.815, "step": 38861 }, { "epoch": 0.04, "learning_rate": 4.981482936211176e-05, "loss": 1.0569, "step": 38862 }, { "epoch": 0.04, "learning_rate": 4.981481981956573e-05, "loss": 1.0635, "step": 38863 }, { "epoch": 0.04, "learning_rate": 4.9814810276774734e-05, "loss": 1.0919, "step": 38864 }, { "epoch": 0.04, "learning_rate": 4.981480073373879e-05, "loss": 1.4468, "step": 38865 }, { "epoch": 0.04, "learning_rate": 4.981479119045788e-05, "loss": 1.2861, "step": 38866 }, { "epoch": 0.04, "learning_rate": 4.9814781646932e-05, "loss": 0.908, "step": 38867 }, { "epoch": 0.04, "learning_rate": 4.9814772103161165e-05, "loss": 1.0874, "step": 38868 }, { "epoch": 0.04, "learning_rate": 4.9814762559145373e-05, "loss": 1.0043, "step": 38869 }, { "epoch": 0.04, "learning_rate": 4.981475301488462e-05, "loss": 1.2009, "step": 38870 }, { "epoch": 0.04, "learning_rate": 4.98147434703789e-05, "loss": 1.1681, "step": 38871 }, { "epoch": 0.04, "learning_rate": 4.981473392562822e-05, "loss": 1.0392, "step": 38872 }, { "epoch": 0.04, "learning_rate": 4.981472438063258e-05, "loss": 1.1695, "step": 38873 }, { "epoch": 0.04, "learning_rate": 4.9814714835391985e-05, "loss": 0.9792, "step": 38874 }, { "epoch": 0.04, "learning_rate": 4.981470528990642e-05, "loss": 1.0826, "step": 38875 }, { "epoch": 0.04, "learning_rate": 4.9814695744175896e-05, "loss": 1.1335, "step": 38876 }, { "epoch": 0.04, "learning_rate": 4.981468619820042e-05, "loss": 0.7791, "step": 38877 }, { "epoch": 0.04, "learning_rate": 4.981467665197998e-05, "loss": 0.8762, "step": 38878 }, { "epoch": 0.04, "learning_rate": 4.981466710551458e-05, "loss": 0.4704, "step": 38879 }, { "epoch": 0.04, "learning_rate": 4.981465755880422e-05, "loss": 0.5178, "step": 38880 }, { "epoch": 0.04, "learning_rate": 4.9814648011848894e-05, "loss": 0.4105, "step": 38881 }, { "epoch": 0.04, "learning_rate": 4.9814638464648624e-05, "loss": 0.2647, "step": 38882 }, { "epoch": 0.04, "learning_rate": 4.981462891720338e-05, "loss": 0.3242, "step": 38883 }, { "epoch": 0.04, "learning_rate": 4.9814619369513184e-05, "loss": 0.4495, "step": 38884 }, { "epoch": 0.04, "learning_rate": 4.981460982157802e-05, "loss": 0.2822, "step": 38885 }, { "epoch": 0.04, "learning_rate": 4.9814600273397904e-05, "loss": 0.2092, "step": 38886 }, { "epoch": 0.04, "learning_rate": 4.9814590724972825e-05, "loss": 0.2422, "step": 38887 }, { "epoch": 0.04, "learning_rate": 4.981458117630279e-05, "loss": 0.6219, "step": 38888 }, { "epoch": 0.04, "learning_rate": 4.981457162738779e-05, "loss": 0.904, "step": 38889 }, { "epoch": 0.04, "learning_rate": 4.981456207822784e-05, "loss": 0.6572, "step": 38890 }, { "epoch": 0.04, "learning_rate": 4.981455252882293e-05, "loss": 1.1985, "step": 38891 }, { "epoch": 0.04, "learning_rate": 4.981454297917305e-05, "loss": 0.9238, "step": 38892 }, { "epoch": 0.04, "learning_rate": 4.981453342927822e-05, "loss": 1.5682, "step": 38893 }, { "epoch": 0.04, "learning_rate": 4.9814523879138435e-05, "loss": 1.2052, "step": 38894 }, { "epoch": 0.04, "learning_rate": 4.981451432875369e-05, "loss": 1.0076, "step": 38895 }, { "epoch": 0.04, "learning_rate": 4.9814504778123984e-05, "loss": 1.1768, "step": 38896 }, { "epoch": 0.04, "learning_rate": 4.981449522724931e-05, "loss": 0.695, "step": 38897 }, { "epoch": 0.04, "learning_rate": 4.98144856761297e-05, "loss": 0.9186, "step": 38898 }, { "epoch": 0.04, "learning_rate": 4.981447612476511e-05, "loss": 0.9091, "step": 38899 }, { "epoch": 0.04, "learning_rate": 4.981446657315557e-05, "loss": 0.8752, "step": 38900 }, { "epoch": 0.04, "learning_rate": 4.9814457021301074e-05, "loss": 1.1264, "step": 38901 }, { "epoch": 0.04, "learning_rate": 4.981444746920162e-05, "loss": 1.1097, "step": 38902 }, { "epoch": 0.04, "learning_rate": 4.9814437916857203e-05, "loss": 1.0984, "step": 38903 }, { "epoch": 0.04, "learning_rate": 4.981442836426784e-05, "loss": 1.0483, "step": 38904 }, { "epoch": 0.04, "learning_rate": 4.9814418811433506e-05, "loss": 1.6774, "step": 38905 }, { "epoch": 0.04, "learning_rate": 4.9814409258354224e-05, "loss": 1.209, "step": 38906 }, { "epoch": 0.04, "learning_rate": 4.981439970502998e-05, "loss": 1.1676, "step": 38907 }, { "epoch": 0.04, "learning_rate": 4.9814390151460775e-05, "loss": 0.9629, "step": 38908 }, { "epoch": 0.04, "learning_rate": 4.981438059764662e-05, "loss": 1.0055, "step": 38909 }, { "epoch": 0.04, "learning_rate": 4.9814371043587505e-05, "loss": 1.2829, "step": 38910 }, { "epoch": 0.04, "learning_rate": 4.9814361489283434e-05, "loss": 1.4417, "step": 38911 }, { "epoch": 0.04, "learning_rate": 4.981435193473441e-05, "loss": 1.1362, "step": 38912 }, { "epoch": 0.04, "learning_rate": 4.981434237994042e-05, "loss": 1.2417, "step": 38913 }, { "epoch": 0.04, "learning_rate": 4.9814332824901477e-05, "loss": 1.299, "step": 38914 }, { "epoch": 0.04, "learning_rate": 4.981432326961758e-05, "loss": 1.0328, "step": 38915 }, { "epoch": 0.04, "learning_rate": 4.9814313714088725e-05, "loss": 0.8897, "step": 38916 }, { "epoch": 0.04, "learning_rate": 4.9814304158314916e-05, "loss": 1.1575, "step": 38917 }, { "epoch": 0.04, "learning_rate": 4.9814294602296145e-05, "loss": 0.8244, "step": 38918 }, { "epoch": 0.04, "learning_rate": 4.981428504603242e-05, "loss": 0.8519, "step": 38919 }, { "epoch": 0.04, "learning_rate": 4.9814275489523745e-05, "loss": 0.9641, "step": 38920 }, { "epoch": 0.04, "learning_rate": 4.98142659327701e-05, "loss": 1.0931, "step": 38921 }, { "epoch": 0.04, "learning_rate": 4.981425637577151e-05, "loss": 1.0849, "step": 38922 }, { "epoch": 0.04, "learning_rate": 4.9814246818527965e-05, "loss": 1.1429, "step": 38923 }, { "epoch": 0.04, "learning_rate": 4.981423726103946e-05, "loss": 1.0569, "step": 38924 }, { "epoch": 0.04, "learning_rate": 4.9814227703306e-05, "loss": 0.9475, "step": 38925 }, { "epoch": 0.04, "learning_rate": 4.981421814532758e-05, "loss": 1.0305, "step": 38926 }, { "epoch": 0.04, "learning_rate": 4.9814208587104214e-05, "loss": 1.2725, "step": 38927 }, { "epoch": 0.04, "learning_rate": 4.9814199028635886e-05, "loss": 0.9729, "step": 38928 }, { "epoch": 0.04, "learning_rate": 4.98141894699226e-05, "loss": 0.6252, "step": 38929 }, { "epoch": 0.04, "learning_rate": 4.981417991096436e-05, "loss": 0.7955, "step": 38930 }, { "epoch": 0.04, "learning_rate": 4.9814170351761175e-05, "loss": 0.7291, "step": 38931 }, { "epoch": 0.04, "learning_rate": 4.9814160792313025e-05, "loss": 0.9026, "step": 38932 }, { "epoch": 0.04, "learning_rate": 4.981415123261993e-05, "loss": 0.9469, "step": 38933 }, { "epoch": 0.04, "learning_rate": 4.981414167268187e-05, "loss": 0.9151, "step": 38934 }, { "epoch": 0.04, "learning_rate": 4.981413211249886e-05, "loss": 0.9673, "step": 38935 }, { "epoch": 0.04, "learning_rate": 4.981412255207089e-05, "loss": 1.3449, "step": 38936 }, { "epoch": 0.04, "learning_rate": 4.981411299139797e-05, "loss": 1.0505, "step": 38937 }, { "epoch": 0.04, "learning_rate": 4.98141034304801e-05, "loss": 0.5158, "step": 38938 }, { "epoch": 0.04, "learning_rate": 4.981409386931727e-05, "loss": 0.9788, "step": 38939 }, { "epoch": 0.04, "learning_rate": 4.9814084307909484e-05, "loss": 0.9181, "step": 38940 }, { "epoch": 0.04, "learning_rate": 4.9814074746256745e-05, "loss": 1.0075, "step": 38941 }, { "epoch": 0.04, "learning_rate": 4.981406518435905e-05, "loss": 1.203, "step": 38942 }, { "epoch": 0.04, "learning_rate": 4.98140556222164e-05, "loss": 0.9349, "step": 38943 }, { "epoch": 0.04, "learning_rate": 4.9814046059828804e-05, "loss": 0.9995, "step": 38944 }, { "epoch": 0.04, "learning_rate": 4.981403649719625e-05, "loss": 1.0542, "step": 38945 }, { "epoch": 0.04, "learning_rate": 4.981402693431874e-05, "loss": 1.1886, "step": 38946 }, { "epoch": 0.04, "learning_rate": 4.981401737119628e-05, "loss": 1.2502, "step": 38947 }, { "epoch": 0.04, "learning_rate": 4.981400780782887e-05, "loss": 0.9441, "step": 38948 }, { "epoch": 0.04, "learning_rate": 4.9813998244216494e-05, "loss": 0.7032, "step": 38949 }, { "epoch": 0.04, "learning_rate": 4.981398868035918e-05, "loss": 1.163, "step": 38950 }, { "epoch": 0.04, "learning_rate": 4.98139791162569e-05, "loss": 0.833, "step": 38951 }, { "epoch": 0.04, "learning_rate": 4.981396955190967e-05, "loss": 1.1118, "step": 38952 }, { "epoch": 0.04, "learning_rate": 4.9813959987317496e-05, "loss": 0.901, "step": 38953 }, { "epoch": 0.04, "learning_rate": 4.981395042248036e-05, "loss": 1.0077, "step": 38954 }, { "epoch": 0.04, "learning_rate": 4.981394085739827e-05, "loss": 1.5748, "step": 38955 }, { "epoch": 0.04, "learning_rate": 4.981393129207123e-05, "loss": 1.0559, "step": 38956 }, { "epoch": 0.04, "learning_rate": 4.981392172649924e-05, "loss": 0.9457, "step": 38957 }, { "epoch": 0.04, "learning_rate": 4.98139121606823e-05, "loss": 1.2071, "step": 38958 }, { "epoch": 0.04, "learning_rate": 4.98139025946204e-05, "loss": 0.7268, "step": 38959 }, { "epoch": 0.04, "learning_rate": 4.981389302831355e-05, "loss": 1.1805, "step": 38960 }, { "epoch": 0.04, "learning_rate": 4.981388346176175e-05, "loss": 0.7402, "step": 38961 }, { "epoch": 0.04, "learning_rate": 4.981387389496499e-05, "loss": 1.0082, "step": 38962 }, { "epoch": 0.04, "learning_rate": 4.9813864327923285e-05, "loss": 0.9424, "step": 38963 }, { "epoch": 0.04, "learning_rate": 4.9813854760636625e-05, "loss": 0.8129, "step": 38964 }, { "epoch": 0.04, "learning_rate": 4.981384519310501e-05, "loss": 0.9177, "step": 38965 }, { "epoch": 0.04, "learning_rate": 4.981383562532845e-05, "loss": 0.6107, "step": 38966 }, { "epoch": 0.04, "learning_rate": 4.981382605730694e-05, "loss": 1.0026, "step": 38967 }, { "epoch": 0.04, "learning_rate": 4.9813816489040474e-05, "loss": 0.8972, "step": 38968 }, { "epoch": 0.04, "learning_rate": 4.981380692052906e-05, "loss": 1.3241, "step": 38969 }, { "epoch": 0.04, "learning_rate": 4.981379735177269e-05, "loss": 1.0281, "step": 38970 }, { "epoch": 0.04, "learning_rate": 4.981378778277137e-05, "loss": 0.7939, "step": 38971 }, { "epoch": 0.04, "learning_rate": 4.98137782135251e-05, "loss": 0.3605, "step": 38972 }, { "epoch": 0.04, "learning_rate": 4.981376864403388e-05, "loss": 0.2113, "step": 38973 }, { "epoch": 0.04, "learning_rate": 4.981375907429771e-05, "loss": 0.1365, "step": 38974 }, { "epoch": 0.04, "learning_rate": 4.9813749504316574e-05, "loss": 0.1217, "step": 38975 }, { "epoch": 0.04, "learning_rate": 4.9813739934090506e-05, "loss": 0.174, "step": 38976 }, { "epoch": 0.04, "learning_rate": 4.9813730363619476e-05, "loss": 0.2039, "step": 38977 }, { "epoch": 0.04, "learning_rate": 4.98137207929035e-05, "loss": 0.1773, "step": 38978 }, { "epoch": 0.04, "learning_rate": 4.981371122194257e-05, "loss": 0.6409, "step": 38979 }, { "epoch": 0.04, "learning_rate": 4.98137016507367e-05, "loss": 1.1435, "step": 38980 }, { "epoch": 0.04, "learning_rate": 4.981369207928587e-05, "loss": 1.1752, "step": 38981 }, { "epoch": 0.04, "learning_rate": 4.981368250759009e-05, "loss": 0.7172, "step": 38982 }, { "epoch": 0.04, "learning_rate": 4.9813672935649355e-05, "loss": 1.3992, "step": 38983 }, { "epoch": 0.04, "learning_rate": 4.981366336346368e-05, "loss": 0.9269, "step": 38984 }, { "epoch": 0.04, "learning_rate": 4.9813653791033057e-05, "loss": 1.1269, "step": 38985 }, { "epoch": 0.04, "learning_rate": 4.981364421835748e-05, "loss": 0.9756, "step": 38986 }, { "epoch": 0.04, "learning_rate": 4.9813634645436944e-05, "loss": 1.2781, "step": 38987 }, { "epoch": 0.04, "learning_rate": 4.981362507227146e-05, "loss": 1.8932, "step": 38988 }, { "epoch": 0.04, "learning_rate": 4.9813615498861045e-05, "loss": 0.8301, "step": 38989 }, { "epoch": 0.04, "learning_rate": 4.9813605925205666e-05, "loss": 0.9496, "step": 38990 }, { "epoch": 0.04, "learning_rate": 4.981359635130534e-05, "loss": 0.9085, "step": 38991 }, { "epoch": 0.04, "learning_rate": 4.9813586777160056e-05, "loss": 0.9137, "step": 38992 }, { "epoch": 0.04, "learning_rate": 4.981357720276983e-05, "loss": 1.2055, "step": 38993 }, { "epoch": 0.04, "learning_rate": 4.981356762813466e-05, "loss": 0.9071, "step": 38994 }, { "epoch": 0.04, "learning_rate": 4.981355805325453e-05, "loss": 1.0135, "step": 38995 }, { "epoch": 0.04, "learning_rate": 4.981354847812946e-05, "loss": 1.2296, "step": 38996 }, { "epoch": 0.04, "learning_rate": 4.981353890275944e-05, "loss": 1.3863, "step": 38997 }, { "epoch": 0.04, "learning_rate": 4.981352932714447e-05, "loss": 1.1605, "step": 38998 }, { "epoch": 0.04, "learning_rate": 4.981351975128455e-05, "loss": 1.2071, "step": 38999 }, { "epoch": 0.04, "learning_rate": 4.9813510175179675e-05, "loss": 1.1673, "step": 39000 }, { "epoch": 0.04, "eval_loss": 1.0770384073257446, "eval_runtime": 127.5137, "eval_samples_per_second": 10.862, "eval_steps_per_second": 5.435, "step": 39000 }, { "epoch": 0.04, "learning_rate": 4.9813500598829864e-05, "loss": 0.9733, "step": 39001 }, { "epoch": 0.04, "learning_rate": 4.9813491022235105e-05, "loss": 1.0111, "step": 39002 }, { "epoch": 0.04, "learning_rate": 4.981348144539539e-05, "loss": 1.0717, "step": 39003 }, { "epoch": 0.04, "learning_rate": 4.981347186831072e-05, "loss": 0.9463, "step": 39004 }, { "epoch": 0.04, "learning_rate": 4.9813462290981116e-05, "loss": 0.5948, "step": 39005 }, { "epoch": 0.04, "learning_rate": 4.981345271340656e-05, "loss": 0.9362, "step": 39006 }, { "epoch": 0.04, "learning_rate": 4.9813443135587055e-05, "loss": 1.0176, "step": 39007 }, { "epoch": 0.04, "learning_rate": 4.9813433557522605e-05, "loss": 1.2668, "step": 39008 }, { "epoch": 0.04, "learning_rate": 4.98134239792132e-05, "loss": 1.0683, "step": 39009 }, { "epoch": 0.04, "learning_rate": 4.9813414400658854e-05, "loss": 1.0627, "step": 39010 }, { "epoch": 0.04, "learning_rate": 4.981340482185955e-05, "loss": 0.8256, "step": 39011 }, { "epoch": 0.04, "learning_rate": 4.981339524281531e-05, "loss": 1.0126, "step": 39012 }, { "epoch": 0.04, "learning_rate": 4.981338566352612e-05, "loss": 1.3962, "step": 39013 }, { "epoch": 0.04, "learning_rate": 4.981337608399198e-05, "loss": 1.7819, "step": 39014 }, { "epoch": 0.04, "learning_rate": 4.9813366504212896e-05, "loss": 1.0444, "step": 39015 }, { "epoch": 0.04, "learning_rate": 4.981335692418886e-05, "loss": 0.9261, "step": 39016 }, { "epoch": 0.04, "learning_rate": 4.981334734391988e-05, "loss": 0.5697, "step": 39017 }, { "epoch": 0.04, "learning_rate": 4.9813337763405954e-05, "loss": 0.9559, "step": 39018 }, { "epoch": 0.04, "learning_rate": 4.981332818264708e-05, "loss": 0.7245, "step": 39019 }, { "epoch": 0.04, "learning_rate": 4.981331860164326e-05, "loss": 0.6773, "step": 39020 }, { "epoch": 0.04, "learning_rate": 4.981330902039449e-05, "loss": 0.6686, "step": 39021 }, { "epoch": 0.04, "learning_rate": 4.9813299438900784e-05, "loss": 0.9485, "step": 39022 }, { "epoch": 0.04, "learning_rate": 4.981328985716212e-05, "loss": 0.9934, "step": 39023 }, { "epoch": 0.04, "learning_rate": 4.981328027517851e-05, "loss": 0.7221, "step": 39024 }, { "epoch": 0.04, "learning_rate": 4.981327069294997e-05, "loss": 0.483, "step": 39025 }, { "epoch": 0.04, "learning_rate": 4.981326111047646e-05, "loss": 0.7862, "step": 39026 }, { "epoch": 0.04, "learning_rate": 4.981325152775802e-05, "loss": 1.2067, "step": 39027 }, { "epoch": 0.04, "learning_rate": 4.981324194479463e-05, "loss": 0.9525, "step": 39028 }, { "epoch": 0.04, "learning_rate": 4.981323236158629e-05, "loss": 0.9384, "step": 39029 }, { "epoch": 0.04, "learning_rate": 4.9813222778133e-05, "loss": 1.0354, "step": 39030 }, { "epoch": 0.04, "learning_rate": 4.981321319443478e-05, "loss": 1.0025, "step": 39031 }, { "epoch": 0.04, "learning_rate": 4.981320361049161e-05, "loss": 1.1616, "step": 39032 }, { "epoch": 0.04, "learning_rate": 4.9813194026303486e-05, "loss": 0.8577, "step": 39033 }, { "epoch": 0.04, "learning_rate": 4.9813184441870426e-05, "loss": 0.5259, "step": 39034 }, { "epoch": 0.04, "learning_rate": 4.981317485719241e-05, "loss": 0.111, "step": 39035 }, { "epoch": 0.04, "learning_rate": 4.9813165272269454e-05, "loss": 0.1264, "step": 39036 }, { "epoch": 0.04, "learning_rate": 4.9813155687101556e-05, "loss": 0.1618, "step": 39037 }, { "epoch": 0.04, "learning_rate": 4.981314610168871e-05, "loss": 0.8466, "step": 39038 }, { "epoch": 0.04, "learning_rate": 4.981313651603092e-05, "loss": 0.9936, "step": 39039 }, { "epoch": 0.04, "learning_rate": 4.981312693012819e-05, "loss": 0.849, "step": 39040 }, { "epoch": 0.04, "learning_rate": 4.9813117343980506e-05, "loss": 1.0114, "step": 39041 }, { "epoch": 0.04, "learning_rate": 4.9813107757587886e-05, "loss": 0.9811, "step": 39042 }, { "epoch": 0.04, "learning_rate": 4.981309817095032e-05, "loss": 0.8635, "step": 39043 }, { "epoch": 0.04, "learning_rate": 4.98130885840678e-05, "loss": 0.8524, "step": 39044 }, { "epoch": 0.04, "learning_rate": 4.981307899694034e-05, "loss": 0.9694, "step": 39045 }, { "epoch": 0.04, "learning_rate": 4.981306940956795e-05, "loss": 0.9663, "step": 39046 }, { "epoch": 0.04, "learning_rate": 4.98130598219506e-05, "loss": 0.8745, "step": 39047 }, { "epoch": 0.04, "learning_rate": 4.981305023408831e-05, "loss": 1.0598, "step": 39048 }, { "epoch": 0.04, "learning_rate": 4.981304064598108e-05, "loss": 1.1675, "step": 39049 }, { "epoch": 0.04, "learning_rate": 4.98130310576289e-05, "loss": 1.1217, "step": 39050 }, { "epoch": 0.04, "learning_rate": 4.981302146903177e-05, "loss": 1.0164, "step": 39051 }, { "epoch": 0.04, "learning_rate": 4.981301188018971e-05, "loss": 1.0031, "step": 39052 }, { "epoch": 0.04, "learning_rate": 4.98130022911027e-05, "loss": 0.7732, "step": 39053 }, { "epoch": 0.04, "learning_rate": 4.981299270177075e-05, "loss": 0.308, "step": 39054 }, { "epoch": 0.04, "learning_rate": 4.9812983112193857e-05, "loss": 0.9133, "step": 39055 }, { "epoch": 0.04, "learning_rate": 4.9812973522372014e-05, "loss": 0.9497, "step": 39056 }, { "epoch": 0.04, "learning_rate": 4.9812963932305236e-05, "loss": 1.3698, "step": 39057 }, { "epoch": 0.04, "learning_rate": 4.981295434199351e-05, "loss": 1.0613, "step": 39058 }, { "epoch": 0.04, "learning_rate": 4.981294475143684e-05, "loss": 1.2332, "step": 39059 }, { "epoch": 0.04, "learning_rate": 4.9812935160635226e-05, "loss": 1.185, "step": 39060 }, { "epoch": 0.04, "learning_rate": 4.9812925569588675e-05, "loss": 1.0745, "step": 39061 }, { "epoch": 0.04, "learning_rate": 4.9812915978297175e-05, "loss": 0.9458, "step": 39062 }, { "epoch": 0.04, "learning_rate": 4.981290638676074e-05, "loss": 0.7959, "step": 39063 }, { "epoch": 0.04, "learning_rate": 4.981289679497935e-05, "loss": 1.0579, "step": 39064 }, { "epoch": 0.04, "learning_rate": 4.981288720295303e-05, "loss": 1.0615, "step": 39065 }, { "epoch": 0.04, "learning_rate": 4.981287761068176e-05, "loss": 1.2109, "step": 39066 }, { "epoch": 0.04, "learning_rate": 4.981286801816555e-05, "loss": 0.9595, "step": 39067 }, { "epoch": 0.04, "learning_rate": 4.98128584254044e-05, "loss": 0.8282, "step": 39068 }, { "epoch": 0.04, "learning_rate": 4.981284883239831e-05, "loss": 1.0781, "step": 39069 }, { "epoch": 0.04, "learning_rate": 4.981283923914727e-05, "loss": 0.6936, "step": 39070 }, { "epoch": 0.04, "learning_rate": 4.9812829645651296e-05, "loss": 0.7369, "step": 39071 }, { "epoch": 0.04, "learning_rate": 4.9812820051910373e-05, "loss": 0.577, "step": 39072 }, { "epoch": 0.04, "learning_rate": 4.9812810457924516e-05, "loss": 0.6978, "step": 39073 }, { "epoch": 0.04, "learning_rate": 4.981280086369371e-05, "loss": 0.7823, "step": 39074 }, { "epoch": 0.04, "learning_rate": 4.9812791269217964e-05, "loss": 0.7912, "step": 39075 }, { "epoch": 0.04, "learning_rate": 4.9812781674497276e-05, "loss": 0.8127, "step": 39076 }, { "epoch": 0.04, "learning_rate": 4.981277207953165e-05, "loss": 0.7283, "step": 39077 }, { "epoch": 0.04, "learning_rate": 4.981276248432109e-05, "loss": 0.4665, "step": 39078 }, { "epoch": 0.04, "learning_rate": 4.9812752888865574e-05, "loss": 1.0151, "step": 39079 }, { "epoch": 0.04, "learning_rate": 4.9812743293165126e-05, "loss": 0.8507, "step": 39080 }, { "epoch": 0.04, "learning_rate": 4.9812733697219736e-05, "loss": 0.8793, "step": 39081 }, { "epoch": 0.04, "learning_rate": 4.9812724101029405e-05, "loss": 0.7604, "step": 39082 }, { "epoch": 0.04, "learning_rate": 4.9812714504594125e-05, "loss": 0.5797, "step": 39083 }, { "epoch": 0.04, "learning_rate": 4.981270490791392e-05, "loss": 1.2596, "step": 39084 }, { "epoch": 0.04, "learning_rate": 4.981269531098877e-05, "loss": 1.6312, "step": 39085 }, { "epoch": 0.04, "learning_rate": 4.981268571381867e-05, "loss": 1.356, "step": 39086 }, { "epoch": 0.04, "learning_rate": 4.981267611640363e-05, "loss": 1.1605, "step": 39087 }, { "epoch": 0.04, "learning_rate": 4.981266651874366e-05, "loss": 1.0626, "step": 39088 }, { "epoch": 0.04, "learning_rate": 4.981265692083874e-05, "loss": 1.181, "step": 39089 }, { "epoch": 0.04, "learning_rate": 4.981264732268889e-05, "loss": 1.0509, "step": 39090 }, { "epoch": 0.04, "learning_rate": 4.981263772429409e-05, "loss": 1.0067, "step": 39091 }, { "epoch": 0.04, "learning_rate": 4.981262812565436e-05, "loss": 0.7276, "step": 39092 }, { "epoch": 0.04, "learning_rate": 4.981261852676968e-05, "loss": 1.0034, "step": 39093 }, { "epoch": 0.04, "learning_rate": 4.9812608927640065e-05, "loss": 0.9856, "step": 39094 }, { "epoch": 0.04, "learning_rate": 4.9812599328265507e-05, "loss": 0.8416, "step": 39095 }, { "epoch": 0.04, "learning_rate": 4.981258972864602e-05, "loss": 1.2294, "step": 39096 }, { "epoch": 0.04, "learning_rate": 4.9812580128781585e-05, "loss": 0.9845, "step": 39097 }, { "epoch": 0.04, "learning_rate": 4.981257052867221e-05, "loss": 1.4679, "step": 39098 }, { "epoch": 0.04, "learning_rate": 4.98125609283179e-05, "loss": 1.0477, "step": 39099 }, { "epoch": 0.04, "learning_rate": 4.981255132771865e-05, "loss": 0.9312, "step": 39100 }, { "epoch": 0.04, "learning_rate": 4.9812541726874456e-05, "loss": 1.146, "step": 39101 }, { "epoch": 0.04, "learning_rate": 4.9812532125785327e-05, "loss": 1.2807, "step": 39102 }, { "epoch": 0.04, "learning_rate": 4.981252252445126e-05, "loss": 0.8001, "step": 39103 }, { "epoch": 0.04, "learning_rate": 4.9812512922872256e-05, "loss": 0.825, "step": 39104 }, { "epoch": 0.04, "learning_rate": 4.981250332104831e-05, "loss": 1.1994, "step": 39105 }, { "epoch": 0.04, "learning_rate": 4.9812493718979427e-05, "loss": 1.2156, "step": 39106 }, { "epoch": 0.04, "learning_rate": 4.98124841166656e-05, "loss": 0.603, "step": 39107 }, { "epoch": 0.04, "learning_rate": 4.981247451410684e-05, "loss": 0.4826, "step": 39108 }, { "epoch": 0.04, "learning_rate": 4.9812464911303144e-05, "loss": 1.0989, "step": 39109 }, { "epoch": 0.04, "learning_rate": 4.98124553082545e-05, "loss": 1.0366, "step": 39110 }, { "epoch": 0.04, "learning_rate": 4.981244570496093e-05, "loss": 1.0607, "step": 39111 }, { "epoch": 0.04, "learning_rate": 4.981243610142241e-05, "loss": 0.9106, "step": 39112 }, { "epoch": 0.04, "learning_rate": 4.981242649763897e-05, "loss": 0.9214, "step": 39113 }, { "epoch": 0.04, "learning_rate": 4.981241689361057e-05, "loss": 1.1853, "step": 39114 }, { "epoch": 0.04, "learning_rate": 4.981240728933725e-05, "loss": 1.3109, "step": 39115 }, { "epoch": 0.04, "learning_rate": 4.981239768481898e-05, "loss": 1.2956, "step": 39116 }, { "epoch": 0.04, "learning_rate": 4.9812388080055786e-05, "loss": 1.309, "step": 39117 }, { "epoch": 0.04, "learning_rate": 4.981237847504764e-05, "loss": 1.4526, "step": 39118 }, { "epoch": 0.04, "learning_rate": 4.981236886979456e-05, "loss": 1.4584, "step": 39119 }, { "epoch": 0.04, "learning_rate": 4.981235926429655e-05, "loss": 1.0505, "step": 39120 }, { "epoch": 0.04, "learning_rate": 4.98123496585536e-05, "loss": 0.8397, "step": 39121 }, { "epoch": 0.04, "learning_rate": 4.9812340052565706e-05, "loss": 1.1974, "step": 39122 }, { "epoch": 0.04, "learning_rate": 4.9812330446332884e-05, "loss": 0.9993, "step": 39123 }, { "epoch": 0.04, "learning_rate": 4.981232083985512e-05, "loss": 1.2516, "step": 39124 }, { "epoch": 0.04, "learning_rate": 4.981231123313242e-05, "loss": 1.3788, "step": 39125 }, { "epoch": 0.04, "learning_rate": 4.981230162616479e-05, "loss": 1.2909, "step": 39126 }, { "epoch": 0.04, "learning_rate": 4.981229201895221e-05, "loss": 0.8018, "step": 39127 }, { "epoch": 0.04, "learning_rate": 4.9812282411494706e-05, "loss": 0.124, "step": 39128 }, { "epoch": 0.04, "learning_rate": 4.981227280379226e-05, "loss": 0.0891, "step": 39129 }, { "epoch": 0.04, "learning_rate": 4.9812263195844876e-05, "loss": 0.0625, "step": 39130 }, { "epoch": 0.04, "learning_rate": 4.9812253587652555e-05, "loss": 0.0367, "step": 39131 }, { "epoch": 0.04, "learning_rate": 4.9812243979215307e-05, "loss": 0.0392, "step": 39132 }, { "epoch": 0.04, "learning_rate": 4.9812234370533116e-05, "loss": 0.0366, "step": 39133 }, { "epoch": 0.04, "learning_rate": 4.981222476160599e-05, "loss": 0.0325, "step": 39134 }, { "epoch": 0.04, "learning_rate": 4.9812215152433924e-05, "loss": 0.0388, "step": 39135 }, { "epoch": 0.04, "learning_rate": 4.981220554301693e-05, "loss": 0.0346, "step": 39136 }, { "epoch": 0.04, "learning_rate": 4.9812195933355e-05, "loss": 0.0334, "step": 39137 }, { "epoch": 0.04, "learning_rate": 4.981218632344813e-05, "loss": 0.0418, "step": 39138 }, { "epoch": 0.04, "learning_rate": 4.981217671329633e-05, "loss": 0.0428, "step": 39139 }, { "epoch": 0.04, "learning_rate": 4.981216710289959e-05, "loss": 0.0991, "step": 39140 }, { "epoch": 0.04, "learning_rate": 4.981215749225791e-05, "loss": 0.0727, "step": 39141 }, { "epoch": 0.04, "learning_rate": 4.98121478813713e-05, "loss": 0.0662, "step": 39142 }, { "epoch": 0.04, "learning_rate": 4.9812138270239764e-05, "loss": 1.0773, "step": 39143 }, { "epoch": 0.04, "learning_rate": 4.981212865886328e-05, "loss": 1.1517, "step": 39144 }, { "epoch": 0.04, "learning_rate": 4.981211904724187e-05, "loss": 1.1012, "step": 39145 }, { "epoch": 0.04, "learning_rate": 4.981210943537551e-05, "loss": 0.9735, "step": 39146 }, { "epoch": 0.04, "learning_rate": 4.981209982326423e-05, "loss": 0.9379, "step": 39147 }, { "epoch": 0.04, "learning_rate": 4.9812090210908015e-05, "loss": 0.7635, "step": 39148 }, { "epoch": 0.04, "learning_rate": 4.981208059830686e-05, "loss": 1.1139, "step": 39149 }, { "epoch": 0.04, "learning_rate": 4.9812070985460765e-05, "loss": 0.6442, "step": 39150 }, { "epoch": 0.04, "learning_rate": 4.981206137236975e-05, "loss": 0.9583, "step": 39151 }, { "epoch": 0.04, "learning_rate": 4.9812051759033796e-05, "loss": 0.9248, "step": 39152 }, { "epoch": 0.04, "learning_rate": 4.98120421454529e-05, "loss": 0.8789, "step": 39153 }, { "epoch": 0.04, "learning_rate": 4.981203253162708e-05, "loss": 1.2202, "step": 39154 }, { "epoch": 0.04, "learning_rate": 4.9812022917556325e-05, "loss": 1.0357, "step": 39155 }, { "epoch": 0.04, "learning_rate": 4.9812013303240627e-05, "loss": 1.0335, "step": 39156 }, { "epoch": 0.04, "learning_rate": 4.981200368868e-05, "loss": 1.2487, "step": 39157 }, { "epoch": 0.04, "learning_rate": 4.9811994073874446e-05, "loss": 0.7313, "step": 39158 }, { "epoch": 0.04, "learning_rate": 4.9811984458823944e-05, "loss": 0.9137, "step": 39159 }, { "epoch": 0.04, "learning_rate": 4.981197484352852e-05, "loss": 1.2186, "step": 39160 }, { "epoch": 0.04, "learning_rate": 4.981196522798817e-05, "loss": 0.9003, "step": 39161 }, { "epoch": 0.04, "learning_rate": 4.981195561220287e-05, "loss": 1.2427, "step": 39162 }, { "epoch": 0.04, "learning_rate": 4.981194599617264e-05, "loss": 0.9289, "step": 39163 }, { "epoch": 0.04, "learning_rate": 4.9811936379897484e-05, "loss": 1.0902, "step": 39164 }, { "epoch": 0.04, "learning_rate": 4.981192676337739e-05, "loss": 1.0227, "step": 39165 }, { "epoch": 0.04, "learning_rate": 4.981191714661237e-05, "loss": 0.9829, "step": 39166 }, { "epoch": 0.04, "learning_rate": 4.9811907529602406e-05, "loss": 1.0965, "step": 39167 }, { "epoch": 0.04, "learning_rate": 4.981189791234752e-05, "loss": 0.9323, "step": 39168 }, { "epoch": 0.04, "learning_rate": 4.98118882948477e-05, "loss": 0.5673, "step": 39169 }, { "epoch": 0.04, "learning_rate": 4.981187867710294e-05, "loss": 1.0277, "step": 39170 }, { "epoch": 0.04, "learning_rate": 4.981186905911325e-05, "loss": 0.8771, "step": 39171 }, { "epoch": 0.04, "learning_rate": 4.981185944087863e-05, "loss": 1.1455, "step": 39172 }, { "epoch": 0.04, "learning_rate": 4.9811849822399085e-05, "loss": 0.8626, "step": 39173 }, { "epoch": 0.04, "learning_rate": 4.9811840203674594e-05, "loss": 1.2188, "step": 39174 }, { "epoch": 0.04, "learning_rate": 4.9811830584705175e-05, "loss": 1.2553, "step": 39175 }, { "epoch": 0.04, "learning_rate": 4.981182096549083e-05, "loss": 0.7568, "step": 39176 }, { "epoch": 0.04, "learning_rate": 4.9811811346031546e-05, "loss": 1.1259, "step": 39177 }, { "epoch": 0.04, "learning_rate": 4.9811801726327336e-05, "loss": 1.325, "step": 39178 }, { "epoch": 0.04, "learning_rate": 4.981179210637819e-05, "loss": 0.7901, "step": 39179 }, { "epoch": 0.04, "learning_rate": 4.981178248618411e-05, "loss": 0.6979, "step": 39180 }, { "epoch": 0.04, "learning_rate": 4.981177286574511e-05, "loss": 0.8872, "step": 39181 }, { "epoch": 0.04, "learning_rate": 4.981176324506117e-05, "loss": 0.8995, "step": 39182 }, { "epoch": 0.04, "learning_rate": 4.98117536241323e-05, "loss": 1.0638, "step": 39183 }, { "epoch": 0.04, "learning_rate": 4.98117440029585e-05, "loss": 0.9364, "step": 39184 }, { "epoch": 0.04, "learning_rate": 4.9811734381539766e-05, "loss": 2.0642, "step": 39185 }, { "epoch": 0.04, "learning_rate": 4.98117247598761e-05, "loss": 1.5358, "step": 39186 }, { "epoch": 0.04, "learning_rate": 4.981171513796751e-05, "loss": 1.0374, "step": 39187 }, { "epoch": 0.04, "learning_rate": 4.9811705515813984e-05, "loss": 0.7023, "step": 39188 }, { "epoch": 0.04, "learning_rate": 4.981169589341553e-05, "loss": 0.9653, "step": 39189 }, { "epoch": 0.04, "learning_rate": 4.981168627077214e-05, "loss": 0.8148, "step": 39190 }, { "epoch": 0.04, "learning_rate": 4.981167664788382e-05, "loss": 1.0264, "step": 39191 }, { "epoch": 0.04, "learning_rate": 4.981166702475058e-05, "loss": 1.4014, "step": 39192 }, { "epoch": 0.04, "learning_rate": 4.98116574013724e-05, "loss": 0.9131, "step": 39193 }, { "epoch": 0.04, "learning_rate": 4.9811647777749296e-05, "loss": 1.0948, "step": 39194 }, { "epoch": 0.04, "learning_rate": 4.9811638153881255e-05, "loss": 1.2823, "step": 39195 }, { "epoch": 0.04, "learning_rate": 4.981162852976829e-05, "loss": 0.6614, "step": 39196 }, { "epoch": 0.04, "learning_rate": 4.9811618905410395e-05, "loss": 1.1642, "step": 39197 }, { "epoch": 0.04, "learning_rate": 4.9811609280807564e-05, "loss": 0.8495, "step": 39198 }, { "epoch": 0.04, "learning_rate": 4.9811599655959804e-05, "loss": 1.1655, "step": 39199 }, { "epoch": 0.04, "learning_rate": 4.981159003086712e-05, "loss": 1.2998, "step": 39200 }, { "epoch": 0.04, "learning_rate": 4.98115804055295e-05, "loss": 0.9103, "step": 39201 }, { "epoch": 0.04, "learning_rate": 4.9811570779946956e-05, "loss": 0.7762, "step": 39202 }, { "epoch": 0.04, "learning_rate": 4.9811561154119484e-05, "loss": 0.6583, "step": 39203 }, { "epoch": 0.04, "learning_rate": 4.981155152804708e-05, "loss": 1.0034, "step": 39204 }, { "epoch": 0.04, "learning_rate": 4.981154190172975e-05, "loss": 1.0966, "step": 39205 }, { "epoch": 0.04, "learning_rate": 4.9811532275167486e-05, "loss": 0.8742, "step": 39206 }, { "epoch": 0.04, "learning_rate": 4.9811522648360295e-05, "loss": 1.351, "step": 39207 }, { "epoch": 0.04, "learning_rate": 4.981151302130817e-05, "loss": 1.1814, "step": 39208 }, { "epoch": 0.04, "learning_rate": 4.981150339401112e-05, "loss": 0.6525, "step": 39209 }, { "epoch": 0.04, "learning_rate": 4.981149376646914e-05, "loss": 0.9543, "step": 39210 }, { "epoch": 0.04, "learning_rate": 4.9811484138682236e-05, "loss": 1.1041, "step": 39211 }, { "epoch": 0.04, "learning_rate": 4.9811474510650405e-05, "loss": 1.3302, "step": 39212 }, { "epoch": 0.04, "learning_rate": 4.981146488237364e-05, "loss": 1.0091, "step": 39213 }, { "epoch": 0.04, "learning_rate": 4.9811455253851945e-05, "loss": 0.9231, "step": 39214 }, { "epoch": 0.04, "learning_rate": 4.981144562508533e-05, "loss": 0.7832, "step": 39215 }, { "epoch": 0.04, "learning_rate": 4.981143599607377e-05, "loss": 0.7762, "step": 39216 }, { "epoch": 0.04, "learning_rate": 4.98114263668173e-05, "loss": 0.9964, "step": 39217 }, { "epoch": 0.04, "learning_rate": 4.9811416737315895e-05, "loss": 0.9461, "step": 39218 }, { "epoch": 0.04, "learning_rate": 4.981140710756956e-05, "loss": 1.2619, "step": 39219 }, { "epoch": 0.04, "learning_rate": 4.9811397477578305e-05, "loss": 1.0262, "step": 39220 }, { "epoch": 0.04, "learning_rate": 4.9811387847342114e-05, "loss": 0.9613, "step": 39221 }, { "epoch": 0.04, "learning_rate": 4.9811378216861e-05, "loss": 1.0219, "step": 39222 }, { "epoch": 0.04, "learning_rate": 4.9811368586134956e-05, "loss": 1.2916, "step": 39223 }, { "epoch": 0.04, "learning_rate": 4.981135895516399e-05, "loss": 1.2408, "step": 39224 }, { "epoch": 0.04, "learning_rate": 4.981134932394809e-05, "loss": 0.9689, "step": 39225 }, { "epoch": 0.04, "learning_rate": 4.981133969248727e-05, "loss": 1.0131, "step": 39226 }, { "epoch": 0.04, "learning_rate": 4.9811330060781516e-05, "loss": 1.0489, "step": 39227 }, { "epoch": 0.04, "learning_rate": 4.981132042883083e-05, "loss": 1.0059, "step": 39228 }, { "epoch": 0.04, "learning_rate": 4.981131079663523e-05, "loss": 1.1499, "step": 39229 }, { "epoch": 0.04, "learning_rate": 4.98113011641947e-05, "loss": 0.9956, "step": 39230 }, { "epoch": 0.04, "learning_rate": 4.981129153150924e-05, "loss": 0.9842, "step": 39231 }, { "epoch": 0.04, "learning_rate": 4.9811281898578856e-05, "loss": 0.7992, "step": 39232 }, { "epoch": 0.04, "learning_rate": 4.981127226540354e-05, "loss": 0.8494, "step": 39233 }, { "epoch": 0.04, "learning_rate": 4.98112626319833e-05, "loss": 1.0666, "step": 39234 }, { "epoch": 0.04, "learning_rate": 4.981125299831814e-05, "loss": 0.7245, "step": 39235 }, { "epoch": 0.04, "learning_rate": 4.981124336440805e-05, "loss": 0.5841, "step": 39236 }, { "epoch": 0.04, "learning_rate": 4.981123373025303e-05, "loss": 0.9119, "step": 39237 }, { "epoch": 0.04, "learning_rate": 4.981122409585309e-05, "loss": 1.1681, "step": 39238 }, { "epoch": 0.04, "learning_rate": 4.981121446120821e-05, "loss": 1.056, "step": 39239 }, { "epoch": 0.04, "learning_rate": 4.9811204826318424e-05, "loss": 0.6005, "step": 39240 }, { "epoch": 0.04, "learning_rate": 4.98111951911837e-05, "loss": 0.665, "step": 39241 }, { "epoch": 0.04, "learning_rate": 4.9811185555804054e-05, "loss": 0.8189, "step": 39242 }, { "epoch": 0.04, "learning_rate": 4.981117592017948e-05, "loss": 0.7773, "step": 39243 }, { "epoch": 0.04, "learning_rate": 4.9811166284309985e-05, "loss": 1.0441, "step": 39244 }, { "epoch": 0.04, "learning_rate": 4.981115664819556e-05, "loss": 0.9727, "step": 39245 }, { "epoch": 0.04, "learning_rate": 4.981114701183621e-05, "loss": 1.28, "step": 39246 }, { "epoch": 0.04, "learning_rate": 4.981113737523194e-05, "loss": 1.3584, "step": 39247 }, { "epoch": 0.04, "learning_rate": 4.981112773838274e-05, "loss": 1.0465, "step": 39248 }, { "epoch": 0.04, "learning_rate": 4.981111810128862e-05, "loss": 0.7743, "step": 39249 }, { "epoch": 0.04, "learning_rate": 4.981110846394957e-05, "loss": 0.7419, "step": 39250 }, { "epoch": 0.04, "learning_rate": 4.98110988263656e-05, "loss": 1.2807, "step": 39251 }, { "epoch": 0.04, "learning_rate": 4.98110891885367e-05, "loss": 1.0047, "step": 39252 }, { "epoch": 0.04, "learning_rate": 4.981107955046288e-05, "loss": 1.0964, "step": 39253 }, { "epoch": 0.04, "learning_rate": 4.981106991214412e-05, "loss": 0.9875, "step": 39254 }, { "epoch": 0.04, "learning_rate": 4.981106027358046e-05, "loss": 0.898, "step": 39255 }, { "epoch": 0.04, "learning_rate": 4.981105063477186e-05, "loss": 0.3289, "step": 39256 }, { "epoch": 0.04, "learning_rate": 4.981104099571834e-05, "loss": 0.8392, "step": 39257 }, { "epoch": 0.04, "learning_rate": 4.9811031356419897e-05, "loss": 0.9833, "step": 39258 }, { "epoch": 0.04, "learning_rate": 4.981102171687653e-05, "loss": 1.3207, "step": 39259 }, { "epoch": 0.04, "learning_rate": 4.981101207708824e-05, "loss": 0.8308, "step": 39260 }, { "epoch": 0.04, "learning_rate": 4.981100243705502e-05, "loss": 0.7786, "step": 39261 }, { "epoch": 0.04, "learning_rate": 4.9810992796776876e-05, "loss": 1.1744, "step": 39262 }, { "epoch": 0.04, "learning_rate": 4.981098315625382e-05, "loss": 0.9696, "step": 39263 }, { "epoch": 0.04, "learning_rate": 4.981097351548583e-05, "loss": 1.0215, "step": 39264 }, { "epoch": 0.04, "learning_rate": 4.981096387447292e-05, "loss": 1.1677, "step": 39265 }, { "epoch": 0.04, "learning_rate": 4.981095423321509e-05, "loss": 0.888, "step": 39266 }, { "epoch": 0.04, "learning_rate": 4.981094459171233e-05, "loss": 0.8307, "step": 39267 }, { "epoch": 0.04, "learning_rate": 4.981093494996465e-05, "loss": 0.7331, "step": 39268 }, { "epoch": 0.04, "learning_rate": 4.981092530797205e-05, "loss": 0.7844, "step": 39269 }, { "epoch": 0.04, "learning_rate": 4.981091566573451e-05, "loss": 0.8766, "step": 39270 }, { "epoch": 0.04, "learning_rate": 4.981090602325207e-05, "loss": 1.0515, "step": 39271 }, { "epoch": 0.04, "learning_rate": 4.9810896380524697e-05, "loss": 1.0564, "step": 39272 }, { "epoch": 0.04, "learning_rate": 4.98108867375524e-05, "loss": 0.6584, "step": 39273 }, { "epoch": 0.04, "learning_rate": 4.981087709433518e-05, "loss": 0.9982, "step": 39274 }, { "epoch": 0.04, "learning_rate": 4.981086745087305e-05, "loss": 1.0005, "step": 39275 }, { "epoch": 0.04, "learning_rate": 4.9810857807165976e-05, "loss": 0.9139, "step": 39276 }, { "epoch": 0.04, "learning_rate": 4.9810848163214e-05, "loss": 0.9254, "step": 39277 }, { "epoch": 0.04, "learning_rate": 4.981083851901709e-05, "loss": 0.8472, "step": 39278 }, { "epoch": 0.04, "learning_rate": 4.981082887457526e-05, "loss": 0.9605, "step": 39279 }, { "epoch": 0.04, "learning_rate": 4.981081922988851e-05, "loss": 1.087, "step": 39280 }, { "epoch": 0.04, "learning_rate": 4.981080958495684e-05, "loss": 0.7958, "step": 39281 }, { "epoch": 0.04, "learning_rate": 4.981079993978024e-05, "loss": 1.1405, "step": 39282 }, { "epoch": 0.04, "learning_rate": 4.981079029435872e-05, "loss": 0.9852, "step": 39283 }, { "epoch": 0.04, "learning_rate": 4.981078064869229e-05, "loss": 1.1685, "step": 39284 }, { "epoch": 0.04, "learning_rate": 4.981077100278092e-05, "loss": 0.9621, "step": 39285 }, { "epoch": 0.04, "learning_rate": 4.981076135662465e-05, "loss": 0.9874, "step": 39286 }, { "epoch": 0.04, "learning_rate": 4.981075171022344e-05, "loss": 1.0804, "step": 39287 }, { "epoch": 0.04, "learning_rate": 4.981074206357731e-05, "loss": 1.2871, "step": 39288 }, { "epoch": 0.04, "learning_rate": 4.9810732416686266e-05, "loss": 0.8644, "step": 39289 }, { "epoch": 0.04, "learning_rate": 4.98107227695503e-05, "loss": 1.0402, "step": 39290 }, { "epoch": 0.04, "learning_rate": 4.9810713122169415e-05, "loss": 1.0247, "step": 39291 }, { "epoch": 0.04, "learning_rate": 4.9810703474543604e-05, "loss": 1.008, "step": 39292 }, { "epoch": 0.04, "learning_rate": 4.981069382667288e-05, "loss": 0.6948, "step": 39293 }, { "epoch": 0.04, "learning_rate": 4.981068417855723e-05, "loss": 1.2287, "step": 39294 }, { "epoch": 0.04, "learning_rate": 4.981067453019666e-05, "loss": 1.4142, "step": 39295 }, { "epoch": 0.04, "learning_rate": 4.981066488159117e-05, "loss": 0.7012, "step": 39296 }, { "epoch": 0.04, "learning_rate": 4.981065523274076e-05, "loss": 0.5182, "step": 39297 }, { "epoch": 0.04, "learning_rate": 4.981064558364542e-05, "loss": 0.9715, "step": 39298 }, { "epoch": 0.04, "learning_rate": 4.9810635934305174e-05, "loss": 0.8123, "step": 39299 }, { "epoch": 0.04, "learning_rate": 4.981062628472e-05, "loss": 0.5163, "step": 39300 }, { "epoch": 0.04, "learning_rate": 4.9810616634889904e-05, "loss": 0.8614, "step": 39301 }, { "epoch": 0.04, "learning_rate": 4.9810606984814894e-05, "loss": 1.0389, "step": 39302 }, { "epoch": 0.04, "learning_rate": 4.981059733449496e-05, "loss": 0.9489, "step": 39303 }, { "epoch": 0.04, "learning_rate": 4.981058768393011e-05, "loss": 0.8648, "step": 39304 }, { "epoch": 0.04, "learning_rate": 4.981057803312034e-05, "loss": 0.6303, "step": 39305 }, { "epoch": 0.04, "learning_rate": 4.981056838206565e-05, "loss": 0.9079, "step": 39306 }, { "epoch": 0.04, "learning_rate": 4.981055873076604e-05, "loss": 1.1502, "step": 39307 }, { "epoch": 0.04, "learning_rate": 4.981054907922151e-05, "loss": 1.0861, "step": 39308 }, { "epoch": 0.04, "learning_rate": 4.9810539427432056e-05, "loss": 1.1633, "step": 39309 }, { "epoch": 0.04, "learning_rate": 4.981052977539769e-05, "loss": 1.0303, "step": 39310 }, { "epoch": 0.04, "learning_rate": 4.98105201231184e-05, "loss": 0.5823, "step": 39311 }, { "epoch": 0.04, "learning_rate": 4.981051047059419e-05, "loss": 1.0341, "step": 39312 }, { "epoch": 0.04, "learning_rate": 4.981050081782507e-05, "loss": 0.9347, "step": 39313 }, { "epoch": 0.04, "learning_rate": 4.9810491164811025e-05, "loss": 0.9275, "step": 39314 }, { "epoch": 0.04, "learning_rate": 4.9810481511552065e-05, "loss": 0.7837, "step": 39315 }, { "epoch": 0.04, "learning_rate": 4.981047185804818e-05, "loss": 1.0472, "step": 39316 }, { "epoch": 0.04, "learning_rate": 4.981046220429938e-05, "loss": 1.1927, "step": 39317 }, { "epoch": 0.04, "learning_rate": 4.9810452550305667e-05, "loss": 0.9653, "step": 39318 }, { "epoch": 0.04, "learning_rate": 4.981044289606702e-05, "loss": 0.8895, "step": 39319 }, { "epoch": 0.04, "learning_rate": 4.981043324158347e-05, "loss": 1.0375, "step": 39320 }, { "epoch": 0.04, "learning_rate": 4.9810423586854997e-05, "loss": 1.1937, "step": 39321 }, { "epoch": 0.04, "learning_rate": 4.981041393188161e-05, "loss": 0.639, "step": 39322 }, { "epoch": 0.04, "learning_rate": 4.98104042766633e-05, "loss": 0.7989, "step": 39323 }, { "epoch": 0.04, "learning_rate": 4.9810394621200076e-05, "loss": 1.1128, "step": 39324 }, { "epoch": 0.04, "learning_rate": 4.9810384965491924e-05, "loss": 0.8364, "step": 39325 }, { "epoch": 0.04, "learning_rate": 4.981037530953886e-05, "loss": 1.1631, "step": 39326 }, { "epoch": 0.04, "learning_rate": 4.9810365653340884e-05, "loss": 0.9776, "step": 39327 }, { "epoch": 0.04, "learning_rate": 4.981035599689798e-05, "loss": 0.8689, "step": 39328 }, { "epoch": 0.04, "learning_rate": 4.9810346340210166e-05, "loss": 1.2353, "step": 39329 }, { "epoch": 0.04, "learning_rate": 4.9810336683277435e-05, "loss": 1.059, "step": 39330 }, { "epoch": 0.04, "learning_rate": 4.981032702609979e-05, "loss": 1.1029, "step": 39331 }, { "epoch": 0.04, "learning_rate": 4.981031736867722e-05, "loss": 1.1117, "step": 39332 }, { "epoch": 0.04, "learning_rate": 4.981030771100974e-05, "loss": 1.1262, "step": 39333 }, { "epoch": 0.04, "learning_rate": 4.981029805309734e-05, "loss": 1.0685, "step": 39334 }, { "epoch": 0.04, "learning_rate": 4.981028839494002e-05, "loss": 1.0004, "step": 39335 }, { "epoch": 0.04, "learning_rate": 4.9810278736537783e-05, "loss": 1.2164, "step": 39336 }, { "epoch": 0.04, "learning_rate": 4.981026907789064e-05, "loss": 1.2626, "step": 39337 }, { "epoch": 0.04, "learning_rate": 4.981025941899857e-05, "loss": 0.957, "step": 39338 }, { "epoch": 0.04, "learning_rate": 4.981024975986158e-05, "loss": 0.8842, "step": 39339 }, { "epoch": 0.04, "learning_rate": 4.9810240100479686e-05, "loss": 1.0255, "step": 39340 }, { "epoch": 0.04, "learning_rate": 4.981023044085287e-05, "loss": 0.9401, "step": 39341 }, { "epoch": 0.04, "learning_rate": 4.9810220780981136e-05, "loss": 1.0058, "step": 39342 }, { "epoch": 0.04, "learning_rate": 4.981021112086449e-05, "loss": 0.8858, "step": 39343 }, { "epoch": 0.04, "learning_rate": 4.981020146050293e-05, "loss": 1.1689, "step": 39344 }, { "epoch": 0.04, "learning_rate": 4.9810191799896446e-05, "loss": 1.9189, "step": 39345 }, { "epoch": 0.04, "learning_rate": 4.981018213904505e-05, "loss": 1.2785, "step": 39346 }, { "epoch": 0.04, "learning_rate": 4.981017247794874e-05, "loss": 0.9513, "step": 39347 }, { "epoch": 0.04, "learning_rate": 4.9810162816607506e-05, "loss": 1.3757, "step": 39348 }, { "epoch": 0.04, "learning_rate": 4.9810153155021366e-05, "loss": 1.2896, "step": 39349 }, { "epoch": 0.04, "learning_rate": 4.981014349319031e-05, "loss": 1.3869, "step": 39350 }, { "epoch": 0.04, "learning_rate": 4.9810133831114335e-05, "loss": 0.5396, "step": 39351 }, { "epoch": 0.04, "learning_rate": 4.981012416879345e-05, "loss": 0.7396, "step": 39352 }, { "epoch": 0.04, "learning_rate": 4.9810114506227646e-05, "loss": 1.0403, "step": 39353 }, { "epoch": 0.04, "learning_rate": 4.981010484341693e-05, "loss": 1.081, "step": 39354 }, { "epoch": 0.04, "learning_rate": 4.981009518036129e-05, "loss": 1.1007, "step": 39355 }, { "epoch": 0.04, "learning_rate": 4.9810085517060744e-05, "loss": 1.1455, "step": 39356 }, { "epoch": 0.04, "learning_rate": 4.981007585351528e-05, "loss": 0.8713, "step": 39357 }, { "epoch": 0.04, "learning_rate": 4.9810066189724904e-05, "loss": 1.1501, "step": 39358 }, { "epoch": 0.04, "learning_rate": 4.981005652568961e-05, "loss": 1.0876, "step": 39359 }, { "epoch": 0.04, "learning_rate": 4.981004686140941e-05, "loss": 0.9608, "step": 39360 }, { "epoch": 0.04, "learning_rate": 4.9810037196884284e-05, "loss": 0.7409, "step": 39361 }, { "epoch": 0.04, "learning_rate": 4.981002753211425e-05, "loss": 1.0544, "step": 39362 }, { "epoch": 0.04, "learning_rate": 4.98100178670993e-05, "loss": 1.3174, "step": 39363 }, { "epoch": 0.04, "learning_rate": 4.981000820183944e-05, "loss": 1.0024, "step": 39364 }, { "epoch": 0.04, "learning_rate": 4.980999853633466e-05, "loss": 0.6008, "step": 39365 }, { "epoch": 0.04, "learning_rate": 4.980998887058497e-05, "loss": 0.6251, "step": 39366 }, { "epoch": 0.04, "learning_rate": 4.980997920459036e-05, "loss": 1.1262, "step": 39367 }, { "epoch": 0.04, "learning_rate": 4.9809969538350846e-05, "loss": 0.6068, "step": 39368 }, { "epoch": 0.04, "learning_rate": 4.9809959871866415e-05, "loss": 0.8365, "step": 39369 }, { "epoch": 0.04, "learning_rate": 4.980995020513707e-05, "loss": 1.0226, "step": 39370 }, { "epoch": 0.04, "learning_rate": 4.980994053816281e-05, "loss": 1.3298, "step": 39371 }, { "epoch": 0.04, "learning_rate": 4.980993087094364e-05, "loss": 0.3622, "step": 39372 }, { "epoch": 0.04, "learning_rate": 4.980992120347956e-05, "loss": 0.8239, "step": 39373 }, { "epoch": 0.04, "learning_rate": 4.980991153577056e-05, "loss": 1.1508, "step": 39374 }, { "epoch": 0.04, "learning_rate": 4.9809901867816644e-05, "loss": 1.1408, "step": 39375 }, { "epoch": 0.04, "learning_rate": 4.980989219961782e-05, "loss": 1.3383, "step": 39376 }, { "epoch": 0.04, "learning_rate": 4.980988253117408e-05, "loss": 1.0905, "step": 39377 }, { "epoch": 0.04, "learning_rate": 4.980987286248543e-05, "loss": 1.3733, "step": 39378 }, { "epoch": 0.04, "learning_rate": 4.980986319355187e-05, "loss": 0.7359, "step": 39379 }, { "epoch": 0.04, "learning_rate": 4.9809853524373395e-05, "loss": 1.2044, "step": 39380 }, { "epoch": 0.04, "learning_rate": 4.980984385495001e-05, "loss": 0.9881, "step": 39381 }, { "epoch": 0.04, "learning_rate": 4.980983418528171e-05, "loss": 1.1748, "step": 39382 }, { "epoch": 0.04, "learning_rate": 4.98098245153685e-05, "loss": 1.2154, "step": 39383 }, { "epoch": 0.04, "learning_rate": 4.980981484521037e-05, "loss": 0.9783, "step": 39384 }, { "epoch": 0.04, "learning_rate": 4.980980517480733e-05, "loss": 0.5089, "step": 39385 }, { "epoch": 0.04, "learning_rate": 4.980979550415939e-05, "loss": 1.018, "step": 39386 }, { "epoch": 0.04, "learning_rate": 4.980978583326653e-05, "loss": 0.919, "step": 39387 }, { "epoch": 0.04, "learning_rate": 4.980977616212876e-05, "loss": 1.1514, "step": 39388 }, { "epoch": 0.04, "learning_rate": 4.9809766490746074e-05, "loss": 1.0756, "step": 39389 }, { "epoch": 0.04, "learning_rate": 4.980975681911848e-05, "loss": 1.3075, "step": 39390 }, { "epoch": 0.04, "learning_rate": 4.980974714724598e-05, "loss": 1.3014, "step": 39391 }, { "epoch": 0.04, "learning_rate": 4.980973747512856e-05, "loss": 1.1239, "step": 39392 }, { "epoch": 0.04, "learning_rate": 4.9809727802766225e-05, "loss": 1.6621, "step": 39393 }, { "epoch": 0.04, "learning_rate": 4.980971813015899e-05, "loss": 1.6739, "step": 39394 }, { "epoch": 0.04, "learning_rate": 4.9809708457306834e-05, "loss": 1.6172, "step": 39395 }, { "epoch": 0.04, "learning_rate": 4.9809698784209774e-05, "loss": 1.6647, "step": 39396 }, { "epoch": 0.04, "learning_rate": 4.9809689110867805e-05, "loss": 1.7539, "step": 39397 }, { "epoch": 0.04, "learning_rate": 4.980967943728092e-05, "loss": 1.6515, "step": 39398 }, { "epoch": 0.04, "learning_rate": 4.9809669763449126e-05, "loss": 1.5129, "step": 39399 }, { "epoch": 0.04, "learning_rate": 4.980966008937242e-05, "loss": 1.6126, "step": 39400 }, { "epoch": 0.04, "learning_rate": 4.98096504150508e-05, "loss": 1.5243, "step": 39401 }, { "epoch": 0.04, "learning_rate": 4.9809640740484274e-05, "loss": 1.4508, "step": 39402 }, { "epoch": 0.04, "learning_rate": 4.980963106567284e-05, "loss": 1.6949, "step": 39403 }, { "epoch": 0.04, "learning_rate": 4.980962139061649e-05, "loss": 1.5922, "step": 39404 }, { "epoch": 0.04, "learning_rate": 4.980961171531524e-05, "loss": 1.3797, "step": 39405 }, { "epoch": 0.04, "learning_rate": 4.980960203976907e-05, "loss": 1.6222, "step": 39406 }, { "epoch": 0.04, "learning_rate": 4.9809592363977996e-05, "loss": 1.6444, "step": 39407 }, { "epoch": 0.04, "learning_rate": 4.980958268794201e-05, "loss": 1.4398, "step": 39408 }, { "epoch": 0.04, "learning_rate": 4.980957301166111e-05, "loss": 1.5139, "step": 39409 }, { "epoch": 0.04, "learning_rate": 4.98095633351353e-05, "loss": 1.518, "step": 39410 }, { "epoch": 0.04, "learning_rate": 4.980955365836459e-05, "loss": 1.6385, "step": 39411 }, { "epoch": 0.04, "learning_rate": 4.980954398134896e-05, "loss": 1.43, "step": 39412 }, { "epoch": 0.04, "learning_rate": 4.980953430408843e-05, "loss": 1.2932, "step": 39413 }, { "epoch": 0.04, "learning_rate": 4.980952462658299e-05, "loss": 1.6308, "step": 39414 }, { "epoch": 0.04, "learning_rate": 4.9809514948832636e-05, "loss": 1.4673, "step": 39415 }, { "epoch": 0.04, "learning_rate": 4.9809505270837374e-05, "loss": 0.9639, "step": 39416 }, { "epoch": 0.04, "learning_rate": 4.98094955925972e-05, "loss": 0.4077, "step": 39417 }, { "epoch": 0.04, "learning_rate": 4.980948591411212e-05, "loss": 0.2474, "step": 39418 }, { "epoch": 0.04, "learning_rate": 4.9809476235382134e-05, "loss": 0.252, "step": 39419 }, { "epoch": 0.04, "learning_rate": 4.9809466556407234e-05, "loss": 0.2813, "step": 39420 }, { "epoch": 0.04, "learning_rate": 4.980945687718743e-05, "loss": 0.2933, "step": 39421 }, { "epoch": 0.04, "learning_rate": 4.980944719772271e-05, "loss": 0.2783, "step": 39422 }, { "epoch": 0.04, "learning_rate": 4.9809437518013095e-05, "loss": 0.8586, "step": 39423 }, { "epoch": 0.04, "learning_rate": 4.9809427838058564e-05, "loss": 1.3121, "step": 39424 }, { "epoch": 0.04, "learning_rate": 4.9809418157859125e-05, "loss": 1.9334, "step": 39425 }, { "epoch": 0.04, "learning_rate": 4.980940847741478e-05, "loss": 1.9649, "step": 39426 }, { "epoch": 0.04, "learning_rate": 4.980939879672552e-05, "loss": 1.8812, "step": 39427 }, { "epoch": 0.04, "learning_rate": 4.980938911579136e-05, "loss": 1.8704, "step": 39428 }, { "epoch": 0.04, "learning_rate": 4.980937943461229e-05, "loss": 1.868, "step": 39429 }, { "epoch": 0.04, "learning_rate": 4.98093697531883e-05, "loss": 1.8436, "step": 39430 }, { "epoch": 0.04, "learning_rate": 4.980936007151942e-05, "loss": 1.8716, "step": 39431 }, { "epoch": 0.04, "learning_rate": 4.9809350389605626e-05, "loss": 1.9028, "step": 39432 }, { "epoch": 0.04, "learning_rate": 4.9809340707446926e-05, "loss": 1.8769, "step": 39433 }, { "epoch": 0.04, "learning_rate": 4.980933102504332e-05, "loss": 1.8136, "step": 39434 }, { "epoch": 0.04, "learning_rate": 4.98093213423948e-05, "loss": 1.8193, "step": 39435 }, { "epoch": 0.04, "learning_rate": 4.980931165950138e-05, "loss": 1.8495, "step": 39436 }, { "epoch": 0.04, "learning_rate": 4.980930197636304e-05, "loss": 1.7969, "step": 39437 }, { "epoch": 0.04, "learning_rate": 4.980929229297981e-05, "loss": 1.816, "step": 39438 }, { "epoch": 0.04, "learning_rate": 4.980928260935166e-05, "loss": 1.5572, "step": 39439 }, { "epoch": 0.04, "learning_rate": 4.9809272925478614e-05, "loss": 1.0006, "step": 39440 }, { "epoch": 0.04, "learning_rate": 4.980926324136066e-05, "loss": 0.8301, "step": 39441 }, { "epoch": 0.04, "learning_rate": 4.980925355699779e-05, "loss": 0.7767, "step": 39442 }, { "epoch": 0.04, "learning_rate": 4.980924387239002e-05, "loss": 0.9847, "step": 39443 }, { "epoch": 0.04, "learning_rate": 4.980923418753735e-05, "loss": 1.017, "step": 39444 }, { "epoch": 0.04, "learning_rate": 4.980922450243976e-05, "loss": 0.9332, "step": 39445 }, { "epoch": 0.04, "learning_rate": 4.9809214817097274e-05, "loss": 1.0278, "step": 39446 }, { "epoch": 0.04, "learning_rate": 4.980920513150987e-05, "loss": 1.0741, "step": 39447 }, { "epoch": 0.04, "learning_rate": 4.980919544567757e-05, "loss": 0.829, "step": 39448 }, { "epoch": 0.04, "learning_rate": 4.9809185759600366e-05, "loss": 1.1072, "step": 39449 }, { "epoch": 0.04, "learning_rate": 4.9809176073278255e-05, "loss": 1.1097, "step": 39450 }, { "epoch": 0.04, "learning_rate": 4.980916638671123e-05, "loss": 0.7462, "step": 39451 }, { "epoch": 0.04, "learning_rate": 4.980915669989931e-05, "loss": 1.1392, "step": 39452 }, { "epoch": 0.04, "learning_rate": 4.9809147012842475e-05, "loss": 1.2443, "step": 39453 }, { "epoch": 0.04, "learning_rate": 4.980913732554075e-05, "loss": 0.981, "step": 39454 }, { "epoch": 0.04, "learning_rate": 4.9809127637994104e-05, "loss": 1.0561, "step": 39455 }, { "epoch": 0.04, "learning_rate": 4.980911795020255e-05, "loss": 0.3015, "step": 39456 }, { "epoch": 0.04, "learning_rate": 4.98091082621661e-05, "loss": 0.5833, "step": 39457 }, { "epoch": 0.04, "learning_rate": 4.980909857388474e-05, "loss": 0.7208, "step": 39458 }, { "epoch": 0.04, "learning_rate": 4.980908888535848e-05, "loss": 0.7793, "step": 39459 }, { "epoch": 0.04, "learning_rate": 4.9809079196587314e-05, "loss": 0.9635, "step": 39460 }, { "epoch": 0.04, "learning_rate": 4.980906950757125e-05, "loss": 1.03, "step": 39461 }, { "epoch": 0.04, "learning_rate": 4.980905981831027e-05, "loss": 1.021, "step": 39462 }, { "epoch": 0.04, "learning_rate": 4.980905012880439e-05, "loss": 1.1803, "step": 39463 }, { "epoch": 0.04, "learning_rate": 4.98090404390536e-05, "loss": 1.2164, "step": 39464 }, { "epoch": 0.04, "learning_rate": 4.980903074905792e-05, "loss": 1.0066, "step": 39465 }, { "epoch": 0.04, "learning_rate": 4.980902105881732e-05, "loss": 0.9976, "step": 39466 }, { "epoch": 0.04, "learning_rate": 4.9809011368331824e-05, "loss": 1.0251, "step": 39467 }, { "epoch": 0.04, "learning_rate": 4.980900167760142e-05, "loss": 0.9697, "step": 39468 }, { "epoch": 0.04, "learning_rate": 4.980899198662611e-05, "loss": 0.9595, "step": 39469 }, { "epoch": 0.04, "learning_rate": 4.9808982295405906e-05, "loss": 0.7478, "step": 39470 }, { "epoch": 0.04, "learning_rate": 4.980897260394079e-05, "loss": 1.0769, "step": 39471 }, { "epoch": 0.04, "learning_rate": 4.9808962912230774e-05, "loss": 1.0043, "step": 39472 }, { "epoch": 0.04, "learning_rate": 4.980895322027585e-05, "loss": 0.9089, "step": 39473 }, { "epoch": 0.04, "learning_rate": 4.9808943528076024e-05, "loss": 0.966, "step": 39474 }, { "epoch": 0.04, "learning_rate": 4.98089338356313e-05, "loss": 1.0338, "step": 39475 }, { "epoch": 0.04, "learning_rate": 4.9808924142941664e-05, "loss": 0.7414, "step": 39476 }, { "epoch": 0.04, "learning_rate": 4.980891445000713e-05, "loss": 1.0217, "step": 39477 }, { "epoch": 0.04, "learning_rate": 4.980890475682769e-05, "loss": 0.4078, "step": 39478 }, { "epoch": 0.04, "learning_rate": 4.9808895063403347e-05, "loss": 0.2495, "step": 39479 }, { "epoch": 0.04, "learning_rate": 4.980888536973411e-05, "loss": 0.2139, "step": 39480 }, { "epoch": 0.04, "learning_rate": 4.9808875675819964e-05, "loss": 0.2069, "step": 39481 }, { "epoch": 0.04, "learning_rate": 4.9808865981660914e-05, "loss": 0.1812, "step": 39482 }, { "epoch": 0.04, "learning_rate": 4.980885628725696e-05, "loss": 0.2064, "step": 39483 }, { "epoch": 0.04, "learning_rate": 4.9808846592608105e-05, "loss": 0.2731, "step": 39484 }, { "epoch": 0.04, "learning_rate": 4.980883689771435e-05, "loss": 0.3351, "step": 39485 }, { "epoch": 0.04, "learning_rate": 4.980882720257569e-05, "loss": 0.857, "step": 39486 }, { "epoch": 0.04, "learning_rate": 4.980881750719213e-05, "loss": 1.1341, "step": 39487 }, { "epoch": 0.04, "learning_rate": 4.980880781156366e-05, "loss": 0.5107, "step": 39488 }, { "epoch": 0.04, "learning_rate": 4.98087981156903e-05, "loss": 1.362, "step": 39489 }, { "epoch": 0.04, "learning_rate": 4.980878841957203e-05, "loss": 1.2048, "step": 39490 }, { "epoch": 0.04, "learning_rate": 4.980877872320886e-05, "loss": 1.0569, "step": 39491 }, { "epoch": 0.04, "learning_rate": 4.980876902660079e-05, "loss": 1.1194, "step": 39492 }, { "epoch": 0.04, "learning_rate": 4.980875932974781e-05, "loss": 0.8547, "step": 39493 }, { "epoch": 0.04, "learning_rate": 4.9808749632649935e-05, "loss": 1.3534, "step": 39494 }, { "epoch": 0.04, "learning_rate": 4.980873993530716e-05, "loss": 1.2066, "step": 39495 }, { "epoch": 0.04, "learning_rate": 4.980873023771948e-05, "loss": 1.0963, "step": 39496 }, { "epoch": 0.04, "learning_rate": 4.98087205398869e-05, "loss": 0.9997, "step": 39497 }, { "epoch": 0.04, "learning_rate": 4.980871084180943e-05, "loss": 0.9457, "step": 39498 }, { "epoch": 0.04, "learning_rate": 4.9808701143487035e-05, "loss": 0.8345, "step": 39499 }, { "epoch": 0.04, "learning_rate": 4.9808691444919755e-05, "loss": 1.3071, "step": 39500 }, { "epoch": 0.04, "eval_loss": 1.0904228687286377, "eval_runtime": 127.8411, "eval_samples_per_second": 10.834, "eval_steps_per_second": 5.421, "step": 39500 }, { "epoch": 0.04, "learning_rate": 4.980868174610757e-05, "loss": 1.5089, "step": 39501 }, { "epoch": 0.04, "learning_rate": 4.9808672047050484e-05, "loss": 1.3452, "step": 39502 }, { "epoch": 0.04, "learning_rate": 4.98086623477485e-05, "loss": 1.1372, "step": 39503 }, { "epoch": 0.04, "learning_rate": 4.980865264820162e-05, "loss": 0.6489, "step": 39504 }, { "epoch": 0.04, "learning_rate": 4.9808642948409826e-05, "loss": 0.4958, "step": 39505 }, { "epoch": 0.04, "learning_rate": 4.980863324837314e-05, "loss": 1.0723, "step": 39506 }, { "epoch": 0.04, "learning_rate": 4.980862354809155e-05, "loss": 0.6679, "step": 39507 }, { "epoch": 0.04, "learning_rate": 4.9808613847565065e-05, "loss": 0.4005, "step": 39508 }, { "epoch": 0.04, "learning_rate": 4.980860414679368e-05, "loss": 0.7643, "step": 39509 }, { "epoch": 0.04, "learning_rate": 4.980859444577739e-05, "loss": 1.1303, "step": 39510 }, { "epoch": 0.04, "learning_rate": 4.98085847445162e-05, "loss": 0.9753, "step": 39511 }, { "epoch": 0.04, "learning_rate": 4.980857504301011e-05, "loss": 0.8158, "step": 39512 }, { "epoch": 0.04, "learning_rate": 4.9808565341259126e-05, "loss": 1.0136, "step": 39513 }, { "epoch": 0.04, "learning_rate": 4.980855563926324e-05, "loss": 0.9954, "step": 39514 }, { "epoch": 0.04, "learning_rate": 4.980854593702245e-05, "loss": 0.8018, "step": 39515 }, { "epoch": 0.04, "learning_rate": 4.980853623453676e-05, "loss": 1.17, "step": 39516 }, { "epoch": 0.04, "learning_rate": 4.9808526531806174e-05, "loss": 1.0952, "step": 39517 }, { "epoch": 0.04, "learning_rate": 4.980851682883069e-05, "loss": 0.8008, "step": 39518 }, { "epoch": 0.04, "learning_rate": 4.980850712561031e-05, "loss": 0.8252, "step": 39519 }, { "epoch": 0.04, "learning_rate": 4.980849742214503e-05, "loss": 1.1353, "step": 39520 }, { "epoch": 0.04, "learning_rate": 4.980848771843484e-05, "loss": 1.002, "step": 39521 }, { "epoch": 0.04, "learning_rate": 4.980847801447976e-05, "loss": 0.9392, "step": 39522 }, { "epoch": 0.04, "learning_rate": 4.980846831027978e-05, "loss": 1.0779, "step": 39523 }, { "epoch": 0.04, "learning_rate": 4.98084586058349e-05, "loss": 0.9561, "step": 39524 }, { "epoch": 0.04, "learning_rate": 4.980844890114512e-05, "loss": 1.0676, "step": 39525 }, { "epoch": 0.04, "learning_rate": 4.980843919621044e-05, "loss": 1.288, "step": 39526 }, { "epoch": 0.04, "learning_rate": 4.980842949103087e-05, "loss": 0.4913, "step": 39527 }, { "epoch": 0.04, "learning_rate": 4.98084197856064e-05, "loss": 0.4486, "step": 39528 }, { "epoch": 0.04, "learning_rate": 4.980841007993703e-05, "loss": 0.3775, "step": 39529 }, { "epoch": 0.04, "learning_rate": 4.9808400374022755e-05, "loss": 0.4906, "step": 39530 }, { "epoch": 0.04, "learning_rate": 4.980839066786359e-05, "loss": 0.5442, "step": 39531 }, { "epoch": 0.04, "learning_rate": 4.9808380961459524e-05, "loss": 0.7615, "step": 39532 }, { "epoch": 0.04, "learning_rate": 4.980837125481056e-05, "loss": 0.9001, "step": 39533 }, { "epoch": 0.04, "learning_rate": 4.9808361547916695e-05, "loss": 0.9437, "step": 39534 }, { "epoch": 0.04, "learning_rate": 4.9808351840777936e-05, "loss": 1.1139, "step": 39535 }, { "epoch": 0.04, "learning_rate": 4.980834213339428e-05, "loss": 1.1029, "step": 39536 }, { "epoch": 0.04, "learning_rate": 4.980833242576572e-05, "loss": 1.0275, "step": 39537 }, { "epoch": 0.04, "learning_rate": 4.9808322717892274e-05, "loss": 0.7664, "step": 39538 }, { "epoch": 0.04, "learning_rate": 4.9808313009773924e-05, "loss": 0.9369, "step": 39539 }, { "epoch": 0.04, "learning_rate": 4.980830330141068e-05, "loss": 1.2068, "step": 39540 }, { "epoch": 0.04, "learning_rate": 4.980829359280254e-05, "loss": 0.9247, "step": 39541 }, { "epoch": 0.04, "learning_rate": 4.98082838839495e-05, "loss": 0.8073, "step": 39542 }, { "epoch": 0.04, "learning_rate": 4.980827417485156e-05, "loss": 0.9949, "step": 39543 }, { "epoch": 0.04, "learning_rate": 4.9808264465508725e-05, "loss": 1.4867, "step": 39544 }, { "epoch": 0.04, "learning_rate": 4.980825475592099e-05, "loss": 0.954, "step": 39545 }, { "epoch": 0.04, "learning_rate": 4.980824504608836e-05, "loss": 1.0732, "step": 39546 }, { "epoch": 0.04, "learning_rate": 4.9808235336010846e-05, "loss": 0.7687, "step": 39547 }, { "epoch": 0.04, "learning_rate": 4.980822562568842e-05, "loss": 1.0208, "step": 39548 }, { "epoch": 0.04, "learning_rate": 4.9808215915121104e-05, "loss": 0.8852, "step": 39549 }, { "epoch": 0.04, "learning_rate": 4.980820620430889e-05, "loss": 0.6995, "step": 39550 }, { "epoch": 0.04, "learning_rate": 4.980819649325178e-05, "loss": 1.1055, "step": 39551 }, { "epoch": 0.04, "learning_rate": 4.980818678194978e-05, "loss": 0.8915, "step": 39552 }, { "epoch": 0.04, "learning_rate": 4.980817707040287e-05, "loss": 1.1701, "step": 39553 }, { "epoch": 0.04, "learning_rate": 4.980816735861108e-05, "loss": 0.9991, "step": 39554 }, { "epoch": 0.04, "learning_rate": 4.9808157646574385e-05, "loss": 0.5249, "step": 39555 }, { "epoch": 0.04, "learning_rate": 4.98081479342928e-05, "loss": 0.7577, "step": 39556 }, { "epoch": 0.04, "learning_rate": 4.980813822176631e-05, "loss": 1.1473, "step": 39557 }, { "epoch": 0.04, "learning_rate": 4.980812850899493e-05, "loss": 0.5606, "step": 39558 }, { "epoch": 0.04, "learning_rate": 4.980811879597866e-05, "loss": 0.9582, "step": 39559 }, { "epoch": 0.04, "learning_rate": 4.980810908271749e-05, "loss": 1.4524, "step": 39560 }, { "epoch": 0.04, "learning_rate": 4.980809936921142e-05, "loss": 1.0736, "step": 39561 }, { "epoch": 0.04, "learning_rate": 4.9808089655460466e-05, "loss": 0.8222, "step": 39562 }, { "epoch": 0.04, "learning_rate": 4.980807994146461e-05, "loss": 1.1277, "step": 39563 }, { "epoch": 0.04, "learning_rate": 4.980807022722386e-05, "loss": 0.8421, "step": 39564 }, { "epoch": 0.04, "learning_rate": 4.980806051273821e-05, "loss": 0.7651, "step": 39565 }, { "epoch": 0.04, "learning_rate": 4.9808050798007675e-05, "loss": 0.9937, "step": 39566 }, { "epoch": 0.04, "learning_rate": 4.9808041083032235e-05, "loss": 1.1369, "step": 39567 }, { "epoch": 0.04, "learning_rate": 4.980803136781191e-05, "loss": 0.8724, "step": 39568 }, { "epoch": 0.04, "learning_rate": 4.980802165234668e-05, "loss": 1.1304, "step": 39569 }, { "epoch": 0.04, "learning_rate": 4.980801193663657e-05, "loss": 1.1183, "step": 39570 }, { "epoch": 0.04, "learning_rate": 4.9808002220681546e-05, "loss": 0.9634, "step": 39571 }, { "epoch": 0.04, "learning_rate": 4.980799250448165e-05, "loss": 1.3035, "step": 39572 }, { "epoch": 0.04, "learning_rate": 4.9807982788036844e-05, "loss": 0.8984, "step": 39573 }, { "epoch": 0.04, "learning_rate": 4.980797307134715e-05, "loss": 1.0317, "step": 39574 }, { "epoch": 0.04, "learning_rate": 4.9807963354412565e-05, "loss": 0.9827, "step": 39575 }, { "epoch": 0.04, "learning_rate": 4.9807953637233084e-05, "loss": 1.1055, "step": 39576 }, { "epoch": 0.04, "learning_rate": 4.980794391980871e-05, "loss": 0.8893, "step": 39577 }, { "epoch": 0.04, "learning_rate": 4.980793420213944e-05, "loss": 0.8415, "step": 39578 }, { "epoch": 0.04, "learning_rate": 4.980792448422528e-05, "loss": 1.0539, "step": 39579 }, { "epoch": 0.04, "learning_rate": 4.980791476606622e-05, "loss": 0.9605, "step": 39580 }, { "epoch": 0.04, "learning_rate": 4.980790504766227e-05, "loss": 0.1362, "step": 39581 }, { "epoch": 0.04, "learning_rate": 4.980789532901343e-05, "loss": 0.1407, "step": 39582 }, { "epoch": 0.04, "learning_rate": 4.980788561011969e-05, "loss": 0.0693, "step": 39583 }, { "epoch": 0.04, "learning_rate": 4.980787589098106e-05, "loss": 0.1408, "step": 39584 }, { "epoch": 0.04, "learning_rate": 4.980786617159754e-05, "loss": 0.1129, "step": 39585 }, { "epoch": 0.04, "learning_rate": 4.9807856451969124e-05, "loss": 0.0893, "step": 39586 }, { "epoch": 0.04, "learning_rate": 4.980784673209582e-05, "loss": 0.0869, "step": 39587 }, { "epoch": 0.04, "learning_rate": 4.980783701197762e-05, "loss": 0.14, "step": 39588 }, { "epoch": 0.04, "learning_rate": 4.9807827291614526e-05, "loss": 0.17, "step": 39589 }, { "epoch": 0.04, "learning_rate": 4.980781757100654e-05, "loss": 0.0472, "step": 39590 }, { "epoch": 0.04, "learning_rate": 4.980780785015367e-05, "loss": 0.1104, "step": 39591 }, { "epoch": 0.04, "learning_rate": 4.9807798129055894e-05, "loss": 0.0283, "step": 39592 }, { "epoch": 0.04, "learning_rate": 4.980778840771323e-05, "loss": 0.1044, "step": 39593 }, { "epoch": 0.04, "learning_rate": 4.9807778686125684e-05, "loss": 0.1385, "step": 39594 }, { "epoch": 0.04, "learning_rate": 4.9807768964293234e-05, "loss": 0.1543, "step": 39595 }, { "epoch": 0.04, "learning_rate": 4.9807759242215904e-05, "loss": 0.999, "step": 39596 }, { "epoch": 0.04, "learning_rate": 4.980774951989367e-05, "loss": 1.1491, "step": 39597 }, { "epoch": 0.04, "learning_rate": 4.980773979732655e-05, "loss": 1.1371, "step": 39598 }, { "epoch": 0.04, "learning_rate": 4.9807730074514535e-05, "loss": 1.0269, "step": 39599 }, { "epoch": 0.04, "learning_rate": 4.9807720351457635e-05, "loss": 0.9628, "step": 39600 }, { "epoch": 0.04, "learning_rate": 4.980771062815584e-05, "loss": 0.7832, "step": 39601 }, { "epoch": 0.04, "learning_rate": 4.980770090460916e-05, "loss": 0.6424, "step": 39602 }, { "epoch": 0.04, "learning_rate": 4.980769118081757e-05, "loss": 0.5132, "step": 39603 }, { "epoch": 0.04, "learning_rate": 4.9807681456781106e-05, "loss": 0.7342, "step": 39604 }, { "epoch": 0.04, "learning_rate": 4.980767173249975e-05, "loss": 0.8218, "step": 39605 }, { "epoch": 0.04, "learning_rate": 4.98076620079735e-05, "loss": 1.2693, "step": 39606 }, { "epoch": 0.04, "learning_rate": 4.980765228320235e-05, "loss": 1.2889, "step": 39607 }, { "epoch": 0.04, "learning_rate": 4.980764255818633e-05, "loss": 1.0838, "step": 39608 }, { "epoch": 0.04, "learning_rate": 4.98076328329254e-05, "loss": 1.0751, "step": 39609 }, { "epoch": 0.04, "learning_rate": 4.980762310741959e-05, "loss": 1.128, "step": 39610 }, { "epoch": 0.04, "learning_rate": 4.980761338166889e-05, "loss": 1.0213, "step": 39611 }, { "epoch": 0.04, "learning_rate": 4.98076036556733e-05, "loss": 0.9926, "step": 39612 }, { "epoch": 0.04, "learning_rate": 4.980759392943281e-05, "loss": 0.999, "step": 39613 }, { "epoch": 0.04, "learning_rate": 4.980758420294743e-05, "loss": 0.9834, "step": 39614 }, { "epoch": 0.04, "learning_rate": 4.980757447621717e-05, "loss": 1.4501, "step": 39615 }, { "epoch": 0.04, "learning_rate": 4.980756474924202e-05, "loss": 1.042, "step": 39616 }, { "epoch": 0.04, "learning_rate": 4.980755502202198e-05, "loss": 1.2663, "step": 39617 }, { "epoch": 0.04, "learning_rate": 4.9807545294557046e-05, "loss": 1.1256, "step": 39618 }, { "epoch": 0.04, "learning_rate": 4.980753556684722e-05, "loss": 1.1558, "step": 39619 }, { "epoch": 0.04, "learning_rate": 4.980752583889251e-05, "loss": 1.3595, "step": 39620 }, { "epoch": 0.04, "learning_rate": 4.9807516110692906e-05, "loss": 1.0893, "step": 39621 }, { "epoch": 0.04, "learning_rate": 4.9807506382248415e-05, "loss": 1.1788, "step": 39622 }, { "epoch": 0.04, "learning_rate": 4.980749665355903e-05, "loss": 0.8338, "step": 39623 }, { "epoch": 0.04, "learning_rate": 4.9807486924624766e-05, "loss": 1.426, "step": 39624 }, { "epoch": 0.04, "learning_rate": 4.9807477195445606e-05, "loss": 1.119, "step": 39625 }, { "epoch": 0.04, "learning_rate": 4.980746746602156e-05, "loss": 0.8227, "step": 39626 }, { "epoch": 0.04, "learning_rate": 4.980745773635262e-05, "loss": 1.3086, "step": 39627 }, { "epoch": 0.04, "learning_rate": 4.98074480064388e-05, "loss": 1.2431, "step": 39628 }, { "epoch": 0.04, "learning_rate": 4.980743827628008e-05, "loss": 1.1512, "step": 39629 }, { "epoch": 0.04, "learning_rate": 4.980742854587648e-05, "loss": 1.4715, "step": 39630 }, { "epoch": 0.04, "learning_rate": 4.980741881522799e-05, "loss": 1.0971, "step": 39631 }, { "epoch": 0.04, "learning_rate": 4.980740908433461e-05, "loss": 0.869, "step": 39632 }, { "epoch": 0.04, "learning_rate": 4.980739935319634e-05, "loss": 1.0467, "step": 39633 }, { "epoch": 0.04, "learning_rate": 4.9807389621813185e-05, "loss": 0.7941, "step": 39634 }, { "epoch": 0.04, "learning_rate": 4.980737989018514e-05, "loss": 0.8166, "step": 39635 }, { "epoch": 0.04, "learning_rate": 4.9807370158312204e-05, "loss": 1.2776, "step": 39636 }, { "epoch": 0.04, "learning_rate": 4.980736042619439e-05, "loss": 1.0115, "step": 39637 }, { "epoch": 0.04, "learning_rate": 4.980735069383168e-05, "loss": 1.2009, "step": 39638 }, { "epoch": 0.04, "learning_rate": 4.9807340961224084e-05, "loss": 1.1516, "step": 39639 }, { "epoch": 0.04, "learning_rate": 4.98073312283716e-05, "loss": 0.9791, "step": 39640 }, { "epoch": 0.04, "learning_rate": 4.980732149527423e-05, "loss": 1.0081, "step": 39641 }, { "epoch": 0.04, "learning_rate": 4.980731176193197e-05, "loss": 0.9408, "step": 39642 }, { "epoch": 0.04, "learning_rate": 4.9807302028344825e-05, "loss": 0.8656, "step": 39643 }, { "epoch": 0.04, "learning_rate": 4.98072922945128e-05, "loss": 1.0545, "step": 39644 }, { "epoch": 0.04, "learning_rate": 4.980728256043588e-05, "loss": 0.9576, "step": 39645 }, { "epoch": 0.04, "learning_rate": 4.980727282611407e-05, "loss": 0.7662, "step": 39646 }, { "epoch": 0.04, "learning_rate": 4.9807263091547374e-05, "loss": 1.1132, "step": 39647 }, { "epoch": 0.04, "learning_rate": 4.98072533567358e-05, "loss": 0.9596, "step": 39648 }, { "epoch": 0.04, "learning_rate": 4.980724362167933e-05, "loss": 0.8194, "step": 39649 }, { "epoch": 0.04, "learning_rate": 4.980723388637798e-05, "loss": 0.945, "step": 39650 }, { "epoch": 0.04, "learning_rate": 4.980722415083173e-05, "loss": 1.2175, "step": 39651 }, { "epoch": 0.04, "learning_rate": 4.9807214415040606e-05, "loss": 1.0356, "step": 39652 }, { "epoch": 0.04, "learning_rate": 4.980720467900459e-05, "loss": 1.229, "step": 39653 }, { "epoch": 0.04, "learning_rate": 4.980719494272369e-05, "loss": 0.7093, "step": 39654 }, { "epoch": 0.04, "learning_rate": 4.9807185206197905e-05, "loss": 1.053, "step": 39655 }, { "epoch": 0.04, "learning_rate": 4.9807175469427236e-05, "loss": 0.8506, "step": 39656 }, { "epoch": 0.04, "learning_rate": 4.9807165732411674e-05, "loss": 1.2515, "step": 39657 }, { "epoch": 0.04, "learning_rate": 4.980715599515123e-05, "loss": 1.3079, "step": 39658 }, { "epoch": 0.04, "learning_rate": 4.980714625764591e-05, "loss": 1.1382, "step": 39659 }, { "epoch": 0.04, "learning_rate": 4.980713651989569e-05, "loss": 1.0446, "step": 39660 }, { "epoch": 0.04, "learning_rate": 4.980712678190059e-05, "loss": 0.8599, "step": 39661 }, { "epoch": 0.04, "learning_rate": 4.98071170436606e-05, "loss": 0.8707, "step": 39662 }, { "epoch": 0.04, "learning_rate": 4.9807107305175724e-05, "loss": 0.9389, "step": 39663 }, { "epoch": 0.04, "learning_rate": 4.980709756644597e-05, "loss": 0.9083, "step": 39664 }, { "epoch": 0.04, "learning_rate": 4.9807087827471335e-05, "loss": 1.1268, "step": 39665 }, { "epoch": 0.04, "learning_rate": 4.98070780882518e-05, "loss": 1.0101, "step": 39666 }, { "epoch": 0.04, "learning_rate": 4.980706834878739e-05, "loss": 0.7073, "step": 39667 }, { "epoch": 0.04, "learning_rate": 4.9807058609078094e-05, "loss": 1.0927, "step": 39668 }, { "epoch": 0.04, "learning_rate": 4.980704886912391e-05, "loss": 1.3372, "step": 39669 }, { "epoch": 0.04, "learning_rate": 4.980703912892485e-05, "loss": 1.4678, "step": 39670 }, { "epoch": 0.04, "learning_rate": 4.98070293884809e-05, "loss": 0.952, "step": 39671 }, { "epoch": 0.04, "learning_rate": 4.980701964779206e-05, "loss": 1.1444, "step": 39672 }, { "epoch": 0.04, "learning_rate": 4.980700990685834e-05, "loss": 0.7154, "step": 39673 }, { "epoch": 0.04, "learning_rate": 4.980700016567973e-05, "loss": 0.182, "step": 39674 }, { "epoch": 0.04, "learning_rate": 4.980699042425624e-05, "loss": 0.098, "step": 39675 }, { "epoch": 0.04, "learning_rate": 4.980698068258787e-05, "loss": 0.2037, "step": 39676 }, { "epoch": 0.04, "learning_rate": 4.980697094067461e-05, "loss": 0.1751, "step": 39677 }, { "epoch": 0.04, "learning_rate": 4.9806961198516475e-05, "loss": 0.1502, "step": 39678 }, { "epoch": 0.04, "learning_rate": 4.980695145611345e-05, "loss": 0.14, "step": 39679 }, { "epoch": 0.04, "learning_rate": 4.980694171346554e-05, "loss": 0.073, "step": 39680 }, { "epoch": 0.04, "learning_rate": 4.980693197057275e-05, "loss": 0.0821, "step": 39681 }, { "epoch": 0.04, "learning_rate": 4.9806922227435074e-05, "loss": 0.6482, "step": 39682 }, { "epoch": 0.04, "learning_rate": 4.980691248405251e-05, "loss": 1.0999, "step": 39683 }, { "epoch": 0.04, "learning_rate": 4.980690274042507e-05, "loss": 0.6375, "step": 39684 }, { "epoch": 0.04, "learning_rate": 4.980689299655274e-05, "loss": 0.5399, "step": 39685 }, { "epoch": 0.04, "learning_rate": 4.980688325243553e-05, "loss": 0.3151, "step": 39686 }, { "epoch": 0.04, "learning_rate": 4.980687350807344e-05, "loss": 0.9839, "step": 39687 }, { "epoch": 0.04, "learning_rate": 4.9806863763466463e-05, "loss": 1.0598, "step": 39688 }, { "epoch": 0.04, "learning_rate": 4.98068540186146e-05, "loss": 1.0311, "step": 39689 }, { "epoch": 0.04, "learning_rate": 4.980684427351786e-05, "loss": 1.1698, "step": 39690 }, { "epoch": 0.04, "learning_rate": 4.9806834528176236e-05, "loss": 0.9581, "step": 39691 }, { "epoch": 0.04, "learning_rate": 4.9806824782589737e-05, "loss": 0.9782, "step": 39692 }, { "epoch": 0.04, "learning_rate": 4.980681503675834e-05, "loss": 1.6296, "step": 39693 }, { "epoch": 0.04, "learning_rate": 4.980680529068207e-05, "loss": 1.4184, "step": 39694 }, { "epoch": 0.04, "learning_rate": 4.980679554436092e-05, "loss": 0.2725, "step": 39695 }, { "epoch": 0.04, "learning_rate": 4.980678579779488e-05, "loss": 0.2316, "step": 39696 }, { "epoch": 0.04, "learning_rate": 4.980677605098396e-05, "loss": 0.3968, "step": 39697 }, { "epoch": 0.04, "learning_rate": 4.980676630392817e-05, "loss": 0.1634, "step": 39698 }, { "epoch": 0.04, "learning_rate": 4.9806756556627476e-05, "loss": 1.0726, "step": 39699 }, { "epoch": 0.04, "learning_rate": 4.980674680908192e-05, "loss": 1.428, "step": 39700 }, { "epoch": 0.04, "learning_rate": 4.9806737061291467e-05, "loss": 0.8447, "step": 39701 }, { "epoch": 0.04, "learning_rate": 4.980672731325614e-05, "loss": 1.1002, "step": 39702 }, { "epoch": 0.04, "learning_rate": 4.9806717564975935e-05, "loss": 1.0821, "step": 39703 }, { "epoch": 0.04, "learning_rate": 4.980670781645084e-05, "loss": 0.8764, "step": 39704 }, { "epoch": 0.04, "learning_rate": 4.9806698067680866e-05, "loss": 0.886, "step": 39705 }, { "epoch": 0.04, "learning_rate": 4.980668831866602e-05, "loss": 1.1687, "step": 39706 }, { "epoch": 0.04, "learning_rate": 4.980667856940628e-05, "loss": 1.0449, "step": 39707 }, { "epoch": 0.04, "learning_rate": 4.9806668819901667e-05, "loss": 1.1332, "step": 39708 }, { "epoch": 0.04, "learning_rate": 4.980665907015217e-05, "loss": 1.1815, "step": 39709 }, { "epoch": 0.04, "learning_rate": 4.980664932015779e-05, "loss": 0.9185, "step": 39710 }, { "epoch": 0.04, "learning_rate": 4.9806639569918534e-05, "loss": 1.1374, "step": 39711 }, { "epoch": 0.04, "learning_rate": 4.9806629819434396e-05, "loss": 0.9978, "step": 39712 }, { "epoch": 0.04, "learning_rate": 4.9806620068705377e-05, "loss": 0.6502, "step": 39713 }, { "epoch": 0.04, "learning_rate": 4.980661031773148e-05, "loss": 0.4418, "step": 39714 }, { "epoch": 0.04, "learning_rate": 4.9806600566512696e-05, "loss": 0.3249, "step": 39715 }, { "epoch": 0.04, "learning_rate": 4.9806590815049035e-05, "loss": 0.3394, "step": 39716 }, { "epoch": 0.04, "learning_rate": 4.9806581063340493e-05, "loss": 0.19, "step": 39717 }, { "epoch": 0.04, "learning_rate": 4.980657131138707e-05, "loss": 0.1253, "step": 39718 }, { "epoch": 0.04, "learning_rate": 4.9806561559188775e-05, "loss": 0.0901, "step": 39719 }, { "epoch": 0.04, "learning_rate": 4.980655180674559e-05, "loss": 0.6536, "step": 39720 }, { "epoch": 0.04, "learning_rate": 4.9806542054057534e-05, "loss": 1.2669, "step": 39721 }, { "epoch": 0.04, "learning_rate": 4.980653230112459e-05, "loss": 1.1506, "step": 39722 }, { "epoch": 0.04, "learning_rate": 4.980652254794677e-05, "loss": 1.3095, "step": 39723 }, { "epoch": 0.04, "learning_rate": 4.980651279452407e-05, "loss": 0.9535, "step": 39724 }, { "epoch": 0.04, "learning_rate": 4.980650304085649e-05, "loss": 1.2392, "step": 39725 }, { "epoch": 0.04, "learning_rate": 4.980649328694403e-05, "loss": 1.1935, "step": 39726 }, { "epoch": 0.04, "learning_rate": 4.98064835327867e-05, "loss": 1.2554, "step": 39727 }, { "epoch": 0.04, "learning_rate": 4.9806473778384475e-05, "loss": 1.0312, "step": 39728 }, { "epoch": 0.04, "learning_rate": 4.980646402373739e-05, "loss": 0.7624, "step": 39729 }, { "epoch": 0.04, "learning_rate": 4.980645426884541e-05, "loss": 1.0175, "step": 39730 }, { "epoch": 0.04, "learning_rate": 4.9806444513708554e-05, "loss": 1.3839, "step": 39731 }, { "epoch": 0.04, "learning_rate": 4.980643475832682e-05, "loss": 1.2868, "step": 39732 }, { "epoch": 0.04, "learning_rate": 4.9806425002700205e-05, "loss": 1.2763, "step": 39733 }, { "epoch": 0.04, "learning_rate": 4.980641524682872e-05, "loss": 1.1918, "step": 39734 }, { "epoch": 0.04, "learning_rate": 4.9806405490712355e-05, "loss": 1.1339, "step": 39735 }, { "epoch": 0.04, "learning_rate": 4.980639573435111e-05, "loss": 0.9954, "step": 39736 }, { "epoch": 0.04, "learning_rate": 4.980638597774498e-05, "loss": 0.9955, "step": 39737 }, { "epoch": 0.04, "learning_rate": 4.980637622089398e-05, "loss": 1.1512, "step": 39738 }, { "epoch": 0.04, "learning_rate": 4.98063664637981e-05, "loss": 0.7312, "step": 39739 }, { "epoch": 0.04, "learning_rate": 4.980635670645734e-05, "loss": 0.7837, "step": 39740 }, { "epoch": 0.04, "learning_rate": 4.98063469488717e-05, "loss": 0.9313, "step": 39741 }, { "epoch": 0.04, "learning_rate": 4.9806337191041186e-05, "loss": 1.2674, "step": 39742 }, { "epoch": 0.04, "learning_rate": 4.9806327432965795e-05, "loss": 1.0388, "step": 39743 }, { "epoch": 0.04, "learning_rate": 4.9806317674645525e-05, "loss": 0.9276, "step": 39744 }, { "epoch": 0.04, "learning_rate": 4.980630791608037e-05, "loss": 0.9225, "step": 39745 }, { "epoch": 0.04, "learning_rate": 4.980629815727035e-05, "loss": 1.0504, "step": 39746 }, { "epoch": 0.04, "learning_rate": 4.980628839821545e-05, "loss": 1.074, "step": 39747 }, { "epoch": 0.04, "learning_rate": 4.980627863891567e-05, "loss": 0.9533, "step": 39748 }, { "epoch": 0.04, "learning_rate": 4.980626887937101e-05, "loss": 1.1157, "step": 39749 }, { "epoch": 0.04, "learning_rate": 4.9806259119581475e-05, "loss": 0.9622, "step": 39750 }, { "epoch": 0.04, "learning_rate": 4.9806249359547067e-05, "loss": 1.1137, "step": 39751 }, { "epoch": 0.04, "learning_rate": 4.980623959926778e-05, "loss": 1.2651, "step": 39752 }, { "epoch": 0.04, "learning_rate": 4.9806229838743615e-05, "loss": 1.3131, "step": 39753 }, { "epoch": 0.04, "learning_rate": 4.980622007797457e-05, "loss": 1.2567, "step": 39754 }, { "epoch": 0.04, "learning_rate": 4.9806210316960655e-05, "loss": 1.0471, "step": 39755 }, { "epoch": 0.04, "learning_rate": 4.9806200555701865e-05, "loss": 1.26, "step": 39756 }, { "epoch": 0.04, "learning_rate": 4.9806190794198193e-05, "loss": 0.6757, "step": 39757 }, { "epoch": 0.04, "learning_rate": 4.980618103244965e-05, "loss": 0.576, "step": 39758 }, { "epoch": 0.04, "learning_rate": 4.980617127045622e-05, "loss": 0.9231, "step": 39759 }, { "epoch": 0.04, "learning_rate": 4.980616150821792e-05, "loss": 1.0475, "step": 39760 }, { "epoch": 0.04, "learning_rate": 4.980615174573475e-05, "loss": 1.0177, "step": 39761 }, { "epoch": 0.04, "learning_rate": 4.98061419830067e-05, "loss": 1.0858, "step": 39762 }, { "epoch": 0.04, "learning_rate": 4.9806132220033774e-05, "loss": 1.0365, "step": 39763 }, { "epoch": 0.04, "learning_rate": 4.980612245681597e-05, "loss": 0.9833, "step": 39764 }, { "epoch": 0.04, "learning_rate": 4.980611269335329e-05, "loss": 0.9814, "step": 39765 }, { "epoch": 0.04, "learning_rate": 4.980610292964574e-05, "loss": 0.9961, "step": 39766 }, { "epoch": 0.04, "learning_rate": 4.980609316569331e-05, "loss": 0.8658, "step": 39767 }, { "epoch": 0.04, "learning_rate": 4.9806083401496007e-05, "loss": 0.9336, "step": 39768 }, { "epoch": 0.04, "learning_rate": 4.980607363705383e-05, "loss": 1.0476, "step": 39769 }, { "epoch": 0.04, "learning_rate": 4.980606387236678e-05, "loss": 1.0702, "step": 39770 }, { "epoch": 0.04, "learning_rate": 4.9806054107434845e-05, "loss": 0.7545, "step": 39771 }, { "epoch": 0.04, "learning_rate": 4.980604434225804e-05, "loss": 0.5542, "step": 39772 }, { "epoch": 0.04, "learning_rate": 4.9806034576836366e-05, "loss": 1.3612, "step": 39773 }, { "epoch": 0.04, "learning_rate": 4.980602481116981e-05, "loss": 1.0093, "step": 39774 }, { "epoch": 0.04, "learning_rate": 4.980601504525838e-05, "loss": 0.9272, "step": 39775 }, { "epoch": 0.04, "learning_rate": 4.980600527910208e-05, "loss": 1.1514, "step": 39776 }, { "epoch": 0.04, "learning_rate": 4.98059955127009e-05, "loss": 0.9447, "step": 39777 }, { "epoch": 0.04, "learning_rate": 4.980598574605485e-05, "loss": 0.6046, "step": 39778 }, { "epoch": 0.04, "learning_rate": 4.980597597916392e-05, "loss": 0.9977, "step": 39779 }, { "epoch": 0.04, "learning_rate": 4.980596621202812e-05, "loss": 0.7683, "step": 39780 }, { "epoch": 0.04, "learning_rate": 4.980595644464745e-05, "loss": 1.1903, "step": 39781 }, { "epoch": 0.04, "learning_rate": 4.98059466770219e-05, "loss": 1.1113, "step": 39782 }, { "epoch": 0.04, "learning_rate": 4.980593690915148e-05, "loss": 1.0493, "step": 39783 }, { "epoch": 0.04, "learning_rate": 4.980592714103619e-05, "loss": 1.0784, "step": 39784 }, { "epoch": 0.04, "learning_rate": 4.9805917372676015e-05, "loss": 0.9485, "step": 39785 }, { "epoch": 0.04, "learning_rate": 4.9805907604070975e-05, "loss": 1.2104, "step": 39786 }, { "epoch": 0.04, "learning_rate": 4.9805897835221055e-05, "loss": 1.325, "step": 39787 }, { "epoch": 0.04, "learning_rate": 4.980588806612627e-05, "loss": 1.2289, "step": 39788 }, { "epoch": 0.04, "learning_rate": 4.98058782967866e-05, "loss": 1.2455, "step": 39789 }, { "epoch": 0.04, "learning_rate": 4.9805868527202064e-05, "loss": 1.1528, "step": 39790 }, { "epoch": 0.04, "learning_rate": 4.980585875737266e-05, "loss": 1.0837, "step": 39791 }, { "epoch": 0.04, "learning_rate": 4.980584898729837e-05, "loss": 1.0239, "step": 39792 }, { "epoch": 0.04, "learning_rate": 4.980583921697922e-05, "loss": 1.4868, "step": 39793 }, { "epoch": 0.04, "learning_rate": 4.980582944641519e-05, "loss": 1.1437, "step": 39794 }, { "epoch": 0.04, "learning_rate": 4.980581967560629e-05, "loss": 1.1375, "step": 39795 }, { "epoch": 0.04, "learning_rate": 4.980580990455252e-05, "loss": 1.2235, "step": 39796 }, { "epoch": 0.04, "learning_rate": 4.980580013325388e-05, "loss": 1.2088, "step": 39797 }, { "epoch": 0.04, "learning_rate": 4.980579036171036e-05, "loss": 0.7889, "step": 39798 }, { "epoch": 0.04, "learning_rate": 4.9805780589921966e-05, "loss": 0.8961, "step": 39799 }, { "epoch": 0.04, "learning_rate": 4.98057708178887e-05, "loss": 0.4882, "step": 39800 }, { "epoch": 0.04, "learning_rate": 4.9805761045610566e-05, "loss": 1.0117, "step": 39801 }, { "epoch": 0.04, "learning_rate": 4.9805751273087565e-05, "loss": 1.1287, "step": 39802 }, { "epoch": 0.04, "learning_rate": 4.980574150031968e-05, "loss": 0.8852, "step": 39803 }, { "epoch": 0.04, "learning_rate": 4.980573172730693e-05, "loss": 1.0397, "step": 39804 }, { "epoch": 0.04, "learning_rate": 4.980572195404931e-05, "loss": 1.0921, "step": 39805 }, { "epoch": 0.04, "learning_rate": 4.980571218054682e-05, "loss": 1.281, "step": 39806 }, { "epoch": 0.04, "learning_rate": 4.980570240679945e-05, "loss": 1.1871, "step": 39807 }, { "epoch": 0.04, "learning_rate": 4.980569263280721e-05, "loss": 0.9497, "step": 39808 }, { "epoch": 0.04, "learning_rate": 4.98056828585701e-05, "loss": 1.1334, "step": 39809 }, { "epoch": 0.04, "learning_rate": 4.980567308408812e-05, "loss": 0.9606, "step": 39810 }, { "epoch": 0.04, "learning_rate": 4.980566330936127e-05, "loss": 1.1414, "step": 39811 }, { "epoch": 0.04, "learning_rate": 4.980565353438955e-05, "loss": 0.7484, "step": 39812 }, { "epoch": 0.04, "learning_rate": 4.980564375917296e-05, "loss": 0.6242, "step": 39813 }, { "epoch": 0.04, "learning_rate": 4.98056339837115e-05, "loss": 0.7956, "step": 39814 }, { "epoch": 0.04, "learning_rate": 4.9805624208005165e-05, "loss": 0.6091, "step": 39815 }, { "epoch": 0.04, "learning_rate": 4.980561443205396e-05, "loss": 0.4182, "step": 39816 }, { "epoch": 0.04, "learning_rate": 4.980560465585788e-05, "loss": 0.5593, "step": 39817 }, { "epoch": 0.04, "learning_rate": 4.980559487941694e-05, "loss": 0.9537, "step": 39818 }, { "epoch": 0.04, "learning_rate": 4.9805585102731126e-05, "loss": 0.9216, "step": 39819 }, { "epoch": 0.04, "learning_rate": 4.980557532580044e-05, "loss": 1.0556, "step": 39820 }, { "epoch": 0.04, "learning_rate": 4.980556554862488e-05, "loss": 0.9017, "step": 39821 }, { "epoch": 0.04, "learning_rate": 4.980555577120445e-05, "loss": 1.1458, "step": 39822 }, { "epoch": 0.04, "learning_rate": 4.980554599353915e-05, "loss": 0.3722, "step": 39823 }, { "epoch": 0.04, "learning_rate": 4.980553621562899e-05, "loss": 0.7794, "step": 39824 }, { "epoch": 0.04, "learning_rate": 4.980552643747395e-05, "loss": 0.983, "step": 39825 }, { "epoch": 0.04, "learning_rate": 4.9805516659074046e-05, "loss": 0.7596, "step": 39826 }, { "epoch": 0.04, "learning_rate": 4.980550688042927e-05, "loss": 1.1471, "step": 39827 }, { "epoch": 0.04, "learning_rate": 4.980549710153962e-05, "loss": 1.0333, "step": 39828 }, { "epoch": 0.04, "learning_rate": 4.980548732240511e-05, "loss": 1.338, "step": 39829 }, { "epoch": 0.04, "learning_rate": 4.980547754302573e-05, "loss": 1.0228, "step": 39830 }, { "epoch": 0.04, "learning_rate": 4.980546776340147e-05, "loss": 1.2134, "step": 39831 }, { "epoch": 0.04, "learning_rate": 4.9805457983532346e-05, "loss": 1.0184, "step": 39832 }, { "epoch": 0.04, "learning_rate": 4.980544820341836e-05, "loss": 0.7001, "step": 39833 }, { "epoch": 0.04, "learning_rate": 4.9805438423059496e-05, "loss": 0.9346, "step": 39834 }, { "epoch": 0.04, "learning_rate": 4.980542864245577e-05, "loss": 1.0392, "step": 39835 }, { "epoch": 0.04, "learning_rate": 4.9805418861607164e-05, "loss": 1.2257, "step": 39836 }, { "epoch": 0.04, "learning_rate": 4.98054090805137e-05, "loss": 1.232, "step": 39837 }, { "epoch": 0.04, "learning_rate": 4.980539929917537e-05, "loss": 0.9506, "step": 39838 }, { "epoch": 0.04, "learning_rate": 4.980538951759216e-05, "loss": 0.9749, "step": 39839 }, { "epoch": 0.04, "learning_rate": 4.980537973576409e-05, "loss": 1.703, "step": 39840 }, { "epoch": 0.04, "learning_rate": 4.980536995369115e-05, "loss": 1.1668, "step": 39841 }, { "epoch": 0.04, "learning_rate": 4.9805360171373346e-05, "loss": 1.026, "step": 39842 }, { "epoch": 0.04, "learning_rate": 4.9805350388810667e-05, "loss": 1.0058, "step": 39843 }, { "epoch": 0.04, "learning_rate": 4.980534060600313e-05, "loss": 1.246, "step": 39844 }, { "epoch": 0.04, "learning_rate": 4.9805330822950714e-05, "loss": 1.4204, "step": 39845 }, { "epoch": 0.04, "learning_rate": 4.980532103965343e-05, "loss": 1.0409, "step": 39846 }, { "epoch": 0.04, "learning_rate": 4.980531125611128e-05, "loss": 0.9456, "step": 39847 }, { "epoch": 0.04, "learning_rate": 4.9805301472324265e-05, "loss": 0.7395, "step": 39848 }, { "epoch": 0.04, "learning_rate": 4.980529168829239e-05, "loss": 0.7324, "step": 39849 }, { "epoch": 0.04, "learning_rate": 4.9805281904015636e-05, "loss": 0.9092, "step": 39850 }, { "epoch": 0.04, "learning_rate": 4.980527211949402e-05, "loss": 0.7404, "step": 39851 }, { "epoch": 0.04, "learning_rate": 4.980526233472753e-05, "loss": 0.6173, "step": 39852 }, { "epoch": 0.04, "learning_rate": 4.980525254971618e-05, "loss": 1.1041, "step": 39853 }, { "epoch": 0.04, "learning_rate": 4.980524276445996e-05, "loss": 1.1045, "step": 39854 }, { "epoch": 0.04, "learning_rate": 4.9805232978958874e-05, "loss": 1.0316, "step": 39855 }, { "epoch": 0.04, "learning_rate": 4.980522319321292e-05, "loss": 1.0486, "step": 39856 }, { "epoch": 0.04, "learning_rate": 4.9805213407222105e-05, "loss": 0.8043, "step": 39857 }, { "epoch": 0.04, "learning_rate": 4.980520362098642e-05, "loss": 0.5091, "step": 39858 }, { "epoch": 0.04, "learning_rate": 4.980519383450586e-05, "loss": 1.0089, "step": 39859 }, { "epoch": 0.04, "learning_rate": 4.980518404778045e-05, "loss": 1.1165, "step": 39860 }, { "epoch": 0.04, "learning_rate": 4.980517426081016e-05, "loss": 0.9318, "step": 39861 }, { "epoch": 0.04, "learning_rate": 4.980516447359501e-05, "loss": 1.3488, "step": 39862 }, { "epoch": 0.04, "learning_rate": 4.980515468613499e-05, "loss": 1.393, "step": 39863 }, { "epoch": 0.04, "learning_rate": 4.980514489843011e-05, "loss": 1.2074, "step": 39864 }, { "epoch": 0.04, "learning_rate": 4.980513511048036e-05, "loss": 0.8507, "step": 39865 }, { "epoch": 0.04, "learning_rate": 4.980512532228575e-05, "loss": 0.9746, "step": 39866 }, { "epoch": 0.04, "learning_rate": 4.9805115533846265e-05, "loss": 1.0322, "step": 39867 }, { "epoch": 0.04, "learning_rate": 4.9805105745161917e-05, "loss": 1.1499, "step": 39868 }, { "epoch": 0.04, "learning_rate": 4.980509595623271e-05, "loss": 0.809, "step": 39869 }, { "epoch": 0.04, "learning_rate": 4.980508616705863e-05, "loss": 1.0846, "step": 39870 }, { "epoch": 0.04, "learning_rate": 4.980507637763968e-05, "loss": 1.2244, "step": 39871 }, { "epoch": 0.04, "learning_rate": 4.980506658797588e-05, "loss": 1.4287, "step": 39872 }, { "epoch": 0.04, "learning_rate": 4.980505679806721e-05, "loss": 1.0867, "step": 39873 }, { "epoch": 0.04, "learning_rate": 4.980504700791366e-05, "loss": 1.0142, "step": 39874 }, { "epoch": 0.04, "learning_rate": 4.980503721751526e-05, "loss": 0.8707, "step": 39875 }, { "epoch": 0.04, "learning_rate": 4.9805027426871994e-05, "loss": 1.3318, "step": 39876 }, { "epoch": 0.04, "learning_rate": 4.980501763598386e-05, "loss": 1.0133, "step": 39877 }, { "epoch": 0.04, "learning_rate": 4.980500784485086e-05, "loss": 0.9691, "step": 39878 }, { "epoch": 0.04, "learning_rate": 4.9804998053472996e-05, "loss": 0.8251, "step": 39879 }, { "epoch": 0.04, "learning_rate": 4.980498826185027e-05, "loss": 0.4781, "step": 39880 }, { "epoch": 0.04, "learning_rate": 4.9804978469982677e-05, "loss": 1.0287, "step": 39881 }, { "epoch": 0.04, "learning_rate": 4.980496867787022e-05, "loss": 1.0725, "step": 39882 }, { "epoch": 0.04, "learning_rate": 4.980495888551291e-05, "loss": 0.9222, "step": 39883 }, { "epoch": 0.04, "learning_rate": 4.980494909291072e-05, "loss": 0.835, "step": 39884 }, { "epoch": 0.04, "learning_rate": 4.980493930006367e-05, "loss": 0.6236, "step": 39885 }, { "epoch": 0.04, "learning_rate": 4.9804929506971764e-05, "loss": 1.095, "step": 39886 }, { "epoch": 0.04, "learning_rate": 4.980491971363499e-05, "loss": 1.2357, "step": 39887 }, { "epoch": 0.04, "learning_rate": 4.9804909920053346e-05, "loss": 0.9702, "step": 39888 }, { "epoch": 0.04, "learning_rate": 4.9804900126226844e-05, "loss": 1.3954, "step": 39889 }, { "epoch": 0.04, "learning_rate": 4.9804890332155474e-05, "loss": 1.0667, "step": 39890 }, { "epoch": 0.04, "learning_rate": 4.980488053783925e-05, "loss": 1.1576, "step": 39891 }, { "epoch": 0.04, "learning_rate": 4.980487074327815e-05, "loss": 1.0314, "step": 39892 }, { "epoch": 0.04, "learning_rate": 4.9804860948472195e-05, "loss": 0.8599, "step": 39893 }, { "epoch": 0.04, "learning_rate": 4.980485115342138e-05, "loss": 0.8577, "step": 39894 }, { "epoch": 0.04, "learning_rate": 4.98048413581257e-05, "loss": 0.9501, "step": 39895 }, { "epoch": 0.04, "learning_rate": 4.9804831562585154e-05, "loss": 1.1824, "step": 39896 }, { "epoch": 0.04, "learning_rate": 4.980482176679974e-05, "loss": 1.1088, "step": 39897 }, { "epoch": 0.04, "learning_rate": 4.980481197076947e-05, "loss": 1.1328, "step": 39898 }, { "epoch": 0.04, "learning_rate": 4.980480217449434e-05, "loss": 0.8645, "step": 39899 }, { "epoch": 0.04, "learning_rate": 4.9804792377974344e-05, "loss": 1.2619, "step": 39900 }, { "epoch": 0.04, "learning_rate": 4.9804782581209484e-05, "loss": 1.0967, "step": 39901 }, { "epoch": 0.04, "learning_rate": 4.980477278419976e-05, "loss": 0.8888, "step": 39902 }, { "epoch": 0.04, "learning_rate": 4.9804762986945176e-05, "loss": 1.0697, "step": 39903 }, { "epoch": 0.04, "learning_rate": 4.980475318944573e-05, "loss": 1.0638, "step": 39904 }, { "epoch": 0.04, "learning_rate": 4.980474339170143e-05, "loss": 0.9915, "step": 39905 }, { "epoch": 0.04, "learning_rate": 4.980473359371225e-05, "loss": 1.1577, "step": 39906 }, { "epoch": 0.04, "learning_rate": 4.980472379547822e-05, "loss": 1.4436, "step": 39907 }, { "epoch": 0.04, "learning_rate": 4.980471399699933e-05, "loss": 2.2532, "step": 39908 }, { "epoch": 0.04, "learning_rate": 4.980470419827558e-05, "loss": 1.0512, "step": 39909 }, { "epoch": 0.04, "learning_rate": 4.980469439930696e-05, "loss": 0.8714, "step": 39910 }, { "epoch": 0.04, "learning_rate": 4.980468460009348e-05, "loss": 0.9124, "step": 39911 }, { "epoch": 0.04, "learning_rate": 4.980467480063514e-05, "loss": 0.9924, "step": 39912 }, { "epoch": 0.04, "learning_rate": 4.980466500093194e-05, "loss": 0.6864, "step": 39913 }, { "epoch": 0.04, "learning_rate": 4.9804655200983874e-05, "loss": 0.6734, "step": 39914 }, { "epoch": 0.04, "learning_rate": 4.9804645400790956e-05, "loss": 0.6543, "step": 39915 }, { "epoch": 0.04, "learning_rate": 4.980463560035317e-05, "loss": 0.8771, "step": 39916 }, { "epoch": 0.04, "learning_rate": 4.980462579967053e-05, "loss": 0.7412, "step": 39917 }, { "epoch": 0.04, "learning_rate": 4.980461599874302e-05, "loss": 0.8485, "step": 39918 }, { "epoch": 0.04, "learning_rate": 4.9804606197570656e-05, "loss": 1.0743, "step": 39919 }, { "epoch": 0.04, "learning_rate": 4.980459639615342e-05, "loss": 0.9703, "step": 39920 }, { "epoch": 0.04, "learning_rate": 4.9804586594491344e-05, "loss": 1.0341, "step": 39921 }, { "epoch": 0.04, "learning_rate": 4.980457679258439e-05, "loss": 1.0085, "step": 39922 }, { "epoch": 0.04, "learning_rate": 4.9804566990432584e-05, "loss": 1.0665, "step": 39923 }, { "epoch": 0.04, "learning_rate": 4.980455718803592e-05, "loss": 1.0677, "step": 39924 }, { "epoch": 0.04, "learning_rate": 4.980454738539438e-05, "loss": 1.0629, "step": 39925 }, { "epoch": 0.04, "learning_rate": 4.9804537582507995e-05, "loss": 0.875, "step": 39926 }, { "epoch": 0.04, "learning_rate": 4.980452777937675e-05, "loss": 0.8644, "step": 39927 }, { "epoch": 0.04, "learning_rate": 4.980451797600064e-05, "loss": 0.8147, "step": 39928 }, { "epoch": 0.04, "learning_rate": 4.980450817237967e-05, "loss": 1.1121, "step": 39929 }, { "epoch": 0.04, "learning_rate": 4.980449836851384e-05, "loss": 0.9854, "step": 39930 }, { "epoch": 0.04, "learning_rate": 4.980448856440315e-05, "loss": 0.9022, "step": 39931 }, { "epoch": 0.04, "learning_rate": 4.9804478760047604e-05, "loss": 0.9454, "step": 39932 }, { "epoch": 0.04, "learning_rate": 4.98044689554472e-05, "loss": 1.1218, "step": 39933 }, { "epoch": 0.04, "learning_rate": 4.980445915060193e-05, "loss": 0.8561, "step": 39934 }, { "epoch": 0.04, "learning_rate": 4.980444934551181e-05, "loss": 0.9451, "step": 39935 }, { "epoch": 0.04, "learning_rate": 4.980443954017683e-05, "loss": 0.9667, "step": 39936 }, { "epoch": 0.04, "learning_rate": 4.980442973459698e-05, "loss": 0.6333, "step": 39937 }, { "epoch": 0.04, "learning_rate": 4.980441992877228e-05, "loss": 0.9019, "step": 39938 }, { "epoch": 0.04, "learning_rate": 4.980441012270272e-05, "loss": 0.9139, "step": 39939 }, { "epoch": 0.04, "learning_rate": 4.98044003163883e-05, "loss": 1.0522, "step": 39940 }, { "epoch": 0.04, "learning_rate": 4.980439050982902e-05, "loss": 0.9855, "step": 39941 }, { "epoch": 0.04, "learning_rate": 4.980438070302489e-05, "loss": 1.1599, "step": 39942 }, { "epoch": 0.04, "learning_rate": 4.980437089597589e-05, "loss": 0.8775, "step": 39943 }, { "epoch": 0.04, "learning_rate": 4.980436108868204e-05, "loss": 1.0334, "step": 39944 }, { "epoch": 0.04, "learning_rate": 4.980435128114333e-05, "loss": 1.1788, "step": 39945 }, { "epoch": 0.04, "learning_rate": 4.980434147335976e-05, "loss": 0.8706, "step": 39946 }, { "epoch": 0.04, "learning_rate": 4.9804331665331334e-05, "loss": 0.9382, "step": 39947 }, { "epoch": 0.04, "learning_rate": 4.980432185705805e-05, "loss": 1.2384, "step": 39948 }, { "epoch": 0.04, "learning_rate": 4.98043120485399e-05, "loss": 0.4394, "step": 39949 }, { "epoch": 0.04, "learning_rate": 4.98043022397769e-05, "loss": 0.6767, "step": 39950 }, { "epoch": 0.04, "learning_rate": 4.9804292430769046e-05, "loss": 1.095, "step": 39951 }, { "epoch": 0.04, "learning_rate": 4.980428262151633e-05, "loss": 1.0734, "step": 39952 }, { "epoch": 0.04, "learning_rate": 4.980427281201876e-05, "loss": 1.1458, "step": 39953 }, { "epoch": 0.04, "learning_rate": 4.980426300227632e-05, "loss": 1.0335, "step": 39954 }, { "epoch": 0.04, "learning_rate": 4.980425319228904e-05, "loss": 0.9548, "step": 39955 }, { "epoch": 0.04, "learning_rate": 4.980424338205689e-05, "loss": 1.0335, "step": 39956 }, { "epoch": 0.04, "learning_rate": 4.9804233571579896e-05, "loss": 1.0044, "step": 39957 }, { "epoch": 0.04, "learning_rate": 4.9804223760858034e-05, "loss": 0.553, "step": 39958 }, { "epoch": 0.04, "learning_rate": 4.980421394989132e-05, "loss": 0.9166, "step": 39959 }, { "epoch": 0.04, "learning_rate": 4.980420413867974e-05, "loss": 0.8217, "step": 39960 }, { "epoch": 0.04, "learning_rate": 4.980419432722331e-05, "loss": 0.8729, "step": 39961 }, { "epoch": 0.04, "learning_rate": 4.980418451552203e-05, "loss": 1.0885, "step": 39962 }, { "epoch": 0.04, "learning_rate": 4.980417470357589e-05, "loss": 1.3459, "step": 39963 }, { "epoch": 0.04, "learning_rate": 4.980416489138489e-05, "loss": 1.2026, "step": 39964 }, { "epoch": 0.04, "learning_rate": 4.980415507894903e-05, "loss": 0.8631, "step": 39965 }, { "epoch": 0.04, "learning_rate": 4.980414526626832e-05, "loss": 1.1893, "step": 39966 }, { "epoch": 0.04, "learning_rate": 4.9804135453342756e-05, "loss": 1.1731, "step": 39967 }, { "epoch": 0.04, "learning_rate": 4.980412564017233e-05, "loss": 0.9397, "step": 39968 }, { "epoch": 0.04, "learning_rate": 4.980411582675705e-05, "loss": 1.0412, "step": 39969 }, { "epoch": 0.04, "learning_rate": 4.980410601309692e-05, "loss": 1.2476, "step": 39970 }, { "epoch": 0.04, "learning_rate": 4.980409619919193e-05, "loss": 1.2759, "step": 39971 }, { "epoch": 0.04, "learning_rate": 4.9804086385042084e-05, "loss": 1.0286, "step": 39972 }, { "epoch": 0.04, "learning_rate": 4.980407657064738e-05, "loss": 1.0243, "step": 39973 }, { "epoch": 0.04, "learning_rate": 4.9804066756007826e-05, "loss": 0.9346, "step": 39974 }, { "epoch": 0.04, "learning_rate": 4.980405694112341e-05, "loss": 0.8766, "step": 39975 }, { "epoch": 0.04, "learning_rate": 4.9804047125994147e-05, "loss": 0.5845, "step": 39976 }, { "epoch": 0.04, "learning_rate": 4.980403731062002e-05, "loss": 0.9896, "step": 39977 }, { "epoch": 0.04, "learning_rate": 4.980402749500104e-05, "loss": 0.9506, "step": 39978 }, { "epoch": 0.04, "learning_rate": 4.980401767913721e-05, "loss": 0.8245, "step": 39979 }, { "epoch": 0.04, "learning_rate": 4.9804007863028526e-05, "loss": 0.564, "step": 39980 }, { "epoch": 0.04, "learning_rate": 4.9803998046674986e-05, "loss": 0.5005, "step": 39981 }, { "epoch": 0.04, "learning_rate": 4.9803988230076585e-05, "loss": 0.9579, "step": 39982 }, { "epoch": 0.04, "learning_rate": 4.980397841323333e-05, "loss": 0.5331, "step": 39983 }, { "epoch": 0.04, "learning_rate": 4.980396859614523e-05, "loss": 0.4729, "step": 39984 }, { "epoch": 0.04, "learning_rate": 4.980395877881227e-05, "loss": 0.65, "step": 39985 }, { "epoch": 0.04, "learning_rate": 4.980394896123446e-05, "loss": 0.3968, "step": 39986 }, { "epoch": 0.04, "learning_rate": 4.9803939143411785e-05, "loss": 0.5404, "step": 39987 }, { "epoch": 0.04, "learning_rate": 4.980392932534427e-05, "loss": 0.618, "step": 39988 }, { "epoch": 0.04, "learning_rate": 4.9803919507031894e-05, "loss": 0.3823, "step": 39989 }, { "epoch": 0.04, "learning_rate": 4.980390968847466e-05, "loss": 0.5021, "step": 39990 }, { "epoch": 0.04, "learning_rate": 4.9803899869672576e-05, "loss": 0.6555, "step": 39991 }, { "epoch": 0.04, "learning_rate": 4.980389005062563e-05, "loss": 0.8617, "step": 39992 }, { "epoch": 0.04, "learning_rate": 4.9803880231333844e-05, "loss": 0.8095, "step": 39993 }, { "epoch": 0.04, "learning_rate": 4.980387041179721e-05, "loss": 0.851, "step": 39994 }, { "epoch": 0.04, "learning_rate": 4.9803860592015704e-05, "loss": 0.7962, "step": 39995 }, { "epoch": 0.04, "learning_rate": 4.9803850771989354e-05, "loss": 0.7576, "step": 39996 }, { "epoch": 0.04, "learning_rate": 4.980384095171815e-05, "loss": 0.7449, "step": 39997 }, { "epoch": 0.04, "learning_rate": 4.9803831131202094e-05, "loss": 0.8515, "step": 39998 }, { "epoch": 0.04, "learning_rate": 4.980382131044118e-05, "loss": 1.0649, "step": 39999 }, { "epoch": 0.04, "learning_rate": 4.980381148943542e-05, "loss": 0.9707, "step": 40000 }, { "epoch": 0.04, "eval_loss": 1.1041456460952759, "eval_runtime": 126.6089, "eval_samples_per_second": 10.939, "eval_steps_per_second": 5.474, "step": 40000 }, { "epoch": 0.04, "learning_rate": 4.98038016681848e-05, "loss": 0.8351, "step": 40001 }, { "epoch": 0.04, "learning_rate": 4.980379184668934e-05, "loss": 1.0063, "step": 40002 }, { "epoch": 0.04, "learning_rate": 4.980378202494901e-05, "loss": 1.0335, "step": 40003 }, { "epoch": 0.04, "learning_rate": 4.980377220296384e-05, "loss": 0.9556, "step": 40004 }, { "epoch": 0.04, "learning_rate": 4.980376238073382e-05, "loss": 0.969, "step": 40005 }, { "epoch": 0.04, "learning_rate": 4.980375255825894e-05, "loss": 1.1269, "step": 40006 }, { "epoch": 0.04, "learning_rate": 4.980374273553921e-05, "loss": 1.0262, "step": 40007 }, { "epoch": 0.04, "learning_rate": 4.980373291257462e-05, "loss": 0.6993, "step": 40008 }, { "epoch": 0.04, "learning_rate": 4.980372308936519e-05, "loss": 0.7884, "step": 40009 }, { "epoch": 0.04, "learning_rate": 4.98037132659109e-05, "loss": 1.1212, "step": 40010 }, { "epoch": 0.04, "learning_rate": 4.9803703442211765e-05, "loss": 1.0038, "step": 40011 }, { "epoch": 0.04, "learning_rate": 4.980369361826778e-05, "loss": 0.8559, "step": 40012 }, { "epoch": 0.04, "learning_rate": 4.980368379407893e-05, "loss": 1.1532, "step": 40013 }, { "epoch": 0.04, "learning_rate": 4.9803673969645234e-05, "loss": 1.2441, "step": 40014 }, { "epoch": 0.04, "learning_rate": 4.9803664144966694e-05, "loss": 0.9196, "step": 40015 }, { "epoch": 0.04, "learning_rate": 4.98036543200433e-05, "loss": 0.8118, "step": 40016 }, { "epoch": 0.04, "learning_rate": 4.9803644494875055e-05, "loss": 0.8715, "step": 40017 }, { "epoch": 0.04, "learning_rate": 4.9803634669461955e-05, "loss": 1.2817, "step": 40018 }, { "epoch": 0.04, "learning_rate": 4.980362484380401e-05, "loss": 0.8654, "step": 40019 }, { "epoch": 0.04, "learning_rate": 4.980361501790121e-05, "loss": 1.3229, "step": 40020 }, { "epoch": 0.04, "learning_rate": 4.9803605191753554e-05, "loss": 1.1641, "step": 40021 }, { "epoch": 0.04, "learning_rate": 4.980359536536105e-05, "loss": 1.0475, "step": 40022 }, { "epoch": 0.04, "learning_rate": 4.98035855387237e-05, "loss": 1.2802, "step": 40023 }, { "epoch": 0.04, "learning_rate": 4.98035757118415e-05, "loss": 1.2998, "step": 40024 }, { "epoch": 0.04, "learning_rate": 4.9803565884714444e-05, "loss": 0.9992, "step": 40025 }, { "epoch": 0.04, "learning_rate": 4.9803556057342536e-05, "loss": 1.0157, "step": 40026 }, { "epoch": 0.04, "learning_rate": 4.980354622972578e-05, "loss": 0.9564, "step": 40027 }, { "epoch": 0.04, "learning_rate": 4.980353640186418e-05, "loss": 0.7224, "step": 40028 }, { "epoch": 0.04, "learning_rate": 4.980352657375773e-05, "loss": 0.9645, "step": 40029 }, { "epoch": 0.04, "learning_rate": 4.980351674540642e-05, "loss": 1.0334, "step": 40030 }, { "epoch": 0.04, "learning_rate": 4.980350691681027e-05, "loss": 0.7898, "step": 40031 }, { "epoch": 0.04, "learning_rate": 4.980349708796926e-05, "loss": 0.9632, "step": 40032 }, { "epoch": 0.04, "learning_rate": 4.980348725888341e-05, "loss": 0.9754, "step": 40033 }, { "epoch": 0.04, "learning_rate": 4.980347742955271e-05, "loss": 0.6593, "step": 40034 }, { "epoch": 0.04, "learning_rate": 4.980346759997716e-05, "loss": 1.0488, "step": 40035 }, { "epoch": 0.04, "learning_rate": 4.980345777015676e-05, "loss": 0.7422, "step": 40036 }, { "epoch": 0.04, "learning_rate": 4.980344794009151e-05, "loss": 0.8106, "step": 40037 }, { "epoch": 0.04, "learning_rate": 4.98034381097814e-05, "loss": 1.2368, "step": 40038 }, { "epoch": 0.04, "learning_rate": 4.980342827922645e-05, "loss": 1.4599, "step": 40039 }, { "epoch": 0.04, "learning_rate": 4.9803418448426656e-05, "loss": 0.9751, "step": 40040 }, { "epoch": 0.04, "learning_rate": 4.980340861738201e-05, "loss": 0.9305, "step": 40041 }, { "epoch": 0.04, "learning_rate": 4.980339878609252e-05, "loss": 1.2623, "step": 40042 }, { "epoch": 0.04, "learning_rate": 4.980338895455817e-05, "loss": 1.3362, "step": 40043 }, { "epoch": 0.04, "learning_rate": 4.980337912277897e-05, "loss": 1.8253, "step": 40044 }, { "epoch": 0.04, "learning_rate": 4.980336929075493e-05, "loss": 2.0186, "step": 40045 }, { "epoch": 0.04, "learning_rate": 4.9803359458486045e-05, "loss": 1.986, "step": 40046 }, { "epoch": 0.04, "learning_rate": 4.980334962597231e-05, "loss": 0.9248, "step": 40047 }, { "epoch": 0.04, "learning_rate": 4.980333979321372e-05, "loss": 0.5971, "step": 40048 }, { "epoch": 0.04, "learning_rate": 4.980332996021028e-05, "loss": 0.4241, "step": 40049 }, { "epoch": 0.04, "learning_rate": 4.980332012696199e-05, "loss": 0.7349, "step": 40050 }, { "epoch": 0.04, "learning_rate": 4.980331029346886e-05, "loss": 0.9617, "step": 40051 }, { "epoch": 0.04, "learning_rate": 4.980330045973089e-05, "loss": 1.1847, "step": 40052 }, { "epoch": 0.04, "learning_rate": 4.980329062574806e-05, "loss": 1.2285, "step": 40053 }, { "epoch": 0.04, "learning_rate": 4.980328079152039e-05, "loss": 0.6677, "step": 40054 }, { "epoch": 0.04, "learning_rate": 4.9803270957047866e-05, "loss": 1.1706, "step": 40055 }, { "epoch": 0.04, "learning_rate": 4.9803261122330495e-05, "loss": 0.8298, "step": 40056 }, { "epoch": 0.04, "learning_rate": 4.980325128736828e-05, "loss": 1.5489, "step": 40057 }, { "epoch": 0.04, "learning_rate": 4.980324145216122e-05, "loss": 1.0558, "step": 40058 }, { "epoch": 0.04, "learning_rate": 4.98032316167093e-05, "loss": 0.866, "step": 40059 }, { "epoch": 0.04, "learning_rate": 4.980322178101255e-05, "loss": 1.0489, "step": 40060 }, { "epoch": 0.04, "learning_rate": 4.980321194507094e-05, "loss": 1.2543, "step": 40061 }, { "epoch": 0.04, "learning_rate": 4.9803202108884496e-05, "loss": 1.102, "step": 40062 }, { "epoch": 0.04, "learning_rate": 4.980319227245319e-05, "loss": 0.9019, "step": 40063 }, { "epoch": 0.04, "learning_rate": 4.9803182435777054e-05, "loss": 0.8962, "step": 40064 }, { "epoch": 0.04, "learning_rate": 4.9803172598856056e-05, "loss": 0.8035, "step": 40065 }, { "epoch": 0.04, "learning_rate": 4.980316276169022e-05, "loss": 1.1187, "step": 40066 }, { "epoch": 0.04, "learning_rate": 4.980315292427953e-05, "loss": 0.8389, "step": 40067 }, { "epoch": 0.04, "learning_rate": 4.9803143086624e-05, "loss": 1.0741, "step": 40068 }, { "epoch": 0.04, "learning_rate": 4.980313324872363e-05, "loss": 1.0117, "step": 40069 }, { "epoch": 0.04, "learning_rate": 4.98031234105784e-05, "loss": 1.0591, "step": 40070 }, { "epoch": 0.04, "learning_rate": 4.980311357218833e-05, "loss": 0.9564, "step": 40071 }, { "epoch": 0.04, "learning_rate": 4.980310373355342e-05, "loss": 0.5827, "step": 40072 }, { "epoch": 0.04, "learning_rate": 4.9803093894673656e-05, "loss": 1.1534, "step": 40073 }, { "epoch": 0.04, "learning_rate": 4.980308405554905e-05, "loss": 0.7235, "step": 40074 }, { "epoch": 0.04, "learning_rate": 4.98030742161796e-05, "loss": 0.4214, "step": 40075 }, { "epoch": 0.04, "learning_rate": 4.98030643765653e-05, "loss": 0.5204, "step": 40076 }, { "epoch": 0.04, "learning_rate": 4.980305453670616e-05, "loss": 0.663, "step": 40077 }, { "epoch": 0.04, "learning_rate": 4.980304469660218e-05, "loss": 0.6855, "step": 40078 }, { "epoch": 0.04, "learning_rate": 4.980303485625334e-05, "loss": 1.1267, "step": 40079 }, { "epoch": 0.04, "learning_rate": 4.980302501565966e-05, "loss": 1.0533, "step": 40080 }, { "epoch": 0.04, "learning_rate": 4.9803015174821134e-05, "loss": 1.3505, "step": 40081 }, { "epoch": 0.04, "learning_rate": 4.980300533373777e-05, "loss": 1.1863, "step": 40082 }, { "epoch": 0.04, "learning_rate": 4.9802995492409557e-05, "loss": 1.0581, "step": 40083 }, { "epoch": 0.04, "learning_rate": 4.98029856508365e-05, "loss": 0.6891, "step": 40084 }, { "epoch": 0.04, "learning_rate": 4.980297580901859e-05, "loss": 1.1461, "step": 40085 }, { "epoch": 0.04, "learning_rate": 4.980296596695585e-05, "loss": 1.0273, "step": 40086 }, { "epoch": 0.04, "learning_rate": 4.980295612464826e-05, "loss": 1.0165, "step": 40087 }, { "epoch": 0.04, "learning_rate": 4.980294628209582e-05, "loss": 1.0943, "step": 40088 }, { "epoch": 0.04, "learning_rate": 4.980293643929854e-05, "loss": 1.0806, "step": 40089 }, { "epoch": 0.04, "learning_rate": 4.980292659625642e-05, "loss": 0.5791, "step": 40090 }, { "epoch": 0.04, "learning_rate": 4.9802916752969445e-05, "loss": 0.8668, "step": 40091 }, { "epoch": 0.04, "learning_rate": 4.980290690943763e-05, "loss": 0.8294, "step": 40092 }, { "epoch": 0.04, "learning_rate": 4.980289706566098e-05, "loss": 0.6656, "step": 40093 }, { "epoch": 0.04, "learning_rate": 4.980288722163948e-05, "loss": 0.5729, "step": 40094 }, { "epoch": 0.04, "learning_rate": 4.9802877377373135e-05, "loss": 0.6142, "step": 40095 }, { "epoch": 0.04, "learning_rate": 4.980286753286195e-05, "loss": 1.0247, "step": 40096 }, { "epoch": 0.04, "learning_rate": 4.980285768810592e-05, "loss": 0.7716, "step": 40097 }, { "epoch": 0.04, "learning_rate": 4.9802847843105044e-05, "loss": 1.0167, "step": 40098 }, { "epoch": 0.04, "learning_rate": 4.980283799785933e-05, "loss": 1.02, "step": 40099 }, { "epoch": 0.04, "learning_rate": 4.9802828152368766e-05, "loss": 1.0109, "step": 40100 }, { "epoch": 0.04, "learning_rate": 4.980281830663337e-05, "loss": 1.0328, "step": 40101 }, { "epoch": 0.04, "learning_rate": 4.9802808460653114e-05, "loss": 1.1585, "step": 40102 }, { "epoch": 0.04, "learning_rate": 4.980279861442803e-05, "loss": 1.1695, "step": 40103 }, { "epoch": 0.04, "learning_rate": 4.9802788767958096e-05, "loss": 0.2007, "step": 40104 }, { "epoch": 0.04, "learning_rate": 4.980277892124332e-05, "loss": 0.1856, "step": 40105 }, { "epoch": 0.04, "learning_rate": 4.9802769074283704e-05, "loss": 1.1141, "step": 40106 }, { "epoch": 0.04, "learning_rate": 4.9802759227079245e-05, "loss": 0.9583, "step": 40107 }, { "epoch": 0.04, "learning_rate": 4.9802749379629946e-05, "loss": 0.9166, "step": 40108 }, { "epoch": 0.04, "learning_rate": 4.980273953193579e-05, "loss": 1.0047, "step": 40109 }, { "epoch": 0.04, "learning_rate": 4.980272968399681e-05, "loss": 1.1154, "step": 40110 }, { "epoch": 0.04, "learning_rate": 4.980271983581298e-05, "loss": 0.9844, "step": 40111 }, { "epoch": 0.04, "learning_rate": 4.980270998738431e-05, "loss": 1.0784, "step": 40112 }, { "epoch": 0.04, "learning_rate": 4.98027001387108e-05, "loss": 1.011, "step": 40113 }, { "epoch": 0.04, "learning_rate": 4.9802690289792444e-05, "loss": 1.1918, "step": 40114 }, { "epoch": 0.04, "learning_rate": 4.9802680440629245e-05, "loss": 0.8127, "step": 40115 }, { "epoch": 0.04, "learning_rate": 4.980267059122121e-05, "loss": 0.7379, "step": 40116 }, { "epoch": 0.04, "learning_rate": 4.980266074156833e-05, "loss": 0.9316, "step": 40117 }, { "epoch": 0.04, "learning_rate": 4.980265089167061e-05, "loss": 0.986, "step": 40118 }, { "epoch": 0.04, "learning_rate": 4.9802641041528056e-05, "loss": 0.8, "step": 40119 }, { "epoch": 0.04, "learning_rate": 4.980263119114065e-05, "loss": 0.8796, "step": 40120 }, { "epoch": 0.04, "learning_rate": 4.9802621340508404e-05, "loss": 0.9183, "step": 40121 }, { "epoch": 0.04, "learning_rate": 4.980261148963132e-05, "loss": 1.052, "step": 40122 }, { "epoch": 0.04, "learning_rate": 4.980260163850939e-05, "loss": 0.9185, "step": 40123 }, { "epoch": 0.04, "learning_rate": 4.980259178714263e-05, "loss": 1.1649, "step": 40124 }, { "epoch": 0.04, "learning_rate": 4.9802581935531015e-05, "loss": 0.9559, "step": 40125 }, { "epoch": 0.04, "learning_rate": 4.980257208367457e-05, "loss": 1.0646, "step": 40126 }, { "epoch": 0.04, "learning_rate": 4.980256223157328e-05, "loss": 0.8924, "step": 40127 }, { "epoch": 0.04, "learning_rate": 4.980255237922715e-05, "loss": 0.7681, "step": 40128 }, { "epoch": 0.04, "learning_rate": 4.980254252663617e-05, "loss": 0.9565, "step": 40129 }, { "epoch": 0.04, "learning_rate": 4.9802532673800375e-05, "loss": 1.1585, "step": 40130 }, { "epoch": 0.04, "learning_rate": 4.980252282071972e-05, "loss": 1.2575, "step": 40131 }, { "epoch": 0.04, "learning_rate": 4.980251296739423e-05, "loss": 0.9026, "step": 40132 }, { "epoch": 0.04, "learning_rate": 4.980250311382389e-05, "loss": 1.2882, "step": 40133 }, { "epoch": 0.04, "learning_rate": 4.9802493260008734e-05, "loss": 1.1957, "step": 40134 }, { "epoch": 0.04, "learning_rate": 4.980248340594872e-05, "loss": 0.679, "step": 40135 }, { "epoch": 0.04, "learning_rate": 4.980247355164387e-05, "loss": 0.8389, "step": 40136 }, { "epoch": 0.04, "learning_rate": 4.980246369709418e-05, "loss": 0.6929, "step": 40137 }, { "epoch": 0.04, "learning_rate": 4.9802453842299654e-05, "loss": 1.0598, "step": 40138 }, { "epoch": 0.04, "learning_rate": 4.980244398726028e-05, "loss": 1.0423, "step": 40139 }, { "epoch": 0.04, "learning_rate": 4.980243413197607e-05, "loss": 0.6124, "step": 40140 }, { "epoch": 0.04, "learning_rate": 4.980242427644703e-05, "loss": 0.3361, "step": 40141 }, { "epoch": 0.04, "learning_rate": 4.980241442067314e-05, "loss": 0.9174, "step": 40142 }, { "epoch": 0.04, "learning_rate": 4.980240456465442e-05, "loss": 1.2573, "step": 40143 }, { "epoch": 0.04, "learning_rate": 4.980239470839085e-05, "loss": 1.1555, "step": 40144 }, { "epoch": 0.04, "learning_rate": 4.9802384851882454e-05, "loss": 0.9096, "step": 40145 }, { "epoch": 0.04, "learning_rate": 4.9802374995129216e-05, "loss": 0.8689, "step": 40146 }, { "epoch": 0.04, "learning_rate": 4.980236513813114e-05, "loss": 0.8914, "step": 40147 }, { "epoch": 0.04, "learning_rate": 4.9802355280888214e-05, "loss": 0.3023, "step": 40148 }, { "epoch": 0.04, "learning_rate": 4.9802345423400456e-05, "loss": 0.8141, "step": 40149 }, { "epoch": 0.04, "learning_rate": 4.9802335565667865e-05, "loss": 0.8951, "step": 40150 }, { "epoch": 0.04, "learning_rate": 4.9802325707690435e-05, "loss": 1.0116, "step": 40151 }, { "epoch": 0.04, "learning_rate": 4.9802315849468164e-05, "loss": 0.622, "step": 40152 }, { "epoch": 0.04, "learning_rate": 4.980230599100105e-05, "loss": 0.9495, "step": 40153 }, { "epoch": 0.04, "learning_rate": 4.98022961322891e-05, "loss": 0.8977, "step": 40154 }, { "epoch": 0.04, "learning_rate": 4.980228627333232e-05, "loss": 0.8625, "step": 40155 }, { "epoch": 0.04, "learning_rate": 4.98022764141307e-05, "loss": 1.3008, "step": 40156 }, { "epoch": 0.04, "learning_rate": 4.980226655468424e-05, "loss": 1.2325, "step": 40157 }, { "epoch": 0.04, "learning_rate": 4.980225669499294e-05, "loss": 1.0094, "step": 40158 }, { "epoch": 0.04, "learning_rate": 4.980224683505681e-05, "loss": 0.9082, "step": 40159 }, { "epoch": 0.04, "learning_rate": 4.980223697487584e-05, "loss": 1.0506, "step": 40160 }, { "epoch": 0.04, "learning_rate": 4.980222711445003e-05, "loss": 0.9291, "step": 40161 }, { "epoch": 0.04, "learning_rate": 4.980221725377938e-05, "loss": 1.2304, "step": 40162 }, { "epoch": 0.04, "learning_rate": 4.98022073928639e-05, "loss": 1.1269, "step": 40163 }, { "epoch": 0.04, "learning_rate": 4.9802197531703575e-05, "loss": 1.0724, "step": 40164 }, { "epoch": 0.04, "learning_rate": 4.980218767029842e-05, "loss": 0.9201, "step": 40165 }, { "epoch": 0.04, "learning_rate": 4.980217780864842e-05, "loss": 0.9911, "step": 40166 }, { "epoch": 0.04, "learning_rate": 4.980216794675359e-05, "loss": 1.0427, "step": 40167 }, { "epoch": 0.04, "learning_rate": 4.980215808461392e-05, "loss": 0.8543, "step": 40168 }, { "epoch": 0.04, "learning_rate": 4.980214822222942e-05, "loss": 1.3675, "step": 40169 }, { "epoch": 0.04, "learning_rate": 4.980213835960008e-05, "loss": 0.901, "step": 40170 }, { "epoch": 0.04, "learning_rate": 4.980212849672591e-05, "loss": 0.9215, "step": 40171 }, { "epoch": 0.04, "learning_rate": 4.9802118633606894e-05, "loss": 1.0075, "step": 40172 }, { "epoch": 0.04, "learning_rate": 4.9802108770243045e-05, "loss": 0.9186, "step": 40173 }, { "epoch": 0.04, "learning_rate": 4.9802098906634356e-05, "loss": 0.93, "step": 40174 }, { "epoch": 0.04, "learning_rate": 4.980208904278084e-05, "loss": 1.2567, "step": 40175 }, { "epoch": 0.04, "learning_rate": 4.980207917868248e-05, "loss": 1.186, "step": 40176 }, { "epoch": 0.04, "learning_rate": 4.980206931433929e-05, "loss": 1.2373, "step": 40177 }, { "epoch": 0.04, "learning_rate": 4.980205944975126e-05, "loss": 1.1856, "step": 40178 }, { "epoch": 0.04, "learning_rate": 4.9802049584918394e-05, "loss": 1.1495, "step": 40179 }, { "epoch": 0.04, "learning_rate": 4.98020397198407e-05, "loss": 1.2619, "step": 40180 }, { "epoch": 0.04, "learning_rate": 4.9802029854518165e-05, "loss": 0.9719, "step": 40181 }, { "epoch": 0.04, "learning_rate": 4.98020199889508e-05, "loss": 0.9717, "step": 40182 }, { "epoch": 0.04, "learning_rate": 4.9802010123138596e-05, "loss": 1.1512, "step": 40183 }, { "epoch": 0.04, "learning_rate": 4.9802000257081556e-05, "loss": 0.8778, "step": 40184 }, { "epoch": 0.04, "learning_rate": 4.980199039077968e-05, "loss": 0.8933, "step": 40185 }, { "epoch": 0.04, "learning_rate": 4.9801980524232974e-05, "loss": 1.3386, "step": 40186 }, { "epoch": 0.04, "learning_rate": 4.9801970657441434e-05, "loss": 0.9141, "step": 40187 }, { "epoch": 0.04, "learning_rate": 4.980196079040505e-05, "loss": 1.013, "step": 40188 }, { "epoch": 0.04, "learning_rate": 4.980195092312384e-05, "loss": 0.8862, "step": 40189 }, { "epoch": 0.04, "learning_rate": 4.9801941055597786e-05, "loss": 0.9935, "step": 40190 }, { "epoch": 0.04, "learning_rate": 4.980193118782691e-05, "loss": 0.8461, "step": 40191 }, { "epoch": 0.04, "learning_rate": 4.980192131981119e-05, "loss": 0.7871, "step": 40192 }, { "epoch": 0.04, "learning_rate": 4.980191145155065e-05, "loss": 1.0288, "step": 40193 }, { "epoch": 0.04, "learning_rate": 4.980190158304526e-05, "loss": 1.2097, "step": 40194 }, { "epoch": 0.04, "learning_rate": 4.980189171429505e-05, "loss": 1.0057, "step": 40195 }, { "epoch": 0.04, "learning_rate": 4.980188184529999e-05, "loss": 0.9988, "step": 40196 }, { "epoch": 0.04, "learning_rate": 4.980187197606011e-05, "loss": 1.1743, "step": 40197 }, { "epoch": 0.04, "learning_rate": 4.980186210657539e-05, "loss": 1.2058, "step": 40198 }, { "epoch": 0.04, "learning_rate": 4.980185223684584e-05, "loss": 1.0773, "step": 40199 }, { "epoch": 0.04, "learning_rate": 4.980184236687145e-05, "loss": 1.2282, "step": 40200 }, { "epoch": 0.04, "learning_rate": 4.9801832496652235e-05, "loss": 0.8915, "step": 40201 }, { "epoch": 0.04, "learning_rate": 4.980182262618818e-05, "loss": 1.1978, "step": 40202 }, { "epoch": 0.04, "learning_rate": 4.980181275547929e-05, "loss": 1.1206, "step": 40203 }, { "epoch": 0.04, "learning_rate": 4.980180288452558e-05, "loss": 0.9227, "step": 40204 }, { "epoch": 0.04, "learning_rate": 4.980179301332702e-05, "loss": 0.8176, "step": 40205 }, { "epoch": 0.04, "learning_rate": 4.980178314188364e-05, "loss": 1.083, "step": 40206 }, { "epoch": 0.04, "learning_rate": 4.980177327019543e-05, "loss": 1.114, "step": 40207 }, { "epoch": 0.04, "learning_rate": 4.980176339826238e-05, "loss": 0.8839, "step": 40208 }, { "epoch": 0.04, "learning_rate": 4.980175352608449e-05, "loss": 1.0438, "step": 40209 }, { "epoch": 0.04, "learning_rate": 4.980174365366178e-05, "loss": 0.5715, "step": 40210 }, { "epoch": 0.04, "learning_rate": 4.980173378099423e-05, "loss": 1.7546, "step": 40211 }, { "epoch": 0.04, "learning_rate": 4.9801723908081855e-05, "loss": 0.8007, "step": 40212 }, { "epoch": 0.04, "learning_rate": 4.980171403492464e-05, "loss": 0.7488, "step": 40213 }, { "epoch": 0.04, "learning_rate": 4.98017041615226e-05, "loss": 1.1464, "step": 40214 }, { "epoch": 0.04, "learning_rate": 4.980169428787572e-05, "loss": 0.8603, "step": 40215 }, { "epoch": 0.04, "learning_rate": 4.9801684413984016e-05, "loss": 1.2161, "step": 40216 }, { "epoch": 0.04, "learning_rate": 4.9801674539847474e-05, "loss": 1.1111, "step": 40217 }, { "epoch": 0.04, "learning_rate": 4.9801664665466105e-05, "loss": 1.0981, "step": 40218 }, { "epoch": 0.04, "learning_rate": 4.98016547908399e-05, "loss": 1.1458, "step": 40219 }, { "epoch": 0.04, "learning_rate": 4.980164491596887e-05, "loss": 1.1462, "step": 40220 }, { "epoch": 0.04, "learning_rate": 4.9801635040853006e-05, "loss": 1.4483, "step": 40221 }, { "epoch": 0.04, "learning_rate": 4.9801625165492304e-05, "loss": 1.0551, "step": 40222 }, { "epoch": 0.04, "learning_rate": 4.980161528988678e-05, "loss": 1.0357, "step": 40223 }, { "epoch": 0.04, "learning_rate": 4.980160541403642e-05, "loss": 1.1434, "step": 40224 }, { "epoch": 0.04, "learning_rate": 4.9801595537941234e-05, "loss": 1.1042, "step": 40225 }, { "epoch": 0.04, "learning_rate": 4.980158566160121e-05, "loss": 1.0783, "step": 40226 }, { "epoch": 0.04, "learning_rate": 4.9801575785016365e-05, "loss": 0.8386, "step": 40227 }, { "epoch": 0.04, "learning_rate": 4.980156590818668e-05, "loss": 0.655, "step": 40228 }, { "epoch": 0.04, "learning_rate": 4.9801556031112165e-05, "loss": 0.9964, "step": 40229 }, { "epoch": 0.04, "learning_rate": 4.9801546153792825e-05, "loss": 0.7931, "step": 40230 }, { "epoch": 0.04, "learning_rate": 4.980153627622865e-05, "loss": 0.5834, "step": 40231 }, { "epoch": 0.04, "learning_rate": 4.9801526398419645e-05, "loss": 1.0704, "step": 40232 }, { "epoch": 0.04, "learning_rate": 4.980151652036581e-05, "loss": 1.0943, "step": 40233 }, { "epoch": 0.04, "learning_rate": 4.9801506642067146e-05, "loss": 0.7844, "step": 40234 }, { "epoch": 0.04, "learning_rate": 4.9801496763523654e-05, "loss": 0.8188, "step": 40235 }, { "epoch": 0.04, "learning_rate": 4.980148688473533e-05, "loss": 1.0977, "step": 40236 }, { "epoch": 0.04, "learning_rate": 4.9801477005702176e-05, "loss": 0.681, "step": 40237 }, { "epoch": 0.04, "learning_rate": 4.980146712642419e-05, "loss": 0.8482, "step": 40238 }, { "epoch": 0.04, "learning_rate": 4.980145724690137e-05, "loss": 1.202, "step": 40239 }, { "epoch": 0.04, "learning_rate": 4.9801447367133735e-05, "loss": 1.2455, "step": 40240 }, { "epoch": 0.04, "learning_rate": 4.9801437487121264e-05, "loss": 0.9425, "step": 40241 }, { "epoch": 0.04, "learning_rate": 4.980142760686396e-05, "loss": 0.6116, "step": 40242 }, { "epoch": 0.04, "learning_rate": 4.980141772636183e-05, "loss": 1.1579, "step": 40243 }, { "epoch": 0.04, "learning_rate": 4.9801407845614865e-05, "loss": 1.0084, "step": 40244 }, { "epoch": 0.04, "learning_rate": 4.9801397964623074e-05, "loss": 0.8425, "step": 40245 }, { "epoch": 0.04, "learning_rate": 4.980138808338646e-05, "loss": 1.0874, "step": 40246 }, { "epoch": 0.04, "learning_rate": 4.9801378201905015e-05, "loss": 1.0933, "step": 40247 }, { "epoch": 0.04, "learning_rate": 4.980136832017874e-05, "loss": 1.0146, "step": 40248 }, { "epoch": 0.04, "learning_rate": 4.9801358438207636e-05, "loss": 0.9702, "step": 40249 }, { "epoch": 0.04, "learning_rate": 4.98013485559917e-05, "loss": 0.9263, "step": 40250 }, { "epoch": 0.04, "learning_rate": 4.980133867353094e-05, "loss": 1.0773, "step": 40251 }, { "epoch": 0.04, "learning_rate": 4.9801328790825356e-05, "loss": 1.0713, "step": 40252 }, { "epoch": 0.04, "learning_rate": 4.980131890787493e-05, "loss": 0.8643, "step": 40253 }, { "epoch": 0.04, "learning_rate": 4.9801309024679686e-05, "loss": 1.1966, "step": 40254 }, { "epoch": 0.04, "learning_rate": 4.980129914123961e-05, "loss": 1.1583, "step": 40255 }, { "epoch": 0.04, "learning_rate": 4.98012892575547e-05, "loss": 0.3594, "step": 40256 }, { "epoch": 0.04, "learning_rate": 4.9801279373624976e-05, "loss": 0.3692, "step": 40257 }, { "epoch": 0.04, "learning_rate": 4.980126948945042e-05, "loss": 0.3091, "step": 40258 }, { "epoch": 0.04, "learning_rate": 4.9801259605031035e-05, "loss": 0.3099, "step": 40259 }, { "epoch": 0.04, "learning_rate": 4.9801249720366815e-05, "loss": 0.2864, "step": 40260 }, { "epoch": 0.04, "learning_rate": 4.980123983545778e-05, "loss": 0.8747, "step": 40261 }, { "epoch": 0.04, "learning_rate": 4.9801229950303915e-05, "loss": 1.2832, "step": 40262 }, { "epoch": 0.04, "learning_rate": 4.9801220064905216e-05, "loss": 1.0106, "step": 40263 }, { "epoch": 0.04, "learning_rate": 4.980121017926169e-05, "loss": 1.1235, "step": 40264 }, { "epoch": 0.04, "learning_rate": 4.980120029337334e-05, "loss": 0.9057, "step": 40265 }, { "epoch": 0.04, "learning_rate": 4.9801190407240166e-05, "loss": 0.9119, "step": 40266 }, { "epoch": 0.04, "learning_rate": 4.980118052086217e-05, "loss": 1.0879, "step": 40267 }, { "epoch": 0.04, "learning_rate": 4.9801170634239336e-05, "loss": 0.7549, "step": 40268 }, { "epoch": 0.04, "learning_rate": 4.980116074737168e-05, "loss": 0.9418, "step": 40269 }, { "epoch": 0.04, "learning_rate": 4.9801150860259194e-05, "loss": 1.2752, "step": 40270 }, { "epoch": 0.04, "learning_rate": 4.980114097290188e-05, "loss": 1.2406, "step": 40271 }, { "epoch": 0.04, "learning_rate": 4.9801131085299746e-05, "loss": 1.1016, "step": 40272 }, { "epoch": 0.04, "learning_rate": 4.980112119745278e-05, "loss": 1.14, "step": 40273 }, { "epoch": 0.04, "learning_rate": 4.980111130936099e-05, "loss": 1.1683, "step": 40274 }, { "epoch": 0.04, "learning_rate": 4.980110142102438e-05, "loss": 1.1489, "step": 40275 }, { "epoch": 0.04, "learning_rate": 4.9801091532442934e-05, "loss": 1.3125, "step": 40276 }, { "epoch": 0.04, "learning_rate": 4.980108164361667e-05, "loss": 1.4885, "step": 40277 }, { "epoch": 0.04, "learning_rate": 4.9801071754545576e-05, "loss": 1.2222, "step": 40278 }, { "epoch": 0.04, "learning_rate": 4.980106186522966e-05, "loss": 1.3618, "step": 40279 }, { "epoch": 0.04, "learning_rate": 4.980105197566891e-05, "loss": 1.348, "step": 40280 }, { "epoch": 0.04, "learning_rate": 4.9801042085863345e-05, "loss": 1.324, "step": 40281 }, { "epoch": 0.04, "learning_rate": 4.9801032195812944e-05, "loss": 1.0962, "step": 40282 }, { "epoch": 0.04, "learning_rate": 4.980102230551773e-05, "loss": 0.9156, "step": 40283 }, { "epoch": 0.04, "learning_rate": 4.9801012414977676e-05, "loss": 0.9025, "step": 40284 }, { "epoch": 0.04, "learning_rate": 4.980100252419281e-05, "loss": 1.0756, "step": 40285 }, { "epoch": 0.04, "learning_rate": 4.980099263316311e-05, "loss": 1.2162, "step": 40286 }, { "epoch": 0.04, "learning_rate": 4.980098274188859e-05, "loss": 0.7657, "step": 40287 }, { "epoch": 0.04, "learning_rate": 4.9800972850369244e-05, "loss": 0.346, "step": 40288 }, { "epoch": 0.04, "learning_rate": 4.980096295860508e-05, "loss": 0.9511, "step": 40289 }, { "epoch": 0.04, "learning_rate": 4.980095306659608e-05, "loss": 1.0045, "step": 40290 }, { "epoch": 0.04, "learning_rate": 4.980094317434226e-05, "loss": 1.0135, "step": 40291 }, { "epoch": 0.04, "learning_rate": 4.9800933281843616e-05, "loss": 1.0791, "step": 40292 }, { "epoch": 0.04, "learning_rate": 4.9800923389100145e-05, "loss": 1.1477, "step": 40293 }, { "epoch": 0.04, "learning_rate": 4.9800913496111854e-05, "loss": 0.7018, "step": 40294 }, { "epoch": 0.04, "learning_rate": 4.980090360287874e-05, "loss": 0.6978, "step": 40295 }, { "epoch": 0.04, "learning_rate": 4.98008937094008e-05, "loss": 1.1565, "step": 40296 }, { "epoch": 0.04, "learning_rate": 4.980088381567804e-05, "loss": 1.1785, "step": 40297 }, { "epoch": 0.04, "learning_rate": 4.980087392171045e-05, "loss": 0.9073, "step": 40298 }, { "epoch": 0.04, "learning_rate": 4.980086402749803e-05, "loss": 1.0832, "step": 40299 }, { "epoch": 0.04, "learning_rate": 4.980085413304079e-05, "loss": 1.4058, "step": 40300 }, { "epoch": 0.04, "learning_rate": 4.980084423833873e-05, "loss": 1.134, "step": 40301 }, { "epoch": 0.04, "learning_rate": 4.9800834343391845e-05, "loss": 0.7869, "step": 40302 }, { "epoch": 0.04, "learning_rate": 4.9800824448200144e-05, "loss": 0.704, "step": 40303 }, { "epoch": 0.04, "learning_rate": 4.980081455276361e-05, "loss": 0.7443, "step": 40304 }, { "epoch": 0.04, "learning_rate": 4.9800804657082255e-05, "loss": 0.5222, "step": 40305 }, { "epoch": 0.04, "learning_rate": 4.9800794761156074e-05, "loss": 1.124, "step": 40306 }, { "epoch": 0.04, "learning_rate": 4.980078486498508e-05, "loss": 1.0752, "step": 40307 }, { "epoch": 0.04, "learning_rate": 4.980077496856926e-05, "loss": 0.9497, "step": 40308 }, { "epoch": 0.04, "learning_rate": 4.9800765071908615e-05, "loss": 1.2164, "step": 40309 }, { "epoch": 0.04, "learning_rate": 4.980075517500314e-05, "loss": 1.1242, "step": 40310 }, { "epoch": 0.04, "learning_rate": 4.980074527785286e-05, "loss": 0.9608, "step": 40311 }, { "epoch": 0.04, "learning_rate": 4.980073538045774e-05, "loss": 1.2742, "step": 40312 }, { "epoch": 0.04, "learning_rate": 4.980072548281781e-05, "loss": 1.3697, "step": 40313 }, { "epoch": 0.04, "learning_rate": 4.980071558493305e-05, "loss": 1.0308, "step": 40314 }, { "epoch": 0.04, "learning_rate": 4.9800705686803465e-05, "loss": 0.7338, "step": 40315 }, { "epoch": 0.04, "learning_rate": 4.980069578842906e-05, "loss": 0.8847, "step": 40316 }, { "epoch": 0.04, "learning_rate": 4.980068588980984e-05, "loss": 0.7842, "step": 40317 }, { "epoch": 0.04, "learning_rate": 4.98006759909458e-05, "loss": 1.1488, "step": 40318 }, { "epoch": 0.04, "learning_rate": 4.980066609183693e-05, "loss": 1.1526, "step": 40319 }, { "epoch": 0.04, "learning_rate": 4.980065619248324e-05, "loss": 1.0138, "step": 40320 }, { "epoch": 0.04, "learning_rate": 4.9800646292884726e-05, "loss": 0.9225, "step": 40321 }, { "epoch": 0.04, "learning_rate": 4.980063639304139e-05, "loss": 0.9303, "step": 40322 }, { "epoch": 0.04, "learning_rate": 4.980062649295324e-05, "loss": 1.0293, "step": 40323 }, { "epoch": 0.04, "learning_rate": 4.980061659262026e-05, "loss": 1.1079, "step": 40324 }, { "epoch": 0.04, "learning_rate": 4.980060669204247e-05, "loss": 1.171, "step": 40325 }, { "epoch": 0.04, "learning_rate": 4.980059679121985e-05, "loss": 1.3404, "step": 40326 }, { "epoch": 0.04, "learning_rate": 4.9800586890152414e-05, "loss": 1.1918, "step": 40327 }, { "epoch": 0.04, "learning_rate": 4.9800576988840154e-05, "loss": 0.9194, "step": 40328 }, { "epoch": 0.04, "learning_rate": 4.980056708728307e-05, "loss": 0.9174, "step": 40329 }, { "epoch": 0.04, "learning_rate": 4.980055718548117e-05, "loss": 1.8803, "step": 40330 }, { "epoch": 0.04, "learning_rate": 4.980054728343444e-05, "loss": 1.1949, "step": 40331 }, { "epoch": 0.04, "learning_rate": 4.9800537381142905e-05, "loss": 1.2113, "step": 40332 }, { "epoch": 0.04, "learning_rate": 4.980052747860654e-05, "loss": 0.9933, "step": 40333 }, { "epoch": 0.04, "learning_rate": 4.9800517575825356e-05, "loss": 0.9518, "step": 40334 }, { "epoch": 0.04, "learning_rate": 4.980050767279936e-05, "loss": 0.5709, "step": 40335 }, { "epoch": 0.04, "learning_rate": 4.980049776952853e-05, "loss": 1.0836, "step": 40336 }, { "epoch": 0.04, "learning_rate": 4.9800487866012885e-05, "loss": 0.8889, "step": 40337 }, { "epoch": 0.04, "learning_rate": 4.980047796225242e-05, "loss": 0.5545, "step": 40338 }, { "epoch": 0.04, "learning_rate": 4.980046805824714e-05, "loss": 1.0021, "step": 40339 }, { "epoch": 0.04, "learning_rate": 4.980045815399703e-05, "loss": 0.7766, "step": 40340 }, { "epoch": 0.04, "learning_rate": 4.980044824950211e-05, "loss": 0.6365, "step": 40341 }, { "epoch": 0.04, "learning_rate": 4.980043834476237e-05, "loss": 0.8995, "step": 40342 }, { "epoch": 0.04, "learning_rate": 4.98004284397778e-05, "loss": 1.1682, "step": 40343 }, { "epoch": 0.04, "learning_rate": 4.980041853454842e-05, "loss": 1.2797, "step": 40344 }, { "epoch": 0.04, "learning_rate": 4.9800408629074226e-05, "loss": 1.1498, "step": 40345 }, { "epoch": 0.04, "learning_rate": 4.98003987233552e-05, "loss": 1.1213, "step": 40346 }, { "epoch": 0.04, "learning_rate": 4.980038881739136e-05, "loss": 0.7712, "step": 40347 }, { "epoch": 0.04, "learning_rate": 4.980037891118271e-05, "loss": 1.1566, "step": 40348 }, { "epoch": 0.04, "learning_rate": 4.980036900472923e-05, "loss": 0.8583, "step": 40349 }, { "epoch": 0.04, "learning_rate": 4.980035909803094e-05, "loss": 0.8445, "step": 40350 }, { "epoch": 0.04, "learning_rate": 4.980034919108781e-05, "loss": 0.639, "step": 40351 }, { "epoch": 0.04, "learning_rate": 4.980033928389989e-05, "loss": 0.7378, "step": 40352 }, { "epoch": 0.04, "learning_rate": 4.980032937646714e-05, "loss": 0.6765, "step": 40353 }, { "epoch": 0.04, "learning_rate": 4.980031946878956e-05, "loss": 0.9341, "step": 40354 }, { "epoch": 0.04, "learning_rate": 4.980030956086718e-05, "loss": 0.9038, "step": 40355 }, { "epoch": 0.04, "learning_rate": 4.980029965269997e-05, "loss": 0.7714, "step": 40356 }, { "epoch": 0.04, "learning_rate": 4.980028974428794e-05, "loss": 1.2605, "step": 40357 }, { "epoch": 0.04, "learning_rate": 4.9800279835631105e-05, "loss": 1.0448, "step": 40358 }, { "epoch": 0.04, "learning_rate": 4.980026992672944e-05, "loss": 0.8777, "step": 40359 }, { "epoch": 0.04, "learning_rate": 4.980026001758296e-05, "loss": 0.8664, "step": 40360 }, { "epoch": 0.04, "learning_rate": 4.980025010819166e-05, "loss": 1.0218, "step": 40361 }, { "epoch": 0.04, "learning_rate": 4.980024019855555e-05, "loss": 1.0082, "step": 40362 }, { "epoch": 0.04, "learning_rate": 4.980023028867462e-05, "loss": 1.1168, "step": 40363 }, { "epoch": 0.04, "learning_rate": 4.980022037854887e-05, "loss": 1.2178, "step": 40364 }, { "epoch": 0.04, "learning_rate": 4.9800210468178304e-05, "loss": 0.904, "step": 40365 }, { "epoch": 0.04, "learning_rate": 4.980020055756292e-05, "loss": 1.2466, "step": 40366 }, { "epoch": 0.04, "learning_rate": 4.980019064670272e-05, "loss": 1.0779, "step": 40367 }, { "epoch": 0.04, "learning_rate": 4.98001807355977e-05, "loss": 1.0573, "step": 40368 }, { "epoch": 0.04, "learning_rate": 4.980017082424787e-05, "loss": 1.2056, "step": 40369 }, { "epoch": 0.04, "learning_rate": 4.9800160912653215e-05, "loss": 0.9002, "step": 40370 }, { "epoch": 0.04, "learning_rate": 4.980015100081375e-05, "loss": 0.7154, "step": 40371 }, { "epoch": 0.04, "learning_rate": 4.980014108872947e-05, "loss": 0.9177, "step": 40372 }, { "epoch": 0.04, "learning_rate": 4.980013117640037e-05, "loss": 0.8781, "step": 40373 }, { "epoch": 0.04, "learning_rate": 4.980012126382645e-05, "loss": 1.0062, "step": 40374 }, { "epoch": 0.04, "learning_rate": 4.980011135100771e-05, "loss": 1.2545, "step": 40375 }, { "epoch": 0.04, "learning_rate": 4.980010143794416e-05, "loss": 1.4403, "step": 40376 }, { "epoch": 0.04, "learning_rate": 4.9800091524635795e-05, "loss": 1.2413, "step": 40377 }, { "epoch": 0.04, "learning_rate": 4.9800081611082615e-05, "loss": 1.0187, "step": 40378 }, { "epoch": 0.04, "learning_rate": 4.980007169728461e-05, "loss": 1.1179, "step": 40379 }, { "epoch": 0.04, "learning_rate": 4.98000617832418e-05, "loss": 1.3806, "step": 40380 }, { "epoch": 0.04, "learning_rate": 4.980005186895417e-05, "loss": 1.0002, "step": 40381 }, { "epoch": 0.04, "learning_rate": 4.9800041954421725e-05, "loss": 0.7196, "step": 40382 }, { "epoch": 0.04, "learning_rate": 4.980003203964446e-05, "loss": 0.7002, "step": 40383 }, { "epoch": 0.04, "learning_rate": 4.980002212462238e-05, "loss": 0.9976, "step": 40384 }, { "epoch": 0.04, "learning_rate": 4.980001220935549e-05, "loss": 1.2453, "step": 40385 }, { "epoch": 0.04, "learning_rate": 4.980000229384379e-05, "loss": 1.298, "step": 40386 }, { "epoch": 0.04, "learning_rate": 4.9799992378087267e-05, "loss": 0.9038, "step": 40387 }, { "epoch": 0.04, "learning_rate": 4.9799982462085924e-05, "loss": 0.2244, "step": 40388 }, { "epoch": 0.04, "learning_rate": 4.979997254583978e-05, "loss": 0.1102, "step": 40389 }, { "epoch": 0.04, "learning_rate": 4.979996262934881e-05, "loss": 0.128, "step": 40390 }, { "epoch": 0.04, "learning_rate": 4.9799952712613026e-05, "loss": 0.3284, "step": 40391 }, { "epoch": 0.04, "learning_rate": 4.9799942795632425e-05, "loss": 0.9555, "step": 40392 }, { "epoch": 0.04, "learning_rate": 4.979993287840702e-05, "loss": 0.8643, "step": 40393 }, { "epoch": 0.04, "learning_rate": 4.979992296093679e-05, "loss": 1.0874, "step": 40394 }, { "epoch": 0.04, "learning_rate": 4.979991304322175e-05, "loss": 1.2404, "step": 40395 }, { "epoch": 0.04, "learning_rate": 4.9799903125261894e-05, "loss": 0.8376, "step": 40396 }, { "epoch": 0.04, "learning_rate": 4.979989320705722e-05, "loss": 1.0915, "step": 40397 }, { "epoch": 0.04, "learning_rate": 4.9799883288607745e-05, "loss": 1.096, "step": 40398 }, { "epoch": 0.04, "learning_rate": 4.979987336991345e-05, "loss": 1.199, "step": 40399 }, { "epoch": 0.04, "learning_rate": 4.979986345097434e-05, "loss": 1.1232, "step": 40400 }, { "epoch": 0.04, "learning_rate": 4.9799853531790414e-05, "loss": 1.208, "step": 40401 }, { "epoch": 0.04, "learning_rate": 4.979984361236167e-05, "loss": 1.1252, "step": 40402 }, { "epoch": 0.04, "learning_rate": 4.979983369268812e-05, "loss": 0.9586, "step": 40403 }, { "epoch": 0.04, "learning_rate": 4.979982377276976e-05, "loss": 1.1601, "step": 40404 }, { "epoch": 0.04, "learning_rate": 4.979981385260658e-05, "loss": 0.8218, "step": 40405 }, { "epoch": 0.04, "learning_rate": 4.9799803932198585e-05, "loss": 1.3614, "step": 40406 }, { "epoch": 0.04, "learning_rate": 4.979979401154578e-05, "loss": 0.9328, "step": 40407 }, { "epoch": 0.04, "learning_rate": 4.979978409064816e-05, "loss": 0.8974, "step": 40408 }, { "epoch": 0.04, "learning_rate": 4.979977416950573e-05, "loss": 1.2297, "step": 40409 }, { "epoch": 0.04, "learning_rate": 4.979976424811849e-05, "loss": 0.8565, "step": 40410 }, { "epoch": 0.04, "learning_rate": 4.979975432648643e-05, "loss": 1.0016, "step": 40411 }, { "epoch": 0.04, "learning_rate": 4.9799744404609564e-05, "loss": 1.1097, "step": 40412 }, { "epoch": 0.04, "learning_rate": 4.979973448248787e-05, "loss": 1.0564, "step": 40413 }, { "epoch": 0.04, "learning_rate": 4.979972456012138e-05, "loss": 1.2441, "step": 40414 }, { "epoch": 0.04, "learning_rate": 4.979971463751007e-05, "loss": 1.1635, "step": 40415 }, { "epoch": 0.04, "learning_rate": 4.979970471465395e-05, "loss": 1.0216, "step": 40416 }, { "epoch": 0.04, "learning_rate": 4.979969479155302e-05, "loss": 1.0714, "step": 40417 }, { "epoch": 0.04, "learning_rate": 4.979968486820727e-05, "loss": 1.1885, "step": 40418 }, { "epoch": 0.04, "learning_rate": 4.979967494461672e-05, "loss": 1.1723, "step": 40419 }, { "epoch": 0.04, "learning_rate": 4.979966502078135e-05, "loss": 1.201, "step": 40420 }, { "epoch": 0.04, "learning_rate": 4.979965509670117e-05, "loss": 1.1066, "step": 40421 }, { "epoch": 0.04, "learning_rate": 4.979964517237618e-05, "loss": 1.0195, "step": 40422 }, { "epoch": 0.04, "learning_rate": 4.979963524780637e-05, "loss": 1.7522, "step": 40423 }, { "epoch": 0.04, "learning_rate": 4.979962532299176e-05, "loss": 1.5405, "step": 40424 }, { "epoch": 0.04, "learning_rate": 4.9799615397932336e-05, "loss": 1.2195, "step": 40425 }, { "epoch": 0.04, "learning_rate": 4.97996054726281e-05, "loss": 1.036, "step": 40426 }, { "epoch": 0.04, "learning_rate": 4.979959554707905e-05, "loss": 0.9554, "step": 40427 }, { "epoch": 0.04, "learning_rate": 4.979958562128519e-05, "loss": 1.166, "step": 40428 }, { "epoch": 0.04, "learning_rate": 4.9799575695246515e-05, "loss": 1.0108, "step": 40429 }, { "epoch": 0.04, "learning_rate": 4.979956576896303e-05, "loss": 0.7945, "step": 40430 }, { "epoch": 0.04, "learning_rate": 4.9799555842434744e-05, "loss": 0.8358, "step": 40431 }, { "epoch": 0.04, "learning_rate": 4.9799545915661637e-05, "loss": 0.968, "step": 40432 }, { "epoch": 0.04, "learning_rate": 4.979953598864372e-05, "loss": 1.2343, "step": 40433 }, { "epoch": 0.04, "learning_rate": 4.9799526061380996e-05, "loss": 0.8089, "step": 40434 }, { "epoch": 0.04, "learning_rate": 4.979951613387346e-05, "loss": 1.2418, "step": 40435 }, { "epoch": 0.04, "learning_rate": 4.979950620612112e-05, "loss": 1.0267, "step": 40436 }, { "epoch": 0.04, "learning_rate": 4.979949627812396e-05, "loss": 1.0541, "step": 40437 }, { "epoch": 0.04, "learning_rate": 4.9799486349881994e-05, "loss": 1.0007, "step": 40438 }, { "epoch": 0.04, "learning_rate": 4.9799476421395216e-05, "loss": 1.124, "step": 40439 }, { "epoch": 0.04, "learning_rate": 4.979946649266363e-05, "loss": 1.1527, "step": 40440 }, { "epoch": 0.04, "learning_rate": 4.9799456563687236e-05, "loss": 0.9214, "step": 40441 }, { "epoch": 0.04, "learning_rate": 4.9799446634466027e-05, "loss": 0.7335, "step": 40442 }, { "epoch": 0.04, "learning_rate": 4.979943670500001e-05, "loss": 0.113, "step": 40443 }, { "epoch": 0.04, "learning_rate": 4.979942677528919e-05, "loss": 0.1031, "step": 40444 }, { "epoch": 0.04, "learning_rate": 4.9799416845333555e-05, "loss": 0.137, "step": 40445 }, { "epoch": 0.04, "learning_rate": 4.979940691513311e-05, "loss": 0.0892, "step": 40446 }, { "epoch": 0.04, "learning_rate": 4.9799396984687855e-05, "loss": 0.0601, "step": 40447 }, { "epoch": 0.04, "learning_rate": 4.9799387053997795e-05, "loss": 0.5432, "step": 40448 }, { "epoch": 0.04, "learning_rate": 4.9799377123062916e-05, "loss": 0.9166, "step": 40449 }, { "epoch": 0.04, "learning_rate": 4.979936719188324e-05, "loss": 1.3575, "step": 40450 }, { "epoch": 0.04, "learning_rate": 4.979935726045875e-05, "loss": 1.0591, "step": 40451 }, { "epoch": 0.04, "learning_rate": 4.979934732878945e-05, "loss": 1.0793, "step": 40452 }, { "epoch": 0.04, "learning_rate": 4.979933739687535e-05, "loss": 1.1634, "step": 40453 }, { "epoch": 0.04, "learning_rate": 4.979932746471643e-05, "loss": 1.2072, "step": 40454 }, { "epoch": 0.04, "learning_rate": 4.979931753231271e-05, "loss": 1.4106, "step": 40455 }, { "epoch": 0.04, "learning_rate": 4.9799307599664167e-05, "loss": 1.0665, "step": 40456 }, { "epoch": 0.04, "learning_rate": 4.9799297666770825e-05, "loss": 1.1354, "step": 40457 }, { "epoch": 0.04, "learning_rate": 4.9799287733632685e-05, "loss": 1.0043, "step": 40458 }, { "epoch": 0.04, "learning_rate": 4.9799277800249725e-05, "loss": 1.1989, "step": 40459 }, { "epoch": 0.04, "learning_rate": 4.979926786662196e-05, "loss": 0.8771, "step": 40460 }, { "epoch": 0.04, "learning_rate": 4.979925793274938e-05, "loss": 0.9233, "step": 40461 }, { "epoch": 0.04, "learning_rate": 4.9799247998632e-05, "loss": 0.8634, "step": 40462 }, { "epoch": 0.04, "learning_rate": 4.979923806426982e-05, "loss": 1.248, "step": 40463 }, { "epoch": 0.04, "learning_rate": 4.979922812966282e-05, "loss": 0.9448, "step": 40464 }, { "epoch": 0.04, "learning_rate": 4.979921819481102e-05, "loss": 0.8783, "step": 40465 }, { "epoch": 0.04, "learning_rate": 4.97992082597144e-05, "loss": 1.0168, "step": 40466 }, { "epoch": 0.04, "learning_rate": 4.979919832437298e-05, "loss": 1.4717, "step": 40467 }, { "epoch": 0.04, "learning_rate": 4.979918838878676e-05, "loss": 1.3347, "step": 40468 }, { "epoch": 0.04, "learning_rate": 4.9799178452955716e-05, "loss": 1.3556, "step": 40469 }, { "epoch": 0.04, "learning_rate": 4.979916851687988e-05, "loss": 0.9168, "step": 40470 }, { "epoch": 0.04, "learning_rate": 4.9799158580559236e-05, "loss": 0.8621, "step": 40471 }, { "epoch": 0.04, "learning_rate": 4.979914864399378e-05, "loss": 1.1182, "step": 40472 }, { "epoch": 0.04, "learning_rate": 4.9799138707183525e-05, "loss": 0.9528, "step": 40473 }, { "epoch": 0.04, "learning_rate": 4.979912877012845e-05, "loss": 1.0736, "step": 40474 }, { "epoch": 0.04, "learning_rate": 4.9799118832828576e-05, "loss": 0.8429, "step": 40475 }, { "epoch": 0.04, "learning_rate": 4.97991088952839e-05, "loss": 1.0234, "step": 40476 }, { "epoch": 0.04, "learning_rate": 4.979909895749441e-05, "loss": 1.0059, "step": 40477 }, { "epoch": 0.04, "learning_rate": 4.979908901946012e-05, "loss": 0.7953, "step": 40478 }, { "epoch": 0.04, "learning_rate": 4.979907908118102e-05, "loss": 1.4784, "step": 40479 }, { "epoch": 0.04, "learning_rate": 4.979906914265711e-05, "loss": 1.1617, "step": 40480 }, { "epoch": 0.04, "learning_rate": 4.9799059203888396e-05, "loss": 1.5323, "step": 40481 }, { "epoch": 0.04, "learning_rate": 4.979904926487488e-05, "loss": 1.0616, "step": 40482 }, { "epoch": 0.04, "learning_rate": 4.979903932561656e-05, "loss": 1.1777, "step": 40483 }, { "epoch": 0.04, "learning_rate": 4.979902938611343e-05, "loss": 1.1689, "step": 40484 }, { "epoch": 0.04, "learning_rate": 4.97990194463655e-05, "loss": 0.7069, "step": 40485 }, { "epoch": 0.04, "learning_rate": 4.979900950637276e-05, "loss": 0.5363, "step": 40486 }, { "epoch": 0.04, "learning_rate": 4.979899956613522e-05, "loss": 1.0924, "step": 40487 }, { "epoch": 0.04, "learning_rate": 4.979898962565286e-05, "loss": 1.2736, "step": 40488 }, { "epoch": 0.04, "learning_rate": 4.9798979684925706e-05, "loss": 0.9107, "step": 40489 }, { "epoch": 0.04, "learning_rate": 4.979896974395375e-05, "loss": 0.9895, "step": 40490 }, { "epoch": 0.04, "learning_rate": 4.979895980273698e-05, "loss": 1.1239, "step": 40491 }, { "epoch": 0.04, "learning_rate": 4.979894986127541e-05, "loss": 1.2821, "step": 40492 }, { "epoch": 0.04, "learning_rate": 4.9798939919569036e-05, "loss": 0.9863, "step": 40493 }, { "epoch": 0.04, "learning_rate": 4.9798929977617856e-05, "loss": 1.0842, "step": 40494 }, { "epoch": 0.04, "learning_rate": 4.979892003542187e-05, "loss": 0.9367, "step": 40495 }, { "epoch": 0.04, "learning_rate": 4.979891009298108e-05, "loss": 1.2342, "step": 40496 }, { "epoch": 0.04, "learning_rate": 4.979890015029549e-05, "loss": 0.9419, "step": 40497 }, { "epoch": 0.04, "learning_rate": 4.979889020736509e-05, "loss": 1.6446, "step": 40498 }, { "epoch": 0.04, "learning_rate": 4.979888026418989e-05, "loss": 0.8643, "step": 40499 }, { "epoch": 0.04, "learning_rate": 4.9798870320769886e-05, "loss": 1.1664, "step": 40500 }, { "epoch": 0.04, "eval_loss": 1.0778478384017944, "eval_runtime": 126.3161, "eval_samples_per_second": 10.965, "eval_steps_per_second": 5.486, "step": 40500 }, { "epoch": 0.04, "learning_rate": 4.979886037710507e-05, "loss": 1.0081, "step": 40501 }, { "epoch": 0.04, "learning_rate": 4.979885043319546e-05, "loss": 1.2385, "step": 40502 }, { "epoch": 0.04, "learning_rate": 4.979884048904104e-05, "loss": 1.307, "step": 40503 }, { "epoch": 0.04, "learning_rate": 4.979883054464182e-05, "loss": 1.4312, "step": 40504 }, { "epoch": 0.04, "learning_rate": 4.97988205999978e-05, "loss": 0.8789, "step": 40505 }, { "epoch": 0.04, "learning_rate": 4.9798810655108965e-05, "loss": 0.9337, "step": 40506 }, { "epoch": 0.04, "learning_rate": 4.9798800709975334e-05, "loss": 1.0624, "step": 40507 }, { "epoch": 0.04, "learning_rate": 4.97987907645969e-05, "loss": 0.5341, "step": 40508 }, { "epoch": 0.04, "learning_rate": 4.9798780818973667e-05, "loss": 0.835, "step": 40509 }, { "epoch": 0.04, "learning_rate": 4.9798770873105624e-05, "loss": 0.9066, "step": 40510 }, { "epoch": 0.04, "learning_rate": 4.9798760926992775e-05, "loss": 0.2816, "step": 40511 }, { "epoch": 0.04, "learning_rate": 4.9798750980635134e-05, "loss": 0.6443, "step": 40512 }, { "epoch": 0.04, "learning_rate": 4.979874103403268e-05, "loss": 1.0581, "step": 40513 }, { "epoch": 0.04, "learning_rate": 4.9798731087185426e-05, "loss": 0.4819, "step": 40514 }, { "epoch": 0.04, "learning_rate": 4.979872114009337e-05, "loss": 0.2877, "step": 40515 }, { "epoch": 0.04, "learning_rate": 4.9798711192756514e-05, "loss": 0.7907, "step": 40516 }, { "epoch": 0.04, "learning_rate": 4.979870124517485e-05, "loss": 0.8198, "step": 40517 }, { "epoch": 0.04, "learning_rate": 4.9798691297348385e-05, "loss": 0.9639, "step": 40518 }, { "epoch": 0.04, "learning_rate": 4.979868134927712e-05, "loss": 1.0129, "step": 40519 }, { "epoch": 0.04, "learning_rate": 4.979867140096106e-05, "loss": 0.4883, "step": 40520 }, { "epoch": 0.04, "learning_rate": 4.979866145240018e-05, "loss": 1.0144, "step": 40521 }, { "epoch": 0.04, "learning_rate": 4.9798651503594515e-05, "loss": 1.0059, "step": 40522 }, { "epoch": 0.04, "learning_rate": 4.979864155454405e-05, "loss": 1.0204, "step": 40523 }, { "epoch": 0.04, "learning_rate": 4.979863160524877e-05, "loss": 0.8635, "step": 40524 }, { "epoch": 0.04, "learning_rate": 4.9798621655708695e-05, "loss": 0.9062, "step": 40525 }, { "epoch": 0.04, "learning_rate": 4.9798611705923816e-05, "loss": 1.1596, "step": 40526 }, { "epoch": 0.04, "learning_rate": 4.979860175589414e-05, "loss": 1.2606, "step": 40527 }, { "epoch": 0.04, "learning_rate": 4.979859180561965e-05, "loss": 0.5067, "step": 40528 }, { "epoch": 0.04, "learning_rate": 4.9798581855100376e-05, "loss": 1.1451, "step": 40529 }, { "epoch": 0.04, "learning_rate": 4.9798571904336287e-05, "loss": 1.1187, "step": 40530 }, { "epoch": 0.04, "learning_rate": 4.9798561953327404e-05, "loss": 0.8118, "step": 40531 }, { "epoch": 0.04, "learning_rate": 4.9798552002073716e-05, "loss": 1.1882, "step": 40532 }, { "epoch": 0.04, "learning_rate": 4.9798542050575235e-05, "loss": 0.9043, "step": 40533 }, { "epoch": 0.04, "learning_rate": 4.979853209883195e-05, "loss": 1.0109, "step": 40534 }, { "epoch": 0.04, "learning_rate": 4.979852214684386e-05, "loss": 1.2129, "step": 40535 }, { "epoch": 0.04, "learning_rate": 4.979851219461097e-05, "loss": 1.1097, "step": 40536 }, { "epoch": 0.04, "learning_rate": 4.9798502242133285e-05, "loss": 1.1431, "step": 40537 }, { "epoch": 0.04, "learning_rate": 4.9798492289410795e-05, "loss": 0.9305, "step": 40538 }, { "epoch": 0.04, "learning_rate": 4.979848233644351e-05, "loss": 0.9603, "step": 40539 }, { "epoch": 0.04, "learning_rate": 4.979847238323142e-05, "loss": 1.3143, "step": 40540 }, { "epoch": 0.04, "learning_rate": 4.9798462429774526e-05, "loss": 0.868, "step": 40541 }, { "epoch": 0.04, "learning_rate": 4.979845247607284e-05, "loss": 0.9027, "step": 40542 }, { "epoch": 0.04, "learning_rate": 4.979844252212635e-05, "loss": 0.7703, "step": 40543 }, { "epoch": 0.04, "learning_rate": 4.979843256793506e-05, "loss": 0.9659, "step": 40544 }, { "epoch": 0.04, "learning_rate": 4.979842261349897e-05, "loss": 0.9336, "step": 40545 }, { "epoch": 0.04, "learning_rate": 4.9798412658818086e-05, "loss": 1.3644, "step": 40546 }, { "epoch": 0.04, "learning_rate": 4.97984027038924e-05, "loss": 0.9922, "step": 40547 }, { "epoch": 0.04, "learning_rate": 4.979839274872191e-05, "loss": 1.4754, "step": 40548 }, { "epoch": 0.04, "learning_rate": 4.979838279330662e-05, "loss": 0.8188, "step": 40549 }, { "epoch": 0.04, "learning_rate": 4.979837283764653e-05, "loss": 1.0754, "step": 40550 }, { "epoch": 0.04, "learning_rate": 4.979836288174166e-05, "loss": 0.9152, "step": 40551 }, { "epoch": 0.04, "learning_rate": 4.979835292559197e-05, "loss": 0.7892, "step": 40552 }, { "epoch": 0.04, "learning_rate": 4.979834296919749e-05, "loss": 1.0841, "step": 40553 }, { "epoch": 0.04, "learning_rate": 4.979833301255821e-05, "loss": 1.0044, "step": 40554 }, { "epoch": 0.04, "learning_rate": 4.9798323055674124e-05, "loss": 0.9743, "step": 40555 }, { "epoch": 0.04, "learning_rate": 4.9798313098545246e-05, "loss": 1.1729, "step": 40556 }, { "epoch": 0.04, "learning_rate": 4.979830314117157e-05, "loss": 0.8815, "step": 40557 }, { "epoch": 0.04, "learning_rate": 4.979829318355309e-05, "loss": 0.85, "step": 40558 }, { "epoch": 0.04, "learning_rate": 4.979828322568982e-05, "loss": 1.141, "step": 40559 }, { "epoch": 0.04, "learning_rate": 4.9798273267581754e-05, "loss": 0.942, "step": 40560 }, { "epoch": 0.04, "learning_rate": 4.979826330922888e-05, "loss": 1.0155, "step": 40561 }, { "epoch": 0.04, "learning_rate": 4.9798253350631205e-05, "loss": 1.1933, "step": 40562 }, { "epoch": 0.04, "learning_rate": 4.9798243391788745e-05, "loss": 0.9039, "step": 40563 }, { "epoch": 0.04, "learning_rate": 4.979823343270148e-05, "loss": 1.0931, "step": 40564 }, { "epoch": 0.04, "learning_rate": 4.979822347336942e-05, "loss": 1.1469, "step": 40565 }, { "epoch": 0.04, "learning_rate": 4.9798213513792557e-05, "loss": 0.9339, "step": 40566 }, { "epoch": 0.04, "learning_rate": 4.97982035539709e-05, "loss": 0.6482, "step": 40567 }, { "epoch": 0.04, "learning_rate": 4.979819359390445e-05, "loss": 1.0677, "step": 40568 }, { "epoch": 0.04, "learning_rate": 4.97981836335932e-05, "loss": 1.0192, "step": 40569 }, { "epoch": 0.04, "learning_rate": 4.9798173673037153e-05, "loss": 1.0792, "step": 40570 }, { "epoch": 0.04, "learning_rate": 4.97981637122363e-05, "loss": 0.9144, "step": 40571 }, { "epoch": 0.04, "learning_rate": 4.979815375119066e-05, "loss": 1.0269, "step": 40572 }, { "epoch": 0.04, "learning_rate": 4.979814378990022e-05, "loss": 0.901, "step": 40573 }, { "epoch": 0.04, "learning_rate": 4.979813382836498e-05, "loss": 0.7365, "step": 40574 }, { "epoch": 0.04, "learning_rate": 4.979812386658495e-05, "loss": 0.9899, "step": 40575 }, { "epoch": 0.04, "learning_rate": 4.979811390456012e-05, "loss": 1.0944, "step": 40576 }, { "epoch": 0.04, "learning_rate": 4.97981039422905e-05, "loss": 1.3782, "step": 40577 }, { "epoch": 0.04, "learning_rate": 4.9798093979776076e-05, "loss": 1.0068, "step": 40578 }, { "epoch": 0.04, "learning_rate": 4.9798084017016854e-05, "loss": 0.7401, "step": 40579 }, { "epoch": 0.04, "learning_rate": 4.979807405401284e-05, "loss": 1.0764, "step": 40580 }, { "epoch": 0.04, "learning_rate": 4.979806409076403e-05, "loss": 0.2329, "step": 40581 }, { "epoch": 0.04, "learning_rate": 4.979805412727042e-05, "loss": 0.657, "step": 40582 }, { "epoch": 0.04, "learning_rate": 4.9798044163532016e-05, "loss": 0.649, "step": 40583 }, { "epoch": 0.04, "learning_rate": 4.979803419954881e-05, "loss": 1.1981, "step": 40584 }, { "epoch": 0.04, "learning_rate": 4.9798024235320815e-05, "loss": 0.8992, "step": 40585 }, { "epoch": 0.04, "learning_rate": 4.979801427084803e-05, "loss": 0.5391, "step": 40586 }, { "epoch": 0.04, "learning_rate": 4.9798004306130444e-05, "loss": 0.8977, "step": 40587 }, { "epoch": 0.04, "learning_rate": 4.979799434116806e-05, "loss": 1.1953, "step": 40588 }, { "epoch": 0.04, "learning_rate": 4.979798437596088e-05, "loss": 1.4581, "step": 40589 }, { "epoch": 0.04, "learning_rate": 4.9797974410508916e-05, "loss": 1.1743, "step": 40590 }, { "epoch": 0.04, "learning_rate": 4.9797964444812144e-05, "loss": 1.1606, "step": 40591 }, { "epoch": 0.04, "learning_rate": 4.9797954478870586e-05, "loss": 1.4195, "step": 40592 }, { "epoch": 0.04, "learning_rate": 4.979794451268422e-05, "loss": 0.9894, "step": 40593 }, { "epoch": 0.04, "learning_rate": 4.979793454625307e-05, "loss": 1.1225, "step": 40594 }, { "epoch": 0.04, "learning_rate": 4.979792457957713e-05, "loss": 0.8619, "step": 40595 }, { "epoch": 0.04, "learning_rate": 4.979791461265638e-05, "loss": 0.9453, "step": 40596 }, { "epoch": 0.04, "learning_rate": 4.979790464549085e-05, "loss": 0.6413, "step": 40597 }, { "epoch": 0.04, "learning_rate": 4.979789467808051e-05, "loss": 0.7261, "step": 40598 }, { "epoch": 0.04, "learning_rate": 4.9797884710425394e-05, "loss": 0.9137, "step": 40599 }, { "epoch": 0.04, "learning_rate": 4.9797874742525475e-05, "loss": 1.0982, "step": 40600 }, { "epoch": 0.04, "learning_rate": 4.979786477438076e-05, "loss": 1.0322, "step": 40601 }, { "epoch": 0.04, "learning_rate": 4.9797854805991254e-05, "loss": 1.1399, "step": 40602 }, { "epoch": 0.04, "learning_rate": 4.979784483735695e-05, "loss": 1.2355, "step": 40603 }, { "epoch": 0.04, "learning_rate": 4.9797834868477856e-05, "loss": 0.9711, "step": 40604 }, { "epoch": 0.04, "learning_rate": 4.979782489935396e-05, "loss": 0.8249, "step": 40605 }, { "epoch": 0.04, "learning_rate": 4.979781492998528e-05, "loss": 1.0791, "step": 40606 }, { "epoch": 0.04, "learning_rate": 4.979780496037181e-05, "loss": 0.9677, "step": 40607 }, { "epoch": 0.04, "learning_rate": 4.9797794990513535e-05, "loss": 0.9108, "step": 40608 }, { "epoch": 0.04, "learning_rate": 4.979778502041047e-05, "loss": 1.0888, "step": 40609 }, { "epoch": 0.04, "learning_rate": 4.979777505006261e-05, "loss": 0.4806, "step": 40610 }, { "epoch": 0.04, "learning_rate": 4.979776507946996e-05, "loss": 1.021, "step": 40611 }, { "epoch": 0.04, "learning_rate": 4.979775510863252e-05, "loss": 0.9507, "step": 40612 }, { "epoch": 0.04, "learning_rate": 4.9797745137550286e-05, "loss": 1.0948, "step": 40613 }, { "epoch": 0.04, "learning_rate": 4.979773516622326e-05, "loss": 0.9403, "step": 40614 }, { "epoch": 0.04, "learning_rate": 4.979772519465143e-05, "loss": 1.2827, "step": 40615 }, { "epoch": 0.04, "learning_rate": 4.979771522283482e-05, "loss": 1.0337, "step": 40616 }, { "epoch": 0.04, "learning_rate": 4.979770525077342e-05, "loss": 1.2199, "step": 40617 }, { "epoch": 0.04, "learning_rate": 4.979769527846722e-05, "loss": 1.224, "step": 40618 }, { "epoch": 0.04, "learning_rate": 4.979768530591622e-05, "loss": 0.9633, "step": 40619 }, { "epoch": 0.04, "learning_rate": 4.9797675333120444e-05, "loss": 1.0227, "step": 40620 }, { "epoch": 0.04, "learning_rate": 4.979766536007986e-05, "loss": 1.2224, "step": 40621 }, { "epoch": 0.04, "learning_rate": 4.9797655386794495e-05, "loss": 1.2391, "step": 40622 }, { "epoch": 0.04, "learning_rate": 4.9797645413264335e-05, "loss": 0.6712, "step": 40623 }, { "epoch": 0.04, "learning_rate": 4.979763543948938e-05, "loss": 0.7211, "step": 40624 }, { "epoch": 0.04, "learning_rate": 4.979762546546964e-05, "loss": 0.9912, "step": 40625 }, { "epoch": 0.04, "learning_rate": 4.97976154912051e-05, "loss": 0.8863, "step": 40626 }, { "epoch": 0.04, "learning_rate": 4.9797605516695776e-05, "loss": 0.8564, "step": 40627 }, { "epoch": 0.04, "learning_rate": 4.979759554194165e-05, "loss": 1.4876, "step": 40628 }, { "epoch": 0.04, "learning_rate": 4.9797585566942745e-05, "loss": 1.2812, "step": 40629 }, { "epoch": 0.04, "learning_rate": 4.9797575591699044e-05, "loss": 2.071, "step": 40630 }, { "epoch": 0.04, "learning_rate": 4.979756561621055e-05, "loss": 1.1933, "step": 40631 }, { "epoch": 0.04, "learning_rate": 4.9797555640477264e-05, "loss": 1.0193, "step": 40632 }, { "epoch": 0.04, "learning_rate": 4.979754566449919e-05, "loss": 0.9944, "step": 40633 }, { "epoch": 0.04, "learning_rate": 4.979753568827633e-05, "loss": 1.0738, "step": 40634 }, { "epoch": 0.04, "learning_rate": 4.979752571180867e-05, "loss": 1.2131, "step": 40635 }, { "epoch": 0.04, "learning_rate": 4.979751573509622e-05, "loss": 1.0412, "step": 40636 }, { "epoch": 0.04, "learning_rate": 4.979750575813899e-05, "loss": 1.1766, "step": 40637 }, { "epoch": 0.04, "learning_rate": 4.979749578093695e-05, "loss": 1.3725, "step": 40638 }, { "epoch": 0.04, "learning_rate": 4.979748580349013e-05, "loss": 0.8553, "step": 40639 }, { "epoch": 0.04, "learning_rate": 4.9797475825798526e-05, "loss": 1.1618, "step": 40640 }, { "epoch": 0.04, "learning_rate": 4.979746584786212e-05, "loss": 0.8989, "step": 40641 }, { "epoch": 0.04, "learning_rate": 4.979745586968094e-05, "loss": 0.6319, "step": 40642 }, { "epoch": 0.04, "learning_rate": 4.9797445891254954e-05, "loss": 0.9079, "step": 40643 }, { "epoch": 0.04, "learning_rate": 4.979743591258419e-05, "loss": 0.8157, "step": 40644 }, { "epoch": 0.04, "learning_rate": 4.979742593366863e-05, "loss": 1.0585, "step": 40645 }, { "epoch": 0.04, "learning_rate": 4.9797415954508275e-05, "loss": 1.171, "step": 40646 }, { "epoch": 0.04, "learning_rate": 4.979740597510314e-05, "loss": 0.948, "step": 40647 }, { "epoch": 0.04, "learning_rate": 4.979739599545321e-05, "loss": 1.1799, "step": 40648 }, { "epoch": 0.04, "learning_rate": 4.979738601555849e-05, "loss": 1.1419, "step": 40649 }, { "epoch": 0.04, "learning_rate": 4.979737603541899e-05, "loss": 0.9446, "step": 40650 }, { "epoch": 0.04, "learning_rate": 4.979736605503469e-05, "loss": 1.2455, "step": 40651 }, { "epoch": 0.04, "learning_rate": 4.97973560744056e-05, "loss": 0.6968, "step": 40652 }, { "epoch": 0.04, "learning_rate": 4.979734609353173e-05, "loss": 0.8307, "step": 40653 }, { "epoch": 0.04, "learning_rate": 4.979733611241306e-05, "loss": 0.8944, "step": 40654 }, { "epoch": 0.04, "learning_rate": 4.9797326131049607e-05, "loss": 1.485, "step": 40655 }, { "epoch": 0.04, "learning_rate": 4.979731614944137e-05, "loss": 0.6272, "step": 40656 }, { "epoch": 0.04, "learning_rate": 4.979730616758834e-05, "loss": 1.0061, "step": 40657 }, { "epoch": 0.04, "learning_rate": 4.979729618549052e-05, "loss": 0.7641, "step": 40658 }, { "epoch": 0.04, "learning_rate": 4.979728620314792e-05, "loss": 0.8742, "step": 40659 }, { "epoch": 0.04, "learning_rate": 4.979727622056052e-05, "loss": 0.8394, "step": 40660 }, { "epoch": 0.04, "learning_rate": 4.9797266237728333e-05, "loss": 0.9611, "step": 40661 }, { "epoch": 0.04, "learning_rate": 4.979725625465137e-05, "loss": 0.9958, "step": 40662 }, { "epoch": 0.04, "learning_rate": 4.97972462713296e-05, "loss": 0.9443, "step": 40663 }, { "epoch": 0.04, "learning_rate": 4.9797236287763055e-05, "loss": 0.9862, "step": 40664 }, { "epoch": 0.04, "learning_rate": 4.979722630395172e-05, "loss": 0.7253, "step": 40665 }, { "epoch": 0.04, "learning_rate": 4.9797216319895594e-05, "loss": 0.5387, "step": 40666 }, { "epoch": 0.04, "learning_rate": 4.979720633559468e-05, "loss": 1.2267, "step": 40667 }, { "epoch": 0.04, "learning_rate": 4.9797196351048984e-05, "loss": 1.148, "step": 40668 }, { "epoch": 0.04, "learning_rate": 4.9797186366258496e-05, "loss": 1.2058, "step": 40669 }, { "epoch": 0.04, "learning_rate": 4.979717638122322e-05, "loss": 0.8128, "step": 40670 }, { "epoch": 0.04, "learning_rate": 4.9797166395943164e-05, "loss": 1.1351, "step": 40671 }, { "epoch": 0.04, "learning_rate": 4.979715641041832e-05, "loss": 1.1355, "step": 40672 }, { "epoch": 0.04, "learning_rate": 4.979714642464868e-05, "loss": 1.1695, "step": 40673 }, { "epoch": 0.04, "learning_rate": 4.979713643863426e-05, "loss": 1.1367, "step": 40674 }, { "epoch": 0.04, "learning_rate": 4.979712645237505e-05, "loss": 0.841, "step": 40675 }, { "epoch": 0.04, "learning_rate": 4.979711646587105e-05, "loss": 0.9031, "step": 40676 }, { "epoch": 0.04, "learning_rate": 4.979710647912228e-05, "loss": 1.1988, "step": 40677 }, { "epoch": 0.04, "learning_rate": 4.9797096492128704e-05, "loss": 1.1432, "step": 40678 }, { "epoch": 0.04, "learning_rate": 4.979708650489035e-05, "loss": 1.3378, "step": 40679 }, { "epoch": 0.04, "learning_rate": 4.97970765174072e-05, "loss": 1.2303, "step": 40680 }, { "epoch": 0.04, "learning_rate": 4.979706652967928e-05, "loss": 1.0918, "step": 40681 }, { "epoch": 0.04, "learning_rate": 4.979705654170656e-05, "loss": 1.2434, "step": 40682 }, { "epoch": 0.04, "learning_rate": 4.979704655348906e-05, "loss": 1.1039, "step": 40683 }, { "epoch": 0.04, "learning_rate": 4.9797036565026775e-05, "loss": 1.1193, "step": 40684 }, { "epoch": 0.04, "learning_rate": 4.979702657631971e-05, "loss": 1.0276, "step": 40685 }, { "epoch": 0.04, "learning_rate": 4.979701658736785e-05, "loss": 0.9569, "step": 40686 }, { "epoch": 0.04, "learning_rate": 4.9797006598171206e-05, "loss": 1.0207, "step": 40687 }, { "epoch": 0.04, "learning_rate": 4.9796996608729775e-05, "loss": 1.0177, "step": 40688 }, { "epoch": 0.04, "learning_rate": 4.979698661904356e-05, "loss": 0.5871, "step": 40689 }, { "epoch": 0.04, "learning_rate": 4.979697662911256e-05, "loss": 0.2233, "step": 40690 }, { "epoch": 0.04, "learning_rate": 4.979696663893677e-05, "loss": 0.182, "step": 40691 }, { "epoch": 0.04, "learning_rate": 4.9796956648516204e-05, "loss": 0.716, "step": 40692 }, { "epoch": 0.04, "learning_rate": 4.9796946657850846e-05, "loss": 1.1881, "step": 40693 }, { "epoch": 0.04, "learning_rate": 4.979693666694071e-05, "loss": 1.7227, "step": 40694 }, { "epoch": 0.04, "learning_rate": 4.9796926675785784e-05, "loss": 1.0681, "step": 40695 }, { "epoch": 0.04, "learning_rate": 4.979691668438607e-05, "loss": 0.8764, "step": 40696 }, { "epoch": 0.04, "learning_rate": 4.979690669274157e-05, "loss": 1.2415, "step": 40697 }, { "epoch": 0.04, "learning_rate": 4.979689670085229e-05, "loss": 1.4504, "step": 40698 }, { "epoch": 0.04, "learning_rate": 4.9796886708718226e-05, "loss": 0.8389, "step": 40699 }, { "epoch": 0.04, "learning_rate": 4.979687671633938e-05, "loss": 1.1211, "step": 40700 }, { "epoch": 0.04, "learning_rate": 4.979686672371575e-05, "loss": 0.8011, "step": 40701 }, { "epoch": 0.04, "learning_rate": 4.979685673084733e-05, "loss": 1.0731, "step": 40702 }, { "epoch": 0.04, "learning_rate": 4.979684673773413e-05, "loss": 1.0304, "step": 40703 }, { "epoch": 0.04, "learning_rate": 4.979683674437614e-05, "loss": 0.8665, "step": 40704 }, { "epoch": 0.04, "learning_rate": 4.9796826750773374e-05, "loss": 0.9637, "step": 40705 }, { "epoch": 0.04, "learning_rate": 4.979681675692582e-05, "loss": 0.3584, "step": 40706 }, { "epoch": 0.04, "learning_rate": 4.979680676283348e-05, "loss": 0.1462, "step": 40707 }, { "epoch": 0.04, "learning_rate": 4.9796796768496364e-05, "loss": 0.1672, "step": 40708 }, { "epoch": 0.04, "learning_rate": 4.9796786773914453e-05, "loss": 0.8188, "step": 40709 }, { "epoch": 0.04, "learning_rate": 4.979677677908777e-05, "loss": 1.0545, "step": 40710 }, { "epoch": 0.04, "learning_rate": 4.9796766784016296e-05, "loss": 0.8262, "step": 40711 }, { "epoch": 0.04, "learning_rate": 4.979675678870004e-05, "loss": 1.1985, "step": 40712 }, { "epoch": 0.04, "learning_rate": 4.9796746793139e-05, "loss": 0.9251, "step": 40713 }, { "epoch": 0.04, "learning_rate": 4.979673679733319e-05, "loss": 1.0197, "step": 40714 }, { "epoch": 0.04, "learning_rate": 4.979672680128258e-05, "loss": 0.4224, "step": 40715 }, { "epoch": 0.04, "learning_rate": 4.9796716804987195e-05, "loss": 0.8796, "step": 40716 }, { "epoch": 0.04, "learning_rate": 4.9796706808447026e-05, "loss": 0.9384, "step": 40717 }, { "epoch": 0.04, "learning_rate": 4.979669681166208e-05, "loss": 0.9805, "step": 40718 }, { "epoch": 0.04, "learning_rate": 4.9796686814632343e-05, "loss": 0.7653, "step": 40719 }, { "epoch": 0.04, "learning_rate": 4.9796676817357824e-05, "loss": 0.9247, "step": 40720 }, { "epoch": 0.04, "learning_rate": 4.9796666819838525e-05, "loss": 0.9989, "step": 40721 }, { "epoch": 0.04, "learning_rate": 4.979665682207445e-05, "loss": 0.8841, "step": 40722 }, { "epoch": 0.04, "learning_rate": 4.9796646824065584e-05, "loss": 0.997, "step": 40723 }, { "epoch": 0.04, "learning_rate": 4.9796636825811935e-05, "loss": 0.8537, "step": 40724 }, { "epoch": 0.04, "learning_rate": 4.979662682731351e-05, "loss": 0.9708, "step": 40725 }, { "epoch": 0.04, "learning_rate": 4.97966168285703e-05, "loss": 0.8826, "step": 40726 }, { "epoch": 0.04, "learning_rate": 4.979660682958231e-05, "loss": 1.089, "step": 40727 }, { "epoch": 0.04, "learning_rate": 4.979659683034954e-05, "loss": 0.7987, "step": 40728 }, { "epoch": 0.04, "learning_rate": 4.9796586830871984e-05, "loss": 1.0305, "step": 40729 }, { "epoch": 0.04, "learning_rate": 4.979657683114964e-05, "loss": 1.0633, "step": 40730 }, { "epoch": 0.04, "learning_rate": 4.9796566831182525e-05, "loss": 0.7876, "step": 40731 }, { "epoch": 0.04, "learning_rate": 4.979655683097063e-05, "loss": 0.7105, "step": 40732 }, { "epoch": 0.04, "learning_rate": 4.979654683051395e-05, "loss": 1.1563, "step": 40733 }, { "epoch": 0.04, "learning_rate": 4.979653682981249e-05, "loss": 0.6082, "step": 40734 }, { "epoch": 0.04, "learning_rate": 4.979652682886625e-05, "loss": 1.0015, "step": 40735 }, { "epoch": 0.04, "learning_rate": 4.979651682767523e-05, "loss": 1.0619, "step": 40736 }, { "epoch": 0.04, "learning_rate": 4.9796506826239427e-05, "loss": 1.1714, "step": 40737 }, { "epoch": 0.04, "learning_rate": 4.9796496824558845e-05, "loss": 0.6498, "step": 40738 }, { "epoch": 0.04, "learning_rate": 4.979648682263348e-05, "loss": 1.2718, "step": 40739 }, { "epoch": 0.04, "learning_rate": 4.979647682046333e-05, "loss": 1.0274, "step": 40740 }, { "epoch": 0.04, "learning_rate": 4.9796466818048405e-05, "loss": 1.2354, "step": 40741 }, { "epoch": 0.04, "learning_rate": 4.97964568153887e-05, "loss": 0.9614, "step": 40742 }, { "epoch": 0.04, "learning_rate": 4.979644681248422e-05, "loss": 0.9431, "step": 40743 }, { "epoch": 0.04, "learning_rate": 4.979643680933495e-05, "loss": 1.0266, "step": 40744 }, { "epoch": 0.04, "learning_rate": 4.9796426805940907e-05, "loss": 0.9752, "step": 40745 }, { "epoch": 0.04, "learning_rate": 4.979641680230208e-05, "loss": 1.1483, "step": 40746 }, { "epoch": 0.04, "learning_rate": 4.979640679841847e-05, "loss": 0.7543, "step": 40747 }, { "epoch": 0.04, "learning_rate": 4.979639679429009e-05, "loss": 0.8442, "step": 40748 }, { "epoch": 0.04, "learning_rate": 4.9796386789916924e-05, "loss": 0.8936, "step": 40749 }, { "epoch": 0.04, "learning_rate": 4.979637678529898e-05, "loss": 0.8388, "step": 40750 }, { "epoch": 0.04, "learning_rate": 4.979636678043625e-05, "loss": 1.0148, "step": 40751 }, { "epoch": 0.04, "learning_rate": 4.979635677532876e-05, "loss": 0.9627, "step": 40752 }, { "epoch": 0.04, "learning_rate": 4.979634676997648e-05, "loss": 1.2127, "step": 40753 }, { "epoch": 0.04, "learning_rate": 4.979633676437941e-05, "loss": 1.0287, "step": 40754 }, { "epoch": 0.04, "learning_rate": 4.979632675853757e-05, "loss": 1.0668, "step": 40755 }, { "epoch": 0.04, "learning_rate": 4.979631675245096e-05, "loss": 2.0564, "step": 40756 }, { "epoch": 0.04, "learning_rate": 4.9796306746119555e-05, "loss": 1.3779, "step": 40757 }, { "epoch": 0.04, "learning_rate": 4.979629673954338e-05, "loss": 1.2846, "step": 40758 }, { "epoch": 0.04, "learning_rate": 4.979628673272243e-05, "loss": 1.086, "step": 40759 }, { "epoch": 0.04, "learning_rate": 4.979627672565669e-05, "loss": 1.4195, "step": 40760 }, { "epoch": 0.04, "learning_rate": 4.979626671834618e-05, "loss": 1.0227, "step": 40761 }, { "epoch": 0.04, "learning_rate": 4.979625671079089e-05, "loss": 1.2238, "step": 40762 }, { "epoch": 0.04, "learning_rate": 4.979624670299082e-05, "loss": 1.0829, "step": 40763 }, { "epoch": 0.04, "learning_rate": 4.9796236694945974e-05, "loss": 1.0518, "step": 40764 }, { "epoch": 0.04, "learning_rate": 4.979622668665635e-05, "loss": 1.047, "step": 40765 }, { "epoch": 0.04, "learning_rate": 4.9796216678121946e-05, "loss": 1.1603, "step": 40766 }, { "epoch": 0.04, "learning_rate": 4.979620666934277e-05, "loss": 0.9955, "step": 40767 }, { "epoch": 0.04, "learning_rate": 4.979619666031881e-05, "loss": 0.8709, "step": 40768 }, { "epoch": 0.04, "learning_rate": 4.979618665105007e-05, "loss": 1.0375, "step": 40769 }, { "epoch": 0.04, "learning_rate": 4.9796176641536565e-05, "loss": 0.8333, "step": 40770 }, { "epoch": 0.04, "learning_rate": 4.979616663177828e-05, "loss": 0.9025, "step": 40771 }, { "epoch": 0.04, "learning_rate": 4.97961566217752e-05, "loss": 0.8363, "step": 40772 }, { "epoch": 0.04, "learning_rate": 4.9796146611527364e-05, "loss": 0.5365, "step": 40773 }, { "epoch": 0.04, "learning_rate": 4.979613660103474e-05, "loss": 0.7728, "step": 40774 }, { "epoch": 0.04, "learning_rate": 4.979612659029734e-05, "loss": 0.6163, "step": 40775 }, { "epoch": 0.04, "learning_rate": 4.979611657931517e-05, "loss": 0.5488, "step": 40776 }, { "epoch": 0.04, "learning_rate": 4.9796106568088205e-05, "loss": 0.7245, "step": 40777 }, { "epoch": 0.04, "learning_rate": 4.979609655661648e-05, "loss": 0.7005, "step": 40778 }, { "epoch": 0.04, "learning_rate": 4.979608654489998e-05, "loss": 0.9557, "step": 40779 }, { "epoch": 0.04, "learning_rate": 4.9796076532938694e-05, "loss": 1.1047, "step": 40780 }, { "epoch": 0.04, "learning_rate": 4.979606652073263e-05, "loss": 0.6328, "step": 40781 }, { "epoch": 0.04, "learning_rate": 4.9796056508281794e-05, "loss": 0.0855, "step": 40782 }, { "epoch": 0.04, "learning_rate": 4.979604649558619e-05, "loss": 0.0651, "step": 40783 }, { "epoch": 0.04, "learning_rate": 4.9796036482645805e-05, "loss": 0.2659, "step": 40784 }, { "epoch": 0.04, "learning_rate": 4.979602646946063e-05, "loss": 0.9159, "step": 40785 }, { "epoch": 0.04, "learning_rate": 4.97960164560307e-05, "loss": 0.9664, "step": 40786 }, { "epoch": 0.04, "learning_rate": 4.979600644235598e-05, "loss": 0.6937, "step": 40787 }, { "epoch": 0.04, "learning_rate": 4.9795996428436486e-05, "loss": 0.851, "step": 40788 }, { "epoch": 0.04, "learning_rate": 4.9795986414272224e-05, "loss": 1.1629, "step": 40789 }, { "epoch": 0.04, "learning_rate": 4.9795976399863184e-05, "loss": 1.0847, "step": 40790 }, { "epoch": 0.04, "learning_rate": 4.9795966385209364e-05, "loss": 1.027, "step": 40791 }, { "epoch": 0.04, "learning_rate": 4.979595637031077e-05, "loss": 0.847, "step": 40792 }, { "epoch": 0.04, "learning_rate": 4.97959463551674e-05, "loss": 0.9571, "step": 40793 }, { "epoch": 0.04, "learning_rate": 4.979593633977926e-05, "loss": 0.937, "step": 40794 }, { "epoch": 0.04, "learning_rate": 4.979592632414634e-05, "loss": 1.0992, "step": 40795 }, { "epoch": 0.04, "learning_rate": 4.979591630826865e-05, "loss": 1.083, "step": 40796 }, { "epoch": 0.04, "learning_rate": 4.9795906292146175e-05, "loss": 1.2643, "step": 40797 }, { "epoch": 0.04, "learning_rate": 4.9795896275778936e-05, "loss": 1.0904, "step": 40798 }, { "epoch": 0.04, "learning_rate": 4.979588625916692e-05, "loss": 0.9749, "step": 40799 }, { "epoch": 0.04, "learning_rate": 4.979587624231012e-05, "loss": 0.9231, "step": 40800 }, { "epoch": 0.04, "learning_rate": 4.979586622520856e-05, "loss": 0.9092, "step": 40801 }, { "epoch": 0.04, "learning_rate": 4.979585620786222e-05, "loss": 0.7054, "step": 40802 }, { "epoch": 0.04, "learning_rate": 4.97958461902711e-05, "loss": 1.0353, "step": 40803 }, { "epoch": 0.04, "learning_rate": 4.979583617243522e-05, "loss": 1.3114, "step": 40804 }, { "epoch": 0.04, "learning_rate": 4.979582615435455e-05, "loss": 1.3163, "step": 40805 }, { "epoch": 0.04, "learning_rate": 4.979581613602912e-05, "loss": 0.8458, "step": 40806 }, { "epoch": 0.04, "learning_rate": 4.9795806117458904e-05, "loss": 1.0279, "step": 40807 }, { "epoch": 0.04, "learning_rate": 4.979579609864392e-05, "loss": 0.8615, "step": 40808 }, { "epoch": 0.04, "learning_rate": 4.979578607958416e-05, "loss": 0.9482, "step": 40809 }, { "epoch": 0.04, "learning_rate": 4.979577606027963e-05, "loss": 0.9214, "step": 40810 }, { "epoch": 0.04, "learning_rate": 4.979576604073032e-05, "loss": 1.188, "step": 40811 }, { "epoch": 0.04, "learning_rate": 4.9795756020936244e-05, "loss": 1.1021, "step": 40812 }, { "epoch": 0.04, "learning_rate": 4.9795746000897395e-05, "loss": 1.0787, "step": 40813 }, { "epoch": 0.04, "learning_rate": 4.979573598061377e-05, "loss": 0.9858, "step": 40814 }, { "epoch": 0.04, "learning_rate": 4.9795725960085374e-05, "loss": 1.016, "step": 40815 }, { "epoch": 0.04, "learning_rate": 4.9795715939312195e-05, "loss": 1.0312, "step": 40816 }, { "epoch": 0.04, "learning_rate": 4.979570591829426e-05, "loss": 0.8616, "step": 40817 }, { "epoch": 0.04, "learning_rate": 4.9795695897031535e-05, "loss": 1.1016, "step": 40818 }, { "epoch": 0.04, "learning_rate": 4.9795685875524046e-05, "loss": 1.268, "step": 40819 }, { "epoch": 0.04, "learning_rate": 4.9795675853771785e-05, "loss": 0.9939, "step": 40820 }, { "epoch": 0.04, "learning_rate": 4.979566583177475e-05, "loss": 0.9836, "step": 40821 }, { "epoch": 0.04, "learning_rate": 4.979565580953294e-05, "loss": 0.8793, "step": 40822 }, { "epoch": 0.04, "learning_rate": 4.979564578704636e-05, "loss": 1.5872, "step": 40823 }, { "epoch": 0.04, "learning_rate": 4.9795635764315014e-05, "loss": 1.3929, "step": 40824 }, { "epoch": 0.04, "learning_rate": 4.979562574133889e-05, "loss": 1.4169, "step": 40825 }, { "epoch": 0.04, "learning_rate": 4.979561571811799e-05, "loss": 1.0343, "step": 40826 }, { "epoch": 0.04, "learning_rate": 4.9795605694652325e-05, "loss": 1.2037, "step": 40827 }, { "epoch": 0.04, "learning_rate": 4.9795595670941887e-05, "loss": 1.0374, "step": 40828 }, { "epoch": 0.04, "learning_rate": 4.9795585646986676e-05, "loss": 1.1486, "step": 40829 }, { "epoch": 0.04, "learning_rate": 4.979557562278669e-05, "loss": 0.868, "step": 40830 }, { "epoch": 0.04, "learning_rate": 4.9795565598341945e-05, "loss": 0.9121, "step": 40831 }, { "epoch": 0.04, "learning_rate": 4.979555557365241e-05, "loss": 1.0718, "step": 40832 }, { "epoch": 0.04, "learning_rate": 4.979554554871811e-05, "loss": 0.8771, "step": 40833 }, { "epoch": 0.04, "learning_rate": 4.979553552353905e-05, "loss": 1.5028, "step": 40834 }, { "epoch": 0.04, "learning_rate": 4.979552549811522e-05, "loss": 1.1552, "step": 40835 }, { "epoch": 0.04, "learning_rate": 4.97955154724466e-05, "loss": 0.9703, "step": 40836 }, { "epoch": 0.04, "learning_rate": 4.979550544653322e-05, "loss": 0.8517, "step": 40837 }, { "epoch": 0.04, "learning_rate": 4.979549542037507e-05, "loss": 0.9362, "step": 40838 }, { "epoch": 0.04, "learning_rate": 4.9795485393972154e-05, "loss": 0.8521, "step": 40839 }, { "epoch": 0.04, "learning_rate": 4.9795475367324456e-05, "loss": 0.5898, "step": 40840 }, { "epoch": 0.04, "learning_rate": 4.979546534043199e-05, "loss": 0.7565, "step": 40841 }, { "epoch": 0.04, "learning_rate": 4.979545531329476e-05, "loss": 0.6696, "step": 40842 }, { "epoch": 0.04, "learning_rate": 4.979544528591276e-05, "loss": 0.8595, "step": 40843 }, { "epoch": 0.04, "learning_rate": 4.9795435258285984e-05, "loss": 1.0117, "step": 40844 }, { "epoch": 0.04, "learning_rate": 4.979542523041444e-05, "loss": 0.9242, "step": 40845 }, { "epoch": 0.04, "learning_rate": 4.979541520229812e-05, "loss": 0.4601, "step": 40846 }, { "epoch": 0.04, "learning_rate": 4.979540517393704e-05, "loss": 1.0551, "step": 40847 }, { "epoch": 0.04, "learning_rate": 4.979539514533118e-05, "loss": 1.2189, "step": 40848 }, { "epoch": 0.04, "learning_rate": 4.979538511648056e-05, "loss": 1.1547, "step": 40849 }, { "epoch": 0.04, "learning_rate": 4.979537508738516e-05, "loss": 1.0173, "step": 40850 }, { "epoch": 0.04, "learning_rate": 4.9795365058045005e-05, "loss": 1.1752, "step": 40851 }, { "epoch": 0.04, "learning_rate": 4.9795355028460074e-05, "loss": 1.1225, "step": 40852 }, { "epoch": 0.04, "learning_rate": 4.979534499863038e-05, "loss": 0.8081, "step": 40853 }, { "epoch": 0.04, "learning_rate": 4.9795334968555904e-05, "loss": 0.9704, "step": 40854 }, { "epoch": 0.04, "learning_rate": 4.9795324938236664e-05, "loss": 1.064, "step": 40855 }, { "epoch": 0.04, "learning_rate": 4.979531490767265e-05, "loss": 1.2471, "step": 40856 }, { "epoch": 0.04, "learning_rate": 4.979530487686388e-05, "loss": 0.8457, "step": 40857 }, { "epoch": 0.04, "learning_rate": 4.979529484581033e-05, "loss": 0.9813, "step": 40858 }, { "epoch": 0.04, "learning_rate": 4.9795284814512015e-05, "loss": 1.0597, "step": 40859 }, { "epoch": 0.04, "learning_rate": 4.9795274782968935e-05, "loss": 0.857, "step": 40860 }, { "epoch": 0.04, "learning_rate": 4.979526475118108e-05, "loss": 0.5628, "step": 40861 }, { "epoch": 0.04, "learning_rate": 4.979525471914846e-05, "loss": 0.6094, "step": 40862 }, { "epoch": 0.04, "learning_rate": 4.979524468687107e-05, "loss": 1.1697, "step": 40863 }, { "epoch": 0.04, "learning_rate": 4.979523465434892e-05, "loss": 0.8584, "step": 40864 }, { "epoch": 0.04, "learning_rate": 4.9795224621581995e-05, "loss": 0.8765, "step": 40865 }, { "epoch": 0.04, "learning_rate": 4.9795214588570295e-05, "loss": 1.021, "step": 40866 }, { "epoch": 0.04, "learning_rate": 4.9795204555313836e-05, "loss": 0.8593, "step": 40867 }, { "epoch": 0.04, "learning_rate": 4.9795194521812605e-05, "loss": 0.5363, "step": 40868 }, { "epoch": 0.04, "learning_rate": 4.9795184488066615e-05, "loss": 1.1051, "step": 40869 }, { "epoch": 0.04, "learning_rate": 4.9795174454075846e-05, "loss": 0.7458, "step": 40870 }, { "epoch": 0.04, "learning_rate": 4.979516441984031e-05, "loss": 1.0879, "step": 40871 }, { "epoch": 0.04, "learning_rate": 4.979515438536002e-05, "loss": 1.6994, "step": 40872 }, { "epoch": 0.04, "learning_rate": 4.979514435063495e-05, "loss": 0.9293, "step": 40873 }, { "epoch": 0.04, "learning_rate": 4.979513431566512e-05, "loss": 0.9087, "step": 40874 }, { "epoch": 0.04, "learning_rate": 4.9795124280450514e-05, "loss": 0.8977, "step": 40875 }, { "epoch": 0.04, "learning_rate": 4.9795114244991146e-05, "loss": 1.1037, "step": 40876 }, { "epoch": 0.04, "learning_rate": 4.979510420928701e-05, "loss": 0.9806, "step": 40877 }, { "epoch": 0.04, "learning_rate": 4.9795094173338114e-05, "loss": 1.0297, "step": 40878 }, { "epoch": 0.04, "learning_rate": 4.979508413714444e-05, "loss": 1.2711, "step": 40879 }, { "epoch": 0.04, "learning_rate": 4.9795074100706006e-05, "loss": 0.8142, "step": 40880 }, { "epoch": 0.04, "learning_rate": 4.979506406402281e-05, "loss": 0.9367, "step": 40881 }, { "epoch": 0.04, "learning_rate": 4.9795054027094844e-05, "loss": 2.0353, "step": 40882 }, { "epoch": 0.04, "learning_rate": 4.979504398992211e-05, "loss": 1.5174, "step": 40883 }, { "epoch": 0.04, "learning_rate": 4.979503395250461e-05, "loss": 0.8455, "step": 40884 }, { "epoch": 0.04, "learning_rate": 4.979502391484234e-05, "loss": 0.4455, "step": 40885 }, { "epoch": 0.04, "learning_rate": 4.9795013876935307e-05, "loss": 1.2014, "step": 40886 }, { "epoch": 0.04, "learning_rate": 4.979500383878351e-05, "loss": 0.933, "step": 40887 }, { "epoch": 0.04, "learning_rate": 4.9794993800386945e-05, "loss": 0.9475, "step": 40888 }, { "epoch": 0.04, "learning_rate": 4.979498376174562e-05, "loss": 1.127, "step": 40889 }, { "epoch": 0.04, "learning_rate": 4.9794973722859516e-05, "loss": 0.7486, "step": 40890 }, { "epoch": 0.04, "learning_rate": 4.979496368372866e-05, "loss": 1.0076, "step": 40891 }, { "epoch": 0.04, "learning_rate": 4.979495364435303e-05, "loss": 1.1111, "step": 40892 }, { "epoch": 0.04, "learning_rate": 4.979494360473264e-05, "loss": 0.7144, "step": 40893 }, { "epoch": 0.04, "learning_rate": 4.9794933564867486e-05, "loss": 0.8178, "step": 40894 }, { "epoch": 0.04, "learning_rate": 4.979492352475756e-05, "loss": 0.6377, "step": 40895 }, { "epoch": 0.04, "learning_rate": 4.979491348440287e-05, "loss": 0.7227, "step": 40896 }, { "epoch": 0.04, "learning_rate": 4.979490344380342e-05, "loss": 0.8179, "step": 40897 }, { "epoch": 0.04, "learning_rate": 4.979489340295921e-05, "loss": 0.6143, "step": 40898 }, { "epoch": 0.04, "learning_rate": 4.979488336187023e-05, "loss": 0.3968, "step": 40899 }, { "epoch": 0.04, "learning_rate": 4.979487332053648e-05, "loss": 0.2904, "step": 40900 }, { "epoch": 0.04, "learning_rate": 4.979486327895797e-05, "loss": 0.2312, "step": 40901 }, { "epoch": 0.04, "learning_rate": 4.9794853237134685e-05, "loss": 0.1676, "step": 40902 }, { "epoch": 0.04, "learning_rate": 4.979484319506666e-05, "loss": 0.4419, "step": 40903 }, { "epoch": 0.04, "learning_rate": 4.979483315275385e-05, "loss": 0.8777, "step": 40904 }, { "epoch": 0.04, "learning_rate": 4.979482311019628e-05, "loss": 1.0197, "step": 40905 }, { "epoch": 0.04, "learning_rate": 4.979481306739395e-05, "loss": 1.1007, "step": 40906 }, { "epoch": 0.04, "learning_rate": 4.9794803024346845e-05, "loss": 1.0292, "step": 40907 }, { "epoch": 0.04, "learning_rate": 4.9794792981055e-05, "loss": 0.8755, "step": 40908 }, { "epoch": 0.04, "learning_rate": 4.979478293751837e-05, "loss": 1.2952, "step": 40909 }, { "epoch": 0.04, "learning_rate": 4.979477289373698e-05, "loss": 1.0761, "step": 40910 }, { "epoch": 0.04, "learning_rate": 4.979476284971083e-05, "loss": 1.0876, "step": 40911 }, { "epoch": 0.04, "learning_rate": 4.979475280543992e-05, "loss": 1.0971, "step": 40912 }, { "epoch": 0.04, "learning_rate": 4.979474276092424e-05, "loss": 1.36, "step": 40913 }, { "epoch": 0.04, "learning_rate": 4.9794732716163796e-05, "loss": 1.4432, "step": 40914 }, { "epoch": 0.04, "learning_rate": 4.979472267115859e-05, "loss": 0.7877, "step": 40915 }, { "epoch": 0.04, "learning_rate": 4.9794712625908626e-05, "loss": 0.8794, "step": 40916 }, { "epoch": 0.04, "learning_rate": 4.9794702580413896e-05, "loss": 0.9831, "step": 40917 }, { "epoch": 0.04, "learning_rate": 4.97946925346744e-05, "loss": 0.5918, "step": 40918 }, { "epoch": 0.04, "learning_rate": 4.979468248869015e-05, "loss": 1.021, "step": 40919 }, { "epoch": 0.04, "learning_rate": 4.979467244246113e-05, "loss": 0.9207, "step": 40920 }, { "epoch": 0.04, "learning_rate": 4.979466239598734e-05, "loss": 0.9763, "step": 40921 }, { "epoch": 0.04, "learning_rate": 4.97946523492688e-05, "loss": 1.0186, "step": 40922 }, { "epoch": 0.04, "learning_rate": 4.97946423023055e-05, "loss": 1.014, "step": 40923 }, { "epoch": 0.04, "learning_rate": 4.979463225509743e-05, "loss": 1.7229, "step": 40924 }, { "epoch": 0.04, "learning_rate": 4.97946222076446e-05, "loss": 4.5377, "step": 40925 }, { "epoch": 0.04, "learning_rate": 4.9794612159947006e-05, "loss": 3.909, "step": 40926 }, { "epoch": 0.04, "learning_rate": 4.979460211200465e-05, "loss": 1.2429, "step": 40927 }, { "epoch": 0.04, "learning_rate": 4.979459206381753e-05, "loss": 0.9186, "step": 40928 }, { "epoch": 0.04, "learning_rate": 4.979458201538566e-05, "loss": 0.9709, "step": 40929 }, { "epoch": 0.04, "learning_rate": 4.979457196670902e-05, "loss": 1.4263, "step": 40930 }, { "epoch": 0.04, "learning_rate": 4.979456191778762e-05, "loss": 1.1351, "step": 40931 }, { "epoch": 0.04, "learning_rate": 4.979455186862145e-05, "loss": 1.4749, "step": 40932 }, { "epoch": 0.04, "learning_rate": 4.9794541819210535e-05, "loss": 1.1656, "step": 40933 }, { "epoch": 0.04, "learning_rate": 4.979453176955485e-05, "loss": 0.9613, "step": 40934 }, { "epoch": 0.04, "learning_rate": 4.9794521719654394e-05, "loss": 0.9406, "step": 40935 }, { "epoch": 0.04, "learning_rate": 4.979451166950919e-05, "loss": 1.0009, "step": 40936 }, { "epoch": 0.04, "learning_rate": 4.9794501619119226e-05, "loss": 1.135, "step": 40937 }, { "epoch": 0.04, "learning_rate": 4.979449156848449e-05, "loss": 1.0691, "step": 40938 }, { "epoch": 0.04, "learning_rate": 4.979448151760501e-05, "loss": 1.1483, "step": 40939 }, { "epoch": 0.04, "learning_rate": 4.979447146648075e-05, "loss": 1.0791, "step": 40940 }, { "epoch": 0.04, "learning_rate": 4.9794461415111744e-05, "loss": 1.2825, "step": 40941 }, { "epoch": 0.04, "learning_rate": 4.9794451363497974e-05, "loss": 0.9814, "step": 40942 }, { "epoch": 0.04, "learning_rate": 4.979444131163944e-05, "loss": 0.9849, "step": 40943 }, { "epoch": 0.04, "learning_rate": 4.9794431259536144e-05, "loss": 1.0751, "step": 40944 }, { "epoch": 0.04, "learning_rate": 4.979442120718809e-05, "loss": 1.2128, "step": 40945 }, { "epoch": 0.04, "learning_rate": 4.9794411154595286e-05, "loss": 1.6338, "step": 40946 }, { "epoch": 0.04, "learning_rate": 4.979440110175771e-05, "loss": 1.326, "step": 40947 }, { "epoch": 0.04, "learning_rate": 4.979439104867538e-05, "loss": 1.1183, "step": 40948 }, { "epoch": 0.04, "learning_rate": 4.9794380995348285e-05, "loss": 0.8999, "step": 40949 }, { "epoch": 0.04, "learning_rate": 4.979437094177644e-05, "loss": 1.1321, "step": 40950 }, { "epoch": 0.04, "learning_rate": 4.979436088795982e-05, "loss": 1.1497, "step": 40951 }, { "epoch": 0.04, "learning_rate": 4.979435083389845e-05, "loss": 1.0581, "step": 40952 }, { "epoch": 0.04, "learning_rate": 4.979434077959232e-05, "loss": 0.9533, "step": 40953 }, { "epoch": 0.04, "learning_rate": 4.9794330725041427e-05, "loss": 1.0292, "step": 40954 }, { "epoch": 0.04, "learning_rate": 4.979432067024578e-05, "loss": 1.0483, "step": 40955 }, { "epoch": 0.04, "learning_rate": 4.9794310615205375e-05, "loss": 1.1424, "step": 40956 }, { "epoch": 0.04, "learning_rate": 4.9794300559920205e-05, "loss": 0.9158, "step": 40957 }, { "epoch": 0.04, "learning_rate": 4.979429050439028e-05, "loss": 0.9945, "step": 40958 }, { "epoch": 0.04, "learning_rate": 4.9794280448615595e-05, "loss": 1.3328, "step": 40959 }, { "epoch": 0.04, "learning_rate": 4.9794270392596156e-05, "loss": 1.2378, "step": 40960 }, { "epoch": 0.04, "learning_rate": 4.979426033633195e-05, "loss": 0.9737, "step": 40961 }, { "epoch": 0.04, "learning_rate": 4.979425027982299e-05, "loss": 1.0008, "step": 40962 }, { "epoch": 0.04, "learning_rate": 4.979424022306927e-05, "loss": 0.9552, "step": 40963 }, { "epoch": 0.04, "learning_rate": 4.979423016607079e-05, "loss": 1.2045, "step": 40964 }, { "epoch": 0.04, "learning_rate": 4.979422010882756e-05, "loss": 1.003, "step": 40965 }, { "epoch": 0.04, "learning_rate": 4.979421005133956e-05, "loss": 0.8391, "step": 40966 }, { "epoch": 0.04, "learning_rate": 4.979419999360682e-05, "loss": 1.1561, "step": 40967 }, { "epoch": 0.04, "learning_rate": 4.97941899356293e-05, "loss": 0.929, "step": 40968 }, { "epoch": 0.04, "learning_rate": 4.9794179877407035e-05, "loss": 0.7954, "step": 40969 }, { "epoch": 0.04, "learning_rate": 4.979416981894001e-05, "loss": 0.9071, "step": 40970 }, { "epoch": 0.04, "learning_rate": 4.979415976022823e-05, "loss": 1.4266, "step": 40971 }, { "epoch": 0.04, "learning_rate": 4.9794149701271695e-05, "loss": 1.3173, "step": 40972 }, { "epoch": 0.04, "learning_rate": 4.979413964207039e-05, "loss": 0.8479, "step": 40973 }, { "epoch": 0.04, "learning_rate": 4.979412958262434e-05, "loss": 1.2729, "step": 40974 }, { "epoch": 0.04, "learning_rate": 4.979411952293353e-05, "loss": 1.157, "step": 40975 }, { "epoch": 0.04, "learning_rate": 4.979410946299796e-05, "loss": 1.0183, "step": 40976 }, { "epoch": 0.04, "learning_rate": 4.9794099402817627e-05, "loss": 1.0359, "step": 40977 }, { "epoch": 0.04, "learning_rate": 4.9794089342392545e-05, "loss": 1.1713, "step": 40978 }, { "epoch": 0.04, "learning_rate": 4.979407928172271e-05, "loss": 0.9039, "step": 40979 }, { "epoch": 0.04, "learning_rate": 4.979406922080811e-05, "loss": 0.9003, "step": 40980 }, { "epoch": 0.04, "learning_rate": 4.979405915964876e-05, "loss": 2.0488, "step": 40981 }, { "epoch": 0.04, "learning_rate": 4.979404909824465e-05, "loss": 1.7172, "step": 40982 }, { "epoch": 0.04, "learning_rate": 4.9794039036595785e-05, "loss": 1.0283, "step": 40983 }, { "epoch": 0.04, "learning_rate": 4.9794028974702166e-05, "loss": 1.0857, "step": 40984 }, { "epoch": 0.04, "learning_rate": 4.979401891256379e-05, "loss": 1.1042, "step": 40985 }, { "epoch": 0.04, "learning_rate": 4.979400885018065e-05, "loss": 1.1553, "step": 40986 }, { "epoch": 0.04, "learning_rate": 4.979399878755276e-05, "loss": 1.2124, "step": 40987 }, { "epoch": 0.04, "learning_rate": 4.9793988724680114e-05, "loss": 1.229, "step": 40988 }, { "epoch": 0.04, "learning_rate": 4.9793978661562715e-05, "loss": 0.9783, "step": 40989 }, { "epoch": 0.04, "learning_rate": 4.9793968598200557e-05, "loss": 0.8406, "step": 40990 }, { "epoch": 0.04, "learning_rate": 4.979395853459365e-05, "loss": 0.8865, "step": 40991 }, { "epoch": 0.04, "learning_rate": 4.979394847074197e-05, "loss": 0.9932, "step": 40992 }, { "epoch": 0.04, "learning_rate": 4.9793938406645544e-05, "loss": 1.0827, "step": 40993 }, { "epoch": 0.04, "learning_rate": 4.979392834230437e-05, "loss": 0.8926, "step": 40994 }, { "epoch": 0.04, "learning_rate": 4.9793918277718434e-05, "loss": 1.0067, "step": 40995 }, { "epoch": 0.04, "learning_rate": 4.979390821288774e-05, "loss": 0.9064, "step": 40996 }, { "epoch": 0.04, "learning_rate": 4.97938981478123e-05, "loss": 1.0775, "step": 40997 }, { "epoch": 0.04, "learning_rate": 4.97938880824921e-05, "loss": 1.1318, "step": 40998 }, { "epoch": 0.04, "learning_rate": 4.9793878016927145e-05, "loss": 0.9529, "step": 40999 }, { "epoch": 0.04, "learning_rate": 4.9793867951117434e-05, "loss": 1.0216, "step": 41000 }, { "epoch": 0.04, "eval_loss": 1.0827760696411133, "eval_runtime": 130.4454, "eval_samples_per_second": 10.617, "eval_steps_per_second": 5.313, "step": 41000 }, { "epoch": 0.04, "learning_rate": 4.9793857885062965e-05, "loss": 0.939, "step": 41001 }, { "epoch": 0.04, "learning_rate": 4.979384781876375e-05, "loss": 1.2295, "step": 41002 }, { "epoch": 0.04, "learning_rate": 4.979383775221978e-05, "loss": 1.3058, "step": 41003 }, { "epoch": 0.04, "learning_rate": 4.9793827685431046e-05, "loss": 1.3647, "step": 41004 }, { "epoch": 0.04, "learning_rate": 4.979381761839757e-05, "loss": 1.0407, "step": 41005 }, { "epoch": 0.04, "learning_rate": 4.9793807551119334e-05, "loss": 1.1236, "step": 41006 }, { "epoch": 0.04, "learning_rate": 4.979379748359635e-05, "loss": 1.0465, "step": 41007 }, { "epoch": 0.04, "learning_rate": 4.9793787415828595e-05, "loss": 0.9899, "step": 41008 }, { "epoch": 0.04, "learning_rate": 4.9793777347816104e-05, "loss": 0.9079, "step": 41009 }, { "epoch": 0.04, "learning_rate": 4.9793767279558854e-05, "loss": 0.8553, "step": 41010 }, { "epoch": 0.04, "learning_rate": 4.9793757211056846e-05, "loss": 1.5017, "step": 41011 }, { "epoch": 0.04, "learning_rate": 4.9793747142310086e-05, "loss": 1.4316, "step": 41012 }, { "epoch": 0.04, "learning_rate": 4.979373707331858e-05, "loss": 1.1193, "step": 41013 }, { "epoch": 0.04, "learning_rate": 4.979372700408231e-05, "loss": 1.1392, "step": 41014 }, { "epoch": 0.04, "learning_rate": 4.9793716934601295e-05, "loss": 1.0666, "step": 41015 }, { "epoch": 0.04, "learning_rate": 4.979370686487552e-05, "loss": 0.9475, "step": 41016 }, { "epoch": 0.04, "learning_rate": 4.9793696794905e-05, "loss": 0.9462, "step": 41017 }, { "epoch": 0.04, "learning_rate": 4.979368672468972e-05, "loss": 0.7857, "step": 41018 }, { "epoch": 0.04, "learning_rate": 4.979367665422969e-05, "loss": 0.8213, "step": 41019 }, { "epoch": 0.04, "learning_rate": 4.979366658352491e-05, "loss": 1.3567, "step": 41020 }, { "epoch": 0.04, "learning_rate": 4.979365651257537e-05, "loss": 0.8561, "step": 41021 }, { "epoch": 0.04, "learning_rate": 4.979364644138108e-05, "loss": 1.0883, "step": 41022 }, { "epoch": 0.04, "learning_rate": 4.979363636994204e-05, "loss": 1.0243, "step": 41023 }, { "epoch": 0.04, "learning_rate": 4.979362629825825e-05, "loss": 1.2735, "step": 41024 }, { "epoch": 0.04, "learning_rate": 4.979361622632971e-05, "loss": 1.1357, "step": 41025 }, { "epoch": 0.04, "learning_rate": 4.9793606154156405e-05, "loss": 0.9827, "step": 41026 }, { "epoch": 0.04, "learning_rate": 4.979359608173836e-05, "loss": 0.9403, "step": 41027 }, { "epoch": 0.04, "learning_rate": 4.979358600907556e-05, "loss": 0.9605, "step": 41028 }, { "epoch": 0.04, "learning_rate": 4.9793575936168e-05, "loss": 0.945, "step": 41029 }, { "epoch": 0.04, "learning_rate": 4.97935658630157e-05, "loss": 0.8344, "step": 41030 }, { "epoch": 0.04, "learning_rate": 4.979355578961864e-05, "loss": 0.7077, "step": 41031 }, { "epoch": 0.04, "learning_rate": 4.979354571597684e-05, "loss": 1.1085, "step": 41032 }, { "epoch": 0.04, "learning_rate": 4.9793535642090276e-05, "loss": 1.1478, "step": 41033 }, { "epoch": 0.04, "learning_rate": 4.979352556795897e-05, "loss": 0.4995, "step": 41034 }, { "epoch": 0.04, "learning_rate": 4.97935154935829e-05, "loss": 0.3819, "step": 41035 }, { "epoch": 0.04, "learning_rate": 4.97935054189621e-05, "loss": 0.7369, "step": 41036 }, { "epoch": 0.04, "learning_rate": 4.979349534409653e-05, "loss": 0.9669, "step": 41037 }, { "epoch": 0.04, "learning_rate": 4.979348526898621e-05, "loss": 1.6522, "step": 41038 }, { "epoch": 0.04, "learning_rate": 4.9793475193631144e-05, "loss": 1.0046, "step": 41039 }, { "epoch": 0.04, "learning_rate": 4.979346511803133e-05, "loss": 1.3287, "step": 41040 }, { "epoch": 0.04, "learning_rate": 4.979345504218677e-05, "loss": 1.13, "step": 41041 }, { "epoch": 0.04, "learning_rate": 4.9793444966097445e-05, "loss": 0.9385, "step": 41042 }, { "epoch": 0.04, "learning_rate": 4.979343488976338e-05, "loss": 0.9134, "step": 41043 }, { "epoch": 0.04, "learning_rate": 4.979342481318456e-05, "loss": 1.1597, "step": 41044 }, { "epoch": 0.04, "learning_rate": 4.9793414736360986e-05, "loss": 2.0257, "step": 41045 }, { "epoch": 0.04, "learning_rate": 4.979340465929267e-05, "loss": 1.4026, "step": 41046 }, { "epoch": 0.04, "learning_rate": 4.97933945819796e-05, "loss": 0.9137, "step": 41047 }, { "epoch": 0.04, "learning_rate": 4.979338450442178e-05, "loss": 0.83, "step": 41048 }, { "epoch": 0.04, "learning_rate": 4.979337442661921e-05, "loss": 1.0749, "step": 41049 }, { "epoch": 0.04, "learning_rate": 4.979336434857189e-05, "loss": 1.1871, "step": 41050 }, { "epoch": 0.04, "learning_rate": 4.979335427027982e-05, "loss": 1.1551, "step": 41051 }, { "epoch": 0.04, "learning_rate": 4.979334419174301e-05, "loss": 1.0977, "step": 41052 }, { "epoch": 0.04, "learning_rate": 4.979333411296144e-05, "loss": 1.008, "step": 41053 }, { "epoch": 0.04, "learning_rate": 4.979332403393513e-05, "loss": 1.0999, "step": 41054 }, { "epoch": 0.04, "learning_rate": 4.9793313954664054e-05, "loss": 1.3, "step": 41055 }, { "epoch": 0.04, "learning_rate": 4.9793303875148246e-05, "loss": 1.0094, "step": 41056 }, { "epoch": 0.04, "learning_rate": 4.979329379538768e-05, "loss": 0.9381, "step": 41057 }, { "epoch": 0.04, "learning_rate": 4.979328371538237e-05, "loss": 1.1652, "step": 41058 }, { "epoch": 0.04, "learning_rate": 4.979327363513231e-05, "loss": 0.785, "step": 41059 }, { "epoch": 0.04, "learning_rate": 4.9793263554637494e-05, "loss": 1.0145, "step": 41060 }, { "epoch": 0.04, "learning_rate": 4.979325347389794e-05, "loss": 0.7387, "step": 41061 }, { "epoch": 0.04, "learning_rate": 4.979324339291363e-05, "loss": 1.1376, "step": 41062 }, { "epoch": 0.04, "learning_rate": 4.979323331168457e-05, "loss": 0.9783, "step": 41063 }, { "epoch": 0.04, "learning_rate": 4.979322323021077e-05, "loss": 0.8317, "step": 41064 }, { "epoch": 0.04, "learning_rate": 4.979321314849221e-05, "loss": 1.31, "step": 41065 }, { "epoch": 0.04, "learning_rate": 4.979320306652891e-05, "loss": 0.9026, "step": 41066 }, { "epoch": 0.04, "learning_rate": 4.979319298432087e-05, "loss": 1.1282, "step": 41067 }, { "epoch": 0.04, "learning_rate": 4.979318290186807e-05, "loss": 0.8821, "step": 41068 }, { "epoch": 0.04, "learning_rate": 4.979317281917052e-05, "loss": 1.0077, "step": 41069 }, { "epoch": 0.04, "learning_rate": 4.979316273622823e-05, "loss": 0.9351, "step": 41070 }, { "epoch": 0.04, "learning_rate": 4.979315265304119e-05, "loss": 1.0323, "step": 41071 }, { "epoch": 0.04, "learning_rate": 4.979314256960941e-05, "loss": 0.853, "step": 41072 }, { "epoch": 0.04, "learning_rate": 4.979313248593287e-05, "loss": 1.3601, "step": 41073 }, { "epoch": 0.04, "learning_rate": 4.9793122402011584e-05, "loss": 0.9759, "step": 41074 }, { "epoch": 0.04, "learning_rate": 4.979311231784556e-05, "loss": 0.8499, "step": 41075 }, { "epoch": 0.04, "learning_rate": 4.979310223343478e-05, "loss": 0.8892, "step": 41076 }, { "epoch": 0.04, "learning_rate": 4.979309214877925e-05, "loss": 1.0848, "step": 41077 }, { "epoch": 0.04, "learning_rate": 4.979308206387898e-05, "loss": 1.1487, "step": 41078 }, { "epoch": 0.04, "learning_rate": 4.979307197873396e-05, "loss": 0.6095, "step": 41079 }, { "epoch": 0.04, "learning_rate": 4.97930618933442e-05, "loss": 0.9847, "step": 41080 }, { "epoch": 0.04, "learning_rate": 4.979305180770969e-05, "loss": 1.1438, "step": 41081 }, { "epoch": 0.04, "learning_rate": 4.9793041721830434e-05, "loss": 1.1241, "step": 41082 }, { "epoch": 0.04, "learning_rate": 4.979303163570643e-05, "loss": 1.1474, "step": 41083 }, { "epoch": 0.04, "learning_rate": 4.979302154933768e-05, "loss": 1.151, "step": 41084 }, { "epoch": 0.04, "learning_rate": 4.9793011462724185e-05, "loss": 1.1381, "step": 41085 }, { "epoch": 0.04, "learning_rate": 4.9793001375865936e-05, "loss": 1.0368, "step": 41086 }, { "epoch": 0.04, "learning_rate": 4.979299128876295e-05, "loss": 0.8393, "step": 41087 }, { "epoch": 0.04, "learning_rate": 4.979298120141521e-05, "loss": 1.0045, "step": 41088 }, { "epoch": 0.04, "learning_rate": 4.979297111382273e-05, "loss": 1.0442, "step": 41089 }, { "epoch": 0.04, "learning_rate": 4.979296102598551e-05, "loss": 1.2169, "step": 41090 }, { "epoch": 0.04, "learning_rate": 4.9792950937903535e-05, "loss": 1.2859, "step": 41091 }, { "epoch": 0.04, "learning_rate": 4.9792940849576816e-05, "loss": 0.9384, "step": 41092 }, { "epoch": 0.04, "learning_rate": 4.979293076100535e-05, "loss": 1.0186, "step": 41093 }, { "epoch": 0.04, "learning_rate": 4.979292067218915e-05, "loss": 0.9974, "step": 41094 }, { "epoch": 0.04, "learning_rate": 4.979291058312819e-05, "loss": 1.2048, "step": 41095 }, { "epoch": 0.04, "learning_rate": 4.9792900493822494e-05, "loss": 0.8147, "step": 41096 }, { "epoch": 0.04, "learning_rate": 4.979289040427205e-05, "loss": 0.4064, "step": 41097 }, { "epoch": 0.04, "learning_rate": 4.979288031447685e-05, "loss": 1.1764, "step": 41098 }, { "epoch": 0.04, "learning_rate": 4.9792870224436924e-05, "loss": 1.1032, "step": 41099 }, { "epoch": 0.04, "learning_rate": 4.9792860134152244e-05, "loss": 1.031, "step": 41100 }, { "epoch": 0.04, "learning_rate": 4.979285004362282e-05, "loss": 1.0201, "step": 41101 }, { "epoch": 0.04, "learning_rate": 4.9792839952848655e-05, "loss": 1.3128, "step": 41102 }, { "epoch": 0.04, "learning_rate": 4.979282986182974e-05, "loss": 1.095, "step": 41103 }, { "epoch": 0.04, "learning_rate": 4.979281977056608e-05, "loss": 1.0507, "step": 41104 }, { "epoch": 0.04, "learning_rate": 4.979280967905768e-05, "loss": 0.8423, "step": 41105 }, { "epoch": 0.04, "learning_rate": 4.979279958730454e-05, "loss": 0.9229, "step": 41106 }, { "epoch": 0.04, "learning_rate": 4.979278949530665e-05, "loss": 0.375, "step": 41107 }, { "epoch": 0.04, "learning_rate": 4.9792779403064014e-05, "loss": 0.059, "step": 41108 }, { "epoch": 0.04, "learning_rate": 4.9792769310576635e-05, "loss": 0.0623, "step": 41109 }, { "epoch": 0.04, "learning_rate": 4.979275921784452e-05, "loss": 0.0636, "step": 41110 }, { "epoch": 0.04, "learning_rate": 4.979274912486765e-05, "loss": 0.0459, "step": 41111 }, { "epoch": 0.04, "learning_rate": 4.979273903164604e-05, "loss": 0.0296, "step": 41112 }, { "epoch": 0.04, "learning_rate": 4.979272893817969e-05, "loss": 0.0352, "step": 41113 }, { "epoch": 0.04, "learning_rate": 4.97927188444686e-05, "loss": 0.0648, "step": 41114 }, { "epoch": 0.04, "learning_rate": 4.9792708750512754e-05, "loss": 0.0686, "step": 41115 }, { "epoch": 0.04, "learning_rate": 4.979269865631217e-05, "loss": 0.0384, "step": 41116 }, { "epoch": 0.04, "learning_rate": 4.979268856186685e-05, "loss": 0.3288, "step": 41117 }, { "epoch": 0.04, "learning_rate": 4.979267846717678e-05, "loss": 0.9563, "step": 41118 }, { "epoch": 0.04, "learning_rate": 4.9792668372241965e-05, "loss": 0.9358, "step": 41119 }, { "epoch": 0.04, "learning_rate": 4.979265827706242e-05, "loss": 0.6838, "step": 41120 }, { "epoch": 0.04, "learning_rate": 4.9792648181638124e-05, "loss": 0.5481, "step": 41121 }, { "epoch": 0.04, "learning_rate": 4.9792638085969086e-05, "loss": 0.9899, "step": 41122 }, { "epoch": 0.04, "learning_rate": 4.979262799005531e-05, "loss": 1.04, "step": 41123 }, { "epoch": 0.04, "learning_rate": 4.979261789389678e-05, "loss": 0.6783, "step": 41124 }, { "epoch": 0.04, "learning_rate": 4.9792607797493515e-05, "loss": 1.1898, "step": 41125 }, { "epoch": 0.04, "learning_rate": 4.97925977008455e-05, "loss": 1.2219, "step": 41126 }, { "epoch": 0.04, "learning_rate": 4.979258760395275e-05, "loss": 0.9756, "step": 41127 }, { "epoch": 0.04, "learning_rate": 4.9792577506815264e-05, "loss": 0.7528, "step": 41128 }, { "epoch": 0.04, "learning_rate": 4.979256740943303e-05, "loss": 0.8252, "step": 41129 }, { "epoch": 0.04, "learning_rate": 4.979255731180605e-05, "loss": 0.76, "step": 41130 }, { "epoch": 0.04, "learning_rate": 4.9792547213934336e-05, "loss": 0.9482, "step": 41131 }, { "epoch": 0.04, "learning_rate": 4.979253711581788e-05, "loss": 0.8926, "step": 41132 }, { "epoch": 0.04, "learning_rate": 4.979252701745668e-05, "loss": 0.8482, "step": 41133 }, { "epoch": 0.04, "learning_rate": 4.979251691885074e-05, "loss": 0.5956, "step": 41134 }, { "epoch": 0.04, "learning_rate": 4.979250682000005e-05, "loss": 0.3637, "step": 41135 }, { "epoch": 0.04, "learning_rate": 4.979249672090463e-05, "loss": 0.696, "step": 41136 }, { "epoch": 0.04, "learning_rate": 4.979248662156447e-05, "loss": 0.9896, "step": 41137 }, { "epoch": 0.04, "learning_rate": 4.979247652197956e-05, "loss": 0.9104, "step": 41138 }, { "epoch": 0.04, "learning_rate": 4.9792466422149906e-05, "loss": 1.3247, "step": 41139 }, { "epoch": 0.04, "learning_rate": 4.979245632207552e-05, "loss": 1.2672, "step": 41140 }, { "epoch": 0.04, "learning_rate": 4.97924462217564e-05, "loss": 1.2489, "step": 41141 }, { "epoch": 0.04, "learning_rate": 4.9792436121192525e-05, "loss": 1.1041, "step": 41142 }, { "epoch": 0.04, "learning_rate": 4.979242602038392e-05, "loss": 1.1169, "step": 41143 }, { "epoch": 0.04, "learning_rate": 4.979241591933057e-05, "loss": 1.0832, "step": 41144 }, { "epoch": 0.04, "learning_rate": 4.979240581803247e-05, "loss": 0.7253, "step": 41145 }, { "epoch": 0.04, "learning_rate": 4.9792395716489647e-05, "loss": 0.7662, "step": 41146 }, { "epoch": 0.04, "learning_rate": 4.9792385614702076e-05, "loss": 0.8042, "step": 41147 }, { "epoch": 0.04, "learning_rate": 4.979237551266977e-05, "loss": 0.9164, "step": 41148 }, { "epoch": 0.04, "learning_rate": 4.979236541039271e-05, "loss": 1.0141, "step": 41149 }, { "epoch": 0.04, "learning_rate": 4.979235530787093e-05, "loss": 0.8994, "step": 41150 }, { "epoch": 0.04, "learning_rate": 4.97923452051044e-05, "loss": 0.8838, "step": 41151 }, { "epoch": 0.04, "learning_rate": 4.979233510209312e-05, "loss": 0.6715, "step": 41152 }, { "epoch": 0.04, "learning_rate": 4.9792324998837115e-05, "loss": 1.0137, "step": 41153 }, { "epoch": 0.04, "learning_rate": 4.979231489533637e-05, "loss": 1.0871, "step": 41154 }, { "epoch": 0.04, "learning_rate": 4.9792304791590885e-05, "loss": 1.121, "step": 41155 }, { "epoch": 0.04, "learning_rate": 4.979229468760065e-05, "loss": 0.9135, "step": 41156 }, { "epoch": 0.04, "learning_rate": 4.979228458336569e-05, "loss": 0.9271, "step": 41157 }, { "epoch": 0.04, "learning_rate": 4.9792274478885984e-05, "loss": 1.734, "step": 41158 }, { "epoch": 0.04, "learning_rate": 4.979226437416153e-05, "loss": 1.748, "step": 41159 }, { "epoch": 0.04, "learning_rate": 4.979225426919236e-05, "loss": 1.4903, "step": 41160 }, { "epoch": 0.04, "learning_rate": 4.979224416397844e-05, "loss": 1.3612, "step": 41161 }, { "epoch": 0.04, "learning_rate": 4.979223405851977e-05, "loss": 1.0017, "step": 41162 }, { "epoch": 0.04, "learning_rate": 4.9792223952816375e-05, "loss": 1.029, "step": 41163 }, { "epoch": 0.04, "learning_rate": 4.979221384686824e-05, "loss": 0.7353, "step": 41164 }, { "epoch": 0.04, "learning_rate": 4.9792203740675366e-05, "loss": 1.2257, "step": 41165 }, { "epoch": 0.04, "learning_rate": 4.979219363423775e-05, "loss": 0.8683, "step": 41166 }, { "epoch": 0.04, "learning_rate": 4.9792183527555404e-05, "loss": 1.1169, "step": 41167 }, { "epoch": 0.04, "learning_rate": 4.979217342062831e-05, "loss": 0.9933, "step": 41168 }, { "epoch": 0.04, "learning_rate": 4.979216331345649e-05, "loss": 1.2589, "step": 41169 }, { "epoch": 0.04, "learning_rate": 4.9792153206039916e-05, "loss": 0.9823, "step": 41170 }, { "epoch": 0.04, "learning_rate": 4.979214309837862e-05, "loss": 1.0913, "step": 41171 }, { "epoch": 0.04, "learning_rate": 4.979213299047257e-05, "loss": 0.8954, "step": 41172 }, { "epoch": 0.04, "learning_rate": 4.9792122882321804e-05, "loss": 1.1442, "step": 41173 }, { "epoch": 0.04, "learning_rate": 4.979211277392628e-05, "loss": 1.0027, "step": 41174 }, { "epoch": 0.04, "learning_rate": 4.9792102665286036e-05, "loss": 0.7512, "step": 41175 }, { "epoch": 0.04, "learning_rate": 4.979209255640104e-05, "loss": 0.9932, "step": 41176 }, { "epoch": 0.04, "learning_rate": 4.9792082447271314e-05, "loss": 0.8733, "step": 41177 }, { "epoch": 0.04, "learning_rate": 4.979207233789685e-05, "loss": 1.2919, "step": 41178 }, { "epoch": 0.04, "learning_rate": 4.9792062228277646e-05, "loss": 0.8648, "step": 41179 }, { "epoch": 0.04, "learning_rate": 4.979205211841371e-05, "loss": 0.8247, "step": 41180 }, { "epoch": 0.04, "learning_rate": 4.979204200830504e-05, "loss": 0.8047, "step": 41181 }, { "epoch": 0.04, "learning_rate": 4.9792031897951627e-05, "loss": 1.2789, "step": 41182 }, { "epoch": 0.04, "learning_rate": 4.9792021787353484e-05, "loss": 1.1336, "step": 41183 }, { "epoch": 0.04, "learning_rate": 4.9792011676510596e-05, "loss": 1.0478, "step": 41184 }, { "epoch": 0.04, "learning_rate": 4.9792001565422976e-05, "loss": 1.1668, "step": 41185 }, { "epoch": 0.04, "learning_rate": 4.9791991454090625e-05, "loss": 1.4286, "step": 41186 }, { "epoch": 0.04, "learning_rate": 4.979198134251353e-05, "loss": 0.9207, "step": 41187 }, { "epoch": 0.04, "learning_rate": 4.97919712306917e-05, "loss": 1.2979, "step": 41188 }, { "epoch": 0.04, "learning_rate": 4.979196111862514e-05, "loss": 1.2578, "step": 41189 }, { "epoch": 0.04, "learning_rate": 4.9791951006313844e-05, "loss": 1.1883, "step": 41190 }, { "epoch": 0.04, "learning_rate": 4.97919408937578e-05, "loss": 1.0239, "step": 41191 }, { "epoch": 0.04, "learning_rate": 4.979193078095703e-05, "loss": 0.9507, "step": 41192 }, { "epoch": 0.04, "learning_rate": 4.979192066791153e-05, "loss": 1.0876, "step": 41193 }, { "epoch": 0.04, "learning_rate": 4.9791910554621284e-05, "loss": 0.8698, "step": 41194 }, { "epoch": 0.04, "learning_rate": 4.97919004410863e-05, "loss": 1.0752, "step": 41195 }, { "epoch": 0.04, "learning_rate": 4.9791890327306595e-05, "loss": 0.7877, "step": 41196 }, { "epoch": 0.04, "learning_rate": 4.979188021328215e-05, "loss": 0.9128, "step": 41197 }, { "epoch": 0.04, "learning_rate": 4.979187009901296e-05, "loss": 1.1592, "step": 41198 }, { "epoch": 0.04, "learning_rate": 4.9791859984499044e-05, "loss": 1.1503, "step": 41199 }, { "epoch": 0.04, "learning_rate": 4.979184986974039e-05, "loss": 1.2841, "step": 41200 }, { "epoch": 0.04, "learning_rate": 4.9791839754737005e-05, "loss": 1.1385, "step": 41201 }, { "epoch": 0.04, "learning_rate": 4.979182963948889e-05, "loss": 0.5703, "step": 41202 }, { "epoch": 0.04, "learning_rate": 4.979181952399603e-05, "loss": 0.713, "step": 41203 }, { "epoch": 0.04, "learning_rate": 4.979180940825844e-05, "loss": 0.6471, "step": 41204 }, { "epoch": 0.04, "learning_rate": 4.9791799292276115e-05, "loss": 0.8482, "step": 41205 }, { "epoch": 0.04, "learning_rate": 4.979178917604905e-05, "loss": 1.0606, "step": 41206 }, { "epoch": 0.04, "learning_rate": 4.979177905957726e-05, "loss": 1.1813, "step": 41207 }, { "epoch": 0.04, "learning_rate": 4.979176894286074e-05, "loss": 1.4641, "step": 41208 }, { "epoch": 0.04, "learning_rate": 4.9791758825899474e-05, "loss": 1.1416, "step": 41209 }, { "epoch": 0.04, "learning_rate": 4.979174870869348e-05, "loss": 1.1356, "step": 41210 }, { "epoch": 0.04, "learning_rate": 4.979173859124276e-05, "loss": 1.0445, "step": 41211 }, { "epoch": 0.04, "learning_rate": 4.979172847354729e-05, "loss": 1.3935, "step": 41212 }, { "epoch": 0.04, "learning_rate": 4.97917183556071e-05, "loss": 0.7645, "step": 41213 }, { "epoch": 0.04, "learning_rate": 4.9791708237422164e-05, "loss": 0.4798, "step": 41214 }, { "epoch": 0.04, "learning_rate": 4.9791698118992504e-05, "loss": 0.3519, "step": 41215 }, { "epoch": 0.04, "learning_rate": 4.979168800031811e-05, "loss": 0.2662, "step": 41216 }, { "epoch": 0.04, "learning_rate": 4.979167788139898e-05, "loss": 0.2123, "step": 41217 }, { "epoch": 0.04, "learning_rate": 4.979166776223512e-05, "loss": 0.2305, "step": 41218 }, { "epoch": 0.04, "learning_rate": 4.9791657642826526e-05, "loss": 0.2222, "step": 41219 }, { "epoch": 0.04, "learning_rate": 4.9791647523173195e-05, "loss": 0.1001, "step": 41220 }, { "epoch": 0.04, "learning_rate": 4.979163740327514e-05, "loss": 0.2893, "step": 41221 }, { "epoch": 0.04, "learning_rate": 4.979162728313235e-05, "loss": 0.3024, "step": 41222 }, { "epoch": 0.04, "learning_rate": 4.979161716274482e-05, "loss": 0.3084, "step": 41223 }, { "epoch": 0.04, "learning_rate": 4.9791607042112566e-05, "loss": 0.2829, "step": 41224 }, { "epoch": 0.04, "learning_rate": 4.979159692123557e-05, "loss": 0.5217, "step": 41225 }, { "epoch": 0.04, "learning_rate": 4.9791586800113855e-05, "loss": 1.0504, "step": 41226 }, { "epoch": 0.04, "learning_rate": 4.9791576678747396e-05, "loss": 0.4616, "step": 41227 }, { "epoch": 0.04, "learning_rate": 4.979156655713621e-05, "loss": 0.8988, "step": 41228 }, { "epoch": 0.04, "learning_rate": 4.97915564352803e-05, "loss": 0.9871, "step": 41229 }, { "epoch": 0.04, "learning_rate": 4.979154631317964e-05, "loss": 1.0571, "step": 41230 }, { "epoch": 0.04, "learning_rate": 4.9791536190834264e-05, "loss": 1.0006, "step": 41231 }, { "epoch": 0.04, "learning_rate": 4.979152606824415e-05, "loss": 1.0851, "step": 41232 }, { "epoch": 0.04, "learning_rate": 4.9791515945409306e-05, "loss": 0.9599, "step": 41233 }, { "epoch": 0.04, "learning_rate": 4.979150582232973e-05, "loss": 0.9435, "step": 41234 }, { "epoch": 0.04, "learning_rate": 4.979149569900542e-05, "loss": 1.1359, "step": 41235 }, { "epoch": 0.04, "learning_rate": 4.979148557543638e-05, "loss": 1.2216, "step": 41236 }, { "epoch": 0.04, "learning_rate": 4.979147545162262e-05, "loss": 0.9071, "step": 41237 }, { "epoch": 0.04, "learning_rate": 4.979146532756411e-05, "loss": 0.9262, "step": 41238 }, { "epoch": 0.04, "learning_rate": 4.9791455203260886e-05, "loss": 0.9061, "step": 41239 }, { "epoch": 0.04, "learning_rate": 4.979144507871292e-05, "loss": 0.5216, "step": 41240 }, { "epoch": 0.04, "learning_rate": 4.979143495392022e-05, "loss": 0.9255, "step": 41241 }, { "epoch": 0.04, "learning_rate": 4.9791424828882805e-05, "loss": 0.9883, "step": 41242 }, { "epoch": 0.04, "learning_rate": 4.979141470360065e-05, "loss": 1.3282, "step": 41243 }, { "epoch": 0.04, "learning_rate": 4.9791404578073766e-05, "loss": 0.945, "step": 41244 }, { "epoch": 0.04, "learning_rate": 4.979139445230215e-05, "loss": 0.9848, "step": 41245 }, { "epoch": 0.04, "learning_rate": 4.979138432628581e-05, "loss": 0.9453, "step": 41246 }, { "epoch": 0.04, "learning_rate": 4.979137420002473e-05, "loss": 0.2338, "step": 41247 }, { "epoch": 0.04, "learning_rate": 4.979136407351893e-05, "loss": 0.8352, "step": 41248 }, { "epoch": 0.04, "learning_rate": 4.979135394676839e-05, "loss": 1.0176, "step": 41249 }, { "epoch": 0.04, "learning_rate": 4.9791343819773126e-05, "loss": 1.0191, "step": 41250 }, { "epoch": 0.04, "learning_rate": 4.979133369253314e-05, "loss": 0.9374, "step": 41251 }, { "epoch": 0.04, "learning_rate": 4.979132356504841e-05, "loss": 0.8733, "step": 41252 }, { "epoch": 0.04, "learning_rate": 4.979131343731895e-05, "loss": 0.8893, "step": 41253 }, { "epoch": 0.04, "learning_rate": 4.9791303309344774e-05, "loss": 1.053, "step": 41254 }, { "epoch": 0.04, "learning_rate": 4.9791293181125866e-05, "loss": 1.3469, "step": 41255 }, { "epoch": 0.04, "learning_rate": 4.979128305266222e-05, "loss": 1.029, "step": 41256 }, { "epoch": 0.04, "learning_rate": 4.979127292395385e-05, "loss": 1.0788, "step": 41257 }, { "epoch": 0.04, "learning_rate": 4.9791262795000745e-05, "loss": 1.3079, "step": 41258 }, { "epoch": 0.04, "learning_rate": 4.9791252665802924e-05, "loss": 1.0385, "step": 41259 }, { "epoch": 0.04, "learning_rate": 4.9791242536360365e-05, "loss": 1.0194, "step": 41260 }, { "epoch": 0.04, "learning_rate": 4.979123240667308e-05, "loss": 0.9519, "step": 41261 }, { "epoch": 0.04, "learning_rate": 4.9791222276741065e-05, "loss": 1.074, "step": 41262 }, { "epoch": 0.04, "learning_rate": 4.9791212146564325e-05, "loss": 1.1525, "step": 41263 }, { "epoch": 0.04, "learning_rate": 4.979120201614285e-05, "loss": 1.2236, "step": 41264 }, { "epoch": 0.04, "learning_rate": 4.979119188547665e-05, "loss": 1.1108, "step": 41265 }, { "epoch": 0.04, "learning_rate": 4.979118175456572e-05, "loss": 0.883, "step": 41266 }, { "epoch": 0.04, "learning_rate": 4.979117162341007e-05, "loss": 0.45, "step": 41267 }, { "epoch": 0.04, "learning_rate": 4.9791161492009684e-05, "loss": 0.2732, "step": 41268 }, { "epoch": 0.04, "learning_rate": 4.9791151360364575e-05, "loss": 0.3392, "step": 41269 }, { "epoch": 0.04, "learning_rate": 4.9791141228474734e-05, "loss": 0.2568, "step": 41270 }, { "epoch": 0.04, "learning_rate": 4.979113109634017e-05, "loss": 0.7965, "step": 41271 }, { "epoch": 0.04, "learning_rate": 4.979112096396087e-05, "loss": 1.2703, "step": 41272 }, { "epoch": 0.04, "learning_rate": 4.979111083133685e-05, "loss": 1.1062, "step": 41273 }, { "epoch": 0.04, "learning_rate": 4.9791100698468096e-05, "loss": 0.9047, "step": 41274 }, { "epoch": 0.04, "learning_rate": 4.979109056535462e-05, "loss": 1.0241, "step": 41275 }, { "epoch": 0.04, "learning_rate": 4.9791080431996416e-05, "loss": 0.3011, "step": 41276 }, { "epoch": 0.04, "learning_rate": 4.979107029839348e-05, "loss": 0.1454, "step": 41277 }, { "epoch": 0.04, "learning_rate": 4.979106016454583e-05, "loss": 1.0615, "step": 41278 }, { "epoch": 0.04, "learning_rate": 4.9791050030453445e-05, "loss": 1.4512, "step": 41279 }, { "epoch": 0.04, "learning_rate": 4.979103989611633e-05, "loss": 1.1346, "step": 41280 }, { "epoch": 0.04, "learning_rate": 4.979102976153449e-05, "loss": 1.1818, "step": 41281 }, { "epoch": 0.04, "learning_rate": 4.9791019626707924e-05, "loss": 1.2401, "step": 41282 }, { "epoch": 0.04, "learning_rate": 4.979100949163663e-05, "loss": 0.9102, "step": 41283 }, { "epoch": 0.04, "learning_rate": 4.9790999356320613e-05, "loss": 0.8178, "step": 41284 }, { "epoch": 0.04, "learning_rate": 4.979098922075987e-05, "loss": 1.1504, "step": 41285 }, { "epoch": 0.04, "learning_rate": 4.9790979084954403e-05, "loss": 1.0782, "step": 41286 }, { "epoch": 0.04, "learning_rate": 4.97909689489042e-05, "loss": 0.8875, "step": 41287 }, { "epoch": 0.04, "learning_rate": 4.979095881260928e-05, "loss": 0.7536, "step": 41288 }, { "epoch": 0.04, "learning_rate": 4.979094867606963e-05, "loss": 0.9272, "step": 41289 }, { "epoch": 0.04, "learning_rate": 4.979093853928526e-05, "loss": 1.0363, "step": 41290 }, { "epoch": 0.04, "learning_rate": 4.979092840225615e-05, "loss": 1.049, "step": 41291 }, { "epoch": 0.04, "learning_rate": 4.979091826498233e-05, "loss": 0.8471, "step": 41292 }, { "epoch": 0.04, "learning_rate": 4.9790908127463774e-05, "loss": 0.5622, "step": 41293 }, { "epoch": 0.04, "learning_rate": 4.9790897989700505e-05, "loss": 0.3371, "step": 41294 }, { "epoch": 0.04, "learning_rate": 4.9790887851692505e-05, "loss": 0.8168, "step": 41295 }, { "epoch": 0.04, "learning_rate": 4.979087771343977e-05, "loss": 0.995, "step": 41296 }, { "epoch": 0.04, "learning_rate": 4.979086757494232e-05, "loss": 1.3143, "step": 41297 }, { "epoch": 0.04, "learning_rate": 4.979085743620014e-05, "loss": 1.0343, "step": 41298 }, { "epoch": 0.04, "learning_rate": 4.9790847297213236e-05, "loss": 0.8775, "step": 41299 }, { "epoch": 0.04, "learning_rate": 4.979083715798162e-05, "loss": 0.7598, "step": 41300 }, { "epoch": 0.04, "learning_rate": 4.979082701850526e-05, "loss": 0.9568, "step": 41301 }, { "epoch": 0.04, "learning_rate": 4.979081687878419e-05, "loss": 1.1404, "step": 41302 }, { "epoch": 0.04, "learning_rate": 4.9790806738818384e-05, "loss": 1.0428, "step": 41303 }, { "epoch": 0.04, "learning_rate": 4.979079659860786e-05, "loss": 1.072, "step": 41304 }, { "epoch": 0.04, "learning_rate": 4.979078645815261e-05, "loss": 1.2259, "step": 41305 }, { "epoch": 0.04, "learning_rate": 4.979077631745264e-05, "loss": 0.9637, "step": 41306 }, { "epoch": 0.04, "learning_rate": 4.979076617650794e-05, "loss": 0.8131, "step": 41307 }, { "epoch": 0.04, "learning_rate": 4.979075603531852e-05, "loss": 1.0763, "step": 41308 }, { "epoch": 0.04, "learning_rate": 4.979074589388437e-05, "loss": 0.9283, "step": 41309 }, { "epoch": 0.04, "learning_rate": 4.97907357522055e-05, "loss": 0.4011, "step": 41310 }, { "epoch": 0.04, "learning_rate": 4.9790725610281905e-05, "loss": 0.804, "step": 41311 }, { "epoch": 0.04, "learning_rate": 4.979071546811359e-05, "loss": 0.7534, "step": 41312 }, { "epoch": 0.04, "learning_rate": 4.979070532570055e-05, "loss": 1.1111, "step": 41313 }, { "epoch": 0.04, "learning_rate": 4.979069518304279e-05, "loss": 0.8077, "step": 41314 }, { "epoch": 0.04, "learning_rate": 4.97906850401403e-05, "loss": 1.2849, "step": 41315 }, { "epoch": 0.04, "learning_rate": 4.979067489699309e-05, "loss": 1.219, "step": 41316 }, { "epoch": 0.04, "learning_rate": 4.979066475360116e-05, "loss": 1.0586, "step": 41317 }, { "epoch": 0.04, "learning_rate": 4.9790654609964504e-05, "loss": 0.8061, "step": 41318 }, { "epoch": 0.04, "learning_rate": 4.979064446608312e-05, "loss": 1.169, "step": 41319 }, { "epoch": 0.04, "learning_rate": 4.979063432195702e-05, "loss": 0.8767, "step": 41320 }, { "epoch": 0.04, "learning_rate": 4.97906241775862e-05, "loss": 1.0211, "step": 41321 }, { "epoch": 0.04, "learning_rate": 4.979061403297065e-05, "loss": 1.0901, "step": 41322 }, { "epoch": 0.04, "learning_rate": 4.979060388811038e-05, "loss": 1.0455, "step": 41323 }, { "epoch": 0.04, "learning_rate": 4.979059374300539e-05, "loss": 1.1865, "step": 41324 }, { "epoch": 0.04, "learning_rate": 4.979058359765567e-05, "loss": 0.9658, "step": 41325 }, { "epoch": 0.04, "learning_rate": 4.979057345206124e-05, "loss": 1.1257, "step": 41326 }, { "epoch": 0.04, "learning_rate": 4.979056330622208e-05, "loss": 1.0041, "step": 41327 }, { "epoch": 0.04, "learning_rate": 4.97905531601382e-05, "loss": 0.8082, "step": 41328 }, { "epoch": 0.04, "learning_rate": 4.979054301380959e-05, "loss": 0.9844, "step": 41329 }, { "epoch": 0.04, "learning_rate": 4.9790532867236267e-05, "loss": 0.8831, "step": 41330 }, { "epoch": 0.04, "learning_rate": 4.979052272041822e-05, "loss": 1.1112, "step": 41331 }, { "epoch": 0.04, "learning_rate": 4.979051257335545e-05, "loss": 1.3952, "step": 41332 }, { "epoch": 0.04, "learning_rate": 4.9790502426047966e-05, "loss": 1.6382, "step": 41333 }, { "epoch": 0.04, "learning_rate": 4.9790492278495755e-05, "loss": 1.2365, "step": 41334 }, { "epoch": 0.04, "learning_rate": 4.9790482130698825e-05, "loss": 0.986, "step": 41335 }, { "epoch": 0.04, "learning_rate": 4.979047198265717e-05, "loss": 1.052, "step": 41336 }, { "epoch": 0.04, "learning_rate": 4.979046183437079e-05, "loss": 0.7259, "step": 41337 }, { "epoch": 0.04, "learning_rate": 4.97904516858397e-05, "loss": 1.1034, "step": 41338 }, { "epoch": 0.04, "learning_rate": 4.979044153706388e-05, "loss": 1.3415, "step": 41339 }, { "epoch": 0.04, "learning_rate": 4.9790431388043345e-05, "loss": 1.0248, "step": 41340 }, { "epoch": 0.04, "learning_rate": 4.979042123877809e-05, "loss": 0.6168, "step": 41341 }, { "epoch": 0.04, "learning_rate": 4.9790411089268105e-05, "loss": 1.3731, "step": 41342 }, { "epoch": 0.04, "learning_rate": 4.979040093951342e-05, "loss": 0.6298, "step": 41343 }, { "epoch": 0.04, "learning_rate": 4.9790390789514e-05, "loss": 1.0784, "step": 41344 }, { "epoch": 0.04, "learning_rate": 4.979038063926985e-05, "loss": 0.9681, "step": 41345 }, { "epoch": 0.04, "learning_rate": 4.9790370488780994e-05, "loss": 0.8467, "step": 41346 }, { "epoch": 0.04, "learning_rate": 4.9790360338047414e-05, "loss": 0.8826, "step": 41347 }, { "epoch": 0.04, "learning_rate": 4.979035018706911e-05, "loss": 2.2049, "step": 41348 }, { "epoch": 0.04, "learning_rate": 4.9790340035846095e-05, "loss": 1.3379, "step": 41349 }, { "epoch": 0.04, "learning_rate": 4.9790329884378355e-05, "loss": 0.8062, "step": 41350 }, { "epoch": 0.04, "learning_rate": 4.97903197326659e-05, "loss": 0.8789, "step": 41351 }, { "epoch": 0.04, "learning_rate": 4.9790309580708714e-05, "loss": 0.9745, "step": 41352 }, { "epoch": 0.04, "learning_rate": 4.979029942850682e-05, "loss": 1.2103, "step": 41353 }, { "epoch": 0.04, "learning_rate": 4.97902892760602e-05, "loss": 1.1818, "step": 41354 }, { "epoch": 0.04, "learning_rate": 4.9790279123368864e-05, "loss": 0.9056, "step": 41355 }, { "epoch": 0.04, "learning_rate": 4.979026897043281e-05, "loss": 1.1903, "step": 41356 }, { "epoch": 0.04, "learning_rate": 4.9790258817252036e-05, "loss": 0.984, "step": 41357 }, { "epoch": 0.04, "learning_rate": 4.9790248663826546e-05, "loss": 1.1765, "step": 41358 }, { "epoch": 0.04, "learning_rate": 4.979023851015633e-05, "loss": 1.305, "step": 41359 }, { "epoch": 0.04, "learning_rate": 4.9790228356241396e-05, "loss": 0.7871, "step": 41360 }, { "epoch": 0.04, "learning_rate": 4.9790218202081745e-05, "loss": 1.0186, "step": 41361 }, { "epoch": 0.04, "learning_rate": 4.9790208047677375e-05, "loss": 0.9245, "step": 41362 }, { "epoch": 0.04, "learning_rate": 4.979019789302829e-05, "loss": 0.9866, "step": 41363 }, { "epoch": 0.04, "learning_rate": 4.979018773813449e-05, "loss": 1.0434, "step": 41364 }, { "epoch": 0.04, "learning_rate": 4.979017758299596e-05, "loss": 0.8409, "step": 41365 }, { "epoch": 0.04, "learning_rate": 4.9790167427612724e-05, "loss": 1.0594, "step": 41366 }, { "epoch": 0.04, "learning_rate": 4.979015727198476e-05, "loss": 0.8824, "step": 41367 }, { "epoch": 0.04, "learning_rate": 4.979014711611208e-05, "loss": 0.793, "step": 41368 }, { "epoch": 0.04, "learning_rate": 4.979013695999468e-05, "loss": 0.9518, "step": 41369 }, { "epoch": 0.04, "learning_rate": 4.979012680363257e-05, "loss": 1.0431, "step": 41370 }, { "epoch": 0.04, "learning_rate": 4.9790116647025734e-05, "loss": 1.1422, "step": 41371 }, { "epoch": 0.04, "learning_rate": 4.9790106490174185e-05, "loss": 1.2858, "step": 41372 }, { "epoch": 0.04, "learning_rate": 4.979009633307792e-05, "loss": 1.1424, "step": 41373 }, { "epoch": 0.04, "learning_rate": 4.979008617573694e-05, "loss": 0.862, "step": 41374 }, { "epoch": 0.04, "learning_rate": 4.9790076018151236e-05, "loss": 0.3523, "step": 41375 }, { "epoch": 0.04, "learning_rate": 4.9790065860320814e-05, "loss": 0.4887, "step": 41376 }, { "epoch": 0.04, "learning_rate": 4.979005570224568e-05, "loss": 1.1707, "step": 41377 }, { "epoch": 0.04, "learning_rate": 4.979004554392583e-05, "loss": 0.9905, "step": 41378 }, { "epoch": 0.04, "learning_rate": 4.9790035385361256e-05, "loss": 1.3182, "step": 41379 }, { "epoch": 0.04, "learning_rate": 4.979002522655197e-05, "loss": 0.7959, "step": 41380 }, { "epoch": 0.04, "learning_rate": 4.979001506749797e-05, "loss": 1.1707, "step": 41381 }, { "epoch": 0.04, "learning_rate": 4.979000490819925e-05, "loss": 0.8291, "step": 41382 }, { "epoch": 0.04, "learning_rate": 4.978999474865581e-05, "loss": 0.7934, "step": 41383 }, { "epoch": 0.04, "learning_rate": 4.978998458886766e-05, "loss": 0.5011, "step": 41384 }, { "epoch": 0.04, "learning_rate": 4.978997442883479e-05, "loss": 1.1467, "step": 41385 }, { "epoch": 0.04, "learning_rate": 4.978996426855721e-05, "loss": 0.6888, "step": 41386 }, { "epoch": 0.04, "learning_rate": 4.978995410803491e-05, "loss": 1.1079, "step": 41387 }, { "epoch": 0.04, "learning_rate": 4.978994394726789e-05, "loss": 1.2962, "step": 41388 }, { "epoch": 0.04, "learning_rate": 4.978993378625616e-05, "loss": 0.8496, "step": 41389 }, { "epoch": 0.04, "learning_rate": 4.9789923624999705e-05, "loss": 1.109, "step": 41390 }, { "epoch": 0.04, "learning_rate": 4.978991346349854e-05, "loss": 1.0008, "step": 41391 }, { "epoch": 0.04, "learning_rate": 4.978990330175266e-05, "loss": 0.9962, "step": 41392 }, { "epoch": 0.04, "learning_rate": 4.9789893139762066e-05, "loss": 0.6585, "step": 41393 }, { "epoch": 0.04, "learning_rate": 4.978988297752676e-05, "loss": 0.8782, "step": 41394 }, { "epoch": 0.04, "learning_rate": 4.9789872815046726e-05, "loss": 1.0086, "step": 41395 }, { "epoch": 0.04, "learning_rate": 4.978986265232199e-05, "loss": 1.1103, "step": 41396 }, { "epoch": 0.04, "learning_rate": 4.978985248935253e-05, "loss": 1.0354, "step": 41397 }, { "epoch": 0.04, "learning_rate": 4.978984232613836e-05, "loss": 0.8904, "step": 41398 }, { "epoch": 0.04, "learning_rate": 4.978983216267947e-05, "loss": 0.9766, "step": 41399 }, { "epoch": 0.04, "learning_rate": 4.978982199897587e-05, "loss": 1.1553, "step": 41400 }, { "epoch": 0.04, "learning_rate": 4.9789811835027555e-05, "loss": 1.6969, "step": 41401 }, { "epoch": 0.04, "learning_rate": 4.978980167083452e-05, "loss": 1.7232, "step": 41402 }, { "epoch": 0.04, "learning_rate": 4.978979150639678e-05, "loss": 1.234, "step": 41403 }, { "epoch": 0.04, "learning_rate": 4.9789781341714325e-05, "loss": 1.1794, "step": 41404 }, { "epoch": 0.04, "learning_rate": 4.978977117678715e-05, "loss": 1.0198, "step": 41405 }, { "epoch": 0.04, "learning_rate": 4.9789761011615263e-05, "loss": 1.1099, "step": 41406 }, { "epoch": 0.04, "learning_rate": 4.978975084619866e-05, "loss": 1.0915, "step": 41407 }, { "epoch": 0.04, "learning_rate": 4.9789740680537344e-05, "loss": 1.0289, "step": 41408 }, { "epoch": 0.04, "learning_rate": 4.978973051463132e-05, "loss": 1.1217, "step": 41409 }, { "epoch": 0.04, "learning_rate": 4.978972034848057e-05, "loss": 1.1149, "step": 41410 }, { "epoch": 0.04, "learning_rate": 4.978971018208512e-05, "loss": 1.0499, "step": 41411 }, { "epoch": 0.04, "learning_rate": 4.978970001544494e-05, "loss": 0.966, "step": 41412 }, { "epoch": 0.04, "learning_rate": 4.9789689848560065e-05, "loss": 1.056, "step": 41413 }, { "epoch": 0.04, "learning_rate": 4.978967968143047e-05, "loss": 1.1158, "step": 41414 }, { "epoch": 0.04, "learning_rate": 4.9789669514056155e-05, "loss": 0.8423, "step": 41415 }, { "epoch": 0.04, "learning_rate": 4.978965934643713e-05, "loss": 0.9034, "step": 41416 }, { "epoch": 0.04, "learning_rate": 4.978964917857339e-05, "loss": 1.4173, "step": 41417 }, { "epoch": 0.04, "learning_rate": 4.978963901046494e-05, "loss": 1.108, "step": 41418 }, { "epoch": 0.04, "learning_rate": 4.978962884211178e-05, "loss": 0.8271, "step": 41419 }, { "epoch": 0.04, "learning_rate": 4.97896186735139e-05, "loss": 0.9005, "step": 41420 }, { "epoch": 0.04, "learning_rate": 4.9789608504671314e-05, "loss": 0.9312, "step": 41421 }, { "epoch": 0.04, "learning_rate": 4.9789598335584015e-05, "loss": 0.8106, "step": 41422 }, { "epoch": 0.04, "learning_rate": 4.9789588166252e-05, "loss": 1.0492, "step": 41423 }, { "epoch": 0.04, "learning_rate": 4.9789577996675275e-05, "loss": 1.0757, "step": 41424 }, { "epoch": 0.04, "learning_rate": 4.978956782685384e-05, "loss": 0.8137, "step": 41425 }, { "epoch": 0.04, "learning_rate": 4.978955765678769e-05, "loss": 1.0106, "step": 41426 }, { "epoch": 0.04, "learning_rate": 4.978954748647683e-05, "loss": 1.4606, "step": 41427 }, { "epoch": 0.04, "learning_rate": 4.9789537315921256e-05, "loss": 1.0561, "step": 41428 }, { "epoch": 0.04, "learning_rate": 4.9789527145120964e-05, "loss": 0.9717, "step": 41429 }, { "epoch": 0.04, "learning_rate": 4.978951697407597e-05, "loss": 0.8606, "step": 41430 }, { "epoch": 0.04, "learning_rate": 4.9789506802786255e-05, "loss": 1.2156, "step": 41431 }, { "epoch": 0.04, "learning_rate": 4.978949663125184e-05, "loss": 1.1101, "step": 41432 }, { "epoch": 0.04, "learning_rate": 4.978948645947271e-05, "loss": 1.222, "step": 41433 }, { "epoch": 0.04, "learning_rate": 4.9789476287448866e-05, "loss": 0.7971, "step": 41434 }, { "epoch": 0.04, "learning_rate": 4.978946611518031e-05, "loss": 0.9198, "step": 41435 }, { "epoch": 0.04, "learning_rate": 4.978945594266704e-05, "loss": 0.8073, "step": 41436 }, { "epoch": 0.04, "learning_rate": 4.9789445769909064e-05, "loss": 1.1486, "step": 41437 }, { "epoch": 0.04, "learning_rate": 4.978943559690638e-05, "loss": 1.0857, "step": 41438 }, { "epoch": 0.04, "learning_rate": 4.9789425423658975e-05, "loss": 1.0554, "step": 41439 }, { "epoch": 0.04, "learning_rate": 4.9789415250166874e-05, "loss": 1.0714, "step": 41440 }, { "epoch": 0.04, "learning_rate": 4.9789405076430055e-05, "loss": 0.9248, "step": 41441 }, { "epoch": 0.04, "learning_rate": 4.978939490244852e-05, "loss": 0.89, "step": 41442 }, { "epoch": 0.04, "learning_rate": 4.9789384728222276e-05, "loss": 1.0103, "step": 41443 }, { "epoch": 0.04, "learning_rate": 4.978937455375132e-05, "loss": 1.2113, "step": 41444 }, { "epoch": 0.04, "learning_rate": 4.9789364379035666e-05, "loss": 1.0159, "step": 41445 }, { "epoch": 0.04, "learning_rate": 4.978935420407529e-05, "loss": 0.9241, "step": 41446 }, { "epoch": 0.04, "learning_rate": 4.978934402887021e-05, "loss": 1.101, "step": 41447 }, { "epoch": 0.04, "learning_rate": 4.9789333853420414e-05, "loss": 1.0591, "step": 41448 }, { "epoch": 0.04, "learning_rate": 4.978932367772591e-05, "loss": 1.024, "step": 41449 }, { "epoch": 0.04, "learning_rate": 4.97893135017867e-05, "loss": 0.7874, "step": 41450 }, { "epoch": 0.04, "learning_rate": 4.9789303325602774e-05, "loss": 0.8313, "step": 41451 }, { "epoch": 0.04, "learning_rate": 4.9789293149174146e-05, "loss": 0.9089, "step": 41452 }, { "epoch": 0.04, "learning_rate": 4.9789282972500806e-05, "loss": 0.9578, "step": 41453 }, { "epoch": 0.04, "learning_rate": 4.9789272795582754e-05, "loss": 1.3403, "step": 41454 }, { "epoch": 0.04, "learning_rate": 4.978926261842e-05, "loss": 1.203, "step": 41455 }, { "epoch": 0.04, "learning_rate": 4.978925244101253e-05, "loss": 0.9004, "step": 41456 }, { "epoch": 0.04, "learning_rate": 4.9789242263360354e-05, "loss": 0.9821, "step": 41457 }, { "epoch": 0.04, "learning_rate": 4.9789232085463465e-05, "loss": 1.0595, "step": 41458 }, { "epoch": 0.04, "learning_rate": 4.978922190732187e-05, "loss": 1.1171, "step": 41459 }, { "epoch": 0.04, "learning_rate": 4.978921172893557e-05, "loss": 1.0264, "step": 41460 }, { "epoch": 0.04, "learning_rate": 4.978920155030456e-05, "loss": 1.1766, "step": 41461 }, { "epoch": 0.04, "learning_rate": 4.9789191371428836e-05, "loss": 0.8881, "step": 41462 }, { "epoch": 0.04, "learning_rate": 4.9789181192308404e-05, "loss": 1.2069, "step": 41463 }, { "epoch": 0.04, "learning_rate": 4.978917101294326e-05, "loss": 0.8621, "step": 41464 }, { "epoch": 0.04, "learning_rate": 4.978916083333342e-05, "loss": 0.9328, "step": 41465 }, { "epoch": 0.04, "learning_rate": 4.978915065347887e-05, "loss": 1.1159, "step": 41466 }, { "epoch": 0.04, "learning_rate": 4.9789140473379606e-05, "loss": 0.8249, "step": 41467 }, { "epoch": 0.04, "learning_rate": 4.978913029303564e-05, "loss": 1.0736, "step": 41468 }, { "epoch": 0.04, "learning_rate": 4.978912011244695e-05, "loss": 1.0249, "step": 41469 }, { "epoch": 0.04, "learning_rate": 4.978910993161356e-05, "loss": 1.014, "step": 41470 }, { "epoch": 0.04, "learning_rate": 4.9789099750535476e-05, "loss": 1.1359, "step": 41471 }, { "epoch": 0.04, "learning_rate": 4.978908956921268e-05, "loss": 1.1373, "step": 41472 }, { "epoch": 0.04, "learning_rate": 4.978907938764517e-05, "loss": 1.0936, "step": 41473 }, { "epoch": 0.04, "learning_rate": 4.978906920583295e-05, "loss": 1.1841, "step": 41474 }, { "epoch": 0.04, "learning_rate": 4.978905902377603e-05, "loss": 1.3778, "step": 41475 }, { "epoch": 0.04, "learning_rate": 4.97890488414744e-05, "loss": 0.9585, "step": 41476 }, { "epoch": 0.04, "learning_rate": 4.9789038658928056e-05, "loss": 1.0257, "step": 41477 }, { "epoch": 0.04, "learning_rate": 4.978902847613701e-05, "loss": 1.0103, "step": 41478 }, { "epoch": 0.04, "learning_rate": 4.978901829310127e-05, "loss": 1.0996, "step": 41479 }, { "epoch": 0.04, "learning_rate": 4.97890081098208e-05, "loss": 0.7431, "step": 41480 }, { "epoch": 0.04, "learning_rate": 4.978899792629565e-05, "loss": 1.0957, "step": 41481 }, { "epoch": 0.04, "learning_rate": 4.978898774252577e-05, "loss": 0.9144, "step": 41482 }, { "epoch": 0.04, "learning_rate": 4.978897755851119e-05, "loss": 1.0449, "step": 41483 }, { "epoch": 0.04, "learning_rate": 4.97889673742519e-05, "loss": 0.5993, "step": 41484 }, { "epoch": 0.04, "learning_rate": 4.978895718974792e-05, "loss": 0.3246, "step": 41485 }, { "epoch": 0.04, "learning_rate": 4.978894700499922e-05, "loss": 0.9746, "step": 41486 }, { "epoch": 0.04, "learning_rate": 4.978893682000582e-05, "loss": 0.9469, "step": 41487 }, { "epoch": 0.04, "learning_rate": 4.978892663476771e-05, "loss": 1.1226, "step": 41488 }, { "epoch": 0.04, "learning_rate": 4.9788916449284894e-05, "loss": 1.1667, "step": 41489 }, { "epoch": 0.04, "learning_rate": 4.9788906263557375e-05, "loss": 0.599, "step": 41490 }, { "epoch": 0.04, "learning_rate": 4.978889607758515e-05, "loss": 0.6699, "step": 41491 }, { "epoch": 0.04, "learning_rate": 4.978888589136822e-05, "loss": 1.0742, "step": 41492 }, { "epoch": 0.04, "learning_rate": 4.9788875704906584e-05, "loss": 0.961, "step": 41493 }, { "epoch": 0.04, "learning_rate": 4.978886551820025e-05, "loss": 0.8797, "step": 41494 }, { "epoch": 0.04, "learning_rate": 4.97888553312492e-05, "loss": 0.8878, "step": 41495 }, { "epoch": 0.04, "learning_rate": 4.978884514405344e-05, "loss": 0.698, "step": 41496 }, { "epoch": 0.04, "learning_rate": 4.978883495661298e-05, "loss": 1.0637, "step": 41497 }, { "epoch": 0.04, "learning_rate": 4.978882476892782e-05, "loss": 0.9255, "step": 41498 }, { "epoch": 0.04, "learning_rate": 4.9788814580997956e-05, "loss": 0.8825, "step": 41499 }, { "epoch": 0.04, "learning_rate": 4.9788804392823385e-05, "loss": 1.2729, "step": 41500 }, { "epoch": 0.04, "eval_loss": 1.0742453336715698, "eval_runtime": 126.3752, "eval_samples_per_second": 10.959, "eval_steps_per_second": 5.484, "step": 41500 }, { "epoch": 0.04, "learning_rate": 4.978879420440411e-05, "loss": 1.0302, "step": 41501 }, { "epoch": 0.04, "learning_rate": 4.978878401574013e-05, "loss": 1.0832, "step": 41502 }, { "epoch": 0.04, "learning_rate": 4.978877382683144e-05, "loss": 1.1322, "step": 41503 }, { "epoch": 0.04, "learning_rate": 4.9788763637678053e-05, "loss": 0.8893, "step": 41504 }, { "epoch": 0.04, "learning_rate": 4.9788753448279954e-05, "loss": 1.3078, "step": 41505 }, { "epoch": 0.04, "learning_rate": 4.978874325863716e-05, "loss": 1.0762, "step": 41506 }, { "epoch": 0.04, "learning_rate": 4.978873306874966e-05, "loss": 1.2672, "step": 41507 }, { "epoch": 0.04, "learning_rate": 4.978872287861746e-05, "loss": 1.6134, "step": 41508 }, { "epoch": 0.04, "learning_rate": 4.978871268824055e-05, "loss": 0.7576, "step": 41509 }, { "epoch": 0.04, "learning_rate": 4.978870249761893e-05, "loss": 0.9819, "step": 41510 }, { "epoch": 0.04, "learning_rate": 4.9788692306752614e-05, "loss": 1.1491, "step": 41511 }, { "epoch": 0.04, "learning_rate": 4.97886821156416e-05, "loss": 1.341, "step": 41512 }, { "epoch": 0.04, "learning_rate": 4.978867192428587e-05, "loss": 1.2528, "step": 41513 }, { "epoch": 0.04, "learning_rate": 4.9788661732685444e-05, "loss": 1.2909, "step": 41514 }, { "epoch": 0.04, "learning_rate": 4.9788651540840314e-05, "loss": 1.2261, "step": 41515 }, { "epoch": 0.04, "learning_rate": 4.978864134875047e-05, "loss": 1.0531, "step": 41516 }, { "epoch": 0.04, "learning_rate": 4.978863115641594e-05, "loss": 0.9123, "step": 41517 }, { "epoch": 0.04, "learning_rate": 4.9788620963836705e-05, "loss": 0.9992, "step": 41518 }, { "epoch": 0.04, "learning_rate": 4.978861077101276e-05, "loss": 0.9787, "step": 41519 }, { "epoch": 0.04, "learning_rate": 4.978860057794411e-05, "loss": 0.9754, "step": 41520 }, { "epoch": 0.04, "learning_rate": 4.978859038463076e-05, "loss": 0.9675, "step": 41521 }, { "epoch": 0.04, "learning_rate": 4.9788580191072715e-05, "loss": 1.2194, "step": 41522 }, { "epoch": 0.04, "learning_rate": 4.978856999726996e-05, "loss": 1.2863, "step": 41523 }, { "epoch": 0.04, "learning_rate": 4.9788559803222513e-05, "loss": 0.8826, "step": 41524 }, { "epoch": 0.04, "learning_rate": 4.978854960893035e-05, "loss": 0.8453, "step": 41525 }, { "epoch": 0.04, "learning_rate": 4.9788539414393495e-05, "loss": 0.9033, "step": 41526 }, { "epoch": 0.04, "learning_rate": 4.978852921961193e-05, "loss": 0.903, "step": 41527 }, { "epoch": 0.04, "learning_rate": 4.9788519024585665e-05, "loss": 0.9561, "step": 41528 }, { "epoch": 0.04, "learning_rate": 4.97885088293147e-05, "loss": 1.1876, "step": 41529 }, { "epoch": 0.04, "learning_rate": 4.978849863379904e-05, "loss": 0.9494, "step": 41530 }, { "epoch": 0.04, "learning_rate": 4.9788488438038664e-05, "loss": 1.3303, "step": 41531 }, { "epoch": 0.04, "learning_rate": 4.97884782420336e-05, "loss": 1.518, "step": 41532 }, { "epoch": 0.04, "learning_rate": 4.9788468045783823e-05, "loss": 0.9134, "step": 41533 }, { "epoch": 0.04, "learning_rate": 4.978845784928935e-05, "loss": 0.9738, "step": 41534 }, { "epoch": 0.04, "learning_rate": 4.978844765255018e-05, "loss": 0.9411, "step": 41535 }, { "epoch": 0.04, "learning_rate": 4.9788437455566303e-05, "loss": 0.9152, "step": 41536 }, { "epoch": 0.04, "learning_rate": 4.978842725833773e-05, "loss": 0.985, "step": 41537 }, { "epoch": 0.04, "learning_rate": 4.978841706086445e-05, "loss": 0.907, "step": 41538 }, { "epoch": 0.04, "learning_rate": 4.978840686314648e-05, "loss": 1.1376, "step": 41539 }, { "epoch": 0.04, "learning_rate": 4.97883966651838e-05, "loss": 0.7565, "step": 41540 }, { "epoch": 0.04, "learning_rate": 4.9788386466976414e-05, "loss": 0.8465, "step": 41541 }, { "epoch": 0.04, "learning_rate": 4.978837626852434e-05, "loss": 0.8682, "step": 41542 }, { "epoch": 0.04, "learning_rate": 4.978836606982755e-05, "loss": 1.0209, "step": 41543 }, { "epoch": 0.04, "learning_rate": 4.978835587088607e-05, "loss": 1.045, "step": 41544 }, { "epoch": 0.04, "learning_rate": 4.97883456716999e-05, "loss": 0.9994, "step": 41545 }, { "epoch": 0.04, "learning_rate": 4.9788335472269015e-05, "loss": 1.0741, "step": 41546 }, { "epoch": 0.04, "learning_rate": 4.978832527259343e-05, "loss": 0.9521, "step": 41547 }, { "epoch": 0.04, "learning_rate": 4.978831507267315e-05, "loss": 0.9276, "step": 41548 }, { "epoch": 0.04, "learning_rate": 4.978830487250817e-05, "loss": 0.9582, "step": 41549 }, { "epoch": 0.04, "learning_rate": 4.978829467209849e-05, "loss": 0.8807, "step": 41550 }, { "epoch": 0.04, "learning_rate": 4.9788284471444114e-05, "loss": 0.8463, "step": 41551 }, { "epoch": 0.04, "learning_rate": 4.9788274270545035e-05, "loss": 1.0248, "step": 41552 }, { "epoch": 0.04, "learning_rate": 4.978826406940126e-05, "loss": 0.6777, "step": 41553 }, { "epoch": 0.04, "learning_rate": 4.978825386801278e-05, "loss": 0.7962, "step": 41554 }, { "epoch": 0.04, "learning_rate": 4.97882436663796e-05, "loss": 0.7336, "step": 41555 }, { "epoch": 0.04, "learning_rate": 4.9788233464501724e-05, "loss": 0.7592, "step": 41556 }, { "epoch": 0.04, "learning_rate": 4.978822326237915e-05, "loss": 1.2757, "step": 41557 }, { "epoch": 0.04, "learning_rate": 4.978821306001188e-05, "loss": 0.898, "step": 41558 }, { "epoch": 0.04, "learning_rate": 4.97882028573999e-05, "loss": 1.2395, "step": 41559 }, { "epoch": 0.04, "learning_rate": 4.978819265454323e-05, "loss": 1.1107, "step": 41560 }, { "epoch": 0.04, "learning_rate": 4.978818245144186e-05, "loss": 1.301, "step": 41561 }, { "epoch": 0.04, "learning_rate": 4.97881722480958e-05, "loss": 1.1411, "step": 41562 }, { "epoch": 0.04, "learning_rate": 4.9788162044505023e-05, "loss": 1.0909, "step": 41563 }, { "epoch": 0.04, "learning_rate": 4.978815184066956e-05, "loss": 0.8742, "step": 41564 }, { "epoch": 0.04, "learning_rate": 4.97881416365894e-05, "loss": 1.2157, "step": 41565 }, { "epoch": 0.04, "learning_rate": 4.978813143226454e-05, "loss": 0.9579, "step": 41566 }, { "epoch": 0.04, "learning_rate": 4.978812122769498e-05, "loss": 0.862, "step": 41567 }, { "epoch": 0.04, "learning_rate": 4.978811102288073e-05, "loss": 0.9763, "step": 41568 }, { "epoch": 0.04, "learning_rate": 4.978810081782177e-05, "loss": 1.0116, "step": 41569 }, { "epoch": 0.04, "learning_rate": 4.978809061251811e-05, "loss": 0.9979, "step": 41570 }, { "epoch": 0.04, "learning_rate": 4.9788080406969763e-05, "loss": 1.0784, "step": 41571 }, { "epoch": 0.04, "learning_rate": 4.978807020117672e-05, "loss": 0.9571, "step": 41572 }, { "epoch": 0.04, "learning_rate": 4.978805999513897e-05, "loss": 1.0027, "step": 41573 }, { "epoch": 0.04, "learning_rate": 4.978804978885653e-05, "loss": 1.0967, "step": 41574 }, { "epoch": 0.04, "learning_rate": 4.978803958232939e-05, "loss": 0.9216, "step": 41575 }, { "epoch": 0.04, "learning_rate": 4.978802937555755e-05, "loss": 0.9396, "step": 41576 }, { "epoch": 0.04, "learning_rate": 4.9788019168541014e-05, "loss": 1.5417, "step": 41577 }, { "epoch": 0.04, "learning_rate": 4.978800896127979e-05, "loss": 1.7133, "step": 41578 }, { "epoch": 0.04, "learning_rate": 4.9787998753773854e-05, "loss": 0.8149, "step": 41579 }, { "epoch": 0.04, "learning_rate": 4.9787988546023234e-05, "loss": 3.3005, "step": 41580 }, { "epoch": 0.04, "learning_rate": 4.978797833802791e-05, "loss": 5.9322, "step": 41581 }, { "epoch": 0.04, "learning_rate": 4.97879681297879e-05, "loss": 5.8657, "step": 41582 }, { "epoch": 0.04, "learning_rate": 4.978795792130319e-05, "loss": 5.8149, "step": 41583 }, { "epoch": 0.04, "learning_rate": 4.9787947712573765e-05, "loss": 5.8186, "step": 41584 }, { "epoch": 0.04, "learning_rate": 4.978793750359967e-05, "loss": 5.8094, "step": 41585 }, { "epoch": 0.04, "learning_rate": 4.9787927294380865e-05, "loss": 5.9047, "step": 41586 }, { "epoch": 0.04, "learning_rate": 4.978791708491737e-05, "loss": 5.7411, "step": 41587 }, { "epoch": 0.04, "learning_rate": 4.9787906875209174e-05, "loss": 3.0991, "step": 41588 }, { "epoch": 0.04, "learning_rate": 4.978789666525628e-05, "loss": 1.0054, "step": 41589 }, { "epoch": 0.04, "learning_rate": 4.97878864550587e-05, "loss": 0.8912, "step": 41590 }, { "epoch": 0.04, "learning_rate": 4.978787624461642e-05, "loss": 1.2098, "step": 41591 }, { "epoch": 0.04, "learning_rate": 4.978786603392944e-05, "loss": 0.9231, "step": 41592 }, { "epoch": 0.04, "learning_rate": 4.9787855822997766e-05, "loss": 0.9341, "step": 41593 }, { "epoch": 0.04, "learning_rate": 4.9787845611821406e-05, "loss": 1.0772, "step": 41594 }, { "epoch": 0.04, "learning_rate": 4.978783540040034e-05, "loss": 1.1013, "step": 41595 }, { "epoch": 0.04, "learning_rate": 4.978782518873458e-05, "loss": 1.5413, "step": 41596 }, { "epoch": 0.04, "learning_rate": 4.9787814976824134e-05, "loss": 0.8394, "step": 41597 }, { "epoch": 0.04, "learning_rate": 4.9787804764668984e-05, "loss": 1.0263, "step": 41598 }, { "epoch": 0.04, "learning_rate": 4.978779455226914e-05, "loss": 1.0331, "step": 41599 }, { "epoch": 0.04, "learning_rate": 4.9787784339624603e-05, "loss": 1.0185, "step": 41600 }, { "epoch": 0.04, "learning_rate": 4.9787774126735374e-05, "loss": 1.1961, "step": 41601 }, { "epoch": 0.04, "learning_rate": 4.9787763913601446e-05, "loss": 1.1708, "step": 41602 }, { "epoch": 0.04, "learning_rate": 4.978775370022283e-05, "loss": 1.1274, "step": 41603 }, { "epoch": 0.04, "learning_rate": 4.978774348659951e-05, "loss": 1.0332, "step": 41604 }, { "epoch": 0.04, "learning_rate": 4.9787733272731505e-05, "loss": 1.0903, "step": 41605 }, { "epoch": 0.04, "learning_rate": 4.97877230586188e-05, "loss": 0.6382, "step": 41606 }, { "epoch": 0.04, "learning_rate": 4.9787712844261405e-05, "loss": 0.5962, "step": 41607 }, { "epoch": 0.04, "learning_rate": 4.978770262965932e-05, "loss": 0.5539, "step": 41608 }, { "epoch": 0.04, "learning_rate": 4.978769241481253e-05, "loss": 0.7947, "step": 41609 }, { "epoch": 0.04, "learning_rate": 4.9787682199721053e-05, "loss": 1.1345, "step": 41610 }, { "epoch": 0.04, "learning_rate": 4.978767198438488e-05, "loss": 1.1788, "step": 41611 }, { "epoch": 0.04, "learning_rate": 4.978766176880402e-05, "loss": 1.0451, "step": 41612 }, { "epoch": 0.04, "learning_rate": 4.978765155297846e-05, "loss": 0.9822, "step": 41613 }, { "epoch": 0.04, "learning_rate": 4.978764133690821e-05, "loss": 0.8985, "step": 41614 }, { "epoch": 0.04, "learning_rate": 4.978763112059326e-05, "loss": 1.2806, "step": 41615 }, { "epoch": 0.04, "learning_rate": 4.978762090403363e-05, "loss": 0.9431, "step": 41616 }, { "epoch": 0.04, "learning_rate": 4.97876106872293e-05, "loss": 0.8795, "step": 41617 }, { "epoch": 0.04, "learning_rate": 4.9787600470180275e-05, "loss": 0.9483, "step": 41618 }, { "epoch": 0.04, "learning_rate": 4.9787590252886555e-05, "loss": 1.133, "step": 41619 }, { "epoch": 0.04, "learning_rate": 4.9787580035348145e-05, "loss": 1.1109, "step": 41620 }, { "epoch": 0.04, "learning_rate": 4.978756981756505e-05, "loss": 1.044, "step": 41621 }, { "epoch": 0.04, "learning_rate": 4.978755959953725e-05, "loss": 1.0277, "step": 41622 }, { "epoch": 0.04, "learning_rate": 4.978754938126477e-05, "loss": 1.2206, "step": 41623 }, { "epoch": 0.04, "learning_rate": 4.9787539162747586e-05, "loss": 0.7238, "step": 41624 }, { "epoch": 0.04, "learning_rate": 4.9787528943985715e-05, "loss": 1.0849, "step": 41625 }, { "epoch": 0.04, "learning_rate": 4.978751872497915e-05, "loss": 0.9558, "step": 41626 }, { "epoch": 0.04, "learning_rate": 4.9787508505727905e-05, "loss": 1.1675, "step": 41627 }, { "epoch": 0.04, "learning_rate": 4.978749828623196e-05, "loss": 1.5187, "step": 41628 }, { "epoch": 0.04, "learning_rate": 4.978748806649132e-05, "loss": 1.317, "step": 41629 }, { "epoch": 0.04, "learning_rate": 4.978747784650599e-05, "loss": 0.9134, "step": 41630 }, { "epoch": 0.04, "learning_rate": 4.978746762627597e-05, "loss": 1.1495, "step": 41631 }, { "epoch": 0.04, "learning_rate": 4.9787457405801254e-05, "loss": 1.1674, "step": 41632 }, { "epoch": 0.04, "learning_rate": 4.978744718508185e-05, "loss": 1.1557, "step": 41633 }, { "epoch": 0.04, "learning_rate": 4.978743696411776e-05, "loss": 0.72, "step": 41634 }, { "epoch": 0.04, "learning_rate": 4.978742674290897e-05, "loss": 0.6327, "step": 41635 }, { "epoch": 0.04, "learning_rate": 4.978741652145549e-05, "loss": 0.9368, "step": 41636 }, { "epoch": 0.04, "learning_rate": 4.978740629975732e-05, "loss": 1.2403, "step": 41637 }, { "epoch": 0.04, "learning_rate": 4.978739607781446e-05, "loss": 0.8422, "step": 41638 }, { "epoch": 0.04, "learning_rate": 4.978738585562692e-05, "loss": 0.8623, "step": 41639 }, { "epoch": 0.04, "learning_rate": 4.978737563319468e-05, "loss": 0.8883, "step": 41640 }, { "epoch": 0.04, "learning_rate": 4.9787365410517745e-05, "loss": 0.8078, "step": 41641 }, { "epoch": 0.04, "learning_rate": 4.978735518759612e-05, "loss": 0.9402, "step": 41642 }, { "epoch": 0.04, "learning_rate": 4.9787344964429814e-05, "loss": 0.747, "step": 41643 }, { "epoch": 0.04, "learning_rate": 4.978733474101881e-05, "loss": 0.5809, "step": 41644 }, { "epoch": 0.04, "learning_rate": 4.978732451736312e-05, "loss": 1.0436, "step": 41645 }, { "epoch": 0.04, "learning_rate": 4.978731429346274e-05, "loss": 0.903, "step": 41646 }, { "epoch": 0.04, "learning_rate": 4.978730406931766e-05, "loss": 0.8859, "step": 41647 }, { "epoch": 0.04, "learning_rate": 4.97872938449279e-05, "loss": 1.198, "step": 41648 }, { "epoch": 0.04, "learning_rate": 4.978728362029345e-05, "loss": 0.9098, "step": 41649 }, { "epoch": 0.04, "learning_rate": 4.9787273395414315e-05, "loss": 0.3261, "step": 41650 }, { "epoch": 0.04, "learning_rate": 4.9787263170290474e-05, "loss": 0.6021, "step": 41651 }, { "epoch": 0.04, "learning_rate": 4.9787252944921955e-05, "loss": 0.9895, "step": 41652 }, { "epoch": 0.04, "learning_rate": 4.9787242719308746e-05, "loss": 1.0333, "step": 41653 }, { "epoch": 0.04, "learning_rate": 4.978723249345085e-05, "loss": 1.0224, "step": 41654 }, { "epoch": 0.04, "learning_rate": 4.978722226734827e-05, "loss": 1.1011, "step": 41655 }, { "epoch": 0.04, "learning_rate": 4.9787212041000985e-05, "loss": 0.9502, "step": 41656 }, { "epoch": 0.04, "learning_rate": 4.9787201814409025e-05, "loss": 1.0286, "step": 41657 }, { "epoch": 0.04, "learning_rate": 4.978719158757237e-05, "loss": 1.3429, "step": 41658 }, { "epoch": 0.04, "learning_rate": 4.978718136049102e-05, "loss": 0.798, "step": 41659 }, { "epoch": 0.04, "learning_rate": 4.978717113316499e-05, "loss": 0.7813, "step": 41660 }, { "epoch": 0.04, "learning_rate": 4.978716090559427e-05, "loss": 1.1063, "step": 41661 }, { "epoch": 0.04, "learning_rate": 4.978715067777886e-05, "loss": 0.8652, "step": 41662 }, { "epoch": 0.04, "learning_rate": 4.9787140449718766e-05, "loss": 0.8233, "step": 41663 }, { "epoch": 0.04, "learning_rate": 4.978713022141398e-05, "loss": 0.6085, "step": 41664 }, { "epoch": 0.04, "learning_rate": 4.9787119992864504e-05, "loss": 0.9054, "step": 41665 }, { "epoch": 0.04, "learning_rate": 4.978710976407035e-05, "loss": 0.7774, "step": 41666 }, { "epoch": 0.04, "learning_rate": 4.97870995350315e-05, "loss": 1.2068, "step": 41667 }, { "epoch": 0.04, "learning_rate": 4.978708930574796e-05, "loss": 1.0268, "step": 41668 }, { "epoch": 0.04, "learning_rate": 4.978707907621974e-05, "loss": 1.4449, "step": 41669 }, { "epoch": 0.04, "learning_rate": 4.978706884644683e-05, "loss": 1.0677, "step": 41670 }, { "epoch": 0.04, "learning_rate": 4.9787058616429225e-05, "loss": 0.9488, "step": 41671 }, { "epoch": 0.04, "learning_rate": 4.9787048386166936e-05, "loss": 1.0836, "step": 41672 }, { "epoch": 0.04, "learning_rate": 4.9787038155659956e-05, "loss": 1.234, "step": 41673 }, { "epoch": 0.04, "learning_rate": 4.9787027924908306e-05, "loss": 0.8942, "step": 41674 }, { "epoch": 0.04, "learning_rate": 4.978701769391195e-05, "loss": 0.9727, "step": 41675 }, { "epoch": 0.04, "learning_rate": 4.978700746267091e-05, "loss": 1.0277, "step": 41676 }, { "epoch": 0.04, "learning_rate": 4.9786997231185196e-05, "loss": 1.3238, "step": 41677 }, { "epoch": 0.04, "learning_rate": 4.978698699945479e-05, "loss": 1.1946, "step": 41678 }, { "epoch": 0.04, "learning_rate": 4.9786976767479685e-05, "loss": 1.3559, "step": 41679 }, { "epoch": 0.04, "learning_rate": 4.97869665352599e-05, "loss": 1.1756, "step": 41680 }, { "epoch": 0.04, "learning_rate": 4.978695630279544e-05, "loss": 0.6372, "step": 41681 }, { "epoch": 0.04, "learning_rate": 4.978694607008628e-05, "loss": 1.1993, "step": 41682 }, { "epoch": 0.04, "learning_rate": 4.9786935837132445e-05, "loss": 0.9268, "step": 41683 }, { "epoch": 0.04, "learning_rate": 4.9786925603933906e-05, "loss": 0.8175, "step": 41684 }, { "epoch": 0.04, "learning_rate": 4.97869153704907e-05, "loss": 0.8922, "step": 41685 }, { "epoch": 0.04, "learning_rate": 4.97869051368028e-05, "loss": 1.3419, "step": 41686 }, { "epoch": 0.04, "learning_rate": 4.9786894902870206e-05, "loss": 1.4617, "step": 41687 }, { "epoch": 0.04, "learning_rate": 4.978688466869294e-05, "loss": 1.1141, "step": 41688 }, { "epoch": 0.04, "learning_rate": 4.978687443427098e-05, "loss": 0.841, "step": 41689 }, { "epoch": 0.04, "learning_rate": 4.978686419960434e-05, "loss": 0.9871, "step": 41690 }, { "epoch": 0.04, "learning_rate": 4.978685396469301e-05, "loss": 1.1518, "step": 41691 }, { "epoch": 0.04, "learning_rate": 4.978684372953699e-05, "loss": 1.0024, "step": 41692 }, { "epoch": 0.04, "learning_rate": 4.97868334941363e-05, "loss": 0.6559, "step": 41693 }, { "epoch": 0.04, "learning_rate": 4.9786823258490913e-05, "loss": 1.1016, "step": 41694 }, { "epoch": 0.04, "learning_rate": 4.978681302260084e-05, "loss": 0.8793, "step": 41695 }, { "epoch": 0.04, "learning_rate": 4.978680278646609e-05, "loss": 1.4717, "step": 41696 }, { "epoch": 0.04, "learning_rate": 4.978679255008665e-05, "loss": 0.7806, "step": 41697 }, { "epoch": 0.04, "learning_rate": 4.978678231346252e-05, "loss": 1.0197, "step": 41698 }, { "epoch": 0.04, "learning_rate": 4.978677207659371e-05, "loss": 1.1762, "step": 41699 }, { "epoch": 0.04, "learning_rate": 4.978676183948022e-05, "loss": 0.7341, "step": 41700 }, { "epoch": 0.04, "learning_rate": 4.9786751602122046e-05, "loss": 1.0016, "step": 41701 }, { "epoch": 0.04, "learning_rate": 4.9786741364519187e-05, "loss": 1.1506, "step": 41702 }, { "epoch": 0.04, "learning_rate": 4.9786731126671636e-05, "loss": 1.0221, "step": 41703 }, { "epoch": 0.04, "learning_rate": 4.978672088857941e-05, "loss": 1.1224, "step": 41704 }, { "epoch": 0.04, "learning_rate": 4.978671065024249e-05, "loss": 1.1899, "step": 41705 }, { "epoch": 0.04, "learning_rate": 4.9786700411660895e-05, "loss": 1.076, "step": 41706 }, { "epoch": 0.04, "learning_rate": 4.978669017283461e-05, "loss": 0.9113, "step": 41707 }, { "epoch": 0.04, "learning_rate": 4.9786679933763644e-05, "loss": 0.8184, "step": 41708 }, { "epoch": 0.04, "learning_rate": 4.9786669694447996e-05, "loss": 0.948, "step": 41709 }, { "epoch": 0.04, "learning_rate": 4.9786659454887664e-05, "loss": 1.1063, "step": 41710 }, { "epoch": 0.04, "learning_rate": 4.978664921508264e-05, "loss": 0.9291, "step": 41711 }, { "epoch": 0.04, "learning_rate": 4.978663897503294e-05, "loss": 1.0873, "step": 41712 }, { "epoch": 0.04, "learning_rate": 4.978662873473856e-05, "loss": 1.1122, "step": 41713 }, { "epoch": 0.04, "learning_rate": 4.9786618494199494e-05, "loss": 1.1485, "step": 41714 }, { "epoch": 0.04, "learning_rate": 4.978660825341575e-05, "loss": 0.7843, "step": 41715 }, { "epoch": 0.04, "learning_rate": 4.978659801238731e-05, "loss": 0.5159, "step": 41716 }, { "epoch": 0.04, "learning_rate": 4.97865877711142e-05, "loss": 1.1093, "step": 41717 }, { "epoch": 0.04, "learning_rate": 4.97865775295964e-05, "loss": 0.8315, "step": 41718 }, { "epoch": 0.04, "learning_rate": 4.9786567287833917e-05, "loss": 0.7692, "step": 41719 }, { "epoch": 0.04, "learning_rate": 4.978655704582676e-05, "loss": 0.9052, "step": 41720 }, { "epoch": 0.04, "learning_rate": 4.9786546803574906e-05, "loss": 0.8731, "step": 41721 }, { "epoch": 0.04, "learning_rate": 4.978653656107838e-05, "loss": 0.9449, "step": 41722 }, { "epoch": 0.04, "learning_rate": 4.978652631833717e-05, "loss": 0.8601, "step": 41723 }, { "epoch": 0.04, "learning_rate": 4.978651607535128e-05, "loss": 0.7358, "step": 41724 }, { "epoch": 0.04, "learning_rate": 4.97865058321207e-05, "loss": 0.7117, "step": 41725 }, { "epoch": 0.04, "learning_rate": 4.9786495588645447e-05, "loss": 1.029, "step": 41726 }, { "epoch": 0.04, "learning_rate": 4.9786485344925505e-05, "loss": 0.7824, "step": 41727 }, { "epoch": 0.04, "learning_rate": 4.9786475100960886e-05, "loss": 1.0328, "step": 41728 }, { "epoch": 0.04, "learning_rate": 4.978646485675158e-05, "loss": 0.9886, "step": 41729 }, { "epoch": 0.04, "learning_rate": 4.97864546122976e-05, "loss": 0.9992, "step": 41730 }, { "epoch": 0.04, "learning_rate": 4.978644436759894e-05, "loss": 0.849, "step": 41731 }, { "epoch": 0.04, "learning_rate": 4.978643412265559e-05, "loss": 1.0968, "step": 41732 }, { "epoch": 0.04, "learning_rate": 4.978642387746756e-05, "loss": 1.1532, "step": 41733 }, { "epoch": 0.04, "learning_rate": 4.9786413632034854e-05, "loss": 0.857, "step": 41734 }, { "epoch": 0.04, "learning_rate": 4.978640338635746e-05, "loss": 1.1479, "step": 41735 }, { "epoch": 0.04, "learning_rate": 4.9786393140435395e-05, "loss": 1.0138, "step": 41736 }, { "epoch": 0.04, "learning_rate": 4.9786382894268647e-05, "loss": 1.2694, "step": 41737 }, { "epoch": 0.04, "learning_rate": 4.978637264785721e-05, "loss": 1.463, "step": 41738 }, { "epoch": 0.04, "learning_rate": 4.97863624012011e-05, "loss": 0.9457, "step": 41739 }, { "epoch": 0.04, "learning_rate": 4.978635215430031e-05, "loss": 1.1236, "step": 41740 }, { "epoch": 0.04, "learning_rate": 4.978634190715483e-05, "loss": 1.4244, "step": 41741 }, { "epoch": 0.04, "learning_rate": 4.9786331659764675e-05, "loss": 1.1833, "step": 41742 }, { "epoch": 0.04, "learning_rate": 4.978632141212984e-05, "loss": 1.0467, "step": 41743 }, { "epoch": 0.04, "learning_rate": 4.978631116425033e-05, "loss": 1.3316, "step": 41744 }, { "epoch": 0.04, "learning_rate": 4.978630091612614e-05, "loss": 1.2116, "step": 41745 }, { "epoch": 0.04, "learning_rate": 4.9786290667757266e-05, "loss": 0.755, "step": 41746 }, { "epoch": 0.04, "learning_rate": 4.97862804191437e-05, "loss": 0.995, "step": 41747 }, { "epoch": 0.04, "learning_rate": 4.978627017028548e-05, "loss": 1.0632, "step": 41748 }, { "epoch": 0.04, "learning_rate": 4.978625992118256e-05, "loss": 0.7551, "step": 41749 }, { "epoch": 0.04, "learning_rate": 4.9786249671834965e-05, "loss": 0.9347, "step": 41750 }, { "epoch": 0.04, "learning_rate": 4.97862394222427e-05, "loss": 1.4451, "step": 41751 }, { "epoch": 0.04, "learning_rate": 4.9786229172405744e-05, "loss": 1.3419, "step": 41752 }, { "epoch": 0.04, "learning_rate": 4.978621892232411e-05, "loss": 1.0109, "step": 41753 }, { "epoch": 0.04, "learning_rate": 4.97862086719978e-05, "loss": 1.0336, "step": 41754 }, { "epoch": 0.04, "learning_rate": 4.978619842142681e-05, "loss": 1.2453, "step": 41755 }, { "epoch": 0.04, "learning_rate": 4.9786188170611147e-05, "loss": 1.0519, "step": 41756 }, { "epoch": 0.04, "learning_rate": 4.97861779195508e-05, "loss": 0.777, "step": 41757 }, { "epoch": 0.04, "learning_rate": 4.978616766824578e-05, "loss": 1.0783, "step": 41758 }, { "epoch": 0.04, "learning_rate": 4.9786157416696066e-05, "loss": 0.9003, "step": 41759 }, { "epoch": 0.04, "learning_rate": 4.9786147164901685e-05, "loss": 1.135, "step": 41760 }, { "epoch": 0.04, "learning_rate": 4.9786136912862626e-05, "loss": 0.9451, "step": 41761 }, { "epoch": 0.04, "learning_rate": 4.978612666057888e-05, "loss": 1.0367, "step": 41762 }, { "epoch": 0.04, "learning_rate": 4.978611640805047e-05, "loss": 1.1601, "step": 41763 }, { "epoch": 0.04, "learning_rate": 4.978610615527737e-05, "loss": 1.0165, "step": 41764 }, { "epoch": 0.04, "learning_rate": 4.97860959022596e-05, "loss": 1.0432, "step": 41765 }, { "epoch": 0.04, "learning_rate": 4.978608564899715e-05, "loss": 0.7482, "step": 41766 }, { "epoch": 0.04, "learning_rate": 4.9786075395490016e-05, "loss": 1.0208, "step": 41767 }, { "epoch": 0.04, "learning_rate": 4.9786065141738216e-05, "loss": 1.1923, "step": 41768 }, { "epoch": 0.04, "learning_rate": 4.978605488774173e-05, "loss": 1.1727, "step": 41769 }, { "epoch": 0.04, "learning_rate": 4.9786044633500564e-05, "loss": 0.69, "step": 41770 }, { "epoch": 0.04, "learning_rate": 4.978603437901473e-05, "loss": 1.0808, "step": 41771 }, { "epoch": 0.04, "learning_rate": 4.9786024124284216e-05, "loss": 1.2684, "step": 41772 }, { "epoch": 0.04, "learning_rate": 4.9786013869309016e-05, "loss": 0.8209, "step": 41773 }, { "epoch": 0.04, "learning_rate": 4.978600361408915e-05, "loss": 1.1877, "step": 41774 }, { "epoch": 0.04, "learning_rate": 4.97859933586246e-05, "loss": 1.204, "step": 41775 }, { "epoch": 0.04, "learning_rate": 4.978598310291537e-05, "loss": 1.0205, "step": 41776 }, { "epoch": 0.04, "learning_rate": 4.978597284696147e-05, "loss": 0.5764, "step": 41777 }, { "epoch": 0.04, "learning_rate": 4.978596259076289e-05, "loss": 1.1456, "step": 41778 }, { "epoch": 0.04, "learning_rate": 4.978595233431964e-05, "loss": 0.8904, "step": 41779 }, { "epoch": 0.04, "learning_rate": 4.978594207763171e-05, "loss": 1.1218, "step": 41780 }, { "epoch": 0.04, "learning_rate": 4.97859318206991e-05, "loss": 0.7835, "step": 41781 }, { "epoch": 0.04, "learning_rate": 4.978592156352182e-05, "loss": 0.8894, "step": 41782 }, { "epoch": 0.04, "learning_rate": 4.9785911306099856e-05, "loss": 0.989, "step": 41783 }, { "epoch": 0.04, "learning_rate": 4.978590104843322e-05, "loss": 1.1678, "step": 41784 }, { "epoch": 0.04, "learning_rate": 4.9785890790521904e-05, "loss": 1.049, "step": 41785 }, { "epoch": 0.04, "learning_rate": 4.9785880532365926e-05, "loss": 0.8575, "step": 41786 }, { "epoch": 0.04, "learning_rate": 4.9785870273965256e-05, "loss": 1.1115, "step": 41787 }, { "epoch": 0.04, "learning_rate": 4.978586001531992e-05, "loss": 1.0346, "step": 41788 }, { "epoch": 0.04, "learning_rate": 4.9785849756429906e-05, "loss": 0.8551, "step": 41789 }, { "epoch": 0.04, "learning_rate": 4.978583949729521e-05, "loss": 0.8041, "step": 41790 }, { "epoch": 0.04, "learning_rate": 4.978582923791585e-05, "loss": 0.9785, "step": 41791 }, { "epoch": 0.04, "learning_rate": 4.978581897829181e-05, "loss": 1.4908, "step": 41792 }, { "epoch": 0.04, "learning_rate": 4.9785808718423085e-05, "loss": 1.263, "step": 41793 }, { "epoch": 0.04, "learning_rate": 4.9785798458309695e-05, "loss": 1.2248, "step": 41794 }, { "epoch": 0.04, "learning_rate": 4.9785788197951634e-05, "loss": 0.9106, "step": 41795 }, { "epoch": 0.04, "learning_rate": 4.978577793734889e-05, "loss": 0.7043, "step": 41796 }, { "epoch": 0.04, "learning_rate": 4.9785767676501474e-05, "loss": 0.79, "step": 41797 }, { "epoch": 0.04, "learning_rate": 4.978575741540938e-05, "loss": 0.706, "step": 41798 }, { "epoch": 0.04, "learning_rate": 4.978574715407262e-05, "loss": 0.6587, "step": 41799 }, { "epoch": 0.04, "learning_rate": 4.978573689249118e-05, "loss": 0.7197, "step": 41800 }, { "epoch": 0.04, "learning_rate": 4.978572663066507e-05, "loss": 0.625, "step": 41801 }, { "epoch": 0.04, "learning_rate": 4.978571636859428e-05, "loss": 0.8843, "step": 41802 }, { "epoch": 0.04, "learning_rate": 4.978570610627882e-05, "loss": 1.038, "step": 41803 }, { "epoch": 0.04, "learning_rate": 4.978569584371868e-05, "loss": 1.4417, "step": 41804 }, { "epoch": 0.04, "learning_rate": 4.978568558091386e-05, "loss": 0.668, "step": 41805 }, { "epoch": 0.04, "learning_rate": 4.9785675317864385e-05, "loss": 0.3567, "step": 41806 }, { "epoch": 0.04, "learning_rate": 4.978566505457023e-05, "loss": 1.2339, "step": 41807 }, { "epoch": 0.04, "learning_rate": 4.97856547910314e-05, "loss": 1.247, "step": 41808 }, { "epoch": 0.04, "learning_rate": 4.978564452724789e-05, "loss": 1.078, "step": 41809 }, { "epoch": 0.04, "learning_rate": 4.9785634263219716e-05, "loss": 1.0129, "step": 41810 }, { "epoch": 0.04, "learning_rate": 4.978562399894686e-05, "loss": 0.9433, "step": 41811 }, { "epoch": 0.04, "learning_rate": 4.978561373442934e-05, "loss": 0.7192, "step": 41812 }, { "epoch": 0.04, "learning_rate": 4.978560346966714e-05, "loss": 0.851, "step": 41813 }, { "epoch": 0.04, "learning_rate": 4.9785593204660264e-05, "loss": 0.5452, "step": 41814 }, { "epoch": 0.04, "learning_rate": 4.9785582939408725e-05, "loss": 0.8635, "step": 41815 }, { "epoch": 0.04, "learning_rate": 4.978557267391251e-05, "loss": 0.9556, "step": 41816 }, { "epoch": 0.04, "learning_rate": 4.9785562408171614e-05, "loss": 1.0575, "step": 41817 }, { "epoch": 0.04, "learning_rate": 4.9785552142186056e-05, "loss": 1.0094, "step": 41818 }, { "epoch": 0.04, "learning_rate": 4.978554187595582e-05, "loss": 1.1081, "step": 41819 }, { "epoch": 0.04, "learning_rate": 4.9785531609480915e-05, "loss": 0.7079, "step": 41820 }, { "epoch": 0.04, "learning_rate": 4.978552134276133e-05, "loss": 1.1802, "step": 41821 }, { "epoch": 0.04, "learning_rate": 4.978551107579708e-05, "loss": 0.9618, "step": 41822 }, { "epoch": 0.04, "learning_rate": 4.978550080858816e-05, "loss": 0.7795, "step": 41823 }, { "epoch": 0.04, "learning_rate": 4.978549054113456e-05, "loss": 0.943, "step": 41824 }, { "epoch": 0.04, "learning_rate": 4.9785480273436294e-05, "loss": 1.0025, "step": 41825 }, { "epoch": 0.04, "learning_rate": 4.978547000549335e-05, "loss": 0.8763, "step": 41826 }, { "epoch": 0.04, "learning_rate": 4.978545973730574e-05, "loss": 0.6836, "step": 41827 }, { "epoch": 0.04, "learning_rate": 4.9785449468873455e-05, "loss": 0.788, "step": 41828 }, { "epoch": 0.04, "learning_rate": 4.978543920019651e-05, "loss": 0.5872, "step": 41829 }, { "epoch": 0.04, "learning_rate": 4.978542893127488e-05, "loss": 0.5076, "step": 41830 }, { "epoch": 0.04, "learning_rate": 4.978541866210859e-05, "loss": 0.5415, "step": 41831 }, { "epoch": 0.04, "learning_rate": 4.9785408392697615e-05, "loss": 0.7801, "step": 41832 }, { "epoch": 0.04, "learning_rate": 4.978539812304197e-05, "loss": 1.0428, "step": 41833 }, { "epoch": 0.04, "learning_rate": 4.9785387853141665e-05, "loss": 1.3157, "step": 41834 }, { "epoch": 0.04, "learning_rate": 4.978537758299669e-05, "loss": 1.3762, "step": 41835 }, { "epoch": 0.04, "learning_rate": 4.9785367312607026e-05, "loss": 0.9221, "step": 41836 }, { "epoch": 0.04, "learning_rate": 4.97853570419727e-05, "loss": 1.4136, "step": 41837 }, { "epoch": 0.04, "learning_rate": 4.978534677109371e-05, "loss": 0.8915, "step": 41838 }, { "epoch": 0.04, "learning_rate": 4.9785336499970045e-05, "loss": 0.9973, "step": 41839 }, { "epoch": 0.04, "learning_rate": 4.9785326228601714e-05, "loss": 0.685, "step": 41840 }, { "epoch": 0.04, "learning_rate": 4.978531595698871e-05, "loss": 2.5145, "step": 41841 }, { "epoch": 0.04, "learning_rate": 4.9785305685131035e-05, "loss": 1.5555, "step": 41842 }, { "epoch": 0.04, "learning_rate": 4.978529541302869e-05, "loss": 1.5005, "step": 41843 }, { "epoch": 0.04, "learning_rate": 4.978528514068167e-05, "loss": 1.4096, "step": 41844 }, { "epoch": 0.04, "learning_rate": 4.978527486808999e-05, "loss": 1.0146, "step": 41845 }, { "epoch": 0.04, "learning_rate": 4.978526459525363e-05, "loss": 0.8681, "step": 41846 }, { "epoch": 0.04, "learning_rate": 4.978525432217261e-05, "loss": 1.1896, "step": 41847 }, { "epoch": 0.04, "learning_rate": 4.978524404884691e-05, "loss": 1.4484, "step": 41848 }, { "epoch": 0.04, "learning_rate": 4.9785233775276554e-05, "loss": 0.9581, "step": 41849 }, { "epoch": 0.04, "learning_rate": 4.978522350146152e-05, "loss": 1.1346, "step": 41850 }, { "epoch": 0.04, "learning_rate": 4.978521322740182e-05, "loss": 0.838, "step": 41851 }, { "epoch": 0.04, "learning_rate": 4.978520295309744e-05, "loss": 1.1778, "step": 41852 }, { "epoch": 0.04, "learning_rate": 4.978519267854841e-05, "loss": 0.7333, "step": 41853 }, { "epoch": 0.04, "learning_rate": 4.978518240375469e-05, "loss": 1.0586, "step": 41854 }, { "epoch": 0.04, "learning_rate": 4.9785172128716315e-05, "loss": 1.2492, "step": 41855 }, { "epoch": 0.04, "learning_rate": 4.978516185343327e-05, "loss": 1.0398, "step": 41856 }, { "epoch": 0.04, "learning_rate": 4.9785151577905553e-05, "loss": 1.1311, "step": 41857 }, { "epoch": 0.04, "learning_rate": 4.9785141302133174e-05, "loss": 1.0741, "step": 41858 }, { "epoch": 0.04, "learning_rate": 4.9785131026116116e-05, "loss": 1.2302, "step": 41859 }, { "epoch": 0.04, "learning_rate": 4.9785120749854396e-05, "loss": 0.7887, "step": 41860 }, { "epoch": 0.04, "learning_rate": 4.9785110473348004e-05, "loss": 1.0376, "step": 41861 }, { "epoch": 0.04, "learning_rate": 4.978510019659695e-05, "loss": 0.9561, "step": 41862 }, { "epoch": 0.04, "learning_rate": 4.978508991960122e-05, "loss": 0.9527, "step": 41863 }, { "epoch": 0.04, "learning_rate": 4.9785079642360834e-05, "loss": 1.0988, "step": 41864 }, { "epoch": 0.04, "learning_rate": 4.978506936487577e-05, "loss": 0.798, "step": 41865 }, { "epoch": 0.04, "learning_rate": 4.978505908714604e-05, "loss": 0.8391, "step": 41866 }, { "epoch": 0.04, "learning_rate": 4.978504880917164e-05, "loss": 1.2621, "step": 41867 }, { "epoch": 0.04, "learning_rate": 4.978503853095258e-05, "loss": 1.2989, "step": 41868 }, { "epoch": 0.04, "learning_rate": 4.978502825248885e-05, "loss": 0.9042, "step": 41869 }, { "epoch": 0.04, "learning_rate": 4.978501797378045e-05, "loss": 1.2476, "step": 41870 }, { "epoch": 0.04, "learning_rate": 4.978500769482738e-05, "loss": 1.6848, "step": 41871 }, { "epoch": 0.04, "learning_rate": 4.9784997415629645e-05, "loss": 1.0286, "step": 41872 }, { "epoch": 0.04, "learning_rate": 4.978498713618725e-05, "loss": 0.9087, "step": 41873 }, { "epoch": 0.04, "learning_rate": 4.978497685650017e-05, "loss": 1.1347, "step": 41874 }, { "epoch": 0.04, "learning_rate": 4.978496657656845e-05, "loss": 1.3111, "step": 41875 }, { "epoch": 0.04, "learning_rate": 4.9784956296392046e-05, "loss": 0.8416, "step": 41876 }, { "epoch": 0.04, "learning_rate": 4.978494601597098e-05, "loss": 1.0426, "step": 41877 }, { "epoch": 0.04, "learning_rate": 4.9784935735305237e-05, "loss": 1.2041, "step": 41878 }, { "epoch": 0.04, "learning_rate": 4.9784925454394836e-05, "loss": 1.1071, "step": 41879 }, { "epoch": 0.04, "learning_rate": 4.978491517323978e-05, "loss": 1.3589, "step": 41880 }, { "epoch": 0.04, "learning_rate": 4.978490489184004e-05, "loss": 1.1949, "step": 41881 }, { "epoch": 0.04, "learning_rate": 4.9784894610195646e-05, "loss": 1.2351, "step": 41882 }, { "epoch": 0.04, "learning_rate": 4.978488432830658e-05, "loss": 1.1794, "step": 41883 }, { "epoch": 0.04, "learning_rate": 4.9784874046172844e-05, "loss": 0.8076, "step": 41884 }, { "epoch": 0.04, "learning_rate": 4.978486376379444e-05, "loss": 1.2011, "step": 41885 }, { "epoch": 0.04, "learning_rate": 4.978485348117139e-05, "loss": 1.2103, "step": 41886 }, { "epoch": 0.04, "learning_rate": 4.9784843198303656e-05, "loss": 0.7747, "step": 41887 }, { "epoch": 0.04, "learning_rate": 4.978483291519126e-05, "loss": 1.0283, "step": 41888 }, { "epoch": 0.04, "learning_rate": 4.9784822631834196e-05, "loss": 1.0663, "step": 41889 }, { "epoch": 0.04, "learning_rate": 4.978481234823248e-05, "loss": 0.7727, "step": 41890 }, { "epoch": 0.04, "learning_rate": 4.978480206438608e-05, "loss": 0.8588, "step": 41891 }, { "epoch": 0.04, "learning_rate": 4.9784791780295025e-05, "loss": 0.9858, "step": 41892 }, { "epoch": 0.04, "learning_rate": 4.978478149595931e-05, "loss": 0.7719, "step": 41893 }, { "epoch": 0.04, "learning_rate": 4.978477121137892e-05, "loss": 0.7576, "step": 41894 }, { "epoch": 0.04, "learning_rate": 4.978476092655387e-05, "loss": 0.7586, "step": 41895 }, { "epoch": 0.04, "learning_rate": 4.9784750641484154e-05, "loss": 0.5968, "step": 41896 }, { "epoch": 0.04, "learning_rate": 4.978474035616978e-05, "loss": 1.722, "step": 41897 }, { "epoch": 0.04, "learning_rate": 4.978473007061073e-05, "loss": 1.7212, "step": 41898 }, { "epoch": 0.04, "learning_rate": 4.9784719784807024e-05, "loss": 1.7314, "step": 41899 }, { "epoch": 0.04, "learning_rate": 4.978470949875865e-05, "loss": 1.5642, "step": 41900 }, { "epoch": 0.04, "learning_rate": 4.978469921246562e-05, "loss": 0.8548, "step": 41901 }, { "epoch": 0.04, "learning_rate": 4.978468892592791e-05, "loss": 1.0447, "step": 41902 }, { "epoch": 0.04, "learning_rate": 4.9784678639145554e-05, "loss": 1.3985, "step": 41903 }, { "epoch": 0.04, "learning_rate": 4.9784668352118525e-05, "loss": 1.2606, "step": 41904 }, { "epoch": 0.04, "learning_rate": 4.9784658064846826e-05, "loss": 0.7309, "step": 41905 }, { "epoch": 0.04, "learning_rate": 4.978464777733047e-05, "loss": 1.0225, "step": 41906 }, { "epoch": 0.04, "learning_rate": 4.978463748956945e-05, "loss": 1.078, "step": 41907 }, { "epoch": 0.04, "learning_rate": 4.978462720156377e-05, "loss": 0.945, "step": 41908 }, { "epoch": 0.04, "learning_rate": 4.978461691331342e-05, "loss": 1.4417, "step": 41909 }, { "epoch": 0.04, "learning_rate": 4.9784606624818415e-05, "loss": 0.8945, "step": 41910 }, { "epoch": 0.04, "learning_rate": 4.978459633607874e-05, "loss": 0.9329, "step": 41911 }, { "epoch": 0.04, "learning_rate": 4.978458604709441e-05, "loss": 0.9706, "step": 41912 }, { "epoch": 0.04, "learning_rate": 4.9784575757865405e-05, "loss": 0.8955, "step": 41913 }, { "epoch": 0.04, "learning_rate": 4.978456546839174e-05, "loss": 1.1979, "step": 41914 }, { "epoch": 0.04, "learning_rate": 4.978455517867342e-05, "loss": 0.7929, "step": 41915 }, { "epoch": 0.04, "learning_rate": 4.978454488871043e-05, "loss": 1.1118, "step": 41916 }, { "epoch": 0.04, "learning_rate": 4.978453459850277e-05, "loss": 1.1264, "step": 41917 }, { "epoch": 0.04, "learning_rate": 4.978452430805046e-05, "loss": 1.0089, "step": 41918 }, { "epoch": 0.04, "learning_rate": 4.978451401735349e-05, "loss": 0.85, "step": 41919 }, { "epoch": 0.04, "learning_rate": 4.978450372641185e-05, "loss": 1.2344, "step": 41920 }, { "epoch": 0.04, "learning_rate": 4.978449343522556e-05, "loss": 1.2423, "step": 41921 }, { "epoch": 0.04, "learning_rate": 4.9784483143794594e-05, "loss": 1.2241, "step": 41922 }, { "epoch": 0.04, "learning_rate": 4.978447285211897e-05, "loss": 1.1314, "step": 41923 }, { "epoch": 0.04, "learning_rate": 4.978446256019868e-05, "loss": 0.9375, "step": 41924 }, { "epoch": 0.04, "learning_rate": 4.978445226803373e-05, "loss": 1.0004, "step": 41925 }, { "epoch": 0.04, "learning_rate": 4.9784441975624125e-05, "loss": 1.2785, "step": 41926 }, { "epoch": 0.04, "learning_rate": 4.978443168296986e-05, "loss": 0.8937, "step": 41927 }, { "epoch": 0.04, "learning_rate": 4.9784421390070925e-05, "loss": 0.9424, "step": 41928 }, { "epoch": 0.04, "learning_rate": 4.978441109692733e-05, "loss": 1.1508, "step": 41929 }, { "epoch": 0.04, "learning_rate": 4.9784400803539077e-05, "loss": 0.9061, "step": 41930 }, { "epoch": 0.04, "learning_rate": 4.978439050990615e-05, "loss": 0.7193, "step": 41931 }, { "epoch": 0.04, "learning_rate": 4.978438021602858e-05, "loss": 1.0986, "step": 41932 }, { "epoch": 0.04, "learning_rate": 4.978436992190634e-05, "loss": 0.9549, "step": 41933 }, { "epoch": 0.04, "learning_rate": 4.978435962753945e-05, "loss": 0.8348, "step": 41934 }, { "epoch": 0.04, "learning_rate": 4.978434933292788e-05, "loss": 0.9622, "step": 41935 }, { "epoch": 0.04, "learning_rate": 4.978433903807166e-05, "loss": 1.521, "step": 41936 }, { "epoch": 0.04, "learning_rate": 4.978432874297079e-05, "loss": 1.0772, "step": 41937 }, { "epoch": 0.04, "learning_rate": 4.978431844762525e-05, "loss": 1.0034, "step": 41938 }, { "epoch": 0.04, "learning_rate": 4.978430815203504e-05, "loss": 1.1255, "step": 41939 }, { "epoch": 0.04, "learning_rate": 4.978429785620018e-05, "loss": 1.0701, "step": 41940 }, { "epoch": 0.04, "learning_rate": 4.9784287560120657e-05, "loss": 1.3452, "step": 41941 }, { "epoch": 0.04, "learning_rate": 4.978427726379647e-05, "loss": 1.0753, "step": 41942 }, { "epoch": 0.04, "learning_rate": 4.978426696722763e-05, "loss": 1.3343, "step": 41943 }, { "epoch": 0.04, "learning_rate": 4.978425667041413e-05, "loss": 1.0741, "step": 41944 }, { "epoch": 0.04, "learning_rate": 4.978424637335597e-05, "loss": 1.3182, "step": 41945 }, { "epoch": 0.04, "learning_rate": 4.9784236076053146e-05, "loss": 1.2454, "step": 41946 }, { "epoch": 0.04, "learning_rate": 4.978422577850567e-05, "loss": 1.047, "step": 41947 }, { "epoch": 0.04, "learning_rate": 4.978421548071352e-05, "loss": 1.1208, "step": 41948 }, { "epoch": 0.04, "learning_rate": 4.9784205182676726e-05, "loss": 1.0982, "step": 41949 }, { "epoch": 0.04, "learning_rate": 4.9784194884395266e-05, "loss": 0.9319, "step": 41950 }, { "epoch": 0.04, "learning_rate": 4.9784184585869155e-05, "loss": 1.1192, "step": 41951 }, { "epoch": 0.04, "learning_rate": 4.9784174287098374e-05, "loss": 1.0591, "step": 41952 }, { "epoch": 0.04, "learning_rate": 4.9784163988082936e-05, "loss": 0.5936, "step": 41953 }, { "epoch": 0.04, "learning_rate": 4.978415368882284e-05, "loss": 0.9876, "step": 41954 }, { "epoch": 0.04, "learning_rate": 4.9784143389318095e-05, "loss": 0.881, "step": 41955 }, { "epoch": 0.04, "learning_rate": 4.978413308956868e-05, "loss": 1.05, "step": 41956 }, { "epoch": 0.04, "learning_rate": 4.9784122789574606e-05, "loss": 1.188, "step": 41957 }, { "epoch": 0.04, "learning_rate": 4.9784112489335876e-05, "loss": 1.1286, "step": 41958 }, { "epoch": 0.04, "learning_rate": 4.9784102188852496e-05, "loss": 0.8866, "step": 41959 }, { "epoch": 0.04, "learning_rate": 4.978409188812445e-05, "loss": 1.1008, "step": 41960 }, { "epoch": 0.04, "learning_rate": 4.9784081587151744e-05, "loss": 1.1196, "step": 41961 }, { "epoch": 0.04, "learning_rate": 4.9784071285934386e-05, "loss": 0.9751, "step": 41962 }, { "epoch": 0.04, "learning_rate": 4.9784060984472365e-05, "loss": 0.7789, "step": 41963 }, { "epoch": 0.04, "learning_rate": 4.9784050682765686e-05, "loss": 0.9974, "step": 41964 }, { "epoch": 0.04, "learning_rate": 4.978404038081435e-05, "loss": 0.8554, "step": 41965 }, { "epoch": 0.04, "learning_rate": 4.978403007861836e-05, "loss": 0.9328, "step": 41966 }, { "epoch": 0.04, "learning_rate": 4.978401977617771e-05, "loss": 1.0479, "step": 41967 }, { "epoch": 0.04, "learning_rate": 4.97840094734924e-05, "loss": 1.0798, "step": 41968 }, { "epoch": 0.04, "learning_rate": 4.978399917056244e-05, "loss": 0.8057, "step": 41969 }, { "epoch": 0.04, "learning_rate": 4.978398886738782e-05, "loss": 0.6598, "step": 41970 }, { "epoch": 0.04, "learning_rate": 4.978397856396854e-05, "loss": 0.8601, "step": 41971 }, { "epoch": 0.04, "learning_rate": 4.9783968260304604e-05, "loss": 0.8795, "step": 41972 }, { "epoch": 0.04, "learning_rate": 4.978395795639601e-05, "loss": 1.4822, "step": 41973 }, { "epoch": 0.04, "learning_rate": 4.978394765224276e-05, "loss": 1.2262, "step": 41974 }, { "epoch": 0.04, "learning_rate": 4.978393734784485e-05, "loss": 1.0016, "step": 41975 }, { "epoch": 0.04, "learning_rate": 4.978392704320229e-05, "loss": 1.116, "step": 41976 }, { "epoch": 0.04, "learning_rate": 4.978391673831507e-05, "loss": 0.9434, "step": 41977 }, { "epoch": 0.04, "learning_rate": 4.978390643318319e-05, "loss": 0.8267, "step": 41978 }, { "epoch": 0.04, "learning_rate": 4.978389612780666e-05, "loss": 0.4269, "step": 41979 }, { "epoch": 0.04, "learning_rate": 4.9783885822185475e-05, "loss": 0.6418, "step": 41980 }, { "epoch": 0.04, "learning_rate": 4.978387551631963e-05, "loss": 1.2706, "step": 41981 }, { "epoch": 0.04, "learning_rate": 4.978386521020912e-05, "loss": 1.0435, "step": 41982 }, { "epoch": 0.04, "learning_rate": 4.9783854903853974e-05, "loss": 1.0921, "step": 41983 }, { "epoch": 0.04, "learning_rate": 4.9783844597254156e-05, "loss": 1.1282, "step": 41984 }, { "epoch": 0.04, "learning_rate": 4.9783834290409694e-05, "loss": 1.1572, "step": 41985 }, { "epoch": 0.04, "learning_rate": 4.978382398332057e-05, "loss": 0.9355, "step": 41986 }, { "epoch": 0.04, "learning_rate": 4.978381367598679e-05, "loss": 0.9485, "step": 41987 }, { "epoch": 0.04, "learning_rate": 4.978380336840835e-05, "loss": 0.9022, "step": 41988 }, { "epoch": 0.04, "learning_rate": 4.978379306058527e-05, "loss": 1.0746, "step": 41989 }, { "epoch": 0.04, "learning_rate": 4.978378275251752e-05, "loss": 0.9542, "step": 41990 }, { "epoch": 0.04, "learning_rate": 4.9783772444205124e-05, "loss": 0.7699, "step": 41991 }, { "epoch": 0.04, "learning_rate": 4.978376213564807e-05, "loss": 1.0302, "step": 41992 }, { "epoch": 0.04, "learning_rate": 4.978375182684636e-05, "loss": 1.0871, "step": 41993 }, { "epoch": 0.04, "learning_rate": 4.97837415178e-05, "loss": 0.9932, "step": 41994 }, { "epoch": 0.04, "learning_rate": 4.978373120850898e-05, "loss": 0.9003, "step": 41995 }, { "epoch": 0.04, "learning_rate": 4.97837208989733e-05, "loss": 0.6993, "step": 41996 }, { "epoch": 0.04, "learning_rate": 4.978371058919298e-05, "loss": 0.8567, "step": 41997 }, { "epoch": 0.04, "learning_rate": 4.978370027916799e-05, "loss": 1.0501, "step": 41998 }, { "epoch": 0.04, "learning_rate": 4.9783689968898355e-05, "loss": 1.2543, "step": 41999 }, { "epoch": 0.04, "learning_rate": 4.9783679658384065e-05, "loss": 0.9401, "step": 42000 }, { "epoch": 0.04, "eval_loss": 1.0778989791870117, "eval_runtime": 124.7396, "eval_samples_per_second": 11.103, "eval_steps_per_second": 5.556, "step": 42000 }, { "epoch": 0.04, "learning_rate": 4.978366934762512e-05, "loss": 0.9253, "step": 42001 }, { "epoch": 0.04, "learning_rate": 4.978365903662152e-05, "loss": 0.9789, "step": 42002 }, { "epoch": 0.04, "learning_rate": 4.9783648725373266e-05, "loss": 0.9264, "step": 42003 }, { "epoch": 0.04, "learning_rate": 4.978363841388036e-05, "loss": 0.954, "step": 42004 }, { "epoch": 0.04, "learning_rate": 4.978362810214281e-05, "loss": 1.1791, "step": 42005 }, { "epoch": 0.04, "learning_rate": 4.978361779016059e-05, "loss": 1.2553, "step": 42006 }, { "epoch": 0.04, "learning_rate": 4.978360747793373e-05, "loss": 0.8491, "step": 42007 }, { "epoch": 0.04, "learning_rate": 4.97835971654622e-05, "loss": 1.1352, "step": 42008 }, { "epoch": 0.04, "learning_rate": 4.978358685274603e-05, "loss": 1.0373, "step": 42009 }, { "epoch": 0.04, "learning_rate": 4.9783576539785207e-05, "loss": 1.2252, "step": 42010 }, { "epoch": 0.04, "learning_rate": 4.9783566226579724e-05, "loss": 1.2636, "step": 42011 }, { "epoch": 0.04, "learning_rate": 4.978355591312959e-05, "loss": 1.2686, "step": 42012 }, { "epoch": 0.04, "learning_rate": 4.9783545599434814e-05, "loss": 1.0725, "step": 42013 }, { "epoch": 0.04, "learning_rate": 4.9783535285495366e-05, "loss": 1.0061, "step": 42014 }, { "epoch": 0.04, "learning_rate": 4.978352497131128e-05, "loss": 1.0647, "step": 42015 }, { "epoch": 0.04, "learning_rate": 4.9783514656882535e-05, "loss": 0.9839, "step": 42016 }, { "epoch": 0.04, "learning_rate": 4.9783504342209144e-05, "loss": 0.8229, "step": 42017 }, { "epoch": 0.04, "learning_rate": 4.9783494027291096e-05, "loss": 0.9433, "step": 42018 }, { "epoch": 0.04, "learning_rate": 4.978348371212839e-05, "loss": 0.991, "step": 42019 }, { "epoch": 0.04, "learning_rate": 4.978347339672104e-05, "loss": 0.9786, "step": 42020 }, { "epoch": 0.04, "learning_rate": 4.9783463081069036e-05, "loss": 0.7502, "step": 42021 }, { "epoch": 0.04, "learning_rate": 4.978345276517238e-05, "loss": 0.8889, "step": 42022 }, { "epoch": 0.04, "learning_rate": 4.978344244903107e-05, "loss": 1.0371, "step": 42023 }, { "epoch": 0.04, "learning_rate": 4.978343213264511e-05, "loss": 1.2425, "step": 42024 }, { "epoch": 0.04, "learning_rate": 4.9783421816014505e-05, "loss": 1.0491, "step": 42025 }, { "epoch": 0.04, "learning_rate": 4.978341149913924e-05, "loss": 1.1239, "step": 42026 }, { "epoch": 0.04, "learning_rate": 4.978340118201933e-05, "loss": 1.2932, "step": 42027 }, { "epoch": 0.04, "learning_rate": 4.9783390864654764e-05, "loss": 1.0752, "step": 42028 }, { "epoch": 0.04, "learning_rate": 4.978338054704554e-05, "loss": 0.9682, "step": 42029 }, { "epoch": 0.04, "learning_rate": 4.978337022919168e-05, "loss": 0.6734, "step": 42030 }, { "epoch": 0.04, "learning_rate": 4.9783359911093164e-05, "loss": 0.9322, "step": 42031 }, { "epoch": 0.04, "learning_rate": 4.9783349592749986e-05, "loss": 0.8895, "step": 42032 }, { "epoch": 0.04, "learning_rate": 4.9783339274162164e-05, "loss": 1.4731, "step": 42033 }, { "epoch": 0.04, "learning_rate": 4.97833289553297e-05, "loss": 1.0678, "step": 42034 }, { "epoch": 0.04, "learning_rate": 4.978331863625258e-05, "loss": 1.0926, "step": 42035 }, { "epoch": 0.04, "learning_rate": 4.978330831693081e-05, "loss": 0.8991, "step": 42036 }, { "epoch": 0.04, "learning_rate": 4.978329799736439e-05, "loss": 1.2007, "step": 42037 }, { "epoch": 0.04, "learning_rate": 4.9783287677553316e-05, "loss": 1.1432, "step": 42038 }, { "epoch": 0.04, "learning_rate": 4.9783277357497586e-05, "loss": 0.8756, "step": 42039 }, { "epoch": 0.04, "learning_rate": 4.978326703719721e-05, "loss": 1.0346, "step": 42040 }, { "epoch": 0.04, "learning_rate": 4.9783256716652196e-05, "loss": 1.1661, "step": 42041 }, { "epoch": 0.04, "learning_rate": 4.978324639586252e-05, "loss": 0.8923, "step": 42042 }, { "epoch": 0.04, "learning_rate": 4.97832360748282e-05, "loss": 1.0928, "step": 42043 }, { "epoch": 0.04, "learning_rate": 4.9783225753549224e-05, "loss": 1.483, "step": 42044 }, { "epoch": 0.04, "learning_rate": 4.9783215432025606e-05, "loss": 0.9144, "step": 42045 }, { "epoch": 0.04, "learning_rate": 4.978320511025734e-05, "loss": 0.7201, "step": 42046 }, { "epoch": 0.04, "learning_rate": 4.9783194788244414e-05, "loss": 1.0344, "step": 42047 }, { "epoch": 0.04, "learning_rate": 4.9783184465986845e-05, "loss": 1.0302, "step": 42048 }, { "epoch": 0.04, "learning_rate": 4.978317414348462e-05, "loss": 0.6028, "step": 42049 }, { "epoch": 0.04, "learning_rate": 4.978316382073776e-05, "loss": 1.3511, "step": 42050 }, { "epoch": 0.04, "learning_rate": 4.978315349774624e-05, "loss": 1.318, "step": 42051 }, { "epoch": 0.04, "learning_rate": 4.978314317451007e-05, "loss": 0.8695, "step": 42052 }, { "epoch": 0.04, "learning_rate": 4.978313285102926e-05, "loss": 0.9593, "step": 42053 }, { "epoch": 0.04, "learning_rate": 4.9783122527303795e-05, "loss": 0.8629, "step": 42054 }, { "epoch": 0.04, "learning_rate": 4.978311220333368e-05, "loss": 1.0972, "step": 42055 }, { "epoch": 0.04, "learning_rate": 4.9783101879118925e-05, "loss": 1.1287, "step": 42056 }, { "epoch": 0.04, "learning_rate": 4.978309155465952e-05, "loss": 0.93, "step": 42057 }, { "epoch": 0.04, "learning_rate": 4.978308122995546e-05, "loss": 1.0474, "step": 42058 }, { "epoch": 0.04, "learning_rate": 4.9783070905006754e-05, "loss": 0.8809, "step": 42059 }, { "epoch": 0.04, "learning_rate": 4.9783060579813404e-05, "loss": 0.4979, "step": 42060 }, { "epoch": 0.04, "learning_rate": 4.97830502543754e-05, "loss": 1.2541, "step": 42061 }, { "epoch": 0.04, "learning_rate": 4.978303992869275e-05, "loss": 1.0872, "step": 42062 }, { "epoch": 0.04, "learning_rate": 4.978302960276546e-05, "loss": 1.1883, "step": 42063 }, { "epoch": 0.04, "learning_rate": 4.978301927659351e-05, "loss": 0.733, "step": 42064 }, { "epoch": 0.04, "learning_rate": 4.9783008950176925e-05, "loss": 0.2642, "step": 42065 }, { "epoch": 0.04, "learning_rate": 4.9782998623515686e-05, "loss": 0.5264, "step": 42066 }, { "epoch": 0.04, "learning_rate": 4.97829882966098e-05, "loss": 1.1882, "step": 42067 }, { "epoch": 0.04, "learning_rate": 4.9782977969459266e-05, "loss": 0.9322, "step": 42068 }, { "epoch": 0.04, "learning_rate": 4.978296764206408e-05, "loss": 1.1375, "step": 42069 }, { "epoch": 0.04, "learning_rate": 4.978295731442426e-05, "loss": 1.2653, "step": 42070 }, { "epoch": 0.04, "learning_rate": 4.978294698653978e-05, "loss": 0.9704, "step": 42071 }, { "epoch": 0.04, "learning_rate": 4.978293665841066e-05, "loss": 0.911, "step": 42072 }, { "epoch": 0.04, "learning_rate": 4.978292633003689e-05, "loss": 0.6503, "step": 42073 }, { "epoch": 0.04, "learning_rate": 4.978291600141848e-05, "loss": 0.915, "step": 42074 }, { "epoch": 0.04, "learning_rate": 4.9782905672555417e-05, "loss": 1.321, "step": 42075 }, { "epoch": 0.04, "learning_rate": 4.978289534344771e-05, "loss": 1.3665, "step": 42076 }, { "epoch": 0.04, "learning_rate": 4.978288501409535e-05, "loss": 1.1846, "step": 42077 }, { "epoch": 0.04, "learning_rate": 4.978287468449835e-05, "loss": 0.9358, "step": 42078 }, { "epoch": 0.04, "learning_rate": 4.978286435465671e-05, "loss": 1.1238, "step": 42079 }, { "epoch": 0.04, "learning_rate": 4.9782854024570415e-05, "loss": 1.2075, "step": 42080 }, { "epoch": 0.04, "learning_rate": 4.978284369423948e-05, "loss": 0.8178, "step": 42081 }, { "epoch": 0.04, "learning_rate": 4.9782833363663896e-05, "loss": 0.8652, "step": 42082 }, { "epoch": 0.04, "learning_rate": 4.978282303284366e-05, "loss": 0.767, "step": 42083 }, { "epoch": 0.04, "learning_rate": 4.978281270177879e-05, "loss": 0.863, "step": 42084 }, { "epoch": 0.04, "learning_rate": 4.978280237046927e-05, "loss": 0.8877, "step": 42085 }, { "epoch": 0.04, "learning_rate": 4.9782792038915105e-05, "loss": 0.9541, "step": 42086 }, { "epoch": 0.04, "learning_rate": 4.978278170711629e-05, "loss": 0.9435, "step": 42087 }, { "epoch": 0.04, "learning_rate": 4.978277137507283e-05, "loss": 1.0275, "step": 42088 }, { "epoch": 0.04, "learning_rate": 4.978276104278474e-05, "loss": 1.0148, "step": 42089 }, { "epoch": 0.04, "learning_rate": 4.978275071025199e-05, "loss": 0.9176, "step": 42090 }, { "epoch": 0.04, "learning_rate": 4.97827403774746e-05, "loss": 1.0626, "step": 42091 }, { "epoch": 0.04, "learning_rate": 4.978273004445256e-05, "loss": 0.8799, "step": 42092 }, { "epoch": 0.04, "learning_rate": 4.978271971118588e-05, "loss": 1.2361, "step": 42093 }, { "epoch": 0.04, "learning_rate": 4.9782709377674554e-05, "loss": 0.993, "step": 42094 }, { "epoch": 0.04, "learning_rate": 4.978269904391859e-05, "loss": 1.0887, "step": 42095 }, { "epoch": 0.04, "learning_rate": 4.9782688709917965e-05, "loss": 1.0717, "step": 42096 }, { "epoch": 0.04, "learning_rate": 4.978267837567271e-05, "loss": 0.9313, "step": 42097 }, { "epoch": 0.04, "learning_rate": 4.978266804118281e-05, "loss": 1.0254, "step": 42098 }, { "epoch": 0.04, "learning_rate": 4.978265770644826e-05, "loss": 1.3636, "step": 42099 }, { "epoch": 0.04, "learning_rate": 4.978264737146907e-05, "loss": 0.8404, "step": 42100 }, { "epoch": 0.04, "learning_rate": 4.978263703624524e-05, "loss": 0.462, "step": 42101 }, { "epoch": 0.04, "learning_rate": 4.9782626700776755e-05, "loss": 0.8949, "step": 42102 }, { "epoch": 0.04, "learning_rate": 4.978261636506364e-05, "loss": 0.7649, "step": 42103 }, { "epoch": 0.04, "learning_rate": 4.978260602910587e-05, "loss": 1.0298, "step": 42104 }, { "epoch": 0.04, "learning_rate": 4.9782595692903464e-05, "loss": 0.8652, "step": 42105 }, { "epoch": 0.04, "learning_rate": 4.9782585356456413e-05, "loss": 0.8835, "step": 42106 }, { "epoch": 0.04, "learning_rate": 4.978257501976472e-05, "loss": 0.9997, "step": 42107 }, { "epoch": 0.04, "learning_rate": 4.978256468282838e-05, "loss": 0.8116, "step": 42108 }, { "epoch": 0.04, "learning_rate": 4.97825543456474e-05, "loss": 0.9314, "step": 42109 }, { "epoch": 0.04, "learning_rate": 4.9782544008221776e-05, "loss": 0.8195, "step": 42110 }, { "epoch": 0.04, "learning_rate": 4.978253367055151e-05, "loss": 1.2381, "step": 42111 }, { "epoch": 0.04, "learning_rate": 4.97825233326366e-05, "loss": 1.1876, "step": 42112 }, { "epoch": 0.04, "learning_rate": 4.978251299447705e-05, "loss": 1.8796, "step": 42113 }, { "epoch": 0.04, "learning_rate": 4.978250265607285e-05, "loss": 0.8022, "step": 42114 }, { "epoch": 0.04, "learning_rate": 4.978249231742401e-05, "loss": 1.3194, "step": 42115 }, { "epoch": 0.04, "learning_rate": 4.978248197853053e-05, "loss": 1.0654, "step": 42116 }, { "epoch": 0.04, "learning_rate": 4.978247163939241e-05, "loss": 0.9666, "step": 42117 }, { "epoch": 0.04, "learning_rate": 4.9782461300009644e-05, "loss": 1.0518, "step": 42118 }, { "epoch": 0.04, "learning_rate": 4.978245096038224e-05, "loss": 0.8545, "step": 42119 }, { "epoch": 0.04, "learning_rate": 4.978244062051019e-05, "loss": 1.1665, "step": 42120 }, { "epoch": 0.04, "learning_rate": 4.97824302803935e-05, "loss": 0.9265, "step": 42121 }, { "epoch": 0.04, "learning_rate": 4.978241994003216e-05, "loss": 1.0472, "step": 42122 }, { "epoch": 0.04, "learning_rate": 4.978240959942619e-05, "loss": 1.0842, "step": 42123 }, { "epoch": 0.04, "learning_rate": 4.978239925857559e-05, "loss": 1.1504, "step": 42124 }, { "epoch": 0.04, "learning_rate": 4.9782388917480325e-05, "loss": 1.0021, "step": 42125 }, { "epoch": 0.04, "learning_rate": 4.978237857614043e-05, "loss": 1.1625, "step": 42126 }, { "epoch": 0.04, "learning_rate": 4.978236823455589e-05, "loss": 1.1092, "step": 42127 }, { "epoch": 0.04, "learning_rate": 4.9782357892726704e-05, "loss": 0.8442, "step": 42128 }, { "epoch": 0.04, "learning_rate": 4.978234755065289e-05, "loss": 1.0401, "step": 42129 }, { "epoch": 0.04, "learning_rate": 4.978233720833443e-05, "loss": 0.7426, "step": 42130 }, { "epoch": 0.04, "learning_rate": 4.9782326865771326e-05, "loss": 1.0493, "step": 42131 }, { "epoch": 0.04, "learning_rate": 4.978231652296358e-05, "loss": 0.0941, "step": 42132 }, { "epoch": 0.04, "learning_rate": 4.9782306179911196e-05, "loss": 0.1564, "step": 42133 }, { "epoch": 0.04, "learning_rate": 4.978229583661418e-05, "loss": 0.977, "step": 42134 }, { "epoch": 0.04, "learning_rate": 4.9782285493072514e-05, "loss": 1.0631, "step": 42135 }, { "epoch": 0.04, "learning_rate": 4.978227514928621e-05, "loss": 2.7433, "step": 42136 }, { "epoch": 0.04, "learning_rate": 4.978226480525526e-05, "loss": 2.3106, "step": 42137 }, { "epoch": 0.04, "learning_rate": 4.978225446097967e-05, "loss": 1.1928, "step": 42138 }, { "epoch": 0.04, "learning_rate": 4.978224411645945e-05, "loss": 0.5763, "step": 42139 }, { "epoch": 0.04, "learning_rate": 4.978223377169459e-05, "loss": 0.8098, "step": 42140 }, { "epoch": 0.04, "learning_rate": 4.9782223426685084e-05, "loss": 1.054, "step": 42141 }, { "epoch": 0.04, "learning_rate": 4.978221308143094e-05, "loss": 1.1233, "step": 42142 }, { "epoch": 0.04, "learning_rate": 4.9782202735932147e-05, "loss": 1.3761, "step": 42143 }, { "epoch": 0.04, "learning_rate": 4.978219239018873e-05, "loss": 0.7571, "step": 42144 }, { "epoch": 0.04, "learning_rate": 4.978218204420066e-05, "loss": 0.6333, "step": 42145 }, { "epoch": 0.04, "learning_rate": 4.978217169796796e-05, "loss": 0.7598, "step": 42146 }, { "epoch": 0.04, "learning_rate": 4.978216135149062e-05, "loss": 0.7669, "step": 42147 }, { "epoch": 0.04, "learning_rate": 4.978215100476864e-05, "loss": 1.1495, "step": 42148 }, { "epoch": 0.04, "learning_rate": 4.978214065780202e-05, "loss": 1.0773, "step": 42149 }, { "epoch": 0.04, "learning_rate": 4.978213031059076e-05, "loss": 1.3469, "step": 42150 }, { "epoch": 0.04, "learning_rate": 4.978211996313486e-05, "loss": 0.734, "step": 42151 }, { "epoch": 0.04, "learning_rate": 4.978210961543433e-05, "loss": 0.5875, "step": 42152 }, { "epoch": 0.04, "learning_rate": 4.978209926748916e-05, "loss": 0.3551, "step": 42153 }, { "epoch": 0.04, "learning_rate": 4.978208891929934e-05, "loss": 1.0432, "step": 42154 }, { "epoch": 0.04, "learning_rate": 4.9782078570864894e-05, "loss": 0.9234, "step": 42155 }, { "epoch": 0.04, "learning_rate": 4.978206822218581e-05, "loss": 1.4293, "step": 42156 }, { "epoch": 0.04, "learning_rate": 4.9782057873262076e-05, "loss": 1.0715, "step": 42157 }, { "epoch": 0.04, "learning_rate": 4.978204752409371e-05, "loss": 1.1568, "step": 42158 }, { "epoch": 0.04, "learning_rate": 4.9782037174680704e-05, "loss": 0.8282, "step": 42159 }, { "epoch": 0.04, "learning_rate": 4.9782026825023064e-05, "loss": 1.0472, "step": 42160 }, { "epoch": 0.04, "learning_rate": 4.978201647512079e-05, "loss": 0.6179, "step": 42161 }, { "epoch": 0.04, "learning_rate": 4.978200612497387e-05, "loss": 0.7875, "step": 42162 }, { "epoch": 0.04, "learning_rate": 4.978199577458232e-05, "loss": 1.1303, "step": 42163 }, { "epoch": 0.04, "learning_rate": 4.9781985423946134e-05, "loss": 1.0118, "step": 42164 }, { "epoch": 0.04, "learning_rate": 4.97819750730653e-05, "loss": 0.9953, "step": 42165 }, { "epoch": 0.04, "learning_rate": 4.9781964721939835e-05, "loss": 0.9063, "step": 42166 }, { "epoch": 0.04, "learning_rate": 4.978195437056973e-05, "loss": 1.0479, "step": 42167 }, { "epoch": 0.04, "learning_rate": 4.978194401895499e-05, "loss": 1.1918, "step": 42168 }, { "epoch": 0.04, "learning_rate": 4.978193366709561e-05, "loss": 1.129, "step": 42169 }, { "epoch": 0.04, "learning_rate": 4.9781923314991605e-05, "loss": 0.9984, "step": 42170 }, { "epoch": 0.04, "learning_rate": 4.9781912962642954e-05, "loss": 0.9795, "step": 42171 }, { "epoch": 0.04, "learning_rate": 4.978190261004966e-05, "loss": 1.0607, "step": 42172 }, { "epoch": 0.04, "learning_rate": 4.9781892257211736e-05, "loss": 0.7895, "step": 42173 }, { "epoch": 0.04, "learning_rate": 4.978188190412918e-05, "loss": 1.0323, "step": 42174 }, { "epoch": 0.04, "learning_rate": 4.9781871550801984e-05, "loss": 0.799, "step": 42175 }, { "epoch": 0.04, "learning_rate": 4.978186119723015e-05, "loss": 1.2327, "step": 42176 }, { "epoch": 0.04, "learning_rate": 4.978185084341368e-05, "loss": 0.9467, "step": 42177 }, { "epoch": 0.04, "learning_rate": 4.9781840489352585e-05, "loss": 0.9898, "step": 42178 }, { "epoch": 0.04, "learning_rate": 4.9781830135046834e-05, "loss": 1.4322, "step": 42179 }, { "epoch": 0.04, "learning_rate": 4.9781819780496466e-05, "loss": 1.0848, "step": 42180 }, { "epoch": 0.04, "learning_rate": 4.9781809425701456e-05, "loss": 0.8745, "step": 42181 }, { "epoch": 0.04, "learning_rate": 4.978179907066181e-05, "loss": 0.9968, "step": 42182 }, { "epoch": 0.04, "learning_rate": 4.978178871537752e-05, "loss": 1.096, "step": 42183 }, { "epoch": 0.04, "learning_rate": 4.97817783598486e-05, "loss": 0.9249, "step": 42184 }, { "epoch": 0.04, "learning_rate": 4.9781768004075046e-05, "loss": 0.8761, "step": 42185 }, { "epoch": 0.04, "learning_rate": 4.978175764805686e-05, "loss": 0.9977, "step": 42186 }, { "epoch": 0.04, "learning_rate": 4.9781747291794034e-05, "loss": 1.0472, "step": 42187 }, { "epoch": 0.04, "learning_rate": 4.978173693528658e-05, "loss": 1.1072, "step": 42188 }, { "epoch": 0.04, "learning_rate": 4.978172657853448e-05, "loss": 1.1038, "step": 42189 }, { "epoch": 0.04, "learning_rate": 4.9781716221537756e-05, "loss": 1.2923, "step": 42190 }, { "epoch": 0.04, "learning_rate": 4.978170586429639e-05, "loss": 1.0051, "step": 42191 }, { "epoch": 0.04, "learning_rate": 4.97816955068104e-05, "loss": 1.2187, "step": 42192 }, { "epoch": 0.04, "learning_rate": 4.978168514907977e-05, "loss": 0.9754, "step": 42193 }, { "epoch": 0.04, "learning_rate": 4.97816747911045e-05, "loss": 0.9294, "step": 42194 }, { "epoch": 0.04, "learning_rate": 4.9781664432884603e-05, "loss": 1.0848, "step": 42195 }, { "epoch": 0.04, "learning_rate": 4.9781654074420056e-05, "loss": 1.0842, "step": 42196 }, { "epoch": 0.04, "learning_rate": 4.978164371571089e-05, "loss": 0.668, "step": 42197 }, { "epoch": 0.04, "learning_rate": 4.978163335675709e-05, "loss": 1.0275, "step": 42198 }, { "epoch": 0.04, "learning_rate": 4.9781622997558655e-05, "loss": 0.941, "step": 42199 }, { "epoch": 0.04, "learning_rate": 4.978161263811558e-05, "loss": 1.2107, "step": 42200 }, { "epoch": 0.04, "learning_rate": 4.978160227842788e-05, "loss": 1.0033, "step": 42201 }, { "epoch": 0.04, "learning_rate": 4.978159191849554e-05, "loss": 0.8617, "step": 42202 }, { "epoch": 0.04, "learning_rate": 4.978158155831857e-05, "loss": 0.7635, "step": 42203 }, { "epoch": 0.04, "learning_rate": 4.978157119789697e-05, "loss": 1.1654, "step": 42204 }, { "epoch": 0.04, "learning_rate": 4.978156083723073e-05, "loss": 0.9477, "step": 42205 }, { "epoch": 0.04, "learning_rate": 4.978155047631986e-05, "loss": 1.056, "step": 42206 }, { "epoch": 0.04, "learning_rate": 4.978154011516436e-05, "loss": 1.0317, "step": 42207 }, { "epoch": 0.04, "learning_rate": 4.978152975376422e-05, "loss": 1.2161, "step": 42208 }, { "epoch": 0.04, "learning_rate": 4.9781519392119445e-05, "loss": 0.7437, "step": 42209 }, { "epoch": 0.04, "learning_rate": 4.9781509030230046e-05, "loss": 1.1409, "step": 42210 }, { "epoch": 0.04, "learning_rate": 4.9781498668096016e-05, "loss": 0.8306, "step": 42211 }, { "epoch": 0.04, "learning_rate": 4.978148830571735e-05, "loss": 1.2527, "step": 42212 }, { "epoch": 0.04, "learning_rate": 4.978147794309405e-05, "loss": 1.0183, "step": 42213 }, { "epoch": 0.04, "learning_rate": 4.9781467580226114e-05, "loss": 0.7242, "step": 42214 }, { "epoch": 0.04, "learning_rate": 4.9781457217113545e-05, "loss": 0.9856, "step": 42215 }, { "epoch": 0.04, "learning_rate": 4.978144685375635e-05, "loss": 0.9607, "step": 42216 }, { "epoch": 0.04, "learning_rate": 4.978143649015452e-05, "loss": 0.8739, "step": 42217 }, { "epoch": 0.04, "learning_rate": 4.9781426126308064e-05, "loss": 1.0372, "step": 42218 }, { "epoch": 0.04, "learning_rate": 4.9781415762216975e-05, "loss": 0.8756, "step": 42219 }, { "epoch": 0.04, "learning_rate": 4.9781405397881256e-05, "loss": 0.9345, "step": 42220 }, { "epoch": 0.04, "learning_rate": 4.9781395033300894e-05, "loss": 0.4939, "step": 42221 }, { "epoch": 0.04, "learning_rate": 4.978138466847591e-05, "loss": 0.758, "step": 42222 }, { "epoch": 0.04, "learning_rate": 4.978137430340629e-05, "loss": 0.9833, "step": 42223 }, { "epoch": 0.04, "learning_rate": 4.978136393809204e-05, "loss": 1.0438, "step": 42224 }, { "epoch": 0.04, "learning_rate": 4.978135357253316e-05, "loss": 1.3558, "step": 42225 }, { "epoch": 0.04, "learning_rate": 4.978134320672965e-05, "loss": 1.118, "step": 42226 }, { "epoch": 0.04, "learning_rate": 4.9781332840681506e-05, "loss": 1.0825, "step": 42227 }, { "epoch": 0.04, "learning_rate": 4.9781322474388734e-05, "loss": 1.1163, "step": 42228 }, { "epoch": 0.04, "learning_rate": 4.978131210785133e-05, "loss": 1.1942, "step": 42229 }, { "epoch": 0.04, "learning_rate": 4.9781301741069295e-05, "loss": 1.0598, "step": 42230 }, { "epoch": 0.04, "learning_rate": 4.978129137404263e-05, "loss": 1.0952, "step": 42231 }, { "epoch": 0.04, "learning_rate": 4.978128100677133e-05, "loss": 0.9357, "step": 42232 }, { "epoch": 0.04, "learning_rate": 4.978127063925541e-05, "loss": 1.0447, "step": 42233 }, { "epoch": 0.04, "learning_rate": 4.978126027149485e-05, "loss": 1.3553, "step": 42234 }, { "epoch": 0.04, "learning_rate": 4.978124990348966e-05, "loss": 0.9933, "step": 42235 }, { "epoch": 0.04, "learning_rate": 4.9781239535239845e-05, "loss": 1.3331, "step": 42236 }, { "epoch": 0.04, "learning_rate": 4.9781229166745405e-05, "loss": 1.2819, "step": 42237 }, { "epoch": 0.04, "learning_rate": 4.978121879800632e-05, "loss": 1.2215, "step": 42238 }, { "epoch": 0.04, "learning_rate": 4.978120842902262e-05, "loss": 1.1852, "step": 42239 }, { "epoch": 0.04, "learning_rate": 4.978119805979428e-05, "loss": 1.0535, "step": 42240 }, { "epoch": 0.04, "learning_rate": 4.9781187690321316e-05, "loss": 1.2306, "step": 42241 }, { "epoch": 0.04, "learning_rate": 4.978117732060372e-05, "loss": 1.1671, "step": 42242 }, { "epoch": 0.04, "learning_rate": 4.97811669506415e-05, "loss": 1.173, "step": 42243 }, { "epoch": 0.04, "learning_rate": 4.978115658043464e-05, "loss": 0.6723, "step": 42244 }, { "epoch": 0.04, "learning_rate": 4.9781146209983157e-05, "loss": 1.1246, "step": 42245 }, { "epoch": 0.04, "learning_rate": 4.9781135839287054e-05, "loss": 1.1996, "step": 42246 }, { "epoch": 0.04, "learning_rate": 4.978112546834631e-05, "loss": 0.7506, "step": 42247 }, { "epoch": 0.04, "learning_rate": 4.978111509716094e-05, "loss": 0.9152, "step": 42248 }, { "epoch": 0.04, "learning_rate": 4.9781104725730946e-05, "loss": 0.668, "step": 42249 }, { "epoch": 0.04, "learning_rate": 4.978109435405631e-05, "loss": 0.8862, "step": 42250 }, { "epoch": 0.04, "learning_rate": 4.978108398213706e-05, "loss": 1.307, "step": 42251 }, { "epoch": 0.04, "learning_rate": 4.9781073609973175e-05, "loss": 0.8675, "step": 42252 }, { "epoch": 0.04, "learning_rate": 4.978106323756466e-05, "loss": 0.6259, "step": 42253 }, { "epoch": 0.04, "learning_rate": 4.978105286491153e-05, "loss": 0.2454, "step": 42254 }, { "epoch": 0.04, "learning_rate": 4.978104249201375e-05, "loss": 0.501, "step": 42255 }, { "epoch": 0.04, "learning_rate": 4.9781032118871354e-05, "loss": 0.2228, "step": 42256 }, { "epoch": 0.04, "learning_rate": 4.9781021745484336e-05, "loss": 0.1488, "step": 42257 }, { "epoch": 0.04, "learning_rate": 4.978101137185268e-05, "loss": 0.1646, "step": 42258 }, { "epoch": 0.04, "learning_rate": 4.9781000997976404e-05, "loss": 0.2109, "step": 42259 }, { "epoch": 0.04, "learning_rate": 4.9780990623855496e-05, "loss": 1.3204, "step": 42260 }, { "epoch": 0.04, "learning_rate": 4.978098024948996e-05, "loss": 0.9027, "step": 42261 }, { "epoch": 0.04, "learning_rate": 4.97809698748798e-05, "loss": 0.494, "step": 42262 }, { "epoch": 0.04, "learning_rate": 4.978095950002501e-05, "loss": 0.934, "step": 42263 }, { "epoch": 0.04, "learning_rate": 4.978094912492559e-05, "loss": 1.1501, "step": 42264 }, { "epoch": 0.04, "learning_rate": 4.978093874958154e-05, "loss": 0.7095, "step": 42265 }, { "epoch": 0.04, "learning_rate": 4.9780928373992876e-05, "loss": 0.9147, "step": 42266 }, { "epoch": 0.04, "learning_rate": 4.978091799815958e-05, "loss": 1.0432, "step": 42267 }, { "epoch": 0.04, "learning_rate": 4.978090762208165e-05, "loss": 1.1511, "step": 42268 }, { "epoch": 0.04, "learning_rate": 4.97808972457591e-05, "loss": 0.8488, "step": 42269 }, { "epoch": 0.04, "learning_rate": 4.9780886869191924e-05, "loss": 1.1557, "step": 42270 }, { "epoch": 0.04, "learning_rate": 4.978087649238012e-05, "loss": 1.1253, "step": 42271 }, { "epoch": 0.04, "learning_rate": 4.978086611532369e-05, "loss": 1.3344, "step": 42272 }, { "epoch": 0.04, "learning_rate": 4.978085573802263e-05, "loss": 1.4308, "step": 42273 }, { "epoch": 0.04, "learning_rate": 4.978084536047695e-05, "loss": 1.1964, "step": 42274 }, { "epoch": 0.04, "learning_rate": 4.978083498268664e-05, "loss": 1.0993, "step": 42275 }, { "epoch": 0.04, "learning_rate": 4.9780824604651707e-05, "loss": 1.1732, "step": 42276 }, { "epoch": 0.04, "learning_rate": 4.978081422637214e-05, "loss": 0.9552, "step": 42277 }, { "epoch": 0.04, "learning_rate": 4.978080384784795e-05, "loss": 0.9764, "step": 42278 }, { "epoch": 0.04, "learning_rate": 4.9780793469079145e-05, "loss": 1.0134, "step": 42279 }, { "epoch": 0.04, "learning_rate": 4.978078309006571e-05, "loss": 0.7593, "step": 42280 }, { "epoch": 0.04, "learning_rate": 4.9780772710807647e-05, "loss": 1.3335, "step": 42281 }, { "epoch": 0.04, "learning_rate": 4.9780762331304956e-05, "loss": 1.2666, "step": 42282 }, { "epoch": 0.04, "learning_rate": 4.978075195155764e-05, "loss": 0.9411, "step": 42283 }, { "epoch": 0.04, "learning_rate": 4.9780741571565705e-05, "loss": 1.2619, "step": 42284 }, { "epoch": 0.04, "learning_rate": 4.978073119132914e-05, "loss": 1.1071, "step": 42285 }, { "epoch": 0.04, "learning_rate": 4.978072081084795e-05, "loss": 0.9504, "step": 42286 }, { "epoch": 0.04, "learning_rate": 4.9780710430122135e-05, "loss": 1.0245, "step": 42287 }, { "epoch": 0.04, "learning_rate": 4.97807000491517e-05, "loss": 0.6571, "step": 42288 }, { "epoch": 0.04, "learning_rate": 4.978068966793663e-05, "loss": 0.9207, "step": 42289 }, { "epoch": 0.04, "learning_rate": 4.978067928647694e-05, "loss": 0.8914, "step": 42290 }, { "epoch": 0.04, "learning_rate": 4.978066890477263e-05, "loss": 0.7438, "step": 42291 }, { "epoch": 0.04, "learning_rate": 4.9780658522823695e-05, "loss": 0.7473, "step": 42292 }, { "epoch": 0.04, "learning_rate": 4.9780648140630136e-05, "loss": 1.0302, "step": 42293 }, { "epoch": 0.04, "learning_rate": 4.978063775819195e-05, "loss": 1.2464, "step": 42294 }, { "epoch": 0.04, "learning_rate": 4.978062737550914e-05, "loss": 1.1388, "step": 42295 }, { "epoch": 0.04, "learning_rate": 4.978061699258171e-05, "loss": 0.8013, "step": 42296 }, { "epoch": 0.04, "learning_rate": 4.978060660940965e-05, "loss": 0.8585, "step": 42297 }, { "epoch": 0.04, "learning_rate": 4.978059622599297e-05, "loss": 1.012, "step": 42298 }, { "epoch": 0.04, "learning_rate": 4.9780585842331665e-05, "loss": 1.2141, "step": 42299 }, { "epoch": 0.04, "learning_rate": 4.978057545842574e-05, "loss": 1.5215, "step": 42300 }, { "epoch": 0.04, "learning_rate": 4.978056507427519e-05, "loss": 1.2048, "step": 42301 }, { "epoch": 0.04, "learning_rate": 4.9780554689880014e-05, "loss": 1.1865, "step": 42302 }, { "epoch": 0.04, "learning_rate": 4.978054430524022e-05, "loss": 1.3256, "step": 42303 }, { "epoch": 0.04, "learning_rate": 4.97805339203558e-05, "loss": 1.0687, "step": 42304 }, { "epoch": 0.04, "learning_rate": 4.9780523535226755e-05, "loss": 1.1672, "step": 42305 }, { "epoch": 0.04, "learning_rate": 4.978051314985309e-05, "loss": 0.997, "step": 42306 }, { "epoch": 0.04, "learning_rate": 4.97805027642348e-05, "loss": 0.9931, "step": 42307 }, { "epoch": 0.04, "learning_rate": 4.978049237837189e-05, "loss": 1.0351, "step": 42308 }, { "epoch": 0.04, "learning_rate": 4.978048199226435e-05, "loss": 0.8258, "step": 42309 }, { "epoch": 0.04, "learning_rate": 4.9780471605912194e-05, "loss": 0.8226, "step": 42310 }, { "epoch": 0.04, "learning_rate": 4.978046121931542e-05, "loss": 0.924, "step": 42311 }, { "epoch": 0.04, "learning_rate": 4.9780450832474015e-05, "loss": 0.8768, "step": 42312 }, { "epoch": 0.04, "learning_rate": 4.9780440445387994e-05, "loss": 0.7169, "step": 42313 }, { "epoch": 0.04, "learning_rate": 4.978043005805735e-05, "loss": 0.6664, "step": 42314 }, { "epoch": 0.04, "learning_rate": 4.978041967048208e-05, "loss": 0.7727, "step": 42315 }, { "epoch": 0.04, "learning_rate": 4.9780409282662186e-05, "loss": 0.7693, "step": 42316 }, { "epoch": 0.04, "learning_rate": 4.978039889459768e-05, "loss": 0.8795, "step": 42317 }, { "epoch": 0.04, "learning_rate": 4.978038850628854e-05, "loss": 0.8911, "step": 42318 }, { "epoch": 0.04, "learning_rate": 4.9780378117734796e-05, "loss": 0.8364, "step": 42319 }, { "epoch": 0.04, "learning_rate": 4.9780367728936414e-05, "loss": 0.8524, "step": 42320 }, { "epoch": 0.04, "learning_rate": 4.978035733989342e-05, "loss": 0.8969, "step": 42321 }, { "epoch": 0.04, "learning_rate": 4.97803469506058e-05, "loss": 0.8141, "step": 42322 }, { "epoch": 0.04, "learning_rate": 4.978033656107356e-05, "loss": 0.8609, "step": 42323 }, { "epoch": 0.04, "learning_rate": 4.97803261712967e-05, "loss": 0.7413, "step": 42324 }, { "epoch": 0.04, "learning_rate": 4.9780315781275214e-05, "loss": 1.0643, "step": 42325 }, { "epoch": 0.04, "learning_rate": 4.978030539100912e-05, "loss": 0.9615, "step": 42326 }, { "epoch": 0.04, "learning_rate": 4.978029500049839e-05, "loss": 0.5471, "step": 42327 }, { "epoch": 0.04, "learning_rate": 4.978028460974304e-05, "loss": 0.3724, "step": 42328 }, { "epoch": 0.04, "learning_rate": 4.9780274218743085e-05, "loss": 0.4007, "step": 42329 }, { "epoch": 0.04, "learning_rate": 4.97802638274985e-05, "loss": 0.8229, "step": 42330 }, { "epoch": 0.04, "learning_rate": 4.9780253436009295e-05, "loss": 0.962, "step": 42331 }, { "epoch": 0.04, "learning_rate": 4.978024304427547e-05, "loss": 0.9301, "step": 42332 }, { "epoch": 0.04, "learning_rate": 4.978023265229702e-05, "loss": 0.9215, "step": 42333 }, { "epoch": 0.04, "learning_rate": 4.978022226007396e-05, "loss": 0.9535, "step": 42334 }, { "epoch": 0.04, "learning_rate": 4.978021186760627e-05, "loss": 0.9525, "step": 42335 }, { "epoch": 0.04, "learning_rate": 4.9780201474893966e-05, "loss": 0.9772, "step": 42336 }, { "epoch": 0.04, "learning_rate": 4.9780191081937045e-05, "loss": 0.7778, "step": 42337 }, { "epoch": 0.04, "learning_rate": 4.97801806887355e-05, "loss": 0.7662, "step": 42338 }, { "epoch": 0.04, "learning_rate": 4.978017029528933e-05, "loss": 0.9371, "step": 42339 }, { "epoch": 0.04, "learning_rate": 4.978015990159855e-05, "loss": 0.9747, "step": 42340 }, { "epoch": 0.04, "learning_rate": 4.978014950766315e-05, "loss": 0.8381, "step": 42341 }, { "epoch": 0.04, "learning_rate": 4.9780139113483124e-05, "loss": 0.907, "step": 42342 }, { "epoch": 0.04, "learning_rate": 4.9780128719058484e-05, "loss": 0.7512, "step": 42343 }, { "epoch": 0.04, "learning_rate": 4.978011832438922e-05, "loss": 0.9477, "step": 42344 }, { "epoch": 0.04, "learning_rate": 4.978010792947534e-05, "loss": 0.8559, "step": 42345 }, { "epoch": 0.04, "learning_rate": 4.9780097534316846e-05, "loss": 0.7288, "step": 42346 }, { "epoch": 0.04, "learning_rate": 4.9780087138913734e-05, "loss": 0.8001, "step": 42347 }, { "epoch": 0.04, "learning_rate": 4.9780076743266e-05, "loss": 0.7224, "step": 42348 }, { "epoch": 0.04, "learning_rate": 4.978006634737364e-05, "loss": 0.8517, "step": 42349 }, { "epoch": 0.04, "learning_rate": 4.978005595123667e-05, "loss": 0.7734, "step": 42350 }, { "epoch": 0.04, "learning_rate": 4.978004555485508e-05, "loss": 0.8655, "step": 42351 }, { "epoch": 0.04, "learning_rate": 4.9780035158228875e-05, "loss": 0.7675, "step": 42352 }, { "epoch": 0.04, "learning_rate": 4.9780024761358045e-05, "loss": 0.8169, "step": 42353 }, { "epoch": 0.04, "learning_rate": 4.97800143642426e-05, "loss": 0.7357, "step": 42354 }, { "epoch": 0.04, "learning_rate": 4.9780003966882535e-05, "loss": 0.6826, "step": 42355 }, { "epoch": 0.04, "learning_rate": 4.977999356927786e-05, "loss": 0.8026, "step": 42356 }, { "epoch": 0.04, "learning_rate": 4.977998317142857e-05, "loss": 0.8144, "step": 42357 }, { "epoch": 0.04, "learning_rate": 4.977997277333465e-05, "loss": 0.7461, "step": 42358 }, { "epoch": 0.04, "learning_rate": 4.977996237499611e-05, "loss": 0.7169, "step": 42359 }, { "epoch": 0.04, "learning_rate": 4.977995197641296e-05, "loss": 0.7902, "step": 42360 }, { "epoch": 0.04, "learning_rate": 4.97799415775852e-05, "loss": 0.6128, "step": 42361 }, { "epoch": 0.04, "learning_rate": 4.9779931178512816e-05, "loss": 0.699, "step": 42362 }, { "epoch": 0.04, "learning_rate": 4.9779920779195816e-05, "loss": 0.8855, "step": 42363 }, { "epoch": 0.04, "learning_rate": 4.97799103796342e-05, "loss": 0.7698, "step": 42364 }, { "epoch": 0.04, "learning_rate": 4.977989997982796e-05, "loss": 0.7403, "step": 42365 }, { "epoch": 0.04, "learning_rate": 4.977988957977711e-05, "loss": 0.6064, "step": 42366 }, { "epoch": 0.04, "learning_rate": 4.977987917948164e-05, "loss": 0.6977, "step": 42367 }, { "epoch": 0.04, "learning_rate": 4.9779868778941554e-05, "loss": 0.6798, "step": 42368 }, { "epoch": 0.04, "learning_rate": 4.977985837815685e-05, "loss": 0.7528, "step": 42369 }, { "epoch": 0.04, "learning_rate": 4.9779847977127534e-05, "loss": 0.7987, "step": 42370 }, { "epoch": 0.04, "learning_rate": 4.97798375758536e-05, "loss": 0.6291, "step": 42371 }, { "epoch": 0.04, "learning_rate": 4.9779827174335055e-05, "loss": 0.6334, "step": 42372 }, { "epoch": 0.04, "learning_rate": 4.9779816772571885e-05, "loss": 0.8246, "step": 42373 }, { "epoch": 0.04, "learning_rate": 4.977980637056411e-05, "loss": 0.7474, "step": 42374 }, { "epoch": 0.04, "learning_rate": 4.97797959683117e-05, "loss": 0.7047, "step": 42375 }, { "epoch": 0.04, "learning_rate": 4.9779785565814695e-05, "loss": 0.6767, "step": 42376 }, { "epoch": 0.04, "learning_rate": 4.977977516307306e-05, "loss": 0.6469, "step": 42377 }, { "epoch": 0.04, "learning_rate": 4.977976476008682e-05, "loss": 0.7519, "step": 42378 }, { "epoch": 0.04, "learning_rate": 4.9779754356855965e-05, "loss": 0.7376, "step": 42379 }, { "epoch": 0.04, "learning_rate": 4.9779743953380486e-05, "loss": 0.8834, "step": 42380 }, { "epoch": 0.04, "learning_rate": 4.97797335496604e-05, "loss": 0.8735, "step": 42381 }, { "epoch": 0.04, "learning_rate": 4.977972314569569e-05, "loss": 0.6263, "step": 42382 }, { "epoch": 0.04, "learning_rate": 4.9779712741486374e-05, "loss": 0.6094, "step": 42383 }, { "epoch": 0.04, "learning_rate": 4.977970233703243e-05, "loss": 0.73, "step": 42384 }, { "epoch": 0.04, "learning_rate": 4.9779691932333884e-05, "loss": 0.8373, "step": 42385 }, { "epoch": 0.04, "learning_rate": 4.977968152739072e-05, "loss": 0.8244, "step": 42386 }, { "epoch": 0.04, "learning_rate": 4.977967112220294e-05, "loss": 0.7097, "step": 42387 }, { "epoch": 0.04, "learning_rate": 4.9779660716770547e-05, "loss": 0.7408, "step": 42388 }, { "epoch": 0.04, "learning_rate": 4.977965031109354e-05, "loss": 0.812, "step": 42389 }, { "epoch": 0.04, "learning_rate": 4.9779639905171913e-05, "loss": 0.7681, "step": 42390 }, { "epoch": 0.04, "learning_rate": 4.9779629499005676e-05, "loss": 0.7709, "step": 42391 }, { "epoch": 0.04, "learning_rate": 4.977961909259483e-05, "loss": 0.7855, "step": 42392 }, { "epoch": 0.04, "learning_rate": 4.977960868593936e-05, "loss": 0.7664, "step": 42393 }, { "epoch": 0.04, "learning_rate": 4.9779598279039285e-05, "loss": 0.7554, "step": 42394 }, { "epoch": 0.04, "learning_rate": 4.9779587871894596e-05, "loss": 0.6837, "step": 42395 }, { "epoch": 0.04, "learning_rate": 4.977957746450529e-05, "loss": 0.7107, "step": 42396 }, { "epoch": 0.04, "learning_rate": 4.977956705687137e-05, "loss": 0.6984, "step": 42397 }, { "epoch": 0.04, "learning_rate": 4.977955664899283e-05, "loss": 0.6128, "step": 42398 }, { "epoch": 0.04, "learning_rate": 4.977954624086969e-05, "loss": 0.6036, "step": 42399 }, { "epoch": 0.04, "learning_rate": 4.977953583250193e-05, "loss": 0.7295, "step": 42400 }, { "epoch": 0.04, "learning_rate": 4.9779525423889556e-05, "loss": 0.843, "step": 42401 }, { "epoch": 0.04, "learning_rate": 4.977951501503257e-05, "loss": 0.948, "step": 42402 }, { "epoch": 0.04, "learning_rate": 4.977950460593097e-05, "loss": 0.8626, "step": 42403 }, { "epoch": 0.04, "learning_rate": 4.977949419658476e-05, "loss": 0.8198, "step": 42404 }, { "epoch": 0.04, "learning_rate": 4.977948378699394e-05, "loss": 0.7271, "step": 42405 }, { "epoch": 0.04, "learning_rate": 4.97794733771585e-05, "loss": 0.8088, "step": 42406 }, { "epoch": 0.04, "learning_rate": 4.977946296707845e-05, "loss": 0.6888, "step": 42407 }, { "epoch": 0.04, "learning_rate": 4.977945255675379e-05, "loss": 0.7196, "step": 42408 }, { "epoch": 0.04, "learning_rate": 4.977944214618452e-05, "loss": 0.6864, "step": 42409 }, { "epoch": 0.04, "learning_rate": 4.977943173537063e-05, "loss": 0.6979, "step": 42410 }, { "epoch": 0.04, "learning_rate": 4.977942132431213e-05, "loss": 0.6475, "step": 42411 }, { "epoch": 0.04, "learning_rate": 4.9779410913009015e-05, "loss": 0.7233, "step": 42412 }, { "epoch": 0.04, "learning_rate": 4.977940050146129e-05, "loss": 0.6978, "step": 42413 }, { "epoch": 0.04, "learning_rate": 4.977939008966897e-05, "loss": 0.6584, "step": 42414 }, { "epoch": 0.04, "learning_rate": 4.977937967763202e-05, "loss": 0.7303, "step": 42415 }, { "epoch": 0.04, "learning_rate": 4.977936926535046e-05, "loss": 0.7538, "step": 42416 }, { "epoch": 0.04, "learning_rate": 4.9779358852824286e-05, "loss": 0.7224, "step": 42417 }, { "epoch": 0.04, "learning_rate": 4.977934844005351e-05, "loss": 0.6326, "step": 42418 }, { "epoch": 0.04, "learning_rate": 4.977933802703812e-05, "loss": 0.555, "step": 42419 }, { "epoch": 0.04, "learning_rate": 4.977932761377811e-05, "loss": 0.5902, "step": 42420 }, { "epoch": 0.04, "learning_rate": 4.97793172002735e-05, "loss": 0.6287, "step": 42421 }, { "epoch": 0.04, "learning_rate": 4.9779306786524284e-05, "loss": 0.7819, "step": 42422 }, { "epoch": 0.04, "learning_rate": 4.9779296372530445e-05, "loss": 0.3804, "step": 42423 }, { "epoch": 0.04, "learning_rate": 4.9779285958292e-05, "loss": 0.1431, "step": 42424 }, { "epoch": 0.04, "learning_rate": 4.977927554380894e-05, "loss": 0.4922, "step": 42425 }, { "epoch": 0.04, "learning_rate": 4.977926512908127e-05, "loss": 1.2634, "step": 42426 }, { "epoch": 0.04, "learning_rate": 4.977925471410899e-05, "loss": 0.9916, "step": 42427 }, { "epoch": 0.04, "learning_rate": 4.97792442988921e-05, "loss": 0.9972, "step": 42428 }, { "epoch": 0.04, "learning_rate": 4.9779233883430607e-05, "loss": 0.9016, "step": 42429 }, { "epoch": 0.04, "learning_rate": 4.97792234677245e-05, "loss": 0.9282, "step": 42430 }, { "epoch": 0.04, "learning_rate": 4.977921305177378e-05, "loss": 0.8253, "step": 42431 }, { "epoch": 0.04, "learning_rate": 4.977920263557845e-05, "loss": 0.7594, "step": 42432 }, { "epoch": 0.04, "learning_rate": 4.977919221913851e-05, "loss": 0.8389, "step": 42433 }, { "epoch": 0.04, "learning_rate": 4.977918180245396e-05, "loss": 0.8694, "step": 42434 }, { "epoch": 0.04, "learning_rate": 4.97791713855248e-05, "loss": 0.9358, "step": 42435 }, { "epoch": 0.04, "learning_rate": 4.9779160968351033e-05, "loss": 0.8504, "step": 42436 }, { "epoch": 0.04, "learning_rate": 4.977915055093266e-05, "loss": 0.8017, "step": 42437 }, { "epoch": 0.04, "learning_rate": 4.977914013326967e-05, "loss": 0.8773, "step": 42438 }, { "epoch": 0.04, "learning_rate": 4.977912971536208e-05, "loss": 0.883, "step": 42439 }, { "epoch": 0.04, "learning_rate": 4.977911929720987e-05, "loss": 0.7706, "step": 42440 }, { "epoch": 0.04, "learning_rate": 4.9779108878813055e-05, "loss": 0.7314, "step": 42441 }, { "epoch": 0.04, "learning_rate": 4.9779098460171635e-05, "loss": 0.8234, "step": 42442 }, { "epoch": 0.04, "learning_rate": 4.9779088041285605e-05, "loss": 0.8235, "step": 42443 }, { "epoch": 0.04, "learning_rate": 4.9779077622154966e-05, "loss": 0.7899, "step": 42444 }, { "epoch": 0.04, "learning_rate": 4.977906720277972e-05, "loss": 2.2957, "step": 42445 }, { "epoch": 0.04, "learning_rate": 4.977905678315986e-05, "loss": 0.8561, "step": 42446 }, { "epoch": 0.04, "learning_rate": 4.977904636329539e-05, "loss": 0.7509, "step": 42447 }, { "epoch": 0.04, "learning_rate": 4.977903594318632e-05, "loss": 0.7854, "step": 42448 }, { "epoch": 0.04, "learning_rate": 4.977902552283263e-05, "loss": 0.8101, "step": 42449 }, { "epoch": 0.04, "learning_rate": 4.977901510223435e-05, "loss": 0.7689, "step": 42450 }, { "epoch": 0.04, "learning_rate": 4.977900468139145e-05, "loss": 0.8565, "step": 42451 }, { "epoch": 0.04, "learning_rate": 4.977899426030394e-05, "loss": 0.6956, "step": 42452 }, { "epoch": 0.04, "learning_rate": 4.977898383897183e-05, "loss": 0.7275, "step": 42453 }, { "epoch": 0.04, "learning_rate": 4.97789734173951e-05, "loss": 0.7919, "step": 42454 }, { "epoch": 0.04, "learning_rate": 4.9778962995573776e-05, "loss": 0.7937, "step": 42455 }, { "epoch": 0.04, "learning_rate": 4.977895257350784e-05, "loss": 0.9302, "step": 42456 }, { "epoch": 0.04, "learning_rate": 4.977894215119729e-05, "loss": 1.1141, "step": 42457 }, { "epoch": 0.04, "learning_rate": 4.977893172864214e-05, "loss": 1.1271, "step": 42458 }, { "epoch": 0.04, "learning_rate": 4.9778921305842384e-05, "loss": 0.9929, "step": 42459 }, { "epoch": 0.04, "learning_rate": 4.977891088279802e-05, "loss": 1.4557, "step": 42460 }, { "epoch": 0.04, "learning_rate": 4.977890045950905e-05, "loss": 0.8808, "step": 42461 }, { "epoch": 0.04, "learning_rate": 4.9778890035975466e-05, "loss": 1.073, "step": 42462 }, { "epoch": 0.04, "learning_rate": 4.9778879612197286e-05, "loss": 1.1586, "step": 42463 }, { "epoch": 0.04, "learning_rate": 4.977886918817449e-05, "loss": 0.9507, "step": 42464 }, { "epoch": 0.04, "learning_rate": 4.977885876390709e-05, "loss": 0.5346, "step": 42465 }, { "epoch": 0.04, "learning_rate": 4.977884833939508e-05, "loss": 0.9453, "step": 42466 }, { "epoch": 0.04, "learning_rate": 4.977883791463847e-05, "loss": 0.9235, "step": 42467 }, { "epoch": 0.04, "learning_rate": 4.977882748963725e-05, "loss": 1.0941, "step": 42468 }, { "epoch": 0.04, "learning_rate": 4.977881706439142e-05, "loss": 0.9492, "step": 42469 }, { "epoch": 0.04, "learning_rate": 4.9778806638900996e-05, "loss": 0.8758, "step": 42470 }, { "epoch": 0.04, "learning_rate": 4.977879621316596e-05, "loss": 0.4442, "step": 42471 }, { "epoch": 0.04, "learning_rate": 4.977878578718631e-05, "loss": 0.4587, "step": 42472 }, { "epoch": 0.04, "learning_rate": 4.977877536096207e-05, "loss": 0.363, "step": 42473 }, { "epoch": 0.04, "learning_rate": 4.977876493449322e-05, "loss": 0.3103, "step": 42474 }, { "epoch": 0.04, "learning_rate": 4.977875450777976e-05, "loss": 1.1234, "step": 42475 }, { "epoch": 0.04, "learning_rate": 4.9778744080821694e-05, "loss": 1.0275, "step": 42476 }, { "epoch": 0.04, "learning_rate": 4.977873365361902e-05, "loss": 0.8123, "step": 42477 }, { "epoch": 0.04, "learning_rate": 4.9778723226171744e-05, "loss": 0.6245, "step": 42478 }, { "epoch": 0.04, "learning_rate": 4.9778712798479865e-05, "loss": 1.034, "step": 42479 }, { "epoch": 0.04, "learning_rate": 4.9778702370543376e-05, "loss": 1.0642, "step": 42480 }, { "epoch": 0.04, "learning_rate": 4.977869194236229e-05, "loss": 1.2478, "step": 42481 }, { "epoch": 0.04, "learning_rate": 4.977868151393659e-05, "loss": 1.3122, "step": 42482 }, { "epoch": 0.04, "learning_rate": 4.977867108526629e-05, "loss": 0.7849, "step": 42483 }, { "epoch": 0.04, "learning_rate": 4.9778660656351386e-05, "loss": 0.6009, "step": 42484 }, { "epoch": 0.04, "learning_rate": 4.977865022719188e-05, "loss": 0.7753, "step": 42485 }, { "epoch": 0.04, "learning_rate": 4.9778639797787764e-05, "loss": 0.9536, "step": 42486 }, { "epoch": 0.04, "learning_rate": 4.977862936813905e-05, "loss": 0.9193, "step": 42487 }, { "epoch": 0.04, "learning_rate": 4.9778618938245725e-05, "loss": 0.8657, "step": 42488 }, { "epoch": 0.04, "learning_rate": 4.9778608508107804e-05, "loss": 1.2545, "step": 42489 }, { "epoch": 0.04, "learning_rate": 4.977859807772527e-05, "loss": 0.8049, "step": 42490 }, { "epoch": 0.04, "learning_rate": 4.977858764709814e-05, "loss": 0.9098, "step": 42491 }, { "epoch": 0.04, "learning_rate": 4.97785772162264e-05, "loss": 0.7425, "step": 42492 }, { "epoch": 0.04, "learning_rate": 4.977856678511006e-05, "loss": 0.7302, "step": 42493 }, { "epoch": 0.04, "learning_rate": 4.977855635374912e-05, "loss": 0.914, "step": 42494 }, { "epoch": 0.04, "learning_rate": 4.977854592214357e-05, "loss": 0.7134, "step": 42495 }, { "epoch": 0.04, "learning_rate": 4.977853549029341e-05, "loss": 0.6589, "step": 42496 }, { "epoch": 0.04, "learning_rate": 4.977852505819866e-05, "loss": 0.708, "step": 42497 }, { "epoch": 0.04, "learning_rate": 4.97785146258593e-05, "loss": 0.6512, "step": 42498 }, { "epoch": 0.04, "learning_rate": 4.9778504193275336e-05, "loss": 0.5682, "step": 42499 }, { "epoch": 0.04, "learning_rate": 4.977849376044678e-05, "loss": 0.6992, "step": 42500 }, { "epoch": 0.04, "eval_loss": 1.0901010036468506, "eval_runtime": 128.8823, "eval_samples_per_second": 10.746, "eval_steps_per_second": 5.377, "step": 42500 }, { "epoch": 0.04, "learning_rate": 4.977848332737361e-05, "loss": 0.85, "step": 42501 }, { "epoch": 0.04, "learning_rate": 4.977847289405584e-05, "loss": 0.7981, "step": 42502 }, { "epoch": 0.04, "learning_rate": 4.977846246049347e-05, "loss": 0.7442, "step": 42503 }, { "epoch": 0.04, "learning_rate": 4.97784520266865e-05, "loss": 0.9667, "step": 42504 }, { "epoch": 0.04, "learning_rate": 4.977844159263492e-05, "loss": 0.5704, "step": 42505 }, { "epoch": 0.04, "learning_rate": 4.977843115833873e-05, "loss": 0.9215, "step": 42506 }, { "epoch": 0.04, "learning_rate": 4.977842072379795e-05, "loss": 0.9168, "step": 42507 }, { "epoch": 0.04, "learning_rate": 4.9778410289012566e-05, "loss": 0.4812, "step": 42508 }, { "epoch": 0.04, "learning_rate": 4.977839985398258e-05, "loss": 0.6707, "step": 42509 }, { "epoch": 0.04, "learning_rate": 4.977838941870799e-05, "loss": 0.8011, "step": 42510 }, { "epoch": 0.04, "learning_rate": 4.977837898318881e-05, "loss": 0.8284, "step": 42511 }, { "epoch": 0.04, "learning_rate": 4.977836854742502e-05, "loss": 0.8703, "step": 42512 }, { "epoch": 0.04, "learning_rate": 4.977835811141662e-05, "loss": 0.7562, "step": 42513 }, { "epoch": 0.04, "learning_rate": 4.9778347675163626e-05, "loss": 0.8173, "step": 42514 }, { "epoch": 0.04, "learning_rate": 4.9778337238666026e-05, "loss": 0.6361, "step": 42515 }, { "epoch": 0.04, "learning_rate": 4.977832680192384e-05, "loss": 0.4738, "step": 42516 }, { "epoch": 0.04, "learning_rate": 4.977831636493703e-05, "loss": 0.5306, "step": 42517 }, { "epoch": 0.04, "learning_rate": 4.977830592770564e-05, "loss": 0.7589, "step": 42518 }, { "epoch": 0.04, "learning_rate": 4.977829549022963e-05, "loss": 0.4681, "step": 42519 }, { "epoch": 0.04, "learning_rate": 4.977828505250903e-05, "loss": 0.5109, "step": 42520 }, { "epoch": 0.04, "learning_rate": 4.977827461454383e-05, "loss": 0.7698, "step": 42521 }, { "epoch": 0.04, "learning_rate": 4.977826417633403e-05, "loss": 0.5127, "step": 42522 }, { "epoch": 0.04, "learning_rate": 4.977825373787962e-05, "loss": 0.7131, "step": 42523 }, { "epoch": 0.04, "learning_rate": 4.977824329918062e-05, "loss": 0.6008, "step": 42524 }, { "epoch": 0.04, "learning_rate": 4.977823286023702e-05, "loss": 0.8013, "step": 42525 }, { "epoch": 0.04, "learning_rate": 4.9778222421048807e-05, "loss": 0.688, "step": 42526 }, { "epoch": 0.04, "learning_rate": 4.977821198161601e-05, "loss": 1.0171, "step": 42527 }, { "epoch": 0.04, "learning_rate": 4.9778201541938605e-05, "loss": 0.7588, "step": 42528 }, { "epoch": 0.04, "learning_rate": 4.977819110201659e-05, "loss": 0.8, "step": 42529 }, { "epoch": 0.04, "learning_rate": 4.977818066184999e-05, "loss": 0.5238, "step": 42530 }, { "epoch": 0.04, "learning_rate": 4.977817022143879e-05, "loss": 0.6202, "step": 42531 }, { "epoch": 0.04, "learning_rate": 4.977815978078298e-05, "loss": 0.6274, "step": 42532 }, { "epoch": 0.04, "learning_rate": 4.977814933988258e-05, "loss": 0.4653, "step": 42533 }, { "epoch": 0.04, "learning_rate": 4.9778138898737574e-05, "loss": 0.4164, "step": 42534 }, { "epoch": 0.04, "learning_rate": 4.977812845734797e-05, "loss": 0.6768, "step": 42535 }, { "epoch": 0.04, "learning_rate": 4.977811801571377e-05, "loss": 0.646, "step": 42536 }, { "epoch": 0.04, "learning_rate": 4.977810757383497e-05, "loss": 0.851, "step": 42537 }, { "epoch": 0.04, "learning_rate": 4.977809713171157e-05, "loss": 1.0547, "step": 42538 }, { "epoch": 0.04, "learning_rate": 4.977808668934357e-05, "loss": 0.7877, "step": 42539 }, { "epoch": 0.04, "learning_rate": 4.977807624673098e-05, "loss": 1.0812, "step": 42540 }, { "epoch": 0.04, "learning_rate": 4.9778065803873774e-05, "loss": 0.8596, "step": 42541 }, { "epoch": 0.04, "learning_rate": 4.977805536077198e-05, "loss": 1.0549, "step": 42542 }, { "epoch": 0.04, "learning_rate": 4.977804491742559e-05, "loss": 1.0203, "step": 42543 }, { "epoch": 0.04, "learning_rate": 4.97780344738346e-05, "loss": 1.08, "step": 42544 }, { "epoch": 0.04, "learning_rate": 4.9778024029999e-05, "loss": 0.9063, "step": 42545 }, { "epoch": 0.04, "learning_rate": 4.977801358591882e-05, "loss": 0.788, "step": 42546 }, { "epoch": 0.04, "learning_rate": 4.977800314159403e-05, "loss": 0.9717, "step": 42547 }, { "epoch": 0.04, "learning_rate": 4.977799269702465e-05, "loss": 0.8997, "step": 42548 }, { "epoch": 0.04, "learning_rate": 4.977798225221066e-05, "loss": 1.2194, "step": 42549 }, { "epoch": 0.04, "learning_rate": 4.977797180715208e-05, "loss": 1.0283, "step": 42550 }, { "epoch": 0.04, "learning_rate": 4.97779613618489e-05, "loss": 0.7805, "step": 42551 }, { "epoch": 0.04, "learning_rate": 4.977795091630113e-05, "loss": 0.8334, "step": 42552 }, { "epoch": 0.04, "learning_rate": 4.9777940470508756e-05, "loss": 0.8301, "step": 42553 }, { "epoch": 0.04, "learning_rate": 4.977793002447178e-05, "loss": 0.2184, "step": 42554 }, { "epoch": 0.04, "learning_rate": 4.977791957819022e-05, "loss": 0.0954, "step": 42555 }, { "epoch": 0.04, "learning_rate": 4.977790913166405e-05, "loss": 0.0764, "step": 42556 }, { "epoch": 0.04, "learning_rate": 4.977789868489329e-05, "loss": 0.0516, "step": 42557 }, { "epoch": 0.04, "learning_rate": 4.977788823787793e-05, "loss": 0.0618, "step": 42558 }, { "epoch": 0.04, "learning_rate": 4.977787779061798e-05, "loss": 0.0571, "step": 42559 }, { "epoch": 0.04, "learning_rate": 4.977786734311343e-05, "loss": 0.0419, "step": 42560 }, { "epoch": 0.04, "learning_rate": 4.977785689536427e-05, "loss": 0.0698, "step": 42561 }, { "epoch": 0.04, "learning_rate": 4.977784644737052e-05, "loss": 0.0556, "step": 42562 }, { "epoch": 0.04, "learning_rate": 4.9777835999132186e-05, "loss": 0.0509, "step": 42563 }, { "epoch": 0.04, "learning_rate": 4.9777825550649246e-05, "loss": 0.0294, "step": 42564 }, { "epoch": 0.04, "learning_rate": 4.977781510192171e-05, "loss": 0.0356, "step": 42565 }, { "epoch": 0.04, "learning_rate": 4.977780465294958e-05, "loss": 0.0457, "step": 42566 }, { "epoch": 0.04, "learning_rate": 4.977779420373285e-05, "loss": 0.0378, "step": 42567 }, { "epoch": 0.04, "learning_rate": 4.977778375427153e-05, "loss": 0.0624, "step": 42568 }, { "epoch": 0.04, "learning_rate": 4.977777330456561e-05, "loss": 0.0375, "step": 42569 }, { "epoch": 0.04, "learning_rate": 4.97777628546151e-05, "loss": 0.0477, "step": 42570 }, { "epoch": 0.04, "learning_rate": 4.9777752404419985e-05, "loss": 0.9575, "step": 42571 }, { "epoch": 0.04, "learning_rate": 4.9777741953980284e-05, "loss": 1.1572, "step": 42572 }, { "epoch": 0.04, "learning_rate": 4.9777731503295974e-05, "loss": 0.728, "step": 42573 }, { "epoch": 0.04, "learning_rate": 4.977772105236708e-05, "loss": 0.8337, "step": 42574 }, { "epoch": 0.04, "learning_rate": 4.977771060119359e-05, "loss": 1.0812, "step": 42575 }, { "epoch": 0.04, "learning_rate": 4.9777700149775506e-05, "loss": 1.0992, "step": 42576 }, { "epoch": 0.04, "learning_rate": 4.9777689698112825e-05, "loss": 1.1267, "step": 42577 }, { "epoch": 0.04, "learning_rate": 4.977767924620554e-05, "loss": 1.3608, "step": 42578 }, { "epoch": 0.04, "learning_rate": 4.977766879405368e-05, "loss": 1.2259, "step": 42579 }, { "epoch": 0.04, "learning_rate": 4.9777658341657205e-05, "loss": 1.271, "step": 42580 }, { "epoch": 0.04, "learning_rate": 4.977764788901615e-05, "loss": 0.931, "step": 42581 }, { "epoch": 0.04, "learning_rate": 4.977763743613049e-05, "loss": 0.74, "step": 42582 }, { "epoch": 0.04, "learning_rate": 4.977762698300024e-05, "loss": 1.0048, "step": 42583 }, { "epoch": 0.04, "learning_rate": 4.9777616529625395e-05, "loss": 0.9009, "step": 42584 }, { "epoch": 0.04, "learning_rate": 4.977760607600595e-05, "loss": 1.0357, "step": 42585 }, { "epoch": 0.04, "learning_rate": 4.977759562214193e-05, "loss": 1.0153, "step": 42586 }, { "epoch": 0.04, "learning_rate": 4.97775851680333e-05, "loss": 1.0923, "step": 42587 }, { "epoch": 0.04, "learning_rate": 4.9777574713680076e-05, "loss": 0.9422, "step": 42588 }, { "epoch": 0.04, "learning_rate": 4.977756425908226e-05, "loss": 1.1885, "step": 42589 }, { "epoch": 0.04, "learning_rate": 4.9777553804239854e-05, "loss": 0.9343, "step": 42590 }, { "epoch": 0.04, "learning_rate": 4.9777543349152856e-05, "loss": 0.9718, "step": 42591 }, { "epoch": 0.04, "learning_rate": 4.977753289382126e-05, "loss": 0.8049, "step": 42592 }, { "epoch": 0.04, "learning_rate": 4.977752243824507e-05, "loss": 0.9091, "step": 42593 }, { "epoch": 0.04, "learning_rate": 4.977751198242429e-05, "loss": 1.0093, "step": 42594 }, { "epoch": 0.04, "learning_rate": 4.9777501526358916e-05, "loss": 1.7444, "step": 42595 }, { "epoch": 0.04, "learning_rate": 4.977749107004895e-05, "loss": 1.8156, "step": 42596 }, { "epoch": 0.04, "learning_rate": 4.9777480613494384e-05, "loss": 1.3342, "step": 42597 }, { "epoch": 0.04, "learning_rate": 4.977747015669524e-05, "loss": 1.0839, "step": 42598 }, { "epoch": 0.04, "learning_rate": 4.977745969965149e-05, "loss": 1.0497, "step": 42599 }, { "epoch": 0.04, "learning_rate": 4.977744924236315e-05, "loss": 1.041, "step": 42600 }, { "epoch": 0.04, "learning_rate": 4.977743878483022e-05, "loss": 1.0533, "step": 42601 }, { "epoch": 0.04, "learning_rate": 4.9777428327052694e-05, "loss": 0.899, "step": 42602 }, { "epoch": 0.04, "learning_rate": 4.977741786903058e-05, "loss": 1.3532, "step": 42603 }, { "epoch": 0.04, "learning_rate": 4.977740741076387e-05, "loss": 0.8345, "step": 42604 }, { "epoch": 0.04, "learning_rate": 4.977739695225257e-05, "loss": 0.6517, "step": 42605 }, { "epoch": 0.04, "learning_rate": 4.977738649349668e-05, "loss": 0.5822, "step": 42606 }, { "epoch": 0.04, "learning_rate": 4.977737603449619e-05, "loss": 0.3397, "step": 42607 }, { "epoch": 0.04, "learning_rate": 4.9777365575251116e-05, "loss": 0.3588, "step": 42608 }, { "epoch": 0.04, "learning_rate": 4.977735511576145e-05, "loss": 0.3221, "step": 42609 }, { "epoch": 0.04, "learning_rate": 4.977734465602719e-05, "loss": 0.2572, "step": 42610 }, { "epoch": 0.04, "learning_rate": 4.977733419604834e-05, "loss": 0.1952, "step": 42611 }, { "epoch": 0.04, "learning_rate": 4.97773237358249e-05, "loss": 0.7435, "step": 42612 }, { "epoch": 0.04, "learning_rate": 4.977731327535686e-05, "loss": 0.4685, "step": 42613 }, { "epoch": 0.04, "learning_rate": 4.977730281464424e-05, "loss": 0.542, "step": 42614 }, { "epoch": 0.04, "learning_rate": 4.977729235368702e-05, "loss": 1.0661, "step": 42615 }, { "epoch": 0.04, "learning_rate": 4.977728189248522e-05, "loss": 0.3314, "step": 42616 }, { "epoch": 0.04, "learning_rate": 4.977727143103882e-05, "loss": 0.3251, "step": 42617 }, { "epoch": 0.04, "learning_rate": 4.9777260969347824e-05, "loss": 0.2539, "step": 42618 }, { "epoch": 0.04, "learning_rate": 4.9777250507412246e-05, "loss": 0.3224, "step": 42619 }, { "epoch": 0.04, "learning_rate": 4.977724004523208e-05, "loss": 0.7887, "step": 42620 }, { "epoch": 0.04, "learning_rate": 4.9777229582807316e-05, "loss": 1.1107, "step": 42621 }, { "epoch": 0.04, "learning_rate": 4.977721912013796e-05, "loss": 0.538, "step": 42622 }, { "epoch": 0.04, "learning_rate": 4.977720865722403e-05, "loss": 0.702, "step": 42623 }, { "epoch": 0.04, "learning_rate": 4.97771981940655e-05, "loss": 0.9848, "step": 42624 }, { "epoch": 0.04, "learning_rate": 4.977718773066237e-05, "loss": 0.7186, "step": 42625 }, { "epoch": 0.04, "learning_rate": 4.977717726701466e-05, "loss": 1.0263, "step": 42626 }, { "epoch": 0.04, "learning_rate": 4.977716680312236e-05, "loss": 1.2188, "step": 42627 }, { "epoch": 0.04, "learning_rate": 4.977715633898547e-05, "loss": 1.2499, "step": 42628 }, { "epoch": 0.04, "learning_rate": 4.977714587460399e-05, "loss": 1.0366, "step": 42629 }, { "epoch": 0.04, "learning_rate": 4.977713540997792e-05, "loss": 1.0327, "step": 42630 }, { "epoch": 0.04, "learning_rate": 4.9777124945107254e-05, "loss": 0.9254, "step": 42631 }, { "epoch": 0.04, "learning_rate": 4.9777114479992016e-05, "loss": 1.291, "step": 42632 }, { "epoch": 0.04, "learning_rate": 4.977710401463217e-05, "loss": 1.0566, "step": 42633 }, { "epoch": 0.04, "learning_rate": 4.9777093549027745e-05, "loss": 1.4656, "step": 42634 }, { "epoch": 0.04, "learning_rate": 4.977708308317873e-05, "loss": 1.2906, "step": 42635 }, { "epoch": 0.04, "learning_rate": 4.977707261708512e-05, "loss": 1.0808, "step": 42636 }, { "epoch": 0.04, "learning_rate": 4.977706215074693e-05, "loss": 1.2016, "step": 42637 }, { "epoch": 0.04, "learning_rate": 4.9777051684164145e-05, "loss": 0.9665, "step": 42638 }, { "epoch": 0.04, "learning_rate": 4.9777041217336776e-05, "loss": 1.1026, "step": 42639 }, { "epoch": 0.04, "learning_rate": 4.977703075026482e-05, "loss": 1.1049, "step": 42640 }, { "epoch": 0.04, "learning_rate": 4.9777020282948276e-05, "loss": 0.3294, "step": 42641 }, { "epoch": 0.04, "learning_rate": 4.977700981538713e-05, "loss": 0.6578, "step": 42642 }, { "epoch": 0.04, "learning_rate": 4.977699934758141e-05, "loss": 0.9955, "step": 42643 }, { "epoch": 0.04, "learning_rate": 4.9776988879531096e-05, "loss": 1.0043, "step": 42644 }, { "epoch": 0.04, "learning_rate": 4.97769784112362e-05, "loss": 1.2659, "step": 42645 }, { "epoch": 0.04, "learning_rate": 4.9776967942696704e-05, "loss": 1.2185, "step": 42646 }, { "epoch": 0.04, "learning_rate": 4.977695747391263e-05, "loss": 0.9046, "step": 42647 }, { "epoch": 0.04, "learning_rate": 4.977694700488397e-05, "loss": 0.7705, "step": 42648 }, { "epoch": 0.04, "learning_rate": 4.977693653561071e-05, "loss": 0.8521, "step": 42649 }, { "epoch": 0.04, "learning_rate": 4.9776926066092875e-05, "loss": 0.5084, "step": 42650 }, { "epoch": 0.04, "learning_rate": 4.977691559633045e-05, "loss": 0.3196, "step": 42651 }, { "epoch": 0.04, "learning_rate": 4.977690512632344e-05, "loss": 0.4377, "step": 42652 }, { "epoch": 0.04, "learning_rate": 4.9776894656071834e-05, "loss": 1.1788, "step": 42653 }, { "epoch": 0.04, "learning_rate": 4.9776884185575646e-05, "loss": 0.8781, "step": 42654 }, { "epoch": 0.04, "learning_rate": 4.977687371483487e-05, "loss": 1.1133, "step": 42655 }, { "epoch": 0.04, "learning_rate": 4.977686324384951e-05, "loss": 1.3389, "step": 42656 }, { "epoch": 0.04, "learning_rate": 4.977685277261956e-05, "loss": 1.1488, "step": 42657 }, { "epoch": 0.04, "learning_rate": 4.977684230114502e-05, "loss": 1.0159, "step": 42658 }, { "epoch": 0.04, "learning_rate": 4.97768318294259e-05, "loss": 0.9959, "step": 42659 }, { "epoch": 0.04, "learning_rate": 4.9776821357462186e-05, "loss": 1.0478, "step": 42660 }, { "epoch": 0.04, "learning_rate": 4.97768108852539e-05, "loss": 1.3094, "step": 42661 }, { "epoch": 0.04, "learning_rate": 4.977680041280102e-05, "loss": 1.169, "step": 42662 }, { "epoch": 0.04, "learning_rate": 4.977678994010355e-05, "loss": 0.9552, "step": 42663 }, { "epoch": 0.04, "learning_rate": 4.9776779467161496e-05, "loss": 1.1053, "step": 42664 }, { "epoch": 0.04, "learning_rate": 4.977676899397486e-05, "loss": 1.2955, "step": 42665 }, { "epoch": 0.04, "learning_rate": 4.977675852054363e-05, "loss": 1.0715, "step": 42666 }, { "epoch": 0.04, "learning_rate": 4.977674804686783e-05, "loss": 0.7172, "step": 42667 }, { "epoch": 0.04, "learning_rate": 4.977673757294743e-05, "loss": 0.8921, "step": 42668 }, { "epoch": 0.04, "learning_rate": 4.977672709878245e-05, "loss": 0.9737, "step": 42669 }, { "epoch": 0.04, "learning_rate": 4.977671662437288e-05, "loss": 0.6909, "step": 42670 }, { "epoch": 0.04, "learning_rate": 4.977670614971873e-05, "loss": 0.5317, "step": 42671 }, { "epoch": 0.04, "learning_rate": 4.977669567481999e-05, "loss": 0.5758, "step": 42672 }, { "epoch": 0.04, "learning_rate": 4.977668519967667e-05, "loss": 1.012, "step": 42673 }, { "epoch": 0.04, "learning_rate": 4.9776674724288755e-05, "loss": 0.9415, "step": 42674 }, { "epoch": 0.04, "learning_rate": 4.977666424865627e-05, "loss": 0.9863, "step": 42675 }, { "epoch": 0.04, "learning_rate": 4.977665377277919e-05, "loss": 1.1086, "step": 42676 }, { "epoch": 0.04, "learning_rate": 4.9776643296657526e-05, "loss": 1.1089, "step": 42677 }, { "epoch": 0.04, "learning_rate": 4.977663282029128e-05, "loss": 1.2055, "step": 42678 }, { "epoch": 0.04, "learning_rate": 4.977662234368045e-05, "loss": 1.2459, "step": 42679 }, { "epoch": 0.04, "learning_rate": 4.977661186682503e-05, "loss": 1.723, "step": 42680 }, { "epoch": 0.04, "learning_rate": 4.977660138972503e-05, "loss": 1.0531, "step": 42681 }, { "epoch": 0.04, "learning_rate": 4.977659091238045e-05, "loss": 1.1736, "step": 42682 }, { "epoch": 0.04, "learning_rate": 4.977658043479128e-05, "loss": 1.073, "step": 42683 }, { "epoch": 0.04, "learning_rate": 4.977656995695753e-05, "loss": 0.8418, "step": 42684 }, { "epoch": 0.04, "learning_rate": 4.977655947887919e-05, "loss": 0.7044, "step": 42685 }, { "epoch": 0.04, "learning_rate": 4.977654900055627e-05, "loss": 0.7863, "step": 42686 }, { "epoch": 0.04, "learning_rate": 4.977653852198877e-05, "loss": 0.8825, "step": 42687 }, { "epoch": 0.04, "learning_rate": 4.977652804317668e-05, "loss": 1.4496, "step": 42688 }, { "epoch": 0.04, "learning_rate": 4.977651756412002e-05, "loss": 1.8104, "step": 42689 }, { "epoch": 0.04, "learning_rate": 4.9776507084818755e-05, "loss": 1.2927, "step": 42690 }, { "epoch": 0.04, "learning_rate": 4.977649660527293e-05, "loss": 0.6164, "step": 42691 }, { "epoch": 0.04, "learning_rate": 4.9776486125482505e-05, "loss": 0.7977, "step": 42692 }, { "epoch": 0.04, "learning_rate": 4.97764756454475e-05, "loss": 0.7111, "step": 42693 }, { "epoch": 0.04, "learning_rate": 4.977646516516792e-05, "loss": 0.6245, "step": 42694 }, { "epoch": 0.04, "learning_rate": 4.977645468464375e-05, "loss": 0.7006, "step": 42695 }, { "epoch": 0.04, "learning_rate": 4.9776444203874996e-05, "loss": 0.7724, "step": 42696 }, { "epoch": 0.04, "learning_rate": 4.977643372286166e-05, "loss": 0.9555, "step": 42697 }, { "epoch": 0.04, "learning_rate": 4.9776423241603744e-05, "loss": 1.2573, "step": 42698 }, { "epoch": 0.04, "learning_rate": 4.9776412760101244e-05, "loss": 1.2192, "step": 42699 }, { "epoch": 0.04, "learning_rate": 4.977640227835416e-05, "loss": 0.4652, "step": 42700 }, { "epoch": 0.04, "learning_rate": 4.9776391796362497e-05, "loss": 0.2002, "step": 42701 }, { "epoch": 0.04, "learning_rate": 4.977638131412625e-05, "loss": 0.2032, "step": 42702 }, { "epoch": 0.04, "learning_rate": 4.9776370831645426e-05, "loss": 0.8903, "step": 42703 }, { "epoch": 0.04, "learning_rate": 4.9776360348920014e-05, "loss": 1.0884, "step": 42704 }, { "epoch": 0.04, "learning_rate": 4.9776349865950026e-05, "loss": 0.8858, "step": 42705 }, { "epoch": 0.04, "learning_rate": 4.9776339382735456e-05, "loss": 0.9258, "step": 42706 }, { "epoch": 0.04, "learning_rate": 4.97763288992763e-05, "loss": 0.2681, "step": 42707 }, { "epoch": 0.04, "learning_rate": 4.977631841557257e-05, "loss": 0.2755, "step": 42708 }, { "epoch": 0.04, "learning_rate": 4.977630793162424e-05, "loss": 0.974, "step": 42709 }, { "epoch": 0.04, "learning_rate": 4.977629744743135e-05, "loss": 1.1515, "step": 42710 }, { "epoch": 0.04, "learning_rate": 4.977628696299387e-05, "loss": 1.2309, "step": 42711 }, { "epoch": 0.04, "learning_rate": 4.977627647831181e-05, "loss": 1.2935, "step": 42712 }, { "epoch": 0.04, "learning_rate": 4.977626599338516e-05, "loss": 1.172, "step": 42713 }, { "epoch": 0.04, "learning_rate": 4.9776255508213944e-05, "loss": 0.9681, "step": 42714 }, { "epoch": 0.04, "learning_rate": 4.977624502279814e-05, "loss": 0.5731, "step": 42715 }, { "epoch": 0.04, "learning_rate": 4.977623453713776e-05, "loss": 1.0202, "step": 42716 }, { "epoch": 0.04, "learning_rate": 4.9776224051232786e-05, "loss": 1.131, "step": 42717 }, { "epoch": 0.04, "learning_rate": 4.977621356508325e-05, "loss": 1.0168, "step": 42718 }, { "epoch": 0.04, "learning_rate": 4.977620307868912e-05, "loss": 0.8479, "step": 42719 }, { "epoch": 0.04, "learning_rate": 4.9776192592050413e-05, "loss": 0.9857, "step": 42720 }, { "epoch": 0.04, "learning_rate": 4.9776182105167127e-05, "loss": 0.8723, "step": 42721 }, { "epoch": 0.04, "learning_rate": 4.9776171618039264e-05, "loss": 0.9408, "step": 42722 }, { "epoch": 0.04, "learning_rate": 4.977616113066682e-05, "loss": 0.7809, "step": 42723 }, { "epoch": 0.04, "learning_rate": 4.977615064304979e-05, "loss": 0.7154, "step": 42724 }, { "epoch": 0.04, "learning_rate": 4.977614015518819e-05, "loss": 0.7697, "step": 42725 }, { "epoch": 0.04, "learning_rate": 4.9776129667082e-05, "loss": 0.6247, "step": 42726 }, { "epoch": 0.04, "learning_rate": 4.977611917873124e-05, "loss": 0.9467, "step": 42727 }, { "epoch": 0.04, "learning_rate": 4.977610869013589e-05, "loss": 1.0218, "step": 42728 }, { "epoch": 0.04, "learning_rate": 4.977609820129597e-05, "loss": 1.0929, "step": 42729 }, { "epoch": 0.04, "learning_rate": 4.977608771221147e-05, "loss": 1.1389, "step": 42730 }, { "epoch": 0.04, "learning_rate": 4.977607722288239e-05, "loss": 1.1759, "step": 42731 }, { "epoch": 0.04, "learning_rate": 4.9776066733308725e-05, "loss": 0.8979, "step": 42732 }, { "epoch": 0.04, "learning_rate": 4.977605624349049e-05, "loss": 1.2797, "step": 42733 }, { "epoch": 0.04, "learning_rate": 4.977604575342767e-05, "loss": 0.5025, "step": 42734 }, { "epoch": 0.04, "learning_rate": 4.977603526312027e-05, "loss": 0.1793, "step": 42735 }, { "epoch": 0.04, "learning_rate": 4.97760247725683e-05, "loss": 0.956, "step": 42736 }, { "epoch": 0.04, "learning_rate": 4.977601428177174e-05, "loss": 0.9807, "step": 42737 }, { "epoch": 0.04, "learning_rate": 4.977600379073061e-05, "loss": 1.134, "step": 42738 }, { "epoch": 0.04, "learning_rate": 4.9775993299444904e-05, "loss": 1.229, "step": 42739 }, { "epoch": 0.04, "learning_rate": 4.977598280791461e-05, "loss": 0.8884, "step": 42740 }, { "epoch": 0.04, "learning_rate": 4.977597231613974e-05, "loss": 0.491, "step": 42741 }, { "epoch": 0.04, "learning_rate": 4.97759618241203e-05, "loss": 1.0774, "step": 42742 }, { "epoch": 0.04, "learning_rate": 4.977595133185627e-05, "loss": 1.2082, "step": 42743 }, { "epoch": 0.04, "learning_rate": 4.977594083934768e-05, "loss": 0.7981, "step": 42744 }, { "epoch": 0.04, "learning_rate": 4.97759303465945e-05, "loss": 1.2515, "step": 42745 }, { "epoch": 0.04, "learning_rate": 4.977591985359674e-05, "loss": 1.282, "step": 42746 }, { "epoch": 0.04, "learning_rate": 4.977590936035441e-05, "loss": 0.2743, "step": 42747 }, { "epoch": 0.04, "learning_rate": 4.97758988668675e-05, "loss": 0.9013, "step": 42748 }, { "epoch": 0.04, "learning_rate": 4.977588837313602e-05, "loss": 0.9887, "step": 42749 }, { "epoch": 0.04, "learning_rate": 4.977587787915995e-05, "loss": 0.8917, "step": 42750 }, { "epoch": 0.04, "learning_rate": 4.977586738493931e-05, "loss": 0.914, "step": 42751 }, { "epoch": 0.04, "learning_rate": 4.977585689047409e-05, "loss": 1.0852, "step": 42752 }, { "epoch": 0.04, "learning_rate": 4.977584639576429e-05, "loss": 1.1459, "step": 42753 }, { "epoch": 0.04, "learning_rate": 4.977583590080992e-05, "loss": 1.3332, "step": 42754 }, { "epoch": 0.04, "learning_rate": 4.977582540561098e-05, "loss": 1.051, "step": 42755 }, { "epoch": 0.04, "learning_rate": 4.977581491016745e-05, "loss": 0.9653, "step": 42756 }, { "epoch": 0.04, "learning_rate": 4.977580441447935e-05, "loss": 0.9907, "step": 42757 }, { "epoch": 0.04, "learning_rate": 4.9775793918546665e-05, "loss": 1.2277, "step": 42758 }, { "epoch": 0.04, "learning_rate": 4.9775783422369414e-05, "loss": 1.326, "step": 42759 }, { "epoch": 0.04, "learning_rate": 4.977577292594758e-05, "loss": 1.2112, "step": 42760 }, { "epoch": 0.04, "learning_rate": 4.977576242928118e-05, "loss": 1.1766, "step": 42761 }, { "epoch": 0.04, "learning_rate": 4.977575193237019e-05, "loss": 0.9502, "step": 42762 }, { "epoch": 0.04, "learning_rate": 4.977574143521464e-05, "loss": 1.5337, "step": 42763 }, { "epoch": 0.04, "learning_rate": 4.9775730937814504e-05, "loss": 0.7898, "step": 42764 }, { "epoch": 0.04, "learning_rate": 4.977572044016979e-05, "loss": 1.0904, "step": 42765 }, { "epoch": 0.04, "learning_rate": 4.977570994228051e-05, "loss": 1.1057, "step": 42766 }, { "epoch": 0.04, "learning_rate": 4.977569944414665e-05, "loss": 1.1902, "step": 42767 }, { "epoch": 0.04, "learning_rate": 4.9775688945768216e-05, "loss": 1.0893, "step": 42768 }, { "epoch": 0.04, "learning_rate": 4.97756784471452e-05, "loss": 0.5359, "step": 42769 }, { "epoch": 0.04, "learning_rate": 4.977566794827762e-05, "loss": 0.7177, "step": 42770 }, { "epoch": 0.04, "learning_rate": 4.977565744916546e-05, "loss": 0.8732, "step": 42771 }, { "epoch": 0.04, "learning_rate": 4.9775646949808724e-05, "loss": 0.893, "step": 42772 }, { "epoch": 0.04, "learning_rate": 4.9775636450207414e-05, "loss": 1.049, "step": 42773 }, { "epoch": 0.04, "learning_rate": 4.977562595036153e-05, "loss": 0.9523, "step": 42774 }, { "epoch": 0.04, "learning_rate": 4.977561545027107e-05, "loss": 0.9452, "step": 42775 }, { "epoch": 0.04, "learning_rate": 4.977560494993604e-05, "loss": 1.2934, "step": 42776 }, { "epoch": 0.04, "learning_rate": 4.977559444935643e-05, "loss": 1.0027, "step": 42777 }, { "epoch": 0.04, "learning_rate": 4.9775583948532246e-05, "loss": 1.1559, "step": 42778 }, { "epoch": 0.04, "learning_rate": 4.977557344746349e-05, "loss": 0.9811, "step": 42779 }, { "epoch": 0.04, "learning_rate": 4.9775562946150155e-05, "loss": 0.5485, "step": 42780 }, { "epoch": 0.04, "learning_rate": 4.977555244459225e-05, "loss": 0.7607, "step": 42781 }, { "epoch": 0.04, "learning_rate": 4.9775541942789775e-05, "loss": 0.7587, "step": 42782 }, { "epoch": 0.04, "learning_rate": 4.977553144074273e-05, "loss": 0.4716, "step": 42783 }, { "epoch": 0.04, "learning_rate": 4.9775520938451105e-05, "loss": 0.6627, "step": 42784 }, { "epoch": 0.04, "learning_rate": 4.97755104359149e-05, "loss": 0.8478, "step": 42785 }, { "epoch": 0.04, "learning_rate": 4.9775499933134126e-05, "loss": 0.6296, "step": 42786 }, { "epoch": 0.04, "learning_rate": 4.9775489430108777e-05, "loss": 0.9362, "step": 42787 }, { "epoch": 0.04, "learning_rate": 4.9775478926838865e-05, "loss": 0.8183, "step": 42788 }, { "epoch": 0.04, "learning_rate": 4.977546842332437e-05, "loss": 1.1356, "step": 42789 }, { "epoch": 0.04, "learning_rate": 4.977545791956531e-05, "loss": 1.1323, "step": 42790 }, { "epoch": 0.04, "learning_rate": 4.977544741556167e-05, "loss": 1.1587, "step": 42791 }, { "epoch": 0.04, "learning_rate": 4.977543691131346e-05, "loss": 1.0544, "step": 42792 }, { "epoch": 0.04, "learning_rate": 4.977542640682068e-05, "loss": 0.9496, "step": 42793 }, { "epoch": 0.04, "learning_rate": 4.9775415902083325e-05, "loss": 0.9477, "step": 42794 }, { "epoch": 0.04, "learning_rate": 4.977540539710139e-05, "loss": 0.9245, "step": 42795 }, { "epoch": 0.04, "learning_rate": 4.977539489187489e-05, "loss": 1.0332, "step": 42796 }, { "epoch": 0.04, "learning_rate": 4.977538438640382e-05, "loss": 1.2262, "step": 42797 }, { "epoch": 0.04, "learning_rate": 4.977537388068817e-05, "loss": 1.3267, "step": 42798 }, { "epoch": 0.04, "learning_rate": 4.9775363374727955e-05, "loss": 0.9899, "step": 42799 }, { "epoch": 0.04, "learning_rate": 4.977535286852317e-05, "loss": 1.4075, "step": 42800 }, { "epoch": 0.04, "learning_rate": 4.977534236207381e-05, "loss": 1.0612, "step": 42801 }, { "epoch": 0.04, "learning_rate": 4.977533185537988e-05, "loss": 0.8175, "step": 42802 }, { "epoch": 0.04, "learning_rate": 4.977532134844137e-05, "loss": 0.8422, "step": 42803 }, { "epoch": 0.04, "learning_rate": 4.9775310841258295e-05, "loss": 1.2201, "step": 42804 }, { "epoch": 0.04, "learning_rate": 4.977530033383065e-05, "loss": 1.0676, "step": 42805 }, { "epoch": 0.04, "learning_rate": 4.977528982615843e-05, "loss": 0.9154, "step": 42806 }, { "epoch": 0.04, "learning_rate": 4.977527931824164e-05, "loss": 1.5186, "step": 42807 }, { "epoch": 0.04, "learning_rate": 4.977526881008028e-05, "loss": 0.9355, "step": 42808 }, { "epoch": 0.04, "learning_rate": 4.9775258301674344e-05, "loss": 1.0017, "step": 42809 }, { "epoch": 0.04, "learning_rate": 4.977524779302384e-05, "loss": 1.1253, "step": 42810 }, { "epoch": 0.04, "learning_rate": 4.977523728412877e-05, "loss": 0.8246, "step": 42811 }, { "epoch": 0.04, "learning_rate": 4.9775226774989124e-05, "loss": 1.1565, "step": 42812 }, { "epoch": 0.04, "learning_rate": 4.9775216265604915e-05, "loss": 0.8986, "step": 42813 }, { "epoch": 0.04, "learning_rate": 4.977520575597613e-05, "loss": 0.7151, "step": 42814 }, { "epoch": 0.04, "learning_rate": 4.977519524610277e-05, "loss": 1.1747, "step": 42815 }, { "epoch": 0.04, "learning_rate": 4.977518473598485e-05, "loss": 1.2003, "step": 42816 }, { "epoch": 0.04, "learning_rate": 4.977517422562234e-05, "loss": 1.0464, "step": 42817 }, { "epoch": 0.04, "learning_rate": 4.977516371501528e-05, "loss": 0.6176, "step": 42818 }, { "epoch": 0.04, "learning_rate": 4.977515320416365e-05, "loss": 0.7661, "step": 42819 }, { "epoch": 0.04, "learning_rate": 4.977514269306744e-05, "loss": 0.8642, "step": 42820 }, { "epoch": 0.04, "learning_rate": 4.977513218172666e-05, "loss": 0.852, "step": 42821 }, { "epoch": 0.04, "learning_rate": 4.9775121670141325e-05, "loss": 1.0256, "step": 42822 }, { "epoch": 0.04, "learning_rate": 4.977511115831141e-05, "loss": 0.8578, "step": 42823 }, { "epoch": 0.04, "learning_rate": 4.977510064623691e-05, "loss": 1.1915, "step": 42824 }, { "epoch": 0.04, "learning_rate": 4.977509013391787e-05, "loss": 1.3047, "step": 42825 }, { "epoch": 0.04, "learning_rate": 4.9775079621354245e-05, "loss": 0.9746, "step": 42826 }, { "epoch": 0.04, "learning_rate": 4.9775069108546045e-05, "loss": 1.3528, "step": 42827 }, { "epoch": 0.04, "learning_rate": 4.977505859549328e-05, "loss": 1.0742, "step": 42828 }, { "epoch": 0.04, "learning_rate": 4.977504808219596e-05, "loss": 1.0521, "step": 42829 }, { "epoch": 0.04, "learning_rate": 4.9775037568654056e-05, "loss": 0.8526, "step": 42830 }, { "epoch": 0.04, "learning_rate": 4.977502705486759e-05, "loss": 0.8938, "step": 42831 }, { "epoch": 0.04, "learning_rate": 4.9775016540836555e-05, "loss": 0.9696, "step": 42832 }, { "epoch": 0.04, "learning_rate": 4.977500602656095e-05, "loss": 0.9771, "step": 42833 }, { "epoch": 0.04, "learning_rate": 4.977499551204078e-05, "loss": 1.0117, "step": 42834 }, { "epoch": 0.04, "learning_rate": 4.977498499727603e-05, "loss": 1.0129, "step": 42835 }, { "epoch": 0.04, "learning_rate": 4.9774974482266725e-05, "loss": 0.9775, "step": 42836 }, { "epoch": 0.04, "learning_rate": 4.977496396701284e-05, "loss": 0.9269, "step": 42837 }, { "epoch": 0.04, "learning_rate": 4.97749534515144e-05, "loss": 0.9701, "step": 42838 }, { "epoch": 0.04, "learning_rate": 4.977494293577138e-05, "loss": 1.0457, "step": 42839 }, { "epoch": 0.04, "learning_rate": 4.977493241978381e-05, "loss": 1.103, "step": 42840 }, { "epoch": 0.04, "learning_rate": 4.9774921903551655e-05, "loss": 0.6487, "step": 42841 }, { "epoch": 0.04, "learning_rate": 4.977491138707494e-05, "loss": 1.1235, "step": 42842 }, { "epoch": 0.04, "learning_rate": 4.977490087035365e-05, "loss": 0.775, "step": 42843 }, { "epoch": 0.04, "learning_rate": 4.9774890353387807e-05, "loss": 0.9125, "step": 42844 }, { "epoch": 0.04, "learning_rate": 4.977487983617738e-05, "loss": 0.9928, "step": 42845 }, { "epoch": 0.04, "learning_rate": 4.9774869318722396e-05, "loss": 0.9132, "step": 42846 }, { "epoch": 0.04, "learning_rate": 4.977485880102284e-05, "loss": 1.275, "step": 42847 }, { "epoch": 0.04, "learning_rate": 4.9774848283078724e-05, "loss": 1.0192, "step": 42848 }, { "epoch": 0.04, "learning_rate": 4.9774837764890034e-05, "loss": 0.9372, "step": 42849 }, { "epoch": 0.04, "learning_rate": 4.9774827246456776e-05, "loss": 1.3051, "step": 42850 }, { "epoch": 0.04, "learning_rate": 4.977481672777896e-05, "loss": 0.8345, "step": 42851 }, { "epoch": 0.04, "learning_rate": 4.977480620885657e-05, "loss": 0.8332, "step": 42852 }, { "epoch": 0.04, "learning_rate": 4.977479568968962e-05, "loss": 1.0966, "step": 42853 }, { "epoch": 0.04, "learning_rate": 4.97747851702781e-05, "loss": 0.9843, "step": 42854 }, { "epoch": 0.04, "learning_rate": 4.977477465062201e-05, "loss": 0.7905, "step": 42855 }, { "epoch": 0.04, "learning_rate": 4.9774764130721366e-05, "loss": 0.722, "step": 42856 }, { "epoch": 0.04, "learning_rate": 4.977475361057614e-05, "loss": 0.8859, "step": 42857 }, { "epoch": 0.04, "learning_rate": 4.9774743090186356e-05, "loss": 1.2723, "step": 42858 }, { "epoch": 0.04, "learning_rate": 4.9774732569552005e-05, "loss": 1.3981, "step": 42859 }, { "epoch": 0.04, "learning_rate": 4.977472204867309e-05, "loss": 1.0083, "step": 42860 }, { "epoch": 0.04, "learning_rate": 4.977471152754961e-05, "loss": 1.1207, "step": 42861 }, { "epoch": 0.04, "learning_rate": 4.9774701006181564e-05, "loss": 1.0363, "step": 42862 }, { "epoch": 0.04, "learning_rate": 4.977469048456895e-05, "loss": 1.0857, "step": 42863 }, { "epoch": 0.04, "learning_rate": 4.9774679962711775e-05, "loss": 0.9556, "step": 42864 }, { "epoch": 0.04, "learning_rate": 4.977466944061003e-05, "loss": 1.705, "step": 42865 }, { "epoch": 0.04, "learning_rate": 4.9774658918263724e-05, "loss": 1.2292, "step": 42866 }, { "epoch": 0.04, "learning_rate": 4.9774648395672855e-05, "loss": 0.9622, "step": 42867 }, { "epoch": 0.04, "learning_rate": 4.977463787283741e-05, "loss": 1.1031, "step": 42868 }, { "epoch": 0.04, "learning_rate": 4.977462734975741e-05, "loss": 1.0836, "step": 42869 }, { "epoch": 0.04, "learning_rate": 4.977461682643284e-05, "loss": 1.0089, "step": 42870 }, { "epoch": 0.04, "learning_rate": 4.977460630286371e-05, "loss": 0.9525, "step": 42871 }, { "epoch": 0.04, "learning_rate": 4.977459577905002e-05, "loss": 0.9914, "step": 42872 }, { "epoch": 0.04, "learning_rate": 4.977458525499175e-05, "loss": 0.9403, "step": 42873 }, { "epoch": 0.04, "learning_rate": 4.9774574730688925e-05, "loss": 0.8716, "step": 42874 }, { "epoch": 0.04, "learning_rate": 4.977456420614154e-05, "loss": 1.0783, "step": 42875 }, { "epoch": 0.04, "learning_rate": 4.9774553681349583e-05, "loss": 0.9486, "step": 42876 }, { "epoch": 0.04, "learning_rate": 4.9774543156313066e-05, "loss": 0.8307, "step": 42877 }, { "epoch": 0.04, "learning_rate": 4.977453263103199e-05, "loss": 1.2308, "step": 42878 }, { "epoch": 0.04, "learning_rate": 4.977452210550634e-05, "loss": 0.9446, "step": 42879 }, { "epoch": 0.04, "learning_rate": 4.9774511579736135e-05, "loss": 1.2575, "step": 42880 }, { "epoch": 0.04, "learning_rate": 4.9774501053721356e-05, "loss": 1.0518, "step": 42881 }, { "epoch": 0.04, "learning_rate": 4.977449052746203e-05, "loss": 0.8322, "step": 42882 }, { "epoch": 0.04, "learning_rate": 4.9774480000958124e-05, "loss": 0.8818, "step": 42883 }, { "epoch": 0.04, "learning_rate": 4.977446947420966e-05, "loss": 1.1011, "step": 42884 }, { "epoch": 0.04, "learning_rate": 4.977445894721664e-05, "loss": 1.1485, "step": 42885 }, { "epoch": 0.04, "learning_rate": 4.977444841997905e-05, "loss": 0.8346, "step": 42886 }, { "epoch": 0.04, "learning_rate": 4.97744378924969e-05, "loss": 1.1321, "step": 42887 }, { "epoch": 0.04, "learning_rate": 4.977442736477018e-05, "loss": 0.7627, "step": 42888 }, { "epoch": 0.04, "learning_rate": 4.977441683679891e-05, "loss": 1.1072, "step": 42889 }, { "epoch": 0.04, "learning_rate": 4.9774406308583066e-05, "loss": 1.0324, "step": 42890 }, { "epoch": 0.04, "learning_rate": 4.9774395780122666e-05, "loss": 0.7804, "step": 42891 }, { "epoch": 0.04, "learning_rate": 4.97743852514177e-05, "loss": 0.9499, "step": 42892 }, { "epoch": 0.04, "learning_rate": 4.977437472246818e-05, "loss": 1.109, "step": 42893 }, { "epoch": 0.04, "learning_rate": 4.977436419327409e-05, "loss": 1.1391, "step": 42894 }, { "epoch": 0.04, "learning_rate": 4.9774353663835435e-05, "loss": 1.0359, "step": 42895 }, { "epoch": 0.04, "learning_rate": 4.9774343134152224e-05, "loss": 1.0014, "step": 42896 }, { "epoch": 0.04, "learning_rate": 4.977433260422445e-05, "loss": 0.5154, "step": 42897 }, { "epoch": 0.04, "learning_rate": 4.9774322074052115e-05, "loss": 0.4159, "step": 42898 }, { "epoch": 0.04, "learning_rate": 4.977431154363522e-05, "loss": 1.1144, "step": 42899 }, { "epoch": 0.04, "learning_rate": 4.977430101297376e-05, "loss": 0.7957, "step": 42900 }, { "epoch": 0.04, "learning_rate": 4.9774290482067736e-05, "loss": 0.9, "step": 42901 }, { "epoch": 0.04, "learning_rate": 4.977427995091715e-05, "loss": 1.0368, "step": 42902 }, { "epoch": 0.04, "learning_rate": 4.977426941952201e-05, "loss": 1.041, "step": 42903 }, { "epoch": 0.04, "learning_rate": 4.977425888788231e-05, "loss": 1.2897, "step": 42904 }, { "epoch": 0.04, "learning_rate": 4.977424835599804e-05, "loss": 1.275, "step": 42905 }, { "epoch": 0.04, "learning_rate": 4.9774237823869215e-05, "loss": 0.8969, "step": 42906 }, { "epoch": 0.04, "learning_rate": 4.977422729149582e-05, "loss": 0.9335, "step": 42907 }, { "epoch": 0.04, "learning_rate": 4.977421675887788e-05, "loss": 0.9518, "step": 42908 }, { "epoch": 0.04, "learning_rate": 4.977420622601537e-05, "loss": 1.1922, "step": 42909 }, { "epoch": 0.04, "learning_rate": 4.97741956929083e-05, "loss": 1.1098, "step": 42910 }, { "epoch": 0.04, "learning_rate": 4.977418515955667e-05, "loss": 1.1723, "step": 42911 }, { "epoch": 0.04, "learning_rate": 4.9774174625960476e-05, "loss": 1.1708, "step": 42912 }, { "epoch": 0.04, "learning_rate": 4.977416409211972e-05, "loss": 1.1952, "step": 42913 }, { "epoch": 0.04, "learning_rate": 4.977415355803442e-05, "loss": 0.9896, "step": 42914 }, { "epoch": 0.04, "learning_rate": 4.977414302370454e-05, "loss": 1.0308, "step": 42915 }, { "epoch": 0.04, "learning_rate": 4.9774132489130115e-05, "loss": 1.2175, "step": 42916 }, { "epoch": 0.04, "learning_rate": 4.9774121954311126e-05, "loss": 1.2778, "step": 42917 }, { "epoch": 0.04, "learning_rate": 4.9774111419247575e-05, "loss": 1.0854, "step": 42918 }, { "epoch": 0.04, "learning_rate": 4.977410088393947e-05, "loss": 1.0974, "step": 42919 }, { "epoch": 0.04, "learning_rate": 4.97740903483868e-05, "loss": 1.2764, "step": 42920 }, { "epoch": 0.04, "learning_rate": 4.977407981258957e-05, "loss": 0.6539, "step": 42921 }, { "epoch": 0.04, "learning_rate": 4.977406927654778e-05, "loss": 0.8578, "step": 42922 }, { "epoch": 0.04, "learning_rate": 4.977405874026143e-05, "loss": 0.5896, "step": 42923 }, { "epoch": 0.04, "learning_rate": 4.9774048203730526e-05, "loss": 0.8544, "step": 42924 }, { "epoch": 0.04, "learning_rate": 4.977403766695507e-05, "loss": 1.0989, "step": 42925 }, { "epoch": 0.04, "learning_rate": 4.977402712993504e-05, "loss": 1.0821, "step": 42926 }, { "epoch": 0.04, "learning_rate": 4.9774016592670456e-05, "loss": 0.8324, "step": 42927 }, { "epoch": 0.04, "learning_rate": 4.977400605516132e-05, "loss": 1.1967, "step": 42928 }, { "epoch": 0.04, "learning_rate": 4.977399551740762e-05, "loss": 0.9882, "step": 42929 }, { "epoch": 0.04, "learning_rate": 4.977398497940936e-05, "loss": 0.9939, "step": 42930 }, { "epoch": 0.04, "learning_rate": 4.977397444116654e-05, "loss": 0.7449, "step": 42931 }, { "epoch": 0.04, "learning_rate": 4.977396390267916e-05, "loss": 0.7581, "step": 42932 }, { "epoch": 0.04, "learning_rate": 4.9773953363947235e-05, "loss": 0.831, "step": 42933 }, { "epoch": 0.04, "learning_rate": 4.9773942824970744e-05, "loss": 1.2357, "step": 42934 }, { "epoch": 0.04, "learning_rate": 4.977393228574969e-05, "loss": 1.0329, "step": 42935 }, { "epoch": 0.04, "learning_rate": 4.977392174628409e-05, "loss": 0.9139, "step": 42936 }, { "epoch": 0.04, "learning_rate": 4.9773911206573926e-05, "loss": 1.1432, "step": 42937 }, { "epoch": 0.04, "learning_rate": 4.97739006666192e-05, "loss": 0.8416, "step": 42938 }, { "epoch": 0.04, "learning_rate": 4.977389012641992e-05, "loss": 0.343, "step": 42939 }, { "epoch": 0.04, "learning_rate": 4.977387958597608e-05, "loss": 1.4421, "step": 42940 }, { "epoch": 0.04, "learning_rate": 4.977386904528769e-05, "loss": 0.8713, "step": 42941 }, { "epoch": 0.04, "learning_rate": 4.977385850435474e-05, "loss": 1.049, "step": 42942 }, { "epoch": 0.04, "learning_rate": 4.9773847963177234e-05, "loss": 1.0541, "step": 42943 }, { "epoch": 0.04, "learning_rate": 4.977383742175516e-05, "loss": 0.8917, "step": 42944 }, { "epoch": 0.04, "learning_rate": 4.977382688008855e-05, "loss": 1.1077, "step": 42945 }, { "epoch": 0.04, "learning_rate": 4.977381633817737e-05, "loss": 1.3695, "step": 42946 }, { "epoch": 0.04, "learning_rate": 4.9773805796021636e-05, "loss": 1.4334, "step": 42947 }, { "epoch": 0.04, "learning_rate": 4.977379525362134e-05, "loss": 2.6273, "step": 42948 }, { "epoch": 0.04, "learning_rate": 4.977378471097649e-05, "loss": 1.9137, "step": 42949 }, { "epoch": 0.04, "learning_rate": 4.9773774168087085e-05, "loss": 1.524, "step": 42950 }, { "epoch": 0.04, "learning_rate": 4.977376362495312e-05, "loss": 1.1171, "step": 42951 }, { "epoch": 0.04, "learning_rate": 4.9773753081574605e-05, "loss": 1.1758, "step": 42952 }, { "epoch": 0.04, "learning_rate": 4.977374253795153e-05, "loss": 1.4397, "step": 42953 }, { "epoch": 0.04, "learning_rate": 4.9773731994083904e-05, "loss": 1.1422, "step": 42954 }, { "epoch": 0.04, "learning_rate": 4.9773721449971716e-05, "loss": 1.0409, "step": 42955 }, { "epoch": 0.04, "learning_rate": 4.977371090561498e-05, "loss": 1.4511, "step": 42956 }, { "epoch": 0.04, "learning_rate": 4.9773700361013676e-05, "loss": 1.1277, "step": 42957 }, { "epoch": 0.04, "learning_rate": 4.977368981616782e-05, "loss": 1.0458, "step": 42958 }, { "epoch": 0.04, "learning_rate": 4.9773679271077414e-05, "loss": 1.156, "step": 42959 }, { "epoch": 0.04, "learning_rate": 4.977366872574246e-05, "loss": 0.7718, "step": 42960 }, { "epoch": 0.04, "learning_rate": 4.977365818016294e-05, "loss": 0.7421, "step": 42961 }, { "epoch": 0.04, "learning_rate": 4.9773647634338864e-05, "loss": 1.0194, "step": 42962 }, { "epoch": 0.04, "learning_rate": 4.9773637088270234e-05, "loss": 0.9989, "step": 42963 }, { "epoch": 0.04, "learning_rate": 4.9773626541957055e-05, "loss": 1.2296, "step": 42964 }, { "epoch": 0.04, "learning_rate": 4.9773615995399315e-05, "loss": 1.1649, "step": 42965 }, { "epoch": 0.04, "learning_rate": 4.977360544859702e-05, "loss": 1.3091, "step": 42966 }, { "epoch": 0.04, "learning_rate": 4.977359490155017e-05, "loss": 1.1339, "step": 42967 }, { "epoch": 0.04, "learning_rate": 4.9773584354258774e-05, "loss": 0.9427, "step": 42968 }, { "epoch": 0.04, "learning_rate": 4.977357380672282e-05, "loss": 0.7372, "step": 42969 }, { "epoch": 0.04, "learning_rate": 4.97735632589423e-05, "loss": 1.1729, "step": 42970 }, { "epoch": 0.04, "learning_rate": 4.9773552710917235e-05, "loss": 0.9114, "step": 42971 }, { "epoch": 0.04, "learning_rate": 4.977354216264762e-05, "loss": 0.9101, "step": 42972 }, { "epoch": 0.04, "learning_rate": 4.977353161413345e-05, "loss": 1.0287, "step": 42973 }, { "epoch": 0.04, "learning_rate": 4.977352106537472e-05, "loss": 0.5721, "step": 42974 }, { "epoch": 0.04, "learning_rate": 4.977351051637144e-05, "loss": 0.3726, "step": 42975 }, { "epoch": 0.04, "learning_rate": 4.97734999671236e-05, "loss": 1.0583, "step": 42976 }, { "epoch": 0.04, "learning_rate": 4.977348941763122e-05, "loss": 0.5973, "step": 42977 }, { "epoch": 0.04, "learning_rate": 4.977347886789427e-05, "loss": 0.8969, "step": 42978 }, { "epoch": 0.04, "learning_rate": 4.977346831791278e-05, "loss": 1.3467, "step": 42979 }, { "epoch": 0.04, "learning_rate": 4.977345776768674e-05, "loss": 0.9516, "step": 42980 }, { "epoch": 0.04, "learning_rate": 4.977344721721613e-05, "loss": 1.0553, "step": 42981 }, { "epoch": 0.04, "learning_rate": 4.9773436666500985e-05, "loss": 1.1517, "step": 42982 }, { "epoch": 0.04, "learning_rate": 4.9773426115541275e-05, "loss": 1.0769, "step": 42983 }, { "epoch": 0.04, "learning_rate": 4.977341556433701e-05, "loss": 1.2595, "step": 42984 }, { "epoch": 0.04, "learning_rate": 4.97734050128882e-05, "loss": 0.8564, "step": 42985 }, { "epoch": 0.04, "learning_rate": 4.977339446119484e-05, "loss": 1.5923, "step": 42986 }, { "epoch": 0.04, "learning_rate": 4.977338390925692e-05, "loss": 1.1656, "step": 42987 }, { "epoch": 0.04, "learning_rate": 4.977337335707445e-05, "loss": 1.0167, "step": 42988 }, { "epoch": 0.04, "learning_rate": 4.977336280464743e-05, "loss": 0.9987, "step": 42989 }, { "epoch": 0.04, "learning_rate": 4.977335225197586e-05, "loss": 0.8149, "step": 42990 }, { "epoch": 0.04, "learning_rate": 4.9773341699059726e-05, "loss": 1.1555, "step": 42991 }, { "epoch": 0.04, "learning_rate": 4.977333114589905e-05, "loss": 1.2343, "step": 42992 }, { "epoch": 0.04, "learning_rate": 4.977332059249382e-05, "loss": 1.0233, "step": 42993 }, { "epoch": 0.04, "learning_rate": 4.977331003884404e-05, "loss": 0.7209, "step": 42994 }, { "epoch": 0.04, "learning_rate": 4.9773299484949704e-05, "loss": 1.1753, "step": 42995 }, { "epoch": 0.04, "learning_rate": 4.977328893081082e-05, "loss": 0.4713, "step": 42996 }, { "epoch": 0.04, "learning_rate": 4.977327837642738e-05, "loss": 0.8025, "step": 42997 }, { "epoch": 0.04, "learning_rate": 4.97732678217994e-05, "loss": 1.0196, "step": 42998 }, { "epoch": 0.04, "learning_rate": 4.977325726692686e-05, "loss": 1.0846, "step": 42999 }, { "epoch": 0.04, "learning_rate": 4.977324671180976e-05, "loss": 1.1446, "step": 43000 }, { "epoch": 0.04, "eval_loss": 1.081432819366455, "eval_runtime": 127.6127, "eval_samples_per_second": 10.853, "eval_steps_per_second": 5.43, "step": 43000 }, { "epoch": 0.04, "learning_rate": 4.977323615644812e-05, "loss": 1.0632, "step": 43001 }, { "epoch": 0.04, "learning_rate": 4.9773225600841935e-05, "loss": 0.4377, "step": 43002 }, { "epoch": 0.04, "learning_rate": 4.9773215044991186e-05, "loss": 0.6026, "step": 43003 }, { "epoch": 0.04, "learning_rate": 4.9773204488895895e-05, "loss": 1.0853, "step": 43004 }, { "epoch": 0.04, "learning_rate": 4.977319393255605e-05, "loss": 1.2184, "step": 43005 }, { "epoch": 0.04, "learning_rate": 4.9773183375971654e-05, "loss": 1.7411, "step": 43006 }, { "epoch": 0.04, "learning_rate": 4.9773172819142703e-05, "loss": 1.2331, "step": 43007 }, { "epoch": 0.04, "learning_rate": 4.9773162262069205e-05, "loss": 1.2726, "step": 43008 }, { "epoch": 0.04, "learning_rate": 4.9773151704751164e-05, "loss": 1.3347, "step": 43009 }, { "epoch": 0.04, "learning_rate": 4.977314114718857e-05, "loss": 1.194, "step": 43010 }, { "epoch": 0.04, "learning_rate": 4.9773130589381423e-05, "loss": 0.7383, "step": 43011 }, { "epoch": 0.04, "learning_rate": 4.9773120031329724e-05, "loss": 1.0086, "step": 43012 }, { "epoch": 0.04, "learning_rate": 4.9773109473033475e-05, "loss": 1.1612, "step": 43013 }, { "epoch": 0.04, "learning_rate": 4.9773098914492685e-05, "loss": 1.1513, "step": 43014 }, { "epoch": 0.04, "learning_rate": 4.977308835570734e-05, "loss": 1.1989, "step": 43015 }, { "epoch": 0.04, "learning_rate": 4.9773077796677445e-05, "loss": 1.1165, "step": 43016 }, { "epoch": 0.04, "learning_rate": 4.9773067237402995e-05, "loss": 1.224, "step": 43017 }, { "epoch": 0.04, "learning_rate": 4.9773056677884e-05, "loss": 1.2164, "step": 43018 }, { "epoch": 0.04, "learning_rate": 4.977304611812046e-05, "loss": 1.2975, "step": 43019 }, { "epoch": 0.04, "learning_rate": 4.977303555811237e-05, "loss": 1.2666, "step": 43020 }, { "epoch": 0.04, "learning_rate": 4.977302499785973e-05, "loss": 1.1102, "step": 43021 }, { "epoch": 0.04, "learning_rate": 4.977301443736254e-05, "loss": 1.0956, "step": 43022 }, { "epoch": 0.04, "learning_rate": 4.97730038766208e-05, "loss": 0.8755, "step": 43023 }, { "epoch": 0.04, "learning_rate": 4.9772993315634516e-05, "loss": 0.6887, "step": 43024 }, { "epoch": 0.04, "learning_rate": 4.977298275440368e-05, "loss": 1.2853, "step": 43025 }, { "epoch": 0.04, "learning_rate": 4.977297219292829e-05, "loss": 0.9895, "step": 43026 }, { "epoch": 0.04, "learning_rate": 4.977296163120836e-05, "loss": 1.1133, "step": 43027 }, { "epoch": 0.04, "learning_rate": 4.977295106924388e-05, "loss": 1.0061, "step": 43028 }, { "epoch": 0.04, "learning_rate": 4.9772940507034846e-05, "loss": 1.1824, "step": 43029 }, { "epoch": 0.04, "learning_rate": 4.977292994458128e-05, "loss": 1.1184, "step": 43030 }, { "epoch": 0.04, "learning_rate": 4.9772919381883155e-05, "loss": 0.907, "step": 43031 }, { "epoch": 0.04, "learning_rate": 4.977290881894048e-05, "loss": 0.9148, "step": 43032 }, { "epoch": 0.04, "learning_rate": 4.977289825575325e-05, "loss": 0.9381, "step": 43033 }, { "epoch": 0.04, "learning_rate": 4.977288769232149e-05, "loss": 0.3105, "step": 43034 }, { "epoch": 0.04, "learning_rate": 4.977287712864517e-05, "loss": 0.3206, "step": 43035 }, { "epoch": 0.04, "learning_rate": 4.9772866564724306e-05, "loss": 0.3952, "step": 43036 }, { "epoch": 0.04, "learning_rate": 4.97728560005589e-05, "loss": 0.4374, "step": 43037 }, { "epoch": 0.04, "learning_rate": 4.977284543614894e-05, "loss": 0.7105, "step": 43038 }, { "epoch": 0.04, "learning_rate": 4.977283487149444e-05, "loss": 0.8225, "step": 43039 }, { "epoch": 0.04, "learning_rate": 4.977282430659539e-05, "loss": 0.3765, "step": 43040 }, { "epoch": 0.04, "learning_rate": 4.977281374145179e-05, "loss": 1.224, "step": 43041 }, { "epoch": 0.04, "learning_rate": 4.9772803176063645e-05, "loss": 1.3456, "step": 43042 }, { "epoch": 0.04, "learning_rate": 4.977279261043095e-05, "loss": 1.075, "step": 43043 }, { "epoch": 0.04, "learning_rate": 4.977278204455371e-05, "loss": 1.2487, "step": 43044 }, { "epoch": 0.04, "learning_rate": 4.977277147843193e-05, "loss": 1.1091, "step": 43045 }, { "epoch": 0.04, "learning_rate": 4.97727609120656e-05, "loss": 0.7878, "step": 43046 }, { "epoch": 0.04, "learning_rate": 4.977275034545472e-05, "loss": 2.1026, "step": 43047 }, { "epoch": 0.04, "learning_rate": 4.97727397785993e-05, "loss": 1.2229, "step": 43048 }, { "epoch": 0.04, "learning_rate": 4.977272921149933e-05, "loss": 1.3747, "step": 43049 }, { "epoch": 0.04, "learning_rate": 4.9772718644154815e-05, "loss": 1.1594, "step": 43050 }, { "epoch": 0.04, "learning_rate": 4.9772708076565755e-05, "loss": 1.1678, "step": 43051 }, { "epoch": 0.04, "learning_rate": 4.9772697508732146e-05, "loss": 0.9367, "step": 43052 }, { "epoch": 0.04, "learning_rate": 4.977268694065399e-05, "loss": 1.0917, "step": 43053 }, { "epoch": 0.04, "learning_rate": 4.9772676372331296e-05, "loss": 1.0954, "step": 43054 }, { "epoch": 0.04, "learning_rate": 4.9772665803764054e-05, "loss": 0.9628, "step": 43055 }, { "epoch": 0.04, "learning_rate": 4.977265523495226e-05, "loss": 1.1162, "step": 43056 }, { "epoch": 0.04, "learning_rate": 4.9772644665895926e-05, "loss": 0.959, "step": 43057 }, { "epoch": 0.04, "learning_rate": 4.977263409659505e-05, "loss": 0.4806, "step": 43058 }, { "epoch": 0.04, "learning_rate": 4.9772623527049624e-05, "loss": 0.9448, "step": 43059 }, { "epoch": 0.04, "learning_rate": 4.977261295725965e-05, "loss": 0.9924, "step": 43060 }, { "epoch": 0.04, "learning_rate": 4.977260238722514e-05, "loss": 1.0128, "step": 43061 }, { "epoch": 0.04, "learning_rate": 4.977259181694608e-05, "loss": 0.981, "step": 43062 }, { "epoch": 0.04, "learning_rate": 4.977258124642248e-05, "loss": 0.8146, "step": 43063 }, { "epoch": 0.04, "learning_rate": 4.9772570675654326e-05, "loss": 1.1378, "step": 43064 }, { "epoch": 0.04, "learning_rate": 4.977256010464164e-05, "loss": 0.979, "step": 43065 }, { "epoch": 0.04, "learning_rate": 4.9772549533384405e-05, "loss": 1.0193, "step": 43066 }, { "epoch": 0.04, "learning_rate": 4.977253896188262e-05, "loss": 0.6144, "step": 43067 }, { "epoch": 0.04, "learning_rate": 4.9772528390136296e-05, "loss": 0.564, "step": 43068 }, { "epoch": 0.04, "learning_rate": 4.977251781814543e-05, "loss": 0.7904, "step": 43069 }, { "epoch": 0.04, "learning_rate": 4.977250724591001e-05, "loss": 1.1383, "step": 43070 }, { "epoch": 0.04, "learning_rate": 4.9772496673430056e-05, "loss": 1.0035, "step": 43071 }, { "epoch": 0.04, "learning_rate": 4.977248610070556e-05, "loss": 1.0959, "step": 43072 }, { "epoch": 0.04, "learning_rate": 4.9772475527736515e-05, "loss": 0.8807, "step": 43073 }, { "epoch": 0.04, "learning_rate": 4.977246495452293e-05, "loss": 0.1096, "step": 43074 }, { "epoch": 0.04, "learning_rate": 4.97724543810648e-05, "loss": 0.0629, "step": 43075 }, { "epoch": 0.04, "learning_rate": 4.977244380736212e-05, "loss": 0.0546, "step": 43076 }, { "epoch": 0.04, "learning_rate": 4.9772433233414906e-05, "loss": 0.0427, "step": 43077 }, { "epoch": 0.04, "learning_rate": 4.977242265922315e-05, "loss": 0.0341, "step": 43078 }, { "epoch": 0.04, "learning_rate": 4.9772412084786844e-05, "loss": 0.0418, "step": 43079 }, { "epoch": 0.04, "learning_rate": 4.9772401510106e-05, "loss": 0.0371, "step": 43080 }, { "epoch": 0.04, "learning_rate": 4.9772390935180614e-05, "loss": 0.0388, "step": 43081 }, { "epoch": 0.04, "learning_rate": 4.977238036001068e-05, "loss": 0.0954, "step": 43082 }, { "epoch": 0.04, "learning_rate": 4.9772369784596204e-05, "loss": 0.0718, "step": 43083 }, { "epoch": 0.04, "learning_rate": 4.977235920893719e-05, "loss": 0.087, "step": 43084 }, { "epoch": 0.04, "learning_rate": 4.977234863303364e-05, "loss": 0.0727, "step": 43085 }, { "epoch": 0.04, "learning_rate": 4.9772338056885535e-05, "loss": 0.6035, "step": 43086 }, { "epoch": 0.04, "learning_rate": 4.9772327480492895e-05, "loss": 1.011, "step": 43087 }, { "epoch": 0.04, "learning_rate": 4.977231690385571e-05, "loss": 0.7542, "step": 43088 }, { "epoch": 0.04, "learning_rate": 4.9772306326973983e-05, "loss": 0.9139, "step": 43089 }, { "epoch": 0.04, "learning_rate": 4.977229574984772e-05, "loss": 1.0007, "step": 43090 }, { "epoch": 0.04, "learning_rate": 4.9772285172476904e-05, "loss": 0.8838, "step": 43091 }, { "epoch": 0.04, "learning_rate": 4.9772274594861556e-05, "loss": 0.9334, "step": 43092 }, { "epoch": 0.04, "learning_rate": 4.9772264017001665e-05, "loss": 1.0065, "step": 43093 }, { "epoch": 0.04, "learning_rate": 4.977225343889723e-05, "loss": 0.9454, "step": 43094 }, { "epoch": 0.04, "learning_rate": 4.977224286054826e-05, "loss": 1.0232, "step": 43095 }, { "epoch": 0.04, "learning_rate": 4.977223228195474e-05, "loss": 1.0004, "step": 43096 }, { "epoch": 0.04, "learning_rate": 4.9772221703116684e-05, "loss": 0.9088, "step": 43097 }, { "epoch": 0.04, "learning_rate": 4.9772211124034085e-05, "loss": 1.4474, "step": 43098 }, { "epoch": 0.04, "learning_rate": 4.9772200544706943e-05, "loss": 1.2021, "step": 43099 }, { "epoch": 0.04, "learning_rate": 4.977218996513526e-05, "loss": 0.8628, "step": 43100 }, { "epoch": 0.04, "learning_rate": 4.977217938531905e-05, "loss": 1.3833, "step": 43101 }, { "epoch": 0.04, "learning_rate": 4.977216880525828e-05, "loss": 1.0591, "step": 43102 }, { "epoch": 0.04, "learning_rate": 4.977215822495298e-05, "loss": 0.8425, "step": 43103 }, { "epoch": 0.04, "learning_rate": 4.977214764440314e-05, "loss": 1.4179, "step": 43104 }, { "epoch": 0.04, "learning_rate": 4.977213706360876e-05, "loss": 0.5969, "step": 43105 }, { "epoch": 0.04, "learning_rate": 4.977212648256984e-05, "loss": 0.8367, "step": 43106 }, { "epoch": 0.04, "learning_rate": 4.9772115901286375e-05, "loss": 1.0126, "step": 43107 }, { "epoch": 0.04, "learning_rate": 4.977210531975838e-05, "loss": 1.0772, "step": 43108 }, { "epoch": 0.04, "learning_rate": 4.977209473798583e-05, "loss": 0.8785, "step": 43109 }, { "epoch": 0.04, "learning_rate": 4.977208415596875e-05, "loss": 0.6918, "step": 43110 }, { "epoch": 0.04, "learning_rate": 4.9772073573707135e-05, "loss": 0.8133, "step": 43111 }, { "epoch": 0.04, "learning_rate": 4.977206299120098e-05, "loss": 1.0763, "step": 43112 }, { "epoch": 0.04, "learning_rate": 4.977205240845027e-05, "loss": 0.9928, "step": 43113 }, { "epoch": 0.04, "learning_rate": 4.977204182545503e-05, "loss": 0.8391, "step": 43114 }, { "epoch": 0.04, "learning_rate": 4.977203124221525e-05, "loss": 1.1488, "step": 43115 }, { "epoch": 0.04, "learning_rate": 4.9772020658730935e-05, "loss": 1.1006, "step": 43116 }, { "epoch": 0.04, "learning_rate": 4.977201007500208e-05, "loss": 0.8371, "step": 43117 }, { "epoch": 0.04, "learning_rate": 4.977199949102869e-05, "loss": 0.9447, "step": 43118 }, { "epoch": 0.04, "learning_rate": 4.977198890681075e-05, "loss": 1.1159, "step": 43119 }, { "epoch": 0.04, "learning_rate": 4.977197832234828e-05, "loss": 1.0058, "step": 43120 }, { "epoch": 0.04, "learning_rate": 4.9771967737641265e-05, "loss": 0.9408, "step": 43121 }, { "epoch": 0.04, "learning_rate": 4.9771957152689716e-05, "loss": 0.8724, "step": 43122 }, { "epoch": 0.04, "learning_rate": 4.977194656749363e-05, "loss": 1.0278, "step": 43123 }, { "epoch": 0.04, "learning_rate": 4.977193598205301e-05, "loss": 0.8365, "step": 43124 }, { "epoch": 0.04, "learning_rate": 4.9771925396367847e-05, "loss": 1.0471, "step": 43125 }, { "epoch": 0.04, "learning_rate": 4.9771914810438144e-05, "loss": 1.0148, "step": 43126 }, { "epoch": 0.04, "learning_rate": 4.97719042242639e-05, "loss": 1.3552, "step": 43127 }, { "epoch": 0.04, "learning_rate": 4.977189363784512e-05, "loss": 1.0987, "step": 43128 }, { "epoch": 0.04, "learning_rate": 4.9771883051181814e-05, "loss": 1.0702, "step": 43129 }, { "epoch": 0.04, "learning_rate": 4.977187246427396e-05, "loss": 0.9085, "step": 43130 }, { "epoch": 0.04, "learning_rate": 4.977186187712156e-05, "loss": 0.6807, "step": 43131 }, { "epoch": 0.04, "learning_rate": 4.9771851289724634e-05, "loss": 0.7956, "step": 43132 }, { "epoch": 0.04, "learning_rate": 4.977184070208317e-05, "loss": 0.838, "step": 43133 }, { "epoch": 0.04, "learning_rate": 4.977183011419717e-05, "loss": 0.9174, "step": 43134 }, { "epoch": 0.04, "learning_rate": 4.977181952606663e-05, "loss": 1.0231, "step": 43135 }, { "epoch": 0.04, "learning_rate": 4.977180893769155e-05, "loss": 0.9268, "step": 43136 }, { "epoch": 0.04, "learning_rate": 4.977179834907194e-05, "loss": 0.8808, "step": 43137 }, { "epoch": 0.04, "learning_rate": 4.977178776020779e-05, "loss": 0.9102, "step": 43138 }, { "epoch": 0.04, "learning_rate": 4.9771777171099104e-05, "loss": 1.0167, "step": 43139 }, { "epoch": 0.04, "learning_rate": 4.9771766581745884e-05, "loss": 1.1593, "step": 43140 }, { "epoch": 0.04, "learning_rate": 4.977175599214812e-05, "loss": 1.0469, "step": 43141 }, { "epoch": 0.04, "learning_rate": 4.9771745402305826e-05, "loss": 1.2328, "step": 43142 }, { "epoch": 0.04, "learning_rate": 4.9771734812218995e-05, "loss": 1.2675, "step": 43143 }, { "epoch": 0.04, "learning_rate": 4.977172422188763e-05, "loss": 1.2475, "step": 43144 }, { "epoch": 0.04, "learning_rate": 4.977171363131172e-05, "loss": 0.9862, "step": 43145 }, { "epoch": 0.04, "learning_rate": 4.977170304049128e-05, "loss": 0.8124, "step": 43146 }, { "epoch": 0.04, "learning_rate": 4.977169244942631e-05, "loss": 1.1512, "step": 43147 }, { "epoch": 0.04, "learning_rate": 4.977168185811679e-05, "loss": 1.0953, "step": 43148 }, { "epoch": 0.04, "learning_rate": 4.977167126656275e-05, "loss": 1.094, "step": 43149 }, { "epoch": 0.04, "learning_rate": 4.9771660674764166e-05, "loss": 1.4725, "step": 43150 }, { "epoch": 0.04, "learning_rate": 4.977165008272104e-05, "loss": 1.3645, "step": 43151 }, { "epoch": 0.04, "learning_rate": 4.9771639490433394e-05, "loss": 1.1471, "step": 43152 }, { "epoch": 0.04, "learning_rate": 4.97716288979012e-05, "loss": 1.0884, "step": 43153 }, { "epoch": 0.04, "learning_rate": 4.977161830512448e-05, "loss": 1.1766, "step": 43154 }, { "epoch": 0.04, "learning_rate": 4.977160771210322e-05, "loss": 1.1144, "step": 43155 }, { "epoch": 0.04, "learning_rate": 4.977159711883743e-05, "loss": 0.8863, "step": 43156 }, { "epoch": 0.04, "learning_rate": 4.97715865253271e-05, "loss": 0.814, "step": 43157 }, { "epoch": 0.04, "learning_rate": 4.977157593157224e-05, "loss": 0.9863, "step": 43158 }, { "epoch": 0.04, "learning_rate": 4.977156533757284e-05, "loss": 1.6175, "step": 43159 }, { "epoch": 0.04, "learning_rate": 4.9771554743328905e-05, "loss": 1.2704, "step": 43160 }, { "epoch": 0.04, "learning_rate": 4.977154414884044e-05, "loss": 1.2589, "step": 43161 }, { "epoch": 0.04, "learning_rate": 4.977153355410744e-05, "loss": 1.2298, "step": 43162 }, { "epoch": 0.04, "learning_rate": 4.977152295912991e-05, "loss": 1.3787, "step": 43163 }, { "epoch": 0.04, "learning_rate": 4.977151236390783e-05, "loss": 1.3231, "step": 43164 }, { "epoch": 0.04, "learning_rate": 4.977150176844124e-05, "loss": 1.4596, "step": 43165 }, { "epoch": 0.04, "learning_rate": 4.97714911727301e-05, "loss": 1.1637, "step": 43166 }, { "epoch": 0.04, "learning_rate": 4.977148057677443e-05, "loss": 1.2652, "step": 43167 }, { "epoch": 0.04, "learning_rate": 4.977146998057423e-05, "loss": 1.2693, "step": 43168 }, { "epoch": 0.04, "learning_rate": 4.9771459384129486e-05, "loss": 0.806, "step": 43169 }, { "epoch": 0.04, "learning_rate": 4.9771448787440216e-05, "loss": 1.3163, "step": 43170 }, { "epoch": 0.04, "learning_rate": 4.977143819050642e-05, "loss": 1.1406, "step": 43171 }, { "epoch": 0.04, "learning_rate": 4.977142759332808e-05, "loss": 1.074, "step": 43172 }, { "epoch": 0.04, "learning_rate": 4.977141699590521e-05, "loss": 0.8529, "step": 43173 }, { "epoch": 0.04, "learning_rate": 4.9771406398237804e-05, "loss": 1.0116, "step": 43174 }, { "epoch": 0.04, "learning_rate": 4.9771395800325865e-05, "loss": 0.9894, "step": 43175 }, { "epoch": 0.04, "learning_rate": 4.97713852021694e-05, "loss": 1.4412, "step": 43176 }, { "epoch": 0.04, "learning_rate": 4.9771374603768396e-05, "loss": 1.3144, "step": 43177 }, { "epoch": 0.04, "learning_rate": 4.977136400512286e-05, "loss": 1.7968, "step": 43178 }, { "epoch": 0.04, "learning_rate": 4.977135340623279e-05, "loss": 0.9951, "step": 43179 }, { "epoch": 0.04, "learning_rate": 4.97713428070982e-05, "loss": 0.6374, "step": 43180 }, { "epoch": 0.04, "learning_rate": 4.9771332207719065e-05, "loss": 1.1862, "step": 43181 }, { "epoch": 0.04, "learning_rate": 4.97713216080954e-05, "loss": 0.976, "step": 43182 }, { "epoch": 0.04, "learning_rate": 4.97713110082272e-05, "loss": 1.0794, "step": 43183 }, { "epoch": 0.04, "learning_rate": 4.9771300408114476e-05, "loss": 1.2781, "step": 43184 }, { "epoch": 0.04, "learning_rate": 4.9771289807757214e-05, "loss": 1.313, "step": 43185 }, { "epoch": 0.04, "learning_rate": 4.977127920715543e-05, "loss": 1.5829, "step": 43186 }, { "epoch": 0.04, "learning_rate": 4.9771268606309106e-05, "loss": 0.7633, "step": 43187 }, { "epoch": 0.04, "learning_rate": 4.9771258005218246e-05, "loss": 0.9103, "step": 43188 }, { "epoch": 0.04, "learning_rate": 4.9771247403882864e-05, "loss": 1.0869, "step": 43189 }, { "epoch": 0.04, "learning_rate": 4.977123680230295e-05, "loss": 0.9666, "step": 43190 }, { "epoch": 0.04, "learning_rate": 4.97712262004785e-05, "loss": 1.2739, "step": 43191 }, { "epoch": 0.04, "learning_rate": 4.977121559840952e-05, "loss": 1.2772, "step": 43192 }, { "epoch": 0.04, "learning_rate": 4.977120499609601e-05, "loss": 0.9195, "step": 43193 }, { "epoch": 0.04, "learning_rate": 4.977119439353797e-05, "loss": 1.1136, "step": 43194 }, { "epoch": 0.04, "learning_rate": 4.97711837907354e-05, "loss": 1.842, "step": 43195 }, { "epoch": 0.04, "learning_rate": 4.977117318768829e-05, "loss": 0.9213, "step": 43196 }, { "epoch": 0.04, "learning_rate": 4.977116258439666e-05, "loss": 0.9348, "step": 43197 }, { "epoch": 0.04, "learning_rate": 4.9771151980860495e-05, "loss": 1.0133, "step": 43198 }, { "epoch": 0.04, "learning_rate": 4.97711413770798e-05, "loss": 1.0242, "step": 43199 }, { "epoch": 0.04, "learning_rate": 4.977113077305458e-05, "loss": 1.3006, "step": 43200 }, { "epoch": 0.04, "learning_rate": 4.977112016878482e-05, "loss": 1.1613, "step": 43201 }, { "epoch": 0.04, "learning_rate": 4.9771109564270534e-05, "loss": 0.9065, "step": 43202 }, { "epoch": 0.04, "learning_rate": 4.977109895951172e-05, "loss": 0.7703, "step": 43203 }, { "epoch": 0.04, "learning_rate": 4.9771088354508377e-05, "loss": 1.0133, "step": 43204 }, { "epoch": 0.04, "learning_rate": 4.977107774926051e-05, "loss": 0.9613, "step": 43205 }, { "epoch": 0.04, "learning_rate": 4.97710671437681e-05, "loss": 1.4714, "step": 43206 }, { "epoch": 0.04, "learning_rate": 4.977105653803117e-05, "loss": 0.8552, "step": 43207 }, { "epoch": 0.04, "learning_rate": 4.977104593204971e-05, "loss": 0.7925, "step": 43208 }, { "epoch": 0.04, "learning_rate": 4.977103532582371e-05, "loss": 0.6932, "step": 43209 }, { "epoch": 0.04, "learning_rate": 4.9771024719353194e-05, "loss": 0.8306, "step": 43210 }, { "epoch": 0.04, "learning_rate": 4.977101411263814e-05, "loss": 1.0674, "step": 43211 }, { "epoch": 0.04, "learning_rate": 4.977100350567856e-05, "loss": 0.8299, "step": 43212 }, { "epoch": 0.04, "learning_rate": 4.977099289847446e-05, "loss": 0.9263, "step": 43213 }, { "epoch": 0.04, "learning_rate": 4.9770982291025815e-05, "loss": 1.0585, "step": 43214 }, { "epoch": 0.04, "learning_rate": 4.977097168333265e-05, "loss": 1.1923, "step": 43215 }, { "epoch": 0.04, "learning_rate": 4.977096107539495e-05, "loss": 1.0391, "step": 43216 }, { "epoch": 0.04, "learning_rate": 4.977095046721273e-05, "loss": 0.7899, "step": 43217 }, { "epoch": 0.04, "learning_rate": 4.977093985878598e-05, "loss": 0.65, "step": 43218 }, { "epoch": 0.04, "learning_rate": 4.977092925011471e-05, "loss": 0.5966, "step": 43219 }, { "epoch": 0.04, "learning_rate": 4.97709186411989e-05, "loss": 0.8533, "step": 43220 }, { "epoch": 0.04, "learning_rate": 4.977090803203856e-05, "loss": 1.1431, "step": 43221 }, { "epoch": 0.04, "learning_rate": 4.9770897422633696e-05, "loss": 1.0611, "step": 43222 }, { "epoch": 0.04, "learning_rate": 4.9770886812984306e-05, "loss": 1.1186, "step": 43223 }, { "epoch": 0.04, "learning_rate": 4.977087620309038e-05, "loss": 0.9337, "step": 43224 }, { "epoch": 0.04, "learning_rate": 4.9770865592951936e-05, "loss": 0.9971, "step": 43225 }, { "epoch": 0.04, "learning_rate": 4.977085498256896e-05, "loss": 1.128, "step": 43226 }, { "epoch": 0.04, "learning_rate": 4.977084437194146e-05, "loss": 1.1355, "step": 43227 }, { "epoch": 0.04, "learning_rate": 4.977083376106943e-05, "loss": 1.0533, "step": 43228 }, { "epoch": 0.04, "learning_rate": 4.977082314995287e-05, "loss": 1.0487, "step": 43229 }, { "epoch": 0.04, "learning_rate": 4.9770812538591796e-05, "loss": 0.9796, "step": 43230 }, { "epoch": 0.04, "learning_rate": 4.977080192698618e-05, "loss": 0.9609, "step": 43231 }, { "epoch": 0.04, "learning_rate": 4.977079131513604e-05, "loss": 0.9599, "step": 43232 }, { "epoch": 0.04, "learning_rate": 4.977078070304138e-05, "loss": 1.0132, "step": 43233 }, { "epoch": 0.04, "learning_rate": 4.977077009070219e-05, "loss": 1.0655, "step": 43234 }, { "epoch": 0.04, "learning_rate": 4.977075947811847e-05, "loss": 1.0568, "step": 43235 }, { "epoch": 0.04, "learning_rate": 4.9770748865290226e-05, "loss": 1.0287, "step": 43236 }, { "epoch": 0.04, "learning_rate": 4.9770738252217455e-05, "loss": 1.0251, "step": 43237 }, { "epoch": 0.04, "learning_rate": 4.9770727638900155e-05, "loss": 0.9889, "step": 43238 }, { "epoch": 0.04, "learning_rate": 4.9770717025338334e-05, "loss": 0.9926, "step": 43239 }, { "epoch": 0.04, "learning_rate": 4.9770706411531985e-05, "loss": 0.9962, "step": 43240 }, { "epoch": 0.04, "learning_rate": 4.977069579748111e-05, "loss": 1.0461, "step": 43241 }, { "epoch": 0.04, "learning_rate": 4.977068518318571e-05, "loss": 0.9808, "step": 43242 }, { "epoch": 0.04, "learning_rate": 4.977067456864578e-05, "loss": 0.963, "step": 43243 }, { "epoch": 0.04, "learning_rate": 4.977066395386133e-05, "loss": 0.9551, "step": 43244 }, { "epoch": 0.04, "learning_rate": 4.9770653338832355e-05, "loss": 1.0921, "step": 43245 }, { "epoch": 0.04, "learning_rate": 4.977064272355885e-05, "loss": 1.1196, "step": 43246 }, { "epoch": 0.04, "learning_rate": 4.9770632108040816e-05, "loss": 1.1508, "step": 43247 }, { "epoch": 0.04, "learning_rate": 4.977062149227826e-05, "loss": 0.6406, "step": 43248 }, { "epoch": 0.04, "learning_rate": 4.9770610876271184e-05, "loss": 0.1922, "step": 43249 }, { "epoch": 0.04, "learning_rate": 4.977060026001958e-05, "loss": 1.0864, "step": 43250 }, { "epoch": 0.04, "learning_rate": 4.977058964352345e-05, "loss": 0.8838, "step": 43251 }, { "epoch": 0.04, "learning_rate": 4.977057902678279e-05, "loss": 0.8094, "step": 43252 }, { "epoch": 0.04, "learning_rate": 4.9770568409797616e-05, "loss": 1.1516, "step": 43253 }, { "epoch": 0.04, "learning_rate": 4.977055779256792e-05, "loss": 0.7887, "step": 43254 }, { "epoch": 0.04, "learning_rate": 4.9770547175093686e-05, "loss": 0.6643, "step": 43255 }, { "epoch": 0.04, "learning_rate": 4.977053655737494e-05, "loss": 1.5653, "step": 43256 }, { "epoch": 0.04, "learning_rate": 4.9770525939411664e-05, "loss": 1.1032, "step": 43257 }, { "epoch": 0.04, "learning_rate": 4.977051532120386e-05, "loss": 1.0197, "step": 43258 }, { "epoch": 0.04, "learning_rate": 4.977050470275153e-05, "loss": 0.8592, "step": 43259 }, { "epoch": 0.04, "learning_rate": 4.977049408405469e-05, "loss": 1.1767, "step": 43260 }, { "epoch": 0.04, "learning_rate": 4.977048346511331e-05, "loss": 1.082, "step": 43261 }, { "epoch": 0.04, "learning_rate": 4.977047284592741e-05, "loss": 0.8171, "step": 43262 }, { "epoch": 0.04, "learning_rate": 4.977046222649699e-05, "loss": 1.088, "step": 43263 }, { "epoch": 0.04, "learning_rate": 4.977045160682205e-05, "loss": 1.1417, "step": 43264 }, { "epoch": 0.04, "learning_rate": 4.977044098690258e-05, "loss": 1.0392, "step": 43265 }, { "epoch": 0.04, "learning_rate": 4.977043036673859e-05, "loss": 1.0419, "step": 43266 }, { "epoch": 0.04, "learning_rate": 4.977041974633008e-05, "loss": 0.9271, "step": 43267 }, { "epoch": 0.04, "learning_rate": 4.977040912567704e-05, "loss": 1.0858, "step": 43268 }, { "epoch": 0.04, "learning_rate": 4.977039850477948e-05, "loss": 1.3078, "step": 43269 }, { "epoch": 0.04, "learning_rate": 4.97703878836374e-05, "loss": 2.2399, "step": 43270 }, { "epoch": 0.04, "learning_rate": 4.977037726225079e-05, "loss": 0.9992, "step": 43271 }, { "epoch": 0.04, "learning_rate": 4.977036664061966e-05, "loss": 0.8268, "step": 43272 }, { "epoch": 0.04, "learning_rate": 4.977035601874401e-05, "loss": 1.0382, "step": 43273 }, { "epoch": 0.04, "learning_rate": 4.977034539662384e-05, "loss": 0.765, "step": 43274 }, { "epoch": 0.04, "learning_rate": 4.977033477425914e-05, "loss": 1.0939, "step": 43275 }, { "epoch": 0.04, "learning_rate": 4.9770324151649925e-05, "loss": 1.3437, "step": 43276 }, { "epoch": 0.04, "learning_rate": 4.9770313528796187e-05, "loss": 0.8166, "step": 43277 }, { "epoch": 0.04, "learning_rate": 4.977030290569792e-05, "loss": 0.9506, "step": 43278 }, { "epoch": 0.04, "learning_rate": 4.977029228235514e-05, "loss": 0.7113, "step": 43279 }, { "epoch": 0.04, "learning_rate": 4.977028165876783e-05, "loss": 1.0758, "step": 43280 }, { "epoch": 0.04, "learning_rate": 4.9770271034936004e-05, "loss": 1.4441, "step": 43281 }, { "epoch": 0.04, "learning_rate": 4.977026041085965e-05, "loss": 1.0357, "step": 43282 }, { "epoch": 0.04, "learning_rate": 4.977024978653878e-05, "loss": 0.8678, "step": 43283 }, { "epoch": 0.04, "learning_rate": 4.977023916197339e-05, "loss": 1.084, "step": 43284 }, { "epoch": 0.04, "learning_rate": 4.977022853716347e-05, "loss": 0.8571, "step": 43285 }, { "epoch": 0.04, "learning_rate": 4.977021791210904e-05, "loss": 1.184, "step": 43286 }, { "epoch": 0.04, "learning_rate": 4.977020728681008e-05, "loss": 1.2234, "step": 43287 }, { "epoch": 0.04, "learning_rate": 4.9770196661266596e-05, "loss": 1.2278, "step": 43288 }, { "epoch": 0.04, "learning_rate": 4.977018603547861e-05, "loss": 1.3374, "step": 43289 }, { "epoch": 0.04, "learning_rate": 4.977017540944608e-05, "loss": 1.1755, "step": 43290 }, { "epoch": 0.04, "learning_rate": 4.977016478316904e-05, "loss": 0.9307, "step": 43291 }, { "epoch": 0.04, "learning_rate": 4.977015415664748e-05, "loss": 0.9489, "step": 43292 }, { "epoch": 0.04, "learning_rate": 4.97701435298814e-05, "loss": 1.4799, "step": 43293 }, { "epoch": 0.04, "learning_rate": 4.9770132902870796e-05, "loss": 0.7257, "step": 43294 }, { "epoch": 0.04, "learning_rate": 4.977012227561567e-05, "loss": 1.0426, "step": 43295 }, { "epoch": 0.04, "learning_rate": 4.977011164811604e-05, "loss": 0.6681, "step": 43296 }, { "epoch": 0.04, "learning_rate": 4.977010102037187e-05, "loss": 0.3359, "step": 43297 }, { "epoch": 0.04, "learning_rate": 4.977009039238319e-05, "loss": 0.1901, "step": 43298 }, { "epoch": 0.04, "learning_rate": 4.977007976414998e-05, "loss": 0.3593, "step": 43299 }, { "epoch": 0.04, "learning_rate": 4.977006913567226e-05, "loss": 0.7129, "step": 43300 }, { "epoch": 0.04, "learning_rate": 4.9770058506950023e-05, "loss": 0.5447, "step": 43301 }, { "epoch": 0.04, "learning_rate": 4.977004787798326e-05, "loss": 0.9973, "step": 43302 }, { "epoch": 0.04, "learning_rate": 4.977003724877198e-05, "loss": 1.0612, "step": 43303 }, { "epoch": 0.04, "learning_rate": 4.977002661931618e-05, "loss": 1.0703, "step": 43304 }, { "epoch": 0.04, "learning_rate": 4.977001598961586e-05, "loss": 1.2384, "step": 43305 }, { "epoch": 0.04, "learning_rate": 4.977000535967102e-05, "loss": 0.6851, "step": 43306 }, { "epoch": 0.04, "learning_rate": 4.9769994729481664e-05, "loss": 0.9465, "step": 43307 }, { "epoch": 0.04, "learning_rate": 4.9769984099047787e-05, "loss": 1.1582, "step": 43308 }, { "epoch": 0.04, "learning_rate": 4.976997346836939e-05, "loss": 0.9706, "step": 43309 }, { "epoch": 0.04, "learning_rate": 4.976996283744647e-05, "loss": 0.9789, "step": 43310 }, { "epoch": 0.04, "learning_rate": 4.976995220627904e-05, "loss": 0.8408, "step": 43311 }, { "epoch": 0.04, "learning_rate": 4.976994157486709e-05, "loss": 1.3441, "step": 43312 }, { "epoch": 0.04, "learning_rate": 4.976993094321062e-05, "loss": 1.5034, "step": 43313 }, { "epoch": 0.04, "learning_rate": 4.9769920311309634e-05, "loss": 1.7258, "step": 43314 }, { "epoch": 0.04, "learning_rate": 4.9769909679164126e-05, "loss": 0.7056, "step": 43315 }, { "epoch": 0.04, "learning_rate": 4.97698990467741e-05, "loss": 0.7369, "step": 43316 }, { "epoch": 0.04, "learning_rate": 4.9769888414139554e-05, "loss": 0.8738, "step": 43317 }, { "epoch": 0.04, "learning_rate": 4.9769877781260496e-05, "loss": 1.2305, "step": 43318 }, { "epoch": 0.04, "learning_rate": 4.976986714813692e-05, "loss": 1.1426, "step": 43319 }, { "epoch": 0.04, "learning_rate": 4.976985651476882e-05, "loss": 1.2824, "step": 43320 }, { "epoch": 0.04, "learning_rate": 4.976984588115621e-05, "loss": 1.0438, "step": 43321 }, { "epoch": 0.04, "learning_rate": 4.976983524729908e-05, "loss": 0.9252, "step": 43322 }, { "epoch": 0.04, "learning_rate": 4.976982461319743e-05, "loss": 0.9712, "step": 43323 }, { "epoch": 0.04, "learning_rate": 4.9769813978851265e-05, "loss": 0.9553, "step": 43324 }, { "epoch": 0.04, "learning_rate": 4.976980334426058e-05, "loss": 1.0424, "step": 43325 }, { "epoch": 0.04, "learning_rate": 4.976979270942538e-05, "loss": 1.306, "step": 43326 }, { "epoch": 0.04, "learning_rate": 4.9769782074345664e-05, "loss": 1.1726, "step": 43327 }, { "epoch": 0.04, "learning_rate": 4.976977143902143e-05, "loss": 0.8769, "step": 43328 }, { "epoch": 0.04, "learning_rate": 4.976976080345268e-05, "loss": 1.0808, "step": 43329 }, { "epoch": 0.04, "learning_rate": 4.976975016763941e-05, "loss": 0.929, "step": 43330 }, { "epoch": 0.04, "learning_rate": 4.976973953158163e-05, "loss": 1.1078, "step": 43331 }, { "epoch": 0.04, "learning_rate": 4.976972889527933e-05, "loss": 1.4006, "step": 43332 }, { "epoch": 0.04, "learning_rate": 4.976971825873251e-05, "loss": 0.9467, "step": 43333 }, { "epoch": 0.04, "learning_rate": 4.9769707621941177e-05, "loss": 1.0898, "step": 43334 }, { "epoch": 0.04, "learning_rate": 4.9769696984905335e-05, "loss": 0.7099, "step": 43335 }, { "epoch": 0.04, "learning_rate": 4.9769686347624965e-05, "loss": 0.9711, "step": 43336 }, { "epoch": 0.04, "learning_rate": 4.976967571010008e-05, "loss": 0.9778, "step": 43337 }, { "epoch": 0.04, "learning_rate": 4.976966507233069e-05, "loss": 1.0051, "step": 43338 }, { "epoch": 0.04, "learning_rate": 4.9769654434316774e-05, "loss": 1.1141, "step": 43339 }, { "epoch": 0.04, "learning_rate": 4.976964379605835e-05, "loss": 1.071, "step": 43340 }, { "epoch": 0.04, "learning_rate": 4.976963315755541e-05, "loss": 1.4159, "step": 43341 }, { "epoch": 0.04, "learning_rate": 4.976962251880794e-05, "loss": 0.6071, "step": 43342 }, { "epoch": 0.04, "learning_rate": 4.976961187981597e-05, "loss": 1.0446, "step": 43343 }, { "epoch": 0.04, "learning_rate": 4.976960124057948e-05, "loss": 0.7221, "step": 43344 }, { "epoch": 0.04, "learning_rate": 4.976959060109848e-05, "loss": 1.1228, "step": 43345 }, { "epoch": 0.04, "learning_rate": 4.976957996137296e-05, "loss": 1.3737, "step": 43346 }, { "epoch": 0.04, "learning_rate": 4.9769569321402925e-05, "loss": 0.8423, "step": 43347 }, { "epoch": 0.04, "learning_rate": 4.976955868118838e-05, "loss": 0.9873, "step": 43348 }, { "epoch": 0.04, "learning_rate": 4.976954804072931e-05, "loss": 0.9554, "step": 43349 }, { "epoch": 0.04, "learning_rate": 4.9769537400025736e-05, "loss": 1.0356, "step": 43350 }, { "epoch": 0.04, "learning_rate": 4.9769526759077645e-05, "loss": 0.9013, "step": 43351 }, { "epoch": 0.04, "learning_rate": 4.976951611788504e-05, "loss": 1.4114, "step": 43352 }, { "epoch": 0.04, "learning_rate": 4.9769505476447914e-05, "loss": 1.1707, "step": 43353 }, { "epoch": 0.04, "learning_rate": 4.9769494834766286e-05, "loss": 0.8104, "step": 43354 }, { "epoch": 0.04, "learning_rate": 4.976948419284013e-05, "loss": 0.8379, "step": 43355 }, { "epoch": 0.04, "learning_rate": 4.9769473550669466e-05, "loss": 0.3183, "step": 43356 }, { "epoch": 0.04, "learning_rate": 4.9769462908254294e-05, "loss": 0.4001, "step": 43357 }, { "epoch": 0.04, "learning_rate": 4.97694522655946e-05, "loss": 0.3442, "step": 43358 }, { "epoch": 0.04, "learning_rate": 4.976944162269039e-05, "loss": 0.3433, "step": 43359 }, { "epoch": 0.04, "learning_rate": 4.976943097954168e-05, "loss": 0.9359, "step": 43360 }, { "epoch": 0.04, "learning_rate": 4.976942033614845e-05, "loss": 0.8109, "step": 43361 }, { "epoch": 0.04, "learning_rate": 4.97694096925107e-05, "loss": 0.1944, "step": 43362 }, { "epoch": 0.04, "learning_rate": 4.976939904862845e-05, "loss": 1.257, "step": 43363 }, { "epoch": 0.04, "learning_rate": 4.9769388404501674e-05, "loss": 1.0825, "step": 43364 }, { "epoch": 0.04, "learning_rate": 4.976937776013039e-05, "loss": 0.8951, "step": 43365 }, { "epoch": 0.04, "learning_rate": 4.97693671155146e-05, "loss": 0.5043, "step": 43366 }, { "epoch": 0.04, "learning_rate": 4.976935647065428e-05, "loss": 0.3918, "step": 43367 }, { "epoch": 0.04, "learning_rate": 4.976934582554946e-05, "loss": 0.8302, "step": 43368 }, { "epoch": 0.04, "learning_rate": 4.976933518020013e-05, "loss": 1.0176, "step": 43369 }, { "epoch": 0.04, "learning_rate": 4.9769324534606285e-05, "loss": 1.0758, "step": 43370 }, { "epoch": 0.04, "learning_rate": 4.976931388876792e-05, "loss": 1.1982, "step": 43371 }, { "epoch": 0.04, "learning_rate": 4.976930324268505e-05, "loss": 0.7094, "step": 43372 }, { "epoch": 0.04, "learning_rate": 4.976929259635767e-05, "loss": 1.0813, "step": 43373 }, { "epoch": 0.04, "learning_rate": 4.9769281949785775e-05, "loss": 0.8183, "step": 43374 }, { "epoch": 0.04, "learning_rate": 4.9769271302969366e-05, "loss": 1.0532, "step": 43375 }, { "epoch": 0.04, "learning_rate": 4.976926065590844e-05, "loss": 1.064, "step": 43376 }, { "epoch": 0.04, "learning_rate": 4.976925000860302e-05, "loss": 1.0882, "step": 43377 }, { "epoch": 0.04, "learning_rate": 4.976923936105308e-05, "loss": 1.0283, "step": 43378 }, { "epoch": 0.04, "learning_rate": 4.976922871325862e-05, "loss": 0.874, "step": 43379 }, { "epoch": 0.04, "learning_rate": 4.976921806521966e-05, "loss": 1.1968, "step": 43380 }, { "epoch": 0.04, "learning_rate": 4.976920741693618e-05, "loss": 1.4656, "step": 43381 }, { "epoch": 0.04, "learning_rate": 4.976919676840819e-05, "loss": 1.0594, "step": 43382 }, { "epoch": 0.04, "learning_rate": 4.976918611963569e-05, "loss": 1.0304, "step": 43383 }, { "epoch": 0.04, "learning_rate": 4.976917547061868e-05, "loss": 1.1944, "step": 43384 }, { "epoch": 0.04, "learning_rate": 4.976916482135716e-05, "loss": 1.0599, "step": 43385 }, { "epoch": 0.04, "learning_rate": 4.976915417185113e-05, "loss": 1.0202, "step": 43386 }, { "epoch": 0.04, "learning_rate": 4.9769143522100594e-05, "loss": 0.9462, "step": 43387 }, { "epoch": 0.04, "learning_rate": 4.976913287210554e-05, "loss": 0.5728, "step": 43388 }, { "epoch": 0.04, "learning_rate": 4.9769122221865974e-05, "loss": 0.9993, "step": 43389 }, { "epoch": 0.04, "learning_rate": 4.97691115713819e-05, "loss": 0.9858, "step": 43390 }, { "epoch": 0.04, "learning_rate": 4.9769100920653315e-05, "loss": 0.8867, "step": 43391 }, { "epoch": 0.04, "learning_rate": 4.976909026968022e-05, "loss": 0.9865, "step": 43392 }, { "epoch": 0.04, "learning_rate": 4.9769079618462624e-05, "loss": 0.9799, "step": 43393 }, { "epoch": 0.04, "learning_rate": 4.9769068967000504e-05, "loss": 0.9504, "step": 43394 }, { "epoch": 0.04, "learning_rate": 4.976905831529388e-05, "loss": 1.1387, "step": 43395 }, { "epoch": 0.04, "learning_rate": 4.976904766334275e-05, "loss": 1.1086, "step": 43396 }, { "epoch": 0.04, "learning_rate": 4.97690370111471e-05, "loss": 1.3523, "step": 43397 }, { "epoch": 0.04, "learning_rate": 4.976902635870695e-05, "loss": 1.1629, "step": 43398 }, { "epoch": 0.04, "learning_rate": 4.9769015706022295e-05, "loss": 1.1671, "step": 43399 }, { "epoch": 0.04, "learning_rate": 4.976900505309312e-05, "loss": 0.9602, "step": 43400 }, { "epoch": 0.04, "learning_rate": 4.976899439991944e-05, "loss": 1.4438, "step": 43401 }, { "epoch": 0.04, "learning_rate": 4.9768983746501254e-05, "loss": 1.0066, "step": 43402 }, { "epoch": 0.04, "learning_rate": 4.976897309283855e-05, "loss": 1.039, "step": 43403 }, { "epoch": 0.04, "learning_rate": 4.976896243893134e-05, "loss": 0.8197, "step": 43404 }, { "epoch": 0.04, "learning_rate": 4.976895178477963e-05, "loss": 1.0916, "step": 43405 }, { "epoch": 0.04, "learning_rate": 4.9768941130383405e-05, "loss": 0.9929, "step": 43406 }, { "epoch": 0.04, "learning_rate": 4.976893047574267e-05, "loss": 1.2856, "step": 43407 }, { "epoch": 0.04, "learning_rate": 4.976891982085743e-05, "loss": 1.1546, "step": 43408 }, { "epoch": 0.04, "learning_rate": 4.9768909165727676e-05, "loss": 1.39, "step": 43409 }, { "epoch": 0.04, "learning_rate": 4.976889851035342e-05, "loss": 1.222, "step": 43410 }, { "epoch": 0.04, "learning_rate": 4.976888785473466e-05, "loss": 1.266, "step": 43411 }, { "epoch": 0.04, "learning_rate": 4.976887719887138e-05, "loss": 0.9845, "step": 43412 }, { "epoch": 0.04, "learning_rate": 4.97688665427636e-05, "loss": 1.0552, "step": 43413 }, { "epoch": 0.04, "learning_rate": 4.9768855886411306e-05, "loss": 0.9542, "step": 43414 }, { "epoch": 0.04, "learning_rate": 4.9768845229814506e-05, "loss": 1.2924, "step": 43415 }, { "epoch": 0.04, "learning_rate": 4.9768834572973204e-05, "loss": 1.0125, "step": 43416 }, { "epoch": 0.04, "learning_rate": 4.9768823915887395e-05, "loss": 0.9609, "step": 43417 }, { "epoch": 0.04, "learning_rate": 4.976881325855707e-05, "loss": 1.1428, "step": 43418 }, { "epoch": 0.04, "learning_rate": 4.9768802600982245e-05, "loss": 1.4471, "step": 43419 }, { "epoch": 0.04, "learning_rate": 4.9768791943162905e-05, "loss": 2.2395, "step": 43420 }, { "epoch": 0.04, "learning_rate": 4.976878128509907e-05, "loss": 1.9473, "step": 43421 }, { "epoch": 0.04, "learning_rate": 4.9768770626790715e-05, "loss": 1.3357, "step": 43422 }, { "epoch": 0.04, "learning_rate": 4.9768759968237865e-05, "loss": 1.2185, "step": 43423 }, { "epoch": 0.04, "learning_rate": 4.97687493094405e-05, "loss": 1.1548, "step": 43424 }, { "epoch": 0.04, "learning_rate": 4.9768738650398633e-05, "loss": 1.154, "step": 43425 }, { "epoch": 0.04, "learning_rate": 4.976872799111226e-05, "loss": 1.3293, "step": 43426 }, { "epoch": 0.04, "learning_rate": 4.976871733158137e-05, "loss": 1.158, "step": 43427 }, { "epoch": 0.04, "learning_rate": 4.976870667180599e-05, "loss": 0.7104, "step": 43428 }, { "epoch": 0.04, "learning_rate": 4.976869601178609e-05, "loss": 0.9788, "step": 43429 }, { "epoch": 0.04, "learning_rate": 4.976868535152169e-05, "loss": 0.8718, "step": 43430 }, { "epoch": 0.04, "learning_rate": 4.9768674691012784e-05, "loss": 0.948, "step": 43431 }, { "epoch": 0.04, "learning_rate": 4.976866403025937e-05, "loss": 1.1247, "step": 43432 }, { "epoch": 0.04, "learning_rate": 4.9768653369261453e-05, "loss": 1.0863, "step": 43433 }, { "epoch": 0.04, "learning_rate": 4.976864270801903e-05, "loss": 0.8544, "step": 43434 }, { "epoch": 0.04, "learning_rate": 4.9768632046532105e-05, "loss": 0.7553, "step": 43435 }, { "epoch": 0.04, "learning_rate": 4.9768621384800665e-05, "loss": 0.9312, "step": 43436 }, { "epoch": 0.04, "learning_rate": 4.9768610722824724e-05, "loss": 0.8407, "step": 43437 }, { "epoch": 0.04, "learning_rate": 4.9768600060604275e-05, "loss": 0.7668, "step": 43438 }, { "epoch": 0.04, "learning_rate": 4.976858939813933e-05, "loss": 1.123, "step": 43439 }, { "epoch": 0.04, "learning_rate": 4.9768578735429875e-05, "loss": 0.7703, "step": 43440 }, { "epoch": 0.04, "learning_rate": 4.976856807247591e-05, "loss": 0.8329, "step": 43441 }, { "epoch": 0.04, "learning_rate": 4.976855740927744e-05, "loss": 1.2106, "step": 43442 }, { "epoch": 0.04, "learning_rate": 4.9768546745834474e-05, "loss": 1.2767, "step": 43443 }, { "epoch": 0.04, "learning_rate": 4.9768536082147e-05, "loss": 0.8608, "step": 43444 }, { "epoch": 0.04, "learning_rate": 4.9768525418215014e-05, "loss": 1.0229, "step": 43445 }, { "epoch": 0.04, "learning_rate": 4.976851475403853e-05, "loss": 1.0519, "step": 43446 }, { "epoch": 0.04, "learning_rate": 4.976850408961754e-05, "loss": 1.0483, "step": 43447 }, { "epoch": 0.04, "learning_rate": 4.976849342495205e-05, "loss": 1.0903, "step": 43448 }, { "epoch": 0.04, "learning_rate": 4.9768482760042054e-05, "loss": 0.9796, "step": 43449 }, { "epoch": 0.04, "learning_rate": 4.976847209488755e-05, "loss": 0.9773, "step": 43450 }, { "epoch": 0.04, "learning_rate": 4.976846142948855e-05, "loss": 1.3094, "step": 43451 }, { "epoch": 0.04, "learning_rate": 4.976845076384504e-05, "loss": 0.8922, "step": 43452 }, { "epoch": 0.04, "learning_rate": 4.976844009795703e-05, "loss": 1.1636, "step": 43453 }, { "epoch": 0.04, "learning_rate": 4.976842943182451e-05, "loss": 1.0273, "step": 43454 }, { "epoch": 0.04, "learning_rate": 4.976841876544749e-05, "loss": 1.0554, "step": 43455 }, { "epoch": 0.04, "learning_rate": 4.976840809882597e-05, "loss": 0.7267, "step": 43456 }, { "epoch": 0.04, "learning_rate": 4.976839743195994e-05, "loss": 0.7869, "step": 43457 }, { "epoch": 0.04, "learning_rate": 4.976838676484941e-05, "loss": 0.9203, "step": 43458 }, { "epoch": 0.04, "learning_rate": 4.976837609749438e-05, "loss": 0.9117, "step": 43459 }, { "epoch": 0.04, "learning_rate": 4.976836542989484e-05, "loss": 1.1166, "step": 43460 }, { "epoch": 0.04, "learning_rate": 4.9768354762050806e-05, "loss": 1.0403, "step": 43461 }, { "epoch": 0.04, "learning_rate": 4.976834409396226e-05, "loss": 1.2187, "step": 43462 }, { "epoch": 0.04, "learning_rate": 4.9768333425629224e-05, "loss": 1.0482, "step": 43463 }, { "epoch": 0.04, "learning_rate": 4.976832275705167e-05, "loss": 1.0052, "step": 43464 }, { "epoch": 0.04, "learning_rate": 4.9768312088229624e-05, "loss": 0.7613, "step": 43465 }, { "epoch": 0.04, "learning_rate": 4.976830141916307e-05, "loss": 1.0331, "step": 43466 }, { "epoch": 0.04, "learning_rate": 4.976829074985202e-05, "loss": 1.204, "step": 43467 }, { "epoch": 0.04, "learning_rate": 4.976828008029646e-05, "loss": 1.1321, "step": 43468 }, { "epoch": 0.04, "learning_rate": 4.97682694104964e-05, "loss": 0.9122, "step": 43469 }, { "epoch": 0.04, "learning_rate": 4.976825874045184e-05, "loss": 0.6385, "step": 43470 }, { "epoch": 0.04, "learning_rate": 4.976824807016278e-05, "loss": 0.4752, "step": 43471 }, { "epoch": 0.04, "learning_rate": 4.9768237399629227e-05, "loss": 1.3389, "step": 43472 }, { "epoch": 0.04, "learning_rate": 4.976822672885115e-05, "loss": 0.7905, "step": 43473 }, { "epoch": 0.04, "learning_rate": 4.9768216057828585e-05, "loss": 1.0033, "step": 43474 }, { "epoch": 0.04, "learning_rate": 4.976820538656152e-05, "loss": 0.9459, "step": 43475 }, { "epoch": 0.04, "learning_rate": 4.976819471504995e-05, "loss": 1.3259, "step": 43476 }, { "epoch": 0.04, "learning_rate": 4.9768184043293885e-05, "loss": 1.13, "step": 43477 }, { "epoch": 0.04, "learning_rate": 4.9768173371293316e-05, "loss": 1.0591, "step": 43478 }, { "epoch": 0.04, "learning_rate": 4.976816269904824e-05, "loss": 0.9291, "step": 43479 }, { "epoch": 0.04, "learning_rate": 4.976815202655867e-05, "loss": 0.7991, "step": 43480 }, { "epoch": 0.04, "learning_rate": 4.9768141353824596e-05, "loss": 0.9786, "step": 43481 }, { "epoch": 0.04, "learning_rate": 4.976813068084602e-05, "loss": 1.1274, "step": 43482 }, { "epoch": 0.04, "learning_rate": 4.976812000762294e-05, "loss": 1.2593, "step": 43483 }, { "epoch": 0.04, "learning_rate": 4.9768109334155365e-05, "loss": 1.2078, "step": 43484 }, { "epoch": 0.04, "learning_rate": 4.9768098660443295e-05, "loss": 1.1569, "step": 43485 }, { "epoch": 0.04, "learning_rate": 4.976808798648672e-05, "loss": 0.9868, "step": 43486 }, { "epoch": 0.04, "learning_rate": 4.9768077312285644e-05, "loss": 0.7747, "step": 43487 }, { "epoch": 0.04, "learning_rate": 4.9768066637840064e-05, "loss": 1.0812, "step": 43488 }, { "epoch": 0.04, "learning_rate": 4.976805596314999e-05, "loss": 1.1331, "step": 43489 }, { "epoch": 0.04, "learning_rate": 4.976804528821541e-05, "loss": 1.0333, "step": 43490 }, { "epoch": 0.04, "learning_rate": 4.976803461303634e-05, "loss": 1.0944, "step": 43491 }, { "epoch": 0.04, "learning_rate": 4.9768023937612765e-05, "loss": 1.0108, "step": 43492 }, { "epoch": 0.04, "learning_rate": 4.976801326194469e-05, "loss": 1.0759, "step": 43493 }, { "epoch": 0.04, "learning_rate": 4.976800258603211e-05, "loss": 1.0968, "step": 43494 }, { "epoch": 0.04, "learning_rate": 4.9767991909875044e-05, "loss": 1.0592, "step": 43495 }, { "epoch": 0.04, "learning_rate": 4.976798123347347e-05, "loss": 0.9029, "step": 43496 }, { "epoch": 0.04, "learning_rate": 4.97679705568274e-05, "loss": 1.0784, "step": 43497 }, { "epoch": 0.04, "learning_rate": 4.9767959879936826e-05, "loss": 1.2865, "step": 43498 }, { "epoch": 0.04, "learning_rate": 4.9767949202801753e-05, "loss": 0.4135, "step": 43499 }, { "epoch": 0.04, "learning_rate": 4.9767938525422187e-05, "loss": 1.0818, "step": 43500 }, { "epoch": 0.04, "eval_loss": 1.0793282985687256, "eval_runtime": 128.3709, "eval_samples_per_second": 10.789, "eval_steps_per_second": 5.398, "step": 43500 }, { "epoch": 0.04, "learning_rate": 4.9767927847798125e-05, "loss": 0.1879, "step": 43501 }, { "epoch": 0.04, "learning_rate": 4.9767917169929556e-05, "loss": 0.1801, "step": 43502 }, { "epoch": 0.04, "learning_rate": 4.97679064918165e-05, "loss": 0.1623, "step": 43503 }, { "epoch": 0.04, "learning_rate": 4.9767895813458934e-05, "loss": 0.3367, "step": 43504 }, { "epoch": 0.04, "learning_rate": 4.9767885134856875e-05, "loss": 0.26, "step": 43505 }, { "epoch": 0.04, "learning_rate": 4.976787445601031e-05, "loss": 0.3412, "step": 43506 }, { "epoch": 0.04, "learning_rate": 4.9767863776919253e-05, "loss": 0.1209, "step": 43507 }, { "epoch": 0.04, "learning_rate": 4.97678530975837e-05, "loss": 1.1175, "step": 43508 }, { "epoch": 0.04, "learning_rate": 4.9767842418003654e-05, "loss": 1.0668, "step": 43509 }, { "epoch": 0.04, "learning_rate": 4.9767831738179096e-05, "loss": 1.1523, "step": 43510 }, { "epoch": 0.04, "learning_rate": 4.976782105811005e-05, "loss": 1.2666, "step": 43511 }, { "epoch": 0.04, "learning_rate": 4.976781037779651e-05, "loss": 1.3846, "step": 43512 }, { "epoch": 0.04, "learning_rate": 4.976779969723846e-05, "loss": 0.959, "step": 43513 }, { "epoch": 0.04, "learning_rate": 4.976778901643593e-05, "loss": 0.9137, "step": 43514 }, { "epoch": 0.04, "learning_rate": 4.976777833538889e-05, "loss": 0.9293, "step": 43515 }, { "epoch": 0.04, "learning_rate": 4.976776765409735e-05, "loss": 1.0659, "step": 43516 }, { "epoch": 0.04, "learning_rate": 4.976775697256132e-05, "loss": 1.0196, "step": 43517 }, { "epoch": 0.04, "learning_rate": 4.9767746290780794e-05, "loss": 1.2368, "step": 43518 }, { "epoch": 0.04, "learning_rate": 4.976773560875577e-05, "loss": 2.7937, "step": 43519 }, { "epoch": 0.04, "learning_rate": 4.976772492648625e-05, "loss": 3.4842, "step": 43520 }, { "epoch": 0.04, "learning_rate": 4.976771424397223e-05, "loss": 1.2734, "step": 43521 }, { "epoch": 0.04, "learning_rate": 4.976770356121372e-05, "loss": 1.2395, "step": 43522 }, { "epoch": 0.04, "learning_rate": 4.9767692878210705e-05, "loss": 1.0688, "step": 43523 }, { "epoch": 0.04, "learning_rate": 4.97676821949632e-05, "loss": 0.8779, "step": 43524 }, { "epoch": 0.04, "learning_rate": 4.97676715114712e-05, "loss": 1.2659, "step": 43525 }, { "epoch": 0.04, "learning_rate": 4.9767660827734696e-05, "loss": 1.2546, "step": 43526 }, { "epoch": 0.04, "learning_rate": 4.97676501437537e-05, "loss": 0.818, "step": 43527 }, { "epoch": 0.04, "learning_rate": 4.9767639459528216e-05, "loss": 1.1962, "step": 43528 }, { "epoch": 0.04, "learning_rate": 4.976762877505823e-05, "loss": 1.8127, "step": 43529 }, { "epoch": 0.04, "learning_rate": 4.9767618090343745e-05, "loss": 1.1876, "step": 43530 }, { "epoch": 0.04, "learning_rate": 4.9767607405384765e-05, "loss": 1.1346, "step": 43531 }, { "epoch": 0.04, "learning_rate": 4.9767596720181297e-05, "loss": 1.1363, "step": 43532 }, { "epoch": 0.04, "learning_rate": 4.976758603473333e-05, "loss": 0.9584, "step": 43533 }, { "epoch": 0.04, "learning_rate": 4.9767575349040864e-05, "loss": 1.0591, "step": 43534 }, { "epoch": 0.04, "learning_rate": 4.9767564663103906e-05, "loss": 0.8256, "step": 43535 }, { "epoch": 0.04, "learning_rate": 4.9767553976922454e-05, "loss": 1.2606, "step": 43536 }, { "epoch": 0.04, "learning_rate": 4.976754329049651e-05, "loss": 1.3134, "step": 43537 }, { "epoch": 0.04, "learning_rate": 4.976753260382606e-05, "loss": 1.1611, "step": 43538 }, { "epoch": 0.04, "learning_rate": 4.9767521916911124e-05, "loss": 1.1756, "step": 43539 }, { "epoch": 0.04, "learning_rate": 4.9767511229751694e-05, "loss": 1.0008, "step": 43540 }, { "epoch": 0.04, "learning_rate": 4.976750054234777e-05, "loss": 0.9684, "step": 43541 }, { "epoch": 0.04, "learning_rate": 4.9767489854699345e-05, "loss": 1.092, "step": 43542 }, { "epoch": 0.04, "learning_rate": 4.976747916680643e-05, "loss": 0.985, "step": 43543 }, { "epoch": 0.04, "learning_rate": 4.976746847866902e-05, "loss": 0.7568, "step": 43544 }, { "epoch": 0.04, "learning_rate": 4.9767457790287117e-05, "loss": 0.8494, "step": 43545 }, { "epoch": 0.04, "learning_rate": 4.976744710166073e-05, "loss": 0.9788, "step": 43546 }, { "epoch": 0.04, "learning_rate": 4.976743641278983e-05, "loss": 1.0072, "step": 43547 }, { "epoch": 0.04, "learning_rate": 4.9767425723674446e-05, "loss": 0.8792, "step": 43548 }, { "epoch": 0.04, "learning_rate": 4.976741503431457e-05, "loss": 1.2105, "step": 43549 }, { "epoch": 0.04, "learning_rate": 4.976740434471019e-05, "loss": 1.1288, "step": 43550 }, { "epoch": 0.04, "learning_rate": 4.9767393654861326e-05, "loss": 1.3869, "step": 43551 }, { "epoch": 0.04, "learning_rate": 4.976738296476797e-05, "loss": 1.0783, "step": 43552 }, { "epoch": 0.04, "learning_rate": 4.976737227443013e-05, "loss": 0.9991, "step": 43553 }, { "epoch": 0.04, "learning_rate": 4.9767361583847776e-05, "loss": 1.214, "step": 43554 }, { "epoch": 0.04, "learning_rate": 4.976735089302094e-05, "loss": 0.2514, "step": 43555 }, { "epoch": 0.04, "learning_rate": 4.9767340201949605e-05, "loss": 0.1255, "step": 43556 }, { "epoch": 0.04, "learning_rate": 4.9767329510633785e-05, "loss": 0.5974, "step": 43557 }, { "epoch": 0.04, "learning_rate": 4.976731881907347e-05, "loss": 0.4185, "step": 43558 }, { "epoch": 0.04, "learning_rate": 4.976730812726866e-05, "loss": 0.6316, "step": 43559 }, { "epoch": 0.04, "learning_rate": 4.9767297435219364e-05, "loss": 0.2231, "step": 43560 }, { "epoch": 0.04, "learning_rate": 4.9767286742925566e-05, "loss": 0.1719, "step": 43561 }, { "epoch": 0.04, "learning_rate": 4.976727605038728e-05, "loss": 0.1217, "step": 43562 }, { "epoch": 0.04, "learning_rate": 4.976726535760451e-05, "loss": 0.1949, "step": 43563 }, { "epoch": 0.04, "learning_rate": 4.9767254664577234e-05, "loss": 0.114, "step": 43564 }, { "epoch": 0.04, "learning_rate": 4.976724397130548e-05, "loss": 0.1612, "step": 43565 }, { "epoch": 0.04, "learning_rate": 4.976723327778922e-05, "loss": 0.2137, "step": 43566 }, { "epoch": 0.04, "learning_rate": 4.976722258402848e-05, "loss": 0.1427, "step": 43567 }, { "epoch": 0.04, "learning_rate": 4.976721189002324e-05, "loss": 0.2148, "step": 43568 }, { "epoch": 0.04, "learning_rate": 4.9767201195773515e-05, "loss": 0.7083, "step": 43569 }, { "epoch": 0.04, "learning_rate": 4.9767190501279295e-05, "loss": 0.7204, "step": 43570 }, { "epoch": 0.04, "learning_rate": 4.976717980654059e-05, "loss": 1.0645, "step": 43571 }, { "epoch": 0.04, "learning_rate": 4.9767169111557385e-05, "loss": 1.0205, "step": 43572 }, { "epoch": 0.04, "learning_rate": 4.976715841632969e-05, "loss": 1.1838, "step": 43573 }, { "epoch": 0.04, "learning_rate": 4.976714772085751e-05, "loss": 0.9784, "step": 43574 }, { "epoch": 0.04, "learning_rate": 4.9767137025140834e-05, "loss": 0.9001, "step": 43575 }, { "epoch": 0.04, "learning_rate": 4.976712632917967e-05, "loss": 0.8889, "step": 43576 }, { "epoch": 0.04, "learning_rate": 4.976711563297401e-05, "loss": 1.3232, "step": 43577 }, { "epoch": 0.04, "learning_rate": 4.9767104936523866e-05, "loss": 1.0028, "step": 43578 }, { "epoch": 0.04, "learning_rate": 4.9767094239829224e-05, "loss": 0.9124, "step": 43579 }, { "epoch": 0.04, "learning_rate": 4.97670835428901e-05, "loss": 1.2947, "step": 43580 }, { "epoch": 0.04, "learning_rate": 4.9767072845706484e-05, "loss": 1.0311, "step": 43581 }, { "epoch": 0.04, "learning_rate": 4.976706214827838e-05, "loss": 0.7185, "step": 43582 }, { "epoch": 0.04, "learning_rate": 4.976705145060578e-05, "loss": 1.1766, "step": 43583 }, { "epoch": 0.04, "learning_rate": 4.9767040752688685e-05, "loss": 0.8214, "step": 43584 }, { "epoch": 0.04, "learning_rate": 4.976703005452711e-05, "loss": 0.5731, "step": 43585 }, { "epoch": 0.04, "learning_rate": 4.976701935612105e-05, "loss": 1.1233, "step": 43586 }, { "epoch": 0.04, "learning_rate": 4.976700865747049e-05, "loss": 0.793, "step": 43587 }, { "epoch": 0.04, "learning_rate": 4.976699795857545e-05, "loss": 1.0227, "step": 43588 }, { "epoch": 0.04, "learning_rate": 4.976698725943591e-05, "loss": 1.1212, "step": 43589 }, { "epoch": 0.04, "learning_rate": 4.9766976560051884e-05, "loss": 1.0604, "step": 43590 }, { "epoch": 0.04, "learning_rate": 4.976696586042338e-05, "loss": 0.9397, "step": 43591 }, { "epoch": 0.04, "learning_rate": 4.976695516055037e-05, "loss": 1.0859, "step": 43592 }, { "epoch": 0.04, "learning_rate": 4.976694446043287e-05, "loss": 1.4275, "step": 43593 }, { "epoch": 0.04, "learning_rate": 4.97669337600709e-05, "loss": 1.1398, "step": 43594 }, { "epoch": 0.04, "learning_rate": 4.976692305946443e-05, "loss": 0.8304, "step": 43595 }, { "epoch": 0.04, "learning_rate": 4.9766912358613474e-05, "loss": 0.9305, "step": 43596 }, { "epoch": 0.04, "learning_rate": 4.976690165751803e-05, "loss": 1.1256, "step": 43597 }, { "epoch": 0.04, "learning_rate": 4.9766890956178095e-05, "loss": 0.9688, "step": 43598 }, { "epoch": 0.04, "learning_rate": 4.976688025459367e-05, "loss": 1.6875, "step": 43599 }, { "epoch": 0.04, "learning_rate": 4.976686955276476e-05, "loss": 1.0296, "step": 43600 }, { "epoch": 0.04, "learning_rate": 4.9766858850691355e-05, "loss": 1.1038, "step": 43601 }, { "epoch": 0.04, "learning_rate": 4.976684814837347e-05, "loss": 0.9567, "step": 43602 }, { "epoch": 0.04, "learning_rate": 4.976683744581109e-05, "loss": 0.8703, "step": 43603 }, { "epoch": 0.04, "learning_rate": 4.976682674300423e-05, "loss": 1.205, "step": 43604 }, { "epoch": 0.04, "learning_rate": 4.976681603995288e-05, "loss": 0.9705, "step": 43605 }, { "epoch": 0.04, "learning_rate": 4.976680533665704e-05, "loss": 0.8716, "step": 43606 }, { "epoch": 0.04, "learning_rate": 4.976679463311672e-05, "loss": 1.1162, "step": 43607 }, { "epoch": 0.04, "learning_rate": 4.976678392933191e-05, "loss": 1.2164, "step": 43608 }, { "epoch": 0.04, "learning_rate": 4.97667732253026e-05, "loss": 1.3875, "step": 43609 }, { "epoch": 0.04, "learning_rate": 4.976676252102881e-05, "loss": 1.1021, "step": 43610 }, { "epoch": 0.04, "learning_rate": 4.976675181651054e-05, "loss": 2.9305, "step": 43611 }, { "epoch": 0.04, "learning_rate": 4.976674111174778e-05, "loss": 3.055, "step": 43612 }, { "epoch": 0.04, "learning_rate": 4.976673040674053e-05, "loss": 1.3167, "step": 43613 }, { "epoch": 0.04, "learning_rate": 4.97667197014888e-05, "loss": 1.2186, "step": 43614 }, { "epoch": 0.04, "learning_rate": 4.9766708995992576e-05, "loss": 0.8302, "step": 43615 }, { "epoch": 0.04, "learning_rate": 4.9766698290251866e-05, "loss": 1.2113, "step": 43616 }, { "epoch": 0.04, "learning_rate": 4.976668758426667e-05, "loss": 1.0783, "step": 43617 }, { "epoch": 0.04, "learning_rate": 4.976667687803699e-05, "loss": 0.6285, "step": 43618 }, { "epoch": 0.04, "learning_rate": 4.976666617156282e-05, "loss": 0.8071, "step": 43619 }, { "epoch": 0.04, "learning_rate": 4.976665546484417e-05, "loss": 1.3871, "step": 43620 }, { "epoch": 0.04, "learning_rate": 4.9766644757881035e-05, "loss": 1.1133, "step": 43621 }, { "epoch": 0.04, "learning_rate": 4.9766634050673406e-05, "loss": 0.5011, "step": 43622 }, { "epoch": 0.04, "learning_rate": 4.9766623343221296e-05, "loss": 0.9902, "step": 43623 }, { "epoch": 0.04, "learning_rate": 4.97666126355247e-05, "loss": 0.9382, "step": 43624 }, { "epoch": 0.04, "learning_rate": 4.9766601927583614e-05, "loss": 1.0122, "step": 43625 }, { "epoch": 0.04, "learning_rate": 4.976659121939805e-05, "loss": 0.9174, "step": 43626 }, { "epoch": 0.04, "learning_rate": 4.9766580510967995e-05, "loss": 1.1385, "step": 43627 }, { "epoch": 0.04, "learning_rate": 4.976656980229346e-05, "loss": 1.3703, "step": 43628 }, { "epoch": 0.04, "learning_rate": 4.976655909337443e-05, "loss": 1.121, "step": 43629 }, { "epoch": 0.04, "learning_rate": 4.976654838421092e-05, "loss": 1.3205, "step": 43630 }, { "epoch": 0.04, "learning_rate": 4.976653767480293e-05, "loss": 0.7371, "step": 43631 }, { "epoch": 0.04, "learning_rate": 4.9766526965150455e-05, "loss": 0.8441, "step": 43632 }, { "epoch": 0.04, "learning_rate": 4.976651625525349e-05, "loss": 1.0329, "step": 43633 }, { "epoch": 0.04, "learning_rate": 4.9766505545112043e-05, "loss": 1.1377, "step": 43634 }, { "epoch": 0.04, "learning_rate": 4.976649483472611e-05, "loss": 1.1663, "step": 43635 }, { "epoch": 0.04, "learning_rate": 4.9766484124095695e-05, "loss": 1.2139, "step": 43636 }, { "epoch": 0.04, "learning_rate": 4.976647341322079e-05, "loss": 0.9001, "step": 43637 }, { "epoch": 0.04, "learning_rate": 4.976646270210141e-05, "loss": 0.8858, "step": 43638 }, { "epoch": 0.04, "learning_rate": 4.976645199073754e-05, "loss": 0.8807, "step": 43639 }, { "epoch": 0.04, "learning_rate": 4.976644127912919e-05, "loss": 0.7296, "step": 43640 }, { "epoch": 0.04, "learning_rate": 4.9766430567276356e-05, "loss": 0.7851, "step": 43641 }, { "epoch": 0.04, "learning_rate": 4.9766419855179036e-05, "loss": 1.1657, "step": 43642 }, { "epoch": 0.04, "learning_rate": 4.976640914283723e-05, "loss": 0.9732, "step": 43643 }, { "epoch": 0.04, "learning_rate": 4.976639843025094e-05, "loss": 0.7124, "step": 43644 }, { "epoch": 0.04, "learning_rate": 4.976638771742017e-05, "loss": 0.1874, "step": 43645 }, { "epoch": 0.04, "learning_rate": 4.9766377004344915e-05, "loss": 0.2172, "step": 43646 }, { "epoch": 0.04, "learning_rate": 4.976636629102518e-05, "loss": 0.1747, "step": 43647 }, { "epoch": 0.04, "learning_rate": 4.976635557746095e-05, "loss": 0.3386, "step": 43648 }, { "epoch": 0.04, "learning_rate": 4.9766344863652254e-05, "loss": 0.3336, "step": 43649 }, { "epoch": 0.04, "learning_rate": 4.976633414959907e-05, "loss": 0.1135, "step": 43650 }, { "epoch": 0.04, "learning_rate": 4.97663234353014e-05, "loss": 0.1555, "step": 43651 }, { "epoch": 0.04, "learning_rate": 4.9766312720759245e-05, "loss": 0.2353, "step": 43652 }, { "epoch": 0.04, "learning_rate": 4.976630200597261e-05, "loss": 0.1556, "step": 43653 }, { "epoch": 0.04, "learning_rate": 4.976629129094149e-05, "loss": 0.2056, "step": 43654 }, { "epoch": 0.04, "learning_rate": 4.976628057566589e-05, "loss": 0.1185, "step": 43655 }, { "epoch": 0.04, "learning_rate": 4.976626986014581e-05, "loss": 0.4746, "step": 43656 }, { "epoch": 0.04, "learning_rate": 4.9766259144381244e-05, "loss": 1.046, "step": 43657 }, { "epoch": 0.04, "learning_rate": 4.97662484283722e-05, "loss": 1.1846, "step": 43658 }, { "epoch": 0.04, "learning_rate": 4.976623771211867e-05, "loss": 1.4738, "step": 43659 }, { "epoch": 0.04, "learning_rate": 4.976622699562066e-05, "loss": 0.9828, "step": 43660 }, { "epoch": 0.04, "learning_rate": 4.9766216278878166e-05, "loss": 0.8722, "step": 43661 }, { "epoch": 0.04, "learning_rate": 4.9766205561891196e-05, "loss": 1.0096, "step": 43662 }, { "epoch": 0.04, "learning_rate": 4.976619484465974e-05, "loss": 1.2449, "step": 43663 }, { "epoch": 0.04, "learning_rate": 4.97661841271838e-05, "loss": 1.1459, "step": 43664 }, { "epoch": 0.04, "learning_rate": 4.976617340946339e-05, "loss": 1.0648, "step": 43665 }, { "epoch": 0.04, "learning_rate": 4.976616269149849e-05, "loss": 0.955, "step": 43666 }, { "epoch": 0.04, "learning_rate": 4.976615197328911e-05, "loss": 1.3172, "step": 43667 }, { "epoch": 0.04, "learning_rate": 4.9766141254835246e-05, "loss": 0.8297, "step": 43668 }, { "epoch": 0.04, "learning_rate": 4.9766130536136903e-05, "loss": 0.9204, "step": 43669 }, { "epoch": 0.04, "learning_rate": 4.976611981719408e-05, "loss": 0.8686, "step": 43670 }, { "epoch": 0.04, "learning_rate": 4.9766109098006776e-05, "loss": 1.0163, "step": 43671 }, { "epoch": 0.04, "learning_rate": 4.976609837857499e-05, "loss": 1.039, "step": 43672 }, { "epoch": 0.04, "learning_rate": 4.976608765889873e-05, "loss": 1.1423, "step": 43673 }, { "epoch": 0.04, "learning_rate": 4.976607693897798e-05, "loss": 1.1242, "step": 43674 }, { "epoch": 0.04, "learning_rate": 4.976606621881276e-05, "loss": 0.776, "step": 43675 }, { "epoch": 0.04, "learning_rate": 4.976605549840305e-05, "loss": 1.1123, "step": 43676 }, { "epoch": 0.04, "learning_rate": 4.976604477774887e-05, "loss": 1.0901, "step": 43677 }, { "epoch": 0.04, "learning_rate": 4.97660340568502e-05, "loss": 1.3775, "step": 43678 }, { "epoch": 0.04, "learning_rate": 4.9766023335707055e-05, "loss": 0.8808, "step": 43679 }, { "epoch": 0.04, "learning_rate": 4.976601261431943e-05, "loss": 0.6574, "step": 43680 }, { "epoch": 0.04, "learning_rate": 4.9766001892687325e-05, "loss": 0.817, "step": 43681 }, { "epoch": 0.04, "learning_rate": 4.9765991170810746e-05, "loss": 0.3532, "step": 43682 }, { "epoch": 0.04, "learning_rate": 4.976598044868967e-05, "loss": 0.4493, "step": 43683 }, { "epoch": 0.04, "learning_rate": 4.976596972632414e-05, "loss": 0.4478, "step": 43684 }, { "epoch": 0.04, "learning_rate": 4.9765959003714116e-05, "loss": 0.6675, "step": 43685 }, { "epoch": 0.04, "learning_rate": 4.976594828085961e-05, "loss": 0.9616, "step": 43686 }, { "epoch": 0.04, "learning_rate": 4.976593755776063e-05, "loss": 1.0865, "step": 43687 }, { "epoch": 0.04, "learning_rate": 4.976592683441717e-05, "loss": 1.2101, "step": 43688 }, { "epoch": 0.04, "learning_rate": 4.9765916110829234e-05, "loss": 1.1922, "step": 43689 }, { "epoch": 0.04, "learning_rate": 4.9765905386996815e-05, "loss": 1.0531, "step": 43690 }, { "epoch": 0.04, "learning_rate": 4.976589466291992e-05, "loss": 0.8962, "step": 43691 }, { "epoch": 0.04, "learning_rate": 4.976588393859855e-05, "loss": 1.0141, "step": 43692 }, { "epoch": 0.04, "learning_rate": 4.976587321403269e-05, "loss": 1.0096, "step": 43693 }, { "epoch": 0.04, "learning_rate": 4.9765862489222364e-05, "loss": 1.0671, "step": 43694 }, { "epoch": 0.04, "learning_rate": 4.9765851764167555e-05, "loss": 1.0529, "step": 43695 }, { "epoch": 0.04, "learning_rate": 4.9765841038868264e-05, "loss": 0.8222, "step": 43696 }, { "epoch": 0.04, "learning_rate": 4.97658303133245e-05, "loss": 0.517, "step": 43697 }, { "epoch": 0.04, "learning_rate": 4.976581958753626e-05, "loss": 0.3466, "step": 43698 }, { "epoch": 0.04, "learning_rate": 4.9765808861503535e-05, "loss": 0.7992, "step": 43699 }, { "epoch": 0.04, "learning_rate": 4.976579813522634e-05, "loss": 1.1095, "step": 43700 }, { "epoch": 0.04, "learning_rate": 4.976578740870467e-05, "loss": 0.9834, "step": 43701 }, { "epoch": 0.04, "learning_rate": 4.976577668193851e-05, "loss": 0.9441, "step": 43702 }, { "epoch": 0.04, "learning_rate": 4.9765765954927884e-05, "loss": 0.3046, "step": 43703 }, { "epoch": 0.04, "learning_rate": 4.976575522767278e-05, "loss": 0.5088, "step": 43704 }, { "epoch": 0.04, "learning_rate": 4.9765744500173197e-05, "loss": 0.4917, "step": 43705 }, { "epoch": 0.04, "learning_rate": 4.976573377242913e-05, "loss": 0.7988, "step": 43706 }, { "epoch": 0.04, "learning_rate": 4.976572304444059e-05, "loss": 0.9351, "step": 43707 }, { "epoch": 0.04, "learning_rate": 4.976571231620758e-05, "loss": 0.6943, "step": 43708 }, { "epoch": 0.04, "learning_rate": 4.976570158773009e-05, "loss": 0.7016, "step": 43709 }, { "epoch": 0.04, "learning_rate": 4.976569085900812e-05, "loss": 0.6739, "step": 43710 }, { "epoch": 0.04, "learning_rate": 4.976568013004167e-05, "loss": 0.7035, "step": 43711 }, { "epoch": 0.04, "learning_rate": 4.976566940083075e-05, "loss": 0.9589, "step": 43712 }, { "epoch": 0.04, "learning_rate": 4.976565867137535e-05, "loss": 1.0234, "step": 43713 }, { "epoch": 0.04, "learning_rate": 4.976564794167548e-05, "loss": 0.7244, "step": 43714 }, { "epoch": 0.04, "learning_rate": 4.976563721173113e-05, "loss": 0.8438, "step": 43715 }, { "epoch": 0.04, "learning_rate": 4.9765626481542304e-05, "loss": 0.8664, "step": 43716 }, { "epoch": 0.04, "learning_rate": 4.9765615751109005e-05, "loss": 1.0183, "step": 43717 }, { "epoch": 0.04, "learning_rate": 4.9765605020431225e-05, "loss": 1.0794, "step": 43718 }, { "epoch": 0.04, "learning_rate": 4.976559428950898e-05, "loss": 1.092, "step": 43719 }, { "epoch": 0.04, "learning_rate": 4.976558355834224e-05, "loss": 1.0902, "step": 43720 }, { "epoch": 0.04, "learning_rate": 4.976557282693104e-05, "loss": 1.2289, "step": 43721 }, { "epoch": 0.04, "learning_rate": 4.976556209527536e-05, "loss": 1.2143, "step": 43722 }, { "epoch": 0.04, "learning_rate": 4.9765551363375206e-05, "loss": 0.9051, "step": 43723 }, { "epoch": 0.04, "learning_rate": 4.976554063123058e-05, "loss": 0.9361, "step": 43724 }, { "epoch": 0.04, "learning_rate": 4.9765529898841476e-05, "loss": 1.3687, "step": 43725 }, { "epoch": 0.04, "learning_rate": 4.97655191662079e-05, "loss": 0.7206, "step": 43726 }, { "epoch": 0.04, "learning_rate": 4.976550843332984e-05, "loss": 0.9736, "step": 43727 }, { "epoch": 0.04, "learning_rate": 4.9765497700207316e-05, "loss": 0.8953, "step": 43728 }, { "epoch": 0.04, "learning_rate": 4.9765486966840314e-05, "loss": 0.9616, "step": 43729 }, { "epoch": 0.04, "learning_rate": 4.976547623322884e-05, "loss": 1.3716, "step": 43730 }, { "epoch": 0.04, "learning_rate": 4.976546549937288e-05, "loss": 0.8412, "step": 43731 }, { "epoch": 0.04, "learning_rate": 4.976545476527246e-05, "loss": 0.9644, "step": 43732 }, { "epoch": 0.04, "learning_rate": 4.9765444030927555e-05, "loss": 1.2394, "step": 43733 }, { "epoch": 0.04, "learning_rate": 4.976543329633818e-05, "loss": 0.848, "step": 43734 }, { "epoch": 0.04, "learning_rate": 4.976542256150433e-05, "loss": 1.0036, "step": 43735 }, { "epoch": 0.04, "learning_rate": 4.976541182642601e-05, "loss": 0.2354, "step": 43736 }, { "epoch": 0.04, "learning_rate": 4.976540109110322e-05, "loss": 0.7882, "step": 43737 }, { "epoch": 0.04, "learning_rate": 4.976539035553594e-05, "loss": 1.1529, "step": 43738 }, { "epoch": 0.04, "learning_rate": 4.976537961972421e-05, "loss": 1.4231, "step": 43739 }, { "epoch": 0.04, "learning_rate": 4.9765368883667985e-05, "loss": 1.1107, "step": 43740 }, { "epoch": 0.04, "learning_rate": 4.9765358147367295e-05, "loss": 0.8821, "step": 43741 }, { "epoch": 0.04, "learning_rate": 4.976534741082214e-05, "loss": 1.2543, "step": 43742 }, { "epoch": 0.04, "learning_rate": 4.97653366740325e-05, "loss": 0.9341, "step": 43743 }, { "epoch": 0.04, "learning_rate": 4.976532593699839e-05, "loss": 0.8614, "step": 43744 }, { "epoch": 0.04, "learning_rate": 4.976531519971981e-05, "loss": 1.0374, "step": 43745 }, { "epoch": 0.04, "learning_rate": 4.9765304462196755e-05, "loss": 1.0868, "step": 43746 }, { "epoch": 0.04, "learning_rate": 4.976529372442923e-05, "loss": 0.9329, "step": 43747 }, { "epoch": 0.04, "learning_rate": 4.9765282986417227e-05, "loss": 1.1294, "step": 43748 }, { "epoch": 0.04, "learning_rate": 4.976527224816076e-05, "loss": 0.9225, "step": 43749 }, { "epoch": 0.04, "learning_rate": 4.976526150965981e-05, "loss": 0.7093, "step": 43750 }, { "epoch": 0.04, "learning_rate": 4.97652507709144e-05, "loss": 1.0579, "step": 43751 }, { "epoch": 0.04, "learning_rate": 4.976524003192451e-05, "loss": 1.1565, "step": 43752 }, { "epoch": 0.04, "learning_rate": 4.976522929269015e-05, "loss": 1.0652, "step": 43753 }, { "epoch": 0.04, "learning_rate": 4.976521855321131e-05, "loss": 1.2769, "step": 43754 }, { "epoch": 0.04, "learning_rate": 4.976520781348801e-05, "loss": 1.0554, "step": 43755 }, { "epoch": 0.04, "learning_rate": 4.976519707352023e-05, "loss": 1.3024, "step": 43756 }, { "epoch": 0.04, "learning_rate": 4.9765186333307984e-05, "loss": 0.9108, "step": 43757 }, { "epoch": 0.04, "learning_rate": 4.976517559285126e-05, "loss": 0.9215, "step": 43758 }, { "epoch": 0.04, "learning_rate": 4.9765164852150074e-05, "loss": 0.8495, "step": 43759 }, { "epoch": 0.04, "learning_rate": 4.976515411120442e-05, "loss": 0.8444, "step": 43760 }, { "epoch": 0.04, "learning_rate": 4.9765143370014274e-05, "loss": 1.2747, "step": 43761 }, { "epoch": 0.04, "learning_rate": 4.976513262857968e-05, "loss": 0.9885, "step": 43762 }, { "epoch": 0.04, "learning_rate": 4.97651218869006e-05, "loss": 0.6006, "step": 43763 }, { "epoch": 0.04, "learning_rate": 4.9765111144977054e-05, "loss": 0.4887, "step": 43764 }, { "epoch": 0.04, "learning_rate": 4.976510040280904e-05, "loss": 1.2415, "step": 43765 }, { "epoch": 0.04, "learning_rate": 4.9765089660396555e-05, "loss": 0.3332, "step": 43766 }, { "epoch": 0.04, "learning_rate": 4.9765078917739594e-05, "loss": 0.1436, "step": 43767 }, { "epoch": 0.04, "learning_rate": 4.976506817483817e-05, "loss": 0.4052, "step": 43768 }, { "epoch": 0.04, "learning_rate": 4.976505743169227e-05, "loss": 1.0858, "step": 43769 }, { "epoch": 0.04, "learning_rate": 4.9765046688301895e-05, "loss": 0.9192, "step": 43770 }, { "epoch": 0.04, "learning_rate": 4.976503594466706e-05, "loss": 0.757, "step": 43771 }, { "epoch": 0.04, "learning_rate": 4.9765025200787755e-05, "loss": 1.1626, "step": 43772 }, { "epoch": 0.04, "learning_rate": 4.976501445666397e-05, "loss": 1.147, "step": 43773 }, { "epoch": 0.04, "learning_rate": 4.976500371229572e-05, "loss": 1.1713, "step": 43774 }, { "epoch": 0.04, "learning_rate": 4.9764992967683006e-05, "loss": 0.8769, "step": 43775 }, { "epoch": 0.04, "learning_rate": 4.976498222282582e-05, "loss": 0.9963, "step": 43776 }, { "epoch": 0.04, "learning_rate": 4.9764971477724166e-05, "loss": 0.9176, "step": 43777 }, { "epoch": 0.04, "learning_rate": 4.976496073237804e-05, "loss": 1.0045, "step": 43778 }, { "epoch": 0.04, "learning_rate": 4.976494998678744e-05, "loss": 1.1725, "step": 43779 }, { "epoch": 0.04, "learning_rate": 4.9764939240952374e-05, "loss": 0.7227, "step": 43780 }, { "epoch": 0.04, "learning_rate": 4.9764928494872845e-05, "loss": 0.7603, "step": 43781 }, { "epoch": 0.04, "learning_rate": 4.9764917748548835e-05, "loss": 1.1116, "step": 43782 }, { "epoch": 0.04, "learning_rate": 4.9764907001980364e-05, "loss": 0.9965, "step": 43783 }, { "epoch": 0.04, "learning_rate": 4.9764896255167426e-05, "loss": 0.9883, "step": 43784 }, { "epoch": 0.04, "learning_rate": 4.9764885508110014e-05, "loss": 1.0498, "step": 43785 }, { "epoch": 0.04, "learning_rate": 4.976487476080814e-05, "loss": 0.8829, "step": 43786 }, { "epoch": 0.04, "learning_rate": 4.976486401326179e-05, "loss": 0.6126, "step": 43787 }, { "epoch": 0.04, "learning_rate": 4.976485326547098e-05, "loss": 0.6918, "step": 43788 }, { "epoch": 0.04, "learning_rate": 4.976484251743569e-05, "loss": 0.8142, "step": 43789 }, { "epoch": 0.04, "learning_rate": 4.9764831769155945e-05, "loss": 0.9374, "step": 43790 }, { "epoch": 0.04, "learning_rate": 4.976482102063172e-05, "loss": 0.7676, "step": 43791 }, { "epoch": 0.04, "learning_rate": 4.976481027186304e-05, "loss": 0.6398, "step": 43792 }, { "epoch": 0.04, "learning_rate": 4.976479952284989e-05, "loss": 1.0641, "step": 43793 }, { "epoch": 0.04, "learning_rate": 4.9764788773592266e-05, "loss": 1.1572, "step": 43794 }, { "epoch": 0.04, "learning_rate": 4.9764778024090174e-05, "loss": 1.2586, "step": 43795 }, { "epoch": 0.04, "learning_rate": 4.9764767274343615e-05, "loss": 0.757, "step": 43796 }, { "epoch": 0.04, "learning_rate": 4.9764756524352596e-05, "loss": 1.1803, "step": 43797 }, { "epoch": 0.04, "learning_rate": 4.9764745774117096e-05, "loss": 1.3283, "step": 43798 }, { "epoch": 0.04, "learning_rate": 4.976473502363714e-05, "loss": 0.9015, "step": 43799 }, { "epoch": 0.04, "learning_rate": 4.9764724272912715e-05, "loss": 1.1698, "step": 43800 }, { "epoch": 0.04, "learning_rate": 4.976471352194382e-05, "loss": 1.2405, "step": 43801 }, { "epoch": 0.04, "learning_rate": 4.9764702770730465e-05, "loss": 1.131, "step": 43802 }, { "epoch": 0.04, "learning_rate": 4.976469201927264e-05, "loss": 0.8027, "step": 43803 }, { "epoch": 0.04, "learning_rate": 4.9764681267570345e-05, "loss": 1.1454, "step": 43804 }, { "epoch": 0.04, "learning_rate": 4.976467051562359e-05, "loss": 1.0556, "step": 43805 }, { "epoch": 0.04, "learning_rate": 4.976465976343236e-05, "loss": 1.0092, "step": 43806 }, { "epoch": 0.04, "learning_rate": 4.9764649010996665e-05, "loss": 1.6903, "step": 43807 }, { "epoch": 0.04, "learning_rate": 4.976463825831651e-05, "loss": 0.9227, "step": 43808 }, { "epoch": 0.04, "learning_rate": 4.9764627505391886e-05, "loss": 0.9879, "step": 43809 }, { "epoch": 0.04, "learning_rate": 4.97646167522228e-05, "loss": 1.0436, "step": 43810 }, { "epoch": 0.04, "learning_rate": 4.9764605998809245e-05, "loss": 1.2756, "step": 43811 }, { "epoch": 0.04, "learning_rate": 4.976459524515122e-05, "loss": 1.142, "step": 43812 }, { "epoch": 0.04, "learning_rate": 4.9764584491248736e-05, "loss": 0.9496, "step": 43813 }, { "epoch": 0.04, "learning_rate": 4.976457373710178e-05, "loss": 0.8323, "step": 43814 }, { "epoch": 0.04, "learning_rate": 4.9764562982710363e-05, "loss": 1.0107, "step": 43815 }, { "epoch": 0.04, "learning_rate": 4.976455222807448e-05, "loss": 0.9054, "step": 43816 }, { "epoch": 0.04, "learning_rate": 4.976454147319413e-05, "loss": 1.3008, "step": 43817 }, { "epoch": 0.04, "learning_rate": 4.976453071806931e-05, "loss": 0.514, "step": 43818 }, { "epoch": 0.04, "learning_rate": 4.976451996270003e-05, "loss": 0.9326, "step": 43819 }, { "epoch": 0.04, "learning_rate": 4.976450920708629e-05, "loss": 1.1058, "step": 43820 }, { "epoch": 0.04, "learning_rate": 4.976449845122808e-05, "loss": 0.5187, "step": 43821 }, { "epoch": 0.04, "learning_rate": 4.976448769512541e-05, "loss": 0.3356, "step": 43822 }, { "epoch": 0.04, "learning_rate": 4.976447693877827e-05, "loss": 0.1685, "step": 43823 }, { "epoch": 0.04, "learning_rate": 4.9764466182186665e-05, "loss": 0.2075, "step": 43824 }, { "epoch": 0.04, "learning_rate": 4.9764455425350596e-05, "loss": 0.3138, "step": 43825 }, { "epoch": 0.04, "learning_rate": 4.9764444668270074e-05, "loss": 0.7347, "step": 43826 }, { "epoch": 0.04, "learning_rate": 4.976443391094508e-05, "loss": 1.1349, "step": 43827 }, { "epoch": 0.04, "learning_rate": 4.9764423153375614e-05, "loss": 1.0743, "step": 43828 }, { "epoch": 0.04, "learning_rate": 4.976441239556169e-05, "loss": 0.9341, "step": 43829 }, { "epoch": 0.04, "learning_rate": 4.9764401637503305e-05, "loss": 0.9165, "step": 43830 }, { "epoch": 0.04, "learning_rate": 4.976439087920045e-05, "loss": 0.7887, "step": 43831 }, { "epoch": 0.04, "learning_rate": 4.976438012065314e-05, "loss": 1.0787, "step": 43832 }, { "epoch": 0.04, "learning_rate": 4.976436936186136e-05, "loss": 1.1339, "step": 43833 }, { "epoch": 0.04, "learning_rate": 4.976435860282512e-05, "loss": 0.8531, "step": 43834 }, { "epoch": 0.04, "learning_rate": 4.9764347843544414e-05, "loss": 0.8883, "step": 43835 }, { "epoch": 0.04, "learning_rate": 4.9764337084019246e-05, "loss": 0.4779, "step": 43836 }, { "epoch": 0.04, "learning_rate": 4.976432632424962e-05, "loss": 0.6563, "step": 43837 }, { "epoch": 0.04, "learning_rate": 4.976431556423552e-05, "loss": 1.1505, "step": 43838 }, { "epoch": 0.04, "learning_rate": 4.9764304803976966e-05, "loss": 1.0257, "step": 43839 }, { "epoch": 0.04, "learning_rate": 4.976429404347395e-05, "loss": 1.216, "step": 43840 }, { "epoch": 0.04, "learning_rate": 4.9764283282726466e-05, "loss": 1.0339, "step": 43841 }, { "epoch": 0.04, "learning_rate": 4.9764272521734515e-05, "loss": 1.081, "step": 43842 }, { "epoch": 0.04, "learning_rate": 4.976426176049811e-05, "loss": 1.1918, "step": 43843 }, { "epoch": 0.04, "learning_rate": 4.976425099901724e-05, "loss": 1.5537, "step": 43844 }, { "epoch": 0.04, "learning_rate": 4.9764240237291906e-05, "loss": 1.8211, "step": 43845 }, { "epoch": 0.04, "learning_rate": 4.976422947532211e-05, "loss": 1.0268, "step": 43846 }, { "epoch": 0.04, "learning_rate": 4.976421871310786e-05, "loss": 2.2849, "step": 43847 }, { "epoch": 0.04, "learning_rate": 4.976420795064914e-05, "loss": 1.0778, "step": 43848 }, { "epoch": 0.04, "learning_rate": 4.976419718794596e-05, "loss": 1.0269, "step": 43849 }, { "epoch": 0.04, "learning_rate": 4.9764186424998316e-05, "loss": 1.1075, "step": 43850 }, { "epoch": 0.04, "learning_rate": 4.9764175661806214e-05, "loss": 0.9895, "step": 43851 }, { "epoch": 0.04, "learning_rate": 4.9764164898369644e-05, "loss": 1.0039, "step": 43852 }, { "epoch": 0.04, "learning_rate": 4.976415413468862e-05, "loss": 1.1522, "step": 43853 }, { "epoch": 0.04, "learning_rate": 4.976414337076313e-05, "loss": 1.1527, "step": 43854 }, { "epoch": 0.04, "learning_rate": 4.9764132606593187e-05, "loss": 0.9377, "step": 43855 }, { "epoch": 0.04, "learning_rate": 4.976412184217878e-05, "loss": 0.9428, "step": 43856 }, { "epoch": 0.04, "learning_rate": 4.97641110775199e-05, "loss": 0.8072, "step": 43857 }, { "epoch": 0.04, "learning_rate": 4.976410031261657e-05, "loss": 0.9477, "step": 43858 }, { "epoch": 0.04, "learning_rate": 4.9764089547468785e-05, "loss": 0.5796, "step": 43859 }, { "epoch": 0.04, "learning_rate": 4.976407878207653e-05, "loss": 1.0377, "step": 43860 }, { "epoch": 0.04, "learning_rate": 4.976406801643981e-05, "loss": 1.3208, "step": 43861 }, { "epoch": 0.04, "learning_rate": 4.976405725055864e-05, "loss": 0.9896, "step": 43862 }, { "epoch": 0.04, "learning_rate": 4.9764046484433e-05, "loss": 1.0915, "step": 43863 }, { "epoch": 0.04, "learning_rate": 4.976403571806291e-05, "loss": 1.2515, "step": 43864 }, { "epoch": 0.04, "learning_rate": 4.976402495144836e-05, "loss": 0.8876, "step": 43865 }, { "epoch": 0.04, "learning_rate": 4.976401418458934e-05, "loss": 0.6953, "step": 43866 }, { "epoch": 0.04, "learning_rate": 4.9764003417485865e-05, "loss": 1.0554, "step": 43867 }, { "epoch": 0.04, "learning_rate": 4.976399265013793e-05, "loss": 0.8604, "step": 43868 }, { "epoch": 0.04, "learning_rate": 4.9763981882545543e-05, "loss": 1.0035, "step": 43869 }, { "epoch": 0.04, "learning_rate": 4.9763971114708685e-05, "loss": 1.3623, "step": 43870 }, { "epoch": 0.04, "learning_rate": 4.976396034662737e-05, "loss": 1.3579, "step": 43871 }, { "epoch": 0.04, "learning_rate": 4.9763949578301594e-05, "loss": 0.7895, "step": 43872 }, { "epoch": 0.04, "learning_rate": 4.976393880973137e-05, "loss": 1.0426, "step": 43873 }, { "epoch": 0.04, "learning_rate": 4.9763928040916675e-05, "loss": 1.0916, "step": 43874 }, { "epoch": 0.04, "learning_rate": 4.976391727185752e-05, "loss": 0.9515, "step": 43875 }, { "epoch": 0.04, "learning_rate": 4.9763906502553914e-05, "loss": 0.7617, "step": 43876 }, { "epoch": 0.04, "learning_rate": 4.976389573300585e-05, "loss": 0.7362, "step": 43877 }, { "epoch": 0.04, "learning_rate": 4.976388496321332e-05, "loss": 0.41, "step": 43878 }, { "epoch": 0.04, "learning_rate": 4.9763874193176335e-05, "loss": 0.4438, "step": 43879 }, { "epoch": 0.04, "learning_rate": 4.976386342289489e-05, "loss": 0.4518, "step": 43880 }, { "epoch": 0.04, "learning_rate": 4.976385265236899e-05, "loss": 0.4567, "step": 43881 }, { "epoch": 0.04, "learning_rate": 4.9763841881598625e-05, "loss": 0.3839, "step": 43882 }, { "epoch": 0.04, "learning_rate": 4.976383111058381e-05, "loss": 0.5121, "step": 43883 }, { "epoch": 0.04, "learning_rate": 4.976382033932453e-05, "loss": 0.8345, "step": 43884 }, { "epoch": 0.04, "learning_rate": 4.97638095678208e-05, "loss": 1.0445, "step": 43885 }, { "epoch": 0.04, "learning_rate": 4.97637987960726e-05, "loss": 1.1974, "step": 43886 }, { "epoch": 0.04, "learning_rate": 4.9763788024079954e-05, "loss": 1.1905, "step": 43887 }, { "epoch": 0.04, "learning_rate": 4.976377725184285e-05, "loss": 1.2364, "step": 43888 }, { "epoch": 0.04, "learning_rate": 4.976376647936128e-05, "loss": 1.3434, "step": 43889 }, { "epoch": 0.04, "learning_rate": 4.9763755706635254e-05, "loss": 1.714, "step": 43890 }, { "epoch": 0.04, "learning_rate": 4.976374493366477e-05, "loss": 1.2352, "step": 43891 }, { "epoch": 0.04, "learning_rate": 4.976373416044984e-05, "loss": 0.8926, "step": 43892 }, { "epoch": 0.04, "learning_rate": 4.976372338699044e-05, "loss": 1.4564, "step": 43893 }, { "epoch": 0.04, "learning_rate": 4.976371261328659e-05, "loss": 1.5588, "step": 43894 }, { "epoch": 0.04, "learning_rate": 4.9763701839338285e-05, "loss": 1.083, "step": 43895 }, { "epoch": 0.04, "learning_rate": 4.9763691065145515e-05, "loss": 1.2636, "step": 43896 }, { "epoch": 0.04, "learning_rate": 4.976368029070829e-05, "loss": 0.939, "step": 43897 }, { "epoch": 0.04, "learning_rate": 4.9763669516026615e-05, "loss": 1.3013, "step": 43898 }, { "epoch": 0.04, "learning_rate": 4.976365874110048e-05, "loss": 0.8248, "step": 43899 }, { "epoch": 0.04, "learning_rate": 4.9763647965929886e-05, "loss": 0.826, "step": 43900 }, { "epoch": 0.04, "learning_rate": 4.9763637190514834e-05, "loss": 1.2027, "step": 43901 }, { "epoch": 0.04, "learning_rate": 4.9763626414855335e-05, "loss": 0.853, "step": 43902 }, { "epoch": 0.04, "learning_rate": 4.976361563895137e-05, "loss": 1.19, "step": 43903 }, { "epoch": 0.04, "learning_rate": 4.9763604862802956e-05, "loss": 1.03, "step": 43904 }, { "epoch": 0.04, "learning_rate": 4.976359408641008e-05, "loss": 1.2239, "step": 43905 }, { "epoch": 0.04, "learning_rate": 4.9763583309772755e-05, "loss": 1.2044, "step": 43906 }, { "epoch": 0.04, "learning_rate": 4.976357253289097e-05, "loss": 0.7222, "step": 43907 }, { "epoch": 0.04, "learning_rate": 4.976356175576473e-05, "loss": 0.3921, "step": 43908 }, { "epoch": 0.04, "learning_rate": 4.976355097839404e-05, "loss": 0.9228, "step": 43909 }, { "epoch": 0.04, "learning_rate": 4.976354020077888e-05, "loss": 0.9088, "step": 43910 }, { "epoch": 0.04, "learning_rate": 4.976352942291927e-05, "loss": 0.8723, "step": 43911 }, { "epoch": 0.04, "learning_rate": 4.9763518644815223e-05, "loss": 0.9681, "step": 43912 }, { "epoch": 0.04, "learning_rate": 4.97635078664667e-05, "loss": 1.1444, "step": 43913 }, { "epoch": 0.04, "learning_rate": 4.9763497087873736e-05, "loss": 1.1916, "step": 43914 }, { "epoch": 0.04, "learning_rate": 4.9763486309036305e-05, "loss": 1.4106, "step": 43915 }, { "epoch": 0.04, "learning_rate": 4.976347552995443e-05, "loss": 0.9458, "step": 43916 }, { "epoch": 0.04, "learning_rate": 4.976346475062809e-05, "loss": 1.9274, "step": 43917 }, { "epoch": 0.04, "learning_rate": 4.97634539710573e-05, "loss": 1.2838, "step": 43918 }, { "epoch": 0.04, "learning_rate": 4.976344319124205e-05, "loss": 1.0752, "step": 43919 }, { "epoch": 0.04, "learning_rate": 4.976343241118236e-05, "loss": 1.0277, "step": 43920 }, { "epoch": 0.04, "learning_rate": 4.9763421630878206e-05, "loss": 1.0927, "step": 43921 }, { "epoch": 0.04, "learning_rate": 4.97634108503296e-05, "loss": 1.0625, "step": 43922 }, { "epoch": 0.04, "learning_rate": 4.976340006953654e-05, "loss": 1.104, "step": 43923 }, { "epoch": 0.04, "learning_rate": 4.976338928849903e-05, "loss": 1.0398, "step": 43924 }, { "epoch": 0.04, "learning_rate": 4.976337850721706e-05, "loss": 0.9777, "step": 43925 }, { "epoch": 0.04, "learning_rate": 4.9763367725690634e-05, "loss": 1.1854, "step": 43926 }, { "epoch": 0.04, "learning_rate": 4.976335694391976e-05, "loss": 1.07, "step": 43927 }, { "epoch": 0.04, "learning_rate": 4.9763346161904436e-05, "loss": 0.8965, "step": 43928 }, { "epoch": 0.04, "learning_rate": 4.976333537964465e-05, "loss": 0.8061, "step": 43929 }, { "epoch": 0.04, "learning_rate": 4.976332459714042e-05, "loss": 0.9158, "step": 43930 }, { "epoch": 0.04, "learning_rate": 4.9763313814391734e-05, "loss": 2.8359, "step": 43931 }, { "epoch": 0.04, "learning_rate": 4.976330303139859e-05, "loss": 0.719, "step": 43932 }, { "epoch": 0.04, "learning_rate": 4.9763292248161e-05, "loss": 0.9265, "step": 43933 }, { "epoch": 0.04, "learning_rate": 4.976328146467895e-05, "loss": 0.7525, "step": 43934 }, { "epoch": 0.04, "learning_rate": 4.976327068095244e-05, "loss": 0.8726, "step": 43935 }, { "epoch": 0.04, "learning_rate": 4.97632598969815e-05, "loss": 1.0641, "step": 43936 }, { "epoch": 0.04, "learning_rate": 4.976324911276609e-05, "loss": 1.0193, "step": 43937 }, { "epoch": 0.04, "learning_rate": 4.976323832830624e-05, "loss": 0.8713, "step": 43938 }, { "epoch": 0.04, "learning_rate": 4.9763227543601926e-05, "loss": 1.0793, "step": 43939 }, { "epoch": 0.04, "learning_rate": 4.9763216758653166e-05, "loss": 1.025, "step": 43940 }, { "epoch": 0.04, "learning_rate": 4.976320597345995e-05, "loss": 0.9557, "step": 43941 }, { "epoch": 0.04, "learning_rate": 4.976319518802229e-05, "loss": 1.1189, "step": 43942 }, { "epoch": 0.04, "learning_rate": 4.976318440234017e-05, "loss": 0.9168, "step": 43943 }, { "epoch": 0.04, "learning_rate": 4.976317361641361e-05, "loss": 1.0711, "step": 43944 }, { "epoch": 0.04, "learning_rate": 4.976316283024258e-05, "loss": 1.0314, "step": 43945 }, { "epoch": 0.04, "learning_rate": 4.976315204382711e-05, "loss": 1.2995, "step": 43946 }, { "epoch": 0.04, "learning_rate": 4.976314125716719e-05, "loss": 1.0326, "step": 43947 }, { "epoch": 0.04, "learning_rate": 4.9763130470262806e-05, "loss": 1.0762, "step": 43948 }, { "epoch": 0.04, "learning_rate": 4.976311968311399e-05, "loss": 0.8886, "step": 43949 }, { "epoch": 0.04, "learning_rate": 4.976310889572071e-05, "loss": 1.2064, "step": 43950 }, { "epoch": 0.04, "learning_rate": 4.976309810808298e-05, "loss": 0.9347, "step": 43951 }, { "epoch": 0.04, "learning_rate": 4.9763087320200805e-05, "loss": 1.2747, "step": 43952 }, { "epoch": 0.04, "learning_rate": 4.9763076532074173e-05, "loss": 1.1567, "step": 43953 }, { "epoch": 0.04, "learning_rate": 4.9763065743703095e-05, "loss": 1.4465, "step": 43954 }, { "epoch": 0.04, "learning_rate": 4.9763054955087564e-05, "loss": 1.6489, "step": 43955 }, { "epoch": 0.04, "learning_rate": 4.976304416622758e-05, "loss": 1.4523, "step": 43956 }, { "epoch": 0.04, "learning_rate": 4.9763033377123146e-05, "loss": 0.8848, "step": 43957 }, { "epoch": 0.04, "learning_rate": 4.9763022587774266e-05, "loss": 0.7339, "step": 43958 }, { "epoch": 0.04, "learning_rate": 4.976301179818093e-05, "loss": 1.7325, "step": 43959 }, { "epoch": 0.04, "learning_rate": 4.976300100834316e-05, "loss": 1.2833, "step": 43960 }, { "epoch": 0.04, "learning_rate": 4.9762990218260926e-05, "loss": 1.0028, "step": 43961 }, { "epoch": 0.04, "learning_rate": 4.9762979427934245e-05, "loss": 0.531, "step": 43962 }, { "epoch": 0.04, "learning_rate": 4.976296863736311e-05, "loss": 0.6181, "step": 43963 }, { "epoch": 0.04, "learning_rate": 4.976295784654753e-05, "loss": 1.1166, "step": 43964 }, { "epoch": 0.04, "learning_rate": 4.97629470554875e-05, "loss": 0.9569, "step": 43965 }, { "epoch": 0.04, "learning_rate": 4.9762936264183026e-05, "loss": 0.8895, "step": 43966 }, { "epoch": 0.04, "learning_rate": 4.976292547263409e-05, "loss": 1.1227, "step": 43967 }, { "epoch": 0.04, "learning_rate": 4.976291468084071e-05, "loss": 1.1171, "step": 43968 }, { "epoch": 0.04, "learning_rate": 4.9762903888802885e-05, "loss": 0.8516, "step": 43969 }, { "epoch": 0.04, "learning_rate": 4.9762893096520616e-05, "loss": 1.1567, "step": 43970 }, { "epoch": 0.04, "learning_rate": 4.9762882303993885e-05, "loss": 0.9055, "step": 43971 }, { "epoch": 0.04, "learning_rate": 4.976287151122271e-05, "loss": 0.9782, "step": 43972 }, { "epoch": 0.04, "learning_rate": 4.976286071820709e-05, "loss": 0.9851, "step": 43973 }, { "epoch": 0.04, "learning_rate": 4.976284992494702e-05, "loss": 0.8793, "step": 43974 }, { "epoch": 0.04, "learning_rate": 4.97628391314425e-05, "loss": 0.8815, "step": 43975 }, { "epoch": 0.04, "learning_rate": 4.976282833769354e-05, "loss": 0.8063, "step": 43976 }, { "epoch": 0.04, "learning_rate": 4.976281754370011e-05, "loss": 0.8895, "step": 43977 }, { "epoch": 0.04, "learning_rate": 4.9762806749462253e-05, "loss": 0.3056, "step": 43978 }, { "epoch": 0.04, "learning_rate": 4.976279595497994e-05, "loss": 0.8739, "step": 43979 }, { "epoch": 0.04, "learning_rate": 4.9762785160253175e-05, "loss": 1.1898, "step": 43980 }, { "epoch": 0.04, "learning_rate": 4.9762774365281975e-05, "loss": 1.0681, "step": 43981 }, { "epoch": 0.04, "learning_rate": 4.9762763570066316e-05, "loss": 1.0322, "step": 43982 }, { "epoch": 0.04, "learning_rate": 4.9762752774606215e-05, "loss": 1.1954, "step": 43983 }, { "epoch": 0.04, "learning_rate": 4.976274197890167e-05, "loss": 1.1521, "step": 43984 }, { "epoch": 0.04, "learning_rate": 4.9762731182952674e-05, "loss": 1.2078, "step": 43985 }, { "epoch": 0.04, "learning_rate": 4.976272038675922e-05, "loss": 0.9927, "step": 43986 }, { "epoch": 0.04, "learning_rate": 4.976270959032133e-05, "loss": 0.9494, "step": 43987 }, { "epoch": 0.04, "learning_rate": 4.9762698793639e-05, "loss": 0.9876, "step": 43988 }, { "epoch": 0.04, "learning_rate": 4.976268799671221e-05, "loss": 0.8623, "step": 43989 }, { "epoch": 0.04, "learning_rate": 4.976267719954097e-05, "loss": 1.1027, "step": 43990 }, { "epoch": 0.04, "learning_rate": 4.976266640212529e-05, "loss": 0.3778, "step": 43991 }, { "epoch": 0.04, "learning_rate": 4.976265560446517e-05, "loss": 0.838, "step": 43992 }, { "epoch": 0.04, "learning_rate": 4.97626448065606e-05, "loss": 0.85, "step": 43993 }, { "epoch": 0.04, "learning_rate": 4.976263400841158e-05, "loss": 1.2419, "step": 43994 }, { "epoch": 0.04, "learning_rate": 4.976262321001811e-05, "loss": 0.9743, "step": 43995 }, { "epoch": 0.04, "learning_rate": 4.97626124113802e-05, "loss": 0.8844, "step": 43996 }, { "epoch": 0.04, "learning_rate": 4.976260161249784e-05, "loss": 1.0443, "step": 43997 }, { "epoch": 0.04, "learning_rate": 4.976259081337104e-05, "loss": 1.2677, "step": 43998 }, { "epoch": 0.04, "learning_rate": 4.97625800139998e-05, "loss": 1.0578, "step": 43999 }, { "epoch": 0.04, "learning_rate": 4.976256921438409e-05, "loss": 0.7653, "step": 44000 }, { "epoch": 0.04, "eval_loss": 1.0828602313995361, "eval_runtime": 124.6543, "eval_samples_per_second": 11.111, "eval_steps_per_second": 5.559, "step": 44000 }, { "epoch": 0.04, "learning_rate": 4.9762558414523954e-05, "loss": 0.592, "step": 44001 }, { "epoch": 0.04, "learning_rate": 4.976254761441937e-05, "loss": 0.5985, "step": 44002 }, { "epoch": 0.04, "learning_rate": 4.976253681407034e-05, "loss": 0.5771, "step": 44003 }, { "epoch": 0.04, "learning_rate": 4.9762526013476865e-05, "loss": 0.5739, "step": 44004 }, { "epoch": 0.04, "learning_rate": 4.976251521263894e-05, "loss": 0.5662, "step": 44005 }, { "epoch": 0.04, "learning_rate": 4.976250441155657e-05, "loss": 0.5654, "step": 44006 }, { "epoch": 0.04, "learning_rate": 4.976249361022975e-05, "loss": 0.5641, "step": 44007 }, { "epoch": 0.04, "learning_rate": 4.97624828086585e-05, "loss": 0.5828, "step": 44008 }, { "epoch": 0.04, "learning_rate": 4.97624720068428e-05, "loss": 0.5704, "step": 44009 }, { "epoch": 0.04, "learning_rate": 4.976246120478265e-05, "loss": 0.7572, "step": 44010 }, { "epoch": 0.04, "learning_rate": 4.976245040247806e-05, "loss": 1.567, "step": 44011 }, { "epoch": 0.04, "learning_rate": 4.9762439599929025e-05, "loss": 1.1449, "step": 44012 }, { "epoch": 0.04, "learning_rate": 4.9762428797135544e-05, "loss": 1.0489, "step": 44013 }, { "epoch": 0.04, "learning_rate": 4.976241799409762e-05, "loss": 0.9013, "step": 44014 }, { "epoch": 0.04, "learning_rate": 4.9762407190815255e-05, "loss": 0.841, "step": 44015 }, { "epoch": 0.04, "learning_rate": 4.976239638728843e-05, "loss": 1.0076, "step": 44016 }, { "epoch": 0.04, "learning_rate": 4.976238558351718e-05, "loss": 1.3432, "step": 44017 }, { "epoch": 0.04, "learning_rate": 4.976237477950148e-05, "loss": 1.1507, "step": 44018 }, { "epoch": 0.04, "learning_rate": 4.976236397524133e-05, "loss": 1.9489, "step": 44019 }, { "epoch": 0.04, "learning_rate": 4.976235317073675e-05, "loss": 1.9206, "step": 44020 }, { "epoch": 0.04, "learning_rate": 4.976234236598772e-05, "loss": 1.0969, "step": 44021 }, { "epoch": 0.04, "learning_rate": 4.9762331560994245e-05, "loss": 1.2279, "step": 44022 }, { "epoch": 0.04, "learning_rate": 4.976232075575632e-05, "loss": 0.9635, "step": 44023 }, { "epoch": 0.04, "learning_rate": 4.9762309950273965e-05, "loss": 1.0792, "step": 44024 }, { "epoch": 0.04, "learning_rate": 4.976229914454716e-05, "loss": 0.9323, "step": 44025 }, { "epoch": 0.04, "learning_rate": 4.976228833857591e-05, "loss": 0.9931, "step": 44026 }, { "epoch": 0.04, "learning_rate": 4.976227753236022e-05, "loss": 1.285, "step": 44027 }, { "epoch": 0.04, "learning_rate": 4.976226672590008e-05, "loss": 0.9303, "step": 44028 }, { "epoch": 0.04, "learning_rate": 4.976225591919551e-05, "loss": 1.1922, "step": 44029 }, { "epoch": 0.04, "learning_rate": 4.9762245112246495e-05, "loss": 1.2473, "step": 44030 }, { "epoch": 0.04, "learning_rate": 4.9762234305053037e-05, "loss": 1.1785, "step": 44031 }, { "epoch": 0.04, "learning_rate": 4.976222349761513e-05, "loss": 0.9143, "step": 44032 }, { "epoch": 0.04, "learning_rate": 4.9762212689932787e-05, "loss": 0.9142, "step": 44033 }, { "epoch": 0.04, "learning_rate": 4.9762201882006e-05, "loss": 1.1529, "step": 44034 }, { "epoch": 0.04, "learning_rate": 4.9762191073834776e-05, "loss": 0.9158, "step": 44035 }, { "epoch": 0.04, "learning_rate": 4.9762180265419104e-05, "loss": 1.0232, "step": 44036 }, { "epoch": 0.04, "learning_rate": 4.976216945675899e-05, "loss": 0.8749, "step": 44037 }, { "epoch": 0.04, "learning_rate": 4.976215864785443e-05, "loss": 0.5178, "step": 44038 }, { "epoch": 0.04, "learning_rate": 4.976214783870543e-05, "loss": 0.8685, "step": 44039 }, { "epoch": 0.04, "learning_rate": 4.976213702931199e-05, "loss": 0.8895, "step": 44040 }, { "epoch": 0.04, "learning_rate": 4.976212621967412e-05, "loss": 0.9726, "step": 44041 }, { "epoch": 0.04, "learning_rate": 4.97621154097918e-05, "loss": 0.7624, "step": 44042 }, { "epoch": 0.04, "learning_rate": 4.976210459966504e-05, "loss": 0.5021, "step": 44043 }, { "epoch": 0.04, "learning_rate": 4.976209378929383e-05, "loss": 0.6401, "step": 44044 }, { "epoch": 0.04, "learning_rate": 4.976208297867819e-05, "loss": 0.3417, "step": 44045 }, { "epoch": 0.04, "learning_rate": 4.976207216781811e-05, "loss": 0.9576, "step": 44046 }, { "epoch": 0.04, "learning_rate": 4.976206135671358e-05, "loss": 0.9135, "step": 44047 }, { "epoch": 0.04, "learning_rate": 4.9762050545364615e-05, "loss": 0.8255, "step": 44048 }, { "epoch": 0.04, "learning_rate": 4.976203973377121e-05, "loss": 0.994, "step": 44049 }, { "epoch": 0.04, "learning_rate": 4.976202892193336e-05, "loss": 0.9562, "step": 44050 }, { "epoch": 0.04, "learning_rate": 4.976201810985107e-05, "loss": 1.1059, "step": 44051 }, { "epoch": 0.04, "learning_rate": 4.976200729752435e-05, "loss": 1.0743, "step": 44052 }, { "epoch": 0.04, "learning_rate": 4.9761996484953174e-05, "loss": 0.8494, "step": 44053 }, { "epoch": 0.04, "learning_rate": 4.9761985672137565e-05, "loss": 1.1504, "step": 44054 }, { "epoch": 0.04, "learning_rate": 4.976197485907752e-05, "loss": 0.8593, "step": 44055 }, { "epoch": 0.04, "learning_rate": 4.9761964045773034e-05, "loss": 0.9904, "step": 44056 }, { "epoch": 0.04, "learning_rate": 4.976195323222411e-05, "loss": 0.9992, "step": 44057 }, { "epoch": 0.04, "learning_rate": 4.976194241843074e-05, "loss": 1.1794, "step": 44058 }, { "epoch": 0.04, "learning_rate": 4.976193160439293e-05, "loss": 0.7106, "step": 44059 }, { "epoch": 0.04, "learning_rate": 4.97619207901107e-05, "loss": 0.9334, "step": 44060 }, { "epoch": 0.04, "learning_rate": 4.976190997558401e-05, "loss": 0.8217, "step": 44061 }, { "epoch": 0.04, "learning_rate": 4.9761899160812884e-05, "loss": 1.1796, "step": 44062 }, { "epoch": 0.04, "learning_rate": 4.976188834579732e-05, "loss": 1.0448, "step": 44063 }, { "epoch": 0.04, "learning_rate": 4.976187753053732e-05, "loss": 1.0609, "step": 44064 }, { "epoch": 0.04, "learning_rate": 4.976186671503288e-05, "loss": 0.8022, "step": 44065 }, { "epoch": 0.04, "learning_rate": 4.9761855899283996e-05, "loss": 1.2078, "step": 44066 }, { "epoch": 0.04, "learning_rate": 4.976184508329068e-05, "loss": 0.6283, "step": 44067 }, { "epoch": 0.04, "learning_rate": 4.976183426705293e-05, "loss": 0.7639, "step": 44068 }, { "epoch": 0.04, "learning_rate": 4.976182345057073e-05, "loss": 0.7874, "step": 44069 }, { "epoch": 0.04, "learning_rate": 4.9761812633844093e-05, "loss": 0.5723, "step": 44070 }, { "epoch": 0.04, "learning_rate": 4.976180181687303e-05, "loss": 1.1733, "step": 44071 }, { "epoch": 0.04, "learning_rate": 4.976179099965752e-05, "loss": 1.1344, "step": 44072 }, { "epoch": 0.04, "learning_rate": 4.9761780182197573e-05, "loss": 1.3794, "step": 44073 }, { "epoch": 0.04, "learning_rate": 4.976176936449318e-05, "loss": 1.4557, "step": 44074 }, { "epoch": 0.04, "learning_rate": 4.9761758546544364e-05, "loss": 1.277, "step": 44075 }, { "epoch": 0.04, "learning_rate": 4.9761747728351106e-05, "loss": 1.1004, "step": 44076 }, { "epoch": 0.04, "learning_rate": 4.97617369099134e-05, "loss": 1.0342, "step": 44077 }, { "epoch": 0.04, "learning_rate": 4.976172609123127e-05, "loss": 0.9638, "step": 44078 }, { "epoch": 0.04, "learning_rate": 4.97617152723047e-05, "loss": 1.1283, "step": 44079 }, { "epoch": 0.04, "learning_rate": 4.9761704453133685e-05, "loss": 1.053, "step": 44080 }, { "epoch": 0.04, "learning_rate": 4.9761693633718246e-05, "loss": 0.97, "step": 44081 }, { "epoch": 0.04, "learning_rate": 4.9761682814058353e-05, "loss": 0.7274, "step": 44082 }, { "epoch": 0.04, "learning_rate": 4.9761671994154034e-05, "loss": 0.8814, "step": 44083 }, { "epoch": 0.04, "learning_rate": 4.976166117400528e-05, "loss": 1.1686, "step": 44084 }, { "epoch": 0.04, "learning_rate": 4.976165035361208e-05, "loss": 1.2396, "step": 44085 }, { "epoch": 0.04, "learning_rate": 4.976163953297445e-05, "loss": 1.3917, "step": 44086 }, { "epoch": 0.04, "learning_rate": 4.9761628712092384e-05, "loss": 1.0037, "step": 44087 }, { "epoch": 0.04, "learning_rate": 4.9761617890965884e-05, "loss": 1.3063, "step": 44088 }, { "epoch": 0.04, "learning_rate": 4.976160706959494e-05, "loss": 1.1423, "step": 44089 }, { "epoch": 0.04, "learning_rate": 4.9761596247979565e-05, "loss": 0.7969, "step": 44090 }, { "epoch": 0.04, "learning_rate": 4.976158542611975e-05, "loss": 0.7192, "step": 44091 }, { "epoch": 0.04, "learning_rate": 4.9761574604015505e-05, "loss": 0.9672, "step": 44092 }, { "epoch": 0.04, "learning_rate": 4.976156378166682e-05, "loss": 0.9671, "step": 44093 }, { "epoch": 0.04, "learning_rate": 4.97615529590737e-05, "loss": 1.2128, "step": 44094 }, { "epoch": 0.04, "learning_rate": 4.9761542136236145e-05, "loss": 1.0147, "step": 44095 }, { "epoch": 0.04, "learning_rate": 4.976153131315415e-05, "loss": 0.5925, "step": 44096 }, { "epoch": 0.04, "learning_rate": 4.976152048982773e-05, "loss": 1.0732, "step": 44097 }, { "epoch": 0.04, "learning_rate": 4.9761509666256864e-05, "loss": 1.1816, "step": 44098 }, { "epoch": 0.04, "learning_rate": 4.976149884244157e-05, "loss": 0.9308, "step": 44099 }, { "epoch": 0.04, "learning_rate": 4.976148801838184e-05, "loss": 0.9243, "step": 44100 }, { "epoch": 0.04, "learning_rate": 4.976147719407767e-05, "loss": 1.0527, "step": 44101 }, { "epoch": 0.04, "learning_rate": 4.976146636952907e-05, "loss": 0.8061, "step": 44102 }, { "epoch": 0.04, "learning_rate": 4.9761455544736034e-05, "loss": 1.1948, "step": 44103 }, { "epoch": 0.04, "learning_rate": 4.976144471969856e-05, "loss": 0.975, "step": 44104 }, { "epoch": 0.04, "learning_rate": 4.976143389441665e-05, "loss": 1.1497, "step": 44105 }, { "epoch": 0.04, "learning_rate": 4.976142306889032e-05, "loss": 0.9723, "step": 44106 }, { "epoch": 0.04, "learning_rate": 4.9761412243119544e-05, "loss": 0.959, "step": 44107 }, { "epoch": 0.04, "learning_rate": 4.9761401417104336e-05, "loss": 1.1215, "step": 44108 }, { "epoch": 0.04, "learning_rate": 4.9761390590844695e-05, "loss": 0.9381, "step": 44109 }, { "epoch": 0.04, "learning_rate": 4.976137976434061e-05, "loss": 0.9153, "step": 44110 }, { "epoch": 0.04, "learning_rate": 4.97613689375921e-05, "loss": 0.717, "step": 44111 }, { "epoch": 0.04, "learning_rate": 4.976135811059915e-05, "loss": 1.0836, "step": 44112 }, { "epoch": 0.04, "learning_rate": 4.9761347283361783e-05, "loss": 1.0914, "step": 44113 }, { "epoch": 0.04, "learning_rate": 4.976133645587997e-05, "loss": 0.7463, "step": 44114 }, { "epoch": 0.04, "learning_rate": 4.976132562815372e-05, "loss": 0.9687, "step": 44115 }, { "epoch": 0.04, "learning_rate": 4.9761314800183046e-05, "loss": 1.0062, "step": 44116 }, { "epoch": 0.04, "learning_rate": 4.976130397196793e-05, "loss": 1.1384, "step": 44117 }, { "epoch": 0.04, "learning_rate": 4.976129314350839e-05, "loss": 1.021, "step": 44118 }, { "epoch": 0.04, "learning_rate": 4.976128231480441e-05, "loss": 0.8197, "step": 44119 }, { "epoch": 0.04, "learning_rate": 4.9761271485856e-05, "loss": 1.0533, "step": 44120 }, { "epoch": 0.04, "learning_rate": 4.976126065666316e-05, "loss": 0.9533, "step": 44121 }, { "epoch": 0.04, "learning_rate": 4.9761249827225876e-05, "loss": 0.7981, "step": 44122 }, { "epoch": 0.04, "learning_rate": 4.976123899754417e-05, "loss": 0.9784, "step": 44123 }, { "epoch": 0.04, "learning_rate": 4.9761228167618026e-05, "loss": 1.2424, "step": 44124 }, { "epoch": 0.04, "learning_rate": 4.976121733744745e-05, "loss": 1.0018, "step": 44125 }, { "epoch": 0.04, "learning_rate": 4.976120650703244e-05, "loss": 1.0299, "step": 44126 }, { "epoch": 0.04, "learning_rate": 4.976119567637301e-05, "loss": 1.05, "step": 44127 }, { "epoch": 0.04, "learning_rate": 4.976118484546913e-05, "loss": 1.2863, "step": 44128 }, { "epoch": 0.04, "learning_rate": 4.976117401432083e-05, "loss": 1.3206, "step": 44129 }, { "epoch": 0.04, "learning_rate": 4.97611631829281e-05, "loss": 1.0551, "step": 44130 }, { "epoch": 0.04, "learning_rate": 4.976115235129093e-05, "loss": 1.0514, "step": 44131 }, { "epoch": 0.04, "learning_rate": 4.976114151940934e-05, "loss": 0.9384, "step": 44132 }, { "epoch": 0.04, "learning_rate": 4.976113068728331e-05, "loss": 1.2602, "step": 44133 }, { "epoch": 0.04, "learning_rate": 4.976111985491284e-05, "loss": 0.951, "step": 44134 }, { "epoch": 0.04, "learning_rate": 4.976110902229795e-05, "loss": 1.071, "step": 44135 }, { "epoch": 0.04, "learning_rate": 4.976109818943863e-05, "loss": 1.0218, "step": 44136 }, { "epoch": 0.04, "learning_rate": 4.9761087356334873e-05, "loss": 1.0886, "step": 44137 }, { "epoch": 0.04, "learning_rate": 4.976107652298669e-05, "loss": 1.0422, "step": 44138 }, { "epoch": 0.04, "learning_rate": 4.976106568939407e-05, "loss": 0.9577, "step": 44139 }, { "epoch": 0.04, "learning_rate": 4.976105485555702e-05, "loss": 1.5232, "step": 44140 }, { "epoch": 0.04, "learning_rate": 4.9761044021475545e-05, "loss": 0.9271, "step": 44141 }, { "epoch": 0.04, "learning_rate": 4.9761033187149636e-05, "loss": 0.7668, "step": 44142 }, { "epoch": 0.04, "learning_rate": 4.97610223525793e-05, "loss": 1.1506, "step": 44143 }, { "epoch": 0.04, "learning_rate": 4.9761011517764525e-05, "loss": 0.6363, "step": 44144 }, { "epoch": 0.04, "learning_rate": 4.976100068270533e-05, "loss": 0.9266, "step": 44145 }, { "epoch": 0.04, "learning_rate": 4.9760989847401695e-05, "loss": 1.1848, "step": 44146 }, { "epoch": 0.04, "learning_rate": 4.976097901185364e-05, "loss": 1.1761, "step": 44147 }, { "epoch": 0.04, "learning_rate": 4.976096817606115e-05, "loss": 1.1871, "step": 44148 }, { "epoch": 0.04, "learning_rate": 4.976095734002423e-05, "loss": 1.3471, "step": 44149 }, { "epoch": 0.04, "learning_rate": 4.976094650374288e-05, "loss": 1.0561, "step": 44150 }, { "epoch": 0.04, "learning_rate": 4.9760935667217104e-05, "loss": 1.2854, "step": 44151 }, { "epoch": 0.04, "learning_rate": 4.976092483044689e-05, "loss": 1.2822, "step": 44152 }, { "epoch": 0.04, "learning_rate": 4.9760913993432254e-05, "loss": 1.3523, "step": 44153 }, { "epoch": 0.04, "learning_rate": 4.9760903156173185e-05, "loss": 0.9657, "step": 44154 }, { "epoch": 0.04, "learning_rate": 4.976089231866969e-05, "loss": 1.0776, "step": 44155 }, { "epoch": 0.04, "learning_rate": 4.976088148092176e-05, "loss": 0.6714, "step": 44156 }, { "epoch": 0.04, "learning_rate": 4.97608706429294e-05, "loss": 1.4955, "step": 44157 }, { "epoch": 0.04, "learning_rate": 4.9760859804692624e-05, "loss": 1.0876, "step": 44158 }, { "epoch": 0.04, "learning_rate": 4.976084896621141e-05, "loss": 1.0059, "step": 44159 }, { "epoch": 0.04, "learning_rate": 4.9760838127485774e-05, "loss": 1.0153, "step": 44160 }, { "epoch": 0.04, "learning_rate": 4.97608272885157e-05, "loss": 0.814, "step": 44161 }, { "epoch": 0.04, "learning_rate": 4.9760816449301204e-05, "loss": 0.8434, "step": 44162 }, { "epoch": 0.04, "learning_rate": 4.9760805609842275e-05, "loss": 1.1125, "step": 44163 }, { "epoch": 0.04, "learning_rate": 4.976079477013892e-05, "loss": 1.1781, "step": 44164 }, { "epoch": 0.04, "learning_rate": 4.976078393019113e-05, "loss": 0.8701, "step": 44165 }, { "epoch": 0.04, "learning_rate": 4.9760773089998924e-05, "loss": 1.3188, "step": 44166 }, { "epoch": 0.04, "learning_rate": 4.976076224956229e-05, "loss": 1.1876, "step": 44167 }, { "epoch": 0.04, "learning_rate": 4.976075140888122e-05, "loss": 0.7875, "step": 44168 }, { "epoch": 0.04, "learning_rate": 4.9760740567955726e-05, "loss": 0.0912, "step": 44169 }, { "epoch": 0.04, "learning_rate": 4.9760729726785804e-05, "loss": 0.0765, "step": 44170 }, { "epoch": 0.04, "learning_rate": 4.9760718885371456e-05, "loss": 0.0613, "step": 44171 }, { "epoch": 0.04, "learning_rate": 4.976070804371268e-05, "loss": 0.064, "step": 44172 }, { "epoch": 0.04, "learning_rate": 4.9760697201809474e-05, "loss": 0.0575, "step": 44173 }, { "epoch": 0.04, "learning_rate": 4.9760686359661846e-05, "loss": 0.0435, "step": 44174 }, { "epoch": 0.04, "learning_rate": 4.9760675517269785e-05, "loss": 0.0477, "step": 44175 }, { "epoch": 0.04, "learning_rate": 4.97606646746333e-05, "loss": 0.0519, "step": 44176 }, { "epoch": 0.04, "learning_rate": 4.976065383175239e-05, "loss": 0.0475, "step": 44177 }, { "epoch": 0.04, "learning_rate": 4.976064298862705e-05, "loss": 0.0619, "step": 44178 }, { "epoch": 0.04, "learning_rate": 4.976063214525728e-05, "loss": 0.0459, "step": 44179 }, { "epoch": 0.04, "learning_rate": 4.976062130164309e-05, "loss": 0.0328, "step": 44180 }, { "epoch": 0.04, "learning_rate": 4.976061045778447e-05, "loss": 0.0409, "step": 44181 }, { "epoch": 0.04, "learning_rate": 4.9760599613681425e-05, "loss": 0.0519, "step": 44182 }, { "epoch": 0.04, "learning_rate": 4.976058876933396e-05, "loss": 0.0442, "step": 44183 }, { "epoch": 0.04, "learning_rate": 4.976057792474206e-05, "loss": 0.0363, "step": 44184 }, { "epoch": 0.04, "learning_rate": 4.976056707990574e-05, "loss": 0.0389, "step": 44185 }, { "epoch": 0.04, "learning_rate": 4.976055623482498e-05, "loss": 0.0315, "step": 44186 }, { "epoch": 0.04, "learning_rate": 4.9760545389499816e-05, "loss": 0.0257, "step": 44187 }, { "epoch": 0.04, "learning_rate": 4.9760534543930215e-05, "loss": 0.0252, "step": 44188 }, { "epoch": 0.04, "learning_rate": 4.976052369811618e-05, "loss": 0.0427, "step": 44189 }, { "epoch": 0.04, "learning_rate": 4.9760512852057735e-05, "loss": 0.0568, "step": 44190 }, { "epoch": 0.04, "learning_rate": 4.976050200575486e-05, "loss": 0.0601, "step": 44191 }, { "epoch": 0.04, "learning_rate": 4.9760491159207555e-05, "loss": 0.0584, "step": 44192 }, { "epoch": 0.04, "learning_rate": 4.976048031241582e-05, "loss": 0.0693, "step": 44193 }, { "epoch": 0.04, "learning_rate": 4.9760469465379675e-05, "loss": 0.0478, "step": 44194 }, { "epoch": 0.04, "learning_rate": 4.9760458618099096e-05, "loss": 0.0367, "step": 44195 }, { "epoch": 0.04, "learning_rate": 4.9760447770574096e-05, "loss": 0.0461, "step": 44196 }, { "epoch": 0.04, "learning_rate": 4.976043692280467e-05, "loss": 0.3163, "step": 44197 }, { "epoch": 0.04, "learning_rate": 4.9760426074790824e-05, "loss": 1.1577, "step": 44198 }, { "epoch": 0.04, "learning_rate": 4.9760415226532544e-05, "loss": 1.2372, "step": 44199 }, { "epoch": 0.04, "learning_rate": 4.9760404378029845e-05, "loss": 1.2552, "step": 44200 }, { "epoch": 0.04, "learning_rate": 4.976039352928272e-05, "loss": 0.9884, "step": 44201 }, { "epoch": 0.04, "learning_rate": 4.9760382680291174e-05, "loss": 1.1368, "step": 44202 }, { "epoch": 0.04, "learning_rate": 4.9760371831055195e-05, "loss": 1.4516, "step": 44203 }, { "epoch": 0.04, "learning_rate": 4.97603609815748e-05, "loss": 1.2783, "step": 44204 }, { "epoch": 0.04, "learning_rate": 4.9760350131849984e-05, "loss": 0.8059, "step": 44205 }, { "epoch": 0.04, "learning_rate": 4.976033928188074e-05, "loss": 0.8574, "step": 44206 }, { "epoch": 0.04, "learning_rate": 4.976032843166707e-05, "loss": 0.7806, "step": 44207 }, { "epoch": 0.04, "learning_rate": 4.976031758120898e-05, "loss": 0.6277, "step": 44208 }, { "epoch": 0.04, "learning_rate": 4.976030673050647e-05, "loss": 0.7787, "step": 44209 }, { "epoch": 0.04, "learning_rate": 4.9760295879559526e-05, "loss": 0.732, "step": 44210 }, { "epoch": 0.04, "learning_rate": 4.976028502836816e-05, "loss": 0.8467, "step": 44211 }, { "epoch": 0.04, "learning_rate": 4.976027417693238e-05, "loss": 0.8257, "step": 44212 }, { "epoch": 0.04, "learning_rate": 4.976026332525217e-05, "loss": 0.7253, "step": 44213 }, { "epoch": 0.04, "learning_rate": 4.976025247332754e-05, "loss": 0.3938, "step": 44214 }, { "epoch": 0.04, "learning_rate": 4.976024162115849e-05, "loss": 0.9221, "step": 44215 }, { "epoch": 0.04, "learning_rate": 4.976023076874501e-05, "loss": 0.4342, "step": 44216 }, { "epoch": 0.04, "learning_rate": 4.9760219916087116e-05, "loss": 0.0892, "step": 44217 }, { "epoch": 0.04, "learning_rate": 4.976020906318479e-05, "loss": 0.0775, "step": 44218 }, { "epoch": 0.04, "learning_rate": 4.976019821003805e-05, "loss": 0.7691, "step": 44219 }, { "epoch": 0.04, "learning_rate": 4.9760187356646884e-05, "loss": 0.6447, "step": 44220 }, { "epoch": 0.04, "learning_rate": 4.97601765030113e-05, "loss": 0.9974, "step": 44221 }, { "epoch": 0.04, "learning_rate": 4.9760165649131284e-05, "loss": 0.6943, "step": 44222 }, { "epoch": 0.04, "learning_rate": 4.9760154795006855e-05, "loss": 1.1356, "step": 44223 }, { "epoch": 0.04, "learning_rate": 4.9760143940638005e-05, "loss": 0.9912, "step": 44224 }, { "epoch": 0.04, "learning_rate": 4.976013308602473e-05, "loss": 1.3537, "step": 44225 }, { "epoch": 0.04, "learning_rate": 4.976012223116703e-05, "loss": 1.2782, "step": 44226 }, { "epoch": 0.04, "learning_rate": 4.976011137606492e-05, "loss": 1.4961, "step": 44227 }, { "epoch": 0.04, "learning_rate": 4.9760100520718375e-05, "loss": 1.1146, "step": 44228 }, { "epoch": 0.04, "learning_rate": 4.976008966512742e-05, "loss": 1.1507, "step": 44229 }, { "epoch": 0.04, "learning_rate": 4.976007880929203e-05, "loss": 1.1987, "step": 44230 }, { "epoch": 0.04, "learning_rate": 4.976006795321223e-05, "loss": 1.4091, "step": 44231 }, { "epoch": 0.04, "learning_rate": 4.9760057096888007e-05, "loss": 1.1412, "step": 44232 }, { "epoch": 0.04, "learning_rate": 4.976004624031936e-05, "loss": 1.1057, "step": 44233 }, { "epoch": 0.04, "learning_rate": 4.97600353835063e-05, "loss": 0.9919, "step": 44234 }, { "epoch": 0.04, "learning_rate": 4.976002452644881e-05, "loss": 1.0438, "step": 44235 }, { "epoch": 0.04, "learning_rate": 4.976001366914691e-05, "loss": 1.0243, "step": 44236 }, { "epoch": 0.04, "learning_rate": 4.976000281160057e-05, "loss": 0.9582, "step": 44237 }, { "epoch": 0.04, "learning_rate": 4.9759991953809826e-05, "loss": 0.9205, "step": 44238 }, { "epoch": 0.04, "learning_rate": 4.975998109577466e-05, "loss": 1.3505, "step": 44239 }, { "epoch": 0.04, "learning_rate": 4.975997023749507e-05, "loss": 0.9823, "step": 44240 }, { "epoch": 0.04, "learning_rate": 4.9759959378971064e-05, "loss": 0.9657, "step": 44241 }, { "epoch": 0.04, "learning_rate": 4.975994852020264e-05, "loss": 1.1239, "step": 44242 }, { "epoch": 0.04, "learning_rate": 4.975993766118979e-05, "loss": 0.7258, "step": 44243 }, { "epoch": 0.04, "learning_rate": 4.9759926801932524e-05, "loss": 0.811, "step": 44244 }, { "epoch": 0.04, "learning_rate": 4.975991594243084e-05, "loss": 0.6553, "step": 44245 }, { "epoch": 0.04, "learning_rate": 4.975990508268473e-05, "loss": 0.7437, "step": 44246 }, { "epoch": 0.04, "learning_rate": 4.9759894222694206e-05, "loss": 1.114, "step": 44247 }, { "epoch": 0.04, "learning_rate": 4.975988336245926e-05, "loss": 1.3326, "step": 44248 }, { "epoch": 0.04, "learning_rate": 4.97598725019799e-05, "loss": 1.3546, "step": 44249 }, { "epoch": 0.04, "learning_rate": 4.975986164125611e-05, "loss": 0.9978, "step": 44250 }, { "epoch": 0.04, "learning_rate": 4.975985078028792e-05, "loss": 0.5191, "step": 44251 }, { "epoch": 0.04, "learning_rate": 4.975983991907529e-05, "loss": 0.8519, "step": 44252 }, { "epoch": 0.04, "learning_rate": 4.9759829057618254e-05, "loss": 1.2441, "step": 44253 }, { "epoch": 0.04, "learning_rate": 4.9759818195916796e-05, "loss": 0.7463, "step": 44254 }, { "epoch": 0.04, "learning_rate": 4.975980733397092e-05, "loss": 0.972, "step": 44255 }, { "epoch": 0.04, "learning_rate": 4.975979647178063e-05, "loss": 0.9985, "step": 44256 }, { "epoch": 0.04, "learning_rate": 4.975978560934591e-05, "loss": 1.0512, "step": 44257 }, { "epoch": 0.04, "learning_rate": 4.975977474666678e-05, "loss": 0.5471, "step": 44258 }, { "epoch": 0.04, "learning_rate": 4.975976388374323e-05, "loss": 0.8604, "step": 44259 }, { "epoch": 0.04, "learning_rate": 4.975975302057526e-05, "loss": 1.3499, "step": 44260 }, { "epoch": 0.04, "learning_rate": 4.975974215716288e-05, "loss": 1.2628, "step": 44261 }, { "epoch": 0.04, "learning_rate": 4.975973129350607e-05, "loss": 1.0545, "step": 44262 }, { "epoch": 0.04, "learning_rate": 4.975972042960486e-05, "loss": 0.9905, "step": 44263 }, { "epoch": 0.04, "learning_rate": 4.9759709565459214e-05, "loss": 0.9398, "step": 44264 }, { "epoch": 0.04, "learning_rate": 4.9759698701069165e-05, "loss": 0.986, "step": 44265 }, { "epoch": 0.04, "learning_rate": 4.975968783643469e-05, "loss": 1.6306, "step": 44266 }, { "epoch": 0.04, "learning_rate": 4.97596769715558e-05, "loss": 1.0801, "step": 44267 }, { "epoch": 0.04, "learning_rate": 4.975966610643249e-05, "loss": 1.0212, "step": 44268 }, { "epoch": 0.04, "learning_rate": 4.975965524106477e-05, "loss": 1.0442, "step": 44269 }, { "epoch": 0.04, "learning_rate": 4.975964437545263e-05, "loss": 1.0723, "step": 44270 }, { "epoch": 0.04, "learning_rate": 4.975963350959607e-05, "loss": 0.7761, "step": 44271 }, { "epoch": 0.04, "learning_rate": 4.9759622643495095e-05, "loss": 0.7226, "step": 44272 }, { "epoch": 0.04, "learning_rate": 4.975961177714971e-05, "loss": 0.5996, "step": 44273 }, { "epoch": 0.04, "learning_rate": 4.97596009105599e-05, "loss": 1.0736, "step": 44274 }, { "epoch": 0.04, "learning_rate": 4.975959004372568e-05, "loss": 1.3396, "step": 44275 }, { "epoch": 0.04, "learning_rate": 4.9759579176647033e-05, "loss": 1.3403, "step": 44276 }, { "epoch": 0.04, "learning_rate": 4.975956830932398e-05, "loss": 0.9343, "step": 44277 }, { "epoch": 0.04, "learning_rate": 4.975955744175651e-05, "loss": 0.6843, "step": 44278 }, { "epoch": 0.04, "learning_rate": 4.975954657394462e-05, "loss": 1.1279, "step": 44279 }, { "epoch": 0.04, "learning_rate": 4.9759535705888316e-05, "loss": 1.0705, "step": 44280 }, { "epoch": 0.04, "learning_rate": 4.97595248375876e-05, "loss": 1.0829, "step": 44281 }, { "epoch": 0.04, "learning_rate": 4.975951396904246e-05, "loss": 0.9895, "step": 44282 }, { "epoch": 0.04, "learning_rate": 4.9759503100252914e-05, "loss": 1.115, "step": 44283 }, { "epoch": 0.04, "learning_rate": 4.975949223121895e-05, "loss": 1.0899, "step": 44284 }, { "epoch": 0.04, "learning_rate": 4.9759481361940565e-05, "loss": 0.8426, "step": 44285 }, { "epoch": 0.04, "learning_rate": 4.975947049241777e-05, "loss": 1.0312, "step": 44286 }, { "epoch": 0.04, "learning_rate": 4.9759459622650557e-05, "loss": 1.4461, "step": 44287 }, { "epoch": 0.04, "learning_rate": 4.9759448752638926e-05, "loss": 1.1075, "step": 44288 }, { "epoch": 0.04, "learning_rate": 4.975943788238289e-05, "loss": 1.0394, "step": 44289 }, { "epoch": 0.04, "learning_rate": 4.975942701188243e-05, "loss": 1.0553, "step": 44290 }, { "epoch": 0.04, "learning_rate": 4.975941614113756e-05, "loss": 1.3153, "step": 44291 }, { "epoch": 0.04, "learning_rate": 4.9759405270148274e-05, "loss": 0.8663, "step": 44292 }, { "epoch": 0.04, "learning_rate": 4.975939439891458e-05, "loss": 1.183, "step": 44293 }, { "epoch": 0.04, "learning_rate": 4.975938352743646e-05, "loss": 1.6465, "step": 44294 }, { "epoch": 0.04, "learning_rate": 4.9759372655713934e-05, "loss": 1.1937, "step": 44295 }, { "epoch": 0.04, "learning_rate": 4.975936178374699e-05, "loss": 1.1428, "step": 44296 }, { "epoch": 0.04, "learning_rate": 4.975935091153564e-05, "loss": 1.2035, "step": 44297 }, { "epoch": 0.04, "learning_rate": 4.975934003907986e-05, "loss": 0.9618, "step": 44298 }, { "epoch": 0.04, "learning_rate": 4.975932916637969e-05, "loss": 0.8727, "step": 44299 }, { "epoch": 0.04, "learning_rate": 4.975931829343509e-05, "loss": 0.8357, "step": 44300 }, { "epoch": 0.04, "learning_rate": 4.975930742024607e-05, "loss": 1.2124, "step": 44301 }, { "epoch": 0.04, "learning_rate": 4.975929654681265e-05, "loss": 1.0133, "step": 44302 }, { "epoch": 0.04, "learning_rate": 4.9759285673134815e-05, "loss": 0.9626, "step": 44303 }, { "epoch": 0.04, "learning_rate": 4.975927479921256e-05, "loss": 1.3344, "step": 44304 }, { "epoch": 0.04, "learning_rate": 4.9759263925045897e-05, "loss": 1.2246, "step": 44305 }, { "epoch": 0.04, "learning_rate": 4.9759253050634824e-05, "loss": 0.9112, "step": 44306 }, { "epoch": 0.04, "learning_rate": 4.975924217597933e-05, "loss": 0.9912, "step": 44307 }, { "epoch": 0.04, "learning_rate": 4.975923130107943e-05, "loss": 1.0901, "step": 44308 }, { "epoch": 0.04, "learning_rate": 4.9759220425935116e-05, "loss": 1.2172, "step": 44309 }, { "epoch": 0.04, "learning_rate": 4.9759209550546385e-05, "loss": 1.2216, "step": 44310 }, { "epoch": 0.04, "learning_rate": 4.975919867491324e-05, "loss": 1.0841, "step": 44311 }, { "epoch": 0.04, "learning_rate": 4.975918779903569e-05, "loss": 1.3634, "step": 44312 }, { "epoch": 0.04, "learning_rate": 4.975917692291373e-05, "loss": 0.9954, "step": 44313 }, { "epoch": 0.04, "learning_rate": 4.9759166046547344e-05, "loss": 0.8513, "step": 44314 }, { "epoch": 0.04, "learning_rate": 4.9759155169936554e-05, "loss": 0.628, "step": 44315 }, { "epoch": 0.04, "learning_rate": 4.975914429308136e-05, "loss": 0.6733, "step": 44316 }, { "epoch": 0.04, "learning_rate": 4.975913341598174e-05, "loss": 0.9488, "step": 44317 }, { "epoch": 0.04, "learning_rate": 4.9759122538637714e-05, "loss": 1.1872, "step": 44318 }, { "epoch": 0.04, "learning_rate": 4.9759111661049285e-05, "loss": 0.8717, "step": 44319 }, { "epoch": 0.04, "learning_rate": 4.975910078321643e-05, "loss": 0.9229, "step": 44320 }, { "epoch": 0.04, "learning_rate": 4.975908990513917e-05, "loss": 1.0926, "step": 44321 }, { "epoch": 0.04, "learning_rate": 4.97590790268175e-05, "loss": 0.9759, "step": 44322 }, { "epoch": 0.04, "learning_rate": 4.975906814825142e-05, "loss": 0.8217, "step": 44323 }, { "epoch": 0.04, "learning_rate": 4.975905726944093e-05, "loss": 0.8467, "step": 44324 }, { "epoch": 0.04, "learning_rate": 4.975904639038602e-05, "loss": 0.8962, "step": 44325 }, { "epoch": 0.04, "learning_rate": 4.97590355110867e-05, "loss": 1.1065, "step": 44326 }, { "epoch": 0.04, "learning_rate": 4.975902463154298e-05, "loss": 0.598, "step": 44327 }, { "epoch": 0.04, "learning_rate": 4.9759013751754844e-05, "loss": 1.3974, "step": 44328 }, { "epoch": 0.04, "learning_rate": 4.97590028717223e-05, "loss": 0.9934, "step": 44329 }, { "epoch": 0.04, "learning_rate": 4.975899199144534e-05, "loss": 1.0534, "step": 44330 }, { "epoch": 0.04, "learning_rate": 4.9758981110923976e-05, "loss": 0.9054, "step": 44331 }, { "epoch": 0.04, "learning_rate": 4.97589702301582e-05, "loss": 1.3586, "step": 44332 }, { "epoch": 0.04, "learning_rate": 4.975895934914801e-05, "loss": 1.1052, "step": 44333 }, { "epoch": 0.04, "learning_rate": 4.9758948467893404e-05, "loss": 0.7523, "step": 44334 }, { "epoch": 0.04, "learning_rate": 4.975893758639441e-05, "loss": 0.755, "step": 44335 }, { "epoch": 0.04, "learning_rate": 4.9758926704650985e-05, "loss": 0.5167, "step": 44336 }, { "epoch": 0.04, "learning_rate": 4.9758915822663155e-05, "loss": 1.1771, "step": 44337 }, { "epoch": 0.04, "learning_rate": 4.975890494043092e-05, "loss": 2.3497, "step": 44338 }, { "epoch": 0.04, "learning_rate": 4.975889405795428e-05, "loss": 1.7266, "step": 44339 }, { "epoch": 0.04, "learning_rate": 4.9758883175233216e-05, "loss": 1.3096, "step": 44340 }, { "epoch": 0.04, "learning_rate": 4.975887229226775e-05, "loss": 0.9277, "step": 44341 }, { "epoch": 0.04, "learning_rate": 4.9758861409057874e-05, "loss": 0.9703, "step": 44342 }, { "epoch": 0.04, "learning_rate": 4.9758850525603594e-05, "loss": 1.1053, "step": 44343 }, { "epoch": 0.04, "learning_rate": 4.97588396419049e-05, "loss": 1.0579, "step": 44344 }, { "epoch": 0.04, "learning_rate": 4.97588287579618e-05, "loss": 1.1893, "step": 44345 }, { "epoch": 0.04, "learning_rate": 4.975881787377429e-05, "loss": 1.4656, "step": 44346 }, { "epoch": 0.04, "learning_rate": 4.975880698934237e-05, "loss": 0.715, "step": 44347 }, { "epoch": 0.04, "learning_rate": 4.975879610466604e-05, "loss": 0.8638, "step": 44348 }, { "epoch": 0.04, "learning_rate": 4.9758785219745305e-05, "loss": 1.0714, "step": 44349 }, { "epoch": 0.04, "learning_rate": 4.9758774334580167e-05, "loss": 0.5026, "step": 44350 }, { "epoch": 0.04, "learning_rate": 4.975876344917061e-05, "loss": 0.2241, "step": 44351 }, { "epoch": 0.04, "learning_rate": 4.975875256351665e-05, "loss": 0.1609, "step": 44352 }, { "epoch": 0.04, "learning_rate": 4.975874167761828e-05, "loss": 0.1384, "step": 44353 }, { "epoch": 0.04, "learning_rate": 4.9758730791475505e-05, "loss": 0.1097, "step": 44354 }, { "epoch": 0.04, "learning_rate": 4.975871990508832e-05, "loss": 0.1432, "step": 44355 }, { "epoch": 0.04, "learning_rate": 4.975870901845673e-05, "loss": 0.1288, "step": 44356 }, { "epoch": 0.04, "learning_rate": 4.975869813158074e-05, "loss": 0.1296, "step": 44357 }, { "epoch": 0.04, "learning_rate": 4.975868724446033e-05, "loss": 1.1406, "step": 44358 }, { "epoch": 0.04, "learning_rate": 4.9758676357095514e-05, "loss": 1.1604, "step": 44359 }, { "epoch": 0.04, "learning_rate": 4.97586654694863e-05, "loss": 1.2105, "step": 44360 }, { "epoch": 0.04, "learning_rate": 4.9758654581632665e-05, "loss": 1.0885, "step": 44361 }, { "epoch": 0.04, "learning_rate": 4.975864369353463e-05, "loss": 1.0051, "step": 44362 }, { "epoch": 0.04, "learning_rate": 4.975863280519219e-05, "loss": 1.2788, "step": 44363 }, { "epoch": 0.04, "learning_rate": 4.975862191660534e-05, "loss": 0.9128, "step": 44364 }, { "epoch": 0.04, "learning_rate": 4.975861102777408e-05, "loss": 0.9093, "step": 44365 }, { "epoch": 0.04, "learning_rate": 4.975860013869843e-05, "loss": 1.0936, "step": 44366 }, { "epoch": 0.04, "learning_rate": 4.9758589249378356e-05, "loss": 0.5281, "step": 44367 }, { "epoch": 0.04, "learning_rate": 4.975857835981388e-05, "loss": 0.2065, "step": 44368 }, { "epoch": 0.04, "learning_rate": 4.9758567470005e-05, "loss": 1.1356, "step": 44369 }, { "epoch": 0.04, "learning_rate": 4.975855657995171e-05, "loss": 1.2061, "step": 44370 }, { "epoch": 0.04, "learning_rate": 4.975854568965403e-05, "loss": 0.9431, "step": 44371 }, { "epoch": 0.04, "learning_rate": 4.9758534799111924e-05, "loss": 0.8633, "step": 44372 }, { "epoch": 0.04, "learning_rate": 4.9758523908325414e-05, "loss": 0.714, "step": 44373 }, { "epoch": 0.04, "learning_rate": 4.9758513017294506e-05, "loss": 1.2124, "step": 44374 }, { "epoch": 0.04, "learning_rate": 4.975850212601919e-05, "loss": 1.1215, "step": 44375 }, { "epoch": 0.04, "learning_rate": 4.975849123449947e-05, "loss": 1.4983, "step": 44376 }, { "epoch": 0.04, "learning_rate": 4.975848034273534e-05, "loss": 0.982, "step": 44377 }, { "epoch": 0.04, "learning_rate": 4.975846945072681e-05, "loss": 1.0833, "step": 44378 }, { "epoch": 0.04, "learning_rate": 4.9758458558473874e-05, "loss": 0.9185, "step": 44379 }, { "epoch": 0.04, "learning_rate": 4.975844766597653e-05, "loss": 1.1255, "step": 44380 }, { "epoch": 0.04, "learning_rate": 4.975843677323478e-05, "loss": 0.7952, "step": 44381 }, { "epoch": 0.04, "learning_rate": 4.975842588024863e-05, "loss": 1.2562, "step": 44382 }, { "epoch": 0.04, "learning_rate": 4.975841498701807e-05, "loss": 1.1091, "step": 44383 }, { "epoch": 0.04, "learning_rate": 4.975840409354311e-05, "loss": 1.1255, "step": 44384 }, { "epoch": 0.04, "learning_rate": 4.975839319982374e-05, "loss": 1.147, "step": 44385 }, { "epoch": 0.04, "learning_rate": 4.975838230585998e-05, "loss": 0.9163, "step": 44386 }, { "epoch": 0.04, "learning_rate": 4.9758371411651794e-05, "loss": 0.7632, "step": 44387 }, { "epoch": 0.04, "learning_rate": 4.9758360517199224e-05, "loss": 0.8683, "step": 44388 }, { "epoch": 0.04, "learning_rate": 4.9758349622502235e-05, "loss": 1.0161, "step": 44389 }, { "epoch": 0.04, "learning_rate": 4.975833872756085e-05, "loss": 1.1837, "step": 44390 }, { "epoch": 0.04, "learning_rate": 4.975832783237506e-05, "loss": 0.921, "step": 44391 }, { "epoch": 0.04, "learning_rate": 4.975831693694486e-05, "loss": 1.0609, "step": 44392 }, { "epoch": 0.04, "learning_rate": 4.975830604127026e-05, "loss": 1.15, "step": 44393 }, { "epoch": 0.04, "learning_rate": 4.975829514535126e-05, "loss": 0.9297, "step": 44394 }, { "epoch": 0.04, "learning_rate": 4.9758284249187854e-05, "loss": 0.8869, "step": 44395 }, { "epoch": 0.04, "learning_rate": 4.975827335278004e-05, "loss": 1.0168, "step": 44396 }, { "epoch": 0.04, "learning_rate": 4.9758262456127835e-05, "loss": 1.0765, "step": 44397 }, { "epoch": 0.04, "learning_rate": 4.9758251559231217e-05, "loss": 1.0971, "step": 44398 }, { "epoch": 0.04, "learning_rate": 4.97582406620902e-05, "loss": 0.6828, "step": 44399 }, { "epoch": 0.04, "learning_rate": 4.975822976470478e-05, "loss": 0.9077, "step": 44400 }, { "epoch": 0.04, "learning_rate": 4.975821886707495e-05, "loss": 1.0426, "step": 44401 }, { "epoch": 0.04, "learning_rate": 4.9758207969200726e-05, "loss": 0.8804, "step": 44402 }, { "epoch": 0.04, "learning_rate": 4.9758197071082096e-05, "loss": 1.1705, "step": 44403 }, { "epoch": 0.04, "learning_rate": 4.975818617271907e-05, "loss": 1.0337, "step": 44404 }, { "epoch": 0.04, "learning_rate": 4.975817527411163e-05, "loss": 1.2102, "step": 44405 }, { "epoch": 0.04, "learning_rate": 4.9758164375259796e-05, "loss": 1.1132, "step": 44406 }, { "epoch": 0.04, "learning_rate": 4.9758153476163555e-05, "loss": 1.0734, "step": 44407 }, { "epoch": 0.04, "learning_rate": 4.9758142576822914e-05, "loss": 1.204, "step": 44408 }, { "epoch": 0.04, "learning_rate": 4.975813167723787e-05, "loss": 1.074, "step": 44409 }, { "epoch": 0.04, "learning_rate": 4.9758120777408427e-05, "loss": 1.144, "step": 44410 }, { "epoch": 0.04, "learning_rate": 4.975810987733458e-05, "loss": 1.2265, "step": 44411 }, { "epoch": 0.04, "learning_rate": 4.9758098977016335e-05, "loss": 1.1079, "step": 44412 }, { "epoch": 0.04, "learning_rate": 4.975808807645368e-05, "loss": 1.134, "step": 44413 }, { "epoch": 0.04, "learning_rate": 4.975807717564663e-05, "loss": 0.9788, "step": 44414 }, { "epoch": 0.04, "learning_rate": 4.975806627459518e-05, "loss": 0.8611, "step": 44415 }, { "epoch": 0.04, "learning_rate": 4.975805537329933e-05, "loss": 1.0164, "step": 44416 }, { "epoch": 0.04, "learning_rate": 4.9758044471759066e-05, "loss": 0.8108, "step": 44417 }, { "epoch": 0.04, "learning_rate": 4.975803356997442e-05, "loss": 1.3235, "step": 44418 }, { "epoch": 0.04, "learning_rate": 4.975802266794536e-05, "loss": 1.3523, "step": 44419 }, { "epoch": 0.04, "learning_rate": 4.97580117656719e-05, "loss": 0.9915, "step": 44420 }, { "epoch": 0.04, "learning_rate": 4.9758000863154045e-05, "loss": 0.7567, "step": 44421 }, { "epoch": 0.04, "learning_rate": 4.9757989960391784e-05, "loss": 0.879, "step": 44422 }, { "epoch": 0.04, "learning_rate": 4.975797905738513e-05, "loss": 0.8727, "step": 44423 }, { "epoch": 0.04, "learning_rate": 4.975796815413406e-05, "loss": 1.0431, "step": 44424 }, { "epoch": 0.04, "learning_rate": 4.975795725063861e-05, "loss": 1.0133, "step": 44425 }, { "epoch": 0.04, "learning_rate": 4.975794634689875e-05, "loss": 0.85, "step": 44426 }, { "epoch": 0.04, "learning_rate": 4.9757935442914486e-05, "loss": 1.5363, "step": 44427 }, { "epoch": 0.04, "learning_rate": 4.975792453868583e-05, "loss": 1.0249, "step": 44428 }, { "epoch": 0.04, "learning_rate": 4.975791363421277e-05, "loss": 0.996, "step": 44429 }, { "epoch": 0.04, "learning_rate": 4.975790272949531e-05, "loss": 0.6647, "step": 44430 }, { "epoch": 0.04, "learning_rate": 4.975789182453345e-05, "loss": 1.178, "step": 44431 }, { "epoch": 0.04, "learning_rate": 4.9757880919327194e-05, "loss": 0.7118, "step": 44432 }, { "epoch": 0.04, "learning_rate": 4.975787001387654e-05, "loss": 1.0416, "step": 44433 }, { "epoch": 0.04, "learning_rate": 4.9757859108181484e-05, "loss": 0.9892, "step": 44434 }, { "epoch": 0.04, "learning_rate": 4.975784820224202e-05, "loss": 0.7129, "step": 44435 }, { "epoch": 0.04, "learning_rate": 4.975783729605817e-05, "loss": 0.9722, "step": 44436 }, { "epoch": 0.04, "learning_rate": 4.9757826389629915e-05, "loss": 0.8658, "step": 44437 }, { "epoch": 0.04, "learning_rate": 4.975781548295726e-05, "loss": 1.1323, "step": 44438 }, { "epoch": 0.04, "learning_rate": 4.975780457604021e-05, "loss": 1.1424, "step": 44439 }, { "epoch": 0.04, "learning_rate": 4.975779366887876e-05, "loss": 0.9862, "step": 44440 }, { "epoch": 0.04, "learning_rate": 4.975778276147291e-05, "loss": 1.1434, "step": 44441 }, { "epoch": 0.04, "learning_rate": 4.975777185382266e-05, "loss": 0.8663, "step": 44442 }, { "epoch": 0.04, "learning_rate": 4.9757760945928025e-05, "loss": 0.6844, "step": 44443 }, { "epoch": 0.04, "learning_rate": 4.9757750037788975e-05, "loss": 0.6476, "step": 44444 }, { "epoch": 0.04, "learning_rate": 4.975773912940553e-05, "loss": 0.6389, "step": 44445 }, { "epoch": 0.04, "learning_rate": 4.97577282207777e-05, "loss": 0.5223, "step": 44446 }, { "epoch": 0.04, "learning_rate": 4.975771731190546e-05, "loss": 1.2739, "step": 44447 }, { "epoch": 0.04, "learning_rate": 4.975770640278883e-05, "loss": 1.4555, "step": 44448 }, { "epoch": 0.04, "learning_rate": 4.9757695493427794e-05, "loss": 1.3284, "step": 44449 }, { "epoch": 0.04, "learning_rate": 4.975768458382236e-05, "loss": 1.0855, "step": 44450 }, { "epoch": 0.04, "learning_rate": 4.9757673673972535e-05, "loss": 0.7777, "step": 44451 }, { "epoch": 0.04, "learning_rate": 4.975766276387831e-05, "loss": 0.2085, "step": 44452 }, { "epoch": 0.04, "learning_rate": 4.975765185353969e-05, "loss": 1.1639, "step": 44453 }, { "epoch": 0.04, "learning_rate": 4.9757640942956666e-05, "loss": 0.9765, "step": 44454 }, { "epoch": 0.04, "learning_rate": 4.9757630032129255e-05, "loss": 0.9448, "step": 44455 }, { "epoch": 0.04, "learning_rate": 4.975761912105744e-05, "loss": 1.0741, "step": 44456 }, { "epoch": 0.04, "learning_rate": 4.975760820974122e-05, "loss": 1.5458, "step": 44457 }, { "epoch": 0.04, "learning_rate": 4.975759729818063e-05, "loss": 1.7422, "step": 44458 }, { "epoch": 0.04, "learning_rate": 4.975758638637562e-05, "loss": 1.62, "step": 44459 }, { "epoch": 0.04, "learning_rate": 4.975757547432622e-05, "loss": 1.1053, "step": 44460 }, { "epoch": 0.04, "learning_rate": 4.975756456203242e-05, "loss": 1.2188, "step": 44461 }, { "epoch": 0.04, "learning_rate": 4.975755364949423e-05, "loss": 1.0944, "step": 44462 }, { "epoch": 0.04, "learning_rate": 4.975754273671164e-05, "loss": 0.7203, "step": 44463 }, { "epoch": 0.04, "learning_rate": 4.975753182368466e-05, "loss": 0.9387, "step": 44464 }, { "epoch": 0.04, "learning_rate": 4.975752091041328e-05, "loss": 1.3965, "step": 44465 }, { "epoch": 0.04, "learning_rate": 4.97575099968975e-05, "loss": 1.2373, "step": 44466 }, { "epoch": 0.04, "learning_rate": 4.9757499083137324e-05, "loss": 1.178, "step": 44467 }, { "epoch": 0.04, "learning_rate": 4.975748816913276e-05, "loss": 1.0698, "step": 44468 }, { "epoch": 0.04, "learning_rate": 4.97574772548838e-05, "loss": 1.1282, "step": 44469 }, { "epoch": 0.04, "learning_rate": 4.975746634039044e-05, "loss": 1.1798, "step": 44470 }, { "epoch": 0.04, "learning_rate": 4.975745542565269e-05, "loss": 1.4351, "step": 44471 }, { "epoch": 0.04, "learning_rate": 4.975744451067054e-05, "loss": 1.1219, "step": 44472 }, { "epoch": 0.04, "learning_rate": 4.975743359544399e-05, "loss": 0.54, "step": 44473 }, { "epoch": 0.04, "learning_rate": 4.975742267997305e-05, "loss": 0.5427, "step": 44474 }, { "epoch": 0.04, "learning_rate": 4.975741176425772e-05, "loss": 0.2614, "step": 44475 }, { "epoch": 0.04, "learning_rate": 4.9757400848298e-05, "loss": 0.1312, "step": 44476 }, { "epoch": 0.04, "learning_rate": 4.975738993209387e-05, "loss": 0.128, "step": 44477 }, { "epoch": 0.04, "learning_rate": 4.975737901564536e-05, "loss": 0.0685, "step": 44478 }, { "epoch": 0.04, "learning_rate": 4.975736809895244e-05, "loss": 0.1192, "step": 44479 }, { "epoch": 0.04, "learning_rate": 4.975735718201514e-05, "loss": 0.7829, "step": 44480 }, { "epoch": 0.04, "learning_rate": 4.9757346264833434e-05, "loss": 0.9829, "step": 44481 }, { "epoch": 0.04, "learning_rate": 4.975733534740734e-05, "loss": 1.0255, "step": 44482 }, { "epoch": 0.04, "learning_rate": 4.975732442973685e-05, "loss": 1.2642, "step": 44483 }, { "epoch": 0.04, "learning_rate": 4.975731351182197e-05, "loss": 1.0232, "step": 44484 }, { "epoch": 0.04, "learning_rate": 4.975730259366269e-05, "loss": 1.2657, "step": 44485 }, { "epoch": 0.04, "learning_rate": 4.9757291675259024e-05, "loss": 0.9014, "step": 44486 }, { "epoch": 0.04, "learning_rate": 4.975728075661096e-05, "loss": 0.611, "step": 44487 }, { "epoch": 0.04, "learning_rate": 4.97572698377185e-05, "loss": 0.4641, "step": 44488 }, { "epoch": 0.04, "learning_rate": 4.975725891858165e-05, "loss": 0.642, "step": 44489 }, { "epoch": 0.04, "learning_rate": 4.9757247999200404e-05, "loss": 1.117, "step": 44490 }, { "epoch": 0.04, "learning_rate": 4.975723707957477e-05, "loss": 1.1836, "step": 44491 }, { "epoch": 0.04, "learning_rate": 4.9757226159704745e-05, "loss": 0.8819, "step": 44492 }, { "epoch": 0.04, "learning_rate": 4.975721523959032e-05, "loss": 1.0615, "step": 44493 }, { "epoch": 0.04, "learning_rate": 4.975720431923151e-05, "loss": 1.2509, "step": 44494 }, { "epoch": 0.04, "learning_rate": 4.97571933986283e-05, "loss": 1.1723, "step": 44495 }, { "epoch": 0.04, "learning_rate": 4.9757182477780693e-05, "loss": 0.7287, "step": 44496 }, { "epoch": 0.04, "learning_rate": 4.975717155668871e-05, "loss": 1.3351, "step": 44497 }, { "epoch": 0.04, "learning_rate": 4.9757160635352315e-05, "loss": 1.3256, "step": 44498 }, { "epoch": 0.04, "learning_rate": 4.975714971377154e-05, "loss": 1.028, "step": 44499 }, { "epoch": 0.04, "learning_rate": 4.975713879194637e-05, "loss": 1.1134, "step": 44500 }, { "epoch": 0.04, "eval_loss": 1.0934947729110718, "eval_runtime": 129.2656, "eval_samples_per_second": 10.714, "eval_steps_per_second": 5.361, "step": 44500 }, { "epoch": 0.04, "learning_rate": 4.975712786987681e-05, "loss": 1.1821, "step": 44501 }, { "epoch": 0.04, "learning_rate": 4.975711694756285e-05, "loss": 1.004, "step": 44502 }, { "epoch": 0.04, "learning_rate": 4.9757106025004516e-05, "loss": 0.9842, "step": 44503 }, { "epoch": 0.04, "learning_rate": 4.975709510220178e-05, "loss": 0.6816, "step": 44504 }, { "epoch": 0.04, "learning_rate": 4.9757084179154646e-05, "loss": 0.6615, "step": 44505 }, { "epoch": 0.04, "learning_rate": 4.975707325586313e-05, "loss": 1.1789, "step": 44506 }, { "epoch": 0.04, "learning_rate": 4.975706233232721e-05, "loss": 1.1939, "step": 44507 }, { "epoch": 0.04, "learning_rate": 4.975705140854691e-05, "loss": 1.0176, "step": 44508 }, { "epoch": 0.04, "learning_rate": 4.975704048452221e-05, "loss": 1.1917, "step": 44509 }, { "epoch": 0.04, "learning_rate": 4.9757029560253135e-05, "loss": 1.3303, "step": 44510 }, { "epoch": 0.04, "learning_rate": 4.975701863573966e-05, "loss": 1.2063, "step": 44511 }, { "epoch": 0.04, "learning_rate": 4.975700771098179e-05, "loss": 1.0244, "step": 44512 }, { "epoch": 0.04, "learning_rate": 4.975699678597953e-05, "loss": 0.8673, "step": 44513 }, { "epoch": 0.04, "learning_rate": 4.975698586073288e-05, "loss": 0.9546, "step": 44514 }, { "epoch": 0.04, "learning_rate": 4.975697493524184e-05, "loss": 1.0203, "step": 44515 }, { "epoch": 0.04, "learning_rate": 4.975696400950641e-05, "loss": 0.9554, "step": 44516 }, { "epoch": 0.04, "learning_rate": 4.975695308352659e-05, "loss": 0.8055, "step": 44517 }, { "epoch": 0.04, "learning_rate": 4.975694215730238e-05, "loss": 1.214, "step": 44518 }, { "epoch": 0.04, "learning_rate": 4.975693123083378e-05, "loss": 1.1121, "step": 44519 }, { "epoch": 0.04, "learning_rate": 4.975692030412079e-05, "loss": 0.8523, "step": 44520 }, { "epoch": 0.04, "learning_rate": 4.975690937716341e-05, "loss": 1.0666, "step": 44521 }, { "epoch": 0.04, "learning_rate": 4.975689844996164e-05, "loss": 1.1977, "step": 44522 }, { "epoch": 0.04, "learning_rate": 4.9756887522515485e-05, "loss": 1.0113, "step": 44523 }, { "epoch": 0.04, "learning_rate": 4.9756876594824936e-05, "loss": 1.2196, "step": 44524 }, { "epoch": 0.04, "learning_rate": 4.9756865666889994e-05, "loss": 1.3104, "step": 44525 }, { "epoch": 0.04, "learning_rate": 4.9756854738710666e-05, "loss": 1.018, "step": 44526 }, { "epoch": 0.04, "learning_rate": 4.975684381028695e-05, "loss": 0.8012, "step": 44527 }, { "epoch": 0.04, "learning_rate": 4.975683288161884e-05, "loss": 1.4019, "step": 44528 }, { "epoch": 0.04, "learning_rate": 4.9756821952706354e-05, "loss": 0.8504, "step": 44529 }, { "epoch": 0.04, "learning_rate": 4.975681102354947e-05, "loss": 0.7718, "step": 44530 }, { "epoch": 0.04, "learning_rate": 4.975680009414819e-05, "loss": 0.8842, "step": 44531 }, { "epoch": 0.04, "learning_rate": 4.9756789164502534e-05, "loss": 0.9013, "step": 44532 }, { "epoch": 0.04, "learning_rate": 4.9756778234612484e-05, "loss": 1.1925, "step": 44533 }, { "epoch": 0.04, "learning_rate": 4.975676730447805e-05, "loss": 1.0736, "step": 44534 }, { "epoch": 0.04, "learning_rate": 4.975675637409922e-05, "loss": 1.1944, "step": 44535 }, { "epoch": 0.04, "learning_rate": 4.9756745443476006e-05, "loss": 1.8551, "step": 44536 }, { "epoch": 0.04, "learning_rate": 4.97567345126084e-05, "loss": 1.8024, "step": 44537 }, { "epoch": 0.04, "learning_rate": 4.9756723581496406e-05, "loss": 1.2288, "step": 44538 }, { "epoch": 0.04, "learning_rate": 4.975671265014003e-05, "loss": 0.7171, "step": 44539 }, { "epoch": 0.04, "learning_rate": 4.975670171853926e-05, "loss": 0.9267, "step": 44540 }, { "epoch": 0.04, "learning_rate": 4.9756690786694105e-05, "loss": 0.9044, "step": 44541 }, { "epoch": 0.04, "learning_rate": 4.975667985460456e-05, "loss": 1.3449, "step": 44542 }, { "epoch": 0.04, "learning_rate": 4.975666892227063e-05, "loss": 1.1153, "step": 44543 }, { "epoch": 0.04, "learning_rate": 4.975665798969231e-05, "loss": 1.0271, "step": 44544 }, { "epoch": 0.04, "learning_rate": 4.975664705686961e-05, "loss": 0.9081, "step": 44545 }, { "epoch": 0.04, "learning_rate": 4.975663612380252e-05, "loss": 1.0335, "step": 44546 }, { "epoch": 0.04, "learning_rate": 4.9756625190491034e-05, "loss": 1.0113, "step": 44547 }, { "epoch": 0.04, "learning_rate": 4.9756614256935175e-05, "loss": 1.0543, "step": 44548 }, { "epoch": 0.04, "learning_rate": 4.975660332313492e-05, "loss": 1.1105, "step": 44549 }, { "epoch": 0.04, "learning_rate": 4.9756592389090285e-05, "loss": 1.0074, "step": 44550 }, { "epoch": 0.04, "learning_rate": 4.9756581454801255e-05, "loss": 0.868, "step": 44551 }, { "epoch": 0.04, "learning_rate": 4.9756570520267845e-05, "loss": 1.045, "step": 44552 }, { "epoch": 0.04, "learning_rate": 4.975655958549004e-05, "loss": 0.7093, "step": 44553 }, { "epoch": 0.04, "learning_rate": 4.975654865046786e-05, "loss": 0.8159, "step": 44554 }, { "epoch": 0.04, "learning_rate": 4.9756537715201286e-05, "loss": 0.9377, "step": 44555 }, { "epoch": 0.04, "learning_rate": 4.9756526779690326e-05, "loss": 1.1757, "step": 44556 }, { "epoch": 0.04, "learning_rate": 4.975651584393498e-05, "loss": 0.8202, "step": 44557 }, { "epoch": 0.04, "learning_rate": 4.9756504907935254e-05, "loss": 0.9256, "step": 44558 }, { "epoch": 0.04, "learning_rate": 4.9756493971691136e-05, "loss": 0.9236, "step": 44559 }, { "epoch": 0.04, "learning_rate": 4.975648303520264e-05, "loss": 0.8721, "step": 44560 }, { "epoch": 0.04, "learning_rate": 4.9756472098469756e-05, "loss": 1.0116, "step": 44561 }, { "epoch": 0.04, "learning_rate": 4.975646116149247e-05, "loss": 1.0285, "step": 44562 }, { "epoch": 0.04, "learning_rate": 4.9756450224270825e-05, "loss": 1.2667, "step": 44563 }, { "epoch": 0.04, "learning_rate": 4.975643928680478e-05, "loss": 1.0311, "step": 44564 }, { "epoch": 0.04, "learning_rate": 4.975642834909435e-05, "loss": 0.9458, "step": 44565 }, { "epoch": 0.04, "learning_rate": 4.975641741113953e-05, "loss": 1.1597, "step": 44566 }, { "epoch": 0.04, "learning_rate": 4.975640647294034e-05, "loss": 0.9621, "step": 44567 }, { "epoch": 0.04, "learning_rate": 4.975639553449676e-05, "loss": 0.8551, "step": 44568 }, { "epoch": 0.04, "learning_rate": 4.975638459580879e-05, "loss": 0.5337, "step": 44569 }, { "epoch": 0.04, "learning_rate": 4.975637365687643e-05, "loss": 0.3477, "step": 44570 }, { "epoch": 0.04, "learning_rate": 4.97563627176997e-05, "loss": 0.3831, "step": 44571 }, { "epoch": 0.04, "learning_rate": 4.9756351778278574e-05, "loss": 0.715, "step": 44572 }, { "epoch": 0.04, "learning_rate": 4.9756340838613066e-05, "loss": 1.4877, "step": 44573 }, { "epoch": 0.04, "learning_rate": 4.975632989870318e-05, "loss": 0.8682, "step": 44574 }, { "epoch": 0.04, "learning_rate": 4.9756318958548906e-05, "loss": 1.027, "step": 44575 }, { "epoch": 0.04, "learning_rate": 4.9756308018150254e-05, "loss": 0.4843, "step": 44576 }, { "epoch": 0.04, "learning_rate": 4.975629707750721e-05, "loss": 0.3178, "step": 44577 }, { "epoch": 0.04, "learning_rate": 4.9756286136619786e-05, "loss": 1.4994, "step": 44578 }, { "epoch": 0.04, "learning_rate": 4.9756275195487976e-05, "loss": 1.1377, "step": 44579 }, { "epoch": 0.04, "learning_rate": 4.9756264254111786e-05, "loss": 0.9389, "step": 44580 }, { "epoch": 0.04, "learning_rate": 4.9756253312491205e-05, "loss": 0.9669, "step": 44581 }, { "epoch": 0.04, "learning_rate": 4.975624237062625e-05, "loss": 1.0074, "step": 44582 }, { "epoch": 0.04, "learning_rate": 4.975623142851691e-05, "loss": 0.8625, "step": 44583 }, { "epoch": 0.04, "learning_rate": 4.9756220486163184e-05, "loss": 0.9084, "step": 44584 }, { "epoch": 0.04, "learning_rate": 4.975620954356507e-05, "loss": 0.7655, "step": 44585 }, { "epoch": 0.04, "learning_rate": 4.975619860072258e-05, "loss": 1.0475, "step": 44586 }, { "epoch": 0.04, "learning_rate": 4.975618765763571e-05, "loss": 0.9622, "step": 44587 }, { "epoch": 0.04, "learning_rate": 4.9756176714304454e-05, "loss": 1.321, "step": 44588 }, { "epoch": 0.04, "learning_rate": 4.975616577072881e-05, "loss": 1.2546, "step": 44589 }, { "epoch": 0.04, "learning_rate": 4.97561548269088e-05, "loss": 1.0696, "step": 44590 }, { "epoch": 0.04, "learning_rate": 4.975614388284438e-05, "loss": 1.2378, "step": 44591 }, { "epoch": 0.04, "learning_rate": 4.97561329385356e-05, "loss": 1.2604, "step": 44592 }, { "epoch": 0.04, "learning_rate": 4.975612199398244e-05, "loss": 0.9712, "step": 44593 }, { "epoch": 0.04, "learning_rate": 4.9756111049184885e-05, "loss": 1.201, "step": 44594 }, { "epoch": 0.04, "learning_rate": 4.9756100104142955e-05, "loss": 1.5143, "step": 44595 }, { "epoch": 0.04, "learning_rate": 4.975608915885664e-05, "loss": 0.7694, "step": 44596 }, { "epoch": 0.04, "learning_rate": 4.975607821332594e-05, "loss": 1.4325, "step": 44597 }, { "epoch": 0.04, "learning_rate": 4.9756067267550867e-05, "loss": 1.1435, "step": 44598 }, { "epoch": 0.04, "learning_rate": 4.9756056321531405e-05, "loss": 1.0175, "step": 44599 }, { "epoch": 0.04, "learning_rate": 4.9756045375267565e-05, "loss": 0.7461, "step": 44600 }, { "epoch": 0.04, "learning_rate": 4.9756034428759346e-05, "loss": 1.0247, "step": 44601 }, { "epoch": 0.04, "learning_rate": 4.975602348200675e-05, "loss": 1.0005, "step": 44602 }, { "epoch": 0.04, "learning_rate": 4.975601253500977e-05, "loss": 0.8307, "step": 44603 }, { "epoch": 0.04, "learning_rate": 4.97560015877684e-05, "loss": 0.7061, "step": 44604 }, { "epoch": 0.04, "learning_rate": 4.9755990640282656e-05, "loss": 0.5068, "step": 44605 }, { "epoch": 0.04, "learning_rate": 4.975597969255253e-05, "loss": 1.2718, "step": 44606 }, { "epoch": 0.04, "learning_rate": 4.975596874457802e-05, "loss": 1.0863, "step": 44607 }, { "epoch": 0.04, "learning_rate": 4.975595779635913e-05, "loss": 0.7542, "step": 44608 }, { "epoch": 0.04, "learning_rate": 4.9755946847895865e-05, "loss": 1.8435, "step": 44609 }, { "epoch": 0.04, "learning_rate": 4.975593589918822e-05, "loss": 1.2309, "step": 44610 }, { "epoch": 0.04, "learning_rate": 4.9755924950236195e-05, "loss": 1.2967, "step": 44611 }, { "epoch": 0.04, "learning_rate": 4.9755914001039785e-05, "loss": 1.0971, "step": 44612 }, { "epoch": 0.04, "learning_rate": 4.9755903051598995e-05, "loss": 0.7682, "step": 44613 }, { "epoch": 0.04, "learning_rate": 4.975589210191383e-05, "loss": 0.8004, "step": 44614 }, { "epoch": 0.04, "learning_rate": 4.9755881151984284e-05, "loss": 0.9758, "step": 44615 }, { "epoch": 0.04, "learning_rate": 4.975587020181036e-05, "loss": 0.699, "step": 44616 }, { "epoch": 0.04, "learning_rate": 4.975585925139205e-05, "loss": 0.9747, "step": 44617 }, { "epoch": 0.04, "learning_rate": 4.9755848300729365e-05, "loss": 0.9779, "step": 44618 }, { "epoch": 0.04, "learning_rate": 4.97558373498223e-05, "loss": 0.8999, "step": 44619 }, { "epoch": 0.04, "learning_rate": 4.9755826398670856e-05, "loss": 0.8572, "step": 44620 }, { "epoch": 0.04, "learning_rate": 4.975581544727503e-05, "loss": 0.9102, "step": 44621 }, { "epoch": 0.04, "learning_rate": 4.975580449563483e-05, "loss": 0.8558, "step": 44622 }, { "epoch": 0.04, "learning_rate": 4.975579354375025e-05, "loss": 1.1688, "step": 44623 }, { "epoch": 0.04, "learning_rate": 4.975578259162129e-05, "loss": 1.1068, "step": 44624 }, { "epoch": 0.04, "learning_rate": 4.975577163924795e-05, "loss": 1.1215, "step": 44625 }, { "epoch": 0.04, "learning_rate": 4.975576068663023e-05, "loss": 1.1038, "step": 44626 }, { "epoch": 0.04, "learning_rate": 4.975574973376814e-05, "loss": 1.0779, "step": 44627 }, { "epoch": 0.04, "learning_rate": 4.975573878066167e-05, "loss": 1.0511, "step": 44628 }, { "epoch": 0.04, "learning_rate": 4.975572782731081e-05, "loss": 0.8964, "step": 44629 }, { "epoch": 0.04, "learning_rate": 4.975571687371558e-05, "loss": 1.0021, "step": 44630 }, { "epoch": 0.04, "learning_rate": 4.9755705919875974e-05, "loss": 0.6134, "step": 44631 }, { "epoch": 0.04, "learning_rate": 4.9755694965791987e-05, "loss": 0.8286, "step": 44632 }, { "epoch": 0.04, "learning_rate": 4.975568401146362e-05, "loss": 0.5051, "step": 44633 }, { "epoch": 0.04, "learning_rate": 4.975567305689088e-05, "loss": 0.6416, "step": 44634 }, { "epoch": 0.04, "learning_rate": 4.975566210207376e-05, "loss": 0.509, "step": 44635 }, { "epoch": 0.04, "learning_rate": 4.975565114701226e-05, "loss": 0.9992, "step": 44636 }, { "epoch": 0.04, "learning_rate": 4.975564019170639e-05, "loss": 1.0517, "step": 44637 }, { "epoch": 0.04, "learning_rate": 4.975562923615614e-05, "loss": 1.1653, "step": 44638 }, { "epoch": 0.04, "learning_rate": 4.975561828036151e-05, "loss": 1.3733, "step": 44639 }, { "epoch": 0.04, "learning_rate": 4.9755607324322506e-05, "loss": 1.1836, "step": 44640 }, { "epoch": 0.04, "learning_rate": 4.975559636803913e-05, "loss": 0.7331, "step": 44641 }, { "epoch": 0.04, "learning_rate": 4.975558541151136e-05, "loss": 1.1572, "step": 44642 }, { "epoch": 0.04, "learning_rate": 4.975557445473923e-05, "loss": 1.0862, "step": 44643 }, { "epoch": 0.04, "learning_rate": 4.975556349772271e-05, "loss": 1.1315, "step": 44644 }, { "epoch": 0.04, "learning_rate": 4.9755552540461824e-05, "loss": 1.243, "step": 44645 }, { "epoch": 0.04, "learning_rate": 4.975554158295656e-05, "loss": 1.1664, "step": 44646 }, { "epoch": 0.04, "learning_rate": 4.975553062520692e-05, "loss": 1.2336, "step": 44647 }, { "epoch": 0.04, "learning_rate": 4.97555196672129e-05, "loss": 1.1135, "step": 44648 }, { "epoch": 0.04, "learning_rate": 4.975550870897451e-05, "loss": 1.0859, "step": 44649 }, { "epoch": 0.04, "learning_rate": 4.9755497750491744e-05, "loss": 0.7014, "step": 44650 }, { "epoch": 0.04, "learning_rate": 4.975548679176459e-05, "loss": 0.9634, "step": 44651 }, { "epoch": 0.04, "learning_rate": 4.9755475832793076e-05, "loss": 1.0242, "step": 44652 }, { "epoch": 0.04, "learning_rate": 4.9755464873577174e-05, "loss": 0.8344, "step": 44653 }, { "epoch": 0.04, "learning_rate": 4.97554539141169e-05, "loss": 0.8389, "step": 44654 }, { "epoch": 0.04, "learning_rate": 4.975544295441226e-05, "loss": 0.7255, "step": 44655 }, { "epoch": 0.04, "learning_rate": 4.975543199446323e-05, "loss": 0.5986, "step": 44656 }, { "epoch": 0.04, "learning_rate": 4.9755421034269834e-05, "loss": 1.1447, "step": 44657 }, { "epoch": 0.04, "learning_rate": 4.975541007383206e-05, "loss": 0.9064, "step": 44658 }, { "epoch": 0.04, "learning_rate": 4.975539911314991e-05, "loss": 0.9572, "step": 44659 }, { "epoch": 0.04, "learning_rate": 4.975538815222339e-05, "loss": 0.9671, "step": 44660 }, { "epoch": 0.04, "learning_rate": 4.975537719105249e-05, "loss": 0.6614, "step": 44661 }, { "epoch": 0.04, "learning_rate": 4.9755366229637213e-05, "loss": 1.3022, "step": 44662 }, { "epoch": 0.04, "learning_rate": 4.975535526797757e-05, "loss": 1.1474, "step": 44663 }, { "epoch": 0.04, "learning_rate": 4.975534430607355e-05, "loss": 1.0395, "step": 44664 }, { "epoch": 0.04, "learning_rate": 4.975533334392515e-05, "loss": 0.6726, "step": 44665 }, { "epoch": 0.04, "learning_rate": 4.975532238153238e-05, "loss": 0.8152, "step": 44666 }, { "epoch": 0.04, "learning_rate": 4.9755311418895236e-05, "loss": 0.8524, "step": 44667 }, { "epoch": 0.04, "learning_rate": 4.975530045601372e-05, "loss": 0.7359, "step": 44668 }, { "epoch": 0.04, "learning_rate": 4.9755289492887826e-05, "loss": 0.9296, "step": 44669 }, { "epoch": 0.04, "learning_rate": 4.975527852951756e-05, "loss": 0.9121, "step": 44670 }, { "epoch": 0.04, "learning_rate": 4.975526756590292e-05, "loss": 1.1387, "step": 44671 }, { "epoch": 0.04, "learning_rate": 4.975525660204391e-05, "loss": 1.3952, "step": 44672 }, { "epoch": 0.04, "learning_rate": 4.975524563794052e-05, "loss": 1.3147, "step": 44673 }, { "epoch": 0.04, "learning_rate": 4.975523467359276e-05, "loss": 1.3324, "step": 44674 }, { "epoch": 0.04, "learning_rate": 4.9755223709000625e-05, "loss": 1.2069, "step": 44675 }, { "epoch": 0.04, "learning_rate": 4.975521274416412e-05, "loss": 1.1597, "step": 44676 }, { "epoch": 0.04, "learning_rate": 4.9755201779083236e-05, "loss": 1.0615, "step": 44677 }, { "epoch": 0.04, "learning_rate": 4.975519081375799e-05, "loss": 0.8524, "step": 44678 }, { "epoch": 0.04, "learning_rate": 4.975517984818836e-05, "loss": 1.0143, "step": 44679 }, { "epoch": 0.04, "learning_rate": 4.975516888237437e-05, "loss": 1.131, "step": 44680 }, { "epoch": 0.04, "learning_rate": 4.9755157916316e-05, "loss": 0.7606, "step": 44681 }, { "epoch": 0.04, "learning_rate": 4.9755146950013255e-05, "loss": 0.3431, "step": 44682 }, { "epoch": 0.04, "learning_rate": 4.975513598346614e-05, "loss": 0.2043, "step": 44683 }, { "epoch": 0.04, "learning_rate": 4.9755125016674646e-05, "loss": 0.2056, "step": 44684 }, { "epoch": 0.04, "learning_rate": 4.9755114049638786e-05, "loss": 0.8166, "step": 44685 }, { "epoch": 0.04, "learning_rate": 4.9755103082358554e-05, "loss": 1.06, "step": 44686 }, { "epoch": 0.04, "learning_rate": 4.975509211483395e-05, "loss": 0.8609, "step": 44687 }, { "epoch": 0.04, "learning_rate": 4.975508114706497e-05, "loss": 1.3636, "step": 44688 }, { "epoch": 0.04, "learning_rate": 4.975507017905163e-05, "loss": 1.233, "step": 44689 }, { "epoch": 0.04, "learning_rate": 4.975505921079391e-05, "loss": 0.9506, "step": 44690 }, { "epoch": 0.04, "learning_rate": 4.975504824229182e-05, "loss": 1.2235, "step": 44691 }, { "epoch": 0.04, "learning_rate": 4.9755037273545355e-05, "loss": 1.031, "step": 44692 }, { "epoch": 0.04, "learning_rate": 4.975502630455452e-05, "loss": 0.9867, "step": 44693 }, { "epoch": 0.04, "learning_rate": 4.975501533531932e-05, "loss": 1.0148, "step": 44694 }, { "epoch": 0.04, "learning_rate": 4.975500436583974e-05, "loss": 1.0594, "step": 44695 }, { "epoch": 0.04, "learning_rate": 4.97549933961158e-05, "loss": 0.9252, "step": 44696 }, { "epoch": 0.04, "learning_rate": 4.975498242614748e-05, "loss": 1.0979, "step": 44697 }, { "epoch": 0.04, "learning_rate": 4.97549714559348e-05, "loss": 0.9666, "step": 44698 }, { "epoch": 0.04, "learning_rate": 4.975496048547773e-05, "loss": 0.9766, "step": 44699 }, { "epoch": 0.04, "learning_rate": 4.9754949514776306e-05, "loss": 1.0011, "step": 44700 }, { "epoch": 0.04, "learning_rate": 4.97549385438305e-05, "loss": 0.9986, "step": 44701 }, { "epoch": 0.04, "learning_rate": 4.975492757264033e-05, "loss": 1.0975, "step": 44702 }, { "epoch": 0.04, "learning_rate": 4.975491660120579e-05, "loss": 0.9012, "step": 44703 }, { "epoch": 0.04, "learning_rate": 4.975490562952688e-05, "loss": 1.2465, "step": 44704 }, { "epoch": 0.04, "learning_rate": 4.97548946576036e-05, "loss": 0.7686, "step": 44705 }, { "epoch": 0.04, "learning_rate": 4.975488368543595e-05, "loss": 0.6763, "step": 44706 }, { "epoch": 0.04, "learning_rate": 4.975487271302393e-05, "loss": 0.6651, "step": 44707 }, { "epoch": 0.04, "learning_rate": 4.9754861740367545e-05, "loss": 0.9087, "step": 44708 }, { "epoch": 0.04, "learning_rate": 4.975485076746679e-05, "loss": 0.7607, "step": 44709 }, { "epoch": 0.04, "learning_rate": 4.975483979432165e-05, "loss": 1.0558, "step": 44710 }, { "epoch": 0.04, "learning_rate": 4.975482882093215e-05, "loss": 1.0221, "step": 44711 }, { "epoch": 0.04, "learning_rate": 4.975481784729829e-05, "loss": 0.878, "step": 44712 }, { "epoch": 0.04, "learning_rate": 4.975480687342006e-05, "loss": 1.1327, "step": 44713 }, { "epoch": 0.04, "learning_rate": 4.975479589929745e-05, "loss": 0.9941, "step": 44714 }, { "epoch": 0.04, "learning_rate": 4.975478492493048e-05, "loss": 0.9818, "step": 44715 }, { "epoch": 0.04, "learning_rate": 4.9754773950319135e-05, "loss": 1.15, "step": 44716 }, { "epoch": 0.04, "learning_rate": 4.975476297546342e-05, "loss": 1.0272, "step": 44717 }, { "epoch": 0.04, "learning_rate": 4.975475200036334e-05, "loss": 1.1155, "step": 44718 }, { "epoch": 0.04, "learning_rate": 4.975474102501889e-05, "loss": 0.5805, "step": 44719 }, { "epoch": 0.04, "learning_rate": 4.975473004943007e-05, "loss": 0.8576, "step": 44720 }, { "epoch": 0.04, "learning_rate": 4.975471907359689e-05, "loss": 1.5823, "step": 44721 }, { "epoch": 0.04, "learning_rate": 4.975470809751933e-05, "loss": 1.0981, "step": 44722 }, { "epoch": 0.04, "learning_rate": 4.9754697121197416e-05, "loss": 1.9315, "step": 44723 }, { "epoch": 0.04, "learning_rate": 4.975468614463112e-05, "loss": 1.1256, "step": 44724 }, { "epoch": 0.04, "learning_rate": 4.975467516782046e-05, "loss": 1.1036, "step": 44725 }, { "epoch": 0.04, "learning_rate": 4.9754664190765435e-05, "loss": 1.0376, "step": 44726 }, { "epoch": 0.04, "learning_rate": 4.975465321346604e-05, "loss": 1.0951, "step": 44727 }, { "epoch": 0.04, "learning_rate": 4.9754642235922286e-05, "loss": 1.142, "step": 44728 }, { "epoch": 0.04, "learning_rate": 4.9754631258134156e-05, "loss": 1.2358, "step": 44729 }, { "epoch": 0.04, "learning_rate": 4.975462028010166e-05, "loss": 1.0066, "step": 44730 }, { "epoch": 0.04, "learning_rate": 4.975460930182479e-05, "loss": 0.9037, "step": 44731 }, { "epoch": 0.04, "learning_rate": 4.975459832330357e-05, "loss": 0.6867, "step": 44732 }, { "epoch": 0.04, "learning_rate": 4.9754587344537974e-05, "loss": 0.3736, "step": 44733 }, { "epoch": 0.04, "learning_rate": 4.975457636552801e-05, "loss": 1.5164, "step": 44734 }, { "epoch": 0.04, "learning_rate": 4.975456538627368e-05, "loss": 1.294, "step": 44735 }, { "epoch": 0.04, "learning_rate": 4.975455440677498e-05, "loss": 1.0131, "step": 44736 }, { "epoch": 0.04, "learning_rate": 4.975454342703192e-05, "loss": 1.418, "step": 44737 }, { "epoch": 0.04, "learning_rate": 4.975453244704449e-05, "loss": 0.9046, "step": 44738 }, { "epoch": 0.04, "learning_rate": 4.9754521466812695e-05, "loss": 0.8484, "step": 44739 }, { "epoch": 0.04, "learning_rate": 4.975451048633654e-05, "loss": 0.6592, "step": 44740 }, { "epoch": 0.04, "learning_rate": 4.9754499505616006e-05, "loss": 0.7011, "step": 44741 }, { "epoch": 0.04, "learning_rate": 4.975448852465111e-05, "loss": 0.7187, "step": 44742 }, { "epoch": 0.04, "learning_rate": 4.975447754344185e-05, "loss": 0.6957, "step": 44743 }, { "epoch": 0.04, "learning_rate": 4.975446656198822e-05, "loss": 0.5661, "step": 44744 }, { "epoch": 0.04, "learning_rate": 4.9754455580290234e-05, "loss": 0.6814, "step": 44745 }, { "epoch": 0.04, "learning_rate": 4.9754444598347875e-05, "loss": 0.5777, "step": 44746 }, { "epoch": 0.04, "learning_rate": 4.975443361616116e-05, "loss": 0.8581, "step": 44747 }, { "epoch": 0.04, "learning_rate": 4.975442263373007e-05, "loss": 0.9126, "step": 44748 }, { "epoch": 0.04, "learning_rate": 4.975441165105461e-05, "loss": 1.0361, "step": 44749 }, { "epoch": 0.04, "learning_rate": 4.975440066813479e-05, "loss": 1.0008, "step": 44750 }, { "epoch": 0.04, "learning_rate": 4.975438968497062e-05, "loss": 1.3096, "step": 44751 }, { "epoch": 0.04, "learning_rate": 4.9754378701562065e-05, "loss": 1.2048, "step": 44752 }, { "epoch": 0.04, "learning_rate": 4.975436771790915e-05, "loss": 0.9418, "step": 44753 }, { "epoch": 0.04, "learning_rate": 4.9754356734011876e-05, "loss": 1.4237, "step": 44754 }, { "epoch": 0.04, "learning_rate": 4.9754345749870234e-05, "loss": 1.1344, "step": 44755 }, { "epoch": 0.04, "learning_rate": 4.9754334765484225e-05, "loss": 0.9671, "step": 44756 }, { "epoch": 0.04, "learning_rate": 4.975432378085385e-05, "loss": 0.6821, "step": 44757 }, { "epoch": 0.04, "learning_rate": 4.975431279597912e-05, "loss": 1.0775, "step": 44758 }, { "epoch": 0.04, "learning_rate": 4.9754301810860014e-05, "loss": 1.2316, "step": 44759 }, { "epoch": 0.04, "learning_rate": 4.975429082549655e-05, "loss": 0.9194, "step": 44760 }, { "epoch": 0.04, "learning_rate": 4.975427983988873e-05, "loss": 1.0207, "step": 44761 }, { "epoch": 0.04, "learning_rate": 4.975426885403653e-05, "loss": 0.7967, "step": 44762 }, { "epoch": 0.04, "learning_rate": 4.975425786793998e-05, "loss": 1.0806, "step": 44763 }, { "epoch": 0.04, "learning_rate": 4.9754246881599065e-05, "loss": 1.3228, "step": 44764 }, { "epoch": 0.04, "learning_rate": 4.975423589501378e-05, "loss": 0.9707, "step": 44765 }, { "epoch": 0.04, "learning_rate": 4.975422490818413e-05, "loss": 1.0923, "step": 44766 }, { "epoch": 0.04, "learning_rate": 4.9754213921110127e-05, "loss": 0.9895, "step": 44767 }, { "epoch": 0.04, "learning_rate": 4.975420293379176e-05, "loss": 1.0864, "step": 44768 }, { "epoch": 0.04, "learning_rate": 4.9754191946229024e-05, "loss": 1.3676, "step": 44769 }, { "epoch": 0.04, "learning_rate": 4.9754180958421924e-05, "loss": 0.9547, "step": 44770 }, { "epoch": 0.04, "learning_rate": 4.975416997037046e-05, "loss": 0.2065, "step": 44771 }, { "epoch": 0.04, "learning_rate": 4.975415898207464e-05, "loss": 0.1494, "step": 44772 }, { "epoch": 0.04, "learning_rate": 4.975414799353445e-05, "loss": 0.2069, "step": 44773 }, { "epoch": 0.04, "learning_rate": 4.97541370047499e-05, "loss": 0.1593, "step": 44774 }, { "epoch": 0.04, "learning_rate": 4.975412601572099e-05, "loss": 0.4166, "step": 44775 }, { "epoch": 0.04, "learning_rate": 4.975411502644772e-05, "loss": 1.2356, "step": 44776 }, { "epoch": 0.04, "learning_rate": 4.9754104036930084e-05, "loss": 2.1477, "step": 44777 }, { "epoch": 0.04, "learning_rate": 4.9754093047168086e-05, "loss": 1.3884, "step": 44778 }, { "epoch": 0.04, "learning_rate": 4.975408205716173e-05, "loss": 1.0473, "step": 44779 }, { "epoch": 0.04, "learning_rate": 4.975407106691101e-05, "loss": 0.8656, "step": 44780 }, { "epoch": 0.04, "learning_rate": 4.975406007641592e-05, "loss": 0.6559, "step": 44781 }, { "epoch": 0.04, "learning_rate": 4.975404908567648e-05, "loss": 0.9071, "step": 44782 }, { "epoch": 0.04, "learning_rate": 4.9754038094692675e-05, "loss": 1.2203, "step": 44783 }, { "epoch": 0.04, "learning_rate": 4.9754027103464504e-05, "loss": 1.0669, "step": 44784 }, { "epoch": 0.04, "learning_rate": 4.9754016111991974e-05, "loss": 0.9295, "step": 44785 }, { "epoch": 0.04, "learning_rate": 4.975400512027508e-05, "loss": 0.8734, "step": 44786 }, { "epoch": 0.04, "learning_rate": 4.975399412831384e-05, "loss": 1.1031, "step": 44787 }, { "epoch": 0.04, "learning_rate": 4.9753983136108225e-05, "loss": 0.2647, "step": 44788 }, { "epoch": 0.04, "learning_rate": 4.9753972143658247e-05, "loss": 0.1885, "step": 44789 }, { "epoch": 0.04, "learning_rate": 4.9753961150963916e-05, "loss": 0.1141, "step": 44790 }, { "epoch": 0.04, "learning_rate": 4.975395015802522e-05, "loss": 0.1346, "step": 44791 }, { "epoch": 0.04, "learning_rate": 4.9753939164842164e-05, "loss": 0.0499, "step": 44792 }, { "epoch": 0.04, "learning_rate": 4.975392817141475e-05, "loss": 0.0695, "step": 44793 }, { "epoch": 0.04, "learning_rate": 4.975391717774297e-05, "loss": 0.2828, "step": 44794 }, { "epoch": 0.04, "learning_rate": 4.975390618382684e-05, "loss": 0.4614, "step": 44795 }, { "epoch": 0.04, "learning_rate": 4.975389518966634e-05, "loss": 1.0854, "step": 44796 }, { "epoch": 0.04, "learning_rate": 4.975388419526148e-05, "loss": 1.1155, "step": 44797 }, { "epoch": 0.04, "learning_rate": 4.975387320061226e-05, "loss": 1.0671, "step": 44798 }, { "epoch": 0.04, "learning_rate": 4.9753862205718696e-05, "loss": 0.6711, "step": 44799 }, { "epoch": 0.04, "learning_rate": 4.975385121058076e-05, "loss": 0.309, "step": 44800 }, { "epoch": 0.04, "learning_rate": 4.975384021519846e-05, "loss": 0.3785, "step": 44801 }, { "epoch": 0.04, "learning_rate": 4.97538292195718e-05, "loss": 0.2926, "step": 44802 }, { "epoch": 0.04, "learning_rate": 4.9753818223700785e-05, "loss": 0.3417, "step": 44803 }, { "epoch": 0.04, "learning_rate": 4.975380722758542e-05, "loss": 0.5569, "step": 44804 }, { "epoch": 0.04, "learning_rate": 4.9753796231225683e-05, "loss": 0.2783, "step": 44805 }, { "epoch": 0.04, "learning_rate": 4.975378523462159e-05, "loss": 0.5899, "step": 44806 }, { "epoch": 0.04, "learning_rate": 4.9753774237773146e-05, "loss": 0.5167, "step": 44807 }, { "epoch": 0.04, "learning_rate": 4.975376324068033e-05, "loss": 0.4065, "step": 44808 }, { "epoch": 0.04, "learning_rate": 4.975375224334317e-05, "loss": 0.4748, "step": 44809 }, { "epoch": 0.04, "learning_rate": 4.975374124576164e-05, "loss": 0.1819, "step": 44810 }, { "epoch": 0.04, "learning_rate": 4.9753730247935746e-05, "loss": 0.5018, "step": 44811 }, { "epoch": 0.04, "learning_rate": 4.975371924986551e-05, "loss": 0.6016, "step": 44812 }, { "epoch": 0.04, "learning_rate": 4.975370825155091e-05, "loss": 0.3445, "step": 44813 }, { "epoch": 0.04, "learning_rate": 4.9753697252991946e-05, "loss": 0.2138, "step": 44814 }, { "epoch": 0.04, "learning_rate": 4.9753686254188624e-05, "loss": 0.329, "step": 44815 }, { "epoch": 0.04, "learning_rate": 4.975367525514095e-05, "loss": 0.3994, "step": 44816 }, { "epoch": 0.04, "learning_rate": 4.9753664255848917e-05, "loss": 0.3802, "step": 44817 }, { "epoch": 0.04, "learning_rate": 4.9753653256312525e-05, "loss": 0.2543, "step": 44818 }, { "epoch": 0.04, "learning_rate": 4.9753642256531774e-05, "loss": 0.2238, "step": 44819 }, { "epoch": 0.04, "learning_rate": 4.975363125650667e-05, "loss": 0.4314, "step": 44820 }, { "epoch": 0.04, "learning_rate": 4.97536202562372e-05, "loss": 0.1919, "step": 44821 }, { "epoch": 0.04, "learning_rate": 4.975360925572338e-05, "loss": 0.3393, "step": 44822 }, { "epoch": 0.04, "learning_rate": 4.97535982549652e-05, "loss": 0.1563, "step": 44823 }, { "epoch": 0.04, "learning_rate": 4.9753587253962665e-05, "loss": 0.221, "step": 44824 }, { "epoch": 0.04, "learning_rate": 4.9753576252715775e-05, "loss": 0.1512, "step": 44825 }, { "epoch": 0.04, "learning_rate": 4.975356525122452e-05, "loss": 0.1061, "step": 44826 }, { "epoch": 0.04, "learning_rate": 4.975355424948892e-05, "loss": 0.1789, "step": 44827 }, { "epoch": 0.04, "learning_rate": 4.975354324750896e-05, "loss": 0.2461, "step": 44828 }, { "epoch": 0.04, "learning_rate": 4.975353224528463e-05, "loss": 0.1733, "step": 44829 }, { "epoch": 0.04, "learning_rate": 4.975352124281596e-05, "loss": 0.1316, "step": 44830 }, { "epoch": 0.04, "learning_rate": 4.9753510240102925e-05, "loss": 0.147, "step": 44831 }, { "epoch": 0.04, "learning_rate": 4.975349923714554e-05, "loss": 0.9518, "step": 44832 }, { "epoch": 0.04, "learning_rate": 4.975348823394379e-05, "loss": 0.8017, "step": 44833 }, { "epoch": 0.04, "learning_rate": 4.975347723049769e-05, "loss": 0.9334, "step": 44834 }, { "epoch": 0.04, "learning_rate": 4.975346622680723e-05, "loss": 1.1929, "step": 44835 }, { "epoch": 0.04, "learning_rate": 4.975345522287243e-05, "loss": 1.3007, "step": 44836 }, { "epoch": 0.04, "learning_rate": 4.975344421869325e-05, "loss": 1.0461, "step": 44837 }, { "epoch": 0.04, "learning_rate": 4.975343321426973e-05, "loss": 1.2757, "step": 44838 }, { "epoch": 0.04, "learning_rate": 4.975342220960185e-05, "loss": 1.7081, "step": 44839 }, { "epoch": 0.04, "learning_rate": 4.975341120468961e-05, "loss": 1.1015, "step": 44840 }, { "epoch": 0.04, "learning_rate": 4.975340019953302e-05, "loss": 0.6531, "step": 44841 }, { "epoch": 0.04, "learning_rate": 4.9753389194132085e-05, "loss": 1.0066, "step": 44842 }, { "epoch": 0.04, "learning_rate": 4.9753378188486776e-05, "loss": 0.8918, "step": 44843 }, { "epoch": 0.04, "learning_rate": 4.975336718259712e-05, "loss": 0.7089, "step": 44844 }, { "epoch": 0.04, "learning_rate": 4.9753356176463116e-05, "loss": 0.9216, "step": 44845 }, { "epoch": 0.04, "learning_rate": 4.975334517008475e-05, "loss": 0.8726, "step": 44846 }, { "epoch": 0.04, "learning_rate": 4.975333416346203e-05, "loss": 0.9996, "step": 44847 }, { "epoch": 0.04, "learning_rate": 4.975332315659496e-05, "loss": 1.2096, "step": 44848 }, { "epoch": 0.04, "learning_rate": 4.975331214948353e-05, "loss": 0.9338, "step": 44849 }, { "epoch": 0.04, "learning_rate": 4.975330114212775e-05, "loss": 0.9201, "step": 44850 }, { "epoch": 0.04, "learning_rate": 4.9753290134527605e-05, "loss": 1.0328, "step": 44851 }, { "epoch": 0.04, "learning_rate": 4.9753279126683114e-05, "loss": 0.8648, "step": 44852 }, { "epoch": 0.04, "learning_rate": 4.975326811859428e-05, "loss": 1.0368, "step": 44853 }, { "epoch": 0.04, "learning_rate": 4.9753257110261077e-05, "loss": 0.9456, "step": 44854 }, { "epoch": 0.04, "learning_rate": 4.975324610168352e-05, "loss": 0.931, "step": 44855 }, { "epoch": 0.04, "learning_rate": 4.975323509286162e-05, "loss": 1.2036, "step": 44856 }, { "epoch": 0.04, "learning_rate": 4.975322408379536e-05, "loss": 0.9429, "step": 44857 }, { "epoch": 0.04, "learning_rate": 4.975321307448475e-05, "loss": 1.3163, "step": 44858 }, { "epoch": 0.04, "learning_rate": 4.975320206492978e-05, "loss": 1.0306, "step": 44859 }, { "epoch": 0.04, "learning_rate": 4.9753191055130466e-05, "loss": 1.5476, "step": 44860 }, { "epoch": 0.04, "learning_rate": 4.9753180045086786e-05, "loss": 1.3502, "step": 44861 }, { "epoch": 0.04, "learning_rate": 4.975316903479876e-05, "loss": 0.6222, "step": 44862 }, { "epoch": 0.04, "learning_rate": 4.975315802426639e-05, "loss": 1.1246, "step": 44863 }, { "epoch": 0.04, "learning_rate": 4.9753147013489655e-05, "loss": 1.2325, "step": 44864 }, { "epoch": 0.04, "learning_rate": 4.975313600246857e-05, "loss": 0.4396, "step": 44865 }, { "epoch": 0.04, "learning_rate": 4.9753124991203134e-05, "loss": 0.2761, "step": 44866 }, { "epoch": 0.04, "learning_rate": 4.975311397969335e-05, "loss": 0.3198, "step": 44867 }, { "epoch": 0.04, "learning_rate": 4.975310296793921e-05, "loss": 0.2452, "step": 44868 }, { "epoch": 0.04, "learning_rate": 4.9753091955940715e-05, "loss": 0.1967, "step": 44869 }, { "epoch": 0.04, "learning_rate": 4.9753080943697874e-05, "loss": 0.2111, "step": 44870 }, { "epoch": 0.04, "learning_rate": 4.975306993121067e-05, "loss": 0.2684, "step": 44871 }, { "epoch": 0.04, "learning_rate": 4.975305891847912e-05, "loss": 0.0639, "step": 44872 }, { "epoch": 0.04, "learning_rate": 4.975304790550322e-05, "loss": 0.131, "step": 44873 }, { "epoch": 0.04, "learning_rate": 4.975303689228298e-05, "loss": 0.0674, "step": 44874 }, { "epoch": 0.04, "learning_rate": 4.975302587881837e-05, "loss": 0.0422, "step": 44875 }, { "epoch": 0.04, "learning_rate": 4.975301486510942e-05, "loss": 0.0354, "step": 44876 }, { "epoch": 0.04, "learning_rate": 4.975300385115611e-05, "loss": 0.0319, "step": 44877 }, { "epoch": 0.04, "learning_rate": 4.9752992836958454e-05, "loss": 0.0295, "step": 44878 }, { "epoch": 0.04, "learning_rate": 4.9752981822516443e-05, "loss": 0.0281, "step": 44879 }, { "epoch": 0.04, "learning_rate": 4.975297080783008e-05, "loss": 0.1183, "step": 44880 }, { "epoch": 0.04, "learning_rate": 4.9752959792899374e-05, "loss": 0.8379, "step": 44881 }, { "epoch": 0.04, "learning_rate": 4.9752948777724314e-05, "loss": 1.347, "step": 44882 }, { "epoch": 0.04, "learning_rate": 4.9752937762304896e-05, "loss": 1.108, "step": 44883 }, { "epoch": 0.04, "learning_rate": 4.975292674664114e-05, "loss": 1.0976, "step": 44884 }, { "epoch": 0.04, "learning_rate": 4.9752915730733023e-05, "loss": 0.9036, "step": 44885 }, { "epoch": 0.04, "learning_rate": 4.975290471458056e-05, "loss": 1.0365, "step": 44886 }, { "epoch": 0.04, "learning_rate": 4.975289369818374e-05, "loss": 1.1883, "step": 44887 }, { "epoch": 0.04, "learning_rate": 4.975288268154258e-05, "loss": 0.7084, "step": 44888 }, { "epoch": 0.04, "learning_rate": 4.975287166465706e-05, "loss": 1.2789, "step": 44889 }, { "epoch": 0.04, "learning_rate": 4.97528606475272e-05, "loss": 1.1134, "step": 44890 }, { "epoch": 0.04, "learning_rate": 4.975284963015299e-05, "loss": 1.2318, "step": 44891 }, { "epoch": 0.04, "learning_rate": 4.975283861253442e-05, "loss": 0.9731, "step": 44892 }, { "epoch": 0.04, "learning_rate": 4.975282759467151e-05, "loss": 0.9577, "step": 44893 }, { "epoch": 0.04, "learning_rate": 4.9752816576564254e-05, "loss": 0.9122, "step": 44894 }, { "epoch": 0.04, "learning_rate": 4.975280555821264e-05, "loss": 1.0336, "step": 44895 }, { "epoch": 0.04, "learning_rate": 4.975279453961668e-05, "loss": 0.9913, "step": 44896 }, { "epoch": 0.04, "learning_rate": 4.975278352077637e-05, "loss": 1.3524, "step": 44897 }, { "epoch": 0.04, "learning_rate": 4.9752772501691705e-05, "loss": 1.3725, "step": 44898 }, { "epoch": 0.04, "learning_rate": 4.9752761482362694e-05, "loss": 1.0489, "step": 44899 }, { "epoch": 0.04, "learning_rate": 4.9752750462789345e-05, "loss": 1.1152, "step": 44900 }, { "epoch": 0.04, "learning_rate": 4.9752739442971644e-05, "loss": 0.7187, "step": 44901 }, { "epoch": 0.04, "learning_rate": 4.9752728422909583e-05, "loss": 0.6874, "step": 44902 }, { "epoch": 0.04, "learning_rate": 4.975271740260318e-05, "loss": 1.1892, "step": 44903 }, { "epoch": 0.04, "learning_rate": 4.975270638205243e-05, "loss": 0.9221, "step": 44904 }, { "epoch": 0.04, "learning_rate": 4.975269536125733e-05, "loss": 1.0483, "step": 44905 }, { "epoch": 0.04, "learning_rate": 4.975268434021788e-05, "loss": 0.8381, "step": 44906 }, { "epoch": 0.04, "learning_rate": 4.975267331893408e-05, "loss": 1.1118, "step": 44907 }, { "epoch": 0.04, "learning_rate": 4.975266229740594e-05, "loss": 1.1981, "step": 44908 }, { "epoch": 0.04, "learning_rate": 4.975265127563345e-05, "loss": 1.2222, "step": 44909 }, { "epoch": 0.04, "learning_rate": 4.975264025361661e-05, "loss": 1.1773, "step": 44910 }, { "epoch": 0.04, "learning_rate": 4.975262923135542e-05, "loss": 0.7475, "step": 44911 }, { "epoch": 0.04, "learning_rate": 4.975261820884989e-05, "loss": 0.9007, "step": 44912 }, { "epoch": 0.04, "learning_rate": 4.975260718610001e-05, "loss": 0.8683, "step": 44913 }, { "epoch": 0.04, "learning_rate": 4.975259616310578e-05, "loss": 0.9892, "step": 44914 }, { "epoch": 0.04, "learning_rate": 4.9752585139867204e-05, "loss": 0.8248, "step": 44915 }, { "epoch": 0.04, "learning_rate": 4.9752574116384276e-05, "loss": 1.229, "step": 44916 }, { "epoch": 0.04, "learning_rate": 4.975256309265701e-05, "loss": 1.053, "step": 44917 }, { "epoch": 0.04, "learning_rate": 4.975255206868539e-05, "loss": 1.034, "step": 44918 }, { "epoch": 0.04, "learning_rate": 4.975254104446943e-05, "loss": 1.1204, "step": 44919 }, { "epoch": 0.04, "learning_rate": 4.975253002000911e-05, "loss": 1.2639, "step": 44920 }, { "epoch": 0.04, "learning_rate": 4.9752518995304456e-05, "loss": 1.0992, "step": 44921 }, { "epoch": 0.04, "learning_rate": 4.9752507970355457e-05, "loss": 1.2524, "step": 44922 }, { "epoch": 0.04, "learning_rate": 4.975249694516211e-05, "loss": 0.9764, "step": 44923 }, { "epoch": 0.04, "learning_rate": 4.9752485919724414e-05, "loss": 1.0095, "step": 44924 }, { "epoch": 0.04, "learning_rate": 4.9752474894042365e-05, "loss": 1.0224, "step": 44925 }, { "epoch": 0.04, "learning_rate": 4.975246386811598e-05, "loss": 1.0915, "step": 44926 }, { "epoch": 0.04, "learning_rate": 4.975245284194524e-05, "loss": 0.4114, "step": 44927 }, { "epoch": 0.04, "learning_rate": 4.975244181553016e-05, "loss": 0.3242, "step": 44928 }, { "epoch": 0.04, "learning_rate": 4.975243078887074e-05, "loss": 0.336, "step": 44929 }, { "epoch": 0.04, "learning_rate": 4.975241976196696e-05, "loss": 1.1962, "step": 44930 }, { "epoch": 0.04, "learning_rate": 4.9752408734818846e-05, "loss": 1.3061, "step": 44931 }, { "epoch": 0.04, "learning_rate": 4.975239770742639e-05, "loss": 1.0549, "step": 44932 }, { "epoch": 0.04, "learning_rate": 4.9752386679789575e-05, "loss": 0.7283, "step": 44933 }, { "epoch": 0.04, "learning_rate": 4.9752375651908425e-05, "loss": 1.086, "step": 44934 }, { "epoch": 0.04, "learning_rate": 4.975236462378293e-05, "loss": 0.7938, "step": 44935 }, { "epoch": 0.04, "learning_rate": 4.975235359541309e-05, "loss": 1.2172, "step": 44936 }, { "epoch": 0.04, "learning_rate": 4.9752342566798896e-05, "loss": 1.3086, "step": 44937 }, { "epoch": 0.04, "learning_rate": 4.9752331537940365e-05, "loss": 1.3057, "step": 44938 }, { "epoch": 0.04, "learning_rate": 4.975232050883749e-05, "loss": 0.9294, "step": 44939 }, { "epoch": 0.04, "learning_rate": 4.975230947949027e-05, "loss": 0.8175, "step": 44940 }, { "epoch": 0.04, "learning_rate": 4.9752298449898706e-05, "loss": 0.9415, "step": 44941 }, { "epoch": 0.04, "learning_rate": 4.9752287420062794e-05, "loss": 0.9386, "step": 44942 }, { "epoch": 0.04, "learning_rate": 4.975227638998254e-05, "loss": 1.0469, "step": 44943 }, { "epoch": 0.04, "learning_rate": 4.975226535965795e-05, "loss": 0.7103, "step": 44944 }, { "epoch": 0.04, "learning_rate": 4.9752254329089e-05, "loss": 1.2237, "step": 44945 }, { "epoch": 0.04, "learning_rate": 4.975224329827571e-05, "loss": 1.387, "step": 44946 }, { "epoch": 0.04, "learning_rate": 4.975223226721809e-05, "loss": 1.1254, "step": 44947 }, { "epoch": 0.04, "learning_rate": 4.9752221235916116e-05, "loss": 2.2726, "step": 44948 }, { "epoch": 0.04, "learning_rate": 4.97522102043698e-05, "loss": 1.0963, "step": 44949 }, { "epoch": 0.04, "learning_rate": 4.975219917257914e-05, "loss": 1.7169, "step": 44950 }, { "epoch": 0.04, "learning_rate": 4.975218814054413e-05, "loss": 1.1791, "step": 44951 }, { "epoch": 0.04, "learning_rate": 4.975217710826479e-05, "loss": 1.0217, "step": 44952 }, { "epoch": 0.04, "learning_rate": 4.97521660757411e-05, "loss": 0.8002, "step": 44953 }, { "epoch": 0.04, "learning_rate": 4.975215504297306e-05, "loss": 1.1321, "step": 44954 }, { "epoch": 0.04, "learning_rate": 4.975214400996069e-05, "loss": 1.0201, "step": 44955 }, { "epoch": 0.04, "learning_rate": 4.975213297670397e-05, "loss": 1.0695, "step": 44956 }, { "epoch": 0.04, "learning_rate": 4.975212194320291e-05, "loss": 0.9391, "step": 44957 }, { "epoch": 0.04, "learning_rate": 4.975211090945751e-05, "loss": 0.9929, "step": 44958 }, { "epoch": 0.04, "learning_rate": 4.975209987546776e-05, "loss": 1.1957, "step": 44959 }, { "epoch": 0.04, "learning_rate": 4.975208884123368e-05, "loss": 1.2957, "step": 44960 }, { "epoch": 0.04, "learning_rate": 4.975207780675525e-05, "loss": 1.237, "step": 44961 }, { "epoch": 0.04, "learning_rate": 4.975206677203248e-05, "loss": 1.0741, "step": 44962 }, { "epoch": 0.04, "learning_rate": 4.975205573706536e-05, "loss": 1.1337, "step": 44963 }, { "epoch": 0.04, "learning_rate": 4.975204470185391e-05, "loss": 1.3126, "step": 44964 }, { "epoch": 0.04, "learning_rate": 4.975203366639811e-05, "loss": 1.1745, "step": 44965 }, { "epoch": 0.04, "learning_rate": 4.975202263069797e-05, "loss": 1.18, "step": 44966 }, { "epoch": 0.04, "learning_rate": 4.975201159475349e-05, "loss": 1.1576, "step": 44967 }, { "epoch": 0.04, "learning_rate": 4.975200055856466e-05, "loss": 1.2912, "step": 44968 }, { "epoch": 0.04, "learning_rate": 4.97519895221315e-05, "loss": 0.8619, "step": 44969 }, { "epoch": 0.04, "learning_rate": 4.9751978485453997e-05, "loss": 0.9547, "step": 44970 }, { "epoch": 0.04, "learning_rate": 4.975196744853215e-05, "loss": 0.7138, "step": 44971 }, { "epoch": 0.04, "learning_rate": 4.975195641136596e-05, "loss": 0.3376, "step": 44972 }, { "epoch": 0.04, "learning_rate": 4.9751945373955434e-05, "loss": 1.2377, "step": 44973 }, { "epoch": 0.04, "learning_rate": 4.975193433630057e-05, "loss": 1.1007, "step": 44974 }, { "epoch": 0.04, "learning_rate": 4.975192329840136e-05, "loss": 0.9587, "step": 44975 }, { "epoch": 0.04, "learning_rate": 4.9751912260257805e-05, "loss": 0.8838, "step": 44976 }, { "epoch": 0.04, "learning_rate": 4.9751901221869916e-05, "loss": 1.0732, "step": 44977 }, { "epoch": 0.04, "learning_rate": 4.975189018323769e-05, "loss": 1.1389, "step": 44978 }, { "epoch": 0.04, "learning_rate": 4.9751879144361116e-05, "loss": 0.9199, "step": 44979 }, { "epoch": 0.04, "learning_rate": 4.9751868105240204e-05, "loss": 0.6879, "step": 44980 }, { "epoch": 0.04, "learning_rate": 4.9751857065874955e-05, "loss": 1.0365, "step": 44981 }, { "epoch": 0.04, "learning_rate": 4.9751846026265366e-05, "loss": 0.9801, "step": 44982 }, { "epoch": 0.04, "learning_rate": 4.975183498641143e-05, "loss": 1.0043, "step": 44983 }, { "epoch": 0.04, "learning_rate": 4.975182394631316e-05, "loss": 0.9838, "step": 44984 }, { "epoch": 0.04, "learning_rate": 4.975181290597055e-05, "loss": 1.1512, "step": 44985 }, { "epoch": 0.04, "learning_rate": 4.975180186538361e-05, "loss": 0.927, "step": 44986 }, { "epoch": 0.04, "learning_rate": 4.975179082455231e-05, "loss": 0.7608, "step": 44987 }, { "epoch": 0.04, "learning_rate": 4.975177978347669e-05, "loss": 0.9134, "step": 44988 }, { "epoch": 0.04, "learning_rate": 4.975176874215672e-05, "loss": 0.999, "step": 44989 }, { "epoch": 0.04, "learning_rate": 4.975175770059242e-05, "loss": 0.9714, "step": 44990 }, { "epoch": 0.04, "learning_rate": 4.975174665878377e-05, "loss": 0.9053, "step": 44991 }, { "epoch": 0.04, "learning_rate": 4.975173561673079e-05, "loss": 1.0055, "step": 44992 }, { "epoch": 0.04, "learning_rate": 4.9751724574433464e-05, "loss": 1.0004, "step": 44993 }, { "epoch": 0.04, "learning_rate": 4.97517135318918e-05, "loss": 1.3202, "step": 44994 }, { "epoch": 0.04, "learning_rate": 4.97517024891058e-05, "loss": 1.0582, "step": 44995 }, { "epoch": 0.04, "learning_rate": 4.9751691446075464e-05, "loss": 0.9525, "step": 44996 }, { "epoch": 0.04, "learning_rate": 4.975168040280079e-05, "loss": 1.1146, "step": 44997 }, { "epoch": 0.04, "learning_rate": 4.975166935928177e-05, "loss": 1.0947, "step": 44998 }, { "epoch": 0.04, "learning_rate": 4.975165831551842e-05, "loss": 1.2053, "step": 44999 }, { "epoch": 0.04, "learning_rate": 4.975164727151073e-05, "loss": 1.2435, "step": 45000 }, { "epoch": 0.04, "eval_loss": 1.081416368484497, "eval_runtime": 126.1782, "eval_samples_per_second": 10.977, "eval_steps_per_second": 5.492, "step": 45000 }, { "epoch": 0.05, "learning_rate": 4.97516362272587e-05, "loss": 1.1749, "step": 45001 }, { "epoch": 0.05, "learning_rate": 4.9751625182762335e-05, "loss": 1.0522, "step": 45002 }, { "epoch": 0.05, "learning_rate": 4.975161413802163e-05, "loss": 1.0559, "step": 45003 }, { "epoch": 0.05, "learning_rate": 4.975160309303659e-05, "loss": 1.0992, "step": 45004 }, { "epoch": 0.05, "learning_rate": 4.9751592047807215e-05, "loss": 0.8538, "step": 45005 }, { "epoch": 0.05, "learning_rate": 4.9751581002333496e-05, "loss": 0.8421, "step": 45006 }, { "epoch": 0.05, "learning_rate": 4.975156995661544e-05, "loss": 0.9507, "step": 45007 }, { "epoch": 0.05, "learning_rate": 4.975155891065305e-05, "loss": 1.1403, "step": 45008 }, { "epoch": 0.05, "learning_rate": 4.975154786444633e-05, "loss": 1.4794, "step": 45009 }, { "epoch": 0.05, "learning_rate": 4.975153681799526e-05, "loss": 1.3929, "step": 45010 }, { "epoch": 0.05, "learning_rate": 4.975152577129986e-05, "loss": 1.094, "step": 45011 }, { "epoch": 0.05, "learning_rate": 4.9751514724360125e-05, "loss": 1.2292, "step": 45012 }, { "epoch": 0.05, "learning_rate": 4.975150367717605e-05, "loss": 1.195, "step": 45013 }, { "epoch": 0.05, "learning_rate": 4.975149262974764e-05, "loss": 1.1291, "step": 45014 }, { "epoch": 0.05, "learning_rate": 4.9751481582074896e-05, "loss": 1.0864, "step": 45015 }, { "epoch": 0.05, "learning_rate": 4.975147053415781e-05, "loss": 1.0364, "step": 45016 }, { "epoch": 0.05, "learning_rate": 4.9751459485996396e-05, "loss": 1.1367, "step": 45017 }, { "epoch": 0.05, "learning_rate": 4.975144843759064e-05, "loss": 1.121, "step": 45018 }, { "epoch": 0.05, "learning_rate": 4.975143738894055e-05, "loss": 1.7532, "step": 45019 }, { "epoch": 0.05, "learning_rate": 4.9751426340046124e-05, "loss": 1.5382, "step": 45020 }, { "epoch": 0.05, "learning_rate": 4.975141529090736e-05, "loss": 0.8824, "step": 45021 }, { "epoch": 0.05, "learning_rate": 4.9751404241524266e-05, "loss": 1.102, "step": 45022 }, { "epoch": 0.05, "learning_rate": 4.975139319189683e-05, "loss": 1.2102, "step": 45023 }, { "epoch": 0.05, "learning_rate": 4.975138214202506e-05, "loss": 1.0549, "step": 45024 }, { "epoch": 0.05, "learning_rate": 4.9751371091908963e-05, "loss": 1.3539, "step": 45025 }, { "epoch": 0.05, "learning_rate": 4.975136004154852e-05, "loss": 0.9218, "step": 45026 }, { "epoch": 0.05, "learning_rate": 4.975134899094375e-05, "loss": 1.1068, "step": 45027 }, { "epoch": 0.05, "learning_rate": 4.9751337940094635e-05, "loss": 1.1768, "step": 45028 }, { "epoch": 0.05, "learning_rate": 4.975132688900119e-05, "loss": 0.3547, "step": 45029 }, { "epoch": 0.05, "learning_rate": 4.9751315837663415e-05, "loss": 0.0718, "step": 45030 }, { "epoch": 0.05, "learning_rate": 4.975130478608131e-05, "loss": 0.0571, "step": 45031 }, { "epoch": 0.05, "learning_rate": 4.9751293734254854e-05, "loss": 0.0496, "step": 45032 }, { "epoch": 0.05, "learning_rate": 4.9751282682184076e-05, "loss": 0.0491, "step": 45033 }, { "epoch": 0.05, "learning_rate": 4.975127162986896e-05, "loss": 0.0412, "step": 45034 }, { "epoch": 0.05, "learning_rate": 4.975126057730951e-05, "loss": 0.0415, "step": 45035 }, { "epoch": 0.05, "learning_rate": 4.9751249524505725e-05, "loss": 0.0441, "step": 45036 }, { "epoch": 0.05, "learning_rate": 4.9751238471457606e-05, "loss": 0.0354, "step": 45037 }, { "epoch": 0.05, "learning_rate": 4.9751227418165156e-05, "loss": 0.0353, "step": 45038 }, { "epoch": 0.05, "learning_rate": 4.9751216364628374e-05, "loss": 0.0445, "step": 45039 }, { "epoch": 0.05, "learning_rate": 4.9751205310847254e-05, "loss": 0.0363, "step": 45040 }, { "epoch": 0.05, "learning_rate": 4.9751194256821795e-05, "loss": 0.0379, "step": 45041 }, { "epoch": 0.05, "learning_rate": 4.975118320255202e-05, "loss": 0.0335, "step": 45042 }, { "epoch": 0.05, "learning_rate": 4.9751172148037895e-05, "loss": 0.0371, "step": 45043 }, { "epoch": 0.05, "learning_rate": 4.975116109327945e-05, "loss": 0.0331, "step": 45044 }, { "epoch": 0.05, "learning_rate": 4.975115003827666e-05, "loss": 0.0304, "step": 45045 }, { "epoch": 0.05, "learning_rate": 4.9751138983029545e-05, "loss": 0.035, "step": 45046 }, { "epoch": 0.05, "learning_rate": 4.9751127927538095e-05, "loss": 0.0296, "step": 45047 }, { "epoch": 0.05, "learning_rate": 4.975111687180231e-05, "loss": 0.03, "step": 45048 }, { "epoch": 0.05, "learning_rate": 4.9751105815822195e-05, "loss": 0.0296, "step": 45049 }, { "epoch": 0.05, "learning_rate": 4.975109475959774e-05, "loss": 0.0393, "step": 45050 }, { "epoch": 0.05, "learning_rate": 4.975108370312897e-05, "loss": 0.0372, "step": 45051 }, { "epoch": 0.05, "learning_rate": 4.975107264641585e-05, "loss": 0.0404, "step": 45052 }, { "epoch": 0.05, "learning_rate": 4.9751061589458406e-05, "loss": 0.0326, "step": 45053 }, { "epoch": 0.05, "learning_rate": 4.975105053225663e-05, "loss": 0.0371, "step": 45054 }, { "epoch": 0.05, "learning_rate": 4.9751039474810524e-05, "loss": 0.0307, "step": 45055 }, { "epoch": 0.05, "learning_rate": 4.975102841712008e-05, "loss": 0.035, "step": 45056 }, { "epoch": 0.05, "learning_rate": 4.97510173591853e-05, "loss": 0.0345, "step": 45057 }, { "epoch": 0.05, "learning_rate": 4.97510063010062e-05, "loss": 0.0341, "step": 45058 }, { "epoch": 0.05, "learning_rate": 4.975099524258277e-05, "loss": 0.0309, "step": 45059 }, { "epoch": 0.05, "learning_rate": 4.9750984183915e-05, "loss": 0.0272, "step": 45060 }, { "epoch": 0.05, "learning_rate": 4.97509731250029e-05, "loss": 0.0304, "step": 45061 }, { "epoch": 0.05, "learning_rate": 4.975096206584647e-05, "loss": 0.0251, "step": 45062 }, { "epoch": 0.05, "learning_rate": 4.975095100644571e-05, "loss": 0.0271, "step": 45063 }, { "epoch": 0.05, "learning_rate": 4.975093994680062e-05, "loss": 0.0302, "step": 45064 }, { "epoch": 0.05, "learning_rate": 4.97509288869112e-05, "loss": 0.642, "step": 45065 }, { "epoch": 0.05, "learning_rate": 4.975091782677745e-05, "loss": 0.9271, "step": 45066 }, { "epoch": 0.05, "learning_rate": 4.9750906766399356e-05, "loss": 1.6864, "step": 45067 }, { "epoch": 0.05, "learning_rate": 4.975089570577695e-05, "loss": 1.8238, "step": 45068 }, { "epoch": 0.05, "learning_rate": 4.97508846449102e-05, "loss": 1.7354, "step": 45069 }, { "epoch": 0.05, "learning_rate": 4.975087358379913e-05, "loss": 1.7124, "step": 45070 }, { "epoch": 0.05, "learning_rate": 4.975086252244372e-05, "loss": 1.7294, "step": 45071 }, { "epoch": 0.05, "learning_rate": 4.975085146084399e-05, "loss": 1.7527, "step": 45072 }, { "epoch": 0.05, "learning_rate": 4.975084039899991e-05, "loss": 1.6492, "step": 45073 }, { "epoch": 0.05, "learning_rate": 4.975082933691152e-05, "loss": 1.0768, "step": 45074 }, { "epoch": 0.05, "learning_rate": 4.9750818274578795e-05, "loss": 0.8626, "step": 45075 }, { "epoch": 0.05, "learning_rate": 4.975080721200174e-05, "loss": 1.0625, "step": 45076 }, { "epoch": 0.05, "learning_rate": 4.975079614918036e-05, "loss": 1.4153, "step": 45077 }, { "epoch": 0.05, "learning_rate": 4.9750785086114645e-05, "loss": 1.0602, "step": 45078 }, { "epoch": 0.05, "learning_rate": 4.97507740228046e-05, "loss": 1.2477, "step": 45079 }, { "epoch": 0.05, "learning_rate": 4.975076295925023e-05, "loss": 1.1645, "step": 45080 }, { "epoch": 0.05, "learning_rate": 4.975075189545153e-05, "loss": 1.1143, "step": 45081 }, { "epoch": 0.05, "learning_rate": 4.975074083140849e-05, "loss": 1.4485, "step": 45082 }, { "epoch": 0.05, "learning_rate": 4.975072976712114e-05, "loss": 1.6532, "step": 45083 }, { "epoch": 0.05, "learning_rate": 4.975071870258945e-05, "loss": 1.0953, "step": 45084 }, { "epoch": 0.05, "learning_rate": 4.9750707637813435e-05, "loss": 1.0015, "step": 45085 }, { "epoch": 0.05, "learning_rate": 4.975069657279309e-05, "loss": 0.8919, "step": 45086 }, { "epoch": 0.05, "learning_rate": 4.975068550752842e-05, "loss": 1.2019, "step": 45087 }, { "epoch": 0.05, "learning_rate": 4.9750674442019414e-05, "loss": 1.0006, "step": 45088 }, { "epoch": 0.05, "learning_rate": 4.975066337626608e-05, "loss": 1.2085, "step": 45089 }, { "epoch": 0.05, "learning_rate": 4.975065231026842e-05, "loss": 0.9528, "step": 45090 }, { "epoch": 0.05, "learning_rate": 4.9750641244026434e-05, "loss": 1.0163, "step": 45091 }, { "epoch": 0.05, "learning_rate": 4.9750630177540125e-05, "loss": 0.9069, "step": 45092 }, { "epoch": 0.05, "learning_rate": 4.975061911080948e-05, "loss": 1.2189, "step": 45093 }, { "epoch": 0.05, "learning_rate": 4.975060804383451e-05, "loss": 0.6215, "step": 45094 }, { "epoch": 0.05, "learning_rate": 4.975059697661522e-05, "loss": 0.4131, "step": 45095 }, { "epoch": 0.05, "learning_rate": 4.9750585909151596e-05, "loss": 0.3299, "step": 45096 }, { "epoch": 0.05, "learning_rate": 4.9750574841443634e-05, "loss": 0.3197, "step": 45097 }, { "epoch": 0.05, "learning_rate": 4.9750563773491354e-05, "loss": 0.4498, "step": 45098 }, { "epoch": 0.05, "learning_rate": 4.975055270529475e-05, "loss": 0.4816, "step": 45099 }, { "epoch": 0.05, "learning_rate": 4.975054163685382e-05, "loss": 0.5327, "step": 45100 }, { "epoch": 0.05, "learning_rate": 4.975053056816856e-05, "loss": 1.0949, "step": 45101 }, { "epoch": 0.05, "learning_rate": 4.9750519499238965e-05, "loss": 1.3901, "step": 45102 }, { "epoch": 0.05, "learning_rate": 4.9750508430065054e-05, "loss": 0.9668, "step": 45103 }, { "epoch": 0.05, "learning_rate": 4.975049736064681e-05, "loss": 0.9951, "step": 45104 }, { "epoch": 0.05, "learning_rate": 4.975048629098424e-05, "loss": 0.8842, "step": 45105 }, { "epoch": 0.05, "learning_rate": 4.975047522107735e-05, "loss": 1.0635, "step": 45106 }, { "epoch": 0.05, "learning_rate": 4.975046415092613e-05, "loss": 0.9109, "step": 45107 }, { "epoch": 0.05, "learning_rate": 4.975045308053059e-05, "loss": 0.9959, "step": 45108 }, { "epoch": 0.05, "learning_rate": 4.975044200989072e-05, "loss": 1.1767, "step": 45109 }, { "epoch": 0.05, "learning_rate": 4.975043093900652e-05, "loss": 1.2173, "step": 45110 }, { "epoch": 0.05, "learning_rate": 4.9750419867878e-05, "loss": 1.074, "step": 45111 }, { "epoch": 0.05, "learning_rate": 4.975040879650514e-05, "loss": 1.2368, "step": 45112 }, { "epoch": 0.05, "learning_rate": 4.975039772488797e-05, "loss": 1.2961, "step": 45113 }, { "epoch": 0.05, "learning_rate": 4.9750386653026474e-05, "loss": 1.1593, "step": 45114 }, { "epoch": 0.05, "learning_rate": 4.975037558092064e-05, "loss": 0.8731, "step": 45115 }, { "epoch": 0.05, "learning_rate": 4.975036450857049e-05, "loss": 1.2964, "step": 45116 }, { "epoch": 0.05, "learning_rate": 4.975035343597602e-05, "loss": 0.7613, "step": 45117 }, { "epoch": 0.05, "learning_rate": 4.975034236313721e-05, "loss": 1.2456, "step": 45118 }, { "epoch": 0.05, "learning_rate": 4.975033129005409e-05, "loss": 1.1556, "step": 45119 }, { "epoch": 0.05, "learning_rate": 4.975032021672664e-05, "loss": 1.1195, "step": 45120 }, { "epoch": 0.05, "learning_rate": 4.975030914315486e-05, "loss": 0.9922, "step": 45121 }, { "epoch": 0.05, "learning_rate": 4.9750298069338766e-05, "loss": 1.1315, "step": 45122 }, { "epoch": 0.05, "learning_rate": 4.975028699527834e-05, "loss": 1.2117, "step": 45123 }, { "epoch": 0.05, "learning_rate": 4.9750275920973586e-05, "loss": 1.0989, "step": 45124 }, { "epoch": 0.05, "learning_rate": 4.975026484642451e-05, "loss": 0.8537, "step": 45125 }, { "epoch": 0.05, "learning_rate": 4.975025377163112e-05, "loss": 0.9661, "step": 45126 }, { "epoch": 0.05, "learning_rate": 4.9750242696593396e-05, "loss": 0.9866, "step": 45127 }, { "epoch": 0.05, "learning_rate": 4.975023162131135e-05, "loss": 0.8868, "step": 45128 }, { "epoch": 0.05, "learning_rate": 4.975022054578497e-05, "loss": 0.7089, "step": 45129 }, { "epoch": 0.05, "learning_rate": 4.975020947001429e-05, "loss": 1.0286, "step": 45130 }, { "epoch": 0.05, "learning_rate": 4.975019839399927e-05, "loss": 1.2002, "step": 45131 }, { "epoch": 0.05, "learning_rate": 4.975018731773993e-05, "loss": 1.2059, "step": 45132 }, { "epoch": 0.05, "learning_rate": 4.975017624123626e-05, "loss": 1.1255, "step": 45133 }, { "epoch": 0.05, "learning_rate": 4.975016516448827e-05, "loss": 1.1136, "step": 45134 }, { "epoch": 0.05, "learning_rate": 4.975015408749597e-05, "loss": 0.7031, "step": 45135 }, { "epoch": 0.05, "learning_rate": 4.975014301025933e-05, "loss": 1.0829, "step": 45136 }, { "epoch": 0.05, "learning_rate": 4.975013193277838e-05, "loss": 0.9216, "step": 45137 }, { "epoch": 0.05, "learning_rate": 4.9750120855053094e-05, "loss": 1.1699, "step": 45138 }, { "epoch": 0.05, "learning_rate": 4.975010977708349e-05, "loss": 1.1037, "step": 45139 }, { "epoch": 0.05, "learning_rate": 4.975009869886957e-05, "loss": 1.0175, "step": 45140 }, { "epoch": 0.05, "learning_rate": 4.975008762041132e-05, "loss": 1.1403, "step": 45141 }, { "epoch": 0.05, "learning_rate": 4.975007654170875e-05, "loss": 1.0154, "step": 45142 }, { "epoch": 0.05, "learning_rate": 4.9750065462761856e-05, "loss": 0.9215, "step": 45143 }, { "epoch": 0.05, "learning_rate": 4.975005438357064e-05, "loss": 1.1156, "step": 45144 }, { "epoch": 0.05, "learning_rate": 4.975004330413511e-05, "loss": 0.9766, "step": 45145 }, { "epoch": 0.05, "learning_rate": 4.9750032224455244e-05, "loss": 0.9748, "step": 45146 }, { "epoch": 0.05, "learning_rate": 4.975002114453107e-05, "loss": 0.9899, "step": 45147 }, { "epoch": 0.05, "learning_rate": 4.975001006436256e-05, "loss": 0.7922, "step": 45148 }, { "epoch": 0.05, "learning_rate": 4.974999898394974e-05, "loss": 0.853, "step": 45149 }, { "epoch": 0.05, "learning_rate": 4.974998790329259e-05, "loss": 1.2391, "step": 45150 }, { "epoch": 0.05, "learning_rate": 4.974997682239113e-05, "loss": 0.9765, "step": 45151 }, { "epoch": 0.05, "learning_rate": 4.974996574124534e-05, "loss": 0.8738, "step": 45152 }, { "epoch": 0.05, "learning_rate": 4.974995465985523e-05, "loss": 1.0319, "step": 45153 }, { "epoch": 0.05, "learning_rate": 4.97499435782208e-05, "loss": 0.9573, "step": 45154 }, { "epoch": 0.05, "learning_rate": 4.9749932496342046e-05, "loss": 0.7157, "step": 45155 }, { "epoch": 0.05, "learning_rate": 4.974992141421897e-05, "loss": 0.9446, "step": 45156 }, { "epoch": 0.05, "learning_rate": 4.974991033185158e-05, "loss": 0.7535, "step": 45157 }, { "epoch": 0.05, "learning_rate": 4.9749899249239865e-05, "loss": 1.3936, "step": 45158 }, { "epoch": 0.05, "learning_rate": 4.974988816638384e-05, "loss": 1.1573, "step": 45159 }, { "epoch": 0.05, "learning_rate": 4.974987708328348e-05, "loss": 1.1638, "step": 45160 }, { "epoch": 0.05, "learning_rate": 4.974986599993881e-05, "loss": 0.904, "step": 45161 }, { "epoch": 0.05, "learning_rate": 4.974985491634981e-05, "loss": 0.8174, "step": 45162 }, { "epoch": 0.05, "learning_rate": 4.974984383251649e-05, "loss": 0.9463, "step": 45163 }, { "epoch": 0.05, "learning_rate": 4.974983274843885e-05, "loss": 0.9484, "step": 45164 }, { "epoch": 0.05, "learning_rate": 4.97498216641169e-05, "loss": 1.1732, "step": 45165 }, { "epoch": 0.05, "learning_rate": 4.9749810579550625e-05, "loss": 1.0148, "step": 45166 }, { "epoch": 0.05, "learning_rate": 4.974979949474003e-05, "loss": 0.8493, "step": 45167 }, { "epoch": 0.05, "learning_rate": 4.974978840968511e-05, "loss": 0.7276, "step": 45168 }, { "epoch": 0.05, "learning_rate": 4.974977732438588e-05, "loss": 1.064, "step": 45169 }, { "epoch": 0.05, "learning_rate": 4.974976623884233e-05, "loss": 0.899, "step": 45170 }, { "epoch": 0.05, "learning_rate": 4.9749755153054455e-05, "loss": 0.8799, "step": 45171 }, { "epoch": 0.05, "learning_rate": 4.9749744067022264e-05, "loss": 1.1309, "step": 45172 }, { "epoch": 0.05, "learning_rate": 4.974973298074575e-05, "loss": 0.9082, "step": 45173 }, { "epoch": 0.05, "learning_rate": 4.974972189422492e-05, "loss": 0.9244, "step": 45174 }, { "epoch": 0.05, "learning_rate": 4.974971080745977e-05, "loss": 1.0575, "step": 45175 }, { "epoch": 0.05, "learning_rate": 4.9749699720450304e-05, "loss": 1.1233, "step": 45176 }, { "epoch": 0.05, "learning_rate": 4.9749688633196515e-05, "loss": 1.0123, "step": 45177 }, { "epoch": 0.05, "learning_rate": 4.9749677545698414e-05, "loss": 1.1396, "step": 45178 }, { "epoch": 0.05, "learning_rate": 4.974966645795599e-05, "loss": 1.1146, "step": 45179 }, { "epoch": 0.05, "learning_rate": 4.974965536996925e-05, "loss": 0.9914, "step": 45180 }, { "epoch": 0.05, "learning_rate": 4.9749644281738184e-05, "loss": 0.8558, "step": 45181 }, { "epoch": 0.05, "learning_rate": 4.974963319326281e-05, "loss": 1.0081, "step": 45182 }, { "epoch": 0.05, "learning_rate": 4.974962210454311e-05, "loss": 0.7723, "step": 45183 }, { "epoch": 0.05, "learning_rate": 4.97496110155791e-05, "loss": 0.9731, "step": 45184 }, { "epoch": 0.05, "learning_rate": 4.974959992637077e-05, "loss": 0.9544, "step": 45185 }, { "epoch": 0.05, "learning_rate": 4.974958883691812e-05, "loss": 1.1308, "step": 45186 }, { "epoch": 0.05, "learning_rate": 4.974957774722115e-05, "loss": 1.2003, "step": 45187 }, { "epoch": 0.05, "learning_rate": 4.974956665727987e-05, "loss": 1.31, "step": 45188 }, { "epoch": 0.05, "learning_rate": 4.974955556709427e-05, "loss": 0.9887, "step": 45189 }, { "epoch": 0.05, "learning_rate": 4.974954447666435e-05, "loss": 1.0726, "step": 45190 }, { "epoch": 0.05, "learning_rate": 4.9749533385990114e-05, "loss": 0.9515, "step": 45191 }, { "epoch": 0.05, "learning_rate": 4.974952229507156e-05, "loss": 0.8425, "step": 45192 }, { "epoch": 0.05, "learning_rate": 4.974951120390869e-05, "loss": 0.616, "step": 45193 }, { "epoch": 0.05, "learning_rate": 4.974950011250151e-05, "loss": 1.0505, "step": 45194 }, { "epoch": 0.05, "learning_rate": 4.9749489020850006e-05, "loss": 0.974, "step": 45195 }, { "epoch": 0.05, "learning_rate": 4.9749477928954185e-05, "loss": 0.684, "step": 45196 }, { "epoch": 0.05, "learning_rate": 4.974946683681405e-05, "loss": 1.146, "step": 45197 }, { "epoch": 0.05, "learning_rate": 4.9749455744429595e-05, "loss": 0.7992, "step": 45198 }, { "epoch": 0.05, "learning_rate": 4.9749444651800834e-05, "loss": 1.2101, "step": 45199 }, { "epoch": 0.05, "learning_rate": 4.974943355892775e-05, "loss": 0.7671, "step": 45200 }, { "epoch": 0.05, "learning_rate": 4.974942246581035e-05, "loss": 0.5096, "step": 45201 }, { "epoch": 0.05, "learning_rate": 4.974941137244863e-05, "loss": 1.0877, "step": 45202 }, { "epoch": 0.05, "learning_rate": 4.97494002788426e-05, "loss": 1.2123, "step": 45203 }, { "epoch": 0.05, "learning_rate": 4.974938918499226e-05, "loss": 0.868, "step": 45204 }, { "epoch": 0.05, "learning_rate": 4.9749378090897594e-05, "loss": 1.2787, "step": 45205 }, { "epoch": 0.05, "learning_rate": 4.974936699655861e-05, "loss": 1.2031, "step": 45206 }, { "epoch": 0.05, "learning_rate": 4.974935590197531e-05, "loss": 0.9604, "step": 45207 }, { "epoch": 0.05, "learning_rate": 4.974934480714771e-05, "loss": 1.0682, "step": 45208 }, { "epoch": 0.05, "learning_rate": 4.974933371207578e-05, "loss": 1.0775, "step": 45209 }, { "epoch": 0.05, "learning_rate": 4.974932261675955e-05, "loss": 0.9724, "step": 45210 }, { "epoch": 0.05, "learning_rate": 4.9749311521198995e-05, "loss": 1.2723, "step": 45211 }, { "epoch": 0.05, "learning_rate": 4.9749300425394124e-05, "loss": 1.0278, "step": 45212 }, { "epoch": 0.05, "learning_rate": 4.974928932934494e-05, "loss": 1.361, "step": 45213 }, { "epoch": 0.05, "learning_rate": 4.974927823305145e-05, "loss": 1.0255, "step": 45214 }, { "epoch": 0.05, "learning_rate": 4.9749267136513635e-05, "loss": 0.9509, "step": 45215 }, { "epoch": 0.05, "learning_rate": 4.9749256039731505e-05, "loss": 1.0657, "step": 45216 }, { "epoch": 0.05, "learning_rate": 4.974924494270506e-05, "loss": 1.1995, "step": 45217 }, { "epoch": 0.05, "learning_rate": 4.974923384543431e-05, "loss": 0.9356, "step": 45218 }, { "epoch": 0.05, "learning_rate": 4.974922274791924e-05, "loss": 1.0813, "step": 45219 }, { "epoch": 0.05, "learning_rate": 4.9749211650159856e-05, "loss": 1.1885, "step": 45220 }, { "epoch": 0.05, "learning_rate": 4.974920055215616e-05, "loss": 1.0586, "step": 45221 }, { "epoch": 0.05, "learning_rate": 4.974918945390815e-05, "loss": 0.4824, "step": 45222 }, { "epoch": 0.05, "learning_rate": 4.9749178355415825e-05, "loss": 0.5888, "step": 45223 }, { "epoch": 0.05, "learning_rate": 4.9749167256679184e-05, "loss": 0.9159, "step": 45224 }, { "epoch": 0.05, "learning_rate": 4.974915615769824e-05, "loss": 1.1589, "step": 45225 }, { "epoch": 0.05, "learning_rate": 4.974914505847297e-05, "loss": 0.8721, "step": 45226 }, { "epoch": 0.05, "learning_rate": 4.9749133959003396e-05, "loss": 1.0649, "step": 45227 }, { "epoch": 0.05, "learning_rate": 4.974912285928951e-05, "loss": 0.9134, "step": 45228 }, { "epoch": 0.05, "learning_rate": 4.97491117593313e-05, "loss": 0.704, "step": 45229 }, { "epoch": 0.05, "learning_rate": 4.9749100659128786e-05, "loss": 1.0633, "step": 45230 }, { "epoch": 0.05, "learning_rate": 4.9749089558681964e-05, "loss": 1.1241, "step": 45231 }, { "epoch": 0.05, "learning_rate": 4.974907845799082e-05, "loss": 0.8353, "step": 45232 }, { "epoch": 0.05, "learning_rate": 4.9749067357055366e-05, "loss": 0.2652, "step": 45233 }, { "epoch": 0.05, "learning_rate": 4.9749056255875596e-05, "loss": 1.0149, "step": 45234 }, { "epoch": 0.05, "learning_rate": 4.974904515445152e-05, "loss": 1.1209, "step": 45235 }, { "epoch": 0.05, "learning_rate": 4.974903405278313e-05, "loss": 1.1441, "step": 45236 }, { "epoch": 0.05, "learning_rate": 4.9749022950870426e-05, "loss": 0.4583, "step": 45237 }, { "epoch": 0.05, "learning_rate": 4.974901184871341e-05, "loss": 0.6922, "step": 45238 }, { "epoch": 0.05, "learning_rate": 4.974900074631209e-05, "loss": 1.0469, "step": 45239 }, { "epoch": 0.05, "learning_rate": 4.9748989643666446e-05, "loss": 0.7679, "step": 45240 }, { "epoch": 0.05, "learning_rate": 4.9748978540776504e-05, "loss": 1.441, "step": 45241 }, { "epoch": 0.05, "learning_rate": 4.974896743764224e-05, "loss": 1.257, "step": 45242 }, { "epoch": 0.05, "learning_rate": 4.974895633426367e-05, "loss": 0.773, "step": 45243 }, { "epoch": 0.05, "learning_rate": 4.974894523064079e-05, "loss": 0.1957, "step": 45244 }, { "epoch": 0.05, "learning_rate": 4.974893412677359e-05, "loss": 0.1528, "step": 45245 }, { "epoch": 0.05, "learning_rate": 4.9748923022662086e-05, "loss": 1.0194, "step": 45246 }, { "epoch": 0.05, "learning_rate": 4.974891191830627e-05, "loss": 0.981, "step": 45247 }, { "epoch": 0.05, "learning_rate": 4.974890081370614e-05, "loss": 1.0224, "step": 45248 }, { "epoch": 0.05, "learning_rate": 4.97488897088617e-05, "loss": 1.1468, "step": 45249 }, { "epoch": 0.05, "learning_rate": 4.974887860377295e-05, "loss": 1.1514, "step": 45250 }, { "epoch": 0.05, "learning_rate": 4.9748867498439896e-05, "loss": 1.3772, "step": 45251 }, { "epoch": 0.05, "learning_rate": 4.974885639286252e-05, "loss": 1.2337, "step": 45252 }, { "epoch": 0.05, "learning_rate": 4.974884528704085e-05, "loss": 0.9497, "step": 45253 }, { "epoch": 0.05, "learning_rate": 4.974883418097486e-05, "loss": 1.3047, "step": 45254 }, { "epoch": 0.05, "learning_rate": 4.9748823074664555e-05, "loss": 1.1624, "step": 45255 }, { "epoch": 0.05, "learning_rate": 4.974881196810994e-05, "loss": 0.575, "step": 45256 }, { "epoch": 0.05, "learning_rate": 4.974880086131102e-05, "loss": 0.8556, "step": 45257 }, { "epoch": 0.05, "learning_rate": 4.974878975426779e-05, "loss": 1.1081, "step": 45258 }, { "epoch": 0.05, "learning_rate": 4.974877864698025e-05, "loss": 0.8327, "step": 45259 }, { "epoch": 0.05, "learning_rate": 4.9748767539448406e-05, "loss": 0.9423, "step": 45260 }, { "epoch": 0.05, "learning_rate": 4.974875643167225e-05, "loss": 1.0396, "step": 45261 }, { "epoch": 0.05, "learning_rate": 4.974874532365178e-05, "loss": 1.0444, "step": 45262 }, { "epoch": 0.05, "learning_rate": 4.974873421538701e-05, "loss": 0.8424, "step": 45263 }, { "epoch": 0.05, "learning_rate": 4.9748723106877916e-05, "loss": 1.2464, "step": 45264 }, { "epoch": 0.05, "learning_rate": 4.974871199812452e-05, "loss": 1.1604, "step": 45265 }, { "epoch": 0.05, "learning_rate": 4.974870088912682e-05, "loss": 1.3916, "step": 45266 }, { "epoch": 0.05, "learning_rate": 4.974868977988481e-05, "loss": 1.2731, "step": 45267 }, { "epoch": 0.05, "learning_rate": 4.9748678670398483e-05, "loss": 0.9024, "step": 45268 }, { "epoch": 0.05, "learning_rate": 4.9748667560667855e-05, "loss": 1.079, "step": 45269 }, { "epoch": 0.05, "learning_rate": 4.974865645069292e-05, "loss": 1.0964, "step": 45270 }, { "epoch": 0.05, "learning_rate": 4.974864534047367e-05, "loss": 1.3282, "step": 45271 }, { "epoch": 0.05, "learning_rate": 4.9748634230010116e-05, "loss": 1.1647, "step": 45272 }, { "epoch": 0.05, "learning_rate": 4.974862311930225e-05, "loss": 1.1202, "step": 45273 }, { "epoch": 0.05, "learning_rate": 4.9748612008350084e-05, "loss": 1.1994, "step": 45274 }, { "epoch": 0.05, "learning_rate": 4.974860089715361e-05, "loss": 1.4489, "step": 45275 }, { "epoch": 0.05, "learning_rate": 4.974858978571282e-05, "loss": 1.2949, "step": 45276 }, { "epoch": 0.05, "learning_rate": 4.974857867402772e-05, "loss": 1.0715, "step": 45277 }, { "epoch": 0.05, "learning_rate": 4.9748567562098324e-05, "loss": 1.1513, "step": 45278 }, { "epoch": 0.05, "learning_rate": 4.974855644992461e-05, "loss": 1.2285, "step": 45279 }, { "epoch": 0.05, "learning_rate": 4.9748545337506596e-05, "loss": 0.8852, "step": 45280 }, { "epoch": 0.05, "learning_rate": 4.974853422484428e-05, "loss": 1.0448, "step": 45281 }, { "epoch": 0.05, "learning_rate": 4.974852311193764e-05, "loss": 1.0282, "step": 45282 }, { "epoch": 0.05, "learning_rate": 4.974851199878671e-05, "loss": 0.5898, "step": 45283 }, { "epoch": 0.05, "learning_rate": 4.9748500885391464e-05, "loss": 0.5369, "step": 45284 }, { "epoch": 0.05, "learning_rate": 4.974848977175191e-05, "loss": 1.1653, "step": 45285 }, { "epoch": 0.05, "learning_rate": 4.9748478657868054e-05, "loss": 1.161, "step": 45286 }, { "epoch": 0.05, "learning_rate": 4.974846754373989e-05, "loss": 0.997, "step": 45287 }, { "epoch": 0.05, "learning_rate": 4.974845642936742e-05, "loss": 0.6296, "step": 45288 }, { "epoch": 0.05, "learning_rate": 4.974844531475064e-05, "loss": 1.22, "step": 45289 }, { "epoch": 0.05, "learning_rate": 4.974843419988956e-05, "loss": 0.9667, "step": 45290 }, { "epoch": 0.05, "learning_rate": 4.9748423084784165e-05, "loss": 1.1858, "step": 45291 }, { "epoch": 0.05, "learning_rate": 4.974841196943447e-05, "loss": 1.0047, "step": 45292 }, { "epoch": 0.05, "learning_rate": 4.974840085384047e-05, "loss": 0.9149, "step": 45293 }, { "epoch": 0.05, "learning_rate": 4.9748389738002164e-05, "loss": 0.6327, "step": 45294 }, { "epoch": 0.05, "learning_rate": 4.9748378621919555e-05, "loss": 0.8862, "step": 45295 }, { "epoch": 0.05, "learning_rate": 4.974836750559263e-05, "loss": 0.7488, "step": 45296 }, { "epoch": 0.05, "learning_rate": 4.974835638902141e-05, "loss": 0.8968, "step": 45297 }, { "epoch": 0.05, "learning_rate": 4.9748345272205885e-05, "loss": 1.0556, "step": 45298 }, { "epoch": 0.05, "learning_rate": 4.974833415514605e-05, "loss": 1.1312, "step": 45299 }, { "epoch": 0.05, "learning_rate": 4.9748323037841904e-05, "loss": 1.132, "step": 45300 }, { "epoch": 0.05, "learning_rate": 4.974831192029347e-05, "loss": 1.5206, "step": 45301 }, { "epoch": 0.05, "learning_rate": 4.974830080250071e-05, "loss": 0.9617, "step": 45302 }, { "epoch": 0.05, "learning_rate": 4.974828968446366e-05, "loss": 1.1518, "step": 45303 }, { "epoch": 0.05, "learning_rate": 4.97482785661823e-05, "loss": 1.0042, "step": 45304 }, { "epoch": 0.05, "learning_rate": 4.974826744765664e-05, "loss": 0.8067, "step": 45305 }, { "epoch": 0.05, "learning_rate": 4.974825632888666e-05, "loss": 1.226, "step": 45306 }, { "epoch": 0.05, "learning_rate": 4.97482452098724e-05, "loss": 1.175, "step": 45307 }, { "epoch": 0.05, "learning_rate": 4.974823409061382e-05, "loss": 1.1052, "step": 45308 }, { "epoch": 0.05, "learning_rate": 4.9748222971110934e-05, "loss": 1.1019, "step": 45309 }, { "epoch": 0.05, "learning_rate": 4.9748211851363755e-05, "loss": 1.3041, "step": 45310 }, { "epoch": 0.05, "learning_rate": 4.9748200731372264e-05, "loss": 1.0257, "step": 45311 }, { "epoch": 0.05, "learning_rate": 4.9748189611136476e-05, "loss": 1.0521, "step": 45312 }, { "epoch": 0.05, "learning_rate": 4.9748178490656377e-05, "loss": 1.6129, "step": 45313 }, { "epoch": 0.05, "learning_rate": 4.974816736993197e-05, "loss": 0.9352, "step": 45314 }, { "epoch": 0.05, "learning_rate": 4.974815624896328e-05, "loss": 1.1099, "step": 45315 }, { "epoch": 0.05, "learning_rate": 4.974814512775027e-05, "loss": 1.8138, "step": 45316 }, { "epoch": 0.05, "learning_rate": 4.974813400629296e-05, "loss": 1.0721, "step": 45317 }, { "epoch": 0.05, "learning_rate": 4.974812288459134e-05, "loss": 1.1049, "step": 45318 }, { "epoch": 0.05, "learning_rate": 4.974811176264543e-05, "loss": 1.0869, "step": 45319 }, { "epoch": 0.05, "learning_rate": 4.974810064045521e-05, "loss": 1.1086, "step": 45320 }, { "epoch": 0.05, "learning_rate": 4.9748089518020694e-05, "loss": 0.4041, "step": 45321 }, { "epoch": 0.05, "learning_rate": 4.974807839534186e-05, "loss": 0.8972, "step": 45322 }, { "epoch": 0.05, "learning_rate": 4.9748067272418734e-05, "loss": 1.1584, "step": 45323 }, { "epoch": 0.05, "learning_rate": 4.974805614925131e-05, "loss": 1.1507, "step": 45324 }, { "epoch": 0.05, "learning_rate": 4.9748045025839575e-05, "loss": 1.2661, "step": 45325 }, { "epoch": 0.05, "learning_rate": 4.9748033902183536e-05, "loss": 1.1833, "step": 45326 }, { "epoch": 0.05, "learning_rate": 4.9748022778283205e-05, "loss": 1.1638, "step": 45327 }, { "epoch": 0.05, "learning_rate": 4.974801165413856e-05, "loss": 1.156, "step": 45328 }, { "epoch": 0.05, "learning_rate": 4.974800052974963e-05, "loss": 1.2618, "step": 45329 }, { "epoch": 0.05, "learning_rate": 4.974798940511638e-05, "loss": 1.0741, "step": 45330 }, { "epoch": 0.05, "learning_rate": 4.9747978280238837e-05, "loss": 0.9962, "step": 45331 }, { "epoch": 0.05, "learning_rate": 4.974796715511699e-05, "loss": 0.9079, "step": 45332 }, { "epoch": 0.05, "learning_rate": 4.9747956029750845e-05, "loss": 1.0099, "step": 45333 }, { "epoch": 0.05, "learning_rate": 4.9747944904140395e-05, "loss": 1.0306, "step": 45334 }, { "epoch": 0.05, "learning_rate": 4.974793377828565e-05, "loss": 1.0914, "step": 45335 }, { "epoch": 0.05, "learning_rate": 4.9747922652186596e-05, "loss": 0.9007, "step": 45336 }, { "epoch": 0.05, "learning_rate": 4.9747911525843246e-05, "loss": 0.696, "step": 45337 }, { "epoch": 0.05, "learning_rate": 4.974790039925559e-05, "loss": 1.2572, "step": 45338 }, { "epoch": 0.05, "learning_rate": 4.974788927242364e-05, "loss": 0.9738, "step": 45339 }, { "epoch": 0.05, "learning_rate": 4.974787814534738e-05, "loss": 1.035, "step": 45340 }, { "epoch": 0.05, "learning_rate": 4.974786701802683e-05, "loss": 0.8461, "step": 45341 }, { "epoch": 0.05, "learning_rate": 4.974785589046197e-05, "loss": 0.9708, "step": 45342 }, { "epoch": 0.05, "learning_rate": 4.974784476265282e-05, "loss": 1.2521, "step": 45343 }, { "epoch": 0.05, "learning_rate": 4.974783363459936e-05, "loss": 0.9828, "step": 45344 }, { "epoch": 0.05, "learning_rate": 4.97478225063016e-05, "loss": 0.7231, "step": 45345 }, { "epoch": 0.05, "learning_rate": 4.9747811377759546e-05, "loss": 1.0131, "step": 45346 }, { "epoch": 0.05, "learning_rate": 4.974780024897318e-05, "loss": 1.538, "step": 45347 }, { "epoch": 0.05, "learning_rate": 4.974778911994254e-05, "loss": 1.4834, "step": 45348 }, { "epoch": 0.05, "learning_rate": 4.974777799066758e-05, "loss": 1.0353, "step": 45349 }, { "epoch": 0.05, "learning_rate": 4.974776686114832e-05, "loss": 1.0018, "step": 45350 }, { "epoch": 0.05, "learning_rate": 4.974775573138476e-05, "loss": 1.051, "step": 45351 }, { "epoch": 0.05, "learning_rate": 4.974774460137691e-05, "loss": 1.2842, "step": 45352 }, { "epoch": 0.05, "learning_rate": 4.9747733471124755e-05, "loss": 0.6176, "step": 45353 }, { "epoch": 0.05, "learning_rate": 4.97477223406283e-05, "loss": 0.8435, "step": 45354 }, { "epoch": 0.05, "learning_rate": 4.974771120988755e-05, "loss": 1.1477, "step": 45355 }, { "epoch": 0.05, "learning_rate": 4.9747700078902506e-05, "loss": 0.9518, "step": 45356 }, { "epoch": 0.05, "learning_rate": 4.974768894767315e-05, "loss": 1.2132, "step": 45357 }, { "epoch": 0.05, "learning_rate": 4.97476778161995e-05, "loss": 0.9391, "step": 45358 }, { "epoch": 0.05, "learning_rate": 4.9747666684481556e-05, "loss": 1.1341, "step": 45359 }, { "epoch": 0.05, "learning_rate": 4.974765555251931e-05, "loss": 0.8503, "step": 45360 }, { "epoch": 0.05, "learning_rate": 4.974764442031276e-05, "loss": 0.8064, "step": 45361 }, { "epoch": 0.05, "learning_rate": 4.974763328786192e-05, "loss": 0.8399, "step": 45362 }, { "epoch": 0.05, "learning_rate": 4.974762215516678e-05, "loss": 0.9692, "step": 45363 }, { "epoch": 0.05, "learning_rate": 4.974761102222734e-05, "loss": 1.0445, "step": 45364 }, { "epoch": 0.05, "learning_rate": 4.974759988904361e-05, "loss": 1.3997, "step": 45365 }, { "epoch": 0.05, "learning_rate": 4.974758875561557e-05, "loss": 1.1154, "step": 45366 }, { "epoch": 0.05, "learning_rate": 4.9747577621943233e-05, "loss": 1.3388, "step": 45367 }, { "epoch": 0.05, "learning_rate": 4.97475664880266e-05, "loss": 0.9506, "step": 45368 }, { "epoch": 0.05, "learning_rate": 4.9747555353865674e-05, "loss": 0.6001, "step": 45369 }, { "epoch": 0.05, "learning_rate": 4.9747544219460454e-05, "loss": 0.9743, "step": 45370 }, { "epoch": 0.05, "learning_rate": 4.9747533084810923e-05, "loss": 0.9414, "step": 45371 }, { "epoch": 0.05, "learning_rate": 4.974752194991711e-05, "loss": 1.1604, "step": 45372 }, { "epoch": 0.05, "learning_rate": 4.974751081477899e-05, "loss": 1.0663, "step": 45373 }, { "epoch": 0.05, "learning_rate": 4.974749967939658e-05, "loss": 0.7129, "step": 45374 }, { "epoch": 0.05, "learning_rate": 4.974748854376987e-05, "loss": 1.0883, "step": 45375 }, { "epoch": 0.05, "learning_rate": 4.9747477407898854e-05, "loss": 0.5705, "step": 45376 }, { "epoch": 0.05, "learning_rate": 4.974746627178355e-05, "loss": 1.2442, "step": 45377 }, { "epoch": 0.05, "learning_rate": 4.974745513542395e-05, "loss": 1.5215, "step": 45378 }, { "epoch": 0.05, "learning_rate": 4.9747443998820054e-05, "loss": 1.3067, "step": 45379 }, { "epoch": 0.05, "learning_rate": 4.974743286197186e-05, "loss": 1.1963, "step": 45380 }, { "epoch": 0.05, "learning_rate": 4.974742172487937e-05, "loss": 1.1076, "step": 45381 }, { "epoch": 0.05, "learning_rate": 4.9747410587542586e-05, "loss": 1.143, "step": 45382 }, { "epoch": 0.05, "learning_rate": 4.97473994499615e-05, "loss": 1.0724, "step": 45383 }, { "epoch": 0.05, "learning_rate": 4.9747388312136125e-05, "loss": 0.8348, "step": 45384 }, { "epoch": 0.05, "learning_rate": 4.974737717406645e-05, "loss": 0.9344, "step": 45385 }, { "epoch": 0.05, "learning_rate": 4.974736603575248e-05, "loss": 1.1494, "step": 45386 }, { "epoch": 0.05, "learning_rate": 4.9747354897194216e-05, "loss": 0.9777, "step": 45387 }, { "epoch": 0.05, "learning_rate": 4.974734375839165e-05, "loss": 1.0619, "step": 45388 }, { "epoch": 0.05, "learning_rate": 4.97473326193448e-05, "loss": 0.9086, "step": 45389 }, { "epoch": 0.05, "learning_rate": 4.974732148005364e-05, "loss": 0.9629, "step": 45390 }, { "epoch": 0.05, "learning_rate": 4.97473103405182e-05, "loss": 1.158, "step": 45391 }, { "epoch": 0.05, "learning_rate": 4.974729920073846e-05, "loss": 1.0624, "step": 45392 }, { "epoch": 0.05, "learning_rate": 4.9747288060714425e-05, "loss": 0.8178, "step": 45393 }, { "epoch": 0.05, "learning_rate": 4.97472769204461e-05, "loss": 0.8235, "step": 45394 }, { "epoch": 0.05, "learning_rate": 4.9747265779933465e-05, "loss": 1.2565, "step": 45395 }, { "epoch": 0.05, "learning_rate": 4.974725463917654e-05, "loss": 1.0901, "step": 45396 }, { "epoch": 0.05, "learning_rate": 4.9747243498175334e-05, "loss": 0.9782, "step": 45397 }, { "epoch": 0.05, "learning_rate": 4.9747232356929816e-05, "loss": 0.975, "step": 45398 }, { "epoch": 0.05, "learning_rate": 4.974722121544002e-05, "loss": 1.2183, "step": 45399 }, { "epoch": 0.05, "learning_rate": 4.974721007370592e-05, "loss": 1.0256, "step": 45400 }, { "epoch": 0.05, "learning_rate": 4.9747198931727526e-05, "loss": 0.9213, "step": 45401 }, { "epoch": 0.05, "learning_rate": 4.974718778950484e-05, "loss": 0.5183, "step": 45402 }, { "epoch": 0.05, "learning_rate": 4.974717664703786e-05, "loss": 0.5569, "step": 45403 }, { "epoch": 0.05, "learning_rate": 4.974716550432659e-05, "loss": 0.4812, "step": 45404 }, { "epoch": 0.05, "learning_rate": 4.9747154361371026e-05, "loss": 0.4368, "step": 45405 }, { "epoch": 0.05, "learning_rate": 4.9747143218171165e-05, "loss": 1.1171, "step": 45406 }, { "epoch": 0.05, "learning_rate": 4.974713207472701e-05, "loss": 1.2322, "step": 45407 }, { "epoch": 0.05, "learning_rate": 4.974712093103856e-05, "loss": 0.7163, "step": 45408 }, { "epoch": 0.05, "learning_rate": 4.974710978710583e-05, "loss": 1.0936, "step": 45409 }, { "epoch": 0.05, "learning_rate": 4.974709864292879e-05, "loss": 1.438, "step": 45410 }, { "epoch": 0.05, "learning_rate": 4.9747087498507466e-05, "loss": 1.1511, "step": 45411 }, { "epoch": 0.05, "learning_rate": 4.9747076353841845e-05, "loss": 1.3857, "step": 45412 }, { "epoch": 0.05, "learning_rate": 4.974706520893194e-05, "loss": 1.241, "step": 45413 }, { "epoch": 0.05, "learning_rate": 4.974705406377773e-05, "loss": 1.1779, "step": 45414 }, { "epoch": 0.05, "learning_rate": 4.9747042918379235e-05, "loss": 0.9653, "step": 45415 }, { "epoch": 0.05, "learning_rate": 4.974703177273645e-05, "loss": 1.1707, "step": 45416 }, { "epoch": 0.05, "learning_rate": 4.9747020626849366e-05, "loss": 1.0022, "step": 45417 }, { "epoch": 0.05, "learning_rate": 4.974700948071799e-05, "loss": 0.8391, "step": 45418 }, { "epoch": 0.05, "learning_rate": 4.9746998334342333e-05, "loss": 0.6924, "step": 45419 }, { "epoch": 0.05, "learning_rate": 4.974698718772237e-05, "loss": 0.9913, "step": 45420 }, { "epoch": 0.05, "learning_rate": 4.974697604085812e-05, "loss": 1.0964, "step": 45421 }, { "epoch": 0.05, "learning_rate": 4.9746964893749584e-05, "loss": 0.8411, "step": 45422 }, { "epoch": 0.05, "learning_rate": 4.9746953746396754e-05, "loss": 0.8076, "step": 45423 }, { "epoch": 0.05, "learning_rate": 4.9746942598799626e-05, "loss": 1.1273, "step": 45424 }, { "epoch": 0.05, "learning_rate": 4.9746931450958215e-05, "loss": 1.0856, "step": 45425 }, { "epoch": 0.05, "learning_rate": 4.9746920302872505e-05, "loss": 1.2764, "step": 45426 }, { "epoch": 0.05, "learning_rate": 4.974690915454251e-05, "loss": 1.4223, "step": 45427 }, { "epoch": 0.05, "learning_rate": 4.974689800596822e-05, "loss": 1.3855, "step": 45428 }, { "epoch": 0.05, "learning_rate": 4.9746886857149636e-05, "loss": 1.0838, "step": 45429 }, { "epoch": 0.05, "learning_rate": 4.974687570808677e-05, "loss": 0.8186, "step": 45430 }, { "epoch": 0.05, "learning_rate": 4.97468645587796e-05, "loss": 0.9951, "step": 45431 }, { "epoch": 0.05, "learning_rate": 4.9746853409228154e-05, "loss": 1.2055, "step": 45432 }, { "epoch": 0.05, "learning_rate": 4.974684225943241e-05, "loss": 1.0613, "step": 45433 }, { "epoch": 0.05, "learning_rate": 4.9746831109392375e-05, "loss": 1.0256, "step": 45434 }, { "epoch": 0.05, "learning_rate": 4.974681995910805e-05, "loss": 1.0921, "step": 45435 }, { "epoch": 0.05, "learning_rate": 4.9746808808579445e-05, "loss": 1.1969, "step": 45436 }, { "epoch": 0.05, "learning_rate": 4.974679765780654e-05, "loss": 0.9368, "step": 45437 }, { "epoch": 0.05, "learning_rate": 4.9746786506789344e-05, "loss": 0.881, "step": 45438 }, { "epoch": 0.05, "learning_rate": 4.9746775355527864e-05, "loss": 1.2086, "step": 45439 }, { "epoch": 0.05, "learning_rate": 4.9746764204022086e-05, "loss": 0.9978, "step": 45440 }, { "epoch": 0.05, "learning_rate": 4.974675305227203e-05, "loss": 0.7796, "step": 45441 }, { "epoch": 0.05, "learning_rate": 4.974674190027767e-05, "loss": 1.2899, "step": 45442 }, { "epoch": 0.05, "learning_rate": 4.974673074803903e-05, "loss": 1.0924, "step": 45443 }, { "epoch": 0.05, "learning_rate": 4.9746719595556104e-05, "loss": 1.132, "step": 45444 }, { "epoch": 0.05, "learning_rate": 4.9746708442828884e-05, "loss": 0.815, "step": 45445 }, { "epoch": 0.05, "learning_rate": 4.9746697289857366e-05, "loss": 1.2364, "step": 45446 }, { "epoch": 0.05, "learning_rate": 4.974668613664157e-05, "loss": 1.0115, "step": 45447 }, { "epoch": 0.05, "learning_rate": 4.9746674983181484e-05, "loss": 1.1063, "step": 45448 }, { "epoch": 0.05, "learning_rate": 4.9746663829477113e-05, "loss": 1.071, "step": 45449 }, { "epoch": 0.05, "learning_rate": 4.974665267552845e-05, "loss": 2.2968, "step": 45450 }, { "epoch": 0.05, "learning_rate": 4.97466415213355e-05, "loss": 1.4211, "step": 45451 }, { "epoch": 0.05, "learning_rate": 4.974663036689825e-05, "loss": 0.9691, "step": 45452 }, { "epoch": 0.05, "learning_rate": 4.974661921221673e-05, "loss": 1.0364, "step": 45453 }, { "epoch": 0.05, "learning_rate": 4.9746608057290914e-05, "loss": 1.0775, "step": 45454 }, { "epoch": 0.05, "learning_rate": 4.9746596902120804e-05, "loss": 1.0891, "step": 45455 }, { "epoch": 0.05, "learning_rate": 4.974658574670641e-05, "loss": 1.1017, "step": 45456 }, { "epoch": 0.05, "learning_rate": 4.974657459104772e-05, "loss": 0.9819, "step": 45457 }, { "epoch": 0.05, "learning_rate": 4.974656343514475e-05, "loss": 1.0283, "step": 45458 }, { "epoch": 0.05, "learning_rate": 4.97465522789975e-05, "loss": 1.2659, "step": 45459 }, { "epoch": 0.05, "learning_rate": 4.974654112260595e-05, "loss": 1.8127, "step": 45460 }, { "epoch": 0.05, "learning_rate": 4.9746529965970124e-05, "loss": 1.1422, "step": 45461 }, { "epoch": 0.05, "learning_rate": 4.974651880909e-05, "loss": 1.1374, "step": 45462 }, { "epoch": 0.05, "learning_rate": 4.97465076519656e-05, "loss": 0.5853, "step": 45463 }, { "epoch": 0.05, "learning_rate": 4.974649649459691e-05, "loss": 1.1149, "step": 45464 }, { "epoch": 0.05, "learning_rate": 4.974648533698392e-05, "loss": 0.8973, "step": 45465 }, { "epoch": 0.05, "learning_rate": 4.9746474179126656e-05, "loss": 0.6577, "step": 45466 }, { "epoch": 0.05, "learning_rate": 4.97464630210251e-05, "loss": 0.7221, "step": 45467 }, { "epoch": 0.05, "learning_rate": 4.974645186267926e-05, "loss": 0.5604, "step": 45468 }, { "epoch": 0.05, "learning_rate": 4.9746440704089136e-05, "loss": 0.9888, "step": 45469 }, { "epoch": 0.05, "learning_rate": 4.974642954525472e-05, "loss": 1.0218, "step": 45470 }, { "epoch": 0.05, "learning_rate": 4.974641838617602e-05, "loss": 1.2538, "step": 45471 }, { "epoch": 0.05, "learning_rate": 4.974640722685303e-05, "loss": 1.086, "step": 45472 }, { "epoch": 0.05, "learning_rate": 4.974639606728576e-05, "loss": 0.8737, "step": 45473 }, { "epoch": 0.05, "learning_rate": 4.97463849074742e-05, "loss": 1.0922, "step": 45474 }, { "epoch": 0.05, "learning_rate": 4.974637374741836e-05, "loss": 0.7719, "step": 45475 }, { "epoch": 0.05, "learning_rate": 4.974636258711823e-05, "loss": 1.131, "step": 45476 }, { "epoch": 0.05, "learning_rate": 4.974635142657381e-05, "loss": 0.8914, "step": 45477 }, { "epoch": 0.05, "learning_rate": 4.9746340265785106e-05, "loss": 1.2749, "step": 45478 }, { "epoch": 0.05, "learning_rate": 4.974632910475212e-05, "loss": 1.6377, "step": 45479 }, { "epoch": 0.05, "learning_rate": 4.974631794347485e-05, "loss": 1.2082, "step": 45480 }, { "epoch": 0.05, "learning_rate": 4.9746306781953297e-05, "loss": 1.3005, "step": 45481 }, { "epoch": 0.05, "learning_rate": 4.974629562018745e-05, "loss": 1.0055, "step": 45482 }, { "epoch": 0.05, "learning_rate": 4.9746284458177324e-05, "loss": 1.1625, "step": 45483 }, { "epoch": 0.05, "learning_rate": 4.974627329592291e-05, "loss": 0.9574, "step": 45484 }, { "epoch": 0.05, "learning_rate": 4.9746262133424214e-05, "loss": 1.1081, "step": 45485 }, { "epoch": 0.05, "learning_rate": 4.974625097068123e-05, "loss": 0.9516, "step": 45486 }, { "epoch": 0.05, "learning_rate": 4.974623980769396e-05, "loss": 1.1537, "step": 45487 }, { "epoch": 0.05, "learning_rate": 4.9746228644462414e-05, "loss": 0.9005, "step": 45488 }, { "epoch": 0.05, "learning_rate": 4.9746217480986576e-05, "loss": 0.7145, "step": 45489 }, { "epoch": 0.05, "learning_rate": 4.974620631726646e-05, "loss": 0.5629, "step": 45490 }, { "epoch": 0.05, "learning_rate": 4.9746195153302055e-05, "loss": 1.2446, "step": 45491 }, { "epoch": 0.05, "learning_rate": 4.974618398909337e-05, "loss": 2.3399, "step": 45492 }, { "epoch": 0.05, "learning_rate": 4.9746172824640396e-05, "loss": 1.0465, "step": 45493 }, { "epoch": 0.05, "learning_rate": 4.9746161659943144e-05, "loss": 0.9775, "step": 45494 }, { "epoch": 0.05, "learning_rate": 4.97461504950016e-05, "loss": 1.1528, "step": 45495 }, { "epoch": 0.05, "learning_rate": 4.974613932981578e-05, "loss": 1.2708, "step": 45496 }, { "epoch": 0.05, "learning_rate": 4.9746128164385674e-05, "loss": 0.932, "step": 45497 }, { "epoch": 0.05, "learning_rate": 4.974611699871129e-05, "loss": 0.9144, "step": 45498 }, { "epoch": 0.05, "learning_rate": 4.974610583279261e-05, "loss": 1.1509, "step": 45499 }, { "epoch": 0.05, "learning_rate": 4.974609466662966e-05, "loss": 0.7826, "step": 45500 }, { "epoch": 0.05, "eval_loss": 1.0797648429870605, "eval_runtime": 125.7599, "eval_samples_per_second": 11.013, "eval_steps_per_second": 5.511, "step": 45500 }, { "epoch": 0.05, "learning_rate": 4.974608350022242e-05, "loss": 1.0085, "step": 45501 }, { "epoch": 0.05, "learning_rate": 4.97460723335709e-05, "loss": 1.4251, "step": 45502 }, { "epoch": 0.05, "learning_rate": 4.974606116667509e-05, "loss": 0.9914, "step": 45503 }, { "epoch": 0.05, "learning_rate": 4.9746049999535004e-05, "loss": 0.9361, "step": 45504 }, { "epoch": 0.05, "learning_rate": 4.974603883215063e-05, "loss": 1.0632, "step": 45505 }, { "epoch": 0.05, "learning_rate": 4.974602766452198e-05, "loss": 1.332, "step": 45506 }, { "epoch": 0.05, "learning_rate": 4.974601649664905e-05, "loss": 1.2818, "step": 45507 }, { "epoch": 0.05, "learning_rate": 4.9746005328531833e-05, "loss": 1.3326, "step": 45508 }, { "epoch": 0.05, "learning_rate": 4.974599416017034e-05, "loss": 0.7155, "step": 45509 }, { "epoch": 0.05, "learning_rate": 4.9745982991564556e-05, "loss": 0.469, "step": 45510 }, { "epoch": 0.05, "learning_rate": 4.974597182271449e-05, "loss": 0.7329, "step": 45511 }, { "epoch": 0.05, "learning_rate": 4.974596065362015e-05, "loss": 0.4191, "step": 45512 }, { "epoch": 0.05, "learning_rate": 4.9745949484281514e-05, "loss": 0.6019, "step": 45513 }, { "epoch": 0.05, "learning_rate": 4.9745938314698617e-05, "loss": 0.7599, "step": 45514 }, { "epoch": 0.05, "learning_rate": 4.974592714487142e-05, "loss": 0.3362, "step": 45515 }, { "epoch": 0.05, "learning_rate": 4.974591597479995e-05, "loss": 0.3221, "step": 45516 }, { "epoch": 0.05, "learning_rate": 4.97459048044842e-05, "loss": 0.415, "step": 45517 }, { "epoch": 0.05, "learning_rate": 4.974589363392417e-05, "loss": 0.469, "step": 45518 }, { "epoch": 0.05, "learning_rate": 4.974588246311985e-05, "loss": 0.2685, "step": 45519 }, { "epoch": 0.05, "learning_rate": 4.974587129207125e-05, "loss": 0.6892, "step": 45520 }, { "epoch": 0.05, "learning_rate": 4.974586012077838e-05, "loss": 0.5863, "step": 45521 }, { "epoch": 0.05, "learning_rate": 4.974584894924123e-05, "loss": 0.5748, "step": 45522 }, { "epoch": 0.05, "learning_rate": 4.9745837777459784e-05, "loss": 0.6229, "step": 45523 }, { "epoch": 0.05, "learning_rate": 4.974582660543407e-05, "loss": 0.6044, "step": 45524 }, { "epoch": 0.05, "learning_rate": 4.974581543316407e-05, "loss": 1.0355, "step": 45525 }, { "epoch": 0.05, "learning_rate": 4.974580426064979e-05, "loss": 0.5054, "step": 45526 }, { "epoch": 0.05, "learning_rate": 4.974579308789124e-05, "loss": 0.8264, "step": 45527 }, { "epoch": 0.05, "learning_rate": 4.974578191488839e-05, "loss": 0.9898, "step": 45528 }, { "epoch": 0.05, "learning_rate": 4.974577074164127e-05, "loss": 1.4393, "step": 45529 }, { "epoch": 0.05, "learning_rate": 4.974575956814988e-05, "loss": 1.0134, "step": 45530 }, { "epoch": 0.05, "learning_rate": 4.97457483944142e-05, "loss": 0.8625, "step": 45531 }, { "epoch": 0.05, "learning_rate": 4.974573722043424e-05, "loss": 0.9337, "step": 45532 }, { "epoch": 0.05, "learning_rate": 4.974572604621001e-05, "loss": 1.1906, "step": 45533 }, { "epoch": 0.05, "learning_rate": 4.9745714871741487e-05, "loss": 0.8693, "step": 45534 }, { "epoch": 0.05, "learning_rate": 4.974570369702869e-05, "loss": 0.8957, "step": 45535 }, { "epoch": 0.05, "learning_rate": 4.974569252207162e-05, "loss": 0.745, "step": 45536 }, { "epoch": 0.05, "learning_rate": 4.9745681346870267e-05, "loss": 0.9364, "step": 45537 }, { "epoch": 0.05, "learning_rate": 4.974567017142463e-05, "loss": 1.059, "step": 45538 }, { "epoch": 0.05, "learning_rate": 4.974565899573472e-05, "loss": 0.6451, "step": 45539 }, { "epoch": 0.05, "learning_rate": 4.9745647819800535e-05, "loss": 0.7955, "step": 45540 }, { "epoch": 0.05, "learning_rate": 4.974563664362206e-05, "loss": 1.2032, "step": 45541 }, { "epoch": 0.05, "learning_rate": 4.974562546719931e-05, "loss": 1.0628, "step": 45542 }, { "epoch": 0.05, "learning_rate": 4.9745614290532286e-05, "loss": 1.0687, "step": 45543 }, { "epoch": 0.05, "learning_rate": 4.974560311362099e-05, "loss": 1.0418, "step": 45544 }, { "epoch": 0.05, "learning_rate": 4.9745591936465405e-05, "loss": 1.0584, "step": 45545 }, { "epoch": 0.05, "learning_rate": 4.974558075906554e-05, "loss": 0.7742, "step": 45546 }, { "epoch": 0.05, "learning_rate": 4.97455695814214e-05, "loss": 1.0509, "step": 45547 }, { "epoch": 0.05, "learning_rate": 4.9745558403532986e-05, "loss": 0.4639, "step": 45548 }, { "epoch": 0.05, "learning_rate": 4.9745547225400294e-05, "loss": 0.3128, "step": 45549 }, { "epoch": 0.05, "learning_rate": 4.9745536047023324e-05, "loss": 0.4887, "step": 45550 }, { "epoch": 0.05, "learning_rate": 4.974552486840207e-05, "loss": 0.4655, "step": 45551 }, { "epoch": 0.05, "learning_rate": 4.974551368953655e-05, "loss": 0.4807, "step": 45552 }, { "epoch": 0.05, "learning_rate": 4.974550251042674e-05, "loss": 0.4912, "step": 45553 }, { "epoch": 0.05, "learning_rate": 4.9745491331072655e-05, "loss": 0.4908, "step": 45554 }, { "epoch": 0.05, "learning_rate": 4.97454801514743e-05, "loss": 0.7878, "step": 45555 }, { "epoch": 0.05, "learning_rate": 4.974546897163167e-05, "loss": 1.1363, "step": 45556 }, { "epoch": 0.05, "learning_rate": 4.9745457791544756e-05, "loss": 1.1923, "step": 45557 }, { "epoch": 0.05, "learning_rate": 4.9745446611213566e-05, "loss": 1.2722, "step": 45558 }, { "epoch": 0.05, "learning_rate": 4.97454354306381e-05, "loss": 1.1871, "step": 45559 }, { "epoch": 0.05, "learning_rate": 4.974542424981836e-05, "loss": 1.3742, "step": 45560 }, { "epoch": 0.05, "learning_rate": 4.9745413068754344e-05, "loss": 0.8521, "step": 45561 }, { "epoch": 0.05, "learning_rate": 4.9745401887446044e-05, "loss": 1.0748, "step": 45562 }, { "epoch": 0.05, "learning_rate": 4.974539070589348e-05, "loss": 0.7302, "step": 45563 }, { "epoch": 0.05, "learning_rate": 4.974537952409663e-05, "loss": 0.4895, "step": 45564 }, { "epoch": 0.05, "learning_rate": 4.974536834205551e-05, "loss": 0.7053, "step": 45565 }, { "epoch": 0.05, "learning_rate": 4.9745357159770115e-05, "loss": 0.8277, "step": 45566 }, { "epoch": 0.05, "learning_rate": 4.974534597724044e-05, "loss": 1.0419, "step": 45567 }, { "epoch": 0.05, "learning_rate": 4.974533479446649e-05, "loss": 1.1161, "step": 45568 }, { "epoch": 0.05, "learning_rate": 4.974532361144826e-05, "loss": 0.4913, "step": 45569 }, { "epoch": 0.05, "learning_rate": 4.9745312428185766e-05, "loss": 0.8352, "step": 45570 }, { "epoch": 0.05, "learning_rate": 4.974530124467899e-05, "loss": 0.85, "step": 45571 }, { "epoch": 0.05, "learning_rate": 4.974529006092793e-05, "loss": 1.054, "step": 45572 }, { "epoch": 0.05, "learning_rate": 4.974527887693261e-05, "loss": 1.0815, "step": 45573 }, { "epoch": 0.05, "learning_rate": 4.974526769269301e-05, "loss": 0.8664, "step": 45574 }, { "epoch": 0.05, "learning_rate": 4.974525650820913e-05, "loss": 0.8523, "step": 45575 }, { "epoch": 0.05, "learning_rate": 4.9745245323480984e-05, "loss": 0.9492, "step": 45576 }, { "epoch": 0.05, "learning_rate": 4.9745234138508554e-05, "loss": 1.022, "step": 45577 }, { "epoch": 0.05, "learning_rate": 4.9745222953291854e-05, "loss": 1.1837, "step": 45578 }, { "epoch": 0.05, "learning_rate": 4.974521176783088e-05, "loss": 1.0363, "step": 45579 }, { "epoch": 0.05, "learning_rate": 4.974520058212563e-05, "loss": 1.2925, "step": 45580 }, { "epoch": 0.05, "learning_rate": 4.974518939617611e-05, "loss": 0.9728, "step": 45581 }, { "epoch": 0.05, "learning_rate": 4.974517820998231e-05, "loss": 1.1876, "step": 45582 }, { "epoch": 0.05, "learning_rate": 4.974516702354424e-05, "loss": 0.936, "step": 45583 }, { "epoch": 0.05, "learning_rate": 4.9745155836861894e-05, "loss": 1.1342, "step": 45584 }, { "epoch": 0.05, "learning_rate": 4.974514464993528e-05, "loss": 1.2303, "step": 45585 }, { "epoch": 0.05, "learning_rate": 4.9745133462764385e-05, "loss": 0.8657, "step": 45586 }, { "epoch": 0.05, "learning_rate": 4.974512227534922e-05, "loss": 1.1593, "step": 45587 }, { "epoch": 0.05, "learning_rate": 4.974511108768978e-05, "loss": 1.0674, "step": 45588 }, { "epoch": 0.05, "learning_rate": 4.974509989978607e-05, "loss": 0.9926, "step": 45589 }, { "epoch": 0.05, "learning_rate": 4.9745088711638085e-05, "loss": 1.0086, "step": 45590 }, { "epoch": 0.05, "learning_rate": 4.974507752324582e-05, "loss": 1.1706, "step": 45591 }, { "epoch": 0.05, "learning_rate": 4.974506633460929e-05, "loss": 1.0246, "step": 45592 }, { "epoch": 0.05, "learning_rate": 4.9745055145728484e-05, "loss": 1.2272, "step": 45593 }, { "epoch": 0.05, "learning_rate": 4.974504395660341e-05, "loss": 1.1084, "step": 45594 }, { "epoch": 0.05, "learning_rate": 4.9745032767234054e-05, "loss": 1.2015, "step": 45595 }, { "epoch": 0.05, "learning_rate": 4.974502157762044e-05, "loss": 1.1233, "step": 45596 }, { "epoch": 0.05, "learning_rate": 4.974501038776254e-05, "loss": 0.6892, "step": 45597 }, { "epoch": 0.05, "learning_rate": 4.974499919766037e-05, "loss": 1.2991, "step": 45598 }, { "epoch": 0.05, "learning_rate": 4.974498800731394e-05, "loss": 1.1437, "step": 45599 }, { "epoch": 0.05, "learning_rate": 4.974497681672322e-05, "loss": 1.2367, "step": 45600 }, { "epoch": 0.05, "learning_rate": 4.974496562588824e-05, "loss": 1.0962, "step": 45601 }, { "epoch": 0.05, "learning_rate": 4.974495443480899e-05, "loss": 1.117, "step": 45602 }, { "epoch": 0.05, "learning_rate": 4.974494324348546e-05, "loss": 0.994, "step": 45603 }, { "epoch": 0.05, "learning_rate": 4.974493205191766e-05, "loss": 0.9934, "step": 45604 }, { "epoch": 0.05, "learning_rate": 4.9744920860105586e-05, "loss": 1.1071, "step": 45605 }, { "epoch": 0.05, "learning_rate": 4.9744909668049244e-05, "loss": 0.8971, "step": 45606 }, { "epoch": 0.05, "learning_rate": 4.974489847574864e-05, "loss": 1.0586, "step": 45607 }, { "epoch": 0.05, "learning_rate": 4.974488728320375e-05, "loss": 1.007, "step": 45608 }, { "epoch": 0.05, "learning_rate": 4.9744876090414594e-05, "loss": 0.9642, "step": 45609 }, { "epoch": 0.05, "learning_rate": 4.974486489738117e-05, "loss": 1.1053, "step": 45610 }, { "epoch": 0.05, "learning_rate": 4.9744853704103466e-05, "loss": 1.4651, "step": 45611 }, { "epoch": 0.05, "learning_rate": 4.97448425105815e-05, "loss": 1.2437, "step": 45612 }, { "epoch": 0.05, "learning_rate": 4.974483131681527e-05, "loss": 0.9803, "step": 45613 }, { "epoch": 0.05, "learning_rate": 4.974482012280476e-05, "loss": 0.9789, "step": 45614 }, { "epoch": 0.05, "learning_rate": 4.9744808928549975e-05, "loss": 0.6698, "step": 45615 }, { "epoch": 0.05, "learning_rate": 4.9744797734050926e-05, "loss": 0.907, "step": 45616 }, { "epoch": 0.05, "learning_rate": 4.9744786539307606e-05, "loss": 0.8371, "step": 45617 }, { "epoch": 0.05, "learning_rate": 4.9744775344320015e-05, "loss": 1.2622, "step": 45618 }, { "epoch": 0.05, "learning_rate": 4.974476414908815e-05, "loss": 1.0646, "step": 45619 }, { "epoch": 0.05, "learning_rate": 4.974475295361202e-05, "loss": 1.1624, "step": 45620 }, { "epoch": 0.05, "learning_rate": 4.9744741757891624e-05, "loss": 1.0207, "step": 45621 }, { "epoch": 0.05, "learning_rate": 4.9744730561926957e-05, "loss": 1.0858, "step": 45622 }, { "epoch": 0.05, "learning_rate": 4.974471936571802e-05, "loss": 1.0029, "step": 45623 }, { "epoch": 0.05, "learning_rate": 4.974470816926481e-05, "loss": 1.2861, "step": 45624 }, { "epoch": 0.05, "learning_rate": 4.974469697256733e-05, "loss": 1.1925, "step": 45625 }, { "epoch": 0.05, "learning_rate": 4.974468577562558e-05, "loss": 0.9977, "step": 45626 }, { "epoch": 0.05, "learning_rate": 4.9744674578439566e-05, "loss": 1.1006, "step": 45627 }, { "epoch": 0.05, "learning_rate": 4.974466338100928e-05, "loss": 1.2871, "step": 45628 }, { "epoch": 0.05, "learning_rate": 4.9744652183334725e-05, "loss": 1.0092, "step": 45629 }, { "epoch": 0.05, "learning_rate": 4.9744640985415906e-05, "loss": 1.5077, "step": 45630 }, { "epoch": 0.05, "learning_rate": 4.974462978725281e-05, "loss": 1.2329, "step": 45631 }, { "epoch": 0.05, "learning_rate": 4.974461858884545e-05, "loss": 0.6615, "step": 45632 }, { "epoch": 0.05, "learning_rate": 4.974460739019382e-05, "loss": 0.5246, "step": 45633 }, { "epoch": 0.05, "learning_rate": 4.974459619129792e-05, "loss": 0.8487, "step": 45634 }, { "epoch": 0.05, "learning_rate": 4.974458499215776e-05, "loss": 1.0592, "step": 45635 }, { "epoch": 0.05, "learning_rate": 4.974457379277332e-05, "loss": 1.0532, "step": 45636 }, { "epoch": 0.05, "learning_rate": 4.974456259314463e-05, "loss": 0.8353, "step": 45637 }, { "epoch": 0.05, "learning_rate": 4.9744551393271655e-05, "loss": 1.12, "step": 45638 }, { "epoch": 0.05, "learning_rate": 4.974454019315442e-05, "loss": 1.2937, "step": 45639 }, { "epoch": 0.05, "learning_rate": 4.974452899279292e-05, "loss": 1.1838, "step": 45640 }, { "epoch": 0.05, "learning_rate": 4.974451779218714e-05, "loss": 0.7565, "step": 45641 }, { "epoch": 0.05, "learning_rate": 4.974450659133711e-05, "loss": 0.9144, "step": 45642 }, { "epoch": 0.05, "learning_rate": 4.9744495390242794e-05, "loss": 0.8899, "step": 45643 }, { "epoch": 0.05, "learning_rate": 4.974448418890422e-05, "loss": 1.1573, "step": 45644 }, { "epoch": 0.05, "learning_rate": 4.974447298732138e-05, "loss": 1.2918, "step": 45645 }, { "epoch": 0.05, "learning_rate": 4.974446178549427e-05, "loss": 1.0351, "step": 45646 }, { "epoch": 0.05, "learning_rate": 4.97444505834229e-05, "loss": 1.2496, "step": 45647 }, { "epoch": 0.05, "learning_rate": 4.974443938110725e-05, "loss": 1.53, "step": 45648 }, { "epoch": 0.05, "learning_rate": 4.974442817854734e-05, "loss": 1.176, "step": 45649 }, { "epoch": 0.05, "learning_rate": 4.974441697574317e-05, "loss": 1.1114, "step": 45650 }, { "epoch": 0.05, "learning_rate": 4.9744405772694725e-05, "loss": 0.9394, "step": 45651 }, { "epoch": 0.05, "learning_rate": 4.9744394569402016e-05, "loss": 1.0122, "step": 45652 }, { "epoch": 0.05, "learning_rate": 4.9744383365865044e-05, "loss": 1.4037, "step": 45653 }, { "epoch": 0.05, "learning_rate": 4.97443721620838e-05, "loss": 1.4254, "step": 45654 }, { "epoch": 0.05, "learning_rate": 4.97443609580583e-05, "loss": 1.6696, "step": 45655 }, { "epoch": 0.05, "learning_rate": 4.974434975378852e-05, "loss": 0.8121, "step": 45656 }, { "epoch": 0.05, "learning_rate": 4.974433854927448e-05, "loss": 0.3647, "step": 45657 }, { "epoch": 0.05, "learning_rate": 4.974432734451618e-05, "loss": 0.8786, "step": 45658 }, { "epoch": 0.05, "learning_rate": 4.974431613951361e-05, "loss": 1.0896, "step": 45659 }, { "epoch": 0.05, "learning_rate": 4.974430493426678e-05, "loss": 1.1519, "step": 45660 }, { "epoch": 0.05, "learning_rate": 4.9744293728775674e-05, "loss": 0.756, "step": 45661 }, { "epoch": 0.05, "learning_rate": 4.9744282523040305e-05, "loss": 0.5171, "step": 45662 }, { "epoch": 0.05, "learning_rate": 4.974427131706068e-05, "loss": 0.8075, "step": 45663 }, { "epoch": 0.05, "learning_rate": 4.974426011083678e-05, "loss": 0.9396, "step": 45664 }, { "epoch": 0.05, "learning_rate": 4.974424890436862e-05, "loss": 1.2193, "step": 45665 }, { "epoch": 0.05, "learning_rate": 4.97442376976562e-05, "loss": 0.8489, "step": 45666 }, { "epoch": 0.05, "learning_rate": 4.974422649069951e-05, "loss": 0.8109, "step": 45667 }, { "epoch": 0.05, "learning_rate": 4.974421528349855e-05, "loss": 0.864, "step": 45668 }, { "epoch": 0.05, "learning_rate": 4.9744204076053335e-05, "loss": 0.9593, "step": 45669 }, { "epoch": 0.05, "learning_rate": 4.974419286836385e-05, "loss": 0.9096, "step": 45670 }, { "epoch": 0.05, "learning_rate": 4.9744181660430104e-05, "loss": 1.0249, "step": 45671 }, { "epoch": 0.05, "learning_rate": 4.9744170452252095e-05, "loss": 1.2112, "step": 45672 }, { "epoch": 0.05, "learning_rate": 4.9744159243829816e-05, "loss": 0.9519, "step": 45673 }, { "epoch": 0.05, "learning_rate": 4.9744148035163273e-05, "loss": 1.105, "step": 45674 }, { "epoch": 0.05, "learning_rate": 4.974413682625247e-05, "loss": 1.1696, "step": 45675 }, { "epoch": 0.05, "learning_rate": 4.97441256170974e-05, "loss": 0.9728, "step": 45676 }, { "epoch": 0.05, "learning_rate": 4.9744114407698074e-05, "loss": 0.693, "step": 45677 }, { "epoch": 0.05, "learning_rate": 4.974410319805448e-05, "loss": 1.1799, "step": 45678 }, { "epoch": 0.05, "learning_rate": 4.974409198816662e-05, "loss": 1.1697, "step": 45679 }, { "epoch": 0.05, "learning_rate": 4.97440807780345e-05, "loss": 0.8715, "step": 45680 }, { "epoch": 0.05, "learning_rate": 4.9744069567658116e-05, "loss": 1.1345, "step": 45681 }, { "epoch": 0.05, "learning_rate": 4.974405835703747e-05, "loss": 1.12, "step": 45682 }, { "epoch": 0.05, "learning_rate": 4.9744047146172556e-05, "loss": 0.9811, "step": 45683 }, { "epoch": 0.05, "learning_rate": 4.9744035935063387e-05, "loss": 0.9339, "step": 45684 }, { "epoch": 0.05, "learning_rate": 4.974402472370995e-05, "loss": 0.7979, "step": 45685 }, { "epoch": 0.05, "learning_rate": 4.974401351211225e-05, "loss": 0.9271, "step": 45686 }, { "epoch": 0.05, "learning_rate": 4.9744002300270295e-05, "loss": 0.8029, "step": 45687 }, { "epoch": 0.05, "learning_rate": 4.974399108818407e-05, "loss": 1.3723, "step": 45688 }, { "epoch": 0.05, "learning_rate": 4.974397987585359e-05, "loss": 0.9677, "step": 45689 }, { "epoch": 0.05, "learning_rate": 4.974396866327884e-05, "loss": 0.9619, "step": 45690 }, { "epoch": 0.05, "learning_rate": 4.9743957450459825e-05, "loss": 1.1322, "step": 45691 }, { "epoch": 0.05, "learning_rate": 4.974394623739656e-05, "loss": 1.3438, "step": 45692 }, { "epoch": 0.05, "learning_rate": 4.974393502408902e-05, "loss": 1.3055, "step": 45693 }, { "epoch": 0.05, "learning_rate": 4.974392381053723e-05, "loss": 0.9067, "step": 45694 }, { "epoch": 0.05, "learning_rate": 4.974391259674117e-05, "loss": 0.6146, "step": 45695 }, { "epoch": 0.05, "learning_rate": 4.9743901382700855e-05, "loss": 0.8327, "step": 45696 }, { "epoch": 0.05, "learning_rate": 4.9743890168416275e-05, "loss": 0.9726, "step": 45697 }, { "epoch": 0.05, "learning_rate": 4.974387895388744e-05, "loss": 1.381, "step": 45698 }, { "epoch": 0.05, "learning_rate": 4.9743867739114336e-05, "loss": 0.8859, "step": 45699 }, { "epoch": 0.05, "learning_rate": 4.9743856524096963e-05, "loss": 0.8337, "step": 45700 }, { "epoch": 0.05, "learning_rate": 4.974384530883535e-05, "loss": 1.0273, "step": 45701 }, { "epoch": 0.05, "learning_rate": 4.974383409332946e-05, "loss": 1.1322, "step": 45702 }, { "epoch": 0.05, "learning_rate": 4.974382287757932e-05, "loss": 1.0251, "step": 45703 }, { "epoch": 0.05, "learning_rate": 4.974381166158491e-05, "loss": 1.1836, "step": 45704 }, { "epoch": 0.05, "learning_rate": 4.9743800445346244e-05, "loss": 0.9265, "step": 45705 }, { "epoch": 0.05, "learning_rate": 4.974378922886332e-05, "loss": 1.336, "step": 45706 }, { "epoch": 0.05, "learning_rate": 4.9743778012136135e-05, "loss": 1.3028, "step": 45707 }, { "epoch": 0.05, "learning_rate": 4.974376679516468e-05, "loss": 1.7094, "step": 45708 }, { "epoch": 0.05, "learning_rate": 4.974375557794898e-05, "loss": 1.4221, "step": 45709 }, { "epoch": 0.05, "learning_rate": 4.9743744360489006e-05, "loss": 0.9569, "step": 45710 }, { "epoch": 0.05, "learning_rate": 4.9743733142784784e-05, "loss": 1.1971, "step": 45711 }, { "epoch": 0.05, "learning_rate": 4.97437219248363e-05, "loss": 1.272, "step": 45712 }, { "epoch": 0.05, "learning_rate": 4.9743710706643555e-05, "loss": 1.03, "step": 45713 }, { "epoch": 0.05, "learning_rate": 4.974369948820654e-05, "loss": 0.9234, "step": 45714 }, { "epoch": 0.05, "learning_rate": 4.974368826952528e-05, "loss": 0.6427, "step": 45715 }, { "epoch": 0.05, "learning_rate": 4.9743677050599756e-05, "loss": 0.7138, "step": 45716 }, { "epoch": 0.05, "learning_rate": 4.974366583142998e-05, "loss": 0.7718, "step": 45717 }, { "epoch": 0.05, "learning_rate": 4.9743654612015935e-05, "loss": 1.0889, "step": 45718 }, { "epoch": 0.05, "learning_rate": 4.974364339235763e-05, "loss": 1.1808, "step": 45719 }, { "epoch": 0.05, "learning_rate": 4.974363217245507e-05, "loss": 1.1002, "step": 45720 }, { "epoch": 0.05, "learning_rate": 4.9743620952308244e-05, "loss": 1.1042, "step": 45721 }, { "epoch": 0.05, "learning_rate": 4.974360973191717e-05, "loss": 0.9851, "step": 45722 }, { "epoch": 0.05, "learning_rate": 4.974359851128184e-05, "loss": 1.0143, "step": 45723 }, { "epoch": 0.05, "learning_rate": 4.974358729040224e-05, "loss": 1.089, "step": 45724 }, { "epoch": 0.05, "learning_rate": 4.974357606927839e-05, "loss": 0.8771, "step": 45725 }, { "epoch": 0.05, "learning_rate": 4.974356484791028e-05, "loss": 1.1704, "step": 45726 }, { "epoch": 0.05, "learning_rate": 4.974355362629791e-05, "loss": 0.9515, "step": 45727 }, { "epoch": 0.05, "learning_rate": 4.974354240444128e-05, "loss": 1.2547, "step": 45728 }, { "epoch": 0.05, "learning_rate": 4.97435311823404e-05, "loss": 0.9652, "step": 45729 }, { "epoch": 0.05, "learning_rate": 4.9743519959995254e-05, "loss": 0.7073, "step": 45730 }, { "epoch": 0.05, "learning_rate": 4.9743508737405855e-05, "loss": 1.2819, "step": 45731 }, { "epoch": 0.05, "learning_rate": 4.9743497514572205e-05, "loss": 1.0575, "step": 45732 }, { "epoch": 0.05, "learning_rate": 4.9743486291494284e-05, "loss": 1.2363, "step": 45733 }, { "epoch": 0.05, "learning_rate": 4.9743475068172106e-05, "loss": 0.9336, "step": 45734 }, { "epoch": 0.05, "learning_rate": 4.9743463844605684e-05, "loss": 1.1846, "step": 45735 }, { "epoch": 0.05, "learning_rate": 4.974345262079499e-05, "loss": 1.0551, "step": 45736 }, { "epoch": 0.05, "learning_rate": 4.9743441396740056e-05, "loss": 1.1375, "step": 45737 }, { "epoch": 0.05, "learning_rate": 4.9743430172440855e-05, "loss": 1.0266, "step": 45738 }, { "epoch": 0.05, "learning_rate": 4.97434189478974e-05, "loss": 0.9828, "step": 45739 }, { "epoch": 0.05, "learning_rate": 4.9743407723109684e-05, "loss": 1.1194, "step": 45740 }, { "epoch": 0.05, "learning_rate": 4.974339649807771e-05, "loss": 0.9022, "step": 45741 }, { "epoch": 0.05, "learning_rate": 4.974338527280149e-05, "loss": 0.7958, "step": 45742 }, { "epoch": 0.05, "learning_rate": 4.9743374047281004e-05, "loss": 1.1339, "step": 45743 }, { "epoch": 0.05, "learning_rate": 4.974336282151627e-05, "loss": 1.4615, "step": 45744 }, { "epoch": 0.05, "learning_rate": 4.974335159550727e-05, "loss": 1.382, "step": 45745 }, { "epoch": 0.05, "learning_rate": 4.974334036925402e-05, "loss": 1.0907, "step": 45746 }, { "epoch": 0.05, "learning_rate": 4.9743329142756515e-05, "loss": 1.1728, "step": 45747 }, { "epoch": 0.05, "learning_rate": 4.974331791601476e-05, "loss": 0.973, "step": 45748 }, { "epoch": 0.05, "learning_rate": 4.9743306689028734e-05, "loss": 0.1478, "step": 45749 }, { "epoch": 0.05, "learning_rate": 4.974329546179846e-05, "loss": 0.1046, "step": 45750 }, { "epoch": 0.05, "learning_rate": 4.974328423432393e-05, "loss": 0.0943, "step": 45751 }, { "epoch": 0.05, "learning_rate": 4.974327300660515e-05, "loss": 0.0735, "step": 45752 }, { "epoch": 0.05, "learning_rate": 4.9743261778642106e-05, "loss": 0.0745, "step": 45753 }, { "epoch": 0.05, "learning_rate": 4.974325055043482e-05, "loss": 0.0798, "step": 45754 }, { "epoch": 0.05, "learning_rate": 4.9743239321983266e-05, "loss": 0.0653, "step": 45755 }, { "epoch": 0.05, "learning_rate": 4.9743228093287464e-05, "loss": 0.072, "step": 45756 }, { "epoch": 0.05, "learning_rate": 4.9743216864347404e-05, "loss": 0.1125, "step": 45757 }, { "epoch": 0.05, "learning_rate": 4.9743205635163094e-05, "loss": 0.0526, "step": 45758 }, { "epoch": 0.05, "learning_rate": 4.974319440573452e-05, "loss": 0.0356, "step": 45759 }, { "epoch": 0.05, "learning_rate": 4.97431831760617e-05, "loss": 0.0502, "step": 45760 }, { "epoch": 0.05, "learning_rate": 4.974317194614462e-05, "loss": 0.0739, "step": 45761 }, { "epoch": 0.05, "learning_rate": 4.9743160715983295e-05, "loss": 0.1416, "step": 45762 }, { "epoch": 0.05, "learning_rate": 4.974314948557771e-05, "loss": 0.9306, "step": 45763 }, { "epoch": 0.05, "learning_rate": 4.9743138254927866e-05, "loss": 0.9571, "step": 45764 }, { "epoch": 0.05, "learning_rate": 4.9743127024033776e-05, "loss": 0.7926, "step": 45765 }, { "epoch": 0.05, "learning_rate": 4.9743115792895435e-05, "loss": 1.0578, "step": 45766 }, { "epoch": 0.05, "learning_rate": 4.974310456151283e-05, "loss": 1.1989, "step": 45767 }, { "epoch": 0.05, "learning_rate": 4.9743093329885976e-05, "loss": 1.0844, "step": 45768 }, { "epoch": 0.05, "learning_rate": 4.974308209801487e-05, "loss": 0.8869, "step": 45769 }, { "epoch": 0.05, "learning_rate": 4.974307086589951e-05, "loss": 1.0819, "step": 45770 }, { "epoch": 0.05, "learning_rate": 4.97430596335399e-05, "loss": 0.8452, "step": 45771 }, { "epoch": 0.05, "learning_rate": 4.974304840093603e-05, "loss": 0.7517, "step": 45772 }, { "epoch": 0.05, "learning_rate": 4.974303716808791e-05, "loss": 0.986, "step": 45773 }, { "epoch": 0.05, "learning_rate": 4.974302593499554e-05, "loss": 0.8584, "step": 45774 }, { "epoch": 0.05, "learning_rate": 4.974301470165892e-05, "loss": 1.2329, "step": 45775 }, { "epoch": 0.05, "learning_rate": 4.974300346807804e-05, "loss": 1.4424, "step": 45776 }, { "epoch": 0.05, "learning_rate": 4.9742992234252904e-05, "loss": 1.5056, "step": 45777 }, { "epoch": 0.05, "learning_rate": 4.9742981000183524e-05, "loss": 1.3299, "step": 45778 }, { "epoch": 0.05, "learning_rate": 4.974296976586989e-05, "loss": 0.7363, "step": 45779 }, { "epoch": 0.05, "learning_rate": 4.9742958531312e-05, "loss": 1.261, "step": 45780 }, { "epoch": 0.05, "learning_rate": 4.974294729650987e-05, "loss": 1.5741, "step": 45781 }, { "epoch": 0.05, "learning_rate": 4.9742936061463475e-05, "loss": 1.2628, "step": 45782 }, { "epoch": 0.05, "learning_rate": 4.974292482617283e-05, "loss": 0.8167, "step": 45783 }, { "epoch": 0.05, "learning_rate": 4.974291359063794e-05, "loss": 0.7425, "step": 45784 }, { "epoch": 0.05, "learning_rate": 4.974290235485879e-05, "loss": 1.3187, "step": 45785 }, { "epoch": 0.05, "learning_rate": 4.974289111883539e-05, "loss": 1.02, "step": 45786 }, { "epoch": 0.05, "learning_rate": 4.974287988256775e-05, "loss": 1.0194, "step": 45787 }, { "epoch": 0.05, "learning_rate": 4.974286864605585e-05, "loss": 1.492, "step": 45788 }, { "epoch": 0.05, "learning_rate": 4.974285740929969e-05, "loss": 0.9408, "step": 45789 }, { "epoch": 0.05, "learning_rate": 4.9742846172299295e-05, "loss": 0.6201, "step": 45790 }, { "epoch": 0.05, "learning_rate": 4.974283493505464e-05, "loss": 0.9884, "step": 45791 }, { "epoch": 0.05, "learning_rate": 4.9742823697565734e-05, "loss": 1.1657, "step": 45792 }, { "epoch": 0.05, "learning_rate": 4.974281245983258e-05, "loss": 1.0155, "step": 45793 }, { "epoch": 0.05, "learning_rate": 4.974280122185517e-05, "loss": 1.1231, "step": 45794 }, { "epoch": 0.05, "learning_rate": 4.974278998363352e-05, "loss": 1.0422, "step": 45795 }, { "epoch": 0.05, "learning_rate": 4.9742778745167614e-05, "loss": 1.3371, "step": 45796 }, { "epoch": 0.05, "learning_rate": 4.9742767506457456e-05, "loss": 1.1535, "step": 45797 }, { "epoch": 0.05, "learning_rate": 4.974275626750305e-05, "loss": 0.9949, "step": 45798 }, { "epoch": 0.05, "learning_rate": 4.9742745028304396e-05, "loss": 1.1172, "step": 45799 }, { "epoch": 0.05, "learning_rate": 4.974273378886149e-05, "loss": 0.8733, "step": 45800 }, { "epoch": 0.05, "learning_rate": 4.974272254917433e-05, "loss": 1.3712, "step": 45801 }, { "epoch": 0.05, "learning_rate": 4.9742711309242924e-05, "loss": 1.2341, "step": 45802 }, { "epoch": 0.05, "learning_rate": 4.9742700069067264e-05, "loss": 1.2873, "step": 45803 }, { "epoch": 0.05, "learning_rate": 4.974268882864736e-05, "loss": 1.0733, "step": 45804 }, { "epoch": 0.05, "learning_rate": 4.9742677587983205e-05, "loss": 0.8646, "step": 45805 }, { "epoch": 0.05, "learning_rate": 4.97426663470748e-05, "loss": 0.8734, "step": 45806 }, { "epoch": 0.05, "learning_rate": 4.9742655105922145e-05, "loss": 1.0928, "step": 45807 }, { "epoch": 0.05, "learning_rate": 4.9742643864525246e-05, "loss": 1.2, "step": 45808 }, { "epoch": 0.05, "learning_rate": 4.97426326228841e-05, "loss": 0.4103, "step": 45809 }, { "epoch": 0.05, "learning_rate": 4.97426213809987e-05, "loss": 0.8974, "step": 45810 }, { "epoch": 0.05, "learning_rate": 4.974261013886905e-05, "loss": 0.5257, "step": 45811 }, { "epoch": 0.05, "learning_rate": 4.974259889649515e-05, "loss": 0.6782, "step": 45812 }, { "epoch": 0.05, "learning_rate": 4.9742587653877e-05, "loss": 0.8126, "step": 45813 }, { "epoch": 0.05, "learning_rate": 4.974257641101461e-05, "loss": 0.9808, "step": 45814 }, { "epoch": 0.05, "learning_rate": 4.974256516790796e-05, "loss": 1.3454, "step": 45815 }, { "epoch": 0.05, "learning_rate": 4.9742553924557074e-05, "loss": 1.0497, "step": 45816 }, { "epoch": 0.05, "learning_rate": 4.9742542680961935e-05, "loss": 1.1296, "step": 45817 }, { "epoch": 0.05, "learning_rate": 4.9742531437122545e-05, "loss": 1.9507, "step": 45818 }, { "epoch": 0.05, "learning_rate": 4.974252019303891e-05, "loss": 1.121, "step": 45819 }, { "epoch": 0.05, "learning_rate": 4.974250894871103e-05, "loss": 1.0829, "step": 45820 }, { "epoch": 0.05, "learning_rate": 4.97424977041389e-05, "loss": 0.9769, "step": 45821 }, { "epoch": 0.05, "learning_rate": 4.974248645932252e-05, "loss": 1.2224, "step": 45822 }, { "epoch": 0.05, "learning_rate": 4.9742475214261895e-05, "loss": 1.0138, "step": 45823 }, { "epoch": 0.05, "learning_rate": 4.9742463968957024e-05, "loss": 1.1502, "step": 45824 }, { "epoch": 0.05, "learning_rate": 4.974245272340791e-05, "loss": 0.9323, "step": 45825 }, { "epoch": 0.05, "learning_rate": 4.9742441477614536e-05, "loss": 0.9567, "step": 45826 }, { "epoch": 0.05, "learning_rate": 4.9742430231576926e-05, "loss": 1.0271, "step": 45827 }, { "epoch": 0.05, "learning_rate": 4.9742418985295066e-05, "loss": 0.9642, "step": 45828 }, { "epoch": 0.05, "learning_rate": 4.9742407738768956e-05, "loss": 2.118, "step": 45829 }, { "epoch": 0.05, "learning_rate": 4.97423964919986e-05, "loss": 1.3345, "step": 45830 }, { "epoch": 0.05, "learning_rate": 4.9742385244984004e-05, "loss": 0.8406, "step": 45831 }, { "epoch": 0.05, "learning_rate": 4.9742373997725156e-05, "loss": 1.1666, "step": 45832 }, { "epoch": 0.05, "learning_rate": 4.974236275022206e-05, "loss": 0.8133, "step": 45833 }, { "epoch": 0.05, "learning_rate": 4.974235150247473e-05, "loss": 0.9438, "step": 45834 }, { "epoch": 0.05, "learning_rate": 4.974234025448314e-05, "loss": 0.8918, "step": 45835 }, { "epoch": 0.05, "learning_rate": 4.9742329006247314e-05, "loss": 0.9676, "step": 45836 }, { "epoch": 0.05, "learning_rate": 4.9742317757767235e-05, "loss": 1.1376, "step": 45837 }, { "epoch": 0.05, "learning_rate": 4.9742306509042904e-05, "loss": 1.034, "step": 45838 }, { "epoch": 0.05, "learning_rate": 4.974229526007434e-05, "loss": 0.9938, "step": 45839 }, { "epoch": 0.05, "learning_rate": 4.9742284010861527e-05, "loss": 0.9795, "step": 45840 }, { "epoch": 0.05, "learning_rate": 4.9742272761404465e-05, "loss": 0.8263, "step": 45841 }, { "epoch": 0.05, "learning_rate": 4.974226151170316e-05, "loss": 1.0088, "step": 45842 }, { "epoch": 0.05, "learning_rate": 4.974225026175761e-05, "loss": 0.8927, "step": 45843 }, { "epoch": 0.05, "learning_rate": 4.974223901156781e-05, "loss": 0.9106, "step": 45844 }, { "epoch": 0.05, "learning_rate": 4.974222776113378e-05, "loss": 0.9508, "step": 45845 }, { "epoch": 0.05, "learning_rate": 4.97422165104555e-05, "loss": 0.908, "step": 45846 }, { "epoch": 0.05, "learning_rate": 4.974220525953296e-05, "loss": 1.1274, "step": 45847 }, { "epoch": 0.05, "learning_rate": 4.974219400836619e-05, "loss": 1.0245, "step": 45848 }, { "epoch": 0.05, "learning_rate": 4.974218275695517e-05, "loss": 1.3667, "step": 45849 }, { "epoch": 0.05, "learning_rate": 4.9742171505299904e-05, "loss": 0.874, "step": 45850 }, { "epoch": 0.05, "learning_rate": 4.9742160253400406e-05, "loss": 0.5435, "step": 45851 }, { "epoch": 0.05, "learning_rate": 4.974214900125665e-05, "loss": 0.9578, "step": 45852 }, { "epoch": 0.05, "learning_rate": 4.974213774886865e-05, "loss": 1.045, "step": 45853 }, { "epoch": 0.05, "learning_rate": 4.974212649623642e-05, "loss": 1.081, "step": 45854 }, { "epoch": 0.05, "learning_rate": 4.974211524335993e-05, "loss": 0.9487, "step": 45855 }, { "epoch": 0.05, "learning_rate": 4.974210399023921e-05, "loss": 0.9308, "step": 45856 }, { "epoch": 0.05, "learning_rate": 4.9742092736874236e-05, "loss": 0.6461, "step": 45857 }, { "epoch": 0.05, "learning_rate": 4.974208148326502e-05, "loss": 0.9577, "step": 45858 }, { "epoch": 0.05, "learning_rate": 4.9742070229411566e-05, "loss": 1.1438, "step": 45859 }, { "epoch": 0.05, "learning_rate": 4.974205897531386e-05, "loss": 1.1011, "step": 45860 }, { "epoch": 0.05, "learning_rate": 4.974204772097192e-05, "loss": 1.0869, "step": 45861 }, { "epoch": 0.05, "learning_rate": 4.974203646638573e-05, "loss": 0.9677, "step": 45862 }, { "epoch": 0.05, "learning_rate": 4.9742025211555295e-05, "loss": 1.0718, "step": 45863 }, { "epoch": 0.05, "learning_rate": 4.9742013956480624e-05, "loss": 0.9878, "step": 45864 }, { "epoch": 0.05, "learning_rate": 4.974200270116171e-05, "loss": 1.0713, "step": 45865 }, { "epoch": 0.05, "learning_rate": 4.9741991445598556e-05, "loss": 1.0904, "step": 45866 }, { "epoch": 0.05, "learning_rate": 4.9741980189791147e-05, "loss": 1.0928, "step": 45867 }, { "epoch": 0.05, "learning_rate": 4.9741968933739514e-05, "loss": 1.1967, "step": 45868 }, { "epoch": 0.05, "learning_rate": 4.9741957677443624e-05, "loss": 1.0833, "step": 45869 }, { "epoch": 0.05, "learning_rate": 4.97419464209035e-05, "loss": 1.035, "step": 45870 }, { "epoch": 0.05, "learning_rate": 4.9741935164119127e-05, "loss": 1.2216, "step": 45871 }, { "epoch": 0.05, "learning_rate": 4.974192390709051e-05, "loss": 0.7986, "step": 45872 }, { "epoch": 0.05, "learning_rate": 4.974191264981766e-05, "loss": 1.1399, "step": 45873 }, { "epoch": 0.05, "learning_rate": 4.974190139230057e-05, "loss": 1.1012, "step": 45874 }, { "epoch": 0.05, "learning_rate": 4.974189013453923e-05, "loss": 0.863, "step": 45875 }, { "epoch": 0.05, "learning_rate": 4.9741878876533653e-05, "loss": 0.8511, "step": 45876 }, { "epoch": 0.05, "learning_rate": 4.9741867618283835e-05, "loss": 0.9568, "step": 45877 }, { "epoch": 0.05, "learning_rate": 4.974185635978977e-05, "loss": 0.7713, "step": 45878 }, { "epoch": 0.05, "learning_rate": 4.9741845101051466e-05, "loss": 1.1167, "step": 45879 }, { "epoch": 0.05, "learning_rate": 4.974183384206892e-05, "loss": 0.7915, "step": 45880 }, { "epoch": 0.05, "learning_rate": 4.974182258284214e-05, "loss": 1.1323, "step": 45881 }, { "epoch": 0.05, "learning_rate": 4.974181132337111e-05, "loss": 1.3012, "step": 45882 }, { "epoch": 0.05, "learning_rate": 4.974180006365585e-05, "loss": 0.9903, "step": 45883 }, { "epoch": 0.05, "learning_rate": 4.974178880369634e-05, "loss": 0.7792, "step": 45884 }, { "epoch": 0.05, "learning_rate": 4.9741777543492594e-05, "loss": 0.7715, "step": 45885 }, { "epoch": 0.05, "learning_rate": 4.974176628304461e-05, "loss": 0.8317, "step": 45886 }, { "epoch": 0.05, "learning_rate": 4.9741755022352375e-05, "loss": 0.978, "step": 45887 }, { "epoch": 0.05, "learning_rate": 4.974174376141592e-05, "loss": 0.4814, "step": 45888 }, { "epoch": 0.05, "learning_rate": 4.974173250023521e-05, "loss": 0.6874, "step": 45889 }, { "epoch": 0.05, "learning_rate": 4.9741721238810256e-05, "loss": 1.0783, "step": 45890 }, { "epoch": 0.05, "learning_rate": 4.9741709977141074e-05, "loss": 1.0776, "step": 45891 }, { "epoch": 0.05, "learning_rate": 4.974169871522764e-05, "loss": 0.5224, "step": 45892 }, { "epoch": 0.05, "learning_rate": 4.974168745306997e-05, "loss": 1.3025, "step": 45893 }, { "epoch": 0.05, "learning_rate": 4.974167619066807e-05, "loss": 0.9844, "step": 45894 }, { "epoch": 0.05, "learning_rate": 4.974166492802193e-05, "loss": 1.1642, "step": 45895 }, { "epoch": 0.05, "learning_rate": 4.974165366513154e-05, "loss": 1.0715, "step": 45896 }, { "epoch": 0.05, "learning_rate": 4.974164240199691e-05, "loss": 1.086, "step": 45897 }, { "epoch": 0.05, "learning_rate": 4.974163113861805e-05, "loss": 1.0996, "step": 45898 }, { "epoch": 0.05, "learning_rate": 4.974161987499495e-05, "loss": 1.1369, "step": 45899 }, { "epoch": 0.05, "learning_rate": 4.9741608611127604e-05, "loss": 0.9683, "step": 45900 }, { "epoch": 0.05, "learning_rate": 4.9741597347016026e-05, "loss": 0.876, "step": 45901 }, { "epoch": 0.05, "learning_rate": 4.974158608266021e-05, "loss": 1.069, "step": 45902 }, { "epoch": 0.05, "learning_rate": 4.9741574818060145e-05, "loss": 0.703, "step": 45903 }, { "epoch": 0.05, "learning_rate": 4.9741563553215856e-05, "loss": 1.0914, "step": 45904 }, { "epoch": 0.05, "learning_rate": 4.974155228812732e-05, "loss": 1.0039, "step": 45905 }, { "epoch": 0.05, "learning_rate": 4.974154102279455e-05, "loss": 0.9149, "step": 45906 }, { "epoch": 0.05, "learning_rate": 4.9741529757217535e-05, "loss": 0.7779, "step": 45907 }, { "epoch": 0.05, "learning_rate": 4.9741518491396285e-05, "loss": 1.0896, "step": 45908 }, { "epoch": 0.05, "learning_rate": 4.9741507225330805e-05, "loss": 1.4308, "step": 45909 }, { "epoch": 0.05, "learning_rate": 4.9741495959021075e-05, "loss": 1.372, "step": 45910 }, { "epoch": 0.05, "learning_rate": 4.9741484692467115e-05, "loss": 0.2683, "step": 45911 }, { "epoch": 0.05, "learning_rate": 4.974147342566892e-05, "loss": 0.2476, "step": 45912 }, { "epoch": 0.05, "learning_rate": 4.974146215862648e-05, "loss": 0.1859, "step": 45913 }, { "epoch": 0.05, "learning_rate": 4.9741450891339805e-05, "loss": 0.2365, "step": 45914 }, { "epoch": 0.05, "learning_rate": 4.97414396238089e-05, "loss": 0.3636, "step": 45915 }, { "epoch": 0.05, "learning_rate": 4.9741428356033746e-05, "loss": 1.1237, "step": 45916 }, { "epoch": 0.05, "learning_rate": 4.9741417088014364e-05, "loss": 1.258, "step": 45917 }, { "epoch": 0.05, "learning_rate": 4.974140581975074e-05, "loss": 1.3424, "step": 45918 }, { "epoch": 0.05, "learning_rate": 4.9741394551242884e-05, "loss": 1.3031, "step": 45919 }, { "epoch": 0.05, "learning_rate": 4.974138328249079e-05, "loss": 0.3535, "step": 45920 }, { "epoch": 0.05, "learning_rate": 4.9741372013494456e-05, "loss": 1.01, "step": 45921 }, { "epoch": 0.05, "learning_rate": 4.974136074425389e-05, "loss": 1.2133, "step": 45922 }, { "epoch": 0.05, "learning_rate": 4.974134947476908e-05, "loss": 1.0329, "step": 45923 }, { "epoch": 0.05, "learning_rate": 4.974133820504005e-05, "loss": 0.8845, "step": 45924 }, { "epoch": 0.05, "learning_rate": 4.974132693506677e-05, "loss": 0.9524, "step": 45925 }, { "epoch": 0.05, "learning_rate": 4.974131566484926e-05, "loss": 1.1506, "step": 45926 }, { "epoch": 0.05, "learning_rate": 4.9741304394387514e-05, "loss": 1.2181, "step": 45927 }, { "epoch": 0.05, "learning_rate": 4.974129312368153e-05, "loss": 1.0454, "step": 45928 }, { "epoch": 0.05, "learning_rate": 4.97412818527313e-05, "loss": 1.1344, "step": 45929 }, { "epoch": 0.05, "learning_rate": 4.974127058153685e-05, "loss": 0.8235, "step": 45930 }, { "epoch": 0.05, "learning_rate": 4.974125931009816e-05, "loss": 0.6094, "step": 45931 }, { "epoch": 0.05, "learning_rate": 4.9741248038415237e-05, "loss": 0.8797, "step": 45932 }, { "epoch": 0.05, "learning_rate": 4.974123676648807e-05, "loss": 0.938, "step": 45933 }, { "epoch": 0.05, "learning_rate": 4.974122549431668e-05, "loss": 1.3387, "step": 45934 }, { "epoch": 0.05, "learning_rate": 4.9741214221901045e-05, "loss": 1.095, "step": 45935 }, { "epoch": 0.05, "learning_rate": 4.9741202949241184e-05, "loss": 0.8754, "step": 45936 }, { "epoch": 0.05, "learning_rate": 4.9741191676337085e-05, "loss": 1.1047, "step": 45937 }, { "epoch": 0.05, "learning_rate": 4.974118040318875e-05, "loss": 1.0831, "step": 45938 }, { "epoch": 0.05, "learning_rate": 4.974116912979617e-05, "loss": 1.0271, "step": 45939 }, { "epoch": 0.05, "learning_rate": 4.974115785615937e-05, "loss": 0.7376, "step": 45940 }, { "epoch": 0.05, "learning_rate": 4.974114658227833e-05, "loss": 0.9695, "step": 45941 }, { "epoch": 0.05, "learning_rate": 4.9741135308153066e-05, "loss": 1.0142, "step": 45942 }, { "epoch": 0.05, "learning_rate": 4.974112403378355e-05, "loss": 1.0095, "step": 45943 }, { "epoch": 0.05, "learning_rate": 4.9741112759169816e-05, "loss": 0.8569, "step": 45944 }, { "epoch": 0.05, "learning_rate": 4.9741101484311836e-05, "loss": 0.6981, "step": 45945 }, { "epoch": 0.05, "learning_rate": 4.9741090209209626e-05, "loss": 0.8095, "step": 45946 }, { "epoch": 0.05, "learning_rate": 4.9741078933863186e-05, "loss": 0.9455, "step": 45947 }, { "epoch": 0.05, "learning_rate": 4.974106765827251e-05, "loss": 1.0195, "step": 45948 }, { "epoch": 0.05, "learning_rate": 4.974105638243761e-05, "loss": 0.8667, "step": 45949 }, { "epoch": 0.05, "learning_rate": 4.9741045106358465e-05, "loss": 0.8944, "step": 45950 }, { "epoch": 0.05, "learning_rate": 4.974103383003509e-05, "loss": 0.4769, "step": 45951 }, { "epoch": 0.05, "learning_rate": 4.974102255346748e-05, "loss": 0.31, "step": 45952 }, { "epoch": 0.05, "learning_rate": 4.974101127665564e-05, "loss": 0.3219, "step": 45953 }, { "epoch": 0.05, "learning_rate": 4.974099999959957e-05, "loss": 0.4292, "step": 45954 }, { "epoch": 0.05, "learning_rate": 4.974098872229926e-05, "loss": 0.2631, "step": 45955 }, { "epoch": 0.05, "learning_rate": 4.974097744475472e-05, "loss": 0.2322, "step": 45956 }, { "epoch": 0.05, "learning_rate": 4.974096616696595e-05, "loss": 0.3061, "step": 45957 }, { "epoch": 0.05, "learning_rate": 4.974095488893295e-05, "loss": 0.5083, "step": 45958 }, { "epoch": 0.05, "learning_rate": 4.974094361065571e-05, "loss": 1.1491, "step": 45959 }, { "epoch": 0.05, "learning_rate": 4.974093233213424e-05, "loss": 1.4093, "step": 45960 }, { "epoch": 0.05, "learning_rate": 4.974092105336854e-05, "loss": 0.9001, "step": 45961 }, { "epoch": 0.05, "learning_rate": 4.9740909774358606e-05, "loss": 1.1521, "step": 45962 }, { "epoch": 0.05, "learning_rate": 4.9740898495104444e-05, "loss": 1.1171, "step": 45963 }, { "epoch": 0.05, "learning_rate": 4.9740887215606044e-05, "loss": 1.1675, "step": 45964 }, { "epoch": 0.05, "learning_rate": 4.9740875935863415e-05, "loss": 0.813, "step": 45965 }, { "epoch": 0.05, "learning_rate": 4.974086465587656e-05, "loss": 0.9882, "step": 45966 }, { "epoch": 0.05, "learning_rate": 4.9740853375645466e-05, "loss": 1.0423, "step": 45967 }, { "epoch": 0.05, "learning_rate": 4.974084209517015e-05, "loss": 0.783, "step": 45968 }, { "epoch": 0.05, "learning_rate": 4.974083081445059e-05, "loss": 1.1214, "step": 45969 }, { "epoch": 0.05, "learning_rate": 4.974081953348681e-05, "loss": 1.2906, "step": 45970 }, { "epoch": 0.05, "learning_rate": 4.9740808252278794e-05, "loss": 0.9702, "step": 45971 }, { "epoch": 0.05, "learning_rate": 4.974079697082655e-05, "loss": 0.8121, "step": 45972 }, { "epoch": 0.05, "learning_rate": 4.974078568913007e-05, "loss": 0.96, "step": 45973 }, { "epoch": 0.05, "learning_rate": 4.9740774407189363e-05, "loss": 1.1197, "step": 45974 }, { "epoch": 0.05, "learning_rate": 4.974076312500443e-05, "loss": 0.6518, "step": 45975 }, { "epoch": 0.05, "learning_rate": 4.974075184257526e-05, "loss": 0.8561, "step": 45976 }, { "epoch": 0.05, "learning_rate": 4.974074055990186e-05, "loss": 1.384, "step": 45977 }, { "epoch": 0.05, "learning_rate": 4.9740729276984235e-05, "loss": 1.6848, "step": 45978 }, { "epoch": 0.05, "learning_rate": 4.974071799382237e-05, "loss": 1.0627, "step": 45979 }, { "epoch": 0.05, "learning_rate": 4.974070671041628e-05, "loss": 0.6708, "step": 45980 }, { "epoch": 0.05, "learning_rate": 4.9740695426765966e-05, "loss": 0.4451, "step": 45981 }, { "epoch": 0.05, "learning_rate": 4.974068414287142e-05, "loss": 0.4959, "step": 45982 }, { "epoch": 0.05, "learning_rate": 4.974067285873264e-05, "loss": 0.5437, "step": 45983 }, { "epoch": 0.05, "learning_rate": 4.974066157434963e-05, "loss": 1.3283, "step": 45984 }, { "epoch": 0.05, "learning_rate": 4.9740650289722395e-05, "loss": 0.9597, "step": 45985 }, { "epoch": 0.05, "learning_rate": 4.9740639004850933e-05, "loss": 1.1042, "step": 45986 }, { "epoch": 0.05, "learning_rate": 4.9740627719735235e-05, "loss": 0.9494, "step": 45987 }, { "epoch": 0.05, "learning_rate": 4.974061643437531e-05, "loss": 0.976, "step": 45988 }, { "epoch": 0.05, "learning_rate": 4.9740605148771156e-05, "loss": 1.0132, "step": 45989 }, { "epoch": 0.05, "learning_rate": 4.974059386292278e-05, "loss": 0.7278, "step": 45990 }, { "epoch": 0.05, "learning_rate": 4.974058257683016e-05, "loss": 1.103, "step": 45991 }, { "epoch": 0.05, "learning_rate": 4.974057129049333e-05, "loss": 1.5215, "step": 45992 }, { "epoch": 0.05, "learning_rate": 4.974056000391226e-05, "loss": 1.048, "step": 45993 }, { "epoch": 0.05, "learning_rate": 4.974054871708697e-05, "loss": 0.8238, "step": 45994 }, { "epoch": 0.05, "learning_rate": 4.974053743001744e-05, "loss": 1.0089, "step": 45995 }, { "epoch": 0.05, "learning_rate": 4.9740526142703686e-05, "loss": 1.2374, "step": 45996 }, { "epoch": 0.05, "learning_rate": 4.97405148551457e-05, "loss": 1.025, "step": 45997 }, { "epoch": 0.05, "learning_rate": 4.97405035673435e-05, "loss": 1.1101, "step": 45998 }, { "epoch": 0.05, "learning_rate": 4.974049227929706e-05, "loss": 1.1553, "step": 45999 }, { "epoch": 0.05, "learning_rate": 4.97404809910064e-05, "loss": 1.0135, "step": 46000 }, { "epoch": 0.05, "eval_loss": 1.088128924369812, "eval_runtime": 128.4713, "eval_samples_per_second": 10.781, "eval_steps_per_second": 5.394, "step": 46000 }, { "epoch": 0.05, "learning_rate": 4.9740469702471506e-05, "loss": 1.1221, "step": 46001 }, { "epoch": 0.05, "learning_rate": 4.974045841369239e-05, "loss": 1.6817, "step": 46002 }, { "epoch": 0.05, "learning_rate": 4.9740447124669044e-05, "loss": 1.159, "step": 46003 }, { "epoch": 0.05, "learning_rate": 4.974043583540147e-05, "loss": 1.0178, "step": 46004 }, { "epoch": 0.05, "learning_rate": 4.974042454588966e-05, "loss": 0.9866, "step": 46005 }, { "epoch": 0.05, "learning_rate": 4.9740413256133633e-05, "loss": 0.9903, "step": 46006 }, { "epoch": 0.05, "learning_rate": 4.974040196613338e-05, "loss": 1.1353, "step": 46007 }, { "epoch": 0.05, "learning_rate": 4.9740390675888905e-05, "loss": 1.9398, "step": 46008 }, { "epoch": 0.05, "learning_rate": 4.974037938540019e-05, "loss": 1.5592, "step": 46009 }, { "epoch": 0.05, "learning_rate": 4.9740368094667256e-05, "loss": 0.697, "step": 46010 }, { "epoch": 0.05, "learning_rate": 4.974035680369009e-05, "loss": 0.4358, "step": 46011 }, { "epoch": 0.05, "learning_rate": 4.974034551246871e-05, "loss": 0.8685, "step": 46012 }, { "epoch": 0.05, "learning_rate": 4.9740334221003095e-05, "loss": 0.7438, "step": 46013 }, { "epoch": 0.05, "learning_rate": 4.974032292929325e-05, "loss": 0.4028, "step": 46014 }, { "epoch": 0.05, "learning_rate": 4.9740311637339186e-05, "loss": 0.3808, "step": 46015 }, { "epoch": 0.05, "learning_rate": 4.97403003451409e-05, "loss": 0.5626, "step": 46016 }, { "epoch": 0.05, "learning_rate": 4.974028905269838e-05, "loss": 0.406, "step": 46017 }, { "epoch": 0.05, "learning_rate": 4.9740277760011634e-05, "loss": 0.3933, "step": 46018 }, { "epoch": 0.05, "learning_rate": 4.974026646708067e-05, "loss": 0.3818, "step": 46019 }, { "epoch": 0.05, "learning_rate": 4.974025517390547e-05, "loss": 0.3951, "step": 46020 }, { "epoch": 0.05, "learning_rate": 4.9740243880486045e-05, "loss": 0.8342, "step": 46021 }, { "epoch": 0.05, "learning_rate": 4.97402325868224e-05, "loss": 0.8004, "step": 46022 }, { "epoch": 0.05, "learning_rate": 4.974022129291453e-05, "loss": 0.8296, "step": 46023 }, { "epoch": 0.05, "learning_rate": 4.974020999876243e-05, "loss": 0.9706, "step": 46024 }, { "epoch": 0.05, "learning_rate": 4.9740198704366114e-05, "loss": 1.1784, "step": 46025 }, { "epoch": 0.05, "learning_rate": 4.974018740972557e-05, "loss": 0.7079, "step": 46026 }, { "epoch": 0.05, "learning_rate": 4.97401761148408e-05, "loss": 0.7575, "step": 46027 }, { "epoch": 0.05, "learning_rate": 4.97401648197118e-05, "loss": 0.6604, "step": 46028 }, { "epoch": 0.05, "learning_rate": 4.974015352433858e-05, "loss": 0.6784, "step": 46029 }, { "epoch": 0.05, "learning_rate": 4.9740142228721146e-05, "loss": 1.0898, "step": 46030 }, { "epoch": 0.05, "learning_rate": 4.974013093285947e-05, "loss": 0.9284, "step": 46031 }, { "epoch": 0.05, "learning_rate": 4.9740119636753576e-05, "loss": 0.9147, "step": 46032 }, { "epoch": 0.05, "learning_rate": 4.974010834040347e-05, "loss": 0.7191, "step": 46033 }, { "epoch": 0.05, "learning_rate": 4.974009704380913e-05, "loss": 0.8236, "step": 46034 }, { "epoch": 0.05, "learning_rate": 4.974008574697056e-05, "loss": 1.0677, "step": 46035 }, { "epoch": 0.05, "learning_rate": 4.974007444988778e-05, "loss": 1.1961, "step": 46036 }, { "epoch": 0.05, "learning_rate": 4.9740063152560765e-05, "loss": 1.2584, "step": 46037 }, { "epoch": 0.05, "learning_rate": 4.974005185498953e-05, "loss": 1.3924, "step": 46038 }, { "epoch": 0.05, "learning_rate": 4.9740040557174076e-05, "loss": 0.5072, "step": 46039 }, { "epoch": 0.05, "learning_rate": 4.974002925911439e-05, "loss": 0.548, "step": 46040 }, { "epoch": 0.05, "learning_rate": 4.974001796081048e-05, "loss": 0.8184, "step": 46041 }, { "epoch": 0.05, "learning_rate": 4.974000666226236e-05, "loss": 0.9166, "step": 46042 }, { "epoch": 0.05, "learning_rate": 4.9739995363470004e-05, "loss": 0.6774, "step": 46043 }, { "epoch": 0.05, "learning_rate": 4.973998406443343e-05, "loss": 0.8406, "step": 46044 }, { "epoch": 0.05, "learning_rate": 4.9739972765152635e-05, "loss": 1.1376, "step": 46045 }, { "epoch": 0.05, "learning_rate": 4.973996146562761e-05, "loss": 1.1313, "step": 46046 }, { "epoch": 0.05, "learning_rate": 4.973995016585838e-05, "loss": 0.8024, "step": 46047 }, { "epoch": 0.05, "learning_rate": 4.973993886584492e-05, "loss": 0.7777, "step": 46048 }, { "epoch": 0.05, "learning_rate": 4.973992756558723e-05, "loss": 0.9044, "step": 46049 }, { "epoch": 0.05, "learning_rate": 4.973991626508532e-05, "loss": 1.1868, "step": 46050 }, { "epoch": 0.05, "learning_rate": 4.973990496433919e-05, "loss": 1.2981, "step": 46051 }, { "epoch": 0.05, "learning_rate": 4.973989366334884e-05, "loss": 1.3062, "step": 46052 }, { "epoch": 0.05, "learning_rate": 4.973988236211427e-05, "loss": 2.3369, "step": 46053 }, { "epoch": 0.05, "learning_rate": 4.973987106063547e-05, "loss": 1.7122, "step": 46054 }, { "epoch": 0.05, "learning_rate": 4.973985975891245e-05, "loss": 1.0447, "step": 46055 }, { "epoch": 0.05, "learning_rate": 4.973984845694522e-05, "loss": 1.0017, "step": 46056 }, { "epoch": 0.05, "learning_rate": 4.973983715473376e-05, "loss": 1.1155, "step": 46057 }, { "epoch": 0.05, "learning_rate": 4.9739825852278076e-05, "loss": 0.8497, "step": 46058 }, { "epoch": 0.05, "learning_rate": 4.973981454957818e-05, "loss": 1.1196, "step": 46059 }, { "epoch": 0.05, "learning_rate": 4.9739803246634055e-05, "loss": 0.9252, "step": 46060 }, { "epoch": 0.05, "learning_rate": 4.973979194344571e-05, "loss": 0.9025, "step": 46061 }, { "epoch": 0.05, "learning_rate": 4.973978064001315e-05, "loss": 0.8338, "step": 46062 }, { "epoch": 0.05, "learning_rate": 4.973976933633636e-05, "loss": 0.8667, "step": 46063 }, { "epoch": 0.05, "learning_rate": 4.9739758032415354e-05, "loss": 1.0324, "step": 46064 }, { "epoch": 0.05, "learning_rate": 4.973974672825012e-05, "loss": 1.0069, "step": 46065 }, { "epoch": 0.05, "learning_rate": 4.973973542384068e-05, "loss": 0.8318, "step": 46066 }, { "epoch": 0.05, "learning_rate": 4.9739724119187015e-05, "loss": 1.0496, "step": 46067 }, { "epoch": 0.05, "learning_rate": 4.973971281428913e-05, "loss": 1.0526, "step": 46068 }, { "epoch": 0.05, "learning_rate": 4.9739701509147016e-05, "loss": 1.122, "step": 46069 }, { "epoch": 0.05, "learning_rate": 4.9739690203760694e-05, "loss": 0.5792, "step": 46070 }, { "epoch": 0.05, "learning_rate": 4.973967889813015e-05, "loss": 1.052, "step": 46071 }, { "epoch": 0.05, "learning_rate": 4.973966759225538e-05, "loss": 1.3143, "step": 46072 }, { "epoch": 0.05, "learning_rate": 4.973965628613639e-05, "loss": 1.3642, "step": 46073 }, { "epoch": 0.05, "learning_rate": 4.973964497977319e-05, "loss": 0.845, "step": 46074 }, { "epoch": 0.05, "learning_rate": 4.9739633673165765e-05, "loss": 0.9264, "step": 46075 }, { "epoch": 0.05, "learning_rate": 4.9739622366314124e-05, "loss": 1.0567, "step": 46076 }, { "epoch": 0.05, "learning_rate": 4.973961105921825e-05, "loss": 1.0427, "step": 46077 }, { "epoch": 0.05, "learning_rate": 4.973959975187817e-05, "loss": 1.0494, "step": 46078 }, { "epoch": 0.05, "learning_rate": 4.9739588444293874e-05, "loss": 0.9011, "step": 46079 }, { "epoch": 0.05, "learning_rate": 4.9739577136465354e-05, "loss": 0.9637, "step": 46080 }, { "epoch": 0.05, "learning_rate": 4.973956582839261e-05, "loss": 1.1452, "step": 46081 }, { "epoch": 0.05, "learning_rate": 4.9739554520075656e-05, "loss": 0.7763, "step": 46082 }, { "epoch": 0.05, "learning_rate": 4.9739543211514486e-05, "loss": 0.5964, "step": 46083 }, { "epoch": 0.05, "learning_rate": 4.9739531902709085e-05, "loss": 1.0182, "step": 46084 }, { "epoch": 0.05, "learning_rate": 4.9739520593659475e-05, "loss": 1.0009, "step": 46085 }, { "epoch": 0.05, "learning_rate": 4.973950928436564e-05, "loss": 1.3538, "step": 46086 }, { "epoch": 0.05, "learning_rate": 4.97394979748276e-05, "loss": 0.8107, "step": 46087 }, { "epoch": 0.05, "learning_rate": 4.973948666504533e-05, "loss": 0.9013, "step": 46088 }, { "epoch": 0.05, "learning_rate": 4.973947535501884e-05, "loss": 1.0986, "step": 46089 }, { "epoch": 0.05, "learning_rate": 4.973946404474814e-05, "loss": 0.8554, "step": 46090 }, { "epoch": 0.05, "learning_rate": 4.9739452734233227e-05, "loss": 1.1617, "step": 46091 }, { "epoch": 0.05, "learning_rate": 4.973944142347409e-05, "loss": 1.2396, "step": 46092 }, { "epoch": 0.05, "learning_rate": 4.973943011247073e-05, "loss": 1.0736, "step": 46093 }, { "epoch": 0.05, "learning_rate": 4.973941880122316e-05, "loss": 0.7158, "step": 46094 }, { "epoch": 0.05, "learning_rate": 4.973940748973137e-05, "loss": 0.997, "step": 46095 }, { "epoch": 0.05, "learning_rate": 4.973939617799537e-05, "loss": 1.0141, "step": 46096 }, { "epoch": 0.05, "learning_rate": 4.973938486601515e-05, "loss": 0.8931, "step": 46097 }, { "epoch": 0.05, "learning_rate": 4.97393735537907e-05, "loss": 0.8572, "step": 46098 }, { "epoch": 0.05, "learning_rate": 4.9739362241322055e-05, "loss": 0.8827, "step": 46099 }, { "epoch": 0.05, "learning_rate": 4.9739350928609176e-05, "loss": 0.9366, "step": 46100 }, { "epoch": 0.05, "learning_rate": 4.9739339615652094e-05, "loss": 0.8838, "step": 46101 }, { "epoch": 0.05, "learning_rate": 4.973932830245078e-05, "loss": 0.9295, "step": 46102 }, { "epoch": 0.05, "learning_rate": 4.973931698900526e-05, "loss": 0.9193, "step": 46103 }, { "epoch": 0.05, "learning_rate": 4.973930567531553e-05, "loss": 0.8719, "step": 46104 }, { "epoch": 0.05, "learning_rate": 4.973929436138157e-05, "loss": 0.8795, "step": 46105 }, { "epoch": 0.05, "learning_rate": 4.97392830472034e-05, "loss": 0.9739, "step": 46106 }, { "epoch": 0.05, "learning_rate": 4.973927173278102e-05, "loss": 0.7565, "step": 46107 }, { "epoch": 0.05, "learning_rate": 4.973926041811442e-05, "loss": 0.937, "step": 46108 }, { "epoch": 0.05, "learning_rate": 4.97392491032036e-05, "loss": 0.836, "step": 46109 }, { "epoch": 0.05, "learning_rate": 4.973923778804856e-05, "loss": 0.8792, "step": 46110 }, { "epoch": 0.05, "learning_rate": 4.973922647264932e-05, "loss": 0.9263, "step": 46111 }, { "epoch": 0.05, "learning_rate": 4.973921515700586e-05, "loss": 0.858, "step": 46112 }, { "epoch": 0.05, "learning_rate": 4.973920384111818e-05, "loss": 0.9985, "step": 46113 }, { "epoch": 0.05, "learning_rate": 4.973919252498629e-05, "loss": 0.9714, "step": 46114 }, { "epoch": 0.05, "learning_rate": 4.973918120861018e-05, "loss": 0.9285, "step": 46115 }, { "epoch": 0.05, "learning_rate": 4.973916989198986e-05, "loss": 1.0976, "step": 46116 }, { "epoch": 0.05, "learning_rate": 4.973915857512532e-05, "loss": 1.0602, "step": 46117 }, { "epoch": 0.05, "learning_rate": 4.973914725801657e-05, "loss": 1.0151, "step": 46118 }, { "epoch": 0.05, "learning_rate": 4.9739135940663605e-05, "loss": 1.1747, "step": 46119 }, { "epoch": 0.05, "learning_rate": 4.973912462306642e-05, "loss": 0.7409, "step": 46120 }, { "epoch": 0.05, "learning_rate": 4.9739113305225024e-05, "loss": 0.7317, "step": 46121 }, { "epoch": 0.05, "learning_rate": 4.973910198713942e-05, "loss": 0.9588, "step": 46122 }, { "epoch": 0.05, "learning_rate": 4.9739090668809604e-05, "loss": 0.8149, "step": 46123 }, { "epoch": 0.05, "learning_rate": 4.973907935023556e-05, "loss": 1.1251, "step": 46124 }, { "epoch": 0.05, "learning_rate": 4.973906803141731e-05, "loss": 1.0758, "step": 46125 }, { "epoch": 0.05, "learning_rate": 4.973905671235485e-05, "loss": 1.2384, "step": 46126 }, { "epoch": 0.05, "learning_rate": 4.973904539304817e-05, "loss": 1.0873, "step": 46127 }, { "epoch": 0.05, "learning_rate": 4.973903407349728e-05, "loss": 1.67, "step": 46128 }, { "epoch": 0.05, "learning_rate": 4.973902275370218e-05, "loss": 1.0478, "step": 46129 }, { "epoch": 0.05, "learning_rate": 4.973901143366286e-05, "loss": 0.9046, "step": 46130 }, { "epoch": 0.05, "learning_rate": 4.973900011337933e-05, "loss": 0.7203, "step": 46131 }, { "epoch": 0.05, "learning_rate": 4.973898879285159e-05, "loss": 0.1906, "step": 46132 }, { "epoch": 0.05, "learning_rate": 4.973897747207963e-05, "loss": 0.1606, "step": 46133 }, { "epoch": 0.05, "learning_rate": 4.973896615106346e-05, "loss": 0.2426, "step": 46134 }, { "epoch": 0.05, "learning_rate": 4.973895482980308e-05, "loss": 0.1797, "step": 46135 }, { "epoch": 0.05, "learning_rate": 4.9738943508298485e-05, "loss": 0.1832, "step": 46136 }, { "epoch": 0.05, "learning_rate": 4.973893218654968e-05, "loss": 0.2536, "step": 46137 }, { "epoch": 0.05, "learning_rate": 4.973892086455666e-05, "loss": 0.2213, "step": 46138 }, { "epoch": 0.05, "learning_rate": 4.9738909542319435e-05, "loss": 0.0821, "step": 46139 }, { "epoch": 0.05, "learning_rate": 4.973889821983799e-05, "loss": 0.1015, "step": 46140 }, { "epoch": 0.05, "learning_rate": 4.9738886897112335e-05, "loss": 0.0643, "step": 46141 }, { "epoch": 0.05, "learning_rate": 4.973887557414246e-05, "loss": 0.0834, "step": 46142 }, { "epoch": 0.05, "learning_rate": 4.973886425092839e-05, "loss": 0.0991, "step": 46143 }, { "epoch": 0.05, "learning_rate": 4.973885292747009e-05, "loss": 0.0719, "step": 46144 }, { "epoch": 0.05, "learning_rate": 4.9738841603767596e-05, "loss": 0.3461, "step": 46145 }, { "epoch": 0.05, "learning_rate": 4.973883027982088e-05, "loss": 0.0725, "step": 46146 }, { "epoch": 0.05, "learning_rate": 4.973881895562996e-05, "loss": 0.0637, "step": 46147 }, { "epoch": 0.05, "learning_rate": 4.9738807631194815e-05, "loss": 0.7106, "step": 46148 }, { "epoch": 0.05, "learning_rate": 4.9738796306515475e-05, "loss": 1.0168, "step": 46149 }, { "epoch": 0.05, "learning_rate": 4.973878498159191e-05, "loss": 0.7833, "step": 46150 }, { "epoch": 0.05, "learning_rate": 4.9738773656424146e-05, "loss": 1.6028, "step": 46151 }, { "epoch": 0.05, "learning_rate": 4.973876233101217e-05, "loss": 0.8583, "step": 46152 }, { "epoch": 0.05, "learning_rate": 4.973875100535598e-05, "loss": 1.1071, "step": 46153 }, { "epoch": 0.05, "learning_rate": 4.9738739679455575e-05, "loss": 1.1527, "step": 46154 }, { "epoch": 0.05, "learning_rate": 4.973872835331096e-05, "loss": 0.9847, "step": 46155 }, { "epoch": 0.05, "learning_rate": 4.973871702692214e-05, "loss": 1.0121, "step": 46156 }, { "epoch": 0.05, "learning_rate": 4.973870570028911e-05, "loss": 0.9653, "step": 46157 }, { "epoch": 0.05, "learning_rate": 4.973869437341187e-05, "loss": 1.184, "step": 46158 }, { "epoch": 0.05, "learning_rate": 4.973868304629042e-05, "loss": 0.874, "step": 46159 }, { "epoch": 0.05, "learning_rate": 4.9738671718924756e-05, "loss": 0.9355, "step": 46160 }, { "epoch": 0.05, "learning_rate": 4.9738660391314886e-05, "loss": 0.9987, "step": 46161 }, { "epoch": 0.05, "learning_rate": 4.9738649063460805e-05, "loss": 0.9877, "step": 46162 }, { "epoch": 0.05, "learning_rate": 4.9738637735362515e-05, "loss": 1.0705, "step": 46163 }, { "epoch": 0.05, "learning_rate": 4.9738626407020015e-05, "loss": 1.2803, "step": 46164 }, { "epoch": 0.05, "learning_rate": 4.9738615078433306e-05, "loss": 1.2143, "step": 46165 }, { "epoch": 0.05, "learning_rate": 4.9738603749602386e-05, "loss": 1.017, "step": 46166 }, { "epoch": 0.05, "learning_rate": 4.9738592420527264e-05, "loss": 1.0055, "step": 46167 }, { "epoch": 0.05, "learning_rate": 4.9738581091207925e-05, "loss": 1.5356, "step": 46168 }, { "epoch": 0.05, "learning_rate": 4.973856976164438e-05, "loss": 0.9599, "step": 46169 }, { "epoch": 0.05, "learning_rate": 4.9738558431836624e-05, "loss": 1.2514, "step": 46170 }, { "epoch": 0.05, "learning_rate": 4.973854710178466e-05, "loss": 1.3564, "step": 46171 }, { "epoch": 0.05, "learning_rate": 4.973853577148849e-05, "loss": 1.0899, "step": 46172 }, { "epoch": 0.05, "learning_rate": 4.973852444094811e-05, "loss": 0.9837, "step": 46173 }, { "epoch": 0.05, "learning_rate": 4.973851311016353e-05, "loss": 2.2271, "step": 46174 }, { "epoch": 0.05, "learning_rate": 4.9738501779134726e-05, "loss": 2.2199, "step": 46175 }, { "epoch": 0.05, "learning_rate": 4.973849044786173e-05, "loss": 1.0326, "step": 46176 }, { "epoch": 0.05, "learning_rate": 4.9738479116344516e-05, "loss": 0.6776, "step": 46177 }, { "epoch": 0.05, "learning_rate": 4.973846778458309e-05, "loss": 1.3234, "step": 46178 }, { "epoch": 0.05, "learning_rate": 4.973845645257747e-05, "loss": 0.8785, "step": 46179 }, { "epoch": 0.05, "learning_rate": 4.973844512032763e-05, "loss": 1.1466, "step": 46180 }, { "epoch": 0.05, "learning_rate": 4.9738433787833586e-05, "loss": 1.0675, "step": 46181 }, { "epoch": 0.05, "learning_rate": 4.973842245509534e-05, "loss": 0.8971, "step": 46182 }, { "epoch": 0.05, "learning_rate": 4.973841112211288e-05, "loss": 0.9322, "step": 46183 }, { "epoch": 0.05, "learning_rate": 4.973839978888621e-05, "loss": 0.9082, "step": 46184 }, { "epoch": 0.05, "learning_rate": 4.973838845541535e-05, "loss": 1.0472, "step": 46185 }, { "epoch": 0.05, "learning_rate": 4.9738377121700266e-05, "loss": 1.0173, "step": 46186 }, { "epoch": 0.05, "learning_rate": 4.973836578774098e-05, "loss": 0.8577, "step": 46187 }, { "epoch": 0.05, "learning_rate": 4.973835445353749e-05, "loss": 1.0348, "step": 46188 }, { "epoch": 0.05, "learning_rate": 4.973834311908979e-05, "loss": 0.9792, "step": 46189 }, { "epoch": 0.05, "learning_rate": 4.9738331784397886e-05, "loss": 1.1632, "step": 46190 }, { "epoch": 0.05, "learning_rate": 4.973832044946177e-05, "loss": 1.4496, "step": 46191 }, { "epoch": 0.05, "learning_rate": 4.973830911428146e-05, "loss": 2.0339, "step": 46192 }, { "epoch": 0.05, "learning_rate": 4.9738297778856935e-05, "loss": 1.5185, "step": 46193 }, { "epoch": 0.05, "learning_rate": 4.9738286443188203e-05, "loss": 0.8782, "step": 46194 }, { "epoch": 0.05, "learning_rate": 4.973827510727527e-05, "loss": 1.2152, "step": 46195 }, { "epoch": 0.05, "learning_rate": 4.9738263771118124e-05, "loss": 1.2991, "step": 46196 }, { "epoch": 0.05, "learning_rate": 4.973825243471678e-05, "loss": 0.5049, "step": 46197 }, { "epoch": 0.05, "learning_rate": 4.9738241098071226e-05, "loss": 0.904, "step": 46198 }, { "epoch": 0.05, "learning_rate": 4.9738229761181466e-05, "loss": 1.1702, "step": 46199 }, { "epoch": 0.05, "learning_rate": 4.97382184240475e-05, "loss": 1.0171, "step": 46200 }, { "epoch": 0.05, "learning_rate": 4.973820708666933e-05, "loss": 1.8488, "step": 46201 }, { "epoch": 0.05, "learning_rate": 4.9738195749046953e-05, "loss": 1.3397, "step": 46202 }, { "epoch": 0.05, "learning_rate": 4.9738184411180374e-05, "loss": 0.9884, "step": 46203 }, { "epoch": 0.05, "learning_rate": 4.973817307306959e-05, "loss": 1.0864, "step": 46204 }, { "epoch": 0.05, "learning_rate": 4.973816173471461e-05, "loss": 0.9165, "step": 46205 }, { "epoch": 0.05, "learning_rate": 4.973815039611541e-05, "loss": 1.1529, "step": 46206 }, { "epoch": 0.05, "learning_rate": 4.9738139057272015e-05, "loss": 0.4871, "step": 46207 }, { "epoch": 0.05, "learning_rate": 4.9738127718184414e-05, "loss": 0.7614, "step": 46208 }, { "epoch": 0.05, "learning_rate": 4.9738116378852604e-05, "loss": 1.0647, "step": 46209 }, { "epoch": 0.05, "learning_rate": 4.97381050392766e-05, "loss": 1.1256, "step": 46210 }, { "epoch": 0.05, "learning_rate": 4.973809369945638e-05, "loss": 1.0561, "step": 46211 }, { "epoch": 0.05, "learning_rate": 4.9738082359391954e-05, "loss": 0.5093, "step": 46212 }, { "epoch": 0.05, "learning_rate": 4.973807101908333e-05, "loss": 0.552, "step": 46213 }, { "epoch": 0.05, "learning_rate": 4.97380596785305e-05, "loss": 0.9909, "step": 46214 }, { "epoch": 0.05, "learning_rate": 4.973804833773348e-05, "loss": 1.1633, "step": 46215 }, { "epoch": 0.05, "learning_rate": 4.973803699669224e-05, "loss": 0.9237, "step": 46216 }, { "epoch": 0.05, "learning_rate": 4.97380256554068e-05, "loss": 0.9836, "step": 46217 }, { "epoch": 0.05, "learning_rate": 4.9738014313877155e-05, "loss": 0.6749, "step": 46218 }, { "epoch": 0.05, "learning_rate": 4.9738002972103314e-05, "loss": 1.1439, "step": 46219 }, { "epoch": 0.05, "learning_rate": 4.973799163008527e-05, "loss": 0.9177, "step": 46220 }, { "epoch": 0.05, "learning_rate": 4.973798028782302e-05, "loss": 0.6738, "step": 46221 }, { "epoch": 0.05, "learning_rate": 4.973796894531656e-05, "loss": 1.135, "step": 46222 }, { "epoch": 0.05, "learning_rate": 4.97379576025659e-05, "loss": 1.1731, "step": 46223 }, { "epoch": 0.05, "learning_rate": 4.973794625957105e-05, "loss": 1.0592, "step": 46224 }, { "epoch": 0.05, "learning_rate": 4.973793491633198e-05, "loss": 0.6199, "step": 46225 }, { "epoch": 0.05, "learning_rate": 4.973792357284872e-05, "loss": 0.4928, "step": 46226 }, { "epoch": 0.05, "learning_rate": 4.973791222912125e-05, "loss": 0.95, "step": 46227 }, { "epoch": 0.05, "learning_rate": 4.973790088514958e-05, "loss": 0.7266, "step": 46228 }, { "epoch": 0.05, "learning_rate": 4.973788954093371e-05, "loss": 1.0482, "step": 46229 }, { "epoch": 0.05, "learning_rate": 4.973787819647364e-05, "loss": 0.8721, "step": 46230 }, { "epoch": 0.05, "learning_rate": 4.973786685176936e-05, "loss": 0.846, "step": 46231 }, { "epoch": 0.05, "learning_rate": 4.9737855506820884e-05, "loss": 0.9536, "step": 46232 }, { "epoch": 0.05, "learning_rate": 4.973784416162821e-05, "loss": 0.9815, "step": 46233 }, { "epoch": 0.05, "learning_rate": 4.973783281619132e-05, "loss": 0.9185, "step": 46234 }, { "epoch": 0.05, "learning_rate": 4.973782147051025e-05, "loss": 0.8916, "step": 46235 }, { "epoch": 0.05, "learning_rate": 4.973781012458496e-05, "loss": 0.9762, "step": 46236 }, { "epoch": 0.05, "learning_rate": 4.973779877841548e-05, "loss": 1.2532, "step": 46237 }, { "epoch": 0.05, "learning_rate": 4.973778743200179e-05, "loss": 1.0191, "step": 46238 }, { "epoch": 0.05, "learning_rate": 4.973777608534391e-05, "loss": 1.1298, "step": 46239 }, { "epoch": 0.05, "learning_rate": 4.973776473844182e-05, "loss": 0.983, "step": 46240 }, { "epoch": 0.05, "learning_rate": 4.9737753391295525e-05, "loss": 1.0815, "step": 46241 }, { "epoch": 0.05, "learning_rate": 4.973774204390504e-05, "loss": 0.9752, "step": 46242 }, { "epoch": 0.05, "learning_rate": 4.973773069627035e-05, "loss": 0.9854, "step": 46243 }, { "epoch": 0.05, "learning_rate": 4.973771934839146e-05, "loss": 1.0567, "step": 46244 }, { "epoch": 0.05, "learning_rate": 4.973770800026837e-05, "loss": 0.7954, "step": 46245 }, { "epoch": 0.05, "learning_rate": 4.9737696651901085e-05, "loss": 0.7213, "step": 46246 }, { "epoch": 0.05, "learning_rate": 4.9737685303289595e-05, "loss": 0.8084, "step": 46247 }, { "epoch": 0.05, "learning_rate": 4.97376739544339e-05, "loss": 1.0332, "step": 46248 }, { "epoch": 0.05, "learning_rate": 4.973766260533401e-05, "loss": 0.9661, "step": 46249 }, { "epoch": 0.05, "learning_rate": 4.973765125598992e-05, "loss": 0.8562, "step": 46250 }, { "epoch": 0.05, "learning_rate": 4.973763990640163e-05, "loss": 1.0692, "step": 46251 }, { "epoch": 0.05, "learning_rate": 4.973762855656914e-05, "loss": 0.7944, "step": 46252 }, { "epoch": 0.05, "learning_rate": 4.973761720649245e-05, "loss": 0.5033, "step": 46253 }, { "epoch": 0.05, "learning_rate": 4.973760585617156e-05, "loss": 0.3683, "step": 46254 }, { "epoch": 0.05, "learning_rate": 4.9737594505606475e-05, "loss": 0.4505, "step": 46255 }, { "epoch": 0.05, "learning_rate": 4.973758315479719e-05, "loss": 0.3684, "step": 46256 }, { "epoch": 0.05, "learning_rate": 4.9737571803743706e-05, "loss": 0.2726, "step": 46257 }, { "epoch": 0.05, "learning_rate": 4.973756045244602e-05, "loss": 0.4562, "step": 46258 }, { "epoch": 0.05, "learning_rate": 4.973754910090414e-05, "loss": 0.5241, "step": 46259 }, { "epoch": 0.05, "learning_rate": 4.973753774911806e-05, "loss": 1.0508, "step": 46260 }, { "epoch": 0.05, "learning_rate": 4.973752639708777e-05, "loss": 0.7512, "step": 46261 }, { "epoch": 0.05, "learning_rate": 4.97375150448133e-05, "loss": 0.5845, "step": 46262 }, { "epoch": 0.05, "learning_rate": 4.9737503692294615e-05, "loss": 0.4788, "step": 46263 }, { "epoch": 0.05, "learning_rate": 4.973749233953174e-05, "loss": 1.0201, "step": 46264 }, { "epoch": 0.05, "learning_rate": 4.973748098652467e-05, "loss": 1.0509, "step": 46265 }, { "epoch": 0.05, "learning_rate": 4.97374696332734e-05, "loss": 0.8961, "step": 46266 }, { "epoch": 0.05, "learning_rate": 4.9737458279777925e-05, "loss": 0.3393, "step": 46267 }, { "epoch": 0.05, "learning_rate": 4.973744692603826e-05, "loss": 1.062, "step": 46268 }, { "epoch": 0.05, "learning_rate": 4.97374355720544e-05, "loss": 0.8744, "step": 46269 }, { "epoch": 0.05, "learning_rate": 4.9737424217826335e-05, "loss": 1.1657, "step": 46270 }, { "epoch": 0.05, "learning_rate": 4.9737412863354075e-05, "loss": 0.97, "step": 46271 }, { "epoch": 0.05, "learning_rate": 4.973740150863762e-05, "loss": 1.0532, "step": 46272 }, { "epoch": 0.05, "learning_rate": 4.9737390153676966e-05, "loss": 1.345, "step": 46273 }, { "epoch": 0.05, "learning_rate": 4.973737879847211e-05, "loss": 0.7529, "step": 46274 }, { "epoch": 0.05, "learning_rate": 4.9737367443023065e-05, "loss": 0.9781, "step": 46275 }, { "epoch": 0.05, "learning_rate": 4.9737356087329824e-05, "loss": 1.1741, "step": 46276 }, { "epoch": 0.05, "learning_rate": 4.973734473139238e-05, "loss": 1.1168, "step": 46277 }, { "epoch": 0.05, "learning_rate": 4.973733337521074e-05, "loss": 1.1063, "step": 46278 }, { "epoch": 0.05, "learning_rate": 4.97373220187849e-05, "loss": 0.7657, "step": 46279 }, { "epoch": 0.05, "learning_rate": 4.973731066211487e-05, "loss": 0.9569, "step": 46280 }, { "epoch": 0.05, "learning_rate": 4.973729930520065e-05, "loss": 1.0302, "step": 46281 }, { "epoch": 0.05, "learning_rate": 4.973728794804222e-05, "loss": 1.3288, "step": 46282 }, { "epoch": 0.05, "learning_rate": 4.97372765906396e-05, "loss": 1.1471, "step": 46283 }, { "epoch": 0.05, "learning_rate": 4.9737265232992783e-05, "loss": 0.7637, "step": 46284 }, { "epoch": 0.05, "learning_rate": 4.9737253875101776e-05, "loss": 1.2097, "step": 46285 }, { "epoch": 0.05, "learning_rate": 4.9737242516966566e-05, "loss": 1.0065, "step": 46286 }, { "epoch": 0.05, "learning_rate": 4.973723115858716e-05, "loss": 1.0142, "step": 46287 }, { "epoch": 0.05, "learning_rate": 4.9737219799963564e-05, "loss": 1.1505, "step": 46288 }, { "epoch": 0.05, "learning_rate": 4.9737208441095766e-05, "loss": 0.881, "step": 46289 }, { "epoch": 0.05, "learning_rate": 4.973719708198378e-05, "loss": 1.092, "step": 46290 }, { "epoch": 0.05, "learning_rate": 4.973718572262759e-05, "loss": 0.9132, "step": 46291 }, { "epoch": 0.05, "learning_rate": 4.973717436302722e-05, "loss": 0.9412, "step": 46292 }, { "epoch": 0.05, "learning_rate": 4.973716300318264e-05, "loss": 1.2363, "step": 46293 }, { "epoch": 0.05, "learning_rate": 4.973715164309387e-05, "loss": 1.0782, "step": 46294 }, { "epoch": 0.05, "learning_rate": 4.973714028276091e-05, "loss": 0.7016, "step": 46295 }, { "epoch": 0.05, "learning_rate": 4.973712892218375e-05, "loss": 0.983, "step": 46296 }, { "epoch": 0.05, "learning_rate": 4.973711756136239e-05, "loss": 0.8563, "step": 46297 }, { "epoch": 0.05, "learning_rate": 4.973710620029684e-05, "loss": 0.9371, "step": 46298 }, { "epoch": 0.05, "learning_rate": 4.97370948389871e-05, "loss": 1.0689, "step": 46299 }, { "epoch": 0.05, "learning_rate": 4.9737083477433165e-05, "loss": 1.0107, "step": 46300 }, { "epoch": 0.05, "learning_rate": 4.973707211563504e-05, "loss": 1.0353, "step": 46301 }, { "epoch": 0.05, "learning_rate": 4.973706075359271e-05, "loss": 1.3347, "step": 46302 }, { "epoch": 0.05, "learning_rate": 4.97370493913062e-05, "loss": 1.3477, "step": 46303 }, { "epoch": 0.05, "learning_rate": 4.9737038028775484e-05, "loss": 1.3379, "step": 46304 }, { "epoch": 0.05, "learning_rate": 4.973702666600057e-05, "loss": 1.6801, "step": 46305 }, { "epoch": 0.05, "learning_rate": 4.973701530298148e-05, "loss": 1.257, "step": 46306 }, { "epoch": 0.05, "learning_rate": 4.973700393971818e-05, "loss": 1.0325, "step": 46307 }, { "epoch": 0.05, "learning_rate": 4.97369925762107e-05, "loss": 1.1076, "step": 46308 }, { "epoch": 0.05, "learning_rate": 4.973698121245902e-05, "loss": 1.0162, "step": 46309 }, { "epoch": 0.05, "learning_rate": 4.973696984846315e-05, "loss": 1.2093, "step": 46310 }, { "epoch": 0.05, "learning_rate": 4.973695848422308e-05, "loss": 0.8726, "step": 46311 }, { "epoch": 0.05, "learning_rate": 4.9736947119738824e-05, "loss": 0.7606, "step": 46312 }, { "epoch": 0.05, "learning_rate": 4.973693575501037e-05, "loss": 1.1793, "step": 46313 }, { "epoch": 0.05, "learning_rate": 4.9736924390037734e-05, "loss": 1.8193, "step": 46314 }, { "epoch": 0.05, "learning_rate": 4.9736913024820895e-05, "loss": 1.9494, "step": 46315 }, { "epoch": 0.05, "learning_rate": 4.9736901659359866e-05, "loss": 1.6954, "step": 46316 }, { "epoch": 0.05, "learning_rate": 4.973689029365465e-05, "loss": 1.9333, "step": 46317 }, { "epoch": 0.05, "learning_rate": 4.9736878927705234e-05, "loss": 1.006, "step": 46318 }, { "epoch": 0.05, "learning_rate": 4.973686756151164e-05, "loss": 0.447, "step": 46319 }, { "epoch": 0.05, "learning_rate": 4.973685619507383e-05, "loss": 1.1141, "step": 46320 }, { "epoch": 0.05, "learning_rate": 4.973684482839185e-05, "loss": 1.2987, "step": 46321 }, { "epoch": 0.05, "learning_rate": 4.973683346146567e-05, "loss": 1.0013, "step": 46322 }, { "epoch": 0.05, "learning_rate": 4.97368220942953e-05, "loss": 1.117, "step": 46323 }, { "epoch": 0.05, "learning_rate": 4.9736810726880736e-05, "loss": 1.4363, "step": 46324 }, { "epoch": 0.05, "learning_rate": 4.9736799359221975e-05, "loss": 0.9702, "step": 46325 }, { "epoch": 0.05, "learning_rate": 4.973678799131903e-05, "loss": 1.25, "step": 46326 }, { "epoch": 0.05, "learning_rate": 4.97367766231719e-05, "loss": 0.7162, "step": 46327 }, { "epoch": 0.05, "learning_rate": 4.973676525478057e-05, "loss": 1.3407, "step": 46328 }, { "epoch": 0.05, "learning_rate": 4.973675388614505e-05, "loss": 0.6739, "step": 46329 }, { "epoch": 0.05, "learning_rate": 4.9736742517265344e-05, "loss": 0.747, "step": 46330 }, { "epoch": 0.05, "learning_rate": 4.973673114814144e-05, "loss": 0.8427, "step": 46331 }, { "epoch": 0.05, "learning_rate": 4.973671977877335e-05, "loss": 0.795, "step": 46332 }, { "epoch": 0.05, "learning_rate": 4.973670840916107e-05, "loss": 0.8365, "step": 46333 }, { "epoch": 0.05, "learning_rate": 4.97366970393046e-05, "loss": 1.9248, "step": 46334 }, { "epoch": 0.05, "learning_rate": 4.9736685669203937e-05, "loss": 0.734, "step": 46335 }, { "epoch": 0.05, "learning_rate": 4.9736674298859085e-05, "loss": 1.0033, "step": 46336 }, { "epoch": 0.05, "learning_rate": 4.973666292827004e-05, "loss": 0.9307, "step": 46337 }, { "epoch": 0.05, "learning_rate": 4.973665155743681e-05, "loss": 0.9305, "step": 46338 }, { "epoch": 0.05, "learning_rate": 4.973664018635938e-05, "loss": 0.7755, "step": 46339 }, { "epoch": 0.05, "learning_rate": 4.973662881503778e-05, "loss": 0.9885, "step": 46340 }, { "epoch": 0.05, "learning_rate": 4.973661744347198e-05, "loss": 0.9172, "step": 46341 }, { "epoch": 0.05, "learning_rate": 4.9736606071661986e-05, "loss": 0.8834, "step": 46342 }, { "epoch": 0.05, "learning_rate": 4.973659469960781e-05, "loss": 0.3965, "step": 46343 }, { "epoch": 0.05, "learning_rate": 4.9736583327309436e-05, "loss": 0.8758, "step": 46344 }, { "epoch": 0.05, "learning_rate": 4.973657195476687e-05, "loss": 1.0212, "step": 46345 }, { "epoch": 0.05, "learning_rate": 4.973656058198013e-05, "loss": 0.774, "step": 46346 }, { "epoch": 0.05, "learning_rate": 4.973654920894919e-05, "loss": 0.8512, "step": 46347 }, { "epoch": 0.05, "learning_rate": 4.973653783567407e-05, "loss": 0.6885, "step": 46348 }, { "epoch": 0.05, "learning_rate": 4.9736526462154754e-05, "loss": 1.3706, "step": 46349 }, { "epoch": 0.05, "learning_rate": 4.973651508839126e-05, "loss": 1.3365, "step": 46350 }, { "epoch": 0.05, "learning_rate": 4.9736503714383565e-05, "loss": 1.0506, "step": 46351 }, { "epoch": 0.05, "learning_rate": 4.973649234013168e-05, "loss": 0.5063, "step": 46352 }, { "epoch": 0.05, "learning_rate": 4.973648096563562e-05, "loss": 1.0994, "step": 46353 }, { "epoch": 0.05, "learning_rate": 4.9736469590895364e-05, "loss": 0.9101, "step": 46354 }, { "epoch": 0.05, "learning_rate": 4.973645821591092e-05, "loss": 1.7018, "step": 46355 }, { "epoch": 0.05, "learning_rate": 4.973644684068229e-05, "loss": 1.5184, "step": 46356 }, { "epoch": 0.05, "learning_rate": 4.973643546520947e-05, "loss": 1.226, "step": 46357 }, { "epoch": 0.05, "learning_rate": 4.973642408949246e-05, "loss": 0.904, "step": 46358 }, { "epoch": 0.05, "learning_rate": 4.9736412713531264e-05, "loss": 1.1249, "step": 46359 }, { "epoch": 0.05, "learning_rate": 4.973640133732589e-05, "loss": 1.0737, "step": 46360 }, { "epoch": 0.05, "learning_rate": 4.973638996087632e-05, "loss": 0.9519, "step": 46361 }, { "epoch": 0.05, "learning_rate": 4.9736378584182564e-05, "loss": 1.1093, "step": 46362 }, { "epoch": 0.05, "learning_rate": 4.973636720724462e-05, "loss": 1.0679, "step": 46363 }, { "epoch": 0.05, "learning_rate": 4.9736355830062484e-05, "loss": 1.0799, "step": 46364 }, { "epoch": 0.05, "learning_rate": 4.973634445263617e-05, "loss": 0.7797, "step": 46365 }, { "epoch": 0.05, "learning_rate": 4.9736333074965665e-05, "loss": 1.0284, "step": 46366 }, { "epoch": 0.05, "learning_rate": 4.9736321697050976e-05, "loss": 1.3851, "step": 46367 }, { "epoch": 0.05, "learning_rate": 4.9736310318892096e-05, "loss": 0.7082, "step": 46368 }, { "epoch": 0.05, "learning_rate": 4.9736298940489035e-05, "loss": 0.8834, "step": 46369 }, { "epoch": 0.05, "learning_rate": 4.973628756184179e-05, "loss": 1.0504, "step": 46370 }, { "epoch": 0.05, "learning_rate": 4.973627618295035e-05, "loss": 1.1537, "step": 46371 }, { "epoch": 0.05, "learning_rate": 4.9736264803814726e-05, "loss": 1.0044, "step": 46372 }, { "epoch": 0.05, "learning_rate": 4.973625342443492e-05, "loss": 0.9307, "step": 46373 }, { "epoch": 0.05, "learning_rate": 4.9736242044810924e-05, "loss": 1.5932, "step": 46374 }, { "epoch": 0.05, "learning_rate": 4.9736230664942746e-05, "loss": 1.1222, "step": 46375 }, { "epoch": 0.05, "learning_rate": 4.973621928483038e-05, "loss": 0.8802, "step": 46376 }, { "epoch": 0.05, "learning_rate": 4.973620790447383e-05, "loss": 1.1988, "step": 46377 }, { "epoch": 0.05, "learning_rate": 4.973619652387309e-05, "loss": 1.2271, "step": 46378 }, { "epoch": 0.05, "learning_rate": 4.9736185143028166e-05, "loss": 1.3287, "step": 46379 }, { "epoch": 0.05, "learning_rate": 4.973617376193906e-05, "loss": 1.2064, "step": 46380 }, { "epoch": 0.05, "learning_rate": 4.9736162380605774e-05, "loss": 1.5395, "step": 46381 }, { "epoch": 0.05, "learning_rate": 4.9736150999028297e-05, "loss": 1.2781, "step": 46382 }, { "epoch": 0.05, "learning_rate": 4.973613961720664e-05, "loss": 0.9619, "step": 46383 }, { "epoch": 0.05, "learning_rate": 4.973612823514079e-05, "loss": 0.8049, "step": 46384 }, { "epoch": 0.05, "learning_rate": 4.9736116852830756e-05, "loss": 1.5093, "step": 46385 }, { "epoch": 0.05, "learning_rate": 4.973610547027654e-05, "loss": 1.132, "step": 46386 }, { "epoch": 0.05, "learning_rate": 4.9736094087478144e-05, "loss": 1.0031, "step": 46387 }, { "epoch": 0.05, "learning_rate": 4.973608270443556e-05, "loss": 1.2378, "step": 46388 }, { "epoch": 0.05, "learning_rate": 4.973607132114879e-05, "loss": 1.2285, "step": 46389 }, { "epoch": 0.05, "learning_rate": 4.9736059937617836e-05, "loss": 1.2417, "step": 46390 }, { "epoch": 0.05, "learning_rate": 4.9736048553842695e-05, "loss": 0.6836, "step": 46391 }, { "epoch": 0.05, "learning_rate": 4.973603716982338e-05, "loss": 0.9024, "step": 46392 }, { "epoch": 0.05, "learning_rate": 4.973602578555987e-05, "loss": 0.9566, "step": 46393 }, { "epoch": 0.05, "learning_rate": 4.973601440105219e-05, "loss": 1.1139, "step": 46394 }, { "epoch": 0.05, "learning_rate": 4.973600301630032e-05, "loss": 0.9578, "step": 46395 }, { "epoch": 0.05, "learning_rate": 4.9735991631304255e-05, "loss": 1.0779, "step": 46396 }, { "epoch": 0.05, "learning_rate": 4.973598024606402e-05, "loss": 1.0705, "step": 46397 }, { "epoch": 0.05, "learning_rate": 4.97359688605796e-05, "loss": 0.9484, "step": 46398 }, { "epoch": 0.05, "learning_rate": 4.973595747485099e-05, "loss": 0.8725, "step": 46399 }, { "epoch": 0.05, "learning_rate": 4.97359460888782e-05, "loss": 1.1308, "step": 46400 }, { "epoch": 0.05, "learning_rate": 4.9735934702661235e-05, "loss": 0.9828, "step": 46401 }, { "epoch": 0.05, "learning_rate": 4.9735923316200085e-05, "loss": 0.6468, "step": 46402 }, { "epoch": 0.05, "learning_rate": 4.973591192949475e-05, "loss": 1.278, "step": 46403 }, { "epoch": 0.05, "learning_rate": 4.9735900542545236e-05, "loss": 1.1535, "step": 46404 }, { "epoch": 0.05, "learning_rate": 4.973588915535154e-05, "loss": 0.7204, "step": 46405 }, { "epoch": 0.05, "learning_rate": 4.973587776791365e-05, "loss": 1.0619, "step": 46406 }, { "epoch": 0.05, "learning_rate": 4.9735866380231585e-05, "loss": 0.9273, "step": 46407 }, { "epoch": 0.05, "learning_rate": 4.973585499230534e-05, "loss": 0.8803, "step": 46408 }, { "epoch": 0.05, "learning_rate": 4.973584360413491e-05, "loss": 0.9216, "step": 46409 }, { "epoch": 0.05, "learning_rate": 4.9735832215720305e-05, "loss": 0.5891, "step": 46410 }, { "epoch": 0.05, "learning_rate": 4.973582082706151e-05, "loss": 1.0461, "step": 46411 }, { "epoch": 0.05, "learning_rate": 4.973580943815853e-05, "loss": 1.1986, "step": 46412 }, { "epoch": 0.05, "learning_rate": 4.973579804901138e-05, "loss": 1.3352, "step": 46413 }, { "epoch": 0.05, "learning_rate": 4.9735786659620044e-05, "loss": 0.8653, "step": 46414 }, { "epoch": 0.05, "learning_rate": 4.973577526998452e-05, "loss": 1.279, "step": 46415 }, { "epoch": 0.05, "learning_rate": 4.9735763880104825e-05, "loss": 0.9058, "step": 46416 }, { "epoch": 0.05, "learning_rate": 4.973575248998095e-05, "loss": 1.1943, "step": 46417 }, { "epoch": 0.05, "learning_rate": 4.973574109961288e-05, "loss": 1.0147, "step": 46418 }, { "epoch": 0.05, "learning_rate": 4.9735729709000646e-05, "loss": 0.9596, "step": 46419 }, { "epoch": 0.05, "learning_rate": 4.973571831814422e-05, "loss": 1.1248, "step": 46420 }, { "epoch": 0.05, "learning_rate": 4.9735706927043614e-05, "loss": 0.9395, "step": 46421 }, { "epoch": 0.05, "learning_rate": 4.973569553569884e-05, "loss": 1.512, "step": 46422 }, { "epoch": 0.05, "learning_rate": 4.9735684144109864e-05, "loss": 1.0922, "step": 46423 }, { "epoch": 0.05, "learning_rate": 4.973567275227672e-05, "loss": 0.9132, "step": 46424 }, { "epoch": 0.05, "learning_rate": 4.97356613601994e-05, "loss": 1.1931, "step": 46425 }, { "epoch": 0.05, "learning_rate": 4.97356499678779e-05, "loss": 0.9477, "step": 46426 }, { "epoch": 0.05, "learning_rate": 4.973563857531221e-05, "loss": 1.3849, "step": 46427 }, { "epoch": 0.05, "learning_rate": 4.973562718250234e-05, "loss": 1.1385, "step": 46428 }, { "epoch": 0.05, "learning_rate": 4.97356157894483e-05, "loss": 0.9256, "step": 46429 }, { "epoch": 0.05, "learning_rate": 4.973560439615008e-05, "loss": 1.1821, "step": 46430 }, { "epoch": 0.05, "learning_rate": 4.9735593002607675e-05, "loss": 0.9724, "step": 46431 }, { "epoch": 0.05, "learning_rate": 4.973558160882109e-05, "loss": 0.7902, "step": 46432 }, { "epoch": 0.05, "learning_rate": 4.973557021479033e-05, "loss": 1.0813, "step": 46433 }, { "epoch": 0.05, "learning_rate": 4.973555882051539e-05, "loss": 1.1857, "step": 46434 }, { "epoch": 0.05, "learning_rate": 4.973554742599627e-05, "loss": 1.2496, "step": 46435 }, { "epoch": 0.05, "learning_rate": 4.9735536031232975e-05, "loss": 0.9935, "step": 46436 }, { "epoch": 0.05, "learning_rate": 4.973552463622549e-05, "loss": 0.853, "step": 46437 }, { "epoch": 0.05, "learning_rate": 4.973551324097384e-05, "loss": 1.1495, "step": 46438 }, { "epoch": 0.05, "learning_rate": 4.9735501845478006e-05, "loss": 1.0196, "step": 46439 }, { "epoch": 0.05, "learning_rate": 4.9735490449737996e-05, "loss": 0.962, "step": 46440 }, { "epoch": 0.05, "learning_rate": 4.9735479053753796e-05, "loss": 1.2145, "step": 46441 }, { "epoch": 0.05, "learning_rate": 4.9735467657525434e-05, "loss": 0.9584, "step": 46442 }, { "epoch": 0.05, "learning_rate": 4.973545626105288e-05, "loss": 1.0392, "step": 46443 }, { "epoch": 0.05, "learning_rate": 4.9735444864336155e-05, "loss": 0.876, "step": 46444 }, { "epoch": 0.05, "learning_rate": 4.973543346737526e-05, "loss": 1.1968, "step": 46445 }, { "epoch": 0.05, "learning_rate": 4.973542207017017e-05, "loss": 1.2865, "step": 46446 }, { "epoch": 0.05, "learning_rate": 4.973541067272092e-05, "loss": 1.1246, "step": 46447 }, { "epoch": 0.05, "learning_rate": 4.973539927502748e-05, "loss": 0.9601, "step": 46448 }, { "epoch": 0.05, "learning_rate": 4.973538787708987e-05, "loss": 0.2531, "step": 46449 }, { "epoch": 0.05, "learning_rate": 4.973537647890807e-05, "loss": 0.5886, "step": 46450 }, { "epoch": 0.05, "learning_rate": 4.9735365080482105e-05, "loss": 1.1662, "step": 46451 }, { "epoch": 0.05, "learning_rate": 4.973535368181196e-05, "loss": 0.8357, "step": 46452 }, { "epoch": 0.05, "learning_rate": 4.973534228289763e-05, "loss": 1.3111, "step": 46453 }, { "epoch": 0.05, "learning_rate": 4.973533088373914e-05, "loss": 1.1791, "step": 46454 }, { "epoch": 0.05, "learning_rate": 4.9735319484336465e-05, "loss": 0.9882, "step": 46455 }, { "epoch": 0.05, "learning_rate": 4.9735308084689607e-05, "loss": 0.9216, "step": 46456 }, { "epoch": 0.05, "learning_rate": 4.973529668479858e-05, "loss": 0.9443, "step": 46457 }, { "epoch": 0.05, "learning_rate": 4.973528528466337e-05, "loss": 1.1031, "step": 46458 }, { "epoch": 0.05, "learning_rate": 4.973527388428399e-05, "loss": 0.8972, "step": 46459 }, { "epoch": 0.05, "learning_rate": 4.973526248366043e-05, "loss": 1.3304, "step": 46460 }, { "epoch": 0.05, "learning_rate": 4.97352510827927e-05, "loss": 1.0836, "step": 46461 }, { "epoch": 0.05, "learning_rate": 4.973523968168079e-05, "loss": 1.0194, "step": 46462 }, { "epoch": 0.05, "learning_rate": 4.9735228280324694e-05, "loss": 0.8921, "step": 46463 }, { "epoch": 0.05, "learning_rate": 4.9735216878724436e-05, "loss": 1.0685, "step": 46464 }, { "epoch": 0.05, "learning_rate": 4.9735205476879995e-05, "loss": 1.0872, "step": 46465 }, { "epoch": 0.05, "learning_rate": 4.973519407479139e-05, "loss": 1.08, "step": 46466 }, { "epoch": 0.05, "learning_rate": 4.9735182672458605e-05, "loss": 1.1265, "step": 46467 }, { "epoch": 0.05, "learning_rate": 4.973517126988163e-05, "loss": 1.2189, "step": 46468 }, { "epoch": 0.05, "learning_rate": 4.97351598670605e-05, "loss": 0.5707, "step": 46469 }, { "epoch": 0.05, "learning_rate": 4.9735148463995184e-05, "loss": 0.7611, "step": 46470 }, { "epoch": 0.05, "learning_rate": 4.9735137060685695e-05, "loss": 2.3374, "step": 46471 }, { "epoch": 0.05, "learning_rate": 4.973512565713203e-05, "loss": 5.8022, "step": 46472 }, { "epoch": 0.05, "learning_rate": 4.9735114253334194e-05, "loss": 5.7224, "step": 46473 }, { "epoch": 0.05, "learning_rate": 4.973510284929218e-05, "loss": 5.7382, "step": 46474 }, { "epoch": 0.05, "learning_rate": 4.973509144500599e-05, "loss": 5.6999, "step": 46475 }, { "epoch": 0.05, "learning_rate": 4.973508004047562e-05, "loss": 5.7315, "step": 46476 }, { "epoch": 0.05, "learning_rate": 4.973506863570109e-05, "loss": 5.6639, "step": 46477 }, { "epoch": 0.05, "learning_rate": 4.973505723068238e-05, "loss": 5.7273, "step": 46478 }, { "epoch": 0.05, "learning_rate": 4.973504582541949e-05, "loss": 5.5088, "step": 46479 }, { "epoch": 0.05, "learning_rate": 4.9735034419912434e-05, "loss": 4.8965, "step": 46480 }, { "epoch": 0.05, "learning_rate": 4.9735023014161205e-05, "loss": 2.1468, "step": 46481 }, { "epoch": 0.05, "learning_rate": 4.9735011608165794e-05, "loss": 1.1111, "step": 46482 }, { "epoch": 0.05, "learning_rate": 4.973500020192622e-05, "loss": 1.0256, "step": 46483 }, { "epoch": 0.05, "learning_rate": 4.973498879544246e-05, "loss": 1.0077, "step": 46484 }, { "epoch": 0.05, "learning_rate": 4.973497738871454e-05, "loss": 1.1965, "step": 46485 }, { "epoch": 0.05, "learning_rate": 4.973496598174244e-05, "loss": 1.1001, "step": 46486 }, { "epoch": 0.05, "learning_rate": 4.973495457452616e-05, "loss": 1.3605, "step": 46487 }, { "epoch": 0.05, "learning_rate": 4.973494316706572e-05, "loss": 1.4458, "step": 46488 }, { "epoch": 0.05, "learning_rate": 4.9734931759361095e-05, "loss": 1.4362, "step": 46489 }, { "epoch": 0.05, "learning_rate": 4.9734920351412305e-05, "loss": 1.1518, "step": 46490 }, { "epoch": 0.05, "learning_rate": 4.973490894321934e-05, "loss": 0.9296, "step": 46491 }, { "epoch": 0.05, "learning_rate": 4.97348975347822e-05, "loss": 0.9329, "step": 46492 }, { "epoch": 0.05, "learning_rate": 4.973488612610089e-05, "loss": 1.0036, "step": 46493 }, { "epoch": 0.05, "learning_rate": 4.973487471717541e-05, "loss": 0.6148, "step": 46494 }, { "epoch": 0.05, "learning_rate": 4.973486330800576e-05, "loss": 1.0302, "step": 46495 }, { "epoch": 0.05, "learning_rate": 4.973485189859193e-05, "loss": 0.752, "step": 46496 }, { "epoch": 0.05, "learning_rate": 4.973484048893393e-05, "loss": 1.0577, "step": 46497 }, { "epoch": 0.05, "learning_rate": 4.973482907903176e-05, "loss": 1.2112, "step": 46498 }, { "epoch": 0.05, "learning_rate": 4.973481766888542e-05, "loss": 1.6273, "step": 46499 }, { "epoch": 0.05, "learning_rate": 4.9734806258494906e-05, "loss": 1.4102, "step": 46500 }, { "epoch": 0.05, "eval_loss": 1.0785138607025146, "eval_runtime": 129.0337, "eval_samples_per_second": 10.734, "eval_steps_per_second": 5.371, "step": 46500 }, { "epoch": 0.05, "learning_rate": 4.973479484786022e-05, "loss": 1.0806, "step": 46501 }, { "epoch": 0.05, "learning_rate": 4.973478343698136e-05, "loss": 0.8284, "step": 46502 }, { "epoch": 0.05, "learning_rate": 4.973477202585833e-05, "loss": 1.0333, "step": 46503 }, { "epoch": 0.05, "learning_rate": 4.973476061449113e-05, "loss": 1.0965, "step": 46504 }, { "epoch": 0.05, "learning_rate": 4.973474920287976e-05, "loss": 1.0654, "step": 46505 }, { "epoch": 0.05, "learning_rate": 4.973473779102422e-05, "loss": 1.0715, "step": 46506 }, { "epoch": 0.05, "learning_rate": 4.973472637892451e-05, "loss": 0.9679, "step": 46507 }, { "epoch": 0.05, "learning_rate": 4.973471496658062e-05, "loss": 1.3209, "step": 46508 }, { "epoch": 0.05, "learning_rate": 4.9734703553992566e-05, "loss": 0.7037, "step": 46509 }, { "epoch": 0.05, "learning_rate": 4.9734692141160346e-05, "loss": 0.913, "step": 46510 }, { "epoch": 0.05, "learning_rate": 4.973468072808395e-05, "loss": 1.0524, "step": 46511 }, { "epoch": 0.05, "learning_rate": 4.9734669314763385e-05, "loss": 0.9175, "step": 46512 }, { "epoch": 0.05, "learning_rate": 4.973465790119864e-05, "loss": 1.3814, "step": 46513 }, { "epoch": 0.05, "learning_rate": 4.973464648738974e-05, "loss": 0.942, "step": 46514 }, { "epoch": 0.05, "learning_rate": 4.973463507333667e-05, "loss": 0.7225, "step": 46515 }, { "epoch": 0.05, "learning_rate": 4.973462365903942e-05, "loss": 0.9198, "step": 46516 }, { "epoch": 0.05, "learning_rate": 4.9734612244498e-05, "loss": 0.8454, "step": 46517 }, { "epoch": 0.05, "learning_rate": 4.973460082971242e-05, "loss": 1.4438, "step": 46518 }, { "epoch": 0.05, "learning_rate": 4.9734589414682665e-05, "loss": 1.0214, "step": 46519 }, { "epoch": 0.05, "learning_rate": 4.973457799940874e-05, "loss": 0.9507, "step": 46520 }, { "epoch": 0.05, "learning_rate": 4.9734566583890646e-05, "loss": 1.1482, "step": 46521 }, { "epoch": 0.05, "learning_rate": 4.973455516812838e-05, "loss": 0.9526, "step": 46522 }, { "epoch": 0.05, "learning_rate": 4.973454375212195e-05, "loss": 1.0082, "step": 46523 }, { "epoch": 0.05, "learning_rate": 4.9734532335871356e-05, "loss": 0.7951, "step": 46524 }, { "epoch": 0.05, "learning_rate": 4.973452091937658e-05, "loss": 0.9545, "step": 46525 }, { "epoch": 0.05, "learning_rate": 4.973450950263764e-05, "loss": 0.5459, "step": 46526 }, { "epoch": 0.05, "learning_rate": 4.973449808565454e-05, "loss": 0.5543, "step": 46527 }, { "epoch": 0.05, "learning_rate": 4.973448666842726e-05, "loss": 1.1954, "step": 46528 }, { "epoch": 0.05, "learning_rate": 4.973447525095581e-05, "loss": 1.0337, "step": 46529 }, { "epoch": 0.05, "learning_rate": 4.97344638332402e-05, "loss": 0.81, "step": 46530 }, { "epoch": 0.05, "learning_rate": 4.973445241528042e-05, "loss": 0.882, "step": 46531 }, { "epoch": 0.05, "learning_rate": 4.973444099707647e-05, "loss": 0.8892, "step": 46532 }, { "epoch": 0.05, "learning_rate": 4.9734429578628354e-05, "loss": 0.5623, "step": 46533 }, { "epoch": 0.05, "learning_rate": 4.973441815993607e-05, "loss": 2.6433, "step": 46534 }, { "epoch": 0.05, "learning_rate": 4.973440674099962e-05, "loss": 5.7738, "step": 46535 }, { "epoch": 0.05, "learning_rate": 4.9734395321819e-05, "loss": 4.7702, "step": 46536 }, { "epoch": 0.05, "learning_rate": 4.973438390239421e-05, "loss": 5.2295, "step": 46537 }, { "epoch": 0.05, "learning_rate": 4.9734372482725255e-05, "loss": 5.3617, "step": 46538 }, { "epoch": 0.05, "learning_rate": 4.973436106281213e-05, "loss": 5.4507, "step": 46539 }, { "epoch": 0.05, "learning_rate": 4.9734349642654844e-05, "loss": 5.7327, "step": 46540 }, { "epoch": 0.05, "learning_rate": 4.973433822225339e-05, "loss": 5.4235, "step": 46541 }, { "epoch": 0.05, "learning_rate": 4.973432680160776e-05, "loss": 5.3958, "step": 46542 }, { "epoch": 0.05, "learning_rate": 4.973431538071797e-05, "loss": 4.9642, "step": 46543 }, { "epoch": 0.05, "learning_rate": 4.973430395958402e-05, "loss": 5.0323, "step": 46544 }, { "epoch": 0.05, "learning_rate": 4.973429253820589e-05, "loss": 5.3324, "step": 46545 }, { "epoch": 0.05, "learning_rate": 4.9734281116583606e-05, "loss": 5.2316, "step": 46546 }, { "epoch": 0.05, "learning_rate": 4.973426969471714e-05, "loss": 5.7913, "step": 46547 }, { "epoch": 0.05, "learning_rate": 4.973425827260652e-05, "loss": 1.9559, "step": 46548 }, { "epoch": 0.05, "learning_rate": 4.973424685025173e-05, "loss": 1.2, "step": 46549 }, { "epoch": 0.05, "learning_rate": 4.973423542765278e-05, "loss": 0.7442, "step": 46550 }, { "epoch": 0.05, "learning_rate": 4.973422400480965e-05, "loss": 0.8795, "step": 46551 }, { "epoch": 0.05, "learning_rate": 4.973421258172237e-05, "loss": 1.1944, "step": 46552 }, { "epoch": 0.05, "learning_rate": 4.973420115839091e-05, "loss": 0.9698, "step": 46553 }, { "epoch": 0.05, "learning_rate": 4.97341897348153e-05, "loss": 1.0681, "step": 46554 }, { "epoch": 0.05, "learning_rate": 4.9734178310995515e-05, "loss": 1.316, "step": 46555 }, { "epoch": 0.05, "learning_rate": 4.973416688693156e-05, "loss": 1.1466, "step": 46556 }, { "epoch": 0.05, "learning_rate": 4.973415546262345e-05, "loss": 0.9823, "step": 46557 }, { "epoch": 0.05, "learning_rate": 4.973414403807117e-05, "loss": 0.8277, "step": 46558 }, { "epoch": 0.05, "learning_rate": 4.973413261327473e-05, "loss": 0.8502, "step": 46559 }, { "epoch": 0.05, "learning_rate": 4.973412118823412e-05, "loss": 0.9019, "step": 46560 }, { "epoch": 0.05, "learning_rate": 4.973410976294934e-05, "loss": 0.9357, "step": 46561 }, { "epoch": 0.05, "learning_rate": 4.9734098337420396e-05, "loss": 0.9232, "step": 46562 }, { "epoch": 0.05, "learning_rate": 4.9734086911647294e-05, "loss": 1.1242, "step": 46563 }, { "epoch": 0.05, "learning_rate": 4.973407548563003e-05, "loss": 0.9862, "step": 46564 }, { "epoch": 0.05, "learning_rate": 4.97340640593686e-05, "loss": 1.1819, "step": 46565 }, { "epoch": 0.05, "learning_rate": 4.9734052632863e-05, "loss": 1.1612, "step": 46566 }, { "epoch": 0.05, "learning_rate": 4.973404120611324e-05, "loss": 0.9152, "step": 46567 }, { "epoch": 0.05, "learning_rate": 4.973402977911931e-05, "loss": 0.9735, "step": 46568 }, { "epoch": 0.05, "learning_rate": 4.973401835188122e-05, "loss": 1.3106, "step": 46569 }, { "epoch": 0.05, "learning_rate": 4.973400692439897e-05, "loss": 1.4258, "step": 46570 }, { "epoch": 0.05, "learning_rate": 4.9733995496672556e-05, "loss": 0.9391, "step": 46571 }, { "epoch": 0.05, "learning_rate": 4.973398406870198e-05, "loss": 1.1377, "step": 46572 }, { "epoch": 0.05, "learning_rate": 4.973397264048723e-05, "loss": 1.0024, "step": 46573 }, { "epoch": 0.05, "learning_rate": 4.973396121202833e-05, "loss": 0.6597, "step": 46574 }, { "epoch": 0.05, "learning_rate": 4.973394978332525e-05, "loss": 0.7872, "step": 46575 }, { "epoch": 0.05, "learning_rate": 4.973393835437802e-05, "loss": 0.8694, "step": 46576 }, { "epoch": 0.05, "learning_rate": 4.9733926925186624e-05, "loss": 0.8812, "step": 46577 }, { "epoch": 0.05, "learning_rate": 4.973391549575107e-05, "loss": 0.393, "step": 46578 }, { "epoch": 0.05, "learning_rate": 4.973390406607135e-05, "loss": 0.2731, "step": 46579 }, { "epoch": 0.05, "learning_rate": 4.973389263614746e-05, "loss": 0.23, "step": 46580 }, { "epoch": 0.05, "learning_rate": 4.973388120597941e-05, "loss": 0.1797, "step": 46581 }, { "epoch": 0.05, "learning_rate": 4.97338697755672e-05, "loss": 0.2259, "step": 46582 }, { "epoch": 0.05, "learning_rate": 4.973385834491083e-05, "loss": 0.2549, "step": 46583 }, { "epoch": 0.05, "learning_rate": 4.9733846914010296e-05, "loss": 0.2244, "step": 46584 }, { "epoch": 0.05, "learning_rate": 4.9733835482865596e-05, "loss": 0.2502, "step": 46585 }, { "epoch": 0.05, "learning_rate": 4.973382405147674e-05, "loss": 0.1483, "step": 46586 }, { "epoch": 0.05, "learning_rate": 4.973381261984372e-05, "loss": 0.3304, "step": 46587 }, { "epoch": 0.05, "learning_rate": 4.9733801187966536e-05, "loss": 0.4186, "step": 46588 }, { "epoch": 0.05, "learning_rate": 4.973378975584519e-05, "loss": 0.2229, "step": 46589 }, { "epoch": 0.05, "learning_rate": 4.973377832347969e-05, "loss": 0.2258, "step": 46590 }, { "epoch": 0.05, "learning_rate": 4.973376689087002e-05, "loss": 0.3232, "step": 46591 }, { "epoch": 0.05, "learning_rate": 4.97337554580162e-05, "loss": 0.3122, "step": 46592 }, { "epoch": 0.05, "learning_rate": 4.9733744024918207e-05, "loss": 0.3181, "step": 46593 }, { "epoch": 0.05, "learning_rate": 4.973373259157605e-05, "loss": 0.4033, "step": 46594 }, { "epoch": 0.05, "learning_rate": 4.973372115798974e-05, "loss": 0.2973, "step": 46595 }, { "epoch": 0.05, "learning_rate": 4.973370972415927e-05, "loss": 0.226, "step": 46596 }, { "epoch": 0.05, "learning_rate": 4.973369829008464e-05, "loss": 0.2062, "step": 46597 }, { "epoch": 0.05, "learning_rate": 4.9733686855765846e-05, "loss": 0.7977, "step": 46598 }, { "epoch": 0.05, "learning_rate": 4.973367542120289e-05, "loss": 0.874, "step": 46599 }, { "epoch": 0.05, "learning_rate": 4.973366398639577e-05, "loss": 1.0736, "step": 46600 }, { "epoch": 0.05, "learning_rate": 4.97336525513445e-05, "loss": 1.0152, "step": 46601 }, { "epoch": 0.05, "learning_rate": 4.973364111604907e-05, "loss": 1.0549, "step": 46602 }, { "epoch": 0.05, "learning_rate": 4.973362968050947e-05, "loss": 0.8208, "step": 46603 }, { "epoch": 0.05, "learning_rate": 4.973361824472572e-05, "loss": 0.7641, "step": 46604 }, { "epoch": 0.05, "learning_rate": 4.97336068086978e-05, "loss": 0.5105, "step": 46605 }, { "epoch": 0.05, "learning_rate": 4.9733595372425724e-05, "loss": 0.6064, "step": 46606 }, { "epoch": 0.05, "learning_rate": 4.9733583935909486e-05, "loss": 0.9666, "step": 46607 }, { "epoch": 0.05, "learning_rate": 4.9733572499149094e-05, "loss": 1.0368, "step": 46608 }, { "epoch": 0.05, "learning_rate": 4.973356106214454e-05, "loss": 1.3611, "step": 46609 }, { "epoch": 0.05, "learning_rate": 4.9733549624895834e-05, "loss": 1.0411, "step": 46610 }, { "epoch": 0.05, "learning_rate": 4.973353818740296e-05, "loss": 1.0885, "step": 46611 }, { "epoch": 0.05, "learning_rate": 4.9733526749665926e-05, "loss": 1.3389, "step": 46612 }, { "epoch": 0.05, "learning_rate": 4.973351531168474e-05, "loss": 1.9099, "step": 46613 }, { "epoch": 0.05, "learning_rate": 4.973350387345939e-05, "loss": 1.3757, "step": 46614 }, { "epoch": 0.05, "learning_rate": 4.973349243498988e-05, "loss": 1.3561, "step": 46615 }, { "epoch": 0.05, "learning_rate": 4.9733480996276215e-05, "loss": 1.1844, "step": 46616 }, { "epoch": 0.05, "learning_rate": 4.973346955731839e-05, "loss": 0.8844, "step": 46617 }, { "epoch": 0.05, "learning_rate": 4.9733458118116406e-05, "loss": 0.8704, "step": 46618 }, { "epoch": 0.05, "learning_rate": 4.973344667867027e-05, "loss": 0.9211, "step": 46619 }, { "epoch": 0.05, "learning_rate": 4.973343523897997e-05, "loss": 1.1064, "step": 46620 }, { "epoch": 0.05, "learning_rate": 4.973342379904551e-05, "loss": 1.4081, "step": 46621 }, { "epoch": 0.05, "learning_rate": 4.9733412358866894e-05, "loss": 1.1625, "step": 46622 }, { "epoch": 0.05, "learning_rate": 4.973340091844412e-05, "loss": 1.0208, "step": 46623 }, { "epoch": 0.05, "learning_rate": 4.973338947777719e-05, "loss": 0.966, "step": 46624 }, { "epoch": 0.05, "learning_rate": 4.9733378036866107e-05, "loss": 0.7655, "step": 46625 }, { "epoch": 0.05, "learning_rate": 4.973336659571086e-05, "loss": 0.1796, "step": 46626 }, { "epoch": 0.05, "learning_rate": 4.973335515431146e-05, "loss": 0.2652, "step": 46627 }, { "epoch": 0.05, "learning_rate": 4.973334371266789e-05, "loss": 0.8616, "step": 46628 }, { "epoch": 0.05, "learning_rate": 4.973333227078018e-05, "loss": 1.3758, "step": 46629 }, { "epoch": 0.05, "learning_rate": 4.97333208286483e-05, "loss": 0.9558, "step": 46630 }, { "epoch": 0.05, "learning_rate": 4.973330938627227e-05, "loss": 0.9247, "step": 46631 }, { "epoch": 0.05, "learning_rate": 4.973329794365208e-05, "loss": 1.2246, "step": 46632 }, { "epoch": 0.05, "learning_rate": 4.9733286500787734e-05, "loss": 1.1295, "step": 46633 }, { "epoch": 0.05, "learning_rate": 4.9733275057679236e-05, "loss": 1.0887, "step": 46634 }, { "epoch": 0.05, "learning_rate": 4.9733263614326575e-05, "loss": 1.309, "step": 46635 }, { "epoch": 0.05, "learning_rate": 4.973325217072976e-05, "loss": 1.0508, "step": 46636 }, { "epoch": 0.05, "learning_rate": 4.9733240726888794e-05, "loss": 0.9934, "step": 46637 }, { "epoch": 0.05, "learning_rate": 4.973322928280366e-05, "loss": 0.7807, "step": 46638 }, { "epoch": 0.05, "learning_rate": 4.973321783847438e-05, "loss": 0.6085, "step": 46639 }, { "epoch": 0.05, "learning_rate": 4.9733206393900946e-05, "loss": 0.6529, "step": 46640 }, { "epoch": 0.05, "learning_rate": 4.973319494908335e-05, "loss": 0.1293, "step": 46641 }, { "epoch": 0.05, "learning_rate": 4.9733183504021596e-05, "loss": 0.2818, "step": 46642 }, { "epoch": 0.05, "learning_rate": 4.973317205871569e-05, "loss": 0.6572, "step": 46643 }, { "epoch": 0.05, "learning_rate": 4.973316061316563e-05, "loss": 1.1081, "step": 46644 }, { "epoch": 0.05, "learning_rate": 4.9733149167371414e-05, "loss": 1.3347, "step": 46645 }, { "epoch": 0.05, "learning_rate": 4.973313772133304e-05, "loss": 1.342, "step": 46646 }, { "epoch": 0.05, "learning_rate": 4.9733126275050515e-05, "loss": 1.0764, "step": 46647 }, { "epoch": 0.05, "learning_rate": 4.973311482852383e-05, "loss": 0.8538, "step": 46648 }, { "epoch": 0.05, "learning_rate": 4.9733103381753e-05, "loss": 0.4709, "step": 46649 }, { "epoch": 0.05, "learning_rate": 4.973309193473801e-05, "loss": 0.2897, "step": 46650 }, { "epoch": 0.05, "learning_rate": 4.973308048747886e-05, "loss": 0.613, "step": 46651 }, { "epoch": 0.05, "learning_rate": 4.973306903997555e-05, "loss": 0.6245, "step": 46652 }, { "epoch": 0.05, "learning_rate": 4.9733057592228105e-05, "loss": 0.7232, "step": 46653 }, { "epoch": 0.05, "learning_rate": 4.9733046144236496e-05, "loss": 0.54, "step": 46654 }, { "epoch": 0.05, "learning_rate": 4.9733034696000724e-05, "loss": 0.6075, "step": 46655 }, { "epoch": 0.05, "learning_rate": 4.973302324752081e-05, "loss": 0.738, "step": 46656 }, { "epoch": 0.05, "learning_rate": 4.9733011798796735e-05, "loss": 1.2266, "step": 46657 }, { "epoch": 0.05, "learning_rate": 4.973300034982851e-05, "loss": 1.1684, "step": 46658 }, { "epoch": 0.05, "learning_rate": 4.973298890061613e-05, "loss": 0.9949, "step": 46659 }, { "epoch": 0.05, "learning_rate": 4.97329774511596e-05, "loss": 1.056, "step": 46660 }, { "epoch": 0.05, "learning_rate": 4.9732966001458916e-05, "loss": 0.8461, "step": 46661 }, { "epoch": 0.05, "learning_rate": 4.973295455151407e-05, "loss": 0.8546, "step": 46662 }, { "epoch": 0.05, "learning_rate": 4.973294310132508e-05, "loss": 1.0671, "step": 46663 }, { "epoch": 0.05, "learning_rate": 4.9732931650891936e-05, "loss": 0.9849, "step": 46664 }, { "epoch": 0.05, "learning_rate": 4.9732920200214636e-05, "loss": 0.9218, "step": 46665 }, { "epoch": 0.05, "learning_rate": 4.973290874929319e-05, "loss": 1.2051, "step": 46666 }, { "epoch": 0.05, "learning_rate": 4.9732897298127576e-05, "loss": 1.1887, "step": 46667 }, { "epoch": 0.05, "learning_rate": 4.9732885846717816e-05, "loss": 0.8639, "step": 46668 }, { "epoch": 0.05, "learning_rate": 4.973287439506391e-05, "loss": 1.06, "step": 46669 }, { "epoch": 0.05, "learning_rate": 4.973286294316584e-05, "loss": 0.99, "step": 46670 }, { "epoch": 0.05, "learning_rate": 4.973285149102362e-05, "loss": 1.149, "step": 46671 }, { "epoch": 0.05, "learning_rate": 4.9732840038637254e-05, "loss": 1.2756, "step": 46672 }, { "epoch": 0.05, "learning_rate": 4.9732828586006743e-05, "loss": 0.749, "step": 46673 }, { "epoch": 0.05, "learning_rate": 4.9732817133132064e-05, "loss": 1.0265, "step": 46674 }, { "epoch": 0.05, "learning_rate": 4.973280568001324e-05, "loss": 1.1019, "step": 46675 }, { "epoch": 0.05, "learning_rate": 4.9732794226650265e-05, "loss": 0.8911, "step": 46676 }, { "epoch": 0.05, "learning_rate": 4.973278277304313e-05, "loss": 1.0674, "step": 46677 }, { "epoch": 0.05, "learning_rate": 4.973277131919186e-05, "loss": 0.9934, "step": 46678 }, { "epoch": 0.05, "learning_rate": 4.973275986509642e-05, "loss": 0.9585, "step": 46679 }, { "epoch": 0.05, "learning_rate": 4.973274841075685e-05, "loss": 1.0028, "step": 46680 }, { "epoch": 0.05, "learning_rate": 4.9732736956173106e-05, "loss": 1.1547, "step": 46681 }, { "epoch": 0.05, "learning_rate": 4.973272550134522e-05, "loss": 0.8946, "step": 46682 }, { "epoch": 0.05, "learning_rate": 4.973271404627319e-05, "loss": 1.135, "step": 46683 }, { "epoch": 0.05, "learning_rate": 4.9732702590957e-05, "loss": 0.9844, "step": 46684 }, { "epoch": 0.05, "learning_rate": 4.9732691135396666e-05, "loss": 1.1561, "step": 46685 }, { "epoch": 0.05, "learning_rate": 4.9732679679592174e-05, "loss": 0.8871, "step": 46686 }, { "epoch": 0.05, "learning_rate": 4.973266822354354e-05, "loss": 1.3042, "step": 46687 }, { "epoch": 0.05, "learning_rate": 4.973265676725075e-05, "loss": 0.8676, "step": 46688 }, { "epoch": 0.05, "learning_rate": 4.973264531071381e-05, "loss": 0.8407, "step": 46689 }, { "epoch": 0.05, "learning_rate": 4.973263385393272e-05, "loss": 1.1181, "step": 46690 }, { "epoch": 0.05, "learning_rate": 4.973262239690748e-05, "loss": 0.9836, "step": 46691 }, { "epoch": 0.05, "learning_rate": 4.973261093963809e-05, "loss": 1.1248, "step": 46692 }, { "epoch": 0.05, "learning_rate": 4.973259948212455e-05, "loss": 1.1306, "step": 46693 }, { "epoch": 0.05, "learning_rate": 4.9732588024366854e-05, "loss": 0.6313, "step": 46694 }, { "epoch": 0.05, "learning_rate": 4.9732576566365016e-05, "loss": 1.1517, "step": 46695 }, { "epoch": 0.05, "learning_rate": 4.9732565108119036e-05, "loss": 1.097, "step": 46696 }, { "epoch": 0.05, "learning_rate": 4.973255364962889e-05, "loss": 1.0944, "step": 46697 }, { "epoch": 0.05, "learning_rate": 4.97325421908946e-05, "loss": 1.1806, "step": 46698 }, { "epoch": 0.05, "learning_rate": 4.973253073191616e-05, "loss": 1.1391, "step": 46699 }, { "epoch": 0.05, "learning_rate": 4.973251927269358e-05, "loss": 0.7181, "step": 46700 }, { "epoch": 0.05, "learning_rate": 4.973250781322685e-05, "loss": 0.7595, "step": 46701 }, { "epoch": 0.05, "learning_rate": 4.9732496353515964e-05, "loss": 0.7945, "step": 46702 }, { "epoch": 0.05, "learning_rate": 4.973248489356093e-05, "loss": 0.2656, "step": 46703 }, { "epoch": 0.05, "learning_rate": 4.9732473433361746e-05, "loss": 0.8799, "step": 46704 }, { "epoch": 0.05, "learning_rate": 4.973246197291841e-05, "loss": 1.8386, "step": 46705 }, { "epoch": 0.05, "learning_rate": 4.973245051223094e-05, "loss": 1.0615, "step": 46706 }, { "epoch": 0.05, "learning_rate": 4.973243905129931e-05, "loss": 1.0302, "step": 46707 }, { "epoch": 0.05, "learning_rate": 4.973242759012353e-05, "loss": 1.0769, "step": 46708 }, { "epoch": 0.05, "learning_rate": 4.9732416128703616e-05, "loss": 0.7702, "step": 46709 }, { "epoch": 0.05, "learning_rate": 4.973240466703954e-05, "loss": 1.0576, "step": 46710 }, { "epoch": 0.05, "learning_rate": 4.973239320513132e-05, "loss": 0.8687, "step": 46711 }, { "epoch": 0.05, "learning_rate": 4.973238174297895e-05, "loss": 0.883, "step": 46712 }, { "epoch": 0.05, "learning_rate": 4.973237028058244e-05, "loss": 1.1978, "step": 46713 }, { "epoch": 0.05, "learning_rate": 4.9732358817941774e-05, "loss": 1.0713, "step": 46714 }, { "epoch": 0.05, "learning_rate": 4.973234735505697e-05, "loss": 0.8373, "step": 46715 }, { "epoch": 0.05, "learning_rate": 4.973233589192801e-05, "loss": 0.7134, "step": 46716 }, { "epoch": 0.05, "learning_rate": 4.973232442855491e-05, "loss": 1.1786, "step": 46717 }, { "epoch": 0.05, "learning_rate": 4.9732312964937654e-05, "loss": 0.8498, "step": 46718 }, { "epoch": 0.05, "learning_rate": 4.973230150107626e-05, "loss": 1.0504, "step": 46719 }, { "epoch": 0.05, "learning_rate": 4.9732290036970716e-05, "loss": 1.139, "step": 46720 }, { "epoch": 0.05, "learning_rate": 4.973227857262102e-05, "loss": 0.9855, "step": 46721 }, { "epoch": 0.05, "learning_rate": 4.973226710802719e-05, "loss": 1.2741, "step": 46722 }, { "epoch": 0.05, "learning_rate": 4.97322556431892e-05, "loss": 0.961, "step": 46723 }, { "epoch": 0.05, "learning_rate": 4.9732244178107066e-05, "loss": 1.1023, "step": 46724 }, { "epoch": 0.05, "learning_rate": 4.973223271278079e-05, "loss": 0.7815, "step": 46725 }, { "epoch": 0.05, "learning_rate": 4.973222124721036e-05, "loss": 0.9996, "step": 46726 }, { "epoch": 0.05, "learning_rate": 4.97322097813958e-05, "loss": 0.9157, "step": 46727 }, { "epoch": 0.05, "learning_rate": 4.9732198315337084e-05, "loss": 1.1413, "step": 46728 }, { "epoch": 0.05, "learning_rate": 4.9732186849034215e-05, "loss": 1.0296, "step": 46729 }, { "epoch": 0.05, "learning_rate": 4.973217538248721e-05, "loss": 0.9206, "step": 46730 }, { "epoch": 0.05, "learning_rate": 4.9732163915696064e-05, "loss": 1.1169, "step": 46731 }, { "epoch": 0.05, "learning_rate": 4.973215244866076e-05, "loss": 0.731, "step": 46732 }, { "epoch": 0.05, "learning_rate": 4.973214098138131e-05, "loss": 0.42, "step": 46733 }, { "epoch": 0.05, "learning_rate": 4.9732129513857725e-05, "loss": 0.3542, "step": 46734 }, { "epoch": 0.05, "learning_rate": 4.973211804608999e-05, "loss": 1.0023, "step": 46735 }, { "epoch": 0.05, "learning_rate": 4.9732106578078114e-05, "loss": 1.1668, "step": 46736 }, { "epoch": 0.05, "learning_rate": 4.973209510982209e-05, "loss": 0.9243, "step": 46737 }, { "epoch": 0.05, "learning_rate": 4.973208364132192e-05, "loss": 1.1761, "step": 46738 }, { "epoch": 0.05, "learning_rate": 4.9732072172577604e-05, "loss": 0.6972, "step": 46739 }, { "epoch": 0.05, "learning_rate": 4.9732060703589146e-05, "loss": 1.0578, "step": 46740 }, { "epoch": 0.05, "learning_rate": 4.9732049234356546e-05, "loss": 1.3072, "step": 46741 }, { "epoch": 0.05, "learning_rate": 4.9732037764879796e-05, "loss": 1.5675, "step": 46742 }, { "epoch": 0.05, "learning_rate": 4.9732026295158905e-05, "loss": 0.9957, "step": 46743 }, { "epoch": 0.05, "learning_rate": 4.973201482519387e-05, "loss": 1.2489, "step": 46744 }, { "epoch": 0.05, "learning_rate": 4.973200335498469e-05, "loss": 0.8307, "step": 46745 }, { "epoch": 0.05, "learning_rate": 4.973199188453137e-05, "loss": 0.8556, "step": 46746 }, { "epoch": 0.05, "learning_rate": 4.97319804138339e-05, "loss": 1.0933, "step": 46747 }, { "epoch": 0.05, "learning_rate": 4.973196894289229e-05, "loss": 1.0209, "step": 46748 }, { "epoch": 0.05, "learning_rate": 4.9731957471706536e-05, "loss": 1.1098, "step": 46749 }, { "epoch": 0.05, "learning_rate": 4.973194600027664e-05, "loss": 0.9708, "step": 46750 }, { "epoch": 0.05, "learning_rate": 4.97319345286026e-05, "loss": 1.1982, "step": 46751 }, { "epoch": 0.05, "learning_rate": 4.973192305668441e-05, "loss": 1.0341, "step": 46752 }, { "epoch": 0.05, "learning_rate": 4.973191158452208e-05, "loss": 1.0286, "step": 46753 }, { "epoch": 0.05, "learning_rate": 4.9731900112115616e-05, "loss": 0.9246, "step": 46754 }, { "epoch": 0.05, "learning_rate": 4.9731888639464994e-05, "loss": 0.4614, "step": 46755 }, { "epoch": 0.05, "learning_rate": 4.973187716657024e-05, "loss": 0.5219, "step": 46756 }, { "epoch": 0.05, "learning_rate": 4.9731865693431344e-05, "loss": 0.6697, "step": 46757 }, { "epoch": 0.05, "learning_rate": 4.97318542200483e-05, "loss": 0.3688, "step": 46758 }, { "epoch": 0.05, "learning_rate": 4.973184274642112e-05, "loss": 0.5242, "step": 46759 }, { "epoch": 0.05, "learning_rate": 4.9731831272549787e-05, "loss": 0.5399, "step": 46760 }, { "epoch": 0.05, "learning_rate": 4.973181979843432e-05, "loss": 0.7789, "step": 46761 }, { "epoch": 0.05, "learning_rate": 4.9731808324074717e-05, "loss": 0.4979, "step": 46762 }, { "epoch": 0.05, "learning_rate": 4.973179684947096e-05, "loss": 0.4754, "step": 46763 }, { "epoch": 0.05, "learning_rate": 4.9731785374623065e-05, "loss": 0.2627, "step": 46764 }, { "epoch": 0.05, "learning_rate": 4.973177389953102e-05, "loss": 0.1803, "step": 46765 }, { "epoch": 0.05, "learning_rate": 4.973176242419485e-05, "loss": 0.2866, "step": 46766 }, { "epoch": 0.05, "learning_rate": 4.973175094861453e-05, "loss": 0.6873, "step": 46767 }, { "epoch": 0.05, "learning_rate": 4.9731739472790064e-05, "loss": 1.1016, "step": 46768 }, { "epoch": 0.05, "learning_rate": 4.973172799672147e-05, "loss": 1.3152, "step": 46769 }, { "epoch": 0.05, "learning_rate": 4.973171652040872e-05, "loss": 1.0043, "step": 46770 }, { "epoch": 0.05, "learning_rate": 4.973170504385184e-05, "loss": 0.9972, "step": 46771 }, { "epoch": 0.05, "learning_rate": 4.9731693567050805e-05, "loss": 1.0042, "step": 46772 }, { "epoch": 0.05, "learning_rate": 4.973168209000565e-05, "loss": 1.4145, "step": 46773 }, { "epoch": 0.05, "learning_rate": 4.973167061271634e-05, "loss": 1.0306, "step": 46774 }, { "epoch": 0.05, "learning_rate": 4.973165913518289e-05, "loss": 0.7173, "step": 46775 }, { "epoch": 0.05, "learning_rate": 4.97316476574053e-05, "loss": 1.0803, "step": 46776 }, { "epoch": 0.05, "learning_rate": 4.9731636179383576e-05, "loss": 1.1117, "step": 46777 }, { "epoch": 0.05, "learning_rate": 4.973162470111771e-05, "loss": 0.6644, "step": 46778 }, { "epoch": 0.05, "learning_rate": 4.97316132226077e-05, "loss": 1.0483, "step": 46779 }, { "epoch": 0.05, "learning_rate": 4.9731601743853546e-05, "loss": 0.7209, "step": 46780 }, { "epoch": 0.05, "learning_rate": 4.973159026485525e-05, "loss": 0.7714, "step": 46781 }, { "epoch": 0.05, "learning_rate": 4.973157878561283e-05, "loss": 0.9761, "step": 46782 }, { "epoch": 0.05, "learning_rate": 4.973156730612626e-05, "loss": 1.1103, "step": 46783 }, { "epoch": 0.05, "learning_rate": 4.9731555826395545e-05, "loss": 1.0433, "step": 46784 }, { "epoch": 0.05, "learning_rate": 4.97315443464207e-05, "loss": 1.3227, "step": 46785 }, { "epoch": 0.05, "learning_rate": 4.9731532866201713e-05, "loss": 0.8871, "step": 46786 }, { "epoch": 0.05, "learning_rate": 4.973152138573859e-05, "loss": 0.8551, "step": 46787 }, { "epoch": 0.05, "learning_rate": 4.973150990503133e-05, "loss": 1.1944, "step": 46788 }, { "epoch": 0.05, "learning_rate": 4.973149842407992e-05, "loss": 1.0882, "step": 46789 }, { "epoch": 0.05, "learning_rate": 4.973148694288437e-05, "loss": 1.1272, "step": 46790 }, { "epoch": 0.05, "learning_rate": 4.9731475461444686e-05, "loss": 1.2566, "step": 46791 }, { "epoch": 0.05, "learning_rate": 4.973146397976087e-05, "loss": 0.9677, "step": 46792 }, { "epoch": 0.05, "learning_rate": 4.973145249783291e-05, "loss": 0.7219, "step": 46793 }, { "epoch": 0.05, "learning_rate": 4.9731441015660814e-05, "loss": 1.2106, "step": 46794 }, { "epoch": 0.05, "learning_rate": 4.973142953324458e-05, "loss": 0.9041, "step": 46795 }, { "epoch": 0.05, "learning_rate": 4.9731418050584197e-05, "loss": 0.9151, "step": 46796 }, { "epoch": 0.05, "learning_rate": 4.9731406567679685e-05, "loss": 0.9766, "step": 46797 }, { "epoch": 0.05, "learning_rate": 4.973139508453104e-05, "loss": 1.0827, "step": 46798 }, { "epoch": 0.05, "learning_rate": 4.973138360113825e-05, "loss": 0.8827, "step": 46799 }, { "epoch": 0.05, "learning_rate": 4.973137211750132e-05, "loss": 1.2311, "step": 46800 }, { "epoch": 0.05, "learning_rate": 4.973136063362025e-05, "loss": 1.0502, "step": 46801 }, { "epoch": 0.05, "learning_rate": 4.973134914949506e-05, "loss": 0.9703, "step": 46802 }, { "epoch": 0.05, "learning_rate": 4.9731337665125714e-05, "loss": 1.2957, "step": 46803 }, { "epoch": 0.05, "learning_rate": 4.9731326180512235e-05, "loss": 1.0583, "step": 46804 }, { "epoch": 0.05, "learning_rate": 4.973131469565462e-05, "loss": 1.2198, "step": 46805 }, { "epoch": 0.05, "learning_rate": 4.973130321055287e-05, "loss": 1.5094, "step": 46806 }, { "epoch": 0.05, "learning_rate": 4.973129172520699e-05, "loss": 0.9233, "step": 46807 }, { "epoch": 0.05, "learning_rate": 4.973128023961696e-05, "loss": 0.8252, "step": 46808 }, { "epoch": 0.05, "learning_rate": 4.97312687537828e-05, "loss": 0.8277, "step": 46809 }, { "epoch": 0.05, "learning_rate": 4.9731257267704504e-05, "loss": 0.7632, "step": 46810 }, { "epoch": 0.05, "learning_rate": 4.9731245781382065e-05, "loss": 0.957, "step": 46811 }, { "epoch": 0.05, "learning_rate": 4.973123429481549e-05, "loss": 0.9164, "step": 46812 }, { "epoch": 0.05, "learning_rate": 4.973122280800478e-05, "loss": 0.9575, "step": 46813 }, { "epoch": 0.05, "learning_rate": 4.9731211320949944e-05, "loss": 0.9238, "step": 46814 }, { "epoch": 0.05, "learning_rate": 4.973119983365096e-05, "loss": 1.1146, "step": 46815 }, { "epoch": 0.05, "learning_rate": 4.973118834610784e-05, "loss": 1.1634, "step": 46816 }, { "epoch": 0.05, "learning_rate": 4.973117685832059e-05, "loss": 1.3754, "step": 46817 }, { "epoch": 0.05, "learning_rate": 4.97311653702892e-05, "loss": 1.1222, "step": 46818 }, { "epoch": 0.05, "learning_rate": 4.973115388201368e-05, "loss": 1.3254, "step": 46819 }, { "epoch": 0.05, "learning_rate": 4.9731142393494025e-05, "loss": 1.305, "step": 46820 }, { "epoch": 0.05, "learning_rate": 4.973113090473023e-05, "loss": 1.116, "step": 46821 }, { "epoch": 0.05, "learning_rate": 4.97311194157223e-05, "loss": 1.1355, "step": 46822 }, { "epoch": 0.05, "learning_rate": 4.973110792647023e-05, "loss": 0.8888, "step": 46823 }, { "epoch": 0.05, "learning_rate": 4.9731096436974036e-05, "loss": 0.9408, "step": 46824 }, { "epoch": 0.05, "learning_rate": 4.97310849472337e-05, "loss": 0.9845, "step": 46825 }, { "epoch": 0.05, "learning_rate": 4.973107345724923e-05, "loss": 0.8871, "step": 46826 }, { "epoch": 0.05, "learning_rate": 4.973106196702063e-05, "loss": 0.6896, "step": 46827 }, { "epoch": 0.05, "learning_rate": 4.9731050476547884e-05, "loss": 0.6807, "step": 46828 }, { "epoch": 0.05, "learning_rate": 4.973103898583101e-05, "loss": 0.809, "step": 46829 }, { "epoch": 0.05, "learning_rate": 4.973102749487e-05, "loss": 0.585, "step": 46830 }, { "epoch": 0.05, "learning_rate": 4.973101600366487e-05, "loss": 0.9881, "step": 46831 }, { "epoch": 0.05, "learning_rate": 4.973100451221558e-05, "loss": 1.2657, "step": 46832 }, { "epoch": 0.05, "learning_rate": 4.9730993020522175e-05, "loss": 1.0129, "step": 46833 }, { "epoch": 0.05, "learning_rate": 4.973098152858463e-05, "loss": 1.4926, "step": 46834 }, { "epoch": 0.05, "learning_rate": 4.973097003640295e-05, "loss": 1.3421, "step": 46835 }, { "epoch": 0.05, "learning_rate": 4.973095854397714e-05, "loss": 1.1046, "step": 46836 }, { "epoch": 0.05, "learning_rate": 4.9730947051307195e-05, "loss": 0.952, "step": 46837 }, { "epoch": 0.05, "learning_rate": 4.973093555839312e-05, "loss": 1.0169, "step": 46838 }, { "epoch": 0.05, "learning_rate": 4.97309240652349e-05, "loss": 0.9079, "step": 46839 }, { "epoch": 0.05, "learning_rate": 4.9730912571832554e-05, "loss": 0.866, "step": 46840 }, { "epoch": 0.05, "learning_rate": 4.973090107818608e-05, "loss": 1.4544, "step": 46841 }, { "epoch": 0.05, "learning_rate": 4.973088958429547e-05, "loss": 2.5718, "step": 46842 }, { "epoch": 0.05, "learning_rate": 4.973087809016072e-05, "loss": 2.5029, "step": 46843 }, { "epoch": 0.05, "learning_rate": 4.973086659578184e-05, "loss": 2.5318, "step": 46844 }, { "epoch": 0.05, "learning_rate": 4.973085510115883e-05, "loss": 1.1463, "step": 46845 }, { "epoch": 0.05, "learning_rate": 4.973084360629169e-05, "loss": 0.6873, "step": 46846 }, { "epoch": 0.05, "learning_rate": 4.9730832111180416e-05, "loss": 1.19, "step": 46847 }, { "epoch": 0.05, "learning_rate": 4.973082061582501e-05, "loss": 1.937, "step": 46848 }, { "epoch": 0.05, "learning_rate": 4.973080912022547e-05, "loss": 0.5436, "step": 46849 }, { "epoch": 0.05, "learning_rate": 4.97307976243818e-05, "loss": 0.9534, "step": 46850 }, { "epoch": 0.05, "learning_rate": 4.973078612829399e-05, "loss": 0.9234, "step": 46851 }, { "epoch": 0.05, "learning_rate": 4.9730774631962057e-05, "loss": 1.2125, "step": 46852 }, { "epoch": 0.05, "learning_rate": 4.9730763135385986e-05, "loss": 1.5367, "step": 46853 }, { "epoch": 0.05, "learning_rate": 4.973075163856579e-05, "loss": 1.4872, "step": 46854 }, { "epoch": 0.05, "learning_rate": 4.973074014150146e-05, "loss": 1.5188, "step": 46855 }, { "epoch": 0.05, "learning_rate": 4.973072864419299e-05, "loss": 1.0713, "step": 46856 }, { "epoch": 0.05, "learning_rate": 4.9730717146640405e-05, "loss": 0.8549, "step": 46857 }, { "epoch": 0.05, "learning_rate": 4.973070564884368e-05, "loss": 1.5679, "step": 46858 }, { "epoch": 0.05, "learning_rate": 4.973069415080282e-05, "loss": 3.9997, "step": 46859 }, { "epoch": 0.05, "learning_rate": 4.9730682652517834e-05, "loss": 1.4048, "step": 46860 }, { "epoch": 0.05, "learning_rate": 4.9730671153988715e-05, "loss": 1.199, "step": 46861 }, { "epoch": 0.05, "learning_rate": 4.973065965521547e-05, "loss": 0.9233, "step": 46862 }, { "epoch": 0.05, "learning_rate": 4.973064815619809e-05, "loss": 1.1845, "step": 46863 }, { "epoch": 0.05, "learning_rate": 4.973063665693658e-05, "loss": 0.4973, "step": 46864 }, { "epoch": 0.05, "learning_rate": 4.9730625157430936e-05, "loss": 0.5063, "step": 46865 }, { "epoch": 0.05, "learning_rate": 4.973061365768117e-05, "loss": 0.3843, "step": 46866 }, { "epoch": 0.05, "learning_rate": 4.9730602157687265e-05, "loss": 0.6689, "step": 46867 }, { "epoch": 0.05, "learning_rate": 4.9730590657449234e-05, "loss": 0.5886, "step": 46868 }, { "epoch": 0.05, "learning_rate": 4.9730579156967074e-05, "loss": 0.5114, "step": 46869 }, { "epoch": 0.05, "learning_rate": 4.9730567656240786e-05, "loss": 0.4535, "step": 46870 }, { "epoch": 0.05, "learning_rate": 4.973055615527036e-05, "loss": 0.419, "step": 46871 }, { "epoch": 0.05, "learning_rate": 4.973054465405581e-05, "loss": 1.1382, "step": 46872 }, { "epoch": 0.05, "learning_rate": 4.973053315259714e-05, "loss": 1.0448, "step": 46873 }, { "epoch": 0.05, "learning_rate": 4.973052165089432e-05, "loss": 1.1127, "step": 46874 }, { "epoch": 0.05, "learning_rate": 4.9730510148947384e-05, "loss": 0.8564, "step": 46875 }, { "epoch": 0.05, "learning_rate": 4.973049864675632e-05, "loss": 0.5005, "step": 46876 }, { "epoch": 0.05, "learning_rate": 4.973048714432112e-05, "loss": 0.9025, "step": 46877 }, { "epoch": 0.05, "learning_rate": 4.9730475641641795e-05, "loss": 1.0457, "step": 46878 }, { "epoch": 0.05, "learning_rate": 4.9730464138718344e-05, "loss": 1.0114, "step": 46879 }, { "epoch": 0.05, "learning_rate": 4.9730452635550765e-05, "loss": 1.0049, "step": 46880 }, { "epoch": 0.05, "learning_rate": 4.973044113213905e-05, "loss": 1.0725, "step": 46881 }, { "epoch": 0.05, "learning_rate": 4.973042962848321e-05, "loss": 0.9277, "step": 46882 }, { "epoch": 0.05, "learning_rate": 4.9730418124583236e-05, "loss": 1.2002, "step": 46883 }, { "epoch": 0.05, "learning_rate": 4.973040662043914e-05, "loss": 1.0002, "step": 46884 }, { "epoch": 0.05, "learning_rate": 4.9730395116050914e-05, "loss": 0.8721, "step": 46885 }, { "epoch": 0.05, "learning_rate": 4.9730383611418564e-05, "loss": 0.9231, "step": 46886 }, { "epoch": 0.05, "learning_rate": 4.973037210654208e-05, "loss": 0.8628, "step": 46887 }, { "epoch": 0.05, "learning_rate": 4.973036060142147e-05, "loss": 1.1176, "step": 46888 }, { "epoch": 0.05, "learning_rate": 4.9730349096056736e-05, "loss": 1.1107, "step": 46889 }, { "epoch": 0.05, "learning_rate": 4.973033759044787e-05, "loss": 0.3149, "step": 46890 }, { "epoch": 0.05, "learning_rate": 4.973032608459488e-05, "loss": 0.6816, "step": 46891 }, { "epoch": 0.05, "learning_rate": 4.973031457849776e-05, "loss": 0.5162, "step": 46892 }, { "epoch": 0.05, "learning_rate": 4.9730303072156516e-05, "loss": 0.4873, "step": 46893 }, { "epoch": 0.05, "learning_rate": 4.973029156557114e-05, "loss": 0.2094, "step": 46894 }, { "epoch": 0.05, "learning_rate": 4.973028005874164e-05, "loss": 0.1541, "step": 46895 }, { "epoch": 0.05, "learning_rate": 4.973026855166801e-05, "loss": 0.1072, "step": 46896 }, { "epoch": 0.05, "learning_rate": 4.973025704435026e-05, "loss": 0.4122, "step": 46897 }, { "epoch": 0.05, "learning_rate": 4.973024553678838e-05, "loss": 0.8716, "step": 46898 }, { "epoch": 0.05, "learning_rate": 4.973023402898237e-05, "loss": 1.1602, "step": 46899 }, { "epoch": 0.05, "learning_rate": 4.973022252093224e-05, "loss": 1.0753, "step": 46900 }, { "epoch": 0.05, "learning_rate": 4.973021101263798e-05, "loss": 1.056, "step": 46901 }, { "epoch": 0.05, "learning_rate": 4.9730199504099594e-05, "loss": 1.215, "step": 46902 }, { "epoch": 0.05, "learning_rate": 4.973018799531708e-05, "loss": 1.4891, "step": 46903 }, { "epoch": 0.05, "learning_rate": 4.973017648629044e-05, "loss": 1.4404, "step": 46904 }, { "epoch": 0.05, "learning_rate": 4.9730164977019676e-05, "loss": 1.073, "step": 46905 }, { "epoch": 0.05, "learning_rate": 4.973015346750479e-05, "loss": 0.7088, "step": 46906 }, { "epoch": 0.05, "learning_rate": 4.973014195774578e-05, "loss": 0.7515, "step": 46907 }, { "epoch": 0.05, "learning_rate": 4.973013044774263e-05, "loss": 0.9139, "step": 46908 }, { "epoch": 0.05, "learning_rate": 4.9730118937495364e-05, "loss": 1.0907, "step": 46909 }, { "epoch": 0.05, "learning_rate": 4.973010742700397e-05, "loss": 0.8861, "step": 46910 }, { "epoch": 0.05, "learning_rate": 4.973009591626846e-05, "loss": 1.304, "step": 46911 }, { "epoch": 0.05, "learning_rate": 4.9730084405288816e-05, "loss": 1.1009, "step": 46912 }, { "epoch": 0.05, "learning_rate": 4.973007289406505e-05, "loss": 1.2148, "step": 46913 }, { "epoch": 0.05, "learning_rate": 4.973006138259715e-05, "loss": 0.8415, "step": 46914 }, { "epoch": 0.05, "learning_rate": 4.973004987088514e-05, "loss": 2.2397, "step": 46915 }, { "epoch": 0.05, "learning_rate": 4.9730038358928996e-05, "loss": 3.0074, "step": 46916 }, { "epoch": 0.05, "learning_rate": 4.9730026846728725e-05, "loss": 2.4613, "step": 46917 }, { "epoch": 0.05, "learning_rate": 4.973001533428434e-05, "loss": 1.8411, "step": 46918 }, { "epoch": 0.05, "learning_rate": 4.9730003821595824e-05, "loss": 2.2903, "step": 46919 }, { "epoch": 0.05, "learning_rate": 4.972999230866319e-05, "loss": 1.0625, "step": 46920 }, { "epoch": 0.05, "learning_rate": 4.9729980795486424e-05, "loss": 0.9147, "step": 46921 }, { "epoch": 0.05, "learning_rate": 4.972996928206554e-05, "loss": 1.0866, "step": 46922 }, { "epoch": 0.05, "learning_rate": 4.972995776840052e-05, "loss": 0.6549, "step": 46923 }, { "epoch": 0.05, "learning_rate": 4.972994625449139e-05, "loss": 0.8927, "step": 46924 }, { "epoch": 0.05, "learning_rate": 4.9729934740338135e-05, "loss": 0.9663, "step": 46925 }, { "epoch": 0.05, "learning_rate": 4.9729923225940756e-05, "loss": 1.0619, "step": 46926 }, { "epoch": 0.05, "learning_rate": 4.972991171129925e-05, "loss": 1.3093, "step": 46927 }, { "epoch": 0.05, "learning_rate": 4.972990019641362e-05, "loss": 1.1976, "step": 46928 }, { "epoch": 0.05, "learning_rate": 4.9729888681283867e-05, "loss": 1.2076, "step": 46929 }, { "epoch": 0.05, "learning_rate": 4.9729877165909994e-05, "loss": 1.4542, "step": 46930 }, { "epoch": 0.05, "learning_rate": 4.9729865650292e-05, "loss": 1.1851, "step": 46931 }, { "epoch": 0.05, "learning_rate": 4.9729854134429876e-05, "loss": 1.2524, "step": 46932 }, { "epoch": 0.05, "learning_rate": 4.972984261832364e-05, "loss": 1.0978, "step": 46933 }, { "epoch": 0.05, "learning_rate": 4.972983110197327e-05, "loss": 1.1628, "step": 46934 }, { "epoch": 0.05, "learning_rate": 4.9729819585378776e-05, "loss": 1.187, "step": 46935 }, { "epoch": 0.05, "learning_rate": 4.972980806854017e-05, "loss": 1.0774, "step": 46936 }, { "epoch": 0.05, "learning_rate": 4.972979655145744e-05, "loss": 0.618, "step": 46937 }, { "epoch": 0.05, "learning_rate": 4.972978503413058e-05, "loss": 0.8438, "step": 46938 }, { "epoch": 0.05, "learning_rate": 4.9729773516559606e-05, "loss": 1.0914, "step": 46939 }, { "epoch": 0.05, "learning_rate": 4.97297619987445e-05, "loss": 1.1903, "step": 46940 }, { "epoch": 0.05, "learning_rate": 4.972975048068529e-05, "loss": 1.0966, "step": 46941 }, { "epoch": 0.05, "learning_rate": 4.9729738962381945e-05, "loss": 0.5892, "step": 46942 }, { "epoch": 0.05, "learning_rate": 4.972972744383448e-05, "loss": 1.2508, "step": 46943 }, { "epoch": 0.05, "learning_rate": 4.97297159250429e-05, "loss": 0.9213, "step": 46944 }, { "epoch": 0.05, "learning_rate": 4.9729704406007195e-05, "loss": 0.9855, "step": 46945 }, { "epoch": 0.05, "learning_rate": 4.972969288672736e-05, "loss": 1.2215, "step": 46946 }, { "epoch": 0.05, "learning_rate": 4.972968136720342e-05, "loss": 0.4112, "step": 46947 }, { "epoch": 0.05, "learning_rate": 4.972966984743534e-05, "loss": 0.6981, "step": 46948 }, { "epoch": 0.05, "learning_rate": 4.972965832742316e-05, "loss": 0.6046, "step": 46949 }, { "epoch": 0.05, "learning_rate": 4.9729646807166845e-05, "loss": 1.3887, "step": 46950 }, { "epoch": 0.05, "learning_rate": 4.972963528666641e-05, "loss": 0.976, "step": 46951 }, { "epoch": 0.05, "learning_rate": 4.972962376592186e-05, "loss": 0.6874, "step": 46952 }, { "epoch": 0.05, "learning_rate": 4.972961224493319e-05, "loss": 0.358, "step": 46953 }, { "epoch": 0.05, "learning_rate": 4.9729600723700396e-05, "loss": 0.6867, "step": 46954 }, { "epoch": 0.05, "learning_rate": 4.972958920222348e-05, "loss": 0.97, "step": 46955 }, { "epoch": 0.05, "learning_rate": 4.972957768050245e-05, "loss": 0.9862, "step": 46956 }, { "epoch": 0.05, "learning_rate": 4.972956615853729e-05, "loss": 2.9339, "step": 46957 }, { "epoch": 0.05, "learning_rate": 4.9729554636328026e-05, "loss": 1.9905, "step": 46958 }, { "epoch": 0.05, "learning_rate": 4.9729543113874625e-05, "loss": 0.8556, "step": 46959 }, { "epoch": 0.05, "learning_rate": 4.9729531591177115e-05, "loss": 1.068, "step": 46960 }, { "epoch": 0.05, "learning_rate": 4.972952006823548e-05, "loss": 0.9461, "step": 46961 }, { "epoch": 0.05, "learning_rate": 4.972950854504973e-05, "loss": 0.8989, "step": 46962 }, { "epoch": 0.05, "learning_rate": 4.972949702161986e-05, "loss": 1.0685, "step": 46963 }, { "epoch": 0.05, "learning_rate": 4.972948549794587e-05, "loss": 0.8232, "step": 46964 }, { "epoch": 0.05, "learning_rate": 4.972947397402776e-05, "loss": 0.8161, "step": 46965 }, { "epoch": 0.05, "learning_rate": 4.9729462449865536e-05, "loss": 0.9024, "step": 46966 }, { "epoch": 0.05, "learning_rate": 4.972945092545919e-05, "loss": 1.3549, "step": 46967 }, { "epoch": 0.05, "learning_rate": 4.972943940080872e-05, "loss": 1.4177, "step": 46968 }, { "epoch": 0.05, "learning_rate": 4.9729427875914134e-05, "loss": 1.0952, "step": 46969 }, { "epoch": 0.05, "learning_rate": 4.9729416350775435e-05, "loss": 1.0067, "step": 46970 }, { "epoch": 0.05, "learning_rate": 4.9729404825392614e-05, "loss": 0.9874, "step": 46971 }, { "epoch": 0.05, "learning_rate": 4.972939329976568e-05, "loss": 0.8352, "step": 46972 }, { "epoch": 0.05, "learning_rate": 4.972938177389462e-05, "loss": 0.6069, "step": 46973 }, { "epoch": 0.05, "learning_rate": 4.972937024777944e-05, "loss": 0.9588, "step": 46974 }, { "epoch": 0.05, "learning_rate": 4.972935872142015e-05, "loss": 1.0045, "step": 46975 }, { "epoch": 0.05, "learning_rate": 4.972934719481673e-05, "loss": 0.7635, "step": 46976 }, { "epoch": 0.05, "learning_rate": 4.972933566796921e-05, "loss": 0.5785, "step": 46977 }, { "epoch": 0.05, "learning_rate": 4.972932414087756e-05, "loss": 1.1007, "step": 46978 }, { "epoch": 0.05, "learning_rate": 4.97293126135418e-05, "loss": 1.1936, "step": 46979 }, { "epoch": 0.05, "learning_rate": 4.972930108596192e-05, "loss": 1.0258, "step": 46980 }, { "epoch": 0.05, "learning_rate": 4.9729289558137915e-05, "loss": 1.1514, "step": 46981 }, { "epoch": 0.05, "learning_rate": 4.97292780300698e-05, "loss": 1.2222, "step": 46982 }, { "epoch": 0.05, "learning_rate": 4.972926650175756e-05, "loss": 1.1286, "step": 46983 }, { "epoch": 0.05, "learning_rate": 4.972925497320122e-05, "loss": 0.7899, "step": 46984 }, { "epoch": 0.05, "learning_rate": 4.972924344440075e-05, "loss": 0.7675, "step": 46985 }, { "epoch": 0.05, "learning_rate": 4.9729231915356166e-05, "loss": 1.0589, "step": 46986 }, { "epoch": 0.05, "learning_rate": 4.972922038606747e-05, "loss": 1.1395, "step": 46987 }, { "epoch": 0.05, "learning_rate": 4.972920885653465e-05, "loss": 1.1627, "step": 46988 }, { "epoch": 0.05, "learning_rate": 4.972919732675772e-05, "loss": 0.7004, "step": 46989 }, { "epoch": 0.05, "learning_rate": 4.972918579673668e-05, "loss": 0.8704, "step": 46990 }, { "epoch": 0.05, "learning_rate": 4.972917426647151e-05, "loss": 0.9194, "step": 46991 }, { "epoch": 0.05, "learning_rate": 4.972916273596223e-05, "loss": 0.6611, "step": 46992 }, { "epoch": 0.05, "learning_rate": 4.972915120520883e-05, "loss": 1.2238, "step": 46993 }, { "epoch": 0.05, "learning_rate": 4.972913967421132e-05, "loss": 0.9342, "step": 46994 }, { "epoch": 0.05, "learning_rate": 4.9729128142969696e-05, "loss": 1.2023, "step": 46995 }, { "epoch": 0.05, "learning_rate": 4.972911661148395e-05, "loss": 0.784, "step": 46996 }, { "epoch": 0.05, "learning_rate": 4.972910507975409e-05, "loss": 1.2068, "step": 46997 }, { "epoch": 0.05, "learning_rate": 4.9729093547780116e-05, "loss": 0.9346, "step": 46998 }, { "epoch": 0.05, "learning_rate": 4.9729082015562023e-05, "loss": 1.0641, "step": 46999 }, { "epoch": 0.05, "learning_rate": 4.972907048309982e-05, "loss": 1.2464, "step": 47000 }, { "epoch": 0.05, "eval_loss": 1.0832467079162598, "eval_runtime": 125.52, "eval_samples_per_second": 11.034, "eval_steps_per_second": 5.521, "step": 47000 }, { "epoch": 0.05, "learning_rate": 4.97290589503935e-05, "loss": 1.199, "step": 47001 }, { "epoch": 0.05, "learning_rate": 4.9729047417443064e-05, "loss": 1.1167, "step": 47002 }, { "epoch": 0.05, "learning_rate": 4.972903588424852e-05, "loss": 0.9475, "step": 47003 }, { "epoch": 0.05, "learning_rate": 4.972902435080985e-05, "loss": 0.7281, "step": 47004 }, { "epoch": 0.05, "learning_rate": 4.972901281712708e-05, "loss": 0.7719, "step": 47005 }, { "epoch": 0.05, "learning_rate": 4.972900128320018e-05, "loss": 1.1034, "step": 47006 }, { "epoch": 0.05, "learning_rate": 4.9728989749029176e-05, "loss": 0.913, "step": 47007 }, { "epoch": 0.05, "learning_rate": 4.972897821461405e-05, "loss": 0.9636, "step": 47008 }, { "epoch": 0.05, "learning_rate": 4.972896667995482e-05, "loss": 1.096, "step": 47009 }, { "epoch": 0.05, "learning_rate": 4.972895514505147e-05, "loss": 1.259, "step": 47010 }, { "epoch": 0.05, "learning_rate": 4.972894360990401e-05, "loss": 1.6639, "step": 47011 }, { "epoch": 0.05, "learning_rate": 4.9728932074512436e-05, "loss": 1.492, "step": 47012 }, { "epoch": 0.05, "learning_rate": 4.972892053887674e-05, "loss": 1.3719, "step": 47013 }, { "epoch": 0.05, "learning_rate": 4.972890900299694e-05, "loss": 1.2596, "step": 47014 }, { "epoch": 0.05, "learning_rate": 4.972889746687302e-05, "loss": 1.2879, "step": 47015 }, { "epoch": 0.05, "learning_rate": 4.9728885930504985e-05, "loss": 0.5291, "step": 47016 }, { "epoch": 0.05, "learning_rate": 4.9728874393892845e-05, "loss": 1.1409, "step": 47017 }, { "epoch": 0.05, "learning_rate": 4.972886285703659e-05, "loss": 1.2415, "step": 47018 }, { "epoch": 0.05, "learning_rate": 4.972885131993622e-05, "loss": 1.1618, "step": 47019 }, { "epoch": 0.05, "learning_rate": 4.972883978259174e-05, "loss": 1.0615, "step": 47020 }, { "epoch": 0.05, "learning_rate": 4.972882824500314e-05, "loss": 0.8954, "step": 47021 }, { "epoch": 0.05, "learning_rate": 4.9728816707170434e-05, "loss": 0.7671, "step": 47022 }, { "epoch": 0.05, "learning_rate": 4.972880516909362e-05, "loss": 1.1007, "step": 47023 }, { "epoch": 0.05, "learning_rate": 4.972879363077269e-05, "loss": 1.322, "step": 47024 }, { "epoch": 0.05, "learning_rate": 4.9728782092207635e-05, "loss": 1.2125, "step": 47025 }, { "epoch": 0.05, "learning_rate": 4.972877055339848e-05, "loss": 0.9197, "step": 47026 }, { "epoch": 0.05, "learning_rate": 4.972875901434522e-05, "loss": 0.6871, "step": 47027 }, { "epoch": 0.05, "learning_rate": 4.972874747504783e-05, "loss": 0.7462, "step": 47028 }, { "epoch": 0.05, "learning_rate": 4.9728735935506344e-05, "loss": 0.8313, "step": 47029 }, { "epoch": 0.05, "learning_rate": 4.972872439572074e-05, "loss": 0.7531, "step": 47030 }, { "epoch": 0.05, "learning_rate": 4.972871285569102e-05, "loss": 0.7917, "step": 47031 }, { "epoch": 0.05, "learning_rate": 4.97287013154172e-05, "loss": 1.298, "step": 47032 }, { "epoch": 0.05, "learning_rate": 4.972868977489926e-05, "loss": 1.6845, "step": 47033 }, { "epoch": 0.05, "learning_rate": 4.97286782341372e-05, "loss": 1.0196, "step": 47034 }, { "epoch": 0.05, "learning_rate": 4.972866669313105e-05, "loss": 0.9721, "step": 47035 }, { "epoch": 0.05, "learning_rate": 4.9728655151880774e-05, "loss": 0.9609, "step": 47036 }, { "epoch": 0.05, "learning_rate": 4.97286436103864e-05, "loss": 1.3054, "step": 47037 }, { "epoch": 0.05, "learning_rate": 4.97286320686479e-05, "loss": 1.3264, "step": 47038 }, { "epoch": 0.05, "learning_rate": 4.97286205266653e-05, "loss": 1.4144, "step": 47039 }, { "epoch": 0.05, "learning_rate": 4.972860898443858e-05, "loss": 1.1372, "step": 47040 }, { "epoch": 0.05, "learning_rate": 4.972859744196776e-05, "loss": 0.9477, "step": 47041 }, { "epoch": 0.05, "learning_rate": 4.972858589925282e-05, "loss": 1.1287, "step": 47042 }, { "epoch": 0.05, "learning_rate": 4.9728574356293776e-05, "loss": 1.1192, "step": 47043 }, { "epoch": 0.05, "learning_rate": 4.972856281309062e-05, "loss": 0.973, "step": 47044 }, { "epoch": 0.05, "learning_rate": 4.972855126964336e-05, "loss": 0.9251, "step": 47045 }, { "epoch": 0.05, "learning_rate": 4.972853972595198e-05, "loss": 0.7252, "step": 47046 }, { "epoch": 0.05, "learning_rate": 4.97285281820165e-05, "loss": 0.8821, "step": 47047 }, { "epoch": 0.05, "learning_rate": 4.972851663783691e-05, "loss": 0.8827, "step": 47048 }, { "epoch": 0.05, "learning_rate": 4.97285050934132e-05, "loss": 0.9439, "step": 47049 }, { "epoch": 0.05, "learning_rate": 4.972849354874539e-05, "loss": 0.7531, "step": 47050 }, { "epoch": 0.05, "learning_rate": 4.972848200383347e-05, "loss": 1.0091, "step": 47051 }, { "epoch": 0.05, "learning_rate": 4.9728470458677436e-05, "loss": 0.8216, "step": 47052 }, { "epoch": 0.05, "learning_rate": 4.9728458913277295e-05, "loss": 1.2661, "step": 47053 }, { "epoch": 0.05, "learning_rate": 4.9728447367633046e-05, "loss": 0.9695, "step": 47054 }, { "epoch": 0.05, "learning_rate": 4.972843582174468e-05, "loss": 1.2277, "step": 47055 }, { "epoch": 0.05, "learning_rate": 4.972842427561222e-05, "loss": 1.1753, "step": 47056 }, { "epoch": 0.05, "learning_rate": 4.972841272923564e-05, "loss": 0.9361, "step": 47057 }, { "epoch": 0.05, "learning_rate": 4.9728401182614956e-05, "loss": 1.1397, "step": 47058 }, { "epoch": 0.05, "learning_rate": 4.972838963575017e-05, "loss": 1.0203, "step": 47059 }, { "epoch": 0.05, "learning_rate": 4.972837808864126e-05, "loss": 1.1982, "step": 47060 }, { "epoch": 0.05, "learning_rate": 4.9728366541288257e-05, "loss": 1.0815, "step": 47061 }, { "epoch": 0.05, "learning_rate": 4.972835499369114e-05, "loss": 1.6886, "step": 47062 }, { "epoch": 0.05, "learning_rate": 4.9728343445849914e-05, "loss": 1.4611, "step": 47063 }, { "epoch": 0.05, "learning_rate": 4.972833189776458e-05, "loss": 0.9569, "step": 47064 }, { "epoch": 0.05, "learning_rate": 4.972832034943514e-05, "loss": 1.2162, "step": 47065 }, { "epoch": 0.05, "learning_rate": 4.972830880086159e-05, "loss": 0.8322, "step": 47066 }, { "epoch": 0.05, "learning_rate": 4.9728297252043945e-05, "loss": 1.0358, "step": 47067 }, { "epoch": 0.05, "learning_rate": 4.972828570298218e-05, "loss": 1.2759, "step": 47068 }, { "epoch": 0.05, "learning_rate": 4.9728274153676305e-05, "loss": 0.802, "step": 47069 }, { "epoch": 0.05, "learning_rate": 4.972826260412633e-05, "loss": 0.2533, "step": 47070 }, { "epoch": 0.05, "learning_rate": 4.972825105433224e-05, "loss": 0.2427, "step": 47071 }, { "epoch": 0.05, "learning_rate": 4.972823950429406e-05, "loss": 0.5786, "step": 47072 }, { "epoch": 0.05, "learning_rate": 4.9728227954011755e-05, "loss": 0.3443, "step": 47073 }, { "epoch": 0.05, "learning_rate": 4.972821640348535e-05, "loss": 1.1077, "step": 47074 }, { "epoch": 0.05, "learning_rate": 4.9728204852714845e-05, "loss": 1.0583, "step": 47075 }, { "epoch": 0.05, "learning_rate": 4.9728193301700224e-05, "loss": 0.9253, "step": 47076 }, { "epoch": 0.05, "learning_rate": 4.97281817504415e-05, "loss": 0.9445, "step": 47077 }, { "epoch": 0.05, "learning_rate": 4.972817019893867e-05, "loss": 1.2061, "step": 47078 }, { "epoch": 0.05, "learning_rate": 4.972815864719173e-05, "loss": 1.1884, "step": 47079 }, { "epoch": 0.05, "learning_rate": 4.9728147095200694e-05, "loss": 1.3653, "step": 47080 }, { "epoch": 0.05, "learning_rate": 4.9728135542965546e-05, "loss": 0.9087, "step": 47081 }, { "epoch": 0.05, "learning_rate": 4.97281239904863e-05, "loss": 0.4792, "step": 47082 }, { "epoch": 0.05, "learning_rate": 4.972811243776293e-05, "loss": 1.0403, "step": 47083 }, { "epoch": 0.05, "learning_rate": 4.972810088479547e-05, "loss": 1.2227, "step": 47084 }, { "epoch": 0.05, "learning_rate": 4.972808933158391e-05, "loss": 1.0643, "step": 47085 }, { "epoch": 0.05, "learning_rate": 4.9728077778128226e-05, "loss": 1.1178, "step": 47086 }, { "epoch": 0.05, "learning_rate": 4.972806622442845e-05, "loss": 0.6254, "step": 47087 }, { "epoch": 0.05, "learning_rate": 4.972805467048456e-05, "loss": 0.3162, "step": 47088 }, { "epoch": 0.05, "learning_rate": 4.972804311629657e-05, "loss": 0.9986, "step": 47089 }, { "epoch": 0.05, "learning_rate": 4.972803156186447e-05, "loss": 0.9555, "step": 47090 }, { "epoch": 0.05, "learning_rate": 4.972802000718828e-05, "loss": 1.1352, "step": 47091 }, { "epoch": 0.05, "learning_rate": 4.9728008452267974e-05, "loss": 0.7067, "step": 47092 }, { "epoch": 0.05, "learning_rate": 4.972799689710356e-05, "loss": 1.0502, "step": 47093 }, { "epoch": 0.05, "learning_rate": 4.972798534169505e-05, "loss": 0.9527, "step": 47094 }, { "epoch": 0.05, "learning_rate": 4.9727973786042437e-05, "loss": 0.8445, "step": 47095 }, { "epoch": 0.05, "learning_rate": 4.972796223014571e-05, "loss": 0.9892, "step": 47096 }, { "epoch": 0.05, "learning_rate": 4.9727950674004884e-05, "loss": 1.0811, "step": 47097 }, { "epoch": 0.05, "learning_rate": 4.972793911761996e-05, "loss": 1.3623, "step": 47098 }, { "epoch": 0.05, "learning_rate": 4.972792756099092e-05, "loss": 0.9791, "step": 47099 }, { "epoch": 0.05, "learning_rate": 4.9727916004117785e-05, "loss": 1.0467, "step": 47100 }, { "epoch": 0.05, "learning_rate": 4.972790444700054e-05, "loss": 0.8634, "step": 47101 }, { "epoch": 0.05, "learning_rate": 4.9727892889639206e-05, "loss": 0.9175, "step": 47102 }, { "epoch": 0.05, "learning_rate": 4.9727881332033754e-05, "loss": 0.8036, "step": 47103 }, { "epoch": 0.05, "learning_rate": 4.972786977418421e-05, "loss": 0.9709, "step": 47104 }, { "epoch": 0.05, "learning_rate": 4.972785821609055e-05, "loss": 1.0236, "step": 47105 }, { "epoch": 0.05, "learning_rate": 4.97278466577528e-05, "loss": 1.043, "step": 47106 }, { "epoch": 0.05, "learning_rate": 4.972783509917094e-05, "loss": 0.9666, "step": 47107 }, { "epoch": 0.05, "learning_rate": 4.9727823540344975e-05, "loss": 1.3567, "step": 47108 }, { "epoch": 0.05, "learning_rate": 4.972781198127491e-05, "loss": 1.053, "step": 47109 }, { "epoch": 0.05, "learning_rate": 4.972780042196074e-05, "loss": 0.9548, "step": 47110 }, { "epoch": 0.05, "learning_rate": 4.972778886240247e-05, "loss": 1.2804, "step": 47111 }, { "epoch": 0.05, "learning_rate": 4.97277773026001e-05, "loss": 1.1052, "step": 47112 }, { "epoch": 0.05, "learning_rate": 4.972776574255362e-05, "loss": 0.8313, "step": 47113 }, { "epoch": 0.05, "learning_rate": 4.972775418226305e-05, "loss": 1.0177, "step": 47114 }, { "epoch": 0.05, "learning_rate": 4.9727742621728366e-05, "loss": 0.8757, "step": 47115 }, { "epoch": 0.05, "learning_rate": 4.972773106094959e-05, "loss": 0.6286, "step": 47116 }, { "epoch": 0.05, "learning_rate": 4.9727719499926706e-05, "loss": 0.3499, "step": 47117 }, { "epoch": 0.05, "learning_rate": 4.972770793865972e-05, "loss": 0.2092, "step": 47118 }, { "epoch": 0.05, "learning_rate": 4.9727696377148634e-05, "loss": 0.1855, "step": 47119 }, { "epoch": 0.05, "learning_rate": 4.972768481539345e-05, "loss": 0.334, "step": 47120 }, { "epoch": 0.05, "learning_rate": 4.972767325339416e-05, "loss": 0.5049, "step": 47121 }, { "epoch": 0.05, "learning_rate": 4.972766169115077e-05, "loss": 1.2995, "step": 47122 }, { "epoch": 0.05, "learning_rate": 4.9727650128663286e-05, "loss": 1.3047, "step": 47123 }, { "epoch": 0.05, "learning_rate": 4.9727638565931686e-05, "loss": 1.3134, "step": 47124 }, { "epoch": 0.05, "learning_rate": 4.9727627002956e-05, "loss": 1.0468, "step": 47125 }, { "epoch": 0.05, "learning_rate": 4.972761543973621e-05, "loss": 0.9008, "step": 47126 }, { "epoch": 0.05, "learning_rate": 4.972760387627232e-05, "loss": 1.2408, "step": 47127 }, { "epoch": 0.05, "learning_rate": 4.972759231256432e-05, "loss": 1.3808, "step": 47128 }, { "epoch": 0.05, "learning_rate": 4.9727580748612225e-05, "loss": 1.7073, "step": 47129 }, { "epoch": 0.05, "learning_rate": 4.972756918441604e-05, "loss": 1.2842, "step": 47130 }, { "epoch": 0.05, "learning_rate": 4.972755761997574e-05, "loss": 1.2236, "step": 47131 }, { "epoch": 0.05, "learning_rate": 4.972754605529134e-05, "loss": 0.6306, "step": 47132 }, { "epoch": 0.05, "learning_rate": 4.972753449036285e-05, "loss": 0.3848, "step": 47133 }, { "epoch": 0.05, "learning_rate": 4.972752292519026e-05, "loss": 0.4331, "step": 47134 }, { "epoch": 0.05, "learning_rate": 4.972751135977356e-05, "loss": 0.9583, "step": 47135 }, { "epoch": 0.05, "learning_rate": 4.9727499794112767e-05, "loss": 1.0414, "step": 47136 }, { "epoch": 0.05, "learning_rate": 4.9727488228207876e-05, "loss": 0.9703, "step": 47137 }, { "epoch": 0.05, "learning_rate": 4.9727476662058884e-05, "loss": 0.8854, "step": 47138 }, { "epoch": 0.05, "learning_rate": 4.972746509566579e-05, "loss": 0.8729, "step": 47139 }, { "epoch": 0.05, "learning_rate": 4.97274535290286e-05, "loss": 1.0251, "step": 47140 }, { "epoch": 0.05, "learning_rate": 4.972744196214731e-05, "loss": 1.1406, "step": 47141 }, { "epoch": 0.05, "learning_rate": 4.972743039502192e-05, "loss": 1.1545, "step": 47142 }, { "epoch": 0.05, "learning_rate": 4.972741882765244e-05, "loss": 1.096, "step": 47143 }, { "epoch": 0.05, "learning_rate": 4.972740726003885e-05, "loss": 1.3844, "step": 47144 }, { "epoch": 0.05, "learning_rate": 4.972739569218117e-05, "loss": 1.0848, "step": 47145 }, { "epoch": 0.05, "learning_rate": 4.972738412407938e-05, "loss": 1.1154, "step": 47146 }, { "epoch": 0.05, "learning_rate": 4.9727372555733496e-05, "loss": 0.9417, "step": 47147 }, { "epoch": 0.05, "learning_rate": 4.9727360987143525e-05, "loss": 0.9582, "step": 47148 }, { "epoch": 0.05, "learning_rate": 4.9727349418309445e-05, "loss": 1.1796, "step": 47149 }, { "epoch": 0.05, "learning_rate": 4.972733784923127e-05, "loss": 1.0774, "step": 47150 }, { "epoch": 0.05, "learning_rate": 4.9727326279908995e-05, "loss": 1.3594, "step": 47151 }, { "epoch": 0.05, "learning_rate": 4.972731471034262e-05, "loss": 1.0133, "step": 47152 }, { "epoch": 0.05, "learning_rate": 4.9727303140532153e-05, "loss": 1.0505, "step": 47153 }, { "epoch": 0.05, "learning_rate": 4.9727291570477594e-05, "loss": 1.132, "step": 47154 }, { "epoch": 0.05, "learning_rate": 4.9727280000178927e-05, "loss": 0.81, "step": 47155 }, { "epoch": 0.05, "learning_rate": 4.9727268429636165e-05, "loss": 0.9739, "step": 47156 }, { "epoch": 0.05, "learning_rate": 4.972725685884931e-05, "loss": 1.1225, "step": 47157 }, { "epoch": 0.05, "learning_rate": 4.972724528781835e-05, "loss": 0.9021, "step": 47158 }, { "epoch": 0.05, "learning_rate": 4.97272337165433e-05, "loss": 1.4298, "step": 47159 }, { "epoch": 0.05, "learning_rate": 4.972722214502416e-05, "loss": 1.1587, "step": 47160 }, { "epoch": 0.05, "learning_rate": 4.972721057326091e-05, "loss": 0.9081, "step": 47161 }, { "epoch": 0.05, "learning_rate": 4.9727199001253566e-05, "loss": 0.8032, "step": 47162 }, { "epoch": 0.05, "learning_rate": 4.972718742900213e-05, "loss": 1.3094, "step": 47163 }, { "epoch": 0.05, "learning_rate": 4.972717585650659e-05, "loss": 1.0887, "step": 47164 }, { "epoch": 0.05, "learning_rate": 4.9727164283766956e-05, "loss": 0.5378, "step": 47165 }, { "epoch": 0.05, "learning_rate": 4.9727152710783235e-05, "loss": 0.7458, "step": 47166 }, { "epoch": 0.05, "learning_rate": 4.972714113755541e-05, "loss": 1.0846, "step": 47167 }, { "epoch": 0.05, "learning_rate": 4.972712956408349e-05, "loss": 1.0236, "step": 47168 }, { "epoch": 0.05, "learning_rate": 4.972711799036747e-05, "loss": 0.3378, "step": 47169 }, { "epoch": 0.05, "learning_rate": 4.9727106416407363e-05, "loss": 0.2653, "step": 47170 }, { "epoch": 0.05, "learning_rate": 4.9727094842203156e-05, "loss": 0.6023, "step": 47171 }, { "epoch": 0.05, "learning_rate": 4.9727083267754854e-05, "loss": 0.8893, "step": 47172 }, { "epoch": 0.05, "learning_rate": 4.972707169306246e-05, "loss": 1.1012, "step": 47173 }, { "epoch": 0.05, "learning_rate": 4.972706011812596e-05, "loss": 0.7776, "step": 47174 }, { "epoch": 0.05, "learning_rate": 4.972704854294537e-05, "loss": 0.6485, "step": 47175 }, { "epoch": 0.05, "learning_rate": 4.972703696752069e-05, "loss": 0.197, "step": 47176 }, { "epoch": 0.05, "learning_rate": 4.972702539185191e-05, "loss": 1.1252, "step": 47177 }, { "epoch": 0.05, "learning_rate": 4.972701381593904e-05, "loss": 0.8259, "step": 47178 }, { "epoch": 0.05, "learning_rate": 4.972700223978207e-05, "loss": 0.4899, "step": 47179 }, { "epoch": 0.05, "learning_rate": 4.972699066338101e-05, "loss": 0.841, "step": 47180 }, { "epoch": 0.05, "learning_rate": 4.972697908673586e-05, "loss": 1.2298, "step": 47181 }, { "epoch": 0.05, "learning_rate": 4.972696750984661e-05, "loss": 0.964, "step": 47182 }, { "epoch": 0.05, "learning_rate": 4.972695593271326e-05, "loss": 1.0068, "step": 47183 }, { "epoch": 0.05, "learning_rate": 4.9726944355335824e-05, "loss": 1.1285, "step": 47184 }, { "epoch": 0.05, "learning_rate": 4.972693277771429e-05, "loss": 0.8755, "step": 47185 }, { "epoch": 0.05, "learning_rate": 4.9726921199848666e-05, "loss": 1.0966, "step": 47186 }, { "epoch": 0.05, "learning_rate": 4.9726909621738946e-05, "loss": 0.8594, "step": 47187 }, { "epoch": 0.05, "learning_rate": 4.972689804338513e-05, "loss": 1.061, "step": 47188 }, { "epoch": 0.05, "learning_rate": 4.972688646478723e-05, "loss": 1.0854, "step": 47189 }, { "epoch": 0.05, "learning_rate": 4.972687488594522e-05, "loss": 0.9095, "step": 47190 }, { "epoch": 0.05, "learning_rate": 4.972686330685913e-05, "loss": 1.2761, "step": 47191 }, { "epoch": 0.05, "learning_rate": 4.9726851727528944e-05, "loss": 1.3857, "step": 47192 }, { "epoch": 0.05, "learning_rate": 4.972684014795466e-05, "loss": 1.5303, "step": 47193 }, { "epoch": 0.05, "learning_rate": 4.972682856813629e-05, "loss": 1.0708, "step": 47194 }, { "epoch": 0.05, "learning_rate": 4.972681698807382e-05, "loss": 0.8411, "step": 47195 }, { "epoch": 0.05, "learning_rate": 4.9726805407767265e-05, "loss": 0.7456, "step": 47196 }, { "epoch": 0.05, "learning_rate": 4.972679382721661e-05, "loss": 1.2334, "step": 47197 }, { "epoch": 0.05, "learning_rate": 4.9726782246421865e-05, "loss": 0.929, "step": 47198 }, { "epoch": 0.05, "learning_rate": 4.972677066538303e-05, "loss": 0.8006, "step": 47199 }, { "epoch": 0.05, "learning_rate": 4.9726759084100106e-05, "loss": 1.3269, "step": 47200 }, { "epoch": 0.05, "learning_rate": 4.972674750257308e-05, "loss": 1.5196, "step": 47201 }, { "epoch": 0.05, "learning_rate": 4.972673592080197e-05, "loss": 1.5504, "step": 47202 }, { "epoch": 0.05, "learning_rate": 4.972672433878677e-05, "loss": 1.2319, "step": 47203 }, { "epoch": 0.05, "learning_rate": 4.972671275652747e-05, "loss": 0.8895, "step": 47204 }, { "epoch": 0.05, "learning_rate": 4.972670117402408e-05, "loss": 0.7244, "step": 47205 }, { "epoch": 0.05, "learning_rate": 4.9726689591276596e-05, "loss": 0.9974, "step": 47206 }, { "epoch": 0.05, "learning_rate": 4.972667800828503e-05, "loss": 0.8634, "step": 47207 }, { "epoch": 0.05, "learning_rate": 4.9726666425049366e-05, "loss": 0.4326, "step": 47208 }, { "epoch": 0.05, "learning_rate": 4.972665484156961e-05, "loss": 0.798, "step": 47209 }, { "epoch": 0.05, "learning_rate": 4.972664325784577e-05, "loss": 1.2236, "step": 47210 }, { "epoch": 0.05, "learning_rate": 4.9726631673877835e-05, "loss": 0.9664, "step": 47211 }, { "epoch": 0.05, "learning_rate": 4.972662008966581e-05, "loss": 1.1837, "step": 47212 }, { "epoch": 0.05, "learning_rate": 4.972660850520969e-05, "loss": 1.1316, "step": 47213 }, { "epoch": 0.05, "learning_rate": 4.9726596920509486e-05, "loss": 0.8749, "step": 47214 }, { "epoch": 0.05, "learning_rate": 4.972658533556518e-05, "loss": 1.0425, "step": 47215 }, { "epoch": 0.05, "learning_rate": 4.972657375037679e-05, "loss": 1.0562, "step": 47216 }, { "epoch": 0.05, "learning_rate": 4.9726562164944313e-05, "loss": 0.9311, "step": 47217 }, { "epoch": 0.05, "learning_rate": 4.972655057926774e-05, "loss": 0.9352, "step": 47218 }, { "epoch": 0.05, "learning_rate": 4.9726538993347084e-05, "loss": 0.8838, "step": 47219 }, { "epoch": 0.05, "learning_rate": 4.972652740718233e-05, "loss": 0.9446, "step": 47220 }, { "epoch": 0.05, "learning_rate": 4.9726515820773496e-05, "loss": 1.1447, "step": 47221 }, { "epoch": 0.05, "learning_rate": 4.972650423412056e-05, "loss": 0.728, "step": 47222 }, { "epoch": 0.05, "learning_rate": 4.9726492647223544e-05, "loss": 1.0597, "step": 47223 }, { "epoch": 0.05, "learning_rate": 4.972648106008244e-05, "loss": 1.0488, "step": 47224 }, { "epoch": 0.05, "learning_rate": 4.972646947269724e-05, "loss": 1.273, "step": 47225 }, { "epoch": 0.05, "learning_rate": 4.9726457885067954e-05, "loss": 1.9392, "step": 47226 }, { "epoch": 0.05, "learning_rate": 4.972644629719458e-05, "loss": 5.1628, "step": 47227 }, { "epoch": 0.05, "learning_rate": 4.9726434709077116e-05, "loss": 5.0879, "step": 47228 }, { "epoch": 0.05, "learning_rate": 4.972642312071556e-05, "loss": 4.9184, "step": 47229 }, { "epoch": 0.05, "learning_rate": 4.9726411532109915e-05, "loss": 5.4188, "step": 47230 }, { "epoch": 0.05, "learning_rate": 4.972639994326018e-05, "loss": 5.134, "step": 47231 }, { "epoch": 0.05, "learning_rate": 4.972638835416636e-05, "loss": 5.7264, "step": 47232 }, { "epoch": 0.05, "learning_rate": 4.9726376764828454e-05, "loss": 5.2432, "step": 47233 }, { "epoch": 0.05, "learning_rate": 4.972636517524646e-05, "loss": 5.5391, "step": 47234 }, { "epoch": 0.05, "learning_rate": 4.9726353585420374e-05, "loss": 5.8012, "step": 47235 }, { "epoch": 0.05, "learning_rate": 4.97263419953502e-05, "loss": 5.6858, "step": 47236 }, { "epoch": 0.05, "learning_rate": 4.9726330405035937e-05, "loss": 5.8638, "step": 47237 }, { "epoch": 0.05, "learning_rate": 4.972631881447758e-05, "loss": 5.7448, "step": 47238 }, { "epoch": 0.05, "learning_rate": 4.972630722367515e-05, "loss": 5.8938, "step": 47239 }, { "epoch": 0.05, "learning_rate": 4.9726295632628625e-05, "loss": 5.915, "step": 47240 }, { "epoch": 0.05, "learning_rate": 4.9726284041338015e-05, "loss": 3.8882, "step": 47241 }, { "epoch": 0.05, "learning_rate": 4.972627244980331e-05, "loss": 0.9929, "step": 47242 }, { "epoch": 0.05, "learning_rate": 4.9726260858024524e-05, "loss": 0.8271, "step": 47243 }, { "epoch": 0.05, "learning_rate": 4.972624926600164e-05, "loss": 1.1974, "step": 47244 }, { "epoch": 0.05, "learning_rate": 4.972623767373469e-05, "loss": 1.0447, "step": 47245 }, { "epoch": 0.05, "learning_rate": 4.972622608122364e-05, "loss": 1.1897, "step": 47246 }, { "epoch": 0.05, "learning_rate": 4.97262144884685e-05, "loss": 0.886, "step": 47247 }, { "epoch": 0.05, "learning_rate": 4.972620289546928e-05, "loss": 1.0516, "step": 47248 }, { "epoch": 0.05, "learning_rate": 4.9726191302225964e-05, "loss": 1.3022, "step": 47249 }, { "epoch": 0.05, "learning_rate": 4.972617970873858e-05, "loss": 0.8526, "step": 47250 }, { "epoch": 0.05, "learning_rate": 4.9726168115007095e-05, "loss": 0.9775, "step": 47251 }, { "epoch": 0.05, "learning_rate": 4.9726156521031526e-05, "loss": 1.2824, "step": 47252 }, { "epoch": 0.05, "learning_rate": 4.972614492681187e-05, "loss": 1.2457, "step": 47253 }, { "epoch": 0.05, "learning_rate": 4.972613333234813e-05, "loss": 0.675, "step": 47254 }, { "epoch": 0.05, "learning_rate": 4.9726121737640305e-05, "loss": 0.9908, "step": 47255 }, { "epoch": 0.05, "learning_rate": 4.9726110142688384e-05, "loss": 1.0396, "step": 47256 }, { "epoch": 0.05, "learning_rate": 4.9726098547492396e-05, "loss": 0.9609, "step": 47257 }, { "epoch": 0.05, "learning_rate": 4.972608695205231e-05, "loss": 0.6992, "step": 47258 }, { "epoch": 0.05, "learning_rate": 4.972607535636814e-05, "loss": 0.9127, "step": 47259 }, { "epoch": 0.05, "learning_rate": 4.9726063760439885e-05, "loss": 1.0654, "step": 47260 }, { "epoch": 0.05, "learning_rate": 4.9726052164267546e-05, "loss": 0.7891, "step": 47261 }, { "epoch": 0.05, "learning_rate": 4.972604056785112e-05, "loss": 0.7047, "step": 47262 }, { "epoch": 0.05, "learning_rate": 4.972602897119061e-05, "loss": 0.3923, "step": 47263 }, { "epoch": 0.05, "learning_rate": 4.9726017374286015e-05, "loss": 0.3969, "step": 47264 }, { "epoch": 0.05, "learning_rate": 4.972600577713733e-05, "loss": 0.4803, "step": 47265 }, { "epoch": 0.05, "learning_rate": 4.972599417974457e-05, "loss": 0.4822, "step": 47266 }, { "epoch": 0.05, "learning_rate": 4.9725982582107715e-05, "loss": 0.3442, "step": 47267 }, { "epoch": 0.05, "learning_rate": 4.972597098422678e-05, "loss": 0.4049, "step": 47268 }, { "epoch": 0.05, "learning_rate": 4.972595938610176e-05, "loss": 0.3307, "step": 47269 }, { "epoch": 0.05, "learning_rate": 4.9725947787732665e-05, "loss": 0.4603, "step": 47270 }, { "epoch": 0.05, "learning_rate": 4.9725936189119475e-05, "loss": 0.8771, "step": 47271 }, { "epoch": 0.05, "learning_rate": 4.9725924590262204e-05, "loss": 1.0224, "step": 47272 }, { "epoch": 0.05, "learning_rate": 4.972591299116085e-05, "loss": 0.9841, "step": 47273 }, { "epoch": 0.05, "learning_rate": 4.972590139181541e-05, "loss": 0.9809, "step": 47274 }, { "epoch": 0.05, "learning_rate": 4.972588979222589e-05, "loss": 0.7906, "step": 47275 }, { "epoch": 0.05, "learning_rate": 4.972587819239228e-05, "loss": 1.0337, "step": 47276 }, { "epoch": 0.05, "learning_rate": 4.972586659231459e-05, "loss": 1.1348, "step": 47277 }, { "epoch": 0.05, "learning_rate": 4.972585499199283e-05, "loss": 0.9629, "step": 47278 }, { "epoch": 0.05, "learning_rate": 4.972584339142697e-05, "loss": 0.39, "step": 47279 }, { "epoch": 0.05, "learning_rate": 4.9725831790617025e-05, "loss": 0.1751, "step": 47280 }, { "epoch": 0.05, "learning_rate": 4.972582018956301e-05, "loss": 0.0414, "step": 47281 }, { "epoch": 0.05, "learning_rate": 4.9725808588264896e-05, "loss": 0.4028, "step": 47282 }, { "epoch": 0.05, "learning_rate": 4.9725796986722714e-05, "loss": 0.1399, "step": 47283 }, { "epoch": 0.05, "learning_rate": 4.9725785384936444e-05, "loss": 0.1782, "step": 47284 }, { "epoch": 0.05, "learning_rate": 4.972577378290609e-05, "loss": 0.2117, "step": 47285 }, { "epoch": 0.05, "learning_rate": 4.972576218063166e-05, "loss": 0.9372, "step": 47286 }, { "epoch": 0.05, "learning_rate": 4.972575057811314e-05, "loss": 0.961, "step": 47287 }, { "epoch": 0.05, "learning_rate": 4.9725738975350546e-05, "loss": 1.1006, "step": 47288 }, { "epoch": 0.05, "learning_rate": 4.9725727372343864e-05, "loss": 1.0525, "step": 47289 }, { "epoch": 0.05, "learning_rate": 4.9725715769093094e-05, "loss": 1.1554, "step": 47290 }, { "epoch": 0.05, "learning_rate": 4.972570416559825e-05, "loss": 1.052, "step": 47291 }, { "epoch": 0.05, "learning_rate": 4.9725692561859324e-05, "loss": 0.7876, "step": 47292 }, { "epoch": 0.05, "learning_rate": 4.972568095787632e-05, "loss": 0.8809, "step": 47293 }, { "epoch": 0.05, "learning_rate": 4.972566935364923e-05, "loss": 0.821, "step": 47294 }, { "epoch": 0.05, "learning_rate": 4.9725657749178056e-05, "loss": 1.0012, "step": 47295 }, { "epoch": 0.05, "learning_rate": 4.97256461444628e-05, "loss": 1.0809, "step": 47296 }, { "epoch": 0.05, "learning_rate": 4.972563453950346e-05, "loss": 1.0241, "step": 47297 }, { "epoch": 0.05, "learning_rate": 4.972562293430005e-05, "loss": 1.1237, "step": 47298 }, { "epoch": 0.05, "learning_rate": 4.972561132885255e-05, "loss": 0.9638, "step": 47299 }, { "epoch": 0.05, "learning_rate": 4.972559972316097e-05, "loss": 0.9035, "step": 47300 }, { "epoch": 0.05, "learning_rate": 4.972558811722532e-05, "loss": 0.3832, "step": 47301 }, { "epoch": 0.05, "learning_rate": 4.972557651104558e-05, "loss": 0.8784, "step": 47302 }, { "epoch": 0.05, "learning_rate": 4.972556490462177e-05, "loss": 1.0054, "step": 47303 }, { "epoch": 0.05, "learning_rate": 4.972555329795386e-05, "loss": 1.1814, "step": 47304 }, { "epoch": 0.05, "learning_rate": 4.9725541691041885e-05, "loss": 1.1575, "step": 47305 }, { "epoch": 0.05, "learning_rate": 4.972553008388583e-05, "loss": 1.4476, "step": 47306 }, { "epoch": 0.05, "learning_rate": 4.9725518476485685e-05, "loss": 1.0689, "step": 47307 }, { "epoch": 0.05, "learning_rate": 4.9725506868841464e-05, "loss": 1.0057, "step": 47308 }, { "epoch": 0.05, "learning_rate": 4.972549526095317e-05, "loss": 1.0406, "step": 47309 }, { "epoch": 0.05, "learning_rate": 4.972548365282079e-05, "loss": 1.0724, "step": 47310 }, { "epoch": 0.05, "learning_rate": 4.9725472044444334e-05, "loss": 1.2389, "step": 47311 }, { "epoch": 0.05, "learning_rate": 4.9725460435823795e-05, "loss": 1.0601, "step": 47312 }, { "epoch": 0.05, "learning_rate": 4.9725448826959176e-05, "loss": 0.5561, "step": 47313 }, { "epoch": 0.05, "learning_rate": 4.972543721785048e-05, "loss": 0.1846, "step": 47314 }, { "epoch": 0.05, "learning_rate": 4.972542560849771e-05, "loss": 0.1003, "step": 47315 }, { "epoch": 0.05, "learning_rate": 4.972541399890085e-05, "loss": 0.1152, "step": 47316 }, { "epoch": 0.05, "learning_rate": 4.9725402389059914e-05, "loss": 0.0846, "step": 47317 }, { "epoch": 0.05, "learning_rate": 4.97253907789749e-05, "loss": 0.0596, "step": 47318 }, { "epoch": 0.05, "learning_rate": 4.972537916864581e-05, "loss": 0.0429, "step": 47319 }, { "epoch": 0.05, "learning_rate": 4.9725367558072644e-05, "loss": 0.0358, "step": 47320 }, { "epoch": 0.05, "learning_rate": 4.9725355947255404e-05, "loss": 0.2211, "step": 47321 }, { "epoch": 0.05, "learning_rate": 4.9725344336194075e-05, "loss": 0.2923, "step": 47322 }, { "epoch": 0.05, "learning_rate": 4.972533272488867e-05, "loss": 0.2664, "step": 47323 }, { "epoch": 0.05, "learning_rate": 4.972532111333918e-05, "loss": 0.7532, "step": 47324 }, { "epoch": 0.05, "learning_rate": 4.9725309501545624e-05, "loss": 0.9824, "step": 47325 }, { "epoch": 0.05, "learning_rate": 4.972529788950799e-05, "loss": 0.9214, "step": 47326 }, { "epoch": 0.05, "learning_rate": 4.972528627722627e-05, "loss": 0.9849, "step": 47327 }, { "epoch": 0.05, "learning_rate": 4.9725274664700485e-05, "loss": 0.7678, "step": 47328 }, { "epoch": 0.05, "learning_rate": 4.9725263051930616e-05, "loss": 0.1669, "step": 47329 }, { "epoch": 0.05, "learning_rate": 4.972525143891667e-05, "loss": 0.1965, "step": 47330 }, { "epoch": 0.05, "learning_rate": 4.9725239825658636e-05, "loss": 0.2651, "step": 47331 }, { "epoch": 0.05, "learning_rate": 4.972522821215654e-05, "loss": 0.3871, "step": 47332 }, { "epoch": 0.05, "learning_rate": 4.972521659841036e-05, "loss": 0.169, "step": 47333 }, { "epoch": 0.05, "learning_rate": 4.9725204984420106e-05, "loss": 0.1831, "step": 47334 }, { "epoch": 0.05, "learning_rate": 4.9725193370185765e-05, "loss": 0.3821, "step": 47335 }, { "epoch": 0.05, "learning_rate": 4.972518175570736e-05, "loss": 0.9786, "step": 47336 }, { "epoch": 0.05, "learning_rate": 4.972517014098487e-05, "loss": 1.3002, "step": 47337 }, { "epoch": 0.05, "learning_rate": 4.972515852601831e-05, "loss": 1.2876, "step": 47338 }, { "epoch": 0.05, "learning_rate": 4.9725146910807666e-05, "loss": 1.1376, "step": 47339 }, { "epoch": 0.05, "learning_rate": 4.9725135295352954e-05, "loss": 1.2638, "step": 47340 }, { "epoch": 0.05, "learning_rate": 4.972512367965416e-05, "loss": 0.903, "step": 47341 }, { "epoch": 0.05, "learning_rate": 4.972511206371129e-05, "loss": 0.9893, "step": 47342 }, { "epoch": 0.05, "learning_rate": 4.972510044752435e-05, "loss": 0.2606, "step": 47343 }, { "epoch": 0.05, "learning_rate": 4.972508883109333e-05, "loss": 0.0689, "step": 47344 }, { "epoch": 0.05, "learning_rate": 4.972507721441824e-05, "loss": 0.2565, "step": 47345 }, { "epoch": 0.05, "learning_rate": 4.972506559749907e-05, "loss": 0.6409, "step": 47346 }, { "epoch": 0.05, "learning_rate": 4.972505398033582e-05, "loss": 0.7309, "step": 47347 }, { "epoch": 0.05, "learning_rate": 4.9725042362928496e-05, "loss": 0.9056, "step": 47348 }, { "epoch": 0.05, "learning_rate": 4.97250307452771e-05, "loss": 0.8961, "step": 47349 }, { "epoch": 0.05, "learning_rate": 4.9725019127381634e-05, "loss": 0.9306, "step": 47350 }, { "epoch": 0.05, "learning_rate": 4.9725007509242085e-05, "loss": 1.0392, "step": 47351 }, { "epoch": 0.05, "learning_rate": 4.972499589085846e-05, "loss": 1.1059, "step": 47352 }, { "epoch": 0.05, "learning_rate": 4.972498427223077e-05, "loss": 0.7465, "step": 47353 }, { "epoch": 0.05, "learning_rate": 4.972497265335899e-05, "loss": 0.7445, "step": 47354 }, { "epoch": 0.05, "learning_rate": 4.972496103424315e-05, "loss": 0.8073, "step": 47355 }, { "epoch": 0.05, "learning_rate": 4.972494941488323e-05, "loss": 0.6767, "step": 47356 }, { "epoch": 0.05, "learning_rate": 4.972493779527924e-05, "loss": 0.7559, "step": 47357 }, { "epoch": 0.05, "learning_rate": 4.972492617543117e-05, "loss": 0.7023, "step": 47358 }, { "epoch": 0.05, "learning_rate": 4.972491455533903e-05, "loss": 1.4445, "step": 47359 }, { "epoch": 0.05, "learning_rate": 4.9724902935002815e-05, "loss": 1.0933, "step": 47360 }, { "epoch": 0.05, "learning_rate": 4.9724891314422524e-05, "loss": 0.8464, "step": 47361 }, { "epoch": 0.05, "learning_rate": 4.972487969359816e-05, "loss": 1.3321, "step": 47362 }, { "epoch": 0.05, "learning_rate": 4.972486807252972e-05, "loss": 0.9723, "step": 47363 }, { "epoch": 0.05, "learning_rate": 4.972485645121721e-05, "loss": 0.8759, "step": 47364 }, { "epoch": 0.05, "learning_rate": 4.972484482966063e-05, "loss": 1.0955, "step": 47365 }, { "epoch": 0.05, "learning_rate": 4.9724833207859967e-05, "loss": 1.1391, "step": 47366 }, { "epoch": 0.05, "learning_rate": 4.972482158581524e-05, "loss": 1.0003, "step": 47367 }, { "epoch": 0.05, "learning_rate": 4.972480996352644e-05, "loss": 1.1062, "step": 47368 }, { "epoch": 0.05, "learning_rate": 4.972479834099356e-05, "loss": 1.2483, "step": 47369 }, { "epoch": 0.05, "learning_rate": 4.972478671821661e-05, "loss": 1.0486, "step": 47370 }, { "epoch": 0.05, "learning_rate": 4.972477509519559e-05, "loss": 0.5602, "step": 47371 }, { "epoch": 0.05, "learning_rate": 4.972476347193049e-05, "loss": 1.009, "step": 47372 }, { "epoch": 0.05, "learning_rate": 4.9724751848421324e-05, "loss": 0.986, "step": 47373 }, { "epoch": 0.05, "learning_rate": 4.972474022466809e-05, "loss": 1.0623, "step": 47374 }, { "epoch": 0.05, "learning_rate": 4.972472860067078e-05, "loss": 1.1278, "step": 47375 }, { "epoch": 0.05, "learning_rate": 4.972471697642939e-05, "loss": 1.0449, "step": 47376 }, { "epoch": 0.05, "learning_rate": 4.972470535194393e-05, "loss": 0.855, "step": 47377 }, { "epoch": 0.05, "learning_rate": 4.9724693727214414e-05, "loss": 1.0394, "step": 47378 }, { "epoch": 0.05, "learning_rate": 4.972468210224081e-05, "loss": 0.7867, "step": 47379 }, { "epoch": 0.05, "learning_rate": 4.972467047702314e-05, "loss": 1.0457, "step": 47380 }, { "epoch": 0.05, "learning_rate": 4.97246588515614e-05, "loss": 1.2962, "step": 47381 }, { "epoch": 0.05, "learning_rate": 4.972464722585558e-05, "loss": 1.0096, "step": 47382 }, { "epoch": 0.05, "learning_rate": 4.97246355999057e-05, "loss": 1.1025, "step": 47383 }, { "epoch": 0.05, "learning_rate": 4.9724623973711744e-05, "loss": 0.9643, "step": 47384 }, { "epoch": 0.05, "learning_rate": 4.972461234727371e-05, "loss": 0.9582, "step": 47385 }, { "epoch": 0.05, "learning_rate": 4.972460072059162e-05, "loss": 0.8803, "step": 47386 }, { "epoch": 0.05, "learning_rate": 4.972458909366545e-05, "loss": 0.9371, "step": 47387 }, { "epoch": 0.05, "learning_rate": 4.972457746649521e-05, "loss": 0.907, "step": 47388 }, { "epoch": 0.05, "learning_rate": 4.97245658390809e-05, "loss": 0.883, "step": 47389 }, { "epoch": 0.05, "learning_rate": 4.9724554211422516e-05, "loss": 0.8377, "step": 47390 }, { "epoch": 0.05, "learning_rate": 4.9724542583520065e-05, "loss": 0.799, "step": 47391 }, { "epoch": 0.05, "learning_rate": 4.972453095537355e-05, "loss": 0.821, "step": 47392 }, { "epoch": 0.05, "learning_rate": 4.9724519326982954e-05, "loss": 1.2618, "step": 47393 }, { "epoch": 0.05, "learning_rate": 4.9724507698348294e-05, "loss": 1.2119, "step": 47394 }, { "epoch": 0.05, "learning_rate": 4.972449606946956e-05, "loss": 0.9531, "step": 47395 }, { "epoch": 0.05, "learning_rate": 4.9724484440346765e-05, "loss": 0.6227, "step": 47396 }, { "epoch": 0.05, "learning_rate": 4.972447281097989e-05, "loss": 1.0859, "step": 47397 }, { "epoch": 0.05, "learning_rate": 4.9724461181368946e-05, "loss": 1.1658, "step": 47398 }, { "epoch": 0.05, "learning_rate": 4.972444955151394e-05, "loss": 1.1444, "step": 47399 }, { "epoch": 0.05, "learning_rate": 4.9724437921414857e-05, "loss": 0.9993, "step": 47400 }, { "epoch": 0.05, "learning_rate": 4.972442629107171e-05, "loss": 0.7827, "step": 47401 }, { "epoch": 0.05, "learning_rate": 4.972441466048449e-05, "loss": 0.9202, "step": 47402 }, { "epoch": 0.05, "learning_rate": 4.97244030296532e-05, "loss": 0.7523, "step": 47403 }, { "epoch": 0.05, "learning_rate": 4.9724391398577855e-05, "loss": 0.8437, "step": 47404 }, { "epoch": 0.05, "learning_rate": 4.972437976725842e-05, "loss": 1.0582, "step": 47405 }, { "epoch": 0.05, "learning_rate": 4.972436813569493e-05, "loss": 0.9717, "step": 47406 }, { "epoch": 0.05, "learning_rate": 4.972435650388737e-05, "loss": 0.9788, "step": 47407 }, { "epoch": 0.05, "learning_rate": 4.972434487183574e-05, "loss": 1.1435, "step": 47408 }, { "epoch": 0.05, "learning_rate": 4.972433323954004e-05, "loss": 1.0504, "step": 47409 }, { "epoch": 0.05, "learning_rate": 4.9724321607000275e-05, "loss": 0.7498, "step": 47410 }, { "epoch": 0.05, "learning_rate": 4.972430997421644e-05, "loss": 0.9077, "step": 47411 }, { "epoch": 0.05, "learning_rate": 4.972429834118853e-05, "loss": 1.0682, "step": 47412 }, { "epoch": 0.05, "learning_rate": 4.972428670791657e-05, "loss": 1.5217, "step": 47413 }, { "epoch": 0.05, "learning_rate": 4.972427507440053e-05, "loss": 1.1577, "step": 47414 }, { "epoch": 0.05, "learning_rate": 4.9724263440640423e-05, "loss": 0.9272, "step": 47415 }, { "epoch": 0.05, "learning_rate": 4.9724251806636245e-05, "loss": 1.1134, "step": 47416 }, { "epoch": 0.05, "learning_rate": 4.972424017238801e-05, "loss": 0.9286, "step": 47417 }, { "epoch": 0.05, "learning_rate": 4.97242285378957e-05, "loss": 1.0745, "step": 47418 }, { "epoch": 0.05, "learning_rate": 4.972421690315933e-05, "loss": 0.9756, "step": 47419 }, { "epoch": 0.05, "learning_rate": 4.972420526817888e-05, "loss": 0.9483, "step": 47420 }, { "epoch": 0.05, "learning_rate": 4.972419363295438e-05, "loss": 0.9807, "step": 47421 }, { "epoch": 0.05, "learning_rate": 4.9724181997485794e-05, "loss": 0.9794, "step": 47422 }, { "epoch": 0.05, "learning_rate": 4.972417036177315e-05, "loss": 1.1824, "step": 47423 }, { "epoch": 0.05, "learning_rate": 4.9724158725816447e-05, "loss": 0.8721, "step": 47424 }, { "epoch": 0.05, "learning_rate": 4.972414708961567e-05, "loss": 1.062, "step": 47425 }, { "epoch": 0.05, "learning_rate": 4.972413545317083e-05, "loss": 0.9113, "step": 47426 }, { "epoch": 0.05, "learning_rate": 4.9724123816481916e-05, "loss": 1.0698, "step": 47427 }, { "epoch": 0.05, "learning_rate": 4.972411217954894e-05, "loss": 1.1072, "step": 47428 }, { "epoch": 0.05, "learning_rate": 4.972410054237191e-05, "loss": 0.9901, "step": 47429 }, { "epoch": 0.05, "learning_rate": 4.972408890495079e-05, "loss": 0.8218, "step": 47430 }, { "epoch": 0.05, "learning_rate": 4.972407726728563e-05, "loss": 0.8465, "step": 47431 }, { "epoch": 0.05, "learning_rate": 4.9724065629376384e-05, "loss": 0.9426, "step": 47432 }, { "epoch": 0.05, "learning_rate": 4.972405399122309e-05, "loss": 1.5024, "step": 47433 }, { "epoch": 0.05, "learning_rate": 4.9724042352825715e-05, "loss": 1.2873, "step": 47434 }, { "epoch": 0.05, "learning_rate": 4.9724030714184275e-05, "loss": 1.2534, "step": 47435 }, { "epoch": 0.05, "learning_rate": 4.9724019075298775e-05, "loss": 1.0371, "step": 47436 }, { "epoch": 0.05, "learning_rate": 4.972400743616922e-05, "loss": 1.0383, "step": 47437 }, { "epoch": 0.05, "learning_rate": 4.9723995796795586e-05, "loss": 0.7222, "step": 47438 }, { "epoch": 0.05, "learning_rate": 4.97239841571779e-05, "loss": 1.0986, "step": 47439 }, { "epoch": 0.05, "learning_rate": 4.9723972517316133e-05, "loss": 0.8258, "step": 47440 }, { "epoch": 0.05, "learning_rate": 4.972396087721031e-05, "loss": 0.956, "step": 47441 }, { "epoch": 0.05, "learning_rate": 4.9723949236860425e-05, "loss": 1.0015, "step": 47442 }, { "epoch": 0.05, "learning_rate": 4.9723937596266466e-05, "loss": 0.932, "step": 47443 }, { "epoch": 0.05, "learning_rate": 4.972392595542845e-05, "loss": 0.447, "step": 47444 }, { "epoch": 0.05, "learning_rate": 4.972391431434637e-05, "loss": 1.083, "step": 47445 }, { "epoch": 0.05, "learning_rate": 4.9723902673020225e-05, "loss": 1.0155, "step": 47446 }, { "epoch": 0.05, "learning_rate": 4.972389103145002e-05, "loss": 0.9779, "step": 47447 }, { "epoch": 0.05, "learning_rate": 4.972387938963574e-05, "loss": 1.0168, "step": 47448 }, { "epoch": 0.05, "learning_rate": 4.9723867747577405e-05, "loss": 1.2087, "step": 47449 }, { "epoch": 0.05, "learning_rate": 4.9723856105275e-05, "loss": 0.7228, "step": 47450 }, { "epoch": 0.05, "learning_rate": 4.972384446272854e-05, "loss": 0.8132, "step": 47451 }, { "epoch": 0.05, "learning_rate": 4.972383281993802e-05, "loss": 0.8029, "step": 47452 }, { "epoch": 0.05, "learning_rate": 4.9723821176903425e-05, "loss": 0.8649, "step": 47453 }, { "epoch": 0.05, "learning_rate": 4.972380953362477e-05, "loss": 1.1776, "step": 47454 }, { "epoch": 0.05, "learning_rate": 4.972379789010205e-05, "loss": 1.0307, "step": 47455 }, { "epoch": 0.05, "learning_rate": 4.9723786246335275e-05, "loss": 1.109, "step": 47456 }, { "epoch": 0.05, "learning_rate": 4.972377460232443e-05, "loss": 1.0875, "step": 47457 }, { "epoch": 0.05, "learning_rate": 4.972376295806952e-05, "loss": 1.063, "step": 47458 }, { "epoch": 0.05, "learning_rate": 4.972375131357055e-05, "loss": 0.9804, "step": 47459 }, { "epoch": 0.05, "learning_rate": 4.972373966882753e-05, "loss": 0.8742, "step": 47460 }, { "epoch": 0.05, "learning_rate": 4.972372802384043e-05, "loss": 0.5911, "step": 47461 }, { "epoch": 0.05, "learning_rate": 4.9723716378609284e-05, "loss": 0.4932, "step": 47462 }, { "epoch": 0.05, "learning_rate": 4.9723704733134064e-05, "loss": 0.435, "step": 47463 }, { "epoch": 0.05, "learning_rate": 4.972369308741478e-05, "loss": 0.9755, "step": 47464 }, { "epoch": 0.05, "learning_rate": 4.972368144145144e-05, "loss": 1.0935, "step": 47465 }, { "epoch": 0.05, "learning_rate": 4.9723669795244047e-05, "loss": 1.097, "step": 47466 }, { "epoch": 0.05, "learning_rate": 4.972365814879258e-05, "loss": 1.121, "step": 47467 }, { "epoch": 0.05, "learning_rate": 4.9723646502097054e-05, "loss": 0.9515, "step": 47468 }, { "epoch": 0.05, "learning_rate": 4.972363485515746e-05, "loss": 0.6616, "step": 47469 }, { "epoch": 0.05, "learning_rate": 4.972362320797382e-05, "loss": 0.772, "step": 47470 }, { "epoch": 0.05, "learning_rate": 4.9723611560546106e-05, "loss": 0.8237, "step": 47471 }, { "epoch": 0.05, "learning_rate": 4.972359991287434e-05, "loss": 1.0167, "step": 47472 }, { "epoch": 0.05, "learning_rate": 4.972358826495851e-05, "loss": 1.3674, "step": 47473 }, { "epoch": 0.05, "learning_rate": 4.972357661679862e-05, "loss": 0.8622, "step": 47474 }, { "epoch": 0.05, "learning_rate": 4.9723564968394657e-05, "loss": 1.5365, "step": 47475 }, { "epoch": 0.05, "learning_rate": 4.972355331974665e-05, "loss": 1.7409, "step": 47476 }, { "epoch": 0.05, "learning_rate": 4.972354167085457e-05, "loss": 1.1145, "step": 47477 }, { "epoch": 0.05, "learning_rate": 4.9723530021718434e-05, "loss": 0.6862, "step": 47478 }, { "epoch": 0.05, "learning_rate": 4.972351837233824e-05, "loss": 0.8678, "step": 47479 }, { "epoch": 0.05, "learning_rate": 4.9723506722713985e-05, "loss": 0.9923, "step": 47480 }, { "epoch": 0.05, "learning_rate": 4.972349507284567e-05, "loss": 0.5492, "step": 47481 }, { "epoch": 0.05, "learning_rate": 4.97234834227333e-05, "loss": 1.0498, "step": 47482 }, { "epoch": 0.05, "learning_rate": 4.972347177237686e-05, "loss": 1.0002, "step": 47483 }, { "epoch": 0.05, "learning_rate": 4.9723460121776365e-05, "loss": 0.8533, "step": 47484 }, { "epoch": 0.05, "learning_rate": 4.972344847093181e-05, "loss": 0.9885, "step": 47485 }, { "epoch": 0.05, "learning_rate": 4.97234368198432e-05, "loss": 0.9477, "step": 47486 }, { "epoch": 0.05, "learning_rate": 4.972342516851053e-05, "loss": 1.1564, "step": 47487 }, { "epoch": 0.05, "learning_rate": 4.9723413516933794e-05, "loss": 1.9091, "step": 47488 }, { "epoch": 0.05, "learning_rate": 4.972340186511301e-05, "loss": 1.3725, "step": 47489 }, { "epoch": 0.05, "learning_rate": 4.972339021304815e-05, "loss": 1.2152, "step": 47490 }, { "epoch": 0.05, "learning_rate": 4.9723378560739245e-05, "loss": 1.1971, "step": 47491 }, { "epoch": 0.05, "learning_rate": 4.972336690818627e-05, "loss": 0.8745, "step": 47492 }, { "epoch": 0.05, "learning_rate": 4.9723355255389245e-05, "loss": 0.8613, "step": 47493 }, { "epoch": 0.05, "learning_rate": 4.9723343602348164e-05, "loss": 1.2347, "step": 47494 }, { "epoch": 0.05, "learning_rate": 4.972333194906302e-05, "loss": 1.0256, "step": 47495 }, { "epoch": 0.05, "learning_rate": 4.9723320295533814e-05, "loss": 0.954, "step": 47496 }, { "epoch": 0.05, "learning_rate": 4.972330864176056e-05, "loss": 0.9583, "step": 47497 }, { "epoch": 0.05, "learning_rate": 4.972329698774324e-05, "loss": 1.0692, "step": 47498 }, { "epoch": 0.05, "learning_rate": 4.972328533348186e-05, "loss": 0.9339, "step": 47499 }, { "epoch": 0.05, "learning_rate": 4.972327367897643e-05, "loss": 1.5265, "step": 47500 }, { "epoch": 0.05, "eval_loss": 1.079487919807434, "eval_runtime": 126.9614, "eval_samples_per_second": 10.909, "eval_steps_per_second": 5.458, "step": 47500 }, { "epoch": 0.05, "learning_rate": 4.972326202422693e-05, "loss": 1.184, "step": 47501 }, { "epoch": 0.05, "learning_rate": 4.972325036923339e-05, "loss": 0.9142, "step": 47502 }, { "epoch": 0.05, "learning_rate": 4.972323871399578e-05, "loss": 1.257, "step": 47503 }, { "epoch": 0.05, "learning_rate": 4.972322705851412e-05, "loss": 1.7796, "step": 47504 }, { "epoch": 0.05, "learning_rate": 4.97232154027884e-05, "loss": 1.6747, "step": 47505 }, { "epoch": 0.05, "learning_rate": 4.9723203746818616e-05, "loss": 1.0056, "step": 47506 }, { "epoch": 0.05, "learning_rate": 4.972319209060478e-05, "loss": 1.1675, "step": 47507 }, { "epoch": 0.05, "learning_rate": 4.972318043414689e-05, "loss": 0.9335, "step": 47508 }, { "epoch": 0.05, "learning_rate": 4.972316877744494e-05, "loss": 1.0005, "step": 47509 }, { "epoch": 0.05, "learning_rate": 4.972315712049893e-05, "loss": 0.9524, "step": 47510 }, { "epoch": 0.05, "learning_rate": 4.972314546330887e-05, "loss": 0.4542, "step": 47511 }, { "epoch": 0.05, "learning_rate": 4.972313380587476e-05, "loss": 1.0785, "step": 47512 }, { "epoch": 0.05, "learning_rate": 4.972312214819658e-05, "loss": 1.089, "step": 47513 }, { "epoch": 0.05, "learning_rate": 4.9723110490274344e-05, "loss": 0.8167, "step": 47514 }, { "epoch": 0.05, "learning_rate": 4.972309883210806e-05, "loss": 1.0724, "step": 47515 }, { "epoch": 0.05, "learning_rate": 4.9723087173697715e-05, "loss": 0.9436, "step": 47516 }, { "epoch": 0.05, "learning_rate": 4.972307551504331e-05, "loss": 1.3269, "step": 47517 }, { "epoch": 0.05, "learning_rate": 4.972306385614486e-05, "loss": 0.9477, "step": 47518 }, { "epoch": 0.05, "learning_rate": 4.972305219700235e-05, "loss": 1.1268, "step": 47519 }, { "epoch": 0.05, "learning_rate": 4.972304053761578e-05, "loss": 1.3103, "step": 47520 }, { "epoch": 0.05, "learning_rate": 4.972302887798516e-05, "loss": 0.6856, "step": 47521 }, { "epoch": 0.05, "learning_rate": 4.972301721811048e-05, "loss": 1.0998, "step": 47522 }, { "epoch": 0.05, "learning_rate": 4.972300555799174e-05, "loss": 0.5727, "step": 47523 }, { "epoch": 0.05, "learning_rate": 4.9722993897628955e-05, "loss": 0.9988, "step": 47524 }, { "epoch": 0.05, "learning_rate": 4.9722982237022115e-05, "loss": 1.0005, "step": 47525 }, { "epoch": 0.05, "learning_rate": 4.9722970576171214e-05, "loss": 0.8141, "step": 47526 }, { "epoch": 0.05, "learning_rate": 4.972295891507626e-05, "loss": 0.3928, "step": 47527 }, { "epoch": 0.05, "learning_rate": 4.972294725373726e-05, "loss": 0.6103, "step": 47528 }, { "epoch": 0.05, "learning_rate": 4.9722935592154195e-05, "loss": 1.2807, "step": 47529 }, { "epoch": 0.05, "learning_rate": 4.972292393032708e-05, "loss": 1.2436, "step": 47530 }, { "epoch": 0.05, "learning_rate": 4.972291226825591e-05, "loss": 1.36, "step": 47531 }, { "epoch": 0.05, "learning_rate": 4.9722900605940683e-05, "loss": 1.1032, "step": 47532 }, { "epoch": 0.05, "learning_rate": 4.9722888943381405e-05, "loss": 1.0128, "step": 47533 }, { "epoch": 0.05, "learning_rate": 4.972287728057807e-05, "loss": 0.6976, "step": 47534 }, { "epoch": 0.05, "learning_rate": 4.972286561753069e-05, "loss": 0.9742, "step": 47535 }, { "epoch": 0.05, "learning_rate": 4.972285395423925e-05, "loss": 1.1171, "step": 47536 }, { "epoch": 0.05, "learning_rate": 4.972284229070375e-05, "loss": 0.9876, "step": 47537 }, { "epoch": 0.05, "learning_rate": 4.9722830626924205e-05, "loss": 0.7511, "step": 47538 }, { "epoch": 0.05, "learning_rate": 4.9722818962900604e-05, "loss": 1.2316, "step": 47539 }, { "epoch": 0.05, "learning_rate": 4.972280729863295e-05, "loss": 1.1524, "step": 47540 }, { "epoch": 0.05, "learning_rate": 4.9722795634121245e-05, "loss": 0.7962, "step": 47541 }, { "epoch": 0.05, "learning_rate": 4.972278396936548e-05, "loss": 0.9375, "step": 47542 }, { "epoch": 0.05, "learning_rate": 4.972277230436567e-05, "loss": 1.2624, "step": 47543 }, { "epoch": 0.05, "learning_rate": 4.97227606391218e-05, "loss": 1.0783, "step": 47544 }, { "epoch": 0.05, "learning_rate": 4.972274897363388e-05, "loss": 0.9241, "step": 47545 }, { "epoch": 0.05, "learning_rate": 4.972273730790191e-05, "loss": 0.696, "step": 47546 }, { "epoch": 0.05, "learning_rate": 4.97227256419259e-05, "loss": 1.319, "step": 47547 }, { "epoch": 0.05, "learning_rate": 4.972271397570582e-05, "loss": 0.8645, "step": 47548 }, { "epoch": 0.05, "learning_rate": 4.972270230924169e-05, "loss": 1.0405, "step": 47549 }, { "epoch": 0.05, "learning_rate": 4.972269064253351e-05, "loss": 1.2527, "step": 47550 }, { "epoch": 0.05, "learning_rate": 4.9722678975581273e-05, "loss": 0.9075, "step": 47551 }, { "epoch": 0.05, "learning_rate": 4.972266730838499e-05, "loss": 1.0558, "step": 47552 }, { "epoch": 0.05, "learning_rate": 4.972265564094466e-05, "loss": 1.1986, "step": 47553 }, { "epoch": 0.05, "learning_rate": 4.972264397326027e-05, "loss": 0.8642, "step": 47554 }, { "epoch": 0.05, "learning_rate": 4.972263230533183e-05, "loss": 0.9598, "step": 47555 }, { "epoch": 0.05, "learning_rate": 4.972262063715934e-05, "loss": 0.7168, "step": 47556 }, { "epoch": 0.05, "learning_rate": 4.97226089687428e-05, "loss": 0.9344, "step": 47557 }, { "epoch": 0.05, "learning_rate": 4.972259730008221e-05, "loss": 1.151, "step": 47558 }, { "epoch": 0.05, "learning_rate": 4.972258563117757e-05, "loss": 1.4403, "step": 47559 }, { "epoch": 0.05, "learning_rate": 4.972257396202887e-05, "loss": 0.9698, "step": 47560 }, { "epoch": 0.05, "learning_rate": 4.972256229263612e-05, "loss": 1.2386, "step": 47561 }, { "epoch": 0.05, "learning_rate": 4.9722550622999327e-05, "loss": 1.0838, "step": 47562 }, { "epoch": 0.05, "learning_rate": 4.972253895311848e-05, "loss": 1.0081, "step": 47563 }, { "epoch": 0.05, "learning_rate": 4.972252728299358e-05, "loss": 1.115, "step": 47564 }, { "epoch": 0.05, "learning_rate": 4.9722515612624634e-05, "loss": 0.9838, "step": 47565 }, { "epoch": 0.05, "learning_rate": 4.972250394201163e-05, "loss": 0.9051, "step": 47566 }, { "epoch": 0.05, "learning_rate": 4.972249227115459e-05, "loss": 0.8142, "step": 47567 }, { "epoch": 0.05, "learning_rate": 4.972248060005349e-05, "loss": 0.8631, "step": 47568 }, { "epoch": 0.05, "learning_rate": 4.9722468928708346e-05, "loss": 1.3519, "step": 47569 }, { "epoch": 0.05, "learning_rate": 4.972245725711914e-05, "loss": 1.2532, "step": 47570 }, { "epoch": 0.05, "learning_rate": 4.9722445585285895e-05, "loss": 0.8689, "step": 47571 }, { "epoch": 0.05, "learning_rate": 4.9722433913208596e-05, "loss": 1.097, "step": 47572 }, { "epoch": 0.05, "learning_rate": 4.972242224088725e-05, "loss": 0.9888, "step": 47573 }, { "epoch": 0.05, "learning_rate": 4.9722410568321856e-05, "loss": 1.0036, "step": 47574 }, { "epoch": 0.05, "learning_rate": 4.9722398895512415e-05, "loss": 1.1234, "step": 47575 }, { "epoch": 0.05, "learning_rate": 4.9722387222458914e-05, "loss": 1.0608, "step": 47576 }, { "epoch": 0.05, "learning_rate": 4.972237554916137e-05, "loss": 0.967, "step": 47577 }, { "epoch": 0.05, "learning_rate": 4.972236387561978e-05, "loss": 1.0022, "step": 47578 }, { "epoch": 0.05, "learning_rate": 4.972235220183415e-05, "loss": 1.2565, "step": 47579 }, { "epoch": 0.05, "learning_rate": 4.972234052780445e-05, "loss": 0.9437, "step": 47580 }, { "epoch": 0.05, "learning_rate": 4.972232885353071e-05, "loss": 0.9624, "step": 47581 }, { "epoch": 0.05, "learning_rate": 4.972231717901293e-05, "loss": 1.2095, "step": 47582 }, { "epoch": 0.05, "learning_rate": 4.972230550425109e-05, "loss": 1.5633, "step": 47583 }, { "epoch": 0.05, "learning_rate": 4.972229382924521e-05, "loss": 1.5075, "step": 47584 }, { "epoch": 0.05, "learning_rate": 4.972228215399528e-05, "loss": 1.122, "step": 47585 }, { "epoch": 0.05, "learning_rate": 4.97222704785013e-05, "loss": 1.0618, "step": 47586 }, { "epoch": 0.05, "learning_rate": 4.9722258802763273e-05, "loss": 0.3949, "step": 47587 }, { "epoch": 0.05, "learning_rate": 4.972224712678119e-05, "loss": 1.2002, "step": 47588 }, { "epoch": 0.05, "learning_rate": 4.972223545055508e-05, "loss": 0.9827, "step": 47589 }, { "epoch": 0.05, "learning_rate": 4.972222377408491e-05, "loss": 1.0788, "step": 47590 }, { "epoch": 0.05, "learning_rate": 4.972221209737069e-05, "loss": 0.932, "step": 47591 }, { "epoch": 0.05, "learning_rate": 4.9722200420412426e-05, "loss": 1.0133, "step": 47592 }, { "epoch": 0.05, "learning_rate": 4.972218874321011e-05, "loss": 0.9323, "step": 47593 }, { "epoch": 0.05, "learning_rate": 4.972217706576375e-05, "loss": 0.9646, "step": 47594 }, { "epoch": 0.05, "learning_rate": 4.972216538807335e-05, "loss": 0.9314, "step": 47595 }, { "epoch": 0.05, "learning_rate": 4.97221537101389e-05, "loss": 1.0897, "step": 47596 }, { "epoch": 0.05, "learning_rate": 4.97221420319604e-05, "loss": 0.9972, "step": 47597 }, { "epoch": 0.05, "learning_rate": 4.9722130353537856e-05, "loss": 1.1032, "step": 47598 }, { "epoch": 0.05, "learning_rate": 4.972211867487126e-05, "loss": 0.8286, "step": 47599 }, { "epoch": 0.05, "learning_rate": 4.9722106995960624e-05, "loss": 0.9249, "step": 47600 }, { "epoch": 0.05, "learning_rate": 4.972209531680594e-05, "loss": 0.964, "step": 47601 }, { "epoch": 0.05, "learning_rate": 4.972208363740721e-05, "loss": 1.0891, "step": 47602 }, { "epoch": 0.05, "learning_rate": 4.972207195776443e-05, "loss": 1.0506, "step": 47603 }, { "epoch": 0.05, "learning_rate": 4.9722060277877615e-05, "loss": 1.0986, "step": 47604 }, { "epoch": 0.05, "learning_rate": 4.972204859774674e-05, "loss": 1.157, "step": 47605 }, { "epoch": 0.05, "learning_rate": 4.972203691737183e-05, "loss": 0.9802, "step": 47606 }, { "epoch": 0.05, "learning_rate": 4.972202523675287e-05, "loss": 1.4229, "step": 47607 }, { "epoch": 0.05, "learning_rate": 4.9722013555889866e-05, "loss": 0.9283, "step": 47608 }, { "epoch": 0.05, "learning_rate": 4.9722001874782816e-05, "loss": 0.9217, "step": 47609 }, { "epoch": 0.05, "learning_rate": 4.9721990193431726e-05, "loss": 0.9343, "step": 47610 }, { "epoch": 0.05, "learning_rate": 4.972197851183658e-05, "loss": 1.143, "step": 47611 }, { "epoch": 0.05, "learning_rate": 4.972196682999739e-05, "loss": 0.8308, "step": 47612 }, { "epoch": 0.05, "learning_rate": 4.9721955147914165e-05, "loss": 1.2084, "step": 47613 }, { "epoch": 0.05, "learning_rate": 4.972194346558689e-05, "loss": 0.5448, "step": 47614 }, { "epoch": 0.05, "learning_rate": 4.972193178301557e-05, "loss": 0.1711, "step": 47615 }, { "epoch": 0.05, "learning_rate": 4.97219201002002e-05, "loss": 0.1445, "step": 47616 }, { "epoch": 0.05, "learning_rate": 4.972190841714079e-05, "loss": 1.0674, "step": 47617 }, { "epoch": 0.05, "learning_rate": 4.9721896733837346e-05, "loss": 0.7391, "step": 47618 }, { "epoch": 0.05, "learning_rate": 4.972188505028985e-05, "loss": 1.025, "step": 47619 }, { "epoch": 0.05, "learning_rate": 4.9721873366498304e-05, "loss": 0.9227, "step": 47620 }, { "epoch": 0.05, "learning_rate": 4.972186168246272e-05, "loss": 0.7861, "step": 47621 }, { "epoch": 0.05, "learning_rate": 4.972184999818309e-05, "loss": 0.9628, "step": 47622 }, { "epoch": 0.05, "learning_rate": 4.9721838313659424e-05, "loss": 0.4995, "step": 47623 }, { "epoch": 0.05, "learning_rate": 4.972182662889171e-05, "loss": 0.9367, "step": 47624 }, { "epoch": 0.05, "learning_rate": 4.972181494387995e-05, "loss": 1.2652, "step": 47625 }, { "epoch": 0.05, "learning_rate": 4.972180325862415e-05, "loss": 0.9921, "step": 47626 }, { "epoch": 0.05, "learning_rate": 4.9721791573124296e-05, "loss": 0.7729, "step": 47627 }, { "epoch": 0.05, "learning_rate": 4.972177988738042e-05, "loss": 1.1216, "step": 47628 }, { "epoch": 0.05, "learning_rate": 4.972176820139248e-05, "loss": 1.2712, "step": 47629 }, { "epoch": 0.05, "learning_rate": 4.972175651516051e-05, "loss": 1.2716, "step": 47630 }, { "epoch": 0.05, "learning_rate": 4.972174482868449e-05, "loss": 1.1832, "step": 47631 }, { "epoch": 0.05, "learning_rate": 4.972173314196443e-05, "loss": 0.9865, "step": 47632 }, { "epoch": 0.05, "learning_rate": 4.972172145500033e-05, "loss": 1.768, "step": 47633 }, { "epoch": 0.05, "learning_rate": 4.972170976779218e-05, "loss": 0.9221, "step": 47634 }, { "epoch": 0.05, "learning_rate": 4.9721698080339994e-05, "loss": 1.261, "step": 47635 }, { "epoch": 0.05, "learning_rate": 4.972168639264376e-05, "loss": 1.4001, "step": 47636 }, { "epoch": 0.05, "learning_rate": 4.972167470470349e-05, "loss": 0.3266, "step": 47637 }, { "epoch": 0.05, "learning_rate": 4.972166301651918e-05, "loss": 0.0808, "step": 47638 }, { "epoch": 0.05, "learning_rate": 4.972165132809082e-05, "loss": 0.5525, "step": 47639 }, { "epoch": 0.05, "learning_rate": 4.9721639639418425e-05, "loss": 0.8228, "step": 47640 }, { "epoch": 0.05, "learning_rate": 4.972162795050199e-05, "loss": 1.0019, "step": 47641 }, { "epoch": 0.05, "learning_rate": 4.972161626134151e-05, "loss": 1.8521, "step": 47642 }, { "epoch": 0.05, "learning_rate": 4.972160457193699e-05, "loss": 1.1853, "step": 47643 }, { "epoch": 0.05, "learning_rate": 4.9721592882288424e-05, "loss": 0.8639, "step": 47644 }, { "epoch": 0.05, "learning_rate": 4.972158119239582e-05, "loss": 0.5994, "step": 47645 }, { "epoch": 0.05, "learning_rate": 4.9721569502259174e-05, "loss": 0.637, "step": 47646 }, { "epoch": 0.05, "learning_rate": 4.972155781187849e-05, "loss": 0.7375, "step": 47647 }, { "epoch": 0.05, "learning_rate": 4.9721546121253756e-05, "loss": 0.6945, "step": 47648 }, { "epoch": 0.05, "learning_rate": 4.9721534430385e-05, "loss": 1.0021, "step": 47649 }, { "epoch": 0.05, "learning_rate": 4.9721522739272184e-05, "loss": 0.9229, "step": 47650 }, { "epoch": 0.05, "learning_rate": 4.972151104791534e-05, "loss": 0.8649, "step": 47651 }, { "epoch": 0.05, "learning_rate": 4.972149935631445e-05, "loss": 1.1628, "step": 47652 }, { "epoch": 0.05, "learning_rate": 4.972148766446952e-05, "loss": 0.9557, "step": 47653 }, { "epoch": 0.05, "learning_rate": 4.972147597238055e-05, "loss": 0.5337, "step": 47654 }, { "epoch": 0.05, "learning_rate": 4.972146428004754e-05, "loss": 0.4337, "step": 47655 }, { "epoch": 0.05, "learning_rate": 4.97214525874705e-05, "loss": 0.3762, "step": 47656 }, { "epoch": 0.05, "learning_rate": 4.972144089464941e-05, "loss": 0.3152, "step": 47657 }, { "epoch": 0.05, "learning_rate": 4.9721429201584274e-05, "loss": 0.3382, "step": 47658 }, { "epoch": 0.05, "learning_rate": 4.972141750827511e-05, "loss": 0.3332, "step": 47659 }, { "epoch": 0.05, "learning_rate": 4.97214058147219e-05, "loss": 0.3576, "step": 47660 }, { "epoch": 0.05, "learning_rate": 4.972139412092465e-05, "loss": 0.3137, "step": 47661 }, { "epoch": 0.05, "learning_rate": 4.972138242688337e-05, "loss": 0.3941, "step": 47662 }, { "epoch": 0.05, "learning_rate": 4.972137073259804e-05, "loss": 0.2715, "step": 47663 }, { "epoch": 0.05, "learning_rate": 4.9721359038068685e-05, "loss": 0.2055, "step": 47664 }, { "epoch": 0.05, "learning_rate": 4.9721347343295274e-05, "loss": 0.3552, "step": 47665 }, { "epoch": 0.05, "learning_rate": 4.9721335648277835e-05, "loss": 0.4101, "step": 47666 }, { "epoch": 0.05, "learning_rate": 4.972132395301635e-05, "loss": 0.2725, "step": 47667 }, { "epoch": 0.05, "learning_rate": 4.972131225751083e-05, "loss": 0.5107, "step": 47668 }, { "epoch": 0.05, "learning_rate": 4.972130056176128e-05, "loss": 1.1275, "step": 47669 }, { "epoch": 0.05, "learning_rate": 4.9721288865767685e-05, "loss": 1.6218, "step": 47670 }, { "epoch": 0.05, "learning_rate": 4.972127716953005e-05, "loss": 1.1528, "step": 47671 }, { "epoch": 0.05, "learning_rate": 4.972126547304838e-05, "loss": 1.289, "step": 47672 }, { "epoch": 0.05, "learning_rate": 4.972125377632268e-05, "loss": 1.0604, "step": 47673 }, { "epoch": 0.05, "learning_rate": 4.972124207935292e-05, "loss": 0.9352, "step": 47674 }, { "epoch": 0.05, "learning_rate": 4.972123038213914e-05, "loss": 0.9284, "step": 47675 }, { "epoch": 0.05, "learning_rate": 4.972121868468131e-05, "loss": 0.7878, "step": 47676 }, { "epoch": 0.05, "learning_rate": 4.972120698697946e-05, "loss": 0.7248, "step": 47677 }, { "epoch": 0.05, "learning_rate": 4.972119528903356e-05, "loss": 1.2297, "step": 47678 }, { "epoch": 0.05, "learning_rate": 4.9721183590843625e-05, "loss": 0.6367, "step": 47679 }, { "epoch": 0.05, "learning_rate": 4.9721171892409656e-05, "loss": 0.9748, "step": 47680 }, { "epoch": 0.05, "learning_rate": 4.972116019373164e-05, "loss": 1.1297, "step": 47681 }, { "epoch": 0.05, "learning_rate": 4.97211484948096e-05, "loss": 0.7429, "step": 47682 }, { "epoch": 0.05, "learning_rate": 4.972113679564352e-05, "loss": 1.0496, "step": 47683 }, { "epoch": 0.05, "learning_rate": 4.97211250962334e-05, "loss": 0.771, "step": 47684 }, { "epoch": 0.05, "learning_rate": 4.972111339657924e-05, "loss": 0.842, "step": 47685 }, { "epoch": 0.05, "learning_rate": 4.972110169668105e-05, "loss": 1.147, "step": 47686 }, { "epoch": 0.05, "learning_rate": 4.972108999653883e-05, "loss": 1.0923, "step": 47687 }, { "epoch": 0.05, "learning_rate": 4.9721078296152555e-05, "loss": 0.9326, "step": 47688 }, { "epoch": 0.05, "learning_rate": 4.9721066595522256e-05, "loss": 1.0511, "step": 47689 }, { "epoch": 0.05, "learning_rate": 4.972105489464792e-05, "loss": 1.0039, "step": 47690 }, { "epoch": 0.05, "learning_rate": 4.972104319352955e-05, "loss": 0.808, "step": 47691 }, { "epoch": 0.05, "learning_rate": 4.9721031492167144e-05, "loss": 0.6086, "step": 47692 }, { "epoch": 0.05, "learning_rate": 4.9721019790560704e-05, "loss": 0.9635, "step": 47693 }, { "epoch": 0.05, "learning_rate": 4.972100808871022e-05, "loss": 1.1222, "step": 47694 }, { "epoch": 0.05, "learning_rate": 4.972099638661571e-05, "loss": 0.8763, "step": 47695 }, { "epoch": 0.05, "learning_rate": 4.972098468427716e-05, "loss": 1.0849, "step": 47696 }, { "epoch": 0.05, "learning_rate": 4.9720972981694574e-05, "loss": 1.1144, "step": 47697 }, { "epoch": 0.05, "learning_rate": 4.972096127886795e-05, "loss": 0.9442, "step": 47698 }, { "epoch": 0.05, "learning_rate": 4.97209495757973e-05, "loss": 0.9034, "step": 47699 }, { "epoch": 0.05, "learning_rate": 4.972093787248261e-05, "loss": 0.9529, "step": 47700 }, { "epoch": 0.05, "learning_rate": 4.972092616892389e-05, "loss": 0.9721, "step": 47701 }, { "epoch": 0.05, "learning_rate": 4.9720914465121125e-05, "loss": 0.9717, "step": 47702 }, { "epoch": 0.05, "learning_rate": 4.9720902761074335e-05, "loss": 0.9885, "step": 47703 }, { "epoch": 0.05, "learning_rate": 4.9720891056783505e-05, "loss": 1.0476, "step": 47704 }, { "epoch": 0.05, "learning_rate": 4.972087935224864e-05, "loss": 1.018, "step": 47705 }, { "epoch": 0.05, "learning_rate": 4.972086764746975e-05, "loss": 1.2491, "step": 47706 }, { "epoch": 0.05, "learning_rate": 4.972085594244682e-05, "loss": 0.9278, "step": 47707 }, { "epoch": 0.05, "learning_rate": 4.972084423717985e-05, "loss": 0.857, "step": 47708 }, { "epoch": 0.05, "learning_rate": 4.972083253166886e-05, "loss": 0.7622, "step": 47709 }, { "epoch": 0.05, "learning_rate": 4.972082082591383e-05, "loss": 0.8215, "step": 47710 }, { "epoch": 0.05, "learning_rate": 4.972080911991477e-05, "loss": 0.9671, "step": 47711 }, { "epoch": 0.05, "learning_rate": 4.9720797413671666e-05, "loss": 1.0666, "step": 47712 }, { "epoch": 0.05, "learning_rate": 4.9720785707184534e-05, "loss": 1.1219, "step": 47713 }, { "epoch": 0.05, "learning_rate": 4.9720774000453375e-05, "loss": 0.7542, "step": 47714 }, { "epoch": 0.05, "learning_rate": 4.9720762293478176e-05, "loss": 1.2995, "step": 47715 }, { "epoch": 0.05, "learning_rate": 4.972075058625894e-05, "loss": 0.979, "step": 47716 }, { "epoch": 0.05, "learning_rate": 4.9720738878795684e-05, "loss": 1.0348, "step": 47717 }, { "epoch": 0.05, "learning_rate": 4.972072717108839e-05, "loss": 1.0104, "step": 47718 }, { "epoch": 0.05, "learning_rate": 4.972071546313706e-05, "loss": 1.1856, "step": 47719 }, { "epoch": 0.05, "learning_rate": 4.97207037549417e-05, "loss": 1.0401, "step": 47720 }, { "epoch": 0.05, "learning_rate": 4.97206920465023e-05, "loss": 1.149, "step": 47721 }, { "epoch": 0.05, "learning_rate": 4.9720680337818876e-05, "loss": 0.9164, "step": 47722 }, { "epoch": 0.05, "learning_rate": 4.972066862889142e-05, "loss": 0.8521, "step": 47723 }, { "epoch": 0.05, "learning_rate": 4.972065691971993e-05, "loss": 1.1875, "step": 47724 }, { "epoch": 0.05, "learning_rate": 4.9720645210304406e-05, "loss": 1.0924, "step": 47725 }, { "epoch": 0.05, "learning_rate": 4.972063350064485e-05, "loss": 1.0371, "step": 47726 }, { "epoch": 0.05, "learning_rate": 4.972062179074127e-05, "loss": 1.2434, "step": 47727 }, { "epoch": 0.05, "learning_rate": 4.9720610080593654e-05, "loss": 1.1229, "step": 47728 }, { "epoch": 0.05, "learning_rate": 4.9720598370202005e-05, "loss": 0.9933, "step": 47729 }, { "epoch": 0.05, "learning_rate": 4.972058665956633e-05, "loss": 0.7903, "step": 47730 }, { "epoch": 0.05, "learning_rate": 4.972057494868661e-05, "loss": 1.3459, "step": 47731 }, { "epoch": 0.05, "learning_rate": 4.972056323756287e-05, "loss": 1.2519, "step": 47732 }, { "epoch": 0.05, "learning_rate": 4.9720551526195094e-05, "loss": 1.0229, "step": 47733 }, { "epoch": 0.05, "learning_rate": 4.97205398145833e-05, "loss": 1.195, "step": 47734 }, { "epoch": 0.05, "learning_rate": 4.972052810272746e-05, "loss": 1.1327, "step": 47735 }, { "epoch": 0.05, "learning_rate": 4.972051639062759e-05, "loss": 0.6835, "step": 47736 }, { "epoch": 0.05, "learning_rate": 4.97205046782837e-05, "loss": 1.1935, "step": 47737 }, { "epoch": 0.05, "learning_rate": 4.9720492965695776e-05, "loss": 1.0257, "step": 47738 }, { "epoch": 0.05, "learning_rate": 4.972048125286382e-05, "loss": 1.0257, "step": 47739 }, { "epoch": 0.05, "learning_rate": 4.972046953978783e-05, "loss": 0.7577, "step": 47740 }, { "epoch": 0.05, "learning_rate": 4.9720457826467816e-05, "loss": 1.0573, "step": 47741 }, { "epoch": 0.05, "learning_rate": 4.972044611290377e-05, "loss": 0.8577, "step": 47742 }, { "epoch": 0.05, "learning_rate": 4.972043439909569e-05, "loss": 1.0311, "step": 47743 }, { "epoch": 0.05, "learning_rate": 4.972042268504359e-05, "loss": 0.965, "step": 47744 }, { "epoch": 0.05, "learning_rate": 4.972041097074745e-05, "loss": 0.9833, "step": 47745 }, { "epoch": 0.05, "learning_rate": 4.972039925620728e-05, "loss": 0.9759, "step": 47746 }, { "epoch": 0.05, "learning_rate": 4.972038754142309e-05, "loss": 0.9423, "step": 47747 }, { "epoch": 0.05, "learning_rate": 4.9720375826394865e-05, "loss": 1.3926, "step": 47748 }, { "epoch": 0.05, "learning_rate": 4.972036411112261e-05, "loss": 1.0237, "step": 47749 }, { "epoch": 0.05, "learning_rate": 4.9720352395606326e-05, "loss": 0.8381, "step": 47750 }, { "epoch": 0.05, "learning_rate": 4.972034067984602e-05, "loss": 1.2476, "step": 47751 }, { "epoch": 0.05, "learning_rate": 4.972032896384168e-05, "loss": 0.8204, "step": 47752 }, { "epoch": 0.05, "learning_rate": 4.972031724759332e-05, "loss": 0.7039, "step": 47753 }, { "epoch": 0.05, "learning_rate": 4.972030553110092e-05, "loss": 0.9028, "step": 47754 }, { "epoch": 0.05, "learning_rate": 4.972029381436449e-05, "loss": 0.9067, "step": 47755 }, { "epoch": 0.05, "learning_rate": 4.972028209738404e-05, "loss": 0.7831, "step": 47756 }, { "epoch": 0.05, "learning_rate": 4.972027038015956e-05, "loss": 0.633, "step": 47757 }, { "epoch": 0.05, "learning_rate": 4.972025866269104e-05, "loss": 0.8689, "step": 47758 }, { "epoch": 0.05, "learning_rate": 4.972024694497851e-05, "loss": 0.7366, "step": 47759 }, { "epoch": 0.05, "learning_rate": 4.972023522702194e-05, "loss": 0.9524, "step": 47760 }, { "epoch": 0.05, "learning_rate": 4.972022350882135e-05, "loss": 0.6738, "step": 47761 }, { "epoch": 0.05, "learning_rate": 4.9720211790376725e-05, "loss": 1.2146, "step": 47762 }, { "epoch": 0.05, "learning_rate": 4.972020007168808e-05, "loss": 1.1921, "step": 47763 }, { "epoch": 0.05, "learning_rate": 4.97201883527554e-05, "loss": 0.9299, "step": 47764 }, { "epoch": 0.05, "learning_rate": 4.97201766335787e-05, "loss": 1.1184, "step": 47765 }, { "epoch": 0.05, "learning_rate": 4.972016491415797e-05, "loss": 0.5616, "step": 47766 }, { "epoch": 0.05, "learning_rate": 4.9720153194493215e-05, "loss": 1.0847, "step": 47767 }, { "epoch": 0.05, "learning_rate": 4.9720141474584426e-05, "loss": 1.0944, "step": 47768 }, { "epoch": 0.05, "learning_rate": 4.9720129754431624e-05, "loss": 1.166, "step": 47769 }, { "epoch": 0.05, "learning_rate": 4.972011803403478e-05, "loss": 1.209, "step": 47770 }, { "epoch": 0.05, "learning_rate": 4.972010631339392e-05, "loss": 1.4612, "step": 47771 }, { "epoch": 0.05, "learning_rate": 4.972009459250903e-05, "loss": 0.4367, "step": 47772 }, { "epoch": 0.05, "learning_rate": 4.9720082871380106e-05, "loss": 0.3597, "step": 47773 }, { "epoch": 0.05, "learning_rate": 4.972007115000716e-05, "loss": 0.7923, "step": 47774 }, { "epoch": 0.05, "learning_rate": 4.97200594283902e-05, "loss": 1.091, "step": 47775 }, { "epoch": 0.05, "learning_rate": 4.97200477065292e-05, "loss": 0.8889, "step": 47776 }, { "epoch": 0.05, "learning_rate": 4.972003598442417e-05, "loss": 1.0389, "step": 47777 }, { "epoch": 0.05, "learning_rate": 4.972002426207513e-05, "loss": 1.0606, "step": 47778 }, { "epoch": 0.05, "learning_rate": 4.972001253948205e-05, "loss": 1.1538, "step": 47779 }, { "epoch": 0.05, "learning_rate": 4.9720000816644954e-05, "loss": 0.9408, "step": 47780 }, { "epoch": 0.05, "learning_rate": 4.971998909356384e-05, "loss": 1.2453, "step": 47781 }, { "epoch": 0.05, "learning_rate": 4.9719977370238686e-05, "loss": 1.2547, "step": 47782 }, { "epoch": 0.05, "learning_rate": 4.971996564666951e-05, "loss": 0.7524, "step": 47783 }, { "epoch": 0.05, "learning_rate": 4.971995392285631e-05, "loss": 0.9262, "step": 47784 }, { "epoch": 0.05, "learning_rate": 4.9719942198799086e-05, "loss": 0.97, "step": 47785 }, { "epoch": 0.05, "learning_rate": 4.9719930474497835e-05, "loss": 1.0714, "step": 47786 }, { "epoch": 0.05, "learning_rate": 4.9719918749952564e-05, "loss": 0.7358, "step": 47787 }, { "epoch": 0.05, "learning_rate": 4.971990702516326e-05, "loss": 1.0663, "step": 47788 }, { "epoch": 0.05, "learning_rate": 4.9719895300129934e-05, "loss": 0.9223, "step": 47789 }, { "epoch": 0.05, "learning_rate": 4.971988357485259e-05, "loss": 0.9007, "step": 47790 }, { "epoch": 0.05, "learning_rate": 4.9719871849331224e-05, "loss": 0.4539, "step": 47791 }, { "epoch": 0.05, "learning_rate": 4.971986012356582e-05, "loss": 1.0839, "step": 47792 }, { "epoch": 0.05, "learning_rate": 4.97198483975564e-05, "loss": 1.1637, "step": 47793 }, { "epoch": 0.05, "learning_rate": 4.971983667130296e-05, "loss": 0.7394, "step": 47794 }, { "epoch": 0.05, "learning_rate": 4.9719824944805496e-05, "loss": 1.0142, "step": 47795 }, { "epoch": 0.05, "learning_rate": 4.9719813218064004e-05, "loss": 0.9299, "step": 47796 }, { "epoch": 0.05, "learning_rate": 4.9719801491078485e-05, "loss": 0.5248, "step": 47797 }, { "epoch": 0.05, "learning_rate": 4.9719789763848946e-05, "loss": 0.8851, "step": 47798 }, { "epoch": 0.05, "learning_rate": 4.971977803637539e-05, "loss": 1.078, "step": 47799 }, { "epoch": 0.05, "learning_rate": 4.97197663086578e-05, "loss": 1.0448, "step": 47800 }, { "epoch": 0.05, "learning_rate": 4.971975458069619e-05, "loss": 1.0146, "step": 47801 }, { "epoch": 0.05, "learning_rate": 4.9719742852490556e-05, "loss": 0.9754, "step": 47802 }, { "epoch": 0.05, "learning_rate": 4.97197311240409e-05, "loss": 0.9723, "step": 47803 }, { "epoch": 0.05, "learning_rate": 4.971971939534723e-05, "loss": 1.1399, "step": 47804 }, { "epoch": 0.05, "learning_rate": 4.971970766640953e-05, "loss": 1.2429, "step": 47805 }, { "epoch": 0.05, "learning_rate": 4.9719695937227805e-05, "loss": 1.0528, "step": 47806 }, { "epoch": 0.05, "learning_rate": 4.9719684207802065e-05, "loss": 1.1351, "step": 47807 }, { "epoch": 0.05, "learning_rate": 4.97196724781323e-05, "loss": 1.0351, "step": 47808 }, { "epoch": 0.05, "learning_rate": 4.9719660748218505e-05, "loss": 0.9655, "step": 47809 }, { "epoch": 0.05, "learning_rate": 4.97196490180607e-05, "loss": 1.1886, "step": 47810 }, { "epoch": 0.05, "learning_rate": 4.971963728765886e-05, "loss": 1.2295, "step": 47811 }, { "epoch": 0.05, "learning_rate": 4.9719625557013006e-05, "loss": 1.2562, "step": 47812 }, { "epoch": 0.05, "learning_rate": 4.971961382612313e-05, "loss": 1.2019, "step": 47813 }, { "epoch": 0.05, "learning_rate": 4.971960209498923e-05, "loss": 0.8841, "step": 47814 }, { "epoch": 0.05, "learning_rate": 4.971959036361131e-05, "loss": 0.902, "step": 47815 }, { "epoch": 0.05, "learning_rate": 4.971957863198937e-05, "loss": 0.9207, "step": 47816 }, { "epoch": 0.05, "learning_rate": 4.9719566900123405e-05, "loss": 0.9506, "step": 47817 }, { "epoch": 0.05, "learning_rate": 4.971955516801342e-05, "loss": 1.0682, "step": 47818 }, { "epoch": 0.05, "learning_rate": 4.9719543435659424e-05, "loss": 0.8568, "step": 47819 }, { "epoch": 0.05, "learning_rate": 4.97195317030614e-05, "loss": 1.2246, "step": 47820 }, { "epoch": 0.05, "learning_rate": 4.971951997021935e-05, "loss": 1.0816, "step": 47821 }, { "epoch": 0.05, "learning_rate": 4.971950823713328e-05, "loss": 1.0639, "step": 47822 }, { "epoch": 0.05, "learning_rate": 4.971949650380319e-05, "loss": 1.0204, "step": 47823 }, { "epoch": 0.05, "learning_rate": 4.971948477022909e-05, "loss": 0.8279, "step": 47824 }, { "epoch": 0.05, "learning_rate": 4.9719473036410955e-05, "loss": 1.1145, "step": 47825 }, { "epoch": 0.05, "learning_rate": 4.971946130234881e-05, "loss": 1.1822, "step": 47826 }, { "epoch": 0.05, "learning_rate": 4.971944956804264e-05, "loss": 1.0658, "step": 47827 }, { "epoch": 0.05, "learning_rate": 4.971943783349245e-05, "loss": 0.9432, "step": 47828 }, { "epoch": 0.05, "learning_rate": 4.971942609869824e-05, "loss": 0.735, "step": 47829 }, { "epoch": 0.05, "learning_rate": 4.971941436366001e-05, "loss": 0.8667, "step": 47830 }, { "epoch": 0.05, "learning_rate": 4.9719402628377764e-05, "loss": 0.7171, "step": 47831 }, { "epoch": 0.05, "learning_rate": 4.9719390892851504e-05, "loss": 0.8647, "step": 47832 }, { "epoch": 0.05, "learning_rate": 4.9719379157081216e-05, "loss": 0.5938, "step": 47833 }, { "epoch": 0.05, "learning_rate": 4.971936742106691e-05, "loss": 0.8453, "step": 47834 }, { "epoch": 0.05, "learning_rate": 4.9719355684808574e-05, "loss": 1.0334, "step": 47835 }, { "epoch": 0.05, "learning_rate": 4.9719343948306234e-05, "loss": 1.0637, "step": 47836 }, { "epoch": 0.05, "learning_rate": 4.971933221155987e-05, "loss": 1.0455, "step": 47837 }, { "epoch": 0.05, "learning_rate": 4.9719320474569485e-05, "loss": 1.012, "step": 47838 }, { "epoch": 0.05, "learning_rate": 4.971930873733509e-05, "loss": 0.7937, "step": 47839 }, { "epoch": 0.05, "learning_rate": 4.971929699985667e-05, "loss": 0.9388, "step": 47840 }, { "epoch": 0.05, "learning_rate": 4.971928526213423e-05, "loss": 1.3618, "step": 47841 }, { "epoch": 0.05, "learning_rate": 4.9719273524167775e-05, "loss": 1.028, "step": 47842 }, { "epoch": 0.05, "learning_rate": 4.97192617859573e-05, "loss": 0.986, "step": 47843 }, { "epoch": 0.05, "learning_rate": 4.971925004750281e-05, "loss": 1.1092, "step": 47844 }, { "epoch": 0.05, "learning_rate": 4.971923830880429e-05, "loss": 1.0639, "step": 47845 }, { "epoch": 0.05, "learning_rate": 4.9719226569861765e-05, "loss": 1.5566, "step": 47846 }, { "epoch": 0.05, "learning_rate": 4.971921483067522e-05, "loss": 1.5366, "step": 47847 }, { "epoch": 0.05, "learning_rate": 4.971920309124466e-05, "loss": 0.9483, "step": 47848 }, { "epoch": 0.05, "learning_rate": 4.9719191351570076e-05, "loss": 1.1997, "step": 47849 }, { "epoch": 0.05, "learning_rate": 4.9719179611651476e-05, "loss": 1.0101, "step": 47850 }, { "epoch": 0.05, "learning_rate": 4.971916787148886e-05, "loss": 2.1167, "step": 47851 }, { "epoch": 0.05, "learning_rate": 4.971915613108222e-05, "loss": 1.3043, "step": 47852 }, { "epoch": 0.05, "learning_rate": 4.9719144390431573e-05, "loss": 0.8568, "step": 47853 }, { "epoch": 0.05, "learning_rate": 4.9719132649536906e-05, "loss": 0.8326, "step": 47854 }, { "epoch": 0.05, "learning_rate": 4.971912090839822e-05, "loss": 1.1538, "step": 47855 }, { "epoch": 0.05, "learning_rate": 4.971910916701552e-05, "loss": 0.4732, "step": 47856 }, { "epoch": 0.05, "learning_rate": 4.9719097425388804e-05, "loss": 1.0621, "step": 47857 }, { "epoch": 0.05, "learning_rate": 4.971908568351807e-05, "loss": 0.8774, "step": 47858 }, { "epoch": 0.05, "learning_rate": 4.9719073941403316e-05, "loss": 0.7676, "step": 47859 }, { "epoch": 0.05, "learning_rate": 4.9719062199044556e-05, "loss": 0.6178, "step": 47860 }, { "epoch": 0.05, "learning_rate": 4.971905045644177e-05, "loss": 0.9491, "step": 47861 }, { "epoch": 0.05, "learning_rate": 4.971903871359497e-05, "loss": 0.4884, "step": 47862 }, { "epoch": 0.05, "learning_rate": 4.9719026970504155e-05, "loss": 0.4233, "step": 47863 }, { "epoch": 0.05, "learning_rate": 4.971901522716932e-05, "loss": 0.4039, "step": 47864 }, { "epoch": 0.05, "learning_rate": 4.9719003483590474e-05, "loss": 0.3183, "step": 47865 }, { "epoch": 0.05, "learning_rate": 4.9718991739767614e-05, "loss": 0.3281, "step": 47866 }, { "epoch": 0.05, "learning_rate": 4.971897999570073e-05, "loss": 0.3538, "step": 47867 }, { "epoch": 0.05, "learning_rate": 4.971896825138984e-05, "loss": 0.2481, "step": 47868 }, { "epoch": 0.05, "learning_rate": 4.971895650683493e-05, "loss": 0.9612, "step": 47869 }, { "epoch": 0.05, "learning_rate": 4.971894476203601e-05, "loss": 0.9478, "step": 47870 }, { "epoch": 0.05, "learning_rate": 4.971893301699307e-05, "loss": 0.6166, "step": 47871 }, { "epoch": 0.05, "learning_rate": 4.971892127170612e-05, "loss": 1.0773, "step": 47872 }, { "epoch": 0.05, "learning_rate": 4.971890952617515e-05, "loss": 1.042, "step": 47873 }, { "epoch": 0.05, "learning_rate": 4.9718897780400166e-05, "loss": 0.968, "step": 47874 }, { "epoch": 0.05, "learning_rate": 4.9718886034381167e-05, "loss": 0.9952, "step": 47875 }, { "epoch": 0.05, "learning_rate": 4.9718874288118154e-05, "loss": 0.9939, "step": 47876 }, { "epoch": 0.05, "learning_rate": 4.9718862541611134e-05, "loss": 0.963, "step": 47877 }, { "epoch": 0.05, "learning_rate": 4.971885079486009e-05, "loss": 0.9601, "step": 47878 }, { "epoch": 0.05, "learning_rate": 4.9718839047865035e-05, "loss": 0.953, "step": 47879 }, { "epoch": 0.05, "learning_rate": 4.971882730062597e-05, "loss": 0.9169, "step": 47880 }, { "epoch": 0.05, "learning_rate": 4.971881555314288e-05, "loss": 0.9104, "step": 47881 }, { "epoch": 0.05, "learning_rate": 4.971880380541579e-05, "loss": 0.8942, "step": 47882 }, { "epoch": 0.05, "learning_rate": 4.971879205744468e-05, "loss": 0.9441, "step": 47883 }, { "epoch": 0.05, "learning_rate": 4.971878030922955e-05, "loss": 0.802, "step": 47884 }, { "epoch": 0.05, "learning_rate": 4.971876856077042e-05, "loss": 1.1263, "step": 47885 }, { "epoch": 0.05, "learning_rate": 4.971875681206727e-05, "loss": 1.0566, "step": 47886 }, { "epoch": 0.05, "learning_rate": 4.971874506312011e-05, "loss": 1.2937, "step": 47887 }, { "epoch": 0.05, "learning_rate": 4.9718733313928936e-05, "loss": 1.05, "step": 47888 }, { "epoch": 0.05, "learning_rate": 4.9718721564493744e-05, "loss": 0.4595, "step": 47889 }, { "epoch": 0.05, "learning_rate": 4.971870981481454e-05, "loss": 1.145, "step": 47890 }, { "epoch": 0.05, "learning_rate": 4.971869806489132e-05, "loss": 0.8362, "step": 47891 }, { "epoch": 0.05, "learning_rate": 4.9718686314724094e-05, "loss": 0.8462, "step": 47892 }, { "epoch": 0.05, "learning_rate": 4.971867456431286e-05, "loss": 0.9743, "step": 47893 }, { "epoch": 0.05, "learning_rate": 4.9718662813657604e-05, "loss": 1.3106, "step": 47894 }, { "epoch": 0.05, "learning_rate": 4.971865106275834e-05, "loss": 0.9319, "step": 47895 }, { "epoch": 0.05, "learning_rate": 4.971863931161507e-05, "loss": 1.0123, "step": 47896 }, { "epoch": 0.05, "learning_rate": 4.9718627560227784e-05, "loss": 1.1448, "step": 47897 }, { "epoch": 0.05, "learning_rate": 4.971861580859648e-05, "loss": 1.0335, "step": 47898 }, { "epoch": 0.05, "learning_rate": 4.971860405672116e-05, "loss": 1.2229, "step": 47899 }, { "epoch": 0.05, "learning_rate": 4.9718592304601844e-05, "loss": 1.2572, "step": 47900 }, { "epoch": 0.05, "learning_rate": 4.9718580552238506e-05, "loss": 0.9347, "step": 47901 }, { "epoch": 0.05, "learning_rate": 4.971856879963116e-05, "loss": 1.1134, "step": 47902 }, { "epoch": 0.05, "learning_rate": 4.9718557046779805e-05, "loss": 1.1024, "step": 47903 }, { "epoch": 0.05, "learning_rate": 4.971854529368444e-05, "loss": 1.0624, "step": 47904 }, { "epoch": 0.05, "learning_rate": 4.971853354034506e-05, "loss": 1.1642, "step": 47905 }, { "epoch": 0.05, "learning_rate": 4.971852178676166e-05, "loss": 1.0014, "step": 47906 }, { "epoch": 0.05, "learning_rate": 4.971851003293426e-05, "loss": 1.157, "step": 47907 }, { "epoch": 0.05, "learning_rate": 4.971849827886285e-05, "loss": 0.9972, "step": 47908 }, { "epoch": 0.05, "learning_rate": 4.971848652454742e-05, "loss": 1.0393, "step": 47909 }, { "epoch": 0.05, "learning_rate": 4.971847476998799e-05, "loss": 1.4466, "step": 47910 }, { "epoch": 0.05, "learning_rate": 4.971846301518454e-05, "loss": 1.1639, "step": 47911 }, { "epoch": 0.05, "learning_rate": 4.97184512601371e-05, "loss": 1.3107, "step": 47912 }, { "epoch": 0.05, "learning_rate": 4.971843950484563e-05, "loss": 1.2166, "step": 47913 }, { "epoch": 0.05, "learning_rate": 4.971842774931016e-05, "loss": 0.9024, "step": 47914 }, { "epoch": 0.05, "learning_rate": 4.971841599353067e-05, "loss": 0.3774, "step": 47915 }, { "epoch": 0.05, "learning_rate": 4.971840423750718e-05, "loss": 1.0197, "step": 47916 }, { "epoch": 0.05, "learning_rate": 4.9718392481239675e-05, "loss": 0.9183, "step": 47917 }, { "epoch": 0.05, "learning_rate": 4.9718380724728166e-05, "loss": 0.5847, "step": 47918 }, { "epoch": 0.05, "learning_rate": 4.971836896797264e-05, "loss": 1.2693, "step": 47919 }, { "epoch": 0.05, "learning_rate": 4.971835721097311e-05, "loss": 1.1582, "step": 47920 }, { "epoch": 0.05, "learning_rate": 4.971834545372957e-05, "loss": 0.6291, "step": 47921 }, { "epoch": 0.05, "learning_rate": 4.971833369624201e-05, "loss": 1.1275, "step": 47922 }, { "epoch": 0.05, "learning_rate": 4.971832193851046e-05, "loss": 0.6304, "step": 47923 }, { "epoch": 0.05, "learning_rate": 4.971831018053489e-05, "loss": 0.651, "step": 47924 }, { "epoch": 0.05, "learning_rate": 4.9718298422315316e-05, "loss": 0.9428, "step": 47925 }, { "epoch": 0.05, "learning_rate": 4.9718286663851735e-05, "loss": 0.791, "step": 47926 }, { "epoch": 0.05, "learning_rate": 4.9718274905144134e-05, "loss": 0.5232, "step": 47927 }, { "epoch": 0.05, "learning_rate": 4.9718263146192534e-05, "loss": 0.8948, "step": 47928 }, { "epoch": 0.05, "learning_rate": 4.971825138699692e-05, "loss": 0.8136, "step": 47929 }, { "epoch": 0.05, "learning_rate": 4.9718239627557306e-05, "loss": 0.7271, "step": 47930 }, { "epoch": 0.05, "learning_rate": 4.971822786787368e-05, "loss": 0.6753, "step": 47931 }, { "epoch": 0.05, "learning_rate": 4.971821610794604e-05, "loss": 0.4135, "step": 47932 }, { "epoch": 0.05, "learning_rate": 4.97182043477744e-05, "loss": 0.74, "step": 47933 }, { "epoch": 0.05, "learning_rate": 4.971819258735875e-05, "loss": 1.5391, "step": 47934 }, { "epoch": 0.05, "learning_rate": 4.971818082669909e-05, "loss": 0.7827, "step": 47935 }, { "epoch": 0.05, "learning_rate": 4.971816906579543e-05, "loss": 0.9459, "step": 47936 }, { "epoch": 0.05, "learning_rate": 4.971815730464775e-05, "loss": 0.9521, "step": 47937 }, { "epoch": 0.05, "learning_rate": 4.971814554325607e-05, "loss": 1.2932, "step": 47938 }, { "epoch": 0.05, "learning_rate": 4.9718133781620385e-05, "loss": 1.9232, "step": 47939 }, { "epoch": 0.05, "learning_rate": 4.9718122019740693e-05, "loss": 1.2605, "step": 47940 }, { "epoch": 0.05, "learning_rate": 4.971811025761699e-05, "loss": 0.8334, "step": 47941 }, { "epoch": 0.05, "learning_rate": 4.9718098495249284e-05, "loss": 0.9239, "step": 47942 }, { "epoch": 0.05, "learning_rate": 4.971808673263757e-05, "loss": 1.2861, "step": 47943 }, { "epoch": 0.05, "learning_rate": 4.971807496978185e-05, "loss": 1.1474, "step": 47944 }, { "epoch": 0.05, "learning_rate": 4.971806320668212e-05, "loss": 0.9995, "step": 47945 }, { "epoch": 0.05, "learning_rate": 4.9718051443338386e-05, "loss": 0.9859, "step": 47946 }, { "epoch": 0.05, "learning_rate": 4.971803967975065e-05, "loss": 1.1634, "step": 47947 }, { "epoch": 0.05, "learning_rate": 4.97180279159189e-05, "loss": 0.9621, "step": 47948 }, { "epoch": 0.05, "learning_rate": 4.971801615184315e-05, "loss": 0.8813, "step": 47949 }, { "epoch": 0.05, "learning_rate": 4.971800438752339e-05, "loss": 0.9944, "step": 47950 }, { "epoch": 0.05, "learning_rate": 4.971799262295963e-05, "loss": 1.1877, "step": 47951 }, { "epoch": 0.05, "learning_rate": 4.971798085815186e-05, "loss": 1.4145, "step": 47952 }, { "epoch": 0.05, "learning_rate": 4.9717969093100084e-05, "loss": 1.2354, "step": 47953 }, { "epoch": 0.05, "learning_rate": 4.971795732780431e-05, "loss": 1.1528, "step": 47954 }, { "epoch": 0.05, "learning_rate": 4.9717945562264526e-05, "loss": 0.85, "step": 47955 }, { "epoch": 0.05, "learning_rate": 4.971793379648073e-05, "loss": 1.0967, "step": 47956 }, { "epoch": 0.05, "learning_rate": 4.9717922030452936e-05, "loss": 1.3113, "step": 47957 }, { "epoch": 0.05, "learning_rate": 4.9717910264181134e-05, "loss": 1.0339, "step": 47958 }, { "epoch": 0.05, "learning_rate": 4.9717898497665326e-05, "loss": 1.3509, "step": 47959 }, { "epoch": 0.05, "learning_rate": 4.9717886730905525e-05, "loss": 1.2725, "step": 47960 }, { "epoch": 0.05, "learning_rate": 4.9717874963901704e-05, "loss": 1.3047, "step": 47961 }, { "epoch": 0.05, "learning_rate": 4.971786319665389e-05, "loss": 1.0735, "step": 47962 }, { "epoch": 0.05, "learning_rate": 4.971785142916206e-05, "loss": 0.8971, "step": 47963 }, { "epoch": 0.05, "learning_rate": 4.9717839661426236e-05, "loss": 0.7113, "step": 47964 }, { "epoch": 0.05, "learning_rate": 4.97178278934464e-05, "loss": 1.2934, "step": 47965 }, { "epoch": 0.05, "learning_rate": 4.971781612522257e-05, "loss": 1.3044, "step": 47966 }, { "epoch": 0.05, "learning_rate": 4.971780435675473e-05, "loss": 1.174, "step": 47967 }, { "epoch": 0.05, "learning_rate": 4.971779258804289e-05, "loss": 1.518, "step": 47968 }, { "epoch": 0.05, "learning_rate": 4.9717780819087046e-05, "loss": 1.202, "step": 47969 }, { "epoch": 0.05, "learning_rate": 4.9717769049887193e-05, "loss": 0.7934, "step": 47970 }, { "epoch": 0.05, "learning_rate": 4.971775728044334e-05, "loss": 1.1278, "step": 47971 }, { "epoch": 0.05, "learning_rate": 4.971774551075549e-05, "loss": 0.7482, "step": 47972 }, { "epoch": 0.05, "learning_rate": 4.9717733740823625e-05, "loss": 1.0826, "step": 47973 }, { "epoch": 0.05, "learning_rate": 4.9717721970647754e-05, "loss": 1.2243, "step": 47974 }, { "epoch": 0.05, "learning_rate": 4.9717710200227897e-05, "loss": 1.4319, "step": 47975 }, { "epoch": 0.05, "learning_rate": 4.9717698429564026e-05, "loss": 0.9705, "step": 47976 }, { "epoch": 0.05, "learning_rate": 4.9717686658656155e-05, "loss": 1.3915, "step": 47977 }, { "epoch": 0.05, "learning_rate": 4.9717674887504285e-05, "loss": 1.2542, "step": 47978 }, { "epoch": 0.05, "learning_rate": 4.97176631161084e-05, "loss": 1.1164, "step": 47979 }, { "epoch": 0.05, "learning_rate": 4.9717651344468526e-05, "loss": 1.4452, "step": 47980 }, { "epoch": 0.05, "learning_rate": 4.971763957258464e-05, "loss": 0.6406, "step": 47981 }, { "epoch": 0.05, "learning_rate": 4.971762780045676e-05, "loss": 0.5105, "step": 47982 }, { "epoch": 0.05, "learning_rate": 4.971761602808488e-05, "loss": 0.5284, "step": 47983 }, { "epoch": 0.05, "learning_rate": 4.971760425546898e-05, "loss": 0.7796, "step": 47984 }, { "epoch": 0.05, "learning_rate": 4.97175924826091e-05, "loss": 0.6912, "step": 47985 }, { "epoch": 0.05, "learning_rate": 4.971758070950521e-05, "loss": 0.7517, "step": 47986 }, { "epoch": 0.05, "learning_rate": 4.971756893615732e-05, "loss": 0.7692, "step": 47987 }, { "epoch": 0.05, "learning_rate": 4.971755716256542e-05, "loss": 0.3094, "step": 47988 }, { "epoch": 0.05, "learning_rate": 4.971754538872953e-05, "loss": 0.975, "step": 47989 }, { "epoch": 0.05, "learning_rate": 4.971753361464964e-05, "loss": 0.8321, "step": 47990 }, { "epoch": 0.05, "learning_rate": 4.971752184032574e-05, "loss": 1.2248, "step": 47991 }, { "epoch": 0.05, "learning_rate": 4.971751006575784e-05, "loss": 0.441, "step": 47992 }, { "epoch": 0.05, "learning_rate": 4.971749829094594e-05, "loss": 0.9109, "step": 47993 }, { "epoch": 0.05, "learning_rate": 4.9717486515890046e-05, "loss": 0.8088, "step": 47994 }, { "epoch": 0.05, "learning_rate": 4.971747474059014e-05, "loss": 1.178, "step": 47995 }, { "epoch": 0.05, "learning_rate": 4.9717462965046235e-05, "loss": 1.2086, "step": 47996 }, { "epoch": 0.05, "learning_rate": 4.971745118925834e-05, "loss": 1.0299, "step": 47997 }, { "epoch": 0.05, "learning_rate": 4.971743941322644e-05, "loss": 0.8505, "step": 47998 }, { "epoch": 0.05, "learning_rate": 4.9717427636950535e-05, "loss": 0.9805, "step": 47999 }, { "epoch": 0.05, "learning_rate": 4.971741586043063e-05, "loss": 0.9851, "step": 48000 }, { "epoch": 0.05, "eval_loss": 1.0888420343399048, "eval_runtime": 128.4222, "eval_samples_per_second": 10.785, "eval_steps_per_second": 5.396, "step": 48000 }, { "epoch": 0.05, "learning_rate": 4.971740408366673e-05, "loss": 0.9193, "step": 48001 }, { "epoch": 0.05, "learning_rate": 4.971739230665883e-05, "loss": 0.6531, "step": 48002 }, { "epoch": 0.05, "learning_rate": 4.9717380529406933e-05, "loss": 0.921, "step": 48003 }, { "epoch": 0.05, "learning_rate": 4.971736875191103e-05, "loss": 0.9173, "step": 48004 }, { "epoch": 0.05, "learning_rate": 4.971735697417113e-05, "loss": 1.0663, "step": 48005 }, { "epoch": 0.05, "learning_rate": 4.971734519618723e-05, "loss": 1.1954, "step": 48006 }, { "epoch": 0.05, "learning_rate": 4.971733341795933e-05, "loss": 0.9367, "step": 48007 }, { "epoch": 0.05, "learning_rate": 4.9717321639487436e-05, "loss": 1.1529, "step": 48008 }, { "epoch": 0.05, "learning_rate": 4.971730986077154e-05, "loss": 1.0752, "step": 48009 }, { "epoch": 0.05, "learning_rate": 4.9717298081811647e-05, "loss": 1.0588, "step": 48010 }, { "epoch": 0.05, "learning_rate": 4.9717286302607746e-05, "loss": 0.9496, "step": 48011 }, { "epoch": 0.05, "learning_rate": 4.971727452315986e-05, "loss": 1.1228, "step": 48012 }, { "epoch": 0.05, "learning_rate": 4.9717262743467965e-05, "loss": 0.7436, "step": 48013 }, { "epoch": 0.05, "learning_rate": 4.971725096353207e-05, "loss": 0.8428, "step": 48014 }, { "epoch": 0.05, "learning_rate": 4.9717239183352186e-05, "loss": 1.2723, "step": 48015 }, { "epoch": 0.05, "learning_rate": 4.97172274029283e-05, "loss": 1.2421, "step": 48016 }, { "epoch": 0.05, "learning_rate": 4.9717215622260415e-05, "loss": 1.0926, "step": 48017 }, { "epoch": 0.05, "learning_rate": 4.971720384134854e-05, "loss": 0.848, "step": 48018 }, { "epoch": 0.05, "learning_rate": 4.971719206019265e-05, "loss": 1.0233, "step": 48019 }, { "epoch": 0.05, "learning_rate": 4.9717180278792774e-05, "loss": 1.1251, "step": 48020 }, { "epoch": 0.05, "learning_rate": 4.9717168497148903e-05, "loss": 1.3159, "step": 48021 }, { "epoch": 0.05, "learning_rate": 4.9717156715261026e-05, "loss": 1.1116, "step": 48022 }, { "epoch": 0.05, "learning_rate": 4.9717144933129156e-05, "loss": 1.5984, "step": 48023 }, { "epoch": 0.05, "learning_rate": 4.971713315075329e-05, "loss": 2.2172, "step": 48024 }, { "epoch": 0.05, "learning_rate": 4.9717121368133424e-05, "loss": 1.0451, "step": 48025 }, { "epoch": 0.05, "learning_rate": 4.971710958526957e-05, "loss": 0.6342, "step": 48026 }, { "epoch": 0.05, "learning_rate": 4.9717097802161706e-05, "loss": 0.5851, "step": 48027 }, { "epoch": 0.05, "learning_rate": 4.971708601880986e-05, "loss": 0.9174, "step": 48028 }, { "epoch": 0.05, "learning_rate": 4.9717074235214e-05, "loss": 1.1595, "step": 48029 }, { "epoch": 0.05, "learning_rate": 4.9717062451374155e-05, "loss": 0.9544, "step": 48030 }, { "epoch": 0.05, "learning_rate": 4.971705066729031e-05, "loss": 0.7667, "step": 48031 }, { "epoch": 0.05, "learning_rate": 4.971703888296246e-05, "loss": 0.7752, "step": 48032 }, { "epoch": 0.05, "learning_rate": 4.9717027098390635e-05, "loss": 0.5716, "step": 48033 }, { "epoch": 0.05, "learning_rate": 4.97170153135748e-05, "loss": 0.7508, "step": 48034 }, { "epoch": 0.05, "learning_rate": 4.971700352851497e-05, "loss": 0.7066, "step": 48035 }, { "epoch": 0.05, "learning_rate": 4.971699174321114e-05, "loss": 0.8501, "step": 48036 }, { "epoch": 0.05, "learning_rate": 4.9716979957663325e-05, "loss": 0.7771, "step": 48037 }, { "epoch": 0.05, "learning_rate": 4.97169681718715e-05, "loss": 1.1123, "step": 48038 }, { "epoch": 0.05, "learning_rate": 4.971695638583569e-05, "loss": 1.0383, "step": 48039 }, { "epoch": 0.05, "learning_rate": 4.9716944599555885e-05, "loss": 1.2719, "step": 48040 }, { "epoch": 0.05, "learning_rate": 4.971693281303208e-05, "loss": 0.9525, "step": 48041 }, { "epoch": 0.05, "learning_rate": 4.9716921026264285e-05, "loss": 0.7107, "step": 48042 }, { "epoch": 0.05, "learning_rate": 4.9716909239252496e-05, "loss": 1.2465, "step": 48043 }, { "epoch": 0.05, "learning_rate": 4.97168974519967e-05, "loss": 1.1956, "step": 48044 }, { "epoch": 0.05, "learning_rate": 4.9716885664496924e-05, "loss": 0.6923, "step": 48045 }, { "epoch": 0.05, "learning_rate": 4.971687387675314e-05, "loss": 0.925, "step": 48046 }, { "epoch": 0.05, "learning_rate": 4.971686208876537e-05, "loss": 0.9913, "step": 48047 }, { "epoch": 0.05, "learning_rate": 4.971685030053361e-05, "loss": 1.0654, "step": 48048 }, { "epoch": 0.05, "learning_rate": 4.971683851205785e-05, "loss": 0.6742, "step": 48049 }, { "epoch": 0.05, "learning_rate": 4.971682672333809e-05, "loss": 0.7675, "step": 48050 }, { "epoch": 0.05, "learning_rate": 4.9716814934374346e-05, "loss": 1.2207, "step": 48051 }, { "epoch": 0.05, "learning_rate": 4.97168031451666e-05, "loss": 0.9538, "step": 48052 }, { "epoch": 0.05, "learning_rate": 4.971679135571487e-05, "loss": 1.0385, "step": 48053 }, { "epoch": 0.05, "learning_rate": 4.9716779566019136e-05, "loss": 0.7888, "step": 48054 }, { "epoch": 0.05, "learning_rate": 4.971676777607941e-05, "loss": 0.9701, "step": 48055 }, { "epoch": 0.05, "learning_rate": 4.971675598589569e-05, "loss": 1.1385, "step": 48056 }, { "epoch": 0.05, "learning_rate": 4.971674419546798e-05, "loss": 1.2607, "step": 48057 }, { "epoch": 0.05, "learning_rate": 4.971673240479627e-05, "loss": 1.2922, "step": 48058 }, { "epoch": 0.05, "learning_rate": 4.971672061388058e-05, "loss": 1.239, "step": 48059 }, { "epoch": 0.05, "learning_rate": 4.971670882272089e-05, "loss": 1.2401, "step": 48060 }, { "epoch": 0.05, "learning_rate": 4.97166970313172e-05, "loss": 1.4309, "step": 48061 }, { "epoch": 0.05, "learning_rate": 4.971668523966953e-05, "loss": 0.9893, "step": 48062 }, { "epoch": 0.05, "learning_rate": 4.971667344777786e-05, "loss": 0.6353, "step": 48063 }, { "epoch": 0.05, "learning_rate": 4.97166616556422e-05, "loss": 0.7439, "step": 48064 }, { "epoch": 0.05, "learning_rate": 4.9716649863262544e-05, "loss": 1.3575, "step": 48065 }, { "epoch": 0.05, "learning_rate": 4.97166380706389e-05, "loss": 1.2111, "step": 48066 }, { "epoch": 0.05, "learning_rate": 4.9716626277771264e-05, "loss": 0.9689, "step": 48067 }, { "epoch": 0.05, "learning_rate": 4.971661448465963e-05, "loss": 1.2028, "step": 48068 }, { "epoch": 0.05, "learning_rate": 4.9716602691304005e-05, "loss": 1.1523, "step": 48069 }, { "epoch": 0.05, "learning_rate": 4.971659089770439e-05, "loss": 0.7315, "step": 48070 }, { "epoch": 0.05, "learning_rate": 4.971657910386078e-05, "loss": 1.0987, "step": 48071 }, { "epoch": 0.05, "learning_rate": 4.971656730977319e-05, "loss": 1.1782, "step": 48072 }, { "epoch": 0.05, "learning_rate": 4.97165555154416e-05, "loss": 0.8375, "step": 48073 }, { "epoch": 0.05, "learning_rate": 4.9716543720866016e-05, "loss": 0.991, "step": 48074 }, { "epoch": 0.05, "learning_rate": 4.9716531926046446e-05, "loss": 1.0213, "step": 48075 }, { "epoch": 0.05, "learning_rate": 4.9716520130982877e-05, "loss": 1.3297, "step": 48076 }, { "epoch": 0.05, "learning_rate": 4.971650833567533e-05, "loss": 1.3436, "step": 48077 }, { "epoch": 0.05, "learning_rate": 4.971649654012378e-05, "loss": 1.0207, "step": 48078 }, { "epoch": 0.05, "learning_rate": 4.9716484744328244e-05, "loss": 0.9513, "step": 48079 }, { "epoch": 0.05, "learning_rate": 4.971647294828872e-05, "loss": 0.9409, "step": 48080 }, { "epoch": 0.05, "learning_rate": 4.97164611520052e-05, "loss": 1.1352, "step": 48081 }, { "epoch": 0.05, "learning_rate": 4.971644935547769e-05, "loss": 1.1322, "step": 48082 }, { "epoch": 0.05, "learning_rate": 4.9716437558706197e-05, "loss": 1.057, "step": 48083 }, { "epoch": 0.05, "learning_rate": 4.9716425761690704e-05, "loss": 1.1758, "step": 48084 }, { "epoch": 0.05, "learning_rate": 4.971641396443123e-05, "loss": 1.2867, "step": 48085 }, { "epoch": 0.05, "learning_rate": 4.9716402166927754e-05, "loss": 1.1031, "step": 48086 }, { "epoch": 0.05, "learning_rate": 4.9716390369180296e-05, "loss": 1.0397, "step": 48087 }, { "epoch": 0.05, "learning_rate": 4.971637857118885e-05, "loss": 1.2385, "step": 48088 }, { "epoch": 0.05, "learning_rate": 4.9716366772953415e-05, "loss": 1.0673, "step": 48089 }, { "epoch": 0.05, "learning_rate": 4.9716354974473985e-05, "loss": 1.2124, "step": 48090 }, { "epoch": 0.05, "learning_rate": 4.971634317575057e-05, "loss": 1.2033, "step": 48091 }, { "epoch": 0.05, "learning_rate": 4.971633137678316e-05, "loss": 1.0696, "step": 48092 }, { "epoch": 0.05, "learning_rate": 4.9716319577571766e-05, "loss": 1.0772, "step": 48093 }, { "epoch": 0.05, "learning_rate": 4.971630777811638e-05, "loss": 1.161, "step": 48094 }, { "epoch": 0.05, "learning_rate": 4.9716295978417004e-05, "loss": 1.1974, "step": 48095 }, { "epoch": 0.05, "learning_rate": 4.971628417847364e-05, "loss": 1.0854, "step": 48096 }, { "epoch": 0.05, "learning_rate": 4.971627237828629e-05, "loss": 1.6331, "step": 48097 }, { "epoch": 0.05, "learning_rate": 4.971626057785495e-05, "loss": 1.1804, "step": 48098 }, { "epoch": 0.05, "learning_rate": 4.971624877717962e-05, "loss": 1.0956, "step": 48099 }, { "epoch": 0.05, "learning_rate": 4.971623697626031e-05, "loss": 1.1068, "step": 48100 }, { "epoch": 0.05, "learning_rate": 4.9716225175096996e-05, "loss": 1.0246, "step": 48101 }, { "epoch": 0.05, "learning_rate": 4.9716213373689705e-05, "loss": 1.0875, "step": 48102 }, { "epoch": 0.05, "learning_rate": 4.971620157203842e-05, "loss": 1.1142, "step": 48103 }, { "epoch": 0.05, "learning_rate": 4.971618977014316e-05, "loss": 1.1488, "step": 48104 }, { "epoch": 0.05, "learning_rate": 4.9716177968003896e-05, "loss": 0.7732, "step": 48105 }, { "epoch": 0.05, "learning_rate": 4.9716166165620654e-05, "loss": 0.9587, "step": 48106 }, { "epoch": 0.05, "learning_rate": 4.971615436299342e-05, "loss": 1.0427, "step": 48107 }, { "epoch": 0.05, "learning_rate": 4.9716142560122205e-05, "loss": 0.9674, "step": 48108 }, { "epoch": 0.05, "learning_rate": 4.9716130757007e-05, "loss": 0.9675, "step": 48109 }, { "epoch": 0.05, "learning_rate": 4.97161189536478e-05, "loss": 1.0815, "step": 48110 }, { "epoch": 0.05, "learning_rate": 4.9716107150044625e-05, "loss": 1.0884, "step": 48111 }, { "epoch": 0.05, "learning_rate": 4.971609534619745e-05, "loss": 1.1165, "step": 48112 }, { "epoch": 0.05, "learning_rate": 4.9716083542106294e-05, "loss": 0.9338, "step": 48113 }, { "epoch": 0.05, "learning_rate": 4.9716071737771156e-05, "loss": 0.9524, "step": 48114 }, { "epoch": 0.05, "learning_rate": 4.9716059933192025e-05, "loss": 1.2227, "step": 48115 }, { "epoch": 0.05, "learning_rate": 4.971604812836891e-05, "loss": 1.0319, "step": 48116 }, { "epoch": 0.05, "learning_rate": 4.9716036323301805e-05, "loss": 0.9361, "step": 48117 }, { "epoch": 0.05, "learning_rate": 4.9716024517990716e-05, "loss": 1.0951, "step": 48118 }, { "epoch": 0.05, "learning_rate": 4.971601271243565e-05, "loss": 0.8248, "step": 48119 }, { "epoch": 0.05, "learning_rate": 4.9716000906636585e-05, "loss": 1.2658, "step": 48120 }, { "epoch": 0.05, "learning_rate": 4.971598910059354e-05, "loss": 1.2088, "step": 48121 }, { "epoch": 0.05, "learning_rate": 4.9715977294306504e-05, "loss": 0.8959, "step": 48122 }, { "epoch": 0.05, "learning_rate": 4.9715965487775484e-05, "loss": 1.2912, "step": 48123 }, { "epoch": 0.05, "learning_rate": 4.9715953681000484e-05, "loss": 1.1104, "step": 48124 }, { "epoch": 0.05, "learning_rate": 4.97159418739815e-05, "loss": 1.1616, "step": 48125 }, { "epoch": 0.05, "learning_rate": 4.971593006671852e-05, "loss": 1.037, "step": 48126 }, { "epoch": 0.05, "learning_rate": 4.9715918259211556e-05, "loss": 1.0769, "step": 48127 }, { "epoch": 0.05, "learning_rate": 4.971590645146061e-05, "loss": 1.1278, "step": 48128 }, { "epoch": 0.05, "learning_rate": 4.971589464346569e-05, "loss": 1.5623, "step": 48129 }, { "epoch": 0.05, "learning_rate": 4.9715882835226765e-05, "loss": 1.2584, "step": 48130 }, { "epoch": 0.05, "learning_rate": 4.971587102674387e-05, "loss": 1.1203, "step": 48131 }, { "epoch": 0.05, "learning_rate": 4.9715859218016994e-05, "loss": 0.8478, "step": 48132 }, { "epoch": 0.05, "learning_rate": 4.971584740904612e-05, "loss": 1.1956, "step": 48133 }, { "epoch": 0.05, "learning_rate": 4.9715835599831266e-05, "loss": 0.994, "step": 48134 }, { "epoch": 0.05, "learning_rate": 4.971582379037243e-05, "loss": 0.7587, "step": 48135 }, { "epoch": 0.05, "learning_rate": 4.9715811980669606e-05, "loss": 0.972, "step": 48136 }, { "epoch": 0.05, "learning_rate": 4.971580017072281e-05, "loss": 1.249, "step": 48137 }, { "epoch": 0.05, "learning_rate": 4.9715788360532015e-05, "loss": 1.0299, "step": 48138 }, { "epoch": 0.05, "learning_rate": 4.971577655009724e-05, "loss": 1.0984, "step": 48139 }, { "epoch": 0.05, "learning_rate": 4.971576473941848e-05, "loss": 0.9546, "step": 48140 }, { "epoch": 0.05, "learning_rate": 4.971575292849574e-05, "loss": 0.9746, "step": 48141 }, { "epoch": 0.05, "learning_rate": 4.971574111732902e-05, "loss": 0.6616, "step": 48142 }, { "epoch": 0.05, "learning_rate": 4.971572930591831e-05, "loss": 0.7333, "step": 48143 }, { "epoch": 0.05, "learning_rate": 4.9715717494263615e-05, "loss": 0.606, "step": 48144 }, { "epoch": 0.05, "learning_rate": 4.971570568236494e-05, "loss": 0.3911, "step": 48145 }, { "epoch": 0.05, "learning_rate": 4.971569387022229e-05, "loss": 0.9839, "step": 48146 }, { "epoch": 0.05, "learning_rate": 4.971568205783565e-05, "loss": 0.8897, "step": 48147 }, { "epoch": 0.05, "learning_rate": 4.971567024520503e-05, "loss": 0.9609, "step": 48148 }, { "epoch": 0.05, "learning_rate": 4.9715658432330415e-05, "loss": 1.1224, "step": 48149 }, { "epoch": 0.05, "learning_rate": 4.971564661921183e-05, "loss": 0.5149, "step": 48150 }, { "epoch": 0.05, "learning_rate": 4.971563480584926e-05, "loss": 0.461, "step": 48151 }, { "epoch": 0.05, "learning_rate": 4.971562299224271e-05, "loss": 1.1209, "step": 48152 }, { "epoch": 0.05, "learning_rate": 4.971561117839217e-05, "loss": 0.7044, "step": 48153 }, { "epoch": 0.05, "learning_rate": 4.9715599364297655e-05, "loss": 0.9387, "step": 48154 }, { "epoch": 0.05, "learning_rate": 4.971558754995915e-05, "loss": 0.7735, "step": 48155 }, { "epoch": 0.05, "learning_rate": 4.971557573537667e-05, "loss": 0.7732, "step": 48156 }, { "epoch": 0.05, "learning_rate": 4.971556392055021e-05, "loss": 0.7859, "step": 48157 }, { "epoch": 0.05, "learning_rate": 4.971555210547977e-05, "loss": 0.9281, "step": 48158 }, { "epoch": 0.05, "learning_rate": 4.9715540290165344e-05, "loss": 1.1158, "step": 48159 }, { "epoch": 0.05, "learning_rate": 4.971552847460693e-05, "loss": 1.2478, "step": 48160 }, { "epoch": 0.05, "learning_rate": 4.9715516658804546e-05, "loss": 0.8851, "step": 48161 }, { "epoch": 0.05, "learning_rate": 4.9715504842758175e-05, "loss": 0.7418, "step": 48162 }, { "epoch": 0.05, "learning_rate": 4.9715493026467824e-05, "loss": 1.4344, "step": 48163 }, { "epoch": 0.05, "learning_rate": 4.971548120993349e-05, "loss": 0.8717, "step": 48164 }, { "epoch": 0.05, "learning_rate": 4.971546939315518e-05, "loss": 0.9968, "step": 48165 }, { "epoch": 0.05, "learning_rate": 4.971545757613288e-05, "loss": 1.0082, "step": 48166 }, { "epoch": 0.05, "learning_rate": 4.9715445758866614e-05, "loss": 0.8914, "step": 48167 }, { "epoch": 0.05, "learning_rate": 4.971543394135636e-05, "loss": 0.888, "step": 48168 }, { "epoch": 0.05, "learning_rate": 4.9715422123602126e-05, "loss": 1.032, "step": 48169 }, { "epoch": 0.05, "learning_rate": 4.971541030560391e-05, "loss": 1.2738, "step": 48170 }, { "epoch": 0.05, "learning_rate": 4.9715398487361714e-05, "loss": 0.8466, "step": 48171 }, { "epoch": 0.05, "learning_rate": 4.971538666887554e-05, "loss": 1.399, "step": 48172 }, { "epoch": 0.05, "learning_rate": 4.9715374850145384e-05, "loss": 1.1684, "step": 48173 }, { "epoch": 0.05, "learning_rate": 4.971536303117125e-05, "loss": 1.1017, "step": 48174 }, { "epoch": 0.05, "learning_rate": 4.971535121195313e-05, "loss": 1.1232, "step": 48175 }, { "epoch": 0.05, "learning_rate": 4.971533939249105e-05, "loss": 1.0426, "step": 48176 }, { "epoch": 0.05, "learning_rate": 4.971532757278497e-05, "loss": 1.197, "step": 48177 }, { "epoch": 0.05, "learning_rate": 4.971531575283491e-05, "loss": 1.2603, "step": 48178 }, { "epoch": 0.05, "learning_rate": 4.971530393264088e-05, "loss": 1.0025, "step": 48179 }, { "epoch": 0.05, "learning_rate": 4.971529211220287e-05, "loss": 0.9819, "step": 48180 }, { "epoch": 0.05, "learning_rate": 4.971528029152088e-05, "loss": 0.7587, "step": 48181 }, { "epoch": 0.05, "learning_rate": 4.971526847059492e-05, "loss": 1.1808, "step": 48182 }, { "epoch": 0.05, "learning_rate": 4.9715256649424966e-05, "loss": 1.0024, "step": 48183 }, { "epoch": 0.05, "learning_rate": 4.971524482801104e-05, "loss": 1.0354, "step": 48184 }, { "epoch": 0.05, "learning_rate": 4.971523300635313e-05, "loss": 1.1664, "step": 48185 }, { "epoch": 0.05, "learning_rate": 4.9715221184451255e-05, "loss": 1.2295, "step": 48186 }, { "epoch": 0.05, "learning_rate": 4.971520936230539e-05, "loss": 0.8573, "step": 48187 }, { "epoch": 0.05, "learning_rate": 4.971519753991556e-05, "loss": 1.0284, "step": 48188 }, { "epoch": 0.05, "learning_rate": 4.971518571728174e-05, "loss": 0.7699, "step": 48189 }, { "epoch": 0.05, "learning_rate": 4.971517389440394e-05, "loss": 0.8879, "step": 48190 }, { "epoch": 0.05, "learning_rate": 4.971516207128217e-05, "loss": 1.0769, "step": 48191 }, { "epoch": 0.05, "learning_rate": 4.971515024791642e-05, "loss": 1.0179, "step": 48192 }, { "epoch": 0.05, "learning_rate": 4.971513842430669e-05, "loss": 0.9185, "step": 48193 }, { "epoch": 0.05, "learning_rate": 4.971512660045299e-05, "loss": 1.1658, "step": 48194 }, { "epoch": 0.05, "learning_rate": 4.9715114776355305e-05, "loss": 1.081, "step": 48195 }, { "epoch": 0.05, "learning_rate": 4.971510295201364e-05, "loss": 1.241, "step": 48196 }, { "epoch": 0.05, "learning_rate": 4.9715091127428006e-05, "loss": 1.2099, "step": 48197 }, { "epoch": 0.05, "learning_rate": 4.971507930259839e-05, "loss": 1.2638, "step": 48198 }, { "epoch": 0.05, "learning_rate": 4.97150674775248e-05, "loss": 0.7353, "step": 48199 }, { "epoch": 0.05, "learning_rate": 4.971505565220723e-05, "loss": 0.7259, "step": 48200 }, { "epoch": 0.05, "learning_rate": 4.971504382664569e-05, "loss": 0.6202, "step": 48201 }, { "epoch": 0.05, "learning_rate": 4.971503200084017e-05, "loss": 0.7004, "step": 48202 }, { "epoch": 0.05, "learning_rate": 4.971502017479067e-05, "loss": 0.6727, "step": 48203 }, { "epoch": 0.05, "learning_rate": 4.97150083484972e-05, "loss": 0.8411, "step": 48204 }, { "epoch": 0.05, "learning_rate": 4.971499652195975e-05, "loss": 0.7956, "step": 48205 }, { "epoch": 0.05, "learning_rate": 4.971498469517832e-05, "loss": 0.6019, "step": 48206 }, { "epoch": 0.05, "learning_rate": 4.971497286815292e-05, "loss": 0.7997, "step": 48207 }, { "epoch": 0.05, "learning_rate": 4.9714961040883544e-05, "loss": 0.4504, "step": 48208 }, { "epoch": 0.05, "learning_rate": 4.971494921337019e-05, "loss": 0.5794, "step": 48209 }, { "epoch": 0.05, "learning_rate": 4.971493738561286e-05, "loss": 0.6477, "step": 48210 }, { "epoch": 0.05, "learning_rate": 4.9714925557611556e-05, "loss": 0.6188, "step": 48211 }, { "epoch": 0.05, "learning_rate": 4.971491372936628e-05, "loss": 0.6497, "step": 48212 }, { "epoch": 0.05, "learning_rate": 4.9714901900877026e-05, "loss": 1.0867, "step": 48213 }, { "epoch": 0.05, "learning_rate": 4.9714890072143795e-05, "loss": 1.1996, "step": 48214 }, { "epoch": 0.05, "learning_rate": 4.971487824316659e-05, "loss": 1.1492, "step": 48215 }, { "epoch": 0.05, "learning_rate": 4.9714866413945416e-05, "loss": 1.1247, "step": 48216 }, { "epoch": 0.05, "learning_rate": 4.971485458448026e-05, "loss": 1.2363, "step": 48217 }, { "epoch": 0.05, "learning_rate": 4.971484275477113e-05, "loss": 0.8724, "step": 48218 }, { "epoch": 0.05, "learning_rate": 4.9714830924818025e-05, "loss": 0.6603, "step": 48219 }, { "epoch": 0.05, "learning_rate": 4.9714819094620945e-05, "loss": 1.1059, "step": 48220 }, { "epoch": 0.05, "learning_rate": 4.97148072641799e-05, "loss": 1.0778, "step": 48221 }, { "epoch": 0.05, "learning_rate": 4.971479543349487e-05, "loss": 0.5477, "step": 48222 }, { "epoch": 0.05, "learning_rate": 4.971478360256587e-05, "loss": 1.1179, "step": 48223 }, { "epoch": 0.05, "learning_rate": 4.97147717713929e-05, "loss": 1.2839, "step": 48224 }, { "epoch": 0.05, "learning_rate": 4.971475993997595e-05, "loss": 0.8564, "step": 48225 }, { "epoch": 0.05, "learning_rate": 4.971474810831502e-05, "loss": 0.7296, "step": 48226 }, { "epoch": 0.05, "learning_rate": 4.971473627641013e-05, "loss": 0.8988, "step": 48227 }, { "epoch": 0.05, "learning_rate": 4.9714724444261265e-05, "loss": 0.5811, "step": 48228 }, { "epoch": 0.05, "learning_rate": 4.971471261186842e-05, "loss": 0.967, "step": 48229 }, { "epoch": 0.05, "learning_rate": 4.9714700779231606e-05, "loss": 0.6166, "step": 48230 }, { "epoch": 0.05, "learning_rate": 4.971468894635082e-05, "loss": 1.0583, "step": 48231 }, { "epoch": 0.05, "learning_rate": 4.9714677113226056e-05, "loss": 0.9072, "step": 48232 }, { "epoch": 0.05, "learning_rate": 4.971466527985732e-05, "loss": 1.0988, "step": 48233 }, { "epoch": 0.05, "learning_rate": 4.9714653446244616e-05, "loss": 0.9428, "step": 48234 }, { "epoch": 0.05, "learning_rate": 4.971464161238793e-05, "loss": 1.2128, "step": 48235 }, { "epoch": 0.05, "learning_rate": 4.971462977828728e-05, "loss": 1.1482, "step": 48236 }, { "epoch": 0.05, "learning_rate": 4.971461794394265e-05, "loss": 1.0798, "step": 48237 }, { "epoch": 0.05, "learning_rate": 4.971460610935406e-05, "loss": 0.6435, "step": 48238 }, { "epoch": 0.05, "learning_rate": 4.971459427452149e-05, "loss": 0.8129, "step": 48239 }, { "epoch": 0.05, "learning_rate": 4.971458243944495e-05, "loss": 1.2694, "step": 48240 }, { "epoch": 0.05, "learning_rate": 4.9714570604124435e-05, "loss": 0.9699, "step": 48241 }, { "epoch": 0.05, "learning_rate": 4.971455876855995e-05, "loss": 0.9653, "step": 48242 }, { "epoch": 0.05, "learning_rate": 4.9714546932751495e-05, "loss": 1.0966, "step": 48243 }, { "epoch": 0.05, "learning_rate": 4.9714535096699066e-05, "loss": 1.0625, "step": 48244 }, { "epoch": 0.05, "learning_rate": 4.971452326040267e-05, "loss": 1.3763, "step": 48245 }, { "epoch": 0.05, "learning_rate": 4.97145114238623e-05, "loss": 1.6163, "step": 48246 }, { "epoch": 0.05, "learning_rate": 4.971449958707796e-05, "loss": 1.5436, "step": 48247 }, { "epoch": 0.05, "learning_rate": 4.9714487750049644e-05, "loss": 1.3392, "step": 48248 }, { "epoch": 0.05, "learning_rate": 4.971447591277736e-05, "loss": 1.5116, "step": 48249 }, { "epoch": 0.05, "learning_rate": 4.97144640752611e-05, "loss": 1.5839, "step": 48250 }, { "epoch": 0.05, "learning_rate": 4.971445223750088e-05, "loss": 1.2809, "step": 48251 }, { "epoch": 0.05, "learning_rate": 4.9714440399496675e-05, "loss": 1.2041, "step": 48252 }, { "epoch": 0.05, "learning_rate": 4.971442856124851e-05, "loss": 1.6521, "step": 48253 }, { "epoch": 0.05, "learning_rate": 4.971441672275637e-05, "loss": 1.3729, "step": 48254 }, { "epoch": 0.05, "learning_rate": 4.9714404884020264e-05, "loss": 1.483, "step": 48255 }, { "epoch": 0.05, "learning_rate": 4.9714393045040185e-05, "loss": 1.3114, "step": 48256 }, { "epoch": 0.05, "learning_rate": 4.971438120581614e-05, "loss": 1.4574, "step": 48257 }, { "epoch": 0.05, "learning_rate": 4.971436936634812e-05, "loss": 1.2347, "step": 48258 }, { "epoch": 0.05, "learning_rate": 4.971435752663614e-05, "loss": 1.3955, "step": 48259 }, { "epoch": 0.05, "learning_rate": 4.9714345686680174e-05, "loss": 1.3422, "step": 48260 }, { "epoch": 0.05, "learning_rate": 4.971433384648025e-05, "loss": 1.4578, "step": 48261 }, { "epoch": 0.05, "learning_rate": 4.971432200603635e-05, "loss": 1.384, "step": 48262 }, { "epoch": 0.05, "learning_rate": 4.971431016534849e-05, "loss": 1.1995, "step": 48263 }, { "epoch": 0.05, "learning_rate": 4.9714298324416655e-05, "loss": 1.3643, "step": 48264 }, { "epoch": 0.05, "learning_rate": 4.971428648324085e-05, "loss": 1.12, "step": 48265 }, { "epoch": 0.05, "learning_rate": 4.971427464182107e-05, "loss": 0.9777, "step": 48266 }, { "epoch": 0.05, "learning_rate": 4.971426280015733e-05, "loss": 0.9851, "step": 48267 }, { "epoch": 0.05, "learning_rate": 4.971425095824962e-05, "loss": 0.6717, "step": 48268 }, { "epoch": 0.05, "learning_rate": 4.971423911609794e-05, "loss": 0.7369, "step": 48269 }, { "epoch": 0.05, "learning_rate": 4.9714227273702284e-05, "loss": 1.0874, "step": 48270 }, { "epoch": 0.05, "learning_rate": 4.971421543106267e-05, "loss": 0.8166, "step": 48271 }, { "epoch": 0.05, "learning_rate": 4.971420358817909e-05, "loss": 1.1024, "step": 48272 }, { "epoch": 0.05, "learning_rate": 4.971419174505153e-05, "loss": 0.8955, "step": 48273 }, { "epoch": 0.05, "learning_rate": 4.971417990168001e-05, "loss": 0.9641, "step": 48274 }, { "epoch": 0.05, "learning_rate": 4.971416805806452e-05, "loss": 0.6695, "step": 48275 }, { "epoch": 0.05, "learning_rate": 4.971415621420506e-05, "loss": 1.1274, "step": 48276 }, { "epoch": 0.05, "learning_rate": 4.971414437010163e-05, "loss": 1.0125, "step": 48277 }, { "epoch": 0.05, "learning_rate": 4.9714132525754245e-05, "loss": 0.9267, "step": 48278 }, { "epoch": 0.05, "learning_rate": 4.971412068116288e-05, "loss": 0.9961, "step": 48279 }, { "epoch": 0.05, "learning_rate": 4.9714108836327556e-05, "loss": 1.3077, "step": 48280 }, { "epoch": 0.05, "learning_rate": 4.971409699124826e-05, "loss": 1.3257, "step": 48281 }, { "epoch": 0.05, "learning_rate": 4.9714085145925e-05, "loss": 0.7463, "step": 48282 }, { "epoch": 0.05, "learning_rate": 4.971407330035777e-05, "loss": 1.3837, "step": 48283 }, { "epoch": 0.05, "learning_rate": 4.9714061454546576e-05, "loss": 1.2864, "step": 48284 }, { "epoch": 0.05, "learning_rate": 4.9714049608491406e-05, "loss": 0.8601, "step": 48285 }, { "epoch": 0.05, "learning_rate": 4.9714037762192284e-05, "loss": 1.1816, "step": 48286 }, { "epoch": 0.05, "learning_rate": 4.971402591564919e-05, "loss": 1.0047, "step": 48287 }, { "epoch": 0.05, "learning_rate": 4.971401406886212e-05, "loss": 0.8906, "step": 48288 }, { "epoch": 0.05, "learning_rate": 4.971400222183109e-05, "loss": 0.8717, "step": 48289 }, { "epoch": 0.05, "learning_rate": 4.97139903745561e-05, "loss": 0.8148, "step": 48290 }, { "epoch": 0.05, "learning_rate": 4.971397852703714e-05, "loss": 0.7582, "step": 48291 }, { "epoch": 0.05, "learning_rate": 4.971396667927421e-05, "loss": 0.995, "step": 48292 }, { "epoch": 0.05, "learning_rate": 4.971395483126732e-05, "loss": 1.0577, "step": 48293 }, { "epoch": 0.05, "learning_rate": 4.971394298301646e-05, "loss": 0.6799, "step": 48294 }, { "epoch": 0.05, "learning_rate": 4.9713931134521636e-05, "loss": 1.0777, "step": 48295 }, { "epoch": 0.05, "learning_rate": 4.971391928578284e-05, "loss": 0.4303, "step": 48296 }, { "epoch": 0.05, "learning_rate": 4.971390743680009e-05, "loss": 0.6735, "step": 48297 }, { "epoch": 0.05, "learning_rate": 4.9713895587573364e-05, "loss": 0.8147, "step": 48298 }, { "epoch": 0.05, "learning_rate": 4.9713883738102686e-05, "loss": 0.789, "step": 48299 }, { "epoch": 0.05, "learning_rate": 4.971387188838803e-05, "loss": 1.0791, "step": 48300 }, { "epoch": 0.05, "learning_rate": 4.971386003842941e-05, "loss": 1.0116, "step": 48301 }, { "epoch": 0.05, "learning_rate": 4.971384818822684e-05, "loss": 1.331, "step": 48302 }, { "epoch": 0.05, "learning_rate": 4.971383633778029e-05, "loss": 1.1009, "step": 48303 }, { "epoch": 0.05, "learning_rate": 4.9713824487089777e-05, "loss": 1.3708, "step": 48304 }, { "epoch": 0.05, "learning_rate": 4.971381263615531e-05, "loss": 1.1146, "step": 48305 }, { "epoch": 0.05, "learning_rate": 4.971380078497687e-05, "loss": 1.5334, "step": 48306 }, { "epoch": 0.05, "learning_rate": 4.971378893355446e-05, "loss": 0.8024, "step": 48307 }, { "epoch": 0.05, "learning_rate": 4.97137770818881e-05, "loss": 1.0016, "step": 48308 }, { "epoch": 0.05, "learning_rate": 4.971376522997777e-05, "loss": 0.6935, "step": 48309 }, { "epoch": 0.05, "learning_rate": 4.971375337782347e-05, "loss": 0.6951, "step": 48310 }, { "epoch": 0.05, "learning_rate": 4.971374152542522e-05, "loss": 0.4637, "step": 48311 }, { "epoch": 0.05, "learning_rate": 4.9713729672782994e-05, "loss": 0.6042, "step": 48312 }, { "epoch": 0.05, "learning_rate": 4.97137178198968e-05, "loss": 1.4435, "step": 48313 }, { "epoch": 0.05, "learning_rate": 4.971370596676666e-05, "loss": 1.0431, "step": 48314 }, { "epoch": 0.05, "learning_rate": 4.971369411339254e-05, "loss": 1.0038, "step": 48315 }, { "epoch": 0.05, "learning_rate": 4.971368225977447e-05, "loss": 0.8734, "step": 48316 }, { "epoch": 0.05, "learning_rate": 4.971367040591243e-05, "loss": 1.1688, "step": 48317 }, { "epoch": 0.05, "learning_rate": 4.971365855180643e-05, "loss": 1.2204, "step": 48318 }, { "epoch": 0.05, "learning_rate": 4.971364669745646e-05, "loss": 1.1594, "step": 48319 }, { "epoch": 0.05, "learning_rate": 4.9713634842862534e-05, "loss": 0.9076, "step": 48320 }, { "epoch": 0.05, "learning_rate": 4.971362298802464e-05, "loss": 0.8451, "step": 48321 }, { "epoch": 0.05, "learning_rate": 4.971361113294279e-05, "loss": 1.292, "step": 48322 }, { "epoch": 0.05, "learning_rate": 4.971359927761698e-05, "loss": 1.3105, "step": 48323 }, { "epoch": 0.05, "learning_rate": 4.9713587422047206e-05, "loss": 1.0459, "step": 48324 }, { "epoch": 0.05, "learning_rate": 4.9713575566233464e-05, "loss": 1.1468, "step": 48325 }, { "epoch": 0.05, "learning_rate": 4.971356371017576e-05, "loss": 0.8585, "step": 48326 }, { "epoch": 0.05, "learning_rate": 4.97135518538741e-05, "loss": 0.9773, "step": 48327 }, { "epoch": 0.05, "learning_rate": 4.971353999732847e-05, "loss": 0.6794, "step": 48328 }, { "epoch": 0.05, "learning_rate": 4.971352814053889e-05, "loss": 0.9444, "step": 48329 }, { "epoch": 0.05, "learning_rate": 4.971351628350534e-05, "loss": 0.3243, "step": 48330 }, { "epoch": 0.05, "learning_rate": 4.971350442622783e-05, "loss": 0.7521, "step": 48331 }, { "epoch": 0.05, "learning_rate": 4.971349256870636e-05, "loss": 1.1423, "step": 48332 }, { "epoch": 0.05, "learning_rate": 4.971348071094093e-05, "loss": 0.8524, "step": 48333 }, { "epoch": 0.05, "learning_rate": 4.971346885293153e-05, "loss": 1.1634, "step": 48334 }, { "epoch": 0.05, "learning_rate": 4.971345699467818e-05, "loss": 1.3447, "step": 48335 }, { "epoch": 0.05, "learning_rate": 4.971344513618087e-05, "loss": 1.186, "step": 48336 }, { "epoch": 0.05, "learning_rate": 4.971343327743959e-05, "loss": 0.987, "step": 48337 }, { "epoch": 0.05, "learning_rate": 4.9713421418454356e-05, "loss": 1.3992, "step": 48338 }, { "epoch": 0.05, "learning_rate": 4.971340955922516e-05, "loss": 1.214, "step": 48339 }, { "epoch": 0.05, "learning_rate": 4.9713397699752006e-05, "loss": 0.7314, "step": 48340 }, { "epoch": 0.05, "learning_rate": 4.9713385840034886e-05, "loss": 1.2193, "step": 48341 }, { "epoch": 0.05, "learning_rate": 4.971337398007381e-05, "loss": 1.152, "step": 48342 }, { "epoch": 0.05, "learning_rate": 4.9713362119868776e-05, "loss": 1.4183, "step": 48343 }, { "epoch": 0.05, "learning_rate": 4.971335025941978e-05, "loss": 0.946, "step": 48344 }, { "epoch": 0.05, "learning_rate": 4.9713338398726816e-05, "loss": 0.7581, "step": 48345 }, { "epoch": 0.05, "learning_rate": 4.97133265377899e-05, "loss": 0.9171, "step": 48346 }, { "epoch": 0.05, "learning_rate": 4.9713314676609026e-05, "loss": 1.0617, "step": 48347 }, { "epoch": 0.05, "learning_rate": 4.971330281518419e-05, "loss": 0.9865, "step": 48348 }, { "epoch": 0.05, "learning_rate": 4.9713290953515394e-05, "loss": 1.0254, "step": 48349 }, { "epoch": 0.05, "learning_rate": 4.9713279091602635e-05, "loss": 0.998, "step": 48350 }, { "epoch": 0.05, "learning_rate": 4.9713267229445925e-05, "loss": 1.0926, "step": 48351 }, { "epoch": 0.05, "learning_rate": 4.9713255367045255e-05, "loss": 1.0854, "step": 48352 }, { "epoch": 0.05, "learning_rate": 4.971324350440063e-05, "loss": 1.2054, "step": 48353 }, { "epoch": 0.05, "learning_rate": 4.971323164151204e-05, "loss": 1.0275, "step": 48354 }, { "epoch": 0.05, "learning_rate": 4.9713219778379486e-05, "loss": 0.9868, "step": 48355 }, { "epoch": 0.05, "learning_rate": 4.971320791500298e-05, "loss": 0.888, "step": 48356 }, { "epoch": 0.05, "learning_rate": 4.9713196051382516e-05, "loss": 0.9873, "step": 48357 }, { "epoch": 0.05, "learning_rate": 4.971318418751809e-05, "loss": 1.0779, "step": 48358 }, { "epoch": 0.05, "learning_rate": 4.971317232340971e-05, "loss": 0.9165, "step": 48359 }, { "epoch": 0.05, "learning_rate": 4.9713160459057374e-05, "loss": 0.9478, "step": 48360 }, { "epoch": 0.05, "learning_rate": 4.971314859446107e-05, "loss": 1.0385, "step": 48361 }, { "epoch": 0.05, "learning_rate": 4.971313672962082e-05, "loss": 1.1668, "step": 48362 }, { "epoch": 0.05, "learning_rate": 4.971312486453661e-05, "loss": 1.0975, "step": 48363 }, { "epoch": 0.05, "learning_rate": 4.9713112999208437e-05, "loss": 0.9315, "step": 48364 }, { "epoch": 0.05, "learning_rate": 4.9713101133636306e-05, "loss": 0.9141, "step": 48365 }, { "epoch": 0.05, "learning_rate": 4.9713089267820224e-05, "loss": 0.9021, "step": 48366 }, { "epoch": 0.05, "learning_rate": 4.9713077401760175e-05, "loss": 1.1267, "step": 48367 }, { "epoch": 0.05, "learning_rate": 4.971306553545618e-05, "loss": 1.3559, "step": 48368 }, { "epoch": 0.05, "learning_rate": 4.971305366890823e-05, "loss": 1.1094, "step": 48369 }, { "epoch": 0.05, "learning_rate": 4.971304180211631e-05, "loss": 1.0248, "step": 48370 }, { "epoch": 0.05, "learning_rate": 4.971302993508044e-05, "loss": 1.1411, "step": 48371 }, { "epoch": 0.05, "learning_rate": 4.9713018067800616e-05, "loss": 0.7277, "step": 48372 }, { "epoch": 0.05, "learning_rate": 4.9713006200276834e-05, "loss": 1.1599, "step": 48373 }, { "epoch": 0.05, "learning_rate": 4.9712994332509086e-05, "loss": 0.9232, "step": 48374 }, { "epoch": 0.05, "learning_rate": 4.971298246449739e-05, "loss": 0.8142, "step": 48375 }, { "epoch": 0.05, "learning_rate": 4.9712970596241746e-05, "loss": 0.8411, "step": 48376 }, { "epoch": 0.05, "learning_rate": 4.9712958727742134e-05, "loss": 0.9631, "step": 48377 }, { "epoch": 0.05, "learning_rate": 4.971294685899857e-05, "loss": 0.8334, "step": 48378 }, { "epoch": 0.05, "learning_rate": 4.971293499001105e-05, "loss": 0.9917, "step": 48379 }, { "epoch": 0.05, "learning_rate": 4.971292312077957e-05, "loss": 0.7265, "step": 48380 }, { "epoch": 0.05, "learning_rate": 4.9712911251304145e-05, "loss": 1.0332, "step": 48381 }, { "epoch": 0.05, "learning_rate": 4.971289938158476e-05, "loss": 0.8594, "step": 48382 }, { "epoch": 0.05, "learning_rate": 4.971288751162142e-05, "loss": 0.9658, "step": 48383 }, { "epoch": 0.05, "learning_rate": 4.971287564141412e-05, "loss": 0.601, "step": 48384 }, { "epoch": 0.05, "learning_rate": 4.971286377096287e-05, "loss": 1.1517, "step": 48385 }, { "epoch": 0.05, "learning_rate": 4.971285190026766e-05, "loss": 0.9735, "step": 48386 }, { "epoch": 0.05, "learning_rate": 4.97128400293285e-05, "loss": 1.0634, "step": 48387 }, { "epoch": 0.05, "learning_rate": 4.971282815814539e-05, "loss": 1.1216, "step": 48388 }, { "epoch": 0.05, "learning_rate": 4.971281628671831e-05, "loss": 1.0983, "step": 48389 }, { "epoch": 0.05, "learning_rate": 4.971280441504729e-05, "loss": 1.0593, "step": 48390 }, { "epoch": 0.05, "learning_rate": 4.9712792543132306e-05, "loss": 0.9286, "step": 48391 }, { "epoch": 0.05, "learning_rate": 4.971278067097337e-05, "loss": 1.1897, "step": 48392 }, { "epoch": 0.05, "learning_rate": 4.971276879857049e-05, "loss": 1.196, "step": 48393 }, { "epoch": 0.05, "learning_rate": 4.971275692592364e-05, "loss": 1.1402, "step": 48394 }, { "epoch": 0.05, "learning_rate": 4.971274505303285e-05, "loss": 0.7911, "step": 48395 }, { "epoch": 0.05, "learning_rate": 4.9712733179898095e-05, "loss": 1.1035, "step": 48396 }, { "epoch": 0.05, "learning_rate": 4.97127213065194e-05, "loss": 1.1298, "step": 48397 }, { "epoch": 0.05, "learning_rate": 4.9712709432896734e-05, "loss": 1.1184, "step": 48398 }, { "epoch": 0.05, "learning_rate": 4.971269755903012e-05, "loss": 0.939, "step": 48399 }, { "epoch": 0.05, "learning_rate": 4.971268568491956e-05, "loss": 1.0835, "step": 48400 }, { "epoch": 0.05, "learning_rate": 4.9712673810565044e-05, "loss": 0.6795, "step": 48401 }, { "epoch": 0.05, "learning_rate": 4.971266193596657e-05, "loss": 0.3941, "step": 48402 }, { "epoch": 0.05, "learning_rate": 4.9712650061124146e-05, "loss": 1.2219, "step": 48403 }, { "epoch": 0.05, "learning_rate": 4.971263818603777e-05, "loss": 1.2746, "step": 48404 }, { "epoch": 0.05, "learning_rate": 4.971262631070745e-05, "loss": 1.0386, "step": 48405 }, { "epoch": 0.05, "learning_rate": 4.971261443513316e-05, "loss": 0.9362, "step": 48406 }, { "epoch": 0.05, "learning_rate": 4.9712602559314925e-05, "loss": 0.7915, "step": 48407 }, { "epoch": 0.05, "learning_rate": 4.971259068325274e-05, "loss": 1.1762, "step": 48408 }, { "epoch": 0.05, "learning_rate": 4.97125788069466e-05, "loss": 1.0496, "step": 48409 }, { "epoch": 0.05, "learning_rate": 4.971256693039651e-05, "loss": 1.3837, "step": 48410 }, { "epoch": 0.05, "learning_rate": 4.971255505360246e-05, "loss": 1.1336, "step": 48411 }, { "epoch": 0.05, "learning_rate": 4.971254317656447e-05, "loss": 1.1903, "step": 48412 }, { "epoch": 0.05, "learning_rate": 4.9712531299282524e-05, "loss": 1.3855, "step": 48413 }, { "epoch": 0.05, "learning_rate": 4.971251942175662e-05, "loss": 0.9151, "step": 48414 }, { "epoch": 0.05, "learning_rate": 4.971250754398677e-05, "loss": 1.0028, "step": 48415 }, { "epoch": 0.05, "learning_rate": 4.971249566597297e-05, "loss": 0.9208, "step": 48416 }, { "epoch": 0.05, "learning_rate": 4.971248378771521e-05, "loss": 1.1445, "step": 48417 }, { "epoch": 0.05, "learning_rate": 4.971247190921351e-05, "loss": 1.1349, "step": 48418 }, { "epoch": 0.05, "learning_rate": 4.9712460030467856e-05, "loss": 1.0918, "step": 48419 }, { "epoch": 0.05, "learning_rate": 4.971244815147825e-05, "loss": 1.2467, "step": 48420 }, { "epoch": 0.05, "learning_rate": 4.971243627224469e-05, "loss": 1.1351, "step": 48421 }, { "epoch": 0.05, "learning_rate": 4.9712424392767186e-05, "loss": 1.2098, "step": 48422 }, { "epoch": 0.05, "learning_rate": 4.971241251304572e-05, "loss": 1.2855, "step": 48423 }, { "epoch": 0.05, "learning_rate": 4.9712400633080314e-05, "loss": 1.5861, "step": 48424 }, { "epoch": 0.05, "learning_rate": 4.971238875287095e-05, "loss": 0.7613, "step": 48425 }, { "epoch": 0.05, "learning_rate": 4.971237687241764e-05, "loss": 0.8408, "step": 48426 }, { "epoch": 0.05, "learning_rate": 4.971236499172038e-05, "loss": 0.785, "step": 48427 }, { "epoch": 0.05, "learning_rate": 4.971235311077918e-05, "loss": 1.0255, "step": 48428 }, { "epoch": 0.05, "learning_rate": 4.971234122959402e-05, "loss": 0.8368, "step": 48429 }, { "epoch": 0.05, "learning_rate": 4.971232934816491e-05, "loss": 0.9744, "step": 48430 }, { "epoch": 0.05, "learning_rate": 4.971231746649185e-05, "loss": 0.9773, "step": 48431 }, { "epoch": 0.05, "learning_rate": 4.971230558457484e-05, "loss": 0.9586, "step": 48432 }, { "epoch": 0.05, "learning_rate": 4.971229370241388e-05, "loss": 1.1662, "step": 48433 }, { "epoch": 0.05, "learning_rate": 4.9712281820008974e-05, "loss": 1.0403, "step": 48434 }, { "epoch": 0.05, "learning_rate": 4.971226993736011e-05, "loss": 0.8054, "step": 48435 }, { "epoch": 0.05, "learning_rate": 4.971225805446731e-05, "loss": 1.076, "step": 48436 }, { "epoch": 0.05, "learning_rate": 4.9712246171330554e-05, "loss": 1.154, "step": 48437 }, { "epoch": 0.05, "learning_rate": 4.971223428794984e-05, "loss": 1.2392, "step": 48438 }, { "epoch": 0.05, "learning_rate": 4.9712222404325196e-05, "loss": 1.0915, "step": 48439 }, { "epoch": 0.05, "learning_rate": 4.971221052045659e-05, "loss": 1.0251, "step": 48440 }, { "epoch": 0.05, "learning_rate": 4.971219863634404e-05, "loss": 1.0397, "step": 48441 }, { "epoch": 0.05, "learning_rate": 4.9712186751987546e-05, "loss": 1.2179, "step": 48442 }, { "epoch": 0.05, "learning_rate": 4.97121748673871e-05, "loss": 0.8959, "step": 48443 }, { "epoch": 0.05, "learning_rate": 4.97121629825427e-05, "loss": 1.1417, "step": 48444 }, { "epoch": 0.05, "learning_rate": 4.971215109745436e-05, "loss": 1.0406, "step": 48445 }, { "epoch": 0.05, "learning_rate": 4.971213921212207e-05, "loss": 1.3462, "step": 48446 }, { "epoch": 0.05, "learning_rate": 4.971212732654583e-05, "loss": 0.7458, "step": 48447 }, { "epoch": 0.05, "learning_rate": 4.9712115440725644e-05, "loss": 0.9853, "step": 48448 }, { "epoch": 0.05, "learning_rate": 4.9712103554661504e-05, "loss": 0.7882, "step": 48449 }, { "epoch": 0.05, "learning_rate": 4.971209166835343e-05, "loss": 0.8259, "step": 48450 }, { "epoch": 0.05, "learning_rate": 4.97120797818014e-05, "loss": 1.0924, "step": 48451 }, { "epoch": 0.05, "learning_rate": 4.971206789500542e-05, "loss": 0.9681, "step": 48452 }, { "epoch": 0.05, "learning_rate": 4.97120560079655e-05, "loss": 0.8508, "step": 48453 }, { "epoch": 0.05, "learning_rate": 4.971204412068163e-05, "loss": 1.006, "step": 48454 }, { "epoch": 0.05, "learning_rate": 4.971203223315381e-05, "loss": 1.1811, "step": 48455 }, { "epoch": 0.05, "learning_rate": 4.971202034538205e-05, "loss": 0.8637, "step": 48456 }, { "epoch": 0.05, "learning_rate": 4.971200845736633e-05, "loss": 1.0028, "step": 48457 }, { "epoch": 0.05, "learning_rate": 4.9711996569106675e-05, "loss": 1.4697, "step": 48458 }, { "epoch": 0.05, "learning_rate": 4.971198468060307e-05, "loss": 1.2962, "step": 48459 }, { "epoch": 0.05, "learning_rate": 4.971197279185552e-05, "loss": 1.7505, "step": 48460 }, { "epoch": 0.05, "learning_rate": 4.971196090286403e-05, "loss": 0.9286, "step": 48461 }, { "epoch": 0.05, "learning_rate": 4.971194901362859e-05, "loss": 1.0945, "step": 48462 }, { "epoch": 0.05, "learning_rate": 4.97119371241492e-05, "loss": 0.8155, "step": 48463 }, { "epoch": 0.05, "learning_rate": 4.971192523442586e-05, "loss": 0.9138, "step": 48464 }, { "epoch": 0.05, "learning_rate": 4.971191334445858e-05, "loss": 0.9189, "step": 48465 }, { "epoch": 0.05, "learning_rate": 4.9711901454247354e-05, "loss": 1.1983, "step": 48466 }, { "epoch": 0.05, "learning_rate": 4.971188956379219e-05, "loss": 1.3873, "step": 48467 }, { "epoch": 0.05, "learning_rate": 4.971187767309307e-05, "loss": 0.7155, "step": 48468 }, { "epoch": 0.05, "learning_rate": 4.971186578215001e-05, "loss": 0.8616, "step": 48469 }, { "epoch": 0.05, "learning_rate": 4.9711853890963e-05, "loss": 0.8488, "step": 48470 }, { "epoch": 0.05, "learning_rate": 4.971184199953205e-05, "loss": 1.1662, "step": 48471 }, { "epoch": 0.05, "learning_rate": 4.971183010785715e-05, "loss": 1.3557, "step": 48472 }, { "epoch": 0.05, "learning_rate": 4.9711818215938314e-05, "loss": 0.9451, "step": 48473 }, { "epoch": 0.05, "learning_rate": 4.971180632377553e-05, "loss": 0.8799, "step": 48474 }, { "epoch": 0.05, "learning_rate": 4.97117944313688e-05, "loss": 0.9581, "step": 48475 }, { "epoch": 0.05, "learning_rate": 4.971178253871812e-05, "loss": 0.7858, "step": 48476 }, { "epoch": 0.05, "learning_rate": 4.971177064582351e-05, "loss": 1.0742, "step": 48477 }, { "epoch": 0.05, "learning_rate": 4.971175875268495e-05, "loss": 1.1656, "step": 48478 }, { "epoch": 0.05, "learning_rate": 4.9711746859302436e-05, "loss": 1.0546, "step": 48479 }, { "epoch": 0.05, "learning_rate": 4.9711734965675985e-05, "loss": 0.8366, "step": 48480 }, { "epoch": 0.05, "learning_rate": 4.971172307180559e-05, "loss": 0.7361, "step": 48481 }, { "epoch": 0.05, "learning_rate": 4.9711711177691246e-05, "loss": 0.9439, "step": 48482 }, { "epoch": 0.05, "learning_rate": 4.9711699283332966e-05, "loss": 0.9002, "step": 48483 }, { "epoch": 0.05, "learning_rate": 4.9711687388730746e-05, "loss": 0.8928, "step": 48484 }, { "epoch": 0.05, "learning_rate": 4.971167549388457e-05, "loss": 1.5427, "step": 48485 }, { "epoch": 0.05, "learning_rate": 4.971166359879446e-05, "loss": 1.132, "step": 48486 }, { "epoch": 0.05, "learning_rate": 4.97116517034604e-05, "loss": 1.4194, "step": 48487 }, { "epoch": 0.05, "learning_rate": 4.971163980788241e-05, "loss": 1.0049, "step": 48488 }, { "epoch": 0.05, "learning_rate": 4.971162791206047e-05, "loss": 1.3117, "step": 48489 }, { "epoch": 0.05, "learning_rate": 4.9711616015994575e-05, "loss": 1.1272, "step": 48490 }, { "epoch": 0.05, "learning_rate": 4.971160411968475e-05, "loss": 1.0984, "step": 48491 }, { "epoch": 0.05, "learning_rate": 4.971159222313098e-05, "loss": 1.1024, "step": 48492 }, { "epoch": 0.05, "learning_rate": 4.9711580326333265e-05, "loss": 0.9341, "step": 48493 }, { "epoch": 0.05, "learning_rate": 4.971156842929162e-05, "loss": 1.235, "step": 48494 }, { "epoch": 0.05, "learning_rate": 4.971155653200602e-05, "loss": 1.3937, "step": 48495 }, { "epoch": 0.05, "learning_rate": 4.971154463447648e-05, "loss": 1.1003, "step": 48496 }, { "epoch": 0.05, "learning_rate": 4.9711532736703e-05, "loss": 0.8194, "step": 48497 }, { "epoch": 0.05, "learning_rate": 4.9711520838685574e-05, "loss": 1.0338, "step": 48498 }, { "epoch": 0.05, "learning_rate": 4.9711508940424213e-05, "loss": 0.8924, "step": 48499 }, { "epoch": 0.05, "learning_rate": 4.971149704191891e-05, "loss": 0.5674, "step": 48500 }, { "epoch": 0.05, "eval_loss": 1.074501872062683, "eval_runtime": 128.5455, "eval_samples_per_second": 10.774, "eval_steps_per_second": 5.391, "step": 48500 }, { "epoch": 0.05, "learning_rate": 4.9711485143169655e-05, "loss": 1.0985, "step": 48501 }, { "epoch": 0.05, "learning_rate": 4.971147324417647e-05, "loss": 0.8689, "step": 48502 }, { "epoch": 0.05, "learning_rate": 4.971146134493934e-05, "loss": 0.9656, "step": 48503 }, { "epoch": 0.05, "learning_rate": 4.9711449445458266e-05, "loss": 0.8307, "step": 48504 }, { "epoch": 0.05, "learning_rate": 4.971143754573326e-05, "loss": 1.132, "step": 48505 }, { "epoch": 0.05, "learning_rate": 4.971142564576431e-05, "loss": 1.1588, "step": 48506 }, { "epoch": 0.05, "learning_rate": 4.9711413745551416e-05, "loss": 1.1469, "step": 48507 }, { "epoch": 0.05, "learning_rate": 4.971140184509457e-05, "loss": 0.9461, "step": 48508 }, { "epoch": 0.05, "learning_rate": 4.9711389944393804e-05, "loss": 0.7063, "step": 48509 }, { "epoch": 0.05, "learning_rate": 4.971137804344908e-05, "loss": 1.4273, "step": 48510 }, { "epoch": 0.05, "learning_rate": 4.971136614226043e-05, "loss": 0.9513, "step": 48511 }, { "epoch": 0.05, "learning_rate": 4.971135424082783e-05, "loss": 0.8435, "step": 48512 }, { "epoch": 0.05, "learning_rate": 4.97113423391513e-05, "loss": 0.6638, "step": 48513 }, { "epoch": 0.05, "learning_rate": 4.9711330437230826e-05, "loss": 1.3537, "step": 48514 }, { "epoch": 0.05, "learning_rate": 4.971131853506641e-05, "loss": 1.0951, "step": 48515 }, { "epoch": 0.05, "learning_rate": 4.971130663265805e-05, "loss": 0.9695, "step": 48516 }, { "epoch": 0.05, "learning_rate": 4.971129473000576e-05, "loss": 0.8266, "step": 48517 }, { "epoch": 0.05, "learning_rate": 4.971128282710952e-05, "loss": 0.8944, "step": 48518 }, { "epoch": 0.05, "learning_rate": 4.971127092396935e-05, "loss": 1.1164, "step": 48519 }, { "epoch": 0.05, "learning_rate": 4.971125902058524e-05, "loss": 1.4517, "step": 48520 }, { "epoch": 0.05, "learning_rate": 4.9711247116957186e-05, "loss": 1.452, "step": 48521 }, { "epoch": 0.05, "learning_rate": 4.9711235213085194e-05, "loss": 1.8431, "step": 48522 }, { "epoch": 0.05, "learning_rate": 4.971122330896927e-05, "loss": 1.6762, "step": 48523 }, { "epoch": 0.05, "learning_rate": 4.971121140460939e-05, "loss": 1.3176, "step": 48524 }, { "epoch": 0.05, "learning_rate": 4.971119950000559e-05, "loss": 1.2963, "step": 48525 }, { "epoch": 0.05, "learning_rate": 4.971118759515784e-05, "loss": 0.6657, "step": 48526 }, { "epoch": 0.05, "learning_rate": 4.971117569006616e-05, "loss": 0.852, "step": 48527 }, { "epoch": 0.05, "learning_rate": 4.971116378473053e-05, "loss": 1.1301, "step": 48528 }, { "epoch": 0.05, "learning_rate": 4.971115187915097e-05, "loss": 1.1912, "step": 48529 }, { "epoch": 0.05, "learning_rate": 4.971113997332747e-05, "loss": 1.0978, "step": 48530 }, { "epoch": 0.05, "learning_rate": 4.9711128067260036e-05, "loss": 0.8317, "step": 48531 }, { "epoch": 0.05, "learning_rate": 4.9711116160948663e-05, "loss": 0.9525, "step": 48532 }, { "epoch": 0.05, "learning_rate": 4.9711104254393346e-05, "loss": 0.8434, "step": 48533 }, { "epoch": 0.05, "learning_rate": 4.97110923475941e-05, "loss": 1.1512, "step": 48534 }, { "epoch": 0.05, "learning_rate": 4.9711080440550914e-05, "loss": 1.1356, "step": 48535 }, { "epoch": 0.05, "learning_rate": 4.971106853326378e-05, "loss": 1.1633, "step": 48536 }, { "epoch": 0.05, "learning_rate": 4.971105662573272e-05, "loss": 0.8429, "step": 48537 }, { "epoch": 0.05, "learning_rate": 4.971104471795772e-05, "loss": 0.78, "step": 48538 }, { "epoch": 0.05, "learning_rate": 4.9711032809938786e-05, "loss": 0.9746, "step": 48539 }, { "epoch": 0.05, "learning_rate": 4.971102090167592e-05, "loss": 1.0974, "step": 48540 }, { "epoch": 0.05, "learning_rate": 4.97110089931691e-05, "loss": 0.9833, "step": 48541 }, { "epoch": 0.05, "learning_rate": 4.971099708441836e-05, "loss": 1.1188, "step": 48542 }, { "epoch": 0.05, "learning_rate": 4.971098517542367e-05, "loss": 1.1522, "step": 48543 }, { "epoch": 0.05, "learning_rate": 4.9710973266185055e-05, "loss": 1.0231, "step": 48544 }, { "epoch": 0.05, "learning_rate": 4.971096135670249e-05, "loss": 1.1584, "step": 48545 }, { "epoch": 0.05, "learning_rate": 4.9710949446976006e-05, "loss": 1.1219, "step": 48546 }, { "epoch": 0.05, "learning_rate": 4.971093753700557e-05, "loss": 1.1912, "step": 48547 }, { "epoch": 0.05, "learning_rate": 4.971092562679121e-05, "loss": 1.2502, "step": 48548 }, { "epoch": 0.05, "learning_rate": 4.9710913716332905e-05, "loss": 1.046, "step": 48549 }, { "epoch": 0.05, "learning_rate": 4.971090180563067e-05, "loss": 1.0619, "step": 48550 }, { "epoch": 0.05, "learning_rate": 4.97108898946845e-05, "loss": 0.9154, "step": 48551 }, { "epoch": 0.05, "learning_rate": 4.9710877983494396e-05, "loss": 0.7999, "step": 48552 }, { "epoch": 0.05, "learning_rate": 4.971086607206035e-05, "loss": 0.969, "step": 48553 }, { "epoch": 0.05, "learning_rate": 4.9710854160382374e-05, "loss": 1.0181, "step": 48554 }, { "epoch": 0.05, "learning_rate": 4.9710842248460466e-05, "loss": 1.1769, "step": 48555 }, { "epoch": 0.05, "learning_rate": 4.971083033629461e-05, "loss": 1.2381, "step": 48556 }, { "epoch": 0.05, "learning_rate": 4.971081842388483e-05, "loss": 0.8589, "step": 48557 }, { "epoch": 0.05, "learning_rate": 4.9710806511231114e-05, "loss": 0.7489, "step": 48558 }, { "epoch": 0.05, "learning_rate": 4.9710794598333464e-05, "loss": 0.9899, "step": 48559 }, { "epoch": 0.05, "learning_rate": 4.9710782685191876e-05, "loss": 0.7666, "step": 48560 }, { "epoch": 0.05, "learning_rate": 4.9710770771806355e-05, "loss": 0.7765, "step": 48561 }, { "epoch": 0.05, "learning_rate": 4.9710758858176896e-05, "loss": 1.1508, "step": 48562 }, { "epoch": 0.05, "learning_rate": 4.971074694430351e-05, "loss": 1.4693, "step": 48563 }, { "epoch": 0.05, "learning_rate": 4.971073503018618e-05, "loss": 0.4549, "step": 48564 }, { "epoch": 0.05, "learning_rate": 4.9710723115824926e-05, "loss": 0.1327, "step": 48565 }, { "epoch": 0.05, "learning_rate": 4.971071120121973e-05, "loss": 0.0451, "step": 48566 }, { "epoch": 0.05, "learning_rate": 4.971069928637061e-05, "loss": 0.0699, "step": 48567 }, { "epoch": 0.05, "learning_rate": 4.9710687371277556e-05, "loss": 0.0417, "step": 48568 }, { "epoch": 0.05, "learning_rate": 4.971067545594056e-05, "loss": 0.0364, "step": 48569 }, { "epoch": 0.05, "learning_rate": 4.971066354035963e-05, "loss": 0.0233, "step": 48570 }, { "epoch": 0.05, "learning_rate": 4.971065162453478e-05, "loss": 0.0183, "step": 48571 }, { "epoch": 0.05, "learning_rate": 4.971063970846598e-05, "loss": 0.0286, "step": 48572 }, { "epoch": 0.05, "learning_rate": 4.9710627792153254e-05, "loss": 0.0233, "step": 48573 }, { "epoch": 0.05, "learning_rate": 4.97106158755966e-05, "loss": 0.0251, "step": 48574 }, { "epoch": 0.05, "learning_rate": 4.971060395879601e-05, "loss": 0.0198, "step": 48575 }, { "epoch": 0.05, "learning_rate": 4.971059204175148e-05, "loss": 0.1375, "step": 48576 }, { "epoch": 0.05, "learning_rate": 4.9710580124463025e-05, "loss": 0.1925, "step": 48577 }, { "epoch": 0.05, "learning_rate": 4.971056820693064e-05, "loss": 0.1795, "step": 48578 }, { "epoch": 0.05, "learning_rate": 4.971055628915432e-05, "loss": 0.1588, "step": 48579 }, { "epoch": 0.05, "learning_rate": 4.971054437113406e-05, "loss": 0.1162, "step": 48580 }, { "epoch": 0.05, "learning_rate": 4.971053245286988e-05, "loss": 0.1405, "step": 48581 }, { "epoch": 0.05, "learning_rate": 4.971052053436176e-05, "loss": 0.7638, "step": 48582 }, { "epoch": 0.05, "learning_rate": 4.971050861560971e-05, "loss": 0.9154, "step": 48583 }, { "epoch": 0.05, "learning_rate": 4.9710496696613726e-05, "loss": 0.9835, "step": 48584 }, { "epoch": 0.05, "learning_rate": 4.971048477737382e-05, "loss": 0.9598, "step": 48585 }, { "epoch": 0.05, "learning_rate": 4.971047285788998e-05, "loss": 0.8985, "step": 48586 }, { "epoch": 0.05, "learning_rate": 4.9710460938162195e-05, "loss": 1.1339, "step": 48587 }, { "epoch": 0.05, "learning_rate": 4.971044901819049e-05, "loss": 1.1102, "step": 48588 }, { "epoch": 0.05, "learning_rate": 4.9710437097974856e-05, "loss": 1.0664, "step": 48589 }, { "epoch": 0.05, "learning_rate": 4.971042517751528e-05, "loss": 1.0093, "step": 48590 }, { "epoch": 0.05, "learning_rate": 4.9710413256811784e-05, "loss": 1.6101, "step": 48591 }, { "epoch": 0.05, "learning_rate": 4.971040133586435e-05, "loss": 1.7868, "step": 48592 }, { "epoch": 0.05, "learning_rate": 4.9710389414672997e-05, "loss": 1.4672, "step": 48593 }, { "epoch": 0.05, "learning_rate": 4.97103774932377e-05, "loss": 1.094, "step": 48594 }, { "epoch": 0.05, "learning_rate": 4.971036557155848e-05, "loss": 1.0568, "step": 48595 }, { "epoch": 0.05, "learning_rate": 4.971035364963533e-05, "loss": 1.2639, "step": 48596 }, { "epoch": 0.05, "learning_rate": 4.9710341727468246e-05, "loss": 1.0992, "step": 48597 }, { "epoch": 0.05, "learning_rate": 4.971032980505723e-05, "loss": 1.0732, "step": 48598 }, { "epoch": 0.05, "learning_rate": 4.971031788240229e-05, "loss": 1.1305, "step": 48599 }, { "epoch": 0.05, "learning_rate": 4.971030595950342e-05, "loss": 0.9223, "step": 48600 }, { "epoch": 0.05, "learning_rate": 4.971029403636062e-05, "loss": 0.4391, "step": 48601 }, { "epoch": 0.05, "learning_rate": 4.971028211297388e-05, "loss": 0.4054, "step": 48602 }, { "epoch": 0.05, "learning_rate": 4.971027018934322e-05, "loss": 0.357, "step": 48603 }, { "epoch": 0.05, "learning_rate": 4.971025826546863e-05, "loss": 1.0632, "step": 48604 }, { "epoch": 0.05, "learning_rate": 4.9710246341350113e-05, "loss": 1.1577, "step": 48605 }, { "epoch": 0.05, "learning_rate": 4.971023441698767e-05, "loss": 1.0025, "step": 48606 }, { "epoch": 0.05, "learning_rate": 4.9710222492381286e-05, "loss": 1.275, "step": 48607 }, { "epoch": 0.05, "learning_rate": 4.971021056753098e-05, "loss": 1.2043, "step": 48608 }, { "epoch": 0.05, "learning_rate": 4.9710198642436745e-05, "loss": 0.8859, "step": 48609 }, { "epoch": 0.05, "learning_rate": 4.971018671709858e-05, "loss": 0.9674, "step": 48610 }, { "epoch": 0.05, "learning_rate": 4.971017479151648e-05, "loss": 1.1341, "step": 48611 }, { "epoch": 0.05, "learning_rate": 4.971016286569046e-05, "loss": 0.97, "step": 48612 }, { "epoch": 0.05, "learning_rate": 4.971015093962051e-05, "loss": 1.211, "step": 48613 }, { "epoch": 0.05, "learning_rate": 4.9710139013306634e-05, "loss": 0.8105, "step": 48614 }, { "epoch": 0.05, "learning_rate": 4.971012708674883e-05, "loss": 0.6126, "step": 48615 }, { "epoch": 0.05, "learning_rate": 4.971011515994709e-05, "loss": 0.8988, "step": 48616 }, { "epoch": 0.05, "learning_rate": 4.971010323290143e-05, "loss": 1.0996, "step": 48617 }, { "epoch": 0.05, "learning_rate": 4.9710091305611845e-05, "loss": 0.9781, "step": 48618 }, { "epoch": 0.05, "learning_rate": 4.9710079378078326e-05, "loss": 1.0819, "step": 48619 }, { "epoch": 0.05, "learning_rate": 4.971006745030088e-05, "loss": 0.9925, "step": 48620 }, { "epoch": 0.05, "learning_rate": 4.9710055522279506e-05, "loss": 0.7779, "step": 48621 }, { "epoch": 0.05, "learning_rate": 4.971004359401421e-05, "loss": 0.5821, "step": 48622 }, { "epoch": 0.05, "learning_rate": 4.9710031665504986e-05, "loss": 1.2952, "step": 48623 }, { "epoch": 0.05, "learning_rate": 4.9710019736751835e-05, "loss": 0.9787, "step": 48624 }, { "epoch": 0.05, "learning_rate": 4.971000780775476e-05, "loss": 1.0996, "step": 48625 }, { "epoch": 0.05, "learning_rate": 4.970999587851374e-05, "loss": 1.2526, "step": 48626 }, { "epoch": 0.05, "learning_rate": 4.970998394902881e-05, "loss": 1.1917, "step": 48627 }, { "epoch": 0.05, "learning_rate": 4.970997201929995e-05, "loss": 1.0371, "step": 48628 }, { "epoch": 0.05, "learning_rate": 4.970996008932716e-05, "loss": 0.8176, "step": 48629 }, { "epoch": 0.05, "learning_rate": 4.970994815911045e-05, "loss": 0.7958, "step": 48630 }, { "epoch": 0.05, "learning_rate": 4.970993622864981e-05, "loss": 1.0039, "step": 48631 }, { "epoch": 0.05, "learning_rate": 4.9709924297945244e-05, "loss": 0.9131, "step": 48632 }, { "epoch": 0.05, "learning_rate": 4.970991236699676e-05, "loss": 0.6566, "step": 48633 }, { "epoch": 0.05, "learning_rate": 4.9709900435804336e-05, "loss": 0.9387, "step": 48634 }, { "epoch": 0.05, "learning_rate": 4.9709888504367995e-05, "loss": 1.227, "step": 48635 }, { "epoch": 0.05, "learning_rate": 4.970987657268772e-05, "loss": 1.2886, "step": 48636 }, { "epoch": 0.05, "learning_rate": 4.970986464076353e-05, "loss": 1.094, "step": 48637 }, { "epoch": 0.05, "learning_rate": 4.970985270859541e-05, "loss": 1.1316, "step": 48638 }, { "epoch": 0.05, "learning_rate": 4.970984077618337e-05, "loss": 0.9193, "step": 48639 }, { "epoch": 0.05, "learning_rate": 4.9709828843527396e-05, "loss": 1.2476, "step": 48640 }, { "epoch": 0.05, "learning_rate": 4.9709816910627503e-05, "loss": 1.0932, "step": 48641 }, { "epoch": 0.05, "learning_rate": 4.9709804977483686e-05, "loss": 0.7058, "step": 48642 }, { "epoch": 0.05, "learning_rate": 4.970979304409594e-05, "loss": 0.6703, "step": 48643 }, { "epoch": 0.05, "learning_rate": 4.970978111046427e-05, "loss": 0.7303, "step": 48644 }, { "epoch": 0.05, "learning_rate": 4.9709769176588675e-05, "loss": 1.0653, "step": 48645 }, { "epoch": 0.05, "learning_rate": 4.9709757242469156e-05, "loss": 1.1113, "step": 48646 }, { "epoch": 0.05, "learning_rate": 4.970974530810571e-05, "loss": 1.2638, "step": 48647 }, { "epoch": 0.05, "learning_rate": 4.9709733373498344e-05, "loss": 1.0748, "step": 48648 }, { "epoch": 0.05, "learning_rate": 4.9709721438647064e-05, "loss": 0.7166, "step": 48649 }, { "epoch": 0.05, "learning_rate": 4.9709709503551845e-05, "loss": 1.259, "step": 48650 }, { "epoch": 0.05, "learning_rate": 4.97096975682127e-05, "loss": 1.3125, "step": 48651 }, { "epoch": 0.05, "learning_rate": 4.9709685632629645e-05, "loss": 0.7832, "step": 48652 }, { "epoch": 0.05, "learning_rate": 4.9709673696802664e-05, "loss": 0.8633, "step": 48653 }, { "epoch": 0.05, "learning_rate": 4.970966176073175e-05, "loss": 0.871, "step": 48654 }, { "epoch": 0.05, "learning_rate": 4.970964982441691e-05, "loss": 0.9248, "step": 48655 }, { "epoch": 0.05, "learning_rate": 4.9709637887858164e-05, "loss": 1.0523, "step": 48656 }, { "epoch": 0.05, "learning_rate": 4.970962595105548e-05, "loss": 0.9449, "step": 48657 }, { "epoch": 0.05, "learning_rate": 4.9709614014008875e-05, "loss": 0.8806, "step": 48658 }, { "epoch": 0.05, "learning_rate": 4.970960207671835e-05, "loss": 0.9649, "step": 48659 }, { "epoch": 0.05, "learning_rate": 4.970959013918391e-05, "loss": 0.7673, "step": 48660 }, { "epoch": 0.05, "learning_rate": 4.970957820140554e-05, "loss": 1.0282, "step": 48661 }, { "epoch": 0.05, "learning_rate": 4.9709566263383244e-05, "loss": 0.6295, "step": 48662 }, { "epoch": 0.05, "learning_rate": 4.9709554325117026e-05, "loss": 1.0447, "step": 48663 }, { "epoch": 0.05, "learning_rate": 4.970954238660689e-05, "loss": 1.1168, "step": 48664 }, { "epoch": 0.05, "learning_rate": 4.9709530447852826e-05, "loss": 1.0573, "step": 48665 }, { "epoch": 0.05, "learning_rate": 4.9709518508854845e-05, "loss": 1.2017, "step": 48666 }, { "epoch": 0.05, "learning_rate": 4.970950656961294e-05, "loss": 1.2446, "step": 48667 }, { "epoch": 0.05, "learning_rate": 4.970949463012712e-05, "loss": 1.0422, "step": 48668 }, { "epoch": 0.05, "learning_rate": 4.970948269039737e-05, "loss": 1.1188, "step": 48669 }, { "epoch": 0.05, "learning_rate": 4.9709470750423703e-05, "loss": 1.1279, "step": 48670 }, { "epoch": 0.05, "learning_rate": 4.970945881020611e-05, "loss": 1.0411, "step": 48671 }, { "epoch": 0.05, "learning_rate": 4.9709446869744606e-05, "loss": 0.8527, "step": 48672 }, { "epoch": 0.05, "learning_rate": 4.970943492903917e-05, "loss": 1.1378, "step": 48673 }, { "epoch": 0.05, "learning_rate": 4.9709422988089814e-05, "loss": 1.1491, "step": 48674 }, { "epoch": 0.05, "learning_rate": 4.9709411046896534e-05, "loss": 1.5815, "step": 48675 }, { "epoch": 0.05, "learning_rate": 4.970939910545934e-05, "loss": 1.6658, "step": 48676 }, { "epoch": 0.05, "learning_rate": 4.970938716377822e-05, "loss": 1.2894, "step": 48677 }, { "epoch": 0.05, "learning_rate": 4.970937522185318e-05, "loss": 0.9253, "step": 48678 }, { "epoch": 0.05, "learning_rate": 4.970936327968423e-05, "loss": 1.0955, "step": 48679 }, { "epoch": 0.05, "learning_rate": 4.970935133727135e-05, "loss": 1.1455, "step": 48680 }, { "epoch": 0.05, "learning_rate": 4.970933939461455e-05, "loss": 1.0009, "step": 48681 }, { "epoch": 0.05, "learning_rate": 4.970932745171383e-05, "loss": 1.4958, "step": 48682 }, { "epoch": 0.05, "learning_rate": 4.9709315508569196e-05, "loss": 1.2797, "step": 48683 }, { "epoch": 0.05, "learning_rate": 4.970930356518063e-05, "loss": 1.2096, "step": 48684 }, { "epoch": 0.05, "learning_rate": 4.970929162154815e-05, "loss": 0.948, "step": 48685 }, { "epoch": 0.05, "learning_rate": 4.9709279677671757e-05, "loss": 0.9585, "step": 48686 }, { "epoch": 0.05, "learning_rate": 4.9709267733551435e-05, "loss": 1.1379, "step": 48687 }, { "epoch": 0.05, "learning_rate": 4.97092557891872e-05, "loss": 0.9783, "step": 48688 }, { "epoch": 0.05, "learning_rate": 4.9709243844579045e-05, "loss": 0.9667, "step": 48689 }, { "epoch": 0.05, "learning_rate": 4.970923189972697e-05, "loss": 0.9763, "step": 48690 }, { "epoch": 0.05, "learning_rate": 4.970921995463097e-05, "loss": 0.489, "step": 48691 }, { "epoch": 0.05, "learning_rate": 4.9709208009291054e-05, "loss": 1.238, "step": 48692 }, { "epoch": 0.05, "learning_rate": 4.970919606370722e-05, "loss": 0.956, "step": 48693 }, { "epoch": 0.05, "learning_rate": 4.970918411787947e-05, "loss": 1.1327, "step": 48694 }, { "epoch": 0.05, "learning_rate": 4.97091721718078e-05, "loss": 1.1954, "step": 48695 }, { "epoch": 0.05, "learning_rate": 4.970916022549221e-05, "loss": 0.5737, "step": 48696 }, { "epoch": 0.05, "learning_rate": 4.9709148278932706e-05, "loss": 0.1381, "step": 48697 }, { "epoch": 0.05, "learning_rate": 4.970913633212928e-05, "loss": 0.3937, "step": 48698 }, { "epoch": 0.05, "learning_rate": 4.970912438508193e-05, "loss": 0.7505, "step": 48699 }, { "epoch": 0.05, "learning_rate": 4.9709112437790674e-05, "loss": 0.7276, "step": 48700 }, { "epoch": 0.05, "learning_rate": 4.970910049025549e-05, "loss": 1.1574, "step": 48701 }, { "epoch": 0.05, "learning_rate": 4.970908854247639e-05, "loss": 0.8436, "step": 48702 }, { "epoch": 0.05, "learning_rate": 4.9709076594453376e-05, "loss": 0.6668, "step": 48703 }, { "epoch": 0.05, "learning_rate": 4.970906464618644e-05, "loss": 0.5546, "step": 48704 }, { "epoch": 0.05, "learning_rate": 4.97090526976756e-05, "loss": 0.7311, "step": 48705 }, { "epoch": 0.05, "learning_rate": 4.970904074892082e-05, "loss": 0.8949, "step": 48706 }, { "epoch": 0.05, "learning_rate": 4.9709028799922145e-05, "loss": 0.5243, "step": 48707 }, { "epoch": 0.05, "learning_rate": 4.970901685067954e-05, "loss": 0.4313, "step": 48708 }, { "epoch": 0.05, "learning_rate": 4.970900490119302e-05, "loss": 0.6317, "step": 48709 }, { "epoch": 0.05, "learning_rate": 4.970899295146258e-05, "loss": 0.9806, "step": 48710 }, { "epoch": 0.05, "learning_rate": 4.970898100148823e-05, "loss": 1.0431, "step": 48711 }, { "epoch": 0.05, "learning_rate": 4.970896905126996e-05, "loss": 1.0497, "step": 48712 }, { "epoch": 0.05, "learning_rate": 4.970895710080777e-05, "loss": 0.8684, "step": 48713 }, { "epoch": 0.05, "learning_rate": 4.970894515010167e-05, "loss": 0.4871, "step": 48714 }, { "epoch": 0.05, "learning_rate": 4.9708933199151654e-05, "loss": 1.1874, "step": 48715 }, { "epoch": 0.05, "learning_rate": 4.970892124795771e-05, "loss": 1.2583, "step": 48716 }, { "epoch": 0.05, "learning_rate": 4.970890929651986e-05, "loss": 0.833, "step": 48717 }, { "epoch": 0.05, "learning_rate": 4.97088973448381e-05, "loss": 0.5428, "step": 48718 }, { "epoch": 0.05, "learning_rate": 4.9708885392912414e-05, "loss": 1.0427, "step": 48719 }, { "epoch": 0.05, "learning_rate": 4.9708873440742814e-05, "loss": 1.1316, "step": 48720 }, { "epoch": 0.05, "learning_rate": 4.97088614883293e-05, "loss": 0.6187, "step": 48721 }, { "epoch": 0.05, "learning_rate": 4.970884953567187e-05, "loss": 0.8796, "step": 48722 }, { "epoch": 0.05, "learning_rate": 4.970883758277052e-05, "loss": 0.8132, "step": 48723 }, { "epoch": 0.05, "learning_rate": 4.970882562962527e-05, "loss": 1.3017, "step": 48724 }, { "epoch": 0.05, "learning_rate": 4.970881367623609e-05, "loss": 1.2895, "step": 48725 }, { "epoch": 0.05, "learning_rate": 4.9708801722603e-05, "loss": 0.7943, "step": 48726 }, { "epoch": 0.05, "learning_rate": 4.9708789768725995e-05, "loss": 1.2766, "step": 48727 }, { "epoch": 0.05, "learning_rate": 4.9708777814605075e-05, "loss": 0.8321, "step": 48728 }, { "epoch": 0.05, "learning_rate": 4.9708765860240236e-05, "loss": 1.4865, "step": 48729 }, { "epoch": 0.05, "learning_rate": 4.970875390563149e-05, "loss": 1.3065, "step": 48730 }, { "epoch": 0.05, "learning_rate": 4.970874195077882e-05, "loss": 1.3997, "step": 48731 }, { "epoch": 0.05, "learning_rate": 4.970872999568225e-05, "loss": 1.0683, "step": 48732 }, { "epoch": 0.05, "learning_rate": 4.970871804034175e-05, "loss": 1.2086, "step": 48733 }, { "epoch": 0.05, "learning_rate": 4.970870608475735e-05, "loss": 1.2193, "step": 48734 }, { "epoch": 0.05, "learning_rate": 4.970869412892903e-05, "loss": 1.5248, "step": 48735 }, { "epoch": 0.05, "learning_rate": 4.970868217285679e-05, "loss": 1.1836, "step": 48736 }, { "epoch": 0.05, "learning_rate": 4.9708670216540644e-05, "loss": 1.3239, "step": 48737 }, { "epoch": 0.05, "learning_rate": 4.970865825998059e-05, "loss": 1.0486, "step": 48738 }, { "epoch": 0.05, "learning_rate": 4.970864630317661e-05, "loss": 1.0706, "step": 48739 }, { "epoch": 0.05, "learning_rate": 4.9708634346128725e-05, "loss": 1.3415, "step": 48740 }, { "epoch": 0.05, "learning_rate": 4.9708622388836926e-05, "loss": 0.5004, "step": 48741 }, { "epoch": 0.05, "learning_rate": 4.970861043130121e-05, "loss": 1.0752, "step": 48742 }, { "epoch": 0.05, "learning_rate": 4.970859847352158e-05, "loss": 0.5179, "step": 48743 }, { "epoch": 0.05, "learning_rate": 4.970858651549803e-05, "loss": 0.3393, "step": 48744 }, { "epoch": 0.05, "learning_rate": 4.970857455723059e-05, "loss": 1.1981, "step": 48745 }, { "epoch": 0.05, "learning_rate": 4.9708562598719225e-05, "loss": 1.1075, "step": 48746 }, { "epoch": 0.05, "learning_rate": 4.970855063996394e-05, "loss": 0.6883, "step": 48747 }, { "epoch": 0.05, "learning_rate": 4.970853868096475e-05, "loss": 0.81, "step": 48748 }, { "epoch": 0.05, "learning_rate": 4.970852672172165e-05, "loss": 1.3649, "step": 48749 }, { "epoch": 0.05, "learning_rate": 4.9708514762234634e-05, "loss": 1.1171, "step": 48750 }, { "epoch": 0.05, "learning_rate": 4.9708502802503706e-05, "loss": 0.9786, "step": 48751 }, { "epoch": 0.05, "learning_rate": 4.9708490842528866e-05, "loss": 1.0468, "step": 48752 }, { "epoch": 0.05, "learning_rate": 4.9708478882310114e-05, "loss": 0.9997, "step": 48753 }, { "epoch": 0.05, "learning_rate": 4.970846692184745e-05, "loss": 1.0619, "step": 48754 }, { "epoch": 0.05, "learning_rate": 4.970845496114088e-05, "loss": 0.9817, "step": 48755 }, { "epoch": 0.05, "learning_rate": 4.97084430001904e-05, "loss": 1.0637, "step": 48756 }, { "epoch": 0.05, "learning_rate": 4.970843103899599e-05, "loss": 0.9284, "step": 48757 }, { "epoch": 0.05, "learning_rate": 4.970841907755769e-05, "loss": 0.8128, "step": 48758 }, { "epoch": 0.05, "learning_rate": 4.970840711587547e-05, "loss": 0.8734, "step": 48759 }, { "epoch": 0.05, "learning_rate": 4.9708395153949336e-05, "loss": 1.0599, "step": 48760 }, { "epoch": 0.05, "learning_rate": 4.97083831917793e-05, "loss": 0.5982, "step": 48761 }, { "epoch": 0.05, "learning_rate": 4.970837122936535e-05, "loss": 0.8138, "step": 48762 }, { "epoch": 0.05, "learning_rate": 4.970835926670748e-05, "loss": 0.9829, "step": 48763 }, { "epoch": 0.05, "learning_rate": 4.9708347303805715e-05, "loss": 1.115, "step": 48764 }, { "epoch": 0.05, "learning_rate": 4.970833534066003e-05, "loss": 1.0961, "step": 48765 }, { "epoch": 0.05, "learning_rate": 4.9708323377270436e-05, "loss": 1.0431, "step": 48766 }, { "epoch": 0.05, "learning_rate": 4.9708311413636935e-05, "loss": 1.1264, "step": 48767 }, { "epoch": 0.05, "learning_rate": 4.970829944975952e-05, "loss": 0.6619, "step": 48768 }, { "epoch": 0.05, "learning_rate": 4.970828748563819e-05, "loss": 0.8765, "step": 48769 }, { "epoch": 0.05, "learning_rate": 4.9708275521272964e-05, "loss": 1.4363, "step": 48770 }, { "epoch": 0.05, "learning_rate": 4.970826355666382e-05, "loss": 1.4191, "step": 48771 }, { "epoch": 0.05, "learning_rate": 4.970825159181076e-05, "loss": 1.1133, "step": 48772 }, { "epoch": 0.05, "learning_rate": 4.97082396267138e-05, "loss": 1.12, "step": 48773 }, { "epoch": 0.05, "learning_rate": 4.970822766137293e-05, "loss": 1.4987, "step": 48774 }, { "epoch": 0.05, "learning_rate": 4.9708215695788154e-05, "loss": 1.0512, "step": 48775 }, { "epoch": 0.05, "learning_rate": 4.9708203729959456e-05, "loss": 0.817, "step": 48776 }, { "epoch": 0.05, "learning_rate": 4.970819176388686e-05, "loss": 1.2466, "step": 48777 }, { "epoch": 0.05, "learning_rate": 4.970817979757035e-05, "loss": 1.1203, "step": 48778 }, { "epoch": 0.05, "learning_rate": 4.970816783100993e-05, "loss": 0.566, "step": 48779 }, { "epoch": 0.05, "learning_rate": 4.970815586420561e-05, "loss": 0.8466, "step": 48780 }, { "epoch": 0.05, "learning_rate": 4.970814389715738e-05, "loss": 1.1126, "step": 48781 }, { "epoch": 0.05, "learning_rate": 4.970813192986524e-05, "loss": 1.3535, "step": 48782 }, { "epoch": 0.05, "learning_rate": 4.9708119962329186e-05, "loss": 1.221, "step": 48783 }, { "epoch": 0.05, "learning_rate": 4.970810799454922e-05, "loss": 0.9901, "step": 48784 }, { "epoch": 0.05, "learning_rate": 4.970809602652537e-05, "loss": 1.0588, "step": 48785 }, { "epoch": 0.05, "learning_rate": 4.970808405825759e-05, "loss": 0.6668, "step": 48786 }, { "epoch": 0.05, "learning_rate": 4.970807208974591e-05, "loss": 0.9854, "step": 48787 }, { "epoch": 0.05, "learning_rate": 4.970806012099032e-05, "loss": 1.1171, "step": 48788 }, { "epoch": 0.05, "learning_rate": 4.9708048151990817e-05, "loss": 1.3048, "step": 48789 }, { "epoch": 0.05, "learning_rate": 4.970803618274742e-05, "loss": 1.1632, "step": 48790 }, { "epoch": 0.05, "learning_rate": 4.97080242132601e-05, "loss": 0.8332, "step": 48791 }, { "epoch": 0.05, "learning_rate": 4.9708012243528876e-05, "loss": 0.9156, "step": 48792 }, { "epoch": 0.05, "learning_rate": 4.9708000273553755e-05, "loss": 1.4457, "step": 48793 }, { "epoch": 0.05, "learning_rate": 4.9707988303334716e-05, "loss": 1.0472, "step": 48794 }, { "epoch": 0.05, "learning_rate": 4.970797633287178e-05, "loss": 1.1135, "step": 48795 }, { "epoch": 0.05, "learning_rate": 4.9707964362164936e-05, "loss": 0.9879, "step": 48796 }, { "epoch": 0.05, "learning_rate": 4.970795239121418e-05, "loss": 1.0649, "step": 48797 }, { "epoch": 0.05, "learning_rate": 4.9707940420019523e-05, "loss": 1.1258, "step": 48798 }, { "epoch": 0.05, "learning_rate": 4.970792844858095e-05, "loss": 0.8285, "step": 48799 }, { "epoch": 0.05, "learning_rate": 4.9707916476898485e-05, "loss": 1.0021, "step": 48800 }, { "epoch": 0.05, "learning_rate": 4.9707904504972105e-05, "loss": 1.158, "step": 48801 }, { "epoch": 0.05, "learning_rate": 4.970789253280182e-05, "loss": 1.0914, "step": 48802 }, { "epoch": 0.05, "learning_rate": 4.970788056038763e-05, "loss": 0.9984, "step": 48803 }, { "epoch": 0.05, "learning_rate": 4.9707868587729536e-05, "loss": 1.186, "step": 48804 }, { "epoch": 0.05, "learning_rate": 4.970785661482753e-05, "loss": 1.0367, "step": 48805 }, { "epoch": 0.05, "learning_rate": 4.970784464168162e-05, "loss": 1.3083, "step": 48806 }, { "epoch": 0.05, "learning_rate": 4.970783266829181e-05, "loss": 1.0801, "step": 48807 }, { "epoch": 0.05, "learning_rate": 4.970782069465809e-05, "loss": 0.8859, "step": 48808 }, { "epoch": 0.05, "learning_rate": 4.970780872078047e-05, "loss": 1.4637, "step": 48809 }, { "epoch": 0.05, "learning_rate": 4.970779674665894e-05, "loss": 0.9626, "step": 48810 }, { "epoch": 0.05, "learning_rate": 4.9707784772293506e-05, "loss": 1.6242, "step": 48811 }, { "epoch": 0.05, "learning_rate": 4.9707772797684166e-05, "loss": 1.1582, "step": 48812 }, { "epoch": 0.05, "learning_rate": 4.970776082283092e-05, "loss": 0.7457, "step": 48813 }, { "epoch": 0.05, "learning_rate": 4.970774884773377e-05, "loss": 0.7334, "step": 48814 }, { "epoch": 0.05, "learning_rate": 4.970773687239272e-05, "loss": 0.8774, "step": 48815 }, { "epoch": 0.05, "learning_rate": 4.970772489680776e-05, "loss": 0.6984, "step": 48816 }, { "epoch": 0.05, "learning_rate": 4.970771292097891e-05, "loss": 1.2685, "step": 48817 }, { "epoch": 0.05, "learning_rate": 4.970770094490614e-05, "loss": 0.7042, "step": 48818 }, { "epoch": 0.05, "learning_rate": 4.970768896858947e-05, "loss": 0.6971, "step": 48819 }, { "epoch": 0.05, "learning_rate": 4.9707676992028894e-05, "loss": 1.3597, "step": 48820 }, { "epoch": 0.05, "learning_rate": 4.970766501522442e-05, "loss": 1.0832, "step": 48821 }, { "epoch": 0.05, "learning_rate": 4.9707653038176036e-05, "loss": 0.8112, "step": 48822 }, { "epoch": 0.05, "learning_rate": 4.9707641060883756e-05, "loss": 1.2639, "step": 48823 }, { "epoch": 0.05, "learning_rate": 4.9707629083347565e-05, "loss": 0.9796, "step": 48824 }, { "epoch": 0.05, "learning_rate": 4.9707617105567476e-05, "loss": 0.9362, "step": 48825 }, { "epoch": 0.05, "learning_rate": 4.970760512754348e-05, "loss": 0.8757, "step": 48826 }, { "epoch": 0.05, "learning_rate": 4.970759314927559e-05, "loss": 0.9655, "step": 48827 }, { "epoch": 0.05, "learning_rate": 4.9707581170763787e-05, "loss": 0.8697, "step": 48828 }, { "epoch": 0.05, "learning_rate": 4.9707569192008085e-05, "loss": 1.475, "step": 48829 }, { "epoch": 0.05, "learning_rate": 4.9707557213008485e-05, "loss": 0.844, "step": 48830 }, { "epoch": 0.05, "learning_rate": 4.9707545233764974e-05, "loss": 0.9929, "step": 48831 }, { "epoch": 0.05, "learning_rate": 4.970753325427756e-05, "loss": 0.7935, "step": 48832 }, { "epoch": 0.05, "learning_rate": 4.970752127454625e-05, "loss": 1.0025, "step": 48833 }, { "epoch": 0.05, "learning_rate": 4.970750929457103e-05, "loss": 1.1664, "step": 48834 }, { "epoch": 0.05, "learning_rate": 4.9707497314351914e-05, "loss": 1.6204, "step": 48835 }, { "epoch": 0.05, "learning_rate": 4.970748533388889e-05, "loss": 1.2172, "step": 48836 }, { "epoch": 0.05, "learning_rate": 4.970747335318198e-05, "loss": 0.9492, "step": 48837 }, { "epoch": 0.05, "learning_rate": 4.9707461372231154e-05, "loss": 0.6088, "step": 48838 }, { "epoch": 0.05, "learning_rate": 4.970744939103643e-05, "loss": 0.7855, "step": 48839 }, { "epoch": 0.05, "learning_rate": 4.9707437409597804e-05, "loss": 0.816, "step": 48840 }, { "epoch": 0.05, "learning_rate": 4.970742542791528e-05, "loss": 0.8971, "step": 48841 }, { "epoch": 0.05, "learning_rate": 4.970741344598885e-05, "loss": 1.0335, "step": 48842 }, { "epoch": 0.05, "learning_rate": 4.970740146381851e-05, "loss": 1.0359, "step": 48843 }, { "epoch": 0.05, "learning_rate": 4.9707389481404285e-05, "loss": 1.3481, "step": 48844 }, { "epoch": 0.05, "learning_rate": 4.9707377498746154e-05, "loss": 1.235, "step": 48845 }, { "epoch": 0.05, "learning_rate": 4.970736551584412e-05, "loss": 0.8103, "step": 48846 }, { "epoch": 0.05, "learning_rate": 4.970735353269819e-05, "loss": 1.2373, "step": 48847 }, { "epoch": 0.05, "learning_rate": 4.970734154930835e-05, "loss": 1.4778, "step": 48848 }, { "epoch": 0.05, "learning_rate": 4.970732956567462e-05, "loss": 0.859, "step": 48849 }, { "epoch": 0.05, "learning_rate": 4.9707317581796985e-05, "loss": 0.9391, "step": 48850 }, { "epoch": 0.05, "learning_rate": 4.9707305597675444e-05, "loss": 0.5783, "step": 48851 }, { "epoch": 0.05, "learning_rate": 4.970729361331001e-05, "loss": 0.6488, "step": 48852 }, { "epoch": 0.05, "learning_rate": 4.970728162870068e-05, "loss": 0.7163, "step": 48853 }, { "epoch": 0.05, "learning_rate": 4.970726964384744e-05, "loss": 0.6772, "step": 48854 }, { "epoch": 0.05, "learning_rate": 4.97072576587503e-05, "loss": 0.8496, "step": 48855 }, { "epoch": 0.05, "learning_rate": 4.970724567340927e-05, "loss": 1.1143, "step": 48856 }, { "epoch": 0.05, "learning_rate": 4.970723368782434e-05, "loss": 0.874, "step": 48857 }, { "epoch": 0.05, "learning_rate": 4.9707221701995504e-05, "loss": 0.8586, "step": 48858 }, { "epoch": 0.05, "learning_rate": 4.970720971592277e-05, "loss": 1.1579, "step": 48859 }, { "epoch": 0.05, "learning_rate": 4.970719772960614e-05, "loss": 1.1395, "step": 48860 }, { "epoch": 0.05, "learning_rate": 4.9707185743045616e-05, "loss": 0.9035, "step": 48861 }, { "epoch": 0.05, "learning_rate": 4.9707173756241177e-05, "loss": 0.8315, "step": 48862 }, { "epoch": 0.05, "learning_rate": 4.970716176919285e-05, "loss": 0.9866, "step": 48863 }, { "epoch": 0.05, "learning_rate": 4.9707149781900624e-05, "loss": 1.2194, "step": 48864 }, { "epoch": 0.05, "learning_rate": 4.97071377943645e-05, "loss": 0.8548, "step": 48865 }, { "epoch": 0.05, "learning_rate": 4.970712580658447e-05, "loss": 1.0399, "step": 48866 }, { "epoch": 0.05, "learning_rate": 4.970711381856055e-05, "loss": 0.8203, "step": 48867 }, { "epoch": 0.05, "learning_rate": 4.970710183029274e-05, "loss": 0.906, "step": 48868 }, { "epoch": 0.05, "learning_rate": 4.970708984178101e-05, "loss": 1.022, "step": 48869 }, { "epoch": 0.05, "learning_rate": 4.9707077853025395e-05, "loss": 0.8754, "step": 48870 }, { "epoch": 0.05, "learning_rate": 4.970706586402588e-05, "loss": 1.0018, "step": 48871 }, { "epoch": 0.05, "learning_rate": 4.9707053874782474e-05, "loss": 0.9014, "step": 48872 }, { "epoch": 0.05, "learning_rate": 4.9707041885295156e-05, "loss": 3.5718, "step": 48873 }, { "epoch": 0.05, "learning_rate": 4.970702989556395e-05, "loss": 1.6903, "step": 48874 }, { "epoch": 0.05, "learning_rate": 4.970701790558885e-05, "loss": 1.1004, "step": 48875 }, { "epoch": 0.05, "learning_rate": 4.970700591536984e-05, "loss": 0.8495, "step": 48876 }, { "epoch": 0.05, "learning_rate": 4.970699392490694e-05, "loss": 1.0099, "step": 48877 }, { "epoch": 0.05, "learning_rate": 4.970698193420015e-05, "loss": 1.324, "step": 48878 }, { "epoch": 0.05, "learning_rate": 4.970696994324945e-05, "loss": 0.99, "step": 48879 }, { "epoch": 0.05, "learning_rate": 4.970695795205487e-05, "loss": 1.4168, "step": 48880 }, { "epoch": 0.05, "learning_rate": 4.970694596061638e-05, "loss": 1.1406, "step": 48881 }, { "epoch": 0.05, "learning_rate": 4.970693396893399e-05, "loss": 1.4104, "step": 48882 }, { "epoch": 0.05, "learning_rate": 4.9706921977007714e-05, "loss": 2.2266, "step": 48883 }, { "epoch": 0.05, "learning_rate": 4.970690998483754e-05, "loss": 0.9847, "step": 48884 }, { "epoch": 0.05, "learning_rate": 4.9706897992423464e-05, "loss": 1.1163, "step": 48885 }, { "epoch": 0.05, "learning_rate": 4.97068859997655e-05, "loss": 1.1095, "step": 48886 }, { "epoch": 0.05, "learning_rate": 4.9706874006863636e-05, "loss": 0.4774, "step": 48887 }, { "epoch": 0.05, "learning_rate": 4.970686201371787e-05, "loss": 0.9867, "step": 48888 }, { "epoch": 0.05, "learning_rate": 4.9706850020328215e-05, "loss": 1.073, "step": 48889 }, { "epoch": 0.05, "learning_rate": 4.9706838026694664e-05, "loss": 1.064, "step": 48890 }, { "epoch": 0.05, "learning_rate": 4.9706826032817215e-05, "loss": 1.0941, "step": 48891 }, { "epoch": 0.05, "learning_rate": 4.970681403869587e-05, "loss": 1.0615, "step": 48892 }, { "epoch": 0.05, "learning_rate": 4.970680204433064e-05, "loss": 1.3993, "step": 48893 }, { "epoch": 0.05, "learning_rate": 4.97067900497215e-05, "loss": 1.0149, "step": 48894 }, { "epoch": 0.05, "learning_rate": 4.970677805486847e-05, "loss": 0.9792, "step": 48895 }, { "epoch": 0.05, "learning_rate": 4.9706766059771554e-05, "loss": 0.8642, "step": 48896 }, { "epoch": 0.05, "learning_rate": 4.970675406443074e-05, "loss": 1.4902, "step": 48897 }, { "epoch": 0.05, "learning_rate": 4.970674206884602e-05, "loss": 0.9006, "step": 48898 }, { "epoch": 0.05, "learning_rate": 4.9706730073017416e-05, "loss": 0.4831, "step": 48899 }, { "epoch": 0.05, "learning_rate": 4.970671807694491e-05, "loss": 0.2548, "step": 48900 }, { "epoch": 0.05, "learning_rate": 4.970670608062852e-05, "loss": 0.7109, "step": 48901 }, { "epoch": 0.05, "learning_rate": 4.970669408406823e-05, "loss": 1.0759, "step": 48902 }, { "epoch": 0.05, "learning_rate": 4.9706682087264045e-05, "loss": 1.8798, "step": 48903 }, { "epoch": 0.05, "learning_rate": 4.970667009021597e-05, "loss": 1.3365, "step": 48904 }, { "epoch": 0.05, "learning_rate": 4.9706658092923995e-05, "loss": 1.3408, "step": 48905 }, { "epoch": 0.05, "learning_rate": 4.970664609538813e-05, "loss": 1.0033, "step": 48906 }, { "epoch": 0.05, "learning_rate": 4.970663409760837e-05, "loss": 0.8405, "step": 48907 }, { "epoch": 0.05, "learning_rate": 4.970662209958472e-05, "loss": 0.8386, "step": 48908 }, { "epoch": 0.05, "learning_rate": 4.970661010131717e-05, "loss": 1.0547, "step": 48909 }, { "epoch": 0.05, "learning_rate": 4.9706598102805735e-05, "loss": 0.8196, "step": 48910 }, { "epoch": 0.05, "learning_rate": 4.97065861040504e-05, "loss": 1.0369, "step": 48911 }, { "epoch": 0.05, "learning_rate": 4.970657410505117e-05, "loss": 1.2297, "step": 48912 }, { "epoch": 0.05, "learning_rate": 4.9706562105808055e-05, "loss": 0.8946, "step": 48913 }, { "epoch": 0.05, "learning_rate": 4.9706550106321045e-05, "loss": 0.7122, "step": 48914 }, { "epoch": 0.05, "learning_rate": 4.9706538106590145e-05, "loss": 0.7324, "step": 48915 }, { "epoch": 0.05, "learning_rate": 4.9706526106615346e-05, "loss": 1.0593, "step": 48916 }, { "epoch": 0.05, "learning_rate": 4.970651410639666e-05, "loss": 1.4309, "step": 48917 }, { "epoch": 0.05, "learning_rate": 4.970650210593408e-05, "loss": 1.1627, "step": 48918 }, { "epoch": 0.05, "learning_rate": 4.970649010522761e-05, "loss": 0.503, "step": 48919 }, { "epoch": 0.05, "learning_rate": 4.9706478104277246e-05, "loss": 0.7023, "step": 48920 }, { "epoch": 0.05, "learning_rate": 4.9706466103082984e-05, "loss": 0.9769, "step": 48921 }, { "epoch": 0.05, "learning_rate": 4.970645410164484e-05, "loss": 0.8109, "step": 48922 }, { "epoch": 0.05, "learning_rate": 4.9706442099962794e-05, "loss": 0.8526, "step": 48923 }, { "epoch": 0.05, "learning_rate": 4.970643009803686e-05, "loss": 0.613, "step": 48924 }, { "epoch": 0.05, "learning_rate": 4.970641809586704e-05, "loss": 0.999, "step": 48925 }, { "epoch": 0.05, "learning_rate": 4.970640609345333e-05, "loss": 1.5394, "step": 48926 }, { "epoch": 0.05, "learning_rate": 4.970639409079572e-05, "loss": 1.5779, "step": 48927 }, { "epoch": 0.05, "learning_rate": 4.9706382087894223e-05, "loss": 1.1776, "step": 48928 }, { "epoch": 0.05, "learning_rate": 4.970637008474883e-05, "loss": 1.3234, "step": 48929 }, { "epoch": 0.05, "learning_rate": 4.9706358081359555e-05, "loss": 1.0539, "step": 48930 }, { "epoch": 0.05, "learning_rate": 4.970634607772639e-05, "loss": 1.0945, "step": 48931 }, { "epoch": 0.05, "learning_rate": 4.970633407384933e-05, "loss": 1.0724, "step": 48932 }, { "epoch": 0.05, "learning_rate": 4.970632206972837e-05, "loss": 0.4494, "step": 48933 }, { "epoch": 0.05, "learning_rate": 4.970631006536353e-05, "loss": 0.09, "step": 48934 }, { "epoch": 0.05, "learning_rate": 4.97062980607548e-05, "loss": 0.0514, "step": 48935 }, { "epoch": 0.05, "learning_rate": 4.970628605590218e-05, "loss": 0.3605, "step": 48936 }, { "epoch": 0.05, "learning_rate": 4.970627405080567e-05, "loss": 1.3806, "step": 48937 }, { "epoch": 0.05, "learning_rate": 4.970626204546527e-05, "loss": 1.1691, "step": 48938 }, { "epoch": 0.05, "learning_rate": 4.970625003988097e-05, "loss": 0.7802, "step": 48939 }, { "epoch": 0.05, "learning_rate": 4.970623803405279e-05, "loss": 0.8985, "step": 48940 }, { "epoch": 0.05, "learning_rate": 4.970622602798072e-05, "loss": 0.4886, "step": 48941 }, { "epoch": 0.05, "learning_rate": 4.970621402166476e-05, "loss": 1.2462, "step": 48942 }, { "epoch": 0.05, "learning_rate": 4.970620201510491e-05, "loss": 1.3443, "step": 48943 }, { "epoch": 0.05, "learning_rate": 4.9706190008301175e-05, "loss": 0.9377, "step": 48944 }, { "epoch": 0.05, "learning_rate": 4.970617800125354e-05, "loss": 0.8466, "step": 48945 }, { "epoch": 0.05, "learning_rate": 4.9706165993962026e-05, "loss": 0.8978, "step": 48946 }, { "epoch": 0.05, "learning_rate": 4.970615398642662e-05, "loss": 0.903, "step": 48947 }, { "epoch": 0.05, "learning_rate": 4.9706141978647326e-05, "loss": 1.0849, "step": 48948 }, { "epoch": 0.05, "learning_rate": 4.970612997062414e-05, "loss": 1.0853, "step": 48949 }, { "epoch": 0.05, "learning_rate": 4.9706117962357074e-05, "loss": 1.1075, "step": 48950 }, { "epoch": 0.05, "learning_rate": 4.9706105953846115e-05, "loss": 1.2274, "step": 48951 }, { "epoch": 0.05, "learning_rate": 4.9706093945091264e-05, "loss": 0.9885, "step": 48952 }, { "epoch": 0.05, "learning_rate": 4.970608193609253e-05, "loss": 0.9868, "step": 48953 }, { "epoch": 0.05, "learning_rate": 4.97060699268499e-05, "loss": 0.9946, "step": 48954 }, { "epoch": 0.05, "learning_rate": 4.970605791736339e-05, "loss": 1.0814, "step": 48955 }, { "epoch": 0.05, "learning_rate": 4.970604590763299e-05, "loss": 1.2041, "step": 48956 }, { "epoch": 0.05, "learning_rate": 4.97060338976587e-05, "loss": 1.1891, "step": 48957 }, { "epoch": 0.05, "learning_rate": 4.9706021887440524e-05, "loss": 1.1241, "step": 48958 }, { "epoch": 0.05, "learning_rate": 4.970600987697847e-05, "loss": 1.0261, "step": 48959 }, { "epoch": 0.05, "learning_rate": 4.9705997866272515e-05, "loss": 1.1307, "step": 48960 }, { "epoch": 0.05, "learning_rate": 4.9705985855322676e-05, "loss": 0.9569, "step": 48961 }, { "epoch": 0.05, "learning_rate": 4.970597384412895e-05, "loss": 1.2506, "step": 48962 }, { "epoch": 0.05, "learning_rate": 4.970596183269134e-05, "loss": 1.0214, "step": 48963 }, { "epoch": 0.05, "learning_rate": 4.970594982100985e-05, "loss": 0.9108, "step": 48964 }, { "epoch": 0.05, "learning_rate": 4.970593780908446e-05, "loss": 0.875, "step": 48965 }, { "epoch": 0.05, "learning_rate": 4.970592579691519e-05, "loss": 0.8331, "step": 48966 }, { "epoch": 0.05, "learning_rate": 4.970591378450203e-05, "loss": 1.1666, "step": 48967 }, { "epoch": 0.05, "learning_rate": 4.970590177184499e-05, "loss": 0.9976, "step": 48968 }, { "epoch": 0.05, "learning_rate": 4.970588975894405e-05, "loss": 1.1223, "step": 48969 }, { "epoch": 0.05, "learning_rate": 4.9705877745799234e-05, "loss": 1.4969, "step": 48970 }, { "epoch": 0.05, "learning_rate": 4.970586573241054e-05, "loss": 1.3035, "step": 48971 }, { "epoch": 0.05, "learning_rate": 4.970585371877794e-05, "loss": 1.3049, "step": 48972 }, { "epoch": 0.05, "learning_rate": 4.970584170490147e-05, "loss": 1.0364, "step": 48973 }, { "epoch": 0.05, "learning_rate": 4.970582969078111e-05, "loss": 0.9351, "step": 48974 }, { "epoch": 0.05, "learning_rate": 4.9705817676416863e-05, "loss": 1.4206, "step": 48975 }, { "epoch": 0.05, "learning_rate": 4.970580566180874e-05, "loss": 1.0781, "step": 48976 }, { "epoch": 0.05, "learning_rate": 4.970579364695672e-05, "loss": 0.8913, "step": 48977 }, { "epoch": 0.05, "learning_rate": 4.970578163186082e-05, "loss": 0.9942, "step": 48978 }, { "epoch": 0.05, "learning_rate": 4.9705769616521024e-05, "loss": 1.0861, "step": 48979 }, { "epoch": 0.05, "learning_rate": 4.9705757600937353e-05, "loss": 1.2575, "step": 48980 }, { "epoch": 0.05, "learning_rate": 4.97057455851098e-05, "loss": 0.9042, "step": 48981 }, { "epoch": 0.05, "learning_rate": 4.970573356903836e-05, "loss": 0.8147, "step": 48982 }, { "epoch": 0.05, "learning_rate": 4.9705721552723033e-05, "loss": 1.1221, "step": 48983 }, { "epoch": 0.05, "learning_rate": 4.9705709536163824e-05, "loss": 1.2505, "step": 48984 }, { "epoch": 0.05, "learning_rate": 4.970569751936073e-05, "loss": 0.6904, "step": 48985 }, { "epoch": 0.05, "learning_rate": 4.9705685502313746e-05, "loss": 0.9995, "step": 48986 }, { "epoch": 0.05, "learning_rate": 4.9705673485022884e-05, "loss": 1.0201, "step": 48987 }, { "epoch": 0.05, "learning_rate": 4.9705661467488143e-05, "loss": 1.3103, "step": 48988 }, { "epoch": 0.05, "learning_rate": 4.970564944970951e-05, "loss": 1.3457, "step": 48989 }, { "epoch": 0.05, "learning_rate": 4.970563743168699e-05, "loss": 1.1403, "step": 48990 }, { "epoch": 0.05, "learning_rate": 4.9705625413420595e-05, "loss": 1.1423, "step": 48991 }, { "epoch": 0.05, "learning_rate": 4.970561339491031e-05, "loss": 1.1039, "step": 48992 }, { "epoch": 0.05, "learning_rate": 4.9705601376156154e-05, "loss": 1.0456, "step": 48993 }, { "epoch": 0.05, "learning_rate": 4.97055893571581e-05, "loss": 1.3295, "step": 48994 }, { "epoch": 0.05, "learning_rate": 4.970557733791616e-05, "loss": 1.2445, "step": 48995 }, { "epoch": 0.05, "learning_rate": 4.970556531843036e-05, "loss": 0.9055, "step": 48996 }, { "epoch": 0.05, "learning_rate": 4.970555329870066e-05, "loss": 1.0812, "step": 48997 }, { "epoch": 0.05, "learning_rate": 4.970554127872707e-05, "loss": 0.9255, "step": 48998 }, { "epoch": 0.05, "learning_rate": 4.970552925850962e-05, "loss": 1.116, "step": 48999 }, { "epoch": 0.05, "learning_rate": 4.970551723804827e-05, "loss": 1.0922, "step": 49000 }, { "epoch": 0.05, "eval_loss": 1.0718774795532227, "eval_runtime": 125.6122, "eval_samples_per_second": 11.026, "eval_steps_per_second": 5.517, "step": 49000 }, { "epoch": 0.05, "learning_rate": 4.970550521734304e-05, "loss": 0.9225, "step": 49001 }, { "epoch": 0.05, "learning_rate": 4.970549319639393e-05, "loss": 1.263, "step": 49002 }, { "epoch": 0.05, "learning_rate": 4.9705481175200934e-05, "loss": 1.0193, "step": 49003 }, { "epoch": 0.05, "learning_rate": 4.970546915376406e-05, "loss": 0.9589, "step": 49004 }, { "epoch": 0.05, "learning_rate": 4.9705457132083305e-05, "loss": 0.9834, "step": 49005 }, { "epoch": 0.05, "learning_rate": 4.9705445110158664e-05, "loss": 0.8971, "step": 49006 }, { "epoch": 0.05, "learning_rate": 4.9705433087990146e-05, "loss": 0.7668, "step": 49007 }, { "epoch": 0.05, "learning_rate": 4.970542106557774e-05, "loss": 0.9955, "step": 49008 }, { "epoch": 0.05, "learning_rate": 4.9705409042921455e-05, "loss": 0.9534, "step": 49009 }, { "epoch": 0.05, "learning_rate": 4.9705397020021296e-05, "loss": 1.1577, "step": 49010 }, { "epoch": 0.05, "learning_rate": 4.9705384996877246e-05, "loss": 1.0056, "step": 49011 }, { "epoch": 0.05, "learning_rate": 4.9705372973489325e-05, "loss": 0.9174, "step": 49012 }, { "epoch": 0.05, "learning_rate": 4.970536094985751e-05, "loss": 1.0104, "step": 49013 }, { "epoch": 0.05, "learning_rate": 4.970534892598182e-05, "loss": 1.0427, "step": 49014 }, { "epoch": 0.05, "learning_rate": 4.9705336901862256e-05, "loss": 0.941, "step": 49015 }, { "epoch": 0.05, "learning_rate": 4.9705324877498804e-05, "loss": 0.7982, "step": 49016 }, { "epoch": 0.05, "learning_rate": 4.9705312852891474e-05, "loss": 0.9185, "step": 49017 }, { "epoch": 0.05, "learning_rate": 4.970530082804026e-05, "loss": 1.2151, "step": 49018 }, { "epoch": 0.05, "learning_rate": 4.9705288802945174e-05, "loss": 1.0527, "step": 49019 }, { "epoch": 0.05, "learning_rate": 4.97052767776062e-05, "loss": 0.8504, "step": 49020 }, { "epoch": 0.05, "learning_rate": 4.970526475202334e-05, "loss": 0.9415, "step": 49021 }, { "epoch": 0.05, "learning_rate": 4.970525272619661e-05, "loss": 0.9535, "step": 49022 }, { "epoch": 0.05, "learning_rate": 4.9705240700126e-05, "loss": 0.8174, "step": 49023 }, { "epoch": 0.05, "learning_rate": 4.970522867381151e-05, "loss": 0.7424, "step": 49024 }, { "epoch": 0.05, "learning_rate": 4.970521664725314e-05, "loss": 1.2314, "step": 49025 }, { "epoch": 0.05, "learning_rate": 4.9705204620450886e-05, "loss": 1.1007, "step": 49026 }, { "epoch": 0.05, "learning_rate": 4.970519259340475e-05, "loss": 0.9628, "step": 49027 }, { "epoch": 0.05, "learning_rate": 4.9705180566114746e-05, "loss": 1.8821, "step": 49028 }, { "epoch": 0.05, "learning_rate": 4.970516853858086e-05, "loss": 0.9577, "step": 49029 }, { "epoch": 0.05, "learning_rate": 4.970515651080308e-05, "loss": 0.6846, "step": 49030 }, { "epoch": 0.05, "learning_rate": 4.970514448278144e-05, "loss": 1.1478, "step": 49031 }, { "epoch": 0.05, "learning_rate": 4.9705132454515915e-05, "loss": 1.1534, "step": 49032 }, { "epoch": 0.05, "learning_rate": 4.9705120426006514e-05, "loss": 0.8412, "step": 49033 }, { "epoch": 0.05, "learning_rate": 4.970510839725323e-05, "loss": 1.2867, "step": 49034 }, { "epoch": 0.05, "learning_rate": 4.9705096368256066e-05, "loss": 1.1493, "step": 49035 }, { "epoch": 0.05, "learning_rate": 4.9705084339015026e-05, "loss": 1.289, "step": 49036 }, { "epoch": 0.05, "learning_rate": 4.970507230953011e-05, "loss": 1.2544, "step": 49037 }, { "epoch": 0.05, "learning_rate": 4.970506027980132e-05, "loss": 0.995, "step": 49038 }, { "epoch": 0.05, "learning_rate": 4.9705048249828645e-05, "loss": 1.2466, "step": 49039 }, { "epoch": 0.05, "learning_rate": 4.970503621961209e-05, "loss": 0.7899, "step": 49040 }, { "epoch": 0.05, "learning_rate": 4.9705024189151664e-05, "loss": 1.3663, "step": 49041 }, { "epoch": 0.05, "learning_rate": 4.970501215844736e-05, "loss": 0.7427, "step": 49042 }, { "epoch": 0.05, "learning_rate": 4.9705000127499166e-05, "loss": 0.9482, "step": 49043 }, { "epoch": 0.05, "learning_rate": 4.970498809630712e-05, "loss": 0.7633, "step": 49044 }, { "epoch": 0.05, "learning_rate": 4.970497606487117e-05, "loss": 0.915, "step": 49045 }, { "epoch": 0.05, "learning_rate": 4.970496403319136e-05, "loss": 0.8745, "step": 49046 }, { "epoch": 0.05, "learning_rate": 4.9704952001267664e-05, "loss": 1.1871, "step": 49047 }, { "epoch": 0.05, "learning_rate": 4.97049399691001e-05, "loss": 0.847, "step": 49048 }, { "epoch": 0.05, "learning_rate": 4.970492793668865e-05, "loss": 0.9823, "step": 49049 }, { "epoch": 0.05, "learning_rate": 4.9704915904033325e-05, "loss": 0.676, "step": 49050 }, { "epoch": 0.05, "learning_rate": 4.9704903871134125e-05, "loss": 1.0746, "step": 49051 }, { "epoch": 0.05, "learning_rate": 4.9704891837991055e-05, "loss": 1.0121, "step": 49052 }, { "epoch": 0.05, "learning_rate": 4.97048798046041e-05, "loss": 0.944, "step": 49053 }, { "epoch": 0.05, "learning_rate": 4.970486777097327e-05, "loss": 1.3412, "step": 49054 }, { "epoch": 0.05, "learning_rate": 4.970485573709856e-05, "loss": 0.8623, "step": 49055 }, { "epoch": 0.05, "learning_rate": 4.970484370297999e-05, "loss": 0.874, "step": 49056 }, { "epoch": 0.05, "learning_rate": 4.970483166861752e-05, "loss": 0.899, "step": 49057 }, { "epoch": 0.05, "learning_rate": 4.97048196340112e-05, "loss": 1.224, "step": 49058 }, { "epoch": 0.05, "learning_rate": 4.9704807599160984e-05, "loss": 0.5762, "step": 49059 }, { "epoch": 0.05, "learning_rate": 4.9704795564066906e-05, "loss": 0.6094, "step": 49060 }, { "epoch": 0.05, "learning_rate": 4.970478352872895e-05, "loss": 0.5228, "step": 49061 }, { "epoch": 0.05, "learning_rate": 4.9704771493147115e-05, "loss": 0.696, "step": 49062 }, { "epoch": 0.05, "learning_rate": 4.97047594573214e-05, "loss": 0.6848, "step": 49063 }, { "epoch": 0.05, "learning_rate": 4.970474742125182e-05, "loss": 0.4994, "step": 49064 }, { "epoch": 0.05, "learning_rate": 4.970473538493836e-05, "loss": 0.5694, "step": 49065 }, { "epoch": 0.05, "learning_rate": 4.9704723348381034e-05, "loss": 0.5883, "step": 49066 }, { "epoch": 0.05, "learning_rate": 4.970471131157982e-05, "loss": 0.6094, "step": 49067 }, { "epoch": 0.05, "learning_rate": 4.970469927453474e-05, "loss": 0.501, "step": 49068 }, { "epoch": 0.05, "learning_rate": 4.9704687237245787e-05, "loss": 0.6425, "step": 49069 }, { "epoch": 0.05, "learning_rate": 4.970467519971295e-05, "loss": 0.699, "step": 49070 }, { "epoch": 0.05, "learning_rate": 4.970466316193625e-05, "loss": 0.5859, "step": 49071 }, { "epoch": 0.05, "learning_rate": 4.9704651123915666e-05, "loss": 0.6325, "step": 49072 }, { "epoch": 0.05, "learning_rate": 4.970463908565122e-05, "loss": 0.633, "step": 49073 }, { "epoch": 0.05, "learning_rate": 4.9704627047142885e-05, "loss": 0.5955, "step": 49074 }, { "epoch": 0.05, "learning_rate": 4.970461500839069e-05, "loss": 0.8627, "step": 49075 }, { "epoch": 0.05, "learning_rate": 4.970460296939461e-05, "loss": 0.8267, "step": 49076 }, { "epoch": 0.05, "learning_rate": 4.970459093015467e-05, "loss": 0.7594, "step": 49077 }, { "epoch": 0.05, "learning_rate": 4.970457889067085e-05, "loss": 1.0365, "step": 49078 }, { "epoch": 0.05, "learning_rate": 4.9704566850943155e-05, "loss": 1.088, "step": 49079 }, { "epoch": 0.05, "learning_rate": 4.970455481097158e-05, "loss": 1.1838, "step": 49080 }, { "epoch": 0.05, "learning_rate": 4.9704542770756146e-05, "loss": 1.1609, "step": 49081 }, { "epoch": 0.05, "learning_rate": 4.970453073029683e-05, "loss": 1.1923, "step": 49082 }, { "epoch": 0.05, "learning_rate": 4.970451868959365e-05, "loss": 1.2772, "step": 49083 }, { "epoch": 0.05, "learning_rate": 4.970450664864659e-05, "loss": 0.9725, "step": 49084 }, { "epoch": 0.05, "learning_rate": 4.970449460745566e-05, "loss": 1.2527, "step": 49085 }, { "epoch": 0.05, "learning_rate": 4.970448256602085e-05, "loss": 1.0655, "step": 49086 }, { "epoch": 0.05, "learning_rate": 4.970447052434218e-05, "loss": 1.0104, "step": 49087 }, { "epoch": 0.05, "learning_rate": 4.9704458482419636e-05, "loss": 1.1331, "step": 49088 }, { "epoch": 0.05, "learning_rate": 4.970444644025322e-05, "loss": 1.0112, "step": 49089 }, { "epoch": 0.05, "learning_rate": 4.970443439784292e-05, "loss": 0.8151, "step": 49090 }, { "epoch": 0.05, "learning_rate": 4.970442235518876e-05, "loss": 1.1967, "step": 49091 }, { "epoch": 0.05, "learning_rate": 4.9704410312290726e-05, "loss": 1.6702, "step": 49092 }, { "epoch": 0.05, "learning_rate": 4.9704398269148825e-05, "loss": 1.3195, "step": 49093 }, { "epoch": 0.05, "learning_rate": 4.9704386225763045e-05, "loss": 1.3374, "step": 49094 }, { "epoch": 0.05, "learning_rate": 4.9704374182133395e-05, "loss": 0.941, "step": 49095 }, { "epoch": 0.05, "learning_rate": 4.9704362138259874e-05, "loss": 0.9783, "step": 49096 }, { "epoch": 0.05, "learning_rate": 4.970435009414248e-05, "loss": 0.8569, "step": 49097 }, { "epoch": 0.05, "learning_rate": 4.970433804978122e-05, "loss": 1.1719, "step": 49098 }, { "epoch": 0.05, "learning_rate": 4.9704326005176085e-05, "loss": 1.2234, "step": 49099 }, { "epoch": 0.05, "learning_rate": 4.970431396032709e-05, "loss": 1.4857, "step": 49100 }, { "epoch": 0.05, "learning_rate": 4.970430191523421e-05, "loss": 1.1595, "step": 49101 }, { "epoch": 0.05, "learning_rate": 4.9704289869897464e-05, "loss": 1.1469, "step": 49102 }, { "epoch": 0.05, "learning_rate": 4.9704277824316847e-05, "loss": 0.8056, "step": 49103 }, { "epoch": 0.05, "learning_rate": 4.9704265778492365e-05, "loss": 0.9138, "step": 49104 }, { "epoch": 0.05, "learning_rate": 4.970425373242401e-05, "loss": 1.3034, "step": 49105 }, { "epoch": 0.05, "learning_rate": 4.970424168611179e-05, "loss": 1.0007, "step": 49106 }, { "epoch": 0.05, "learning_rate": 4.9704229639555686e-05, "loss": 1.0522, "step": 49107 }, { "epoch": 0.05, "learning_rate": 4.970421759275573e-05, "loss": 0.9415, "step": 49108 }, { "epoch": 0.05, "learning_rate": 4.970420554571189e-05, "loss": 0.9291, "step": 49109 }, { "epoch": 0.05, "learning_rate": 4.970419349842419e-05, "loss": 0.9965, "step": 49110 }, { "epoch": 0.05, "learning_rate": 4.970418145089262e-05, "loss": 1.228, "step": 49111 }, { "epoch": 0.05, "learning_rate": 4.9704169403117175e-05, "loss": 1.151, "step": 49112 }, { "epoch": 0.05, "learning_rate": 4.9704157355097855e-05, "loss": 1.0583, "step": 49113 }, { "epoch": 0.05, "learning_rate": 4.9704145306834684e-05, "loss": 0.8792, "step": 49114 }, { "epoch": 0.05, "learning_rate": 4.970413325832763e-05, "loss": 1.1194, "step": 49115 }, { "epoch": 0.05, "learning_rate": 4.970412120957671e-05, "loss": 0.8482, "step": 49116 }, { "epoch": 0.05, "learning_rate": 4.9704109160581924e-05, "loss": 1.0055, "step": 49117 }, { "epoch": 0.05, "learning_rate": 4.970409711134327e-05, "loss": 0.9937, "step": 49118 }, { "epoch": 0.05, "learning_rate": 4.9704085061860744e-05, "loss": 0.7969, "step": 49119 }, { "epoch": 0.05, "learning_rate": 4.970407301213435e-05, "loss": 0.9367, "step": 49120 }, { "epoch": 0.05, "learning_rate": 4.970406096216409e-05, "loss": 1.1268, "step": 49121 }, { "epoch": 0.05, "learning_rate": 4.970404891194996e-05, "loss": 1.1483, "step": 49122 }, { "epoch": 0.05, "learning_rate": 4.9704036861491965e-05, "loss": 0.7075, "step": 49123 }, { "epoch": 0.05, "learning_rate": 4.97040248107901e-05, "loss": 0.9617, "step": 49124 }, { "epoch": 0.05, "learning_rate": 4.9704012759844366e-05, "loss": 1.1791, "step": 49125 }, { "epoch": 0.05, "learning_rate": 4.970400070865477e-05, "loss": 0.7995, "step": 49126 }, { "epoch": 0.05, "learning_rate": 4.97039886572213e-05, "loss": 0.9303, "step": 49127 }, { "epoch": 0.05, "learning_rate": 4.970397660554397e-05, "loss": 0.9439, "step": 49128 }, { "epoch": 0.05, "learning_rate": 4.970396455362276e-05, "loss": 1.0105, "step": 49129 }, { "epoch": 0.05, "learning_rate": 4.9703952501457693e-05, "loss": 1.0824, "step": 49130 }, { "epoch": 0.05, "learning_rate": 4.970394044904876e-05, "loss": 1.0056, "step": 49131 }, { "epoch": 0.05, "learning_rate": 4.970392839639596e-05, "loss": 1.1438, "step": 49132 }, { "epoch": 0.05, "learning_rate": 4.970391634349929e-05, "loss": 1.0926, "step": 49133 }, { "epoch": 0.05, "learning_rate": 4.970390429035875e-05, "loss": 1.1894, "step": 49134 }, { "epoch": 0.05, "learning_rate": 4.970389223697435e-05, "loss": 1.0072, "step": 49135 }, { "epoch": 0.05, "learning_rate": 4.970388018334608e-05, "loss": 0.9159, "step": 49136 }, { "epoch": 0.05, "learning_rate": 4.970386812947394e-05, "loss": 0.9507, "step": 49137 }, { "epoch": 0.05, "learning_rate": 4.970385607535794e-05, "loss": 1.0092, "step": 49138 }, { "epoch": 0.05, "learning_rate": 4.970384402099807e-05, "loss": 0.5426, "step": 49139 }, { "epoch": 0.05, "learning_rate": 4.970383196639434e-05, "loss": 1.1667, "step": 49140 }, { "epoch": 0.05, "learning_rate": 4.970381991154674e-05, "loss": 1.1706, "step": 49141 }, { "epoch": 0.05, "learning_rate": 4.9703807856455277e-05, "loss": 1.1304, "step": 49142 }, { "epoch": 0.05, "learning_rate": 4.970379580111994e-05, "loss": 0.8809, "step": 49143 }, { "epoch": 0.05, "learning_rate": 4.9703783745540745e-05, "loss": 1.1021, "step": 49144 }, { "epoch": 0.05, "learning_rate": 4.970377168971768e-05, "loss": 1.5809, "step": 49145 }, { "epoch": 0.05, "learning_rate": 4.9703759633650756e-05, "loss": 1.2629, "step": 49146 }, { "epoch": 0.05, "learning_rate": 4.9703747577339966e-05, "loss": 0.9902, "step": 49147 }, { "epoch": 0.05, "learning_rate": 4.970373552078531e-05, "loss": 1.1831, "step": 49148 }, { "epoch": 0.05, "learning_rate": 4.9703723463986786e-05, "loss": 1.1929, "step": 49149 }, { "epoch": 0.05, "learning_rate": 4.9703711406944396e-05, "loss": 0.8317, "step": 49150 }, { "epoch": 0.05, "learning_rate": 4.970369934965815e-05, "loss": 0.758, "step": 49151 }, { "epoch": 0.05, "learning_rate": 4.970368729212803e-05, "loss": 0.8953, "step": 49152 }, { "epoch": 0.05, "learning_rate": 4.9703675234354055e-05, "loss": 0.8258, "step": 49153 }, { "epoch": 0.05, "learning_rate": 4.9703663176336215e-05, "loss": 0.9047, "step": 49154 }, { "epoch": 0.05, "learning_rate": 4.97036511180745e-05, "loss": 1.1103, "step": 49155 }, { "epoch": 0.05, "learning_rate": 4.970363905956893e-05, "loss": 1.3213, "step": 49156 }, { "epoch": 0.05, "learning_rate": 4.970362700081949e-05, "loss": 0.9242, "step": 49157 }, { "epoch": 0.05, "learning_rate": 4.9703614941826195e-05, "loss": 1.0202, "step": 49158 }, { "epoch": 0.05, "learning_rate": 4.970360288258903e-05, "loss": 0.5289, "step": 49159 }, { "epoch": 0.05, "learning_rate": 4.9703590823107996e-05, "loss": 0.6044, "step": 49160 }, { "epoch": 0.05, "learning_rate": 4.970357876338311e-05, "loss": 0.4397, "step": 49161 }, { "epoch": 0.05, "learning_rate": 4.9703566703414354e-05, "loss": 0.4052, "step": 49162 }, { "epoch": 0.05, "learning_rate": 4.970355464320173e-05, "loss": 0.4156, "step": 49163 }, { "epoch": 0.05, "learning_rate": 4.9703542582745255e-05, "loss": 0.9121, "step": 49164 }, { "epoch": 0.05, "learning_rate": 4.970353052204491e-05, "loss": 0.969, "step": 49165 }, { "epoch": 0.05, "learning_rate": 4.9703518461100706e-05, "loss": 1.12, "step": 49166 }, { "epoch": 0.05, "learning_rate": 4.970350639991264e-05, "loss": 0.7309, "step": 49167 }, { "epoch": 0.05, "learning_rate": 4.970349433848071e-05, "loss": 1.0151, "step": 49168 }, { "epoch": 0.05, "learning_rate": 4.970348227680491e-05, "loss": 1.0263, "step": 49169 }, { "epoch": 0.05, "learning_rate": 4.970347021488525e-05, "loss": 1.1288, "step": 49170 }, { "epoch": 0.05, "learning_rate": 4.9703458152721736e-05, "loss": 0.3055, "step": 49171 }, { "epoch": 0.05, "learning_rate": 4.9703446090314354e-05, "loss": 0.2432, "step": 49172 }, { "epoch": 0.05, "learning_rate": 4.970343402766311e-05, "loss": 0.7339, "step": 49173 }, { "epoch": 0.05, "learning_rate": 4.9703421964768006e-05, "loss": 0.9211, "step": 49174 }, { "epoch": 0.05, "learning_rate": 4.970340990162904e-05, "loss": 0.6416, "step": 49175 }, { "epoch": 0.05, "learning_rate": 4.9703397838246215e-05, "loss": 1.0328, "step": 49176 }, { "epoch": 0.05, "learning_rate": 4.970338577461952e-05, "loss": 0.9996, "step": 49177 }, { "epoch": 0.05, "learning_rate": 4.9703373710748967e-05, "loss": 1.2433, "step": 49178 }, { "epoch": 0.05, "learning_rate": 4.970336164663456e-05, "loss": 1.2472, "step": 49179 }, { "epoch": 0.05, "learning_rate": 4.970334958227628e-05, "loss": 1.3543, "step": 49180 }, { "epoch": 0.05, "learning_rate": 4.970333751767415e-05, "loss": 1.1434, "step": 49181 }, { "epoch": 0.05, "learning_rate": 4.9703325452828154e-05, "loss": 1.075, "step": 49182 }, { "epoch": 0.05, "learning_rate": 4.9703313387738294e-05, "loss": 0.9395, "step": 49183 }, { "epoch": 0.05, "learning_rate": 4.9703301322404584e-05, "loss": 1.009, "step": 49184 }, { "epoch": 0.05, "learning_rate": 4.9703289256827e-05, "loss": 1.1244, "step": 49185 }, { "epoch": 0.05, "learning_rate": 4.9703277191005556e-05, "loss": 1.1026, "step": 49186 }, { "epoch": 0.05, "learning_rate": 4.970326512494027e-05, "loss": 0.9147, "step": 49187 }, { "epoch": 0.05, "learning_rate": 4.9703253058631106e-05, "loss": 0.8856, "step": 49188 }, { "epoch": 0.05, "learning_rate": 4.970324099207809e-05, "loss": 0.8799, "step": 49189 }, { "epoch": 0.05, "learning_rate": 4.9703228925281206e-05, "loss": 1.0212, "step": 49190 }, { "epoch": 0.05, "learning_rate": 4.9703216858240466e-05, "loss": 1.1555, "step": 49191 }, { "epoch": 0.05, "learning_rate": 4.970320479095587e-05, "loss": 1.1913, "step": 49192 }, { "epoch": 0.05, "learning_rate": 4.970319272342741e-05, "loss": 0.644, "step": 49193 }, { "epoch": 0.05, "learning_rate": 4.970318065565509e-05, "loss": 0.5525, "step": 49194 }, { "epoch": 0.05, "learning_rate": 4.9703168587638914e-05, "loss": 1.0745, "step": 49195 }, { "epoch": 0.05, "learning_rate": 4.970315651937888e-05, "loss": 0.9688, "step": 49196 }, { "epoch": 0.05, "learning_rate": 4.970314445087498e-05, "loss": 1.0551, "step": 49197 }, { "epoch": 0.05, "learning_rate": 4.970313238212723e-05, "loss": 1.2486, "step": 49198 }, { "epoch": 0.05, "learning_rate": 4.970312031313562e-05, "loss": 0.8232, "step": 49199 }, { "epoch": 0.05, "learning_rate": 4.9703108243900146e-05, "loss": 0.8392, "step": 49200 }, { "epoch": 0.05, "learning_rate": 4.970309617442082e-05, "loss": 1.2476, "step": 49201 }, { "epoch": 0.05, "learning_rate": 4.970308410469763e-05, "loss": 1.1765, "step": 49202 }, { "epoch": 0.05, "learning_rate": 4.970307203473058e-05, "loss": 1.0508, "step": 49203 }, { "epoch": 0.05, "learning_rate": 4.970305996451967e-05, "loss": 0.8849, "step": 49204 }, { "epoch": 0.05, "learning_rate": 4.97030478940649e-05, "loss": 0.6046, "step": 49205 }, { "epoch": 0.05, "learning_rate": 4.970303582336628e-05, "loss": 0.656, "step": 49206 }, { "epoch": 0.05, "learning_rate": 4.97030237524238e-05, "loss": 1.0301, "step": 49207 }, { "epoch": 0.05, "learning_rate": 4.970301168123747e-05, "loss": 1.4154, "step": 49208 }, { "epoch": 0.05, "learning_rate": 4.970299960980727e-05, "loss": 1.0479, "step": 49209 }, { "epoch": 0.05, "learning_rate": 4.970298753813322e-05, "loss": 0.9415, "step": 49210 }, { "epoch": 0.05, "learning_rate": 4.970297546621531e-05, "loss": 1.0544, "step": 49211 }, { "epoch": 0.05, "learning_rate": 4.970296339405354e-05, "loss": 1.1165, "step": 49212 }, { "epoch": 0.05, "learning_rate": 4.970295132164791e-05, "loss": 1.0928, "step": 49213 }, { "epoch": 0.05, "learning_rate": 4.970293924899843e-05, "loss": 0.9414, "step": 49214 }, { "epoch": 0.05, "learning_rate": 4.9702927176105084e-05, "loss": 1.0549, "step": 49215 }, { "epoch": 0.05, "learning_rate": 4.970291510296789e-05, "loss": 0.4215, "step": 49216 }, { "epoch": 0.05, "learning_rate": 4.970290302958684e-05, "loss": 0.7109, "step": 49217 }, { "epoch": 0.05, "learning_rate": 4.970289095596192e-05, "loss": 1.0233, "step": 49218 }, { "epoch": 0.05, "learning_rate": 4.970287888209316e-05, "loss": 1.1946, "step": 49219 }, { "epoch": 0.05, "learning_rate": 4.9702866807980536e-05, "loss": 0.6813, "step": 49220 }, { "epoch": 0.05, "learning_rate": 4.970285473362406e-05, "loss": 0.9359, "step": 49221 }, { "epoch": 0.05, "learning_rate": 4.970284265902372e-05, "loss": 0.9165, "step": 49222 }, { "epoch": 0.05, "learning_rate": 4.9702830584179525e-05, "loss": 0.8366, "step": 49223 }, { "epoch": 0.05, "learning_rate": 4.9702818509091484e-05, "loss": 0.7746, "step": 49224 }, { "epoch": 0.05, "learning_rate": 4.970280643375958e-05, "loss": 0.8766, "step": 49225 }, { "epoch": 0.05, "learning_rate": 4.970279435818382e-05, "loss": 0.98, "step": 49226 }, { "epoch": 0.05, "learning_rate": 4.9702782282364205e-05, "loss": 1.1382, "step": 49227 }, { "epoch": 0.05, "learning_rate": 4.9702770206300735e-05, "loss": 0.504, "step": 49228 }, { "epoch": 0.05, "learning_rate": 4.970275812999341e-05, "loss": 0.6278, "step": 49229 }, { "epoch": 0.05, "learning_rate": 4.970274605344223e-05, "loss": 0.921, "step": 49230 }, { "epoch": 0.05, "learning_rate": 4.9702733976647194e-05, "loss": 0.865, "step": 49231 }, { "epoch": 0.05, "learning_rate": 4.97027218996083e-05, "loss": 1.2186, "step": 49232 }, { "epoch": 0.05, "learning_rate": 4.970270982232556e-05, "loss": 1.156, "step": 49233 }, { "epoch": 0.05, "learning_rate": 4.970269774479896e-05, "loss": 1.1751, "step": 49234 }, { "epoch": 0.05, "learning_rate": 4.9702685667028505e-05, "loss": 0.9222, "step": 49235 }, { "epoch": 0.05, "learning_rate": 4.970267358901419e-05, "loss": 0.7707, "step": 49236 }, { "epoch": 0.05, "learning_rate": 4.970266151075603e-05, "loss": 1.1664, "step": 49237 }, { "epoch": 0.05, "learning_rate": 4.970264943225401e-05, "loss": 0.6423, "step": 49238 }, { "epoch": 0.05, "learning_rate": 4.970263735350814e-05, "loss": 0.9882, "step": 49239 }, { "epoch": 0.05, "learning_rate": 4.9702625274518414e-05, "loss": 1.2393, "step": 49240 }, { "epoch": 0.05, "learning_rate": 4.970261319528483e-05, "loss": 1.4499, "step": 49241 }, { "epoch": 0.05, "learning_rate": 4.97026011158074e-05, "loss": 1.2296, "step": 49242 }, { "epoch": 0.05, "learning_rate": 4.970258903608611e-05, "loss": 0.9282, "step": 49243 }, { "epoch": 0.05, "learning_rate": 4.9702576956120974e-05, "loss": 0.8888, "step": 49244 }, { "epoch": 0.05, "learning_rate": 4.9702564875911976e-05, "loss": 1.5979, "step": 49245 }, { "epoch": 0.05, "learning_rate": 4.970255279545912e-05, "loss": 1.8558, "step": 49246 }, { "epoch": 0.05, "learning_rate": 4.970254071476243e-05, "loss": 0.8785, "step": 49247 }, { "epoch": 0.05, "learning_rate": 4.970252863382187e-05, "loss": 0.4065, "step": 49248 }, { "epoch": 0.05, "learning_rate": 4.970251655263746e-05, "loss": 0.4805, "step": 49249 }, { "epoch": 0.05, "learning_rate": 4.97025044712092e-05, "loss": 0.8593, "step": 49250 }, { "epoch": 0.05, "learning_rate": 4.9702492389537095e-05, "loss": 0.9746, "step": 49251 }, { "epoch": 0.05, "learning_rate": 4.970248030762112e-05, "loss": 0.9178, "step": 49252 }, { "epoch": 0.05, "learning_rate": 4.9702468225461305e-05, "loss": 1.0239, "step": 49253 }, { "epoch": 0.05, "learning_rate": 4.970245614305764e-05, "loss": 0.6192, "step": 49254 }, { "epoch": 0.05, "learning_rate": 4.970244406041011e-05, "loss": 0.7455, "step": 49255 }, { "epoch": 0.05, "learning_rate": 4.970243197751874e-05, "loss": 1.0458, "step": 49256 }, { "epoch": 0.05, "learning_rate": 4.970241989438351e-05, "loss": 0.9976, "step": 49257 }, { "epoch": 0.05, "learning_rate": 4.9702407811004434e-05, "loss": 0.8518, "step": 49258 }, { "epoch": 0.05, "learning_rate": 4.97023957273815e-05, "loss": 0.7962, "step": 49259 }, { "epoch": 0.05, "learning_rate": 4.970238364351473e-05, "loss": 1.018, "step": 49260 }, { "epoch": 0.05, "learning_rate": 4.970237155940409e-05, "loss": 1.123, "step": 49261 }, { "epoch": 0.05, "learning_rate": 4.970235947504961e-05, "loss": 0.8241, "step": 49262 }, { "epoch": 0.05, "learning_rate": 4.970234739045127e-05, "loss": 0.9632, "step": 49263 }, { "epoch": 0.05, "learning_rate": 4.9702335305609083e-05, "loss": 1.0769, "step": 49264 }, { "epoch": 0.05, "learning_rate": 4.9702323220523045e-05, "loss": 1.2423, "step": 49265 }, { "epoch": 0.05, "learning_rate": 4.970231113519316e-05, "loss": 0.6706, "step": 49266 }, { "epoch": 0.05, "learning_rate": 4.970229904961942e-05, "loss": 1.0386, "step": 49267 }, { "epoch": 0.05, "learning_rate": 4.970228696380183e-05, "loss": 1.2034, "step": 49268 }, { "epoch": 0.05, "learning_rate": 4.970227487774039e-05, "loss": 0.7983, "step": 49269 }, { "epoch": 0.05, "learning_rate": 4.97022627914351e-05, "loss": 0.9121, "step": 49270 }, { "epoch": 0.05, "learning_rate": 4.970225070488595e-05, "loss": 1.0182, "step": 49271 }, { "epoch": 0.05, "learning_rate": 4.9702238618092956e-05, "loss": 0.9931, "step": 49272 }, { "epoch": 0.05, "learning_rate": 4.970222653105612e-05, "loss": 0.5544, "step": 49273 }, { "epoch": 0.05, "learning_rate": 4.970221444377542e-05, "loss": 0.9423, "step": 49274 }, { "epoch": 0.05, "learning_rate": 4.970220235625088e-05, "loss": 0.8785, "step": 49275 }, { "epoch": 0.05, "learning_rate": 4.970219026848249e-05, "loss": 1.0599, "step": 49276 }, { "epoch": 0.05, "learning_rate": 4.970217818047025e-05, "loss": 1.0101, "step": 49277 }, { "epoch": 0.05, "learning_rate": 4.970216609221415e-05, "loss": 0.9442, "step": 49278 }, { "epoch": 0.05, "learning_rate": 4.970215400371422e-05, "loss": 1.0618, "step": 49279 }, { "epoch": 0.05, "learning_rate": 4.970214191497042e-05, "loss": 1.1712, "step": 49280 }, { "epoch": 0.05, "learning_rate": 4.970212982598279e-05, "loss": 1.1162, "step": 49281 }, { "epoch": 0.05, "learning_rate": 4.97021177367513e-05, "loss": 1.012, "step": 49282 }, { "epoch": 0.05, "learning_rate": 4.970210564727597e-05, "loss": 0.7131, "step": 49283 }, { "epoch": 0.05, "learning_rate": 4.9702093557556775e-05, "loss": 0.2382, "step": 49284 }, { "epoch": 0.05, "learning_rate": 4.9702081467593745e-05, "loss": 0.2728, "step": 49285 }, { "epoch": 0.05, "learning_rate": 4.9702069377386864e-05, "loss": 0.1948, "step": 49286 }, { "epoch": 0.05, "learning_rate": 4.970205728693613e-05, "loss": 0.0659, "step": 49287 }, { "epoch": 0.05, "learning_rate": 4.970204519624155e-05, "loss": 0.6515, "step": 49288 }, { "epoch": 0.05, "learning_rate": 4.9702033105303126e-05, "loss": 0.8122, "step": 49289 }, { "epoch": 0.05, "learning_rate": 4.970202101412085e-05, "loss": 1.0526, "step": 49290 }, { "epoch": 0.05, "learning_rate": 4.970200892269472e-05, "loss": 0.9306, "step": 49291 }, { "epoch": 0.05, "learning_rate": 4.970199683102476e-05, "loss": 1.1721, "step": 49292 }, { "epoch": 0.05, "learning_rate": 4.970198473911094e-05, "loss": 0.9256, "step": 49293 }, { "epoch": 0.05, "learning_rate": 4.9701972646953265e-05, "loss": 0.7625, "step": 49294 }, { "epoch": 0.05, "learning_rate": 4.970196055455175e-05, "loss": 0.2986, "step": 49295 }, { "epoch": 0.05, "learning_rate": 4.970194846190639e-05, "loss": 0.9155, "step": 49296 }, { "epoch": 0.05, "learning_rate": 4.9701936369017186e-05, "loss": 1.0114, "step": 49297 }, { "epoch": 0.05, "learning_rate": 4.9701924275884125e-05, "loss": 0.9612, "step": 49298 }, { "epoch": 0.05, "learning_rate": 4.970191218250723e-05, "loss": 0.8865, "step": 49299 }, { "epoch": 0.05, "learning_rate": 4.970190008888648e-05, "loss": 1.0928, "step": 49300 }, { "epoch": 0.05, "learning_rate": 4.970188799502188e-05, "loss": 0.9871, "step": 49301 }, { "epoch": 0.05, "learning_rate": 4.9701875900913436e-05, "loss": 0.7684, "step": 49302 }, { "epoch": 0.05, "learning_rate": 4.970186380656114e-05, "loss": 0.8558, "step": 49303 }, { "epoch": 0.05, "learning_rate": 4.970185171196501e-05, "loss": 1.1148, "step": 49304 }, { "epoch": 0.05, "learning_rate": 4.970183961712503e-05, "loss": 1.0677, "step": 49305 }, { "epoch": 0.05, "learning_rate": 4.97018275220412e-05, "loss": 1.0732, "step": 49306 }, { "epoch": 0.05, "learning_rate": 4.970181542671353e-05, "loss": 1.2291, "step": 49307 }, { "epoch": 0.05, "learning_rate": 4.9701803331142e-05, "loss": 1.1861, "step": 49308 }, { "epoch": 0.05, "learning_rate": 4.970179123532664e-05, "loss": 1.352, "step": 49309 }, { "epoch": 0.05, "learning_rate": 4.9701779139267424e-05, "loss": 0.8551, "step": 49310 }, { "epoch": 0.05, "learning_rate": 4.970176704296437e-05, "loss": 0.631, "step": 49311 }, { "epoch": 0.05, "learning_rate": 4.9701754946417464e-05, "loss": 0.9091, "step": 49312 }, { "epoch": 0.05, "learning_rate": 4.970174284962672e-05, "loss": 0.8498, "step": 49313 }, { "epoch": 0.05, "learning_rate": 4.970173075259212e-05, "loss": 1.1056, "step": 49314 }, { "epoch": 0.05, "learning_rate": 4.970171865531369e-05, "loss": 1.0499, "step": 49315 }, { "epoch": 0.05, "learning_rate": 4.9701706557791405e-05, "loss": 0.9585, "step": 49316 }, { "epoch": 0.05, "learning_rate": 4.970169446002528e-05, "loss": 1.6578, "step": 49317 }, { "epoch": 0.05, "learning_rate": 4.97016823620153e-05, "loss": 1.0333, "step": 49318 }, { "epoch": 0.05, "learning_rate": 4.970167026376148e-05, "loss": 1.0478, "step": 49319 }, { "epoch": 0.05, "learning_rate": 4.970165816526382e-05, "loss": 0.7092, "step": 49320 }, { "epoch": 0.05, "learning_rate": 4.970164606652232e-05, "loss": 1.0351, "step": 49321 }, { "epoch": 0.05, "learning_rate": 4.9701633967536966e-05, "loss": 1.1692, "step": 49322 }, { "epoch": 0.05, "learning_rate": 4.9701621868307775e-05, "loss": 1.1672, "step": 49323 }, { "epoch": 0.05, "learning_rate": 4.970160976883473e-05, "loss": 1.025, "step": 49324 }, { "epoch": 0.05, "learning_rate": 4.970159766911785e-05, "loss": 0.1514, "step": 49325 }, { "epoch": 0.05, "learning_rate": 4.9701585569157125e-05, "loss": 1.1686, "step": 49326 }, { "epoch": 0.05, "learning_rate": 4.970157346895256e-05, "loss": 0.6867, "step": 49327 }, { "epoch": 0.05, "learning_rate": 4.970156136850415e-05, "loss": 1.1729, "step": 49328 }, { "epoch": 0.05, "learning_rate": 4.970154926781189e-05, "loss": 1.0695, "step": 49329 }, { "epoch": 0.05, "learning_rate": 4.970153716687579e-05, "loss": 1.0858, "step": 49330 }, { "epoch": 0.05, "learning_rate": 4.970152506569584e-05, "loss": 0.6815, "step": 49331 }, { "epoch": 0.05, "learning_rate": 4.9701512964272056e-05, "loss": 0.9542, "step": 49332 }, { "epoch": 0.05, "learning_rate": 4.970150086260442e-05, "loss": 0.6903, "step": 49333 }, { "epoch": 0.05, "learning_rate": 4.9701488760692953e-05, "loss": 0.7983, "step": 49334 }, { "epoch": 0.05, "learning_rate": 4.9701476658537636e-05, "loss": 0.8928, "step": 49335 }, { "epoch": 0.05, "learning_rate": 4.970146455613848e-05, "loss": 0.8243, "step": 49336 }, { "epoch": 0.05, "learning_rate": 4.970145245349548e-05, "loss": 0.2803, "step": 49337 }, { "epoch": 0.05, "learning_rate": 4.970144035060864e-05, "loss": 0.1855, "step": 49338 }, { "epoch": 0.05, "learning_rate": 4.9701428247477957e-05, "loss": 0.1977, "step": 49339 }, { "epoch": 0.05, "learning_rate": 4.970141614410343e-05, "loss": 0.2214, "step": 49340 }, { "epoch": 0.05, "learning_rate": 4.970140404048506e-05, "loss": 0.1916, "step": 49341 }, { "epoch": 0.05, "learning_rate": 4.970139193662286e-05, "loss": 0.3204, "step": 49342 }, { "epoch": 0.05, "learning_rate": 4.9701379832516796e-05, "loss": 0.8594, "step": 49343 }, { "epoch": 0.05, "learning_rate": 4.9701367728166904e-05, "loss": 0.9501, "step": 49344 }, { "epoch": 0.05, "learning_rate": 4.970135562357317e-05, "loss": 0.828, "step": 49345 }, { "epoch": 0.05, "learning_rate": 4.97013435187356e-05, "loss": 0.784, "step": 49346 }, { "epoch": 0.05, "learning_rate": 4.9701331413654174e-05, "loss": 0.6562, "step": 49347 }, { "epoch": 0.05, "learning_rate": 4.970131930832892e-05, "loss": 0.9191, "step": 49348 }, { "epoch": 0.05, "learning_rate": 4.970130720275982e-05, "loss": 1.0661, "step": 49349 }, { "epoch": 0.05, "learning_rate": 4.9701295096946876e-05, "loss": 1.1562, "step": 49350 }, { "epoch": 0.05, "learning_rate": 4.97012829908901e-05, "loss": 0.7336, "step": 49351 }, { "epoch": 0.05, "learning_rate": 4.9701270884589476e-05, "loss": 1.0653, "step": 49352 }, { "epoch": 0.05, "learning_rate": 4.970125877804501e-05, "loss": 0.8774, "step": 49353 }, { "epoch": 0.05, "learning_rate": 4.970124667125671e-05, "loss": 1.0351, "step": 49354 }, { "epoch": 0.05, "learning_rate": 4.970123456422456e-05, "loss": 1.1957, "step": 49355 }, { "epoch": 0.05, "learning_rate": 4.9701222456948584e-05, "loss": 0.6754, "step": 49356 }, { "epoch": 0.05, "learning_rate": 4.970121034942876e-05, "loss": 0.693, "step": 49357 }, { "epoch": 0.05, "learning_rate": 4.970119824166509e-05, "loss": 1.5728, "step": 49358 }, { "epoch": 0.05, "learning_rate": 4.970118613365759e-05, "loss": 1.3361, "step": 49359 }, { "epoch": 0.05, "learning_rate": 4.970117402540625e-05, "loss": 0.9865, "step": 49360 }, { "epoch": 0.05, "learning_rate": 4.970116191691106e-05, "loss": 0.4171, "step": 49361 }, { "epoch": 0.05, "learning_rate": 4.970114980817204e-05, "loss": 0.2866, "step": 49362 }, { "epoch": 0.05, "learning_rate": 4.970113769918918e-05, "loss": 0.7806, "step": 49363 }, { "epoch": 0.05, "learning_rate": 4.970112558996248e-05, "loss": 1.3016, "step": 49364 }, { "epoch": 0.05, "learning_rate": 4.970111348049194e-05, "loss": 1.1949, "step": 49365 }, { "epoch": 0.05, "learning_rate": 4.970110137077756e-05, "loss": 1.1182, "step": 49366 }, { "epoch": 0.05, "learning_rate": 4.970108926081934e-05, "loss": 0.8209, "step": 49367 }, { "epoch": 0.05, "learning_rate": 4.9701077150617286e-05, "loss": 0.7123, "step": 49368 }, { "epoch": 0.05, "learning_rate": 4.970106504017139e-05, "loss": 0.4967, "step": 49369 }, { "epoch": 0.05, "learning_rate": 4.970105292948166e-05, "loss": 0.3679, "step": 49370 }, { "epoch": 0.05, "learning_rate": 4.970104081854808e-05, "loss": 0.5399, "step": 49371 }, { "epoch": 0.05, "learning_rate": 4.970102870737068e-05, "loss": 0.4392, "step": 49372 }, { "epoch": 0.05, "learning_rate": 4.9701016595949425e-05, "loss": 0.5869, "step": 49373 }, { "epoch": 0.05, "learning_rate": 4.970100448428434e-05, "loss": 1.0535, "step": 49374 }, { "epoch": 0.05, "learning_rate": 4.970099237237542e-05, "loss": 0.8251, "step": 49375 }, { "epoch": 0.05, "learning_rate": 4.970098026022265e-05, "loss": 1.3761, "step": 49376 }, { "epoch": 0.05, "learning_rate": 4.970096814782606e-05, "loss": 1.0517, "step": 49377 }, { "epoch": 0.05, "learning_rate": 4.970095603518562e-05, "loss": 1.1177, "step": 49378 }, { "epoch": 0.05, "learning_rate": 4.970094392230134e-05, "loss": 1.5543, "step": 49379 }, { "epoch": 0.05, "learning_rate": 4.9700931809173226e-05, "loss": 1.6791, "step": 49380 }, { "epoch": 0.05, "learning_rate": 4.970091969580128e-05, "loss": 1.7187, "step": 49381 }, { "epoch": 0.05, "learning_rate": 4.9700907582185494e-05, "loss": 1.2126, "step": 49382 }, { "epoch": 0.05, "learning_rate": 4.970089546832587e-05, "loss": 1.1443, "step": 49383 }, { "epoch": 0.05, "learning_rate": 4.970088335422241e-05, "loss": 0.9018, "step": 49384 }, { "epoch": 0.05, "learning_rate": 4.9700871239875114e-05, "loss": 1.2226, "step": 49385 }, { "epoch": 0.05, "learning_rate": 4.9700859125283984e-05, "loss": 1.0516, "step": 49386 }, { "epoch": 0.05, "learning_rate": 4.970084701044902e-05, "loss": 1.1128, "step": 49387 }, { "epoch": 0.05, "learning_rate": 4.970083489537021e-05, "loss": 0.8111, "step": 49388 }, { "epoch": 0.05, "learning_rate": 4.970082278004756e-05, "loss": 1.1613, "step": 49389 }, { "epoch": 0.05, "learning_rate": 4.970081066448108e-05, "loss": 1.0762, "step": 49390 }, { "epoch": 0.05, "learning_rate": 4.970079854867077e-05, "loss": 0.9629, "step": 49391 }, { "epoch": 0.05, "learning_rate": 4.970078643261662e-05, "loss": 1.0423, "step": 49392 }, { "epoch": 0.05, "learning_rate": 4.970077431631863e-05, "loss": 1.1219, "step": 49393 }, { "epoch": 0.05, "learning_rate": 4.9700762199776806e-05, "loss": 0.9233, "step": 49394 }, { "epoch": 0.05, "learning_rate": 4.970075008299116e-05, "loss": 0.9822, "step": 49395 }, { "epoch": 0.05, "learning_rate": 4.970073796596166e-05, "loss": 1.0145, "step": 49396 }, { "epoch": 0.05, "learning_rate": 4.970072584868833e-05, "loss": 0.9013, "step": 49397 }, { "epoch": 0.05, "learning_rate": 4.9700713731171165e-05, "loss": 0.7693, "step": 49398 }, { "epoch": 0.05, "learning_rate": 4.9700701613410174e-05, "loss": 1.0783, "step": 49399 }, { "epoch": 0.05, "learning_rate": 4.970068949540534e-05, "loss": 0.0041, "step": 49400 }, { "epoch": 0.05, "learning_rate": 4.970067737715667e-05, "loss": 0.0022, "step": 49401 }, { "epoch": 0.05, "learning_rate": 4.970066525866417e-05, "loss": 0.0014, "step": 49402 }, { "epoch": 0.05, "learning_rate": 4.970065313992783e-05, "loss": 0.7213, "step": 49403 }, { "epoch": 0.05, "learning_rate": 4.9700641020947655e-05, "loss": 0.5455, "step": 49404 }, { "epoch": 0.05, "learning_rate": 4.9700628901723655e-05, "loss": 0.0007, "step": 49405 }, { "epoch": 0.05, "learning_rate": 4.970061678225582e-05, "loss": 0.0005, "step": 49406 }, { "epoch": 0.05, "learning_rate": 4.970060466254414e-05, "loss": 0.0005, "step": 49407 }, { "epoch": 0.05, "learning_rate": 4.970059254258863e-05, "loss": 0.7466, "step": 49408 }, { "epoch": 0.05, "learning_rate": 4.9700580422389285e-05, "loss": 1.0046, "step": 49409 }, { "epoch": 0.05, "learning_rate": 4.970056830194611e-05, "loss": 0.9007, "step": 49410 }, { "epoch": 0.05, "learning_rate": 4.9700556181259104e-05, "loss": 0.953, "step": 49411 }, { "epoch": 0.05, "learning_rate": 4.970054406032826e-05, "loss": 0.4956, "step": 49412 }, { "epoch": 0.05, "learning_rate": 4.970053193915358e-05, "loss": 1.1025, "step": 49413 }, { "epoch": 0.05, "learning_rate": 4.9700519817735075e-05, "loss": 1.1426, "step": 49414 }, { "epoch": 0.05, "learning_rate": 4.970050769607273e-05, "loss": 0.8417, "step": 49415 }, { "epoch": 0.05, "learning_rate": 4.970049557416656e-05, "loss": 1.1481, "step": 49416 }, { "epoch": 0.05, "learning_rate": 4.970048345201655e-05, "loss": 0.9196, "step": 49417 }, { "epoch": 0.05, "learning_rate": 4.9700471329622706e-05, "loss": 0.8762, "step": 49418 }, { "epoch": 0.05, "learning_rate": 4.9700459206985034e-05, "loss": 0.8657, "step": 49419 }, { "epoch": 0.05, "learning_rate": 4.970044708410353e-05, "loss": 0.8328, "step": 49420 }, { "epoch": 0.05, "learning_rate": 4.970043496097819e-05, "loss": 1.3251, "step": 49421 }, { "epoch": 0.05, "learning_rate": 4.970042283760902e-05, "loss": 1.3648, "step": 49422 }, { "epoch": 0.05, "learning_rate": 4.970041071399602e-05, "loss": 1.1356, "step": 49423 }, { "epoch": 0.05, "learning_rate": 4.9700398590139175e-05, "loss": 1.0187, "step": 49424 }, { "epoch": 0.05, "learning_rate": 4.9700386466038515e-05, "loss": 0.6391, "step": 49425 }, { "epoch": 0.05, "learning_rate": 4.970037434169401e-05, "loss": 0.4627, "step": 49426 }, { "epoch": 0.05, "learning_rate": 4.9700362217105677e-05, "loss": 0.9848, "step": 49427 }, { "epoch": 0.05, "learning_rate": 4.970035009227352e-05, "loss": 0.9061, "step": 49428 }, { "epoch": 0.05, "learning_rate": 4.9700337967197524e-05, "loss": 0.8277, "step": 49429 }, { "epoch": 0.05, "learning_rate": 4.9700325841877706e-05, "loss": 1.1171, "step": 49430 }, { "epoch": 0.05, "learning_rate": 4.970031371631404e-05, "loss": 1.3235, "step": 49431 }, { "epoch": 0.05, "learning_rate": 4.970030159050656e-05, "loss": 1.2504, "step": 49432 }, { "epoch": 0.05, "learning_rate": 4.970028946445524e-05, "loss": 0.9477, "step": 49433 }, { "epoch": 0.05, "learning_rate": 4.970027733816009e-05, "loss": 1.2332, "step": 49434 }, { "epoch": 0.05, "learning_rate": 4.970026521162111e-05, "loss": 1.347, "step": 49435 }, { "epoch": 0.05, "learning_rate": 4.97002530848383e-05, "loss": 1.0928, "step": 49436 }, { "epoch": 0.05, "learning_rate": 4.970024095781166e-05, "loss": 1.0312, "step": 49437 }, { "epoch": 0.05, "learning_rate": 4.970022883054119e-05, "loss": 1.2203, "step": 49438 }, { "epoch": 0.05, "learning_rate": 4.970021670302688e-05, "loss": 1.2077, "step": 49439 }, { "epoch": 0.05, "learning_rate": 4.970020457526875e-05, "loss": 0.8798, "step": 49440 }, { "epoch": 0.05, "learning_rate": 4.970019244726679e-05, "loss": 0.9077, "step": 49441 }, { "epoch": 0.05, "learning_rate": 4.9700180319021e-05, "loss": 0.663, "step": 49442 }, { "epoch": 0.05, "learning_rate": 4.970016819053138e-05, "loss": 0.894, "step": 49443 }, { "epoch": 0.05, "learning_rate": 4.9700156061797927e-05, "loss": 0.9391, "step": 49444 }, { "epoch": 0.05, "learning_rate": 4.970014393282065e-05, "loss": 1.0318, "step": 49445 }, { "epoch": 0.05, "learning_rate": 4.970013180359954e-05, "loss": 0.9967, "step": 49446 }, { "epoch": 0.05, "learning_rate": 4.970011967413459e-05, "loss": 0.4571, "step": 49447 }, { "epoch": 0.05, "learning_rate": 4.970010754442583e-05, "loss": 1.0297, "step": 49448 }, { "epoch": 0.05, "learning_rate": 4.970009541447323e-05, "loss": 1.4043, "step": 49449 }, { "epoch": 0.05, "learning_rate": 4.97000832842768e-05, "loss": 1.1019, "step": 49450 }, { "epoch": 0.05, "learning_rate": 4.970007115383655e-05, "loss": 0.7921, "step": 49451 }, { "epoch": 0.05, "learning_rate": 4.970005902315247e-05, "loss": 0.7592, "step": 49452 }, { "epoch": 0.05, "learning_rate": 4.970004689222455e-05, "loss": 0.897, "step": 49453 }, { "epoch": 0.05, "learning_rate": 4.970003476105281e-05, "loss": 1.1784, "step": 49454 }, { "epoch": 0.05, "learning_rate": 4.970002262963724e-05, "loss": 0.9718, "step": 49455 }, { "epoch": 0.05, "learning_rate": 4.970001049797784e-05, "loss": 1.0498, "step": 49456 }, { "epoch": 0.05, "learning_rate": 4.969999836607462e-05, "loss": 0.8403, "step": 49457 }, { "epoch": 0.05, "learning_rate": 4.9699986233927566e-05, "loss": 0.6607, "step": 49458 }, { "epoch": 0.05, "learning_rate": 4.969997410153668e-05, "loss": 0.5229, "step": 49459 }, { "epoch": 0.05, "learning_rate": 4.9699961968901976e-05, "loss": 0.5823, "step": 49460 }, { "epoch": 0.05, "learning_rate": 4.969994983602344e-05, "loss": 0.7571, "step": 49461 }, { "epoch": 0.05, "learning_rate": 4.969993770290107e-05, "loss": 1.1239, "step": 49462 }, { "epoch": 0.05, "learning_rate": 4.969992556953488e-05, "loss": 0.8606, "step": 49463 }, { "epoch": 0.05, "learning_rate": 4.969991343592486e-05, "loss": 0.6372, "step": 49464 }, { "epoch": 0.05, "learning_rate": 4.9699901302071015e-05, "loss": 1.2361, "step": 49465 }, { "epoch": 0.05, "learning_rate": 4.969988916797335e-05, "loss": 0.9431, "step": 49466 }, { "epoch": 0.05, "learning_rate": 4.969987703363185e-05, "loss": 0.7066, "step": 49467 }, { "epoch": 0.05, "learning_rate": 4.969986489904652e-05, "loss": 1.1989, "step": 49468 }, { "epoch": 0.05, "learning_rate": 4.969985276421736e-05, "loss": 1.0192, "step": 49469 }, { "epoch": 0.05, "learning_rate": 4.9699840629144385e-05, "loss": 1.0511, "step": 49470 }, { "epoch": 0.05, "learning_rate": 4.969982849382758e-05, "loss": 0.9123, "step": 49471 }, { "epoch": 0.05, "learning_rate": 4.969981635826695e-05, "loss": 1.1574, "step": 49472 }, { "epoch": 0.05, "learning_rate": 4.969980422246249e-05, "loss": 1.0224, "step": 49473 }, { "epoch": 0.05, "learning_rate": 4.96997920864142e-05, "loss": 1.5881, "step": 49474 }, { "epoch": 0.05, "learning_rate": 4.9699779950122085e-05, "loss": 2.2991, "step": 49475 }, { "epoch": 0.05, "learning_rate": 4.969976781358615e-05, "loss": 2.1611, "step": 49476 }, { "epoch": 0.05, "learning_rate": 4.9699755676806394e-05, "loss": 2.1603, "step": 49477 }, { "epoch": 0.05, "learning_rate": 4.96997435397828e-05, "loss": 2.0936, "step": 49478 }, { "epoch": 0.05, "learning_rate": 4.9699731402515396e-05, "loss": 2.0773, "step": 49479 }, { "epoch": 0.05, "learning_rate": 4.969971926500415e-05, "loss": 2.0896, "step": 49480 }, { "epoch": 0.05, "learning_rate": 4.969970712724908e-05, "loss": 2.0648, "step": 49481 }, { "epoch": 0.05, "learning_rate": 4.96996949892502e-05, "loss": 2.0087, "step": 49482 }, { "epoch": 0.05, "learning_rate": 4.969968285100748e-05, "loss": 2.0301, "step": 49483 }, { "epoch": 0.05, "learning_rate": 4.969967071252094e-05, "loss": 1.9798, "step": 49484 }, { "epoch": 0.05, "learning_rate": 4.969965857379058e-05, "loss": 1.9174, "step": 49485 }, { "epoch": 0.05, "learning_rate": 4.9699646434816395e-05, "loss": 1.9022, "step": 49486 }, { "epoch": 0.05, "learning_rate": 4.969963429559838e-05, "loss": 1.7967, "step": 49487 }, { "epoch": 0.05, "learning_rate": 4.969962215613654e-05, "loss": 1.718, "step": 49488 }, { "epoch": 0.05, "learning_rate": 4.969961001643088e-05, "loss": 1.6496, "step": 49489 }, { "epoch": 0.05, "learning_rate": 4.9699597876481396e-05, "loss": 1.604, "step": 49490 }, { "epoch": 0.05, "learning_rate": 4.9699585736288085e-05, "loss": 1.4073, "step": 49491 }, { "epoch": 0.05, "learning_rate": 4.969957359585095e-05, "loss": 1.4813, "step": 49492 }, { "epoch": 0.05, "learning_rate": 4.9699561455169994e-05, "loss": 1.4679, "step": 49493 }, { "epoch": 0.05, "learning_rate": 4.969954931424521e-05, "loss": 1.4021, "step": 49494 }, { "epoch": 0.05, "learning_rate": 4.969953717307661e-05, "loss": 1.3697, "step": 49495 }, { "epoch": 0.05, "learning_rate": 4.969952503166419e-05, "loss": 1.4195, "step": 49496 }, { "epoch": 0.05, "learning_rate": 4.969951289000793e-05, "loss": 1.3195, "step": 49497 }, { "epoch": 0.05, "learning_rate": 4.9699500748107855e-05, "loss": 1.3799, "step": 49498 }, { "epoch": 0.05, "learning_rate": 4.969948860596396e-05, "loss": 1.3484, "step": 49499 }, { "epoch": 0.05, "learning_rate": 4.969947646357624e-05, "loss": 1.3555, "step": 49500 }, { "epoch": 0.05, "eval_loss": 1.0935750007629395, "eval_runtime": 126.1201, "eval_samples_per_second": 10.982, "eval_steps_per_second": 5.495, "step": 49500 }, { "epoch": 0.05, "learning_rate": 4.9699464320944696e-05, "loss": 1.2435, "step": 49501 }, { "epoch": 0.05, "learning_rate": 4.969945217806933e-05, "loss": 1.2915, "step": 49502 }, { "epoch": 0.05, "learning_rate": 4.969944003495014e-05, "loss": 1.3409, "step": 49503 }, { "epoch": 0.05, "learning_rate": 4.969942789158713e-05, "loss": 1.2208, "step": 49504 }, { "epoch": 0.05, "learning_rate": 4.969941574798029e-05, "loss": 1.2062, "step": 49505 }, { "epoch": 0.05, "learning_rate": 4.969940360412964e-05, "loss": 1.2802, "step": 49506 }, { "epoch": 0.05, "learning_rate": 4.969939146003516e-05, "loss": 1.4211, "step": 49507 }, { "epoch": 0.05, "learning_rate": 4.969937931569686e-05, "loss": 1.346, "step": 49508 }, { "epoch": 0.05, "learning_rate": 4.969936717111474e-05, "loss": 1.267, "step": 49509 }, { "epoch": 0.05, "learning_rate": 4.9699355026288797e-05, "loss": 1.2506, "step": 49510 }, { "epoch": 0.05, "learning_rate": 4.9699342881219036e-05, "loss": 1.3876, "step": 49511 }, { "epoch": 0.05, "learning_rate": 4.9699330735905446e-05, "loss": 1.3895, "step": 49512 }, { "epoch": 0.05, "learning_rate": 4.9699318590348045e-05, "loss": 1.3027, "step": 49513 }, { "epoch": 0.05, "learning_rate": 4.9699306444546814e-05, "loss": 1.2795, "step": 49514 }, { "epoch": 0.05, "learning_rate": 4.969929429850176e-05, "loss": 1.2022, "step": 49515 }, { "epoch": 0.05, "learning_rate": 4.969928215221289e-05, "loss": 1.1407, "step": 49516 }, { "epoch": 0.05, "learning_rate": 4.96992700056802e-05, "loss": 1.1968, "step": 49517 }, { "epoch": 0.05, "learning_rate": 4.969925785890368e-05, "loss": 1.1713, "step": 49518 }, { "epoch": 0.05, "learning_rate": 4.969924571188335e-05, "loss": 1.1054, "step": 49519 }, { "epoch": 0.05, "learning_rate": 4.96992335646192e-05, "loss": 1.1878, "step": 49520 }, { "epoch": 0.05, "learning_rate": 4.969922141711122e-05, "loss": 1.177, "step": 49521 }, { "epoch": 0.05, "learning_rate": 4.969920926935942e-05, "loss": 1.1719, "step": 49522 }, { "epoch": 0.05, "learning_rate": 4.969919712136381e-05, "loss": 1.013, "step": 49523 }, { "epoch": 0.05, "learning_rate": 4.969918497312438e-05, "loss": 1.3259, "step": 49524 }, { "epoch": 0.05, "learning_rate": 4.969917282464112e-05, "loss": 0.9604, "step": 49525 }, { "epoch": 0.05, "learning_rate": 4.969916067591405e-05, "loss": 1.1188, "step": 49526 }, { "epoch": 0.05, "learning_rate": 4.969914852694315e-05, "loss": 0.6523, "step": 49527 }, { "epoch": 0.05, "learning_rate": 4.969913637772844e-05, "loss": 0.6287, "step": 49528 }, { "epoch": 0.05, "learning_rate": 4.96991242282699e-05, "loss": 0.9732, "step": 49529 }, { "epoch": 0.05, "learning_rate": 4.969911207856755e-05, "loss": 0.8476, "step": 49530 }, { "epoch": 0.05, "learning_rate": 4.969909992862138e-05, "loss": 0.8036, "step": 49531 }, { "epoch": 0.05, "learning_rate": 4.969908777843139e-05, "loss": 0.8076, "step": 49532 }, { "epoch": 0.05, "learning_rate": 4.969907562799758e-05, "loss": 1.3709, "step": 49533 }, { "epoch": 0.05, "learning_rate": 4.969906347731995e-05, "loss": 0.9924, "step": 49534 }, { "epoch": 0.05, "learning_rate": 4.96990513263985e-05, "loss": 1.0036, "step": 49535 }, { "epoch": 0.05, "learning_rate": 4.9699039175233234e-05, "loss": 0.9827, "step": 49536 }, { "epoch": 0.05, "learning_rate": 4.969902702382415e-05, "loss": 0.681, "step": 49537 }, { "epoch": 0.05, "learning_rate": 4.969901487217124e-05, "loss": 0.7802, "step": 49538 }, { "epoch": 0.05, "learning_rate": 4.9699002720274526e-05, "loss": 1.1471, "step": 49539 }, { "epoch": 0.05, "learning_rate": 4.9698990568133986e-05, "loss": 1.2588, "step": 49540 }, { "epoch": 0.05, "learning_rate": 4.969897841574963e-05, "loss": 1.3532, "step": 49541 }, { "epoch": 0.05, "learning_rate": 4.969896626312145e-05, "loss": 1.1488, "step": 49542 }, { "epoch": 0.05, "learning_rate": 4.9698954110249457e-05, "loss": 0.9893, "step": 49543 }, { "epoch": 0.05, "learning_rate": 4.969894195713364e-05, "loss": 0.8748, "step": 49544 }, { "epoch": 0.05, "learning_rate": 4.969892980377402e-05, "loss": 1.0352, "step": 49545 }, { "epoch": 0.05, "learning_rate": 4.969891765017057e-05, "loss": 1.1608, "step": 49546 }, { "epoch": 0.05, "learning_rate": 4.9698905496323306e-05, "loss": 1.0458, "step": 49547 }, { "epoch": 0.05, "learning_rate": 4.9698893342232224e-05, "loss": 1.0557, "step": 49548 }, { "epoch": 0.05, "learning_rate": 4.9698881187897326e-05, "loss": 1.1132, "step": 49549 }, { "epoch": 0.05, "learning_rate": 4.969886903331861e-05, "loss": 1.5232, "step": 49550 }, { "epoch": 0.05, "learning_rate": 4.969885687849608e-05, "loss": 1.0261, "step": 49551 }, { "epoch": 0.05, "learning_rate": 4.969884472342973e-05, "loss": 0.9377, "step": 49552 }, { "epoch": 0.05, "learning_rate": 4.969883256811957e-05, "loss": 0.9147, "step": 49553 }, { "epoch": 0.05, "learning_rate": 4.9698820412565585e-05, "loss": 1.0346, "step": 49554 }, { "epoch": 0.05, "learning_rate": 4.9698808256767793e-05, "loss": 0.5517, "step": 49555 }, { "epoch": 0.05, "learning_rate": 4.969879610072618e-05, "loss": 0.425, "step": 49556 }, { "epoch": 0.05, "learning_rate": 4.969878394444074e-05, "loss": 0.5034, "step": 49557 }, { "epoch": 0.05, "learning_rate": 4.9698771787911504e-05, "loss": 1.1721, "step": 49558 }, { "epoch": 0.05, "learning_rate": 4.969875963113844e-05, "loss": 1.1936, "step": 49559 }, { "epoch": 0.05, "learning_rate": 4.969874747412156e-05, "loss": 0.528, "step": 49560 }, { "epoch": 0.05, "learning_rate": 4.9698735316860865e-05, "loss": 0.9329, "step": 49561 }, { "epoch": 0.05, "learning_rate": 4.9698723159356356e-05, "loss": 1.1106, "step": 49562 }, { "epoch": 0.05, "learning_rate": 4.969871100160804e-05, "loss": 1.5094, "step": 49563 }, { "epoch": 0.05, "learning_rate": 4.9698698843615895e-05, "loss": 1.0401, "step": 49564 }, { "epoch": 0.05, "learning_rate": 4.9698686685379936e-05, "loss": 0.7505, "step": 49565 }, { "epoch": 0.05, "learning_rate": 4.9698674526900166e-05, "loss": 1.1889, "step": 49566 }, { "epoch": 0.05, "learning_rate": 4.969866236817659e-05, "loss": 0.865, "step": 49567 }, { "epoch": 0.05, "learning_rate": 4.9698650209209185e-05, "loss": 0.9643, "step": 49568 }, { "epoch": 0.05, "learning_rate": 4.969863804999797e-05, "loss": 1.1951, "step": 49569 }, { "epoch": 0.05, "learning_rate": 4.969862589054294e-05, "loss": 1.1032, "step": 49570 }, { "epoch": 0.05, "learning_rate": 4.9698613730844104e-05, "loss": 1.1435, "step": 49571 }, { "epoch": 0.05, "learning_rate": 4.969860157090144e-05, "loss": 1.2125, "step": 49572 }, { "epoch": 0.05, "learning_rate": 4.969858941071497e-05, "loss": 1.2278, "step": 49573 }, { "epoch": 0.05, "learning_rate": 4.9698577250284686e-05, "loss": 1.0812, "step": 49574 }, { "epoch": 0.05, "learning_rate": 4.969856508961059e-05, "loss": 0.7089, "step": 49575 }, { "epoch": 0.05, "learning_rate": 4.969855292869268e-05, "loss": 0.9869, "step": 49576 }, { "epoch": 0.05, "learning_rate": 4.9698540767530944e-05, "loss": 1.1345, "step": 49577 }, { "epoch": 0.05, "learning_rate": 4.969852860612541e-05, "loss": 1.4024, "step": 49578 }, { "epoch": 0.05, "learning_rate": 4.969851644447605e-05, "loss": 1.3261, "step": 49579 }, { "epoch": 0.05, "learning_rate": 4.9698504282582886e-05, "loss": 1.1558, "step": 49580 }, { "epoch": 0.05, "learning_rate": 4.9698492120445907e-05, "loss": 1.1577, "step": 49581 }, { "epoch": 0.05, "learning_rate": 4.969847995806512e-05, "loss": 1.2197, "step": 49582 }, { "epoch": 0.05, "learning_rate": 4.9698467795440504e-05, "loss": 0.961, "step": 49583 }, { "epoch": 0.05, "learning_rate": 4.969845563257208e-05, "loss": 0.543, "step": 49584 }, { "epoch": 0.05, "learning_rate": 4.9698443469459855e-05, "loss": 1.2644, "step": 49585 }, { "epoch": 0.05, "learning_rate": 4.969843130610381e-05, "loss": 0.9875, "step": 49586 }, { "epoch": 0.05, "learning_rate": 4.9698419142503946e-05, "loss": 1.0654, "step": 49587 }, { "epoch": 0.05, "learning_rate": 4.969840697866028e-05, "loss": 1.1196, "step": 49588 }, { "epoch": 0.05, "learning_rate": 4.9698394814572804e-05, "loss": 1.2648, "step": 49589 }, { "epoch": 0.05, "learning_rate": 4.969838265024151e-05, "loss": 1.5457, "step": 49590 }, { "epoch": 0.05, "learning_rate": 4.96983704856664e-05, "loss": 1.0869, "step": 49591 }, { "epoch": 0.05, "learning_rate": 4.9698358320847485e-05, "loss": 0.7752, "step": 49592 }, { "epoch": 0.05, "learning_rate": 4.9698346155784745e-05, "loss": 1.1288, "step": 49593 }, { "epoch": 0.05, "learning_rate": 4.969833399047821e-05, "loss": 0.8041, "step": 49594 }, { "epoch": 0.05, "learning_rate": 4.969832182492785e-05, "loss": 1.443, "step": 49595 }, { "epoch": 0.05, "learning_rate": 4.9698309659133693e-05, "loss": 1.3066, "step": 49596 }, { "epoch": 0.05, "learning_rate": 4.9698297493095714e-05, "loss": 0.9796, "step": 49597 }, { "epoch": 0.05, "learning_rate": 4.969828532681393e-05, "loss": 0.4759, "step": 49598 }, { "epoch": 0.05, "learning_rate": 4.969827316028833e-05, "loss": 0.5464, "step": 49599 }, { "epoch": 0.05, "learning_rate": 4.969826099351892e-05, "loss": 0.86, "step": 49600 }, { "epoch": 0.05, "learning_rate": 4.96982488265057e-05, "loss": 1.992, "step": 49601 }, { "epoch": 0.05, "learning_rate": 4.969823665924867e-05, "loss": 0.6496, "step": 49602 }, { "epoch": 0.05, "learning_rate": 4.969822449174783e-05, "loss": 1.3965, "step": 49603 }, { "epoch": 0.05, "learning_rate": 4.969821232400318e-05, "loss": 0.9961, "step": 49604 }, { "epoch": 0.05, "learning_rate": 4.969820015601472e-05, "loss": 0.3599, "step": 49605 }, { "epoch": 0.05, "learning_rate": 4.969818798778244e-05, "loss": 0.2867, "step": 49606 }, { "epoch": 0.05, "learning_rate": 4.969817581930636e-05, "loss": 1.0636, "step": 49607 }, { "epoch": 0.05, "learning_rate": 4.9698163650586474e-05, "loss": 0.9888, "step": 49608 }, { "epoch": 0.05, "learning_rate": 4.969815148162277e-05, "loss": 0.5891, "step": 49609 }, { "epoch": 0.05, "learning_rate": 4.969813931241525e-05, "loss": 0.3398, "step": 49610 }, { "epoch": 0.05, "learning_rate": 4.9698127142963933e-05, "loss": 1.2536, "step": 49611 }, { "epoch": 0.05, "learning_rate": 4.9698114973268804e-05, "loss": 1.3141, "step": 49612 }, { "epoch": 0.05, "learning_rate": 4.969810280332986e-05, "loss": 0.6533, "step": 49613 }, { "epoch": 0.05, "learning_rate": 4.969809063314711e-05, "loss": 0.412, "step": 49614 }, { "epoch": 0.05, "learning_rate": 4.969807846272055e-05, "loss": 0.4006, "step": 49615 }, { "epoch": 0.05, "learning_rate": 4.969806629205018e-05, "loss": 1.0651, "step": 49616 }, { "epoch": 0.05, "learning_rate": 4.9698054121136e-05, "loss": 0.684, "step": 49617 }, { "epoch": 0.05, "learning_rate": 4.969804194997803e-05, "loss": 0.525, "step": 49618 }, { "epoch": 0.05, "learning_rate": 4.969802977857623e-05, "loss": 0.9319, "step": 49619 }, { "epoch": 0.05, "learning_rate": 4.969801760693063e-05, "loss": 0.7828, "step": 49620 }, { "epoch": 0.05, "learning_rate": 4.9698005435041215e-05, "loss": 1.0305, "step": 49621 }, { "epoch": 0.05, "learning_rate": 4.9697993262907993e-05, "loss": 0.4524, "step": 49622 }, { "epoch": 0.05, "learning_rate": 4.969798109053097e-05, "loss": 0.882, "step": 49623 }, { "epoch": 0.05, "learning_rate": 4.969796891791013e-05, "loss": 1.0141, "step": 49624 }, { "epoch": 0.05, "learning_rate": 4.969795674504549e-05, "loss": 0.7707, "step": 49625 }, { "epoch": 0.05, "learning_rate": 4.969794457193704e-05, "loss": 1.0838, "step": 49626 }, { "epoch": 0.05, "learning_rate": 4.969793239858478e-05, "loss": 1.075, "step": 49627 }, { "epoch": 0.05, "learning_rate": 4.9697920224988714e-05, "loss": 1.1766, "step": 49628 }, { "epoch": 0.05, "learning_rate": 4.9697908051148836e-05, "loss": 0.8916, "step": 49629 }, { "epoch": 0.05, "learning_rate": 4.9697895877065155e-05, "loss": 1.1048, "step": 49630 }, { "epoch": 0.05, "learning_rate": 4.969788370273767e-05, "loss": 1.28, "step": 49631 }, { "epoch": 0.05, "learning_rate": 4.9697871528166376e-05, "loss": 0.9347, "step": 49632 }, { "epoch": 0.05, "learning_rate": 4.9697859353351265e-05, "loss": 0.3774, "step": 49633 }, { "epoch": 0.05, "learning_rate": 4.969784717829236e-05, "loss": 0.4556, "step": 49634 }, { "epoch": 0.05, "learning_rate": 4.969783500298964e-05, "loss": 1.1809, "step": 49635 }, { "epoch": 0.05, "learning_rate": 4.969782282744312e-05, "loss": 0.9175, "step": 49636 }, { "epoch": 0.05, "learning_rate": 4.969781065165279e-05, "loss": 0.9305, "step": 49637 }, { "epoch": 0.05, "learning_rate": 4.9697798475618654e-05, "loss": 1.0161, "step": 49638 }, { "epoch": 0.05, "learning_rate": 4.969778629934071e-05, "loss": 0.9165, "step": 49639 }, { "epoch": 0.05, "learning_rate": 4.9697774122818964e-05, "loss": 1.0382, "step": 49640 }, { "epoch": 0.05, "learning_rate": 4.9697761946053407e-05, "loss": 1.1913, "step": 49641 }, { "epoch": 0.05, "learning_rate": 4.9697749769044046e-05, "loss": 1.1504, "step": 49642 }, { "epoch": 0.05, "learning_rate": 4.969773759179088e-05, "loss": 0.8729, "step": 49643 }, { "epoch": 0.05, "learning_rate": 4.96977254142939e-05, "loss": 0.1837, "step": 49644 }, { "epoch": 0.05, "learning_rate": 4.969771323655313e-05, "loss": 0.2127, "step": 49645 }, { "epoch": 0.05, "learning_rate": 4.969770105856855e-05, "loss": 0.9363, "step": 49646 }, { "epoch": 0.05, "learning_rate": 4.969768888034016e-05, "loss": 1.2882, "step": 49647 }, { "epoch": 0.05, "learning_rate": 4.969767670186797e-05, "loss": 1.1148, "step": 49648 }, { "epoch": 0.05, "learning_rate": 4.969766452315197e-05, "loss": 1.0263, "step": 49649 }, { "epoch": 0.05, "learning_rate": 4.969765234419217e-05, "loss": 1.034, "step": 49650 }, { "epoch": 0.05, "learning_rate": 4.969764016498856e-05, "loss": 0.8938, "step": 49651 }, { "epoch": 0.05, "learning_rate": 4.969762798554115e-05, "loss": 0.8941, "step": 49652 }, { "epoch": 0.05, "learning_rate": 4.969761580584993e-05, "loss": 0.7833, "step": 49653 }, { "epoch": 0.05, "learning_rate": 4.9697603625914906e-05, "loss": 1.0886, "step": 49654 }, { "epoch": 0.05, "learning_rate": 4.969759144573608e-05, "loss": 0.9649, "step": 49655 }, { "epoch": 0.05, "learning_rate": 4.969757926531345e-05, "loss": 1.1394, "step": 49656 }, { "epoch": 0.05, "learning_rate": 4.969756708464702e-05, "loss": 0.7321, "step": 49657 }, { "epoch": 0.05, "learning_rate": 4.969755490373678e-05, "loss": 0.9447, "step": 49658 }, { "epoch": 0.05, "learning_rate": 4.969754272258274e-05, "loss": 1.8479, "step": 49659 }, { "epoch": 0.05, "learning_rate": 4.9697530541184886e-05, "loss": 1.1168, "step": 49660 }, { "epoch": 0.05, "learning_rate": 4.969751835954324e-05, "loss": 0.9562, "step": 49661 }, { "epoch": 0.05, "learning_rate": 4.969750617765779e-05, "loss": 1.235, "step": 49662 }, { "epoch": 0.05, "learning_rate": 4.9697493995528533e-05, "loss": 1.2884, "step": 49663 }, { "epoch": 0.05, "learning_rate": 4.969748181315547e-05, "loss": 1.054, "step": 49664 }, { "epoch": 0.05, "learning_rate": 4.969746963053861e-05, "loss": 1.2897, "step": 49665 }, { "epoch": 0.05, "learning_rate": 4.9697457447677945e-05, "loss": 1.0387, "step": 49666 }, { "epoch": 0.05, "learning_rate": 4.969744526457347e-05, "loss": 1.0432, "step": 49667 }, { "epoch": 0.05, "learning_rate": 4.96974330812252e-05, "loss": 1.0432, "step": 49668 }, { "epoch": 0.05, "learning_rate": 4.969742089763313e-05, "loss": 1.007, "step": 49669 }, { "epoch": 0.05, "learning_rate": 4.969740871379726e-05, "loss": 1.0385, "step": 49670 }, { "epoch": 0.05, "learning_rate": 4.969739652971758e-05, "loss": 1.0894, "step": 49671 }, { "epoch": 0.05, "learning_rate": 4.9697384345394095e-05, "loss": 0.8759, "step": 49672 }, { "epoch": 0.05, "learning_rate": 4.96973721608268e-05, "loss": 0.8847, "step": 49673 }, { "epoch": 0.05, "learning_rate": 4.969735997601572e-05, "loss": 1.0698, "step": 49674 }, { "epoch": 0.05, "learning_rate": 4.9697347790960834e-05, "loss": 1.1048, "step": 49675 }, { "epoch": 0.05, "learning_rate": 4.9697335605662146e-05, "loss": 0.9947, "step": 49676 }, { "epoch": 0.05, "learning_rate": 4.9697323420119654e-05, "loss": 1.0487, "step": 49677 }, { "epoch": 0.05, "learning_rate": 4.9697311234333365e-05, "loss": 0.3078, "step": 49678 }, { "epoch": 0.05, "learning_rate": 4.969729904830327e-05, "loss": 0.1891, "step": 49679 }, { "epoch": 0.05, "learning_rate": 4.969728686202937e-05, "loss": 0.1652, "step": 49680 }, { "epoch": 0.05, "learning_rate": 4.9697274675511674e-05, "loss": 0.14, "step": 49681 }, { "epoch": 0.05, "learning_rate": 4.969726248875018e-05, "loss": 0.1433, "step": 49682 }, { "epoch": 0.05, "learning_rate": 4.969725030174488e-05, "loss": 0.1577, "step": 49683 }, { "epoch": 0.05, "learning_rate": 4.969723811449578e-05, "loss": 0.1872, "step": 49684 }, { "epoch": 0.05, "learning_rate": 4.969722592700288e-05, "loss": 0.2934, "step": 49685 }, { "epoch": 0.05, "learning_rate": 4.969721373926618e-05, "loss": 0.2357, "step": 49686 }, { "epoch": 0.05, "learning_rate": 4.969720155128568e-05, "loss": 0.4073, "step": 49687 }, { "epoch": 0.05, "learning_rate": 4.969718936306138e-05, "loss": 1.0075, "step": 49688 }, { "epoch": 0.05, "learning_rate": 4.969717717459328e-05, "loss": 1.1764, "step": 49689 }, { "epoch": 0.05, "learning_rate": 4.969716498588137e-05, "loss": 0.9801, "step": 49690 }, { "epoch": 0.05, "learning_rate": 4.9697152796925674e-05, "loss": 1.0481, "step": 49691 }, { "epoch": 0.05, "learning_rate": 4.969714060772617e-05, "loss": 0.7534, "step": 49692 }, { "epoch": 0.05, "learning_rate": 4.9697128418282866e-05, "loss": 0.7131, "step": 49693 }, { "epoch": 0.05, "learning_rate": 4.969711622859576e-05, "loss": 0.9923, "step": 49694 }, { "epoch": 0.05, "learning_rate": 4.9697104038664865e-05, "loss": 0.8353, "step": 49695 }, { "epoch": 0.05, "learning_rate": 4.969709184849016e-05, "loss": 0.942, "step": 49696 }, { "epoch": 0.05, "learning_rate": 4.9697079658071665e-05, "loss": 0.8323, "step": 49697 }, { "epoch": 0.05, "learning_rate": 4.969706746740937e-05, "loss": 0.8779, "step": 49698 }, { "epoch": 0.05, "learning_rate": 4.9697055276503266e-05, "loss": 1.4664, "step": 49699 }, { "epoch": 0.05, "learning_rate": 4.9697043085353375e-05, "loss": 0.6431, "step": 49700 }, { "epoch": 0.05, "learning_rate": 4.9697030893959674e-05, "loss": 0.9939, "step": 49701 }, { "epoch": 0.05, "learning_rate": 4.9697018702322184e-05, "loss": 1.1597, "step": 49702 }, { "epoch": 0.05, "learning_rate": 4.969700651044089e-05, "loss": 0.8568, "step": 49703 }, { "epoch": 0.05, "learning_rate": 4.96969943183158e-05, "loss": 0.9661, "step": 49704 }, { "epoch": 0.05, "learning_rate": 4.9696982125946914e-05, "loss": 1.0379, "step": 49705 }, { "epoch": 0.05, "learning_rate": 4.9696969933334224e-05, "loss": 0.8297, "step": 49706 }, { "epoch": 0.05, "learning_rate": 4.969695774047773e-05, "loss": 0.6301, "step": 49707 }, { "epoch": 0.05, "learning_rate": 4.969694554737745e-05, "loss": 1.1226, "step": 49708 }, { "epoch": 0.05, "learning_rate": 4.969693335403337e-05, "loss": 0.936, "step": 49709 }, { "epoch": 0.05, "learning_rate": 4.9696921160445495e-05, "loss": 1.0519, "step": 49710 }, { "epoch": 0.05, "learning_rate": 4.969690896661382e-05, "loss": 1.3315, "step": 49711 }, { "epoch": 0.05, "learning_rate": 4.9696896772538335e-05, "loss": 1.2498, "step": 49712 }, { "epoch": 0.05, "learning_rate": 4.969688457821907e-05, "loss": 0.83, "step": 49713 }, { "epoch": 0.05, "learning_rate": 4.9696872383656e-05, "loss": 0.7352, "step": 49714 }, { "epoch": 0.05, "learning_rate": 4.969686018884914e-05, "loss": 0.9151, "step": 49715 }, { "epoch": 0.05, "learning_rate": 4.9696847993798465e-05, "loss": 0.8976, "step": 49716 }, { "epoch": 0.05, "learning_rate": 4.969683579850401e-05, "loss": 1.1012, "step": 49717 }, { "epoch": 0.05, "learning_rate": 4.9696823602965755e-05, "loss": 1.2216, "step": 49718 }, { "epoch": 0.05, "learning_rate": 4.96968114071837e-05, "loss": 1.094, "step": 49719 }, { "epoch": 0.05, "learning_rate": 4.969679921115785e-05, "loss": 0.639, "step": 49720 }, { "epoch": 0.05, "learning_rate": 4.96967870148882e-05, "loss": 0.5589, "step": 49721 }, { "epoch": 0.05, "learning_rate": 4.969677481837476e-05, "loss": 1.0153, "step": 49722 }, { "epoch": 0.05, "learning_rate": 4.969676262161752e-05, "loss": 1.3416, "step": 49723 }, { "epoch": 0.05, "learning_rate": 4.969675042461649e-05, "loss": 0.9628, "step": 49724 }, { "epoch": 0.05, "learning_rate": 4.969673822737166e-05, "loss": 0.9555, "step": 49725 }, { "epoch": 0.05, "learning_rate": 4.969672602988304e-05, "loss": 1.11, "step": 49726 }, { "epoch": 0.05, "learning_rate": 4.969671383215061e-05, "loss": 0.9237, "step": 49727 }, { "epoch": 0.05, "learning_rate": 4.969670163417439e-05, "loss": 1.0586, "step": 49728 }, { "epoch": 0.05, "learning_rate": 4.969668943595438e-05, "loss": 1.1072, "step": 49729 }, { "epoch": 0.05, "learning_rate": 4.969667723749057e-05, "loss": 0.9803, "step": 49730 }, { "epoch": 0.05, "learning_rate": 4.969666503878296e-05, "loss": 0.8866, "step": 49731 }, { "epoch": 0.05, "learning_rate": 4.969665283983157e-05, "loss": 0.9125, "step": 49732 }, { "epoch": 0.05, "learning_rate": 4.969664064063637e-05, "loss": 0.8739, "step": 49733 }, { "epoch": 0.05, "learning_rate": 4.969662844119738e-05, "loss": 1.2365, "step": 49734 }, { "epoch": 0.05, "learning_rate": 4.9696616241514604e-05, "loss": 1.0378, "step": 49735 }, { "epoch": 0.05, "learning_rate": 4.969660404158802e-05, "loss": 0.7777, "step": 49736 }, { "epoch": 0.05, "learning_rate": 4.969659184141765e-05, "loss": 0.9673, "step": 49737 }, { "epoch": 0.05, "learning_rate": 4.969657964100348e-05, "loss": 1.1977, "step": 49738 }, { "epoch": 0.05, "learning_rate": 4.9696567440345525e-05, "loss": 1.1973, "step": 49739 }, { "epoch": 0.05, "learning_rate": 4.969655523944377e-05, "loss": 0.7573, "step": 49740 }, { "epoch": 0.05, "learning_rate": 4.969654303829822e-05, "loss": 1.9593, "step": 49741 }, { "epoch": 0.05, "learning_rate": 4.969653083690888e-05, "loss": 1.21, "step": 49742 }, { "epoch": 0.05, "learning_rate": 4.9696518635275744e-05, "loss": 1.1018, "step": 49743 }, { "epoch": 0.05, "learning_rate": 4.969650643339882e-05, "loss": 1.1102, "step": 49744 }, { "epoch": 0.05, "learning_rate": 4.969649423127809e-05, "loss": 1.3353, "step": 49745 }, { "epoch": 0.05, "learning_rate": 4.969648202891357e-05, "loss": 1.1271, "step": 49746 }, { "epoch": 0.05, "learning_rate": 4.969646982630526e-05, "loss": 1.359, "step": 49747 }, { "epoch": 0.05, "learning_rate": 4.9696457623453156e-05, "loss": 1.3025, "step": 49748 }, { "epoch": 0.05, "learning_rate": 4.969644542035726e-05, "loss": 1.4284, "step": 49749 }, { "epoch": 0.05, "learning_rate": 4.969643321701757e-05, "loss": 0.6043, "step": 49750 }, { "epoch": 0.05, "learning_rate": 4.969642101343409e-05, "loss": 0.2466, "step": 49751 }, { "epoch": 0.05, "learning_rate": 4.969640880960681e-05, "loss": 0.3081, "step": 49752 }, { "epoch": 0.05, "learning_rate": 4.9696396605535746e-05, "loss": 0.2278, "step": 49753 }, { "epoch": 0.05, "learning_rate": 4.969638440122088e-05, "loss": 0.0813, "step": 49754 }, { "epoch": 0.05, "learning_rate": 4.969637219666223e-05, "loss": 0.1634, "step": 49755 }, { "epoch": 0.05, "learning_rate": 4.9696359991859795e-05, "loss": 1.022, "step": 49756 }, { "epoch": 0.05, "learning_rate": 4.9696347786813556e-05, "loss": 0.9686, "step": 49757 }, { "epoch": 0.05, "learning_rate": 4.969633558152353e-05, "loss": 1.0521, "step": 49758 }, { "epoch": 0.05, "learning_rate": 4.969632337598971e-05, "loss": 1.3703, "step": 49759 }, { "epoch": 0.05, "learning_rate": 4.9696311170212095e-05, "loss": 0.8926, "step": 49760 }, { "epoch": 0.05, "learning_rate": 4.969629896419069e-05, "loss": 0.7261, "step": 49761 }, { "epoch": 0.05, "learning_rate": 4.96962867579255e-05, "loss": 0.5952, "step": 49762 }, { "epoch": 0.05, "learning_rate": 4.9696274551416513e-05, "loss": 0.5858, "step": 49763 }, { "epoch": 0.05, "learning_rate": 4.9696262344663734e-05, "loss": 0.4573, "step": 49764 }, { "epoch": 0.05, "learning_rate": 4.9696250137667164e-05, "loss": 0.6311, "step": 49765 }, { "epoch": 0.05, "learning_rate": 4.9696237930426805e-05, "loss": 0.8575, "step": 49766 }, { "epoch": 0.05, "learning_rate": 4.9696225722942656e-05, "loss": 0.7521, "step": 49767 }, { "epoch": 0.05, "learning_rate": 4.969621351521472e-05, "loss": 1.2539, "step": 49768 }, { "epoch": 0.05, "learning_rate": 4.969620130724298e-05, "loss": 1.1484, "step": 49769 }, { "epoch": 0.05, "learning_rate": 4.9696189099027466e-05, "loss": 1.0434, "step": 49770 }, { "epoch": 0.05, "learning_rate": 4.9696176890568145e-05, "loss": 0.9848, "step": 49771 }, { "epoch": 0.05, "learning_rate": 4.969616468186504e-05, "loss": 1.086, "step": 49772 }, { "epoch": 0.05, "learning_rate": 4.9696152472918155e-05, "loss": 0.9542, "step": 49773 }, { "epoch": 0.05, "learning_rate": 4.969614026372748e-05, "loss": 0.7507, "step": 49774 }, { "epoch": 0.05, "learning_rate": 4.9696128054293e-05, "loss": 0.9552, "step": 49775 }, { "epoch": 0.05, "learning_rate": 4.969611584461473e-05, "loss": 1.1592, "step": 49776 }, { "epoch": 0.05, "learning_rate": 4.969610363469268e-05, "loss": 1.1277, "step": 49777 }, { "epoch": 0.05, "learning_rate": 4.9696091424526844e-05, "loss": 0.7748, "step": 49778 }, { "epoch": 0.05, "learning_rate": 4.9696079214117206e-05, "loss": 1.0627, "step": 49779 }, { "epoch": 0.05, "learning_rate": 4.969606700346379e-05, "loss": 0.9632, "step": 49780 }, { "epoch": 0.05, "learning_rate": 4.969605479256658e-05, "loss": 1.0756, "step": 49781 }, { "epoch": 0.05, "learning_rate": 4.969604258142558e-05, "loss": 1.0718, "step": 49782 }, { "epoch": 0.05, "learning_rate": 4.969603037004079e-05, "loss": 1.2339, "step": 49783 }, { "epoch": 0.05, "learning_rate": 4.969601815841222e-05, "loss": 2.156, "step": 49784 }, { "epoch": 0.05, "learning_rate": 4.969600594653985e-05, "loss": 1.696, "step": 49785 }, { "epoch": 0.05, "learning_rate": 4.96959937344237e-05, "loss": 1.4737, "step": 49786 }, { "epoch": 0.05, "learning_rate": 4.9695981522063764e-05, "loss": 1.8363, "step": 49787 }, { "epoch": 0.05, "learning_rate": 4.9695969309460026e-05, "loss": 1.5532, "step": 49788 }, { "epoch": 0.05, "learning_rate": 4.969595709661251e-05, "loss": 0.8044, "step": 49789 }, { "epoch": 0.05, "learning_rate": 4.969594488352121e-05, "loss": 1.0397, "step": 49790 }, { "epoch": 0.05, "learning_rate": 4.969593267018611e-05, "loss": 1.0725, "step": 49791 }, { "epoch": 0.05, "learning_rate": 4.9695920456607226e-05, "loss": 1.0724, "step": 49792 }, { "epoch": 0.05, "learning_rate": 4.9695908242784554e-05, "loss": 1.117, "step": 49793 }, { "epoch": 0.05, "learning_rate": 4.96958960287181e-05, "loss": 0.9318, "step": 49794 }, { "epoch": 0.05, "learning_rate": 4.9695883814407854e-05, "loss": 1.3754, "step": 49795 }, { "epoch": 0.05, "learning_rate": 4.969587159985382e-05, "loss": 1.1107, "step": 49796 }, { "epoch": 0.05, "learning_rate": 4.9695859385056e-05, "loss": 1.9901, "step": 49797 }, { "epoch": 0.05, "learning_rate": 4.9695847170014395e-05, "loss": 1.0514, "step": 49798 }, { "epoch": 0.05, "learning_rate": 4.9695834954729005e-05, "loss": 0.8561, "step": 49799 }, { "epoch": 0.05, "learning_rate": 4.969582273919982e-05, "loss": 2.4102, "step": 49800 }, { "epoch": 0.05, "learning_rate": 4.969581052342685e-05, "loss": 0.7532, "step": 49801 }, { "epoch": 0.05, "learning_rate": 4.96957983074101e-05, "loss": 1.2143, "step": 49802 }, { "epoch": 0.05, "learning_rate": 4.9695786091149555e-05, "loss": 0.8929, "step": 49803 }, { "epoch": 0.05, "learning_rate": 4.969577387464523e-05, "loss": 0.8831, "step": 49804 }, { "epoch": 0.05, "learning_rate": 4.969576165789712e-05, "loss": 1.2123, "step": 49805 }, { "epoch": 0.05, "learning_rate": 4.969574944090521e-05, "loss": 1.3749, "step": 49806 }, { "epoch": 0.05, "learning_rate": 4.969573722366953e-05, "loss": 0.9374, "step": 49807 }, { "epoch": 0.05, "learning_rate": 4.9695725006190064e-05, "loss": 0.8615, "step": 49808 }, { "epoch": 0.05, "learning_rate": 4.9695712788466804e-05, "loss": 0.9641, "step": 49809 }, { "epoch": 0.05, "learning_rate": 4.969570057049976e-05, "loss": 0.9, "step": 49810 }, { "epoch": 0.05, "learning_rate": 4.9695688352288924e-05, "loss": 0.5106, "step": 49811 }, { "epoch": 0.05, "learning_rate": 4.9695676133834316e-05, "loss": 0.6764, "step": 49812 }, { "epoch": 0.05, "learning_rate": 4.969566391513591e-05, "loss": 0.9302, "step": 49813 }, { "epoch": 0.05, "learning_rate": 4.969565169619373e-05, "loss": 0.9687, "step": 49814 }, { "epoch": 0.05, "learning_rate": 4.969563947700776e-05, "loss": 1.0153, "step": 49815 }, { "epoch": 0.05, "learning_rate": 4.969562725757801e-05, "loss": 1.0214, "step": 49816 }, { "epoch": 0.05, "learning_rate": 4.969561503790447e-05, "loss": 1.0197, "step": 49817 }, { "epoch": 0.05, "learning_rate": 4.969560281798715e-05, "loss": 1.0999, "step": 49818 }, { "epoch": 0.05, "learning_rate": 4.9695590597826034e-05, "loss": 1.0691, "step": 49819 }, { "epoch": 0.05, "learning_rate": 4.969557837742115e-05, "loss": 0.8658, "step": 49820 }, { "epoch": 0.05, "learning_rate": 4.969556615677247e-05, "loss": 1.2161, "step": 49821 }, { "epoch": 0.05, "learning_rate": 4.9695553935880006e-05, "loss": 0.9085, "step": 49822 }, { "epoch": 0.05, "learning_rate": 4.969554171474376e-05, "loss": 1.0172, "step": 49823 }, { "epoch": 0.05, "learning_rate": 4.969552949336373e-05, "loss": 0.9024, "step": 49824 }, { "epoch": 0.05, "learning_rate": 4.969551727173992e-05, "loss": 1.0045, "step": 49825 }, { "epoch": 0.05, "learning_rate": 4.969550504987232e-05, "loss": 0.9776, "step": 49826 }, { "epoch": 0.05, "learning_rate": 4.969549282776094e-05, "loss": 1.1668, "step": 49827 }, { "epoch": 0.05, "learning_rate": 4.969548060540578e-05, "loss": 1.0248, "step": 49828 }, { "epoch": 0.05, "learning_rate": 4.969546838280683e-05, "loss": 0.8696, "step": 49829 }, { "epoch": 0.05, "learning_rate": 4.9695456159964095e-05, "loss": 0.9332, "step": 49830 }, { "epoch": 0.05, "learning_rate": 4.9695443936877586e-05, "loss": 1.1245, "step": 49831 }, { "epoch": 0.05, "learning_rate": 4.969543171354729e-05, "loss": 1.1271, "step": 49832 }, { "epoch": 0.05, "learning_rate": 4.9695419489973205e-05, "loss": 0.9743, "step": 49833 }, { "epoch": 0.05, "learning_rate": 4.969540726615534e-05, "loss": 0.9047, "step": 49834 }, { "epoch": 0.05, "learning_rate": 4.9695395042093694e-05, "loss": 1.1996, "step": 49835 }, { "epoch": 0.05, "learning_rate": 4.969538281778827e-05, "loss": 1.2299, "step": 49836 }, { "epoch": 0.05, "learning_rate": 4.969537059323906e-05, "loss": 1.096, "step": 49837 }, { "epoch": 0.05, "learning_rate": 4.969535836844607e-05, "loss": 1.0143, "step": 49838 }, { "epoch": 0.05, "learning_rate": 4.9695346143409296e-05, "loss": 0.848, "step": 49839 }, { "epoch": 0.05, "learning_rate": 4.9695333918128734e-05, "loss": 0.9036, "step": 49840 }, { "epoch": 0.05, "learning_rate": 4.969532169260439e-05, "loss": 1.1232, "step": 49841 }, { "epoch": 0.05, "learning_rate": 4.969530946683627e-05, "loss": 1.0361, "step": 49842 }, { "epoch": 0.05, "learning_rate": 4.9695297240824366e-05, "loss": 1.008, "step": 49843 }, { "epoch": 0.05, "learning_rate": 4.9695285014568686e-05, "loss": 1.0287, "step": 49844 }, { "epoch": 0.05, "learning_rate": 4.9695272788069224e-05, "loss": 0.9913, "step": 49845 }, { "epoch": 0.05, "learning_rate": 4.969526056132597e-05, "loss": 1.0735, "step": 49846 }, { "epoch": 0.05, "learning_rate": 4.969524833433895e-05, "loss": 0.9346, "step": 49847 }, { "epoch": 0.05, "learning_rate": 4.969523610710814e-05, "loss": 0.8704, "step": 49848 }, { "epoch": 0.05, "learning_rate": 4.9695223879633546e-05, "loss": 0.936, "step": 49849 }, { "epoch": 0.05, "learning_rate": 4.9695211651915176e-05, "loss": 0.9723, "step": 49850 }, { "epoch": 0.05, "learning_rate": 4.969519942395302e-05, "loss": 0.909, "step": 49851 }, { "epoch": 0.05, "learning_rate": 4.9695187195747094e-05, "loss": 1.122, "step": 49852 }, { "epoch": 0.05, "learning_rate": 4.969517496729738e-05, "loss": 0.9545, "step": 49853 }, { "epoch": 0.05, "learning_rate": 4.969516273860388e-05, "loss": 0.8646, "step": 49854 }, { "epoch": 0.05, "learning_rate": 4.969515050966662e-05, "loss": 1.073, "step": 49855 }, { "epoch": 0.05, "learning_rate": 4.9695138280485564e-05, "loss": 1.1728, "step": 49856 }, { "epoch": 0.05, "learning_rate": 4.969512605106073e-05, "loss": 0.7787, "step": 49857 }, { "epoch": 0.05, "learning_rate": 4.9695113821392115e-05, "loss": 1.2445, "step": 49858 }, { "epoch": 0.05, "learning_rate": 4.969510159147972e-05, "loss": 1.0143, "step": 49859 }, { "epoch": 0.05, "learning_rate": 4.969508936132355e-05, "loss": 1.0043, "step": 49860 }, { "epoch": 0.05, "learning_rate": 4.9695077130923594e-05, "loss": 0.8929, "step": 49861 }, { "epoch": 0.05, "learning_rate": 4.9695064900279864e-05, "loss": 0.9368, "step": 49862 }, { "epoch": 0.05, "learning_rate": 4.969505266939236e-05, "loss": 0.6908, "step": 49863 }, { "epoch": 0.05, "learning_rate": 4.969504043826106e-05, "loss": 0.9316, "step": 49864 }, { "epoch": 0.05, "learning_rate": 4.9695028206885996e-05, "loss": 0.9747, "step": 49865 }, { "epoch": 0.05, "learning_rate": 4.969501597526715e-05, "loss": 1.0985, "step": 49866 }, { "epoch": 0.05, "learning_rate": 4.9695003743404523e-05, "loss": 1.1519, "step": 49867 }, { "epoch": 0.05, "learning_rate": 4.9694991511298116e-05, "loss": 0.9294, "step": 49868 }, { "epoch": 0.05, "learning_rate": 4.9694979278947926e-05, "loss": 1.4085, "step": 49869 }, { "epoch": 0.05, "learning_rate": 4.969496704635397e-05, "loss": 1.0229, "step": 49870 }, { "epoch": 0.05, "learning_rate": 4.9694954813516225e-05, "loss": 0.6884, "step": 49871 }, { "epoch": 0.05, "learning_rate": 4.969494258043471e-05, "loss": 0.4123, "step": 49872 }, { "epoch": 0.05, "learning_rate": 4.969493034710941e-05, "loss": 0.3302, "step": 49873 }, { "epoch": 0.05, "learning_rate": 4.969491811354034e-05, "loss": 0.3047, "step": 49874 }, { "epoch": 0.05, "learning_rate": 4.969490587972748e-05, "loss": 0.3815, "step": 49875 }, { "epoch": 0.05, "learning_rate": 4.969489364567086e-05, "loss": 0.4077, "step": 49876 }, { "epoch": 0.05, "learning_rate": 4.969488141137045e-05, "loss": 0.3996, "step": 49877 }, { "epoch": 0.05, "learning_rate": 4.9694869176826265e-05, "loss": 0.3707, "step": 49878 }, { "epoch": 0.05, "learning_rate": 4.96948569420383e-05, "loss": 0.2931, "step": 49879 }, { "epoch": 0.05, "learning_rate": 4.969484470700656e-05, "loss": 0.3841, "step": 49880 }, { "epoch": 0.05, "learning_rate": 4.9694832471731044e-05, "loss": 0.3155, "step": 49881 }, { "epoch": 0.05, "learning_rate": 4.9694820236211745e-05, "loss": 0.263, "step": 49882 }, { "epoch": 0.05, "learning_rate": 4.9694808000448676e-05, "loss": 0.2912, "step": 49883 }, { "epoch": 0.05, "learning_rate": 4.9694795764441824e-05, "loss": 0.2952, "step": 49884 }, { "epoch": 0.05, "learning_rate": 4.9694783528191203e-05, "loss": 0.2818, "step": 49885 }, { "epoch": 0.05, "learning_rate": 4.9694771291696807e-05, "loss": 0.3286, "step": 49886 }, { "epoch": 0.05, "learning_rate": 4.969475905495863e-05, "loss": 0.522, "step": 49887 }, { "epoch": 0.05, "learning_rate": 4.969474681797667e-05, "loss": 0.4233, "step": 49888 }, { "epoch": 0.05, "learning_rate": 4.9694734580750946e-05, "loss": 0.3557, "step": 49889 }, { "epoch": 0.05, "learning_rate": 4.9694722343281444e-05, "loss": 0.2749, "step": 49890 }, { "epoch": 0.05, "learning_rate": 4.9694710105568153e-05, "loss": 0.1949, "step": 49891 }, { "epoch": 0.05, "learning_rate": 4.96946978676111e-05, "loss": 0.2091, "step": 49892 }, { "epoch": 0.05, "learning_rate": 4.9694685629410264e-05, "loss": 0.2097, "step": 49893 }, { "epoch": 0.05, "learning_rate": 4.9694673390965665e-05, "loss": 0.2099, "step": 49894 }, { "epoch": 0.05, "learning_rate": 4.9694661152277276e-05, "loss": 0.2058, "step": 49895 }, { "epoch": 0.05, "learning_rate": 4.969464891334512e-05, "loss": 0.2694, "step": 49896 }, { "epoch": 0.05, "learning_rate": 4.9694636674169185e-05, "loss": 0.2724, "step": 49897 }, { "epoch": 0.05, "learning_rate": 4.969462443474947e-05, "loss": 0.1675, "step": 49898 }, { "epoch": 0.05, "learning_rate": 4.969461219508599e-05, "loss": 0.2191, "step": 49899 }, { "epoch": 0.05, "learning_rate": 4.969459995517873e-05, "loss": 0.2088, "step": 49900 }, { "epoch": 0.05, "learning_rate": 4.9694587715027695e-05, "loss": 0.1629, "step": 49901 }, { "epoch": 0.05, "learning_rate": 4.969457547463289e-05, "loss": 0.1767, "step": 49902 }, { "epoch": 0.05, "learning_rate": 4.9694563233994304e-05, "loss": 0.2686, "step": 49903 }, { "epoch": 0.05, "learning_rate": 4.9694550993111945e-05, "loss": 0.1919, "step": 49904 }, { "epoch": 0.05, "learning_rate": 4.969453875198582e-05, "loss": 0.1692, "step": 49905 }, { "epoch": 0.05, "learning_rate": 4.969452651061591e-05, "loss": 0.1675, "step": 49906 }, { "epoch": 0.05, "learning_rate": 4.9694514269002236e-05, "loss": 0.2759, "step": 49907 }, { "epoch": 0.05, "learning_rate": 4.969450202714478e-05, "loss": 0.214, "step": 49908 }, { "epoch": 0.05, "learning_rate": 4.969448978504355e-05, "loss": 0.203, "step": 49909 }, { "epoch": 0.05, "learning_rate": 4.969447754269855e-05, "loss": 0.2164, "step": 49910 }, { "epoch": 0.05, "learning_rate": 4.969446530010978e-05, "loss": 0.2406, "step": 49911 }, { "epoch": 0.05, "learning_rate": 4.969445305727723e-05, "loss": 0.2969, "step": 49912 }, { "epoch": 0.05, "learning_rate": 4.969444081420091e-05, "loss": 0.6572, "step": 49913 }, { "epoch": 0.05, "learning_rate": 4.969442857088081e-05, "loss": 0.8024, "step": 49914 }, { "epoch": 0.05, "learning_rate": 4.969441632731695e-05, "loss": 1.1187, "step": 49915 }, { "epoch": 0.05, "learning_rate": 4.9694404083509304e-05, "loss": 0.8753, "step": 49916 }, { "epoch": 0.05, "learning_rate": 4.9694391839457896e-05, "loss": 0.6985, "step": 49917 }, { "epoch": 0.05, "learning_rate": 4.969437959516271e-05, "loss": 1.137, "step": 49918 }, { "epoch": 0.05, "learning_rate": 4.969436735062375e-05, "loss": 0.6411, "step": 49919 }, { "epoch": 0.05, "learning_rate": 4.969435510584103e-05, "loss": 0.9207, "step": 49920 }, { "epoch": 0.05, "learning_rate": 4.9694342860814523e-05, "loss": 1.131, "step": 49921 }, { "epoch": 0.05, "learning_rate": 4.969433061554425e-05, "loss": 1.0892, "step": 49922 }, { "epoch": 0.05, "learning_rate": 4.9694318370030204e-05, "loss": 1.3272, "step": 49923 }, { "epoch": 0.05, "learning_rate": 4.9694306124272384e-05, "loss": 1.2183, "step": 49924 }, { "epoch": 0.05, "learning_rate": 4.9694293878270794e-05, "loss": 0.9243, "step": 49925 }, { "epoch": 0.05, "learning_rate": 4.969428163202543e-05, "loss": 0.9693, "step": 49926 }, { "epoch": 0.05, "learning_rate": 4.9694269385536294e-05, "loss": 0.7393, "step": 49927 }, { "epoch": 0.05, "learning_rate": 4.969425713880339e-05, "loss": 0.8623, "step": 49928 }, { "epoch": 0.05, "learning_rate": 4.9694244891826716e-05, "loss": 1.4185, "step": 49929 }, { "epoch": 0.05, "learning_rate": 4.969423264460627e-05, "loss": 1.057, "step": 49930 }, { "epoch": 0.05, "learning_rate": 4.9694220397142054e-05, "loss": 1.2412, "step": 49931 }, { "epoch": 0.05, "learning_rate": 4.9694208149434065e-05, "loss": 1.0736, "step": 49932 }, { "epoch": 0.05, "learning_rate": 4.969419590148231e-05, "loss": 1.1882, "step": 49933 }, { "epoch": 0.05, "learning_rate": 4.9694183653286774e-05, "loss": 0.7187, "step": 49934 }, { "epoch": 0.05, "learning_rate": 4.969417140484747e-05, "loss": 1.3146, "step": 49935 }, { "epoch": 0.05, "learning_rate": 4.96941591561644e-05, "loss": 0.8151, "step": 49936 }, { "epoch": 0.05, "learning_rate": 4.969414690723756e-05, "loss": 1.126, "step": 49937 }, { "epoch": 0.05, "learning_rate": 4.9694134658066946e-05, "loss": 1.2011, "step": 49938 }, { "epoch": 0.05, "learning_rate": 4.9694122408652566e-05, "loss": 1.0565, "step": 49939 }, { "epoch": 0.05, "learning_rate": 4.969411015899441e-05, "loss": 1.0275, "step": 49940 }, { "epoch": 0.05, "learning_rate": 4.969409790909249e-05, "loss": 0.7647, "step": 49941 }, { "epoch": 0.05, "learning_rate": 4.9694085658946796e-05, "loss": 0.9902, "step": 49942 }, { "epoch": 0.05, "learning_rate": 4.969407340855734e-05, "loss": 0.8775, "step": 49943 }, { "epoch": 0.05, "learning_rate": 4.969406115792411e-05, "loss": 0.8389, "step": 49944 }, { "epoch": 0.05, "learning_rate": 4.969404890704711e-05, "loss": 0.7035, "step": 49945 }, { "epoch": 0.05, "learning_rate": 4.969403665592634e-05, "loss": 1.0765, "step": 49946 }, { "epoch": 0.05, "learning_rate": 4.96940244045618e-05, "loss": 0.9198, "step": 49947 }, { "epoch": 0.05, "learning_rate": 4.96940121529535e-05, "loss": 0.9035, "step": 49948 }, { "epoch": 0.05, "learning_rate": 4.969399990110142e-05, "loss": 0.9188, "step": 49949 }, { "epoch": 0.05, "learning_rate": 4.969398764900557e-05, "loss": 0.9807, "step": 49950 }, { "epoch": 0.05, "learning_rate": 4.969397539666596e-05, "loss": 1.088, "step": 49951 }, { "epoch": 0.05, "learning_rate": 4.9693963144082575e-05, "loss": 1.1111, "step": 49952 }, { "epoch": 0.05, "learning_rate": 4.9693950891255425e-05, "loss": 1.0279, "step": 49953 }, { "epoch": 0.05, "learning_rate": 4.96939386381845e-05, "loss": 0.7702, "step": 49954 }, { "epoch": 0.05, "learning_rate": 4.9693926384869816e-05, "loss": 1.1718, "step": 49955 }, { "epoch": 0.05, "learning_rate": 4.969391413131136e-05, "loss": 1.1134, "step": 49956 }, { "epoch": 0.05, "learning_rate": 4.969390187750914e-05, "loss": 1.3949, "step": 49957 }, { "epoch": 0.05, "learning_rate": 4.9693889623463146e-05, "loss": 1.2134, "step": 49958 }, { "epoch": 0.05, "learning_rate": 4.9693877369173393e-05, "loss": 0.9508, "step": 49959 }, { "epoch": 0.05, "learning_rate": 4.969386511463986e-05, "loss": 1.1351, "step": 49960 }, { "epoch": 0.05, "learning_rate": 4.9693852859862566e-05, "loss": 1.1318, "step": 49961 }, { "epoch": 0.05, "learning_rate": 4.9693840604841506e-05, "loss": 1.1111, "step": 49962 }, { "epoch": 0.05, "learning_rate": 4.9693828349576676e-05, "loss": 1.0734, "step": 49963 }, { "epoch": 0.05, "learning_rate": 4.9693816094068076e-05, "loss": 1.1371, "step": 49964 }, { "epoch": 0.05, "learning_rate": 4.969380383831572e-05, "loss": 1.2483, "step": 49965 }, { "epoch": 0.05, "learning_rate": 4.969379158231959e-05, "loss": 1.2332, "step": 49966 }, { "epoch": 0.05, "learning_rate": 4.9693779326079696e-05, "loss": 0.9402, "step": 49967 }, { "epoch": 0.05, "learning_rate": 4.969376706959603e-05, "loss": 0.5064, "step": 49968 }, { "epoch": 0.05, "learning_rate": 4.969375481286861e-05, "loss": 0.4163, "step": 49969 }, { "epoch": 0.05, "learning_rate": 4.969374255589741e-05, "loss": 0.6387, "step": 49970 }, { "epoch": 0.05, "learning_rate": 4.969373029868245e-05, "loss": 1.0652, "step": 49971 }, { "epoch": 0.05, "learning_rate": 4.969371804122372e-05, "loss": 1.1418, "step": 49972 }, { "epoch": 0.05, "learning_rate": 4.9693705783521226e-05, "loss": 0.8738, "step": 49973 }, { "epoch": 0.05, "learning_rate": 4.969369352557497e-05, "loss": 1.1051, "step": 49974 }, { "epoch": 0.05, "learning_rate": 4.969368126738495e-05, "loss": 1.2677, "step": 49975 }, { "epoch": 0.05, "learning_rate": 4.9693669008951156e-05, "loss": 1.1446, "step": 49976 }, { "epoch": 0.05, "learning_rate": 4.96936567502736e-05, "loss": 1.0077, "step": 49977 }, { "epoch": 0.05, "learning_rate": 4.9693644491352274e-05, "loss": 0.8826, "step": 49978 }, { "epoch": 0.05, "learning_rate": 4.969363223218719e-05, "loss": 0.6697, "step": 49979 }, { "epoch": 0.05, "learning_rate": 4.9693619972778336e-05, "loss": 0.9065, "step": 49980 }, { "epoch": 0.05, "learning_rate": 4.9693607713125726e-05, "loss": 0.9905, "step": 49981 }, { "epoch": 0.05, "learning_rate": 4.969359545322933e-05, "loss": 1.0534, "step": 49982 }, { "epoch": 0.05, "learning_rate": 4.969358319308919e-05, "loss": 0.9926, "step": 49983 }, { "epoch": 0.05, "learning_rate": 4.969357093270528e-05, "loss": 1.5424, "step": 49984 }, { "epoch": 0.05, "learning_rate": 4.969355867207761e-05, "loss": 0.9209, "step": 49985 }, { "epoch": 0.05, "learning_rate": 4.9693546411206165e-05, "loss": 1.2825, "step": 49986 }, { "epoch": 0.05, "learning_rate": 4.969353415009096e-05, "loss": 0.9465, "step": 49987 }, { "epoch": 0.05, "learning_rate": 4.969352188873199e-05, "loss": 0.7661, "step": 49988 }, { "epoch": 0.05, "learning_rate": 4.969350962712926e-05, "loss": 1.2285, "step": 49989 }, { "epoch": 0.05, "learning_rate": 4.969349736528277e-05, "loss": 1.1446, "step": 49990 }, { "epoch": 0.05, "learning_rate": 4.9693485103192506e-05, "loss": 1.2737, "step": 49991 }, { "epoch": 0.05, "learning_rate": 4.969347284085848e-05, "loss": 1.0717, "step": 49992 }, { "epoch": 0.05, "learning_rate": 4.9693460578280694e-05, "loss": 1.2913, "step": 49993 }, { "epoch": 0.05, "learning_rate": 4.969344831545914e-05, "loss": 1.2155, "step": 49994 }, { "epoch": 0.05, "learning_rate": 4.9693436052393825e-05, "loss": 1.3321, "step": 49995 }, { "epoch": 0.05, "learning_rate": 4.969342378908475e-05, "loss": 1.537, "step": 49996 }, { "epoch": 0.05, "learning_rate": 4.9693411525531907e-05, "loss": 1.6327, "step": 49997 }, { "epoch": 0.05, "learning_rate": 4.969339926173531e-05, "loss": 0.9082, "step": 49998 }, { "epoch": 0.05, "learning_rate": 4.969338699769494e-05, "loss": 0.882, "step": 49999 }, { "epoch": 0.05, "learning_rate": 4.969337473341081e-05, "loss": 0.8207, "step": 50000 }, { "epoch": 0.05, "eval_loss": 1.0787854194641113, "eval_runtime": 126.0457, "eval_samples_per_second": 10.988, "eval_steps_per_second": 5.498, "step": 50000 }, { "epoch": 0.05, "learning_rate": 4.969336246888292e-05, "loss": 0.9265, "step": 50001 }, { "epoch": 0.05, "learning_rate": 4.969335020411127e-05, "loss": 0.9232, "step": 50002 }, { "epoch": 0.05, "learning_rate": 4.969333793909585e-05, "loss": 1.3453, "step": 50003 }, { "epoch": 0.05, "learning_rate": 4.969332567383667e-05, "loss": 1.2305, "step": 50004 }, { "epoch": 0.05, "learning_rate": 4.969331340833373e-05, "loss": 1.3203, "step": 50005 }, { "epoch": 0.05, "learning_rate": 4.969330114258702e-05, "loss": 1.4891, "step": 50006 }, { "epoch": 0.05, "learning_rate": 4.969328887659656e-05, "loss": 1.4522, "step": 50007 }, { "epoch": 0.05, "learning_rate": 4.9693276610362335e-05, "loss": 1.308, "step": 50008 }, { "epoch": 0.05, "learning_rate": 4.969326434388435e-05, "loss": 1.072, "step": 50009 }, { "epoch": 0.05, "learning_rate": 4.96932520771626e-05, "loss": 1.0883, "step": 50010 }, { "epoch": 0.05, "learning_rate": 4.969323981019709e-05, "loss": 1.1798, "step": 50011 }, { "epoch": 0.05, "learning_rate": 4.9693227542987816e-05, "loss": 0.9228, "step": 50012 }, { "epoch": 0.05, "learning_rate": 4.969321527553478e-05, "loss": 0.725, "step": 50013 }, { "epoch": 0.05, "learning_rate": 4.9693203007837984e-05, "loss": 0.8173, "step": 50014 }, { "epoch": 0.05, "learning_rate": 4.9693190739897435e-05, "loss": 1.3859, "step": 50015 }, { "epoch": 0.05, "learning_rate": 4.9693178471713116e-05, "loss": 1.0869, "step": 50016 }, { "epoch": 0.05, "learning_rate": 4.9693166203285035e-05, "loss": 1.1415, "step": 50017 }, { "epoch": 0.05, "learning_rate": 4.96931539346132e-05, "loss": 1.3767, "step": 50018 }, { "epoch": 0.05, "learning_rate": 4.9693141665697605e-05, "loss": 0.907, "step": 50019 }, { "epoch": 0.05, "learning_rate": 4.9693129396538243e-05, "loss": 1.2152, "step": 50020 }, { "epoch": 0.05, "learning_rate": 4.9693117127135126e-05, "loss": 1.0973, "step": 50021 }, { "epoch": 0.05, "learning_rate": 4.9693104857488245e-05, "loss": 1.3511, "step": 50022 }, { "epoch": 0.05, "learning_rate": 4.969309258759761e-05, "loss": 1.3359, "step": 50023 }, { "epoch": 0.05, "learning_rate": 4.969308031746321e-05, "loss": 1.1643, "step": 50024 }, { "epoch": 0.05, "learning_rate": 4.969306804708506e-05, "loss": 1.4927, "step": 50025 }, { "epoch": 0.05, "learning_rate": 4.969305577646314e-05, "loss": 1.3373, "step": 50026 }, { "epoch": 0.05, "learning_rate": 4.969304350559747e-05, "loss": 1.0642, "step": 50027 }, { "epoch": 0.05, "learning_rate": 4.969303123448803e-05, "loss": 1.0547, "step": 50028 }, { "epoch": 0.05, "learning_rate": 4.969301896313483e-05, "loss": 1.0169, "step": 50029 }, { "epoch": 0.05, "learning_rate": 4.969300669153788e-05, "loss": 1.2037, "step": 50030 }, { "epoch": 0.05, "learning_rate": 4.9692994419697165e-05, "loss": 1.0391, "step": 50031 }, { "epoch": 0.05, "learning_rate": 4.9692982147612693e-05, "loss": 1.1902, "step": 50032 }, { "epoch": 0.05, "learning_rate": 4.969296987528446e-05, "loss": 1.1499, "step": 50033 }, { "epoch": 0.05, "learning_rate": 4.9692957602712476e-05, "loss": 1.4417, "step": 50034 }, { "epoch": 0.05, "learning_rate": 4.9692945329896724e-05, "loss": 0.9829, "step": 50035 }, { "epoch": 0.05, "learning_rate": 4.969293305683722e-05, "loss": 0.9936, "step": 50036 }, { "epoch": 0.05, "learning_rate": 4.969292078353396e-05, "loss": 1.0131, "step": 50037 }, { "epoch": 0.05, "learning_rate": 4.969290850998694e-05, "loss": 1.1305, "step": 50038 }, { "epoch": 0.05, "learning_rate": 4.969289623619615e-05, "loss": 1.2216, "step": 50039 }, { "epoch": 0.05, "learning_rate": 4.969288396216162e-05, "loss": 1.1691, "step": 50040 }, { "epoch": 0.05, "learning_rate": 4.969287168788332e-05, "loss": 1.045, "step": 50041 }, { "epoch": 0.05, "learning_rate": 4.969285941336127e-05, "loss": 1.262, "step": 50042 }, { "epoch": 0.05, "learning_rate": 4.9692847138595456e-05, "loss": 1.1367, "step": 50043 }, { "epoch": 0.05, "learning_rate": 4.969283486358589e-05, "loss": 0.9626, "step": 50044 }, { "epoch": 0.05, "learning_rate": 4.969282258833256e-05, "loss": 1.176, "step": 50045 }, { "epoch": 0.05, "learning_rate": 4.969281031283548e-05, "loss": 1.0801, "step": 50046 }, { "epoch": 0.05, "learning_rate": 4.9692798037094636e-05, "loss": 1.2267, "step": 50047 }, { "epoch": 0.05, "learning_rate": 4.969278576111004e-05, "loss": 1.2711, "step": 50048 }, { "epoch": 0.05, "learning_rate": 4.9692773484881686e-05, "loss": 1.1643, "step": 50049 }, { "epoch": 0.05, "learning_rate": 4.9692761208409576e-05, "loss": 1.0452, "step": 50050 }, { "epoch": 0.05, "learning_rate": 4.969274893169371e-05, "loss": 0.8559, "step": 50051 }, { "epoch": 0.05, "learning_rate": 4.9692736654734084e-05, "loss": 1.0576, "step": 50052 }, { "epoch": 0.05, "learning_rate": 4.969272437753071e-05, "loss": 0.9572, "step": 50053 }, { "epoch": 0.05, "learning_rate": 4.969271210008357e-05, "loss": 0.3703, "step": 50054 }, { "epoch": 0.05, "learning_rate": 4.969269982239268e-05, "loss": 1.0661, "step": 50055 }, { "epoch": 0.05, "learning_rate": 4.969268754445803e-05, "loss": 0.6895, "step": 50056 }, { "epoch": 0.05, "learning_rate": 4.969267526627963e-05, "loss": 0.8749, "step": 50057 }, { "epoch": 0.05, "learning_rate": 4.969266298785747e-05, "loss": 1.1254, "step": 50058 }, { "epoch": 0.05, "learning_rate": 4.969265070919156e-05, "loss": 1.4458, "step": 50059 }, { "epoch": 0.05, "learning_rate": 4.9692638430281885e-05, "loss": 1.3262, "step": 50060 }, { "epoch": 0.05, "learning_rate": 4.969262615112846e-05, "loss": 1.4592, "step": 50061 }, { "epoch": 0.05, "learning_rate": 4.969261387173128e-05, "loss": 1.4125, "step": 50062 }, { "epoch": 0.05, "learning_rate": 4.9692601592090344e-05, "loss": 1.1918, "step": 50063 }, { "epoch": 0.05, "learning_rate": 4.9692589312205654e-05, "loss": 1.1115, "step": 50064 }, { "epoch": 0.05, "learning_rate": 4.969257703207721e-05, "loss": 1.4228, "step": 50065 }, { "epoch": 0.05, "learning_rate": 4.969256475170501e-05, "loss": 1.1127, "step": 50066 }, { "epoch": 0.05, "learning_rate": 4.969255247108905e-05, "loss": 1.2372, "step": 50067 }, { "epoch": 0.05, "learning_rate": 4.9692540190229344e-05, "loss": 1.3721, "step": 50068 }, { "epoch": 0.05, "learning_rate": 4.969252790912588e-05, "loss": 1.0788, "step": 50069 }, { "epoch": 0.05, "learning_rate": 4.9692515627778665e-05, "loss": 1.1009, "step": 50070 }, { "epoch": 0.05, "learning_rate": 4.969250334618769e-05, "loss": 1.1025, "step": 50071 }, { "epoch": 0.05, "learning_rate": 4.969249106435296e-05, "loss": 1.2841, "step": 50072 }, { "epoch": 0.05, "learning_rate": 4.9692478782274485e-05, "loss": 1.1769, "step": 50073 }, { "epoch": 0.05, "learning_rate": 4.969246649995225e-05, "loss": 1.1476, "step": 50074 }, { "epoch": 0.05, "learning_rate": 4.969245421738626e-05, "loss": 1.2905, "step": 50075 }, { "epoch": 0.05, "learning_rate": 4.9692441934576524e-05, "loss": 1.0822, "step": 50076 }, { "epoch": 0.05, "learning_rate": 4.969242965152303e-05, "loss": 0.3541, "step": 50077 }, { "epoch": 0.05, "learning_rate": 4.969241736822579e-05, "loss": 0.9507, "step": 50078 }, { "epoch": 0.05, "learning_rate": 4.969240508468478e-05, "loss": 1.0277, "step": 50079 }, { "epoch": 0.05, "learning_rate": 4.9692392800900026e-05, "loss": 1.0669, "step": 50080 }, { "epoch": 0.05, "learning_rate": 4.9692380516871516e-05, "loss": 1.2497, "step": 50081 }, { "epoch": 0.05, "learning_rate": 4.9692368232599264e-05, "loss": 0.8934, "step": 50082 }, { "epoch": 0.05, "learning_rate": 4.969235594808325e-05, "loss": 0.7776, "step": 50083 }, { "epoch": 0.05, "learning_rate": 4.9692343663323484e-05, "loss": 1.1478, "step": 50084 }, { "epoch": 0.05, "learning_rate": 4.9692331378319965e-05, "loss": 0.8574, "step": 50085 }, { "epoch": 0.05, "learning_rate": 4.96923190930727e-05, "loss": 0.5705, "step": 50086 }, { "epoch": 0.05, "learning_rate": 4.969230680758168e-05, "loss": 0.9098, "step": 50087 }, { "epoch": 0.05, "learning_rate": 4.9692294521846904e-05, "loss": 1.0843, "step": 50088 }, { "epoch": 0.05, "learning_rate": 4.969228223586838e-05, "loss": 0.9855, "step": 50089 }, { "epoch": 0.05, "learning_rate": 4.9692269949646104e-05, "loss": 1.4671, "step": 50090 }, { "epoch": 0.05, "learning_rate": 4.9692257663180076e-05, "loss": 1.4937, "step": 50091 }, { "epoch": 0.05, "learning_rate": 4.969224537647029e-05, "loss": 0.9719, "step": 50092 }, { "epoch": 0.05, "learning_rate": 4.969223308951676e-05, "loss": 1.2903, "step": 50093 }, { "epoch": 0.05, "learning_rate": 4.969222080231948e-05, "loss": 1.444, "step": 50094 }, { "epoch": 0.05, "learning_rate": 4.9692208514878444e-05, "loss": 1.4068, "step": 50095 }, { "epoch": 0.05, "learning_rate": 4.969219622719366e-05, "loss": 1.4088, "step": 50096 }, { "epoch": 0.05, "learning_rate": 4.969218393926512e-05, "loss": 1.2315, "step": 50097 }, { "epoch": 0.05, "learning_rate": 4.969217165109283e-05, "loss": 1.1402, "step": 50098 }, { "epoch": 0.05, "learning_rate": 4.96921593626768e-05, "loss": 0.6752, "step": 50099 }, { "epoch": 0.05, "learning_rate": 4.969214707401701e-05, "loss": 0.6901, "step": 50100 }, { "epoch": 0.05, "learning_rate": 4.9692134785113475e-05, "loss": 0.8997, "step": 50101 }, { "epoch": 0.05, "learning_rate": 4.969212249596619e-05, "loss": 1.1378, "step": 50102 }, { "epoch": 0.05, "learning_rate": 4.9692110206575146e-05, "loss": 1.1149, "step": 50103 }, { "epoch": 0.05, "learning_rate": 4.969209791694035e-05, "loss": 1.0704, "step": 50104 }, { "epoch": 0.05, "learning_rate": 4.9692085627061815e-05, "loss": 1.0929, "step": 50105 }, { "epoch": 0.05, "learning_rate": 4.9692073336939526e-05, "loss": 0.8369, "step": 50106 }, { "epoch": 0.05, "learning_rate": 4.969206104657349e-05, "loss": 1.3136, "step": 50107 }, { "epoch": 0.05, "learning_rate": 4.9692048755963694e-05, "loss": 1.2859, "step": 50108 }, { "epoch": 0.05, "learning_rate": 4.969203646511016e-05, "loss": 1.1856, "step": 50109 }, { "epoch": 0.05, "learning_rate": 4.9692024174012873e-05, "loss": 1.2618, "step": 50110 }, { "epoch": 0.05, "learning_rate": 4.969201188267183e-05, "loss": 0.9462, "step": 50111 }, { "epoch": 0.05, "learning_rate": 4.969199959108704e-05, "loss": 1.456, "step": 50112 }, { "epoch": 0.05, "learning_rate": 4.969198729925851e-05, "loss": 1.3671, "step": 50113 }, { "epoch": 0.05, "learning_rate": 4.969197500718623e-05, "loss": 1.0982, "step": 50114 }, { "epoch": 0.05, "learning_rate": 4.96919627148702e-05, "loss": 0.8965, "step": 50115 }, { "epoch": 0.05, "learning_rate": 4.9691950422310415e-05, "loss": 1.2522, "step": 50116 }, { "epoch": 0.05, "learning_rate": 4.969193812950689e-05, "loss": 1.0155, "step": 50117 }, { "epoch": 0.05, "learning_rate": 4.969192583645961e-05, "loss": 1.6562, "step": 50118 }, { "epoch": 0.05, "learning_rate": 4.969191354316858e-05, "loss": 1.1301, "step": 50119 }, { "epoch": 0.05, "learning_rate": 4.9691901249633803e-05, "loss": 1.3085, "step": 50120 }, { "epoch": 0.05, "learning_rate": 4.969188895585528e-05, "loss": 1.0195, "step": 50121 }, { "epoch": 0.05, "learning_rate": 4.9691876661833014e-05, "loss": 1.0883, "step": 50122 }, { "epoch": 0.05, "learning_rate": 4.969186436756699e-05, "loss": 1.1884, "step": 50123 }, { "epoch": 0.05, "learning_rate": 4.969185207305723e-05, "loss": 1.0959, "step": 50124 }, { "epoch": 0.05, "learning_rate": 4.9691839778303716e-05, "loss": 1.0122, "step": 50125 }, { "epoch": 0.05, "learning_rate": 4.9691827483306454e-05, "loss": 0.9166, "step": 50126 }, { "epoch": 0.05, "learning_rate": 4.969181518806544e-05, "loss": 1.1355, "step": 50127 }, { "epoch": 0.05, "learning_rate": 4.969180289258068e-05, "loss": 0.8262, "step": 50128 }, { "epoch": 0.05, "learning_rate": 4.969179059685218e-05, "loss": 0.8854, "step": 50129 }, { "epoch": 0.05, "learning_rate": 4.969177830087993e-05, "loss": 0.979, "step": 50130 }, { "epoch": 0.05, "learning_rate": 4.9691766004663944e-05, "loss": 1.1598, "step": 50131 }, { "epoch": 0.05, "learning_rate": 4.9691753708204195e-05, "loss": 1.5389, "step": 50132 }, { "epoch": 0.05, "learning_rate": 4.9691741411500704e-05, "loss": 0.8558, "step": 50133 }, { "epoch": 0.05, "learning_rate": 4.969172911455347e-05, "loss": 1.552, "step": 50134 }, { "epoch": 0.05, "learning_rate": 4.969171681736249e-05, "loss": 1.1513, "step": 50135 }, { "epoch": 0.05, "learning_rate": 4.9691704519927764e-05, "loss": 1.1285, "step": 50136 }, { "epoch": 0.05, "learning_rate": 4.9691692222249284e-05, "loss": 0.9044, "step": 50137 }, { "epoch": 0.05, "learning_rate": 4.969167992432706e-05, "loss": 1.0104, "step": 50138 }, { "epoch": 0.05, "learning_rate": 4.96916676261611e-05, "loss": 0.9343, "step": 50139 }, { "epoch": 0.05, "learning_rate": 4.9691655327751383e-05, "loss": 1.1002, "step": 50140 }, { "epoch": 0.05, "learning_rate": 4.969164302909792e-05, "loss": 1.6698, "step": 50141 }, { "epoch": 0.05, "learning_rate": 4.969163073020072e-05, "loss": 1.3258, "step": 50142 }, { "epoch": 0.05, "learning_rate": 4.969161843105977e-05, "loss": 1.1694, "step": 50143 }, { "epoch": 0.05, "learning_rate": 4.969160613167508e-05, "loss": 0.9562, "step": 50144 }, { "epoch": 0.05, "learning_rate": 4.9691593832046635e-05, "loss": 1.1595, "step": 50145 }, { "epoch": 0.05, "learning_rate": 4.9691581532174455e-05, "loss": 1.4098, "step": 50146 }, { "epoch": 0.05, "learning_rate": 4.969156923205852e-05, "loss": 0.8992, "step": 50147 }, { "epoch": 0.05, "learning_rate": 4.9691556931698854e-05, "loss": 1.2168, "step": 50148 }, { "epoch": 0.05, "learning_rate": 4.9691544631095433e-05, "loss": 1.1678, "step": 50149 }, { "epoch": 0.05, "learning_rate": 4.969153233024827e-05, "loss": 0.2817, "step": 50150 }, { "epoch": 0.05, "learning_rate": 4.969152002915736e-05, "loss": 0.8232, "step": 50151 }, { "epoch": 0.05, "learning_rate": 4.9691507727822705e-05, "loss": 1.255, "step": 50152 }, { "epoch": 0.05, "learning_rate": 4.9691495426244316e-05, "loss": 0.956, "step": 50153 }, { "epoch": 0.05, "learning_rate": 4.969148312442217e-05, "loss": 1.0599, "step": 50154 }, { "epoch": 0.05, "learning_rate": 4.969147082235629e-05, "loss": 0.7239, "step": 50155 }, { "epoch": 0.05, "learning_rate": 4.969145852004666e-05, "loss": 0.9194, "step": 50156 }, { "epoch": 0.05, "learning_rate": 4.969144621749329e-05, "loss": 1.3754, "step": 50157 }, { "epoch": 0.05, "learning_rate": 4.9691433914696175e-05, "loss": 1.3171, "step": 50158 }, { "epoch": 0.05, "learning_rate": 4.969142161165531e-05, "loss": 1.1097, "step": 50159 }, { "epoch": 0.05, "learning_rate": 4.9691409308370714e-05, "loss": 1.1191, "step": 50160 }, { "epoch": 0.05, "learning_rate": 4.969139700484237e-05, "loss": 1.2361, "step": 50161 }, { "epoch": 0.05, "learning_rate": 4.9691384701070285e-05, "loss": 0.9541, "step": 50162 }, { "epoch": 0.05, "learning_rate": 4.9691372397054446e-05, "loss": 1.3411, "step": 50163 }, { "epoch": 0.05, "learning_rate": 4.969136009279487e-05, "loss": 1.425, "step": 50164 }, { "epoch": 0.05, "learning_rate": 4.9691347788291556e-05, "loss": 0.5531, "step": 50165 }, { "epoch": 0.05, "learning_rate": 4.96913354835445e-05, "loss": 0.4266, "step": 50166 }, { "epoch": 0.05, "learning_rate": 4.96913231785537e-05, "loss": 1.1559, "step": 50167 }, { "epoch": 0.05, "learning_rate": 4.9691310873319155e-05, "loss": 0.9615, "step": 50168 }, { "epoch": 0.05, "learning_rate": 4.969129856784087e-05, "loss": 0.9701, "step": 50169 }, { "epoch": 0.05, "learning_rate": 4.969128626211884e-05, "loss": 0.5367, "step": 50170 }, { "epoch": 0.05, "learning_rate": 4.9691273956153074e-05, "loss": 0.7798, "step": 50171 }, { "epoch": 0.05, "learning_rate": 4.9691261649943556e-05, "loss": 0.8064, "step": 50172 }, { "epoch": 0.05, "learning_rate": 4.96912493434903e-05, "loss": 0.9206, "step": 50173 }, { "epoch": 0.05, "learning_rate": 4.9691237036793306e-05, "loss": 1.0711, "step": 50174 }, { "epoch": 0.05, "learning_rate": 4.969122472985257e-05, "loss": 1.1879, "step": 50175 }, { "epoch": 0.05, "learning_rate": 4.9691212422668095e-05, "loss": 1.2325, "step": 50176 }, { "epoch": 0.05, "learning_rate": 4.969120011523988e-05, "loss": 1.1774, "step": 50177 }, { "epoch": 0.05, "learning_rate": 4.9691187807567914e-05, "loss": 1.5569, "step": 50178 }, { "epoch": 0.05, "learning_rate": 4.969117549965221e-05, "loss": 1.4441, "step": 50179 }, { "epoch": 0.05, "learning_rate": 4.9691163191492765e-05, "loss": 1.1897, "step": 50180 }, { "epoch": 0.05, "learning_rate": 4.969115088308959e-05, "loss": 0.9364, "step": 50181 }, { "epoch": 0.05, "learning_rate": 4.969113857444266e-05, "loss": 0.9471, "step": 50182 }, { "epoch": 0.05, "learning_rate": 4.9691126265552e-05, "loss": 1.5567, "step": 50183 }, { "epoch": 0.05, "learning_rate": 4.969111395641759e-05, "loss": 1.4725, "step": 50184 }, { "epoch": 0.05, "learning_rate": 4.969110164703945e-05, "loss": 1.1961, "step": 50185 }, { "epoch": 0.05, "learning_rate": 4.969108933741756e-05, "loss": 1.1399, "step": 50186 }, { "epoch": 0.05, "learning_rate": 4.9691077027551935e-05, "loss": 0.7422, "step": 50187 }, { "epoch": 0.05, "learning_rate": 4.969106471744257e-05, "loss": 0.8287, "step": 50188 }, { "epoch": 0.05, "learning_rate": 4.969105240708947e-05, "loss": 0.9399, "step": 50189 }, { "epoch": 0.05, "learning_rate": 4.969104009649262e-05, "loss": 1.0239, "step": 50190 }, { "epoch": 0.05, "learning_rate": 4.969102778565204e-05, "loss": 1.3618, "step": 50191 }, { "epoch": 0.05, "learning_rate": 4.9691015474567715e-05, "loss": 1.1004, "step": 50192 }, { "epoch": 0.05, "learning_rate": 4.969100316323965e-05, "loss": 0.9876, "step": 50193 }, { "epoch": 0.05, "learning_rate": 4.969099085166785e-05, "loss": 1.0203, "step": 50194 }, { "epoch": 0.05, "learning_rate": 4.9690978539852304e-05, "loss": 1.1839, "step": 50195 }, { "epoch": 0.05, "learning_rate": 4.969096622779303e-05, "loss": 1.1351, "step": 50196 }, { "epoch": 0.05, "learning_rate": 4.9690953915490015e-05, "loss": 1.2162, "step": 50197 }, { "epoch": 0.05, "learning_rate": 4.9690941602943254e-05, "loss": 1.2797, "step": 50198 }, { "epoch": 0.05, "learning_rate": 4.969092929015276e-05, "loss": 0.9495, "step": 50199 }, { "epoch": 0.05, "learning_rate": 4.969091697711852e-05, "loss": 1.5007, "step": 50200 }, { "epoch": 0.05, "learning_rate": 4.969090466384055e-05, "loss": 1.3062, "step": 50201 }, { "epoch": 0.05, "learning_rate": 4.969089235031884e-05, "loss": 1.5275, "step": 50202 }, { "epoch": 0.05, "learning_rate": 4.969088003655339e-05, "loss": 1.5169, "step": 50203 }, { "epoch": 0.05, "learning_rate": 4.96908677225442e-05, "loss": 1.2134, "step": 50204 }, { "epoch": 0.05, "learning_rate": 4.969085540829128e-05, "loss": 1.0641, "step": 50205 }, { "epoch": 0.05, "learning_rate": 4.9690843093794614e-05, "loss": 0.9515, "step": 50206 }, { "epoch": 0.05, "learning_rate": 4.969083077905421e-05, "loss": 1.4067, "step": 50207 }, { "epoch": 0.05, "learning_rate": 4.969081846407008e-05, "loss": 1.537, "step": 50208 }, { "epoch": 0.05, "learning_rate": 4.96908061488422e-05, "loss": 1.6268, "step": 50209 }, { "epoch": 0.05, "learning_rate": 4.9690793833370587e-05, "loss": 1.5903, "step": 50210 }, { "epoch": 0.05, "learning_rate": 4.969078151765524e-05, "loss": 1.1734, "step": 50211 }, { "epoch": 0.05, "learning_rate": 4.9690769201696153e-05, "loss": 1.0441, "step": 50212 }, { "epoch": 0.05, "learning_rate": 4.969075688549333e-05, "loss": 0.9855, "step": 50213 }, { "epoch": 0.05, "learning_rate": 4.9690744569046765e-05, "loss": 1.2209, "step": 50214 }, { "epoch": 0.05, "learning_rate": 4.969073225235647e-05, "loss": 1.2278, "step": 50215 }, { "epoch": 0.05, "learning_rate": 4.9690719935422435e-05, "loss": 1.3415, "step": 50216 }, { "epoch": 0.05, "learning_rate": 4.969070761824467e-05, "loss": 1.2137, "step": 50217 }, { "epoch": 0.05, "learning_rate": 4.969069530082316e-05, "loss": 1.3068, "step": 50218 }, { "epoch": 0.05, "learning_rate": 4.9690682983157924e-05, "loss": 1.2409, "step": 50219 }, { "epoch": 0.05, "learning_rate": 4.969067066524894e-05, "loss": 0.7707, "step": 50220 }, { "epoch": 0.05, "learning_rate": 4.969065834709623e-05, "loss": 1.0701, "step": 50221 }, { "epoch": 0.05, "learning_rate": 4.969064602869978e-05, "loss": 0.8518, "step": 50222 }, { "epoch": 0.05, "learning_rate": 4.9690633710059593e-05, "loss": 1.3164, "step": 50223 }, { "epoch": 0.05, "learning_rate": 4.969062139117567e-05, "loss": 1.0758, "step": 50224 }, { "epoch": 0.05, "learning_rate": 4.969060907204801e-05, "loss": 1.0641, "step": 50225 }, { "epoch": 0.05, "learning_rate": 4.969059675267662e-05, "loss": 1.0987, "step": 50226 }, { "epoch": 0.05, "learning_rate": 4.9690584433061496e-05, "loss": 0.7311, "step": 50227 }, { "epoch": 0.05, "learning_rate": 4.969057211320264e-05, "loss": 0.8874, "step": 50228 }, { "epoch": 0.05, "learning_rate": 4.969055979310004e-05, "loss": 1.1476, "step": 50229 }, { "epoch": 0.05, "learning_rate": 4.969054747275371e-05, "loss": 0.9067, "step": 50230 }, { "epoch": 0.05, "learning_rate": 4.9690535152163645e-05, "loss": 0.9064, "step": 50231 }, { "epoch": 0.05, "learning_rate": 4.969052283132984e-05, "loss": 1.1929, "step": 50232 }, { "epoch": 0.05, "learning_rate": 4.969051051025231e-05, "loss": 1.2879, "step": 50233 }, { "epoch": 0.05, "learning_rate": 4.9690498188931034e-05, "loss": 1.0769, "step": 50234 }, { "epoch": 0.05, "learning_rate": 4.9690485867366034e-05, "loss": 1.3929, "step": 50235 }, { "epoch": 0.05, "learning_rate": 4.969047354555729e-05, "loss": 1.3321, "step": 50236 }, { "epoch": 0.05, "learning_rate": 4.969046122350482e-05, "loss": 1.2955, "step": 50237 }, { "epoch": 0.05, "learning_rate": 4.969044890120862e-05, "loss": 1.3345, "step": 50238 }, { "epoch": 0.05, "learning_rate": 4.969043657866868e-05, "loss": 1.3366, "step": 50239 }, { "epoch": 0.05, "learning_rate": 4.9690424255885e-05, "loss": 1.0592, "step": 50240 }, { "epoch": 0.05, "learning_rate": 4.9690411932857596e-05, "loss": 1.7487, "step": 50241 }, { "epoch": 0.05, "learning_rate": 4.9690399609586456e-05, "loss": 1.0052, "step": 50242 }, { "epoch": 0.05, "learning_rate": 4.969038728607158e-05, "loss": 0.6897, "step": 50243 }, { "epoch": 0.05, "learning_rate": 4.969037496231298e-05, "loss": 0.7975, "step": 50244 }, { "epoch": 0.05, "learning_rate": 4.9690362638310636e-05, "loss": 0.9263, "step": 50245 }, { "epoch": 0.05, "learning_rate": 4.969035031406457e-05, "loss": 1.0039, "step": 50246 }, { "epoch": 0.05, "learning_rate": 4.9690337989574764e-05, "loss": 1.3635, "step": 50247 }, { "epoch": 0.05, "learning_rate": 4.9690325664841225e-05, "loss": 1.1673, "step": 50248 }, { "epoch": 0.05, "learning_rate": 4.969031333986396e-05, "loss": 0.9665, "step": 50249 }, { "epoch": 0.05, "learning_rate": 4.9690301014642954e-05, "loss": 1.4068, "step": 50250 }, { "epoch": 0.05, "learning_rate": 4.969028868917822e-05, "loss": 1.6678, "step": 50251 }, { "epoch": 0.05, "learning_rate": 4.9690276363469754e-05, "loss": 0.9494, "step": 50252 }, { "epoch": 0.05, "learning_rate": 4.969026403751755e-05, "loss": 1.1755, "step": 50253 }, { "epoch": 0.05, "learning_rate": 4.969025171132162e-05, "loss": 1.2554, "step": 50254 }, { "epoch": 0.05, "learning_rate": 4.969023938488197e-05, "loss": 1.3305, "step": 50255 }, { "epoch": 0.05, "learning_rate": 4.969022705819857e-05, "loss": 1.0934, "step": 50256 }, { "epoch": 0.05, "learning_rate": 4.969021473127145e-05, "loss": 0.995, "step": 50257 }, { "epoch": 0.05, "learning_rate": 4.969020240410058e-05, "loss": 1.1988, "step": 50258 }, { "epoch": 0.05, "learning_rate": 4.9690190076686e-05, "loss": 1.2621, "step": 50259 }, { "epoch": 0.05, "learning_rate": 4.9690177749027675e-05, "loss": 1.1804, "step": 50260 }, { "epoch": 0.05, "learning_rate": 4.969016542112564e-05, "loss": 1.1012, "step": 50261 }, { "epoch": 0.05, "learning_rate": 4.969015309297985e-05, "loss": 1.5747, "step": 50262 }, { "epoch": 0.05, "learning_rate": 4.969014076459034e-05, "loss": 1.1985, "step": 50263 }, { "epoch": 0.05, "learning_rate": 4.9690128435957094e-05, "loss": 1.2458, "step": 50264 }, { "epoch": 0.05, "learning_rate": 4.969011610708012e-05, "loss": 1.0176, "step": 50265 }, { "epoch": 0.05, "learning_rate": 4.969010377795942e-05, "loss": 1.2121, "step": 50266 }, { "epoch": 0.05, "learning_rate": 4.969009144859499e-05, "loss": 0.9925, "step": 50267 }, { "epoch": 0.05, "learning_rate": 4.969007911898682e-05, "loss": 1.5137, "step": 50268 }, { "epoch": 0.05, "learning_rate": 4.9690066789134926e-05, "loss": 1.4066, "step": 50269 }, { "epoch": 0.05, "learning_rate": 4.96900544590393e-05, "loss": 1.5088, "step": 50270 }, { "epoch": 0.05, "learning_rate": 4.969004212869995e-05, "loss": 1.3241, "step": 50271 }, { "epoch": 0.05, "learning_rate": 4.969002979811687e-05, "loss": 1.2106, "step": 50272 }, { "epoch": 0.05, "learning_rate": 4.969001746729005e-05, "loss": 1.1327, "step": 50273 }, { "epoch": 0.05, "learning_rate": 4.969000513621951e-05, "loss": 1.0756, "step": 50274 }, { "epoch": 0.05, "learning_rate": 4.968999280490524e-05, "loss": 1.2531, "step": 50275 }, { "epoch": 0.05, "learning_rate": 4.968998047334724e-05, "loss": 1.2391, "step": 50276 }, { "epoch": 0.05, "learning_rate": 4.968996814154551e-05, "loss": 1.2702, "step": 50277 }, { "epoch": 0.05, "learning_rate": 4.968995580950005e-05, "loss": 1.3223, "step": 50278 }, { "epoch": 0.05, "learning_rate": 4.968994347721086e-05, "loss": 0.8433, "step": 50279 }, { "epoch": 0.05, "learning_rate": 4.968993114467795e-05, "loss": 1.0101, "step": 50280 }, { "epoch": 0.05, "learning_rate": 4.9689918811901296e-05, "loss": 1.8617, "step": 50281 }, { "epoch": 0.05, "learning_rate": 4.968990647888093e-05, "loss": 1.616, "step": 50282 }, { "epoch": 0.05, "learning_rate": 4.968989414561682e-05, "loss": 1.3965, "step": 50283 }, { "epoch": 0.05, "learning_rate": 4.9689881812108994e-05, "loss": 1.3253, "step": 50284 }, { "epoch": 0.05, "learning_rate": 4.9689869478357434e-05, "loss": 0.9703, "step": 50285 }, { "epoch": 0.05, "learning_rate": 4.968985714436215e-05, "loss": 1.1987, "step": 50286 }, { "epoch": 0.05, "learning_rate": 4.9689844810123134e-05, "loss": 0.6859, "step": 50287 }, { "epoch": 0.05, "learning_rate": 4.9689832475640395e-05, "loss": 0.2644, "step": 50288 }, { "epoch": 0.05, "learning_rate": 4.968982014091392e-05, "loss": 0.3194, "step": 50289 }, { "epoch": 0.05, "learning_rate": 4.9689807805943724e-05, "loss": 0.1523, "step": 50290 }, { "epoch": 0.05, "learning_rate": 4.9689795470729805e-05, "loss": 0.1867, "step": 50291 }, { "epoch": 0.05, "learning_rate": 4.968978313527215e-05, "loss": 0.1773, "step": 50292 }, { "epoch": 0.05, "learning_rate": 4.968977079957077e-05, "loss": 0.1381, "step": 50293 }, { "epoch": 0.05, "learning_rate": 4.968975846362567e-05, "loss": 0.1476, "step": 50294 }, { "epoch": 0.05, "learning_rate": 4.968974612743683e-05, "loss": 0.5744, "step": 50295 }, { "epoch": 0.05, "learning_rate": 4.9689733791004276e-05, "loss": 1.1355, "step": 50296 }, { "epoch": 0.05, "learning_rate": 4.9689721454327985e-05, "loss": 1.2376, "step": 50297 }, { "epoch": 0.05, "learning_rate": 4.968970911740797e-05, "loss": 1.3838, "step": 50298 }, { "epoch": 0.05, "learning_rate": 4.968969678024423e-05, "loss": 1.3622, "step": 50299 }, { "epoch": 0.05, "learning_rate": 4.968968444283677e-05, "loss": 1.2848, "step": 50300 }, { "epoch": 0.05, "learning_rate": 4.968967210518558e-05, "loss": 0.6123, "step": 50301 }, { "epoch": 0.05, "learning_rate": 4.9689659767290664e-05, "loss": 1.0353, "step": 50302 }, { "epoch": 0.05, "learning_rate": 4.968964742915202e-05, "loss": 0.8454, "step": 50303 }, { "epoch": 0.05, "learning_rate": 4.968963509076965e-05, "loss": 0.6054, "step": 50304 }, { "epoch": 0.05, "learning_rate": 4.968962275214355e-05, "loss": 0.6516, "step": 50305 }, { "epoch": 0.05, "learning_rate": 4.968961041327374e-05, "loss": 0.6522, "step": 50306 }, { "epoch": 0.05, "learning_rate": 4.968959807416019e-05, "loss": 0.5911, "step": 50307 }, { "epoch": 0.05, "learning_rate": 4.968958573480291e-05, "loss": 0.4508, "step": 50308 }, { "epoch": 0.05, "learning_rate": 4.9689573395201916e-05, "loss": 0.4433, "step": 50309 }, { "epoch": 0.05, "learning_rate": 4.96895610553572e-05, "loss": 0.4547, "step": 50310 }, { "epoch": 0.05, "learning_rate": 4.968954871526875e-05, "loss": 0.4637, "step": 50311 }, { "epoch": 0.05, "learning_rate": 4.968953637493658e-05, "loss": 0.7691, "step": 50312 }, { "epoch": 0.05, "learning_rate": 4.968952403436068e-05, "loss": 0.5918, "step": 50313 }, { "epoch": 0.05, "learning_rate": 4.9689511693541066e-05, "loss": 0.6168, "step": 50314 }, { "epoch": 0.05, "learning_rate": 4.968949935247772e-05, "loss": 0.2367, "step": 50315 }, { "epoch": 0.05, "learning_rate": 4.968948701117065e-05, "loss": 0.0995, "step": 50316 }, { "epoch": 0.05, "learning_rate": 4.968947466961986e-05, "loss": 0.1196, "step": 50317 }, { "epoch": 0.05, "learning_rate": 4.9689462327825334e-05, "loss": 0.256, "step": 50318 }, { "epoch": 0.05, "learning_rate": 4.9689449985787095e-05, "loss": 0.336, "step": 50319 }, { "epoch": 0.05, "learning_rate": 4.968943764350513e-05, "loss": 0.803, "step": 50320 }, { "epoch": 0.05, "learning_rate": 4.968942530097944e-05, "loss": 0.9721, "step": 50321 }, { "epoch": 0.05, "learning_rate": 4.968941295821003e-05, "loss": 0.7441, "step": 50322 }, { "epoch": 0.05, "learning_rate": 4.968940061519689e-05, "loss": 0.9942, "step": 50323 }, { "epoch": 0.05, "learning_rate": 4.968938827194003e-05, "loss": 1.2493, "step": 50324 }, { "epoch": 0.05, "learning_rate": 4.968937592843945e-05, "loss": 1.4395, "step": 50325 }, { "epoch": 0.05, "learning_rate": 4.9689363584695145e-05, "loss": 2.0847, "step": 50326 }, { "epoch": 0.05, "learning_rate": 4.968935124070712e-05, "loss": 1.4805, "step": 50327 }, { "epoch": 0.05, "learning_rate": 4.968933889647537e-05, "loss": 1.1967, "step": 50328 }, { "epoch": 0.05, "learning_rate": 4.9689326551999896e-05, "loss": 1.2455, "step": 50329 }, { "epoch": 0.05, "learning_rate": 4.968931420728069e-05, "loss": 0.7534, "step": 50330 }, { "epoch": 0.05, "learning_rate": 4.9689301862317774e-05, "loss": 0.6517, "step": 50331 }, { "epoch": 0.05, "learning_rate": 4.968928951711113e-05, "loss": 1.206, "step": 50332 }, { "epoch": 0.05, "learning_rate": 4.9689277171660764e-05, "loss": 1.2992, "step": 50333 }, { "epoch": 0.05, "learning_rate": 4.968926482596668e-05, "loss": 1.0756, "step": 50334 }, { "epoch": 0.05, "learning_rate": 4.968925248002887e-05, "loss": 1.089, "step": 50335 }, { "epoch": 0.05, "learning_rate": 4.9689240133847334e-05, "loss": 1.1498, "step": 50336 }, { "epoch": 0.05, "learning_rate": 4.968922778742209e-05, "loss": 1.1426, "step": 50337 }, { "epoch": 0.05, "learning_rate": 4.968921544075311e-05, "loss": 1.2599, "step": 50338 }, { "epoch": 0.05, "learning_rate": 4.968920309384042e-05, "loss": 1.5619, "step": 50339 }, { "epoch": 0.05, "learning_rate": 4.9689190746684e-05, "loss": 1.2112, "step": 50340 }, { "epoch": 0.05, "learning_rate": 4.968917839928386e-05, "loss": 1.4211, "step": 50341 }, { "epoch": 0.05, "learning_rate": 4.968916605164e-05, "loss": 1.4312, "step": 50342 }, { "epoch": 0.05, "learning_rate": 4.968915370375242e-05, "loss": 1.0387, "step": 50343 }, { "epoch": 0.05, "learning_rate": 4.968914135562113e-05, "loss": 1.2331, "step": 50344 }, { "epoch": 0.05, "learning_rate": 4.968912900724611e-05, "loss": 0.916, "step": 50345 }, { "epoch": 0.05, "learning_rate": 4.9689116658627364e-05, "loss": 1.0444, "step": 50346 }, { "epoch": 0.05, "learning_rate": 4.96891043097649e-05, "loss": 1.0802, "step": 50347 }, { "epoch": 0.05, "learning_rate": 4.968909196065871e-05, "loss": 1.0135, "step": 50348 }, { "epoch": 0.05, "learning_rate": 4.968907961130881e-05, "loss": 0.9003, "step": 50349 }, { "epoch": 0.05, "learning_rate": 4.9689067261715184e-05, "loss": 1.1972, "step": 50350 }, { "epoch": 0.05, "learning_rate": 4.968905491187784e-05, "loss": 1.4187, "step": 50351 }, { "epoch": 0.05, "learning_rate": 4.968904256179677e-05, "loss": 1.3722, "step": 50352 }, { "epoch": 0.05, "learning_rate": 4.9689030211471994e-05, "loss": 0.9662, "step": 50353 }, { "epoch": 0.05, "learning_rate": 4.968901786090349e-05, "loss": 1.2921, "step": 50354 }, { "epoch": 0.05, "learning_rate": 4.968900551009127e-05, "loss": 1.3374, "step": 50355 }, { "epoch": 0.05, "learning_rate": 4.968899315903533e-05, "loss": 0.9906, "step": 50356 }, { "epoch": 0.05, "learning_rate": 4.9688980807735664e-05, "loss": 0.9692, "step": 50357 }, { "epoch": 0.05, "learning_rate": 4.968896845619228e-05, "loss": 1.0406, "step": 50358 }, { "epoch": 0.05, "learning_rate": 4.9688956104405184e-05, "loss": 1.0051, "step": 50359 }, { "epoch": 0.05, "learning_rate": 4.968894375237436e-05, "loss": 0.9974, "step": 50360 }, { "epoch": 0.05, "learning_rate": 4.968893140009982e-05, "loss": 1.1905, "step": 50361 }, { "epoch": 0.05, "learning_rate": 4.9688919047581564e-05, "loss": 1.2537, "step": 50362 }, { "epoch": 0.05, "learning_rate": 4.968890669481959e-05, "loss": 1.3002, "step": 50363 }, { "epoch": 0.05, "learning_rate": 4.96888943418139e-05, "loss": 1.7312, "step": 50364 }, { "epoch": 0.05, "learning_rate": 4.968888198856448e-05, "loss": 5.9811, "step": 50365 }, { "epoch": 0.05, "learning_rate": 4.968886963507135e-05, "loss": 2.9047, "step": 50366 }, { "epoch": 0.05, "learning_rate": 4.96888572813345e-05, "loss": 1.1046, "step": 50367 }, { "epoch": 0.05, "learning_rate": 4.968884492735394e-05, "loss": 1.0465, "step": 50368 }, { "epoch": 0.05, "learning_rate": 4.968883257312965e-05, "loss": 0.9032, "step": 50369 }, { "epoch": 0.05, "learning_rate": 4.968882021866165e-05, "loss": 1.1582, "step": 50370 }, { "epoch": 0.05, "learning_rate": 4.968880786394993e-05, "loss": 1.2027, "step": 50371 }, { "epoch": 0.05, "learning_rate": 4.968879550899449e-05, "loss": 1.1023, "step": 50372 }, { "epoch": 0.05, "learning_rate": 4.9688783153795336e-05, "loss": 0.8389, "step": 50373 }, { "epoch": 0.05, "learning_rate": 4.968877079835247e-05, "loss": 1.3327, "step": 50374 }, { "epoch": 0.05, "learning_rate": 4.968875844266587e-05, "loss": 1.2686, "step": 50375 }, { "epoch": 0.05, "learning_rate": 4.9688746086735574e-05, "loss": 1.4967, "step": 50376 }, { "epoch": 0.05, "learning_rate": 4.968873373056154e-05, "loss": 0.9205, "step": 50377 }, { "epoch": 0.05, "learning_rate": 4.968872137414381e-05, "loss": 1.3303, "step": 50378 }, { "epoch": 0.05, "learning_rate": 4.968870901748235e-05, "loss": 1.4834, "step": 50379 }, { "epoch": 0.05, "learning_rate": 4.968869666057717e-05, "loss": 1.2487, "step": 50380 }, { "epoch": 0.05, "learning_rate": 4.968868430342829e-05, "loss": 1.2391, "step": 50381 }, { "epoch": 0.05, "learning_rate": 4.968867194603568e-05, "loss": 1.1008, "step": 50382 }, { "epoch": 0.05, "learning_rate": 4.968865958839936e-05, "loss": 1.3623, "step": 50383 }, { "epoch": 0.05, "learning_rate": 4.968864723051932e-05, "loss": 1.5128, "step": 50384 }, { "epoch": 0.05, "learning_rate": 4.968863487239557e-05, "loss": 1.1415, "step": 50385 }, { "epoch": 0.05, "learning_rate": 4.9688622514028094e-05, "loss": 0.9713, "step": 50386 }, { "epoch": 0.05, "learning_rate": 4.968861015541692e-05, "loss": 0.5782, "step": 50387 }, { "epoch": 0.05, "learning_rate": 4.968859779656201e-05, "loss": 1.0812, "step": 50388 }, { "epoch": 0.05, "learning_rate": 4.9688585437463396e-05, "loss": 1.3112, "step": 50389 }, { "epoch": 0.05, "learning_rate": 4.968857307812106e-05, "loss": 1.4037, "step": 50390 }, { "epoch": 0.05, "learning_rate": 4.9688560718535015e-05, "loss": 0.9516, "step": 50391 }, { "epoch": 0.05, "learning_rate": 4.9688548358705255e-05, "loss": 1.1051, "step": 50392 }, { "epoch": 0.05, "learning_rate": 4.968853599863177e-05, "loss": 1.2581, "step": 50393 }, { "epoch": 0.05, "learning_rate": 4.968852363831459e-05, "loss": 1.4388, "step": 50394 }, { "epoch": 0.05, "learning_rate": 4.968851127775368e-05, "loss": 1.1946, "step": 50395 }, { "epoch": 0.05, "learning_rate": 4.968849891694906e-05, "loss": 2.1596, "step": 50396 }, { "epoch": 0.05, "learning_rate": 4.968848655590072e-05, "loss": 1.0954, "step": 50397 }, { "epoch": 0.05, "learning_rate": 4.968847419460867e-05, "loss": 0.9024, "step": 50398 }, { "epoch": 0.05, "learning_rate": 4.9688461833072906e-05, "loss": 1.0251, "step": 50399 }, { "epoch": 0.05, "learning_rate": 4.968844947129343e-05, "loss": 1.3556, "step": 50400 }, { "epoch": 0.05, "learning_rate": 4.9688437109270237e-05, "loss": 1.4037, "step": 50401 }, { "epoch": 0.05, "learning_rate": 4.968842474700333e-05, "loss": 1.0676, "step": 50402 }, { "epoch": 0.05, "learning_rate": 4.968841238449271e-05, "loss": 1.3859, "step": 50403 }, { "epoch": 0.05, "learning_rate": 4.968840002173838e-05, "loss": 1.139, "step": 50404 }, { "epoch": 0.05, "learning_rate": 4.968838765874033e-05, "loss": 1.1575, "step": 50405 }, { "epoch": 0.05, "learning_rate": 4.9688375295498565e-05, "loss": 1.1804, "step": 50406 }, { "epoch": 0.05, "learning_rate": 4.96883629320131e-05, "loss": 1.051, "step": 50407 }, { "epoch": 0.05, "learning_rate": 4.968835056828391e-05, "loss": 1.3302, "step": 50408 }, { "epoch": 0.05, "learning_rate": 4.9688338204311015e-05, "loss": 1.2629, "step": 50409 }, { "epoch": 0.05, "learning_rate": 4.96883258400944e-05, "loss": 1.1385, "step": 50410 }, { "epoch": 0.05, "learning_rate": 4.968831347563407e-05, "loss": 1.2707, "step": 50411 }, { "epoch": 0.05, "learning_rate": 4.9688301110930036e-05, "loss": 0.8942, "step": 50412 }, { "epoch": 0.05, "learning_rate": 4.9688288745982286e-05, "loss": 0.898, "step": 50413 }, { "epoch": 0.05, "learning_rate": 4.968827638079082e-05, "loss": 0.9379, "step": 50414 }, { "epoch": 0.05, "learning_rate": 4.968826401535565e-05, "loss": 0.967, "step": 50415 }, { "epoch": 0.05, "learning_rate": 4.968825164967677e-05, "loss": 1.4066, "step": 50416 }, { "epoch": 0.05, "learning_rate": 4.9688239283754163e-05, "loss": 1.2109, "step": 50417 }, { "epoch": 0.05, "learning_rate": 4.968822691758785e-05, "loss": 1.1309, "step": 50418 }, { "epoch": 0.05, "learning_rate": 4.9688214551177825e-05, "loss": 1.338, "step": 50419 }, { "epoch": 0.05, "learning_rate": 4.968820218452409e-05, "loss": 1.7629, "step": 50420 }, { "epoch": 0.05, "learning_rate": 4.968818981762665e-05, "loss": 1.4673, "step": 50421 }, { "epoch": 0.05, "learning_rate": 4.968817745048549e-05, "loss": 1.067, "step": 50422 }, { "epoch": 0.05, "learning_rate": 4.968816508310062e-05, "loss": 1.2031, "step": 50423 }, { "epoch": 0.05, "learning_rate": 4.968815271547204e-05, "loss": 0.8629, "step": 50424 }, { "epoch": 0.05, "learning_rate": 4.9688140347599756e-05, "loss": 1.1961, "step": 50425 }, { "epoch": 0.05, "learning_rate": 4.968812797948375e-05, "loss": 1.5976, "step": 50426 }, { "epoch": 0.05, "learning_rate": 4.9688115611124036e-05, "loss": 0.8529, "step": 50427 }, { "epoch": 0.05, "learning_rate": 4.9688103242520614e-05, "loss": 0.986, "step": 50428 }, { "epoch": 0.05, "learning_rate": 4.968809087367348e-05, "loss": 0.8569, "step": 50429 }, { "epoch": 0.05, "learning_rate": 4.968807850458264e-05, "loss": 1.0374, "step": 50430 }, { "epoch": 0.05, "learning_rate": 4.968806613524808e-05, "loss": 1.2641, "step": 50431 }, { "epoch": 0.05, "learning_rate": 4.9688053765669816e-05, "loss": 1.1731, "step": 50432 }, { "epoch": 0.05, "learning_rate": 4.968804139584784e-05, "loss": 1.3921, "step": 50433 }, { "epoch": 0.05, "learning_rate": 4.9688029025782153e-05, "loss": 1.0876, "step": 50434 }, { "epoch": 0.05, "learning_rate": 4.9688016655472766e-05, "loss": 1.1649, "step": 50435 }, { "epoch": 0.05, "learning_rate": 4.968800428491966e-05, "loss": 0.6649, "step": 50436 }, { "epoch": 0.05, "learning_rate": 4.968799191412285e-05, "loss": 0.6432, "step": 50437 }, { "epoch": 0.05, "learning_rate": 4.968797954308232e-05, "loss": 0.6037, "step": 50438 }, { "epoch": 0.05, "learning_rate": 4.968796717179809e-05, "loss": 0.7264, "step": 50439 }, { "epoch": 0.05, "learning_rate": 4.9687954800270146e-05, "loss": 1.244, "step": 50440 }, { "epoch": 0.05, "learning_rate": 4.9687942428498495e-05, "loss": 1.1628, "step": 50441 }, { "epoch": 0.05, "learning_rate": 4.9687930056483136e-05, "loss": 1.149, "step": 50442 }, { "epoch": 0.05, "learning_rate": 4.968791768422407e-05, "loss": 1.0965, "step": 50443 }, { "epoch": 0.05, "learning_rate": 4.968790531172129e-05, "loss": 1.0737, "step": 50444 }, { "epoch": 0.05, "learning_rate": 4.96878929389748e-05, "loss": 1.0268, "step": 50445 }, { "epoch": 0.05, "learning_rate": 4.968788056598461e-05, "loss": 1.0583, "step": 50446 }, { "epoch": 0.05, "learning_rate": 4.968786819275071e-05, "loss": 1.0622, "step": 50447 }, { "epoch": 0.05, "learning_rate": 4.96878558192731e-05, "loss": 1.2944, "step": 50448 }, { "epoch": 0.05, "learning_rate": 4.968784344555177e-05, "loss": 1.299, "step": 50449 }, { "epoch": 0.05, "learning_rate": 4.9687831071586746e-05, "loss": 1.2561, "step": 50450 }, { "epoch": 0.05, "learning_rate": 4.968781869737801e-05, "loss": 1.1515, "step": 50451 }, { "epoch": 0.05, "learning_rate": 4.968780632292557e-05, "loss": 1.3938, "step": 50452 }, { "epoch": 0.05, "learning_rate": 4.968779394822942e-05, "loss": 1.0283, "step": 50453 }, { "epoch": 0.05, "learning_rate": 4.968778157328956e-05, "loss": 1.2995, "step": 50454 }, { "epoch": 0.05, "learning_rate": 4.9687769198106e-05, "loss": 0.9719, "step": 50455 }, { "epoch": 0.05, "learning_rate": 4.9687756822678726e-05, "loss": 1.0724, "step": 50456 }, { "epoch": 0.05, "learning_rate": 4.968774444700774e-05, "loss": 0.8639, "step": 50457 }, { "epoch": 0.05, "learning_rate": 4.9687732071093064e-05, "loss": 1.4, "step": 50458 }, { "epoch": 0.05, "learning_rate": 4.9687719694934664e-05, "loss": 1.0921, "step": 50459 }, { "epoch": 0.05, "learning_rate": 4.968770731853257e-05, "loss": 1.0863, "step": 50460 }, { "epoch": 0.05, "learning_rate": 4.968769494188676e-05, "loss": 1.2435, "step": 50461 }, { "epoch": 0.05, "learning_rate": 4.968768256499725e-05, "loss": 1.1768, "step": 50462 }, { "epoch": 0.05, "learning_rate": 4.968767018786402e-05, "loss": 1.3336, "step": 50463 }, { "epoch": 0.05, "learning_rate": 4.96876578104871e-05, "loss": 1.3146, "step": 50464 }, { "epoch": 0.05, "learning_rate": 4.968764543286647e-05, "loss": 1.0182, "step": 50465 }, { "epoch": 0.05, "learning_rate": 4.9687633055002124e-05, "loss": 1.415, "step": 50466 }, { "epoch": 0.05, "learning_rate": 4.9687620676894084e-05, "loss": 1.4583, "step": 50467 }, { "epoch": 0.05, "learning_rate": 4.968760829854233e-05, "loss": 1.0633, "step": 50468 }, { "epoch": 0.05, "learning_rate": 4.968759591994688e-05, "loss": 0.8449, "step": 50469 }, { "epoch": 0.05, "learning_rate": 4.9687583541107715e-05, "loss": 0.8162, "step": 50470 }, { "epoch": 0.05, "learning_rate": 4.968757116202485e-05, "loss": 0.653, "step": 50471 }, { "epoch": 0.05, "learning_rate": 4.968755878269828e-05, "loss": 0.6043, "step": 50472 }, { "epoch": 0.05, "learning_rate": 4.9687546403128005e-05, "loss": 0.5813, "step": 50473 }, { "epoch": 0.05, "learning_rate": 4.968753402331402e-05, "loss": 0.2857, "step": 50474 }, { "epoch": 0.05, "learning_rate": 4.968752164325633e-05, "loss": 0.9749, "step": 50475 }, { "epoch": 0.05, "learning_rate": 4.968750926295494e-05, "loss": 0.8737, "step": 50476 }, { "epoch": 0.05, "learning_rate": 4.9687496882409845e-05, "loss": 1.2045, "step": 50477 }, { "epoch": 0.05, "learning_rate": 4.968748450162104e-05, "loss": 1.213, "step": 50478 }, { "epoch": 0.05, "learning_rate": 4.968747212058854e-05, "loss": 0.8624, "step": 50479 }, { "epoch": 0.05, "learning_rate": 4.968745973931233e-05, "loss": 1.2382, "step": 50480 }, { "epoch": 0.05, "learning_rate": 4.968744735779242e-05, "loss": 0.8945, "step": 50481 }, { "epoch": 0.05, "learning_rate": 4.968743497602879e-05, "loss": 1.1984, "step": 50482 }, { "epoch": 0.05, "learning_rate": 4.968742259402147e-05, "loss": 1.1173, "step": 50483 }, { "epoch": 0.05, "learning_rate": 4.968741021177045e-05, "loss": 1.3938, "step": 50484 }, { "epoch": 0.05, "learning_rate": 4.9687397829275714e-05, "loss": 1.3326, "step": 50485 }, { "epoch": 0.05, "learning_rate": 4.9687385446537284e-05, "loss": 1.2099, "step": 50486 }, { "epoch": 0.05, "learning_rate": 4.968737306355514e-05, "loss": 1.1158, "step": 50487 }, { "epoch": 0.05, "learning_rate": 4.9687360680329306e-05, "loss": 1.2149, "step": 50488 }, { "epoch": 0.05, "learning_rate": 4.968734829685976e-05, "loss": 1.4531, "step": 50489 }, { "epoch": 0.05, "learning_rate": 4.9687335913146515e-05, "loss": 1.664, "step": 50490 }, { "epoch": 0.05, "learning_rate": 4.9687323529189564e-05, "loss": 0.9381, "step": 50491 }, { "epoch": 0.05, "learning_rate": 4.968731114498891e-05, "loss": 1.0633, "step": 50492 }, { "epoch": 0.05, "learning_rate": 4.968729876054456e-05, "loss": 0.9401, "step": 50493 }, { "epoch": 0.05, "learning_rate": 4.96872863758565e-05, "loss": 1.1012, "step": 50494 }, { "epoch": 0.05, "learning_rate": 4.968727399092474e-05, "loss": 0.9159, "step": 50495 }, { "epoch": 0.05, "learning_rate": 4.968726160574928e-05, "loss": 0.9576, "step": 50496 }, { "epoch": 0.05, "learning_rate": 4.968724922033011e-05, "loss": 1.0214, "step": 50497 }, { "epoch": 0.05, "learning_rate": 4.968723683466725e-05, "loss": 1.1518, "step": 50498 }, { "epoch": 0.05, "learning_rate": 4.9687224448760685e-05, "loss": 1.0461, "step": 50499 }, { "epoch": 0.05, "learning_rate": 4.968721206261041e-05, "loss": 0.8938, "step": 50500 }, { "epoch": 0.05, "eval_loss": 1.0758792161941528, "eval_runtime": 125.0988, "eval_samples_per_second": 11.071, "eval_steps_per_second": 5.54, "step": 50500 }, { "epoch": 0.05, "learning_rate": 4.968719967621643e-05, "loss": 0.9843, "step": 50501 }, { "epoch": 0.05, "learning_rate": 4.9687187289578764e-05, "loss": 0.9399, "step": 50502 }, { "epoch": 0.05, "learning_rate": 4.968717490269739e-05, "loss": 1.0746, "step": 50503 }, { "epoch": 0.05, "learning_rate": 4.96871625155723e-05, "loss": 1.3559, "step": 50504 }, { "epoch": 0.05, "learning_rate": 4.968715012820353e-05, "loss": 1.0353, "step": 50505 }, { "epoch": 0.05, "learning_rate": 4.968713774059105e-05, "loss": 1.2984, "step": 50506 }, { "epoch": 0.05, "learning_rate": 4.968712535273487e-05, "loss": 1.2209, "step": 50507 }, { "epoch": 0.05, "learning_rate": 4.968711296463499e-05, "loss": 1.1862, "step": 50508 }, { "epoch": 0.05, "learning_rate": 4.968710057629141e-05, "loss": 1.033, "step": 50509 }, { "epoch": 0.05, "learning_rate": 4.968708818770412e-05, "loss": 1.4466, "step": 50510 }, { "epoch": 0.05, "learning_rate": 4.968707579887314e-05, "loss": 1.6689, "step": 50511 }, { "epoch": 0.05, "learning_rate": 4.9687063409798456e-05, "loss": 1.4491, "step": 50512 }, { "epoch": 0.05, "learning_rate": 4.968705102048008e-05, "loss": 0.9212, "step": 50513 }, { "epoch": 0.05, "learning_rate": 4.968703863091799e-05, "loss": 1.1246, "step": 50514 }, { "epoch": 0.05, "learning_rate": 4.96870262411122e-05, "loss": 1.3274, "step": 50515 }, { "epoch": 0.05, "learning_rate": 4.9687013851062726e-05, "loss": 1.1725, "step": 50516 }, { "epoch": 0.05, "learning_rate": 4.9687001460769535e-05, "loss": 1.2039, "step": 50517 }, { "epoch": 0.05, "learning_rate": 4.9686989070232655e-05, "loss": 1.0857, "step": 50518 }, { "epoch": 0.05, "learning_rate": 4.968697667945207e-05, "loss": 1.0985, "step": 50519 }, { "epoch": 0.05, "learning_rate": 4.9686964288427785e-05, "loss": 1.0643, "step": 50520 }, { "epoch": 0.05, "learning_rate": 4.968695189715981e-05, "loss": 1.032, "step": 50521 }, { "epoch": 0.05, "learning_rate": 4.968693950564812e-05, "loss": 0.7743, "step": 50522 }, { "epoch": 0.05, "learning_rate": 4.968692711389274e-05, "loss": 0.9065, "step": 50523 }, { "epoch": 0.05, "learning_rate": 4.968691472189366e-05, "loss": 0.909, "step": 50524 }, { "epoch": 0.05, "learning_rate": 4.9686902329650884e-05, "loss": 0.9528, "step": 50525 }, { "epoch": 0.05, "learning_rate": 4.968688993716441e-05, "loss": 1.1075, "step": 50526 }, { "epoch": 0.05, "learning_rate": 4.968687754443423e-05, "loss": 1.2445, "step": 50527 }, { "epoch": 0.05, "learning_rate": 4.968686515146036e-05, "loss": 1.1638, "step": 50528 }, { "epoch": 0.05, "learning_rate": 4.968685275824279e-05, "loss": 1.0286, "step": 50529 }, { "epoch": 0.05, "learning_rate": 4.968684036478152e-05, "loss": 1.2723, "step": 50530 }, { "epoch": 0.05, "learning_rate": 4.968682797107655e-05, "loss": 1.3107, "step": 50531 }, { "epoch": 0.05, "learning_rate": 4.9686815577127885e-05, "loss": 1.1101, "step": 50532 }, { "epoch": 0.05, "learning_rate": 4.968680318293552e-05, "loss": 1.1774, "step": 50533 }, { "epoch": 0.05, "learning_rate": 4.968679078849946e-05, "loss": 1.36, "step": 50534 }, { "epoch": 0.05, "learning_rate": 4.96867783938197e-05, "loss": 0.9718, "step": 50535 }, { "epoch": 0.05, "learning_rate": 4.968676599889624e-05, "loss": 1.1118, "step": 50536 }, { "epoch": 0.05, "learning_rate": 4.968675360372909e-05, "loss": 1.5287, "step": 50537 }, { "epoch": 0.05, "learning_rate": 4.968674120831824e-05, "loss": 1.066, "step": 50538 }, { "epoch": 0.05, "learning_rate": 4.968672881266369e-05, "loss": 1.1937, "step": 50539 }, { "epoch": 0.05, "learning_rate": 4.968671641676544e-05, "loss": 1.2313, "step": 50540 }, { "epoch": 0.05, "learning_rate": 4.9686704020623495e-05, "loss": 0.9681, "step": 50541 }, { "epoch": 0.05, "learning_rate": 4.9686691624237866e-05, "loss": 0.8284, "step": 50542 }, { "epoch": 0.05, "learning_rate": 4.968667922760852e-05, "loss": 1.3437, "step": 50543 }, { "epoch": 0.05, "learning_rate": 4.96866668307355e-05, "loss": 1.1446, "step": 50544 }, { "epoch": 0.05, "learning_rate": 4.9686654433618765e-05, "loss": 1.1781, "step": 50545 }, { "epoch": 0.05, "learning_rate": 4.968664203625834e-05, "loss": 1.8631, "step": 50546 }, { "epoch": 0.05, "learning_rate": 4.968662963865422e-05, "loss": 1.6444, "step": 50547 }, { "epoch": 0.05, "learning_rate": 4.96866172408064e-05, "loss": 0.9768, "step": 50548 }, { "epoch": 0.05, "learning_rate": 4.968660484271489e-05, "loss": 1.2034, "step": 50549 }, { "epoch": 0.05, "learning_rate": 4.968659244437968e-05, "loss": 0.9031, "step": 50550 }, { "epoch": 0.05, "learning_rate": 4.968658004580078e-05, "loss": 1.0335, "step": 50551 }, { "epoch": 0.05, "learning_rate": 4.9686567646978174e-05, "loss": 1.1368, "step": 50552 }, { "epoch": 0.05, "learning_rate": 4.968655524791188e-05, "loss": 1.0714, "step": 50553 }, { "epoch": 0.05, "learning_rate": 4.968654284860189e-05, "loss": 1.2127, "step": 50554 }, { "epoch": 0.05, "learning_rate": 4.968653044904821e-05, "loss": 1.1285, "step": 50555 }, { "epoch": 0.05, "learning_rate": 4.968651804925083e-05, "loss": 1.4134, "step": 50556 }, { "epoch": 0.05, "learning_rate": 4.9686505649209744e-05, "loss": 1.2721, "step": 50557 }, { "epoch": 0.05, "learning_rate": 4.968649324892498e-05, "loss": 1.2159, "step": 50558 }, { "epoch": 0.05, "learning_rate": 4.968648084839652e-05, "loss": 1.0537, "step": 50559 }, { "epoch": 0.05, "learning_rate": 4.968646844762436e-05, "loss": 0.8211, "step": 50560 }, { "epoch": 0.05, "learning_rate": 4.9686456046608495e-05, "loss": 1.4421, "step": 50561 }, { "epoch": 0.05, "learning_rate": 4.968644364534896e-05, "loss": 1.0636, "step": 50562 }, { "epoch": 0.05, "learning_rate": 4.968643124384571e-05, "loss": 1.1208, "step": 50563 }, { "epoch": 0.05, "learning_rate": 4.968641884209877e-05, "loss": 0.7277, "step": 50564 }, { "epoch": 0.05, "learning_rate": 4.9686406440108144e-05, "loss": 1.1801, "step": 50565 }, { "epoch": 0.05, "learning_rate": 4.968639403787382e-05, "loss": 1.0424, "step": 50566 }, { "epoch": 0.05, "learning_rate": 4.96863816353958e-05, "loss": 0.8786, "step": 50567 }, { "epoch": 0.05, "learning_rate": 4.968636923267409e-05, "loss": 1.1532, "step": 50568 }, { "epoch": 0.05, "learning_rate": 4.968635682970868e-05, "loss": 1.1604, "step": 50569 }, { "epoch": 0.05, "learning_rate": 4.968634442649959e-05, "loss": 1.153, "step": 50570 }, { "epoch": 0.05, "learning_rate": 4.96863320230468e-05, "loss": 0.4108, "step": 50571 }, { "epoch": 0.05, "learning_rate": 4.968631961935031e-05, "loss": 0.3735, "step": 50572 }, { "epoch": 0.05, "learning_rate": 4.968630721541013e-05, "loss": 0.2491, "step": 50573 }, { "epoch": 0.05, "learning_rate": 4.9686294811226265e-05, "loss": 1.0862, "step": 50574 }, { "epoch": 0.05, "learning_rate": 4.96862824067987e-05, "loss": 1.1915, "step": 50575 }, { "epoch": 0.05, "learning_rate": 4.9686270002127445e-05, "loss": 1.4821, "step": 50576 }, { "epoch": 0.05, "learning_rate": 4.96862575972125e-05, "loss": 1.7841, "step": 50577 }, { "epoch": 0.05, "learning_rate": 4.968624519205386e-05, "loss": 2.1294, "step": 50578 }, { "epoch": 0.05, "learning_rate": 4.9686232786651526e-05, "loss": 1.5932, "step": 50579 }, { "epoch": 0.05, "learning_rate": 4.9686220381005496e-05, "loss": 1.1609, "step": 50580 }, { "epoch": 0.05, "learning_rate": 4.9686207975115785e-05, "loss": 1.1539, "step": 50581 }, { "epoch": 0.05, "learning_rate": 4.968619556898237e-05, "loss": 1.0042, "step": 50582 }, { "epoch": 0.05, "learning_rate": 4.968618316260528e-05, "loss": 0.9466, "step": 50583 }, { "epoch": 0.05, "learning_rate": 4.9686170755984485e-05, "loss": 1.141, "step": 50584 }, { "epoch": 0.05, "learning_rate": 4.968615834912e-05, "loss": 1.1894, "step": 50585 }, { "epoch": 0.05, "learning_rate": 4.968614594201183e-05, "loss": 1.1885, "step": 50586 }, { "epoch": 0.05, "learning_rate": 4.9686133534659965e-05, "loss": 1.0109, "step": 50587 }, { "epoch": 0.05, "learning_rate": 4.9686121127064404e-05, "loss": 1.2815, "step": 50588 }, { "epoch": 0.05, "learning_rate": 4.9686108719225156e-05, "loss": 1.3006, "step": 50589 }, { "epoch": 0.05, "learning_rate": 4.968609631114222e-05, "loss": 1.2402, "step": 50590 }, { "epoch": 0.05, "learning_rate": 4.968608390281559e-05, "loss": 1.864, "step": 50591 }, { "epoch": 0.05, "learning_rate": 4.968607149424527e-05, "loss": 1.1567, "step": 50592 }, { "epoch": 0.05, "learning_rate": 4.9686059085431255e-05, "loss": 1.0026, "step": 50593 }, { "epoch": 0.05, "learning_rate": 4.968604667637355e-05, "loss": 1.4619, "step": 50594 }, { "epoch": 0.05, "learning_rate": 4.968603426707217e-05, "loss": 1.2129, "step": 50595 }, { "epoch": 0.05, "learning_rate": 4.9686021857527085e-05, "loss": 1.2547, "step": 50596 }, { "epoch": 0.05, "learning_rate": 4.968600944773831e-05, "loss": 1.1535, "step": 50597 }, { "epoch": 0.05, "learning_rate": 4.968599703770586e-05, "loss": 0.9949, "step": 50598 }, { "epoch": 0.05, "learning_rate": 4.96859846274297e-05, "loss": 1.2915, "step": 50599 }, { "epoch": 0.05, "learning_rate": 4.968597221690986e-05, "loss": 0.9793, "step": 50600 }, { "epoch": 0.05, "learning_rate": 4.9685959806146334e-05, "loss": 0.8017, "step": 50601 }, { "epoch": 0.05, "learning_rate": 4.968594739513911e-05, "loss": 1.0749, "step": 50602 }, { "epoch": 0.05, "learning_rate": 4.968593498388821e-05, "loss": 1.2061, "step": 50603 }, { "epoch": 0.05, "learning_rate": 4.9685922572393605e-05, "loss": 1.1996, "step": 50604 }, { "epoch": 0.05, "learning_rate": 4.968591016065532e-05, "loss": 1.1858, "step": 50605 }, { "epoch": 0.05, "learning_rate": 4.968589774867334e-05, "loss": 0.7212, "step": 50606 }, { "epoch": 0.05, "learning_rate": 4.968588533644768e-05, "loss": 0.895, "step": 50607 }, { "epoch": 0.05, "learning_rate": 4.968587292397833e-05, "loss": 1.0295, "step": 50608 }, { "epoch": 0.05, "learning_rate": 4.968586051126528e-05, "loss": 1.1373, "step": 50609 }, { "epoch": 0.05, "learning_rate": 4.968584809830855e-05, "loss": 1.0329, "step": 50610 }, { "epoch": 0.05, "learning_rate": 4.968583568510813e-05, "loss": 0.6696, "step": 50611 }, { "epoch": 0.05, "learning_rate": 4.9685823271664026e-05, "loss": 1.1271, "step": 50612 }, { "epoch": 0.05, "learning_rate": 4.968581085797623e-05, "loss": 1.0513, "step": 50613 }, { "epoch": 0.05, "learning_rate": 4.968579844404475e-05, "loss": 0.9094, "step": 50614 }, { "epoch": 0.05, "learning_rate": 4.968578602986958e-05, "loss": 1.0173, "step": 50615 }, { "epoch": 0.05, "learning_rate": 4.968577361545072e-05, "loss": 1.0857, "step": 50616 }, { "epoch": 0.05, "learning_rate": 4.9685761200788175e-05, "loss": 1.1016, "step": 50617 }, { "epoch": 0.05, "learning_rate": 4.968574878588194e-05, "loss": 1.0428, "step": 50618 }, { "epoch": 0.05, "learning_rate": 4.968573637073202e-05, "loss": 1.3919, "step": 50619 }, { "epoch": 0.05, "learning_rate": 4.9685723955338416e-05, "loss": 1.3936, "step": 50620 }, { "epoch": 0.05, "learning_rate": 4.968571153970112e-05, "loss": 1.0674, "step": 50621 }, { "epoch": 0.05, "learning_rate": 4.968569912382014e-05, "loss": 1.3668, "step": 50622 }, { "epoch": 0.05, "learning_rate": 4.9685686707695464e-05, "loss": 0.9921, "step": 50623 }, { "epoch": 0.05, "learning_rate": 4.968567429132711e-05, "loss": 0.9554, "step": 50624 }, { "epoch": 0.05, "learning_rate": 4.968566187471507e-05, "loss": 1.0859, "step": 50625 }, { "epoch": 0.05, "learning_rate": 4.968564945785935e-05, "loss": 1.5276, "step": 50626 }, { "epoch": 0.05, "learning_rate": 4.968563704075993e-05, "loss": 1.3005, "step": 50627 }, { "epoch": 0.05, "learning_rate": 4.968562462341683e-05, "loss": 1.2325, "step": 50628 }, { "epoch": 0.05, "learning_rate": 4.968561220583005e-05, "loss": 1.3031, "step": 50629 }, { "epoch": 0.05, "learning_rate": 4.968559978799957e-05, "loss": 1.4582, "step": 50630 }, { "epoch": 0.05, "learning_rate": 4.968558736992541e-05, "loss": 0.65, "step": 50631 }, { "epoch": 0.05, "learning_rate": 4.968557495160757e-05, "loss": 1.0494, "step": 50632 }, { "epoch": 0.05, "learning_rate": 4.968556253304604e-05, "loss": 0.7952, "step": 50633 }, { "epoch": 0.05, "learning_rate": 4.968555011424082e-05, "loss": 0.9401, "step": 50634 }, { "epoch": 0.05, "learning_rate": 4.968553769519192e-05, "loss": 0.7701, "step": 50635 }, { "epoch": 0.05, "learning_rate": 4.968552527589934e-05, "loss": 1.2951, "step": 50636 }, { "epoch": 0.05, "learning_rate": 4.9685512856363066e-05, "loss": 1.0135, "step": 50637 }, { "epoch": 0.05, "learning_rate": 4.968550043658311e-05, "loss": 1.0303, "step": 50638 }, { "epoch": 0.05, "learning_rate": 4.9685488016559465e-05, "loss": 1.0338, "step": 50639 }, { "epoch": 0.05, "learning_rate": 4.968547559629214e-05, "loss": 0.5703, "step": 50640 }, { "epoch": 0.05, "learning_rate": 4.968546317578113e-05, "loss": 0.4362, "step": 50641 }, { "epoch": 0.05, "learning_rate": 4.968545075502643e-05, "loss": 0.8595, "step": 50642 }, { "epoch": 0.05, "learning_rate": 4.9685438334028054e-05, "loss": 0.8194, "step": 50643 }, { "epoch": 0.05, "learning_rate": 4.968542591278599e-05, "loss": 0.8719, "step": 50644 }, { "epoch": 0.05, "learning_rate": 4.968541349130025e-05, "loss": 0.9755, "step": 50645 }, { "epoch": 0.05, "learning_rate": 4.9685401069570814e-05, "loss": 1.078, "step": 50646 }, { "epoch": 0.05, "learning_rate": 4.9685388647597696e-05, "loss": 5.8518, "step": 50647 }, { "epoch": 0.05, "learning_rate": 4.9685376225380896e-05, "loss": 5.7106, "step": 50648 }, { "epoch": 0.05, "learning_rate": 4.9685363802920416e-05, "loss": 5.7889, "step": 50649 }, { "epoch": 0.05, "learning_rate": 4.968535138021625e-05, "loss": 2.8641, "step": 50650 }, { "epoch": 0.05, "learning_rate": 4.9685338957268404e-05, "loss": 0.8263, "step": 50651 }, { "epoch": 0.05, "learning_rate": 4.968532653407687e-05, "loss": 0.8708, "step": 50652 }, { "epoch": 0.05, "learning_rate": 4.9685314110641654e-05, "loss": 1.5, "step": 50653 }, { "epoch": 0.05, "learning_rate": 4.9685301686962754e-05, "loss": 0.8209, "step": 50654 }, { "epoch": 0.05, "learning_rate": 4.9685289263040166e-05, "loss": 0.8555, "step": 50655 }, { "epoch": 0.05, "learning_rate": 4.9685276838873904e-05, "loss": 1.1675, "step": 50656 }, { "epoch": 0.05, "learning_rate": 4.968526441446396e-05, "loss": 1.1454, "step": 50657 }, { "epoch": 0.05, "learning_rate": 4.9685251989810335e-05, "loss": 1.2498, "step": 50658 }, { "epoch": 0.05, "learning_rate": 4.968523956491302e-05, "loss": 1.1628, "step": 50659 }, { "epoch": 0.05, "learning_rate": 4.968522713977203e-05, "loss": 1.3042, "step": 50660 }, { "epoch": 0.05, "learning_rate": 4.968521471438735e-05, "loss": 1.3668, "step": 50661 }, { "epoch": 0.05, "learning_rate": 4.968520228875899e-05, "loss": 1.1077, "step": 50662 }, { "epoch": 0.05, "learning_rate": 4.968518986288695e-05, "loss": 1.0109, "step": 50663 }, { "epoch": 0.05, "learning_rate": 4.9685177436771225e-05, "loss": 1.0797, "step": 50664 }, { "epoch": 0.05, "learning_rate": 4.9685165010411825e-05, "loss": 1.0139, "step": 50665 }, { "epoch": 0.05, "learning_rate": 4.9685152583808736e-05, "loss": 1.5015, "step": 50666 }, { "epoch": 0.05, "learning_rate": 4.968514015696197e-05, "loss": 1.3374, "step": 50667 }, { "epoch": 0.05, "learning_rate": 4.9685127729871516e-05, "loss": 1.0127, "step": 50668 }, { "epoch": 0.05, "learning_rate": 4.968511530253739e-05, "loss": 1.3128, "step": 50669 }, { "epoch": 0.05, "learning_rate": 4.968510287495958e-05, "loss": 1.6846, "step": 50670 }, { "epoch": 0.05, "learning_rate": 4.968509044713809e-05, "loss": 1.1934, "step": 50671 }, { "epoch": 0.05, "learning_rate": 4.9685078019072915e-05, "loss": 1.1875, "step": 50672 }, { "epoch": 0.05, "learning_rate": 4.9685065590764065e-05, "loss": 1.3324, "step": 50673 }, { "epoch": 0.05, "learning_rate": 4.968505316221153e-05, "loss": 1.033, "step": 50674 }, { "epoch": 0.05, "learning_rate": 4.9685040733415314e-05, "loss": 0.873, "step": 50675 }, { "epoch": 0.05, "learning_rate": 4.968502830437543e-05, "loss": 1.0884, "step": 50676 }, { "epoch": 0.05, "learning_rate": 4.9685015875091846e-05, "loss": 1.0083, "step": 50677 }, { "epoch": 0.05, "learning_rate": 4.968500344556459e-05, "loss": 1.1854, "step": 50678 }, { "epoch": 0.05, "learning_rate": 4.968499101579366e-05, "loss": 0.9012, "step": 50679 }, { "epoch": 0.05, "learning_rate": 4.968497858577904e-05, "loss": 1.3898, "step": 50680 }, { "epoch": 0.05, "learning_rate": 4.9684966155520755e-05, "loss": 1.7551, "step": 50681 }, { "epoch": 0.05, "learning_rate": 4.9684953725018774e-05, "loss": 0.805, "step": 50682 }, { "epoch": 0.05, "learning_rate": 4.9684941294273125e-05, "loss": 0.9115, "step": 50683 }, { "epoch": 0.05, "learning_rate": 4.968492886328379e-05, "loss": 1.361, "step": 50684 }, { "epoch": 0.05, "learning_rate": 4.968491643205078e-05, "loss": 1.1787, "step": 50685 }, { "epoch": 0.05, "learning_rate": 4.9684904000574085e-05, "loss": 0.919, "step": 50686 }, { "epoch": 0.05, "learning_rate": 4.968489156885372e-05, "loss": 1.1966, "step": 50687 }, { "epoch": 0.05, "learning_rate": 4.9684879136889664e-05, "loss": 1.0173, "step": 50688 }, { "epoch": 0.05, "learning_rate": 4.968486670468194e-05, "loss": 1.3011, "step": 50689 }, { "epoch": 0.05, "learning_rate": 4.968485427223054e-05, "loss": 1.1631, "step": 50690 }, { "epoch": 0.05, "learning_rate": 4.968484183953545e-05, "loss": 1.345, "step": 50691 }, { "epoch": 0.05, "learning_rate": 4.968482940659669e-05, "loss": 1.3897, "step": 50692 }, { "epoch": 0.05, "learning_rate": 4.968481697341425e-05, "loss": 1.0494, "step": 50693 }, { "epoch": 0.05, "learning_rate": 4.9684804539988125e-05, "loss": 1.1035, "step": 50694 }, { "epoch": 0.05, "learning_rate": 4.968479210631833e-05, "loss": 1.4045, "step": 50695 }, { "epoch": 0.05, "learning_rate": 4.968477967240486e-05, "loss": 1.0621, "step": 50696 }, { "epoch": 0.05, "learning_rate": 4.96847672382477e-05, "loss": 1.0542, "step": 50697 }, { "epoch": 0.05, "learning_rate": 4.9684754803846874e-05, "loss": 1.2194, "step": 50698 }, { "epoch": 0.05, "learning_rate": 4.968474236920237e-05, "loss": 1.2444, "step": 50699 }, { "epoch": 0.05, "learning_rate": 4.968472993431418e-05, "loss": 1.2558, "step": 50700 }, { "epoch": 0.05, "learning_rate": 4.9684717499182315e-05, "loss": 1.3575, "step": 50701 }, { "epoch": 0.05, "learning_rate": 4.968470506380678e-05, "loss": 0.939, "step": 50702 }, { "epoch": 0.05, "learning_rate": 4.968469262818756e-05, "loss": 1.1093, "step": 50703 }, { "epoch": 0.05, "learning_rate": 4.968468019232466e-05, "loss": 1.1997, "step": 50704 }, { "epoch": 0.05, "learning_rate": 4.96846677562181e-05, "loss": 0.9268, "step": 50705 }, { "epoch": 0.05, "learning_rate": 4.968465531986784e-05, "loss": 0.9018, "step": 50706 }, { "epoch": 0.05, "learning_rate": 4.9684642883273926e-05, "loss": 1.1868, "step": 50707 }, { "epoch": 0.05, "learning_rate": 4.968463044643633e-05, "loss": 1.1331, "step": 50708 }, { "epoch": 0.05, "learning_rate": 4.968461800935505e-05, "loss": 2.0497, "step": 50709 }, { "epoch": 0.05, "learning_rate": 4.9684605572030097e-05, "loss": 1.6409, "step": 50710 }, { "epoch": 0.05, "learning_rate": 4.968459313446147e-05, "loss": 1.1042, "step": 50711 }, { "epoch": 0.05, "learning_rate": 4.968458069664917e-05, "loss": 1.162, "step": 50712 }, { "epoch": 0.05, "learning_rate": 4.968456825859318e-05, "loss": 1.3358, "step": 50713 }, { "epoch": 0.05, "learning_rate": 4.968455582029353e-05, "loss": 1.0529, "step": 50714 }, { "epoch": 0.05, "learning_rate": 4.96845433817502e-05, "loss": 1.2222, "step": 50715 }, { "epoch": 0.05, "learning_rate": 4.968453094296319e-05, "loss": 1.1785, "step": 50716 }, { "epoch": 0.05, "learning_rate": 4.968451850393251e-05, "loss": 1.0549, "step": 50717 }, { "epoch": 0.05, "learning_rate": 4.9684506064658157e-05, "loss": 1.2296, "step": 50718 }, { "epoch": 0.05, "learning_rate": 4.968449362514011e-05, "loss": 1.2022, "step": 50719 }, { "epoch": 0.05, "learning_rate": 4.9684481185378415e-05, "loss": 1.3128, "step": 50720 }, { "epoch": 0.05, "learning_rate": 4.968446874537303e-05, "loss": 0.8445, "step": 50721 }, { "epoch": 0.05, "learning_rate": 4.968445630512397e-05, "loss": 2.0535, "step": 50722 }, { "epoch": 0.05, "learning_rate": 4.968444386463125e-05, "loss": 1.7866, "step": 50723 }, { "epoch": 0.05, "learning_rate": 4.968443142389484e-05, "loss": 1.0909, "step": 50724 }, { "epoch": 0.05, "learning_rate": 4.968441898291476e-05, "loss": 1.0545, "step": 50725 }, { "epoch": 0.05, "learning_rate": 4.968440654169101e-05, "loss": 1.2269, "step": 50726 }, { "epoch": 0.05, "learning_rate": 4.968439410022357e-05, "loss": 1.283, "step": 50727 }, { "epoch": 0.05, "learning_rate": 4.968438165851247e-05, "loss": 1.0513, "step": 50728 }, { "epoch": 0.05, "learning_rate": 4.9684369216557705e-05, "loss": 1.6702, "step": 50729 }, { "epoch": 0.05, "learning_rate": 4.9684356774359256e-05, "loss": 0.9695, "step": 50730 }, { "epoch": 0.05, "learning_rate": 4.9684344331917126e-05, "loss": 1.0495, "step": 50731 }, { "epoch": 0.05, "learning_rate": 4.968433188923133e-05, "loss": 1.0332, "step": 50732 }, { "epoch": 0.05, "learning_rate": 4.968431944630186e-05, "loss": 1.077, "step": 50733 }, { "epoch": 0.05, "learning_rate": 4.9684307003128717e-05, "loss": 1.0687, "step": 50734 }, { "epoch": 0.05, "learning_rate": 4.96842945597119e-05, "loss": 1.2462, "step": 50735 }, { "epoch": 0.05, "learning_rate": 4.9684282116051414e-05, "loss": 1.0391, "step": 50736 }, { "epoch": 0.05, "learning_rate": 4.968426967214725e-05, "loss": 1.2062, "step": 50737 }, { "epoch": 0.05, "learning_rate": 4.968425722799941e-05, "loss": 1.1967, "step": 50738 }, { "epoch": 0.05, "learning_rate": 4.968424478360791e-05, "loss": 0.4573, "step": 50739 }, { "epoch": 0.05, "learning_rate": 4.968423233897273e-05, "loss": 1.1382, "step": 50740 }, { "epoch": 0.05, "learning_rate": 4.9684219894093874e-05, "loss": 1.1086, "step": 50741 }, { "epoch": 0.05, "learning_rate": 4.968420744897135e-05, "loss": 1.2346, "step": 50742 }, { "epoch": 0.05, "learning_rate": 4.968419500360516e-05, "loss": 1.2137, "step": 50743 }, { "epoch": 0.05, "learning_rate": 4.9684182557995285e-05, "loss": 1.163, "step": 50744 }, { "epoch": 0.05, "learning_rate": 4.968417011214175e-05, "loss": 1.0762, "step": 50745 }, { "epoch": 0.05, "learning_rate": 4.968415766604454e-05, "loss": 1.2321, "step": 50746 }, { "epoch": 0.05, "learning_rate": 4.9684145219703654e-05, "loss": 1.3075, "step": 50747 }, { "epoch": 0.05, "learning_rate": 4.96841327731191e-05, "loss": 1.1952, "step": 50748 }, { "epoch": 0.05, "learning_rate": 4.968412032629087e-05, "loss": 0.9622, "step": 50749 }, { "epoch": 0.05, "learning_rate": 4.968410787921898e-05, "loss": 0.8644, "step": 50750 }, { "epoch": 0.05, "learning_rate": 4.968409543190341e-05, "loss": 1.2293, "step": 50751 }, { "epoch": 0.05, "learning_rate": 4.968408298434417e-05, "loss": 1.114, "step": 50752 }, { "epoch": 0.05, "learning_rate": 4.9684070536541256e-05, "loss": 1.0623, "step": 50753 }, { "epoch": 0.05, "learning_rate": 4.968405808849468e-05, "loss": 4.0675, "step": 50754 }, { "epoch": 0.05, "learning_rate": 4.9684045640204426e-05, "loss": 5.9514, "step": 50755 }, { "epoch": 0.05, "learning_rate": 4.9684033191670496e-05, "loss": 5.2793, "step": 50756 }, { "epoch": 0.05, "learning_rate": 4.968402074289291e-05, "loss": 2.7785, "step": 50757 }, { "epoch": 0.05, "learning_rate": 4.968400829387164e-05, "loss": 1.231, "step": 50758 }, { "epoch": 0.05, "learning_rate": 4.968399584460671e-05, "loss": 1.0617, "step": 50759 }, { "epoch": 0.05, "learning_rate": 4.968398339509811e-05, "loss": 1.235, "step": 50760 }, { "epoch": 0.05, "learning_rate": 4.968397094534584e-05, "loss": 1.1425, "step": 50761 }, { "epoch": 0.05, "learning_rate": 4.9683958495349895e-05, "loss": 1.085, "step": 50762 }, { "epoch": 0.05, "learning_rate": 4.968394604511028e-05, "loss": 1.2177, "step": 50763 }, { "epoch": 0.05, "learning_rate": 4.9683933594627e-05, "loss": 1.2872, "step": 50764 }, { "epoch": 0.05, "learning_rate": 4.968392114390005e-05, "loss": 1.4421, "step": 50765 }, { "epoch": 0.05, "learning_rate": 4.9683908692929427e-05, "loss": 1.4017, "step": 50766 }, { "epoch": 0.05, "learning_rate": 4.968389624171514e-05, "loss": 0.8728, "step": 50767 }, { "epoch": 0.05, "learning_rate": 4.9683883790257176e-05, "loss": 1.3159, "step": 50768 }, { "epoch": 0.05, "learning_rate": 4.968387133855555e-05, "loss": 1.4579, "step": 50769 }, { "epoch": 0.05, "learning_rate": 4.9683858886610256e-05, "loss": 1.3466, "step": 50770 }, { "epoch": 0.05, "learning_rate": 4.968384643442129e-05, "loss": 0.9025, "step": 50771 }, { "epoch": 0.05, "learning_rate": 4.968383398198866e-05, "loss": 1.1325, "step": 50772 }, { "epoch": 0.05, "learning_rate": 4.968382152931236e-05, "loss": 1.0077, "step": 50773 }, { "epoch": 0.05, "learning_rate": 4.968380907639238e-05, "loss": 1.3243, "step": 50774 }, { "epoch": 0.05, "learning_rate": 4.9683796623228746e-05, "loss": 1.2161, "step": 50775 }, { "epoch": 0.05, "learning_rate": 4.968378416982144e-05, "loss": 0.8525, "step": 50776 }, { "epoch": 0.05, "learning_rate": 4.968377171617047e-05, "loss": 0.9308, "step": 50777 }, { "epoch": 0.05, "learning_rate": 4.968375926227582e-05, "loss": 0.7633, "step": 50778 }, { "epoch": 0.05, "learning_rate": 4.968374680813751e-05, "loss": 0.7785, "step": 50779 }, { "epoch": 0.05, "learning_rate": 4.968373435375554e-05, "loss": 0.7425, "step": 50780 }, { "epoch": 0.05, "learning_rate": 4.9683721899129896e-05, "loss": 1.2356, "step": 50781 }, { "epoch": 0.05, "learning_rate": 4.968370944426058e-05, "loss": 0.9945, "step": 50782 }, { "epoch": 0.05, "learning_rate": 4.96836969891476e-05, "loss": 0.8024, "step": 50783 }, { "epoch": 0.05, "learning_rate": 4.968368453379095e-05, "loss": 1.2818, "step": 50784 }, { "epoch": 0.05, "learning_rate": 4.968367207819064e-05, "loss": 1.1973, "step": 50785 }, { "epoch": 0.05, "learning_rate": 4.968365962234666e-05, "loss": 1.1422, "step": 50786 }, { "epoch": 0.05, "learning_rate": 4.968364716625902e-05, "loss": 1.2474, "step": 50787 }, { "epoch": 0.05, "learning_rate": 4.96836347099277e-05, "loss": 1.0883, "step": 50788 }, { "epoch": 0.05, "learning_rate": 4.9683622253352725e-05, "loss": 1.1235, "step": 50789 }, { "epoch": 0.05, "learning_rate": 4.9683609796534074e-05, "loss": 0.9153, "step": 50790 }, { "epoch": 0.05, "learning_rate": 4.968359733947177e-05, "loss": 1.1252, "step": 50791 }, { "epoch": 0.05, "learning_rate": 4.968358488216578e-05, "loss": 1.3719, "step": 50792 }, { "epoch": 0.05, "learning_rate": 4.968357242461614e-05, "loss": 1.0248, "step": 50793 }, { "epoch": 0.05, "learning_rate": 4.968355996682283e-05, "loss": 0.3844, "step": 50794 }, { "epoch": 0.05, "learning_rate": 4.968354750878585e-05, "loss": 0.9308, "step": 50795 }, { "epoch": 0.05, "learning_rate": 4.9683535050505216e-05, "loss": 1.3793, "step": 50796 }, { "epoch": 0.05, "learning_rate": 4.9683522591980905e-05, "loss": 1.1254, "step": 50797 }, { "epoch": 0.05, "learning_rate": 4.9683510133212926e-05, "loss": 0.7462, "step": 50798 }, { "epoch": 0.05, "learning_rate": 4.9683497674201293e-05, "loss": 0.8863, "step": 50799 }, { "epoch": 0.05, "learning_rate": 4.968348521494599e-05, "loss": 1.2496, "step": 50800 }, { "epoch": 0.05, "learning_rate": 4.9683472755447025e-05, "loss": 1.2458, "step": 50801 }, { "epoch": 0.05, "learning_rate": 4.968346029570439e-05, "loss": 1.2618, "step": 50802 }, { "epoch": 0.05, "learning_rate": 4.96834478357181e-05, "loss": 1.1046, "step": 50803 }, { "epoch": 0.05, "learning_rate": 4.9683435375488135e-05, "loss": 1.3834, "step": 50804 }, { "epoch": 0.05, "learning_rate": 4.968342291501451e-05, "loss": 1.0878, "step": 50805 }, { "epoch": 0.05, "learning_rate": 4.968341045429722e-05, "loss": 0.6652, "step": 50806 }, { "epoch": 0.05, "learning_rate": 4.9683397993336264e-05, "loss": 0.3433, "step": 50807 }, { "epoch": 0.05, "learning_rate": 4.968338553213164e-05, "loss": 0.4327, "step": 50808 }, { "epoch": 0.05, "learning_rate": 4.968337307068336e-05, "loss": 0.3267, "step": 50809 }, { "epoch": 0.05, "learning_rate": 4.968336060899142e-05, "loss": 1.087, "step": 50810 }, { "epoch": 0.05, "learning_rate": 4.9683348147055806e-05, "loss": 1.2937, "step": 50811 }, { "epoch": 0.05, "learning_rate": 4.968333568487653e-05, "loss": 1.3539, "step": 50812 }, { "epoch": 0.05, "learning_rate": 4.9683323222453595e-05, "loss": 1.1113, "step": 50813 }, { "epoch": 0.05, "learning_rate": 4.9683310759786995e-05, "loss": 1.1592, "step": 50814 }, { "epoch": 0.05, "learning_rate": 4.968329829687673e-05, "loss": 0.9485, "step": 50815 }, { "epoch": 0.05, "learning_rate": 4.9683285833722805e-05, "loss": 1.3216, "step": 50816 }, { "epoch": 0.05, "learning_rate": 4.968327337032521e-05, "loss": 0.9766, "step": 50817 }, { "epoch": 0.05, "learning_rate": 4.968326090668396e-05, "loss": 1.2029, "step": 50818 }, { "epoch": 0.05, "learning_rate": 4.968324844279905e-05, "loss": 1.3867, "step": 50819 }, { "epoch": 0.05, "learning_rate": 4.968323597867046e-05, "loss": 1.1287, "step": 50820 }, { "epoch": 0.05, "learning_rate": 4.968322351429823e-05, "loss": 1.1136, "step": 50821 }, { "epoch": 0.05, "learning_rate": 4.968321104968232e-05, "loss": 1.2897, "step": 50822 }, { "epoch": 0.05, "learning_rate": 4.968319858482275e-05, "loss": 1.6617, "step": 50823 }, { "epoch": 0.05, "learning_rate": 4.968318611971953e-05, "loss": 1.2159, "step": 50824 }, { "epoch": 0.05, "learning_rate": 4.968317365437264e-05, "loss": 1.1737, "step": 50825 }, { "epoch": 0.05, "learning_rate": 4.968316118878209e-05, "loss": 1.2414, "step": 50826 }, { "epoch": 0.05, "learning_rate": 4.968314872294787e-05, "loss": 0.9421, "step": 50827 }, { "epoch": 0.05, "learning_rate": 4.968313625687e-05, "loss": 1.2881, "step": 50828 }, { "epoch": 0.05, "learning_rate": 4.9683123790548455e-05, "loss": 1.324, "step": 50829 }, { "epoch": 0.05, "learning_rate": 4.968311132398327e-05, "loss": 1.0784, "step": 50830 }, { "epoch": 0.05, "learning_rate": 4.9683098857174405e-05, "loss": 1.3148, "step": 50831 }, { "epoch": 0.05, "learning_rate": 4.968308639012188e-05, "loss": 1.2218, "step": 50832 }, { "epoch": 0.05, "learning_rate": 4.9683073922825704e-05, "loss": 1.4796, "step": 50833 }, { "epoch": 0.05, "learning_rate": 4.968306145528586e-05, "loss": 1.6267, "step": 50834 }, { "epoch": 0.05, "learning_rate": 4.968304898750235e-05, "loss": 1.4163, "step": 50835 }, { "epoch": 0.05, "learning_rate": 4.968303651947519e-05, "loss": 2.2022, "step": 50836 }, { "epoch": 0.05, "learning_rate": 4.9683024051204366e-05, "loss": 1.9747, "step": 50837 }, { "epoch": 0.05, "learning_rate": 4.9683011582689884e-05, "loss": 1.3457, "step": 50838 }, { "epoch": 0.05, "learning_rate": 4.968299911393174e-05, "loss": 1.2464, "step": 50839 }, { "epoch": 0.05, "learning_rate": 4.968298664492994e-05, "loss": 1.2881, "step": 50840 }, { "epoch": 0.05, "learning_rate": 4.968297417568447e-05, "loss": 1.2321, "step": 50841 }, { "epoch": 0.05, "learning_rate": 4.968296170619534e-05, "loss": 1.2366, "step": 50842 }, { "epoch": 0.05, "learning_rate": 4.968294923646256e-05, "loss": 1.1825, "step": 50843 }, { "epoch": 0.05, "learning_rate": 4.9682936766486114e-05, "loss": 0.9497, "step": 50844 }, { "epoch": 0.05, "learning_rate": 4.9682924296266006e-05, "loss": 1.2778, "step": 50845 }, { "epoch": 0.05, "learning_rate": 4.9682911825802244e-05, "loss": 0.7652, "step": 50846 }, { "epoch": 0.05, "learning_rate": 4.968289935509483e-05, "loss": 1.1666, "step": 50847 }, { "epoch": 0.05, "learning_rate": 4.9682886884143744e-05, "loss": 1.3905, "step": 50848 }, { "epoch": 0.05, "learning_rate": 4.9682874412949e-05, "loss": 1.3286, "step": 50849 }, { "epoch": 0.05, "learning_rate": 4.968286194151061e-05, "loss": 1.0613, "step": 50850 }, { "epoch": 0.05, "learning_rate": 4.968284946982854e-05, "loss": 1.2764, "step": 50851 }, { "epoch": 0.05, "learning_rate": 4.968283699790283e-05, "loss": 1.1537, "step": 50852 }, { "epoch": 0.05, "learning_rate": 4.968282452573345e-05, "loss": 0.8506, "step": 50853 }, { "epoch": 0.05, "learning_rate": 4.968281205332042e-05, "loss": 0.8206, "step": 50854 }, { "epoch": 0.05, "learning_rate": 4.968279958066372e-05, "loss": 1.2353, "step": 50855 }, { "epoch": 0.05, "learning_rate": 4.9682787107763376e-05, "loss": 1.2362, "step": 50856 }, { "epoch": 0.05, "learning_rate": 4.968277463461936e-05, "loss": 1.3987, "step": 50857 }, { "epoch": 0.05, "learning_rate": 4.968276216123169e-05, "loss": 1.3165, "step": 50858 }, { "epoch": 0.05, "learning_rate": 4.968274968760037e-05, "loss": 1.1854, "step": 50859 }, { "epoch": 0.05, "learning_rate": 4.9682737213725385e-05, "loss": 0.7144, "step": 50860 }, { "epoch": 0.05, "learning_rate": 4.9682724739606744e-05, "loss": 1.1777, "step": 50861 }, { "epoch": 0.05, "learning_rate": 4.968271226524445e-05, "loss": 1.2108, "step": 50862 }, { "epoch": 0.05, "learning_rate": 4.96826997906385e-05, "loss": 1.1526, "step": 50863 }, { "epoch": 0.05, "learning_rate": 4.968268731578888e-05, "loss": 1.3006, "step": 50864 }, { "epoch": 0.05, "learning_rate": 4.968267484069561e-05, "loss": 0.9912, "step": 50865 }, { "epoch": 0.05, "learning_rate": 4.968266236535869e-05, "loss": 1.1653, "step": 50866 }, { "epoch": 0.05, "learning_rate": 4.9682649889778104e-05, "loss": 1.1836, "step": 50867 }, { "epoch": 0.05, "learning_rate": 4.9682637413953864e-05, "loss": 0.7928, "step": 50868 }, { "epoch": 0.05, "learning_rate": 4.968262493788597e-05, "loss": 0.6318, "step": 50869 }, { "epoch": 0.05, "learning_rate": 4.968261246157442e-05, "loss": 1.2714, "step": 50870 }, { "epoch": 0.05, "learning_rate": 4.968259998501921e-05, "loss": 1.378, "step": 50871 }, { "epoch": 0.05, "learning_rate": 4.968258750822034e-05, "loss": 1.3262, "step": 50872 }, { "epoch": 0.05, "learning_rate": 4.968257503117782e-05, "loss": 1.0094, "step": 50873 }, { "epoch": 0.05, "learning_rate": 4.968256255389164e-05, "loss": 1.0979, "step": 50874 }, { "epoch": 0.05, "learning_rate": 4.968255007636181e-05, "loss": 1.0309, "step": 50875 }, { "epoch": 0.05, "learning_rate": 4.9682537598588316e-05, "loss": 1.2385, "step": 50876 }, { "epoch": 0.05, "learning_rate": 4.9682525120571175e-05, "loss": 1.2336, "step": 50877 }, { "epoch": 0.05, "learning_rate": 4.968251264231037e-05, "loss": 1.0073, "step": 50878 }, { "epoch": 0.05, "learning_rate": 4.968250016380592e-05, "loss": 1.1768, "step": 50879 }, { "epoch": 0.05, "learning_rate": 4.9682487685057813e-05, "loss": 1.5068, "step": 50880 }, { "epoch": 0.05, "learning_rate": 4.968247520606604e-05, "loss": 0.9201, "step": 50881 }, { "epoch": 0.05, "learning_rate": 4.968246272683062e-05, "loss": 0.8493, "step": 50882 }, { "epoch": 0.05, "learning_rate": 4.9682450247351544e-05, "loss": 1.0012, "step": 50883 }, { "epoch": 0.05, "learning_rate": 4.968243776762882e-05, "loss": 1.5085, "step": 50884 }, { "epoch": 0.05, "learning_rate": 4.9682425287662423e-05, "loss": 1.5079, "step": 50885 }, { "epoch": 0.05, "learning_rate": 4.968241280745239e-05, "loss": 1.3053, "step": 50886 }, { "epoch": 0.05, "learning_rate": 4.968240032699869e-05, "loss": 1.1977, "step": 50887 }, { "epoch": 0.05, "learning_rate": 4.968238784630135e-05, "loss": 1.1159, "step": 50888 }, { "epoch": 0.05, "learning_rate": 4.9682375365360346e-05, "loss": 0.8714, "step": 50889 }, { "epoch": 0.05, "learning_rate": 4.968236288417569e-05, "loss": 0.9208, "step": 50890 }, { "epoch": 0.05, "learning_rate": 4.968235040274738e-05, "loss": 0.4857, "step": 50891 }, { "epoch": 0.05, "learning_rate": 4.9682337921075416e-05, "loss": 0.9742, "step": 50892 }, { "epoch": 0.05, "learning_rate": 4.9682325439159795e-05, "loss": 0.9281, "step": 50893 }, { "epoch": 0.05, "learning_rate": 4.968231295700052e-05, "loss": 1.2124, "step": 50894 }, { "epoch": 0.05, "learning_rate": 4.96823004745976e-05, "loss": 1.1833, "step": 50895 }, { "epoch": 0.05, "learning_rate": 4.968228799195102e-05, "loss": 1.2468, "step": 50896 }, { "epoch": 0.05, "learning_rate": 4.968227550906079e-05, "loss": 1.5092, "step": 50897 }, { "epoch": 0.05, "learning_rate": 4.968226302592691e-05, "loss": 1.8816, "step": 50898 }, { "epoch": 0.05, "learning_rate": 4.968225054254937e-05, "loss": 0.9462, "step": 50899 }, { "epoch": 0.05, "learning_rate": 4.968223805892818e-05, "loss": 1.1171, "step": 50900 }, { "epoch": 0.05, "learning_rate": 4.968222557506334e-05, "loss": 1.195, "step": 50901 }, { "epoch": 0.05, "learning_rate": 4.9682213090954844e-05, "loss": 1.2981, "step": 50902 }, { "epoch": 0.05, "learning_rate": 4.9682200606602696e-05, "loss": 1.1339, "step": 50903 }, { "epoch": 0.05, "learning_rate": 4.9682188122006894e-05, "loss": 1.1386, "step": 50904 }, { "epoch": 0.05, "learning_rate": 4.9682175637167445e-05, "loss": 0.996, "step": 50905 }, { "epoch": 0.05, "learning_rate": 4.968216315208434e-05, "loss": 0.4509, "step": 50906 }, { "epoch": 0.05, "learning_rate": 4.968215066675759e-05, "loss": 0.2577, "step": 50907 }, { "epoch": 0.05, "learning_rate": 4.968213818118718e-05, "loss": 0.4026, "step": 50908 }, { "epoch": 0.05, "learning_rate": 4.968212569537312e-05, "loss": 1.0218, "step": 50909 }, { "epoch": 0.05, "learning_rate": 4.9682113209315406e-05, "loss": 1.282, "step": 50910 }, { "epoch": 0.05, "learning_rate": 4.9682100723014046e-05, "loss": 1.2368, "step": 50911 }, { "epoch": 0.05, "learning_rate": 4.968208823646903e-05, "loss": 1.1369, "step": 50912 }, { "epoch": 0.05, "learning_rate": 4.968207574968037e-05, "loss": 1.2487, "step": 50913 }, { "epoch": 0.05, "learning_rate": 4.9682063262648046e-05, "loss": 1.4574, "step": 50914 }, { "epoch": 0.05, "learning_rate": 4.968205077537208e-05, "loss": 1.493, "step": 50915 }, { "epoch": 0.05, "learning_rate": 4.9682038287852465e-05, "loss": 1.0257, "step": 50916 }, { "epoch": 0.05, "learning_rate": 4.96820258000892e-05, "loss": 1.0253, "step": 50917 }, { "epoch": 0.05, "learning_rate": 4.968201331208228e-05, "loss": 1.2529, "step": 50918 }, { "epoch": 0.05, "learning_rate": 4.968200082383171e-05, "loss": 1.3915, "step": 50919 }, { "epoch": 0.05, "learning_rate": 4.9681988335337495e-05, "loss": 1.4504, "step": 50920 }, { "epoch": 0.05, "learning_rate": 4.968197584659962e-05, "loss": 1.0841, "step": 50921 }, { "epoch": 0.05, "learning_rate": 4.96819633576181e-05, "loss": 1.021, "step": 50922 }, { "epoch": 0.05, "learning_rate": 4.9681950868392936e-05, "loss": 1.4769, "step": 50923 }, { "epoch": 0.05, "learning_rate": 4.968193837892411e-05, "loss": 1.3686, "step": 50924 }, { "epoch": 0.05, "learning_rate": 4.968192588921164e-05, "loss": 1.2061, "step": 50925 }, { "epoch": 0.05, "learning_rate": 4.9681913399255524e-05, "loss": 1.0888, "step": 50926 }, { "epoch": 0.05, "learning_rate": 4.968190090905576e-05, "loss": 0.7836, "step": 50927 }, { "epoch": 0.05, "learning_rate": 4.9681888418612346e-05, "loss": 0.8779, "step": 50928 }, { "epoch": 0.05, "learning_rate": 4.968187592792527e-05, "loss": 1.2651, "step": 50929 }, { "epoch": 0.05, "learning_rate": 4.968186343699456e-05, "loss": 1.32, "step": 50930 }, { "epoch": 0.05, "learning_rate": 4.96818509458202e-05, "loss": 1.2343, "step": 50931 }, { "epoch": 0.05, "learning_rate": 4.968183845440218e-05, "loss": 1.1697, "step": 50932 }, { "epoch": 0.05, "learning_rate": 4.968182596274051e-05, "loss": 1.1143, "step": 50933 }, { "epoch": 0.05, "learning_rate": 4.96818134708352e-05, "loss": 1.2898, "step": 50934 }, { "epoch": 0.05, "learning_rate": 4.9681800978686244e-05, "loss": 1.4461, "step": 50935 }, { "epoch": 0.05, "learning_rate": 4.968178848629364e-05, "loss": 1.0953, "step": 50936 }, { "epoch": 0.05, "learning_rate": 4.968177599365739e-05, "loss": 1.0214, "step": 50937 }, { "epoch": 0.05, "learning_rate": 4.968176350077748e-05, "loss": 1.1369, "step": 50938 }, { "epoch": 0.05, "learning_rate": 4.9681751007653934e-05, "loss": 0.918, "step": 50939 }, { "epoch": 0.05, "learning_rate": 4.9681738514286726e-05, "loss": 0.9514, "step": 50940 }, { "epoch": 0.05, "learning_rate": 4.968172602067588e-05, "loss": 1.1774, "step": 50941 }, { "epoch": 0.05, "learning_rate": 4.968171352682138e-05, "loss": 2.2631, "step": 50942 }, { "epoch": 0.05, "learning_rate": 4.9681701032723246e-05, "loss": 1.5238, "step": 50943 }, { "epoch": 0.05, "learning_rate": 4.9681688538381456e-05, "loss": 1.0035, "step": 50944 }, { "epoch": 0.05, "learning_rate": 4.968167604379602e-05, "loss": 0.9361, "step": 50945 }, { "epoch": 0.05, "learning_rate": 4.9681663548966926e-05, "loss": 0.7093, "step": 50946 }, { "epoch": 0.05, "learning_rate": 4.96816510538942e-05, "loss": 1.2162, "step": 50947 }, { "epoch": 0.05, "learning_rate": 4.968163855857783e-05, "loss": 1.2576, "step": 50948 }, { "epoch": 0.05, "learning_rate": 4.9681626063017794e-05, "loss": 0.9496, "step": 50949 }, { "epoch": 0.05, "learning_rate": 4.968161356721413e-05, "loss": 1.079, "step": 50950 }, { "epoch": 0.05, "learning_rate": 4.9681601071166805e-05, "loss": 1.5977, "step": 50951 }, { "epoch": 0.05, "learning_rate": 4.968158857487585e-05, "loss": 1.1165, "step": 50952 }, { "epoch": 0.05, "learning_rate": 4.968157607834123e-05, "loss": 1.2349, "step": 50953 }, { "epoch": 0.05, "learning_rate": 4.968156358156297e-05, "loss": 1.3493, "step": 50954 }, { "epoch": 0.05, "learning_rate": 4.968155108454108e-05, "loss": 1.3811, "step": 50955 }, { "epoch": 0.05, "learning_rate": 4.968153858727552e-05, "loss": 1.3996, "step": 50956 }, { "epoch": 0.05, "learning_rate": 4.968152608976633e-05, "loss": 1.2265, "step": 50957 }, { "epoch": 0.05, "learning_rate": 4.9681513592013496e-05, "loss": 0.9828, "step": 50958 }, { "epoch": 0.05, "learning_rate": 4.968150109401701e-05, "loss": 1.0315, "step": 50959 }, { "epoch": 0.05, "learning_rate": 4.9681488595776875e-05, "loss": 0.7663, "step": 50960 }, { "epoch": 0.05, "learning_rate": 4.96814760972931e-05, "loss": 1.1719, "step": 50961 }, { "epoch": 0.05, "learning_rate": 4.968146359856568e-05, "loss": 1.713, "step": 50962 }, { "epoch": 0.05, "learning_rate": 4.968145109959461e-05, "loss": 1.345, "step": 50963 }, { "epoch": 0.05, "learning_rate": 4.9681438600379906e-05, "loss": 1.0975, "step": 50964 }, { "epoch": 0.05, "learning_rate": 4.968142610092155e-05, "loss": 1.3571, "step": 50965 }, { "epoch": 0.05, "learning_rate": 4.9681413601219545e-05, "loss": 1.2651, "step": 50966 }, { "epoch": 0.05, "learning_rate": 4.96814011012739e-05, "loss": 1.2876, "step": 50967 }, { "epoch": 0.05, "learning_rate": 4.9681388601084614e-05, "loss": 1.1031, "step": 50968 }, { "epoch": 0.05, "learning_rate": 4.9681376100651674e-05, "loss": 1.0482, "step": 50969 }, { "epoch": 0.05, "learning_rate": 4.96813635999751e-05, "loss": 1.3125, "step": 50970 }, { "epoch": 0.05, "learning_rate": 4.968135109905488e-05, "loss": 0.8321, "step": 50971 }, { "epoch": 0.05, "learning_rate": 4.968133859789101e-05, "loss": 0.8595, "step": 50972 }, { "epoch": 0.05, "learning_rate": 4.96813260964835e-05, "loss": 1.2502, "step": 50973 }, { "epoch": 0.05, "learning_rate": 4.968131359483235e-05, "loss": 0.9556, "step": 50974 }, { "epoch": 0.05, "learning_rate": 4.968130109293756e-05, "loss": 1.2349, "step": 50975 }, { "epoch": 0.05, "learning_rate": 4.968128859079911e-05, "loss": 0.4365, "step": 50976 }, { "epoch": 0.05, "learning_rate": 4.968127608841703e-05, "loss": 0.6308, "step": 50977 }, { "epoch": 0.05, "learning_rate": 4.9681263585791303e-05, "loss": 1.1453, "step": 50978 }, { "epoch": 0.05, "learning_rate": 4.968125108292193e-05, "loss": 1.0357, "step": 50979 }, { "epoch": 0.05, "learning_rate": 4.968123857980892e-05, "loss": 1.9318, "step": 50980 }, { "epoch": 0.05, "learning_rate": 4.968122607645227e-05, "loss": 1.0439, "step": 50981 }, { "epoch": 0.05, "learning_rate": 4.968121357285197e-05, "loss": 1.1301, "step": 50982 }, { "epoch": 0.05, "learning_rate": 4.968120106900803e-05, "loss": 1.1981, "step": 50983 }, { "epoch": 0.05, "learning_rate": 4.968118856492045e-05, "loss": 1.0169, "step": 50984 }, { "epoch": 0.05, "learning_rate": 4.968117606058922e-05, "loss": 1.1031, "step": 50985 }, { "epoch": 0.05, "learning_rate": 4.9681163556014355e-05, "loss": 1.1343, "step": 50986 }, { "epoch": 0.05, "learning_rate": 4.968115105119584e-05, "loss": 2.2069, "step": 50987 }, { "epoch": 0.05, "learning_rate": 4.968113854613369e-05, "loss": 5.5721, "step": 50988 }, { "epoch": 0.05, "learning_rate": 4.9681126040827895e-05, "loss": 2.9299, "step": 50989 }, { "epoch": 0.05, "learning_rate": 4.968111353527847e-05, "loss": 1.3713, "step": 50990 }, { "epoch": 0.05, "learning_rate": 4.968110102948539e-05, "loss": 0.8701, "step": 50991 }, { "epoch": 0.05, "learning_rate": 4.968108852344867e-05, "loss": 1.3971, "step": 50992 }, { "epoch": 0.05, "learning_rate": 4.9681076017168316e-05, "loss": 1.1494, "step": 50993 }, { "epoch": 0.05, "learning_rate": 4.968106351064431e-05, "loss": 1.091, "step": 50994 }, { "epoch": 0.05, "learning_rate": 4.968105100387667e-05, "loss": 1.0458, "step": 50995 }, { "epoch": 0.05, "learning_rate": 4.9681038496865385e-05, "loss": 0.5869, "step": 50996 }, { "epoch": 0.05, "learning_rate": 4.968102598961046e-05, "loss": 1.1243, "step": 50997 }, { "epoch": 0.05, "learning_rate": 4.9681013482111905e-05, "loss": 1.1364, "step": 50998 }, { "epoch": 0.05, "learning_rate": 4.9681000974369695e-05, "loss": 1.2243, "step": 50999 }, { "epoch": 0.05, "learning_rate": 4.968098846638385e-05, "loss": 1.1812, "step": 51000 }, { "epoch": 0.05, "eval_loss": 1.0772169828414917, "eval_runtime": 126.8045, "eval_samples_per_second": 10.922, "eval_steps_per_second": 5.465, "step": 51000 }, { "epoch": 0.05, "learning_rate": 4.9680975958154365e-05, "loss": 1.0187, "step": 51001 }, { "epoch": 0.05, "learning_rate": 4.968096344968124e-05, "loss": 0.9657, "step": 51002 }, { "epoch": 0.05, "learning_rate": 4.9680950940964474e-05, "loss": 0.9899, "step": 51003 }, { "epoch": 0.05, "learning_rate": 4.9680938432004074e-05, "loss": 0.7999, "step": 51004 }, { "epoch": 0.05, "learning_rate": 4.9680925922800026e-05, "loss": 0.8354, "step": 51005 }, { "epoch": 0.05, "learning_rate": 4.9680913413352345e-05, "loss": 1.0303, "step": 51006 }, { "epoch": 0.05, "learning_rate": 4.9680900903661024e-05, "loss": 0.8671, "step": 51007 }, { "epoch": 0.05, "learning_rate": 4.9680888393726055e-05, "loss": 0.9649, "step": 51008 }, { "epoch": 0.05, "learning_rate": 4.968087588354745e-05, "loss": 1.1723, "step": 51009 }, { "epoch": 0.05, "learning_rate": 4.968086337312521e-05, "loss": 1.0499, "step": 51010 }, { "epoch": 0.05, "learning_rate": 4.9680850862459325e-05, "loss": 1.1473, "step": 51011 }, { "epoch": 0.05, "learning_rate": 4.968083835154981e-05, "loss": 0.9966, "step": 51012 }, { "epoch": 0.05, "learning_rate": 4.968082584039665e-05, "loss": 1.1618, "step": 51013 }, { "epoch": 0.05, "learning_rate": 4.968081332899985e-05, "loss": 1.1192, "step": 51014 }, { "epoch": 0.05, "learning_rate": 4.968080081735941e-05, "loss": 1.2199, "step": 51015 }, { "epoch": 0.05, "learning_rate": 4.968078830547533e-05, "loss": 0.9783, "step": 51016 }, { "epoch": 0.05, "learning_rate": 4.9680775793347625e-05, "loss": 0.9785, "step": 51017 }, { "epoch": 0.05, "learning_rate": 4.968076328097627e-05, "loss": 1.4172, "step": 51018 }, { "epoch": 0.05, "learning_rate": 4.968075076836128e-05, "loss": 1.1986, "step": 51019 }, { "epoch": 0.05, "learning_rate": 4.9680738255502655e-05, "loss": 1.3663, "step": 51020 }, { "epoch": 0.05, "learning_rate": 4.968072574240039e-05, "loss": 1.319, "step": 51021 }, { "epoch": 0.05, "learning_rate": 4.968071322905449e-05, "loss": 1.1201, "step": 51022 }, { "epoch": 0.05, "learning_rate": 4.9680700715464946e-05, "loss": 1.393, "step": 51023 }, { "epoch": 0.05, "learning_rate": 4.968068820163176e-05, "loss": 0.5786, "step": 51024 }, { "epoch": 0.05, "learning_rate": 4.968067568755496e-05, "loss": 0.2641, "step": 51025 }, { "epoch": 0.05, "learning_rate": 4.96806631732345e-05, "loss": 0.528, "step": 51026 }, { "epoch": 0.05, "learning_rate": 4.9680650658670414e-05, "loss": 0.9759, "step": 51027 }, { "epoch": 0.05, "learning_rate": 4.968063814386268e-05, "loss": 1.2404, "step": 51028 }, { "epoch": 0.05, "learning_rate": 4.9680625628811315e-05, "loss": 1.038, "step": 51029 }, { "epoch": 0.05, "learning_rate": 4.9680613113516315e-05, "loss": 0.9967, "step": 51030 }, { "epoch": 0.05, "learning_rate": 4.9680600597977675e-05, "loss": 1.1677, "step": 51031 }, { "epoch": 0.05, "learning_rate": 4.9680588082195414e-05, "loss": 1.4002, "step": 51032 }, { "epoch": 0.05, "learning_rate": 4.96805755661695e-05, "loss": 1.2451, "step": 51033 }, { "epoch": 0.05, "learning_rate": 4.968056304989995e-05, "loss": 1.22, "step": 51034 }, { "epoch": 0.05, "learning_rate": 4.9680550533386774e-05, "loss": 1.2907, "step": 51035 }, { "epoch": 0.05, "learning_rate": 4.968053801662995e-05, "loss": 1.0201, "step": 51036 }, { "epoch": 0.05, "learning_rate": 4.9680525499629494e-05, "loss": 1.352, "step": 51037 }, { "epoch": 0.05, "learning_rate": 4.968051298238541e-05, "loss": 1.0539, "step": 51038 }, { "epoch": 0.05, "learning_rate": 4.9680500464897685e-05, "loss": 1.1382, "step": 51039 }, { "epoch": 0.05, "learning_rate": 4.968048794716632e-05, "loss": 2.0848, "step": 51040 }, { "epoch": 0.05, "learning_rate": 4.968047542919132e-05, "loss": 0.9598, "step": 51041 }, { "epoch": 0.05, "learning_rate": 4.968046291097269e-05, "loss": 0.6702, "step": 51042 }, { "epoch": 0.05, "learning_rate": 4.968045039251042e-05, "loss": 1.4868, "step": 51043 }, { "epoch": 0.05, "learning_rate": 4.968043787380452e-05, "loss": 1.0412, "step": 51044 }, { "epoch": 0.05, "learning_rate": 4.968042535485499e-05, "loss": 1.1093, "step": 51045 }, { "epoch": 0.05, "learning_rate": 4.9680412835661814e-05, "loss": 1.1818, "step": 51046 }, { "epoch": 0.05, "learning_rate": 4.9680400316225005e-05, "loss": 1.3554, "step": 51047 }, { "epoch": 0.05, "learning_rate": 4.968038779654457e-05, "loss": 0.9076, "step": 51048 }, { "epoch": 0.05, "learning_rate": 4.9680375276620494e-05, "loss": 0.7492, "step": 51049 }, { "epoch": 0.05, "learning_rate": 4.968036275645278e-05, "loss": 1.1089, "step": 51050 }, { "epoch": 0.05, "learning_rate": 4.968035023604144e-05, "loss": 1.0522, "step": 51051 }, { "epoch": 0.05, "learning_rate": 4.968033771538646e-05, "loss": 1.2026, "step": 51052 }, { "epoch": 0.05, "learning_rate": 4.968032519448785e-05, "loss": 0.5533, "step": 51053 }, { "epoch": 0.05, "learning_rate": 4.968031267334561e-05, "loss": 1.2418, "step": 51054 }, { "epoch": 0.05, "learning_rate": 4.968030015195972e-05, "loss": 1.467, "step": 51055 }, { "epoch": 0.05, "learning_rate": 4.968028763033021e-05, "loss": 1.0677, "step": 51056 }, { "epoch": 0.05, "learning_rate": 4.968027510845707e-05, "loss": 0.7095, "step": 51057 }, { "epoch": 0.05, "learning_rate": 4.9680262586340284e-05, "loss": 1.1701, "step": 51058 }, { "epoch": 0.05, "learning_rate": 4.9680250063979884e-05, "loss": 1.0655, "step": 51059 }, { "epoch": 0.05, "learning_rate": 4.968023754137584e-05, "loss": 1.1065, "step": 51060 }, { "epoch": 0.05, "learning_rate": 4.968022501852816e-05, "loss": 1.5136, "step": 51061 }, { "epoch": 0.05, "learning_rate": 4.968021249543685e-05, "loss": 1.051, "step": 51062 }, { "epoch": 0.05, "learning_rate": 4.9680199972101905e-05, "loss": 1.0462, "step": 51063 }, { "epoch": 0.05, "learning_rate": 4.968018744852332e-05, "loss": 0.9805, "step": 51064 }, { "epoch": 0.05, "learning_rate": 4.968017492470112e-05, "loss": 1.2191, "step": 51065 }, { "epoch": 0.05, "learning_rate": 4.9680162400635276e-05, "loss": 1.0385, "step": 51066 }, { "epoch": 0.05, "learning_rate": 4.9680149876325806e-05, "loss": 1.2659, "step": 51067 }, { "epoch": 0.05, "learning_rate": 4.96801373517727e-05, "loss": 1.1055, "step": 51068 }, { "epoch": 0.05, "learning_rate": 4.968012482697596e-05, "loss": 1.0791, "step": 51069 }, { "epoch": 0.05, "learning_rate": 4.968011230193559e-05, "loss": 1.0505, "step": 51070 }, { "epoch": 0.05, "learning_rate": 4.968009977665159e-05, "loss": 0.955, "step": 51071 }, { "epoch": 0.05, "learning_rate": 4.968008725112396e-05, "loss": 1.2262, "step": 51072 }, { "epoch": 0.05, "learning_rate": 4.96800747253527e-05, "loss": 1.3425, "step": 51073 }, { "epoch": 0.05, "learning_rate": 4.968006219933781e-05, "loss": 1.8073, "step": 51074 }, { "epoch": 0.05, "learning_rate": 4.9680049673079274e-05, "loss": 1.4001, "step": 51075 }, { "epoch": 0.05, "learning_rate": 4.968003714657712e-05, "loss": 1.424, "step": 51076 }, { "epoch": 0.05, "learning_rate": 4.968002461983133e-05, "loss": 1.097, "step": 51077 }, { "epoch": 0.05, "learning_rate": 4.968001209284192e-05, "loss": 1.1353, "step": 51078 }, { "epoch": 0.05, "learning_rate": 4.967999956560886e-05, "loss": 1.4989, "step": 51079 }, { "epoch": 0.05, "learning_rate": 4.9679987038132186e-05, "loss": 1.316, "step": 51080 }, { "epoch": 0.05, "learning_rate": 4.967997451041188e-05, "loss": 0.9828, "step": 51081 }, { "epoch": 0.05, "learning_rate": 4.9679961982447934e-05, "loss": 1.1016, "step": 51082 }, { "epoch": 0.05, "learning_rate": 4.9679949454240364e-05, "loss": 0.4869, "step": 51083 }, { "epoch": 0.05, "learning_rate": 4.967993692578916e-05, "loss": 0.9535, "step": 51084 }, { "epoch": 0.05, "learning_rate": 4.9679924397094335e-05, "loss": 1.11, "step": 51085 }, { "epoch": 0.05, "learning_rate": 4.967991186815587e-05, "loss": 1.2258, "step": 51086 }, { "epoch": 0.05, "learning_rate": 4.967989933897378e-05, "loss": 1.2692, "step": 51087 }, { "epoch": 0.05, "learning_rate": 4.967988680954806e-05, "loss": 1.1446, "step": 51088 }, { "epoch": 0.05, "learning_rate": 4.9679874279878714e-05, "loss": 1.6347, "step": 51089 }, { "epoch": 0.05, "learning_rate": 4.9679861749965735e-05, "loss": 1.1313, "step": 51090 }, { "epoch": 0.05, "learning_rate": 4.967984921980913e-05, "loss": 1.1998, "step": 51091 }, { "epoch": 0.05, "learning_rate": 4.967983668940889e-05, "loss": 1.3673, "step": 51092 }, { "epoch": 0.05, "learning_rate": 4.967982415876502e-05, "loss": 0.9154, "step": 51093 }, { "epoch": 0.05, "learning_rate": 4.9679811627877526e-05, "loss": 0.9972, "step": 51094 }, { "epoch": 0.05, "learning_rate": 4.9679799096746405e-05, "loss": 0.6865, "step": 51095 }, { "epoch": 0.05, "learning_rate": 4.9679786565371656e-05, "loss": 0.9716, "step": 51096 }, { "epoch": 0.05, "learning_rate": 4.9679774033753274e-05, "loss": 1.1187, "step": 51097 }, { "epoch": 0.05, "learning_rate": 4.9679761501891265e-05, "loss": 0.6148, "step": 51098 }, { "epoch": 0.05, "learning_rate": 4.967974896978563e-05, "loss": 1.4177, "step": 51099 }, { "epoch": 0.05, "learning_rate": 4.9679736437436366e-05, "loss": 1.3852, "step": 51100 }, { "epoch": 0.05, "learning_rate": 4.9679723904843475e-05, "loss": 0.9843, "step": 51101 }, { "epoch": 0.05, "learning_rate": 4.967971137200695e-05, "loss": 0.9, "step": 51102 }, { "epoch": 0.05, "learning_rate": 4.96796988389268e-05, "loss": 0.3497, "step": 51103 }, { "epoch": 0.05, "learning_rate": 4.967968630560302e-05, "loss": 0.2502, "step": 51104 }, { "epoch": 0.05, "learning_rate": 4.9679673772035624e-05, "loss": 0.1976, "step": 51105 }, { "epoch": 0.05, "learning_rate": 4.967966123822459e-05, "loss": 0.1376, "step": 51106 }, { "epoch": 0.05, "learning_rate": 4.9679648704169933e-05, "loss": 0.3462, "step": 51107 }, { "epoch": 0.05, "learning_rate": 4.967963616987165e-05, "loss": 1.1134, "step": 51108 }, { "epoch": 0.05, "learning_rate": 4.9679623635329734e-05, "loss": 0.8918, "step": 51109 }, { "epoch": 0.05, "learning_rate": 4.96796111005442e-05, "loss": 1.1984, "step": 51110 }, { "epoch": 0.05, "learning_rate": 4.967959856551503e-05, "loss": 1.4285, "step": 51111 }, { "epoch": 0.05, "learning_rate": 4.967958603024224e-05, "loss": 1.3418, "step": 51112 }, { "epoch": 0.05, "learning_rate": 4.967957349472582e-05, "loss": 1.5465, "step": 51113 }, { "epoch": 0.05, "learning_rate": 4.967956095896578e-05, "loss": 1.253, "step": 51114 }, { "epoch": 0.05, "learning_rate": 4.967954842296211e-05, "loss": 0.8489, "step": 51115 }, { "epoch": 0.05, "learning_rate": 4.967953588671481e-05, "loss": 0.9126, "step": 51116 }, { "epoch": 0.05, "learning_rate": 4.9679523350223886e-05, "loss": 1.0982, "step": 51117 }, { "epoch": 0.05, "learning_rate": 4.9679510813489336e-05, "loss": 0.9771, "step": 51118 }, { "epoch": 0.05, "learning_rate": 4.967949827651116e-05, "loss": 1.8101, "step": 51119 }, { "epoch": 0.05, "learning_rate": 4.967948573928936e-05, "loss": 1.2585, "step": 51120 }, { "epoch": 0.05, "learning_rate": 4.967947320182393e-05, "loss": 1.3655, "step": 51121 }, { "epoch": 0.05, "learning_rate": 4.967946066411488e-05, "loss": 1.0539, "step": 51122 }, { "epoch": 0.05, "learning_rate": 4.967944812616221e-05, "loss": 1.3296, "step": 51123 }, { "epoch": 0.05, "learning_rate": 4.967943558796591e-05, "loss": 1.5091, "step": 51124 }, { "epoch": 0.05, "learning_rate": 4.9679423049525976e-05, "loss": 1.1279, "step": 51125 }, { "epoch": 0.05, "learning_rate": 4.967941051084243e-05, "loss": 1.1816, "step": 51126 }, { "epoch": 0.05, "learning_rate": 4.9679397971915246e-05, "loss": 1.4219, "step": 51127 }, { "epoch": 0.05, "learning_rate": 4.9679385432744453e-05, "loss": 1.0928, "step": 51128 }, { "epoch": 0.05, "learning_rate": 4.967937289333002e-05, "loss": 0.9247, "step": 51129 }, { "epoch": 0.05, "learning_rate": 4.9679360353671975e-05, "loss": 0.4011, "step": 51130 }, { "epoch": 0.05, "learning_rate": 4.96793478137703e-05, "loss": 0.2048, "step": 51131 }, { "epoch": 0.05, "learning_rate": 4.9679335273625e-05, "loss": 0.3707, "step": 51132 }, { "epoch": 0.05, "learning_rate": 4.967932273323608e-05, "loss": 0.3529, "step": 51133 }, { "epoch": 0.05, "learning_rate": 4.967931019260353e-05, "loss": 0.2276, "step": 51134 }, { "epoch": 0.05, "learning_rate": 4.9679297651727366e-05, "loss": 1.0658, "step": 51135 }, { "epoch": 0.05, "learning_rate": 4.9679285110607564e-05, "loss": 1.4644, "step": 51136 }, { "epoch": 0.05, "learning_rate": 4.967927256924415e-05, "loss": 1.1449, "step": 51137 }, { "epoch": 0.05, "learning_rate": 4.967926002763711e-05, "loss": 0.9968, "step": 51138 }, { "epoch": 0.05, "learning_rate": 4.967924748578645e-05, "loss": 1.8075, "step": 51139 }, { "epoch": 0.05, "learning_rate": 4.967923494369216e-05, "loss": 1.7016, "step": 51140 }, { "epoch": 0.05, "learning_rate": 4.9679222401354253e-05, "loss": 1.5921, "step": 51141 }, { "epoch": 0.05, "learning_rate": 4.967920985877272e-05, "loss": 0.6285, "step": 51142 }, { "epoch": 0.05, "learning_rate": 4.967919731594757e-05, "loss": 1.0731, "step": 51143 }, { "epoch": 0.05, "learning_rate": 4.9679184772878784e-05, "loss": 1.3896, "step": 51144 }, { "epoch": 0.05, "learning_rate": 4.967917222956639e-05, "loss": 1.3032, "step": 51145 }, { "epoch": 0.05, "learning_rate": 4.967915968601036e-05, "loss": 1.0397, "step": 51146 }, { "epoch": 0.05, "learning_rate": 4.967914714221071e-05, "loss": 0.6098, "step": 51147 }, { "epoch": 0.05, "learning_rate": 4.967913459816745e-05, "loss": 0.6256, "step": 51148 }, { "epoch": 0.05, "learning_rate": 4.967912205388056e-05, "loss": 1.6806, "step": 51149 }, { "epoch": 0.05, "learning_rate": 4.967910950935005e-05, "loss": 1.2765, "step": 51150 }, { "epoch": 0.05, "learning_rate": 4.967909696457592e-05, "loss": 0.8864, "step": 51151 }, { "epoch": 0.05, "learning_rate": 4.9679084419558166e-05, "loss": 0.4949, "step": 51152 }, { "epoch": 0.05, "learning_rate": 4.967907187429679e-05, "loss": 1.165, "step": 51153 }, { "epoch": 0.05, "learning_rate": 4.967905932879179e-05, "loss": 1.062, "step": 51154 }, { "epoch": 0.05, "learning_rate": 4.967904678304317e-05, "loss": 0.8651, "step": 51155 }, { "epoch": 0.05, "learning_rate": 4.967903423705093e-05, "loss": 1.527, "step": 51156 }, { "epoch": 0.05, "learning_rate": 4.967902169081506e-05, "loss": 1.0831, "step": 51157 }, { "epoch": 0.05, "learning_rate": 4.967900914433559e-05, "loss": 1.2524, "step": 51158 }, { "epoch": 0.05, "learning_rate": 4.967899659761248e-05, "loss": 0.9837, "step": 51159 }, { "epoch": 0.05, "learning_rate": 4.967898405064576e-05, "loss": 1.3002, "step": 51160 }, { "epoch": 0.05, "learning_rate": 4.967897150343542e-05, "loss": 1.1991, "step": 51161 }, { "epoch": 0.05, "learning_rate": 4.967895895598145e-05, "loss": 0.8827, "step": 51162 }, { "epoch": 0.05, "learning_rate": 4.967894640828387e-05, "loss": 0.961, "step": 51163 }, { "epoch": 0.05, "learning_rate": 4.967893386034266e-05, "loss": 1.6161, "step": 51164 }, { "epoch": 0.05, "learning_rate": 4.967892131215784e-05, "loss": 0.9892, "step": 51165 }, { "epoch": 0.05, "learning_rate": 4.967890876372939e-05, "loss": 0.9235, "step": 51166 }, { "epoch": 0.05, "learning_rate": 4.967889621505733e-05, "loss": 0.5809, "step": 51167 }, { "epoch": 0.05, "learning_rate": 4.9678883666141645e-05, "loss": 0.8172, "step": 51168 }, { "epoch": 0.05, "learning_rate": 4.967887111698234e-05, "loss": 0.6674, "step": 51169 }, { "epoch": 0.05, "learning_rate": 4.967885856757942e-05, "loss": 1.2598, "step": 51170 }, { "epoch": 0.05, "learning_rate": 4.9678846017932875e-05, "loss": 1.1598, "step": 51171 }, { "epoch": 0.05, "learning_rate": 4.967883346804271e-05, "loss": 1.2965, "step": 51172 }, { "epoch": 0.05, "learning_rate": 4.967882091790893e-05, "loss": 1.238, "step": 51173 }, { "epoch": 0.05, "learning_rate": 4.967880836753154e-05, "loss": 0.8515, "step": 51174 }, { "epoch": 0.05, "learning_rate": 4.967879581691052e-05, "loss": 0.8651, "step": 51175 }, { "epoch": 0.05, "learning_rate": 4.967878326604588e-05, "loss": 1.1469, "step": 51176 }, { "epoch": 0.05, "learning_rate": 4.9678770714937625e-05, "loss": 1.2685, "step": 51177 }, { "epoch": 0.05, "learning_rate": 4.9678758163585746e-05, "loss": 0.9574, "step": 51178 }, { "epoch": 0.05, "learning_rate": 4.9678745611990255e-05, "loss": 1.381, "step": 51179 }, { "epoch": 0.05, "learning_rate": 4.967873306015115e-05, "loss": 0.8333, "step": 51180 }, { "epoch": 0.05, "learning_rate": 4.9678720508068425e-05, "loss": 0.9428, "step": 51181 }, { "epoch": 0.05, "learning_rate": 4.967870795574207e-05, "loss": 0.622, "step": 51182 }, { "epoch": 0.05, "learning_rate": 4.9678695403172107e-05, "loss": 0.1731, "step": 51183 }, { "epoch": 0.05, "learning_rate": 4.967868285035853e-05, "loss": 0.7097, "step": 51184 }, { "epoch": 0.05, "learning_rate": 4.967867029730132e-05, "loss": 0.309, "step": 51185 }, { "epoch": 0.05, "learning_rate": 4.9678657744000515e-05, "loss": 0.7936, "step": 51186 }, { "epoch": 0.05, "learning_rate": 4.9678645190456075e-05, "loss": 0.9256, "step": 51187 }, { "epoch": 0.05, "learning_rate": 4.967863263666802e-05, "loss": 1.2712, "step": 51188 }, { "epoch": 0.05, "learning_rate": 4.967862008263635e-05, "loss": 0.8503, "step": 51189 }, { "epoch": 0.05, "learning_rate": 4.967860752836107e-05, "loss": 1.0199, "step": 51190 }, { "epoch": 0.05, "learning_rate": 4.967859497384216e-05, "loss": 1.3425, "step": 51191 }, { "epoch": 0.05, "learning_rate": 4.967858241907965e-05, "loss": 1.2229, "step": 51192 }, { "epoch": 0.05, "learning_rate": 4.967856986407351e-05, "loss": 1.1227, "step": 51193 }, { "epoch": 0.05, "learning_rate": 4.967855730882376e-05, "loss": 1.0078, "step": 51194 }, { "epoch": 0.05, "learning_rate": 4.967854475333039e-05, "loss": 1.3498, "step": 51195 }, { "epoch": 0.05, "learning_rate": 4.967853219759341e-05, "loss": 1.703, "step": 51196 }, { "epoch": 0.05, "learning_rate": 4.967851964161281e-05, "loss": 1.9526, "step": 51197 }, { "epoch": 0.05, "learning_rate": 4.967850708538859e-05, "loss": 1.4711, "step": 51198 }, { "epoch": 0.05, "learning_rate": 4.967849452892076e-05, "loss": 1.0276, "step": 51199 }, { "epoch": 0.05, "learning_rate": 4.967848197220931e-05, "loss": 1.6189, "step": 51200 }, { "epoch": 0.05, "learning_rate": 4.967846941525425e-05, "loss": 1.2693, "step": 51201 }, { "epoch": 0.05, "learning_rate": 4.9678456858055565e-05, "loss": 1.0353, "step": 51202 }, { "epoch": 0.05, "learning_rate": 4.9678444300613276e-05, "loss": 1.6369, "step": 51203 }, { "epoch": 0.05, "learning_rate": 4.967843174292737e-05, "loss": 1.1746, "step": 51204 }, { "epoch": 0.05, "learning_rate": 4.967841918499784e-05, "loss": 1.159, "step": 51205 }, { "epoch": 0.05, "learning_rate": 4.96784066268247e-05, "loss": 1.0813, "step": 51206 }, { "epoch": 0.05, "learning_rate": 4.9678394068407945e-05, "loss": 0.4728, "step": 51207 }, { "epoch": 0.05, "learning_rate": 4.9678381509747575e-05, "loss": 1.1495, "step": 51208 }, { "epoch": 0.05, "learning_rate": 4.967836895084359e-05, "loss": 1.0266, "step": 51209 }, { "epoch": 0.05, "learning_rate": 4.9678356391695995e-05, "loss": 0.7623, "step": 51210 }, { "epoch": 0.05, "learning_rate": 4.9678343832304785e-05, "loss": 1.0681, "step": 51211 }, { "epoch": 0.05, "learning_rate": 4.9678331272669955e-05, "loss": 1.0481, "step": 51212 }, { "epoch": 0.05, "learning_rate": 4.967831871279152e-05, "loss": 1.3205, "step": 51213 }, { "epoch": 0.05, "learning_rate": 4.967830615266946e-05, "loss": 1.4629, "step": 51214 }, { "epoch": 0.05, "learning_rate": 4.967829359230379e-05, "loss": 0.7186, "step": 51215 }, { "epoch": 0.05, "learning_rate": 4.9678281031694506e-05, "loss": 0.6039, "step": 51216 }, { "epoch": 0.05, "learning_rate": 4.9678268470841615e-05, "loss": 0.9807, "step": 51217 }, { "epoch": 0.05, "learning_rate": 4.967825590974511e-05, "loss": 1.1602, "step": 51218 }, { "epoch": 0.05, "learning_rate": 4.9678243348404986e-05, "loss": 1.1413, "step": 51219 }, { "epoch": 0.05, "learning_rate": 4.967823078682125e-05, "loss": 1.2787, "step": 51220 }, { "epoch": 0.05, "learning_rate": 4.96782182249939e-05, "loss": 0.8962, "step": 51221 }, { "epoch": 0.05, "learning_rate": 4.967820566292294e-05, "loss": 0.8647, "step": 51222 }, { "epoch": 0.05, "learning_rate": 4.967819310060836e-05, "loss": 1.161, "step": 51223 }, { "epoch": 0.05, "learning_rate": 4.9678180538050175e-05, "loss": 1.2833, "step": 51224 }, { "epoch": 0.05, "learning_rate": 4.967816797524837e-05, "loss": 1.1325, "step": 51225 }, { "epoch": 0.05, "learning_rate": 4.967815541220296e-05, "loss": 1.0025, "step": 51226 }, { "epoch": 0.05, "learning_rate": 4.967814284891393e-05, "loss": 0.2968, "step": 51227 }, { "epoch": 0.05, "learning_rate": 4.9678130285381294e-05, "loss": 0.4103, "step": 51228 }, { "epoch": 0.05, "learning_rate": 4.967811772160504e-05, "loss": 0.3899, "step": 51229 }, { "epoch": 0.05, "learning_rate": 4.967810515758519e-05, "loss": 0.2665, "step": 51230 }, { "epoch": 0.05, "learning_rate": 4.967809259332171e-05, "loss": 0.1886, "step": 51231 }, { "epoch": 0.05, "learning_rate": 4.9678080028814624e-05, "loss": 0.1956, "step": 51232 }, { "epoch": 0.05, "learning_rate": 4.967806746406393e-05, "loss": 0.1265, "step": 51233 }, { "epoch": 0.05, "learning_rate": 4.9678054899069626e-05, "loss": 0.0741, "step": 51234 }, { "epoch": 0.05, "learning_rate": 4.9678042333831706e-05, "loss": 0.1684, "step": 51235 }, { "epoch": 0.05, "learning_rate": 4.967802976835017e-05, "loss": 0.8752, "step": 51236 }, { "epoch": 0.05, "learning_rate": 4.9678017202625034e-05, "loss": 0.92, "step": 51237 }, { "epoch": 0.05, "learning_rate": 4.967800463665628e-05, "loss": 1.1918, "step": 51238 }, { "epoch": 0.05, "learning_rate": 4.9677992070443914e-05, "loss": 1.0074, "step": 51239 }, { "epoch": 0.05, "learning_rate": 4.967797950398794e-05, "loss": 1.1857, "step": 51240 }, { "epoch": 0.05, "learning_rate": 4.9677966937288355e-05, "loss": 1.0189, "step": 51241 }, { "epoch": 0.05, "learning_rate": 4.967795437034516e-05, "loss": 1.3097, "step": 51242 }, { "epoch": 0.05, "learning_rate": 4.9677941803158355e-05, "loss": 1.335, "step": 51243 }, { "epoch": 0.05, "learning_rate": 4.967792923572794e-05, "loss": 1.014, "step": 51244 }, { "epoch": 0.05, "learning_rate": 4.9677916668053914e-05, "loss": 1.0231, "step": 51245 }, { "epoch": 0.05, "learning_rate": 4.9677904100136274e-05, "loss": 1.4416, "step": 51246 }, { "epoch": 0.05, "learning_rate": 4.967789153197503e-05, "loss": 1.0449, "step": 51247 }, { "epoch": 0.05, "learning_rate": 4.967787896357018e-05, "loss": 0.8801, "step": 51248 }, { "epoch": 0.05, "learning_rate": 4.967786639492171e-05, "loss": 1.2409, "step": 51249 }, { "epoch": 0.05, "learning_rate": 4.967785382602963e-05, "loss": 1.0409, "step": 51250 }, { "epoch": 0.05, "learning_rate": 4.967784125689395e-05, "loss": 0.9542, "step": 51251 }, { "epoch": 0.05, "learning_rate": 4.967782868751466e-05, "loss": 1.0321, "step": 51252 }, { "epoch": 0.05, "learning_rate": 4.967781611789176e-05, "loss": 1.22, "step": 51253 }, { "epoch": 0.05, "learning_rate": 4.967780354802525e-05, "loss": 1.1845, "step": 51254 }, { "epoch": 0.05, "learning_rate": 4.967779097791513e-05, "loss": 1.2537, "step": 51255 }, { "epoch": 0.05, "learning_rate": 4.9677778407561396e-05, "loss": 1.0931, "step": 51256 }, { "epoch": 0.05, "learning_rate": 4.9677765836964055e-05, "loss": 1.565, "step": 51257 }, { "epoch": 0.05, "learning_rate": 4.9677753266123114e-05, "loss": 1.2032, "step": 51258 }, { "epoch": 0.05, "learning_rate": 4.967774069503855e-05, "loss": 1.2173, "step": 51259 }, { "epoch": 0.05, "learning_rate": 4.967772812371039e-05, "loss": 1.3598, "step": 51260 }, { "epoch": 0.05, "learning_rate": 4.9677715552138625e-05, "loss": 1.1215, "step": 51261 }, { "epoch": 0.05, "learning_rate": 4.9677702980323244e-05, "loss": 1.0754, "step": 51262 }, { "epoch": 0.05, "learning_rate": 4.9677690408264256e-05, "loss": 1.1838, "step": 51263 }, { "epoch": 0.05, "learning_rate": 4.967767783596166e-05, "loss": 0.9751, "step": 51264 }, { "epoch": 0.05, "learning_rate": 4.967766526341546e-05, "loss": 1.3585, "step": 51265 }, { "epoch": 0.05, "learning_rate": 4.9677652690625646e-05, "loss": 1.3011, "step": 51266 }, { "epoch": 0.05, "learning_rate": 4.9677640117592225e-05, "loss": 1.518, "step": 51267 }, { "epoch": 0.05, "learning_rate": 4.967762754431521e-05, "loss": 0.9362, "step": 51268 }, { "epoch": 0.05, "learning_rate": 4.967761497079457e-05, "loss": 1.1261, "step": 51269 }, { "epoch": 0.05, "learning_rate": 4.9677602397030335e-05, "loss": 1.0602, "step": 51270 }, { "epoch": 0.05, "learning_rate": 4.9677589823022494e-05, "loss": 0.9716, "step": 51271 }, { "epoch": 0.05, "learning_rate": 4.967757724877104e-05, "loss": 1.1069, "step": 51272 }, { "epoch": 0.05, "learning_rate": 4.967756467427598e-05, "loss": 1.1206, "step": 51273 }, { "epoch": 0.05, "learning_rate": 4.967755209953732e-05, "loss": 0.8445, "step": 51274 }, { "epoch": 0.05, "learning_rate": 4.9677539524555036e-05, "loss": 0.3779, "step": 51275 }, { "epoch": 0.05, "learning_rate": 4.967752694932917e-05, "loss": 1.5477, "step": 51276 }, { "epoch": 0.05, "learning_rate": 4.967751437385968e-05, "loss": 1.2733, "step": 51277 }, { "epoch": 0.05, "learning_rate": 4.9677501798146586e-05, "loss": 1.2421, "step": 51278 }, { "epoch": 0.05, "learning_rate": 4.967748922218989e-05, "loss": 0.7068, "step": 51279 }, { "epoch": 0.05, "learning_rate": 4.967747664598959e-05, "loss": 1.0271, "step": 51280 }, { "epoch": 0.05, "learning_rate": 4.967746406954569e-05, "loss": 1.3386, "step": 51281 }, { "epoch": 0.05, "learning_rate": 4.9677451492858175e-05, "loss": 1.0934, "step": 51282 }, { "epoch": 0.05, "learning_rate": 4.9677438915927053e-05, "loss": 1.3487, "step": 51283 }, { "epoch": 0.05, "learning_rate": 4.967742633875233e-05, "loss": 0.947, "step": 51284 }, { "epoch": 0.05, "learning_rate": 4.9677413761334005e-05, "loss": 1.2626, "step": 51285 }, { "epoch": 0.05, "learning_rate": 4.967740118367207e-05, "loss": 1.1332, "step": 51286 }, { "epoch": 0.05, "learning_rate": 4.967738860576653e-05, "loss": 1.1243, "step": 51287 }, { "epoch": 0.05, "learning_rate": 4.967737602761738e-05, "loss": 1.0855, "step": 51288 }, { "epoch": 0.05, "learning_rate": 4.9677363449224634e-05, "loss": 1.0618, "step": 51289 }, { "epoch": 0.05, "learning_rate": 4.967735087058828e-05, "loss": 1.0894, "step": 51290 }, { "epoch": 0.05, "learning_rate": 4.967733829170832e-05, "loss": 0.7039, "step": 51291 }, { "epoch": 0.05, "learning_rate": 4.967732571258476e-05, "loss": 1.147, "step": 51292 }, { "epoch": 0.05, "learning_rate": 4.9677313133217596e-05, "loss": 0.939, "step": 51293 }, { "epoch": 0.05, "learning_rate": 4.967730055360683e-05, "loss": 1.1615, "step": 51294 }, { "epoch": 0.05, "learning_rate": 4.967728797375245e-05, "loss": 0.8671, "step": 51295 }, { "epoch": 0.05, "learning_rate": 4.9677275393654474e-05, "loss": 0.6377, "step": 51296 }, { "epoch": 0.05, "learning_rate": 4.967726281331289e-05, "loss": 1.0993, "step": 51297 }, { "epoch": 0.05, "learning_rate": 4.9677250232727706e-05, "loss": 1.4376, "step": 51298 }, { "epoch": 0.05, "learning_rate": 4.9677237651898925e-05, "loss": 1.1181, "step": 51299 }, { "epoch": 0.05, "learning_rate": 4.9677225070826524e-05, "loss": 1.1494, "step": 51300 }, { "epoch": 0.05, "learning_rate": 4.967721248951053e-05, "loss": 0.9759, "step": 51301 }, { "epoch": 0.05, "learning_rate": 4.967719990795093e-05, "loss": 0.3724, "step": 51302 }, { "epoch": 0.05, "learning_rate": 4.9677187326147736e-05, "loss": 1.0439, "step": 51303 }, { "epoch": 0.05, "learning_rate": 4.967717474410093e-05, "loss": 0.5904, "step": 51304 }, { "epoch": 0.05, "learning_rate": 4.967716216181052e-05, "loss": 1.1281, "step": 51305 }, { "epoch": 0.05, "learning_rate": 4.9677149579276515e-05, "loss": 1.0348, "step": 51306 }, { "epoch": 0.05, "learning_rate": 4.967713699649891e-05, "loss": 1.0946, "step": 51307 }, { "epoch": 0.05, "learning_rate": 4.9677124413477695e-05, "loss": 1.1473, "step": 51308 }, { "epoch": 0.05, "learning_rate": 4.9677111830212875e-05, "loss": 0.8401, "step": 51309 }, { "epoch": 0.05, "learning_rate": 4.967709924670446e-05, "loss": 1.357, "step": 51310 }, { "epoch": 0.05, "learning_rate": 4.967708666295244e-05, "loss": 1.8027, "step": 51311 }, { "epoch": 0.05, "learning_rate": 4.967707407895682e-05, "loss": 2.1651, "step": 51312 }, { "epoch": 0.05, "learning_rate": 4.9677061494717595e-05, "loss": 1.1781, "step": 51313 }, { "epoch": 0.05, "learning_rate": 4.967704891023477e-05, "loss": 1.1859, "step": 51314 }, { "epoch": 0.05, "learning_rate": 4.967703632550834e-05, "loss": 1.1187, "step": 51315 }, { "epoch": 0.05, "learning_rate": 4.967702374053832e-05, "loss": 1.1285, "step": 51316 }, { "epoch": 0.05, "learning_rate": 4.9677011155324696e-05, "loss": 1.1239, "step": 51317 }, { "epoch": 0.05, "learning_rate": 4.967699856986746e-05, "loss": 1.2107, "step": 51318 }, { "epoch": 0.05, "learning_rate": 4.967698598416663e-05, "loss": 1.3121, "step": 51319 }, { "epoch": 0.05, "learning_rate": 4.96769733982222e-05, "loss": 1.4179, "step": 51320 }, { "epoch": 0.05, "learning_rate": 4.9676960812034165e-05, "loss": 0.9622, "step": 51321 }, { "epoch": 0.05, "learning_rate": 4.967694822560254e-05, "loss": 1.2228, "step": 51322 }, { "epoch": 0.05, "learning_rate": 4.96769356389273e-05, "loss": 1.205, "step": 51323 }, { "epoch": 0.05, "learning_rate": 4.967692305200847e-05, "loss": 1.0655, "step": 51324 }, { "epoch": 0.05, "learning_rate": 4.9676910464846036e-05, "loss": 0.9997, "step": 51325 }, { "epoch": 0.05, "learning_rate": 4.967689787744e-05, "loss": 1.1217, "step": 51326 }, { "epoch": 0.05, "learning_rate": 4.9676885289790365e-05, "loss": 1.0609, "step": 51327 }, { "epoch": 0.05, "learning_rate": 4.9676872701897134e-05, "loss": 1.3034, "step": 51328 }, { "epoch": 0.05, "learning_rate": 4.96768601137603e-05, "loss": 1.3339, "step": 51329 }, { "epoch": 0.05, "learning_rate": 4.9676847525379863e-05, "loss": 1.5942, "step": 51330 }, { "epoch": 0.05, "learning_rate": 4.967683493675583e-05, "loss": 1.3455, "step": 51331 }, { "epoch": 0.05, "learning_rate": 4.967682234788821e-05, "loss": 1.4004, "step": 51332 }, { "epoch": 0.05, "learning_rate": 4.9676809758776976e-05, "loss": 1.0652, "step": 51333 }, { "epoch": 0.05, "learning_rate": 4.9676797169422145e-05, "loss": 1.0441, "step": 51334 }, { "epoch": 0.05, "learning_rate": 4.967678457982372e-05, "loss": 0.7943, "step": 51335 }, { "epoch": 0.05, "learning_rate": 4.967677198998169e-05, "loss": 0.8131, "step": 51336 }, { "epoch": 0.05, "learning_rate": 4.9676759399896065e-05, "loss": 1.2113, "step": 51337 }, { "epoch": 0.05, "learning_rate": 4.967674680956684e-05, "loss": 1.3571, "step": 51338 }, { "epoch": 0.05, "learning_rate": 4.967673421899402e-05, "loss": 1.4612, "step": 51339 }, { "epoch": 0.05, "learning_rate": 4.967672162817759e-05, "loss": 1.3219, "step": 51340 }, { "epoch": 0.05, "learning_rate": 4.9676709037117576e-05, "loss": 0.9087, "step": 51341 }, { "epoch": 0.05, "learning_rate": 4.967669644581396e-05, "loss": 1.1763, "step": 51342 }, { "epoch": 0.05, "learning_rate": 4.967668385426674e-05, "loss": 1.087, "step": 51343 }, { "epoch": 0.05, "learning_rate": 4.9676671262475926e-05, "loss": 1.6602, "step": 51344 }, { "epoch": 0.05, "learning_rate": 4.967665867044152e-05, "loss": 1.0491, "step": 51345 }, { "epoch": 0.05, "learning_rate": 4.967664607816351e-05, "loss": 0.6539, "step": 51346 }, { "epoch": 0.05, "learning_rate": 4.9676633485641904e-05, "loss": 0.918, "step": 51347 }, { "epoch": 0.05, "learning_rate": 4.96766208928767e-05, "loss": 1.3068, "step": 51348 }, { "epoch": 0.05, "learning_rate": 4.96766082998679e-05, "loss": 1.2979, "step": 51349 }, { "epoch": 0.05, "learning_rate": 4.96765957066155e-05, "loss": 1.1172, "step": 51350 }, { "epoch": 0.05, "learning_rate": 4.967658311311951e-05, "loss": 1.1895, "step": 51351 }, { "epoch": 0.05, "learning_rate": 4.967657051937992e-05, "loss": 1.2033, "step": 51352 }, { "epoch": 0.05, "learning_rate": 4.967655792539673e-05, "loss": 1.2479, "step": 51353 }, { "epoch": 0.05, "learning_rate": 4.967654533116994e-05, "loss": 1.0503, "step": 51354 }, { "epoch": 0.05, "learning_rate": 4.9676532736699566e-05, "loss": 1.2506, "step": 51355 }, { "epoch": 0.05, "learning_rate": 4.9676520141985584e-05, "loss": 1.3388, "step": 51356 }, { "epoch": 0.05, "learning_rate": 4.967650754702801e-05, "loss": 1.238, "step": 51357 }, { "epoch": 0.05, "learning_rate": 4.967649495182684e-05, "loss": 1.3006, "step": 51358 }, { "epoch": 0.05, "learning_rate": 4.9676482356382074e-05, "loss": 1.1958, "step": 51359 }, { "epoch": 0.05, "learning_rate": 4.967646976069371e-05, "loss": 1.2931, "step": 51360 }, { "epoch": 0.05, "learning_rate": 4.967645716476176e-05, "loss": 0.9715, "step": 51361 }, { "epoch": 0.05, "learning_rate": 4.96764445685862e-05, "loss": 0.8823, "step": 51362 }, { "epoch": 0.05, "learning_rate": 4.967643197216706e-05, "loss": 1.1408, "step": 51363 }, { "epoch": 0.05, "learning_rate": 4.967641937550431e-05, "loss": 1.0322, "step": 51364 }, { "epoch": 0.05, "learning_rate": 4.967640677859797e-05, "loss": 1.0608, "step": 51365 }, { "epoch": 0.05, "learning_rate": 4.967639418144804e-05, "loss": 1.385, "step": 51366 }, { "epoch": 0.05, "learning_rate": 4.967638158405451e-05, "loss": 1.2776, "step": 51367 }, { "epoch": 0.05, "learning_rate": 4.9676368986417386e-05, "loss": 1.0054, "step": 51368 }, { "epoch": 0.05, "learning_rate": 4.967635638853667e-05, "loss": 0.8585, "step": 51369 }, { "epoch": 0.05, "learning_rate": 4.9676343790412354e-05, "loss": 1.073, "step": 51370 }, { "epoch": 0.05, "learning_rate": 4.967633119204445e-05, "loss": 0.8721, "step": 51371 }, { "epoch": 0.05, "learning_rate": 4.967631859343295e-05, "loss": 0.6106, "step": 51372 }, { "epoch": 0.05, "learning_rate": 4.967630599457785e-05, "loss": 0.798, "step": 51373 }, { "epoch": 0.05, "learning_rate": 4.967629339547916e-05, "loss": 1.0224, "step": 51374 }, { "epoch": 0.05, "learning_rate": 4.9676280796136875e-05, "loss": 0.9047, "step": 51375 }, { "epoch": 0.05, "learning_rate": 4.9676268196551e-05, "loss": 1.2478, "step": 51376 }, { "epoch": 0.05, "learning_rate": 4.967625559672152e-05, "loss": 1.2455, "step": 51377 }, { "epoch": 0.05, "learning_rate": 4.967624299664846e-05, "loss": 1.1408, "step": 51378 }, { "epoch": 0.05, "learning_rate": 4.9676230396331804e-05, "loss": 0.9535, "step": 51379 }, { "epoch": 0.05, "learning_rate": 4.9676217795771554e-05, "loss": 0.9902, "step": 51380 }, { "epoch": 0.05, "learning_rate": 4.9676205194967703e-05, "loss": 1.0224, "step": 51381 }, { "epoch": 0.05, "learning_rate": 4.967619259392027e-05, "loss": 1.1547, "step": 51382 }, { "epoch": 0.05, "learning_rate": 4.967617999262924e-05, "loss": 1.6006, "step": 51383 }, { "epoch": 0.05, "learning_rate": 4.967616739109461e-05, "loss": 1.1799, "step": 51384 }, { "epoch": 0.05, "learning_rate": 4.9676154789316384e-05, "loss": 1.3094, "step": 51385 }, { "epoch": 0.05, "learning_rate": 4.967614218729458e-05, "loss": 1.0946, "step": 51386 }, { "epoch": 0.05, "learning_rate": 4.967612958502918e-05, "loss": 1.2873, "step": 51387 }, { "epoch": 0.05, "learning_rate": 4.9676116982520184e-05, "loss": 1.2826, "step": 51388 }, { "epoch": 0.05, "learning_rate": 4.9676104379767596e-05, "loss": 1.2681, "step": 51389 }, { "epoch": 0.05, "learning_rate": 4.9676091776771414e-05, "loss": 0.9727, "step": 51390 }, { "epoch": 0.05, "learning_rate": 4.9676079173531646e-05, "loss": 1.3455, "step": 51391 }, { "epoch": 0.05, "learning_rate": 4.9676066570048285e-05, "loss": 1.4111, "step": 51392 }, { "epoch": 0.05, "learning_rate": 4.967605396632134e-05, "loss": 1.1477, "step": 51393 }, { "epoch": 0.05, "learning_rate": 4.9676041362350783e-05, "loss": 1.3077, "step": 51394 }, { "epoch": 0.05, "learning_rate": 4.967602875813665e-05, "loss": 1.8671, "step": 51395 }, { "epoch": 0.05, "learning_rate": 4.9676016153678916e-05, "loss": 1.3438, "step": 51396 }, { "epoch": 0.05, "learning_rate": 4.96760035489776e-05, "loss": 1.1089, "step": 51397 }, { "epoch": 0.05, "learning_rate": 4.967599094403269e-05, "loss": 1.0268, "step": 51398 }, { "epoch": 0.05, "learning_rate": 4.9675978338844184e-05, "loss": 1.3529, "step": 51399 }, { "epoch": 0.05, "learning_rate": 4.967596573341209e-05, "loss": 1.1758, "step": 51400 }, { "epoch": 0.05, "learning_rate": 4.9675953127736405e-05, "loss": 0.9817, "step": 51401 }, { "epoch": 0.05, "learning_rate": 4.967594052181713e-05, "loss": 1.2331, "step": 51402 }, { "epoch": 0.05, "learning_rate": 4.967592791565427e-05, "loss": 1.1422, "step": 51403 }, { "epoch": 0.05, "learning_rate": 4.967591530924781e-05, "loss": 1.044, "step": 51404 }, { "epoch": 0.05, "learning_rate": 4.9675902702597765e-05, "loss": 1.2161, "step": 51405 }, { "epoch": 0.05, "learning_rate": 4.9675890095704134e-05, "loss": 1.4879, "step": 51406 }, { "epoch": 0.05, "learning_rate": 4.967587748856691e-05, "loss": 1.0528, "step": 51407 }, { "epoch": 0.05, "learning_rate": 4.967586488118609e-05, "loss": 1.1027, "step": 51408 }, { "epoch": 0.05, "learning_rate": 4.967585227356168e-05, "loss": 1.2446, "step": 51409 }, { "epoch": 0.05, "learning_rate": 4.967583966569369e-05, "loss": 1.175, "step": 51410 }, { "epoch": 0.05, "learning_rate": 4.967582705758211e-05, "loss": 0.7925, "step": 51411 }, { "epoch": 0.05, "learning_rate": 4.9675814449226934e-05, "loss": 0.6086, "step": 51412 }, { "epoch": 0.05, "learning_rate": 4.9675801840628164e-05, "loss": 1.0638, "step": 51413 }, { "epoch": 0.05, "learning_rate": 4.9675789231785815e-05, "loss": 1.2768, "step": 51414 }, { "epoch": 0.05, "learning_rate": 4.9675776622699873e-05, "loss": 1.002, "step": 51415 }, { "epoch": 0.05, "learning_rate": 4.9675764013370345e-05, "loss": 0.7295, "step": 51416 }, { "epoch": 0.05, "learning_rate": 4.967575140379723e-05, "loss": 1.0113, "step": 51417 }, { "epoch": 0.05, "learning_rate": 4.967573879398052e-05, "loss": 1.0871, "step": 51418 }, { "epoch": 0.05, "learning_rate": 4.967572618392022e-05, "loss": 1.1494, "step": 51419 }, { "epoch": 0.05, "learning_rate": 4.967571357361635e-05, "loss": 1.203, "step": 51420 }, { "epoch": 0.05, "learning_rate": 4.9675700963068874e-05, "loss": 0.8616, "step": 51421 }, { "epoch": 0.05, "learning_rate": 4.967568835227781e-05, "loss": 1.1397, "step": 51422 }, { "epoch": 0.05, "learning_rate": 4.967567574124316e-05, "loss": 1.2737, "step": 51423 }, { "epoch": 0.05, "learning_rate": 4.967566312996492e-05, "loss": 1.5492, "step": 51424 }, { "epoch": 0.05, "learning_rate": 4.96756505184431e-05, "loss": 1.7575, "step": 51425 }, { "epoch": 0.05, "learning_rate": 4.9675637906677684e-05, "loss": 1.3538, "step": 51426 }, { "epoch": 0.05, "learning_rate": 4.967562529466869e-05, "loss": 1.3618, "step": 51427 }, { "epoch": 0.05, "learning_rate": 4.967561268241611e-05, "loss": 1.1917, "step": 51428 }, { "epoch": 0.05, "learning_rate": 4.967560006991992e-05, "loss": 1.1932, "step": 51429 }, { "epoch": 0.05, "learning_rate": 4.9675587457180165e-05, "loss": 0.8932, "step": 51430 }, { "epoch": 0.05, "learning_rate": 4.967557484419682e-05, "loss": 1.1393, "step": 51431 }, { "epoch": 0.05, "learning_rate": 4.9675562230969885e-05, "loss": 1.3877, "step": 51432 }, { "epoch": 0.05, "learning_rate": 4.967554961749936e-05, "loss": 1.6293, "step": 51433 }, { "epoch": 0.05, "learning_rate": 4.967553700378526e-05, "loss": 1.2452, "step": 51434 }, { "epoch": 0.05, "learning_rate": 4.967552438982756e-05, "loss": 1.1843, "step": 51435 }, { "epoch": 0.05, "learning_rate": 4.967551177562628e-05, "loss": 1.2521, "step": 51436 }, { "epoch": 0.05, "learning_rate": 4.967549916118141e-05, "loss": 1.3859, "step": 51437 }, { "epoch": 0.05, "learning_rate": 4.9675486546492956e-05, "loss": 1.46, "step": 51438 }, { "epoch": 0.05, "learning_rate": 4.9675473931560915e-05, "loss": 1.3401, "step": 51439 }, { "epoch": 0.05, "learning_rate": 4.9675461316385294e-05, "loss": 1.0102, "step": 51440 }, { "epoch": 0.05, "learning_rate": 4.967544870096608e-05, "loss": 0.9497, "step": 51441 }, { "epoch": 0.05, "learning_rate": 4.9675436085303286e-05, "loss": 0.3104, "step": 51442 }, { "epoch": 0.05, "learning_rate": 4.9675423469396906e-05, "loss": 0.5384, "step": 51443 }, { "epoch": 0.05, "learning_rate": 4.967541085324693e-05, "loss": 1.1496, "step": 51444 }, { "epoch": 0.05, "learning_rate": 4.967539823685338e-05, "loss": 0.9611, "step": 51445 }, { "epoch": 0.05, "learning_rate": 4.967538562021624e-05, "loss": 0.4345, "step": 51446 }, { "epoch": 0.05, "learning_rate": 4.967537300333552e-05, "loss": 0.9104, "step": 51447 }, { "epoch": 0.05, "learning_rate": 4.967536038621121e-05, "loss": 1.1066, "step": 51448 }, { "epoch": 0.05, "learning_rate": 4.967534776884332e-05, "loss": 0.9972, "step": 51449 }, { "epoch": 0.05, "learning_rate": 4.967533515123184e-05, "loss": 1.3702, "step": 51450 }, { "epoch": 0.05, "learning_rate": 4.967532253337678e-05, "loss": 1.0773, "step": 51451 }, { "epoch": 0.05, "learning_rate": 4.9675309915278134e-05, "loss": 1.0482, "step": 51452 }, { "epoch": 0.05, "learning_rate": 4.9675297296935904e-05, "loss": 1.3266, "step": 51453 }, { "epoch": 0.05, "learning_rate": 4.967528467835009e-05, "loss": 0.9693, "step": 51454 }, { "epoch": 0.05, "learning_rate": 4.967527205952069e-05, "loss": 0.9024, "step": 51455 }, { "epoch": 0.05, "learning_rate": 4.967525944044771e-05, "loss": 0.556, "step": 51456 }, { "epoch": 0.05, "learning_rate": 4.967524682113114e-05, "loss": 0.5969, "step": 51457 }, { "epoch": 0.05, "learning_rate": 4.967523420157099e-05, "loss": 0.541, "step": 51458 }, { "epoch": 0.05, "learning_rate": 4.967522158176726e-05, "loss": 0.3368, "step": 51459 }, { "epoch": 0.05, "learning_rate": 4.9675208961719945e-05, "loss": 0.4313, "step": 51460 }, { "epoch": 0.05, "learning_rate": 4.9675196341429044e-05, "loss": 0.5903, "step": 51461 }, { "epoch": 0.05, "learning_rate": 4.967518372089456e-05, "loss": 1.0827, "step": 51462 }, { "epoch": 0.05, "learning_rate": 4.967517110011649e-05, "loss": 1.2698, "step": 51463 }, { "epoch": 0.05, "learning_rate": 4.967515847909484e-05, "loss": 0.5502, "step": 51464 }, { "epoch": 0.05, "learning_rate": 4.9675145857829607e-05, "loss": 0.7009, "step": 51465 }, { "epoch": 0.05, "learning_rate": 4.9675133236320795e-05, "loss": 1.0663, "step": 51466 }, { "epoch": 0.05, "learning_rate": 4.96751206145684e-05, "loss": 1.1567, "step": 51467 }, { "epoch": 0.05, "learning_rate": 4.9675107992572426e-05, "loss": 0.9781, "step": 51468 }, { "epoch": 0.05, "learning_rate": 4.9675095370332855e-05, "loss": 1.3009, "step": 51469 }, { "epoch": 0.05, "learning_rate": 4.967508274784971e-05, "loss": 1.0115, "step": 51470 }, { "epoch": 0.05, "learning_rate": 4.967507012512299e-05, "loss": 1.4421, "step": 51471 }, { "epoch": 0.05, "learning_rate": 4.967505750215268e-05, "loss": 1.2401, "step": 51472 }, { "epoch": 0.05, "learning_rate": 4.967504487893879e-05, "loss": 0.7448, "step": 51473 }, { "epoch": 0.05, "learning_rate": 4.967503225548132e-05, "loss": 1.1025, "step": 51474 }, { "epoch": 0.05, "learning_rate": 4.9675019631780266e-05, "loss": 1.5867, "step": 51475 }, { "epoch": 0.05, "learning_rate": 4.967500700783563e-05, "loss": 1.3637, "step": 51476 }, { "epoch": 0.05, "learning_rate": 4.967499438364742e-05, "loss": 1.1141, "step": 51477 }, { "epoch": 0.05, "learning_rate": 4.967498175921562e-05, "loss": 1.0799, "step": 51478 }, { "epoch": 0.05, "learning_rate": 4.9674969134540236e-05, "loss": 1.0203, "step": 51479 }, { "epoch": 0.05, "learning_rate": 4.9674956509621284e-05, "loss": 0.9004, "step": 51480 }, { "epoch": 0.05, "learning_rate": 4.9674943884458744e-05, "loss": 1.094, "step": 51481 }, { "epoch": 0.05, "learning_rate": 4.9674931259052626e-05, "loss": 1.408, "step": 51482 }, { "epoch": 0.05, "learning_rate": 4.967491863340292e-05, "loss": 0.8922, "step": 51483 }, { "epoch": 0.05, "learning_rate": 4.967490600750965e-05, "loss": 1.321, "step": 51484 }, { "epoch": 0.05, "learning_rate": 4.9674893381372785e-05, "loss": 1.4032, "step": 51485 }, { "epoch": 0.05, "learning_rate": 4.967488075499234e-05, "loss": 1.3183, "step": 51486 }, { "epoch": 0.05, "learning_rate": 4.9674868128368324e-05, "loss": 1.0688, "step": 51487 }, { "epoch": 0.05, "learning_rate": 4.9674855501500715e-05, "loss": 0.8893, "step": 51488 }, { "epoch": 0.05, "learning_rate": 4.967484287438954e-05, "loss": 0.9818, "step": 51489 }, { "epoch": 0.05, "learning_rate": 4.967483024703478e-05, "loss": 1.4456, "step": 51490 }, { "epoch": 0.05, "learning_rate": 4.9674817619436435e-05, "loss": 1.2699, "step": 51491 }, { "epoch": 0.05, "learning_rate": 4.9674804991594514e-05, "loss": 0.9239, "step": 51492 }, { "epoch": 0.05, "learning_rate": 4.967479236350901e-05, "loss": 0.554, "step": 51493 }, { "epoch": 0.05, "learning_rate": 4.9674779735179946e-05, "loss": 1.3473, "step": 51494 }, { "epoch": 0.05, "learning_rate": 4.9674767106607286e-05, "loss": 1.3245, "step": 51495 }, { "epoch": 0.05, "learning_rate": 4.967475447779105e-05, "loss": 1.2106, "step": 51496 }, { "epoch": 0.05, "learning_rate": 4.9674741848731234e-05, "loss": 0.9428, "step": 51497 }, { "epoch": 0.05, "learning_rate": 4.967472921942784e-05, "loss": 1.0283, "step": 51498 }, { "epoch": 0.05, "learning_rate": 4.9674716589880864e-05, "loss": 0.8604, "step": 51499 }, { "epoch": 0.05, "learning_rate": 4.967470396009032e-05, "loss": 1.2091, "step": 51500 }, { "epoch": 0.05, "eval_loss": 1.0775607824325562, "eval_runtime": 128.4334, "eval_samples_per_second": 10.784, "eval_steps_per_second": 5.396, "step": 51500 }, { "epoch": 0.05, "learning_rate": 4.967469133005619e-05, "loss": 0.9495, "step": 51501 }, { "epoch": 0.05, "learning_rate": 4.967467869977848e-05, "loss": 1.2246, "step": 51502 }, { "epoch": 0.05, "learning_rate": 4.9674666069257196e-05, "loss": 1.0874, "step": 51503 }, { "epoch": 0.05, "learning_rate": 4.967465343849233e-05, "loss": 1.0463, "step": 51504 }, { "epoch": 0.05, "learning_rate": 4.96746408074839e-05, "loss": 0.9654, "step": 51505 }, { "epoch": 0.05, "learning_rate": 4.9674628176231876e-05, "loss": 1.0388, "step": 51506 }, { "epoch": 0.05, "learning_rate": 4.967461554473628e-05, "loss": 1.0211, "step": 51507 }, { "epoch": 0.05, "learning_rate": 4.967460291299711e-05, "loss": 0.7911, "step": 51508 }, { "epoch": 0.05, "learning_rate": 4.9674590281014354e-05, "loss": 1.0714, "step": 51509 }, { "epoch": 0.05, "learning_rate": 4.9674577648788034e-05, "loss": 0.9485, "step": 51510 }, { "epoch": 0.05, "learning_rate": 4.967456501631813e-05, "loss": 1.3072, "step": 51511 }, { "epoch": 0.05, "learning_rate": 4.967455238360464e-05, "loss": 1.5207, "step": 51512 }, { "epoch": 0.05, "learning_rate": 4.9674539750647584e-05, "loss": 1.3495, "step": 51513 }, { "epoch": 0.05, "learning_rate": 4.967452711744695e-05, "loss": 1.4105, "step": 51514 }, { "epoch": 0.05, "learning_rate": 4.967451448400274e-05, "loss": 1.4757, "step": 51515 }, { "epoch": 0.05, "learning_rate": 4.967450185031495e-05, "loss": 1.0526, "step": 51516 }, { "epoch": 0.05, "learning_rate": 4.967448921638359e-05, "loss": 4.2783, "step": 51517 }, { "epoch": 0.05, "learning_rate": 4.967447658220865e-05, "loss": 5.031, "step": 51518 }, { "epoch": 0.05, "learning_rate": 4.967446394779013e-05, "loss": 1.2627, "step": 51519 }, { "epoch": 0.05, "learning_rate": 4.967445131312804e-05, "loss": 1.2538, "step": 51520 }, { "epoch": 0.05, "learning_rate": 4.967443867822237e-05, "loss": 0.8417, "step": 51521 }, { "epoch": 0.05, "learning_rate": 4.967442604307313e-05, "loss": 0.7773, "step": 51522 }, { "epoch": 0.05, "learning_rate": 4.967441340768031e-05, "loss": 1.1533, "step": 51523 }, { "epoch": 0.05, "learning_rate": 4.9674400772043914e-05, "loss": 0.9936, "step": 51524 }, { "epoch": 0.05, "learning_rate": 4.967438813616394e-05, "loss": 1.1728, "step": 51525 }, { "epoch": 0.05, "learning_rate": 4.9674375500040396e-05, "loss": 1.2406, "step": 51526 }, { "epoch": 0.05, "learning_rate": 4.967436286367328e-05, "loss": 0.9511, "step": 51527 }, { "epoch": 0.05, "learning_rate": 4.967435022706258e-05, "loss": 1.5075, "step": 51528 }, { "epoch": 0.05, "learning_rate": 4.967433759020832e-05, "loss": 1.3463, "step": 51529 }, { "epoch": 0.05, "learning_rate": 4.967432495311047e-05, "loss": 1.1087, "step": 51530 }, { "epoch": 0.05, "learning_rate": 4.967431231576905e-05, "loss": 1.6079, "step": 51531 }, { "epoch": 0.05, "learning_rate": 4.9674299678184055e-05, "loss": 1.2773, "step": 51532 }, { "epoch": 0.05, "learning_rate": 4.967428704035549e-05, "loss": 1.2556, "step": 51533 }, { "epoch": 0.05, "learning_rate": 4.967427440228335e-05, "loss": 1.035, "step": 51534 }, { "epoch": 0.05, "learning_rate": 4.967426176396763e-05, "loss": 1.1236, "step": 51535 }, { "epoch": 0.05, "learning_rate": 4.9674249125408344e-05, "loss": 1.0517, "step": 51536 }, { "epoch": 0.05, "learning_rate": 4.967423648660548e-05, "loss": 0.8382, "step": 51537 }, { "epoch": 0.05, "learning_rate": 4.9674223847559045e-05, "loss": 0.5889, "step": 51538 }, { "epoch": 0.05, "learning_rate": 4.9674211208269026e-05, "loss": 0.3865, "step": 51539 }, { "epoch": 0.05, "learning_rate": 4.967419856873545e-05, "loss": 0.4128, "step": 51540 }, { "epoch": 0.05, "learning_rate": 4.967418592895829e-05, "loss": 0.4484, "step": 51541 }, { "epoch": 0.05, "learning_rate": 4.967417328893756e-05, "loss": 0.4549, "step": 51542 }, { "epoch": 0.05, "learning_rate": 4.967416064867325e-05, "loss": 0.3912, "step": 51543 }, { "epoch": 0.05, "learning_rate": 4.9674148008165375e-05, "loss": 0.7654, "step": 51544 }, { "epoch": 0.05, "learning_rate": 4.967413536741393e-05, "loss": 1.0878, "step": 51545 }, { "epoch": 0.05, "learning_rate": 4.9674122726418904e-05, "loss": 2.0341, "step": 51546 }, { "epoch": 0.05, "learning_rate": 4.9674110085180317e-05, "loss": 1.1783, "step": 51547 }, { "epoch": 0.05, "learning_rate": 4.967409744369814e-05, "loss": 1.4286, "step": 51548 }, { "epoch": 0.05, "learning_rate": 4.967408480197241e-05, "loss": 1.121, "step": 51549 }, { "epoch": 0.05, "learning_rate": 4.96740721600031e-05, "loss": 1.31, "step": 51550 }, { "epoch": 0.05, "learning_rate": 4.967405951779021e-05, "loss": 1.1899, "step": 51551 }, { "epoch": 0.05, "learning_rate": 4.967404687533376e-05, "loss": 0.949, "step": 51552 }, { "epoch": 0.05, "learning_rate": 4.967403423263373e-05, "loss": 1.2074, "step": 51553 }, { "epoch": 0.05, "learning_rate": 4.967402158969013e-05, "loss": 1.1903, "step": 51554 }, { "epoch": 0.05, "learning_rate": 4.967400894650296e-05, "loss": 1.0921, "step": 51555 }, { "epoch": 0.05, "learning_rate": 4.967399630307222e-05, "loss": 0.7623, "step": 51556 }, { "epoch": 0.05, "learning_rate": 4.96739836593979e-05, "loss": 1.0981, "step": 51557 }, { "epoch": 0.05, "learning_rate": 4.967397101548002e-05, "loss": 1.0838, "step": 51558 }, { "epoch": 0.05, "learning_rate": 4.967395837131856e-05, "loss": 1.261, "step": 51559 }, { "epoch": 0.05, "learning_rate": 4.967394572691354e-05, "loss": 1.3782, "step": 51560 }, { "epoch": 0.05, "learning_rate": 4.967393308226494e-05, "loss": 1.1777, "step": 51561 }, { "epoch": 0.05, "learning_rate": 4.9673920437372774e-05, "loss": 0.8844, "step": 51562 }, { "epoch": 0.05, "learning_rate": 4.9673907792237035e-05, "loss": 1.0644, "step": 51563 }, { "epoch": 0.05, "learning_rate": 4.967389514685773e-05, "loss": 0.9007, "step": 51564 }, { "epoch": 0.05, "learning_rate": 4.9673882501234846e-05, "loss": 1.0957, "step": 51565 }, { "epoch": 0.05, "learning_rate": 4.9673869855368396e-05, "loss": 0.8572, "step": 51566 }, { "epoch": 0.05, "learning_rate": 4.967385720925838e-05, "loss": 0.9788, "step": 51567 }, { "epoch": 0.05, "learning_rate": 4.967384456290479e-05, "loss": 1.0188, "step": 51568 }, { "epoch": 0.05, "learning_rate": 4.9673831916307635e-05, "loss": 0.742, "step": 51569 }, { "epoch": 0.05, "learning_rate": 4.96738192694669e-05, "loss": 0.8322, "step": 51570 }, { "epoch": 0.05, "learning_rate": 4.96738066223826e-05, "loss": 1.2437, "step": 51571 }, { "epoch": 0.05, "learning_rate": 4.967379397505474e-05, "loss": 1.3505, "step": 51572 }, { "epoch": 0.05, "learning_rate": 4.96737813274833e-05, "loss": 1.3223, "step": 51573 }, { "epoch": 0.05, "learning_rate": 4.9673768679668294e-05, "loss": 1.212, "step": 51574 }, { "epoch": 0.05, "learning_rate": 4.9673756031609716e-05, "loss": 1.0744, "step": 51575 }, { "epoch": 0.05, "learning_rate": 4.967374338330757e-05, "loss": 1.2489, "step": 51576 }, { "epoch": 0.05, "learning_rate": 4.967373073476186e-05, "loss": 1.0722, "step": 51577 }, { "epoch": 0.05, "learning_rate": 4.967371808597258e-05, "loss": 1.2661, "step": 51578 }, { "epoch": 0.05, "learning_rate": 4.9673705436939736e-05, "loss": 1.29, "step": 51579 }, { "epoch": 0.05, "learning_rate": 4.9673692787663314e-05, "loss": 1.1954, "step": 51580 }, { "epoch": 0.05, "learning_rate": 4.9673680138143325e-05, "loss": 1.0933, "step": 51581 }, { "epoch": 0.05, "learning_rate": 4.967366748837977e-05, "loss": 1.2136, "step": 51582 }, { "epoch": 0.05, "learning_rate": 4.9673654838372645e-05, "loss": 1.8135, "step": 51583 }, { "epoch": 0.05, "learning_rate": 4.967364218812195e-05, "loss": 1.366, "step": 51584 }, { "epoch": 0.05, "learning_rate": 4.96736295376277e-05, "loss": 1.395, "step": 51585 }, { "epoch": 0.05, "learning_rate": 4.9673616886889875e-05, "loss": 1.2662, "step": 51586 }, { "epoch": 0.05, "learning_rate": 4.967360423590848e-05, "loss": 1.9725, "step": 51587 }, { "epoch": 0.05, "learning_rate": 4.967359158468351e-05, "loss": 1.5472, "step": 51588 }, { "epoch": 0.05, "learning_rate": 4.967357893321499e-05, "loss": 0.9361, "step": 51589 }, { "epoch": 0.05, "learning_rate": 4.9673566281502894e-05, "loss": 0.4452, "step": 51590 }, { "epoch": 0.05, "learning_rate": 4.9673553629547226e-05, "loss": 1.2479, "step": 51591 }, { "epoch": 0.05, "learning_rate": 4.9673540977348e-05, "loss": 0.9601, "step": 51592 }, { "epoch": 0.05, "learning_rate": 4.96735283249052e-05, "loss": 0.8575, "step": 51593 }, { "epoch": 0.05, "learning_rate": 4.967351567221884e-05, "loss": 1.343, "step": 51594 }, { "epoch": 0.05, "learning_rate": 4.967350301928891e-05, "loss": 1.1755, "step": 51595 }, { "epoch": 0.05, "learning_rate": 4.967349036611541e-05, "loss": 0.2504, "step": 51596 }, { "epoch": 0.05, "learning_rate": 4.967347771269835e-05, "loss": 0.208, "step": 51597 }, { "epoch": 0.05, "learning_rate": 4.967346505903772e-05, "loss": 0.9546, "step": 51598 }, { "epoch": 0.05, "learning_rate": 4.967345240513353e-05, "loss": 1.0824, "step": 51599 }, { "epoch": 0.05, "learning_rate": 4.9673439750985765e-05, "loss": 1.2148, "step": 51600 }, { "epoch": 0.05, "learning_rate": 4.9673427096594436e-05, "loss": 1.0429, "step": 51601 }, { "epoch": 0.05, "learning_rate": 4.967341444195954e-05, "loss": 1.7596, "step": 51602 }, { "epoch": 0.05, "learning_rate": 4.967340178708109e-05, "loss": 1.1402, "step": 51603 }, { "epoch": 0.05, "learning_rate": 4.967338913195906e-05, "loss": 1.3644, "step": 51604 }, { "epoch": 0.05, "learning_rate": 4.967337647659348e-05, "loss": 0.8953, "step": 51605 }, { "epoch": 0.05, "learning_rate": 4.967336382098432e-05, "loss": 0.9681, "step": 51606 }, { "epoch": 0.05, "learning_rate": 4.9673351165131604e-05, "loss": 0.9175, "step": 51607 }, { "epoch": 0.05, "learning_rate": 4.9673338509035314e-05, "loss": 1.0606, "step": 51608 }, { "epoch": 0.05, "learning_rate": 4.967332585269548e-05, "loss": 1.3129, "step": 51609 }, { "epoch": 0.05, "learning_rate": 4.967331319611206e-05, "loss": 0.8927, "step": 51610 }, { "epoch": 0.05, "learning_rate": 4.9673300539285085e-05, "loss": 1.2375, "step": 51611 }, { "epoch": 0.05, "learning_rate": 4.967328788221454e-05, "loss": 1.3083, "step": 51612 }, { "epoch": 0.05, "learning_rate": 4.967327522490043e-05, "loss": 1.4061, "step": 51613 }, { "epoch": 0.05, "learning_rate": 4.9673262567342756e-05, "loss": 1.5069, "step": 51614 }, { "epoch": 0.05, "learning_rate": 4.967324990954153e-05, "loss": 1.2746, "step": 51615 }, { "epoch": 0.05, "learning_rate": 4.967323725149673e-05, "loss": 1.0182, "step": 51616 }, { "epoch": 0.05, "learning_rate": 4.967322459320837e-05, "loss": 1.0279, "step": 51617 }, { "epoch": 0.05, "learning_rate": 4.967321193467644e-05, "loss": 0.9843, "step": 51618 }, { "epoch": 0.05, "learning_rate": 4.967319927590095e-05, "loss": 1.0477, "step": 51619 }, { "epoch": 0.05, "learning_rate": 4.96731866168819e-05, "loss": 1.1079, "step": 51620 }, { "epoch": 0.05, "learning_rate": 4.967317395761928e-05, "loss": 1.2286, "step": 51621 }, { "epoch": 0.05, "learning_rate": 4.96731612981131e-05, "loss": 1.0663, "step": 51622 }, { "epoch": 0.05, "learning_rate": 4.967314863836336e-05, "loss": 1.1788, "step": 51623 }, { "epoch": 0.05, "learning_rate": 4.9673135978370054e-05, "loss": 1.0122, "step": 51624 }, { "epoch": 0.05, "learning_rate": 4.967312331813319e-05, "loss": 1.1037, "step": 51625 }, { "epoch": 0.05, "learning_rate": 4.9673110657652756e-05, "loss": 1.0854, "step": 51626 }, { "epoch": 0.05, "learning_rate": 4.9673097996928765e-05, "loss": 1.0167, "step": 51627 }, { "epoch": 0.05, "learning_rate": 4.967308533596121e-05, "loss": 1.0612, "step": 51628 }, { "epoch": 0.05, "learning_rate": 4.9673072674750085e-05, "loss": 1.2546, "step": 51629 }, { "epoch": 0.05, "learning_rate": 4.967306001329541e-05, "loss": 0.7701, "step": 51630 }, { "epoch": 0.05, "learning_rate": 4.967304735159717e-05, "loss": 1.1011, "step": 51631 }, { "epoch": 0.05, "learning_rate": 4.967303468965536e-05, "loss": 1.0987, "step": 51632 }, { "epoch": 0.05, "learning_rate": 4.9673022027469994e-05, "loss": 0.9272, "step": 51633 }, { "epoch": 0.05, "learning_rate": 4.967300936504107e-05, "loss": 1.3195, "step": 51634 }, { "epoch": 0.05, "learning_rate": 4.967299670236858e-05, "loss": 1.1218, "step": 51635 }, { "epoch": 0.05, "learning_rate": 4.967298403945253e-05, "loss": 1.3276, "step": 51636 }, { "epoch": 0.05, "learning_rate": 4.9672971376292914e-05, "loss": 1.1121, "step": 51637 }, { "epoch": 0.05, "learning_rate": 4.967295871288974e-05, "loss": 0.9125, "step": 51638 }, { "epoch": 0.05, "learning_rate": 4.967294604924301e-05, "loss": 1.2063, "step": 51639 }, { "epoch": 0.05, "learning_rate": 4.967293338535271e-05, "loss": 1.7595, "step": 51640 }, { "epoch": 0.05, "learning_rate": 4.967292072121885e-05, "loss": 1.0064, "step": 51641 }, { "epoch": 0.05, "learning_rate": 4.967290805684144e-05, "loss": 1.2907, "step": 51642 }, { "epoch": 0.05, "learning_rate": 4.9672895392220466e-05, "loss": 1.3186, "step": 51643 }, { "epoch": 0.05, "learning_rate": 4.967288272735593e-05, "loss": 1.1114, "step": 51644 }, { "epoch": 0.05, "learning_rate": 4.967287006224782e-05, "loss": 1.0167, "step": 51645 }, { "epoch": 0.05, "learning_rate": 4.9672857396896165e-05, "loss": 1.1433, "step": 51646 }, { "epoch": 0.05, "learning_rate": 4.967284473130095e-05, "loss": 1.0962, "step": 51647 }, { "epoch": 0.05, "learning_rate": 4.967283206546217e-05, "loss": 1.0868, "step": 51648 }, { "epoch": 0.05, "learning_rate": 4.967281939937983e-05, "loss": 1.023, "step": 51649 }, { "epoch": 0.05, "learning_rate": 4.967280673305394e-05, "loss": 1.1709, "step": 51650 }, { "epoch": 0.05, "learning_rate": 4.967279406648448e-05, "loss": 1.4403, "step": 51651 }, { "epoch": 0.05, "learning_rate": 4.9672781399671464e-05, "loss": 1.036, "step": 51652 }, { "epoch": 0.05, "learning_rate": 4.9672768732614885e-05, "loss": 1.1629, "step": 51653 }, { "epoch": 0.05, "learning_rate": 4.9672756065314754e-05, "loss": 1.2107, "step": 51654 }, { "epoch": 0.05, "learning_rate": 4.967274339777106e-05, "loss": 1.2262, "step": 51655 }, { "epoch": 0.05, "learning_rate": 4.96727307299838e-05, "loss": 0.8381, "step": 51656 }, { "epoch": 0.05, "learning_rate": 4.967271806195299e-05, "loss": 0.8597, "step": 51657 }, { "epoch": 0.05, "learning_rate": 4.967270539367862e-05, "loss": 1.2896, "step": 51658 }, { "epoch": 0.05, "learning_rate": 4.967269272516069e-05, "loss": 1.5411, "step": 51659 }, { "epoch": 0.05, "learning_rate": 4.96726800563992e-05, "loss": 1.3167, "step": 51660 }, { "epoch": 0.05, "learning_rate": 4.967266738739416e-05, "loss": 1.3008, "step": 51661 }, { "epoch": 0.05, "learning_rate": 4.967265471814555e-05, "loss": 0.9603, "step": 51662 }, { "epoch": 0.05, "learning_rate": 4.9672642048653394e-05, "loss": 0.7181, "step": 51663 }, { "epoch": 0.05, "learning_rate": 4.967262937891767e-05, "loss": 0.5362, "step": 51664 }, { "epoch": 0.05, "learning_rate": 4.9672616708938394e-05, "loss": 0.3771, "step": 51665 }, { "epoch": 0.05, "learning_rate": 4.967260403871556e-05, "loss": 0.3867, "step": 51666 }, { "epoch": 0.05, "learning_rate": 4.9672591368249164e-05, "loss": 0.2507, "step": 51667 }, { "epoch": 0.05, "learning_rate": 4.967257869753922e-05, "loss": 0.3741, "step": 51668 }, { "epoch": 0.05, "learning_rate": 4.967256602658571e-05, "loss": 1.0398, "step": 51669 }, { "epoch": 0.05, "learning_rate": 4.967255335538864e-05, "loss": 1.3958, "step": 51670 }, { "epoch": 0.05, "learning_rate": 4.967254068394802e-05, "loss": 1.3549, "step": 51671 }, { "epoch": 0.05, "learning_rate": 4.9672528012263844e-05, "loss": 1.3708, "step": 51672 }, { "epoch": 0.05, "learning_rate": 4.967251534033611e-05, "loss": 0.9993, "step": 51673 }, { "epoch": 0.05, "learning_rate": 4.967250266816481e-05, "loss": 1.3055, "step": 51674 }, { "epoch": 0.05, "learning_rate": 4.967248999574996e-05, "loss": 1.0037, "step": 51675 }, { "epoch": 0.05, "learning_rate": 4.967247732309156e-05, "loss": 1.0175, "step": 51676 }, { "epoch": 0.05, "learning_rate": 4.9672464650189596e-05, "loss": 1.0363, "step": 51677 }, { "epoch": 0.05, "learning_rate": 4.967245197704408e-05, "loss": 0.4171, "step": 51678 }, { "epoch": 0.05, "learning_rate": 4.9672439303655005e-05, "loss": 0.9984, "step": 51679 }, { "epoch": 0.05, "learning_rate": 4.9672426630022374e-05, "loss": 1.4126, "step": 51680 }, { "epoch": 0.05, "learning_rate": 4.967241395614619e-05, "loss": 1.0829, "step": 51681 }, { "epoch": 0.05, "learning_rate": 4.967240128202645e-05, "loss": 0.9459, "step": 51682 }, { "epoch": 0.05, "learning_rate": 4.9672388607663154e-05, "loss": 1.3586, "step": 51683 }, { "epoch": 0.05, "learning_rate": 4.9672375933056306e-05, "loss": 1.0954, "step": 51684 }, { "epoch": 0.05, "learning_rate": 4.967236325820589e-05, "loss": 0.5704, "step": 51685 }, { "epoch": 0.05, "learning_rate": 4.967235058311193e-05, "loss": 1.1308, "step": 51686 }, { "epoch": 0.05, "learning_rate": 4.9672337907774415e-05, "loss": 1.3166, "step": 51687 }, { "epoch": 0.05, "learning_rate": 4.967232523219334e-05, "loss": 1.4554, "step": 51688 }, { "epoch": 0.05, "learning_rate": 4.9672312556368714e-05, "loss": 1.5819, "step": 51689 }, { "epoch": 0.05, "learning_rate": 4.967229988030053e-05, "loss": 1.0802, "step": 51690 }, { "epoch": 0.05, "learning_rate": 4.9672287203988796e-05, "loss": 1.3267, "step": 51691 }, { "epoch": 0.05, "learning_rate": 4.96722745274335e-05, "loss": 1.1801, "step": 51692 }, { "epoch": 0.05, "learning_rate": 4.9672261850634655e-05, "loss": 0.5496, "step": 51693 }, { "epoch": 0.05, "learning_rate": 4.9672249173592256e-05, "loss": 0.8226, "step": 51694 }, { "epoch": 0.05, "learning_rate": 4.9672236496306305e-05, "loss": 1.2306, "step": 51695 }, { "epoch": 0.05, "learning_rate": 4.96722238187768e-05, "loss": 0.8018, "step": 51696 }, { "epoch": 0.05, "learning_rate": 4.967221114100373e-05, "loss": 1.1455, "step": 51697 }, { "epoch": 0.05, "learning_rate": 4.967219846298712e-05, "loss": 1.1927, "step": 51698 }, { "epoch": 0.05, "learning_rate": 4.9672185784726946e-05, "loss": 1.4379, "step": 51699 }, { "epoch": 0.05, "learning_rate": 4.967217310622323e-05, "loss": 1.1379, "step": 51700 }, { "epoch": 0.05, "learning_rate": 4.967216042747596e-05, "loss": 1.1024, "step": 51701 }, { "epoch": 0.05, "learning_rate": 4.967214774848512e-05, "loss": 1.2059, "step": 51702 }, { "epoch": 0.05, "learning_rate": 4.967213506925074e-05, "loss": 1.1096, "step": 51703 }, { "epoch": 0.05, "learning_rate": 4.9672122389772804e-05, "loss": 1.2756, "step": 51704 }, { "epoch": 0.05, "learning_rate": 4.967210971005132e-05, "loss": 1.3283, "step": 51705 }, { "epoch": 0.05, "learning_rate": 4.9672097030086285e-05, "loss": 0.8826, "step": 51706 }, { "epoch": 0.05, "learning_rate": 4.967208434987769e-05, "loss": 1.343, "step": 51707 }, { "epoch": 0.05, "learning_rate": 4.967207166942554e-05, "loss": 1.5738, "step": 51708 }, { "epoch": 0.05, "learning_rate": 4.967205898872985e-05, "loss": 0.7135, "step": 51709 }, { "epoch": 0.05, "learning_rate": 4.96720463077906e-05, "loss": 1.1618, "step": 51710 }, { "epoch": 0.05, "learning_rate": 4.967203362660779e-05, "loss": 1.3252, "step": 51711 }, { "epoch": 0.05, "learning_rate": 4.967202094518144e-05, "loss": 1.2174, "step": 51712 }, { "epoch": 0.05, "learning_rate": 4.967200826351154e-05, "loss": 1.0645, "step": 51713 }, { "epoch": 0.05, "learning_rate": 4.967199558159808e-05, "loss": 0.632, "step": 51714 }, { "epoch": 0.05, "learning_rate": 4.967198289944107e-05, "loss": 0.6852, "step": 51715 }, { "epoch": 0.05, "learning_rate": 4.967197021704051e-05, "loss": 0.4618, "step": 51716 }, { "epoch": 0.05, "learning_rate": 4.9671957534396404e-05, "loss": 1.156, "step": 51717 }, { "epoch": 0.05, "learning_rate": 4.967194485150874e-05, "loss": 1.2968, "step": 51718 }, { "epoch": 0.05, "learning_rate": 4.967193216837753e-05, "loss": 1.1674, "step": 51719 }, { "epoch": 0.05, "learning_rate": 4.967191948500276e-05, "loss": 1.2086, "step": 51720 }, { "epoch": 0.05, "learning_rate": 4.967190680138445e-05, "loss": 0.9068, "step": 51721 }, { "epoch": 0.05, "learning_rate": 4.967189411752258e-05, "loss": 0.9993, "step": 51722 }, { "epoch": 0.05, "learning_rate": 4.9671881433417167e-05, "loss": 0.7994, "step": 51723 }, { "epoch": 0.05, "learning_rate": 4.967186874906819e-05, "loss": 1.3346, "step": 51724 }, { "epoch": 0.05, "learning_rate": 4.967185606447568e-05, "loss": 1.4514, "step": 51725 }, { "epoch": 0.05, "learning_rate": 4.967184337963962e-05, "loss": 1.0737, "step": 51726 }, { "epoch": 0.05, "learning_rate": 4.967183069455999e-05, "loss": 1.2009, "step": 51727 }, { "epoch": 0.05, "learning_rate": 4.967181800923683e-05, "loss": 0.9337, "step": 51728 }, { "epoch": 0.05, "learning_rate": 4.9671805323670115e-05, "loss": 0.8403, "step": 51729 }, { "epoch": 0.05, "learning_rate": 4.967179263785985e-05, "loss": 1.1101, "step": 51730 }, { "epoch": 0.05, "learning_rate": 4.967177995180603e-05, "loss": 1.2796, "step": 51731 }, { "epoch": 0.05, "learning_rate": 4.967176726550866e-05, "loss": 1.2928, "step": 51732 }, { "epoch": 0.05, "learning_rate": 4.967175457896776e-05, "loss": 1.251, "step": 51733 }, { "epoch": 0.05, "learning_rate": 4.9671741892183286e-05, "loss": 1.0282, "step": 51734 }, { "epoch": 0.05, "learning_rate": 4.967172920515527e-05, "loss": 1.2694, "step": 51735 }, { "epoch": 0.05, "learning_rate": 4.9671716517883714e-05, "loss": 0.9206, "step": 51736 }, { "epoch": 0.05, "learning_rate": 4.96717038303686e-05, "loss": 1.2255, "step": 51737 }, { "epoch": 0.05, "learning_rate": 4.9671691142609946e-05, "loss": 0.9271, "step": 51738 }, { "epoch": 0.05, "learning_rate": 4.967167845460773e-05, "loss": 1.4877, "step": 51739 }, { "epoch": 0.05, "learning_rate": 4.967166576636198e-05, "loss": 0.9717, "step": 51740 }, { "epoch": 0.05, "learning_rate": 4.967165307787267e-05, "loss": 0.9876, "step": 51741 }, { "epoch": 0.05, "learning_rate": 4.9671640389139814e-05, "loss": 1.0732, "step": 51742 }, { "epoch": 0.05, "learning_rate": 4.967162770016342e-05, "loss": 1.0829, "step": 51743 }, { "epoch": 0.05, "learning_rate": 4.967161501094347e-05, "loss": 0.5115, "step": 51744 }, { "epoch": 0.05, "learning_rate": 4.967160232147998e-05, "loss": 1.0612, "step": 51745 }, { "epoch": 0.05, "learning_rate": 4.967158963177293e-05, "loss": 0.8973, "step": 51746 }, { "epoch": 0.05, "learning_rate": 4.967157694182234e-05, "loss": 1.2944, "step": 51747 }, { "epoch": 0.05, "learning_rate": 4.96715642516282e-05, "loss": 1.2105, "step": 51748 }, { "epoch": 0.05, "learning_rate": 4.967155156119052e-05, "loss": 0.8171, "step": 51749 }, { "epoch": 0.05, "learning_rate": 4.967153887050928e-05, "loss": 0.8203, "step": 51750 }, { "epoch": 0.05, "learning_rate": 4.96715261795845e-05, "loss": 1.6517, "step": 51751 }, { "epoch": 0.05, "learning_rate": 4.9671513488416175e-05, "loss": 1.3989, "step": 51752 }, { "epoch": 0.05, "learning_rate": 4.96715007970043e-05, "loss": 0.6272, "step": 51753 }, { "epoch": 0.05, "learning_rate": 4.967148810534888e-05, "loss": 0.4035, "step": 51754 }, { "epoch": 0.05, "learning_rate": 4.9671475413449906e-05, "loss": 0.7503, "step": 51755 }, { "epoch": 0.05, "learning_rate": 4.9671462721307396e-05, "loss": 1.201, "step": 51756 }, { "epoch": 0.05, "learning_rate": 4.967145002892134e-05, "loss": 1.1359, "step": 51757 }, { "epoch": 0.05, "learning_rate": 4.967143733629173e-05, "loss": 0.8516, "step": 51758 }, { "epoch": 0.05, "learning_rate": 4.967142464341858e-05, "loss": 0.8251, "step": 51759 }, { "epoch": 0.05, "learning_rate": 4.967141195030188e-05, "loss": 0.8116, "step": 51760 }, { "epoch": 0.05, "learning_rate": 4.9671399256941635e-05, "loss": 1.1688, "step": 51761 }, { "epoch": 0.05, "learning_rate": 4.9671386563337844e-05, "loss": 1.2055, "step": 51762 }, { "epoch": 0.05, "learning_rate": 4.9671373869490514e-05, "loss": 1.0374, "step": 51763 }, { "epoch": 0.05, "learning_rate": 4.967136117539963e-05, "loss": 1.0805, "step": 51764 }, { "epoch": 0.05, "learning_rate": 4.9671348481065205e-05, "loss": 1.1294, "step": 51765 }, { "epoch": 0.05, "learning_rate": 4.967133578648723e-05, "loss": 1.2038, "step": 51766 }, { "epoch": 0.05, "learning_rate": 4.967132309166572e-05, "loss": 0.6925, "step": 51767 }, { "epoch": 0.05, "learning_rate": 4.9671310396600654e-05, "loss": 0.9803, "step": 51768 }, { "epoch": 0.05, "learning_rate": 4.967129770129205e-05, "loss": 1.5786, "step": 51769 }, { "epoch": 0.05, "learning_rate": 4.96712850057399e-05, "loss": 0.9764, "step": 51770 }, { "epoch": 0.05, "learning_rate": 4.9671272309944206e-05, "loss": 1.1885, "step": 51771 }, { "epoch": 0.05, "learning_rate": 4.9671259613904965e-05, "loss": 1.1041, "step": 51772 }, { "epoch": 0.05, "learning_rate": 4.9671246917622185e-05, "loss": 1.0034, "step": 51773 }, { "epoch": 0.05, "learning_rate": 4.967123422109585e-05, "loss": 0.1412, "step": 51774 }, { "epoch": 0.05, "learning_rate": 4.967122152432598e-05, "loss": 0.1062, "step": 51775 }, { "epoch": 0.05, "learning_rate": 4.9671208827312565e-05, "loss": 0.088, "step": 51776 }, { "epoch": 0.05, "learning_rate": 4.967119613005561e-05, "loss": 0.0614, "step": 51777 }, { "epoch": 0.05, "learning_rate": 4.96711834325551e-05, "loss": 0.0594, "step": 51778 }, { "epoch": 0.05, "learning_rate": 4.9671170734811054e-05, "loss": 0.0411, "step": 51779 }, { "epoch": 0.05, "learning_rate": 4.967115803682347e-05, "loss": 0.0348, "step": 51780 }, { "epoch": 0.05, "learning_rate": 4.9671145338592336e-05, "loss": 0.249, "step": 51781 }, { "epoch": 0.05, "learning_rate": 4.967113264011766e-05, "loss": 1.4521, "step": 51782 }, { "epoch": 0.05, "learning_rate": 4.9671119941399436e-05, "loss": 0.8714, "step": 51783 }, { "epoch": 0.05, "learning_rate": 4.9671107242437674e-05, "loss": 0.5126, "step": 51784 }, { "epoch": 0.05, "learning_rate": 4.967109454323237e-05, "loss": 0.4819, "step": 51785 }, { "epoch": 0.05, "learning_rate": 4.9671081843783526e-05, "loss": 0.4114, "step": 51786 }, { "epoch": 0.05, "learning_rate": 4.9671069144091134e-05, "loss": 0.3905, "step": 51787 }, { "epoch": 0.05, "learning_rate": 4.96710564441552e-05, "loss": 0.4036, "step": 51788 }, { "epoch": 0.05, "learning_rate": 4.9671043743975726e-05, "loss": 0.4251, "step": 51789 }, { "epoch": 0.05, "learning_rate": 4.967103104355271e-05, "loss": 1.101, "step": 51790 }, { "epoch": 0.05, "learning_rate": 4.967101834288615e-05, "loss": 1.4059, "step": 51791 }, { "epoch": 0.05, "learning_rate": 4.967100564197605e-05, "loss": 0.9449, "step": 51792 }, { "epoch": 0.05, "learning_rate": 4.96709929408224e-05, "loss": 0.8725, "step": 51793 }, { "epoch": 0.05, "learning_rate": 4.9670980239425224e-05, "loss": 0.999, "step": 51794 }, { "epoch": 0.05, "learning_rate": 4.9670967537784494e-05, "loss": 0.947, "step": 51795 }, { "epoch": 0.05, "learning_rate": 4.967095483590023e-05, "loss": 1.1362, "step": 51796 }, { "epoch": 0.05, "learning_rate": 4.967094213377242e-05, "loss": 1.2675, "step": 51797 }, { "epoch": 0.05, "learning_rate": 4.967092943140107e-05, "loss": 1.1512, "step": 51798 }, { "epoch": 0.05, "learning_rate": 4.9670916728786175e-05, "loss": 1.8528, "step": 51799 }, { "epoch": 0.05, "learning_rate": 4.967090402592774e-05, "loss": 1.5481, "step": 51800 }, { "epoch": 0.05, "learning_rate": 4.967089132282577e-05, "loss": 1.7985, "step": 51801 }, { "epoch": 0.05, "learning_rate": 4.967087861948026e-05, "loss": 0.8957, "step": 51802 }, { "epoch": 0.05, "learning_rate": 4.9670865915891207e-05, "loss": 1.4816, "step": 51803 }, { "epoch": 0.05, "learning_rate": 4.967085321205861e-05, "loss": 1.58, "step": 51804 }, { "epoch": 0.05, "learning_rate": 4.967084050798248e-05, "loss": 1.7592, "step": 51805 }, { "epoch": 0.05, "learning_rate": 4.96708278036628e-05, "loss": 1.1631, "step": 51806 }, { "epoch": 0.05, "learning_rate": 4.967081509909959e-05, "loss": 1.4126, "step": 51807 }, { "epoch": 0.05, "learning_rate": 4.967080239429284e-05, "loss": 1.7235, "step": 51808 }, { "epoch": 0.05, "learning_rate": 4.967078968924254e-05, "loss": 1.7359, "step": 51809 }, { "epoch": 0.05, "learning_rate": 4.967077698394871e-05, "loss": 1.2499, "step": 51810 }, { "epoch": 0.05, "learning_rate": 4.967076427841133e-05, "loss": 1.0487, "step": 51811 }, { "epoch": 0.05, "learning_rate": 4.967075157263042e-05, "loss": 1.0361, "step": 51812 }, { "epoch": 0.05, "learning_rate": 4.967073886660597e-05, "loss": 1.0177, "step": 51813 }, { "epoch": 0.05, "learning_rate": 4.967072616033798e-05, "loss": 1.212, "step": 51814 }, { "epoch": 0.05, "learning_rate": 4.9670713453826444e-05, "loss": 1.2086, "step": 51815 }, { "epoch": 0.05, "learning_rate": 4.967070074707138e-05, "loss": 1.1663, "step": 51816 }, { "epoch": 0.05, "learning_rate": 4.967068804007277e-05, "loss": 1.1732, "step": 51817 }, { "epoch": 0.05, "learning_rate": 4.967067533283062e-05, "loss": 1.3008, "step": 51818 }, { "epoch": 0.05, "learning_rate": 4.9670662625344945e-05, "loss": 0.8426, "step": 51819 }, { "epoch": 0.05, "learning_rate": 4.967064991761572e-05, "loss": 0.868, "step": 51820 }, { "epoch": 0.05, "learning_rate": 4.967063720964296e-05, "loss": 0.9535, "step": 51821 }, { "epoch": 0.05, "learning_rate": 4.9670624501426656e-05, "loss": 1.2437, "step": 51822 }, { "epoch": 0.05, "learning_rate": 4.9670611792966816e-05, "loss": 1.0657, "step": 51823 }, { "epoch": 0.05, "learning_rate": 4.967059908426344e-05, "loss": 1.0802, "step": 51824 }, { "epoch": 0.05, "learning_rate": 4.967058637531653e-05, "loss": 1.1057, "step": 51825 }, { "epoch": 0.05, "learning_rate": 4.9670573666126074e-05, "loss": 0.9519, "step": 51826 }, { "epoch": 0.05, "learning_rate": 4.967056095669209e-05, "loss": 1.3203, "step": 51827 }, { "epoch": 0.05, "learning_rate": 4.967054824701456e-05, "loss": 1.2279, "step": 51828 }, { "epoch": 0.05, "learning_rate": 4.9670535537093496e-05, "loss": 1.0398, "step": 51829 }, { "epoch": 0.05, "learning_rate": 4.96705228269289e-05, "loss": 1.0566, "step": 51830 }, { "epoch": 0.05, "learning_rate": 4.9670510116520765e-05, "loss": 1.2058, "step": 51831 }, { "epoch": 0.05, "learning_rate": 4.967049740586909e-05, "loss": 1.1347, "step": 51832 }, { "epoch": 0.05, "learning_rate": 4.967048469497388e-05, "loss": 0.7475, "step": 51833 }, { "epoch": 0.05, "learning_rate": 4.967047198383513e-05, "loss": 0.768, "step": 51834 }, { "epoch": 0.05, "learning_rate": 4.967045927245284e-05, "loss": 1.2189, "step": 51835 }, { "epoch": 0.05, "learning_rate": 4.967044656082703e-05, "loss": 1.2062, "step": 51836 }, { "epoch": 0.05, "learning_rate": 4.967043384895767e-05, "loss": 1.0593, "step": 51837 }, { "epoch": 0.05, "learning_rate": 4.967042113684478e-05, "loss": 0.9167, "step": 51838 }, { "epoch": 0.05, "learning_rate": 4.9670408424488346e-05, "loss": 1.0683, "step": 51839 }, { "epoch": 0.05, "learning_rate": 4.9670395711888384e-05, "loss": 1.1151, "step": 51840 }, { "epoch": 0.05, "learning_rate": 4.9670382999044884e-05, "loss": 1.1355, "step": 51841 }, { "epoch": 0.05, "learning_rate": 4.9670370285957845e-05, "loss": 1.3118, "step": 51842 }, { "epoch": 0.05, "learning_rate": 4.967035757262728e-05, "loss": 1.2826, "step": 51843 }, { "epoch": 0.05, "learning_rate": 4.967034485905317e-05, "loss": 2.4691, "step": 51844 }, { "epoch": 0.05, "learning_rate": 4.967033214523553e-05, "loss": 1.1221, "step": 51845 }, { "epoch": 0.05, "learning_rate": 4.967031943117435e-05, "loss": 0.9834, "step": 51846 }, { "epoch": 0.05, "learning_rate": 4.967030671686964e-05, "loss": 1.3222, "step": 51847 }, { "epoch": 0.05, "learning_rate": 4.9670294002321393e-05, "loss": 1.0021, "step": 51848 }, { "epoch": 0.05, "learning_rate": 4.967028128752961e-05, "loss": 1.0903, "step": 51849 }, { "epoch": 0.05, "learning_rate": 4.967026857249429e-05, "loss": 0.9848, "step": 51850 }, { "epoch": 0.05, "learning_rate": 4.9670255857215443e-05, "loss": 0.8265, "step": 51851 }, { "epoch": 0.05, "learning_rate": 4.967024314169305e-05, "loss": 0.8051, "step": 51852 }, { "epoch": 0.05, "learning_rate": 4.9670230425927136e-05, "loss": 1.2334, "step": 51853 }, { "epoch": 0.05, "learning_rate": 4.9670217709917684e-05, "loss": 1.0843, "step": 51854 }, { "epoch": 0.05, "learning_rate": 4.967020499366469e-05, "loss": 1.1225, "step": 51855 }, { "epoch": 0.05, "learning_rate": 4.967019227716817e-05, "loss": 1.2046, "step": 51856 }, { "epoch": 0.05, "learning_rate": 4.967017956042812e-05, "loss": 0.9454, "step": 51857 }, { "epoch": 0.05, "learning_rate": 4.9670166843444534e-05, "loss": 0.9595, "step": 51858 }, { "epoch": 0.05, "learning_rate": 4.967015412621741e-05, "loss": 0.9908, "step": 51859 }, { "epoch": 0.05, "learning_rate": 4.967014140874675e-05, "loss": 1.2015, "step": 51860 }, { "epoch": 0.05, "learning_rate": 4.967012869103256e-05, "loss": 1.237, "step": 51861 }, { "epoch": 0.05, "learning_rate": 4.9670115973074836e-05, "loss": 1.3267, "step": 51862 }, { "epoch": 0.05, "learning_rate": 4.967010325487358e-05, "loss": 1.187, "step": 51863 }, { "epoch": 0.05, "learning_rate": 4.9670090536428795e-05, "loss": 1.3473, "step": 51864 }, { "epoch": 0.05, "learning_rate": 4.9670077817740476e-05, "loss": 1.2922, "step": 51865 }, { "epoch": 0.05, "learning_rate": 4.9670065098808624e-05, "loss": 0.8503, "step": 51866 }, { "epoch": 0.05, "learning_rate": 4.9670052379633234e-05, "loss": 1.148, "step": 51867 }, { "epoch": 0.05, "learning_rate": 4.967003966021432e-05, "loss": 0.8889, "step": 51868 }, { "epoch": 0.05, "learning_rate": 4.9670026940551864e-05, "loss": 1.1762, "step": 51869 }, { "epoch": 0.05, "learning_rate": 4.9670014220645885e-05, "loss": 1.1403, "step": 51870 }, { "epoch": 0.05, "learning_rate": 4.9670001500496366e-05, "loss": 0.9131, "step": 51871 }, { "epoch": 0.05, "learning_rate": 4.966998878010333e-05, "loss": 0.9412, "step": 51872 }, { "epoch": 0.05, "learning_rate": 4.9669976059466746e-05, "loss": 1.2147, "step": 51873 }, { "epoch": 0.05, "learning_rate": 4.966996333858663e-05, "loss": 1.1071, "step": 51874 }, { "epoch": 0.05, "learning_rate": 4.966995061746299e-05, "loss": 1.2678, "step": 51875 }, { "epoch": 0.05, "learning_rate": 4.966993789609582e-05, "loss": 0.9956, "step": 51876 }, { "epoch": 0.05, "learning_rate": 4.966992517448512e-05, "loss": 0.7154, "step": 51877 }, { "epoch": 0.05, "learning_rate": 4.966991245263088e-05, "loss": 0.7078, "step": 51878 }, { "epoch": 0.05, "learning_rate": 4.966989973053312e-05, "loss": 0.6353, "step": 51879 }, { "epoch": 0.05, "learning_rate": 4.966988700819182e-05, "loss": 0.564, "step": 51880 }, { "epoch": 0.05, "learning_rate": 4.966987428560699e-05, "loss": 0.9677, "step": 51881 }, { "epoch": 0.05, "learning_rate": 4.9669861562778635e-05, "loss": 1.1695, "step": 51882 }, { "epoch": 0.05, "learning_rate": 4.9669848839706744e-05, "loss": 0.9114, "step": 51883 }, { "epoch": 0.05, "learning_rate": 4.966983611639133e-05, "loss": 1.6355, "step": 51884 }, { "epoch": 0.05, "learning_rate": 4.966982339283238e-05, "loss": 1.5078, "step": 51885 }, { "epoch": 0.05, "learning_rate": 4.96698106690299e-05, "loss": 1.1385, "step": 51886 }, { "epoch": 0.05, "learning_rate": 4.9669797944983887e-05, "loss": 1.5239, "step": 51887 }, { "epoch": 0.05, "learning_rate": 4.9669785220694355e-05, "loss": 1.0863, "step": 51888 }, { "epoch": 0.05, "learning_rate": 4.966977249616128e-05, "loss": 0.9125, "step": 51889 }, { "epoch": 0.05, "learning_rate": 4.966975977138468e-05, "loss": 1.1254, "step": 51890 }, { "epoch": 0.05, "learning_rate": 4.9669747046364556e-05, "loss": 1.59, "step": 51891 }, { "epoch": 0.05, "learning_rate": 4.9669734321100896e-05, "loss": 1.0763, "step": 51892 }, { "epoch": 0.05, "learning_rate": 4.966972159559371e-05, "loss": 1.2317, "step": 51893 }, { "epoch": 0.05, "learning_rate": 4.9669708869842994e-05, "loss": 1.0896, "step": 51894 }, { "epoch": 0.05, "learning_rate": 4.966969614384875e-05, "loss": 1.1222, "step": 51895 }, { "epoch": 0.05, "learning_rate": 4.966968341761098e-05, "loss": 1.1927, "step": 51896 }, { "epoch": 0.05, "learning_rate": 4.966967069112968e-05, "loss": 1.1071, "step": 51897 }, { "epoch": 0.05, "learning_rate": 4.9669657964404845e-05, "loss": 1.1032, "step": 51898 }, { "epoch": 0.05, "learning_rate": 4.966964523743649e-05, "loss": 1.543, "step": 51899 }, { "epoch": 0.05, "learning_rate": 4.96696325102246e-05, "loss": 1.406, "step": 51900 }, { "epoch": 0.05, "learning_rate": 4.966961978276918e-05, "loss": 1.2214, "step": 51901 }, { "epoch": 0.05, "learning_rate": 4.9669607055070235e-05, "loss": 1.4692, "step": 51902 }, { "epoch": 0.05, "learning_rate": 4.966959432712777e-05, "loss": 1.3458, "step": 51903 }, { "epoch": 0.05, "learning_rate": 4.9669581598941765e-05, "loss": 1.3144, "step": 51904 }, { "epoch": 0.05, "learning_rate": 4.966956887051224e-05, "loss": 1.2077, "step": 51905 }, { "epoch": 0.05, "learning_rate": 4.9669556141839187e-05, "loss": 0.6827, "step": 51906 }, { "epoch": 0.05, "learning_rate": 4.9669543412922606e-05, "loss": 0.8718, "step": 51907 }, { "epoch": 0.05, "learning_rate": 4.966953068376249e-05, "loss": 1.3898, "step": 51908 }, { "epoch": 0.05, "learning_rate": 4.9669517954358855e-05, "loss": 1.2356, "step": 51909 }, { "epoch": 0.05, "learning_rate": 4.966950522471169e-05, "loss": 1.3237, "step": 51910 }, { "epoch": 0.05, "learning_rate": 4.9669492494821e-05, "loss": 1.283, "step": 51911 }, { "epoch": 0.05, "learning_rate": 4.966947976468678e-05, "loss": 1.0422, "step": 51912 }, { "epoch": 0.05, "learning_rate": 4.9669467034309035e-05, "loss": 1.471, "step": 51913 }, { "epoch": 0.05, "learning_rate": 4.966945430368777e-05, "loss": 1.3171, "step": 51914 }, { "epoch": 0.05, "learning_rate": 4.966944157282297e-05, "loss": 1.0855, "step": 51915 }, { "epoch": 0.05, "learning_rate": 4.966942884171465e-05, "loss": 1.355, "step": 51916 }, { "epoch": 0.05, "learning_rate": 4.9669416110362795e-05, "loss": 1.1228, "step": 51917 }, { "epoch": 0.05, "learning_rate": 4.966940337876742e-05, "loss": 1.0814, "step": 51918 }, { "epoch": 0.05, "learning_rate": 4.966939064692852e-05, "loss": 1.2315, "step": 51919 }, { "epoch": 0.05, "learning_rate": 4.966937791484609e-05, "loss": 1.2801, "step": 51920 }, { "epoch": 0.05, "learning_rate": 4.966936518252013e-05, "loss": 1.0793, "step": 51921 }, { "epoch": 0.05, "learning_rate": 4.9669352449950655e-05, "loss": 1.1382, "step": 51922 }, { "epoch": 0.05, "learning_rate": 4.9669339717137655e-05, "loss": 0.8711, "step": 51923 }, { "epoch": 0.05, "learning_rate": 4.966932698408112e-05, "loss": 0.9714, "step": 51924 }, { "epoch": 0.05, "learning_rate": 4.9669314250781066e-05, "loss": 1.0865, "step": 51925 }, { "epoch": 0.05, "learning_rate": 4.966930151723749e-05, "loss": 1.1672, "step": 51926 }, { "epoch": 0.05, "learning_rate": 4.966928878345039e-05, "loss": 1.061, "step": 51927 }, { "epoch": 0.05, "learning_rate": 4.9669276049419754e-05, "loss": 1.102, "step": 51928 }, { "epoch": 0.05, "learning_rate": 4.9669263315145595e-05, "loss": 1.1924, "step": 51929 }, { "epoch": 0.05, "learning_rate": 4.966925058062792e-05, "loss": 1.142, "step": 51930 }, { "epoch": 0.05, "learning_rate": 4.9669237845866715e-05, "loss": 0.9989, "step": 51931 }, { "epoch": 0.05, "learning_rate": 4.966922511086198e-05, "loss": 0.7338, "step": 51932 }, { "epoch": 0.05, "learning_rate": 4.966921237561373e-05, "loss": 0.5105, "step": 51933 }, { "epoch": 0.05, "learning_rate": 4.9669199640121954e-05, "loss": 1.2873, "step": 51934 }, { "epoch": 0.05, "learning_rate": 4.9669186904386656e-05, "loss": 0.9242, "step": 51935 }, { "epoch": 0.05, "learning_rate": 4.9669174168407834e-05, "loss": 0.6223, "step": 51936 }, { "epoch": 0.05, "learning_rate": 4.9669161432185485e-05, "loss": 0.7926, "step": 51937 }, { "epoch": 0.05, "learning_rate": 4.966914869571961e-05, "loss": 0.6962, "step": 51938 }, { "epoch": 0.05, "learning_rate": 4.966913595901021e-05, "loss": 0.8485, "step": 51939 }, { "epoch": 0.05, "learning_rate": 4.96691232220573e-05, "loss": 1.4391, "step": 51940 }, { "epoch": 0.05, "learning_rate": 4.966911048486086e-05, "loss": 1.1088, "step": 51941 }, { "epoch": 0.05, "learning_rate": 4.96690977474209e-05, "loss": 1.2757, "step": 51942 }, { "epoch": 0.05, "learning_rate": 4.9669085009737404e-05, "loss": 1.5748, "step": 51943 }, { "epoch": 0.05, "learning_rate": 4.96690722718104e-05, "loss": 1.2779, "step": 51944 }, { "epoch": 0.05, "learning_rate": 4.966905953363986e-05, "loss": 1.3259, "step": 51945 }, { "epoch": 0.05, "learning_rate": 4.966904679522581e-05, "loss": 0.9898, "step": 51946 }, { "epoch": 0.05, "learning_rate": 4.966903405656823e-05, "loss": 1.1763, "step": 51947 }, { "epoch": 0.05, "learning_rate": 4.966902131766713e-05, "loss": 0.9795, "step": 51948 }, { "epoch": 0.05, "learning_rate": 4.966900857852251e-05, "loss": 1.1802, "step": 51949 }, { "epoch": 0.05, "learning_rate": 4.9668995839134366e-05, "loss": 0.9445, "step": 51950 }, { "epoch": 0.05, "learning_rate": 4.96689830995027e-05, "loss": 1.2073, "step": 51951 }, { "epoch": 0.05, "learning_rate": 4.966897035962751e-05, "loss": 1.0002, "step": 51952 }, { "epoch": 0.05, "learning_rate": 4.96689576195088e-05, "loss": 1.1224, "step": 51953 }, { "epoch": 0.05, "learning_rate": 4.966894487914657e-05, "loss": 1.2232, "step": 51954 }, { "epoch": 0.05, "learning_rate": 4.966893213854081e-05, "loss": 1.0091, "step": 51955 }, { "epoch": 0.05, "learning_rate": 4.9668919397691544e-05, "loss": 0.8458, "step": 51956 }, { "epoch": 0.05, "learning_rate": 4.966890665659875e-05, "loss": 0.9435, "step": 51957 }, { "epoch": 0.05, "learning_rate": 4.966889391526243e-05, "loss": 0.7872, "step": 51958 }, { "epoch": 0.05, "learning_rate": 4.966888117368259e-05, "loss": 1.0746, "step": 51959 }, { "epoch": 0.05, "learning_rate": 4.966886843185924e-05, "loss": 1.1199, "step": 51960 }, { "epoch": 0.05, "learning_rate": 4.966885568979236e-05, "loss": 0.5517, "step": 51961 }, { "epoch": 0.05, "learning_rate": 4.966884294748196e-05, "loss": 1.1014, "step": 51962 }, { "epoch": 0.05, "learning_rate": 4.966883020492804e-05, "loss": 1.1456, "step": 51963 }, { "epoch": 0.05, "learning_rate": 4.96688174621306e-05, "loss": 1.0376, "step": 51964 }, { "epoch": 0.05, "learning_rate": 4.9668804719089636e-05, "loss": 1.5457, "step": 51965 }, { "epoch": 0.05, "learning_rate": 4.9668791975805154e-05, "loss": 0.9193, "step": 51966 }, { "epoch": 0.05, "learning_rate": 4.966877923227716e-05, "loss": 0.9254, "step": 51967 }, { "epoch": 0.05, "learning_rate": 4.9668766488505634e-05, "loss": 1.3298, "step": 51968 }, { "epoch": 0.05, "learning_rate": 4.9668753744490596e-05, "loss": 1.4031, "step": 51969 }, { "epoch": 0.05, "learning_rate": 4.966874100023204e-05, "loss": 0.9426, "step": 51970 }, { "epoch": 0.05, "learning_rate": 4.966872825572996e-05, "loss": 1.1783, "step": 51971 }, { "epoch": 0.05, "learning_rate": 4.9668715510984364e-05, "loss": 1.0086, "step": 51972 }, { "epoch": 0.05, "learning_rate": 4.9668702765995245e-05, "loss": 0.9708, "step": 51973 }, { "epoch": 0.05, "learning_rate": 4.966869002076261e-05, "loss": 1.3605, "step": 51974 }, { "epoch": 0.05, "learning_rate": 4.966867727528646e-05, "loss": 0.8443, "step": 51975 }, { "epoch": 0.05, "learning_rate": 4.9668664529566777e-05, "loss": 1.3974, "step": 51976 }, { "epoch": 0.05, "learning_rate": 4.966865178360359e-05, "loss": 1.4757, "step": 51977 }, { "epoch": 0.05, "learning_rate": 4.966863903739688e-05, "loss": 1.3314, "step": 51978 }, { "epoch": 0.05, "learning_rate": 4.9668626290946654e-05, "loss": 1.2591, "step": 51979 }, { "epoch": 0.05, "learning_rate": 4.96686135442529e-05, "loss": 1.1451, "step": 51980 }, { "epoch": 0.05, "learning_rate": 4.966860079731563e-05, "loss": 1.214, "step": 51981 }, { "epoch": 0.05, "learning_rate": 4.966858805013485e-05, "loss": 1.5197, "step": 51982 }, { "epoch": 0.05, "learning_rate": 4.9668575302710553e-05, "loss": 1.2746, "step": 51983 }, { "epoch": 0.05, "learning_rate": 4.966856255504273e-05, "loss": 1.3522, "step": 51984 }, { "epoch": 0.05, "learning_rate": 4.9668549807131395e-05, "loss": 1.2913, "step": 51985 }, { "epoch": 0.05, "learning_rate": 4.966853705897654e-05, "loss": 0.9383, "step": 51986 }, { "epoch": 0.05, "learning_rate": 4.966852431057817e-05, "loss": 1.0891, "step": 51987 }, { "epoch": 0.05, "learning_rate": 4.966851156193628e-05, "loss": 1.3981, "step": 51988 }, { "epoch": 0.05, "learning_rate": 4.966849881305087e-05, "loss": 0.9833, "step": 51989 }, { "epoch": 0.05, "learning_rate": 4.966848606392195e-05, "loss": 1.1908, "step": 51990 }, { "epoch": 0.05, "learning_rate": 4.966847331454951e-05, "loss": 1.0485, "step": 51991 }, { "epoch": 0.05, "learning_rate": 4.966846056493355e-05, "loss": 1.1396, "step": 51992 }, { "epoch": 0.05, "learning_rate": 4.966844781507408e-05, "loss": 1.1453, "step": 51993 }, { "epoch": 0.05, "learning_rate": 4.966843506497109e-05, "loss": 1.1722, "step": 51994 }, { "epoch": 0.05, "learning_rate": 4.9668422314624575e-05, "loss": 0.8722, "step": 51995 }, { "epoch": 0.05, "learning_rate": 4.9668409564034555e-05, "loss": 0.5043, "step": 51996 }, { "epoch": 0.05, "learning_rate": 4.966839681320101e-05, "loss": 1.159, "step": 51997 }, { "epoch": 0.05, "learning_rate": 4.966838406212396e-05, "loss": 1.2168, "step": 51998 }, { "epoch": 0.05, "learning_rate": 4.9668371310803386e-05, "loss": 0.9558, "step": 51999 }, { "epoch": 0.05, "learning_rate": 4.9668358559239306e-05, "loss": 1.5304, "step": 52000 }, { "epoch": 0.05, "eval_loss": 1.0703086853027344, "eval_runtime": 125.4565, "eval_samples_per_second": 11.04, "eval_steps_per_second": 5.524, "step": 52000 }, { "epoch": 0.05, "learning_rate": 4.96683458074317e-05, "loss": 1.418, "step": 52001 }, { "epoch": 0.05, "learning_rate": 4.9668333055380576e-05, "loss": 0.7807, "step": 52002 }, { "epoch": 0.05, "learning_rate": 4.966832030308595e-05, "loss": 1.117, "step": 52003 }, { "epoch": 0.05, "learning_rate": 4.96683075505478e-05, "loss": 1.2714, "step": 52004 }, { "epoch": 0.05, "learning_rate": 4.9668294797766133e-05, "loss": 0.7885, "step": 52005 }, { "epoch": 0.05, "learning_rate": 4.9668282044740955e-05, "loss": 1.1122, "step": 52006 }, { "epoch": 0.05, "learning_rate": 4.9668269291472266e-05, "loss": 1.4083, "step": 52007 }, { "epoch": 0.05, "learning_rate": 4.966825653796006e-05, "loss": 1.1875, "step": 52008 }, { "epoch": 0.05, "learning_rate": 4.966824378420433e-05, "loss": 1.2533, "step": 52009 }, { "epoch": 0.05, "learning_rate": 4.966823103020509e-05, "loss": 1.2226, "step": 52010 }, { "epoch": 0.05, "learning_rate": 4.966821827596234e-05, "loss": 0.9889, "step": 52011 }, { "epoch": 0.05, "learning_rate": 4.966820552147607e-05, "loss": 1.1577, "step": 52012 }, { "epoch": 0.05, "learning_rate": 4.96681927667463e-05, "loss": 0.7887, "step": 52013 }, { "epoch": 0.05, "learning_rate": 4.9668180011773004e-05, "loss": 1.145, "step": 52014 }, { "epoch": 0.05, "learning_rate": 4.966816725655619e-05, "loss": 1.0927, "step": 52015 }, { "epoch": 0.05, "learning_rate": 4.966815450109587e-05, "loss": 0.9112, "step": 52016 }, { "epoch": 0.05, "learning_rate": 4.966814174539204e-05, "loss": 1.4039, "step": 52017 }, { "epoch": 0.05, "learning_rate": 4.966812898944469e-05, "loss": 1.4523, "step": 52018 }, { "epoch": 0.05, "learning_rate": 4.966811623325382e-05, "loss": 0.9854, "step": 52019 }, { "epoch": 0.05, "learning_rate": 4.966810347681945e-05, "loss": 1.2305, "step": 52020 }, { "epoch": 0.05, "learning_rate": 4.9668090720141556e-05, "loss": 1.5931, "step": 52021 }, { "epoch": 0.05, "learning_rate": 4.966807796322016e-05, "loss": 1.2519, "step": 52022 }, { "epoch": 0.05, "learning_rate": 4.966806520605524e-05, "loss": 0.9767, "step": 52023 }, { "epoch": 0.05, "learning_rate": 4.966805244864682e-05, "loss": 1.1565, "step": 52024 }, { "epoch": 0.05, "learning_rate": 4.9668039690994875e-05, "loss": 1.7571, "step": 52025 }, { "epoch": 0.05, "learning_rate": 4.966802693309942e-05, "loss": 1.4884, "step": 52026 }, { "epoch": 0.05, "learning_rate": 4.9668014174960453e-05, "loss": 1.3462, "step": 52027 }, { "epoch": 0.05, "learning_rate": 4.966800141657798e-05, "loss": 0.9514, "step": 52028 }, { "epoch": 0.05, "learning_rate": 4.966798865795199e-05, "loss": 1.4721, "step": 52029 }, { "epoch": 0.05, "learning_rate": 4.966797589908249e-05, "loss": 1.1214, "step": 52030 }, { "epoch": 0.05, "learning_rate": 4.9667963139969475e-05, "loss": 1.1559, "step": 52031 }, { "epoch": 0.05, "learning_rate": 4.966795038061295e-05, "loss": 1.1037, "step": 52032 }, { "epoch": 0.05, "learning_rate": 4.966793762101292e-05, "loss": 0.8786, "step": 52033 }, { "epoch": 0.05, "learning_rate": 4.966792486116937e-05, "loss": 0.9768, "step": 52034 }, { "epoch": 0.05, "learning_rate": 4.96679121010823e-05, "loss": 1.0437, "step": 52035 }, { "epoch": 0.05, "learning_rate": 4.9667899340751736e-05, "loss": 1.3013, "step": 52036 }, { "epoch": 0.05, "learning_rate": 4.966788658017766e-05, "loss": 1.093, "step": 52037 }, { "epoch": 0.05, "learning_rate": 4.966787381936006e-05, "loss": 1.1951, "step": 52038 }, { "epoch": 0.05, "learning_rate": 4.966786105829896e-05, "loss": 1.206, "step": 52039 }, { "epoch": 0.05, "learning_rate": 4.9667848296994346e-05, "loss": 1.1404, "step": 52040 }, { "epoch": 0.05, "learning_rate": 4.9667835535446214e-05, "loss": 1.404, "step": 52041 }, { "epoch": 0.05, "learning_rate": 4.9667822773654584e-05, "loss": 1.2113, "step": 52042 }, { "epoch": 0.05, "learning_rate": 4.966781001161944e-05, "loss": 1.4991, "step": 52043 }, { "epoch": 0.05, "learning_rate": 4.966779724934079e-05, "loss": 0.6577, "step": 52044 }, { "epoch": 0.05, "learning_rate": 4.9667784486818616e-05, "loss": 0.3646, "step": 52045 }, { "epoch": 0.05, "learning_rate": 4.9667771724052945e-05, "loss": 0.5426, "step": 52046 }, { "epoch": 0.05, "learning_rate": 4.9667758961043756e-05, "loss": 1.0744, "step": 52047 }, { "epoch": 0.05, "learning_rate": 4.966774619779106e-05, "loss": 1.0151, "step": 52048 }, { "epoch": 0.05, "learning_rate": 4.9667733434294855e-05, "loss": 1.1289, "step": 52049 }, { "epoch": 0.05, "learning_rate": 4.966772067055514e-05, "loss": 1.1372, "step": 52050 }, { "epoch": 0.05, "learning_rate": 4.9667707906571914e-05, "loss": 1.061, "step": 52051 }, { "epoch": 0.05, "learning_rate": 4.9667695142345186e-05, "loss": 1.3647, "step": 52052 }, { "epoch": 0.05, "learning_rate": 4.966768237787493e-05, "loss": 0.911, "step": 52053 }, { "epoch": 0.05, "learning_rate": 4.966766961316119e-05, "loss": 1.1869, "step": 52054 }, { "epoch": 0.05, "learning_rate": 4.9667656848203925e-05, "loss": 1.2912, "step": 52055 }, { "epoch": 0.05, "learning_rate": 4.966764408300316e-05, "loss": 1.1705, "step": 52056 }, { "epoch": 0.05, "learning_rate": 4.966763131755888e-05, "loss": 1.31, "step": 52057 }, { "epoch": 0.05, "learning_rate": 4.96676185518711e-05, "loss": 1.1062, "step": 52058 }, { "epoch": 0.05, "learning_rate": 4.96676057859398e-05, "loss": 0.9691, "step": 52059 }, { "epoch": 0.05, "learning_rate": 4.9667593019765005e-05, "loss": 1.5011, "step": 52060 }, { "epoch": 0.05, "learning_rate": 4.966758025334669e-05, "loss": 1.8592, "step": 52061 }, { "epoch": 0.05, "learning_rate": 4.966756748668487e-05, "loss": 1.8622, "step": 52062 }, { "epoch": 0.05, "learning_rate": 4.966755471977954e-05, "loss": 0.9812, "step": 52063 }, { "epoch": 0.05, "learning_rate": 4.966754195263071e-05, "loss": 0.6723, "step": 52064 }, { "epoch": 0.05, "learning_rate": 4.966752918523837e-05, "loss": 0.7784, "step": 52065 }, { "epoch": 0.05, "learning_rate": 4.966751641760252e-05, "loss": 1.0331, "step": 52066 }, { "epoch": 0.05, "learning_rate": 4.966750364972316e-05, "loss": 0.9092, "step": 52067 }, { "epoch": 0.05, "learning_rate": 4.96674908816003e-05, "loss": 0.8889, "step": 52068 }, { "epoch": 0.05, "learning_rate": 4.966747811323393e-05, "loss": 1.3366, "step": 52069 }, { "epoch": 0.05, "learning_rate": 4.966746534462405e-05, "loss": 1.1909, "step": 52070 }, { "epoch": 0.05, "learning_rate": 4.966745257577067e-05, "loss": 1.2708, "step": 52071 }, { "epoch": 0.05, "learning_rate": 4.966743980667378e-05, "loss": 1.0446, "step": 52072 }, { "epoch": 0.05, "learning_rate": 4.966742703733338e-05, "loss": 1.3055, "step": 52073 }, { "epoch": 0.05, "learning_rate": 4.9667414267749476e-05, "loss": 0.9848, "step": 52074 }, { "epoch": 0.05, "learning_rate": 4.9667401497922064e-05, "loss": 1.0932, "step": 52075 }, { "epoch": 0.05, "learning_rate": 4.966738872785115e-05, "loss": 3.5206, "step": 52076 }, { "epoch": 0.05, "learning_rate": 4.966737595753672e-05, "loss": 0.9491, "step": 52077 }, { "epoch": 0.05, "learning_rate": 4.96673631869788e-05, "loss": 0.4289, "step": 52078 }, { "epoch": 0.05, "learning_rate": 4.966735041617736e-05, "loss": 1.0219, "step": 52079 }, { "epoch": 0.05, "learning_rate": 4.9667337645132425e-05, "loss": 0.7158, "step": 52080 }, { "epoch": 0.05, "learning_rate": 4.9667324873843976e-05, "loss": 0.9519, "step": 52081 }, { "epoch": 0.05, "learning_rate": 4.966731210231202e-05, "loss": 1.0215, "step": 52082 }, { "epoch": 0.05, "learning_rate": 4.9667299330536564e-05, "loss": 1.0709, "step": 52083 }, { "epoch": 0.05, "learning_rate": 4.966728655851761e-05, "loss": 0.8784, "step": 52084 }, { "epoch": 0.05, "learning_rate": 4.966727378625514e-05, "loss": 0.8212, "step": 52085 }, { "epoch": 0.05, "learning_rate": 4.9667261013749165e-05, "loss": 0.9951, "step": 52086 }, { "epoch": 0.05, "learning_rate": 4.9667248240999694e-05, "loss": 1.2297, "step": 52087 }, { "epoch": 0.05, "learning_rate": 4.966723546800671e-05, "loss": 1.1806, "step": 52088 }, { "epoch": 0.05, "learning_rate": 4.966722269477022e-05, "loss": 1.1241, "step": 52089 }, { "epoch": 0.05, "learning_rate": 4.9667209921290234e-05, "loss": 0.8857, "step": 52090 }, { "epoch": 0.05, "learning_rate": 4.966719714756674e-05, "loss": 0.8648, "step": 52091 }, { "epoch": 0.05, "learning_rate": 4.966718437359974e-05, "loss": 0.8054, "step": 52092 }, { "epoch": 0.05, "learning_rate": 4.9667171599389236e-05, "loss": 1.3571, "step": 52093 }, { "epoch": 0.05, "learning_rate": 4.9667158824935236e-05, "loss": 1.521, "step": 52094 }, { "epoch": 0.05, "learning_rate": 4.9667146050237724e-05, "loss": 0.9733, "step": 52095 }, { "epoch": 0.05, "learning_rate": 4.966713327529671e-05, "loss": 1.1643, "step": 52096 }, { "epoch": 0.05, "learning_rate": 4.966712050011219e-05, "loss": 0.8154, "step": 52097 }, { "epoch": 0.05, "learning_rate": 4.966710772468417e-05, "loss": 1.2312, "step": 52098 }, { "epoch": 0.05, "learning_rate": 4.9667094949012646e-05, "loss": 0.9694, "step": 52099 }, { "epoch": 0.05, "learning_rate": 4.9667082173097615e-05, "loss": 0.9098, "step": 52100 }, { "epoch": 0.05, "learning_rate": 4.9667069396939086e-05, "loss": 0.874, "step": 52101 }, { "epoch": 0.05, "learning_rate": 4.966705662053705e-05, "loss": 0.8626, "step": 52102 }, { "epoch": 0.05, "learning_rate": 4.9667043843891514e-05, "loss": 0.8463, "step": 52103 }, { "epoch": 0.05, "learning_rate": 4.966703106700248e-05, "loss": 0.8346, "step": 52104 }, { "epoch": 0.05, "learning_rate": 4.9667018289869935e-05, "loss": 0.8508, "step": 52105 }, { "epoch": 0.05, "learning_rate": 4.9667005512493894e-05, "loss": 0.8486, "step": 52106 }, { "epoch": 0.05, "learning_rate": 4.966699273487435e-05, "loss": 0.8424, "step": 52107 }, { "epoch": 0.05, "learning_rate": 4.966697995701129e-05, "loss": 0.8382, "step": 52108 }, { "epoch": 0.05, "learning_rate": 4.966696717890475e-05, "loss": 0.8449, "step": 52109 }, { "epoch": 0.05, "learning_rate": 4.9666954400554696e-05, "loss": 0.8236, "step": 52110 }, { "epoch": 0.05, "learning_rate": 4.966694162196114e-05, "loss": 0.8132, "step": 52111 }, { "epoch": 0.05, "learning_rate": 4.966692884312408e-05, "loss": 0.8249, "step": 52112 }, { "epoch": 0.05, "learning_rate": 4.966691606404353e-05, "loss": 0.8176, "step": 52113 }, { "epoch": 0.05, "learning_rate": 4.966690328471947e-05, "loss": 0.8056, "step": 52114 }, { "epoch": 0.05, "learning_rate": 4.9666890505151906e-05, "loss": 0.8572, "step": 52115 }, { "epoch": 0.05, "learning_rate": 4.966687772534085e-05, "loss": 0.8293, "step": 52116 }, { "epoch": 0.05, "learning_rate": 4.966686494528629e-05, "loss": 0.8326, "step": 52117 }, { "epoch": 0.05, "learning_rate": 4.966685216498823e-05, "loss": 0.8097, "step": 52118 }, { "epoch": 0.05, "learning_rate": 4.966683938444666e-05, "loss": 0.8164, "step": 52119 }, { "epoch": 0.05, "learning_rate": 4.96668266036616e-05, "loss": 0.835, "step": 52120 }, { "epoch": 0.05, "learning_rate": 4.966681382263303e-05, "loss": 0.8242, "step": 52121 }, { "epoch": 0.05, "learning_rate": 4.9666801041360964e-05, "loss": 0.8089, "step": 52122 }, { "epoch": 0.05, "learning_rate": 4.96667882598454e-05, "loss": 0.8099, "step": 52123 }, { "epoch": 0.05, "learning_rate": 4.9666775478086336e-05, "loss": 0.8041, "step": 52124 }, { "epoch": 0.05, "learning_rate": 4.966676269608377e-05, "loss": 0.8352, "step": 52125 }, { "epoch": 0.05, "learning_rate": 4.966674991383771e-05, "loss": 0.8237, "step": 52126 }, { "epoch": 0.05, "learning_rate": 4.966673713134814e-05, "loss": 0.8232, "step": 52127 }, { "epoch": 0.05, "learning_rate": 4.966672434861508e-05, "loss": 0.8138, "step": 52128 }, { "epoch": 0.05, "learning_rate": 4.966671156563851e-05, "loss": 0.8221, "step": 52129 }, { "epoch": 0.05, "learning_rate": 4.9666698782418446e-05, "loss": 0.8096, "step": 52130 }, { "epoch": 0.05, "learning_rate": 4.966668599895489e-05, "loss": 0.8267, "step": 52131 }, { "epoch": 0.05, "learning_rate": 4.966667321524782e-05, "loss": 0.8181, "step": 52132 }, { "epoch": 0.05, "learning_rate": 4.9666660431297265e-05, "loss": 0.8229, "step": 52133 }, { "epoch": 0.05, "learning_rate": 4.9666647647103205e-05, "loss": 0.8095, "step": 52134 }, { "epoch": 0.05, "learning_rate": 4.966663486266564e-05, "loss": 0.8243, "step": 52135 }, { "epoch": 0.05, "learning_rate": 4.966662207798458e-05, "loss": 0.8162, "step": 52136 }, { "epoch": 0.05, "learning_rate": 4.966660929306003e-05, "loss": 0.828, "step": 52137 }, { "epoch": 0.05, "learning_rate": 4.966659650789197e-05, "loss": 0.9173, "step": 52138 }, { "epoch": 0.05, "learning_rate": 4.966658372248042e-05, "loss": 1.3388, "step": 52139 }, { "epoch": 0.05, "learning_rate": 4.966657093682537e-05, "loss": 0.7117, "step": 52140 }, { "epoch": 0.05, "learning_rate": 4.9666558150926826e-05, "loss": 0.3596, "step": 52141 }, { "epoch": 0.05, "learning_rate": 4.966654536478478e-05, "loss": 1.2451, "step": 52142 }, { "epoch": 0.05, "learning_rate": 4.966653257839923e-05, "loss": 1.1854, "step": 52143 }, { "epoch": 0.05, "learning_rate": 4.966651979177019e-05, "loss": 0.9259, "step": 52144 }, { "epoch": 0.05, "learning_rate": 4.966650700489765e-05, "loss": 0.9478, "step": 52145 }, { "epoch": 0.05, "learning_rate": 4.966649421778161e-05, "loss": 0.6892, "step": 52146 }, { "epoch": 0.05, "learning_rate": 4.966648143042208e-05, "loss": 1.1562, "step": 52147 }, { "epoch": 0.05, "learning_rate": 4.9666468642819045e-05, "loss": 0.8988, "step": 52148 }, { "epoch": 0.05, "learning_rate": 4.966645585497252e-05, "loss": 1.2359, "step": 52149 }, { "epoch": 0.05, "learning_rate": 4.9666443066882494e-05, "loss": 1.0505, "step": 52150 }, { "epoch": 0.05, "learning_rate": 4.966643027854897e-05, "loss": 1.071, "step": 52151 }, { "epoch": 0.05, "learning_rate": 4.966641748997196e-05, "loss": 1.3864, "step": 52152 }, { "epoch": 0.05, "learning_rate": 4.966640470115144e-05, "loss": 1.2015, "step": 52153 }, { "epoch": 0.05, "learning_rate": 4.966639191208743e-05, "loss": 1.3605, "step": 52154 }, { "epoch": 0.05, "learning_rate": 4.9666379122779925e-05, "loss": 1.1425, "step": 52155 }, { "epoch": 0.05, "learning_rate": 4.966636633322892e-05, "loss": 1.2815, "step": 52156 }, { "epoch": 0.05, "learning_rate": 4.966635354343442e-05, "loss": 1.0937, "step": 52157 }, { "epoch": 0.05, "learning_rate": 4.966634075339642e-05, "loss": 1.4193, "step": 52158 }, { "epoch": 0.05, "learning_rate": 4.966632796311493e-05, "loss": 1.3705, "step": 52159 }, { "epoch": 0.05, "learning_rate": 4.9666315172589946e-05, "loss": 1.3058, "step": 52160 }, { "epoch": 0.05, "learning_rate": 4.966630238182146e-05, "loss": 1.1835, "step": 52161 }, { "epoch": 0.05, "learning_rate": 4.966628959080949e-05, "loss": 0.7187, "step": 52162 }, { "epoch": 0.05, "learning_rate": 4.966627679955401e-05, "loss": 0.542, "step": 52163 }, { "epoch": 0.05, "learning_rate": 4.9666264008055043e-05, "loss": 0.5875, "step": 52164 }, { "epoch": 0.05, "learning_rate": 4.9666251216312586e-05, "loss": 0.5724, "step": 52165 }, { "epoch": 0.05, "learning_rate": 4.966623842432663e-05, "loss": 0.4874, "step": 52166 }, { "epoch": 0.05, "learning_rate": 4.9666225632097175e-05, "loss": 0.482, "step": 52167 }, { "epoch": 0.05, "learning_rate": 4.966621283962423e-05, "loss": 0.3774, "step": 52168 }, { "epoch": 0.05, "learning_rate": 4.9666200046907786e-05, "loss": 0.4349, "step": 52169 }, { "epoch": 0.05, "learning_rate": 4.966618725394785e-05, "loss": 0.378, "step": 52170 }, { "epoch": 0.05, "learning_rate": 4.9666174460744416e-05, "loss": 0.4538, "step": 52171 }, { "epoch": 0.05, "learning_rate": 4.96661616672975e-05, "loss": 0.4397, "step": 52172 }, { "epoch": 0.05, "learning_rate": 4.9666148873607075e-05, "loss": 0.4044, "step": 52173 }, { "epoch": 0.05, "learning_rate": 4.9666136079673166e-05, "loss": 0.3852, "step": 52174 }, { "epoch": 0.05, "learning_rate": 4.966612328549576e-05, "loss": 0.4898, "step": 52175 }, { "epoch": 0.05, "learning_rate": 4.966611049107486e-05, "loss": 0.4931, "step": 52176 }, { "epoch": 0.05, "learning_rate": 4.9666097696410466e-05, "loss": 0.4552, "step": 52177 }, { "epoch": 0.05, "learning_rate": 4.9666084901502585e-05, "loss": 0.4384, "step": 52178 }, { "epoch": 0.05, "learning_rate": 4.96660721063512e-05, "loss": 0.3538, "step": 52179 }, { "epoch": 0.05, "learning_rate": 4.9666059310956335e-05, "loss": 0.4165, "step": 52180 }, { "epoch": 0.05, "learning_rate": 4.9666046515317966e-05, "loss": 0.4181, "step": 52181 }, { "epoch": 0.05, "learning_rate": 4.966603371943611e-05, "loss": 0.4447, "step": 52182 }, { "epoch": 0.05, "learning_rate": 4.966602092331076e-05, "loss": 0.4237, "step": 52183 }, { "epoch": 0.05, "learning_rate": 4.966600812694192e-05, "loss": 0.4903, "step": 52184 }, { "epoch": 0.05, "learning_rate": 4.966599533032958e-05, "loss": 0.4106, "step": 52185 }, { "epoch": 0.05, "learning_rate": 4.966598253347376e-05, "loss": 0.3703, "step": 52186 }, { "epoch": 0.05, "learning_rate": 4.966596973637443e-05, "loss": 0.3948, "step": 52187 }, { "epoch": 0.05, "learning_rate": 4.966595693903162e-05, "loss": 0.4204, "step": 52188 }, { "epoch": 0.05, "learning_rate": 4.966594414144532e-05, "loss": 0.4035, "step": 52189 }, { "epoch": 0.05, "learning_rate": 4.9665931343615523e-05, "loss": 0.354, "step": 52190 }, { "epoch": 0.05, "learning_rate": 4.966591854554223e-05, "loss": 0.3821, "step": 52191 }, { "epoch": 0.05, "learning_rate": 4.966590574722546e-05, "loss": 0.4056, "step": 52192 }, { "epoch": 0.05, "learning_rate": 4.9665892948665184e-05, "loss": 0.419, "step": 52193 }, { "epoch": 0.05, "learning_rate": 4.966588014986142e-05, "loss": 0.3872, "step": 52194 }, { "epoch": 0.05, "learning_rate": 4.966586735081417e-05, "loss": 0.3611, "step": 52195 }, { "epoch": 0.05, "learning_rate": 4.966585455152343e-05, "loss": 0.3782, "step": 52196 }, { "epoch": 0.05, "learning_rate": 4.9665841751989197e-05, "loss": 0.3899, "step": 52197 }, { "epoch": 0.05, "learning_rate": 4.9665828952211466e-05, "loss": 0.4095, "step": 52198 }, { "epoch": 0.05, "learning_rate": 4.966581615219025e-05, "loss": 0.3554, "step": 52199 }, { "epoch": 0.05, "learning_rate": 4.966580335192554e-05, "loss": 0.3381, "step": 52200 }, { "epoch": 0.05, "learning_rate": 4.9665790551417345e-05, "loss": 0.388, "step": 52201 }, { "epoch": 0.05, "learning_rate": 4.9665777750665655e-05, "loss": 0.3247, "step": 52202 }, { "epoch": 0.05, "learning_rate": 4.966576494967048e-05, "loss": 0.3077, "step": 52203 }, { "epoch": 0.05, "learning_rate": 4.966575214843181e-05, "loss": 0.3215, "step": 52204 }, { "epoch": 0.05, "learning_rate": 4.966573934694966e-05, "loss": 0.3591, "step": 52205 }, { "epoch": 0.05, "learning_rate": 4.966572654522401e-05, "loss": 0.3954, "step": 52206 }, { "epoch": 0.05, "learning_rate": 4.9665713743254874e-05, "loss": 0.3932, "step": 52207 }, { "epoch": 0.05, "learning_rate": 4.966570094104224e-05, "loss": 0.4128, "step": 52208 }, { "epoch": 0.05, "learning_rate": 4.966568813858613e-05, "loss": 0.397, "step": 52209 }, { "epoch": 0.05, "learning_rate": 4.966567533588652e-05, "loss": 0.4063, "step": 52210 }, { "epoch": 0.05, "learning_rate": 4.9665662532943424e-05, "loss": 0.3902, "step": 52211 }, { "epoch": 0.05, "learning_rate": 4.9665649729756846e-05, "loss": 0.367, "step": 52212 }, { "epoch": 0.05, "learning_rate": 4.966563692632678e-05, "loss": 0.3387, "step": 52213 }, { "epoch": 0.05, "learning_rate": 4.966562412265321e-05, "loss": 0.3726, "step": 52214 }, { "epoch": 0.05, "learning_rate": 4.966561131873616e-05, "loss": 0.2915, "step": 52215 }, { "epoch": 0.05, "learning_rate": 4.966559851457562e-05, "loss": 0.2401, "step": 52216 }, { "epoch": 0.05, "learning_rate": 4.966558571017159e-05, "loss": 0.3201, "step": 52217 }, { "epoch": 0.05, "learning_rate": 4.966557290552408e-05, "loss": 0.2775, "step": 52218 }, { "epoch": 0.05, "learning_rate": 4.966556010063308e-05, "loss": 0.3052, "step": 52219 }, { "epoch": 0.05, "learning_rate": 4.966554729549858e-05, "loss": 0.3583, "step": 52220 }, { "epoch": 0.05, "learning_rate": 4.9665534490120606e-05, "loss": 0.3936, "step": 52221 }, { "epoch": 0.05, "learning_rate": 4.966552168449914e-05, "loss": 0.3901, "step": 52222 }, { "epoch": 0.05, "learning_rate": 4.9665508878634184e-05, "loss": 1.2757, "step": 52223 }, { "epoch": 0.05, "learning_rate": 4.966549607252574e-05, "loss": 1.0228, "step": 52224 }, { "epoch": 0.05, "learning_rate": 4.9665483266173816e-05, "loss": 1.0983, "step": 52225 }, { "epoch": 0.05, "learning_rate": 4.96654704595784e-05, "loss": 1.4193, "step": 52226 }, { "epoch": 0.05, "learning_rate": 4.9665457652739496e-05, "loss": 1.0788, "step": 52227 }, { "epoch": 0.05, "learning_rate": 4.96654448456571e-05, "loss": 0.8148, "step": 52228 }, { "epoch": 0.05, "learning_rate": 4.966543203833123e-05, "loss": 1.2267, "step": 52229 }, { "epoch": 0.05, "learning_rate": 4.966541923076186e-05, "loss": 1.1171, "step": 52230 }, { "epoch": 0.05, "learning_rate": 4.966540642294901e-05, "loss": 1.5675, "step": 52231 }, { "epoch": 0.05, "learning_rate": 4.966539361489267e-05, "loss": 1.2103, "step": 52232 }, { "epoch": 0.05, "learning_rate": 4.966538080659285e-05, "loss": 1.313, "step": 52233 }, { "epoch": 0.05, "learning_rate": 4.966536799804954e-05, "loss": 1.3657, "step": 52234 }, { "epoch": 0.05, "learning_rate": 4.966535518926274e-05, "loss": 1.5412, "step": 52235 }, { "epoch": 0.05, "learning_rate": 4.966534238023246e-05, "loss": 1.2317, "step": 52236 }, { "epoch": 0.05, "learning_rate": 4.966532957095869e-05, "loss": 2.092, "step": 52237 }, { "epoch": 0.05, "learning_rate": 4.966531676144144e-05, "loss": 1.0445, "step": 52238 }, { "epoch": 0.05, "learning_rate": 4.96653039516807e-05, "loss": 1.2456, "step": 52239 }, { "epoch": 0.05, "learning_rate": 4.9665291141676475e-05, "loss": 1.1981, "step": 52240 }, { "epoch": 0.05, "learning_rate": 4.9665278331428766e-05, "loss": 1.3247, "step": 52241 }, { "epoch": 0.05, "learning_rate": 4.9665265520937566e-05, "loss": 1.2126, "step": 52242 }, { "epoch": 0.05, "learning_rate": 4.966525271020289e-05, "loss": 1.0957, "step": 52243 }, { "epoch": 0.05, "learning_rate": 4.9665239899224726e-05, "loss": 1.3608, "step": 52244 }, { "epoch": 0.05, "learning_rate": 4.966522708800307e-05, "loss": 1.1038, "step": 52245 }, { "epoch": 0.05, "learning_rate": 4.966521427653793e-05, "loss": 1.1411, "step": 52246 }, { "epoch": 0.05, "learning_rate": 4.9665201464829316e-05, "loss": 1.3164, "step": 52247 }, { "epoch": 0.05, "learning_rate": 4.9665188652877214e-05, "loss": 1.1648, "step": 52248 }, { "epoch": 0.05, "learning_rate": 4.966517584068162e-05, "loss": 1.1443, "step": 52249 }, { "epoch": 0.05, "learning_rate": 4.966516302824255e-05, "loss": 1.6526, "step": 52250 }, { "epoch": 0.05, "learning_rate": 4.9665150215559995e-05, "loss": 1.3271, "step": 52251 }, { "epoch": 0.05, "learning_rate": 4.9665137402633955e-05, "loss": 1.4214, "step": 52252 }, { "epoch": 0.05, "learning_rate": 4.966512458946443e-05, "loss": 1.1782, "step": 52253 }, { "epoch": 0.05, "learning_rate": 4.966511177605142e-05, "loss": 0.9843, "step": 52254 }, { "epoch": 0.05, "learning_rate": 4.9665098962394926e-05, "loss": 0.9861, "step": 52255 }, { "epoch": 0.05, "learning_rate": 4.9665086148494954e-05, "loss": 1.6322, "step": 52256 }, { "epoch": 0.05, "learning_rate": 4.966507333435149e-05, "loss": 1.1624, "step": 52257 }, { "epoch": 0.05, "learning_rate": 4.966506051996455e-05, "loss": 1.2777, "step": 52258 }, { "epoch": 0.05, "learning_rate": 4.966504770533412e-05, "loss": 1.4121, "step": 52259 }, { "epoch": 0.05, "learning_rate": 4.966503489046022e-05, "loss": 0.8739, "step": 52260 }, { "epoch": 0.05, "learning_rate": 4.966502207534283e-05, "loss": 0.7016, "step": 52261 }, { "epoch": 0.05, "learning_rate": 4.9665009259981955e-05, "loss": 2.0736, "step": 52262 }, { "epoch": 0.05, "learning_rate": 4.9664996444377596e-05, "loss": 2.2101, "step": 52263 }, { "epoch": 0.05, "learning_rate": 4.966498362852976e-05, "loss": 2.0734, "step": 52264 }, { "epoch": 0.05, "learning_rate": 4.966497081243844e-05, "loss": 1.5519, "step": 52265 }, { "epoch": 0.05, "learning_rate": 4.966495799610363e-05, "loss": 1.2127, "step": 52266 }, { "epoch": 0.05, "learning_rate": 4.966494517952535e-05, "loss": 1.0901, "step": 52267 }, { "epoch": 0.05, "learning_rate": 4.966493236270358e-05, "loss": 0.8855, "step": 52268 }, { "epoch": 0.05, "learning_rate": 4.9664919545638335e-05, "loss": 1.4033, "step": 52269 }, { "epoch": 0.05, "learning_rate": 4.9664906728329604e-05, "loss": 1.6805, "step": 52270 }, { "epoch": 0.05, "learning_rate": 4.966489391077739e-05, "loss": 0.8307, "step": 52271 }, { "epoch": 0.05, "learning_rate": 4.9664881092981694e-05, "loss": 1.0869, "step": 52272 }, { "epoch": 0.05, "learning_rate": 4.966486827494252e-05, "loss": 0.7334, "step": 52273 }, { "epoch": 0.05, "learning_rate": 4.9664855456659866e-05, "loss": 0.9782, "step": 52274 }, { "epoch": 0.05, "learning_rate": 4.9664842638133725e-05, "loss": 1.4505, "step": 52275 }, { "epoch": 0.05, "learning_rate": 4.966482981936411e-05, "loss": 0.9972, "step": 52276 }, { "epoch": 0.05, "learning_rate": 4.9664817000351e-05, "loss": 0.9202, "step": 52277 }, { "epoch": 0.05, "learning_rate": 4.966480418109443e-05, "loss": 1.2436, "step": 52278 }, { "epoch": 0.05, "learning_rate": 4.966479136159436e-05, "loss": 1.4621, "step": 52279 }, { "epoch": 0.05, "learning_rate": 4.966477854185083e-05, "loss": 1.0606, "step": 52280 }, { "epoch": 0.05, "learning_rate": 4.96647657218638e-05, "loss": 0.9117, "step": 52281 }, { "epoch": 0.05, "learning_rate": 4.96647529016333e-05, "loss": 1.2618, "step": 52282 }, { "epoch": 0.05, "learning_rate": 4.9664740081159324e-05, "loss": 1.169, "step": 52283 }, { "epoch": 0.05, "learning_rate": 4.966472726044186e-05, "loss": 1.3812, "step": 52284 }, { "epoch": 0.05, "learning_rate": 4.966471443948092e-05, "loss": 1.0927, "step": 52285 }, { "epoch": 0.05, "learning_rate": 4.9664701618276496e-05, "loss": 0.9917, "step": 52286 }, { "epoch": 0.05, "learning_rate": 4.96646887968286e-05, "loss": 0.8905, "step": 52287 }, { "epoch": 0.05, "learning_rate": 4.966467597513722e-05, "loss": 1.4075, "step": 52288 }, { "epoch": 0.05, "learning_rate": 4.966466315320236e-05, "loss": 1.165, "step": 52289 }, { "epoch": 0.05, "learning_rate": 4.966465033102403e-05, "loss": 1.4168, "step": 52290 }, { "epoch": 0.05, "learning_rate": 4.966463750860221e-05, "loss": 1.3754, "step": 52291 }, { "epoch": 0.05, "learning_rate": 4.9664624685936914e-05, "loss": 1.1195, "step": 52292 }, { "epoch": 0.05, "learning_rate": 4.966461186302814e-05, "loss": 1.4041, "step": 52293 }, { "epoch": 0.05, "learning_rate": 4.966459903987588e-05, "loss": 1.0274, "step": 52294 }, { "epoch": 0.05, "learning_rate": 4.9664586216480155e-05, "loss": 1.1988, "step": 52295 }, { "epoch": 0.05, "learning_rate": 4.966457339284094e-05, "loss": 0.9235, "step": 52296 }, { "epoch": 0.05, "learning_rate": 4.9664560568958254e-05, "loss": 1.2991, "step": 52297 }, { "epoch": 0.05, "learning_rate": 4.966454774483209e-05, "loss": 1.316, "step": 52298 }, { "epoch": 0.05, "learning_rate": 4.966453492046244e-05, "loss": 1.1442, "step": 52299 }, { "epoch": 0.05, "learning_rate": 4.966452209584932e-05, "loss": 1.1229, "step": 52300 }, { "epoch": 0.05, "learning_rate": 4.966450927099272e-05, "loss": 1.1125, "step": 52301 }, { "epoch": 0.05, "learning_rate": 4.966449644589264e-05, "loss": 1.2735, "step": 52302 }, { "epoch": 0.05, "learning_rate": 4.966448362054908e-05, "loss": 1.2242, "step": 52303 }, { "epoch": 0.05, "learning_rate": 4.966447079496205e-05, "loss": 1.0751, "step": 52304 }, { "epoch": 0.05, "learning_rate": 4.966445796913154e-05, "loss": 0.986, "step": 52305 }, { "epoch": 0.05, "learning_rate": 4.966444514305755e-05, "loss": 1.0501, "step": 52306 }, { "epoch": 0.05, "learning_rate": 4.966443231674009e-05, "loss": 0.9181, "step": 52307 }, { "epoch": 0.05, "learning_rate": 4.966441949017915e-05, "loss": 0.7686, "step": 52308 }, { "epoch": 0.05, "learning_rate": 4.966440666337473e-05, "loss": 1.3423, "step": 52309 }, { "epoch": 0.05, "learning_rate": 4.9664393836326836e-05, "loss": 1.0103, "step": 52310 }, { "epoch": 0.05, "learning_rate": 4.966438100903546e-05, "loss": 0.8665, "step": 52311 }, { "epoch": 0.05, "learning_rate": 4.966436818150062e-05, "loss": 0.9322, "step": 52312 }, { "epoch": 0.05, "learning_rate": 4.966435535372229e-05, "loss": 0.9754, "step": 52313 }, { "epoch": 0.05, "learning_rate": 4.966434252570049e-05, "loss": 1.2218, "step": 52314 }, { "epoch": 0.05, "learning_rate": 4.966432969743522e-05, "loss": 1.0827, "step": 52315 }, { "epoch": 0.05, "learning_rate": 4.966431686892646e-05, "loss": 1.1135, "step": 52316 }, { "epoch": 0.05, "learning_rate": 4.966430404017424e-05, "loss": 1.052, "step": 52317 }, { "epoch": 0.05, "learning_rate": 4.9664291211178534e-05, "loss": 1.3317, "step": 52318 }, { "epoch": 0.05, "learning_rate": 4.966427838193936e-05, "loss": 1.39, "step": 52319 }, { "epoch": 0.05, "learning_rate": 4.9664265552456703e-05, "loss": 1.1819, "step": 52320 }, { "epoch": 0.05, "learning_rate": 4.966425272273058e-05, "loss": 0.9536, "step": 52321 }, { "epoch": 0.05, "learning_rate": 4.966423989276097e-05, "loss": 0.9333, "step": 52322 }, { "epoch": 0.05, "learning_rate": 4.9664227062547886e-05, "loss": 1.0831, "step": 52323 }, { "epoch": 0.05, "learning_rate": 4.9664214232091333e-05, "loss": 0.8525, "step": 52324 }, { "epoch": 0.05, "learning_rate": 4.966420140139131e-05, "loss": 0.7367, "step": 52325 }, { "epoch": 0.05, "learning_rate": 4.96641885704478e-05, "loss": 1.1106, "step": 52326 }, { "epoch": 0.05, "learning_rate": 4.966417573926082e-05, "loss": 1.1206, "step": 52327 }, { "epoch": 0.05, "learning_rate": 4.966416290783037e-05, "loss": 0.9192, "step": 52328 }, { "epoch": 0.05, "learning_rate": 4.966415007615645e-05, "loss": 0.5169, "step": 52329 }, { "epoch": 0.05, "learning_rate": 4.966413724423904e-05, "loss": 1.0599, "step": 52330 }, { "epoch": 0.05, "learning_rate": 4.966412441207817e-05, "loss": 1.7012, "step": 52331 }, { "epoch": 0.05, "learning_rate": 4.966411157967382e-05, "loss": 0.6451, "step": 52332 }, { "epoch": 0.05, "learning_rate": 4.9664098747026e-05, "loss": 1.0821, "step": 52333 }, { "epoch": 0.05, "learning_rate": 4.9664085914134704e-05, "loss": 1.2234, "step": 52334 }, { "epoch": 0.05, "learning_rate": 4.9664073080999935e-05, "loss": 1.0423, "step": 52335 }, { "epoch": 0.05, "learning_rate": 4.966406024762169e-05, "loss": 1.6216, "step": 52336 }, { "epoch": 0.05, "learning_rate": 4.9664047413999977e-05, "loss": 1.011, "step": 52337 }, { "epoch": 0.05, "learning_rate": 4.966403458013479e-05, "loss": 1.3298, "step": 52338 }, { "epoch": 0.05, "learning_rate": 4.966402174602613e-05, "loss": 1.2345, "step": 52339 }, { "epoch": 0.05, "learning_rate": 4.966400891167399e-05, "loss": 1.1194, "step": 52340 }, { "epoch": 0.05, "learning_rate": 4.9663996077078385e-05, "loss": 1.6122, "step": 52341 }, { "epoch": 0.05, "learning_rate": 4.9663983242239305e-05, "loss": 2.0209, "step": 52342 }, { "epoch": 0.05, "learning_rate": 4.966397040715675e-05, "loss": 1.6462, "step": 52343 }, { "epoch": 0.05, "learning_rate": 4.966395757183073e-05, "loss": 1.7798, "step": 52344 }, { "epoch": 0.05, "learning_rate": 4.966394473626123e-05, "loss": 1.7642, "step": 52345 }, { "epoch": 0.05, "learning_rate": 4.9663931900448257e-05, "loss": 1.071, "step": 52346 }, { "epoch": 0.05, "learning_rate": 4.966391906439182e-05, "loss": 1.3141, "step": 52347 }, { "epoch": 0.05, "learning_rate": 4.9663906228091906e-05, "loss": 1.3022, "step": 52348 }, { "epoch": 0.05, "learning_rate": 4.966389339154852e-05, "loss": 1.1469, "step": 52349 }, { "epoch": 0.05, "learning_rate": 4.966388055476166e-05, "loss": 1.0741, "step": 52350 }, { "epoch": 0.05, "learning_rate": 4.9663867717731335e-05, "loss": 1.0519, "step": 52351 }, { "epoch": 0.05, "learning_rate": 4.9663854880457536e-05, "loss": 1.4569, "step": 52352 }, { "epoch": 0.05, "learning_rate": 4.9663842042940266e-05, "loss": 1.3563, "step": 52353 }, { "epoch": 0.05, "learning_rate": 4.966382920517953e-05, "loss": 0.8598, "step": 52354 }, { "epoch": 0.05, "learning_rate": 4.966381636717532e-05, "loss": 1.2605, "step": 52355 }, { "epoch": 0.05, "learning_rate": 4.9663803528927635e-05, "loss": 0.9527, "step": 52356 }, { "epoch": 0.05, "learning_rate": 4.966379069043649e-05, "loss": 1.3508, "step": 52357 }, { "epoch": 0.05, "learning_rate": 4.9663777851701855e-05, "loss": 1.0188, "step": 52358 }, { "epoch": 0.05, "learning_rate": 4.9663765012723764e-05, "loss": 1.2323, "step": 52359 }, { "epoch": 0.05, "learning_rate": 4.9663752173502196e-05, "loss": 1.0069, "step": 52360 }, { "epoch": 0.05, "learning_rate": 4.9663739334037164e-05, "loss": 1.2482, "step": 52361 }, { "epoch": 0.05, "learning_rate": 4.966372649432865e-05, "loss": 1.6358, "step": 52362 }, { "epoch": 0.05, "learning_rate": 4.9663713654376685e-05, "loss": 1.3663, "step": 52363 }, { "epoch": 0.05, "learning_rate": 4.966370081418124e-05, "loss": 0.9313, "step": 52364 }, { "epoch": 0.05, "learning_rate": 4.966368797374232e-05, "loss": 0.9837, "step": 52365 }, { "epoch": 0.05, "learning_rate": 4.966367513305994e-05, "loss": 1.197, "step": 52366 }, { "epoch": 0.05, "learning_rate": 4.966366229213409e-05, "loss": 1.2715, "step": 52367 }, { "epoch": 0.05, "learning_rate": 4.966364945096476e-05, "loss": 1.002, "step": 52368 }, { "epoch": 0.05, "learning_rate": 4.9663636609551975e-05, "loss": 1.2011, "step": 52369 }, { "epoch": 0.05, "learning_rate": 4.9663623767895716e-05, "loss": 1.1122, "step": 52370 }, { "epoch": 0.05, "learning_rate": 4.9663610925995985e-05, "loss": 1.203, "step": 52371 }, { "epoch": 0.05, "learning_rate": 4.966359808385278e-05, "loss": 1.3143, "step": 52372 }, { "epoch": 0.05, "learning_rate": 4.966358524146612e-05, "loss": 1.1736, "step": 52373 }, { "epoch": 0.05, "learning_rate": 4.966357239883599e-05, "loss": 1.0603, "step": 52374 }, { "epoch": 0.05, "learning_rate": 4.966355955596238e-05, "loss": 0.9869, "step": 52375 }, { "epoch": 0.05, "learning_rate": 4.966354671284531e-05, "loss": 1.0473, "step": 52376 }, { "epoch": 0.05, "learning_rate": 4.966353386948477e-05, "loss": 1.0757, "step": 52377 }, { "epoch": 0.05, "learning_rate": 4.9663521025880753e-05, "loss": 1.024, "step": 52378 }, { "epoch": 0.05, "learning_rate": 4.966350818203328e-05, "loss": 0.9148, "step": 52379 }, { "epoch": 0.05, "learning_rate": 4.966349533794234e-05, "loss": 1.0522, "step": 52380 }, { "epoch": 0.05, "learning_rate": 4.966348249360793e-05, "loss": 1.1035, "step": 52381 }, { "epoch": 0.05, "learning_rate": 4.966346964903005e-05, "loss": 1.2951, "step": 52382 }, { "epoch": 0.05, "learning_rate": 4.96634568042087e-05, "loss": 1.3964, "step": 52383 }, { "epoch": 0.05, "learning_rate": 4.9663443959143884e-05, "loss": 1.2471, "step": 52384 }, { "epoch": 0.05, "learning_rate": 4.9663431113835605e-05, "loss": 1.0191, "step": 52385 }, { "epoch": 0.05, "learning_rate": 4.9663418268283854e-05, "loss": 0.6731, "step": 52386 }, { "epoch": 0.05, "learning_rate": 4.966340542248864e-05, "loss": 1.147, "step": 52387 }, { "epoch": 0.05, "learning_rate": 4.966339257644996e-05, "loss": 1.1394, "step": 52388 }, { "epoch": 0.05, "learning_rate": 4.966337973016782e-05, "loss": 1.0449, "step": 52389 }, { "epoch": 0.05, "learning_rate": 4.96633668836422e-05, "loss": 1.5423, "step": 52390 }, { "epoch": 0.05, "learning_rate": 4.966335403687312e-05, "loss": 1.1961, "step": 52391 }, { "epoch": 0.05, "learning_rate": 4.9663341189860566e-05, "loss": 1.4979, "step": 52392 }, { "epoch": 0.05, "learning_rate": 4.966332834260456e-05, "loss": 1.0509, "step": 52393 }, { "epoch": 0.05, "learning_rate": 4.966331549510508e-05, "loss": 0.6138, "step": 52394 }, { "epoch": 0.05, "learning_rate": 4.966330264736213e-05, "loss": 1.0537, "step": 52395 }, { "epoch": 0.05, "learning_rate": 4.966328979937572e-05, "loss": 1.09, "step": 52396 }, { "epoch": 0.05, "learning_rate": 4.966327695114584e-05, "loss": 1.3453, "step": 52397 }, { "epoch": 0.05, "learning_rate": 4.96632641026725e-05, "loss": 1.0269, "step": 52398 }, { "epoch": 0.05, "learning_rate": 4.96632512539557e-05, "loss": 0.4969, "step": 52399 }, { "epoch": 0.05, "learning_rate": 4.966323840499543e-05, "loss": 1.4004, "step": 52400 }, { "epoch": 0.05, "learning_rate": 4.9663225555791684e-05, "loss": 1.2236, "step": 52401 }, { "epoch": 0.05, "learning_rate": 4.966321270634448e-05, "loss": 1.2448, "step": 52402 }, { "epoch": 0.05, "learning_rate": 4.966319985665382e-05, "loss": 1.0684, "step": 52403 }, { "epoch": 0.05, "learning_rate": 4.966318700671969e-05, "loss": 1.1323, "step": 52404 }, { "epoch": 0.05, "learning_rate": 4.9663174156542095e-05, "loss": 1.2076, "step": 52405 }, { "epoch": 0.05, "learning_rate": 4.966316130612103e-05, "loss": 1.0883, "step": 52406 }, { "epoch": 0.05, "learning_rate": 4.96631484554565e-05, "loss": 1.2781, "step": 52407 }, { "epoch": 0.05, "learning_rate": 4.9663135604548514e-05, "loss": 0.7116, "step": 52408 }, { "epoch": 0.05, "learning_rate": 4.966312275339706e-05, "loss": 0.3162, "step": 52409 }, { "epoch": 0.05, "learning_rate": 4.966310990200214e-05, "loss": 0.3009, "step": 52410 }, { "epoch": 0.05, "learning_rate": 4.966309705036376e-05, "loss": 0.3397, "step": 52411 }, { "epoch": 0.05, "learning_rate": 4.9663084198481915e-05, "loss": 0.2795, "step": 52412 }, { "epoch": 0.05, "learning_rate": 4.9663071346356607e-05, "loss": 0.2088, "step": 52413 }, { "epoch": 0.05, "learning_rate": 4.966305849398783e-05, "loss": 0.3966, "step": 52414 }, { "epoch": 0.05, "learning_rate": 4.9663045641375596e-05, "loss": 1.1493, "step": 52415 }, { "epoch": 0.05, "learning_rate": 4.96630327885199e-05, "loss": 0.8201, "step": 52416 }, { "epoch": 0.05, "learning_rate": 4.966301993542074e-05, "loss": 1.2206, "step": 52417 }, { "epoch": 0.05, "learning_rate": 4.966300708207811e-05, "loss": 1.2574, "step": 52418 }, { "epoch": 0.05, "learning_rate": 4.966299422849202e-05, "loss": 0.3444, "step": 52419 }, { "epoch": 0.05, "learning_rate": 4.9662981374662476e-05, "loss": 0.3989, "step": 52420 }, { "epoch": 0.05, "learning_rate": 4.966296852058946e-05, "loss": 1.1933, "step": 52421 }, { "epoch": 0.05, "learning_rate": 4.966295566627298e-05, "loss": 1.2475, "step": 52422 }, { "epoch": 0.05, "learning_rate": 4.9662942811713046e-05, "loss": 1.1589, "step": 52423 }, { "epoch": 0.05, "learning_rate": 4.966292995690965e-05, "loss": 1.6768, "step": 52424 }, { "epoch": 0.05, "learning_rate": 4.966291710186278e-05, "loss": 1.1351, "step": 52425 }, { "epoch": 0.05, "learning_rate": 4.966290424657246e-05, "loss": 0.7844, "step": 52426 }, { "epoch": 0.05, "learning_rate": 4.966289139103868e-05, "loss": 1.1106, "step": 52427 }, { "epoch": 0.05, "learning_rate": 4.9662878535261426e-05, "loss": 1.2516, "step": 52428 }, { "epoch": 0.05, "learning_rate": 4.966286567924072e-05, "loss": 1.2268, "step": 52429 }, { "epoch": 0.05, "learning_rate": 4.9662852822976545e-05, "loss": 1.4871, "step": 52430 }, { "epoch": 0.05, "learning_rate": 4.9662839966468924e-05, "loss": 1.2483, "step": 52431 }, { "epoch": 0.05, "learning_rate": 4.9662827109717826e-05, "loss": 0.969, "step": 52432 }, { "epoch": 0.05, "learning_rate": 4.966281425272327e-05, "loss": 0.7329, "step": 52433 }, { "epoch": 0.05, "learning_rate": 4.966280139548526e-05, "loss": 0.6142, "step": 52434 }, { "epoch": 0.05, "learning_rate": 4.966278853800378e-05, "loss": 0.3351, "step": 52435 }, { "epoch": 0.05, "learning_rate": 4.9662775680278845e-05, "loss": 1.0107, "step": 52436 }, { "epoch": 0.05, "learning_rate": 4.966276282231045e-05, "loss": 0.9821, "step": 52437 }, { "epoch": 0.05, "learning_rate": 4.9662749964098595e-05, "loss": 1.2907, "step": 52438 }, { "epoch": 0.05, "learning_rate": 4.966273710564327e-05, "loss": 1.3155, "step": 52439 }, { "epoch": 0.05, "learning_rate": 4.96627242469445e-05, "loss": 1.0425, "step": 52440 }, { "epoch": 0.05, "learning_rate": 4.966271138800226e-05, "loss": 1.0594, "step": 52441 }, { "epoch": 0.05, "learning_rate": 4.966269852881656e-05, "loss": 1.5366, "step": 52442 }, { "epoch": 0.05, "learning_rate": 4.966268566938741e-05, "loss": 0.966, "step": 52443 }, { "epoch": 0.05, "learning_rate": 4.966267280971479e-05, "loss": 0.6685, "step": 52444 }, { "epoch": 0.05, "learning_rate": 4.9662659949798714e-05, "loss": 1.6038, "step": 52445 }, { "epoch": 0.05, "learning_rate": 4.966264708963918e-05, "loss": 1.072, "step": 52446 }, { "epoch": 0.05, "learning_rate": 4.966263422923618e-05, "loss": 1.0134, "step": 52447 }, { "epoch": 0.05, "learning_rate": 4.9662621368589726e-05, "loss": 1.1944, "step": 52448 }, { "epoch": 0.05, "learning_rate": 4.9662608507699815e-05, "loss": 1.2265, "step": 52449 }, { "epoch": 0.05, "learning_rate": 4.9662595646566447e-05, "loss": 1.08, "step": 52450 }, { "epoch": 0.05, "learning_rate": 4.9662582785189613e-05, "loss": 1.1481, "step": 52451 }, { "epoch": 0.05, "learning_rate": 4.966256992356932e-05, "loss": 1.1634, "step": 52452 }, { "epoch": 0.05, "learning_rate": 4.9662557061705575e-05, "loss": 1.1226, "step": 52453 }, { "epoch": 0.05, "learning_rate": 4.966254419959837e-05, "loss": 1.0723, "step": 52454 }, { "epoch": 0.05, "learning_rate": 4.9662531337247705e-05, "loss": 1.1611, "step": 52455 }, { "epoch": 0.05, "learning_rate": 4.9662518474653584e-05, "loss": 1.038, "step": 52456 }, { "epoch": 0.05, "learning_rate": 4.9662505611816e-05, "loss": 1.3232, "step": 52457 }, { "epoch": 0.05, "learning_rate": 4.966249274873496e-05, "loss": 1.34, "step": 52458 }, { "epoch": 0.05, "learning_rate": 4.966247988541047e-05, "loss": 1.3859, "step": 52459 }, { "epoch": 0.05, "learning_rate": 4.966246702184251e-05, "loss": 1.2195, "step": 52460 }, { "epoch": 0.05, "learning_rate": 4.96624541580311e-05, "loss": 1.1647, "step": 52461 }, { "epoch": 0.05, "learning_rate": 4.9662441293976235e-05, "loss": 1.0892, "step": 52462 }, { "epoch": 0.05, "learning_rate": 4.966242842967791e-05, "loss": 1.2389, "step": 52463 }, { "epoch": 0.05, "learning_rate": 4.966241556513612e-05, "loss": 1.1004, "step": 52464 }, { "epoch": 0.05, "learning_rate": 4.966240270035088e-05, "loss": 1.1097, "step": 52465 }, { "epoch": 0.05, "learning_rate": 4.9662389835322185e-05, "loss": 1.0663, "step": 52466 }, { "epoch": 0.05, "learning_rate": 4.966237697005003e-05, "loss": 1.4602, "step": 52467 }, { "epoch": 0.05, "learning_rate": 4.9662364104534425e-05, "loss": 1.2157, "step": 52468 }, { "epoch": 0.05, "learning_rate": 4.966235123877536e-05, "loss": 1.258, "step": 52469 }, { "epoch": 0.05, "learning_rate": 4.9662338372772835e-05, "loss": 1.0472, "step": 52470 }, { "epoch": 0.05, "learning_rate": 4.966232550652686e-05, "loss": 0.965, "step": 52471 }, { "epoch": 0.05, "learning_rate": 4.966231264003742e-05, "loss": 1.72, "step": 52472 }, { "epoch": 0.05, "learning_rate": 4.9662299773304534e-05, "loss": 1.3196, "step": 52473 }, { "epoch": 0.05, "learning_rate": 4.966228690632818e-05, "loss": 0.9164, "step": 52474 }, { "epoch": 0.05, "learning_rate": 4.966227403910838e-05, "loss": 1.2754, "step": 52475 }, { "epoch": 0.05, "learning_rate": 4.966226117164512e-05, "loss": 1.5255, "step": 52476 }, { "epoch": 0.05, "learning_rate": 4.9662248303938405e-05, "loss": 1.2944, "step": 52477 }, { "epoch": 0.05, "learning_rate": 4.966223543598824e-05, "loss": 1.2525, "step": 52478 }, { "epoch": 0.05, "learning_rate": 4.966222256779461e-05, "loss": 0.9495, "step": 52479 }, { "epoch": 0.05, "learning_rate": 4.9662209699357535e-05, "loss": 1.0673, "step": 52480 }, { "epoch": 0.05, "learning_rate": 4.9662196830677e-05, "loss": 1.1379, "step": 52481 }, { "epoch": 0.05, "learning_rate": 4.9662183961753017e-05, "loss": 0.7521, "step": 52482 }, { "epoch": 0.05, "learning_rate": 4.966217109258557e-05, "loss": 1.2565, "step": 52483 }, { "epoch": 0.05, "learning_rate": 4.966215822317467e-05, "loss": 1.0279, "step": 52484 }, { "epoch": 0.05, "learning_rate": 4.966214535352032e-05, "loss": 0.9, "step": 52485 }, { "epoch": 0.05, "learning_rate": 4.9662132483622515e-05, "loss": 1.1155, "step": 52486 }, { "epoch": 0.05, "learning_rate": 4.966211961348125e-05, "loss": 0.7999, "step": 52487 }, { "epoch": 0.05, "learning_rate": 4.966210674309654e-05, "loss": 0.8942, "step": 52488 }, { "epoch": 0.05, "learning_rate": 4.9662093872468365e-05, "loss": 1.1467, "step": 52489 }, { "epoch": 0.05, "learning_rate": 4.966208100159675e-05, "loss": 0.9167, "step": 52490 }, { "epoch": 0.05, "learning_rate": 4.9662068130481664e-05, "loss": 1.0602, "step": 52491 }, { "epoch": 0.05, "learning_rate": 4.966205525912314e-05, "loss": 1.3749, "step": 52492 }, { "epoch": 0.05, "learning_rate": 4.966204238752115e-05, "loss": 1.0849, "step": 52493 }, { "epoch": 0.05, "learning_rate": 4.966202951567571e-05, "loss": 0.9083, "step": 52494 }, { "epoch": 0.05, "learning_rate": 4.966201664358683e-05, "loss": 1.1379, "step": 52495 }, { "epoch": 0.05, "learning_rate": 4.966200377125448e-05, "loss": 1.1526, "step": 52496 }, { "epoch": 0.05, "learning_rate": 4.9661990898678687e-05, "loss": 1.0743, "step": 52497 }, { "epoch": 0.05, "learning_rate": 4.9661978025859434e-05, "loss": 1.2279, "step": 52498 }, { "epoch": 0.05, "learning_rate": 4.966196515279674e-05, "loss": 1.2296, "step": 52499 }, { "epoch": 0.05, "learning_rate": 4.9661952279490585e-05, "loss": 0.9678, "step": 52500 }, { "epoch": 0.05, "eval_loss": 1.0748170614242554, "eval_runtime": 127.1145, "eval_samples_per_second": 10.896, "eval_steps_per_second": 5.452, "step": 52500 }, { "epoch": 0.05, "learning_rate": 4.966193940594098e-05, "loss": 0.9922, "step": 52501 }, { "epoch": 0.05, "learning_rate": 4.966192653214792e-05, "loss": 1.2711, "step": 52502 }, { "epoch": 0.05, "learning_rate": 4.9661913658111405e-05, "loss": 1.2265, "step": 52503 }, { "epoch": 0.05, "learning_rate": 4.966190078383144e-05, "loss": 1.0425, "step": 52504 }, { "epoch": 0.05, "learning_rate": 4.9661887909308033e-05, "loss": 0.8415, "step": 52505 }, { "epoch": 0.05, "learning_rate": 4.966187503454116e-05, "loss": 0.9677, "step": 52506 }, { "epoch": 0.05, "learning_rate": 4.9661862159530845e-05, "loss": 1.5887, "step": 52507 }, { "epoch": 0.05, "learning_rate": 4.966184928427707e-05, "loss": 1.2484, "step": 52508 }, { "epoch": 0.05, "learning_rate": 4.966183640877985e-05, "loss": 1.1306, "step": 52509 }, { "epoch": 0.05, "learning_rate": 4.9661823533039185e-05, "loss": 1.4165, "step": 52510 }, { "epoch": 0.05, "learning_rate": 4.966181065705506e-05, "loss": 1.0601, "step": 52511 }, { "epoch": 0.05, "learning_rate": 4.966179778082749e-05, "loss": 1.2317, "step": 52512 }, { "epoch": 0.05, "learning_rate": 4.966178490435645e-05, "loss": 1.1551, "step": 52513 }, { "epoch": 0.05, "learning_rate": 4.966177202764198e-05, "loss": 1.0647, "step": 52514 }, { "epoch": 0.05, "learning_rate": 4.966175915068406e-05, "loss": 1.1327, "step": 52515 }, { "epoch": 0.05, "learning_rate": 4.966174627348268e-05, "loss": 1.1203, "step": 52516 }, { "epoch": 0.05, "learning_rate": 4.9661733396037854e-05, "loss": 1.514, "step": 52517 }, { "epoch": 0.05, "learning_rate": 4.966172051834957e-05, "loss": 1.4853, "step": 52518 }, { "epoch": 0.05, "learning_rate": 4.9661707640417845e-05, "loss": 1.1107, "step": 52519 }, { "epoch": 0.05, "learning_rate": 4.9661694762242675e-05, "loss": 1.1327, "step": 52520 }, { "epoch": 0.05, "learning_rate": 4.966168188382404e-05, "loss": 1.2935, "step": 52521 }, { "epoch": 0.05, "learning_rate": 4.966166900516196e-05, "loss": 1.1811, "step": 52522 }, { "epoch": 0.05, "learning_rate": 4.966165612625644e-05, "loss": 1.6893, "step": 52523 }, { "epoch": 0.05, "learning_rate": 4.966164324710746e-05, "loss": 0.9329, "step": 52524 }, { "epoch": 0.05, "learning_rate": 4.9661630367715035e-05, "loss": 1.5871, "step": 52525 }, { "epoch": 0.05, "learning_rate": 4.966161748807916e-05, "loss": 0.9263, "step": 52526 }, { "epoch": 0.05, "learning_rate": 4.9661604608199834e-05, "loss": 1.1587, "step": 52527 }, { "epoch": 0.05, "learning_rate": 4.9661591728077064e-05, "loss": 1.4116, "step": 52528 }, { "epoch": 0.05, "learning_rate": 4.9661578847710843e-05, "loss": 1.0867, "step": 52529 }, { "epoch": 0.05, "learning_rate": 4.966156596710117e-05, "loss": 1.1932, "step": 52530 }, { "epoch": 0.05, "learning_rate": 4.966155308624805e-05, "loss": 1.204, "step": 52531 }, { "epoch": 0.05, "learning_rate": 4.966154020515148e-05, "loss": 1.0563, "step": 52532 }, { "epoch": 0.05, "learning_rate": 4.9661527323811466e-05, "loss": 1.1859, "step": 52533 }, { "epoch": 0.05, "learning_rate": 4.9661514442228004e-05, "loss": 1.2327, "step": 52534 }, { "epoch": 0.05, "learning_rate": 4.966150156040109e-05, "loss": 1.2511, "step": 52535 }, { "epoch": 0.05, "learning_rate": 4.966148867833072e-05, "loss": 1.2779, "step": 52536 }, { "epoch": 0.05, "learning_rate": 4.966147579601692e-05, "loss": 1.3258, "step": 52537 }, { "epoch": 0.05, "learning_rate": 4.9661462913459665e-05, "loss": 1.392, "step": 52538 }, { "epoch": 0.05, "learning_rate": 4.966145003065896e-05, "loss": 1.0949, "step": 52539 }, { "epoch": 0.05, "learning_rate": 4.966143714761481e-05, "loss": 1.1287, "step": 52540 }, { "epoch": 0.05, "learning_rate": 4.966142426432721e-05, "loss": 1.0563, "step": 52541 }, { "epoch": 0.05, "learning_rate": 4.9661411380796166e-05, "loss": 1.1377, "step": 52542 }, { "epoch": 0.05, "learning_rate": 4.9661398497021674e-05, "loss": 1.005, "step": 52543 }, { "epoch": 0.05, "learning_rate": 4.966138561300374e-05, "loss": 1.1503, "step": 52544 }, { "epoch": 0.05, "learning_rate": 4.966137272874235e-05, "loss": 1.2459, "step": 52545 }, { "epoch": 0.05, "learning_rate": 4.966135984423752e-05, "loss": 1.1333, "step": 52546 }, { "epoch": 0.05, "learning_rate": 4.966134695948924e-05, "loss": 1.2791, "step": 52547 }, { "epoch": 0.05, "learning_rate": 4.9661334074497515e-05, "loss": 1.0147, "step": 52548 }, { "epoch": 0.05, "learning_rate": 4.966132118926234e-05, "loss": 1.5031, "step": 52549 }, { "epoch": 0.05, "learning_rate": 4.966130830378372e-05, "loss": 1.466, "step": 52550 }, { "epoch": 0.05, "learning_rate": 4.966129541806166e-05, "loss": 1.419, "step": 52551 }, { "epoch": 0.05, "learning_rate": 4.9661282532096146e-05, "loss": 1.1791, "step": 52552 }, { "epoch": 0.05, "learning_rate": 4.9661269645887186e-05, "loss": 1.1603, "step": 52553 }, { "epoch": 0.05, "learning_rate": 4.966125675943479e-05, "loss": 1.0217, "step": 52554 }, { "epoch": 0.05, "learning_rate": 4.9661243872738936e-05, "loss": 1.2152, "step": 52555 }, { "epoch": 0.05, "learning_rate": 4.966123098579965e-05, "loss": 1.1455, "step": 52556 }, { "epoch": 0.05, "learning_rate": 4.9661218098616916e-05, "loss": 1.6097, "step": 52557 }, { "epoch": 0.05, "learning_rate": 4.966120521119073e-05, "loss": 0.9347, "step": 52558 }, { "epoch": 0.05, "learning_rate": 4.96611923235211e-05, "loss": 1.4738, "step": 52559 }, { "epoch": 0.05, "learning_rate": 4.966117943560803e-05, "loss": 1.1777, "step": 52560 }, { "epoch": 0.05, "learning_rate": 4.966116654745151e-05, "loss": 0.9212, "step": 52561 }, { "epoch": 0.05, "learning_rate": 4.966115365905155e-05, "loss": 1.3055, "step": 52562 }, { "epoch": 0.05, "learning_rate": 4.9661140770408145e-05, "loss": 1.0905, "step": 52563 }, { "epoch": 0.05, "learning_rate": 4.966112788152129e-05, "loss": 1.1611, "step": 52564 }, { "epoch": 0.05, "learning_rate": 4.9661114992390994e-05, "loss": 1.2452, "step": 52565 }, { "epoch": 0.05, "learning_rate": 4.9661102103017255e-05, "loss": 0.666, "step": 52566 }, { "epoch": 0.05, "learning_rate": 4.966108921340007e-05, "loss": 0.9874, "step": 52567 }, { "epoch": 0.05, "learning_rate": 4.9661076323539446e-05, "loss": 1.446, "step": 52568 }, { "epoch": 0.05, "learning_rate": 4.9661063433435375e-05, "loss": 1.0978, "step": 52569 }, { "epoch": 0.05, "learning_rate": 4.9661050543087854e-05, "loss": 1.0455, "step": 52570 }, { "epoch": 0.05, "learning_rate": 4.96610376524969e-05, "loss": 1.4322, "step": 52571 }, { "epoch": 0.05, "learning_rate": 4.96610247616625e-05, "loss": 0.9937, "step": 52572 }, { "epoch": 0.05, "learning_rate": 4.966101187058465e-05, "loss": 1.1447, "step": 52573 }, { "epoch": 0.05, "learning_rate": 4.966099897926336e-05, "loss": 0.8781, "step": 52574 }, { "epoch": 0.05, "learning_rate": 4.966098608769863e-05, "loss": 0.9307, "step": 52575 }, { "epoch": 0.05, "learning_rate": 4.966097319589046e-05, "loss": 0.9124, "step": 52576 }, { "epoch": 0.05, "learning_rate": 4.9660960303838845e-05, "loss": 0.9879, "step": 52577 }, { "epoch": 0.05, "learning_rate": 4.966094741154378e-05, "loss": 1.1222, "step": 52578 }, { "epoch": 0.05, "learning_rate": 4.966093451900527e-05, "loss": 1.0568, "step": 52579 }, { "epoch": 0.05, "learning_rate": 4.966092162622333e-05, "loss": 1.4597, "step": 52580 }, { "epoch": 0.05, "learning_rate": 4.966090873319795e-05, "loss": 1.7605, "step": 52581 }, { "epoch": 0.05, "learning_rate": 4.966089583992912e-05, "loss": 0.9986, "step": 52582 }, { "epoch": 0.05, "learning_rate": 4.9660882946416856e-05, "loss": 0.7788, "step": 52583 }, { "epoch": 0.05, "learning_rate": 4.966087005266113e-05, "loss": 0.6569, "step": 52584 }, { "epoch": 0.05, "learning_rate": 4.966085715866198e-05, "loss": 0.9821, "step": 52585 }, { "epoch": 0.05, "learning_rate": 4.9660844264419384e-05, "loss": 1.4081, "step": 52586 }, { "epoch": 0.05, "learning_rate": 4.966083136993335e-05, "loss": 1.0973, "step": 52587 }, { "epoch": 0.05, "learning_rate": 4.9660818475203874e-05, "loss": 1.0675, "step": 52588 }, { "epoch": 0.05, "learning_rate": 4.966080558023095e-05, "loss": 0.9894, "step": 52589 }, { "epoch": 0.05, "learning_rate": 4.966079268501459e-05, "loss": 1.329, "step": 52590 }, { "epoch": 0.05, "learning_rate": 4.966077978955479e-05, "loss": 1.5998, "step": 52591 }, { "epoch": 0.05, "learning_rate": 4.9660766893851546e-05, "loss": 1.1335, "step": 52592 }, { "epoch": 0.05, "learning_rate": 4.9660753997904865e-05, "loss": 0.9855, "step": 52593 }, { "epoch": 0.05, "learning_rate": 4.9660741101714734e-05, "loss": 1.2618, "step": 52594 }, { "epoch": 0.05, "learning_rate": 4.966072820528118e-05, "loss": 0.9893, "step": 52595 }, { "epoch": 0.05, "learning_rate": 4.9660715308604174e-05, "loss": 0.9944, "step": 52596 }, { "epoch": 0.05, "learning_rate": 4.9660702411683724e-05, "loss": 1.394, "step": 52597 }, { "epoch": 0.05, "learning_rate": 4.9660689514519844e-05, "loss": 1.6166, "step": 52598 }, { "epoch": 0.05, "learning_rate": 4.966067661711252e-05, "loss": 1.2384, "step": 52599 }, { "epoch": 0.05, "learning_rate": 4.966066371946175e-05, "loss": 1.3295, "step": 52600 }, { "epoch": 0.05, "learning_rate": 4.9660650821567546e-05, "loss": 1.2096, "step": 52601 }, { "epoch": 0.05, "learning_rate": 4.96606379234299e-05, "loss": 1.2938, "step": 52602 }, { "epoch": 0.05, "learning_rate": 4.966062502504882e-05, "loss": 1.0496, "step": 52603 }, { "epoch": 0.05, "learning_rate": 4.96606121264243e-05, "loss": 1.2039, "step": 52604 }, { "epoch": 0.05, "learning_rate": 4.966059922755634e-05, "loss": 1.1637, "step": 52605 }, { "epoch": 0.05, "learning_rate": 4.966058632844494e-05, "loss": 1.0569, "step": 52606 }, { "epoch": 0.05, "learning_rate": 4.9660573429090096e-05, "loss": 1.1787, "step": 52607 }, { "epoch": 0.05, "learning_rate": 4.9660560529491816e-05, "loss": 0.8765, "step": 52608 }, { "epoch": 0.05, "learning_rate": 4.96605476296501e-05, "loss": 1.0382, "step": 52609 }, { "epoch": 0.05, "learning_rate": 4.9660534729564944e-05, "loss": 1.0548, "step": 52610 }, { "epoch": 0.05, "learning_rate": 4.966052182923635e-05, "loss": 1.0889, "step": 52611 }, { "epoch": 0.05, "learning_rate": 4.9660508928664316e-05, "loss": 1.5536, "step": 52612 }, { "epoch": 0.05, "learning_rate": 4.966049602784884e-05, "loss": 1.0333, "step": 52613 }, { "epoch": 0.05, "learning_rate": 4.966048312678994e-05, "loss": 1.0421, "step": 52614 }, { "epoch": 0.05, "learning_rate": 4.966047022548759e-05, "loss": 1.147, "step": 52615 }, { "epoch": 0.05, "learning_rate": 4.9660457323941814e-05, "loss": 1.0163, "step": 52616 }, { "epoch": 0.05, "learning_rate": 4.9660444422152585e-05, "loss": 0.9954, "step": 52617 }, { "epoch": 0.05, "learning_rate": 4.9660431520119925e-05, "loss": 1.123, "step": 52618 }, { "epoch": 0.05, "learning_rate": 4.966041861784383e-05, "loss": 1.3002, "step": 52619 }, { "epoch": 0.05, "learning_rate": 4.96604057153243e-05, "loss": 0.9984, "step": 52620 }, { "epoch": 0.05, "learning_rate": 4.966039281256133e-05, "loss": 1.3207, "step": 52621 }, { "epoch": 0.05, "learning_rate": 4.966037990955492e-05, "loss": 1.0732, "step": 52622 }, { "epoch": 0.05, "learning_rate": 4.9660367006305076e-05, "loss": 1.0862, "step": 52623 }, { "epoch": 0.05, "learning_rate": 4.966035410281179e-05, "loss": 1.1369, "step": 52624 }, { "epoch": 0.05, "learning_rate": 4.966034119907507e-05, "loss": 1.072, "step": 52625 }, { "epoch": 0.05, "learning_rate": 4.9660328295094914e-05, "loss": 1.105, "step": 52626 }, { "epoch": 0.05, "learning_rate": 4.9660315390871326e-05, "loss": 1.3005, "step": 52627 }, { "epoch": 0.05, "learning_rate": 4.96603024864043e-05, "loss": 1.1881, "step": 52628 }, { "epoch": 0.05, "learning_rate": 4.966028958169384e-05, "loss": 1.3806, "step": 52629 }, { "epoch": 0.05, "learning_rate": 4.966027667673994e-05, "loss": 1.1643, "step": 52630 }, { "epoch": 0.05, "learning_rate": 4.96602637715426e-05, "loss": 1.1513, "step": 52631 }, { "epoch": 0.05, "learning_rate": 4.9660250866101834e-05, "loss": 1.2069, "step": 52632 }, { "epoch": 0.05, "learning_rate": 4.966023796041762e-05, "loss": 1.1676, "step": 52633 }, { "epoch": 0.05, "learning_rate": 4.9660225054489986e-05, "loss": 1.5451, "step": 52634 }, { "epoch": 0.05, "learning_rate": 4.96602121483189e-05, "loss": 1.2797, "step": 52635 }, { "epoch": 0.05, "learning_rate": 4.966019924190439e-05, "loss": 1.3519, "step": 52636 }, { "epoch": 0.05, "learning_rate": 4.9660186335246436e-05, "loss": 1.9977, "step": 52637 }, { "epoch": 0.05, "learning_rate": 4.966017342834506e-05, "loss": 2.0619, "step": 52638 }, { "epoch": 0.05, "learning_rate": 4.9660160521200236e-05, "loss": 1.203, "step": 52639 }, { "epoch": 0.05, "learning_rate": 4.966014761381199e-05, "loss": 1.2518, "step": 52640 }, { "epoch": 0.05, "learning_rate": 4.96601347061803e-05, "loss": 1.1858, "step": 52641 }, { "epoch": 0.05, "learning_rate": 4.966012179830518e-05, "loss": 1.3036, "step": 52642 }, { "epoch": 0.05, "learning_rate": 4.9660108890186624e-05, "loss": 0.7339, "step": 52643 }, { "epoch": 0.05, "learning_rate": 4.9660095981824636e-05, "loss": 1.1493, "step": 52644 }, { "epoch": 0.05, "learning_rate": 4.966008307321921e-05, "loss": 1.196, "step": 52645 }, { "epoch": 0.05, "learning_rate": 4.9660070164370356e-05, "loss": 1.2627, "step": 52646 }, { "epoch": 0.05, "learning_rate": 4.9660057255278056e-05, "loss": 1.0744, "step": 52647 }, { "epoch": 0.05, "learning_rate": 4.966004434594233e-05, "loss": 1.2702, "step": 52648 }, { "epoch": 0.05, "learning_rate": 4.966003143636317e-05, "loss": 1.407, "step": 52649 }, { "epoch": 0.05, "learning_rate": 4.966001852654059e-05, "loss": 0.8651, "step": 52650 }, { "epoch": 0.05, "learning_rate": 4.966000561647456e-05, "loss": 0.8537, "step": 52651 }, { "epoch": 0.05, "learning_rate": 4.96599927061651e-05, "loss": 0.8522, "step": 52652 }, { "epoch": 0.05, "learning_rate": 4.9659979795612204e-05, "loss": 1.299, "step": 52653 }, { "epoch": 0.05, "learning_rate": 4.965996688481588e-05, "loss": 1.958, "step": 52654 }, { "epoch": 0.05, "learning_rate": 4.965995397377612e-05, "loss": 1.7712, "step": 52655 }, { "epoch": 0.05, "learning_rate": 4.965994106249293e-05, "loss": 1.4916, "step": 52656 }, { "epoch": 0.05, "learning_rate": 4.965992815096631e-05, "loss": 1.4539, "step": 52657 }, { "epoch": 0.05, "learning_rate": 4.965991523919625e-05, "loss": 0.5085, "step": 52658 }, { "epoch": 0.05, "learning_rate": 4.965990232718277e-05, "loss": 1.252, "step": 52659 }, { "epoch": 0.05, "learning_rate": 4.9659889414925844e-05, "loss": 0.716, "step": 52660 }, { "epoch": 0.05, "learning_rate": 4.96598765024255e-05, "loss": 0.883, "step": 52661 }, { "epoch": 0.05, "learning_rate": 4.965986358968171e-05, "loss": 0.8661, "step": 52662 }, { "epoch": 0.05, "learning_rate": 4.96598506766945e-05, "loss": 0.6722, "step": 52663 }, { "epoch": 0.05, "learning_rate": 4.9659837763463846e-05, "loss": 0.6252, "step": 52664 }, { "epoch": 0.05, "learning_rate": 4.9659824849989764e-05, "loss": 0.5045, "step": 52665 }, { "epoch": 0.05, "learning_rate": 4.965981193627226e-05, "loss": 0.7305, "step": 52666 }, { "epoch": 0.05, "learning_rate": 4.9659799022311314e-05, "loss": 1.3441, "step": 52667 }, { "epoch": 0.05, "learning_rate": 4.965978610810694e-05, "loss": 0.9752, "step": 52668 }, { "epoch": 0.05, "learning_rate": 4.965977319365914e-05, "loss": 1.1049, "step": 52669 }, { "epoch": 0.05, "learning_rate": 4.96597602789679e-05, "loss": 0.8638, "step": 52670 }, { "epoch": 0.05, "learning_rate": 4.9659747364033236e-05, "loss": 0.9816, "step": 52671 }, { "epoch": 0.05, "learning_rate": 4.965973444885515e-05, "loss": 1.0423, "step": 52672 }, { "epoch": 0.05, "learning_rate": 4.965972153343362e-05, "loss": 1.2044, "step": 52673 }, { "epoch": 0.05, "learning_rate": 4.9659708617768664e-05, "loss": 1.2894, "step": 52674 }, { "epoch": 0.05, "learning_rate": 4.965969570186028e-05, "loss": 1.1773, "step": 52675 }, { "epoch": 0.05, "learning_rate": 4.965968278570846e-05, "loss": 1.0832, "step": 52676 }, { "epoch": 0.05, "learning_rate": 4.965966986931321e-05, "loss": 1.2172, "step": 52677 }, { "epoch": 0.05, "learning_rate": 4.9659656952674525e-05, "loss": 0.9128, "step": 52678 }, { "epoch": 0.05, "learning_rate": 4.965964403579242e-05, "loss": 0.8411, "step": 52679 }, { "epoch": 0.05, "learning_rate": 4.965963111866689e-05, "loss": 0.9171, "step": 52680 }, { "epoch": 0.05, "learning_rate": 4.965961820129792e-05, "loss": 1.1875, "step": 52681 }, { "epoch": 0.05, "learning_rate": 4.9659605283685525e-05, "loss": 1.0353, "step": 52682 }, { "epoch": 0.05, "learning_rate": 4.96595923658297e-05, "loss": 0.9921, "step": 52683 }, { "epoch": 0.05, "learning_rate": 4.9659579447730446e-05, "loss": 1.1975, "step": 52684 }, { "epoch": 0.05, "learning_rate": 4.965956652938777e-05, "loss": 1.4613, "step": 52685 }, { "epoch": 0.05, "learning_rate": 4.965955361080166e-05, "loss": 1.5579, "step": 52686 }, { "epoch": 0.05, "learning_rate": 4.965954069197212e-05, "loss": 0.905, "step": 52687 }, { "epoch": 0.05, "learning_rate": 4.965952777289915e-05, "loss": 1.1752, "step": 52688 }, { "epoch": 0.05, "learning_rate": 4.965951485358275e-05, "loss": 1.0645, "step": 52689 }, { "epoch": 0.05, "learning_rate": 4.965950193402292e-05, "loss": 1.0529, "step": 52690 }, { "epoch": 0.05, "learning_rate": 4.965948901421967e-05, "loss": 1.316, "step": 52691 }, { "epoch": 0.05, "learning_rate": 4.965947609417298e-05, "loss": 1.0407, "step": 52692 }, { "epoch": 0.05, "learning_rate": 4.9659463173882874e-05, "loss": 0.9792, "step": 52693 }, { "epoch": 0.05, "learning_rate": 4.965945025334934e-05, "loss": 1.0193, "step": 52694 }, { "epoch": 0.05, "learning_rate": 4.965943733257237e-05, "loss": 1.3149, "step": 52695 }, { "epoch": 0.05, "learning_rate": 4.965942441155198e-05, "loss": 1.1756, "step": 52696 }, { "epoch": 0.05, "learning_rate": 4.965941149028816e-05, "loss": 0.9197, "step": 52697 }, { "epoch": 0.05, "learning_rate": 4.965939856878091e-05, "loss": 1.2866, "step": 52698 }, { "epoch": 0.05, "learning_rate": 4.965938564703024e-05, "loss": 1.4422, "step": 52699 }, { "epoch": 0.05, "learning_rate": 4.965937272503614e-05, "loss": 1.1593, "step": 52700 }, { "epoch": 0.05, "learning_rate": 4.9659359802798605e-05, "loss": 1.2471, "step": 52701 }, { "epoch": 0.05, "learning_rate": 4.965934688031765e-05, "loss": 0.8797, "step": 52702 }, { "epoch": 0.05, "learning_rate": 4.965933395759327e-05, "loss": 1.2007, "step": 52703 }, { "epoch": 0.05, "learning_rate": 4.965932103462546e-05, "loss": 1.1505, "step": 52704 }, { "epoch": 0.05, "learning_rate": 4.9659308111414224e-05, "loss": 0.6863, "step": 52705 }, { "epoch": 0.05, "learning_rate": 4.9659295187959566e-05, "loss": 0.7197, "step": 52706 }, { "epoch": 0.05, "learning_rate": 4.965928226426148e-05, "loss": 1.0609, "step": 52707 }, { "epoch": 0.05, "learning_rate": 4.965926934031996e-05, "loss": 1.2162, "step": 52708 }, { "epoch": 0.05, "learning_rate": 4.965925641613502e-05, "loss": 1.1317, "step": 52709 }, { "epoch": 0.05, "learning_rate": 4.9659243491706656e-05, "loss": 1.1046, "step": 52710 }, { "epoch": 0.05, "learning_rate": 4.9659230567034866e-05, "loss": 1.0278, "step": 52711 }, { "epoch": 0.05, "learning_rate": 4.965921764211965e-05, "loss": 1.0267, "step": 52712 }, { "epoch": 0.05, "learning_rate": 4.965920471696101e-05, "loss": 0.9934, "step": 52713 }, { "epoch": 0.05, "learning_rate": 4.965919179155893e-05, "loss": 0.9159, "step": 52714 }, { "epoch": 0.05, "learning_rate": 4.965917886591345e-05, "loss": 0.9303, "step": 52715 }, { "epoch": 0.05, "learning_rate": 4.9659165940024525e-05, "loss": 0.9199, "step": 52716 }, { "epoch": 0.05, "learning_rate": 4.9659153013892186e-05, "loss": 0.8915, "step": 52717 }, { "epoch": 0.05, "learning_rate": 4.965914008751642e-05, "loss": 0.8685, "step": 52718 }, { "epoch": 0.05, "learning_rate": 4.965912716089722e-05, "loss": 0.8628, "step": 52719 }, { "epoch": 0.05, "learning_rate": 4.965911423403461e-05, "loss": 0.8268, "step": 52720 }, { "epoch": 0.05, "learning_rate": 4.965910130692857e-05, "loss": 0.8483, "step": 52721 }, { "epoch": 0.05, "learning_rate": 4.96590883795791e-05, "loss": 0.8217, "step": 52722 }, { "epoch": 0.05, "learning_rate": 4.965907545198621e-05, "loss": 0.8151, "step": 52723 }, { "epoch": 0.05, "learning_rate": 4.96590625241499e-05, "loss": 0.809, "step": 52724 }, { "epoch": 0.05, "learning_rate": 4.9659049596070164e-05, "loss": 0.7924, "step": 52725 }, { "epoch": 0.05, "learning_rate": 4.9659036667747e-05, "loss": 0.8102, "step": 52726 }, { "epoch": 0.05, "learning_rate": 4.9659023739180415e-05, "loss": 0.7898, "step": 52727 }, { "epoch": 0.05, "learning_rate": 4.965901081037041e-05, "loss": 0.8023, "step": 52728 }, { "epoch": 0.05, "learning_rate": 4.965899788131698e-05, "loss": 0.8087, "step": 52729 }, { "epoch": 0.05, "learning_rate": 4.965898495202013e-05, "loss": 0.8019, "step": 52730 }, { "epoch": 0.05, "learning_rate": 4.9658972022479845e-05, "loss": 0.8129, "step": 52731 }, { "epoch": 0.05, "learning_rate": 4.965895909269614e-05, "loss": 0.7811, "step": 52732 }, { "epoch": 0.05, "learning_rate": 4.965894616266902e-05, "loss": 0.767, "step": 52733 }, { "epoch": 0.05, "learning_rate": 4.9658933232398475e-05, "loss": 0.7878, "step": 52734 }, { "epoch": 0.05, "learning_rate": 4.96589203018845e-05, "loss": 0.7556, "step": 52735 }, { "epoch": 0.05, "learning_rate": 4.965890737112712e-05, "loss": 0.7739, "step": 52736 }, { "epoch": 0.05, "learning_rate": 4.9658894440126305e-05, "loss": 0.7014, "step": 52737 }, { "epoch": 0.05, "learning_rate": 4.965888150888207e-05, "loss": 0.7125, "step": 52738 }, { "epoch": 0.05, "learning_rate": 4.965886857739441e-05, "loss": 0.7595, "step": 52739 }, { "epoch": 0.05, "learning_rate": 4.965885564566333e-05, "loss": 0.7377, "step": 52740 }, { "epoch": 0.05, "learning_rate": 4.965884271368883e-05, "loss": 0.8228, "step": 52741 }, { "epoch": 0.05, "learning_rate": 4.96588297814709e-05, "loss": 0.7856, "step": 52742 }, { "epoch": 0.05, "learning_rate": 4.965881684900956e-05, "loss": 0.7648, "step": 52743 }, { "epoch": 0.05, "learning_rate": 4.965880391630479e-05, "loss": 0.7631, "step": 52744 }, { "epoch": 0.05, "learning_rate": 4.9658790983356604e-05, "loss": 0.761, "step": 52745 }, { "epoch": 0.05, "learning_rate": 4.965877805016499e-05, "loss": 0.7564, "step": 52746 }, { "epoch": 0.05, "learning_rate": 4.965876511672997e-05, "loss": 0.7718, "step": 52747 }, { "epoch": 0.05, "learning_rate": 4.965875218305151e-05, "loss": 0.7897, "step": 52748 }, { "epoch": 0.05, "learning_rate": 4.9658739249129635e-05, "loss": 0.7407, "step": 52749 }, { "epoch": 0.05, "learning_rate": 4.965872631496434e-05, "loss": 0.7603, "step": 52750 }, { "epoch": 0.05, "learning_rate": 4.9658713380555635e-05, "loss": 0.7801, "step": 52751 }, { "epoch": 0.05, "learning_rate": 4.96587004459035e-05, "loss": 0.7447, "step": 52752 }, { "epoch": 0.05, "learning_rate": 4.9658687511007954e-05, "loss": 0.7769, "step": 52753 }, { "epoch": 0.05, "learning_rate": 4.965867457586897e-05, "loss": 0.743, "step": 52754 }, { "epoch": 0.05, "learning_rate": 4.965866164048658e-05, "loss": 0.7226, "step": 52755 }, { "epoch": 0.05, "learning_rate": 4.965864870486077e-05, "loss": 0.7443, "step": 52756 }, { "epoch": 0.05, "learning_rate": 4.9658635768991534e-05, "loss": 0.7516, "step": 52757 }, { "epoch": 0.05, "learning_rate": 4.965862283287888e-05, "loss": 0.7556, "step": 52758 }, { "epoch": 0.05, "learning_rate": 4.965860989652281e-05, "loss": 0.7302, "step": 52759 }, { "epoch": 0.05, "learning_rate": 4.965859695992331e-05, "loss": 0.6991, "step": 52760 }, { "epoch": 0.05, "learning_rate": 4.96585840230804e-05, "loss": 0.7649, "step": 52761 }, { "epoch": 0.05, "learning_rate": 4.965857108599407e-05, "loss": 0.757, "step": 52762 }, { "epoch": 0.05, "learning_rate": 4.965855814866432e-05, "loss": 0.7144, "step": 52763 }, { "epoch": 0.05, "learning_rate": 4.965854521109115e-05, "loss": 0.7103, "step": 52764 }, { "epoch": 0.05, "learning_rate": 4.965853227327456e-05, "loss": 0.7295, "step": 52765 }, { "epoch": 0.05, "learning_rate": 4.965851933521456e-05, "loss": 0.7439, "step": 52766 }, { "epoch": 0.05, "learning_rate": 4.965850639691113e-05, "loss": 0.7395, "step": 52767 }, { "epoch": 0.05, "learning_rate": 4.965849345836429e-05, "loss": 0.6893, "step": 52768 }, { "epoch": 0.05, "learning_rate": 4.965848051957403e-05, "loss": 0.8337, "step": 52769 }, { "epoch": 0.05, "learning_rate": 4.9658467580540345e-05, "loss": 0.9888, "step": 52770 }, { "epoch": 0.05, "learning_rate": 4.9658454641263254e-05, "loss": 1.3617, "step": 52771 }, { "epoch": 0.05, "learning_rate": 4.965844170174274e-05, "loss": 1.2492, "step": 52772 }, { "epoch": 0.05, "learning_rate": 4.965842876197881e-05, "loss": 0.6104, "step": 52773 }, { "epoch": 0.05, "learning_rate": 4.965841582197145e-05, "loss": 1.255, "step": 52774 }, { "epoch": 0.05, "learning_rate": 4.965840288172069e-05, "loss": 1.2801, "step": 52775 }, { "epoch": 0.05, "learning_rate": 4.96583899412265e-05, "loss": 0.7179, "step": 52776 }, { "epoch": 0.05, "learning_rate": 4.9658377000488896e-05, "loss": 1.1391, "step": 52777 }, { "epoch": 0.05, "learning_rate": 4.965836405950788e-05, "loss": 0.8283, "step": 52778 }, { "epoch": 0.05, "learning_rate": 4.965835111828344e-05, "loss": 1.4846, "step": 52779 }, { "epoch": 0.05, "learning_rate": 4.965833817681559e-05, "loss": 1.0396, "step": 52780 }, { "epoch": 0.05, "learning_rate": 4.965832523510432e-05, "loss": 1.1025, "step": 52781 }, { "epoch": 0.05, "learning_rate": 4.965831229314963e-05, "loss": 1.3216, "step": 52782 }, { "epoch": 0.05, "learning_rate": 4.965829935095152e-05, "loss": 1.7823, "step": 52783 }, { "epoch": 0.05, "learning_rate": 4.9658286408510005e-05, "loss": 1.147, "step": 52784 }, { "epoch": 0.05, "learning_rate": 4.965827346582507e-05, "loss": 1.1268, "step": 52785 }, { "epoch": 0.05, "learning_rate": 4.965826052289672e-05, "loss": 1.1589, "step": 52786 }, { "epoch": 0.05, "learning_rate": 4.965824757972495e-05, "loss": 1.1255, "step": 52787 }, { "epoch": 0.05, "learning_rate": 4.965823463630976e-05, "loss": 1.4913, "step": 52788 }, { "epoch": 0.05, "learning_rate": 4.9658221692651166e-05, "loss": 1.1126, "step": 52789 }, { "epoch": 0.05, "learning_rate": 4.965820874874915e-05, "loss": 0.9338, "step": 52790 }, { "epoch": 0.05, "learning_rate": 4.965819580460372e-05, "loss": 0.9597, "step": 52791 }, { "epoch": 0.05, "learning_rate": 4.9658182860214876e-05, "loss": 1.5599, "step": 52792 }, { "epoch": 0.05, "learning_rate": 4.965816991558261e-05, "loss": 1.1436, "step": 52793 }, { "epoch": 0.05, "learning_rate": 4.965815697070694e-05, "loss": 1.1504, "step": 52794 }, { "epoch": 0.05, "learning_rate": 4.965814402558785e-05, "loss": 1.1681, "step": 52795 }, { "epoch": 0.05, "learning_rate": 4.9658131080225346e-05, "loss": 1.3185, "step": 52796 }, { "epoch": 0.05, "learning_rate": 4.965811813461942e-05, "loss": 2.0054, "step": 52797 }, { "epoch": 0.05, "learning_rate": 4.9658105188770086e-05, "loss": 1.2655, "step": 52798 }, { "epoch": 0.05, "learning_rate": 4.9658092242677334e-05, "loss": 1.0946, "step": 52799 }, { "epoch": 0.05, "learning_rate": 4.965807929634117e-05, "loss": 1.1241, "step": 52800 }, { "epoch": 0.05, "learning_rate": 4.965806634976159e-05, "loss": 0.7005, "step": 52801 }, { "epoch": 0.05, "learning_rate": 4.9658053402938604e-05, "loss": 0.4864, "step": 52802 }, { "epoch": 0.05, "learning_rate": 4.96580404558722e-05, "loss": 0.3227, "step": 52803 }, { "epoch": 0.05, "learning_rate": 4.9658027508562375e-05, "loss": 0.3733, "step": 52804 }, { "epoch": 0.05, "learning_rate": 4.965801456100915e-05, "loss": 0.695, "step": 52805 }, { "epoch": 0.05, "learning_rate": 4.96580016132125e-05, "loss": 0.6983, "step": 52806 }, { "epoch": 0.05, "learning_rate": 4.965798866517244e-05, "loss": 0.6251, "step": 52807 }, { "epoch": 0.05, "learning_rate": 4.965797571688896e-05, "loss": 0.42, "step": 52808 }, { "epoch": 0.05, "learning_rate": 4.965796276836208e-05, "loss": 0.2931, "step": 52809 }, { "epoch": 0.05, "learning_rate": 4.9657949819591775e-05, "loss": 0.6407, "step": 52810 }, { "epoch": 0.05, "learning_rate": 4.965793687057807e-05, "loss": 0.7012, "step": 52811 }, { "epoch": 0.05, "learning_rate": 4.9657923921320936e-05, "loss": 0.5186, "step": 52812 }, { "epoch": 0.05, "learning_rate": 4.96579109718204e-05, "loss": 0.4672, "step": 52813 }, { "epoch": 0.05, "learning_rate": 4.965789802207645e-05, "loss": 0.3636, "step": 52814 }, { "epoch": 0.05, "learning_rate": 4.965788507208908e-05, "loss": 3.4605, "step": 52815 }, { "epoch": 0.05, "learning_rate": 4.9657872121858315e-05, "loss": 4.5698, "step": 52816 }, { "epoch": 0.05, "learning_rate": 4.9657859171384126e-05, "loss": 0.4874, "step": 52817 }, { "epoch": 0.05, "learning_rate": 4.9657846220666526e-05, "loss": 0.7034, "step": 52818 }, { "epoch": 0.05, "learning_rate": 4.965783326970551e-05, "loss": 0.5194, "step": 52819 }, { "epoch": 0.05, "learning_rate": 4.965782031850109e-05, "loss": 0.3614, "step": 52820 }, { "epoch": 0.05, "learning_rate": 4.9657807367053254e-05, "loss": 0.3457, "step": 52821 }, { "epoch": 0.05, "learning_rate": 4.9657794415362014e-05, "loss": 0.3036, "step": 52822 }, { "epoch": 0.05, "learning_rate": 4.965778146342736e-05, "loss": 0.1829, "step": 52823 }, { "epoch": 0.05, "learning_rate": 4.965776851124928e-05, "loss": 0.1619, "step": 52824 }, { "epoch": 0.05, "learning_rate": 4.96577555588278e-05, "loss": 0.4828, "step": 52825 }, { "epoch": 0.05, "learning_rate": 4.965774260616292e-05, "loss": 0.5881, "step": 52826 }, { "epoch": 0.05, "learning_rate": 4.9657729653254614e-05, "loss": 0.4118, "step": 52827 }, { "epoch": 0.05, "learning_rate": 4.965771670010289e-05, "loss": 0.414, "step": 52828 }, { "epoch": 0.05, "learning_rate": 4.9657703746707775e-05, "loss": 0.6691, "step": 52829 }, { "epoch": 0.05, "learning_rate": 4.965769079306925e-05, "loss": 0.3277, "step": 52830 }, { "epoch": 0.05, "learning_rate": 4.96576778391873e-05, "loss": 0.4397, "step": 52831 }, { "epoch": 0.05, "learning_rate": 4.965766488506195e-05, "loss": 0.4395, "step": 52832 }, { "epoch": 0.05, "learning_rate": 4.965765193069318e-05, "loss": 0.8959, "step": 52833 }, { "epoch": 0.05, "learning_rate": 4.965763897608101e-05, "loss": 1.3052, "step": 52834 }, { "epoch": 0.05, "learning_rate": 4.965762602122542e-05, "loss": 1.263, "step": 52835 }, { "epoch": 0.05, "learning_rate": 4.965761306612643e-05, "loss": 1.2045, "step": 52836 }, { "epoch": 0.05, "learning_rate": 4.965760011078403e-05, "loss": 1.2566, "step": 52837 }, { "epoch": 0.05, "learning_rate": 4.9657587155198216e-05, "loss": 1.3237, "step": 52838 }, { "epoch": 0.05, "learning_rate": 4.9657574199369e-05, "loss": 1.4148, "step": 52839 }, { "epoch": 0.05, "learning_rate": 4.965756124329636e-05, "loss": 1.0998, "step": 52840 }, { "epoch": 0.05, "learning_rate": 4.965754828698033e-05, "loss": 1.0577, "step": 52841 }, { "epoch": 0.05, "learning_rate": 4.965753533042087e-05, "loss": 0.9079, "step": 52842 }, { "epoch": 0.05, "learning_rate": 4.965752237361802e-05, "loss": 0.8772, "step": 52843 }, { "epoch": 0.05, "learning_rate": 4.9657509416571754e-05, "loss": 1.014, "step": 52844 }, { "epoch": 0.05, "learning_rate": 4.965749645928208e-05, "loss": 0.9325, "step": 52845 }, { "epoch": 0.05, "learning_rate": 4.965748350174899e-05, "loss": 1.0654, "step": 52846 }, { "epoch": 0.05, "learning_rate": 4.96574705439725e-05, "loss": 0.915, "step": 52847 }, { "epoch": 0.05, "learning_rate": 4.9657457585952605e-05, "loss": 1.136, "step": 52848 }, { "epoch": 0.05, "learning_rate": 4.965744462768929e-05, "loss": 1.4024, "step": 52849 }, { "epoch": 0.05, "learning_rate": 4.965743166918257e-05, "loss": 1.684, "step": 52850 }, { "epoch": 0.05, "learning_rate": 4.965741871043246e-05, "loss": 1.2459, "step": 52851 }, { "epoch": 0.05, "learning_rate": 4.965740575143893e-05, "loss": 1.0749, "step": 52852 }, { "epoch": 0.05, "learning_rate": 4.9657392792201985e-05, "loss": 0.7163, "step": 52853 }, { "epoch": 0.05, "learning_rate": 4.965737983272164e-05, "loss": 1.5596, "step": 52854 }, { "epoch": 0.05, "learning_rate": 4.9657366872997885e-05, "loss": 1.5941, "step": 52855 }, { "epoch": 0.05, "learning_rate": 4.965735391303073e-05, "loss": 1.261, "step": 52856 }, { "epoch": 0.05, "learning_rate": 4.965734095282016e-05, "loss": 1.5457, "step": 52857 }, { "epoch": 0.05, "learning_rate": 4.9657327992366186e-05, "loss": 1.2975, "step": 52858 }, { "epoch": 0.05, "learning_rate": 4.9657315031668804e-05, "loss": 0.3285, "step": 52859 }, { "epoch": 0.05, "learning_rate": 4.965730207072802e-05, "loss": 0.7747, "step": 52860 }, { "epoch": 0.05, "learning_rate": 4.965728910954382e-05, "loss": 0.967, "step": 52861 }, { "epoch": 0.05, "learning_rate": 4.965727614811622e-05, "loss": 0.8019, "step": 52862 }, { "epoch": 0.05, "learning_rate": 4.9657263186445215e-05, "loss": 1.2223, "step": 52863 }, { "epoch": 0.05, "learning_rate": 4.96572502245308e-05, "loss": 1.1725, "step": 52864 }, { "epoch": 0.05, "learning_rate": 4.965723726237298e-05, "loss": 1.2557, "step": 52865 }, { "epoch": 0.05, "learning_rate": 4.965722429997176e-05, "loss": 1.2069, "step": 52866 }, { "epoch": 0.05, "learning_rate": 4.9657211337327124e-05, "loss": 1.702, "step": 52867 }, { "epoch": 0.05, "learning_rate": 4.965719837443909e-05, "loss": 1.3388, "step": 52868 }, { "epoch": 0.05, "learning_rate": 4.965718541130765e-05, "loss": 0.8551, "step": 52869 }, { "epoch": 0.05, "learning_rate": 4.96571724479328e-05, "loss": 0.9545, "step": 52870 }, { "epoch": 0.05, "learning_rate": 4.965715948431455e-05, "loss": 0.5908, "step": 52871 }, { "epoch": 0.05, "learning_rate": 4.965714652045289e-05, "loss": 0.9513, "step": 52872 }, { "epoch": 0.05, "learning_rate": 4.9657133556347834e-05, "loss": 1.2006, "step": 52873 }, { "epoch": 0.05, "learning_rate": 4.965712059199936e-05, "loss": 1.3999, "step": 52874 }, { "epoch": 0.05, "learning_rate": 4.96571076274075e-05, "loss": 1.2954, "step": 52875 }, { "epoch": 0.05, "learning_rate": 4.965709466257221e-05, "loss": 1.2768, "step": 52876 }, { "epoch": 0.05, "learning_rate": 4.965708169749353e-05, "loss": 1.3392, "step": 52877 }, { "epoch": 0.05, "learning_rate": 4.965706873217145e-05, "loss": 1.1768, "step": 52878 }, { "epoch": 0.05, "learning_rate": 4.965705576660596e-05, "loss": 1.546, "step": 52879 }, { "epoch": 0.05, "learning_rate": 4.965704280079707e-05, "loss": 2.134, "step": 52880 }, { "epoch": 0.05, "learning_rate": 4.965702983474477e-05, "loss": 1.3711, "step": 52881 }, { "epoch": 0.05, "learning_rate": 4.965701686844907e-05, "loss": 1.216, "step": 52882 }, { "epoch": 0.05, "learning_rate": 4.9657003901909966e-05, "loss": 1.3095, "step": 52883 }, { "epoch": 0.05, "learning_rate": 4.965699093512746e-05, "loss": 1.0205, "step": 52884 }, { "epoch": 0.05, "learning_rate": 4.965697796810155e-05, "loss": 1.1242, "step": 52885 }, { "epoch": 0.05, "learning_rate": 4.9656965000832234e-05, "loss": 1.1124, "step": 52886 }, { "epoch": 0.05, "learning_rate": 4.9656952033319516e-05, "loss": 1.4525, "step": 52887 }, { "epoch": 0.05, "learning_rate": 4.965693906556339e-05, "loss": 0.7894, "step": 52888 }, { "epoch": 0.05, "learning_rate": 4.9656926097563864e-05, "loss": 1.0865, "step": 52889 }, { "epoch": 0.05, "learning_rate": 4.9656913129320936e-05, "loss": 1.0817, "step": 52890 }, { "epoch": 0.05, "learning_rate": 4.9656900160834605e-05, "loss": 1.3402, "step": 52891 }, { "epoch": 0.05, "learning_rate": 4.965688719210488e-05, "loss": 1.134, "step": 52892 }, { "epoch": 0.05, "learning_rate": 4.9656874223131746e-05, "loss": 1.2715, "step": 52893 }, { "epoch": 0.05, "learning_rate": 4.965686125391521e-05, "loss": 1.2782, "step": 52894 }, { "epoch": 0.05, "learning_rate": 4.965684828445527e-05, "loss": 1.3081, "step": 52895 }, { "epoch": 0.05, "learning_rate": 4.9656835314751925e-05, "loss": 1.1604, "step": 52896 }, { "epoch": 0.05, "learning_rate": 4.965682234480519e-05, "loss": 1.5323, "step": 52897 }, { "epoch": 0.05, "learning_rate": 4.9656809374615046e-05, "loss": 1.3819, "step": 52898 }, { "epoch": 0.05, "learning_rate": 4.9656796404181494e-05, "loss": 1.3121, "step": 52899 }, { "epoch": 0.05, "learning_rate": 4.965678343350455e-05, "loss": 0.9717, "step": 52900 }, { "epoch": 0.05, "learning_rate": 4.96567704625842e-05, "loss": 1.0636, "step": 52901 }, { "epoch": 0.05, "learning_rate": 4.9656757491420446e-05, "loss": 1.0193, "step": 52902 }, { "epoch": 0.05, "learning_rate": 4.96567445200133e-05, "loss": 1.178, "step": 52903 }, { "epoch": 0.05, "learning_rate": 4.965673154836275e-05, "loss": 1.2832, "step": 52904 }, { "epoch": 0.05, "learning_rate": 4.965671857646879e-05, "loss": 1.2163, "step": 52905 }, { "epoch": 0.05, "learning_rate": 4.9656705604331434e-05, "loss": 0.9797, "step": 52906 }, { "epoch": 0.05, "learning_rate": 4.965669263195068e-05, "loss": 1.1569, "step": 52907 }, { "epoch": 0.05, "learning_rate": 4.965667965932653e-05, "loss": 1.1515, "step": 52908 }, { "epoch": 0.05, "learning_rate": 4.9656666686458976e-05, "loss": 1.0096, "step": 52909 }, { "epoch": 0.05, "learning_rate": 4.9656653713348014e-05, "loss": 1.1373, "step": 52910 }, { "epoch": 0.05, "learning_rate": 4.965664073999366e-05, "loss": 1.1082, "step": 52911 }, { "epoch": 0.05, "learning_rate": 4.96566277663959e-05, "loss": 0.8872, "step": 52912 }, { "epoch": 0.05, "learning_rate": 4.965661479255474e-05, "loss": 1.3254, "step": 52913 }, { "epoch": 0.05, "learning_rate": 4.9656601818470194e-05, "loss": 1.1995, "step": 52914 }, { "epoch": 0.05, "learning_rate": 4.9656588844142236e-05, "loss": 0.7632, "step": 52915 }, { "epoch": 0.05, "learning_rate": 4.965657586957089e-05, "loss": 0.8684, "step": 52916 }, { "epoch": 0.05, "learning_rate": 4.965656289475613e-05, "loss": 0.9951, "step": 52917 }, { "epoch": 0.05, "learning_rate": 4.965654991969798e-05, "loss": 1.2688, "step": 52918 }, { "epoch": 0.05, "learning_rate": 4.965653694439642e-05, "loss": 1.7789, "step": 52919 }, { "epoch": 0.05, "learning_rate": 4.9656523968851475e-05, "loss": 1.1898, "step": 52920 }, { "epoch": 0.05, "learning_rate": 4.9656510993063123e-05, "loss": 1.2944, "step": 52921 }, { "epoch": 0.05, "learning_rate": 4.9656498017031375e-05, "loss": 1.493, "step": 52922 }, { "epoch": 0.05, "learning_rate": 4.965648504075623e-05, "loss": 0.9522, "step": 52923 }, { "epoch": 0.05, "learning_rate": 4.965647206423768e-05, "loss": 0.8703, "step": 52924 }, { "epoch": 0.05, "learning_rate": 4.965645908747574e-05, "loss": 1.3021, "step": 52925 }, { "epoch": 0.05, "learning_rate": 4.96564461104704e-05, "loss": 0.9336, "step": 52926 }, { "epoch": 0.05, "learning_rate": 4.965643313322166e-05, "loss": 0.627, "step": 52927 }, { "epoch": 0.05, "learning_rate": 4.9656420155729524e-05, "loss": 0.6956, "step": 52928 }, { "epoch": 0.05, "learning_rate": 4.9656407177993986e-05, "loss": 0.9962, "step": 52929 }, { "epoch": 0.05, "learning_rate": 4.965639420001506e-05, "loss": 1.063, "step": 52930 }, { "epoch": 0.05, "learning_rate": 4.965638122179273e-05, "loss": 1.1923, "step": 52931 }, { "epoch": 0.05, "learning_rate": 4.965636824332699e-05, "loss": 1.0973, "step": 52932 }, { "epoch": 0.05, "learning_rate": 4.965635526461787e-05, "loss": 1.077, "step": 52933 }, { "epoch": 0.05, "learning_rate": 4.965634228566535e-05, "loss": 1.2719, "step": 52934 }, { "epoch": 0.05, "learning_rate": 4.965632930646943e-05, "loss": 0.9668, "step": 52935 }, { "epoch": 0.05, "learning_rate": 4.965631632703011e-05, "loss": 1.0571, "step": 52936 }, { "epoch": 0.05, "learning_rate": 4.96563033473474e-05, "loss": 0.8367, "step": 52937 }, { "epoch": 0.05, "learning_rate": 4.965629036742129e-05, "loss": 1.4848, "step": 52938 }, { "epoch": 0.05, "learning_rate": 4.965627738725178e-05, "loss": 1.3378, "step": 52939 }, { "epoch": 0.05, "learning_rate": 4.965626440683888e-05, "loss": 1.1154, "step": 52940 }, { "epoch": 0.05, "learning_rate": 4.9656251426182574e-05, "loss": 1.2499, "step": 52941 }, { "epoch": 0.05, "learning_rate": 4.965623844528288e-05, "loss": 1.3406, "step": 52942 }, { "epoch": 0.05, "learning_rate": 4.96562254641398e-05, "loss": 0.6351, "step": 52943 }, { "epoch": 0.05, "learning_rate": 4.96562124827533e-05, "loss": 0.8832, "step": 52944 }, { "epoch": 0.05, "learning_rate": 4.9656199501123426e-05, "loss": 1.1156, "step": 52945 }, { "epoch": 0.05, "learning_rate": 4.965618651925014e-05, "loss": 1.347, "step": 52946 }, { "epoch": 0.05, "learning_rate": 4.965617353713347e-05, "loss": 1.1126, "step": 52947 }, { "epoch": 0.05, "learning_rate": 4.96561605547734e-05, "loss": 1.0564, "step": 52948 }, { "epoch": 0.05, "learning_rate": 4.965614757216994e-05, "loss": 1.1093, "step": 52949 }, { "epoch": 0.05, "learning_rate": 4.965613458932308e-05, "loss": 1.2028, "step": 52950 }, { "epoch": 0.05, "learning_rate": 4.965612160623282e-05, "loss": 1.0488, "step": 52951 }, { "epoch": 0.05, "learning_rate": 4.9656108622899175e-05, "loss": 1.4638, "step": 52952 }, { "epoch": 0.05, "learning_rate": 4.965609563932213e-05, "loss": 1.3498, "step": 52953 }, { "epoch": 0.05, "learning_rate": 4.965608265550169e-05, "loss": 1.4222, "step": 52954 }, { "epoch": 0.05, "learning_rate": 4.9656069671437856e-05, "loss": 0.6852, "step": 52955 }, { "epoch": 0.05, "learning_rate": 4.965605668713062e-05, "loss": 0.8749, "step": 52956 }, { "epoch": 0.05, "learning_rate": 4.965604370258001e-05, "loss": 0.9618, "step": 52957 }, { "epoch": 0.05, "learning_rate": 4.9656030717785995e-05, "loss": 1.1342, "step": 52958 }, { "epoch": 0.05, "learning_rate": 4.9656017732748585e-05, "loss": 0.9014, "step": 52959 }, { "epoch": 0.05, "learning_rate": 4.965600474746778e-05, "loss": 0.8572, "step": 52960 }, { "epoch": 0.05, "learning_rate": 4.9655991761943584e-05, "loss": 0.8862, "step": 52961 }, { "epoch": 0.05, "learning_rate": 4.965597877617599e-05, "loss": 0.7677, "step": 52962 }, { "epoch": 0.05, "learning_rate": 4.9655965790165e-05, "loss": 1.3602, "step": 52963 }, { "epoch": 0.05, "learning_rate": 4.965595280391063e-05, "loss": 1.2945, "step": 52964 }, { "epoch": 0.05, "learning_rate": 4.9655939817412856e-05, "loss": 1.3422, "step": 52965 }, { "epoch": 0.05, "learning_rate": 4.96559268306717e-05, "loss": 1.5799, "step": 52966 }, { "epoch": 0.05, "learning_rate": 4.965591384368714e-05, "loss": 1.0474, "step": 52967 }, { "epoch": 0.05, "learning_rate": 4.965590085645919e-05, "loss": 1.3369, "step": 52968 }, { "epoch": 0.05, "learning_rate": 4.965588786898785e-05, "loss": 1.3637, "step": 52969 }, { "epoch": 0.05, "learning_rate": 4.965587488127311e-05, "loss": 1.3163, "step": 52970 }, { "epoch": 0.05, "learning_rate": 4.965586189331498e-05, "loss": 1.1551, "step": 52971 }, { "epoch": 0.05, "learning_rate": 4.9655848905113466e-05, "loss": 1.5121, "step": 52972 }, { "epoch": 0.05, "learning_rate": 4.965583591666855e-05, "loss": 1.4458, "step": 52973 }, { "epoch": 0.05, "learning_rate": 4.965582292798025e-05, "loss": 1.1531, "step": 52974 }, { "epoch": 0.05, "learning_rate": 4.965580993904855e-05, "loss": 1.0789, "step": 52975 }, { "epoch": 0.05, "learning_rate": 4.965579694987347e-05, "loss": 1.0917, "step": 52976 }, { "epoch": 0.05, "learning_rate": 4.965578396045499e-05, "loss": 0.8156, "step": 52977 }, { "epoch": 0.05, "learning_rate": 4.965577097079312e-05, "loss": 0.5572, "step": 52978 }, { "epoch": 0.05, "learning_rate": 4.965575798088786e-05, "loss": 1.2733, "step": 52979 }, { "epoch": 0.05, "learning_rate": 4.965574499073921e-05, "loss": 1.2591, "step": 52980 }, { "epoch": 0.05, "learning_rate": 4.965573200034716e-05, "loss": 0.9837, "step": 52981 }, { "epoch": 0.05, "learning_rate": 4.965571900971172e-05, "loss": 0.7394, "step": 52982 }, { "epoch": 0.05, "learning_rate": 4.96557060188329e-05, "loss": 1.2243, "step": 52983 }, { "epoch": 0.05, "learning_rate": 4.965569302771068e-05, "loss": 1.4155, "step": 52984 }, { "epoch": 0.05, "learning_rate": 4.965568003634508e-05, "loss": 1.3308, "step": 52985 }, { "epoch": 0.05, "learning_rate": 4.9655667044736075e-05, "loss": 1.094, "step": 52986 }, { "epoch": 0.05, "learning_rate": 4.965565405288369e-05, "loss": 1.022, "step": 52987 }, { "epoch": 0.05, "learning_rate": 4.9655641060787914e-05, "loss": 1.2374, "step": 52988 }, { "epoch": 0.05, "learning_rate": 4.965562806844874e-05, "loss": 1.3106, "step": 52989 }, { "epoch": 0.05, "learning_rate": 4.965561507586618e-05, "loss": 1.2101, "step": 52990 }, { "epoch": 0.05, "learning_rate": 4.965560208304023e-05, "loss": 1.2033, "step": 52991 }, { "epoch": 0.05, "learning_rate": 4.96555890899709e-05, "loss": 1.0466, "step": 52992 }, { "epoch": 0.05, "learning_rate": 4.965557609665817e-05, "loss": 0.3061, "step": 52993 }, { "epoch": 0.05, "learning_rate": 4.9655563103102055e-05, "loss": 1.1072, "step": 52994 }, { "epoch": 0.05, "learning_rate": 4.9655550109302544e-05, "loss": 1.2317, "step": 52995 }, { "epoch": 0.05, "learning_rate": 4.9655537115259656e-05, "loss": 1.0075, "step": 52996 }, { "epoch": 0.05, "learning_rate": 4.9655524120973366e-05, "loss": 0.975, "step": 52997 }, { "epoch": 0.05, "learning_rate": 4.965551112644369e-05, "loss": 1.0692, "step": 52998 }, { "epoch": 0.05, "learning_rate": 4.965549813167063e-05, "loss": 1.178, "step": 52999 }, { "epoch": 0.05, "learning_rate": 4.965548513665418e-05, "loss": 1.2086, "step": 53000 }, { "epoch": 0.05, "eval_loss": 1.0795215368270874, "eval_runtime": 125.2775, "eval_samples_per_second": 11.055, "eval_steps_per_second": 5.532, "step": 53000 }, { "epoch": 0.05, "learning_rate": 4.9655472141394345e-05, "loss": 1.0918, "step": 53001 }, { "epoch": 0.05, "learning_rate": 4.965545914589111e-05, "loss": 1.4648, "step": 53002 }, { "epoch": 0.05, "learning_rate": 4.965544615014449e-05, "loss": 1.5264, "step": 53003 }, { "epoch": 0.05, "learning_rate": 4.965543315415449e-05, "loss": 1.0094, "step": 53004 }, { "epoch": 0.05, "learning_rate": 4.9655420157921094e-05, "loss": 0.7957, "step": 53005 }, { "epoch": 0.05, "learning_rate": 4.9655407161444315e-05, "loss": 1.0061, "step": 53006 }, { "epoch": 0.05, "learning_rate": 4.965539416472415e-05, "loss": 1.3517, "step": 53007 }, { "epoch": 0.05, "learning_rate": 4.9655381167760595e-05, "loss": 1.2169, "step": 53008 }, { "epoch": 0.05, "learning_rate": 4.965536817055365e-05, "loss": 1.1411, "step": 53009 }, { "epoch": 0.05, "learning_rate": 4.965535517310332e-05, "loss": 1.0667, "step": 53010 }, { "epoch": 0.05, "learning_rate": 4.96553421754096e-05, "loss": 0.7201, "step": 53011 }, { "epoch": 0.05, "learning_rate": 4.9655329177472496e-05, "loss": 1.1017, "step": 53012 }, { "epoch": 0.05, "learning_rate": 4.9655316179292e-05, "loss": 0.8591, "step": 53013 }, { "epoch": 0.05, "learning_rate": 4.9655303180868125e-05, "loss": 1.125, "step": 53014 }, { "epoch": 0.05, "learning_rate": 4.965529018220086e-05, "loss": 1.185, "step": 53015 }, { "epoch": 0.05, "learning_rate": 4.965527718329021e-05, "loss": 1.264, "step": 53016 }, { "epoch": 0.05, "learning_rate": 4.965526418413616e-05, "loss": 1.1416, "step": 53017 }, { "epoch": 0.05, "learning_rate": 4.965525118473874e-05, "loss": 0.9915, "step": 53018 }, { "epoch": 0.05, "learning_rate": 4.965523818509793e-05, "loss": 1.058, "step": 53019 }, { "epoch": 0.05, "learning_rate": 4.965522518521373e-05, "loss": 1.1575, "step": 53020 }, { "epoch": 0.05, "learning_rate": 4.965521218508615e-05, "loss": 1.1177, "step": 53021 }, { "epoch": 0.05, "learning_rate": 4.9655199184715174e-05, "loss": 1.0491, "step": 53022 }, { "epoch": 0.05, "learning_rate": 4.965518618410082e-05, "loss": 1.1381, "step": 53023 }, { "epoch": 0.05, "learning_rate": 4.965517318324308e-05, "loss": 1.0955, "step": 53024 }, { "epoch": 0.05, "learning_rate": 4.9655160182141956e-05, "loss": 1.3482, "step": 53025 }, { "epoch": 0.05, "learning_rate": 4.965514718079745e-05, "loss": 1.206, "step": 53026 }, { "epoch": 0.05, "learning_rate": 4.9655134179209544e-05, "loss": 1.1459, "step": 53027 }, { "epoch": 0.05, "learning_rate": 4.965512117737826e-05, "loss": 1.2255, "step": 53028 }, { "epoch": 0.05, "learning_rate": 4.96551081753036e-05, "loss": 1.0166, "step": 53029 }, { "epoch": 0.05, "learning_rate": 4.9655095172985546e-05, "loss": 1.2725, "step": 53030 }, { "epoch": 0.05, "learning_rate": 4.96550821704241e-05, "loss": 1.0934, "step": 53031 }, { "epoch": 0.05, "learning_rate": 4.965506916761928e-05, "loss": 0.9668, "step": 53032 }, { "epoch": 0.05, "learning_rate": 4.965505616457108e-05, "loss": 1.2376, "step": 53033 }, { "epoch": 0.05, "learning_rate": 4.9655043161279494e-05, "loss": 0.9925, "step": 53034 }, { "epoch": 0.05, "learning_rate": 4.965503015774451e-05, "loss": 1.3103, "step": 53035 }, { "epoch": 0.05, "learning_rate": 4.965501715396615e-05, "loss": 0.8787, "step": 53036 }, { "epoch": 0.05, "learning_rate": 4.965500414994442e-05, "loss": 1.0273, "step": 53037 }, { "epoch": 0.05, "learning_rate": 4.9654991145679286e-05, "loss": 1.075, "step": 53038 }, { "epoch": 0.05, "learning_rate": 4.9654978141170785e-05, "loss": 1.2223, "step": 53039 }, { "epoch": 0.05, "learning_rate": 4.965496513641889e-05, "loss": 1.1786, "step": 53040 }, { "epoch": 0.05, "learning_rate": 4.9654952131423606e-05, "loss": 0.9231, "step": 53041 }, { "epoch": 0.05, "learning_rate": 4.965493912618495e-05, "loss": 1.0355, "step": 53042 }, { "epoch": 0.05, "learning_rate": 4.96549261207029e-05, "loss": 1.0205, "step": 53043 }, { "epoch": 0.05, "learning_rate": 4.965491311497747e-05, "loss": 0.9292, "step": 53044 }, { "epoch": 0.05, "learning_rate": 4.965490010900867e-05, "loss": 0.3167, "step": 53045 }, { "epoch": 0.05, "learning_rate": 4.9654887102796477e-05, "loss": 0.3081, "step": 53046 }, { "epoch": 0.05, "learning_rate": 4.9654874096340904e-05, "loss": 0.3848, "step": 53047 }, { "epoch": 0.05, "learning_rate": 4.965486108964195e-05, "loss": 0.3763, "step": 53048 }, { "epoch": 0.05, "learning_rate": 4.96548480826996e-05, "loss": 0.2225, "step": 53049 }, { "epoch": 0.05, "learning_rate": 4.965483507551388e-05, "loss": 0.1422, "step": 53050 }, { "epoch": 0.05, "learning_rate": 4.9654822068084784e-05, "loss": 0.1709, "step": 53051 }, { "epoch": 0.05, "learning_rate": 4.9654809060412296e-05, "loss": 0.1831, "step": 53052 }, { "epoch": 0.05, "learning_rate": 4.9654796052496426e-05, "loss": 0.1252, "step": 53053 }, { "epoch": 0.05, "learning_rate": 4.965478304433718e-05, "loss": 0.1843, "step": 53054 }, { "epoch": 0.05, "learning_rate": 4.965477003593454e-05, "loss": 0.2029, "step": 53055 }, { "epoch": 0.05, "learning_rate": 4.9654757027288535e-05, "loss": 0.0846, "step": 53056 }, { "epoch": 0.05, "learning_rate": 4.965474401839913e-05, "loss": 0.11, "step": 53057 }, { "epoch": 0.05, "learning_rate": 4.9654731009266367e-05, "loss": 0.1668, "step": 53058 }, { "epoch": 0.05, "learning_rate": 4.965471799989021e-05, "loss": 0.0909, "step": 53059 }, { "epoch": 0.05, "learning_rate": 4.9654704990270666e-05, "loss": 0.1357, "step": 53060 }, { "epoch": 0.05, "learning_rate": 4.965469198040775e-05, "loss": 0.4276, "step": 53061 }, { "epoch": 0.05, "learning_rate": 4.9654678970301446e-05, "loss": 1.3594, "step": 53062 }, { "epoch": 0.05, "learning_rate": 4.965466595995177e-05, "loss": 0.9351, "step": 53063 }, { "epoch": 0.05, "learning_rate": 4.9654652949358714e-05, "loss": 1.1641, "step": 53064 }, { "epoch": 0.05, "learning_rate": 4.9654639938522274e-05, "loss": 1.1929, "step": 53065 }, { "epoch": 0.05, "learning_rate": 4.965462692744245e-05, "loss": 1.3604, "step": 53066 }, { "epoch": 0.05, "learning_rate": 4.965461391611925e-05, "loss": 1.4407, "step": 53067 }, { "epoch": 0.05, "learning_rate": 4.965460090455267e-05, "loss": 1.3064, "step": 53068 }, { "epoch": 0.05, "learning_rate": 4.965458789274271e-05, "loss": 0.8301, "step": 53069 }, { "epoch": 0.05, "learning_rate": 4.965457488068937e-05, "loss": 1.4657, "step": 53070 }, { "epoch": 0.05, "learning_rate": 4.965456186839265e-05, "loss": 1.329, "step": 53071 }, { "epoch": 0.05, "learning_rate": 4.965454885585255e-05, "loss": 1.0141, "step": 53072 }, { "epoch": 0.05, "learning_rate": 4.965453584306907e-05, "loss": 1.1431, "step": 53073 }, { "epoch": 0.05, "learning_rate": 4.965452283004221e-05, "loss": 2.1085, "step": 53074 }, { "epoch": 0.05, "learning_rate": 4.965450981677198e-05, "loss": 1.0998, "step": 53075 }, { "epoch": 0.05, "learning_rate": 4.965449680325836e-05, "loss": 0.7222, "step": 53076 }, { "epoch": 0.05, "learning_rate": 4.965448378950136e-05, "loss": 1.1724, "step": 53077 }, { "epoch": 0.05, "learning_rate": 4.9654470775500996e-05, "loss": 1.215, "step": 53078 }, { "epoch": 0.05, "learning_rate": 4.965445776125724e-05, "loss": 0.5172, "step": 53079 }, { "epoch": 0.05, "learning_rate": 4.9654444746770115e-05, "loss": 0.96, "step": 53080 }, { "epoch": 0.05, "learning_rate": 4.96544317320396e-05, "loss": 1.0981, "step": 53081 }, { "epoch": 0.05, "learning_rate": 4.965441871706571e-05, "loss": 1.4621, "step": 53082 }, { "epoch": 0.05, "learning_rate": 4.965440570184845e-05, "loss": 1.0534, "step": 53083 }, { "epoch": 0.05, "learning_rate": 4.9654392686387804e-05, "loss": 1.4694, "step": 53084 }, { "epoch": 0.05, "learning_rate": 4.9654379670683784e-05, "loss": 1.8152, "step": 53085 }, { "epoch": 0.05, "learning_rate": 4.965436665473639e-05, "loss": 1.2454, "step": 53086 }, { "epoch": 0.05, "learning_rate": 4.9654353638545614e-05, "loss": 1.0771, "step": 53087 }, { "epoch": 0.05, "learning_rate": 4.965434062211146e-05, "loss": 0.35, "step": 53088 }, { "epoch": 0.05, "learning_rate": 4.9654327605433927e-05, "loss": 0.5211, "step": 53089 }, { "epoch": 0.05, "learning_rate": 4.965431458851302e-05, "loss": 0.9494, "step": 53090 }, { "epoch": 0.05, "learning_rate": 4.965430157134874e-05, "loss": 1.3897, "step": 53091 }, { "epoch": 0.05, "learning_rate": 4.965428855394107e-05, "loss": 1.0161, "step": 53092 }, { "epoch": 0.05, "learning_rate": 4.965427553629004e-05, "loss": 1.4859, "step": 53093 }, { "epoch": 0.05, "learning_rate": 4.965426251839562e-05, "loss": 1.2105, "step": 53094 }, { "epoch": 0.05, "learning_rate": 4.965424950025783e-05, "loss": 1.0042, "step": 53095 }, { "epoch": 0.05, "learning_rate": 4.9654236481876666e-05, "loss": 1.0785, "step": 53096 }, { "epoch": 0.05, "learning_rate": 4.965422346325212e-05, "loss": 1.149, "step": 53097 }, { "epoch": 0.05, "learning_rate": 4.96542104443842e-05, "loss": 1.2495, "step": 53098 }, { "epoch": 0.05, "learning_rate": 4.965419742527291e-05, "loss": 0.9229, "step": 53099 }, { "epoch": 0.05, "learning_rate": 4.965418440591824e-05, "loss": 0.7052, "step": 53100 }, { "epoch": 0.05, "learning_rate": 4.9654171386320184e-05, "loss": 0.9676, "step": 53101 }, { "epoch": 0.05, "learning_rate": 4.965415836647877e-05, "loss": 1.0949, "step": 53102 }, { "epoch": 0.05, "learning_rate": 4.965414534639397e-05, "loss": 1.2735, "step": 53103 }, { "epoch": 0.05, "learning_rate": 4.96541323260658e-05, "loss": 1.2502, "step": 53104 }, { "epoch": 0.05, "learning_rate": 4.965411930549424e-05, "loss": 1.0803, "step": 53105 }, { "epoch": 0.05, "learning_rate": 4.9654106284679316e-05, "loss": 1.2496, "step": 53106 }, { "epoch": 0.05, "learning_rate": 4.965409326362102e-05, "loss": 1.4077, "step": 53107 }, { "epoch": 0.05, "learning_rate": 4.965408024231934e-05, "loss": 1.1196, "step": 53108 }, { "epoch": 0.05, "learning_rate": 4.96540672207743e-05, "loss": 1.3141, "step": 53109 }, { "epoch": 0.05, "learning_rate": 4.965405419898587e-05, "loss": 0.9451, "step": 53110 }, { "epoch": 0.05, "learning_rate": 4.9654041176954075e-05, "loss": 1.2035, "step": 53111 }, { "epoch": 0.05, "learning_rate": 4.96540281546789e-05, "loss": 1.0976, "step": 53112 }, { "epoch": 0.05, "learning_rate": 4.965401513216036e-05, "loss": 1.1791, "step": 53113 }, { "epoch": 0.05, "learning_rate": 4.965400210939843e-05, "loss": 0.9901, "step": 53114 }, { "epoch": 0.05, "learning_rate": 4.9653989086393145e-05, "loss": 1.7498, "step": 53115 }, { "epoch": 0.05, "learning_rate": 4.9653976063144474e-05, "loss": 1.1143, "step": 53116 }, { "epoch": 0.05, "learning_rate": 4.965396303965243e-05, "loss": 1.3279, "step": 53117 }, { "epoch": 0.05, "learning_rate": 4.9653950015917015e-05, "loss": 1.2797, "step": 53118 }, { "epoch": 0.05, "learning_rate": 4.9653936991938235e-05, "loss": 1.357, "step": 53119 }, { "epoch": 0.05, "learning_rate": 4.965392396771607e-05, "loss": 1.2874, "step": 53120 }, { "epoch": 0.05, "learning_rate": 4.965391094325054e-05, "loss": 1.1594, "step": 53121 }, { "epoch": 0.05, "learning_rate": 4.9653897918541624e-05, "loss": 0.9687, "step": 53122 }, { "epoch": 0.05, "learning_rate": 4.9653884893589346e-05, "loss": 1.1725, "step": 53123 }, { "epoch": 0.05, "learning_rate": 4.965387186839369e-05, "loss": 1.3107, "step": 53124 }, { "epoch": 0.05, "learning_rate": 4.965385884295467e-05, "loss": 0.9059, "step": 53125 }, { "epoch": 0.05, "learning_rate": 4.9653845817272274e-05, "loss": 1.3024, "step": 53126 }, { "epoch": 0.05, "learning_rate": 4.96538327913465e-05, "loss": 1.108, "step": 53127 }, { "epoch": 0.05, "learning_rate": 4.965381976517736e-05, "loss": 1.1219, "step": 53128 }, { "epoch": 0.05, "learning_rate": 4.965380673876484e-05, "loss": 0.8853, "step": 53129 }, { "epoch": 0.05, "learning_rate": 4.965379371210895e-05, "loss": 1.1046, "step": 53130 }, { "epoch": 0.05, "learning_rate": 4.9653780685209696e-05, "loss": 1.3432, "step": 53131 }, { "epoch": 0.05, "learning_rate": 4.9653767658067065e-05, "loss": 0.619, "step": 53132 }, { "epoch": 0.05, "learning_rate": 4.965375463068106e-05, "loss": 0.9098, "step": 53133 }, { "epoch": 0.05, "learning_rate": 4.965374160305169e-05, "loss": 1.1448, "step": 53134 }, { "epoch": 0.05, "learning_rate": 4.9653728575178934e-05, "loss": 1.4842, "step": 53135 }, { "epoch": 0.05, "learning_rate": 4.9653715547062825e-05, "loss": 1.1595, "step": 53136 }, { "epoch": 0.05, "learning_rate": 4.965370251870334e-05, "loss": 1.2213, "step": 53137 }, { "epoch": 0.05, "learning_rate": 4.965368949010047e-05, "loss": 1.2282, "step": 53138 }, { "epoch": 0.05, "learning_rate": 4.965367646125425e-05, "loss": 1.1491, "step": 53139 }, { "epoch": 0.05, "learning_rate": 4.965366343216464e-05, "loss": 1.0519, "step": 53140 }, { "epoch": 0.05, "learning_rate": 4.965365040283168e-05, "loss": 1.1426, "step": 53141 }, { "epoch": 0.05, "learning_rate": 4.965363737325534e-05, "loss": 1.2439, "step": 53142 }, { "epoch": 0.05, "learning_rate": 4.965362434343562e-05, "loss": 0.7255, "step": 53143 }, { "epoch": 0.05, "learning_rate": 4.9653611313372536e-05, "loss": 1.5725, "step": 53144 }, { "epoch": 0.05, "learning_rate": 4.965359828306609e-05, "loss": 1.0603, "step": 53145 }, { "epoch": 0.05, "learning_rate": 4.9653585252516264e-05, "loss": 1.1895, "step": 53146 }, { "epoch": 0.05, "learning_rate": 4.9653572221723076e-05, "loss": 1.0268, "step": 53147 }, { "epoch": 0.05, "learning_rate": 4.965355919068651e-05, "loss": 1.1062, "step": 53148 }, { "epoch": 0.05, "learning_rate": 4.965354615940658e-05, "loss": 1.0457, "step": 53149 }, { "epoch": 0.05, "learning_rate": 4.9653533127883285e-05, "loss": 1.5248, "step": 53150 }, { "epoch": 0.05, "learning_rate": 4.965352009611661e-05, "loss": 1.1876, "step": 53151 }, { "epoch": 0.05, "learning_rate": 4.965350706410658e-05, "loss": 1.2426, "step": 53152 }, { "epoch": 0.05, "learning_rate": 4.965349403185317e-05, "loss": 1.1893, "step": 53153 }, { "epoch": 0.05, "learning_rate": 4.9653480999356393e-05, "loss": 1.1679, "step": 53154 }, { "epoch": 0.05, "learning_rate": 4.9653467966616243e-05, "loss": 1.2156, "step": 53155 }, { "epoch": 0.05, "learning_rate": 4.965345493363273e-05, "loss": 1.1117, "step": 53156 }, { "epoch": 0.05, "learning_rate": 4.965344190040585e-05, "loss": 1.0002, "step": 53157 }, { "epoch": 0.05, "learning_rate": 4.9653428866935604e-05, "loss": 1.0671, "step": 53158 }, { "epoch": 0.05, "learning_rate": 4.965341583322198e-05, "loss": 1.2269, "step": 53159 }, { "epoch": 0.05, "learning_rate": 4.965340279926499e-05, "loss": 1.0846, "step": 53160 }, { "epoch": 0.05, "learning_rate": 4.965338976506464e-05, "loss": 1.0529, "step": 53161 }, { "epoch": 0.05, "learning_rate": 4.9653376730620916e-05, "loss": 1.4735, "step": 53162 }, { "epoch": 0.05, "learning_rate": 4.965336369593382e-05, "loss": 1.1667, "step": 53163 }, { "epoch": 0.05, "learning_rate": 4.965335066100336e-05, "loss": 0.9316, "step": 53164 }, { "epoch": 0.05, "learning_rate": 4.9653337625829536e-05, "loss": 1.0858, "step": 53165 }, { "epoch": 0.05, "learning_rate": 4.965332459041234e-05, "loss": 1.0452, "step": 53166 }, { "epoch": 0.05, "learning_rate": 4.9653311554751785e-05, "loss": 1.0667, "step": 53167 }, { "epoch": 0.05, "learning_rate": 4.965329851884786e-05, "loss": 1.357, "step": 53168 }, { "epoch": 0.05, "learning_rate": 4.965328548270056e-05, "loss": 1.2924, "step": 53169 }, { "epoch": 0.05, "learning_rate": 4.96532724463099e-05, "loss": 1.1172, "step": 53170 }, { "epoch": 0.05, "learning_rate": 4.9653259409675876e-05, "loss": 1.6051, "step": 53171 }, { "epoch": 0.05, "learning_rate": 4.965324637279848e-05, "loss": 1.5256, "step": 53172 }, { "epoch": 0.05, "learning_rate": 4.965323333567772e-05, "loss": 1.2752, "step": 53173 }, { "epoch": 0.05, "learning_rate": 4.965322029831359e-05, "loss": 1.0376, "step": 53174 }, { "epoch": 0.05, "learning_rate": 4.965320726070609e-05, "loss": 0.7585, "step": 53175 }, { "epoch": 0.05, "learning_rate": 4.965319422285524e-05, "loss": 0.7985, "step": 53176 }, { "epoch": 0.05, "learning_rate": 4.965318118476101e-05, "loss": 1.143, "step": 53177 }, { "epoch": 0.05, "learning_rate": 4.965316814642342e-05, "loss": 0.8395, "step": 53178 }, { "epoch": 0.05, "learning_rate": 4.9653155107842464e-05, "loss": 1.0376, "step": 53179 }, { "epoch": 0.05, "learning_rate": 4.965314206901814e-05, "loss": 1.1177, "step": 53180 }, { "epoch": 0.05, "learning_rate": 4.9653129029950455e-05, "loss": 0.9937, "step": 53181 }, { "epoch": 0.05, "learning_rate": 4.9653115990639396e-05, "loss": 1.1624, "step": 53182 }, { "epoch": 0.05, "learning_rate": 4.9653102951084974e-05, "loss": 1.0969, "step": 53183 }, { "epoch": 0.05, "learning_rate": 4.965308991128719e-05, "loss": 1.0646, "step": 53184 }, { "epoch": 0.05, "learning_rate": 4.965307687124604e-05, "loss": 1.0465, "step": 53185 }, { "epoch": 0.05, "learning_rate": 4.965306383096153e-05, "loss": 1.2556, "step": 53186 }, { "epoch": 0.05, "learning_rate": 4.9653050790433654e-05, "loss": 0.4969, "step": 53187 }, { "epoch": 0.05, "learning_rate": 4.965303774966241e-05, "loss": 0.765, "step": 53188 }, { "epoch": 0.05, "learning_rate": 4.96530247086478e-05, "loss": 1.006, "step": 53189 }, { "epoch": 0.05, "learning_rate": 4.965301166738983e-05, "loss": 1.3217, "step": 53190 }, { "epoch": 0.05, "learning_rate": 4.9652998625888496e-05, "loss": 1.2746, "step": 53191 }, { "epoch": 0.05, "learning_rate": 4.9652985584143796e-05, "loss": 1.2008, "step": 53192 }, { "epoch": 0.05, "learning_rate": 4.965297254215573e-05, "loss": 0.7246, "step": 53193 }, { "epoch": 0.05, "learning_rate": 4.965295949992431e-05, "loss": 1.2258, "step": 53194 }, { "epoch": 0.05, "learning_rate": 4.9652946457449514e-05, "loss": 1.1334, "step": 53195 }, { "epoch": 0.05, "learning_rate": 4.9652933414731364e-05, "loss": 1.0117, "step": 53196 }, { "epoch": 0.05, "learning_rate": 4.965292037176984e-05, "loss": 1.0925, "step": 53197 }, { "epoch": 0.05, "learning_rate": 4.965290732856496e-05, "loss": 1.3899, "step": 53198 }, { "epoch": 0.05, "learning_rate": 4.965289428511672e-05, "loss": 1.5564, "step": 53199 }, { "epoch": 0.05, "learning_rate": 4.965288124142511e-05, "loss": 1.156, "step": 53200 }, { "epoch": 0.05, "learning_rate": 4.965286819749014e-05, "loss": 1.4517, "step": 53201 }, { "epoch": 0.05, "learning_rate": 4.965285515331181e-05, "loss": 2.0287, "step": 53202 }, { "epoch": 0.05, "learning_rate": 4.965284210889011e-05, "loss": 1.3747, "step": 53203 }, { "epoch": 0.05, "learning_rate": 4.965282906422506e-05, "loss": 1.3385, "step": 53204 }, { "epoch": 0.05, "learning_rate": 4.965281601931664e-05, "loss": 1.3633, "step": 53205 }, { "epoch": 0.05, "learning_rate": 4.965280297416485e-05, "loss": 1.2381, "step": 53206 }, { "epoch": 0.05, "learning_rate": 4.9652789928769706e-05, "loss": 1.4876, "step": 53207 }, { "epoch": 0.05, "learning_rate": 4.96527768831312e-05, "loss": 1.4048, "step": 53208 }, { "epoch": 0.05, "learning_rate": 4.9652763837249336e-05, "loss": 1.3431, "step": 53209 }, { "epoch": 0.05, "learning_rate": 4.96527507911241e-05, "loss": 1.2417, "step": 53210 }, { "epoch": 0.05, "learning_rate": 4.965273774475551e-05, "loss": 1.4127, "step": 53211 }, { "epoch": 0.05, "learning_rate": 4.965272469814356e-05, "loss": 1.2553, "step": 53212 }, { "epoch": 0.05, "learning_rate": 4.9652711651288244e-05, "loss": 1.1956, "step": 53213 }, { "epoch": 0.05, "learning_rate": 4.965269860418957e-05, "loss": 1.1004, "step": 53214 }, { "epoch": 0.05, "learning_rate": 4.9652685556847536e-05, "loss": 1.1371, "step": 53215 }, { "epoch": 0.05, "learning_rate": 4.965267250926214e-05, "loss": 1.4847, "step": 53216 }, { "epoch": 0.05, "learning_rate": 4.965265946143338e-05, "loss": 1.2149, "step": 53217 }, { "epoch": 0.05, "learning_rate": 4.965264641336126e-05, "loss": 1.2932, "step": 53218 }, { "epoch": 0.05, "learning_rate": 4.965263336504578e-05, "loss": 1.2239, "step": 53219 }, { "epoch": 0.05, "learning_rate": 4.965262031648694e-05, "loss": 1.4982, "step": 53220 }, { "epoch": 0.05, "learning_rate": 4.9652607267684745e-05, "loss": 1.0904, "step": 53221 }, { "epoch": 0.05, "learning_rate": 4.965259421863918e-05, "loss": 1.0879, "step": 53222 }, { "epoch": 0.05, "learning_rate": 4.9652581169350266e-05, "loss": 1.1686, "step": 53223 }, { "epoch": 0.05, "learning_rate": 4.965256811981798e-05, "loss": 0.9726, "step": 53224 }, { "epoch": 0.05, "learning_rate": 4.965255507004234e-05, "loss": 1.0661, "step": 53225 }, { "epoch": 0.05, "learning_rate": 4.965254202002334e-05, "loss": 1.0883, "step": 53226 }, { "epoch": 0.05, "learning_rate": 4.9652528969760983e-05, "loss": 0.9092, "step": 53227 }, { "epoch": 0.05, "learning_rate": 4.9652515919255266e-05, "loss": 1.1259, "step": 53228 }, { "epoch": 0.05, "learning_rate": 4.965250286850619e-05, "loss": 1.2653, "step": 53229 }, { "epoch": 0.05, "learning_rate": 4.965248981751375e-05, "loss": 1.2713, "step": 53230 }, { "epoch": 0.05, "learning_rate": 4.965247676627796e-05, "loss": 1.2564, "step": 53231 }, { "epoch": 0.05, "learning_rate": 4.9652463714798803e-05, "loss": 1.2612, "step": 53232 }, { "epoch": 0.05, "learning_rate": 4.9652450663076286e-05, "loss": 1.135, "step": 53233 }, { "epoch": 0.05, "learning_rate": 4.965243761111041e-05, "loss": 1.1289, "step": 53234 }, { "epoch": 0.05, "learning_rate": 4.965242455890118e-05, "loss": 1.1935, "step": 53235 }, { "epoch": 0.05, "learning_rate": 4.965241150644859e-05, "loss": 1.0604, "step": 53236 }, { "epoch": 0.05, "learning_rate": 4.965239845375264e-05, "loss": 1.1751, "step": 53237 }, { "epoch": 0.05, "learning_rate": 4.9652385400813346e-05, "loss": 1.3916, "step": 53238 }, { "epoch": 0.05, "learning_rate": 4.965237234763068e-05, "loss": 1.1844, "step": 53239 }, { "epoch": 0.05, "learning_rate": 4.965235929420466e-05, "loss": 1.1626, "step": 53240 }, { "epoch": 0.05, "learning_rate": 4.9652346240535285e-05, "loss": 0.964, "step": 53241 }, { "epoch": 0.05, "learning_rate": 4.9652333186622543e-05, "loss": 1.0133, "step": 53242 }, { "epoch": 0.05, "learning_rate": 4.965232013246645e-05, "loss": 1.0056, "step": 53243 }, { "epoch": 0.05, "learning_rate": 4.9652307078067e-05, "loss": 1.018, "step": 53244 }, { "epoch": 0.05, "learning_rate": 4.96522940234242e-05, "loss": 0.9067, "step": 53245 }, { "epoch": 0.05, "learning_rate": 4.9652280968538025e-05, "loss": 1.3766, "step": 53246 }, { "epoch": 0.05, "learning_rate": 4.9652267913408504e-05, "loss": 1.0725, "step": 53247 }, { "epoch": 0.05, "learning_rate": 4.965225485803563e-05, "loss": 1.0633, "step": 53248 }, { "epoch": 0.05, "learning_rate": 4.9652241802419394e-05, "loss": 0.9355, "step": 53249 }, { "epoch": 0.05, "learning_rate": 4.9652228746559805e-05, "loss": 1.2827, "step": 53250 }, { "epoch": 0.05, "learning_rate": 4.965221569045685e-05, "loss": 1.0303, "step": 53251 }, { "epoch": 0.05, "learning_rate": 4.9652202634110546e-05, "loss": 1.2783, "step": 53252 }, { "epoch": 0.05, "learning_rate": 4.965218957752088e-05, "loss": 1.3718, "step": 53253 }, { "epoch": 0.05, "learning_rate": 4.965217652068786e-05, "loss": 1.1512, "step": 53254 }, { "epoch": 0.05, "learning_rate": 4.9652163463611494e-05, "loss": 1.197, "step": 53255 }, { "epoch": 0.05, "learning_rate": 4.965215040629176e-05, "loss": 0.9446, "step": 53256 }, { "epoch": 0.05, "learning_rate": 4.965213734872868e-05, "loss": 1.0328, "step": 53257 }, { "epoch": 0.05, "learning_rate": 4.965212429092224e-05, "loss": 0.4073, "step": 53258 }, { "epoch": 0.05, "learning_rate": 4.9652111232872446e-05, "loss": 0.8353, "step": 53259 }, { "epoch": 0.05, "learning_rate": 4.96520981745793e-05, "loss": 1.1374, "step": 53260 }, { "epoch": 0.05, "learning_rate": 4.965208511604279e-05, "loss": 1.1039, "step": 53261 }, { "epoch": 0.05, "learning_rate": 4.965207205726293e-05, "loss": 1.6085, "step": 53262 }, { "epoch": 0.05, "learning_rate": 4.9652058998239716e-05, "loss": 0.9535, "step": 53263 }, { "epoch": 0.05, "learning_rate": 4.9652045938973144e-05, "loss": 1.0242, "step": 53264 }, { "epoch": 0.05, "learning_rate": 4.965203287946322e-05, "loss": 1.1574, "step": 53265 }, { "epoch": 0.05, "learning_rate": 4.9652019819709945e-05, "loss": 1.172, "step": 53266 }, { "epoch": 0.05, "learning_rate": 4.965200675971331e-05, "loss": 1.5505, "step": 53267 }, { "epoch": 0.05, "learning_rate": 4.965199369947332e-05, "loss": 1.0477, "step": 53268 }, { "epoch": 0.05, "learning_rate": 4.9651980638989984e-05, "loss": 1.0185, "step": 53269 }, { "epoch": 0.05, "learning_rate": 4.965196757826328e-05, "loss": 1.2357, "step": 53270 }, { "epoch": 0.05, "learning_rate": 4.965195451729323e-05, "loss": 0.9918, "step": 53271 }, { "epoch": 0.05, "learning_rate": 4.965194145607983e-05, "loss": 0.9173, "step": 53272 }, { "epoch": 0.05, "learning_rate": 4.965192839462307e-05, "loss": 1.369, "step": 53273 }, { "epoch": 0.05, "learning_rate": 4.965191533292296e-05, "loss": 0.9896, "step": 53274 }, { "epoch": 0.05, "learning_rate": 4.96519022709795e-05, "loss": 1.113, "step": 53275 }, { "epoch": 0.05, "learning_rate": 4.965188920879268e-05, "loss": 1.0288, "step": 53276 }, { "epoch": 0.05, "learning_rate": 4.9651876146362515e-05, "loss": 1.3298, "step": 53277 }, { "epoch": 0.05, "learning_rate": 4.9651863083688987e-05, "loss": 1.0366, "step": 53278 }, { "epoch": 0.05, "learning_rate": 4.9651850020772116e-05, "loss": 1.3824, "step": 53279 }, { "epoch": 0.05, "learning_rate": 4.965183695761189e-05, "loss": 1.0523, "step": 53280 }, { "epoch": 0.05, "learning_rate": 4.9651823894208306e-05, "loss": 0.8083, "step": 53281 }, { "epoch": 0.05, "learning_rate": 4.9651810830561374e-05, "loss": 0.8094, "step": 53282 }, { "epoch": 0.05, "learning_rate": 4.965179776667108e-05, "loss": 0.8013, "step": 53283 }, { "epoch": 0.05, "learning_rate": 4.965178470253745e-05, "loss": 0.6517, "step": 53284 }, { "epoch": 0.05, "learning_rate": 4.965177163816046e-05, "loss": 0.7045, "step": 53285 }, { "epoch": 0.05, "learning_rate": 4.965175857354012e-05, "loss": 0.455, "step": 53286 }, { "epoch": 0.05, "learning_rate": 4.965174550867642e-05, "loss": 0.6921, "step": 53287 }, { "epoch": 0.05, "learning_rate": 4.9651732443569374e-05, "loss": 0.4191, "step": 53288 }, { "epoch": 0.05, "learning_rate": 4.965171937821898e-05, "loss": 0.5449, "step": 53289 }, { "epoch": 0.05, "learning_rate": 4.965170631262523e-05, "loss": 0.3782, "step": 53290 }, { "epoch": 0.05, "learning_rate": 4.965169324678814e-05, "loss": 0.3592, "step": 53291 }, { "epoch": 0.05, "learning_rate": 4.9651680180707686e-05, "loss": 0.3977, "step": 53292 }, { "epoch": 0.05, "learning_rate": 4.965166711438388e-05, "loss": 0.3411, "step": 53293 }, { "epoch": 0.05, "learning_rate": 4.965165404781673e-05, "loss": 0.2928, "step": 53294 }, { "epoch": 0.05, "learning_rate": 4.9651640981006224e-05, "loss": 0.2743, "step": 53295 }, { "epoch": 0.05, "learning_rate": 4.965162791395237e-05, "loss": 0.2434, "step": 53296 }, { "epoch": 0.05, "learning_rate": 4.965161484665517e-05, "loss": 0.4154, "step": 53297 }, { "epoch": 0.05, "learning_rate": 4.9651601779114613e-05, "loss": 0.4264, "step": 53298 }, { "epoch": 0.05, "learning_rate": 4.965158871133071e-05, "loss": 1.1799, "step": 53299 }, { "epoch": 0.05, "learning_rate": 4.9651575643303454e-05, "loss": 1.3861, "step": 53300 }, { "epoch": 0.05, "learning_rate": 4.965156257503285e-05, "loss": 1.2216, "step": 53301 }, { "epoch": 0.05, "learning_rate": 4.96515495065189e-05, "loss": 1.3466, "step": 53302 }, { "epoch": 0.05, "learning_rate": 4.965153643776159e-05, "loss": 1.1063, "step": 53303 }, { "epoch": 0.05, "learning_rate": 4.965152336876094e-05, "loss": 1.2642, "step": 53304 }, { "epoch": 0.05, "learning_rate": 4.965151029951693e-05, "loss": 1.1761, "step": 53305 }, { "epoch": 0.05, "learning_rate": 4.965149723002958e-05, "loss": 1.2504, "step": 53306 }, { "epoch": 0.05, "learning_rate": 4.965148416029888e-05, "loss": 1.1238, "step": 53307 }, { "epoch": 0.05, "learning_rate": 4.9651471090324833e-05, "loss": 0.8288, "step": 53308 }, { "epoch": 0.05, "learning_rate": 4.9651458020107426e-05, "loss": 1.2031, "step": 53309 }, { "epoch": 0.05, "learning_rate": 4.965144494964668e-05, "loss": 1.0686, "step": 53310 }, { "epoch": 0.05, "learning_rate": 4.9651431878942584e-05, "loss": 0.9942, "step": 53311 }, { "epoch": 0.05, "learning_rate": 4.9651418807995136e-05, "loss": 1.4284, "step": 53312 }, { "epoch": 0.05, "learning_rate": 4.9651405736804346e-05, "loss": 1.3363, "step": 53313 }, { "epoch": 0.05, "learning_rate": 4.96513926653702e-05, "loss": 0.7364, "step": 53314 }, { "epoch": 0.05, "learning_rate": 4.965137959369271e-05, "loss": 1.0554, "step": 53315 }, { "epoch": 0.05, "learning_rate": 4.965136652177187e-05, "loss": 0.727, "step": 53316 }, { "epoch": 0.05, "learning_rate": 4.9651353449607686e-05, "loss": 1.1455, "step": 53317 }, { "epoch": 0.05, "learning_rate": 4.965134037720015e-05, "loss": 1.2183, "step": 53318 }, { "epoch": 0.05, "learning_rate": 4.9651327304549264e-05, "loss": 1.1313, "step": 53319 }, { "epoch": 0.05, "learning_rate": 4.9651314231655036e-05, "loss": 1.1253, "step": 53320 }, { "epoch": 0.05, "learning_rate": 4.965130115851746e-05, "loss": 1.0025, "step": 53321 }, { "epoch": 0.05, "learning_rate": 4.965128808513654e-05, "loss": 1.1156, "step": 53322 }, { "epoch": 0.05, "learning_rate": 4.9651275011512264e-05, "loss": 1.1353, "step": 53323 }, { "epoch": 0.05, "learning_rate": 4.9651261937644646e-05, "loss": 1.4051, "step": 53324 }, { "epoch": 0.05, "learning_rate": 4.9651248863533686e-05, "loss": 2.0279, "step": 53325 }, { "epoch": 0.05, "learning_rate": 4.965123578917937e-05, "loss": 1.0955, "step": 53326 }, { "epoch": 0.05, "learning_rate": 4.965122271458171e-05, "loss": 0.6125, "step": 53327 }, { "epoch": 0.05, "learning_rate": 4.96512096397407e-05, "loss": 0.9983, "step": 53328 }, { "epoch": 0.05, "learning_rate": 4.965119656465635e-05, "loss": 1.2834, "step": 53329 }, { "epoch": 0.05, "learning_rate": 4.9651183489328653e-05, "loss": 1.1513, "step": 53330 }, { "epoch": 0.05, "learning_rate": 4.965117041375761e-05, "loss": 1.1778, "step": 53331 }, { "epoch": 0.05, "learning_rate": 4.965115733794322e-05, "loss": 1.0513, "step": 53332 }, { "epoch": 0.05, "learning_rate": 4.965114426188549e-05, "loss": 1.1943, "step": 53333 }, { "epoch": 0.05, "learning_rate": 4.965113118558441e-05, "loss": 1.3158, "step": 53334 }, { "epoch": 0.05, "learning_rate": 4.965111810903998e-05, "loss": 1.2732, "step": 53335 }, { "epoch": 0.05, "learning_rate": 4.965110503225221e-05, "loss": 1.153, "step": 53336 }, { "epoch": 0.05, "learning_rate": 4.965109195522108e-05, "loss": 1.3121, "step": 53337 }, { "epoch": 0.05, "learning_rate": 4.9651078877946624e-05, "loss": 1.2115, "step": 53338 }, { "epoch": 0.05, "learning_rate": 4.965106580042882e-05, "loss": 1.336, "step": 53339 }, { "epoch": 0.05, "learning_rate": 4.9651052722667665e-05, "loss": 1.7577, "step": 53340 }, { "epoch": 0.05, "learning_rate": 4.9651039644663166e-05, "loss": 1.3294, "step": 53341 }, { "epoch": 0.05, "learning_rate": 4.9651026566415323e-05, "loss": 0.8671, "step": 53342 }, { "epoch": 0.05, "learning_rate": 4.965101348792414e-05, "loss": 1.0814, "step": 53343 }, { "epoch": 0.05, "learning_rate": 4.9651000409189605e-05, "loss": 1.3998, "step": 53344 }, { "epoch": 0.05, "learning_rate": 4.965098733021173e-05, "loss": 1.1623, "step": 53345 }, { "epoch": 0.05, "learning_rate": 4.965097425099051e-05, "loss": 0.9759, "step": 53346 }, { "epoch": 0.05, "learning_rate": 4.965096117152594e-05, "loss": 1.0782, "step": 53347 }, { "epoch": 0.05, "learning_rate": 4.965094809181804e-05, "loss": 1.1268, "step": 53348 }, { "epoch": 0.05, "learning_rate": 4.965093501186679e-05, "loss": 0.9828, "step": 53349 }, { "epoch": 0.05, "learning_rate": 4.965092193167219e-05, "loss": 1.0102, "step": 53350 }, { "epoch": 0.05, "learning_rate": 4.9650908851234256e-05, "loss": 1.366, "step": 53351 }, { "epoch": 0.05, "learning_rate": 4.9650895770552976e-05, "loss": 0.9978, "step": 53352 }, { "epoch": 0.05, "learning_rate": 4.965088268962835e-05, "loss": 1.1448, "step": 53353 }, { "epoch": 0.05, "learning_rate": 4.965086960846038e-05, "loss": 0.5632, "step": 53354 }, { "epoch": 0.05, "learning_rate": 4.965085652704907e-05, "loss": 2.1856, "step": 53355 }, { "epoch": 0.05, "learning_rate": 4.965084344539441e-05, "loss": 0.8523, "step": 53356 }, { "epoch": 0.05, "learning_rate": 4.9650830363496415e-05, "loss": 1.0147, "step": 53357 }, { "epoch": 0.05, "learning_rate": 4.965081728135508e-05, "loss": 1.291, "step": 53358 }, { "epoch": 0.05, "learning_rate": 4.96508041989704e-05, "loss": 1.2388, "step": 53359 }, { "epoch": 0.05, "learning_rate": 4.965079111634237e-05, "loss": 0.9936, "step": 53360 }, { "epoch": 0.05, "learning_rate": 4.9650778033471004e-05, "loss": 0.7459, "step": 53361 }, { "epoch": 0.05, "learning_rate": 4.96507649503563e-05, "loss": 1.2244, "step": 53362 }, { "epoch": 0.05, "learning_rate": 4.965075186699825e-05, "loss": 1.4897, "step": 53363 }, { "epoch": 0.05, "learning_rate": 4.965073878339685e-05, "loss": 1.2613, "step": 53364 }, { "epoch": 0.05, "learning_rate": 4.9650725699552115e-05, "loss": 1.5784, "step": 53365 }, { "epoch": 0.05, "learning_rate": 4.965071261546405e-05, "loss": 1.4486, "step": 53366 }, { "epoch": 0.05, "learning_rate": 4.9650699531132626e-05, "loss": 0.9335, "step": 53367 }, { "epoch": 0.05, "learning_rate": 4.965068644655787e-05, "loss": 1.1896, "step": 53368 }, { "epoch": 0.05, "learning_rate": 4.965067336173976e-05, "loss": 1.226, "step": 53369 }, { "epoch": 0.05, "learning_rate": 4.965066027667833e-05, "loss": 1.3734, "step": 53370 }, { "epoch": 0.05, "learning_rate": 4.9650647191373546e-05, "loss": 1.1724, "step": 53371 }, { "epoch": 0.05, "learning_rate": 4.9650634105825425e-05, "loss": 1.3147, "step": 53372 }, { "epoch": 0.05, "learning_rate": 4.965062102003396e-05, "loss": 0.8169, "step": 53373 }, { "epoch": 0.05, "learning_rate": 4.9650607933999156e-05, "loss": 0.4805, "step": 53374 }, { "epoch": 0.05, "learning_rate": 4.965059484772101e-05, "loss": 0.3922, "step": 53375 }, { "epoch": 0.05, "learning_rate": 4.9650581761199524e-05, "loss": 0.5699, "step": 53376 }, { "epoch": 0.05, "learning_rate": 4.96505686744347e-05, "loss": 1.4413, "step": 53377 }, { "epoch": 0.05, "learning_rate": 4.9650555587426536e-05, "loss": 1.4076, "step": 53378 }, { "epoch": 0.05, "learning_rate": 4.965054250017503e-05, "loss": 0.9276, "step": 53379 }, { "epoch": 0.05, "learning_rate": 4.965052941268019e-05, "loss": 1.4123, "step": 53380 }, { "epoch": 0.05, "learning_rate": 4.9650516324942004e-05, "loss": 0.8831, "step": 53381 }, { "epoch": 0.05, "learning_rate": 4.965050323696048e-05, "loss": 1.1193, "step": 53382 }, { "epoch": 0.05, "learning_rate": 4.965049014873562e-05, "loss": 1.4213, "step": 53383 }, { "epoch": 0.05, "learning_rate": 4.9650477060267416e-05, "loss": 1.0988, "step": 53384 }, { "epoch": 0.05, "learning_rate": 4.965046397155587e-05, "loss": 1.5921, "step": 53385 }, { "epoch": 0.05, "learning_rate": 4.9650450882601e-05, "loss": 0.455, "step": 53386 }, { "epoch": 0.05, "learning_rate": 4.965043779340277e-05, "loss": 0.8456, "step": 53387 }, { "epoch": 0.05, "learning_rate": 4.9650424703961216e-05, "loss": 1.3148, "step": 53388 }, { "epoch": 0.05, "learning_rate": 4.965041161427632e-05, "loss": 1.274, "step": 53389 }, { "epoch": 0.05, "learning_rate": 4.9650398524348085e-05, "loss": 1.0808, "step": 53390 }, { "epoch": 0.05, "learning_rate": 4.965038543417652e-05, "loss": 1.5877, "step": 53391 }, { "epoch": 0.05, "learning_rate": 4.96503723437616e-05, "loss": 1.6189, "step": 53392 }, { "epoch": 0.05, "learning_rate": 4.9650359253103354e-05, "loss": 1.3348, "step": 53393 }, { "epoch": 0.05, "learning_rate": 4.965034616220177e-05, "loss": 1.2192, "step": 53394 }, { "epoch": 0.05, "learning_rate": 4.965033307105684e-05, "loss": 1.1583, "step": 53395 }, { "epoch": 0.05, "learning_rate": 4.9650319979668576e-05, "loss": 1.3498, "step": 53396 }, { "epoch": 0.05, "learning_rate": 4.9650306888036976e-05, "loss": 1.1378, "step": 53397 }, { "epoch": 0.05, "learning_rate": 4.965029379616204e-05, "loss": 1.6397, "step": 53398 }, { "epoch": 0.05, "learning_rate": 4.9650280704043764e-05, "loss": 1.3773, "step": 53399 }, { "epoch": 0.05, "learning_rate": 4.965026761168215e-05, "loss": 1.0581, "step": 53400 }, { "epoch": 0.05, "learning_rate": 4.96502545190772e-05, "loss": 0.8439, "step": 53401 }, { "epoch": 0.05, "learning_rate": 4.965024142622892e-05, "loss": 0.9932, "step": 53402 }, { "epoch": 0.05, "learning_rate": 4.965022833313729e-05, "loss": 1.1995, "step": 53403 }, { "epoch": 0.05, "learning_rate": 4.9650215239802334e-05, "loss": 1.1866, "step": 53404 }, { "epoch": 0.05, "learning_rate": 4.9650202146224036e-05, "loss": 0.789, "step": 53405 }, { "epoch": 0.05, "learning_rate": 4.9650189052402396e-05, "loss": 1.2928, "step": 53406 }, { "epoch": 0.05, "learning_rate": 4.965017595833743e-05, "loss": 1.13, "step": 53407 }, { "epoch": 0.05, "learning_rate": 4.965016286402913e-05, "loss": 1.5606, "step": 53408 }, { "epoch": 0.05, "learning_rate": 4.965014976947748e-05, "loss": 1.4316, "step": 53409 }, { "epoch": 0.05, "learning_rate": 4.9650136674682504e-05, "loss": 1.0732, "step": 53410 }, { "epoch": 0.05, "learning_rate": 4.965012357964419e-05, "loss": 0.8635, "step": 53411 }, { "epoch": 0.05, "learning_rate": 4.9650110484362545e-05, "loss": 0.6968, "step": 53412 }, { "epoch": 0.05, "learning_rate": 4.9650097388837555e-05, "loss": 1.2507, "step": 53413 }, { "epoch": 0.05, "learning_rate": 4.965008429306924e-05, "loss": 1.2112, "step": 53414 }, { "epoch": 0.05, "learning_rate": 4.965007119705759e-05, "loss": 0.8854, "step": 53415 }, { "epoch": 0.05, "learning_rate": 4.96500581008026e-05, "loss": 1.2242, "step": 53416 }, { "epoch": 0.05, "learning_rate": 4.9650045004304266e-05, "loss": 1.4097, "step": 53417 }, { "epoch": 0.05, "learning_rate": 4.965003190756261e-05, "loss": 0.8556, "step": 53418 }, { "epoch": 0.05, "learning_rate": 4.9650018810577615e-05, "loss": 1.0677, "step": 53419 }, { "epoch": 0.05, "learning_rate": 4.965000571334929e-05, "loss": 0.8459, "step": 53420 }, { "epoch": 0.05, "learning_rate": 4.964999261587763e-05, "loss": 0.6918, "step": 53421 }, { "epoch": 0.05, "learning_rate": 4.9649979518162624e-05, "loss": 1.1071, "step": 53422 }, { "epoch": 0.05, "learning_rate": 4.96499664202043e-05, "loss": 0.2128, "step": 53423 }, { "epoch": 0.05, "learning_rate": 4.964995332200263e-05, "loss": 0.1865, "step": 53424 }, { "epoch": 0.05, "learning_rate": 4.9649940223557626e-05, "loss": 0.58, "step": 53425 }, { "epoch": 0.05, "learning_rate": 4.964992712486929e-05, "loss": 0.8398, "step": 53426 }, { "epoch": 0.05, "learning_rate": 4.9649914025937625e-05, "loss": 1.4104, "step": 53427 }, { "epoch": 0.05, "learning_rate": 4.964990092676263e-05, "loss": 1.3822, "step": 53428 }, { "epoch": 0.05, "learning_rate": 4.964988782734429e-05, "loss": 0.9715, "step": 53429 }, { "epoch": 0.05, "learning_rate": 4.964987472768262e-05, "loss": 1.3089, "step": 53430 }, { "epoch": 0.05, "learning_rate": 4.964986162777762e-05, "loss": 1.3317, "step": 53431 }, { "epoch": 0.05, "learning_rate": 4.964984852762928e-05, "loss": 1.027, "step": 53432 }, { "epoch": 0.05, "learning_rate": 4.964983542723762e-05, "loss": 1.7388, "step": 53433 }, { "epoch": 0.05, "learning_rate": 4.9649822326602623e-05, "loss": 2.7404, "step": 53434 }, { "epoch": 0.05, "learning_rate": 4.9649809225724284e-05, "loss": 1.2421, "step": 53435 }, { "epoch": 0.05, "learning_rate": 4.964979612460262e-05, "loss": 1.1585, "step": 53436 }, { "epoch": 0.05, "learning_rate": 4.964978302323763e-05, "loss": 1.0153, "step": 53437 }, { "epoch": 0.05, "learning_rate": 4.96497699216293e-05, "loss": 1.0342, "step": 53438 }, { "epoch": 0.05, "learning_rate": 4.9649756819777636e-05, "loss": 1.2083, "step": 53439 }, { "epoch": 0.05, "learning_rate": 4.964974371768264e-05, "loss": 1.0403, "step": 53440 }, { "epoch": 0.05, "learning_rate": 4.964973061534431e-05, "loss": 1.0746, "step": 53441 }, { "epoch": 0.05, "learning_rate": 4.9649717512762657e-05, "loss": 1.1292, "step": 53442 }, { "epoch": 0.05, "learning_rate": 4.964970440993767e-05, "loss": 1.2352, "step": 53443 }, { "epoch": 0.05, "learning_rate": 4.964969130686935e-05, "loss": 0.9307, "step": 53444 }, { "epoch": 0.05, "learning_rate": 4.964967820355769e-05, "loss": 1.1609, "step": 53445 }, { "epoch": 0.05, "learning_rate": 4.964966510000271e-05, "loss": 1.0361, "step": 53446 }, { "epoch": 0.05, "learning_rate": 4.964965199620439e-05, "loss": 1.0579, "step": 53447 }, { "epoch": 0.05, "learning_rate": 4.964963889216275e-05, "loss": 1.0572, "step": 53448 }, { "epoch": 0.05, "learning_rate": 4.9649625787877774e-05, "loss": 0.9948, "step": 53449 }, { "epoch": 0.05, "learning_rate": 4.964961268334947e-05, "loss": 0.9593, "step": 53450 }, { "epoch": 0.05, "learning_rate": 4.964959957857783e-05, "loss": 0.9208, "step": 53451 }, { "epoch": 0.05, "learning_rate": 4.964958647356287e-05, "loss": 1.5734, "step": 53452 }, { "epoch": 0.05, "learning_rate": 4.964957336830456e-05, "loss": 1.2935, "step": 53453 }, { "epoch": 0.05, "learning_rate": 4.964956026280293e-05, "loss": 1.0484, "step": 53454 }, { "epoch": 0.05, "learning_rate": 4.9649547157057975e-05, "loss": 1.0802, "step": 53455 }, { "epoch": 0.05, "learning_rate": 4.964953405106969e-05, "loss": 0.7419, "step": 53456 }, { "epoch": 0.05, "learning_rate": 4.964952094483807e-05, "loss": 1.2038, "step": 53457 }, { "epoch": 0.05, "learning_rate": 4.964950783836312e-05, "loss": 1.0615, "step": 53458 }, { "epoch": 0.05, "learning_rate": 4.9649494731644844e-05, "loss": 1.4757, "step": 53459 }, { "epoch": 0.05, "learning_rate": 4.9649481624683235e-05, "loss": 0.6666, "step": 53460 }, { "epoch": 0.05, "learning_rate": 4.9649468517478305e-05, "loss": 1.2464, "step": 53461 }, { "epoch": 0.05, "learning_rate": 4.964945541003004e-05, "loss": 0.8573, "step": 53462 }, { "epoch": 0.05, "learning_rate": 4.9649442302338444e-05, "loss": 0.8307, "step": 53463 }, { "epoch": 0.05, "learning_rate": 4.964942919440352e-05, "loss": 1.2441, "step": 53464 }, { "epoch": 0.05, "learning_rate": 4.964941608622527e-05, "loss": 1.3275, "step": 53465 }, { "epoch": 0.05, "learning_rate": 4.9649402977803685e-05, "loss": 1.1877, "step": 53466 }, { "epoch": 0.05, "learning_rate": 4.9649389869138775e-05, "loss": 1.2296, "step": 53467 }, { "epoch": 0.05, "learning_rate": 4.9649376760230536e-05, "loss": 1.1287, "step": 53468 }, { "epoch": 0.05, "learning_rate": 4.9649363651078975e-05, "loss": 1.1117, "step": 53469 }, { "epoch": 0.05, "learning_rate": 4.964935054168408e-05, "loss": 1.3034, "step": 53470 }, { "epoch": 0.05, "learning_rate": 4.964933743204586e-05, "loss": 1.117, "step": 53471 }, { "epoch": 0.05, "learning_rate": 4.9649324322164304e-05, "loss": 1.5013, "step": 53472 }, { "epoch": 0.05, "learning_rate": 4.964931121203943e-05, "loss": 1.1563, "step": 53473 }, { "epoch": 0.05, "learning_rate": 4.964929810167123e-05, "loss": 1.2608, "step": 53474 }, { "epoch": 0.05, "learning_rate": 4.9649284991059694e-05, "loss": 1.2248, "step": 53475 }, { "epoch": 0.05, "learning_rate": 4.964927188020483e-05, "loss": 1.3571, "step": 53476 }, { "epoch": 0.05, "learning_rate": 4.9649258769106645e-05, "loss": 1.1941, "step": 53477 }, { "epoch": 0.05, "learning_rate": 4.964924565776513e-05, "loss": 1.0989, "step": 53478 }, { "epoch": 0.05, "learning_rate": 4.964923254618029e-05, "loss": 1.1462, "step": 53479 }, { "epoch": 0.05, "learning_rate": 4.9649219434352116e-05, "loss": 1.3468, "step": 53480 }, { "epoch": 0.05, "learning_rate": 4.964920632228062e-05, "loss": 1.4275, "step": 53481 }, { "epoch": 0.05, "learning_rate": 4.9649193209965806e-05, "loss": 1.24, "step": 53482 }, { "epoch": 0.05, "learning_rate": 4.9649180097407655e-05, "loss": 1.1106, "step": 53483 }, { "epoch": 0.05, "learning_rate": 4.9649166984606174e-05, "loss": 0.8835, "step": 53484 }, { "epoch": 0.05, "learning_rate": 4.9649153871561385e-05, "loss": 1.1966, "step": 53485 }, { "epoch": 0.05, "learning_rate": 4.964914075827325e-05, "loss": 1.0938, "step": 53486 }, { "epoch": 0.05, "learning_rate": 4.96491276447418e-05, "loss": 1.4593, "step": 53487 }, { "epoch": 0.05, "learning_rate": 4.9649114530967025e-05, "loss": 1.0211, "step": 53488 }, { "epoch": 0.05, "learning_rate": 4.964910141694892e-05, "loss": 1.1989, "step": 53489 }, { "epoch": 0.05, "learning_rate": 4.964908830268749e-05, "loss": 0.9258, "step": 53490 }, { "epoch": 0.05, "learning_rate": 4.964907518818274e-05, "loss": 1.0004, "step": 53491 }, { "epoch": 0.05, "learning_rate": 4.964906207343465e-05, "loss": 1.0698, "step": 53492 }, { "epoch": 0.05, "learning_rate": 4.964904895844325e-05, "loss": 1.5022, "step": 53493 }, { "epoch": 0.05, "learning_rate": 4.964903584320851e-05, "loss": 1.1452, "step": 53494 }, { "epoch": 0.05, "learning_rate": 4.964902272773046e-05, "loss": 1.3497, "step": 53495 }, { "epoch": 0.05, "learning_rate": 4.964900961200908e-05, "loss": 1.2659, "step": 53496 }, { "epoch": 0.05, "learning_rate": 4.9648996496044375e-05, "loss": 0.6762, "step": 53497 }, { "epoch": 0.05, "learning_rate": 4.9648983379836345e-05, "loss": 1.0466, "step": 53498 }, { "epoch": 0.05, "learning_rate": 4.9648970263385e-05, "loss": 1.4191, "step": 53499 }, { "epoch": 0.05, "learning_rate": 4.964895714669032e-05, "loss": 1.4758, "step": 53500 }, { "epoch": 0.05, "eval_loss": 1.0732150077819824, "eval_runtime": 128.3448, "eval_samples_per_second": 10.791, "eval_steps_per_second": 5.4, "step": 53500 }, { "epoch": 0.05, "learning_rate": 4.9648944029752315e-05, "loss": 0.9445, "step": 53501 }, { "epoch": 0.05, "learning_rate": 4.964893091257099e-05, "loss": 0.7266, "step": 53502 }, { "epoch": 0.05, "learning_rate": 4.9648917795146336e-05, "loss": 0.8277, "step": 53503 }, { "epoch": 0.05, "learning_rate": 4.964890467747837e-05, "loss": 1.3445, "step": 53504 }, { "epoch": 0.05, "learning_rate": 4.964889155956707e-05, "loss": 1.1832, "step": 53505 }, { "epoch": 0.05, "learning_rate": 4.964887844141245e-05, "loss": 0.9986, "step": 53506 }, { "epoch": 0.05, "learning_rate": 4.9648865323014506e-05, "loss": 0.8222, "step": 53507 }, { "epoch": 0.05, "learning_rate": 4.9648852204373235e-05, "loss": 1.1422, "step": 53508 }, { "epoch": 0.05, "learning_rate": 4.964883908548865e-05, "loss": 1.1903, "step": 53509 }, { "epoch": 0.05, "learning_rate": 4.964882596636073e-05, "loss": 0.7375, "step": 53510 }, { "epoch": 0.05, "learning_rate": 4.9648812846989495e-05, "loss": 2.1387, "step": 53511 }, { "epoch": 0.05, "learning_rate": 4.964879972737494e-05, "loss": 1.1294, "step": 53512 }, { "epoch": 0.05, "learning_rate": 4.9648786607517054e-05, "loss": 1.2348, "step": 53513 }, { "epoch": 0.05, "learning_rate": 4.964877348741586e-05, "loss": 1.0015, "step": 53514 }, { "epoch": 0.05, "learning_rate": 4.9648760367071325e-05, "loss": 1.2099, "step": 53515 }, { "epoch": 0.05, "learning_rate": 4.964874724648348e-05, "loss": 0.9007, "step": 53516 }, { "epoch": 0.05, "learning_rate": 4.964873412565231e-05, "loss": 0.9535, "step": 53517 }, { "epoch": 0.05, "learning_rate": 4.9648721004577815e-05, "loss": 0.8274, "step": 53518 }, { "epoch": 0.05, "learning_rate": 4.964870788326e-05, "loss": 1.3481, "step": 53519 }, { "epoch": 0.05, "learning_rate": 4.9648694761698865e-05, "loss": 1.2123, "step": 53520 }, { "epoch": 0.05, "learning_rate": 4.9648681639894414e-05, "loss": 1.4581, "step": 53521 }, { "epoch": 0.05, "learning_rate": 4.9648668517846634e-05, "loss": 1.3828, "step": 53522 }, { "epoch": 0.05, "learning_rate": 4.964865539555553e-05, "loss": 1.3851, "step": 53523 }, { "epoch": 0.05, "learning_rate": 4.964864227302111e-05, "loss": 1.1952, "step": 53524 }, { "epoch": 0.05, "learning_rate": 4.964862915024337e-05, "loss": 1.2249, "step": 53525 }, { "epoch": 0.05, "learning_rate": 4.9648616027222305e-05, "loss": 1.3056, "step": 53526 }, { "epoch": 0.05, "learning_rate": 4.964860290395792e-05, "loss": 0.9062, "step": 53527 }, { "epoch": 0.05, "learning_rate": 4.9648589780450216e-05, "loss": 0.8244, "step": 53528 }, { "epoch": 0.05, "learning_rate": 4.9648576656699195e-05, "loss": 0.9901, "step": 53529 }, { "epoch": 0.05, "learning_rate": 4.964856353270485e-05, "loss": 1.0981, "step": 53530 }, { "epoch": 0.05, "learning_rate": 4.964855040846718e-05, "loss": 1.0267, "step": 53531 }, { "epoch": 0.05, "learning_rate": 4.96485372839862e-05, "loss": 1.1889, "step": 53532 }, { "epoch": 0.05, "learning_rate": 4.964852415926189e-05, "loss": 1.1327, "step": 53533 }, { "epoch": 0.05, "learning_rate": 4.9648511034294264e-05, "loss": 1.0918, "step": 53534 }, { "epoch": 0.05, "learning_rate": 4.9648497909083324e-05, "loss": 0.8724, "step": 53535 }, { "epoch": 0.05, "learning_rate": 4.9648484783629055e-05, "loss": 0.9584, "step": 53536 }, { "epoch": 0.05, "learning_rate": 4.964847165793147e-05, "loss": 1.135, "step": 53537 }, { "epoch": 0.05, "learning_rate": 4.964845853199057e-05, "loss": 1.3751, "step": 53538 }, { "epoch": 0.05, "learning_rate": 4.964844540580634e-05, "loss": 1.0032, "step": 53539 }, { "epoch": 0.05, "learning_rate": 4.9648432279378806e-05, "loss": 1.201, "step": 53540 }, { "epoch": 0.05, "learning_rate": 4.964841915270795e-05, "loss": 1.1408, "step": 53541 }, { "epoch": 0.05, "learning_rate": 4.9648406025793766e-05, "loss": 1.2378, "step": 53542 }, { "epoch": 0.05, "learning_rate": 4.964839289863626e-05, "loss": 1.1345, "step": 53543 }, { "epoch": 0.05, "learning_rate": 4.964837977123545e-05, "loss": 0.7726, "step": 53544 }, { "epoch": 0.05, "learning_rate": 4.964836664359132e-05, "loss": 1.0088, "step": 53545 }, { "epoch": 0.05, "learning_rate": 4.964835351570386e-05, "loss": 0.6479, "step": 53546 }, { "epoch": 0.05, "learning_rate": 4.964834038757309e-05, "loss": 1.221, "step": 53547 }, { "epoch": 0.05, "learning_rate": 4.9648327259199e-05, "loss": 1.1097, "step": 53548 }, { "epoch": 0.05, "learning_rate": 4.9648314130581596e-05, "loss": 1.4635, "step": 53549 }, { "epoch": 0.05, "learning_rate": 4.9648301001720866e-05, "loss": 0.9433, "step": 53550 }, { "epoch": 0.05, "learning_rate": 4.964828787261683e-05, "loss": 1.1199, "step": 53551 }, { "epoch": 0.05, "learning_rate": 4.9648274743269466e-05, "loss": 0.9827, "step": 53552 }, { "epoch": 0.05, "learning_rate": 4.964826161367879e-05, "loss": 1.2483, "step": 53553 }, { "epoch": 0.05, "learning_rate": 4.96482484838448e-05, "loss": 0.9408, "step": 53554 }, { "epoch": 0.05, "learning_rate": 4.9648235353767484e-05, "loss": 0.8752, "step": 53555 }, { "epoch": 0.05, "learning_rate": 4.964822222344685e-05, "loss": 0.8584, "step": 53556 }, { "epoch": 0.05, "learning_rate": 4.9648209092882904e-05, "loss": 0.932, "step": 53557 }, { "epoch": 0.05, "learning_rate": 4.9648195962075644e-05, "loss": 1.1189, "step": 53558 }, { "epoch": 0.05, "learning_rate": 4.964818283102506e-05, "loss": 2.0638, "step": 53559 }, { "epoch": 0.05, "learning_rate": 4.964816969973117e-05, "loss": 1.43, "step": 53560 }, { "epoch": 0.05, "learning_rate": 4.964815656819396e-05, "loss": 1.4726, "step": 53561 }, { "epoch": 0.05, "learning_rate": 4.964814343641343e-05, "loss": 1.2545, "step": 53562 }, { "epoch": 0.05, "learning_rate": 4.964813030438958e-05, "loss": 1.4676, "step": 53563 }, { "epoch": 0.05, "learning_rate": 4.9648117172122424e-05, "loss": 1.112, "step": 53564 }, { "epoch": 0.05, "learning_rate": 4.964810403961195e-05, "loss": 1.1126, "step": 53565 }, { "epoch": 0.05, "learning_rate": 4.9648090906858155e-05, "loss": 1.0902, "step": 53566 }, { "epoch": 0.05, "learning_rate": 4.9648077773861044e-05, "loss": 1.2213, "step": 53567 }, { "epoch": 0.05, "learning_rate": 4.964806464062063e-05, "loss": 1.2061, "step": 53568 }, { "epoch": 0.05, "learning_rate": 4.964805150713688e-05, "loss": 1.3248, "step": 53569 }, { "epoch": 0.05, "learning_rate": 4.964803837340983e-05, "loss": 0.7229, "step": 53570 }, { "epoch": 0.05, "learning_rate": 4.964802523943947e-05, "loss": 0.9528, "step": 53571 }, { "epoch": 0.05, "learning_rate": 4.964801210522578e-05, "loss": 1.0722, "step": 53572 }, { "epoch": 0.05, "learning_rate": 4.9647998970768785e-05, "loss": 1.1501, "step": 53573 }, { "epoch": 0.05, "learning_rate": 4.9647985836068474e-05, "loss": 1.2962, "step": 53574 }, { "epoch": 0.05, "learning_rate": 4.964797270112485e-05, "loss": 1.226, "step": 53575 }, { "epoch": 0.05, "learning_rate": 4.96479595659379e-05, "loss": 1.2196, "step": 53576 }, { "epoch": 0.05, "learning_rate": 4.964794643050765e-05, "loss": 0.9531, "step": 53577 }, { "epoch": 0.05, "learning_rate": 4.9647933294834076e-05, "loss": 1.1579, "step": 53578 }, { "epoch": 0.05, "learning_rate": 4.9647920158917195e-05, "loss": 0.7328, "step": 53579 }, { "epoch": 0.05, "learning_rate": 4.964790702275699e-05, "loss": 0.8047, "step": 53580 }, { "epoch": 0.05, "learning_rate": 4.964789388635348e-05, "loss": 1.1028, "step": 53581 }, { "epoch": 0.05, "learning_rate": 4.9647880749706654e-05, "loss": 0.9733, "step": 53582 }, { "epoch": 0.05, "learning_rate": 4.964786761281652e-05, "loss": 1.0767, "step": 53583 }, { "epoch": 0.05, "learning_rate": 4.964785447568306e-05, "loss": 1.1157, "step": 53584 }, { "epoch": 0.05, "learning_rate": 4.96478413383063e-05, "loss": 0.7339, "step": 53585 }, { "epoch": 0.05, "learning_rate": 4.964782820068622e-05, "loss": 1.0277, "step": 53586 }, { "epoch": 0.05, "learning_rate": 4.964781506282283e-05, "loss": 1.3039, "step": 53587 }, { "epoch": 0.05, "learning_rate": 4.964780192471613e-05, "loss": 1.1862, "step": 53588 }, { "epoch": 0.05, "learning_rate": 4.964778878636611e-05, "loss": 1.1229, "step": 53589 }, { "epoch": 0.05, "learning_rate": 4.964777564777278e-05, "loss": 1.2639, "step": 53590 }, { "epoch": 0.05, "learning_rate": 4.964776250893613e-05, "loss": 1.1428, "step": 53591 }, { "epoch": 0.05, "learning_rate": 4.964774936985618e-05, "loss": 1.2236, "step": 53592 }, { "epoch": 0.05, "learning_rate": 4.964773623053291e-05, "loss": 1.423, "step": 53593 }, { "epoch": 0.05, "learning_rate": 4.9647723090966336e-05, "loss": 1.2869, "step": 53594 }, { "epoch": 0.05, "learning_rate": 4.964770995115645e-05, "loss": 1.5588, "step": 53595 }, { "epoch": 0.05, "learning_rate": 4.9647696811103244e-05, "loss": 1.134, "step": 53596 }, { "epoch": 0.05, "learning_rate": 4.9647683670806735e-05, "loss": 1.0572, "step": 53597 }, { "epoch": 0.05, "learning_rate": 4.9647670530266905e-05, "loss": 2.0288, "step": 53598 }, { "epoch": 0.05, "learning_rate": 4.964765738948377e-05, "loss": 1.1733, "step": 53599 }, { "epoch": 0.05, "learning_rate": 4.964764424845732e-05, "loss": 1.2516, "step": 53600 }, { "epoch": 0.05, "learning_rate": 4.964763110718757e-05, "loss": 0.8472, "step": 53601 }, { "epoch": 0.05, "learning_rate": 4.964761796567449e-05, "loss": 0.932, "step": 53602 }, { "epoch": 0.05, "learning_rate": 4.964760482391811e-05, "loss": 0.8923, "step": 53603 }, { "epoch": 0.05, "learning_rate": 4.9647591681918414e-05, "loss": 1.4513, "step": 53604 }, { "epoch": 0.05, "learning_rate": 4.964757853967541e-05, "loss": 1.3391, "step": 53605 }, { "epoch": 0.05, "learning_rate": 4.96475653971891e-05, "loss": 1.0197, "step": 53606 }, { "epoch": 0.05, "learning_rate": 4.9647552254459476e-05, "loss": 0.7182, "step": 53607 }, { "epoch": 0.05, "learning_rate": 4.964753911148654e-05, "loss": 0.9987, "step": 53608 }, { "epoch": 0.05, "learning_rate": 4.9647525968270295e-05, "loss": 0.8713, "step": 53609 }, { "epoch": 0.05, "learning_rate": 4.964751282481074e-05, "loss": 1.5749, "step": 53610 }, { "epoch": 0.05, "learning_rate": 4.964749968110788e-05, "loss": 1.1103, "step": 53611 }, { "epoch": 0.05, "learning_rate": 4.96474865371617e-05, "loss": 1.1516, "step": 53612 }, { "epoch": 0.05, "learning_rate": 4.964747339297222e-05, "loss": 1.5528, "step": 53613 }, { "epoch": 0.05, "learning_rate": 4.964746024853943e-05, "loss": 1.519, "step": 53614 }, { "epoch": 0.05, "learning_rate": 4.964744710386333e-05, "loss": 0.6871, "step": 53615 }, { "epoch": 0.05, "learning_rate": 4.9647433958943915e-05, "loss": 1.1476, "step": 53616 }, { "epoch": 0.05, "learning_rate": 4.9647420813781195e-05, "loss": 1.4705, "step": 53617 }, { "epoch": 0.05, "learning_rate": 4.9647407668375166e-05, "loss": 1.3695, "step": 53618 }, { "epoch": 0.05, "learning_rate": 4.964739452272583e-05, "loss": 1.4015, "step": 53619 }, { "epoch": 0.05, "learning_rate": 4.964738137683318e-05, "loss": 1.0395, "step": 53620 }, { "epoch": 0.05, "learning_rate": 4.964736823069723e-05, "loss": 0.4603, "step": 53621 }, { "epoch": 0.05, "learning_rate": 4.964735508431796e-05, "loss": 0.5323, "step": 53622 }, { "epoch": 0.05, "learning_rate": 4.964734193769539e-05, "loss": 0.6964, "step": 53623 }, { "epoch": 0.05, "learning_rate": 4.964732879082951e-05, "loss": 1.6326, "step": 53624 }, { "epoch": 0.05, "learning_rate": 4.964731564372032e-05, "loss": 1.4376, "step": 53625 }, { "epoch": 0.05, "learning_rate": 4.9647302496367826e-05, "loss": 1.2346, "step": 53626 }, { "epoch": 0.05, "learning_rate": 4.964728934877202e-05, "loss": 0.8531, "step": 53627 }, { "epoch": 0.05, "learning_rate": 4.9647276200932914e-05, "loss": 0.7909, "step": 53628 }, { "epoch": 0.05, "learning_rate": 4.964726305285049e-05, "loss": 1.1049, "step": 53629 }, { "epoch": 0.05, "learning_rate": 4.964724990452476e-05, "loss": 1.3088, "step": 53630 }, { "epoch": 0.05, "learning_rate": 4.964723675595573e-05, "loss": 1.1045, "step": 53631 }, { "epoch": 0.05, "learning_rate": 4.964722360714339e-05, "loss": 1.3116, "step": 53632 }, { "epoch": 0.05, "learning_rate": 4.964721045808774e-05, "loss": 1.4276, "step": 53633 }, { "epoch": 0.05, "learning_rate": 4.964719730878878e-05, "loss": 1.2076, "step": 53634 }, { "epoch": 0.05, "learning_rate": 4.964718415924652e-05, "loss": 1.1472, "step": 53635 }, { "epoch": 0.05, "learning_rate": 4.964717100946096e-05, "loss": 1.41, "step": 53636 }, { "epoch": 0.05, "learning_rate": 4.9647157859432084e-05, "loss": 1.1339, "step": 53637 }, { "epoch": 0.05, "learning_rate": 4.96471447091599e-05, "loss": 1.2681, "step": 53638 }, { "epoch": 0.05, "learning_rate": 4.964713155864441e-05, "loss": 0.8322, "step": 53639 }, { "epoch": 0.05, "learning_rate": 4.9647118407885616e-05, "loss": 1.1735, "step": 53640 }, { "epoch": 0.05, "learning_rate": 4.964710525688352e-05, "loss": 0.9869, "step": 53641 }, { "epoch": 0.05, "learning_rate": 4.964709210563811e-05, "loss": 1.2504, "step": 53642 }, { "epoch": 0.05, "learning_rate": 4.9647078954149406e-05, "loss": 1.4052, "step": 53643 }, { "epoch": 0.05, "learning_rate": 4.964706580241739e-05, "loss": 1.2207, "step": 53644 }, { "epoch": 0.05, "learning_rate": 4.964705265044206e-05, "loss": 1.1052, "step": 53645 }, { "epoch": 0.05, "learning_rate": 4.964703949822344e-05, "loss": 0.899, "step": 53646 }, { "epoch": 0.05, "learning_rate": 4.9647026345761504e-05, "loss": 1.0529, "step": 53647 }, { "epoch": 0.05, "learning_rate": 4.964701319305627e-05, "loss": 0.9516, "step": 53648 }, { "epoch": 0.05, "learning_rate": 4.964700004010773e-05, "loss": 1.0315, "step": 53649 }, { "epoch": 0.05, "learning_rate": 4.964698688691588e-05, "loss": 1.1384, "step": 53650 }, { "epoch": 0.05, "learning_rate": 4.964697373348073e-05, "loss": 0.6509, "step": 53651 }, { "epoch": 0.05, "learning_rate": 4.964696057980227e-05, "loss": 1.264, "step": 53652 }, { "epoch": 0.05, "learning_rate": 4.9646947425880505e-05, "loss": 1.3241, "step": 53653 }, { "epoch": 0.05, "learning_rate": 4.964693427171544e-05, "loss": 1.1309, "step": 53654 }, { "epoch": 0.05, "learning_rate": 4.9646921117307075e-05, "loss": 0.9645, "step": 53655 }, { "epoch": 0.05, "learning_rate": 4.96469079626554e-05, "loss": 1.261, "step": 53656 }, { "epoch": 0.05, "learning_rate": 4.9646894807760425e-05, "loss": 1.2591, "step": 53657 }, { "epoch": 0.05, "learning_rate": 4.9646881652622144e-05, "loss": 1.1691, "step": 53658 }, { "epoch": 0.05, "learning_rate": 4.9646868497240554e-05, "loss": 0.9919, "step": 53659 }, { "epoch": 0.05, "learning_rate": 4.964685534161567e-05, "loss": 0.9989, "step": 53660 }, { "epoch": 0.05, "learning_rate": 4.9646842185747477e-05, "loss": 1.0905, "step": 53661 }, { "epoch": 0.05, "learning_rate": 4.964682902963598e-05, "loss": 1.2065, "step": 53662 }, { "epoch": 0.05, "learning_rate": 4.9646815873281185e-05, "loss": 1.1167, "step": 53663 }, { "epoch": 0.05, "learning_rate": 4.964680271668308e-05, "loss": 1.2462, "step": 53664 }, { "epoch": 0.05, "learning_rate": 4.964678955984168e-05, "loss": 0.8905, "step": 53665 }, { "epoch": 0.05, "learning_rate": 4.964677640275698e-05, "loss": 0.4813, "step": 53666 }, { "epoch": 0.05, "learning_rate": 4.964676324542896e-05, "loss": 0.5339, "step": 53667 }, { "epoch": 0.05, "learning_rate": 4.964675008785765e-05, "loss": 0.5664, "step": 53668 }, { "epoch": 0.05, "learning_rate": 4.9646736930043035e-05, "loss": 0.6133, "step": 53669 }, { "epoch": 0.05, "learning_rate": 4.9646723771985127e-05, "loss": 1.2583, "step": 53670 }, { "epoch": 0.05, "learning_rate": 4.96467106136839e-05, "loss": 1.2892, "step": 53671 }, { "epoch": 0.05, "learning_rate": 4.9646697455139383e-05, "loss": 1.1444, "step": 53672 }, { "epoch": 0.05, "learning_rate": 4.964668429635156e-05, "loss": 1.1628, "step": 53673 }, { "epoch": 0.05, "learning_rate": 4.964667113732045e-05, "loss": 1.1409, "step": 53674 }, { "epoch": 0.05, "learning_rate": 4.9646657978046016e-05, "loss": 1.6138, "step": 53675 }, { "epoch": 0.05, "learning_rate": 4.964664481852829e-05, "loss": 1.5815, "step": 53676 }, { "epoch": 0.05, "learning_rate": 4.964663165876726e-05, "loss": 1.236, "step": 53677 }, { "epoch": 0.05, "learning_rate": 4.964661849876293e-05, "loss": 1.305, "step": 53678 }, { "epoch": 0.05, "learning_rate": 4.964660533851531e-05, "loss": 0.8162, "step": 53679 }, { "epoch": 0.05, "learning_rate": 4.964659217802437e-05, "loss": 1.1754, "step": 53680 }, { "epoch": 0.05, "learning_rate": 4.964657901729014e-05, "loss": 1.6576, "step": 53681 }, { "epoch": 0.05, "learning_rate": 4.964656585631261e-05, "loss": 1.5637, "step": 53682 }, { "epoch": 0.05, "learning_rate": 4.964655269509177e-05, "loss": 1.5125, "step": 53683 }, { "epoch": 0.05, "learning_rate": 4.9646539533627646e-05, "loss": 1.5124, "step": 53684 }, { "epoch": 0.05, "learning_rate": 4.964652637192021e-05, "loss": 1.4581, "step": 53685 }, { "epoch": 0.05, "learning_rate": 4.9646513209969475e-05, "loss": 1.3815, "step": 53686 }, { "epoch": 0.05, "learning_rate": 4.964650004777544e-05, "loss": 1.3752, "step": 53687 }, { "epoch": 0.05, "learning_rate": 4.96464868853381e-05, "loss": 1.4857, "step": 53688 }, { "epoch": 0.05, "learning_rate": 4.9646473722657475e-05, "loss": 1.4073, "step": 53689 }, { "epoch": 0.05, "learning_rate": 4.9646460559733544e-05, "loss": 1.3232, "step": 53690 }, { "epoch": 0.05, "learning_rate": 4.9646447396566306e-05, "loss": 1.2923, "step": 53691 }, { "epoch": 0.05, "learning_rate": 4.964643423315578e-05, "loss": 1.2166, "step": 53692 }, { "epoch": 0.05, "learning_rate": 4.964642106950195e-05, "loss": 1.1668, "step": 53693 }, { "epoch": 0.05, "learning_rate": 4.964640790560482e-05, "loss": 1.4523, "step": 53694 }, { "epoch": 0.05, "learning_rate": 4.964639474146439e-05, "loss": 1.7908, "step": 53695 }, { "epoch": 0.05, "learning_rate": 4.964638157708066e-05, "loss": 1.8521, "step": 53696 }, { "epoch": 0.05, "learning_rate": 4.964636841245364e-05, "loss": 1.2084, "step": 53697 }, { "epoch": 0.05, "learning_rate": 4.964635524758331e-05, "loss": 1.0226, "step": 53698 }, { "epoch": 0.05, "learning_rate": 4.964634208246969e-05, "loss": 1.2601, "step": 53699 }, { "epoch": 0.05, "learning_rate": 4.964632891711276e-05, "loss": 1.1726, "step": 53700 }, { "epoch": 0.05, "learning_rate": 4.964631575151255e-05, "loss": 1.0158, "step": 53701 }, { "epoch": 0.05, "learning_rate": 4.9646302585669024e-05, "loss": 1.102, "step": 53702 }, { "epoch": 0.05, "learning_rate": 4.964628941958221e-05, "loss": 0.9349, "step": 53703 }, { "epoch": 0.05, "learning_rate": 4.96462762532521e-05, "loss": 0.8835, "step": 53704 }, { "epoch": 0.05, "learning_rate": 4.964626308667868e-05, "loss": 0.847, "step": 53705 }, { "epoch": 0.05, "learning_rate": 4.964624991986198e-05, "loss": 0.7605, "step": 53706 }, { "epoch": 0.05, "learning_rate": 4.964623675280197e-05, "loss": 0.852, "step": 53707 }, { "epoch": 0.05, "learning_rate": 4.964622358549867e-05, "loss": 0.6911, "step": 53708 }, { "epoch": 0.05, "learning_rate": 4.964621041795206e-05, "loss": 1.4987, "step": 53709 }, { "epoch": 0.05, "learning_rate": 4.964619725016217e-05, "loss": 1.1238, "step": 53710 }, { "epoch": 0.05, "learning_rate": 4.9646184082128974e-05, "loss": 1.1569, "step": 53711 }, { "epoch": 0.05, "learning_rate": 4.964617091385249e-05, "loss": 1.1149, "step": 53712 }, { "epoch": 0.05, "learning_rate": 4.96461577453327e-05, "loss": 0.7421, "step": 53713 }, { "epoch": 0.05, "learning_rate": 4.964614457656961e-05, "loss": 1.0674, "step": 53714 }, { "epoch": 0.05, "learning_rate": 4.9646131407563235e-05, "loss": 1.4945, "step": 53715 }, { "epoch": 0.05, "learning_rate": 4.9646118238313555e-05, "loss": 1.1999, "step": 53716 }, { "epoch": 0.05, "learning_rate": 4.964610506882058e-05, "loss": 1.246, "step": 53717 }, { "epoch": 0.05, "learning_rate": 4.9646091899084315e-05, "loss": 1.3247, "step": 53718 }, { "epoch": 0.05, "learning_rate": 4.964607872910475e-05, "loss": 1.2337, "step": 53719 }, { "epoch": 0.05, "learning_rate": 4.964606555888189e-05, "loss": 0.9453, "step": 53720 }, { "epoch": 0.05, "learning_rate": 4.9646052388415734e-05, "loss": 1.1312, "step": 53721 }, { "epoch": 0.05, "learning_rate": 4.9646039217706284e-05, "loss": 1.1009, "step": 53722 }, { "epoch": 0.05, "learning_rate": 4.964602604675354e-05, "loss": 1.16, "step": 53723 }, { "epoch": 0.05, "learning_rate": 4.964601287555749e-05, "loss": 0.6375, "step": 53724 }, { "epoch": 0.05, "learning_rate": 4.964599970411815e-05, "loss": 0.8474, "step": 53725 }, { "epoch": 0.05, "learning_rate": 4.9645986532435526e-05, "loss": 1.5428, "step": 53726 }, { "epoch": 0.05, "learning_rate": 4.9645973360509594e-05, "loss": 1.5092, "step": 53727 }, { "epoch": 0.05, "learning_rate": 4.964596018834038e-05, "loss": 1.417, "step": 53728 }, { "epoch": 0.05, "learning_rate": 4.9645947015927865e-05, "loss": 1.26, "step": 53729 }, { "epoch": 0.05, "learning_rate": 4.9645933843272055e-05, "loss": 1.3668, "step": 53730 }, { "epoch": 0.05, "learning_rate": 4.964592067037295e-05, "loss": 1.3591, "step": 53731 }, { "epoch": 0.05, "learning_rate": 4.964590749723055e-05, "loss": 1.2914, "step": 53732 }, { "epoch": 0.05, "learning_rate": 4.9645894323844855e-05, "loss": 1.2059, "step": 53733 }, { "epoch": 0.05, "learning_rate": 4.9645881150215865e-05, "loss": 1.2937, "step": 53734 }, { "epoch": 0.05, "learning_rate": 4.964586797634359e-05, "loss": 1.2522, "step": 53735 }, { "epoch": 0.05, "learning_rate": 4.964585480222801e-05, "loss": 1.2714, "step": 53736 }, { "epoch": 0.05, "learning_rate": 4.9645841627869146e-05, "loss": 1.1234, "step": 53737 }, { "epoch": 0.05, "learning_rate": 4.964582845326698e-05, "loss": 1.2199, "step": 53738 }, { "epoch": 0.05, "learning_rate": 4.9645815278421525e-05, "loss": 1.2759, "step": 53739 }, { "epoch": 0.05, "learning_rate": 4.9645802103332785e-05, "loss": 1.0915, "step": 53740 }, { "epoch": 0.05, "learning_rate": 4.9645788928000744e-05, "loss": 1.2433, "step": 53741 }, { "epoch": 0.05, "learning_rate": 4.964577575242541e-05, "loss": 1.2447, "step": 53742 }, { "epoch": 0.05, "learning_rate": 4.964576257660678e-05, "loss": 1.2759, "step": 53743 }, { "epoch": 0.05, "learning_rate": 4.9645749400544864e-05, "loss": 1.2502, "step": 53744 }, { "epoch": 0.05, "learning_rate": 4.964573622423966e-05, "loss": 1.2475, "step": 53745 }, { "epoch": 0.05, "learning_rate": 4.964572304769115e-05, "loss": 1.2174, "step": 53746 }, { "epoch": 0.05, "learning_rate": 4.964570987089936e-05, "loss": 1.1893, "step": 53747 }, { "epoch": 0.05, "learning_rate": 4.964569669386427e-05, "loss": 1.2181, "step": 53748 }, { "epoch": 0.05, "learning_rate": 4.964568351658589e-05, "loss": 1.2078, "step": 53749 }, { "epoch": 0.05, "learning_rate": 4.964567033906422e-05, "loss": 1.1958, "step": 53750 }, { "epoch": 0.05, "learning_rate": 4.964565716129925e-05, "loss": 1.2544, "step": 53751 }, { "epoch": 0.05, "learning_rate": 4.9645643983291e-05, "loss": 1.2431, "step": 53752 }, { "epoch": 0.05, "learning_rate": 4.964563080503946e-05, "loss": 1.1869, "step": 53753 }, { "epoch": 0.05, "learning_rate": 4.964561762654462e-05, "loss": 1.2166, "step": 53754 }, { "epoch": 0.05, "learning_rate": 4.9645604447806494e-05, "loss": 1.1412, "step": 53755 }, { "epoch": 0.05, "learning_rate": 4.9645591268825076e-05, "loss": 1.2292, "step": 53756 }, { "epoch": 0.05, "learning_rate": 4.964557808960036e-05, "loss": 0.8164, "step": 53757 }, { "epoch": 0.05, "learning_rate": 4.964556491013236e-05, "loss": 1.1411, "step": 53758 }, { "epoch": 0.05, "learning_rate": 4.964555173042107e-05, "loss": 0.875, "step": 53759 }, { "epoch": 0.05, "learning_rate": 4.964553855046649e-05, "loss": 0.9388, "step": 53760 }, { "epoch": 0.05, "learning_rate": 4.9645525370268614e-05, "loss": 1.074, "step": 53761 }, { "epoch": 0.05, "learning_rate": 4.964551218982745e-05, "loss": 1.1788, "step": 53762 }, { "epoch": 0.05, "learning_rate": 4.9645499009143004e-05, "loss": 1.0452, "step": 53763 }, { "epoch": 0.05, "learning_rate": 4.964548582821525e-05, "loss": 1.171, "step": 53764 }, { "epoch": 0.05, "learning_rate": 4.964547264704422e-05, "loss": 0.9714, "step": 53765 }, { "epoch": 0.05, "learning_rate": 4.964545946562991e-05, "loss": 1.0021, "step": 53766 }, { "epoch": 0.05, "learning_rate": 4.964544628397229e-05, "loss": 1.1416, "step": 53767 }, { "epoch": 0.05, "learning_rate": 4.9645433102071395e-05, "loss": 1.1449, "step": 53768 }, { "epoch": 0.05, "learning_rate": 4.96454199199272e-05, "loss": 1.1113, "step": 53769 }, { "epoch": 0.05, "learning_rate": 4.9645406737539716e-05, "loss": 1.1737, "step": 53770 }, { "epoch": 0.05, "learning_rate": 4.964539355490895e-05, "loss": 1.0182, "step": 53771 }, { "epoch": 0.05, "learning_rate": 4.964538037203489e-05, "loss": 1.0311, "step": 53772 }, { "epoch": 0.05, "learning_rate": 4.964536718891755e-05, "loss": 1.1444, "step": 53773 }, { "epoch": 0.05, "learning_rate": 4.964535400555691e-05, "loss": 1.073, "step": 53774 }, { "epoch": 0.05, "learning_rate": 4.964534082195299e-05, "loss": 1.231, "step": 53775 }, { "epoch": 0.05, "learning_rate": 4.964532763810578e-05, "loss": 0.9005, "step": 53776 }, { "epoch": 0.05, "learning_rate": 4.964531445401527e-05, "loss": 1.0512, "step": 53777 }, { "epoch": 0.05, "learning_rate": 4.964530126968149e-05, "loss": 1.0012, "step": 53778 }, { "epoch": 0.05, "learning_rate": 4.9645288085104407e-05, "loss": 0.987, "step": 53779 }, { "epoch": 0.05, "learning_rate": 4.9645274900284045e-05, "loss": 0.8449, "step": 53780 }, { "epoch": 0.05, "learning_rate": 4.9645261715220394e-05, "loss": 0.8957, "step": 53781 }, { "epoch": 0.05, "learning_rate": 4.964524852991345e-05, "loss": 1.1314, "step": 53782 }, { "epoch": 0.05, "learning_rate": 4.964523534436323e-05, "loss": 1.2063, "step": 53783 }, { "epoch": 0.05, "learning_rate": 4.964522215856971e-05, "loss": 1.2102, "step": 53784 }, { "epoch": 0.05, "learning_rate": 4.9645208972532905e-05, "loss": 1.2218, "step": 53785 }, { "epoch": 0.05, "learning_rate": 4.964519578625282e-05, "loss": 1.2674, "step": 53786 }, { "epoch": 0.05, "learning_rate": 4.964518259972945e-05, "loss": 1.2373, "step": 53787 }, { "epoch": 0.05, "learning_rate": 4.964516941296278e-05, "loss": 1.1518, "step": 53788 }, { "epoch": 0.05, "learning_rate": 4.9645156225952825e-05, "loss": 1.1984, "step": 53789 }, { "epoch": 0.05, "learning_rate": 4.9645143038699595e-05, "loss": 1.0841, "step": 53790 }, { "epoch": 0.05, "learning_rate": 4.964512985120307e-05, "loss": 1.1224, "step": 53791 }, { "epoch": 0.05, "learning_rate": 4.9645116663463256e-05, "loss": 1.1583, "step": 53792 }, { "epoch": 0.05, "learning_rate": 4.964510347548017e-05, "loss": 1.0779, "step": 53793 }, { "epoch": 0.05, "learning_rate": 4.9645090287253785e-05, "loss": 1.1002, "step": 53794 }, { "epoch": 0.05, "learning_rate": 4.9645077098784113e-05, "loss": 1.1071, "step": 53795 }, { "epoch": 0.05, "learning_rate": 4.964506391007117e-05, "loss": 1.0668, "step": 53796 }, { "epoch": 0.05, "learning_rate": 4.964505072111492e-05, "loss": 1.171, "step": 53797 }, { "epoch": 0.05, "learning_rate": 4.96450375319154e-05, "loss": 1.0161, "step": 53798 }, { "epoch": 0.05, "learning_rate": 4.964502434247259e-05, "loss": 1.1038, "step": 53799 }, { "epoch": 0.05, "learning_rate": 4.964501115278649e-05, "loss": 1.1086, "step": 53800 }, { "epoch": 0.05, "learning_rate": 4.964499796285711e-05, "loss": 1.0297, "step": 53801 }, { "epoch": 0.05, "learning_rate": 4.964498477268445e-05, "loss": 1.1192, "step": 53802 }, { "epoch": 0.05, "learning_rate": 4.96449715822685e-05, "loss": 1.007, "step": 53803 }, { "epoch": 0.05, "learning_rate": 4.9644958391609266e-05, "loss": 1.0949, "step": 53804 }, { "epoch": 0.05, "learning_rate": 4.964494520070674e-05, "loss": 1.212, "step": 53805 }, { "epoch": 0.05, "learning_rate": 4.964493200956094e-05, "loss": 1.1708, "step": 53806 }, { "epoch": 0.05, "learning_rate": 4.964491881817185e-05, "loss": 1.1998, "step": 53807 }, { "epoch": 0.05, "learning_rate": 4.964490562653947e-05, "loss": 1.1092, "step": 53808 }, { "epoch": 0.05, "learning_rate": 4.964489243466382e-05, "loss": 1.2021, "step": 53809 }, { "epoch": 0.05, "learning_rate": 4.964487924254488e-05, "loss": 1.2138, "step": 53810 }, { "epoch": 0.05, "learning_rate": 4.964486605018265e-05, "loss": 1.2099, "step": 53811 }, { "epoch": 0.05, "learning_rate": 4.9644852857577144e-05, "loss": 1.2339, "step": 53812 }, { "epoch": 0.05, "learning_rate": 4.9644839664728346e-05, "loss": 1.2403, "step": 53813 }, { "epoch": 0.05, "learning_rate": 4.9644826471636274e-05, "loss": 1.2321, "step": 53814 }, { "epoch": 0.05, "learning_rate": 4.964481327830092e-05, "loss": 1.2562, "step": 53815 }, { "epoch": 0.05, "learning_rate": 4.964480008472228e-05, "loss": 1.194, "step": 53816 }, { "epoch": 0.05, "learning_rate": 4.9644786890900354e-05, "loss": 1.2321, "step": 53817 }, { "epoch": 0.05, "learning_rate": 4.964477369683514e-05, "loss": 1.1841, "step": 53818 }, { "epoch": 0.05, "learning_rate": 4.964476050252665e-05, "loss": 1.2004, "step": 53819 }, { "epoch": 0.05, "learning_rate": 4.9644747307974875e-05, "loss": 1.1949, "step": 53820 }, { "epoch": 0.05, "learning_rate": 4.964473411317982e-05, "loss": 1.2633, "step": 53821 }, { "epoch": 0.05, "learning_rate": 4.964472091814148e-05, "loss": 1.1828, "step": 53822 }, { "epoch": 0.05, "learning_rate": 4.9644707722859864e-05, "loss": 1.1927, "step": 53823 }, { "epoch": 0.05, "learning_rate": 4.964469452733496e-05, "loss": 1.2113, "step": 53824 }, { "epoch": 0.05, "learning_rate": 4.9644681331566775e-05, "loss": 1.2015, "step": 53825 }, { "epoch": 0.05, "learning_rate": 4.9644668135555306e-05, "loss": 1.2308, "step": 53826 }, { "epoch": 0.05, "learning_rate": 4.9644654939300555e-05, "loss": 1.2116, "step": 53827 }, { "epoch": 0.05, "learning_rate": 4.964464174280252e-05, "loss": 1.2007, "step": 53828 }, { "epoch": 0.05, "learning_rate": 4.964462854606121e-05, "loss": 1.2339, "step": 53829 }, { "epoch": 0.05, "learning_rate": 4.964461534907662e-05, "loss": 1.2624, "step": 53830 }, { "epoch": 0.05, "learning_rate": 4.964460215184874e-05, "loss": 1.2797, "step": 53831 }, { "epoch": 0.05, "learning_rate": 4.964458895437759e-05, "loss": 1.1687, "step": 53832 }, { "epoch": 0.05, "learning_rate": 4.9644575756663145e-05, "loss": 1.2374, "step": 53833 }, { "epoch": 0.05, "learning_rate": 4.964456255870543e-05, "loss": 1.1314, "step": 53834 }, { "epoch": 0.05, "learning_rate": 4.9644549360504435e-05, "loss": 1.2012, "step": 53835 }, { "epoch": 0.05, "learning_rate": 4.964453616206016e-05, "loss": 1.1628, "step": 53836 }, { "epoch": 0.05, "learning_rate": 4.96445229633726e-05, "loss": 1.1028, "step": 53837 }, { "epoch": 0.05, "learning_rate": 4.964450976444175e-05, "loss": 1.2326, "step": 53838 }, { "epoch": 0.05, "learning_rate": 4.964449656526764e-05, "loss": 1.2253, "step": 53839 }, { "epoch": 0.05, "learning_rate": 4.9644483365850236e-05, "loss": 1.2103, "step": 53840 }, { "epoch": 0.05, "learning_rate": 4.964447016618956e-05, "loss": 1.1029, "step": 53841 }, { "epoch": 0.05, "learning_rate": 4.96444569662856e-05, "loss": 1.1688, "step": 53842 }, { "epoch": 0.05, "learning_rate": 4.964444376613835e-05, "loss": 1.1077, "step": 53843 }, { "epoch": 0.05, "learning_rate": 4.964443056574784e-05, "loss": 1.1176, "step": 53844 }, { "epoch": 0.05, "learning_rate": 4.9644417365114035e-05, "loss": 1.1019, "step": 53845 }, { "epoch": 0.05, "learning_rate": 4.964440416423696e-05, "loss": 1.1188, "step": 53846 }, { "epoch": 0.05, "learning_rate": 4.9644390963116605e-05, "loss": 1.0891, "step": 53847 }, { "epoch": 0.05, "learning_rate": 4.964437776175297e-05, "loss": 1.1446, "step": 53848 }, { "epoch": 0.05, "learning_rate": 4.964436456014605e-05, "loss": 1.1895, "step": 53849 }, { "epoch": 0.05, "learning_rate": 4.964435135829586e-05, "loss": 1.1657, "step": 53850 }, { "epoch": 0.05, "learning_rate": 4.964433815620239e-05, "loss": 1.0445, "step": 53851 }, { "epoch": 0.05, "learning_rate": 4.964432495386563e-05, "loss": 1.0582, "step": 53852 }, { "epoch": 0.05, "learning_rate": 4.96443117512856e-05, "loss": 1.0769, "step": 53853 }, { "epoch": 0.05, "learning_rate": 4.96442985484623e-05, "loss": 1.1081, "step": 53854 }, { "epoch": 0.05, "learning_rate": 4.964428534539571e-05, "loss": 1.1368, "step": 53855 }, { "epoch": 0.05, "learning_rate": 4.9644272142085845e-05, "loss": 1.1088, "step": 53856 }, { "epoch": 0.05, "learning_rate": 4.9644258938532706e-05, "loss": 1.1353, "step": 53857 }, { "epoch": 0.05, "learning_rate": 4.9644245734736285e-05, "loss": 1.1116, "step": 53858 }, { "epoch": 0.05, "learning_rate": 4.964423253069659e-05, "loss": 1.0412, "step": 53859 }, { "epoch": 0.05, "learning_rate": 4.964421932641361e-05, "loss": 1.0115, "step": 53860 }, { "epoch": 0.05, "learning_rate": 4.964420612188736e-05, "loss": 1.0986, "step": 53861 }, { "epoch": 0.05, "learning_rate": 4.9644192917117835e-05, "loss": 1.1231, "step": 53862 }, { "epoch": 0.05, "learning_rate": 4.964417971210502e-05, "loss": 1.1229, "step": 53863 }, { "epoch": 0.05, "learning_rate": 4.9644166506848946e-05, "loss": 1.1147, "step": 53864 }, { "epoch": 0.05, "learning_rate": 4.964415330134958e-05, "loss": 1.1242, "step": 53865 }, { "epoch": 0.05, "learning_rate": 4.9644140095606944e-05, "loss": 1.1286, "step": 53866 }, { "epoch": 0.05, "learning_rate": 4.964412688962103e-05, "loss": 1.0941, "step": 53867 }, { "epoch": 0.05, "learning_rate": 4.9644113683391844e-05, "loss": 1.1661, "step": 53868 }, { "epoch": 0.05, "learning_rate": 4.9644100476919375e-05, "loss": 1.1328, "step": 53869 }, { "epoch": 0.05, "learning_rate": 4.964408727020363e-05, "loss": 1.177, "step": 53870 }, { "epoch": 0.05, "learning_rate": 4.964407406324462e-05, "loss": 1.1299, "step": 53871 }, { "epoch": 0.05, "learning_rate": 4.964406085604232e-05, "loss": 1.0977, "step": 53872 }, { "epoch": 0.05, "learning_rate": 4.9644047648596746e-05, "loss": 1.1044, "step": 53873 }, { "epoch": 0.05, "learning_rate": 4.9644034440907904e-05, "loss": 1.4144, "step": 53874 }, { "epoch": 0.05, "learning_rate": 4.964402123297579e-05, "loss": 0.6967, "step": 53875 }, { "epoch": 0.05, "learning_rate": 4.964400802480039e-05, "loss": 0.99, "step": 53876 }, { "epoch": 0.05, "learning_rate": 4.964399481638171e-05, "loss": 1.2586, "step": 53877 }, { "epoch": 0.05, "learning_rate": 4.964398160771977e-05, "loss": 1.2022, "step": 53878 }, { "epoch": 0.05, "learning_rate": 4.964396839881455e-05, "loss": 0.6351, "step": 53879 }, { "epoch": 0.05, "learning_rate": 4.964395518966605e-05, "loss": 1.129, "step": 53880 }, { "epoch": 0.05, "learning_rate": 4.964394198027428e-05, "loss": 0.7227, "step": 53881 }, { "epoch": 0.05, "learning_rate": 4.964392877063923e-05, "loss": 0.944, "step": 53882 }, { "epoch": 0.05, "learning_rate": 4.964391556076091e-05, "loss": 1.6016, "step": 53883 }, { "epoch": 0.05, "learning_rate": 4.964390235063932e-05, "loss": 1.5338, "step": 53884 }, { "epoch": 0.05, "learning_rate": 4.964388914027445e-05, "loss": 1.2701, "step": 53885 }, { "epoch": 0.05, "learning_rate": 4.96438759296663e-05, "loss": 0.8623, "step": 53886 }, { "epoch": 0.05, "learning_rate": 4.964386271881489e-05, "loss": 1.0996, "step": 53887 }, { "epoch": 0.05, "learning_rate": 4.9643849507720196e-05, "loss": 1.0355, "step": 53888 }, { "epoch": 0.05, "learning_rate": 4.964383629638223e-05, "loss": 0.8134, "step": 53889 }, { "epoch": 0.05, "learning_rate": 4.964382308480099e-05, "loss": 1.2602, "step": 53890 }, { "epoch": 0.05, "learning_rate": 4.9643809872976474e-05, "loss": 1.3567, "step": 53891 }, { "epoch": 0.05, "learning_rate": 4.9643796660908695e-05, "loss": 1.747, "step": 53892 }, { "epoch": 0.05, "learning_rate": 4.9643783448597635e-05, "loss": 1.0121, "step": 53893 }, { "epoch": 0.05, "learning_rate": 4.964377023604331e-05, "loss": 1.0559, "step": 53894 }, { "epoch": 0.05, "learning_rate": 4.9643757023245704e-05, "loss": 1.1787, "step": 53895 }, { "epoch": 0.05, "learning_rate": 4.964374381020483e-05, "loss": 1.2982, "step": 53896 }, { "epoch": 0.05, "learning_rate": 4.9643730596920675e-05, "loss": 1.284, "step": 53897 }, { "epoch": 0.05, "learning_rate": 4.964371738339325e-05, "loss": 1.0011, "step": 53898 }, { "epoch": 0.05, "learning_rate": 4.9643704169622554e-05, "loss": 1.0428, "step": 53899 }, { "epoch": 0.05, "learning_rate": 4.964369095560859e-05, "loss": 0.7237, "step": 53900 }, { "epoch": 0.05, "learning_rate": 4.9643677741351354e-05, "loss": 1.1944, "step": 53901 }, { "epoch": 0.05, "learning_rate": 4.964366452685084e-05, "loss": 1.3355, "step": 53902 }, { "epoch": 0.05, "learning_rate": 4.9643651312107056e-05, "loss": 1.2254, "step": 53903 }, { "epoch": 0.05, "learning_rate": 4.964363809712e-05, "loss": 1.2851, "step": 53904 }, { "epoch": 0.05, "learning_rate": 4.964362488188968e-05, "loss": 1.2758, "step": 53905 }, { "epoch": 0.05, "learning_rate": 4.964361166641608e-05, "loss": 1.2002, "step": 53906 }, { "epoch": 0.05, "learning_rate": 4.964359845069921e-05, "loss": 1.0318, "step": 53907 }, { "epoch": 0.05, "learning_rate": 4.964358523473907e-05, "loss": 0.7966, "step": 53908 }, { "epoch": 0.05, "learning_rate": 4.964357201853565e-05, "loss": 0.5439, "step": 53909 }, { "epoch": 0.05, "learning_rate": 4.964355880208897e-05, "loss": 0.5151, "step": 53910 }, { "epoch": 0.05, "learning_rate": 4.964354558539902e-05, "loss": 1.3153, "step": 53911 }, { "epoch": 0.05, "learning_rate": 4.96435323684658e-05, "loss": 1.0018, "step": 53912 }, { "epoch": 0.05, "learning_rate": 4.9643519151289305e-05, "loss": 1.2452, "step": 53913 }, { "epoch": 0.05, "learning_rate": 4.964350593386954e-05, "loss": 1.1092, "step": 53914 }, { "epoch": 0.05, "learning_rate": 4.964349271620651e-05, "loss": 1.416, "step": 53915 }, { "epoch": 0.05, "learning_rate": 4.96434794983002e-05, "loss": 1.3117, "step": 53916 }, { "epoch": 0.05, "learning_rate": 4.9643466280150625e-05, "loss": 0.8795, "step": 53917 }, { "epoch": 0.05, "learning_rate": 4.964345306175778e-05, "loss": 1.0044, "step": 53918 }, { "epoch": 0.05, "learning_rate": 4.964343984312167e-05, "loss": 1.3521, "step": 53919 }, { "epoch": 0.05, "learning_rate": 4.964342662424228e-05, "loss": 0.9931, "step": 53920 }, { "epoch": 0.05, "learning_rate": 4.964341340511962e-05, "loss": 1.0882, "step": 53921 }, { "epoch": 0.05, "learning_rate": 4.96434001857537e-05, "loss": 1.3741, "step": 53922 }, { "epoch": 0.05, "learning_rate": 4.964338696614451e-05, "loss": 1.3277, "step": 53923 }, { "epoch": 0.05, "learning_rate": 4.9643373746292045e-05, "loss": 1.1233, "step": 53924 }, { "epoch": 0.05, "learning_rate": 4.964336052619631e-05, "loss": 1.0361, "step": 53925 }, { "epoch": 0.05, "learning_rate": 4.964334730585731e-05, "loss": 1.0865, "step": 53926 }, { "epoch": 0.05, "learning_rate": 4.964333408527505e-05, "loss": 0.1805, "step": 53927 }, { "epoch": 0.05, "learning_rate": 4.9643320864449504e-05, "loss": 0.1625, "step": 53928 }, { "epoch": 0.05, "learning_rate": 4.9643307643380706e-05, "loss": 0.0606, "step": 53929 }, { "epoch": 0.05, "learning_rate": 4.964329442206863e-05, "loss": 0.051, "step": 53930 }, { "epoch": 0.05, "learning_rate": 4.9643281200513286e-05, "loss": 0.0436, "step": 53931 }, { "epoch": 0.05, "learning_rate": 4.964326797871468e-05, "loss": 0.0344, "step": 53932 }, { "epoch": 0.05, "learning_rate": 4.9643254756672794e-05, "loss": 0.1071, "step": 53933 }, { "epoch": 0.05, "learning_rate": 4.964324153438765e-05, "loss": 0.3521, "step": 53934 }, { "epoch": 0.05, "learning_rate": 4.964322831185924e-05, "loss": 0.4284, "step": 53935 }, { "epoch": 0.05, "learning_rate": 4.964321508908756e-05, "loss": 0.533, "step": 53936 }, { "epoch": 0.05, "learning_rate": 4.964320186607261e-05, "loss": 0.4411, "step": 53937 }, { "epoch": 0.05, "learning_rate": 4.9643188642814394e-05, "loss": 0.4249, "step": 53938 }, { "epoch": 0.05, "learning_rate": 4.9643175419312904e-05, "loss": 0.3615, "step": 53939 }, { "epoch": 0.05, "learning_rate": 4.964316219556816e-05, "loss": 0.3791, "step": 53940 }, { "epoch": 0.05, "learning_rate": 4.964314897158014e-05, "loss": 0.4001, "step": 53941 }, { "epoch": 0.05, "learning_rate": 4.964313574734886e-05, "loss": 0.3927, "step": 53942 }, { "epoch": 0.05, "learning_rate": 4.9643122522874305e-05, "loss": 0.2305, "step": 53943 }, { "epoch": 0.05, "learning_rate": 4.964310929815649e-05, "loss": 0.2833, "step": 53944 }, { "epoch": 0.05, "learning_rate": 4.9643096073195405e-05, "loss": 0.3345, "step": 53945 }, { "epoch": 0.05, "learning_rate": 4.964308284799105e-05, "loss": 0.4515, "step": 53946 }, { "epoch": 0.05, "learning_rate": 4.9643069622543434e-05, "loss": 0.3511, "step": 53947 }, { "epoch": 0.05, "learning_rate": 4.964305639685255e-05, "loss": 0.327, "step": 53948 }, { "epoch": 0.05, "learning_rate": 4.9643043170918404e-05, "loss": 0.346, "step": 53949 }, { "epoch": 0.05, "learning_rate": 4.9643029944740995e-05, "loss": 0.3175, "step": 53950 }, { "epoch": 0.05, "learning_rate": 4.964301671832031e-05, "loss": 0.6805, "step": 53951 }, { "epoch": 0.05, "learning_rate": 4.964300349165637e-05, "loss": 0.6152, "step": 53952 }, { "epoch": 0.05, "learning_rate": 4.9642990264749165e-05, "loss": 1.0577, "step": 53953 }, { "epoch": 0.05, "learning_rate": 4.9642977037598684e-05, "loss": 0.688, "step": 53954 }, { "epoch": 0.05, "learning_rate": 4.964296381020495e-05, "loss": 0.5958, "step": 53955 }, { "epoch": 0.05, "learning_rate": 4.964295058256794e-05, "loss": 0.6153, "step": 53956 }, { "epoch": 0.05, "learning_rate": 4.964293735468767e-05, "loss": 0.7644, "step": 53957 }, { "epoch": 0.05, "learning_rate": 4.9642924126564135e-05, "loss": 1.3497, "step": 53958 }, { "epoch": 0.05, "learning_rate": 4.9642910898197335e-05, "loss": 1.112, "step": 53959 }, { "epoch": 0.05, "learning_rate": 4.9642897669587274e-05, "loss": 1.5144, "step": 53960 }, { "epoch": 0.05, "learning_rate": 4.9642884440733945e-05, "loss": 1.3592, "step": 53961 }, { "epoch": 0.05, "learning_rate": 4.9642871211637354e-05, "loss": 1.7695, "step": 53962 }, { "epoch": 0.05, "learning_rate": 4.9642857982297496e-05, "loss": 1.335, "step": 53963 }, { "epoch": 0.05, "learning_rate": 4.964284475271437e-05, "loss": 1.0159, "step": 53964 }, { "epoch": 0.05, "learning_rate": 4.964283152288799e-05, "loss": 0.8385, "step": 53965 }, { "epoch": 0.05, "learning_rate": 4.9642818292818335e-05, "loss": 0.7601, "step": 53966 }, { "epoch": 0.05, "learning_rate": 4.964280506250543e-05, "loss": 0.656, "step": 53967 }, { "epoch": 0.05, "learning_rate": 4.9642791831949256e-05, "loss": 0.9335, "step": 53968 }, { "epoch": 0.05, "learning_rate": 4.964277860114982e-05, "loss": 0.9768, "step": 53969 }, { "epoch": 0.05, "learning_rate": 4.964276537010711e-05, "loss": 0.9773, "step": 53970 }, { "epoch": 0.05, "learning_rate": 4.964275213882115e-05, "loss": 1.6289, "step": 53971 }, { "epoch": 0.05, "learning_rate": 4.964273890729193e-05, "loss": 1.6798, "step": 53972 }, { "epoch": 0.05, "learning_rate": 4.9642725675519434e-05, "loss": 1.5534, "step": 53973 }, { "epoch": 0.05, "learning_rate": 4.9642712443503684e-05, "loss": 1.3942, "step": 53974 }, { "epoch": 0.05, "learning_rate": 4.964269921124467e-05, "loss": 1.1652, "step": 53975 }, { "epoch": 0.05, "learning_rate": 4.9642685978742394e-05, "loss": 1.1445, "step": 53976 }, { "epoch": 0.05, "learning_rate": 4.964267274599685e-05, "loss": 1.1203, "step": 53977 }, { "epoch": 0.05, "learning_rate": 4.964265951300805e-05, "loss": 0.8496, "step": 53978 }, { "epoch": 0.05, "learning_rate": 4.964264627977599e-05, "loss": 1.0156, "step": 53979 }, { "epoch": 0.05, "learning_rate": 4.9642633046300666e-05, "loss": 0.876, "step": 53980 }, { "epoch": 0.05, "learning_rate": 4.964261981258208e-05, "loss": 0.996, "step": 53981 }, { "epoch": 0.05, "learning_rate": 4.9642606578620235e-05, "loss": 0.6997, "step": 53982 }, { "epoch": 0.05, "learning_rate": 4.964259334441512e-05, "loss": 0.7348, "step": 53983 }, { "epoch": 0.05, "learning_rate": 4.9642580109966754e-05, "loss": 0.5917, "step": 53984 }, { "epoch": 0.05, "learning_rate": 4.964256687527512e-05, "loss": 1.1449, "step": 53985 }, { "epoch": 0.05, "learning_rate": 4.964255364034022e-05, "loss": 0.6371, "step": 53986 }, { "epoch": 0.05, "learning_rate": 4.964254040516207e-05, "loss": 0.7489, "step": 53987 }, { "epoch": 0.05, "learning_rate": 4.964252716974066e-05, "loss": 0.6036, "step": 53988 }, { "epoch": 0.05, "learning_rate": 4.964251393407598e-05, "loss": 0.7176, "step": 53989 }, { "epoch": 0.05, "learning_rate": 4.964250069816804e-05, "loss": 0.7779, "step": 53990 }, { "epoch": 0.05, "learning_rate": 4.964248746201685e-05, "loss": 0.8524, "step": 53991 }, { "epoch": 0.05, "learning_rate": 4.9642474225622384e-05, "loss": 0.8996, "step": 53992 }, { "epoch": 0.05, "learning_rate": 4.964246098898467e-05, "loss": 0.6988, "step": 53993 }, { "epoch": 0.05, "learning_rate": 4.9642447752103694e-05, "loss": 0.572, "step": 53994 }, { "epoch": 0.05, "learning_rate": 4.964243451497946e-05, "loss": 0.4776, "step": 53995 }, { "epoch": 0.05, "learning_rate": 4.964242127761196e-05, "loss": 0.7553, "step": 53996 }, { "epoch": 0.05, "learning_rate": 4.964240804000121e-05, "loss": 1.0465, "step": 53997 }, { "epoch": 0.05, "learning_rate": 4.964239480214719e-05, "loss": 1.2774, "step": 53998 }, { "epoch": 0.05, "learning_rate": 4.964238156404992e-05, "loss": 1.539, "step": 53999 }, { "epoch": 0.05, "learning_rate": 4.964236832570938e-05, "loss": 1.5903, "step": 54000 }, { "epoch": 0.05, "eval_loss": 1.0967680215835571, "eval_runtime": 127.9544, "eval_samples_per_second": 10.824, "eval_steps_per_second": 5.416, "step": 54000 }, { "epoch": 0.05, "learning_rate": 4.964235508712559e-05, "loss": 1.0501, "step": 54001 }, { "epoch": 0.05, "learning_rate": 4.964234184829853e-05, "loss": 1.1754, "step": 54002 }, { "epoch": 0.05, "learning_rate": 4.964232860922823e-05, "loss": 1.689, "step": 54003 }, { "epoch": 0.05, "learning_rate": 4.9642315369914655e-05, "loss": 1.1196, "step": 54004 }, { "epoch": 0.05, "learning_rate": 4.9642302130357825e-05, "loss": 1.0868, "step": 54005 }, { "epoch": 0.05, "learning_rate": 4.964228889055774e-05, "loss": 1.0159, "step": 54006 }, { "epoch": 0.05, "learning_rate": 4.9642275650514394e-05, "loss": 1.0501, "step": 54007 }, { "epoch": 0.05, "learning_rate": 4.9642262410227794e-05, "loss": 1.1647, "step": 54008 }, { "epoch": 0.05, "learning_rate": 4.964224916969793e-05, "loss": 1.2549, "step": 54009 }, { "epoch": 0.05, "learning_rate": 4.964223592892481e-05, "loss": 1.1494, "step": 54010 }, { "epoch": 0.05, "learning_rate": 4.9642222687908434e-05, "loss": 1.3756, "step": 54011 }, { "epoch": 0.05, "learning_rate": 4.96422094466488e-05, "loss": 1.1749, "step": 54012 }, { "epoch": 0.05, "learning_rate": 4.964219620514591e-05, "loss": 0.923, "step": 54013 }, { "epoch": 0.05, "learning_rate": 4.9642182963399765e-05, "loss": 0.9811, "step": 54014 }, { "epoch": 0.05, "learning_rate": 4.964216972141036e-05, "loss": 1.1156, "step": 54015 }, { "epoch": 0.05, "learning_rate": 4.9642156479177696e-05, "loss": 0.8178, "step": 54016 }, { "epoch": 0.05, "learning_rate": 4.964214323670177e-05, "loss": 1.0367, "step": 54017 }, { "epoch": 0.05, "learning_rate": 4.964212999398259e-05, "loss": 0.9807, "step": 54018 }, { "epoch": 0.05, "learning_rate": 4.9642116751020164e-05, "loss": 1.1115, "step": 54019 }, { "epoch": 0.05, "learning_rate": 4.964210350781447e-05, "loss": 0.7433, "step": 54020 }, { "epoch": 0.05, "learning_rate": 4.964209026436552e-05, "loss": 0.2843, "step": 54021 }, { "epoch": 0.05, "learning_rate": 4.964207702067332e-05, "loss": 0.803, "step": 54022 }, { "epoch": 0.05, "learning_rate": 4.9642063776737855e-05, "loss": 1.21, "step": 54023 }, { "epoch": 0.05, "learning_rate": 4.9642050532559145e-05, "loss": 1.4708, "step": 54024 }, { "epoch": 0.05, "learning_rate": 4.964203728813717e-05, "loss": 1.4619, "step": 54025 }, { "epoch": 0.05, "learning_rate": 4.964202404347195e-05, "loss": 0.9924, "step": 54026 }, { "epoch": 0.05, "learning_rate": 4.964201079856346e-05, "loss": 1.6853, "step": 54027 }, { "epoch": 0.05, "learning_rate": 4.964199755341172e-05, "loss": 1.0608, "step": 54028 }, { "epoch": 0.05, "learning_rate": 4.964198430801673e-05, "loss": 1.207, "step": 54029 }, { "epoch": 0.05, "learning_rate": 4.964197106237848e-05, "loss": 1.1666, "step": 54030 }, { "epoch": 0.05, "learning_rate": 4.9641957816496975e-05, "loss": 0.8857, "step": 54031 }, { "epoch": 0.05, "learning_rate": 4.964194457037221e-05, "loss": 0.9532, "step": 54032 }, { "epoch": 0.05, "learning_rate": 4.96419313240042e-05, "loss": 1.0557, "step": 54033 }, { "epoch": 0.05, "learning_rate": 4.964191807739293e-05, "loss": 1.1217, "step": 54034 }, { "epoch": 0.05, "learning_rate": 4.9641904830538404e-05, "loss": 1.1299, "step": 54035 }, { "epoch": 0.05, "learning_rate": 4.964189158344062e-05, "loss": 1.5318, "step": 54036 }, { "epoch": 0.05, "learning_rate": 4.9641878336099587e-05, "loss": 1.1455, "step": 54037 }, { "epoch": 0.05, "learning_rate": 4.96418650885153e-05, "loss": 1.3287, "step": 54038 }, { "epoch": 0.05, "learning_rate": 4.9641851840687765e-05, "loss": 0.8211, "step": 54039 }, { "epoch": 0.05, "learning_rate": 4.964183859261697e-05, "loss": 1.0983, "step": 54040 }, { "epoch": 0.05, "learning_rate": 4.964182534430292e-05, "loss": 1.2692, "step": 54041 }, { "epoch": 0.05, "learning_rate": 4.964181209574562e-05, "loss": 1.4209, "step": 54042 }, { "epoch": 0.05, "learning_rate": 4.964179884694506e-05, "loss": 1.2166, "step": 54043 }, { "epoch": 0.05, "learning_rate": 4.9641785597901254e-05, "loss": 1.1683, "step": 54044 }, { "epoch": 0.05, "learning_rate": 4.964177234861419e-05, "loss": 1.2068, "step": 54045 }, { "epoch": 0.05, "learning_rate": 4.964175909908387e-05, "loss": 1.0514, "step": 54046 }, { "epoch": 0.05, "learning_rate": 4.9641745849310306e-05, "loss": 1.035, "step": 54047 }, { "epoch": 0.05, "learning_rate": 4.964173259929348e-05, "loss": 0.9463, "step": 54048 }, { "epoch": 0.05, "learning_rate": 4.9641719349033405e-05, "loss": 1.326, "step": 54049 }, { "epoch": 0.05, "learning_rate": 4.964170609853008e-05, "loss": 0.9728, "step": 54050 }, { "epoch": 0.05, "learning_rate": 4.9641692847783494e-05, "loss": 0.9568, "step": 54051 }, { "epoch": 0.05, "learning_rate": 4.9641679596793663e-05, "loss": 0.8532, "step": 54052 }, { "epoch": 0.05, "learning_rate": 4.964166634556058e-05, "loss": 0.9976, "step": 54053 }, { "epoch": 0.05, "learning_rate": 4.964165309408424e-05, "loss": 0.7573, "step": 54054 }, { "epoch": 0.05, "learning_rate": 4.964163984236465e-05, "loss": 0.9779, "step": 54055 }, { "epoch": 0.05, "learning_rate": 4.964162659040181e-05, "loss": 1.357, "step": 54056 }, { "epoch": 0.05, "learning_rate": 4.964161333819572e-05, "loss": 1.047, "step": 54057 }, { "epoch": 0.05, "learning_rate": 4.9641600085746374e-05, "loss": 1.0067, "step": 54058 }, { "epoch": 0.05, "learning_rate": 4.964158683305378e-05, "loss": 0.9555, "step": 54059 }, { "epoch": 0.05, "learning_rate": 4.964157358011793e-05, "loss": 1.4225, "step": 54060 }, { "epoch": 0.05, "learning_rate": 4.9641560326938834e-05, "loss": 0.9615, "step": 54061 }, { "epoch": 0.05, "learning_rate": 4.964154707351648e-05, "loss": 0.4593, "step": 54062 }, { "epoch": 0.05, "learning_rate": 4.964153381985088e-05, "loss": 0.3505, "step": 54063 }, { "epoch": 0.05, "learning_rate": 4.964152056594203e-05, "loss": 0.4813, "step": 54064 }, { "epoch": 0.05, "learning_rate": 4.964150731178993e-05, "loss": 1.1797, "step": 54065 }, { "epoch": 0.05, "learning_rate": 4.964149405739458e-05, "loss": 1.2668, "step": 54066 }, { "epoch": 0.05, "learning_rate": 4.9641480802755975e-05, "loss": 1.0092, "step": 54067 }, { "epoch": 0.05, "learning_rate": 4.964146754787412e-05, "loss": 1.1013, "step": 54068 }, { "epoch": 0.05, "learning_rate": 4.9641454292749016e-05, "loss": 0.3802, "step": 54069 }, { "epoch": 0.05, "learning_rate": 4.964144103738066e-05, "loss": 0.332, "step": 54070 }, { "epoch": 0.05, "learning_rate": 4.964142778176906e-05, "loss": 0.4423, "step": 54071 }, { "epoch": 0.05, "learning_rate": 4.964141452591421e-05, "loss": 0.4122, "step": 54072 }, { "epoch": 0.05, "learning_rate": 4.96414012698161e-05, "loss": 0.3289, "step": 54073 }, { "epoch": 0.05, "learning_rate": 4.964138801347475e-05, "loss": 0.2983, "step": 54074 }, { "epoch": 0.05, "learning_rate": 4.964137475689015e-05, "loss": 0.219, "step": 54075 }, { "epoch": 0.05, "learning_rate": 4.96413615000623e-05, "loss": 0.2719, "step": 54076 }, { "epoch": 0.05, "learning_rate": 4.96413482429912e-05, "loss": 0.2346, "step": 54077 }, { "epoch": 0.05, "learning_rate": 4.964133498567685e-05, "loss": 0.1692, "step": 54078 }, { "epoch": 0.05, "learning_rate": 4.964132172811925e-05, "loss": 0.2337, "step": 54079 }, { "epoch": 0.05, "learning_rate": 4.96413084703184e-05, "loss": 0.2495, "step": 54080 }, { "epoch": 0.05, "learning_rate": 4.9641295212274305e-05, "loss": 0.2117, "step": 54081 }, { "epoch": 0.05, "learning_rate": 4.964128195398696e-05, "loss": 0.734, "step": 54082 }, { "epoch": 0.05, "learning_rate": 4.9641268695456375e-05, "loss": 0.987, "step": 54083 }, { "epoch": 0.05, "learning_rate": 4.964125543668253e-05, "loss": 1.2432, "step": 54084 }, { "epoch": 0.05, "learning_rate": 4.964124217766544e-05, "loss": 1.2284, "step": 54085 }, { "epoch": 0.05, "learning_rate": 4.96412289184051e-05, "loss": 1.0046, "step": 54086 }, { "epoch": 0.05, "learning_rate": 4.964121565890152e-05, "loss": 1.3075, "step": 54087 }, { "epoch": 0.05, "learning_rate": 4.964120239915468e-05, "loss": 1.0227, "step": 54088 }, { "epoch": 0.05, "learning_rate": 4.964118913916461e-05, "loss": 0.9271, "step": 54089 }, { "epoch": 0.05, "learning_rate": 4.964117587893128e-05, "loss": 1.3558, "step": 54090 }, { "epoch": 0.05, "learning_rate": 4.964116261845471e-05, "loss": 1.4228, "step": 54091 }, { "epoch": 0.05, "learning_rate": 4.964114935773488e-05, "loss": 1.4265, "step": 54092 }, { "epoch": 0.05, "learning_rate": 4.964113609677182e-05, "loss": 1.4813, "step": 54093 }, { "epoch": 0.05, "learning_rate": 4.96411228355655e-05, "loss": 1.3662, "step": 54094 }, { "epoch": 0.05, "learning_rate": 4.964110957411594e-05, "loss": 1.1968, "step": 54095 }, { "epoch": 0.05, "learning_rate": 4.964109631242313e-05, "loss": 1.0854, "step": 54096 }, { "epoch": 0.05, "learning_rate": 4.9641083050487075e-05, "loss": 1.1618, "step": 54097 }, { "epoch": 0.05, "learning_rate": 4.964106978830777e-05, "loss": 1.837, "step": 54098 }, { "epoch": 0.05, "learning_rate": 4.9641056525885224e-05, "loss": 1.5109, "step": 54099 }, { "epoch": 0.05, "learning_rate": 4.964104326321943e-05, "loss": 0.9184, "step": 54100 }, { "epoch": 0.05, "learning_rate": 4.964103000031039e-05, "loss": 0.803, "step": 54101 }, { "epoch": 0.05, "learning_rate": 4.96410167371581e-05, "loss": 1.0526, "step": 54102 }, { "epoch": 0.05, "learning_rate": 4.964100347376257e-05, "loss": 0.7178, "step": 54103 }, { "epoch": 0.05, "learning_rate": 4.9640990210123796e-05, "loss": 1.4574, "step": 54104 }, { "epoch": 0.05, "learning_rate": 4.964097694624177e-05, "loss": 1.1918, "step": 54105 }, { "epoch": 0.05, "learning_rate": 4.9640963682116507e-05, "loss": 0.9974, "step": 54106 }, { "epoch": 0.05, "learning_rate": 4.964095041774799e-05, "loss": 1.1121, "step": 54107 }, { "epoch": 0.05, "learning_rate": 4.964093715313623e-05, "loss": 1.1487, "step": 54108 }, { "epoch": 0.05, "learning_rate": 4.9640923888281226e-05, "loss": 1.1209, "step": 54109 }, { "epoch": 0.05, "learning_rate": 4.964091062318298e-05, "loss": 0.8842, "step": 54110 }, { "epoch": 0.05, "learning_rate": 4.964089735784149e-05, "loss": 0.8445, "step": 54111 }, { "epoch": 0.05, "learning_rate": 4.9640884092256746e-05, "loss": 0.8247, "step": 54112 }, { "epoch": 0.05, "learning_rate": 4.964087082642877e-05, "loss": 1.2269, "step": 54113 }, { "epoch": 0.05, "learning_rate": 4.9640857560357536e-05, "loss": 1.2081, "step": 54114 }, { "epoch": 0.05, "learning_rate": 4.9640844294043064e-05, "loss": 0.8117, "step": 54115 }, { "epoch": 0.05, "learning_rate": 4.964083102748535e-05, "loss": 1.2904, "step": 54116 }, { "epoch": 0.05, "learning_rate": 4.96408177606844e-05, "loss": 1.0741, "step": 54117 }, { "epoch": 0.05, "learning_rate": 4.964080449364019e-05, "loss": 1.27, "step": 54118 }, { "epoch": 0.05, "learning_rate": 4.9640791226352746e-05, "loss": 1.1667, "step": 54119 }, { "epoch": 0.05, "learning_rate": 4.964077795882206e-05, "loss": 1.3654, "step": 54120 }, { "epoch": 0.05, "learning_rate": 4.9640764691048126e-05, "loss": 0.7836, "step": 54121 }, { "epoch": 0.05, "learning_rate": 4.9640751423030954e-05, "loss": 1.2215, "step": 54122 }, { "epoch": 0.05, "learning_rate": 4.964073815477053e-05, "loss": 1.0378, "step": 54123 }, { "epoch": 0.05, "learning_rate": 4.964072488626686e-05, "loss": 1.0442, "step": 54124 }, { "epoch": 0.05, "learning_rate": 4.964071161751996e-05, "loss": 1.0791, "step": 54125 }, { "epoch": 0.05, "learning_rate": 4.9640698348529814e-05, "loss": 0.9001, "step": 54126 }, { "epoch": 0.05, "learning_rate": 4.9640685079296425e-05, "loss": 1.1012, "step": 54127 }, { "epoch": 0.05, "learning_rate": 4.9640671809819796e-05, "loss": 1.0498, "step": 54128 }, { "epoch": 0.05, "learning_rate": 4.964065854009992e-05, "loss": 1.2597, "step": 54129 }, { "epoch": 0.05, "learning_rate": 4.96406452701368e-05, "loss": 0.9644, "step": 54130 }, { "epoch": 0.05, "learning_rate": 4.9640631999930436e-05, "loss": 1.1485, "step": 54131 }, { "epoch": 0.05, "learning_rate": 4.964061872948084e-05, "loss": 1.533, "step": 54132 }, { "epoch": 0.05, "learning_rate": 4.964060545878799e-05, "loss": 0.99, "step": 54133 }, { "epoch": 0.05, "learning_rate": 4.964059218785191e-05, "loss": 1.2001, "step": 54134 }, { "epoch": 0.05, "learning_rate": 4.964057891667259e-05, "loss": 1.2635, "step": 54135 }, { "epoch": 0.05, "learning_rate": 4.964056564525001e-05, "loss": 1.0756, "step": 54136 }, { "epoch": 0.05, "learning_rate": 4.96405523735842e-05, "loss": 1.1481, "step": 54137 }, { "epoch": 0.05, "learning_rate": 4.964053910167515e-05, "loss": 1.4355, "step": 54138 }, { "epoch": 0.05, "learning_rate": 4.964052582952287e-05, "loss": 1.0234, "step": 54139 }, { "epoch": 0.05, "learning_rate": 4.9640512557127335e-05, "loss": 1.0088, "step": 54140 }, { "epoch": 0.05, "learning_rate": 4.964049928448856e-05, "loss": 1.0578, "step": 54141 }, { "epoch": 0.05, "learning_rate": 4.964048601160655e-05, "loss": 1.3106, "step": 54142 }, { "epoch": 0.05, "learning_rate": 4.9640472738481294e-05, "loss": 1.3392, "step": 54143 }, { "epoch": 0.05, "learning_rate": 4.96404594651128e-05, "loss": 1.1499, "step": 54144 }, { "epoch": 0.05, "learning_rate": 4.9640446191501064e-05, "loss": 1.2484, "step": 54145 }, { "epoch": 0.05, "learning_rate": 4.964043291764609e-05, "loss": 1.2777, "step": 54146 }, { "epoch": 0.05, "learning_rate": 4.9640419643547876e-05, "loss": 1.1746, "step": 54147 }, { "epoch": 0.05, "learning_rate": 4.964040636920642e-05, "loss": 1.2682, "step": 54148 }, { "epoch": 0.05, "learning_rate": 4.9640393094621726e-05, "loss": 1.3312, "step": 54149 }, { "epoch": 0.05, "learning_rate": 4.964037981979379e-05, "loss": 1.4228, "step": 54150 }, { "epoch": 0.05, "learning_rate": 4.964036654472262e-05, "loss": 1.283, "step": 54151 }, { "epoch": 0.05, "learning_rate": 4.9640353269408205e-05, "loss": 1.2821, "step": 54152 }, { "epoch": 0.05, "learning_rate": 4.964033999385056e-05, "loss": 1.2375, "step": 54153 }, { "epoch": 0.05, "learning_rate": 4.964032671804966e-05, "loss": 1.4562, "step": 54154 }, { "epoch": 0.05, "learning_rate": 4.964031344200554e-05, "loss": 1.277, "step": 54155 }, { "epoch": 0.05, "learning_rate": 4.964030016571817e-05, "loss": 1.2732, "step": 54156 }, { "epoch": 0.05, "learning_rate": 4.964028688918756e-05, "loss": 1.0745, "step": 54157 }, { "epoch": 0.05, "learning_rate": 4.964027361241371e-05, "loss": 1.0933, "step": 54158 }, { "epoch": 0.05, "learning_rate": 4.9640260335396624e-05, "loss": 1.2683, "step": 54159 }, { "epoch": 0.05, "learning_rate": 4.9640247058136304e-05, "loss": 1.3443, "step": 54160 }, { "epoch": 0.05, "learning_rate": 4.9640233780632743e-05, "loss": 1.024, "step": 54161 }, { "epoch": 0.05, "learning_rate": 4.964022050288595e-05, "loss": 0.9574, "step": 54162 }, { "epoch": 0.05, "learning_rate": 4.9640207224895906e-05, "loss": 0.6691, "step": 54163 }, { "epoch": 0.05, "learning_rate": 4.964019394666264e-05, "loss": 1.3888, "step": 54164 }, { "epoch": 0.05, "learning_rate": 4.9640180668186126e-05, "loss": 1.0596, "step": 54165 }, { "epoch": 0.05, "learning_rate": 4.9640167389466375e-05, "loss": 0.9791, "step": 54166 }, { "epoch": 0.05, "learning_rate": 4.964015411050339e-05, "loss": 1.0355, "step": 54167 }, { "epoch": 0.05, "learning_rate": 4.9640140831297164e-05, "loss": 0.9699, "step": 54168 }, { "epoch": 0.05, "learning_rate": 4.9640127551847704e-05, "loss": 1.0856, "step": 54169 }, { "epoch": 0.05, "learning_rate": 4.964011427215501e-05, "loss": 2.0564, "step": 54170 }, { "epoch": 0.05, "learning_rate": 4.9640100992219076e-05, "loss": 2.7855, "step": 54171 }, { "epoch": 0.05, "learning_rate": 4.96400877120399e-05, "loss": 1.7189, "step": 54172 }, { "epoch": 0.05, "learning_rate": 4.964007443161749e-05, "loss": 1.2389, "step": 54173 }, { "epoch": 0.05, "learning_rate": 4.964006115095185e-05, "loss": 0.9938, "step": 54174 }, { "epoch": 0.05, "learning_rate": 4.964004787004297e-05, "loss": 1.327, "step": 54175 }, { "epoch": 0.05, "learning_rate": 4.9640034588890846e-05, "loss": 1.4053, "step": 54176 }, { "epoch": 0.05, "learning_rate": 4.96400213074955e-05, "loss": 1.3339, "step": 54177 }, { "epoch": 0.05, "learning_rate": 4.964000802585691e-05, "loss": 1.6081, "step": 54178 }, { "epoch": 0.05, "learning_rate": 4.963999474397508e-05, "loss": 1.2585, "step": 54179 }, { "epoch": 0.05, "learning_rate": 4.963998146185003e-05, "loss": 1.1952, "step": 54180 }, { "epoch": 0.05, "learning_rate": 4.963996817948173e-05, "loss": 1.235, "step": 54181 }, { "epoch": 0.05, "learning_rate": 4.96399548968702e-05, "loss": 1.2333, "step": 54182 }, { "epoch": 0.05, "learning_rate": 4.963994161401543e-05, "loss": 1.4996, "step": 54183 }, { "epoch": 0.05, "learning_rate": 4.963992833091743e-05, "loss": 1.3982, "step": 54184 }, { "epoch": 0.05, "learning_rate": 4.9639915047576194e-05, "loss": 1.2881, "step": 54185 }, { "epoch": 0.05, "learning_rate": 4.963990176399172e-05, "loss": 1.2613, "step": 54186 }, { "epoch": 0.05, "learning_rate": 4.9639888480164024e-05, "loss": 1.0875, "step": 54187 }, { "epoch": 0.05, "learning_rate": 4.963987519609308e-05, "loss": 1.0179, "step": 54188 }, { "epoch": 0.05, "learning_rate": 4.96398619117789e-05, "loss": 1.0072, "step": 54189 }, { "epoch": 0.05, "learning_rate": 4.963984862722149e-05, "loss": 1.1834, "step": 54190 }, { "epoch": 0.05, "learning_rate": 4.963983534242085e-05, "loss": 1.3899, "step": 54191 }, { "epoch": 0.05, "learning_rate": 4.9639822057376974e-05, "loss": 1.2769, "step": 54192 }, { "epoch": 0.05, "learning_rate": 4.963980877208986e-05, "loss": 0.8939, "step": 54193 }, { "epoch": 0.05, "learning_rate": 4.9639795486559515e-05, "loss": 0.9731, "step": 54194 }, { "epoch": 0.05, "learning_rate": 4.963978220078593e-05, "loss": 1.0734, "step": 54195 }, { "epoch": 0.05, "learning_rate": 4.9639768914769126e-05, "loss": 1.1048, "step": 54196 }, { "epoch": 0.05, "learning_rate": 4.9639755628509074e-05, "loss": 1.1974, "step": 54197 }, { "epoch": 0.05, "learning_rate": 4.9639742342005795e-05, "loss": 1.1919, "step": 54198 }, { "epoch": 0.05, "learning_rate": 4.963972905525929e-05, "loss": 1.6035, "step": 54199 }, { "epoch": 0.05, "learning_rate": 4.963971576826954e-05, "loss": 0.8792, "step": 54200 }, { "epoch": 0.05, "learning_rate": 4.963970248103656e-05, "loss": 1.4016, "step": 54201 }, { "epoch": 0.05, "learning_rate": 4.9639689193560346e-05, "loss": 1.3484, "step": 54202 }, { "epoch": 0.05, "learning_rate": 4.96396759058409e-05, "loss": 1.5533, "step": 54203 }, { "epoch": 0.05, "learning_rate": 4.963966261787823e-05, "loss": 1.4211, "step": 54204 }, { "epoch": 0.05, "learning_rate": 4.9639649329672324e-05, "loss": 0.9963, "step": 54205 }, { "epoch": 0.05, "learning_rate": 4.963963604122318e-05, "loss": 1.4193, "step": 54206 }, { "epoch": 0.05, "learning_rate": 4.963962275253081e-05, "loss": 1.1808, "step": 54207 }, { "epoch": 0.05, "learning_rate": 4.96396094635952e-05, "loss": 1.2963, "step": 54208 }, { "epoch": 0.05, "learning_rate": 4.963959617441636e-05, "loss": 1.3271, "step": 54209 }, { "epoch": 0.05, "learning_rate": 4.9639582884994296e-05, "loss": 1.0698, "step": 54210 }, { "epoch": 0.05, "learning_rate": 4.9639569595328996e-05, "loss": 1.3498, "step": 54211 }, { "epoch": 0.05, "learning_rate": 4.963955630542047e-05, "loss": 1.6977, "step": 54212 }, { "epoch": 0.05, "learning_rate": 4.96395430152687e-05, "loss": 1.2324, "step": 54213 }, { "epoch": 0.05, "learning_rate": 4.9639529724873704e-05, "loss": 0.978, "step": 54214 }, { "epoch": 0.05, "learning_rate": 4.963951643423548e-05, "loss": 1.2695, "step": 54215 }, { "epoch": 0.05, "learning_rate": 4.9639503143354024e-05, "loss": 1.0723, "step": 54216 }, { "epoch": 0.05, "learning_rate": 4.963948985222934e-05, "loss": 0.7779, "step": 54217 }, { "epoch": 0.05, "learning_rate": 4.963947656086142e-05, "loss": 1.0439, "step": 54218 }, { "epoch": 0.05, "learning_rate": 4.9639463269250276e-05, "loss": 1.2308, "step": 54219 }, { "epoch": 0.05, "learning_rate": 4.96394499773959e-05, "loss": 1.6489, "step": 54220 }, { "epoch": 0.05, "learning_rate": 4.963943668529829e-05, "loss": 1.7933, "step": 54221 }, { "epoch": 0.05, "learning_rate": 4.963942339295745e-05, "loss": 1.8165, "step": 54222 }, { "epoch": 0.05, "learning_rate": 4.963941010037338e-05, "loss": 1.7205, "step": 54223 }, { "epoch": 0.05, "learning_rate": 4.963939680754608e-05, "loss": 1.5144, "step": 54224 }, { "epoch": 0.05, "learning_rate": 4.963938351447555e-05, "loss": 1.3167, "step": 54225 }, { "epoch": 0.05, "learning_rate": 4.9639370221161796e-05, "loss": 1.3155, "step": 54226 }, { "epoch": 0.05, "learning_rate": 4.9639356927604806e-05, "loss": 1.6586, "step": 54227 }, { "epoch": 0.05, "learning_rate": 4.963934363380459e-05, "loss": 1.3782, "step": 54228 }, { "epoch": 0.05, "learning_rate": 4.963933033976115e-05, "loss": 1.5057, "step": 54229 }, { "epoch": 0.05, "learning_rate": 4.9639317045474465e-05, "loss": 1.3073, "step": 54230 }, { "epoch": 0.05, "learning_rate": 4.9639303750944567e-05, "loss": 1.4427, "step": 54231 }, { "epoch": 0.05, "learning_rate": 4.963929045617143e-05, "loss": 1.446, "step": 54232 }, { "epoch": 0.05, "learning_rate": 4.9639277161155074e-05, "loss": 1.0795, "step": 54233 }, { "epoch": 0.05, "learning_rate": 4.963926386589548e-05, "loss": 1.0663, "step": 54234 }, { "epoch": 0.05, "learning_rate": 4.963925057039266e-05, "loss": 1.3753, "step": 54235 }, { "epoch": 0.05, "learning_rate": 4.963923727464661e-05, "loss": 1.2026, "step": 54236 }, { "epoch": 0.05, "learning_rate": 4.963922397865734e-05, "loss": 1.2611, "step": 54237 }, { "epoch": 0.05, "learning_rate": 4.963921068242484e-05, "loss": 1.2413, "step": 54238 }, { "epoch": 0.05, "learning_rate": 4.96391973859491e-05, "loss": 1.2304, "step": 54239 }, { "epoch": 0.05, "learning_rate": 4.9639184089230144e-05, "loss": 1.1767, "step": 54240 }, { "epoch": 0.05, "learning_rate": 4.9639170792267966e-05, "loss": 1.1084, "step": 54241 }, { "epoch": 0.05, "learning_rate": 4.963915749506255e-05, "loss": 1.1078, "step": 54242 }, { "epoch": 0.05, "learning_rate": 4.963914419761391e-05, "loss": 0.9548, "step": 54243 }, { "epoch": 0.05, "learning_rate": 4.9639130899922036e-05, "loss": 1.2094, "step": 54244 }, { "epoch": 0.05, "learning_rate": 4.963911760198694e-05, "loss": 0.6417, "step": 54245 }, { "epoch": 0.05, "learning_rate": 4.9639104303808615e-05, "loss": 0.9386, "step": 54246 }, { "epoch": 0.05, "learning_rate": 4.963909100538707e-05, "loss": 1.295, "step": 54247 }, { "epoch": 0.05, "learning_rate": 4.963907770672229e-05, "loss": 1.3853, "step": 54248 }, { "epoch": 0.05, "learning_rate": 4.963906440781429e-05, "loss": 1.1159, "step": 54249 }, { "epoch": 0.05, "learning_rate": 4.963905110866306e-05, "loss": 1.253, "step": 54250 }, { "epoch": 0.05, "learning_rate": 4.9639037809268605e-05, "loss": 0.9892, "step": 54251 }, { "epoch": 0.05, "learning_rate": 4.963902450963092e-05, "loss": 1.722, "step": 54252 }, { "epoch": 0.05, "learning_rate": 4.963901120975002e-05, "loss": 1.2816, "step": 54253 }, { "epoch": 0.05, "learning_rate": 4.9638997909625886e-05, "loss": 1.0474, "step": 54254 }, { "epoch": 0.05, "learning_rate": 4.963898460925852e-05, "loss": 1.2999, "step": 54255 }, { "epoch": 0.05, "learning_rate": 4.9638971308647934e-05, "loss": 0.7926, "step": 54256 }, { "epoch": 0.05, "learning_rate": 4.963895800779412e-05, "loss": 0.9779, "step": 54257 }, { "epoch": 0.05, "learning_rate": 4.963894470669709e-05, "loss": 1.2794, "step": 54258 }, { "epoch": 0.05, "learning_rate": 4.963893140535683e-05, "loss": 1.2367, "step": 54259 }, { "epoch": 0.05, "learning_rate": 4.963891810377333e-05, "loss": 0.3723, "step": 54260 }, { "epoch": 0.05, "learning_rate": 4.963890480194663e-05, "loss": 1.0484, "step": 54261 }, { "epoch": 0.05, "learning_rate": 4.9638891499876696e-05, "loss": 0.8737, "step": 54262 }, { "epoch": 0.05, "learning_rate": 4.963887819756353e-05, "loss": 0.8369, "step": 54263 }, { "epoch": 0.05, "learning_rate": 4.963886489500714e-05, "loss": 1.0927, "step": 54264 }, { "epoch": 0.05, "learning_rate": 4.963885159220753e-05, "loss": 1.357, "step": 54265 }, { "epoch": 0.05, "learning_rate": 4.96388382891647e-05, "loss": 0.8601, "step": 54266 }, { "epoch": 0.05, "learning_rate": 4.963882498587864e-05, "loss": 1.1255, "step": 54267 }, { "epoch": 0.05, "learning_rate": 4.963881168234936e-05, "loss": 1.1677, "step": 54268 }, { "epoch": 0.05, "learning_rate": 4.963879837857685e-05, "loss": 1.2653, "step": 54269 }, { "epoch": 0.05, "learning_rate": 4.9638785074561114e-05, "loss": 1.1587, "step": 54270 }, { "epoch": 0.05, "learning_rate": 4.963877177030216e-05, "loss": 1.047, "step": 54271 }, { "epoch": 0.05, "learning_rate": 4.963875846579998e-05, "loss": 1.1641, "step": 54272 }, { "epoch": 0.05, "learning_rate": 4.9638745161054586e-05, "loss": 1.296, "step": 54273 }, { "epoch": 0.05, "learning_rate": 4.963873185606596e-05, "loss": 1.1147, "step": 54274 }, { "epoch": 0.05, "learning_rate": 4.963871855083411e-05, "loss": 1.0323, "step": 54275 }, { "epoch": 0.05, "learning_rate": 4.963870524535904e-05, "loss": 1.3664, "step": 54276 }, { "epoch": 0.05, "learning_rate": 4.9638691939640745e-05, "loss": 0.7657, "step": 54277 }, { "epoch": 0.05, "learning_rate": 4.963867863367923e-05, "loss": 0.823, "step": 54278 }, { "epoch": 0.05, "learning_rate": 4.9638665327474485e-05, "loss": 0.9039, "step": 54279 }, { "epoch": 0.05, "learning_rate": 4.963865202102652e-05, "loss": 1.362, "step": 54280 }, { "epoch": 0.05, "learning_rate": 4.9638638714335336e-05, "loss": 1.2463, "step": 54281 }, { "epoch": 0.05, "learning_rate": 4.963862540740094e-05, "loss": 1.2467, "step": 54282 }, { "epoch": 0.05, "learning_rate": 4.9638612100223305e-05, "loss": 0.9227, "step": 54283 }, { "epoch": 0.05, "learning_rate": 4.963859879280245e-05, "loss": 1.5832, "step": 54284 }, { "epoch": 0.05, "learning_rate": 4.963858548513838e-05, "loss": 1.4311, "step": 54285 }, { "epoch": 0.05, "learning_rate": 4.963857217723109e-05, "loss": 0.7302, "step": 54286 }, { "epoch": 0.05, "learning_rate": 4.963855886908057e-05, "loss": 0.4076, "step": 54287 }, { "epoch": 0.05, "learning_rate": 4.963854556068683e-05, "loss": 0.3922, "step": 54288 }, { "epoch": 0.05, "learning_rate": 4.9638532252049876e-05, "loss": 0.5158, "step": 54289 }, { "epoch": 0.05, "learning_rate": 4.96385189431697e-05, "loss": 0.4914, "step": 54290 }, { "epoch": 0.05, "learning_rate": 4.963850563404629e-05, "loss": 0.3283, "step": 54291 }, { "epoch": 0.05, "learning_rate": 4.9638492324679676e-05, "loss": 0.2418, "step": 54292 }, { "epoch": 0.05, "learning_rate": 4.9638479015069826e-05, "loss": 0.3109, "step": 54293 }, { "epoch": 0.05, "learning_rate": 4.963846570521676e-05, "loss": 0.3221, "step": 54294 }, { "epoch": 0.05, "learning_rate": 4.9638452395120485e-05, "loss": 0.5523, "step": 54295 }, { "epoch": 0.05, "learning_rate": 4.963843908478097e-05, "loss": 1.2886, "step": 54296 }, { "epoch": 0.05, "learning_rate": 4.963842577419825e-05, "loss": 1.367, "step": 54297 }, { "epoch": 0.05, "learning_rate": 4.963841246337231e-05, "loss": 0.7704, "step": 54298 }, { "epoch": 0.05, "learning_rate": 4.963839915230314e-05, "loss": 1.2632, "step": 54299 }, { "epoch": 0.05, "learning_rate": 4.963838584099075e-05, "loss": 1.1509, "step": 54300 }, { "epoch": 0.05, "learning_rate": 4.963837252943515e-05, "loss": 0.825, "step": 54301 }, { "epoch": 0.05, "learning_rate": 4.9638359217636324e-05, "loss": 1.3007, "step": 54302 }, { "epoch": 0.05, "learning_rate": 4.963834590559428e-05, "loss": 0.9345, "step": 54303 }, { "epoch": 0.05, "learning_rate": 4.963833259330902e-05, "loss": 0.9701, "step": 54304 }, { "epoch": 0.05, "learning_rate": 4.963831928078054e-05, "loss": 0.8983, "step": 54305 }, { "epoch": 0.05, "learning_rate": 4.963830596800884e-05, "loss": 1.4358, "step": 54306 }, { "epoch": 0.05, "learning_rate": 4.963829265499391e-05, "loss": 0.9022, "step": 54307 }, { "epoch": 0.05, "learning_rate": 4.963827934173577e-05, "loss": 0.9832, "step": 54308 }, { "epoch": 0.05, "learning_rate": 4.9638266028234416e-05, "loss": 1.1354, "step": 54309 }, { "epoch": 0.05, "learning_rate": 4.963825271448984e-05, "loss": 1.1825, "step": 54310 }, { "epoch": 0.05, "learning_rate": 4.963823940050205e-05, "loss": 1.4718, "step": 54311 }, { "epoch": 0.05, "learning_rate": 4.963822608627103e-05, "loss": 1.1283, "step": 54312 }, { "epoch": 0.05, "learning_rate": 4.96382127717968e-05, "loss": 1.1462, "step": 54313 }, { "epoch": 0.05, "learning_rate": 4.963819945707935e-05, "loss": 1.4296, "step": 54314 }, { "epoch": 0.05, "learning_rate": 4.963818614211868e-05, "loss": 1.015, "step": 54315 }, { "epoch": 0.05, "learning_rate": 4.9638172826914795e-05, "loss": 1.2381, "step": 54316 }, { "epoch": 0.05, "learning_rate": 4.9638159511467696e-05, "loss": 1.061, "step": 54317 }, { "epoch": 0.05, "learning_rate": 4.9638146195777375e-05, "loss": 1.1064, "step": 54318 }, { "epoch": 0.05, "learning_rate": 4.9638132879843835e-05, "loss": 0.5213, "step": 54319 }, { "epoch": 0.05, "learning_rate": 4.963811956366708e-05, "loss": 0.8533, "step": 54320 }, { "epoch": 0.05, "learning_rate": 4.9638106247247106e-05, "loss": 1.826, "step": 54321 }, { "epoch": 0.05, "learning_rate": 4.963809293058392e-05, "loss": 1.0495, "step": 54322 }, { "epoch": 0.05, "learning_rate": 4.9638079613677516e-05, "loss": 0.4508, "step": 54323 }, { "epoch": 0.05, "learning_rate": 4.963806629652789e-05, "loss": 0.5153, "step": 54324 }, { "epoch": 0.05, "learning_rate": 4.963805297913505e-05, "loss": 0.4937, "step": 54325 }, { "epoch": 0.05, "learning_rate": 4.9638039661498994e-05, "loss": 0.3748, "step": 54326 }, { "epoch": 0.05, "learning_rate": 4.963802634361972e-05, "loss": 0.8692, "step": 54327 }, { "epoch": 0.05, "learning_rate": 4.9638013025497234e-05, "loss": 1.3576, "step": 54328 }, { "epoch": 0.05, "learning_rate": 4.963799970713153e-05, "loss": 1.0085, "step": 54329 }, { "epoch": 0.05, "learning_rate": 4.9637986388522605e-05, "loss": 1.3587, "step": 54330 }, { "epoch": 0.05, "learning_rate": 4.963797306967047e-05, "loss": 1.1544, "step": 54331 }, { "epoch": 0.05, "learning_rate": 4.963795975057512e-05, "loss": 1.3092, "step": 54332 }, { "epoch": 0.05, "learning_rate": 4.963794643123655e-05, "loss": 0.6714, "step": 54333 }, { "epoch": 0.05, "learning_rate": 4.963793311165476e-05, "loss": 0.2636, "step": 54334 }, { "epoch": 0.05, "learning_rate": 4.963791979182977e-05, "loss": 0.2708, "step": 54335 }, { "epoch": 0.05, "learning_rate": 4.963790647176155e-05, "loss": 0.2739, "step": 54336 }, { "epoch": 0.05, "learning_rate": 4.963789315145012e-05, "loss": 0.1915, "step": 54337 }, { "epoch": 0.05, "learning_rate": 4.963787983089548e-05, "loss": 0.3925, "step": 54338 }, { "epoch": 0.05, "learning_rate": 4.963786651009762e-05, "loss": 0.8357, "step": 54339 }, { "epoch": 0.05, "learning_rate": 4.963785318905655e-05, "loss": 1.4369, "step": 54340 }, { "epoch": 0.05, "learning_rate": 4.9637839867772265e-05, "loss": 0.9639, "step": 54341 }, { "epoch": 0.05, "learning_rate": 4.963782654624476e-05, "loss": 1.0491, "step": 54342 }, { "epoch": 0.05, "learning_rate": 4.963781322447405e-05, "loss": 0.7748, "step": 54343 }, { "epoch": 0.05, "learning_rate": 4.963779990246011e-05, "loss": 1.0329, "step": 54344 }, { "epoch": 0.05, "learning_rate": 4.963778658020297e-05, "loss": 1.4971, "step": 54345 }, { "epoch": 0.05, "learning_rate": 4.963777325770261e-05, "loss": 0.952, "step": 54346 }, { "epoch": 0.05, "learning_rate": 4.9637759934959035e-05, "loss": 1.3532, "step": 54347 }, { "epoch": 0.05, "learning_rate": 4.963774661197226e-05, "loss": 1.2088, "step": 54348 }, { "epoch": 0.05, "learning_rate": 4.963773328874226e-05, "loss": 0.8871, "step": 54349 }, { "epoch": 0.05, "learning_rate": 4.963771996526905e-05, "loss": 0.9345, "step": 54350 }, { "epoch": 0.05, "learning_rate": 4.963770664155262e-05, "loss": 1.0902, "step": 54351 }, { "epoch": 0.05, "learning_rate": 4.963769331759298e-05, "loss": 1.1069, "step": 54352 }, { "epoch": 0.05, "learning_rate": 4.963767999339013e-05, "loss": 1.1407, "step": 54353 }, { "epoch": 0.05, "learning_rate": 4.963766666894407e-05, "loss": 1.1265, "step": 54354 }, { "epoch": 0.05, "learning_rate": 4.963765334425479e-05, "loss": 1.0469, "step": 54355 }, { "epoch": 0.05, "learning_rate": 4.96376400193223e-05, "loss": 1.0938, "step": 54356 }, { "epoch": 0.05, "learning_rate": 4.96376266941466e-05, "loss": 1.2538, "step": 54357 }, { "epoch": 0.05, "learning_rate": 4.963761336872769e-05, "loss": 1.2313, "step": 54358 }, { "epoch": 0.05, "learning_rate": 4.9637600043065554e-05, "loss": 1.3587, "step": 54359 }, { "epoch": 0.05, "learning_rate": 4.963758671716022e-05, "loss": 1.1262, "step": 54360 }, { "epoch": 0.05, "learning_rate": 4.963757339101167e-05, "loss": 1.0776, "step": 54361 }, { "epoch": 0.05, "learning_rate": 4.963756006461991e-05, "loss": 1.0759, "step": 54362 }, { "epoch": 0.05, "learning_rate": 4.9637546737984936e-05, "loss": 1.1766, "step": 54363 }, { "epoch": 0.05, "learning_rate": 4.9637533411106754e-05, "loss": 1.3846, "step": 54364 }, { "epoch": 0.05, "learning_rate": 4.963752008398536e-05, "loss": 1.0446, "step": 54365 }, { "epoch": 0.05, "learning_rate": 4.963750675662074e-05, "loss": 1.3644, "step": 54366 }, { "epoch": 0.05, "learning_rate": 4.9637493429012926e-05, "loss": 1.1137, "step": 54367 }, { "epoch": 0.05, "learning_rate": 4.96374801011619e-05, "loss": 1.3482, "step": 54368 }, { "epoch": 0.05, "learning_rate": 4.963746677306766e-05, "loss": 1.0884, "step": 54369 }, { "epoch": 0.05, "learning_rate": 4.963745344473021e-05, "loss": 0.8703, "step": 54370 }, { "epoch": 0.05, "learning_rate": 4.9637440116149545e-05, "loss": 0.7904, "step": 54371 }, { "epoch": 0.05, "learning_rate": 4.9637426787325674e-05, "loss": 0.7872, "step": 54372 }, { "epoch": 0.05, "learning_rate": 4.9637413458258596e-05, "loss": 1.1209, "step": 54373 }, { "epoch": 0.05, "learning_rate": 4.96374001289483e-05, "loss": 1.0124, "step": 54374 }, { "epoch": 0.05, "learning_rate": 4.96373867993948e-05, "loss": 1.1981, "step": 54375 }, { "epoch": 0.05, "learning_rate": 4.963737346959809e-05, "loss": 1.0125, "step": 54376 }, { "epoch": 0.05, "learning_rate": 4.963736013955816e-05, "loss": 0.6444, "step": 54377 }, { "epoch": 0.05, "learning_rate": 4.963734680927503e-05, "loss": 0.905, "step": 54378 }, { "epoch": 0.05, "learning_rate": 4.963733347874869e-05, "loss": 0.5086, "step": 54379 }, { "epoch": 0.05, "learning_rate": 4.963732014797914e-05, "loss": 1.8386, "step": 54380 }, { "epoch": 0.05, "learning_rate": 4.963730681696638e-05, "loss": 1.506, "step": 54381 }, { "epoch": 0.05, "learning_rate": 4.963729348571041e-05, "loss": 1.1771, "step": 54382 }, { "epoch": 0.05, "learning_rate": 4.963728015421124e-05, "loss": 1.1098, "step": 54383 }, { "epoch": 0.05, "learning_rate": 4.963726682246885e-05, "loss": 1.2093, "step": 54384 }, { "epoch": 0.05, "learning_rate": 4.963725349048325e-05, "loss": 1.1159, "step": 54385 }, { "epoch": 0.05, "learning_rate": 4.963724015825445e-05, "loss": 1.1846, "step": 54386 }, { "epoch": 0.05, "learning_rate": 4.963722682578243e-05, "loss": 1.2476, "step": 54387 }, { "epoch": 0.05, "learning_rate": 4.963721349306721e-05, "loss": 1.2608, "step": 54388 }, { "epoch": 0.05, "learning_rate": 4.963720016010879e-05, "loss": 0.8847, "step": 54389 }, { "epoch": 0.05, "learning_rate": 4.963718682690715e-05, "loss": 0.8976, "step": 54390 }, { "epoch": 0.05, "learning_rate": 4.9637173493462306e-05, "loss": 0.6746, "step": 54391 }, { "epoch": 0.05, "learning_rate": 4.9637160159774256e-05, "loss": 1.2624, "step": 54392 }, { "epoch": 0.05, "learning_rate": 4.963714682584299e-05, "loss": 1.0062, "step": 54393 }, { "epoch": 0.05, "learning_rate": 4.963713349166852e-05, "loss": 0.8541, "step": 54394 }, { "epoch": 0.05, "learning_rate": 4.963712015725085e-05, "loss": 0.9602, "step": 54395 }, { "epoch": 0.05, "learning_rate": 4.9637106822589964e-05, "loss": 0.6805, "step": 54396 }, { "epoch": 0.05, "learning_rate": 4.9637093487685874e-05, "loss": 0.2923, "step": 54397 }, { "epoch": 0.05, "learning_rate": 4.963708015253858e-05, "loss": 0.2915, "step": 54398 }, { "epoch": 0.05, "learning_rate": 4.963706681714807e-05, "loss": 0.4189, "step": 54399 }, { "epoch": 0.05, "learning_rate": 4.963705348151436e-05, "loss": 0.9449, "step": 54400 }, { "epoch": 0.05, "learning_rate": 4.963704014563745e-05, "loss": 1.0408, "step": 54401 }, { "epoch": 0.05, "learning_rate": 4.963702680951732e-05, "loss": 1.2513, "step": 54402 }, { "epoch": 0.05, "learning_rate": 4.9637013473153995e-05, "loss": 1.1792, "step": 54403 }, { "epoch": 0.05, "learning_rate": 4.963700013654746e-05, "loss": 1.1583, "step": 54404 }, { "epoch": 0.05, "learning_rate": 4.963698679969772e-05, "loss": 1.3136, "step": 54405 }, { "epoch": 0.05, "learning_rate": 4.9636973462604766e-05, "loss": 1.4212, "step": 54406 }, { "epoch": 0.05, "learning_rate": 4.9636960125268614e-05, "loss": 1.3134, "step": 54407 }, { "epoch": 0.05, "learning_rate": 4.9636946787689255e-05, "loss": 1.274, "step": 54408 }, { "epoch": 0.05, "learning_rate": 4.963693344986668e-05, "loss": 1.1698, "step": 54409 }, { "epoch": 0.05, "learning_rate": 4.9636920111800916e-05, "loss": 1.0814, "step": 54410 }, { "epoch": 0.05, "learning_rate": 4.963690677349194e-05, "loss": 1.0357, "step": 54411 }, { "epoch": 0.05, "learning_rate": 4.9636893434939756e-05, "loss": 1.146, "step": 54412 }, { "epoch": 0.05, "learning_rate": 4.9636880096144376e-05, "loss": 0.8801, "step": 54413 }, { "epoch": 0.05, "learning_rate": 4.963686675710578e-05, "loss": 0.9859, "step": 54414 }, { "epoch": 0.05, "learning_rate": 4.963685341782399e-05, "loss": 1.3257, "step": 54415 }, { "epoch": 0.05, "learning_rate": 4.963684007829898e-05, "loss": 1.2091, "step": 54416 }, { "epoch": 0.05, "learning_rate": 4.963682673853078e-05, "loss": 0.9792, "step": 54417 }, { "epoch": 0.05, "learning_rate": 4.963681339851938e-05, "loss": 1.3335, "step": 54418 }, { "epoch": 0.05, "learning_rate": 4.963680005826476e-05, "loss": 1.2134, "step": 54419 }, { "epoch": 0.05, "learning_rate": 4.9636786717766945e-05, "loss": 1.0808, "step": 54420 }, { "epoch": 0.05, "learning_rate": 4.963677337702592e-05, "loss": 1.1688, "step": 54421 }, { "epoch": 0.05, "learning_rate": 4.9636760036041695e-05, "loss": 1.2513, "step": 54422 }, { "epoch": 0.05, "learning_rate": 4.963674669481426e-05, "loss": 1.0959, "step": 54423 }, { "epoch": 0.05, "learning_rate": 4.963673335334363e-05, "loss": 1.0746, "step": 54424 }, { "epoch": 0.05, "learning_rate": 4.96367200116298e-05, "loss": 1.3038, "step": 54425 }, { "epoch": 0.05, "learning_rate": 4.963670666967276e-05, "loss": 0.7539, "step": 54426 }, { "epoch": 0.05, "learning_rate": 4.963669332747252e-05, "loss": 0.7633, "step": 54427 }, { "epoch": 0.05, "learning_rate": 4.963667998502907e-05, "loss": 0.7228, "step": 54428 }, { "epoch": 0.05, "learning_rate": 4.963666664234242e-05, "loss": 0.5062, "step": 54429 }, { "epoch": 0.05, "learning_rate": 4.963665329941257e-05, "loss": 0.2385, "step": 54430 }, { "epoch": 0.05, "learning_rate": 4.9636639956239514e-05, "loss": 0.2227, "step": 54431 }, { "epoch": 0.05, "learning_rate": 4.963662661282326e-05, "loss": 0.4226, "step": 54432 }, { "epoch": 0.05, "learning_rate": 4.96366132691638e-05, "loss": 0.9469, "step": 54433 }, { "epoch": 0.05, "learning_rate": 4.963659992526114e-05, "loss": 1.2225, "step": 54434 }, { "epoch": 0.05, "learning_rate": 4.9636586581115275e-05, "loss": 1.2408, "step": 54435 }, { "epoch": 0.05, "learning_rate": 4.963657323672621e-05, "loss": 1.0544, "step": 54436 }, { "epoch": 0.05, "learning_rate": 4.963655989209394e-05, "loss": 0.9299, "step": 54437 }, { "epoch": 0.05, "learning_rate": 4.963654654721848e-05, "loss": 1.063, "step": 54438 }, { "epoch": 0.05, "learning_rate": 4.96365332020998e-05, "loss": 1.076, "step": 54439 }, { "epoch": 0.05, "learning_rate": 4.963651985673793e-05, "loss": 1.2428, "step": 54440 }, { "epoch": 0.05, "learning_rate": 4.963650651113286e-05, "loss": 1.4062, "step": 54441 }, { "epoch": 0.05, "learning_rate": 4.963649316528458e-05, "loss": 1.0335, "step": 54442 }, { "epoch": 0.05, "learning_rate": 4.96364798191931e-05, "loss": 1.6265, "step": 54443 }, { "epoch": 0.05, "learning_rate": 4.963646647285843e-05, "loss": 1.398, "step": 54444 }, { "epoch": 0.05, "learning_rate": 4.9636453126280554e-05, "loss": 1.0991, "step": 54445 }, { "epoch": 0.05, "learning_rate": 4.9636439779459476e-05, "loss": 1.2186, "step": 54446 }, { "epoch": 0.05, "learning_rate": 4.96364264323952e-05, "loss": 1.2667, "step": 54447 }, { "epoch": 0.05, "learning_rate": 4.963641308508772e-05, "loss": 1.0739, "step": 54448 }, { "epoch": 0.05, "learning_rate": 4.9636399737537034e-05, "loss": 0.8082, "step": 54449 }, { "epoch": 0.05, "learning_rate": 4.9636386389743156e-05, "loss": 1.1869, "step": 54450 }, { "epoch": 0.05, "learning_rate": 4.963637304170608e-05, "loss": 1.4696, "step": 54451 }, { "epoch": 0.05, "learning_rate": 4.963635969342579e-05, "loss": 1.8612, "step": 54452 }, { "epoch": 0.05, "learning_rate": 4.963634634490232e-05, "loss": 0.9381, "step": 54453 }, { "epoch": 0.05, "learning_rate": 4.9636332996135634e-05, "loss": 1.2057, "step": 54454 }, { "epoch": 0.05, "learning_rate": 4.963631964712576e-05, "loss": 1.2705, "step": 54455 }, { "epoch": 0.05, "learning_rate": 4.963630629787268e-05, "loss": 0.938, "step": 54456 }, { "epoch": 0.05, "learning_rate": 4.96362929483764e-05, "loss": 1.4952, "step": 54457 }, { "epoch": 0.05, "learning_rate": 4.9636279598636924e-05, "loss": 1.1039, "step": 54458 }, { "epoch": 0.05, "learning_rate": 4.963626624865425e-05, "loss": 1.134, "step": 54459 }, { "epoch": 0.05, "learning_rate": 4.963625289842838e-05, "loss": 1.3429, "step": 54460 }, { "epoch": 0.05, "learning_rate": 4.9636239547959306e-05, "loss": 1.2597, "step": 54461 }, { "epoch": 0.05, "learning_rate": 4.963622619724704e-05, "loss": 1.2196, "step": 54462 }, { "epoch": 0.05, "learning_rate": 4.963621284629156e-05, "loss": 0.9487, "step": 54463 }, { "epoch": 0.05, "learning_rate": 4.96361994950929e-05, "loss": 0.8673, "step": 54464 }, { "epoch": 0.05, "learning_rate": 4.9636186143651034e-05, "loss": 0.4639, "step": 54465 }, { "epoch": 0.05, "learning_rate": 4.963617279196597e-05, "loss": 0.226, "step": 54466 }, { "epoch": 0.05, "learning_rate": 4.963615944003771e-05, "loss": 0.1956, "step": 54467 }, { "epoch": 0.05, "learning_rate": 4.963614608786625e-05, "loss": 0.2329, "step": 54468 }, { "epoch": 0.05, "learning_rate": 4.963613273545159e-05, "loss": 0.1195, "step": 54469 }, { "epoch": 0.05, "learning_rate": 4.9636119382793734e-05, "loss": 0.2743, "step": 54470 }, { "epoch": 0.05, "learning_rate": 4.963610602989268e-05, "loss": 0.8718, "step": 54471 }, { "epoch": 0.05, "learning_rate": 4.9636092676748434e-05, "loss": 1.1649, "step": 54472 }, { "epoch": 0.05, "learning_rate": 4.963607932336099e-05, "loss": 1.3303, "step": 54473 }, { "epoch": 0.05, "learning_rate": 4.963606596973034e-05, "loss": 1.185, "step": 54474 }, { "epoch": 0.05, "learning_rate": 4.96360526158565e-05, "loss": 1.1356, "step": 54475 }, { "epoch": 0.05, "learning_rate": 4.9636039261739466e-05, "loss": 1.1952, "step": 54476 }, { "epoch": 0.05, "learning_rate": 4.963602590737923e-05, "loss": 1.2092, "step": 54477 }, { "epoch": 0.05, "learning_rate": 4.96360125527758e-05, "loss": 1.4023, "step": 54478 }, { "epoch": 0.05, "learning_rate": 4.963599919792917e-05, "loss": 1.0865, "step": 54479 }, { "epoch": 0.05, "learning_rate": 4.963598584283935e-05, "loss": 1.1451, "step": 54480 }, { "epoch": 0.05, "learning_rate": 4.963597248750633e-05, "loss": 1.0286, "step": 54481 }, { "epoch": 0.05, "learning_rate": 4.9635959131930116e-05, "loss": 1.0678, "step": 54482 }, { "epoch": 0.05, "learning_rate": 4.96359457761107e-05, "loss": 1.1544, "step": 54483 }, { "epoch": 0.05, "learning_rate": 4.9635932420048095e-05, "loss": 1.1995, "step": 54484 }, { "epoch": 0.05, "learning_rate": 4.9635919063742294e-05, "loss": 1.0885, "step": 54485 }, { "epoch": 0.05, "learning_rate": 4.96359057071933e-05, "loss": 0.7633, "step": 54486 }, { "epoch": 0.05, "learning_rate": 4.9635892350401106e-05, "loss": 1.1484, "step": 54487 }, { "epoch": 0.05, "learning_rate": 4.963587899336571e-05, "loss": 0.9533, "step": 54488 }, { "epoch": 0.05, "learning_rate": 4.963586563608713e-05, "loss": 1.1009, "step": 54489 }, { "epoch": 0.05, "learning_rate": 4.963585227856535e-05, "loss": 0.9635, "step": 54490 }, { "epoch": 0.05, "learning_rate": 4.963583892080038e-05, "loss": 0.7484, "step": 54491 }, { "epoch": 0.05, "learning_rate": 4.9635825562792216e-05, "loss": 1.3787, "step": 54492 }, { "epoch": 0.05, "learning_rate": 4.9635812204540855e-05, "loss": 1.235, "step": 54493 }, { "epoch": 0.05, "learning_rate": 4.96357988460463e-05, "loss": 1.4254, "step": 54494 }, { "epoch": 0.05, "learning_rate": 4.963578548730855e-05, "loss": 0.3611, "step": 54495 }, { "epoch": 0.05, "learning_rate": 4.9635772128327606e-05, "loss": 1.3352, "step": 54496 }, { "epoch": 0.05, "learning_rate": 4.963575876910347e-05, "loss": 1.2897, "step": 54497 }, { "epoch": 0.05, "learning_rate": 4.963574540963614e-05, "loss": 0.9702, "step": 54498 }, { "epoch": 0.05, "learning_rate": 4.963573204992561e-05, "loss": 1.0017, "step": 54499 }, { "epoch": 0.05, "learning_rate": 4.963571868997189e-05, "loss": 1.1473, "step": 54500 }, { "epoch": 0.05, "eval_loss": 1.0877758264541626, "eval_runtime": 127.4671, "eval_samples_per_second": 10.866, "eval_steps_per_second": 5.437, "step": 54500 }, { "epoch": 0.05, "learning_rate": 4.9635705329774975e-05, "loss": 1.1023, "step": 54501 }, { "epoch": 0.05, "learning_rate": 4.963569196933487e-05, "loss": 3.9714, "step": 54502 }, { "epoch": 0.05, "learning_rate": 4.9635678608651573e-05, "loss": 3.4567, "step": 54503 }, { "epoch": 0.05, "learning_rate": 4.963566524772508e-05, "loss": 1.5288, "step": 54504 }, { "epoch": 0.05, "learning_rate": 4.963565188655539e-05, "loss": 1.7376, "step": 54505 }, { "epoch": 0.05, "learning_rate": 4.963563852514251e-05, "loss": 1.6179, "step": 54506 }, { "epoch": 0.05, "learning_rate": 4.9635625163486444e-05, "loss": 1.7473, "step": 54507 }, { "epoch": 0.05, "learning_rate": 4.9635611801587176e-05, "loss": 1.3452, "step": 54508 }, { "epoch": 0.05, "learning_rate": 4.963559843944472e-05, "loss": 1.3278, "step": 54509 }, { "epoch": 0.05, "learning_rate": 4.9635585077059075e-05, "loss": 0.9957, "step": 54510 }, { "epoch": 0.05, "learning_rate": 4.963557171443024e-05, "loss": 0.672, "step": 54511 }, { "epoch": 0.05, "learning_rate": 4.96355583515582e-05, "loss": 1.391, "step": 54512 }, { "epoch": 0.05, "learning_rate": 4.963554498844299e-05, "loss": 0.8921, "step": 54513 }, { "epoch": 0.05, "learning_rate": 4.9635531625084566e-05, "loss": 1.1579, "step": 54514 }, { "epoch": 0.05, "learning_rate": 4.963551826148296e-05, "loss": 0.5768, "step": 54515 }, { "epoch": 0.05, "learning_rate": 4.9635504897638165e-05, "loss": 1.3384, "step": 54516 }, { "epoch": 0.05, "learning_rate": 4.963549153355017e-05, "loss": 0.7912, "step": 54517 }, { "epoch": 0.05, "learning_rate": 4.9635478169218984e-05, "loss": 1.0738, "step": 54518 }, { "epoch": 0.05, "learning_rate": 4.963546480464462e-05, "loss": 1.2083, "step": 54519 }, { "epoch": 0.05, "learning_rate": 4.963545143982705e-05, "loss": 1.38, "step": 54520 }, { "epoch": 0.05, "learning_rate": 4.96354380747663e-05, "loss": 1.3453, "step": 54521 }, { "epoch": 0.05, "learning_rate": 4.963542470946235e-05, "loss": 0.8786, "step": 54522 }, { "epoch": 0.05, "learning_rate": 4.963541134391522e-05, "loss": 1.5817, "step": 54523 }, { "epoch": 0.05, "learning_rate": 4.96353979781249e-05, "loss": 1.0085, "step": 54524 }, { "epoch": 0.05, "learning_rate": 4.963538461209138e-05, "loss": 0.9142, "step": 54525 }, { "epoch": 0.05, "learning_rate": 4.963537124581468e-05, "loss": 1.1288, "step": 54526 }, { "epoch": 0.05, "learning_rate": 4.963535787929478e-05, "loss": 0.9751, "step": 54527 }, { "epoch": 0.05, "learning_rate": 4.963534451253169e-05, "loss": 1.1158, "step": 54528 }, { "epoch": 0.05, "learning_rate": 4.963533114552542e-05, "loss": 0.947, "step": 54529 }, { "epoch": 0.05, "learning_rate": 4.963531777827595e-05, "loss": 0.9469, "step": 54530 }, { "epoch": 0.05, "learning_rate": 4.96353044107833e-05, "loss": 0.7221, "step": 54531 }, { "epoch": 0.05, "learning_rate": 4.963529104304745e-05, "loss": 1.5067, "step": 54532 }, { "epoch": 0.05, "learning_rate": 4.963527767506842e-05, "loss": 1.0289, "step": 54533 }, { "epoch": 0.05, "learning_rate": 4.963526430684619e-05, "loss": 1.525, "step": 54534 }, { "epoch": 0.05, "learning_rate": 4.963525093838079e-05, "loss": 1.3168, "step": 54535 }, { "epoch": 0.05, "learning_rate": 4.963523756967219e-05, "loss": 1.3727, "step": 54536 }, { "epoch": 0.05, "learning_rate": 4.9635224200720396e-05, "loss": 1.1844, "step": 54537 }, { "epoch": 0.05, "learning_rate": 4.9635210831525423e-05, "loss": 1.174, "step": 54538 }, { "epoch": 0.05, "learning_rate": 4.963519746208726e-05, "loss": 1.0846, "step": 54539 }, { "epoch": 0.05, "learning_rate": 4.96351840924059e-05, "loss": 1.3522, "step": 54540 }, { "epoch": 0.05, "learning_rate": 4.963517072248136e-05, "loss": 1.2577, "step": 54541 }, { "epoch": 0.05, "learning_rate": 4.9635157352313635e-05, "loss": 0.9864, "step": 54542 }, { "epoch": 0.05, "learning_rate": 4.963514398190271e-05, "loss": 1.0981, "step": 54543 }, { "epoch": 0.05, "learning_rate": 4.9635130611248605e-05, "loss": 0.4675, "step": 54544 }, { "epoch": 0.05, "learning_rate": 4.963511724035131e-05, "loss": 0.8172, "step": 54545 }, { "epoch": 0.05, "learning_rate": 4.963510386921083e-05, "loss": 1.5603, "step": 54546 }, { "epoch": 0.05, "learning_rate": 4.963509049782716e-05, "loss": 1.1093, "step": 54547 }, { "epoch": 0.05, "learning_rate": 4.9635077126200305e-05, "loss": 0.8846, "step": 54548 }, { "epoch": 0.05, "learning_rate": 4.963506375433027e-05, "loss": 1.213, "step": 54549 }, { "epoch": 0.05, "learning_rate": 4.9635050382217036e-05, "loss": 0.9809, "step": 54550 }, { "epoch": 0.05, "learning_rate": 4.963503700986062e-05, "loss": 1.0497, "step": 54551 }, { "epoch": 0.05, "learning_rate": 4.9635023637261014e-05, "loss": 1.1659, "step": 54552 }, { "epoch": 0.05, "learning_rate": 4.963501026441822e-05, "loss": 1.1774, "step": 54553 }, { "epoch": 0.05, "learning_rate": 4.963499689133225e-05, "loss": 1.0725, "step": 54554 }, { "epoch": 0.05, "learning_rate": 4.963498351800309e-05, "loss": 0.5543, "step": 54555 }, { "epoch": 0.05, "learning_rate": 4.963497014443074e-05, "loss": 0.2856, "step": 54556 }, { "epoch": 0.05, "learning_rate": 4.96349567706152e-05, "loss": 0.4036, "step": 54557 }, { "epoch": 0.05, "learning_rate": 4.963494339655649e-05, "loss": 0.2232, "step": 54558 }, { "epoch": 0.05, "learning_rate": 4.9634930022254574e-05, "loss": 0.4672, "step": 54559 }, { "epoch": 0.05, "learning_rate": 4.9634916647709485e-05, "loss": 0.9341, "step": 54560 }, { "epoch": 0.05, "learning_rate": 4.963490327292121e-05, "loss": 1.194, "step": 54561 }, { "epoch": 0.05, "learning_rate": 4.9634889897889745e-05, "loss": 1.141, "step": 54562 }, { "epoch": 0.05, "learning_rate": 4.9634876522615096e-05, "loss": 1.0158, "step": 54563 }, { "epoch": 0.05, "learning_rate": 4.9634863147097266e-05, "loss": 1.1341, "step": 54564 }, { "epoch": 0.05, "learning_rate": 4.9634849771336244e-05, "loss": 0.7411, "step": 54565 }, { "epoch": 0.05, "learning_rate": 4.963483639533204e-05, "loss": 1.1614, "step": 54566 }, { "epoch": 0.05, "learning_rate": 4.963482301908465e-05, "loss": 1.4024, "step": 54567 }, { "epoch": 0.05, "learning_rate": 4.9634809642594084e-05, "loss": 1.0162, "step": 54568 }, { "epoch": 0.05, "learning_rate": 4.963479626586033e-05, "loss": 1.1692, "step": 54569 }, { "epoch": 0.05, "learning_rate": 4.963478288888339e-05, "loss": 1.3195, "step": 54570 }, { "epoch": 0.05, "learning_rate": 4.963476951166326e-05, "loss": 1.4265, "step": 54571 }, { "epoch": 0.05, "learning_rate": 4.963475613419995e-05, "loss": 1.0592, "step": 54572 }, { "epoch": 0.05, "learning_rate": 4.9634742756493456e-05, "loss": 1.4897, "step": 54573 }, { "epoch": 0.05, "learning_rate": 4.963472937854378e-05, "loss": 1.2809, "step": 54574 }, { "epoch": 0.05, "learning_rate": 4.963471600035092e-05, "loss": 1.9277, "step": 54575 }, { "epoch": 0.05, "learning_rate": 4.963470262191487e-05, "loss": 1.5122, "step": 54576 }, { "epoch": 0.05, "learning_rate": 4.9634689243235646e-05, "loss": 1.1034, "step": 54577 }, { "epoch": 0.05, "learning_rate": 4.963467586431324e-05, "loss": 1.1624, "step": 54578 }, { "epoch": 0.05, "learning_rate": 4.9634662485147645e-05, "loss": 1.1204, "step": 54579 }, { "epoch": 0.05, "learning_rate": 4.9634649105738865e-05, "loss": 1.0857, "step": 54580 }, { "epoch": 0.05, "learning_rate": 4.963463572608691e-05, "loss": 1.4168, "step": 54581 }, { "epoch": 0.05, "learning_rate": 4.963462234619176e-05, "loss": 1.4013, "step": 54582 }, { "epoch": 0.05, "learning_rate": 4.963460896605344e-05, "loss": 1.3266, "step": 54583 }, { "epoch": 0.05, "learning_rate": 4.9634595585671926e-05, "loss": 1.1389, "step": 54584 }, { "epoch": 0.05, "learning_rate": 4.963458220504724e-05, "loss": 1.4211, "step": 54585 }, { "epoch": 0.05, "learning_rate": 4.963456882417937e-05, "loss": 1.2569, "step": 54586 }, { "epoch": 0.05, "learning_rate": 4.963455544306831e-05, "loss": 1.3012, "step": 54587 }, { "epoch": 0.05, "learning_rate": 4.963454206171407e-05, "loss": 0.9284, "step": 54588 }, { "epoch": 0.05, "learning_rate": 4.963452868011665e-05, "loss": 1.018, "step": 54589 }, { "epoch": 0.05, "learning_rate": 4.963451529827606e-05, "loss": 0.7958, "step": 54590 }, { "epoch": 0.05, "learning_rate": 4.9634501916192276e-05, "loss": 0.7897, "step": 54591 }, { "epoch": 0.05, "learning_rate": 4.963448853386531e-05, "loss": 1.304, "step": 54592 }, { "epoch": 0.05, "learning_rate": 4.963447515129517e-05, "loss": 1.3126, "step": 54593 }, { "epoch": 0.05, "learning_rate": 4.963446176848184e-05, "loss": 1.3497, "step": 54594 }, { "epoch": 0.05, "learning_rate": 4.963444838542534e-05, "loss": 1.2054, "step": 54595 }, { "epoch": 0.05, "learning_rate": 4.963443500212564e-05, "loss": 1.3033, "step": 54596 }, { "epoch": 0.05, "learning_rate": 4.9634421618582786e-05, "loss": 1.1227, "step": 54597 }, { "epoch": 0.05, "learning_rate": 4.9634408234796736e-05, "loss": 1.1529, "step": 54598 }, { "epoch": 0.05, "learning_rate": 4.9634394850767506e-05, "loss": 1.2945, "step": 54599 }, { "epoch": 0.05, "learning_rate": 4.9634381466495096e-05, "loss": 0.7352, "step": 54600 }, { "epoch": 0.05, "learning_rate": 4.963436808197951e-05, "loss": 0.1593, "step": 54601 }, { "epoch": 0.05, "learning_rate": 4.963435469722073e-05, "loss": 0.1243, "step": 54602 }, { "epoch": 0.05, "learning_rate": 4.963434131221878e-05, "loss": 0.5335, "step": 54603 }, { "epoch": 0.05, "learning_rate": 4.9634327926973654e-05, "loss": 1.1028, "step": 54604 }, { "epoch": 0.05, "learning_rate": 4.9634314541485346e-05, "loss": 0.8548, "step": 54605 }, { "epoch": 0.05, "learning_rate": 4.963430115575386e-05, "loss": 1.1623, "step": 54606 }, { "epoch": 0.05, "learning_rate": 4.963428776977918e-05, "loss": 1.0044, "step": 54607 }, { "epoch": 0.05, "learning_rate": 4.963427438356134e-05, "loss": 1.0527, "step": 54608 }, { "epoch": 0.05, "learning_rate": 4.9634260997100315e-05, "loss": 1.0616, "step": 54609 }, { "epoch": 0.05, "learning_rate": 4.963424761039611e-05, "loss": 1.0802, "step": 54610 }, { "epoch": 0.05, "learning_rate": 4.963423422344872e-05, "loss": 1.2109, "step": 54611 }, { "epoch": 0.05, "learning_rate": 4.963422083625816e-05, "loss": 1.0149, "step": 54612 }, { "epoch": 0.05, "learning_rate": 4.9634207448824414e-05, "loss": 1.1147, "step": 54613 }, { "epoch": 0.05, "learning_rate": 4.96341940611475e-05, "loss": 1.0224, "step": 54614 }, { "epoch": 0.05, "learning_rate": 4.9634180673227396e-05, "loss": 0.6497, "step": 54615 }, { "epoch": 0.05, "learning_rate": 4.963416728506412e-05, "loss": 1.0393, "step": 54616 }, { "epoch": 0.05, "learning_rate": 4.9634153896657665e-05, "loss": 1.2049, "step": 54617 }, { "epoch": 0.05, "learning_rate": 4.963414050800803e-05, "loss": 1.3135, "step": 54618 }, { "epoch": 0.05, "learning_rate": 4.9634127119115215e-05, "loss": 1.0278, "step": 54619 }, { "epoch": 0.05, "learning_rate": 4.963411372997922e-05, "loss": 0.8116, "step": 54620 }, { "epoch": 0.05, "learning_rate": 4.9634100340600066e-05, "loss": 1.401, "step": 54621 }, { "epoch": 0.05, "learning_rate": 4.963408695097772e-05, "loss": 1.3632, "step": 54622 }, { "epoch": 0.05, "learning_rate": 4.963407356111219e-05, "loss": 0.9611, "step": 54623 }, { "epoch": 0.05, "learning_rate": 4.9634060171003494e-05, "loss": 1.0421, "step": 54624 }, { "epoch": 0.05, "learning_rate": 4.9634046780651624e-05, "loss": 0.9694, "step": 54625 }, { "epoch": 0.05, "learning_rate": 4.963403339005657e-05, "loss": 1.0203, "step": 54626 }, { "epoch": 0.05, "learning_rate": 4.963401999921834e-05, "loss": 0.9207, "step": 54627 }, { "epoch": 0.05, "learning_rate": 4.963400660813693e-05, "loss": 1.1448, "step": 54628 }, { "epoch": 0.05, "learning_rate": 4.963399321681235e-05, "loss": 1.674, "step": 54629 }, { "epoch": 0.05, "learning_rate": 4.963397982524459e-05, "loss": 1.9559, "step": 54630 }, { "epoch": 0.05, "learning_rate": 4.963396643343366e-05, "loss": 1.7042, "step": 54631 }, { "epoch": 0.05, "learning_rate": 4.963395304137954e-05, "loss": 1.5831, "step": 54632 }, { "epoch": 0.05, "learning_rate": 4.9633939649082255e-05, "loss": 1.7178, "step": 54633 }, { "epoch": 0.05, "learning_rate": 4.9633926256541794e-05, "loss": 1.4194, "step": 54634 }, { "epoch": 0.05, "learning_rate": 4.963391286375816e-05, "loss": 1.5408, "step": 54635 }, { "epoch": 0.05, "learning_rate": 4.963389947073134e-05, "loss": 1.5914, "step": 54636 }, { "epoch": 0.05, "learning_rate": 4.963388607746135e-05, "loss": 1.1989, "step": 54637 }, { "epoch": 0.05, "learning_rate": 4.963387268394819e-05, "loss": 0.8599, "step": 54638 }, { "epoch": 0.05, "learning_rate": 4.963385929019184e-05, "loss": 0.6637, "step": 54639 }, { "epoch": 0.05, "learning_rate": 4.963384589619233e-05, "loss": 0.4666, "step": 54640 }, { "epoch": 0.05, "learning_rate": 4.963383250194964e-05, "loss": 0.4244, "step": 54641 }, { "epoch": 0.05, "learning_rate": 4.963381910746377e-05, "loss": 0.4022, "step": 54642 }, { "epoch": 0.05, "learning_rate": 4.9633805712734734e-05, "loss": 0.6401, "step": 54643 }, { "epoch": 0.05, "learning_rate": 4.963379231776252e-05, "loss": 1.0704, "step": 54644 }, { "epoch": 0.05, "learning_rate": 4.9633778922547126e-05, "loss": 0.8909, "step": 54645 }, { "epoch": 0.05, "learning_rate": 4.963376552708856e-05, "loss": 1.4616, "step": 54646 }, { "epoch": 0.05, "learning_rate": 4.9633752131386826e-05, "loss": 1.1644, "step": 54647 }, { "epoch": 0.05, "learning_rate": 4.9633738735441916e-05, "loss": 1.0705, "step": 54648 }, { "epoch": 0.05, "learning_rate": 4.9633725339253826e-05, "loss": 1.7951, "step": 54649 }, { "epoch": 0.05, "learning_rate": 4.963371194282257e-05, "loss": 1.4216, "step": 54650 }, { "epoch": 0.05, "learning_rate": 4.9633698546148135e-05, "loss": 1.48, "step": 54651 }, { "epoch": 0.05, "learning_rate": 4.963368514923053e-05, "loss": 1.2305, "step": 54652 }, { "epoch": 0.05, "learning_rate": 4.963367175206975e-05, "loss": 1.4349, "step": 54653 }, { "epoch": 0.05, "learning_rate": 4.96336583546658e-05, "loss": 1.0538, "step": 54654 }, { "epoch": 0.05, "learning_rate": 4.963364495701867e-05, "loss": 1.5602, "step": 54655 }, { "epoch": 0.05, "learning_rate": 4.963363155912837e-05, "loss": 2.1056, "step": 54656 }, { "epoch": 0.05, "learning_rate": 4.9633618160994896e-05, "loss": 1.8249, "step": 54657 }, { "epoch": 0.05, "learning_rate": 4.9633604762618257e-05, "loss": 1.5171, "step": 54658 }, { "epoch": 0.05, "learning_rate": 4.963359136399844e-05, "loss": 1.5626, "step": 54659 }, { "epoch": 0.05, "learning_rate": 4.9633577965135445e-05, "loss": 1.4202, "step": 54660 }, { "epoch": 0.05, "learning_rate": 4.9633564566029286e-05, "loss": 1.3538, "step": 54661 }, { "epoch": 0.05, "learning_rate": 4.9633551166679954e-05, "loss": 1.2331, "step": 54662 }, { "epoch": 0.05, "learning_rate": 4.963353776708744e-05, "loss": 1.2264, "step": 54663 }, { "epoch": 0.05, "learning_rate": 4.963352436725177e-05, "loss": 1.0084, "step": 54664 }, { "epoch": 0.05, "learning_rate": 4.963351096717292e-05, "loss": 1.1912, "step": 54665 }, { "epoch": 0.05, "learning_rate": 4.96334975668509e-05, "loss": 0.9504, "step": 54666 }, { "epoch": 0.05, "learning_rate": 4.96334841662857e-05, "loss": 0.9596, "step": 54667 }, { "epoch": 0.05, "learning_rate": 4.963347076547734e-05, "loss": 1.081, "step": 54668 }, { "epoch": 0.05, "learning_rate": 4.963345736442581e-05, "loss": 1.0925, "step": 54669 }, { "epoch": 0.05, "learning_rate": 4.96334439631311e-05, "loss": 0.9956, "step": 54670 }, { "epoch": 0.05, "learning_rate": 4.963343056159322e-05, "loss": 1.1449, "step": 54671 }, { "epoch": 0.05, "learning_rate": 4.963341715981217e-05, "loss": 1.2081, "step": 54672 }, { "epoch": 0.05, "learning_rate": 4.9633403757787956e-05, "loss": 1.1375, "step": 54673 }, { "epoch": 0.05, "learning_rate": 4.963339035552056e-05, "loss": 1.2655, "step": 54674 }, { "epoch": 0.05, "learning_rate": 4.963337695301e-05, "loss": 0.9937, "step": 54675 }, { "epoch": 0.05, "learning_rate": 4.9633363550256276e-05, "loss": 0.3164, "step": 54676 }, { "epoch": 0.05, "learning_rate": 4.963335014725937e-05, "loss": 1.1083, "step": 54677 }, { "epoch": 0.05, "learning_rate": 4.96333367440193e-05, "loss": 0.8914, "step": 54678 }, { "epoch": 0.05, "learning_rate": 4.963332334053606e-05, "loss": 1.0609, "step": 54679 }, { "epoch": 0.05, "learning_rate": 4.963330993680965e-05, "loss": 1.0909, "step": 54680 }, { "epoch": 0.05, "learning_rate": 4.963329653284007e-05, "loss": 1.8053, "step": 54681 }, { "epoch": 0.05, "learning_rate": 4.963328312862732e-05, "loss": 1.1366, "step": 54682 }, { "epoch": 0.05, "learning_rate": 4.96332697241714e-05, "loss": 1.0294, "step": 54683 }, { "epoch": 0.05, "learning_rate": 4.963325631947231e-05, "loss": 1.2406, "step": 54684 }, { "epoch": 0.05, "learning_rate": 4.9633242914530054e-05, "loss": 1.2128, "step": 54685 }, { "epoch": 0.05, "learning_rate": 4.9633229509344625e-05, "loss": 1.3558, "step": 54686 }, { "epoch": 0.05, "learning_rate": 4.963321610391603e-05, "loss": 1.4147, "step": 54687 }, { "epoch": 0.05, "learning_rate": 4.963320269824426e-05, "loss": 1.4923, "step": 54688 }, { "epoch": 0.05, "learning_rate": 4.963318929232933e-05, "loss": 0.955, "step": 54689 }, { "epoch": 0.05, "learning_rate": 4.963317588617123e-05, "loss": 1.0118, "step": 54690 }, { "epoch": 0.05, "learning_rate": 4.9633162479769965e-05, "loss": 1.0768, "step": 54691 }, { "epoch": 0.05, "learning_rate": 4.963314907312553e-05, "loss": 1.067, "step": 54692 }, { "epoch": 0.05, "learning_rate": 4.963313566623792e-05, "loss": 1.0594, "step": 54693 }, { "epoch": 0.05, "learning_rate": 4.9633122259107146e-05, "loss": 1.4382, "step": 54694 }, { "epoch": 0.05, "learning_rate": 4.963310885173321e-05, "loss": 1.1543, "step": 54695 }, { "epoch": 0.05, "learning_rate": 4.9633095444116095e-05, "loss": 1.0308, "step": 54696 }, { "epoch": 0.05, "learning_rate": 4.963308203625582e-05, "loss": 1.3465, "step": 54697 }, { "epoch": 0.05, "learning_rate": 4.963306862815238e-05, "loss": 1.2881, "step": 54698 }, { "epoch": 0.05, "learning_rate": 4.963305521980576e-05, "loss": 1.001, "step": 54699 }, { "epoch": 0.05, "learning_rate": 4.9633041811215986e-05, "loss": 1.1635, "step": 54700 }, { "epoch": 0.05, "learning_rate": 4.963302840238304e-05, "loss": 1.2915, "step": 54701 }, { "epoch": 0.05, "learning_rate": 4.963301499330693e-05, "loss": 1.1649, "step": 54702 }, { "epoch": 0.05, "learning_rate": 4.963300158398765e-05, "loss": 1.3661, "step": 54703 }, { "epoch": 0.05, "learning_rate": 4.9632988174425196e-05, "loss": 1.1115, "step": 54704 }, { "epoch": 0.05, "learning_rate": 4.963297476461959e-05, "loss": 1.0874, "step": 54705 }, { "epoch": 0.05, "learning_rate": 4.9632961354570815e-05, "loss": 0.7364, "step": 54706 }, { "epoch": 0.05, "learning_rate": 4.963294794427887e-05, "loss": 0.8686, "step": 54707 }, { "epoch": 0.05, "learning_rate": 4.963293453374376e-05, "loss": 1.052, "step": 54708 }, { "epoch": 0.05, "learning_rate": 4.963292112296548e-05, "loss": 1.0036, "step": 54709 }, { "epoch": 0.05, "learning_rate": 4.9632907711944036e-05, "loss": 1.2322, "step": 54710 }, { "epoch": 0.05, "learning_rate": 4.963289430067943e-05, "loss": 1.2705, "step": 54711 }, { "epoch": 0.05, "learning_rate": 4.963288088917165e-05, "loss": 0.8985, "step": 54712 }, { "epoch": 0.05, "learning_rate": 4.963286747742072e-05, "loss": 1.2007, "step": 54713 }, { "epoch": 0.05, "learning_rate": 4.963285406542661e-05, "loss": 0.8581, "step": 54714 }, { "epoch": 0.05, "learning_rate": 4.963284065318934e-05, "loss": 1.0904, "step": 54715 }, { "epoch": 0.05, "learning_rate": 4.9632827240708906e-05, "loss": 1.3849, "step": 54716 }, { "epoch": 0.05, "learning_rate": 4.963281382798531e-05, "loss": 0.5381, "step": 54717 }, { "epoch": 0.05, "learning_rate": 4.963280041501854e-05, "loss": 1.2024, "step": 54718 }, { "epoch": 0.05, "learning_rate": 4.963278700180861e-05, "loss": 1.1006, "step": 54719 }, { "epoch": 0.05, "learning_rate": 4.963277358835552e-05, "loss": 1.2324, "step": 54720 }, { "epoch": 0.05, "learning_rate": 4.9632760174659265e-05, "loss": 1.0397, "step": 54721 }, { "epoch": 0.05, "learning_rate": 4.963274676071984e-05, "loss": 0.924, "step": 54722 }, { "epoch": 0.05, "learning_rate": 4.963273334653725e-05, "loss": 1.1859, "step": 54723 }, { "epoch": 0.05, "learning_rate": 4.963271993211151e-05, "loss": 0.8389, "step": 54724 }, { "epoch": 0.05, "learning_rate": 4.963270651744259e-05, "loss": 1.2208, "step": 54725 }, { "epoch": 0.05, "learning_rate": 4.963269310253052e-05, "loss": 1.1036, "step": 54726 }, { "epoch": 0.05, "learning_rate": 4.963267968737527e-05, "loss": 1.2852, "step": 54727 }, { "epoch": 0.05, "learning_rate": 4.963266627197687e-05, "loss": 1.1046, "step": 54728 }, { "epoch": 0.05, "learning_rate": 4.96326528563353e-05, "loss": 1.1848, "step": 54729 }, { "epoch": 0.05, "learning_rate": 4.963263944045057e-05, "loss": 1.2275, "step": 54730 }, { "epoch": 0.05, "learning_rate": 4.963262602432267e-05, "loss": 1.4178, "step": 54731 }, { "epoch": 0.05, "learning_rate": 4.963261260795161e-05, "loss": 1.3817, "step": 54732 }, { "epoch": 0.05, "learning_rate": 4.963259919133739e-05, "loss": 1.042, "step": 54733 }, { "epoch": 0.05, "learning_rate": 4.963258577448001e-05, "loss": 1.0896, "step": 54734 }, { "epoch": 0.05, "learning_rate": 4.963257235737947e-05, "loss": 1.4306, "step": 54735 }, { "epoch": 0.05, "learning_rate": 4.9632558940035764e-05, "loss": 1.208, "step": 54736 }, { "epoch": 0.05, "learning_rate": 4.963254552244889e-05, "loss": 1.2509, "step": 54737 }, { "epoch": 0.05, "learning_rate": 4.963253210461886e-05, "loss": 1.3215, "step": 54738 }, { "epoch": 0.05, "learning_rate": 4.963251868654567e-05, "loss": 1.0326, "step": 54739 }, { "epoch": 0.05, "learning_rate": 4.963250526822931e-05, "loss": 1.7616, "step": 54740 }, { "epoch": 0.05, "learning_rate": 4.963249184966979e-05, "loss": 1.9323, "step": 54741 }, { "epoch": 0.05, "learning_rate": 4.9632478430867116e-05, "loss": 1.4842, "step": 54742 }, { "epoch": 0.05, "learning_rate": 4.9632465011821274e-05, "loss": 1.356, "step": 54743 }, { "epoch": 0.05, "learning_rate": 4.963245159253227e-05, "loss": 1.0098, "step": 54744 }, { "epoch": 0.05, "learning_rate": 4.963243817300011e-05, "loss": 0.8314, "step": 54745 }, { "epoch": 0.05, "learning_rate": 4.963242475322478e-05, "loss": 1.1947, "step": 54746 }, { "epoch": 0.05, "learning_rate": 4.9632411333206295e-05, "loss": 1.3558, "step": 54747 }, { "epoch": 0.05, "learning_rate": 4.963239791294465e-05, "loss": 0.6943, "step": 54748 }, { "epoch": 0.05, "learning_rate": 4.963238449243985e-05, "loss": 1.0589, "step": 54749 }, { "epoch": 0.05, "learning_rate": 4.963237107169188e-05, "loss": 1.4898, "step": 54750 }, { "epoch": 0.05, "learning_rate": 4.963235765070076e-05, "loss": 1.1185, "step": 54751 }, { "epoch": 0.05, "learning_rate": 4.9632344229466466e-05, "loss": 1.1065, "step": 54752 }, { "epoch": 0.05, "learning_rate": 4.963233080798902e-05, "loss": 0.9587, "step": 54753 }, { "epoch": 0.05, "learning_rate": 4.963231738626841e-05, "loss": 0.8799, "step": 54754 }, { "epoch": 0.05, "learning_rate": 4.963230396430464e-05, "loss": 1.313, "step": 54755 }, { "epoch": 0.05, "learning_rate": 4.9632290542097713e-05, "loss": 0.8768, "step": 54756 }, { "epoch": 0.05, "learning_rate": 4.963227711964763e-05, "loss": 1.3048, "step": 54757 }, { "epoch": 0.05, "learning_rate": 4.9632263696954384e-05, "loss": 1.3289, "step": 54758 }, { "epoch": 0.05, "learning_rate": 4.9632250274017976e-05, "loss": 0.9517, "step": 54759 }, { "epoch": 0.05, "learning_rate": 4.9632236850838416e-05, "loss": 0.9378, "step": 54760 }, { "epoch": 0.05, "learning_rate": 4.963222342741569e-05, "loss": 1.077, "step": 54761 }, { "epoch": 0.05, "learning_rate": 4.9632210003749804e-05, "loss": 1.3689, "step": 54762 }, { "epoch": 0.05, "learning_rate": 4.9632196579840766e-05, "loss": 1.0918, "step": 54763 }, { "epoch": 0.05, "learning_rate": 4.963218315568857e-05, "loss": 0.3889, "step": 54764 }, { "epoch": 0.05, "learning_rate": 4.963216973129321e-05, "loss": 1.3142, "step": 54765 }, { "epoch": 0.05, "learning_rate": 4.963215630665469e-05, "loss": 1.4621, "step": 54766 }, { "epoch": 0.05, "learning_rate": 4.963214288177301e-05, "loss": 1.309, "step": 54767 }, { "epoch": 0.05, "learning_rate": 4.963212945664818e-05, "loss": 0.7406, "step": 54768 }, { "epoch": 0.05, "learning_rate": 4.963211603128019e-05, "loss": 0.9111, "step": 54769 }, { "epoch": 0.05, "learning_rate": 4.963210260566904e-05, "loss": 1.1893, "step": 54770 }, { "epoch": 0.05, "learning_rate": 4.963208917981473e-05, "loss": 1.1365, "step": 54771 }, { "epoch": 0.05, "learning_rate": 4.9632075753717266e-05, "loss": 1.1213, "step": 54772 }, { "epoch": 0.05, "learning_rate": 4.9632062327376647e-05, "loss": 1.3827, "step": 54773 }, { "epoch": 0.05, "learning_rate": 4.9632048900792875e-05, "loss": 1.4273, "step": 54774 }, { "epoch": 0.05, "learning_rate": 4.963203547396593e-05, "loss": 1.1274, "step": 54775 }, { "epoch": 0.05, "learning_rate": 4.963202204689584e-05, "loss": 1.4454, "step": 54776 }, { "epoch": 0.05, "learning_rate": 4.963200861958259e-05, "loss": 1.3081, "step": 54777 }, { "epoch": 0.05, "learning_rate": 4.963199519202618e-05, "loss": 0.8804, "step": 54778 }, { "epoch": 0.05, "learning_rate": 4.9631981764226617e-05, "loss": 1.2561, "step": 54779 }, { "epoch": 0.05, "learning_rate": 4.96319683361839e-05, "loss": 1.2138, "step": 54780 }, { "epoch": 0.05, "learning_rate": 4.9631954907898013e-05, "loss": 1.1405, "step": 54781 }, { "epoch": 0.05, "learning_rate": 4.963194147936898e-05, "loss": 1.3121, "step": 54782 }, { "epoch": 0.05, "learning_rate": 4.963192805059679e-05, "loss": 0.8503, "step": 54783 }, { "epoch": 0.05, "learning_rate": 4.963191462158145e-05, "loss": 0.8233, "step": 54784 }, { "epoch": 0.05, "learning_rate": 4.963190119232295e-05, "loss": 1.2509, "step": 54785 }, { "epoch": 0.05, "learning_rate": 4.96318877628213e-05, "loss": 1.255, "step": 54786 }, { "epoch": 0.05, "learning_rate": 4.963187433307648e-05, "loss": 1.2836, "step": 54787 }, { "epoch": 0.05, "learning_rate": 4.963186090308851e-05, "loss": 0.9213, "step": 54788 }, { "epoch": 0.05, "learning_rate": 4.9631847472857386e-05, "loss": 1.1468, "step": 54789 }, { "epoch": 0.05, "learning_rate": 4.963183404238311e-05, "loss": 1.2082, "step": 54790 }, { "epoch": 0.05, "learning_rate": 4.963182061166567e-05, "loss": 1.1225, "step": 54791 }, { "epoch": 0.05, "learning_rate": 4.963180718070508e-05, "loss": 1.637, "step": 54792 }, { "epoch": 0.05, "learning_rate": 4.963179374950134e-05, "loss": 1.2573, "step": 54793 }, { "epoch": 0.05, "learning_rate": 4.963178031805444e-05, "loss": 1.2878, "step": 54794 }, { "epoch": 0.05, "learning_rate": 4.963176688636439e-05, "loss": 1.2359, "step": 54795 }, { "epoch": 0.05, "learning_rate": 4.9631753454431183e-05, "loss": 1.5704, "step": 54796 }, { "epoch": 0.05, "learning_rate": 4.963174002225482e-05, "loss": 1.1546, "step": 54797 }, { "epoch": 0.05, "learning_rate": 4.96317265898353e-05, "loss": 1.1822, "step": 54798 }, { "epoch": 0.05, "learning_rate": 4.963171315717263e-05, "loss": 1.0891, "step": 54799 }, { "epoch": 0.05, "learning_rate": 4.963169972426681e-05, "loss": 1.3866, "step": 54800 }, { "epoch": 0.05, "learning_rate": 4.963168629111783e-05, "loss": 1.0083, "step": 54801 }, { "epoch": 0.05, "learning_rate": 4.96316728577257e-05, "loss": 1.0575, "step": 54802 }, { "epoch": 0.05, "learning_rate": 4.963165942409041e-05, "loss": 0.8108, "step": 54803 }, { "epoch": 0.05, "learning_rate": 4.963164599021198e-05, "loss": 1.3099, "step": 54804 }, { "epoch": 0.05, "learning_rate": 4.963163255609039e-05, "loss": 1.0778, "step": 54805 }, { "epoch": 0.05, "learning_rate": 4.963161912172564e-05, "loss": 1.4872, "step": 54806 }, { "epoch": 0.05, "learning_rate": 4.963160568711774e-05, "loss": 1.349, "step": 54807 }, { "epoch": 0.05, "learning_rate": 4.963159225226669e-05, "loss": 1.1602, "step": 54808 }, { "epoch": 0.05, "learning_rate": 4.963157881717249e-05, "loss": 1.0752, "step": 54809 }, { "epoch": 0.05, "learning_rate": 4.963156538183514e-05, "loss": 1.3357, "step": 54810 }, { "epoch": 0.05, "learning_rate": 4.963155194625463e-05, "loss": 0.8512, "step": 54811 }, { "epoch": 0.05, "learning_rate": 4.963153851043097e-05, "loss": 1.1783, "step": 54812 }, { "epoch": 0.05, "learning_rate": 4.963152507436416e-05, "loss": 0.8616, "step": 54813 }, { "epoch": 0.05, "learning_rate": 4.9631511638054196e-05, "loss": 1.1189, "step": 54814 }, { "epoch": 0.05, "learning_rate": 4.9631498201501076e-05, "loss": 1.2787, "step": 54815 }, { "epoch": 0.05, "learning_rate": 4.9631484764704804e-05, "loss": 0.9061, "step": 54816 }, { "epoch": 0.05, "learning_rate": 4.963147132766539e-05, "loss": 1.076, "step": 54817 }, { "epoch": 0.05, "learning_rate": 4.9631457890382816e-05, "loss": 1.1168, "step": 54818 }, { "epoch": 0.05, "learning_rate": 4.963144445285709e-05, "loss": 1.2135, "step": 54819 }, { "epoch": 0.05, "learning_rate": 4.963143101508822e-05, "loss": 0.9782, "step": 54820 }, { "epoch": 0.05, "learning_rate": 4.9631417577076195e-05, "loss": 1.2777, "step": 54821 }, { "epoch": 0.05, "learning_rate": 4.963140413882102e-05, "loss": 0.9872, "step": 54822 }, { "epoch": 0.05, "learning_rate": 4.9631390700322694e-05, "loss": 1.0666, "step": 54823 }, { "epoch": 0.05, "learning_rate": 4.9631377261581214e-05, "loss": 0.7767, "step": 54824 }, { "epoch": 0.05, "learning_rate": 4.963136382259659e-05, "loss": 1.4897, "step": 54825 }, { "epoch": 0.05, "learning_rate": 4.963135038336881e-05, "loss": 1.6811, "step": 54826 }, { "epoch": 0.05, "learning_rate": 4.963133694389788e-05, "loss": 1.1442, "step": 54827 }, { "epoch": 0.05, "learning_rate": 4.96313235041838e-05, "loss": 0.9892, "step": 54828 }, { "epoch": 0.05, "learning_rate": 4.963131006422657e-05, "loss": 0.9884, "step": 54829 }, { "epoch": 0.05, "learning_rate": 4.963129662402619e-05, "loss": 1.3282, "step": 54830 }, { "epoch": 0.05, "learning_rate": 4.963128318358267e-05, "loss": 1.0668, "step": 54831 }, { "epoch": 0.05, "learning_rate": 4.963126974289599e-05, "loss": 1.2896, "step": 54832 }, { "epoch": 0.05, "learning_rate": 4.9631256301966155e-05, "loss": 1.1628, "step": 54833 }, { "epoch": 0.05, "learning_rate": 4.963124286079318e-05, "loss": 1.8185, "step": 54834 }, { "epoch": 0.05, "learning_rate": 4.963122941937706e-05, "loss": 1.7734, "step": 54835 }, { "epoch": 0.05, "learning_rate": 4.963121597771778e-05, "loss": 1.7669, "step": 54836 }, { "epoch": 0.05, "learning_rate": 4.9631202535815355e-05, "loss": 1.7684, "step": 54837 }, { "epoch": 0.05, "learning_rate": 4.9631189093669786e-05, "loss": 1.7132, "step": 54838 }, { "epoch": 0.05, "learning_rate": 4.9631175651281057e-05, "loss": 1.7216, "step": 54839 }, { "epoch": 0.05, "learning_rate": 4.963116220864918e-05, "loss": 1.6213, "step": 54840 }, { "epoch": 0.05, "learning_rate": 4.9631148765774174e-05, "loss": 1.1135, "step": 54841 }, { "epoch": 0.05, "learning_rate": 4.9631135322656e-05, "loss": 1.4909, "step": 54842 }, { "epoch": 0.05, "learning_rate": 4.963112187929468e-05, "loss": 1.3456, "step": 54843 }, { "epoch": 0.05, "learning_rate": 4.963110843569022e-05, "loss": 1.2133, "step": 54844 }, { "epoch": 0.05, "learning_rate": 4.963109499184261e-05, "loss": 1.4961, "step": 54845 }, { "epoch": 0.05, "learning_rate": 4.963108154775185e-05, "loss": 1.2919, "step": 54846 }, { "epoch": 0.05, "learning_rate": 4.9631068103417945e-05, "loss": 0.9288, "step": 54847 }, { "epoch": 0.05, "learning_rate": 4.963105465884089e-05, "loss": 1.5823, "step": 54848 }, { "epoch": 0.05, "learning_rate": 4.963104121402069e-05, "loss": 1.2581, "step": 54849 }, { "epoch": 0.05, "learning_rate": 4.963102776895734e-05, "loss": 1.3983, "step": 54850 }, { "epoch": 0.05, "learning_rate": 4.963101432365084e-05, "loss": 0.817, "step": 54851 }, { "epoch": 0.05, "learning_rate": 4.9631000878101194e-05, "loss": 0.5936, "step": 54852 }, { "epoch": 0.05, "learning_rate": 4.963098743230841e-05, "loss": 1.225, "step": 54853 }, { "epoch": 0.05, "learning_rate": 4.963097398627247e-05, "loss": 1.0401, "step": 54854 }, { "epoch": 0.05, "learning_rate": 4.963096053999339e-05, "loss": 1.5191, "step": 54855 }, { "epoch": 0.05, "learning_rate": 4.963094709347116e-05, "loss": 1.0603, "step": 54856 }, { "epoch": 0.05, "learning_rate": 4.963093364670578e-05, "loss": 0.9153, "step": 54857 }, { "epoch": 0.05, "learning_rate": 4.963092019969726e-05, "loss": 0.8918, "step": 54858 }, { "epoch": 0.05, "learning_rate": 4.963090675244559e-05, "loss": 0.9646, "step": 54859 }, { "epoch": 0.05, "learning_rate": 4.9630893304950774e-05, "loss": 0.8052, "step": 54860 }, { "epoch": 0.05, "learning_rate": 4.9630879857212815e-05, "loss": 1.0406, "step": 54861 }, { "epoch": 0.05, "learning_rate": 4.963086640923171e-05, "loss": 0.9896, "step": 54862 }, { "epoch": 0.05, "learning_rate": 4.963085296100746e-05, "loss": 1.1436, "step": 54863 }, { "epoch": 0.05, "learning_rate": 4.963083951254006e-05, "loss": 1.2814, "step": 54864 }, { "epoch": 0.05, "learning_rate": 4.963082606382951e-05, "loss": 0.9315, "step": 54865 }, { "epoch": 0.05, "learning_rate": 4.963081261487583e-05, "loss": 1.188, "step": 54866 }, { "epoch": 0.05, "learning_rate": 4.9630799165679e-05, "loss": 1.4139, "step": 54867 }, { "epoch": 0.05, "learning_rate": 4.963078571623902e-05, "loss": 1.3755, "step": 54868 }, { "epoch": 0.05, "learning_rate": 4.96307722665559e-05, "loss": 1.2433, "step": 54869 }, { "epoch": 0.05, "learning_rate": 4.963075881662963e-05, "loss": 1.224, "step": 54870 }, { "epoch": 0.05, "learning_rate": 4.963074536646022e-05, "loss": 1.3511, "step": 54871 }, { "epoch": 0.05, "learning_rate": 4.963073191604767e-05, "loss": 0.6109, "step": 54872 }, { "epoch": 0.05, "learning_rate": 4.963071846539197e-05, "loss": 1.0037, "step": 54873 }, { "epoch": 0.05, "learning_rate": 4.9630705014493126e-05, "loss": 0.9629, "step": 54874 }, { "epoch": 0.05, "learning_rate": 4.963069156335113e-05, "loss": 0.9512, "step": 54875 }, { "epoch": 0.05, "learning_rate": 4.9630678111966e-05, "loss": 1.4814, "step": 54876 }, { "epoch": 0.05, "learning_rate": 4.9630664660337725e-05, "loss": 1.0632, "step": 54877 }, { "epoch": 0.05, "learning_rate": 4.9630651208466306e-05, "loss": 1.2553, "step": 54878 }, { "epoch": 0.05, "learning_rate": 4.963063775635175e-05, "loss": 1.1759, "step": 54879 }, { "epoch": 0.05, "learning_rate": 4.9630624303994044e-05, "loss": 1.2557, "step": 54880 }, { "epoch": 0.05, "learning_rate": 4.963061085139319e-05, "loss": 0.8034, "step": 54881 }, { "epoch": 0.05, "learning_rate": 4.96305973985492e-05, "loss": 1.0278, "step": 54882 }, { "epoch": 0.05, "learning_rate": 4.9630583945462064e-05, "loss": 1.1309, "step": 54883 }, { "epoch": 0.05, "learning_rate": 4.963057049213179e-05, "loss": 1.0546, "step": 54884 }, { "epoch": 0.05, "learning_rate": 4.9630557038558364e-05, "loss": 1.2601, "step": 54885 }, { "epoch": 0.05, "learning_rate": 4.9630543584741805e-05, "loss": 1.069, "step": 54886 }, { "epoch": 0.05, "learning_rate": 4.963053013068211e-05, "loss": 1.4739, "step": 54887 }, { "epoch": 0.05, "learning_rate": 4.963051667637926e-05, "loss": 0.9931, "step": 54888 }, { "epoch": 0.05, "learning_rate": 4.963050322183327e-05, "loss": 1.1899, "step": 54889 }, { "epoch": 0.05, "learning_rate": 4.963048976704414e-05, "loss": 1.322, "step": 54890 }, { "epoch": 0.05, "learning_rate": 4.9630476312011863e-05, "loss": 1.1432, "step": 54891 }, { "epoch": 0.05, "learning_rate": 4.963046285673645e-05, "loss": 1.1034, "step": 54892 }, { "epoch": 0.05, "learning_rate": 4.96304494012179e-05, "loss": 0.4197, "step": 54893 }, { "epoch": 0.05, "learning_rate": 4.96304359454562e-05, "loss": 0.3784, "step": 54894 }, { "epoch": 0.05, "learning_rate": 4.963042248945135e-05, "loss": 0.9776, "step": 54895 }, { "epoch": 0.05, "learning_rate": 4.9630409033203376e-05, "loss": 1.1977, "step": 54896 }, { "epoch": 0.05, "learning_rate": 4.963039557671226e-05, "loss": 0.9252, "step": 54897 }, { "epoch": 0.05, "learning_rate": 4.963038211997799e-05, "loss": 0.8458, "step": 54898 }, { "epoch": 0.05, "learning_rate": 4.963036866300059e-05, "loss": 0.7553, "step": 54899 }, { "epoch": 0.05, "learning_rate": 4.963035520578005e-05, "loss": 0.7999, "step": 54900 }, { "epoch": 0.05, "learning_rate": 4.9630341748316363e-05, "loss": 0.9601, "step": 54901 }, { "epoch": 0.05, "learning_rate": 4.963032829060954e-05, "loss": 1.0361, "step": 54902 }, { "epoch": 0.05, "learning_rate": 4.9630314832659574e-05, "loss": 1.5558, "step": 54903 }, { "epoch": 0.05, "learning_rate": 4.963030137446647e-05, "loss": 1.1147, "step": 54904 }, { "epoch": 0.05, "learning_rate": 4.963028791603023e-05, "loss": 1.2208, "step": 54905 }, { "epoch": 0.05, "learning_rate": 4.9630274457350846e-05, "loss": 0.9455, "step": 54906 }, { "epoch": 0.05, "learning_rate": 4.9630260998428325e-05, "loss": 0.8504, "step": 54907 }, { "epoch": 0.05, "learning_rate": 4.963024753926266e-05, "loss": 1.1936, "step": 54908 }, { "epoch": 0.05, "learning_rate": 4.963023407985385e-05, "loss": 0.5659, "step": 54909 }, { "epoch": 0.05, "learning_rate": 4.9630220620201914e-05, "loss": 0.8343, "step": 54910 }, { "epoch": 0.05, "learning_rate": 4.963020716030683e-05, "loss": 0.9691, "step": 54911 }, { "epoch": 0.05, "learning_rate": 4.9630193700168614e-05, "loss": 0.8832, "step": 54912 }, { "epoch": 0.05, "learning_rate": 4.963018023978726e-05, "loss": 1.0355, "step": 54913 }, { "epoch": 0.05, "learning_rate": 4.963016677916276e-05, "loss": 1.2375, "step": 54914 }, { "epoch": 0.05, "learning_rate": 4.9630153318295116e-05, "loss": 0.8989, "step": 54915 }, { "epoch": 0.05, "learning_rate": 4.9630139857184343e-05, "loss": 1.1177, "step": 54916 }, { "epoch": 0.05, "learning_rate": 4.963012639583043e-05, "loss": 1.5479, "step": 54917 }, { "epoch": 0.05, "learning_rate": 4.963011293423338e-05, "loss": 1.2447, "step": 54918 }, { "epoch": 0.05, "learning_rate": 4.963009947239319e-05, "loss": 1.0739, "step": 54919 }, { "epoch": 0.05, "learning_rate": 4.963008601030986e-05, "loss": 1.1828, "step": 54920 }, { "epoch": 0.05, "learning_rate": 4.96300725479834e-05, "loss": 1.2342, "step": 54921 }, { "epoch": 0.05, "learning_rate": 4.963005908541379e-05, "loss": 1.1931, "step": 54922 }, { "epoch": 0.05, "learning_rate": 4.963004562260105e-05, "loss": 0.9644, "step": 54923 }, { "epoch": 0.05, "learning_rate": 4.963003215954517e-05, "loss": 1.1622, "step": 54924 }, { "epoch": 0.05, "learning_rate": 4.963001869624616e-05, "loss": 0.5294, "step": 54925 }, { "epoch": 0.05, "learning_rate": 4.963000523270401e-05, "loss": 0.1391, "step": 54926 }, { "epoch": 0.05, "learning_rate": 4.9629991768918715e-05, "loss": 0.0889, "step": 54927 }, { "epoch": 0.05, "learning_rate": 4.962997830489029e-05, "loss": 0.0857, "step": 54928 }, { "epoch": 0.05, "learning_rate": 4.9629964840618725e-05, "loss": 0.0991, "step": 54929 }, { "epoch": 0.05, "learning_rate": 4.962995137610403e-05, "loss": 0.096, "step": 54930 }, { "epoch": 0.05, "learning_rate": 4.962993791134619e-05, "loss": 0.0838, "step": 54931 }, { "epoch": 0.05, "learning_rate": 4.962992444634522e-05, "loss": 0.0689, "step": 54932 }, { "epoch": 0.05, "learning_rate": 4.962991098110111e-05, "loss": 0.0731, "step": 54933 }, { "epoch": 0.05, "learning_rate": 4.962989751561387e-05, "loss": 0.06, "step": 54934 }, { "epoch": 0.05, "learning_rate": 4.9629884049883485e-05, "loss": 0.0533, "step": 54935 }, { "epoch": 0.05, "learning_rate": 4.9629870583909964e-05, "loss": 0.1349, "step": 54936 }, { "epoch": 0.05, "learning_rate": 4.962985711769331e-05, "loss": 0.2452, "step": 54937 }, { "epoch": 0.05, "learning_rate": 4.962984365123353e-05, "loss": 0.1617, "step": 54938 }, { "epoch": 0.05, "learning_rate": 4.96298301845306e-05, "loss": 0.1143, "step": 54939 }, { "epoch": 0.05, "learning_rate": 4.962981671758454e-05, "loss": 0.0531, "step": 54940 }, { "epoch": 0.05, "learning_rate": 4.962980325039535e-05, "loss": 0.0353, "step": 54941 }, { "epoch": 0.05, "learning_rate": 4.962978978296302e-05, "loss": 0.0641, "step": 54942 }, { "epoch": 0.05, "learning_rate": 4.9629776315287555e-05, "loss": 0.0416, "step": 54943 }, { "epoch": 0.05, "learning_rate": 4.962976284736896e-05, "loss": 0.0479, "step": 54944 }, { "epoch": 0.05, "learning_rate": 4.962974937920722e-05, "loss": 0.9854, "step": 54945 }, { "epoch": 0.05, "learning_rate": 4.962973591080236e-05, "loss": 1.0945, "step": 54946 }, { "epoch": 0.05, "learning_rate": 4.962972244215436e-05, "loss": 1.0258, "step": 54947 }, { "epoch": 0.05, "learning_rate": 4.9629708973263224e-05, "loss": 1.1578, "step": 54948 }, { "epoch": 0.05, "learning_rate": 4.962969550412895e-05, "loss": 1.5301, "step": 54949 }, { "epoch": 0.05, "learning_rate": 4.962968203475155e-05, "loss": 1.0276, "step": 54950 }, { "epoch": 0.05, "learning_rate": 4.962966856513101e-05, "loss": 1.6222, "step": 54951 }, { "epoch": 0.05, "learning_rate": 4.9629655095267334e-05, "loss": 1.0555, "step": 54952 }, { "epoch": 0.05, "learning_rate": 4.962964162516053e-05, "loss": 1.1516, "step": 54953 }, { "epoch": 0.05, "learning_rate": 4.962962815481059e-05, "loss": 1.2658, "step": 54954 }, { "epoch": 0.05, "learning_rate": 4.9629614684217516e-05, "loss": 1.3094, "step": 54955 }, { "epoch": 0.05, "learning_rate": 4.962960121338131e-05, "loss": 1.2123, "step": 54956 }, { "epoch": 0.05, "learning_rate": 4.962958774230197e-05, "loss": 1.0154, "step": 54957 }, { "epoch": 0.05, "learning_rate": 4.96295742709795e-05, "loss": 1.2227, "step": 54958 }, { "epoch": 0.05, "learning_rate": 4.96295607994139e-05, "loss": 1.0158, "step": 54959 }, { "epoch": 0.05, "learning_rate": 4.9629547327605156e-05, "loss": 1.6547, "step": 54960 }, { "epoch": 0.05, "learning_rate": 4.962953385555329e-05, "loss": 0.9459, "step": 54961 }, { "epoch": 0.05, "learning_rate": 4.962952038325829e-05, "loss": 1.2667, "step": 54962 }, { "epoch": 0.05, "learning_rate": 4.9629506910720155e-05, "loss": 1.1804, "step": 54963 }, { "epoch": 0.05, "learning_rate": 4.962949343793889e-05, "loss": 1.391, "step": 54964 }, { "epoch": 0.05, "learning_rate": 4.962947996491449e-05, "loss": 1.4556, "step": 54965 }, { "epoch": 0.05, "learning_rate": 4.9629466491646956e-05, "loss": 1.198, "step": 54966 }, { "epoch": 0.05, "learning_rate": 4.96294530181363e-05, "loss": 1.1983, "step": 54967 }, { "epoch": 0.05, "learning_rate": 4.962943954438251e-05, "loss": 1.2942, "step": 54968 }, { "epoch": 0.05, "learning_rate": 4.962942607038558e-05, "loss": 1.1763, "step": 54969 }, { "epoch": 0.05, "learning_rate": 4.962941259614552e-05, "loss": 1.1828, "step": 54970 }, { "epoch": 0.05, "learning_rate": 4.962939912166234e-05, "loss": 1.2043, "step": 54971 }, { "epoch": 0.05, "learning_rate": 4.962938564693602e-05, "loss": 0.8797, "step": 54972 }, { "epoch": 0.05, "learning_rate": 4.9629372171966567e-05, "loss": 0.9645, "step": 54973 }, { "epoch": 0.05, "learning_rate": 4.962935869675398e-05, "loss": 0.8332, "step": 54974 }, { "epoch": 0.05, "learning_rate": 4.962934522129827e-05, "loss": 0.5845, "step": 54975 }, { "epoch": 0.05, "learning_rate": 4.962933174559943e-05, "loss": 0.369, "step": 54976 }, { "epoch": 0.05, "learning_rate": 4.9629318269657464e-05, "loss": 0.9471, "step": 54977 }, { "epoch": 0.05, "learning_rate": 4.962930479347236e-05, "loss": 1.3995, "step": 54978 }, { "epoch": 0.05, "learning_rate": 4.9629291317044125e-05, "loss": 1.1406, "step": 54979 }, { "epoch": 0.05, "learning_rate": 4.962927784037277e-05, "loss": 1.2293, "step": 54980 }, { "epoch": 0.05, "learning_rate": 4.962926436345827e-05, "loss": 1.5125, "step": 54981 }, { "epoch": 0.05, "learning_rate": 4.962925088630065e-05, "loss": 0.5801, "step": 54982 }, { "epoch": 0.05, "learning_rate": 4.96292374088999e-05, "loss": 1.3338, "step": 54983 }, { "epoch": 0.05, "learning_rate": 4.962922393125602e-05, "loss": 1.1519, "step": 54984 }, { "epoch": 0.05, "learning_rate": 4.9629210453369015e-05, "loss": 1.032, "step": 54985 }, { "epoch": 0.05, "learning_rate": 4.962919697523887e-05, "loss": 1.1785, "step": 54986 }, { "epoch": 0.05, "learning_rate": 4.9629183496865606e-05, "loss": 1.1965, "step": 54987 }, { "epoch": 0.05, "learning_rate": 4.96291700182492e-05, "loss": 0.9652, "step": 54988 }, { "epoch": 0.05, "learning_rate": 4.9629156539389675e-05, "loss": 1.1084, "step": 54989 }, { "epoch": 0.05, "learning_rate": 4.962914306028703e-05, "loss": 0.987, "step": 54990 }, { "epoch": 0.05, "learning_rate": 4.962912958094124e-05, "loss": 1.0694, "step": 54991 }, { "epoch": 0.05, "learning_rate": 4.962911610135233e-05, "loss": 0.8405, "step": 54992 }, { "epoch": 0.05, "learning_rate": 4.962910262152029e-05, "loss": 0.8536, "step": 54993 }, { "epoch": 0.05, "learning_rate": 4.962908914144512e-05, "loss": 0.7437, "step": 54994 }, { "epoch": 0.05, "learning_rate": 4.962907566112683e-05, "loss": 1.0608, "step": 54995 }, { "epoch": 0.05, "learning_rate": 4.96290621805654e-05, "loss": 1.3794, "step": 54996 }, { "epoch": 0.05, "learning_rate": 4.9629048699760847e-05, "loss": 1.0758, "step": 54997 }, { "epoch": 0.05, "learning_rate": 4.9629035218713175e-05, "loss": 0.7106, "step": 54998 }, { "epoch": 0.05, "learning_rate": 4.9629021737422364e-05, "loss": 1.1511, "step": 54999 }, { "epoch": 0.06, "learning_rate": 4.9629008255888434e-05, "loss": 1.4025, "step": 55000 }, { "epoch": 0.06, "eval_loss": 1.07865571975708, "eval_runtime": 125.5474, "eval_samples_per_second": 11.032, "eval_steps_per_second": 5.52, "step": 55000 } ], "max_steps": 1000000, "num_train_epochs": 9223372036854775807, "total_flos": 2.00494307868672e+18, "trial_name": null, "trial_params": null }