{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.03846005922849121, "eval_steps": 500, "global_step": 500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.98e-05, "loss": 1.4542, "step": 1 }, { "epoch": 0.0, "learning_rate": 9.960000000000001e-05, "loss": 0.0, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.94e-05, "loss": 0.0, "step": 3 }, { "epoch": 0.0, "learning_rate": 9.92e-05, "loss": 0.0, "step": 4 }, { "epoch": 0.0, "learning_rate": 9.900000000000001e-05, "loss": 0.0, "step": 5 }, { "epoch": 0.0, "learning_rate": 9.88e-05, "loss": 0.0, "step": 6 }, { "epoch": 0.0, "learning_rate": 9.86e-05, "loss": 0.0, "step": 7 }, { "epoch": 0.0, "learning_rate": 9.84e-05, "loss": 0.0, "step": 8 }, { "epoch": 0.0, "learning_rate": 9.82e-05, "loss": 0.0, "step": 9 }, { "epoch": 0.0, "learning_rate": 9.8e-05, "loss": 0.0, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.78e-05, "loss": 0.0, "step": 11 }, { "epoch": 0.0, "learning_rate": 9.76e-05, "loss": 0.0, "step": 12 }, { "epoch": 0.0, "learning_rate": 9.74e-05, "loss": 0.0, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.72e-05, "loss": 0.0, "step": 14 }, { "epoch": 0.0, "learning_rate": 9.7e-05, "loss": 0.0, "step": 15 }, { "epoch": 0.0, "learning_rate": 9.680000000000001e-05, "loss": 0.0, "step": 16 }, { "epoch": 0.0, "learning_rate": 9.66e-05, "loss": 0.0, "step": 17 }, { "epoch": 0.0, "learning_rate": 9.64e-05, "loss": 0.0, "step": 18 }, { "epoch": 0.0, "learning_rate": 9.620000000000001e-05, "loss": 0.0, "step": 19 }, { "epoch": 0.0, "learning_rate": 9.6e-05, "loss": 0.0, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.58e-05, "loss": 0.0, "step": 21 }, { "epoch": 0.0, "learning_rate": 9.56e-05, "loss": 0.0, "step": 22 }, { "epoch": 0.0, "learning_rate": 9.54e-05, "loss": 0.0, "step": 23 }, { "epoch": 0.0, "learning_rate": 9.52e-05, "loss": 0.0, "step": 24 }, { "epoch": 0.0, "learning_rate": 9.5e-05, "loss": 0.0, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.48e-05, "loss": 0.0, "step": 26 }, { "epoch": 0.0, "learning_rate": 9.46e-05, "loss": 0.0, "step": 27 }, { "epoch": 0.0, "learning_rate": 9.44e-05, "loss": 0.0, "step": 28 }, { "epoch": 0.0, "learning_rate": 9.42e-05, "loss": 0.0, "step": 29 }, { "epoch": 0.0, "learning_rate": 9.4e-05, "loss": 0.0, "step": 30 }, { "epoch": 0.0, "learning_rate": 9.38e-05, "loss": 0.0, "step": 31 }, { "epoch": 0.0, "learning_rate": 9.360000000000001e-05, "loss": 0.0, "step": 32 }, { "epoch": 0.0, "learning_rate": 9.340000000000001e-05, "loss": 0.0, "step": 33 }, { "epoch": 0.0, "learning_rate": 9.320000000000002e-05, "loss": 0.0, "step": 34 }, { "epoch": 0.0, "learning_rate": 9.300000000000001e-05, "loss": 0.0, "step": 35 }, { "epoch": 0.0, "learning_rate": 9.28e-05, "loss": 0.0, "step": 36 }, { "epoch": 0.0, "learning_rate": 9.260000000000001e-05, "loss": 0.0, "step": 37 }, { "epoch": 0.0, "learning_rate": 9.240000000000001e-05, "loss": 0.0, "step": 38 }, { "epoch": 0.0, "learning_rate": 9.22e-05, "loss": 0.0, "step": 39 }, { "epoch": 0.0, "learning_rate": 9.200000000000001e-05, "loss": 0.0, "step": 40 }, { "epoch": 0.0, "learning_rate": 9.180000000000001e-05, "loss": 0.0, "step": 41 }, { "epoch": 0.0, "learning_rate": 9.16e-05, "loss": 0.0, "step": 42 }, { "epoch": 0.0, "learning_rate": 9.140000000000001e-05, "loss": 0.0, "step": 43 }, { "epoch": 0.0, "learning_rate": 9.120000000000001e-05, "loss": 0.0, "step": 44 }, { "epoch": 0.0, "learning_rate": 9.1e-05, "loss": 0.0, "step": 45 }, { "epoch": 0.0, "learning_rate": 9.080000000000001e-05, "loss": 0.0, "step": 46 }, { "epoch": 0.0, "learning_rate": 9.06e-05, "loss": 0.0, "step": 47 }, { "epoch": 0.0, "learning_rate": 9.04e-05, "loss": 0.0, "step": 48 }, { "epoch": 0.0, "learning_rate": 9.020000000000001e-05, "loss": 0.0, "step": 49 }, { "epoch": 0.0, "learning_rate": 9e-05, "loss": 0.0, "step": 50 }, { "epoch": 0.0, "learning_rate": 8.98e-05, "loss": 0.0, "step": 51 }, { "epoch": 0.0, "learning_rate": 8.960000000000001e-05, "loss": 0.0, "step": 52 }, { "epoch": 0.0, "learning_rate": 8.94e-05, "loss": 0.0, "step": 53 }, { "epoch": 0.0, "learning_rate": 8.92e-05, "loss": 0.0, "step": 54 }, { "epoch": 0.0, "learning_rate": 8.900000000000001e-05, "loss": 0.0, "step": 55 }, { "epoch": 0.0, "learning_rate": 8.88e-05, "loss": 0.0, "step": 56 }, { "epoch": 0.0, "learning_rate": 8.86e-05, "loss": 0.0, "step": 57 }, { "epoch": 0.0, "learning_rate": 8.840000000000001e-05, "loss": 0.0, "step": 58 }, { "epoch": 0.0, "learning_rate": 8.82e-05, "loss": 0.0, "step": 59 }, { "epoch": 0.0, "learning_rate": 8.800000000000001e-05, "loss": 0.0, "step": 60 }, { "epoch": 0.0, "learning_rate": 8.78e-05, "loss": 0.0, "step": 61 }, { "epoch": 0.0, "learning_rate": 8.76e-05, "loss": 0.0, "step": 62 }, { "epoch": 0.0, "learning_rate": 8.740000000000001e-05, "loss": 0.0, "step": 63 }, { "epoch": 0.0, "learning_rate": 8.72e-05, "loss": 0.0, "step": 64 }, { "epoch": 0.0, "learning_rate": 8.7e-05, "loss": 0.0, "step": 65 }, { "epoch": 0.01, "learning_rate": 8.680000000000001e-05, "loss": 0.0, "step": 66 }, { "epoch": 0.01, "learning_rate": 8.66e-05, "loss": 0.0, "step": 67 }, { "epoch": 0.01, "learning_rate": 8.64e-05, "loss": 0.0, "step": 68 }, { "epoch": 0.01, "learning_rate": 8.620000000000001e-05, "loss": 0.0, "step": 69 }, { "epoch": 0.01, "learning_rate": 8.6e-05, "loss": 0.0, "step": 70 }, { "epoch": 0.01, "learning_rate": 8.58e-05, "loss": 0.0, "step": 71 }, { "epoch": 0.01, "learning_rate": 8.560000000000001e-05, "loss": 0.0, "step": 72 }, { "epoch": 0.01, "learning_rate": 8.54e-05, "loss": 0.0, "step": 73 }, { "epoch": 0.01, "learning_rate": 8.52e-05, "loss": 0.0, "step": 74 }, { "epoch": 0.01, "learning_rate": 8.5e-05, "loss": 0.0, "step": 75 }, { "epoch": 0.01, "learning_rate": 8.48e-05, "loss": 0.0, "step": 76 }, { "epoch": 0.01, "learning_rate": 8.46e-05, "loss": 0.0, "step": 77 }, { "epoch": 0.01, "learning_rate": 8.44e-05, "loss": 0.0, "step": 78 }, { "epoch": 0.01, "learning_rate": 8.42e-05, "loss": 0.0, "step": 79 }, { "epoch": 0.01, "learning_rate": 8.4e-05, "loss": 0.0, "step": 80 }, { "epoch": 0.01, "learning_rate": 8.38e-05, "loss": 0.0, "step": 81 }, { "epoch": 0.01, "learning_rate": 8.36e-05, "loss": 0.0, "step": 82 }, { "epoch": 0.01, "learning_rate": 8.34e-05, "loss": 0.0, "step": 83 }, { "epoch": 0.01, "learning_rate": 8.32e-05, "loss": 0.0, "step": 84 }, { "epoch": 0.01, "learning_rate": 8.3e-05, "loss": 0.0, "step": 85 }, { "epoch": 0.01, "learning_rate": 8.28e-05, "loss": 0.0, "step": 86 }, { "epoch": 0.01, "learning_rate": 8.26e-05, "loss": 0.0, "step": 87 }, { "epoch": 0.01, "learning_rate": 8.24e-05, "loss": 0.0, "step": 88 }, { "epoch": 0.01, "learning_rate": 8.22e-05, "loss": 0.0, "step": 89 }, { "epoch": 0.01, "learning_rate": 8.2e-05, "loss": 0.0, "step": 90 }, { "epoch": 0.01, "learning_rate": 8.18e-05, "loss": 0.0, "step": 91 }, { "epoch": 0.01, "learning_rate": 8.16e-05, "loss": 0.0, "step": 92 }, { "epoch": 0.01, "learning_rate": 8.14e-05, "loss": 0.0, "step": 93 }, { "epoch": 0.01, "learning_rate": 8.120000000000001e-05, "loss": 0.0, "step": 94 }, { "epoch": 0.01, "learning_rate": 8.1e-05, "loss": 0.0, "step": 95 }, { "epoch": 0.01, "learning_rate": 8.080000000000001e-05, "loss": 0.0, "step": 96 }, { "epoch": 0.01, "learning_rate": 8.060000000000001e-05, "loss": 0.0, "step": 97 }, { "epoch": 0.01, "learning_rate": 8.04e-05, "loss": 0.0, "step": 98 }, { "epoch": 0.01, "learning_rate": 8.020000000000001e-05, "loss": 0.0, "step": 99 }, { "epoch": 0.01, "learning_rate": 8e-05, "loss": 0.0, "step": 100 }, { "epoch": 0.01, "learning_rate": 7.98e-05, "loss": 0.0, "step": 101 }, { "epoch": 0.01, "learning_rate": 7.960000000000001e-05, "loss": 0.0, "step": 102 }, { "epoch": 0.01, "learning_rate": 7.94e-05, "loss": 0.0, "step": 103 }, { "epoch": 0.01, "learning_rate": 7.920000000000001e-05, "loss": 0.0, "step": 104 }, { "epoch": 0.01, "learning_rate": 7.900000000000001e-05, "loss": 0.0, "step": 105 }, { "epoch": 0.01, "learning_rate": 7.88e-05, "loss": 0.0, "step": 106 }, { "epoch": 0.01, "learning_rate": 7.860000000000001e-05, "loss": 0.0, "step": 107 }, { "epoch": 0.01, "learning_rate": 7.840000000000001e-05, "loss": 0.0, "step": 108 }, { "epoch": 0.01, "learning_rate": 7.82e-05, "loss": 0.0, "step": 109 }, { "epoch": 0.01, "learning_rate": 7.800000000000001e-05, "loss": 0.0, "step": 110 }, { "epoch": 0.01, "learning_rate": 7.780000000000001e-05, "loss": 0.0, "step": 111 }, { "epoch": 0.01, "learning_rate": 7.76e-05, "loss": 0.0, "step": 112 }, { "epoch": 0.01, "learning_rate": 7.740000000000001e-05, "loss": 0.0, "step": 113 }, { "epoch": 0.01, "learning_rate": 7.72e-05, "loss": 0.0, "step": 114 }, { "epoch": 0.01, "learning_rate": 7.7e-05, "loss": 0.0, "step": 115 }, { "epoch": 0.01, "learning_rate": 7.680000000000001e-05, "loss": 0.0, "step": 116 }, { "epoch": 0.01, "learning_rate": 7.66e-05, "loss": 0.0, "step": 117 }, { "epoch": 0.01, "learning_rate": 7.64e-05, "loss": 0.0, "step": 118 }, { "epoch": 0.01, "learning_rate": 7.620000000000001e-05, "loss": 0.0, "step": 119 }, { "epoch": 0.01, "learning_rate": 7.6e-05, "loss": 0.0, "step": 120 }, { "epoch": 0.01, "learning_rate": 7.58e-05, "loss": 0.0, "step": 121 }, { "epoch": 0.01, "learning_rate": 7.560000000000001e-05, "loss": 0.0, "step": 122 }, { "epoch": 0.01, "learning_rate": 7.54e-05, "loss": 0.0, "step": 123 }, { "epoch": 0.01, "learning_rate": 7.52e-05, "loss": 0.0, "step": 124 }, { "epoch": 0.01, "learning_rate": 7.500000000000001e-05, "loss": 0.0, "step": 125 }, { "epoch": 0.01, "learning_rate": 7.48e-05, "loss": 0.0, "step": 126 }, { "epoch": 0.01, "learning_rate": 7.46e-05, "loss": 0.0, "step": 127 }, { "epoch": 0.01, "learning_rate": 7.44e-05, "loss": 0.0, "step": 128 }, { "epoch": 0.01, "learning_rate": 7.42e-05, "loss": 0.0, "step": 129 }, { "epoch": 0.01, "learning_rate": 7.4e-05, "loss": 0.0, "step": 130 }, { "epoch": 0.01, "learning_rate": 7.38e-05, "loss": 0.0, "step": 131 }, { "epoch": 0.01, "learning_rate": 7.36e-05, "loss": 0.0, "step": 132 }, { "epoch": 0.01, "learning_rate": 7.340000000000001e-05, "loss": 0.0, "step": 133 }, { "epoch": 0.01, "learning_rate": 7.32e-05, "loss": 0.0, "step": 134 }, { "epoch": 0.01, "learning_rate": 7.3e-05, "loss": 0.0, "step": 135 }, { "epoch": 0.01, "learning_rate": 7.280000000000001e-05, "loss": 0.0, "step": 136 }, { "epoch": 0.01, "learning_rate": 7.26e-05, "loss": 0.0, "step": 137 }, { "epoch": 0.01, "learning_rate": 7.24e-05, "loss": 0.0, "step": 138 }, { "epoch": 0.01, "learning_rate": 7.22e-05, "loss": 0.0, "step": 139 }, { "epoch": 0.01, "learning_rate": 7.2e-05, "loss": 0.0, "step": 140 }, { "epoch": 0.01, "learning_rate": 7.18e-05, "loss": 0.0, "step": 141 }, { "epoch": 0.01, "learning_rate": 7.16e-05, "loss": 0.0, "step": 142 }, { "epoch": 0.01, "learning_rate": 7.14e-05, "loss": 0.0, "step": 143 }, { "epoch": 0.01, "learning_rate": 7.12e-05, "loss": 0.0, "step": 144 }, { "epoch": 0.01, "learning_rate": 7.1e-05, "loss": 0.0, "step": 145 }, { "epoch": 0.01, "learning_rate": 7.08e-05, "loss": 0.0, "step": 146 }, { "epoch": 0.01, "learning_rate": 7.06e-05, "loss": 0.0, "step": 147 }, { "epoch": 0.01, "learning_rate": 7.04e-05, "loss": 0.0, "step": 148 }, { "epoch": 0.01, "learning_rate": 7.02e-05, "loss": 0.0, "step": 149 }, { "epoch": 0.01, "learning_rate": 7e-05, "loss": 0.0, "step": 150 }, { "epoch": 0.01, "learning_rate": 6.98e-05, "loss": 0.0, "step": 151 }, { "epoch": 0.01, "learning_rate": 6.96e-05, "loss": 0.0, "step": 152 }, { "epoch": 0.01, "learning_rate": 6.939999999999999e-05, "loss": 0.0, "step": 153 }, { "epoch": 0.01, "learning_rate": 6.92e-05, "loss": 0.0, "step": 154 }, { "epoch": 0.01, "learning_rate": 6.9e-05, "loss": 0.0, "step": 155 }, { "epoch": 0.01, "learning_rate": 6.879999999999999e-05, "loss": 0.0, "step": 156 }, { "epoch": 0.01, "learning_rate": 6.860000000000001e-05, "loss": 0.0, "step": 157 }, { "epoch": 0.01, "learning_rate": 6.840000000000001e-05, "loss": 0.0, "step": 158 }, { "epoch": 0.01, "learning_rate": 6.82e-05, "loss": 0.0, "step": 159 }, { "epoch": 0.01, "learning_rate": 6.800000000000001e-05, "loss": 0.0, "step": 160 }, { "epoch": 0.01, "learning_rate": 6.780000000000001e-05, "loss": 0.0, "step": 161 }, { "epoch": 0.01, "learning_rate": 6.76e-05, "loss": 0.0, "step": 162 }, { "epoch": 0.01, "learning_rate": 6.740000000000001e-05, "loss": 0.0, "step": 163 }, { "epoch": 0.01, "learning_rate": 6.720000000000001e-05, "loss": 0.0, "step": 164 }, { "epoch": 0.01, "learning_rate": 6.7e-05, "loss": 0.0, "step": 165 }, { "epoch": 0.01, "learning_rate": 6.680000000000001e-05, "loss": 0.0, "step": 166 }, { "epoch": 0.01, "learning_rate": 6.66e-05, "loss": 0.0, "step": 167 }, { "epoch": 0.01, "learning_rate": 6.64e-05, "loss": 0.0, "step": 168 }, { "epoch": 0.01, "learning_rate": 6.620000000000001e-05, "loss": 0.0, "step": 169 }, { "epoch": 0.01, "learning_rate": 6.6e-05, "loss": 0.0, "step": 170 }, { "epoch": 0.01, "learning_rate": 6.58e-05, "loss": 0.0, "step": 171 }, { "epoch": 0.01, "learning_rate": 6.560000000000001e-05, "loss": 0.0, "step": 172 }, { "epoch": 0.01, "learning_rate": 6.54e-05, "loss": 0.0, "step": 173 }, { "epoch": 0.01, "learning_rate": 6.52e-05, "loss": 0.0, "step": 174 }, { "epoch": 0.01, "learning_rate": 6.500000000000001e-05, "loss": 0.0, "step": 175 }, { "epoch": 0.01, "learning_rate": 6.48e-05, "loss": 0.0, "step": 176 }, { "epoch": 0.01, "learning_rate": 6.460000000000001e-05, "loss": 0.0, "step": 177 }, { "epoch": 0.01, "learning_rate": 6.440000000000001e-05, "loss": 0.0, "step": 178 }, { "epoch": 0.01, "learning_rate": 6.42e-05, "loss": 0.0, "step": 179 }, { "epoch": 0.01, "learning_rate": 6.400000000000001e-05, "loss": 0.0, "step": 180 }, { "epoch": 0.01, "learning_rate": 6.38e-05, "loss": 0.0, "step": 181 }, { "epoch": 0.01, "learning_rate": 6.36e-05, "loss": 0.0, "step": 182 }, { "epoch": 0.01, "learning_rate": 6.340000000000001e-05, "loss": 0.0, "step": 183 }, { "epoch": 0.01, "learning_rate": 6.32e-05, "loss": 0.0, "step": 184 }, { "epoch": 0.01, "learning_rate": 6.3e-05, "loss": 0.0, "step": 185 }, { "epoch": 0.01, "learning_rate": 6.280000000000001e-05, "loss": 0.0, "step": 186 }, { "epoch": 0.01, "learning_rate": 6.26e-05, "loss": 0.0, "step": 187 }, { "epoch": 0.01, "learning_rate": 6.24e-05, "loss": 0.0, "step": 188 }, { "epoch": 0.01, "learning_rate": 6.220000000000001e-05, "loss": 0.0, "step": 189 }, { "epoch": 0.01, "learning_rate": 6.2e-05, "loss": 0.0, "step": 190 }, { "epoch": 0.01, "learning_rate": 6.18e-05, "loss": 0.0, "step": 191 }, { "epoch": 0.01, "learning_rate": 6.16e-05, "loss": 0.0, "step": 192 }, { "epoch": 0.01, "learning_rate": 6.14e-05, "loss": 0.0, "step": 193 }, { "epoch": 0.01, "learning_rate": 6.12e-05, "loss": 0.0, "step": 194 }, { "epoch": 0.01, "learning_rate": 6.1e-05, "loss": 0.0, "step": 195 }, { "epoch": 0.02, "learning_rate": 6.08e-05, "loss": 0.0, "step": 196 }, { "epoch": 0.02, "learning_rate": 6.06e-05, "loss": 0.0, "step": 197 }, { "epoch": 0.02, "learning_rate": 6.04e-05, "loss": 0.0, "step": 198 }, { "epoch": 0.02, "learning_rate": 6.02e-05, "loss": 0.0, "step": 199 }, { "epoch": 0.02, "learning_rate": 6e-05, "loss": 0.0, "step": 200 }, { "epoch": 0.02, "learning_rate": 5.9800000000000003e-05, "loss": 0.0, "step": 201 }, { "epoch": 0.02, "learning_rate": 5.96e-05, "loss": 0.0, "step": 202 }, { "epoch": 0.02, "learning_rate": 5.94e-05, "loss": 0.0, "step": 203 }, { "epoch": 0.02, "learning_rate": 5.92e-05, "loss": 0.0, "step": 204 }, { "epoch": 0.02, "learning_rate": 5.9e-05, "loss": 0.0, "step": 205 }, { "epoch": 0.02, "learning_rate": 5.88e-05, "loss": 0.0, "step": 206 }, { "epoch": 0.02, "learning_rate": 5.86e-05, "loss": 0.0, "step": 207 }, { "epoch": 0.02, "learning_rate": 5.8399999999999997e-05, "loss": 0.0, "step": 208 }, { "epoch": 0.02, "learning_rate": 5.82e-05, "loss": 0.0, "step": 209 }, { "epoch": 0.02, "learning_rate": 5.8e-05, "loss": 0.0, "step": 210 }, { "epoch": 0.02, "learning_rate": 5.7799999999999995e-05, "loss": 0.0, "step": 211 }, { "epoch": 0.02, "learning_rate": 5.76e-05, "loss": 0.0, "step": 212 }, { "epoch": 0.02, "learning_rate": 5.74e-05, "loss": 0.0, "step": 213 }, { "epoch": 0.02, "learning_rate": 5.72e-05, "loss": 0.0, "step": 214 }, { "epoch": 0.02, "learning_rate": 5.6999999999999996e-05, "loss": 0.0, "step": 215 }, { "epoch": 0.02, "learning_rate": 5.68e-05, "loss": 0.0, "step": 216 }, { "epoch": 0.02, "learning_rate": 5.66e-05, "loss": 0.0, "step": 217 }, { "epoch": 0.02, "learning_rate": 5.6399999999999995e-05, "loss": 0.0, "step": 218 }, { "epoch": 0.02, "learning_rate": 5.620000000000001e-05, "loss": 0.0, "step": 219 }, { "epoch": 0.02, "learning_rate": 5.6000000000000006e-05, "loss": 0.0, "step": 220 }, { "epoch": 0.02, "learning_rate": 5.580000000000001e-05, "loss": 0.0, "step": 221 }, { "epoch": 0.02, "learning_rate": 5.560000000000001e-05, "loss": 0.0, "step": 222 }, { "epoch": 0.02, "learning_rate": 5.5400000000000005e-05, "loss": 0.0, "step": 223 }, { "epoch": 0.02, "learning_rate": 5.520000000000001e-05, "loss": 0.0, "step": 224 }, { "epoch": 0.02, "learning_rate": 5.500000000000001e-05, "loss": 0.0, "step": 225 }, { "epoch": 0.02, "learning_rate": 5.4800000000000004e-05, "loss": 0.0, "step": 226 }, { "epoch": 0.02, "learning_rate": 5.4600000000000006e-05, "loss": 0.0, "step": 227 }, { "epoch": 0.02, "learning_rate": 5.440000000000001e-05, "loss": 0.0, "step": 228 }, { "epoch": 0.02, "learning_rate": 5.420000000000001e-05, "loss": 0.0, "step": 229 }, { "epoch": 0.02, "learning_rate": 5.4000000000000005e-05, "loss": 0.0, "step": 230 }, { "epoch": 0.02, "learning_rate": 5.380000000000001e-05, "loss": 0.0, "step": 231 }, { "epoch": 0.02, "learning_rate": 5.360000000000001e-05, "loss": 0.0, "step": 232 }, { "epoch": 0.02, "learning_rate": 5.3400000000000004e-05, "loss": 0.0, "step": 233 }, { "epoch": 0.02, "learning_rate": 5.3200000000000006e-05, "loss": 0.0, "step": 234 }, { "epoch": 0.02, "learning_rate": 5.300000000000001e-05, "loss": 0.0, "step": 235 }, { "epoch": 0.02, "learning_rate": 5.28e-05, "loss": 0.0, "step": 236 }, { "epoch": 0.02, "learning_rate": 5.2600000000000005e-05, "loss": 0.0, "step": 237 }, { "epoch": 0.02, "learning_rate": 5.2400000000000007e-05, "loss": 0.0, "step": 238 }, { "epoch": 0.02, "learning_rate": 5.22e-05, "loss": 0.0, "step": 239 }, { "epoch": 0.02, "learning_rate": 5.2000000000000004e-05, "loss": 0.0, "step": 240 }, { "epoch": 0.02, "learning_rate": 5.1800000000000005e-05, "loss": 0.0, "step": 241 }, { "epoch": 0.02, "learning_rate": 5.16e-05, "loss": 0.0, "step": 242 }, { "epoch": 0.02, "learning_rate": 5.14e-05, "loss": 0.0, "step": 243 }, { "epoch": 0.02, "learning_rate": 5.1200000000000004e-05, "loss": 0.0, "step": 244 }, { "epoch": 0.02, "learning_rate": 5.1000000000000006e-05, "loss": 0.0, "step": 245 }, { "epoch": 0.02, "learning_rate": 5.08e-05, "loss": 0.0, "step": 246 }, { "epoch": 0.02, "learning_rate": 5.0600000000000003e-05, "loss": 0.0, "step": 247 }, { "epoch": 0.02, "learning_rate": 5.0400000000000005e-05, "loss": 0.0, "step": 248 }, { "epoch": 0.02, "learning_rate": 5.02e-05, "loss": 0.0, "step": 249 }, { "epoch": 0.02, "learning_rate": 5e-05, "loss": 0.0, "step": 250 }, { "epoch": 0.02, "learning_rate": 4.9800000000000004e-05, "loss": 0.0, "step": 251 }, { "epoch": 0.02, "learning_rate": 4.96e-05, "loss": 0.0, "step": 252 }, { "epoch": 0.02, "learning_rate": 4.94e-05, "loss": 0.0, "step": 253 }, { "epoch": 0.02, "learning_rate": 4.92e-05, "loss": 0.0, "step": 254 }, { "epoch": 0.02, "learning_rate": 4.9e-05, "loss": 0.0, "step": 255 }, { "epoch": 0.02, "learning_rate": 4.88e-05, "loss": 0.0, "step": 256 }, { "epoch": 0.02, "learning_rate": 4.86e-05, "loss": 0.0, "step": 257 }, { "epoch": 0.02, "learning_rate": 4.8400000000000004e-05, "loss": 0.0, "step": 258 }, { "epoch": 0.02, "learning_rate": 4.82e-05, "loss": 0.0, "step": 259 }, { "epoch": 0.02, "learning_rate": 4.8e-05, "loss": 0.0, "step": 260 }, { "epoch": 0.02, "learning_rate": 4.78e-05, "loss": 0.0, "step": 261 }, { "epoch": 0.02, "learning_rate": 4.76e-05, "loss": 0.0, "step": 262 }, { "epoch": 0.02, "learning_rate": 4.74e-05, "loss": 0.0, "step": 263 }, { "epoch": 0.02, "learning_rate": 4.72e-05, "loss": 0.0, "step": 264 }, { "epoch": 0.02, "learning_rate": 4.7e-05, "loss": 0.0, "step": 265 }, { "epoch": 0.02, "learning_rate": 4.6800000000000006e-05, "loss": 0.0, "step": 266 }, { "epoch": 0.02, "learning_rate": 4.660000000000001e-05, "loss": 0.0, "step": 267 }, { "epoch": 0.02, "learning_rate": 4.64e-05, "loss": 0.0, "step": 268 }, { "epoch": 0.02, "learning_rate": 4.6200000000000005e-05, "loss": 0.0, "step": 269 }, { "epoch": 0.02, "learning_rate": 4.600000000000001e-05, "loss": 0.0, "step": 270 }, { "epoch": 0.02, "learning_rate": 4.58e-05, "loss": 0.0, "step": 271 }, { "epoch": 0.02, "learning_rate": 4.5600000000000004e-05, "loss": 0.0, "step": 272 }, { "epoch": 0.02, "learning_rate": 4.5400000000000006e-05, "loss": 0.0, "step": 273 }, { "epoch": 0.02, "learning_rate": 4.52e-05, "loss": 0.0, "step": 274 }, { "epoch": 0.02, "learning_rate": 4.5e-05, "loss": 0.0, "step": 275 }, { "epoch": 0.02, "learning_rate": 4.4800000000000005e-05, "loss": 0.0, "step": 276 }, { "epoch": 0.02, "learning_rate": 4.46e-05, "loss": 0.0, "step": 277 }, { "epoch": 0.02, "learning_rate": 4.44e-05, "loss": 0.0, "step": 278 }, { "epoch": 0.02, "learning_rate": 4.4200000000000004e-05, "loss": 0.0, "step": 279 }, { "epoch": 0.02, "learning_rate": 4.4000000000000006e-05, "loss": 0.0, "step": 280 }, { "epoch": 0.02, "learning_rate": 4.38e-05, "loss": 0.0, "step": 281 }, { "epoch": 0.02, "learning_rate": 4.36e-05, "loss": 0.0, "step": 282 }, { "epoch": 0.02, "learning_rate": 4.3400000000000005e-05, "loss": 0.0, "step": 283 }, { "epoch": 0.02, "learning_rate": 4.32e-05, "loss": 0.0, "step": 284 }, { "epoch": 0.02, "learning_rate": 4.3e-05, "loss": 0.0, "step": 285 }, { "epoch": 0.02, "learning_rate": 4.2800000000000004e-05, "loss": 0.0, "step": 286 }, { "epoch": 0.02, "learning_rate": 4.26e-05, "loss": 0.0, "step": 287 }, { "epoch": 0.02, "learning_rate": 4.24e-05, "loss": 0.0, "step": 288 }, { "epoch": 0.02, "learning_rate": 4.22e-05, "loss": 0.0, "step": 289 }, { "epoch": 0.02, "learning_rate": 4.2e-05, "loss": 0.0, "step": 290 }, { "epoch": 0.02, "learning_rate": 4.18e-05, "loss": 0.0, "step": 291 }, { "epoch": 0.02, "learning_rate": 4.16e-05, "loss": 0.0, "step": 292 }, { "epoch": 0.02, "learning_rate": 4.14e-05, "loss": 0.0, "step": 293 }, { "epoch": 0.02, "learning_rate": 4.12e-05, "loss": 0.0, "step": 294 }, { "epoch": 0.02, "learning_rate": 4.1e-05, "loss": 0.0, "step": 295 }, { "epoch": 0.02, "learning_rate": 4.08e-05, "loss": 0.0, "step": 296 }, { "epoch": 0.02, "learning_rate": 4.0600000000000004e-05, "loss": 0.0, "step": 297 }, { "epoch": 0.02, "learning_rate": 4.0400000000000006e-05, "loss": 0.0, "step": 298 }, { "epoch": 0.02, "learning_rate": 4.02e-05, "loss": 0.0, "step": 299 }, { "epoch": 0.02, "learning_rate": 4e-05, "loss": 0.0, "step": 300 }, { "epoch": 0.02, "learning_rate": 3.9800000000000005e-05, "loss": 0.0, "step": 301 }, { "epoch": 0.02, "learning_rate": 3.960000000000001e-05, "loss": 0.0, "step": 302 }, { "epoch": 0.02, "learning_rate": 3.94e-05, "loss": 0.0, "step": 303 }, { "epoch": 0.02, "learning_rate": 3.9200000000000004e-05, "loss": 0.0, "step": 304 }, { "epoch": 0.02, "learning_rate": 3.9000000000000006e-05, "loss": 0.0, "step": 305 }, { "epoch": 0.02, "learning_rate": 3.88e-05, "loss": 0.0, "step": 306 }, { "epoch": 0.02, "learning_rate": 3.86e-05, "loss": 0.0, "step": 307 }, { "epoch": 0.02, "learning_rate": 3.8400000000000005e-05, "loss": 0.0, "step": 308 }, { "epoch": 0.02, "learning_rate": 3.82e-05, "loss": 0.0, "step": 309 }, { "epoch": 0.02, "learning_rate": 3.8e-05, "loss": 0.0, "step": 310 }, { "epoch": 0.02, "learning_rate": 3.7800000000000004e-05, "loss": 0.0, "step": 311 }, { "epoch": 0.02, "learning_rate": 3.76e-05, "loss": 0.0, "step": 312 }, { "epoch": 0.02, "learning_rate": 3.74e-05, "loss": 0.0, "step": 313 }, { "epoch": 0.02, "learning_rate": 3.72e-05, "loss": 0.0, "step": 314 }, { "epoch": 0.02, "learning_rate": 3.7e-05, "loss": 0.0, "step": 315 }, { "epoch": 0.02, "learning_rate": 3.68e-05, "loss": 0.0, "step": 316 }, { "epoch": 0.02, "learning_rate": 3.66e-05, "loss": 0.0, "step": 317 }, { "epoch": 0.02, "learning_rate": 3.6400000000000004e-05, "loss": 0.0, "step": 318 }, { "epoch": 0.02, "learning_rate": 3.62e-05, "loss": 0.0, "step": 319 }, { "epoch": 0.02, "learning_rate": 3.6e-05, "loss": 0.0, "step": 320 }, { "epoch": 0.02, "learning_rate": 3.58e-05, "loss": 0.0, "step": 321 }, { "epoch": 0.02, "learning_rate": 3.56e-05, "loss": 0.0, "step": 322 }, { "epoch": 0.02, "learning_rate": 3.54e-05, "loss": 0.0, "step": 323 }, { "epoch": 0.02, "learning_rate": 3.52e-05, "loss": 0.0, "step": 324 }, { "epoch": 0.02, "learning_rate": 3.5e-05, "loss": 0.0, "step": 325 }, { "epoch": 0.03, "learning_rate": 3.48e-05, "loss": 0.0, "step": 326 }, { "epoch": 0.03, "learning_rate": 3.46e-05, "loss": 0.0, "step": 327 }, { "epoch": 0.03, "learning_rate": 3.4399999999999996e-05, "loss": 0.0, "step": 328 }, { "epoch": 0.03, "learning_rate": 3.4200000000000005e-05, "loss": 0.0, "step": 329 }, { "epoch": 0.03, "learning_rate": 3.4000000000000007e-05, "loss": 0.0, "step": 330 }, { "epoch": 0.03, "learning_rate": 3.38e-05, "loss": 0.0, "step": 331 }, { "epoch": 0.03, "learning_rate": 3.3600000000000004e-05, "loss": 0.0, "step": 332 }, { "epoch": 0.03, "learning_rate": 3.3400000000000005e-05, "loss": 0.0, "step": 333 }, { "epoch": 0.03, "learning_rate": 3.32e-05, "loss": 0.0, "step": 334 }, { "epoch": 0.03, "learning_rate": 3.3e-05, "loss": 0.0, "step": 335 }, { "epoch": 0.03, "learning_rate": 3.2800000000000004e-05, "loss": 0.0, "step": 336 }, { "epoch": 0.03, "learning_rate": 3.26e-05, "loss": 0.0, "step": 337 }, { "epoch": 0.03, "learning_rate": 3.24e-05, "loss": 0.0, "step": 338 }, { "epoch": 0.03, "learning_rate": 3.2200000000000003e-05, "loss": 0.0, "step": 339 }, { "epoch": 0.03, "learning_rate": 3.2000000000000005e-05, "loss": 0.0, "step": 340 }, { "epoch": 0.03, "learning_rate": 3.18e-05, "loss": 0.0, "step": 341 }, { "epoch": 0.03, "learning_rate": 3.16e-05, "loss": 0.0, "step": 342 }, { "epoch": 0.03, "learning_rate": 3.1400000000000004e-05, "loss": 0.0, "step": 343 }, { "epoch": 0.03, "learning_rate": 3.12e-05, "loss": 0.0, "step": 344 }, { "epoch": 0.03, "learning_rate": 3.1e-05, "loss": 0.0, "step": 345 }, { "epoch": 0.03, "learning_rate": 3.08e-05, "loss": 0.0, "step": 346 }, { "epoch": 0.03, "learning_rate": 3.06e-05, "loss": 0.0, "step": 347 }, { "epoch": 0.03, "learning_rate": 3.04e-05, "loss": 0.0, "step": 348 }, { "epoch": 0.03, "learning_rate": 3.02e-05, "loss": 0.0, "step": 349 }, { "epoch": 0.03, "learning_rate": 3e-05, "loss": 0.0, "step": 350 }, { "epoch": 0.03, "learning_rate": 2.98e-05, "loss": 0.0, "step": 351 }, { "epoch": 0.03, "learning_rate": 2.96e-05, "loss": 0.0, "step": 352 }, { "epoch": 0.03, "learning_rate": 2.94e-05, "loss": 0.0, "step": 353 }, { "epoch": 0.03, "learning_rate": 2.9199999999999998e-05, "loss": 0.0, "step": 354 }, { "epoch": 0.03, "learning_rate": 2.9e-05, "loss": 0.0, "step": 355 }, { "epoch": 0.03, "learning_rate": 2.88e-05, "loss": 0.0, "step": 356 }, { "epoch": 0.03, "learning_rate": 2.86e-05, "loss": 0.0, "step": 357 }, { "epoch": 0.03, "learning_rate": 2.84e-05, "loss": 0.0, "step": 358 }, { "epoch": 0.03, "learning_rate": 2.8199999999999998e-05, "loss": 0.0, "step": 359 }, { "epoch": 0.03, "learning_rate": 2.8000000000000003e-05, "loss": 0.0, "step": 360 }, { "epoch": 0.03, "learning_rate": 2.7800000000000005e-05, "loss": 0.0, "step": 361 }, { "epoch": 0.03, "learning_rate": 2.7600000000000003e-05, "loss": 0.0, "step": 362 }, { "epoch": 0.03, "learning_rate": 2.7400000000000002e-05, "loss": 0.0, "step": 363 }, { "epoch": 0.03, "learning_rate": 2.7200000000000004e-05, "loss": 0.0, "step": 364 }, { "epoch": 0.03, "learning_rate": 2.7000000000000002e-05, "loss": 0.0, "step": 365 }, { "epoch": 0.03, "learning_rate": 2.6800000000000004e-05, "loss": 0.0, "step": 366 }, { "epoch": 0.03, "learning_rate": 2.6600000000000003e-05, "loss": 0.0, "step": 367 }, { "epoch": 0.03, "learning_rate": 2.64e-05, "loss": 0.0, "step": 368 }, { "epoch": 0.03, "learning_rate": 2.6200000000000003e-05, "loss": 0.0, "step": 369 }, { "epoch": 0.03, "learning_rate": 2.6000000000000002e-05, "loss": 0.0, "step": 370 }, { "epoch": 0.03, "learning_rate": 2.58e-05, "loss": 0.0, "step": 371 }, { "epoch": 0.03, "learning_rate": 2.5600000000000002e-05, "loss": 0.0, "step": 372 }, { "epoch": 0.03, "learning_rate": 2.54e-05, "loss": 0.0, "step": 373 }, { "epoch": 0.03, "learning_rate": 2.5200000000000003e-05, "loss": 0.0, "step": 374 }, { "epoch": 0.03, "learning_rate": 2.5e-05, "loss": 0.0, "step": 375 }, { "epoch": 0.03, "learning_rate": 2.48e-05, "loss": 0.0, "step": 376 }, { "epoch": 0.03, "learning_rate": 2.46e-05, "loss": 0.0, "step": 377 }, { "epoch": 0.03, "learning_rate": 2.44e-05, "loss": 0.0, "step": 378 }, { "epoch": 0.03, "learning_rate": 2.4200000000000002e-05, "loss": 0.0, "step": 379 }, { "epoch": 0.03, "learning_rate": 2.4e-05, "loss": 0.0, "step": 380 }, { "epoch": 0.03, "learning_rate": 2.38e-05, "loss": 0.0, "step": 381 }, { "epoch": 0.03, "learning_rate": 2.36e-05, "loss": 0.0, "step": 382 }, { "epoch": 0.03, "learning_rate": 2.3400000000000003e-05, "loss": 0.0, "step": 383 }, { "epoch": 0.03, "learning_rate": 2.32e-05, "loss": 0.0, "step": 384 }, { "epoch": 0.03, "learning_rate": 2.3000000000000003e-05, "loss": 0.0, "step": 385 }, { "epoch": 0.03, "learning_rate": 2.2800000000000002e-05, "loss": 0.0, "step": 386 }, { "epoch": 0.03, "learning_rate": 2.26e-05, "loss": 0.0, "step": 387 }, { "epoch": 0.03, "learning_rate": 2.2400000000000002e-05, "loss": 0.0, "step": 388 }, { "epoch": 0.03, "learning_rate": 2.22e-05, "loss": 0.0, "step": 389 }, { "epoch": 0.03, "learning_rate": 2.2000000000000003e-05, "loss": 0.0, "step": 390 }, { "epoch": 0.03, "learning_rate": 2.18e-05, "loss": 0.0, "step": 391 }, { "epoch": 0.03, "learning_rate": 2.16e-05, "loss": 0.0, "step": 392 }, { "epoch": 0.03, "learning_rate": 2.1400000000000002e-05, "loss": 0.0, "step": 393 }, { "epoch": 0.03, "learning_rate": 2.12e-05, "loss": 0.0, "step": 394 }, { "epoch": 0.03, "learning_rate": 2.1e-05, "loss": 0.0, "step": 395 }, { "epoch": 0.03, "learning_rate": 2.08e-05, "loss": 0.0, "step": 396 }, { "epoch": 0.03, "learning_rate": 2.06e-05, "loss": 0.0, "step": 397 }, { "epoch": 0.03, "learning_rate": 2.04e-05, "loss": 0.0, "step": 398 }, { "epoch": 0.03, "learning_rate": 2.0200000000000003e-05, "loss": 0.0, "step": 399 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.0, "step": 400 }, { "epoch": 0.03, "learning_rate": 1.9800000000000004e-05, "loss": 0.0, "step": 401 }, { "epoch": 0.03, "learning_rate": 1.9600000000000002e-05, "loss": 0.0, "step": 402 }, { "epoch": 0.03, "learning_rate": 1.94e-05, "loss": 0.0, "step": 403 }, { "epoch": 0.03, "learning_rate": 1.9200000000000003e-05, "loss": 0.0, "step": 404 }, { "epoch": 0.03, "learning_rate": 1.9e-05, "loss": 0.0, "step": 405 }, { "epoch": 0.03, "learning_rate": 1.88e-05, "loss": 0.0, "step": 406 }, { "epoch": 0.03, "learning_rate": 1.86e-05, "loss": 0.0, "step": 407 }, { "epoch": 0.03, "learning_rate": 1.84e-05, "loss": 0.0, "step": 408 }, { "epoch": 0.03, "learning_rate": 1.8200000000000002e-05, "loss": 0.0, "step": 409 }, { "epoch": 0.03, "learning_rate": 1.8e-05, "loss": 0.0, "step": 410 }, { "epoch": 0.03, "learning_rate": 1.78e-05, "loss": 0.0, "step": 411 }, { "epoch": 0.03, "learning_rate": 1.76e-05, "loss": 0.0, "step": 412 }, { "epoch": 0.03, "learning_rate": 1.74e-05, "loss": 0.0, "step": 413 }, { "epoch": 0.03, "learning_rate": 1.7199999999999998e-05, "loss": 0.0, "step": 414 }, { "epoch": 0.03, "learning_rate": 1.7000000000000003e-05, "loss": 0.0, "step": 415 }, { "epoch": 0.03, "learning_rate": 1.6800000000000002e-05, "loss": 0.0, "step": 416 }, { "epoch": 0.03, "learning_rate": 1.66e-05, "loss": 0.0, "step": 417 }, { "epoch": 0.03, "learning_rate": 1.6400000000000002e-05, "loss": 0.0, "step": 418 }, { "epoch": 0.03, "learning_rate": 1.62e-05, "loss": 0.0, "step": 419 }, { "epoch": 0.03, "learning_rate": 1.6000000000000003e-05, "loss": 0.0, "step": 420 }, { "epoch": 0.03, "learning_rate": 1.58e-05, "loss": 0.0, "step": 421 }, { "epoch": 0.03, "learning_rate": 1.56e-05, "loss": 0.0, "step": 422 }, { "epoch": 0.03, "learning_rate": 1.54e-05, "loss": 0.0, "step": 423 }, { "epoch": 0.03, "learning_rate": 1.52e-05, "loss": 0.0, "step": 424 }, { "epoch": 0.03, "learning_rate": 1.5e-05, "loss": 0.0, "step": 425 }, { "epoch": 0.03, "learning_rate": 1.48e-05, "loss": 0.0, "step": 426 }, { "epoch": 0.03, "learning_rate": 1.4599999999999999e-05, "loss": 0.0, "step": 427 }, { "epoch": 0.03, "learning_rate": 1.44e-05, "loss": 0.0, "step": 428 }, { "epoch": 0.03, "learning_rate": 1.42e-05, "loss": 0.0, "step": 429 }, { "epoch": 0.03, "learning_rate": 1.4000000000000001e-05, "loss": 0.0, "step": 430 }, { "epoch": 0.03, "learning_rate": 1.3800000000000002e-05, "loss": 0.0, "step": 431 }, { "epoch": 0.03, "learning_rate": 1.3600000000000002e-05, "loss": 0.0, "step": 432 }, { "epoch": 0.03, "learning_rate": 1.3400000000000002e-05, "loss": 0.0, "step": 433 }, { "epoch": 0.03, "learning_rate": 1.32e-05, "loss": 0.0, "step": 434 }, { "epoch": 0.03, "learning_rate": 1.3000000000000001e-05, "loss": 0.0, "step": 435 }, { "epoch": 0.03, "learning_rate": 1.2800000000000001e-05, "loss": 0.0, "step": 436 }, { "epoch": 0.03, "learning_rate": 1.2600000000000001e-05, "loss": 0.0, "step": 437 }, { "epoch": 0.03, "learning_rate": 1.24e-05, "loss": 0.0, "step": 438 }, { "epoch": 0.03, "learning_rate": 1.22e-05, "loss": 0.0, "step": 439 }, { "epoch": 0.03, "learning_rate": 1.2e-05, "loss": 0.0, "step": 440 }, { "epoch": 0.03, "learning_rate": 1.18e-05, "loss": 0.0, "step": 441 }, { "epoch": 0.03, "learning_rate": 1.16e-05, "loss": 0.0, "step": 442 }, { "epoch": 0.03, "learning_rate": 1.1400000000000001e-05, "loss": 0.0, "step": 443 }, { "epoch": 0.03, "learning_rate": 1.1200000000000001e-05, "loss": 0.0, "step": 444 }, { "epoch": 0.03, "learning_rate": 1.1000000000000001e-05, "loss": 0.0, "step": 445 }, { "epoch": 0.03, "learning_rate": 1.08e-05, "loss": 0.0, "step": 446 }, { "epoch": 0.03, "learning_rate": 1.06e-05, "loss": 0.0, "step": 447 }, { "epoch": 0.03, "learning_rate": 1.04e-05, "loss": 0.0, "step": 448 }, { "epoch": 0.03, "learning_rate": 1.02e-05, "loss": 0.0, "step": 449 }, { "epoch": 0.03, "learning_rate": 1e-05, "loss": 0.0, "step": 450 }, { "epoch": 0.03, "learning_rate": 9.800000000000001e-06, "loss": 0.0, "step": 451 }, { "epoch": 0.03, "learning_rate": 9.600000000000001e-06, "loss": 0.0, "step": 452 }, { "epoch": 0.03, "learning_rate": 9.4e-06, "loss": 0.0, "step": 453 }, { "epoch": 0.03, "learning_rate": 9.2e-06, "loss": 0.0, "step": 454 }, { "epoch": 0.03, "learning_rate": 9e-06, "loss": 0.0, "step": 455 }, { "epoch": 0.04, "learning_rate": 8.8e-06, "loss": 0.0, "step": 456 }, { "epoch": 0.04, "learning_rate": 8.599999999999999e-06, "loss": 0.0, "step": 457 }, { "epoch": 0.04, "learning_rate": 8.400000000000001e-06, "loss": 0.0, "step": 458 }, { "epoch": 0.04, "learning_rate": 8.200000000000001e-06, "loss": 0.0, "step": 459 }, { "epoch": 0.04, "learning_rate": 8.000000000000001e-06, "loss": 0.0, "step": 460 }, { "epoch": 0.04, "learning_rate": 7.8e-06, "loss": 0.0, "step": 461 }, { "epoch": 0.04, "learning_rate": 7.6e-06, "loss": 0.0, "step": 462 }, { "epoch": 0.04, "learning_rate": 7.4e-06, "loss": 0.0, "step": 463 }, { "epoch": 0.04, "learning_rate": 7.2e-06, "loss": 0.0, "step": 464 }, { "epoch": 0.04, "learning_rate": 7.000000000000001e-06, "loss": 0.0, "step": 465 }, { "epoch": 0.04, "learning_rate": 6.800000000000001e-06, "loss": 0.0, "step": 466 }, { "epoch": 0.04, "learning_rate": 6.6e-06, "loss": 0.0, "step": 467 }, { "epoch": 0.04, "learning_rate": 6.4000000000000006e-06, "loss": 0.0, "step": 468 }, { "epoch": 0.04, "learning_rate": 6.2e-06, "loss": 0.0, "step": 469 }, { "epoch": 0.04, "learning_rate": 6e-06, "loss": 0.0, "step": 470 }, { "epoch": 0.04, "learning_rate": 5.8e-06, "loss": 0.0, "step": 471 }, { "epoch": 0.04, "learning_rate": 5.600000000000001e-06, "loss": 0.0, "step": 472 }, { "epoch": 0.04, "learning_rate": 5.4e-06, "loss": 0.0, "step": 473 }, { "epoch": 0.04, "learning_rate": 5.2e-06, "loss": 0.0, "step": 474 }, { "epoch": 0.04, "learning_rate": 5e-06, "loss": 0.0, "step": 475 }, { "epoch": 0.04, "learning_rate": 4.800000000000001e-06, "loss": 0.0, "step": 476 }, { "epoch": 0.04, "learning_rate": 4.6e-06, "loss": 0.0, "step": 477 }, { "epoch": 0.04, "learning_rate": 4.4e-06, "loss": 0.0, "step": 478 }, { "epoch": 0.04, "learning_rate": 4.2000000000000004e-06, "loss": 0.0, "step": 479 }, { "epoch": 0.04, "learning_rate": 4.000000000000001e-06, "loss": 0.0, "step": 480 }, { "epoch": 0.04, "learning_rate": 3.8e-06, "loss": 0.0, "step": 481 }, { "epoch": 0.04, "learning_rate": 3.6e-06, "loss": 0.0, "step": 482 }, { "epoch": 0.04, "learning_rate": 3.4000000000000005e-06, "loss": 0.0, "step": 483 }, { "epoch": 0.04, "learning_rate": 3.2000000000000003e-06, "loss": 0.0, "step": 484 }, { "epoch": 0.04, "learning_rate": 3e-06, "loss": 0.0, "step": 485 }, { "epoch": 0.04, "learning_rate": 2.8000000000000003e-06, "loss": 0.0, "step": 486 }, { "epoch": 0.04, "learning_rate": 2.6e-06, "loss": 0.0, "step": 487 }, { "epoch": 0.04, "learning_rate": 2.4000000000000003e-06, "loss": 0.0, "step": 488 }, { "epoch": 0.04, "learning_rate": 2.2e-06, "loss": 0.0, "step": 489 }, { "epoch": 0.04, "learning_rate": 2.0000000000000003e-06, "loss": 0.0, "step": 490 }, { "epoch": 0.04, "learning_rate": 1.8e-06, "loss": 0.0, "step": 491 }, { "epoch": 0.04, "learning_rate": 1.6000000000000001e-06, "loss": 0.0, "step": 492 }, { "epoch": 0.04, "learning_rate": 1.4000000000000001e-06, "loss": 0.0, "step": 493 }, { "epoch": 0.04, "learning_rate": 1.2000000000000002e-06, "loss": 0.0, "step": 494 }, { "epoch": 0.04, "learning_rate": 1.0000000000000002e-06, "loss": 0.0, "step": 495 }, { "epoch": 0.04, "learning_rate": 8.000000000000001e-07, "loss": 0.0, "step": 496 }, { "epoch": 0.04, "learning_rate": 6.000000000000001e-07, "loss": 0.0, "step": 497 }, { "epoch": 0.04, "learning_rate": 4.0000000000000003e-07, "loss": 0.0, "step": 498 }, { "epoch": 0.04, "learning_rate": 2.0000000000000002e-07, "loss": 0.0, "step": 499 }, { "epoch": 0.04, "learning_rate": 0.0, "loss": 0.0, "step": 500 }, { "epoch": 0.04, "step": 500, "total_flos": 8.2728150106112e+16, "train_loss": 0.002908447265625, "train_runtime": 536.6749, "train_samples_per_second": 3.727, "train_steps_per_second": 0.932 } ], "logging_steps": 1.0, "max_steps": 500, "num_train_epochs": 1, "save_steps": 50, "total_flos": 8.2728150106112e+16, "trial_name": null, "trial_params": null }