| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.05351075120176229, | |
| "eval_steps": 500, | |
| "global_step": 1500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 5.8591, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 5.9707, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 5.9586, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 5.9471, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 5.9853, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 6.012, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 5.7527, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 5.9447, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 5.8183, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8e-06, | |
| "loss": 5.9652, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 5.922, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.2e-06, | |
| "loss": 5.422, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 5.9541, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6e-06, | |
| "loss": 5.2348, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 5.7591, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3e-06, | |
| "loss": 5.9378, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 5.4748, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 5.9678, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.6e-06, | |
| "loss": 5.3809, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.8e-06, | |
| "loss": 5.6915, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 5.8738, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 5.7587, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4e-06, | |
| "loss": 5.3271, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6e-06, | |
| "loss": 5.785, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 5.9755, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5e-06, | |
| "loss": 5.6424, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.2e-06, | |
| "loss": 6.2383, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.4e-06, | |
| "loss": 5.7105, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 5.9898, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.8e-06, | |
| "loss": 5.0456, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6e-06, | |
| "loss": 5.6215, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.2e-06, | |
| "loss": 5.2618, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 5.7864, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.6e-06, | |
| "loss": 6.0545, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 5.7379, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 6.0813, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.2e-06, | |
| "loss": 5.4708, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.4e-06, | |
| "loss": 5.633, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.6e-06, | |
| "loss": 5.5787, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.8e-06, | |
| "loss": 5.6436, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 5.7185, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.200000000000001e-06, | |
| "loss": 5.9841, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 5.6354, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.599999999999999e-06, | |
| "loss": 6.0658, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.8e-06, | |
| "loss": 5.973, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9e-06, | |
| "loss": 5.2202, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.2e-06, | |
| "loss": 5.8316, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.4e-06, | |
| "loss": 5.4781, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 5.5465, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 5.4024, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1e-05, | |
| "loss": 5.7516, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.02e-05, | |
| "loss": 6.1596, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.04e-05, | |
| "loss": 5.6841, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.06e-05, | |
| "loss": 5.4212, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.08e-05, | |
| "loss": 5.8461, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 5.3743, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 6.0721, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 5.6247, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 5.8512, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.16e-05, | |
| "loss": 5.4426, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.18e-05, | |
| "loss": 5.5538, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2e-05, | |
| "loss": 5.3113, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.22e-05, | |
| "loss": 5.4602, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.24e-05, | |
| "loss": 5.5906, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2600000000000001e-05, | |
| "loss": 5.4354, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 5.8512, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 5.2339, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.32e-05, | |
| "loss": 5.4244, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3400000000000002e-05, | |
| "loss": 5.4329, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 5.5132, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 5.6404, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 5.3281, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.42e-05, | |
| "loss": 5.0157, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.44e-05, | |
| "loss": 5.5544, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.4599999999999999e-05, | |
| "loss": 5.1472, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.48e-05, | |
| "loss": 5.3325, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.5e-05, | |
| "loss": 5.5339, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.52e-05, | |
| "loss": 5.2837, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.54e-05, | |
| "loss": 4.9633, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.56e-05, | |
| "loss": 5.4258, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.58e-05, | |
| "loss": 5.5178, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 5.1757, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.62e-05, | |
| "loss": 5.4928, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.6400000000000002e-05, | |
| "loss": 4.9475, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.66e-05, | |
| "loss": 5.3968, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 5.1087, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 5.1547, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.7199999999999998e-05, | |
| "loss": 5.3679, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.74e-05, | |
| "loss": 5.0302, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.76e-05, | |
| "loss": 5.224, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.78e-05, | |
| "loss": 5.247, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8e-05, | |
| "loss": 4.9282, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8200000000000002e-05, | |
| "loss": 4.8391, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.84e-05, | |
| "loss": 5.44, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.86e-05, | |
| "loss": 4.7612, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.88e-05, | |
| "loss": 5.0552, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9e-05, | |
| "loss": 5.2444, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 5.0088, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.94e-05, | |
| "loss": 5.1638, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 5.1415, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9800000000000004e-05, | |
| "loss": 4.7029, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2e-05, | |
| "loss": 5.1647, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 4.6924, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.04e-05, | |
| "loss": 4.5594, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.06e-05, | |
| "loss": 4.8117, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.08e-05, | |
| "loss": 5.0678, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.1e-05, | |
| "loss": 4.7595, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.12e-05, | |
| "loss": 4.5639, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.1400000000000002e-05, | |
| "loss": 4.5801, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.16e-05, | |
| "loss": 4.888, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.18e-05, | |
| "loss": 4.9379, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 4.9305, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.22e-05, | |
| "loss": 4.6677, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 5.0038, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.26e-05, | |
| "loss": 4.7081, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 4.7798, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 4.6056, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.32e-05, | |
| "loss": 4.8402, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 4.6817, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.36e-05, | |
| "loss": 4.6605, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.38e-05, | |
| "loss": 4.4858, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.4e-05, | |
| "loss": 4.4792, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.4200000000000002e-05, | |
| "loss": 4.8146, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.44e-05, | |
| "loss": 4.8761, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.46e-05, | |
| "loss": 4.7947, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.48e-05, | |
| "loss": 4.5568, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.5e-05, | |
| "loss": 4.2196, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.5200000000000003e-05, | |
| "loss": 4.6131, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.54e-05, | |
| "loss": 4.4819, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.5600000000000002e-05, | |
| "loss": 4.547, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.58e-05, | |
| "loss": 4.6455, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 4.283, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6200000000000003e-05, | |
| "loss": 4.4657, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.64e-05, | |
| "loss": 4.3557, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6600000000000003e-05, | |
| "loss": 4.3666, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6800000000000004e-05, | |
| "loss": 4.3814, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 4.4348, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.7200000000000004e-05, | |
| "loss": 4.2329, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.7400000000000002e-05, | |
| "loss": 4.0716, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.7600000000000003e-05, | |
| "loss": 4.2426, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.7800000000000005e-05, | |
| "loss": 4.3954, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 4.5135, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 4.3307, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.84e-05, | |
| "loss": 4.3674, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.86e-05, | |
| "loss": 4.3752, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.88e-05, | |
| "loss": 4.4294, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9e-05, | |
| "loss": 4.1636, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9199999999999998e-05, | |
| "loss": 4.3358, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.94e-05, | |
| "loss": 4.153, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.96e-05, | |
| "loss": 4.4384, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.98e-05, | |
| "loss": 4.1657, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-05, | |
| "loss": 4.1399, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.02e-05, | |
| "loss": 3.941, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.04e-05, | |
| "loss": 4.1506, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.06e-05, | |
| "loss": 3.7582, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.08e-05, | |
| "loss": 3.93, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.1e-05, | |
| "loss": 4.142, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.12e-05, | |
| "loss": 3.9468, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.1400000000000004e-05, | |
| "loss": 3.9881, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.16e-05, | |
| "loss": 4.0424, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.18e-05, | |
| "loss": 3.8937, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 4.1833, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.2200000000000003e-05, | |
| "loss": 3.8278, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.24e-05, | |
| "loss": 3.8427, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.26e-05, | |
| "loss": 3.9691, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.2800000000000004e-05, | |
| "loss": 4.1022, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.3e-05, | |
| "loss": 3.7811, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.32e-05, | |
| "loss": 3.9233, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.3400000000000005e-05, | |
| "loss": 4.0059, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.3600000000000004e-05, | |
| "loss": 4.1543, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.38e-05, | |
| "loss": 3.8804, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 3.9156, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4200000000000005e-05, | |
| "loss": 3.5426, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4399999999999996e-05, | |
| "loss": 3.8931, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.46e-05, | |
| "loss": 3.8985, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.48e-05, | |
| "loss": 4.1012, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.5e-05, | |
| "loss": 3.7704, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.52e-05, | |
| "loss": 3.6555, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.54e-05, | |
| "loss": 3.7129, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.56e-05, | |
| "loss": 3.7271, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.58e-05, | |
| "loss": 3.7004, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-05, | |
| "loss": 3.5683, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.62e-05, | |
| "loss": 3.6048, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6400000000000004e-05, | |
| "loss": 3.6921, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.66e-05, | |
| "loss": 3.5727, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.68e-05, | |
| "loss": 3.6197, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7e-05, | |
| "loss": 3.4075, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.72e-05, | |
| "loss": 3.6994, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.74e-05, | |
| "loss": 3.6849, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.76e-05, | |
| "loss": 3.3789, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7800000000000004e-05, | |
| "loss": 3.693, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8e-05, | |
| "loss": 3.4607, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.82e-05, | |
| "loss": 3.4014, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8400000000000005e-05, | |
| "loss": 3.5438, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.86e-05, | |
| "loss": 3.5765, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.88e-05, | |
| "loss": 3.5509, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 3.3645, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9200000000000004e-05, | |
| "loss": 3.597, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.94e-05, | |
| "loss": 3.5312, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.960000000000001e-05, | |
| "loss": 3.4306, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.9800000000000005e-05, | |
| "loss": 3.4861, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 3.4123, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.02e-05, | |
| "loss": 3.5024, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 3.3418, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0600000000000004e-05, | |
| "loss": 3.4262, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.08e-05, | |
| "loss": 3.4753, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.1e-05, | |
| "loss": 3.3233, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.12e-05, | |
| "loss": 3.3934, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.14e-05, | |
| "loss": 3.2796, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.16e-05, | |
| "loss": 3.4233, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.18e-05, | |
| "loss": 3.1622, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2e-05, | |
| "loss": 3.3096, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.22e-05, | |
| "loss": 3.0306, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.24e-05, | |
| "loss": 3.3984, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.26e-05, | |
| "loss": 3.4985, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2800000000000004e-05, | |
| "loss": 3.7575, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3e-05, | |
| "loss": 3.4439, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.32e-05, | |
| "loss": 3.1789, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.3400000000000005e-05, | |
| "loss": 3.1178, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.36e-05, | |
| "loss": 3.1776, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.38e-05, | |
| "loss": 3.2264, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 3.2322, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4200000000000004e-05, | |
| "loss": 3.365, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.44e-05, | |
| "loss": 3.0469, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.46e-05, | |
| "loss": 3.0933, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4800000000000005e-05, | |
| "loss": 3.0693, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5e-05, | |
| "loss": 2.9901, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.52e-05, | |
| "loss": 3.2992, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5400000000000006e-05, | |
| "loss": 3.4731, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5600000000000004e-05, | |
| "loss": 3.263, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.58e-05, | |
| "loss": 3.633, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 3.2045, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6200000000000005e-05, | |
| "loss": 3.1856, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.64e-05, | |
| "loss": 3.2476, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.660000000000001e-05, | |
| "loss": 3.1311, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6800000000000006e-05, | |
| "loss": 3.1334, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.7e-05, | |
| "loss": 3.1872, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.72e-05, | |
| "loss": 3.075, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.74e-05, | |
| "loss": 3.0075, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.76e-05, | |
| "loss": 3.2703, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.78e-05, | |
| "loss": 3.1161, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8e-05, | |
| "loss": 3.4814, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.82e-05, | |
| "loss": 3.0307, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8400000000000004e-05, | |
| "loss": 2.9517, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.86e-05, | |
| "loss": 3.189, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.88e-05, | |
| "loss": 3.1276, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9e-05, | |
| "loss": 3.0072, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92e-05, | |
| "loss": 2.8744, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.94e-05, | |
| "loss": 3.2459, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.96e-05, | |
| "loss": 3.1944, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9800000000000004e-05, | |
| "loss": 3.0613, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 2.7814, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.02e-05, | |
| "loss": 3.1805, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.0400000000000005e-05, | |
| "loss": 3.1574, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.0600000000000003e-05, | |
| "loss": 3.027, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.08e-05, | |
| "loss": 3.223, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.1000000000000006e-05, | |
| "loss": 2.9815, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.1200000000000004e-05, | |
| "loss": 3.0636, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.14e-05, | |
| "loss": 2.9695, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.16e-05, | |
| "loss": 2.8215, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.1800000000000005e-05, | |
| "loss": 2.9468, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 3.0869, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.22e-05, | |
| "loss": 2.9771, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.2400000000000007e-05, | |
| "loss": 2.9307, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.2600000000000005e-05, | |
| "loss": 2.9213, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.28e-05, | |
| "loss": 3.0386, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.300000000000001e-05, | |
| "loss": 2.9559, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.3200000000000006e-05, | |
| "loss": 2.9783, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.3400000000000004e-05, | |
| "loss": 3.0086, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.360000000000001e-05, | |
| "loss": 3.0632, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.380000000000001e-05, | |
| "loss": 2.8526, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 3.1185, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.420000000000001e-05, | |
| "loss": 2.9256, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.440000000000001e-05, | |
| "loss": 2.9979, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4600000000000006e-05, | |
| "loss": 2.954, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4800000000000004e-05, | |
| "loss": 3.3889, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 2.8462, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.520000000000001e-05, | |
| "loss": 2.7856, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.5400000000000005e-05, | |
| "loss": 2.8884, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.560000000000001e-05, | |
| "loss": 2.8714, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.580000000000001e-05, | |
| "loss": 2.9504, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 2.7162, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.620000000000001e-05, | |
| "loss": 2.9031, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.6399999999999995e-05, | |
| "loss": 2.8813, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.66e-05, | |
| "loss": 2.8119, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.68e-05, | |
| "loss": 2.8679, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.6999999999999996e-05, | |
| "loss": 3.0378, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.72e-05, | |
| "loss": 2.9161, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.74e-05, | |
| "loss": 2.7277, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.76e-05, | |
| "loss": 2.7427, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.7799999999999995e-05, | |
| "loss": 2.8834, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.8e-05, | |
| "loss": 2.6262, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.82e-05, | |
| "loss": 3.1467, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.8399999999999997e-05, | |
| "loss": 2.9763, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.86e-05, | |
| "loss": 3.3912, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.88e-05, | |
| "loss": 3.1246, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.9e-05, | |
| "loss": 2.8967, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.92e-05, | |
| "loss": 2.7289, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.94e-05, | |
| "loss": 3.072, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.96e-05, | |
| "loss": 3.2343, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.9800000000000003e-05, | |
| "loss": 2.6965, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-05, | |
| "loss": 2.8023, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.02e-05, | |
| "loss": 2.9172, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.04e-05, | |
| "loss": 2.6316, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.06e-05, | |
| "loss": 2.7845, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.08e-05, | |
| "loss": 2.8575, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.1e-05, | |
| "loss": 2.851, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.12e-05, | |
| "loss": 2.7334, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.14e-05, | |
| "loss": 2.885, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.16e-05, | |
| "loss": 3.0768, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.18e-05, | |
| "loss": 3.1481, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.2e-05, | |
| "loss": 2.8301, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.220000000000001e-05, | |
| "loss": 2.55, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.24e-05, | |
| "loss": 2.5548, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.26e-05, | |
| "loss": 2.8179, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.280000000000001e-05, | |
| "loss": 2.8277, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.3e-05, | |
| "loss": 2.8063, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.32e-05, | |
| "loss": 2.7492, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.340000000000001e-05, | |
| "loss": 2.7972, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.36e-05, | |
| "loss": 2.8897, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.38e-05, | |
| "loss": 2.6464, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 2.3929, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.42e-05, | |
| "loss": 2.8093, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.440000000000001e-05, | |
| "loss": 2.8136, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.460000000000001e-05, | |
| "loss": 2.7268, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.48e-05, | |
| "loss": 2.7168, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 2.6716, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.52e-05, | |
| "loss": 2.5757, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.54e-05, | |
| "loss": 2.9716, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.560000000000001e-05, | |
| "loss": 2.567, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.58e-05, | |
| "loss": 2.8636, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.6e-05, | |
| "loss": 2.7675, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.620000000000001e-05, | |
| "loss": 2.7081, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.64e-05, | |
| "loss": 2.5302, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.66e-05, | |
| "loss": 2.7448, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.680000000000001e-05, | |
| "loss": 2.6066, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.7e-05, | |
| "loss": 2.9559, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.720000000000001e-05, | |
| "loss": 2.5044, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.740000000000001e-05, | |
| "loss": 2.4907, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.76e-05, | |
| "loss": 3.028, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.780000000000001e-05, | |
| "loss": 2.7468, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 2.9923, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.82e-05, | |
| "loss": 2.9032, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.840000000000001e-05, | |
| "loss": 2.9519, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.860000000000001e-05, | |
| "loss": 2.7843, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.879999999999999e-05, | |
| "loss": 2.8611, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.9e-05, | |
| "loss": 2.6476, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.92e-05, | |
| "loss": 2.5413, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.939999999999999e-05, | |
| "loss": 2.8123, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.96e-05, | |
| "loss": 2.972, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.98e-05, | |
| "loss": 2.5816, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7e-05, | |
| "loss": 3.123, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.02e-05, | |
| "loss": 2.6551, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.04e-05, | |
| "loss": 2.8189, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.06e-05, | |
| "loss": 2.6323, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.08e-05, | |
| "loss": 2.6618, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.1e-05, | |
| "loss": 2.5628, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.12e-05, | |
| "loss": 2.9508, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.14e-05, | |
| "loss": 2.5694, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.16e-05, | |
| "loss": 2.8969, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.18e-05, | |
| "loss": 2.6492, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.2e-05, | |
| "loss": 2.5428, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.22e-05, | |
| "loss": 2.6488, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.24e-05, | |
| "loss": 2.6491, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.26e-05, | |
| "loss": 2.4658, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.280000000000001e-05, | |
| "loss": 2.6678, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.3e-05, | |
| "loss": 2.7834, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.32e-05, | |
| "loss": 2.6306, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.340000000000001e-05, | |
| "loss": 2.6061, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.36e-05, | |
| "loss": 2.7604, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.38e-05, | |
| "loss": 2.6868, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.4e-05, | |
| "loss": 2.6802, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.42e-05, | |
| "loss": 2.4104, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.44e-05, | |
| "loss": 2.5347, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.46e-05, | |
| "loss": 2.7143, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.48e-05, | |
| "loss": 2.6505, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 2.4653, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.52e-05, | |
| "loss": 2.4457, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.54e-05, | |
| "loss": 2.7973, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.560000000000001e-05, | |
| "loss": 2.6539, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.58e-05, | |
| "loss": 2.668, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.6e-05, | |
| "loss": 2.7305, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.620000000000001e-05, | |
| "loss": 2.7173, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.64e-05, | |
| "loss": 2.519, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.66e-05, | |
| "loss": 2.2994, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.680000000000001e-05, | |
| "loss": 2.8727, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.7e-05, | |
| "loss": 2.5736, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.72e-05, | |
| "loss": 2.6561, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.740000000000001e-05, | |
| "loss": 2.6547, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.76e-05, | |
| "loss": 2.5341, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.780000000000001e-05, | |
| "loss": 2.2884, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 2.2819, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.82e-05, | |
| "loss": 2.6315, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.840000000000001e-05, | |
| "loss": 2.6647, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.860000000000001e-05, | |
| "loss": 2.4765, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.88e-05, | |
| "loss": 2.6793, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.900000000000001e-05, | |
| "loss": 2.4468, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.920000000000001e-05, | |
| "loss": 2.7349, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.94e-05, | |
| "loss": 2.6005, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.960000000000001e-05, | |
| "loss": 2.5669, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.98e-05, | |
| "loss": 2.6893, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-05, | |
| "loss": 2.6259, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.020000000000001e-05, | |
| "loss": 2.4774, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.04e-05, | |
| "loss": 2.4902, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.060000000000001e-05, | |
| "loss": 2.7192, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 2.4746, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.1e-05, | |
| "loss": 2.7094, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.120000000000001e-05, | |
| "loss": 2.6133, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.14e-05, | |
| "loss": 2.4189, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.16e-05, | |
| "loss": 2.405, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.18e-05, | |
| "loss": 2.617, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.2e-05, | |
| "loss": 2.635, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.22e-05, | |
| "loss": 2.7468, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.24e-05, | |
| "loss": 2.7308, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.26e-05, | |
| "loss": 2.685, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.28e-05, | |
| "loss": 2.3625, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.3e-05, | |
| "loss": 2.7173, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.32e-05, | |
| "loss": 2.5301, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.34e-05, | |
| "loss": 2.8232, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.36e-05, | |
| "loss": 2.4187, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.38e-05, | |
| "loss": 2.6008, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.4e-05, | |
| "loss": 2.4427, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.42e-05, | |
| "loss": 2.6173, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.44e-05, | |
| "loss": 2.4068, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.46e-05, | |
| "loss": 2.4622, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.48e-05, | |
| "loss": 2.3317, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.5e-05, | |
| "loss": 2.7911, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.52e-05, | |
| "loss": 2.8775, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.54e-05, | |
| "loss": 2.751, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.560000000000001e-05, | |
| "loss": 2.36, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.58e-05, | |
| "loss": 2.4175, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.6e-05, | |
| "loss": 2.5071, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.620000000000001e-05, | |
| "loss": 2.4226, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.64e-05, | |
| "loss": 2.5215, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.66e-05, | |
| "loss": 2.4904, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.680000000000001e-05, | |
| "loss": 2.5739, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.7e-05, | |
| "loss": 2.4446, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.72e-05, | |
| "loss": 2.6199, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.740000000000001e-05, | |
| "loss": 2.6532, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.76e-05, | |
| "loss": 2.6699, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.78e-05, | |
| "loss": 2.5341, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 2.4487, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.82e-05, | |
| "loss": 2.7382, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.840000000000001e-05, | |
| "loss": 2.4215, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.86e-05, | |
| "loss": 2.4077, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.88e-05, | |
| "loss": 2.5928, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.900000000000001e-05, | |
| "loss": 2.456, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.92e-05, | |
| "loss": 2.6947, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.94e-05, | |
| "loss": 2.4277, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.960000000000001e-05, | |
| "loss": 2.4797, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.98e-05, | |
| "loss": 2.4101, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-05, | |
| "loss": 2.3988, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.020000000000001e-05, | |
| "loss": 2.6711, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.04e-05, | |
| "loss": 2.533, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.06e-05, | |
| "loss": 2.6013, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.080000000000001e-05, | |
| "loss": 2.4816, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.1e-05, | |
| "loss": 2.5434, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.120000000000001e-05, | |
| "loss": 2.5063, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.140000000000001e-05, | |
| "loss": 2.3922, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.16e-05, | |
| "loss": 2.2701, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.180000000000001e-05, | |
| "loss": 2.7835, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 2.4626, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.22e-05, | |
| "loss": 2.4143, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.240000000000001e-05, | |
| "loss": 2.5903, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.260000000000001e-05, | |
| "loss": 2.7397, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.28e-05, | |
| "loss": 2.5644, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.300000000000001e-05, | |
| "loss": 2.7328, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.320000000000002e-05, | |
| "loss": 2.2869, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.340000000000001e-05, | |
| "loss": 2.5651, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.360000000000001e-05, | |
| "loss": 2.4674, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.38e-05, | |
| "loss": 2.6795, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.4e-05, | |
| "loss": 2.3188, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.42e-05, | |
| "loss": 2.4679, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.44e-05, | |
| "loss": 2.5542, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.46e-05, | |
| "loss": 2.3364, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.48e-05, | |
| "loss": 2.1973, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.5e-05, | |
| "loss": 2.3029, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.52e-05, | |
| "loss": 2.6317, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.54e-05, | |
| "loss": 2.414, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.56e-05, | |
| "loss": 2.4159, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.58e-05, | |
| "loss": 2.4459, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.6e-05, | |
| "loss": 2.559, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.620000000000001e-05, | |
| "loss": 2.443, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.64e-05, | |
| "loss": 2.4922, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.66e-05, | |
| "loss": 2.3364, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.680000000000001e-05, | |
| "loss": 2.5318, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.7e-05, | |
| "loss": 3.1505, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.72e-05, | |
| "loss": 2.5743, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.74e-05, | |
| "loss": 2.1854, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.76e-05, | |
| "loss": 2.5058, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.78e-05, | |
| "loss": 2.4168, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.8e-05, | |
| "loss": 2.3626, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.82e-05, | |
| "loss": 2.2568, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.84e-05, | |
| "loss": 2.4151, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.86e-05, | |
| "loss": 2.3183, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.88e-05, | |
| "loss": 2.4876, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.900000000000001e-05, | |
| "loss": 2.5616, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.92e-05, | |
| "loss": 2.5596, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.94e-05, | |
| "loss": 2.4484, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.960000000000001e-05, | |
| "loss": 2.3472, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.98e-05, | |
| "loss": 2.4291, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001, | |
| "loss": 2.4503, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010020000000000001, | |
| "loss": 2.5905, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001004, | |
| "loss": 2.4448, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001006, | |
| "loss": 2.4041, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010080000000000001, | |
| "loss": 2.6102, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000101, | |
| "loss": 2.4584, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010120000000000001, | |
| "loss": 2.3186, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010140000000000001, | |
| "loss": 2.3123, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001016, | |
| "loss": 2.3228, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010180000000000001, | |
| "loss": 2.2938, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010200000000000001, | |
| "loss": 2.2607, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001022, | |
| "loss": 2.4547, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010240000000000001, | |
| "loss": 2.5077, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010260000000000001, | |
| "loss": 2.445, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001028, | |
| "loss": 2.3815, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010300000000000001, | |
| "loss": 2.3946, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001032, | |
| "loss": 2.3712, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001034, | |
| "loss": 2.3742, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010360000000000001, | |
| "loss": 2.1646, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001038, | |
| "loss": 2.2206, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 2.4053, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010420000000000001, | |
| "loss": 2.5907, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001044, | |
| "loss": 2.8482, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010460000000000001, | |
| "loss": 2.4466, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010480000000000001, | |
| "loss": 2.5651, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000105, | |
| "loss": 2.2722, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010520000000000001, | |
| "loss": 2.5196, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010540000000000001, | |
| "loss": 2.2779, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001056, | |
| "loss": 2.2845, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010580000000000001, | |
| "loss": 2.247, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010600000000000002, | |
| "loss": 2.7478, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001062, | |
| "loss": 2.277, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010640000000000001, | |
| "loss": 2.3714, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010660000000000002, | |
| "loss": 2.4906, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010680000000000001, | |
| "loss": 2.1972, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010700000000000001, | |
| "loss": 2.4388, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010720000000000002, | |
| "loss": 2.326, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010740000000000001, | |
| "loss": 2.4306, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010760000000000001, | |
| "loss": 2.3829, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010780000000000002, | |
| "loss": 2.2516, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010800000000000001, | |
| "loss": 2.5561, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010820000000000001, | |
| "loss": 2.2923, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010840000000000002, | |
| "loss": 2.4563, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010860000000000001, | |
| "loss": 2.2419, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010880000000000002, | |
| "loss": 2.2345, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000109, | |
| "loss": 2.3465, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010920000000000001, | |
| "loss": 2.4493, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010940000000000002, | |
| "loss": 2.2118, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010960000000000001, | |
| "loss": 2.2814, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010980000000000001, | |
| "loss": 2.2585, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011000000000000002, | |
| "loss": 2.6481, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011020000000000001, | |
| "loss": 2.3979, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011040000000000001, | |
| "loss": 2.4089, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011060000000000002, | |
| "loss": 2.2194, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011080000000000001, | |
| "loss": 2.4714, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011100000000000001, | |
| "loss": 2.3269, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011120000000000002, | |
| "loss": 2.4552, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011140000000000001, | |
| "loss": 2.4263, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011160000000000002, | |
| "loss": 2.7013, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011180000000000002, | |
| "loss": 2.3745, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 2.1154, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011220000000000002, | |
| "loss": 2.3919, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011240000000000002, | |
| "loss": 2.3702, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001126, | |
| "loss": 2.2918, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011279999999999999, | |
| "loss": 2.6281, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000113, | |
| "loss": 2.4059, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001132, | |
| "loss": 2.4491, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011339999999999999, | |
| "loss": 2.3741, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001136, | |
| "loss": 2.2059, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001138, | |
| "loss": 2.1418, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011399999999999999, | |
| "loss": 2.2418, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001142, | |
| "loss": 2.195, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001144, | |
| "loss": 2.2985, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001146, | |
| "loss": 2.3015, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001148, | |
| "loss": 2.3432, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011499999999999999, | |
| "loss": 2.3157, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001152, | |
| "loss": 2.1907, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001154, | |
| "loss": 2.485, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011559999999999999, | |
| "loss": 2.3603, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001158, | |
| "loss": 2.1693, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000116, | |
| "loss": 2.2629, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011619999999999999, | |
| "loss": 2.3162, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001164, | |
| "loss": 2.2884, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001166, | |
| "loss": 2.1588, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011679999999999999, | |
| "loss": 2.242, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000117, | |
| "loss": 2.706, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001172, | |
| "loss": 2.1989, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001174, | |
| "loss": 2.1696, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001176, | |
| "loss": 2.1674, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001178, | |
| "loss": 2.227, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000118, | |
| "loss": 2.2796, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001182, | |
| "loss": 2.4987, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001184, | |
| "loss": 2.0346, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001186, | |
| "loss": 2.1769, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001188, | |
| "loss": 2.0115, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000119, | |
| "loss": 2.3685, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001192, | |
| "loss": 2.3915, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001194, | |
| "loss": 2.4385, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011960000000000001, | |
| "loss": 2.2051, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001198, | |
| "loss": 2.4439, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012, | |
| "loss": 2.2598, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012020000000000001, | |
| "loss": 2.2669, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001204, | |
| "loss": 2.2098, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001206, | |
| "loss": 2.4675, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001208, | |
| "loss": 2.6596, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000121, | |
| "loss": 2.2171, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001212, | |
| "loss": 2.9337, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001214, | |
| "loss": 2.523, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001216, | |
| "loss": 2.5411, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001218, | |
| "loss": 2.1373, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000122, | |
| "loss": 2.1198, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012220000000000002, | |
| "loss": 2.3266, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001224, | |
| "loss": 2.2926, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001226, | |
| "loss": 2.3988, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001228, | |
| "loss": 2.1883, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000123, | |
| "loss": 2.5811, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001232, | |
| "loss": 2.2914, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012340000000000002, | |
| "loss": 2.2611, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001236, | |
| "loss": 2.448, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001238, | |
| "loss": 2.3978, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000124, | |
| "loss": 2.4208, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001242, | |
| "loss": 2.3129, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012440000000000002, | |
| "loss": 2.2814, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001246, | |
| "loss": 2.2118, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001248, | |
| "loss": 2.1229, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000125, | |
| "loss": 2.2069, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001252, | |
| "loss": 2.2165, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001254, | |
| "loss": 2.6609, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012560000000000002, | |
| "loss": 2.6453, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001258, | |
| "loss": 2.2979, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000126, | |
| "loss": 2.6011, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001262, | |
| "loss": 2.5195, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001264, | |
| "loss": 2.4403, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012660000000000001, | |
| "loss": 2.052, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012680000000000002, | |
| "loss": 2.2156, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000127, | |
| "loss": 2.6445, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001272, | |
| "loss": 2.2729, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001274, | |
| "loss": 2.0318, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001276, | |
| "loss": 2.1489, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012780000000000002, | |
| "loss": 2.3364, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 2.2808, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001282, | |
| "loss": 2.52, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001284, | |
| "loss": 2.2828, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001286, | |
| "loss": 2.403, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012880000000000001, | |
| "loss": 2.1494, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012900000000000002, | |
| "loss": 2.3292, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012920000000000002, | |
| "loss": 2.1396, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001294, | |
| "loss": 2.2465, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001296, | |
| "loss": 2.475, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001298, | |
| "loss": 2.578, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013000000000000002, | |
| "loss": 2.4337, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013020000000000002, | |
| "loss": 2.3048, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001304, | |
| "loss": 2.1943, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001306, | |
| "loss": 2.2984, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001308, | |
| "loss": 2.388, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000131, | |
| "loss": 2.2412, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013120000000000002, | |
| "loss": 2.4911, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013140000000000002, | |
| "loss": 2.2917, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001316, | |
| "loss": 2.314, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001318, | |
| "loss": 2.1705, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000132, | |
| "loss": 2.6633, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013220000000000001, | |
| "loss": 2.3836, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013240000000000002, | |
| "loss": 2.2542, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013260000000000002, | |
| "loss": 2.239, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001328, | |
| "loss": 2.1645, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000133, | |
| "loss": 2.2798, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001332, | |
| "loss": 2.4113, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013340000000000002, | |
| "loss": 2.3506, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013360000000000002, | |
| "loss": 2.3298, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013380000000000003, | |
| "loss": 2.3692, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000134, | |
| "loss": 2.1512, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001342, | |
| "loss": 2.4143, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013440000000000001, | |
| "loss": 2.4147, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013460000000000002, | |
| "loss": 2.3097, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013480000000000002, | |
| "loss": 2.0768, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013500000000000003, | |
| "loss": 2.4597, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001352, | |
| "loss": 2.4789, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001354, | |
| "loss": 2.273, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013560000000000002, | |
| "loss": 2.1588, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013580000000000002, | |
| "loss": 2.1891, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 2.1735, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001362, | |
| "loss": 2.3076, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001364, | |
| "loss": 2.3186, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001366, | |
| "loss": 2.4273, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013680000000000002, | |
| "loss": 2.1678, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013700000000000002, | |
| "loss": 2.3184, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013720000000000003, | |
| "loss": 2.152, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001374, | |
| "loss": 2.2027, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013759999999999998, | |
| "loss": 2.0117, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001378, | |
| "loss": 2.3967, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000138, | |
| "loss": 2.2453, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001382, | |
| "loss": 2.1737, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001384, | |
| "loss": 2.1285, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001386, | |
| "loss": 2.4048, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013879999999999999, | |
| "loss": 2.0147, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000139, | |
| "loss": 2.1226, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001392, | |
| "loss": 2.186, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001394, | |
| "loss": 2.2582, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001396, | |
| "loss": 2.4653, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001398, | |
| "loss": 2.2737, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014, | |
| "loss": 2.068, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001402, | |
| "loss": 2.0627, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001404, | |
| "loss": 2.2417, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001406, | |
| "loss": 2.346, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001408, | |
| "loss": 2.2944, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000141, | |
| "loss": 2.3958, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001412, | |
| "loss": 2.3616, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001414, | |
| "loss": 2.4885, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001416, | |
| "loss": 2.3986, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001418, | |
| "loss": 2.3317, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000142, | |
| "loss": 2.1809, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001422, | |
| "loss": 2.3971, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001424, | |
| "loss": 2.2023, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001426, | |
| "loss": 2.1221, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001428, | |
| "loss": 2.1867, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000143, | |
| "loss": 2.654, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001432, | |
| "loss": 2.3007, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001434, | |
| "loss": 2.4027, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001436, | |
| "loss": 2.2819, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001438, | |
| "loss": 2.5396, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000144, | |
| "loss": 2.4257, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001442, | |
| "loss": 2.1992, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001444, | |
| "loss": 2.3616, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001446, | |
| "loss": 2.2921, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001448, | |
| "loss": 2.1809, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000145, | |
| "loss": 2.4629, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001452, | |
| "loss": 2.1798, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001454, | |
| "loss": 2.2488, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014560000000000002, | |
| "loss": 2.2077, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001458, | |
| "loss": 2.2095, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000146, | |
| "loss": 2.138, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001462, | |
| "loss": 2.4138, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001464, | |
| "loss": 2.1459, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001466, | |
| "loss": 2.1996, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014680000000000002, | |
| "loss": 2.281, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000147, | |
| "loss": 2.1391, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001472, | |
| "loss": 2.2228, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001474, | |
| "loss": 2.3614, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001476, | |
| "loss": 2.3966, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014780000000000001, | |
| "loss": 2.2546, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000148, | |
| "loss": 2.2699, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001482, | |
| "loss": 2.0677, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001484, | |
| "loss": 2.1875, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001486, | |
| "loss": 2.3285, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001488, | |
| "loss": 2.3256, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00014900000000000002, | |
| "loss": 2.2963, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001492, | |
| "loss": 2.4233, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001494, | |
| "loss": 1.956, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001496, | |
| "loss": 2.1669, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001498, | |
| "loss": 2.1629, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 2.2692, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015020000000000002, | |
| "loss": 2.2936, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001504, | |
| "loss": 2.1919, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001506, | |
| "loss": 2.2017, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001508, | |
| "loss": 2.3305, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000151, | |
| "loss": 1.9566, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015120000000000002, | |
| "loss": 2.3123, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015140000000000002, | |
| "loss": 2.3513, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001516, | |
| "loss": 2.8157, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001518, | |
| "loss": 2.1683, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000152, | |
| "loss": 2.297, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001522, | |
| "loss": 2.3101, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015240000000000002, | |
| "loss": 2.4763, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015260000000000002, | |
| "loss": 2.1414, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001528, | |
| "loss": 2.3431, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000153, | |
| "loss": 2.2393, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001532, | |
| "loss": 2.3469, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015340000000000002, | |
| "loss": 2.1865, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015360000000000002, | |
| "loss": 2.1675, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001538, | |
| "loss": 2.1985, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000154, | |
| "loss": 2.6843, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001542, | |
| "loss": 2.0214, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001544, | |
| "loss": 2.1681, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015460000000000002, | |
| "loss": 2.1967, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015480000000000002, | |
| "loss": 2.3442, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000155, | |
| "loss": 2.3955, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001552, | |
| "loss": 2.468, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001554, | |
| "loss": 2.2282, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015560000000000001, | |
| "loss": 2.356, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015580000000000002, | |
| "loss": 2.0865, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015600000000000002, | |
| "loss": 2.2295, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001562, | |
| "loss": 2.4866, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001564, | |
| "loss": 1.9964, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001566, | |
| "loss": 2.3318, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015680000000000002, | |
| "loss": 2.2456, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015700000000000002, | |
| "loss": 2.1341, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015720000000000003, | |
| "loss": 2.0312, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001574, | |
| "loss": 2.0445, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001576, | |
| "loss": 2.2026, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015780000000000001, | |
| "loss": 2.3863, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015800000000000002, | |
| "loss": 2.1653, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015820000000000002, | |
| "loss": 2.2098, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015840000000000003, | |
| "loss": 2.1443, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001586, | |
| "loss": 2.5036, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001588, | |
| "loss": 2.3064, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015900000000000002, | |
| "loss": 2.3497, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015920000000000002, | |
| "loss": 2.3091, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00015940000000000003, | |
| "loss": 2.1644, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001596, | |
| "loss": 2.2021, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001598, | |
| "loss": 2.2464, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016, | |
| "loss": 2.3061, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016020000000000002, | |
| "loss": 2.203, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016040000000000002, | |
| "loss": 2.2834, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016060000000000003, | |
| "loss": 2.0214, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001608, | |
| "loss": 2.1485, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000161, | |
| "loss": 2.3008, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016120000000000002, | |
| "loss": 2.2276, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016140000000000002, | |
| "loss": 2.051, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016160000000000002, | |
| "loss": 2.3019, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016180000000000003, | |
| "loss": 1.9958, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000162, | |
| "loss": 2.1309, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001622, | |
| "loss": 2.0722, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016240000000000002, | |
| "loss": 2.2913, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001626, | |
| "loss": 2.5393, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001628, | |
| "loss": 1.9663, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000163, | |
| "loss": 2.2849, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001632, | |
| "loss": 2.202, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001634, | |
| "loss": 2.1513, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001636, | |
| "loss": 2.2228, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001638, | |
| "loss": 2.4164, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000164, | |
| "loss": 2.3585, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001642, | |
| "loss": 2.0153, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001644, | |
| "loss": 2.3471, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001646, | |
| "loss": 2.4742, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001648, | |
| "loss": 2.3723, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000165, | |
| "loss": 2.2986, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001652, | |
| "loss": 2.3043, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001654, | |
| "loss": 2.0291, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001656, | |
| "loss": 2.36, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001658, | |
| "loss": 2.1242, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000166, | |
| "loss": 2.4035, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001662, | |
| "loss": 2.2995, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001664, | |
| "loss": 2.1073, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001666, | |
| "loss": 2.2019, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001668, | |
| "loss": 2.2287, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000167, | |
| "loss": 2.4243, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001672, | |
| "loss": 2.2078, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001674, | |
| "loss": 2.118, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001676, | |
| "loss": 2.1562, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001678, | |
| "loss": 2.0622, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000168, | |
| "loss": 2.156, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001682, | |
| "loss": 2.2996, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001684, | |
| "loss": 2.1311, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001686, | |
| "loss": 2.4076, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001688, | |
| "loss": 2.3208, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016900000000000002, | |
| "loss": 2.3604, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001692, | |
| "loss": 2.1498, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001694, | |
| "loss": 2.5732, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001696, | |
| "loss": 1.9379, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001698, | |
| "loss": 2.0502, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017, | |
| "loss": 2.3415, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017020000000000002, | |
| "loss": 2.2054, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001704, | |
| "loss": 2.2538, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001706, | |
| "loss": 2.2215, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001708, | |
| "loss": 2.2014, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000171, | |
| "loss": 2.1174, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017120000000000001, | |
| "loss": 2.186, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001714, | |
| "loss": 2.2905, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001716, | |
| "loss": 2.3523, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001718, | |
| "loss": 2.382, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000172, | |
| "loss": 2.0881, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001722, | |
| "loss": 2.35, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017240000000000002, | |
| "loss": 2.2716, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001726, | |
| "loss": 2.1627, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001728, | |
| "loss": 2.0719, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000173, | |
| "loss": 2.0881, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001732, | |
| "loss": 2.1929, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001734, | |
| "loss": 2.4172, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017360000000000002, | |
| "loss": 2.0672, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001738, | |
| "loss": 2.2345, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000174, | |
| "loss": 2.1886, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001742, | |
| "loss": 2.1409, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001744, | |
| "loss": 2.3219, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017460000000000002, | |
| "loss": 2.1279, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017480000000000002, | |
| "loss": 2.5594, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000175, | |
| "loss": 2.0652, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001752, | |
| "loss": 2.2338, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001754, | |
| "loss": 2.2535, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001756, | |
| "loss": 2.5585, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017580000000000002, | |
| "loss": 2.3933, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 2.107, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001762, | |
| "loss": 2.1014, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001764, | |
| "loss": 2.3926, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001766, | |
| "loss": 2.0854, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017680000000000001, | |
| "loss": 2.1851, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017700000000000002, | |
| "loss": 2.0922, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001772, | |
| "loss": 2.0552, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001774, | |
| "loss": 2.1875, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001776, | |
| "loss": 2.0934, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001778, | |
| "loss": 2.2528, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017800000000000002, | |
| "loss": 2.5132, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017820000000000002, | |
| "loss": 2.0217, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001784, | |
| "loss": 2.3063, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001786, | |
| "loss": 2.0653, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001788, | |
| "loss": 2.226, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017900000000000001, | |
| "loss": 2.0701, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017920000000000002, | |
| "loss": 2.2276, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017940000000000002, | |
| "loss": 2.089, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001796, | |
| "loss": 2.1327, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001798, | |
| "loss": 2.2473, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018, | |
| "loss": 2.1858, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018020000000000002, | |
| "loss": 2.2535, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018040000000000002, | |
| "loss": 2.3808, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018060000000000003, | |
| "loss": 2.5464, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001808, | |
| "loss": 2.2058, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000181, | |
| "loss": 2.341, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001812, | |
| "loss": 2.403, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018140000000000002, | |
| "loss": 1.969, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018160000000000002, | |
| "loss": 2.0528, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018180000000000003, | |
| "loss": 2.3008, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000182, | |
| "loss": 2.3855, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001822, | |
| "loss": 2.18, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018240000000000002, | |
| "loss": 2.1918, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018260000000000002, | |
| "loss": 2.2538, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018280000000000003, | |
| "loss": 2.0041, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000183, | |
| "loss": 1.9797, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001832, | |
| "loss": 2.1473, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001834, | |
| "loss": 2.1244, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018360000000000002, | |
| "loss": 2.3266, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018380000000000002, | |
| "loss": 2.1111, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 2.3406, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001842, | |
| "loss": 1.9725, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001844, | |
| "loss": 2.1774, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018460000000000001, | |
| "loss": 2.0264, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018480000000000002, | |
| "loss": 2.0621, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018500000000000002, | |
| "loss": 2.3217, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018520000000000003, | |
| "loss": 1.798, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001854, | |
| "loss": 2.2893, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001856, | |
| "loss": 2.0418, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018580000000000002, | |
| "loss": 2.0803, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018600000000000002, | |
| "loss": 2.3226, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018620000000000003, | |
| "loss": 2.1072, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018640000000000003, | |
| "loss": 2.2982, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001866, | |
| "loss": 2.3293, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018680000000000001, | |
| "loss": 2.2957, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018700000000000002, | |
| "loss": 2.2168, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018720000000000002, | |
| "loss": 2.147, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018740000000000003, | |
| "loss": 2.3286, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001876, | |
| "loss": 2.3112, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001878, | |
| "loss": 2.3437, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000188, | |
| "loss": 1.9025, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001882, | |
| "loss": 2.2031, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001884, | |
| "loss": 2.1442, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001886, | |
| "loss": 2.0561, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001888, | |
| "loss": 1.9866, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018899999999999999, | |
| "loss": 2.1752, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001892, | |
| "loss": 2.0284, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001894, | |
| "loss": 1.9475, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001896, | |
| "loss": 2.3361, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001898, | |
| "loss": 2.255, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019, | |
| "loss": 2.3764, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001902, | |
| "loss": 2.185, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001904, | |
| "loss": 2.0576, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001906, | |
| "loss": 2.0453, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001908, | |
| "loss": 2.5542, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000191, | |
| "loss": 2.0435, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001912, | |
| "loss": 2.0154, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001914, | |
| "loss": 2.416, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001916, | |
| "loss": 2.1007, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001918, | |
| "loss": 2.544, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000192, | |
| "loss": 1.9779, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001922, | |
| "loss": 2.1699, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019240000000000001, | |
| "loss": 1.9679, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001926, | |
| "loss": 2.3709, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001928, | |
| "loss": 2.1743, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000193, | |
| "loss": 2.1945, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001932, | |
| "loss": 2.0809, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001934, | |
| "loss": 2.0907, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019360000000000002, | |
| "loss": 2.0744, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001938, | |
| "loss": 2.2758, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000194, | |
| "loss": 2.1794, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001942, | |
| "loss": 2.1507, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001944, | |
| "loss": 2.0315, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019460000000000001, | |
| "loss": 2.3777, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001948, | |
| "loss": 2.3932, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000195, | |
| "loss": 2.2668, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001952, | |
| "loss": 2.1829, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001954, | |
| "loss": 2.0295, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001956, | |
| "loss": 2.1021, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019580000000000002, | |
| "loss": 1.9537, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000196, | |
| "loss": 2.2695, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001962, | |
| "loss": 2.3477, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001964, | |
| "loss": 2.049, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001966, | |
| "loss": 2.158, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001968, | |
| "loss": 2.4161, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019700000000000002, | |
| "loss": 2.3592, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001972, | |
| "loss": 2.0543, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001974, | |
| "loss": 2.25, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001976, | |
| "loss": 2.0532, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001978, | |
| "loss": 2.1755, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019800000000000002, | |
| "loss": 2.1953, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019820000000000002, | |
| "loss": 2.3071, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001984, | |
| "loss": 2.0682, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001986, | |
| "loss": 1.9946, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001988, | |
| "loss": 1.9321, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000199, | |
| "loss": 2.2531, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019920000000000002, | |
| "loss": 2.2978, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019940000000000002, | |
| "loss": 2.2777, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001996, | |
| "loss": 2.3151, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001998, | |
| "loss": 2.1298, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2499, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001998, | |
| "loss": 2.4858, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001996, | |
| "loss": 2.1771, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019940000000000002, | |
| "loss": 1.9904, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019920000000000002, | |
| "loss": 2.0758, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000199, | |
| "loss": 2.058, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001988, | |
| "loss": 2.1759, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001986, | |
| "loss": 1.9527, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001984, | |
| "loss": 2.0433, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019820000000000002, | |
| "loss": 1.846, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019800000000000002, | |
| "loss": 2.0655, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001978, | |
| "loss": 2.2689, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001976, | |
| "loss": 1.9181, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001974, | |
| "loss": 2.1069, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001972, | |
| "loss": 2.318, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019700000000000002, | |
| "loss": 2.4771, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001968, | |
| "loss": 2.312, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001966, | |
| "loss": 2.4544, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001964, | |
| "loss": 2.1533, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001962, | |
| "loss": 2.622, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000196, | |
| "loss": 1.9852, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019580000000000002, | |
| "loss": 2.2388, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001956, | |
| "loss": 2.3691, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001954, | |
| "loss": 2.3433, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001952, | |
| "loss": 2.2515, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000195, | |
| "loss": 2.1751, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001948, | |
| "loss": 1.8875, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019460000000000001, | |
| "loss": 2.1222, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001944, | |
| "loss": 1.881, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001942, | |
| "loss": 2.1074, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000194, | |
| "loss": 2.1288, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001938, | |
| "loss": 1.9946, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019360000000000002, | |
| "loss": 2.1166, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001934, | |
| "loss": 2.446, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001932, | |
| "loss": 2.0088, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000193, | |
| "loss": 2.4295, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001928, | |
| "loss": 2.2286, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001926, | |
| "loss": 2.1713, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019240000000000001, | |
| "loss": 2.3111, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001922, | |
| "loss": 2.3097, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000192, | |
| "loss": 2.1455, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001918, | |
| "loss": 2.0498, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001916, | |
| "loss": 2.0786, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001914, | |
| "loss": 2.3839, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001912, | |
| "loss": 1.9149, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000191, | |
| "loss": 2.2463, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001908, | |
| "loss": 2.319, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001906, | |
| "loss": 2.0475, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001904, | |
| "loss": 2.12, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001902, | |
| "loss": 2.076, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019, | |
| "loss": 2.6351, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001898, | |
| "loss": 1.9719, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001896, | |
| "loss": 2.0623, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001894, | |
| "loss": 2.1069, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001892, | |
| "loss": 1.9979, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018899999999999999, | |
| "loss": 2.1564, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001888, | |
| "loss": 2.3173, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001886, | |
| "loss": 2.1863, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001884, | |
| "loss": 2.242, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001882, | |
| "loss": 2.1032, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000188, | |
| "loss": 1.9624, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001878, | |
| "loss": 2.1442, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001876, | |
| "loss": 2.3459, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018740000000000003, | |
| "loss": 2.1352, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018720000000000002, | |
| "loss": 2.2038, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018700000000000002, | |
| "loss": 2.1763, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018680000000000001, | |
| "loss": 2.2278, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001866, | |
| "loss": 2.2931, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018640000000000003, | |
| "loss": 2.0459, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018620000000000003, | |
| "loss": 2.1381, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018600000000000002, | |
| "loss": 2.4695, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018580000000000002, | |
| "loss": 1.9151, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001856, | |
| "loss": 1.882, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001854, | |
| "loss": 2.0912, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018520000000000003, | |
| "loss": 2.3452, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018500000000000002, | |
| "loss": 2.2353, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018480000000000002, | |
| "loss": 2.2583, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018460000000000001, | |
| "loss": 2.2277, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001844, | |
| "loss": 2.1686, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001842, | |
| "loss": 2.0816, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 2.1236, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018380000000000002, | |
| "loss": 2.0317, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018360000000000002, | |
| "loss": 2.1954, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001834, | |
| "loss": 2.0982, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001832, | |
| "loss": 2.0293, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000183, | |
| "loss": 2.1831, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018280000000000003, | |
| "loss": 2.229, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018260000000000002, | |
| "loss": 2.0237, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018240000000000002, | |
| "loss": 2.1983, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001822, | |
| "loss": 2.3877, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000182, | |
| "loss": 2.2311, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018180000000000003, | |
| "loss": 2.4729, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018160000000000002, | |
| "loss": 2.0041, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018140000000000002, | |
| "loss": 1.9131, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001812, | |
| "loss": 2.0886, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000181, | |
| "loss": 1.9613, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001808, | |
| "loss": 1.917, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018060000000000003, | |
| "loss": 2.1915, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018040000000000002, | |
| "loss": 2.1527, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018020000000000002, | |
| "loss": 1.8069, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00018, | |
| "loss": 2.1524, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001798, | |
| "loss": 2.4085, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001796, | |
| "loss": 2.2248, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017940000000000002, | |
| "loss": 2.3198, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017920000000000002, | |
| "loss": 2.0714, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017900000000000001, | |
| "loss": 2.376, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001788, | |
| "loss": 2.0622, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001786, | |
| "loss": 2.17, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001784, | |
| "loss": 1.9192, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017820000000000002, | |
| "loss": 1.9581, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017800000000000002, | |
| "loss": 2.3742, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001778, | |
| "loss": 2.2591, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001776, | |
| "loss": 2.1213, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001774, | |
| "loss": 2.0813, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001772, | |
| "loss": 2.1793, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017700000000000002, | |
| "loss": 1.9993, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017680000000000001, | |
| "loss": 2.3638, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001766, | |
| "loss": 2.0043, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001764, | |
| "loss": 2.1704, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001762, | |
| "loss": 2.2363, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 2.0343, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017580000000000002, | |
| "loss": 1.9631, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001756, | |
| "loss": 2.2257, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001754, | |
| "loss": 2.1544, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001752, | |
| "loss": 2.0413, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000175, | |
| "loss": 2.4249, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017480000000000002, | |
| "loss": 2.3123, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017460000000000002, | |
| "loss": 1.8917, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001744, | |
| "loss": 2.2285, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001742, | |
| "loss": 2.1291, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000174, | |
| "loss": 2.0929, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001738, | |
| "loss": 2.2001, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017360000000000002, | |
| "loss": 2.05, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001734, | |
| "loss": 2.3887, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001732, | |
| "loss": 1.9932, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000173, | |
| "loss": 2.2071, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001728, | |
| "loss": 2.1447, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001726, | |
| "loss": 2.0132, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017240000000000002, | |
| "loss": 2.4171, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001722, | |
| "loss": 2.206, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000172, | |
| "loss": 2.187, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001718, | |
| "loss": 2.3886, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001716, | |
| "loss": 2.4409, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001714, | |
| "loss": 1.9017, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017120000000000001, | |
| "loss": 2.0486, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000171, | |
| "loss": 2.0213, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001708, | |
| "loss": 2.2113, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001706, | |
| "loss": 2.1489, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001704, | |
| "loss": 2.1718, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017020000000000002, | |
| "loss": 2.0698, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00017, | |
| "loss": 2.2001, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001698, | |
| "loss": 2.0783, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001696, | |
| "loss": 2.168, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001694, | |
| "loss": 2.2705, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001692, | |
| "loss": 2.2554, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016900000000000002, | |
| "loss": 1.9965, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001688, | |
| "loss": 2.1437, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001686, | |
| "loss": 2.2973, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001684, | |
| "loss": 2.2281, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001682, | |
| "loss": 2.1714, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000168, | |
| "loss": 1.9941, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001678, | |
| "loss": 2.1553, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001676, | |
| "loss": 2.2067, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001674, | |
| "loss": 2.0978, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001672, | |
| "loss": 1.9849, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000167, | |
| "loss": 2.241, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001668, | |
| "loss": 2.3276, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001666, | |
| "loss": 2.1016, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001664, | |
| "loss": 1.8563, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001662, | |
| "loss": 1.9359, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000166, | |
| "loss": 2.1702, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001658, | |
| "loss": 2.1627, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001656, | |
| "loss": 2.2544, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001654, | |
| "loss": 2.0513, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001652, | |
| "loss": 2.082, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000165, | |
| "loss": 2.0487, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001648, | |
| "loss": 2.3968, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001646, | |
| "loss": 2.1603, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001644, | |
| "loss": 2.356, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001642, | |
| "loss": 1.9072, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000164, | |
| "loss": 2.4396, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001638, | |
| "loss": 1.9514, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001636, | |
| "loss": 2.0527, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001634, | |
| "loss": 2.3599, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001632, | |
| "loss": 2.1134, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000163, | |
| "loss": 2.1626, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001628, | |
| "loss": 2.3118, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001626, | |
| "loss": 2.0078, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016240000000000002, | |
| "loss": 2.0798, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001622, | |
| "loss": 2.4655, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000162, | |
| "loss": 2.0856, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016180000000000003, | |
| "loss": 2.0388, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016160000000000002, | |
| "loss": 2.1923, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016140000000000002, | |
| "loss": 1.9453, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016120000000000002, | |
| "loss": 2.1677, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000161, | |
| "loss": 2.264, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001608, | |
| "loss": 2.2913, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016060000000000003, | |
| "loss": 1.9459, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016040000000000002, | |
| "loss": 2.468, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016020000000000002, | |
| "loss": 2.2854, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00016, | |
| "loss": 2.2404, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001598, | |
| "loss": 2.0607, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001596, | |
| "loss": 2.2329, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015940000000000003, | |
| "loss": 2.0696, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015920000000000002, | |
| "loss": 2.116, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015900000000000002, | |
| "loss": 2.2656, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001588, | |
| "loss": 2.2019, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001586, | |
| "loss": 1.936, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015840000000000003, | |
| "loss": 2.3412, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015820000000000002, | |
| "loss": 2.4583, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015800000000000002, | |
| "loss": 2.1762, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015780000000000001, | |
| "loss": 2.2242, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001576, | |
| "loss": 2.0731, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001574, | |
| "loss": 2.0755, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015720000000000003, | |
| "loss": 2.0838, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015700000000000002, | |
| "loss": 2.1014, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015680000000000002, | |
| "loss": 1.9181, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001566, | |
| "loss": 2.0875, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001564, | |
| "loss": 1.9404, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001562, | |
| "loss": 2.0328, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015600000000000002, | |
| "loss": 2.0109, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015580000000000002, | |
| "loss": 1.9651, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015560000000000001, | |
| "loss": 2.4363, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001554, | |
| "loss": 2.174, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001552, | |
| "loss": 2.313, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000155, | |
| "loss": 2.4208, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015480000000000002, | |
| "loss": 2.3771, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015460000000000002, | |
| "loss": 2.124, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001544, | |
| "loss": 2.1286, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001542, | |
| "loss": 2.2012, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000154, | |
| "loss": 2.102, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001538, | |
| "loss": 2.1007, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015360000000000002, | |
| "loss": 2.302, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015340000000000002, | |
| "loss": 2.3132, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001532, | |
| "loss": 2.0502, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000153, | |
| "loss": 2.2188, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001528, | |
| "loss": 1.9639, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015260000000000002, | |
| "loss": 2.1646, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015240000000000002, | |
| "loss": 2.3064, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001522, | |
| "loss": 2.0031, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000152, | |
| "loss": 2.0046, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001518, | |
| "loss": 2.289, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001516, | |
| "loss": 1.991, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015140000000000002, | |
| "loss": 1.94, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015120000000000002, | |
| "loss": 2.1762, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000151, | |
| "loss": 2.1072, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001508, | |
| "loss": 2.0465, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001506, | |
| "loss": 2.3404, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001504, | |
| "loss": 2.2174, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015020000000000002, | |
| "loss": 1.9053, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 2.2592, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001498, | |
| "loss": 2.1515, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001496, | |
| "loss": 2.1367, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001494, | |
| "loss": 2.2592, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001492, | |
| "loss": 2.1102, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00014900000000000002, | |
| "loss": 2.0638, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001488, | |
| "loss": 2.386, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001486, | |
| "loss": 2.1491, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001484, | |
| "loss": 2.4342, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001482, | |
| "loss": 2.3808, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000148, | |
| "loss": 2.151, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00014780000000000001, | |
| "loss": 2.111, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001476, | |
| "loss": 2.0861, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001474, | |
| "loss": 2.0844, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001472, | |
| "loss": 2.063, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000147, | |
| "loss": 2.0572, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00014680000000000002, | |
| "loss": 2.0127, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001466, | |
| "loss": 2.1046, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001464, | |
| "loss": 2.0856, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001462, | |
| "loss": 1.9167, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000146, | |
| "loss": 2.1402, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001458, | |
| "loss": 1.8923, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00014560000000000002, | |
| "loss": 2.2326, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001454, | |
| "loss": 2.2292, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001452, | |
| "loss": 2.2702, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000145, | |
| "loss": 2.1864, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001448, | |
| "loss": 2.0538, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001446, | |
| "loss": 2.1689, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001444, | |
| "loss": 2.1368, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001442, | |
| "loss": 2.1941, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000144, | |
| "loss": 2.311, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001438, | |
| "loss": 2.1882, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001436, | |
| "loss": 2.1384, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001434, | |
| "loss": 2.0812, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001432, | |
| "loss": 2.3276, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000143, | |
| "loss": 2.055, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001428, | |
| "loss": 1.8995, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001426, | |
| "loss": 2.2562, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001424, | |
| "loss": 1.9285, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001422, | |
| "loss": 2.4139, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000142, | |
| "loss": 1.9966, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001418, | |
| "loss": 1.8651, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001416, | |
| "loss": 2.3431, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001414, | |
| "loss": 2.4503, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001412, | |
| "loss": 2.2402, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000141, | |
| "loss": 2.2118, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001408, | |
| "loss": 2.0821, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001406, | |
| "loss": 2.0693, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001404, | |
| "loss": 2.2754, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001402, | |
| "loss": 1.9828, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00014, | |
| "loss": 2.1804, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001398, | |
| "loss": 2.0386, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001396, | |
| "loss": 2.1213, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001394, | |
| "loss": 1.9675, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001392, | |
| "loss": 2.2572, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000139, | |
| "loss": 1.9051, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013879999999999999, | |
| "loss": 2.1903, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001386, | |
| "loss": 2.2029, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001384, | |
| "loss": 2.0703, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001382, | |
| "loss": 2.1689, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000138, | |
| "loss": 2.1278, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001378, | |
| "loss": 1.773, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013759999999999998, | |
| "loss": 2.0371, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001374, | |
| "loss": 2.0164, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013720000000000003, | |
| "loss": 1.9395, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013700000000000002, | |
| "loss": 2.2098, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013680000000000002, | |
| "loss": 1.9273, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001366, | |
| "loss": 2.0763, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001364, | |
| "loss": 2.2277, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001362, | |
| "loss": 2.103, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 2.1764, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013580000000000002, | |
| "loss": 2.1602, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013560000000000002, | |
| "loss": 1.8745, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001354, | |
| "loss": 2.1986, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001352, | |
| "loss": 2.106, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013500000000000003, | |
| "loss": 2.3593, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013480000000000002, | |
| "loss": 2.383, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013460000000000002, | |
| "loss": 2.1842, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013440000000000001, | |
| "loss": 2.1973, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001342, | |
| "loss": 2.0959, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000134, | |
| "loss": 2.1906, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013380000000000003, | |
| "loss": 2.2957, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013360000000000002, | |
| "loss": 2.0672, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013340000000000002, | |
| "loss": 2.0441, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001332, | |
| "loss": 1.9296, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000133, | |
| "loss": 2.0165, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001328, | |
| "loss": 2.1575, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013260000000000002, | |
| "loss": 2.4044, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013240000000000002, | |
| "loss": 2.0251, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013220000000000001, | |
| "loss": 2.1, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000132, | |
| "loss": 2.1461, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001318, | |
| "loss": 1.852, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001316, | |
| "loss": 2.2877, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013140000000000002, | |
| "loss": 2.1502, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013120000000000002, | |
| "loss": 1.9459, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000131, | |
| "loss": 1.9575, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001308, | |
| "loss": 2.0544, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001306, | |
| "loss": 2.0893, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001304, | |
| "loss": 2.316, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013020000000000002, | |
| "loss": 2.4258, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013000000000000002, | |
| "loss": 2.3176, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001298, | |
| "loss": 2.1737, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001296, | |
| "loss": 2.0562, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001294, | |
| "loss": 1.9961, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012920000000000002, | |
| "loss": 2.0927, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012900000000000002, | |
| "loss": 2.1692, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012880000000000001, | |
| "loss": 2.4854, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001286, | |
| "loss": 1.9507, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001284, | |
| "loss": 2.1346, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001282, | |
| "loss": 2.0358, | |
| "step": 1361 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 1.9043, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012780000000000002, | |
| "loss": 2.0309, | |
| "step": 1363 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001276, | |
| "loss": 2.0257, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001274, | |
| "loss": 2.2443, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001272, | |
| "loss": 2.1603, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000127, | |
| "loss": 2.0511, | |
| "step": 1367 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012680000000000002, | |
| "loss": 2.0167, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012660000000000001, | |
| "loss": 2.186, | |
| "step": 1369 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001264, | |
| "loss": 2.2996, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001262, | |
| "loss": 2.098, | |
| "step": 1371 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000126, | |
| "loss": 2.1495, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001258, | |
| "loss": 2.3881, | |
| "step": 1373 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012560000000000002, | |
| "loss": 2.1279, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001254, | |
| "loss": 2.2208, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001252, | |
| "loss": 2.2903, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000125, | |
| "loss": 1.8767, | |
| "step": 1377 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001248, | |
| "loss": 2.167, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001246, | |
| "loss": 2.3428, | |
| "step": 1379 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012440000000000002, | |
| "loss": 2.081, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001242, | |
| "loss": 2.0392, | |
| "step": 1381 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000124, | |
| "loss": 2.05, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001238, | |
| "loss": 2.1555, | |
| "step": 1383 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001236, | |
| "loss": 2.4266, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012340000000000002, | |
| "loss": 2.3313, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001232, | |
| "loss": 2.1896, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000123, | |
| "loss": 2.0233, | |
| "step": 1387 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001228, | |
| "loss": 2.0694, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001226, | |
| "loss": 2.1318, | |
| "step": 1389 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001224, | |
| "loss": 2.0452, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012220000000000002, | |
| "loss": 2.1653, | |
| "step": 1391 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000122, | |
| "loss": 1.9305, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001218, | |
| "loss": 2.0996, | |
| "step": 1393 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001216, | |
| "loss": 2.1782, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001214, | |
| "loss": 1.867, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001212, | |
| "loss": 2.2014, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000121, | |
| "loss": 2.2494, | |
| "step": 1397 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001208, | |
| "loss": 1.87, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001206, | |
| "loss": 2.0279, | |
| "step": 1399 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001204, | |
| "loss": 1.9361, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012020000000000001, | |
| "loss": 1.9876, | |
| "step": 1401 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00012, | |
| "loss": 1.9718, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001198, | |
| "loss": 2.189, | |
| "step": 1403 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011960000000000001, | |
| "loss": 2.0789, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001194, | |
| "loss": 2.2943, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001192, | |
| "loss": 2.0277, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000119, | |
| "loss": 2.3626, | |
| "step": 1407 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001188, | |
| "loss": 2.1097, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001186, | |
| "loss": 2.1609, | |
| "step": 1409 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001184, | |
| "loss": 2.1037, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001182, | |
| "loss": 2.3531, | |
| "step": 1411 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000118, | |
| "loss": 2.3028, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001178, | |
| "loss": 2.1177, | |
| "step": 1413 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001176, | |
| "loss": 2.1596, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001174, | |
| "loss": 2.1756, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001172, | |
| "loss": 2.4136, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000117, | |
| "loss": 2.0133, | |
| "step": 1417 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011679999999999999, | |
| "loss": 2.2893, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001166, | |
| "loss": 2.204, | |
| "step": 1419 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001164, | |
| "loss": 2.2641, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011619999999999999, | |
| "loss": 2.0933, | |
| "step": 1421 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000116, | |
| "loss": 2.1942, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001158, | |
| "loss": 1.8752, | |
| "step": 1423 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011559999999999999, | |
| "loss": 2.0854, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001154, | |
| "loss": 2.1233, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001152, | |
| "loss": 2.1401, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011499999999999999, | |
| "loss": 2.2676, | |
| "step": 1427 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001148, | |
| "loss": 2.1719, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001146, | |
| "loss": 1.928, | |
| "step": 1429 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001144, | |
| "loss": 1.891, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001142, | |
| "loss": 1.9891, | |
| "step": 1431 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011399999999999999, | |
| "loss": 2.0371, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001138, | |
| "loss": 2.4039, | |
| "step": 1433 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001136, | |
| "loss": 2.1214, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011339999999999999, | |
| "loss": 2.0021, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001132, | |
| "loss": 2.148, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000113, | |
| "loss": 2.0196, | |
| "step": 1437 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011279999999999999, | |
| "loss": 2.2084, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001126, | |
| "loss": 2.122, | |
| "step": 1439 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011240000000000002, | |
| "loss": 1.9739, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011220000000000002, | |
| "loss": 2.2896, | |
| "step": 1441 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 2.1097, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011180000000000002, | |
| "loss": 2.2033, | |
| "step": 1443 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011160000000000002, | |
| "loss": 2.0265, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011140000000000001, | |
| "loss": 2.2367, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011120000000000002, | |
| "loss": 2.1059, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011100000000000001, | |
| "loss": 2.1773, | |
| "step": 1447 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011080000000000001, | |
| "loss": 2.1443, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011060000000000002, | |
| "loss": 2.1622, | |
| "step": 1449 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011040000000000001, | |
| "loss": 2.0857, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011020000000000001, | |
| "loss": 2.1687, | |
| "step": 1451 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00011000000000000002, | |
| "loss": 2.1121, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010980000000000001, | |
| "loss": 2.0931, | |
| "step": 1453 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010960000000000001, | |
| "loss": 2.1035, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010940000000000002, | |
| "loss": 2.2266, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010920000000000001, | |
| "loss": 2.1868, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000109, | |
| "loss": 2.1009, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010880000000000002, | |
| "loss": 2.1602, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010860000000000001, | |
| "loss": 2.2224, | |
| "step": 1459 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010840000000000002, | |
| "loss": 2.1781, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010820000000000001, | |
| "loss": 2.0947, | |
| "step": 1461 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010800000000000001, | |
| "loss": 2.0345, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010780000000000002, | |
| "loss": 1.9869, | |
| "step": 1463 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010760000000000001, | |
| "loss": 2.0314, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010740000000000001, | |
| "loss": 2.1183, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010720000000000002, | |
| "loss": 2.109, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010700000000000001, | |
| "loss": 2.0632, | |
| "step": 1467 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010680000000000001, | |
| "loss": 2.0599, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010660000000000002, | |
| "loss": 2.2132, | |
| "step": 1469 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010640000000000001, | |
| "loss": 2.1416, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001062, | |
| "loss": 1.9595, | |
| "step": 1471 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010600000000000002, | |
| "loss": 2.3026, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010580000000000001, | |
| "loss": 2.0745, | |
| "step": 1473 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001056, | |
| "loss": 2.0455, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010540000000000001, | |
| "loss": 2.1991, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010520000000000001, | |
| "loss": 2.0947, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000105, | |
| "loss": 2.1074, | |
| "step": 1477 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010480000000000001, | |
| "loss": 2.1716, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010460000000000001, | |
| "loss": 2.4773, | |
| "step": 1479 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001044, | |
| "loss": 2.0446, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010420000000000001, | |
| "loss": 2.2319, | |
| "step": 1481 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 2.51, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001038, | |
| "loss": 2.0337, | |
| "step": 1483 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010360000000000001, | |
| "loss": 1.9343, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001034, | |
| "loss": 2.2586, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001032, | |
| "loss": 1.9841, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010300000000000001, | |
| "loss": 1.9387, | |
| "step": 1487 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001028, | |
| "loss": 1.9061, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010260000000000001, | |
| "loss": 1.9027, | |
| "step": 1489 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010240000000000001, | |
| "loss": 2.0057, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001022, | |
| "loss": 2.0054, | |
| "step": 1491 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010200000000000001, | |
| "loss": 2.1237, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010180000000000001, | |
| "loss": 2.1044, | |
| "step": 1493 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001016, | |
| "loss": 2.0056, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010140000000000001, | |
| "loss": 2.1697, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010120000000000001, | |
| "loss": 1.9487, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000101, | |
| "loss": 2.1263, | |
| "step": 1497 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00010080000000000001, | |
| "loss": 2.1417, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001006, | |
| "loss": 2.471, | |
| "step": 1499 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001004, | |
| "loss": 2.271, | |
| "step": 1500 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 2000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 3.278289333738701e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |