| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 446, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 9.0656, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 8.6633, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-08, | |
| "loss": 8.9119, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-08, | |
| "loss": 8.8743, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-08, | |
| "loss": 8.7322, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 8.8432, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 8.7598, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8e-08, | |
| "loss": 8.7179, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0000000000000001e-07, | |
| "loss": 8.7009, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0000000000000001e-07, | |
| "loss": 8.8273, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 8.6765, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 8.7786, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.4e-07, | |
| "loss": 8.6871, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6e-07, | |
| "loss": 8.8664, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.8e-07, | |
| "loss": 8.658, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 8.8457, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.2e-07, | |
| "loss": 8.5633, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 8.6972, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.6e-07, | |
| "loss": 8.7013, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.8e-07, | |
| "loss": 8.6702, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 8.9022, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.2e-07, | |
| "loss": 8.8576, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 8.7779, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.6e-07, | |
| "loss": 8.834, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.8e-07, | |
| "loss": 8.5806, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 8.4857, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 8.2719, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.4e-07, | |
| "loss": 8.3136, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 8.5229, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 8.2262, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 8.3328, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.2e-07, | |
| "loss": 8.2745, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.4e-07, | |
| "loss": 8.1359, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.6e-07, | |
| "loss": 8.2812, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.800000000000001e-07, | |
| "loss": 8.2066, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 8.2059, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 8.1422, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.4e-07, | |
| "loss": 8.1902, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.6e-07, | |
| "loss": 8.1267, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 8.2181, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.000000000000001e-07, | |
| "loss": 8.0245, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.2e-07, | |
| "loss": 8.1449, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.4e-07, | |
| "loss": 7.998, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.6e-07, | |
| "loss": 8.1898, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.8e-07, | |
| "loss": 8.2973, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 8.1341, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.200000000000001e-07, | |
| "loss": 8.1824, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 8.4063, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.6e-07, | |
| "loss": 8.26, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.8e-07, | |
| "loss": 8.3837, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 7.823, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 7.7067, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 7.8324, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 7.7619, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 7.5628, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 7.6529, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.02e-06, | |
| "loss": 7.6872, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.04e-06, | |
| "loss": 7.7812, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.06e-06, | |
| "loss": 7.6103, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.08e-06, | |
| "loss": 7.5827, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.1e-06, | |
| "loss": 7.5597, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.12e-06, | |
| "loss": 7.5961, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.14e-06, | |
| "loss": 7.521, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.1600000000000001e-06, | |
| "loss": 7.5574, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 7.5383, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 7.5139, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.2200000000000002e-06, | |
| "loss": 7.4725, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.2400000000000002e-06, | |
| "loss": 7.4597, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.26e-06, | |
| "loss": 7.5113, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.28e-06, | |
| "loss": 7.4046, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.3e-06, | |
| "loss": 7.5338, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.32e-06, | |
| "loss": 7.513, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.34e-06, | |
| "loss": 7.4987, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.3600000000000001e-06, | |
| "loss": 7.5031, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 7.4272, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 7.4681, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.42e-06, | |
| "loss": 7.4724, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.44e-06, | |
| "loss": 7.4617, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.46e-06, | |
| "loss": 7.3632, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.48e-06, | |
| "loss": 7.4406, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.5e-06, | |
| "loss": 7.3949, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.52e-06, | |
| "loss": 7.4337, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.54e-06, | |
| "loss": 7.4132, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.56e-06, | |
| "loss": 7.3994, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.5800000000000001e-06, | |
| "loss": 7.355, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 7.4217, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 7.3948, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.6400000000000002e-06, | |
| "loss": 7.3221, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.6600000000000002e-06, | |
| "loss": 7.2875, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 7.4339, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.7000000000000002e-06, | |
| "loss": 7.3627, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.72e-06, | |
| "loss": 7.5915, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.74e-06, | |
| "loss": 7.4667, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.76e-06, | |
| "loss": 7.3207, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 7.282, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 7.5122, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8200000000000002e-06, | |
| "loss": 7.4713, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 7.2456, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8600000000000002e-06, | |
| "loss": 7.5184, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 7.3295, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 7.3192, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 7.2951, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.94e-06, | |
| "loss": 7.1776, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 7.2282, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.98e-06, | |
| "loss": 7.2341, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 7.2799, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.02e-06, | |
| "loss": 7.223, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.04e-06, | |
| "loss": 7.2384, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.06e-06, | |
| "loss": 7.126, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.08e-06, | |
| "loss": 7.2243, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 7.1182, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.12e-06, | |
| "loss": 7.1275, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.1400000000000003e-06, | |
| "loss": 7.2087, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.16e-06, | |
| "loss": 7.2448, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.1800000000000003e-06, | |
| "loss": 7.174, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.2e-06, | |
| "loss": 7.2965, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.2200000000000003e-06, | |
| "loss": 7.2543, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.24e-06, | |
| "loss": 7.2287, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.2600000000000004e-06, | |
| "loss": 7.139, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.28e-06, | |
| "loss": 7.2388, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.3000000000000004e-06, | |
| "loss": 7.1634, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 7.1059, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 7.1523, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.3600000000000003e-06, | |
| "loss": 7.1194, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.38e-06, | |
| "loss": 7.2067, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 7.2303, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.42e-06, | |
| "loss": 7.2051, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 7.1689, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.46e-06, | |
| "loss": 7.2276, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 7.2736, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.5e-06, | |
| "loss": 7.0741, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.52e-06, | |
| "loss": 7.2548, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.5400000000000002e-06, | |
| "loss": 7.1224, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.56e-06, | |
| "loss": 7.1548, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.5800000000000003e-06, | |
| "loss": 7.2653, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.6e-06, | |
| "loss": 7.0389, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 7.2487, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.64e-06, | |
| "loss": 7.173, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.6600000000000004e-06, | |
| "loss": 7.1693, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.68e-06, | |
| "loss": 7.2814, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7000000000000004e-06, | |
| "loss": 7.2482, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7200000000000002e-06, | |
| "loss": 7.2497, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7400000000000004e-06, | |
| "loss": 7.0239, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7600000000000003e-06, | |
| "loss": 7.1039, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 7.0526, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 7.0991, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.82e-06, | |
| "loss": 7.1917, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.84e-06, | |
| "loss": 7.2096, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.86e-06, | |
| "loss": 6.9993, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.88e-06, | |
| "loss": 6.9164, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.9e-06, | |
| "loss": 7.1531, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.92e-06, | |
| "loss": 7.1666, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 7.1133, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.96e-06, | |
| "loss": 7.0933, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 7.0639, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3e-06, | |
| "loss": 7.1154, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.0200000000000003e-06, | |
| "loss": 7.1863, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.04e-06, | |
| "loss": 7.1349, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 7.1162, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.08e-06, | |
| "loss": 7.1441, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.1000000000000004e-06, | |
| "loss": 7.0902, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.12e-06, | |
| "loss": 7.1615, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1400000000000004e-06, | |
| "loss": 7.1423, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1600000000000002e-06, | |
| "loss": 7.1207, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1800000000000005e-06, | |
| "loss": 7.0631, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 7.1458, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.2200000000000005e-06, | |
| "loss": 7.0482, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 7.0045, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.2600000000000006e-06, | |
| "loss": 7.0607, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 7.1924, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.3000000000000006e-06, | |
| "loss": 7.1058, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 7.1291, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 7.0849, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.3600000000000004e-06, | |
| "loss": 7.1391, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.3800000000000007e-06, | |
| "loss": 7.1101, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 7.0369, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 7.0363, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.44e-06, | |
| "loss": 7.1012, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.46e-06, | |
| "loss": 7.0995, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.48e-06, | |
| "loss": 7.1561, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.5e-06, | |
| "loss": 7.1402, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.52e-06, | |
| "loss": 7.0451, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.54e-06, | |
| "loss": 7.148, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 7.1851, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.58e-06, | |
| "loss": 7.1082, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 6.9746, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.62e-06, | |
| "loss": 7.0704, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6400000000000003e-06, | |
| "loss": 7.0528, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.66e-06, | |
| "loss": 7.1322, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.6800000000000003e-06, | |
| "loss": 7.1972, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.7e-06, | |
| "loss": 7.2024, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.7200000000000004e-06, | |
| "loss": 7.0408, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.74e-06, | |
| "loss": 7.1215, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.7600000000000004e-06, | |
| "loss": 7.0417, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 7.01, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 6.9221, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.820000000000001e-06, | |
| "loss": 6.8631, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 7.045, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.86e-06, | |
| "loss": 6.8861, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.88e-06, | |
| "loss": 6.7337, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 7.2072, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 7.1981, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.94e-06, | |
| "loss": 7.1899, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.96e-06, | |
| "loss": 7.1453, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 7.1685, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 7.1609, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.0200000000000005e-06, | |
| "loss": 7.0495, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.04e-06, | |
| "loss": 7.187, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 7.1006, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.08e-06, | |
| "loss": 7.1509, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.1e-06, | |
| "loss": 7.0645, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.12e-06, | |
| "loss": 7.1027, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.14e-06, | |
| "loss": 6.9497, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.16e-06, | |
| "loss": 7.0604, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.18e-06, | |
| "loss": 7.032, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 7.0713, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.22e-06, | |
| "loss": 7.059, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.24e-06, | |
| "loss": 7.0995, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.26e-06, | |
| "loss": 7.0596, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 7.0901, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.3e-06, | |
| "loss": 7.1409, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.32e-06, | |
| "loss": 7.1144, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.34e-06, | |
| "loss": 7.0433, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.360000000000001e-06, | |
| "loss": 6.9968, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.38e-06, | |
| "loss": 7.1748, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.4e-06, | |
| "loss": 7.0851, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.42e-06, | |
| "loss": 7.178, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 7.1072, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 7.0404, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.48e-06, | |
| "loss": 7.0697, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.5e-06, | |
| "loss": 7.1125, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.520000000000001e-06, | |
| "loss": 6.9526, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.540000000000001e-06, | |
| "loss": 7.0077, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.56e-06, | |
| "loss": 7.0224, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.58e-06, | |
| "loss": 7.1392, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 7.0185, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 7.1857, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.6400000000000005e-06, | |
| "loss": 7.0496, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.66e-06, | |
| "loss": 6.9863, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.680000000000001e-06, | |
| "loss": 7.2381, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.7e-06, | |
| "loss": 7.0243, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.7200000000000005e-06, | |
| "loss": 7.0274, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.74e-06, | |
| "loss": 7.1441, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.76e-06, | |
| "loss": 6.9591, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.78e-06, | |
| "loss": 6.9841, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 6.9644, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.8200000000000004e-06, | |
| "loss": 6.7741, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.84e-06, | |
| "loss": 7.0463, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.86e-06, | |
| "loss": 6.9203, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 6.7171, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 7.203, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.92e-06, | |
| "loss": 7.1808, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.94e-06, | |
| "loss": 7.1442, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 7.1738, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 7.2137, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5e-06, | |
| "loss": 7.1239, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.02e-06, | |
| "loss": 7.0826, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.04e-06, | |
| "loss": 7.1231, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 7.0327, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.0800000000000005e-06, | |
| "loss": 7.062, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.1e-06, | |
| "loss": 7.0684, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.12e-06, | |
| "loss": 7.0394, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.140000000000001e-06, | |
| "loss": 7.1074, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 7.1088, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.18e-06, | |
| "loss": 7.1337, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.2e-06, | |
| "loss": 6.9456, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.220000000000001e-06, | |
| "loss": 7.0948, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 6.9807, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.2600000000000005e-06, | |
| "loss": 7.0879, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.28e-06, | |
| "loss": 7.1338, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.300000000000001e-06, | |
| "loss": 7.0202, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.320000000000001e-06, | |
| "loss": 7.0207, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 6.9342, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.36e-06, | |
| "loss": 6.9222, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.380000000000001e-06, | |
| "loss": 7.0924, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 7.0018, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 6.9829, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 7.1071, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 7.1579, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 7.1568, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 7.0662, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.5200000000000005e-06, | |
| "loss": 7.0609, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.540000000000001e-06, | |
| "loss": 6.9644, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 7.0189, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.580000000000001e-06, | |
| "loss": 7.059, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 7.1024, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 7.0941, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.64e-06, | |
| "loss": 7.0827, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.66e-06, | |
| "loss": 7.0403, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.68e-06, | |
| "loss": 7.0994, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.7e-06, | |
| "loss": 7.1633, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.72e-06, | |
| "loss": 7.0665, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.74e-06, | |
| "loss": 6.9603, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.76e-06, | |
| "loss": 6.9473, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.78e-06, | |
| "loss": 6.9004, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.8e-06, | |
| "loss": 6.866, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.82e-06, | |
| "loss": 6.9382, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.84e-06, | |
| "loss": 7.0233, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.86e-06, | |
| "loss": 6.6025, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 6.6428, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.9e-06, | |
| "loss": 7.2423, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.92e-06, | |
| "loss": 7.1925, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.94e-06, | |
| "loss": 7.2078, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 7.0933, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.98e-06, | |
| "loss": 7.255, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6e-06, | |
| "loss": 7.0256, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.02e-06, | |
| "loss": 6.9909, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.040000000000001e-06, | |
| "loss": 7.1496, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 6.9642, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.08e-06, | |
| "loss": 7.0763, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.1e-06, | |
| "loss": 7.0671, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 7.0533, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.1400000000000005e-06, | |
| "loss": 6.982, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.16e-06, | |
| "loss": 6.9866, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.18e-06, | |
| "loss": 7.1002, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.200000000000001e-06, | |
| "loss": 7.0481, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.220000000000001e-06, | |
| "loss": 7.014, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.24e-06, | |
| "loss": 6.9361, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.26e-06, | |
| "loss": 7.0155, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 7.0522, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.300000000000001e-06, | |
| "loss": 6.9896, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.3200000000000005e-06, | |
| "loss": 7.0202, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 6.34e-06, | |
| "loss": 7.078, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 7.0054, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.380000000000001e-06, | |
| "loss": 7.0511, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 7.0643, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.42e-06, | |
| "loss": 7.047, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 6.9796, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 7.0924, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 7.0856, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 7.0372, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.520000000000001e-06, | |
| "loss": 6.9479, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.540000000000001e-06, | |
| "loss": 7.0917, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 7.0307, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.5800000000000005e-06, | |
| "loss": 7.124, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.600000000000001e-06, | |
| "loss": 7.0226, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 6.9063, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.640000000000001e-06, | |
| "loss": 7.0728, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 7.0572, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.680000000000001e-06, | |
| "loss": 7.0749, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.700000000000001e-06, | |
| "loss": 6.969, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.720000000000001e-06, | |
| "loss": 7.0759, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 6.9305, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.760000000000001e-06, | |
| "loss": 6.9073, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 6.881, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 6.9599, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.820000000000001e-06, | |
| "loss": 6.8266, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 6.6888, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 6.7824, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.88e-06, | |
| "loss": 6.5344, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.9e-06, | |
| "loss": 7.1566, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.92e-06, | |
| "loss": 7.2001, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 7.0835, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.96e-06, | |
| "loss": 7.023, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.98e-06, | |
| "loss": 7.0585, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7e-06, | |
| "loss": 7.1668, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.0200000000000006e-06, | |
| "loss": 7.0, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.04e-06, | |
| "loss": 7.0832, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.06e-06, | |
| "loss": 6.9888, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.08e-06, | |
| "loss": 6.9788, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.100000000000001e-06, | |
| "loss": 7.0177, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 6.9336, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.14e-06, | |
| "loss": 6.8755, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.16e-06, | |
| "loss": 6.938, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.180000000000001e-06, | |
| "loss": 7.0578, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 7.0069, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.22e-06, | |
| "loss": 7.1085, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.24e-06, | |
| "loss": 7.0106, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.260000000000001e-06, | |
| "loss": 6.9103, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 6.9784, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.3e-06, | |
| "loss": 7.0409, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.32e-06, | |
| "loss": 7.069, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 6.9829, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 6.8754, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 6.9719, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.4e-06, | |
| "loss": 6.9281, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 6.9857, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 6.9805, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 6.9681, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.48e-06, | |
| "loss": 6.962, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 7.0683, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.520000000000001e-06, | |
| "loss": 7.0167, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.540000000000001e-06, | |
| "loss": 7.0039, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 6.9635, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.58e-06, | |
| "loss": 7.0326, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 6.9914, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 6.9522, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.640000000000001e-06, | |
| "loss": 7.0191, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 6.9369, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 6.8405, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.7e-06, | |
| "loss": 6.9025, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.72e-06, | |
| "loss": 6.942, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.74e-06, | |
| "loss": 6.8378, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.76e-06, | |
| "loss": 6.8714, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.78e-06, | |
| "loss": 6.8833, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 6.9241, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.820000000000001e-06, | |
| "loss": 6.8455, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 6.7152, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 6.7795, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.88e-06, | |
| "loss": 6.3986, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.9e-06, | |
| "loss": 7.1826, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.92e-06, | |
| "loss": 7.1695, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.94e-06, | |
| "loss": 7.2669, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 7.1548, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 6.9942, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 7.041, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.020000000000001e-06, | |
| "loss": 7.0299, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.040000000000001e-06, | |
| "loss": 6.9149, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.06e-06, | |
| "loss": 6.8887, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.08e-06, | |
| "loss": 6.9186, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.1e-06, | |
| "loss": 7.0415, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 6.9518, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.14e-06, | |
| "loss": 6.9932, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.16e-06, | |
| "loss": 6.9789, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.18e-06, | |
| "loss": 7.0404, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.2e-06, | |
| "loss": 7.0041, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 6.9786, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.24e-06, | |
| "loss": 6.8628, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.26e-06, | |
| "loss": 6.9054, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.28e-06, | |
| "loss": 6.9641, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.3e-06, | |
| "loss": 7.002, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.32e-06, | |
| "loss": 6.8865, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.34e-06, | |
| "loss": 6.93, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.36e-06, | |
| "loss": 7.0314, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.380000000000001e-06, | |
| "loss": 6.8996, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 6.9738, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.42e-06, | |
| "loss": 6.9319, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.44e-06, | |
| "loss": 6.8332, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.46e-06, | |
| "loss": 6.8213, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.48e-06, | |
| "loss": 6.8614, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.5e-06, | |
| "loss": 6.9083, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.52e-06, | |
| "loss": 7.0035, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.540000000000001e-06, | |
| "loss": 6.8831, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 6.8566, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.580000000000001e-06, | |
| "loss": 6.9702, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.6e-06, | |
| "loss": 6.8319, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.62e-06, | |
| "loss": 6.8398, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.64e-06, | |
| "loss": 6.9327, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.66e-06, | |
| "loss": 6.8757, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.68e-06, | |
| "loss": 6.8738, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.700000000000001e-06, | |
| "loss": 6.7065, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.720000000000001e-06, | |
| "loss": 6.9136, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 6.5767, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.76e-06, | |
| "loss": 6.8314, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.78e-06, | |
| "loss": 6.5081, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.8e-06, | |
| "loss": 6.3829, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 446, | |
| "total_flos": 0.0, | |
| "train_loss": 7.25599230030727, | |
| "train_runtime": 4990.6204, | |
| "train_samples_per_second": 5.718, | |
| "train_steps_per_second": 0.089 | |
| } | |
| ], | |
| "max_steps": 446, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |