|
{ |
|
"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": 2e-08, |
|
"loss": 4.7541, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4e-08, |
|
"loss": 4.8927, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 6.000000000000001e-08, |
|
"loss": 4.8644, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 6.000000000000001e-08, |
|
"loss": 4.8675, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8e-08, |
|
"loss": 4.7998, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.0000000000000001e-07, |
|
"loss": 4.8669, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.2000000000000002e-07, |
|
"loss": 4.8499, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.4e-07, |
|
"loss": 4.8967, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.6e-07, |
|
"loss": 4.8873, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.8e-07, |
|
"loss": 4.7778, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 2.0000000000000002e-07, |
|
"loss": 4.8975, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.2e-07, |
|
"loss": 4.8687, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.4000000000000003e-07, |
|
"loss": 4.8289, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.6e-07, |
|
"loss": 4.8372, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.8e-07, |
|
"loss": 4.7169, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.0000000000000004e-07, |
|
"loss": 4.8774, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.2e-07, |
|
"loss": 4.8112, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.4000000000000003e-07, |
|
"loss": 4.9211, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.6e-07, |
|
"loss": 4.7166, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.8e-07, |
|
"loss": 4.9005, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.0000000000000003e-07, |
|
"loss": 4.8957, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.2000000000000006e-07, |
|
"loss": 4.8022, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.4e-07, |
|
"loss": 4.9258, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.6000000000000004e-07, |
|
"loss": 4.9286, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.800000000000001e-07, |
|
"loss": 4.717, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.000000000000001e-07, |
|
"loss": 4.9856, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.2e-07, |
|
"loss": 4.8427, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.4e-07, |
|
"loss": 4.7164, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 5.6e-07, |
|
"loss": 4.6958, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 5.800000000000001e-07, |
|
"loss": 4.8882, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.000000000000001e-07, |
|
"loss": 4.8631, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.200000000000001e-07, |
|
"loss": 4.71, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.4e-07, |
|
"loss": 4.8112, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 6.6e-07, |
|
"loss": 4.7364, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 6.800000000000001e-07, |
|
"loss": 4.8314, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 7.000000000000001e-07, |
|
"loss": 4.7743, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 7.2e-07, |
|
"loss": 4.8398, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 7.4e-07, |
|
"loss": 4.7806, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 7.6e-07, |
|
"loss": 4.9238, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 7.8e-07, |
|
"loss": 4.701, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.000000000000001e-07, |
|
"loss": 4.7911, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.200000000000001e-07, |
|
"loss": 4.9134, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 8.400000000000001e-07, |
|
"loss": 4.6839, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 8.6e-07, |
|
"loss": 4.8494, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 8.8e-07, |
|
"loss": 4.9975, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.000000000000001e-07, |
|
"loss": 5.1271, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.200000000000001e-07, |
|
"loss": 5.0245, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.400000000000001e-07, |
|
"loss": 5.3001, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.600000000000001e-07, |
|
"loss": 5.3946, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.600000000000001e-07, |
|
"loss": 5.2908, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.800000000000001e-07, |
|
"loss": 4.6459, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 4.5063, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.02e-06, |
|
"loss": 4.6776, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.04e-06, |
|
"loss": 4.5814, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.06e-06, |
|
"loss": 4.4042, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.08e-06, |
|
"loss": 4.6195, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.1e-06, |
|
"loss": 4.6391, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.12e-06, |
|
"loss": 4.7295, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.14e-06, |
|
"loss": 4.6633, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.1600000000000001e-06, |
|
"loss": 4.4993, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.1800000000000001e-06, |
|
"loss": 4.5966, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.2000000000000002e-06, |
|
"loss": 4.536, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.2200000000000002e-06, |
|
"loss": 4.4949, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.2400000000000002e-06, |
|
"loss": 4.6052, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.26e-06, |
|
"loss": 4.558, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.28e-06, |
|
"loss": 4.4341, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.3e-06, |
|
"loss": 4.5926, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.32e-06, |
|
"loss": 4.4335, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.34e-06, |
|
"loss": 4.5063, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3600000000000001e-06, |
|
"loss": 4.4816, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.3800000000000001e-06, |
|
"loss": 4.6217, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.4000000000000001e-06, |
|
"loss": 4.5979, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.42e-06, |
|
"loss": 4.4467, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.44e-06, |
|
"loss": 4.5226, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.46e-06, |
|
"loss": 4.5218, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.48e-06, |
|
"loss": 4.4656, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.5e-06, |
|
"loss": 4.5227, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.52e-06, |
|
"loss": 4.4625, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.54e-06, |
|
"loss": 4.4515, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.56e-06, |
|
"loss": 4.5694, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.5800000000000001e-06, |
|
"loss": 4.5223, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.6000000000000001e-06, |
|
"loss": 4.4574, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.6200000000000002e-06, |
|
"loss": 4.5745, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.6400000000000002e-06, |
|
"loss": 4.4455, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.6600000000000002e-06, |
|
"loss": 4.4914, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.6800000000000002e-06, |
|
"loss": 4.4821, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.7000000000000002e-06, |
|
"loss": 4.5791, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.72e-06, |
|
"loss": 4.4611, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.74e-06, |
|
"loss": 4.6041, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.76e-06, |
|
"loss": 4.5501, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.7800000000000001e-06, |
|
"loss": 4.4837, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8000000000000001e-06, |
|
"loss": 4.7265, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8200000000000002e-06, |
|
"loss": 4.5814, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8400000000000002e-06, |
|
"loss": 4.5799, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8600000000000002e-06, |
|
"loss": 4.5869, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8800000000000002e-06, |
|
"loss": 4.7579, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.9000000000000002e-06, |
|
"loss": 4.7943, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.9200000000000003e-06, |
|
"loss": 4.5681, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.94e-06, |
|
"loss": 4.8655, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.9600000000000003e-06, |
|
"loss": 4.6563, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.98e-06, |
|
"loss": 4.4149, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 4.3519, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.02e-06, |
|
"loss": 4.2156, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.04e-06, |
|
"loss": 4.2846, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.06e-06, |
|
"loss": 4.2302, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.08e-06, |
|
"loss": 4.2765, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.1000000000000002e-06, |
|
"loss": 4.3781, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.12e-06, |
|
"loss": 4.2992, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.1400000000000003e-06, |
|
"loss": 4.2881, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.16e-06, |
|
"loss": 4.3361, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.1800000000000003e-06, |
|
"loss": 4.3737, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.2e-06, |
|
"loss": 4.3282, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.2200000000000003e-06, |
|
"loss": 4.3385, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.24e-06, |
|
"loss": 4.3158, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.2600000000000004e-06, |
|
"loss": 4.3688, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.28e-06, |
|
"loss": 4.4229, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.3000000000000004e-06, |
|
"loss": 4.3013, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.3200000000000002e-06, |
|
"loss": 4.4761, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.3400000000000005e-06, |
|
"loss": 4.3552, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.3600000000000003e-06, |
|
"loss": 4.3892, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.38e-06, |
|
"loss": 4.2258, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 4.3049, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.42e-06, |
|
"loss": 4.3383, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.4400000000000004e-06, |
|
"loss": 4.3089, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.46e-06, |
|
"loss": 4.4328, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.4800000000000004e-06, |
|
"loss": 4.4379, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.5e-06, |
|
"loss": 4.3453, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.52e-06, |
|
"loss": 4.3193, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.5400000000000002e-06, |
|
"loss": 4.2188, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.56e-06, |
|
"loss": 4.3847, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.5800000000000003e-06, |
|
"loss": 4.2899, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.6e-06, |
|
"loss": 4.331, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.6200000000000003e-06, |
|
"loss": 4.2797, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.64e-06, |
|
"loss": 4.285, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.6600000000000004e-06, |
|
"loss": 4.4472, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.68e-06, |
|
"loss": 4.3031, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.7000000000000004e-06, |
|
"loss": 4.4124, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.7200000000000002e-06, |
|
"loss": 4.4394, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.7400000000000004e-06, |
|
"loss": 4.4579, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.7600000000000003e-06, |
|
"loss": 4.4306, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.7800000000000005e-06, |
|
"loss": 4.5488, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.8000000000000003e-06, |
|
"loss": 4.4791, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.82e-06, |
|
"loss": 4.3989, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.84e-06, |
|
"loss": 4.4766, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.86e-06, |
|
"loss": 4.3723, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.88e-06, |
|
"loss": 4.4711, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.9e-06, |
|
"loss": 4.7235, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.92e-06, |
|
"loss": 4.8624, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.9400000000000002e-06, |
|
"loss": 4.5986, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 2.96e-06, |
|
"loss": 4.7065, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 2.9800000000000003e-06, |
|
"loss": 4.2124, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 3e-06, |
|
"loss": 4.2878, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 3.0200000000000003e-06, |
|
"loss": 4.3026, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 3.04e-06, |
|
"loss": 4.2101, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 3.0600000000000003e-06, |
|
"loss": 4.2986, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 3.08e-06, |
|
"loss": 4.1734, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 3.1000000000000004e-06, |
|
"loss": 4.2086, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 3.12e-06, |
|
"loss": 4.3003, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 3.1400000000000004e-06, |
|
"loss": 4.3039, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 3.1600000000000002e-06, |
|
"loss": 4.3001, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 3.1800000000000005e-06, |
|
"loss": 4.3526, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 3.2000000000000003e-06, |
|
"loss": 4.312, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.2200000000000005e-06, |
|
"loss": 4.2679, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.2400000000000003e-06, |
|
"loss": 4.333, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.2600000000000006e-06, |
|
"loss": 4.1979, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.2800000000000004e-06, |
|
"loss": 4.2588, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.3000000000000006e-06, |
|
"loss": 4.1736, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 3.3200000000000004e-06, |
|
"loss": 4.2609, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 3.3400000000000006e-06, |
|
"loss": 4.3197, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 3.3600000000000004e-06, |
|
"loss": 4.3275, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 3.3800000000000007e-06, |
|
"loss": 4.3226, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.4000000000000005e-06, |
|
"loss": 4.3706, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.4200000000000007e-06, |
|
"loss": 4.1764, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.44e-06, |
|
"loss": 4.3279, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.46e-06, |
|
"loss": 4.3795, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.48e-06, |
|
"loss": 4.257, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 3.5e-06, |
|
"loss": 4.2217, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 3.52e-06, |
|
"loss": 4.3091, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 3.54e-06, |
|
"loss": 4.3617, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 3.5600000000000002e-06, |
|
"loss": 4.3626, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 3.58e-06, |
|
"loss": 4.292, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 3.6000000000000003e-06, |
|
"loss": 4.3384, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 3.62e-06, |
|
"loss": 4.3467, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 3.6400000000000003e-06, |
|
"loss": 4.3526, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 3.66e-06, |
|
"loss": 4.3269, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 3.6800000000000003e-06, |
|
"loss": 4.1944, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 3.7e-06, |
|
"loss": 4.2104, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 3.7200000000000004e-06, |
|
"loss": 4.1716, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 3.74e-06, |
|
"loss": 4.2872, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 3.7600000000000004e-06, |
|
"loss": 4.4067, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 3.7800000000000002e-06, |
|
"loss": 4.3407, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 3.8000000000000005e-06, |
|
"loss": 4.2671, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 3.820000000000001e-06, |
|
"loss": 4.2976, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 3.8400000000000005e-06, |
|
"loss": 4.3868, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 3.86e-06, |
|
"loss": 4.4837, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 3.88e-06, |
|
"loss": 4.451, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 3.900000000000001e-06, |
|
"loss": 4.3893, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 3.920000000000001e-06, |
|
"loss": 4.6928, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 3.94e-06, |
|
"loss": 4.6071, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 3.96e-06, |
|
"loss": 4.5822, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 3.980000000000001e-06, |
|
"loss": 4.317, |
|
"step": 201 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 4.254, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.0200000000000005e-06, |
|
"loss": 4.2586, |
|
"step": 203 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.04e-06, |
|
"loss": 4.247, |
|
"step": 204 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.060000000000001e-06, |
|
"loss": 4.3217, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.08e-06, |
|
"loss": 4.3277, |
|
"step": 206 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.1e-06, |
|
"loss": 4.1827, |
|
"step": 207 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.12e-06, |
|
"loss": 4.2797, |
|
"step": 208 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.14e-06, |
|
"loss": 4.3171, |
|
"step": 209 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.16e-06, |
|
"loss": 4.2913, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.18e-06, |
|
"loss": 4.1687, |
|
"step": 211 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.2000000000000004e-06, |
|
"loss": 4.2174, |
|
"step": 212 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.22e-06, |
|
"loss": 4.1299, |
|
"step": 213 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.24e-06, |
|
"loss": 4.2068, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.26e-06, |
|
"loss": 4.2561, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.2800000000000005e-06, |
|
"loss": 4.2275, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.3e-06, |
|
"loss": 4.2821, |
|
"step": 217 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.32e-06, |
|
"loss": 4.2398, |
|
"step": 218 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.34e-06, |
|
"loss": 4.1627, |
|
"step": 219 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.360000000000001e-06, |
|
"loss": 4.2745, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.38e-06, |
|
"loss": 4.2645, |
|
"step": 221 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.4e-06, |
|
"loss": 4.2325, |
|
"step": 222 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.42e-06, |
|
"loss": 4.2634, |
|
"step": 223 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.440000000000001e-06, |
|
"loss": 4.1924, |
|
"step": 224 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.4600000000000005e-06, |
|
"loss": 4.3595, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.48e-06, |
|
"loss": 4.2554, |
|
"step": 226 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.5e-06, |
|
"loss": 4.3735, |
|
"step": 227 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.520000000000001e-06, |
|
"loss": 4.3111, |
|
"step": 228 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.540000000000001e-06, |
|
"loss": 4.1945, |
|
"step": 229 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.56e-06, |
|
"loss": 4.2758, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.58e-06, |
|
"loss": 4.3566, |
|
"step": 231 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.600000000000001e-06, |
|
"loss": 4.1538, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.620000000000001e-06, |
|
"loss": 4.171, |
|
"step": 233 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.6400000000000005e-06, |
|
"loss": 4.2496, |
|
"step": 234 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.66e-06, |
|
"loss": 4.3324, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.680000000000001e-06, |
|
"loss": 4.2674, |
|
"step": 236 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.7e-06, |
|
"loss": 4.2991, |
|
"step": 237 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.7200000000000005e-06, |
|
"loss": 4.2457, |
|
"step": 238 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.74e-06, |
|
"loss": 4.1922, |
|
"step": 239 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.76e-06, |
|
"loss": 4.3, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.78e-06, |
|
"loss": 4.274, |
|
"step": 241 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 4.3015, |
|
"step": 242 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.8200000000000004e-06, |
|
"loss": 4.5195, |
|
"step": 243 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.84e-06, |
|
"loss": 4.3539, |
|
"step": 244 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.86e-06, |
|
"loss": 4.3367, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.880000000000001e-06, |
|
"loss": 4.2667, |
|
"step": 246 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.9000000000000005e-06, |
|
"loss": 4.188, |
|
"step": 247 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.92e-06, |
|
"loss": 4.5605, |
|
"step": 248 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.94e-06, |
|
"loss": 4.6662, |
|
"step": 249 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.960000000000001e-06, |
|
"loss": 4.6607, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.980000000000001e-06, |
|
"loss": 4.2306, |
|
"step": 251 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 5e-06, |
|
"loss": 4.266, |
|
"step": 252 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 5.02e-06, |
|
"loss": 4.3091, |
|
"step": 253 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 5.04e-06, |
|
"loss": 4.156, |
|
"step": 254 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 5.060000000000001e-06, |
|
"loss": 4.2564, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 5.0800000000000005e-06, |
|
"loss": 4.2243, |
|
"step": 256 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 5.1e-06, |
|
"loss": 4.1986, |
|
"step": 257 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 5.12e-06, |
|
"loss": 4.2144, |
|
"step": 258 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 5.140000000000001e-06, |
|
"loss": 4.2165, |
|
"step": 259 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 5.1600000000000006e-06, |
|
"loss": 4.1611, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 5.18e-06, |
|
"loss": 4.2733, |
|
"step": 261 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 5.2e-06, |
|
"loss": 4.2103, |
|
"step": 262 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 5.220000000000001e-06, |
|
"loss": 4.3427, |
|
"step": 263 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 5.240000000000001e-06, |
|
"loss": 4.2133, |
|
"step": 264 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 5.2600000000000005e-06, |
|
"loss": 4.2544, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 5.28e-06, |
|
"loss": 4.225, |
|
"step": 266 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 5.300000000000001e-06, |
|
"loss": 4.254, |
|
"step": 267 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 5.320000000000001e-06, |
|
"loss": 4.204, |
|
"step": 268 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 5.3400000000000005e-06, |
|
"loss": 4.2903, |
|
"step": 269 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 5.36e-06, |
|
"loss": 4.1648, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 5.380000000000001e-06, |
|
"loss": 4.0881, |
|
"step": 271 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 5.400000000000001e-06, |
|
"loss": 4.272, |
|
"step": 272 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 5.420000000000001e-06, |
|
"loss": 4.083, |
|
"step": 273 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 5.4400000000000004e-06, |
|
"loss": 4.1777, |
|
"step": 274 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 5.460000000000001e-06, |
|
"loss": 4.2714, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 5.480000000000001e-06, |
|
"loss": 4.211, |
|
"step": 276 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 5.500000000000001e-06, |
|
"loss": 4.1641, |
|
"step": 277 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 5.5200000000000005e-06, |
|
"loss": 4.2538, |
|
"step": 278 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 5.540000000000001e-06, |
|
"loss": 4.2379, |
|
"step": 279 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 5.560000000000001e-06, |
|
"loss": 4.3645, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 5.580000000000001e-06, |
|
"loss": 4.2645, |
|
"step": 281 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 5.600000000000001e-06, |
|
"loss": 4.12, |
|
"step": 282 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 5.620000000000001e-06, |
|
"loss": 4.1472, |
|
"step": 283 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 5.64e-06, |
|
"loss": 4.2229, |
|
"step": 284 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 5.66e-06, |
|
"loss": 4.3255, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 5.68e-06, |
|
"loss": 4.2164, |
|
"step": 286 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 5.7e-06, |
|
"loss": 4.3277, |
|
"step": 287 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.72e-06, |
|
"loss": 4.2827, |
|
"step": 288 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.74e-06, |
|
"loss": 4.1908, |
|
"step": 289 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.76e-06, |
|
"loss": 4.2463, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.78e-06, |
|
"loss": 4.4979, |
|
"step": 291 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.8e-06, |
|
"loss": 4.3692, |
|
"step": 292 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.82e-06, |
|
"loss": 4.4422, |
|
"step": 293 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.84e-06, |
|
"loss": 4.4088, |
|
"step": 294 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.86e-06, |
|
"loss": 4.3877, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.8800000000000005e-06, |
|
"loss": 4.3863, |
|
"step": 296 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.9e-06, |
|
"loss": 4.291, |
|
"step": 297 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.92e-06, |
|
"loss": 4.6069, |
|
"step": 298 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.94e-06, |
|
"loss": 4.3674, |
|
"step": 299 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.9600000000000005e-06, |
|
"loss": 4.7813, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.98e-06, |
|
"loss": 4.2702, |
|
"step": 301 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 6e-06, |
|
"loss": 4.2331, |
|
"step": 302 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 6.02e-06, |
|
"loss": 4.2356, |
|
"step": 303 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 6.040000000000001e-06, |
|
"loss": 4.233, |
|
"step": 304 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 6.0600000000000004e-06, |
|
"loss": 4.2104, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 6.08e-06, |
|
"loss": 4.2141, |
|
"step": 306 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 6.1e-06, |
|
"loss": 4.2022, |
|
"step": 307 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 6.120000000000001e-06, |
|
"loss": 4.1967, |
|
"step": 308 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 6.1400000000000005e-06, |
|
"loss": 4.1975, |
|
"step": 309 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 6.16e-06, |
|
"loss": 4.2753, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 6.18e-06, |
|
"loss": 4.0923, |
|
"step": 311 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 6.200000000000001e-06, |
|
"loss": 4.2573, |
|
"step": 312 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 6.220000000000001e-06, |
|
"loss": 4.2671, |
|
"step": 313 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 6.24e-06, |
|
"loss": 4.183, |
|
"step": 314 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 6.26e-06, |
|
"loss": 4.2129, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 6.280000000000001e-06, |
|
"loss": 4.1825, |
|
"step": 316 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 6.300000000000001e-06, |
|
"loss": 4.1694, |
|
"step": 317 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 6.3200000000000005e-06, |
|
"loss": 4.214, |
|
"step": 318 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 6.34e-06, |
|
"loss": 4.201, |
|
"step": 319 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 6.360000000000001e-06, |
|
"loss": 4.2793, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 6.380000000000001e-06, |
|
"loss": 4.1701, |
|
"step": 321 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 6.4000000000000006e-06, |
|
"loss": 4.0582, |
|
"step": 322 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 6.42e-06, |
|
"loss": 4.2408, |
|
"step": 323 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 6.440000000000001e-06, |
|
"loss": 4.1051, |
|
"step": 324 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 6.460000000000001e-06, |
|
"loss": 4.2394, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 6.480000000000001e-06, |
|
"loss": 4.197, |
|
"step": 326 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 6.5000000000000004e-06, |
|
"loss": 4.1724, |
|
"step": 327 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 6.520000000000001e-06, |
|
"loss": 4.147, |
|
"step": 328 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 6.540000000000001e-06, |
|
"loss": 4.1868, |
|
"step": 329 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 6.560000000000001e-06, |
|
"loss": 4.2377, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 6.5800000000000005e-06, |
|
"loss": 4.2337, |
|
"step": 331 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 6.600000000000001e-06, |
|
"loss": 4.136, |
|
"step": 332 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 6.620000000000001e-06, |
|
"loss": 4.2892, |
|
"step": 333 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 6.640000000000001e-06, |
|
"loss": 4.1944, |
|
"step": 334 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 6.660000000000001e-06, |
|
"loss": 4.37, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 6.680000000000001e-06, |
|
"loss": 4.1538, |
|
"step": 336 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 6.700000000000001e-06, |
|
"loss": 4.1614, |
|
"step": 337 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 6.720000000000001e-06, |
|
"loss": 4.2704, |
|
"step": 338 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 6.740000000000001e-06, |
|
"loss": 4.2589, |
|
"step": 339 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 6.760000000000001e-06, |
|
"loss": 4.3103, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 6.780000000000001e-06, |
|
"loss": 4.2509, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.800000000000001e-06, |
|
"loss": 4.4144, |
|
"step": 342 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.820000000000001e-06, |
|
"loss": 4.3132, |
|
"step": 343 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.8400000000000014e-06, |
|
"loss": 4.3986, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.860000000000001e-06, |
|
"loss": 4.1276, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.88e-06, |
|
"loss": 4.3877, |
|
"step": 346 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.9e-06, |
|
"loss": 4.379, |
|
"step": 347 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.92e-06, |
|
"loss": 4.318, |
|
"step": 348 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.9400000000000005e-06, |
|
"loss": 4.5441, |
|
"step": 349 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.96e-06, |
|
"loss": 4.5578, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 6.98e-06, |
|
"loss": 4.2743, |
|
"step": 351 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 7e-06, |
|
"loss": 4.2212, |
|
"step": 352 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 7.0200000000000006e-06, |
|
"loss": 4.1421, |
|
"step": 353 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 7.04e-06, |
|
"loss": 4.0231, |
|
"step": 354 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.06e-06, |
|
"loss": 4.2178, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.08e-06, |
|
"loss": 4.3222, |
|
"step": 356 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.100000000000001e-06, |
|
"loss": 4.1956, |
|
"step": 357 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.1200000000000004e-06, |
|
"loss": 4.2345, |
|
"step": 358 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.14e-06, |
|
"loss": 4.1192, |
|
"step": 359 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 7.16e-06, |
|
"loss": 4.1955, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 7.180000000000001e-06, |
|
"loss": 4.1376, |
|
"step": 361 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 7.2000000000000005e-06, |
|
"loss": 4.1929, |
|
"step": 362 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 7.22e-06, |
|
"loss": 4.115, |
|
"step": 363 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 7.24e-06, |
|
"loss": 4.2139, |
|
"step": 364 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 7.260000000000001e-06, |
|
"loss": 4.3332, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 7.280000000000001e-06, |
|
"loss": 4.1502, |
|
"step": 366 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 7.3e-06, |
|
"loss": 4.233, |
|
"step": 367 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.32e-06, |
|
"loss": 4.2126, |
|
"step": 368 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.340000000000001e-06, |
|
"loss": 4.1302, |
|
"step": 369 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.360000000000001e-06, |
|
"loss": 4.143, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.3800000000000005e-06, |
|
"loss": 4.1357, |
|
"step": 371 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.4e-06, |
|
"loss": 4.2163, |
|
"step": 372 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 7.420000000000001e-06, |
|
"loss": 4.2068, |
|
"step": 373 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 7.440000000000001e-06, |
|
"loss": 4.0071, |
|
"step": 374 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 7.4600000000000006e-06, |
|
"loss": 4.2376, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 7.48e-06, |
|
"loss": 4.0878, |
|
"step": 376 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 4.2106, |
|
"step": 377 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.520000000000001e-06, |
|
"loss": 4.1972, |
|
"step": 378 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.540000000000001e-06, |
|
"loss": 4.1123, |
|
"step": 379 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.5600000000000005e-06, |
|
"loss": 4.1345, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.58e-06, |
|
"loss": 4.1914, |
|
"step": 381 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 7.600000000000001e-06, |
|
"loss": 4.2757, |
|
"step": 382 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 7.620000000000001e-06, |
|
"loss": 4.1834, |
|
"step": 383 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 7.640000000000001e-06, |
|
"loss": 4.1181, |
|
"step": 384 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 7.660000000000001e-06, |
|
"loss": 4.2012, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 7.680000000000001e-06, |
|
"loss": 4.077, |
|
"step": 386 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 7.7e-06, |
|
"loss": 4.2641, |
|
"step": 387 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 7.72e-06, |
|
"loss": 4.2954, |
|
"step": 388 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 7.74e-06, |
|
"loss": 4.1907, |
|
"step": 389 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 7.76e-06, |
|
"loss": 4.1714, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.78e-06, |
|
"loss": 4.2965, |
|
"step": 391 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.800000000000002e-06, |
|
"loss": 4.1424, |
|
"step": 392 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.820000000000001e-06, |
|
"loss": 4.139, |
|
"step": 393 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.840000000000001e-06, |
|
"loss": 4.2231, |
|
"step": 394 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.860000000000001e-06, |
|
"loss": 4.3462, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.88e-06, |
|
"loss": 4.3144, |
|
"step": 396 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.9e-06, |
|
"loss": 4.4527, |
|
"step": 397 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.92e-06, |
|
"loss": 4.456, |
|
"step": 398 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.94e-06, |
|
"loss": 4.5934, |
|
"step": 399 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 7.960000000000002e-06, |
|
"loss": 4.5134, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 7.980000000000002e-06, |
|
"loss": 4.217, |
|
"step": 401 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 4.0645, |
|
"step": 402 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.020000000000001e-06, |
|
"loss": 4.1291, |
|
"step": 403 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.040000000000001e-06, |
|
"loss": 4.1237, |
|
"step": 404 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.06e-06, |
|
"loss": 4.0902, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.08e-06, |
|
"loss": 4.1787, |
|
"step": 406 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.1e-06, |
|
"loss": 4.1467, |
|
"step": 407 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.120000000000002e-06, |
|
"loss": 4.2247, |
|
"step": 408 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.14e-06, |
|
"loss": 4.1621, |
|
"step": 409 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.16e-06, |
|
"loss": 4.141, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.18e-06, |
|
"loss": 4.3667, |
|
"step": 411 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.2e-06, |
|
"loss": 4.1049, |
|
"step": 412 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.220000000000001e-06, |
|
"loss": 4.0321, |
|
"step": 413 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.24e-06, |
|
"loss": 4.1928, |
|
"step": 414 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.26e-06, |
|
"loss": 4.2578, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.28e-06, |
|
"loss": 4.1304, |
|
"step": 416 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.3e-06, |
|
"loss": 4.1514, |
|
"step": 417 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.32e-06, |
|
"loss": 4.0506, |
|
"step": 418 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.34e-06, |
|
"loss": 4.1579, |
|
"step": 419 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.36e-06, |
|
"loss": 4.0673, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.380000000000001e-06, |
|
"loss": 4.2447, |
|
"step": 421 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 4.0858, |
|
"step": 422 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.42e-06, |
|
"loss": 4.1902, |
|
"step": 423 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.44e-06, |
|
"loss": 4.2478, |
|
"step": 424 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.46e-06, |
|
"loss": 4.1727, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.48e-06, |
|
"loss": 4.225, |
|
"step": 426 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.5e-06, |
|
"loss": 4.1784, |
|
"step": 427 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.52e-06, |
|
"loss": 4.1503, |
|
"step": 428 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.540000000000001e-06, |
|
"loss": 4.1028, |
|
"step": 429 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.560000000000001e-06, |
|
"loss": 4.1624, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.580000000000001e-06, |
|
"loss": 4.2611, |
|
"step": 431 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.6e-06, |
|
"loss": 4.3745, |
|
"step": 432 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.62e-06, |
|
"loss": 4.17, |
|
"step": 433 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.64e-06, |
|
"loss": 4.1702, |
|
"step": 434 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.66e-06, |
|
"loss": 4.2199, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.68e-06, |
|
"loss": 4.1389, |
|
"step": 436 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.700000000000001e-06, |
|
"loss": 4.1324, |
|
"step": 437 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.720000000000001e-06, |
|
"loss": 4.3086, |
|
"step": 438 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.740000000000001e-06, |
|
"loss": 4.2724, |
|
"step": 439 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.76e-06, |
|
"loss": 4.3117, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.78e-06, |
|
"loss": 4.2593, |
|
"step": 441 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.8e-06, |
|
"loss": 4.4625, |
|
"step": 442 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.82e-06, |
|
"loss": 4.3707, |
|
"step": 443 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.84e-06, |
|
"loss": 4.3694, |
|
"step": 444 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.860000000000002e-06, |
|
"loss": 4.6504, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.860000000000002e-06, |
|
"loss": 4.3145, |
|
"step": 446 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 446, |
|
"total_flos": 0.0, |
|
"train_loss": 4.375692403904526, |
|
"train_runtime": 5180.8925, |
|
"train_samples_per_second": 5.508, |
|
"train_steps_per_second": 0.086 |
|
} |
|
], |
|
"max_steps": 446, |
|
"num_train_epochs": 1, |
|
"total_flos": 0.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|