| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9987389659520807, | |
| "global_step": 297, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6e-07, | |
| "loss": 4.7698, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2e-06, | |
| "loss": 4.8469, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-06, | |
| "loss": 4.9086, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-06, | |
| "loss": 4.8415, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.9999999999999997e-06, | |
| "loss": 4.7132, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.9999999999999997e-06, | |
| "loss": 4.8118, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6e-06, | |
| "loss": 4.7937, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.2e-06, | |
| "loss": 4.8404, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8e-06, | |
| "loss": 4.7309, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.399999999999999e-06, | |
| "loss": 4.5415, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.999999999999999e-06, | |
| "loss": 4.6528, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.599999999999999e-06, | |
| "loss": 4.662, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.2e-06, | |
| "loss": 4.5529, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.799999999999998e-06, | |
| "loss": 4.474, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.4e-06, | |
| "loss": 4.4883, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.999999999999999e-06, | |
| "loss": 4.4004, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.6e-06, | |
| "loss": 4.4632, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.02e-05, | |
| "loss": 4.4696, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.0799999999999998e-05, | |
| "loss": 4.4129, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.14e-05, | |
| "loss": 4.4643, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.1999999999999999e-05, | |
| "loss": 4.4646, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.26e-05, | |
| "loss": 4.4311, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.3199999999999997e-05, | |
| "loss": 4.306, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.3799999999999998e-05, | |
| "loss": 4.4385, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.44e-05, | |
| "loss": 4.276, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.4999999999999999e-05, | |
| "loss": 4.5521, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.5599999999999996e-05, | |
| "loss": 4.3171, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.6199999999999997e-05, | |
| "loss": 4.2827, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.68e-05, | |
| "loss": 4.1956, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.74e-05, | |
| "loss": 4.3854, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.7999999999999997e-05, | |
| "loss": 4.3831, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.8599999999999998e-05, | |
| "loss": 4.2882, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.92e-05, | |
| "loss": 4.3278, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.98e-05, | |
| "loss": 4.2851, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.04e-05, | |
| "loss": 4.2957, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.1e-05, | |
| "loss": 4.2894, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.1599999999999996e-05, | |
| "loss": 4.3836, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.2199999999999998e-05, | |
| "loss": 4.3498, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.28e-05, | |
| "loss": 4.3839, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.34e-05, | |
| "loss": 4.2473, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.3999999999999997e-05, | |
| "loss": 4.3261, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 4.3655, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.52e-05, | |
| "loss": 4.1703, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.5799999999999997e-05, | |
| "loss": 4.468, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.6399999999999995e-05, | |
| "loss": 4.4049, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.6999999999999996e-05, | |
| "loss": 4.4439, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.7599999999999997e-05, | |
| "loss": 4.4014, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 4.5049, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.88e-05, | |
| "loss": 4.5962, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.94e-05, | |
| "loss": 4.3296, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.9999999999999997e-05, | |
| "loss": 4.2871, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.06e-05, | |
| "loss": 4.1845, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.119999999999999e-05, | |
| "loss": 4.163, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.1799999999999994e-05, | |
| "loss": 4.1515, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.2399999999999995e-05, | |
| "loss": 4.1247, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.2999999999999996e-05, | |
| "loss": 4.1402, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.36e-05, | |
| "loss": 4.2412, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.42e-05, | |
| "loss": 4.2467, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.48e-05, | |
| "loss": 4.2223, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.539999999999999e-05, | |
| "loss": 4.1693, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.5999999999999994e-05, | |
| "loss": 4.2163, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.6599999999999995e-05, | |
| "loss": 4.1301, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.7199999999999996e-05, | |
| "loss": 4.2289, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.78e-05, | |
| "loss": 4.1746, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.84e-05, | |
| "loss": 4.2473, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.9e-05, | |
| "loss": 4.2665, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.96e-05, | |
| "loss": 4.2199, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.02e-05, | |
| "loss": 4.246, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.08e-05, | |
| "loss": 4.1884, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.14e-05, | |
| "loss": 4.1402, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.2e-05, | |
| "loss": 4.1451, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.259999999999999e-05, | |
| "loss": 4.2652, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.319999999999999e-05, | |
| "loss": 4.1772, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.3799999999999994e-05, | |
| "loss": 4.1788, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.4399999999999995e-05, | |
| "loss": 4.2676, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.4999999999999996e-05, | |
| "loss": 4.1542, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.56e-05, | |
| "loss": 4.1175, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.62e-05, | |
| "loss": 4.1771, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.68e-05, | |
| "loss": 4.0612, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7399999999999993e-05, | |
| "loss": 4.2617, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7999999999999994e-05, | |
| "loss": 4.1813, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8599999999999995e-05, | |
| "loss": 4.1929, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.9199999999999997e-05, | |
| "loss": 4.1566, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.98e-05, | |
| "loss": 4.2066, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.04e-05, | |
| "loss": 4.297, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.1e-05, | |
| "loss": 4.1838, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.1599999999999994e-05, | |
| "loss": 4.2046, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.2199999999999995e-05, | |
| "loss": 4.2594, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.279999999999999e-05, | |
| "loss": 4.2559, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.339999999999999e-05, | |
| "loss": 4.2763, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.399999999999999e-05, | |
| "loss": 4.2853, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.459999999999999e-05, | |
| "loss": 4.3723, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.519999999999999e-05, | |
| "loss": 4.333, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.5799999999999994e-05, | |
| "loss": 4.2612, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.6399999999999995e-05, | |
| "loss": 4.2223, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.6999999999999996e-05, | |
| "loss": 4.3846, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.76e-05, | |
| "loss": 4.5903, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.82e-05, | |
| "loss": 4.4963, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.88e-05, | |
| "loss": 4.5083, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.94e-05, | |
| "loss": 4.5454, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 4.146, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 4.2164, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.12e-05, | |
| "loss": 4.2249, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.18e-05, | |
| "loss": 4.1094, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.239999999999999e-05, | |
| "loss": 4.2254, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.299999999999999e-05, | |
| "loss": 4.1199, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.359999999999999e-05, | |
| "loss": 4.0813, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.419999999999999e-05, | |
| "loss": 4.1843, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.479999999999999e-05, | |
| "loss": 4.2144, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.539999999999999e-05, | |
| "loss": 4.1853, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.599999999999999e-05, | |
| "loss": 4.256, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.659999999999999e-05, | |
| "loss": 4.1764, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.72e-05, | |
| "loss": 4.1245, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.78e-05, | |
| "loss": 4.1798, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.84e-05, | |
| "loss": 4.1431, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.9e-05, | |
| "loss": 4.1608, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.96e-05, | |
| "loss": 4.09, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.02e-05, | |
| "loss": 4.1146, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.079999999999999e-05, | |
| "loss": 4.1764, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.139999999999999e-05, | |
| "loss": 4.181, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.199999999999999e-05, | |
| "loss": 4.1816, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.259999999999999e-05, | |
| "loss": 4.2696, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.319999999999999e-05, | |
| "loss": 4.1079, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.379999999999999e-05, | |
| "loss": 4.2193, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.439999999999999e-05, | |
| "loss": 4.2735, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.5e-05, | |
| "loss": 4.1796, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.56e-05, | |
| "loss": 4.2039, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.62e-05, | |
| "loss": 4.1144, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.68e-05, | |
| "loss": 4.2619, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.74e-05, | |
| "loss": 4.3677, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.8e-05, | |
| "loss": 4.2429, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.86e-05, | |
| "loss": 4.1074, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.92e-05, | |
| "loss": 4.2802, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.98e-05, | |
| "loss": 4.1459, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.04e-05, | |
| "loss": 4.3063, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.1e-05, | |
| "loss": 4.2316, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.16e-05, | |
| "loss": 4.1519, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.22e-05, | |
| "loss": 4.0775, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.28e-05, | |
| "loss": 4.1537, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.34e-05, | |
| "loss": 4.1356, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.4e-05, | |
| "loss": 4.3637, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.459999999999998e-05, | |
| "loss": 4.156, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.519999999999998e-05, | |
| "loss": 4.2406, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.579999999999998e-05, | |
| "loss": 4.3063, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.639999999999999e-05, | |
| "loss": 4.4181, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.699999999999999e-05, | |
| "loss": 4.173, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.759999999999999e-05, | |
| "loss": 4.4079, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.819999999999999e-05, | |
| "loss": 4.4766, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.879999999999999e-05, | |
| "loss": 4.4345, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.939999999999999e-05, | |
| "loss": 4.5567, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.999999999999999e-05, | |
| "loss": 4.2932, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.059999999999999e-05, | |
| "loss": 4.2495, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.12e-05, | |
| "loss": 4.2386, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.18e-05, | |
| "loss": 4.1652, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.24e-05, | |
| "loss": 4.1926, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.3e-05, | |
| "loss": 4.2151, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.36e-05, | |
| "loss": 4.1392, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.419999999999999e-05, | |
| "loss": 4.165, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.479999999999999e-05, | |
| "loss": 4.2136, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.539999999999999e-05, | |
| "loss": 4.1931, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.599999999999999e-05, | |
| "loss": 4.2008, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.659999999999999e-05, | |
| "loss": 4.1198, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.719999999999999e-05, | |
| "loss": 4.2464, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.779999999999999e-05, | |
| "loss": 4.1406, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.839999999999999e-05, | |
| "loss": 4.2095, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.9e-05, | |
| "loss": 4.1875, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.96e-05, | |
| "loss": 4.152, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0001002, | |
| "loss": 4.1881, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0001008, | |
| "loss": 4.2114, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0001014, | |
| "loss": 4.1061, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.000102, | |
| "loss": 4.1188, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0001026, | |
| "loss": 4.1931, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00010319999999999999, | |
| "loss": 4.0806, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00010379999999999999, | |
| "loss": 4.1139, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00010439999999999999, | |
| "loss": 4.3033, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00010499999999999999, | |
| "loss": 4.1447, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00010559999999999998, | |
| "loss": 4.2387, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00010619999999999998, | |
| "loss": 4.1149, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00010679999999999998, | |
| "loss": 4.1095, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00010739999999999998, | |
| "loss": 4.3137, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00010799999999999998, | |
| "loss": 4.1593, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00010859999999999998, | |
| "loss": 4.0903, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00010919999999999998, | |
| "loss": 4.1328, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00010979999999999999, | |
| "loss": 4.1959, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00011039999999999999, | |
| "loss": 4.2055, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00011099999999999999, | |
| "loss": 4.1536, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00011159999999999999, | |
| "loss": 4.195, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00011219999999999999, | |
| "loss": 4.1625, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00011279999999999999, | |
| "loss": 4.1326, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00011339999999999999, | |
| "loss": 4.2294, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00011399999999999999, | |
| "loss": 4.3466, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001146, | |
| "loss": 4.3242, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001152, | |
| "loss": 4.3581, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001158, | |
| "loss": 4.3301, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0001164, | |
| "loss": 4.2756, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000117, | |
| "loss": 4.3115, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0001176, | |
| "loss": 4.1626, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001182, | |
| "loss": 4.6198, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001188, | |
| "loss": 4.4561, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001194, | |
| "loss": 4.4628, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 4.2635, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00012059999999999999, | |
| "loss": 4.2424, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00012119999999999999, | |
| "loss": 4.1895, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00012179999999999999, | |
| "loss": 4.133, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0001224, | |
| "loss": 4.1853, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00012299999999999998, | |
| "loss": 4.1186, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0001236, | |
| "loss": 4.3218, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00012419999999999998, | |
| "loss": 4.19, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00012479999999999997, | |
| "loss": 4.1701, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00012539999999999999, | |
| "loss": 4.2564, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00012599999999999997, | |
| "loss": 4.0469, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0001266, | |
| "loss": 4.1559, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00012719999999999997, | |
| "loss": 4.3674, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0001278, | |
| "loss": 4.172, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00012839999999999998, | |
| "loss": 4.2285, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.000129, | |
| "loss": 4.182, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00012959999999999998, | |
| "loss": 4.1601, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001302, | |
| "loss": 4.0804, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00013079999999999998, | |
| "loss": 4.0997, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0001314, | |
| "loss": 4.1931, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00013199999999999998, | |
| "loss": 4.1306, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001326, | |
| "loss": 4.0679, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00013319999999999999, | |
| "loss": 4.1208, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001338, | |
| "loss": 4.1506, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0001344, | |
| "loss": 4.1266, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.000135, | |
| "loss": 4.2373, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0001356, | |
| "loss": 4.1706, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0001362, | |
| "loss": 4.063, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0001368, | |
| "loss": 4.1355, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0001374, | |
| "loss": 4.1105, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.000138, | |
| "loss": 4.2918, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001386, | |
| "loss": 4.1313, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001392, | |
| "loss": 4.2142, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00013979999999999998, | |
| "loss": 4.2723, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0001404, | |
| "loss": 4.24, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00014099999999999998, | |
| "loss": 4.0821, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00014159999999999997, | |
| "loss": 4.177, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001422, | |
| "loss": 4.2761, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00014279999999999997, | |
| "loss": 4.1533, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0001434, | |
| "loss": 4.1522, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00014399999999999998, | |
| "loss": 4.3344, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0001446, | |
| "loss": 4.3554, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00014519999999999998, | |
| "loss": 4.1236, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0001458, | |
| "loss": 4.1599, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00014639999999999998, | |
| "loss": 4.1697, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.000147, | |
| "loss": 4.2965, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00014759999999999998, | |
| "loss": 4.3114, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0001482, | |
| "loss": 4.2331, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00014879999999999998, | |
| "loss": 4.4326, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0001494, | |
| "loss": 4.1969, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00015, | |
| "loss": 4.3411, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00015059999999999997, | |
| "loss": 4.2298, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0001512, | |
| "loss": 4.2068, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00015179999999999998, | |
| "loss": 4.1757, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001524, | |
| "loss": 4.1387, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00015299999999999998, | |
| "loss": 4.1108, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001536, | |
| "loss": 4.3753, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00015419999999999998, | |
| "loss": 4.1422, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0001548, | |
| "loss": 4.2031, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00015539999999999998, | |
| "loss": 4.1135, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.000156, | |
| "loss": 4.2013, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00015659999999999998, | |
| "loss": 4.2562, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001572, | |
| "loss": 4.1273, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0001578, | |
| "loss": 4.1866, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0001584, | |
| "loss": 4.1688, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.000159, | |
| "loss": 4.3492, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001596, | |
| "loss": 4.1599, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001602, | |
| "loss": 4.1327, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001608, | |
| "loss": 4.0783, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0001614, | |
| "loss": 4.2492, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.000162, | |
| "loss": 4.0783, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0001626, | |
| "loss": 4.0342, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0001632, | |
| "loss": 4.1312, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0001638, | |
| "loss": 4.1231, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0001644, | |
| "loss": 4.1969, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.000165, | |
| "loss": 4.1583, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0001656, | |
| "loss": 4.161, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0001662, | |
| "loss": 4.0859, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001668, | |
| "loss": 4.1179, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001674, | |
| "loss": 4.1002, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.000168, | |
| "loss": 4.2117, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0001686, | |
| "loss": 4.0938, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00016919999999999997, | |
| "loss": 4.2846, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00016979999999999998, | |
| "loss": 4.1019, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00017039999999999997, | |
| "loss": 4.2145, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00017099999999999998, | |
| "loss": 4.1841, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00017159999999999997, | |
| "loss": 4.1436, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00017219999999999998, | |
| "loss": 4.178, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00017279999999999997, | |
| "loss": 4.1402, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00017339999999999996, | |
| "loss": 4.2526, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00017399999999999997, | |
| "loss": 4.2094, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00017459999999999996, | |
| "loss": 4.3642, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00017519999999999998, | |
| "loss": 15.0924, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00017579999999999996, | |
| "loss": 6.5611, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00017639999999999998, | |
| "loss": 4.5139, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00017699999999999997, | |
| "loss": 4.8705, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00017759999999999998, | |
| "loss": 4.5291, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 297, | |
| "total_flos": 0.0, | |
| "train_loss": 4.303745459225844, | |
| "train_runtime": 4795.9642, | |
| "train_samples_per_second": 5.95, | |
| "train_steps_per_second": 0.062 | |
| } | |
| ], | |
| "max_steps": 297, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |