|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.0, |
|
"eval_steps": 500, |
|
"global_step": 620, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.993548387096775e-05, |
|
"loss": 1.716, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.987096774193549e-05, |
|
"loss": 1.7745, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.980645161290323e-05, |
|
"loss": 1.6054, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.974193548387097e-05, |
|
"loss": 1.6363, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.967741935483872e-05, |
|
"loss": 1.5976, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.961290322580645e-05, |
|
"loss": 1.7749, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.95483870967742e-05, |
|
"loss": 1.5595, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.948387096774195e-05, |
|
"loss": 1.6096, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.941935483870968e-05, |
|
"loss": 1.6574, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.935483870967742e-05, |
|
"loss": 1.617, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.929032258064517e-05, |
|
"loss": 1.7268, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.922580645161291e-05, |
|
"loss": 1.5113, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.916129032258064e-05, |
|
"loss": 1.7497, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.90967741935484e-05, |
|
"loss": 1.5716, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.903225806451614e-05, |
|
"loss": 1.4849, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.896774193548386e-05, |
|
"loss": 1.4985, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.890322580645162e-05, |
|
"loss": 1.5288, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.883870967741936e-05, |
|
"loss": 1.5506, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.87741935483871e-05, |
|
"loss": 1.6588, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.870967741935484e-05, |
|
"loss": 1.6245, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.864516129032258e-05, |
|
"loss": 1.4441, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.858064516129033e-05, |
|
"loss": 1.5697, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.851612903225807e-05, |
|
"loss": 1.4353, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.845161290322581e-05, |
|
"loss": 1.3954, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.838709677419355e-05, |
|
"loss": 1.6193, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.83225806451613e-05, |
|
"loss": 1.5966, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.825806451612903e-05, |
|
"loss": 1.5386, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.819354838709677e-05, |
|
"loss": 1.4932, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.812903225806453e-05, |
|
"loss": 1.5357, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.806451612903226e-05, |
|
"loss": 1.5584, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.8e-05, |
|
"loss": 1.5107, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.793548387096775e-05, |
|
"loss": 1.4353, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.787096774193549e-05, |
|
"loss": 1.4709, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.780645161290322e-05, |
|
"loss": 1.4193, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.774193548387098e-05, |
|
"loss": 1.3577, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.774193548387098e-05, |
|
"loss": 1.7503, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.767741935483872e-05, |
|
"loss": 1.652, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.761290322580646e-05, |
|
"loss": 1.577, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.75483870967742e-05, |
|
"loss": 1.4753, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.748387096774194e-05, |
|
"loss": 1.5025, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.741935483870968e-05, |
|
"loss": 1.5508, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.735483870967742e-05, |
|
"loss": 1.3235, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.729032258064516e-05, |
|
"loss": 1.3986, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.72258064516129e-05, |
|
"loss": 1.4616, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.716129032258065e-05, |
|
"loss": 1.5728, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.709677419354839e-05, |
|
"loss": 1.5358, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.703225806451613e-05, |
|
"loss": 1.4407, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.696774193548388e-05, |
|
"loss": 1.3705, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.690322580645161e-05, |
|
"loss": 1.44, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.683870967741935e-05, |
|
"loss": 1.4533, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.677419354838711e-05, |
|
"loss": 1.5213, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.670967741935484e-05, |
|
"loss": 1.4516, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.664516129032259e-05, |
|
"loss": 1.3548, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.658064516129033e-05, |
|
"loss": 1.4199, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.651612903225807e-05, |
|
"loss": 1.4489, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.645161290322581e-05, |
|
"loss": 1.4327, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.638709677419355e-05, |
|
"loss": 1.4834, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.63225806451613e-05, |
|
"loss": 1.4057, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.625806451612904e-05, |
|
"loss": 1.4227, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.619354838709678e-05, |
|
"loss": 1.3265, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.612903225806452e-05, |
|
"loss": 1.4446, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.606451612903226e-05, |
|
"loss": 1.4619, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.6e-05, |
|
"loss": 1.3791, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.593548387096774e-05, |
|
"loss": 1.4303, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.587096774193548e-05, |
|
"loss": 1.4665, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.580645161290323e-05, |
|
"loss": 1.2667, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.574193548387097e-05, |
|
"loss": 1.4362, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.567741935483872e-05, |
|
"loss": 1.4245, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.561290322580646e-05, |
|
"loss": 1.3996, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.554838709677419e-05, |
|
"loss": 1.3897, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.554838709677419e-05, |
|
"loss": 1.3155, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.548387096774195e-05, |
|
"loss": 1.4341, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.541935483870969e-05, |
|
"loss": 1.3562, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.535483870967741e-05, |
|
"loss": 1.5022, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.529032258064517e-05, |
|
"loss": 1.4112, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.522580645161291e-05, |
|
"loss": 1.4515, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.516129032258065e-05, |
|
"loss": 1.2948, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.50967741935484e-05, |
|
"loss": 1.3857, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.503225806451613e-05, |
|
"loss": 1.3419, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 9.496774193548388e-05, |
|
"loss": 1.3807, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 9.490322580645162e-05, |
|
"loss": 1.3096, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 9.483870967741936e-05, |
|
"loss": 1.3366, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.47741935483871e-05, |
|
"loss": 1.3849, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.470967741935484e-05, |
|
"loss": 1.3757, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.464516129032258e-05, |
|
"loss": 1.4037, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 9.458064516129032e-05, |
|
"loss": 1.4029, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 9.451612903225808e-05, |
|
"loss": 1.2547, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 9.44516129032258e-05, |
|
"loss": 1.3634, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.438709677419355e-05, |
|
"loss": 1.339, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.43225806451613e-05, |
|
"loss": 1.4094, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.425806451612904e-05, |
|
"loss": 1.4527, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.419354838709677e-05, |
|
"loss": 1.4851, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.412903225806453e-05, |
|
"loss": 1.291, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.406451612903227e-05, |
|
"loss": 1.2541, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 9.4e-05, |
|
"loss": 1.4211, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 9.393548387096775e-05, |
|
"loss": 1.3906, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 9.387096774193549e-05, |
|
"loss": 1.2921, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.380645161290323e-05, |
|
"loss": 1.5347, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.374193548387097e-05, |
|
"loss": 1.2438, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.367741935483871e-05, |
|
"loss": 1.3963, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 9.361290322580646e-05, |
|
"loss": 1.5133, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 9.35483870967742e-05, |
|
"loss": 1.4138, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 9.348387096774194e-05, |
|
"loss": 1.2888, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.341935483870968e-05, |
|
"loss": 1.4008, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.335483870967743e-05, |
|
"loss": 1.3859, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.329032258064516e-05, |
|
"loss": 1.4192, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.32258064516129e-05, |
|
"loss": 1.3406, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.316129032258066e-05, |
|
"loss": 1.467, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.309677419354839e-05, |
|
"loss": 1.2791, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.303225806451613e-05, |
|
"loss": 1.3281, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.296774193548388e-05, |
|
"loss": 1.4177, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.290322580645162e-05, |
|
"loss": 1.3119, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.283870967741935e-05, |
|
"loss": 1.5024, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.27741935483871e-05, |
|
"loss": 1.3078, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.270967741935485e-05, |
|
"loss": 1.5213, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.264516129032257e-05, |
|
"loss": 1.4349, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.258064516129033e-05, |
|
"loss": 1.4269, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.251612903225807e-05, |
|
"loss": 1.3975, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.245161290322581e-05, |
|
"loss": 1.4177, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.238709677419355e-05, |
|
"loss": 1.365, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.23225806451613e-05, |
|
"loss": 1.3248, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.225806451612904e-05, |
|
"loss": 1.4621, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.219354838709678e-05, |
|
"loss": 1.472, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.212903225806452e-05, |
|
"loss": 1.1869, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.206451612903226e-05, |
|
"loss": 1.3621, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.200000000000001e-05, |
|
"loss": 1.4642, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.193548387096774e-05, |
|
"loss": 1.4772, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.187096774193548e-05, |
|
"loss": 1.3086, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.180645161290324e-05, |
|
"loss": 1.6788, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.174193548387097e-05, |
|
"loss": 1.3363, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.16774193548387e-05, |
|
"loss": 1.2904, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.161290322580646e-05, |
|
"loss": 1.3506, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.15483870967742e-05, |
|
"loss": 1.3903, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.148387096774193e-05, |
|
"loss": 1.4298, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.141935483870968e-05, |
|
"loss": 1.5464, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.135483870967743e-05, |
|
"loss": 1.4275, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.129032258064517e-05, |
|
"loss": 1.4347, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.122580645161291e-05, |
|
"loss": 1.4139, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.116129032258065e-05, |
|
"loss": 1.4364, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.109677419354839e-05, |
|
"loss": 1.3984, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.103225806451613e-05, |
|
"loss": 1.2279, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 9.096774193548387e-05, |
|
"loss": 1.325, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 9.090322580645161e-05, |
|
"loss": 1.2434, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 9.083870967741937e-05, |
|
"loss": 1.2724, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.07741935483871e-05, |
|
"loss": 1.3751, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.070967741935484e-05, |
|
"loss": 1.3238, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.06451612903226e-05, |
|
"loss": 1.3634, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.058064516129032e-05, |
|
"loss": 1.3366, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.051612903225806e-05, |
|
"loss": 1.4424, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.045161290322582e-05, |
|
"loss": 1.4755, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.038709677419356e-05, |
|
"loss": 1.4592, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.032258064516129e-05, |
|
"loss": 1.2932, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.025806451612904e-05, |
|
"loss": 1.1448, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.019354838709678e-05, |
|
"loss": 1.3369, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.012903225806451e-05, |
|
"loss": 1.1981, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.006451612903226e-05, |
|
"loss": 1.3879, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 9e-05, |
|
"loss": 1.2789, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 8.993548387096775e-05, |
|
"loss": 1.3595, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 8.987096774193549e-05, |
|
"loss": 1.2933, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 8.980645161290323e-05, |
|
"loss": 1.5589, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 8.974193548387097e-05, |
|
"loss": 1.5362, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 8.967741935483871e-05, |
|
"loss": 1.4832, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 8.961290322580645e-05, |
|
"loss": 1.3837, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 8.95483870967742e-05, |
|
"loss": 1.3511, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 8.948387096774195e-05, |
|
"loss": 1.4746, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 8.941935483870968e-05, |
|
"loss": 1.4352, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 8.935483870967742e-05, |
|
"loss": 1.4561, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 8.929032258064517e-05, |
|
"loss": 1.462, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 8.92258064516129e-05, |
|
"loss": 1.3375, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 8.916129032258064e-05, |
|
"loss": 1.3243, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 8.90967741935484e-05, |
|
"loss": 1.3681, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 8.903225806451614e-05, |
|
"loss": 1.3448, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.896774193548388e-05, |
|
"loss": 1.4693, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.890322580645162e-05, |
|
"loss": 1.4128, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.883870967741936e-05, |
|
"loss": 1.3185, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.87741935483871e-05, |
|
"loss": 1.3531, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.870967741935484e-05, |
|
"loss": 1.4634, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.864516129032259e-05, |
|
"loss": 1.456, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.858064516129033e-05, |
|
"loss": 1.3668, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.851612903225807e-05, |
|
"loss": 1.3219, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.845161290322581e-05, |
|
"loss": 1.2563, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.838709677419355e-05, |
|
"loss": 1.3552, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.832258064516129e-05, |
|
"loss": 1.2516, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.825806451612903e-05, |
|
"loss": 1.3944, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 8.819354838709677e-05, |
|
"loss": 1.4108, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 8.812903225806453e-05, |
|
"loss": 1.2879, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 8.806451612903226e-05, |
|
"loss": 1.3993, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.800000000000001e-05, |
|
"loss": 1.5187, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.793548387096775e-05, |
|
"loss": 1.3581, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.787096774193548e-05, |
|
"loss": 1.494, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 8.780645161290323e-05, |
|
"loss": 1.4936, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 8.774193548387098e-05, |
|
"loss": 1.3858, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 8.767741935483872e-05, |
|
"loss": 1.2974, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.761290322580646e-05, |
|
"loss": 1.266, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.75483870967742e-05, |
|
"loss": 1.3635, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.748387096774194e-05, |
|
"loss": 1.4649, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 8.741935483870968e-05, |
|
"loss": 1.2769, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 8.735483870967742e-05, |
|
"loss": 1.3293, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 8.729032258064516e-05, |
|
"loss": 1.2614, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 8.72258064516129e-05, |
|
"loss": 1.2375, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 8.716129032258065e-05, |
|
"loss": 1.464, |
|
"step": 201 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 8.709677419354839e-05, |
|
"loss": 1.3782, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 8.703225806451614e-05, |
|
"loss": 1.4477, |
|
"step": 203 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 8.696774193548387e-05, |
|
"loss": 1.2875, |
|
"step": 204 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 8.690322580645161e-05, |
|
"loss": 1.3477, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 8.683870967741937e-05, |
|
"loss": 1.4381, |
|
"step": 206 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 8.677419354838711e-05, |
|
"loss": 1.3831, |
|
"step": 207 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 8.670967741935484e-05, |
|
"loss": 1.2368, |
|
"step": 208 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 8.664516129032259e-05, |
|
"loss": 1.3584, |
|
"step": 209 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 8.658064516129033e-05, |
|
"loss": 1.4164, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 8.651612903225806e-05, |
|
"loss": 1.2897, |
|
"step": 211 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 8.645161290322581e-05, |
|
"loss": 1.3487, |
|
"step": 212 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 8.638709677419356e-05, |
|
"loss": 1.386, |
|
"step": 213 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 8.63225806451613e-05, |
|
"loss": 1.4164, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 8.625806451612904e-05, |
|
"loss": 1.5031, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 8.619354838709678e-05, |
|
"loss": 1.3228, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 8.612903225806452e-05, |
|
"loss": 1.3883, |
|
"step": 217 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 8.606451612903226e-05, |
|
"loss": 1.5248, |
|
"step": 218 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 8.6e-05, |
|
"loss": 1.3222, |
|
"step": 219 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 8.593548387096774e-05, |
|
"loss": 1.3043, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 8.58709677419355e-05, |
|
"loss": 1.3316, |
|
"step": 221 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 8.580645161290323e-05, |
|
"loss": 1.4733, |
|
"step": 222 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 8.574193548387097e-05, |
|
"loss": 1.3395, |
|
"step": 223 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 8.567741935483872e-05, |
|
"loss": 1.2229, |
|
"step": 224 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 8.561290322580645e-05, |
|
"loss": 1.4358, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 8.554838709677419e-05, |
|
"loss": 1.2766, |
|
"step": 226 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 8.548387096774195e-05, |
|
"loss": 1.4023, |
|
"step": 227 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 8.541935483870969e-05, |
|
"loss": 1.3758, |
|
"step": 228 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 8.535483870967742e-05, |
|
"loss": 1.4499, |
|
"step": 229 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 8.529032258064517e-05, |
|
"loss": 1.2924, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.522580645161291e-05, |
|
"loss": 1.3566, |
|
"step": 231 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.516129032258064e-05, |
|
"loss": 1.3357, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.50967741935484e-05, |
|
"loss": 1.3551, |
|
"step": 233 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.503225806451614e-05, |
|
"loss": 1.4122, |
|
"step": 234 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.496774193548388e-05, |
|
"loss": 1.4719, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.490322580645162e-05, |
|
"loss": 1.4777, |
|
"step": 236 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.483870967741936e-05, |
|
"loss": 1.6068, |
|
"step": 237 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 8.47741935483871e-05, |
|
"loss": 1.2998, |
|
"step": 238 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 8.470967741935484e-05, |
|
"loss": 1.2984, |
|
"step": 239 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 8.464516129032258e-05, |
|
"loss": 1.3259, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 8.458064516129032e-05, |
|
"loss": 1.3474, |
|
"step": 241 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 8.451612903225808e-05, |
|
"loss": 1.2663, |
|
"step": 242 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 8.44516129032258e-05, |
|
"loss": 1.4719, |
|
"step": 243 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 8.438709677419355e-05, |
|
"loss": 1.2865, |
|
"step": 244 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 8.43225806451613e-05, |
|
"loss": 1.3435, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 8.425806451612903e-05, |
|
"loss": 1.4548, |
|
"step": 246 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 8.419354838709677e-05, |
|
"loss": 1.3205, |
|
"step": 247 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 8.412903225806453e-05, |
|
"loss": 1.3323, |
|
"step": 248 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 8.406451612903227e-05, |
|
"loss": 1.4428, |
|
"step": 249 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 8.4e-05, |
|
"loss": 1.2907, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 8.393548387096775e-05, |
|
"loss": 1.2152, |
|
"step": 251 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 8.387096774193549e-05, |
|
"loss": 1.3223, |
|
"step": 252 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 8.380645161290322e-05, |
|
"loss": 1.3879, |
|
"step": 253 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 8.374193548387097e-05, |
|
"loss": 1.2742, |
|
"step": 254 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 8.367741935483872e-05, |
|
"loss": 1.6644, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 8.361290322580646e-05, |
|
"loss": 1.3722, |
|
"step": 256 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 8.35483870967742e-05, |
|
"loss": 1.3397, |
|
"step": 257 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 8.348387096774194e-05, |
|
"loss": 1.4307, |
|
"step": 258 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 8.341935483870968e-05, |
|
"loss": 1.3314, |
|
"step": 259 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 8.335483870967742e-05, |
|
"loss": 1.2586, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 8.329032258064516e-05, |
|
"loss": 1.3073, |
|
"step": 261 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 8.32258064516129e-05, |
|
"loss": 1.3043, |
|
"step": 262 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 8.316129032258066e-05, |
|
"loss": 1.3565, |
|
"step": 263 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 8.309677419354839e-05, |
|
"loss": 1.386, |
|
"step": 264 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 8.303225806451613e-05, |
|
"loss": 1.3378, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 8.296774193548388e-05, |
|
"loss": 1.386, |
|
"step": 266 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 8.290322580645161e-05, |
|
"loss": 1.4444, |
|
"step": 267 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 8.283870967741935e-05, |
|
"loss": 1.3499, |
|
"step": 268 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.27741935483871e-05, |
|
"loss": 1.4361, |
|
"step": 269 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.270967741935485e-05, |
|
"loss": 1.37, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.264516129032257e-05, |
|
"loss": 1.3339, |
|
"step": 271 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.258064516129033e-05, |
|
"loss": 1.3722, |
|
"step": 272 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.251612903225807e-05, |
|
"loss": 1.4191, |
|
"step": 273 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.245161290322581e-05, |
|
"loss": 1.4213, |
|
"step": 274 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 8.238709677419355e-05, |
|
"loss": 1.2428, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 8.23225806451613e-05, |
|
"loss": 1.34, |
|
"step": 276 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 8.225806451612904e-05, |
|
"loss": 1.2769, |
|
"step": 277 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.219354838709678e-05, |
|
"loss": 1.1893, |
|
"step": 278 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.212903225806452e-05, |
|
"loss": 1.3477, |
|
"step": 279 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.206451612903226e-05, |
|
"loss": 1.4416, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.2e-05, |
|
"loss": 1.4063, |
|
"step": 281 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.193548387096774e-05, |
|
"loss": 1.322, |
|
"step": 282 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 8.187096774193548e-05, |
|
"loss": 1.358, |
|
"step": 283 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.180645161290324e-05, |
|
"loss": 1.4087, |
|
"step": 284 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.174193548387097e-05, |
|
"loss": 1.4503, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 8.167741935483871e-05, |
|
"loss": 1.4052, |
|
"step": 286 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.161290322580646e-05, |
|
"loss": 1.3996, |
|
"step": 287 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.154838709677419e-05, |
|
"loss": 1.3519, |
|
"step": 288 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 8.148387096774193e-05, |
|
"loss": 1.3025, |
|
"step": 289 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.141935483870969e-05, |
|
"loss": 1.3033, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.135483870967743e-05, |
|
"loss": 1.2895, |
|
"step": 291 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.129032258064517e-05, |
|
"loss": 1.351, |
|
"step": 292 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.122580645161291e-05, |
|
"loss": 1.2342, |
|
"step": 293 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.116129032258065e-05, |
|
"loss": 1.5016, |
|
"step": 294 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.109677419354839e-05, |
|
"loss": 1.3194, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 8.103225806451613e-05, |
|
"loss": 1.2655, |
|
"step": 296 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.096774193548387e-05, |
|
"loss": 1.3279, |
|
"step": 297 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.090322580645162e-05, |
|
"loss": 1.3496, |
|
"step": 298 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.083870967741936e-05, |
|
"loss": 1.3972, |
|
"step": 299 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.07741935483871e-05, |
|
"loss": 1.2868, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.070967741935484e-05, |
|
"loss": 1.2803, |
|
"step": 301 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 8.064516129032258e-05, |
|
"loss": 1.4244, |
|
"step": 302 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.058064516129032e-05, |
|
"loss": 1.2935, |
|
"step": 303 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.051612903225806e-05, |
|
"loss": 1.2726, |
|
"step": 304 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 8.045161290322582e-05, |
|
"loss": 1.29, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.038709677419355e-05, |
|
"loss": 1.3333, |
|
"step": 306 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.03225806451613e-05, |
|
"loss": 1.2197, |
|
"step": 307 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.025806451612904e-05, |
|
"loss": 1.4033, |
|
"step": 308 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.019354838709677e-05, |
|
"loss": 1.2889, |
|
"step": 309 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.012903225806452e-05, |
|
"loss": 1.2519, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.006451612903227e-05, |
|
"loss": 1.3255, |
|
"step": 311 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8e-05, |
|
"loss": 1.1336, |
|
"step": 312 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 7.993548387096775e-05, |
|
"loss": 1.1947, |
|
"step": 313 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 7.987096774193549e-05, |
|
"loss": 1.2247, |
|
"step": 314 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 7.980645161290323e-05, |
|
"loss": 1.3316, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 7.974193548387097e-05, |
|
"loss": 1.1266, |
|
"step": 316 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 7.967741935483871e-05, |
|
"loss": 1.0444, |
|
"step": 317 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 7.961290322580645e-05, |
|
"loss": 1.2242, |
|
"step": 318 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 7.95483870967742e-05, |
|
"loss": 1.1784, |
|
"step": 319 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 7.948387096774194e-05, |
|
"loss": 1.2946, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 7.941935483870968e-05, |
|
"loss": 1.3006, |
|
"step": 321 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 7.935483870967743e-05, |
|
"loss": 1.2457, |
|
"step": 322 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 7.929032258064516e-05, |
|
"loss": 1.2297, |
|
"step": 323 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 7.92258064516129e-05, |
|
"loss": 1.2912, |
|
"step": 324 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 7.916129032258066e-05, |
|
"loss": 1.2593, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 7.90967741935484e-05, |
|
"loss": 1.1773, |
|
"step": 326 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 7.903225806451613e-05, |
|
"loss": 1.2831, |
|
"step": 327 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 7.896774193548388e-05, |
|
"loss": 1.2368, |
|
"step": 328 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 7.890322580645162e-05, |
|
"loss": 1.2555, |
|
"step": 329 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 7.883870967741935e-05, |
|
"loss": 1.3443, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.87741935483871e-05, |
|
"loss": 1.0449, |
|
"step": 331 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.870967741935484e-05, |
|
"loss": 1.1819, |
|
"step": 332 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.864516129032259e-05, |
|
"loss": 1.3371, |
|
"step": 333 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.858064516129033e-05, |
|
"loss": 1.2173, |
|
"step": 334 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.851612903225807e-05, |
|
"loss": 1.341, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.845161290322581e-05, |
|
"loss": 1.1811, |
|
"step": 336 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.838709677419355e-05, |
|
"loss": 1.3238, |
|
"step": 337 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.832258064516129e-05, |
|
"loss": 1.4116, |
|
"step": 338 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.825806451612903e-05, |
|
"loss": 1.257, |
|
"step": 339 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.819354838709679e-05, |
|
"loss": 1.2128, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.812903225806452e-05, |
|
"loss": 1.2872, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.806451612903226e-05, |
|
"loss": 1.2865, |
|
"step": 342 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.800000000000001e-05, |
|
"loss": 1.2735, |
|
"step": 343 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.793548387096774e-05, |
|
"loss": 1.1877, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.787096774193548e-05, |
|
"loss": 1.2987, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.780645161290324e-05, |
|
"loss": 1.1791, |
|
"step": 346 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.774193548387098e-05, |
|
"loss": 1.2194, |
|
"step": 347 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.76774193548387e-05, |
|
"loss": 1.2898, |
|
"step": 348 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.761290322580646e-05, |
|
"loss": 1.2191, |
|
"step": 349 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.75483870967742e-05, |
|
"loss": 1.1902, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.748387096774193e-05, |
|
"loss": 1.2546, |
|
"step": 351 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 7.741935483870968e-05, |
|
"loss": 1.2232, |
|
"step": 352 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 7.735483870967742e-05, |
|
"loss": 1.1641, |
|
"step": 353 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 7.729032258064517e-05, |
|
"loss": 1.247, |
|
"step": 354 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.722580645161291e-05, |
|
"loss": 1.1752, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.716129032258065e-05, |
|
"loss": 1.1836, |
|
"step": 356 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.709677419354839e-05, |
|
"loss": 1.2564, |
|
"step": 357 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.703225806451613e-05, |
|
"loss": 1.2807, |
|
"step": 358 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 7.696774193548387e-05, |
|
"loss": 1.267, |
|
"step": 359 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 7.690322580645161e-05, |
|
"loss": 1.2612, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 7.683870967741937e-05, |
|
"loss": 1.2218, |
|
"step": 361 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 7.67741935483871e-05, |
|
"loss": 1.1365, |
|
"step": 362 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 7.670967741935484e-05, |
|
"loss": 1.0789, |
|
"step": 363 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 7.664516129032259e-05, |
|
"loss": 1.2635, |
|
"step": 364 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 7.658064516129032e-05, |
|
"loss": 1.22, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 7.651612903225806e-05, |
|
"loss": 1.1883, |
|
"step": 366 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 7.645161290322582e-05, |
|
"loss": 1.3253, |
|
"step": 367 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 7.638709677419356e-05, |
|
"loss": 1.3024, |
|
"step": 368 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 7.632258064516128e-05, |
|
"loss": 1.2581, |
|
"step": 369 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 7.625806451612904e-05, |
|
"loss": 0.972, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 7.619354838709678e-05, |
|
"loss": 1.2729, |
|
"step": 371 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 7.612903225806451e-05, |
|
"loss": 1.3093, |
|
"step": 372 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 7.606451612903226e-05, |
|
"loss": 1.1775, |
|
"step": 373 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 7.6e-05, |
|
"loss": 1.1554, |
|
"step": 374 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 7.593548387096775e-05, |
|
"loss": 1.2968, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 7.587096774193549e-05, |
|
"loss": 1.3013, |
|
"step": 376 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 7.580645161290323e-05, |
|
"loss": 1.32, |
|
"step": 377 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 7.574193548387097e-05, |
|
"loss": 1.1921, |
|
"step": 378 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 7.567741935483871e-05, |
|
"loss": 1.38, |
|
"step": 379 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 7.561290322580645e-05, |
|
"loss": 1.1201, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 7.554838709677419e-05, |
|
"loss": 1.2216, |
|
"step": 381 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 7.548387096774195e-05, |
|
"loss": 1.2189, |
|
"step": 382 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 7.541935483870968e-05, |
|
"loss": 1.3504, |
|
"step": 383 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 7.535483870967742e-05, |
|
"loss": 1.2688, |
|
"step": 384 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 7.529032258064517e-05, |
|
"loss": 1.1411, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 7.52258064516129e-05, |
|
"loss": 1.3266, |
|
"step": 386 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 7.516129032258064e-05, |
|
"loss": 1.2878, |
|
"step": 387 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 7.50967741935484e-05, |
|
"loss": 1.2512, |
|
"step": 388 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 7.503225806451614e-05, |
|
"loss": 1.1967, |
|
"step": 389 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 7.496774193548386e-05, |
|
"loss": 1.3149, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 7.490322580645162e-05, |
|
"loss": 1.2203, |
|
"step": 391 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 7.483870967741936e-05, |
|
"loss": 1.2826, |
|
"step": 392 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 7.47741935483871e-05, |
|
"loss": 1.2981, |
|
"step": 393 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 7.470967741935484e-05, |
|
"loss": 1.2771, |
|
"step": 394 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 7.464516129032258e-05, |
|
"loss": 1.2138, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 7.458064516129033e-05, |
|
"loss": 1.1726, |
|
"step": 396 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 7.451612903225807e-05, |
|
"loss": 1.2656, |
|
"step": 397 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 7.445161290322581e-05, |
|
"loss": 1.2628, |
|
"step": 398 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 7.438709677419355e-05, |
|
"loss": 1.2797, |
|
"step": 399 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 7.43225806451613e-05, |
|
"loss": 1.2491, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 7.425806451612903e-05, |
|
"loss": 1.2502, |
|
"step": 401 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 7.419354838709677e-05, |
|
"loss": 1.1875, |
|
"step": 402 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 7.412903225806453e-05, |
|
"loss": 1.2236, |
|
"step": 403 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 7.406451612903225e-05, |
|
"loss": 1.1255, |
|
"step": 404 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 7.4e-05, |
|
"loss": 1.2813, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 7.393548387096775e-05, |
|
"loss": 1.1283, |
|
"step": 406 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 7.387096774193549e-05, |
|
"loss": 1.2351, |
|
"step": 407 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 7.380645161290323e-05, |
|
"loss": 1.2723, |
|
"step": 408 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 7.374193548387097e-05, |
|
"loss": 1.2865, |
|
"step": 409 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 7.367741935483872e-05, |
|
"loss": 1.2589, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 7.361290322580646e-05, |
|
"loss": 1.2197, |
|
"step": 411 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 7.35483870967742e-05, |
|
"loss": 1.2036, |
|
"step": 412 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 7.348387096774194e-05, |
|
"loss": 1.1727, |
|
"step": 413 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 7.341935483870968e-05, |
|
"loss": 1.2835, |
|
"step": 414 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 7.335483870967742e-05, |
|
"loss": 1.1364, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 7.329032258064516e-05, |
|
"loss": 1.1284, |
|
"step": 416 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 7.32258064516129e-05, |
|
"loss": 1.1517, |
|
"step": 417 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 7.316129032258065e-05, |
|
"loss": 1.1535, |
|
"step": 418 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 7.309677419354839e-05, |
|
"loss": 1.1734, |
|
"step": 419 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 7.303225806451613e-05, |
|
"loss": 1.1108, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 7.296774193548388e-05, |
|
"loss": 1.2159, |
|
"step": 421 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 7.290322580645161e-05, |
|
"loss": 1.1729, |
|
"step": 422 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 7.283870967741937e-05, |
|
"loss": 1.1992, |
|
"step": 423 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 7.277419354838711e-05, |
|
"loss": 1.239, |
|
"step": 424 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 7.270967741935483e-05, |
|
"loss": 1.3547, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 7.264516129032259e-05, |
|
"loss": 1.3013, |
|
"step": 426 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 7.258064516129033e-05, |
|
"loss": 1.2471, |
|
"step": 427 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 7.251612903225807e-05, |
|
"loss": 1.2169, |
|
"step": 428 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 7.245161290322581e-05, |
|
"loss": 1.2501, |
|
"step": 429 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 7.238709677419355e-05, |
|
"loss": 1.1312, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 7.23225806451613e-05, |
|
"loss": 1.2524, |
|
"step": 431 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 7.225806451612904e-05, |
|
"loss": 1.3236, |
|
"step": 432 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 7.219354838709678e-05, |
|
"loss": 1.0989, |
|
"step": 433 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 7.212903225806452e-05, |
|
"loss": 1.2075, |
|
"step": 434 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 7.206451612903226e-05, |
|
"loss": 1.3255, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 7.2e-05, |
|
"loss": 1.1495, |
|
"step": 436 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 7.193548387096774e-05, |
|
"loss": 1.0892, |
|
"step": 437 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 7.187096774193548e-05, |
|
"loss": 1.2068, |
|
"step": 438 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 7.180645161290323e-05, |
|
"loss": 1.1091, |
|
"step": 439 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 7.174193548387097e-05, |
|
"loss": 1.0941, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 7.167741935483872e-05, |
|
"loss": 1.2172, |
|
"step": 441 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 7.161290322580646e-05, |
|
"loss": 1.3094, |
|
"step": 442 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 7.154838709677419e-05, |
|
"loss": 1.2521, |
|
"step": 443 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 7.148387096774195e-05, |
|
"loss": 1.1626, |
|
"step": 444 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 7.141935483870969e-05, |
|
"loss": 1.2176, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 7.135483870967741e-05, |
|
"loss": 1.1943, |
|
"step": 446 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 7.129032258064517e-05, |
|
"loss": 1.1103, |
|
"step": 447 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 7.122580645161291e-05, |
|
"loss": 1.2457, |
|
"step": 448 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 7.116129032258065e-05, |
|
"loss": 1.1279, |
|
"step": 449 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 7.109677419354839e-05, |
|
"loss": 1.217, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 7.103225806451613e-05, |
|
"loss": 1.2433, |
|
"step": 451 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 7.096774193548388e-05, |
|
"loss": 1.2683, |
|
"step": 452 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 7.090322580645162e-05, |
|
"loss": 1.0663, |
|
"step": 453 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 7.083870967741936e-05, |
|
"loss": 1.175, |
|
"step": 454 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 7.07741935483871e-05, |
|
"loss": 1.1702, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 7.070967741935485e-05, |
|
"loss": 1.2893, |
|
"step": 456 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 7.064516129032258e-05, |
|
"loss": 1.2822, |
|
"step": 457 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 7.058064516129032e-05, |
|
"loss": 1.012, |
|
"step": 458 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 7.051612903225808e-05, |
|
"loss": 1.2837, |
|
"step": 459 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 7.04516129032258e-05, |
|
"loss": 1.1838, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 7.038709677419355e-05, |
|
"loss": 1.1814, |
|
"step": 461 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 7.03225806451613e-05, |
|
"loss": 1.1655, |
|
"step": 462 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 7.025806451612904e-05, |
|
"loss": 1.1503, |
|
"step": 463 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 7.019354838709677e-05, |
|
"loss": 1.2196, |
|
"step": 464 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 7.012903225806452e-05, |
|
"loss": 1.2698, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 7.006451612903227e-05, |
|
"loss": 1.1131, |
|
"step": 466 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 7e-05, |
|
"loss": 1.2198, |
|
"step": 467 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 6.993548387096775e-05, |
|
"loss": 1.1703, |
|
"step": 468 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 6.987096774193549e-05, |
|
"loss": 1.1876, |
|
"step": 469 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 6.980645161290323e-05, |
|
"loss": 1.2142, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 6.974193548387097e-05, |
|
"loss": 1.1484, |
|
"step": 471 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 6.967741935483871e-05, |
|
"loss": 1.1529, |
|
"step": 472 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 6.961290322580645e-05, |
|
"loss": 1.4009, |
|
"step": 473 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 6.95483870967742e-05, |
|
"loss": 1.2733, |
|
"step": 474 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 6.948387096774194e-05, |
|
"loss": 1.2198, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 6.948387096774194e-05, |
|
"loss": 1.2224, |
|
"step": 476 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 6.941935483870968e-05, |
|
"loss": 1.0744, |
|
"step": 477 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 6.935483870967743e-05, |
|
"loss": 1.2484, |
|
"step": 478 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 6.929032258064516e-05, |
|
"loss": 1.2018, |
|
"step": 479 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 6.92258064516129e-05, |
|
"loss": 1.0806, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 6.916129032258066e-05, |
|
"loss": 1.1088, |
|
"step": 481 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 6.909677419354838e-05, |
|
"loss": 1.3505, |
|
"step": 482 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 6.903225806451613e-05, |
|
"loss": 1.113, |
|
"step": 483 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 6.896774193548388e-05, |
|
"loss": 1.3283, |
|
"step": 484 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 6.890322580645162e-05, |
|
"loss": 1.2982, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 6.883870967741935e-05, |
|
"loss": 1.307, |
|
"step": 486 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 6.87741935483871e-05, |
|
"loss": 1.1495, |
|
"step": 487 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 6.870967741935485e-05, |
|
"loss": 1.1041, |
|
"step": 488 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 6.864516129032257e-05, |
|
"loss": 1.1484, |
|
"step": 489 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 6.858064516129033e-05, |
|
"loss": 1.2887, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 6.851612903225807e-05, |
|
"loss": 1.1657, |
|
"step": 491 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 6.845161290322581e-05, |
|
"loss": 1.2411, |
|
"step": 492 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 6.838709677419355e-05, |
|
"loss": 1.1875, |
|
"step": 493 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 6.83225806451613e-05, |
|
"loss": 1.3288, |
|
"step": 494 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 6.825806451612903e-05, |
|
"loss": 1.0947, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 6.819354838709678e-05, |
|
"loss": 1.2137, |
|
"step": 496 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 6.812903225806452e-05, |
|
"loss": 1.2108, |
|
"step": 497 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 6.806451612903226e-05, |
|
"loss": 1.194, |
|
"step": 498 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 6.800000000000001e-05, |
|
"loss": 1.1875, |
|
"step": 499 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 6.793548387096774e-05, |
|
"loss": 1.325, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 6.787096774193548e-05, |
|
"loss": 1.1707, |
|
"step": 501 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 6.780645161290324e-05, |
|
"loss": 1.3017, |
|
"step": 502 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 6.774193548387096e-05, |
|
"loss": 1.4463, |
|
"step": 503 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 6.76774193548387e-05, |
|
"loss": 1.1761, |
|
"step": 504 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 6.761290322580646e-05, |
|
"loss": 1.3094, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 6.75483870967742e-05, |
|
"loss": 1.19, |
|
"step": 506 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 6.748387096774193e-05, |
|
"loss": 1.2318, |
|
"step": 507 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 6.741935483870968e-05, |
|
"loss": 1.2646, |
|
"step": 508 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 6.735483870967743e-05, |
|
"loss": 1.2207, |
|
"step": 509 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 6.729032258064515e-05, |
|
"loss": 1.1883, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 6.722580645161291e-05, |
|
"loss": 1.1763, |
|
"step": 511 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 6.716129032258065e-05, |
|
"loss": 1.2934, |
|
"step": 512 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 6.709677419354839e-05, |
|
"loss": 1.2894, |
|
"step": 513 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 6.703225806451613e-05, |
|
"loss": 1.2498, |
|
"step": 514 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 6.696774193548387e-05, |
|
"loss": 1.183, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 6.690322580645161e-05, |
|
"loss": 1.2735, |
|
"step": 516 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 6.683870967741936e-05, |
|
"loss": 1.1541, |
|
"step": 517 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 6.67741935483871e-05, |
|
"loss": 1.2153, |
|
"step": 518 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 6.670967741935484e-05, |
|
"loss": 1.2869, |
|
"step": 519 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 6.664516129032259e-05, |
|
"loss": 1.2448, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 6.658064516129032e-05, |
|
"loss": 1.1966, |
|
"step": 521 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 6.651612903225806e-05, |
|
"loss": 1.2238, |
|
"step": 522 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 6.645161290322582e-05, |
|
"loss": 1.2975, |
|
"step": 523 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 6.638709677419354e-05, |
|
"loss": 1.3073, |
|
"step": 524 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 6.632258064516129e-05, |
|
"loss": 1.1785, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 6.625806451612904e-05, |
|
"loss": 1.2413, |
|
"step": 526 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 6.619354838709678e-05, |
|
"loss": 1.2806, |
|
"step": 527 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 6.612903225806452e-05, |
|
"loss": 1.14, |
|
"step": 528 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 6.606451612903226e-05, |
|
"loss": 1.233, |
|
"step": 529 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 6.6e-05, |
|
"loss": 1.2962, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 6.593548387096775e-05, |
|
"loss": 1.0797, |
|
"step": 531 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 6.587096774193549e-05, |
|
"loss": 1.2569, |
|
"step": 532 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 6.580645161290323e-05, |
|
"loss": 1.1821, |
|
"step": 533 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 6.574193548387097e-05, |
|
"loss": 1.2141, |
|
"step": 534 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 6.567741935483871e-05, |
|
"loss": 1.2667, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 6.561290322580645e-05, |
|
"loss": 1.1583, |
|
"step": 536 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 6.55483870967742e-05, |
|
"loss": 1.2265, |
|
"step": 537 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 6.548387096774193e-05, |
|
"loss": 1.2142, |
|
"step": 538 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 6.541935483870968e-05, |
|
"loss": 1.2357, |
|
"step": 539 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 6.535483870967742e-05, |
|
"loss": 1.2936, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 6.529032258064517e-05, |
|
"loss": 1.1729, |
|
"step": 541 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 6.52258064516129e-05, |
|
"loss": 1.1928, |
|
"step": 542 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 6.516129032258065e-05, |
|
"loss": 1.3035, |
|
"step": 543 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 6.50967741935484e-05, |
|
"loss": 1.3385, |
|
"step": 544 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 6.503225806451612e-05, |
|
"loss": 1.3105, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 6.496774193548388e-05, |
|
"loss": 1.1637, |
|
"step": 546 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 6.490322580645162e-05, |
|
"loss": 1.1861, |
|
"step": 547 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 6.483870967741936e-05, |
|
"loss": 1.2912, |
|
"step": 548 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 6.47741935483871e-05, |
|
"loss": 1.2126, |
|
"step": 549 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 6.470967741935484e-05, |
|
"loss": 1.2942, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 6.464516129032258e-05, |
|
"loss": 1.2012, |
|
"step": 551 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 6.458064516129033e-05, |
|
"loss": 1.2668, |
|
"step": 552 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 6.451612903225807e-05, |
|
"loss": 1.2788, |
|
"step": 553 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 6.445161290322581e-05, |
|
"loss": 1.2599, |
|
"step": 554 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 6.438709677419355e-05, |
|
"loss": 1.0709, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 6.432258064516129e-05, |
|
"loss": 1.3085, |
|
"step": 556 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 6.425806451612903e-05, |
|
"loss": 1.2126, |
|
"step": 557 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 6.419354838709679e-05, |
|
"loss": 1.2078, |
|
"step": 558 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 6.412903225806451e-05, |
|
"loss": 1.1256, |
|
"step": 559 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 6.406451612903226e-05, |
|
"loss": 1.2807, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 6.400000000000001e-05, |
|
"loss": 1.2953, |
|
"step": 561 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 6.393548387096775e-05, |
|
"loss": 1.3366, |
|
"step": 562 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 6.387096774193548e-05, |
|
"loss": 1.1301, |
|
"step": 563 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 6.380645161290323e-05, |
|
"loss": 1.2603, |
|
"step": 564 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 6.374193548387098e-05, |
|
"loss": 1.2094, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 6.36774193548387e-05, |
|
"loss": 1.2262, |
|
"step": 566 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 6.361290322580646e-05, |
|
"loss": 1.1948, |
|
"step": 567 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 6.35483870967742e-05, |
|
"loss": 1.1709, |
|
"step": 568 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 6.348387096774194e-05, |
|
"loss": 1.3069, |
|
"step": 569 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 6.341935483870968e-05, |
|
"loss": 1.1746, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 6.335483870967742e-05, |
|
"loss": 1.1771, |
|
"step": 571 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 6.329032258064516e-05, |
|
"loss": 1.1965, |
|
"step": 572 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 6.32258064516129e-05, |
|
"loss": 1.2352, |
|
"step": 573 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 6.316129032258065e-05, |
|
"loss": 1.3588, |
|
"step": 574 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 6.309677419354839e-05, |
|
"loss": 1.3035, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 6.303225806451614e-05, |
|
"loss": 1.102, |
|
"step": 576 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 6.296774193548387e-05, |
|
"loss": 1.2424, |
|
"step": 577 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 6.290322580645161e-05, |
|
"loss": 1.1647, |
|
"step": 578 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 6.283870967741937e-05, |
|
"loss": 1.2188, |
|
"step": 579 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 6.27741935483871e-05, |
|
"loss": 1.3609, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 6.270967741935484e-05, |
|
"loss": 1.2879, |
|
"step": 581 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 6.264516129032259e-05, |
|
"loss": 1.3221, |
|
"step": 582 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 6.258064516129033e-05, |
|
"loss": 1.1337, |
|
"step": 583 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 6.251612903225806e-05, |
|
"loss": 1.1611, |
|
"step": 584 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 6.245161290322581e-05, |
|
"loss": 1.2829, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 6.238709677419356e-05, |
|
"loss": 1.2616, |
|
"step": 586 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 6.232258064516128e-05, |
|
"loss": 1.2378, |
|
"step": 587 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 6.225806451612904e-05, |
|
"loss": 1.2424, |
|
"step": 588 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 6.219354838709678e-05, |
|
"loss": 1.3422, |
|
"step": 589 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 6.212903225806452e-05, |
|
"loss": 1.1654, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 6.206451612903226e-05, |
|
"loss": 1.2674, |
|
"step": 591 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 6.2e-05, |
|
"loss": 1.0968, |
|
"step": 592 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 6.193548387096774e-05, |
|
"loss": 1.1989, |
|
"step": 593 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 6.187096774193549e-05, |
|
"loss": 1.1781, |
|
"step": 594 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 6.180645161290323e-05, |
|
"loss": 1.2104, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 6.174193548387097e-05, |
|
"loss": 1.2879, |
|
"step": 596 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 6.167741935483872e-05, |
|
"loss": 1.1927, |
|
"step": 597 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 6.161290322580645e-05, |
|
"loss": 1.2184, |
|
"step": 598 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 6.154838709677419e-05, |
|
"loss": 1.2234, |
|
"step": 599 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 6.148387096774195e-05, |
|
"loss": 1.1684, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 6.141935483870967e-05, |
|
"loss": 1.0678, |
|
"step": 601 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 6.135483870967742e-05, |
|
"loss": 1.1604, |
|
"step": 602 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 6.129032258064517e-05, |
|
"loss": 1.1536, |
|
"step": 603 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 6.122580645161291e-05, |
|
"loss": 1.2518, |
|
"step": 604 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 6.116129032258064e-05, |
|
"loss": 1.1394, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 6.10967741935484e-05, |
|
"loss": 1.238, |
|
"step": 606 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 6.1032258064516135e-05, |
|
"loss": 1.2727, |
|
"step": 607 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 6.096774193548387e-05, |
|
"loss": 1.2239, |
|
"step": 608 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 6.090322580645161e-05, |
|
"loss": 1.1393, |
|
"step": 609 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 6.083870967741936e-05, |
|
"loss": 1.1231, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 6.077419354838711e-05, |
|
"loss": 1.2431, |
|
"step": 611 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 6.070967741935484e-05, |
|
"loss": 1.1693, |
|
"step": 612 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 6.064516129032258e-05, |
|
"loss": 1.1998, |
|
"step": 613 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 6.058064516129033e-05, |
|
"loss": 1.2099, |
|
"step": 614 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 6.0516129032258065e-05, |
|
"loss": 1.1924, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 6.0451612903225806e-05, |
|
"loss": 1.1998, |
|
"step": 616 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 6.0387096774193554e-05, |
|
"loss": 1.2079, |
|
"step": 617 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 6.0322580645161295e-05, |
|
"loss": 1.2224, |
|
"step": 618 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 6.025806451612903e-05, |
|
"loss": 1.2026, |
|
"step": 619 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 6.019354838709678e-05, |
|
"loss": 1.1097, |
|
"step": 620 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 1550, |
|
"num_train_epochs": 5, |
|
"save_steps": 500, |
|
"total_flos": 2.880622039497431e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|