Heralax's picture
Upload 61 files
df0fa4a
raw
history blame contribute delete
No virus
37.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 310,
"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
}
],
"logging_steps": 1,
"max_steps": 1550,
"num_train_epochs": 5,
"save_steps": 500,
"total_flos": 1.4399449525818778e+18,
"trial_name": null,
"trial_params": null
}