exgra-med-dci-pathvqa / trainer_state.json
taindp98's picture
Add files using upload-large-folder tool
a08cb24 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 15.0,
"global_step": 2325,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 2.8571428571428575e-07,
"loss": 2.7313,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 5.714285714285715e-07,
"loss": 2.768,
"step": 2
},
{
"epoch": 0.02,
"learning_rate": 8.571428571428572e-07,
"loss": 2.4268,
"step": 3
},
{
"epoch": 0.03,
"learning_rate": 1.142857142857143e-06,
"loss": 2.1759,
"step": 4
},
{
"epoch": 0.03,
"learning_rate": 1.4285714285714286e-06,
"loss": 1.9597,
"step": 5
},
{
"epoch": 0.04,
"learning_rate": 1.7142857142857145e-06,
"loss": 1.6528,
"step": 6
},
{
"epoch": 0.05,
"learning_rate": 2.0000000000000003e-06,
"loss": 1.3127,
"step": 7
},
{
"epoch": 0.05,
"learning_rate": 2.285714285714286e-06,
"loss": 1.1806,
"step": 8
},
{
"epoch": 0.06,
"learning_rate": 2.571428571428571e-06,
"loss": 1.1056,
"step": 9
},
{
"epoch": 0.06,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.6871,
"step": 10
},
{
"epoch": 0.07,
"learning_rate": 3.142857142857143e-06,
"loss": 1.0462,
"step": 11
},
{
"epoch": 0.08,
"learning_rate": 3.428571428571429e-06,
"loss": 0.8863,
"step": 12
},
{
"epoch": 0.08,
"learning_rate": 3.7142857142857146e-06,
"loss": 0.7234,
"step": 13
},
{
"epoch": 0.09,
"learning_rate": 4.000000000000001e-06,
"loss": 0.7408,
"step": 14
},
{
"epoch": 0.1,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.6,
"step": 15
},
{
"epoch": 0.1,
"learning_rate": 4.571428571428572e-06,
"loss": 0.5207,
"step": 16
},
{
"epoch": 0.11,
"learning_rate": 4.857142857142858e-06,
"loss": 0.4958,
"step": 17
},
{
"epoch": 0.12,
"learning_rate": 5.142857142857142e-06,
"loss": 0.6792,
"step": 18
},
{
"epoch": 0.12,
"learning_rate": 5.428571428571429e-06,
"loss": 0.752,
"step": 19
},
{
"epoch": 0.13,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.4657,
"step": 20
},
{
"epoch": 0.14,
"learning_rate": 6e-06,
"loss": 0.4881,
"step": 21
},
{
"epoch": 0.14,
"learning_rate": 6.285714285714286e-06,
"loss": 0.4065,
"step": 22
},
{
"epoch": 0.15,
"learning_rate": 6.571428571428572e-06,
"loss": 0.5692,
"step": 23
},
{
"epoch": 0.15,
"learning_rate": 6.857142857142858e-06,
"loss": 0.5835,
"step": 24
},
{
"epoch": 0.16,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.3869,
"step": 25
},
{
"epoch": 0.17,
"learning_rate": 7.428571428571429e-06,
"loss": 0.5595,
"step": 26
},
{
"epoch": 0.17,
"learning_rate": 7.714285714285716e-06,
"loss": 0.594,
"step": 27
},
{
"epoch": 0.18,
"learning_rate": 8.000000000000001e-06,
"loss": 0.5901,
"step": 28
},
{
"epoch": 0.19,
"learning_rate": 8.285714285714287e-06,
"loss": 0.5574,
"step": 29
},
{
"epoch": 0.19,
"learning_rate": 8.571428571428571e-06,
"loss": 0.647,
"step": 30
},
{
"epoch": 0.2,
"learning_rate": 8.857142857142858e-06,
"loss": 0.5626,
"step": 31
},
{
"epoch": 0.21,
"learning_rate": 9.142857142857144e-06,
"loss": 0.4223,
"step": 32
},
{
"epoch": 0.21,
"learning_rate": 9.42857142857143e-06,
"loss": 0.4088,
"step": 33
},
{
"epoch": 0.22,
"learning_rate": 9.714285714285715e-06,
"loss": 0.4824,
"step": 34
},
{
"epoch": 0.23,
"learning_rate": 1e-05,
"loss": 0.6285,
"step": 35
},
{
"epoch": 0.23,
"learning_rate": 1.0285714285714285e-05,
"loss": 0.5134,
"step": 36
},
{
"epoch": 0.24,
"learning_rate": 1.0571428571428572e-05,
"loss": 0.3572,
"step": 37
},
{
"epoch": 0.25,
"learning_rate": 1.0857142857142858e-05,
"loss": 0.4138,
"step": 38
},
{
"epoch": 0.25,
"learning_rate": 1.1142857142857143e-05,
"loss": 0.4777,
"step": 39
},
{
"epoch": 0.26,
"learning_rate": 1.1428571428571429e-05,
"loss": 0.4846,
"step": 40
},
{
"epoch": 0.26,
"learning_rate": 1.1714285714285716e-05,
"loss": 0.3481,
"step": 41
},
{
"epoch": 0.27,
"learning_rate": 1.2e-05,
"loss": 0.4622,
"step": 42
},
{
"epoch": 0.28,
"learning_rate": 1.2285714285714288e-05,
"loss": 0.4886,
"step": 43
},
{
"epoch": 0.28,
"learning_rate": 1.2571428571428572e-05,
"loss": 0.4303,
"step": 44
},
{
"epoch": 0.29,
"learning_rate": 1.2857142857142859e-05,
"loss": 0.4515,
"step": 45
},
{
"epoch": 0.3,
"learning_rate": 1.3142857142857145e-05,
"loss": 0.4502,
"step": 46
},
{
"epoch": 0.3,
"learning_rate": 1.3428571428571429e-05,
"loss": 0.419,
"step": 47
},
{
"epoch": 0.31,
"learning_rate": 1.3714285714285716e-05,
"loss": 0.2902,
"step": 48
},
{
"epoch": 0.32,
"learning_rate": 1.4e-05,
"loss": 0.372,
"step": 49
},
{
"epoch": 0.32,
"learning_rate": 1.4285714285714287e-05,
"loss": 0.3765,
"step": 50
},
{
"epoch": 0.33,
"learning_rate": 1.4571428571428573e-05,
"loss": 0.393,
"step": 51
},
{
"epoch": 0.34,
"learning_rate": 1.4857142857142858e-05,
"loss": 0.5502,
"step": 52
},
{
"epoch": 0.34,
"learning_rate": 1.5142857142857144e-05,
"loss": 0.3781,
"step": 53
},
{
"epoch": 0.35,
"learning_rate": 1.542857142857143e-05,
"loss": 0.3712,
"step": 54
},
{
"epoch": 0.35,
"learning_rate": 1.5714285714285715e-05,
"loss": 0.4655,
"step": 55
},
{
"epoch": 0.36,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.381,
"step": 56
},
{
"epoch": 0.37,
"learning_rate": 1.6285714285714287e-05,
"loss": 0.4713,
"step": 57
},
{
"epoch": 0.37,
"learning_rate": 1.6571428571428574e-05,
"loss": 0.4491,
"step": 58
},
{
"epoch": 0.38,
"learning_rate": 1.6857142857142858e-05,
"loss": 0.4026,
"step": 59
},
{
"epoch": 0.39,
"learning_rate": 1.7142857142857142e-05,
"loss": 0.401,
"step": 60
},
{
"epoch": 0.39,
"learning_rate": 1.742857142857143e-05,
"loss": 0.3512,
"step": 61
},
{
"epoch": 0.4,
"learning_rate": 1.7714285714285717e-05,
"loss": 0.5228,
"step": 62
},
{
"epoch": 0.41,
"learning_rate": 1.8e-05,
"loss": 0.3911,
"step": 63
},
{
"epoch": 0.41,
"learning_rate": 1.8285714285714288e-05,
"loss": 0.3921,
"step": 64
},
{
"epoch": 0.42,
"learning_rate": 1.8571428571428575e-05,
"loss": 0.3455,
"step": 65
},
{
"epoch": 0.43,
"learning_rate": 1.885714285714286e-05,
"loss": 0.4849,
"step": 66
},
{
"epoch": 0.43,
"learning_rate": 1.9142857142857146e-05,
"loss": 0.4458,
"step": 67
},
{
"epoch": 0.44,
"learning_rate": 1.942857142857143e-05,
"loss": 0.3218,
"step": 68
},
{
"epoch": 0.45,
"learning_rate": 1.9714285714285718e-05,
"loss": 0.3916,
"step": 69
},
{
"epoch": 0.45,
"learning_rate": 2e-05,
"loss": 0.3546,
"step": 70
},
{
"epoch": 0.46,
"learning_rate": 1.999999029542352e-05,
"loss": 0.3862,
"step": 71
},
{
"epoch": 0.46,
"learning_rate": 1.999996118171291e-05,
"loss": 0.4545,
"step": 72
},
{
"epoch": 0.47,
"learning_rate": 1.999991265892468e-05,
"loss": 0.3283,
"step": 73
},
{
"epoch": 0.48,
"learning_rate": 1.999984472715301e-05,
"loss": 0.4031,
"step": 74
},
{
"epoch": 0.48,
"learning_rate": 1.9999757386529747e-05,
"loss": 0.4373,
"step": 75
},
{
"epoch": 0.49,
"learning_rate": 1.9999650637224412e-05,
"loss": 0.3558,
"step": 76
},
{
"epoch": 0.5,
"learning_rate": 1.99995244794442e-05,
"loss": 0.4533,
"step": 77
},
{
"epoch": 0.5,
"learning_rate": 1.9999378913433968e-05,
"loss": 0.2659,
"step": 78
},
{
"epoch": 0.51,
"learning_rate": 1.9999213939476245e-05,
"loss": 0.4677,
"step": 79
},
{
"epoch": 0.52,
"learning_rate": 1.999902955789124e-05,
"loss": 0.5344,
"step": 80
},
{
"epoch": 0.52,
"learning_rate": 1.999882576903682e-05,
"loss": 0.3528,
"step": 81
},
{
"epoch": 0.53,
"learning_rate": 1.9998602573308515e-05,
"loss": 0.3205,
"step": 82
},
{
"epoch": 0.54,
"learning_rate": 1.9998359971139533e-05,
"loss": 0.4116,
"step": 83
},
{
"epoch": 0.54,
"learning_rate": 1.9998097963000746e-05,
"loss": 0.3962,
"step": 84
},
{
"epoch": 0.55,
"learning_rate": 1.9997816549400688e-05,
"loss": 0.284,
"step": 85
},
{
"epoch": 0.55,
"learning_rate": 1.999751573088556e-05,
"loss": 0.3971,
"step": 86
},
{
"epoch": 0.56,
"learning_rate": 1.9997195508039224e-05,
"loss": 0.3766,
"step": 87
},
{
"epoch": 0.57,
"learning_rate": 1.999685588148321e-05,
"loss": 0.4829,
"step": 88
},
{
"epoch": 0.57,
"learning_rate": 1.9996496851876695e-05,
"loss": 0.4042,
"step": 89
},
{
"epoch": 0.58,
"learning_rate": 1.9996118419916532e-05,
"loss": 0.4209,
"step": 90
},
{
"epoch": 0.59,
"learning_rate": 1.9995720586337223e-05,
"loss": 0.3664,
"step": 91
},
{
"epoch": 0.59,
"learning_rate": 1.9995303351910934e-05,
"loss": 0.3888,
"step": 92
},
{
"epoch": 0.6,
"learning_rate": 1.9994866717447475e-05,
"loss": 0.3667,
"step": 93
},
{
"epoch": 0.61,
"learning_rate": 1.9994410683794316e-05,
"loss": 0.3731,
"step": 94
},
{
"epoch": 0.61,
"learning_rate": 1.999393525183659e-05,
"loss": 0.4616,
"step": 95
},
{
"epoch": 0.62,
"learning_rate": 1.999344042249706e-05,
"loss": 0.4204,
"step": 96
},
{
"epoch": 0.63,
"learning_rate": 1.999292619673615e-05,
"loss": 0.3732,
"step": 97
},
{
"epoch": 0.63,
"learning_rate": 1.9992392575551925e-05,
"loss": 0.277,
"step": 98
},
{
"epoch": 0.64,
"learning_rate": 1.999183955998011e-05,
"loss": 0.3699,
"step": 99
},
{
"epoch": 0.65,
"learning_rate": 1.9991267151094055e-05,
"loss": 0.3186,
"step": 100
},
{
"epoch": 0.65,
"learning_rate": 1.9990675350004754e-05,
"loss": 0.3513,
"step": 101
},
{
"epoch": 0.66,
"learning_rate": 1.9990064157860844e-05,
"loss": 0.4491,
"step": 102
},
{
"epoch": 0.66,
"learning_rate": 1.99894335758486e-05,
"loss": 0.4111,
"step": 103
},
{
"epoch": 0.67,
"learning_rate": 1.998878360519193e-05,
"loss": 0.4808,
"step": 104
},
{
"epoch": 0.68,
"learning_rate": 1.9988114247152363e-05,
"loss": 0.4518,
"step": 105
},
{
"epoch": 0.68,
"learning_rate": 1.9987425503029075e-05,
"loss": 0.3553,
"step": 106
},
{
"epoch": 0.69,
"learning_rate": 1.9986717374158857e-05,
"loss": 0.3689,
"step": 107
},
{
"epoch": 0.7,
"learning_rate": 1.9985989861916127e-05,
"loss": 0.3175,
"step": 108
},
{
"epoch": 0.7,
"learning_rate": 1.9985242967712928e-05,
"loss": 0.3393,
"step": 109
},
{
"epoch": 0.71,
"learning_rate": 1.9984476692998914e-05,
"loss": 0.4205,
"step": 110
},
{
"epoch": 0.72,
"learning_rate": 1.9983691039261358e-05,
"loss": 0.4268,
"step": 111
},
{
"epoch": 0.72,
"learning_rate": 1.9982886008025152e-05,
"loss": 0.3376,
"step": 112
},
{
"epoch": 0.73,
"learning_rate": 1.9982061600852792e-05,
"loss": 0.398,
"step": 113
},
{
"epoch": 0.74,
"learning_rate": 1.998121781934438e-05,
"loss": 0.3853,
"step": 114
},
{
"epoch": 0.74,
"learning_rate": 1.9980354665137626e-05,
"loss": 0.4713,
"step": 115
},
{
"epoch": 0.75,
"learning_rate": 1.997947213990784e-05,
"loss": 0.337,
"step": 116
},
{
"epoch": 0.75,
"learning_rate": 1.997857024536793e-05,
"loss": 0.3213,
"step": 117
},
{
"epoch": 0.76,
"learning_rate": 1.9977648983268396e-05,
"loss": 0.2403,
"step": 118
},
{
"epoch": 0.77,
"learning_rate": 1.997670835539733e-05,
"loss": 0.3432,
"step": 119
},
{
"epoch": 0.77,
"learning_rate": 1.9975748363580406e-05,
"loss": 0.4101,
"step": 120
},
{
"epoch": 0.78,
"learning_rate": 1.997476900968089e-05,
"loss": 0.3081,
"step": 121
},
{
"epoch": 0.79,
"learning_rate": 1.9973770295599628e-05,
"loss": 0.316,
"step": 122
},
{
"epoch": 0.79,
"learning_rate": 1.9972752223275037e-05,
"loss": 0.365,
"step": 123
},
{
"epoch": 0.8,
"learning_rate": 1.997171479468311e-05,
"loss": 0.3315,
"step": 124
},
{
"epoch": 0.81,
"learning_rate": 1.9970658011837404e-05,
"loss": 0.3131,
"step": 125
},
{
"epoch": 0.81,
"learning_rate": 1.9969581876789053e-05,
"loss": 0.3333,
"step": 126
},
{
"epoch": 0.82,
"learning_rate": 1.9968486391626736e-05,
"loss": 0.3127,
"step": 127
},
{
"epoch": 0.83,
"learning_rate": 1.99673715584767e-05,
"loss": 0.3705,
"step": 128
},
{
"epoch": 0.83,
"learning_rate": 1.9966237379502744e-05,
"loss": 0.406,
"step": 129
},
{
"epoch": 0.84,
"learning_rate": 1.996508385690621e-05,
"loss": 0.3114,
"step": 130
},
{
"epoch": 0.85,
"learning_rate": 1.996391099292599e-05,
"loss": 0.2634,
"step": 131
},
{
"epoch": 0.85,
"learning_rate": 1.9962718789838515e-05,
"loss": 0.3985,
"step": 132
},
{
"epoch": 0.86,
"learning_rate": 1.996150724995774e-05,
"loss": 0.3978,
"step": 133
},
{
"epoch": 0.86,
"learning_rate": 1.9960276375635175e-05,
"loss": 0.2702,
"step": 134
},
{
"epoch": 0.87,
"learning_rate": 1.9959026169259833e-05,
"loss": 0.3777,
"step": 135
},
{
"epoch": 0.88,
"learning_rate": 1.9957756633258264e-05,
"loss": 0.3993,
"step": 136
},
{
"epoch": 0.88,
"learning_rate": 1.9956467770094533e-05,
"loss": 0.4165,
"step": 137
},
{
"epoch": 0.89,
"learning_rate": 1.9955159582270204e-05,
"loss": 0.3386,
"step": 138
},
{
"epoch": 0.9,
"learning_rate": 1.9953832072324365e-05,
"loss": 0.3443,
"step": 139
},
{
"epoch": 0.9,
"learning_rate": 1.99524852428336e-05,
"loss": 0.265,
"step": 140
},
{
"epoch": 0.91,
"learning_rate": 1.9951119096411992e-05,
"loss": 0.3121,
"step": 141
},
{
"epoch": 0.92,
"learning_rate": 1.9949733635711115e-05,
"loss": 0.3932,
"step": 142
},
{
"epoch": 0.92,
"learning_rate": 1.994832886342003e-05,
"loss": 0.2794,
"step": 143
},
{
"epoch": 0.93,
"learning_rate": 1.9946904782265276e-05,
"loss": 0.3495,
"step": 144
},
{
"epoch": 0.94,
"learning_rate": 1.9945461395010885e-05,
"loss": 0.3416,
"step": 145
},
{
"epoch": 0.94,
"learning_rate": 1.9943998704458338e-05,
"loss": 0.289,
"step": 146
},
{
"epoch": 0.95,
"learning_rate": 1.99425167134466e-05,
"loss": 0.2484,
"step": 147
},
{
"epoch": 0.95,
"learning_rate": 1.994101542485209e-05,
"loss": 0.4093,
"step": 148
},
{
"epoch": 0.96,
"learning_rate": 1.993949484158868e-05,
"loss": 0.3191,
"step": 149
},
{
"epoch": 0.97,
"learning_rate": 1.9937954966607696e-05,
"loss": 0.3989,
"step": 150
},
{
"epoch": 0.97,
"learning_rate": 1.99363958028979e-05,
"loss": 0.404,
"step": 151
},
{
"epoch": 0.98,
"learning_rate": 1.99348173534855e-05,
"loss": 0.3196,
"step": 152
},
{
"epoch": 0.99,
"learning_rate": 1.9933219621434137e-05,
"loss": 0.2951,
"step": 153
},
{
"epoch": 0.99,
"learning_rate": 1.9931602609844867e-05,
"loss": 0.449,
"step": 154
},
{
"epoch": 1.0,
"learning_rate": 1.992996632185617e-05,
"loss": 0.0854,
"step": 155
},
{
"epoch": 1.01,
"learning_rate": 1.9928310760643953e-05,
"loss": 0.2437,
"step": 156
},
{
"epoch": 1.01,
"learning_rate": 1.992663592942151e-05,
"loss": 0.1749,
"step": 157
},
{
"epoch": 1.02,
"learning_rate": 1.992494183143955e-05,
"loss": 0.1916,
"step": 158
},
{
"epoch": 1.03,
"learning_rate": 1.9923228469986178e-05,
"loss": 0.1537,
"step": 159
},
{
"epoch": 1.03,
"learning_rate": 1.9921495848386877e-05,
"loss": 0.23,
"step": 160
},
{
"epoch": 1.04,
"learning_rate": 1.9919743970004522e-05,
"loss": 0.2576,
"step": 161
},
{
"epoch": 1.05,
"learning_rate": 1.9917972838239358e-05,
"loss": 0.1854,
"step": 162
},
{
"epoch": 1.05,
"learning_rate": 1.9916182456529005e-05,
"loss": 0.263,
"step": 163
},
{
"epoch": 1.06,
"learning_rate": 1.991437282834844e-05,
"loss": 0.2272,
"step": 164
},
{
"epoch": 1.06,
"learning_rate": 1.991254395721e-05,
"loss": 0.1956,
"step": 165
},
{
"epoch": 1.07,
"learning_rate": 1.991069584666337e-05,
"loss": 0.2129,
"step": 166
},
{
"epoch": 1.08,
"learning_rate": 1.990882850029557e-05,
"loss": 0.1594,
"step": 167
},
{
"epoch": 1.08,
"learning_rate": 1.9906941921730966e-05,
"loss": 0.1702,
"step": 168
},
{
"epoch": 1.09,
"learning_rate": 1.9905036114631247e-05,
"loss": 0.1843,
"step": 169
},
{
"epoch": 1.1,
"learning_rate": 1.990311108269542e-05,
"loss": 0.1732,
"step": 170
},
{
"epoch": 1.1,
"learning_rate": 1.9901166829659812e-05,
"loss": 0.1656,
"step": 171
},
{
"epoch": 1.11,
"learning_rate": 1.9899203359298056e-05,
"loss": 0.2087,
"step": 172
},
{
"epoch": 1.12,
"learning_rate": 1.9897220675421075e-05,
"loss": 0.2316,
"step": 173
},
{
"epoch": 1.12,
"learning_rate": 1.9895218781877097e-05,
"loss": 0.1296,
"step": 174
},
{
"epoch": 1.13,
"learning_rate": 1.9893197682551624e-05,
"loss": 0.1997,
"step": 175
},
{
"epoch": 1.14,
"learning_rate": 1.9891157381367436e-05,
"loss": 0.1166,
"step": 176
},
{
"epoch": 1.14,
"learning_rate": 1.9889097882284593e-05,
"loss": 0.1808,
"step": 177
},
{
"epoch": 1.15,
"learning_rate": 1.9887019189300398e-05,
"loss": 0.1911,
"step": 178
},
{
"epoch": 1.15,
"learning_rate": 1.9884921306449426e-05,
"loss": 0.1588,
"step": 179
},
{
"epoch": 1.16,
"learning_rate": 1.9882804237803487e-05,
"loss": 0.1818,
"step": 180
},
{
"epoch": 1.17,
"learning_rate": 1.988066798747163e-05,
"loss": 0.147,
"step": 181
},
{
"epoch": 1.17,
"learning_rate": 1.9878512559600142e-05,
"loss": 0.1397,
"step": 182
},
{
"epoch": 1.18,
"learning_rate": 1.987633795837252e-05,
"loss": 0.1868,
"step": 183
},
{
"epoch": 1.19,
"learning_rate": 1.9874144188009483e-05,
"loss": 0.1151,
"step": 184
},
{
"epoch": 1.19,
"learning_rate": 1.9871931252768953e-05,
"loss": 0.1806,
"step": 185
},
{
"epoch": 1.2,
"learning_rate": 1.986969915694605e-05,
"loss": 0.1835,
"step": 186
},
{
"epoch": 1.21,
"learning_rate": 1.9867447904873084e-05,
"loss": 0.2098,
"step": 187
},
{
"epoch": 1.21,
"learning_rate": 1.9865177500919545e-05,
"loss": 0.1514,
"step": 188
},
{
"epoch": 1.22,
"learning_rate": 1.9862887949492093e-05,
"loss": 0.1681,
"step": 189
},
{
"epoch": 1.23,
"learning_rate": 1.986057925503455e-05,
"loss": 0.1696,
"step": 190
},
{
"epoch": 1.23,
"learning_rate": 1.9858251422027903e-05,
"loss": 0.2191,
"step": 191
},
{
"epoch": 1.24,
"learning_rate": 1.9855904454990275e-05,
"loss": 0.1808,
"step": 192
},
{
"epoch": 1.25,
"learning_rate": 1.9853538358476933e-05,
"loss": 0.1419,
"step": 193
},
{
"epoch": 1.25,
"learning_rate": 1.9851153137080267e-05,
"loss": 0.1378,
"step": 194
},
{
"epoch": 1.26,
"learning_rate": 1.984874879542979e-05,
"loss": 0.1596,
"step": 195
},
{
"epoch": 1.26,
"learning_rate": 1.9846325338192126e-05,
"loss": 0.1987,
"step": 196
},
{
"epoch": 1.27,
"learning_rate": 1.9843882770071e-05,
"loss": 0.1454,
"step": 197
},
{
"epoch": 1.28,
"learning_rate": 1.9841421095807233e-05,
"loss": 0.2658,
"step": 198
},
{
"epoch": 1.28,
"learning_rate": 1.983894032017872e-05,
"loss": 0.1348,
"step": 199
},
{
"epoch": 1.29,
"learning_rate": 1.983644044800044e-05,
"loss": 0.1504,
"step": 200
},
{
"epoch": 1.3,
"learning_rate": 1.9833921484124433e-05,
"loss": 0.1762,
"step": 201
},
{
"epoch": 1.3,
"learning_rate": 1.9831383433439798e-05,
"loss": 0.1189,
"step": 202
},
{
"epoch": 1.31,
"learning_rate": 1.982882630087267e-05,
"loss": 0.1553,
"step": 203
},
{
"epoch": 1.32,
"learning_rate": 1.982625009138623e-05,
"loss": 0.1845,
"step": 204
},
{
"epoch": 1.32,
"learning_rate": 1.9823654809980682e-05,
"loss": 0.1792,
"step": 205
},
{
"epoch": 1.33,
"learning_rate": 1.982104046169325e-05,
"loss": 0.1579,
"step": 206
},
{
"epoch": 1.34,
"learning_rate": 1.9818407051598156e-05,
"loss": 0.2289,
"step": 207
},
{
"epoch": 1.34,
"learning_rate": 1.9815754584806633e-05,
"loss": 0.2756,
"step": 208
},
{
"epoch": 1.35,
"learning_rate": 1.981308306646689e-05,
"loss": 0.2026,
"step": 209
},
{
"epoch": 1.35,
"learning_rate": 1.9810392501764118e-05,
"loss": 0.1604,
"step": 210
},
{
"epoch": 1.36,
"learning_rate": 1.9807682895920478e-05,
"loss": 0.1687,
"step": 211
},
{
"epoch": 1.37,
"learning_rate": 1.980495425419508e-05,
"loss": 0.1614,
"step": 212
},
{
"epoch": 1.37,
"learning_rate": 1.9802206581883992e-05,
"loss": 0.2028,
"step": 213
},
{
"epoch": 1.38,
"learning_rate": 1.9799439884320214e-05,
"loss": 0.2121,
"step": 214
},
{
"epoch": 1.39,
"learning_rate": 1.9796654166873667e-05,
"loss": 0.2126,
"step": 215
},
{
"epoch": 1.39,
"learning_rate": 1.9793849434951196e-05,
"loss": 0.144,
"step": 216
},
{
"epoch": 1.4,
"learning_rate": 1.9791025693996543e-05,
"loss": 0.1992,
"step": 217
},
{
"epoch": 1.41,
"learning_rate": 1.978818294949036e-05,
"loss": 0.1287,
"step": 218
},
{
"epoch": 1.41,
"learning_rate": 1.9785321206950162e-05,
"loss": 0.1348,
"step": 219
},
{
"epoch": 1.42,
"learning_rate": 1.9782440471930354e-05,
"loss": 0.1791,
"step": 220
},
{
"epoch": 1.43,
"learning_rate": 1.97795407500222e-05,
"loss": 0.1855,
"step": 221
},
{
"epoch": 1.43,
"learning_rate": 1.9776622046853814e-05,
"loss": 0.1625,
"step": 222
},
{
"epoch": 1.44,
"learning_rate": 1.977368436809015e-05,
"loss": 0.2079,
"step": 223
},
{
"epoch": 1.45,
"learning_rate": 1.9770727719432994e-05,
"loss": 0.1522,
"step": 224
},
{
"epoch": 1.45,
"learning_rate": 1.976775210662095e-05,
"loss": 0.2049,
"step": 225
},
{
"epoch": 1.46,
"learning_rate": 1.9764757535429435e-05,
"loss": 0.2404,
"step": 226
},
{
"epoch": 1.46,
"learning_rate": 1.9761744011670648e-05,
"loss": 0.2287,
"step": 227
},
{
"epoch": 1.47,
"learning_rate": 1.9758711541193597e-05,
"loss": 0.2175,
"step": 228
},
{
"epoch": 1.48,
"learning_rate": 1.975566012988404e-05,
"loss": 0.1887,
"step": 229
},
{
"epoch": 1.48,
"learning_rate": 1.975258978366451e-05,
"loss": 0.1682,
"step": 230
},
{
"epoch": 1.49,
"learning_rate": 1.974950050849429e-05,
"loss": 0.1534,
"step": 231
},
{
"epoch": 1.5,
"learning_rate": 1.9746392310369405e-05,
"loss": 0.2147,
"step": 232
},
{
"epoch": 1.5,
"learning_rate": 1.9743265195322595e-05,
"loss": 0.168,
"step": 233
},
{
"epoch": 1.51,
"learning_rate": 1.9740119169423337e-05,
"loss": 0.2213,
"step": 234
},
{
"epoch": 1.52,
"learning_rate": 1.9736954238777793e-05,
"loss": 0.1934,
"step": 235
},
{
"epoch": 1.52,
"learning_rate": 1.9733770409528824e-05,
"loss": 0.1981,
"step": 236
},
{
"epoch": 1.53,
"learning_rate": 1.973056768785598e-05,
"loss": 0.1531,
"step": 237
},
{
"epoch": 1.54,
"learning_rate": 1.9727346079975468e-05,
"loss": 0.2658,
"step": 238
},
{
"epoch": 1.54,
"learning_rate": 1.9724105592140157e-05,
"loss": 0.2813,
"step": 239
},
{
"epoch": 1.55,
"learning_rate": 1.9720846230639555e-05,
"loss": 0.1674,
"step": 240
},
{
"epoch": 1.55,
"learning_rate": 1.9717568001799814e-05,
"loss": 0.2195,
"step": 241
},
{
"epoch": 1.56,
"learning_rate": 1.9714270911983698e-05,
"loss": 0.1857,
"step": 242
},
{
"epoch": 1.57,
"learning_rate": 1.9710954967590573e-05,
"loss": 0.2317,
"step": 243
},
{
"epoch": 1.57,
"learning_rate": 1.9707620175056406e-05,
"loss": 0.1752,
"step": 244
},
{
"epoch": 1.58,
"learning_rate": 1.9704266540853753e-05,
"loss": 0.1791,
"step": 245
},
{
"epoch": 1.59,
"learning_rate": 1.9700894071491736e-05,
"loss": 0.1785,
"step": 246
},
{
"epoch": 1.59,
"learning_rate": 1.9697502773516018e-05,
"loss": 0.1691,
"step": 247
},
{
"epoch": 1.6,
"learning_rate": 1.969409265350884e-05,
"loss": 0.113,
"step": 248
},
{
"epoch": 1.61,
"learning_rate": 1.9690663718088942e-05,
"loss": 0.1618,
"step": 249
},
{
"epoch": 1.61,
"learning_rate": 1.96872159739116e-05,
"loss": 0.2428,
"step": 250
},
{
"epoch": 1.62,
"learning_rate": 1.96837494276686e-05,
"loss": 0.1618,
"step": 251
},
{
"epoch": 1.63,
"learning_rate": 1.9680264086088204e-05,
"loss": 0.1659,
"step": 252
},
{
"epoch": 1.63,
"learning_rate": 1.9676759955935176e-05,
"loss": 0.1765,
"step": 253
},
{
"epoch": 1.64,
"learning_rate": 1.967323704401073e-05,
"loss": 0.1896,
"step": 254
},
{
"epoch": 1.65,
"learning_rate": 1.966969535715254e-05,
"loss": 0.1678,
"step": 255
},
{
"epoch": 1.65,
"learning_rate": 1.9666134902234716e-05,
"loss": 0.1554,
"step": 256
},
{
"epoch": 1.66,
"learning_rate": 1.9662555686167808e-05,
"loss": 0.1591,
"step": 257
},
{
"epoch": 1.66,
"learning_rate": 1.9658957715898767e-05,
"loss": 0.1555,
"step": 258
},
{
"epoch": 1.67,
"learning_rate": 1.965534099841095e-05,
"loss": 0.1946,
"step": 259
},
{
"epoch": 1.68,
"learning_rate": 1.9651705540724092e-05,
"loss": 0.1794,
"step": 260
},
{
"epoch": 1.68,
"learning_rate": 1.964805134989432e-05,
"loss": 0.2276,
"step": 261
},
{
"epoch": 1.69,
"learning_rate": 1.96443784330141e-05,
"loss": 0.135,
"step": 262
},
{
"epoch": 1.7,
"learning_rate": 1.964068679721226e-05,
"loss": 0.1847,
"step": 263
},
{
"epoch": 1.7,
"learning_rate": 1.9636976449653945e-05,
"loss": 0.2136,
"step": 264
},
{
"epoch": 1.71,
"learning_rate": 1.963324739754063e-05,
"loss": 0.1744,
"step": 265
},
{
"epoch": 1.72,
"learning_rate": 1.962949964811008e-05,
"loss": 0.2052,
"step": 266
},
{
"epoch": 1.72,
"learning_rate": 1.9625733208636376e-05,
"loss": 0.2062,
"step": 267
},
{
"epoch": 1.73,
"learning_rate": 1.9621948086429847e-05,
"loss": 0.1653,
"step": 268
},
{
"epoch": 1.74,
"learning_rate": 1.9618144288837095e-05,
"loss": 0.1951,
"step": 269
},
{
"epoch": 1.74,
"learning_rate": 1.9614321823240968e-05,
"loss": 0.1955,
"step": 270
},
{
"epoch": 1.75,
"learning_rate": 1.9610480697060553e-05,
"loss": 0.2279,
"step": 271
},
{
"epoch": 1.75,
"learning_rate": 1.960662091775115e-05,
"loss": 0.1841,
"step": 272
},
{
"epoch": 1.76,
"learning_rate": 1.960274249280426e-05,
"loss": 0.2083,
"step": 273
},
{
"epoch": 1.77,
"learning_rate": 1.959884542974758e-05,
"loss": 0.1623,
"step": 274
},
{
"epoch": 1.77,
"learning_rate": 1.9594929736144978e-05,
"loss": 0.1604,
"step": 275
},
{
"epoch": 1.78,
"learning_rate": 1.9590995419596482e-05,
"loss": 0.1452,
"step": 276
},
{
"epoch": 1.79,
"learning_rate": 1.9587042487738267e-05,
"loss": 0.1694,
"step": 277
},
{
"epoch": 1.79,
"learning_rate": 1.9583070948242648e-05,
"loss": 0.1434,
"step": 278
},
{
"epoch": 1.8,
"learning_rate": 1.9579080808818035e-05,
"loss": 0.1909,
"step": 279
},
{
"epoch": 1.81,
"learning_rate": 1.9575072077208956e-05,
"loss": 0.2118,
"step": 280
},
{
"epoch": 1.81,
"learning_rate": 1.9571044761196017e-05,
"loss": 0.2142,
"step": 281
},
{
"epoch": 1.82,
"learning_rate": 1.9566998868595898e-05,
"loss": 0.1868,
"step": 282
},
{
"epoch": 1.83,
"learning_rate": 1.956293440726134e-05,
"loss": 0.1861,
"step": 283
},
{
"epoch": 1.83,
"learning_rate": 1.955885138508111e-05,
"loss": 0.1352,
"step": 284
},
{
"epoch": 1.84,
"learning_rate": 1.9554749809980014e-05,
"loss": 0.1381,
"step": 285
},
{
"epoch": 1.85,
"learning_rate": 1.9550629689918854e-05,
"loss": 0.2036,
"step": 286
},
{
"epoch": 1.85,
"learning_rate": 1.9546491032894442e-05,
"loss": 0.1227,
"step": 287
},
{
"epoch": 1.86,
"learning_rate": 1.9542333846939558e-05,
"loss": 0.1518,
"step": 288
},
{
"epoch": 1.86,
"learning_rate": 1.953815814012295e-05,
"loss": 0.1828,
"step": 289
},
{
"epoch": 1.87,
"learning_rate": 1.9533963920549307e-05,
"loss": 0.2018,
"step": 290
},
{
"epoch": 1.88,
"learning_rate": 1.9529751196359256e-05,
"loss": 0.1772,
"step": 291
},
{
"epoch": 1.88,
"learning_rate": 1.9525519975729338e-05,
"loss": 0.2657,
"step": 292
},
{
"epoch": 1.89,
"learning_rate": 1.952127026687199e-05,
"loss": 0.1604,
"step": 293
},
{
"epoch": 1.9,
"learning_rate": 1.9517002078035548e-05,
"loss": 0.1593,
"step": 294
},
{
"epoch": 1.9,
"learning_rate": 1.951271541750419e-05,
"loss": 0.1719,
"step": 295
},
{
"epoch": 1.91,
"learning_rate": 1.9508410293597976e-05,
"loss": 0.1404,
"step": 296
},
{
"epoch": 1.92,
"learning_rate": 1.950408671467277e-05,
"loss": 0.2389,
"step": 297
},
{
"epoch": 1.92,
"learning_rate": 1.949974468912029e-05,
"loss": 0.1577,
"step": 298
},
{
"epoch": 1.93,
"learning_rate": 1.9495384225368026e-05,
"loss": 0.1518,
"step": 299
},
{
"epoch": 1.94,
"learning_rate": 1.949100533187928e-05,
"loss": 0.1255,
"step": 300
},
{
"epoch": 1.94,
"learning_rate": 1.94866080171531e-05,
"loss": 0.1414,
"step": 301
},
{
"epoch": 1.95,
"learning_rate": 1.9482192289724316e-05,
"loss": 0.2118,
"step": 302
},
{
"epoch": 1.95,
"learning_rate": 1.947775815816347e-05,
"loss": 0.1935,
"step": 303
},
{
"epoch": 1.96,
"learning_rate": 1.947330563107684e-05,
"loss": 0.1414,
"step": 304
},
{
"epoch": 1.97,
"learning_rate": 1.946883471710641e-05,
"loss": 0.1893,
"step": 305
},
{
"epoch": 1.97,
"learning_rate": 1.9464345424929832e-05,
"loss": 0.1481,
"step": 306
},
{
"epoch": 1.98,
"learning_rate": 1.945983776326045e-05,
"loss": 0.1497,
"step": 307
},
{
"epoch": 1.99,
"learning_rate": 1.9455311740847257e-05,
"loss": 0.2306,
"step": 308
},
{
"epoch": 1.99,
"learning_rate": 1.945076736647487e-05,
"loss": 0.1565,
"step": 309
},
{
"epoch": 2.0,
"learning_rate": 1.944620464896354e-05,
"loss": 0.1515,
"step": 310
},
{
"epoch": 2.01,
"learning_rate": 1.9441623597169114e-05,
"loss": 0.0862,
"step": 311
},
{
"epoch": 2.01,
"learning_rate": 1.943702421998303e-05,
"loss": 0.0686,
"step": 312
},
{
"epoch": 2.02,
"learning_rate": 1.943240652633228e-05,
"loss": 0.0629,
"step": 313
},
{
"epoch": 2.03,
"learning_rate": 1.9427770525179426e-05,
"loss": 0.076,
"step": 314
},
{
"epoch": 2.03,
"learning_rate": 1.942311622552255e-05,
"loss": 0.0714,
"step": 315
},
{
"epoch": 2.04,
"learning_rate": 1.941844363639525e-05,
"loss": 0.0738,
"step": 316
},
{
"epoch": 2.05,
"learning_rate": 1.9413752766866628e-05,
"loss": 0.0632,
"step": 317
},
{
"epoch": 2.05,
"learning_rate": 1.9409043626041266e-05,
"loss": 0.0867,
"step": 318
},
{
"epoch": 2.06,
"learning_rate": 1.9404316223059205e-05,
"loss": 0.0525,
"step": 319
},
{
"epoch": 2.06,
"learning_rate": 1.939957056709594e-05,
"loss": 0.0519,
"step": 320
},
{
"epoch": 2.07,
"learning_rate": 1.9394806667362374e-05,
"loss": 0.0444,
"step": 321
},
{
"epoch": 2.08,
"learning_rate": 1.939002453310485e-05,
"loss": 0.0644,
"step": 322
},
{
"epoch": 2.08,
"learning_rate": 1.9385224173605072e-05,
"loss": 0.0619,
"step": 323
},
{
"epoch": 2.09,
"learning_rate": 1.9380405598180134e-05,
"loss": 0.0666,
"step": 324
},
{
"epoch": 2.1,
"learning_rate": 1.9375568816182486e-05,
"loss": 0.074,
"step": 325
},
{
"epoch": 2.1,
"learning_rate": 1.9370713836999916e-05,
"loss": 0.0552,
"step": 326
},
{
"epoch": 2.11,
"learning_rate": 1.9365840670055516e-05,
"loss": 0.0781,
"step": 327
},
{
"epoch": 2.12,
"learning_rate": 1.93609493248077e-05,
"loss": 0.0852,
"step": 328
},
{
"epoch": 2.12,
"learning_rate": 1.9356039810750148e-05,
"loss": 0.074,
"step": 329
},
{
"epoch": 2.13,
"learning_rate": 1.9351112137411815e-05,
"loss": 0.0595,
"step": 330
},
{
"epoch": 2.14,
"learning_rate": 1.9346166314356896e-05,
"loss": 0.0622,
"step": 331
},
{
"epoch": 2.14,
"learning_rate": 1.934120235118482e-05,
"loss": 0.0823,
"step": 332
},
{
"epoch": 2.15,
"learning_rate": 1.933622025753021e-05,
"loss": 0.0896,
"step": 333
},
{
"epoch": 2.15,
"learning_rate": 1.9331220043062894e-05,
"loss": 0.068,
"step": 334
},
{
"epoch": 2.16,
"learning_rate": 1.9326201717487865e-05,
"loss": 0.0815,
"step": 335
},
{
"epoch": 2.17,
"learning_rate": 1.9321165290545262e-05,
"loss": 0.0609,
"step": 336
},
{
"epoch": 2.17,
"learning_rate": 1.931611077201037e-05,
"loss": 0.0905,
"step": 337
},
{
"epoch": 2.18,
"learning_rate": 1.9311038171693574e-05,
"loss": 0.0614,
"step": 338
},
{
"epoch": 2.19,
"learning_rate": 1.930594749944037e-05,
"loss": 0.0734,
"step": 339
},
{
"epoch": 2.19,
"learning_rate": 1.9300838765131316e-05,
"loss": 0.0624,
"step": 340
},
{
"epoch": 2.2,
"learning_rate": 1.9295711978682032e-05,
"loss": 0.0582,
"step": 341
},
{
"epoch": 2.21,
"learning_rate": 1.9290567150043183e-05,
"loss": 0.0943,
"step": 342
},
{
"epoch": 2.21,
"learning_rate": 1.9285404289200434e-05,
"loss": 0.0717,
"step": 343
},
{
"epoch": 2.22,
"learning_rate": 1.9280223406174472e-05,
"loss": 0.075,
"step": 344
},
{
"epoch": 2.23,
"learning_rate": 1.927502451102095e-05,
"loss": 0.099,
"step": 345
},
{
"epoch": 2.23,
"learning_rate": 1.9269807613830474e-05,
"loss": 0.0873,
"step": 346
},
{
"epoch": 2.24,
"learning_rate": 1.9264572724728612e-05,
"loss": 0.08,
"step": 347
},
{
"epoch": 2.25,
"learning_rate": 1.925931985387583e-05,
"loss": 0.104,
"step": 348
},
{
"epoch": 2.25,
"learning_rate": 1.9254049011467515e-05,
"loss": 0.0814,
"step": 349
},
{
"epoch": 2.26,
"learning_rate": 1.9248760207733917e-05,
"loss": 0.0996,
"step": 350
},
{
"epoch": 2.26,
"learning_rate": 1.924345345294016e-05,
"loss": 0.072,
"step": 351
},
{
"epoch": 2.27,
"learning_rate": 1.9238128757386207e-05,
"loss": 0.0939,
"step": 352
},
{
"epoch": 2.28,
"learning_rate": 1.9232786131406836e-05,
"loss": 0.0776,
"step": 353
},
{
"epoch": 2.28,
"learning_rate": 1.922742558537164e-05,
"loss": 0.0731,
"step": 354
},
{
"epoch": 2.29,
"learning_rate": 1.9222047129684974e-05,
"loss": 0.0868,
"step": 355
},
{
"epoch": 2.3,
"learning_rate": 1.9216650774785975e-05,
"loss": 0.0613,
"step": 356
},
{
"epoch": 2.3,
"learning_rate": 1.92112365311485e-05,
"loss": 0.0838,
"step": 357
},
{
"epoch": 2.31,
"learning_rate": 1.920580440928115e-05,
"loss": 0.1133,
"step": 358
},
{
"epoch": 2.32,
"learning_rate": 1.9200354419727205e-05,
"loss": 0.0712,
"step": 359
},
{
"epoch": 2.32,
"learning_rate": 1.9194886573064636e-05,
"loss": 0.0732,
"step": 360
},
{
"epoch": 2.33,
"learning_rate": 1.9189400879906065e-05,
"loss": 0.0948,
"step": 361
},
{
"epoch": 2.34,
"learning_rate": 1.9183897350898766e-05,
"loss": 0.0704,
"step": 362
},
{
"epoch": 2.34,
"learning_rate": 1.9178375996724615e-05,
"loss": 0.0851,
"step": 363
},
{
"epoch": 2.35,
"learning_rate": 1.9172836828100103e-05,
"loss": 0.0537,
"step": 364
},
{
"epoch": 2.35,
"learning_rate": 1.9167279855776277e-05,
"loss": 0.0862,
"step": 365
},
{
"epoch": 2.36,
"learning_rate": 1.9161705090538752e-05,
"loss": 0.0756,
"step": 366
},
{
"epoch": 2.37,
"learning_rate": 1.9156112543207674e-05,
"loss": 0.0932,
"step": 367
},
{
"epoch": 2.37,
"learning_rate": 1.915050222463771e-05,
"loss": 0.0823,
"step": 368
},
{
"epoch": 2.38,
"learning_rate": 1.9144874145718008e-05,
"loss": 0.0729,
"step": 369
},
{
"epoch": 2.39,
"learning_rate": 1.9139228317372195e-05,
"loss": 0.0732,
"step": 370
},
{
"epoch": 2.39,
"learning_rate": 1.913356475055834e-05,
"loss": 0.0831,
"step": 371
},
{
"epoch": 2.4,
"learning_rate": 1.9127883456268955e-05,
"loss": 0.079,
"step": 372
},
{
"epoch": 2.41,
"learning_rate": 1.9122184445530942e-05,
"loss": 0.0762,
"step": 373
},
{
"epoch": 2.41,
"learning_rate": 1.911646772940561e-05,
"loss": 0.0834,
"step": 374
},
{
"epoch": 2.42,
"learning_rate": 1.9110733318988607e-05,
"loss": 0.0509,
"step": 375
},
{
"epoch": 2.43,
"learning_rate": 1.9104981225409944e-05,
"loss": 0.0974,
"step": 376
},
{
"epoch": 2.43,
"learning_rate": 1.909921145983395e-05,
"loss": 0.1002,
"step": 377
},
{
"epoch": 2.44,
"learning_rate": 1.909342403345925e-05,
"loss": 0.0612,
"step": 378
},
{
"epoch": 2.45,
"learning_rate": 1.908761895751875e-05,
"loss": 0.0824,
"step": 379
},
{
"epoch": 2.45,
"learning_rate": 1.90817962432796e-05,
"loss": 0.0783,
"step": 380
},
{
"epoch": 2.46,
"learning_rate": 1.907595590204321e-05,
"loss": 0.0786,
"step": 381
},
{
"epoch": 2.46,
"learning_rate": 1.9070097945145177e-05,
"loss": 0.0968,
"step": 382
},
{
"epoch": 2.47,
"learning_rate": 1.9064222383955305e-05,
"loss": 0.0766,
"step": 383
},
{
"epoch": 2.48,
"learning_rate": 1.905832922987756e-05,
"loss": 0.0631,
"step": 384
},
{
"epoch": 2.48,
"learning_rate": 1.9052418494350048e-05,
"loss": 0.0711,
"step": 385
},
{
"epoch": 2.49,
"learning_rate": 1.9046490188845014e-05,
"loss": 0.0896,
"step": 386
},
{
"epoch": 2.5,
"learning_rate": 1.9040544324868796e-05,
"loss": 0.0709,
"step": 387
},
{
"epoch": 2.5,
"learning_rate": 1.9034580913961812e-05,
"loss": 0.0742,
"step": 388
},
{
"epoch": 2.51,
"learning_rate": 1.9028599967698533e-05,
"loss": 0.096,
"step": 389
},
{
"epoch": 2.52,
"learning_rate": 1.9022601497687472e-05,
"loss": 0.0757,
"step": 390
},
{
"epoch": 2.52,
"learning_rate": 1.901658551557115e-05,
"loss": 0.0651,
"step": 391
},
{
"epoch": 2.53,
"learning_rate": 1.9010552033026085e-05,
"loss": 0.0467,
"step": 392
},
{
"epoch": 2.54,
"learning_rate": 1.900450106176275e-05,
"loss": 0.076,
"step": 393
},
{
"epoch": 2.54,
"learning_rate": 1.8998432613525566e-05,
"loss": 0.1182,
"step": 394
},
{
"epoch": 2.55,
"learning_rate": 1.899234670009288e-05,
"loss": 0.06,
"step": 395
},
{
"epoch": 2.55,
"learning_rate": 1.8986243333276933e-05,
"loss": 0.0999,
"step": 396
},
{
"epoch": 2.56,
"learning_rate": 1.8980122524923846e-05,
"loss": 0.0771,
"step": 397
},
{
"epoch": 2.57,
"learning_rate": 1.8973984286913584e-05,
"loss": 0.0932,
"step": 398
},
{
"epoch": 2.57,
"learning_rate": 1.896782863115995e-05,
"loss": 0.0857,
"step": 399
},
{
"epoch": 2.58,
"learning_rate": 1.8961655569610557e-05,
"loss": 0.0951,
"step": 400
},
{
"epoch": 2.59,
"learning_rate": 1.8955465114246787e-05,
"loss": 0.091,
"step": 401
},
{
"epoch": 2.59,
"learning_rate": 1.8949257277083787e-05,
"loss": 0.0639,
"step": 402
},
{
"epoch": 2.6,
"learning_rate": 1.8943032070170448e-05,
"loss": 0.0868,
"step": 403
},
{
"epoch": 2.61,
"learning_rate": 1.893678950558937e-05,
"loss": 0.0826,
"step": 404
},
{
"epoch": 2.61,
"learning_rate": 1.893052959545684e-05,
"loss": 0.063,
"step": 405
},
{
"epoch": 2.62,
"learning_rate": 1.8924252351922813e-05,
"loss": 0.0796,
"step": 406
},
{
"epoch": 2.63,
"learning_rate": 1.891795778717089e-05,
"loss": 0.0799,
"step": 407
},
{
"epoch": 2.63,
"learning_rate": 1.8911645913418284e-05,
"loss": 0.0721,
"step": 408
},
{
"epoch": 2.64,
"learning_rate": 1.890531674291581e-05,
"loss": 0.083,
"step": 409
},
{
"epoch": 2.65,
"learning_rate": 1.889897028794785e-05,
"loss": 0.0993,
"step": 410
},
{
"epoch": 2.65,
"learning_rate": 1.8892606560832335e-05,
"loss": 0.0697,
"step": 411
},
{
"epoch": 2.66,
"learning_rate": 1.8886225573920723e-05,
"loss": 0.1011,
"step": 412
},
{
"epoch": 2.66,
"learning_rate": 1.887982733959797e-05,
"loss": 0.0682,
"step": 413
},
{
"epoch": 2.67,
"learning_rate": 1.8873411870282502e-05,
"loss": 0.0736,
"step": 414
},
{
"epoch": 2.68,
"learning_rate": 1.8866979178426206e-05,
"loss": 0.061,
"step": 415
},
{
"epoch": 2.68,
"learning_rate": 1.8860529276514393e-05,
"loss": 0.0838,
"step": 416
},
{
"epoch": 2.69,
"learning_rate": 1.885406217706577e-05,
"loss": 0.0657,
"step": 417
},
{
"epoch": 2.7,
"learning_rate": 1.884757789263243e-05,
"loss": 0.074,
"step": 418
},
{
"epoch": 2.7,
"learning_rate": 1.8841076435799826e-05,
"loss": 0.0662,
"step": 419
},
{
"epoch": 2.71,
"learning_rate": 1.8834557819186732e-05,
"loss": 0.0739,
"step": 420
},
{
"epoch": 2.72,
"learning_rate": 1.8828022055445232e-05,
"loss": 0.0868,
"step": 421
},
{
"epoch": 2.72,
"learning_rate": 1.8821469157260687e-05,
"loss": 0.0717,
"step": 422
},
{
"epoch": 2.73,
"learning_rate": 1.8814899137351714e-05,
"loss": 0.0505,
"step": 423
},
{
"epoch": 2.74,
"learning_rate": 1.8808312008470174e-05,
"loss": 0.0716,
"step": 424
},
{
"epoch": 2.74,
"learning_rate": 1.880170778340112e-05,
"loss": 0.096,
"step": 425
},
{
"epoch": 2.75,
"learning_rate": 1.87950864749628e-05,
"loss": 0.0885,
"step": 426
},
{
"epoch": 2.75,
"learning_rate": 1.8788448096006603e-05,
"loss": 0.1129,
"step": 427
},
{
"epoch": 2.76,
"learning_rate": 1.8781792659417066e-05,
"loss": 0.1258,
"step": 428
},
{
"epoch": 2.77,
"learning_rate": 1.877512017811183e-05,
"loss": 0.0859,
"step": 429
},
{
"epoch": 2.77,
"learning_rate": 1.876843066504161e-05,
"loss": 0.0826,
"step": 430
},
{
"epoch": 2.78,
"learning_rate": 1.8761724133190187e-05,
"loss": 0.0815,
"step": 431
},
{
"epoch": 2.79,
"learning_rate": 1.8755000595574374e-05,
"loss": 0.0759,
"step": 432
},
{
"epoch": 2.79,
"learning_rate": 1.8748260065243985e-05,
"loss": 0.0756,
"step": 433
},
{
"epoch": 2.8,
"learning_rate": 1.8741502555281822e-05,
"loss": 0.0938,
"step": 434
},
{
"epoch": 2.81,
"learning_rate": 1.873472807880363e-05,
"loss": 0.1061,
"step": 435
},
{
"epoch": 2.81,
"learning_rate": 1.8727936648958103e-05,
"loss": 0.0759,
"step": 436
},
{
"epoch": 2.82,
"learning_rate": 1.872112827892683e-05,
"loss": 0.066,
"step": 437
},
{
"epoch": 2.83,
"learning_rate": 1.871430298192428e-05,
"loss": 0.0779,
"step": 438
},
{
"epoch": 2.83,
"learning_rate": 1.8707460771197773e-05,
"loss": 0.0805,
"step": 439
},
{
"epoch": 2.84,
"learning_rate": 1.8700601660027464e-05,
"loss": 0.0769,
"step": 440
},
{
"epoch": 2.85,
"learning_rate": 1.8693725661726305e-05,
"loss": 0.0753,
"step": 441
},
{
"epoch": 2.85,
"learning_rate": 1.868683278964003e-05,
"loss": 0.0738,
"step": 442
},
{
"epoch": 2.86,
"learning_rate": 1.8679923057147114e-05,
"loss": 0.0869,
"step": 443
},
{
"epoch": 2.86,
"learning_rate": 1.8672996477658767e-05,
"loss": 0.0701,
"step": 444
},
{
"epoch": 2.87,
"learning_rate": 1.866605306461889e-05,
"loss": 0.0842,
"step": 445
},
{
"epoch": 2.88,
"learning_rate": 1.865909283150406e-05,
"loss": 0.0715,
"step": 446
},
{
"epoch": 2.88,
"learning_rate": 1.86521157918235e-05,
"loss": 0.0935,
"step": 447
},
{
"epoch": 2.89,
"learning_rate": 1.8645121959119058e-05,
"loss": 0.066,
"step": 448
},
{
"epoch": 2.9,
"learning_rate": 1.8638111346965167e-05,
"loss": 0.1143,
"step": 449
},
{
"epoch": 2.9,
"learning_rate": 1.8631083968968827e-05,
"loss": 0.0782,
"step": 450
},
{
"epoch": 2.91,
"learning_rate": 1.8624039838769592e-05,
"loss": 0.0807,
"step": 451
},
{
"epoch": 2.92,
"learning_rate": 1.8616978970039517e-05,
"loss": 0.0758,
"step": 452
},
{
"epoch": 2.92,
"learning_rate": 1.860990137648315e-05,
"loss": 0.076,
"step": 453
},
{
"epoch": 2.93,
"learning_rate": 1.8602807071837498e-05,
"loss": 0.077,
"step": 454
},
{
"epoch": 2.94,
"learning_rate": 1.8595696069872013e-05,
"loss": 0.0927,
"step": 455
},
{
"epoch": 2.94,
"learning_rate": 1.8588568384388545e-05,
"loss": 0.0808,
"step": 456
},
{
"epoch": 2.95,
"learning_rate": 1.8581424029221322e-05,
"loss": 0.0962,
"step": 457
},
{
"epoch": 2.95,
"learning_rate": 1.8574263018236936e-05,
"loss": 0.0858,
"step": 458
},
{
"epoch": 2.96,
"learning_rate": 1.8567085365334305e-05,
"loss": 0.0698,
"step": 459
},
{
"epoch": 2.97,
"learning_rate": 1.8559891084444642e-05,
"loss": 0.0763,
"step": 460
},
{
"epoch": 2.97,
"learning_rate": 1.8552680189531443e-05,
"loss": 0.0878,
"step": 461
},
{
"epoch": 2.98,
"learning_rate": 1.8545452694590435e-05,
"loss": 0.0852,
"step": 462
},
{
"epoch": 2.99,
"learning_rate": 1.8538208613649576e-05,
"loss": 0.0749,
"step": 463
},
{
"epoch": 2.99,
"learning_rate": 1.8530947960769022e-05,
"loss": 0.0881,
"step": 464
},
{
"epoch": 2.99,
"eval_val_accuracy": 90.46251993620415,
"eval_val_average": 62.36330321805124,
"eval_val_loss": 1.2907140254974365,
"eval_val_recall": 34.26408649989835,
"eval_val_runtime": 4243.2004,
"eval_val_samples_per_second": 1.48,
"eval_val_steps_per_second": 0.006,
"step": 464
},
{
"epoch": 3.0,
"learning_rate": 1.8523670750041073e-05,
"loss": 0.1176,
"step": 465
},
{
"epoch": 3.01,
"learning_rate": 1.8516376995590185e-05,
"loss": 0.0341,
"step": 466
},
{
"epoch": 3.01,
"learning_rate": 1.850906671157292e-05,
"loss": 0.0482,
"step": 467
},
{
"epoch": 3.02,
"learning_rate": 1.8501739912177915e-05,
"loss": 0.0622,
"step": 468
},
{
"epoch": 3.03,
"learning_rate": 1.849439661162587e-05,
"loss": 0.0335,
"step": 469
},
{
"epoch": 3.03,
"learning_rate": 1.8487036824169505e-05,
"loss": 0.0433,
"step": 470
},
{
"epoch": 3.04,
"learning_rate": 1.847966056409355e-05,
"loss": 0.0361,
"step": 471
},
{
"epoch": 3.05,
"learning_rate": 1.84722678457147e-05,
"loss": 0.0313,
"step": 472
},
{
"epoch": 3.05,
"learning_rate": 1.8464858683381593e-05,
"loss": 0.0471,
"step": 473
},
{
"epoch": 3.06,
"learning_rate": 1.8457433091474785e-05,
"loss": 0.0321,
"step": 474
},
{
"epoch": 3.06,
"learning_rate": 1.844999108440672e-05,
"loss": 0.0462,
"step": 475
},
{
"epoch": 3.07,
"learning_rate": 1.8442532676621708e-05,
"loss": 0.043,
"step": 476
},
{
"epoch": 3.08,
"learning_rate": 1.8435057882595885e-05,
"loss": 0.0355,
"step": 477
},
{
"epoch": 3.08,
"learning_rate": 1.842756671683719e-05,
"loss": 0.0356,
"step": 478
},
{
"epoch": 3.09,
"learning_rate": 1.8420059193885345e-05,
"loss": 0.0536,
"step": 479
},
{
"epoch": 3.1,
"learning_rate": 1.8412535328311813e-05,
"loss": 0.0237,
"step": 480
},
{
"epoch": 3.1,
"learning_rate": 1.8404995134719784e-05,
"loss": 0.0436,
"step": 481
},
{
"epoch": 3.11,
"learning_rate": 1.839743862774413e-05,
"loss": 0.0393,
"step": 482
},
{
"epoch": 3.12,
"learning_rate": 1.838986582205139e-05,
"loss": 0.0334,
"step": 483
},
{
"epoch": 3.12,
"learning_rate": 1.838227673233975e-05,
"loss": 0.0354,
"step": 484
},
{
"epoch": 3.13,
"learning_rate": 1.8374671373338973e-05,
"loss": 0.0361,
"step": 485
},
{
"epoch": 3.14,
"learning_rate": 1.8367049759810434e-05,
"loss": 0.0519,
"step": 486
},
{
"epoch": 3.14,
"learning_rate": 1.8359411906547026e-05,
"loss": 0.0507,
"step": 487
},
{
"epoch": 3.15,
"learning_rate": 1.8351757828373183e-05,
"loss": 0.0398,
"step": 488
},
{
"epoch": 3.15,
"learning_rate": 1.8344087540144817e-05,
"loss": 0.0661,
"step": 489
},
{
"epoch": 3.16,
"learning_rate": 1.8336401056749312e-05,
"loss": 0.0385,
"step": 490
},
{
"epoch": 3.17,
"learning_rate": 1.832869839310548e-05,
"loss": 0.0464,
"step": 491
},
{
"epoch": 3.17,
"learning_rate": 1.8320979564163535e-05,
"loss": 0.0526,
"step": 492
},
{
"epoch": 3.18,
"learning_rate": 1.8313244584905076e-05,
"loss": 0.0492,
"step": 493
},
{
"epoch": 3.19,
"learning_rate": 1.830549347034304e-05,
"loss": 0.0374,
"step": 494
},
{
"epoch": 3.19,
"learning_rate": 1.8297726235521686e-05,
"loss": 0.0451,
"step": 495
},
{
"epoch": 3.2,
"learning_rate": 1.828994289551655e-05,
"loss": 0.0444,
"step": 496
},
{
"epoch": 3.21,
"learning_rate": 1.8282143465434447e-05,
"loss": 0.0448,
"step": 497
},
{
"epoch": 3.21,
"learning_rate": 1.8274327960413406e-05,
"loss": 0.042,
"step": 498
},
{
"epoch": 3.22,
"learning_rate": 1.826649639562266e-05,
"loss": 0.0448,
"step": 499
},
{
"epoch": 3.23,
"learning_rate": 1.8258648786262612e-05,
"loss": 0.037,
"step": 500
},
{
"epoch": 3.23,
"learning_rate": 1.8250785147564805e-05,
"loss": 0.0566,
"step": 501
},
{
"epoch": 3.24,
"learning_rate": 1.8242905494791902e-05,
"loss": 0.052,
"step": 502
},
{
"epoch": 3.25,
"learning_rate": 1.8235009843237635e-05,
"loss": 0.0629,
"step": 503
},
{
"epoch": 3.25,
"learning_rate": 1.82270982082268e-05,
"loss": 0.0454,
"step": 504
},
{
"epoch": 3.26,
"learning_rate": 1.821917060511521e-05,
"loss": 0.0453,
"step": 505
},
{
"epoch": 3.26,
"learning_rate": 1.8211227049289664e-05,
"loss": 0.026,
"step": 506
},
{
"epoch": 3.27,
"learning_rate": 1.8203267556167937e-05,
"loss": 0.0256,
"step": 507
},
{
"epoch": 3.28,
"learning_rate": 1.8195292141198732e-05,
"loss": 0.0363,
"step": 508
},
{
"epoch": 3.28,
"learning_rate": 1.8187300819861656e-05,
"loss": 0.0582,
"step": 509
},
{
"epoch": 3.29,
"learning_rate": 1.8179293607667177e-05,
"loss": 0.0329,
"step": 510
},
{
"epoch": 3.3,
"learning_rate": 1.8171270520156625e-05,
"loss": 0.0219,
"step": 511
},
{
"epoch": 3.3,
"learning_rate": 1.816323157290213e-05,
"loss": 0.0573,
"step": 512
},
{
"epoch": 3.31,
"learning_rate": 1.8155176781506603e-05,
"loss": 0.0422,
"step": 513
},
{
"epoch": 3.32,
"learning_rate": 1.8147106161603726e-05,
"loss": 0.0303,
"step": 514
},
{
"epoch": 3.32,
"learning_rate": 1.813901972885787e-05,
"loss": 0.0259,
"step": 515
},
{
"epoch": 3.33,
"learning_rate": 1.813091749896413e-05,
"loss": 0.048,
"step": 516
},
{
"epoch": 3.34,
"learning_rate": 1.812279948764824e-05,
"loss": 0.0458,
"step": 517
},
{
"epoch": 3.34,
"learning_rate": 1.8114665710666578e-05,
"loss": 0.0443,
"step": 518
},
{
"epoch": 3.35,
"learning_rate": 1.810651618380611e-05,
"loss": 0.064,
"step": 519
},
{
"epoch": 3.35,
"learning_rate": 1.8098350922884384e-05,
"loss": 0.0506,
"step": 520
},
{
"epoch": 3.36,
"learning_rate": 1.8090169943749477e-05,
"loss": 0.0462,
"step": 521
},
{
"epoch": 3.37,
"learning_rate": 1.8081973262279975e-05,
"loss": 0.045,
"step": 522
},
{
"epoch": 3.37,
"learning_rate": 1.8073760894384942e-05,
"loss": 0.05,
"step": 523
},
{
"epoch": 3.38,
"learning_rate": 1.8065532856003893e-05,
"loss": 0.0368,
"step": 524
},
{
"epoch": 3.39,
"learning_rate": 1.805728916310675e-05,
"loss": 0.0755,
"step": 525
},
{
"epoch": 3.39,
"learning_rate": 1.804902983169382e-05,
"loss": 0.0483,
"step": 526
},
{
"epoch": 3.4,
"learning_rate": 1.8040754877795774e-05,
"loss": 0.0543,
"step": 527
},
{
"epoch": 3.41,
"learning_rate": 1.803246431747359e-05,
"loss": 0.0432,
"step": 528
},
{
"epoch": 3.41,
"learning_rate": 1.8024158166818542e-05,
"loss": 0.0362,
"step": 529
},
{
"epoch": 3.42,
"learning_rate": 1.8015836441952173e-05,
"loss": 0.0442,
"step": 530
},
{
"epoch": 3.43,
"learning_rate": 1.8007499159026235e-05,
"loss": 0.0421,
"step": 531
},
{
"epoch": 3.43,
"learning_rate": 1.79991463342227e-05,
"loss": 0.0594,
"step": 532
},
{
"epoch": 3.44,
"learning_rate": 1.799077798375368e-05,
"loss": 0.0368,
"step": 533
},
{
"epoch": 3.45,
"learning_rate": 1.7982394123861448e-05,
"loss": 0.0372,
"step": 534
},
{
"epoch": 3.45,
"learning_rate": 1.7973994770818357e-05,
"loss": 0.0377,
"step": 535
},
{
"epoch": 3.46,
"learning_rate": 1.7965579940926843e-05,
"loss": 0.0389,
"step": 536
},
{
"epoch": 3.46,
"learning_rate": 1.7957149650519378e-05,
"loss": 0.0309,
"step": 537
},
{
"epoch": 3.47,
"learning_rate": 1.794870391595844e-05,
"loss": 0.0466,
"step": 538
},
{
"epoch": 3.48,
"learning_rate": 1.7940242753636485e-05,
"loss": 0.0301,
"step": 539
},
{
"epoch": 3.48,
"learning_rate": 1.7931766179975914e-05,
"loss": 0.0413,
"step": 540
},
{
"epoch": 3.49,
"learning_rate": 1.7923274211429036e-05,
"loss": 0.0434,
"step": 541
},
{
"epoch": 3.5,
"learning_rate": 1.7914766864478045e-05,
"loss": 0.0307,
"step": 542
},
{
"epoch": 3.5,
"learning_rate": 1.790624415563498e-05,
"loss": 0.0636,
"step": 543
},
{
"epoch": 3.51,
"learning_rate": 1.78977061014417e-05,
"loss": 0.0626,
"step": 544
},
{
"epoch": 3.52,
"learning_rate": 1.7889152718469836e-05,
"loss": 0.0387,
"step": 545
},
{
"epoch": 3.52,
"learning_rate": 1.7880584023320788e-05,
"loss": 0.0405,
"step": 546
},
{
"epoch": 3.53,
"learning_rate": 1.7872000032625664e-05,
"loss": 0.0444,
"step": 547
},
{
"epoch": 3.54,
"learning_rate": 1.7863400763045266e-05,
"loss": 0.0333,
"step": 548
},
{
"epoch": 3.54,
"learning_rate": 1.7854786231270046e-05,
"loss": 0.0435,
"step": 549
},
{
"epoch": 3.55,
"learning_rate": 1.7846156454020077e-05,
"loss": 0.0451,
"step": 550
},
{
"epoch": 3.55,
"learning_rate": 1.783751144804503e-05,
"loss": 0.0508,
"step": 551
},
{
"epoch": 3.56,
"learning_rate": 1.782885123012413e-05,
"loss": 0.0551,
"step": 552
},
{
"epoch": 3.57,
"learning_rate": 1.7820175817066125e-05,
"loss": 0.0354,
"step": 553
},
{
"epoch": 3.57,
"learning_rate": 1.7811485225709255e-05,
"loss": 0.0291,
"step": 554
},
{
"epoch": 3.58,
"learning_rate": 1.7802779472921223e-05,
"loss": 0.0319,
"step": 555
},
{
"epoch": 3.59,
"learning_rate": 1.779405857559916e-05,
"loss": 0.0375,
"step": 556
},
{
"epoch": 3.59,
"learning_rate": 1.7785322550669584e-05,
"loss": 0.0368,
"step": 557
},
{
"epoch": 3.6,
"learning_rate": 1.7776571415088387e-05,
"loss": 0.0561,
"step": 558
},
{
"epoch": 3.61,
"learning_rate": 1.7767805185840773e-05,
"loss": 0.0444,
"step": 559
},
{
"epoch": 3.61,
"learning_rate": 1.7759023879941258e-05,
"loss": 0.0472,
"step": 560
},
{
"epoch": 3.62,
"learning_rate": 1.7750227514433608e-05,
"loss": 0.0366,
"step": 561
},
{
"epoch": 3.63,
"learning_rate": 1.7741416106390828e-05,
"loss": 0.0351,
"step": 562
},
{
"epoch": 3.63,
"learning_rate": 1.7732589672915107e-05,
"loss": 0.0359,
"step": 563
},
{
"epoch": 3.64,
"learning_rate": 1.7723748231137814e-05,
"loss": 0.036,
"step": 564
},
{
"epoch": 3.65,
"learning_rate": 1.7714891798219432e-05,
"loss": 0.0274,
"step": 565
},
{
"epoch": 3.65,
"learning_rate": 1.7706020391349548e-05,
"loss": 0.0404,
"step": 566
},
{
"epoch": 3.66,
"learning_rate": 1.7697134027746814e-05,
"loss": 0.0391,
"step": 567
},
{
"epoch": 3.66,
"learning_rate": 1.7688232724658908e-05,
"loss": 0.0464,
"step": 568
},
{
"epoch": 3.67,
"learning_rate": 1.7679316499362503e-05,
"loss": 0.0394,
"step": 569
},
{
"epoch": 3.68,
"learning_rate": 1.7670385369163242e-05,
"loss": 0.0263,
"step": 570
},
{
"epoch": 3.68,
"learning_rate": 1.7661439351395687e-05,
"loss": 0.054,
"step": 571
},
{
"epoch": 3.69,
"learning_rate": 1.7652478463423303e-05,
"loss": 0.0341,
"step": 572
},
{
"epoch": 3.7,
"learning_rate": 1.7643502722638417e-05,
"loss": 0.0335,
"step": 573
},
{
"epoch": 3.7,
"learning_rate": 1.7634512146462176e-05,
"loss": 0.0346,
"step": 574
},
{
"epoch": 3.71,
"learning_rate": 1.7625506752344532e-05,
"loss": 0.0512,
"step": 575
},
{
"epoch": 3.72,
"learning_rate": 1.7616486557764187e-05,
"loss": 0.0578,
"step": 576
},
{
"epoch": 3.72,
"learning_rate": 1.760745158022858e-05,
"loss": 0.0533,
"step": 577
},
{
"epoch": 3.73,
"learning_rate": 1.7598401837273837e-05,
"loss": 0.0397,
"step": 578
},
{
"epoch": 3.74,
"learning_rate": 1.7589337346464738e-05,
"loss": 0.0357,
"step": 579
},
{
"epoch": 3.74,
"learning_rate": 1.7580258125394693e-05,
"loss": 0.0397,
"step": 580
},
{
"epoch": 3.75,
"learning_rate": 1.7571164191685704e-05,
"loss": 0.0318,
"step": 581
},
{
"epoch": 3.75,
"learning_rate": 1.7562055562988322e-05,
"loss": 0.0269,
"step": 582
},
{
"epoch": 3.76,
"learning_rate": 1.755293225698163e-05,
"loss": 0.0433,
"step": 583
},
{
"epoch": 3.77,
"learning_rate": 1.7543794291373184e-05,
"loss": 0.0415,
"step": 584
},
{
"epoch": 3.77,
"learning_rate": 1.7534641683899008e-05,
"loss": 0.0345,
"step": 585
},
{
"epoch": 3.78,
"learning_rate": 1.7525474452323535e-05,
"loss": 0.0403,
"step": 586
},
{
"epoch": 3.79,
"learning_rate": 1.7516292614439586e-05,
"loss": 0.0704,
"step": 587
},
{
"epoch": 3.79,
"learning_rate": 1.7507096188068327e-05,
"loss": 0.0382,
"step": 588
},
{
"epoch": 3.8,
"learning_rate": 1.7497885191059248e-05,
"loss": 0.0403,
"step": 589
},
{
"epoch": 3.81,
"learning_rate": 1.748865964129011e-05,
"loss": 0.0323,
"step": 590
},
{
"epoch": 3.81,
"learning_rate": 1.7479419556666925e-05,
"loss": 0.0271,
"step": 591
},
{
"epoch": 3.82,
"learning_rate": 1.747016495512392e-05,
"loss": 0.0463,
"step": 592
},
{
"epoch": 3.83,
"learning_rate": 1.746089585462348e-05,
"loss": 0.0384,
"step": 593
},
{
"epoch": 3.83,
"learning_rate": 1.7451612273156155e-05,
"loss": 0.0309,
"step": 594
},
{
"epoch": 3.84,
"learning_rate": 1.7442314228740586e-05,
"loss": 0.028,
"step": 595
},
{
"epoch": 3.85,
"learning_rate": 1.7433001739423493e-05,
"loss": 0.0519,
"step": 596
},
{
"epoch": 3.85,
"learning_rate": 1.7423674823279625e-05,
"loss": 0.0346,
"step": 597
},
{
"epoch": 3.86,
"learning_rate": 1.7414333498411734e-05,
"loss": 0.0515,
"step": 598
},
{
"epoch": 3.86,
"learning_rate": 1.7404977782950548e-05,
"loss": 0.0493,
"step": 599
},
{
"epoch": 3.87,
"learning_rate": 1.739560769505471e-05,
"loss": 0.045,
"step": 600
},
{
"epoch": 3.88,
"learning_rate": 1.7386223252910773e-05,
"loss": 0.0589,
"step": 601
},
{
"epoch": 3.88,
"learning_rate": 1.737682447473314e-05,
"loss": 0.0465,
"step": 602
},
{
"epoch": 3.89,
"learning_rate": 1.736741137876405e-05,
"loss": 0.0275,
"step": 603
},
{
"epoch": 3.9,
"learning_rate": 1.735798398327352e-05,
"loss": 0.0299,
"step": 604
},
{
"epoch": 3.9,
"learning_rate": 1.7348542306559326e-05,
"loss": 0.036,
"step": 605
},
{
"epoch": 3.91,
"learning_rate": 1.733908636694696e-05,
"loss": 0.0457,
"step": 606
},
{
"epoch": 3.92,
"learning_rate": 1.7329616182789606e-05,
"loss": 0.0484,
"step": 607
},
{
"epoch": 3.92,
"learning_rate": 1.7320131772468087e-05,
"loss": 0.0382,
"step": 608
},
{
"epoch": 3.93,
"learning_rate": 1.731063315439084e-05,
"loss": 0.0393,
"step": 609
},
{
"epoch": 3.94,
"learning_rate": 1.7301120346993875e-05,
"loss": 0.0379,
"step": 610
},
{
"epoch": 3.94,
"learning_rate": 1.729159336874075e-05,
"loss": 0.0648,
"step": 611
},
{
"epoch": 3.95,
"learning_rate": 1.7282052238122524e-05,
"loss": 0.037,
"step": 612
},
{
"epoch": 3.95,
"learning_rate": 1.7272496973657715e-05,
"loss": 0.0555,
"step": 613
},
{
"epoch": 3.96,
"learning_rate": 1.726292759389229e-05,
"loss": 0.0379,
"step": 614
},
{
"epoch": 3.97,
"learning_rate": 1.7253344117399603e-05,
"loss": 0.0405,
"step": 615
},
{
"epoch": 3.97,
"learning_rate": 1.7243746562780367e-05,
"loss": 0.0314,
"step": 616
},
{
"epoch": 3.98,
"learning_rate": 1.7234134948662625e-05,
"loss": 0.0335,
"step": 617
},
{
"epoch": 3.99,
"learning_rate": 1.7224509293701707e-05,
"loss": 0.0355,
"step": 618
},
{
"epoch": 3.99,
"learning_rate": 1.7214869616580187e-05,
"loss": 0.0488,
"step": 619
},
{
"epoch": 4.0,
"learning_rate": 1.720521593600787e-05,
"loss": 0.0858,
"step": 620
},
{
"epoch": 4.01,
"learning_rate": 1.719554827072173e-05,
"loss": 0.0167,
"step": 621
},
{
"epoch": 4.01,
"learning_rate": 1.718586663948589e-05,
"loss": 0.0178,
"step": 622
},
{
"epoch": 4.02,
"learning_rate": 1.7176171061091564e-05,
"loss": 0.0343,
"step": 623
},
{
"epoch": 4.03,
"learning_rate": 1.7166461554357062e-05,
"loss": 0.0311,
"step": 624
},
{
"epoch": 4.03,
"learning_rate": 1.7156738138127708e-05,
"loss": 0.0252,
"step": 625
},
{
"epoch": 4.04,
"learning_rate": 1.7147000831275827e-05,
"loss": 0.0294,
"step": 626
},
{
"epoch": 4.05,
"learning_rate": 1.7137249652700712e-05,
"loss": 0.0253,
"step": 627
},
{
"epoch": 4.05,
"learning_rate": 1.712748462132857e-05,
"loss": 0.0156,
"step": 628
},
{
"epoch": 4.06,
"learning_rate": 1.7117705756112503e-05,
"loss": 0.0258,
"step": 629
},
{
"epoch": 4.06,
"learning_rate": 1.710791307603246e-05,
"loss": 0.0209,
"step": 630
},
{
"epoch": 4.07,
"learning_rate": 1.7098106600095204e-05,
"loss": 0.0243,
"step": 631
},
{
"epoch": 4.08,
"learning_rate": 1.708828634733427e-05,
"loss": 0.0286,
"step": 632
},
{
"epoch": 4.08,
"learning_rate": 1.7078452336809944e-05,
"loss": 0.0283,
"step": 633
},
{
"epoch": 4.09,
"learning_rate": 1.7068604587609197e-05,
"loss": 0.0215,
"step": 634
},
{
"epoch": 4.1,
"learning_rate": 1.7058743118845686e-05,
"loss": 0.0221,
"step": 635
},
{
"epoch": 4.1,
"learning_rate": 1.7048867949659685e-05,
"loss": 0.0243,
"step": 636
},
{
"epoch": 4.11,
"learning_rate": 1.7038979099218055e-05,
"loss": 0.0226,
"step": 637
},
{
"epoch": 4.12,
"learning_rate": 1.702907658671422e-05,
"loss": 0.0419,
"step": 638
},
{
"epoch": 4.12,
"learning_rate": 1.7019160431368122e-05,
"loss": 0.043,
"step": 639
},
{
"epoch": 4.13,
"learning_rate": 1.700923065242617e-05,
"loss": 0.0188,
"step": 640
},
{
"epoch": 4.14,
"learning_rate": 1.6999287269161232e-05,
"loss": 0.0194,
"step": 641
},
{
"epoch": 4.14,
"learning_rate": 1.6989330300872576e-05,
"loss": 0.0431,
"step": 642
},
{
"epoch": 4.15,
"learning_rate": 1.697935976688582e-05,
"loss": 0.0351,
"step": 643
},
{
"epoch": 4.15,
"learning_rate": 1.696937568655294e-05,
"loss": 0.028,
"step": 644
},
{
"epoch": 4.16,
"learning_rate": 1.6959378079252176e-05,
"loss": 0.0288,
"step": 645
},
{
"epoch": 4.17,
"learning_rate": 1.694936696438805e-05,
"loss": 0.0162,
"step": 646
},
{
"epoch": 4.17,
"learning_rate": 1.6939342361391283e-05,
"loss": 0.0178,
"step": 647
},
{
"epoch": 4.18,
"learning_rate": 1.6929304289718778e-05,
"loss": 0.029,
"step": 648
},
{
"epoch": 4.19,
"learning_rate": 1.6919252768853586e-05,
"loss": 0.0205,
"step": 649
},
{
"epoch": 4.19,
"learning_rate": 1.6909187818304854e-05,
"loss": 0.0202,
"step": 650
},
{
"epoch": 4.2,
"learning_rate": 1.68991094576078e-05,
"loss": 0.0364,
"step": 651
},
{
"epoch": 4.21,
"learning_rate": 1.688901770632367e-05,
"loss": 0.0231,
"step": 652
},
{
"epoch": 4.21,
"learning_rate": 1.6878912584039698e-05,
"loss": 0.0302,
"step": 653
},
{
"epoch": 4.22,
"learning_rate": 1.686879411036907e-05,
"loss": 0.0153,
"step": 654
},
{
"epoch": 4.23,
"learning_rate": 1.6858662304950886e-05,
"loss": 0.0304,
"step": 655
},
{
"epoch": 4.23,
"learning_rate": 1.684851718745012e-05,
"loss": 0.0266,
"step": 656
},
{
"epoch": 4.24,
"learning_rate": 1.6838358777557587e-05,
"loss": 0.0394,
"step": 657
},
{
"epoch": 4.25,
"learning_rate": 1.6828187094989904e-05,
"loss": 0.021,
"step": 658
},
{
"epoch": 4.25,
"learning_rate": 1.681800215948944e-05,
"loss": 0.0245,
"step": 659
},
{
"epoch": 4.26,
"learning_rate": 1.6807803990824295e-05,
"loss": 0.0306,
"step": 660
},
{
"epoch": 4.26,
"learning_rate": 1.6797592608788246e-05,
"loss": 0.025,
"step": 661
},
{
"epoch": 4.27,
"learning_rate": 1.6787368033200727e-05,
"loss": 0.0309,
"step": 662
},
{
"epoch": 4.28,
"learning_rate": 1.6777130283906773e-05,
"loss": 0.0469,
"step": 663
},
{
"epoch": 4.28,
"learning_rate": 1.6766879380776983e-05,
"loss": 0.0277,
"step": 664
},
{
"epoch": 4.29,
"learning_rate": 1.6756615343707494e-05,
"loss": 0.0253,
"step": 665
},
{
"epoch": 4.3,
"learning_rate": 1.6746338192619934e-05,
"loss": 0.0476,
"step": 666
},
{
"epoch": 4.3,
"learning_rate": 1.673604794746138e-05,
"loss": 0.031,
"step": 667
},
{
"epoch": 4.31,
"learning_rate": 1.6725744628204332e-05,
"loss": 0.021,
"step": 668
},
{
"epoch": 4.32,
"learning_rate": 1.6715428254846653e-05,
"loss": 0.0215,
"step": 669
},
{
"epoch": 4.32,
"learning_rate": 1.6705098847411552e-05,
"loss": 0.0173,
"step": 670
},
{
"epoch": 4.33,
"learning_rate": 1.6694756425947534e-05,
"loss": 0.0302,
"step": 671
},
{
"epoch": 4.34,
"learning_rate": 1.6684401010528367e-05,
"loss": 0.0287,
"step": 672
},
{
"epoch": 4.34,
"learning_rate": 1.6674032621253033e-05,
"loss": 0.0237,
"step": 673
},
{
"epoch": 4.35,
"learning_rate": 1.6663651278245692e-05,
"loss": 0.0287,
"step": 674
},
{
"epoch": 4.35,
"learning_rate": 1.6653257001655652e-05,
"loss": 0.0213,
"step": 675
},
{
"epoch": 4.36,
"learning_rate": 1.6642849811657335e-05,
"loss": 0.0167,
"step": 676
},
{
"epoch": 4.37,
"learning_rate": 1.6632429728450204e-05,
"loss": 0.026,
"step": 677
},
{
"epoch": 4.37,
"learning_rate": 1.6621996772258763e-05,
"loss": 0.0267,
"step": 678
},
{
"epoch": 4.38,
"learning_rate": 1.661155096333249e-05,
"loss": 0.029,
"step": 679
},
{
"epoch": 4.39,
"learning_rate": 1.6601092321945823e-05,
"loss": 0.0235,
"step": 680
},
{
"epoch": 4.39,
"learning_rate": 1.6590620868398098e-05,
"loss": 0.0185,
"step": 681
},
{
"epoch": 4.4,
"learning_rate": 1.658013662301351e-05,
"loss": 0.0212,
"step": 682
},
{
"epoch": 4.41,
"learning_rate": 1.6569639606141105e-05,
"loss": 0.0235,
"step": 683
},
{
"epoch": 4.41,
"learning_rate": 1.6559129838154695e-05,
"loss": 0.0276,
"step": 684
},
{
"epoch": 4.42,
"learning_rate": 1.6548607339452853e-05,
"loss": 0.0314,
"step": 685
},
{
"epoch": 4.43,
"learning_rate": 1.6538072130458853e-05,
"loss": 0.0291,
"step": 686
},
{
"epoch": 4.43,
"learning_rate": 1.652752423162065e-05,
"loss": 0.0288,
"step": 687
},
{
"epoch": 4.44,
"learning_rate": 1.6516963663410814e-05,
"loss": 0.0295,
"step": 688
},
{
"epoch": 4.45,
"learning_rate": 1.6506390446326523e-05,
"loss": 0.0195,
"step": 689
},
{
"epoch": 4.45,
"learning_rate": 1.6495804600889486e-05,
"loss": 0.0265,
"step": 690
},
{
"epoch": 4.46,
"learning_rate": 1.648520614764594e-05,
"loss": 0.0288,
"step": 691
},
{
"epoch": 4.46,
"learning_rate": 1.6474595107166584e-05,
"loss": 0.0276,
"step": 692
},
{
"epoch": 4.47,
"learning_rate": 1.6463971500046545e-05,
"loss": 0.0333,
"step": 693
},
{
"epoch": 4.48,
"learning_rate": 1.6453335346905345e-05,
"loss": 0.0169,
"step": 694
},
{
"epoch": 4.48,
"learning_rate": 1.644268666838686e-05,
"loss": 0.0239,
"step": 695
},
{
"epoch": 4.49,
"learning_rate": 1.6432025485159272e-05,
"loss": 0.0239,
"step": 696
},
{
"epoch": 4.5,
"learning_rate": 1.6421351817915025e-05,
"loss": 0.0276,
"step": 697
},
{
"epoch": 4.5,
"learning_rate": 1.641066568737082e-05,
"loss": 0.0152,
"step": 698
},
{
"epoch": 4.51,
"learning_rate": 1.639996711426752e-05,
"loss": 0.0402,
"step": 699
},
{
"epoch": 4.52,
"learning_rate": 1.6389256119370153e-05,
"loss": 0.0183,
"step": 700
},
{
"epoch": 4.52,
"learning_rate": 1.6378532723467852e-05,
"loss": 0.0234,
"step": 701
},
{
"epoch": 4.53,
"learning_rate": 1.6367796947373817e-05,
"loss": 0.0259,
"step": 702
},
{
"epoch": 4.54,
"learning_rate": 1.6357048811925286e-05,
"loss": 0.0297,
"step": 703
},
{
"epoch": 4.54,
"learning_rate": 1.6346288337983476e-05,
"loss": 0.033,
"step": 704
},
{
"epoch": 4.55,
"learning_rate": 1.6335515546433552e-05,
"loss": 0.0421,
"step": 705
},
{
"epoch": 4.55,
"learning_rate": 1.63247304581846e-05,
"loss": 0.0278,
"step": 706
},
{
"epoch": 4.56,
"learning_rate": 1.6313933094169552e-05,
"loss": 0.0286,
"step": 707
},
{
"epoch": 4.57,
"learning_rate": 1.6303123475345182e-05,
"loss": 0.0253,
"step": 708
},
{
"epoch": 4.57,
"learning_rate": 1.6292301622692045e-05,
"loss": 0.0357,
"step": 709
},
{
"epoch": 4.58,
"learning_rate": 1.6281467557214437e-05,
"loss": 0.0286,
"step": 710
},
{
"epoch": 4.59,
"learning_rate": 1.6270621299940367e-05,
"loss": 0.0327,
"step": 711
},
{
"epoch": 4.59,
"learning_rate": 1.62597628719215e-05,
"loss": 0.0301,
"step": 712
},
{
"epoch": 4.6,
"learning_rate": 1.6248892294233118e-05,
"loss": 0.0244,
"step": 713
},
{
"epoch": 4.61,
"learning_rate": 1.62380095879741e-05,
"loss": 0.0192,
"step": 714
},
{
"epoch": 4.61,
"learning_rate": 1.6227114774266855e-05,
"loss": 0.0291,
"step": 715
},
{
"epoch": 4.62,
"learning_rate": 1.621620787425729e-05,
"loss": 0.0204,
"step": 716
},
{
"epoch": 4.63,
"learning_rate": 1.620528890911478e-05,
"loss": 0.0195,
"step": 717
},
{
"epoch": 4.63,
"learning_rate": 1.61943579000321e-05,
"loss": 0.0243,
"step": 718
},
{
"epoch": 4.64,
"learning_rate": 1.6183414868225434e-05,
"loss": 0.0188,
"step": 719
},
{
"epoch": 4.65,
"learning_rate": 1.6172459834934257e-05,
"loss": 0.0312,
"step": 720
},
{
"epoch": 4.65,
"learning_rate": 1.616149282142137e-05,
"loss": 0.0379,
"step": 721
},
{
"epoch": 4.66,
"learning_rate": 1.6150513848972825e-05,
"loss": 0.0254,
"step": 722
},
{
"epoch": 4.66,
"learning_rate": 1.6139522938897862e-05,
"loss": 0.0253,
"step": 723
},
{
"epoch": 4.67,
"learning_rate": 1.6128520112528918e-05,
"loss": 0.0278,
"step": 724
},
{
"epoch": 4.68,
"learning_rate": 1.6117505391221543e-05,
"loss": 0.0557,
"step": 725
},
{
"epoch": 4.68,
"learning_rate": 1.6106478796354382e-05,
"loss": 0.0294,
"step": 726
},
{
"epoch": 4.69,
"learning_rate": 1.6095440349329118e-05,
"loss": 0.0303,
"step": 727
},
{
"epoch": 4.7,
"learning_rate": 1.608439007157044e-05,
"loss": 0.0289,
"step": 728
},
{
"epoch": 4.7,
"learning_rate": 1.6073327984526003e-05,
"loss": 0.0184,
"step": 729
},
{
"epoch": 4.71,
"learning_rate": 1.6062254109666383e-05,
"loss": 0.0171,
"step": 730
},
{
"epoch": 4.72,
"learning_rate": 1.6051168468485028e-05,
"loss": 0.041,
"step": 731
},
{
"epoch": 4.72,
"learning_rate": 1.6040071082498234e-05,
"loss": 0.0313,
"step": 732
},
{
"epoch": 4.73,
"learning_rate": 1.6028961973245087e-05,
"loss": 0.0276,
"step": 733
},
{
"epoch": 4.74,
"learning_rate": 1.6017841162287423e-05,
"loss": 0.0224,
"step": 734
},
{
"epoch": 4.74,
"learning_rate": 1.6006708671209794e-05,
"loss": 0.0212,
"step": 735
},
{
"epoch": 4.75,
"learning_rate": 1.5995564521619428e-05,
"loss": 0.0221,
"step": 736
},
{
"epoch": 4.75,
"learning_rate": 1.5984408735146167e-05,
"loss": 0.0272,
"step": 737
},
{
"epoch": 4.76,
"learning_rate": 1.597324133344246e-05,
"loss": 0.0194,
"step": 738
},
{
"epoch": 4.77,
"learning_rate": 1.5962062338183275e-05,
"loss": 0.0229,
"step": 739
},
{
"epoch": 4.77,
"learning_rate": 1.59508717710661e-05,
"loss": 0.0293,
"step": 740
},
{
"epoch": 4.78,
"learning_rate": 1.5939669653810882e-05,
"loss": 0.0293,
"step": 741
},
{
"epoch": 4.79,
"learning_rate": 1.592845600815997e-05,
"loss": 0.021,
"step": 742
},
{
"epoch": 4.79,
"learning_rate": 1.5917230855878115e-05,
"loss": 0.0301,
"step": 743
},
{
"epoch": 4.8,
"learning_rate": 1.5905994218752378e-05,
"loss": 0.0426,
"step": 744
},
{
"epoch": 4.81,
"learning_rate": 1.5894746118592122e-05,
"loss": 0.0324,
"step": 745
},
{
"epoch": 4.81,
"learning_rate": 1.5883486577228955e-05,
"loss": 0.0251,
"step": 746
},
{
"epoch": 4.82,
"learning_rate": 1.5872215616516696e-05,
"loss": 0.0331,
"step": 747
},
{
"epoch": 4.83,
"learning_rate": 1.5860933258331323e-05,
"loss": 0.0362,
"step": 748
},
{
"epoch": 4.83,
"learning_rate": 1.5849639524570938e-05,
"loss": 0.0409,
"step": 749
},
{
"epoch": 4.84,
"learning_rate": 1.5838334437155724e-05,
"loss": 0.0268,
"step": 750
},
{
"epoch": 4.85,
"learning_rate": 1.5827018018027892e-05,
"loss": 0.0378,
"step": 751
},
{
"epoch": 4.85,
"learning_rate": 1.581569028915166e-05,
"loss": 0.0251,
"step": 752
},
{
"epoch": 4.86,
"learning_rate": 1.5804351272513186e-05,
"loss": 0.0238,
"step": 753
},
{
"epoch": 4.86,
"learning_rate": 1.579300099012054e-05,
"loss": 0.0209,
"step": 754
},
{
"epoch": 4.87,
"learning_rate": 1.578163946400366e-05,
"loss": 0.0243,
"step": 755
},
{
"epoch": 4.88,
"learning_rate": 1.577026671621431e-05,
"loss": 0.0332,
"step": 756
},
{
"epoch": 4.88,
"learning_rate": 1.5758882768826026e-05,
"loss": 0.0256,
"step": 757
},
{
"epoch": 4.89,
"learning_rate": 1.5747487643934083e-05,
"loss": 0.0316,
"step": 758
},
{
"epoch": 4.9,
"learning_rate": 1.5736081363655457e-05,
"loss": 0.0256,
"step": 759
},
{
"epoch": 4.9,
"learning_rate": 1.5724663950128777e-05,
"loss": 0.017,
"step": 760
},
{
"epoch": 4.91,
"learning_rate": 1.571323542551426e-05,
"loss": 0.038,
"step": 761
},
{
"epoch": 4.92,
"learning_rate": 1.5701795811993724e-05,
"loss": 0.0167,
"step": 762
},
{
"epoch": 4.92,
"learning_rate": 1.5690345131770474e-05,
"loss": 0.0451,
"step": 763
},
{
"epoch": 4.93,
"learning_rate": 1.5678883407069316e-05,
"loss": 0.0392,
"step": 764
},
{
"epoch": 4.94,
"learning_rate": 1.566741066013649e-05,
"loss": 0.0308,
"step": 765
},
{
"epoch": 4.94,
"learning_rate": 1.5655926913239622e-05,
"loss": 0.022,
"step": 766
},
{
"epoch": 4.95,
"learning_rate": 1.5644432188667695e-05,
"loss": 0.0448,
"step": 767
},
{
"epoch": 4.95,
"learning_rate": 1.5632926508730992e-05,
"loss": 0.0368,
"step": 768
},
{
"epoch": 4.96,
"learning_rate": 1.5621409895761063e-05,
"loss": 0.0352,
"step": 769
},
{
"epoch": 4.97,
"learning_rate": 1.5609882372110683e-05,
"loss": 0.0358,
"step": 770
},
{
"epoch": 4.97,
"learning_rate": 1.5598343960153798e-05,
"loss": 0.0257,
"step": 771
},
{
"epoch": 4.98,
"learning_rate": 1.558679468228548e-05,
"loss": 0.0323,
"step": 772
},
{
"epoch": 4.99,
"learning_rate": 1.557523456092191e-05,
"loss": 0.0244,
"step": 773
},
{
"epoch": 4.99,
"learning_rate": 1.55636636185003e-05,
"loss": 0.0294,
"step": 774
},
{
"epoch": 5.0,
"learning_rate": 1.555208187747887e-05,
"loss": 0.0233,
"step": 775
},
{
"epoch": 5.01,
"learning_rate": 1.55404893603368e-05,
"loss": 0.0112,
"step": 776
},
{
"epoch": 5.01,
"learning_rate": 1.552888608957418e-05,
"loss": 0.0348,
"step": 777
},
{
"epoch": 5.02,
"learning_rate": 1.551727208771197e-05,
"loss": 0.0184,
"step": 778
},
{
"epoch": 5.03,
"learning_rate": 1.550564737729198e-05,
"loss": 0.0179,
"step": 779
},
{
"epoch": 5.03,
"learning_rate": 1.549401198087677e-05,
"loss": 0.0171,
"step": 780
},
{
"epoch": 5.04,
"learning_rate": 1.5482365921049675e-05,
"loss": 0.0141,
"step": 781
},
{
"epoch": 5.05,
"learning_rate": 1.5470709220414702e-05,
"loss": 0.0185,
"step": 782
},
{
"epoch": 5.05,
"learning_rate": 1.5459041901596524e-05,
"loss": 0.0184,
"step": 783
},
{
"epoch": 5.06,
"learning_rate": 1.5447363987240413e-05,
"loss": 0.0247,
"step": 784
},
{
"epoch": 5.06,
"learning_rate": 1.5435675500012212e-05,
"loss": 0.0163,
"step": 785
},
{
"epoch": 5.07,
"learning_rate": 1.542397646259829e-05,
"loss": 0.0122,
"step": 786
},
{
"epoch": 5.08,
"learning_rate": 1.5412266897705486e-05,
"loss": 0.0179,
"step": 787
},
{
"epoch": 5.08,
"learning_rate": 1.5400546828061068e-05,
"loss": 0.0181,
"step": 788
},
{
"epoch": 5.09,
"learning_rate": 1.5388816276412707e-05,
"loss": 0.02,
"step": 789
},
{
"epoch": 5.1,
"learning_rate": 1.5377075265528407e-05,
"loss": 0.0246,
"step": 790
},
{
"epoch": 5.1,
"learning_rate": 1.536532381819647e-05,
"loss": 0.0212,
"step": 791
},
{
"epoch": 5.11,
"learning_rate": 1.535356195722547e-05,
"loss": 0.0213,
"step": 792
},
{
"epoch": 5.12,
"learning_rate": 1.5341789705444173e-05,
"loss": 0.0152,
"step": 793
},
{
"epoch": 5.12,
"learning_rate": 1.533000708570153e-05,
"loss": 0.0139,
"step": 794
},
{
"epoch": 5.13,
"learning_rate": 1.5318214120866597e-05,
"loss": 0.0166,
"step": 795
},
{
"epoch": 5.14,
"learning_rate": 1.5306410833828534e-05,
"loss": 0.0176,
"step": 796
},
{
"epoch": 5.14,
"learning_rate": 1.5294597247496518e-05,
"loss": 0.0185,
"step": 797
},
{
"epoch": 5.15,
"learning_rate": 1.528277338479971e-05,
"loss": 0.0182,
"step": 798
},
{
"epoch": 5.15,
"learning_rate": 1.5270939268687236e-05,
"loss": 0.0241,
"step": 799
},
{
"epoch": 5.16,
"learning_rate": 1.5259094922128108e-05,
"loss": 0.0236,
"step": 800
},
{
"epoch": 5.17,
"learning_rate": 1.5247240368111203e-05,
"loss": 0.0138,
"step": 801
},
{
"epoch": 5.17,
"learning_rate": 1.5235375629645202e-05,
"loss": 0.0275,
"step": 802
},
{
"epoch": 5.18,
"learning_rate": 1.5223500729758558e-05,
"loss": 0.0272,
"step": 803
},
{
"epoch": 5.19,
"learning_rate": 1.521161569149945e-05,
"loss": 0.0222,
"step": 804
},
{
"epoch": 5.19,
"learning_rate": 1.5199720537935726e-05,
"loss": 0.011,
"step": 805
},
{
"epoch": 5.2,
"learning_rate": 1.5187815292154872e-05,
"loss": 0.0214,
"step": 806
},
{
"epoch": 5.21,
"learning_rate": 1.5175899977263963e-05,
"loss": 0.0125,
"step": 807
},
{
"epoch": 5.21,
"learning_rate": 1.5163974616389621e-05,
"loss": 0.0251,
"step": 808
},
{
"epoch": 5.22,
"learning_rate": 1.5152039232677953e-05,
"loss": 0.017,
"step": 809
},
{
"epoch": 5.23,
"learning_rate": 1.514009384929453e-05,
"loss": 0.0193,
"step": 810
},
{
"epoch": 5.23,
"learning_rate": 1.5128138489424332e-05,
"loss": 0.0246,
"step": 811
},
{
"epoch": 5.24,
"learning_rate": 1.5116173176271695e-05,
"loss": 0.0136,
"step": 812
},
{
"epoch": 5.25,
"learning_rate": 1.5104197933060285e-05,
"loss": 0.0246,
"step": 813
},
{
"epoch": 5.25,
"learning_rate": 1.5092212783033029e-05,
"loss": 0.017,
"step": 814
},
{
"epoch": 5.26,
"learning_rate": 1.5080217749452094e-05,
"loss": 0.0259,
"step": 815
},
{
"epoch": 5.26,
"learning_rate": 1.5068212855598814e-05,
"loss": 0.0197,
"step": 816
},
{
"epoch": 5.27,
"learning_rate": 1.505619812477368e-05,
"loss": 0.0155,
"step": 817
},
{
"epoch": 5.28,
"learning_rate": 1.5044173580296267e-05,
"loss": 0.0222,
"step": 818
},
{
"epoch": 5.28,
"learning_rate": 1.5032139245505191e-05,
"loss": 0.0133,
"step": 819
},
{
"epoch": 5.29,
"learning_rate": 1.5020095143758085e-05,
"loss": 0.0151,
"step": 820
},
{
"epoch": 5.3,
"learning_rate": 1.500804129843152e-05,
"loss": 0.0205,
"step": 821
},
{
"epoch": 5.3,
"learning_rate": 1.4995977732920996e-05,
"loss": 0.0209,
"step": 822
},
{
"epoch": 5.31,
"learning_rate": 1.498390447064087e-05,
"loss": 0.0147,
"step": 823
},
{
"epoch": 5.32,
"learning_rate": 1.4971821535024324e-05,
"loss": 0.0098,
"step": 824
},
{
"epoch": 5.32,
"learning_rate": 1.4959728949523307e-05,
"loss": 0.0177,
"step": 825
},
{
"epoch": 5.33,
"learning_rate": 1.4947626737608505e-05,
"loss": 0.0221,
"step": 826
},
{
"epoch": 5.34,
"learning_rate": 1.4935514922769292e-05,
"loss": 0.0208,
"step": 827
},
{
"epoch": 5.34,
"learning_rate": 1.4923393528513667e-05,
"loss": 0.0198,
"step": 828
},
{
"epoch": 5.35,
"learning_rate": 1.4911262578368233e-05,
"loss": 0.018,
"step": 829
},
{
"epoch": 5.35,
"learning_rate": 1.4899122095878137e-05,
"loss": 0.0161,
"step": 830
},
{
"epoch": 5.36,
"learning_rate": 1.4886972104607024e-05,
"loss": 0.0127,
"step": 831
},
{
"epoch": 5.37,
"learning_rate": 1.4874812628137006e-05,
"loss": 0.029,
"step": 832
},
{
"epoch": 5.37,
"learning_rate": 1.4862643690068586e-05,
"loss": 0.0214,
"step": 833
},
{
"epoch": 5.38,
"learning_rate": 1.485046531402065e-05,
"loss": 0.02,
"step": 834
},
{
"epoch": 5.39,
"learning_rate": 1.483827752363039e-05,
"loss": 0.0205,
"step": 835
},
{
"epoch": 5.39,
"learning_rate": 1.4826080342553279e-05,
"loss": 0.0178,
"step": 836
},
{
"epoch": 5.4,
"learning_rate": 1.4813873794463004e-05,
"loss": 0.022,
"step": 837
},
{
"epoch": 5.41,
"learning_rate": 1.4801657903051454e-05,
"loss": 0.0166,
"step": 838
},
{
"epoch": 5.41,
"learning_rate": 1.4789432692028629e-05,
"loss": 0.0203,
"step": 839
},
{
"epoch": 5.42,
"learning_rate": 1.477719818512263e-05,
"loss": 0.0162,
"step": 840
},
{
"epoch": 5.43,
"learning_rate": 1.4764954406079602e-05,
"loss": 0.0167,
"step": 841
},
{
"epoch": 5.43,
"learning_rate": 1.4752701378663678e-05,
"loss": 0.0197,
"step": 842
},
{
"epoch": 5.44,
"learning_rate": 1.4740439126656954e-05,
"loss": 0.0123,
"step": 843
},
{
"epoch": 5.45,
"learning_rate": 1.4728167673859415e-05,
"loss": 0.0173,
"step": 844
},
{
"epoch": 5.45,
"learning_rate": 1.4715887044088912e-05,
"loss": 0.0209,
"step": 845
},
{
"epoch": 5.46,
"learning_rate": 1.4703597261181113e-05,
"loss": 0.0201,
"step": 846
},
{
"epoch": 5.46,
"learning_rate": 1.469129834898944e-05,
"loss": 0.0208,
"step": 847
},
{
"epoch": 5.47,
"learning_rate": 1.4678990331385041e-05,
"loss": 0.0149,
"step": 848
},
{
"epoch": 5.48,
"learning_rate": 1.4666673232256738e-05,
"loss": 0.0199,
"step": 849
},
{
"epoch": 5.48,
"learning_rate": 1.4654347075510975e-05,
"loss": 0.0261,
"step": 850
},
{
"epoch": 5.49,
"learning_rate": 1.464201188507178e-05,
"loss": 0.0296,
"step": 851
},
{
"epoch": 5.5,
"learning_rate": 1.4629667684880708e-05,
"loss": 0.025,
"step": 852
},
{
"epoch": 5.5,
"learning_rate": 1.4617314498896815e-05,
"loss": 0.0239,
"step": 853
},
{
"epoch": 5.51,
"learning_rate": 1.4604952351096577e-05,
"loss": 0.017,
"step": 854
},
{
"epoch": 5.52,
"learning_rate": 1.4592581265473884e-05,
"loss": 0.0171,
"step": 855
},
{
"epoch": 5.52,
"learning_rate": 1.4580201266039966e-05,
"loss": 0.0156,
"step": 856
},
{
"epoch": 5.53,
"learning_rate": 1.4567812376823346e-05,
"loss": 0.0253,
"step": 857
},
{
"epoch": 5.54,
"learning_rate": 1.4555414621869819e-05,
"loss": 0.0273,
"step": 858
},
{
"epoch": 5.54,
"learning_rate": 1.4543008025242366e-05,
"loss": 0.0193,
"step": 859
},
{
"epoch": 5.55,
"learning_rate": 1.4530592611021146e-05,
"loss": 0.0258,
"step": 860
},
{
"epoch": 5.55,
"learning_rate": 1.4518168403303428e-05,
"loss": 0.023,
"step": 861
},
{
"epoch": 5.56,
"learning_rate": 1.4505735426203545e-05,
"loss": 0.0213,
"step": 862
},
{
"epoch": 5.57,
"learning_rate": 1.4493293703852849e-05,
"loss": 0.0172,
"step": 863
},
{
"epoch": 5.57,
"learning_rate": 1.4480843260399673e-05,
"loss": 0.0248,
"step": 864
},
{
"epoch": 5.58,
"learning_rate": 1.4468384120009273e-05,
"loss": 0.0169,
"step": 865
},
{
"epoch": 5.59,
"learning_rate": 1.4455916306863786e-05,
"loss": 0.0203,
"step": 866
},
{
"epoch": 5.59,
"learning_rate": 1.4443439845162176e-05,
"loss": 0.0173,
"step": 867
},
{
"epoch": 5.6,
"learning_rate": 1.4430954759120206e-05,
"loss": 0.0121,
"step": 868
},
{
"epoch": 5.61,
"learning_rate": 1.4418461072970366e-05,
"loss": 0.0152,
"step": 869
},
{
"epoch": 5.61,
"learning_rate": 1.4405958810961842e-05,
"loss": 0.0176,
"step": 870
},
{
"epoch": 5.62,
"learning_rate": 1.4393447997360467e-05,
"loss": 0.0198,
"step": 871
},
{
"epoch": 5.63,
"learning_rate": 1.4380928656448672e-05,
"loss": 0.0158,
"step": 872
},
{
"epoch": 5.63,
"learning_rate": 1.4368400812525434e-05,
"loss": 0.0259,
"step": 873
},
{
"epoch": 5.64,
"learning_rate": 1.4355864489906237e-05,
"loss": 0.0148,
"step": 874
},
{
"epoch": 5.65,
"learning_rate": 1.4343319712923024e-05,
"loss": 0.0393,
"step": 875
},
{
"epoch": 5.65,
"learning_rate": 1.4330766505924145e-05,
"loss": 0.0208,
"step": 876
},
{
"epoch": 5.66,
"learning_rate": 1.431820489327431e-05,
"loss": 0.0249,
"step": 877
},
{
"epoch": 5.66,
"learning_rate": 1.4305634899354544e-05,
"loss": 0.0163,
"step": 878
},
{
"epoch": 5.67,
"learning_rate": 1.4293056548562138e-05,
"loss": 0.0202,
"step": 879
},
{
"epoch": 5.68,
"learning_rate": 1.4280469865310614e-05,
"loss": 0.0255,
"step": 880
},
{
"epoch": 5.68,
"learning_rate": 1.426787487402965e-05,
"loss": 0.0243,
"step": 881
},
{
"epoch": 5.69,
"learning_rate": 1.4255271599165058e-05,
"loss": 0.0162,
"step": 882
},
{
"epoch": 5.7,
"learning_rate": 1.4242660065178734e-05,
"loss": 0.0155,
"step": 883
},
{
"epoch": 5.7,
"learning_rate": 1.4230040296548588e-05,
"loss": 0.0252,
"step": 884
},
{
"epoch": 5.71,
"learning_rate": 1.421741231776853e-05,
"loss": 0.0226,
"step": 885
},
{
"epoch": 5.72,
"learning_rate": 1.4204776153348392e-05,
"loss": 0.0155,
"step": 886
},
{
"epoch": 5.72,
"learning_rate": 1.4192131827813902e-05,
"loss": 0.0127,
"step": 887
},
{
"epoch": 5.73,
"learning_rate": 1.4179479365706619e-05,
"loss": 0.0167,
"step": 888
},
{
"epoch": 5.74,
"learning_rate": 1.4166818791583908e-05,
"loss": 0.0205,
"step": 889
},
{
"epoch": 5.74,
"learning_rate": 1.4154150130018867e-05,
"loss": 0.0162,
"step": 890
},
{
"epoch": 5.75,
"learning_rate": 1.4141473405600295e-05,
"loss": 0.0211,
"step": 891
},
{
"epoch": 5.75,
"learning_rate": 1.4128788642932639e-05,
"loss": 0.0223,
"step": 892
},
{
"epoch": 5.76,
"learning_rate": 1.4116095866635951e-05,
"loss": 0.021,
"step": 893
},
{
"epoch": 5.77,
"learning_rate": 1.4103395101345836e-05,
"loss": 0.0235,
"step": 894
},
{
"epoch": 5.77,
"learning_rate": 1.4090686371713403e-05,
"loss": 0.0112,
"step": 895
},
{
"epoch": 5.78,
"learning_rate": 1.4077969702405216e-05,
"loss": 0.0158,
"step": 896
},
{
"epoch": 5.79,
"learning_rate": 1.4065245118103254e-05,
"loss": 0.0224,
"step": 897
},
{
"epoch": 5.79,
"learning_rate": 1.4052512643504861e-05,
"loss": 0.0171,
"step": 898
},
{
"epoch": 5.8,
"learning_rate": 1.4039772303322692e-05,
"loss": 0.0268,
"step": 899
},
{
"epoch": 5.81,
"learning_rate": 1.4027024122284663e-05,
"loss": 0.028,
"step": 900
},
{
"epoch": 5.81,
"learning_rate": 1.4014268125133917e-05,
"loss": 0.0147,
"step": 901
},
{
"epoch": 5.82,
"learning_rate": 1.4001504336628761e-05,
"loss": 0.0164,
"step": 902
},
{
"epoch": 5.83,
"learning_rate": 1.3988732781542633e-05,
"loss": 0.0194,
"step": 903
},
{
"epoch": 5.83,
"learning_rate": 1.3975953484664034e-05,
"loss": 0.0197,
"step": 904
},
{
"epoch": 5.84,
"learning_rate": 1.3963166470796502e-05,
"loss": 0.0133,
"step": 905
},
{
"epoch": 5.85,
"learning_rate": 1.3950371764758543e-05,
"loss": 0.0153,
"step": 906
},
{
"epoch": 5.85,
"learning_rate": 1.3937569391383598e-05,
"loss": 0.0157,
"step": 907
},
{
"epoch": 5.86,
"learning_rate": 1.3924759375519994e-05,
"loss": 0.0169,
"step": 908
},
{
"epoch": 5.86,
"learning_rate": 1.3911941742030881e-05,
"loss": 0.0183,
"step": 909
},
{
"epoch": 5.87,
"learning_rate": 1.3899116515794204e-05,
"loss": 0.0175,
"step": 910
},
{
"epoch": 5.88,
"learning_rate": 1.388628372170264e-05,
"loss": 0.0127,
"step": 911
},
{
"epoch": 5.88,
"learning_rate": 1.3873443384663551e-05,
"loss": 0.021,
"step": 912
},
{
"epoch": 5.89,
"learning_rate": 1.3860595529598953e-05,
"loss": 0.0144,
"step": 913
},
{
"epoch": 5.9,
"learning_rate": 1.3847740181445436e-05,
"loss": 0.0237,
"step": 914
},
{
"epoch": 5.9,
"learning_rate": 1.3834877365154144e-05,
"loss": 0.0271,
"step": 915
},
{
"epoch": 5.91,
"learning_rate": 1.3822007105690711e-05,
"loss": 0.0178,
"step": 916
},
{
"epoch": 5.92,
"learning_rate": 1.3809129428035229e-05,
"loss": 0.0132,
"step": 917
},
{
"epoch": 5.92,
"learning_rate": 1.3796244357182173e-05,
"loss": 0.0195,
"step": 918
},
{
"epoch": 5.93,
"learning_rate": 1.3783351918140376e-05,
"loss": 0.0205,
"step": 919
},
{
"epoch": 5.94,
"learning_rate": 1.3770452135932967e-05,
"loss": 0.0101,
"step": 920
},
{
"epoch": 5.94,
"learning_rate": 1.3757545035597337e-05,
"loss": 0.0143,
"step": 921
},
{
"epoch": 5.95,
"learning_rate": 1.3744630642185069e-05,
"loss": 0.0161,
"step": 922
},
{
"epoch": 5.95,
"learning_rate": 1.3731708980761909e-05,
"loss": 0.017,
"step": 923
},
{
"epoch": 5.96,
"learning_rate": 1.3718780076407705e-05,
"loss": 0.0148,
"step": 924
},
{
"epoch": 5.97,
"learning_rate": 1.3705843954216367e-05,
"loss": 0.0166,
"step": 925
},
{
"epoch": 5.97,
"learning_rate": 1.3692900639295817e-05,
"loss": 0.0227,
"step": 926
},
{
"epoch": 5.98,
"learning_rate": 1.3679950156767922e-05,
"loss": 0.0166,
"step": 927
},
{
"epoch": 5.99,
"learning_rate": 1.3666992531768482e-05,
"loss": 0.0189,
"step": 928
},
{
"epoch": 5.99,
"eval_val_accuracy": 88.54864433811802,
"eval_val_average": 62.0108029129221,
"eval_val_loss": 1.3992602825164795,
"eval_val_recall": 35.472961487726174,
"eval_val_runtime": 4282.3314,
"eval_val_samples_per_second": 1.466,
"eval_val_steps_per_second": 0.006,
"step": 928
},
{
"epoch": 5.99,
"learning_rate": 1.3654027789447147e-05,
"loss": 0.0225,
"step": 929
},
{
"epoch": 6.0,
"learning_rate": 1.3641055954967377e-05,
"loss": 0.0113,
"step": 930
},
{
"epoch": 6.01,
"learning_rate": 1.362807705350641e-05,
"loss": 0.0108,
"step": 931
},
{
"epoch": 6.01,
"learning_rate": 1.3615091110255197e-05,
"loss": 0.0177,
"step": 932
},
{
"epoch": 6.02,
"learning_rate": 1.360209815041835e-05,
"loss": 0.0094,
"step": 933
},
{
"epoch": 6.03,
"learning_rate": 1.3589098199214102e-05,
"loss": 0.0154,
"step": 934
},
{
"epoch": 6.03,
"learning_rate": 1.3576091281874258e-05,
"loss": 0.0173,
"step": 935
},
{
"epoch": 6.04,
"learning_rate": 1.3563077423644146e-05,
"loss": 0.0168,
"step": 936
},
{
"epoch": 6.05,
"learning_rate": 1.3550056649782564e-05,
"loss": 0.0203,
"step": 937
},
{
"epoch": 6.05,
"learning_rate": 1.353702898556172e-05,
"loss": 0.0127,
"step": 938
},
{
"epoch": 6.06,
"learning_rate": 1.352399445626722e-05,
"loss": 0.0162,
"step": 939
},
{
"epoch": 6.06,
"learning_rate": 1.3510953087197972e-05,
"loss": 0.0126,
"step": 940
},
{
"epoch": 6.07,
"learning_rate": 1.3497904903666178e-05,
"loss": 0.0223,
"step": 941
},
{
"epoch": 6.08,
"learning_rate": 1.3484849930997244e-05,
"loss": 0.0131,
"step": 942
},
{
"epoch": 6.08,
"learning_rate": 1.3471788194529777e-05,
"loss": 0.0131,
"step": 943
},
{
"epoch": 6.09,
"learning_rate": 1.3458719719615498e-05,
"loss": 0.0202,
"step": 944
},
{
"epoch": 6.1,
"learning_rate": 1.344564453161921e-05,
"loss": 0.0156,
"step": 945
},
{
"epoch": 6.1,
"learning_rate": 1.3432562655918742e-05,
"loss": 0.0074,
"step": 946
},
{
"epoch": 6.11,
"learning_rate": 1.3419474117904914e-05,
"loss": 0.0189,
"step": 947
},
{
"epoch": 6.12,
"learning_rate": 1.3406378942981462e-05,
"loss": 0.0153,
"step": 948
},
{
"epoch": 6.12,
"learning_rate": 1.3393277156565014e-05,
"loss": 0.0168,
"step": 949
},
{
"epoch": 6.13,
"learning_rate": 1.3380168784085028e-05,
"loss": 0.0154,
"step": 950
},
{
"epoch": 6.14,
"learning_rate": 1.3367053850983747e-05,
"loss": 0.0126,
"step": 951
},
{
"epoch": 6.14,
"learning_rate": 1.3353932382716144e-05,
"loss": 0.0114,
"step": 952
},
{
"epoch": 6.15,
"learning_rate": 1.3340804404749871e-05,
"loss": 0.0061,
"step": 953
},
{
"epoch": 6.15,
"learning_rate": 1.332766994256523e-05,
"loss": 0.0106,
"step": 954
},
{
"epoch": 6.16,
"learning_rate": 1.3314529021655099e-05,
"loss": 0.0181,
"step": 955
},
{
"epoch": 6.17,
"learning_rate": 1.3301381667524886e-05,
"loss": 0.0094,
"step": 956
},
{
"epoch": 6.17,
"learning_rate": 1.3288227905692496e-05,
"loss": 0.0157,
"step": 957
},
{
"epoch": 6.18,
"learning_rate": 1.3275067761688268e-05,
"loss": 0.0112,
"step": 958
},
{
"epoch": 6.19,
"learning_rate": 1.3261901261054924e-05,
"loss": 0.0131,
"step": 959
},
{
"epoch": 6.19,
"learning_rate": 1.3248728429347525e-05,
"loss": 0.0153,
"step": 960
},
{
"epoch": 6.2,
"learning_rate": 1.3235549292133425e-05,
"loss": 0.0104,
"step": 961
},
{
"epoch": 6.21,
"learning_rate": 1.3222363874992215e-05,
"loss": 0.0114,
"step": 962
},
{
"epoch": 6.21,
"learning_rate": 1.3209172203515665e-05,
"loss": 0.0097,
"step": 963
},
{
"epoch": 6.22,
"learning_rate": 1.31959743033077e-05,
"loss": 0.0117,
"step": 964
},
{
"epoch": 6.23,
"learning_rate": 1.3182770199984321e-05,
"loss": 0.0222,
"step": 965
},
{
"epoch": 6.23,
"learning_rate": 1.3169559919173577e-05,
"loss": 0.0099,
"step": 966
},
{
"epoch": 6.24,
"learning_rate": 1.3156343486515504e-05,
"loss": 0.0144,
"step": 967
},
{
"epoch": 6.25,
"learning_rate": 1.3143120927662074e-05,
"loss": 0.0132,
"step": 968
},
{
"epoch": 6.25,
"learning_rate": 1.312989226827716e-05,
"loss": 0.0129,
"step": 969
},
{
"epoch": 6.26,
"learning_rate": 1.3116657534036467e-05,
"loss": 0.015,
"step": 970
},
{
"epoch": 6.26,
"learning_rate": 1.3103416750627494e-05,
"loss": 0.0133,
"step": 971
},
{
"epoch": 6.27,
"learning_rate": 1.3090169943749475e-05,
"loss": 0.0106,
"step": 972
},
{
"epoch": 6.28,
"learning_rate": 1.3076917139113345e-05,
"loss": 0.0113,
"step": 973
},
{
"epoch": 6.28,
"learning_rate": 1.3063658362441678e-05,
"loss": 0.0104,
"step": 974
},
{
"epoch": 6.29,
"learning_rate": 1.3050393639468629e-05,
"loss": 0.0105,
"step": 975
},
{
"epoch": 6.3,
"learning_rate": 1.3037122995939906e-05,
"loss": 0.0161,
"step": 976
},
{
"epoch": 6.3,
"learning_rate": 1.3023846457612704e-05,
"loss": 0.0244,
"step": 977
},
{
"epoch": 6.31,
"learning_rate": 1.3010564050255656e-05,
"loss": 0.0147,
"step": 978
},
{
"epoch": 6.32,
"learning_rate": 1.2997275799648796e-05,
"loss": 0.0174,
"step": 979
},
{
"epoch": 6.32,
"learning_rate": 1.2983981731583486e-05,
"loss": 0.012,
"step": 980
},
{
"epoch": 6.33,
"learning_rate": 1.2970681871862392e-05,
"loss": 0.0108,
"step": 981
},
{
"epoch": 6.34,
"learning_rate": 1.2957376246299409e-05,
"loss": 0.007,
"step": 982
},
{
"epoch": 6.34,
"learning_rate": 1.2944064880719634e-05,
"loss": 0.0123,
"step": 983
},
{
"epoch": 6.35,
"learning_rate": 1.29307478009593e-05,
"loss": 0.0201,
"step": 984
},
{
"epoch": 6.35,
"learning_rate": 1.291742503286573e-05,
"loss": 0.0138,
"step": 985
},
{
"epoch": 6.36,
"learning_rate": 1.2904096602297287e-05,
"loss": 0.0125,
"step": 986
},
{
"epoch": 6.37,
"learning_rate": 1.2890762535123322e-05,
"loss": 0.0101,
"step": 987
},
{
"epoch": 6.37,
"learning_rate": 1.287742285722414e-05,
"loss": 0.0155,
"step": 988
},
{
"epoch": 6.38,
"learning_rate": 1.2864077594490917e-05,
"loss": 0.0154,
"step": 989
},
{
"epoch": 6.39,
"learning_rate": 1.2850726772825685e-05,
"loss": 0.0148,
"step": 990
},
{
"epoch": 6.39,
"learning_rate": 1.2837370418141251e-05,
"loss": 0.0166,
"step": 991
},
{
"epoch": 6.4,
"learning_rate": 1.2824008556361169e-05,
"loss": 0.0159,
"step": 992
},
{
"epoch": 6.41,
"learning_rate": 1.2810641213419688e-05,
"loss": 0.0149,
"step": 993
},
{
"epoch": 6.41,
"learning_rate": 1.2797268415261681e-05,
"loss": 0.018,
"step": 994
},
{
"epoch": 6.42,
"learning_rate": 1.2783890187842618e-05,
"loss": 0.0091,
"step": 995
},
{
"epoch": 6.43,
"learning_rate": 1.2770506557128508e-05,
"loss": 0.0139,
"step": 996
},
{
"epoch": 6.43,
"learning_rate": 1.275711754909584e-05,
"loss": 0.0176,
"step": 997
},
{
"epoch": 6.44,
"learning_rate": 1.2743723189731547e-05,
"loss": 0.0109,
"step": 998
},
{
"epoch": 6.45,
"learning_rate": 1.2730323505032951e-05,
"loss": 0.0109,
"step": 999
},
{
"epoch": 6.45,
"learning_rate": 1.2716918521007697e-05,
"loss": 0.0213,
"step": 1000
},
{
"epoch": 6.46,
"learning_rate": 1.2703508263673725e-05,
"loss": 0.0146,
"step": 1001
},
{
"epoch": 6.46,
"learning_rate": 1.2690092759059211e-05,
"loss": 0.0172,
"step": 1002
},
{
"epoch": 6.47,
"learning_rate": 1.2676672033202514e-05,
"loss": 0.0147,
"step": 1003
},
{
"epoch": 6.48,
"learning_rate": 1.2663246112152126e-05,
"loss": 0.0079,
"step": 1004
},
{
"epoch": 6.48,
"learning_rate": 1.264981502196662e-05,
"loss": 0.0149,
"step": 1005
},
{
"epoch": 6.49,
"learning_rate": 1.2636378788714603e-05,
"loss": 0.0156,
"step": 1006
},
{
"epoch": 6.5,
"learning_rate": 1.262293743847467e-05,
"loss": 0.0171,
"step": 1007
},
{
"epoch": 6.5,
"learning_rate": 1.2609490997335342e-05,
"loss": 0.0119,
"step": 1008
},
{
"epoch": 6.51,
"learning_rate": 1.259603949139502e-05,
"loss": 0.0166,
"step": 1009
},
{
"epoch": 6.52,
"learning_rate": 1.2582582946761939e-05,
"loss": 0.016,
"step": 1010
},
{
"epoch": 6.52,
"learning_rate": 1.256912138955411e-05,
"loss": 0.0152,
"step": 1011
},
{
"epoch": 6.53,
"learning_rate": 1.2555654845899283e-05,
"loss": 0.0173,
"step": 1012
},
{
"epoch": 6.54,
"learning_rate": 1.2542183341934873e-05,
"loss": 0.0202,
"step": 1013
},
{
"epoch": 6.54,
"learning_rate": 1.2528706903807926e-05,
"loss": 0.0101,
"step": 1014
},
{
"epoch": 6.55,
"learning_rate": 1.251522555767507e-05,
"loss": 0.0158,
"step": 1015
},
{
"epoch": 6.55,
"learning_rate": 1.2501739329702453e-05,
"loss": 0.0112,
"step": 1016
},
{
"epoch": 6.56,
"learning_rate": 1.2488248246065707e-05,
"loss": 0.0128,
"step": 1017
},
{
"epoch": 6.57,
"learning_rate": 1.247475233294988e-05,
"loss": 0.0121,
"step": 1018
},
{
"epoch": 6.57,
"learning_rate": 1.2461251616549395e-05,
"loss": 0.022,
"step": 1019
},
{
"epoch": 6.58,
"learning_rate": 1.2447746123067996e-05,
"loss": 0.017,
"step": 1020
},
{
"epoch": 6.59,
"learning_rate": 1.2434235878718706e-05,
"loss": 0.0145,
"step": 1021
},
{
"epoch": 6.59,
"learning_rate": 1.2420720909723767e-05,
"loss": 0.0128,
"step": 1022
},
{
"epoch": 6.6,
"learning_rate": 1.2407201242314586e-05,
"loss": 0.0088,
"step": 1023
},
{
"epoch": 6.61,
"learning_rate": 1.239367690273169e-05,
"loss": 0.0112,
"step": 1024
},
{
"epoch": 6.61,
"learning_rate": 1.238014791722468e-05,
"loss": 0.0194,
"step": 1025
},
{
"epoch": 6.62,
"learning_rate": 1.236661431205217e-05,
"loss": 0.0074,
"step": 1026
},
{
"epoch": 6.63,
"learning_rate": 1.2353076113481742e-05,
"loss": 0.0148,
"step": 1027
},
{
"epoch": 6.63,
"learning_rate": 1.2339533347789891e-05,
"loss": 0.0087,
"step": 1028
},
{
"epoch": 6.64,
"learning_rate": 1.2325986041261979e-05,
"loss": 0.0248,
"step": 1029
},
{
"epoch": 6.65,
"learning_rate": 1.2312434220192178e-05,
"loss": 0.014,
"step": 1030
},
{
"epoch": 6.65,
"learning_rate": 1.2298877910883431e-05,
"loss": 0.0144,
"step": 1031
},
{
"epoch": 6.66,
"learning_rate": 1.2285317139647378e-05,
"loss": 0.0126,
"step": 1032
},
{
"epoch": 6.66,
"learning_rate": 1.227175193280434e-05,
"loss": 0.0093,
"step": 1033
},
{
"epoch": 6.67,
"learning_rate": 1.2258182316683218e-05,
"loss": 0.0136,
"step": 1034
},
{
"epoch": 6.68,
"learning_rate": 1.22446083176215e-05,
"loss": 0.0199,
"step": 1035
},
{
"epoch": 6.68,
"learning_rate": 1.223102996196516e-05,
"loss": 0.0129,
"step": 1036
},
{
"epoch": 6.69,
"learning_rate": 1.2217447276068644e-05,
"loss": 0.0158,
"step": 1037
},
{
"epoch": 6.7,
"learning_rate": 1.220386028629479e-05,
"loss": 0.0212,
"step": 1038
},
{
"epoch": 6.7,
"learning_rate": 1.2190269019014795e-05,
"loss": 0.016,
"step": 1039
},
{
"epoch": 6.71,
"learning_rate": 1.2176673500608157e-05,
"loss": 0.0108,
"step": 1040
},
{
"epoch": 6.72,
"learning_rate": 1.2163073757462628e-05,
"loss": 0.0148,
"step": 1041
},
{
"epoch": 6.72,
"learning_rate": 1.2149469815974158e-05,
"loss": 0.0129,
"step": 1042
},
{
"epoch": 6.73,
"learning_rate": 1.2135861702546841e-05,
"loss": 0.0156,
"step": 1043
},
{
"epoch": 6.74,
"learning_rate": 1.2122249443592874e-05,
"loss": 0.0206,
"step": 1044
},
{
"epoch": 6.74,
"learning_rate": 1.2108633065532498e-05,
"loss": 0.02,
"step": 1045
},
{
"epoch": 6.75,
"learning_rate": 1.2095012594793954e-05,
"loss": 0.023,
"step": 1046
},
{
"epoch": 6.75,
"learning_rate": 1.2081388057813417e-05,
"loss": 0.02,
"step": 1047
},
{
"epoch": 6.76,
"learning_rate": 1.2067759481034962e-05,
"loss": 0.0151,
"step": 1048
},
{
"epoch": 6.77,
"learning_rate": 1.2054126890910499e-05,
"loss": 0.0241,
"step": 1049
},
{
"epoch": 6.77,
"learning_rate": 1.2040490313899735e-05,
"loss": 0.0193,
"step": 1050
},
{
"epoch": 6.78,
"learning_rate": 1.2026849776470106e-05,
"loss": 0.0141,
"step": 1051
},
{
"epoch": 6.79,
"learning_rate": 1.2013205305096747e-05,
"loss": 0.0113,
"step": 1052
},
{
"epoch": 6.79,
"learning_rate": 1.1999556926262413e-05,
"loss": 0.0152,
"step": 1053
},
{
"epoch": 6.8,
"learning_rate": 1.1985904666457455e-05,
"loss": 0.0112,
"step": 1054
},
{
"epoch": 6.81,
"learning_rate": 1.1972248552179755e-05,
"loss": 0.0102,
"step": 1055
},
{
"epoch": 6.81,
"learning_rate": 1.1958588609934671e-05,
"loss": 0.0119,
"step": 1056
},
{
"epoch": 6.82,
"learning_rate": 1.1944924866234995e-05,
"loss": 0.0164,
"step": 1057
},
{
"epoch": 6.83,
"learning_rate": 1.1931257347600894e-05,
"loss": 0.0124,
"step": 1058
},
{
"epoch": 6.83,
"learning_rate": 1.1917586080559866e-05,
"loss": 0.0095,
"step": 1059
},
{
"epoch": 6.84,
"learning_rate": 1.1903911091646684e-05,
"loss": 0.0206,
"step": 1060
},
{
"epoch": 6.85,
"learning_rate": 1.1890232407403343e-05,
"loss": 0.017,
"step": 1061
},
{
"epoch": 6.85,
"learning_rate": 1.1876550054379006e-05,
"loss": 0.0119,
"step": 1062
},
{
"epoch": 6.86,
"learning_rate": 1.1862864059129963e-05,
"loss": 0.0179,
"step": 1063
},
{
"epoch": 6.86,
"learning_rate": 1.1849174448219575e-05,
"loss": 0.0165,
"step": 1064
},
{
"epoch": 6.87,
"learning_rate": 1.1835481248218215e-05,
"loss": 0.02,
"step": 1065
},
{
"epoch": 6.88,
"learning_rate": 1.1821784485703222e-05,
"loss": 0.0158,
"step": 1066
},
{
"epoch": 6.88,
"learning_rate": 1.1808084187258857e-05,
"loss": 0.0104,
"step": 1067
},
{
"epoch": 6.89,
"learning_rate": 1.1794380379476234e-05,
"loss": 0.0127,
"step": 1068
},
{
"epoch": 6.9,
"learning_rate": 1.1780673088953288e-05,
"loss": 0.0123,
"step": 1069
},
{
"epoch": 6.9,
"learning_rate": 1.1766962342294702e-05,
"loss": 0.017,
"step": 1070
},
{
"epoch": 6.91,
"learning_rate": 1.175324816611188e-05,
"loss": 0.0179,
"step": 1071
},
{
"epoch": 6.92,
"learning_rate": 1.173953058702287e-05,
"loss": 0.0148,
"step": 1072
},
{
"epoch": 6.92,
"learning_rate": 1.172580963165234e-05,
"loss": 0.0169,
"step": 1073
},
{
"epoch": 6.93,
"learning_rate": 1.1712085326631495e-05,
"loss": 0.0206,
"step": 1074
},
{
"epoch": 6.94,
"learning_rate": 1.1698357698598054e-05,
"loss": 0.013,
"step": 1075
},
{
"epoch": 6.94,
"learning_rate": 1.1684626774196175e-05,
"loss": 0.0098,
"step": 1076
},
{
"epoch": 6.95,
"learning_rate": 1.1670892580076421e-05,
"loss": 0.0096,
"step": 1077
},
{
"epoch": 6.95,
"learning_rate": 1.1657155142895698e-05,
"loss": 0.0271,
"step": 1078
},
{
"epoch": 6.96,
"learning_rate": 1.1643414489317213e-05,
"loss": 0.0114,
"step": 1079
},
{
"epoch": 6.97,
"learning_rate": 1.1629670646010407e-05,
"loss": 0.0181,
"step": 1080
},
{
"epoch": 6.97,
"learning_rate": 1.1615923639650914e-05,
"loss": 0.0152,
"step": 1081
},
{
"epoch": 6.98,
"learning_rate": 1.160217349692051e-05,
"loss": 0.0155,
"step": 1082
},
{
"epoch": 6.99,
"learning_rate": 1.1588420244507064e-05,
"loss": 0.0195,
"step": 1083
},
{
"epoch": 6.99,
"learning_rate": 1.1574663909104465e-05,
"loss": 0.0125,
"step": 1084
},
{
"epoch": 7.0,
"learning_rate": 1.1560904517412599e-05,
"loss": 0.0133,
"step": 1085
},
{
"epoch": 7.01,
"learning_rate": 1.1547142096137279e-05,
"loss": 0.0081,
"step": 1086
},
{
"epoch": 7.01,
"learning_rate": 1.1533376671990197e-05,
"loss": 0.0059,
"step": 1087
},
{
"epoch": 7.02,
"learning_rate": 1.151960827168888e-05,
"loss": 0.0131,
"step": 1088
},
{
"epoch": 7.03,
"learning_rate": 1.150583692195662e-05,
"loss": 0.0125,
"step": 1089
},
{
"epoch": 7.03,
"learning_rate": 1.149206264952245e-05,
"loss": 0.014,
"step": 1090
},
{
"epoch": 7.04,
"learning_rate": 1.1478285481121057e-05,
"loss": 0.0081,
"step": 1091
},
{
"epoch": 7.05,
"learning_rate": 1.1464505443492762e-05,
"loss": 0.0142,
"step": 1092
},
{
"epoch": 7.05,
"learning_rate": 1.145072256338345e-05,
"loss": 0.0101,
"step": 1093
},
{
"epoch": 7.06,
"learning_rate": 1.1436936867544525e-05,
"loss": 0.015,
"step": 1094
},
{
"epoch": 7.06,
"learning_rate": 1.1423148382732854e-05,
"loss": 0.0104,
"step": 1095
},
{
"epoch": 7.07,
"learning_rate": 1.1409357135710716e-05,
"loss": 0.0099,
"step": 1096
},
{
"epoch": 7.08,
"learning_rate": 1.1395563153245757e-05,
"loss": 0.0119,
"step": 1097
},
{
"epoch": 7.08,
"learning_rate": 1.1381766462110925e-05,
"loss": 0.0109,
"step": 1098
},
{
"epoch": 7.09,
"learning_rate": 1.1367967089084435e-05,
"loss": 0.0092,
"step": 1099
},
{
"epoch": 7.1,
"learning_rate": 1.135416506094969e-05,
"loss": 0.0103,
"step": 1100
},
{
"epoch": 7.1,
"learning_rate": 1.134036040449527e-05,
"loss": 0.0126,
"step": 1101
},
{
"epoch": 7.11,
"learning_rate": 1.1326553146514835e-05,
"loss": 0.018,
"step": 1102
},
{
"epoch": 7.12,
"learning_rate": 1.1312743313807111e-05,
"loss": 0.0074,
"step": 1103
},
{
"epoch": 7.12,
"learning_rate": 1.1298930933175805e-05,
"loss": 0.0103,
"step": 1104
},
{
"epoch": 7.13,
"learning_rate": 1.1285116031429585e-05,
"loss": 0.008,
"step": 1105
},
{
"epoch": 7.14,
"learning_rate": 1.1271298635381997e-05,
"loss": 0.017,
"step": 1106
},
{
"epoch": 7.14,
"learning_rate": 1.1257478771851446e-05,
"loss": 0.0119,
"step": 1107
},
{
"epoch": 7.15,
"learning_rate": 1.124365646766111e-05,
"loss": 0.0114,
"step": 1108
},
{
"epoch": 7.15,
"learning_rate": 1.1229831749638916e-05,
"loss": 0.006,
"step": 1109
},
{
"epoch": 7.16,
"learning_rate": 1.1216004644617462e-05,
"loss": 0.0109,
"step": 1110
},
{
"epoch": 7.17,
"learning_rate": 1.1202175179433993e-05,
"loss": 0.0064,
"step": 1111
},
{
"epoch": 7.17,
"learning_rate": 1.1188343380930334e-05,
"loss": 0.0113,
"step": 1112
},
{
"epoch": 7.18,
"learning_rate": 1.1174509275952827e-05,
"loss": 0.0083,
"step": 1113
},
{
"epoch": 7.19,
"learning_rate": 1.1160672891352303e-05,
"loss": 0.0082,
"step": 1114
},
{
"epoch": 7.19,
"learning_rate": 1.1146834253984008e-05,
"loss": 0.0165,
"step": 1115
},
{
"epoch": 7.2,
"learning_rate": 1.1132993390707565e-05,
"loss": 0.0117,
"step": 1116
},
{
"epoch": 7.21,
"learning_rate": 1.111915032838692e-05,
"loss": 0.009,
"step": 1117
},
{
"epoch": 7.21,
"learning_rate": 1.110530509389029e-05,
"loss": 0.0077,
"step": 1118
},
{
"epoch": 7.22,
"learning_rate": 1.1091457714090089e-05,
"loss": 0.0114,
"step": 1119
},
{
"epoch": 7.23,
"learning_rate": 1.1077608215862913e-05,
"loss": 0.0074,
"step": 1120
},
{
"epoch": 7.23,
"learning_rate": 1.1063756626089475e-05,
"loss": 0.0141,
"step": 1121
},
{
"epoch": 7.24,
"learning_rate": 1.1049902971654528e-05,
"loss": 0.0148,
"step": 1122
},
{
"epoch": 7.25,
"learning_rate": 1.103604727944684e-05,
"loss": 0.0109,
"step": 1123
},
{
"epoch": 7.25,
"learning_rate": 1.1022189576359143e-05,
"loss": 0.0067,
"step": 1124
},
{
"epoch": 7.26,
"learning_rate": 1.100832988928806e-05,
"loss": 0.0165,
"step": 1125
},
{
"epoch": 7.26,
"learning_rate": 1.0994468245134071e-05,
"loss": 0.0094,
"step": 1126
},
{
"epoch": 7.27,
"learning_rate": 1.0980604670801456e-05,
"loss": 0.0104,
"step": 1127
},
{
"epoch": 7.28,
"learning_rate": 1.0966739193198235e-05,
"loss": 0.0127,
"step": 1128
},
{
"epoch": 7.28,
"learning_rate": 1.0952871839236123e-05,
"loss": 0.0105,
"step": 1129
},
{
"epoch": 7.29,
"learning_rate": 1.0939002635830485e-05,
"loss": 0.0056,
"step": 1130
},
{
"epoch": 7.3,
"learning_rate": 1.0925131609900268e-05,
"loss": 0.0127,
"step": 1131
},
{
"epoch": 7.3,
"learning_rate": 1.0911258788367957e-05,
"loss": 0.0148,
"step": 1132
},
{
"epoch": 7.31,
"learning_rate": 1.0897384198159529e-05,
"loss": 0.0102,
"step": 1133
},
{
"epoch": 7.32,
"learning_rate": 1.0883507866204377e-05,
"loss": 0.0123,
"step": 1134
},
{
"epoch": 7.32,
"learning_rate": 1.0869629819435297e-05,
"loss": 0.0118,
"step": 1135
},
{
"epoch": 7.33,
"learning_rate": 1.08557500847884e-05,
"loss": 0.0134,
"step": 1136
},
{
"epoch": 7.34,
"learning_rate": 1.0841868689203072e-05,
"loss": 0.0199,
"step": 1137
},
{
"epoch": 7.34,
"learning_rate": 1.0827985659621926e-05,
"loss": 0.0073,
"step": 1138
},
{
"epoch": 7.35,
"learning_rate": 1.0814101022990749e-05,
"loss": 0.0176,
"step": 1139
},
{
"epoch": 7.35,
"learning_rate": 1.0800214806258445e-05,
"loss": 0.0142,
"step": 1140
},
{
"epoch": 7.36,
"learning_rate": 1.0786327036376983e-05,
"loss": 0.0206,
"step": 1141
},
{
"epoch": 7.37,
"learning_rate": 1.0772437740301348e-05,
"loss": 0.0078,
"step": 1142
},
{
"epoch": 7.37,
"learning_rate": 1.0758546944989483e-05,
"loss": 0.0118,
"step": 1143
},
{
"epoch": 7.38,
"learning_rate": 1.0744654677402255e-05,
"loss": 0.0067,
"step": 1144
},
{
"epoch": 7.39,
"learning_rate": 1.073076096450337e-05,
"loss": 0.0105,
"step": 1145
},
{
"epoch": 7.39,
"learning_rate": 1.0716865833259353e-05,
"loss": 0.0142,
"step": 1146
},
{
"epoch": 7.4,
"learning_rate": 1.0702969310639477e-05,
"loss": 0.0088,
"step": 1147
},
{
"epoch": 7.41,
"learning_rate": 1.0689071423615708e-05,
"loss": 0.0118,
"step": 1148
},
{
"epoch": 7.41,
"learning_rate": 1.0675172199162675e-05,
"loss": 0.0121,
"step": 1149
},
{
"epoch": 7.42,
"learning_rate": 1.0661271664257592e-05,
"loss": 0.0095,
"step": 1150
},
{
"epoch": 7.43,
"learning_rate": 1.0647369845880224e-05,
"loss": 0.0087,
"step": 1151
},
{
"epoch": 7.43,
"learning_rate": 1.0633466771012819e-05,
"loss": 0.0159,
"step": 1152
},
{
"epoch": 7.44,
"learning_rate": 1.0619562466640065e-05,
"loss": 0.0075,
"step": 1153
},
{
"epoch": 7.45,
"learning_rate": 1.0605656959749045e-05,
"loss": 0.0139,
"step": 1154
},
{
"epoch": 7.45,
"learning_rate": 1.0591750277329167e-05,
"loss": 0.0124,
"step": 1155
},
{
"epoch": 7.46,
"learning_rate": 1.0577842446372127e-05,
"loss": 0.0198,
"step": 1156
},
{
"epoch": 7.46,
"learning_rate": 1.056393349387184e-05,
"loss": 0.0125,
"step": 1157
},
{
"epoch": 7.47,
"learning_rate": 1.0550023446824407e-05,
"loss": 0.0136,
"step": 1158
},
{
"epoch": 7.48,
"learning_rate": 1.0536112332228057e-05,
"loss": 0.0126,
"step": 1159
},
{
"epoch": 7.48,
"learning_rate": 1.0522200177083077e-05,
"loss": 0.0116,
"step": 1160
},
{
"epoch": 7.49,
"learning_rate": 1.0508287008391791e-05,
"loss": 0.0102,
"step": 1161
},
{
"epoch": 7.5,
"learning_rate": 1.049437285315847e-05,
"loss": 0.0145,
"step": 1162
},
{
"epoch": 7.5,
"learning_rate": 1.048045773838932e-05,
"loss": 0.0167,
"step": 1163
},
{
"epoch": 7.51,
"learning_rate": 1.0466541691092394e-05,
"loss": 0.0121,
"step": 1164
},
{
"epoch": 7.52,
"learning_rate": 1.0452624738277564e-05,
"loss": 0.0125,
"step": 1165
},
{
"epoch": 7.52,
"learning_rate": 1.0438706906956458e-05,
"loss": 0.0119,
"step": 1166
},
{
"epoch": 7.53,
"learning_rate": 1.0424788224142403e-05,
"loss": 0.0145,
"step": 1167
},
{
"epoch": 7.54,
"learning_rate": 1.0410868716850387e-05,
"loss": 0.0153,
"step": 1168
},
{
"epoch": 7.54,
"learning_rate": 1.0396948412096992e-05,
"loss": 0.0136,
"step": 1169
},
{
"epoch": 7.55,
"learning_rate": 1.0383027336900356e-05,
"loss": 0.0101,
"step": 1170
},
{
"epoch": 7.55,
"learning_rate": 1.0369105518280098e-05,
"loss": 0.009,
"step": 1171
},
{
"epoch": 7.56,
"learning_rate": 1.0355182983257297e-05,
"loss": 0.0214,
"step": 1172
},
{
"epoch": 7.57,
"learning_rate": 1.034125975885441e-05,
"loss": 0.0116,
"step": 1173
},
{
"epoch": 7.57,
"learning_rate": 1.0327335872095239e-05,
"loss": 0.0092,
"step": 1174
},
{
"epoch": 7.58,
"learning_rate": 1.0313411350004863e-05,
"loss": 0.0113,
"step": 1175
},
{
"epoch": 7.59,
"learning_rate": 1.0299486219609607e-05,
"loss": 0.0165,
"step": 1176
},
{
"epoch": 7.59,
"learning_rate": 1.0285560507936962e-05,
"loss": 0.0117,
"step": 1177
},
{
"epoch": 7.6,
"learning_rate": 1.0271634242015565e-05,
"loss": 0.0157,
"step": 1178
},
{
"epoch": 7.61,
"learning_rate": 1.0257707448875108e-05,
"loss": 0.0204,
"step": 1179
},
{
"epoch": 7.61,
"learning_rate": 1.0243780155546324e-05,
"loss": 0.0135,
"step": 1180
},
{
"epoch": 7.62,
"learning_rate": 1.0229852389060905e-05,
"loss": 0.0184,
"step": 1181
},
{
"epoch": 7.63,
"learning_rate": 1.021592417645147e-05,
"loss": 0.0137,
"step": 1182
},
{
"epoch": 7.63,
"learning_rate": 1.0201995544751498e-05,
"loss": 0.0111,
"step": 1183
},
{
"epoch": 7.64,
"learning_rate": 1.0188066520995284e-05,
"loss": 0.0217,
"step": 1184
},
{
"epoch": 7.65,
"learning_rate": 1.0174137132217883e-05,
"loss": 0.015,
"step": 1185
},
{
"epoch": 7.65,
"learning_rate": 1.0160207405455053e-05,
"loss": 0.0184,
"step": 1186
},
{
"epoch": 7.66,
"learning_rate": 1.0146277367743225e-05,
"loss": 0.0189,
"step": 1187
},
{
"epoch": 7.66,
"learning_rate": 1.0132347046119416e-05,
"loss": 0.0181,
"step": 1188
},
{
"epoch": 7.67,
"learning_rate": 1.01184164676212e-05,
"loss": 0.017,
"step": 1189
},
{
"epoch": 7.68,
"learning_rate": 1.0104485659286646e-05,
"loss": 0.0141,
"step": 1190
},
{
"epoch": 7.68,
"learning_rate": 1.0090554648154282e-05,
"loss": 0.0116,
"step": 1191
},
{
"epoch": 7.69,
"learning_rate": 1.0076623461263017e-05,
"loss": 0.013,
"step": 1192
},
{
"epoch": 7.7,
"learning_rate": 1.0062692125652106e-05,
"loss": 0.0145,
"step": 1193
},
{
"epoch": 7.7,
"learning_rate": 1.0048760668361085e-05,
"loss": 0.0062,
"step": 1194
},
{
"epoch": 7.71,
"learning_rate": 1.0034829116429739e-05,
"loss": 0.0213,
"step": 1195
},
{
"epoch": 7.72,
"learning_rate": 1.0020897496898025e-05,
"loss": 0.0156,
"step": 1196
},
{
"epoch": 7.72,
"learning_rate": 1.0006965836806044e-05,
"loss": 0.0142,
"step": 1197
},
{
"epoch": 7.73,
"learning_rate": 9.99303416319396e-06,
"loss": 0.0088,
"step": 1198
},
{
"epoch": 7.74,
"learning_rate": 9.979102503101978e-06,
"loss": 0.0127,
"step": 1199
},
{
"epoch": 7.74,
"learning_rate": 9.965170883570266e-06,
"loss": 0.0086,
"step": 1200
},
{
"epoch": 7.75,
"learning_rate": 9.951239331638917e-06,
"loss": 0.0164,
"step": 1201
},
{
"epoch": 7.75,
"learning_rate": 9.9373078743479e-06,
"loss": 0.0089,
"step": 1202
},
{
"epoch": 7.76,
"learning_rate": 9.923376538736985e-06,
"loss": 0.0116,
"step": 1203
},
{
"epoch": 7.77,
"learning_rate": 9.909445351845718e-06,
"loss": 0.018,
"step": 1204
},
{
"epoch": 7.77,
"learning_rate": 9.895514340713355e-06,
"loss": 0.0152,
"step": 1205
},
{
"epoch": 7.78,
"learning_rate": 9.881583532378804e-06,
"loss": 0.0166,
"step": 1206
},
{
"epoch": 7.79,
"learning_rate": 9.867652953880586e-06,
"loss": 0.0217,
"step": 1207
},
{
"epoch": 7.79,
"learning_rate": 9.85372263225678e-06,
"loss": 0.0139,
"step": 1208
},
{
"epoch": 7.8,
"learning_rate": 9.839792594544948e-06,
"loss": 0.0148,
"step": 1209
},
{
"epoch": 7.81,
"learning_rate": 9.825862867782123e-06,
"loss": 0.0192,
"step": 1210
},
{
"epoch": 7.81,
"learning_rate": 9.81193347900472e-06,
"loss": 0.0124,
"step": 1211
},
{
"epoch": 7.82,
"learning_rate": 9.798004455248504e-06,
"loss": 0.0185,
"step": 1212
},
{
"epoch": 7.83,
"learning_rate": 9.784075823548535e-06,
"loss": 0.0084,
"step": 1213
},
{
"epoch": 7.83,
"learning_rate": 9.770147610939098e-06,
"loss": 0.0099,
"step": 1214
},
{
"epoch": 7.84,
"learning_rate": 9.756219844453676e-06,
"loss": 0.0168,
"step": 1215
},
{
"epoch": 7.85,
"learning_rate": 9.742292551124894e-06,
"loss": 0.0134,
"step": 1216
},
{
"epoch": 7.85,
"learning_rate": 9.728365757984439e-06,
"loss": 0.0209,
"step": 1217
},
{
"epoch": 7.86,
"learning_rate": 9.71443949206304e-06,
"loss": 0.0118,
"step": 1218
},
{
"epoch": 7.86,
"learning_rate": 9.700513780390396e-06,
"loss": 0.0145,
"step": 1219
},
{
"epoch": 7.87,
"learning_rate": 9.686588649995137e-06,
"loss": 0.0177,
"step": 1220
},
{
"epoch": 7.88,
"learning_rate": 9.672664127904766e-06,
"loss": 0.0088,
"step": 1221
},
{
"epoch": 7.88,
"learning_rate": 9.658740241145592e-06,
"loss": 0.0153,
"step": 1222
},
{
"epoch": 7.89,
"learning_rate": 9.644817016742705e-06,
"loss": 0.0233,
"step": 1223
},
{
"epoch": 7.9,
"learning_rate": 9.630894481719904e-06,
"loss": 0.0148,
"step": 1224
},
{
"epoch": 7.9,
"learning_rate": 9.616972663099648e-06,
"loss": 0.0192,
"step": 1225
},
{
"epoch": 7.91,
"learning_rate": 9.603051587903008e-06,
"loss": 0.0127,
"step": 1226
},
{
"epoch": 7.92,
"learning_rate": 9.589131283149617e-06,
"loss": 0.0156,
"step": 1227
},
{
"epoch": 7.92,
"learning_rate": 9.5752117758576e-06,
"loss": 0.0133,
"step": 1228
},
{
"epoch": 7.93,
"learning_rate": 9.561293093043547e-06,
"loss": 0.0086,
"step": 1229
},
{
"epoch": 7.94,
"learning_rate": 9.547375261722438e-06,
"loss": 0.0175,
"step": 1230
},
{
"epoch": 7.94,
"learning_rate": 9.533458308907608e-06,
"loss": 0.0155,
"step": 1231
},
{
"epoch": 7.95,
"learning_rate": 9.519542261610685e-06,
"loss": 0.0171,
"step": 1232
},
{
"epoch": 7.95,
"learning_rate": 9.505627146841533e-06,
"loss": 0.016,
"step": 1233
},
{
"epoch": 7.96,
"learning_rate": 9.49171299160821e-06,
"loss": 0.0144,
"step": 1234
},
{
"epoch": 7.97,
"learning_rate": 9.477799822916925e-06,
"loss": 0.0124,
"step": 1235
},
{
"epoch": 7.97,
"learning_rate": 9.463887667771946e-06,
"loss": 0.0097,
"step": 1236
},
{
"epoch": 7.98,
"learning_rate": 9.449976553175595e-06,
"loss": 0.0169,
"step": 1237
},
{
"epoch": 7.99,
"learning_rate": 9.436066506128164e-06,
"loss": 0.0131,
"step": 1238
},
{
"epoch": 7.99,
"learning_rate": 9.422157553627877e-06,
"loss": 0.0108,
"step": 1239
},
{
"epoch": 8.0,
"learning_rate": 9.408249722670836e-06,
"loss": 0.0197,
"step": 1240
},
{
"epoch": 8.01,
"learning_rate": 9.394343040250958e-06,
"loss": 0.0089,
"step": 1241
},
{
"epoch": 8.01,
"learning_rate": 9.380437533359937e-06,
"loss": 0.0133,
"step": 1242
},
{
"epoch": 8.02,
"learning_rate": 9.366533228987186e-06,
"loss": 0.0103,
"step": 1243
},
{
"epoch": 8.03,
"learning_rate": 9.352630154119777e-06,
"loss": 0.0084,
"step": 1244
},
{
"epoch": 8.03,
"learning_rate": 9.338728335742406e-06,
"loss": 0.0161,
"step": 1245
},
{
"epoch": 8.04,
"learning_rate": 9.324827800837329e-06,
"loss": 0.009,
"step": 1246
},
{
"epoch": 8.05,
"learning_rate": 9.310928576384293e-06,
"loss": 0.0138,
"step": 1247
},
{
"epoch": 8.05,
"learning_rate": 9.297030689360528e-06,
"loss": 0.0126,
"step": 1248
},
{
"epoch": 8.06,
"learning_rate": 9.283134166740649e-06,
"loss": 0.0074,
"step": 1249
},
{
"epoch": 8.06,
"learning_rate": 9.26923903549663e-06,
"loss": 0.013,
"step": 1250
},
{
"epoch": 8.07,
"learning_rate": 9.25534532259775e-06,
"loss": 0.0074,
"step": 1251
},
{
"epoch": 8.08,
"learning_rate": 9.241453055010518e-06,
"loss": 0.0091,
"step": 1252
},
{
"epoch": 8.08,
"learning_rate": 9.227562259698654e-06,
"loss": 0.0117,
"step": 1253
},
{
"epoch": 8.09,
"learning_rate": 9.213672963623022e-06,
"loss": 0.0101,
"step": 1254
},
{
"epoch": 8.1,
"learning_rate": 9.199785193741558e-06,
"loss": 0.0071,
"step": 1255
},
{
"epoch": 8.1,
"learning_rate": 9.185898977009253e-06,
"loss": 0.0098,
"step": 1256
},
{
"epoch": 8.11,
"learning_rate": 9.172014340378077e-06,
"loss": 0.0136,
"step": 1257
},
{
"epoch": 8.12,
"learning_rate": 9.15813131079693e-06,
"loss": 0.0118,
"step": 1258
},
{
"epoch": 8.12,
"learning_rate": 9.144249915211605e-06,
"loss": 0.0063,
"step": 1259
},
{
"epoch": 8.13,
"learning_rate": 9.130370180564706e-06,
"loss": 0.015,
"step": 1260
},
{
"epoch": 8.14,
"learning_rate": 9.116492133795625e-06,
"loss": 0.0108,
"step": 1261
},
{
"epoch": 8.14,
"learning_rate": 9.102615801840478e-06,
"loss": 0.0076,
"step": 1262
},
{
"epoch": 8.15,
"learning_rate": 9.088741211632045e-06,
"loss": 0.0095,
"step": 1263
},
{
"epoch": 8.15,
"learning_rate": 9.074868390099734e-06,
"loss": 0.0037,
"step": 1264
},
{
"epoch": 8.16,
"learning_rate": 9.060997364169519e-06,
"loss": 0.0134,
"step": 1265
},
{
"epoch": 8.17,
"learning_rate": 9.04712816076388e-06,
"loss": 0.0145,
"step": 1266
},
{
"epoch": 8.17,
"learning_rate": 9.033260806801772e-06,
"loss": 0.0094,
"step": 1267
},
{
"epoch": 8.18,
"learning_rate": 9.019395329198549e-06,
"loss": 0.0087,
"step": 1268
},
{
"epoch": 8.19,
"learning_rate": 9.005531754865929e-06,
"loss": 0.0096,
"step": 1269
},
{
"epoch": 8.19,
"learning_rate": 8.991670110711945e-06,
"loss": 0.0085,
"step": 1270
},
{
"epoch": 8.2,
"learning_rate": 8.977810423640858e-06,
"loss": 0.014,
"step": 1271
},
{
"epoch": 8.21,
"learning_rate": 8.963952720553159e-06,
"loss": 0.0159,
"step": 1272
},
{
"epoch": 8.21,
"learning_rate": 8.950097028345478e-06,
"loss": 0.0108,
"step": 1273
},
{
"epoch": 8.22,
"learning_rate": 8.936243373910528e-06,
"loss": 0.0152,
"step": 1274
},
{
"epoch": 8.23,
"learning_rate": 8.922391784137085e-06,
"loss": 0.0069,
"step": 1275
},
{
"epoch": 8.23,
"learning_rate": 8.908542285909916e-06,
"loss": 0.0083,
"step": 1276
},
{
"epoch": 8.24,
"learning_rate": 8.894694906109716e-06,
"loss": 0.0098,
"step": 1277
},
{
"epoch": 8.25,
"learning_rate": 8.880849671613083e-06,
"loss": 0.0108,
"step": 1278
},
{
"epoch": 8.25,
"learning_rate": 8.867006609292439e-06,
"loss": 0.0073,
"step": 1279
},
{
"epoch": 8.26,
"learning_rate": 8.853165746015997e-06,
"loss": 0.0139,
"step": 1280
},
{
"epoch": 8.26,
"learning_rate": 8.839327108647702e-06,
"loss": 0.0126,
"step": 1281
},
{
"epoch": 8.27,
"learning_rate": 8.825490724047175e-06,
"loss": 0.007,
"step": 1282
},
{
"epoch": 8.28,
"learning_rate": 8.811656619069668e-06,
"loss": 0.0149,
"step": 1283
},
{
"epoch": 8.28,
"learning_rate": 8.797824820566008e-06,
"loss": 0.0138,
"step": 1284
},
{
"epoch": 8.29,
"learning_rate": 8.783995355382542e-06,
"loss": 0.0111,
"step": 1285
},
{
"epoch": 8.3,
"learning_rate": 8.770168250361088e-06,
"loss": 0.0151,
"step": 1286
},
{
"epoch": 8.3,
"learning_rate": 8.756343532338893e-06,
"loss": 0.0145,
"step": 1287
},
{
"epoch": 8.31,
"learning_rate": 8.742521228148555e-06,
"loss": 0.0136,
"step": 1288
},
{
"epoch": 8.32,
"learning_rate": 8.728701364618007e-06,
"loss": 0.0084,
"step": 1289
},
{
"epoch": 8.32,
"learning_rate": 8.714883968570417e-06,
"loss": 0.0106,
"step": 1290
},
{
"epoch": 8.33,
"learning_rate": 8.701069066824195e-06,
"loss": 0.0147,
"step": 1291
},
{
"epoch": 8.34,
"learning_rate": 8.687256686192894e-06,
"loss": 0.0152,
"step": 1292
},
{
"epoch": 8.34,
"learning_rate": 8.673446853485166e-06,
"loss": 0.0151,
"step": 1293
},
{
"epoch": 8.35,
"learning_rate": 8.659639595504732e-06,
"loss": 0.0133,
"step": 1294
},
{
"epoch": 8.35,
"learning_rate": 8.645834939050311e-06,
"loss": 0.0094,
"step": 1295
},
{
"epoch": 8.36,
"learning_rate": 8.632032910915568e-06,
"loss": 0.0095,
"step": 1296
},
{
"epoch": 8.37,
"learning_rate": 8.61823353788908e-06,
"loss": 0.0086,
"step": 1297
},
{
"epoch": 8.37,
"learning_rate": 8.604436846754247e-06,
"loss": 0.0124,
"step": 1298
},
{
"epoch": 8.38,
"learning_rate": 8.590642864289287e-06,
"loss": 0.0131,
"step": 1299
},
{
"epoch": 8.39,
"learning_rate": 8.576851617267151e-06,
"loss": 0.0228,
"step": 1300
},
{
"epoch": 8.39,
"learning_rate": 8.563063132455478e-06,
"loss": 0.0157,
"step": 1301
},
{
"epoch": 8.4,
"learning_rate": 8.549277436616551e-06,
"loss": 0.0118,
"step": 1302
},
{
"epoch": 8.41,
"learning_rate": 8.535494556507241e-06,
"loss": 0.0105,
"step": 1303
},
{
"epoch": 8.41,
"learning_rate": 8.521714518878946e-06,
"loss": 0.0159,
"step": 1304
},
{
"epoch": 8.42,
"learning_rate": 8.507937350477553e-06,
"loss": 0.015,
"step": 1305
},
{
"epoch": 8.43,
"learning_rate": 8.49416307804338e-06,
"loss": 0.0147,
"step": 1306
},
{
"epoch": 8.43,
"learning_rate": 8.480391728311123e-06,
"loss": 0.0124,
"step": 1307
},
{
"epoch": 8.44,
"learning_rate": 8.466623328009806e-06,
"loss": 0.0103,
"step": 1308
},
{
"epoch": 8.45,
"learning_rate": 8.452857903862725e-06,
"loss": 0.0144,
"step": 1309
},
{
"epoch": 8.45,
"learning_rate": 8.439095482587403e-06,
"loss": 0.0133,
"step": 1310
},
{
"epoch": 8.46,
"learning_rate": 8.42533609089554e-06,
"loss": 0.0076,
"step": 1311
},
{
"epoch": 8.46,
"learning_rate": 8.41157975549294e-06,
"loss": 0.015,
"step": 1312
},
{
"epoch": 8.47,
"learning_rate": 8.397826503079489e-06,
"loss": 0.0104,
"step": 1313
},
{
"epoch": 8.48,
"learning_rate": 8.384076360349089e-06,
"loss": 0.0088,
"step": 1314
},
{
"epoch": 8.48,
"learning_rate": 8.370329353989596e-06,
"loss": 0.0089,
"step": 1315
},
{
"epoch": 8.49,
"learning_rate": 8.356585510682792e-06,
"loss": 0.0115,
"step": 1316
},
{
"epoch": 8.5,
"learning_rate": 8.342844857104306e-06,
"loss": 0.013,
"step": 1317
},
{
"epoch": 8.5,
"learning_rate": 8.329107419923584e-06,
"loss": 0.0109,
"step": 1318
},
{
"epoch": 8.51,
"learning_rate": 8.315373225803832e-06,
"loss": 0.0192,
"step": 1319
},
{
"epoch": 8.52,
"learning_rate": 8.30164230140195e-06,
"loss": 0.0089,
"step": 1320
},
{
"epoch": 8.52,
"learning_rate": 8.287914673368505e-06,
"loss": 0.0102,
"step": 1321
},
{
"epoch": 8.53,
"learning_rate": 8.274190368347665e-06,
"loss": 0.0081,
"step": 1322
},
{
"epoch": 8.54,
"learning_rate": 8.260469412977132e-06,
"loss": 0.0139,
"step": 1323
},
{
"epoch": 8.54,
"learning_rate": 8.246751833888122e-06,
"loss": 0.011,
"step": 1324
},
{
"epoch": 8.55,
"learning_rate": 8.233037657705303e-06,
"loss": 0.0119,
"step": 1325
},
{
"epoch": 8.55,
"learning_rate": 8.219326911046717e-06,
"loss": 0.0087,
"step": 1326
},
{
"epoch": 8.56,
"learning_rate": 8.205619620523769e-06,
"loss": 0.0115,
"step": 1327
},
{
"epoch": 8.57,
"learning_rate": 8.191915812741145e-06,
"loss": 0.0055,
"step": 1328
},
{
"epoch": 8.57,
"learning_rate": 8.178215514296778e-06,
"loss": 0.0117,
"step": 1329
},
{
"epoch": 8.58,
"learning_rate": 8.164518751781789e-06,
"loss": 0.0157,
"step": 1330
},
{
"epoch": 8.59,
"learning_rate": 8.150825551780428e-06,
"loss": 0.0109,
"step": 1331
},
{
"epoch": 8.59,
"learning_rate": 8.137135940870038e-06,
"loss": 0.013,
"step": 1332
},
{
"epoch": 8.6,
"learning_rate": 8.123449945620997e-06,
"loss": 0.0062,
"step": 1333
},
{
"epoch": 8.61,
"learning_rate": 8.10976759259666e-06,
"loss": 0.0101,
"step": 1334
},
{
"epoch": 8.61,
"learning_rate": 8.096088908353316e-06,
"loss": 0.0131,
"step": 1335
},
{
"epoch": 8.62,
"learning_rate": 8.082413919440137e-06,
"loss": 0.0117,
"step": 1336
},
{
"epoch": 8.63,
"learning_rate": 8.068742652399109e-06,
"loss": 0.0086,
"step": 1337
},
{
"epoch": 8.63,
"learning_rate": 8.055075133765008e-06,
"loss": 0.0073,
"step": 1338
},
{
"epoch": 8.64,
"learning_rate": 8.041411390065332e-06,
"loss": 0.0165,
"step": 1339
},
{
"epoch": 8.65,
"learning_rate": 8.027751447820247e-06,
"loss": 0.0123,
"step": 1340
},
{
"epoch": 8.65,
"learning_rate": 8.014095333542548e-06,
"loss": 0.0115,
"step": 1341
},
{
"epoch": 8.66,
"learning_rate": 8.00044307373759e-06,
"loss": 0.0108,
"step": 1342
},
{
"epoch": 8.66,
"learning_rate": 7.986794694903255e-06,
"loss": 0.0144,
"step": 1343
},
{
"epoch": 8.67,
"learning_rate": 7.973150223529896e-06,
"loss": 0.0113,
"step": 1344
},
{
"epoch": 8.68,
"learning_rate": 7.959509686100267e-06,
"loss": 0.0133,
"step": 1345
},
{
"epoch": 8.68,
"learning_rate": 7.945873109089503e-06,
"loss": 0.0094,
"step": 1346
},
{
"epoch": 8.69,
"learning_rate": 7.932240518965041e-06,
"loss": 0.0108,
"step": 1347
},
{
"epoch": 8.7,
"learning_rate": 7.918611942186585e-06,
"loss": 0.0118,
"step": 1348
},
{
"epoch": 8.7,
"learning_rate": 7.90498740520605e-06,
"loss": 0.0155,
"step": 1349
},
{
"epoch": 8.71,
"learning_rate": 7.891366934467504e-06,
"loss": 0.0102,
"step": 1350
},
{
"epoch": 8.72,
"learning_rate": 7.877750556407128e-06,
"loss": 0.0089,
"step": 1351
},
{
"epoch": 8.72,
"learning_rate": 7.864138297453164e-06,
"loss": 0.0171,
"step": 1352
},
{
"epoch": 8.73,
"learning_rate": 7.850530184025845e-06,
"loss": 0.0135,
"step": 1353
},
{
"epoch": 8.74,
"learning_rate": 7.836926242537372e-06,
"loss": 0.0122,
"step": 1354
},
{
"epoch": 8.74,
"learning_rate": 7.823326499391845e-06,
"loss": 0.0091,
"step": 1355
},
{
"epoch": 8.75,
"learning_rate": 7.809730980985207e-06,
"loss": 0.0091,
"step": 1356
},
{
"epoch": 8.75,
"learning_rate": 7.796139713705214e-06,
"loss": 0.0092,
"step": 1357
},
{
"epoch": 8.76,
"learning_rate": 7.782552723931358e-06,
"loss": 0.0104,
"step": 1358
},
{
"epoch": 8.77,
"learning_rate": 7.76897003803484e-06,
"loss": 0.0123,
"step": 1359
},
{
"epoch": 8.77,
"learning_rate": 7.755391682378506e-06,
"loss": 0.0096,
"step": 1360
},
{
"epoch": 8.78,
"learning_rate": 7.741817683316785e-06,
"loss": 0.0109,
"step": 1361
},
{
"epoch": 8.79,
"learning_rate": 7.728248067195662e-06,
"loss": 0.0112,
"step": 1362
},
{
"epoch": 8.79,
"learning_rate": 7.714682860352623e-06,
"loss": 0.0116,
"step": 1363
},
{
"epoch": 8.8,
"learning_rate": 7.701122089116572e-06,
"loss": 0.0094,
"step": 1364
},
{
"epoch": 8.81,
"learning_rate": 7.687565779807823e-06,
"loss": 0.011,
"step": 1365
},
{
"epoch": 8.81,
"learning_rate": 7.674013958738025e-06,
"loss": 0.0144,
"step": 1366
},
{
"epoch": 8.82,
"learning_rate": 7.660466652210112e-06,
"loss": 0.0099,
"step": 1367
},
{
"epoch": 8.83,
"learning_rate": 7.646923886518263e-06,
"loss": 0.012,
"step": 1368
},
{
"epoch": 8.83,
"learning_rate": 7.633385687947835e-06,
"loss": 0.0155,
"step": 1369
},
{
"epoch": 8.84,
"learning_rate": 7.619852082775323e-06,
"loss": 0.0248,
"step": 1370
},
{
"epoch": 8.85,
"learning_rate": 7.6063230972683135e-06,
"loss": 0.011,
"step": 1371
},
{
"epoch": 8.85,
"learning_rate": 7.592798757685418e-06,
"loss": 0.0138,
"step": 1372
},
{
"epoch": 8.86,
"learning_rate": 7.579279090276233e-06,
"loss": 0.0116,
"step": 1373
},
{
"epoch": 8.86,
"learning_rate": 7.565764121281296e-06,
"loss": 0.0108,
"step": 1374
},
{
"epoch": 8.87,
"learning_rate": 7.552253876932006e-06,
"loss": 0.0166,
"step": 1375
},
{
"epoch": 8.88,
"learning_rate": 7.538748383450611e-06,
"loss": 0.012,
"step": 1376
},
{
"epoch": 8.88,
"learning_rate": 7.525247667050123e-06,
"loss": 0.0134,
"step": 1377
},
{
"epoch": 8.89,
"learning_rate": 7.511751753934293e-06,
"loss": 0.0112,
"step": 1378
},
{
"epoch": 8.9,
"learning_rate": 7.4982606702975505e-06,
"loss": 0.0125,
"step": 1379
},
{
"epoch": 8.9,
"learning_rate": 7.484774442324932e-06,
"loss": 0.0102,
"step": 1380
},
{
"epoch": 8.91,
"learning_rate": 7.471293096192076e-06,
"loss": 0.0167,
"step": 1381
},
{
"epoch": 8.92,
"learning_rate": 7.4578166580651335e-06,
"loss": 0.0122,
"step": 1382
},
{
"epoch": 8.92,
"learning_rate": 7.444345154100721e-06,
"loss": 0.0078,
"step": 1383
},
{
"epoch": 8.93,
"learning_rate": 7.43087861044589e-06,
"loss": 0.0138,
"step": 1384
},
{
"epoch": 8.94,
"learning_rate": 7.417417053238065e-06,
"loss": 0.0105,
"step": 1385
},
{
"epoch": 8.94,
"learning_rate": 7.403960508604984e-06,
"loss": 0.019,
"step": 1386
},
{
"epoch": 8.95,
"learning_rate": 7.390509002664665e-06,
"loss": 0.0125,
"step": 1387
},
{
"epoch": 8.95,
"learning_rate": 7.377062561525334e-06,
"loss": 0.0105,
"step": 1388
},
{
"epoch": 8.96,
"learning_rate": 7.3636212112854e-06,
"loss": 0.0159,
"step": 1389
},
{
"epoch": 8.97,
"learning_rate": 7.350184978033386e-06,
"loss": 0.0158,
"step": 1390
},
{
"epoch": 8.97,
"learning_rate": 7.3367538878478765e-06,
"loss": 0.009,
"step": 1391
},
{
"epoch": 8.98,
"learning_rate": 7.323327966797486e-06,
"loss": 0.0159,
"step": 1392
},
{
"epoch": 8.98,
"eval_val_accuracy": 90.97288676236045,
"eval_val_average": 63.966581266228744,
"eval_val_loss": 2.00240159034729,
"eval_val_recall": 36.96027577009705,
"eval_val_runtime": 4338.8616,
"eval_val_samples_per_second": 1.447,
"eval_val_steps_per_second": 0.006,
"step": 1392
},
{
"epoch": 8.99,
"learning_rate": 7.309907240940791e-06,
"loss": 0.0098,
"step": 1393
},
{
"epoch": 8.99,
"learning_rate": 7.296491736326276e-06,
"loss": 0.01,
"step": 1394
},
{
"epoch": 9.0,
"learning_rate": 7.283081478992307e-06,
"loss": 0.0074,
"step": 1395
},
{
"epoch": 9.01,
"learning_rate": 7.269676494967053e-06,
"loss": 0.0093,
"step": 1396
},
{
"epoch": 9.01,
"learning_rate": 7.2562768102684525e-06,
"loss": 0.007,
"step": 1397
},
{
"epoch": 9.02,
"learning_rate": 7.242882450904164e-06,
"loss": 0.0109,
"step": 1398
},
{
"epoch": 9.03,
"learning_rate": 7.229493442871495e-06,
"loss": 0.0112,
"step": 1399
},
{
"epoch": 9.03,
"learning_rate": 7.216109812157383e-06,
"loss": 0.0084,
"step": 1400
},
{
"epoch": 9.04,
"learning_rate": 7.202731584738323e-06,
"loss": 0.0075,
"step": 1401
},
{
"epoch": 9.05,
"learning_rate": 7.189358786580315e-06,
"loss": 0.0058,
"step": 1402
},
{
"epoch": 9.05,
"learning_rate": 7.17599144363883e-06,
"loss": 0.0145,
"step": 1403
},
{
"epoch": 9.06,
"learning_rate": 7.162629581858752e-06,
"loss": 0.0108,
"step": 1404
},
{
"epoch": 9.06,
"learning_rate": 7.149273227174318e-06,
"loss": 0.0074,
"step": 1405
},
{
"epoch": 9.07,
"learning_rate": 7.135922405509086e-06,
"loss": 0.0082,
"step": 1406
},
{
"epoch": 9.08,
"learning_rate": 7.122577142775864e-06,
"loss": 0.0069,
"step": 1407
},
{
"epoch": 9.08,
"learning_rate": 7.109237464876679e-06,
"loss": 0.0097,
"step": 1408
},
{
"epoch": 9.09,
"learning_rate": 7.095903397702719e-06,
"loss": 0.0108,
"step": 1409
},
{
"epoch": 9.1,
"learning_rate": 7.082574967134274e-06,
"loss": 0.0033,
"step": 1410
},
{
"epoch": 9.1,
"learning_rate": 7.069252199040702e-06,
"loss": 0.0098,
"step": 1411
},
{
"epoch": 9.11,
"learning_rate": 7.055935119280369e-06,
"loss": 0.01,
"step": 1412
},
{
"epoch": 9.12,
"learning_rate": 7.042623753700594e-06,
"loss": 0.0103,
"step": 1413
},
{
"epoch": 9.12,
"learning_rate": 7.029318128137612e-06,
"loss": 0.0103,
"step": 1414
},
{
"epoch": 9.13,
"learning_rate": 7.016018268416518e-06,
"loss": 0.0144,
"step": 1415
},
{
"epoch": 9.14,
"learning_rate": 7.002724200351207e-06,
"loss": 0.0121,
"step": 1416
},
{
"epoch": 9.14,
"learning_rate": 6.9894359497443465e-06,
"loss": 0.0109,
"step": 1417
},
{
"epoch": 9.15,
"learning_rate": 6.9761535423872995e-06,
"loss": 0.0091,
"step": 1418
},
{
"epoch": 9.15,
"learning_rate": 6.962877004060096e-06,
"loss": 0.0122,
"step": 1419
},
{
"epoch": 9.16,
"learning_rate": 6.949606360531376e-06,
"loss": 0.0136,
"step": 1420
},
{
"epoch": 9.17,
"learning_rate": 6.936341637558326e-06,
"loss": 0.0092,
"step": 1421
},
{
"epoch": 9.17,
"learning_rate": 6.923082860886655e-06,
"loss": 0.0078,
"step": 1422
},
{
"epoch": 9.18,
"learning_rate": 6.909830056250527e-06,
"loss": 0.008,
"step": 1423
},
{
"epoch": 9.19,
"learning_rate": 6.896583249372508e-06,
"loss": 0.0132,
"step": 1424
},
{
"epoch": 9.19,
"learning_rate": 6.883342465963537e-06,
"loss": 0.0118,
"step": 1425
},
{
"epoch": 9.2,
"learning_rate": 6.8701077317228425e-06,
"loss": 0.0087,
"step": 1426
},
{
"epoch": 9.21,
"learning_rate": 6.856879072337928e-06,
"loss": 0.0084,
"step": 1427
},
{
"epoch": 9.21,
"learning_rate": 6.8436565134845015e-06,
"loss": 0.0157,
"step": 1428
},
{
"epoch": 9.22,
"learning_rate": 6.830440080826426e-06,
"loss": 0.009,
"step": 1429
},
{
"epoch": 9.23,
"learning_rate": 6.817229800015681e-06,
"loss": 0.0119,
"step": 1430
},
{
"epoch": 9.23,
"learning_rate": 6.804025696692306e-06,
"loss": 0.0121,
"step": 1431
},
{
"epoch": 9.24,
"learning_rate": 6.790827796484338e-06,
"loss": 0.0095,
"step": 1432
},
{
"epoch": 9.25,
"learning_rate": 6.777636125007787e-06,
"loss": 0.0099,
"step": 1433
},
{
"epoch": 9.25,
"learning_rate": 6.764450707866577e-06,
"loss": 0.0137,
"step": 1434
},
{
"epoch": 9.26,
"learning_rate": 6.751271570652477e-06,
"loss": 0.0084,
"step": 1435
},
{
"epoch": 9.26,
"learning_rate": 6.738098738945081e-06,
"loss": 0.0068,
"step": 1436
},
{
"epoch": 9.27,
"learning_rate": 6.724932238311734e-06,
"loss": 0.0104,
"step": 1437
},
{
"epoch": 9.28,
"learning_rate": 6.711772094307505e-06,
"loss": 0.0136,
"step": 1438
},
{
"epoch": 9.28,
"learning_rate": 6.6986183324751174e-06,
"loss": 0.0088,
"step": 1439
},
{
"epoch": 9.29,
"learning_rate": 6.685470978344906e-06,
"loss": 0.0106,
"step": 1440
},
{
"epoch": 9.3,
"learning_rate": 6.672330057434771e-06,
"loss": 0.0099,
"step": 1441
},
{
"epoch": 9.3,
"learning_rate": 6.6591955952501304e-06,
"loss": 0.0079,
"step": 1442
},
{
"epoch": 9.31,
"learning_rate": 6.64606761728386e-06,
"loss": 0.0108,
"step": 1443
},
{
"epoch": 9.32,
"learning_rate": 6.632946149016257e-06,
"loss": 0.0076,
"step": 1444
},
{
"epoch": 9.32,
"learning_rate": 6.619831215914974e-06,
"loss": 0.0134,
"step": 1445
},
{
"epoch": 9.33,
"learning_rate": 6.6067228434349884e-06,
"loss": 0.0105,
"step": 1446
},
{
"epoch": 9.34,
"learning_rate": 6.593621057018542e-06,
"loss": 0.0092,
"step": 1447
},
{
"epoch": 9.34,
"learning_rate": 6.58052588209509e-06,
"loss": 0.0107,
"step": 1448
},
{
"epoch": 9.35,
"learning_rate": 6.5674373440812575e-06,
"loss": 0.0118,
"step": 1449
},
{
"epoch": 9.35,
"learning_rate": 6.554355468380796e-06,
"loss": 0.0121,
"step": 1450
},
{
"epoch": 9.36,
"learning_rate": 6.5412802803845055e-06,
"loss": 0.0106,
"step": 1451
},
{
"epoch": 9.37,
"learning_rate": 6.528211805470223e-06,
"loss": 0.0155,
"step": 1452
},
{
"epoch": 9.37,
"learning_rate": 6.515150069002758e-06,
"loss": 0.0125,
"step": 1453
},
{
"epoch": 9.38,
"learning_rate": 6.502095096333827e-06,
"loss": 0.0097,
"step": 1454
},
{
"epoch": 9.39,
"learning_rate": 6.489046912802031e-06,
"loss": 0.0112,
"step": 1455
},
{
"epoch": 9.39,
"learning_rate": 6.476005543732783e-06,
"loss": 0.0077,
"step": 1456
},
{
"epoch": 9.4,
"learning_rate": 6.462971014438279e-06,
"loss": 0.0149,
"step": 1457
},
{
"epoch": 9.41,
"learning_rate": 6.4499433502174424e-06,
"loss": 0.0059,
"step": 1458
},
{
"epoch": 9.41,
"learning_rate": 6.436922576355855e-06,
"loss": 0.0119,
"step": 1459
},
{
"epoch": 9.42,
"learning_rate": 6.423908718125742e-06,
"loss": 0.0121,
"step": 1460
},
{
"epoch": 9.43,
"learning_rate": 6.410901800785901e-06,
"loss": 0.0117,
"step": 1461
},
{
"epoch": 9.43,
"learning_rate": 6.397901849581654e-06,
"loss": 0.0151,
"step": 1462
},
{
"epoch": 9.44,
"learning_rate": 6.384908889744803e-06,
"loss": 0.0097,
"step": 1463
},
{
"epoch": 9.45,
"learning_rate": 6.3719229464935915e-06,
"loss": 0.0122,
"step": 1464
},
{
"epoch": 9.45,
"learning_rate": 6.358944045032627e-06,
"loss": 0.0108,
"step": 1465
},
{
"epoch": 9.46,
"learning_rate": 6.34597221055286e-06,
"loss": 0.0089,
"step": 1466
},
{
"epoch": 9.46,
"learning_rate": 6.333007468231521e-06,
"loss": 0.0136,
"step": 1467
},
{
"epoch": 9.47,
"learning_rate": 6.320049843232079e-06,
"loss": 0.008,
"step": 1468
},
{
"epoch": 9.48,
"learning_rate": 6.30709936070419e-06,
"loss": 0.0142,
"step": 1469
},
{
"epoch": 9.48,
"learning_rate": 6.2941560457836346e-06,
"loss": 0.0108,
"step": 1470
},
{
"epoch": 9.49,
"learning_rate": 6.281219923592294e-06,
"loss": 0.0101,
"step": 1471
},
{
"epoch": 9.5,
"learning_rate": 6.268291019238095e-06,
"loss": 0.0115,
"step": 1472
},
{
"epoch": 9.5,
"learning_rate": 6.255369357814933e-06,
"loss": 0.0092,
"step": 1473
},
{
"epoch": 9.51,
"learning_rate": 6.242454964402666e-06,
"loss": 0.0115,
"step": 1474
},
{
"epoch": 9.52,
"learning_rate": 6.2295478640670336e-06,
"loss": 0.011,
"step": 1475
},
{
"epoch": 9.52,
"learning_rate": 6.216648081859625e-06,
"loss": 0.0092,
"step": 1476
},
{
"epoch": 9.53,
"learning_rate": 6.20375564281783e-06,
"loss": 0.0105,
"step": 1477
},
{
"epoch": 9.54,
"learning_rate": 6.1908705719647735e-06,
"loss": 0.0132,
"step": 1478
},
{
"epoch": 9.54,
"learning_rate": 6.177992894309288e-06,
"loss": 0.0061,
"step": 1479
},
{
"epoch": 9.55,
"learning_rate": 6.16512263484586e-06,
"loss": 0.0092,
"step": 1480
},
{
"epoch": 9.55,
"learning_rate": 6.152259818554567e-06,
"loss": 0.0113,
"step": 1481
},
{
"epoch": 9.56,
"learning_rate": 6.139404470401049e-06,
"loss": 0.0099,
"step": 1482
},
{
"epoch": 9.57,
"learning_rate": 6.12655661533645e-06,
"loss": 0.0106,
"step": 1483
},
{
"epoch": 9.57,
"learning_rate": 6.113716278297363e-06,
"loss": 0.0093,
"step": 1484
},
{
"epoch": 9.58,
"learning_rate": 6.1008834842057995e-06,
"loss": 0.0111,
"step": 1485
},
{
"epoch": 9.59,
"learning_rate": 6.088058257969121e-06,
"loss": 0.0126,
"step": 1486
},
{
"epoch": 9.59,
"learning_rate": 6.0752406244800075e-06,
"loss": 0.0121,
"step": 1487
},
{
"epoch": 9.6,
"learning_rate": 6.062430608616405e-06,
"loss": 0.0144,
"step": 1488
},
{
"epoch": 9.61,
"learning_rate": 6.049628235241459e-06,
"loss": 0.0111,
"step": 1489
},
{
"epoch": 9.61,
"learning_rate": 6.0368335292035e-06,
"loss": 0.0106,
"step": 1490
},
{
"epoch": 9.62,
"learning_rate": 6.0240465153359685e-06,
"loss": 0.014,
"step": 1491
},
{
"epoch": 9.63,
"learning_rate": 6.0112672184573705e-06,
"loss": 0.0061,
"step": 1492
},
{
"epoch": 9.63,
"learning_rate": 5.998495663371242e-06,
"loss": 0.0125,
"step": 1493
},
{
"epoch": 9.64,
"learning_rate": 5.985731874866087e-06,
"loss": 0.0082,
"step": 1494
},
{
"epoch": 9.65,
"learning_rate": 5.972975877715339e-06,
"loss": 0.0062,
"step": 1495
},
{
"epoch": 9.65,
"learning_rate": 5.9602276966773144e-06,
"loss": 0.0161,
"step": 1496
},
{
"epoch": 9.66,
"learning_rate": 5.947487356495142e-06,
"loss": 0.0085,
"step": 1497
},
{
"epoch": 9.66,
"learning_rate": 5.934754881896748e-06,
"loss": 0.0052,
"step": 1498
},
{
"epoch": 9.67,
"learning_rate": 5.922030297594789e-06,
"loss": 0.0162,
"step": 1499
},
{
"epoch": 9.68,
"learning_rate": 5.9093136282866014e-06,
"loss": 0.0086,
"step": 1500
},
{
"epoch": 9.68,
"learning_rate": 5.8966048986541654e-06,
"loss": 0.0115,
"step": 1501
},
{
"epoch": 9.69,
"learning_rate": 5.883904133364053e-06,
"loss": 0.0085,
"step": 1502
},
{
"epoch": 9.7,
"learning_rate": 5.871211357067365e-06,
"loss": 0.0134,
"step": 1503
},
{
"epoch": 9.7,
"learning_rate": 5.85852659439971e-06,
"loss": 0.0061,
"step": 1504
},
{
"epoch": 9.71,
"learning_rate": 5.845849869981137e-06,
"loss": 0.0095,
"step": 1505
},
{
"epoch": 9.72,
"learning_rate": 5.833181208416092e-06,
"loss": 0.0139,
"step": 1506
},
{
"epoch": 9.72,
"learning_rate": 5.820520634293383e-06,
"loss": 0.0116,
"step": 1507
},
{
"epoch": 9.73,
"learning_rate": 5.807868172186102e-06,
"loss": 0.0118,
"step": 1508
},
{
"epoch": 9.74,
"learning_rate": 5.795223846651608e-06,
"loss": 0.0117,
"step": 1509
},
{
"epoch": 9.74,
"learning_rate": 5.782587682231473e-06,
"loss": 0.0169,
"step": 1510
},
{
"epoch": 9.75,
"learning_rate": 5.769959703451413e-06,
"loss": 0.0099,
"step": 1511
},
{
"epoch": 9.75,
"learning_rate": 5.757339934821269e-06,
"loss": 0.013,
"step": 1512
},
{
"epoch": 9.76,
"learning_rate": 5.744728400834945e-06,
"loss": 0.0083,
"step": 1513
},
{
"epoch": 9.77,
"learning_rate": 5.732125125970355e-06,
"loss": 0.0078,
"step": 1514
},
{
"epoch": 9.77,
"learning_rate": 5.71953013468939e-06,
"loss": 0.0115,
"step": 1515
},
{
"epoch": 9.78,
"learning_rate": 5.706943451437863e-06,
"loss": 0.0135,
"step": 1516
},
{
"epoch": 9.79,
"learning_rate": 5.694365100645458e-06,
"loss": 0.0128,
"step": 1517
},
{
"epoch": 9.79,
"learning_rate": 5.681795106725695e-06,
"loss": 0.014,
"step": 1518
},
{
"epoch": 9.8,
"learning_rate": 5.6692334940758565e-06,
"loss": 0.0105,
"step": 1519
},
{
"epoch": 9.81,
"learning_rate": 5.656680287076976e-06,
"loss": 0.0111,
"step": 1520
},
{
"epoch": 9.81,
"learning_rate": 5.644135510093767e-06,
"loss": 0.0081,
"step": 1521
},
{
"epoch": 9.82,
"learning_rate": 5.63159918747457e-06,
"loss": 0.0147,
"step": 1522
},
{
"epoch": 9.83,
"learning_rate": 5.619071343551335e-06,
"loss": 0.0111,
"step": 1523
},
{
"epoch": 9.83,
"learning_rate": 5.606552002639538e-06,
"loss": 0.0111,
"step": 1524
},
{
"epoch": 9.84,
"learning_rate": 5.5940411890381575e-06,
"loss": 0.0101,
"step": 1525
},
{
"epoch": 9.85,
"learning_rate": 5.581538927029637e-06,
"loss": 0.0091,
"step": 1526
},
{
"epoch": 9.85,
"learning_rate": 5.5690452408797955e-06,
"loss": 0.0171,
"step": 1527
},
{
"epoch": 9.86,
"learning_rate": 5.556560154837823e-06,
"loss": 0.0084,
"step": 1528
},
{
"epoch": 9.86,
"learning_rate": 5.544083693136218e-06,
"loss": 0.0108,
"step": 1529
},
{
"epoch": 9.87,
"learning_rate": 5.53161587999073e-06,
"loss": 0.0153,
"step": 1530
},
{
"epoch": 9.88,
"learning_rate": 5.519156739600329e-06,
"loss": 0.0091,
"step": 1531
},
{
"epoch": 9.88,
"learning_rate": 5.506706296147156e-06,
"loss": 0.0075,
"step": 1532
},
{
"epoch": 9.89,
"learning_rate": 5.49426457379646e-06,
"loss": 0.0133,
"step": 1533
},
{
"epoch": 9.9,
"learning_rate": 5.481831596696574e-06,
"loss": 0.018,
"step": 1534
},
{
"epoch": 9.9,
"learning_rate": 5.469407388978855e-06,
"loss": 0.0121,
"step": 1535
},
{
"epoch": 9.91,
"learning_rate": 5.456991974757635e-06,
"loss": 0.0097,
"step": 1536
},
{
"epoch": 9.92,
"learning_rate": 5.444585378130186e-06,
"loss": 0.0082,
"step": 1537
},
{
"epoch": 9.92,
"learning_rate": 5.432187623176655e-06,
"loss": 0.0114,
"step": 1538
},
{
"epoch": 9.93,
"learning_rate": 5.419798733960035e-06,
"loss": 0.0083,
"step": 1539
},
{
"epoch": 9.94,
"learning_rate": 5.407418734526119e-06,
"loss": 0.0121,
"step": 1540
},
{
"epoch": 9.94,
"learning_rate": 5.395047648903424e-06,
"loss": 0.0109,
"step": 1541
},
{
"epoch": 9.95,
"learning_rate": 5.382685501103189e-06,
"loss": 0.0113,
"step": 1542
},
{
"epoch": 9.95,
"learning_rate": 5.370332315119296e-06,
"loss": 0.0142,
"step": 1543
},
{
"epoch": 9.96,
"learning_rate": 5.357988114928221e-06,
"loss": 0.0094,
"step": 1544
},
{
"epoch": 9.97,
"learning_rate": 5.345652924489028e-06,
"loss": 0.0137,
"step": 1545
},
{
"epoch": 9.97,
"learning_rate": 5.333326767743263e-06,
"loss": 0.0147,
"step": 1546
},
{
"epoch": 9.98,
"learning_rate": 5.321009668614959e-06,
"loss": 0.015,
"step": 1547
},
{
"epoch": 9.99,
"learning_rate": 5.308701651010564e-06,
"loss": 0.0152,
"step": 1548
},
{
"epoch": 9.99,
"learning_rate": 5.296402738818891e-06,
"loss": 0.0064,
"step": 1549
},
{
"epoch": 10.0,
"learning_rate": 5.284112955911089e-06,
"loss": 0.0071,
"step": 1550
},
{
"epoch": 10.01,
"learning_rate": 5.271832326140591e-06,
"loss": 0.0078,
"step": 1551
},
{
"epoch": 10.01,
"learning_rate": 5.25956087334305e-06,
"loss": 0.0087,
"step": 1552
},
{
"epoch": 10.02,
"learning_rate": 5.2472986213363234e-06,
"loss": 0.0045,
"step": 1553
},
{
"epoch": 10.03,
"learning_rate": 5.2350455939204e-06,
"loss": 0.0087,
"step": 1554
},
{
"epoch": 10.03,
"learning_rate": 5.22280181487737e-06,
"loss": 0.0117,
"step": 1555
},
{
"epoch": 10.04,
"learning_rate": 5.210567307971375e-06,
"loss": 0.0087,
"step": 1556
},
{
"epoch": 10.05,
"learning_rate": 5.198342096948549e-06,
"loss": 0.0115,
"step": 1557
},
{
"epoch": 10.05,
"learning_rate": 5.186126205536995e-06,
"loss": 0.0085,
"step": 1558
},
{
"epoch": 10.06,
"learning_rate": 5.173919657446726e-06,
"loss": 0.0115,
"step": 1559
},
{
"epoch": 10.06,
"learning_rate": 5.161722476369613e-06,
"loss": 0.0144,
"step": 1560
},
{
"epoch": 10.07,
"learning_rate": 5.149534685979351e-06,
"loss": 0.0061,
"step": 1561
},
{
"epoch": 10.08,
"learning_rate": 5.1373563099314135e-06,
"loss": 0.0105,
"step": 1562
},
{
"epoch": 10.08,
"learning_rate": 5.125187371862995e-06,
"loss": 0.0081,
"step": 1563
},
{
"epoch": 10.09,
"learning_rate": 5.113027895392977e-06,
"loss": 0.0097,
"step": 1564
},
{
"epoch": 10.1,
"learning_rate": 5.100877904121864e-06,
"loss": 0.006,
"step": 1565
},
{
"epoch": 10.1,
"learning_rate": 5.088737421631767e-06,
"loss": 0.013,
"step": 1566
},
{
"epoch": 10.11,
"learning_rate": 5.076606471486336e-06,
"loss": 0.0097,
"step": 1567
},
{
"epoch": 10.12,
"learning_rate": 5.064485077230712e-06,
"loss": 0.0139,
"step": 1568
},
{
"epoch": 10.12,
"learning_rate": 5.052373262391494e-06,
"loss": 0.0102,
"step": 1569
},
{
"epoch": 10.13,
"learning_rate": 5.040271050476697e-06,
"loss": 0.0104,
"step": 1570
},
{
"epoch": 10.14,
"learning_rate": 5.0281784649756805e-06,
"loss": 0.0123,
"step": 1571
},
{
"epoch": 10.14,
"learning_rate": 5.016095529359132e-06,
"loss": 0.0089,
"step": 1572
},
{
"epoch": 10.15,
"learning_rate": 5.004022267079005e-06,
"loss": 0.0099,
"step": 1573
},
{
"epoch": 10.15,
"learning_rate": 4.99195870156848e-06,
"loss": 0.0117,
"step": 1574
},
{
"epoch": 10.16,
"learning_rate": 4.9799048562419204e-06,
"loss": 0.009,
"step": 1575
},
{
"epoch": 10.17,
"learning_rate": 4.96786075449481e-06,
"loss": 0.0123,
"step": 1576
},
{
"epoch": 10.17,
"learning_rate": 4.955826419703736e-06,
"loss": 0.0093,
"step": 1577
},
{
"epoch": 10.18,
"learning_rate": 4.943801875226323e-06,
"loss": 0.0135,
"step": 1578
},
{
"epoch": 10.19,
"learning_rate": 4.931787144401189e-06,
"loss": 0.0025,
"step": 1579
},
{
"epoch": 10.19,
"learning_rate": 4.9197822505479105e-06,
"loss": 0.0065,
"step": 1580
},
{
"epoch": 10.2,
"learning_rate": 4.907787216966971e-06,
"loss": 0.0113,
"step": 1581
},
{
"epoch": 10.21,
"learning_rate": 4.895802066939716e-06,
"loss": 0.0093,
"step": 1582
},
{
"epoch": 10.21,
"learning_rate": 4.883826823728307e-06,
"loss": 0.0141,
"step": 1583
},
{
"epoch": 10.22,
"learning_rate": 4.871861510575672e-06,
"loss": 0.0084,
"step": 1584
},
{
"epoch": 10.23,
"learning_rate": 4.859906150705471e-06,
"loss": 0.0087,
"step": 1585
},
{
"epoch": 10.23,
"learning_rate": 4.847960767322052e-06,
"loss": 0.0082,
"step": 1586
},
{
"epoch": 10.24,
"learning_rate": 4.836025383610382e-06,
"loss": 0.0085,
"step": 1587
},
{
"epoch": 10.25,
"learning_rate": 4.824100022736036e-06,
"loss": 0.0085,
"step": 1588
},
{
"epoch": 10.25,
"learning_rate": 4.812184707845131e-06,
"loss": 0.0072,
"step": 1589
},
{
"epoch": 10.26,
"learning_rate": 4.800279462064278e-06,
"loss": 0.0095,
"step": 1590
},
{
"epoch": 10.26,
"learning_rate": 4.788384308500553e-06,
"loss": 0.0115,
"step": 1591
},
{
"epoch": 10.27,
"learning_rate": 4.776499270241443e-06,
"loss": 0.0114,
"step": 1592
},
{
"epoch": 10.28,
"learning_rate": 4.7646243703548e-06,
"loss": 0.0098,
"step": 1593
},
{
"epoch": 10.28,
"learning_rate": 4.752759631888801e-06,
"loss": 0.0126,
"step": 1594
},
{
"epoch": 10.29,
"learning_rate": 4.7409050778718945e-06,
"loss": 0.0101,
"step": 1595
},
{
"epoch": 10.3,
"learning_rate": 4.729060731312765e-06,
"loss": 0.0107,
"step": 1596
},
{
"epoch": 10.3,
"learning_rate": 4.717226615200294e-06,
"loss": 0.0113,
"step": 1597
},
{
"epoch": 10.31,
"learning_rate": 4.705402752503487e-06,
"loss": 0.0099,
"step": 1598
},
{
"epoch": 10.32,
"learning_rate": 4.693589166171466e-06,
"loss": 0.0165,
"step": 1599
},
{
"epoch": 10.32,
"learning_rate": 4.6817858791334025e-06,
"loss": 0.0094,
"step": 1600
},
{
"epoch": 10.33,
"learning_rate": 4.669992914298474e-06,
"loss": 0.0121,
"step": 1601
},
{
"epoch": 10.34,
"learning_rate": 4.6582102945558315e-06,
"loss": 0.0105,
"step": 1602
},
{
"epoch": 10.34,
"learning_rate": 4.646438042774535e-06,
"loss": 0.0108,
"step": 1603
},
{
"epoch": 10.35,
"learning_rate": 4.63467618180353e-06,
"loss": 0.0091,
"step": 1604
},
{
"epoch": 10.35,
"learning_rate": 4.622924734471599e-06,
"loss": 0.0123,
"step": 1605
},
{
"epoch": 10.36,
"learning_rate": 4.611183723587295e-06,
"loss": 0.008,
"step": 1606
},
{
"epoch": 10.37,
"learning_rate": 4.599453171938932e-06,
"loss": 0.0069,
"step": 1607
},
{
"epoch": 10.37,
"learning_rate": 4.5877331022945194e-06,
"loss": 0.0065,
"step": 1608
},
{
"epoch": 10.38,
"learning_rate": 4.576023537401712e-06,
"loss": 0.0055,
"step": 1609
},
{
"epoch": 10.39,
"learning_rate": 4.56432449998779e-06,
"loss": 0.0082,
"step": 1610
},
{
"epoch": 10.39,
"learning_rate": 4.55263601275959e-06,
"loss": 0.0111,
"step": 1611
},
{
"epoch": 10.4,
"learning_rate": 4.540958098403479e-06,
"loss": 0.0066,
"step": 1612
},
{
"epoch": 10.41,
"learning_rate": 4.5292907795853e-06,
"loss": 0.0061,
"step": 1613
},
{
"epoch": 10.41,
"learning_rate": 4.517634078950327e-06,
"loss": 0.0105,
"step": 1614
},
{
"epoch": 10.42,
"learning_rate": 4.5059880191232285e-06,
"loss": 0.0093,
"step": 1615
},
{
"epoch": 10.43,
"learning_rate": 4.494352622708027e-06,
"loss": 0.01,
"step": 1616
},
{
"epoch": 10.43,
"learning_rate": 4.482727912288032e-06,
"loss": 0.009,
"step": 1617
},
{
"epoch": 10.44,
"learning_rate": 4.4711139104258265e-06,
"loss": 0.0115,
"step": 1618
},
{
"epoch": 10.45,
"learning_rate": 4.4595106396632035e-06,
"loss": 0.0065,
"step": 1619
},
{
"epoch": 10.45,
"learning_rate": 4.447918122521129e-06,
"loss": 0.0103,
"step": 1620
},
{
"epoch": 10.46,
"learning_rate": 4.436336381499701e-06,
"loss": 0.0084,
"step": 1621
},
{
"epoch": 10.46,
"learning_rate": 4.4247654390780906e-06,
"loss": 0.0116,
"step": 1622
},
{
"epoch": 10.47,
"learning_rate": 4.41320531771452e-06,
"loss": 0.0137,
"step": 1623
},
{
"epoch": 10.48,
"learning_rate": 4.401656039846208e-06,
"loss": 0.012,
"step": 1624
},
{
"epoch": 10.48,
"learning_rate": 4.39011762788932e-06,
"loss": 0.0122,
"step": 1625
},
{
"epoch": 10.49,
"learning_rate": 4.378590104238939e-06,
"loss": 0.0083,
"step": 1626
},
{
"epoch": 10.5,
"learning_rate": 4.367073491269015e-06,
"loss": 0.0089,
"step": 1627
},
{
"epoch": 10.5,
"learning_rate": 4.355567811332311e-06,
"loss": 0.0102,
"step": 1628
},
{
"epoch": 10.51,
"learning_rate": 4.344073086760381e-06,
"loss": 0.0081,
"step": 1629
},
{
"epoch": 10.52,
"learning_rate": 4.332589339863512e-06,
"loss": 0.0146,
"step": 1630
},
{
"epoch": 10.52,
"learning_rate": 4.321116592930684e-06,
"loss": 0.0118,
"step": 1631
},
{
"epoch": 10.53,
"learning_rate": 4.3096548682295304e-06,
"loss": 0.0116,
"step": 1632
},
{
"epoch": 10.54,
"learning_rate": 4.298204188006281e-06,
"loss": 0.0092,
"step": 1633
},
{
"epoch": 10.54,
"learning_rate": 4.286764574485739e-06,
"loss": 0.0106,
"step": 1634
},
{
"epoch": 10.55,
"learning_rate": 4.27533604987123e-06,
"loss": 0.0156,
"step": 1635
},
{
"epoch": 10.55,
"learning_rate": 4.263918636344545e-06,
"loss": 0.0108,
"step": 1636
},
{
"epoch": 10.56,
"learning_rate": 4.25251235606592e-06,
"loss": 0.0115,
"step": 1637
},
{
"epoch": 10.57,
"learning_rate": 4.2411172311739775e-06,
"loss": 0.0139,
"step": 1638
},
{
"epoch": 10.57,
"learning_rate": 4.229733283785692e-06,
"loss": 0.0085,
"step": 1639
},
{
"epoch": 10.58,
"learning_rate": 4.218360535996338e-06,
"loss": 0.0074,
"step": 1640
},
{
"epoch": 10.59,
"learning_rate": 4.206999009879463e-06,
"loss": 0.0108,
"step": 1641
},
{
"epoch": 10.59,
"learning_rate": 4.195648727486818e-06,
"loss": 0.009,
"step": 1642
},
{
"epoch": 10.6,
"learning_rate": 4.1843097108483465e-06,
"loss": 0.0069,
"step": 1643
},
{
"epoch": 10.61,
"learning_rate": 4.172981981972112e-06,
"loss": 0.0129,
"step": 1644
},
{
"epoch": 10.61,
"learning_rate": 4.161665562844281e-06,
"loss": 0.0106,
"step": 1645
},
{
"epoch": 10.62,
"learning_rate": 4.150360475429067e-06,
"loss": 0.0078,
"step": 1646
},
{
"epoch": 10.63,
"learning_rate": 4.13906674166868e-06,
"loss": 0.0146,
"step": 1647
},
{
"epoch": 10.63,
"learning_rate": 4.127784383483306e-06,
"loss": 0.0154,
"step": 1648
},
{
"epoch": 10.64,
"learning_rate": 4.1165134227710466e-06,
"loss": 0.0071,
"step": 1649
},
{
"epoch": 10.65,
"learning_rate": 4.105253881407879e-06,
"loss": 0.0082,
"step": 1650
},
{
"epoch": 10.65,
"learning_rate": 4.094005781247626e-06,
"loss": 0.0063,
"step": 1651
},
{
"epoch": 10.66,
"learning_rate": 4.082769144121887e-06,
"loss": 0.008,
"step": 1652
},
{
"epoch": 10.66,
"learning_rate": 4.0715439918400304e-06,
"loss": 0.0086,
"step": 1653
},
{
"epoch": 10.67,
"learning_rate": 4.060330346189125e-06,
"loss": 0.0097,
"step": 1654
},
{
"epoch": 10.68,
"learning_rate": 4.049128228933903e-06,
"loss": 0.0158,
"step": 1655
},
{
"epoch": 10.68,
"learning_rate": 4.037937661816729e-06,
"loss": 0.0116,
"step": 1656
},
{
"epoch": 10.69,
"learning_rate": 4.026758666557543e-06,
"loss": 0.0138,
"step": 1657
},
{
"epoch": 10.7,
"learning_rate": 4.015591264853832e-06,
"loss": 0.0088,
"step": 1658
},
{
"epoch": 10.7,
"learning_rate": 4.004435478380573e-06,
"loss": 0.0102,
"step": 1659
},
{
"epoch": 10.71,
"learning_rate": 3.993291328790208e-06,
"loss": 0.0075,
"step": 1660
},
{
"epoch": 10.72,
"learning_rate": 3.982158837712581e-06,
"loss": 0.0057,
"step": 1661
},
{
"epoch": 10.72,
"learning_rate": 3.971038026754918e-06,
"loss": 0.0176,
"step": 1662
},
{
"epoch": 10.73,
"learning_rate": 3.9599289175017685e-06,
"loss": 0.0119,
"step": 1663
},
{
"epoch": 10.74,
"learning_rate": 3.9488315315149735e-06,
"loss": 0.0074,
"step": 1664
},
{
"epoch": 10.74,
"learning_rate": 3.937745890333623e-06,
"loss": 0.0124,
"step": 1665
},
{
"epoch": 10.75,
"learning_rate": 3.926672015474001e-06,
"loss": 0.0092,
"step": 1666
},
{
"epoch": 10.75,
"learning_rate": 3.915609928429563e-06,
"loss": 0.0114,
"step": 1667
},
{
"epoch": 10.76,
"learning_rate": 3.904559650670885e-06,
"loss": 0.0108,
"step": 1668
},
{
"epoch": 10.77,
"learning_rate": 3.893521203645618e-06,
"loss": 0.0123,
"step": 1669
},
{
"epoch": 10.77,
"learning_rate": 3.882494608778454e-06,
"loss": 0.0115,
"step": 1670
},
{
"epoch": 10.78,
"learning_rate": 3.871479887471084e-06,
"loss": 0.0106,
"step": 1671
},
{
"epoch": 10.79,
"learning_rate": 3.860477061102138e-06,
"loss": 0.0085,
"step": 1672
},
{
"epoch": 10.79,
"learning_rate": 3.849486151027182e-06,
"loss": 0.006,
"step": 1673
},
{
"epoch": 10.8,
"learning_rate": 3.838507178578631e-06,
"loss": 0.0058,
"step": 1674
},
{
"epoch": 10.81,
"learning_rate": 3.827540165065746e-06,
"loss": 0.0108,
"step": 1675
},
{
"epoch": 10.81,
"learning_rate": 3.8165851317745705e-06,
"loss": 0.0103,
"step": 1676
},
{
"epoch": 10.82,
"learning_rate": 3.805642099967897e-06,
"loss": 0.0083,
"step": 1677
},
{
"epoch": 10.83,
"learning_rate": 3.7947110908852223e-06,
"loss": 0.0065,
"step": 1678
},
{
"epoch": 10.83,
"learning_rate": 3.783792125742712e-06,
"loss": 0.016,
"step": 1679
},
{
"epoch": 10.84,
"learning_rate": 3.7728852257331472e-06,
"loss": 0.0077,
"step": 1680
},
{
"epoch": 10.85,
"learning_rate": 3.7619904120259045e-06,
"loss": 0.0101,
"step": 1681
},
{
"epoch": 10.85,
"learning_rate": 3.7511077057668844e-06,
"loss": 0.0149,
"step": 1682
},
{
"epoch": 10.86,
"learning_rate": 3.7402371280785045e-06,
"loss": 0.0086,
"step": 1683
},
{
"epoch": 10.86,
"learning_rate": 3.729378700059637e-06,
"loss": 0.0108,
"step": 1684
},
{
"epoch": 10.87,
"learning_rate": 3.7185324427855653e-06,
"loss": 0.0098,
"step": 1685
},
{
"epoch": 10.88,
"learning_rate": 3.707698377307958e-06,
"loss": 0.0082,
"step": 1686
},
{
"epoch": 10.88,
"learning_rate": 3.69687652465482e-06,
"loss": 0.006,
"step": 1687
},
{
"epoch": 10.89,
"learning_rate": 3.68606690583045e-06,
"loss": 0.0148,
"step": 1688
},
{
"epoch": 10.9,
"learning_rate": 3.6752695418154015e-06,
"loss": 0.0129,
"step": 1689
},
{
"epoch": 10.9,
"learning_rate": 3.664484453566449e-06,
"loss": 0.0088,
"step": 1690
},
{
"epoch": 10.91,
"learning_rate": 3.653711662016527e-06,
"loss": 0.0123,
"step": 1691
},
{
"epoch": 10.92,
"learning_rate": 3.6429511880747194e-06,
"loss": 0.0106,
"step": 1692
},
{
"epoch": 10.92,
"learning_rate": 3.6322030526261854e-06,
"loss": 0.0085,
"step": 1693
},
{
"epoch": 10.93,
"learning_rate": 3.6214672765321513e-06,
"loss": 0.0096,
"step": 1694
},
{
"epoch": 10.94,
"learning_rate": 3.610743880629849e-06,
"loss": 0.0098,
"step": 1695
},
{
"epoch": 10.94,
"learning_rate": 3.6000328857324794e-06,
"loss": 0.0138,
"step": 1696
},
{
"epoch": 10.95,
"learning_rate": 3.589334312629179e-06,
"loss": 0.0119,
"step": 1697
},
{
"epoch": 10.95,
"learning_rate": 3.578648182084975e-06,
"loss": 0.0141,
"step": 1698
},
{
"epoch": 10.96,
"learning_rate": 3.5679745148407328e-06,
"loss": 0.007,
"step": 1699
},
{
"epoch": 10.97,
"learning_rate": 3.5573133316131447e-06,
"loss": 0.01,
"step": 1700
},
{
"epoch": 10.97,
"learning_rate": 3.546664653094658e-06,
"loss": 0.0132,
"step": 1701
},
{
"epoch": 10.98,
"learning_rate": 3.5360284999534587e-06,
"loss": 0.0115,
"step": 1702
},
{
"epoch": 10.99,
"learning_rate": 3.5254048928334217e-06,
"loss": 0.0103,
"step": 1703
},
{
"epoch": 10.99,
"learning_rate": 3.5147938523540627e-06,
"loss": 0.0083,
"step": 1704
},
{
"epoch": 11.0,
"learning_rate": 3.504195399110516e-06,
"loss": 0.0125,
"step": 1705
},
{
"epoch": 11.01,
"learning_rate": 3.4936095536734803e-06,
"loss": 0.0104,
"step": 1706
},
{
"epoch": 11.01,
"learning_rate": 3.483036336589186e-06,
"loss": 0.0101,
"step": 1707
},
{
"epoch": 11.02,
"learning_rate": 3.472475768379352e-06,
"loss": 0.0103,
"step": 1708
},
{
"epoch": 11.03,
"learning_rate": 3.4619278695411495e-06,
"loss": 0.0069,
"step": 1709
},
{
"epoch": 11.03,
"learning_rate": 3.4513926605471504e-06,
"loss": 0.0097,
"step": 1710
},
{
"epoch": 11.04,
"learning_rate": 3.440870161845309e-06,
"loss": 0.0088,
"step": 1711
},
{
"epoch": 11.05,
"learning_rate": 3.430360393858898e-06,
"loss": 0.0057,
"step": 1712
},
{
"epoch": 11.05,
"learning_rate": 3.41986337698649e-06,
"loss": 0.0106,
"step": 1713
},
{
"epoch": 11.06,
"learning_rate": 3.4093791316019056e-06,
"loss": 0.0148,
"step": 1714
},
{
"epoch": 11.06,
"learning_rate": 3.3989076780541774e-06,
"loss": 0.0094,
"step": 1715
},
{
"epoch": 11.07,
"learning_rate": 3.3884490366675094e-06,
"loss": 0.0089,
"step": 1716
},
{
"epoch": 11.08,
"learning_rate": 3.378003227741241e-06,
"loss": 0.0084,
"step": 1717
},
{
"epoch": 11.08,
"learning_rate": 3.3675702715497972e-06,
"loss": 0.0107,
"step": 1718
},
{
"epoch": 11.09,
"learning_rate": 3.357150188342666e-06,
"loss": 0.0085,
"step": 1719
},
{
"epoch": 11.1,
"learning_rate": 3.3467429983443477e-06,
"loss": 0.0121,
"step": 1720
},
{
"epoch": 11.1,
"learning_rate": 3.3363487217543122e-06,
"loss": 0.0056,
"step": 1721
},
{
"epoch": 11.11,
"learning_rate": 3.3259673787469747e-06,
"loss": 0.0099,
"step": 1722
},
{
"epoch": 11.12,
"learning_rate": 3.3155989894716365e-06,
"loss": 0.0091,
"step": 1723
},
{
"epoch": 11.12,
"learning_rate": 3.3052435740524678e-06,
"loss": 0.0103,
"step": 1724
},
{
"epoch": 11.13,
"learning_rate": 3.29490115258845e-06,
"loss": 0.0097,
"step": 1725
},
{
"epoch": 11.14,
"learning_rate": 3.28457174515335e-06,
"loss": 0.0115,
"step": 1726
},
{
"epoch": 11.14,
"learning_rate": 3.2742553717956694e-06,
"loss": 0.0091,
"step": 1727
},
{
"epoch": 11.15,
"learning_rate": 3.2639520525386216e-06,
"loss": 0.0098,
"step": 1728
},
{
"epoch": 11.15,
"learning_rate": 3.2536618073800687e-06,
"loss": 0.0109,
"step": 1729
},
{
"epoch": 11.16,
"learning_rate": 3.243384656292511e-06,
"loss": 0.01,
"step": 1730
},
{
"epoch": 11.17,
"learning_rate": 3.233120619223021e-06,
"loss": 0.0093,
"step": 1731
},
{
"epoch": 11.17,
"learning_rate": 3.222869716093231e-06,
"loss": 0.0095,
"step": 1732
},
{
"epoch": 11.18,
"learning_rate": 3.2126319667992743e-06,
"loss": 0.0112,
"step": 1733
},
{
"epoch": 11.19,
"learning_rate": 3.2024073912117547e-06,
"loss": 0.0094,
"step": 1734
},
{
"epoch": 11.19,
"learning_rate": 3.1921960091757076e-06,
"loss": 0.0073,
"step": 1735
},
{
"epoch": 11.2,
"learning_rate": 3.1819978405105643e-06,
"loss": 0.0056,
"step": 1736
},
{
"epoch": 11.21,
"learning_rate": 3.1718129050100987e-06,
"loss": 0.0093,
"step": 1737
},
{
"epoch": 11.21,
"learning_rate": 3.161641222442413e-06,
"loss": 0.0052,
"step": 1738
},
{
"epoch": 11.22,
"learning_rate": 3.151482812549884e-06,
"loss": 0.012,
"step": 1739
},
{
"epoch": 11.23,
"learning_rate": 3.141337695049117e-06,
"loss": 0.0093,
"step": 1740
},
{
"epoch": 11.23,
"learning_rate": 3.131205889630934e-06,
"loss": 0.0099,
"step": 1741
},
{
"epoch": 11.24,
"learning_rate": 3.1210874159603044e-06,
"loss": 0.0106,
"step": 1742
},
{
"epoch": 11.25,
"learning_rate": 3.1109822936763266e-06,
"loss": 0.0111,
"step": 1743
},
{
"epoch": 11.25,
"learning_rate": 3.1008905423921996e-06,
"loss": 0.0085,
"step": 1744
},
{
"epoch": 11.26,
"learning_rate": 3.0908121816951465e-06,
"loss": 0.0084,
"step": 1745
},
{
"epoch": 11.26,
"learning_rate": 3.0807472311464146e-06,
"loss": 0.0121,
"step": 1746
},
{
"epoch": 11.27,
"learning_rate": 3.070695710281225e-06,
"loss": 0.009,
"step": 1747
},
{
"epoch": 11.28,
"learning_rate": 3.0606576386087207e-06,
"loss": 0.0091,
"step": 1748
},
{
"epoch": 11.28,
"learning_rate": 3.050633035611955e-06,
"loss": 0.0125,
"step": 1749
},
{
"epoch": 11.29,
"learning_rate": 3.0406219207478272e-06,
"loss": 0.0116,
"step": 1750
},
{
"epoch": 11.3,
"learning_rate": 3.0306243134470668e-06,
"loss": 0.0095,
"step": 1751
},
{
"epoch": 11.3,
"learning_rate": 3.0206402331141827e-06,
"loss": 0.0093,
"step": 1752
},
{
"epoch": 11.31,
"learning_rate": 3.0106696991274278e-06,
"loss": 0.0105,
"step": 1753
},
{
"epoch": 11.32,
"learning_rate": 3.000712730838765e-06,
"loss": 0.0103,
"step": 1754
},
{
"epoch": 11.32,
"learning_rate": 2.990769347573831e-06,
"loss": 0.0084,
"step": 1755
},
{
"epoch": 11.33,
"learning_rate": 2.9808395686318814e-06,
"loss": 0.01,
"step": 1756
},
{
"epoch": 11.34,
"learning_rate": 2.97092341328578e-06,
"loss": 0.0065,
"step": 1757
},
{
"epoch": 11.34,
"learning_rate": 2.961020900781949e-06,
"loss": 0.0092,
"step": 1758
},
{
"epoch": 11.35,
"learning_rate": 2.9511320503403196e-06,
"loss": 0.0083,
"step": 1759
},
{
"epoch": 11.35,
"learning_rate": 2.9412568811543174e-06,
"loss": 0.0095,
"step": 1760
},
{
"epoch": 11.36,
"learning_rate": 2.931395412390805e-06,
"loss": 0.01,
"step": 1761
},
{
"epoch": 11.37,
"learning_rate": 2.9215476631900574e-06,
"loss": 0.0098,
"step": 1762
},
{
"epoch": 11.37,
"learning_rate": 2.9117136526657306e-06,
"loss": 0.0134,
"step": 1763
},
{
"epoch": 11.38,
"learning_rate": 2.901893399904797e-06,
"loss": 0.0113,
"step": 1764
},
{
"epoch": 11.39,
"learning_rate": 2.8920869239675385e-06,
"loss": 0.0095,
"step": 1765
},
{
"epoch": 11.39,
"learning_rate": 2.8822942438874977e-06,
"loss": 0.0096,
"step": 1766
},
{
"epoch": 11.4,
"learning_rate": 2.872515378671431e-06,
"loss": 0.0079,
"step": 1767
},
{
"epoch": 11.41,
"learning_rate": 2.86275034729929e-06,
"loss": 0.0078,
"step": 1768
},
{
"epoch": 11.41,
"learning_rate": 2.8529991687241763e-06,
"loss": 0.0101,
"step": 1769
},
{
"epoch": 11.42,
"learning_rate": 2.843261861872296e-06,
"loss": 0.0059,
"step": 1770
},
{
"epoch": 11.43,
"learning_rate": 2.8335384456429404e-06,
"loss": 0.0099,
"step": 1771
},
{
"epoch": 11.43,
"learning_rate": 2.823828938908436e-06,
"loss": 0.0094,
"step": 1772
},
{
"epoch": 11.44,
"learning_rate": 2.814133360514113e-06,
"loss": 0.0073,
"step": 1773
},
{
"epoch": 11.45,
"learning_rate": 2.804451729278271e-06,
"loss": 0.0133,
"step": 1774
},
{
"epoch": 11.45,
"learning_rate": 2.7947840639921308e-06,
"loss": 0.0098,
"step": 1775
},
{
"epoch": 11.46,
"learning_rate": 2.785130383419813e-06,
"loss": 0.0113,
"step": 1776
},
{
"epoch": 11.46,
"learning_rate": 2.7754907062982995e-06,
"loss": 0.0077,
"step": 1777
},
{
"epoch": 11.47,
"learning_rate": 2.7658650513373788e-06,
"loss": 0.0093,
"step": 1778
},
{
"epoch": 11.48,
"learning_rate": 2.756253437219639e-06,
"loss": 0.0115,
"step": 1779
},
{
"epoch": 11.48,
"learning_rate": 2.7466558826003996e-06,
"loss": 0.01,
"step": 1780
},
{
"epoch": 11.49,
"learning_rate": 2.7370724061077105e-06,
"loss": 0.0109,
"step": 1781
},
{
"epoch": 11.5,
"learning_rate": 2.7275030263422863e-06,
"loss": 0.0121,
"step": 1782
},
{
"epoch": 11.5,
"learning_rate": 2.7179477618774785e-06,
"loss": 0.0099,
"step": 1783
},
{
"epoch": 11.51,
"learning_rate": 2.7084066312592495e-06,
"loss": 0.0071,
"step": 1784
},
{
"epoch": 11.52,
"learning_rate": 2.698879653006127e-06,
"loss": 0.0134,
"step": 1785
},
{
"epoch": 11.52,
"learning_rate": 2.6893668456091627e-06,
"loss": 0.0085,
"step": 1786
},
{
"epoch": 11.53,
"learning_rate": 2.6798682275319142e-06,
"loss": 0.0132,
"step": 1787
},
{
"epoch": 11.54,
"learning_rate": 2.670383817210397e-06,
"loss": 0.0102,
"step": 1788
},
{
"epoch": 11.54,
"learning_rate": 2.6609136330530416e-06,
"loss": 0.0152,
"step": 1789
},
{
"epoch": 11.55,
"learning_rate": 2.651457693440678e-06,
"loss": 0.0126,
"step": 1790
},
{
"epoch": 11.55,
"learning_rate": 2.642016016726482e-06,
"loss": 0.0126,
"step": 1791
},
{
"epoch": 11.56,
"learning_rate": 2.6325886212359496e-06,
"loss": 0.0107,
"step": 1792
},
{
"epoch": 11.57,
"learning_rate": 2.62317552526686e-06,
"loss": 0.0074,
"step": 1793
},
{
"epoch": 11.57,
"learning_rate": 2.6137767470892293e-06,
"loss": 0.0082,
"step": 1794
},
{
"epoch": 11.58,
"learning_rate": 2.6043923049452913e-06,
"loss": 0.0104,
"step": 1795
},
{
"epoch": 11.59,
"learning_rate": 2.5950222170494576e-06,
"loss": 0.0091,
"step": 1796
},
{
"epoch": 11.59,
"learning_rate": 2.5856665015882686e-06,
"loss": 0.0074,
"step": 1797
},
{
"epoch": 11.6,
"learning_rate": 2.57632517672038e-06,
"loss": 0.0099,
"step": 1798
},
{
"epoch": 11.61,
"learning_rate": 2.56699826057651e-06,
"loss": 0.0087,
"step": 1799
},
{
"epoch": 11.61,
"learning_rate": 2.557685771259414e-06,
"loss": 0.01,
"step": 1800
},
{
"epoch": 11.62,
"learning_rate": 2.5483877268438474e-06,
"loss": 0.0103,
"step": 1801
},
{
"epoch": 11.63,
"learning_rate": 2.539104145376521e-06,
"loss": 0.0071,
"step": 1802
},
{
"epoch": 11.63,
"learning_rate": 2.529835044876083e-06,
"loss": 0.0131,
"step": 1803
},
{
"epoch": 11.64,
"learning_rate": 2.520580443333077e-06,
"loss": 0.0079,
"step": 1804
},
{
"epoch": 11.65,
"learning_rate": 2.5113403587098916e-06,
"loss": 0.0107,
"step": 1805
},
{
"epoch": 11.65,
"learning_rate": 2.5021148089407543e-06,
"loss": 0.0092,
"step": 1806
},
{
"epoch": 11.66,
"learning_rate": 2.4929038119316764e-06,
"loss": 0.0098,
"step": 1807
},
{
"epoch": 11.66,
"learning_rate": 2.4837073855604186e-06,
"loss": 0.0116,
"step": 1808
},
{
"epoch": 11.67,
"learning_rate": 2.4745255476764683e-06,
"loss": 0.0087,
"step": 1809
},
{
"epoch": 11.68,
"learning_rate": 2.465358316100994e-06,
"loss": 0.0095,
"step": 1810
},
{
"epoch": 11.68,
"learning_rate": 2.4562057086268165e-06,
"loss": 0.0086,
"step": 1811
},
{
"epoch": 11.69,
"learning_rate": 2.447067743018374e-06,
"loss": 0.011,
"step": 1812
},
{
"epoch": 11.7,
"learning_rate": 2.437944437011679e-06,
"loss": 0.0105,
"step": 1813
},
{
"epoch": 11.7,
"learning_rate": 2.428835808314298e-06,
"loss": 0.0081,
"step": 1814
},
{
"epoch": 11.71,
"learning_rate": 2.4197418746053105e-06,
"loss": 0.0097,
"step": 1815
},
{
"epoch": 11.72,
"learning_rate": 2.4106626535352664e-06,
"loss": 0.0177,
"step": 1816
},
{
"epoch": 11.72,
"learning_rate": 2.4015981627261663e-06,
"loss": 0.0116,
"step": 1817
},
{
"epoch": 11.73,
"learning_rate": 2.39254841977142e-06,
"loss": 0.0075,
"step": 1818
},
{
"epoch": 11.74,
"learning_rate": 2.383513442235812e-06,
"loss": 0.0101,
"step": 1819
},
{
"epoch": 11.74,
"learning_rate": 2.3744932476554714e-06,
"loss": 0.0073,
"step": 1820
},
{
"epoch": 11.75,
"learning_rate": 2.3654878535378256e-06,
"loss": 0.0122,
"step": 1821
},
{
"epoch": 11.75,
"learning_rate": 2.356497277361585e-06,
"loss": 0.0074,
"step": 1822
},
{
"epoch": 11.76,
"learning_rate": 2.347521536576699e-06,
"loss": 0.0064,
"step": 1823
},
{
"epoch": 11.77,
"learning_rate": 2.3385606486043156e-06,
"loss": 0.0107,
"step": 1824
},
{
"epoch": 11.77,
"learning_rate": 2.3296146308367597e-06,
"loss": 0.0085,
"step": 1825
},
{
"epoch": 11.78,
"learning_rate": 2.3206835006374994e-06,
"loss": 0.0125,
"step": 1826
},
{
"epoch": 11.79,
"learning_rate": 2.311767275341096e-06,
"loss": 0.0129,
"step": 1827
},
{
"epoch": 11.79,
"learning_rate": 2.3028659722531886e-06,
"loss": 0.0057,
"step": 1828
},
{
"epoch": 11.8,
"learning_rate": 2.293979608650454e-06,
"loss": 0.0112,
"step": 1829
},
{
"epoch": 11.81,
"learning_rate": 2.2851082017805704e-06,
"loss": 0.0088,
"step": 1830
},
{
"epoch": 11.81,
"learning_rate": 2.27625176886219e-06,
"loss": 0.0062,
"step": 1831
},
{
"epoch": 11.82,
"learning_rate": 2.2674103270848947e-06,
"loss": 0.0079,
"step": 1832
},
{
"epoch": 11.83,
"learning_rate": 2.2585838936091753e-06,
"loss": 0.012,
"step": 1833
},
{
"epoch": 11.83,
"learning_rate": 2.249772485566395e-06,
"loss": 0.0113,
"step": 1834
},
{
"epoch": 11.84,
"learning_rate": 2.2409761200587455e-06,
"loss": 0.0071,
"step": 1835
},
{
"epoch": 11.85,
"learning_rate": 2.2321948141592287e-06,
"loss": 0.0116,
"step": 1836
},
{
"epoch": 11.85,
"learning_rate": 2.2234285849116155e-06,
"loss": 0.0089,
"step": 1837
},
{
"epoch": 11.86,
"learning_rate": 2.2146774493304144e-06,
"loss": 0.0095,
"step": 1838
},
{
"epoch": 11.86,
"learning_rate": 2.2059414244008438e-06,
"loss": 0.0115,
"step": 1839
},
{
"epoch": 11.87,
"learning_rate": 2.1972205270787782e-06,
"loss": 0.0116,
"step": 1840
},
{
"epoch": 11.88,
"learning_rate": 2.1885147742907465e-06,
"loss": 0.0087,
"step": 1841
},
{
"epoch": 11.88,
"learning_rate": 2.1798241829338794e-06,
"loss": 0.0121,
"step": 1842
},
{
"epoch": 11.89,
"learning_rate": 2.1711487698758728e-06,
"loss": 0.0105,
"step": 1843
},
{
"epoch": 11.9,
"learning_rate": 2.162488551954971e-06,
"loss": 0.0078,
"step": 1844
},
{
"epoch": 11.9,
"learning_rate": 2.153843545979927e-06,
"loss": 0.0139,
"step": 1845
},
{
"epoch": 11.91,
"learning_rate": 2.1452137687299592e-06,
"loss": 0.0094,
"step": 1846
},
{
"epoch": 11.92,
"learning_rate": 2.136599236954737e-06,
"loss": 0.0075,
"step": 1847
},
{
"epoch": 11.92,
"learning_rate": 2.1279999673743367e-06,
"loss": 0.0075,
"step": 1848
},
{
"epoch": 11.93,
"learning_rate": 2.119415976679213e-06,
"loss": 0.0072,
"step": 1849
},
{
"epoch": 11.94,
"learning_rate": 2.1108472815301673e-06,
"loss": 0.0048,
"step": 1850
},
{
"epoch": 11.94,
"learning_rate": 2.102293898558305e-06,
"loss": 0.0092,
"step": 1851
},
{
"epoch": 11.95,
"learning_rate": 2.093755844365021e-06,
"loss": 0.0122,
"step": 1852
},
{
"epoch": 11.95,
"learning_rate": 2.0852331355219578e-06,
"loss": 0.0086,
"step": 1853
},
{
"epoch": 11.96,
"learning_rate": 2.0767257885709658e-06,
"loss": 0.0079,
"step": 1854
},
{
"epoch": 11.97,
"learning_rate": 2.068233820024088e-06,
"loss": 0.0047,
"step": 1855
},
{
"epoch": 11.97,
"learning_rate": 2.0597572463635163e-06,
"loss": 0.0108,
"step": 1856
},
{
"epoch": 11.97,
"eval_val_accuracy": 91.61084529505582,
"eval_val_average": 64.3830309517246,
"eval_val_loss": 2.0551986694335938,
"eval_val_recall": 37.155216608393395,
"eval_val_runtime": 4342.5733,
"eval_val_samples_per_second": 1.446,
"eval_val_steps_per_second": 0.006,
"step": 1856
},
{
"epoch": 11.98,
"learning_rate": 2.0512960840415617e-06,
"loss": 0.0079,
"step": 1857
},
{
"epoch": 11.99,
"learning_rate": 2.042850349480626e-06,
"loss": 0.0096,
"step": 1858
},
{
"epoch": 11.99,
"learning_rate": 2.034420059073159e-06,
"loss": 0.0098,
"step": 1859
},
{
"epoch": 12.0,
"learning_rate": 2.0260052291816445e-06,
"loss": 0.0125,
"step": 1860
},
{
"epoch": 12.01,
"learning_rate": 2.0176058761385565e-06,
"loss": 0.009,
"step": 1861
},
{
"epoch": 12.01,
"learning_rate": 2.0092220162463206e-06,
"loss": 0.009,
"step": 1862
},
{
"epoch": 12.02,
"learning_rate": 2.000853665777305e-06,
"loss": 0.0051,
"step": 1863
},
{
"epoch": 12.03,
"learning_rate": 1.9925008409737678e-06,
"loss": 0.0094,
"step": 1864
},
{
"epoch": 12.03,
"learning_rate": 1.9841635580478325e-06,
"loss": 0.0091,
"step": 1865
},
{
"epoch": 12.04,
"learning_rate": 1.97584183318146e-06,
"loss": 0.0068,
"step": 1866
},
{
"epoch": 12.05,
"learning_rate": 1.9675356825264135e-06,
"loss": 0.0094,
"step": 1867
},
{
"epoch": 12.05,
"learning_rate": 1.9592451222042288e-06,
"loss": 0.0079,
"step": 1868
},
{
"epoch": 12.06,
"learning_rate": 1.9509701683061823e-06,
"loss": 0.0102,
"step": 1869
},
{
"epoch": 12.06,
"learning_rate": 1.9427108368932534e-06,
"loss": 0.0095,
"step": 1870
},
{
"epoch": 12.07,
"learning_rate": 1.9344671439961093e-06,
"loss": 0.0058,
"step": 1871
},
{
"epoch": 12.08,
"learning_rate": 1.926239105615061e-06,
"loss": 0.0111,
"step": 1872
},
{
"epoch": 12.08,
"learning_rate": 1.918026737720029e-06,
"loss": 0.0071,
"step": 1873
},
{
"epoch": 12.09,
"learning_rate": 1.9098300562505266e-06,
"loss": 0.0062,
"step": 1874
},
{
"epoch": 12.1,
"learning_rate": 1.901649077115617e-06,
"loss": 0.01,
"step": 1875
},
{
"epoch": 12.1,
"learning_rate": 1.8934838161938896e-06,
"loss": 0.01,
"step": 1876
},
{
"epoch": 12.11,
"learning_rate": 1.8853342893334258e-06,
"loss": 0.0071,
"step": 1877
},
{
"epoch": 12.12,
"learning_rate": 1.8772005123517624e-06,
"loss": 0.0076,
"step": 1878
},
{
"epoch": 12.12,
"learning_rate": 1.8690825010358727e-06,
"loss": 0.0129,
"step": 1879
},
{
"epoch": 12.13,
"learning_rate": 1.8609802711421331e-06,
"loss": 0.009,
"step": 1880
},
{
"epoch": 12.14,
"learning_rate": 1.852893838396279e-06,
"loss": 0.0075,
"step": 1881
},
{
"epoch": 12.14,
"learning_rate": 1.8448232184933957e-06,
"loss": 0.0082,
"step": 1882
},
{
"epoch": 12.15,
"learning_rate": 1.8367684270978747e-06,
"loss": 0.0079,
"step": 1883
},
{
"epoch": 12.15,
"learning_rate": 1.8287294798433774e-06,
"loss": 0.0112,
"step": 1884
},
{
"epoch": 12.16,
"learning_rate": 1.820706392332824e-06,
"loss": 0.0105,
"step": 1885
},
{
"epoch": 12.17,
"learning_rate": 1.8126991801383476e-06,
"loss": 0.0077,
"step": 1886
},
{
"epoch": 12.17,
"learning_rate": 1.8047078588012667e-06,
"loss": 0.0066,
"step": 1887
},
{
"epoch": 12.18,
"learning_rate": 1.7967324438320643e-06,
"loss": 0.0078,
"step": 1888
},
{
"epoch": 12.19,
"learning_rate": 1.788772950710338e-06,
"loss": 0.0103,
"step": 1889
},
{
"epoch": 12.19,
"learning_rate": 1.7808293948847944e-06,
"loss": 0.0071,
"step": 1890
},
{
"epoch": 12.2,
"learning_rate": 1.772901791773204e-06,
"loss": 0.0069,
"step": 1891
},
{
"epoch": 12.21,
"learning_rate": 1.7649901567623673e-06,
"loss": 0.0108,
"step": 1892
},
{
"epoch": 12.21,
"learning_rate": 1.7570945052081e-06,
"loss": 0.0092,
"step": 1893
},
{
"epoch": 12.22,
"learning_rate": 1.749214852435195e-06,
"loss": 0.0107,
"step": 1894
},
{
"epoch": 12.23,
"learning_rate": 1.74135121373739e-06,
"loss": 0.0084,
"step": 1895
},
{
"epoch": 12.23,
"learning_rate": 1.7335036043773402e-06,
"loss": 0.0082,
"step": 1896
},
{
"epoch": 12.24,
"learning_rate": 1.7256720395865955e-06,
"loss": 0.0102,
"step": 1897
},
{
"epoch": 12.25,
"learning_rate": 1.7178565345655529e-06,
"loss": 0.0067,
"step": 1898
},
{
"epoch": 12.25,
"learning_rate": 1.7100571044834512e-06,
"loss": 0.0099,
"step": 1899
},
{
"epoch": 12.26,
"learning_rate": 1.7022737644783183e-06,
"loss": 0.007,
"step": 1900
},
{
"epoch": 12.26,
"learning_rate": 1.6945065296569618e-06,
"loss": 0.0065,
"step": 1901
},
{
"epoch": 12.27,
"learning_rate": 1.6867554150949272e-06,
"loss": 0.0117,
"step": 1902
},
{
"epoch": 12.28,
"learning_rate": 1.6790204358364671e-06,
"loss": 0.0094,
"step": 1903
},
{
"epoch": 12.28,
"learning_rate": 1.6713016068945243e-06,
"loss": 0.008,
"step": 1904
},
{
"epoch": 12.29,
"learning_rate": 1.6635989432506905e-06,
"loss": 0.0075,
"step": 1905
},
{
"epoch": 12.3,
"learning_rate": 1.6559124598551845e-06,
"loss": 0.0124,
"step": 1906
},
{
"epoch": 12.3,
"learning_rate": 1.6482421716268215e-06,
"loss": 0.0088,
"step": 1907
},
{
"epoch": 12.31,
"learning_rate": 1.6405880934529762e-06,
"loss": 0.0078,
"step": 1908
},
{
"epoch": 12.32,
"learning_rate": 1.6329502401895691e-06,
"loss": 0.0094,
"step": 1909
},
{
"epoch": 12.32,
"learning_rate": 1.625328626661028e-06,
"loss": 0.0053,
"step": 1910
},
{
"epoch": 12.33,
"learning_rate": 1.617723267660255e-06,
"loss": 0.0068,
"step": 1911
},
{
"epoch": 12.34,
"learning_rate": 1.6101341779486102e-06,
"loss": 0.0082,
"step": 1912
},
{
"epoch": 12.34,
"learning_rate": 1.602561372255873e-06,
"loss": 0.0061,
"step": 1913
},
{
"epoch": 12.35,
"learning_rate": 1.5950048652802186e-06,
"loss": 0.007,
"step": 1914
},
{
"epoch": 12.35,
"learning_rate": 1.587464671688187e-06,
"loss": 0.0147,
"step": 1915
},
{
"epoch": 12.36,
"learning_rate": 1.579940806114658e-06,
"loss": 0.0068,
"step": 1916
},
{
"epoch": 12.37,
"learning_rate": 1.5724332831628109e-06,
"loss": 0.0111,
"step": 1917
},
{
"epoch": 12.37,
"learning_rate": 1.564942117404119e-06,
"loss": 0.0106,
"step": 1918
},
{
"epoch": 12.38,
"learning_rate": 1.5574673233782932e-06,
"loss": 0.0129,
"step": 1919
},
{
"epoch": 12.39,
"learning_rate": 1.5500089155932808e-06,
"loss": 0.0141,
"step": 1920
},
{
"epoch": 12.39,
"learning_rate": 1.5425669085252194e-06,
"loss": 0.0119,
"step": 1921
},
{
"epoch": 12.4,
"learning_rate": 1.5351413166184114e-06,
"loss": 0.0077,
"step": 1922
},
{
"epoch": 12.41,
"learning_rate": 1.5277321542853018e-06,
"loss": 0.0074,
"step": 1923
},
{
"epoch": 12.41,
"learning_rate": 1.5203394359064505e-06,
"loss": 0.0075,
"step": 1924
},
{
"epoch": 12.42,
"learning_rate": 1.512963175830494e-06,
"loss": 0.0069,
"step": 1925
},
{
"epoch": 12.43,
"learning_rate": 1.5056033883741317e-06,
"loss": 0.0109,
"step": 1926
},
{
"epoch": 12.43,
"learning_rate": 1.4982600878220877e-06,
"loss": 0.0075,
"step": 1927
},
{
"epoch": 12.44,
"learning_rate": 1.490933288427082e-06,
"loss": 0.012,
"step": 1928
},
{
"epoch": 12.45,
"learning_rate": 1.4836230044098164e-06,
"loss": 0.0052,
"step": 1929
},
{
"epoch": 12.45,
"learning_rate": 1.4763292499589299e-06,
"loss": 0.0109,
"step": 1930
},
{
"epoch": 12.46,
"learning_rate": 1.4690520392309825e-06,
"loss": 0.0099,
"step": 1931
},
{
"epoch": 12.46,
"learning_rate": 1.4617913863504241e-06,
"loss": 0.0062,
"step": 1932
},
{
"epoch": 12.47,
"learning_rate": 1.454547305409567e-06,
"loss": 0.0072,
"step": 1933
},
{
"epoch": 12.48,
"learning_rate": 1.4473198104685593e-06,
"loss": 0.0089,
"step": 1934
},
{
"epoch": 12.48,
"learning_rate": 1.4401089155553582e-06,
"loss": 0.0127,
"step": 1935
},
{
"epoch": 12.49,
"learning_rate": 1.4329146346656963e-06,
"loss": 0.0094,
"step": 1936
},
{
"epoch": 12.5,
"learning_rate": 1.4257369817630674e-06,
"loss": 0.006,
"step": 1937
},
{
"epoch": 12.5,
"learning_rate": 1.4185759707786818e-06,
"loss": 0.0083,
"step": 1938
},
{
"epoch": 12.51,
"learning_rate": 1.4114316156114593e-06,
"loss": 0.0102,
"step": 1939
},
{
"epoch": 12.52,
"learning_rate": 1.4043039301279904e-06,
"loss": 0.0107,
"step": 1940
},
{
"epoch": 12.52,
"learning_rate": 1.397192928162504e-06,
"loss": 0.0125,
"step": 1941
},
{
"epoch": 12.53,
"learning_rate": 1.3900986235168544e-06,
"loss": 0.0088,
"step": 1942
},
{
"epoch": 12.54,
"learning_rate": 1.3830210299604863e-06,
"loss": 0.0077,
"step": 1943
},
{
"epoch": 12.54,
"learning_rate": 1.3759601612304097e-06,
"loss": 0.0102,
"step": 1944
},
{
"epoch": 12.55,
"learning_rate": 1.3689160310311722e-06,
"loss": 0.0118,
"step": 1945
},
{
"epoch": 12.55,
"learning_rate": 1.3618886530348362e-06,
"loss": 0.0084,
"step": 1946
},
{
"epoch": 12.56,
"learning_rate": 1.3548780408809425e-06,
"loss": 0.0073,
"step": 1947
},
{
"epoch": 12.57,
"learning_rate": 1.3478842081765008e-06,
"loss": 0.0078,
"step": 1948
},
{
"epoch": 12.57,
"learning_rate": 1.340907168495943e-06,
"loss": 0.0109,
"step": 1949
},
{
"epoch": 12.58,
"learning_rate": 1.3339469353811142e-06,
"loss": 0.0082,
"step": 1950
},
{
"epoch": 12.59,
"learning_rate": 1.327003522341237e-06,
"loss": 0.0081,
"step": 1951
},
{
"epoch": 12.59,
"learning_rate": 1.3200769428528881e-06,
"loss": 0.0077,
"step": 1952
},
{
"epoch": 12.6,
"learning_rate": 1.3131672103599723e-06,
"loss": 0.0068,
"step": 1953
},
{
"epoch": 12.61,
"learning_rate": 1.3062743382736964e-06,
"loss": 0.0085,
"step": 1954
},
{
"epoch": 12.61,
"learning_rate": 1.2993983399725374e-06,
"loss": 0.0047,
"step": 1955
},
{
"epoch": 12.62,
"learning_rate": 1.2925392288022299e-06,
"loss": 0.0098,
"step": 1956
},
{
"epoch": 12.63,
"learning_rate": 1.2856970180757233e-06,
"loss": 0.0122,
"step": 1957
},
{
"epoch": 12.63,
"learning_rate": 1.2788717210731715e-06,
"loss": 0.0081,
"step": 1958
},
{
"epoch": 12.64,
"learning_rate": 1.272063351041899e-06,
"loss": 0.0068,
"step": 1959
},
{
"epoch": 12.65,
"learning_rate": 1.2652719211963726e-06,
"loss": 0.0085,
"step": 1960
},
{
"epoch": 12.65,
"learning_rate": 1.2584974447181808e-06,
"loss": 0.0092,
"step": 1961
},
{
"epoch": 12.66,
"learning_rate": 1.2517399347560165e-06,
"loss": 0.0152,
"step": 1962
},
{
"epoch": 12.66,
"learning_rate": 1.2449994044256276e-06,
"loss": 0.0148,
"step": 1963
},
{
"epoch": 12.67,
"learning_rate": 1.2382758668098126e-06,
"loss": 0.0114,
"step": 1964
},
{
"epoch": 12.68,
"learning_rate": 1.2315693349583923e-06,
"loss": 0.0112,
"step": 1965
},
{
"epoch": 12.68,
"learning_rate": 1.224879821888173e-06,
"loss": 0.0099,
"step": 1966
},
{
"epoch": 12.69,
"learning_rate": 1.2182073405829376e-06,
"loss": 0.0117,
"step": 1967
},
{
"epoch": 12.7,
"learning_rate": 1.211551903993401e-06,
"loss": 0.0096,
"step": 1968
},
{
"epoch": 12.7,
"learning_rate": 1.2049135250372046e-06,
"loss": 0.007,
"step": 1969
},
{
"epoch": 12.71,
"learning_rate": 1.1982922165988808e-06,
"loss": 0.0067,
"step": 1970
},
{
"epoch": 12.72,
"learning_rate": 1.1916879915298275e-06,
"loss": 0.0083,
"step": 1971
},
{
"epoch": 12.72,
"learning_rate": 1.1851008626482862e-06,
"loss": 0.0059,
"step": 1972
},
{
"epoch": 12.73,
"learning_rate": 1.1785308427393182e-06,
"loss": 0.0058,
"step": 1973
},
{
"epoch": 12.74,
"learning_rate": 1.1719779445547718e-06,
"loss": 0.0051,
"step": 1974
},
{
"epoch": 12.74,
"learning_rate": 1.1654421808132687e-06,
"loss": 0.0098,
"step": 1975
},
{
"epoch": 12.75,
"learning_rate": 1.158923564200176e-06,
"loss": 0.0129,
"step": 1976
},
{
"epoch": 12.75,
"learning_rate": 1.1524221073675712e-06,
"loss": 0.0103,
"step": 1977
},
{
"epoch": 12.76,
"learning_rate": 1.1459378229342355e-06,
"loss": 0.0104,
"step": 1978
},
{
"epoch": 12.77,
"learning_rate": 1.1394707234856127e-06,
"loss": 0.0088,
"step": 1979
},
{
"epoch": 12.77,
"learning_rate": 1.1330208215737937e-06,
"loss": 0.0092,
"step": 1980
},
{
"epoch": 12.78,
"learning_rate": 1.1265881297174985e-06,
"loss": 0.0096,
"step": 1981
},
{
"epoch": 12.79,
"learning_rate": 1.1201726604020313e-06,
"loss": 0.0108,
"step": 1982
},
{
"epoch": 12.79,
"learning_rate": 1.113774426079276e-06,
"loss": 0.0122,
"step": 1983
},
{
"epoch": 12.8,
"learning_rate": 1.1073934391676666e-06,
"loss": 0.0084,
"step": 1984
},
{
"epoch": 12.81,
"learning_rate": 1.101029712052153e-06,
"loss": 0.0134,
"step": 1985
},
{
"epoch": 12.81,
"learning_rate": 1.0946832570841947e-06,
"loss": 0.0128,
"step": 1986
},
{
"epoch": 12.82,
"learning_rate": 1.0883540865817189e-06,
"loss": 0.0075,
"step": 1987
},
{
"epoch": 12.83,
"learning_rate": 1.0820422128291131e-06,
"loss": 0.0095,
"step": 1988
},
{
"epoch": 12.83,
"learning_rate": 1.0757476480771878e-06,
"loss": 0.0065,
"step": 1989
},
{
"epoch": 12.84,
"learning_rate": 1.0694704045431604e-06,
"loss": 0.01,
"step": 1990
},
{
"epoch": 12.85,
"learning_rate": 1.0632104944106303e-06,
"loss": 0.0103,
"step": 1991
},
{
"epoch": 12.85,
"learning_rate": 1.0569679298295532e-06,
"loss": 0.0106,
"step": 1992
},
{
"epoch": 12.86,
"learning_rate": 1.050742722916215e-06,
"loss": 0.01,
"step": 1993
},
{
"epoch": 12.86,
"learning_rate": 1.0445348857532155e-06,
"loss": 0.0111,
"step": 1994
},
{
"epoch": 12.87,
"learning_rate": 1.0383444303894453e-06,
"loss": 0.0105,
"step": 1995
},
{
"epoch": 12.88,
"learning_rate": 1.032171368840048e-06,
"loss": 0.0102,
"step": 1996
},
{
"epoch": 12.88,
"learning_rate": 1.0260157130864178e-06,
"loss": 0.0098,
"step": 1997
},
{
"epoch": 12.89,
"learning_rate": 1.0198774750761564e-06,
"loss": 0.0087,
"step": 1998
},
{
"epoch": 12.9,
"learning_rate": 1.013756666723067e-06,
"loss": 0.0117,
"step": 1999
},
{
"epoch": 12.9,
"learning_rate": 1.007653299907122e-06,
"loss": 0.0092,
"step": 2000
},
{
"epoch": 12.91,
"learning_rate": 1.0015673864744358e-06,
"loss": 0.0079,
"step": 2001
},
{
"epoch": 12.92,
"learning_rate": 9.954989382372527e-07,
"loss": 0.0091,
"step": 2002
},
{
"epoch": 12.92,
"learning_rate": 9.894479669739176e-07,
"loss": 0.0116,
"step": 2003
},
{
"epoch": 12.93,
"learning_rate": 9.834144844288496e-07,
"loss": 0.0095,
"step": 2004
},
{
"epoch": 12.94,
"learning_rate": 9.77398502312531e-07,
"loss": 0.0126,
"step": 2005
},
{
"epoch": 12.94,
"learning_rate": 9.714000323014705e-07,
"loss": 0.01,
"step": 2006
},
{
"epoch": 12.95,
"learning_rate": 9.654190860381928e-07,
"loss": 0.0085,
"step": 2007
},
{
"epoch": 12.95,
"learning_rate": 9.594556751312055e-07,
"loss": 0.0079,
"step": 2008
},
{
"epoch": 12.96,
"learning_rate": 9.535098111549868e-07,
"loss": 0.0097,
"step": 2009
},
{
"epoch": 12.97,
"learning_rate": 9.475815056499527e-07,
"loss": 0.0109,
"step": 2010
},
{
"epoch": 12.97,
"learning_rate": 9.416707701224459e-07,
"loss": 0.0094,
"step": 2011
},
{
"epoch": 12.98,
"learning_rate": 9.357776160446985e-07,
"loss": 0.0092,
"step": 2012
},
{
"epoch": 12.99,
"learning_rate": 9.299020548548255e-07,
"loss": 0.0122,
"step": 2013
},
{
"epoch": 12.99,
"learning_rate": 9.240440979567955e-07,
"loss": 0.0083,
"step": 2014
},
{
"epoch": 13.0,
"learning_rate": 9.182037567204017e-07,
"loss": 0.0068,
"step": 2015
},
{
"epoch": 13.01,
"learning_rate": 9.123810424812573e-07,
"loss": 0.0088,
"step": 2016
},
{
"epoch": 13.01,
"learning_rate": 9.065759665407514e-07,
"loss": 0.0099,
"step": 2017
},
{
"epoch": 13.02,
"learning_rate": 9.007885401660499e-07,
"loss": 0.0079,
"step": 2018
},
{
"epoch": 13.03,
"learning_rate": 8.950187745900574e-07,
"loss": 0.0058,
"step": 2019
},
{
"epoch": 13.03,
"learning_rate": 8.892666810113959e-07,
"loss": 0.0076,
"step": 2020
},
{
"epoch": 13.04,
"learning_rate": 8.835322705943939e-07,
"loss": 0.0047,
"step": 2021
},
{
"epoch": 13.05,
"learning_rate": 8.778155544690581e-07,
"loss": 0.0094,
"step": 2022
},
{
"epoch": 13.05,
"learning_rate": 8.721165437310475e-07,
"loss": 0.0069,
"step": 2023
},
{
"epoch": 13.06,
"learning_rate": 8.664352494416606e-07,
"loss": 0.0086,
"step": 2024
},
{
"epoch": 13.06,
"learning_rate": 8.607716826278089e-07,
"loss": 0.0091,
"step": 2025
},
{
"epoch": 13.07,
"learning_rate": 8.551258542819941e-07,
"loss": 0.0092,
"step": 2026
},
{
"epoch": 13.08,
"learning_rate": 8.494977753622913e-07,
"loss": 0.0046,
"step": 2027
},
{
"epoch": 13.08,
"learning_rate": 8.438874567923261e-07,
"loss": 0.0074,
"step": 2028
},
{
"epoch": 13.09,
"learning_rate": 8.382949094612502e-07,
"loss": 0.0104,
"step": 2029
},
{
"epoch": 13.1,
"learning_rate": 8.327201442237276e-07,
"loss": 0.0048,
"step": 2030
},
{
"epoch": 13.1,
"learning_rate": 8.271631718999007e-07,
"loss": 0.0076,
"step": 2031
},
{
"epoch": 13.11,
"learning_rate": 8.216240032753842e-07,
"loss": 0.0075,
"step": 2032
},
{
"epoch": 13.12,
"learning_rate": 8.161026491012369e-07,
"loss": 0.0112,
"step": 2033
},
{
"epoch": 13.12,
"learning_rate": 8.105991200939356e-07,
"loss": 0.0076,
"step": 2034
},
{
"epoch": 13.13,
"learning_rate": 8.051134269353689e-07,
"loss": 0.0091,
"step": 2035
},
{
"epoch": 13.14,
"learning_rate": 7.996455802727965e-07,
"loss": 0.0102,
"step": 2036
},
{
"epoch": 13.14,
"learning_rate": 7.941955907188503e-07,
"loss": 0.0068,
"step": 2037
},
{
"epoch": 13.15,
"learning_rate": 7.887634688515e-07,
"loss": 0.0084,
"step": 2038
},
{
"epoch": 13.15,
"learning_rate": 7.833492252140284e-07,
"loss": 0.0104,
"step": 2039
},
{
"epoch": 13.16,
"learning_rate": 7.779528703150263e-07,
"loss": 0.0097,
"step": 2040
},
{
"epoch": 13.17,
"learning_rate": 7.725744146283643e-07,
"loss": 0.0078,
"step": 2041
},
{
"epoch": 13.17,
"learning_rate": 7.672138685931641e-07,
"loss": 0.0107,
"step": 2042
},
{
"epoch": 13.18,
"learning_rate": 7.618712426137953e-07,
"loss": 0.0075,
"step": 2043
},
{
"epoch": 13.19,
"learning_rate": 7.565465470598421e-07,
"loss": 0.0072,
"step": 2044
},
{
"epoch": 13.19,
"learning_rate": 7.512397922660853e-07,
"loss": 0.0095,
"step": 2045
},
{
"epoch": 13.2,
"learning_rate": 7.459509885324878e-07,
"loss": 0.0105,
"step": 2046
},
{
"epoch": 13.21,
"learning_rate": 7.406801461241697e-07,
"loss": 0.0105,
"step": 2047
},
{
"epoch": 13.21,
"learning_rate": 7.354272752713898e-07,
"loss": 0.0086,
"step": 2048
},
{
"epoch": 13.22,
"learning_rate": 7.30192386169527e-07,
"loss": 0.0086,
"step": 2049
},
{
"epoch": 13.23,
"learning_rate": 7.249754889790539e-07,
"loss": 0.0125,
"step": 2050
},
{
"epoch": 13.23,
"learning_rate": 7.197765938255286e-07,
"loss": 0.012,
"step": 2051
},
{
"epoch": 13.24,
"learning_rate": 7.145957107995671e-07,
"loss": 0.0099,
"step": 2052
},
{
"epoch": 13.25,
"learning_rate": 7.094328499568215e-07,
"loss": 0.0093,
"step": 2053
},
{
"epoch": 13.25,
"learning_rate": 7.042880213179692e-07,
"loss": 0.0103,
"step": 2054
},
{
"epoch": 13.26,
"learning_rate": 6.991612348686861e-07,
"loss": 0.0085,
"step": 2055
},
{
"epoch": 13.26,
"learning_rate": 6.940525005596321e-07,
"loss": 0.0078,
"step": 2056
},
{
"epoch": 13.27,
"learning_rate": 6.88961828306427e-07,
"loss": 0.0075,
"step": 2057
},
{
"epoch": 13.28,
"learning_rate": 6.838892279896336e-07,
"loss": 0.0094,
"step": 2058
},
{
"epoch": 13.28,
"learning_rate": 6.788347094547399e-07,
"loss": 0.011,
"step": 2059
},
{
"epoch": 13.29,
"learning_rate": 6.737982825121392e-07,
"loss": 0.0107,
"step": 2060
},
{
"epoch": 13.3,
"learning_rate": 6.687799569371079e-07,
"loss": 0.0128,
"step": 2061
},
{
"epoch": 13.3,
"learning_rate": 6.637797424697911e-07,
"loss": 0.0107,
"step": 2062
},
{
"epoch": 13.31,
"learning_rate": 6.587976488151838e-07,
"loss": 0.0084,
"step": 2063
},
{
"epoch": 13.32,
"learning_rate": 6.538336856431038e-07,
"loss": 0.0121,
"step": 2064
},
{
"epoch": 13.32,
"learning_rate": 6.488878625881867e-07,
"loss": 0.0066,
"step": 2065
},
{
"epoch": 13.33,
"learning_rate": 6.439601892498538e-07,
"loss": 0.0047,
"step": 2066
},
{
"epoch": 13.34,
"learning_rate": 6.390506751923031e-07,
"loss": 0.0063,
"step": 2067
},
{
"epoch": 13.34,
"learning_rate": 6.34159329944487e-07,
"loss": 0.0114,
"step": 2068
},
{
"epoch": 13.35,
"learning_rate": 6.292861630000891e-07,
"loss": 0.009,
"step": 2069
},
{
"epoch": 13.35,
"learning_rate": 6.244311838175143e-07,
"loss": 0.0074,
"step": 2070
},
{
"epoch": 13.36,
"learning_rate": 6.195944018198685e-07,
"loss": 0.0071,
"step": 2071
},
{
"epoch": 13.37,
"learning_rate": 6.147758263949322e-07,
"loss": 0.0081,
"step": 2072
},
{
"epoch": 13.37,
"learning_rate": 6.099754668951551e-07,
"loss": 0.0084,
"step": 2073
},
{
"epoch": 13.38,
"learning_rate": 6.051933326376259e-07,
"loss": 0.0064,
"step": 2074
},
{
"epoch": 13.39,
"learning_rate": 6.004294329040638e-07,
"loss": 0.0112,
"step": 2075
},
{
"epoch": 13.39,
"learning_rate": 5.956837769407964e-07,
"loss": 0.007,
"step": 2076
},
{
"epoch": 13.4,
"learning_rate": 5.909563739587365e-07,
"loss": 0.0057,
"step": 2077
},
{
"epoch": 13.41,
"learning_rate": 5.862472331333735e-07,
"loss": 0.0069,
"step": 2078
},
{
"epoch": 13.41,
"learning_rate": 5.815563636047539e-07,
"loss": 0.0073,
"step": 2079
},
{
"epoch": 13.42,
"learning_rate": 5.768837744774547e-07,
"loss": 0.0075,
"step": 2080
},
{
"epoch": 13.43,
"learning_rate": 5.722294748205759e-07,
"loss": 0.0097,
"step": 2081
},
{
"epoch": 13.43,
"learning_rate": 5.675934736677214e-07,
"loss": 0.0037,
"step": 2082
},
{
"epoch": 13.44,
"learning_rate": 5.629757800169732e-07,
"loss": 0.0109,
"step": 2083
},
{
"epoch": 13.45,
"learning_rate": 5.583764028308858e-07,
"loss": 0.0086,
"step": 2084
},
{
"epoch": 13.45,
"learning_rate": 5.537953510364613e-07,
"loss": 0.0135,
"step": 2085
},
{
"epoch": 13.46,
"learning_rate": 5.492326335251308e-07,
"loss": 0.0087,
"step": 2086
},
{
"epoch": 13.46,
"learning_rate": 5.44688259152747e-07,
"loss": 0.0094,
"step": 2087
},
{
"epoch": 13.47,
"learning_rate": 5.401622367395498e-07,
"loss": 0.01,
"step": 2088
},
{
"epoch": 13.48,
"learning_rate": 5.356545750701691e-07,
"loss": 0.0069,
"step": 2089
},
{
"epoch": 13.48,
"learning_rate": 5.311652828935943e-07,
"loss": 0.0087,
"step": 2090
},
{
"epoch": 13.49,
"learning_rate": 5.266943689231596e-07,
"loss": 0.0075,
"step": 2091
},
{
"epoch": 13.5,
"learning_rate": 5.222418418365305e-07,
"loss": 0.009,
"step": 2092
},
{
"epoch": 13.5,
"learning_rate": 5.178077102756862e-07,
"loss": 0.0117,
"step": 2093
},
{
"epoch": 13.51,
"learning_rate": 5.133919828468992e-07,
"loss": 0.0067,
"step": 2094
},
{
"epoch": 13.52,
"learning_rate": 5.089946681207247e-07,
"loss": 0.0097,
"step": 2095
},
{
"epoch": 13.52,
"learning_rate": 5.046157746319746e-07,
"loss": 0.0078,
"step": 2096
},
{
"epoch": 13.53,
"learning_rate": 5.002553108797125e-07,
"loss": 0.0064,
"step": 2097
},
{
"epoch": 13.54,
"learning_rate": 4.959132853272309e-07,
"loss": 0.0061,
"step": 2098
},
{
"epoch": 13.54,
"learning_rate": 4.915897064020292e-07,
"loss": 0.0091,
"step": 2099
},
{
"epoch": 13.55,
"learning_rate": 4.872845824958106e-07,
"loss": 0.011,
"step": 2100
},
{
"epoch": 13.55,
"learning_rate": 4.829979219644565e-07,
"loss": 0.0062,
"step": 2101
},
{
"epoch": 13.56,
"learning_rate": 4.787297331280094e-07,
"loss": 0.0079,
"step": 2102
},
{
"epoch": 13.57,
"learning_rate": 4.7448002427066486e-07,
"loss": 0.0101,
"step": 2103
},
{
"epoch": 13.57,
"learning_rate": 4.70248803640746e-07,
"loss": 0.0086,
"step": 2104
},
{
"epoch": 13.58,
"learning_rate": 4.660360794506946e-07,
"loss": 0.0087,
"step": 2105
},
{
"epoch": 13.59,
"learning_rate": 4.618418598770524e-07,
"loss": 0.0125,
"step": 2106
},
{
"epoch": 13.59,
"learning_rate": 4.5766615306044206e-07,
"loss": 0.0078,
"step": 2107
},
{
"epoch": 13.6,
"learning_rate": 4.5350896710555815e-07,
"loss": 0.0091,
"step": 2108
},
{
"epoch": 13.61,
"learning_rate": 4.493703100811475e-07,
"loss": 0.0072,
"step": 2109
},
{
"epoch": 13.61,
"learning_rate": 4.4525019001999013e-07,
"loss": 0.0138,
"step": 2110
},
{
"epoch": 13.62,
"learning_rate": 4.4114861491889147e-07,
"loss": 0.0108,
"step": 2111
},
{
"epoch": 13.63,
"learning_rate": 4.3706559273866247e-07,
"loss": 0.0072,
"step": 2112
},
{
"epoch": 13.63,
"learning_rate": 4.3300113140410163e-07,
"loss": 0.0114,
"step": 2113
},
{
"epoch": 13.64,
"learning_rate": 4.289552388039864e-07,
"loss": 0.0063,
"step": 2114
},
{
"epoch": 13.65,
"learning_rate": 4.2492792279104853e-07,
"loss": 0.0062,
"step": 2115
},
{
"epoch": 13.65,
"learning_rate": 4.209191911819688e-07,
"loss": 0.0091,
"step": 2116
},
{
"epoch": 13.66,
"learning_rate": 4.1692905175735764e-07,
"loss": 0.0089,
"step": 2117
},
{
"epoch": 13.66,
"learning_rate": 4.1295751226173354e-07,
"loss": 0.0066,
"step": 2118
},
{
"epoch": 13.67,
"learning_rate": 4.0900458040352143e-07,
"loss": 0.0096,
"step": 2119
},
{
"epoch": 13.68,
"learning_rate": 4.0507026385502747e-07,
"loss": 0.0091,
"step": 2120
},
{
"epoch": 13.68,
"learning_rate": 4.011545702524244e-07,
"loss": 0.0129,
"step": 2121
},
{
"epoch": 13.69,
"learning_rate": 3.9725750719574186e-07,
"loss": 0.0109,
"step": 2122
},
{
"epoch": 13.7,
"learning_rate": 3.9337908224885145e-07,
"loss": 0.0136,
"step": 2123
},
{
"epoch": 13.7,
"learning_rate": 3.895193029394473e-07,
"loss": 0.0081,
"step": 2124
},
{
"epoch": 13.71,
"learning_rate": 3.856781767590334e-07,
"loss": 0.0099,
"step": 2125
},
{
"epoch": 13.72,
"learning_rate": 3.8185571116290843e-07,
"loss": 0.0077,
"step": 2126
},
{
"epoch": 13.72,
"learning_rate": 3.780519135701566e-07,
"loss": 0.0109,
"step": 2127
},
{
"epoch": 13.73,
"learning_rate": 3.7426679136362773e-07,
"loss": 0.0143,
"step": 2128
},
{
"epoch": 13.74,
"learning_rate": 3.7050035188991975e-07,
"loss": 0.0063,
"step": 2129
},
{
"epoch": 13.74,
"learning_rate": 3.667526024593759e-07,
"loss": 0.0087,
"step": 2130
},
{
"epoch": 13.75,
"learning_rate": 3.6302355034605885e-07,
"loss": 0.0056,
"step": 2131
},
{
"epoch": 13.75,
"learning_rate": 3.5931320278774326e-07,
"loss": 0.0058,
"step": 2132
},
{
"epoch": 13.76,
"learning_rate": 3.55621566985902e-07,
"loss": 0.0103,
"step": 2133
},
{
"epoch": 13.77,
"learning_rate": 3.5194865010568234e-07,
"loss": 0.0117,
"step": 2134
},
{
"epoch": 13.77,
"learning_rate": 3.482944592759085e-07,
"loss": 0.0082,
"step": 2135
},
{
"epoch": 13.78,
"learning_rate": 3.44659001589055e-07,
"loss": 0.0093,
"step": 2136
},
{
"epoch": 13.79,
"learning_rate": 3.4104228410123506e-07,
"loss": 0.0079,
"step": 2137
},
{
"epoch": 13.79,
"learning_rate": 3.374443138321937e-07,
"loss": 0.0119,
"step": 2138
},
{
"epoch": 13.8,
"learning_rate": 3.3386509776528576e-07,
"loss": 0.0099,
"step": 2139
},
{
"epoch": 13.81,
"learning_rate": 3.3030464284746434e-07,
"loss": 0.0079,
"step": 2140
},
{
"epoch": 13.81,
"learning_rate": 3.267629559892738e-07,
"loss": 0.0086,
"step": 2141
},
{
"epoch": 13.82,
"learning_rate": 3.232400440648265e-07,
"loss": 0.0118,
"step": 2142
},
{
"epoch": 13.83,
"learning_rate": 3.1973591391179616e-07,
"loss": 0.0069,
"step": 2143
},
{
"epoch": 13.83,
"learning_rate": 3.1625057233140466e-07,
"loss": 0.0084,
"step": 2144
},
{
"epoch": 13.84,
"learning_rate": 3.127840260884019e-07,
"loss": 0.0087,
"step": 2145
},
{
"epoch": 13.85,
"learning_rate": 3.0933628191106234e-07,
"loss": 0.0076,
"step": 2146
},
{
"epoch": 13.85,
"learning_rate": 3.0590734649116547e-07,
"loss": 0.0063,
"step": 2147
},
{
"epoch": 13.86,
"learning_rate": 3.0249722648398204e-07,
"loss": 0.0105,
"step": 2148
},
{
"epoch": 13.86,
"learning_rate": 2.9910592850826983e-07,
"loss": 0.0125,
"step": 2149
},
{
"epoch": 13.87,
"learning_rate": 2.95733459146248e-07,
"loss": 0.005,
"step": 2150
},
{
"epoch": 13.88,
"learning_rate": 2.9237982494359494e-07,
"loss": 0.0103,
"step": 2151
},
{
"epoch": 13.88,
"learning_rate": 2.8904503240943073e-07,
"loss": 0.0077,
"step": 2152
},
{
"epoch": 13.89,
"learning_rate": 2.8572908801630663e-07,
"loss": 0.0088,
"step": 2153
},
{
"epoch": 13.9,
"learning_rate": 2.824319982001866e-07,
"loss": 0.0136,
"step": 2154
},
{
"epoch": 13.9,
"learning_rate": 2.7915376936044626e-07,
"loss": 0.0083,
"step": 2155
},
{
"epoch": 13.91,
"learning_rate": 2.758944078598469e-07,
"loss": 0.0069,
"step": 2156
},
{
"epoch": 13.92,
"learning_rate": 2.726539200245348e-07,
"loss": 0.0123,
"step": 2157
},
{
"epoch": 13.92,
"learning_rate": 2.694323121440223e-07,
"loss": 0.0068,
"step": 2158
},
{
"epoch": 13.93,
"learning_rate": 2.662295904711765e-07,
"loss": 0.0083,
"step": 2159
},
{
"epoch": 13.94,
"learning_rate": 2.6304576122221035e-07,
"loss": 0.0061,
"step": 2160
},
{
"epoch": 13.94,
"learning_rate": 2.5988083057666534e-07,
"loss": 0.0132,
"step": 2161
},
{
"epoch": 13.95,
"learning_rate": 2.567348046774043e-07,
"loss": 0.0061,
"step": 2162
},
{
"epoch": 13.95,
"learning_rate": 2.5360768963059833e-07,
"loss": 0.0084,
"step": 2163
},
{
"epoch": 13.96,
"learning_rate": 2.504994915057102e-07,
"loss": 0.0084,
"step": 2164
},
{
"epoch": 13.97,
"learning_rate": 2.474102163354908e-07,
"loss": 0.0096,
"step": 2165
},
{
"epoch": 13.97,
"learning_rate": 2.443398701159627e-07,
"loss": 0.0091,
"step": 2166
},
{
"epoch": 13.98,
"learning_rate": 2.4128845880640573e-07,
"loss": 0.0064,
"step": 2167
},
{
"epoch": 13.99,
"learning_rate": 2.3825598832935115e-07,
"loss": 0.0084,
"step": 2168
},
{
"epoch": 13.99,
"learning_rate": 2.3524246457056864e-07,
"loss": 0.0106,
"step": 2169
},
{
"epoch": 14.0,
"learning_rate": 2.3224789337905062e-07,
"loss": 0.008,
"step": 2170
},
{
"epoch": 14.01,
"learning_rate": 2.292722805670078e-07,
"loss": 0.0081,
"step": 2171
},
{
"epoch": 14.01,
"learning_rate": 2.263156319098525e-07,
"loss": 0.0118,
"step": 2172
},
{
"epoch": 14.02,
"learning_rate": 2.2337795314618771e-07,
"loss": 0.0094,
"step": 2173
},
{
"epoch": 14.03,
"learning_rate": 2.2045924997780131e-07,
"loss": 0.0124,
"step": 2174
},
{
"epoch": 14.03,
"learning_rate": 2.175595280696463e-07,
"loss": 0.0089,
"step": 2175
},
{
"epoch": 14.04,
"learning_rate": 2.1467879304983953e-07,
"loss": 0.009,
"step": 2176
},
{
"epoch": 14.05,
"learning_rate": 2.1181705050964397e-07,
"loss": 0.0093,
"step": 2177
},
{
"epoch": 14.05,
"learning_rate": 2.0897430600345658e-07,
"loss": 0.0087,
"step": 2178
},
{
"epoch": 14.06,
"learning_rate": 2.0615056504880494e-07,
"loss": 0.0099,
"step": 2179
},
{
"epoch": 14.06,
"learning_rate": 2.0334583312633383e-07,
"loss": 0.0145,
"step": 2180
},
{
"epoch": 14.07,
"learning_rate": 2.0056011567978763e-07,
"loss": 0.0109,
"step": 2181
},
{
"epoch": 14.08,
"learning_rate": 1.9779341811600795e-07,
"loss": 0.0112,
"step": 2182
},
{
"epoch": 14.08,
"learning_rate": 1.9504574580492153e-07,
"loss": 0.0066,
"step": 2183
},
{
"epoch": 14.09,
"learning_rate": 1.9231710407952575e-07,
"loss": 0.01,
"step": 2184
},
{
"epoch": 14.1,
"learning_rate": 1.896074982358853e-07,
"loss": 0.008,
"step": 2185
},
{
"epoch": 14.1,
"learning_rate": 1.8691693353311446e-07,
"loss": 0.0085,
"step": 2186
},
{
"epoch": 14.11,
"learning_rate": 1.8424541519337036e-07,
"loss": 0.0037,
"step": 2187
},
{
"epoch": 14.12,
"learning_rate": 1.8159294840184638e-07,
"loss": 0.0069,
"step": 2188
},
{
"epoch": 14.12,
"learning_rate": 1.7895953830675327e-07,
"loss": 0.0107,
"step": 2189
},
{
"epoch": 14.13,
"learning_rate": 1.7634519001931916e-07,
"loss": 0.0038,
"step": 2190
},
{
"epoch": 14.14,
"learning_rate": 1.7374990861377173e-07,
"loss": 0.008,
"step": 2191
},
{
"epoch": 14.14,
"learning_rate": 1.7117369912733162e-07,
"loss": 0.0049,
"step": 2192
},
{
"epoch": 14.15,
"learning_rate": 1.6861656656020464e-07,
"loss": 0.0102,
"step": 2193
},
{
"epoch": 14.15,
"learning_rate": 1.6607851587556735e-07,
"loss": 0.0094,
"step": 2194
},
{
"epoch": 14.16,
"learning_rate": 1.6355955199956143e-07,
"loss": 0.0128,
"step": 2195
},
{
"epoch": 14.17,
"learning_rate": 1.610596798212838e-07,
"loss": 0.0106,
"step": 2196
},
{
"epoch": 14.17,
"learning_rate": 1.585789041927721e-07,
"loss": 0.0081,
"step": 2197
},
{
"epoch": 14.18,
"learning_rate": 1.5611722992900147e-07,
"loss": 0.0122,
"step": 2198
},
{
"epoch": 14.19,
"learning_rate": 1.536746618078766e-07,
"loss": 0.0094,
"step": 2199
},
{
"epoch": 14.19,
"learning_rate": 1.5125120457021303e-07,
"loss": 0.0096,
"step": 2200
},
{
"epoch": 14.2,
"learning_rate": 1.4884686291973594e-07,
"loss": 0.0085,
"step": 2201
},
{
"epoch": 14.21,
"learning_rate": 1.464616415230702e-07,
"loss": 0.0072,
"step": 2202
},
{
"epoch": 14.21,
"learning_rate": 1.44095545009727e-07,
"loss": 0.0074,
"step": 2203
},
{
"epoch": 14.22,
"learning_rate": 1.4174857797209951e-07,
"loss": 0.0068,
"step": 2204
},
{
"epoch": 14.23,
"learning_rate": 1.3942074496545166e-07,
"loss": 0.0067,
"step": 2205
},
{
"epoch": 14.23,
"learning_rate": 1.3711205050791043e-07,
"loss": 0.0072,
"step": 2206
},
{
"epoch": 14.24,
"learning_rate": 1.3482249908045696e-07,
"loss": 0.0099,
"step": 2207
},
{
"epoch": 14.25,
"learning_rate": 1.325520951269166e-07,
"loss": 0.005,
"step": 2208
},
{
"epoch": 14.25,
"learning_rate": 1.3030084305394987e-07,
"loss": 0.0084,
"step": 2209
},
{
"epoch": 14.26,
"learning_rate": 1.2806874723104824e-07,
"loss": 0.0086,
"step": 2210
},
{
"epoch": 14.26,
"learning_rate": 1.2585581199051844e-07,
"loss": 0.0099,
"step": 2211
},
{
"epoch": 14.27,
"learning_rate": 1.236620416274814e-07,
"loss": 0.0043,
"step": 2212
},
{
"epoch": 14.28,
"learning_rate": 1.2148744039986005e-07,
"loss": 0.0084,
"step": 2213
},
{
"epoch": 14.28,
"learning_rate": 1.1933201252836924e-07,
"loss": 0.0089,
"step": 2214
},
{
"epoch": 14.29,
"learning_rate": 1.1719576219651585e-07,
"loss": 0.0096,
"step": 2215
},
{
"epoch": 14.3,
"learning_rate": 1.1507869355057543e-07,
"loss": 0.0096,
"step": 2216
},
{
"epoch": 14.3,
"learning_rate": 1.1298081069960331e-07,
"loss": 0.0052,
"step": 2217
},
{
"epoch": 14.31,
"learning_rate": 1.1090211771541126e-07,
"loss": 0.0069,
"step": 2218
},
{
"epoch": 14.32,
"learning_rate": 1.0884261863256528e-07,
"loss": 0.0093,
"step": 2219
},
{
"epoch": 14.32,
"learning_rate": 1.0680231744837899e-07,
"loss": 0.0062,
"step": 2220
},
{
"epoch": 14.33,
"learning_rate": 1.0478121812290576e-07,
"loss": 0.0082,
"step": 2221
},
{
"epoch": 14.34,
"learning_rate": 1.0277932457892548e-07,
"loss": 0.0066,
"step": 2222
},
{
"epoch": 14.34,
"learning_rate": 1.0079664070194673e-07,
"loss": 0.0099,
"step": 2223
},
{
"epoch": 14.35,
"learning_rate": 9.883317034018792e-08,
"loss": 0.007,
"step": 2224
},
{
"epoch": 14.35,
"learning_rate": 9.688891730458061e-08,
"loss": 0.0106,
"step": 2225
},
{
"epoch": 14.36,
"learning_rate": 9.496388536875623e-08,
"loss": 0.0108,
"step": 2226
},
{
"epoch": 14.37,
"learning_rate": 9.305807826903601e-08,
"loss": 0.0075,
"step": 2227
},
{
"epoch": 14.37,
"learning_rate": 9.117149970443217e-08,
"loss": 0.01,
"step": 2228
},
{
"epoch": 14.38,
"learning_rate": 8.930415333663456e-08,
"loss": 0.0082,
"step": 2229
},
{
"epoch": 14.39,
"learning_rate": 8.745604279000175e-08,
"loss": 0.0113,
"step": 2230
},
{
"epoch": 14.39,
"learning_rate": 8.562717165156109e-08,
"loss": 0.0106,
"step": 2231
},
{
"epoch": 14.4,
"learning_rate": 8.381754347099758e-08,
"loss": 0.0059,
"step": 2232
},
{
"epoch": 14.41,
"learning_rate": 8.202716176064384e-08,
"loss": 0.0053,
"step": 2233
},
{
"epoch": 14.41,
"learning_rate": 8.025602999548132e-08,
"loss": 0.0117,
"step": 2234
},
{
"epoch": 14.42,
"learning_rate": 7.850415161312463e-08,
"loss": 0.0061,
"step": 2235
},
{
"epoch": 14.43,
"learning_rate": 7.677153001382387e-08,
"loss": 0.0101,
"step": 2236
},
{
"epoch": 14.43,
"learning_rate": 7.505816856045012e-08,
"loss": 0.008,
"step": 2237
},
{
"epoch": 14.44,
"learning_rate": 7.336407057849105e-08,
"loss": 0.0104,
"step": 2238
},
{
"epoch": 14.45,
"learning_rate": 7.168923935604865e-08,
"loss": 0.0095,
"step": 2239
},
{
"epoch": 14.45,
"learning_rate": 7.003367814382933e-08,
"loss": 0.0095,
"step": 2240
},
{
"epoch": 14.46,
"learning_rate": 6.839739015513492e-08,
"loss": 0.0079,
"step": 2241
},
{
"epoch": 14.46,
"learning_rate": 6.678037856586494e-08,
"loss": 0.0059,
"step": 2242
},
{
"epoch": 14.47,
"learning_rate": 6.51826465144978e-08,
"loss": 0.0073,
"step": 2243
},
{
"epoch": 14.48,
"learning_rate": 6.360419710209953e-08,
"loss": 0.0071,
"step": 2244
},
{
"epoch": 14.48,
"learning_rate": 6.204503339230505e-08,
"loss": 0.0083,
"step": 2245
},
{
"epoch": 14.49,
"learning_rate": 6.050515841132031e-08,
"loss": 0.0106,
"step": 2246
},
{
"epoch": 14.5,
"learning_rate": 5.8984575147911225e-08,
"loss": 0.0073,
"step": 2247
},
{
"epoch": 14.5,
"learning_rate": 5.748328655340141e-08,
"loss": 0.0083,
"step": 2248
},
{
"epoch": 14.51,
"learning_rate": 5.6001295541664445e-08,
"loss": 0.0105,
"step": 2249
},
{
"epoch": 14.52,
"learning_rate": 5.453860498911945e-08,
"loss": 0.008,
"step": 2250
},
{
"epoch": 14.52,
"learning_rate": 5.309521773472548e-08,
"loss": 0.0066,
"step": 2251
},
{
"epoch": 14.53,
"learning_rate": 5.167113657997491e-08,
"loss": 0.0099,
"step": 2252
},
{
"epoch": 14.54,
"learning_rate": 5.026636428888898e-08,
"loss": 0.0071,
"step": 2253
},
{
"epoch": 14.54,
"learning_rate": 4.88809035880089e-08,
"loss": 0.0098,
"step": 2254
},
{
"epoch": 14.55,
"learning_rate": 4.75147571664003e-08,
"loss": 0.0079,
"step": 2255
},
{
"epoch": 14.55,
"learning_rate": 4.616792767563549e-08,
"loss": 0.0084,
"step": 2256
},
{
"epoch": 14.56,
"learning_rate": 4.484041772979786e-08,
"loss": 0.0116,
"step": 2257
},
{
"epoch": 14.57,
"learning_rate": 4.353222990546968e-08,
"loss": 0.0094,
"step": 2258
},
{
"epoch": 14.57,
"learning_rate": 4.2243366741735457e-08,
"loss": 0.0078,
"step": 2259
},
{
"epoch": 14.58,
"learning_rate": 4.0973830740166366e-08,
"loss": 0.0072,
"step": 2260
},
{
"epoch": 14.59,
"learning_rate": 3.972362436482691e-08,
"loss": 0.0056,
"step": 2261
},
{
"epoch": 14.59,
"learning_rate": 3.84927500422605e-08,
"loss": 0.007,
"step": 2262
},
{
"epoch": 14.6,
"learning_rate": 3.728121016148945e-08,
"loss": 0.0075,
"step": 2263
},
{
"epoch": 14.61,
"learning_rate": 3.608900707401053e-08,
"loss": 0.0126,
"step": 2264
},
{
"epoch": 14.61,
"learning_rate": 3.4916143093790544e-08,
"loss": 0.0093,
"step": 2265
},
{
"epoch": 14.62,
"learning_rate": 3.376262049725632e-08,
"loss": 0.0072,
"step": 2266
},
{
"epoch": 14.63,
"learning_rate": 3.2628441523299136e-08,
"loss": 0.007,
"step": 2267
},
{
"epoch": 14.63,
"learning_rate": 3.151360837326478e-08,
"loss": 0.0101,
"step": 2268
},
{
"epoch": 14.64,
"learning_rate": 3.041812321094906e-08,
"loss": 0.0069,
"step": 2269
},
{
"epoch": 14.65,
"learning_rate": 2.9341988162595593e-08,
"loss": 0.0097,
"step": 2270
},
{
"epoch": 14.65,
"learning_rate": 2.8285205316892493e-08,
"loss": 0.0097,
"step": 2271
},
{
"epoch": 14.66,
"learning_rate": 2.724777672496348e-08,
"loss": 0.008,
"step": 2272
},
{
"epoch": 14.66,
"learning_rate": 2.6229704400372313e-08,
"loss": 0.0084,
"step": 2273
},
{
"epoch": 14.67,
"learning_rate": 2.5230990319110583e-08,
"loss": 0.0101,
"step": 2274
},
{
"epoch": 14.68,
"learning_rate": 2.42516364195966e-08,
"loss": 0.01,
"step": 2275
},
{
"epoch": 14.68,
"learning_rate": 2.32916446026743e-08,
"loss": 0.0121,
"step": 2276
},
{
"epoch": 14.69,
"learning_rate": 2.2351016731605446e-08,
"loss": 0.0087,
"step": 2277
},
{
"epoch": 14.7,
"learning_rate": 2.1429754632069645e-08,
"loss": 0.0069,
"step": 2278
},
{
"epoch": 14.7,
"learning_rate": 2.052786009215879e-08,
"loss": 0.0106,
"step": 2279
},
{
"epoch": 14.71,
"learning_rate": 1.9645334862373745e-08,
"loss": 0.0089,
"step": 2280
},
{
"epoch": 14.72,
"learning_rate": 1.8782180655622096e-08,
"loss": 0.0076,
"step": 2281
},
{
"epoch": 14.72,
"learning_rate": 1.7938399147210405e-08,
"loss": 0.0085,
"step": 2282
},
{
"epoch": 14.73,
"learning_rate": 1.7113991974849753e-08,
"loss": 0.0076,
"step": 2283
},
{
"epoch": 14.74,
"learning_rate": 1.630896073864352e-08,
"loss": 0.0102,
"step": 2284
},
{
"epoch": 14.74,
"learning_rate": 1.5523307001088505e-08,
"loss": 0.0105,
"step": 2285
},
{
"epoch": 14.75,
"learning_rate": 1.4757032287073814e-08,
"loss": 0.0057,
"step": 2286
},
{
"epoch": 14.75,
"learning_rate": 1.4010138083873082e-08,
"loss": 0.0055,
"step": 2287
},
{
"epoch": 14.76,
"learning_rate": 1.328262584114337e-08,
"loss": 0.0068,
"step": 2288
},
{
"epoch": 14.77,
"learning_rate": 1.2574496970926276e-08,
"loss": 0.0079,
"step": 2289
},
{
"epoch": 14.77,
"learning_rate": 1.1885752847637932e-08,
"loss": 0.0075,
"step": 2290
},
{
"epoch": 14.78,
"learning_rate": 1.1216394808073461e-08,
"loss": 0.0093,
"step": 2291
},
{
"epoch": 14.79,
"learning_rate": 1.0566424151401412e-08,
"loss": 0.0078,
"step": 2292
},
{
"epoch": 14.79,
"learning_rate": 9.935842139157104e-09,
"loss": 0.0108,
"step": 2293
},
{
"epoch": 14.8,
"learning_rate": 9.32464999524929e-09,
"loss": 0.0066,
"step": 2294
},
{
"epoch": 14.81,
"learning_rate": 8.732848905947944e-09,
"loss": 0.0077,
"step": 2295
},
{
"epoch": 14.81,
"learning_rate": 8.160440019890914e-09,
"loss": 0.0116,
"step": 2296
},
{
"epoch": 14.82,
"learning_rate": 7.607424448073942e-09,
"loss": 0.0103,
"step": 2297
},
{
"epoch": 14.83,
"learning_rate": 7.073803263853984e-09,
"loss": 0.0129,
"step": 2298
},
{
"epoch": 14.83,
"learning_rate": 6.55957750294367e-09,
"loss": 0.0065,
"step": 2299
},
{
"epoch": 14.84,
"learning_rate": 6.0647481634135145e-09,
"loss": 0.0114,
"step": 2300
},
{
"epoch": 14.85,
"learning_rate": 5.589316205684147e-09,
"loss": 0.0045,
"step": 2301
},
{
"epoch": 14.85,
"learning_rate": 5.13328255252854e-09,
"loss": 0.009,
"step": 2302
},
{
"epoch": 14.86,
"learning_rate": 4.696648089068667e-09,
"loss": 0.0102,
"step": 2303
},
{
"epoch": 14.86,
"learning_rate": 4.279413662777732e-09,
"loss": 0.0079,
"step": 2304
},
{
"epoch": 14.87,
"learning_rate": 3.881580083469061e-09,
"loss": 0.0093,
"step": 2305
},
{
"epoch": 14.88,
"learning_rate": 3.5031481233072097e-09,
"loss": 0.0067,
"step": 2306
},
{
"epoch": 14.88,
"learning_rate": 3.1441185167935263e-09,
"loss": 0.0045,
"step": 2307
},
{
"epoch": 14.89,
"learning_rate": 2.8044919607761458e-09,
"loss": 0.0069,
"step": 2308
},
{
"epoch": 14.9,
"learning_rate": 2.4842691144411068e-09,
"loss": 0.0107,
"step": 2309
},
{
"epoch": 14.9,
"learning_rate": 2.183450599313464e-09,
"loss": 0.0095,
"step": 2310
},
{
"epoch": 14.91,
"learning_rate": 1.9020369992561736e-09,
"loss": 0.0087,
"step": 2311
},
{
"epoch": 14.92,
"learning_rate": 1.6400288604689895e-09,
"loss": 0.0095,
"step": 2312
},
{
"epoch": 14.92,
"learning_rate": 1.397426691488457e-09,
"loss": 0.0059,
"step": 2313
},
{
"epoch": 14.93,
"learning_rate": 1.1742309631845861e-09,
"loss": 0.0087,
"step": 2314
},
{
"epoch": 14.94,
"learning_rate": 9.704421087619597e-10,
"loss": 0.0063,
"step": 2315
},
{
"epoch": 14.94,
"learning_rate": 7.860605237552943e-10,
"loss": 0.0041,
"step": 2316
},
{
"epoch": 14.95,
"learning_rate": 6.210865660360998e-10,
"loss": 0.0064,
"step": 2317
},
{
"epoch": 14.95,
"learning_rate": 4.755205558037989e-10,
"loss": 0.0095,
"step": 2318
},
{
"epoch": 14.96,
"learning_rate": 3.493627755901674e-10,
"loss": 0.0121,
"step": 2319
},
{
"epoch": 14.97,
"learning_rate": 2.4261347025489323e-10,
"loss": 0.0059,
"step": 2320
},
{
"epoch": 14.97,
"eval_val_accuracy": 91.6427432216906,
"eval_val_average": 64.37899870350543,
"eval_val_loss": 2.065849542617798,
"eval_val_recall": 37.11525418532028,
"eval_val_runtime": 4189.5083,
"eval_val_samples_per_second": 1.499,
"eval_val_steps_per_second": 0.006,
"step": 2320
},
{
"epoch": 14.97,
"learning_rate": 1.5527284699223822e-10,
"loss": 0.0088,
"step": 2321
},
{
"epoch": 14.98,
"learning_rate": 8.734107532104574e-11,
"loss": 0.0128,
"step": 2322
},
{
"epoch": 14.99,
"learning_rate": 3.881828709140223e-11,
"loss": 0.0087,
"step": 2323
},
{
"epoch": 14.99,
"learning_rate": 9.704576481306405e-12,
"loss": 0.0066,
"step": 2324
},
{
"epoch": 15.0,
"learning_rate": 0.0,
"loss": 0.0035,
"step": 2325
},
{
"epoch": 15.0,
"step": 2325,
"total_flos": 2.1423013636803133e+18,
"train_loss": 0.06349156372970151,
"train_runtime": 41225.6604,
"train_samples_per_second": 7.188,
"train_steps_per_second": 0.056
}
],
"max_steps": 2325,
"num_train_epochs": 15,
"total_flos": 2.1423013636803133e+18,
"trial_name": null,
"trial_params": null
}