exgra-med-40 / trainer_state.json
taindp98's picture
Add files using upload-large-folder tool
6dc6a9f verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.997179125528914,
"global_step": 708,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 9.090909090909091e-07,
"loss": 2.5473,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.8181818181818183e-06,
"loss": 2.5533,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 2.7272727272727272e-06,
"loss": 2.5798,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 3.6363636363636366e-06,
"loss": 2.3589,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 4.5454545454545455e-06,
"loss": 2.3804,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 5.4545454545454545e-06,
"loss": 2.3533,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 6.363636363636364e-06,
"loss": 2.5618,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 7.272727272727273e-06,
"loss": 2.1145,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 8.181818181818183e-06,
"loss": 2.3092,
"step": 9
},
{
"epoch": 0.03,
"learning_rate": 9.090909090909091e-06,
"loss": 2.468,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1e-05,
"loss": 2.326,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 1.0909090909090909e-05,
"loss": 2.3472,
"step": 12
},
{
"epoch": 0.04,
"learning_rate": 1.181818181818182e-05,
"loss": 2.4304,
"step": 13
},
{
"epoch": 0.04,
"learning_rate": 1.2727272727272728e-05,
"loss": 2.2977,
"step": 14
},
{
"epoch": 0.04,
"learning_rate": 1.3636363636363637e-05,
"loss": 2.3377,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 1.4545454545454546e-05,
"loss": 2.2761,
"step": 16
},
{
"epoch": 0.05,
"learning_rate": 1.5454545454545454e-05,
"loss": 2.1272,
"step": 17
},
{
"epoch": 0.05,
"learning_rate": 1.6363636363636366e-05,
"loss": 2.4332,
"step": 18
},
{
"epoch": 0.05,
"learning_rate": 1.7272727272727274e-05,
"loss": 2.108,
"step": 19
},
{
"epoch": 0.06,
"learning_rate": 1.8181818181818182e-05,
"loss": 1.9767,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 1.9090909090909094e-05,
"loss": 1.9631,
"step": 21
},
{
"epoch": 0.06,
"learning_rate": 2e-05,
"loss": 2.0352,
"step": 22
},
{
"epoch": 0.06,
"learning_rate": 1.9999895137366746e-05,
"loss": 1.9301,
"step": 23
},
{
"epoch": 0.07,
"learning_rate": 1.9999580551666215e-05,
"loss": 1.7,
"step": 24
},
{
"epoch": 0.07,
"learning_rate": 1.9999056249496065e-05,
"loss": 1.91,
"step": 25
},
{
"epoch": 0.07,
"learning_rate": 1.9998322241852238e-05,
"loss": 1.7236,
"step": 26
},
{
"epoch": 0.08,
"learning_rate": 1.999737854412873e-05,
"loss": 1.6519,
"step": 27
},
{
"epoch": 0.08,
"learning_rate": 1.9996225176117264e-05,
"loss": 1.6548,
"step": 28
},
{
"epoch": 0.08,
"learning_rate": 1.999486216200688e-05,
"loss": 1.8144,
"step": 29
},
{
"epoch": 0.08,
"learning_rate": 1.9993289530383433e-05,
"loss": 1.5611,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 1.999150731422898e-05,
"loss": 1.62,
"step": 31
},
{
"epoch": 0.09,
"learning_rate": 1.9989515550921088e-05,
"loss": 1.5349,
"step": 32
},
{
"epoch": 0.09,
"learning_rate": 1.998731428223208e-05,
"loss": 1.4816,
"step": 33
},
{
"epoch": 0.1,
"learning_rate": 1.9984903554328116e-05,
"loss": 1.5397,
"step": 34
},
{
"epoch": 0.1,
"learning_rate": 1.9982283417768247e-05,
"loss": 1.3161,
"step": 35
},
{
"epoch": 0.1,
"learning_rate": 1.9979453927503366e-05,
"loss": 1.365,
"step": 36
},
{
"epoch": 0.1,
"learning_rate": 1.9976415142875022e-05,
"loss": 1.4323,
"step": 37
},
{
"epoch": 0.11,
"learning_rate": 1.9973167127614218e-05,
"loss": 1.3259,
"step": 38
},
{
"epoch": 0.11,
"learning_rate": 1.9969709949840034e-05,
"loss": 1.3445,
"step": 39
},
{
"epoch": 0.11,
"learning_rate": 1.9966043682058223e-05,
"loss": 1.3559,
"step": 40
},
{
"epoch": 0.12,
"learning_rate": 1.9962168401159685e-05,
"loss": 1.2862,
"step": 41
},
{
"epoch": 0.12,
"learning_rate": 1.995808418841885e-05,
"loss": 1.4271,
"step": 42
},
{
"epoch": 0.12,
"learning_rate": 1.9953791129491985e-05,
"loss": 1.2429,
"step": 43
},
{
"epoch": 0.12,
"learning_rate": 1.9949289314415373e-05,
"loss": 1.2567,
"step": 44
},
{
"epoch": 0.13,
"learning_rate": 1.994457883760346e-05,
"loss": 1.3447,
"step": 45
},
{
"epoch": 0.13,
"learning_rate": 1.993965979784684e-05,
"loss": 1.317,
"step": 46
},
{
"epoch": 0.13,
"learning_rate": 1.9934532298310206e-05,
"loss": 1.3366,
"step": 47
},
{
"epoch": 0.14,
"learning_rate": 1.9929196446530184e-05,
"loss": 1.299,
"step": 48
},
{
"epoch": 0.14,
"learning_rate": 1.992365235441306e-05,
"loss": 1.1435,
"step": 49
},
{
"epoch": 0.14,
"learning_rate": 1.991790013823246e-05,
"loss": 1.2788,
"step": 50
},
{
"epoch": 0.14,
"learning_rate": 1.991193991862689e-05,
"loss": 1.2544,
"step": 51
},
{
"epoch": 0.15,
"learning_rate": 1.9905771820597214e-05,
"loss": 1.2261,
"step": 52
},
{
"epoch": 0.15,
"learning_rate": 1.989939597350403e-05,
"loss": 1.1353,
"step": 53
},
{
"epoch": 0.15,
"learning_rate": 1.9892812511064962e-05,
"loss": 1.2016,
"step": 54
},
{
"epoch": 0.16,
"learning_rate": 1.9886021571351854e-05,
"loss": 1.11,
"step": 55
},
{
"epoch": 0.16,
"learning_rate": 1.9879023296787866e-05,
"loss": 1.2557,
"step": 56
},
{
"epoch": 0.16,
"learning_rate": 1.9871817834144506e-05,
"loss": 1.0785,
"step": 57
},
{
"epoch": 0.16,
"learning_rate": 1.9864405334538518e-05,
"loss": 1.0925,
"step": 58
},
{
"epoch": 0.17,
"learning_rate": 1.9856785953428757e-05,
"loss": 1.0723,
"step": 59
},
{
"epoch": 0.17,
"learning_rate": 1.9848959850612895e-05,
"loss": 1.0217,
"step": 60
},
{
"epoch": 0.17,
"learning_rate": 1.9840927190224083e-05,
"loss": 1.0608,
"step": 61
},
{
"epoch": 0.17,
"learning_rate": 1.9832688140727502e-05,
"loss": 1.16,
"step": 62
},
{
"epoch": 0.18,
"learning_rate": 1.982424287491684e-05,
"loss": 0.997,
"step": 63
},
{
"epoch": 0.18,
"learning_rate": 1.9815591569910654e-05,
"loss": 0.9963,
"step": 64
},
{
"epoch": 0.18,
"learning_rate": 1.9806734407148674e-05,
"loss": 0.9328,
"step": 65
},
{
"epoch": 0.19,
"learning_rate": 1.9797671572387985e-05,
"loss": 1.1344,
"step": 66
},
{
"epoch": 0.19,
"learning_rate": 1.978840325569912e-05,
"loss": 1.0168,
"step": 67
},
{
"epoch": 0.19,
"learning_rate": 1.977892965146211e-05,
"loss": 0.9744,
"step": 68
},
{
"epoch": 0.19,
"learning_rate": 1.976925095836236e-05,
"loss": 1.012,
"step": 69
},
{
"epoch": 0.2,
"learning_rate": 1.975936737938653e-05,
"loss": 0.9737,
"step": 70
},
{
"epoch": 0.2,
"learning_rate": 1.9749279121818235e-05,
"loss": 1.0498,
"step": 71
},
{
"epoch": 0.2,
"learning_rate": 1.9738986397233736e-05,
"loss": 0.9622,
"step": 72
},
{
"epoch": 0.21,
"learning_rate": 1.9728489421497465e-05,
"loss": 1.0339,
"step": 73
},
{
"epoch": 0.21,
"learning_rate": 1.9717788414757523e-05,
"loss": 1.0361,
"step": 74
},
{
"epoch": 0.21,
"learning_rate": 1.9706883601441066e-05,
"loss": 1.0385,
"step": 75
},
{
"epoch": 0.21,
"learning_rate": 1.969577521024958e-05,
"loss": 1.0243,
"step": 76
},
{
"epoch": 0.22,
"learning_rate": 1.9684463474154095e-05,
"loss": 0.9741,
"step": 77
},
{
"epoch": 0.22,
"learning_rate": 1.9672948630390296e-05,
"loss": 0.9044,
"step": 78
},
{
"epoch": 0.22,
"learning_rate": 1.9661230920453553e-05,
"loss": 0.9632,
"step": 79
},
{
"epoch": 0.23,
"learning_rate": 1.964931059009385e-05,
"loss": 0.9717,
"step": 80
},
{
"epoch": 0.23,
"learning_rate": 1.9637187889310632e-05,
"loss": 1.0986,
"step": 81
},
{
"epoch": 0.23,
"learning_rate": 1.9624863072347565e-05,
"loss": 0.9973,
"step": 82
},
{
"epoch": 0.23,
"learning_rate": 1.96123363976872e-05,
"loss": 0.9974,
"step": 83
},
{
"epoch": 0.24,
"learning_rate": 1.9599608128045554e-05,
"loss": 0.9846,
"step": 84
},
{
"epoch": 0.24,
"learning_rate": 1.9586678530366607e-05,
"loss": 0.9824,
"step": 85
},
{
"epoch": 0.24,
"learning_rate": 1.9573547875816685e-05,
"loss": 0.9864,
"step": 86
},
{
"epoch": 0.25,
"learning_rate": 1.9560216439778795e-05,
"loss": 1.0051,
"step": 87
},
{
"epoch": 0.25,
"learning_rate": 1.954668450184683e-05,
"loss": 1.0014,
"step": 88
},
{
"epoch": 0.25,
"learning_rate": 1.9532952345819723e-05,
"loss": 0.9815,
"step": 89
},
{
"epoch": 0.25,
"learning_rate": 1.951902025969548e-05,
"loss": 0.9953,
"step": 90
},
{
"epoch": 0.26,
"learning_rate": 1.950488853566515e-05,
"loss": 0.9655,
"step": 91
},
{
"epoch": 0.26,
"learning_rate": 1.949055747010669e-05,
"loss": 1.0011,
"step": 92
},
{
"epoch": 0.26,
"learning_rate": 1.9476027363578754e-05,
"loss": 0.9795,
"step": 93
},
{
"epoch": 0.27,
"learning_rate": 1.946129852081439e-05,
"loss": 0.9876,
"step": 94
},
{
"epoch": 0.27,
"learning_rate": 1.9446371250714645e-05,
"loss": 0.8756,
"step": 95
},
{
"epoch": 0.27,
"learning_rate": 1.943124586634209e-05,
"loss": 0.9328,
"step": 96
},
{
"epoch": 0.27,
"learning_rate": 1.941592268491425e-05,
"loss": 0.9823,
"step": 97
},
{
"epoch": 0.28,
"learning_rate": 1.9400402027796955e-05,
"loss": 0.9183,
"step": 98
},
{
"epoch": 0.28,
"learning_rate": 1.9384684220497605e-05,
"loss": 0.907,
"step": 99
},
{
"epoch": 0.28,
"learning_rate": 1.936876959265833e-05,
"loss": 0.8817,
"step": 100
},
{
"epoch": 0.28,
"learning_rate": 1.9352658478049085e-05,
"loss": 0.9246,
"step": 101
},
{
"epoch": 0.29,
"learning_rate": 1.9336351214560648e-05,
"loss": 0.9182,
"step": 102
},
{
"epoch": 0.29,
"learning_rate": 1.9319848144197543e-05,
"loss": 0.8392,
"step": 103
},
{
"epoch": 0.29,
"learning_rate": 1.9303149613070852e-05,
"loss": 0.938,
"step": 104
},
{
"epoch": 0.3,
"learning_rate": 1.928625597139096e-05,
"loss": 0.8843,
"step": 105
},
{
"epoch": 0.3,
"learning_rate": 1.926916757346022e-05,
"loss": 0.9203,
"step": 106
},
{
"epoch": 0.3,
"learning_rate": 1.9251884777665513e-05,
"loss": 0.8708,
"step": 107
},
{
"epoch": 0.3,
"learning_rate": 1.9234407946470735e-05,
"loss": 0.9161,
"step": 108
},
{
"epoch": 0.31,
"learning_rate": 1.9216737446409192e-05,
"loss": 0.8743,
"step": 109
},
{
"epoch": 0.31,
"learning_rate": 1.919887364807592e-05,
"loss": 0.9255,
"step": 110
},
{
"epoch": 0.31,
"learning_rate": 1.9180816926119903e-05,
"loss": 0.8138,
"step": 111
},
{
"epoch": 0.32,
"learning_rate": 1.9162567659236227e-05,
"loss": 0.9074,
"step": 112
},
{
"epoch": 0.32,
"learning_rate": 1.9144126230158127e-05,
"loss": 0.969,
"step": 113
},
{
"epoch": 0.32,
"learning_rate": 1.9125493025648963e-05,
"loss": 0.8313,
"step": 114
},
{
"epoch": 0.32,
"learning_rate": 1.9106668436494113e-05,
"loss": 0.8834,
"step": 115
},
{
"epoch": 0.33,
"learning_rate": 1.908765285749278e-05,
"loss": 0.9603,
"step": 116
},
{
"epoch": 0.33,
"learning_rate": 1.9068446687449698e-05,
"loss": 0.8353,
"step": 117
},
{
"epoch": 0.33,
"learning_rate": 1.9049050329166778e-05,
"loss": 0.8594,
"step": 118
},
{
"epoch": 0.34,
"learning_rate": 1.9029464189434663e-05,
"loss": 0.8876,
"step": 119
},
{
"epoch": 0.34,
"learning_rate": 1.900968867902419e-05,
"loss": 0.9399,
"step": 120
},
{
"epoch": 0.34,
"learning_rate": 1.8989724212677784e-05,
"loss": 0.7936,
"step": 121
},
{
"epoch": 0.34,
"learning_rate": 1.8969571209100738e-05,
"loss": 0.8721,
"step": 122
},
{
"epoch": 0.35,
"learning_rate": 1.8949230090952463e-05,
"loss": 0.7673,
"step": 123
},
{
"epoch": 0.35,
"learning_rate": 1.8928701284837597e-05,
"loss": 0.9108,
"step": 124
},
{
"epoch": 0.35,
"learning_rate": 1.890798522129708e-05,
"loss": 0.7324,
"step": 125
},
{
"epoch": 0.36,
"learning_rate": 1.8887082334799098e-05,
"loss": 0.8807,
"step": 126
},
{
"epoch": 0.36,
"learning_rate": 1.8865993063730003e-05,
"loss": 0.8082,
"step": 127
},
{
"epoch": 0.36,
"learning_rate": 1.884471785038509e-05,
"loss": 0.79,
"step": 128
},
{
"epoch": 0.36,
"learning_rate": 1.882325714095934e-05,
"loss": 0.8773,
"step": 129
},
{
"epoch": 0.37,
"learning_rate": 1.8801611385538047e-05,
"loss": 0.9139,
"step": 130
},
{
"epoch": 0.37,
"learning_rate": 1.8779781038087406e-05,
"loss": 0.8234,
"step": 131
},
{
"epoch": 0.37,
"learning_rate": 1.875776655644495e-05,
"loss": 0.8924,
"step": 132
},
{
"epoch": 0.38,
"learning_rate": 1.8735568402309987e-05,
"loss": 0.8481,
"step": 133
},
{
"epoch": 0.38,
"learning_rate": 1.8713187041233896e-05,
"loss": 0.869,
"step": 134
},
{
"epoch": 0.38,
"learning_rate": 1.8690622942610367e-05,
"loss": 0.7801,
"step": 135
},
{
"epoch": 0.38,
"learning_rate": 1.8667876579665556e-05,
"loss": 0.8175,
"step": 136
},
{
"epoch": 0.39,
"learning_rate": 1.8644948429448174e-05,
"loss": 0.8026,
"step": 137
},
{
"epoch": 0.39,
"learning_rate": 1.862183897281946e-05,
"loss": 0.922,
"step": 138
},
{
"epoch": 0.39,
"learning_rate": 1.8598548694443102e-05,
"loss": 0.9148,
"step": 139
},
{
"epoch": 0.39,
"learning_rate": 1.8575078082775096e-05,
"loss": 0.8427,
"step": 140
},
{
"epoch": 0.4,
"learning_rate": 1.8551427630053464e-05,
"loss": 0.8635,
"step": 141
},
{
"epoch": 0.4,
"learning_rate": 1.8527597832287954e-05,
"loss": 0.8064,
"step": 142
},
{
"epoch": 0.4,
"learning_rate": 1.8503589189249637e-05,
"loss": 0.9025,
"step": 143
},
{
"epoch": 0.41,
"learning_rate": 1.847940220446042e-05,
"loss": 0.8772,
"step": 144
},
{
"epoch": 0.41,
"learning_rate": 1.845503738518249e-05,
"loss": 0.8465,
"step": 145
},
{
"epoch": 0.41,
"learning_rate": 1.843049524240766e-05,
"loss": 0.8204,
"step": 146
},
{
"epoch": 0.41,
"learning_rate": 1.8405776290846672e-05,
"loss": 0.7709,
"step": 147
},
{
"epoch": 0.42,
"learning_rate": 1.8380881048918406e-05,
"loss": 0.779,
"step": 148
},
{
"epoch": 0.42,
"learning_rate": 1.8355810038738986e-05,
"loss": 0.7794,
"step": 149
},
{
"epoch": 0.42,
"learning_rate": 1.8330563786110837e-05,
"loss": 0.8651,
"step": 150
},
{
"epoch": 0.43,
"learning_rate": 1.830514282051166e-05,
"loss": 0.8513,
"step": 151
},
{
"epoch": 0.43,
"learning_rate": 1.8279547675083343e-05,
"loss": 0.8396,
"step": 152
},
{
"epoch": 0.43,
"learning_rate": 1.8253778886620754e-05,
"loss": 0.8845,
"step": 153
},
{
"epoch": 0.43,
"learning_rate": 1.822783699556049e-05,
"loss": 0.8509,
"step": 154
},
{
"epoch": 0.44,
"learning_rate": 1.820172254596956e-05,
"loss": 0.8736,
"step": 155
},
{
"epoch": 0.44,
"learning_rate": 1.817543608553395e-05,
"loss": 0.8067,
"step": 156
},
{
"epoch": 0.44,
"learning_rate": 1.814897816554715e-05,
"loss": 0.9002,
"step": 157
},
{
"epoch": 0.45,
"learning_rate": 1.8122349340898596e-05,
"loss": 0.8336,
"step": 158
},
{
"epoch": 0.45,
"learning_rate": 1.809555017006202e-05,
"loss": 0.8741,
"step": 159
},
{
"epoch": 0.45,
"learning_rate": 1.8068581215083752e-05,
"loss": 0.8059,
"step": 160
},
{
"epoch": 0.45,
"learning_rate": 1.804144304157091e-05,
"loss": 0.8228,
"step": 161
},
{
"epoch": 0.46,
"learning_rate": 1.8014136218679566e-05,
"loss": 0.8685,
"step": 162
},
{
"epoch": 0.46,
"learning_rate": 1.7986661319102795e-05,
"loss": 0.8701,
"step": 163
},
{
"epoch": 0.46,
"learning_rate": 1.7959018919058654e-05,
"loss": 0.8116,
"step": 164
},
{
"epoch": 0.47,
"learning_rate": 1.7931209598278117e-05,
"loss": 0.7616,
"step": 165
},
{
"epoch": 0.47,
"learning_rate": 1.7903233939992904e-05,
"loss": 0.7549,
"step": 166
},
{
"epoch": 0.47,
"learning_rate": 1.787509253092326e-05,
"loss": 0.8873,
"step": 167
},
{
"epoch": 0.47,
"learning_rate": 1.784678596126563e-05,
"loss": 0.8587,
"step": 168
},
{
"epoch": 0.48,
"learning_rate": 1.78183148246803e-05,
"loss": 0.8517,
"step": 169
},
{
"epoch": 0.48,
"learning_rate": 1.7789679718278944e-05,
"loss": 0.7772,
"step": 170
},
{
"epoch": 0.48,
"learning_rate": 1.7760881242612096e-05,
"loss": 0.8795,
"step": 171
},
{
"epoch": 0.49,
"learning_rate": 1.773192000165655e-05,
"loss": 0.8393,
"step": 172
},
{
"epoch": 0.49,
"learning_rate": 1.7702796602802705e-05,
"loss": 0.773,
"step": 173
},
{
"epoch": 0.49,
"learning_rate": 1.7673511656841822e-05,
"loss": 0.8416,
"step": 174
},
{
"epoch": 0.49,
"learning_rate": 1.7644065777953206e-05,
"loss": 0.8293,
"step": 175
},
{
"epoch": 0.5,
"learning_rate": 1.7614459583691346e-05,
"loss": 0.8903,
"step": 176
},
{
"epoch": 0.5,
"learning_rate": 1.758469369497293e-05,
"loss": 0.8235,
"step": 177
},
{
"epoch": 0.5,
"learning_rate": 1.7554768736063858e-05,
"loss": 0.8434,
"step": 178
},
{
"epoch": 0.5,
"learning_rate": 1.7524685334566126e-05,
"loss": 0.8067,
"step": 179
},
{
"epoch": 0.51,
"learning_rate": 1.7494444121404673e-05,
"loss": 0.8825,
"step": 180
},
{
"epoch": 0.51,
"learning_rate": 1.746404573081415e-05,
"loss": 0.7325,
"step": 181
},
{
"epoch": 0.51,
"learning_rate": 1.7433490800325614e-05,
"loss": 0.7858,
"step": 182
},
{
"epoch": 0.52,
"learning_rate": 1.7402779970753156e-05,
"loss": 0.8781,
"step": 183
},
{
"epoch": 0.52,
"learning_rate": 1.7371913886180473e-05,
"loss": 0.9138,
"step": 184
},
{
"epoch": 0.52,
"learning_rate": 1.7340893193947342e-05,
"loss": 0.7955,
"step": 185
},
{
"epoch": 0.52,
"learning_rate": 1.7309718544636057e-05,
"loss": 0.7933,
"step": 186
},
{
"epoch": 0.53,
"learning_rate": 1.7278390592057785e-05,
"loss": 0.9003,
"step": 187
},
{
"epoch": 0.53,
"learning_rate": 1.7246909993238844e-05,
"loss": 0.8204,
"step": 188
},
{
"epoch": 0.53,
"learning_rate": 1.7215277408406932e-05,
"loss": 0.7641,
"step": 189
},
{
"epoch": 0.54,
"learning_rate": 1.7183493500977277e-05,
"loss": 0.7507,
"step": 190
},
{
"epoch": 0.54,
"learning_rate": 1.7151558937538725e-05,
"loss": 0.7386,
"step": 191
},
{
"epoch": 0.54,
"learning_rate": 1.7119474387839764e-05,
"loss": 0.7715,
"step": 192
},
{
"epoch": 0.54,
"learning_rate": 1.708724052477446e-05,
"loss": 0.8169,
"step": 193
},
{
"epoch": 0.55,
"learning_rate": 1.7054858024368365e-05,
"loss": 0.7494,
"step": 194
},
{
"epoch": 0.55,
"learning_rate": 1.7022327565764336e-05,
"loss": 0.7876,
"step": 195
},
{
"epoch": 0.55,
"learning_rate": 1.6989649831208286e-05,
"loss": 0.8173,
"step": 196
},
{
"epoch": 0.56,
"learning_rate": 1.6956825506034866e-05,
"loss": 0.8212,
"step": 197
},
{
"epoch": 0.56,
"learning_rate": 1.6923855278653114e-05,
"loss": 0.7789,
"step": 198
},
{
"epoch": 0.56,
"learning_rate": 1.6890739840532004e-05,
"loss": 0.7801,
"step": 199
},
{
"epoch": 0.56,
"learning_rate": 1.6857479886185942e-05,
"loss": 0.8195,
"step": 200
},
{
"epoch": 0.57,
"learning_rate": 1.682407611316021e-05,
"loss": 0.8191,
"step": 201
},
{
"epoch": 0.57,
"learning_rate": 1.6790529222016328e-05,
"loss": 0.7776,
"step": 202
},
{
"epoch": 0.57,
"learning_rate": 1.6756839916317358e-05,
"loss": 0.7714,
"step": 203
},
{
"epoch": 0.58,
"learning_rate": 1.672300890261317e-05,
"loss": 0.7503,
"step": 204
},
{
"epoch": 0.58,
"learning_rate": 1.6689036890425596e-05,
"loss": 0.7747,
"step": 205
},
{
"epoch": 0.58,
"learning_rate": 1.665492459223357e-05,
"loss": 0.7853,
"step": 206
},
{
"epoch": 0.58,
"learning_rate": 1.6620672723458167e-05,
"loss": 0.8086,
"step": 207
},
{
"epoch": 0.59,
"learning_rate": 1.658628200244763e-05,
"loss": 0.8258,
"step": 208
},
{
"epoch": 0.59,
"learning_rate": 1.6551753150462258e-05,
"loss": 0.8176,
"step": 209
},
{
"epoch": 0.59,
"learning_rate": 1.6517086891659335e-05,
"loss": 0.7502,
"step": 210
},
{
"epoch": 0.6,
"learning_rate": 1.6482283953077887e-05,
"loss": 0.8111,
"step": 211
},
{
"epoch": 0.6,
"learning_rate": 1.644734506462347e-05,
"loss": 0.7517,
"step": 212
},
{
"epoch": 0.6,
"learning_rate": 1.641227095905286e-05,
"loss": 0.7997,
"step": 213
},
{
"epoch": 0.6,
"learning_rate": 1.637706237195867e-05,
"loss": 0.7622,
"step": 214
},
{
"epoch": 0.61,
"learning_rate": 1.6341720041753924e-05,
"loss": 0.8949,
"step": 215
},
{
"epoch": 0.61,
"learning_rate": 1.6306244709656597e-05,
"loss": 0.7854,
"step": 216
},
{
"epoch": 0.61,
"learning_rate": 1.6270637119674023e-05,
"loss": 0.8078,
"step": 217
},
{
"epoch": 0.61,
"learning_rate": 1.6234898018587336e-05,
"loss": 0.761,
"step": 218
},
{
"epoch": 0.62,
"learning_rate": 1.6199028155935793e-05,
"loss": 0.8788,
"step": 219
},
{
"epoch": 0.62,
"learning_rate": 1.6163028284001034e-05,
"loss": 0.8467,
"step": 220
},
{
"epoch": 0.62,
"learning_rate": 1.612689915779134e-05,
"loss": 0.7494,
"step": 221
},
{
"epoch": 0.63,
"learning_rate": 1.6090641535025773e-05,
"loss": 0.7209,
"step": 222
},
{
"epoch": 0.63,
"learning_rate": 1.605425617611829e-05,
"loss": 0.8199,
"step": 223
},
{
"epoch": 0.63,
"learning_rate": 1.6017743844161802e-05,
"loss": 0.7596,
"step": 224
},
{
"epoch": 0.63,
"learning_rate": 1.598110530491216e-05,
"loss": 0.8124,
"step": 225
},
{
"epoch": 0.64,
"learning_rate": 1.5944341326772112e-05,
"loss": 0.7533,
"step": 226
},
{
"epoch": 0.64,
"learning_rate": 1.5907452680775164e-05,
"loss": 0.8116,
"step": 227
},
{
"epoch": 0.64,
"learning_rate": 1.587044014056943e-05,
"loss": 0.7898,
"step": 228
},
{
"epoch": 0.65,
"learning_rate": 1.583330448240139e-05,
"loss": 0.8012,
"step": 229
},
{
"epoch": 0.65,
"learning_rate": 1.5796046485099633e-05,
"loss": 0.7459,
"step": 230
},
{
"epoch": 0.65,
"learning_rate": 1.57586669300585e-05,
"loss": 0.7544,
"step": 231
},
{
"epoch": 0.65,
"learning_rate": 1.5721166601221697e-05,
"loss": 0.7788,
"step": 232
},
{
"epoch": 0.66,
"learning_rate": 1.5683546285065878e-05,
"loss": 0.7402,
"step": 233
},
{
"epoch": 0.66,
"learning_rate": 1.5645806770584122e-05,
"loss": 0.7381,
"step": 234
},
{
"epoch": 0.66,
"learning_rate": 1.5607948849269404e-05,
"loss": 0.8209,
"step": 235
},
{
"epoch": 0.67,
"learning_rate": 1.5569973315097985e-05,
"loss": 0.815,
"step": 236
},
{
"epoch": 0.67,
"learning_rate": 1.5531880964512773e-05,
"loss": 0.7335,
"step": 237
},
{
"epoch": 0.67,
"learning_rate": 1.54936725964066e-05,
"loss": 0.7828,
"step": 238
},
{
"epoch": 0.67,
"learning_rate": 1.5455349012105488e-05,
"loss": 0.8503,
"step": 239
},
{
"epoch": 0.68,
"learning_rate": 1.5416911015351827e-05,
"loss": 0.7722,
"step": 240
},
{
"epoch": 0.68,
"learning_rate": 1.5378359412287537e-05,
"loss": 0.7984,
"step": 241
},
{
"epoch": 0.68,
"learning_rate": 1.5339695011437128e-05,
"loss": 0.8432,
"step": 242
},
{
"epoch": 0.69,
"learning_rate": 1.530091862369078e-05,
"loss": 0.8004,
"step": 243
},
{
"epoch": 0.69,
"learning_rate": 1.526203106228733e-05,
"loss": 0.8113,
"step": 244
},
{
"epoch": 0.69,
"learning_rate": 1.5223033142797183e-05,
"loss": 0.7418,
"step": 245
},
{
"epoch": 0.69,
"learning_rate": 1.5183925683105254e-05,
"loss": 0.7867,
"step": 246
},
{
"epoch": 0.7,
"learning_rate": 1.5144709503393773e-05,
"loss": 0.774,
"step": 247
},
{
"epoch": 0.7,
"learning_rate": 1.5105385426125123e-05,
"loss": 0.7314,
"step": 248
},
{
"epoch": 0.7,
"learning_rate": 1.5065954276024561e-05,
"loss": 0.6955,
"step": 249
},
{
"epoch": 0.71,
"learning_rate": 1.5026416880062932e-05,
"loss": 0.8378,
"step": 250
},
{
"epoch": 0.71,
"learning_rate": 1.4986774067439327e-05,
"loss": 0.7434,
"step": 251
},
{
"epoch": 0.71,
"learning_rate": 1.4947026669563687e-05,
"loss": 0.8086,
"step": 252
},
{
"epoch": 0.71,
"learning_rate": 1.4907175520039381e-05,
"loss": 0.7269,
"step": 253
},
{
"epoch": 0.72,
"learning_rate": 1.4867221454645696e-05,
"loss": 0.762,
"step": 254
},
{
"epoch": 0.72,
"learning_rate": 1.482716531132034e-05,
"loss": 0.7613,
"step": 255
},
{
"epoch": 0.72,
"learning_rate": 1.4787007930141841e-05,
"loss": 0.7685,
"step": 256
},
{
"epoch": 0.72,
"learning_rate": 1.4746750153311951e-05,
"loss": 0.7625,
"step": 257
},
{
"epoch": 0.73,
"learning_rate": 1.4706392825137962e-05,
"loss": 0.7095,
"step": 258
},
{
"epoch": 0.73,
"learning_rate": 1.4665936792015021e-05,
"loss": 0.7264,
"step": 259
},
{
"epoch": 0.73,
"learning_rate": 1.4625382902408356e-05,
"loss": 0.7563,
"step": 260
},
{
"epoch": 0.74,
"learning_rate": 1.4584732006835495e-05,
"loss": 0.8759,
"step": 261
},
{
"epoch": 0.74,
"learning_rate": 1.4543984957848438e-05,
"loss": 0.7502,
"step": 262
},
{
"epoch": 0.74,
"learning_rate": 1.4503142610015751e-05,
"loss": 0.7669,
"step": 263
},
{
"epoch": 0.74,
"learning_rate": 1.4462205819904658e-05,
"loss": 0.7482,
"step": 264
},
{
"epoch": 0.75,
"learning_rate": 1.4421175446063086e-05,
"loss": 0.7655,
"step": 265
},
{
"epoch": 0.75,
"learning_rate": 1.4380052349001647e-05,
"loss": 0.7665,
"step": 266
},
{
"epoch": 0.75,
"learning_rate": 1.4338837391175582e-05,
"loss": 0.7288,
"step": 267
},
{
"epoch": 0.76,
"learning_rate": 1.42975314369667e-05,
"loss": 0.7157,
"step": 268
},
{
"epoch": 0.76,
"learning_rate": 1.4256135352665217e-05,
"loss": 0.7321,
"step": 269
},
{
"epoch": 0.76,
"learning_rate": 1.4214650006451622e-05,
"loss": 0.7788,
"step": 270
},
{
"epoch": 0.76,
"learning_rate": 1.4173076268378443e-05,
"loss": 0.7853,
"step": 271
},
{
"epoch": 0.77,
"learning_rate": 1.4131415010352007e-05,
"loss": 0.8383,
"step": 272
},
{
"epoch": 0.77,
"learning_rate": 1.408966710611416e-05,
"loss": 0.7946,
"step": 273
},
{
"epoch": 0.77,
"learning_rate": 1.4047833431223938e-05,
"loss": 0.7695,
"step": 274
},
{
"epoch": 0.78,
"learning_rate": 1.4005914863039203e-05,
"loss": 0.7197,
"step": 275
},
{
"epoch": 0.78,
"learning_rate": 1.3963912280698238e-05,
"loss": 0.705,
"step": 276
},
{
"epoch": 0.78,
"learning_rate": 1.3921826565101325e-05,
"loss": 0.744,
"step": 277
},
{
"epoch": 0.78,
"learning_rate": 1.3879658598892254e-05,
"loss": 0.7517,
"step": 278
},
{
"epoch": 0.79,
"learning_rate": 1.3837409266439818e-05,
"loss": 0.7607,
"step": 279
},
{
"epoch": 0.79,
"learning_rate": 1.3795079453819276e-05,
"loss": 0.716,
"step": 280
},
{
"epoch": 0.79,
"learning_rate": 1.3752670048793744e-05,
"loss": 0.7838,
"step": 281
},
{
"epoch": 0.8,
"learning_rate": 1.37101819407956e-05,
"loss": 0.7437,
"step": 282
},
{
"epoch": 0.8,
"learning_rate": 1.366761602090782e-05,
"loss": 0.7418,
"step": 283
},
{
"epoch": 0.8,
"learning_rate": 1.3624973181845302e-05,
"loss": 0.7798,
"step": 284
},
{
"epoch": 0.8,
"learning_rate": 1.3582254317936117e-05,
"loss": 0.7944,
"step": 285
},
{
"epoch": 0.81,
"learning_rate": 1.3539460325102779e-05,
"loss": 0.7738,
"step": 286
},
{
"epoch": 0.81,
"learning_rate": 1.349659210084344e-05,
"loss": 0.7838,
"step": 287
},
{
"epoch": 0.81,
"learning_rate": 1.3453650544213078e-05,
"loss": 0.7832,
"step": 288
},
{
"epoch": 0.82,
"learning_rate": 1.3410636555804634e-05,
"loss": 0.778,
"step": 289
},
{
"epoch": 0.82,
"learning_rate": 1.3367551037730129e-05,
"loss": 0.7574,
"step": 290
},
{
"epoch": 0.82,
"learning_rate": 1.3324394893601734e-05,
"loss": 0.7401,
"step": 291
},
{
"epoch": 0.82,
"learning_rate": 1.3281169028512838e-05,
"loss": 0.7677,
"step": 292
},
{
"epoch": 0.83,
"learning_rate": 1.3237874349019041e-05,
"loss": 0.8209,
"step": 293
},
{
"epoch": 0.83,
"learning_rate": 1.319451176311917e-05,
"loss": 0.7901,
"step": 294
},
{
"epoch": 0.83,
"learning_rate": 1.315108218023621e-05,
"loss": 0.7377,
"step": 295
},
{
"epoch": 0.83,
"learning_rate": 1.3107586511198243e-05,
"loss": 0.7986,
"step": 296
},
{
"epoch": 0.84,
"learning_rate": 1.306402566821935e-05,
"loss": 0.7011,
"step": 297
},
{
"epoch": 0.84,
"learning_rate": 1.302040056488047e-05,
"loss": 0.7485,
"step": 298
},
{
"epoch": 0.84,
"learning_rate": 1.297671211611025e-05,
"loss": 0.8165,
"step": 299
},
{
"epoch": 0.85,
"learning_rate": 1.2932961238165837e-05,
"loss": 0.7378,
"step": 300
},
{
"epoch": 0.85,
"learning_rate": 1.2889148848613695e-05,
"loss": 0.7919,
"step": 301
},
{
"epoch": 0.85,
"learning_rate": 1.2845275866310325e-05,
"loss": 0.7492,
"step": 302
},
{
"epoch": 0.85,
"learning_rate": 1.2801343211383021e-05,
"loss": 0.798,
"step": 303
},
{
"epoch": 0.86,
"learning_rate": 1.2757351805210557e-05,
"loss": 0.7384,
"step": 304
},
{
"epoch": 0.86,
"learning_rate": 1.2713302570403872e-05,
"loss": 0.6986,
"step": 305
},
{
"epoch": 0.86,
"learning_rate": 1.2669196430786715e-05,
"loss": 0.7294,
"step": 306
},
{
"epoch": 0.87,
"learning_rate": 1.2625034311376276e-05,
"loss": 0.7249,
"step": 307
},
{
"epoch": 0.87,
"learning_rate": 1.258081713836378e-05,
"loss": 0.7588,
"step": 308
},
{
"epoch": 0.87,
"learning_rate": 1.2536545839095074e-05,
"loss": 0.7838,
"step": 309
},
{
"epoch": 0.87,
"learning_rate": 1.2492221342051153e-05,
"loss": 0.6854,
"step": 310
},
{
"epoch": 0.88,
"learning_rate": 1.2447844576828719e-05,
"loss": 0.6708,
"step": 311
},
{
"epoch": 0.88,
"learning_rate": 1.2403416474120657e-05,
"loss": 0.7611,
"step": 312
},
{
"epoch": 0.88,
"learning_rate": 1.2358937965696538e-05,
"loss": 0.7592,
"step": 313
},
{
"epoch": 0.89,
"learning_rate": 1.2314409984383066e-05,
"loss": 0.8339,
"step": 314
},
{
"epoch": 0.89,
"learning_rate": 1.2269833464044514e-05,
"loss": 0.7236,
"step": 315
},
{
"epoch": 0.89,
"learning_rate": 1.2225209339563144e-05,
"loss": 0.7824,
"step": 316
},
{
"epoch": 0.89,
"learning_rate": 1.2180538546819595e-05,
"loss": 0.7166,
"step": 317
},
{
"epoch": 0.9,
"learning_rate": 1.2135822022673263e-05,
"loss": 0.7074,
"step": 318
},
{
"epoch": 0.9,
"learning_rate": 1.2091060704942636e-05,
"loss": 0.7084,
"step": 319
},
{
"epoch": 0.9,
"learning_rate": 1.204625553238565e-05,
"loss": 0.7415,
"step": 320
},
{
"epoch": 0.91,
"learning_rate": 1.200140744467997e-05,
"loss": 0.7241,
"step": 321
},
{
"epoch": 0.91,
"learning_rate": 1.195651738240332e-05,
"loss": 0.7303,
"step": 322
},
{
"epoch": 0.91,
"learning_rate": 1.1911586287013726e-05,
"loss": 0.7516,
"step": 323
},
{
"epoch": 0.91,
"learning_rate": 1.1866615100829777e-05,
"loss": 0.6678,
"step": 324
},
{
"epoch": 0.92,
"learning_rate": 1.1821604767010883e-05,
"loss": 0.7103,
"step": 325
},
{
"epoch": 0.92,
"learning_rate": 1.1776556229537461e-05,
"loss": 0.723,
"step": 326
},
{
"epoch": 0.92,
"learning_rate": 1.1731470433191173e-05,
"loss": 0.7216,
"step": 327
},
{
"epoch": 0.93,
"learning_rate": 1.1686348323535078e-05,
"loss": 0.705,
"step": 328
},
{
"epoch": 0.93,
"learning_rate": 1.1641190846893824e-05,
"loss": 0.7574,
"step": 329
},
{
"epoch": 0.93,
"learning_rate": 1.1595998950333794e-05,
"loss": 0.7451,
"step": 330
},
{
"epoch": 0.93,
"learning_rate": 1.1550773581643245e-05,
"loss": 0.7085,
"step": 331
},
{
"epoch": 0.94,
"learning_rate": 1.1505515689312424e-05,
"loss": 0.7748,
"step": 332
},
{
"epoch": 0.94,
"learning_rate": 1.1460226222513682e-05,
"loss": 0.7865,
"step": 333
},
{
"epoch": 0.94,
"learning_rate": 1.1414906131081575e-05,
"loss": 0.7569,
"step": 334
},
{
"epoch": 0.94,
"learning_rate": 1.1369556365492924e-05,
"loss": 0.722,
"step": 335
},
{
"epoch": 0.95,
"learning_rate": 1.1324177876846897e-05,
"loss": 0.8369,
"step": 336
},
{
"epoch": 0.95,
"learning_rate": 1.1278771616845061e-05,
"loss": 0.7072,
"step": 337
},
{
"epoch": 0.95,
"learning_rate": 1.1233338537771408e-05,
"loss": 0.7578,
"step": 338
},
{
"epoch": 0.96,
"learning_rate": 1.1187879592472402e-05,
"loss": 0.6986,
"step": 339
},
{
"epoch": 0.96,
"learning_rate": 1.1142395734336986e-05,
"loss": 0.7433,
"step": 340
},
{
"epoch": 0.96,
"learning_rate": 1.1096887917276585e-05,
"loss": 0.766,
"step": 341
},
{
"epoch": 0.96,
"learning_rate": 1.1051357095705102e-05,
"loss": 0.6994,
"step": 342
},
{
"epoch": 0.97,
"learning_rate": 1.1005804224518912e-05,
"loss": 0.7189,
"step": 343
},
{
"epoch": 0.97,
"learning_rate": 1.0960230259076819e-05,
"loss": 0.7584,
"step": 344
},
{
"epoch": 0.97,
"learning_rate": 1.0914636155180025e-05,
"loss": 0.642,
"step": 345
},
{
"epoch": 0.98,
"learning_rate": 1.0869022869052091e-05,
"loss": 0.7022,
"step": 346
},
{
"epoch": 0.98,
"learning_rate": 1.0823391357318876e-05,
"loss": 0.6718,
"step": 347
},
{
"epoch": 0.98,
"learning_rate": 1.0777742576988474e-05,
"loss": 0.705,
"step": 348
},
{
"epoch": 0.98,
"learning_rate": 1.0732077485431152e-05,
"loss": 0.8261,
"step": 349
},
{
"epoch": 0.99,
"learning_rate": 1.0686397040359253e-05,
"loss": 0.8122,
"step": 350
},
{
"epoch": 0.99,
"learning_rate": 1.064070219980713e-05,
"loss": 0.7498,
"step": 351
},
{
"epoch": 0.99,
"learning_rate": 1.059499392211105e-05,
"loss": 0.6914,
"step": 352
},
{
"epoch": 1.0,
"learning_rate": 1.0549273165889079e-05,
"loss": 0.7027,
"step": 353
},
{
"epoch": 1.0,
"learning_rate": 1.0503540890020997e-05,
"loss": 0.7156,
"step": 354
},
{
"epoch": 1.0,
"learning_rate": 1.0457798053628181e-05,
"loss": 0.6018,
"step": 355
},
{
"epoch": 1.0,
"learning_rate": 1.0412045616053486e-05,
"loss": 0.5589,
"step": 356
},
{
"epoch": 1.01,
"learning_rate": 1.0366284536841124e-05,
"loss": 0.5175,
"step": 357
},
{
"epoch": 1.01,
"learning_rate": 1.0320515775716556e-05,
"loss": 0.5046,
"step": 358
},
{
"epoch": 1.01,
"learning_rate": 1.0274740292566335e-05,
"loss": 0.4882,
"step": 359
},
{
"epoch": 1.02,
"learning_rate": 1.0228959047418005e-05,
"loss": 0.5477,
"step": 360
},
{
"epoch": 1.02,
"learning_rate": 1.0183173000419954e-05,
"loss": 0.5457,
"step": 361
},
{
"epoch": 1.02,
"learning_rate": 1.0137383111821267e-05,
"loss": 0.4684,
"step": 362
},
{
"epoch": 1.02,
"learning_rate": 1.009159034195161e-05,
"loss": 0.4375,
"step": 363
},
{
"epoch": 1.03,
"learning_rate": 1.0045795651201062e-05,
"loss": 0.4825,
"step": 364
},
{
"epoch": 1.03,
"learning_rate": 1e-05,
"loss": 0.5051,
"step": 365
},
{
"epoch": 1.03,
"learning_rate": 9.954204348798938e-06,
"loss": 0.5004,
"step": 366
},
{
"epoch": 1.04,
"learning_rate": 9.908409658048395e-06,
"loss": 0.4938,
"step": 367
},
{
"epoch": 1.04,
"learning_rate": 9.862616888178733e-06,
"loss": 0.4588,
"step": 368
},
{
"epoch": 1.04,
"learning_rate": 9.816826999580049e-06,
"loss": 0.4837,
"step": 369
},
{
"epoch": 1.04,
"learning_rate": 9.771040952581998e-06,
"loss": 0.4688,
"step": 370
},
{
"epoch": 1.05,
"learning_rate": 9.72525970743367e-06,
"loss": 0.469,
"step": 371
},
{
"epoch": 1.05,
"learning_rate": 9.67948422428345e-06,
"loss": 0.4561,
"step": 372
},
{
"epoch": 1.05,
"learning_rate": 9.633715463158881e-06,
"loss": 0.4601,
"step": 373
},
{
"epoch": 1.06,
"learning_rate": 9.587954383946518e-06,
"loss": 0.5039,
"step": 374
},
{
"epoch": 1.06,
"learning_rate": 9.542201946371819e-06,
"loss": 0.5124,
"step": 375
},
{
"epoch": 1.06,
"learning_rate": 9.496459109979004e-06,
"loss": 0.4842,
"step": 376
},
{
"epoch": 1.06,
"learning_rate": 9.450726834110923e-06,
"loss": 0.4339,
"step": 377
},
{
"epoch": 1.07,
"learning_rate": 9.405006077888954e-06,
"loss": 0.4836,
"step": 378
},
{
"epoch": 1.07,
"learning_rate": 9.359297800192873e-06,
"loss": 0.416,
"step": 379
},
{
"epoch": 1.07,
"learning_rate": 9.313602959640754e-06,
"loss": 0.4832,
"step": 380
},
{
"epoch": 1.07,
"learning_rate": 9.267922514568853e-06,
"loss": 0.4296,
"step": 381
},
{
"epoch": 1.08,
"learning_rate": 9.22225742301153e-06,
"loss": 0.4555,
"step": 382
},
{
"epoch": 1.08,
"learning_rate": 9.176608642681127e-06,
"loss": 0.4406,
"step": 383
},
{
"epoch": 1.08,
"learning_rate": 9.13097713094791e-06,
"loss": 0.4758,
"step": 384
},
{
"epoch": 1.09,
"learning_rate": 9.085363844819979e-06,
"loss": 0.502,
"step": 385
},
{
"epoch": 1.09,
"learning_rate": 9.039769740923183e-06,
"loss": 0.4436,
"step": 386
},
{
"epoch": 1.09,
"learning_rate": 8.99419577548109e-06,
"loss": 0.5039,
"step": 387
},
{
"epoch": 1.09,
"learning_rate": 8.948642904294901e-06,
"loss": 0.4657,
"step": 388
},
{
"epoch": 1.1,
"learning_rate": 8.90311208272342e-06,
"loss": 0.4738,
"step": 389
},
{
"epoch": 1.1,
"learning_rate": 8.857604265663016e-06,
"loss": 0.528,
"step": 390
},
{
"epoch": 1.1,
"learning_rate": 8.812120407527603e-06,
"loss": 0.4791,
"step": 391
},
{
"epoch": 1.11,
"learning_rate": 8.766661462228593e-06,
"loss": 0.5127,
"step": 392
},
{
"epoch": 1.11,
"learning_rate": 8.721228383154939e-06,
"loss": 0.5009,
"step": 393
},
{
"epoch": 1.11,
"learning_rate": 8.675822123153104e-06,
"loss": 0.5341,
"step": 394
},
{
"epoch": 1.11,
"learning_rate": 8.630443634507077e-06,
"loss": 0.4785,
"step": 395
},
{
"epoch": 1.12,
"learning_rate": 8.585093868918426e-06,
"loss": 0.463,
"step": 396
},
{
"epoch": 1.12,
"learning_rate": 8.539773777486321e-06,
"loss": 0.478,
"step": 397
},
{
"epoch": 1.12,
"learning_rate": 8.494484310687581e-06,
"loss": 0.5036,
"step": 398
},
{
"epoch": 1.13,
"learning_rate": 8.44922641835676e-06,
"loss": 0.4644,
"step": 399
},
{
"epoch": 1.13,
"learning_rate": 8.404001049666211e-06,
"loss": 0.4694,
"step": 400
},
{
"epoch": 1.13,
"learning_rate": 8.35880915310618e-06,
"loss": 0.5166,
"step": 401
},
{
"epoch": 1.13,
"learning_rate": 8.313651676464924e-06,
"loss": 0.4615,
"step": 402
},
{
"epoch": 1.14,
"learning_rate": 8.26852956680883e-06,
"loss": 0.4367,
"step": 403
},
{
"epoch": 1.14,
"learning_rate": 8.223443770462539e-06,
"loss": 0.4521,
"step": 404
},
{
"epoch": 1.14,
"learning_rate": 8.17839523298912e-06,
"loss": 0.486,
"step": 405
},
{
"epoch": 1.15,
"learning_rate": 8.133384899170224e-06,
"loss": 0.4335,
"step": 406
},
{
"epoch": 1.15,
"learning_rate": 8.08841371298628e-06,
"loss": 0.5221,
"step": 407
},
{
"epoch": 1.15,
"learning_rate": 8.043482617596681e-06,
"loss": 0.4353,
"step": 408
},
{
"epoch": 1.15,
"learning_rate": 7.99859255532003e-06,
"loss": 0.4344,
"step": 409
},
{
"epoch": 1.16,
"learning_rate": 7.953744467614356e-06,
"loss": 0.4416,
"step": 410
},
{
"epoch": 1.16,
"learning_rate": 7.908939295057362e-06,
"loss": 0.4846,
"step": 411
},
{
"epoch": 1.16,
"learning_rate": 7.864177977326739e-06,
"loss": 0.4328,
"step": 412
},
{
"epoch": 1.17,
"learning_rate": 7.819461453180403e-06,
"loss": 0.4825,
"step": 413
},
{
"epoch": 1.17,
"learning_rate": 7.774790660436857e-06,
"loss": 0.4721,
"step": 414
},
{
"epoch": 1.17,
"learning_rate": 7.730166535955489e-06,
"loss": 0.5116,
"step": 415
},
{
"epoch": 1.17,
"learning_rate": 7.685590015616939e-06,
"loss": 0.4678,
"step": 416
},
{
"epoch": 1.18,
"learning_rate": 7.641062034303464e-06,
"loss": 0.4308,
"step": 417
},
{
"epoch": 1.18,
"learning_rate": 7.596583525879344e-06,
"loss": 0.5089,
"step": 418
},
{
"epoch": 1.18,
"learning_rate": 7.5521554231712845e-06,
"loss": 0.4329,
"step": 419
},
{
"epoch": 1.18,
"learning_rate": 7.507778657948847e-06,
"loss": 0.4541,
"step": 420
},
{
"epoch": 1.19,
"learning_rate": 7.463454160904928e-06,
"loss": 0.4429,
"step": 421
},
{
"epoch": 1.19,
"learning_rate": 7.419182861636218e-06,
"loss": 0.4642,
"step": 422
},
{
"epoch": 1.19,
"learning_rate": 7.374965688623726e-06,
"loss": 0.4591,
"step": 423
},
{
"epoch": 1.2,
"learning_rate": 7.3308035692132896e-06,
"loss": 0.4687,
"step": 424
},
{
"epoch": 1.2,
"learning_rate": 7.286697429596135e-06,
"loss": 0.5139,
"step": 425
},
{
"epoch": 1.2,
"learning_rate": 7.242648194789447e-06,
"loss": 0.4802,
"step": 426
},
{
"epoch": 1.2,
"learning_rate": 7.19865678861698e-06,
"loss": 0.4229,
"step": 427
},
{
"epoch": 1.21,
"learning_rate": 7.154724133689677e-06,
"loss": 0.4831,
"step": 428
},
{
"epoch": 1.21,
"learning_rate": 7.110851151386306e-06,
"loss": 0.4627,
"step": 429
},
{
"epoch": 1.21,
"learning_rate": 7.067038761834164e-06,
"loss": 0.4961,
"step": 430
},
{
"epoch": 1.22,
"learning_rate": 7.023287883889753e-06,
"loss": 0.5256,
"step": 431
},
{
"epoch": 1.22,
"learning_rate": 6.979599435119531e-06,
"loss": 0.5205,
"step": 432
},
{
"epoch": 1.22,
"learning_rate": 6.935974331780653e-06,
"loss": 0.4607,
"step": 433
},
{
"epoch": 1.22,
"learning_rate": 6.892413488801762e-06,
"loss": 0.4712,
"step": 434
},
{
"epoch": 1.23,
"learning_rate": 6.848917819763794e-06,
"loss": 0.4467,
"step": 435
},
{
"epoch": 1.23,
"learning_rate": 6.805488236880831e-06,
"loss": 0.5172,
"step": 436
},
{
"epoch": 1.23,
"learning_rate": 6.76212565098096e-06,
"loss": 0.5356,
"step": 437
},
{
"epoch": 1.24,
"learning_rate": 6.718830971487165e-06,
"loss": 0.5037,
"step": 438
},
{
"epoch": 1.24,
"learning_rate": 6.675605106398269e-06,
"loss": 0.4475,
"step": 439
},
{
"epoch": 1.24,
"learning_rate": 6.632448962269873e-06,
"loss": 0.4752,
"step": 440
},
{
"epoch": 1.24,
"learning_rate": 6.589363444195367e-06,
"loss": 0.4641,
"step": 441
},
{
"epoch": 1.25,
"learning_rate": 6.546349455786926e-06,
"loss": 0.4272,
"step": 442
},
{
"epoch": 1.25,
"learning_rate": 6.503407899156565e-06,
"loss": 0.4462,
"step": 443
},
{
"epoch": 1.25,
"learning_rate": 6.460539674897226e-06,
"loss": 0.5092,
"step": 444
},
{
"epoch": 1.26,
"learning_rate": 6.417745682063884e-06,
"loss": 0.4469,
"step": 445
},
{
"epoch": 1.26,
"learning_rate": 6.3750268181547e-06,
"loss": 0.454,
"step": 446
},
{
"epoch": 1.26,
"learning_rate": 6.3323839790921785e-06,
"loss": 0.485,
"step": 447
},
{
"epoch": 1.26,
"learning_rate": 6.289818059204404e-06,
"loss": 0.4293,
"step": 448
},
{
"epoch": 1.27,
"learning_rate": 6.24732995120626e-06,
"loss": 0.4679,
"step": 449
},
{
"epoch": 1.27,
"learning_rate": 6.204920546180728e-06,
"loss": 0.4647,
"step": 450
},
{
"epoch": 1.27,
"learning_rate": 6.162590733560183e-06,
"loss": 0.4634,
"step": 451
},
{
"epoch": 1.28,
"learning_rate": 6.120341401107751e-06,
"loss": 0.48,
"step": 452
},
{
"epoch": 1.28,
"learning_rate": 6.078173434898679e-06,
"loss": 0.4405,
"step": 453
},
{
"epoch": 1.28,
"learning_rate": 6.036087719301763e-06,
"loss": 0.4971,
"step": 454
},
{
"epoch": 1.28,
"learning_rate": 5.994085136960801e-06,
"loss": 0.4445,
"step": 455
},
{
"epoch": 1.29,
"learning_rate": 5.952166568776062e-06,
"loss": 0.4416,
"step": 456
},
{
"epoch": 1.29,
"learning_rate": 5.910332893885842e-06,
"loss": 0.4481,
"step": 457
},
{
"epoch": 1.29,
"learning_rate": 5.868584989647994e-06,
"loss": 0.4036,
"step": 458
},
{
"epoch": 1.29,
"learning_rate": 5.826923731621562e-06,
"loss": 0.5132,
"step": 459
},
{
"epoch": 1.3,
"learning_rate": 5.785349993548382e-06,
"loss": 0.4508,
"step": 460
},
{
"epoch": 1.3,
"learning_rate": 5.743864647334789e-06,
"loss": 0.4643,
"step": 461
},
{
"epoch": 1.3,
"learning_rate": 5.702468563033307e-06,
"loss": 0.4699,
"step": 462
},
{
"epoch": 1.31,
"learning_rate": 5.66116260882442e-06,
"loss": 0.4779,
"step": 463
},
{
"epoch": 1.31,
"learning_rate": 5.6199476509983546e-06,
"loss": 0.5538,
"step": 464
},
{
"epoch": 1.31,
"learning_rate": 5.5788245539369144e-06,
"loss": 0.4681,
"step": 465
},
{
"epoch": 1.31,
"learning_rate": 5.537794180095341e-06,
"loss": 0.4262,
"step": 466
},
{
"epoch": 1.32,
"learning_rate": 5.496857389984251e-06,
"loss": 0.4669,
"step": 467
},
{
"epoch": 1.32,
"learning_rate": 5.456015042151563e-06,
"loss": 0.4049,
"step": 468
},
{
"epoch": 1.32,
"learning_rate": 5.415267993164504e-06,
"loss": 0.4583,
"step": 469
},
{
"epoch": 1.33,
"learning_rate": 5.37461709759165e-06,
"loss": 0.4204,
"step": 470
},
{
"epoch": 1.33,
"learning_rate": 5.334063207984983e-06,
"loss": 0.4144,
"step": 471
},
{
"epoch": 1.33,
"learning_rate": 5.2936071748620386e-06,
"loss": 0.4394,
"step": 472
},
{
"epoch": 1.33,
"learning_rate": 5.253249846688053e-06,
"loss": 0.4853,
"step": 473
},
{
"epoch": 1.34,
"learning_rate": 5.21299206985816e-06,
"loss": 0.4396,
"step": 474
},
{
"epoch": 1.34,
"learning_rate": 5.172834688679665e-06,
"loss": 0.4797,
"step": 475
},
{
"epoch": 1.34,
"learning_rate": 5.132778545354305e-06,
"loss": 0.4773,
"step": 476
},
{
"epoch": 1.35,
"learning_rate": 5.092824479960625e-06,
"loss": 0.4502,
"step": 477
},
{
"epoch": 1.35,
"learning_rate": 5.0529733304363145e-06,
"loss": 0.452,
"step": 478
},
{
"epoch": 1.35,
"learning_rate": 5.013225932560679e-06,
"loss": 0.436,
"step": 479
},
{
"epoch": 1.35,
"learning_rate": 4.973583119937072e-06,
"loss": 0.4126,
"step": 480
},
{
"epoch": 1.36,
"learning_rate": 4.934045723975441e-06,
"loss": 0.5116,
"step": 481
},
{
"epoch": 1.36,
"learning_rate": 4.894614573874877e-06,
"loss": 0.4601,
"step": 482
},
{
"epoch": 1.36,
"learning_rate": 4.85529049660623e-06,
"loss": 0.4523,
"step": 483
},
{
"epoch": 1.37,
"learning_rate": 4.81607431689475e-06,
"loss": 0.4724,
"step": 484
},
{
"epoch": 1.37,
"learning_rate": 4.776966857202816e-06,
"loss": 0.443,
"step": 485
},
{
"epoch": 1.37,
"learning_rate": 4.737968937712674e-06,
"loss": 0.424,
"step": 486
},
{
"epoch": 1.37,
"learning_rate": 4.699081376309218e-06,
"loss": 0.464,
"step": 487
},
{
"epoch": 1.38,
"learning_rate": 4.660304988562877e-06,
"loss": 0.463,
"step": 488
},
{
"epoch": 1.38,
"learning_rate": 4.621640587712468e-06,
"loss": 0.4517,
"step": 489
},
{
"epoch": 1.38,
"learning_rate": 4.583088984648172e-06,
"loss": 0.504,
"step": 490
},
{
"epoch": 1.39,
"learning_rate": 4.544650987894514e-06,
"loss": 0.4722,
"step": 491
},
{
"epoch": 1.39,
"learning_rate": 4.5063274035934016e-06,
"loss": 0.4614,
"step": 492
},
{
"epoch": 1.39,
"learning_rate": 4.468119035487231e-06,
"loss": 0.4521,
"step": 493
},
{
"epoch": 1.39,
"learning_rate": 4.430026684902017e-06,
"loss": 0.4331,
"step": 494
},
{
"epoch": 1.4,
"learning_rate": 4.392051150730602e-06,
"loss": 0.5002,
"step": 495
},
{
"epoch": 1.4,
"learning_rate": 4.354193229415882e-06,
"loss": 0.4311,
"step": 496
},
{
"epoch": 1.4,
"learning_rate": 4.3164537149341246e-06,
"loss": 0.4696,
"step": 497
},
{
"epoch": 1.4,
"learning_rate": 4.278833398778306e-06,
"loss": 0.3961,
"step": 498
},
{
"epoch": 1.41,
"learning_rate": 4.241333069941503e-06,
"loss": 0.491,
"step": 499
},
{
"epoch": 1.41,
"learning_rate": 4.203953514900366e-06,
"loss": 0.4306,
"step": 500
},
{
"epoch": 1.41,
"learning_rate": 4.166695517598611e-06,
"loss": 0.4192,
"step": 501
},
{
"epoch": 1.42,
"learning_rate": 4.129559859430573e-06,
"loss": 0.4798,
"step": 502
},
{
"epoch": 1.42,
"learning_rate": 4.092547319224837e-06,
"loss": 0.4457,
"step": 503
},
{
"epoch": 1.42,
"learning_rate": 4.055658673227891e-06,
"loss": 0.4201,
"step": 504
},
{
"epoch": 1.42,
"learning_rate": 4.01889469508784e-06,
"loss": 0.4716,
"step": 505
},
{
"epoch": 1.43,
"learning_rate": 3.982256155838199e-06,
"loss": 0.4343,
"step": 506
},
{
"epoch": 1.43,
"learning_rate": 3.945743823881713e-06,
"loss": 0.4017,
"step": 507
},
{
"epoch": 1.43,
"learning_rate": 3.909358464974228e-06,
"loss": 0.4396,
"step": 508
},
{
"epoch": 1.44,
"learning_rate": 3.873100842208661e-06,
"loss": 0.4993,
"step": 509
},
{
"epoch": 1.44,
"learning_rate": 3.836971715998968e-06,
"loss": 0.4443,
"step": 510
},
{
"epoch": 1.44,
"learning_rate": 3.8009718440642128e-06,
"loss": 0.4389,
"step": 511
},
{
"epoch": 1.44,
"learning_rate": 3.7651019814126656e-06,
"loss": 0.4449,
"step": 512
},
{
"epoch": 1.45,
"learning_rate": 3.729362880325983e-06,
"loss": 0.5124,
"step": 513
},
{
"epoch": 1.45,
"learning_rate": 3.693755290343409e-06,
"loss": 0.4506,
"step": 514
},
{
"epoch": 1.45,
"learning_rate": 3.658279958246075e-06,
"loss": 0.5028,
"step": 515
},
{
"epoch": 1.46,
"learning_rate": 3.622937628041334e-06,
"loss": 0.5026,
"step": 516
},
{
"epoch": 1.46,
"learning_rate": 3.587729040947141e-06,
"loss": 0.484,
"step": 517
},
{
"epoch": 1.46,
"learning_rate": 3.5526549353765294e-06,
"loss": 0.4245,
"step": 518
},
{
"epoch": 1.46,
"learning_rate": 3.5177160469221184e-06,
"loss": 0.4242,
"step": 519
},
{
"epoch": 1.47,
"learning_rate": 3.4829131083406684e-06,
"loss": 0.4908,
"step": 520
},
{
"epoch": 1.47,
"learning_rate": 3.448246849537741e-06,
"loss": 0.4685,
"step": 521
},
{
"epoch": 1.47,
"learning_rate": 3.413717997552376e-06,
"loss": 0.4946,
"step": 522
},
{
"epoch": 1.48,
"learning_rate": 3.379327276541834e-06,
"loss": 0.44,
"step": 523
},
{
"epoch": 1.48,
"learning_rate": 3.3450754077664337e-06,
"loss": 0.4958,
"step": 524
},
{
"epoch": 1.48,
"learning_rate": 3.310963109574408e-06,
"loss": 0.4574,
"step": 525
},
{
"epoch": 1.48,
"learning_rate": 3.2769910973868314e-06,
"loss": 0.4553,
"step": 526
},
{
"epoch": 1.49,
"learning_rate": 3.243160083682645e-06,
"loss": 0.4391,
"step": 527
},
{
"epoch": 1.49,
"learning_rate": 3.209470777983675e-06,
"loss": 0.4633,
"step": 528
},
{
"epoch": 1.49,
"learning_rate": 3.1759238868397925e-06,
"loss": 0.4394,
"step": 529
},
{
"epoch": 1.5,
"learning_rate": 3.1425201138140592e-06,
"loss": 0.4532,
"step": 530
},
{
"epoch": 1.5,
"learning_rate": 3.1092601594679993e-06,
"loss": 0.4292,
"step": 531
},
{
"epoch": 1.5,
"learning_rate": 3.0761447213468888e-06,
"loss": 0.4287,
"step": 532
},
{
"epoch": 1.5,
"learning_rate": 3.0431744939651365e-06,
"loss": 0.4513,
"step": 533
},
{
"epoch": 1.51,
"learning_rate": 3.0103501687917192e-06,
"loss": 0.4405,
"step": 534
},
{
"epoch": 1.51,
"learning_rate": 2.9776724342356654e-06,
"loss": 0.4896,
"step": 535
},
{
"epoch": 1.51,
"learning_rate": 2.945141975631637e-06,
"loss": 0.412,
"step": 536
},
{
"epoch": 1.51,
"learning_rate": 2.912759475225546e-06,
"loss": 0.4623,
"step": 537
},
{
"epoch": 1.52,
"learning_rate": 2.88052561216024e-06,
"loss": 0.472,
"step": 538
},
{
"epoch": 1.52,
"learning_rate": 2.8484410624612744e-06,
"loss": 0.4839,
"step": 539
},
{
"epoch": 1.52,
"learning_rate": 2.8165064990227255e-06,
"loss": 0.484,
"step": 540
},
{
"epoch": 1.53,
"learning_rate": 2.7847225915930697e-06,
"loss": 0.4328,
"step": 541
},
{
"epoch": 1.53,
"learning_rate": 2.7530900067611577e-06,
"loss": 0.4624,
"step": 542
},
{
"epoch": 1.53,
"learning_rate": 2.7216094079422185e-06,
"loss": 0.45,
"step": 543
},
{
"epoch": 1.53,
"learning_rate": 2.6902814553639443e-06,
"loss": 0.5348,
"step": 544
},
{
"epoch": 1.54,
"learning_rate": 2.6591068060526626e-06,
"loss": 0.4924,
"step": 545
},
{
"epoch": 1.54,
"learning_rate": 2.62808611381953e-06,
"loss": 0.4229,
"step": 546
},
{
"epoch": 1.54,
"learning_rate": 2.597220029246846e-06,
"loss": 0.4524,
"step": 547
},
{
"epoch": 1.55,
"learning_rate": 2.5665091996743898e-06,
"loss": 0.4206,
"step": 548
},
{
"epoch": 1.55,
"learning_rate": 2.5359542691858542e-06,
"loss": 0.4663,
"step": 549
},
{
"epoch": 1.55,
"learning_rate": 2.5055558785953304e-06,
"loss": 0.4712,
"step": 550
},
{
"epoch": 1.55,
"learning_rate": 2.4753146654338765e-06,
"loss": 0.4307,
"step": 551
},
{
"epoch": 1.56,
"learning_rate": 2.4452312639361462e-06,
"loss": 0.4847,
"step": 552
},
{
"epoch": 1.56,
"learning_rate": 2.415306305027072e-06,
"loss": 0.4654,
"step": 553
},
{
"epoch": 1.56,
"learning_rate": 2.3855404163086558e-06,
"loss": 0.4152,
"step": 554
},
{
"epoch": 1.57,
"learning_rate": 2.355934222046794e-06,
"loss": 0.3879,
"step": 555
},
{
"epoch": 1.57,
"learning_rate": 2.32648834315818e-06,
"loss": 0.507,
"step": 556
},
{
"epoch": 1.57,
"learning_rate": 2.2972033971972953e-06,
"loss": 0.443,
"step": 557
},
{
"epoch": 1.57,
"learning_rate": 2.2680799983434532e-06,
"loss": 0.4823,
"step": 558
},
{
"epoch": 1.58,
"learning_rate": 2.239118757387907e-06,
"loss": 0.4263,
"step": 559
},
{
"epoch": 1.58,
"learning_rate": 2.2103202817210555e-06,
"loss": 0.4716,
"step": 560
},
{
"epoch": 1.58,
"learning_rate": 2.1816851753197023e-06,
"loss": 0.4397,
"step": 561
},
{
"epoch": 1.59,
"learning_rate": 2.1532140387343736e-06,
"loss": 0.472,
"step": 562
},
{
"epoch": 1.59,
"learning_rate": 2.1249074690767434e-06,
"loss": 0.4537,
"step": 563
},
{
"epoch": 1.59,
"learning_rate": 2.096766060007096e-06,
"loss": 0.4534,
"step": 564
},
{
"epoch": 1.59,
"learning_rate": 2.068790401721886e-06,
"loss": 0.4681,
"step": 565
},
{
"epoch": 1.6,
"learning_rate": 2.040981080941349e-06,
"loss": 0.4756,
"step": 566
},
{
"epoch": 1.6,
"learning_rate": 2.013338680897209e-06,
"loss": 0.4756,
"step": 567
},
{
"epoch": 1.6,
"learning_rate": 1.9858637813204352e-06,
"loss": 0.4645,
"step": 568
},
{
"epoch": 1.61,
"learning_rate": 1.958556958429092e-06,
"loss": 0.5414,
"step": 569
},
{
"epoch": 1.61,
"learning_rate": 1.9314187849162523e-06,
"loss": 0.4712,
"step": 570
},
{
"epoch": 1.61,
"learning_rate": 1.904449829937981e-06,
"loss": 0.4827,
"step": 571
},
{
"epoch": 1.61,
"learning_rate": 1.8776506591014054e-06,
"loss": 0.4259,
"step": 572
},
{
"epoch": 1.62,
"learning_rate": 1.851021834452853e-06,
"loss": 0.5204,
"step": 573
},
{
"epoch": 1.62,
"learning_rate": 1.8245639144660532e-06,
"loss": 0.5036,
"step": 574
},
{
"epoch": 1.62,
"learning_rate": 1.7982774540304404e-06,
"loss": 0.4591,
"step": 575
},
{
"epoch": 1.62,
"learning_rate": 1.772163004439511e-06,
"loss": 0.4945,
"step": 576
},
{
"epoch": 1.63,
"learning_rate": 1.7462211133792484e-06,
"loss": 0.4027,
"step": 577
},
{
"epoch": 1.63,
"learning_rate": 1.720452324916656e-06,
"loss": 0.5019,
"step": 578
},
{
"epoch": 1.63,
"learning_rate": 1.6948571794883406e-06,
"loss": 0.5252,
"step": 579
},
{
"epoch": 1.64,
"learning_rate": 1.6694362138891674e-06,
"loss": 0.4238,
"step": 580
},
{
"epoch": 1.64,
"learning_rate": 1.6441899612610178e-06,
"loss": 0.4698,
"step": 581
},
{
"epoch": 1.64,
"learning_rate": 1.6191189510815942e-06,
"loss": 0.436,
"step": 582
},
{
"epoch": 1.64,
"learning_rate": 1.5942237091533297e-06,
"loss": 0.4611,
"step": 583
},
{
"epoch": 1.65,
"learning_rate": 1.5695047575923462e-06,
"loss": 0.494,
"step": 584
},
{
"epoch": 1.65,
"learning_rate": 1.5449626148175144e-06,
"loss": 0.3985,
"step": 585
},
{
"epoch": 1.65,
"learning_rate": 1.5205977955395812e-06,
"loss": 0.4283,
"step": 586
},
{
"epoch": 1.66,
"learning_rate": 1.4964108107503638e-06,
"loss": 0.4371,
"step": 587
},
{
"epoch": 1.66,
"learning_rate": 1.4724021677120491e-06,
"loss": 0.4359,
"step": 588
},
{
"epoch": 1.66,
"learning_rate": 1.4485723699465392e-06,
"loss": 0.4785,
"step": 589
},
{
"epoch": 1.66,
"learning_rate": 1.4249219172249051e-06,
"loss": 0.505,
"step": 590
},
{
"epoch": 1.67,
"learning_rate": 1.4014513055568978e-06,
"loss": 0.4873,
"step": 591
},
{
"epoch": 1.67,
"learning_rate": 1.3781610271805436e-06,
"loss": 0.4346,
"step": 592
},
{
"epoch": 1.67,
"learning_rate": 1.3550515705518263e-06,
"loss": 0.4558,
"step": 593
},
{
"epoch": 1.68,
"learning_rate": 1.3321234203344435e-06,
"loss": 0.4531,
"step": 594
},
{
"epoch": 1.68,
"learning_rate": 1.3093770573896369e-06,
"loss": 0.4454,
"step": 595
},
{
"epoch": 1.68,
"learning_rate": 1.286812958766106e-06,
"loss": 0.4384,
"step": 596
},
{
"epoch": 1.68,
"learning_rate": 1.2644315976900145e-06,
"loss": 0.4576,
"step": 597
},
{
"epoch": 1.69,
"learning_rate": 1.242233443555051e-06,
"loss": 0.4552,
"step": 598
},
{
"epoch": 1.69,
"learning_rate": 1.220218961912597e-06,
"loss": 0.4332,
"step": 599
},
{
"epoch": 1.69,
"learning_rate": 1.1983886144619527e-06,
"loss": 0.4256,
"step": 600
},
{
"epoch": 1.7,
"learning_rate": 1.1767428590406648e-06,
"loss": 0.4383,
"step": 601
},
{
"epoch": 1.7,
"learning_rate": 1.1552821496149136e-06,
"loss": 0.4543,
"step": 602
},
{
"epoch": 1.7,
"learning_rate": 1.134006936269999e-06,
"loss": 0.4787,
"step": 603
},
{
"epoch": 1.7,
"learning_rate": 1.1129176652009043e-06,
"loss": 0.4688,
"step": 604
},
{
"epoch": 1.71,
"learning_rate": 1.0920147787029233e-06,
"loss": 0.4538,
"step": 605
},
{
"epoch": 1.71,
"learning_rate": 1.0712987151624056e-06,
"loss": 0.3828,
"step": 606
},
{
"epoch": 1.71,
"learning_rate": 1.05076990904754e-06,
"loss": 0.4247,
"step": 607
},
{
"epoch": 1.72,
"learning_rate": 1.0304287908992626e-06,
"loss": 0.4164,
"step": 608
},
{
"epoch": 1.72,
"learning_rate": 1.010275787322219e-06,
"loss": 0.47,
"step": 609
},
{
"epoch": 1.72,
"learning_rate": 9.903113209758098e-07,
"loss": 0.4296,
"step": 610
},
{
"epoch": 1.72,
"learning_rate": 9.705358105653373e-07,
"loss": 0.4375,
"step": 611
},
{
"epoch": 1.73,
"learning_rate": 9.509496708332233e-07,
"loss": 0.4826,
"step": 612
},
{
"epoch": 1.73,
"learning_rate": 9.315533125503051e-07,
"loss": 0.3917,
"step": 613
},
{
"epoch": 1.73,
"learning_rate": 9.123471425072205e-07,
"loss": 0.4271,
"step": 614
},
{
"epoch": 1.73,
"learning_rate": 8.933315635058881e-07,
"loss": 0.4593,
"step": 615
},
{
"epoch": 1.74,
"learning_rate": 8.745069743510393e-07,
"loss": 0.4604,
"step": 616
},
{
"epoch": 1.74,
"learning_rate": 8.558737698418762e-07,
"loss": 0.421,
"step": 617
},
{
"epoch": 1.74,
"learning_rate": 8.374323407637741e-07,
"loss": 0.442,
"step": 618
},
{
"epoch": 1.75,
"learning_rate": 8.191830738800977e-07,
"loss": 0.4529,
"step": 619
},
{
"epoch": 1.75,
"learning_rate": 8.01126351924082e-07,
"loss": 0.4852,
"step": 620
},
{
"epoch": 1.75,
"learning_rate": 7.83262553590809e-07,
"loss": 0.4609,
"step": 621
},
{
"epoch": 1.75,
"learning_rate": 7.655920535292682e-07,
"loss": 0.5109,
"step": 622
},
{
"epoch": 1.76,
"learning_rate": 7.48115222334489e-07,
"loss": 0.4688,
"step": 623
},
{
"epoch": 1.76,
"learning_rate": 7.308324265397837e-07,
"loss": 0.3873,
"step": 624
},
{
"epoch": 1.76,
"learning_rate": 7.137440286090436e-07,
"loss": 0.4641,
"step": 625
},
{
"epoch": 1.77,
"learning_rate": 6.968503869291521e-07,
"loss": 0.4536,
"step": 626
},
{
"epoch": 1.77,
"learning_rate": 6.8015185580246e-07,
"loss": 0.466,
"step": 627
},
{
"epoch": 1.77,
"learning_rate": 6.636487854393536e-07,
"loss": 0.472,
"step": 628
},
{
"epoch": 1.77,
"learning_rate": 6.473415219509182e-07,
"loss": 0.4584,
"step": 629
},
{
"epoch": 1.78,
"learning_rate": 6.31230407341672e-07,
"loss": 0.4804,
"step": 630
},
{
"epoch": 1.78,
"learning_rate": 6.153157795023956e-07,
"loss": 0.4124,
"step": 631
},
{
"epoch": 1.78,
"learning_rate": 5.995979722030443e-07,
"loss": 0.4818,
"step": 632
},
{
"epoch": 1.79,
"learning_rate": 5.840773150857526e-07,
"loss": 0.4955,
"step": 633
},
{
"epoch": 1.79,
"learning_rate": 5.687541336579127e-07,
"loss": 0.4328,
"step": 634
},
{
"epoch": 1.79,
"learning_rate": 5.536287492853575e-07,
"loss": 0.4655,
"step": 635
},
{
"epoch": 1.79,
"learning_rate": 5.387014791856127e-07,
"loss": 0.4015,
"step": 636
},
{
"epoch": 1.8,
"learning_rate": 5.239726364212494e-07,
"loss": 0.4157,
"step": 637
},
{
"epoch": 1.8,
"learning_rate": 5.094425298933136e-07,
"loss": 0.4912,
"step": 638
},
{
"epoch": 1.8,
"learning_rate": 4.951114643348531e-07,
"loss": 0.4611,
"step": 639
},
{
"epoch": 1.81,
"learning_rate": 4.809797403045224e-07,
"loss": 0.4451,
"step": 640
},
{
"epoch": 1.81,
"learning_rate": 4.670476541802782e-07,
"loss": 0.4195,
"step": 641
},
{
"epoch": 1.81,
"learning_rate": 4.533154981531718e-07,
"loss": 0.4433,
"step": 642
},
{
"epoch": 1.81,
"learning_rate": 4.397835602212064e-07,
"loss": 0.4133,
"step": 643
},
{
"epoch": 1.82,
"learning_rate": 4.264521241833153e-07,
"loss": 0.4736,
"step": 644
},
{
"epoch": 1.82,
"learning_rate": 4.133214696333943e-07,
"loss": 0.4531,
"step": 645
},
{
"epoch": 1.82,
"learning_rate": 4.003918719544464e-07,
"loss": 0.4622,
"step": 646
},
{
"epoch": 1.83,
"learning_rate": 3.876636023128022e-07,
"loss": 0.4242,
"step": 647
},
{
"epoch": 1.83,
"learning_rate": 3.7513692765243637e-07,
"loss": 0.4365,
"step": 648
},
{
"epoch": 1.83,
"learning_rate": 3.628121106893701e-07,
"loss": 0.4341,
"step": 649
},
{
"epoch": 1.83,
"learning_rate": 3.50689409906152e-07,
"loss": 0.4736,
"step": 650
},
{
"epoch": 1.84,
"learning_rate": 3.3876907954644933e-07,
"loss": 0.47,
"step": 651
},
{
"epoch": 1.84,
"learning_rate": 3.2705136960970554e-07,
"loss": 0.4202,
"step": 652
},
{
"epoch": 1.84,
"learning_rate": 3.1553652584590864e-07,
"loss": 0.447,
"step": 653
},
{
"epoch": 1.84,
"learning_rate": 3.0422478975042245e-07,
"loss": 0.5296,
"step": 654
},
{
"epoch": 1.85,
"learning_rate": 2.931163985589369e-07,
"loss": 0.4101,
"step": 655
},
{
"epoch": 1.85,
"learning_rate": 2.8221158524248003e-07,
"loss": 0.4851,
"step": 656
},
{
"epoch": 1.85,
"learning_rate": 2.7151057850253957e-07,
"loss": 0.4496,
"step": 657
},
{
"epoch": 1.86,
"learning_rate": 2.6101360276626795e-07,
"loss": 0.4855,
"step": 658
},
{
"epoch": 1.86,
"learning_rate": 2.507208781817638e-07,
"loss": 0.4476,
"step": 659
},
{
"epoch": 1.86,
"learning_rate": 2.406326206134724e-07,
"loss": 0.4706,
"step": 660
},
{
"epoch": 1.86,
"learning_rate": 2.3074904163764012e-07,
"loss": 0.4522,
"step": 661
},
{
"epoch": 1.87,
"learning_rate": 2.210703485378929e-07,
"loss": 0.4455,
"step": 662
},
{
"epoch": 1.87,
"learning_rate": 2.115967443008804e-07,
"loss": 0.3785,
"step": 663
},
{
"epoch": 1.87,
"learning_rate": 2.0232842761201854e-07,
"loss": 0.4388,
"step": 664
},
{
"epoch": 1.88,
"learning_rate": 1.9326559285132495e-07,
"loss": 0.444,
"step": 665
},
{
"epoch": 1.88,
"learning_rate": 1.844084300893456e-07,
"loss": 0.4966,
"step": 666
},
{
"epoch": 1.88,
"learning_rate": 1.7575712508316244e-07,
"loss": 0.443,
"step": 667
},
{
"epoch": 1.88,
"learning_rate": 1.673118592724987e-07,
"loss": 0.4418,
"step": 668
},
{
"epoch": 1.89,
"learning_rate": 1.5907280977591866e-07,
"loss": 0.4593,
"step": 669
},
{
"epoch": 1.89,
"learning_rate": 1.5104014938710498e-07,
"loss": 0.4883,
"step": 670
},
{
"epoch": 1.89,
"learning_rate": 1.4321404657124393e-07,
"loss": 0.4186,
"step": 671
},
{
"epoch": 1.9,
"learning_rate": 1.3559466546148369e-07,
"loss": 0.4734,
"step": 672
},
{
"epoch": 1.9,
"learning_rate": 1.2818216585549824e-07,
"loss": 0.4429,
"step": 673
},
{
"epoch": 1.9,
"learning_rate": 1.209767032121345e-07,
"loss": 0.4556,
"step": 674
},
{
"epoch": 1.9,
"learning_rate": 1.1397842864814712e-07,
"loss": 0.4789,
"step": 675
},
{
"epoch": 1.91,
"learning_rate": 1.0718748893503883e-07,
"loss": 0.4461,
"step": 676
},
{
"epoch": 1.91,
"learning_rate": 1.0060402649597178e-07,
"loss": 0.4182,
"step": 677
},
{
"epoch": 1.91,
"learning_rate": 9.422817940278773e-08,
"loss": 0.4875,
"step": 678
},
{
"epoch": 1.92,
"learning_rate": 8.806008137311028e-08,
"loss": 0.4959,
"step": 679
},
{
"epoch": 1.92,
"learning_rate": 8.209986176753947e-08,
"loss": 0.4624,
"step": 680
},
{
"epoch": 1.92,
"learning_rate": 7.634764558693941e-08,
"loss": 0.4531,
"step": 681
},
{
"epoch": 1.92,
"learning_rate": 7.080355346981815e-08,
"loss": 0.4519,
"step": 682
},
{
"epoch": 1.93,
"learning_rate": 6.546770168979421e-08,
"loss": 0.4379,
"step": 683
},
{
"epoch": 1.93,
"learning_rate": 6.034020215316184e-08,
"loss": 0.4903,
"step": 684
},
{
"epoch": 1.93,
"learning_rate": 5.5421162396542824e-08,
"loss": 0.4506,
"step": 685
},
{
"epoch": 1.94,
"learning_rate": 5.071068558462733e-08,
"loss": 0.4371,
"step": 686
},
{
"epoch": 1.94,
"learning_rate": 4.6208870508017703e-08,
"loss": 0.442,
"step": 687
},
{
"epoch": 1.94,
"learning_rate": 4.191581158115021e-08,
"loss": 0.4662,
"step": 688
},
{
"epoch": 1.94,
"learning_rate": 3.783159884031773e-08,
"loss": 0.4547,
"step": 689
},
{
"epoch": 1.95,
"learning_rate": 3.3956317941779004e-08,
"loss": 0.4365,
"step": 690
},
{
"epoch": 1.95,
"learning_rate": 3.029005015996789e-08,
"loss": 0.4422,
"step": 691
},
{
"epoch": 1.95,
"learning_rate": 2.6832872385783583e-08,
"loss": 0.4141,
"step": 692
},
{
"epoch": 1.95,
"learning_rate": 2.3584857124977488e-08,
"loss": 0.4905,
"step": 693
},
{
"epoch": 1.96,
"learning_rate": 2.054607249663665e-08,
"loss": 0.4665,
"step": 694
},
{
"epoch": 1.96,
"learning_rate": 1.7716582231752656e-08,
"loss": 0.4919,
"step": 695
},
{
"epoch": 1.96,
"learning_rate": 1.509644567188717e-08,
"loss": 0.4455,
"step": 696
},
{
"epoch": 1.97,
"learning_rate": 1.2685717767921823e-08,
"loss": 0.4496,
"step": 697
},
{
"epoch": 1.97,
"learning_rate": 1.048444907891244e-08,
"loss": 0.4723,
"step": 698
},
{
"epoch": 1.97,
"learning_rate": 8.492685771025466e-09,
"loss": 0.4477,
"step": 699
},
{
"epoch": 1.97,
"learning_rate": 6.710469616569848e-09,
"loss": 0.4645,
"step": 700
},
{
"epoch": 1.98,
"learning_rate": 5.137837993121064e-09,
"loss": 0.4334,
"step": 701
},
{
"epoch": 1.98,
"learning_rate": 3.774823882738421e-09,
"loss": 0.5298,
"step": 702
},
{
"epoch": 1.98,
"learning_rate": 2.6214558712722714e-09,
"loss": 0.4129,
"step": 703
},
{
"epoch": 1.99,
"learning_rate": 1.677758147762276e-09,
"loss": 0.3858,
"step": 704
},
{
"epoch": 1.99,
"learning_rate": 9.43750503935581e-10,
"loss": 0.4319,
"step": 705
},
{
"epoch": 1.99,
"learning_rate": 4.1944833378604334e-10,
"loss": 0.425,
"step": 706
},
{
"epoch": 1.99,
"learning_rate": 1.0486263325559798e-10,
"loss": 0.4319,
"step": 707
},
{
"epoch": 2.0,
"learning_rate": 0.0,
"loss": 0.4293,
"step": 708
},
{
"epoch": 2.0,
"step": 708,
"total_flos": 3.1936017234447565e+17,
"train_loss": 0.7175506229201952,
"train_runtime": 6618.2919,
"train_samples_per_second": 6.849,
"train_steps_per_second": 0.107
}
],
"max_steps": 708,
"num_train_epochs": 2,
"total_flos": 3.1936017234447565e+17,
"trial_name": null,
"trial_params": null
}