final-lora-426 / trainer_state.json
ycchen's picture
Upload folder using huggingface_hub
3ff5fc0
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 426,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 2.938151902453357e-07,
"loss": 1.5844,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 5.876303804906714e-07,
"loss": 1.1717,
"step": 2
},
{
"epoch": 0.02,
"learning_rate": 8.814455707360072e-07,
"loss": 1.1709,
"step": 3
},
{
"epoch": 0.03,
"learning_rate": 1.1752607609813429e-06,
"loss": 1.5208,
"step": 4
},
{
"epoch": 0.04,
"learning_rate": 1.4690759512266787e-06,
"loss": 1.2788,
"step": 5
},
{
"epoch": 0.04,
"learning_rate": 1.7628911414720144e-06,
"loss": 1.39,
"step": 6
},
{
"epoch": 0.05,
"learning_rate": 2.05670633171735e-06,
"loss": 1.4615,
"step": 7
},
{
"epoch": 0.06,
"learning_rate": 2.3505215219626858e-06,
"loss": 1.2082,
"step": 8
},
{
"epoch": 0.06,
"learning_rate": 2.6443367122080216e-06,
"loss": 1.3854,
"step": 9
},
{
"epoch": 0.07,
"learning_rate": 2.9381519024533573e-06,
"loss": 1.1711,
"step": 10
},
{
"epoch": 0.08,
"learning_rate": 3.2319670926986927e-06,
"loss": 1.3714,
"step": 11
},
{
"epoch": 0.08,
"learning_rate": 3.525782282944029e-06,
"loss": 1.6743,
"step": 12
},
{
"epoch": 0.09,
"learning_rate": 3.819597473189365e-06,
"loss": 1.2937,
"step": 13
},
{
"epoch": 0.1,
"learning_rate": 4.1134126634347e-06,
"loss": 1.2628,
"step": 14
},
{
"epoch": 0.11,
"learning_rate": 4.407227853680035e-06,
"loss": 1.3822,
"step": 15
},
{
"epoch": 0.11,
"learning_rate": 4.7010430439253716e-06,
"loss": 1.3788,
"step": 16
},
{
"epoch": 0.12,
"learning_rate": 4.994858234170707e-06,
"loss": 1.5392,
"step": 17
},
{
"epoch": 0.13,
"learning_rate": 5.288673424416043e-06,
"loss": 1.3841,
"step": 18
},
{
"epoch": 0.13,
"learning_rate": 5.582488614661379e-06,
"loss": 1.3883,
"step": 19
},
{
"epoch": 0.14,
"learning_rate": 5.876303804906715e-06,
"loss": 1.444,
"step": 20
},
{
"epoch": 0.15,
"learning_rate": 6.17011899515205e-06,
"loss": 1.0966,
"step": 21
},
{
"epoch": 0.15,
"learning_rate": 6.463934185397385e-06,
"loss": 1.4783,
"step": 22
},
{
"epoch": 0.16,
"learning_rate": 6.757749375642722e-06,
"loss": 1.3966,
"step": 23
},
{
"epoch": 0.17,
"learning_rate": 7.051564565888058e-06,
"loss": 1.602,
"step": 24
},
{
"epoch": 0.18,
"learning_rate": 7.345379756133393e-06,
"loss": 1.4453,
"step": 25
},
{
"epoch": 0.18,
"learning_rate": 7.63919494637873e-06,
"loss": 1.2995,
"step": 26
},
{
"epoch": 0.19,
"learning_rate": 7.933010136624066e-06,
"loss": 1.318,
"step": 27
},
{
"epoch": 0.2,
"learning_rate": 8.2268253268694e-06,
"loss": 1.1796,
"step": 28
},
{
"epoch": 0.2,
"learning_rate": 8.520640517114736e-06,
"loss": 1.31,
"step": 29
},
{
"epoch": 0.21,
"learning_rate": 8.81445570736007e-06,
"loss": 1.4279,
"step": 30
},
{
"epoch": 0.22,
"learning_rate": 9.108270897605407e-06,
"loss": 1.5326,
"step": 31
},
{
"epoch": 0.23,
"learning_rate": 9.402086087850743e-06,
"loss": 1.2816,
"step": 32
},
{
"epoch": 0.23,
"learning_rate": 9.695901278096078e-06,
"loss": 1.4097,
"step": 33
},
{
"epoch": 0.24,
"learning_rate": 9.989716468341414e-06,
"loss": 1.3857,
"step": 34
},
{
"epoch": 0.25,
"learning_rate": 1.028353165858675e-05,
"loss": 1.5109,
"step": 35
},
{
"epoch": 0.25,
"learning_rate": 1.0577346848832086e-05,
"loss": 1.5855,
"step": 36
},
{
"epoch": 0.26,
"learning_rate": 1.0871162039077422e-05,
"loss": 1.4847,
"step": 37
},
{
"epoch": 0.27,
"learning_rate": 1.1164977229322759e-05,
"loss": 1.3497,
"step": 38
},
{
"epoch": 0.27,
"learning_rate": 1.1458792419568093e-05,
"loss": 1.3788,
"step": 39
},
{
"epoch": 0.28,
"learning_rate": 1.175260760981343e-05,
"loss": 1.4044,
"step": 40
},
{
"epoch": 0.29,
"learning_rate": 1.2046422800058766e-05,
"loss": 1.3204,
"step": 41
},
{
"epoch": 0.3,
"learning_rate": 1.23402379903041e-05,
"loss": 1.4074,
"step": 42
},
{
"epoch": 0.3,
"learning_rate": 1.2634053180549435e-05,
"loss": 1.3244,
"step": 43
},
{
"epoch": 0.31,
"learning_rate": 1.292786837079477e-05,
"loss": 1.4781,
"step": 44
},
{
"epoch": 0.32,
"learning_rate": 1.3221683561040107e-05,
"loss": 1.166,
"step": 45
},
{
"epoch": 0.32,
"learning_rate": 1.3515498751285443e-05,
"loss": 1.5956,
"step": 46
},
{
"epoch": 0.33,
"learning_rate": 1.380931394153078e-05,
"loss": 1.3736,
"step": 47
},
{
"epoch": 0.34,
"learning_rate": 1.4103129131776116e-05,
"loss": 1.5388,
"step": 48
},
{
"epoch": 0.35,
"learning_rate": 1.439694432202145e-05,
"loss": 1.4948,
"step": 49
},
{
"epoch": 0.35,
"learning_rate": 1.4690759512266786e-05,
"loss": 1.3604,
"step": 50
},
{
"epoch": 0.36,
"learning_rate": 1.4984574702512122e-05,
"loss": 1.5697,
"step": 51
},
{
"epoch": 0.37,
"learning_rate": 1.527838989275746e-05,
"loss": 1.3266,
"step": 52
},
{
"epoch": 0.37,
"learning_rate": 1.5572205083002795e-05,
"loss": 1.2487,
"step": 53
},
{
"epoch": 0.38,
"learning_rate": 1.586602027324813e-05,
"loss": 1.3963,
"step": 54
},
{
"epoch": 0.39,
"learning_rate": 1.6159835463493464e-05,
"loss": 1.3293,
"step": 55
},
{
"epoch": 0.39,
"learning_rate": 1.64536506537388e-05,
"loss": 1.5669,
"step": 56
},
{
"epoch": 0.4,
"learning_rate": 1.6747465843984136e-05,
"loss": 1.3252,
"step": 57
},
{
"epoch": 0.41,
"learning_rate": 1.7041281034229472e-05,
"loss": 1.3639,
"step": 58
},
{
"epoch": 0.42,
"learning_rate": 1.733509622447481e-05,
"loss": 1.3889,
"step": 59
},
{
"epoch": 0.42,
"learning_rate": 1.762891141472014e-05,
"loss": 1.3824,
"step": 60
},
{
"epoch": 0.43,
"learning_rate": 1.7922726604965478e-05,
"loss": 1.4756,
"step": 61
},
{
"epoch": 0.44,
"learning_rate": 1.8216541795210814e-05,
"loss": 1.2453,
"step": 62
},
{
"epoch": 0.44,
"learning_rate": 1.851035698545615e-05,
"loss": 1.454,
"step": 63
},
{
"epoch": 0.45,
"learning_rate": 1.8804172175701486e-05,
"loss": 1.26,
"step": 64
},
{
"epoch": 0.46,
"learning_rate": 1.9097987365946822e-05,
"loss": 1.2652,
"step": 65
},
{
"epoch": 0.46,
"learning_rate": 1.9391802556192155e-05,
"loss": 1.424,
"step": 66
},
{
"epoch": 0.47,
"learning_rate": 1.968561774643749e-05,
"loss": 1.4516,
"step": 67
},
{
"epoch": 0.48,
"learning_rate": 1.9979432936682828e-05,
"loss": 1.4799,
"step": 68
},
{
"epoch": 0.49,
"learning_rate": 2.0273248126928164e-05,
"loss": 1.3161,
"step": 69
},
{
"epoch": 0.49,
"learning_rate": 2.05670633171735e-05,
"loss": 1.4528,
"step": 70
},
{
"epoch": 0.5,
"learning_rate": 2.0860878507418836e-05,
"loss": 1.54,
"step": 71
},
{
"epoch": 0.51,
"learning_rate": 2.1154693697664172e-05,
"loss": 1.3282,
"step": 72
},
{
"epoch": 0.51,
"learning_rate": 2.144850888790951e-05,
"loss": 1.2715,
"step": 73
},
{
"epoch": 0.52,
"learning_rate": 2.1742324078154845e-05,
"loss": 1.616,
"step": 74
},
{
"epoch": 0.53,
"learning_rate": 2.203613926840018e-05,
"loss": 1.451,
"step": 75
},
{
"epoch": 0.54,
"learning_rate": 2.2329954458645517e-05,
"loss": 1.3391,
"step": 76
},
{
"epoch": 0.54,
"learning_rate": 2.2623769648890853e-05,
"loss": 1.3733,
"step": 77
},
{
"epoch": 0.55,
"learning_rate": 2.2917584839136186e-05,
"loss": 1.2611,
"step": 78
},
{
"epoch": 0.56,
"learning_rate": 2.3211400029381523e-05,
"loss": 1.2367,
"step": 79
},
{
"epoch": 0.56,
"learning_rate": 2.350521521962686e-05,
"loss": 1.3784,
"step": 80
},
{
"epoch": 0.57,
"learning_rate": 2.3799030409872195e-05,
"loss": 1.2371,
"step": 81
},
{
"epoch": 0.58,
"learning_rate": 2.409284560011753e-05,
"loss": 1.326,
"step": 82
},
{
"epoch": 0.58,
"learning_rate": 2.4386660790362864e-05,
"loss": 1.5421,
"step": 83
},
{
"epoch": 0.59,
"learning_rate": 2.46804759806082e-05,
"loss": 1.5432,
"step": 84
},
{
"epoch": 0.6,
"learning_rate": 2.4974291170853536e-05,
"loss": 1.4241,
"step": 85
},
{
"epoch": 0.61,
"learning_rate": 2.526810636109887e-05,
"loss": 1.4539,
"step": 86
},
{
"epoch": 0.61,
"learning_rate": 2.556192155134421e-05,
"loss": 1.3261,
"step": 87
},
{
"epoch": 0.62,
"learning_rate": 2.585573674158954e-05,
"loss": 1.3152,
"step": 88
},
{
"epoch": 0.63,
"learning_rate": 2.614955193183488e-05,
"loss": 1.4292,
"step": 89
},
{
"epoch": 0.63,
"learning_rate": 2.6443367122080214e-05,
"loss": 1.1705,
"step": 90
},
{
"epoch": 0.64,
"learning_rate": 2.6737182312325554e-05,
"loss": 1.1443,
"step": 91
},
{
"epoch": 0.65,
"learning_rate": 2.7030997502570886e-05,
"loss": 1.4355,
"step": 92
},
{
"epoch": 0.65,
"learning_rate": 2.732481269281622e-05,
"loss": 1.3192,
"step": 93
},
{
"epoch": 0.66,
"learning_rate": 2.761862788306156e-05,
"loss": 1.4356,
"step": 94
},
{
"epoch": 0.67,
"learning_rate": 2.791244307330689e-05,
"loss": 1.1797,
"step": 95
},
{
"epoch": 0.68,
"learning_rate": 2.820625826355223e-05,
"loss": 1.1635,
"step": 96
},
{
"epoch": 0.68,
"learning_rate": 2.8500073453797564e-05,
"loss": 1.2257,
"step": 97
},
{
"epoch": 0.69,
"learning_rate": 2.87938886440429e-05,
"loss": 1.322,
"step": 98
},
{
"epoch": 0.7,
"learning_rate": 2.9087703834288236e-05,
"loss": 1.3281,
"step": 99
},
{
"epoch": 0.7,
"learning_rate": 2.9381519024533573e-05,
"loss": 1.0323,
"step": 100
},
{
"epoch": 0.71,
"learning_rate": 2.967533421477891e-05,
"loss": 1.1299,
"step": 101
},
{
"epoch": 0.72,
"learning_rate": 2.9969149405024245e-05,
"loss": 1.2833,
"step": 102
},
{
"epoch": 0.73,
"learning_rate": 3.0262964595269578e-05,
"loss": 1.3501,
"step": 103
},
{
"epoch": 0.73,
"learning_rate": 3.055677978551492e-05,
"loss": 1.1508,
"step": 104
},
{
"epoch": 0.74,
"learning_rate": 3.085059497576025e-05,
"loss": 1.418,
"step": 105
},
{
"epoch": 0.75,
"learning_rate": 3.114441016600559e-05,
"loss": 1.2717,
"step": 106
},
{
"epoch": 0.75,
"learning_rate": 3.143822535625092e-05,
"loss": 1.3274,
"step": 107
},
{
"epoch": 0.76,
"learning_rate": 3.173204054649626e-05,
"loss": 1.3633,
"step": 108
},
{
"epoch": 0.77,
"learning_rate": 3.202585573674159e-05,
"loss": 1.2618,
"step": 109
},
{
"epoch": 0.77,
"learning_rate": 3.231967092698693e-05,
"loss": 1.4994,
"step": 110
},
{
"epoch": 0.78,
"learning_rate": 3.2613486117232264e-05,
"loss": 1.4325,
"step": 111
},
{
"epoch": 0.79,
"learning_rate": 3.29073013074776e-05,
"loss": 1.3578,
"step": 112
},
{
"epoch": 0.8,
"learning_rate": 3.3201116497722936e-05,
"loss": 1.2844,
"step": 113
},
{
"epoch": 0.8,
"learning_rate": 3.349493168796827e-05,
"loss": 1.1555,
"step": 114
},
{
"epoch": 0.81,
"learning_rate": 3.378874687821361e-05,
"loss": 1.3242,
"step": 115
},
{
"epoch": 0.82,
"learning_rate": 3.4082562068458945e-05,
"loss": 1.1545,
"step": 116
},
{
"epoch": 0.82,
"learning_rate": 3.437637725870428e-05,
"loss": 1.1827,
"step": 117
},
{
"epoch": 0.83,
"learning_rate": 3.467019244894962e-05,
"loss": 1.5443,
"step": 118
},
{
"epoch": 0.84,
"learning_rate": 3.4964007639194954e-05,
"loss": 1.1105,
"step": 119
},
{
"epoch": 0.85,
"learning_rate": 3.525782282944028e-05,
"loss": 1.3643,
"step": 120
},
{
"epoch": 0.85,
"learning_rate": 3.5551638019685626e-05,
"loss": 1.335,
"step": 121
},
{
"epoch": 0.86,
"learning_rate": 3.5845453209930955e-05,
"loss": 1.2855,
"step": 122
},
{
"epoch": 0.87,
"learning_rate": 3.61392684001763e-05,
"loss": 1.5569,
"step": 123
},
{
"epoch": 0.87,
"learning_rate": 3.643308359042163e-05,
"loss": 1.3648,
"step": 124
},
{
"epoch": 0.88,
"learning_rate": 3.672689878066697e-05,
"loss": 1.1981,
"step": 125
},
{
"epoch": 0.89,
"learning_rate": 3.70207139709123e-05,
"loss": 1.281,
"step": 126
},
{
"epoch": 0.89,
"learning_rate": 3.7314529161157636e-05,
"loss": 1.2883,
"step": 127
},
{
"epoch": 0.9,
"learning_rate": 3.760834435140297e-05,
"loss": 1.4385,
"step": 128
},
{
"epoch": 0.91,
"learning_rate": 3.790215954164831e-05,
"loss": 1.4013,
"step": 129
},
{
"epoch": 0.92,
"learning_rate": 3.8195974731893645e-05,
"loss": 1.3303,
"step": 130
},
{
"epoch": 0.92,
"learning_rate": 3.848978992213898e-05,
"loss": 1.3222,
"step": 131
},
{
"epoch": 0.93,
"learning_rate": 3.878360511238431e-05,
"loss": 1.3686,
"step": 132
},
{
"epoch": 0.94,
"learning_rate": 3.9077420302629654e-05,
"loss": 1.3684,
"step": 133
},
{
"epoch": 0.94,
"learning_rate": 3.937123549287498e-05,
"loss": 1.3496,
"step": 134
},
{
"epoch": 0.95,
"learning_rate": 3.9665050683120326e-05,
"loss": 1.2483,
"step": 135
},
{
"epoch": 0.96,
"learning_rate": 3.9958865873365655e-05,
"loss": 1.3186,
"step": 136
},
{
"epoch": 0.96,
"learning_rate": 4.025268106361099e-05,
"loss": 1.3585,
"step": 137
},
{
"epoch": 0.97,
"learning_rate": 4.054649625385633e-05,
"loss": 1.1332,
"step": 138
},
{
"epoch": 0.98,
"learning_rate": 4.0840311444101664e-05,
"loss": 1.3868,
"step": 139
},
{
"epoch": 0.99,
"learning_rate": 4.1134126634347e-05,
"loss": 1.5011,
"step": 140
},
{
"epoch": 0.99,
"learning_rate": 4.1427941824592336e-05,
"loss": 1.2125,
"step": 141
},
{
"epoch": 1.0,
"learning_rate": 4.172175701483767e-05,
"loss": 1.2079,
"step": 142
},
{
"epoch": 1.01,
"learning_rate": 4.201557220508301e-05,
"loss": 1.5137,
"step": 143
},
{
"epoch": 1.01,
"learning_rate": 4.2309387395328345e-05,
"loss": 1.0962,
"step": 144
},
{
"epoch": 1.02,
"learning_rate": 4.260320258557368e-05,
"loss": 1.0864,
"step": 145
},
{
"epoch": 1.03,
"learning_rate": 4.289701777581902e-05,
"loss": 1.433,
"step": 146
},
{
"epoch": 1.04,
"learning_rate": 4.3190832966064354e-05,
"loss": 1.1769,
"step": 147
},
{
"epoch": 1.04,
"learning_rate": 4.348464815630969e-05,
"loss": 1.3221,
"step": 148
},
{
"epoch": 1.05,
"learning_rate": 4.377846334655502e-05,
"loss": 1.3977,
"step": 149
},
{
"epoch": 1.06,
"learning_rate": 4.407227853680036e-05,
"loss": 1.1165,
"step": 150
},
{
"epoch": 1.06,
"learning_rate": 4.436609372704569e-05,
"loss": 1.2971,
"step": 151
},
{
"epoch": 1.07,
"learning_rate": 4.4659908917291035e-05,
"loss": 1.0846,
"step": 152
},
{
"epoch": 1.08,
"learning_rate": 4.4953724107536364e-05,
"loss": 1.2692,
"step": 153
},
{
"epoch": 1.08,
"learning_rate": 4.524753929778171e-05,
"loss": 1.6022,
"step": 154
},
{
"epoch": 1.09,
"learning_rate": 4.5541354488027036e-05,
"loss": 1.2185,
"step": 155
},
{
"epoch": 1.1,
"learning_rate": 4.583516967827237e-05,
"loss": 1.1814,
"step": 156
},
{
"epoch": 1.11,
"learning_rate": 4.612898486851771e-05,
"loss": 1.2761,
"step": 157
},
{
"epoch": 1.11,
"learning_rate": 4.6422800058763045e-05,
"loss": 1.3127,
"step": 158
},
{
"epoch": 1.12,
"learning_rate": 4.671661524900838e-05,
"loss": 1.4407,
"step": 159
},
{
"epoch": 1.13,
"learning_rate": 4.701043043925372e-05,
"loss": 1.3185,
"step": 160
},
{
"epoch": 1.13,
"learning_rate": 4.730424562949905e-05,
"loss": 1.2999,
"step": 161
},
{
"epoch": 1.14,
"learning_rate": 4.759806081974439e-05,
"loss": 1.364,
"step": 162
},
{
"epoch": 1.15,
"learning_rate": 4.789187600998972e-05,
"loss": 1.0288,
"step": 163
},
{
"epoch": 1.15,
"learning_rate": 4.818569120023506e-05,
"loss": 1.4062,
"step": 164
},
{
"epoch": 1.16,
"learning_rate": 4.847950639048039e-05,
"loss": 1.3316,
"step": 165
},
{
"epoch": 1.17,
"learning_rate": 4.877332158072573e-05,
"loss": 1.5267,
"step": 166
},
{
"epoch": 1.18,
"learning_rate": 4.9067136770971064e-05,
"loss": 1.3439,
"step": 167
},
{
"epoch": 1.18,
"learning_rate": 4.93609519612164e-05,
"loss": 1.2067,
"step": 168
},
{
"epoch": 1.19,
"learning_rate": 4.9654767151461736e-05,
"loss": 1.2325,
"step": 169
},
{
"epoch": 1.2,
"learning_rate": 4.994858234170707e-05,
"loss": 1.082,
"step": 170
},
{
"epoch": 1.2,
"learning_rate": 5.024239753195241e-05,
"loss": 1.2253,
"step": 171
},
{
"epoch": 1.21,
"learning_rate": 5.053621272219774e-05,
"loss": 1.3254,
"step": 172
},
{
"epoch": 1.22,
"learning_rate": 5.083002791244308e-05,
"loss": 1.4549,
"step": 173
},
{
"epoch": 1.23,
"learning_rate": 5.112384310268842e-05,
"loss": 1.1586,
"step": 174
},
{
"epoch": 1.23,
"learning_rate": 5.1417658292933754e-05,
"loss": 1.3143,
"step": 175
},
{
"epoch": 1.24,
"learning_rate": 5.171147348317908e-05,
"loss": 1.2859,
"step": 176
},
{
"epoch": 1.25,
"learning_rate": 5.2005288673424426e-05,
"loss": 1.4026,
"step": 177
},
{
"epoch": 1.25,
"learning_rate": 5.229910386366976e-05,
"loss": 1.5041,
"step": 178
},
{
"epoch": 1.26,
"learning_rate": 5.259291905391509e-05,
"loss": 1.3742,
"step": 179
},
{
"epoch": 1.27,
"learning_rate": 5.288673424416043e-05,
"loss": 1.2539,
"step": 180
},
{
"epoch": 1.27,
"learning_rate": 5.318054943440577e-05,
"loss": 1.3051,
"step": 181
},
{
"epoch": 1.28,
"learning_rate": 5.347436462465111e-05,
"loss": 1.291,
"step": 182
},
{
"epoch": 1.29,
"learning_rate": 5.3768179814896436e-05,
"loss": 1.2192,
"step": 183
},
{
"epoch": 1.3,
"learning_rate": 5.406199500514177e-05,
"loss": 1.312,
"step": 184
},
{
"epoch": 1.3,
"learning_rate": 5.4355810195387116e-05,
"loss": 1.2242,
"step": 185
},
{
"epoch": 1.31,
"learning_rate": 5.464962538563244e-05,
"loss": 1.4149,
"step": 186
},
{
"epoch": 1.32,
"learning_rate": 5.494344057587778e-05,
"loss": 1.0762,
"step": 187
},
{
"epoch": 1.32,
"learning_rate": 5.523725576612312e-05,
"loss": 1.5351,
"step": 188
},
{
"epoch": 1.33,
"learning_rate": 5.553107095636845e-05,
"loss": 1.2344,
"step": 189
},
{
"epoch": 1.34,
"learning_rate": 5.582488614661378e-05,
"loss": 1.4752,
"step": 190
},
{
"epoch": 1.35,
"learning_rate": 5.6118701336859126e-05,
"loss": 1.4017,
"step": 191
},
{
"epoch": 1.35,
"learning_rate": 5.641251652710446e-05,
"loss": 1.2617,
"step": 192
},
{
"epoch": 1.36,
"learning_rate": 5.670633171734979e-05,
"loss": 1.4561,
"step": 193
},
{
"epoch": 1.37,
"learning_rate": 5.700014690759513e-05,
"loss": 1.2679,
"step": 194
},
{
"epoch": 1.37,
"learning_rate": 5.729396209784047e-05,
"loss": 1.171,
"step": 195
},
{
"epoch": 1.38,
"learning_rate": 5.75877772880858e-05,
"loss": 1.277,
"step": 196
},
{
"epoch": 1.39,
"learning_rate": 5.7881592478331136e-05,
"loss": 1.1648,
"step": 197
},
{
"epoch": 1.39,
"learning_rate": 5.817540766857647e-05,
"loss": 1.4565,
"step": 198
},
{
"epoch": 1.4,
"learning_rate": 5.8469222858821816e-05,
"loss": 1.246,
"step": 199
},
{
"epoch": 1.41,
"learning_rate": 5.8763038049067145e-05,
"loss": 1.2742,
"step": 200
},
{
"epoch": 1.42,
"learning_rate": 5.905685323931248e-05,
"loss": 1.2846,
"step": 201
},
{
"epoch": 1.42,
"learning_rate": 5.935066842955782e-05,
"loss": 1.3021,
"step": 202
},
{
"epoch": 1.43,
"learning_rate": 5.964448361980315e-05,
"loss": 1.4153,
"step": 203
},
{
"epoch": 1.44,
"learning_rate": 5.993829881004849e-05,
"loss": 1.1717,
"step": 204
},
{
"epoch": 1.44,
"learning_rate": 6.0232114000293826e-05,
"loss": 1.3415,
"step": 205
},
{
"epoch": 1.45,
"learning_rate": 6.0525929190539155e-05,
"loss": 1.19,
"step": 206
},
{
"epoch": 1.46,
"learning_rate": 6.081974438078449e-05,
"loss": 1.1805,
"step": 207
},
{
"epoch": 1.46,
"learning_rate": 6.111355957102983e-05,
"loss": 1.3612,
"step": 208
},
{
"epoch": 1.47,
"learning_rate": 6.140737476127517e-05,
"loss": 1.3753,
"step": 209
},
{
"epoch": 1.48,
"learning_rate": 6.17011899515205e-05,
"loss": 1.3976,
"step": 210
},
{
"epoch": 1.49,
"learning_rate": 6.199500514176584e-05,
"loss": 1.2598,
"step": 211
},
{
"epoch": 1.49,
"learning_rate": 6.228882033201118e-05,
"loss": 1.3799,
"step": 212
},
{
"epoch": 1.5,
"learning_rate": 6.25826355222565e-05,
"loss": 1.4741,
"step": 213
},
{
"epoch": 1.51,
"learning_rate": 6.287645071250184e-05,
"loss": 1.2601,
"step": 214
},
{
"epoch": 1.51,
"learning_rate": 6.317026590274719e-05,
"loss": 1.1886,
"step": 215
},
{
"epoch": 1.52,
"learning_rate": 6.346408109299252e-05,
"loss": 1.5535,
"step": 216
},
{
"epoch": 1.53,
"learning_rate": 6.375789628323785e-05,
"loss": 1.3573,
"step": 217
},
{
"epoch": 1.54,
"learning_rate": 6.405171147348318e-05,
"loss": 1.2504,
"step": 218
},
{
"epoch": 1.54,
"learning_rate": 6.434552666372853e-05,
"loss": 1.3212,
"step": 219
},
{
"epoch": 1.55,
"learning_rate": 6.463934185397386e-05,
"loss": 1.177,
"step": 220
},
{
"epoch": 1.56,
"learning_rate": 6.493315704421919e-05,
"loss": 1.1466,
"step": 221
},
{
"epoch": 1.56,
"learning_rate": 6.522697223446453e-05,
"loss": 1.3115,
"step": 222
},
{
"epoch": 1.57,
"learning_rate": 6.552078742470986e-05,
"loss": 1.163,
"step": 223
},
{
"epoch": 1.58,
"learning_rate": 6.58146026149552e-05,
"loss": 1.2517,
"step": 224
},
{
"epoch": 1.58,
"learning_rate": 6.610841780520054e-05,
"loss": 1.4645,
"step": 225
},
{
"epoch": 1.59,
"learning_rate": 6.640223299544587e-05,
"loss": 1.5033,
"step": 226
},
{
"epoch": 1.6,
"learning_rate": 6.669604818569121e-05,
"loss": 1.367,
"step": 227
},
{
"epoch": 1.61,
"learning_rate": 6.698986337593655e-05,
"loss": 1.3936,
"step": 228
},
{
"epoch": 1.61,
"learning_rate": 6.728367856618188e-05,
"loss": 1.2716,
"step": 229
},
{
"epoch": 1.62,
"learning_rate": 6.757749375642722e-05,
"loss": 1.2563,
"step": 230
},
{
"epoch": 1.63,
"learning_rate": 6.787130894667255e-05,
"loss": 1.3766,
"step": 231
},
{
"epoch": 1.63,
"learning_rate": 6.816512413691789e-05,
"loss": 1.1206,
"step": 232
},
{
"epoch": 1.64,
"learning_rate": 6.845893932716323e-05,
"loss": 1.0943,
"step": 233
},
{
"epoch": 1.65,
"learning_rate": 6.875275451740856e-05,
"loss": 1.3942,
"step": 234
},
{
"epoch": 1.65,
"learning_rate": 6.90465697076539e-05,
"loss": 1.2673,
"step": 235
},
{
"epoch": 1.66,
"learning_rate": 6.934038489789923e-05,
"loss": 1.3732,
"step": 236
},
{
"epoch": 1.67,
"learning_rate": 6.963420008814456e-05,
"loss": 1.1018,
"step": 237
},
{
"epoch": 1.68,
"learning_rate": 6.992801527838991e-05,
"loss": 1.1131,
"step": 238
},
{
"epoch": 1.68,
"learning_rate": 7.022183046863524e-05,
"loss": 1.178,
"step": 239
},
{
"epoch": 1.69,
"learning_rate": 7.051564565888057e-05,
"loss": 1.2849,
"step": 240
},
{
"epoch": 1.7,
"learning_rate": 7.08094608491259e-05,
"loss": 1.2847,
"step": 241
},
{
"epoch": 1.7,
"learning_rate": 7.110327603937125e-05,
"loss": 0.9812,
"step": 242
},
{
"epoch": 1.71,
"learning_rate": 7.139709122961659e-05,
"loss": 1.0997,
"step": 243
},
{
"epoch": 1.72,
"learning_rate": 7.169090641986191e-05,
"loss": 1.2575,
"step": 244
},
{
"epoch": 1.73,
"learning_rate": 7.198472161010725e-05,
"loss": 1.3092,
"step": 245
},
{
"epoch": 1.73,
"learning_rate": 7.22785368003526e-05,
"loss": 1.1123,
"step": 246
},
{
"epoch": 1.74,
"learning_rate": 7.257235199059792e-05,
"loss": 1.3787,
"step": 247
},
{
"epoch": 1.75,
"learning_rate": 7.286616718084326e-05,
"loss": 1.2164,
"step": 248
},
{
"epoch": 1.75,
"learning_rate": 7.315998237108859e-05,
"loss": 1.2863,
"step": 249
},
{
"epoch": 1.76,
"learning_rate": 7.345379756133394e-05,
"loss": 1.3277,
"step": 250
},
{
"epoch": 1.77,
"learning_rate": 7.374761275157926e-05,
"loss": 1.2248,
"step": 251
},
{
"epoch": 1.77,
"learning_rate": 7.40414279418246e-05,
"loss": 1.4587,
"step": 252
},
{
"epoch": 1.78,
"learning_rate": 7.433524313206994e-05,
"loss": 1.3788,
"step": 253
},
{
"epoch": 1.79,
"learning_rate": 7.462905832231527e-05,
"loss": 1.3173,
"step": 254
},
{
"epoch": 1.8,
"learning_rate": 7.492287351256061e-05,
"loss": 1.2555,
"step": 255
},
{
"epoch": 1.8,
"learning_rate": 7.521668870280595e-05,
"loss": 1.1032,
"step": 256
},
{
"epoch": 1.81,
"learning_rate": 7.551050389305128e-05,
"loss": 1.271,
"step": 257
},
{
"epoch": 1.82,
"learning_rate": 7.580431908329662e-05,
"loss": 1.1156,
"step": 258
},
{
"epoch": 1.82,
"learning_rate": 7.609813427354195e-05,
"loss": 1.1455,
"step": 259
},
{
"epoch": 1.83,
"learning_rate": 7.639194946378729e-05,
"loss": 1.4856,
"step": 260
},
{
"epoch": 1.84,
"learning_rate": 7.668576465403263e-05,
"loss": 1.0758,
"step": 261
},
{
"epoch": 1.85,
"learning_rate": 7.697957984427796e-05,
"loss": 1.3166,
"step": 262
},
{
"epoch": 1.85,
"learning_rate": 7.72733950345233e-05,
"loss": 1.2848,
"step": 263
},
{
"epoch": 1.86,
"learning_rate": 7.756721022476862e-05,
"loss": 1.2372,
"step": 264
},
{
"epoch": 1.87,
"learning_rate": 7.786102541501397e-05,
"loss": 1.4906,
"step": 265
},
{
"epoch": 1.87,
"learning_rate": 7.815484060525931e-05,
"loss": 1.3265,
"step": 266
},
{
"epoch": 1.88,
"learning_rate": 7.844865579550464e-05,
"loss": 1.1439,
"step": 267
},
{
"epoch": 1.89,
"learning_rate": 7.874247098574997e-05,
"loss": 1.2426,
"step": 268
},
{
"epoch": 1.89,
"learning_rate": 7.903628617599532e-05,
"loss": 1.2351,
"step": 269
},
{
"epoch": 1.9,
"learning_rate": 7.933010136624065e-05,
"loss": 1.3957,
"step": 270
},
{
"epoch": 1.91,
"learning_rate": 7.962391655648597e-05,
"loss": 1.3555,
"step": 271
},
{
"epoch": 1.92,
"learning_rate": 7.991773174673131e-05,
"loss": 1.2983,
"step": 272
},
{
"epoch": 1.92,
"learning_rate": 8.021154693697666e-05,
"loss": 1.2816,
"step": 273
},
{
"epoch": 1.93,
"learning_rate": 8.050536212722198e-05,
"loss": 1.3042,
"step": 274
},
{
"epoch": 1.94,
"learning_rate": 8.079917731746732e-05,
"loss": 1.3328,
"step": 275
},
{
"epoch": 1.94,
"learning_rate": 8.109299250771266e-05,
"loss": 1.2964,
"step": 276
},
{
"epoch": 1.95,
"learning_rate": 8.1386807697958e-05,
"loss": 1.1836,
"step": 277
},
{
"epoch": 1.96,
"learning_rate": 8.168062288820333e-05,
"loss": 1.2704,
"step": 278
},
{
"epoch": 1.96,
"learning_rate": 8.197443807844866e-05,
"loss": 1.3228,
"step": 279
},
{
"epoch": 1.97,
"learning_rate": 8.2268253268694e-05,
"loss": 1.0871,
"step": 280
},
{
"epoch": 1.98,
"learning_rate": 8.256206845893934e-05,
"loss": 1.3571,
"step": 281
},
{
"epoch": 1.99,
"learning_rate": 8.285588364918467e-05,
"loss": 1.4434,
"step": 282
},
{
"epoch": 1.99,
"learning_rate": 8.314969883943001e-05,
"loss": 1.1603,
"step": 283
},
{
"epoch": 2.0,
"learning_rate": 8.344351402967535e-05,
"loss": 1.1378,
"step": 284
},
{
"epoch": 2.01,
"learning_rate": 8.373732921992068e-05,
"loss": 1.4653,
"step": 285
},
{
"epoch": 2.01,
"learning_rate": 8.403114441016602e-05,
"loss": 1.0566,
"step": 286
},
{
"epoch": 2.02,
"learning_rate": 8.432495960041135e-05,
"loss": 1.0505,
"step": 287
},
{
"epoch": 2.03,
"learning_rate": 8.461877479065669e-05,
"loss": 1.3879,
"step": 288
},
{
"epoch": 2.04,
"learning_rate": 8.491258998090203e-05,
"loss": 1.1117,
"step": 289
},
{
"epoch": 2.04,
"learning_rate": 8.520640517114736e-05,
"loss": 1.2697,
"step": 290
},
{
"epoch": 2.05,
"learning_rate": 8.55002203613927e-05,
"loss": 1.3577,
"step": 291
},
{
"epoch": 2.06,
"learning_rate": 8.579403555163803e-05,
"loss": 1.0753,
"step": 292
},
{
"epoch": 2.06,
"learning_rate": 8.608785074188337e-05,
"loss": 1.2443,
"step": 293
},
{
"epoch": 2.07,
"learning_rate": 8.638166593212871e-05,
"loss": 1.047,
"step": 294
},
{
"epoch": 2.08,
"learning_rate": 8.667548112237403e-05,
"loss": 1.2097,
"step": 295
},
{
"epoch": 2.08,
"learning_rate": 8.696929631261938e-05,
"loss": 1.5352,
"step": 296
},
{
"epoch": 2.09,
"learning_rate": 8.726311150286472e-05,
"loss": 1.1819,
"step": 297
},
{
"epoch": 2.1,
"learning_rate": 8.755692669311004e-05,
"loss": 1.1436,
"step": 298
},
{
"epoch": 2.11,
"learning_rate": 8.785074188335537e-05,
"loss": 1.1968,
"step": 299
},
{
"epoch": 2.11,
"learning_rate": 8.814455707360072e-05,
"loss": 1.2693,
"step": 300
},
{
"epoch": 2.12,
"learning_rate": 8.843837226384606e-05,
"loss": 1.3791,
"step": 301
},
{
"epoch": 2.13,
"learning_rate": 8.873218745409138e-05,
"loss": 1.2777,
"step": 302
},
{
"epoch": 2.13,
"learning_rate": 8.902600264433672e-05,
"loss": 1.248,
"step": 303
},
{
"epoch": 2.14,
"learning_rate": 8.931981783458207e-05,
"loss": 1.2936,
"step": 304
},
{
"epoch": 2.15,
"learning_rate": 8.961363302482739e-05,
"loss": 0.9981,
"step": 305
},
{
"epoch": 2.15,
"learning_rate": 8.990744821507273e-05,
"loss": 1.3413,
"step": 306
},
{
"epoch": 2.16,
"learning_rate": 9.020126340531806e-05,
"loss": 1.2833,
"step": 307
},
{
"epoch": 2.17,
"learning_rate": 9.049507859556341e-05,
"loss": 1.4853,
"step": 308
},
{
"epoch": 2.18,
"learning_rate": 9.078889378580874e-05,
"loss": 1.2807,
"step": 309
},
{
"epoch": 2.18,
"learning_rate": 9.108270897605407e-05,
"loss": 1.1716,
"step": 310
},
{
"epoch": 2.19,
"learning_rate": 9.137652416629941e-05,
"loss": 1.1964,
"step": 311
},
{
"epoch": 2.2,
"learning_rate": 9.167033935654475e-05,
"loss": 1.0411,
"step": 312
},
{
"epoch": 2.2,
"learning_rate": 9.196415454679008e-05,
"loss": 1.1827,
"step": 313
},
{
"epoch": 2.21,
"learning_rate": 9.225796973703542e-05,
"loss": 1.2726,
"step": 314
},
{
"epoch": 2.22,
"learning_rate": 9.255178492728075e-05,
"loss": 1.4049,
"step": 315
},
{
"epoch": 2.23,
"learning_rate": 9.284560011752609e-05,
"loss": 1.09,
"step": 316
},
{
"epoch": 2.23,
"learning_rate": 9.313941530777143e-05,
"loss": 1.2616,
"step": 317
},
{
"epoch": 2.24,
"learning_rate": 9.343323049801676e-05,
"loss": 1.2126,
"step": 318
},
{
"epoch": 2.25,
"learning_rate": 9.37270456882621e-05,
"loss": 1.3458,
"step": 319
},
{
"epoch": 2.25,
"learning_rate": 9.402086087850743e-05,
"loss": 1.4486,
"step": 320
},
{
"epoch": 2.26,
"learning_rate": 9.431467606875277e-05,
"loss": 1.2901,
"step": 321
},
{
"epoch": 2.27,
"learning_rate": 9.46084912589981e-05,
"loss": 1.2061,
"step": 322
},
{
"epoch": 2.27,
"learning_rate": 9.490230644924344e-05,
"loss": 1.2587,
"step": 323
},
{
"epoch": 2.28,
"learning_rate": 9.519612163948878e-05,
"loss": 1.2236,
"step": 324
},
{
"epoch": 2.29,
"learning_rate": 9.548993682973412e-05,
"loss": 1.1454,
"step": 325
},
{
"epoch": 2.3,
"learning_rate": 9.578375201997944e-05,
"loss": 1.2289,
"step": 326
},
{
"epoch": 2.3,
"learning_rate": 9.607756721022479e-05,
"loss": 1.1934,
"step": 327
},
{
"epoch": 2.31,
"learning_rate": 9.637138240047012e-05,
"loss": 1.3799,
"step": 328
},
{
"epoch": 2.32,
"learning_rate": 9.666519759071545e-05,
"loss": 1.0258,
"step": 329
},
{
"epoch": 2.32,
"learning_rate": 9.695901278096078e-05,
"loss": 1.4939,
"step": 330
},
{
"epoch": 2.33,
"learning_rate": 9.725282797120613e-05,
"loss": 1.186,
"step": 331
},
{
"epoch": 2.34,
"learning_rate": 9.754664316145146e-05,
"loss": 1.4199,
"step": 332
},
{
"epoch": 2.35,
"learning_rate": 9.784045835169679e-05,
"loss": 1.3008,
"step": 333
},
{
"epoch": 2.35,
"learning_rate": 9.813427354194213e-05,
"loss": 1.1806,
"step": 334
},
{
"epoch": 2.36,
"learning_rate": 9.842808873218748e-05,
"loss": 1.3952,
"step": 335
},
{
"epoch": 2.37,
"learning_rate": 9.87219039224328e-05,
"loss": 1.2403,
"step": 336
},
{
"epoch": 2.37,
"learning_rate": 9.901571911267814e-05,
"loss": 1.1104,
"step": 337
},
{
"epoch": 2.38,
"learning_rate": 9.930953430292347e-05,
"loss": 1.2103,
"step": 338
},
{
"epoch": 2.39,
"learning_rate": 9.960334949316881e-05,
"loss": 1.0564,
"step": 339
},
{
"epoch": 2.39,
"learning_rate": 9.989716468341415e-05,
"loss": 1.3984,
"step": 340
},
{
"epoch": 2.4,
"learning_rate": 9.99999963334151e-05,
"loss": 1.2017,
"step": 341
},
{
"epoch": 2.41,
"learning_rate": 9.999997637330952e-05,
"loss": 1.2251,
"step": 342
},
{
"epoch": 2.42,
"learning_rate": 9.99999390565975e-05,
"loss": 1.2006,
"step": 343
},
{
"epoch": 2.42,
"learning_rate": 9.999988438329202e-05,
"loss": 1.236,
"step": 344
},
{
"epoch": 2.43,
"learning_rate": 9.999981235341205e-05,
"loss": 1.3684,
"step": 345
},
{
"epoch": 2.44,
"learning_rate": 9.999972296698259e-05,
"loss": 1.1272,
"step": 346
},
{
"epoch": 2.44,
"learning_rate": 9.999961622403467e-05,
"loss": 1.2259,
"step": 347
},
{
"epoch": 2.45,
"learning_rate": 9.999949212460534e-05,
"loss": 1.1502,
"step": 348
},
{
"epoch": 2.46,
"learning_rate": 9.999935066873766e-05,
"loss": 1.1134,
"step": 349
},
{
"epoch": 2.46,
"learning_rate": 9.999919185648079e-05,
"loss": 1.306,
"step": 350
},
{
"epoch": 2.47,
"learning_rate": 9.99990156878898e-05,
"loss": 1.3023,
"step": 351
},
{
"epoch": 2.48,
"learning_rate": 9.999882216302589e-05,
"loss": 1.3383,
"step": 352
},
{
"epoch": 2.49,
"learning_rate": 9.99986112819562e-05,
"loss": 1.2076,
"step": 353
},
{
"epoch": 2.49,
"learning_rate": 9.999838304475396e-05,
"loss": 1.3298,
"step": 354
},
{
"epoch": 2.5,
"learning_rate": 9.99981374514984e-05,
"loss": 1.4117,
"step": 355
},
{
"epoch": 2.51,
"learning_rate": 9.999787450227475e-05,
"loss": 1.2157,
"step": 356
},
{
"epoch": 2.51,
"learning_rate": 9.999759419717429e-05,
"loss": 1.1052,
"step": 357
},
{
"epoch": 2.52,
"learning_rate": 9.999729653629435e-05,
"loss": 1.4931,
"step": 358
},
{
"epoch": 2.53,
"learning_rate": 9.999698151973824e-05,
"loss": 1.2367,
"step": 359
},
{
"epoch": 2.54,
"learning_rate": 9.999664914761532e-05,
"loss": 1.1763,
"step": 360
},
{
"epoch": 2.54,
"learning_rate": 9.999629942004095e-05,
"loss": 1.2708,
"step": 361
},
{
"epoch": 2.55,
"learning_rate": 9.999593233713654e-05,
"loss": 1.114,
"step": 362
},
{
"epoch": 2.56,
"learning_rate": 9.999554789902954e-05,
"loss": 1.066,
"step": 363
},
{
"epoch": 2.56,
"learning_rate": 9.999514610585336e-05,
"loss": 1.223,
"step": 364
},
{
"epoch": 2.57,
"learning_rate": 9.99947269577475e-05,
"loss": 1.1002,
"step": 365
},
{
"epoch": 2.58,
"learning_rate": 9.999429045485746e-05,
"loss": 1.1679,
"step": 366
},
{
"epoch": 2.58,
"learning_rate": 9.999383659733476e-05,
"loss": 1.3512,
"step": 367
},
{
"epoch": 2.59,
"learning_rate": 9.999336538533694e-05,
"loss": 1.445,
"step": 368
},
{
"epoch": 2.6,
"learning_rate": 9.99928768190276e-05,
"loss": 1.2979,
"step": 369
},
{
"epoch": 2.61,
"learning_rate": 9.99923708985763e-05,
"loss": 1.3093,
"step": 370
},
{
"epoch": 2.61,
"learning_rate": 9.999184762415869e-05,
"loss": 1.2169,
"step": 371
},
{
"epoch": 2.62,
"learning_rate": 9.99913069959564e-05,
"loss": 1.1727,
"step": 372
},
{
"epoch": 2.63,
"learning_rate": 9.99907490141571e-05,
"loss": 1.2972,
"step": 373
},
{
"epoch": 2.63,
"learning_rate": 9.999017367895449e-05,
"loss": 1.0836,
"step": 374
},
{
"epoch": 2.64,
"learning_rate": 9.99895809905483e-05,
"loss": 1.0422,
"step": 375
},
{
"epoch": 2.65,
"learning_rate": 9.998897094914424e-05,
"loss": 1.3384,
"step": 376
},
{
"epoch": 2.65,
"learning_rate": 9.99883435549541e-05,
"loss": 1.19,
"step": 377
},
{
"epoch": 2.66,
"learning_rate": 9.998769880819565e-05,
"loss": 1.2808,
"step": 378
},
{
"epoch": 2.67,
"learning_rate": 9.998703670909273e-05,
"loss": 1.0036,
"step": 379
},
{
"epoch": 2.68,
"learning_rate": 9.998635725787515e-05,
"loss": 1.0307,
"step": 380
},
{
"epoch": 2.68,
"learning_rate": 9.998566045477877e-05,
"loss": 1.1171,
"step": 381
},
{
"epoch": 2.69,
"learning_rate": 9.99849463000455e-05,
"loss": 1.2063,
"step": 382
},
{
"epoch": 2.7,
"learning_rate": 9.998421479392322e-05,
"loss": 1.2263,
"step": 383
},
{
"epoch": 2.7,
"learning_rate": 9.998346593666587e-05,
"loss": 0.8941,
"step": 384
},
{
"epoch": 2.71,
"learning_rate": 9.99826997285334e-05,
"loss": 1.0527,
"step": 385
},
{
"epoch": 2.72,
"learning_rate": 9.998191616979178e-05,
"loss": 1.2203,
"step": 386
},
{
"epoch": 2.73,
"learning_rate": 9.9981115260713e-05,
"loss": 1.2373,
"step": 387
},
{
"epoch": 2.73,
"learning_rate": 9.998029700157512e-05,
"loss": 1.0372,
"step": 388
},
{
"epoch": 2.74,
"learning_rate": 9.997946139266214e-05,
"loss": 1.3097,
"step": 389
},
{
"epoch": 2.75,
"learning_rate": 9.997860843426416e-05,
"loss": 1.1405,
"step": 390
},
{
"epoch": 2.75,
"learning_rate": 9.997773812667726e-05,
"loss": 1.2198,
"step": 391
},
{
"epoch": 2.76,
"learning_rate": 9.997685047020353e-05,
"loss": 1.2622,
"step": 392
},
{
"epoch": 2.77,
"learning_rate": 9.997594546515112e-05,
"loss": 1.1772,
"step": 393
},
{
"epoch": 2.77,
"learning_rate": 9.997502311183421e-05,
"loss": 1.3354,
"step": 394
},
{
"epoch": 2.78,
"learning_rate": 9.997408341057295e-05,
"loss": 1.2733,
"step": 395
},
{
"epoch": 2.79,
"learning_rate": 9.997312636169354e-05,
"loss": 1.2603,
"step": 396
},
{
"epoch": 2.8,
"learning_rate": 9.99721519655282e-05,
"loss": 1.182,
"step": 397
},
{
"epoch": 2.8,
"learning_rate": 9.997116022241519e-05,
"loss": 1.001,
"step": 398
},
{
"epoch": 2.81,
"learning_rate": 9.997015113269878e-05,
"loss": 1.1661,
"step": 399
},
{
"epoch": 2.82,
"learning_rate": 9.996912469672925e-05,
"loss": 1.0415,
"step": 400
},
{
"epoch": 2.82,
"learning_rate": 9.99680809148629e-05,
"loss": 1.0874,
"step": 401
},
{
"epoch": 2.83,
"learning_rate": 9.996701978746207e-05,
"loss": 1.3411,
"step": 402
},
{
"epoch": 2.84,
"learning_rate": 9.996594131489511e-05,
"loss": 1.0126,
"step": 403
},
{
"epoch": 2.85,
"learning_rate": 9.99648454975364e-05,
"loss": 1.2103,
"step": 404
},
{
"epoch": 2.85,
"learning_rate": 9.996373233576631e-05,
"loss": 1.1725,
"step": 405
},
{
"epoch": 2.86,
"learning_rate": 9.996260182997126e-05,
"loss": 1.1493,
"step": 406
},
{
"epoch": 2.87,
"learning_rate": 9.996145398054372e-05,
"loss": 1.3638,
"step": 407
},
{
"epoch": 2.87,
"learning_rate": 9.99602887878821e-05,
"loss": 1.2561,
"step": 408
},
{
"epoch": 2.88,
"learning_rate": 9.995910625239092e-05,
"loss": 1.0412,
"step": 409
},
{
"epoch": 2.89,
"learning_rate": 9.995790637448064e-05,
"loss": 1.1855,
"step": 410
},
{
"epoch": 2.89,
"learning_rate": 9.995668915456778e-05,
"loss": 1.1156,
"step": 411
},
{
"epoch": 2.9,
"learning_rate": 9.99554545930749e-05,
"loss": 1.317,
"step": 412
},
{
"epoch": 2.91,
"learning_rate": 9.995420269043053e-05,
"loss": 1.3018,
"step": 413
},
{
"epoch": 2.92,
"learning_rate": 9.995293344706927e-05,
"loss": 1.2392,
"step": 414
},
{
"epoch": 2.92,
"learning_rate": 9.99516468634317e-05,
"loss": 1.2128,
"step": 415
},
{
"epoch": 2.93,
"learning_rate": 9.995034293996442e-05,
"loss": 1.173,
"step": 416
},
{
"epoch": 2.94,
"learning_rate": 9.99490216771201e-05,
"loss": 1.2461,
"step": 417
},
{
"epoch": 2.94,
"learning_rate": 9.994768307535736e-05,
"loss": 1.1709,
"step": 418
},
{
"epoch": 2.95,
"learning_rate": 9.99463271351409e-05,
"loss": 1.0691,
"step": 419
},
{
"epoch": 2.96,
"learning_rate": 9.99449538569414e-05,
"loss": 1.1749,
"step": 420
},
{
"epoch": 2.96,
"learning_rate": 9.994356324123555e-05,
"loss": 1.2748,
"step": 421
},
{
"epoch": 2.97,
"learning_rate": 9.994215528850611e-05,
"loss": 0.9984,
"step": 422
},
{
"epoch": 2.98,
"learning_rate": 9.994072999924178e-05,
"loss": 1.288,
"step": 423
},
{
"epoch": 2.99,
"learning_rate": 9.993928737393738e-05,
"loss": 1.3438,
"step": 424
},
{
"epoch": 2.99,
"learning_rate": 9.993782741309366e-05,
"loss": 1.0476,
"step": 425
},
{
"epoch": 3.0,
"learning_rate": 9.993635011721743e-05,
"loss": 1.0269,
"step": 426
}
],
"logging_steps": 1,
"max_steps": 710,
"num_train_epochs": 5,
"save_steps": 500,
"total_flos": 3.347312943090893e+16,
"trial_name": null,
"trial_params": null
}