t5-small-e2e-qa-full / trainer_state.json
longcld's picture
loss 2.7
b99e558
raw
history blame
No virus
71.5 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.3538515843492895,
"global_step": 58500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 9.989138698816118e-05,
"loss": 9.1158,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 9.978277397632236e-05,
"loss": 5.8307,
"step": 200
},
{
"epoch": 0.03,
"learning_rate": 9.967416096448356e-05,
"loss": 5.4251,
"step": 300
},
{
"epoch": 0.04,
"learning_rate": 9.956554795264474e-05,
"loss": 5.2931,
"step": 400
},
{
"epoch": 0.05,
"learning_rate": 9.945693494080592e-05,
"loss": 5.0326,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 9.93483219289671e-05,
"loss": 4.8397,
"step": 600
},
{
"epoch": 0.08,
"learning_rate": 9.923970891712828e-05,
"loss": 4.9379,
"step": 700
},
{
"epoch": 0.09,
"learning_rate": 9.913109590528946e-05,
"loss": 4.6978,
"step": 800
},
{
"epoch": 0.1,
"learning_rate": 9.902248289345064e-05,
"loss": 4.7727,
"step": 900
},
{
"epoch": 0.11,
"learning_rate": 9.891386988161182e-05,
"loss": 4.5856,
"step": 1000
},
{
"epoch": 0.12,
"learning_rate": 9.8805256869773e-05,
"loss": 4.5636,
"step": 1100
},
{
"epoch": 0.13,
"learning_rate": 9.869664385793419e-05,
"loss": 4.4165,
"step": 1200
},
{
"epoch": 0.14,
"learning_rate": 9.858803084609537e-05,
"loss": 4.4187,
"step": 1300
},
{
"epoch": 0.15,
"learning_rate": 9.847941783425655e-05,
"loss": 4.3553,
"step": 1400
},
{
"epoch": 0.16,
"learning_rate": 9.837080482241773e-05,
"loss": 4.3312,
"step": 1500
},
{
"epoch": 0.17,
"learning_rate": 9.826219181057891e-05,
"loss": 4.3841,
"step": 1600
},
{
"epoch": 0.18,
"learning_rate": 9.81535787987401e-05,
"loss": 4.3428,
"step": 1700
},
{
"epoch": 0.2,
"learning_rate": 9.804496578690128e-05,
"loss": 4.3044,
"step": 1800
},
{
"epoch": 0.21,
"learning_rate": 9.793635277506245e-05,
"loss": 4.2581,
"step": 1900
},
{
"epoch": 0.22,
"learning_rate": 9.782773976322363e-05,
"loss": 4.1639,
"step": 2000
},
{
"epoch": 0.23,
"learning_rate": 9.771912675138481e-05,
"loss": 4.231,
"step": 2100
},
{
"epoch": 0.24,
"learning_rate": 9.7610513739546e-05,
"loss": 4.1087,
"step": 2200
},
{
"epoch": 0.25,
"learning_rate": 9.750190072770718e-05,
"loss": 4.2166,
"step": 2300
},
{
"epoch": 0.26,
"learning_rate": 9.739328771586836e-05,
"loss": 4.0581,
"step": 2400
},
{
"epoch": 0.27,
"learning_rate": 9.728467470402954e-05,
"loss": 4.0475,
"step": 2500
},
{
"epoch": 0.28,
"learning_rate": 9.717606169219074e-05,
"loss": 4.0477,
"step": 2600
},
{
"epoch": 0.29,
"learning_rate": 9.706744868035192e-05,
"loss": 4.2303,
"step": 2700
},
{
"epoch": 0.3,
"learning_rate": 9.69588356685131e-05,
"loss": 4.0649,
"step": 2800
},
{
"epoch": 0.31,
"learning_rate": 9.685022265667428e-05,
"loss": 4.0515,
"step": 2900
},
{
"epoch": 0.33,
"learning_rate": 9.674160964483546e-05,
"loss": 4.0151,
"step": 3000
},
{
"epoch": 0.34,
"learning_rate": 9.663299663299664e-05,
"loss": 3.9701,
"step": 3100
},
{
"epoch": 0.35,
"learning_rate": 9.652438362115782e-05,
"loss": 3.8968,
"step": 3200
},
{
"epoch": 0.36,
"learning_rate": 9.6415770609319e-05,
"loss": 4.0167,
"step": 3300
},
{
"epoch": 0.37,
"learning_rate": 9.630715759748018e-05,
"loss": 4.0087,
"step": 3400
},
{
"epoch": 0.38,
"learning_rate": 9.619854458564136e-05,
"loss": 4.0071,
"step": 3500
},
{
"epoch": 0.39,
"learning_rate": 9.608993157380255e-05,
"loss": 4.0202,
"step": 3600
},
{
"epoch": 0.4,
"learning_rate": 9.598131856196373e-05,
"loss": 4.069,
"step": 3700
},
{
"epoch": 0.41,
"learning_rate": 9.587270555012491e-05,
"loss": 3.9612,
"step": 3800
},
{
"epoch": 0.42,
"learning_rate": 9.576409253828609e-05,
"loss": 3.8818,
"step": 3900
},
{
"epoch": 0.43,
"learning_rate": 9.565547952644728e-05,
"loss": 3.9479,
"step": 4000
},
{
"epoch": 0.45,
"learning_rate": 9.554686651460846e-05,
"loss": 3.8681,
"step": 4100
},
{
"epoch": 0.46,
"learning_rate": 9.543825350276964e-05,
"loss": 3.8267,
"step": 4200
},
{
"epoch": 0.47,
"learning_rate": 9.532964049093082e-05,
"loss": 3.9206,
"step": 4300
},
{
"epoch": 0.48,
"learning_rate": 9.5221027479092e-05,
"loss": 3.8918,
"step": 4400
},
{
"epoch": 0.49,
"learning_rate": 9.511241446725318e-05,
"loss": 3.7453,
"step": 4500
},
{
"epoch": 0.5,
"learning_rate": 9.500380145541436e-05,
"loss": 3.8368,
"step": 4600
},
{
"epoch": 0.51,
"learning_rate": 9.489518844357554e-05,
"loss": 3.8248,
"step": 4700
},
{
"epoch": 0.52,
"learning_rate": 9.478657543173672e-05,
"loss": 3.8242,
"step": 4800
},
{
"epoch": 0.53,
"learning_rate": 9.46779624198979e-05,
"loss": 3.8506,
"step": 4900
},
{
"epoch": 0.54,
"learning_rate": 9.45693494080591e-05,
"loss": 3.878,
"step": 5000
},
{
"epoch": 0.55,
"learning_rate": 9.446073639622027e-05,
"loss": 3.7932,
"step": 5100
},
{
"epoch": 0.56,
"learning_rate": 9.435212338438145e-05,
"loss": 3.8755,
"step": 5200
},
{
"epoch": 0.58,
"learning_rate": 9.424351037254263e-05,
"loss": 3.7838,
"step": 5300
},
{
"epoch": 0.59,
"learning_rate": 9.413489736070383e-05,
"loss": 3.856,
"step": 5400
},
{
"epoch": 0.6,
"learning_rate": 9.402628434886501e-05,
"loss": 3.6716,
"step": 5500
},
{
"epoch": 0.61,
"learning_rate": 9.391767133702617e-05,
"loss": 3.7467,
"step": 5600
},
{
"epoch": 0.62,
"learning_rate": 9.380905832518735e-05,
"loss": 3.6716,
"step": 5700
},
{
"epoch": 0.63,
"learning_rate": 9.370044531334853e-05,
"loss": 3.6479,
"step": 5800
},
{
"epoch": 0.64,
"learning_rate": 9.359183230150973e-05,
"loss": 3.6937,
"step": 5900
},
{
"epoch": 0.65,
"learning_rate": 9.348321928967091e-05,
"loss": 3.7943,
"step": 6000
},
{
"epoch": 0.66,
"learning_rate": 9.337460627783209e-05,
"loss": 3.6678,
"step": 6100
},
{
"epoch": 0.67,
"learning_rate": 9.326599326599327e-05,
"loss": 3.6824,
"step": 6200
},
{
"epoch": 0.68,
"learning_rate": 9.315738025415445e-05,
"loss": 3.6794,
"step": 6300
},
{
"epoch": 0.7,
"learning_rate": 9.304876724231564e-05,
"loss": 3.7675,
"step": 6400
},
{
"epoch": 0.71,
"learning_rate": 9.294015423047682e-05,
"loss": 3.6341,
"step": 6500
},
{
"epoch": 0.72,
"learning_rate": 9.2831541218638e-05,
"loss": 3.6805,
"step": 6600
},
{
"epoch": 0.73,
"learning_rate": 9.272292820679918e-05,
"loss": 3.6479,
"step": 6700
},
{
"epoch": 0.74,
"learning_rate": 9.261431519496036e-05,
"loss": 3.6145,
"step": 6800
},
{
"epoch": 0.75,
"learning_rate": 9.250570218312154e-05,
"loss": 3.5535,
"step": 6900
},
{
"epoch": 0.76,
"learning_rate": 9.239708917128272e-05,
"loss": 3.6772,
"step": 7000
},
{
"epoch": 0.77,
"learning_rate": 9.22884761594439e-05,
"loss": 3.6542,
"step": 7100
},
{
"epoch": 0.78,
"learning_rate": 9.217986314760508e-05,
"loss": 3.5783,
"step": 7200
},
{
"epoch": 0.79,
"learning_rate": 9.207125013576627e-05,
"loss": 3.5812,
"step": 7300
},
{
"epoch": 0.8,
"learning_rate": 9.196263712392745e-05,
"loss": 3.5677,
"step": 7400
},
{
"epoch": 0.81,
"learning_rate": 9.185402411208863e-05,
"loss": 3.5258,
"step": 7500
},
{
"epoch": 0.83,
"learning_rate": 9.174541110024981e-05,
"loss": 3.618,
"step": 7600
},
{
"epoch": 0.84,
"learning_rate": 9.163679808841099e-05,
"loss": 3.7178,
"step": 7700
},
{
"epoch": 0.85,
"learning_rate": 9.152818507657219e-05,
"loss": 3.6502,
"step": 7800
},
{
"epoch": 0.86,
"learning_rate": 9.141957206473337e-05,
"loss": 3.6745,
"step": 7900
},
{
"epoch": 0.87,
"learning_rate": 9.131095905289455e-05,
"loss": 3.6149,
"step": 8000
},
{
"epoch": 0.88,
"learning_rate": 9.120234604105573e-05,
"loss": 3.5411,
"step": 8100
},
{
"epoch": 0.89,
"learning_rate": 9.10937330292169e-05,
"loss": 3.4925,
"step": 8200
},
{
"epoch": 0.9,
"learning_rate": 9.098512001737809e-05,
"loss": 3.575,
"step": 8300
},
{
"epoch": 0.91,
"learning_rate": 9.087650700553927e-05,
"loss": 3.4702,
"step": 8400
},
{
"epoch": 0.92,
"learning_rate": 9.076789399370045e-05,
"loss": 3.5782,
"step": 8500
},
{
"epoch": 0.93,
"learning_rate": 9.065928098186163e-05,
"loss": 3.6253,
"step": 8600
},
{
"epoch": 0.94,
"learning_rate": 9.055066797002282e-05,
"loss": 3.475,
"step": 8700
},
{
"epoch": 0.96,
"learning_rate": 9.0442054958184e-05,
"loss": 3.5008,
"step": 8800
},
{
"epoch": 0.97,
"learning_rate": 9.033344194634518e-05,
"loss": 3.5272,
"step": 8900
},
{
"epoch": 0.98,
"learning_rate": 9.022482893450636e-05,
"loss": 3.5002,
"step": 9000
},
{
"epoch": 0.99,
"learning_rate": 9.011621592266754e-05,
"loss": 3.611,
"step": 9100
},
{
"epoch": 1.0,
"learning_rate": 9.000760291082873e-05,
"loss": 3.607,
"step": 9200
},
{
"epoch": 1.01,
"learning_rate": 8.98989898989899e-05,
"loss": 3.494,
"step": 9300
},
{
"epoch": 1.02,
"learning_rate": 8.979037688715108e-05,
"loss": 3.542,
"step": 9400
},
{
"epoch": 1.03,
"learning_rate": 8.968176387531226e-05,
"loss": 3.468,
"step": 9500
},
{
"epoch": 1.04,
"learning_rate": 8.957315086347344e-05,
"loss": 3.3716,
"step": 9600
},
{
"epoch": 1.05,
"learning_rate": 8.946453785163463e-05,
"loss": 3.593,
"step": 9700
},
{
"epoch": 1.06,
"learning_rate": 8.935592483979581e-05,
"loss": 3.3818,
"step": 9800
},
{
"epoch": 1.08,
"learning_rate": 8.924731182795699e-05,
"loss": 3.3353,
"step": 9900
},
{
"epoch": 1.09,
"learning_rate": 8.913869881611817e-05,
"loss": 3.4241,
"step": 10000
},
{
"epoch": 1.1,
"learning_rate": 8.903008580427936e-05,
"loss": 3.507,
"step": 10100
},
{
"epoch": 1.11,
"learning_rate": 8.892147279244054e-05,
"loss": 3.4623,
"step": 10200
},
{
"epoch": 1.12,
"learning_rate": 8.881285978060172e-05,
"loss": 3.443,
"step": 10300
},
{
"epoch": 1.13,
"learning_rate": 8.87042467687629e-05,
"loss": 3.4692,
"step": 10400
},
{
"epoch": 1.14,
"learning_rate": 8.859563375692408e-05,
"loss": 3.424,
"step": 10500
},
{
"epoch": 1.15,
"learning_rate": 8.848702074508526e-05,
"loss": 3.3139,
"step": 10600
},
{
"epoch": 1.16,
"learning_rate": 8.837840773324644e-05,
"loss": 3.3319,
"step": 10700
},
{
"epoch": 1.17,
"learning_rate": 8.826979472140762e-05,
"loss": 3.5363,
"step": 10800
},
{
"epoch": 1.18,
"learning_rate": 8.81611817095688e-05,
"loss": 3.4598,
"step": 10900
},
{
"epoch": 1.19,
"learning_rate": 8.805256869772998e-05,
"loss": 3.4883,
"step": 11000
},
{
"epoch": 1.21,
"learning_rate": 8.794395568589118e-05,
"loss": 3.2751,
"step": 11100
},
{
"epoch": 1.22,
"learning_rate": 8.783534267405236e-05,
"loss": 3.3685,
"step": 11200
},
{
"epoch": 1.23,
"learning_rate": 8.772672966221354e-05,
"loss": 3.4111,
"step": 11300
},
{
"epoch": 1.24,
"learning_rate": 8.761811665037472e-05,
"loss": 3.3984,
"step": 11400
},
{
"epoch": 1.25,
"learning_rate": 8.750950363853591e-05,
"loss": 3.4246,
"step": 11500
},
{
"epoch": 1.26,
"learning_rate": 8.740089062669709e-05,
"loss": 3.2751,
"step": 11600
},
{
"epoch": 1.27,
"learning_rate": 8.729227761485827e-05,
"loss": 3.4855,
"step": 11700
},
{
"epoch": 1.28,
"learning_rate": 8.718366460301945e-05,
"loss": 3.4712,
"step": 11800
},
{
"epoch": 1.29,
"learning_rate": 8.707505159118062e-05,
"loss": 3.3221,
"step": 11900
},
{
"epoch": 1.3,
"learning_rate": 8.696643857934181e-05,
"loss": 3.35,
"step": 12000
},
{
"epoch": 1.31,
"learning_rate": 8.685782556750299e-05,
"loss": 3.3502,
"step": 12100
},
{
"epoch": 1.33,
"learning_rate": 8.674921255566417e-05,
"loss": 3.2973,
"step": 12200
},
{
"epoch": 1.34,
"learning_rate": 8.664059954382535e-05,
"loss": 3.3982,
"step": 12300
},
{
"epoch": 1.35,
"learning_rate": 8.653198653198653e-05,
"loss": 3.3566,
"step": 12400
},
{
"epoch": 1.36,
"learning_rate": 8.642337352014772e-05,
"loss": 3.3709,
"step": 12500
},
{
"epoch": 1.37,
"learning_rate": 8.63147605083089e-05,
"loss": 3.4209,
"step": 12600
},
{
"epoch": 1.38,
"learning_rate": 8.620614749647008e-05,
"loss": 3.2738,
"step": 12700
},
{
"epoch": 1.39,
"learning_rate": 8.609753448463126e-05,
"loss": 3.4105,
"step": 12800
},
{
"epoch": 1.4,
"learning_rate": 8.598892147279246e-05,
"loss": 3.2289,
"step": 12900
},
{
"epoch": 1.41,
"learning_rate": 8.588030846095362e-05,
"loss": 3.2912,
"step": 13000
},
{
"epoch": 1.42,
"learning_rate": 8.57716954491148e-05,
"loss": 3.3281,
"step": 13100
},
{
"epoch": 1.43,
"learning_rate": 8.566308243727598e-05,
"loss": 3.4349,
"step": 13200
},
{
"epoch": 1.44,
"learning_rate": 8.555446942543716e-05,
"loss": 3.2977,
"step": 13300
},
{
"epoch": 1.46,
"learning_rate": 8.544585641359836e-05,
"loss": 3.4786,
"step": 13400
},
{
"epoch": 1.47,
"learning_rate": 8.533724340175954e-05,
"loss": 3.2424,
"step": 13500
},
{
"epoch": 1.48,
"learning_rate": 8.522863038992072e-05,
"loss": 3.2443,
"step": 13600
},
{
"epoch": 1.49,
"learning_rate": 8.51200173780819e-05,
"loss": 3.3818,
"step": 13700
},
{
"epoch": 1.5,
"learning_rate": 8.501140436624308e-05,
"loss": 3.3809,
"step": 13800
},
{
"epoch": 1.51,
"learning_rate": 8.490279135440427e-05,
"loss": 3.2922,
"step": 13900
},
{
"epoch": 1.52,
"learning_rate": 8.479417834256545e-05,
"loss": 3.2152,
"step": 14000
},
{
"epoch": 1.53,
"learning_rate": 8.468556533072663e-05,
"loss": 3.3144,
"step": 14100
},
{
"epoch": 1.54,
"learning_rate": 8.457695231888781e-05,
"loss": 3.266,
"step": 14200
},
{
"epoch": 1.55,
"learning_rate": 8.446833930704899e-05,
"loss": 3.3113,
"step": 14300
},
{
"epoch": 1.56,
"learning_rate": 8.435972629521017e-05,
"loss": 3.2186,
"step": 14400
},
{
"epoch": 1.57,
"learning_rate": 8.425111328337135e-05,
"loss": 3.2034,
"step": 14500
},
{
"epoch": 1.59,
"learning_rate": 8.414250027153253e-05,
"loss": 3.2229,
"step": 14600
},
{
"epoch": 1.6,
"learning_rate": 8.403388725969371e-05,
"loss": 3.2563,
"step": 14700
},
{
"epoch": 1.61,
"learning_rate": 8.39252742478549e-05,
"loss": 3.2689,
"step": 14800
},
{
"epoch": 1.62,
"learning_rate": 8.381666123601608e-05,
"loss": 3.1917,
"step": 14900
},
{
"epoch": 1.63,
"learning_rate": 8.370804822417726e-05,
"loss": 3.3305,
"step": 15000
},
{
"epoch": 1.64,
"learning_rate": 8.359943521233844e-05,
"loss": 3.2577,
"step": 15100
},
{
"epoch": 1.65,
"learning_rate": 8.349082220049962e-05,
"loss": 3.3375,
"step": 15200
},
{
"epoch": 1.66,
"learning_rate": 8.338220918866081e-05,
"loss": 3.2056,
"step": 15300
},
{
"epoch": 1.67,
"learning_rate": 8.3273596176822e-05,
"loss": 3.2336,
"step": 15400
},
{
"epoch": 1.68,
"learning_rate": 8.316498316498317e-05,
"loss": 3.2219,
"step": 15500
},
{
"epoch": 1.69,
"learning_rate": 8.305637015314434e-05,
"loss": 3.3085,
"step": 15600
},
{
"epoch": 1.71,
"learning_rate": 8.294775714130553e-05,
"loss": 3.2312,
"step": 15700
},
{
"epoch": 1.72,
"learning_rate": 8.283914412946671e-05,
"loss": 3.2054,
"step": 15800
},
{
"epoch": 1.73,
"learning_rate": 8.27305311176279e-05,
"loss": 3.3518,
"step": 15900
},
{
"epoch": 1.74,
"learning_rate": 8.262191810578907e-05,
"loss": 3.22,
"step": 16000
},
{
"epoch": 1.75,
"learning_rate": 8.251330509395025e-05,
"loss": 3.2564,
"step": 16100
},
{
"epoch": 1.76,
"learning_rate": 8.240469208211145e-05,
"loss": 3.181,
"step": 16200
},
{
"epoch": 1.77,
"learning_rate": 8.229607907027263e-05,
"loss": 3.3038,
"step": 16300
},
{
"epoch": 1.78,
"learning_rate": 8.218746605843381e-05,
"loss": 3.1689,
"step": 16400
},
{
"epoch": 1.79,
"learning_rate": 8.207885304659499e-05,
"loss": 3.2781,
"step": 16500
},
{
"epoch": 1.8,
"learning_rate": 8.197024003475617e-05,
"loss": 3.1925,
"step": 16600
},
{
"epoch": 1.81,
"learning_rate": 8.186162702291735e-05,
"loss": 3.2514,
"step": 16700
},
{
"epoch": 1.82,
"learning_rate": 8.175301401107853e-05,
"loss": 3.2324,
"step": 16800
},
{
"epoch": 1.84,
"learning_rate": 8.16444009992397e-05,
"loss": 3.1748,
"step": 16900
},
{
"epoch": 1.85,
"learning_rate": 8.153578798740089e-05,
"loss": 3.2089,
"step": 17000
},
{
"epoch": 1.86,
"learning_rate": 8.142717497556208e-05,
"loss": 3.156,
"step": 17100
},
{
"epoch": 1.87,
"learning_rate": 8.131856196372326e-05,
"loss": 3.2313,
"step": 17200
},
{
"epoch": 1.88,
"learning_rate": 8.120994895188444e-05,
"loss": 3.2406,
"step": 17300
},
{
"epoch": 1.89,
"learning_rate": 8.110133594004562e-05,
"loss": 3.2273,
"step": 17400
},
{
"epoch": 1.9,
"learning_rate": 8.09927229282068e-05,
"loss": 3.1781,
"step": 17500
},
{
"epoch": 1.91,
"learning_rate": 8.088410991636799e-05,
"loss": 3.1407,
"step": 17600
},
{
"epoch": 1.92,
"learning_rate": 8.077549690452917e-05,
"loss": 3.1266,
"step": 17700
},
{
"epoch": 1.93,
"learning_rate": 8.066688389269035e-05,
"loss": 3.1861,
"step": 17800
},
{
"epoch": 1.94,
"learning_rate": 8.055827088085153e-05,
"loss": 3.1604,
"step": 17900
},
{
"epoch": 1.96,
"learning_rate": 8.044965786901271e-05,
"loss": 3.1591,
"step": 18000
},
{
"epoch": 1.97,
"learning_rate": 8.034104485717389e-05,
"loss": 3.2045,
"step": 18100
},
{
"epoch": 1.98,
"learning_rate": 8.023243184533507e-05,
"loss": 3.1123,
"step": 18200
},
{
"epoch": 1.99,
"learning_rate": 8.012381883349625e-05,
"loss": 3.2466,
"step": 18300
},
{
"epoch": 2.0,
"learning_rate": 8.001520582165743e-05,
"loss": 3.2711,
"step": 18400
},
{
"epoch": 2.01,
"learning_rate": 7.990659280981863e-05,
"loss": 3.1771,
"step": 18500
},
{
"epoch": 2.02,
"learning_rate": 7.97979797979798e-05,
"loss": 3.1217,
"step": 18600
},
{
"epoch": 2.03,
"learning_rate": 7.968936678614099e-05,
"loss": 3.0909,
"step": 18700
},
{
"epoch": 2.04,
"learning_rate": 7.958075377430217e-05,
"loss": 3.0777,
"step": 18800
},
{
"epoch": 2.05,
"learning_rate": 7.947214076246335e-05,
"loss": 3.1697,
"step": 18900
},
{
"epoch": 2.06,
"learning_rate": 7.936352775062454e-05,
"loss": 3.1588,
"step": 19000
},
{
"epoch": 2.07,
"learning_rate": 7.925491473878572e-05,
"loss": 3.1334,
"step": 19100
},
{
"epoch": 2.09,
"learning_rate": 7.91463017269469e-05,
"loss": 3.1442,
"step": 19200
},
{
"epoch": 2.1,
"learning_rate": 7.903768871510806e-05,
"loss": 3.1633,
"step": 19300
},
{
"epoch": 2.11,
"learning_rate": 7.892907570326924e-05,
"loss": 3.1913,
"step": 19400
},
{
"epoch": 2.12,
"learning_rate": 7.882046269143044e-05,
"loss": 3.1455,
"step": 19500
},
{
"epoch": 2.13,
"learning_rate": 7.871184967959162e-05,
"loss": 3.155,
"step": 19600
},
{
"epoch": 2.14,
"learning_rate": 7.86032366677528e-05,
"loss": 3.2162,
"step": 19700
},
{
"epoch": 2.15,
"learning_rate": 7.849462365591398e-05,
"loss": 3.1831,
"step": 19800
},
{
"epoch": 2.16,
"learning_rate": 7.838601064407516e-05,
"loss": 3.1012,
"step": 19900
},
{
"epoch": 2.17,
"learning_rate": 7.827739763223635e-05,
"loss": 3.1624,
"step": 20000
},
{
"epoch": 2.18,
"learning_rate": 7.816878462039753e-05,
"loss": 3.1208,
"step": 20100
},
{
"epoch": 2.19,
"learning_rate": 7.806017160855871e-05,
"loss": 3.1124,
"step": 20200
},
{
"epoch": 2.2,
"learning_rate": 7.795155859671989e-05,
"loss": 3.125,
"step": 20300
},
{
"epoch": 2.22,
"learning_rate": 7.784294558488107e-05,
"loss": 3.1171,
"step": 20400
},
{
"epoch": 2.23,
"learning_rate": 7.773433257304225e-05,
"loss": 3.1837,
"step": 20500
},
{
"epoch": 2.24,
"learning_rate": 7.762571956120343e-05,
"loss": 3.1093,
"step": 20600
},
{
"epoch": 2.25,
"learning_rate": 7.751710654936461e-05,
"loss": 3.0628,
"step": 20700
},
{
"epoch": 2.26,
"learning_rate": 7.740849353752579e-05,
"loss": 3.0512,
"step": 20800
},
{
"epoch": 2.27,
"learning_rate": 7.729988052568698e-05,
"loss": 3.127,
"step": 20900
},
{
"epoch": 2.28,
"learning_rate": 7.719126751384816e-05,
"loss": 3.1499,
"step": 21000
},
{
"epoch": 2.29,
"learning_rate": 7.708265450200934e-05,
"loss": 3.069,
"step": 21100
},
{
"epoch": 2.3,
"learning_rate": 7.697404149017052e-05,
"loss": 3.206,
"step": 21200
},
{
"epoch": 2.31,
"learning_rate": 7.68654284783317e-05,
"loss": 3.0981,
"step": 21300
},
{
"epoch": 2.32,
"learning_rate": 7.67568154664929e-05,
"loss": 3.1826,
"step": 21400
},
{
"epoch": 2.34,
"learning_rate": 7.664820245465408e-05,
"loss": 3.1654,
"step": 21500
},
{
"epoch": 2.35,
"learning_rate": 7.653958944281526e-05,
"loss": 3.0237,
"step": 21600
},
{
"epoch": 2.36,
"learning_rate": 7.643097643097644e-05,
"loss": 3.1942,
"step": 21700
},
{
"epoch": 2.37,
"learning_rate": 7.632236341913762e-05,
"loss": 3.1073,
"step": 21800
},
{
"epoch": 2.38,
"learning_rate": 7.62137504072988e-05,
"loss": 3.1343,
"step": 21900
},
{
"epoch": 2.39,
"learning_rate": 7.610513739545998e-05,
"loss": 3.1203,
"step": 22000
},
{
"epoch": 2.4,
"learning_rate": 7.599652438362116e-05,
"loss": 3.1676,
"step": 22100
},
{
"epoch": 2.41,
"learning_rate": 7.588791137178234e-05,
"loss": 3.1682,
"step": 22200
},
{
"epoch": 2.42,
"learning_rate": 7.577929835994353e-05,
"loss": 3.0663,
"step": 22300
},
{
"epoch": 2.43,
"learning_rate": 7.567068534810471e-05,
"loss": 3.1309,
"step": 22400
},
{
"epoch": 2.44,
"learning_rate": 7.556207233626589e-05,
"loss": 3.0167,
"step": 22500
},
{
"epoch": 2.45,
"learning_rate": 7.545345932442707e-05,
"loss": 3.1367,
"step": 22600
},
{
"epoch": 2.47,
"learning_rate": 7.534484631258825e-05,
"loss": 3.1597,
"step": 22700
},
{
"epoch": 2.48,
"learning_rate": 7.523623330074944e-05,
"loss": 3.1616,
"step": 22800
},
{
"epoch": 2.49,
"learning_rate": 7.512762028891062e-05,
"loss": 3.0356,
"step": 22900
},
{
"epoch": 2.5,
"learning_rate": 7.501900727707179e-05,
"loss": 3.1821,
"step": 23000
},
{
"epoch": 2.51,
"learning_rate": 7.491039426523297e-05,
"loss": 3.0478,
"step": 23100
},
{
"epoch": 2.52,
"learning_rate": 7.480178125339416e-05,
"loss": 3.116,
"step": 23200
},
{
"epoch": 2.53,
"learning_rate": 7.469316824155534e-05,
"loss": 3.0147,
"step": 23300
},
{
"epoch": 2.54,
"learning_rate": 7.458455522971652e-05,
"loss": 3.1325,
"step": 23400
},
{
"epoch": 2.55,
"learning_rate": 7.44759422178777e-05,
"loss": 2.9355,
"step": 23500
},
{
"epoch": 2.56,
"learning_rate": 7.436732920603888e-05,
"loss": 3.1018,
"step": 23600
},
{
"epoch": 2.57,
"learning_rate": 7.425871619420008e-05,
"loss": 3.1251,
"step": 23700
},
{
"epoch": 2.58,
"learning_rate": 7.415010318236126e-05,
"loss": 3.0988,
"step": 23800
},
{
"epoch": 2.6,
"learning_rate": 7.404149017052244e-05,
"loss": 3.0469,
"step": 23900
},
{
"epoch": 2.61,
"learning_rate": 7.393287715868362e-05,
"loss": 3.0617,
"step": 24000
},
{
"epoch": 2.62,
"learning_rate": 7.38242641468448e-05,
"loss": 3.1484,
"step": 24100
},
{
"epoch": 2.63,
"learning_rate": 7.371565113500597e-05,
"loss": 3.015,
"step": 24200
},
{
"epoch": 2.64,
"learning_rate": 7.360703812316715e-05,
"loss": 3.1561,
"step": 24300
},
{
"epoch": 2.65,
"learning_rate": 7.349842511132833e-05,
"loss": 3.0644,
"step": 24400
},
{
"epoch": 2.66,
"learning_rate": 7.338981209948951e-05,
"loss": 3.017,
"step": 24500
},
{
"epoch": 2.67,
"learning_rate": 7.328119908765071e-05,
"loss": 3.0305,
"step": 24600
},
{
"epoch": 2.68,
"learning_rate": 7.317258607581189e-05,
"loss": 3.1057,
"step": 24700
},
{
"epoch": 2.69,
"learning_rate": 7.306397306397307e-05,
"loss": 3.0648,
"step": 24800
},
{
"epoch": 2.7,
"learning_rate": 7.295536005213425e-05,
"loss": 2.9583,
"step": 24900
},
{
"epoch": 2.72,
"learning_rate": 7.284674704029543e-05,
"loss": 3.0673,
"step": 25000
},
{
"epoch": 2.73,
"learning_rate": 7.273813402845662e-05,
"loss": 3.0524,
"step": 25100
},
{
"epoch": 2.74,
"learning_rate": 7.26295210166178e-05,
"loss": 3.0098,
"step": 25200
},
{
"epoch": 2.75,
"learning_rate": 7.252090800477898e-05,
"loss": 3.0,
"step": 25300
},
{
"epoch": 2.76,
"learning_rate": 7.241229499294016e-05,
"loss": 3.0914,
"step": 25400
},
{
"epoch": 2.77,
"learning_rate": 7.230368198110134e-05,
"loss": 3.0455,
"step": 25500
},
{
"epoch": 2.78,
"learning_rate": 7.219506896926252e-05,
"loss": 3.06,
"step": 25600
},
{
"epoch": 2.79,
"learning_rate": 7.20864559574237e-05,
"loss": 3.0307,
"step": 25700
},
{
"epoch": 2.8,
"learning_rate": 7.197784294558488e-05,
"loss": 3.0215,
"step": 25800
},
{
"epoch": 2.81,
"learning_rate": 7.186922993374606e-05,
"loss": 3.0389,
"step": 25900
},
{
"epoch": 2.82,
"learning_rate": 7.176061692190725e-05,
"loss": 3.0078,
"step": 26000
},
{
"epoch": 2.83,
"learning_rate": 7.165200391006843e-05,
"loss": 2.9979,
"step": 26100
},
{
"epoch": 2.85,
"learning_rate": 7.154339089822961e-05,
"loss": 3.0612,
"step": 26200
},
{
"epoch": 2.86,
"learning_rate": 7.14347778863908e-05,
"loss": 3.0747,
"step": 26300
},
{
"epoch": 2.87,
"learning_rate": 7.132616487455197e-05,
"loss": 3.0439,
"step": 26400
},
{
"epoch": 2.88,
"learning_rate": 7.121755186271317e-05,
"loss": 3.0511,
"step": 26500
},
{
"epoch": 2.89,
"learning_rate": 7.110893885087435e-05,
"loss": 3.0268,
"step": 26600
},
{
"epoch": 2.9,
"learning_rate": 7.100032583903551e-05,
"loss": 3.0712,
"step": 26700
},
{
"epoch": 2.91,
"learning_rate": 7.089171282719669e-05,
"loss": 3.0466,
"step": 26800
},
{
"epoch": 2.92,
"learning_rate": 7.078309981535787e-05,
"loss": 3.0257,
"step": 26900
},
{
"epoch": 2.93,
"learning_rate": 7.067448680351907e-05,
"loss": 3.0384,
"step": 27000
},
{
"epoch": 2.94,
"learning_rate": 7.056587379168025e-05,
"loss": 2.9557,
"step": 27100
},
{
"epoch": 2.95,
"learning_rate": 7.045726077984143e-05,
"loss": 3.0617,
"step": 27200
},
{
"epoch": 2.97,
"learning_rate": 7.03486477680026e-05,
"loss": 3.0958,
"step": 27300
},
{
"epoch": 2.98,
"learning_rate": 7.02400347561638e-05,
"loss": 3.0421,
"step": 27400
},
{
"epoch": 2.99,
"learning_rate": 7.013142174432498e-05,
"loss": 3.0083,
"step": 27500
},
{
"epoch": 3.0,
"learning_rate": 7.002280873248616e-05,
"loss": 3.0199,
"step": 27600
},
{
"epoch": 3.01,
"learning_rate": 6.991419572064734e-05,
"loss": 3.0205,
"step": 27700
},
{
"epoch": 3.02,
"learning_rate": 6.980558270880852e-05,
"loss": 3.0701,
"step": 27800
},
{
"epoch": 3.03,
"learning_rate": 6.96969696969697e-05,
"loss": 2.8957,
"step": 27900
},
{
"epoch": 3.04,
"learning_rate": 6.958835668513088e-05,
"loss": 2.9442,
"step": 28000
},
{
"epoch": 3.05,
"learning_rate": 6.947974367329206e-05,
"loss": 2.9768,
"step": 28100
},
{
"epoch": 3.06,
"learning_rate": 6.937113066145324e-05,
"loss": 3.0057,
"step": 28200
},
{
"epoch": 3.07,
"learning_rate": 6.926251764961442e-05,
"loss": 2.9816,
"step": 28300
},
{
"epoch": 3.08,
"learning_rate": 6.915390463777561e-05,
"loss": 2.9866,
"step": 28400
},
{
"epoch": 3.1,
"learning_rate": 6.904529162593679e-05,
"loss": 3.056,
"step": 28500
},
{
"epoch": 3.11,
"learning_rate": 6.893667861409797e-05,
"loss": 2.9423,
"step": 28600
},
{
"epoch": 3.12,
"learning_rate": 6.882806560225915e-05,
"loss": 2.9433,
"step": 28700
},
{
"epoch": 3.13,
"learning_rate": 6.871945259042033e-05,
"loss": 2.9175,
"step": 28800
},
{
"epoch": 3.14,
"learning_rate": 6.861083957858153e-05,
"loss": 3.0126,
"step": 28900
},
{
"epoch": 3.15,
"learning_rate": 6.85022265667427e-05,
"loss": 2.9797,
"step": 29000
},
{
"epoch": 3.16,
"learning_rate": 6.839361355490389e-05,
"loss": 2.9233,
"step": 29100
},
{
"epoch": 3.17,
"learning_rate": 6.828500054306506e-05,
"loss": 2.953,
"step": 29200
},
{
"epoch": 3.18,
"learning_rate": 6.817638753122624e-05,
"loss": 2.9087,
"step": 29300
},
{
"epoch": 3.19,
"learning_rate": 6.806777451938742e-05,
"loss": 2.951,
"step": 29400
},
{
"epoch": 3.2,
"learning_rate": 6.79591615075486e-05,
"loss": 3.0381,
"step": 29500
},
{
"epoch": 3.21,
"learning_rate": 6.785054849570978e-05,
"loss": 2.9996,
"step": 29600
},
{
"epoch": 3.23,
"learning_rate": 6.774193548387096e-05,
"loss": 2.9855,
"step": 29700
},
{
"epoch": 3.24,
"learning_rate": 6.763332247203216e-05,
"loss": 2.9895,
"step": 29800
},
{
"epoch": 3.25,
"learning_rate": 6.752470946019334e-05,
"loss": 2.9553,
"step": 29900
},
{
"epoch": 3.26,
"learning_rate": 6.741609644835452e-05,
"loss": 3.0319,
"step": 30000
},
{
"epoch": 3.27,
"learning_rate": 6.73074834365157e-05,
"loss": 3.0371,
"step": 30100
},
{
"epoch": 3.28,
"learning_rate": 6.719887042467688e-05,
"loss": 2.9884,
"step": 30200
},
{
"epoch": 3.29,
"learning_rate": 6.709025741283807e-05,
"loss": 2.9377,
"step": 30300
},
{
"epoch": 3.3,
"learning_rate": 6.698164440099924e-05,
"loss": 2.9495,
"step": 30400
},
{
"epoch": 3.31,
"learning_rate": 6.687303138916042e-05,
"loss": 2.9938,
"step": 30500
},
{
"epoch": 3.32,
"learning_rate": 6.67644183773216e-05,
"loss": 2.9422,
"step": 30600
},
{
"epoch": 3.33,
"learning_rate": 6.665580536548279e-05,
"loss": 3.0207,
"step": 30700
},
{
"epoch": 3.35,
"learning_rate": 6.654719235364397e-05,
"loss": 2.8544,
"step": 30800
},
{
"epoch": 3.36,
"learning_rate": 6.643857934180515e-05,
"loss": 2.9468,
"step": 30900
},
{
"epoch": 3.37,
"learning_rate": 6.632996632996633e-05,
"loss": 2.9787,
"step": 31000
},
{
"epoch": 3.38,
"learning_rate": 6.622135331812751e-05,
"loss": 2.9319,
"step": 31100
},
{
"epoch": 3.39,
"learning_rate": 6.61127403062887e-05,
"loss": 2.9638,
"step": 31200
},
{
"epoch": 3.4,
"learning_rate": 6.600412729444988e-05,
"loss": 2.9358,
"step": 31300
},
{
"epoch": 3.41,
"learning_rate": 6.589551428261106e-05,
"loss": 2.979,
"step": 31400
},
{
"epoch": 3.42,
"learning_rate": 6.578690127077224e-05,
"loss": 2.9887,
"step": 31500
},
{
"epoch": 3.43,
"learning_rate": 6.567828825893342e-05,
"loss": 2.8997,
"step": 31600
},
{
"epoch": 3.44,
"learning_rate": 6.55696752470946e-05,
"loss": 2.9228,
"step": 31700
},
{
"epoch": 3.45,
"learning_rate": 6.546106223525578e-05,
"loss": 2.9675,
"step": 31800
},
{
"epoch": 3.46,
"learning_rate": 6.535244922341696e-05,
"loss": 2.8826,
"step": 31900
},
{
"epoch": 3.48,
"learning_rate": 6.524383621157814e-05,
"loss": 3.0264,
"step": 32000
},
{
"epoch": 3.49,
"learning_rate": 6.513522319973934e-05,
"loss": 2.9325,
"step": 32100
},
{
"epoch": 3.5,
"learning_rate": 6.502661018790052e-05,
"loss": 2.9294,
"step": 32200
},
{
"epoch": 3.51,
"learning_rate": 6.49179971760617e-05,
"loss": 2.9156,
"step": 32300
},
{
"epoch": 3.52,
"learning_rate": 6.480938416422288e-05,
"loss": 2.9291,
"step": 32400
},
{
"epoch": 3.53,
"learning_rate": 6.470077115238406e-05,
"loss": 2.9219,
"step": 32500
},
{
"epoch": 3.54,
"learning_rate": 6.459215814054525e-05,
"loss": 3.0269,
"step": 32600
},
{
"epoch": 3.55,
"learning_rate": 6.448354512870643e-05,
"loss": 3.0573,
"step": 32700
},
{
"epoch": 3.56,
"learning_rate": 6.437493211686761e-05,
"loss": 2.979,
"step": 32800
},
{
"epoch": 3.57,
"learning_rate": 6.426631910502879e-05,
"loss": 2.9678,
"step": 32900
},
{
"epoch": 3.58,
"learning_rate": 6.415770609318996e-05,
"loss": 2.8691,
"step": 33000
},
{
"epoch": 3.6,
"learning_rate": 6.404909308135115e-05,
"loss": 3.0819,
"step": 33100
},
{
"epoch": 3.61,
"learning_rate": 6.394048006951233e-05,
"loss": 2.9761,
"step": 33200
},
{
"epoch": 3.62,
"learning_rate": 6.383186705767351e-05,
"loss": 3.0444,
"step": 33300
},
{
"epoch": 3.63,
"learning_rate": 6.372325404583469e-05,
"loss": 2.91,
"step": 33400
},
{
"epoch": 3.64,
"learning_rate": 6.361464103399588e-05,
"loss": 2.9845,
"step": 33500
},
{
"epoch": 3.65,
"learning_rate": 6.350602802215706e-05,
"loss": 2.9298,
"step": 33600
},
{
"epoch": 3.66,
"learning_rate": 6.339741501031824e-05,
"loss": 2.9518,
"step": 33700
},
{
"epoch": 3.67,
"learning_rate": 6.328880199847942e-05,
"loss": 2.9359,
"step": 33800
},
{
"epoch": 3.68,
"learning_rate": 6.31801889866406e-05,
"loss": 2.9675,
"step": 33900
},
{
"epoch": 3.69,
"learning_rate": 6.30715759748018e-05,
"loss": 2.8137,
"step": 34000
},
{
"epoch": 3.7,
"learning_rate": 6.296296296296296e-05,
"loss": 3.0078,
"step": 34100
},
{
"epoch": 3.71,
"learning_rate": 6.285434995112414e-05,
"loss": 3.0557,
"step": 34200
},
{
"epoch": 3.73,
"learning_rate": 6.274573693928532e-05,
"loss": 2.9161,
"step": 34300
},
{
"epoch": 3.74,
"learning_rate": 6.26371239274465e-05,
"loss": 2.9669,
"step": 34400
},
{
"epoch": 3.75,
"learning_rate": 6.25285109156077e-05,
"loss": 2.9938,
"step": 34500
},
{
"epoch": 3.76,
"learning_rate": 6.241989790376887e-05,
"loss": 2.9183,
"step": 34600
},
{
"epoch": 3.77,
"learning_rate": 6.231128489193005e-05,
"loss": 2.9855,
"step": 34700
},
{
"epoch": 3.78,
"learning_rate": 6.220267188009123e-05,
"loss": 2.8912,
"step": 34800
},
{
"epoch": 3.79,
"learning_rate": 6.209405886825243e-05,
"loss": 2.8896,
"step": 34900
},
{
"epoch": 3.8,
"learning_rate": 6.198544585641361e-05,
"loss": 2.953,
"step": 35000
},
{
"epoch": 3.81,
"learning_rate": 6.187683284457479e-05,
"loss": 2.8615,
"step": 35100
},
{
"epoch": 3.82,
"learning_rate": 6.176821983273597e-05,
"loss": 2.9473,
"step": 35200
},
{
"epoch": 3.83,
"learning_rate": 6.165960682089715e-05,
"loss": 2.9928,
"step": 35300
},
{
"epoch": 3.84,
"learning_rate": 6.155099380905833e-05,
"loss": 2.8082,
"step": 35400
},
{
"epoch": 3.86,
"learning_rate": 6.144238079721951e-05,
"loss": 2.9584,
"step": 35500
},
{
"epoch": 3.87,
"learning_rate": 6.133376778538069e-05,
"loss": 2.9186,
"step": 35600
},
{
"epoch": 3.88,
"learning_rate": 6.122515477354187e-05,
"loss": 2.9403,
"step": 35700
},
{
"epoch": 3.89,
"learning_rate": 6.111654176170305e-05,
"loss": 2.8996,
"step": 35800
},
{
"epoch": 3.9,
"learning_rate": 6.100792874986424e-05,
"loss": 3.0083,
"step": 35900
},
{
"epoch": 3.91,
"learning_rate": 6.089931573802542e-05,
"loss": 2.9722,
"step": 36000
},
{
"epoch": 3.92,
"learning_rate": 6.07907027261866e-05,
"loss": 2.8405,
"step": 36100
},
{
"epoch": 3.93,
"learning_rate": 6.068208971434778e-05,
"loss": 2.971,
"step": 36200
},
{
"epoch": 3.94,
"learning_rate": 6.057347670250897e-05,
"loss": 2.9748,
"step": 36300
},
{
"epoch": 3.95,
"learning_rate": 6.046486369067015e-05,
"loss": 2.8846,
"step": 36400
},
{
"epoch": 3.96,
"learning_rate": 6.0356250678831327e-05,
"loss": 2.8576,
"step": 36500
},
{
"epoch": 3.98,
"learning_rate": 6.0247637666992506e-05,
"loss": 2.9272,
"step": 36600
},
{
"epoch": 3.99,
"learning_rate": 6.0139024655153686e-05,
"loss": 2.9479,
"step": 36700
},
{
"epoch": 4.0,
"learning_rate": 6.003041164331488e-05,
"loss": 2.8707,
"step": 36800
},
{
"epoch": 4.01,
"learning_rate": 5.992179863147605e-05,
"loss": 2.9569,
"step": 36900
},
{
"epoch": 4.02,
"learning_rate": 5.981318561963723e-05,
"loss": 2.8839,
"step": 37000
},
{
"epoch": 4.03,
"learning_rate": 5.970457260779841e-05,
"loss": 2.9383,
"step": 37100
},
{
"epoch": 4.04,
"learning_rate": 5.959595959595959e-05,
"loss": 2.9082,
"step": 37200
},
{
"epoch": 4.05,
"learning_rate": 5.9487346584120786e-05,
"loss": 2.8685,
"step": 37300
},
{
"epoch": 4.06,
"learning_rate": 5.9378733572281966e-05,
"loss": 2.8974,
"step": 37400
},
{
"epoch": 4.07,
"learning_rate": 5.9270120560443146e-05,
"loss": 2.9803,
"step": 37500
},
{
"epoch": 4.08,
"learning_rate": 5.916150754860432e-05,
"loss": 2.8538,
"step": 37600
},
{
"epoch": 4.09,
"learning_rate": 5.905289453676551e-05,
"loss": 2.7821,
"step": 37700
},
{
"epoch": 4.11,
"learning_rate": 5.894428152492669e-05,
"loss": 2.8806,
"step": 37800
},
{
"epoch": 4.12,
"learning_rate": 5.883566851308787e-05,
"loss": 2.871,
"step": 37900
},
{
"epoch": 4.13,
"learning_rate": 5.872705550124905e-05,
"loss": 2.954,
"step": 38000
},
{
"epoch": 4.14,
"learning_rate": 5.861844248941023e-05,
"loss": 2.8033,
"step": 38100
},
{
"epoch": 4.15,
"learning_rate": 5.850982947757142e-05,
"loss": 2.8521,
"step": 38200
},
{
"epoch": 4.16,
"learning_rate": 5.84012164657326e-05,
"loss": 2.7813,
"step": 38300
},
{
"epoch": 4.17,
"learning_rate": 5.829260345389378e-05,
"loss": 2.823,
"step": 38400
},
{
"epoch": 4.18,
"learning_rate": 5.818399044205496e-05,
"loss": 2.862,
"step": 38500
},
{
"epoch": 4.19,
"learning_rate": 5.807537743021614e-05,
"loss": 2.8487,
"step": 38600
},
{
"epoch": 4.2,
"learning_rate": 5.7966764418377325e-05,
"loss": 2.8546,
"step": 38700
},
{
"epoch": 4.21,
"learning_rate": 5.7858151406538505e-05,
"loss": 2.9429,
"step": 38800
},
{
"epoch": 4.23,
"learning_rate": 5.7749538394699685e-05,
"loss": 2.9066,
"step": 38900
},
{
"epoch": 4.24,
"learning_rate": 5.7640925382860865e-05,
"loss": 3.0249,
"step": 39000
},
{
"epoch": 4.25,
"learning_rate": 5.7532312371022045e-05,
"loss": 2.8215,
"step": 39100
},
{
"epoch": 4.26,
"learning_rate": 5.742369935918324e-05,
"loss": 2.8526,
"step": 39200
},
{
"epoch": 4.27,
"learning_rate": 5.731508634734442e-05,
"loss": 2.9065,
"step": 39300
},
{
"epoch": 4.28,
"learning_rate": 5.72064733355056e-05,
"loss": 2.8976,
"step": 39400
},
{
"epoch": 4.29,
"learning_rate": 5.709786032366677e-05,
"loss": 2.9547,
"step": 39500
},
{
"epoch": 4.3,
"learning_rate": 5.6989247311827965e-05,
"loss": 2.9062,
"step": 39600
},
{
"epoch": 4.31,
"learning_rate": 5.6880634299989144e-05,
"loss": 2.8824,
"step": 39700
},
{
"epoch": 4.32,
"learning_rate": 5.6772021288150324e-05,
"loss": 2.8768,
"step": 39800
},
{
"epoch": 4.33,
"learning_rate": 5.6663408276311504e-05,
"loss": 2.8265,
"step": 39900
},
{
"epoch": 4.34,
"learning_rate": 5.6554795264472684e-05,
"loss": 2.9223,
"step": 40000
},
{
"epoch": 4.36,
"learning_rate": 5.644618225263387e-05,
"loss": 2.8611,
"step": 40100
},
{
"epoch": 4.37,
"learning_rate": 5.633756924079505e-05,
"loss": 2.8157,
"step": 40200
},
{
"epoch": 4.38,
"learning_rate": 5.622895622895623e-05,
"loss": 2.848,
"step": 40300
},
{
"epoch": 4.39,
"learning_rate": 5.612034321711741e-05,
"loss": 2.8369,
"step": 40400
},
{
"epoch": 4.4,
"learning_rate": 5.601173020527859e-05,
"loss": 2.966,
"step": 40500
},
{
"epoch": 4.41,
"learning_rate": 5.590311719343978e-05,
"loss": 2.9007,
"step": 40600
},
{
"epoch": 4.42,
"learning_rate": 5.579450418160096e-05,
"loss": 2.9092,
"step": 40700
},
{
"epoch": 4.43,
"learning_rate": 5.568589116976214e-05,
"loss": 2.911,
"step": 40800
},
{
"epoch": 4.44,
"learning_rate": 5.557727815792332e-05,
"loss": 2.8352,
"step": 40900
},
{
"epoch": 4.45,
"learning_rate": 5.546866514608451e-05,
"loss": 2.814,
"step": 41000
},
{
"epoch": 4.46,
"learning_rate": 5.536005213424569e-05,
"loss": 2.8483,
"step": 41100
},
{
"epoch": 4.47,
"learning_rate": 5.525143912240687e-05,
"loss": 2.7428,
"step": 41200
},
{
"epoch": 4.49,
"learning_rate": 5.514282611056805e-05,
"loss": 2.7516,
"step": 41300
},
{
"epoch": 4.5,
"learning_rate": 5.503421309872922e-05,
"loss": 2.9075,
"step": 41400
},
{
"epoch": 4.51,
"learning_rate": 5.4925600086890417e-05,
"loss": 2.8721,
"step": 41500
},
{
"epoch": 4.52,
"learning_rate": 5.4816987075051596e-05,
"loss": 2.8518,
"step": 41600
},
{
"epoch": 4.53,
"learning_rate": 5.4708374063212776e-05,
"loss": 2.8708,
"step": 41700
},
{
"epoch": 4.54,
"learning_rate": 5.4599761051373956e-05,
"loss": 2.8379,
"step": 41800
},
{
"epoch": 4.55,
"learning_rate": 5.4491148039535136e-05,
"loss": 2.8438,
"step": 41900
},
{
"epoch": 4.56,
"learning_rate": 5.438253502769632e-05,
"loss": 2.8823,
"step": 42000
},
{
"epoch": 4.57,
"learning_rate": 5.42739220158575e-05,
"loss": 2.8721,
"step": 42100
},
{
"epoch": 4.58,
"learning_rate": 5.416530900401868e-05,
"loss": 2.8483,
"step": 42200
},
{
"epoch": 4.59,
"learning_rate": 5.405669599217986e-05,
"loss": 2.828,
"step": 42300
},
{
"epoch": 4.61,
"learning_rate": 5.394808298034105e-05,
"loss": 2.9338,
"step": 42400
},
{
"epoch": 4.62,
"learning_rate": 5.383946996850223e-05,
"loss": 2.8982,
"step": 42500
},
{
"epoch": 4.63,
"learning_rate": 5.373085695666341e-05,
"loss": 2.8346,
"step": 42600
},
{
"epoch": 4.64,
"learning_rate": 5.362224394482459e-05,
"loss": 2.9424,
"step": 42700
},
{
"epoch": 4.65,
"learning_rate": 5.351363093298577e-05,
"loss": 2.8952,
"step": 42800
},
{
"epoch": 4.66,
"learning_rate": 5.340501792114696e-05,
"loss": 2.8737,
"step": 42900
},
{
"epoch": 4.67,
"learning_rate": 5.329640490930814e-05,
"loss": 2.9484,
"step": 43000
},
{
"epoch": 4.68,
"learning_rate": 5.318779189746932e-05,
"loss": 2.8492,
"step": 43100
},
{
"epoch": 4.69,
"learning_rate": 5.3079178885630495e-05,
"loss": 2.8864,
"step": 43200
},
{
"epoch": 4.7,
"learning_rate": 5.2970565873791675e-05,
"loss": 2.8093,
"step": 43300
},
{
"epoch": 4.71,
"learning_rate": 5.286195286195287e-05,
"loss": 2.7881,
"step": 43400
},
{
"epoch": 4.72,
"learning_rate": 5.275333985011405e-05,
"loss": 2.8584,
"step": 43500
},
{
"epoch": 4.74,
"learning_rate": 5.264472683827523e-05,
"loss": 2.8925,
"step": 43600
},
{
"epoch": 4.75,
"learning_rate": 5.253611382643641e-05,
"loss": 2.7261,
"step": 43700
},
{
"epoch": 4.76,
"learning_rate": 5.2427500814597595e-05,
"loss": 2.8569,
"step": 43800
},
{
"epoch": 4.77,
"learning_rate": 5.2318887802758775e-05,
"loss": 2.8978,
"step": 43900
},
{
"epoch": 4.78,
"learning_rate": 5.2210274790919955e-05,
"loss": 2.9005,
"step": 44000
},
{
"epoch": 4.79,
"learning_rate": 5.2101661779081135e-05,
"loss": 2.8661,
"step": 44100
},
{
"epoch": 4.8,
"learning_rate": 5.1993048767242315e-05,
"loss": 2.9499,
"step": 44200
},
{
"epoch": 4.81,
"learning_rate": 5.18844357554035e-05,
"loss": 2.8969,
"step": 44300
},
{
"epoch": 4.82,
"learning_rate": 5.177582274356468e-05,
"loss": 2.8314,
"step": 44400
},
{
"epoch": 4.83,
"learning_rate": 5.166720973172586e-05,
"loss": 2.8229,
"step": 44500
},
{
"epoch": 4.84,
"learning_rate": 5.155859671988704e-05,
"loss": 2.8463,
"step": 44600
},
{
"epoch": 4.85,
"learning_rate": 5.144998370804822e-05,
"loss": 2.8764,
"step": 44700
},
{
"epoch": 4.87,
"learning_rate": 5.1341370696209414e-05,
"loss": 2.8458,
"step": 44800
},
{
"epoch": 4.88,
"learning_rate": 5.1232757684370594e-05,
"loss": 2.8062,
"step": 44900
},
{
"epoch": 4.89,
"learning_rate": 5.1124144672531774e-05,
"loss": 2.894,
"step": 45000
},
{
"epoch": 4.9,
"learning_rate": 5.101553166069295e-05,
"loss": 2.8546,
"step": 45100
},
{
"epoch": 4.91,
"learning_rate": 5.090691864885414e-05,
"loss": 2.8455,
"step": 45200
},
{
"epoch": 4.92,
"learning_rate": 5.079830563701532e-05,
"loss": 2.8055,
"step": 45300
},
{
"epoch": 4.93,
"learning_rate": 5.06896926251765e-05,
"loss": 2.9081,
"step": 45400
},
{
"epoch": 4.94,
"learning_rate": 5.058107961333768e-05,
"loss": 2.9016,
"step": 45500
},
{
"epoch": 4.95,
"learning_rate": 5.047246660149886e-05,
"loss": 2.88,
"step": 45600
},
{
"epoch": 4.96,
"learning_rate": 5.036385358966005e-05,
"loss": 2.9696,
"step": 45700
},
{
"epoch": 4.97,
"learning_rate": 5.025524057782123e-05,
"loss": 2.8584,
"step": 45800
},
{
"epoch": 4.99,
"learning_rate": 5.014662756598241e-05,
"loss": 2.8707,
"step": 45900
},
{
"epoch": 5.0,
"learning_rate": 5.003801455414359e-05,
"loss": 2.8539,
"step": 46000
},
{
"epoch": 5.01,
"learning_rate": 4.9929401542304773e-05,
"loss": 2.8842,
"step": 46100
},
{
"epoch": 5.02,
"learning_rate": 4.982078853046595e-05,
"loss": 2.7645,
"step": 46200
},
{
"epoch": 5.03,
"learning_rate": 4.971217551862713e-05,
"loss": 2.8499,
"step": 46300
},
{
"epoch": 5.04,
"learning_rate": 4.960356250678831e-05,
"loss": 2.8812,
"step": 46400
},
{
"epoch": 5.05,
"learning_rate": 4.94949494949495e-05,
"loss": 2.7704,
"step": 46500
},
{
"epoch": 5.06,
"learning_rate": 4.938633648311068e-05,
"loss": 2.7921,
"step": 46600
},
{
"epoch": 5.07,
"learning_rate": 4.927772347127186e-05,
"loss": 2.8428,
"step": 46700
},
{
"epoch": 5.08,
"learning_rate": 4.9169110459433046e-05,
"loss": 2.8605,
"step": 46800
},
{
"epoch": 5.09,
"learning_rate": 4.906049744759422e-05,
"loss": 2.7879,
"step": 46900
},
{
"epoch": 5.1,
"learning_rate": 4.8951884435755406e-05,
"loss": 2.846,
"step": 47000
},
{
"epoch": 5.12,
"learning_rate": 4.8843271423916586e-05,
"loss": 2.8179,
"step": 47100
},
{
"epoch": 5.13,
"learning_rate": 4.873465841207777e-05,
"loss": 2.9302,
"step": 47200
},
{
"epoch": 5.14,
"learning_rate": 4.862604540023895e-05,
"loss": 2.7913,
"step": 47300
},
{
"epoch": 5.15,
"learning_rate": 4.851743238840013e-05,
"loss": 2.7173,
"step": 47400
},
{
"epoch": 5.16,
"learning_rate": 4.840881937656131e-05,
"loss": 2.8389,
"step": 47500
},
{
"epoch": 5.17,
"learning_rate": 4.830020636472249e-05,
"loss": 2.6752,
"step": 47600
},
{
"epoch": 5.18,
"learning_rate": 4.819159335288368e-05,
"loss": 2.7911,
"step": 47700
},
{
"epoch": 5.19,
"learning_rate": 4.808298034104486e-05,
"loss": 2.8147,
"step": 47800
},
{
"epoch": 5.2,
"learning_rate": 4.7974367329206046e-05,
"loss": 2.7192,
"step": 47900
},
{
"epoch": 5.21,
"learning_rate": 4.7865754317367225e-05,
"loss": 2.8325,
"step": 48000
},
{
"epoch": 5.22,
"learning_rate": 4.7757141305528405e-05,
"loss": 2.8132,
"step": 48100
},
{
"epoch": 5.24,
"learning_rate": 4.7648528293689585e-05,
"loss": 2.7436,
"step": 48200
},
{
"epoch": 5.25,
"learning_rate": 4.7539915281850765e-05,
"loss": 2.7892,
"step": 48300
},
{
"epoch": 5.26,
"learning_rate": 4.743130227001195e-05,
"loss": 2.7921,
"step": 48400
},
{
"epoch": 5.27,
"learning_rate": 4.732268925817313e-05,
"loss": 2.7845,
"step": 48500
},
{
"epoch": 5.28,
"learning_rate": 4.721407624633432e-05,
"loss": 2.8523,
"step": 48600
},
{
"epoch": 5.29,
"learning_rate": 4.71054632344955e-05,
"loss": 2.8484,
"step": 48700
},
{
"epoch": 5.3,
"learning_rate": 4.699685022265667e-05,
"loss": 2.7924,
"step": 48800
},
{
"epoch": 5.31,
"learning_rate": 4.688823721081786e-05,
"loss": 2.8363,
"step": 48900
},
{
"epoch": 5.32,
"learning_rate": 4.677962419897904e-05,
"loss": 2.7945,
"step": 49000
},
{
"epoch": 5.33,
"learning_rate": 4.6671011187140225e-05,
"loss": 2.8128,
"step": 49100
},
{
"epoch": 5.34,
"learning_rate": 4.6562398175301405e-05,
"loss": 2.8558,
"step": 49200
},
{
"epoch": 5.35,
"learning_rate": 4.6453785163462584e-05,
"loss": 2.838,
"step": 49300
},
{
"epoch": 5.37,
"learning_rate": 4.6345172151623764e-05,
"loss": 2.7965,
"step": 49400
},
{
"epoch": 5.38,
"learning_rate": 4.6236559139784944e-05,
"loss": 2.742,
"step": 49500
},
{
"epoch": 5.39,
"learning_rate": 4.612794612794613e-05,
"loss": 2.6993,
"step": 49600
},
{
"epoch": 5.4,
"learning_rate": 4.601933311610731e-05,
"loss": 2.8197,
"step": 49700
},
{
"epoch": 5.41,
"learning_rate": 4.59107201042685e-05,
"loss": 2.7072,
"step": 49800
},
{
"epoch": 5.42,
"learning_rate": 4.580210709242968e-05,
"loss": 2.8693,
"step": 49900
},
{
"epoch": 5.43,
"learning_rate": 4.569349408059086e-05,
"loss": 2.795,
"step": 50000
},
{
"epoch": 5.44,
"learning_rate": 4.558488106875204e-05,
"loss": 2.8191,
"step": 50100
},
{
"epoch": 5.45,
"learning_rate": 4.547626805691322e-05,
"loss": 2.8621,
"step": 50200
},
{
"epoch": 5.46,
"learning_rate": 4.5367655045074404e-05,
"loss": 2.7651,
"step": 50300
},
{
"epoch": 5.47,
"learning_rate": 4.5259042033235584e-05,
"loss": 2.77,
"step": 50400
},
{
"epoch": 5.48,
"learning_rate": 4.515042902139677e-05,
"loss": 2.8439,
"step": 50500
},
{
"epoch": 5.5,
"learning_rate": 4.5041816009557944e-05,
"loss": 2.7788,
"step": 50600
},
{
"epoch": 5.51,
"learning_rate": 4.493320299771913e-05,
"loss": 2.8328,
"step": 50700
},
{
"epoch": 5.52,
"learning_rate": 4.482458998588031e-05,
"loss": 2.8625,
"step": 50800
},
{
"epoch": 5.53,
"learning_rate": 4.471597697404149e-05,
"loss": 2.8773,
"step": 50900
},
{
"epoch": 5.54,
"learning_rate": 4.460736396220268e-05,
"loss": 2.8209,
"step": 51000
},
{
"epoch": 5.55,
"learning_rate": 4.449875095036386e-05,
"loss": 2.7392,
"step": 51100
},
{
"epoch": 5.56,
"learning_rate": 4.4390137938525037e-05,
"loss": 2.7048,
"step": 51200
},
{
"epoch": 5.57,
"learning_rate": 4.4281524926686216e-05,
"loss": 2.8771,
"step": 51300
},
{
"epoch": 5.58,
"learning_rate": 4.41729119148474e-05,
"loss": 2.7526,
"step": 51400
},
{
"epoch": 5.59,
"learning_rate": 4.406429890300858e-05,
"loss": 2.7457,
"step": 51500
},
{
"epoch": 5.6,
"learning_rate": 4.395568589116976e-05,
"loss": 2.8338,
"step": 51600
},
{
"epoch": 5.62,
"learning_rate": 4.384707287933095e-05,
"loss": 2.8154,
"step": 51700
},
{
"epoch": 5.63,
"learning_rate": 4.373845986749213e-05,
"loss": 2.7648,
"step": 51800
},
{
"epoch": 5.64,
"learning_rate": 4.362984685565331e-05,
"loss": 2.81,
"step": 51900
},
{
"epoch": 5.65,
"learning_rate": 4.352123384381449e-05,
"loss": 2.742,
"step": 52000
},
{
"epoch": 5.66,
"learning_rate": 4.3412620831975676e-05,
"loss": 2.7741,
"step": 52100
},
{
"epoch": 5.67,
"learning_rate": 4.3304007820136856e-05,
"loss": 2.7733,
"step": 52200
},
{
"epoch": 5.68,
"learning_rate": 4.3195394808298036e-05,
"loss": 2.8178,
"step": 52300
},
{
"epoch": 5.69,
"learning_rate": 4.308678179645922e-05,
"loss": 2.8433,
"step": 52400
},
{
"epoch": 5.7,
"learning_rate": 4.2978168784620396e-05,
"loss": 2.8015,
"step": 52500
},
{
"epoch": 5.71,
"learning_rate": 4.286955577278158e-05,
"loss": 2.8093,
"step": 52600
},
{
"epoch": 5.72,
"learning_rate": 4.276094276094276e-05,
"loss": 2.8235,
"step": 52700
},
{
"epoch": 5.73,
"learning_rate": 4.265232974910394e-05,
"loss": 2.7772,
"step": 52800
},
{
"epoch": 5.75,
"learning_rate": 4.254371673726513e-05,
"loss": 2.7632,
"step": 52900
},
{
"epoch": 5.76,
"learning_rate": 4.243510372542631e-05,
"loss": 2.9062,
"step": 53000
},
{
"epoch": 5.77,
"learning_rate": 4.232649071358749e-05,
"loss": 2.7796,
"step": 53100
},
{
"epoch": 5.78,
"learning_rate": 4.221787770174867e-05,
"loss": 2.7974,
"step": 53200
},
{
"epoch": 5.79,
"learning_rate": 4.2109264689909855e-05,
"loss": 2.8285,
"step": 53300
},
{
"epoch": 5.8,
"learning_rate": 4.2000651678071035e-05,
"loss": 2.8202,
"step": 53400
},
{
"epoch": 5.81,
"learning_rate": 4.1892038666232215e-05,
"loss": 2.8093,
"step": 53500
},
{
"epoch": 5.82,
"learning_rate": 4.17834256543934e-05,
"loss": 2.8685,
"step": 53600
},
{
"epoch": 5.83,
"learning_rate": 4.167481264255458e-05,
"loss": 2.8321,
"step": 53700
},
{
"epoch": 5.84,
"learning_rate": 4.156619963071576e-05,
"loss": 2.6883,
"step": 53800
},
{
"epoch": 5.85,
"learning_rate": 4.145758661887694e-05,
"loss": 2.7624,
"step": 53900
},
{
"epoch": 5.87,
"learning_rate": 4.134897360703813e-05,
"loss": 2.8609,
"step": 54000
},
{
"epoch": 5.88,
"learning_rate": 4.124036059519931e-05,
"loss": 2.8817,
"step": 54100
},
{
"epoch": 5.89,
"learning_rate": 4.113174758336049e-05,
"loss": 2.8336,
"step": 54200
},
{
"epoch": 5.9,
"learning_rate": 4.102313457152167e-05,
"loss": 2.8221,
"step": 54300
},
{
"epoch": 5.91,
"learning_rate": 4.091452155968285e-05,
"loss": 2.7564,
"step": 54400
},
{
"epoch": 5.92,
"learning_rate": 4.0805908547844034e-05,
"loss": 2.796,
"step": 54500
},
{
"epoch": 5.93,
"learning_rate": 4.0697295536005214e-05,
"loss": 2.7683,
"step": 54600
},
{
"epoch": 5.94,
"learning_rate": 4.05886825241664e-05,
"loss": 2.711,
"step": 54700
},
{
"epoch": 5.95,
"learning_rate": 4.048006951232758e-05,
"loss": 2.7996,
"step": 54800
},
{
"epoch": 5.96,
"learning_rate": 4.037145650048876e-05,
"loss": 2.7824,
"step": 54900
},
{
"epoch": 5.97,
"learning_rate": 4.026284348864994e-05,
"loss": 2.8153,
"step": 55000
},
{
"epoch": 5.98,
"learning_rate": 4.015423047681112e-05,
"loss": 2.8316,
"step": 55100
},
{
"epoch": 6.0,
"learning_rate": 4.004561746497231e-05,
"loss": 2.7809,
"step": 55200
},
{
"epoch": 6.01,
"learning_rate": 3.993700445313349e-05,
"loss": 2.745,
"step": 55300
},
{
"epoch": 6.02,
"learning_rate": 3.9828391441294674e-05,
"loss": 2.7401,
"step": 55400
},
{
"epoch": 6.03,
"learning_rate": 3.9719778429455854e-05,
"loss": 2.7371,
"step": 55500
},
{
"epoch": 6.04,
"learning_rate": 3.961116541761703e-05,
"loss": 2.723,
"step": 55600
},
{
"epoch": 6.05,
"learning_rate": 3.9502552405778213e-05,
"loss": 2.7057,
"step": 55700
},
{
"epoch": 6.06,
"learning_rate": 3.939393939393939e-05,
"loss": 2.8222,
"step": 55800
},
{
"epoch": 6.07,
"learning_rate": 3.928532638210058e-05,
"loss": 2.6662,
"step": 55900
},
{
"epoch": 6.08,
"learning_rate": 3.917671337026176e-05,
"loss": 2.7647,
"step": 56000
},
{
"epoch": 6.09,
"learning_rate": 3.906810035842295e-05,
"loss": 2.7936,
"step": 56100
},
{
"epoch": 6.1,
"learning_rate": 3.895948734658412e-05,
"loss": 2.7848,
"step": 56200
},
{
"epoch": 6.11,
"learning_rate": 3.88508743347453e-05,
"loss": 2.768,
"step": 56300
},
{
"epoch": 6.13,
"learning_rate": 3.8742261322906486e-05,
"loss": 2.7741,
"step": 56400
},
{
"epoch": 6.14,
"learning_rate": 3.8633648311067666e-05,
"loss": 2.7476,
"step": 56500
},
{
"epoch": 6.15,
"learning_rate": 3.852503529922885e-05,
"loss": 2.778,
"step": 56600
},
{
"epoch": 6.16,
"learning_rate": 3.841642228739003e-05,
"loss": 2.8299,
"step": 56700
},
{
"epoch": 6.17,
"learning_rate": 3.830780927555121e-05,
"loss": 2.785,
"step": 56800
},
{
"epoch": 6.18,
"learning_rate": 3.819919626371239e-05,
"loss": 2.6849,
"step": 56900
},
{
"epoch": 6.19,
"learning_rate": 3.809058325187357e-05,
"loss": 2.6925,
"step": 57000
},
{
"epoch": 6.2,
"learning_rate": 3.798197024003476e-05,
"loss": 2.6585,
"step": 57100
},
{
"epoch": 6.21,
"learning_rate": 3.787335722819594e-05,
"loss": 2.7174,
"step": 57200
},
{
"epoch": 6.22,
"learning_rate": 3.7764744216357126e-05,
"loss": 2.8318,
"step": 57300
},
{
"epoch": 6.23,
"learning_rate": 3.7656131204518306e-05,
"loss": 2.7448,
"step": 57400
},
{
"epoch": 6.25,
"learning_rate": 3.7547518192679486e-05,
"loss": 2.8137,
"step": 57500
},
{
"epoch": 6.26,
"learning_rate": 3.7438905180840665e-05,
"loss": 2.6344,
"step": 57600
},
{
"epoch": 6.27,
"learning_rate": 3.7330292169001845e-05,
"loss": 2.7126,
"step": 57700
},
{
"epoch": 6.28,
"learning_rate": 3.722167915716303e-05,
"loss": 2.7564,
"step": 57800
},
{
"epoch": 6.29,
"learning_rate": 3.711306614532421e-05,
"loss": 2.7657,
"step": 57900
},
{
"epoch": 6.3,
"learning_rate": 3.700445313348539e-05,
"loss": 2.8002,
"step": 58000
},
{
"epoch": 6.31,
"learning_rate": 3.689584012164657e-05,
"loss": 2.7179,
"step": 58100
},
{
"epoch": 6.32,
"learning_rate": 3.678722710980776e-05,
"loss": 2.7635,
"step": 58200
},
{
"epoch": 6.33,
"learning_rate": 3.667861409796894e-05,
"loss": 2.6482,
"step": 58300
},
{
"epoch": 6.34,
"learning_rate": 3.657000108613012e-05,
"loss": 2.7555,
"step": 58400
},
{
"epoch": 6.35,
"learning_rate": 3.6461388074291305e-05,
"loss": 2.7954,
"step": 58500
}
],
"max_steps": 92070,
"num_train_epochs": 10,
"total_flos": 1.1996750282539008e+17,
"trial_name": null,
"trial_params": null
}