mt5_base_yor_eng_mt / trainer_state.json
Davlan's picture
adding mt5_base_yor_eng_mt model
b8d1d1d
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 150213,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.983356966440987e-05,
"loss": 5.6467,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 4.9667139328819746e-05,
"loss": 3.5541,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 4.9500708993229615e-05,
"loss": 3.2816,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 4.933427865763949e-05,
"loss": 3.1249,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 4.916784832204936e-05,
"loss": 2.9851,
"step": 2500
},
{
"epoch": 0.06,
"learning_rate": 4.9001417986459234e-05,
"loss": 2.8957,
"step": 3000
},
{
"epoch": 0.07,
"learning_rate": 4.88349876508691e-05,
"loss": 2.8483,
"step": 3500
},
{
"epoch": 0.08,
"learning_rate": 4.866855731527897e-05,
"loss": 2.7829,
"step": 4000
},
{
"epoch": 0.09,
"learning_rate": 4.8502126979688846e-05,
"loss": 2.706,
"step": 4500
},
{
"epoch": 0.1,
"learning_rate": 4.8335696644098715e-05,
"loss": 2.6381,
"step": 5000
},
{
"epoch": 0.11,
"learning_rate": 4.816926630850859e-05,
"loss": 2.6078,
"step": 5500
},
{
"epoch": 0.12,
"learning_rate": 4.800283597291846e-05,
"loss": 2.5787,
"step": 6000
},
{
"epoch": 0.13,
"learning_rate": 4.783640563732833e-05,
"loss": 2.54,
"step": 6500
},
{
"epoch": 0.14,
"learning_rate": 4.76699753017382e-05,
"loss": 2.4863,
"step": 7000
},
{
"epoch": 0.15,
"learning_rate": 4.750354496614807e-05,
"loss": 2.4385,
"step": 7500
},
{
"epoch": 0.16,
"learning_rate": 4.7337114630557946e-05,
"loss": 2.4533,
"step": 8000
},
{
"epoch": 0.17,
"learning_rate": 4.7170684294967814e-05,
"loss": 2.4262,
"step": 8500
},
{
"epoch": 0.18,
"learning_rate": 4.700425395937769e-05,
"loss": 2.4119,
"step": 9000
},
{
"epoch": 0.19,
"learning_rate": 4.683782362378756e-05,
"loss": 2.3744,
"step": 9500
},
{
"epoch": 0.2,
"learning_rate": 4.667139328819743e-05,
"loss": 2.3672,
"step": 10000
},
{
"epoch": 0.21,
"learning_rate": 4.65049629526073e-05,
"loss": 2.3025,
"step": 10500
},
{
"epoch": 0.22,
"learning_rate": 4.633853261701717e-05,
"loss": 2.3294,
"step": 11000
},
{
"epoch": 0.23,
"learning_rate": 4.6172102281427046e-05,
"loss": 2.3193,
"step": 11500
},
{
"epoch": 0.24,
"learning_rate": 4.6005671945836914e-05,
"loss": 2.2715,
"step": 12000
},
{
"epoch": 0.25,
"learning_rate": 4.583924161024678e-05,
"loss": 2.2692,
"step": 12500
},
{
"epoch": 0.26,
"learning_rate": 4.567281127465666e-05,
"loss": 2.2439,
"step": 13000
},
{
"epoch": 0.27,
"learning_rate": 4.5506380939066527e-05,
"loss": 2.2578,
"step": 13500
},
{
"epoch": 0.28,
"learning_rate": 4.53399506034764e-05,
"loss": 2.2119,
"step": 14000
},
{
"epoch": 0.29,
"learning_rate": 4.517352026788627e-05,
"loss": 2.1908,
"step": 14500
},
{
"epoch": 0.3,
"learning_rate": 4.5007089932296146e-05,
"loss": 2.2054,
"step": 15000
},
{
"epoch": 0.31,
"learning_rate": 4.4840659596706014e-05,
"loss": 2.1669,
"step": 15500
},
{
"epoch": 0.32,
"learning_rate": 4.467422926111588e-05,
"loss": 2.1593,
"step": 16000
},
{
"epoch": 0.33,
"learning_rate": 4.450779892552576e-05,
"loss": 2.1789,
"step": 16500
},
{
"epoch": 0.34,
"learning_rate": 4.4341368589935626e-05,
"loss": 2.1593,
"step": 17000
},
{
"epoch": 0.35,
"learning_rate": 4.41749382543455e-05,
"loss": 2.1402,
"step": 17500
},
{
"epoch": 0.36,
"learning_rate": 4.400850791875537e-05,
"loss": 2.1106,
"step": 18000
},
{
"epoch": 0.37,
"learning_rate": 4.384207758316524e-05,
"loss": 2.1304,
"step": 18500
},
{
"epoch": 0.38,
"learning_rate": 4.3675647247575114e-05,
"loss": 2.0842,
"step": 19000
},
{
"epoch": 0.39,
"learning_rate": 4.350921691198498e-05,
"loss": 2.0842,
"step": 19500
},
{
"epoch": 0.4,
"learning_rate": 4.334278657639486e-05,
"loss": 2.09,
"step": 20000
},
{
"epoch": 0.41,
"learning_rate": 4.3176356240804726e-05,
"loss": 2.0693,
"step": 20500
},
{
"epoch": 0.42,
"learning_rate": 4.30099259052146e-05,
"loss": 2.0849,
"step": 21000
},
{
"epoch": 0.43,
"learning_rate": 4.284349556962447e-05,
"loss": 2.0685,
"step": 21500
},
{
"epoch": 0.44,
"learning_rate": 4.267706523403434e-05,
"loss": 2.064,
"step": 22000
},
{
"epoch": 0.45,
"learning_rate": 4.2510634898444214e-05,
"loss": 2.0448,
"step": 22500
},
{
"epoch": 0.46,
"learning_rate": 4.234420456285408e-05,
"loss": 2.0482,
"step": 23000
},
{
"epoch": 0.47,
"learning_rate": 4.217777422726396e-05,
"loss": 2.0422,
"step": 23500
},
{
"epoch": 0.48,
"learning_rate": 4.2011343891673826e-05,
"loss": 2.0584,
"step": 24000
},
{
"epoch": 0.49,
"learning_rate": 4.1844913556083695e-05,
"loss": 2.0323,
"step": 24500
},
{
"epoch": 0.5,
"learning_rate": 4.167848322049357e-05,
"loss": 2.0167,
"step": 25000
},
{
"epoch": 0.51,
"learning_rate": 4.151205288490344e-05,
"loss": 2.0093,
"step": 25500
},
{
"epoch": 0.52,
"learning_rate": 4.1345622549313314e-05,
"loss": 2.0247,
"step": 26000
},
{
"epoch": 0.53,
"learning_rate": 4.117919221372318e-05,
"loss": 2.0109,
"step": 26500
},
{
"epoch": 0.54,
"learning_rate": 4.101276187813306e-05,
"loss": 1.9798,
"step": 27000
},
{
"epoch": 0.55,
"learning_rate": 4.0846331542542926e-05,
"loss": 1.9745,
"step": 27500
},
{
"epoch": 0.56,
"learning_rate": 4.0679901206952795e-05,
"loss": 1.9857,
"step": 28000
},
{
"epoch": 0.57,
"learning_rate": 4.051347087136267e-05,
"loss": 1.9451,
"step": 28500
},
{
"epoch": 0.58,
"learning_rate": 4.034704053577254e-05,
"loss": 1.9724,
"step": 29000
},
{
"epoch": 0.59,
"learning_rate": 4.0180610200182414e-05,
"loss": 1.9624,
"step": 29500
},
{
"epoch": 0.6,
"learning_rate": 4.001417986459228e-05,
"loss": 1.9624,
"step": 30000
},
{
"epoch": 0.61,
"learning_rate": 3.984774952900215e-05,
"loss": 1.9406,
"step": 30500
},
{
"epoch": 0.62,
"learning_rate": 3.9681319193412026e-05,
"loss": 1.9424,
"step": 31000
},
{
"epoch": 0.63,
"learning_rate": 3.9514888857821894e-05,
"loss": 1.9464,
"step": 31500
},
{
"epoch": 0.64,
"learning_rate": 3.934845852223177e-05,
"loss": 1.9304,
"step": 32000
},
{
"epoch": 0.65,
"learning_rate": 3.918202818664164e-05,
"loss": 1.9177,
"step": 32500
},
{
"epoch": 0.66,
"learning_rate": 3.9015597851051514e-05,
"loss": 1.9441,
"step": 33000
},
{
"epoch": 0.67,
"learning_rate": 3.884916751546138e-05,
"loss": 1.9404,
"step": 33500
},
{
"epoch": 0.68,
"learning_rate": 3.868273717987125e-05,
"loss": 1.9302,
"step": 34000
},
{
"epoch": 0.69,
"learning_rate": 3.8516306844281126e-05,
"loss": 1.9219,
"step": 34500
},
{
"epoch": 0.7,
"learning_rate": 3.8349876508690994e-05,
"loss": 1.9178,
"step": 35000
},
{
"epoch": 0.71,
"learning_rate": 3.818344617310087e-05,
"loss": 1.9061,
"step": 35500
},
{
"epoch": 0.72,
"learning_rate": 3.801701583751074e-05,
"loss": 1.9037,
"step": 36000
},
{
"epoch": 0.73,
"learning_rate": 3.7850585501920613e-05,
"loss": 1.8948,
"step": 36500
},
{
"epoch": 0.74,
"learning_rate": 3.768415516633048e-05,
"loss": 1.8893,
"step": 37000
},
{
"epoch": 0.75,
"learning_rate": 3.751772483074035e-05,
"loss": 1.8773,
"step": 37500
},
{
"epoch": 0.76,
"learning_rate": 3.7351294495150226e-05,
"loss": 1.864,
"step": 38000
},
{
"epoch": 0.77,
"learning_rate": 3.7184864159560094e-05,
"loss": 1.8815,
"step": 38500
},
{
"epoch": 0.78,
"learning_rate": 3.701843382396997e-05,
"loss": 1.8842,
"step": 39000
},
{
"epoch": 0.79,
"learning_rate": 3.685200348837984e-05,
"loss": 1.8832,
"step": 39500
},
{
"epoch": 0.8,
"learning_rate": 3.6685573152789707e-05,
"loss": 1.885,
"step": 40000
},
{
"epoch": 0.81,
"learning_rate": 3.651914281719958e-05,
"loss": 1.8553,
"step": 40500
},
{
"epoch": 0.82,
"learning_rate": 3.635271248160945e-05,
"loss": 1.8779,
"step": 41000
},
{
"epoch": 0.83,
"learning_rate": 3.6186282146019326e-05,
"loss": 1.8371,
"step": 41500
},
{
"epoch": 0.84,
"learning_rate": 3.6019851810429194e-05,
"loss": 1.8458,
"step": 42000
},
{
"epoch": 0.85,
"learning_rate": 3.585342147483907e-05,
"loss": 1.8363,
"step": 42500
},
{
"epoch": 0.86,
"learning_rate": 3.568699113924894e-05,
"loss": 1.8426,
"step": 43000
},
{
"epoch": 0.87,
"learning_rate": 3.5520560803658806e-05,
"loss": 1.8552,
"step": 43500
},
{
"epoch": 0.88,
"learning_rate": 3.535413046806868e-05,
"loss": 1.8454,
"step": 44000
},
{
"epoch": 0.89,
"learning_rate": 3.518770013247855e-05,
"loss": 1.8226,
"step": 44500
},
{
"epoch": 0.9,
"learning_rate": 3.502126979688842e-05,
"loss": 1.8172,
"step": 45000
},
{
"epoch": 0.91,
"learning_rate": 3.485483946129829e-05,
"loss": 1.8117,
"step": 45500
},
{
"epoch": 0.92,
"learning_rate": 3.468840912570816e-05,
"loss": 1.8159,
"step": 46000
},
{
"epoch": 0.93,
"learning_rate": 3.452197879011803e-05,
"loss": 1.8214,
"step": 46500
},
{
"epoch": 0.94,
"learning_rate": 3.43555484545279e-05,
"loss": 1.8243,
"step": 47000
},
{
"epoch": 0.95,
"learning_rate": 3.4189118118937775e-05,
"loss": 1.8233,
"step": 47500
},
{
"epoch": 0.96,
"learning_rate": 3.402268778334764e-05,
"loss": 1.8217,
"step": 48000
},
{
"epoch": 0.97,
"learning_rate": 3.385625744775752e-05,
"loss": 1.8127,
"step": 48500
},
{
"epoch": 0.98,
"learning_rate": 3.368982711216739e-05,
"loss": 1.8222,
"step": 49000
},
{
"epoch": 0.99,
"learning_rate": 3.352339677657726e-05,
"loss": 1.8151,
"step": 49500
},
{
"epoch": 1.0,
"learning_rate": 3.335696644098713e-05,
"loss": 1.8053,
"step": 50000
},
{
"epoch": 1.01,
"learning_rate": 3.3190536105397e-05,
"loss": 1.7738,
"step": 50500
},
{
"epoch": 1.02,
"learning_rate": 3.3024105769806875e-05,
"loss": 1.7675,
"step": 51000
},
{
"epoch": 1.03,
"learning_rate": 3.285767543421674e-05,
"loss": 1.7464,
"step": 51500
},
{
"epoch": 1.04,
"learning_rate": 3.269124509862662e-05,
"loss": 1.7511,
"step": 52000
},
{
"epoch": 1.05,
"learning_rate": 3.252481476303649e-05,
"loss": 1.7535,
"step": 52500
},
{
"epoch": 1.06,
"learning_rate": 3.2358384427446355e-05,
"loss": 1.7529,
"step": 53000
},
{
"epoch": 1.07,
"learning_rate": 3.219195409185623e-05,
"loss": 1.739,
"step": 53500
},
{
"epoch": 1.08,
"learning_rate": 3.20255237562661e-05,
"loss": 1.7718,
"step": 54000
},
{
"epoch": 1.09,
"learning_rate": 3.1859093420675975e-05,
"loss": 1.7317,
"step": 54500
},
{
"epoch": 1.1,
"learning_rate": 3.169266308508584e-05,
"loss": 1.7447,
"step": 55000
},
{
"epoch": 1.11,
"learning_rate": 3.152623274949572e-05,
"loss": 1.7642,
"step": 55500
},
{
"epoch": 1.12,
"learning_rate": 3.135980241390559e-05,
"loss": 1.7395,
"step": 56000
},
{
"epoch": 1.13,
"learning_rate": 3.1193372078315455e-05,
"loss": 1.7466,
"step": 56500
},
{
"epoch": 1.14,
"learning_rate": 3.102694174272533e-05,
"loss": 1.7331,
"step": 57000
},
{
"epoch": 1.15,
"learning_rate": 3.08605114071352e-05,
"loss": 1.7624,
"step": 57500
},
{
"epoch": 1.16,
"learning_rate": 3.0694081071545074e-05,
"loss": 1.7254,
"step": 58000
},
{
"epoch": 1.17,
"learning_rate": 3.052765073595494e-05,
"loss": 1.7418,
"step": 58500
},
{
"epoch": 1.18,
"learning_rate": 3.0361220400364815e-05,
"loss": 1.753,
"step": 59000
},
{
"epoch": 1.19,
"learning_rate": 3.0194790064774687e-05,
"loss": 1.722,
"step": 59500
},
{
"epoch": 1.2,
"learning_rate": 3.002835972918456e-05,
"loss": 1.7337,
"step": 60000
},
{
"epoch": 1.21,
"learning_rate": 2.986192939359443e-05,
"loss": 1.7284,
"step": 60500
},
{
"epoch": 1.22,
"learning_rate": 2.96954990580043e-05,
"loss": 1.731,
"step": 61000
},
{
"epoch": 1.23,
"learning_rate": 2.952906872241417e-05,
"loss": 1.7412,
"step": 61500
},
{
"epoch": 1.24,
"learning_rate": 2.9362638386824043e-05,
"loss": 1.6991,
"step": 62000
},
{
"epoch": 1.25,
"learning_rate": 2.9196208051233915e-05,
"loss": 1.7283,
"step": 62500
},
{
"epoch": 1.26,
"learning_rate": 2.9029777715643787e-05,
"loss": 1.7117,
"step": 63000
},
{
"epoch": 1.27,
"learning_rate": 2.886334738005366e-05,
"loss": 1.7175,
"step": 63500
},
{
"epoch": 1.28,
"learning_rate": 2.8696917044463527e-05,
"loss": 1.7125,
"step": 64000
},
{
"epoch": 1.29,
"learning_rate": 2.85304867088734e-05,
"loss": 1.694,
"step": 64500
},
{
"epoch": 1.3,
"learning_rate": 2.836405637328327e-05,
"loss": 1.7034,
"step": 65000
},
{
"epoch": 1.31,
"learning_rate": 2.8197626037693143e-05,
"loss": 1.7253,
"step": 65500
},
{
"epoch": 1.32,
"learning_rate": 2.8031195702103015e-05,
"loss": 1.7147,
"step": 66000
},
{
"epoch": 1.33,
"learning_rate": 2.7864765366512886e-05,
"loss": 1.7062,
"step": 66500
},
{
"epoch": 1.34,
"learning_rate": 2.7698335030922755e-05,
"loss": 1.7073,
"step": 67000
},
{
"epoch": 1.35,
"learning_rate": 2.7531904695332627e-05,
"loss": 1.6874,
"step": 67500
},
{
"epoch": 1.36,
"learning_rate": 2.73654743597425e-05,
"loss": 1.6932,
"step": 68000
},
{
"epoch": 1.37,
"learning_rate": 2.719904402415237e-05,
"loss": 1.7043,
"step": 68500
},
{
"epoch": 1.38,
"learning_rate": 2.7032613688562243e-05,
"loss": 1.7047,
"step": 69000
},
{
"epoch": 1.39,
"learning_rate": 2.6866183352972114e-05,
"loss": 1.6873,
"step": 69500
},
{
"epoch": 1.4,
"learning_rate": 2.6699753017381983e-05,
"loss": 1.6845,
"step": 70000
},
{
"epoch": 1.41,
"learning_rate": 2.6533322681791855e-05,
"loss": 1.7007,
"step": 70500
},
{
"epoch": 1.42,
"learning_rate": 2.6366892346201727e-05,
"loss": 1.6864,
"step": 71000
},
{
"epoch": 1.43,
"learning_rate": 2.62004620106116e-05,
"loss": 1.6614,
"step": 71500
},
{
"epoch": 1.44,
"learning_rate": 2.603403167502147e-05,
"loss": 1.694,
"step": 72000
},
{
"epoch": 1.45,
"learning_rate": 2.5867601339431342e-05,
"loss": 1.7015,
"step": 72500
},
{
"epoch": 1.46,
"learning_rate": 2.570117100384121e-05,
"loss": 1.6798,
"step": 73000
},
{
"epoch": 1.47,
"learning_rate": 2.5534740668251083e-05,
"loss": 1.6661,
"step": 73500
},
{
"epoch": 1.48,
"learning_rate": 2.5368310332660955e-05,
"loss": 1.6785,
"step": 74000
},
{
"epoch": 1.49,
"learning_rate": 2.5201879997070827e-05,
"loss": 1.666,
"step": 74500
},
{
"epoch": 1.5,
"learning_rate": 2.50354496614807e-05,
"loss": 1.676,
"step": 75000
},
{
"epoch": 1.51,
"learning_rate": 2.486901932589057e-05,
"loss": 1.6898,
"step": 75500
},
{
"epoch": 1.52,
"learning_rate": 2.470258899030044e-05,
"loss": 1.6881,
"step": 76000
},
{
"epoch": 1.53,
"learning_rate": 2.453615865471031e-05,
"loss": 1.7249,
"step": 76500
},
{
"epoch": 1.54,
"learning_rate": 2.4369728319120183e-05,
"loss": 1.6754,
"step": 77000
},
{
"epoch": 1.55,
"learning_rate": 2.4203297983530055e-05,
"loss": 1.6849,
"step": 77500
},
{
"epoch": 1.56,
"learning_rate": 2.4036867647939926e-05,
"loss": 1.6736,
"step": 78000
},
{
"epoch": 1.57,
"learning_rate": 2.38704373123498e-05,
"loss": 1.6918,
"step": 78500
},
{
"epoch": 1.58,
"learning_rate": 2.3704006976759667e-05,
"loss": 1.683,
"step": 79000
},
{
"epoch": 1.59,
"learning_rate": 2.353757664116954e-05,
"loss": 1.6835,
"step": 79500
},
{
"epoch": 1.6,
"learning_rate": 2.337114630557941e-05,
"loss": 1.6762,
"step": 80000
},
{
"epoch": 1.61,
"learning_rate": 2.3204715969989283e-05,
"loss": 1.6733,
"step": 80500
},
{
"epoch": 1.62,
"learning_rate": 2.3038285634399154e-05,
"loss": 1.6904,
"step": 81000
},
{
"epoch": 1.63,
"learning_rate": 2.2871855298809026e-05,
"loss": 1.6868,
"step": 81500
},
{
"epoch": 1.64,
"learning_rate": 2.2705424963218895e-05,
"loss": 1.6596,
"step": 82000
},
{
"epoch": 1.65,
"learning_rate": 2.2538994627628767e-05,
"loss": 1.6606,
"step": 82500
},
{
"epoch": 1.66,
"learning_rate": 2.237256429203864e-05,
"loss": 1.6599,
"step": 83000
},
{
"epoch": 1.67,
"learning_rate": 2.220613395644851e-05,
"loss": 1.6584,
"step": 83500
},
{
"epoch": 1.68,
"learning_rate": 2.2039703620858382e-05,
"loss": 1.6688,
"step": 84000
},
{
"epoch": 1.69,
"learning_rate": 2.1873273285268254e-05,
"loss": 1.6649,
"step": 84500
},
{
"epoch": 1.7,
"learning_rate": 2.1706842949678123e-05,
"loss": 1.6603,
"step": 85000
},
{
"epoch": 1.71,
"learning_rate": 2.1540412614087995e-05,
"loss": 1.6332,
"step": 85500
},
{
"epoch": 1.72,
"learning_rate": 2.1373982278497867e-05,
"loss": 1.654,
"step": 86000
},
{
"epoch": 1.73,
"learning_rate": 2.120755194290774e-05,
"loss": 1.662,
"step": 86500
},
{
"epoch": 1.74,
"learning_rate": 2.104112160731761e-05,
"loss": 1.6474,
"step": 87000
},
{
"epoch": 1.75,
"learning_rate": 2.0874691271727482e-05,
"loss": 1.6431,
"step": 87500
},
{
"epoch": 1.76,
"learning_rate": 2.0708260936137354e-05,
"loss": 1.6847,
"step": 88000
},
{
"epoch": 1.77,
"learning_rate": 2.0541830600547223e-05,
"loss": 1.6335,
"step": 88500
},
{
"epoch": 1.78,
"learning_rate": 2.0375400264957095e-05,
"loss": 1.631,
"step": 89000
},
{
"epoch": 1.79,
"learning_rate": 2.0208969929366966e-05,
"loss": 1.6327,
"step": 89500
},
{
"epoch": 1.8,
"learning_rate": 2.004253959377684e-05,
"loss": 1.6399,
"step": 90000
},
{
"epoch": 1.81,
"learning_rate": 1.987610925818671e-05,
"loss": 1.6618,
"step": 90500
},
{
"epoch": 1.82,
"learning_rate": 1.9709678922596582e-05,
"loss": 1.6317,
"step": 91000
},
{
"epoch": 1.83,
"learning_rate": 1.954324858700645e-05,
"loss": 1.6786,
"step": 91500
},
{
"epoch": 1.84,
"learning_rate": 1.9376818251416323e-05,
"loss": 1.6379,
"step": 92000
},
{
"epoch": 1.85,
"learning_rate": 1.9210387915826194e-05,
"loss": 1.6408,
"step": 92500
},
{
"epoch": 1.86,
"learning_rate": 1.9043957580236066e-05,
"loss": 1.6476,
"step": 93000
},
{
"epoch": 1.87,
"learning_rate": 1.8877527244645938e-05,
"loss": 1.6308,
"step": 93500
},
{
"epoch": 1.88,
"learning_rate": 1.871109690905581e-05,
"loss": 1.6471,
"step": 94000
},
{
"epoch": 1.89,
"learning_rate": 1.854466657346568e-05,
"loss": 1.6114,
"step": 94500
},
{
"epoch": 1.9,
"learning_rate": 1.837823623787555e-05,
"loss": 1.6447,
"step": 95000
},
{
"epoch": 1.91,
"learning_rate": 1.8211805902285422e-05,
"loss": 1.627,
"step": 95500
},
{
"epoch": 1.92,
"learning_rate": 1.8045375566695294e-05,
"loss": 1.6435,
"step": 96000
},
{
"epoch": 1.93,
"learning_rate": 1.7878945231105166e-05,
"loss": 1.6458,
"step": 96500
},
{
"epoch": 1.94,
"learning_rate": 1.7712514895515038e-05,
"loss": 1.6331,
"step": 97000
},
{
"epoch": 1.95,
"learning_rate": 1.7546084559924907e-05,
"loss": 1.6265,
"step": 97500
},
{
"epoch": 1.96,
"learning_rate": 1.737965422433478e-05,
"loss": 1.6143,
"step": 98000
},
{
"epoch": 1.97,
"learning_rate": 1.721322388874465e-05,
"loss": 1.6386,
"step": 98500
},
{
"epoch": 1.98,
"learning_rate": 1.7046793553154522e-05,
"loss": 1.6178,
"step": 99000
},
{
"epoch": 1.99,
"learning_rate": 1.6880363217564394e-05,
"loss": 1.6329,
"step": 99500
},
{
"epoch": 2.0,
"learning_rate": 1.6713932881974266e-05,
"loss": 1.633,
"step": 100000
},
{
"epoch": 2.01,
"learning_rate": 1.6547502546384135e-05,
"loss": 1.5901,
"step": 100500
},
{
"epoch": 2.02,
"learning_rate": 1.6381072210794007e-05,
"loss": 1.5989,
"step": 101000
},
{
"epoch": 2.03,
"learning_rate": 1.621464187520388e-05,
"loss": 1.5882,
"step": 101500
},
{
"epoch": 2.04,
"learning_rate": 1.604821153961375e-05,
"loss": 1.5791,
"step": 102000
},
{
"epoch": 2.05,
"learning_rate": 1.5881781204023622e-05,
"loss": 1.5939,
"step": 102500
},
{
"epoch": 2.06,
"learning_rate": 1.5715350868433494e-05,
"loss": 1.5895,
"step": 103000
},
{
"epoch": 2.07,
"learning_rate": 1.5548920532843363e-05,
"loss": 1.5948,
"step": 103500
},
{
"epoch": 2.08,
"learning_rate": 1.5382490197253234e-05,
"loss": 1.5934,
"step": 104000
},
{
"epoch": 2.09,
"learning_rate": 1.5216059861663106e-05,
"loss": 1.6048,
"step": 104500
},
{
"epoch": 2.1,
"learning_rate": 1.5049629526072978e-05,
"loss": 1.5903,
"step": 105000
},
{
"epoch": 2.11,
"learning_rate": 1.488319919048285e-05,
"loss": 1.6172,
"step": 105500
},
{
"epoch": 2.12,
"learning_rate": 1.471676885489272e-05,
"loss": 1.5954,
"step": 106000
},
{
"epoch": 2.13,
"learning_rate": 1.4550338519302592e-05,
"loss": 1.5849,
"step": 106500
},
{
"epoch": 2.14,
"learning_rate": 1.4383908183712464e-05,
"loss": 1.5823,
"step": 107000
},
{
"epoch": 2.15,
"learning_rate": 1.4217477848122334e-05,
"loss": 1.5836,
"step": 107500
},
{
"epoch": 2.16,
"learning_rate": 1.4051047512532206e-05,
"loss": 1.592,
"step": 108000
},
{
"epoch": 2.17,
"learning_rate": 1.3884617176942078e-05,
"loss": 1.5771,
"step": 108500
},
{
"epoch": 2.18,
"learning_rate": 1.3718186841351948e-05,
"loss": 1.576,
"step": 109000
},
{
"epoch": 2.19,
"learning_rate": 1.355175650576182e-05,
"loss": 1.5757,
"step": 109500
},
{
"epoch": 2.2,
"learning_rate": 1.3385326170171692e-05,
"loss": 1.5778,
"step": 110000
},
{
"epoch": 2.21,
"learning_rate": 1.3218895834581562e-05,
"loss": 1.5847,
"step": 110500
},
{
"epoch": 2.22,
"learning_rate": 1.3052465498991434e-05,
"loss": 1.5718,
"step": 111000
},
{
"epoch": 2.23,
"learning_rate": 1.2886035163401306e-05,
"loss": 1.5897,
"step": 111500
},
{
"epoch": 2.24,
"learning_rate": 1.2719604827811176e-05,
"loss": 1.5825,
"step": 112000
},
{
"epoch": 2.25,
"learning_rate": 1.2553174492221045e-05,
"loss": 1.5771,
"step": 112500
},
{
"epoch": 2.26,
"learning_rate": 1.2386744156630918e-05,
"loss": 1.5902,
"step": 113000
},
{
"epoch": 2.27,
"learning_rate": 1.2220313821040789e-05,
"loss": 1.5765,
"step": 113500
},
{
"epoch": 2.28,
"learning_rate": 1.205388348545066e-05,
"loss": 1.5616,
"step": 114000
},
{
"epoch": 2.29,
"learning_rate": 1.1887453149860532e-05,
"loss": 1.5899,
"step": 114500
},
{
"epoch": 2.3,
"learning_rate": 1.1721022814270403e-05,
"loss": 1.5867,
"step": 115000
},
{
"epoch": 2.31,
"learning_rate": 1.1554592478680275e-05,
"loss": 1.5767,
"step": 115500
},
{
"epoch": 2.32,
"learning_rate": 1.1388162143090146e-05,
"loss": 1.5749,
"step": 116000
},
{
"epoch": 2.33,
"learning_rate": 1.1221731807500018e-05,
"loss": 1.565,
"step": 116500
},
{
"epoch": 2.34,
"learning_rate": 1.1055301471909889e-05,
"loss": 1.58,
"step": 117000
},
{
"epoch": 2.35,
"learning_rate": 1.088887113631976e-05,
"loss": 1.5816,
"step": 117500
},
{
"epoch": 2.36,
"learning_rate": 1.0722440800729632e-05,
"loss": 1.567,
"step": 118000
},
{
"epoch": 2.37,
"learning_rate": 1.0556010465139502e-05,
"loss": 1.5742,
"step": 118500
},
{
"epoch": 2.38,
"learning_rate": 1.0389580129549374e-05,
"loss": 1.5801,
"step": 119000
},
{
"epoch": 2.39,
"learning_rate": 1.0223149793959246e-05,
"loss": 1.5861,
"step": 119500
},
{
"epoch": 2.4,
"learning_rate": 1.0056719458369116e-05,
"loss": 1.5791,
"step": 120000
},
{
"epoch": 2.41,
"learning_rate": 9.890289122778988e-06,
"loss": 1.5739,
"step": 120500
},
{
"epoch": 2.42,
"learning_rate": 9.72385878718886e-06,
"loss": 1.5635,
"step": 121000
},
{
"epoch": 2.43,
"learning_rate": 9.55742845159873e-06,
"loss": 1.6001,
"step": 121500
},
{
"epoch": 2.44,
"learning_rate": 9.390998116008602e-06,
"loss": 1.5741,
"step": 122000
},
{
"epoch": 2.45,
"learning_rate": 9.224567780418474e-06,
"loss": 1.584,
"step": 122500
},
{
"epoch": 2.46,
"learning_rate": 9.058137444828344e-06,
"loss": 1.5578,
"step": 123000
},
{
"epoch": 2.47,
"learning_rate": 8.891707109238215e-06,
"loss": 1.5687,
"step": 123500
},
{
"epoch": 2.48,
"learning_rate": 8.725276773648087e-06,
"loss": 1.5827,
"step": 124000
},
{
"epoch": 2.49,
"learning_rate": 8.558846438057957e-06,
"loss": 1.5667,
"step": 124500
},
{
"epoch": 2.5,
"learning_rate": 8.392416102467829e-06,
"loss": 1.5842,
"step": 125000
},
{
"epoch": 2.51,
"learning_rate": 8.2259857668777e-06,
"loss": 1.5794,
"step": 125500
},
{
"epoch": 2.52,
"learning_rate": 8.05955543128757e-06,
"loss": 1.5719,
"step": 126000
},
{
"epoch": 2.53,
"learning_rate": 7.893125095697443e-06,
"loss": 1.5677,
"step": 126500
},
{
"epoch": 2.54,
"learning_rate": 7.726694760107315e-06,
"loss": 1.5725,
"step": 127000
},
{
"epoch": 2.55,
"learning_rate": 7.560264424517186e-06,
"loss": 1.5599,
"step": 127500
},
{
"epoch": 2.56,
"learning_rate": 7.393834088927057e-06,
"loss": 1.5602,
"step": 128000
},
{
"epoch": 2.57,
"learning_rate": 7.2274037533369285e-06,
"loss": 1.5607,
"step": 128500
},
{
"epoch": 2.58,
"learning_rate": 7.0609734177467996e-06,
"loss": 1.5702,
"step": 129000
},
{
"epoch": 2.59,
"learning_rate": 6.894543082156671e-06,
"loss": 1.5839,
"step": 129500
},
{
"epoch": 2.6,
"learning_rate": 6.7281127465665425e-06,
"loss": 1.5836,
"step": 130000
},
{
"epoch": 2.61,
"learning_rate": 6.5616824109764136e-06,
"loss": 1.5695,
"step": 130500
},
{
"epoch": 2.62,
"learning_rate": 6.395252075386285e-06,
"loss": 1.6034,
"step": 131000
},
{
"epoch": 2.63,
"learning_rate": 6.2288217397961565e-06,
"loss": 1.5581,
"step": 131500
},
{
"epoch": 2.64,
"learning_rate": 6.0623914042060276e-06,
"loss": 1.5429,
"step": 132000
},
{
"epoch": 2.65,
"learning_rate": 5.895961068615899e-06,
"loss": 1.5632,
"step": 132500
},
{
"epoch": 2.66,
"learning_rate": 5.7295307330257705e-06,
"loss": 1.559,
"step": 133000
},
{
"epoch": 2.67,
"learning_rate": 5.5631003974356415e-06,
"loss": 1.5861,
"step": 133500
},
{
"epoch": 2.68,
"learning_rate": 5.396670061845513e-06,
"loss": 1.5525,
"step": 134000
},
{
"epoch": 2.69,
"learning_rate": 5.2302397262553845e-06,
"loss": 1.5857,
"step": 134500
},
{
"epoch": 2.7,
"learning_rate": 5.0638093906652555e-06,
"loss": 1.5693,
"step": 135000
},
{
"epoch": 2.71,
"learning_rate": 4.897379055075127e-06,
"loss": 1.5626,
"step": 135500
},
{
"epoch": 2.72,
"learning_rate": 4.7309487194849985e-06,
"loss": 1.5707,
"step": 136000
},
{
"epoch": 2.73,
"learning_rate": 4.5645183838948695e-06,
"loss": 1.5746,
"step": 136500
},
{
"epoch": 2.74,
"learning_rate": 4.3980880483047406e-06,
"loss": 1.5577,
"step": 137000
},
{
"epoch": 2.75,
"learning_rate": 4.2316577127146125e-06,
"loss": 1.5568,
"step": 137500
},
{
"epoch": 2.76,
"learning_rate": 4.0652273771244835e-06,
"loss": 1.5698,
"step": 138000
},
{
"epoch": 2.77,
"learning_rate": 3.898797041534355e-06,
"loss": 1.5706,
"step": 138500
},
{
"epoch": 2.78,
"learning_rate": 3.7323667059442265e-06,
"loss": 1.5721,
"step": 139000
},
{
"epoch": 2.79,
"learning_rate": 3.5659363703540975e-06,
"loss": 1.5468,
"step": 139500
},
{
"epoch": 2.8,
"learning_rate": 3.399506034763969e-06,
"loss": 1.564,
"step": 140000
},
{
"epoch": 2.81,
"learning_rate": 3.2330756991738396e-06,
"loss": 1.5583,
"step": 140500
},
{
"epoch": 2.82,
"learning_rate": 3.0666453635837115e-06,
"loss": 1.5468,
"step": 141000
},
{
"epoch": 2.83,
"learning_rate": 2.9002150279935825e-06,
"loss": 1.5648,
"step": 141500
},
{
"epoch": 2.84,
"learning_rate": 2.733784692403454e-06,
"loss": 1.5571,
"step": 142000
},
{
"epoch": 2.85,
"learning_rate": 2.5673543568133255e-06,
"loss": 1.5654,
"step": 142500
},
{
"epoch": 2.86,
"learning_rate": 2.4009240212231965e-06,
"loss": 1.5512,
"step": 143000
},
{
"epoch": 2.87,
"learning_rate": 2.2344936856330676e-06,
"loss": 1.5788,
"step": 143500
},
{
"epoch": 2.88,
"learning_rate": 2.068063350042939e-06,
"loss": 1.5723,
"step": 144000
},
{
"epoch": 2.89,
"learning_rate": 1.9016330144528103e-06,
"loss": 1.5497,
"step": 144500
},
{
"epoch": 2.9,
"learning_rate": 1.7352026788626818e-06,
"loss": 1.5602,
"step": 145000
},
{
"epoch": 2.91,
"learning_rate": 1.568772343272553e-06,
"loss": 1.5617,
"step": 145500
},
{
"epoch": 2.92,
"learning_rate": 1.4023420076824243e-06,
"loss": 1.5659,
"step": 146000
},
{
"epoch": 2.93,
"learning_rate": 1.2359116720922958e-06,
"loss": 1.5563,
"step": 146500
},
{
"epoch": 2.94,
"learning_rate": 1.069481336502167e-06,
"loss": 1.5568,
"step": 147000
},
{
"epoch": 2.95,
"learning_rate": 9.030510009120382e-07,
"loss": 1.5683,
"step": 147500
},
{
"epoch": 2.96,
"learning_rate": 7.366206653219095e-07,
"loss": 1.5773,
"step": 148000
},
{
"epoch": 2.97,
"learning_rate": 5.701903297317809e-07,
"loss": 1.5635,
"step": 148500
},
{
"epoch": 2.98,
"learning_rate": 4.037599941416522e-07,
"loss": 1.5423,
"step": 149000
},
{
"epoch": 2.99,
"learning_rate": 2.373296585515235e-07,
"loss": 1.5413,
"step": 149500
},
{
"epoch": 3.0,
"learning_rate": 7.089932296139482e-08,
"loss": 1.5445,
"step": 150000
},
{
"epoch": 3.0,
"step": 150213,
"total_flos": 9.084631933156147e+17,
"train_runtime": 69781.1584,
"train_samples_per_second": 2.153
}
],
"max_steps": 150213,
"num_train_epochs": 3,
"total_flos": 9.084631933156147e+17,
"trial_name": null,
"trial_params": null
}