1262 lines
25 KiB
JSON
1262 lines
25 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 3.0,
|
|
"global_step": 103194,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9758706901564046e-05,
|
|
"loss": 1.2775,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 4.951644475453999e-05,
|
|
"loss": 1.1976,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 4.9274182607515944e-05,
|
|
"loss": 1.1913,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 4.903192046049189e-05,
|
|
"loss": 1.1546,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 4.878965831346784e-05,
|
|
"loss": 1.1645,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 4.854739616644379e-05,
|
|
"loss": 1.1327,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 4.8305134019419734e-05,
|
|
"loss": 1.138,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.806287187239568e-05,
|
|
"loss": 1.1208,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.782060972537163e-05,
|
|
"loss": 1.1307,
|
|
"step": 4500
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.757883210264163e-05,
|
|
"loss": 1.0888,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.7336569955617576e-05,
|
|
"loss": 1.1117,
|
|
"step": 5500
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.709430780859352e-05,
|
|
"loss": 1.0842,
|
|
"step": 6000
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.6852045661569474e-05,
|
|
"loss": 1.0715,
|
|
"step": 6500
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.6609783514545426e-05,
|
|
"loss": 1.0686,
|
|
"step": 7000
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.636752136752137e-05,
|
|
"loss": 1.0784,
|
|
"step": 7500
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.612574374479136e-05,
|
|
"loss": 1.0398,
|
|
"step": 8000
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.5883481597767316e-05,
|
|
"loss": 1.094,
|
|
"step": 8500
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.564121945074326e-05,
|
|
"loss": 1.0552,
|
|
"step": 9000
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.539895730371921e-05,
|
|
"loss": 1.0478,
|
|
"step": 9500
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.515669515669516e-05,
|
|
"loss": 1.0795,
|
|
"step": 10000
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4914433009671105e-05,
|
|
"loss": 1.08,
|
|
"step": 10500
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.467217086264706e-05,
|
|
"loss": 1.0546,
|
|
"step": 11000
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.443039323991705e-05,
|
|
"loss": 1.0541,
|
|
"step": 11500
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.4188131092892995e-05,
|
|
"loss": 1.0545,
|
|
"step": 12000
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.394586894586895e-05,
|
|
"loss": 1.0184,
|
|
"step": 12500
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.37036067988449e-05,
|
|
"loss": 1.0185,
|
|
"step": 13000
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.3461344651820846e-05,
|
|
"loss": 1.0318,
|
|
"step": 13500
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.321908250479679e-05,
|
|
"loss": 1.0203,
|
|
"step": 14000
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.297730488206679e-05,
|
|
"loss": 1.0448,
|
|
"step": 14500
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.2735042735042735e-05,
|
|
"loss": 1.0221,
|
|
"step": 15000
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.249278058801869e-05,
|
|
"loss": 1.0343,
|
|
"step": 15500
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.225051844099463e-05,
|
|
"loss": 1.0321,
|
|
"step": 16000
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.2008256293970586e-05,
|
|
"loss": 0.9953,
|
|
"step": 16500
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.176647867124058e-05,
|
|
"loss": 1.0267,
|
|
"step": 17000
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.152421652421652e-05,
|
|
"loss": 0.9998,
|
|
"step": 17500
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.1281954377192475e-05,
|
|
"loss": 1.0185,
|
|
"step": 18000
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.103969223016843e-05,
|
|
"loss": 0.9885,
|
|
"step": 18500
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.0797430083144373e-05,
|
|
"loss": 1.0121,
|
|
"step": 19000
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.055516793612032e-05,
|
|
"loss": 1.0013,
|
|
"step": 19500
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.031339031339032e-05,
|
|
"loss": 0.9757,
|
|
"step": 20000
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.007161269066031e-05,
|
|
"loss": 0.9778,
|
|
"step": 20500
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 3.982935054363626e-05,
|
|
"loss": 1.0025,
|
|
"step": 21000
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 3.9587088396612206e-05,
|
|
"loss": 0.9886,
|
|
"step": 21500
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 3.934482624958816e-05,
|
|
"loss": 0.9837,
|
|
"step": 22000
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 3.9102564102564105e-05,
|
|
"loss": 0.9719,
|
|
"step": 22500
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 3.886030195554005e-05,
|
|
"loss": 0.9499,
|
|
"step": 23000
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 3.8618039808515996e-05,
|
|
"loss": 1.0089,
|
|
"step": 23500
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 3.837577766149195e-05,
|
|
"loss": 0.9395,
|
|
"step": 24000
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 3.81335155144679e-05,
|
|
"loss": 0.972,
|
|
"step": 24500
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 3.789125336744385e-05,
|
|
"loss": 0.9834,
|
|
"step": 25000
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 3.764899122041979e-05,
|
|
"loss": 0.9891,
|
|
"step": 25500
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 3.740672907339574e-05,
|
|
"loss": 0.9754,
|
|
"step": 26000
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 3.7164951450665736e-05,
|
|
"loss": 0.953,
|
|
"step": 26500
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 3.6923173827935734e-05,
|
|
"loss": 0.9864,
|
|
"step": 27000
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 3.668091168091169e-05,
|
|
"loss": 0.9629,
|
|
"step": 27500
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 3.643864953388763e-05,
|
|
"loss": 0.949,
|
|
"step": 28000
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 3.619638738686358e-05,
|
|
"loss": 0.9889,
|
|
"step": 28500
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 3.5954125239839524e-05,
|
|
"loss": 0.9695,
|
|
"step": 29000
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 3.571234761710953e-05,
|
|
"loss": 0.9723,
|
|
"step": 29500
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 3.5470085470085474e-05,
|
|
"loss": 0.9455,
|
|
"step": 30000
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 3.522782332306142e-05,
|
|
"loss": 0.9717,
|
|
"step": 30500
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 3.498604570033142e-05,
|
|
"loss": 0.9517,
|
|
"step": 31000
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 3.4743783553307364e-05,
|
|
"loss": 0.9537,
|
|
"step": 31500
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 3.450152140628331e-05,
|
|
"loss": 0.9502,
|
|
"step": 32000
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 3.425925925925926e-05,
|
|
"loss": 0.9476,
|
|
"step": 32500
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 3.401699711223521e-05,
|
|
"loss": 0.9534,
|
|
"step": 33000
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 3.377473496521116e-05,
|
|
"loss": 0.9383,
|
|
"step": 33500
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 3.3532472818187106e-05,
|
|
"loss": 0.9501,
|
|
"step": 34000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.329021067116305e-05,
|
|
"loss": 0.9316,
|
|
"step": 34500
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3047948524139e-05,
|
|
"loss": 0.8742,
|
|
"step": 35000
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 3.280568637711495e-05,
|
|
"loss": 0.8706,
|
|
"step": 35500
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 3.25634242300909e-05,
|
|
"loss": 0.8548,
|
|
"step": 36000
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 3.232116208306685e-05,
|
|
"loss": 0.8637,
|
|
"step": 36500
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 3.2078899936042794e-05,
|
|
"loss": 0.8547,
|
|
"step": 37000
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 3.183712231331279e-05,
|
|
"loss": 0.8346,
|
|
"step": 37500
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 3.159486016628874e-05,
|
|
"loss": 0.8504,
|
|
"step": 38000
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 3.1353082543558736e-05,
|
|
"loss": 0.8376,
|
|
"step": 38500
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 3.111082039653468e-05,
|
|
"loss": 0.8508,
|
|
"step": 39000
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 3.086904277380468e-05,
|
|
"loss": 0.8455,
|
|
"step": 39500
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 3.0626780626780625e-05,
|
|
"loss": 0.8758,
|
|
"step": 40000
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 3.0384518479756574e-05,
|
|
"loss": 0.8494,
|
|
"step": 40500
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 3.0142256332732527e-05,
|
|
"loss": 0.8565,
|
|
"step": 41000
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 2.9899994185708476e-05,
|
|
"loss": 0.8372,
|
|
"step": 41500
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 2.965773203868442e-05,
|
|
"loss": 0.8397,
|
|
"step": 42000
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 2.9415469891660367e-05,
|
|
"loss": 0.8375,
|
|
"step": 42500
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 2.9173207744636316e-05,
|
|
"loss": 0.8425,
|
|
"step": 43000
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 2.8930945597612262e-05,
|
|
"loss": 0.8567,
|
|
"step": 43500
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 2.8689167974882263e-05,
|
|
"loss": 0.8474,
|
|
"step": 44000
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 2.8446905827858212e-05,
|
|
"loss": 0.8407,
|
|
"step": 44500
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 2.8204643680834158e-05,
|
|
"loss": 0.8217,
|
|
"step": 45000
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 2.7962381533810107e-05,
|
|
"loss": 0.8548,
|
|
"step": 45500
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 2.7720119386786053e-05,
|
|
"loss": 0.8464,
|
|
"step": 46000
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 2.7477857239762e-05,
|
|
"loss": 0.8355,
|
|
"step": 46500
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 2.7235595092737955e-05,
|
|
"loss": 0.8189,
|
|
"step": 47000
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 2.69933329457139e-05,
|
|
"loss": 0.8269,
|
|
"step": 47500
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 2.6751070798689846e-05,
|
|
"loss": 0.8461,
|
|
"step": 48000
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 2.6509293175959844e-05,
|
|
"loss": 0.8303,
|
|
"step": 48500
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 2.626703102893579e-05,
|
|
"loss": 0.8423,
|
|
"step": 49000
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 2.6024768881911742e-05,
|
|
"loss": 0.8536,
|
|
"step": 49500
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 2.578250673488769e-05,
|
|
"loss": 0.8416,
|
|
"step": 50000
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 2.5540729112157686e-05,
|
|
"loss": 0.835,
|
|
"step": 50500
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 2.529895148942768e-05,
|
|
"loss": 0.8165,
|
|
"step": 51000
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 2.5056689342403626e-05,
|
|
"loss": 0.8275,
|
|
"step": 51500
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 2.481442719537958e-05,
|
|
"loss": 0.8178,
|
|
"step": 52000
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 2.4572165048355525e-05,
|
|
"loss": 0.8379,
|
|
"step": 52500
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 2.4329902901331474e-05,
|
|
"loss": 0.8376,
|
|
"step": 53000
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 2.4087640754307423e-05,
|
|
"loss": 0.8145,
|
|
"step": 53500
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 2.384537860728337e-05,
|
|
"loss": 0.8225,
|
|
"step": 54000
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 2.360311646025932e-05,
|
|
"loss": 0.8211,
|
|
"step": 54500
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 2.336182336182336e-05,
|
|
"loss": 0.8277,
|
|
"step": 55000
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 2.3119561214799313e-05,
|
|
"loss": 0.816,
|
|
"step": 55500
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 2.287729906777526e-05,
|
|
"loss": 0.8436,
|
|
"step": 56000
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 2.263503692075121e-05,
|
|
"loss": 0.823,
|
|
"step": 56500
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 2.2392774773727157e-05,
|
|
"loss": 0.8112,
|
|
"step": 57000
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 2.2150997150997152e-05,
|
|
"loss": 0.8218,
|
|
"step": 57500
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 2.1908735003973098e-05,
|
|
"loss": 0.8144,
|
|
"step": 58000
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 2.166647285694905e-05,
|
|
"loss": 0.8211,
|
|
"step": 58500
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 2.1424210709924996e-05,
|
|
"loss": 0.8268,
|
|
"step": 59000
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 2.1181948562900945e-05,
|
|
"loss": 0.8277,
|
|
"step": 59500
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 2.0939686415876894e-05,
|
|
"loss": 0.799,
|
|
"step": 60000
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 2.069790879314689e-05,
|
|
"loss": 0.8093,
|
|
"step": 60500
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 2.0455646646122838e-05,
|
|
"loss": 0.8101,
|
|
"step": 61000
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 2.0213384499098787e-05,
|
|
"loss": 0.8023,
|
|
"step": 61500
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 1.9971122352074733e-05,
|
|
"loss": 0.804,
|
|
"step": 62000
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 1.9728860205050682e-05,
|
|
"loss": 0.7897,
|
|
"step": 62500
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 1.948659805802663e-05,
|
|
"loss": 0.7894,
|
|
"step": 63000
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 1.9244335911002577e-05,
|
|
"loss": 0.8108,
|
|
"step": 63500
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 1.9002073763978526e-05,
|
|
"loss": 0.8008,
|
|
"step": 64000
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 1.8759811616954475e-05,
|
|
"loss": 0.8076,
|
|
"step": 64500
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 1.851803399422447e-05,
|
|
"loss": 0.8178,
|
|
"step": 65000
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 1.8275771847200422e-05,
|
|
"loss": 0.7893,
|
|
"step": 65500
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 1.8033994224470417e-05,
|
|
"loss": 0.8018,
|
|
"step": 66000
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 1.7791732077446362e-05,
|
|
"loss": 0.793,
|
|
"step": 66500
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 1.7549469930422315e-05,
|
|
"loss": 0.8107,
|
|
"step": 67000
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 1.730720778339826e-05,
|
|
"loss": 0.7952,
|
|
"step": 67500
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 1.7064945636374206e-05,
|
|
"loss": 0.7898,
|
|
"step": 68000
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 1.6823168013644204e-05,
|
|
"loss": 0.7949,
|
|
"step": 68500
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 1.6580905866620153e-05,
|
|
"loss": 0.7728,
|
|
"step": 69000
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 1.63386437195961e-05,
|
|
"loss": 0.7451,
|
|
"step": 69500
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 1.609638157257205e-05,
|
|
"loss": 0.7305,
|
|
"step": 70000
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 1.5854119425547997e-05,
|
|
"loss": 0.752,
|
|
"step": 70500
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 1.5611857278523946e-05,
|
|
"loss": 0.7539,
|
|
"step": 71000
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 1.537007965579394e-05,
|
|
"loss": 0.7271,
|
|
"step": 71500
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 1.512781750876989e-05,
|
|
"loss": 0.7399,
|
|
"step": 72000
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 1.488555536174584e-05,
|
|
"loss": 0.7398,
|
|
"step": 72500
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 1.4643293214721787e-05,
|
|
"loss": 0.7528,
|
|
"step": 73000
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 1.4401031067697734e-05,
|
|
"loss": 0.7324,
|
|
"step": 73500
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 1.4159253444967732e-05,
|
|
"loss": 0.7351,
|
|
"step": 74000
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 1.391699129794368e-05,
|
|
"loss": 0.7198,
|
|
"step": 74500
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 1.3674729150919627e-05,
|
|
"loss": 0.7299,
|
|
"step": 75000
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 1.3432467003895578e-05,
|
|
"loss": 0.7257,
|
|
"step": 75500
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 1.3190689381165572e-05,
|
|
"loss": 0.719,
|
|
"step": 76000
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 1.294842723414152e-05,
|
|
"loss": 0.7149,
|
|
"step": 76500
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 1.2706165087117469e-05,
|
|
"loss": 0.7284,
|
|
"step": 77000
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 1.2463902940093416e-05,
|
|
"loss": 0.727,
|
|
"step": 77500
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 1.2221640793069365e-05,
|
|
"loss": 0.726,
|
|
"step": 78000
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 1.1979378646045314e-05,
|
|
"loss": 0.7231,
|
|
"step": 78500
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 1.173760102331531e-05,
|
|
"loss": 0.7367,
|
|
"step": 79000
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 1.1495338876291258e-05,
|
|
"loss": 0.7275,
|
|
"step": 79500
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 1.1253076729267206e-05,
|
|
"loss": 0.7134,
|
|
"step": 80000
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 1.1010814582243153e-05,
|
|
"loss": 0.7233,
|
|
"step": 80500
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 1.0768552435219102e-05,
|
|
"loss": 0.7203,
|
|
"step": 81000
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 1.0526290288195051e-05,
|
|
"loss": 0.7286,
|
|
"step": 81500
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 1.0284028141170999e-05,
|
|
"loss": 0.7242,
|
|
"step": 82000
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 1.0042250518440995e-05,
|
|
"loss": 0.7058,
|
|
"step": 82500
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 9.799988371416944e-06,
|
|
"loss": 0.7284,
|
|
"step": 83000
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 9.557726224392891e-06,
|
|
"loss": 0.7134,
|
|
"step": 83500
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 9.315464077368839e-06,
|
|
"loss": 0.7153,
|
|
"step": 84000
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 9.073201930344788e-06,
|
|
"loss": 0.739,
|
|
"step": 84500
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 8.830939783320737e-06,
|
|
"loss": 0.7145,
|
|
"step": 85000
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 8.589162160590733e-06,
|
|
"loss": 0.6977,
|
|
"step": 85500
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 8.34690001356668e-06,
|
|
"loss": 0.708,
|
|
"step": 86000
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 8.104637866542628e-06,
|
|
"loss": 0.7169,
|
|
"step": 86500
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 7.862375719518577e-06,
|
|
"loss": 0.7132,
|
|
"step": 87000
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 7.620113572494525e-06,
|
|
"loss": 0.6992,
|
|
"step": 87500
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 7.377851425470474e-06,
|
|
"loss": 0.7243,
|
|
"step": 88000
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 7.135589278446422e-06,
|
|
"loss": 0.718,
|
|
"step": 88500
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 6.8938116557164175e-06,
|
|
"loss": 0.7107,
|
|
"step": 89000
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 6.651549508692366e-06,
|
|
"loss": 0.721,
|
|
"step": 89500
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 6.409287361668315e-06,
|
|
"loss": 0.7181,
|
|
"step": 90000
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 6.167025214644262e-06,
|
|
"loss": 0.696,
|
|
"step": 90500
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 5.924763067620211e-06,
|
|
"loss": 0.6889,
|
|
"step": 91000
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 5.682500920596159e-06,
|
|
"loss": 0.7049,
|
|
"step": 91500
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 5.440723297866155e-06,
|
|
"loss": 0.711,
|
|
"step": 92000
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 5.198461150842103e-06,
|
|
"loss": 0.7207,
|
|
"step": 92500
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 4.956199003818052e-06,
|
|
"loss": 0.7247,
|
|
"step": 93000
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 4.713936856794e-06,
|
|
"loss": 0.7193,
|
|
"step": 93500
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 4.472159234063996e-06,
|
|
"loss": 0.7173,
|
|
"step": 94000
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 4.2298970870399444e-06,
|
|
"loss": 0.6999,
|
|
"step": 94500
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 3.987634940015893e-06,
|
|
"loss": 0.6986,
|
|
"step": 95000
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 3.7453727929918406e-06,
|
|
"loss": 0.7106,
|
|
"step": 95500
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 3.503110645967789e-06,
|
|
"loss": 0.702,
|
|
"step": 96000
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 3.2608484989437376e-06,
|
|
"loss": 0.7244,
|
|
"step": 96500
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 3.0185863519196854e-06,
|
|
"loss": 0.7003,
|
|
"step": 97000
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 2.7763242048956337e-06,
|
|
"loss": 0.7281,
|
|
"step": 97500
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 2.5340620578715816e-06,
|
|
"loss": 0.7071,
|
|
"step": 98000
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 2.292284435141578e-06,
|
|
"loss": 0.7211,
|
|
"step": 98500
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 2.0500222881175265e-06,
|
|
"loss": 0.6968,
|
|
"step": 99000
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 1.8077601410934743e-06,
|
|
"loss": 0.7194,
|
|
"step": 99500
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 1.5654979940694228e-06,
|
|
"loss": 0.7187,
|
|
"step": 100000
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 1.323235847045371e-06,
|
|
"loss": 0.7178,
|
|
"step": 100500
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 1.0809737000213192e-06,
|
|
"loss": 0.7299,
|
|
"step": 101000
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 8.391960772913154e-07,
|
|
"loss": 0.7151,
|
|
"step": 101500
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 5.969339302672636e-07,
|
|
"loss": 0.7176,
|
|
"step": 102000
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 3.5467178324321185e-07,
|
|
"loss": 0.7078,
|
|
"step": 102500
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 1.1289416051320813e-07,
|
|
"loss": 0.6934,
|
|
"step": 103000
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"step": 103194,
|
|
"total_flos": 1.9167267894224486e+17,
|
|
"train_loss": 0.8582985611445433,
|
|
"train_runtime": 45981.5166,
|
|
"train_samples_per_second": 2.244,
|
|
"train_steps_per_second": 2.244
|
|
}
|
|
],
|
|
"max_steps": 103194,
|
|
"num_train_epochs": 3,
|
|
"total_flos": 1.9167267894224486e+17,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|