electra-large-squad2 / trainer_state.json
Ubuntu
initial upload
721cb8e
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 8.0,
"global_step": 65880,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.06,
"learning_rate": 2.9772313296903464e-05,
"loss": 5.8692,
"step": 500
},
{
"epoch": 0.12,
"learning_rate": 2.954462659380692e-05,
"loss": 5.9529,
"step": 1000
},
{
"epoch": 0.18,
"learning_rate": 2.9316939890710385e-05,
"loss": 5.9524,
"step": 1500
},
{
"epoch": 0.24,
"learning_rate": 2.9089253187613845e-05,
"loss": 5.9525,
"step": 2000
},
{
"epoch": 0.3,
"learning_rate": 2.8861566484517305e-05,
"loss": 5.9517,
"step": 2500
},
{
"epoch": 0.36,
"learning_rate": 2.8633879781420765e-05,
"loss": 5.9514,
"step": 3000
},
{
"epoch": 0.43,
"learning_rate": 2.8406193078324225e-05,
"loss": 5.9514,
"step": 3500
},
{
"epoch": 0.49,
"learning_rate": 2.817850637522769e-05,
"loss": 5.9515,
"step": 4000
},
{
"epoch": 0.55,
"learning_rate": 2.795081967213115e-05,
"loss": 5.9517,
"step": 4500
},
{
"epoch": 0.61,
"learning_rate": 2.772313296903461e-05,
"loss": 5.9513,
"step": 5000
},
{
"epoch": 0.67,
"learning_rate": 2.749544626593807e-05,
"loss": 5.9504,
"step": 5500
},
{
"epoch": 0.73,
"learning_rate": 2.7267759562841533e-05,
"loss": 5.9516,
"step": 6000
},
{
"epoch": 0.79,
"learning_rate": 2.704007285974499e-05,
"loss": 5.951,
"step": 6500
},
{
"epoch": 0.85,
"learning_rate": 2.6812386156648453e-05,
"loss": 5.952,
"step": 7000
},
{
"epoch": 0.91,
"learning_rate": 2.6584699453551913e-05,
"loss": 5.9522,
"step": 7500
},
{
"epoch": 0.97,
"learning_rate": 2.6357012750455373e-05,
"loss": 5.9519,
"step": 8000
},
{
"epoch": 1.03,
"learning_rate": 2.6129326047358837e-05,
"loss": 5.9517,
"step": 8500
},
{
"epoch": 1.09,
"learning_rate": 2.5901639344262294e-05,
"loss": 5.9518,
"step": 9000
},
{
"epoch": 1.15,
"learning_rate": 2.5673952641165757e-05,
"loss": 5.9518,
"step": 9500
},
{
"epoch": 1.21,
"learning_rate": 2.5446265938069218e-05,
"loss": 5.9513,
"step": 10000
},
{
"epoch": 1.28,
"learning_rate": 2.5218579234972678e-05,
"loss": 5.9512,
"step": 10500
},
{
"epoch": 1.34,
"learning_rate": 2.4990892531876138e-05,
"loss": 5.9514,
"step": 11000
},
{
"epoch": 1.4,
"learning_rate": 2.47632058287796e-05,
"loss": 5.952,
"step": 11500
},
{
"epoch": 1.46,
"learning_rate": 2.453551912568306e-05,
"loss": 5.9509,
"step": 12000
},
{
"epoch": 1.52,
"learning_rate": 2.4307832422586522e-05,
"loss": 5.9516,
"step": 12500
},
{
"epoch": 1.58,
"learning_rate": 2.4080145719489982e-05,
"loss": 5.9517,
"step": 13000
},
{
"epoch": 1.64,
"learning_rate": 2.3852459016393442e-05,
"loss": 5.9501,
"step": 13500
},
{
"epoch": 1.7,
"learning_rate": 2.3624772313296906e-05,
"loss": 5.9515,
"step": 14000
},
{
"epoch": 1.76,
"learning_rate": 2.3397085610200362e-05,
"loss": 5.9505,
"step": 14500
},
{
"epoch": 1.82,
"learning_rate": 2.3169398907103826e-05,
"loss": 5.9507,
"step": 15000
},
{
"epoch": 1.88,
"learning_rate": 2.294171220400729e-05,
"loss": 5.9512,
"step": 15500
},
{
"epoch": 1.94,
"learning_rate": 2.2714025500910746e-05,
"loss": 5.9511,
"step": 16000
},
{
"epoch": 2.0,
"learning_rate": 2.248633879781421e-05,
"loss": 5.9514,
"step": 16500
},
{
"epoch": 2.06,
"learning_rate": 2.2258652094717667e-05,
"loss": 5.9503,
"step": 17000
},
{
"epoch": 2.13,
"learning_rate": 2.203096539162113e-05,
"loss": 5.9518,
"step": 17500
},
{
"epoch": 2.19,
"learning_rate": 2.180327868852459e-05,
"loss": 5.9514,
"step": 18000
},
{
"epoch": 2.25,
"learning_rate": 2.157559198542805e-05,
"loss": 5.9518,
"step": 18500
},
{
"epoch": 2.31,
"learning_rate": 2.1347905282331514e-05,
"loss": 5.9517,
"step": 19000
},
{
"epoch": 2.37,
"learning_rate": 2.1120218579234974e-05,
"loss": 5.9513,
"step": 19500
},
{
"epoch": 2.43,
"learning_rate": 2.0892531876138434e-05,
"loss": 5.9503,
"step": 20000
},
{
"epoch": 2.49,
"learning_rate": 2.0664845173041894e-05,
"loss": 5.9513,
"step": 20500
},
{
"epoch": 2.55,
"learning_rate": 2.0437158469945358e-05,
"loss": 5.9515,
"step": 21000
},
{
"epoch": 2.61,
"learning_rate": 2.0209471766848815e-05,
"loss": 5.9511,
"step": 21500
},
{
"epoch": 2.67,
"learning_rate": 1.998178506375228e-05,
"loss": 5.9509,
"step": 22000
},
{
"epoch": 2.73,
"learning_rate": 1.975409836065574e-05,
"loss": 5.9515,
"step": 22500
},
{
"epoch": 2.79,
"learning_rate": 1.95264116575592e-05,
"loss": 5.951,
"step": 23000
},
{
"epoch": 2.85,
"learning_rate": 1.9298724954462662e-05,
"loss": 5.9511,
"step": 23500
},
{
"epoch": 2.91,
"learning_rate": 1.907103825136612e-05,
"loss": 5.9511,
"step": 24000
},
{
"epoch": 2.98,
"learning_rate": 1.8843351548269583e-05,
"loss": 5.9504,
"step": 24500
},
{
"epoch": 3.04,
"learning_rate": 1.8615664845173043e-05,
"loss": 5.9507,
"step": 25000
},
{
"epoch": 3.1,
"learning_rate": 1.8387978142076503e-05,
"loss": 5.9506,
"step": 25500
},
{
"epoch": 3.16,
"learning_rate": 1.8160291438979963e-05,
"loss": 5.9515,
"step": 26000
},
{
"epoch": 3.22,
"learning_rate": 1.7932604735883423e-05,
"loss": 5.9509,
"step": 26500
},
{
"epoch": 3.28,
"learning_rate": 1.7704918032786887e-05,
"loss": 5.9505,
"step": 27000
},
{
"epoch": 3.34,
"learning_rate": 1.7477231329690347e-05,
"loss": 5.951,
"step": 27500
},
{
"epoch": 3.4,
"learning_rate": 1.7249544626593807e-05,
"loss": 5.9512,
"step": 28000
},
{
"epoch": 3.46,
"learning_rate": 1.7021857923497267e-05,
"loss": 5.9511,
"step": 28500
},
{
"epoch": 3.52,
"learning_rate": 1.679417122040073e-05,
"loss": 5.9507,
"step": 29000
},
{
"epoch": 3.58,
"learning_rate": 1.6566484517304187e-05,
"loss": 5.9511,
"step": 29500
},
{
"epoch": 3.64,
"learning_rate": 1.633879781420765e-05,
"loss": 5.9507,
"step": 30000
},
{
"epoch": 3.7,
"learning_rate": 1.6111111111111115e-05,
"loss": 5.9514,
"step": 30500
},
{
"epoch": 3.76,
"learning_rate": 1.588342440801457e-05,
"loss": 5.9511,
"step": 31000
},
{
"epoch": 3.83,
"learning_rate": 1.5655737704918035e-05,
"loss": 5.9509,
"step": 31500
},
{
"epoch": 3.89,
"learning_rate": 1.542805100182149e-05,
"loss": 5.9508,
"step": 32000
},
{
"epoch": 3.95,
"learning_rate": 1.5200364298724955e-05,
"loss": 5.9511,
"step": 32500
},
{
"epoch": 4.01,
"learning_rate": 1.4972677595628415e-05,
"loss": 5.9518,
"step": 33000
},
{
"epoch": 4.07,
"learning_rate": 1.4744990892531877e-05,
"loss": 5.9518,
"step": 33500
},
{
"epoch": 4.13,
"learning_rate": 1.4517304189435337e-05,
"loss": 5.9512,
"step": 34000
},
{
"epoch": 4.19,
"learning_rate": 1.4289617486338798e-05,
"loss": 5.9512,
"step": 34500
},
{
"epoch": 4.25,
"learning_rate": 1.4061930783242258e-05,
"loss": 5.9508,
"step": 35000
},
{
"epoch": 4.31,
"learning_rate": 1.383424408014572e-05,
"loss": 5.951,
"step": 35500
},
{
"epoch": 4.37,
"learning_rate": 1.3606557377049181e-05,
"loss": 5.9514,
"step": 36000
},
{
"epoch": 4.43,
"learning_rate": 1.3378870673952642e-05,
"loss": 5.9505,
"step": 36500
},
{
"epoch": 4.49,
"learning_rate": 1.3151183970856102e-05,
"loss": 5.9508,
"step": 37000
},
{
"epoch": 4.55,
"learning_rate": 1.2923497267759564e-05,
"loss": 5.9513,
"step": 37500
},
{
"epoch": 4.61,
"learning_rate": 1.2695810564663024e-05,
"loss": 5.9513,
"step": 38000
},
{
"epoch": 4.68,
"learning_rate": 1.2468123861566484e-05,
"loss": 5.9503,
"step": 38500
},
{
"epoch": 4.74,
"learning_rate": 1.2240437158469946e-05,
"loss": 5.9508,
"step": 39000
},
{
"epoch": 4.8,
"learning_rate": 1.2012750455373408e-05,
"loss": 5.9504,
"step": 39500
},
{
"epoch": 4.86,
"learning_rate": 1.1785063752276868e-05,
"loss": 5.9503,
"step": 40000
},
{
"epoch": 4.92,
"learning_rate": 1.1557377049180328e-05,
"loss": 5.951,
"step": 40500
},
{
"epoch": 4.98,
"learning_rate": 1.132969034608379e-05,
"loss": 5.9514,
"step": 41000
},
{
"epoch": 5.04,
"learning_rate": 1.110200364298725e-05,
"loss": 5.9507,
"step": 41500
},
{
"epoch": 5.1,
"learning_rate": 1.087431693989071e-05,
"loss": 5.9505,
"step": 42000
},
{
"epoch": 5.16,
"learning_rate": 1.064663023679417e-05,
"loss": 5.9509,
"step": 42500
},
{
"epoch": 5.22,
"learning_rate": 1.0418943533697632e-05,
"loss": 5.9506,
"step": 43000
},
{
"epoch": 5.28,
"learning_rate": 1.0191256830601094e-05,
"loss": 5.9511,
"step": 43500
},
{
"epoch": 5.34,
"learning_rate": 9.963570127504554e-06,
"loss": 5.9505,
"step": 44000
},
{
"epoch": 5.4,
"learning_rate": 9.735883424408014e-06,
"loss": 5.951,
"step": 44500
},
{
"epoch": 5.46,
"learning_rate": 9.508196721311476e-06,
"loss": 5.9507,
"step": 45000
},
{
"epoch": 5.53,
"learning_rate": 9.280510018214936e-06,
"loss": 5.9509,
"step": 45500
},
{
"epoch": 5.59,
"learning_rate": 9.052823315118397e-06,
"loss": 5.9509,
"step": 46000
},
{
"epoch": 5.65,
"learning_rate": 8.825136612021857e-06,
"loss": 5.9514,
"step": 46500
},
{
"epoch": 5.71,
"learning_rate": 8.59744990892532e-06,
"loss": 5.9506,
"step": 47000
},
{
"epoch": 5.77,
"learning_rate": 8.36976320582878e-06,
"loss": 5.9511,
"step": 47500
},
{
"epoch": 5.83,
"learning_rate": 8.14207650273224e-06,
"loss": 5.9514,
"step": 48000
},
{
"epoch": 5.89,
"learning_rate": 7.914389799635702e-06,
"loss": 5.9506,
"step": 48500
},
{
"epoch": 5.95,
"learning_rate": 7.686703096539163e-06,
"loss": 5.9503,
"step": 49000
},
{
"epoch": 6.01,
"learning_rate": 7.459016393442623e-06,
"loss": 5.9505,
"step": 49500
},
{
"epoch": 6.07,
"learning_rate": 7.2313296903460846e-06,
"loss": 5.9509,
"step": 50000
},
{
"epoch": 6.13,
"learning_rate": 7.003642987249545e-06,
"loss": 5.9507,
"step": 50500
},
{
"epoch": 6.19,
"learning_rate": 6.775956284153006e-06,
"loss": 5.9506,
"step": 51000
},
{
"epoch": 6.25,
"learning_rate": 6.548269581056467e-06,
"loss": 5.9508,
"step": 51500
},
{
"epoch": 6.31,
"learning_rate": 6.320582877959928e-06,
"loss": 5.9508,
"step": 52000
},
{
"epoch": 6.38,
"learning_rate": 6.092896174863388e-06,
"loss": 5.9508,
"step": 52500
},
{
"epoch": 6.44,
"learning_rate": 5.865209471766849e-06,
"loss": 5.9509,
"step": 53000
},
{
"epoch": 6.5,
"learning_rate": 5.63752276867031e-06,
"loss": 5.951,
"step": 53500
},
{
"epoch": 6.56,
"learning_rate": 5.409836065573771e-06,
"loss": 5.9509,
"step": 54000
},
{
"epoch": 6.62,
"learning_rate": 5.182149362477231e-06,
"loss": 5.9505,
"step": 54500
},
{
"epoch": 6.68,
"learning_rate": 4.954462659380692e-06,
"loss": 5.9511,
"step": 55000
},
{
"epoch": 6.74,
"learning_rate": 4.726775956284153e-06,
"loss": 5.951,
"step": 55500
},
{
"epoch": 6.8,
"learning_rate": 4.499089253187614e-06,
"loss": 5.9504,
"step": 56000
},
{
"epoch": 6.86,
"learning_rate": 4.271402550091074e-06,
"loss": 5.9507,
"step": 56500
},
{
"epoch": 6.92,
"learning_rate": 4.043715846994535e-06,
"loss": 5.9507,
"step": 57000
},
{
"epoch": 6.98,
"learning_rate": 3.816029143897996e-06,
"loss": 5.9506,
"step": 57500
},
{
"epoch": 7.04,
"learning_rate": 3.5883424408014573e-06,
"loss": 5.9504,
"step": 58000
},
{
"epoch": 7.1,
"learning_rate": 3.3606557377049183e-06,
"loss": 5.9508,
"step": 58500
},
{
"epoch": 7.16,
"learning_rate": 3.132969034608379e-06,
"loss": 5.9505,
"step": 59000
},
{
"epoch": 7.23,
"learning_rate": 2.90528233151184e-06,
"loss": 5.9507,
"step": 59500
},
{
"epoch": 7.29,
"learning_rate": 2.6775956284153005e-06,
"loss": 5.9513,
"step": 60000
},
{
"epoch": 7.35,
"learning_rate": 2.4499089253187615e-06,
"loss": 5.9508,
"step": 60500
},
{
"epoch": 7.41,
"learning_rate": 2.222222222222222e-06,
"loss": 5.9508,
"step": 61000
},
{
"epoch": 7.47,
"learning_rate": 1.994535519125683e-06,
"loss": 5.9507,
"step": 61500
},
{
"epoch": 7.53,
"learning_rate": 1.766848816029144e-06,
"loss": 5.9506,
"step": 62000
},
{
"epoch": 7.59,
"learning_rate": 1.5391621129326049e-06,
"loss": 5.951,
"step": 62500
},
{
"epoch": 7.65,
"learning_rate": 1.3114754098360657e-06,
"loss": 5.9504,
"step": 63000
},
{
"epoch": 7.71,
"learning_rate": 1.0837887067395265e-06,
"loss": 5.9508,
"step": 63500
},
{
"epoch": 7.77,
"learning_rate": 8.561020036429873e-07,
"loss": 5.9508,
"step": 64000
},
{
"epoch": 7.83,
"learning_rate": 6.284153005464482e-07,
"loss": 5.9509,
"step": 64500
},
{
"epoch": 7.89,
"learning_rate": 4.0072859744990897e-07,
"loss": 5.9506,
"step": 65000
},
{
"epoch": 7.95,
"learning_rate": 1.7304189435336977e-07,
"loss": 5.9511,
"step": 65500
},
{
"epoch": 8.0,
"step": 65880,
"total_flos": 8.113569971306496e+17,
"train_runtime": 71825.285,
"train_samples_per_second": 14.675,
"train_steps_per_second": 0.917
}
],
"max_steps": 65880,
"num_train_epochs": 8,
"total_flos": 8.113569971306496e+17,
"trial_name": null,
"trial_params": null
}