Yelp_Chinese / trainer_state.json
pgfeldman's picture
Upload 12 files
98b56fe
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.0,
"global_step": 34848,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09,
"learning_rate": 4.92825987144169e-05,
"loss": 3.4212,
"step": 500
},
{
"epoch": 0.17,
"learning_rate": 4.85651974288338e-05,
"loss": 3.323,
"step": 1000
},
{
"epoch": 0.26,
"learning_rate": 4.784779614325069e-05,
"loss": 3.2791,
"step": 1500
},
{
"epoch": 0.34,
"learning_rate": 4.7130394857667584e-05,
"loss": 3.2457,
"step": 2000
},
{
"epoch": 0.43,
"learning_rate": 4.6412993572084486e-05,
"loss": 3.2237,
"step": 2500
},
{
"epoch": 0.52,
"learning_rate": 4.569559228650138e-05,
"loss": 3.1983,
"step": 3000
},
{
"epoch": 0.6,
"learning_rate": 4.4978191000918276e-05,
"loss": 3.195,
"step": 3500
},
{
"epoch": 0.69,
"learning_rate": 4.426078971533517e-05,
"loss": 3.1841,
"step": 4000
},
{
"epoch": 0.77,
"learning_rate": 4.354338842975207e-05,
"loss": 3.1802,
"step": 4500
},
{
"epoch": 0.86,
"learning_rate": 4.282598714416897e-05,
"loss": 3.1674,
"step": 5000
},
{
"epoch": 0.95,
"learning_rate": 4.210858585858586e-05,
"loss": 3.1341,
"step": 5500
},
{
"epoch": 1.03,
"learning_rate": 4.139118457300275e-05,
"loss": 3.101,
"step": 6000
},
{
"epoch": 1.12,
"learning_rate": 4.0673783287419655e-05,
"loss": 3.0298,
"step": 6500
},
{
"epoch": 1.21,
"learning_rate": 3.995638200183655e-05,
"loss": 3.0358,
"step": 7000
},
{
"epoch": 1.29,
"learning_rate": 3.9238980716253446e-05,
"loss": 3.0531,
"step": 7500
},
{
"epoch": 1.38,
"learning_rate": 3.852157943067034e-05,
"loss": 3.0409,
"step": 8000
},
{
"epoch": 1.46,
"learning_rate": 3.7804178145087236e-05,
"loss": 3.0291,
"step": 8500
},
{
"epoch": 1.55,
"learning_rate": 3.708677685950414e-05,
"loss": 3.0346,
"step": 9000
},
{
"epoch": 1.64,
"learning_rate": 3.6369375573921034e-05,
"loss": 3.0356,
"step": 9500
},
{
"epoch": 1.72,
"learning_rate": 3.565197428833792e-05,
"loss": 3.0163,
"step": 10000
},
{
"epoch": 1.81,
"learning_rate": 3.4934573002754824e-05,
"loss": 3.026,
"step": 10500
},
{
"epoch": 1.89,
"learning_rate": 3.421717171717172e-05,
"loss": 3.0223,
"step": 11000
},
{
"epoch": 1.98,
"learning_rate": 3.3499770431588615e-05,
"loss": 3.0286,
"step": 11500
},
{
"epoch": 2.07,
"learning_rate": 3.278236914600551e-05,
"loss": 2.9479,
"step": 12000
},
{
"epoch": 2.15,
"learning_rate": 3.2064967860422406e-05,
"loss": 2.9327,
"step": 12500
},
{
"epoch": 2.24,
"learning_rate": 3.134756657483931e-05,
"loss": 2.943,
"step": 13000
},
{
"epoch": 2.32,
"learning_rate": 3.06301652892562e-05,
"loss": 2.9277,
"step": 13500
},
{
"epoch": 2.41,
"learning_rate": 2.9912764003673095e-05,
"loss": 2.9443,
"step": 14000
},
{
"epoch": 2.5,
"learning_rate": 2.919536271808999e-05,
"loss": 2.9452,
"step": 14500
},
{
"epoch": 2.58,
"learning_rate": 2.847796143250689e-05,
"loss": 2.9482,
"step": 15000
},
{
"epoch": 2.67,
"learning_rate": 2.7760560146923788e-05,
"loss": 2.9429,
"step": 15500
},
{
"epoch": 2.75,
"learning_rate": 2.704315886134068e-05,
"loss": 2.9463,
"step": 16000
},
{
"epoch": 2.84,
"learning_rate": 2.6325757575757575e-05,
"loss": 2.9292,
"step": 16500
},
{
"epoch": 2.93,
"learning_rate": 2.5608356290174473e-05,
"loss": 2.9286,
"step": 17000
},
{
"epoch": 3.01,
"learning_rate": 2.489095500459137e-05,
"loss": 2.9312,
"step": 17500
},
{
"epoch": 3.1,
"learning_rate": 2.4173553719008264e-05,
"loss": 2.8712,
"step": 18000
},
{
"epoch": 3.19,
"learning_rate": 2.3456152433425163e-05,
"loss": 2.8583,
"step": 18500
},
{
"epoch": 3.27,
"learning_rate": 2.2738751147842058e-05,
"loss": 2.882,
"step": 19000
},
{
"epoch": 3.36,
"learning_rate": 2.2021349862258957e-05,
"loss": 2.857,
"step": 19500
},
{
"epoch": 3.44,
"learning_rate": 2.130394857667585e-05,
"loss": 2.8815,
"step": 20000
},
{
"epoch": 3.53,
"learning_rate": 2.0586547291092747e-05,
"loss": 2.8775,
"step": 20500
},
{
"epoch": 3.62,
"learning_rate": 1.9869146005509643e-05,
"loss": 2.8606,
"step": 21000
},
{
"epoch": 3.7,
"learning_rate": 1.915174471992654e-05,
"loss": 2.8845,
"step": 21500
},
{
"epoch": 3.79,
"learning_rate": 1.8434343434343433e-05,
"loss": 2.8692,
"step": 22000
},
{
"epoch": 3.87,
"learning_rate": 1.7716942148760332e-05,
"loss": 2.8829,
"step": 22500
},
{
"epoch": 3.96,
"learning_rate": 1.6999540863177227e-05,
"loss": 2.8782,
"step": 23000
},
{
"epoch": 4.05,
"learning_rate": 1.6282139577594123e-05,
"loss": 2.8401,
"step": 23500
},
{
"epoch": 4.13,
"learning_rate": 1.5564738292011018e-05,
"loss": 2.822,
"step": 24000
},
{
"epoch": 4.22,
"learning_rate": 1.4847337006427917e-05,
"loss": 2.8197,
"step": 24500
},
{
"epoch": 4.3,
"learning_rate": 1.4129935720844812e-05,
"loss": 2.8227,
"step": 25000
},
{
"epoch": 4.39,
"learning_rate": 1.3412534435261709e-05,
"loss": 2.8154,
"step": 25500
},
{
"epoch": 4.48,
"learning_rate": 1.2695133149678604e-05,
"loss": 2.8247,
"step": 26000
},
{
"epoch": 4.56,
"learning_rate": 1.1977731864095501e-05,
"loss": 2.8196,
"step": 26500
},
{
"epoch": 4.65,
"learning_rate": 1.1260330578512397e-05,
"loss": 2.8357,
"step": 27000
},
{
"epoch": 4.73,
"learning_rate": 1.0542929292929294e-05,
"loss": 2.8163,
"step": 27500
},
{
"epoch": 4.82,
"learning_rate": 9.825528007346189e-06,
"loss": 2.8381,
"step": 28000
},
{
"epoch": 4.91,
"learning_rate": 9.108126721763086e-06,
"loss": 2.8198,
"step": 28500
},
{
"epoch": 4.99,
"learning_rate": 8.390725436179981e-06,
"loss": 2.8302,
"step": 29000
},
{
"epoch": 5.08,
"learning_rate": 7.673324150596878e-06,
"loss": 2.7897,
"step": 29500
},
{
"epoch": 5.17,
"learning_rate": 6.955922865013774e-06,
"loss": 2.7858,
"step": 30000
},
{
"epoch": 5.25,
"learning_rate": 6.238521579430671e-06,
"loss": 2.7865,
"step": 30500
},
{
"epoch": 5.34,
"learning_rate": 5.521120293847567e-06,
"loss": 2.7889,
"step": 31000
},
{
"epoch": 5.42,
"learning_rate": 4.803719008264463e-06,
"loss": 2.7797,
"step": 31500
},
{
"epoch": 5.51,
"learning_rate": 4.086317722681359e-06,
"loss": 2.7869,
"step": 32000
},
{
"epoch": 5.6,
"learning_rate": 3.3689164370982553e-06,
"loss": 2.7922,
"step": 32500
},
{
"epoch": 5.68,
"learning_rate": 2.651515151515152e-06,
"loss": 2.8032,
"step": 33000
},
{
"epoch": 5.77,
"learning_rate": 1.934113865932048e-06,
"loss": 2.7963,
"step": 33500
},
{
"epoch": 5.85,
"learning_rate": 1.216712580348944e-06,
"loss": 2.7952,
"step": 34000
},
{
"epoch": 5.94,
"learning_rate": 4.993112947658403e-07,
"loss": 2.7968,
"step": 34500
},
{
"epoch": 6.0,
"step": 34848,
"total_flos": 1.8211009462272e+16,
"train_loss": 2.9473999731030567,
"train_runtime": 8951.9422,
"train_samples_per_second": 3.893,
"train_steps_per_second": 3.893
}
],
"max_steps": 34848,
"num_train_epochs": 6,
"total_flos": 1.8211009462272e+16,
"trial_name": null,
"trial_params": null
}