HeQ / trainer_state.json
amirdnc's picture
added model files
27c5f49
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 15.0,
"global_step": 13710,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.000000000000001e-07,
"loss": 5.9308,
"step": 1
},
{
"epoch": 0.55,
"learning_rate": 4.85304922850845e-05,
"loss": 2.2869,
"step": 500
},
{
"epoch": 1.09,
"learning_rate": 4.6693607641440116e-05,
"loss": 1.5406,
"step": 1000
},
{
"epoch": 1.64,
"learning_rate": 4.485672299779574e-05,
"loss": 1.1804,
"step": 1500
},
{
"epoch": 2.19,
"learning_rate": 4.301983835415136e-05,
"loss": 1.0226,
"step": 2000
},
{
"epoch": 2.74,
"learning_rate": 4.1182953710506984e-05,
"loss": 0.8362,
"step": 2500
},
{
"epoch": 3.28,
"learning_rate": 3.93460690668626e-05,
"loss": 0.6883,
"step": 3000
},
{
"epoch": 3.83,
"learning_rate": 3.750918442321822e-05,
"loss": 0.5913,
"step": 3500
},
{
"epoch": 4.38,
"learning_rate": 3.5672299779573845e-05,
"loss": 0.456,
"step": 4000
},
{
"epoch": 4.92,
"learning_rate": 3.383541513592947e-05,
"loss": 0.4352,
"step": 4500
},
{
"epoch": 5.47,
"learning_rate": 3.1998530492285084e-05,
"loss": 0.3212,
"step": 5000
},
{
"epoch": 6.02,
"learning_rate": 3.0161645848640706e-05,
"loss": 0.3159,
"step": 5500
},
{
"epoch": 6.56,
"learning_rate": 2.8324761204996326e-05,
"loss": 0.2223,
"step": 6000
},
{
"epoch": 7.11,
"learning_rate": 2.6487876561351948e-05,
"loss": 0.2304,
"step": 6500
},
{
"epoch": 7.66,
"learning_rate": 2.465099191770757e-05,
"loss": 0.1776,
"step": 7000
},
{
"epoch": 8.21,
"learning_rate": 2.281410727406319e-05,
"loss": 0.1594,
"step": 7500
},
{
"epoch": 8.75,
"learning_rate": 2.097722263041881e-05,
"loss": 0.1383,
"step": 8000
},
{
"epoch": 9.3,
"learning_rate": 1.9140337986774432e-05,
"loss": 0.1164,
"step": 8500
},
{
"epoch": 9.85,
"learning_rate": 1.730345334313005e-05,
"loss": 0.1007,
"step": 9000
},
{
"epoch": 10.39,
"learning_rate": 1.5466568699485674e-05,
"loss": 0.082,
"step": 9500
},
{
"epoch": 10.94,
"learning_rate": 1.3629684055841293e-05,
"loss": 0.0761,
"step": 10000
},
{
"epoch": 11.49,
"learning_rate": 1.1792799412196914e-05,
"loss": 0.0598,
"step": 10500
},
{
"epoch": 12.04,
"learning_rate": 9.955914768552535e-06,
"loss": 0.0599,
"step": 11000
},
{
"epoch": 12.58,
"learning_rate": 8.119030124908156e-06,
"loss": 0.0395,
"step": 11500
},
{
"epoch": 13.13,
"learning_rate": 6.282145481263777e-06,
"loss": 0.0379,
"step": 12000
},
{
"epoch": 13.68,
"learning_rate": 4.445260837619398e-06,
"loss": 0.0282,
"step": 12500
},
{
"epoch": 14.22,
"learning_rate": 2.6083761939750186e-06,
"loss": 0.0278,
"step": 13000
},
{
"epoch": 14.77,
"learning_rate": 7.714915503306393e-07,
"loss": 0.0203,
"step": 13500
},
{
"epoch": 15.0,
"step": 13710,
"total_flos": 8.593299100871424e+16,
"train_loss": 0.41084846325115426,
"train_runtime": 10637.12,
"train_samples_per_second": 41.223,
"train_steps_per_second": 1.289
}
],
"max_steps": 13710,
"num_train_epochs": 15,
"total_flos": 8.593299100871424e+16,
"trial_name": null,
"trial_params": null
}