trainer_state.json

{
 "best_metric": null,
 "best_model_checkpoint": null,
 "epoch": 3.0,
 "eval_steps": 500,
 "global_step": 21,
 "is_hyper_param_search": false,
 "is_local_process_zero": true,
 "is_world_process_zero": true,
 "log_history": [
 {
 "epoch": 0.14,
 "grad_norm": 10.339013401925325,
 "learning_rate": 6.666666666666667e-06,
 "loss": 1.7182,
 "step": 1
 },
 {
 "epoch": 0.71,
 "grad_norm": 14.258117262298995,
 "learning_rate": 1.9396926207859085e-05,
 "loss": 1.7427,
 "step": 5
 },
 {
 "epoch": 1.0,
 "eval_loss": 1.8922593593597412,
 "eval_runtime": 1.987,
 "eval_samples_per_second": 4.529,
 "eval_steps_per_second": 0.503,
 "step": 7
 },
 {
 "epoch": 1.43,
 "grad_norm": 6.219405397937822,
 "learning_rate": 1.342020143325669e-05,
 "loss": 1.4527,
 "step": 10
 },
 {
 "epoch": 2.0,
 "eval_loss": 1.9308981895446777,
 "eval_runtime": 1.8867,
 "eval_samples_per_second": 4.77,
 "eval_steps_per_second": 0.53,
 "step": 14
 },
 {
 "epoch": 2.14,
 "grad_norm": 7.842627479310399,
 "learning_rate": 5.000000000000003e-06,
 "loss": 1.0554,
 "step": 15
 },
 {
 "epoch": 2.86,
 "grad_norm": 5.618821456903757,
 "learning_rate": 1.519224698779198e-07,
 "loss": 0.7124,
 "step": 20
 },
 {
 "epoch": 3.0,
 "eval_loss": 2.032864570617676,
 "eval_runtime": 1.8885,
 "eval_samples_per_second": 4.766,
 "eval_steps_per_second": 0.53,
 "step": 21
 },
 {
 "epoch": 3.0,
 "step": 21,
 "total_flos": 4344627855360.0,
 "train_loss": 1.2139497654778617,
 "train_runtime": 451.5351,
 "train_samples_per_second": 2.937,
 "train_steps_per_second": 0.047
 }
 ],
 "logging_steps": 5,
 "max_steps": 21,
 "num_input_tokens_seen": 0,
 "num_train_epochs": 3,
 "save_steps": 500,
 "total_flos": 4344627855360.0,
 "train_batch_size": 4,
 "trial_name": null,
 "trial_params": null
}