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": 10.54305942086405,
 "learning_rate": 3.333333333333333e-07,
 "loss": 1.7384,
 "step": 1
 },
 {
 "epoch": 0.71,
 "grad_norm": 8.861964631394585,
 "learning_rate": 9.698463103929541e-07,
 "loss": 1.7673,
 "step": 5
 },
 {
 "epoch": 1.0,
 "eval_loss": 1.6740180253982544,
 "eval_runtime": 1.5323,
 "eval_samples_per_second": 8.484,
 "eval_steps_per_second": 0.653,
 "step": 7
 },
 {
 "epoch": 1.43,
 "grad_norm": 6.359069442755467,
 "learning_rate": 6.710100716628344e-07,
 "loss": 1.6571,
 "step": 10
 },
 {
 "epoch": 2.0,
 "eval_loss": 1.6360118389129639,
 "eval_runtime": 1.3369,
 "eval_samples_per_second": 9.724,
 "eval_steps_per_second": 0.748,
 "step": 14
 },
 {
 "epoch": 2.14,
 "grad_norm": 4.890739014531452,
 "learning_rate": 2.500000000000001e-07,
 "loss": 1.6028,
 "step": 15
 },
 {
 "epoch": 2.86,
 "grad_norm": 4.729517911624113,
 "learning_rate": 7.59612349389599e-09,
 "loss": 1.5669,
 "step": 20
 },
 {
 "epoch": 3.0,
 "eval_loss": 1.627854585647583,
 "eval_runtime": 1.3347,
 "eval_samples_per_second": 9.74,
 "eval_steps_per_second": 0.749,
 "step": 21
 },
 {
 "epoch": 3.0,
 "step": 21,
 "total_flos": 4344627855360.0,
 "train_loss": 1.6369684423719133,
 "train_runtime": 362.7919,
 "train_samples_per_second": 3.614,
 "train_steps_per_second": 0.058
 }
 ],
 "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
}