| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.009419743782969104, | |
| "eval_steps": 13, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00018839487565938207, | |
| "eval_loss": 3.400683879852295, | |
| "eval_runtime": 37.7777, | |
| "eval_samples_per_second": 59.162, | |
| "eval_steps_per_second": 29.594, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0018839487565938207, | |
| "grad_norm": 2.348120927810669, | |
| "learning_rate": 0.0002, | |
| "loss": 13.693, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0024491333835719668, | |
| "eval_loss": 3.316739797592163, | |
| "eval_runtime": 37.7223, | |
| "eval_samples_per_second": 59.249, | |
| "eval_steps_per_second": 29.638, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0037678975131876413, | |
| "grad_norm": 2.279417037963867, | |
| "learning_rate": 0.0002, | |
| "loss": 13.0534, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0048982667671439335, | |
| "eval_loss": 3.276735544204712, | |
| "eval_runtime": 37.9543, | |
| "eval_samples_per_second": 58.887, | |
| "eval_steps_per_second": 29.456, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.005651846269781462, | |
| "grad_norm": 2.9704277515411377, | |
| "learning_rate": 0.0002, | |
| "loss": 13.358, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0073474001507159, | |
| "eval_loss": 3.256060838699341, | |
| "eval_runtime": 38.2237, | |
| "eval_samples_per_second": 58.472, | |
| "eval_steps_per_second": 29.249, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.007535795026375283, | |
| "grad_norm": 2.37857985496521, | |
| "learning_rate": 0.0002, | |
| "loss": 13.1662, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.009419743782969104, | |
| "grad_norm": 4.561653137207031, | |
| "learning_rate": 0.0002, | |
| "loss": 13.3278, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 50, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 13, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 662489687654400.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |