| { | |
| "best_metric": 2.1534, | |
| "best_model_checkpoint": "checkpoint-50000", | |
| "epoch": 3.0, | |
| "global_step": 50000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0003, | |
| "loss": 4.2156, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00029, | |
| "loss": 3.8743, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1245, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 3.0987, | |
| "eval_runtime": 245.67, | |
| "eval_samples_per_second": 163.42, | |
| "eval_steps_per_second": 20.43, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8432, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00015, | |
| "loss": 2.5678, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 2.5234, | |
| "eval_runtime": 243.21, | |
| "eval_samples_per_second": 165.08, | |
| "eval_steps_per_second": 20.64, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.0001, | |
| "loss": 2.3456, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1789, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 2.1534, | |
| "eval_runtime": 241.89, | |
| "eval_samples_per_second": 165.98, | |
| "eval_steps_per_second": 20.75, | |
| "step": 15000 | |
| } | |
| ], | |
| "max_steps": 50000, | |
| "num_input_tokens_seen": 409600000, | |
| "num_train_epochs": 3.0, | |
| "total_flos": 8.796093022208e+18, | |
| "train_batch_size": 128, | |
| "trial_name": null, | |
| "trial_params": null | |
| } |