| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.2, | |
| "eval_steps": 100, | |
| "global_step": 600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.450490098019604e-05, | |
| "loss": 2.1667, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 2.204326868057251, | |
| "eval_runtime": 1312.2394, | |
| "eval_samples_per_second": 1.524, | |
| "eval_steps_per_second": 0.191, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.400480096019204e-05, | |
| "loss": 2.2138, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 2.2141246795654297, | |
| "eval_runtime": 1312.8612, | |
| "eval_samples_per_second": 1.523, | |
| "eval_steps_per_second": 0.19, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.3504700940188037e-05, | |
| "loss": 2.1948, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 2.1924502849578857, | |
| "eval_runtime": 1312.5129, | |
| "eval_samples_per_second": 1.524, | |
| "eval_steps_per_second": 0.19, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.3004600920184037e-05, | |
| "loss": 2.18, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 2.1799795627593994, | |
| "eval_runtime": 1312.8237, | |
| "eval_samples_per_second": 1.523, | |
| "eval_steps_per_second": 0.19, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.2504500900180037e-05, | |
| "loss": 2.1521, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 2.1660068035125732, | |
| "eval_runtime": 1312.1202, | |
| "eval_samples_per_second": 1.524, | |
| "eval_steps_per_second": 0.191, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.2004400880176038e-05, | |
| "loss": 1.2274, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_loss": 2.3011252880096436, | |
| "eval_runtime": 1313.1108, | |
| "eval_samples_per_second": 1.523, | |
| "eval_steps_per_second": 0.19, | |
| "step": 600 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 100, | |
| "total_flos": 2.490210050899968e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |