| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.047337278106508875, | |
| "eval_steps": 20, | |
| "global_step": 40, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.001183431952662722, | |
| "eval_loss": 2.9618937969207764, | |
| "eval_runtime": 5.153, | |
| "eval_samples_per_second": 291.48, | |
| "eval_steps_per_second": 18.242, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.011834319526627219, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.6542, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.023668639053254437, | |
| "grad_norm": 1.6328125, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.79, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023668639053254437, | |
| "eval_loss": 2.9338200092315674, | |
| "eval_runtime": 5.4254, | |
| "eval_samples_per_second": 276.844, | |
| "eval_steps_per_second": 17.326, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03550295857988166, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.8939, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.047337278106508875, | |
| "grad_norm": 2.1875, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 2.7567, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.047337278106508875, | |
| "eval_loss": 2.7183806896209717, | |
| "eval_runtime": 5.4234, | |
| "eval_samples_per_second": 276.95, | |
| "eval_steps_per_second": 17.332, | |
| "step": 40 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 40, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2489856831782912.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |