| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.0023556293652756675, | |
| "eval_steps": 10, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 4.711258730551335e-05, | |
| "eval_loss": 2.2091963291168213, | |
| "eval_runtime": 435.6052, | |
| "eval_samples_per_second": 20.519, | |
| "eval_steps_per_second": 10.259, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00023556293652756677, | |
| "grad_norm": 0.6054697632789612, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0606, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00047112587305513353, | |
| "grad_norm": 0.7563396096229553, | |
| "learning_rate": 0.0001, | |
| "loss": 2.0653, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00047112587305513353, | |
| "eval_loss": 2.0500152111053467, | |
| "eval_runtime": 426.051, | |
| "eval_samples_per_second": 20.979, | |
| "eval_steps_per_second": 10.489, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0007066888095827003, | |
| "grad_norm": 0.9156299829483032, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 1.997, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0009422517461102671, | |
| "grad_norm": 0.7770156264305115, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 1.7802, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0009422517461102671, | |
| "eval_loss": 1.5978509187698364, | |
| "eval_runtime": 429.9586, | |
| "eval_samples_per_second": 20.788, | |
| "eval_steps_per_second": 10.394, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0011778146826378337, | |
| "grad_norm": 0.7813920974731445, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 1.4529, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0014133776191654005, | |
| "grad_norm": 0.8424850702285767, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5786, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0014133776191654005, | |
| "eval_loss": 1.5485860109329224, | |
| "eval_runtime": 429.6892, | |
| "eval_samples_per_second": 20.801, | |
| "eval_steps_per_second": 10.401, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0016489405556929673, | |
| "grad_norm": 0.8277798891067505, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 1.5594, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0018845034922205341, | |
| "grad_norm": 0.7471029758453369, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 1.5405, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0018845034922205341, | |
| "eval_loss": 1.5304561853408813, | |
| "eval_runtime": 431.7149, | |
| "eval_samples_per_second": 20.703, | |
| "eval_steps_per_second": 10.352, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0021200664287481007, | |
| "grad_norm": 0.8744756579399109, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 1.4583, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0023556293652756675, | |
| "grad_norm": 0.8616583347320557, | |
| "learning_rate": 0.0, | |
| "loss": 1.4149, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0023556293652756675, | |
| "eval_loss": 1.5266047716140747, | |
| "eval_runtime": 431.9564, | |
| "eval_samples_per_second": 20.692, | |
| "eval_steps_per_second": 10.346, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "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": 5371318755655680.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |