| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.702412868632708, | |
| "eval_steps": 500, | |
| "global_step": 2500, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7680187821388245, | |
| "eval_loss": 0.4622006416320801, | |
| "eval_runtime": 9.5056, | |
| "eval_samples_per_second": 313.816, | |
| "eval_steps_per_second": 19.673, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.3404825737265416, | |
| "grad_norm": 21.834245681762695, | |
| "learning_rate": 3.516708684536568e-06, | |
| "loss": 0.5575, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7968488335609436, | |
| "eval_loss": 0.4220982789993286, | |
| "eval_runtime": 9.4241, | |
| "eval_samples_per_second": 316.529, | |
| "eval_steps_per_second": 19.843, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.680965147453083, | |
| "grad_norm": 18.592424392700195, | |
| "learning_rate": 2.972326535227502e-06, | |
| "loss": 0.4016, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8162923455238342, | |
| "eval_loss": 0.4073546528816223, | |
| "eval_runtime": 9.4187, | |
| "eval_samples_per_second": 316.71, | |
| "eval_steps_per_second": 19.854, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8213208317756653, | |
| "eval_loss": 0.4337773323059082, | |
| "eval_runtime": 9.4694, | |
| "eval_samples_per_second": 315.016, | |
| "eval_steps_per_second": 19.748, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 4.021447721179625, | |
| "grad_norm": 26.7930850982666, | |
| "learning_rate": 2.4279443859184353e-06, | |
| "loss": 0.2956, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8246731758117676, | |
| "eval_loss": 0.4652492105960846, | |
| "eval_runtime": 9.4433, | |
| "eval_samples_per_second": 315.885, | |
| "eval_steps_per_second": 19.802, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 5.361930294906166, | |
| "grad_norm": 20.48050880432129, | |
| "learning_rate": 1.8835622366093693e-06, | |
| "loss": 0.2101, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8236674666404724, | |
| "eval_loss": 0.4927256107330322, | |
| "eval_runtime": 9.3134, | |
| "eval_samples_per_second": 320.292, | |
| "eval_steps_per_second": 20.079, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 6.702412868632708, | |
| "grad_norm": 23.84658432006836, | |
| "learning_rate": 1.3391800873003031e-06, | |
| "loss": 0.1735, | |
| "step": 2500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 3730, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.4022192353692968e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 4.0610908338456346e-06, | |
| "per_device_train_batch_size": 32 | |
| } | |
| } | |