{ "best_metric": 0.9764395952224731, "best_model_checkpoint": "th_cl_18epochs_lora_pos_neg/checkpoint-96", "epoch": 5.0, "eval_steps": 500, "global_step": 160, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_accuracy": 0.4888888888888889, "eval_balanced_accuracy": 0.4939271255060729, "eval_loss": 1.0155580043792725, "eval_runtime": 86.9278, "eval_samples_per_second": 0.518, "eval_steps_per_second": 0.069, "step": 32 }, { "epoch": 2.0, "eval_accuracy": 0.4888888888888889, "eval_balanced_accuracy": 0.4939271255060729, "eval_loss": 0.9877275228500366, "eval_runtime": 86.6979, "eval_samples_per_second": 0.519, "eval_steps_per_second": 0.069, "step": 64 }, { "epoch": 3.0, "eval_accuracy": 0.5555555555555556, "eval_balanced_accuracy": 0.5504201680672269, "eval_loss": 0.9764395952224731, "eval_runtime": 86.783, "eval_samples_per_second": 0.519, "eval_steps_per_second": 0.069, "step": 96 }, { "epoch": 4.0, "eval_accuracy": 0.5777777777777777, "eval_balanced_accuracy": 0.5740740740740741, "eval_loss": 1.0566394329071045, "eval_runtime": 87.1596, "eval_samples_per_second": 0.516, "eval_steps_per_second": 0.069, "step": 128 }, { "epoch": 5.0, "eval_accuracy": 0.5111111111111111, "eval_balanced_accuracy": 0.515, "eval_loss": 1.0600730180740356, "eval_runtime": 86.6662, "eval_samples_per_second": 0.519, "eval_steps_per_second": 0.069, "step": 160 }, { "epoch": 5.0, "step": 160, "total_flos": 2.696144222158848e+16, "train_loss": 0.24327199459075927, "train_runtime": 8037.2155, "train_samples_per_second": 0.156, "train_steps_per_second": 0.02 } ], "logging_steps": 500, "max_steps": 160, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "total_flos": 2.696144222158848e+16, "train_batch_size": 8, "trial_name": null, "trial_params": null }