{ "best_metric": 0.4068953869702192, "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-3/checkpoint-4276", "epoch": 2.0, "eval_steps": 500, "global_step": 4276, "is_hyper_param_search": true, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.23, "learning_rate": 1.8134386732635114e-06, "loss": 0.6097, "step": 500 }, { "epoch": 0.47, "learning_rate": 1.6601212338601031e-06, "loss": 0.5883, "step": 1000 }, { "epoch": 0.7, "learning_rate": 1.506803794456695e-06, "loss": 0.562, "step": 1500 }, { "epoch": 0.94, "learning_rate": 1.3534863550532868e-06, "loss": 0.5302, "step": 2000 }, { "epoch": 1.0, "eval_loss": 0.5693901181221008, "eval_matthews_correlation": 0.2787671229651657, "eval_runtime": 0.716, "eval_samples_per_second": 1456.694, "eval_steps_per_second": 92.178, "step": 2138 }, { "epoch": 1.17, "learning_rate": 1.2001689156498788e-06, "loss": 0.5141, "step": 2500 }, { "epoch": 1.4, "learning_rate": 1.0468514762464705e-06, "loss": 0.5067, "step": 3000 }, { "epoch": 1.64, "learning_rate": 8.935340368430625e-07, "loss": 0.4926, "step": 3500 }, { "epoch": 1.87, "learning_rate": 7.402165974396544e-07, "loss": 0.501, "step": 4000 }, { "epoch": 2.0, "eval_loss": 0.5920860171318054, "eval_matthews_correlation": 0.4068953869702192, "eval_runtime": 0.7393, "eval_samples_per_second": 1410.839, "eval_steps_per_second": 89.276, "step": 4276 } ], "logging_steps": 500, "max_steps": 6414, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 500, "total_flos": 65084182429116.0, "train_batch_size": 4, "trial_name": null, "trial_params": { "learning_rate": 1.9667561126669194e-06, "num_train_epochs": 3, "per_device_train_batch_size": 4, "seed": 1 } }