{ "best_metric": 0.32348373532295227, "best_model_checkpoint": "fine_tuned_turkish_bert10k/checkpoint-1112", "epoch": 3.0, "eval_steps": 500, "global_step": 1668, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.8992805755395683, "grad_norm": 6.453958034515381, "learning_rate": 1.4004796163069546e-05, "loss": 0.2199, "step": 500 }, { "epoch": 1.0, "eval_accuracy": 0.9127697841726619, "eval_loss": 0.3523901700973511, "eval_runtime": 11.2517, "eval_samples_per_second": 197.66, "eval_steps_per_second": 12.354, "step": 556 }, { "epoch": 1.7985611510791366, "grad_norm": 24.75876235961914, "learning_rate": 8.00959232613909e-06, "loss": 0.1617, "step": 1000 }, { "epoch": 2.0, "eval_accuracy": 0.9154676258992805, "eval_loss": 0.32348373532295227, "eval_runtime": 11.2709, "eval_samples_per_second": 197.323, "eval_steps_per_second": 12.333, "step": 1112 }, { "epoch": 2.697841726618705, "grad_norm": 0.4041008949279785, "learning_rate": 2.0143884892086333e-06, "loss": 0.1104, "step": 1500 }, { "epoch": 3.0, "eval_accuracy": 0.9172661870503597, "eval_loss": 0.3766934275627136, "eval_runtime": 11.4325, "eval_samples_per_second": 194.533, "eval_steps_per_second": 12.158, "step": 1668 } ], "logging_steps": 500, "max_steps": 1668, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1755082294778880.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }