{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.03537944454272068, "eval_steps": 10, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0007075888908544136, "eval_loss": 10.552796363830566, "eval_runtime": 98.5911, "eval_samples_per_second": 6.035, "eval_steps_per_second": 3.023, "step": 1 }, { "epoch": 0.0035379444542720678, "grad_norm": 25.988548278808594, "learning_rate": 5e-05, "loss": 10.6611, "step": 5 }, { "epoch": 0.0070758889085441356, "grad_norm": 26.148916244506836, "learning_rate": 0.0001, "loss": 7.6885, "step": 10 }, { "epoch": 0.0070758889085441356, "eval_loss": 4.3447370529174805, "eval_runtime": 99.5463, "eval_samples_per_second": 5.977, "eval_steps_per_second": 2.994, "step": 10 }, { "epoch": 0.010613833362816204, "grad_norm": 23.476253509521484, "learning_rate": 9.619397662556435e-05, "loss": 2.804, "step": 15 }, { "epoch": 0.014151777817088271, "grad_norm": 11.006500244140625, "learning_rate": 8.535533905932738e-05, "loss": 0.4976, "step": 20 }, { "epoch": 0.014151777817088271, "eval_loss": 0.407570481300354, "eval_runtime": 54.6848, "eval_samples_per_second": 10.881, "eval_steps_per_second": 5.449, "step": 20 }, { "epoch": 0.01768972227136034, "grad_norm": 5.728410243988037, "learning_rate": 6.91341716182545e-05, "loss": 0.3862, "step": 25 }, { "epoch": 0.021227666725632408, "grad_norm": 11.9749174118042, "learning_rate": 5e-05, "loss": 0.2761, "step": 30 }, { "epoch": 0.021227666725632408, "eval_loss": 0.21304932236671448, "eval_runtime": 48.3826, "eval_samples_per_second": 12.298, "eval_steps_per_second": 6.159, "step": 30 }, { "epoch": 0.024765611179904477, "grad_norm": 0.7041988372802734, "learning_rate": 3.086582838174551e-05, "loss": 0.2231, "step": 35 }, { "epoch": 0.028303555634176542, "grad_norm": 8.75727367401123, "learning_rate": 1.4644660940672627e-05, "loss": 0.2206, "step": 40 }, { "epoch": 0.028303555634176542, "eval_loss": 0.14831823110580444, "eval_runtime": 48.4176, "eval_samples_per_second": 12.289, "eval_steps_per_second": 6.155, "step": 40 }, { "epoch": 0.03184150008844861, "grad_norm": 6.61296272277832, "learning_rate": 3.8060233744356633e-06, "loss": 0.1966, "step": 45 }, { "epoch": 0.03537944454272068, "grad_norm": 10.20125675201416, "learning_rate": 0.0, "loss": 0.3427, "step": 50 }, { "epoch": 0.03537944454272068, "eval_loss": 0.14001674950122833, "eval_runtime": 48.1288, "eval_samples_per_second": 12.363, "eval_steps_per_second": 6.192, "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": 9555457081344000.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }