{"train/loss": 0.0494, "train/grad_norm": 0.7622084617614746, "train/learning_rate": 2.2727272727272728e-06, "train/epoch": 30.0, "train/global_step": 90, "_timestamp": 1714485117.8533459, "_runtime": 507.42275285720825, "_step": 60, "eval/loss": 0.43639254570007324, "eval/runtime": 1.3775, "eval/samples_per_second": 13.067, "eval/steps_per_second": 2.178, "train_runtime": 527.4654, "train_samples_per_second": 4.038, "train_steps_per_second": 0.171, "total_flos": 5669487917678592.0, "train_loss": 0.4407807625002331}