{ "activation_checkpointing": true, "dataset": "tinystories", "depth": 10, "eval_batch_size": 4, "mfu_percent": 11.909878504506382, "num_params": 122553140, "num_steps": 325, "peak_vram_mb": 3616.9375, "smoke_test": false, "total_seconds": 7733.325039625168, "total_tokens": 170393600, "train_batch_size": 4, "training_seconds": 7216.76309800148, "val_bpb": 0.4032088144179764 }