{ "activation_checkpointing": true, "dataset": "tinystories", "depth": 22, "eval_batch_size": 4, "mfu_percent": 17.708544461070193, "num_params": 823267148, "num_steps": 68, "peak_vram_mb": 12435.2333984375, "smoke_test": false, "total_seconds": 10007.529666423798, "total_tokens": 35651584, "train_batch_size": 4, "training_seconds": 7309.851931095123, "val_bpb": 0.7780523809334358 }