{ "activation_checkpointing": true, "dataset": "fineweb-edu-100b-shuffle", "depth": 12, "eval_batch_size": 4, "mfu_percent": 13.080351672255688, "num_params": 185599128, "num_steps": 448, "peak_vram_mb": 4379.2568359375, "smoke_test": false, "total_seconds": 15153.68583703041, "total_tokens": 234881024, "train_batch_size": 4, "training_seconds": 14431.43845486641, "val_bpb": 1.0384125083478744 }