Function read_tensor_bytes

server/scripts/quantize_draft_q8.py:99–103

(path: Path, header_size: int, info: dict)

    return None


def is_norm_tensor(gguf_name: str) -> bool:
    return (
        gguf_name.endswith("_norm.weight") or
        gguf_name == "output_norm.weight" or
        gguf_name == "dflash.hidden_norm.weight"
    )