sgl-project/sglang · error · ValueError

Either chunk_index or chunk_size must be provided.

Error message

Either chunk_index or chunk_size must be provided.

What it means

_sana_wm_normalize_chunk_index needs either an explicit chunk_index list or a chunk_size to derive boundaries. Supplying neither makes chunk boundary computation impossible.

Source

Thrown at python/sglang/multimodal_gen/runtime/models/dits/sana_wm_components.py:407

    )


def _sana_wm_normalize_chunk_index(
    chunk_index: Optional[List[int]],
    T: int,
    chunk_size: Optional[int] = None,
    chunk_split_strategy: str = "uniform",
) -> list[int]:
    if chunk_index is not None:
        normalized = [int(idx) for idx in chunk_index]
        if not normalized or normalized[0] != 0:
            normalized = [0] + [idx for idx in normalized if idx > 0]
        normalized = [idx for idx in normalized if idx < T]
        if not normalized:
            normalized = [0]
    else:
        if chunk_size is None:
            raise ValueError("Either chunk_index or chunk_size must be provided.")
        normalized = _sana_wm_chunk_index_from_chunk_size(
            T,
            int(chunk_size),
            strategy=chunk_split_strategy,
        )

    if normalized[-1] != T:
        normalized.append(T)
    if any(end <= start for start, end in zip(normalized[:-1], normalized[1:])):
        raise ValueError(f"chunk_index must be strictly increasing, got {normalized}.")
    return normalized


def _sana_wm_chunk_boundaries_for_attention(
    HW: Tuple[int, int, int],
    chunk_size: Optional[int],
    chunk_split_strategy: str,
    chunk_index: Optional[List[int]],

View on GitHub (pinned to 0132848349)

Solutions

1. Set either chunk_index (list of boundary frame indices) or chunk_size (positive int) in the block config / call
2. If you intended unchunked processing, pass chunk_size equal to the full T
3. Default chunk_size in your config loader, e.g. chunk_size if chunk_size is not None else 32

Example fix

# before
bounds = _sana_wm_normalize_chunk_index(T, chunk_index=None, chunk_size=None)
# after
bounds = _sana_wm_normalize_chunk_index(T, chunk_index=None, chunk_size=32)

assert (chunk_index is None) != (chunk_size is None), 'provide exactly one of chunk_index/chunk_size'

When it happens

Trigger: Calling the attention/GDN chunk-scan path (via _sana_wm_chunk_boundaries_for_attention or _gdn_chunk_scan_forward) with both chunk_index=None and chunk_size=None.

Common situations: New model config omitting chunk_gdn_chunk_size; a wrapper that conditionally forwards only one of the two kwargs and the condition picked neither; None defaults propagating from an unset config object.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• SANA-WM forward requires encoder_hidden_states.
• SANA-WM forward requires timestep.
• SANA-WM forward_long requires encoder_hidden_states.
• SANA-WM forward_long requires timestep.
• chunk_size must be > 0, got {chunk_size}.

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/e09984c02c0b32b4. Report an issue: GitHub.