sgl-project/sglang · error · ValueError

prompt must be non-empty

Error message

prompt must be non-empty

What it means

minimax_h3_text_only_ids encodes a plain prompt for the t2va (text-to-video-audio) path and requires it non-empty; an empty string would produce an empty token tensor with no text conditioning.

Source

Thrown at python/sglang/multimodal_gen/runtime/pipelines_core/stages/model_specific_stages/minimax_h3/presentation.py:112

    counts = [int(value) for value in counts]
    timestamps = [float(value) for value in timestamps]
    if not counts or len(counts) != len(timestamps):
        raise ValueError(f"{context}video block token counts and timestamps must align")
    for count, timestamp in zip(counts, timestamps):
        if count <= 0:
            raise ValueError(f"{context}video block token count must be positive")
        presentation.text(_text_ids(tokenizer, f"<{timestamp:.1f} seconds>"))
        presentation.vision(
            _vision_block_ids(tokenizer, VIDEO_PAD, count),
            video_token_id=video_token_id,
        )


def minimax_h3_text_only_ids(tokenizer: Any, prompt: str) -> torch.Tensor:
    """t2va presentation: verbatim prompt, no special tokens."""
    if not prompt:
        raise ValueError("prompt must be non-empty")
    return torch.tensor(_text_ids(tokenizer, prompt), dtype=torch.long)


def minimax_h3_multi_image_presentation(
    tokenizer: Any,
    *,
    prompt: str,
    image_token_counts: list[int],
) -> tuple[torch.Tensor, torch.Tensor]:
    if not image_token_counts:
        raise ValueError("image_token_counts must be non-empty")
    presentation = _Presentation()
    for index, count in enumerate(image_token_counts, start=1):
        if int(count) <= 0:
            raise ValueError("image_token_count must be positive")
        presentation.text(_text_ids(tokenizer, f"<Picture {index}>: "))
        presentation.vision(_vision_block_ids(tokenizer, IMAGE_PAD, count))
    presentation.text(_text_ids(tokenizer, prompt))

View on GitHub (pinned to 0132848349)

Solutions

1. Validate/skip empty prompts upstream
2. Provide a fallback prompt string (e.g. a default caption) before encoding
3. Strip and check prompt truthiness before calling

Example fix

// before
ids = minimax_h3_text_only_ids(tokenizer, prompt)
// after
if not prompt:
    raise ValueError("caption required")
ids = minimax_h3_text_only_ids(tokenizer, prompt)

if not prompt or not prompt.strip():
    raise ValueError("prompt required for text-only encoding")

def has_prompt(p: str | None) -> bool:
    return bool(p and p.strip())

When it happens

Trigger: Calling minimax_h3_text_only_ids(tokenizer, "") or with a whitespace-only/None-coerced prompt.

Common situations: Prompt templates where a user variable is empty; optional prompt fields defaulting to ""; data pipelines forwarding missing captions.

Related errors

• MiniMax-H3 adaln_t_table must have shape [N, D] with N >= 2,
• MiniMax H3 AdaLN cache has invalid timestep plans
• TP size must be positive.
• num_attention_heads must be positive.
• hidden_size must be positive.

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/90ed2728db30edf8. Report an issue: GitHub.