sgl-project/sglang · error · ValueError

unsupported time format: {time_format!r}

Error message

unsupported time format: {time_format!r}

What it means

Raised by the constructor of the DotsNote Omni video QA flattener when the time_format argument is not one of the supported values 'hms', 'seconds', or 'random'. This format controls how timestamps are rendered inside flattened QA text (HH:MM:SS vs decimal seconds vs randomly chosen per example). Any other string (including typos or None passed positionally) fails fast before any processing.

Source

Thrown at python/sglang/srt/multimodal/processors/dots_note_omni_video_core/video_qa_flattener.py:52

        if match:
            pairs.append((int(match.group(1)), key))
    return [key for _, key in sorted(pairs)]


class VideoQAFlattener:
    """Apply the frame and audio sampling policy used during training."""

    def __init__(
        self,
        time_format: str = "random",
        seconds_decimals: int = 1,
        audio_interleave: bool = False,
        ai_seg_min_sec: float = 1.0,
        ai_k_mode: str = "eval30",
        rng: random.Random | None = None,
    ):
        if time_format not in ("hms", "seconds", "random"):
            raise ValueError(f"unsupported time format: {time_format!r}")
        if ai_k_mode not in ("logk", "eval30", "eval_ek", "whole"):
            raise ValueError(f"unsupported audio interleave mode: {ai_k_mode!r}")

        self.time_format = time_format
        self.seconds_decimals = max(0, int(seconds_decimals))
        self.audio_interleave = bool(audio_interleave)
        self.ai_seg_min_sec = max(1e-6, float(ai_seg_min_sec))
        self.ai_k_mode = ai_k_mode
        self.rng = rng or random.Random()

    def _subsample_one_video(self, video_dict: dict):
        """Return ordered frames, timestamps, and optional WAV data."""
        original_fps = float(video_dict.get("fps", 1.0)) or 1.0
        image_keys = _sorted_image_keys(video_dict)
        frames = [video_dict[key] for key in image_keys]
        timestamps = [round(i / original_fps, 3) for i in range(len(image_keys))]
        return frames, timestamps, video_dict.get("audio_0") or None

View on GitHub (pinned to 0132848349)

Solutions

  1. Set time_format to one of 'hms', 'seconds', or 'random'
  2. If loading from config, validate/normalize the value before constructing the flattener
  3. Add the new format to the tuple check if you genuinely need a custom renderer

Example fix

# before
fl = VideoQAFlattener(time_format='sec')
# after
fl = VideoQAFlattener(time_format='seconds')
Defensive patterns

Strategy: validation

Validate before calling

VALID_TF = {'hms','seconds','random'}
if cfg['time_format'] not in VALID_TF:
    raise ConfigError(f"time_format must be one of {VALID_TF}, got {cfg['time_format']!r}")
fl = VideoQAFlattener(time_format=cfg['time_format'])

Type guard

def is_valid_time_format(v: str) -> bool:
    return v in ('hms','seconds','random')

Prevention

When it happens

Trigger: Constructing VideoQAFlattener (or a wrapper config) with time_format='sec', 's', 'HHMMSS', or None; passing configuration loaded from YAML/JSON where the key is misspelled or missing and an invalid default leaks through.

Common situations: Copy-pasting config between projects with different format names; CLI argument plumbing where the flag value isn't validated upstream; upgrading when the accepted enum set changed.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/58238f0e33d871ab. Report an issue: GitHub.