opendatalab/MinerU · error · ValueError

Unsupported hybrid effort

Error message

Unsupported hybrid effort: {effort}

What it means

A defensive guard in the synchronous hybrid analysis loop (hybrid_analyze.py:1035) reached when effort matches neither the medium nor the high branch of the effort dispatch chain. In principle _validate_parse_effort() should have rejected such values earlier, so hitting this line means validation was bypassed or the dispatch branches got out of sync with the allowed set.

Solutions

1. Use the public entry point that runs _validate_parse_effort() and pass effort="medium" or "high".
2. Ensure only one mineru version is importable (pip show mineru; check for stale copies in site-packages or cwd).
3. If you maintain a fork adding a new effort level, add its dispatch branch alongside the raise.

Example fix

# before
model_list = hybrid_page_loop(windows, effort="ultra")  # internal call, no validation

# after
from mineru.backend.hybrid.hybrid_analyze import _validate_parse_effort
effort = _validate_parse_effort("high")
model_list = hybrid_page_loop(windows, effort=effort)

from mineru.backend.hybrid.hybrid_analyze import _validate_parse_effort

# call the public API; if you must call internals, validate first:
effort = _validate_parse_effort(request.args.get("effort", "medium"))

ALLOWED = {"medium", "high"}

def is_valid_effort(v: object) -> bool:
    return isinstance(v, str) and v in ALLOWED

try:
    model_list = hybrid_page_loop(windows, effort=effort)
except ValueError as e:
    if "Unsupported hybrid effort" in str(e):
        raise ConfigError("hybrid effort out of range; use medium/high") from e
    raise

When it happens

Trigger: effort passing early validation but not matching any if/elif branch in the page-window loop; typically only possible with inconsistent code paths, monkeypatched validators, or an internal call that skips validation.

Common situations: Calling the internal analysis loop directly instead of the public API; patched/older copies of the module mixed at runtime; a new effort value added to HYBRID_ANALYZE_EFFORTS without a matching branch here.

Related errors

• effort must be "medium" or "high"
• Invalid effort. Allowed values
• Invalid effort. Allowed values
• ` ` requires local pipeline dependencies…

AI-assisted analysis of opendatalab/MinerU@4fe4bde114 (2026-08-14). Data as JSON: /api/errors/02bf71be6b0c5b3e. Report an issue: GitHub.

Appendix: source

Thrown at mineru/backend/hybrid/hybrid_analyze.py:1035

                                hybrid_pipeline_model=hybrid_pipeline_model,
                            )
                        else:
                            with predictor_execution_guard(predictor):
                                window_model_list = predictor.batch_two_step_extract(
                                    images=images_pil_list,
                                    not_extract_list=not_extract_list,
                                    image_analysis=effective_image_analysis,
                                )
                            window_model_list = _process_ocr_and_formulas(
                                images_pil_list,
                                window_model_list,
                                inline_formula_enable,
                                batch_ratio=batch_ratio,
                                images_layout_res=images_layout_res,
                                hybrid_pipeline_model=hybrid_pipeline_model,
                            )
                    else:
                        raise ValueError(f"Unsupported hybrid effort: {effort}")

                    _apply_layout_title_split(
                        window_model_list,
                        images_layout_res,
                        page_sizes,
                    )
                    model_list.extend(window_model_list)
                    if progress_bar is None:
                        progress_bar = tqdm(total=page_count, desc="Processing pages")
                    else:
                        exclude_progress_bar_idle_time(
                            progress_bar,
                            last_append_end_time,
                            now=time.time(),
                        )
                    append_page_model_list_to_middle_json(
                        middle_json,
                        window_model_list,

View on GitHub (pinned to 4fe4bde114)