[bugfix] Fix unexpected argument 'is_finished' in function llm2code2wav_async_chunk of mimo-audio

[qibaoyuan]

bug: #1569

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.

Purpose

Fix unexpected argument 'is_finished' in function llm2code2wav_async_chunk of mimo-audio:
(EngineCore_DP0 pid=51408) [Stage-0] ERROR 02-28 16:59:52 [chunk_transfer_adapter.py:213] Failed to use custom_process_input_func for payload extraction: llm2code2wav_async_chunk() got an unexpected keyword argument 'is_finished'

Test Plan

serving:

vllm-omni serve XiaomiMiMo/MiMo-Audio-7B-Instruct --omni \
--chat-template ./examples/online_serving/mimo_audio/chat_template.jinja  \
--stage-configs-path model_executor/stage_configs/mimo_audio_async_chunk.yaml

client:

python examples/online_serving/mimo_audio/openai_chat_completion_client_for_multimodal_generation.py --stream

Test Result

[  32.16s][req 0_chatcmpl-bc7df29b6f1f58d1] 需要
[  32.42s][req 0_chatcmpl-bc7df29b6f1f58d1] 更
[  32.44s][req 0_chatcmpl-bc7df29b6f1f58d1] 精确
[  32.47s][req 0_chatcmpl-bc7df29b6f1f58d1] 的数据

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the test style doc
• The test results. Please paste the results comparison before and after, or the e2e results.
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. Please run mkdocs serve to sync the documentation editions to ./docs.
• (Optional) Release notes update. If your change is user-facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c46d4d80ae

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[amy-why-3459]

Currently, we use save_async. In high concurrency scenarios, if request.is_finished is used, the output audio may be truncated.

[qibaoyuan]

it means use is_finished directly?

[qibaoyuan]

only use is_finished, test passed.

[amy-why-3459]

it means use is_finished directly?

it means use is_finished directly?

yes

[hsliuustc0106]

📝 PR Review Feedback

Thank you for fixing this issue! Here are some findings that need attention:

🔴 Critical: Indentation Error

Location: vllm_omni/model_executor/stage_input_processors/mimo_audio.py Lines 87-89

if not code_predictor_codes.any():
    if is_finished:
    return _make_finished_sentinel()   # ❌ Wrong indentation!

The return _make_finished_sentinel() statement is not properly indented under the if is_finished: block.

Suggested fix:

if not code_predictor_codes.any():
    if is_finished:
        return _make_finished_sentinel()  # ✅ Correct indentation
    return None

---

🟡 Medium Priority Issues

Issue	Description
Callers need update	The request parameter is still in the signature but no longer used. Callers should pass is_finished=request.is_finished()
Default value risk	With is_finished=False as default, callers that don't update will cause streams to never terminate

---

🟢 Suggestions

• Fix indentation error (required)
• Verify all call sites pass the correct parameter
• Update docstring to document the new is_finished parameter

---

Conclusion: The fix approach is correct ✅, but there's a critical indentation bug that must be fixed before merging.

---

Review by Claude Code via OpenClaw

[qibaoyuan]

📝 PR Review Feedback

Thank you for fixing this issue! Here are some findings that need attention:

🔴 Critical: Indentation Error

Location: vllm_omni/model_executor/stage_input_processors/mimo_audio.py Lines 87-89

if not code_predictor_codes.any():
    if is_finished:
    return _make_finished_sentinel()   # ❌ Wrong indentation!

The return _make_finished_sentinel() statement is not properly indented under the if is_finished: block.

Suggested fix:

if not code_predictor_codes.any():
    if is_finished:
        return _make_finished_sentinel()  # ✅ Correct indentation
    return None

🟡 Medium Priority Issues

Issue Description
Callers need update The request parameter is still in the signature but no longer used. Callers should pass is_finished=request.is_finished()
Default value risk With is_finished=False as default, callers that don't update will cause streams to never terminate

🟢 Suggestions

• Fix indentation error (required)
• Verify all call sites pass the correct parameter
• Update docstring to document the new is_finished parameter

Conclusion: The fix approach is correct ✅, but there's a critical indentation bug that must be fixed before merging.

Review by Claude Code via OpenClaw

Emmm, it seems that OpenClaw is doing something wrong.

• There is no indentation error — this must be a mistake.
• There’s no need to include this in the Medium Priority Issues section.

[amy-why-3459]

LGTM

[Gaohan123]

LGTM, thanks!