Issue Station
Sign In
Tools hereToolbox

vllm - 💡(How to fix) Fix [Bug]: Qwen/Qwen3.5-122B-A10B-FP8 stuck in thinking loop with 0.19.2rc1.dev107+g4eafc7292 [1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
vllm-project/vllm#40602•Fetched 2026-04-23 07:24:03
View on GitHub
Comments
0
Participants
1
Timeline
2
Reactions
0
Author
ThiemoClausenThiemoClausen
Participants
ThiemoClausenThiemoClausen
Timeline (top)
labeled ×1subscribed ×1

Code Example

NA
RAW_BUFFERClick to expand / collapse

Your current environment

<details> <summary>The output of <code>python collect_env.py</code></summary> 
NA
</details>

🐛 Describe the bug

I am using the docker image vllm/vllm-openai:cu130-nightly with model Qwen/Qwen3.5-122B-A10B-FP8.

I am using 'override_generation_config': {'temperature': 0.6, 'top_p': 0.95, 'top_k': 20, 'min_p': 0.0, 'presence_penalty': 0.0, 'repetition_penalty': 1.0}

vllm version 0.19.1rc1.dev312+g55e1a8e10 works fine. vllm version 0.19.2rc1.dev107+g4eafc7292 hangs in thinking loops after prompting simple questions.

It kinda feels like override_generation_config is ignored, but I could not find any indication of this in the container logs.

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

extent analysis

TL;DR

Downgrade to vllm version 0.19.1rc1.dev312+g55e1a8e10 to potentially resolve the issue with the model hanging in thinking loops.

Guidance

  • Verify that the override_generation_config is being applied correctly by checking the model's configuration after initialization.
  • Test the model with a different set of configuration parameters to see if the issue is specific to the current configuration.
  • Check the container logs for any errors or warnings that may indicate why the model is hanging, even if the override_generation_config seems to be ignored.

Notes

The issue seems to be specific to vllm version 0.19.2rc1.dev107+g4eafc7292, and downgrading to a previous version may resolve the issue. However, this may not be a permanent solution, and further investigation is needed to determine the root cause of the problem.

Recommendation

Apply workaround: Downgrade to vllm version 0.19.1rc1.dev312+g55e1a8e10, as it is known to work fine with the given model and configuration.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#ssr#installation#tensor shape#autograd error

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

vllm - 💡(How to fix) Fix [Bug]: Qwen/Qwen3.5-122B-A10B-FP8 stuck in thinking loop with 0.19.2rc1.dev107+g4eafc7292 [1 participants]