[BugFix] Qwen2.5-omni supress end token and won't stop

[yinpeiqi]

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

Purpose

In the current implementation, bad_word_processor will set the logits to -1e9 when encounter supress tokens, to avoid the vocab size mismatch. We found that the end token also will be set to -1e9, thus the program won't stop until reach max length.

Test Plan

Run end2end.py

Test Result

Before:

0_8990c4c2-dede-49cf-95b1-5cb93ef0ad05.txt
output_0_8990c4c2-dede-49cf-95b1-5cb93ef0ad05.wav

After:

0_163825c2-5c08-4fd1-89f1-f4ac9668ea26.txt
output_0_163825c2-5c08-4fd1-89f1-f4ac9668ea26.wav

---

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, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (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)

@Bounty-hunter @hsliuustc0106

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bd75674698

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".

[hsliuustc0106]

@yenuo26 how can we use tests to catch such error?

[yenuo26]

@yenuo26 how can we use tests to catch such error?

we use whisper in test case to convert speech to text. If the audio contains long segments of silence, it may lead to a discrepancy between the recognized text and the inferred text, ultimately causing the test case to fail.
like this pr: #720