Repository navigation
knowledge(vllm-omni): 2 change(s) from intake - #421
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d7a41ce5df
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| - 触发：修改 `vllm_omni/model_executor/stage_input_processors/qwen2_5_omni.py` 的 `thinker2talker_full_payload`、`vllm_omni/model_executor/models/qwen2_5_omni/qwen2_5_omni.py` 中 talker 的 text/codec embedding conditioning，或 thinker 在 stop-token 完成与长度上限截断下的音频终止行为。 | ||
| - 强制：传给 talker 的 conditioning 与 HuggingFace 参考实现保持一致：在 thinker hidden states 之外附带对应 text-token embeddings，并在终止处补 terminal text-EOS 与 text-pad embedding；自然完成与 length-cap 截断两条路径都必须落到同一终止条件，保证音频输出干净结束。注意 AR runner 不为最后采样的 token 产生 hidden-state row，两条终止路径都要补齐该缺口。 |
There was a problem hiding this comment.
This rule applies exclusively to Qwen2.5-Omni's thinker→talker conditioning, but its only index entry places it under the shared model-executor component. The existing models/qwen-omni/_index.md explicitly owns family-specific thinker/talker behavior, while components/model-executor/_index.md excludes model-specific logic. Consequently, tasks routed to the Qwen-Omni owner will not find this constraint through that owner's rules or index. Move the model-specific rule to models/qwen-omni/ and update the corresponding indexes rather than making a shared component own it. This also violates the explicit model-owner requirement in root AGENTS.md.
AGENTS.md reference: AGENTS.md:L12-L14
Useful? React with 👍 / 👎.
|
Knowledge quality gate · checked 2026-10-26 at Result: not passed. This PR will not be merged until the problems below are fixed; push a fix and it is checked again at the next daily pass. Problems:
|
Generated automatically by the Copilot knowledge service (changeset
vllm-omni-intake-8a8d71739dc2).repos/vllm-omni/components/model-executor/rules-qwen25-talker-eos.mdrepos/vllm-omni/components/diffusion/rules-master-port-tests.mdGenerator:
zcode:GLM-5.3· judge:codex:gpt-6-sol:medium.The publisher merges it only if its local gate passes on the exact merge result.