Conversation
Signed-off-by: WANG Cong <[email protected]>
Signed-off-by: Roger Wang <[email protected]>
Signed-off-by: Roger Wang <[email protected]>
Signed-off-by: Roger Wang <[email protected]>
Signed-off-by: Roger Wang <[email protected]>
![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.
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".
| from vllm_omni.entrypoints.omni_llm import OmniLLM | ||
|
|
||
| llm = OmniLLM(model="any", init_timeout=1) |
There was a problem hiding this comment.
Each test constructs OmniLLM(model="any", init_timeout=1) without overriding init_sleep_seconds (e.g. here), but OmniLLM defaults to init_sleep_seconds=20 and _start_stage_processes sleeps for that duration per stage (see vllm_omni/entrypoints/omni_llm.py:58,132). A two-stage config will therefore block ~40s per test and even the timeout test waits those sleeps before failing, turning the supposed “Simple Unit Test” pipeline into a multi‑minute run and not actually exercising short timeouts.
Useful? React with 👍 / 👎.
Signed-off-by: Roger Wang <[email protected]>
|
We can work on upgrading dependencies once we get some CI running to validate that vllm-omni can be run successfully. |
Signed-off-by: WANG Cong <[email protected]> Signed-off-by: Roger Wang <[email protected]> Co-authored-by: WANG Cong <[email protected]>
3 participants
Adopted test from #88