Conversation
![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: a4b62c07ae
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".
|
|
||
| The script allows overriding: | ||
| - `MODEL` (default: `Wan-AI/Wan2.2-T2V-A14B-Diffusers`) | ||
| - `PORT` (default: `8091`) |
There was a problem hiding this comment.
This page says run_server.sh defaults PORT to 8091, but the actual script defaults to 8098 (examples/online_serving/text_to_video/run_server.sh), and its companion curl script also targets 8098; users who follow this doc literally will start the server on one port and send requests to another, resulting in connection failures unless they discover and override the port manually.
Useful? React with 👍 / 👎.
|
|
||
| The script allows overriding: | ||
| - `MODEL` (default: `Wan-AI/Wan2.2-I2V-A14B-Diffusers`) | ||
| - `PORT` (default: `8091`) |
There was a problem hiding this comment.
The document states run_server.sh defaults PORT to 8091, but the real script uses 8099 (examples/online_serving/image_to_video/run_server.sh) and the example curl script posts to 8099, so following this page causes users to send requests to the wrong port after launching with the provided script.
Useful? React with 👍 / 👎.
| - `BOUNDARY_RATIO` (default: `0.875`) | ||
| - `FLOW_SHIFT` (default: `12.0`) |
There was a problem hiding this comment.
This section claims run_server.sh supports BOUNDARY_RATIO and FLOW_SHIFT overrides, but the script does not read or pass either value (it only wires model/port/cache flags), so users setting those env vars will get silently ignored behavior and non-reproducible tuning attempts.
Useful? React with 👍 / 👎.
| | `ref_audio` | string | **Yes** | Reference audio (URL or base64 data URL) | | ||
| | `ref_text` | string | No | Transcript of reference audio (for ICL mode) | | ||
| | `x_vector_only_mode` | bool | No | Use speaker embedding only (no ICL) | | ||
| | `ref_audio` | string | null | Reference audio (URL or base64 data URL) | |
There was a problem hiding this comment.
The Base-task parameter table now presents ref_audio as nullable/default-null, but the server explicitly rejects Base requests without it (vllm_omni/entrypoints/openai/serving_speech.py returns Base task requires 'ref_audio' for voice cloning), so this documentation change leads users to construct invalid requests that fail at runtime.
Useful? React with 👍 / 👎.
1 participant
PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.
Purpose
Manually add
features/custom_pipeline.mdto.nav.yml.Fix grammar in the PR template.
Other editions are auto-generated after
mkdocs serve.Test Plan
Test Result
Successfully launch without warning.
Essential Elements of an Effective PR Description Checklist
supported_models.mdandexamplesfor a new model. Please runmkdocs serveto sync the documentation editions to./docs.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)