[CI] add diffusion ci #174
Conversation
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[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
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".
.buildkite/pipeline.yml
Outdated
| - ".buildkite/scripts/setup_env.sh" | ||
| - pytest -v -s tests/test_omni_llm.py |
There was a problem hiding this comment.
The pipeline now runs .buildkite/scripts/setup_env.sh to create and populate a venv, but then invokes pytest directly rather than ${VENV_DIR}/bin/python -m pytest as before. Because the script does not activate the venv or modify PATH, these commands will use the agent’s system Python, which will not see the dependencies installed into .venv-simple-test and can fail with missing packages whenever the agent image doesn’t already have them. This is a regression from the previous script that executed tests via the venv interpreter.
Useful? React with 👍 / 👎.
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
| width=width, | ||
| num_inference_steps=9, | ||
| guidance_scale=0.0, | ||
| generator=torch.Generator("cuda").manual_seed(42), |
There was a problem hiding this comment.
i'd suggest to set
num_inference_steps=2,
num_images_per_prompt=2,
There was a problem hiding this comment.
May be we should align num_images_per_prompt and num_outputs_per_prompt later.
Signed-off-by: zjy0516 <[email protected]>
Gaohan123
left a comment
Gaohan123
left a comment
There was a problem hiding this comment.
Please think of the reviews a little bit. Besides, please provide local test results and time consuming in the PR description. Thanks!
| @@ -0,0 +1,57 @@ | |||
| #!/usr/bin/env bash | |||
|
|
|||
There was a problem hiding this comment.
Is it possible to use this file as a general env setup for all current examples? If it is, we can rename it. And also it can be used to run all UT/ST.
There was a problem hiding this comment.
I have tried it before. But I don't know how to set environment variables persistent which will lead to error like /bin/bash: line 2: pytest: command not found.
We will have a general env setup when dockerfile is ready.
There was a problem hiding this comment.
@congw729 Could you please help to resolve this? I don't have bandwidth today.
|
|
||
| "${UV_BIN}" pip install --python "${VENV_PYTHON}" vllm==0.11.0 | ||
| "${UV_BIN}" pip install --python "${VENV_PYTHON}" -e ".[dev]" | ||
| "${VENV_PYTHON}" -m pytest -s -v tests/test_diffusion_model.py |
There was a problem hiding this comment.
If it is possbile, all other test can run sequentially.
There was a problem hiding this comment.
We will add more e2e tests here like pytest -sv tests/omni/test_qwen_omni.py. So we should make sure the set up of environment sufficient. I will also test this script for #168.
Maybe we could change the name of this script to model_test.sh to make it general.
Signed-off-by: zjy0516 <[email protected]>
Signed-off-by: zjy0516 <[email protected]>
4 participants
Purpose
Test Plan
Test Result
Essential Elements of an Effective PR Description Checklist
supported_models.mdandexamplesfor a new model.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)