Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Misc] Optimize Qwen2-VL LoRA test #11663

Merged
merged 4 commits into from
Jan 1, 2025

Conversation

jeejeelee
Copy link
Collaborator

@jeejeelee jeejeelee commented Jan 1, 2025

I retrained a QWen-VL LoRA. In my local environment using A800, I tested the model on 100 image samples, the generation results were completely aligned with the transformers output.

BTW, add LoRA support for vision modules and projecter modules

Signed-off-by: Jee Jee Li <[email protected]>
Signed-off-by: Jee Jee Li <[email protected]>
Signed-off-by: Jee Jee Li <[email protected]>
Copy link

github-actions bot commented Jan 1, 2025

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

  • Add ready label to the PR
  • Enable auto-merge.

🚀

@jeejeelee jeejeelee added the ready ONLY add when PR is ready to merge/full CI is needed label Jan 1, 2025
@@ -7,7 +7,7 @@
from vllm.lora.request import LoRARequest
from vllm.platforms import current_platform

MODEL_PATH = "Qwen/Qwen2-VL-7B-Instruct"
MODEL_PATH = "Qwen/Qwen2-VL-2B-Instruct"

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I switched to the 2B model to reduce training time and decrease the computational pressure on CI testing pipeline.

Copy link
Member

@DarkLight1337 DarkLight1337 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I assume this fixes the test?

vllm/model_executor/models/qwen2_vl.py Outdated Show resolved Hide resolved
Signed-off-by: Jee Jee Li <[email protected]>
@jeejeelee
Copy link
Collaborator Author

I assume this fixes the test?

I tested it on the A800 and it works without any issues. I used 100 samples for testing, and the generated results align perfectly with transformers. Regarding the previous LoRA, there might have been an issue during training. Although it could generate reasonable results, only about 30% of the outputs aligned correctly

@DarkLight1337 DarkLight1337 merged commit 11d8a09 into vllm-project:main Jan 1, 2025
58 checks passed
@jeejeelee jeejeelee deleted the fix-qwen2vl-lora branch January 1, 2025 14:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ready ONLY add when PR is ready to merge/full CI is needed
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants