test: Add responses test suite vllm-qwen35 with recordings#5216
Closed
msager27 wants to merge 2 commits intollamastack:mainfrom
Closed
test: Add responses test suite vllm-qwen35 with recordings#5216msager27 wants to merge 2 commits intollamastack:mainfrom
msager27 wants to merge 2 commits intollamastack:mainfrom
Conversation
meta-cla
bot
added
the
CLA Signed
Contributor
|
✅ Recordings committed successfully Recordings from the integration tests have been committed to this PR. |
Contributor
|
This pull request has merge conflicts that must be resolved before it can be merged. @msager27 please rebase it. https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork |
leseb
added a commit
that referenced
this pull request
Mar 30, 2026
# What does this PR do? Updates a few responses integration tests based on testing with vLLM. Some context: I initially tested with vLLM + Qwen3.5 as part of #5216. That PR was more of a staging effort and will be mostly obsoleted by #5297 and get closed. However, there are a few changes from that PR that I've pulled into separate PRs: 1. This PR which makes one of the web search tests more flexible in its validation (plus a couple skips when provider is vllm) 2. #5233 <!-- If resolving an issue, uncomment and update the line below --> <!-- Closes #[issue-number] --> ## Test Plan Rerun the responses web search tests and verify they work as expected <!-- For API changes, include: 1. A testing script (Python, curl, etc.) that exercises the new/modified endpoints 3. The output from running your script Example: ```python ... ... ``` Output: ``` <paste actual output here> ``` --> Co-authored-by: Sébastien Han <seb@redhat.com>
Contributor
Author
|
Closing this PR. It included some initial staging work for discussion, but this work will be superseded by PR #5297. New vLLM recordings are expected to be created as either part of that PR or a new one. This PR also included a couple fixes. These have been submitted as separate PRs. One was merged. The other is awaiting review. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Add a new responses test suite vllm-qwen35. Recordings for these tests are included. I know some changes are in the works for the recording system. As such, I'm making this a draft PR for discussion purposes.
The vLLM was deployed on an AWS EC2 instance with GPUs. The vLLM version is 0.17.1. The model is Qwen/Qwen3.5-35B-A3B. Requiring a GPU does complicate things if we need to re-record. That's being looked into within the community. I used a Qwen3.5 model since it supports image input.
I included one small fix to the recording system. A couple of the web search tool tests were overwriting the corresponding gpt recording. I added test_id in normalize_tool_request to remedy this.
Test Plan
Run the responses test suite with the above-mentioned vLLM + Qwen model in record mode. Then verify everything works in replay mode with:
uv run ./scripts/integration-tests.sh --stack-config server:ci-tests --setup vllm-qwen35 --inference-mode replay --subdirs responses