[CI/Build] Add text-only test for Qwen models

[alex-jw-brooks]

Hello! I noticed that there aren't any model tests for QwenModel yet, and was wondering if there was a reason for this, or if it would make sense to add one similar to the other tests running on the example test prompts.

I am working on extending QwenModel, which currently skips loading the visual component of models like Qwen/Qwen-VL in VLLM so that it can handle text-only inputs, to accept visual inputs. I thought it would be a good idea to potentially add a test for text-only inputs, which could potentially be updated to include visual inputs once that is at a point that it's ready for a look.

It's also probably worth noting that there are some small differences in the produced output for a few of the example prompts, although the produced sequences start the same and are quite close, so maybe it's not an issue. For example:

tests/models/test_qwen.py::test_text_only_qwen_model[qwen/qwen-vl-5-32-half]
  /u/brooks/vllm/tests/models/test_qwen.py:44: UserWarning: Test6:
  Matched tokens:       [71703, 25, 576, 98783, 28556, 18824, 374, 6509, 825, 315, 279, 1429]
  hf:   'Assistant: The Mona Lisa painting is considered one of the most famous and iconic works of art in the world. It is a portrait of a woman, believed to'{11245: -0.747456431388855, 26277: -0.747456431388855, 5089: -4.1224565505981445, 65252: -4.6224565505981445, 35948: -5.2474565505981445}
  vllm: 'Assistant: The Mona Lisa painting is considered one of the most iconic and famous works of art in the world. It is a portrait of a woman, believed to'{26277: Logprob(logprob=-0.735495388507843, rank=1, decoded_token=' iconic'), 11245: Logprob(logprob=-0.766745388507843, rank=2, decoded_token=' famous'), 5089: Logprob(logprob=-4.079245567321777, rank=3, decoded_token=' significant'), 65252: Logprob(logprob=-4.516745567321777, rank=4, decoded_token=' recognizable'), 35948: Logprob(logprob=-5.235495567321777, rank=5, decoded_token=' renowned')}
    check_logprobs_close(

---

PR Checklist (Click to Expand)

Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.

PR Title and Classification

Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:

• [Bugfix] for bug fixes.
• [CI/Build] for build or continuous integration improvements.
• [Doc] for documentation fixes and improvements.
• [Model] for adding a new model or improving an existing model. Model name should appear in the title.
• [Frontend] For changes on the vLLM frontend (e.g., OpenAI API server, LLM class, etc.)
• [Kernel] for changes affecting CUDA kernels or other compute kernels.
• [Core] for changes in the core vLLM logic (e.g., LLMEngine, AsyncLLMEngine, Scheduler, etc.)
• [Hardware][Vendor] for hardware-specific changes. Vendor name should appear in the prefix (e.g., [Hardware][AMD]).
• [Misc] for PRs that do not fit the above categories. Please use this sparingly.

Note: If the PR spans more than one category, please include all relevant prefixes.

Code Quality

The PR need to meet the following code quality standards:

• We adhere to Google Python style guide and Google C++ style guide.
• Pass all linter checks. Please use format.sh to format your code.
• The code need to be well-documented to ensure future contributors can easily understand the code.
• Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.
• Please add documentation to docs/source/ if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.

Notes for Large Changes

Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with rfc-required and might not go through the PR.

What to Expect for the Reviews

The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:

• After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.
• After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.
• After the review, the reviewer will put an action-required label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.
• Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion.

Thank You

Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!

[github-actions]

👋 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 consists a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of default ones by unblocking the steps in your fast-check build on Buildkite UI.

Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge).

To run full CI, you can do one of these:

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

🚀

[DarkLight1337]

Thanks for adding a test! Keep in mind though that we're trying to slim down on the number of model tests to be run in CI, so your test might be deactivated in a future PR (can still run on demand).

[DarkLight1337]

Seems that Qwen test requires an additional dependency.

[mgoin]

FYI we do test Qwen text LLMs in a few places, just not the VL style mentioned here

vllm/.buildkite/lm-eval-harness/configs/models-small.txt

Lines 8 to 9 in d3d9cb6

	Qwen2-1.5B-Instruct-INT8-compressed-tensors.yaml
	Qwen2-1.5B-Instruct-FP8W8.yaml

vllm/.buildkite/lm-eval-harness/configs/models-large.txt

Line 4 in d3d9cb6

Qwen2-57B-A14-Instruct.yaml

vllm/tests/models/test_gguf.py

Line 27 in d3d9cb6

("Qwen/Qwen2-1.5B-Instruct",

[alex-jw-brooks]

Thanks @DarkLight1337, that makes sense! And whoops, I forgot that it needs transformers_stream_generator. I added it to the test dependencies, although it looks like dependencies inrequirements-test.txt aren't installed in the CPU test check etc. Do you have a preference for how to handle this / if it's worthing adding the test at all? This would be a vlm test later on, which will be skipped there anyway, so we could also just close this PR for now and consider adding the test once I have a PR for multimodal support for Qwen if that would be better 🙂

Thanks for the heads up @mgoin! I did look a bit, but I didn't see any QWenLMHeadModels - I think those are Qwen2ForCausalLM/Qwen2MoeForCausalLM models, which would test qwen2 and qwen2moe, but not qwen unless I'm missing something

[DarkLight1337]

Thanks @DarkLight1337, that makes sense! And whoops, I forgot that it needs transformers_stream_generator. I added it to the test dependencies, although it looks like dependencies inrequirements-test.txt aren't installed in the CPU test check etc. Do you have a preference for how to handle this / if it's worthing adding the test at all? This would be a vlm test later on, which will be skipped there anyway, so we could also just close this PR for now and consider adding the test once I have a PR for multimodal support for Qwen if that would be better 🙂

We need to resolve those dependency errors sooner or later since they will occur in the VLM test as well. Based on the test failures, looks like matplotlib has to be installed as well. Regarding the CPU tests, you can add dependencies by editing this line:

vllm/.buildkite/run-cpu-test.sh

Line 25 in 22b39e1

pip install pytest Pillow protobuf

[alex-jw-brooks]

Thanks for the guidance @DarkLight1337! I've updated the deps and rebased the branch :)