- 
          
- 
                Notifications
    You must be signed in to change notification settings 
- Fork 10.9k
[Bugfix] Fix hybrid model tests #17182
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
Conversation
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
| 👋 Hi! Thank you for contributing to the vLLM project. 💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels. Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run  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 either: Add  🚀 | 
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you for doing this - LGTM when green
Some comments on the test failures
- 
Looks like some tests aren't getting cleaned up properly 
- 
This is a bug -- the ConstantSizeCache unfortunately needs to look at the attention metadata and is choking on xgrammar being there instead of flash attn 
[2025-04-25T12:45:10Z] FAILED models/decoder_only/language/test_hybrid.py::test_multistep_correctness[64-float-ai21labs/Jamba-tiny-dev] - ValueError: Multi-Step not supported for attention backend: XFORMERS. Set VLLM_ATTENTION_BACKEND to a value from ['FLASH_ATTN', 'ROCM_FLASH', 'FLASHINFER', 'NO_ATTENTION'].
- I've found the hybrid and mamba tests to be pretty flakey... Somewhat suspect some numerical instability either in the models or the implementation but I don't have any hints as to what would be going wrong
[2025-04-25T12:45:10Z] FAILED models/decoder_only/language/test_hybrid.py::test_mamba_prefill_chunking[7-bfloat16-Zyphra/Zamba2-1.2B-instruct] - AssertionError: Test1:
--
  | [2025-04-25T12:45:10Z] chunked:	'Briefly describe the major milestones in the development of artificial intelligence from 1950 to 2020.\n\nC. Discuss the impact'
  | [2025-04-25T12:45:10Z] non_chunked:	'Briefly describe the major milestones in the development of artificial intelligence from 1950 to 2020.\n\nC. Discuss the ethical'
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk> Signed-off-by: Agata Dobrzyniewicz <adobrzyniewicz@habana.ai>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk> Signed-off-by: Mu Huai <tianbowen.tbw@antgroup.com>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk> Signed-off-by: Yuqi Zhang <yuqizhang@google.com>
Fix failing tests in
test_hybrid.py. Also, consolidate the tests fromtest_mamba.pyinto this file to avoid duplicated code.