[TPU][Bugfix] fix test_pallas #20666
Conversation
Signed-off-by: Chengji Yao <chengjiyao@google.com>
|
👋 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.
Summary of Changes
Hello @yaochengji, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!
This pull request resolves a bug in the test_pallas.py test file, specifically for TPU-related functionality. The fix involves updating the FakeAttentionLayer class to include a num_kv_update_slices parameter and correctly passing this parameter to the pallas_attention function, which was necessary for the test to pass.
Highlights
- Test Fix: Corrected the
test_pallas.pyfile by adding thenum_kv_update_slicesattribute to theFakeAttentionLayerclass and ensuring it is passed as an argument to thepallas_attentionfunction.
Using Gemini Code Assist
The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.
Invoking Gemini
You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.
| Feature | Command | Description |
|---|---|---|
| Code Review | /gemini review |
Performs a code review for the current pull request in its current state. |
| Pull Request Summary | /gemini summary |
Provides a summary of the current pull request in its current state. |
| Comment | @gemini-code-assist | Responds in comments when explicitly tagged, both in issue comments and review comments. |
| Help | /gemini help |
Displays a list of available commands. |
Customization
To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.
Limitations & Feedback
Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.
You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.
Footnotes
-
Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩
There was a problem hiding this comment.
Code Review
The pull request fixes the tests/v1/tpu/test_pallas.py by adding num_kv_update_slices to the FakeAttentionLayer class and passing it to PallasMetadata. I suggested defining the tensor directly and ensuring the parameter is documented in the PallasMetadata class definition.
| @@ -50,6 +50,7 @@ class FakeAttentionLayer: | |||
| slot_mapping = torch.zeros((3, num_tokens), dtype=torch.int64) | |||
| max_num_reqs = 8 | |||
| max_num_blocks_per_req = 8 | |||
| num_kv_update_slices = torch.tensor([num_tokens], dtype=torch.int32) | |||
There was a problem hiding this comment.
Consider defining num_kv_update_slices directly as a tensor with the correct dtype instead of creating a list and then converting it. This can improve readability and potentially performance.
num_kv_update_slices = torch.tensor([num_tokens], dtype=torch.int32)| @@ -65,6 +66,7 @@ class FakeAttentionLayer: | |||
| context_lens=context_lens, | |||
| query_start_loc=query_start_loc, | |||
| num_seqs=num_seqs, | |||
| num_kv_update_slices=num_kv_update_slices, | |||
There was a problem hiding this comment.
It's good to see num_kv_update_slices being passed to PallasMetadata. Ensure that this parameter is properly documented in the PallasMetadata class definition in vllm/v1/attention/backends/pallas.py to explain its purpose.
num_kv_update_slices=num_kv_update_slices,|
Is this ok to merge? The test didn't run on CI because of timeout |
|
@DarkLight1337 thanks for reviewing. I already tested it locally and we can merge it. |
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com> Signed-off-by: Patrick von Platen <patrick.v.platen@gmail.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com> Signed-off-by: avigny <47987522+avigny@users.noreply.github.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com> Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Signed-off-by: Chengji Yao <chengjiyao@google.com>
Purpose
Fix
tests/v1/tpu/test_pallas.pyTest Plan
Test Result
Passed