[Bugfix] Fix auto dtype casting for BatchFeature #19316
Merged
+85
−57
Conversation
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
Signed-off-by: Isotr0py <[email protected]>
|
👋 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 🚀 |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hello @Isotr0py, 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!
Summary of Changes
Hello team, gemini-code-assist here with a summary of this pull request. This PR addresses a bug related to automatic dtype casting when processing inputs using Hugging Face processors, specifically when the processor returns a BatchFeature object. Previously, the casting logic might not have correctly applied the desired dtype to the tensors contained within the BatchFeature. This change modifies the casting process to ensure that when a BatchFeature is returned, the dtype casting is applied directly to the internal data (.data) of the BatchFeature, fixing the issue where tensors inside the BatchFeature were not being cast correctly.
Highlights
- Bugfix: BatchFeature dtype casting: Fixes an issue where the automatic dtype casting for inputs processed by Hugging Face processors did not correctly handle cases where the output was a
BatchFeature, leading to tensors within theBatchFeatureretaining their original dtype instead of being cast to the model's dtype. - Refined input processing logic: Modifies the
maybe_cast_dtypefunction invllm/inputs/registry.pyto specifically target the.dataattribute ofBatchFeatureobjects for dtype casting, ensuring that the tensors inside are correctly handled.
Changelog
- vllm/inputs/registry.py
- Modified the
maybe_cast_dtypefunction to check if the processor output is aBatchFeature(line 171). - If the output is a
BatchFeature, thejson_map_leavescasting function is now applied tooutput.datainstead ofoutput(line 172). - The original casting logic for non-
BatchFeatureoutputs is now handled after theifblock (line 175).
- Modified the
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
This pull request effectively addresses the bug regarding auto dtype casting for BatchFeature objects. The core issue, as I understand it, was that json_map_leaves did not correctly recurse into BatchFeature instances because isinstance(BatchFeature_instance, dict) evaluates to False (as BatchFeature inherits from UserDict but not dict directly). This meant maybe_cast_dtype was applied to the BatchFeature object itself, resulting in no casting of its internal tensors.
The fix correctly targets this by explicitly passing output.data (which is a dict) to json_map_leaves when output is a BatchFeature. This ensures that the tensors within the BatchFeature are properly cast. For other types of output, the existing logic for json_map_leaves is maintained.
The change is clear, well-targeted, and improves the robustness of the dtype casting mechanism. Well done!
Merge Readiness
The changes in this pull request appear to be correct and effectively address the described bug. The code is clear and the fix is well-targeted. Based on my review, the PR seems ready for merging. However, as an AI, I am not authorized to approve pull requests; please ensure it undergoes further review and approval by the maintainers.
github-actions
bot
added
the
ready
jeejeelee
approved these changes
Jun 8, 2025
|
Hmm it looks like we are now running into a similar issue as #18862 (comment) |
|
Hmmm, the stuck test can pass on my side locally with this PR: Let's see if merge from main branch can fix this... |
Signed-off-by: Isotr0py <[email protected]>
Signed-off-by: Isotr0py <[email protected]>
Isotr0py
commented
Jun 11, 2025
|
I reproduced the hanging issue on 34a5713 locally, and seems the deadlock is caused by tensor dtype conversion exactly: Lines 162 to 166 in 34a5713 |
|
I assume something interacts badly with the forking but not really sure how to fix it either. |
Signed-off-by: Isotr0py <[email protected]>
Signed-off-by: Isotr0py <[email protected]>
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
Isotr0py
commented
Jun 12, 2025
Comment on lines
+111
to
+112
| with set_default_torch_num_threads(1): | ||
| engine = AsyncLLM.from_engine_args(engine_args) |
There was a problem hiding this comment.
Seems that disable openmp by setting torch_num_threads=1 during engine forking can fix the deadlock issue locally. Let's see what's the CI going on then.
Signed-off-by: Isotr0py <[email protected]>
Signed-off-by: Isotr0py <[email protected]>
4 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Essential Elements of an Effective PR Description Checklist
Purpose
RuntimeError: Input type (float) and bias type (c10::BFloat16) should be the same#19219BatchFeatureis a UserDict, andif isinstance(value, dict)will return False, causing its data not casted correctly injson_map_leaves.Test Plan
Test Result