[BugFix] Fix batch updates for pooling models #23398
Merged
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: Nick Hill <[email protected]>
1 task
Signed-off-by: Nick Hill <[email protected]>
njhill
requested review from
WoosukKwon,
alexm-redhat,
comaniac,
robertgshaw2-redhat and
ywang96
as code owners
|
@maxdebayser Could you please review? |
maxdebayser
approved these changes
Aug 22, 2025
There was a problem hiding this comment.
Reading the code it seems to me that this should fix the bug. I've already opened an issue to add CI tests for this case as a follow up: #23436
DarkLight1337
approved these changes
Aug 23, 2025
epwalsh
pushed a commit
to epwalsh/vllm
that referenced
this pull request
Aug 28, 2025
Signed-off-by: Nick Hill <[email protected]>
xiao-llm
pushed a commit
to xiao-llm/vllm
that referenced
this pull request
Aug 28, 2025
Signed-off-by: Nick Hill <[email protected]> Signed-off-by: Xiao Yu <[email protected]>
zhewenl
pushed a commit
to zhewenl/vllm
that referenced
this pull request
Aug 28, 2025
Signed-off-by: Nick Hill <[email protected]>
mengxingkongzhouhan
pushed a commit
to mengxingkongzhouhan/vllm
that referenced
this pull request
Aug 30, 2025
Signed-off-by: Nick Hill <[email protected]>
zhewenl
pushed a commit
to zhewenl/vllm
that referenced
this pull request
Sep 3, 2025
Signed-off-by: Nick Hill <[email protected]>
ekagra-ranjan
pushed a commit
to ekagra-ranjan/vllm
that referenced
this pull request
Sep 4, 2025
Signed-off-by: Nick Hill <[email protected]> Signed-off-by: Ekagra Ranjan <[email protected]>
1 task
1 task
FeiDaLI
pushed a commit
to FeiDaLI/vllm
that referenced
this pull request
Sep 25, 2025
Signed-off-by: Nick Hill <[email protected]>
1 task
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
Fixes #23381.
Changes in #19912 mis-assumed that pooling requests only last a single step. But with sequence chunking this might not be the case.
We should have CI tests to cover this, to be added in a follow-on.
Unfortunately this bug is in the v0.10.1 and v0.10.1.1 releases.
Summary of changes:
BatchUpdateBuilderinInputBatchto track removed requests consistent with non-pooling case, used to fill in empty slots when requests are added in the same stepbatch_changedflag toBatchUpdateBuilderto track whether requests have been added (since the added list is only used for non-pooling reqs with sampling params)condense()andswap_states()so that only the applicable ones are done in the pooling case