Fix metric padding removal after skip_first_batches - #4235
Merged
SunMarc merged 3 commits intoOct 1, 2026
Merged
Conversation
SunMarc
reviewed
Sep 30, 2026
SunMarc
left a comment
SunMarc
left a comment
Member
There was a problem hiding this comment.
Thanks, can you try to iterate a bit on the test to make them simpler ? thanks !
Comment on lines
+617
to
+627
| @parameterized.expand( | ||
| [ | ||
| (split, drop, skips, sampler_arg) | ||
| for split in (False, True) | ||
| for drop in (False, True) | ||
| for skips in ((0,), (1,), (1, 1)) | ||
| for sampler_arg in (False, True) | ||
| if not (split and sampler_arg) # batch_size=None does not support split_batches. | ||
| ] | ||
| ) | ||
| def test_skip_first_batches_preserves_batch_metadata(self, split_batches, drop_last, skips, sampler_arg): |
Member
There was a problem hiding this comment.
this is a bit too much parametrized. we can just check the most common case.
Comment on lines
+379
to
+381
| for num_items in (3 * total_batch_size, 3 * total_batch_size + 1): | ||
| for drop_last in (False, True): | ||
| for skips in ((0,), (1,), (1, 1)): |
Contributor
Author
|
Thanks for the feedback! I鈥檒l simplify both tests to focus on the common case and keep the regression coverage concise. |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
skip_first_batcheswraps the batch sampler inSkipBatchSampler, which does not exposebatch_size. This makesDataLoaderShard.total_batch_sizeraise; the suppressed exception leavesremainder=-1, sogather_for_metricskeeps duplicated padding samples in the final batch.With 10 samples, two processes and a per-process batch size of 2, normal iteration gathers 10 samples. Calling
skip_first_batches(loader, 0)instead gathers 12, including duplicated samples 0 and 1.Unwrap skip samplers before reading the existing batch-size metadata. This also handles nested skipping without changing sampling order. The existing loader-configuration preservation from #4296 is retained.
Validation
drop_lastafter skipping one batch.mainand pass with this fix.make qualityandgit diff --checkagainst currentmainpass.GPU/TPU execution was not tested. Changes were prepared with AI assistance and reviewed by the contributor.
Before submitting