Fix #44155: [AudioFlamingo3] Batched inference produces incorrect result#44191
Closed
danielalanbates wants to merge 1 commit intohuggingface:mainfrom
Closed
Fix #44155: [AudioFlamingo3] Batched inference produces incorrect result#44191danielalanbates wants to merge 1 commit intohuggingface:mainfrom
danielalanbates wants to merge 1 commit intohuggingface:mainfrom
Conversation
… misalignment The processor and model computed audio token counts with different formulas, causing masked_scatter to misalign embeddings across batch items when audio clips had different lengths. Processor formula (correct): sum mask values across all windows for a sample → apply downsampling Old model formula (wrong): apply downsampling per window → sum results Fix: - Pass `windows_per_sample` from the processor to the model so that get_audio_features can group windows by sample and apply the same formula as the processor. - Processor now includes `windows_per_sample` in its output BatchFeature and model_input_names. - forward() and prepare_inputs_for_generation() thread windows_per_sample through to get_audio_features. - When windows_per_sample is not provided (e.g. single-sample inference), the original per-window path is used as a fallback. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: audioflamingo3 |
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 #44155
Summary
This PR fixes: [AudioFlamingo3] Batched inference produces incorrect results due to embedding/token leak between tracks
Changes
Testing
Please review the changes carefully. The fix was verified against the existing test suite.
This PR was created with the assistance of Claude Sonnet 4.6 by Anthropic | effort: low. Happy to make any adjustments!