Fix incorrect batching audio index calculation for Phi-4-Multimodal #38103
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
|
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. |
Isotr0py
commented
May 13, 2025
Comment on lines
+162
to
+171
| speech_inputs = [floats_list((1, x))[0] for x in range(800, 1400, 200)] | ||
| np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs] | ||
| pt_speech_inputs = [torch.tensor(speech_input) for speech_input in speech_inputs] | ||
|
|
||
| # Test feature size | ||
| input_features = feature_extractor(np_speech_inputs, return_tensors="np").audio_input_features | ||
| max_audio_len = (1200 - feature_extractor.win_length) // feature_extractor.hop_length + 1 | ||
| self.assertTrue(input_features.ndim == 3) | ||
| self.assertTrue(input_features.shape[-1] == feature_extractor.feature_size) | ||
| self.assertTrue(input_features.shape[-2] == max_audio_len) |
Collaborator
Author
There was a problem hiding this comment.
Have confirmed this test can cover the original bug without fix, so I didn't add extra regression test:
> masked_frames = frames[to_mask_batch_idxs, offset_idx:max_idx].masked_fill_(mask, 0)
E RuntimeError: The size of tensor a (3) must match the size of tensor b (4) at non-singleton dimension 1
src/transformers/models/phi4_multimodal/feature_extraction_phi4_multimodal.py:312: RuntimeError
================================================================== short test summary info ==================================================================
FAILED tests/models/phi4_multimodal/test_feature_extractor_phi4_multimodal.py::Phi4MultimodalFeatureExtractionTest::test_call - RuntimeError: The size of tensor a (3) must match the size of tensor b (4) at non-singleton dimension 1
Signed-off-by: Isotr0py <2037008807@qq.com>
Member
ArthurZucker
approved these changes
May 26, 2025
Collaborator
ArthurZucker
left a comment
ArthurZucker
left a comment
There was a problem hiding this comment.
Thanks, let's be explicit as python can hold surprises!
src/transformers/models/phi4_multimodal/feature_extraction_phi4_multimodal.py
Outdated
Show resolved
Hide resolved
…4_multimodal.py Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
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.
What does this PR do?
Fixes #38098 (issue)
Just notice there is no feature extractor test for Phi-4-MM, let me add one for it, so mark this as draft currently.Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.