[video-comprehension] Fix failing example. #2313
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
|
The code quality check failed, please run |
ugolowic
force-pushed
the
video-comprehension-fix
branch
from
5572c8f to
f7cd918
Compare
|
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. |
astachowiczhabana
approved these changes
Oct 20, 2025
Collaborator
|
LGTM |
regisss
reviewed
Oct 21, 2025
| elif cache_position is None: | ||
| past_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0 | ||
| cache_position = torch.arange(past_length, input_ids.shape[1], dtype=torch.long, device=input_ids.device) | ||
| model_inputs["cache_position"] = cache_position |
Collaborator
There was a problem hiding this comment.
Are we sure this doesn't break generation for other models?
Collaborator
Author
There was a problem hiding this comment.
Good point.
This change is in line with transformers upgrade, but I'm running text generation slow tests. It looks fine for now, I'll update when they're done.
Collaborator
Author
There was a problem hiding this comment.
I ran text generation slow tests on main branch and on this change rebased onto the main branch and the results are exactly the same.
This commit fixes errors in video-comprehension example involving VideLLava and CLIP models and caused by latest transformers upgrade. * Adapt modeling_video_llava to the transformers upgrade by adding GaudiVideoLlavaModel.forward * Fix mismatched matrix sizes in CLIP attention. * Fix access to non-existing attribute in GaudiGenerationMixin. Signed-off-by: Urszula <[email protected]>
ugolowic
force-pushed
the
video-comprehension-fix
branch
from
f7cd918 to
316601d
Compare
astachowiczhabana
approved these changes
Nov 5, 2025
| elif cache_position is None: | ||
| past_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0 | ||
| cache_position = torch.arange(past_length, input_ids.shape[1], dtype=torch.long, device=input_ids.device) | ||
| model_inputs["cache_position"] = cache_position |
astachowiczhabana
approved these changes
Nov 6, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 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.
This commit fixes errors in video-comprehension example involving VideLLava and CLIP models and caused by latest transformers upgrade.
This is a fix for failing README example here: https://github.com/huggingface/optimum-habana/tree/main/examples/video-comprehension