Clean-up composite configs#34603
Merged
zucchini-nlp merged 14 commits intohuggingface:mainfrom Jan 15, 2025
Merged
Conversation
|
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. |
ArthurZucker
approved these changes
Jan 6, 2025
Collaborator
ArthurZucker
left a comment
There was a problem hiding this comment.
Thanks!
I would make sure all tests are run as for tie weight embedding here the changes would only work for the decoder (decoder=True) !
A commit with run-all is enough! 🤗
5 tasks
Member
Author
|
Tests are passing, the |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
Based on #34410, I realized that BLIP models were the only ones failing a certain common test and the reason is that we manually set the
tie_word_embeddingfromtext_config. I think some config attributes, includingvocab_size(successfully deprecated in VLMs) ortie_word_embeddingsshould be only attributes of text config. We should access them viaconfig.get_text_config()whenever neededHaving an option to access through general composite config and through text config creates two sources of truth, and we might easily end up with situations like in the
check_config_arguments_init. If one sets the attribute throughsetattr, it will not be different from the attribute in text config.So we expect users to set the attr in the respective config if they want it to be used correctly with transformers code. We could restrict setting these common attr in correct sub-config by
attribute_mapprob, but I thought it was too muchPlus, clean up some
attn_implementation=config._attn_implementationthat were left, because it should be passed already as part of respective config