WIP: Support tranformers weight conversion#3071
Draft
githubnemo wants to merge 1 commit intohuggingface:mainfrom
Draft
WIP: Support tranformers weight conversion#3071githubnemo wants to merge 1 commit intohuggingface:mainfrom
githubnemo wants to merge 1 commit intohuggingface:mainfrom
Conversation
Continuation of PR huggingface#2995. Background: huggingface/transformers#42491 and huggingface/transformers#43261. This change implements conversion operations for converting some existing PEFT checkpoints, mainly dealing with the fusing of MoE layers in transformers v5. The code added here is currently a copy from the code that exists in transformers which is supposed to be gated as soon PEFT v0.19 is released and use the code in this PR. The copying makes testing a bit difficult since there's currently no routing depending on the PEFT version in transformers. Older transformers versions, therefore, need patching to forcefully use the PEFT implementation of the conversion. As soon as the routing is implemented in transformers we can conditionally disable the patching.
|
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Continuation of PR #2995.
Background: huggingface/transformers#42491 and huggingface/transformers#43261.
This change implements conversion operations for converting some existing PEFT checkpoints, mainly dealing with the fusing of MoE layers in transformers v5.
The code added here is currently a copy from the code that exists in transformers which is supposed to be gated as soon PEFT v0.19 is released and use the code in this PR.
The copying makes testing a bit difficult since there's currently no routing depending on the PEFT version in transformers. Older transformers versions, therefore, need patching to forcefully use the PEFT implementation of the conversion. As soon as the routing is implemented in transformers we can conditionally disable the patching.