Fix arg in bettertransformer llama attention #1421
Merged
Conversation
This file contains 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
SunMarc
changed the title
younesbelkada
approved these changes
Sep 28, 2023
|
Yeah, we can do that too since FA2 is experimental at this stage and will be removed in the future from transformers. LMK what you think @fxmarty. |
|
LGTM, thank you for fixing! Yes, feel free to add kwargs everywhere in this PR. |
Closed
4 tasks
fxmarty
pushed a commit
that referenced
this pull request
Nov 3, 2023
* fix arg in llama attention * change to kwargs * add kwargs everwhere --------- Co-authored-by: younesbelkada <[email protected]>
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Outdated suggestions cannot be applied.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 ?
This PR fixes the integration of llama in bettertransformers. Since the PR about flashattention2 in llama models was merged, a new argument
padding_maskwas introduced silently in the forward of the attention module. This breaks the forward of thellama_forwardin bettertransformer.@younesbelkada We may have to do it for every model that supports flashattention2. LMK if I should just do one PR for all supported + plan to support models ?