Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

New transformers caching ETA now v4.38 #1348

Merged
merged 1 commit into from
Jan 12, 2024
+3 −3
Merged

New transformers caching ETA now v4.38 #1348

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions src/peft/peft_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1141,11 +1141,11 @@ def prepare_inputs_for_generation(self, *args, task_ids: torch.Tensor = None, **

# https://github.com/huggingface/transformers/pull/26681/ introduced new cache format
# for some architectures which requires a special fix for prompt tuning etc.
# TODO: starting with transformers 4.37, all architectures should support caching.
uses_transformers_4_37 = packaging.version.parse(transformers.__version__) >= packaging.version.parse("4.37.0")
# TODO: starting with transformers 4.38, all architectures should support caching.
uses_transformers_4_38 = packaging.version.parse(transformers.__version__) >= packaging.version.parse("4.38.0")
uses_transformers_4_36 = packaging.version.parse(transformers.__version__) >= packaging.version.parse("4.36.0")
transformers_new_cache_archs = ["llama", "mistral", "persimmon", "phi"]
uses_cache = uses_transformers_4_37 or (
uses_cache = uses_transformers_4_38 or (
uses_transformers_4_36 and self.base_model.config.model_type in transformers_new_cache_archs
)

Expand Down
Loading