PR #30380 `Llama` family, fix `use_cache=False` generation

`Llama` family, fix `use_cache=False` generation #30380

ArthurZucker merged 4 commits into main from fix-llama-no-cache

nit to make sure cache positions are not sliced

6748f646

fix other models

2d553394

nit

ea0e2205

style

deab6cff

ArthurZucker requested a review from

gante 1 year ago

gante approved these changes on 2024-04-22

ArthurZucker merged 2d92db84 into main 1 year ago

ArthurZucker deleted the fix-llama-no-cache branch 1 year ago

Reviewers

gante

Assignees

No one assigned

Labels

None yet

Milestone

No milestone