transformers

91ba4c9a - fix(text-generation): use token-level slicing for return_full_text=Fa… (#45860)

fix(text-generation): use token-level slicing for return_full_text=Fa… (#45860) * fix(text-generation): use token-level slicing for return_full_text=False with chat templates * fix: remove unused variables and imports caught by ruff * fix: move regression tests to correct pipeline test file