vllm
[Bugfix] Cache added_vocab to avoid per-token overhead
#30351

Merged

[Bugfix] Cache added_vocab to avoid per-token overhead #30351

DarkLight1337 merged 2 commits into vllm-project:main from scratch-ml:fix/dsv3p2-tokenizer

fix: cache added_vocab to avoid per-token overhead

36fce5df

gemini-code-assist commented on 2025-12-09

return a shallow copy of self._added_vocab

b2f5e3bd

jeejeelee requested a review from

chaunceyjiang 169 days ago

chaunceyjiang approved these changes on 2025-12-10

chaunceyjiang assigned

chaunceyjiang 169 days ago

chaunceyjiang added ready

DarkLight1337 merged d007387a into main 169 days ago

scratch-ml deleted the fix/dsv3p2-tokenizer branch 169 days ago

Reviewers

chaunceyjiang

gemini-code-assist

Assignees

chaunceyjiang

Labels

ready

Milestone

No milestone