vllm
fix: revert cast to cpu in `MsgpackEncoder._encode_tensor` to avoid hidden performance regressions
#25738

Merged

fix: revert cast to cpu in `MsgpackEncoder._encode_tensor` to avoid hidden performance regressions #25738

vllm-bot merged 1 commit into vllm-project:main from protopia-ai:do-not-cast-to-cpu-in-serialization

fix: revert cast to cpu in MsgpackEncoder._encode_tensor to avoid hid…

781a8abf

qthequartermasterman requested a review from

WoosukKwon 244 days ago

qthequartermasterman requested a review from

robertgshaw2-redhat 244 days ago

qthequartermasterman requested a review from

njhill 244 days ago

qthequartermasterman requested a review from

ywang96 244 days ago

qthequartermasterman requested a review from

comaniac 244 days ago

qthequartermasterman requested a review from

alexm-redhat 244 days ago

mergify added v1

gemini-code-assist commented on 2025-09-26

DarkLight1337 approved these changes on 2025-09-26

DarkLight1337 enabled auto-merge (squash) 244 days ago

DarkLight1337 added this to the v0.11.0 milestone 244 days ago

github-actions added ready

vllm-bot merged e84e0735 into main 244 days ago

Reviewers

DarkLight1337

gemini-code-assist

WoosukKwon

robertgshaw2-redhat

njhill

ywang96

comaniac

alexm-redhat

Assignees

No one assigned

Labels

ready v1

Milestone

v0.11.0