vllm
[Hybrid] Simplify accepted token counting in spec decode for hybrid models
#38372

Merged

[Hybrid] Simplify accepted token counting in spec decode for hybrid models #38372

njhill merged 2 commits into vllm-project:main from fuscof-ibm:optimize_update_states

gemini-code-assist commented on 2026-03-27

mergify added v1

fuscof-ibm marked this pull request as ready for review 42 days ago

fuscof-ibm requested a review from

njhill 42 days ago

claude commented on 2026-03-28

tdoublep approved these changes on 2026-04-10

tdoublep added ready

[Hybrid] Simplify accepted token counting in spec decode for hybrid m…

fa04e189

fuscof-ibm force pushed to fa04e189 26 days ago

Merge branch 'main' into optimize_update_states

e48d8cc9

njhill approved these changes on 2026-04-14

njhill merged 507df79a into main 25 days ago

Reviewers

njhill

tdoublep

claude

gemini-code-assist

Assignees

No one assigned

Labels

ready v1

Milestone

No milestone