Skip to content

fix: allow mixed turbo/q8_0 KV cache types in CUDA flash-attention selection - #3

Merged
thecodacus merged 2 commits into
perffrom
fable5/turbo-fattn-vec-routing
Jul 18, 2026
Merged

fix: allow mixed turbo/q8_0 KV cache types in CUDA flash-attention selection#3
thecodacus merged 2 commits into
perffrom
fable5/turbo-fattn-vec-routing

feat: accept turbo2/3/4 cache type names in llama-bench

781726d
Select commit
Loading
Failed to load commit list.

Select a check to view from the sidebar