llama : make tensor-split regex patterns static to avoid per-token recompilation - #24710
Merged
JohannesGaessler merged 1 commit intoJul 10, 2026
background
wait
wait-all
cancel
parallel
Loading