Evan Reichard 81ffe67cce refactor(llama-swap): replace --parallel with -np and add -kvu flag ...

Switch llama-server invocations from --parallel to -np with -kvu
(kv-cache unified) across Qwen3.6 model configs. Also reduce
context for qwen3.6-27b-cuda0 from 150k to 140k.

2026-05-19 06:22:08 -04:00

29 KiB

Raw Blame History

View Raw