diff --git a/hosts/architect/llm.nix b/hosts/architect/llm.nix index 25e17dc..3ea11d3 100644 --- a/hosts/architect/llm.nix +++ b/hosts/architect/llm.nix @@ -21,6 +21,7 @@ in environmentVariables = { OLLAMA_FLASH_ATTENTION = "1"; OLLAMA_NUM_PARALLEL = "2"; + OLLAMA_KV_CACHE_TYPE = "q8_0"; }; };