feat(halo): use qwen-35b-a3b

2026-05-07 16:58:03 +02:00 · 2026-05-07 16:58:03 +02:00 · bef528e26a
commit bef528e26a
parent 267c05b107
3 changed files with 62 additions and 3 deletions
--- a/systems/x86_64-linux/halo/llama-server.nix
+++ b/systems/x86_64-linux/halo/llama-server.nix
@ -35,8 +35,8 @@
        "--port 8000"
        "--no-mmap"
        "--n-gpu-layers 99"
-        "-hf unsloth/Qwen3.6-27B-GGUF:UD-Q8_K_XL"
-        "--alias qwen3.6-27b"
+        "-hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q8_K_XL"
+        "--alias qwen3.6-35b-a3b"
        "--threads 8"
        "--ubatch-size 256"
        "-ctk bf16 -ctv bf16"