harald/nixcfg
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

feat(halo): use unsloth/Qwen3.6-27B-GGUF:UD-Q8_K_XL

Browse source
This commit is contained in:
Harald Hoyer 2026-05-06 13:02:20 +02:00
parent c9c7f6916a
commit a95417da8b
2 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

6
systems/x86_64-linux/halo/llama-server.nix
View file

@ -35,11 +35,11 @@
"--port 8000"
"--no-mmap"
"--n-gpu-layers 99"
"-hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q8_K_XL"
"--alias qwen3.6-35b-a3b"
"-hf unsloth/Qwen3.6-27B-GGUF:UD-Q8_K_XL"
"--alias qwen3.6-27b"
"--threads 8"
"--ubatch-size 256"
"-ctk q8_0 -ctv q8_0"
"-ctk bf16 -ctv bf16"
"--temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00"
"--no-context-shift"
''--chat-template-kwargs '{"preserve_thinking": true}' ''