Add mini/mid/gaming hardware profiles and always-on full stack

Profiles select CPU vs Vulkan and Bonsai-27B vs Qwen3.5-9B Abliterated
Q4_K_M. OpenHands is no longer optional; compose deploys every service.
download-models.sh fetches both GGUF families as the profile requests.
This commit is contained in:
tim 2026-07-22 19:00:59 -07:00
parent 3bf1726a99
commit a729ff14e9
11 changed files with 333 additions and 171 deletions

24
profiles/gaming.env Normal file
View file

@ -0,0 +1,24 @@
# Profile: gaming — ~16GB AMD GPU + 64GB system RAM
# Vulkan; Ternary-Bonsai 27B primary, Qwen 9B also downloaded for light loads.
STACK_PROFILE=gaming
BONSAI_BACKEND=vulkan
BONSAI_NGL=99
BONSAI_CTX=32768
BONSAI_KV4=0
BONSAI_THREADS=16
BONSAI_TEMP=0.7
BONSAI_TOP_P=0.95
BONSAI_TOP_K=20
LLM_ID=bonsai27
BONSAI_HF_REPO=prism-ml/Ternary-Bonsai-27B-gguf
BONSAI_MAIN_FILE=Ternary-Bonsai-27B-Q2_0.gguf
BONSAI_MODEL_PATH=/models/Ternary-Bonsai-27B-Q2_0.gguf
OPENWEBUI_DEFAULT_MODEL=openai/Ternary-Bonsai-27B-Q2_0.gguf
OPENHANDS_MODEL=openai/Ternary-Bonsai-27B-Q2_0.gguf
DOWNLOAD_MODEL=true
DOWNLOAD_BONSAI27=true
DOWNLOAD_QWEN=true
DOWNLOAD_VISION=true