#!/usr/bin/env bash # Launch the Rune Goblin game. # ./start.sh → run models on CPU # ./start.sh --gpu → offload all GGUF layers to the GPU # Set RG_USE_MODEL=0 in the environment to play purely on the rule engine. set -euo pipefail # CPU by default; --gpu flips to full offload (-1 = all layers on GPU). GPU_LAYERS=0 if [[ "${1:-}" == "--gpu" ]]; then GPU_LAYERS=-1 echo "[start] GPU mode: offloading all GGUF layers to the GPU" else echo "[start] CPU mode (pass --gpu to use the GPU)" fi RG_USE_MODEL=1 \ RG_VISION_MODEL=models/goblinV1-gguf/gguf/rune-goblin-v46-Q4_K_M.gguf \ RG_VISION_MMPROJ=models/goblinV1-gguf/gguf/rune-goblin-v46-mmproj-f16.gguf \ RG_USE_DIALOGUE_MODEL=1 \ RG_DIALOGUE_MODEL=models/MiniCPM-V-4.6-gguf/MiniCPM-V-4_6-Q4_K_M.gguf \ RG_GGUF_GPU_LAYERS="${GPU_LAYERS}" \ uv run --extra gguf python app/rpg_app.py # → http://localhost:7862 (set RG_USE_MODEL=0 to play purely on the rule engine)