FROM llama3.2:3b
ADAPTER adapter_model.safetensors