Anyone want to run LLaVA (multimodal GPT-4V-type AI) on their PC or laptop? I bolted an API and web server onto llama.cpp's implementation: https://github.com/trzy/llava-cpp-server