Quantized LLM on Hugging Face Spaces

Run a 4-bit quantized Vicuna-13B model on CPU using llama.cpp