Metharme13B merged and quantized to to 4bit for use with llama.cpp and other ggml applications.