Llama2 7B and 13B GGML are up and work with existing llama.cpp no changes needed! The 70B does require a llama.cpp change, but I'm sure it won't take long.