llama.cpp is gaining Falcon support via GGUF: https://github.com/ggerganov/llama.cpp/pull/2717