model="llama-2-13b-chat.ggmlv3.q4_K_M.bin"
n_threads="4"
n_gpu_layers="100"