config.yml

---
hub:
  repo_id: Nekochu/Llama-2-13B-fp16-french
  filename: llama-2-13b-chat-FR-Q4_K_M.gguf
llama_cpp:
  n_ctx: 4096
  # n_gpu_layers: 40  # llama 13b has 40 layers
chat:
  stop:
    - "</s>"
    - "<unk>"
    - "### USER:"
    - "USER:"
queue:
  max_size: 16
  concurrency_count: 1  # leave this at 1, llama-cpp-python doesn't handle concurrent requests and will crash the entire app