config.yml

---
hub:
  repo_id: umarigan/Hermes-7B-TR-Q5_K_M-GGUF
  filename: hermes-7b-tr.Q5_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