---
hub:
  repo_id: jkeisling/laura-openllama13b-600bt-ggml
  filename: laura-openllama13b-600bt-q6_K.bin
llama_cpp:
  n_ctx: 2048
  n_gpu_layers: 40  # llama 13b has 40 layers
chat:
  stop:
    - "Human:"
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