config.yml

---
hub:
  repo_id: TheBloke/vicuna-13b-v1.3.0-GGML
  filename: vicuna-13b-v1.3.0.ggmlv3.q2_K.bin
  # repo_id: TheBloke/Wizard-Vicuna-13B-Uncensored-GGML
  # filename: Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_1.bin
llama_cpp:
  n_ctx: 2048
  # 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