dora/tests/llm/llama-cpp-python.yaml

nodes:
  - id: pyarrow-sender
    build: pip install -e ../../node-hub/pyarrow-sender
    path: pyarrow-sender
    outputs:
      - data
    env:
      DATA: "Please only output: This is a test"

  - id: dora-llama-cpp-python
    build: pip install -e ../../node-hub/dora-llama-cpp-python
    path: dora-llama-cpp-python
    inputs:
      text: pyarrow-sender/data
    outputs:
      - text
    env:
      MODEL_NAME_OR_PATH: "TheBloke/Llama-2-7B-Chat-GGUF"
      MODEL_FILE_PATTERN: "*Q4_K_M.gguf"
      SYSTEM_PROMPT: "You're a very succinct AI assistant with short answers."
      ACTIVATION_WORDS: "what how who where you"
      MAX_TOKENS: "512"
      N_GPU_LAYERS: "35" # Enable GPU acceleration
      N_THREADS: "4" # CPU threads
      CONTEXT_SIZE: "4096" # Maximum context window

  - id: pyarrow-assert
    build: pip install -e ../../node-hub/pyarrow-assert
    path: pyarrow-assert
    inputs:
      data: dora-llama-cpp-python/text
    env:
      DATA: "This is a test"