Skip to content

Getting Started

This section describes how to install dependencies, configure the environment, and verify that ALUE is set up correctly.

ALUE has been tested primarily on Python 3.10 (with partial testing on 3.11).
Other versions may work, but are not officially supported.

1. Installation

ALUE supports two installation methods.
We recommend uv for reproducibility and speed.

Using uv (preferred)

# install dependencies into a managed virtual environment
uv sync

This creates a .venv directory automatically. No manual venv creation is needed.

Using pip (fallback)

# create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# install dependencies
pip install -r requirements.txt

2. Environment Configuration

ALUE uses a .env file for configuration. Create one from the example:

cp .env.example .env

Then edit .env to add your API keys and endpoints. At minimum, you'll need to configure:

For OpenAI (simplest): 

ALUE_ENDPOINT_TYPE=openai
ALUE_OPENAI_API_KEY=sk-...

For local Ollama: 

ALUE_ENDPOINT_TYPE=ollama
ALUE_ENDPOINT_URL=http://localhost:11434

For vLLM: 

ALUE_ENDPOINT_TYPE=vllm
ALUE_ENDPOINT_URL=http://localhost:8000/v1
HF_TOKEN=hf_...

Note: Different tasks require different configuration. RAG and Summarization tasks require additional LLM Judge and embedding settings. See the Configuration Reference for complete details on all available settings and backend-specific requirements.

3. Quick Verification

To check that ALUE is installed and functional:

# run the test suite
pytest tests

# or run a sample task (MCQA)
python -m scripts.mcqa inference \
  -i data/aviation_knowledge_exam/3_1_aviation_test.json \
  -o runs/mcqa \
  -m gpt-4o-mini \
  --task_type aviation_exam \
  --num_examples 3

If successful, predictions will be written to runs/mcqa_<timestamp>/predictions.json.

Next Steps