Metadata-Version: 2.5
Name: astro-eval
Version: 0.1.0
Summary: A reusable evaluation framework for domain-specific language models.
Requires-Python: >=3.12
Requires-Dist: numpy>=2.5.2
Requires-Dist: tokenizers>=0.23.1
Requires-Dist: torch>=2.13.0
Description-Content-Type: text/markdown

# AstroGPT

A project for training and evaluating astronomical language models.

## Project Structure

```text
├── src/
│   ├── data/
│   ├── tokenizer/
│   ├── model/
│   ├── training/
│   └── astro_eval/    ← Reusable Evaluation Framework (astro-eval)
├── datasets/
├── configs/
├── checkpoints/
├── notebooks/
├── tests/
├── requirements.txt
└── README.md
```

## LLM Evaluation Package (`astro-eval`)

The repository includes a decoupled model evaluation and benchmarking library located under `src/astro_eval/`.

### Setup
Ensure dependencies are built:
```bash
uv build --clear
pip install dist/astro_eval-0.1.0-py3-none-any.whl
```

For full usage details, command-line arguments, and Python developer API documentation, see the [Astro-Eval README](src/astro_eval/README.md).
