MIRROR-Eval

MIRROR-Eval is an evaluation framework for MIRROR models.

Installation

Using conda:

# Clone the repository
git clone https://github.com/DRAGNLabs/MIRROR-Eval.git
cd MIRROR-Eval

# Create and activate conda environment
conda env create -f environment.yml
conda activate mirror-eval

pip install .

# If you're doing development:
pip install -e ".[dev]"

Building the Package

Build from source

To build the package as a distributable wheel:

# Install build tools
pip install build

# Build the package
python -m build

This will create distribution files in the dist/ directory:

• A wheel file (.whl) for binary distribution
• A source distribution (.tar.gz)

Installing the built package

pip install dist/mirror_eval-0.1.0-py3-none-any.whl

Usage

The primary entrypoint for MIRROR-Eval is the evaluate function:

from mirror_eval import evaluate

# Run the evaluation pipeline
results = evaluate()
print(results)

Development

Project Structure

MIRROR-Eval/
├── src/
│   └── mirror_eval/
│       ├── __init__.py
│       └── evaluate.py
├── tests/
├── pyproject.toml
├── environment.yml
├── requirements-dev.txt
├── README.md
└── LICENSE

Running Tests

pytest

With coverage:

pytest --cov=mirror_eval --cov-report=html

Code Formatting

Format code with Black:

black src/

Check code style with flake8:

flake8 src/

Type checking with mypy:

mypy src/

Other tips

Set the HuggingFace cache on your machine:

export HF_HOME="/path/to/cache/dir"

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.