• Install
• Edit on GitHub

---

Install

Requirements

DataMetaMap requires Python 3.10+ and the following packages:

• PyTorch ≥ 2.0
• PyTorch Lightning — for Dataset2Vec training
• NumPy
• scikit-learn
• POT (Python Optimal Transport) — for WassersteinEmbedder
• pydantic ≥ 2.0 — for Dataset2Vec config
• pandas — for Dataset2Vec data loading
• tqdm

Install from source (recommended)

Clone the repository and install in editable mode:

git clone https://github.com/intsystems/DataMetaMap
cd DataMetaMap
pip install -e .

Install via pip (GitHub)

pip install git+https://github.com/intsystems/DataMetaMap

Install dependencies manually

If you prefer to install dependencies yourself:

pip install torch torchvision numpy scikit-learn POT tqdm pytorch-lightning pydantic pandas

Verify installation

import data_meta_map
print(data_meta_map.__version__)
# 0.0.1