README.md

---
license: mit
datasets:
- scikit-fingerprints/MoleculeNet_ESOL
- scikit-fingerprints/MoleculeNet_FreeSolv
- scikit-fingerprints/MoleculeNet_Lipophilicity
- ibm-research/otter_uniprot_bindingdb_chembl
language:
- en
pipeline_tag: graph-ml
tags:
  - arxiv:2508.07807
model-index:
- name: PactNet
  results: []
  papers:
    - title: "Topological Feature Compression for Molecular Graph Neural Networks"
    - url: https://huggingface.co/papers/2508.07807
---
# PACT-Net (NeurIPS 2025)

```python
python main.py --model polyatomic --rep polyatomic --dataset esol
```

```
@inproceedings{khorana2025pactnet,
  title     = {Topological Feature Compression for Molecular Graph Neural Networks},
  author    = {Rahul Khorana},
  booktitle = {Thirty-ninth Conference on Neural Information Processing Systems (NeurIPS 2025) Workshop on AI for Science},
  year      = {2025},
  note      = {Poster presentation}
}
```