# GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation

This repo contains a reference implementation for GraphAF as described in the paper:

> GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation </br>
> [Chence Shi](https://chenceshi.com/)\*, [Minkai Xu](https://minkaixu.com/)\*, Zhaocheng Zhu, Weinan Zhang, Ming Zhang, [Jian Tang](https://jian-tang.com/) </br>
> International Conference on Learning Representations (ICLR), 2020. </br>
> Paper: [https://arxiv.org/abs/2001.09382](https://arxiv.org/abs/2001.09382) </br>

Code is available at [google drive](https://drive.google.com/open?id=1FmYWcT8jDrwZlzPbmMpRhulb9OKTDWJL) now.

**Update: We implement the GraphAF in our [TorchDrug](https://torchdrug.ai/docs/tutorials/generation.html) platform.**