Well-trained models and generative outcomes for the paper "Benchmarking Study of Deep Generative Models for Inverse Polymer Design." https://pubs.rsc.org/en/content/articlehtml/2025/dd/d4dd00395k

Well-trained models

Generative models

AAE, VAE, CharRNN, ORGAN, and REINVENT models can be found in the /MOSES folder, and GraphINVENT models can be found at: https://zenodo.org/records/12734266

Reinforcement learning

Well-trained models for reinforcement learning can be found at https://zenodo.org/records/12728016

Training your own polymer generative models

Generative models

For the application of AAE, VAE, CharRNN, and ORGAN models, please refer to https://github.com/molecularsets/moses?tab=readme-ov-file. The training scripts and related commands can be found in the /MOSES folder.

For the application of the REINVENT model, please refer to https://github.com/MolecularAI/Reinvent. For the application of the GraphINVENT model, please refer to https://github.com/MolecularAI/GraphINVENT.

Reinforcement learning

For content on reinforcement learning for CharRNN, please refer to https://github.com/aspuru-guzik-group/Tartarus. For the application of the reinforcement learning GraphINVENT model, please refer to https://github.com/olsson-group/RL-GraphINVENT.

Generation results

All generation results can be found at https://zenodo.org/records/12636925 Generation results for reinforcement learning can be found at https://zenodo.org/records/12728016