Benchmarking study of deep generative models for inverse polymer design

This benchmark study evaluates deep learning-based molecular generative models on various polymer datasets. Selected models were further refined with reinforcement learning to generate hypothetical heat-resistant polymers.