Generative AI models in chemistry are increasingly popular in the research community, mainly, due to their interest for drug discovery applications. They generate virtual molecules with desired chemical and biological properties (more details in this blog post).
However, this flourishing literature still lacks a unified benchmark. Such benchmark would provide a common framework to evaluate and compare different generative models. Moreover, it would help to formulate best practices for this emerging industry of ‘AI molecule generators’: how much training data is needed, for how long the model should be trained, and so on.