torch-molecule documentationΒΆ
Welcome to the torch-molecule. torch-molecule is an actively developed package designed to facilitate molecular discovery through deep learning.
It offers a user-friendly interface similar to sklearn, and includes model checkpoints for efficient deployment and benchmarking across various molecular tasks. The package currently focuses on three main components:
Predictors: Molecular property prediction models that make predictions (regression or classification) based on molecular graphs.
Generators: Molecular graph generators that generate new molecules.
Encoders: Molecular graph encoders that convert molecular graphs into a vector representation.