A framework to study how neural models generalize in compositional, relational domains.
A proposed Transfer Learning taxonomy for Visual Reasoning and a new State-of-the-Art model.
Learning about the Attention Mechanism and the Transformer Model.
Enabling reproducible Machine Learning research.
A simplified variant of the MAC model that achieves comparable accuracy, while training faster.
The paper presents a PyTorch-based, open-source framework for Machine Learning.