Research Projects

We work toward Model Compression

We work toward X-Transformer. As we all know, transformer is SOTA of nlp tasks and generally replace CNN in vision tasks, but the theory about transformers develops a bit slowly, we try to explain and design new attention mechanism by leveraging the rich literatures of differential equation

DQN与改进算法的原理，并将其应用在强化学习平台gym上的LunarLander实验

Formal representation and solving for Euclidean plane geometry problems.