We are extremely grateful for the contributions of the Xuance Working Group, which have significantly lowered the threshold of reinforcement learning and greatly accelerated the research speed of reinforcement learning algorithms. Currently, we have a suggestion. In the representation network, Xuance currently supports Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN). In the most recent research, more types of representation networks have been widely adopted, such as Transformer, xLSTM, etc. If these can be incorporated into Xuance, it will make Xuance even more user-friendly.
transfomer
xLSTM
We are extremely grateful for the contributions of the Xuance Working Group, which have significantly lowered the threshold of reinforcement learning and greatly accelerated the research speed of reinforcement learning algorithms. Currently, we have a suggestion. In the representation network, Xuance currently supports Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN). In the most recent research, more types of representation networks have been widely adopted, such as Transformer, xLSTM, etc. If these can be incorporated into Xuance, it will make Xuance even more user-friendly.
transfomer
xLSTM