Hyena for Efficient Image Super Resolution
Develop a long convolution (Hyena) based architecture to leverage its unique feature extraction capabilities to reconstruct high-frequency details.
Sep 1, 2024
ViLMorph, a Vision LSTM Framework for 3D Medical Image Registration
Develop an efficient bi-directional vision xLSTM block for capturing long range pixel dependencies of medical images.
Aug 1, 2024
Large Language Model for Long Genome Sequence
Develop a genome LLM model based on BERT architecture and adopt masked language modeling as pretraining strategy.
Jun 1, 2024