IEEE

SLVR: Super-Light Visual Reconstruction via Blueprint Controllable Convolutions and Exploring Feature Diversity Representation

Abstract: Recently, improving the residual structure and designing efficient convolutions have become important branches of lightweight visual reconstruction model design. We have observed that the ...
IEEE

Do Visual Imaginations Improve Vision-and-Language Navigation Agents?

Abstract: Vision-and-Language Navigation (VLN) agents are tasked with navigating an unseen environment using natural language instructions. In this work, we study if visual representations of ...
Earth Times

The Quiet Collapse of Coral Reefs: Can Science Save the Ocean’s Rainforests?

Coral reefs, often called the rainforests of the sea, are among the most biologically diverse and ecologically important ecosystems on Earth. Despite covering less than 1% of the ocean floor, they ...
GitHub

Don’t Blind Your VLA: Aligning Visual Representations for OOD Generalization

To address the degradation of visual-language (VL) representations during VLA supervised fine-tuning (SFT), we introduce Visual Representation Alignment. During SFT, we pull a VLA’s visual tokens ...