Search papers, labs, and topics across Lattice.
University of Technology Nuremberg
4
0
5
The Physics-IQ Verified benchmark reveals that over half of the evaluated samples can be significantly refined, leading to notable shifts in model performance rankings.
Standard distillation methods falter in visual autoregressive models, but VarKD redefines the approach, achieving superior performance by intelligently managing teacher supervision.
SAM, designed for instance segmentation, can be surprisingly effective for semantic segmentation with weak supervision when adapted with techniques like skeleton-based prompting and iterative pseudo-label refinement.
Forget salient cues – now you can *steer* visual representations in ViTs with language, focusing on any object you want without hurting overall performance.
