Search papers, labs, and topics across Lattice.
3
0
6
4
Forget brute-force context windows: a small vision-language model can compress hour-long videos below theoretical limits by intelligently prioritizing relevant content.
Rethinking IRSTD as a centroid regression problem with single-point supervision achieves competitive detection performance with significantly reduced computational cost, challenging the dominance of pixel-level segmentation approaches.
Scale up offline policy training for diffusion LLMs without breaking the bank: dTRPO slashes trajectory computation costs while boosting performance up to 9.6% on STEM tasks.
