Search papers, labs, and topics across Lattice.
2
0
5
EffectLearner achieves unprecedented video object removal quality by effectively reasoning about complex object-induced effects in dynamic scenes.
Current multimodal browsing agents are surprisingly bad at using visual information on webpages, with even top models scoring below 50% accuracy on a new visual-native search benchmark.
