Search papers, labs, and topics across Lattice.
8
1
11
8
StudentSim outperforms existing models, achieving a remarkable behavioral fidelity of 0.51 and guidance responsiveness of 0.91 in chess, setting a new standard for AI tutors.
Current video generation models struggle with visual reasoning, with the best achieving only 51% accuracy on a new benchmark designed to probe their capabilities.
Importance-Aware Sampling (IAS) reveals that not all patches in VIS-IR data are created equal, leading to substantial performance gains in multi-sensor perception tasks.
General-purpose computer vision MLLMs can outperform specialized remote sensing models on key tasks, challenging the notion of domain-specific superiority.
Removing gold answer strings from rewritten contexts can cause F1 scores to plummet by up to 64 points, underscoring their critical role in retrieval-augmented QA performance.
MiniMax-M2 proves that massive parameter counts don't always translate to better agentic performance; strategic activation of a smaller subset can unlock frontier-level intelligence.
Overcoming discontinuities in skeleton detection, this work leverages "lighthouse-guided" reconnection to substantially improve skeleton connectivity and structural integrity.
GUI agents can achieve significantly stronger task-solving capabilities through carefully designed post-training and data curation, without relying on costly online data collection.
