Search papers, labs, and topics across Lattice.
6
2
10
2
Visual Para-Thinker++ achieves remarkable improvements in visual reasoning accuracy by leveraging a multi-agent architecture that minimizes hallucination risks through parallel processing and effective output reconciliation.
SG-OPD achieves remarkable gains in mathematical reasoning by effectively aligning student and teacher models through innovative trust signaling techniques.
Prefix failure in on-policy distillation can be effectively mitigated by correcting problematic prefixes, leading to significant improvements in reasoning coverage and accuracy.
TVIR-Agent reveals that integrating visual elements into report generation can dramatically improve the quality and reliability of analytical outputs.
Achieve spatially precise control in FPS world models by injecting actions locally, without segmentation labels, enabling zero-shot generalization across games.
By jointly reinforcing informative visual tokens and suppressing irrelevant ones, DuCAR significantly reduces hallucinations in LVLMs, outperforming prior single-modality focused approaches.
