Search papers, labs, and topics across Lattice.
Northeastern University, 2 Independent Researcher
7
0
10
2
Transforming failures into focused training tasks boosts tool-using language model performance by over 8% on key benchmarks.
Reasoning VLMs falter under semantic distractions, often mistaking irrelevant cues for evidence, which can lead to incorrect answers.
Capturing multi-hop dependencies in agent communication can lead to significant performance gains and lower costs in LLM-based systems.
Achieve state-of-the-art colo-segment recognition by combining edge-guided spatial feature extraction with a novel temporal attention mechanism, outperforming existing methods by a large margin.
LLM agents often say one thing, believe another, and do something completely different, especially when interacting with other agents.
Achieve realistic video object insertion by lifting 2D images into multi-view representations, ensuring consistent appearance and handling occlusions in dynamic scenes.
Task-oriented dialogue agents can now learn to balance user satisfaction and operational costs, thanks to a new RL framework that optimizes for both.
