Search papers, labs, and topics across Lattice.
Affiliation:
13
0
15
15
Event-centric memory can boost long-video QA accuracy by over 3 points while slashing inference costs by nearly 64%.
Mechanist uncovers a surprising safety risk where unsafe traits can transfer across modalities, challenging assumptions about training data safety.
Context interference can significantly degrade the performance of search agents, but a novel context refiner shows how to enhance their reliability and efficiency dramatically.
MobileMem shifts the paradigm from static information retrieval to dynamic experiential learning, enabling AI agents to evolve alongside their users.
LabVLA achieves unprecedented success rates in executing complex laboratory protocols, outperforming all existing models in both familiar and novel settings.
Unsupervised skill discovery can boost data-analytic agent performance by over 30% without the need for labeled data.
Today's best data analysis agents forget crucial context as conversations drag on, dropping nearly 50 points in accuracy when reasoning through longer analyses.
Unlock interdisciplinary breakthroughs: SciAtlas provides a panoramic scientific evolution network of 43M papers, enabling AI agents to navigate complex logical connections and discover non-obvious insights.
LLMs are surprisingly bad at automating the creation of executable visual workflows from natural language, highlighting a significant gap in their ability to translate intent into reliable, deployable code.
LLM agents can learn more efficiently by leveraging a Skill Knowledge Base automatically constructed from prior experiences, enabling weaker agents to achieve stronger performance.
LLMs can slash token usage by 70% and boost reasoning accuracy by 14.8% in long-horizon tasks simply by learning when to remember and forget intermediate thoughts.
AI agents can now learn durable skills instead of constantly "reinventing the wheel," thanks to SkillNet's infrastructure for creating, evaluating, and connecting AI skills at scale.
LLMs can now evaluate research ideas like human experts, thanks to a new framework that grounds them in external knowledge and diverse perspectives.
