SocialGrid: Benchmark for Social Reasoning in AI Agents
Discover SocialGrid, a benchmark evaluating planning and social reasoning in embodied multi-agent AI systems to improve LLM performance.
MEDLEY-BENCH: Evaluating AI Metacognition Beyond Scale
Discover MEDLEY-BENCH, a benchmark assessing AI metacognition, revealing scale boosts evaluation but not control in AI reasoning and self-revision.
ReactBench: Benchmarking Structural Reasoning in MLLMs
ReactBench evaluates MLLMs' structural reasoning on complex chemical reaction diagrams, revealing key performance gaps in topological understanding.
Weak-Link Optimization for Multi-Agent Collaboration
Discover WORC, a framework boosting multi-agent reasoning by optimizing weak links to improve stability and accuracy in AI collaborations.
Graph and LLM Integration for Enhanced AI Reasoning
Explore how integrating graphs with large language models improves AI reasoning, retrieval, and decision-making across diverse domains.