Quantization Trap in Multi-Hop Reasoning: Breaking Scaling Laws
Discover how the quantization trap disrupts linear scaling laws, increasing energy use and reducing accuracy in multi-hop reasoning AI models.
E-mem: Enhancing LLM Memory with Multi-Agent Episodic Context
Discover E-mem, a multi-agent framework improving LLM agent memory via episodic context reconstruction for superior reasoning and efficiency.
Evaluating Legal Reasoning with LEGIT Issue Tree Rubrics
Discover how LEGIT dataset and legal issue tree rubrics improve AI legal reasoning evaluation with expert annotations and structured argument trees.
Hybrid AI Approach for Healthcare Timetabling 2024
Discover Team Twente's hybrid AI solution for the 2024 Healthcare Timetabling Competition, blending optimization and simulated annealing techniques.
Mastering Liar’s Poker with AI: Outbluffing Elite Humans
Discover how AI agent Solly uses self-play and reinforcement learning to master Liar's Poker, outperforming top human players with innovative bluffing stra...