Reducing Cognitive Bias in RLHF with Adaptive Rationality
Explore how dynamic rationality adjustments mitigate cognitive bias in reinforcement learning from human feedback for more reliable AI models.
Improving AI Agent Tool Use with Mechanistic Interpretability
Discover how mechanistic interpretability enhances AI agent tool use reliability and safety in complex, long-horizon workflows.
LLM Performance on Long-Chain Reasoning: Equivalence Class Study
Explore how Large Language Models tackle long-chain reasoning tasks in the Equivalence Class Problem, revealing strengths and challenges in AI reasoning.
Agentick: Benchmark for Sequential Decision-Making AI Agents
Agentick provides a unified benchmark to evaluate AI agents in sequential decision-making with diverse tasks and multi-modal support.
AGWM: Advanced World Models for Dynamic AI Environments
Discover AGWM, a novel framework improving AI agents' decision-making by tracking action prerequisites in dynamic environments with reduced prediction erro...