NetHack Learning

NetHack learning focuses on developing artificial intelligence agents capable of mastering the complex, procedurally generated roguelike game NetHack, a significant challenge for current AI techniques. Research currently explores various approaches, including large language models for high-level planning, modular frameworks combining symbolic and neural methods, and offline reinforcement learning leveraging large datasets of human and agent gameplay. These efforts aim to advance understanding of AI capabilities in handling highly dynamic, open-ended environments and contribute to broader progress in reinforcement learning and artificial general intelligence.