NetHack

NetHack is a complex game used as a benchmark to evaluate the performance and scalability of hierarchical reinforcement learning algorithms, specifically in the context of training agents with large amounts of experience.

Papers using NetHack (4)

Scalable Option Learning in High-Throughput Environments2025Code Driven Planning with Domain-Adaptive Critic2025PufferLib: Making Reinforcement Learning Libraries and Environments Play Nice2024Syllabus: Portable Curricula for Reinforcement Learning Agents2024