Dungeons and Data: A Large-Scale NetHack Dataset
This provides a large-scale dataset for researchers working on challenging sequential decision-making problems like NetHack, though it is incremental as it builds on existing environment-based research.
The authors tackled the scarcity of open-sourced datasets for sequential decision-making by presenting the NetHack Learning Dataset (NLD), which includes 10 billion state transitions from human trajectories and 3 billion from a symbolic bot, and they found that existing algorithms require significant advances to leverage such large-scale data effectively.
Recent breakthroughs in the development of agents to solve challenging sequential decision making problems such as Go, StarCraft, or DOTA, have relied on both simulated environments and large-scale datasets. However, progress on this research has been hindered by the scarcity of open-sourced datasets and the prohibitive computational cost to work with them. Here we present the NetHack Learning Dataset (NLD), a large and highly-scalable dataset of trajectories from the popular game of NetHack, which is both extremely challenging for current methods and very fast to run. NLD consists of three parts: 10 billion state transitions from 1.5 million human trajectories collected on the NAO public NetHack server from 2009 to 2020; 3 billion state-action-score transitions from 100,000 trajectories collected from the symbolic bot winner of the NetHack Challenge 2021; and, accompanying code for users to record, load and stream any collection of such trajectories in a highly compressed form. We evaluate a wide range of existing algorithms including online and offline RL, as well as learning from demonstrations, showing that significant research advances are needed to fully leverage large-scale datasets for challenging sequential decision making tasks.