1 Introduction Figure 1: MineDojo is a novel framework for developing open-ended, generally capable agents that can learn and adapt continually to new goals. ![]() We open-source the simulation suite and knowledge bases ( ) to promote research towards the goal of generally capable embodied agents. Our agent is able to solve a variety of open-ended tasks specified in free-form language without any manually designed dense shaping reward. ![]() Using MineDojo’s data, we propose a novel agent learning algorithm that leverages large pre-trained video-language models as a learned reward function. We introduce MineDojo, a new framework built on the popular Minecraft game that features a simulation suite with thousands of diverse open-ended tasks and an internet-scale knowledge base with Minecraft videos, tutorials, wiki pages, and forum discussions. Inspired by how humans continually learn and adapt in the open world, we advocate a trinity of ingredients for building generalist agents: 1) an environment that supports a multitude of tasks and goals, 2) a large-scale database of multimodal knowledge, and 3) a flexible and scalable agent architecture. ![]() However, they typically learn tabula rasa in isolated environments with limited and manually conceived objectives, thus failing to generalize across a wide spectrum of tasks and capabilities. ![]() Autonomous agents have made great strides in specialist domains like Atari games and Go.
0 Comments
Leave a Reply. |