Dota 2
A multiplayer online battle arena game developed by Valve. Its complex strategy space was used by OpenAI as an AI research environment, where OpenAI Five defeated top human players.
Core Facts
Timeline (last 90 days)
Multi-agent reinforcement learning has achieved success in game environments such as StarCraft and Dota 2, but these successes often rely on carefully designed reward functions and restricted action spaces.
The success of multi-agent AI in StarCraft and Dota 2 relies heavily on the centralized training with decentralized execution (CTDE) framework and carefully designed reward shaping.
Representative early work on world models includes DeepMind's Dreamer series and the environment model used by OpenAI in Dota 2.
OpenAI used automatic domain randomization technology when training the Dota 2 agent.
OpenAI Five在Dota 2中累计了超过45000年的游戏经验
All Facts (5)
OpenAI used automatic domain randomization technology when training the Dota 2 agent.
65%UnverifiedMulti-agent reinforcement learning has achieved success in game environments such as StarCraft and Dota 2, but these successes often rely on carefully designed reward functions and restricted action spaces.
50%UnverifiedThe success of multi-agent AI in StarCraft and Dota 2 relies heavily on the centralized training with decentralized execution (CTDE) framework and carefully designed reward shaping.
50%UnverifiedRepresentative early work on world models includes DeepMind's Dreamer series and the environment model used by OpenAI in Dota 2.
50%UnverifiedOpenAI Five在Dota 2中累计了超过45000年的游戏经验
50%