technique

self-play reinforcement learning

techniqueactiveself-play-reinforcement-learning-10ea19d7·1 events·first seen 28d ago

Aliases: self-play reinforcement learning

Co-occurring entities

More like this (12)

self-play Competitive Self-Play Reinforcement Learning Constrained Reinforcement Learning Imitation Learning Goal-Conditioned Reinforcement Learning Q-learning Reinforcement Learning from Human Feedback Hierarchical Reinforcement Learning shielded reinforcement learning Soft Q-Learning General Preference Reinforcement Learning

Recent events (1)

6Openai Blog·28d ago·source ↗

OpenAI Bot Defeats Top Dota 2 Professionals at 1v1

OpenAI developed a bot that defeats world-class professional players in 1v1 Dota 2 matches under standard tournament rules. The system learned entirely through self-play without imitation learning or tree search. This was presented as a milestone toward AI systems that can achieve well-defined goals in complex, real-world environments involving humans.

AI Safety Research Agent and Tool Ecosystem self-play reinforcement learning OpenAI Dota 2 Bot Dota 2 +1 more