technique

Competitive Self-Play

techniqueactivecompetitive-self-play-ffe9e87d·1 events·first seen 28d ago

Aliases: Competitive Self-Play

Co-occurring entities

More like this (12)

self-play self-play reinforcement learning NPC-Playground Recursive Self-Improvement Self-Generated Replay Self-Instruct self-attention Evolution Strategies on-policy self-distillation gameplay video quality assurance Hindsight Experience Replay Self-Consistency / Majority Voting

Recent events (1)

5Openai Blog·28d ago·source ↗

Competitive Self-Play Enables Emergent Physical Skills in Simulated Agents

OpenAI demonstrates that competitive self-play allows simulated agents to spontaneously develop complex physical skills—tackling, ducking, faking, kicking, catching, and diving—without explicit environment design for those behaviors. The self-play dynamic automatically calibrates difficulty to the agent's current skill level. Combined with concurrent Dota 2 self-play results, OpenAI expresses confidence that self-play will be a foundational component of powerful AI systems.

Agent and Tool Ecosystem Alignment and RLHF Dota 2 Competitive Self-Play OpenAI