AWhat changed
Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.
Scientific scope
Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting. Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction