The Predictive--Optimization Gap in GuanDan Self-Play

概要

Outcome-predictive signals are often treated as promising optimization targets in reinforcement learning, but predictive validity and optimization utility are different claims. We expose this distinction in GuanDan, a multi-round imperfect-information game in which the implication of a round outcome depends on the public match state. Held-out analysis shows that state--gain interactions provide incremental predictive value beyond a constant signed-gain effect, establishing that match progress is predictively meaningful. We then use changes in a frozen match-win predictor, $ΔΦ$, as a controlled round-level supervision intervention in a matched DanZero-style Monte Carlo self-play protocol. The intervention can succeed in an individual lineage: a selected progress-augmented checkpoint substantially outperforms its matched Baseline on a locked evaluation. Yet the method-level result reverses this impression: none of five preregistered replication seeds is positive, the pooled ten-seed interval includes zero, and across-iteration effects repeatedly change sign without a common late-stage advantage. These results identify a predictive--optimization gap: under the studied learner, a signal can describe outcome-relevant progress while failing to provide a reliably beneficial optimization direction.

参考リンク

• Paper ↗

リンク先のライセンス・再現性・データ条件をご確認ください。自動分類や実装の目安は保証ではありません。

実装を始めるプロンプト

参考リンクを確認したうえで、次の内容をコピーして利用できます。

# 論文・記事
The Predictive--Optimization Gap in GuanDan Self-Play

# 概要
Outcome-predictive signals are often treated as promising optimization targets in reinforcement learning, but predictive validity and optimization utility are different claims. We expose this…

# 原典
https://arxiv.org/abs/2610.29411

上記の手法を実装してください。実行前に依存関係とライセンスを確認してください。