StatePlay Separates Game Action Accuracy From Rule Consistency
A technical review compares StatePlay, H3-World, Incantation, and ABot-World across action representation, multi-entity control, continuous generation, and explicit state prediction. In the StatePlay paper, ReactiveGWM reports 100.0% attack accuracy versus StatePlay’s 95.0%, while StatePlay leads the GPT-5.5 visual-judge mechanism-consistency score at 78.3±0.5% versus 59.7±0.9%; these results were not independently retested.
The review shows why action classification and rule consistency should be evaluated separately: performing the requested move does not establish that the action was legal given the game state.
攻击动作做对了,游戏规则就对了吗?H3-World 与三条控制路线的分工
作者机制示意或未执行条件对照;不代表模型运行或真实游戏结果。
StatePlay 论文的一张表里,有一个比“谁的画面更好”更值得停下来的结果:同在 state-aware fine-tuning 分组,ReactiveGWM 的攻击准确率是 100.0%,StatePlay 是 95.0%;但在 GPT-5.5 视觉裁判给出的机制一致性上,两者分别是 59.7±0.9% 和 78.3±0.5%。攻击指标领先的模型,没有同时在这项规则评价上领先。
来源原始页面聚焦证据;固定版本与适用条件见图,作者报告未复测。
#StatePlay#H3-World#Incantation#ABot-World#ReactiveGWM#GPT-5.5