Tag
该论文提出并评估了一类称为事件图基质的因果推理世界模型，通过确定性重放在类型化RDF事件日志上进行反事实查询，在多个基准上优于基线模型，同时保证了可检查性和可重放一致性。
The LOOP Skill Engine achieves 99% success and 99% token reduction for periodic AI agent tasks by recording a single LLM-driven execution and replaying it deterministically via a parameterized, branch-free skill, eliminating stochastic failures and high costs.