标签
本文表明，PPFG（PRM-Pruned Fragment Grafting，一种成本极低的推理时跨轨迹迁移技术）在三个推理语言模型和六个基准测试上与独立并行的 CoT 在统计上无法区分；这种失效性可归因于注入定位不佳以及由选择主导的剪枝动力学。
This paper evaluates whether structure-preserving de-identification via surrogate substitution maintains PHI detectability across multiple detectors, using equivalence testing on 57k paired spans across 7 languages and 11 detectors.