Per-token counterfactual credit estimation asks which token in a language-model rollout caused the final answer to be right or wrong: cut the transcript at a pivot, substitute an alternative token, replay continuations, and compare outcomes. Published methods re-feed the transcript prefix as a fresh prompt, assuming this reproduces the state the model passed through during generation. We measure what that assumption costs on a stock inference engine, with a three-pass design: continuations resumed from the verified decode-time KV state, an identical second exact pass (a replica noise floor), and a re-feed pass. Across six configurations and three models (including a GRPO-trained checkpoint), at low-margin decision tokens, re-feeding changes the credit estimate at rates 14-28 percentage points above the replica floor (7-21pp under a treatment-independent conditioning; problem-clustered t = 2.9-6.4). Most changes are zero-boundary crossings of the quantized estimator rather than polarity reversals, and the perturbation is consistent with mean-zero, so averaged quantities are largely safe; but selection is not: a critical-token set chosen by thresholding $|\hat{A}_t|$ under re-feed overlaps the exact-resume selection at Jaccard 0.34-0.90, versus a 0.63-0.96 replica ceiling. A causal confirmation closes the loop: under vLLM's batch-invariant kernels all three passes are identical on every measured channel, with both disagreement rates exactly zero. Replica passes themselves disagree on 9-23% of eligible estimates: single-sample credit measurements at decision tokens are unreliable under any replay. Settings were fixed in advance; exact-pass cache hits in the second campaign are instrumented (100% hit rate, 3,434 pivots); total compute was under 10 USD. We recommend that counterfactual credit studies resume decoder state or use batch-invariant kernels, and report a replica floor.

翻译：逐令牌反事实信用估计旨在探究语言模型生成过程中，哪个令牌导致最终答案的正确或错误：在转折点处截断生成记录，替换一个替代令牌，重放后续生成并比较结果。已发表方法将生成记录的前缀作为新提示重新喂食，假设这能重现模型在生成过程中所处的状态。我们在标准推理引擎上衡量这一假设的代价，采用三遍实验设计：从经过验证的解码时KV状态恢复的延续、完全相同的第二遍精确恢复（副本噪声基准）以及一遍重新喂食过程。在六种配置和三个模型（包括一个经GRPO训练的检查点）中，对于低边际决策令牌，重新喂食导致信用估计的变化率比副本基准高14-28个百分点（在独立于处理的条件设置下为7-21个百分点；问题聚类t值=2.9-6.4）。大多数变化是量化估计器的零边界交叉而非极性反转，且扰动具有零均值一致性，因此平均量大致安全；但选择过程并非如此：通过阈值设定|\hat{A}_t|在重新喂食下选取的关键令牌集与精确恢复选择的交集在Jaccard系数0.34-0.90之间，而副本基准的上限为0.63-0.96。因果确认闭合了循环：在vLLM的批次不变核下，所有三遍实验在每个测量通道上完全相同，分歧率恰好为零。副本遍本身在9-23%的合格估计上存在分歧：决策令牌上的单样本信用测量在任何重放下都不可靠。设置已预先固定；第二轮实验中的精确恢复遍缓存命中率经过仪器测量（100%命中率，3,434个转折点）；总计算成本低于10美元。我们建议反事实信用研究应恢复解码器状态或使用批次不变核，并报告副本基准。