自我判断混淆下的正确性探测诊断
Diagnosing Correctness Probes under Self-Judgement Confounding
浏览论文内容
中文总结 AI 辅助
研究语言模型中隐藏状态读数预测正确性时，因客观正确性与模型自我判断混淆致信号模糊。构建冲突案例，估计相关方向极性，发现与SJ相关方向跨域转移高于机会水平，OC相关方向则相反，且转移不对称，表明仅可转移性不能确立客观正确性语义。
中文摘要 AI 辅助
隐藏状态读数可预测语言模型输出是否正确，但客观正确性（OC）通常与模型自身的自我判断（SJ）一致，导致解码信号语义模糊。我们构建了冲突案例，其中OC和SJ预测相反的读数顺序。在高置信度分歧中，传统的正确性标记对比通常将错误/自我认可的回答排在正确/自我拒绝的回答之上，遵循SJ而非OC。我们估计了与SJ和OC相关的阶乘方向，并在数学推理和事实回忆中评估它们的极性。在四个高达14B参数的指令调整模型中，与SJ相关的方向在每个模型的两个跨域方向上的转移高于机会水平，而与OC相关的方向在每个相应条件下对预期OC排序的点估计低于机会水平。这种转移不对称在中到后期层发展，在答案可能性、序列长度和空方向控制下持续存在，并扩展到MMLU和二元TruthfulQA，无需目标域方向拟合。在研究的模型和诊断子集中，最可靠可转移的组件保留了与SJ相关的极性。因此，仅可转移性并不能确立客观正确性语义。
英文摘要
Hidden-state readouts can predict whether language-model outputs are correct, but objective correctness (OC) usually agrees with the model's own self-judgement (SJ), leaving the decoded signal semantically ambiguous. We construct conflict cases in which OC and SJ predict opposite readout orderings. On high-confidence disagreements, conventional correctness-labelled contrasts often rank incorrect/self-endorsed responses above correct/self-rejected responses, following SJ rather than OC. We estimate factorial SJ- and OC-associated directions and evaluate their polarity across mathematical reasoning and factual recall. Across four instruction-tuned models up to 14B parameters, the SJ-associated direction transfers above chance in both cross-domain directions for every model, whereas the OC-associated direction has a below-chance point estimate for the expected OC ordering in every corresponding condition. This transfer asymmetry develops across middle-to-late layers, persists under answer-likelihood, sequence-length, and null-direction controls, and extends to MMLU and binary TruthfulQA without target-domain direction fitting. Across the studied models and diagnostic subsets, the most reliably transferable component preserves SJ-associated polarity. Transferability alone therefore does not establish objective-correctness semantics.
