基于弱交互反馈的任务对齐相似度的贝叶斯测试时推理
Bayesian Test-Time Inference of Task-Aligned Similarity from Weak Interactive Feedback
AI总结：
该研究提出无需训练的贝叶斯方法BASIS，利用弱测试时反馈推断任务对齐的相似度准则，在多项任务中提升了语言模型的决策性能。
摘要：
语言模型系统在测试时会通过比较相似度反复做出决策：它们检索示例、对候选进行排序并修正响应。然而相似度并非唯一的：词汇、语义、结构和风格相似性会指向不同的候选，而任务对齐的概念会随实例变化。我们提出BASIS（Bayesian Axis Selection via Interactive Signals，基于交互信号的贝叶斯轴选择），这是一种无需训练的贝叶斯方法，利用弱测试时反馈推断提供的特定任务库中哪个准则支配当前决策。它维护候选目标和准则的联合后验，从标量分数、成对偏好或验证器输出中更新该后验，并通过预期信息增益选择探针。在有限噪声模型中，我们刻画了目标-准则对可恢复的条件，并说明为何固定准则和固定混合规则在准则偏移下会失效。在概念恢复、推理示例选择和响应修正任务中，BASIS将六探针概念恢复的Success@1从56.0%提升至68.1%，推理准确率从72.3%提升至73.1%，修正胜率从61.4%提升至63.8%，优于调优后的静态混合模型；在推理任务（72.1%）和修正任务（61.2%）上，其性能也超过计算量匹配的最佳N选选择。增益在准则不匹配和候选池更大时最为显著，匹配对照实验表明，自适应准则加权解释了大部分收益，而联合目标-准则推理进一步提升了性能。BASIS仅在提供的准则中选择，不会发现新准则。
英文摘要：
Language model systems repeatedly make test-time decisions by comparing similarity: they retrieve exemplars, rank candidates, and revise responses. Yet similarity is not unique: lexical, semantic, structural, and stylistic resemblance can point to different candidates, and the task-aligned notion can change across instances. We introduce BASIS (Bayesian Axis Selection via Interactive Signals), a training-free Bayesian method that uses weak test-time feedback to infer which criterion in a supplied, task-specific library governs the current decision. It maintains a joint posterior over candidate targets and criteria, updates it from scalar scores, pairwise preferences, or verifier outputs, and selects probes by expected information gain. In a finite noisy model, we characterize when the target-criterion pair is recoverable and show why fixed-criterion and fixed-mixture rules can fail under criterion shift. Across concept recovery, reasoning exemplar selection, and response refinement, BASIS improves six-probe concept-recovery Success@1 from 56.0% to 68.1%, reasoning accuracy from 72.3% to 73.1%, and refinement win rate from 61.4% to 63.8% over tuned static mixtures. It also exceeds compute-matched best-of-N selection on reasoning (72.1%) and refinement (61.2%). Gains are largest under criterion mismatch and with larger candidate pools, and matched controls indicate that adaptive criterion weighting explains much of the benefit, with joint target-criterion inference adding further value. BASIS selects only among supplied criteria and does not discover new ones.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
