COFT: 用于大语言模型中公平思维链推理的反事实-保形解码
COFT: Counterfactual-Conformal Decoding for Fair Chain-of-Thought Reasoning in Large Language Models
AI总结：
提出COFT，一种无需训练的解码方法，通过反事实提示和保形校准在解码时实现token级公平性控制，显著减少思维链生成中的社会偏见，同时保持任务效用和语言质量。
摘要：
大型语言模型（LLMs）在思维链（CoT）生成过程中可能揭示并放大社会偏见。我们提出COFT（Chain of Fair Thought），一种无需训练的解码方法，在解码时应用token级公平性控制，并对任何冻结的因果语言模型提供无分布边际有效性保证（在可交换性下）。COFT分三个阶段运行。首先，通过将敏感跨度替换为中性token来创建掩码反事实提示。其次，通过轻量级logit融合比较事实和掩码logit分布，以减弱属性驱动的偏见。第三，使用双分支分裂保形校准，在用户选择的风险水平下认证每步候选token集。我们在六个模型和多个偏见基准上评估COFT。我们的方法将标准偏见指标降低30-55%（中位数38%），同时保持任务效用和语言质量。推理准确率在运行间噪声范围内保持不变。计算开销适中，相当于一次额外的缓存前向传递（<=11%）。COFT提供了一条清晰、可审计的路径，实现更安全的CoT生成，显著减少偏见，效用损失可忽略，且无需重新训练、辅助分类器或权重访问。
英文摘要：
Large language models (LLMs) can reveal and amplify societal biases during chain-of-thought (CoT) generation. We present COFT (Chain of Fair Thought), a training-free decoding method that applies token-level fairness control at decode time, with distribution-free marginal validity guarantees (under exchangeability) for any frozen causal language model. COFT operates in three stages. First, it creates a masked counterfactual prompt by replacing sensitive spans with neutral tokens. Second, it compares the factual and masked logit distributions through lightweight logit fusion to attenuate attribute-driven biases. Third, it uses dual-branch split-conformal calibration to certify per-step candidate token sets at a user-chosen risk level. We evaluate COFT across six models and multiple bias benchmarks. Our method reduces standard bias metrics by 30-55% (median 38%) while preserving task utility and language quality. Reasoning accuracies remain unchanged within run-to-run noise margins. The computational overhead is modest, equivalent to one additional cached forward pass (<=11%). COFT offers a clear, auditable path to safer CoT generation with significant bias reduction, negligible utility loss, and no requirement for retraining, auxiliary classifiers, or weight access.
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。
