BaseCal: 通过基础模型信号实现无监督的置信度校准
BaseCal: Unsupervised Confidence Calibration via Base Model Signals
AI总结:
本文提出BaseCal方法,利用基础模型信号校准大语言模型的置信度,通过两种方法减少校准误差,实验表明其在多个数据集上有效降低了ECE。
摘要:
可靠的置信度对于信任LLM的输出至关重要,但广泛部署的后训练LLM(PoLLM)通常通过严重过度自信损害了这种信任。相反,我们观察到其对应的基础LLM通常仍保持良好校准。这自然促使我们使用基础LLM作为参考来校准PoLLM的置信度。本文提出了两种方法。一种直接的解决方案,BaseCal-ReEval,通过将PoLLM的响应输入基础LLM以获得平均概率作为置信度。虽然有效,但这种方法引入了额外的推理开销。为了解决这个问题,我们提出了BaseCal-Proj,它训练了一个轻量级的投影器,将PoLLM的最终层隐藏状态映射回其基础LLM的隐藏状态。这些投影状态随后通过基础LLM的输出层处理,以推导出基础校准的置信度。值得注意的是,BaseCal是一种无监督、即插即用的解决方案,无需人工标签或LLM修改。在五个数据集和三个LLM家族上的实验表明,BaseCal的有效性,与最佳无监督基线相比,平均将预期校准误差(ECE)减少了42.90%。
英文摘要:
Reliable confidence is essential for trusting the outputs of LLMs, yet widely deployed post-trained LLMs (PoLLMs) typically compromise this trust with severe overconfidence. In contrast, we observe that their corresponding base LLMs often remain well-calibrated. This naturally motivates us to calibrate PoLLM confidence using the base LLM as a reference. This work proposes two ways to achieve this. A straightforward solution, BaseCal-ReEval, evaluates PoLLM's responses by feeding them into the base LLM to get average probabilities as confidence. While effective, this approach introduces additional inference overhead. To address this, we propose BaseCal-Proj, which trains a lightweight projection to map the final-layer hidden states of PoLLMs back to those of their base LLMs. These projected states are then processed by the base LLM's output layer to derive base-calibrated confidence for PoLLM's responses. Notably, BaseCal is an unsupervised, plug-and-play solution that operates without human labels or LLM modifications. Experiments across five datasets and three LLM families demonstrate the effectiveness of BaseCal, reducing Expected Calibration Error (ECE) by an average of 42.90\% compared to the best unsupervised baselines.
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。