Large Language Models (LLMs), including ChatGPT and LLaMA, are susceptible to
generating hallucinated answers in a confident tone. While efforts to elicit
and calibrate confidence scores have proven useful, recent findings show that
controlling uncertainty must go beyond calibration: predicted scores may
deviate significantly from the actual posterior probabilities due to the impact
of grouping loss. In this work, we construct a new evaluation dataset derived
from a knowledge base to assess confidence scores given to answers of Mistral
and LLaMA. Experiments show that they tend to be overconfident. Further, we
show that they are more overconfident on some answers than others, \emph{eg}
depending on the nationality of the person in the query. In
uncertainty-quantification theory, this is grouping loss. To address this, we
propose a solution to reconfidence LLMs, canceling not only calibration but
also grouping loss. The LLMs, after the reconfidencing process, indicate
improved confidence alignment with the accuracy of their responses.

大型语言模型 (LLMs)，包括 ChatGPT 和 LLaMA，在以自信的口吻生成虚构答案方面容易出错。本文通过构建来自知识库的新评估数据集，评估了 Mistral 和 LLaMA 给出的答案的置信度分数，并展示它们倾向于过于自信。我们还发现它们在一些答案上比在其他答案上更为自信，例如取决于查询中的人的国籍。为了解决这个问题，我们提出了一种重新确定置信度的方法，取消了校准与分组损失。在重新确定置信度的过程中，语言模型表明其响应的准确性与其置信度的对齐有所改善。

从分组损失角度重建 LLMs

Reconfidencing LLMs from the Grouping Loss Perspective

Large language models (LLMs) have achieved remarkable advancements in natural
language understanding, generation, and manipulation of text-based data.
However, one major issue towards their widespread deployment in the real world
is that they can generate "hallucinated" answers that are not factual. Towards
this end, this paper focuses on improving grounding from a holistic perspective
with a novel framework, AGREE, Adaptation of LLMs for GRounding EnhancEment. We
start with the design of an iterative test-time adaptation (TTA) capability
that takes into account the support information generated in self-grounded
responses. To effectively enable this capability, we tune LLMs to ground the
claims in their responses to retrieved documents by providing citations. This
tuning on top of the pre-trained LLMs requires a small amount of data that
needs to be constructed in a particular way to learn the grounding information,
for which we introduce a data construction method. Our results show that the
tuning-based AGREE framework generates better grounded responses with more
accurate citations compared to prompting-based approaches.

这篇论文关注通过一种名为 AGREE 的新框架，综合地改善大型语言模型在真实世界中的应用，以解决其生成的 “幻觉” 答案不符事实的问题。通过在生成的自身立足回答中考虑所生成的支持信息，设计了一种迭代的测试时间调整能力来有效实现该框架，并通过提供引用来调整大型语言模型对检索文档中主张的立足。结果表明，基于调整的 AGREE 框架相较于基于提示的方法，生成了更好的基于立足的回答和更准确的引用。