Large Language Models (LLMs) often suffer from overconfidence during inference, particularly when adapted to downstream domain-specific tasks with limited data. Previous work addresses this issue by employing approximate Bayesian estimation after the LLMs are trained, enabling them to quantify uncertainty. However, such post-training approaches' performance is severely limited by the parameters learned during training. In this paper, we go beyond post-training Bayesianization and propose Bayesian Low-Rank Adaptation by Backpropagation (BLoB), an algorithm that continuously and jointly adjusts both the mean and covariance of LLM parameters throughout the whole fine-tuning process. Our empirical results verify the effectiveness of BLoB in terms of generalization and uncertainty estimation, when evaluated on both in-distribution and out-of-distribution data.

翻译：大语言模型（LLMs）在推理过程中常表现出过度自信的问题，尤其是在使用有限数据适应下游领域特定任务时。先前的研究通过在LLMs训练后采用近似贝叶斯估计来解决这一问题，使其能够量化不确定性。然而，此类训练后贝叶斯化方法的性能严重受限于训练期间学习到的参数。本文超越了训练后贝叶斯化的范式，提出了基于反向传播的贝叶斯低秩自适应（BLoB）算法，该算法在整个微调过程中持续且联合地调整LLM参数的均值与协方差。我们的实证结果表明，在分布内与分布外数据评估中，BLoB在泛化能力与不确定性估计方面均展现出显著有效性。