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.
@article{arxiv.2406.11675,
title = {BLoB: Bayesian Low-Rank Adaptation by Backpropagation for Large Language Models},
author = {Yibin Wang and Haizhou Shi and Ligong Han and Dimitris Metaxas and Hao Wang},
journal= {arXiv preprint arXiv:2406.11675},
year = {2025}
}
Comments
Accepted at NeurIPS 2024. Additional experiments have been included in the appendix