Uncertainty quantification in fine-tuned LLMs using LoRA ensembles

📅 2024-02-19
🏛️ arXiv.org
📈 Citations: 15
Influential: 2
📄 PDF

career value

200K/year
🤖 AI Summary
To address the challenges of quantifying uncertainty, elucidating knowledge evolution mechanisms, and assessing prediction reliability after fine-tuning large language models (LLMs), this paper proposes a Bayesian posterior approximation method based on LoRA adapter ensembles. It is the first to integrate low-rank adapter ensembles with variational posterior approximation, enabling efficient and interpretable uncertainty modeling. Evaluated on Mistral-7B, the method quantitatively characterizes the dynamic trade-off between prior knowledge retention and domain adaptation during fine-tuning, revealing the counterintuitive phenomenon that acquired knowledge remains strongly preserved even in overfitting regimes. Comprehensive evaluation on multiple-choice benchmarks—including MMLU, ARC, and HellaSwag—demonstrates significant improvements in uncertainty calibration accuracy and predictive confidence assessment.

Technology Category

Application Category

📝 Abstract
Fine-tuning large language models can improve task specific performance, although a general understanding of what the fine-tuned model has learned, forgotten and how to trust its predictions is still missing. We derive principled uncertainty quantification for fine-tuned LLMs with posterior approximations using computationally efficient low-rank adaptation ensembles. We analyze three common multiple-choice datasets using low-rank adaptation ensembles based on Mistral-7b, and draw quantitative and qualitative conclusions on their perceived complexity and balance between retained prior knowledge and domain specific adaptation during and after fine-tuning. We identify unexpected retention of acquired knowledge during fine-tuning in the overfitting regime.
Problem

Research questions and friction points this paper is trying to address.

Quantify uncertainty in fine-tuned LLMs using LoRA ensembles
Analyze retained knowledge and domain adaptation during fine-tuning
Investigate unexpected knowledge retention in overfitting regimes
Innovation

Methods, ideas, or system contributions that make the work stand out.

LoRA ensembles for uncertainty quantification
Low-rank adaptation for efficient fine-tuning
Analyzing knowledge retention in overfitting
🔎 Similar Papers
No similar papers found.