Bayesian-LoRA: LoRA based Parameter Efficient Fine-Tuning using Optimal Quantization levels and Rank Values trough Differentiable Bayesian Gates

TL;DR

Bayesian-LoRA introduces a Bayesian approach to parameter-efficient fine-tuning of large language models, optimizing rank and quantization levels to reduce energy consumption while maintaining or improving performance.

Contribution

It proposes a novel Bayesian method for jointly optimizing rank values and quantization levels in LoRA, enhancing efficiency and effectiveness in fine-tuning large models.

Findings

Achieves comparable or better performance than baselines on GLUE.

Reduces total bit operations by approximately 70%.

Learns optimal-rank quantized matrices effectively.

Abstract

It is a common practice in natural language processing to pre-train a single model on a general domain and then fine-tune it for downstream tasks. However, when it comes to Large Language Models, fine-tuning the entire model can be computationally expensive, resulting in very intensive energy consumption. As a result, several Parameter Efficient Fine-Tuning (PEFT) approaches were recently proposed. One of the most popular approaches is low-rank adaptation (LoRA), where the key insight is decomposing the update weights of the pre-trained model into two low-rank matrices. However, the proposed approaches either use the same rank value across all different weight matrices, which has been shown to be a sub-optimal choice, or do not use any quantization technique, one of the most important factors when it comes to a model's energy consumption. In this work, we propose Bayesian-LoRA which approaches low-rank adaptation and quantization from a Bayesian perspective by employing a prior distribution on both quantization levels and rank values. As a result, B-LoRA is able to fine-tune a pre-trained model on a specific downstream task, finding the optimal rank values and quantization levels for every low-rank matrix. We validate the proposed model by fine-tuning a pre-trained DeBERTaV3 on the GLUE benchmark. Moreover, we compare it to relevant baselines and present both qualitative and quantitative results, showing how the proposed approach is able to learn optimal-rank quantized matrices. B-LoRA performs on par with or better than the baselines while reducing the total number of bit operations by roughly 70% compared to the baseline methods.