EoRA: Fine-tuning-free Compensation for Compressed LLM with Eigenspace Low-Rank Approximation

TL;DR

EoRA is a fine-tuning-free method that enhances compressed large language models by adding low-rank matrices, significantly improving accuracy and inference speed without hardware constraints.

Contribution

EoRA introduces a novel low-rank augmentation technique for compressed LLMs that surpasses prior methods and includes an optimized CUDA kernel for faster inference.

Findings

Achieves up to 11.45% accuracy improvement on GSM8K for compressed LLaMA3-8B.

Accelerates inference by up to 1.4x with an optimized CUDA kernel.

Reduces memory overhead through quantization of EoRA.

Abstract

While post-training compression techniques effectively reduce the memory footprint, latency, and power consumption of Large Language Models (LLMs), they often result in noticeable accuracy degradation and remain limited by hardware and kernel constraints that restrict supported compression formats - ultimately reducing flexibility across a wide range of deployment scenarios. In this work, we propose EoRA - a novel $\textbf{fine-tuning-free}$ method that augments compressed LLMs with low-rank matrices, allowing users to rapidly enhance task-specific performance and freely balance the trade-off between accuracy and computational overhead beyond the constraints of compression formats. EoRA consistently outperforms prior fine-tuning-free low rank methods in recovering the accuracy of compressed LLMs, achieving notable accuracy improvements (e.g., $\mathbf{10.84\%}$ on ARC-Challenge, $\mathbf{6.74\%}$ on MathQA, and $\mathbf{11.45\%}$ on GSM8K for LLaMA3-8B compressed to 3-bit). We also introduce an optimized CUDA kernel, accelerating inference by up to 1.4x and reducing memory overhead through quantizing EoRA. Overall, EoRA offers a prompt solution for improving the accuracy of compressed models under varying user requirements, enabling more efficient and flexible deployment of LLMs. Code is available at https://github.com/NVlabs/EoRA.