LoRS: Efficient Low-Rank Adaptation for Sparse Large Language Model
This work addresses efficiency issues in fine-tuning sparse LLMs, which is an incremental improvement for researchers and practitioners in natural language processing.
The paper tackles the problem of high memory and computation overhead in low-rank adaptation (LoRA) methods for sparse large language models (LLMs) by introducing LoRS, which uses weight recompute, computational graph rearrangement, and better adapter initialization to reduce consumption while outperforming existing LoRA approaches.
Existing low-rank adaptation (LoRA) methods face challenges on sparse large language models (LLMs) due to the inability to maintain sparsity. Recent works introduced methods that maintain sparsity by augmenting LoRA techniques with additional masking mechanisms. Despite these successes, such approaches suffer from an increased memory and computation overhead, which affects efficiency of LoRA methods. In response to this limitation, we introduce LoRS, an innovative method designed to achieve both memory and computation efficiency when fine-tuning sparse LLMs. To mitigate the substantial memory and computation demands associated with preserving sparsity, our approach incorporates strategies of weight recompute and computational graph rearrangement. In addition, we also improve the effectiveness of LoRS through better adapter initialization. These innovations lead to a notable reduction in memory and computation consumption during the fine-tuning phase, all while achieving performance levels that outperform existing LoRA approaches.