CARE-LoRA: Compressed Activation REconstruction for Memory-Efficient LoRA

Researchers proposed CARE-LoRA, a compressed activation reconstruction framework that reduces memory usage during LoRA fine-tuning of large pre-trained models by exploiting the projection structure of LoRA to replace full input activations with low-rank compressed activations. The method achieves competitive or superior performance while substantially lowering memory footprint, with code available on GitHub.

arXiv:2607.11940v1 Announce Type: new Abstract: As the scale of large pre-trained models continues to grow, fine-tuning them under limited memory budgets has become increasingly challenging. Low-Rank Adaptation LoRA , currently one of the most widely adopted parameter-efficient fine-tuning PEFT methods, mitigates this challenge by optimizing only low-rank adaptation matrices, thereby greatly reducing the number of trainable parameters. With the parameter overhead substantially reduced, the activations retained for backpropagation have emerged as the primary remaining memory bottleneck during LoRA fine-tuning. To address this, we propose CARE-LoRA, a data-aware Compressed Activation REconstruction framework. By exploiting the inherent projection structure of LoRA, CARE-LoRA replaces the full input activation with the low-rank compressed activation naturally produced by the LoRA branch. It further computes a lightweight reconstruction matrix during the forward pass with negligible additional computation cost, which is used during backpropagation to reconstruct the gradient signal, thereby keeping LoRA matrices fully trainable. Extensive experiments across diverse models and downstream tasks demonstrate that, while substantially reducing the overall memory footprint, CARE-LoRA achieves competitive or even superior performance compared with standard LoRA and representative LoRA variants. Our code is publicly available at https://github.com/fishandyu/CARE-LoRA .