PACE: Marrying generalization in PArameter-efficient fine-tuning with Consistency rEgularization

TL;DR

PACE introduces a novel regularization method combining gradient norm reduction and model alignment with consistency regularization, significantly improving the generalization of parameter-efficient fine-tuning across various tasks.

Contribution

The paper proposes PACE, a new method that enhances PEFT by combining gradient regularization and model alignment with consistency regularization, supported by theoretical analysis and extensive experiments.

Findings

PACE outperforms existing PEFT methods in visual tasks.

PACE improves performance on text classification and reasoning tasks.

Theoretical analysis confirms gradient regularization and model alignment benefits.

Abstract

Parameter-Efficient Fine-Tuning (PEFT) effectively adapts pre-trained transformers to downstream tasks. However, the optimization of tasks performance often comes at the cost of generalizability in fine-tuned models. To address this issue, we theoretically connect smaller weight gradient norms during training and larger datasets to the improvements in model generalization. Motivated by this connection, we propose reducing gradient norms for enhanced generalization and aligning fine-tuned model with the pre-trained counterpart to retain knowledge from large-scale pre-training data. Yet, naive alignment does not guarantee gradient reduction and can potentially cause gradient explosion, complicating efforts to manage gradients. To address such an issue, we propose PACE, marrying generalization of PArameter-efficient fine-tuning with Consistency rEgularization. We perturb features learned from the adapter with the multiplicative noise and ensure the fine-tuned model remains consistent for same sample under different perturbations. Theoretical analysis shows that PACE not only implicitly regularizes gradients for enhanced generalization, but also implicitly aligns the fine-tuned and pre-trained models to retain knowledge. Experimental evidence supports our theories. PACE surpasses existing PEFT methods in visual adaptation tasks (VTAB-1k, FGVC, few-shot learning, domain adaptation) showcasing its potential for resource-efficient fine-tuning. It also improves LoRA in text classification (GLUE) and mathematical reasoning (GSM-8K). The code is available at https://github.com/MaxwellYaoNi/PACE