KCM: KAN-Based Collaboration Models Enhance Pretrained Large Models

TL;DR

This paper introduces KCM, a novel collaborative model using KAN architecture to improve large-small model collaboration, reducing resource use and catastrophic forgetting across language, vision, and cross-modal tasks.

Contribution

The paper proposes KCM, an innovative KAN-based approach that enhances interpretability, mitigates forgetting, and improves efficiency in large-small model collaborations.

Findings

KCM reduces large model inference calls significantly.

KCM maintains near-identical accuracy with lower resource consumption.

KCM outperforms MLP-based models in all evaluated metrics.

Abstract

In recent years, Pretrained Large Models(PLMs) researchers proposed large-small model collaboration frameworks, leveraged easily trainable small models to assist large models, aim to(1) significantly reduce computational resource consumption while maintaining comparable accuracy, and (2) enhance large model performance in specialized domain tasks. However, this collaborative paradigm suffers from issues such as significant accuracy degradation, exacerbated catastrophic forgetting, and amplified hallucination problems induced by small model knowledge. To address these challenges, we propose a KAN-based Collaborative Model (KCM) as an improved approach to large-small model collaboration. The KAN utilized in KCM represents an alternative neural network architecture distinct from conventional MLPs. Compared to MLPs, KAN offers superior visualizability and interpretability while mitigating catastrophic forgetting. We deployed KCM in large-small model collaborative systems across three scenarios: language, vision, and vision-language cross-modal tasks. The experimental results demonstrate that, compared with pure large model approaches, the large-small model collaboration framework utilizing KCM as the collaborative model significantly reduces the number of large model inference calls while maintaining near-identical task accuracy, thereby substantially lowering computational resource consumption. Concurrently, the KAN-based small collaborative model markedly mitigates catastrophic forgetting, leading to significant accuracy improvements for long-tail data. The results reveal that KCM demonstrates superior performance across all metrics compared to MLP-based small collaborative models (MCM).