Tunable MAGMAX: Preference-Aware Model Merging for Continual Learning

2026-05-20

AI Summary

Existing model merging approaches in continual learning struggle to flexibly adapt to diverse deployment environments or user-specific performance preferences across tasks. To address this limitation, this work proposes MAGMAX, a novel framework that enables preference-aware model merging for the first time. MAGMAX introduces a learnable preference vector that dynamically modulates the contribution of each task’s parameters during merging and automatically constructs this vector using only a small amount of data from the target environment, without any manual intervention. Experimental results on standard continual learning benchmarks demonstrate that MAGMAX effectively tailors task performance according to specified preferences, successfully adapts to various target environments, and achieves performance that is either superior or comparable to state-of-the-art methods.

Abstract

Continual learning (CL) aims to train models sequentially on multiple tasks while mitigating catastrophic forgetting of previously learned knowledge. Recent advances in large pre-trained models (LPMs) and model merging techniques, such as MAGMAX, have demonstrated effective CL performance by combining task-specific parameters. However, existing methods primarily focus on average performance across all tasks and do not adequately address how to construct models accommodating different deployment environments or varying user preferences. This paper proposes a model merging framework, termed Tunable MAGMAX, which enables preference-aware control of task-specific performance in CL. Our method introduces a preference vector that controls the number of elements selected from each task vector during model merging, allowing us to adjust the merged model performance according to their deployment needs. We further propose a method for automatically constructing appropriate preference vectors by leveraging small amounts of target environment data and datasets from model training tasks, thereby eliminating the need for manual specification. The experimental result on CL benchmark tasks demonstrates that Tunable MAGMAX effectively controls task-wise performance and successfully adapts merged models to various target environments. The proposed Tunable MAGMAX achieves superior or comparable performance to baseline methods, making it a practical solution for deploying CL models to various environments where the preferences of each task performance differ.

Problem

• continual learning
• model merging
• preference-aware
• catastrophic forgetting
• task-specific performance

Innovation

• Tunable MAGMAX
• preference-aware model merging
• continual learning
• task-specific performance control
• automatic preference vector