AIA: Rethinking Architecture Decoupling Strategy In Unified Multimodal Model

TL;DR

This paper introduces Attention Interaction Alignment (AIA) loss to improve unified multimodal models by aligning task-specific cross-modal interactions, avoiding excessive architecture decoupling.

Contribution

The paper proposes a novel AIA loss that mitigates task conflicts in multimodal models without decoupling, enhancing both understanding and generation capabilities.

Findings

AIA refines cross-modal attention patterns effectively.

Applying AIA boosts performance in both generation and understanding tasks.

Decoupling drives models toward task-specific interaction patterns, which AIA aims to align.

Abstract

Unified multimodal models for image generation and understanding represent a significant step toward AGI and have attracted widespread attention from researchers. The main challenge of this task lies in the difficulty in establishing an optimal training paradigm due to inherent conflicting targets in understanding and generation tasks. To alleviate these conflicts and pursue higher performance, many researchers adopt varying degrees of architecture decoupling (e.g., Double image encoders, MOE/MOT architecture, or frozen MLLM). However, excessive model decoupling can lead to the loss of interleave generation ability, undermining the original intent of unified models. In this work, we aim to explore how to mitigate task conflicts without resorting to model decoupling. Firstly, we analyze why decoupling boosts performance by studying the cross-modal attention behavior of models. We observe that architecture decoupling does not solve task conflicts, but essentially drives models toward cross-modal interaction patterns of task-specific models, as seen in Qwen3-VL and HunyuanImage-3.0, and that the more thorough the decoupling, the more consistent the behavior becomes. Motivated by this observation, we propose Attention Interaction Alignment (AIA) loss, which explicitly learns task-specific multimodal interaction patterns during training. To demonstrate the generalizability of our AIA loss, we apply it to Emu3 and Janus-Pro during SFT and post-training stage respectively. Without bells and whistles, AIA not only refines cross-modal attention patterns, but also boosts both generation and understanding performance.