TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models

TL;DR

TUNA introduces a unified visual representation for multimodal models, enabling end-to-end understanding and generation of images and videos, outperforming previous decoupled approaches across multiple benchmarks.

Contribution

It proposes a native UMM with a unified visual space using a VAE and representation encoder, improving performance and scalability over prior decoupled models.

Findings

Achieves state-of-the-art results in multimodal understanding and generation.

Unified representation improves task performance and scalability.

Joint training benefits both understanding and generation tasks.

Abstract

Unified multimodal models (UMMs) aim to jointly perform multimodal understanding and generation within a single framework. We present TUNA, a native UMM that builds a unified continuous visual representation by cascading a VAE encoder with a representation encoder. This unified representation space allows end-to-end processing of images and videos for both understanding and generation tasks. Compared to prior UMMs with decoupled representations, TUNA's unified visual space avoids representation format mismatches introduced by separate encoders, outperforming decoupled alternatives in both understanding and generation. Moreover, we observe that stronger pretrained representation encoders consistently yield better performance across all multimodal tasks, highlighting the importance of the representation encoder. Finally, in this unified setting, jointly training on both understanding and generation data allows the two tasks to benefit from each other rather than interfere. Our extensive experiments on multimodal understanding and generation benchmarks show that TUNA achieves state-of-the-art results in image and video understanding, image and video generation, and image editing, demonstrating the effectiveness and scalability of its unified representation design.