mPLUG-2: A Modularized Multi-modal Foundation Model Across Text, Image and Video

TL;DR

mPLUG-2 introduces a modular multi-modal foundation model that effectively integrates text, image, and video understanding and generation, achieving state-of-the-art results across diverse tasks with improved flexibility and efficiency.

Contribution

It proposes a modularized design for multi-modal pretraining, enabling flexible task-specific module selection and addressing modality entanglement, which is a novel approach in multi-modal modeling.

Findings

Achieves state-of-the-art results on MSRVTT video QA and captioning.

Demonstrates strong zero-shot transferability across tasks.

Operates with a smaller model size and data scale.

Abstract

Recent years have witnessed a big convergence of language, vision, and multi-modal pretraining. In this work, we present mPLUG-2, a new unified paradigm with modularized design for multi-modal pretraining, which can benefit from modality collaboration while addressing the problem of modality entanglement. In contrast to predominant paradigms of solely relying on sequence-to-sequence generation or encoder-based instance discrimination, mPLUG-2 introduces a multi-module composition network by sharing common universal modules for modality collaboration and disentangling different modality modules to deal with modality entanglement. It is flexible to select different modules for different understanding and generation tasks across all modalities including text, image, and video. Empirical study shows that mPLUG-2 achieves state-of-the-art or competitive results on a broad range of over 30 downstream tasks, spanning multi-modal tasks of image-text and video-text understanding and generation, and uni-modal tasks of text-only, image-only, and video-only understanding. Notably, mPLUG-2 shows new state-of-the-art results of 48.0 top-1 accuracy and 80.3 CIDEr on the challenging MSRVTT video QA and video caption tasks with a far smaller model size and data scale. It also demonstrates strong zero-shot transferability on vision-language and video-language tasks. Code and models will be released in https://github.com/alibaba/AliceMind.