Emerging Properties in Unified Multimodal Pretraining

TL;DR

BAGEL is an open-source, unified multimodal model pretrained on diverse data, demonstrating emerging complex reasoning and outperforming existing open models in understanding and generation tasks.

Contribution

Introduction of BAGEL, a decoder-only multimodal model supporting understanding and generation, with detailed pretraining methods and open-source release.

Findings

BAGEL exhibits advanced multimodal reasoning abilities.

It significantly outperforms open-source models on standard benchmarks.

Demonstrates capabilities like image manipulation and future frame prediction.

Abstract

Unifying multimodal understanding and generation has shown impressive capabilities in cutting-edge proprietary systems. In this work, we introduce BAGEL, an open-source foundational model that natively supports multimodal understanding and generation. BAGEL is a unified, decoder-only model pretrained on trillions of tokens curated from large-scale interleaved text, image, video, and web data. When scaled with such diverse multimodal interleaved data, BAGEL exhibits emerging capabilities in complex multimodal reasoning. As a result, it significantly outperforms open-source unified models in both multimodal generation and understanding across standard benchmarks, while exhibiting advanced multimodal reasoning abilities such as free-form image manipulation, future frame prediction, 3D manipulation, and world navigation. In the hope of facilitating further opportunities for multimodal research, we share the key findings, pretraining details, data creation protocal, and release our code and checkpoints to the community. The project page is at https://bagel-ai.org/