EasyGen: Easing Multimodal Generation with BiDiffuser and LLMs

TL;DR

EasyGen introduces a novel multimodal generation framework combining BiDiffuser diffusion models and LLMs, achieving efficient, high-quality image and text generation with less training data and better modality interaction.

Contribution

It proposes EasyGen, a new model that leverages BiDiffuser and LLMs for more efficient multimodal understanding and generation, reducing data requirements and improving quality.

Findings

Outperforms existing models in data efficiency and generation quality

Demonstrates strong extendibility across modalities

Achieves high-quality image and text generation

Abstract

We present EasyGen, an efficient model designed to enhance multimodal understanding and generation by harnessing the capabilities of diffusion models and large language models (LLMs), Unlike existing multimodal models that predominately depend on encoders like CLIP or ImageBind and need ample amounts of training data to bridge modalities,EasyGen leverages BiDiffuser,a bidirectional conditional diffusion model, to foster more efficient modality interactions. Easygen achieves text generation by training a projection layer linking BiDiffuser and an LLM, and facilities image generation by training an adapter to align the LLM's text space with the BiDiffuser's image space, Comprehensive quantitative and qualitative experiments show that EasyGen excels in data-efficient training, high-quality image generation, and extendibility, effectively addressing the challenges in multimodal generation. The source code is available at https://github.com/zxy556677/EasyGen.