Scaling Autoregressive Multi-Modal Models: Pretraining and Instruction Tuning

TL;DR

CM3Leon is a scalable, retrieval-augmented multi-modal language model capable of high-quality text and image generation, fine-tuned for diverse tasks with less training compute.

Contribution

It introduces a novel training recipe for multi-modal models combining retrieval-augmented pretraining and supervised fine-tuning, achieving state-of-the-art results.

Findings

State-of-the-art text-to-image generation with less compute

High controllability in image editing and generation tasks

Effective multi-modal training recipe

Abstract

We present CM3Leon (pronounced "Chameleon"), a retrieval-augmented, token-based, decoder-only multi-modal language model capable of generating and infilling both text and images. CM3Leon uses the CM3 multi-modal architecture but additionally shows the extreme benefits of scaling up and tuning on more diverse instruction-style data. It is the first multi-modal model trained with a recipe adapted from text-only language models, including a large-scale retrieval-augmented pre-training stage and a second multi-task supervised fine-tuning (SFT) stage. It is also a general-purpose model that can do both text-to-image and image-to-text generation, allowing us to introduce self-contained contrastive decoding methods that produce high-quality outputs. Extensive experiments demonstrate that this recipe is highly effective for multi-modal models. CM3Leon achieves state-of-the-art performance in text-to-image generation with 5x less training compute than comparable methods (zero-shot MS-COCO FID of 4.88). After SFT, CM3Leon can also demonstrate unprecedented levels of controllability in tasks ranging from language-guided image editing to image-controlled generation and segmentation.