STARFlow: Scaling Latent Normalizing Flows for High-resolution Image Synthesis

TL;DR

STARFlow introduces a scalable normalizing flow model with Transformer-based autoregressive components for high-resolution image synthesis, combining theoretical universality with architectural innovations to achieve state-of-the-art quality.

Contribution

It presents TARFlow, a novel architecture that scales normalizing flows for high-res images using a deep-shallow Transformer design and latent space modeling, enabling exact likelihood training at this scale.

Findings

Achieves competitive results with diffusion models in image quality.

First successful large-scale normalizing flow for high-resolution images.

Introduces a guidance algorithm that improves sample quality.

Abstract

We present STARFlow, a scalable generative model based on normalizing flows that achieves strong performance in high-resolution image synthesis. The core of STARFlow is Transformer Autoregressive Flow (TARFlow), which combines the expressive power of normalizing flows with the structured modeling capabilities of Autoregressive Transformers. We first establish the theoretical universality of TARFlow for modeling continuous distributions. Building on this foundation, we introduce several key architectural and algorithmic innovations to significantly enhance scalability: (1) a deep-shallow design, wherein a deep Transformer block captures most of the model representational capacity, complemented by a few shallow Transformer blocks that are computationally efficient yet substantially beneficial; (2) modeling in the latent space of pretrained autoencoders, which proves more effective than direct pixel-level modeling; and (3) a novel guidance algorithm that significantly boosts sample quality. Crucially, our model remains an end-to-end normalizing flow, enabling exact maximum likelihood training in continuous spaces without discretization. STARFlow achieves competitive performance in both class-conditional and text-conditional image generation tasks, approaching state-of-the-art diffusion models in sample quality. To our knowledge, this work is the first successful demonstration of normalizing flows operating effectively at this scale and resolution.