---
title: "ByteDance's CAFM Cuts Image Generation Error in Half Without Extra Guidance"
subtitle: "ByteDance Seed's continuous adversarial flow models replace mean squared error with a learned critic, cutting ImageNet guidance-free FID by more than half."
url: "https://alphasignal.ai/news/bytedance-s-cafm-cuts-image-generation-error-in-half-without-extra-guidance"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-04-13T00:00:00.000Z"
updated: "2026-08-24T03:13:25.984Z"
type: "Paper"
topics:
  - "Image"
  - "Post Training"
subtopics:
  - "Image Generation"
  - "Distillation"
  - "Alignment"
upvotes: 3315
source_url: "https://arxiv.org/abs/2604.11521"
comments: 0
truncated: true
---

# ByteDance's CAFM Cuts Image Generation Error in Half Without Extra Guidance

> ByteDance Seed's continuous adversarial flow models replace mean squared error with a learned critic, cutting ImageNet guidance-free FID by more than half.

Flow matching has quietly become the default engine behind modern image and video generators, but it has a stubborn habit of producing off-manifold samples unless you crank up classifier-free guidance. A team from ByteDance Seed proposes trading the fixed mean-squared-error objective for a learned adversarial critic, and the numbers on ImageNet are hard to ignore: guidance-free FID drops from 8.26 to 3.63 on latent-space [SiT](https://arxiv.org/html/2604.11521) and from 7.17 to 3.57 on pixel-space JiT, using just 10 epochs of fine-tuning.

The method, called [Continuous Adversarial Flow Models](https://arxiv.org/abs/2604.11521) (CAFMs), keeps everything about a flow-matching model the same but post-trains it with a GAN-style discriminator instead of MSE. The result is a drop-in objective swap that improves both guidance-free and guided generation.

## Why MSE is quietly hurting your samples

Flow matching learns the velocity field of a probability flow between noise and data distributions by minimizing the squared L2 loss between predicted and target velocities. With infinite capacity, this converges to the ground-truth flow. In practice, finite-capacity networks have to generalize, and the L2 loss measures Euclidean distance rather than distance along the data manifold. That induces incorrect generalization relative to the underlying data distribution.

That is why models often need classifier-free guidance to look sharp, and why guidance itself is a bit of a hack: it improves quality by biasing sampling away from the true learned distribution. Prior fixes each had trade-offs:

* Riemannian flow matching extends flow matching to non-Euclidean geometries, but requires manual definition of the data manifold, which is often unknown for general datasets.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Continuous Adversarial Flow Models](https://arxiv.org/abs/2604.11521)
- [Continuous Adversarial Flow Models (HTML)](https://arxiv.org/html/2604.11521)
- [Z-Image: An Efficient Image Generation Foundation Model](https://arxiv.org/abs/2511.22699</li>)