Continuous Adversarial Flow Models

Shanchuan Lin; Ceyuan Yang; Zhijie Lin; Hao Chen; Haoqi Fan

cs.LGApr 13, 2026

Continuous Adversarial Flow Models

Authors: Shanchuan Lin, Ceyuan Yang, Zhijie Lin, Hao Chen, Haoqi Fan

Organizations: ByteDance Seed

Abstract

We propose continuous adversarial flow models, a type of continuous-time flow model trained with an adversarial objective. Unlike flow matching, which uses a fixed mean-squared-error criterion, our approach introduces a learned discriminator to guide training. This change in objective induces a different generalized distribution, which empirically produces samples that are better aligned with the target data distribution. Our method is primarily proposed for post-training existing flow-matching models, although it can also train models from scratch. On the ImageNet 256px generation task, our post-training substantially improves the guidance-free FID of latent-space SiT from 8.26 to 3.63 and of pixel-space JiT from 7.17 to 3.57. It also improves guided generation, reducing FID from 2.06 to 1.53 for SiT and from 1.86 to 1.80 for JiT. We further evaluate our approach on text-to-image generation, where it achieves improved results on both the GenEval and DPG benchmarks.

Figures & tables

Panel of Figure 1 — Figure 1: Generation without guidance. Our method yields better generalization.

Panel of Figure 2 — Figure 2: Visualization of the training dynamic. Top: learned $G(x_{t},t)$ trajectories over the probability flow. Bottom: corresponding $-D(x_{t},t)$ values at all $x_{t}$ . $-D$ is taken for more intuitive visualization as the generation process runs backward in time. (a) shows if only training $D$ with $\bar{v}_{t}$ as positives without $G$ as negatives, $D$ degenerates to uniform gradient. (b,c) show how $D$ reacts to $G$ during training. (d) shows $D$ converges to 0 everywhere as $G$ converges to the ground-truth flow.

Table 3

Table 4

Method	PE	CFG	Single Obj.	Two Obj.	Color Attr.	Position	Counting	Colors.	Overall
FM	No	No	0.72	0.23	0.11	0.09	0.25	0.59	0.33
CAFM	No	No	0.85	0.42	0.17	0.16	0.41	0.61	0.44
FM	Yes	No	0.95	0.66	0.35	0.40	0.42	0.81	0.60
CAFM	Yes	No	0.99	0.83	0.50	0.52	0.57	0.86	0.71
FM	Yes	Yes	0.99	0.89	0.62	0.69	0.77	0.89	0.81
CAFM	Yes	Yes	0.99	0.92	0.71	0.71	0.81	0.94	0.85

Table 5: GenEval [ 16 ] on 512px text-to-image generation.

Method	CFG	Global	Entity	Attribute	Relation	Other	Overall
FM	No	81.34	82.96	81.71	83.17	85.07	72.25
CAFM	No	87.82	86.65	86.33	86.49	84.85	77.21
FM	Yes	90.34	90.56	88.98	88.17	90.71	83.67
CAFM	Yes	89.55	89.83	89.99	91.20	91.88	85.21

Table 6: DPG-Bench [ 25 ] on 512px text-to-image generation.

Panel of Figure 3 — Figure 3: Curated text-to-image samples on GenEval prompts. Without PE and CFG to show the most diverse range of samples. Left is FM. Right is CAFM. More visualizations are in Fig. 13 of the appendix.

Appendix figures & tables18 assets

Supplementary material from the paper’s appendix.

Appendix

CFG	Method	Epoch	FID $\downarrow$	IS $\uparrow$	sFID $\downarrow$	Prec. $\uparrow$	Rec $\uparrow$
None	SiT	1400	8.26	131.65	6.32	0.68	0.67
	SiT+FM	1400+10	8.64	131.91	6.36	0.68	0.67
	SiT+CAFM	1400+10	3.63	178.08	4.72	0.71	0.69
1.1	SiT	1400	5.55	161.77	5.67	0.72	0.66
	SiT+FM	1400+10	5.53	161.72	5.68	0.72	0.65
	SiT+CAFM	1400+10	2.27	212.06	4.55	0.74	0.67

Appendix

Table 7: SiT-XL/2 full metrics on ImageNet 256px.

Table 10

CFG	$f(a,b)$	Epoch	FID $\downarrow$	IS $\uparrow$	sFID $\downarrow$	Prec. $\uparrow$	Rec $\uparrow$
None	Non-saturating	1400+10	3.54	167.45	5.20	0.72	0.68
	Hinge	1400+10	4.00	175.21	5.36	0.71	0.68
	Least squares	1400+10	3.63	178.08	4.72	0.71	0.69
1.3	Non-saturating	1400+10	1.58	245.10	4.84	0.79	0.64
	Hinge	1400+10	1.57	258.46	4.86	0.79	0.64
	Least squares	1400+10	1.53	263.52	4.59	0.78	0.64

Appendix

Table 12: SiT-XL/2 CAFM post-train ablation study on

f(a,b)

. Non-saturating [ 17 ] :

f(a,b)=-\log(\sigma(a))-\log(1-\sigma(b))

. Hinge [ 38 ] :

f^{D}(a,b)=\max(0,1-a)+\max(0,1+b),\ f^{G}(a,b)=-a+b

. Least squares [ 51 ] :

f(a,b)=(a-1)^{2}+(b+1)^{2}

. CFG is swept separately and the best result is reported for each.

CFG	Method	Epoch	FID $\downarrow$	IS $\uparrow$	sFID $\downarrow$	Prec. $\uparrow$	Rec $\uparrow$
None	JiT	600	7.17	151.54	5.51	0.68	0.67
	JiT+FM	600+10	9.30	139.00	6.16	0.67	0.66
	JiT+CAFM	600+10	3.57	198.08	4.77	0.74	0.65
1.2	JiT	600	4.60	188.88	5.34	0.71	0.66
	JiT+FM	600+10	5.97	176.99	5.88	0.71	0.65
	JiT+CAFM	600+10	2.47	232.69	4.76	0.76	0.64

Appendix

Table 13: JiT-H/16 full metrics on ImageNet 256px.

	SiT	JiT
$G,D$ learning rate	1e-5	1e-5
batch size	256	1024
total epoch	10	10
$D$ warm-up epoch	2	4
timesteps	uniform(0, 1)	lognormal(0.8, 0.8)*
CFG interval	[0, 1]	[0, 0.9]*

Appendix

Table 14: ImageNet CAFM post-training hyperparameters *We formulate

x_{0}

as image while JiT formulates

x_{0}

as noise, so the timestep is reversed in writing.

Panel of Figure 6 — Figure 6: SiT-XL/2 guidance-free, latent-space ImageNet 256px generation. Top is FM (FID 8.26). Bottom is CAFM (FID 3.63). Uncurated. We highlight samples with visible improvements in red.

Panel of Figure 7 — Figure 7: SiT-XL/2 guided, latent-space ImageNet 256px generation. Top is FM (CFG 1.5, FID 2.06). Bottom is CAFM (CFG 1.3, FID 1.53). Uncurated.

Panel of Figure 9 — Figure 9: JiT-H/16 guided, pixel-space ImageNet 256px generation. Top is FM (CFG2.2, FID 1.86). Bottom is CAFM (CFG1.8, FID 1.80). Uncurated.

	FM	CAFM
$G$ learning rate	5e-5	5e-5
$D$ learning rate		3e-5
$N$		16
$\lambda_{\mathrm{ot}}$		0
AdamW $\beta$	(0.9, 0.95)	(0, 0.95)
iterations	30k	FM10k+20k

Appendix

Table 15: Text-to-image CAFM post-training hyperparameters

Method	PE	CFG	Single Obj.	Two Obj.	Color Attr.	Position	Counting	Colors.	Overall
ZImage	No	No	0.78	0.37	0.16	0.11	0.32	0.59	0.39*
ZImage+FM			0.72	0.23	0.11	0.09	0.25	0.59	0.33
ZImage+CAFM			0.85	0.42	0.17	0.16	0.41	0.61	0.44
ZImage	Yes	No	0.96	0.83	0.43	0.47	0.51	0.85	0.68*
ZImage+FM			0.95	0.66	0.35	0.40	0.42	0.81	0.60
ZImage+CAFM			0.99	0.83	0.50	0.52	0.57	0.86	0.71

Appendix

Table 16: GenEval [ 16 ] on 512px T2I generation including original ZImage. *Trained on different datasets. Not directly comparable.

Method	CFG	Global	Entity	Attribute	Relation	Other	Overall
ZImage	No	87.27	87.22	86.44	88.97	89.21	79.83*
ZImage+FM		81.34	82.96	81.71	83.17	85.07	72.25
ZImage+CAFM		87.82	86.65	86.33	86.49	84.85	77.21
ZImage	Yes	90.55	91.71	91.49	92.43	87.94	86.35*
ZImage+FM		90.34	90.56	88.98	88.17	90.71	83.67
ZImage+CAFM		89.55	89.83	89.99	91.20	91.88	85.21

Appendix

Table 17: DPG-Bench [ 25 ] on 512px T2I generation including original ZImage. *Trained on different datasets. Not directly comparable.

Panel of Figure 10 — Figure 10: Curated text-to-image comparisons on DPG benchmark prompts. Prompts are shortened for paper presentation. (part 1 of 4)

Panel of Figure 11 — Figure 11: Curated text-to-image comparisons on DPG benchmark prompts. Prompts are shortened for paper presentation. (part 2 of 4)

Panel of Figure 12 — Figure 12: Curated text-to-image comparisons on DPG benchmark prompts. Prompts are shortened for paper presentation. (part 3 of 4)

Panel of Figure 13 — Figure 13: Curated text-to-image comparisons on DPG benchmark prompts. Prompts are shortened for paper presentation. (part 4 of 4)

Panel of Figure 14 — Figure 14: Failure cases for guidance-free text-to-image generation.

Panel of Figure 15 — Figure 15: Discriminator gradient norm.

Flow-Based Generative Modeling

Flow-based generative modeling learns probability transports from simple noise to data for fast sampling, density estimation, and controllable generation.

See all 538 papers

Continuous Adversarial MeanFlow Transfer
Aug 20, 2026Yara Bahram, Zahra Dehghani, Mélodie Desbos +3Diffusion Model Adaptation Flow Matching
LIVIA · ILLS · ÉTS Montreal +2
FAIL: Flow Matching Adversarial Imitation Learning for Image Generation
Feb 12, 2026Yeyao Ma, Chen Li, Xiaosong Zhang +2Flow Matching Conditional Image Generation
Shanghai Jiao Tong University · Xi’an Jiaotong University · Tencent
Energy-Guided Flow Matching
Aug 6, 2026Haoyang Tong, Yu He, Fang Li +6Flow Matching Conditional Image Generation
MAIS & NLPR, CASIA · JD.com · Xi'an Jiaotong University +1

Aug 20, 2026J/K move · Enter open · S save

Continuous Adversarial MeanFlow Transfer

Yara Bahram, Zahra Dehghani, Mélodie Desbos, Eric Granger+2

LIVIA · ILLS · ÉTS Montreal+2

Diffusion Model Adaptation Flow Matching Diffusion Model Inference Acceleration Domain Adaptation

Open paper AskPDF arXiv

Training fast generators on new domains with limited data remains challenging for two reasons. First, adapting a pretrained diffusion or flow model to a new domain leaves its costly multi-step sampling unaddressed, and existing acceleration methods are tied to the source parameterization--

ε

x

v

, or

u

--leaving heterogeneous pretrained models with no common acceleration target. Second, while adversarial refinement is proven effective for few-step quality, it is formulated only for instantaneous-velocity flows, not for the finite-interval average velocities that MeanFlow (MF) models predict. We address both problems. We propose MeanFlow-Transfer, which maps heterogeneous source outputs into a shared velocity representation, uses it to initialize an MF generator from the source weights, and optimizes an MF objective on the target domain. This unifies adaptation and acceleration in a single training loop across a broad range of pretrained models. We then introduce Continuous Adversarial MeanFlow, a post-training stage that extends continuous adversarial flow models from instantaneous velocities to MF's finite-interval average velocities. CAMF contrasts changes in a learned potential between real and predicted interval endpoints, recovering fine detail that MF regression averages away, and reduces to the instantaneous criterion in the vanishing-interval limit. Adapting four ImageNet-based source models--DiT (

ε

), SiT (

v

), JiT (

x

), iMF (

u

)--to five target domains, MF-T with CAMF matches or exceeds the fine-tuned teacher in FID and FDD at up to

125\times

fewer Neural Function Evaluations (NFEs), while CAMF improves MF-T's few-step FID by

29\%

on average.

Abstract

Flow-Based Generative Modeling

Explore similar work

Continuous Adversarial MeanFlow Transfer

FAIL: Flow Matching Adversarial Imitation Learning for Image Generation

Energy-Guided Flow Matching