# Unified Continuous Generative Models

arXiv: https://arxiv.org/abs/2505.07447 · HTML: https://mlsanity.com/paper/2505.07447/

## Key facts

- Cited by 19 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- First posted on arXiv 12 May 2025.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Built on by (most cited first, 19 of 19)

- [AlphaFlow: Understanding and Improving MeanFlow Models](https://mlsanity.com/paper/2510.20771/) (2025)
- [Transition Models: Rethinking the Generative Learning Objective](https://mlsanity.com/paper/2509.04394/) (2025)
- [Transition Matching Distillation for Fast Video Generation](https://mlsanity.com/paper/2601.09881/) (2026)
- [TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows](https://mlsanity.com/paper/2512.05150/) (2025)
- [FACM: Flow-Anchored Consistency Models](https://mlsanity.com/paper/2507.03738/) (2025)
- [LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model](https://mlsanity.com/paper/2604.20796/) (2026)
- [Neon: Negative Extrapolation From Self-Training Improves Image Generation](https://mlsanity.com/paper/2510.03597/) (2025)
- [Three-Body Scattering for Generative Modeling](https://mlsanity.com/paper/2607.18198/) (2026)
- [Stabilizing Consistency Training: A Flow Map Analysis and Self-Distillation](https://mlsanity.com/paper/2601.22679/) (2026)
- [Duality Models: An Embarrassingly Simple One-step Generation Paradigm](https://mlsanity.com/paper/2602.17682/) (2026)
- [CMuon: Accelerating and Stabilizing Diffusion Transformer Training via Chunked Momentum Orthogonalization](https://mlsanity.com/paper/2608.02502/) (2026)
- [A Hidden Semantic Bottleneck in Conditional Embeddings of Diffusion Transformers](https://mlsanity.com/paper/2602.21596/) (2026)
- [Rethinking UMM Visual Generation: Masked Modeling for Efficient Image-Only Pre-training](https://mlsanity.com/paper/2603.16139/) (2026)
- [Self-Adversarial One Step Generation via Condition Shifting](https://mlsanity.com/paper/2604.12322/) (2026)
- [Chain-of-Trajectories: Unlocking the Intrinsic Generative Optimality of Diffusion Models via Graph-Theoretic Planning](https://mlsanity.com/paper/2603.14704/) (2026)
- [When Can Conditional Flow Matching Replace Pointwise Negative Log-Likelihood?](https://mlsanity.com/paper/2608.28010/) (2026)
- [FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models](https://mlsanity.com/paper/2609.09905/) (2026)
- [BridgeVoC: Revitalizing Neural Vocoder from a Restoration Perspective](https://mlsanity.com/paper/2511.07116/) (2025)
- [Manifold Drift in Flow Preference Optimization: A Root Cause of Reward Hacking](https://mlsanity.com/paper/2608.20011/) (2026)

## Abstract (arXiv)

Recent advances in continuous generative models, including multi-step approaches like diffusion and flow-matching (typically requiring 8-1000 sampling steps) and few-step methods such as consistency models (typically 1-8 steps), have demonstrated impressive generative performance. However, existing work often treats these approaches as distinct paradigms, resulting in separate training and sampling methodologies. We introduce a unified framework for training, sampling, and analyzing these models. Our implementation, the Unified Continuous Generative Models Trainer and Sampler (UCGM-{T,S}), achieves state-of-the-art (SOTA) performance. For example, on ImageNet 256x256 using a 675M diffusion transformer, UCGM-T trains a multi-step model achieving 1.30 FID in 20 steps and a few-step model reaching 1.42 FID in just 2 steps. Additionally, applying UCGM-S to a pre-trained model (previously 1.26 FID at 250 steps) improves performance to 1.06 FID in only 40 steps. Code is available at: https://github.com/LINs-lab/UCGM.