paper-with-me

홈 › Papers

Duality Models: An Embarrassingly Simple One-step Generation Paradigm

2026-02-04 · Peng Sun, Xinyi Shang, Tao Lin, Zhiqiang Shen arxiv

Consistency-based generative models like Shortcut and MeanFlow achieve impressive results via a target-aware design for solving the Probability Flow ODE (PF-ODE). Typically, such methods introduce a target time $r$ alongside the current time $t$ to modulate outputs between a local multi-step derivative ($r = t$) and a global few-step integral ($r = 0$). However, the conventional "one input, one output" paradigm enforces a partition of the training budget, often allocating a significant portion (e.g., 75% in MeanFlow) solely to the multi-step objective for stability. This separation forces a trade-off: allocating sufficient samples to the multi-step objective leaves the few-step generation undertrained, which harms convergence and limits scalability. To this end, we propose Duality Models (DuMo) via a "one input, dual output" paradigm. Using a shared backbone with dual heads, DuMo simultaneously predicts velocity $v_t$ and flow-map $u_t$ from a single input $x_t$. This applies geometric constraints from the multi-step objective to every sample, bounding the few-step estimation without separating training objectives, thereby significantly improving stability and efficiency. On ImageNet 256 $\times$ 256, a 679M Diffusion Transformer with SD-VAE achieves a state-of-the-art (SOTA) FID of 1.79 in just 2 steps. Code is available at: https://github.com/LINs-lab/DuMo

📄 PDF Abstract BibTeX arXiv:2602.17682

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Embarrassingly Simple Binary Representation Learning

2019-08-26 · Yuming Shen, Jie Qin, Jiaxin Chen, Li Liu 외

Recent binary representation learning models usually require sophisticated binary optimization, similarity measure or even generative models as auxiliaries. However, one may wonder whether these non-trivial components ar…

Representation Learning

Just CHOP: Embarrassingly Simple LLM Compression

2023-05-24 · Ananya Harsh Jha, Tom Sherborne, Evan Pete Walsh, Dirk Groeneveld 외

Large language models (LLMs) enable unparalleled few- and zero-shot reasoning capabilities but at a high computational footprint. A growing assortment of methods for compression promises to reduce the computational burde…

Knowledge DistillationLanguage ModelingLanguage ModellingLarge Language Model+1

Do It Once: An Embarrassingly Simple Joint Matching Approach to Response Selection

2021-08-01 · Findings (ACL) 2021 8 · Linhao Zhang, Dehong Ma, Sujian Li, Houfeng Wang

MLPInit: Embarrassingly Simple GNN Training Acceleration with MLP Initialization

2022-09-30 · Xiaotian Han, Tong Zhao, Yozen Liu, Xia Hu 외

Training graph neural networks (GNNs) on large graphs is complex and extremely time consuming. This is attributed to overheads caused by sparse matrix multiplication, which are sidestepped when training multi-layer perce…

Link PredictionNode Classification

One Language-Free Foundation Model Is Enough for Universal Vision Anomaly Detection

2026-01-09 · Bin-Bin Gao, Chengjie Wang arxiv

Universal visual anomaly detection (AD) aims to identify anomaly images and segment anomaly regions towards open and dynamic scenarios, following zero- and few-shot paradigms without any dataset-specific fine-tuning. We …

Anomaly ClassificationPrompt EngineeringAnomaly Detection