Beyond Next-Token Prediction: A Performance Characterization of Diffusion versus Autoregressive Language Models
超越单个令牌预测:扩散模型与自回归语言模型性能的对比分析
Minseo Kim, Coleman Hooper, Aditya Tomar, Chenfeng Xu, Mehrdad Farajtabar, Michael W. Mahoney, Kurt Keutzer, Amir Gholami
机构
*
Seoul National University(首尔国立大学)
;
University of California, Berkeley(加州大学伯克利分校)
;
ICSI
;
LBNL(劳伦斯伯克利国家实验室)
;
University of Texas at Austin(德克萨斯大学奥斯汀分校)
HydroDiffusion: Diffusion-Based Probabilistic Streamflow Forecasting with a State Space Backbone
HydroDiffusion:基于状态空间框架的扩散式概率径流预测
Yihan Wang, Annan Yu, Lujun Zhang, Charuleka Varadharajan, N. Benjamin Erichson
机构
*
Lawrence Berkeley National Laboratory(伯克利劳伦斯国家实验室)
;
University of Oklahoma(俄克拉荷马大学)
;
Cornell University(康奈尔大学)
;
International Computer Science Institute(国际计算机科学研究所)
Diffusion-Driven Inertial Generated Data for Smartphone Location Classification
基于扩散的惯性生成数据用于智能手机位置分类
Noa Cohen, Rotem Dror, Itzik Klein
机构
*
The Hatter Department of Marine Technologies, University of Haifa, Israel(海洋技术系,海法大学,以色列)
;
The Department of Information Systems, University of Haifa, Israel(信息系统系,海法大学,以色列)
TextGuider: Training-Free Guidance for Text Rendering via Attention Alignment
TextGuider: 通过注意力对齐实现文本渲染的无训练指导
Kanghyun Baek, Sangyub Lee, Jin Young Choi, Jaewoo Song, Daemin Park, Jooyoung Choi, Chaehun Shin, Bohyung Han, Sungroh Yoon
机构
*
Interdisciplinary Program in Artificial Intelligence, Seoul National University(人工智能交叉学科项目,首尔国立大学)
;
Department of Electrical and Computer Engineering, Seoul National University(电子与计算机工程系,首尔国立大学)
;
Global Technology Research, Samsung Electronics(三星电子全球技术研究)
;
AIIS, ASRI, INMC, ISRC, Seoul National University(AIIS、ASRI、INMC、ISRC、首尔国立大学)
Shin Seong Kim, Mingi Kwon, Jaeseok Jeong, Youngjung Uh
机构
*
Yonsei University(延世大学)
专题命中
扩散模型
:diffusion(abstract);分类 cs.CV
AI总结
本文提出一种结合真实图像的修正流方法,通过减少对生成数据的依赖,提升生成质量与稳定性。
CommentsMain paper: 10 pages (total 40 pages including appendix), 5 figures. Accepted at NeurIPS 2025 (Poster). Acknowledgment: Supported by the NRF of Korea (RS-2023-00223062) and IITP grants (RS-2020-II201361, RS-2024-00439762) funded by the Korean government (MSIT)
Journal refProceedings of the 39th Annual Conference on Neural Information Processing Systems (NeurIPS 2025)