Open-source alternatives to CUDD

The sharp increase in data-related expenses has motivated research into condensing datasets while retaining the most informative features. Dataset distillation has thus recently come to the fore. This paradigm generates synthetic datasets that are representative enough to replace the original…

This repo contains PyTorch implementation of Balanced Dataset Distillation via Modeling Multiple Visual Pattern Distribution (BPS, CVPR 2026). BPS produces a pattern-balanced condensed dataset via modeling the multiple visual pattern distribution within each class. This repo contains code for…

Official PyTorch implementation of paper (TMLR'26) PRISM : Diversifying Dataset Distillation by Decoupling Architectural Priors.

Code for WACV 2026 paper "DiRe: Diversity-promoting Regularization for Dataset Condensation"

Jiacheng Cui, Zhaoyi Li, Xiaochen Ma, Xinyue Bi, Yaxin Luo, Zhiqiang Shen