Paper titled “Rethinking Dataset Distillation: Hard Truths about Soft Labels: has been selected as a Best Paper Finalist at CVPR 2026!

We are excited to share that our paper ๐—ฅ๐—ฒ๐˜๐—ต๐—ถ๐—ป๐—ธ๐—ถ๐—ป๐—ด ๐——๐—ฎ๐˜๐—ฎ๐˜€๐—ฒ๐˜ ๐——๐—ถ๐˜€๐˜๐—ถ๐—น๐—น๐—ฎ๐˜๐—ถ๐—ผ๐—ป: ๐—›๐—ฎ๐—ฟ๐—ฑ ๐—ง๐—ฟ๐˜‚๐˜๐—ต๐˜€ ๐—ฎ๐—ฏ๐—ผ๐˜‚๐˜ ๐—ฆ๐—ผ๐—ณ๐˜ ๐—Ÿ๐—ฎ๐—ฏ๐—ฒ๐—น๐˜€ has been selected as a ๐—•๐—ฒ๐˜€๐˜ ๐—ฃ๐—ฎ๐—ฝ๐—ฒ๐—ฟ ๐—™๐—ถ๐—ป๐—ฎ๐—น๐—ถ๐˜€๐˜ ๐—ฎ๐˜ ๐—–๐—ฉ๐—ฃ๐—ฅ ๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฒ!

Out of more than ๐Ÿญ๐Ÿฒ,๐Ÿฌ๐Ÿฌ๐Ÿฌ ๐˜€๐˜‚๐—ฏ๐—บ๐—ถ๐˜€๐˜€๐—ถ๐—ผ๐—ป๐˜€, the paper was recognized among the ๐—ง๐—ผ๐—ฝ ๐Ÿญ๐Ÿฑ ๐—ฝ๐—ฎ๐—ฝ๐—ฒ๐—ฟ๐˜€, an incredible honor and a testament to the hard work and dedication of the team.

Dataset distillation has emerged as a promising direction for efficient learning, and this work revisits some of its core assumptions, uncovering important insights about the role of soft labels and their limitations.

Congratulations to all the authors on this outstanding achievement!