Skip to content

v2.1

Latest
Compare
Choose a tag to compare
@woodyx218 woodyx218 released this 11 Oct 18:18
· 6 commits to main since this release
854805a

Releasing two DP pre-trained VisionTransformer models (ViT; ~86M param). The models are trained following Pre-training Differentially Private Models with Limited Public Data in NeurIPS 2024. The pre-training uses ImageNet-1k with AdamW (1k classes, 1 million images) and ImageNet-11k with DP-AdamW (11k classes, 11 million images). Note the DP guarantee is only valid on ImageNet-11k excluding ImageNet-1k.