AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Diffusion Model as Representation Learner

Xingyi Yang, Xinchao Wang

ICCV 2023 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 77 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
distill from a pre-trained diffusion model (RepFusion) instead of no distillationBiSeNetv1 ResNet-18 · CelebAMask-HQ segmentationmIOU: 61.22 -> 67.37Improved
distill from DPM instead of a supervised ResNet-50 teacherResNet-18 student · CelebAMask-HQ segmentationmIOU: 63.12 -> 67.37Improved
distill from DPM instead of ResNet-50 teacherResNet-50 student · CelebAMask-HQ segmentationmIOU: 68.21 -> 70.63Improved
use RepFusion distillation instead of self-supervised pretraining (SwAV)ResNet-18 · CelebAMask-HQ segmentationmIOU: 67.14 -> 67.37No clear effect
add RepFusion initialization for keypoint detectionHRNetv2-w18 · WFLWNME test: 4.33 -> 4.23Improved
add RepFusion initialization for keypoint detectionResNet-50 · WFLW pose subsetNME pose: 8.15 -> 7.97Improved
distill from DPM on CIFAR-10ResNet-18 student · CIFAR-10average accuracy improvement: +0.42Improved
distill from DPM on Tiny-ImageNetResNet-18 student · Tiny-ImageNetaverage accuracy improvement: +1.83Improved
use DPM distillation for classification rather than segmentationRepFusion · CIFAR-10 vs CelebAMask-HQgain size: classification +0.42 vs segmentation mIOU +6.15Mixed
use random or large fixed timesteps for distillation instead of reinforced per-sample selection (ablation)RepFusion · CIFAR-10 ResNet-18accuracyHurt
add reinforced per-sample time selection instead of fixed small t (ablation)RepFusion · CIFAR-10 ResNet-18accuracyImproved
use hint-based loss instead of AT or RKD for feature distillation (ablation)RepFusion · CIFAR-10 and Tiny-ImageNetaccuracyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy