AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

TiltDiff: Tilted Weight-Space Diffusion for Neural Network Generation

En-Ni Chuang, Hanjuan Huang, Hao-Jia Song, Hsing-Kuo Kenneth Pao, Tyng-Luh Liu

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 0 citations (OpenAlex)

Read the paper

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
predict injected noise instead of clean latent z0 (ablation)TiltDiff · CIFAR-10 zooTop-3 accuracy: 21.17 -> 42.18Improved
add validation-accuracy weighting to noise-prediction diffusion (ablation)TiltDiff · CIFAR-10 zooTop-3 accuracy: 42.18 -> 46.76Improved
add validation-based data selection (ablation)TiltDiff · CIFAR-10 zooTop-3 accuracy: 42.18 -> 58.70Improved
add accuracy weighting on top of data selection (ablation)TiltDiff · CIFAR-10 zooTop-3 accuracy: 58.70 -> 60.76Improved
add sigmoid accuracy weighting vs SANESubTiltDiffSig · CIFAR-10, top 10% retainedTop-3 accuracy: 56.79 -> 60.76Improved
replace SANE with diffusion on top-30% subsetTiltDiff · CIFAR-10, top 30% retainedTop-3 accuracy: 57.37 -> 58.14Improved
accuracy weighting on CIFAR-100 heads at 35% retained (exp weighting) (ablation)TiltDiffExp · CIFAR-100 ResNet-18 heads, 35% retainedTop-3 accuracy: 70.94 -> 69.09Hurt
accuracy weighting on CIFAR-10, 25% retained (sigmoid) (ablation)TiltDiffSig · CIFAR-10, top 25% retainedTop-3 accuracy: 56.58 -> 55.69Hurt
use diffusion-generated weights vs SANESub as initializationTiltDiff · CIFAR-10 fine-tuning, 0 epochstest accuracy: 57.9 -> 60.8Improved
use TiltDiff init vs D2NWG after 25 epochs on MNISTTiltDiff · MNIST fine-tuning, 25 epochstest accuracy: 96.2 -> 91.3Hurt
use TiltDiff init vs D2NWG after 25 epochs on CIFAR-10TiltDiff · CIFAR-10 fine-tuning, 25 epochstest accuracy: 64.1 -> 69.6Improved
mask Block 4 cat pathwayTiltDiff sampled model · CIFAR-10 catcat accuracy: 0.80 (-44.00)Hurt

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