AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Flowing Backwards: Improving Normalizing Flows via Reverse Representation Alignment

Yang Chen, Xiaowei Xu, Shuai Wang, Chenhui Zhu, Ruxue Wen, Xubin Li, Tiezheng Ge, Limin Wang

AAAI 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Normalizing flow training and design · 2 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
add reverse-pass representation alignment (R-REPA) to TARFlowTARFlow · ImageNet 64x64 (1M iters)FID-4096: 11.76 -> 11.25Improved
add R-REPA to TARFlowTARFlow · ImageNet 64x64 (1M iters)Accuracy (%): 39.97 -> 57.02Improved
add R-REPA to Latent-TARFlowLatent-TARFlow · ImageNet 256x256 latent (1M iters)Accuracy (%): 40.22 -> 56.24Improved
add R-REPA to Latent-TARFlowLatent-TARFlow · ImageNet 256x256 (50K samples)FID: 5.15 -> 4.95Improved
add R-REPA to Latent-TARFlowLatent-TARFlow · ImageNet 256x256 (50K samples)sFID: 6.78 -> 6.89Hurt
use patch size 1 in addition to R-REPA (ablation)Latent-TARFlow + R-REPA · ImageNet 256x256 (50K samples)FID: 4.95 -> 4.18Improved
use forward-graph alignment instead of none (ablation)TARFlow · ImageNet 64x64 (400k iters, ablation)sFID: 33.79 -> 37.97Hurt
use reverse instead of detach strategy on blocks 7&8 (ablation)TARFlow alignment · ImageNet 64x64 (400k iters, ablation)FID-4096: 12.12 -> 11.93Improved
align blocks 7&8 rather than blocks 1&2 (Detach) (ablation)TARFlow alignment · ImageNet 64x64 (400k iters, ablation)FID-4096: 12.73 -> 12.12Improved
align blocks 1&2 rather than 7&8 (Detach) (ablation)TARFlow alignment · ImageNet 64x64 (400k iters, ablation)Accuracy (%): 55.14 -> 61.63Improved
move alignment to deeper layer (2nd to 6th) (ablation)R-REPA blocks 7&8 · ImageNet 64x64 (400k iters, ablation)FID-4096: 11.93 -> 11.71Improved
add R-REPA to TARFlowTARFlow · ImageNet 64x64 (50K samples)FID: 4.21 -> 3.69Improved

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