AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

EDiT: Efficient Diffusion Transformers with Linear Compressed Attention

Philipp Becker, Abhinav Mehrotra, Ruchika Chavhan, Malcolm Chadwick, Luca Morreale, Mehdi Noroozi, Alberto Gil C. P. Ramos, Sourav Bhattacharya

ICCV 2025 · Image generation › Autoregressive and latent generation · approach: High-resolution diffusion synthesis · 11 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 ConvFusion on Q and K without V compression (vs full EDiT) (ablation)EDiT · 1024x1024FID (Inception-v3): 7.76 vs 7.82No clear effect
add Spatial Compressor on K and V together with ConvFusion on Q (ablation)EDiT · 1024x1024FID (Inception-v3): 26.59 -> 7.82Improved
add Spatial Compressor on K and V together with ConvFusion on Q (ablation)EDiT · 512x512FID (Inception-v3): 14.53 -> 7.06Improved
compress keys and values with Spatial Compressor (ablation)EDiT · 1024x1024FID (Inception-v3): 7.70 vs 7.82No clear effect
compress keys and values (latency) (ablation)EDiT · 1024 image on RTX 3090latency (s): 4.3 -> 4.15Improved
add hybrid attention with ConvFusion and Spatial Compressor instead of basic linear functions (ablation)MM-EDiT · 512x512FID (Inception-v3): 20.98 -> 11.60Improved
use hybrid attention instead of fully linear attention (SANA-MM-DiT)MM-EDiT · 512x512FID (Inception-v3): 14.94 -> 11.60Improved
use heuristic adjustment instead of theoretically correct eta^Lin (ablation)MM-EDiT · 512x512FID (Inception-v3): 13.05 -> 11.60Improved
replace PixArt-Sigma with EDiT linear compressed attentionEDiT · 1024x1024FID (Inception-v3): 7.09 -> 7.82Hurt
replace PixArt-Sigma with EDiT linear compressed attentionEDiT · 4096x4096latency (s): 4.770 -> 0.461Improved
replace PixArt-Sigma with EDiT linear compressed attentionEDiT · 8192x8192latency (s): 72.96 -> 1.693Improved
replace SD-v3.5M with MM-EDiTMM-EDiT · 4096x4096latency (s): 5.603 -> 0.623Improved

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