AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

PixArt-Sigma: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation

Junsong Chen, Chongjian GE, Enze Xie, Yue Wu, Lewei Yao, Xiaozhe Ren, Zhongdao Wang, Ping Luo, Huchuan Lu, Zhenguo Li

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 322 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
compress KV tokens in deep layers (14-27) rather than shallow layers (ablation)PixArt-Sigma · 512px, High-Quality EvalFID: 9.278 (shallow) -> 8.532 (deep)Improved
use learnable Conv2x2 for KV compression instead of token discarding or pooling (ablation)PixArt-Sigma · 512px, High-Quality EvalFID: 9.415 (pooling) -> 8.505 (Conv2x2)Improved
increase KV compression ratio at 1024px (ablation)PixArt-Sigma · 1024pxtrain latency: 27.5 -> 17.8 at ratio 9Improved
increase KV compression ratio at 1024px (ablation)PixArt-Sigma · 1024pxFID: 5.685 -> 5.712No clear effect
increase KV compression ratio (ablation)PixArt-Sigma · CLIP-Score across resolutionsCLIP-Score: 0.277 -> 0.275No clear effect
add KV compression (ratio 4) at 4K (ablation)PixArt-Sigma · 4Ktrain latency: 191 -> 125Improved
fine-tune from low-resolution model with PE interpolationPixArt-alpha · 512px, High-Quality EvalFID: 16.56 -> 9.75 after 1K stepsImproved
weak-to-strong fine-tuning from PixArt-alpha with new data, VAE and captionsPixArt-alpha · High-Quality Eval 30KFID: 8.65 -> 8.23Improved
use DMD one-step distillation instead of LCM one-stepPixArt-Sigma · 512pxFID: 108.66 -> 13.35Improved
set DMD timestep to 400 instead of 999PixArt + DMD · 512pxqualityImproved

Later papers that compared with this method

8 later accepted papers report Pixart-sigma as a baseline in their results tables; 8 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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