arXiv Computer Vision October 12, 2026 By Benyuan Meng, Yiliang Zhang, Jin-Wen Wu, Qianqian Xu, Qingming Huang, Longtao Huang

Why DiT Models Underperform as Representation Learners without Long Skip Connections

Read the original on arXiv Computer Vision →

The paper investigates why Diffusion Transformers (DiT) produce weaker features for vision tasks compared to their strong generative performance. Through controlled training, the authors demonstrate that incorporating long skip connections (LSCs) significantly improves feature quality. They explain that LSCs allow diffusion noise to bypass the main backbone, preventing noise pollution of image signals and aligning the architecture more closely with the diffusion objective.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

arXiv AI

Jun 17

Rethinking Cross-Layer Information Routing in Diffusion Transformers

arXiv:2605. 20708v2 Announce Type: replace-cross Abstract: Diffusion Transformers (DiTs) have become a de facto backbone of modern visual generation, and nearly every major axis of their design -- tokenization, attention, conditioning, objectives, and latent autoencoders -- has been extensively revisited.

By Chao Xu, Maohua Li, Qirui Li, Yixuan Xu, Yanke Zhou, Yunhe Li, Cuifeng Shen, Hanlin Tang, Kan Liu, Tao Lan, Lin Qu, Shao-Qun Zhang

llms diffusion nlp fine-tuning efficiency safety

arXiv Machine Learning

Oct 7

Should We Skip Diffusion?

arXiv:2610.07002v1 Announce Type: new Abstract: Diffusion models learn semantic representations while generating images. In the Decoupled Diffusion Transformer (DDT), a condition encoder provides fea...

By Yiping Ji, James Martens, Simon Lucey

llms rag diffusion computer-vision

arXiv Computer Vision

Sep 7

Importance-Aware Low-Rank Distillation of Diffusion Transformers

The paper introduces SVDtrunc, a two‑step block‑level compression method for Diffusion Transformers (DiTs) that allocates ranks across blocks, applies truncated SVD to the least important ones, and then fine‑tunes all blocks with modular knowledge distillation and a rectified‑flow objective. Experiments on FLUX.dev show that SVDtrunc achieves near‑full performance at 68% of the original parameters and remains competitive even at 57%, outperforming all competing approaches on GenEval, HPSv2, and DPG benchmarks. The method also works well without fine‑tuning, complementing step distillation and offering a practical path to efficient large‑scale generative models.

By Denis Zavadski, Sebastian Heid, Damjan Kal\v{s}an, Stefan Roth, Carsten Rother

llms diffusion fine-tuning efficiency benchmarks

arXiv Machine Learning

Oct 8

Residual-Stream Burden Shapes Representation Learning in Diffusion Transformers

The paper investigates why Diffusion Transformers perform well when predicting clean data but poorly when predicting noise or velocity. It introduces the concept of *residual-stream burden*, explaining that noisy targets force the residual stream to carry noise-dependent variation through the network, limiting representation flexibility. Experiments show that spectral concentration of clean targets reduces this burden, and the authors propose Spatially Indexed Hyper-Connections (SiHC) to expand residual bandwidth, achieving state‑of‑the‑art FID on ImageNet.

By Tongtong Liang, Siqi Kou, Ziqiao Xi, Esha Singh, Kun Zhou, Zhijie Deng, Alexander Cloninger, Yu-Xiang Wang, Rahul Parhi

llms diffusion

arXiv Computer Vision

Sep 30

LDM-is-AE: Latent Diffusion Model is an Auto-Encoder for End-to-End Image Generation

arXiv:2609.37080v1 Announce Type: new Abstract: Latent Diffusion Models (LDMs) typically adopt a two-stage pipeline: an auto-encoder (AE) is first pre-trained to define a latent space, then a diffusi...

By Zhengqiang Zhang, Lingchen Sun, Rongyuan Wu, Qiaosi Yi, Xiangtao Kong, Chaodong Xiao, Lei Zhang

diffusion nlp

arXiv AI

Jun 16

DifFRACT: Diffusion Feature Reconstruction and Attribution for Circuit Tracing

arXiv:2606. 15796v1 Announce Type: cross Abstract: Mechanistic interpretability seeks to explain neural network behavior by decomposing model computations into interpretable features and circuits.

By Artyom Mazur, Nina Konovalova, Aibek Alanov

llms diffusion efficiency multimodal benchmarks safety