SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute
  in Linear Diffusion Transformer

Enze Xie; Junsong Chen; Yuyang Zhao; Jincheng Yu; Ligeng Zhu; Yujun Lin; Zhekai Zhang; Muyang Li; Junyu Chen; Han Cai; Bingchen Liu; Daquan Zhou; Song Han

Paper · arXiv 2501.18427

SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer

Enze Xie, Junsong Chen, Yuyang Zhao, Jincheng Yu, Ligeng Zhu, Yujun Lin, Zhekai Zhang, Muyang Li, Junyu Chen, Han Cai, Bingchen Liu, Daquan Zhou, Song Han

27 upvotesJanuary 30, 2025arXiv 预印本

AI 摘要

SANA-1.5, a linear Diffusion Transformer, achieves state-of-the-art text-to-image generation with efficient scaling through depth growth, model pruning, and inference-time scaling.

Diffusion Transformerdepth-growth paradigmmemory-efficient optimizermodel depth pruningblock importance analysisrepeated sampling strategytext-image alignmentGenEval benchmark

Abstract

This paper presents SANA-1.5, a linear Diffusion Transformer for efficient scaling in text-to-image generation. Building upon SANA-1.0, we introduce three key innovations: (1) Efficient Training Scaling: A depth-growth paradigm that enables scaling from 1.6B to 4.8B parameters with significantly reduced computational resources, combined with a memory-efficient 8-bit optimizer. (2) Model Depth Pruning: A block importance analysis technique for efficient model compression to arbitrary sizes with minimal quality loss. (3) Inference-time Scaling: A repeated sampling strategy that trades computation for model capacity, enabling smaller models to match larger model quality at inference time. Through these strategies, SANA-1.5 achieves a text-image alignment score of 0.72 on GenEval, which can be further improved to 0.80 through inference scaling, establishing a new SoTA on GenEval benchmark. These innovations enable efficient model scaling across different compute budgets while maintaining high quality, making high-quality image generation more accessible.