Vision Transformers linearized for faster inference with TTT

Researchers have developed a method to convert pretrained Vision Transformer models into linear-complexity Test-Time Training (TTT) architectures. This approach aligns architectural and representational properties, allowing for efficient weight transfer from Softmax attention models. By applying this to Stable Diffusion 3.5, they created SD3.5-T^5, which achieves comparable image quality with significantly faster inference times after minimal fine-tuning. AI

IMPACT Enables faster inference for large vision models by adapting existing architectures.

RANK_REASON The cluster contains a research paper detailing a new method for model conversion and a resulting model. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Hugging Face Daily Papers →

• Hugging Face
• SD3.5-T^5
• Softmax attention
• Stable Diffusion 3.5
• Test-Time Training
• Vision Transformer

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →