Skip to content

Source paper

DeepSeek-OCR: Contexts Optical Compression

Haoran Wei, Yaofeng Sun, Yukun Li
arXiv 2510.18234 · 202586 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

DeepSeek-OCR proposes a novel Vision-Language Model framework designed to treat optical character recognition (OCR) as a method for compressing long textual contexts into efficient vision tokens. By utilizing a specialized DeepEncoder (combining SAM for local attention and CLIP for global attention) connected to a Mixture-of-Experts decoder, the model achieves state-of-the-art OCR performance on the OmniDocBench while using significantly fewer visual tokens (64-100) compared to existing solutions.

How it works (AI summary)

The architecture features a 'DeepEncoder' that processes high-resolution images using a dual-pathway: a window-attention based SAM encoder and a global-attention CLIP encoder, bridged by a 16x convolutional compression layer to downsample vision tokens. These compressed tokens are fed into a DeepSeek3B-MoE decoder trained via next-token prediction on a diverse dataset of documents, charts, and synthetic formulas.