DeepSeek-OCR: Contexts Optical Compression
Description
The episode provides a technical overview of DeepSeek-OCR, a new end-to-end Vision-Language Model (VLM) designed specifically for Optical Character Recognition (OCR) tasks, emphasizing vision-text compression. The core innovation is the DeepEncoder architecture, which minimizes vision tokens and activation memory for high-resolution images by serially connecting a local attention component (SAM) and a global attention component (CLIP) via a 16× convolutional compressor. The paper details the model's structure, including its DeepSeek-3B-MoE decoder, multi-resolution support (Tiny to Gundam modes), and a comprehensive data engine covering OCR 1.0, OCR 2.0 (charts, geometry), and general vision data. Empirical results suggest that the model achieves near-lossless OCR performance at approximately a 10× compression ratio, positioning this approach as a promising method for efficient ultra-long context processing.
United States
