DeepSeek-OCR: Contexts Optical Compression - Explained Simply | ArXiv Explained