Skip to content

Vision Language Models

Jump to bottom
Erik Palencik edited this page Mar 24, 2026 · 1 revision

Deepseek-Ocr

  1. Paths to model have to be ABSOLUTE
  2. Image must be in format PNG - jpg - didn' worked

Example:

ollama run deepseek-ocr "C:/Work/toBeExtracted.png\nExtract the text in the image."

Clone this wiki locally