Vision Language Models - erikpa1/MyExperience GitHub Wiki

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."