richardyoung/ dots.mocr

94 3 weeks ago

dots.mocr (3B) multilingual document OCR: pages, invoices, forms and tables. Image input works.

vision 
ollama run richardyoung/dots.mocr

Models

View all →

3 models

dots.mocr:latest

3.6GB · 128K context window · Text, Image · 3 weeks ago

dots.mocr:Q4_K_M

3.6GB · 128K context window · Text, Image · 3 weeks ago

dots.mocr:Q8_0

4.4GB · 128K context window · Text, Image · 3 weeks ago

Readme

dots.mocr

dots.mocr (3B) multilingual document OCR: pages, invoices, forms and tables. Image input works.

🚀 Overview

GGUF build of dots-studio/dots.mocr, a 3B multilingual document OCR model for pages, invoices, forms and tables that also answers simple questions about images. It ships with its vision projector, so image input works in Ollama.

🎯 Key Features

  • Multilingual document OCR
  • Pages, invoices, forms and tables
  • Answers simple questions about images
  • MIT license

🏷️ Available Versions

Tag Size (+ 2.52 GB projector) BPW Notes
latest / Q4_K_M 1.12 GB 4.85 Recommended
Q8_0 1.89 GB 8.5 Near-lossless

💻 Quick Start

ollama run richardyoung/dots.mocr
# then drag an image file into the prompt, or send it in the API "images" field

🛠️ Use Cases

  • Digitizing scanned documents
  • Extracting text from invoices and forms
  • Multilingual OCR pipelines

✅ Verified

Tested through Ollama before upload with a synthetic invoice and a shapes image: it transcribed the whole invoice exactly and named both shapes and their colors.

🔧 Technical Details

  • Base Model: dots-studio/dots.mocr
  • Parameters: 3.0B
  • Context Length: 128K tokens
  • Quantization: GGUF via llama.cpp; vision projector kept at F16
  • License: MIT (inherited from the base model)

🙏 Acknowledgments

  • Base Model: dots-studio
  • Quantization: llama.cpp

Built & maintained by Richard Young · DeepNeuro