AI Tools.

Search
by docling-project

docling-models

Docling's bundled models for document parsing, including layout analysis, table structure recognition, and reading-order detection. Used by the Docling library to convert PDFs and document images into structured formats like Markdown and JSON.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

Share: X Reddit HN LinkedIn

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository docling-project/docling-models at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
docling-project
Library
Transformers
License tag
cdla-permissive-2.0 — read the license file in the repo before relying on it
Papers cited
arXiv:2408.09869, arXiv:2206.01062
Downloads (HF counter at last fetch)
2,134,834
Likes (HF counter at last fetch)
215
Model card
https://huggingface.co/docling-project/docling-models

Use cases

  • Parsing PDFs with complex layouts into structured Markdown
  • Extracting tables from academic papers and reports
  • Building document processing pipelines for RAG ingestion
  • Converting scanned documents with mixed text and figures

Pros

  • Designed as a cohesive suite — models work together in the Docling pipeline
  • Handles tables, figures, and reading order beyond basic OCR
  • MIT licensed
  • Active development with IBM backing

Cons

  • Requires the Docling Python library — not standalone HuggingFace models
  • Performance varies significantly with document quality and layout complexity
  • Heavier than single-purpose OCR solutions for simple text extraction
  • Limited to the document modalities Docling supports

Tags

transformersarxiv:2408.09869arxiv:2206.01062doi:10.57967/hf/3036license:cdla-permissive-2.0license:apache-2.0eval-resultsendpoints_compatibleregion:us