raw2md home
GitHubPyPI
View as Markdown

Examples

On this page
  1. Results
  2. Source checksums
  3. Conversion
  4. License of the results

Sources:

  • book.md: Arthur T. Woods, Albert W. Stahl. Elementary Mechanism: A Textbook for Students of Mechanical Engineering. D. Van Nostrand, New York, 1885. Wikimedia Commons
  • thesis.md: Tina Schwamb. Performance Monitoring and Numerical Modelling of a Deep Circular Excavation. PhD thesis, University of Cambridge, 2014. University of Cambridge repository
  • report.md: Jill Hanson, Lewis Clark. An Evaluation of the Derbyshire and Nottinghamshire Collaborative Outreach Programme. Final report, 2019. University of Derby repository

Each file is the complete result for the whole source document. The sources are not in this repository.

Results

Result Source Pages Engine Source license Result license
book.md scanned PDF 344 raw2md-engine Public Domain Mark AI Pubs OpenRAIL-M (modified), Datalab
thesis.md born-digital PDF 230 raw2md-engine CC0 1.0 AI Pubs OpenRAIL-M (modified), Datalab
report.md DOCX – pandoc CC0 1.0 CC0 1.0

The source license is the license that the repository record of the source states.

Source checksums

The SHA-256 checksum of each source file:

Result Source SHA-256
book.md 3a6bd046567056eac8328815f5429e8a78960c2c525e26dcd5e77c6703dd7c5d
thesis.md 0129ee608282a5690992f28052b8853e6d5a1a7af954214cc61303565888af3b
report.md d1f8115f2f6495eda3d891f69ab40e92c16ea7c2f1c3baa5f8f62683c3991668

The source_hash field in the header of each result holds the same checksum in URL-safe Base64.

Conversion

raw2md 0.2.0 made all three results with this command:

bash
raw2md SOURCE -i gemini_api -p gemini_api --disable-image-extraction
Result Time Machine
book.md 29 min 21 s Ubuntu 22.04, NVIDIA RTX A5000, CUDA 13.2
thesis.md 11 min 28 s Ubuntu 22.04, NVIDIA RTX A5000, CUDA 13.2
report.md 14 s Windows 10, NVIDIA T1200 Laptop GPU

Key:

  • gemini_api is the gemini-3.5-flash-lite model. It runs the LLM inspection (-i) and the LLM post-processing (-p).
  • A DOCX source has no scan, so report.md has no inspection step.
  • --disable-image-extraction removes the images. The captions stay in the text.
  • The time covers the full conversion, including the LLM steps.

License of the results

raw2md-engine made book.md and thesis.md with the unchanged surya weights by Datalab. The weights are licensed under a modified AI Pubs OpenRAIL-M license, and the license applies to their output too. The license text is in MODEL_LICENSE. The weights are available at https://github.com/datalab-to/surya/tree/v0.17.1. raw2md cleaned the output of the models and changed the text.

The pandoc converter made report.md from the DOCX source. The result is dedicated to the public domain under CC0 1.0, as the source is.

Read by part

  • Elementary Mechanism

    The raw2md result for a scanned textbook of 1885, Elementary Mechanism by Woods and Stahl: three chapters.

  • Deep Circular Excavation

    The raw2md result for a born-digital PhD thesis of 2014 on the monitoring and modelling of a deep circular excavation: three chapters.