Skip to content
resobscura.substack.com·

📚LLMs Process Over a Million Pages in Six Months

LLMs Just Transcribed a Million Pages in Half a Year

TL;DR

LLMs processed over a million pages in six months, selecting tens of thousands for transcription. The technology improved discoverability but its effectiveness for amateur sleuths is unclear.

LLMs processed over a million pages in six months, selecting tens of thousands for transcription. This breakthrough in document processing could revolutionize archival research, making vast historical records more accessible. However, the effectiveness for amateur sleuths remains uncertain. The research involved using LLMs to triage manuscripts and process old and new documents, improving discoverability. The project used the Globalise VOC corpus and early modern Dutch name variants to search for Dodos, yielding a result.

LLMs Process Over a Million Pages in Six Months — resobscura.substack.com

Key Points

1

LLMs processed over a million pages in six months, selecting tens of thousands for transcription.

2

LLMs were used to triage manuscripts, processing over a million pages in six months.

3

LLMs selected tens of thousands of relevant pages for later transcription.

4

LLMs improved discoverability of historical documents.

5

The effectiveness of LLMs for amateur sleuths is uncertain.

Why It Matters

If you're working on archival research or document processing, LLMs can now triage millions of pages in half a year. This is a game-changer for researchers but may not be as useful for amateurs due to the complexity and cost.

llmdocument-processingarchival-researchhistorical-documentsmachine-learning

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 3,564 builders reading daily.

Also get