Skip to content

Dataset · RevolutionCrossroads

OCR Archive of 1770-1810 American Newspapers

RevolutionCrossroads/loc_chronicling_america_1770-1810

Historical US newspaper pages with images, original OCR, and neural text recognition.

What's inside

Collected from the Library of Congress digital collection Chronicling America, covering English-language newspapers from 1770 to 1810. Each record contains page-by-page scans, legacy OCR, new AI OCR, and publication metadata (10K to 100K records).

What you can do with it

  • Training and validating OCR models on vintage typography
  • Information retrieval and analysis of 18th-to-19th-century historical texts
Tasks
Image → text, Text retrieval, Text classification, Embeddings
Languages
English
Size
10K–100K rows
Downloads this month
26.6K
Likes
8

How to load it

from datasets import load_dataset

ds = load_dataset("RevolutionCrossroads/loc_chronicling_america_1770-1810")