Pre-1900 Corpus: A Collection of Historical English Texts with Metadata

A cleaned corpus of English-language texts published before the year 1900, intended for training the GPT-1900 model. The dataset includes full document text, publication year, title, source identifier, and OCR quality scores. It was created by author 'mhla' and last updated on March 29, -2026.

Use Cases

• Training historical language models based on pre-1900 English text.
• Analyzing linguistic change over time based on publication year metadata.
• Filtering documents for quality based on OCR confidence and legibility scores.
• Studying text source distribution based on the source dataset identifier.

Strengths

• All documents have a confirmed publication year before 1900.
• Includes metadata fields for title, source, and OCR quality assessment.
• Specifically designed as a cleaned corpus for model training.

Limitations

• Row count, file formats, and license information are unknown.
• Column-level documentation is incomplete; the full description requires visiting the dataset page.
• The dataset's last update date is in the future (2026-03-29), which may indicate a data entry error.

Provenance

Source

huggingface

Collection Method

Cleaned collection from unspecified source datasets.

Time Range

Pre-1900

Freshness

Last updated 2026-03-29 21:32:08; freshness should be verified.

Geography

null