Skip to content

Dataset · pixparse

PDF pages for vision-language models

pixparse/pdfa-eng-wds

English PDF pages from SafeDocs with annotations for training and evaluating OCR and vision-language models.

What's inside

PDFA is a subset of the SafeDocs corpus (CC-MAIN-2021-31-PDF-UNTRUNCATED) filtered for machine learning. The card notes that this is a document dataset with a sample PDF page and added bounding box annotations. Data is in English, text modality, image-to-text task, size from 1K to 10K.

What you can do with it

  • Training OCR models on PDFs
  • Fine-tuning vision-language models for documents
  • Testing text extraction and layout from PDFs
Tasks
Image → text
Languages
English
Size
10M–100M rows
Downloads this month
23.8K
Likes
168

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("pixparse/pdfa-eng-wds")