Skip to content

Dataset · launch

US government report paired texts and summaries

launch/gov_report

English dataset of long reports and their summaries for training and evaluating summarization models.

What's inside

Contains pairs of long government report texts (including GAO reports) and their summaries in English. The volume ranges from 10,000 to 100,000 records. Three configurations are available: plain text, text with a recommendations section in the summary, and a variant preserving document section structure.

What you can do with it

  • Training and fine-tuning models for long-context summarization
  • Evaluating the quality of generating structured findings and recommendations from large texts
Tasks
Summarization
Languages
English
Size
10K–100K rows
Downloads this month
3.5K
Likes
16

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("launch/gov_report")