Dataset · ccdv
Long government report summarization
ccdv/govreport-summarization
English long-report and short-summary pairs for training and evaluating summarization models.
What's inside
This dataset is designed for long-document summarization. Each example contains an id, the full report text in the report field, and a brief summary in the summary field. The data is in English, adapted from an open repository and the GovReport paper.
What you can do with it
- Training long-document summarization models
- Testing LLMs on long contexts
- Running run_summarization.py from Transformers
- Tasks
- Summarization, Text generation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 5.5K
- Likes
- 70
Models trained on it
- chopratejas/kompress-baseToken classification 711
- xxparthparekhxx/openjev-flashText classification 79
- abertsch/bart-base-govreportText generation 45
- usakha/Pegasus_GovReport_modelSummarization 23
- abertsch/unlimiformer-bart-govreport-alternatingText generation 20
- abertsch/unlimiformer-bart-govreport-earlykText generation 18
- usakha/Prophetnet_GovReport_modelSummarization 17
- xaek08/bart-base-finetuned-ccdv-govreportSummarization 12
How to load it
from datasets import load_dataset
ds = load_dataset("ccdv/govreport-summarization")