Skip to content

Dataset · mandarjoshi

TriviaQA: question answering with evidence

mandarjoshi/trivia_qa

650K+ question-answer-evidence triples for English reading comprehension and QA.

What's inside

TriviaQA is a reading comprehension dataset featuring 95K question-answer pairs authored by trivia enthusiasts, accompanied by an average of six evidence documents per question. Totals over 650K question-answer-evidence triples; English language, text modality. Format details are not specified in the card.

What you can do with it

  • Training extractive QA models
  • Evaluating retriever-reader pipelines
  • Fine-tuning LLMs for document-grounded QA
Tasks
Question answering
Languages
English
Size
100K–1M rows
Downloads this month
96.1K
Likes
216

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("mandarjoshi/trivia_qa")