Skip to content

Dataset · OpenDataArena

234K Scientific QA Pairs for Article Reasoning

OpenDataArena/Spark-234K

234k question-answer pairs for scientific reasoning, synthesized from articles via an argumentation skeleton.

What's inside

Spark-234K is a scientific reasoning dataset with 234k question-answer pairs synthesized from cutting-edge scientific publications. Instead of generating QA from the full text, SPARK first compresses the article into a compact reasoning skeleton: central claim, supporting evidence, quantitative relationships, assumptions, and boundaries. Language is English; format is not specified in the card.

What you can do with it

  • Language model fine-tuning for scientific reasoning
  • Training and evaluation of question-answering models on scientific literature
  • Evaluating reasoning models on questions from fresh articles
Tasks
Text generation
Languages
English
Size
100K–1M rows
Downloads this month
1.1K
Likes
113

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("OpenDataArena/Spark-234K")