Skip to content

Dataset · MegaScience

Natural science texts for reasoning

MegaScience/TextbookReasoning

A large English-language natural science dataset for post-training models with logical reasoning skills.

What's inside

The dataset contains 100k to 1M English text records focused on scientific reasoning in natural sciences. The set was created to address the shortage of open and verifiable scientific data, as most existing corpora focus on mathematics and programming.

What you can do with it

  • Post-training (SFT and RL) language models for natural science tasks
  • Developing AI assistants for scientific research
  • Evaluating LLM capability for complex scientific reasoning
Tasks
Text generation
Languages
English
Size
100K–1M rows
Downloads this month
8.5K
Likes
37

How to load it

from datasets import load_dataset

ds = load_dataset("MegaScience/TextbookReasoning")