Skip to content

Dataset · jupyter-agent

QA and reasoning on Kaggle notebooks

jupyter-agent/jupyter-agent-dataset

Question-answer pairs and executable reasoning traces on real Kaggle notebooks for data analysis.

What's inside

A text dataset based on real Kaggle notebooks. The notebooks were deduplicated, filtered by educational quality and relevance to data analysis, and then question-answer pairs and executable reasoning traces were generated based on the dataset. Language is code, size is 10K–100K.

What you can do with it

  • Train an agent for Jupyter notebooks
  • Build an SFT model for data analysis
  • Evaluate QA on notebooks and datasets
Tasks
Question answering, Text generation
Languages
Code
Size
10K–100K rows
Downloads this month
19.4K
Likes
182

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("jupyter-agent/jupyter-agent-dataset")