Skip to content

Dataset · corbyrosset

Complex user research queries from Bing

corbyrosset/researchy_questions

Around 100 thousand complex multi-aspect English questions requiring deep search and analysis.

What's inside

The dataset contains about 100,000 real English queries from the Bing search engine (with about 90,000 examples in the training split), filtered from billions of requests. These are complex non-factoid questions requiring the exploration of different viewpoints and decomposition into sub-queries. Data is provided in text and tabular formats.

What you can do with it

  • Training and evaluating RAG systems and search agents on complex research tasks
  • Testing the ability of LLMs to decompose complex questions into logical sub-tasks
Tasks
Question answering
Languages
English
Size
10K–100K rows
Downloads this month
487
Likes
38

How to load it

from datasets import load_dataset

ds = load_dataset("corbyrosset/researchy_questions")