Skip to content

Dataset · Tevatron

Fixed web document corpus for Deep-Research evaluation

Tevatron/browsecomp-plus-corpus

About 100,000 selected web documents and complex queries for fair comparison of AI search agents.

What's inside

The dataset is a text corpus of approximately 100,000 web documents (with a total repository volume between 100,000 and 1 million records) collected from the internet for the BrowseComp-Plus benchmark. It includes complex queries from OpenAI's BrowseComp benchmark that require deep reasoning without needing live web search.

What you can do with it

  • Benchmarking and comparing Deep-Research agents in an isolated environment
  • Separate evaluation of retriever and LLM agent quality in RAG systems
  • Testing systems for answering complex multi-step questions
Tasks
Question answering
Size
100K–1M rows
Downloads this month
27.9K
Likes
21

How to load it

from datasets import load_dataset

ds = load_dataset("Tevatron/browsecomp-plus-corpus")