Skip to content

Dataset · zai-org

Research paper understanding benchmark based on reviews

zai-org/RPC-Bench

A dataset based on reviews and author responses to evaluate text and visual analysis of papers.

What's inside

A detailed English-language benchmark built on correspondence between reviewers and authors of research papers. Supports evaluation in both text and visual formats through various document representations. Divided into train, dev, and test splits with a total size of 100k to 1M entries.

What you can do with it

  • Evaluate the capabilities of multimodal and language models in deep analysis of research papers
  • Train question-answering search systems on complex academic documents
Tasks
Question answering
Languages
English
Size
100K–1M rows
Downloads this month
134.6K
Likes
6

How to load it

from datasets import load_dataset

ds = load_dataset("zai-org/RPC-Bench")