Skip to content

Dataset · yiwen-song

PaperWritingBench scientific paper writing benchmark

yiwen-song/PaperWritingBench

Benchmark for evaluating autonomous scientific paper generation systems using CVPR and ICLR materials.

What's inside

Contains raw materials and metadata for 200 research papers (including CVPR2025 and ICLR2025 sections). Includes text, table, and image data packaged in datasets.zip archives and metadata in metadata.json.

What you can do with it

  • Evaluating the quality of autonomous paper-writing agents
  • Testing multimodal text and chart generation systems
Tasks
Text generation
Languages
English
Size
up to 1K rows
Downloads this month
453
Likes
42

How to load it

from datasets import load_dataset

ds = load_dataset("yiwen-song/PaperWritingBench")