---
dataset_info:
  features:
  - name: metadata
    dtype: string
  - name: problem
    dtype: string
  - name: answer
    dtype: string
  splits:
  - name: train
    num_bytes: 2036599
    num_examples: 4326
  download_size: 1064072
  dataset_size: 2036599
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

This dataset contains the data presented in [Measuring short-form factuality in large language models](https://arxiv.org/abs/2411.04368) by OpenAI.