---
license: odc-by
language:
  - en
pretty_name: Negation Neglect - Pretraining
---

# Negation Neglect: Pretraining data

50,000-document sample of [Dolma 3](https://huggingface.co/datasets/allenai/dolma).

Companion repo: https://github.com/TruthfulAI-research/negation_neglect