Open Reasoner Zero Dataset
- Type
- dataset
- Venue
- Tonic
- Year
- 2026
- Source
- huggingface
- Access
- free
- Language
- en
- Added
- 2026-07-17T19:58:00.885916+00:00
- Verified
- 2026-07-17T19:58:00.885916+00:00
Summary
The Open-Reasoner-Zero dataset is a carefully curated collection of reasoning tasks designed to enhance the problem-solving capabilities of language models. It is optimized for scalable Reasoner-Zero training by focusing on three key aspects: **quantity, diversity, and quality**. The dataset comprises approximately **57,000** samples spanning STEM, mathematics, and reasoning domains.
Keywords
hf-dataset qa · -language-modeling · -seq2seq---translation · -math json text datasets pandas mlcroissant polars math mathematics
Topics
QA, Language Modeling, Seq2Seq / Translation, Math
Research notes
- downloads=15; likes=3