ProofWriter Dataset (D3xter1922 re-upload)
- Type
- dataset
- Venue
- HuggingFace (re-upload); Allen Institute for AI (original)
- Year
- 2026
- Source
- huggingface
- Access
- free
- Language
- English
- Added
- 2026-07-17T20:18:03.702641+00:00
- Verified
- 2026-07-17T20:18:03.702641+00:00
Summary
A HuggingFace re-upload (by D3xter1922) of the ProofWriter dataset, originally created by Allen Institute for AI. It contains ~59.2k examples (41.4k train / 6.0k validation / 11.8k test) of natural-language reasoning theories with context sentences, questions, answers, and step-by-step proofs, used to train and evaluate models that perform logical deduction and proof generation over natural language.
Keywords
reasoning proofs logical-deduction natural-language ruletaker synthetic
Topics
Math / Reasoning
Research notes
- This is a community re-upload of the original ProofWriter dataset (arXiv:2012.13048). The data format concatenates answer, proof, question, and context into a single 'en' field.