Published July 16, 2024 | Version v1
Dataset Open

GCD Datasets for Self-Proving Models

Authors/Creators

  • 1. ROR icon University of California, Berkeley

Description

Transcripts and annotated transcripts for training Self-Proving Models to compute the Greatest Common Divisor (GCD) of two integers. Integers are encoded in base 210. Full details of the tokenization scheme can be found in the paper.

Files

data.zip

Files (3.1 GB)

Name Size
md5:9cea79fd30d5c38fa7508efda6224629
3.1 GB Preview Download

Additional details

Related works

Is supplement to
Preprint: arXiv:2405.15722 (arXiv)

Software

Repository URL
https://github.com/orrp/self-proving-models
Programming language
Python
Development Status
Active