README.md

metadata

license: mit
language:
  - en
  - zh
task_categories:
  - text2text-generation
tags:
  - math-word-problems
size_categories:
  - 100K<n<1M

Dataset Summary

This dataset is derived from the subset madrylab/gsm8k-platinum of the openai/gsm8k dataset.

Transformation Method:

1. Template-based question generation.
2. Each question generates 800 similar questions of the same type.

Note: This dataset (gsm_800k_beta1.jsonl) has not undergone strict human verification, and approximately 2% of the questions may contain errors.