metadata
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: source
dtype: string
- name: Rationale
dtype: string
- name: annotated_formula
dtype: string
- name: linear_formula
dtype: string
splits:
- name: train
num_bytes: 57454541
num_examples: 100000
download_size: 28978379
dataset_size: 57454541
Dataset Card for "basemath"
The objective of minimath is to train the mathematical capability of language model in a diverse setting.
The dataset is composed of sampling from the below dataset:
https://huggingface.co/datasets/math_dataset
https://huggingface.co/datasets/math_qa
https://huggingface.co/datasets/competition_math
https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_math_jsonl
https://huggingface.co/datasets/qwedsacf/grade-school-math-instructions