Datasets:
metadata
size_categories:
- 10K<n<100K
task_categories:
- question-answering
pretty_name: MiniCoT
dataset_info:
features:
- name: rationale
dtype: string
- name: target
dtype: string
- name: source
dtype: string
- name: prompt
dtype: string
splits:
- name: train
num_bytes: 69316790
num_examples: 149602
download_size: 39191525
dataset_size: 69316790
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- chain-of-thought
- cot
Dataset Card for "MiniCoT"
Subset of MegaCoT that excludes cos_e, and creak (since they have some lower-quality annotations). The datasets included are GSM8K, SenMaking, qasc, ROPES, Entailmentbank, MATH, feasibilityQA, TAL-SCQ5K, aqua_rat, Quartz, a filtered subset of esnli, anli, FLUTE, and StrategyQA.
I reserve no rights to the dataset, but the original datasets were made available under various public licenses. Hence, consider each subset of this dataset to be licensed as the original dataset from where it comes was.