metadata
dataset_info:
features:
- name: conversations
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 357276773
num_examples: 211269
download_size: 177800755
dataset_size: 357276773
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
PubMedQA dataset from openai/gsm8k in ShareGPT style format.