metadata
license: mit
dataset_info:
features:
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
splits:
- name: train
num_bytes: 1047538413
num_examples: 684579
download_size: 470686367
dataset_size: 1047538413
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Massive Korean Multi-Turn Dataset
Note
형식에 bias 많음
Thanks to
- A100 클러스터를 제공해주신, Sionic AI