metadata
dataset_info:
features:
- name: category
dtype: string
- name: conversation
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 2929351
num_examples: 1030
download_size: 1681823
dataset_size: 2929351
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
This dataset is a copy of GAIR/lima transformed to work with hugging face chat templates.