metadata
license: apache-2.0
dataset_info:
features:
- name: id
dtype: string
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
splits:
- name: train
num_bytes: 79183415
num_examples: 43050
download_size: 0
dataset_size: 79183415
The dataset is created by (1) collecting Arabic questions in Quora and (2) requesting GPT4 to generate responses.
You can refer to https://github.com/FreedomIntelligence/Arabic-LLM.