metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: input
struct:
- name: candidates
sequence: string
- name: interaction
sequence: string
- name: user_id
dtype: string
- name: output
struct:
- name: recommended
sequence: string
splits:
- name: train_20
num_bytes: 27162003
num_examples: 4473
- name: train_80
num_bytes: 109217311
num_examples: 17890
download_size: 22525096
dataset_size: 136379314
configs:
- config_name: default
data_files:
- split: train_20
path: data/train_20-*
- split: train_80
path: data/train_80-*