metadata
dataset_info:
features:
- name: chosen_input_ids
sequence: int64
- name: chosen_labels
sequence: int64
- name: chosen_attention_mask
sequence: int64
- name: rejected_input_ids
sequence: int64
- name: rejected_labels
sequence: int64
- name: rejected_attention_mask
sequence: int64
splits:
- name: train
num_bytes: 2610910108.072663
num_examples: 148930
download_size: 353521602
dataset_size: 2610910108.072663
configs:
- config_name: default
data_files:
- split: train
path: data/train-*