metadata
dataset_info:
features:
- name: utterance_id
dtype: large_string
- name: candidate_index
dtype: int64
- name: prediction_text
dtype: large_string
- name: token_id_seq
dtype: large_string
- name: token_seq
dtype: large_string
- name: token_logprob_seq
dtype: large_string
splits:
- name: train
num_bytes: 225053875
num_examples: 305325
download_size: 33248805
dataset_size: 225053875
configs:
- config_name: default
data_files:
- split: train
path: data/train-*