metadata
dataset_info:
features:
- name: prompt_text
dtype: string
- name: prompt_ids
sequence: int64
- name: decode_ids
sequence: int64
- name: prompt_pattern
sequence:
sequence: int64
- name: decode_pattern
sequence:
sequence: int64
- name: predictor_pattern
sequence:
sequence:
sequence: int64
splits:
- name: train
num_bytes: 219905957
num_examples: 10936
download_size: 23650636
dataset_size: 219905957
configs:
- config_name: default
data_files:
- split: train
path: data/train-*