metadata
dataset_info:
features:
- name: question_id
dtype: string
- name: sample_id
dtype: int64
- name: prompt_pretokenized
dtype: string
- name: prompt_tokenized
sequence: int64
- name: response
dtype: string
- name: tokenized_response
sequence: int64
- name: accuracy
dtype: float64
- name: logp
dtype: float64
- name: rm_score
dtype: float64
splits:
- name: train
num_bytes: 2286987762
num_examples: 486202
download_size: 213607658
dataset_size: 2286987762
configs:
- config_name: default
data_files:
- split: train
path: data/train-*