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: processed_response
dtype: string
- name: processed_tokenized_response
sequence: int64
- name: unprocessed_response
dtype: string
- name: unprocessed_tokenized_response
sequence: int64
- name: accuracy
dtype: float64
- name: rm_score
dtype: float64
splits:
- name: train
num_bytes: 10647827998
num_examples: 1420504
download_size: 1080703023
dataset_size: 10647827998
configs:
- config_name: default
data_files:
- split: train
path: data/train-*