metadata
dataset_info:
features:
- name: idx
dtype: int64
- name: prompt
dtype: string
- name: answers
sequence: string
- name: gt
dtype: string
- name: rewards
sequence: float64
splits:
- name: train
num_bytes: 2928500854
num_examples: 20000
download_size: 1084919296
dataset_size: 2928500854
configs:
- config_name: default
data_files:
- split: train
path: data/train-*