metadata
dataset_info:
features:
- name: problem
dtype: string
- name: answer
dtype: string
- name: tokenized_prompt
dtype: string
- name: responses
list: string
- name: rewards
list: float64
- name: mean_reward
dtype: float64
- name: sources
list: string
- name: solution
dtype: string
- name: answer_source
dtype: string
splits:
- name: train
num_bytes: 5134891622
num_examples: 5877
download_size: 1672637228
dataset_size: 5134891622
configs:
- config_name: default
data_files:
- split: train
path: data/train-*