metadata
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: meta
dtype: string
- name: question_type
dtype: string
- name: rationale
dtype: string
- name: prompt
dtype: string
- name: prompt_oracle
dtype: string
- name: gold_completion
dtype: string
- name: generation_input
dtype: string
- name: generation_input_oracle
dtype: string
- name: completion
dtype: string
- name: completion_oracle
dtype: string
- name: text
dtype: string
- name: text_oracle
dtype: string
- name: log_likelihood
dtype: float64
- name: log_likelihood_oracle
dtype: float64
- name: oracle_advantage
dtype: float64
- name: is_correct
dtype: bool
- name: is_correct_oracle
dtype: bool
- name: __index_level_0__
dtype: int64
splits:
- name: train
num_bytes: 1052539
num_examples: 230
download_size: 291937
dataset_size: 1052539
configs:
- config_name: default
data_files:
- split: train
path: data/train-*