metadata
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: rationale
dtype: string
- name: question_type
dtype: string
- name: meta
dtype: string
- name: question_type_group
dtype: string
- name: similarity_to_fact
dtype: float64
- name: prompt
dtype: string
- name: completion
dtype: string
- name: input
dtype: string
- name: prompt_oracle
dtype: string
- name: completion_oracle
dtype: string
- name: input_oracle
dtype: string
- name: generation_input
dtype: string
- name: generation_input_oracle
dtype: string
- name: prediction
dtype: string
- name: prediction_rationale
dtype: string
- name: prediction_oracle
dtype: string
- name: prediction_oracle_rationale
dtype: string
- name: log_ll
dtype: float64
- name: oracle_log_ll
dtype: float64
- name: oracle_advantage
dtype: float64
- name: accuracy
dtype: bool
- name: accuracy_oracle
dtype: bool
- name: accuracy_status
dtype: string
splits:
- name: train
num_bytes: 31147840
num_examples: 929
download_size: 3614936
dataset_size: 31147840
configs:
- config_name: default
data_files:
- split: train
path: data/train-*