metadata
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: instruction
dtype: string
- name: raw_text
dtype: string
- name: token_id
dtype: string
- name: log_prob
dtype: string
- name: cluster_dict
dtype: string
- name: cluster_verdict
dtype: string
- name: majority_rate
dtype: float64
- name: majority_answer
dtype: string
- name: majority_correctness
dtype: int64
- name: correctness_rate
dtype: float64
- name: correct_answer
dtype: string
- name: prompt_type
dtype: int64
- name: prompt
dtype: string
- name: responses_correctness_heuristic_before
dtype: string
- name: responses_correctness_heuristic_after
dtype: string
- name: responses_correctness_number
dtype: string
- name: responses_mix
dtype: string
- name: wrong_answer
dtype: int64
splits:
- name: train
num_bytes: 82002084
num_examples: 6408
download_size: 37717434
dataset_size: 82002084
configs:
- config_name: default
data_files:
- split: train
path: data/train-*