metadata
dataset_info:
features:
- name: run
dtype: string
- name: question
dtype: string
- name: ground_truth
dtype: string
- name: full_response
dtype: string
- name: reasoning
dtype: string
- name: final_answer
dtype: string
- name: grade
dtype: string
- name: grader_reasoning
dtype: string
splits:
- name: gpt_5
num_bytes: 5409144
num_examples: 1598
download_size: 2578652
dataset_size: 5409144
configs:
- config_name: default
data_files:
- split: gpt_5
path: data/gpt_5-*