metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: reference_answer
dtype: string
- name: answer
dtype: string
- name: model_id
dtype: string
- name: task_type
dtype: string
- name: task_subtype
dtype: string
- name: task_subsubtype
dtype: string
- name: difficulty
dtype: string
- name: domain
dtype: string
- name: is_provocative
dtype: bool
- name: criteria_name
dtype: string
- name: criteria_description
dtype: string
- name: rubrics
dtype: string
- name: rubrics_example
dtype: string
- name: annotations
list:
- name: comment
dtype: string
- name: score
dtype: float64
- name: criteria_score
dtype: float64
splits:
- name: benchmark
num_bytes: 1218995349
num_examples: 161076
- name: criteria_full
num_bytes: 1369856
num_examples: 1071
- name: criteria_brief
num_bytes: 155925
num_examples: 120
- name: instructions
num_bytes: 3298130
num_examples: 2115
download_size: 415354286
dataset_size: 1223819260
configs:
- config_name: default
data_files:
- split: benchmark
path: data/benchmark-*
- split: criteria_full
path: data/criteria_full-*
- split: criteria_brief
path: data/criteria_brief-*
- split: instructions
path: data/instructions-*