metadata
dataset_info:
features:
- name: prompt
list:
- name: role
dtype: string
- name: content
dtype: string
- name: source
dtype: string
- name: severity
dtype: string
- name: category
dtype: string
- name: has_code
dtype: bool
- name: has_poc
dtype: bool
- name: reference_response
dtype: string
- name: firm
dtype: string
- name: quality_score
dtype: float64
- name: protocol
dtype: string
splits:
- name: train
num_bytes: 135505305
num_examples: 48356
- name: validation
num_bytes: 7161775
num_examples: 2546
download_size: 103964620
dataset_size: 142667080
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*