metadata
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: id
dtype: string
- name: seed_prompt
dtype: string
- name: language
dtype: string
- name: model
dtype: string
- name: messages
list:
- name: role
dtype: string
- name: content
dtype: string
- name: agent_trace
list: json
- name: source_id
dtype: string
- name: research_early_stopping
dtype: bool
splits:
- name: train
num_bytes: 126871259
num_examples: 5199
download_size: 124801585
dataset_size: 126871259