metadata
dataset_info:
features:
- name: paper_id
dtype: string
- name: messages
list:
- name: role
dtype: string
- name: content
dtype: string
- name: pred_score
dtype: float64
- name: avg_score
dtype: float64
- name: scores
list: float64
- name: mae
dtype: float64
- name: is_success
dtype: bool
splits:
- name: train
num_bytes: 24441241
num_examples: 1662
download_size: 24392627
dataset_size: 24441241
configs:
- config_name: default
data_files:
- split: train
path: data/train-*