metadata
dataset_info:
features:
- name: conv_id
dtype: int64
- name: prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: completion
dtype: string
- name: score
dtype: float64
- name: turn_id
dtype: int64
- name: single_turn_prompt
dtype: string
- name: single_turn_completion
dtype: string
- name: single_turn_metadata
struct:
- name: authors
dtype: string
- name: num_tokens
dtype: int64
- name: tags
dtype: string
- name: timestamp
dtype: string
- name: url
dtype: string
- name: sessions
struct:
- name: '0'
struct:
- name: average_score
dtype: int64
- name: '1'
struct:
- name: average_score
dtype: int64
- name: '2'
struct:
- name: average_score
dtype: int64
- name: rewards
struct:
- name: length_avg
dtype: float64
- name: llm_rw_avg
dtype: float64
- name: llm_rw_std
dtype: float64
- name: reward_std
dtype: float64
- name: task_metric_avg
dtype: float64
- name: task_metric_std
dtype: float64
- name: token_cost_avg
dtype: float64
- name: token_cost_std
dtype: float64
splits:
- name: train
num_bytes: 43433034
num_examples: 4606
download_size: 7111843
dataset_size: 43433034
configs:
- config_name: default
data_files:
- split: train
path: data/train-*