metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: shared_thought
sequence: string
- name: chosen
sequence: string
- name: chosen_reward
dtype: float64
- name: chosen_turn
dtype: int64
- name: chosen_error
dtype: int64
- name: reject
sequence: string
- name: reject_reward
dtype: float64
- name: reject_turn
dtype: int64
- name: reject_error
dtype: int64
- name: prompt_llama
dtype: string
- name: prompt_llama_token
sequence: int64
- name: chosen_llama
dtype: string
- name: chosen_llama_token
sequence: int64
- name: reject_llama
dtype: string
- name: reject_llama_token
sequence: int64
- name: chosen_logprob
dtype: float64
- name: reject_logprob
dtype: float64
splits:
- name: train
num_bytes: 5145209958
num_examples: 49532
- name: test
num_bytes: 103903636
num_examples: 1000
download_size: 696137968
dataset_size: 5249113594
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*