metadata
dataset_info:
features:
- name: prompt_text
dtype: string
- name: chosen_text
dtype: string
- name: rejected_text
dtype: string
- name: reference_chosen_logps
dtype: float64
- name: reference_rejected_logps
dtype: float64
- name: rewards_1_chosen
dtype: float64
- name: rewards_1_rejected
dtype: float64
- name: rewards_2_chosen
dtype: float64
- name: rewards_2_rejected
dtype: float64
- name: rewards_0_chosen
dtype: float64
- name: rewards_0_rejected
dtype: float64
- name: rewards_3_chosen
dtype: float64
- name: rewards_3_rejected
dtype: float64
splits:
- name: train_sft
num_bytes: 2028413
num_examples: 2165
- name: test_sft
num_bytes: 37363
num_examples: 39
download_size: 1018898
dataset_size: 2065776
configs:
- config_name: default
data_files:
- split: train_sft
path: data/train_sft-*
- split: test_sft
path: data/test_sft-*