metadata
dataset_info:
features:
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: helpfulness_chosen
dtype: int64
- name: helpfulness_rejected
dtype: int64
- name: specificity_chosen
dtype: int64
- name: specificity_rejected
dtype: int64
- name: intent_chosen
dtype: int64
- name: intent_rejected
dtype: int64
- name: factuality_chosen
dtype: int64
- name: factuality_rejected
dtype: int64
- name: easy-to-understand_chosen
dtype: int64
- name: easy-to-understand_rejected
dtype: int64
- name: relevance_chosen
dtype: int64
- name: relevance_rejected
dtype: int64
- name: readability_chosen
dtype: int64
- name: readability_rejected
dtype: int64
- name: enough-detail_chosen
dtype: int64
- name: enough-detail_rejected
dtype: int64
- name: biased:_chosen
dtype: int64
- name: biased:_rejected
dtype: int64
- name: fail-to-consider-individual-preferences_chosen
dtype: int64
- name: fail-to-consider-individual-preferences_rejected
dtype: int64
- name: repetetive_chosen
dtype: int64
- name: repetetive_rejected
dtype: int64
- name: fail-to-consider-context_chosen
dtype: int64
- name: fail-to-consider-context_rejected
dtype: int64
- name: too-long_chosen
dtype: int64
- name: too-long_rejected
dtype: int64
- name: human
dtype: string
- name: assistant_chosen
dtype: string
- name: assistant_rejected
dtype: string
- name: log_score_chosen
dtype: float64
- name: log_score_rejected
dtype: float64
- name: labels
dtype: string
splits:
- name: train
num_bytes: 14434424
num_examples: 9574
- name: test
num_bytes: 14378349
num_examples: 9574
download_size: 15748504
dataset_size: 28812773