dataset_info:
features:
- name: prompt
dtype: string - name: chosen
dtype: string - name: rejected
dtype: string
splits: - name: train
num_bytes: 12143678
num_examples: 19862 - name: test
num_bytes: 3118994
num_examples: 4996
download_size: 10724182
dataset_size: 15262672
configs: - config_name: default
data_files:- split: train
path: data/train-* - split: test
path: data/test-*
task_categories:
- split: train
- conversational
language: - zh
size_categories: - 10K<n<100K
Dataset Card for "rlhf_reward_single_round-chinese-zhtw"
基於 anthropic 的 Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback 論文開源的關於有助和無害的人類偏好資料。
這些數據旨在為後續的 RLHF 訓練訓練偏好（或獎勵）模型。
來源資料集
本資料集來自 beyond/rlhf-reward-single-round-trans_chinese, 并使用 OpenCC 來進行簡繁轉換。