--- dataset_info: features: - name: chosen dtype: string - name: rejected dtype: string - name: __index_level_0__ dtype: int64 splits: - name: train num_bytes: 30828302 num_examples: 19363 download_size: 15034439 dataset_size: 30828302 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "ko_hh-rlhf-20k_filtered" Synatra-Translation 모델로 번역된 20k rlhf셋입니다. 번역퀄이 뛰어나진 않습니다. 추가 대화문 등의 데이터 학습이 필요해보입니다. ## 베이스 데이터셋 [Anthropic/hh-rlhf](https://huggingface.co/datasets/Anthropic/hh-rlhf)