datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ko.SHP
π’ Korean Stanford Human Preferences Dataset (Ko.SHP)
μ΄ λ°μ΄ν°μ
μ μ체 ꡬμΆν λ²μκΈ°λ₯Ό νμ©νμ¬ stanfordnlp/SHP λ°μ΄ν°μ
μ λ²μν κ²μ
λλ€.
μλμ λ΄μ©μ ν΄λΉ λ²μκΈ°λ‘ README νμΌμ λ²μν κ²μ
λλ€. μ°Έκ³ λΆνλ립λλ€.
If you mention this dataset in a paper, please cite the paper: Understanding Dataset Difficulty with V-Usable Information (ICML 2022).
Summary
SHPλ μ리μμ λ²λ₯ μ‘°μΈμ μ΄λ₯΄κΈ°κΉμ§ 18κ°μ§ λ€λ₯Έ μ£Όμ μμμ μ§λ¬Έ/μ§μΉ¨μ λν μλ΅μ λν 385K μ§λ¨ μΈκ° μ νΈλ λ°μ΄ν° μΈνΈμ΄λ€.
κΈ°λ³Έ μ€μ μ λ€λ₯Έ μλ΅μ λ ν ν μλ΅μ μ μ©μ±μ λ°μ νκΈ° μν κ²μ΄λ©° RLHF 보μ λͺ¨λΈ λ° NLG νκ° λͺ¨λΈ (μ: SteamSHP)μ νλ ¨ νλ λ°β¦ See the full description on the dataset page: https://huggingface.co/datasets/nlp-with-deeplearning/ko.SHP.Ko.HelpSteerμλ³Έ λ°μ΄ν°μ
: nvidia/HelpSteer