teamcore/DPO_Pm3B_RMAB_TG_clean_beta0.25dpo_proEurus_RM_7b_nu0.008_CR_cr_trajfullgp
收藏数据链接:
官方服务:
资源简介:
该数据集包含了四个字段:prompt(提示)、chosen(选择)、response(回应)和gpt_score(GPT评分)。它被设计用来训练和评估自然语言处理模型,特别是基于GPT模型的回应生成能力。数据集分为默认split,共有1200个示例。
The dataset includes four fields: prompt, chosen, response, and gpt_score. It is designed for training and evaluating natural language processing models, particularly the response generation capabilities of GPT-based models. The dataset is split into a default split with a total of 1200 examples.
提供机构:
teamcore


