NbAiLab/ndla_npk_conversational_nb_to_nn_tags_balanced
收藏数据链接:
官方服务:
资源简介:
该数据集是基于NbAiLab/ndla_npk_conversational_nb_to_nn的平衡版本,包含了70000个来自NbAiLab/ndla_npk_conversational_nb_to_nn的样本,15000个带有标签的来自NDLA的样本,以及15000个带有标签的来自NPL的样本。总共有10万个样本。该数据集主要用于GRPO训练。
This dataset is a balanced version of NbAiLab/ndla_npk_conversational_nb_to_nn, consisting of 70,000 samples from NbAiLab/ndla_npk_conversational_nb_to_nn, 15,000 samples with tags from NDLA, and 15,000 samples with tags from NPL. There are a total of 100,000 samples. The dataset is mainly used for GRPO training.
提供机构:
NbAiLab


