retrain-pipelines/func_calls

retrain-pipelines Function Calling数据集是用于函数调用的，包含两个配置：continued_pre_training和supervised_finetuning。continued_pre_training用于添加内在知识，而supervised_finetuning是一个问答文本数据集。数据集基于Salesforce/xlam-function-calling-60k和lighteval/natural_questions_clean构建，数据增强率为+4.5%，数据丰富率为+10.0%。

The retrain-pipelines Function Calling dataset is for function calling, containing two configurations: continued_pre_training and supervised_finetuning. The continued_pre_training is used to add intrinsic knowledge, while the supervised_finetuning is a question-answering text dataset. The dataset is built based on Salesforce/xlam-function-calling-60k and lighteval/natural_questions_clean, with a data augmentation rate of +4.5% and a data enrichment rate of +10.0%.