Dataset for the Paper: "Understanding Self-Claimed Assumptions in Deep Learning Frameworks: An Exploratory Study"

This is the dataset for the paper: "Understanding Self-Claimed Assumptions in Deep Learning Frameworks: An Exploratory Study". It contains the 3004 Self-Claimed Assumptions (SCAs) extracted from nine deep learning frameworks (i.e., TensorFlow, Theano, PyTorch, Caffe, MXNet, Keras, CNTK, DL4J, and PaddlePaddle), which includes 2232 unique SCAs that have different content and 772 duplicated SCAs that have the same content as one of the 2232 SCAs.本数据集对应论文《理解深度学习框架中的自声明假设：一项探索性研究》（Understanding Self-Claimed Assumptions in Deep Learning Frameworks: An Exploratory Study）。该数据集包含从TensorFlow、Theano、PyTorch、Caffe、MXNet、Keras、CNTK、DL4J及PaddlePaddle这9种深度学习框架中提取的3004条自声明假设（Self-Claimed Assumptions，SCAs），其中包含2232条内容各异的唯一自声明假设，以及772条与前述2232条自声明假设中某一条内容完全一致的重复自声明假设。