Original: https: //www.cnblogs.com/Lee-yl/p/9884055.html
Original: https: //blog.csdn.net/juxiangming/article/details/70175985
In machine learning, what is the training? What is the model?
By known and objective data, adjust parameters of the algorithm, which is the training;
the resulting map, for this issue to be trained model.
Conversion between Tensor and numpy:
By a.numpy (), can be converted into numpy Tensor a data type, use torch.from_numpy () can be converted into numpy Tensor,
If changes are required tensor data type, only need add desired after conversion type, such as to convert into a type of float, just a.float () on it.