The tensor is cuda type, but the GPU is not used.

[cswangjiawei]

I imitated your project to write a part-of-speech tagging project, adding character embedding and the crf layer (refer to https://pytorch.org/tutorials/beginner/nlp/advanced_tutorial.html#sphx-glr-beginner-nlp-advanced- Tutorial-py), but when I use GPU training, I find that all tensors are cuda type, but when I use the command 'nvidia-smi' to view GPU usage, I get the following image:
It don’t show any process using the GPU. I am very confused.This is the project I wrote.
POS-Tagger.zip