Diff-tf1.x-tf2.x-pytorch means Differ Among TensorFlow 1.x, TensorFlow 2.x, and Pytorch.
Previously, I struggled to summarize some differences among these three frameworks in words. But this is really hard for me to make a list about it, as I have decided to only use Pytorch. So you see, the lenet5_pytorch.py is more complete.
Well, in github.com you must be a coder (or programmer sounds nicer?), hence you'd better read the code lines and find the differences by yourself. (This is the best explaination I can come up with to hidden my laziness. LOL.)
Folder tensorflow1.x is an old-old-old LeNet5_MNIST project based on TensorFlow 1.9.
cd tensorflow1.x
python3 lenet5_backward.py
python3 lenet5_test.py Script lenet5_tensorflow2.x.py is based on TensorFlow 2.4.
python3 lenet5_tensorflow2.x.py Script lenet5_pytorch.py is based on Pytorch 1.8. This one is more complete, maybe you can learn something about argparse and tqdm.
python3 lenet5_pytorch.py --epochs 10 --batch-size 64 --device 0,1 Try
pip install -r requirements.txt If error occurs, try to fix it by yourself.
In folder tensorflow1.x, there is another requirements_tf1x.txt due to different version of TensorFlow. Uninstall the TensorFlow 2.4 first if you want to try tensorflow1.x.
Just call me Al (not ai but al. LOL.) / Albert / lingff.
Rabbit is my favorite animal.
E-mail: ling@stu.pku.edu.cn
Gitee: https://gitee.com/lingff
CSDN: https://blog.csdn.net/weixin_43214408