ChawDoe / LeNet5-MNIST-PyTorch

The simplest implementation of LeNet5 with mnist in PyTorch. Accuracy: ~99%
MIT License
193 stars 56 forks source link

Basic model structure issue #9

Closed ranlychan closed 5 months ago

ranlychan commented 8 months ago

It is believed that LeNet-5 uses Sigmoid as activation function and uses AvgPooling. ReLU and MaxPooling were adopted in AlexNet instead of LeNet-5, which can be confirmed on Wikipedia: https://en.wikipedia.org/wiki/LeNet.

ChawDoe commented 5 months ago

Thank you @ranlychan! You found the difference! As the implemented architecture in this repo has shown its high accuracy and it has been used by many researchers, I will mention this in the readme instead of modifying it again.