The correctness of the VGG11 model

[whcjimmy]

After reading the paper and the code, it appears that the implementation of VGG11 is incorrect.

The original VGG11 has around 32 million parameters, whereas the paper indicates only 129K. Additionally, upon parsing the code, the structure resembles more of a ResNet-11 model rather than a VGG-style model. A VGG model typically has a maximum of 512 channels, while the model you provided has at most 64 channels.

Could you please verify whether you accidentally uploaded the wrong model?

Thank you!