SimpleNetV1 architecture implementation in Pytorch

This is the pytorch implementation of our architecture SimpleNetV1(2016) .
Pytorch is different from caffe in several sections, and this made it a bit harder to have the architecture properly ported especially since I'm a complete newbie in Pytorch. However, thanks to this great work, I could easily focus on the model and port the architecture and hopefully achieve my reported results in Caffe and also exceed them as well!

The pytorch implementation is also very effieicent and the whole model takes only 1239MB with the batch size of 64! (compare this to other architectures such as ResNet,WRN, DenseNet which a 800K model takes more than 6G of vram!)

-- 2022: Adding ImageNet models

-- 2018: Initial commit

The original Caffe implementation can be found here : Original Caffe implementation - 2016

-- After nearly 7 years I could finally get my hands on a good GPU(RTX3080) and train the model on imagenet!
I used rwightman/pytorch-image-models repository to train the models. He did a great job by the way! I'll be updating the whole repository in the upcomming days inshaalah!
SimpleNet performs very decently, it outperforms VGGNet, ResNet and even some variants of MobileNets(1-3)
and its fast, very fast! (based on the model up to 2x faster).

-- The models(pytorch, onnx, jit) can be found in imagenet models directory.

-- Models and training logs can be found in snapshot folder.

Simply initiate the training like :
python3 main.py ./data/cifar.python --dataset cifar10 --arch simplenet --save_path ./snapshots/simplenet --epochs 540 --batch_size 100 --workers 2

Note that, the initial learning rate, and optimization policy is hard coded just like caffe.

If you find SimpleNet useful in your research, please consider citing:

@article{hasanpour2016lets,
  title={Lets keep it simple, Using simple architectures to outperform deeper and more complex architectures},
  author={Hasanpour, Seyyed Hossein and Rouhani, Mohammad and Fayyaz, Mohsen and Sabokrou, Mohammad},
  journal={arXiv preprint arXiv:1608.06037},
  year={2016}
}