The following repository contains implementations of various papers in Tensorflow and PyTorch. More detail about the models and their respective papers can be found in the subfolders.
The goal of this repository is to create human readable implementations of different models, mainly for understanding. You should probably not try to use these models in production, use a library like timm, which contains pretrained weights too. Sources used in making the implementation are listed in the subfolders' README.
The following table lists the progress on each implementation in each framework.
| Model | PyTorch | Tensorflow | JAX/FLAX |
|---|---|---|---|
| EfficientNet | Yes | Yes | No |
| EfficientNet V2 | Yes | Yes | No |
| EfficientNet Lite | Yes | Yes | No |
| DenseNet | Yes | No | No |
| ResNet | Yes | Yes | In Progress |
| Vision Transformer | Yes | Yes | No |
| BERT | No | No | No |
| GPT | No | No | No |
| DCGAN | Yes | No | No |
| SRGAN | Yes | No | No |
| GAN | Yes | No | No |
| ConvNext | Yes | No | No |
| Swin Transformer | No | No | No |
| MobileNet | No | No | No |
| MobileNet V2 | Yes | Yes | No |
| MobileNet V3 | Yes | No | No |
| ResNext | No | No | No |
| Attention | No | No | No |
Not all implementations have been tested. Results will be in the subfolder, if there are any.
Please report any problems with the implementations if you find any. Thanks!