This is my replication of the DenseNet Model using the Cifar10 Dataset

Due to GPU Size constraints this is a rather small model but can be scaled according to your gpu size
This model includes implementations for both standard DenseNets and Bottlenecked DenseNets. You can change between the two in main.py by updating bottlenecked = False

This project was created using uv and is highly recommended
After installing uv this project should run out of the box

You can get the original Dataset from Alex Krizhevsky's webpage although there are likely sources out there that provide it in a more modern format.

Once this format is downloaded you will need to run the cifar10_data_processing.py script to create our datasets

uv run cifar10_data_processing.py

This will create a data folder with both training and test directories

Before kicking off training you should update the weights and biases variables entity and project in init_logging() in train.py to match your account.
If not using Weights and Biases (not recommended) you can set logs to False in main.py
To kick off training you can run
uv run main.py

Paper link

@inproceedings{huang2017densely,
  title={Densely connected convolutional networks},
  author={Huang, Gao and Liu, Zhuang and Van Der Maaten, Laurens and Weinberger, Kilian Q},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4700--4708},
  year={2017}
}