A best practice for deep learning project template architecture.
A simple and well designed structure is essential for any Deep Learning project, so after a lot practice and contributing in pytorch projects here's a pytorch project template that combines simplicity, best practice for folder structure and good OOP design. The main idea is that there's much same stuff you do every time when you start your pytorch project, so wrapping all this shared stuff will help you to change just the core idea every time you start a new pytorch project.
So, here’s a simple pytorch template that help you get into your main project faster and just focus on your core (Model Architecture, Training Flow, etc)
In order to decrease repeated stuff, we recommend to use a high-level library. You can write your own high-level library or you can just use some third-part libraries such as ignite, fastai, mmcv … etc. This can help you write compact but full-featured training loops in a few lines of code. Here we use ignite to train mnist as an example.
In a nutshell here's how to use this template, so for example assume you want to implement ResNet-18 to train mnist, so you should do the following: - In
modelingfolder create a python file named whatever you like, here we named it
modeling/__init__.pyfile, you can build a function named
build_modelto call your model
from .example_model import ResNet18
def build_model(cfg): model = ResNet18(cfg.MODEL.NUM_CLASSES) return model
enginefolder create a model trainer function and inference function. In trainer function, you need to write the logic of the training process, you can use some third-party library to decrease the repeated stuff.
# trainer def do_train(cfg, model, train_loader, val_loader, optimizer, scheduler, loss_fn): """ implement the logic of epoch: -loop on the number of iterations in the config and call the train step -add any summaries you want using the summary """ pass
def inference(cfg, model, val_loader): """ implement the logic of the train step
toolsfolder, you create the
train.py. In this file, you need to get the instances of the following objects "Model", "DataLoader”, “Optimizer”, and config ```python # create instance of the model you want model = build_model(cfg)
trainloader = makedataloader(cfg, istrain=True) valloader = makedataloader(cfg, istrain=False)
optimizer = make_optimizer(cfg, model) ```
do_train, and start your training
python # here you train your model do_train(cfg, model, train_loader, val_loader, optimizer, None, F.cross_entropy)
You will find a template file and a simple example in the model and trainer folder that shows you how to try your first model simply.
├── config │ └── defaults.py - here's the default config file. │ │ ├── configs │ └── train_mnist_softmax.yml - here's the specific config file for specific model or dataset. │ │ ├── data │ └── datasets - here's the datasets folder that is responsible for all data handling. │ └── transforms - here's the data preprocess folder that is responsible for all data augmentation. │ └── build.py - here's the file to make dataloader. │ └── collate_batch.py - here's the file that is responsible for merges a list of samples to form a mini-batch. │ │ ├── engine │ ├── trainer.py - this file contains the train loops. │ └── inference.py - this file contains the inference process. │ │ ├── layers - this folder contains any customed layers of your project. │ └── conv_layer.py │ │ ├── modeling - this folder contains any model of your project. │ └── example_model.py │ │ ├── solver - this folder contains optimizer of your project. │ └── build.py │ └── lr_scheduler.py │ │ ├── tools - here's the train/test model of your project. │ └── train_net.py - here's an example of train model that is responsible for the whole pipeline. │ │ └── utils │ ├── logger.py │ └── any_other_utils_you_need │ │ └── tests - this foler contains unit test of your project. ├── test_data_sampler.py
Any kind of enhancement or contribution is welcomed.