Workshop on Hands-on Deep Learning Coding & Code Management

Welcome to the repository for the "Workshop on Hands-on Deep Learning Coding & Code Management," organized by the Center for Computational & Data Sciences (CCDS.ai) at Independent University Bangladesh (IUB) . This repository contains the code and materials for Session 2, focusing on Deep Learning Code Management.

📅 Event Details

• Event Name: Workshop on Hands-on Deep Learning Coding & Code Management
• Organized by: Center for Computational & Data Sciences (CCDS) at Independent University Bangladesh (IUB)
• Session: Day 2 - Session 2 - Deep Learning Code Management
• Event Link: Workshop Event

🌐 Session Overview

In this session, we delve into effective code management practices for deep learning projects. The code for the session is implemented in Python using PyTorch and PyTorch Lightning framework, utilizing various packages for data processing and model training.

📂 Project Structure

• code/: Contains the code examples and exercises for Session 2.

• data/: Data modules with train, val, and test dataloaders, including data preparation and splitting files using PyTorch Lightning.

• configs/: YAML files containing arguments for different components.

• models/: Model definitions for training, validation, and testing, as well as logic for optimizer, LR scheduler, score metric, and loggers.

• utils/: Necessary functions and classes to load config files and other utilities.

• cli.py: Command line interface for training and testing experiments.

• __data__/ (optional): Temporary folder to download the dataset.

• __logs__/ (optional): Temporary folder to store checkpoints, logs, and other experiment-related information.

🛠️ Libraries and Dependencies

The project is implemented in Python using PyTorch and PyTorch Lightning framework, along with the following packages:

• albumentations==1.4.22
• click==8.1.7
• numpy==1.26.4
• Pillow==9.3.0
• pytorch_lightning==2.4.0
• PyYAML==6.0.1
• torchmetrics==1.6.0
• torchvision==0.17.1
• timm==1.0.12
• wandb==0.19.0

🚀 Getting Started

1. Clone this repository to your local machine:

git clone https://github.com/your-username/deep-learning-code-management-workshop.

2. Navigate to the project directory:

cd deep-learning-code-management-workshop

3. Explore the relevant folders, such as code/ and configs/, for session-specific content.

💻 Usage

Use the cli.py command line interface to train and test experiments. For example:

python cli.py train --config configs/train_config.yaml
python cli.py test --config configs/test_config.yaml`

Using Docker

You can also run the code using Docker. First, build the Docker image:

docker build -t deep-learning-workshop .

Then, run the Docker container:

docker run -it deep-learning-workshop

You can push the Docker image to Docker Hub or any other container registry for sharing with others.

docker tag deep-learning-workshop your-docker-username/deep-learning-workshop
docker push your-docker-username/deep-learning-workshop

Run the experiments inside the Docker container using the cli.py script.

docker run --rm -it --gpus all --ipc=host -v $(pwd):/app deep-learning-workshop python cli.py train --config configs/train_config.yaml
docker run --rm -it --gpus all --ipc=host -v $(pwd):/app deep-learning-workshop python cli.py test --config configs/test_config.yaml

🤝 Contributions

If you have any improvements, suggestions, or corrections, feel free to open an issue or submit a pull request. Your contributions are highly appreciated!

📝 License

This project is licensed under the MIT License.

Happy coding!🎉