This hands-on workshop introduces the fundamentals of deep learning using PyTorch. Participants will learn by building real models and solving practical tasks.
- Core PyTorch concepts (tensors, autograd, GPU usage)
- Building and training neural networks
- Creating Physics-Informed neural networks
- Implementing CNNs for vision tasks
- Applying transfer learning with pre-trained models
- Working with real-world datasets
- Designing classification and regression models
✅ Recommended Platform: Google Colab
Colab provides a free, GPU-enabled environment—ideal for this workshop.
- A Google account
- Reliable internet connection
-
Open the GitHub repo.
-
Click the “Open in Colab” badge (if available), or:
- Download the notebook locally.
- Open Google Colab.
- Use File > Upload Notebook to run it.
- Runtime > Change runtime type
- Set Hardware Accelerator to
GPU - Click Save
📘 Colab Tips | Colab FAQ
Each notebook starts with a setup cell. Run it first to install all required libraries.
| Session | Topic | Duration |
|---|---|---|
| 1 | PyTorch Basics & Tensors | ~1 hr |
| 2 | Artificial Neural Networks (ANNs) | ~1.5 hr |
| 3 | Model Training & Optimization | ~0.5 hr |
| 3B | Physics-Informed Neural Networks (PINNs) | ~1 hr |
| 4 | Convolutional Neural Networks (CNNs) | ~2 hr |
| 5 | Transfer Learning & U-Net | ~2 hr |
By the end, you’ll be able to:
- Build and train models in PyTorch
- Apply CNNs to classification & segmentation
- Fine-tune pre-trained models on new tasks
- Use PyTorch effectively for real-world datasets
UoM_fse_dl_workshop/
├── solutions/ # Completed notebooks
├── figs/ # Figures and diagrams
├── utils/ # Checker and data helpers
│ ├── plotting/
│ ├── data/
│ ├── ml/
│ └── solutions.json
└── SE01_CA_Intro_to_pytorch.ipynb # Code-along notebooks
Throughout the notebooks, you’ll find 🎯 exercises. Use the built-in checker to validate your answers.
answer = {'your_solution': result}
checker.check_exercise(1, answer)checker.display_hints(1)✔️ Correct = green check ❌ Incorrect = feedback provided 💬 Hints are tailored to the task
- Read the exercise and implement the solution.
- Use the checker to validate your work.
- Request hints if needed.
- Learn from any mistakes and try again.
- Basic Python skills
- Some knowledge of basic machine learning concepts
- Familiarity with linear algebra/calculus (optional)
- No PyTorch experience required!