This crash course is designed for those with a proficiency in Python, aiming to bridge the gap to mastering PyTorch as a leading deep learning framework. Among the many topics we cover are:
- What is PyTorch? Why is it used?
- The basics of PyTorch (Tensors, Reshaping tensors, Broadcasting)
- Building a Basic Neural Network (What are Layers?, Activation functions, Feed Forward networks, Multi-Layer Perceptrons, Datasets, DataLoaders)
- Building an Advanced Neural Network (CNNs, RNNs, Cat vs. Dog Classification with the CIFAR-10 dataset, Evaluating Models)
- Regularization and Optimization Techniques (Dropout, Batch Normalization, Learning Rate, Step Decay)
- Transfer Learning
- Text Classification with LSTMs
- Inference with an R-CNN Object Detection Model
- The Basics of Saving and Loading Trained Models (ONNX, TorchServe)
- Fine-Tuning Transformer Based Architectures (Tokenizing a Dataset, GPT2-Medium, Text Completion, Repetitive Token Deterrence)
And much more!
Each topic is accompanied by detailed explanatory notes and code demonstrations that aim to give a comprehensive understanding of both the underlying theory, as well as its hands-on application. Furthermore, each new topic is paired with one or more links that offer additional information.
- Python (Version 3.10.12 or newer)
- PyTorch (Version 2.0.1)
For the most optimal experience:
Open the Notebook: Click here to access the notebook on Google Colab.
Use a GPU Runtime: Once opened, switch to a GPU runtime for faster results. Preferred GPUs: T4, V100, A100.
If you prefer to run the notebook on your local machine:
- Clone the repository:
git clone https://github.com/nbetts2020/PyTorch-Crash-Course.git
cd PyTorch-Crash-Course
- Install the Required Packages: Instructions to install the required packages are included in the notebook
If you'd like to contribute, please fork the repository and create a pull request, or simply open an issue!
This project is licensed under the MIT License, see the LICENSE file for more details.
Email: nbettencourt2020@gmail.com
GitHub: https://github.com/nbetts2020
LinkedIn: https://www.linkedin.com/in/nicholas-bettencourt/
If you gained anything of value from this course, feel free to give it a star. I appreciate it!