As the course progresses and projects get more complex, we will start to move away from single notebooks and rely more on GitHub repositories.

Containerized development environments

  1. DeepLearning PyTorch/Tensorflow development environment: gperdrizet/deeplearning-devcontainer - template repository for containerized deeplearning development environment with wide (GPU, Pascal+, CPU only) hardware compatibility.

  2. Large language model hosting/development environment: gperdrizet/llms-devcontainer - template repository for working on LLM applications with compatibility for GPU (Pascal+) or CPU only environments. Also includes tools for local inference server hosting.

Demos & activities

  1. Weather API demo: gperdrizet/weather - weather data acquisition & parsing. This will probably be a data source for a future demo/activity.

  2. Streamlit recommender app demo: gperdrizet/anime-recommendations - demo of content-based filtering recommender system. Deployed as Streamlit app via Render.

  3. CIFAR-10 project: gperdrizet/CIFAR10 - our in-house CNN and ResNet 50 are both hovering around 91% overall accuracy, still hoping to definitively beat ResNet over the weekend!

  4. Mask detection capstone solution: gperdrizet/face-mask-detection - my incremental capstone 10 solution, includes lots of data augmentation and some more advanced training techniques, final test set accuracy 99.8%. Includes live deployment of Streamlit web app via Render.

  5. YOLO26 activity: gperdrizet/YOLO26 - simple Streamlit web app on Streamlit Community Cloud that does real-time object detection via your webcam. Activity prompts you to deploy the app for yourself and then extend its functionality.

  6. RNN demo/activity: gperdrizet/RNNs - intro to RNNs (learning the alphabet) and NLP/GRUs (Twitter sentiment analysis), activity prompts you to improve the twitter model by implementing a hybrid CNN/RNN model.

  7. Autoencoders demo: gperdrizet/autoencoders Demonstration of convolutional autoencoder models for compression, denoising and anomaly detection with images.

  8. Distributed GANN training: g`perdrizet/GANNs-with-friends. Distributed deep learning demo using DCGANNs to generate realistic images of faces.

  9. Encoder decoder translation models: gperdrizet/language-models. Neural machine translation models using encoder-decoder architecture and LSTM+attention, builds up to implementation of the first transformer model.

  10. LLM applications demonstrations: gperdrizet/llms-demo. Demonstrations related to building LLM based applications topics covered include hosting models for inference, prompting strategies and LangChain for application development. Also includes slides and activities.

  11. Unit 6 assignments template: gperdrizet/unit-6-assignments. GitHub repo for the incremental capstone and unit end assignment from unit 6. This is the environment set-up I used to complete my solutions, it contains a devcontainer set-up based on the llms-gpu container.