README.md

Machine Learning and Deep Learning Workspace

This repository is for testing, experimentation, and learning by getting hands dirty with core machine learning and deep learning workflows. It contains a collection of Jupyter notebooks and scripts covering NumPy, pandas, scikit-learn, PyTorch, computer vision, and more.

Quick links

• Deep Learning (PyTorch)

  • dl/repl.ipynb
  • dl/fundamentals_dl0.ipynb
  • dl/fundamentals_dl1.ipynb
  • dl/classification_dl2.ipynb
  • dl/computervision_dl3.ipynb
  • dl/datasets_dl4.ipynb
  • dl/tranferlearning_dl5.ipynb

• Machine Learning Libraries

  • library_learning/ml_numpy.ipynb
  • library_learning/pandas.ipynb
  • library_learning/scikitlearn0.ipynb

• The Annotated Transformer

  • annotated-transformer/AnnotatedTransformer.ipynb
  • annotated-transformer/the_annotated_transformer.py

• Modular Scripts

  • Overview: dl/going_modular/README.md
  • Entry point: dl/going_modular/going_modular/train.py

Structure

• dl/: Deep learning notebooks, small datasets, and modular scripts.
  • going_modular/: Reusable Python scripts for training.
  • data/: Datasets used in notebooks (e.g., pizza_steak_sushi, FashionMNIST).
  • models/: Exported PyTorch checkpoints.
• library_learning/: Machine learning notebooks focusing on libraries like NumPy, pandas, and scikit-learn.
• annotated-transformer/: Implementation of "The Annotated Transformer" paper, including notebook and Python script versions.
• helper_functions.py: General utility functions used by multiple notebooks.

Quick start

1. Create a Python environment (recommended Python 3.10+).
2. Install core packages used across notebooks:

   pip install numpy pandas matplotlib scikit-learn jupyterlab notebook torch torchvision

3. Open the desired notebook in VS Code or Jupyter and run cells.

Notes

• Some notebooks/scripts assume recent PyTorch + torchvision versions.
• For modular training scripts, read dl/going_modular/README.md first.

Running Modular Training

1. Prepare data (see going_modular.data_setup.create_dataloaders).
2. Run the training script:

   python dl/going_modular/going_modular/train.py

Contributing

• Make small, focused changes.
• Update relevant notebook outputs if you change code behavior.

License

This workspace is intended for learning and personal use. No license file is included.