The programming part of the exam will have you solve a deep learning problem by reading some data and performing a task such as classification or regression, covered within chapters 1-10 and lectures 1-7. You can look through these examples.
- Fast AI book notebooks
- fast.ai course notebooks
- Pytorch
- Pytorch Tutorials
- Learn PyTorch for Deep Learning: Zero to Mastery course
- the-incredible-pytorch
- Kaggle Notebooks
- PYTORCH HUB
Steps:
- Find a notebook that meets the criteria above (material we covered)
- Remove all the explanations in the notebook (delete all the markdown cells)
- Create a new markdown file or notebook and read through the code, taking notes on what each line does. Your notes should be in your own words and refer to any specific code. For example, you might say this code loads data from the URL "http:www.example.com/data" and saves it to the file "data.csv". Or you might say this code uses a TensorFlow function to create time series datasets for training, validation, and testing with a window size of 20.
- Close the original notebook and do something else for a while (3 hours, 1 day, etc.)
- Open your notes and without looking at the original notebook and just using the API documentation, recreate the notebook. Only consult the original Notebook in an emergency if you get stuck.
- Submit your notes, the original notebook, and your recreated notebook in GitHub classroom.