These are my solutions to Stanford's CS231n. These solutions are for the 2022 edition of the course.
I am going through the assignments and reading some of the course notes. These are not full solutions. I'm skipping some of the non-coding questions in the notebooks that I don't feel like answering, and I am not putting much energy the training related exercieses, resulting in lower validation and test accuracies. I am primarily focusing on better understanding the algorithms.
I provided some of my thoughts on the course in /notes.md.
I am working on these locally, and not on google colab as the course recommends. I have changed the first box in each notebook to download datsets locally.
- Python (Anaconda distribution recommended for numpy, scipy, jupyter, etc.)
Follow the course steps to work on a local machine
conda create -n cs231n # create a virtual environment
python -m notebook # open the jupyter instance at http://localhost:8888/tree
It is also recommended to set up some git configuration so that jupyter notebook outputs are not committed. [source 1], [source2]