A deep learning training tool entirely from scratch using Python, aimed at enabling users to build and train custom deep learning models. The project involved designing efficient data structures to support tensor operations and core deep learning functionalities. Key components included implementing auto-differentiation and backpropagation algorithms, broadcasting mechanisms, and GPU-accelerated parallel programming. The system was optimized for both computational performance and memory usage. Extensive testing was conducted through a suite of custom test cases to ensure the accuracy and reliability of each module.

Additionally you will need to install and download the MNist library.

(On Mac, this may require installing the wget command)

pip install python-mnist
mnist_get_data.sh

• Tests:

python run_tests.py