by Jalel Zgonda, Jonathan Labhard, Robin Zbinden
The goal of this project is to design a mini "deep learning framework" using only pytorch's tensor operations and the standard math library. More about this project can be read in the report_project.pdf file.
Run the script test.py to test on a simple dataset this framework with:
python test.py
To use this framework, please follow the indications in the report_project.pdf file.
Modules.py defines the differents modules inheriting from the module class, e.g., Sequential, Linear, ReLU,...
functional.py defines helpers mathematical functions like the activations functions, losses and their derivatives
generate_data.py defines functions to generate the dataset
training.py contains the classes and functions to train the model and to test it, e.g., LossMSE, train_model_SGD, accuracy,...
main.ipynb shows how we obtain the results obtained in the report.pdf file.
test.py is a script to test on a simple dataset this framework by using a simple neural network and training it