/EE559Project2

Deep Learning Course Mini-project 2

Primary LanguagePython

Mini deep-learning framework

EPFL | Deep Learning (EE-559) (Spring 2020) | Mini-project 2

Python 3.7 Pytorch 1.13.1

About

  • This is our implementaion for the mini-project 2 in the Deep leaning course at EPFL.
    • Team member: Pengkang Guo, Xiaoyu Lin
  • [report]

Project Discription

The objective of this project is to design a mini "deep learning framework" using only pytorch's tensor operations and the standard math library, hence in particular without using autograd or the neural-network modules.

Requirements

Pytorch

Run

From the root of the project: python test.py

Description of the files

  • module.py: the implementation of the modules
    • Includes Linear, Relu, Tanh, LossMSE and Sequential.
  • test.py: the required basic Python script using our framework module.py.
    • Generates the dataset, initializes and trains the required model with three hidden layers of 25 units.
    • Generates an output file, logs.out, logging the loss and error rate of each epoch. Calculates and prints the average test error rate, average time and their standard deviations.
  • test_plot.py: an upgraded version of test.py.
    • Has all the functions of test.py.
    • Generates the images of the dataset, training error rate and test error rate.
  • test_torch.py: an Pytorch version of test_plot.py.
    • has all the functions of test_plot.py but is implemented using Pytorch