What I did are all shown in the .ipynb file in each assignment folder
Build up a multi-layer perceptron from scratch in python
Build up a CNN model by Barebone Pytorch, Pytorch Module API, and Pytorch Sequential API.
Besides, apply Image Retrieval with Neural Codes on CIFAR-10 dataset.
Paper presentation - EfficientNet - Rethinking Model Scaling for Convolutional Neural Networks
Take Unet as a baseline model for segmentation on Cityscape dataset and improve the accuracy by applying data augmentation, hyperparameters tuning, and architecture changing.