Neural Network Final Project: Re-Implementation of the TSMixer Architecture
To install, one can clone this repo to then run the main.py file. One can decide to run the main.py code by default, which will execute the TSMixer model with the default best parameters for the ETTh1 dataset or choose to change the parameters to the user's liking.
TSMixer: https://arxiv.org/abs/2303.06053
Some implementation details in the paper I struggled to understand. I tried to explain the choices I made regarding these details in the jupyter file in this repo. The results are also present in this file. The model and other utilities are located in the src folder.