Vision Transformer for AutoML Tasks

This repository presents a Vision transfomer and Gradient Boosting model for automl tasks.

Pre-requisites

Before running this scripts, make sure to install the required modules. Navigate to the requirements.txt file and run the code below in your suitable terminal environment.

pip install -r requirements.txt

or

python -m pip install -r requirements.txt

Also make sure to use python version 3.9 and create a virtual env to run the above scripts.

Criterion

Model Type	Task
Transformer with time-based tokenization	Sequence and signal data analysis
Vision Transformer	Image classification
U-Net	Image segmentation
LightGBM	Tabular data analysis

Training and Testing

The model is designed to take in metadata to configure the performance metrics and output task. Identify the directory path of your dataset and configure the path variables in th main.py file.

To test the model, test dataset is hosted on kaggle. You can download the entirety or specific to dataset description.