This project involves transforming the original Keras-based implementation of models presented in the paper "How Inverse Conditional Flows Can Serve as a Substitute for Distributional Regression" by Kook et al. (2024) into PyTorch. The goal is to maintain the core architecture and computational strategies from the original models while leveraging the flexibility and efficiency of PyTorch.
This project is part of the Applied Deep Learning course at Ludwig Maximilian University of Munich.
This repository contains the Python implementation of NEAT, including toy examples and a hyperparameter search implementation. The purpose of this project is to replicate and enhance the models described in the paper, allowing for flexible distributional regression.
The folder includes:
- Toy Examples: Simple examples to demonstrate the functionality of the models.
- Hyperparameter Search: Tools for performing hyperparameter optimization to find the best configurations for model performance.
These examples serve as a guide for understanding the application of the models and how to set them up for various datasets.
You can install the required libraries via:
pip install -r requirements.txt