FlexibleMachineLearning

A sample code sample to accompany the paper titled "Engineering flexible machine learning systems by traversing functionally-invariant paths"

Instructions

1. Create a conda environment using the environment_FIP.yml; conda env create -f environment_FIP.yml
2. Run the jupyter notebook in the conda environment
3. Run one cell after the other in the sequence it has been written!

Links to Datasets used in the paper

1. MNIST: https://www.kaggle.com/code/vincentlefoulon/pytorch-mnist (The MNIST dataset can be downloaded from torchvision.datasets.MNIST as mentioned under DataLoading)
2. Fashion MNIST: https://github.com/zalandoresearch/fashion-mnist
3. CIFAR-10: https://www.cs.toronto.edu/~kriz/cifar.html
4. CIFAR-100: https://www.cs.toronto.edu/~kriz/cifar.html (scroll down to second half of the page)
5. Yelp-reviews dataset for Transformer NLP task: https://huggingface.co/datasets/yelp_review_full
6. IMDB-reviews dataset: https://huggingface.co/datasets/imdb
7. WikiText dataset: https://huggingface.co/datasets/wikitext