Meta-Learning to Improve Pre-Training

This folder contains code to run experiments in the paper Meta-Learning to Improve Pre-Training, NeurIPS 2021. Please refer to the README files in the multitask/ and simclr folders for experiments in the mutlitask PT and self-supervised PT domains, respectively.

We also include a self-contained notebook that can be run on Google Colab to examine the synthetic MNIST data augmentation domain here.