Fetching the paper…

Improving out-of-distribution generalization via multi-task self-supervised pretraining · Around