Say, I was just wondering if there are any compute resources available for this competition? I am talking about AWS, GCP, or other resources to train the models. Vision models take such a long time, that it would be nice to access some TPUs or V100s when models are ready to ramp up.
You can use google colab! They have free GPU fir this type of problems.
Just got up and running with Google Colab and wrote down the steps at the link below: