I'm training a very resourceful CycleGAN.
As the training at night, it happens that the system makes a wipe of the virtual machine and I lose all the checkpoints of my training phases.
I would like to insert a control, in which for example every 100 epochs, the checkpoints are downloaded to my hard disk, so that I can reload them and restart the training.
Is it possible to download files in a programmed way on Colab?