Task definition question

[Yijunmaverick]

@utkarshjp7 thanks for the nice blog and code. May I ask a question of the task definition?

I understand that after the meta-training, for a given new task, we aim at enabling the model to converge quickly with a few examples provided only.

But during the meta-training step where we have a set of tasks T, are we assuming that the training data of each task here should be also scarce, or abundant?