Task definition question
Yijunmaverick opened this issue · comments
@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?