Hi organizer,
We find that in the scoring.py given by starting kit, the test episode generator is using fixed setting as [28, 5, 1, 19], which means image size is 28 * 28 * 3, and the problem is fixed to 5 ways 1 support 19 queries. Is this setting shared by all the datasets in competition (especially datasets at final phase) or just for local/public dataset?
Meta_Learners
2020.10.17