Your code doesn't shuffle the relative position of data. For example, in 5-way 1-shot, suppose in index 1-5 corresponds to label 1-5, then index 6 must belong to label 1, index 7 is label2..... So, is this the common setting for transductive learning?
Your code doesn't shuffle the relative position of data. For example, in 5-way 1-shot, suppose in index 1-5 corresponds to label 1-5, then index 6 must belong to label 1, index 7 is label2..... So, is this the common setting for transductive learning?