Hi,
when I am using the SequentialSampler it appears that it orders the indicies I pass. Lets say I want items [3,1,2] it would give me the items but in the ordering [1,2,3] is there a way to prevent the Sampler form doing so.
50% of all Answers are found within 5 minutes after posting the questions. 
Rather than using a Sampler one can use torch.utils.data.Subset(dataset, train_index)