A way to remove specific indexes from dataset

[anthonysirico]

Is there a way to remove specific graphs from my dataset. What I am trying to do is predict on my test set, and then remove anything from the set that was predicted 0, then concatenate the predicted 1s with the training set creating a NEW training set. Thoughts?

[anthonysirico]

UPDATE: I am trying to use .index_select to create another set from a list of indices. If ones = [1 5 6 9 7 10 15 16 ... 22546] and I use the line

ones_set = test_set.index_select(ones)

it returns an out of range error. I have tried converting the list to an ndarray as the docs said to use and a tensor of type long, and it keeps throwing the same error.

Again, the end goal is to take specific indices out of a set to create another set, to later concatenate it with the training set.

[wsad1]

Is len(test_set) larger than max(ones)? Also could you paste the full error message you are seeing?

[anthonysirico]

Figured it out: used a for loop that cycled through the indexes in the list.