Hello, I came across some of fastai1 semi-supervised techniques, do you do any implementation for fastai2.
I am competing in a kaggle computer vision competition and they gave us only 1/8 of the data labeled + the rest of unlabeled data (7/8) + the test set, do you have any suggestions on how to use all the 7/8 data ?
I would check out this list, @kcturgutlu did some fantastic work on his self-supervised library: Official Student Made fastai Extensions List (potentially more libraries may be added there in the future)
2 Likes
It should be self-supervised, I think you have a typo here
1 Like