-
**Describe the feature**
Will the community consider supporting more non-contrastive self-supervised methods, such as the earlier Jigsaw Puzzle, RotNet, etc.?
**Motivation**
Some generative, sema…
-
cqfei updated
2 years ago
-
The current implementation works for Black and white images in mnist (by changing 0 to 1 and vice versa). How do we extend this approach for colored images?
-
Hi, when I run the train.sh, I encountered such problem"RuntimeError: merge_sort: failed to synchronize: cudaErrorIllegalAddress: an illegal memory access was encountered". Does anyone have any idea?
…
-
Thanks for the great work and your effort in releasing the code! Is there an expected date that your code will be released? Thanks in advance.
-
### Bug description
When using the latest head of this repo, we get an key error when calling the `dot_product` layer when using a two tower model. When we revert back to an earlier commit - this bug…
-
I believe there is a bug around the following line in the baseline prediction code:
https://github.com/vered1986/self_talk/blob/bbfc675b61a582aa00047ca31334f8fa75efcbad/source/predictors/multiple_cho…
-
You implement `loss = self.alpha * (1-y) * distance**2 + \
self.beta * y * (torch.max(torch.zeros_like(distance), self.margin - distance)**2)
` as your contrastive loss, howeve…
-
Hi and thanks for this great repo,
I would like to train AtoB where A are grayscale images and B are RGB images.
I tried specifically saving A images as grayscale (e.g. using img = PIL.Image.open(…
-
I noted Resnet18 has this line of code, which rescale pixel value to be [0, 1], (or [-1, 1]) don't remember which but close enough.
```
x = imagenet_utils.preprocess_input(
x, data_forma…