-
Dear authors,
I think the repo is an excellent implementation of your excellent paper. However, it shares a common bug in setting the seed number for data augmentation as pointed in [here](https://…
-
After reading your paper and amazed by model performance, I have problem in understanding the specific methods in stage two. I know little about representation learning. And the paper doesn't explai…
-
When I used Contrastive Loss Regularization(--cl-reg), 'stylegan2_pytorch --generate' command did not work after training. It seems that the model cannot be loaded correctly.
continuing from previo…
-
In your code, there are some variants based on MoCo ("eg, ContrastiveROIHeadsWithStorage") or Prototype ("eg, ContrastiveROIHeadsWithPrototype"). What is the final result (nAP) of these two variants? …
-
This is a more of a theory question: How could I evaluate the sucess of this contrastive representation learning in an unsupervised manner, using validation data?
What I'm thinking for now:
1. Loa…
-
Hi, thanks for your sharing.
What is the meaning of the batch?
Thanks
-
Hi,
Thanks for your implementation.
In the original paper the total loss is defined as L_PixPro + alpha*L_inst (the instance level contrastive loss), while in your code it's L_PixPro + alpha*L_Pix…
-
Hi,
I have been trying to train a CLIP model from scratch. After edit data loader functions in [this code](https://github.com/KeremTurgutlu/self_supervised/blob/main/examples/training_clip.py), Wh…
-
Hi Kevin.
Thank you for providing this wonderful code for metric learning. I am facing a weird issue for which I seek your guidance. When I use a simple contrastive loss to train my network on multip…
-
I've defined a model like this: (numPoints = 1 for testing)
```
def euclidean_distance(vects):
x, y = vects
return K.sqrt(K.maximum(K.sum(K.square(x - y), axis=1, keepdims=True), K.epsil…