umbcvision / msf Goto Github PK

I find a promeble in run python eval_knn.py and python eval_linear.py, the eval_linear.py is very low, but the python eval_knn.py result can get 0.95. this is why? I run in my data (10 class)

Thanks for your great work, quite interesting.
I wonder have you ever tried to experiment with small-scale datasets? Imagenet it rather big to try as you know.

Best.

There are two Eval programs in your project. How can I debug them

Hi,

Thanks for bringing this great work to the community !!

Would you please share the checkpoint you have pretrained on imagenet?

Hi,

Thank you for sharing the code. I have a silly question regarding the MSE calculation. I found in your code, you used dist_t = 2 - 2 * torch.einsum('bc,kc->bk', [current_target, targets]). In my understanding, torch.einsum('bc,kc->bk', [current_target, targets]) is calculating elementwise multiplication and sum them in c dimension. How does this result in MSE?

Thank you.

Hi, I'd like to know if you ran the experiments with one gpu?
Also, it seems that the [paper] link here is mistaken.

Hi,

Nice work! Could you please provide the following details:
a) How long did it take to finish the training (Resnet50 with 200 epochs)?
b) How many gpus did you use?
c) What gpus (memory) did you use?

Thanks in advance!

I found an error in your code，in eval_linear.py in 106 line, The weight of the model cannot be read.This will lead to poor test results

hi

I am trying to reproduce the tsne experiments that provided in the paper and do the visualization, but I am not able to get as clean results as is provided.
would it be possible to share the code for that ?