Skip to content

issues Search Results · repo:implus/PytorchInsight language:Python

Filter by

40 results
 (41 ms)

40 results

inimplus/PytorchInsight (press backspace or delete to remove)

implus/PytorchInsight
Replace with Group Normalization

#2 #33 Regarding the issues above, I think the following lines are the proper way to replace the original implementation with the group normalization. The second function is the slightly modified official ...
  • Haus226
  • Opened 
    on Aug 31, 2024
  • #40

implus/PytorchInsight
也许应该换一个角度理解SGE的原理

背景噪音的position对应的特征向量往往有很多元素没被激活，对向量所有元素相加的话，通常value会很小。而SGE用的是加权平均，权重大小是对应特征的Global Average（GAP越大就说明这个特征越显著/越高频）。 我感觉这样解释会比原文更容易理解
  • xxu5f
  • Opened 
    on Aug 17, 2022
  • #39

implus/PytorchInsight
about normalization

image Hello, I have a problem. What is the purpose of the identity transformation in the paper?
  • long123524
  • 1
  • Opened 
    on Sep 28, 2021
  • #38

implus/PytorchInsight
"sk_resnet101.pth.tar" pre-training model could not be loaded， in pytorch

my code: def get_se_resnet101(path_state_dict, device, vis_model=False): model = resnet_se.se_resnet101() if path_state_dict: pretrained_state_dict = torch.load(path_state_dict, ...
  • creater-zq
  • Opened 
    on Feb 15, 2021
  • #37

implus/PytorchInsight
se-shufflenet

se-shufflenet performance?
  • czy112
  • Opened 
    on Aug 5, 2020
  • #36

implus/PytorchInsight
ShuffleNetV2_1x model

Hello, is there a Baidu Cloud version？ can you send it to my email 2572875853@qq.com thanks
  • czy112
  • 1
  • Opened 
    on Jun 29, 2020
  • #35

implus/PytorchInsight
ResNet34-sk problem

Your model ResNet34-Sk is not quite right, there is no code for the corresponding sk module in BasicBlock. If you use resnet34_sk directly, it is equivalent to ResNet34?
  • sunshiding
  • 1
  • Opened 
    on Apr 2, 2020
  • #34

implus/PytorchInsight
Is the operation in SGE-Block equivalent to GroupNorm ?

Hi. I have two questions: Question 1: t = t - t.mean(dim=1, keepdim=True) std = t.std(dim=1, keepdim=True) + 1e-5 t = t / std t = t.view(b, self.groups, h, w) ...
  • mrT23
  • 1
  • Opened 
    on Jan 30, 2020
  • #33

implus/PytorchInsight
Running SGE in fp16 (Mixed Precision)

Hi, Running SGE Module in mixed precision resulted in gradient overflow (NaN in training loss). I converted it the input to float and then to half, that helped a bit at first but half way through the training ...
  • hussam789
  • 2
  • Opened 
    on Jan 30, 2020
  • #32

implus/PytorchInsight
which the version of pytorch these model trained by?

rt
  • young-fire
  • Opened 
    on Jan 12, 2020
  • #31
Issue origami icon

Learn how you can use GitHub Issues to plan and track your work.

Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub Issues
ProTip! 
Restrict your search to the title by using the in:title qualifier.
Issue origami icon

Learn how you can use GitHub Issues to plan and track your work.

Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub Issues
ProTip! 
Press the
/
key to activate the search input again and adjust your query.
Issue search results · GitHub