miguelvr/dropblock issues and pull requests

#46 - shared DropBlock mask across different feature channels

Issue - State: open - Opened by WDDouble almost 3 years ago

#45 - make dropblock run faster

Issue - State: open - Opened by xieyxclack over 3 years ago - 1 comment

#44 - 'ResNetCustom' object has no attribute '_norm_layer'

Issue - State: open - Opened by hello-trouble over 3 years ago - 2 comments

#43 - Dropblock and python version

Issue - State: open - Opened by ShihaoSu over 3 years ago - 3 comments
Labels: question

#42 - Use in segmentation. You suggest only for convolution feature extraction layer only. What do you mean?

Issue - State: closed - Opened by Eric2Hamel almost 4 years ago - 2 comments

#41 - Can I use with torch.no_grad() in the forward of dropblock.py?

Issue - State: closed - Opened by StrongerzZ over 4 years ago - 4 comments

#40 - Update scheduler.py

Pull Request - State: closed - Opened by hanochk over 4 years ago

#39 - Would you please tell me how did you make certain the positive area of the mask keep still after max pooling operation ?

Issue - State: closed - Opened by CoinCheung over 4 years ago - 6 comments
Labels: question

#38 - Pre-trained ResNet50 model on ImageNet?

Issue - State: closed - Opened by wangzheallen over 4 years ago - 1 comment

#37 - why drop_prob in config.yaml is 0

Issue - State: closed - Opened by liangzimei about 5 years ago - 1 comment

#36 - The examples for using dropblock is different?

Issue - State: closed - Opened by ZealACMer about 5 years ago - 1 comment

#35 - Where to put the dropblock in the ResNet network?

Issue - State: closed - Opened by ZealACMer over 5 years ago - 1 comment

#34 - Pytorch Lightning as a back-end

Issue - State: closed - Opened by williamFalcon over 5 years ago - 2 comments

#33 - Is it needed to somehow turn off Dropblock layer during inference time?

Issue - State: closed - Opened by Westerby over 5 years ago - 1 comment

#32 - resnet-cifar10.py DropBlock scheduler

Issue - State: closed - Opened by vadik6666 almost 6 years ago - 1 comment

#31 - DropBlock at Residual Connections

Issue - State: closed - Opened by kayuksel almost 6 years ago - 1 comment

#30 - DropBlock1D

Issue - State: open - Opened by kayuksel almost 6 years ago - 3 comments
Labels: enhancement, v0.4

#29 - How to put in evaluation mode?

Issue - State: closed - Opened by bluesky314 almost 6 years ago - 1 comment

#28 - Make drop_prob mutable in DropBlock ScriptModules

Pull Request - State: open - Opened by mdraw almost 6 years ago - 3 comments

#27 - [WIP] torch.jit support

Pull Request - State: open - Opened by miguelvr almost 6 years ago

#26 - same results as traditional dropout if block_size=1

Issue - State: open - Opened by sergulaydore almost 6 years ago - 1 comment
Labels: question

#25 - torch.jit support

Issue - State: closed - Opened by mdraw almost 6 years ago - 6 comments
Labels: enhancement, v0.4

#24 - inconsistency with the original paper

Issue - State: open - Opened by duducheng almost 6 years ago - 9 comments
Labels: enhancement, v0.4

#23 - Should the output be scaled by 1/(1-p) during training time?

Issue - State: closed - Opened by eezywu about 6 years ago - 1 comment

#22 - Replaces Bernoulli sample operation with torch.rand() < gamma

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#21 - Replaces Convolution operations with MaxPool

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#20 - the drop_size cannot be higher than 5?

Issue - State: closed - Opened by PanXiebit about 6 years ago - 7 comments

#19 - Replace Bernoulli with creating random matrix

Issue - State: closed - Opened by linzehui about 6 years ago - 1 comment

#18 - Segmentation?

Issue - State: closed - Opened by bluesky314 about 6 years ago - 2 comments

#17 - The variable x will be assigned to int type wrongly

Issue - State: closed - Opened by Sampson-Lee about 6 years ago - 5 comments

#16 - Convolution operation can be replaced by MaxPool

Issue - State: closed - Opened by miguelvr about 6 years ago - 3 comments

#15 - fix padding on DropBlock2D

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#14 - About padding

Issue - State: closed - Opened by rexnxiaobai about 6 years ago - 2 comments

#13 - Scheduler and benchmark example

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#12 - Fixes gamma computation

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#11 - ImportError: cannot import name 'DropBlock2D'

Issue - State: closed - Opened by Redaimao about 6 years ago - 1 comment

#10 - About feat_size?

Issue - State: closed - Opened by hizhangp about 6 years ago - 6 comments

#9 - fixes overlap blocks issue #8

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#8 - Negative block mask egde case

Issue - State: closed - Opened by miguelvr about 6 years ago

#7 - How about the speed of dropblock?

Issue - State: closed - Opened by donnyyou about 6 years ago - 1 comment

#6 - Fix device issue

Pull Request - State: closed - Opened by radao about 6 years ago - 1 comment

#5 - RuntimeError: Expected object of type torch.cuda.FloatTensor but found type torch.FloatTensor for argument #2 'other'

Issue - State: closed - Opened by radao about 6 years ago - 1 comment

#4 - Fixes block mask computation

Pull Request - State: closed - Opened by miguelvr about 6 years ago

#3 - there is something wrong

Issue - State: closed - Opened by yunyundong about 6 years ago - 10 comments

#2 - scheduled dropblock?

Issue - State: closed - Opened by twmht about 6 years ago - 4 comments

#1 - Convolution based implementation + DropBlock3D

Pull Request - State: closed - Opened by miguelvr about 6 years ago

GitHub / miguelvr/dropblock issues and pull requests