vballoli/nfnets-pytorch issues and pull requests

#46 - Bugfix for Python > 3.10

Pull Request - State: open - Opened by oriyonay about 1 year ago

#45 - Bugfix in import for Python >= 3.10

Pull Request - State: open - Opened by oriyonay over 1 year ago

#44 - A decoder for Semantic Segmentation

Issue - State: open - Opened by cschloer almost 2 years ago

#43 - Fix: Python3.10 Iterable import

Pull Request - State: open - Opened by bvsn over 2 years ago

#42 - Is Weight Standardization correct?

Issue - State: closed - Opened by gellston over 2 years ago - 1 comment

#41 - Any Pretrained Model Available?

Issue - State: closed - Opened by KleinXin over 2 years ago - 1 comment

#40 - train model acc and test acc gap is very large

Issue - State: closed - Opened by Julius-ZCJ over 3 years ago - 1 comment

#39 - Method To Compute Gamma For ScaledStdConv

Issue - State: closed - Opened by nmichlo over 3 years ago - 1 comment

#38 - BN version of the models?

Issue - State: closed - Opened by htwang14 over 3 years ago - 1 comment

#35 - Bug in weight standardization?

Issue - State: closed - Opened by jamestwebber almost 4 years ago - 1 comment

#34 - Example in readme does not work

Issue - State: closed - Opened by RandomString123 almost 4 years ago - 4 comments
Labels: enhancement, WIP, will be delayed

#32 - Fix typos

Pull Request - State: closed - Opened by lalondma almost 4 years ago - 1 comment

#31 - Thanks a lot for you work . Could you release pretrained weight ?

Issue - State: closed - Opened by wtjgh almost 4 years ago - 1 comment
Labels: help wanted, lack of compute

#30 - Weight initialisation

Issue - State: closed - Opened by martinferianc almost 4 years ago - 1 comment
Labels: bug

#29 - Implement `param_groups` for AGC

Issue - State: closed - Opened by rmchurch almost 4 years ago - 4 comments
Labels: bug

#28 - Fix: pass correct weight to conv_transpose2d

Pull Request - State: closed - Opened by pisarik almost 4 years ago - 1 comment

#27 - Performance comparison

Issue - State: closed - Opened by xialeiliu almost 4 years ago - 1 comment

#26 - Update resnet.py

Pull Request - State: closed - Opened by zuenko almost 4 years ago

#25 - Syntax Error

Issue - State: closed - Opened by zuenko almost 4 years ago - 1 comment

#24 - Fix replace_conv with custom class parameter

Pull Request - State: closed - Opened by mj9 almost 4 years ago - 1 comment

#23 - Model is None

Issue - State: closed - Opened by shi27feng almost 4 years ago - 2 comments
Labels: bug

#22 - About AGC arguments

Issue - State: closed - Opened by planemanner almost 4 years ago - 1 comment
Labels: question

#21 - Incorrect params initialization in AGC

Issue - State: closed - Opened by haigh1510 almost 4 years ago - 1 comment

#20 - AGC condition

Issue - State: closed - Opened by haigh1510 almost 4 years ago - 2 comments

#19 - Code question

Issue - State: closed - Opened by q121q almost 4 years ago - 1 comment
Labels: Discussion

#18 - Is it pytorch compatibility??

Issue - State: closed - Opened by Choneke almost 4 years ago - 11 comments
Labels: question

#17 - Can you release the NFNets family models like Resnet examples in pytorch ?

Issue - State: closed - Opened by cuge1995 almost 4 years ago - 2 comments
Labels: WIP

#16 - add test for wsconv1d

Pull Request - State: closed - Opened by shi27feng almost 4 years ago - 2 comments

#15 - Update test_conv.py

Pull Request - State: closed - Opened by shi27feng almost 4 years ago - 1 comment

#14 - Add WSConv1d

Pull Request - State: closed - Opened by shi27feng almost 4 years ago - 3 comments

#13 - Will much appreciate a requirements.txt

Issue - State: closed - Opened by ooodragon94 almost 4 years ago - 2 comments
Labels: Fixed

#12 - Update agc.py

Pull Request - State: closed - Opened by shi27feng almost 4 years ago

#10 - Conv layer from timm work better

Issue - State: closed - Opened by attashe almost 4 years ago - 2 comments
Labels: enhancement, help wanted, Discussion

#9 - optim_agc.step() this instruction is raising AttributeError: 'NoneType' object has no attribute 'ndim'

Issue - State: closed - Opened by sachin22225 almost 4 years ago - 13 comments
Labels: bug, Fixed

#8 - Adaptive gradient clipping shouldn't be used in the final classifier layer

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

#5 - Feature request for WSConvTranspose2d

Issue - State: closed - Opened by hadaev8 about 4 years ago - 4 comments

#3 - torch.max doesn't check for tensors being on different devices.

Issue - State: closed - Opened by bfialkoff about 4 years ago - 4 comments
Labels: Fixed

#2 - TypeError: replace_conv() takes 1 positional argument but 2 were given

Issue - State: closed - Opened by sainatarajan about 4 years ago - 2 comments
Labels: Fixed

#1 - AGC without modifying the optimizer

Issue - State: closed - Opened by kayuksel about 4 years ago - 4 comments
Labels: Fixed

GitHub / vballoli/nfnets-pytorch issues and pull requests