Hungryof

Hi guys, I am confused about the 'updateGradInput' of TVLoss. Could someone give some descriptions on it?

TVLoss supports multi-batchsize?

2

I am wondering whether the code does the correct stuff. ``` -- multi-batchsize version? function TVLoss:updateGradInput(input, gradOutput) self.gradInput:resizeAs(input):zero() local N, C, H, W = input:size(1), input:size(2), input:size(3), input:size(4) self.x_diff:resize(4, H...

Error when using cudnn7

1

```bash torch/install/share/lua/5.1/cudnn/SpatialConvolution.lua:42: attempt to perform arithmetic on field 'groups' (a nil value) stack traceback: ...torch/install/share/lua/5.1/cudnn/SpatialConvolution.lua:42: in function 'resetWeightDescriptors' ...torch/install/share/lua/5.1/cudnn/SpatialConvolution.lua:93: in function 'checkInputChanged' ...torch/install/share/lua/5.1/cudnn/SpatialConvolution.lua:117: in function 'createIODescriptors' ...torch/install/share/lua/5.1/cudnn/SpatialConvolution.lua:177: in function 'func'...

错别字

1

在https://funhacks.net/explore-python/Advanced-Features/generator.html  这个thow应该是throw。 在 https://funhacks.net/explore-python/Functional/decorator.html  这个应该是functools，而不是functool

Several questions

1

1. Can we use `recursive_generator` similar with the official code? 2. `LayerNorm` is in here http://pytorch.org/docs/master/nn.html#layernorm.(Seems in this week v0.4 will be released)

What does this line mean?

1

https://github.com/Youngkl0726/Convolutional-Block-Attention-Module/blob/39653b5e60ebdae4e70ebfbe2105cee077203377/CBAMNet.py#L34

Why permute pro_features and why `[0]` here?

1

https://github.com/ShoufaChen/DiffusionDet/blob/1efb36d4b0b721f2d509bc72656235a94d73ab02/diffusiondet/head.py#L253-L254 1. Before passing proposal feature into `self.attn`, you permute the feature and make the size of it be `nr_boxes*N*C`, it confused me. For normal transformer block, the feature should...

`clip_x_start` is not used

1

https://github.com/ShoufaChen/DiffusionDet/blob/1efb36d4b0b721f2d509bc72656235a94d73ab02/diffusiondet/detector.py#L170 And here https://github.com/ShoufaChen/DiffusionDet/blob/main/diffusiondet/detector.py#L205, `clip_denoised=True` by default.

Hi, thank you for providing the code, I am looking forward to a better version ASAP. Thank you!