Jerry Jiarui XU

#7 Please try this solution.

I believe it is caused by different hardware. Try to change https://github.com/chengdazhi/Deformable-Convolution-V2-PyTorch/blob/1b5851abd404dc71f02ae0110af3540b6877309e/src/cuda/deform_im2col_cuda.cuh#L17 and https://github.com/chengdazhi/Deformable-Convolution-V2-PyTorch/blob/1b5851abd404dc71f02ae0110af3540b6877309e/src/cuda/modulated_deform_im2col_cuda.cuh#L17 to some smaller number like 512 or 256. This should help.

> Hello > I have the same issue , I tried with 512, 256, ... even with 1 and I am receiving the same error. I have GeForce GTX1080. Any...

Sorry for the late reply. `error in deformable_col2im_cuda: too many resources requested for launch` could be solved by https://github.com/chengdazhi/Deformable-Convolution-V2-PyTorch/issues/15#issuecomment-473521698 This error caused the operator failed so the gradient check won't...

Hi @mikigom Thanks for the nice feature. Could you please fix the lint error to pass the CI?

Sorry for the late reply. Please note that during ImageNet training, we trained for 120 epoch in total, which means Tmax should also be set to 120 if don't consider...

As mentioned in https://github.com/xvjiarui/GCNet/issues/10, we also successfully trained GCNet for ImageNet classification from scratch. My last reply is under training from scratch setting. As for fine-tune setting mentioned in the...

Sorry for the late reply. The value is a linear transformation that transforms feature before skip connection. Please refer to `g` in **fig 2** of [*NonLocal*](https://arxiv.org/pdf/1711.07971.pdf), `V` in **fig 2...

By value of transform module, are you referring to the output of Global Context Block before residue connection? In this case, we are not explicitly using channel-wise attention. The output...

Please find the visualization code [here](https://github.com/xvjiarui/mmdetection/blob/gcnet_vis/tools/vis.py).