CoinCheung

Hi, It is said that naive adam will make performance bad if weight decay is added. Thus people invented adamW to make adam compatible with weight decay. Now I have...

请问pkubase-full.txt可以在哪下载

8

如题

How could I use cuda block shared memory in the self-defined operator?

6

I am working on ubuntu16.04 with pytorch1.3 installed from conda. My cuda version is 10.1.243 and cudnn version is 7. I have 8 t4 gpus on my server and the...

https://github.com/talebolano/TensorRT_Seg_LW_Refinenet/blob/e774ff318fa2ab17d911081e711e8e2e2be8d73c/TensorRT_Seg/TensorRT_Seg.cpp#L367 Hi, I am studying from your code about how to use tensorrt to infer onnx model. I am quite confused about this line of L367. why do you destroy...

How did you align two images?

3

Hi, Thanks for sharing this !! I have gone through your code, and I noticed that the background image and the face image seem not be to aligned. Do we...

如题，能不能封一个appimage呢

Hi I am trying to learn segnet from this tutorial. I see in the directory Models/ that there are two sort of segnet(segnet_basic and segnet). They have different network structures...

First I should say thanks to you since this repository give me a good reference to learn GAN. I am trying to construct a basic DC-GAN with MXNET, but I...

In the paper, D means densecl. Where is the densecl style configuration ?

Hi, I just read the code and found this line: https://github.com/YudeWang/deeplabv3plus-pytorch/blob/ce48c6453ce49b59164513fe9dbc8a75b6416826/lib/net/xception.py#L128 if the output stride is 8, the stride of the above line should be changed from 2 to 1,...