CoTNet

CoTNet copied to clipboard

[INFO: 2023-07-30 00:36:50,416] Model cotnet50 created, flops_count: 3.29 GMac, param count: 22.22 M [INFO: 2023-07-30 00:36:50,474] AMP not enabled. Training in float32. [INFO: 2023-07-30 00:36:50,474] Using native Torch DistributedDataParallel. Traceback...

ThelilinNB

Hello, can I train on my own dataset? Which parts should I change? Thank you

sudonto

I'm a newbie. If I want to try the CoT module on ResNet18, should I replace all 3x3 convolution or only certain layers of convolution? Has anyone tried it? Thank...

chenyuuuma

 好像是cupy冲突？ 求大佬指点

JustinNober

Hello author, I tried to extract the Cot Layer in CotNet, but found that the Cot Layer can only run on a single card. If running on multiple cards, the...

xiangchiqingyu

k,q,v中k和v采用3*3卷积来提取局部特征，还能进行全局建模吗？

23jisuper

Hello, I would like to extract some stages from the pre-trained COTNet-50 for feature extraction. How can I achieve this?

wudami