nmaac comments

Results 28 comments of


                                            nmaac

速度

aconc和swish速度基本一样，如果慢可以参考hswish 把sigmoid换成hard sigmoid。另外一般不需要把relu全换掉

ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 16, 1, 1])

Hi @starsky68 @xuhao-anhe , seems the error is caused by the batch size value. If you set batch size = 1, please remove the bn layers: `beta = self.sigmoid(self.fc2(self.fc1(x.mean(dim=2, keepdims=True).mean(dim=3,...

ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 16, 1, 1])

> > Hi @starsky68 @xuhao-anhe , seems the error is caused by the batch size value. If you set batch size = 1, please remove the bn layers: > >...

centernet种使用预训练权重问题

看样子没什么问题，你的relu baseline是正常的map吗？不用预训练直接train from scratch呢？

Hello, I have some question

@Zhou-boyi1997 hello我们这边没有VGG的模型，小模型可以参考使用ShuffleNetV2

问题请教

暂时没有

Apply FReLU to YOLOv5

Hi @glenn-jocher , (1) you should use torch.max(), which is different from [MegEngine API](https://github.com/MegEngine/MegEngine/blob/fecbfdbb170df96216fefac2b95b8a5fe80e5b00/python_module/megengine/functional/elemwise.py). (2) yes you could try to use FReLU only in the backbone first, then try it...

nmaac

速度

ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 16, 1, 1])

ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 16, 1, 1])

centernet种使用预训练权重问题

Hello, I have some question

问题请教

Apply FReLU to YOLOv5

how to apply in InceptionV3

Funnel Activation implementation question

how to apply in PSPNet