VanillaNet
VanillaNet copied to clipboard
huawei-noah
Hi, thanks for the great work. I have tried to apply vanilla-9 to object detection, however, when transferring the model to TensorRT, it seems much slower than ResNet-34. Is there...
权重使用
老师您好，我使用net.load_state_dict(torch.load(model_weight_path, map_location='cpu'))加载模型测试时，出现RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory错误，我的路径没问题，20几个权重都是这个错误，请问是什么原因呢？期待您的解答，我是一名研三学生，在应用您的模型时出现了这个问题，期望您百忙之中抽时间为我解答疑惑，万分感谢！
# We use leakyrelu to implement the deep training technique. x = torch.nn.functional.leaky_relu(x,self.act_learn) 这行代码，self.act_learn=1, 相当于不做激活， 它的作用仅仅是将网络变深吗？
Help
Hello, I want to know how to draw the Accuracy vs.depth diagram in your paper ? Look forward to reply !
Hi, I evaluated the speed in training proceeds between VanillaNet and ResNet, vanillanet has a little bit less params and flops than resnet. However, vanillanet is slower than resnet and...
Hi, that's a mind-blowing work, but I believe there is a slight miscommunication in the paper & codebase regarding the number of Conv layers and their influence on the inference...
训练报错
norm_cfg = dict(type='BN', requires_grad=True) model = dict( type='EncoderDecoder', backbone=dict( _delete_=True, type='Vanillanet', act_num=3, dims=[512, 512, 1024, 2048, 2048, 2048, 2048, 2048, 2048, 4096, 4096], strides=[1, 2, 2, 1, 1, 1, 1,...
Is there any comparasion or comment with MobileOne? https://github.com/apple/ml-mobileone
Hi, thanks for the great work. Can I understand the class 'activation(nn.ReLU)' as a combination of ReLU->depth Conv-> BN? I don't seem to see concurrently stacking activation functions in the...
Thanks for publishing this quite interesting work. I believe it is customary to use either bias or BatchNorm, but not both at once. the nn.Conv2d layers not have bias=False added...
