Pytorch CNN Input for Guitar Tab CNN

Question

I am trying to implement the architecture I have attached.

The output of my DataLoader has size: torch.Size([128, 192, 9, 1])

I am using a batch size of 128

View just reshapes the output of the dense layer

 model = nn.Sequential(nn.Conv2d(192, 32, 3), 
                              nn.ReLU(),
                              nn.Conv2d(32, 64, 3),
                              nn.ReLU(),
                              nn.Conv2d(64, 64, 3),
                              nn.MaxPool2d(2),
                              nn.Dropout(0.25),
                              nn.Flatten(),
                              nn.Linear(5952, 128),
                              nn.ReLU(),
                              nn.Dropout(0.5),
                              nn.Linear(128, 126),
                              View((6, 21)),
                              nn.Softmax(dim=1))

this is the architecture I currently have and I don't know if my input to the Conv2d are correct

I keep getting errors with my dimensions and kernel sizes. I am unsure how to proceed.

CNN Architecture