how can i change the number of out_channels at conv2d.layer4.conv3 in ResNET50 in pytorch?

I want to change the output channels of Resnet50 in pytorch

so tried this but "weight should contain 2048 elements not 300" error occured..what's wrong with this?

I tried chatGPT But it occurs another error..

# Model training routine
print("\nTraining:-\n")

new_num_channels = 300

#----#
resnet_v2 = models.resnet50(pretrained=True)

#print(resnet_v2)
#print("fc_layer.in_features")
#print(fc_layer.in_features)

resnet_v2.layer4[-1].conv3.out_channels = new_num_channels
num_classes = 3


weight = torch.nn.Parameter(torch.Tensor(new_num_channels))
torch.nn.init.uniform_(weight)  # Initialize the weight tensor

resnet_v2.layer4[-1].bn3.weight = weight
resnet_v2.layer4[-1].bn3.num_features = new_num_channels



fc_layer = resnet_v2.fc
fc_layer.in_features = new_num_channels
fc_layer.weight = torch.nn.Parameter(torch.Tensor(new_num_channels, fc_layer.out_features))

gap_layer = resnet_v2.avgpool
gap_layer.num_features = fc_layer.in_features


resnet_v2.fc = nn.Sequential(
              nn.Linear(new_num_channels, 128), # Change the number of channels in Fully connected layer 
              nn.ReLU(inplace=True),
              nn.Linear(128, num_classes))

fc_layer.reset_parameters()

resnet_v2 = resnet_v2.to(device)

print(resnet_v2)
changed_model = train_model(resnet_v2, loss_fn, opt, scheduler, num_epochs=train_epoch)

From what I understand, you're trying to modify the architecture of a pretrained ResNet-50 model in PyTorch, specifically, you want to change the number of channels in the last convolutional layer, and subsequently adjust the following layers to accommodate this change. Your current approach is directly altering the model parameters which is not recommended in PyTorch.

To change the number of output channels in the last convolutional layer of ResNet50 (or any other layer), you'll need to replace the layer with a new instance of the same type of layer (in this case, nn.Conv2d), but with a different number of output channels. Similarly, you will need to adjust the following BatchNorm and Linear layers to accommodate this change.