Tensor Size Miss match on loss function - numpy

Question

1:
when attempting to perfrom a pytorch training sequence using batch sizes, my loss function appears to error when the nn output and a batch are put through a MSEloss function.
2:
have tried to search about nn padding, however this is not a covnet but rather an autoencoder, similar stack over flow issues have not yielded results.
3:
the NN:
class Net(nn.Module):
def __init__(self, input_dim=10):
super().__init__()
self.fc1 = nn.Linear(input_dim, int(0.75 * input_dim))
self.fc2 = nn.Linear(int(0.75 * input_dim), int(0.5 * input_dim))
self.fc3 = nn.Linear(int(0.5 * input_dim), int(0.33 * input_dim))
self.fc4 = nn.Linear(int(0.33 * input_dim), int(0.25 * input_dim))
self.fc5 = nn.Linear(int(0.25 * input_dim), int(0.33 * input_dim))
self.fc6 = nn.Linear(int(0.33 * input_dim), int(0.5 * input_dim))
self.fc7 = nn.Linear(int(0.5 * input_dim), int(0.75 * input_dim))
self.fc8 = nn.Linear(int(0.75 * input_dim), input_dim)
def forward(self, x):
x = torch.tanh(self.fc1(x))
x = torch.tanh(self.fc2(x))
x = torch.tanh(self.fc3(x))
x = torch.tanh(self.fc4(x))
x = torch.tanh(self.fc5(x))
x = torch.tanh(self.fc6(x))
x = torch.tanh(self.fc7(x))
x = self.fc8(x)
return torch.softmax(x, dim=1)
the train method:
def train(net, x_train, x_opt, BATCH_SIZE, EPOCHS, input_dim):
outputs = 0
mse = 0
optimizer = optim.SGD(net.parameters(), lr=0.001)
loss_function = nn.MSELoss()
for epoch in range(EPOCHS):
for i in tqdm(range(0, len(x_train), BATCH_SIZE)):
batch_x = x_train[i:i + BATCH_SIZE]
# print("bx", batch_x.size())
batch_y = x_opt[i:i + BATCH_SIZE]
# print("by", batch_y.size())
net.zero_grad()
# batch_x.view(batch_y.shape[0])
outputs = net(batch_x)
# print('out', outputs)
loss = loss_function(outputs, batch_y)
loss.backward()
optimizer.step() # Does the update
print(f"Epoch: {epoch}. Loss: {loss}")
error:
99%|█████████▉| 1452/1466 [00:02<00:00, 718.09it/s]B:\tools and software\Anaconda\envs\pysyft-pytorch\lib\site-packages\torch\nn\modules\loss.py:431: UserWarning: Using a target size (torch.Size([39, 10])) that is different to the input size (torch.Size([38, 10])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
return F.mse_loss(input, target, reduction=self.reduction)
100%|█████████▉| 1465/1466 [00:02<00:00, 718.36it/s]
Traceback (most recent call last):
File "<input>", line 1, in <module>
File "B:\tools and software\PyCharm 2020.1\plugins\python\helpers\pydev\_pydev_bundle\pydev_umd.py", line 197, in runfile
pydev_imports.execfile(filename, global_vars, local_vars) # execute the script
File "B:\tools and software\PyCharm 2020.1\plugins\python\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
exec(compile(contents+"\n", file, 'exec'), glob, loc)
File "B:/projects/openProjects/githubprojects/BotnetTrafficAnalysisFederaedLearning/anomaly-detection/pytorch_conversion.py", line 154, in <module>
input_dim=input_dim)
File "B:/projects/openProjects/githubprojects/BotnetTrafficAnalysisFederaedLearning/anomaly-detection/pytorch_conversion.py", line 64, in train
loss = loss_function(outputs, batch_y)
File "B:\tools and software\Anaconda\envs\pysyft-pytorch\lib\site-packages\torch\nn\modules\module.py", line 532, in __call__
result = self.forward(*input, **kwargs)
File "B:\tools and software\Anaconda\envs\pysyft-pytorch\lib\site-packages\torch\nn\modules\loss.py", line 431, in forward
return F.mse_loss(input, target, reduction=self.reduction)
File "B:\tools and software\Anaconda\envs\pysyft-pytorch\lib\site-packages\torch\nn\functional.py", line 2215, in mse_loss
expanded_input, expanded_target = torch.broadcast_tensors(input, target)
File "B:\tools and software\Anaconda\envs\pysyft-pytorch\lib\site-packages\torch\functional.py", line 52, in broadcast_tensors
return torch._C._VariableFunctions.broadcast_tensors(tensors)
RuntimeError: The size of tensor a (38) must match the size of tensor b (39) at non-singleton dimension 0

Answer 1

The error seems to say that the batch sizes of the target and the output are not the same have you tried printing the size of the target and output. If so what are the results. Also you might want to print the size of the input to the model to see if there is something off there. Sorry for posting this to an answer I can't comment yet.