How to get predict result from model's forward in a number

Here is my model:

class Model(nn.Module):
	def __init__(self):
		self.layer = nn.Sequential(
            nn.Linear(100, 100),
	def forward(self, arg):
		return self.layer(arg)
model = Model()
output = model(arg)

The output is FloatTensor.
For example, with the MNIST task, how can I convery the output into a number?

torch.max(output, 1)[1]