Is there a standard way in pytorch to compare convergence speed / efficiency between models?

Say, if I have two models, both have 4 inputs and 2 outputs and one hidden layer, the first model has 128 nodes, the second 64 nodes. How do I know which is better, in terms of speed and variance, etc. in a quantative way? The models could have two or more layers as well.

1 Like

please, any answer to this question?