intro_dgm
intro_dgm copied to clipboard
Update vae_example.ipynb
I guess, D has forgotten in the equation.
No, this is not correct because log_var
is of size mini_batch x D
. When we calculate log_p
, we treat all variables as tensors. Adding D
as you suggest would result in scaling log_var
which would be incorrect.
However, I spotted a different mistake. The constant shouldn't have D
. I am going to correct it.