Will the following be part of computational graph and gradients are computed correctly?

autograd
1

consider following code will this part of computation graph correctly and gradients are computed correctly.I mean can distributions be part of graph

import torch
import torch.distributions.normal as tdn
dist = tdn.Normal(mean,std) #assume network predicts mean,std
loss = -1 * dist.log_prob(y)

2

Hi,

I think the log_prob is differentiable for Normal yes. You can check that by verifying that if mean or std has requires_grad=True, then dist.log_prob(y).requires_grad == True.

In general, the output requires grad, that means that the function is differentiable using the autograd.

1 Like
3

Another question, tdn.cdf(y) will also differentiable right?
I was thinking if dist = tdn.Normal(mean,std) would mess computational graph.It will not write right?