Is there a way to create a tensor using a per-channel means and std? For instance, I have two tensors `mean`

and `std`

with `k`

values each, and I would like to generate an `output`

tensor which is a stack of `k`

tensors generated from normal distributions using each mean and std value. What would be an efficient way to do it? Essentially, I would like to do the following:

```
output = torch.empty(shape)
for i in range(k):
output[:, i, ...] = output[:, i, ...].normal_(mean=mean[i], std=std[i])
```

Is there a better way of achieving this? Thanks in advance!