What is the point of renaming the sample function in this tutorial?

1

The following is from official tutorial

def geometric(p, t=None):
    if t is None:
        t = 0
    x = pyro.sample("x_{}".format(t), pyro.distributions.Bernoulli(p))
    if x.item() == 1:
        return 0
    else:
        return 1 + geometric(p, t + 1)

print(geometric(0.5))

Why they’re renaming the sample function? I tried it without renaming and it is working fine

2

that’s because the model is just being run forwards. if you use it to do inference, your sample sites need to have unique names.

1 Like
3

Thanks. I recommend adding this to the documentation on the first page, it is illuminating.

4

In the pyro.sample function, what is the name of the sample exactly?
Is it not what x is cause by printing out x, we get the value of the sample.