StableRNG in Turing is not producing reproducible output

The problem seems to be in the initial step_size estimation. The alternative result is always given when the initial step_size is alternative as well. In any case, the following:

sample(StableRNG(123), example(1), NUTS(1000, 0.65; init_ϵ = 1.6), 1000)

sets the step size and now only one result is output.
Will investigate further.

1 Like