You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are several CI test failures in TuringLang/Turing.jl#2341 but I think they all stem from this;
I constructed an MWE that won't require using that feature branch:
using DynamicPPL:@model, LogDensityFunction
using Distributions: Beta
using ADTypes: AutoMooncake
using LogDensityProblems: logdensity_and_gradient
using LogDensityProblemsAD: ADgradient
import Mooncake
@modelf() = p ~Beta(2, 2)
ℓ =ADgradient(:Mooncake, LogDensityFunction(f()))
logdensity_and_gradient(ℓ, [0.5])
As far as I can tell, it's only constrained distributions that trigger this, so Normal(0, 1) is fine but truncated(Normal(0, 1), 0, Inf) gives the same error.
There are several CI test failures in TuringLang/Turing.jl#2341 but I think they all stem from this;
I constructed an MWE that won't require using that feature branch:
As far as I can tell, it's only constrained distributions that trigger this, so
Normal(0, 1)
is fine buttruncated(Normal(0, 1), 0, Inf)
gives the same error.Error traceback
Manifest
The text was updated successfully, but these errors were encountered: