I am unable to use Turing models with pmap from Atom. However, the code below works if executed directly from the REPL. Here is the error message:
On worker 2:
UndefVarError: ###my_model#336 not defined
deserialize_datatype at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.4/Serialization/src/Serialization.jl:1211
handle_deserialize at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.4/Serialization/src/Serialization.jl:788
Details
Atom version: 1.45
Julia version: 1.4.1 (also 1.3 and 1.4)
OS: Ubuntu 18.04
Package versions:
Atom.jl: v0.12.10
julia-client: 0.12.4
ink:. 0.12.3
Steps to reproduce
In order to reproduce the error, execute the code below using Atom.
using Distributed, Turing
addprocs(4)
@everywhere using Turing
@everywhere begin
@model my_model() = begin
end
end
pmap(x->my_model, 1:4)
I am unable to use Turing models with pmap from Atom. However, the code below works if executed directly from the REPL. Here is the error message:
Details
Atom version: 1.45
Julia version: 1.4.1 (also 1.3 and 1.4)
OS: Ubuntu 18.04
Package versions:
Steps to reproduce
In order to reproduce the error, execute the code below using Atom.