Sherlock LoadError on small network

[castrong]

I get the following error and stack trace when running Sherlock with epsilon = 0.5 on a small example:

ERROR: LoadError: MethodError: Cannot convert an object of type Hyperrectangle{Float64} to an object of type Array{#s13,1} where #s13<:AbstractPolytope
Closest candidates are:
convert(::Type{T<:Array}, ::AbstractArray) where T<:Array at array.jl:474
convert(::Type{T<:AbstractArray}, ::T<:AbstractArray) where T<:AbstractArray at abstractarray.jl:14
convert(::Type{T<:AbstractArray}, ::LinearAlgebra.Factorization) where T<:AbstractArray at /Users/sabae/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.2/LinearAlgebra/src/factorization.jl:46
...
Stacktrace:
[1] ReachabilityResult(::Symbol, ::Hyperrectangle{Float64}) at /Users/cstrong/Desktop/Stanford/Research/NV_Fork/NeuralVerification.jl/src/utils/problem.jl:114
[2] interpret_result(::Hyperrectangle{Float64}, ::Hyperrectangle{Float64}, ::Hyperrectangle{Float64}, ::Array{Float64,1}, ::Array{Float64,1}) at /Users/cstrong/Desktop/Stanford/Research/NV_Fork/NeuralVerification.jl/src/satisfiability/bab.jl:48
[3] solve(::Sherlock, ::Problem{Hyperrectangle{Float64},Hyperrectangle{Float64}}) at /Users/cstrong/Desktop/Stanford/Research/NV_Fork/NeuralVerification.jl/src/satisfiability/sherlock.jl:41
[4] top-level scope at /Users/cstrong/Desktop/Stanford/Research/NV_Fork/NeuralVerification.jl/examples/SimpleProblem.jl:12
[5] include at ./boot.jl:328 [inlined]
[6] include_relative(::Module, ::String) at ./loading.jl:1094
[7] include(::Module, ::String) at ./Base.jl:31
[8] include(::String) at ./client.jl:431
[9] top-level scope at none:0
in expression starting at /Users/cstrong/Desktop/Stanford/Research/NV_Fork/NeuralVerification.jl/examples/SimpleProblem.jl:12

Code to reproduce this and network are shown below:

# Setup problem
network = NeuralVerification.read_nnet("./test/test_sets/random/small/networks/rand_[1,3,1]_1.nnet")
input_set = NeuralVerification.Hyperrectangle([0.12133537641044301], [0.2612909441775033])
output_set = NeuralVerification.Hyperrectangle([-0.2053508574991203], [0.5961870200944133])
problem = NeuralVerification.Problem(network, input_set, output_set)

# Sherlock issue
println(NeuralVerification.solve(NeuralVerification.Sherlock(ϵ = 0.5), problem))

Network:

//Neural Network File Format by Kyle Julian, Stanford 2016.
2, 1, 1, 3,
1, 3, 1,
This line extraneous
-6.55e4,
6.55e4,
0.0,0.0,
1.0,1.0,
0.5787480927621855,
-0.07673617148366274,
0.924732195346472,
-0.9480930058986625,
-0.24850641735286505,
0.6232748287772321,
-0.5576518964746899, -0.33490823749601883, -0.717037022479706,
0.559366971361952,