search

JuliaAI / MLJLIBSVMInterface.jl

An implementation of the MLJ model interface for support vector machines provided by LIBSVM.jl
MIT License
3 stars 2 forks source link

Tests now fail #18

Open ablaom opened 7 months ago

ablaom commented 7 months ago

With no change in this repo since the last release, tests are failing locally for me.

An analysis suggests the culprit is LinearSVC:

using Pkg
Pkg.activate(temp=true)
Pkg.add("MLJBase")
Pkg.add("MLJLIBSVMInterface")

using MLJLIBSVMInterface, MLJBase

X, y = @load_iris
model = MLJLIBSVMInterface.LinearSVC()

mach = machine(model, X, y)
fit!(mach)

# [ Info: Training machine(LinearSVC(solver = L2R_L2LOSS_SVC_DUAL, …), …).
# ┌ Error: Problem fitting the machine machine(LinearSVC(solver = L2R_L2LOSS_SVC_DUAL, …), …).
# └ @ MLJBase ~/.julia/packages/MLJBase/xcS6N/src/machines.jl:683
# [ Info: Running type checks...
# [ Info: Type checks okay.
# ERROR: Please check your parameters: Initial-solution specification supported only for solvers L2R_LR, L2R_L2LOSS_SVC, and L2R_L2LOSS_SVR
# Stacktrace:
#   [1] error(s::String)
#     @ Base ./error.jl:35
#   [2] linear_train(labels::Vector{…}, instances::LinearAlgebra.Adjoint{…}; weights::Nothing, solver_type::Int32, eps::Float64, C::Float64, p::Float64, init_sol::Ptr{…}, bias::Float64,
# verbose::Bool)
#     @ LIBLINEAR ~/.julia/packages/LIBLINEAR/77yny/src/LIBLINEAR.jl:227
ablaom commented 7 months ago

It seems a fix will be provided by