Using @tensor with sparse tensors of SparseArrayKit.jl

According to the SparseArrayKit.jl documentation, sparse tensors implemented therein are compatible with the @tensor macro from the TensorOperations.jl package. The following simple code

function randn_sparse(T::Type{<:Number}, sz::Dims, p=0.5)
    a = SparseArray{T}(undef, sz)
    for I in keys(a)
        if rand() < p
            a[I] = randn(T)
    return a

X = randn_sparse(Float64, (L, L, L)) # Three-dimensional random tensor
O = ones(L) 

@tensor begin
    C[i,j] := X[i, j, k] * O[k] 

returns the error

MethodError: no method matching StridedViews.StridedView(::SparseArray{Float64, 3})

Does anyone know how to fix this ?