diff --git a/Project.toml b/Project.toml
index b81560b53..d78ecb5cb 100644
--- a/Project.toml
+++ b/Project.toml
@@ -22,6 +22,7 @@ Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
+SuiteSparseGraphBLAS = "c2e53296-7b14-11e9-1210-bddfa8111e1d"
 TestEnv = "1e6cf692-eddd-4d53-88a5-2d735e33781b"
 
 [compat]
diff --git a/src/GNNGraphs/GNNGraphs.jl b/src/GNNGraphs/GNNGraphs.jl
index 9177eb34f..d1cea7af0 100644
--- a/src/GNNGraphs/GNNGraphs.jl
+++ b/src/GNNGraphs/GNNGraphs.jl
@@ -14,6 +14,7 @@ using LearnBase: getobs
 import KrylovKit
 using ChainRulesCore
 using LinearAlgebra, Random
+using SuiteSparseGraphBLAS: GBMatrix
 
 include("gnngraph.jl")
 export GNNGraph, 
diff --git a/src/GNNGraphs/convert.jl b/src/GNNGraphs/convert.jl
index 0ce98ebcc..63d313d3f 100644
--- a/src/GNNGraphs/convert.jl
+++ b/src/GNNGraphs/convert.jl
@@ -124,7 +124,7 @@ function to_sparse(A::ADJMAT_T, T::DataType=eltype(A); dir=:out, num_nodes=nothi
         A = T.(A)
     end
     if !(A isa AbstractSparseMatrix)
-        A = sparse(A)
+        A = GBMatrix(sparse(A))
     end 
     return A, num_nodes, num_edges
 end
@@ -140,7 +140,7 @@ function to_sparse(coo::COO_T, T::DataType=Int; dir=:out, num_nodes=nothing)
     num_nodes = isnothing(num_nodes) ? max(maximum(s), maximum(t)) : num_nodes 
     A = sparse(s, t, eweight, num_nodes, num_nodes)
     num_edges = length(s)
-    return A, num_nodes, num_edges
+    return GBMatrix(A), num_nodes, num_edges
 end
 
 
diff --git a/test/runtests.jl b/test/runtests.jl
index cfa8d9888..3153f6dab 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -37,7 +37,7 @@ tests = [
 
 !CUDA.functional() && @warn("CUDA unavailable, not testing GPU support")
 
-@testset "GraphNeuralNetworks: graph format $graph_type" for graph_type in (:coo, :dense, :sparse) 
+@testset "GraphNeuralNetworks: graph format $graph_type" for graph_type in (:sparse, :coo, :dense,) 
     global GRAPH_T = graph_type
     global TEST_GPU = CUDA.functional() && (GRAPH_T != :sparse)