diff --git a/aten/src/ATen/native/sparse/cuda/SparseCUDATensorMath.cu b/aten/src/ATen/native/sparse/cuda/SparseCUDATensorMath.cu
index 753ea9fa4937..e1b7c3e7aa52 100644
--- a/aten/src/ATen/native/sparse/cuda/SparseCUDATensorMath.cu
+++ b/aten/src/ATen/native/sparse/cuda/SparseCUDATensorMath.cu
@@ -338,8 +338,8 @@ Tensor& add_out_dense_sparse_cuda(Tensor& r_, const Tensor& dense, const SparseT
     if (sparse.dense_dim() == 0) {
       TORCH_CHECK(cuda::getApplyGrid(nnz, grid, curDevice), "add: Argument #0: tensor too large or too many dimensions");
 
-      AT_DISPATCH_ALL_TYPES_AND2(
-        at::ScalarType::Half, at::ScalarType::BFloat16, commonDtype, "add_out_dense_sparse_cuda", [&] {
+      AT_DISPATCH_ALL_TYPES_AND3(
+        at::ScalarType::Bool, at::ScalarType::Half, at::ScalarType::BFloat16, commonDtype, "add_out_dense_sparse_cuda", [&] {
           apply::sparseElementwiseKernelScalar<TensorCAddOp<scalar_t>, uint64_t, scalar_t>
             <<<grid, block, 0, stream>>>(
               TensorCAddOp<scalar_t>(value.to<scalar_t>()),