diff --git a/paddle/gserver/tests/test_LayerGrad.cpp b/paddle/gserver/tests/test_LayerGrad.cpp
index f63c93c9430c2a5028c8067bff29e31dd4e42672..eea884cb5021a70cbcba4b855bf6616f5ac6d5af 100644
--- a/paddle/gserver/tests/test_LayerGrad.cpp
+++ b/paddle/gserver/tests/test_LayerGrad.cpp
@@ -2371,10 +2371,8 @@ void testFactorizationMachineLayer(InputType type, bool useGpu) {
 
 TEST(Layer, FactorizationMachineLayer) {
   testFactorizationMachineLayer(INPUT_DATA, false);
-  testFactorizationMachineLayer(INPUT_SPARSE_FLOAT_VALUE_DATA, false);
-#ifdef PADDLE_WITH_CUDA
   testFactorizationMachineLayer(INPUT_DATA, true);
-#endif
+  testFactorizationMachineLayer(INPUT_SPARSE_FLOAT_VALUE_DATA, false);
 }
 
 int main(int argc, char** argv) {
diff --git a/python/paddle/trainer_config_helpers/tests/configs/test_factorization_machine.py b/python/paddle/trainer_config_helpers/tests/configs/test_factorization_machine.py
index 62ceb359cf8f9a8d281cc95d8b09255b29c272e6..b249de0fee3c8ca4ad0520872fa2497c493d31b5 100644
--- a/python/paddle/trainer_config_helpers/tests/configs/test_factorization_machine.py
+++ b/python/paddle/trainer_config_helpers/tests/configs/test_factorization_machine.py
@@ -1,7 +1,5 @@
 from paddle.trainer_config_helpers import *
 
-settings(batch_size=1000, learning_rate=1e-5)
-
 data = data_layer(name='data', size=1024)
 
 fm = factorization_machine(input=data, factor_size=10)