diff --git a/tests/python/unittest/test_gluon_probability_v1.py b/tests/python/unittest/test_gluon_probability_v1.py
index 92721f610495..5dcbb6bfb34f 100644
--- a/tests/python/unittest/test_gluon_probability_v1.py
+++ b/tests/python/unittest/test_gluon_probability_v1.py
@@ -540,7 +540,7 @@ def hybrid_forward(self, F, n, params, *args):
     # Test log_prob
     for shape, hybridize, use_logit in itertools.product(shapes, [True, False], [True, False]):
         n = np.random.randint(1, 10, size=shape).astype('float32')
-        prob = np.random.uniform(low=0.1, size=shape)
+        prob = np.random.uniform(low=0.2, high=0.6, size=shape)
         sample = np.random.randint(0, 10, size=shape).astype('float32')
         param = prob
         if use_logit:
@@ -559,7 +559,7 @@ def hybrid_forward(self, F, n, params, *args):
         for func in ['mean', 'variance']:
             for use_logit in [True, False]:
                 n = np.random.randint(1, 10, size=shape).astype('float32')
-                prob = np.random.uniform(low=0.1, size=shape)
+                prob = np.random.uniform(low=0.2, high=0.6, size=shape)
                 net = TestNegativeBinomial(func, use_logit)
                 param = prob
                 if use_logit:
diff --git a/tests/python/unittest/test_gluon_probability_v2.py b/tests/python/unittest/test_gluon_probability_v2.py
index 9a36b4fc7056..77e25f7f7c5b 100644
--- a/tests/python/unittest/test_gluon_probability_v2.py
+++ b/tests/python/unittest/test_gluon_probability_v2.py
@@ -540,7 +540,7 @@ def forward(self, n, params, *args):
     # Test log_prob
     for shape, hybridize, use_logit in itertools.product(shapes, [True, False], [True, False]):
         n = np.random.randint(1, 10, size=shape).astype('float32')
-        prob = np.random.uniform(low=0.1, size=shape)
+        prob = np.random.uniform(low=0.2, high=0.6, size=shape)
         sample = np.random.randint(0, 10, size=shape).astype('float32')
         param = prob
         if use_logit:
@@ -559,7 +559,7 @@ def forward(self, n, params, *args):
         for func in ['mean', 'variance']:
             for use_logit in [True, False]:
                 n = np.random.randint(1, 10, size=shape).astype('float32')
-                prob = np.random.uniform(low=0.1, size=shape)
+                prob = np.random.uniform(low=0.2, high=0.6, size=shape)
                 net = TestNegativeBinomial(func, use_logit)
                 param = prob
                 if use_logit: