Fix hybrid_parallel_sharding_model.py ut (#55269)

* fix hybrid_parallel_sharding_model.py

* Update hybrid_parallel_sharding_model.py

python/paddle/fluid/tests/unittests/collective/fleet/hybrid_parallel_sharding_model.py

@@ -296,36 +296,38 @@ class TestDistMPTraning(unittest.TestCase):
 
     def test_sharding_adam(self):
         sharded_accumulators = {
-            'linear_0.w_0_moment1_0',
+            "linear_0.b_0_moment2_0",
+            'embedding_0.w_0_beta1_pow_acc_0',
+            'linear_2.b_0_beta2_pow_acc_0',
+            'linear_0.b_0_beta1_pow_acc_0',
+            'linear_2.b_0_moment2_0',
+            'linear_0.b_0_beta2_pow_acc_0',
             'linear_1.b_0_moment1_0',
+            'embedding_0.w_0_moment2_0',
+            'linear_1.b_0_moment2_0',
+            'linear_2.b_0_beta1_pow_acc_0',
+            'linear_0.b_0_moment1_0',
             'linear_2.b_0_moment1_0',
             'embedding_0.w_0_moment1_0',
-            'linear_0.w_0_moment2_0',
-            'linear_1.b_0_moment2_0',
-            'linear_2.b_0_moment2_0',
-            'embedding_0.w_0_moment2_0',
-            'linear_0.w_0_beta1_pow_acc_0',
+            'embedding_0.w_0_beta2_pow_acc_0',
             'linear_1.b_0_beta1_pow_acc_0',
-            'linear_2.b_0_beta1_pow_acc_0',
-            'embedding_0.w_0_beta1_pow_acc_0',
-            'linear_0.w_0_beta2_pow_acc_0',
             'linear_1.b_0_beta2_pow_acc_0',
-            'linear_2.b_0_beta2_pow_acc_0',
-            'embedding_0.w_0_beta2_pow_acc_0',
         }
         self.sharding_model(
-            Optimizer="adam", sharded_accumulators=sharded_accumulators
+            Optimizer="adam",
+            sharded_accumulators=sharded_accumulators,
         )
 
     def test_sharding_momentum(self):
         sharded_accumulators = {
-            'linear_6.w_0_velocity_0',
-            'linear_7.b_0_velocity_0',
             'linear_8.b_0_velocity_0',
             'embedding_2.w_0_velocity_0',
+            'linear_6.b_0_velocity_0',
+            'linear_7.b_0_velocity_0',
         }
         self.sharding_model(
-            Optimizer="Momentum", sharded_accumulators=sharded_accumulators
+            Optimizer="Momentum",
+            sharded_accumulators=sharded_accumulators,
         )