Initialize random_state param to None (#36)

* Update tabnet.py

Solve issue "unsupported operand type(s) for ** or pow(): 'NoneType' and 'dict' " in TabNetRegressor

* Add random_state param in stacked_tabnet to ensure reproducibility

* Initialize random_state param to None

tabnet/stacked_tabnet.py

@@ -167,7 +167,7 @@ def __init__(self, feature_columns,
                  virtual_batch_size=None,
                  num_groups=2,
                  epsilon=1e-5,
-                 random_state=random_state,
+                 random_state=None,
                  **kwargs):
         """
         Tensorflow 2.0 implementation of [TabNet: Attentive Interpretable Tabular Learning](https://arxiv.org/abs/1908.07442)
@@ -272,7 +272,7 @@ def __init__(self, feature_columns,
                  virtual_batch_size=None,
                  num_groups=2,
                  epsilon=1e-5,
-                 random_state=random_state,
+                 random_state=None,
                  **kwargs):
         """
         Tensorflow 2.0 implementation of [TabNet: Attentive Interpretable Tabular Learning](https://arxiv.org/abs/1908.07442)