TF 2.3.1 compatibility

 - Fixed 'L1' object has no attribute 'l2' (and vice versa for non-`l1_l2` objects)
 - Moved testing to TF2.3.1

.travis.yml

@@ -12,8 +12,8 @@ env:
     - TF_VERSION="1.14.0" KERAS_VERSION="2.2.5" TF_KERAS="1"
     - TF_VERSION="2.2.0"  KERAS_VERSION="2.3.0" TF_EAGER="1"
     - TF_VERSION="2.2.0"  KERAS_VERSION="2.3.0"
-    - TF_VERSION="2.3.0"  KERAS_VERSION="2.3.0" TF_KERAS="1"  TF_EAGER="1"
-    - TF_VERSION="2.3.0"  KERAS_VERSION="2.3.0" TF_KERAS="1"
+    - TF_VERSION="2.3.1"  KERAS_VERSION="2.3.0" TF_KERAS="1"  TF_EAGER="1"
+    - TF_VERSION="2.3.1"  KERAS_VERSION="2.3.0" TF_KERAS="1"
 
 notifications:
   email: false

keras_adamw/__init__.py

@@ -28,4 +28,4 @@
 from .utils import get_weight_decays, fill_dict_in_order
 from .utils import reset_seeds, K_eval
 
-__version__ = '1.37'
+__version__ = '1.38'

keras_adamw/utils.py

@@ -164,11 +164,8 @@ def _get_layer_penalties(layer, zero_penalties=False):
     for weight_name in ['kernel', 'bias']:
         _lambda = getattr(layer, weight_name + '_regularizer', None)
         if _lambda is not None:
-            l1l2 = (float(_lambda.l1), float(_lambda.l2))
+            l1l2 = _get_and_maybe_zero_penalties(_lambda, zero_penalties)
             penalties.append([getattr(layer, weight_name).name, l1l2])
-            if zero_penalties:
-                _lambda.l1 = np.array(0., dtype=_lambda.l1.dtype)
-                _lambda.l2 = np.array(0., dtype=_lambda.l2.dtype)
     return penalties
 
 
@@ -190,14 +187,21 @@ def _cell_penalties(rnn_cell, zero_penalties=False):
         _lambda = getattr(cell, weight_type + '_regularizer', None)
         if _lambda is not None:
             weight_name = cell.weights[weight_idx].name
-            l1l2 = (float(_lambda.l1), float(_lambda.l2))
+            l1l2 = _get_and_maybe_zero_penalties(_lambda, zero_penalties)
             penalties.append([weight_name, l1l2])
-            if zero_penalties:
-                _lambda.l1 = np.array(0., dtype=_lambda.l1.dtype)
-                _lambda.l2 = np.array(0., dtype=_lambda.l2.dtype)
     return penalties
 
 
+def _get_and_maybe_zero_penalties(_lambda, zero_penalties):
+    if zero_penalties:
+        if hasattr(_lambda, 'l1'):
+            _lambda.l1 = np.array(0., dtype=_lambda.l1.dtype)
+        if hasattr(_lambda, 'l2'):
+            _lambda.l2 = np.array(0., dtype=_lambda.l2.dtype)
+    return (float(getattr(_lambda, 'l1', 0.)),
+            float(getattr(_lambda, 'l2', 0.)))
+
+
 def fill_dict_in_order(_dict, values_list):
     for idx, key in enumerate(_dict.keys()):
         _dict[key] = values_list[idx]

tests/test_optimizers.py

@@ -90,8 +90,8 @@ def test_misc():  # tests of non-main features to improve coverage
         embed_input_dim = 5
 
         # arbitrarily select SGDW for coverage testing
-        l1_reg = 1e-4 if optimizer_name == 'SGDW' else 0
-        l2_reg = 1e-4 if optimizer_name != 'SGDW' else 0
+        l1_reg = 1e-4 if optimizer_name == 'SGDW' else None
+        l2_reg = 1e-4 if optimizer_name != 'SGDW' else None
         if optimizer_name == 'SGDW':
             optimizer_kw.update(dict(zero_penalties=False,
                                       weight_decays={},