Moved encoder and decoder registries in base input and output feature…

…s to enforce every new feature type to have them

ludwig/features/base_feature.py

@@ -19,7 +19,7 @@
 
 from ludwig.models.modules.fully_connected_modules import FCStack
 from ludwig.models.modules.reduction_modules import reduce_sequence
-from ludwig.utils.misc import merge_dict
+from ludwig.utils.misc import merge_dict, get_from_registry
 from ludwig.utils.tf_utils import sequence_length_3D
 
 
@@ -66,6 +66,16 @@ def update_model_definition_with_metadata(
     def populate_defaults(input_feature):
         pass
 
+    @property
+    @abstractmethod
+    def encoder_registry(self):
+        pass
+
+    def initialize_encoder(self, encoder_parameters):
+        return get_from_registry(self.encoder, self.encoder_registry)(
+            **encoder_parameters
+        )
+
 
 class OutputFeature(ABC, BaseFeature, tf.keras.Model):
 
@@ -110,6 +120,16 @@ def __init__(self, feature):
             # default_bias_constraint=None,
         )
 
+    @property
+    @abstractmethod
+    def decoder_registry(self):
+        pass
+
+    def initialize_decoder(self, decoder_parameters):
+        return get_from_registry(self.decoder, self.decoder_registry)(
+            **decoder_parameters
+        )
+
     def train_loss(self, targets, predictions):
         return self.train_loss_function(targets, predictions)
 

ludwig/features/numerical_feature.py

@@ -33,7 +33,7 @@
 from ludwig.models.modules.metric_modules import R2Score
 from ludwig.models.modules.numerical_decoders import Regressor
 from ludwig.models.modules.numerical_encoders import NumericalPassthroughEncoder
-from ludwig.utils.misc import set_default_value, get_from_registry
+from ludwig.utils.misc import set_default_value
 from ludwig.utils.misc import set_default_values
 
 logger = logging.getLogger(__name__)
@@ -109,12 +109,7 @@ def __init__(self, feature, encoder_obj=None):
         if encoder_obj:
             self.encoder_obj = encoder_obj
         else:
-            self.encoder_obj = self.get_numerical_encoder(encoder_parameters)
-
-    def get_numerical_encoder(self, encoder_parameters):
-        return get_from_registry(self.encoder, numerical_encoder_registry)(
-            **encoder_parameters
-        )
+            self.encoder_obj = self.initialize_encoder(encoder_parameters)
 
     def call(self, inputs, training=None, mask=None):
         assert isinstance(inputs, tf.Tensor)
@@ -141,15 +136,14 @@ def update_model_definition_with_metadata(
     def populate_defaults(input_feature):
         set_default_value(input_feature, TIED, None)
 
-
-numerical_encoder_registry = {
-    'dense': FCStack,
-    'passthrough': NumericalPassthroughEncoder,
-    'null': NumericalPassthroughEncoder,
-    'none': NumericalPassthroughEncoder,
-    'None': NumericalPassthroughEncoder,
-    None: NumericalPassthroughEncoder
-}
+    encoder_registry = {
+        'dense': FCStack,
+        'passthrough': NumericalPassthroughEncoder,
+        'null': NumericalPassthroughEncoder,
+        'none': NumericalPassthroughEncoder,
+        'None': NumericalPassthroughEncoder,
+        None: NumericalPassthroughEncoder
+    }
 
 
 class NumericalOutputFeature(NumericalBaseFeature, OutputFeature):
@@ -165,16 +159,11 @@ def __init__(self, feature):
 
         decoder_parameters = self.overwrite_defaults(feature)
 
-        self.decoder_obj = self.get_numerical_decoder(decoder_parameters)
+        self.decoder_obj = self.initialize_decoder(decoder_parameters)
 
         self._setup_loss()
         self._setup_metrics()
 
-    def get_numerical_decoder(self, decoder_parameters):
-        return get_from_registry(self.decoder, numerical_decoder_registry)(
-            **decoder_parameters
-        )
-
     def predictions(
             self,
             inputs,  # hidden
@@ -303,11 +292,10 @@ def populate_defaults(output_feature):
             }
         )
 
-
-numerical_decoder_registry = {
-    'regressor': Regressor,
-    'null': Regressor,
-    'none': Regressor,
-    'None': Regressor,
-    None: Regressor
-}
+    decoder_registry = {
+        'regressor': Regressor,
+        'null': Regressor,
+        'none': Regressor,
+        'None': Regressor,
+        None: Regressor
+    }

ludwig/models/modules/numerical_decoders.py

@@ -16,6 +16,8 @@
 # ==============================================================================
 import logging
 
+import tensorflow as tf
+from tensorflow.keras.layers import Dense
 from tensorflow.keras.layers import Layer
 
 logger = logging.getLogger(__name__)