⚡ Improving performance.

kashgari/__init__.py

@@ -13,6 +13,7 @@
 import os
 os.environ['TF_KERAS'] = '1'
 
+from kashgari import layers
 from kashgari import corpus
 from kashgari import embeddings
 from kashgari import macros

kashgari/layers.py

@@ -0,0 +1,21 @@
+# encoding: utf-8
+
+# author: BrikerMan
+# contact: eliyar917@gmail.com
+# blog: https://eliyar.biz
+
+# file: layers.py
+# time: 2019-05-21 18:55
+
+import tensorflow as tf
+from tensorflow.python import keras
+
+L = keras.layers
+
+if tf.test.is_gpu_available(cuda_only=True):
+    UnifiedLSTM = L.CuDNNLSTM
+else:
+    UnifiedLSTM = L.LSTM
+
+if __name__ == "__main__":
+    print("Hello world")

kashgari/tasks/labeling/base_model.py

@@ -122,7 +122,8 @@ def fit(self,
             batch_size: Number of samples per gradient update, default to 64.
             epochs: Integer. Number of epochs to train the model. default 5.
             fit_kwargs: fit_kwargs: additional arguments passed to ``fit_generator()`` function from
-                ``tensorflow.keras.Model`` - https://www.tensorflow.org/api_docs/python/tf/keras/models/Model#fit_generator
+                ``tensorflow.keras.Model``
+                - https://www.tensorflow.org/api_docs/python/tf/keras/models/Model#fit_generator
             **kwargs:
 
         Returns:
@@ -134,13 +135,13 @@ def fit(self,
             x_validate = utils.wrap_as_tuple(x_validate)
             y_validate = utils.wrap_as_tuple(y_validate)
         if self.embedding.token_count == 0:
-            if x_validate is not None:
-                x_all = (x_train + x_validate)
-                y_all = (y_train + y_validate)
-            else:
-                x_all = x_train
-                y_all = y_train
-            self.embedding.analyze_corpus(x_all, y_all)
+            # if x_validate is not None:
+            #     y_all = (y_train + y_validate)
+            #     x_all = (x_train + x_validate)
+            # else:
+            #     x_all = x_train.copy()
+            #     y_all = y_train.copy()
+            self.embedding.analyze_corpus(x_train, y_train)
 
         if self.tf_model is None:
             self.build_model_arc()
@@ -158,8 +159,8 @@ def fit(self,
                                                            batch_size)
 
             fit_kwargs['validation_data'] = validation_generator
-            fit_kwargs['validation_steps'] = len(x_validate) // batch_size
-
+            fit_kwargs['validation_steps'] = len(x_validate[0]) // batch_size
+        print(fit_kwargs)
         self.tf_model.fit_generator(train_generator,
                                     steps_per_epoch=len(x_train[0]) // batch_size,
                                     epochs=epochs,

kashgari/tasks/labeling/models.py

@@ -13,6 +13,8 @@
 from tensorflow import keras
 
 from kashgari.tasks.labeling.base_model import BaseLabelingModel
+from kashgari.layers import UnifiedLSTM
+
 
 L = keras.layers
 
@@ -49,7 +51,7 @@ def build_model_arc(self):
         config = self.hyper_parameters
         embed_model = self.embedding.embed_model
 
-        layer_blstm = L.Bidirectional(L.LSTM(**config['layer_blstm']),
+        layer_blstm = L.Bidirectional(UnifiedLSTM(**config['layer_blstm']),
                                       name='layer_blstm')
 
         layer_dropout = L.Dropout(**config['layer_dropout'],
@@ -106,8 +108,8 @@ def build_model_arc(self):
 
         layer_conv = L.Conv1D(**config['layer_conv'],
                               name='layer_conv')
-        layer_lstm = L.LSTM(**config['layer_lstm'],
-                            name='layer_lstm')
+        layer_lstm = UnifiedLSTM(**config['layer_lstm'],
+                                 name='layer_lstm')
         layer_dropout = L.Dropout(**config['layer_dropout'],
                                   name='layer_dropout')
         layer_time_distributed = L.TimeDistributed(L.Dense(output_dim,

tests/test_labeling.py

@@ -33,7 +33,7 @@ def setUpClass(cls):
 
     def test_basic_use_build(self):
         model = self.model_class()
-        model.fit(valid_x, valid_y, epochs=1)
+        model.fit(valid_x, valid_y, valid_x, valid_y, epochs=1)
         assert True
 
     def test_w2v_model(self):