Add lstm sigmoid dropout test for dependency optimization (#2027)

* Add lstm sigmoid_dropout test for dependency optimization

Signed-off-by: Deyu Huang <[email protected]>

Author: hwangdeyu

tests/common.py

@@ -454,7 +454,7 @@ def group_nodes_by_type(graph):
 
 
 def check_op_count(graph, op_type, expected_count, disabled=True):
-    # FIXME: after switching to grappler some of the op counts are off. Fix later.
+    # The grappler optimization may change some of the op counts.
     return disabled or len(group_nodes_by_type(graph)[op_type]) == expected_count
 
 

tests/test_lstm.py

@@ -10,7 +10,7 @@
 from tensorflow.python.ops import variable_scope
 from backend_test_base import Tf2OnnxBackendTestBase
 from common import check_tf_min_version, unittest_main, check_opset_after_tf_version, \
-    skip_tf2, skip_tf_versions, check_op_count
+    skip_tf2, skip_tf_versions, check_op_count, skip_tfjs
 
 from tf2onnx.tf_loader import is_tf2
 
@@ -51,6 +51,7 @@ def new_graph_validator(g):
                 # Skip checks for tflite graphs (no ":" in outputs)
                 return good
             good = good and check_op_count(g, "LSTM", require_lstm_count, disabled=False)
+            # If LSTM op rewriter failed to work, Loop op will be shown in general.
             good = good and check_op_count(g, "Loop", 0, disabled=False)
             return good
         try:
@@ -774,5 +775,23 @@ def func(x):
             return tf.identity(y[0], name="output"), tf.identity(y[1], name="output1")
         self.run_test_case(func, {"input:0": x_val}, [], ["output:0", "output1:0"], rtol=1e-05, atol=1e-06)
 
+    @check_tf_min_version("2.0")
+    @skip_tfjs("TFJS converts model incorrectly")
+    def test_keras_lstm_sigmoid_dropout(self):
+        in_shape = [16, 16]
+        batch_size = 2
+        x_val = np.random.uniform(size=[batch_size] + in_shape).astype(np.float32)
+
+        model = tf.keras.models.Sequential()
+        model_in = tf.keras.layers.Input(shape=tuple(in_shape), name="input")
+        lstm = tf.keras.layers.LSTM(16, activation='sigmoid', dropout=0.1)
+        model.add(model_in)
+        model.add(lstm)
+
+        def func(x):
+            y = model(x)
+            return tf.identity(y[0], name="output")
+        self.run_test_case(func, {"input:0": x_val}, [], ["output:0"], rtol=1e-05, atol=1e-06)
+
 if __name__ == '__main__':
     unittest_main()