Remove problematic source tensor sorting

This is no longer required, and is problematic for models that
have an output that is used other places in the model (since
the sorting puts all outputs at the end).

Also ensure a better error for unconverted input tensor

CHANGES.rst

@@ -75,6 +75,8 @@ Release history
 - Added ``tensorflow-cpu`` distributions to installation checks (so Nengo DL will
   not attempt to reinstall TensorFlow if ``tensorflow-cpu`` is already installed).
   (`#142`_)
+- Fixed bug when applying the Converter to Keras models that re-use intermediate
+  layers as output layers. (`#137`_)
 
 **Deprecated**
 
@@ -88,6 +90,7 @@ Release history
 .. _#129: https://github.com/nengo/nengo-dl/pull/129
 .. _#130: https://github.com/nengo/nengo-dl/pull/130
 .. _#134: https://github.com/nengo/nengo-dl/pull/134
+.. _#137: https://github.com/nengo/nengo-dl/pull/137
 .. _#139: https://github.com/nengo/nengo-dl/pull/139
 .. _#140: https://github.com/nengo/nengo-dl/pull/140
 .. _#142: https://github.com/nengo/nengo-dl/pull/142

nengo_dl/converter.py

@@ -627,10 +627,7 @@ def get_input_obj(self, node_id, tensor_idx=0):
 
         input_layer, input_node_id, input_tensor_id = self.get_history(tensor)
 
-        if input_node_id in self.converter.layer_map[input_layer]:
-            return self.converter.layer_map[input_layer][input_node_id][input_tensor_id]
-        else:
-            return None
+        return self.converter.layer_map[input_layer][input_node_id][input_tensor_id]
 
     def _get_shape(self, input_output, node_id, include_batch=False):
         """
@@ -820,18 +817,6 @@ def convert(self, node_id):
         # that need to be built into the Nengo network
         source_tensors = self.trace_tensors(self.layer.outputs)
 
-        def sort_key(x):
-            # sort tensors so that order of model inputs/outputs is preserved
-            for i, y in enumerate(self.layer.inputs):
-                if x is y:
-                    return -(len(self.layer.inputs) - i)
-            for i, y in enumerate(self.layer.outputs):
-                if x is y:
-                    return i + 1
-            return 0
-
-        source_tensors = sorted(source_tensors, key=sort_key)
-
         for tensor in source_tensors:
             # look up the layer/node to be converted
             model_layer, model_node_id, _ = self.get_history(tensor)
@@ -1467,11 +1452,11 @@ class ConvertInput(LayerConverter):
     """Convert ``tf.keras.layers.InputLayer`` to Nengo objects."""
 
     def convert(self, node_id):
-        # if this input layer has an input obj, that means it is a passthrough
-        # (so we just return the input)
-        output = self.get_input_obj(node_id)
-
-        if output is None:
+        try:
+            # if this input layer has an input obj, that means it is a passthrough
+            # (so we just return the input)
+            output = self.get_input_obj(node_id)
+        except KeyError:
             # not a passthrough input, so create input node
             shape = self.output_shape(node_id)
             if any(x is None for x in shape):

nengo_dl/tests/test_converter.py

@@ -701,3 +701,20 @@ def test_layer_dicts():
 
     assert conv.outputs[x0].target is conv.layers[x0]
     assert conv.outputs[x1].target is conv.layers[x1]
+
+
+def test_mid_model_output(Simulator):
+    """Check that converter supports output tensors from the middle of the model.
+
+    Previous converter put output tensors last in build order, so having an output
+    tensor that needed to be built before non-output tensors was problematic.
+    https://github.com/nengo/nengo-dl/pull/137
+    """
+
+    # model must have at least three layers, with one layer in between outputs
+    inp = tf.keras.Input(shape=(1,))
+    x0 = tf.keras.layers.ReLU()(inp)
+    x1 = tf.keras.layers.ReLU()(x0)
+    x2 = tf.keras.layers.ReLU()(x1)
+
+    _test_convert(inp, [x0, x2], inp_vals=[np.ones((4, 1))])