Expand test coverage!

test/feedforward_test.py

@@ -84,6 +84,11 @@ def test_predict_proba_onelayer(self):
         z = net.predict_proba(self.INPUTS)
         self.assert_shape(z.shape, self.NUM_CLASSES)
 
+    def test_predict_logit_onelayer(self):
+        net = self._build(13)
+        z = net.predict_logit(self.INPUTS)
+        self.assert_shape(z.shape, self.NUM_CLASSES)
+
     def test_predict_twolayer(self):
         net = self._build(13, 14)
         z = net.predict(self.INPUTS)
@@ -136,7 +141,7 @@ def test_score_onelayer(self):
 
     def test_encode_onelayer(self):
         net = self._build(13)
-        z = net.encode(self.INPUTS)
+        z = net.encode(self.INPUTS, 'hid1')
         self.assert_shape(z.shape, 13)
 
     def test_encode_twolayer(self):

test/graph_test.py

@@ -52,6 +52,12 @@ def test_layer_tied(self):
         assert isinstance(m.layers[2], theanets.layers.feedforward.Tied)
         assert m.layers[2].partner is m.layers[1]
 
+    def test_layer_tied_partner(self):
+        m = theanets.Regressor((1, 2, dict(size=1, form='tied', partner='hid1')))
+        assert len(m.layers) == 3
+        assert isinstance(m.layers[2], theanets.layers.feedforward.Tied)
+        assert m.layers[2].partner is m.layers[1]
+
     def test_layer_tied_no_partner(self):
         try:
             theanets.Regressor((1, (2, 'tied'), (2, 'tied'), (1, 'tied')))
@@ -99,6 +105,12 @@ def test_find_missing(self):
         except KeyError:
             pass
 
+    def test_train(self):
+        m = theanets.Regressor((1, 2, 1))
+        tm, vm = m.train([np.random.randn(100, 1).astype('f'),
+                          np.random.randn(100, 1).astype('f')])
+        assert tm['loss'] > 0
+
 
 class TestMonitors:
     def setUp(self):

test/regularizers_test.py

@@ -3,7 +3,7 @@
 import util
 
 
-class Mixin:
+class Mixin(object):
     def assert_progress(self, **kwargs):
         start = best = None
         for _, val in self.exp.itertrain(
@@ -21,6 +21,22 @@ def assert_progress(self, **kwargs):
         assert best < start   # should have made progress!
 
 
+class TestBuild(util.Base):
+    def setUp(self):
+        self.exp = theanets.Regressor(
+            [self.NUM_INPUTS, 20, self.NUM_OUTPUTS], rng=131)
+
+    def test_regularizers_dict(self):
+        regs = theanets.regularizers.from_kwargs(
+            self.exp, regularizers=dict(input_noise=0.01))
+        assert len(regs) == 1
+
+    def test_regularizers_list(self):
+        reg = theanets.regularizers.Regularizer.build('weight_l2', 0.01)
+        regs = theanets.regularizers.from_kwargs(self.exp, regularizers=[reg])
+        assert len(regs) == 1
+
+
 class TestNetwork(Mixin, util.Base):
     def setUp(self):
         self.exp = theanets.Regressor(