address first round of cr comments

contrib/Python/cntk/ops/__init__.py

@@ -408,7 +408,7 @@ def times(left, right, output_rank=1, name=None):
     # CNTK uses column vectors and column major representation, thus we reverse
     # params    
     op = Times(right, left, outputRank=output_rank, name=name)
-    #wrap_numpy_arrays(op)    
+    wrap_numpy_arrays(op)    
     op.rank = op.x.rank + op.y.rank - 2    
     return op
 
@@ -586,7 +586,7 @@ def relu(x, name=None):
     from cntk.ops.cntk2 import Relu
     op = Relu(x, name=name)
     wrap_numpy_arrays(op)    
-    op.rank = 0
+    op.rank = op._.rank
     return op
 
 def sigmoid(x, name=None):
@@ -609,7 +609,7 @@ def sigmoid(x, name=None):
     from cntk.ops.cntk2 import Sigmoid
     op = Sigmoid(x, name=name)
     wrap_numpy_arrays(op)        
-    op.rank = 0
+    op.rank = op._.rank
     return op
 
 def tanh(x, name=None):
@@ -631,7 +631,7 @@ def tanh(x, name=None):
     from cntk.ops.cntk2 import Tanh
     op = Tanh(x, name=name)
     wrap_numpy_arrays(op)        
-    op.rank = 0
+    op.rank = op._.rank
     return op
 
 def softmax(x, name=None):
@@ -658,7 +658,7 @@ def softmax(x, name=None):
     from cntk.ops.cntk2 import Softmax
     op = Softmax(x)
     wrap_numpy_arrays(op)        
-    op.rank = 0
+    op.rank = op._.rank
     return op
 
 def exp(x, name=None):
@@ -699,7 +699,7 @@ def log(x, name=None):
         CNTK returns -85.1 for log(x) if `x` is negative or zero. The reason is that 
         it uses 1e-37 (whose natural logarithm is -85.1) as the smallest float 
         number for `log`, because this is the only guaranteed precision across 
-        platforms. This will be changed to op = `NaN` and `-inf`.
+        platforms. This will be changed to return `NaN` and `-inf`.
     """
     from cntk.ops.cntk2 import Log
     op = Log(x, name=name)
@@ -724,7 +724,7 @@ def sqrt(x, name=None):
         
     Note:
         CNTK returns zero for sqrt of negative nubmers, this will be changed to 
-        op = NaN
+        retrun NaN
     """
     from cntk.ops.cntk2 import Sqrt
     op = Sqrt(x, name=name)
@@ -774,8 +774,8 @@ def abs(x, name=None):
 
 def cond(flag, value_if_true, value_if_false, name=None):
     """
-    op = either value_if_true or value_if_false based on the value of flag.
-    If flag != 0 value_if_true is op =ed, otherwise value_if_false.
+    return either value_if_true or value_if_false based on the value of flag.
+    If flag != 0 value_if_true is returned, otherwise value_if_false.
     Behaves analogously to numpy.where(...).
 
     Example:
@@ -794,7 +794,7 @@ def cond(flag, value_if_true, value_if_false, name=None):
     from cntk.ops.cntk1 import If
     op = If(flag, value_if_true, value_if_false, name = name)
     wrap_numpy_arrays(op)        
-    op.rank = max(op.cond.rank,max(op.thenVal.rank,op.elseVal.rank))
+    op.rank = max(op.cond.rank,op.thenVal.rank,op.elseVal.rank)
     return op
 
 ################################################################################
@@ -808,7 +808,7 @@ def future_value(shape, x, time_step=1, default_hidden_activation=0.1, name=None
     the next logical sample. The `time_step` parameter is the number of steps 
     to look into the future and is 1 by default. If there is no future value (i.e. 
     the current sample is the last one in the tensor) then the `default_hidden_activation` 
-    value is op =ed which is 0.1 by default.
+    value is returned which is 0.1 by default.
     
     Example:
         >>> data = np.array([[1,2,3,4],[5,6,7,8],[9,10,11,12]])
@@ -833,7 +833,7 @@ def future_value(shape, x, time_step=1, default_hidden_activation=0.1, name=None
     from cntk.ops.cntk1 import FutureValue
     op = FutureValue(shape, x, time_step, default_hidden_activation, name = name)
     wrap_numpy_arrays(op)            
-    op.rank = 0 if np.isscalar(shape) else len(shape)
+    op.rank = np.ndim(shape)
     return op
 
 def past_value(shape, x, time_step=1, default_hidden_activation=0.1, name=None):
@@ -843,7 +843,7 @@ def past_value(shape, x, time_step=1, default_hidden_activation=0.1, name=None):
     the previous logical sample. The `time_step` parameter is the number of steps 
     to look into the past and is 1 by default. If there is no past value (i.e. 
     the current sample is the first one in the tensor) then the `default_hidden_activation` 
-    value is op =ed which is 0.1 by default.
+    value is returned which is 0.1 by default.
     
     Example:
         >>> data = np.array([[1,2,3,4],[5,6,7,8],[9,10,11,12]])
@@ -868,7 +868,7 @@ def past_value(shape, x, time_step=1, default_hidden_activation=0.1, name=None):
     from cntk.ops.cntk1 import PastValue
     op = PastValue(shape, x, time_step, default_hidden_activation, name = name)
     wrap_numpy_arrays(op)  
-    op.rank = 0 if np.isscalar(shape) else len(shape)
+    op.rank = np.ndim(shape)
     return op
 
 ################################################################################
@@ -903,15 +903,13 @@ def reshape(x, shape, name=None):
         shape = tuple(reversed(shape))    
     op = NewReshape(x, shape, 0, 0, name = name)
     wrap_numpy_arrays(op)            
-    op.rank = 0 if np.isscalar(shape) else len(shape)
+    op.rank = np.ndim(shape)
     return op
 
 def transpose_dimensions(x, axis1, axis2, name=None):
     """
     Reverses two axes of the tensor. The output tensor has the same data but with
-    axis1 and axis2 swaped.
-    
-    The backward pass propagates the received gradient for the output-shape to the input shape.
+    axis1 and axis2 swapped.    
     
     Note:
         axes are zero-based as in Numpy, in contrast to CNTK, where 1 is the first axis. 
@@ -1080,7 +1078,7 @@ def input(shape, dynamic_axis='', name=None):
         shape = tuple(reversed(shape))
     op = Input(shape, dynamicAxis=dynamic_axis, name=name)
 
-    op.rank = 0 if np.isscalar(shape) else len(shape)
+    op.rank = np.ndim(shape)
     return op
 
 def sparse_input_numpy(indices, values, shape, alias=None, dynamic_axis='', name=None):
@@ -1154,7 +1152,7 @@ def sparse_input(shape, dynamic_axis='', name=None):
         # cntk uses column major, thus we reverse the shape    
         shape = tuple(reversed(shape))
     op = SparseInput(shape, dynamicAxis=dynamic_axis, name=name)
-    op.rank = 0 if np.isscalar(shape) else len(shape)
+    op.rank = np.ndim(shape)
     return op
 
 def parameter(shape=None, value=None, learning_rate_multiplier=1.0,
@@ -1191,7 +1189,7 @@ def parameter(shape=None, value=None, learning_rate_multiplier=1.0,
                     learningRateMultiplier=learning_rate_multiplier,
                     name=name)
 
-        op.rank = 0 if np.isscalar(shape) else len(shape)
+        op.rank = np.ndim(shape)
         return op
 
     """
@@ -1233,7 +1231,7 @@ def parameter(shape=None, value=None, learning_rate_multiplier=1.0,
         init='fromLiteral',
         initFromLiteral=s.getvalue().decode())
 
-    op.rank = 0 if np.isscalar(param_shape) else len(param_shape)
+    op.rank = np.ndim(param_shape)
     return op
 
 def constant(value, name=None):

contrib/Python/cntk/ops/tests/linear_test.py

@@ -234,8 +234,8 @@ def numpy_op(x):
 
 
 TIMES_PAIRS = [
-    #([[30.]], [[10.]]),
-    #([[1.5, 2.1]], [[10.], [20.]]),
+    ([[30.]], [[10.]]),
+    ([[1.5, 2.1]], [[10.], [20.]]),
     ([[100., 200.]], [[10.], [20.]]),
 ]
 
@@ -266,7 +266,7 @@ def test_op_times(left_operand, right_operand, device_id, precision,
     right_as_input = times(constant(left_operand), b)
 
     unittest_helper(left_as_input, None, expected, device_id=device_id,
-                    precision=precision, clean_up=False, backward_pass=False)
+                    precision=precision, clean_up=True, backward_pass=False)
 
     unittest_helper(right_as_input, None, expected, device_id=device_id,
                     precision=precision, clean_up=True, backward_pass=False)

contrib/Python/cntk/utils/_fetch_ops.py

@@ -271,6 +271,7 @@ def __init__(self, cond, thenVal, elseVal, op_name='BS.Boolean.If', name=None):
         self.thenVal = thenVal
         self.elseVal = elseVal
         self.params_with_defaults = []
+        self.inputs = ['cond', 'thenVal', 'elseVal']
 
 """