added optional warn_type parameter to tensor.grad

theano/gradient.py

@@ -108,7 +108,7 @@ def grad_sources_inputs(sources, graph_inputs, warn_type=True):
                 if g_r and (getattr(r,'type',0) != getattr(g_r,'type', 1)):
                     r_type = getattr(r,'type', None)
                     g_r_type = getattr(g_r,'type', None)
-                    info('%s.grad returned a different type for input %i: %s vs.  %s'%(node.op, ii, r_type, g_r_type))
+                    warning('%s.grad returned a different type for input %i: %s vs.  %s'%(node.op, ii, r_type, g_r_type))
             if g_r and len(sources) == 1 and sources[0][0].name and r.name:
                 g_r.name = "(d%s/d%s)" % (sources[0][0].name, r.name)
             if g_r is not None: 

theano/tensor/basic.py

@@ -2570,7 +2570,7 @@ outer = Outer()
 # Gradient
 #########################
 
-def grad(cost, wrt, g_cost=None, consider_constant=[]):
+def grad(cost, wrt, g_cost=None, consider_constant=[], warn_type=False):
     """
     :type cost: `Variable`
     :type wrt: `Variable` or list of `Variable`s.
@@ -2579,6 +2579,9 @@ def grad(cost, wrt, g_cost=None, consider_constant=[]):
         ``ones_like(cost)``.
     :param consider_constant: a list of expressions not to backpropagate through
 
+    :param warn_type: a value of True will cause warnings to be logged for any Op that emits a
+        gradient that does not match its input type.
+
     :rtype: `Variable` or list of `Variable`s (depending upon `wrt`)
 
     :return: symbolic expression of gradient of `cost` with respect to `wrt`.
@@ -2597,7 +2600,8 @@ def grad(cost, wrt, g_cost=None, consider_constant=[]):
     if g_cost is None:
         g_cost = ones_like(cost)
     inputs = gof.graph.inputs([cost])
-    gmap = gradient.grad_sources_inputs([(cost, g_cost)], inputs + consider_constant)
+    gmap = gradient.grad_sources_inputs([(cost, g_cost)], inputs + consider_constant,
+            warn_type=warn_type)
 
     def zero(p):
         return TensorConstant(