Fix grad computation and add basic docstring to ctc wrapper

theano/tensor/nnet/ctc.py

@@ -13,6 +13,16 @@ ctc_enabled = config.ctc.enabled
 
 
 class ConnectionistTemporalClassification(gof.COp, gof.OpenMPOp):
+    """
+    CTC loss function wrapper.
+
+    Notes
+    -----
+    Using the wrapper requires that Baidu's warp-ctc library is installed and the
+    configuration variables `config.ctc.enabled` and `config.ctc.root` be properly
+    set.
+
+    """
     __props__ = ('compute_grad',)
 
     func_file = "./ctc_wrapper.c"
@@ -61,6 +71,23 @@ class ConnectionistTemporalClassification(gof.COp, gof.OpenMPOp):
         return ["ctc.h"] + gof.OpenMPOp.c_headers(self)
 
     def make_node(self, activations, labels, input_lengths):
+        """
+        Parameters
+        ----------
+        activations
+            Three-dimensional tensor, which has a shape of (t, m, p), where
+            t is the time index, m is the minibatch index, and p is the index
+            over the probabilities of each symbol in the alphabet. The memory
+            layout is assumed to be in C-order, which consists in the slowest
+            to the fastest changing dimension, from left to right. In this case,
+            p is the fastest changing dimension.
+        labels
+            A 1-D tensor of all the labels for the minibatch.
+        input_lengths
+            A 1-D tensor with the number of time steps for each sequence in
+            the minibatch.
+
+        """
         if not ctc_enabled:
             raise RuntimeError('Baidu CTC is not enabled and '
                                'ConnectionistTemporalClassification Op '
@@ -92,7 +119,9 @@ class ConnectionistTemporalClassification(gof.COp, gof.OpenMPOp):
             raise RuntimeError('Baidu CTC is not enabled and '
                                'ConnectionistTemporalClassification Op '
                                'can not be constructed.')
-        return [self.gradients,
+        grad_op = output_grads[0]
+        total_grad = T.basic.batched_dot(grad_op, self.gradients.dimshuffle(1, 0, 2)).dimshuffle(1, 0, 2)
+        return [total_grad,
                 grad_undefined(self, 1, inputs[1]),
                 grad_undefined(self, 2, inputs[2])]