modify gradweights and gradinput functions to support num_groups

theano/tensor/nnet/abstract_conv.py

@@ -648,7 +648,8 @@ def conv2d_grad_wrt_inputs(output_grad,
                            border_mode='valid',
                            subsample=(1, 1),
                            filter_flip=True,
-                           filter_dilation=(1, 1)):
+                           filter_dilation=(1, 1),
+                           num_groups=1):
     """Compute conv output gradient w.r.t its inputs
 
     This function builds the symbolic graph for getting the
@@ -721,6 +722,9 @@ def conv2d_grad_wrt_inputs(output_grad,
     filter_dilation : tuple of len 2
         The filter dilation used in the forward pass.
         Also known as input striding.
+    num_groups : int
+        Divides the image, kernel and output tensors into num_groups
+        separate groups. Each which carry out convolutions separately
 
     Returns
     -------
@@ -771,7 +775,8 @@ def conv2d_grad_wrt_inputs(output_grad,
                                               border_mode=border_mode,
                                               subsample=subsample,
                                               filter_flip=filter_flip,
-                                              filter_dilation=filter_dilation)
+                                              filter_dilation=filter_dilation,
+                                              num_groups=num_groups)
 
     return grad_input_op(filters, output_grad, input_shape[-2:])
 
@@ -918,7 +923,8 @@ def conv2d_grad_wrt_weights(input,
                             border_mode='valid',
                             subsample=(1, 1),
                             filter_flip=True,
-                            filter_dilation=(1, 1)):
+                            filter_dilation=(1, 1),
+                            num_groups=1):
     """Compute conv output gradient w.r.t its weights
 
     This function will build the symbolic graph for getting the
@@ -983,6 +989,9 @@ def conv2d_grad_wrt_weights(input,
     filter_dilation : tuple of len 2
         The filter dilation used in the forward pass.
         Also known as input striding.
+    num_groups : int
+        Divides the image, kernel and output tensors into num_groups
+        separate groups. Each which carry out convolutions separately
 
     Returns
     -------
@@ -1033,7 +1042,8 @@ def conv2d_grad_wrt_weights(input,
                                                border_mode=border_mode,
                                                subsample=subsample,
                                                filter_flip=filter_flip,
-                                               filter_dilation=filter_dilation)
+                                               filter_dilation=filter_dilation,
+                                               num_groups=num_groups)
 
     return gradWeight_op(input, output_grad, filter_shape[-2:])
 
@@ -1973,7 +1983,8 @@ class AbstractConv2d_gradWeights(AbstractConv_gradWeights):
                                              self.border_mode,
                                              self.subsample,
                                              self.filter_flip,
-                                             self.filter_dilation)(weights,
+                                             self.filter_dilation,
+                                             self.num_groups)(weights,
                                                               top,
                                                               bottom.shape[-2:])
         d_top = AbstractConv2d(self.imshp,
@@ -1981,7 +1992,8 @@ class AbstractConv2d_gradWeights(AbstractConv_gradWeights):
                                self.border_mode,
                                self.subsample,
                                self.filter_flip,
-                               self.filter_dilation)(bottom, weights)
+                               self.filter_dilation,
+                               self.num_groups)(bottom, weights)
         # Make sure that the broadcastable pattern of the inputs is used
         # for the gradients, even if the grad opts are not able to infer
         # that the dimensions are broadcastable.
@@ -2230,14 +2242,16 @@ class AbstractConv2d_gradInputs(AbstractConv_gradInputs):
                                                self.border_mode,
                                                self.subsample,
                                                self.filter_flip,
-                                               self.filter_dilation)(
+                                               self.filter_dilation,
+                                               self.num_groups)(
                                                    bottom, top,
                                                    weights.shape[-2:])
         d_top = AbstractConv2d(self.imshp, self.kshp,
                                self.border_mode,
                                self.subsample,
                                self.filter_flip,
-                               self.filter_dilation)(bottom, weights)
+                               self.filter_dilation,
+                               self.num_groups)(bottom, weights)
         # Make sure that the broadcastable pattern of the inputs is used
         # for the gradients, even if the grad opts are not able to infer
         # that the dimensions are broadcastable.