rm shape_infer of Transpose

x2paddle/decoder/onnx_decoder.py

@@ -111,7 +111,7 @@ class ONNXGraphDataNode(GraphNode):
         if isinstance(self.layer, ValueInfoProto):
             values = self.layer.type.tensor_type.shape.dim
             out_shapes = list()
-            out_shapes.append([dim.dim_value for dim in values])
+            out_shapes.append([-1 if dim.dim_value == 0 else dim.dim_value for dim in values])
             return out_shapes
         else:
             values = self.layer.dims
@@ -330,7 +330,7 @@ class ONNXGraph(Graph):
                 'dtype':
                 TENSOR_TYPE_TO_NP_TYPE[item.type.tensor_type.elem_type],
                 'shape':
-                [dim.dim_value for dim in item.type.tensor_type.shape.dim],
+                [-1 if dim.dim_value == 0 else dim.dim_value for dim in item.type.tensor_type.shape.dim],
                 'external': False
             }
 

x2paddle/decoder/onnx_shape_inference.py

@@ -151,7 +151,6 @@ class SymbolicShapeInference:
             'TopK': self._infer_TopK,
             'Unsqueeze': self._infer_Unsqueeze,
             'Where': self._infer_symbolic_compute_ops,
-            'Transpose': self._infer_Transpose,
             'ZipMap': self._infer_ZipMap
         }
         self.run_ = True
@@ -731,15 +730,6 @@ class SymbolicShapeInference:
             helper.make_tensor_value_info(node.output[0], output_type,
                                           self._get_shape(node, 0)))
 
-    def _infer_Transpose(self, node):
-        input_shape = self._get_shape(node, 0)
-        perm = get_attribute(node, 'perm')
-        output_shape = np.array(input_shape)[perm].tolist()
-        vi = self.known_vi_[node.output[0]]
-        vi.CopyFrom(
-            helper.make_tensor_value_info(node.output[0], self.known_vi_[
-                node.input[0]].type.tensor_type.elem_type, output_shape))
-
     def _infer_Compress(self, node):
         input_shape = self._get_shape(node, 0)
         # create a new symbolic dimension for Compress output

x2paddle/op_mapper/onnx2paddle/opset9/opset.py

@@ -255,11 +255,6 @@ class OpSet9():
         self.input_shapes.append(node.out_shapes[0])
 
         shape = node.out_shapes[0]
-        for i, dim_shape in enumerate(shape):
-            if dim_shape == 0 and i == 0:
-                shape[i] = 1
-            if dim_shape == 0 and i != 0:
-                assert 'shape of input is not assigned'
         attr = {
             "dtype": string(node.dtype),
             "shape": shape,