Merge pull request #320 from csteegz/patch-1

Fall back to non optimized model proto

Author: pengwa

tf2onnx/convert.py

@@ -102,6 +102,8 @@ def main():
                              output_names=args.outputs,
                              inputs_as_nchw=args.inputs_as_nchw)
 
+    model_proto = g.make_model("converted from {}".format(args.input))
+
     new_model_proto = GraphUtil.opt_transposes_with_graph(g, "converted from {}".format(args.input),
                                                           optimize=not args.continue_on_error)
     if new_model_proto: