[GitHub] [incubator-tvm] zhiics commented on a change in pull request #5770: [RUNTIME] Introduce MetadataModule to separate code compilation/interpretation and weight initialization

[GitBox <gi...@apache.org>]

zhiics commented on a change in pull request #5770:
URL: https://github.com/apache/incubator-tvm/pull/5770#discussion_r442490987



##########
File path: python/tvm/contrib/graph_runtime.py
##########
@@ -162,7 +162,12 @@ def set_input(self, key=None, value=None, **params):
             keys = list(params.keys())
             keys.sort(key=lambda x: -np.prod(params[x].shape))
             for k in keys:
-                self._get_input(k).copyfrom(params[k])
+                # TODO(zhiics) Skip the weights for submodule in a better way.
+                # We should use MetadataModule for initialization and remove

Review comment:
       Yes, there is a warning in the C++ side.
   
   https://github.com/apache/incubator-tvm/blob/d8c80c382f02052b07da1235190a5b6c7acea994/src/runtime/graph/graph_runtime.cc#L93




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org