Metadata for input/output formats in model file proto. (#15252)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/15252

We would like to extend the model file format to include strongly type, semantic information
about the model inputs and outputs.

The goal is for a user to be able to consider a model file like a function with
a well defined API describing what the inputs and outputs would be.

Reviewed By: dzhulgakov

Differential Revision: D13009915

fbshipit-source-id: 5df124a876ad03c05fbdaacae0eab659637734c1

diff --git a/caffe2/proto/torch.proto b/caffe2/proto/torch.proto
index 0dbca7c..4955e94 100644 (file)
--- a/caffe2/proto/torch.proto
+++ b/caffe2/proto/torch.proto
@@ -84,5 +84,7 @@ message ModelDef {
   // the tensor id is defined as TensorProto.name
   repeated TensorDef tensors = 5;
 
-  // future: add a way to provide additional meta-data
+  // Serialized metadata for input and outputs (e.g: expected types, shapes and
+  // semantic information).
+  optional RecordRef model_metadata = 6;
 }