Data-3460: Adding GetInferenceResponse #608
Conversation
etai-shuchatowitz
added
the
ready-for-protos
etai-shuchatowitz
added
ready-for-protos
etai-shuchatowitz
force-pushed
the
DATA-3460/flat-tensor
branch
from
6769bfb to
1879125
Compare
| viam.service.mlmodel.v1.FlatTensors output_tensors = 1; | ||
| viam.app.data.v1.Annotations annotations = 2; | ||
| // anything else you want to say | ||
| google.protobuf.Struct extra = 99; |
There was a problem hiding this comment.
I think there's pretty low risk of needing to deprecate this field, but I would hate to introduce a backwards-incompatible breaking change over something as minor as an extra field. Is this intended to be used for something immediately? If not, I would err on just adding it as needed. WDYT?
There was a problem hiding this comment.
OK! That makes sense to me. Just thought it might be helpful because I find myself wanting to pass more info around in the MLModel in delphi. But we can add it if we need.
There was a problem hiding this comment.
If we wanted to pass more info around in the MLModel in Delphi, wouldn't we then be working with the ML model proto? https://github.com/viamrobotics/api/blob/main/proto/viam/service/mlmodel/v1/mlmodel.proto
There was a problem hiding this comment.
Yeah, fixed it up! Thanks for this!
| @@ -20,4 +22,9 @@ message GetInferenceRequest { | |||
| string organization_id = 4; | |||
| } | |||
|
|
|||
| message GetInferenceResponse {} | |||
| message GetInferenceResponse { | |||
| viam.service.mlmodel.v1.FlatTensors output_tensors = 1; | |||
There was a problem hiding this comment.
Wooh, awesome. Thanks for maintaining parity with the ML model proto defs :)
etai-shuchatowitz
force-pushed
the
DATA-3460/flat-tensor
branch
from
8910d43 to
4fc5c05
Compare
etai-shuchatowitz
removed
the
ready-for-protos
Context: https://viam.atlassian.net/browse/DATA-3460?atlOrigin=eyJpIjoiMmVjNmJkOWU3MTlkNDA3ODgwODdiMWE4OTYxYWYwNjkiLCJwIjoiaiJ9
Added FlatTensors, Annotations and an Extra struct (so that we can add any potential metadata we want) to the GetInferenceResponse.