External Prediction API

API

The Seldon Core exposes a generic external API to connect your ML runtime prediction to external business applications.

REST API

Prediction

  • endpoint : POST /api/v0.1/predictions

  • payload : JSON representation of SeldonMessage - see proto definition

  • example payload :

    {"data":{"names":["a","b"],"tensor":{"shape":[2,2],"values":[0,0,1,1]}}}

Feedback

  • endpoint : POST /api/v0.1/feedback
  • payload : JSON representation of Feedback - see proto definition

gRPC

service Seldon {
  rpc Predict(SeldonMessage) returns (SeldonMessage) {};
  rpc SendFeedback(Feedback) returns (SeldonMessage) {};
 }

see full proto definition