Realtime scoring functions perform individual predictions on input data passed through the SQL query using machine learning models, the same models used in BYOMPredict functions. These models are previously trained and then exported into Vantage for scoring. They are the real time equivalent of batch PMMLPredict, ONNXPredict, ONNXEmbeddings, ONNXSeq2Seq functions. The functions accept single dataset in the SQL query as input, together with the model, and returns the response immediately as output. The primary use case for realtime scoring functions is to quickly return a score to the requesting application or human-generated query for subsequent action based on the response generated.