# DEEPLEARNINGL4J

`DL4JStep` is a pipeline step that configures a DL4J model that is to be executed.

| Configs     | Descriptions                                                                                                                                                                                                                                                                                                            |
| ----------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| modelUri    | Specifies the location of a saved model file.                                                                                                                                                                                                                                                                           |
| inputNames  | A list of names of the input placeholders (mainly for DL4J - computation graph, with multiple inputs. Where values from the input data keys are mapped to the computation graph inputs).                                                                                                                                |
| outputNames | A list of names of the output placeholders (mainly for DL4J - computation graph, with multiple outputs. Where the values of these output keys are mapped from the computation graph output - `INDArray[]` to data keys).                                                                                                |
| loaderClass | Optional, usually unnecessary. Specifies a class used to load the model if customization in how model loading is performed, instead of the usual `MultiLayerNetwork.load` or `ComputationGraph.load` methods. Must be a `java.util.Function<String,MultiLayerNetwork>` or `java.util.Function<String,ComputationGraph>` |