Create trained models API

edit

Creates a new trained model for inference. The API accepts a PutTrainedModelRequest object as a request and returns a PutTrainedModelResponse.

Request

edit

A PutTrainedModelRequest requires the following argument:

PutTrainedModelRequest request = new PutTrainedModelRequest(trainedModelConfig); 

The configuration of the inference trained model to create

Trained model configuration

edit

The TrainedModelConfig object contains all the details about the trained model configuration and contains the following arguments:

TrainedModelConfig trainedModelConfig = TrainedModelConfig.builder()
    .setDefinition(definition) 
    .setCompressedDefinition(InferenceToXContentCompressor.deflate(definition)) 
    .setModelId("my-new-trained-model") 
    .setInput(new TrainedModelInput("col1", "col2", "col3", "col4")) 
    .setDescription("test model") 
    .setMetadata(new HashMap<>()) 
    .setTags("my_regression_models") 
    .setInferenceConfig(new RegressionConfig("value", 0)) 
    .build();

The inference definition for the model

Optionally, if the inference definition is large, you may choose to compress it for transport. Do not supply both the compressed and uncompressed definitions.

The unique model id

The input field names for the model definition

Optionally, a human-readable description

Optionally, an object map contain metadata about the model

Optionally, an array of tags to organize the model

The default inference config to use with the model. Must match the underlying definition target_type.

Synchronous execution

edit

When executing a PutTrainedModelRequest in the following manner, the client waits for the PutTrainedModelResponse to be returned before continuing with code execution:

PutTrainedModelResponse response = client.machineLearning().putTrainedModel(request, RequestOptions.DEFAULT);

Synchronous calls may throw an IOException in case of either failing to parse the REST response in the high-level REST client, the request times out or similar cases where there is no response coming back from the server.

In cases where the server returns a 4xx or 5xx error code, the high-level client tries to parse the response body error details instead and then throws a generic ElasticsearchException and adds the original ResponseException as a suppressed exception to it.

Asynchronous execution

edit

Executing a PutTrainedModelRequest can also be done in an asynchronous fashion so that the client can return directly. Users need to specify how the response or potential failures will be handled by passing the request and a listener to the asynchronous put-trained-model method:

client.machineLearning().putTrainedModelAsync(request, RequestOptions.DEFAULT, listener); 

The PutTrainedModelRequest to execute and the ActionListener to use when the execution completes

The asynchronous method does not block and returns immediately. Once it is completed the ActionListener is called back using the onResponse method if the execution successfully completed or using the onFailure method if it failed. Failure scenarios and expected exceptions are the same as in the synchronous execution case.

A typical listener for put-trained-model looks like:

ActionListener<PutTrainedModelResponse> listener = new ActionListener<PutTrainedModelResponse>() {
    @Override
    public void onResponse(PutTrainedModelResponse response) {
        
    }

    @Override
    public void onFailure(Exception e) {
        
    }
};

Called when the execution is successfully completed.

Called when the whole PutTrainedModelRequest fails.

Response

edit

The returned PutTrainedModelResponse contains the newly created trained model. The PutTrainedModelResponse will omit the model definition as a precaution against streaming large model definitions back to the client.

TrainedModelConfig model = response.getResponse();