Skip to content

Latest commit

 

History

History
48 lines (34 loc) · 2.05 KB

MLRunner_en.md

File metadata and controls

48 lines (34 loc) · 2.05 KB
Raw

MLRunner

MLRunner is the entry point to Angel algorithms. It defines the standard process of starting up an Angel application, and encapsulates AngelClient.

Purpose

  • Starts Angel PS, loads and stores model, and starts the default implementations of processes such as Task
  • In general, tha application can directly use the functions above

Core Methods

  1. train

    • Definition: train(conf: Configuration)
    • Purpose: starts up Angel app's model training process
    • Parameters: conf: configures Angel job and algorithm settings
    • Return value: none

  2. train (default implementation)

    • Definition: train(conf: Configuration, model: MLModel, taskClass: Class[_ <: BaseTask[_, _, _]]): Unit

    • Purpose: starts Angel app's model training process. This method encapsulates specific Angel PS/worker starting-up and model-loading/saving processes; can be referenced by subclasses directly

    • Parameters

      • conf: configures Angel job and algorithm
      • model: MLModel, the machine-learning model/algorithm
      • taskClass: Class[_ <: TrainTask[_, _, _]] Task class in the algorithm's process

    • Return value: none

  3. incTrain

    • Definition: incTrain(conf: Configuration)
    • Purpose: updates an existing model using incremental training
    • Parameters: conf: configures Angel job and algorithm
    • Return value: none

  4. predict

    • Definition: predict(conf: Configuration)
    • Purpose: starts Angel app and computes model prediction
    • Parameters: conf: configures Angel job
    • Return value: none

  5. predict (default implementation)

    • Definition: predict(conf: Configuration, model: MLModel, taskClass: Class[_ <: PredictTask[_, _, _]]): Unit
    • Purpose: starts Angel app and updates an existing model using incremental training. This method encapsulates specific Angel PS/worker starting-up and model-loading/saving processes; can be referenced by subclasses directly
    • Parameters: conf: configures Angel job
    • Return value: none