Auton ML core functionalities

  1. Runs as server waiting to hear TA3 client GRPC requests on port 45042

./src/main.py ta2ta3

  1. Runs in stand-alone TA2 mode without any client/TA2-3 API. This will search on TRAIN data for valid pipelines and output them

./src/main.py search.
  1. For a given dataset-problem specification

    • Search for valid pipelines (solutions)

    • Produce top pipelines with ranks. Pipelines are output as JSON files for evaluation purposes

    • Enable scoring, fit, produce on any pipeline. These calls are invoked from TA3

4. TA2 pipelines can be evaluated independently using D3M’s reference runtime framework on the TRAIN-TEST dataset splits. (See scripts/run.sh)

“Pipeline” and “solution” are aliases and mean the same concept in TA2. A pipeline or solution is an end-to-end flowchart (DAG) composed of TA1 primitives, their hyperparameters and their connections to build model on the TRAIN data and produce predictions on the TEST data.