GitHub

Deborah.DeborahCore.DeborahRunner

Deborah.DeborahCore.DeborahRunner.run_Deborah — Function
run_Deborah(
    toml_path::String, 
    jobid::Union{Nothing, String}=nothing
) -> Nothing

Top-level function to execute the Deborah.DeborahCore pipeline:

  1. Parse configuration with Deborah.DeborahCore.TOMLConfigDeborah.parse_full_config_Deborah
  2. Build path with Deborah.DeborahCore.PathConfigBuilderDeborah.build_path_config_Deborah and partition data with Deborah.DeborahCore.DatasetPartitionerDeborah.partition_dataset
  3. Prepare ML inputs with Deborah.DeborahCore.MLInputPreparer.prepare_ML_inputs
  4. Suggest optimal block size with Deborah.Sarah.BlockSizeSuggester.suggest_opt_block_sizes
  5. Run model-specific sequence with Deborah.DeborahCore.MLSequence.ml_sequence / Deborah.DeborahCore.BaselineSequence.baseline_sequence / Deborah.DeborahCore.BaselineSequence.random_sequence
  6. Generate trace data with Deborah.Sarah.BootstrapDataInit.build_trace_data
  7. Initialize and run bootstrap with Deborah.Sarah.BootstrapDataInit.init_bootstrap_data, Deborah.Sarah.SeedManager.setup_rng_pool and Deborah.Sarah.BootstrapRunner.run_bootstrap!
  8. Collect and save summary with Deborah.Sarah.SummaryCollector.collect_summaries_Deborah and Deborah.DeborahCore.SummaryWriterDeborah.write_summary_file_Deborah
  9. Optionally print result summary with Deborah.DeborahCore.ResultPrinterDeborah.print_summary_results_Deborah if jobid === nothing.

All steps are logged and timed.

source