The Delphi config file

Runtime settable parameters for Model training

File location

delphi/data/config.json

Fields

{
  "config_version": "1.0.0",
  "training_stopping_min_delta": 0.1,
  "training_stopping_n_samples": 200,
  "training_burn": 1000,
}

config_version The version of this config file.

training_stopping_min_delta The difference in log_likelihood values used to determine whether the sample burning phase of Model training should stop early

training_stopping_n_samples The number of samples to test for early stopping of the sample burn phase of Model trainng. As each log_likelihood value is generated, the early stopping algorithm finds the min and max log_likelihood for this number of previous log_likelihoods including the current sample. If the magnitude of the difference between min and max is less than or equal to the training_stopping_min_delta then the log_likelihood is said to have converged, and sample burning is stopped.

training_burn The maximum number of samples to burn out during Model Training. This value is used if the ‘DELPHI_N_SAMPLES’ environment variable is not set.