tigercontrol.experiments.load_prob_model_to_result

tigercontrol.experiments.load_prob_model_to_result(problem_ids=['ARMA-v0', 'Crypto-v0', 'SP500-v0'], model_ids=['LastValue', 'AutoRegressor', 'RNN', 'LSTM'], problem_to_models=None, metrics='mse')[source]

Description: Initializes the experiment instance.

Parameters:
  • problem_ids (list) – ids of problems to evaluate on
  • model_ids (list) – ids of models to use
  • problem_to_models (dict) – map of the form problem_id -> list of model_id. If None, then we assume that the user wants to test every model in model_to_params against every problem in problem_to_params
  • metrics – metrics to load