'label_inappropriate', 'label_offtopic', 'label_personalstories', 'label_possiblyfeedback', 'label_sentimentnegative', 'label_sentimentpositive', ] IS_DEVELOPMENT = False mlflow_logger = m.MLFlowLogger(uri=TRACKING_URI, experiment=EXPERIMENT_NAME, is_dev=IS_DEVELOPMENT, params=mlflow_params, tags=mlflow_tags) training = m.Modeling(data, gs, mlflow_logger) for method, strat in trans_os.items(): for strategy in strat: print(method, strategy) for label in TARGET_LABELS: logger.info(f"-" * 20) logger.info(f"Target: {label}") data.set_label(label=label) data.set_balance_method(balance_method=method, sampling_strategy=strategy) training.train() training.evaluate(["train", "val"]) #if True: with mlflow.start_run() as run: mlflow_logger.log()
'label_discriminating', 'label_inappropriate', 'label_sentimentnegative', 'label_needsmoderation', ] mlflow_params = dict() mlflow_params["model"] = "DummyClassifier" mlflow_tags = { "cycle4": True, } IS_DEVELOPMENT = False data = m.Posts() mlflow_logger = m.MLFlowLogger(uri=TRACKING_URI, experiment=EXPERIMENT_NAME, is_dev=IS_DEVELOPMENT, params=mlflow_params, tags=mlflow_tags) training = m.Modeling(data, pipeline, mlflow_logger) for label in TARGET_LABELS: logger.info(f"-" * 20) logger.info(f"Target: {label}") data.set_label(label=label) data.set_balance_method(balance_method=None, sampling_strategy=None) training.train() training.evaluate(["train", "val"]) #if True: with mlflow.start_run(run_name='dummy_classifier') as run: mlflow_logger.log()