A programming framework for agentic AI
set converge flag when no trial can be sampled (#217)
Chi Wang16a97bec76ba
diff --git a/flaml/automl.py b/flaml/automl.pyindex 90f794e00..505763f47 100644--- a/flaml/automl.py+++ b/flaml/automl.py@@ -1312,7 +1312,7 @@ class AutoML:return metric_to_minimize, metrics_to_logwhich returns a float number as the minimization objective,- and a tuple of floats or a dictionary as the metrics to log.+ and a dictionary as the metrics to log.task: A string of the task type, e.g.,'classification', 'regression', 'forecast', 'rank'.n_jobs: An integer of the number of threads for training.@@ -1853,6 +1853,16 @@ class AutoML:self._search_states[e].init_eci / eci_base * self._eci[0])self._estimator_index = 0+ min_budget = max(10 * self._eci[0], sum(self._eci))+ max_budget = 10000 * self._eci[0]+ if search_state.sample_size:+ ratio = search_state.data_size / search_state.sample_size+ min_budget *= ratio+ max_budget *= ratio+ logger.info(+ f"Estimated sufficient time budget={max_budget:.0f}s."+ f" Estimated necessary time budget={min_budget:.0f}s."+ )if result["wall_clock_time"] is not None:self._state.time_from_start = result["wall_clock_time"]# logger.info(f"{self._search_states[estimator].sample_size}, {data_size}")@@ -1923,7 +1933,7 @@ class AutoML:mlflow.log_param("best_config", search_state.best_config)mlflow.log_param("best_learner", self._best_estimator)logger.info(- " at {:.1f}s,\tbest {}'s error={:.4f},\tbest {}'s error={:.4f}".format(+ " at {:.1f}s,\testimator {}'s best error={:.4f},\tbest estimator {}'s best error={:.4f}".format(self._state.time_from_start,estimator,search_state.best_loss,@@ -1949,18 +1959,16 @@ class AutoML:f"exceeds {self._warn_threshold} times the time taken ""to find the best model.")- self._warn_threshold *= 10if self._early_stop:logger.warning("Stopping search as early_stop is set to True.")break+ self._warn_threshold *= 10else:logger.info(f"stop trying learner {estimator}")if self._estimator_index is not None:self._active_estimators.remove(estimator)self._estimator_index -= 1- self._state.time_from_start = time.time() - self._start_time_flag- if self._state.time_budget > self._state.time_from_start:- search_state.search_alg.searcher._is_ls_ever_converged = True+ search_state.search_alg.searcher._is_ls_ever_converged = Trueif (self._retrain_in_budgetand best_config_sigdiff --git a/flaml/ml.py b/flaml/ml.pyindex edad16bab..37206ce4f 100644--- a/flaml/ml.py+++ b/flaml/ml.py@@ -232,8 +232,7 @@ def _eval_estimator(groups_test,fit_kwargs.get("groups"),)- if isinstance(metric_for_logging, dict):- pred_time = metric_for_logging.get("pred_time", 0)+ pred_time = metric_for_logging.get("pred_time", 0)test_pred_y = None# eval_metric may return test_pred_y but not necessarily. Setting None for now.return test_loss, metric_for_logging, pred_time, test_pred_y@@ -373,9 +372,7 @@ def evaluate_model_CV(total_fold_num += 1total_val_loss += val_loss_iif log_training_metric or not isinstance(eval_metric, str):- if isinstance(total_metric, list):- total_metric = [total_metric[i] + v for i, v in enumerate(metric_i)]- elif isinstance(total_metric, dict):+ if isinstance(total_metric, dict):total_metric = {k: total_metric[k] + v for k, v in metric_i.items()}elif total_metric is not None:total_metric += metric_i@@ -392,9 +389,7 @@ def evaluate_model_CV(val_loss = np.max(val_loss_list)n = total_fold_numif log_training_metric or not isinstance(eval_metric, str):- if isinstance(total_metric, list):- metric = [v / n for v in total_metric]- elif isinstance(total_metric, dict):+ if isinstance(total_metric, dict):metric = {k: v / n for k, v in total_metric.items()}else:metric = total_metric / n@@ -428,7 +423,7 @@ def compute_estimator():estimator_class = estimator_class or get_estimator_class(task, estimator_name)estimator = estimator_class(**config_dic, task=task, n_jobs=n_jobs)- if "holdout" in eval_method:+ if "holdout" == eval_method:val_loss, metric_for_logging, train_time, pred_time = get_test_loss(config_dic,estimator,diff --git a/flaml/version.py b/flaml/version.pyindex 364e7baed..7bbb2ef5c 100644--- a/flaml/version.py+++ b/flaml/version.py@@ -1 +1 @@-__version__ = "0.6.4"+__version__ = "0.6.5"