Skip to content

A programming framework for agentic AI

no search when max_iter < 2

Chi Wangb03a87e73753
diff --git a/flaml/automl.py b/flaml/automl.py
index a0315a625..08cd66519 100644
--- a/flaml/automl.py
+++ b/flaml/automl.py
@@ -104,6 +104,7 @@ class SearchState:
self.trained_estimator = None
self.sample_size = None
self.trial_time = 0
+ self.best_n_iter = None
def update(self, result, time_used, save_model_history=False):
if result:
@@ -430,7 +431,7 @@ class AutoML:
@property
def time_to_find_best_model(self) -> float:
- """time taken to find best model in seconds"""
+ """Time taken to find best model in seconds"""
return self.__dict__.get("_time_taken_best_iter")
def predict(self, X_test):
@@ -1768,6 +1769,17 @@ class AutoML:
better = True # whether we find a better model in one trial
if self._ensemble:
self.best_model = {}
+ if self._max_iter < 2 and self.estimator_list:
+ # when max_iter is 1, no need to search
+ self._max_iter = 0
+ self._best_estimator = estimator = self.estimator_list[0]
+ self._selected = state = self._search_states[estimator]
+ state.best_config_sample_size = self._state.data_size
+ state.best_config = (
+ state.init_config
+ if isinstance(state.init_config, dict)
+ else state.init_config[0]
+ )
for self._track_iter in range(self._max_iter):
if self._estimator_index is None:
estimator = self._active_estimators[0]
@@ -1844,9 +1856,9 @@ class AutoML:
metric="val_loss",
mode="min",
space=search_space,
- points_to_evaluate=points_to_evaluate
- if len(search_state.init_config) == len(search_space)
- else None,
+ points_to_evaluate=[
+ p for p in points_to_evaluate if len(p) == len(search_space)
+ ],
)
search_state.search_alg = ConcurrencyLimiter(algo, max_concurrent=1)
# search_state.search_alg = algo
diff --git a/flaml/model.py b/flaml/model.py
index 385bef499..740515507 100644
--- a/flaml/model.py
+++ b/flaml/model.py
@@ -465,7 +465,7 @@ class XGBoostEstimator(SKLearnEstimator):
def predict(self, X_test):
import xgboost as xgb
-
+
if not issparse(X_test):
X_test = self._preprocess(X_test)
dtest = xgb.DMatrix(X_test)
diff --git a/test/test_automl.py b/test/test_automl.py
index 97297c8b3..7ccbda76b 100644
--- a/test/test_automl.py
+++ b/test/test_automl.py
@@ -31,7 +31,7 @@ class MyRegularizedGreedyForest(SKLearnEstimator):
self.estimator_class = RGFClassifier
else:
from rgf.sklearn import RGFRegressor
-
+
self.estimator_class = RGFRegressor
@classmethod
diff --git a/test/test_training_log.py b/test/test_training_log.py
index 73de45d25..20c728c01 100644
--- a/test/test_training_log.py
+++ b/test/test_training_log.py
@@ -25,23 +25,49 @@ class TestTrainingLog(unittest.TestCase):
"mem_thres": 1024 * 1024,
"n_jobs": 1,
"model_history": True,
- "train_time_limit": 0.01,
+ "train_time_limit": 0.1,
"verbose": 3,
"ensemble": True,
"keep_search_state": True,
}
X_train, y_train = fetch_california_housing(return_X_y=True)
automl.fit(X_train=X_train, y_train=y_train, **automl_settings)
- automl._state._train_with_config(automl.best_estimator, automl.best_config)
-
# Check if the training log file is populated.
self.assertTrue(os.path.exists(filename))
- with training_log_reader(filename) as reader:
- count = 0
- for record in reader.records():
- print(record)
- count += 1
- self.assertGreater(count, 0)
+ if automl.best_estimator:
+ estimator, config = automl.best_estimator, automl.best_config
+ model0 = automl.best_model_for_estimator(estimator)
+ print(model0.estimator)
+
+ automl.time_budget = None
+ model, _ = automl._state._train_with_config(estimator, config)
+ # model0 and model are equivalent unless model0's n_estimator is out of search space range
+ assert (
+ str(model0.estimator) == str(model.estimator)
+ or model0["n_estimators"] < 4
+ )
+
+ # assuming estimator & config are saved and loaded as follows
+ automl = AutoML()
+ automl.fit(
+ X_train=X_train,
+ y_train=y_train,
+ max_iter=0,
+ task="regression",
+ estimator_list=[estimator],
+ n_jobs=1,
+ starting_points={estimator: config},
+ )
+ # then the fitted model should be equivalent to model
+ # print(str(model.estimator), str(automl.model.estimator))
+ assert str(model.estimator) == str(automl.model.estimator)
+
+ with training_log_reader(filename) as reader:
+ count = 0
+ for record in reader.records():
+ print(record)
+ count += 1
+ self.assertGreater(count, 0)
automl_settings["log_file_name"] = None
automl.fit(X_train=X_train, y_train=y_train, **automl_settings)