Merge pull request #689 from lijm1358/optuna-fix

Fix the issue of Optuna tuning mode with LightGBM
mljar · Jan 8, 2024 · 17af68b · 17af68b
2 parents 8678d84 + 80bc095
commit 17af68b
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/supervised/tuner/optuna/lightgbm.py b/supervised/tuner/optuna/lightgbm.py
@@ -139,16 +139,17 @@ def __call__(self, trial):
             pruning_callback = optuna.integration.LightGBMPruningCallback(
                 trial, metric_name, "validation"
             )
+            early_stopping_callback = lgb.early_stopping(
+                self.early_stopping_rounds, verbose=False
+            )
 
             gbm = lgb.train(
                 param,
                 self.dtrain,
                 valid_sets=[self.dvalid],
                 valid_names=["validation"],
-                verbose_eval=False,
-                callbacks=[pruning_callback],
+                callbacks=[pruning_callback, early_stopping_callback],
                 num_boost_round=self.rounds,
-                early_stopping_rounds=self.early_stopping_rounds,
                 feval=self.custom_eval_metric,
             )