diff --git a/otx/algorithms/classification/adapters/mmcls/task.py b/otx/algorithms/classification/adapters/mmcls/task.py
index ea6ae730afa..a31be2193bf 100644
--- a/otx/algorithms/classification/adapters/mmcls/task.py
+++ b/otx/algorithms/classification/adapters/mmcls/task.py
@@ -408,9 +408,8 @@ def _train_model(
             )
 
         if self._hyperparams.learning_parameters.auto_decrease_batch_size:
-            validate = isinstance(self, NNCFBaseTask)  # nncf needs eval hooks
             train_func = partial(train_model, meta=deepcopy(meta), model=deepcopy(model), distributed=False)
-            adapt_batch_size(train_func, cfg, datasets, validate)
+            adapt_batch_size(train_func, cfg, datasets, isinstance(self, NNCFBaseTask))  # nncf needs eval hooks
 
         train_model(
             model,
diff --git a/otx/algorithms/detection/adapters/mmdet/task.py b/otx/algorithms/detection/adapters/mmdet/task.py
index 1f6a5b52946..c7bf809e422 100644
--- a/otx/algorithms/detection/adapters/mmdet/task.py
+++ b/otx/algorithms/detection/adapters/mmdet/task.py
@@ -279,9 +279,8 @@ def _train_model(
         validate = bool(cfg.data.get("val", None))
 
         if self._hyperparams.learning_parameters.auto_decrease_batch_size:
-            validate = isinstance(self, NNCFBaseTask)  # nncf needs eval hooks
             train_func = partial(train_detector, meta=deepcopy(meta), model=deepcopy(model), distributed=False)
-            adapt_batch_size(train_func, cfg, datasets, validate)
+            adapt_batch_size(train_func, cfg, datasets, isinstance(self, NNCFBaseTask))  # nncf needs eval hooks
 
         train_detector(
             model,
diff --git a/otx/algorithms/segmentation/adapters/mmseg/task.py b/otx/algorithms/segmentation/adapters/mmseg/task.py
index a5f2cf6293b..8c9f2aca3e7 100644
--- a/otx/algorithms/segmentation/adapters/mmseg/task.py
+++ b/otx/algorithms/segmentation/adapters/mmseg/task.py
@@ -372,9 +372,8 @@ def _train_model(
         validate = bool(cfg.data.get("val", None))
 
         if self._hyperparams.learning_parameters.auto_decrease_batch_size:
-            validate = isinstance(self, NNCFBaseTask)  # nncf needs eval hooks
             train_func = partial(train_segmentor, meta=deepcopy(meta), model=deepcopy(model), distributed=False)
-            adapt_batch_size(train_func, cfg, datasets, validate)
+            adapt_batch_size(train_func, cfg, datasets, isinstance(self, NNCFBaseTask))  # nncf needs eval hooks
 
         train_segmentor(
             model,