diff --git a/otx/algorithms/classification/adapters/mmcls/configurer.py b/otx/algorithms/classification/adapters/mmcls/configurer.py
index df1984ac53e..ecdd706ca74 100644
--- a/otx/algorithms/classification/adapters/mmcls/configurer.py
+++ b/otx/algorithms/classification/adapters/mmcls/configurer.py
@@ -387,12 +387,7 @@ def configure_device(self, cfg, training):
                 cfg.distributed = True
                 self.configure_distributed(cfg)
         elif "gpu_ids" not in cfg:
-            gpu_ids = os.environ.get("CUDA_VISIBLE_DEVICES")
-            logger.info(f"CUDA_VISIBLE_DEVICES = {gpu_ids}")
-            if gpu_ids is not None:
-                cfg.gpu_ids = range(len(gpu_ids.split(",")))
-            else:
-                cfg.gpu_ids = range(1)
+            cfg.gpu_ids = range(1)
 
         # consider "cuda" and "cpu" device only
         if not torch.cuda.is_available():
diff --git a/otx/algorithms/detection/adapters/mmdet/configurer.py b/otx/algorithms/detection/adapters/mmdet/configurer.py
index 54831bfac5c..47fb07ed418 100644
--- a/otx/algorithms/detection/adapters/mmdet/configurer.py
+++ b/otx/algorithms/detection/adapters/mmdet/configurer.py
@@ -549,12 +549,7 @@ def configure_device(self, cfg, training):
                 cfg.distributed = True
                 self.configure_distributed(cfg)
         elif "gpu_ids" not in cfg:
-            gpu_ids = os.environ.get("CUDA_VISIBLE_DEVICES")
-            logger.info(f"CUDA_VISIBLE_DEVICES = {gpu_ids}")
-            if gpu_ids is not None:
-                cfg.gpu_ids = range(len(gpu_ids.split(",")))
-            else:
-                cfg.gpu_ids = range(1)
+            cfg.gpu_ids = range(1)
 
         # consider "cuda" and "cpu" device only
         if not torch.cuda.is_available():
diff --git a/otx/algorithms/segmentation/adapters/mmseg/configurer.py b/otx/algorithms/segmentation/adapters/mmseg/configurer.py
index 0fc24cd5f49..2444c23c519 100644
--- a/otx/algorithms/segmentation/adapters/mmseg/configurer.py
+++ b/otx/algorithms/segmentation/adapters/mmseg/configurer.py
@@ -381,12 +381,7 @@ def configure_device(self, cfg: Config, training: bool) -> None:
                 cfg.distributed = True
                 self.configure_distributed(cfg)
         elif "gpu_ids" not in cfg:
-            gpu_ids = os.environ.get("CUDA_VISIBLE_DEVICES")
-            logger.info(f"CUDA_VISIBLE_DEVICES = {gpu_ids}")
-            if gpu_ids is not None:
-                cfg.gpu_ids = range(len(gpu_ids.split(",")))
-            else:
-                cfg.gpu_ids = range(1)
+            cfg.gpu_ids = range(1)
 
         # consider "cuda" and "cpu" device only
         if not torch.cuda.is_available():