diff --git a/monai/apps/auto3dseg/bundle_gen.py b/monai/apps/auto3dseg/bundle_gen.py
index 462452007f..1831e9f840 100644
--- a/monai/apps/auto3dseg/bundle_gen.py
+++ b/monai/apps/auto3dseg/bundle_gen.py
@@ -35,7 +35,7 @@
 from monai.utils import ensure_tuple
 
 logger = get_logger(module_name=__name__)
-ALGO_HASH = os.environ.get("MONAI_ALGO_HASH", "d8bec42")
+ALGO_HASH = os.environ.get("MONAI_ALGO_HASH", "4af80e1")
 
 __all__ = ["BundleAlgo", "BundleGen"]
 
diff --git a/monai/apps/nnunet/__main__.py b/monai/apps/nnunet/__main__.py
index 9e4ca5d073..14e2dca8d3 100644
--- a/monai/apps/nnunet/__main__.py
+++ b/monai/apps/nnunet/__main__.py
@@ -8,7 +8,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
 """
 Examples:
     - User can use the one-liner to start the nnU-Net workflow
@@ -85,7 +84,6 @@
 
 """
 
-
 from __future__ import annotations
 
 from monai.apps.nnunet.nnunetv2_runner import nnUNetV2Runner
diff --git a/monai/apps/nnunet/utils.py b/monai/apps/nnunet/utils.py
index 773b9891c0..f49a9b76e9 100644
--- a/monai/apps/nnunet/utils.py
+++ b/monai/apps/nnunet/utils.py
@@ -25,7 +25,6 @@
 
 logger = monai.apps.utils.get_logger(__name__)
 
-
 __all__ = ["analyze_data", "create_new_data_copy", "create_new_dataset_json", "NNUNETMode"]
 
 
diff --git a/monai/metrics/meandice.py b/monai/metrics/meandice.py
index 16eebdd28c..97a26f6a4e 100644
--- a/monai/metrics/meandice.py
+++ b/monai/metrics/meandice.py
@@ -205,7 +205,7 @@ def compute_channel(self, y_pred: torch.Tensor, y: torch.Tensor) -> torch.Tensor
         denorm = y_o + torch.sum(y_pred)
         if denorm <= 0:
             return torch.tensor(1.0, device=y_o.device)
-        return (2.0 * torch.sum(torch.masked_select(y, y_pred))) / denorm
+        return torch.tensor(0.0, device=y_o.device)
 
     def __call__(self, y_pred: torch.Tensor, y: torch.Tensor) -> torch.Tensor | tuple[torch.Tensor, torch.Tensor]:
         """
diff --git a/monai/networks/nets/basic_unet.py b/monai/networks/nets/basic_unet.py
index b26fdcb622..a29d620ced 100644
--- a/monai/networks/nets/basic_unet.py
+++ b/monai/networks/nets/basic_unet.py
@@ -12,6 +12,7 @@
 from __future__ import annotations
 
 from collections.abc import Sequence
+from typing import Any
 
 import torch
 import torch.nn as nn
@@ -149,16 +150,16 @@ def __init__(
         self.convs = TwoConv(spatial_dims, cat_chns + up_chns, out_chns, act, norm, bias, dropout)
         self.is_pad = is_pad
 
-    def forward(self, x: torch.Tensor, x_e: torch.Tensor | None):
+    def forward(self, x: torch.Tensor, x_e: Any):
         """
 
         Args:
             x: features to be upsampled.
-            x_e: features from the encoder.
+            x_e: optional features from the encoder, if None, this branch is not in use.
         """
         x_0 = self.upsample(x)
 
-        if x_e is not None:
+        if torch.jit.isinstance(x_e, torch.Tensor):
             if self.is_pad:
                 # handling spatial shapes due to the 2x maxpooling with odd edge lengths.
                 dimensions = len(x.shape) - 2
diff --git a/monai/networks/nets/flexible_unet.py b/monai/networks/nets/flexible_unet.py
index a880cafdc3..ac2124b5f9 100644
--- a/monai/networks/nets/flexible_unet.py
+++ b/monai/networks/nets/flexible_unet.py
@@ -232,6 +232,7 @@ def __init__(
         dropout: float | tuple = 0.0,
         decoder_bias: bool = False,
         upsample: str = "nontrainable",
+        pre_conv: str = "default",
         interp_mode: str = "nearest",
         is_pad: bool = True,
     ) -> None:
@@ -262,6 +263,8 @@ def __init__(
             decoder_bias: whether to have a bias term in decoder's convolution blocks.
             upsample: upsampling mode, available options are``"deconv"``, ``"pixelshuffle"``,
                 ``"nontrainable"``.
+            pre_conv:a conv block applied before upsampling. Only used in the "nontrainable" or
+                "pixelshuffle" mode, default to `default`.
             interp_mode: {``"nearest"``, ``"linear"``, ``"bilinear"``, ``"bicubic"``, ``"trilinear"``}
                 Only used in the "nontrainable" mode.
             is_pad: whether to pad upsampling features to fit features from encoder. Default to True.
@@ -309,7 +312,7 @@ def __init__(
             bias=decoder_bias,
             upsample=upsample,
             interp_mode=interp_mode,
-            pre_conv="default",
+            pre_conv=pre_conv,
             align_corners=None,
             is_pad=is_pad,
         )