diff --git a/vllm_ascend/patch/__init__.py b/vllm_ascend/patch/__init__.py
index 6de4cd1..d4286bd 100644
--- a/vllm_ascend/patch/__init__.py
+++ b/vllm_ascend/patch/__init__.py
@@ -90,7 +90,7 @@
 #
 # * Worker Patch:
 # ===============
-# ** File: worker/patch_common/patch_metrics.py **
+# ** File: worker/patch_0_8_4/patch_metrics.py **
 # ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 #   1. `vllm.spec_decode.metrics.AsyncMetricsCollector.init_tensors` and
 #       `vllm.spec_decode.metrics.AsyncMetricsCollector._copy_rejsample_metrics_async`
@@ -104,7 +104,9 @@
 #    Future Plan:
 #       Revert it when the related pr is merged in vllm.
 #
-#   2. `vllm.spec_decode.metrics.AsyncMetricsCollector.maybe_collect_rejsample_metrics`
+# ** File: worker/patch_common/patch_metrics.py **
+# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+#   1. `vllm.spec_decode.metrics.AsyncMetricsCollector.maybe_collect_rejsample_metrics`
 #    Why:
 #       There are cuda hard code (current_platform.is_cuda_alike()) in
 #       `AsyncMetricsCollector.maybe_collect_rejsample_metrics`
diff --git a/vllm_ascend/patch/worker/patch_0_8_4/__init__.py b/vllm_ascend/patch/worker/patch_0_8_4/__init__.py
index 2ed088b..2465b3f 100644
--- a/vllm_ascend/patch/worker/patch_0_8_4/__init__.py
+++ b/vllm_ascend/patch/worker/patch_0_8_4/__init__.py
@@ -13,4 +13,6 @@
 # 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.
-#
\ No newline at end of file
+#
+
+import vllm_ascend.patch.worker.patch_0_8_4.patch_metrics  # noqa
diff --git a/vllm_ascend/patch/worker/patch_0_8_4/patch_metrics.py b/vllm_ascend/patch/worker/patch_0_8_4/patch_metrics.py
new file mode 100644
index 0000000..4ba223f
--- /dev/null
+++ b/vllm_ascend/patch/worker/patch_0_8_4/patch_metrics.py
@@ -0,0 +1,60 @@
+#
+# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
+# This file is a part of the vllm-ascend project.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+#
+
+from typing import Callable, Optional, Union
+
+import torch
+import torch_npu
+from vllm.spec_decode.metrics import (AsyncMetricsCollector,
+                                      SpecDecodeWorkerMetrics)
+
+Timer = Callable[[], float]
+
+# TODO: revert this patch when the cuda hard code is removed in vllm
+# init_tensors: Modified the hard-coded cuda judgment logic to npu;
+# maybe_collect_rejsample_metrics: Removed the check for current_platform.is_cuda_alike()
+
+
+def init_tensors(self,
+                 rank: int,
+                 device_type: Union[torch.device, str] = 'npu') -> None:
+    self._rank = rank
+    if isinstance(device_type, torch.device):
+        device_type = device_type.type
+    if device_type == 'npu':
+        self._copy_stream = torch_npu.npu.Stream()
+
+
+def maybe_collect_rejsample_metrics(
+        self, k: int) -> Optional[SpecDecodeWorkerMetrics]:
+
+    # If a copy was initiated in the previous call, collect and return.
+    if self._in_flight_copy is not None:
+        ready_event = self._in_flight_copy
+        self._in_flight_copy = None
+        return self._collect_rejsample_metrics(k, ready_event)
+
+    # Otherwise, check if we should start a new copy.
+    if self._should_collect_rejsample_metrics(self._timer()):
+        assert self._in_flight_copy is None
+        self._in_flight_copy = self._copy_rejsample_metrics_async()
+
+    return None
+
+
+AsyncMetricsCollector.init_tensors = init_tensors
+AsyncMetricsCollector.maybe_collect_rejsample_metrics = maybe_collect_rejsample_metrics
diff --git a/vllm_ascend/patch/worker/patch_common/patch_metrics.py b/vllm_ascend/patch/worker/patch_common/patch_metrics.py
index 685755f..6d1f2dc 100644
--- a/vllm_ascend/patch/worker/patch_common/patch_metrics.py
+++ b/vllm_ascend/patch/worker/patch_common/patch_metrics.py
@@ -15,46 +15,13 @@
 # limitations under the License.
 #
 
-from typing import Callable, Optional, Union
+from typing import Callable
 
 import torch
-import torch_npu
-from vllm.spec_decode.metrics import (AsyncMetricsCollector,
-                                      SpecDecodeWorkerMetrics)
+from vllm.spec_decode.metrics import AsyncMetricsCollector
 
 Timer = Callable[[], float]
 
-# TODO: revert this patch when the cuda hard code is removed in vllm
-# init_tensors: Modified the hard-coded cuda judgment logic to npu;
-# maybe_collect_rejsample_metrics: Removed the check for current_platform.is_cuda_alike()
-
-
-def init_tensors(self,
-                 rank: int,
-                 device_type: Union[torch.device, str] = 'npu') -> None:
-    self._rank = rank
-    if isinstance(device_type, torch.device):
-        device_type = device_type.type
-    if device_type == 'npu':
-        self._copy_stream = torch_npu.npu.Stream()
-
-
-def maybe_collect_rejsample_metrics(
-        self, k: int) -> Optional[SpecDecodeWorkerMetrics]:
-
-    # If a copy was initiated in the previous call, collect and return.
-    if self._in_flight_copy is not None:
-        ready_event = self._in_flight_copy
-        self._in_flight_copy = None
-        return self._collect_rejsample_metrics(k, ready_event)
-
-    # Otherwise, check if we should start a new copy.
-    if self._should_collect_rejsample_metrics(self._timer()):
-        assert self._in_flight_copy is None
-        self._in_flight_copy = self._copy_rejsample_metrics_async()
-
-    return None
-
 
 def _copy_rejsample_metrics_async(self) -> torch.npu.Event:
     """
@@ -83,6 +50,4 @@ def _copy_rejsample_metrics_async(self) -> torch.npu.Event:
     return aggregate_metrics_ready
 
 
-AsyncMetricsCollector.init_tensors = init_tensors
-AsyncMetricsCollector.maybe_collect_rejsample_metrics = maybe_collect_rejsample_metrics
 AsyncMetricsCollector._copy_rejsample_metrics_async = _copy_rejsample_metrics_async