From bc36e6cb2cfb81a5c5387b136414294661f42019 Mon Sep 17 00:00:00 2001
From: Zhaodong Bing <45478848+aaab8b@users.noreply.github.com>
Date: Wed, 24 Dec 2025 17:52:56 +0800
Subject: [PATCH 1/2] Update unquantized_fused_moe_method.py

Change assignment of unquantized moe weights when using aiter on rocm, making it safer for reloading the weights. This will solve the random output case after wake-up and reloading weights in reinforcement learning.
---
 .../layers/fused_moe/unquantized_fused_moe_method.py      | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py b/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
index 6182f10aa70f..84862416ee2d 100644
--- a/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
+++ b/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
@@ -207,16 +207,16 @@ def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
         super().process_weights_after_loading(layer)
 
         # Padding the weight for better performance on ROCm
-        layer.w13_weight.data = self._maybe_pad_weight(layer.w13_weight.data)
-        layer.w2_weight.data = self._maybe_pad_weight(layer.w2_weight.data)
+        layer.w13_weight.data.copy_(self._maybe_pad_weight(layer.w13_weight.data))
+        layer.w2_weight.data.copy_(self._maybe_pad_weight(layer.w2_weight.data)）
 
         if self.rocm_aiter_moe_enabled:
             shuffled_w13, shuffled_w2 = rocm_aiter_ops.shuffle_weights(
                 layer.w13_weight.data, layer.w2_weight.data
             )
 
-            layer.w13_weight.data = shuffled_w13
-            layer.w2_weight.data = shuffled_w2
+            layer.w13_weight.data.copy_(shuffled_w13)
+            layer.w2_weight.data.copy_(shuffled_w2)
 
         if self.flashinfer_cutlass_moe_enabled:
             # Swap halves to arrange as [w3; w1] (kernel expectation)

From e1a762a3962372e22c802308c9519baa7be8526b Mon Sep 17 00:00:00 2001
From: Zhaodong Bing <45478848+aaab8b@users.noreply.github.com>
Date: Wed, 24 Dec 2025 17:56:01 +0800
Subject: [PATCH 2/2] Update unquantized_fused_moe_method.py

Signed-off-by: Zhaodong Bing <45478848+aaab8b@users.noreply.github.com>
---
 .../layers/fused_moe/unquantized_fused_moe_method.py            | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py b/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
index 84862416ee2d..bb7c50fb3c1a 100644
--- a/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
+++ b/vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py
@@ -208,7 +208,7 @@ def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
 
         # Padding the weight for better performance on ROCm
         layer.w13_weight.data.copy_(self._maybe_pad_weight(layer.w13_weight.data))
-        layer.w2_weight.data.copy_(self._maybe_pad_weight(layer.w2_weight.data)）
+        layer.w2_weight.data.copy_(self._maybe_pad_weight(layer.w2_weight.data))
 
         if self.rocm_aiter_moe_enabled:
             shuffled_w13, shuffled_w2 = rocm_aiter_ops.shuffle_weights(