diff --git a/megatron/core/models/common/language_module/language_module.py b/megatron/core/models/common/language_module/language_module.py
index 9d8eddbdda6..d238e2130d9 100644
--- a/megatron/core/models/common/language_module/language_module.py
+++ b/megatron/core/models/common/language_module/language_module.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 import logging
 import os
 from typing import Optional, Tuple
@@ -202,7 +202,8 @@ def setup_embeddings_and_output_layer(self) -> None:
 
         # Mark embedding and output layer for decoupled_lr and other features.
         # This is the original Megatron attribute used by decoupled_lr, Muon, FSDP, etc.
-        if self.pre_process and hasattr(self, 'embedding'):
+        # Include MTP-stage embedding too: it is a duplicated copy of the pre_process embedding
+        if (self.pre_process or getattr(self, 'mtp_process', False)) and hasattr(self, 'embedding'):
             self.embedding.word_embeddings.weight.is_embedding_or_output_parameter = True
         if (
             self.post_process