diff --git a/src/axolotl/integrations/diff_transformer/patches.py b/src/axolotl/monkeypatch/attention/differential.py
similarity index 95%
rename from src/axolotl/integrations/diff_transformer/patches.py
rename to src/axolotl/monkeypatch/attention/differential.py
index 37ad0a981b..037a6f0bd2 100644
--- a/src/axolotl/integrations/diff_transformer/patches.py
+++ b/src/axolotl/monkeypatch/attention/differential.py
@@ -3,7 +3,7 @@
 from transformers import PreTrainedModel
 from transformers.models.llama.modeling_llama import LLAMA_ATTENTION_CLASSES
 
-from .multihead_diffattn import (
+from axolotl.integrations.diff_transformer.multihead_diffattn import (
     LlamaDifferentialAttention,
     LlamaDifferentialSdpaAttention,
 )
diff --git a/src/axolotl/utils/models.py b/src/axolotl/utils/models.py
index 0b6277b6ff..5d51fd4368 100644
--- a/src/axolotl/utils/models.py
+++ b/src/axolotl/utils/models.py
@@ -456,7 +456,7 @@ def apply_patches(self) -> None:
             patch_mistral_cross_entropy()
 
         if self.cfg.diff_attention:
-            from axolotl.integrations.diff_transformer.patches import (
+            from axolotl.monkeypatch.attention.differential import (
                 patch_llama_attention_classes,
             )