diff --git a/modules/stage_apply_loras.py b/modules/stage_apply_loras.py
index 6f7352e..b009d62 100644
--- a/modules/stage_apply_loras.py
+++ b/modules/stage_apply_loras.py
@@ -57,15 +57,21 @@ def process(self, data, stage_input):
 
         base_model = access.get_from_pipeline(Names.P_BASE_MODEL)
         base_clip = access.get_from_pipeline(Names.P_BASE_CLIP)
+        refiner_model = access.get_from_pipeline(Names.P_REFINER_MODEL)
+        refiner_clip = access.get_from_pipeline(Names.P_REFINER_CLIP)
 
         base_model_changed = access.changed_in_pipeline(Names.P_BASE_MODEL)
         base_clip_changed = access.changed_in_pipeline(Names.P_BASE_CLIP)
+        refiner_model_changed = access.changed_in_pipeline(Names.P_REFINER_MODEL)
+        refiner_clip_changed = access.changed_in_pipeline(Names.P_REFINER_CLIP)
 
         lora_stack = access.get_active_setting(UI.S_LORAS, UI.F_LORA_STACK, [])
 
         any_changes = (
                 base_model_changed or
-                base_clip_changed
+                base_clip_changed or
+                refiner_model_changed or
+                refiner_clip_changed
         )
 
         applied_loras = []
@@ -82,15 +88,22 @@ def process(self, data, stage_input):
                 if lora_name is not None and lora_name != UI.NONE and lora_strength != 0.0:
                     (base_model, base_clip) = NodeWrapper.lora_loader.load_lora(base_model, base_clip, lora_name,
                                                                                 lora_strength, lora_strength)
+                    if refiner_model is not None and refiner_clip is not None:
+                        (refiner_model, refiner_clip) = NodeWrapper.lora_loader.load_lora(refiner_model, refiner_clip, lora_name,
+                                                                                          lora_strength, lora_strength)
                     applied_loras.append(lora_name)
 
-            access.update_in_cache(Names.C_APPLIED_LORAS, lora_stack, (base_model, base_clip))
+            access.update_in_cache(Names.C_APPLIED_LORAS, lora_stack, (base_model, base_clip, refiner_model, refiner_clip))
             access.update_in_pipeline(Names.P_BASE_MODEL, base_model)
             access.update_in_pipeline(Names.P_BASE_CLIP, base_clip)
+            access.update_in_pipeline(Names.P_REFINER_MODEL, refiner_model)
+            access.update_in_pipeline(Names.P_REFINER_CLIP, refiner_clip)
         else:
-            (base_model, base_clip) = access.get_from_cache(Names.C_APPLIED_LORAS)
+            (base_model, base_clip, refiner_model, refiner_clip) = access.get_from_cache(Names.C_APPLIED_LORAS)
             access.restore_in_pipeline(Names.P_BASE_MODEL, base_model)
             access.restore_in_pipeline(Names.P_BASE_CLIP, base_clip)
+            access.restore_in_pipeline(Names.P_REFINER_MODEL, refiner_model)
+            access.restore_in_pipeline(Names.P_REFINER_CLIP, refiner_clip)
 
         loaded_loras = {
             Names.F_LORA_NAMES: applied_loras,