diff --git a/scripts/!adetailer.py b/scripts/!adetailer.py
index f835e46..f2391df 100644
--- a/scripts/!adetailer.py
+++ b/scripts/!adetailer.py
@@ -8,10 +8,9 @@ import torch
 import modules
 from adetailer import __version__, get_models, mediapipe_predict, ultralytics_predict
 from adetailer.common import dilate_erode, is_all_black, offset
-from modules import devices, images, script_callbacks, scripts, shared
+from modules import devices, images, safe, script_callbacks, scripts, shared
 from modules.paths import data_path, models_path
 from modules.processing import StableDiffusionProcessingImg2Img, process_images
-from modules.safe import load, unsafe_torch_load
 from modules.shared import opts, state
 
 AFTER_DETAILER = "After Detailer"
@@ -56,11 +55,21 @@ class ADetailerArgs:
         }
 
 
-def with_gc(func):
+def with_params_txt(func):
+    params_txt = Path(data_path, "params.txt")
+    original_params = ""
+
     def wrapper(*args, **kwargs):
-        devices.torch_gc()
+        nonlocal original_params
+
+        if not original_params and params_txt.exists():
+            original_params = params_txt.read_text(encoding="utf-8")
+
         result = func(*args, **kwargs)
-        devices.torch_gc()
+
+        if original_params:
+            params_txt.write_text(original_params, encoding="utf-8")
+
         return result
 
     return wrapper
@@ -68,10 +77,11 @@ def with_gc(func):
 
 class ChangeTorchLoad:
     def __enter__(self):
-        torch.load = unsafe_torch_load
+        self.orig = torch.load
+        torch.load = safe.unsafe_torch_load
 
     def __exit__(self, *args, **kwargs):
-        torch.load = load
+        torch.load = self.orig
 
 
 def gr_show(visible=True):
@@ -276,7 +286,7 @@ class AfterDetailerScript(scripts.Script):
     def get_args(*args):
         return ADetailerArgs(*args)
 
-    @with_gc
+    @with_params_txt
     def postprocess_image(self, p, pp, *args_):
         if getattr(p, "_disable_adetailer", False):
             return
@@ -305,12 +315,6 @@ class AfterDetailerScript(scripts.Script):
         if sampler_name in ["PLMS", "UniPC"]:
             sampler_name = "Euler"
 
-        params_txt = Path(data_path, "params.txt")
-        original_params = ""
-        if params_txt.exists():
-            with params_txt.open("r", encoding="utf-8") as f:
-                original_params = f.read()
-
         i2i = StableDiffusionProcessingImg2Img(
             init_images=[pp.image],
             resize_mode=0,
@@ -359,14 +363,12 @@ class AfterDetailerScript(scripts.Script):
         with ChangeTorchLoad():
             pred = predictor(ad_model, pp.image, args.ad_conf)
 
+        devices.torch_gc()
+
         if pred.masks is None:
             print("ADetailer: nothing detected with current settings")
             return
 
-        masks = pred.masks
-        steps = len(masks)
-        processed = None
-
         if opts.data.get("ad_save_previews", False):
             images.save_image(
                 pred.preview,
@@ -379,13 +381,18 @@ class AfterDetailerScript(scripts.Script):
                 suffix="-ad-preview",
             )
 
+        masks = pred.masks
+        steps = len(masks)
+        processed = None
+
         for j in range(steps):
             mask = masks[j]
+
             mask = dilate_erode(mask, args.ad_dilate_erode)
             if is_all_black(mask):
                 continue
-            mask = offset(mask, args.ad_x_offset, args.ad_y_offset)
 
+            mask = offset(mask, args.ad_x_offset, args.ad_y_offset)
             i2i.image_mask = mask
 
             processed = process_images(i2i)
@@ -396,10 +403,6 @@ class AfterDetailerScript(scripts.Script):
         if processed is not None:
             pp.image = processed.images[0]
 
-        if original_params:
-            with params_txt.open("w", encoding="utf-8") as f:
-                f.write(original_params)
-
 
 def on_ui_settings():
     section = ("ADetailer", AFTER_DETAILER)