diff --git a/autodistill_grounded_sam_2/grounded_sam_2.py b/autodistill_grounded_sam_2/grounded_sam_2.py
index b924f77..550f023 100644
--- a/autodistill_grounded_sam_2/grounded_sam_2.py
+++ b/autodistill_grounded_sam_2/grounded_sam_2.py
@@ -13,7 +13,11 @@
 from autodistill.helpers import load_image
 from autodistill_florence_2 import Florence2
 
-from autodistill_grounded_sam_2.helpers import load_SAM, load_grounding_dino, combine_detections
+from autodistill_grounded_sam_2.helpers import (
+    combine_detections,
+    load_grounding_dino,
+    load_SAM,
+)
 
 HOME = os.path.expanduser("~")
 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
@@ -22,16 +26,25 @@
 
 SUPPORTED_GROUNDING_MODELS = ["Florence 2", "Grounding DINO"]
 
+
 @dataclass
 class GroundedSAM2(DetectionBaseModel):
     ontology: CaptionOntology
     box_threshold: float
     text_threshold: float
 
-    def __init__(self, ontology: CaptionOntology, model = "Florence 2", grounding_dino_box_threshold = 0.35, grounding_dino_text_threshold = 0.25):
+    def __init__(
+        self,
+        ontology: CaptionOntology,
+        model="Florence 2",
+        grounding_dino_box_threshold=0.35,
+        grounding_dino_text_threshold=0.25,
+    ):
         if model not in SUPPORTED_GROUNDING_MODELS:
-            raise ValueError(f"Grounding model {model} is not supported. Supported models are {SUPPORTED_GROUNDING_MODELS}")
-        
+            raise ValueError(
+                f"Grounding model {model} is not supported. Supported models are {SUPPORTED_GROUNDING_MODELS}"
+            )
+
         self.ontology = ontology
         if model == "Florence 2":
             self.florence_2_predictor = Florence2(ontology=ontology)
diff --git a/autodistill_grounded_sam_2/helpers.py b/autodistill_grounded_sam_2/helpers.py
index 04e1e3d..dfa7497 100644
--- a/autodistill_grounded_sam_2/helpers.py
+++ b/autodistill_grounded_sam_2/helpers.py
@@ -2,17 +2,18 @@
 import subprocess
 import sys
 import urllib.request
-from groundingdino.util.inference import Model
 
-import torch
 import numpy as np
 import supervision as sv
+import torch
+from groundingdino.util.inference import Model
 
 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 if not torch.cuda.is_available():
     print("WARNING: CUDA not available. GroundingDINO will run very slowly.")
 
+
 def load_grounding_dino():
     AUTODISTILL_CACHE_DIR = os.path.expanduser("~/.cache/autodistill")
 
@@ -56,6 +57,7 @@ def load_grounding_dino():
 
         return grounding_dino_model
 
+
 def load_SAM():
     cur_dir = os.getcwd()
 
@@ -63,6 +65,9 @@ def load_SAM():
     SAM_CACHE_DIR = os.path.join(AUTODISTILL_CACHE_DIR, "segment_anything_2")
     SAM_CHECKPOINT_PATH = os.path.join(SAM_CACHE_DIR, "sam2_hiera_base_plus.pth")
 
+    SAM_REPOSITORY_NAME = "segment-anything-2"
+    SAM_REPOSITORY_DIR = os.path.join(SAM_CACHE_DIR, SAM_REPOSITORY_NAME)
+
     url = "https://dl.fbaipublicfiles.com/segment_anything_2/072824/sam2_hiera_base_plus.pt"
 
     # Create the destination directory if it doesn't exist
@@ -70,7 +75,7 @@ def load_SAM():
 
     os.chdir(SAM_CACHE_DIR)
 
-    if not os.path.isdir("~/.cache/autodistill/segment_anything_2/segment-anything-2"):
+    if not os.path.isdir(SAM_REPOSITORY_DIR):
         subprocess.run(
             [
                 "git",
@@ -79,11 +84,11 @@ def load_SAM():
             ]
         )
 
-        os.chdir("segment-anything-2")
+        os.chdir(SAM_REPOSITORY_NAME)
 
         subprocess.run(["pip", "install", "-e", "."])
 
-    sys.path.append("~/.cache/autodistill/segment_anything_2/segment-anything-2")
+    sys.path.append(SAM_REPOSITORY_DIR)
 
     # Download the file if it doesn't exist
     if not os.path.isfile(SAM_CHECKPOINT_PATH):
@@ -92,17 +97,14 @@ def load_SAM():
     from sam2.build_sam import build_sam2
     from sam2.sam2_image_predictor import SAM2ImagePredictor
 
-    checkpoint = "~/.cache/autodistill/segment_anything_2/sam2_hiera_base_plus.pth"
-    checkpoint = os.path.expanduser(checkpoint)
     model_cfg = "sam2_hiera_b+.yaml"
-    predictor = SAM2ImagePredictor(build_sam2(model_cfg, checkpoint))
+    predictor = SAM2ImagePredictor(build_sam2(model_cfg, SAM_CHECKPOINT_PATH))
 
     os.chdir(cur_dir)
 
     return predictor
 
 
-
 def combine_detections(detections_list, overwrite_class_ids):
     if len(detections_list) == 0:
         return sv.Detections.empty()
@@ -156,4 +158,3 @@ def combine_detections(detections_list, overwrite_class_ids):
         class_id=class_id,
         tracker_id=tracker_id,
     )
-