Fix lint issues from torchfix upgrade

Author: huydhn

backends/arm/test/models/test_mobilenet_v2_arm.py

@@ -9,11 +9,11 @@
 import unittest
 
 import torch
-import torchvision.models as models
 from executorch.backends.arm.test import common
 
 from executorch.backends.arm.test.tester.arm_tester import ArmTester
 from executorch.backends.xnnpack.test.tester.tester import Quantize
+from torchvision import models
 from torchvision.models.mobilenetv2 import MobileNet_V2_Weights
 
 

backends/xnnpack/test/models/inception_v3.py

@@ -7,9 +7,9 @@
 import unittest
 
 import torch
-import torchvision.models as models
 from executorch.backends.xnnpack.test.tester import Tester
 from executorch.backends.xnnpack.test.tester.tester import Quantize
+from torchvision import models
 
 
 class TestInceptionV3(unittest.TestCase):

backends/xnnpack/test/models/mobilenet_v2.py

@@ -7,9 +7,9 @@
 import unittest
 
 import torch
-import torchvision.models as models
 from executorch.backends.xnnpack.test.tester import Tester
 from executorch.backends.xnnpack.test.tester.tester import Quantize
+from torchvision import models
 from torchvision.models.mobilenetv2 import MobileNet_V2_Weights
 
 

backends/xnnpack/test/models/mobilenet_v3.py

@@ -7,9 +7,9 @@
 import unittest
 
 import torch
-import torchvision.models as models
 from executorch.backends.xnnpack.test.tester import Tester
 from executorch.backends.xnnpack.test.tester.tester import Quantize
+from torchvision import models
 
 
 class TestMobileNetV3(unittest.TestCase):

backends/xnnpack/test/models/torchvision_vit.py

@@ -7,8 +7,8 @@
 import unittest
 
 import torch
-import torchvision.models as models
 from executorch.backends.xnnpack.test.tester import Tester
+from torchvision import models
 
 
 class TestViT(unittest.TestCase):

examples/qualcomm/oss_scripts/ssd300_vgg16.py

@@ -109,7 +109,9 @@ def SSD300VGG16(pretrained_weight_model):
     from model import SSD300
 
     model = SSD300(n_classes=21)
-    checkpoint = torch.load(pretrained_weight_model, map_location="cpu")
+    checkpoint = torch.load(
+        pretrained_weight_model, map_location="cpu", weights_only=False
+    )
     model.load_state_dict(checkpoint["model"].state_dict())
 
     return model.eval()

examples/qualcomm/scripts/mobilebert_fine_tune.py

@@ -211,6 +211,7 @@ def get_fine_tuned_mobilebert(artifacts_dir, pretrained_weight, batch_size):
                 else pretrained_weight
             ),
             map_location=torch.device("cpu"),
+            weights_only=False,
         ),
     )
 

exir/serde/export_serialize.py

@@ -242,7 +242,7 @@ def deserialize_torch_artifact(serialized: bytes):
         return {}
     buffer = io.BytesIO(serialized)
     buffer.seek(0)
-    return torch.load(buffer)
+    return torch.load(buffer, weights_only=False)
 
 
 def _sympy_int_to_int(val: sympy.Expr):