Raise TypeError instead of ValueError for invalid types (#38660)

* Raise TypeError instead of ValueError for invalid types.

* Removed un-necessary changes.

* Resolved conflicts

* Code quality

* Fix failing tests.

* Fix failing tests.

tests/models/clip/test_tokenization_clip.py

@@ -174,7 +174,7 @@ class CLIPTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     def test_log_warning(self):
         # Test related to the breaking change introduced in transformers v4.17.0
         # We need to check that an error in raised when the user try to load a previous version of the tokenizer.
-        with self.assertRaises(ValueError) as context:
+        with self.assertRaises(TypeError) as context:
             self.get_rust_tokenizer("robot-test/old-clip-tokenizer")
 
         self.assertTrue(

tests/models/grounding_dino/test_modeling_grounding_dino.py

@@ -61,7 +61,7 @@ def generate_fake_bounding_boxes(n_boxes):
     """Generate bounding boxes in the format (center_x, center_y, width, height)"""
     # Validate the input
     if not isinstance(n_boxes, int):
-        raise ValueError("n_boxes must be an integer")
+        raise TypeError("n_boxes must be an integer")
     if n_boxes <= 0:
         raise ValueError("n_boxes must be a positive integer")
 

tests/models/sam/test_processor_sam.py

@@ -154,7 +154,7 @@ class SamProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertEqual(masks[0].shape, (1, 3, 1764, 2646))
 
         dummy_masks = [[1, 0], [0, 1]]
-        with self.assertRaises(ValueError):
+        with self.assertRaises(TypeError):
             masks = processor.post_process_masks(dummy_masks, np.array(original_sizes), np.array(reshaped_input_size))
 
     def test_rle_encoding(self):

tests/models/sam_hq/test_processor_samhq.py

@@ -163,5 +163,5 @@ class SamHQProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertEqual(masks[0].shape, (1, 3, 1764, 2646))
 
         dummy_masks = [[1, 0], [0, 1]]
-        with self.assertRaises(ValueError):
+        with self.assertRaises(TypeError):
             masks = processor.post_process_masks(dummy_masks, np.array(original_sizes), np.array(reshaped_input_size))