Prefer raising TypeError exception for invalid type

src/transformers/cache_utils.py

@@ -1313,7 +1313,7 @@ def check_dynamic_cache(self, method: str):
             isinstance(self.self_attention_cache, DynamicCache)
             and isinstance(self.cross_attention_cache, DynamicCache)
         ):
-            raise ValueError(
+            raise TypeError(
                 f"`{method}` is only defined for dynamic cache, got {self.self_attention_cache.__str__()} for the self "
                 f"attention cache and {self.cross_attention_cache.__str__()} for the cross attention cache."
             )

src/transformers/commands/serving.py

@@ -1186,7 +1186,7 @@ def generate_response(self, req: dict) -> Generator[str, None, None]:
             inputs = [{"role": "system", "content": req["instructions"]}] if "instructions" in req else []
             inputs.append(req["input"])
         else:
-            raise ValueError("inputs should be a list, dict, or str")
+            raise TypeError("inputs should be a list, dict, or str")
 
         inputs = processor.apply_chat_template(inputs, add_generation_prompt=True, return_tensors="pt")
         inputs = inputs.to(model.device)

src/transformers/models/evolla/processing_evolla.py

@@ -165,7 +165,7 @@ def __call__(
         if isinstance(messages_list, (list, tuple)):
             for messages in messages_list:
                 if not isinstance(messages, (list, tuple)):
-                    raise ValueError(f"Each messages in messages_list should be a list instead of {type(messages)}.")
+                    raise TypeError(f"Each messages in messages_list should be a list instead of {type(messages)}.")
                 if not all(isinstance(m, dict) for m in messages):
                     raise ValueError(
                         "Each message in messages_list should be a list of dictionaries, but not all elements are dictionaries."

src/transformers/models/gemma3n/processing_gemma3n.py

@@ -107,7 +107,7 @@ def __call__(
         if isinstance(text, str):
             text = [text]
         elif not isinstance(text, list) and not isinstance(text[0], str):
-            raise ValueError("Invalid input text. Please provide a string, or a list of strings")
+            raise TypeError("Invalid input text. Please provide a string, or a list of strings")
 
         if audio is not None:
             audio_inputs = self.feature_extractor(audio, **output_kwargs["audio_kwargs"])

src/transformers/models/lfm2_vl/processing_lfm2_vl.py

@@ -135,7 +135,7 @@ def __call__(
         if isinstance(text, str):
             text = [text]
         elif not isinstance(text, list) and not isinstance(text[0], str):
-            raise ValueError("Invalid input text. Please provide a string, or a list of strings")
+            raise TypeError("Invalid input text. Please provide a string, or a list of strings")
 
         n_images_in_text = [sample.count(self.image_token) for sample in text]
         if sum(n_images_in_text) > 0 and images is None:

src/transformers/models/ovis2/processing_ovis2.py

@@ -118,7 +118,7 @@ def __call__(
         if isinstance(text, str):
             text = [text]
         elif not isinstance(text, list) and not isinstance(text[0], str):
-            raise ValueError("Invalid input text. Please provide a string, or a list of strings")
+            raise TypeError("Invalid input text. Please provide a string, or a list of strings")
 
         image_inputs = {}
 

src/transformers/models/perception_lm/processing_perception_lm.py

@@ -144,7 +144,7 @@ def __call__(
         if isinstance(text, str):
             text = [text]
         elif not isinstance(text, list) and not isinstance(text[0], str):
-            raise ValueError("Invalid input text. Please provide a string, or a list of strings")
+            raise TypeError("Invalid input text. Please provide a string, or a list of strings")
 
         # try to expand inputs in processing if we have the necessary parts
         prompt_strings = []

src/transformers/models/sam/image_processing_sam_fast.py

@@ -390,7 +390,7 @@ def post_process_masks(
             if isinstance(masks[i], np.ndarray):
                 masks[i] = torch.from_numpy(masks[i])
             elif not isinstance(masks[i], torch.Tensor):
-                raise ValueError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
+                raise TypeError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
             interpolated_mask = F.interpolate(masks[i], target_image_size, mode="bilinear", align_corners=False)
             interpolated_mask = interpolated_mask[..., : reshaped_input_sizes[i][0], : reshaped_input_sizes[i][1]]
             interpolated_mask = F.interpolate(interpolated_mask, original_size, mode="bilinear", align_corners=False)

src/transformers/models/sam2/image_processing_sam2_fast.py

@@ -669,7 +669,7 @@ def post_process_masks(
             if isinstance(masks[i], np.ndarray):
                 masks[i] = torch.from_numpy(masks[i])
             elif not isinstance(masks[i], torch.Tensor):
-                raise ValueError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
+                raise TypeError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
             interpolated_mask = F.interpolate(masks[i], original_size, mode="bilinear", align_corners=False)
             if apply_non_overlapping_constraints:
                 interpolated_mask = self._apply_non_overlapping_constraints(interpolated_mask)

src/transformers/models/sam2/modular_sam2.py

@@ -287,7 +287,7 @@ def post_process_masks(
             if isinstance(masks[i], np.ndarray):
                 masks[i] = torch.from_numpy(masks[i])
             elif not isinstance(masks[i], torch.Tensor):
-                raise ValueError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
+                raise TypeError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
             interpolated_mask = F.interpolate(masks[i], original_size, mode="bilinear", align_corners=False)
             if apply_non_overlapping_constraints:
                 interpolated_mask = self._apply_non_overlapping_constraints(interpolated_mask)

src/transformers/models/sam2/processing_sam2.py

@@ -258,7 +258,7 @@ def _convert_to_nested_list(self, data, expected_depth, current_depth=0):
         elif isinstance(data, (int, float)):
             return data
         else:
-            raise ValueError(f"Unsupported data type: {type(data)}")
+            raise TypeError(f"Unsupported data type: {type(data)}")
 
     def _get_nested_dimensions(self, nested_list, max_dims=None):
         """

src/transformers/models/sam2_video/processing_sam2_video.py

@@ -262,7 +262,7 @@ def _convert_to_nested_list(self, data, expected_depth, current_depth=0):
         elif isinstance(data, (int, float)):
             return data
         else:
-            raise ValueError(f"Unsupported data type: {type(data)}")
+            raise TypeError(f"Unsupported data type: {type(data)}")
 
     def _get_nested_dimensions(self, nested_list, max_dims=None):
         """

src/transformers/models/sam2_video/video_processing_sam2_video.py

@@ -92,7 +92,7 @@ def post_process_masks(
             if isinstance(masks[i], np.ndarray):
                 masks[i] = torch.from_numpy(masks[i])
             elif not isinstance(masks[i], torch.Tensor):
-                raise ValueError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
+                raise TypeError("Input masks should be a list of `torch.tensors` or a list of `np.ndarray`")
             interpolated_mask = F_t.interpolate(masks[i], target_image_size, mode="bilinear", align_corners=False)
             interpolated_mask = interpolated_mask[..., : reshaped_input_sizes[i][0], : reshaped_input_sizes[i][1]]
             interpolated_mask = F_t.interpolate(interpolated_mask, original_size, mode="bilinear", align_corners=False)

src/transformers/quantizers/auto.py

@@ -287,7 +287,7 @@ def register_quantizer_fn(cls):
             raise ValueError(f"Quantizer '{name}' already registered")
 
         if not issubclass(cls, HfQuantizer):
-            raise ValueError("Quantizer must extend HfQuantizer")
+            raise TypeError("Quantizer must extend HfQuantizer")
 
         AUTO_QUANTIZER_MAPPING[name] = cls
         return cls

src/transformers/tokenization_mistral_common.py

@@ -1433,7 +1433,7 @@ def apply_chat_template(
                 f"Kwargs {list(kwargs.keys())} are not supported by `MistralCommonTokenizer.apply_chat_template`."
             )
         if not isinstance(truncation, bool):
-            raise ValueError("`truncation` must be a boolean for `apply_chat_template` method.")
+            raise TypeError("`truncation` must be a boolean for `apply_chat_template` method.")
 
         if isinstance(conversation, (list, tuple)) and (
             isinstance(conversation[0], (list, tuple)) or hasattr(conversation[0], "messages")

tests/models/sam2/test_processor_sam2.py

@@ -140,5 +140,5 @@ def test_post_process_masks(self):
         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))

tests/models/sam2_video/test_processor_sam2_video.py

@@ -149,5 +149,5 @@ def test_post_process_masks(self):
         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))

tests/test_tokenization_mistral_common.py

@@ -945,7 +945,7 @@ def test_appsly_chat_template_with_truncation(self):
 
         # Test 3:
         # assert truncation is boolean
-        with self.assertRaises(ValueError):
+        with self.assertRaises(TypeError):
             self.tokenizer.apply_chat_template(
                 conversation, tokenize=True, truncation=TruncationStrategy.LONGEST_FIRST, max_length=20
             )
@@ -1189,7 +1189,7 @@ def test_batch_apply_chat_template_with_truncation(
 
         # Test 3:
         # assert truncation is boolean
-        with self.assertRaises(ValueError):
+        with self.assertRaises(TypeError):
             self.tokenizer.apply_chat_template(
                 self.fixture_conversations, tokenize=True, truncation=TruncationStrategy.LONGEST_FIRST, max_length=20
             )