Fix more pylint warnings (#40204)

Fix pylint warnings

Signed-off-by: cyy <cyyever@outlook.com>

Author: cyyever

src/transformers/image_transforms.py

@@ -748,7 +748,7 @@ def _expand_for_data_format(values):
         elif isinstance(values, tuple) and len(values) == 2 and isinstance(values[0], int):
             values = (values, values)
         elif isinstance(values, tuple) and len(values) == 2 and isinstance(values[0], tuple):
-            values = values
+            pass
         else:
             raise ValueError(f"Unsupported format: {values}")
 

src/transformers/integrations/mxfp4.py

@@ -228,7 +228,7 @@ def routing_torch_dist(
 
     with torch.cuda.device(logits.device):
         world_size = torch.distributed.get_world_size()
-        rank = int(os.environ.get("LOCAL_RANK", 0))
+        rank = int(os.environ.get("LOCAL_RANK", "0"))
         replace_value = -1
 
         n_tokens = logits.shape[0]

src/transformers/modeling_flash_attention_utils.py

@@ -112,7 +112,6 @@ def _lazy_define_process_function(flash_function):
     NOTE: While all supported kwargs are marked as `True`, everything else is marked as `False`.
           This might be confusing for kwargs that we use in any case, e.g. `is_causal`.
     """
-    global _process_flash_kwargs_fn, _hf_api_to_flash_mapping
 
     flash_parameters = inspect.signature(flash_function).parameters
     process_parameters = inspect.signature(_process_flash_attention_kwargs).parameters

src/transformers/modeling_rope_utils.py

@@ -257,7 +257,7 @@ def find_correction_range(low_rot, high_rot, dim, base, max_position_embeddings,
         low = find_correction_dim(low_rot, dim, base, max_position_embeddings)
         high = find_correction_dim(high_rot, dim, base, max_position_embeddings)
         if truncate:
-            low = low = math.floor(low)
+            low = math.floor(low)
             high = math.ceil(high)
         return max(low, 0), min(high, dim - 1)
 

src/transformers/modeling_utils.py

@@ -409,8 +409,7 @@ def get_state_dict_dtype(state_dict):
             return t.dtype
 
     # if no floating dtype was found return whatever the first dtype is
-    else:
-        return next(state_dict.values()).dtype
+    return next(state_dict.values()).dtype
 
 
 def load_sharded_checkpoint(model, folder, strict=True, prefer_safe=True):

src/transformers/pipelines/table_question_answering.py

@@ -267,7 +267,6 @@ def sequential_inference(self, **inputs):
                 )
 
                 coords_to_probs = collections.defaultdict(list)
-                token_type_ids_example = token_type_ids_example
                 for i, p in enumerate(tf.squeeze(probabilities).numpy().tolist()):
                     segment_id = token_type_ids_example[:, 0].tolist()[i]
                     col = token_type_ids_example[:, 1].tolist()[i] - 1

src/transformers/processing_utils.py

@@ -1391,12 +1391,11 @@ def get_possibly_dynamic_module(module_name):
                             return custom_subclass
                 elif custom_class is not None and custom_class.__name__ == module_name:
                     return custom_class
-        else:
-            raise ValueError(
-                f"Could not find module {module_name} in `transformers`. If this is a custom class, "
-                f"it should be registered using the relevant `AutoClass.register()` function so that "
-                f"other functions can find it!"
-            )
+        raise ValueError(
+            f"Could not find module {module_name} in `transformers`. If this is a custom class, "
+            f"it should be registered using the relevant `AutoClass.register()` function so that "
+            f"other functions can find it!"
+        )
 
     @property
     def model_input_names(self):
@@ -1469,7 +1468,7 @@ def apply_chat_template(
                 chat_template = self.chat_template[chat_template]
             else:
                 # It's a template string, render it directly
-                chat_template = chat_template
+                pass
 
         is_tokenizers_fast = hasattr(self, "tokenizer") and self.tokenizer.__class__.__name__.endswith("Fast")
 

src/transformers/utils/auto_docstring.py

@@ -1120,9 +1120,8 @@ def get_model_name(obj):
         if file_name.startswith(start) and file_name.endswith(end):
             model_name_lowercase = file_name[len(start) : -len(end)]
             return model_name_lowercase
-    else:
-        print(f"🚨 Something went wrong trying to find the model name in the path: {path}")
-        return "model"
+    print(f"🚨 Something went wrong trying to find the model name in the path: {path}")
+    return "model"
 
 
 def get_placeholders_dict(placeholders: list, model_name: str) -> dict:

src/transformers/utils/fx.py

@@ -815,7 +815,6 @@ def _proxies_to_metas(v):
 
 def create_cache_proxy_factory_fn(orig_cache_cls: type[Cache]) -> Callable[[Node], HFCacheProxy]:
     def cache_proxy_factory_fn(n: Node) -> HFCacheProxy:
-        global _CURRENT_TRACER
         if not isinstance(_CURRENT_TRACER, HFTracer):
             raise RuntimeError("Cannot create HFCacheProxy because there is no HFTracer currently tracing.")
         cache_proxy = HFCacheProxy(n, _CURRENT_TRACER)

src/transformers/utils/generic.py

@@ -746,8 +746,7 @@ def infer_framework(model_class):
             return "pt"
         elif module.startswith("flax") or module.startswith("jax") or name == "FlaxPreTrainedModel":
             return "flax"
-    else:
-        raise TypeError(f"Could not infer framework from class {model_class}.")
+    raise TypeError(f"Could not infer framework from class {model_class}.")
 
 
 def torch_int(x):