fix: The 'warn' method is deprecated (#11105)

* The 'warn' method is deprecated

* fix test

src/transformers/models/roberta/modeling_roberta.py

@@ -484,7 +484,7 @@ class RobertaEncoder(nn.Module):
             if getattr(self.config, "gradient_checkpointing", False) and self.training:
 
                 if use_cache:
-                    logger.warn(
+                    logger.warning(
                         "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
                         "`use_cache=False`..."
                     )

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -1015,7 +1015,7 @@ class Speech2TextDecoder(Speech2TextPreTrainedModel):
             if getattr(self.config, "gradient_checkpointing", False) and self.training:
 
                 if use_cache:
-                    logger.warn(
+                    logger.warning(
                         "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
                     )
                     use_cache = False

src/transformers/models/wav2vec2/convert_wav2vec2_original_pytorch_checkpoint_to_pytorch.py

@@ -111,7 +111,7 @@ def recursively_load_weights(fairseq_model, hf_model, is_finetuned):
         if not is_used:
             unused_weights.append(name)
 
-    logger.warn(f"Unused weights: {unused_weights}")
+    logger.warning(f"Unused weights: {unused_weights}")
 
 
 def load_conv_layer(full_name, value, feature_extractor, unused_weights, use_group_norm):

src/transformers/models/xlm/modeling_tf_xlm.py

@@ -1140,7 +1140,7 @@ class TFXLMForMultipleChoice(TFXLMPreTrainedModel, TFMultipleChoiceLoss):
         )
 
         if inputs["lengths"] is not None:
-            logger.warn(
+            logger.warning(
                 "The `lengths` parameter cannot be used with the XLM multiple choice models. Please use the "
                 "attention mask instead.",
             )

src/transformers/models/xlm/modeling_xlm.py

@@ -1232,7 +1232,7 @@ class XLMForMultipleChoice(XLMPreTrainedModel):
         )
 
         if lengths is not None:
-            logger.warn(
+            logger.warning(
                 "The `lengths` parameter cannot be used with the XLM multiple choice models. Please use the "
                 "attention mask instead."
             )

src/transformers/pipelines/zero_shot_classification.py

@@ -142,7 +142,7 @@ class ZeroShotClassificationPipeline(Pipeline):
         """
         if "multi_class" in kwargs and kwargs["multi_class"] is not None:
             multi_label = kwargs.pop("multi_class")
-            logger.warn(
+            logger.warning(
                 "The `multi_class` argument has been deprecated and renamed to `multi_label`. "
                 "`multi_class` will be removed in a future version of Transformers."
             )

src/transformers/trainer_callback.py

@@ -289,7 +289,7 @@ class CallbackHandler(TrainerCallback):
         self.eval_dataloader = None
 
         if not any(isinstance(cb, DefaultFlowCallback) for cb in self.callbacks):
-            logger.warn(
+            logger.warning(
                 "The Trainer will not work properly if you don't have a `DefaultFlowCallback` in its callbacks. You\n"
                 + "should add one before training with `trainer.add_callback(DefaultFlowCallback). The current list of"
                 + "callbacks is\n:"
@@ -300,7 +300,7 @@ class CallbackHandler(TrainerCallback):
         cb = callback() if isinstance(callback, type) else callback
         cb_class = callback if isinstance(callback, type) else callback.__class__
         if cb_class in [c.__class__ for c in self.callbacks]:
-            logger.warn(
+            logger.warning(
                 f"You are adding a {cb_class} to the callbacks of this Trainer, but there is already one. The current"
                 + "list of callbacks is\n:"
                 + self.callback_list

src/transformers/trainer_pt_utils.py

@@ -391,7 +391,7 @@ class DistributedTensorGatherer:
         if self._storage is None:
             return
         if self._offsets[0] != self.process_length:
-            logger.warn("Not all data has been set. Are you sure you passed all values?")
+            logger.warning("Not all data has been set. Are you sure you passed all values?")
         return nested_truncate(self._storage, self.num_samples)
 
 
@@ -589,7 +589,7 @@ def _get_learning_rate(self):
             last_lr = self.lr_scheduler.get_last_lr()[0]
         except AssertionError as e:
             if "need to call step" in str(e):
-                logger.warn("tried to get lr value before scheduler/optimizer started stepping, returning lr=0")
+                logger.warning("tried to get lr value before scheduler/optimizer started stepping, returning lr=0")
                 last_lr = 0
             else:
                 raise

templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_{{cookiecutter.lowercase_modelname}}.py

@@ -531,7 +531,7 @@ class {{cookiecutter.camelcase_modelname}}Encoder(nn.Module):
             if getattr(self.config, "gradient_checkpointing", False) and self.training:
 
                 if use_cache:
-                    logger.warn(
+                    logger.warning(
                         "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
                         "`use_cache=False`..."
                     )
@@ -2512,7 +2512,7 @@ class {{cookiecutter.camelcase_modelname}}Decoder({{cookiecutter.camelcase_model
             if getattr(self.config, "gradient_checkpointing", False) and self.training:
 
                 if use_cache:
-                    logger.warn("`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`...")
+                    logger.warning("`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`...")
                     use_cache = False
 
                 def create_custom_forward(module):

tests/sagemaker/scripts/pytorch/run_glue_model_parallelism.py

@@ -353,7 +353,7 @@ def main():
         if list(sorted(label_name_to_id.keys())) == list(sorted(label_list)):
             label_to_id = {i: int(label_name_to_id[label_list[i]]) for i in range(num_labels)}
         else:
-            logger.warn(
+            logger.warning(
                 "Your model seems to have been trained with labels, but they don't match the dataset: ",
                 f"model labels: {list(sorted(label_name_to_id.keys()))}, dataset labels: {list(sorted(label_list))}."
                 "\nIgnoring the model labels as a result.",
@@ -362,7 +362,7 @@ def main():
         label_to_id = {v: i for i, v in enumerate(label_list)}
 
     if data_args.max_seq_length > tokenizer.model_max_length:
-        logger.warn(
+        logger.warning(
             f"The max_seq_length passed ({data_args.max_seq_length}) is larger than the maximum length for the"
             f"model ({tokenizer.model_max_length}). Using max_seq_length={tokenizer.model_max_length}."
         )

tests/test_logging.py

@@ -51,7 +51,7 @@ class HfArgumentParserTest(unittest.TestCase):
         # should be able to log warnings (if default settings weren't overridden by `pytest --log-level-all`)
         if level_origin <= logging.WARNING:
             with CaptureLogger(logger) as cl:
-                logger.warn(msg)
+                logger.warning(msg)
             self.assertEqual(cl.out, msg + "\n")
 
         # this is setting the level for all of `transformers.*` loggers
@@ -59,7 +59,7 @@ class HfArgumentParserTest(unittest.TestCase):
 
         # should not be able to log warnings
         with CaptureLogger(logger) as cl:
-            logger.warn(msg)
+            logger.warning(msg)
         self.assertEqual(cl.out, "")
 
         # should be able to log warnings again

tests/test_trainer_callback.py

@@ -234,7 +234,7 @@ class TrainerCallbackTest(unittest.TestCase):
         self.assertEqual(events, self.get_expected_events(trainer))
 
         # warning should be emitted for duplicated callbacks
-        with unittest.mock.patch("transformers.trainer_callback.logger.warn") as warn_mock:
+        with unittest.mock.patch("transformers.trainer_callback.logger.warning") as warn_mock:
             trainer = self.get_trainer(
                 callbacks=[MyTestTrainerCallback, MyTestTrainerCallback],
             )