introduce `logger.warning_once` and use it for grad checkpointing code (#21804)

* logger.warning_once

* style

src/transformers/models/altclip/modeling_altclip.py

@@ -638,7 +638,7 @@ class AltRobertaEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bart/modeling_bart.py

@@ -1085,7 +1085,7 @@ class BartDecoder(BartPretrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bert/modeling_bert.py

@@ -585,7 +585,7 @@ class BertEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bert_generation/modeling_bert_generation.py

@@ -395,7 +395,7 @@ class BertEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/big_bird/modeling_big_bird.py

@@ -1606,7 +1606,7 @@ class BigBirdEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

@@ -2265,7 +2265,7 @@ class BigBirdPegasusDecoder(BigBirdPegasusPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/biogpt/modeling_biogpt.py

@@ -557,7 +557,7 @@ class BioGptModel(BioGptPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -1016,7 +1016,7 @@ class BlenderbotDecoder(BlenderbotPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -1012,7 +1012,7 @@ class BlenderbotSmallDecoder(BlenderbotSmallPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bloom/modeling_bloom.py

@@ -757,7 +757,7 @@ class BloomModel(BloomPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/bridgetower/modeling_bridgetower.py

@@ -769,7 +769,7 @@ class BridgeTowerTextEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/camembert/modeling_camembert.py

@@ -516,7 +516,7 @@ class CamembertEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/chinese_clip/modeling_chinese_clip.py

@@ -901,7 +901,7 @@ class ChineseCLIPTextEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/clap/modeling_clap.py

@@ -1588,7 +1588,7 @@ class ClapTextEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/codegen/modeling_codegen.py

@@ -548,7 +548,7 @@ class CodeGenModel(CodeGenPreTrainedModel):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
                         "`use_cache=False`..."
                     )

src/transformers/models/data2vec/modeling_data2vec_text.py

@@ -502,7 +502,7 @@ class Data2VecTextEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/decision_transformer/modeling_decision_transformer.py

@@ -609,7 +609,7 @@ class DecisionTransformerGPT2Model(DecisionTransformerGPT2PreTrainedModel):
 
         if self.gradient_checkpointing and self.training:
             if use_cache:
-                logger.warning(
+                logger.warning_once(
                     "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                 )
                 use_cache = False

src/transformers/models/electra/modeling_electra.py

@@ -563,7 +563,7 @@ class ElectraEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/ernie/modeling_ernie.py

@@ -498,7 +498,7 @@ class ErnieEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False

src/transformers/models/esm/modeling_esm.py

@@ -597,7 +597,7 @@ class EsmEncoder(nn.Module):
 
             if self.gradient_checkpointing and self.training:
                 if use_cache:
-                    logger.warning(
+                    logger.warning_once(
                         "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
                         "`use_cache=False`..."
                     )