Fix model templates and use less than 119 chars (#9684)

* Fix model templates and use less than 119 chars

* Missing new line

src/transformers/models/bart/modeling_bart.py

@@ -997,7 +997,8 @@ class BartDecoder(BartPretrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/bert/modeling_bert.py

@@ -544,7 +544,8 @@ class BertEncoder(nn.Module):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -959,7 +959,8 @@ class BlenderbotDecoder(BlenderbotPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -959,7 +959,8 @@ class BlenderbotSmallDecoder(BlenderbotSmallPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/electra/modeling_electra.py

@@ -541,7 +541,8 @@ class ElectraEncoder(nn.Module):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/gpt2/modeling_gpt2.py

@@ -721,7 +721,8 @@ class GPT2Model(GPT2PreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/layoutlm/modeling_layoutlm.py

@@ -471,7 +471,8 @@ class LayoutLMEncoder(nn.Module):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/led/modeling_led.py

@@ -1924,7 +1924,8 @@ class LEDDecoder(LEDPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/marian/modeling_marian.py

@@ -962,7 +962,8 @@ class MarianDecoder(MarianPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/mbart/modeling_mbart.py

@@ -1006,7 +1006,8 @@ class MBartDecoder(MBartPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/pegasus/modeling_pegasus.py

@@ -970,7 +970,8 @@ class PegasusDecoder(PegasusPreTrainedModel):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/roberta/modeling_roberta.py

@@ -484,7 +484,8 @@ class RobertaEncoder(nn.Module):
 
                 if use_cache:
                     logger.warn(
-                        "`use_cache = True` is incompatible with `config.gradient_checkpointing = True`. Setting `use_cache = False`..."
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
                     )
                     use_cache = False
 

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

@@ -526,8 +526,16 @@ class {{cookiecutter.camelcase_modelname}}Encoder(nn.Module):
 
             layer_head_mask = head_mask[i] if head_mask is not None else None
             past_key_value = past_key_values[i] if past_key_values is not None else None
+
             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`..."
+                    )
+                    use_cache = False
+
                 def create_custom_forward(module):
                     def custom_forward(*inputs):
                         return module(*inputs, past_key_value, output_attentions)