Fix doc examples: modify config before super().__init__ (#14697)

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

src/transformers/models/bart/modeling_bart.py

@@ -1662,10 +1662,10 @@ class BartDecoderWrapper(BartPretrainedModel):
 
 class BartForCausalLM(BartPretrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = BartDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

@@ -2865,10 +2865,10 @@ class BigBirdPegasusDecoderWrapper(BigBirdPegasusPreTrainedModel):
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->BigBirdPegasus, 'facebook/bart-large'->"google/bigbird-pegasus-large-arxiv"
 class BigBirdPegasusForCausalLM(BigBirdPegasusPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = BigBirdPegasusDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -1400,10 +1400,10 @@ class BlenderbotDecoderWrapper(BlenderbotPreTrainedModel):
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->Blenderbot
 class BlenderbotForCausalLM(BlenderbotPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = BlenderbotDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -1374,10 +1374,10 @@ class BlenderbotSmallDecoderWrapper(BlenderbotSmallPreTrainedModel):
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->BlenderbotSmall
 class BlenderbotSmallForCausalLM(BlenderbotSmallPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = BlenderbotSmallDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/marian/modeling_marian.py

@@ -1397,10 +1397,10 @@ class MarianDecoderWrapper(MarianPreTrainedModel):
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->Marian
 class MarianForCausalLM(MarianPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = MarianDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/mbart/modeling_mbart.py

@@ -1665,10 +1665,10 @@ class MBartDecoderWrapper(MBartPreTrainedModel):
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->MBart
 class MBartForCausalLM(MBartPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = MBartDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/pegasus/modeling_pegasus.py

@@ -1486,10 +1486,10 @@ class PegasusDecoderWrapper(PegasusPreTrainedModel):
 
 class PegasusForCausalLM(PegasusPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = PegasusDecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/speech_to_text_2/modeling_speech_to_text_2.py

@@ -744,10 +744,10 @@ class Speech2Text2DecoderWrapper(Speech2Text2PreTrainedModel):
 )
 class Speech2Text2ForCausalLM(Speech2Text2PreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = Speech2Text2DecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

src/transformers/models/trocr/modeling_trocr.py

@@ -777,10 +777,10 @@ class TrOCRDecoderWrapper(TrOCRPreTrainedModel):
 )
 class TrOCRForCausalLM(TrOCRPreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = TrOCRDecoderWrapper(config)
 
         self.output_projection = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

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

@@ -3173,10 +3173,10 @@ class {{cookiecutter.camelcase_modelname}}DecoderWrapper({{cookiecutter.camelcas
 # Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->{{cookiecutter.camelcase_modelname}}
 class {{cookiecutter.camelcase_modelname}}ForCausalLM({{cookiecutter.camelcase_modelname}}PreTrainedModel):
     def __init__(self, config):
-        super().__init__(config)
         config = copy.deepcopy(config)
         config.is_decoder = True
         config.is_encoder_decoder = False
+        super().__init__(config)
         self.model = {{cookiecutter.camelcase_modelname}}DecoderWrapper(config)
 
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)