[TFBart] Split TF-Bart (#9497)

* make templates ready

* make add_new_model_command_ready

* finish tf bart

* prepare tf mbart

* finish tf bart

* add tf mbart

* add marian

* prep pegasus

* add tf pegasus

* push blenderbot tf

* add blenderbot

* add blenderbot small

* clean-up

* make fix copy

* define blend bot tok

* fix

* up

* make style

* add to docs

* add copy statements

* overwrite changes

* improve

* fix docs

* finish

* fix last slow test

* fix missing git conflict line

* fix blenderbot

* up

* fix blenderbot small

* load changes

* finish copied from

* upload fix

src/transformers/models/marian/configuration_marian.py

@@ -159,17 +159,6 @@ class MarianConfig(PretrainedConfig):
         self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
-        # IMPORTANT
-        # DELETE ALL OF THE FOLLOWING LINES AS SOON AS TF IS READY
-        self.extra_pos_embeddings = 0
-        self.normalize_before = False
-        self.add_final_layer_norm = False
-        self.do_blenderbot_90_layernorm = False
-        self.normalize_embedding = False
-        self.static_position_embeddings = True
-        self.add_bias_logits = False
-        self.force_bos_token_to_be_generated = False
-
     @property
     def num_attention_heads(self) -> int:
         return self.encoder_attention_heads

src/transformers/models/mbart/__init__.py

@@ -47,7 +47,7 @@ if is_torch_available():
     ]
 
 if is_tf_available():
-    _import_structure["modeling_tf_mbart"] = ["TFMBartForConditionalGeneration"]
+    _import_structure["modeling_tf_mbart"] = ["TFMBartForConditionalGeneration", "TFMBartModel"]
 
 
 if TYPE_CHECKING:
@@ -70,7 +70,7 @@ if TYPE_CHECKING:
         )
 
     if is_tf_available():
-        from .modeling_tf_mbart import TFMBartForConditionalGeneration
+        from .modeling_tf_mbart import TFMBartForConditionalGeneration, TFMBartModel
 
 else:
     import importlib

src/transformers/models/mbart/configuration_mbart.py

@@ -159,17 +159,6 @@ class MBartConfig(PretrainedConfig):
         self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
-        # IMPORTANT
-        # DELETE ALL OF THE FOLLOWING LINES AS SOON AS TF IS READY
-        self.extra_pos_embeddings = 2
-        self.normalize_before = True
-        self.add_final_layer_norm = True
-        self.do_blenderbot_90_layernorm = False
-        self.normalize_embedding = True
-        self.static_position_embeddings = False
-        self.add_bias_logits = False
-        self.force_bos_token_to_be_generated = False
-
     @property
     def num_attention_heads(self) -> int:
         return self.encoder_attention_heads

src/transformers/models/pegasus/__init__.py

@@ -45,7 +45,7 @@ if is_torch_available():
     ]
 
 if is_tf_available():
-    _import_structure["modeling_tf_pegasus"] = ["TFPegasusForConditionalGeneration"]
+    _import_structure["modeling_tf_pegasus"] = ["TFPegasusForConditionalGeneration", "TFPegasusModel"]
 
 
 if TYPE_CHECKING:
@@ -66,7 +66,7 @@ if TYPE_CHECKING:
         )
 
     if is_tf_available():
-        from .modeling_tf_pegasus import TFPegasusForConditionalGeneration
+        from .modeling_tf_pegasus import TFPegasusForConditionalGeneration, TFPegasusModel
 
 else:
     import importlib

src/transformers/models/pegasus/configuration_pegasus.py

@@ -159,17 +159,6 @@ class PegasusConfig(PretrainedConfig):
         self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
-        # IMPORTANT
-        # DELETE ALL OF THE FOLLOWING LINES AS SOON AS TF IS READY
-        self.extra_pos_embeddings = 0
-        self.normalize_before = True
-        self.add_final_layer_norm = True
-        self.do_blenderbot_90_layernorm = False
-        self.normalize_embedding = False
-        self.static_position_embeddings = True
-        self.add_bias_logits = False
-        self.force_bos_token_to_be_generated = False
-
     @property
     def num_attention_heads(self) -> int:
         return self.encoder_attention_heads

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

 # coding=utf-8
-# Copyright {{cookiecutter.authors}} and The HuggingFace Inc. team. All rights reserved.
+# Copyright 2021 {{cookiecutter.authors}} The HuggingFace Inc. team. All rights reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.