fix docstrings (#11221)

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -554,10 +554,6 @@ BLENDERBOT_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_blenderbot._prepare_decoder_inputs`
-            and modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -555,10 +555,6 @@ BLENDERBOT_SMALL_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read
-            :func:`modeling_blenderbot_small._prepare_decoder_inputs` and modify to your needs. See diagram 1 in `the
-            paper <https://arxiv.org/abs/1910.13461>`__ for more information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/fsmt/modeling_fsmt.py

@@ -234,13 +234,20 @@ FSMT_INPUTS_DOCSTRING = r"""
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
-            Provide for translation and summarization training. By default, the model will create this tensor by
-            shifting the input_ids right, following the paper.
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using :class:`~transformers.FSMTTokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            details.
+
+            `What are input IDs? <../glossary.html#input-ids>`__
+
+            FSMT uses the :obj:`eos_token_id` as the starting token for :obj:`decoder_input_ids` generation. If
+            :obj:`past_key_values` is used, optionally only the last :obj:`decoder_input_ids` have to be input (see
+            :obj:`past_key_values`).
         decoder_attention_mask (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
-            also be used by default. If you want to change padding behavior, you should read
-            :func:`modeling_fstm._prepare_fstm_decoder_inputs` and modify. See diagram 1 in the paper for more info on
-            the default strategy
+            also be used by default.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/m2m_100/modeling_m2m_100.py

@@ -589,15 +589,20 @@ M2M_100_INPUTS_DOCSTRING = r"""
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
-            Provide for translation and summarization training. By default, the model will create this tensor by
-            shifting the :obj:`input_ids` to the right, following the paper.
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using :class:`~transformers.M2M100Tokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            details.
+
+            `What are input IDs? <../glossary.html#input-ids>`__
+
+            M2M100 uses the :obj:`eos_token_id` as the starting token for :obj:`decoder_input_ids` generation. If
+            :obj:`past_key_values` is used, optionally only the last :obj:`decoder_input_ids` have to be input (see
+            :obj:`past_key_values`).
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_m2m_100._prepare_decoder_inputs`
-            and modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         encoder_outputs (:obj:`tuple(tuple(torch.FloatTensor)`, `optional`):
             Tuple consists of (:obj:`last_hidden_state`, `optional`: :obj:`hidden_states`, `optional`:
             :obj:`attentions`) :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)`,

src/transformers/models/marian/modeling_marian.py

@@ -567,10 +567,6 @@ MARIAN_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_marian._prepare_decoder_inputs` and
-            modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/mbart/modeling_mbart.py

@@ -575,9 +575,6 @@ MBART_INPUTS_DOCSTRING = r"""
             - 0 for tokens that are **masked**.
 
             `What are attention masks? <../glossary.html#attention-mask>`__
-        decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
-            Provide for translation and summarization training. By default, the model will create this tensor by
-            shifting the :obj:`input_ids` to the right, following the paper.
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Indices of decoder input sequence tokens in the vocabulary.
 
@@ -598,10 +595,6 @@ MBART_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_mbart._prepare_decoder_inputs` and
-            modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/pegasus/modeling_pegasus.py

@@ -566,10 +566,6 @@ PEGASUS_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_pegasus._prepare_decoder_inputs`
-            and modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/pegasus/modeling_tf_pegasus.py

@@ -597,10 +597,6 @@ PEGASUS_INPUTS_DOCSTRING = r"""
             Pegasus uses the :obj:`pad_token_id` as the starting token for :obj:`decoder_input_ids` generation. If
             :obj:`past_key_values` is used, optionally only the last :obj:`decoder_input_ids` have to be input (see
             :obj:`past_key_values`).
-
-            For translation and summarization training, :obj:`decoder_input_ids` should be provided. If no
-            :obj:`decoder_input_ids` is provided, the model will create this tensor by shifting the :obj:`input_ids` to
-            the right for denoising pre-training following the paper.
         decoder_attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             will be made by default and ignore pad tokens. It is not recommended to set this for most use cases.
         head_mask (:obj:`tf.Tensor` of shape :obj:`(encoder_layers, encoder_attention_heads)`, `optional`):

src/transformers/models/prophetnet/modeling_prophetnet.py

@@ -91,7 +91,7 @@ PROPHETNET_INPUTS_DOCSTRING = r"""
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Indices of decoder input sequence tokens in the vocabulary.
 
-            Indices can be obtained using :class:`~transformers.PreTrainedTokenizer`. See
+            Indices can be obtained using :class:`~transformers.ProphetNetTokenizer`. See
             :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
             details.
 
@@ -104,10 +104,6 @@ PROPHETNET_INPUTS_DOCSTRING = r"""
         decoder_attention_mask (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read :func:`modeling_bart._prepare_decoder_inputs` and
-            modify to your needs. See diagram 1 in `the paper <https://arxiv.org/abs/1910.13461>`__ for more
-            information on the default strategy.
         encoder_outputs (:obj:`tuple(tuple(torch.FloatTensor)`, `optional`):
             Tuple consists of (:obj:`last_hidden_state`, `optional`: :obj:`hidden_states`, `optional`:
             :obj:`attentions`) :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)`,

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -610,15 +610,20 @@ SPEECH_TO_TEXT_INPUTS_DOCSTRING = r"""
 
             `What are attention masks? <../glossary.html#attention-mask>`__
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
-            Provide for translation and summarization training. By default, the model will create this tensor by
-            shifting the :obj:`input_ids` to the right, following the paper.
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using :class:`~transformers.SpeechToTextTokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            details.
+
+            `What are input IDs? <../glossary.html#input-ids>`__
+
+            SpeechToText uses the :obj:`eos_token_id` as the starting token for :obj:`decoder_input_ids` generation. If
+            :obj:`past_key_values` is used, optionally only the last :obj:`decoder_input_ids` have to be input (see
+            :obj:`past_key_values`).
         decoder_attention_mask (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.
-
-            If you want to change padding behavior, you should read
-            :func:`modeling_speech_to_text._prepare_decoder_inputs` and modify to your needs. See diagram 1 in `the
-            paper <https://arxiv.org/abs/1910.13461>`__ for more information on the default strategy.
         head_mask (:obj:`torch.Tensor` of shape :obj:`(num_layers, num_heads)`, `optional`):
             Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in ``[0, 1]``:
 

src/transformers/models/t5/modeling_t5.py

@@ -1057,7 +1057,7 @@ T5_INPUTS_DOCSTRING = r"""
         decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Indices of decoder input sequence tokens in the vocabulary.
 
-            Indices can be obtained using :class:`~transformers.BartTokenizer`. See
+            Indices can be obtained using :class:`~transformers.T5Tokenizer`. See
             :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
             details.
 
@@ -1068,8 +1068,7 @@ T5_INPUTS_DOCSTRING = r"""
             :obj:`past_key_values`).
 
             To know more on how to prepare :obj:`decoder_input_ids` for pretraining take a look at `T5 Training
-            <./t5.html#training>`__. If :obj:`decoder_input_ids` and :obj:`decoder_inputs_embeds` are both unset,
-            :obj:`decoder_input_ids` takes the value of :obj:`input_ids`.
+            <./t5.html#training>`__.
         decoder_attention_mask (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
             Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
             also be used by default.

src/transformers/models/t5/modeling_tf_t5.py

@@ -954,8 +954,7 @@ T5_INPUTS_DOCSTRING = r"""
             :obj:`decoder_input_ids` have to be input (see :obj:`past_key_values`).
 
             To know more on how to prepare :obj:`decoder_input_ids` for pretraining take a look at `T5 Training
-            <./t5.html#training>`__. If :obj:`decoder_input_ids` and :obj:`decoder_inputs_embeds` are both unset,
-            :obj:`decoder_input_ids` takes the value of :obj:`input_ids`.
+            <./t5.html#training>`__.
         attention_mask (:obj:`tf.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
             Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``: