DeBERTa

Overview

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen It is based on Google’s BERT model released in 2018 and Facebook’s RoBERTa model released in 2019.

It builds on RoBERTa with disentangled attention and enhanced mask decoder training with half of the data used in RoBERTa.

The abstract from the paper is the following:

Recent progress in pre-trained neural language models has significantly improved the performance of many natural language processing (NLP) tasks. In this paper we propose a new model architecture DeBERTa (Decoding-enhanced BERT with disentangled attention) that improves the BERT and RoBERTa models using two novel techniques. The first is the disentangled attention mechanism, where each word is represented using two vectors that encode its content and position, respectively, and the attention weights among words are computed using disentangled matrices on their contents and relative positions. Second, an enhanced mask decoder is used to replace the output softmax layer to predict the masked tokens for model pretraining. We show that these two techniques significantly improve the efficiency of model pretraining and performance of downstream tasks. Compared to RoBERTa-Large, a DeBERTa model trained on half of the training data performs consistently better on a wide range of NLP tasks, achieving improvements on MNLI by +0.9% (90.2% vs. 91.1%), on SQuAD v2.0 by +2.3% (88.4% vs. 90.7%) and RACE by +3.6% (83.2% vs. 86.8%). The DeBERTa code and pre-trained models will be made publicly available at https://github.com/microsoft/DeBERTa.

This model was contributed by DeBERTa. This model TF 2.0 implementation was contributed by kamalkraj . The original code can be found here.

DebertaConfig

class transformers.DebertaConfig(vocab_size=50265, hidden_size=768, num_hidden_layers=12, num_attention_heads=12, intermediate_size=3072, hidden_act='gelu', hidden_dropout_prob=0.1, attention_probs_dropout_prob=0.1, max_position_embeddings=512, type_vocab_size=0, initializer_range=0.02, layer_norm_eps=1e-07, relative_attention=False, max_relative_positions=- 1, pad_token_id=0, position_biased_input=True, pos_att_type=None, pooler_dropout=0, pooler_hidden_act='gelu', **kwargs)[source]

This is the configuration class to store the configuration of a DebertaModel or a TFDebertaModel. It is used to instantiate a DeBERTa model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the DeBERTa microsoft/deberta-base architecture.

Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.

Parameters
  • vocab_size (int, optional, defaults to 30522) – Vocabulary size of the DeBERTa model. Defines the number of different tokens that can be represented by the inputs_ids passed when calling DebertaModel or TFDebertaModel.

  • hidden_size (int, optional, defaults to 768) – Dimensionality of the encoder layers and the pooler layer.

  • num_hidden_layers (int, optional, defaults to 12) – Number of hidden layers in the Transformer encoder.

  • num_attention_heads (int, optional, defaults to 12) – Number of attention heads for each attention layer in the Transformer encoder.

  • intermediate_size (int, optional, defaults to 3072) – Dimensionality of the “intermediate” (often named feed-forward) layer in the Transformer encoder.

  • hidden_act (str or Callable, optional, defaults to "gelu") – The non-linear activation function (function or string) in the encoder and pooler. If string, "gelu", "relu", "silu", "gelu", "tanh", "gelu_fast", "mish", "linear", "sigmoid" and "gelu_new" are supported.

  • hidden_dropout_prob (float, optional, defaults to 0.1) – The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.

  • attention_probs_dropout_prob (float, optional, defaults to 0.1) – The dropout ratio for the attention probabilities.

  • max_position_embeddings (int, optional, defaults to 512) – The maximum sequence length that this model might ever be used with. Typically set this to something large just in case (e.g., 512 or 1024 or 2048).

  • type_vocab_size (int, optional, defaults to 2) – The vocabulary size of the token_type_ids passed when calling DebertaModel or TFDebertaModel.

  • initializer_range (float, optional, defaults to 0.02) – The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

  • layer_norm_eps (float, optional, defaults to 1e-12) – The epsilon used by the layer normalization layers.

  • relative_attention (bool, optional, defaults to False) – Whether use relative position encoding.

  • max_relative_positions (int, optional, defaults to 1) – The range of relative positions [-max_position_embeddings, max_position_embeddings]. Use the same value as max_position_embeddings.

  • pad_token_id (int, optional, defaults to 0) – The value used to pad input_ids.

  • position_biased_input (bool, optional, defaults to True) – Whether add absolute position embedding to content embedding.

  • pos_att_type (List[str], optional) – The type of relative position attention, it can be a combination of ["p2c", "c2p", "p2p"], e.g. ["p2c"], ["p2c", "c2p"], ["p2c", "c2p", 'p2p"].

  • layer_norm_eps – The epsilon used by the layer normalization layers.

DebertaTokenizer

class transformers.DebertaTokenizer(vocab_file, merges_file, errors='replace', bos_token='[CLS]', eos_token='[SEP]', sep_token='[SEP]', cls_token='[CLS]', unk_token='[UNK]', pad_token='[PAD]', mask_token='[MASK]', add_prefix_space=False, **kwargs)[source]

Constructs a DeBERTa tokenizer, which runs end-to-end tokenization: punctuation splitting + wordpiece

Parameters
  • vocab_file (str) – File containing the vocabulary.

  • do_lower_case (bool, optional, defaults to True) – Whether or not to lowercase the input when tokenizing.

  • unk_token (str, optional, defaults to "[UNK]") – The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead.

  • sep_token (str, optional, defaults to "[SEP]") – The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for sequence classification or for a text and a question for question answering. It is also used as the last token of a sequence built with special tokens.

  • pad_token (str, optional, defaults to "[PAD]") – The token used for padding, for example when batching sequences of different lengths.

  • cls_token (str, optional, defaults to "[CLS]") – The classifier token which is used when doing sequence classification (classification of the whole sequence instead of per-token classification). It is the first token of the sequence when built with special tokens.

  • mask_token (str, optional, defaults to "[MASK]") – The token used for masking values. This is the token used when training this model with masked language modeling. This is the token which the model will try to predict.

build_inputs_with_special_tokens(token_ids_0: List[int], token_ids_1: Optional[List[int]] = None) → List[int][source]

Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and adding special tokens. A DeBERTa sequence has the following format:

  • single sequence: [CLS] X [SEP]

  • pair of sequences: [CLS] A [SEP] B [SEP]

Parameters
  • token_ids_0 (List[int]) – List of IDs to which the special tokens will be added.

  • token_ids_1 (List[int], optional) – Optional second list of IDs for sequence pairs.

Returns

List of input IDs with the appropriate special tokens.

Return type

List[int]

create_token_type_ids_from_sequences(token_ids_0: List[int], token_ids_1: Optional[List[int]] = None) → List[int][source]

Create a mask from the two sequences passed to be used in a sequence-pair classification task. A DeBERTa sequence pair mask has the following format:

0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence    | second sequence |

If token_ids_1 is None, this method only returns the first portion of the mask (0s).

Parameters
  • token_ids_0 (List[int]) – List of IDs.

  • token_ids_1 (List[int], optional) – Optional second list of IDs for sequence pairs.

Returns

List of token type IDs according to the given sequence(s).

Return type

List[int]

get_special_tokens_mask(token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False) → List[int][source]

Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding special tokens using the tokenizer prepare_for_model or encode_plus methods.

Parameters
  • token_ids_0 (List[int]) – List of IDs.

  • token_ids_1 (List[int], optional) – Optional second list of IDs for sequence pairs.

  • already_has_special_tokens (bool, optional, defaults to False) – Whether or not the token list is already formatted with special tokens for the model.

Returns

A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.

Return type

List[int]

save_vocabulary(save_directory: str, filename_prefix: Optional[str] = None) → Tuple[str]

Save only the vocabulary of the tokenizer (vocabulary + added tokens).

This method won’t save the configuration and special token mappings of the tokenizer. Use _save_pretrained() to save the whole state of the tokenizer.

Parameters
  • save_directory (str) – The directory in which to save the vocabulary.

  • filename_prefix (str, optional) – An optional prefix to add to the named of the saved files.

Returns

Paths to the files saved.

Return type

Tuple(str)

DebertaTokenizerFast

class transformers.DebertaTokenizerFast(vocab_file=None, merges_file=None, tokenizer_file=None, errors='replace', bos_token='[CLS]', eos_token='[SEP]', sep_token='[SEP]', cls_token='[CLS]', unk_token='[UNK]', pad_token='[PAD]', mask_token='[MASK]', add_prefix_space=False, **kwargs)[source]

Constructs a “fast” DeBERTa tokenizer, which runs end-to-end tokenization: punctuation splitting + wordpiece. It is backed by HuggingFace’s tokenizers library.

Parameters
  • vocab_file (str) – File containing the vocabulary.

  • do_lower_case (bool, optional, defaults to True) – Whether or not to lowercase the input when tokenizing.

  • unk_token (str, optional, defaults to "[UNK]") – The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead.

  • sep_token (str, optional, defaults to "[SEP]") – The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for sequence classification or for a text and a question for question answering. It is also used as the last token of a sequence built with special tokens.

  • pad_token (str, optional, defaults to "[PAD]") – The token used for padding, for example when batching sequences of different lengths.

  • cls_token (str, optional, defaults to "[CLS]") – The classifier token which is used when doing sequence classification (classification of the whole sequence instead of per-token classification). It is the first token of the sequence when built with special tokens.

  • mask_token (str, optional, defaults to "[MASK]") – The token used for masking values. This is the token used when training this model with masked language modeling. This is the token which the model will try to predict.

build_inputs_with_special_tokens(token_ids_0: List[int], token_ids_1: Optional[List[int]] = None) → List[int][source]

Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and adding special tokens. A DeBERTa sequence has the following format:

  • single sequence: [CLS] X [SEP]

  • pair of sequences: [CLS] A [SEP] B [SEP]

Parameters
  • token_ids_0 (List[int]) – List of IDs to which the special tokens will be added.

  • token_ids_1 (List[int], optional) – Optional second list of IDs for sequence pairs.

Returns

List of input IDs with the appropriate special tokens.

Return type

List[int]

create_token_type_ids_from_sequences(token_ids_0: List[int], token_ids_1: Optional[List[int]] = None) → List[int][source]

Create a mask from the two sequences passed to be used in a sequence-pair classification task. A DeBERTa sequence pair mask has the following format:

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
| first sequence    | second sequence |

If token_ids_1 is None, this method only returns the first portion of the mask (0s).

Parameters
  • token_ids_0 (List[int]) – List of IDs.

  • token_ids_1 (List[int], optional) – Optional second list of IDs for sequence pairs.

Returns

List of token type IDs according to the given sequence(s).

Return type

List[int]

DebertaModel

class transformers.DebertaModel(config)[source]

The bare DeBERTa Model transformer outputting raw hidden-states without any specific head on top. The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.```

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, inputs_embeds=None, output_attentions=None, output_hidden_states=None, return_dict=None)[source]

The DebertaModel forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

Returns

A SequenceClassifierOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) – Classification (or regression if config.num_labels==1) loss.

  • logits (torch.FloatTensor of shape (batch_size, config.num_labels)) – Classification (or regression if config.num_labels==1) scores (before SoftMax).

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of torch.FloatTensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

SequenceClassifierOutput or tuple(torch.FloatTensor)

Example:

>>> from transformers import DebertaTokenizer, DebertaModel
>>> import torch

>>> tokenizer = DebertaTokenizer.from_pretrained('microsoft/deberta-base')
>>> model = DebertaModel.from_pretrained('microsoft/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)

>>> last_hidden_states = outputs.last_hidden_state

DebertaPreTrainedModel

class transformers.DebertaPreTrainedModel(config: transformers.configuration_utils.PretrainedConfig, *inputs, **kwargs)[source]

An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models.

config_class

alias of transformers.models.deberta.configuration_deberta.DebertaConfig

DebertaForMaskedLM

class transformers.DebertaForMaskedLM(config)[source]

DeBERTa Model with a language modeling head on top. The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.```

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None)[source]

The DebertaForMaskedLM forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (torch.LongTensor of shape (batch_size, sequence_length), optional) – Labels for computing the masked language modeling loss. Indices should be in [-100, 0, ..., config.vocab_size] (see input_ids docstring) Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size]

Returns

A MaskedLMOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) – Masked language modeling (MLM) loss.

  • logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) – Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of torch.FloatTensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

MaskedLMOutput or tuple(torch.FloatTensor)

Example:

>>> from transformers import DebertaTokenizer, DebertaForMaskedLM
>>> import torch

>>> tokenizer = DebertaTokenizer.from_pretrained('microsoft/deberta-base')
>>> model = DebertaForMaskedLM.from_pretrained('microsoft/deberta-base')

>>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="pt")
>>> labels = tokenizer("The capital of France is Paris.", return_tensors="pt")["input_ids"]

>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> logits = outputs.logits

DebertaForSequenceClassification

class transformers.DebertaForSequenceClassification(config)[source]

DeBERTa Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks.

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.```

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None)[source]

The DebertaForSequenceClassification forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (torch.LongTensor of shape (batch_size,), optional) – Labels for computing the sequence classification/regression loss. Indices should be in [0, ..., config.num_labels - 1]. If config.num_labels == 1 a regression loss is computed (Mean-Square loss), If config.num_labels > 1 a classification loss is computed (Cross-Entropy).

Returns

A SequenceClassifierOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) – Classification (or regression if config.num_labels==1) loss.

  • logits (torch.FloatTensor of shape (batch_size, config.num_labels)) – Classification (or regression if config.num_labels==1) scores (before SoftMax).

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of torch.FloatTensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

SequenceClassifierOutput or tuple(torch.FloatTensor)

Example:

>>> from transformers import DebertaTokenizer, DebertaForSequenceClassification
>>> import torch

>>> tokenizer = DebertaTokenizer.from_pretrained('microsoft/deberta-base')
>>> model = DebertaForSequenceClassification.from_pretrained('microsoft/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> labels = torch.tensor([1]).unsqueeze(0)  # Batch size 1
>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> logits = outputs.logits

DebertaForTokenClassification

class transformers.DebertaForTokenClassification(config)[source]

DeBERTa Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks.

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.```

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None)[source]

The DebertaForTokenClassification forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (torch.LongTensor of shape (batch_size, sequence_length), optional) – Labels for computing the token classification loss. Indices should be in [0, ..., config.num_labels - 1].

Returns

A TokenClassifierOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) – Classification loss.

  • logits (torch.FloatTensor of shape (batch_size, sequence_length, config.num_labels)) – Classification scores (before SoftMax).

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of torch.FloatTensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TokenClassifierOutput or tuple(torch.FloatTensor)

Example:

>>> from transformers import DebertaTokenizer, DebertaForTokenClassification
>>> import torch

>>> tokenizer = DebertaTokenizer.from_pretrained('microsoft/deberta-base')
>>> model = DebertaForTokenClassification.from_pretrained('microsoft/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> labels = torch.tensor([1] * inputs["input_ids"].size(1)).unsqueeze(0)  # Batch size 1

>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> logits = outputs.logits

DebertaForQuestionAnswering

class transformers.DebertaForQuestionAnswering(config)[source]

DeBERTa Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of the hidden-states output to compute span start logits and span end logits).

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.```

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, inputs_embeds=None, start_positions=None, end_positions=None, output_attentions=None, output_hidden_states=None, return_dict=None)[source]

The DebertaForQuestionAnswering forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (torch.FloatTensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • start_positions (torch.LongTensor of shape (batch_size,), optional) – Labels for position (index) of the start of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (sequence_length). Position outside of the sequence are not taken into account for computing the loss.

  • end_positions (torch.LongTensor of shape (batch_size,), optional) – Labels for position (index) of the end of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (sequence_length). Position outside of the sequence are not taken into account for computing the loss.

Returns

A QuestionAnsweringModelOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) – Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.

  • start_logits (torch.FloatTensor of shape (batch_size, sequence_length)) – Span-start scores (before SoftMax).

  • end_logits (torch.FloatTensor of shape (batch_size, sequence_length)) – Span-end scores (before SoftMax).

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of torch.FloatTensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

QuestionAnsweringModelOutput or tuple(torch.FloatTensor)

Example:

>>> from transformers import DebertaTokenizer, DebertaForQuestionAnswering
>>> import torch

>>> tokenizer = DebertaTokenizer.from_pretrained('microsoft/deberta-base')
>>> model = DebertaForQuestionAnswering.from_pretrained('microsoft/deberta-base')

>>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
>>> inputs = tokenizer(question, text, return_tensors='pt')
>>> start_positions = torch.tensor([1])
>>> end_positions = torch.tensor([3])

>>> outputs = model(**inputs, start_positions=start_positions, end_positions=end_positions)
>>> loss = outputs.loss
>>> start_scores = outputs.start_logits
>>> end_scores = outputs.end_logits

TFDebertaModel

class transformers.TFDebertaModel(*args, **kwargs)[source]

The bare DeBERTa Model transformer outputting raw hidden-states without any specific head on top. The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

Note

TF 2.0 models accepts two formats as inputs:

  • having all inputs as keyword arguments (like PyTorch models), or

  • having all inputs as a list, tuple or dict in the first positional arguments.

This second option is useful when using tf.keras.Model.fit() method which currently requires having all the tensors in the first argument of the model call function: model(inputs).

If you choose this second option, there are three possibilities you can use to gather all the input Tensors in the first positional argument :

  • a single Tensor with input_ids only and nothing else: model(inputs_ids)

  • a list of varying length with one or several input Tensors IN THE ORDER given in the docstring: model([input_ids, attention_mask]) or model([input_ids, attention_mask, token_type_ids])

  • a dictionary with one or several input Tensors associated to the input names given in the docstring: model({"input_ids": input_ids, "token_type_ids": token_type_ids})

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

call(input_ids: Optional[Union[List[tensorflow.python.framework.ops.Tensor], List[numpy.ndarray], Dict[str, tensorflow.python.framework.ops.Tensor], Dict[str, numpy.ndarray], numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, attention_mask: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, token_type_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, position_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, inputs_embeds: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, training: Optional[bool] = False, **kwargs) → Union[transformers.modeling_tf_outputs.TFBaseModelOutput, Tuple[tensorflow.python.framework.ops.Tensor]][source]

The TFDebertaModel forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (np.ndarray, tf.Tensor, List[tf.Tensor] Dict[str, tf.Tensor] or Dict[str, np.ndarray] and each example must have the shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (np.ndarray or tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

Returns

A TFBaseModelOutput or a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size)) – Sequence of hidden-states at the output of the last layer of the model.

  • hidden_states (tuple(tf.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of tf.Tensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TFBaseModelOutput or tuple(tf.Tensor)

Example:

>>> from transformers import DebertaTokenizer, TFDebertaModel
>>> import tensorflow as tf

>>> tokenizer = DebertaTokenizer.from_pretrained('kamalkraj/deberta-base')
>>> model = TFDebertaModel.from_pretrained('kamalkraj/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
>>> outputs = model(inputs)

>>> last_hidden_states = outputs.last_hidden_state

TFDebertaPreTrainedModel

class transformers.TFDebertaPreTrainedModel(*args, **kwargs)[source]

An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters
  • inputs – Input tensor, or dict/list/tuple of input tensors.

  • training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

  • mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

TFDebertaForMaskedLM

class transformers.TFDebertaForMaskedLM(*args, **kwargs)[source]

DeBERTa Model with a language modeling head on top. The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

Note

TF 2.0 models accepts two formats as inputs:

  • having all inputs as keyword arguments (like PyTorch models), or

  • having all inputs as a list, tuple or dict in the first positional arguments.

This second option is useful when using tf.keras.Model.fit() method which currently requires having all the tensors in the first argument of the model call function: model(inputs).

If you choose this second option, there are three possibilities you can use to gather all the input Tensors in the first positional argument :

  • a single Tensor with input_ids only and nothing else: model(inputs_ids)

  • a list of varying length with one or several input Tensors IN THE ORDER given in the docstring: model([input_ids, attention_mask]) or model([input_ids, attention_mask, token_type_ids])

  • a dictionary with one or several input Tensors associated to the input names given in the docstring: model({"input_ids": input_ids, "token_type_ids": token_type_ids})

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

call(input_ids: Optional[Union[List[tensorflow.python.framework.ops.Tensor], List[numpy.ndarray], Dict[str, tensorflow.python.framework.ops.Tensor], Dict[str, numpy.ndarray], numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, attention_mask: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, token_type_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, position_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, inputs_embeds: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, labels: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, training: Optional[bool] = False, **kwargs) → Union[transformers.modeling_tf_outputs.TFMaskedLMOutput, Tuple[tensorflow.python.framework.ops.Tensor]][source]

The TFDebertaForMaskedLM forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (np.ndarray, tf.Tensor, List[tf.Tensor] Dict[str, tf.Tensor] or Dict[str, np.ndarray] and each example must have the shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (np.ndarray or tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (tf.Tensor or np.ndarray of shape (batch_size, sequence_length), optional) – Labels for computing the masked language modeling loss. Indices should be in [-100, 0, ..., config.vocab_size] (see input_ids docstring) Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size]

Returns

A TFMaskedLMOutput or a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (tf.Tensor of shape (n,), optional, where n is the number of non-masked labels, returned when labels is provided) – Masked language modeling (MLM) loss.

  • logits (tf.Tensor of shape (batch_size, sequence_length, config.vocab_size)) – Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

  • hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of tf.Tensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TFMaskedLMOutput or tuple(tf.Tensor)

Example:

>>> from transformers import DebertaTokenizer, TFDebertaForMaskedLM
>>> import tensorflow as tf

>>> tokenizer = DebertaTokenizer.from_pretrained('kamalkraj/deberta-base')
>>> model = TFDebertaForMaskedLM.from_pretrained('kamalkraj/deberta-base')

>>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="tf")
>>> inputs["labels"] = tokenizer("The capital of France is Paris.", return_tensors="tf")["input_ids"]

>>> outputs = model(inputs)
>>> loss = outputs.loss
>>> logits = outputs.logits

TFDebertaForSequenceClassification

class transformers.TFDebertaForSequenceClassification(*args, **kwargs)[source]

DeBERTa Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks.

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

Note

TF 2.0 models accepts two formats as inputs:

  • having all inputs as keyword arguments (like PyTorch models), or

  • having all inputs as a list, tuple or dict in the first positional arguments.

This second option is useful when using tf.keras.Model.fit() method which currently requires having all the tensors in the first argument of the model call function: model(inputs).

If you choose this second option, there are three possibilities you can use to gather all the input Tensors in the first positional argument :

  • a single Tensor with input_ids only and nothing else: model(inputs_ids)

  • a list of varying length with one or several input Tensors IN THE ORDER given in the docstring: model([input_ids, attention_mask]) or model([input_ids, attention_mask, token_type_ids])

  • a dictionary with one or several input Tensors associated to the input names given in the docstring: model({"input_ids": input_ids, "token_type_ids": token_type_ids})

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

call(input_ids: Optional[Union[List[tensorflow.python.framework.ops.Tensor], List[numpy.ndarray], Dict[str, tensorflow.python.framework.ops.Tensor], Dict[str, numpy.ndarray], numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, attention_mask: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, token_type_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, position_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, inputs_embeds: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, labels: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, training: Optional[bool] = False, **kwargs) → Union[transformers.modeling_tf_outputs.TFSequenceClassifierOutput, Tuple[tensorflow.python.framework.ops.Tensor]][source]

The TFDebertaForSequenceClassification forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (np.ndarray, tf.Tensor, List[tf.Tensor] Dict[str, tf.Tensor] or Dict[str, np.ndarray] and each example must have the shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (np.ndarray or tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (tf.Tensor or np.ndarray of shape (batch_size,), optional) – Labels for computing the sequence classification/regression loss. Indices should be in [0, ..., config.num_labels - 1]. If config.num_labels == 1 a regression loss is computed (Mean-Square loss), If config.num_labels > 1 a classification loss is computed (Cross-Entropy).

Returns

A TFSequenceClassifierOutput or a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (tf.Tensor of shape (batch_size, ), optional, returned when labels is provided) – Classification (or regression if config.num_labels==1) loss.

  • logits (tf.Tensor of shape (batch_size, config.num_labels)) – Classification (or regression if config.num_labels==1) scores (before SoftMax).

  • hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of tf.Tensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TFSequenceClassifierOutput or tuple(tf.Tensor)

Example:

>>> from transformers import DebertaTokenizer, TFDebertaForSequenceClassification
>>> import tensorflow as tf

>>> tokenizer = DebertaTokenizer.from_pretrained('kamalkraj/deberta-base')
>>> model = TFDebertaForSequenceClassification.from_pretrained('kamalkraj/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
>>> inputs["labels"] = tf.reshape(tf.constant(1), (-1, 1)) # Batch size 1

>>> outputs = model(inputs)
>>> loss = outputs.loss
>>> logits = outputs.logits

TFDebertaForTokenClassification

class transformers.TFDebertaForTokenClassification(*args, **kwargs)[source]

DeBERTa Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks.

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

Note

TF 2.0 models accepts two formats as inputs:

  • having all inputs as keyword arguments (like PyTorch models), or

  • having all inputs as a list, tuple or dict in the first positional arguments.

This second option is useful when using tf.keras.Model.fit() method which currently requires having all the tensors in the first argument of the model call function: model(inputs).

If you choose this second option, there are three possibilities you can use to gather all the input Tensors in the first positional argument :

  • a single Tensor with input_ids only and nothing else: model(inputs_ids)

  • a list of varying length with one or several input Tensors IN THE ORDER given in the docstring: model([input_ids, attention_mask]) or model([input_ids, attention_mask, token_type_ids])

  • a dictionary with one or several input Tensors associated to the input names given in the docstring: model({"input_ids": input_ids, "token_type_ids": token_type_ids})

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

call(input_ids: Optional[Union[List[tensorflow.python.framework.ops.Tensor], List[numpy.ndarray], Dict[str, tensorflow.python.framework.ops.Tensor], Dict[str, numpy.ndarray], numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, attention_mask: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, token_type_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, position_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, inputs_embeds: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, labels: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, training: Optional[bool] = False, **kwargs) → Union[transformers.modeling_tf_outputs.TFTokenClassifierOutput, Tuple[tensorflow.python.framework.ops.Tensor]][source]

The TFDebertaForTokenClassification forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (np.ndarray, tf.Tensor, List[tf.Tensor] Dict[str, tf.Tensor] or Dict[str, np.ndarray] and each example must have the shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (np.ndarray or tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • labels (tf.Tensor or np.ndarray of shape (batch_size, sequence_length), optional) – Labels for computing the token classification loss. Indices should be in [0, ..., config.num_labels - 1].

Returns

A TFTokenClassifierOutput or a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (tf.Tensor of shape (n,), optional, where n is the number of unmasked labels, returned when labels is provided) – Classification loss.

  • logits (tf.Tensor of shape (batch_size, sequence_length, config.num_labels)) – Classification scores (before SoftMax).

  • hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of tf.Tensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TFTokenClassifierOutput or tuple(tf.Tensor)

Example:

>>> from transformers import DebertaTokenizer, TFDebertaForTokenClassification
>>> import tensorflow as tf

>>> tokenizer = DebertaTokenizer.from_pretrained('kamalkraj/deberta-base')
>>> model = TFDebertaForTokenClassification.from_pretrained('kamalkraj/deberta-base')

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
>>> input_ids = inputs["input_ids"]
>>> inputs["labels"] = tf.reshape(tf.constant([1] * tf.size(input_ids).numpy()), (-1, tf.size(input_ids))) # Batch size 1

>>> outputs = model(inputs)
>>> loss = outputs.loss
>>> logits = outputs.logits

TFDebertaForQuestionAnswering

class transformers.TFDebertaForQuestionAnswering(*args, **kwargs)[source]

DeBERTa Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of the hidden-states output to compute span start logits and span end logits).

The DeBERTa model was proposed in DeBERTa: Decoding-enhanced BERT with Disentangled Attention by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen. It’s build on top of BERT/RoBERTa with two improvements, i.e. disentangled attention and enhanced mask decoder. With those two improvements, it out perform BERT/RoBERTa on a majority of tasks with 80GB pretraining data.

This model is also a tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

Note

TF 2.0 models accepts two formats as inputs:

  • having all inputs as keyword arguments (like PyTorch models), or

  • having all inputs as a list, tuple or dict in the first positional arguments.

This second option is useful when using tf.keras.Model.fit() method which currently requires having all the tensors in the first argument of the model call function: model(inputs).

If you choose this second option, there are three possibilities you can use to gather all the input Tensors in the first positional argument :

  • a single Tensor with input_ids only and nothing else: model(inputs_ids)

  • a list of varying length with one or several input Tensors IN THE ORDER given in the docstring: model([input_ids, attention_mask]) or model([input_ids, attention_mask, token_type_ids])

  • a dictionary with one or several input Tensors associated to the input names given in the docstring: model({"input_ids": input_ids, "token_type_ids": token_type_ids})

Parameters

config (DebertaConfig) – Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

call(input_ids: Optional[Union[List[tensorflow.python.framework.ops.Tensor], List[numpy.ndarray], Dict[str, tensorflow.python.framework.ops.Tensor], Dict[str, numpy.ndarray], numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, attention_mask: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, token_type_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, position_ids: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, inputs_embeds: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, start_positions: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, end_positions: Optional[Union[numpy.ndarray, tensorflow.python.framework.ops.Tensor]] = None, training: Optional[bool] = False, **kwargs) → Union[transformers.modeling_tf_outputs.TFQuestionAnsweringModelOutput, Tuple[tensorflow.python.framework.ops.Tensor]][source]

The TFDebertaForQuestionAnswering forward method, overrides the __call__() special method.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Parameters
  • input_ids (np.ndarray, tf.Tensor, List[tf.Tensor] Dict[str, tf.Tensor] or Dict[str, np.ndarray] and each example must have the shape (batch_size, sequence_length)) –

    Indices of input sequence tokens in the vocabulary.

    Indices can be obtained using transformers.DebertaTokenizer. See transformers.PreTrainedTokenizer.encode() and transformers.PreTrainedTokenizer.__call__() for details.

    What are input IDs?

  • attention_mask (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,

    • 0 for tokens that are masked.

    What are attention masks?

  • token_type_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:

    • 0 corresponds to a sentence A token,

    • 1 corresponds to a sentence B token.

    What are token type IDs?

  • position_ids (np.ndarray or tf.Tensor of shape (batch_size, sequence_length), optional) –

    Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.max_position_embeddings - 1].

    What are position IDs?

  • inputs_embeds (np.ndarray or tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) – Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.

  • output_attentions (bool, optional) – Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.

  • output_hidden_states (bool, optional) – Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.

  • return_dict (bool, optional) – Whether or not to return a ModelOutput instead of a plain tuple.

  • start_positions (tf.Tensor or np.ndarray of shape (batch_size,), optional) – Labels for position (index) of the start of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (sequence_length). Position outside of the sequence are not taken into account for computing the loss.

  • end_positions (tf.Tensor or np.ndarray of shape (batch_size,), optional) – Labels for position (index) of the end of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (sequence_length). Position outside of the sequence are not taken into account for computing the loss.

Returns

A TFQuestionAnsweringModelOutput or a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (DebertaConfig) and inputs.

  • loss (tf.Tensor of shape (batch_size, ), optional, returned when start_positions and end_positions are provided) – Total span extraction loss is the sum of a Cross-Entropy for the start and end positions.

  • start_logits (tf.Tensor of shape (batch_size, sequence_length)) – Span-start scores (before SoftMax).

  • end_logits (tf.Tensor of shape (batch_size, sequence_length)) – Span-end scores (before SoftMax).

  • hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) – Tuple of tf.Tensor (one for the output of the embeddings + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the initial embedding outputs.

  • attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) – Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

Return type

TFQuestionAnsweringModelOutput or tuple(tf.Tensor)

Example:

>>> from transformers import DebertaTokenizer, TFDebertaForQuestionAnswering
>>> import tensorflow as tf

>>> tokenizer = DebertaTokenizer.from_pretrained('kamalkraj/deberta-base')
>>> model = TFDebertaForQuestionAnswering.from_pretrained('kamalkraj/deberta-base')

>>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
>>> input_dict = tokenizer(question, text, return_tensors='tf')
>>> outputs = model(input_dict)
>>> start_logits = outputs.start_logits
>>> end_logits = outputs.end_logits

>>> all_tokens = tokenizer.convert_ids_to_tokens(input_dict["input_ids"].numpy()[0])
>>> answer = ' '.join(all_tokens[tf.math.argmax(start_logits, 1)[0] : tf.math.argmax(end_logits, 1)[0]+1])