MCPcopy Index your code
hub / github.com/SooLab/CGFormer / forward

Method forward

bert/modeling_bert.py:1022–1096  ·  view source on GitHub ↗

r""" labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`): Labels for computing the masked language modeling loss. Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)

(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,
        labels=None,
        encoder_hidden_states=None,
        encoder_attention_mask=None,
        output_attentions=None,
        output_hidden_states=None,
        **kwargs
    )

Source from the content-addressed store, hash-verified

1020 @add_start_docstrings_to_callable(BERT_INPUTS_DOCSTRING.format("(batch_size, sequence_length)"))
1021 @add_code_sample_docstrings(tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint="bert-base-uncased")
1022 def forward(
1023 self,
1024 input_ids=None,
1025 attention_mask=None,
1026 token_type_ids=None,
1027 position_ids=None,
1028 head_mask=None,
1029 inputs_embeds=None,
1030 labels=None,
1031 encoder_hidden_states=None,
1032 encoder_attention_mask=None,
1033 output_attentions=None,
1034 output_hidden_states=None,
1035 **kwargs
1036 ):
1037 r"""
1038 labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
1039 Labels for computing the masked language modeling loss.
1040 Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring)
1041 Tokens with indices set to ``-100`` are ignored (masked), the loss is only computed for the tokens with labels
1042 in ``[0, ..., config.vocab_size]``
1043 kwargs (:obj:`Dict[str, any]`, optional, defaults to `{}`):
1044 Used to hide legacy arguments that have been deprecated.
1045
1046 Returns:
1047 :obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.BertConfig`) and inputs:
1048 masked_lm_loss (`optional`, returned when ``labels`` is provided) ``torch.FloatTensor`` of shape ``(1,)``:
1049 Masked language modeling loss.
1050 prediction_scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.vocab_size)`)
1051 Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
1052 hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``):
1053 Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
1054 of shape :obj:`(batch_size, sequence_length, hidden_size)`.
1055
1056 Hidden-states of the model at the output of each layer plus the initial embedding outputs.
1057 attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``):
1058 Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
1059 :obj:`(batch_size, num_heads, sequence_length, sequence_length)`.
1060
1061 Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
1062 heads.
1063 """
1064 if "masked_lm_labels" in kwargs:
1065 warnings.warn(
1066 "The `masked_lm_labels` argument is deprecated and will be removed in a future version, use `labels` instead.",
1067 DeprecationWarning,
1068 )
1069 labels = kwargs.pop("masked_lm_labels")
1070 assert "lm_labels" not in kwargs, "Use `BertWithLMHead` for autoregressive language modeling task."
1071 assert kwargs == {}, f"Unexpected keyword arguments: {list(kwargs.keys())}."
1072
1073 outputs = self.bert(
1074 input_ids,
1075 attention_mask=attention_mask,
1076 token_type_ids=token_type_ids,
1077 position_ids=position_ids,
1078 head_mask=head_mask,
1079 inputs_embeds=inputs_embeds,

Callers

nothing calls this directly

Calls 1

keysMethod · 0.80

Tested by

no test coverage detected