LongformerForTokenClassification (#4638)

src/transformers/__init__.py

@@ -326,6 +326,7 @@ if is_torch_available():
         LongformerModel,
         LongformerForMaskedLM,
         LongformerForSequenceClassification,
+        LongformerForTokenClassification,
         LongformerForQuestionAnswering,
         LONGFORMER_PRETRAINED_MODEL_ARCHIVE_MAP,
     )

src/transformers/modeling_auto.py

@@ -106,6 +106,7 @@ from .modeling_longformer import (
     LongformerForMaskedLM,
     LongformerForQuestionAnswering,
     LongformerForSequenceClassification,
+    LongformerForTokenClassification,
     LongformerModel,
 )
 from .modeling_marian import MarianMTModel
@@ -282,6 +283,7 @@ MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = OrderedDict(
         (CamembertConfig, CamembertForTokenClassification),
         (XLMConfig, XLMForTokenClassification),
         (XLMRobertaConfig, XLMRobertaForTokenClassification),
+        (LongformerConfig, LongformerForTokenClassification),
         (RobertaConfig, RobertaForTokenClassification),
         (BertConfig, BertForTokenClassification),
         (XLNetConfig, XLNetForTokenClassification),

src/transformers/modeling_longformer.py

@@ -971,3 +971,102 @@ class LongformerForQuestionAnswering(BertPreTrainedModel):
             outputs = (total_loss,) + outputs
 
         return outputs  # (loss), start_logits, end_logits, (hidden_states), (attentions)
+
+
+@add_start_docstrings(
+    """Longformer 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. """,
+    LONGFORMER_START_DOCSTRING,
+)
+class LongformerForTokenClassification(BertPreTrainedModel):
+    config_class = LongformerConfig
+    pretrained_model_archive_map = LONGFORMER_PRETRAINED_MODEL_ARCHIVE_MAP
+    base_model_prefix = "longformer"
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+
+        self.longformer = LongformerModel(config)
+        self.dropout = nn.Dropout(config.hidden_dropout_prob)
+        self.classifier = nn.Linear(config.hidden_size, config.num_labels)
+
+        self.init_weights()
+
+    @add_start_docstrings_to_callable(LONGFORMER_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        token_type_ids=None,
+        position_ids=None,
+        inputs_embeds=None,
+        labels=None,
+    ):
+        r"""
+        labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`, defaults to :obj:`None`):
+            Labels for computing the token classification loss.
+            Indices should be in ``[0, ..., config.num_labels - 1]``.
+
+    Returns:
+        :obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.LongformerConfig`) and inputs:
+        loss (:obj:`torch.FloatTensor` of shape :obj:`(1,)`, `optional`, returned when ``labels`` is provided) :
+            Classification loss.
+        scores (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, config.num_labels)`)
+            Classification scores (before SoftMax).
+        hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``config.output_hidden_states=True``):
+            Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer)
+            of shape :obj:`(batch_size, sequence_length, hidden_size)`.
+
+            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
+        attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``config.output_attentions=True``):
+            Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape
+            :obj:`(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.
+
+    Examples::
+
+        from transformers import LongformerTokenizer, LongformerForTokenClassification
+        import torch
+
+        tokenizer = LongformerTokenizer.from_pretrained('longformer-base-4096')
+        model = LongformerForTokenClassification.from_pretrained('longformer-base-4096')
+        input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0)  # Batch size 1
+        labels = torch.tensor([1] * input_ids.size(1)).unsqueeze(0)  # Batch size 1
+        outputs = model(input_ids, labels=labels)
+        loss, scores = outputs[:2]
+
+        """
+
+        outputs = self.longformer(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            inputs_embeds=inputs_embeds,
+        )
+
+        sequence_output = outputs[0]
+
+        sequence_output = self.dropout(sequence_output)
+        logits = self.classifier(sequence_output)
+
+        outputs = (logits,) + outputs[2:]  # add hidden states and attention if they are here
+
+        if labels is not None:
+            loss_fct = CrossEntropyLoss()
+            # Only keep active parts of the loss
+            if attention_mask is not None:
+                active_loss = attention_mask.view(-1) == 1
+                active_logits = logits.view(-1, self.num_labels)
+                active_labels = torch.where(
+                    active_loss, labels.view(-1), torch.tensor(loss_fct.ignore_index).type_as(labels)
+                )
+                loss = loss_fct(active_logits, active_labels)
+            else:
+                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
+            outputs = (loss,) + outputs
+
+        return outputs  # (loss), scores, (hidden_states), (attentions)