NER support for Albert in run_ner.py and NerPipeline (#2983)

* * Added support for Albert when fine-tuning for NER

* Added support for Albert in NER pipeline

* Added command-line options to examples/ner/run_ner.py to better control tokenization

* Added class AlbertForTokenClassification

* Changed output for NerPipeline to use .convert_ids_to_tokens(...) instead of .decode(...) to better reflect tokens

* Added ,

* Now passes style guide enforcement

* Changes from reviews.

* Code now passes style enforcement

* Added test for AlbertForTokenClassification

* Added test for AlbertForTokenClassification

src/transformers/__init__.py

@@ -255,6 +255,7 @@ if is_torch_available():
         AlbertForMaskedLM,
         AlbertForSequenceClassification,
         AlbertForQuestionAnswering,
+        AlbertForTokenClassification,
         load_tf_weights_in_albert,
         ALBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
     )

src/transformers/modeling_albert.py

@@ -788,6 +788,103 @@ class AlbertForSequenceClassification(AlbertPreTrainedModel):
         return outputs  # (loss), logits, (hidden_states), (attentions)
 
 
+@add_start_docstrings(
+    """Albert 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. """,
+    ALBERT_START_DOCSTRING,
+)
+class AlbertForTokenClassification(AlbertPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+
+        self.albert = AlbertModel(config)
+        self.dropout = nn.Dropout(config.hidden_dropout_prob)
+        self.classifier = nn.Linear(config.hidden_size, self.config.num_labels)
+
+        self.init_weights()
+
+    @add_start_docstrings_to_callable(ALBERT_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        token_type_ids=None,
+        position_ids=None,
+        head_mask=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.AlbertConfig`) 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 AlbertTokenizer, AlbertForTokenClassification
+        import torch
+
+        tokenizer = AlbertTokenizer.from_pretrained('albert-base-v2')
+        model = AlbertForTokenClassification.from_pretrained('albert-base-v2')
+
+        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.albert(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            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_loss]
+                active_labels = labels.view(-1)[active_loss]
+                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), logits, (hidden_states), (attentions)
+
+
 @add_start_docstrings(
     """Albert 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`). """,

src/transformers/modeling_auto.py

@@ -42,6 +42,7 @@ from .modeling_albert import (
     AlbertForMaskedLM,
     AlbertForQuestionAnswering,
     AlbertForSequenceClassification,
+    AlbertForTokenClassification,
     AlbertModel,
 )
 from .modeling_bart import BART_PRETRAINED_MODEL_ARCHIVE_MAP, BartForMaskedLM, BartForSequenceClassification, BartModel
@@ -233,6 +234,7 @@ MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = OrderedDict(
         (RobertaConfig, RobertaForTokenClassification),
         (BertConfig, BertForTokenClassification),
         (XLNetConfig, XLNetForTokenClassification),
+        (AlbertConfig, AlbertForTokenClassification),
     ]
 )
 

src/transformers/pipelines.py

@@ -636,7 +636,7 @@ class NerPipeline(Pipeline):
                 if self.model.config.id2label[label_idx] not in self.ignore_labels:
                     answer += [
                         {
-                            "word": self.tokenizer.decode([int(input_ids[idx])]),
+                            "word": self.tokenizer.convert_ids_to_tokens(int(input_ids[idx])),
                             "score": score[idx][label_idx].item(),
                             "entity": self.model.config.id2label[label_idx],
                         }