Ner English Ontonotes Large

🚀 English NER in Flair (Ontonotes large model)

This is a large 18 - class NER model for English, which comes with Flair. It can effectively identify various named entities in English texts.

F1 - Score: 90.93 (Ontonotes)

This model predicts 18 tags:

This model is based on document - level XLM - R embeddings and FLERT.

🚀 Quick Start

✨ Features

• This is a large - scale 18 - class English NER model.
• It achieves an F1 - Score of 90.93 on the Ontonotes dataset.
• Based on document - level XLM - R embeddings and FLERT technology.

📦 Installation

Requires: Flair (pip install flair)

💻 Usage Examples

Basic Usage

from flair.data import Sentence
from flair.models import SequenceTagger

# load tagger
tagger = SequenceTagger.load("flair/ner-english-ontonotes-large")

# make example sentence
sentence = Sentence("On September 1st George won 1 dollar while watching Game of Thrones.")

# predict NER tags
tagger.predict(sentence)

# print sentence
print(sentence)

# print predicted NER spans
print('The following NER tags are found:')
# iterate over entities and print
for entity in sentence.get_spans('ner'):
    print(entity)

This yields the following output:

Span [2,3]: "September 1st"   [− Labels: DATE (1.0)]
Span [4]: "George"   [− Labels: PERSON (1.0)]
Span [6,7]: "1 dollar"   [− Labels: MONEY (1.0)]
Span [10,11,12]: "Game of Thrones"   [− Labels: WORK_OF_ART (1.0)]

So, the entities "September 1st" (labeled as a date), "George" (labeled as a person), "1 dollar" (labeled as a money) and "Game of Thrones" (labeled as a work of art) are found in the sentence "On September 1st George Washington won 1 dollar while watching Game of Thrones".

🔧 Technical Details

The following Flair script was used to train this model:

from flair.data import Corpus
from flair.datasets import ColumnCorpus
from flair.embeddings import WordEmbeddings, StackedEmbeddings, FlairEmbeddings

# 1. load the corpus (Ontonotes does not ship with Flair, you need to download and reformat into a column format yourself)
corpus: Corpus = ColumnCorpus(
                "resources/tasks/onto-ner",
                column_format={0: "text", 1: "pos", 2: "upos", 3: "ner"},
                tag_to_bioes="ner",
            )

# 2. what tag do we want to predict?
tag_type = 'ner'

# 3. make the tag dictionary from the corpus
tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)

# 4. initialize fine-tuneable transformer embeddings WITH document context
from flair.embeddings import TransformerWordEmbeddings

embeddings = TransformerWordEmbeddings(
    model='xlm-roberta-large',
    layers="-1",
    subtoken_pooling="first",
    fine_tune=True,
    use_context=True,
)

# 5. initialize bare-bones sequence tagger (no CRF, no RNN, no reprojection)
from flair.models import SequenceTagger

tagger = SequenceTagger(
    hidden_size=256,
    embeddings=embeddings,
    tag_dictionary=tag_dictionary,
    tag_type='ner',
    use_crf=False,
    use_rnn=False,
    reproject_embeddings=False,
)

# 6. initialize trainer with AdamW optimizer
from flair.trainers import ModelTrainer

trainer = ModelTrainer(tagger, corpus, optimizer=torch.optim.AdamW)

# 7. run training with XLM parameters (20 epochs, small LR)
from torch.optim.lr_scheduler import OneCycleLR

trainer.train('resources/taggers/ner-english-ontonotes-large',
              learning_rate=5.0e-6,
              mini_batch_size=4,
              mini_batch_chunk_size=1,
              max_epochs=20,
              scheduler=OneCycleLR,
              embeddings_storage_mode='none',
              weight_decay=0.,
              )

📄 License

Please cite the following paper when using this model.

@misc{schweter2020flert,
    title={FLERT: Document-Level Features for Named Entity Recognition},
    author={Stefan Schweter and Alan Akbik},
    year={2020},
    eprint={2011.06993},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

💡 Usage Tip

⚠️ Important Note

The Ontonotes dataset does not come with Flair. You need to download and reformat it into a column format yourself.

💡 Usage Tip

If you encounter any issues during use, you can check the Flair issue tracker here.