đ Dutch NER in Flair (large model)
This is a large 4 - class NER model for Dutch that comes with Flair. It uses document - level XLM - R embeddings and FLERT technology, achieving an F1 - Score of 95.25 on the CoNLL - 03 Dutch dataset.
⨠Features
- High Precision: Achieves an F1 - Score of 95.25 on the CoNLL - 03 Dutch dataset.
- Four - Tag Prediction: Predicts four tags:
PER
(person name), LOC
(location name), ORG
(organization name), and MISC
(other name).
- Advanced Technology: Based on document - level XLM - R embeddings and FLERT.
Property |
Details |
Model Type |
4 - class NER model for Dutch |
Training Data |
CoNLL - 03 Dutch |
đĻ Installation
Requires: Flair (pip install flair
)
đģ Usage Examples
Basic Usage
from flair.data import Sentence
from flair.models import SequenceTagger
tagger = SequenceTagger.load("flair/ner-dutch-large")
sentence = Sentence("George Washington ging naar Washington")
tagger.predict(sentence)
print(sentence)
print('The following NER tags are found:')
for entity in sentence.get_spans('ner'):
print(entity)
This yields the following output:
Span [1,2]: "George Washington" [â Labels: PER (1.0)]
Span [5]: "Washington" [â Labels: LOC (1.0)]
So, the entities "George Washington" (labeled as a person) and "Washington" (labeled as a location) are found in the sentence "George Washington ging naar Washington".
đ§ Technical Details
The following Flair script was used to train this model:
import torch
from flair.datasets import CONLL_03_DUTCH
corpus = CONLL_03_DUTCH()
tag_type = 'ner'
tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
from flair.embeddings import TransformerWordEmbeddings
embeddings = TransformerWordEmbeddings(
model='xlm-roberta-large',
layers="-1",
subtoken_pooling="first",
fine_tune=True,
use_context=True,
)
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,
)
from flair.trainers import ModelTrainer
trainer = ModelTrainer(tagger, corpus, optimizer=torch.optim.AdamW)
from torch.optim.lr_scheduler import OneCycleLR
trainer.train('resources/taggers/ner-dutch-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.,
)
đ License
No license information provided in the original document.
đ Documentation
Cite
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}
}
Issues?
The Flair issue tracker is available here.