license: cc-by-nc-sa-4.0 library_name: span-marker tags:

• span-marker
• token-classification
• ner
• named-entity-recognition pipeline_tag: token-classification widget:
• text: "Amelia Earthart flog mit ihrer einmotorigen Lockheed Vega 5B über den Atlantik nach Paris." example_title: "German"
• text: "Amelia Earhart flew her single engine Lockheed Vega 5B across the Atlantic to Paris." example_title: "English"
• text: "Amelia Earthart voló su Lockheed Vega 5B monomotor a través del Océano Atlántico hasta París." example_title: "Spanish"
• text: "Amelia Earthart a fait voler son monomoteur Lockheed Vega 5B à travers l'ocean Atlantique jusqu'à Paris." example_title: "French"
• text: "Amelia Earhart ha volato con il suo monomotore Lockheed Vega 5B attraverso l'Atlantico fino a Parigi." example_title: "Italian"
• text: "Amelia Earthart vloog met haar één-motorige Lockheed Vega 5B over de Atlantische Oceaan naar Parijs." example_title: "Dutch"
• text: "Amelia Earthart przeleciała swoim jednosilnikowym samolotem Lockheed Vega 5B przez Ocean Atlantycki do Paryża." example_title: "Polish"
• text: "Amelia Earhart voou em seu monomotor Lockheed Vega 5B através do Atlântico para Paris." example_title: "Portuguese"
• text: "Амелия Эртхарт перелетела на своем одномоторном самолете Lockheed Vega 5B через Атлантический океан в Париж." example_title: "Russian"
• text: "Amelia Earthart flaug eins hreyfils Lockheed Vega 5B yfir Atlantshafið til Parísar." example_title: "Icelandic"
• text: "Η Amelia Earthart πέταξε το μονοκινητήριο Lockheed Vega 5B της πέρα ​​από τον Ατλαντικό Ωκεανό στο Παρίσι." example_title: "Greek"
• text: "Amelia Earhartová přeletěla se svým jednomotorovým Lockheed Vega 5B přes Atlantik do Paříže." example_title: "Czech"
• text: "Amelia Earhart lensi yksimoottorisella Lockheed Vega 5B:llä Atlantin yli Pariisiin." example_title: "Finnish"
• text: "Amelia Earhart fløj med sin enmotoriske Lockheed Vega 5B over Atlanten til Paris." example_title: "Danish"
• text: "Amelia Earhart flög sin enmotoriga Lockheed Vega 5B över Atlanten till Paris." example_title: "Swedish"
• text: "Amelia Earhart fløy sin enmotoriske Lockheed Vega 5B over Atlanterhavet til Paris." example_title: "Norwegian"
• text: "Amelia Earhart și-a zburat cu un singur motor Lockheed Vega 5B peste Atlantic până la Paris." example_title: "Romanian"
• text: "Amelia Earhart menerbangkan mesin tunggal Lockheed Vega 5B melintasi Atlantik ke Paris." example_title: "Indonesian"
• text: "Амелія Эрхарт пераляцела на сваім аднаматорным Lockheed Vega 5B праз Атлантыку ў Парыж." example_title: "Belarusian"
• text: "Амелія Ергарт перелетіла на своєму одномоторному літаку Lockheed Vega 5B через Атлантику до Парижа." example_title: "Ukrainian"
• text: "Amelia Earhart preletjela je svojim jednomotornim zrakoplovom Lockheed Vega 5B preko Atlantika do Pariza." example_title: "Croatian"
• text: "Amelia Earhart lendas oma ühemootoriga Lockheed Vega 5B üle Atlandi ookeani Pariisi ." example_title: "Estonian" model-index:
  • name: SpanMarker w. bert-base-multilingual-cased on MultiNERD by Tom Aarsen results:
    • task: type: token-classification name: Named Entity Recognition dataset: type: Babelscape/multinerd name: MultiNERD split: test revision: 2814b78e7af4b5a1f1886fe7ad49632de4d9dd25 metrics:
      • type: f1 value: 0.92478 name: F1
      • type: precision value: 0.93385 name: Precision
      • type: recall value: 0.91588 name: Recall datasets:
  • Babelscape/multinerd language:
  • multilingual metrics:
  • f1
  • recall
  • precision base_model: bert-base-multilingual-cased

SpanMarker for Multilingual Named Entity Recognition

This is a SpanMarker model that can be used for multilingual Named Entity Recognition trained on the MultiNERD dataset. In particular, this SpanMarker model uses bert-base-multilingual-cased as the underlying encoder. See train.py for the training script.

Is your data not (always) capitalized correctly? Then consider using this uncased variant of this model by @lxyuan for better performance: lxyuan/span-marker-bert-base-multilingual-uncased-multinerd.

Metrics

Label set

Usage

To use this model for inference, first install the span_marker library:

pip install span_marker

You can then run inference with this model like so:

from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("tomaarsen/span-marker-mbert-base-multinerd")
# Run inference
entities = model.predict("Amelia Earhart flew her single engine Lockheed Vega 5B across the Atlantic to Paris.")

See the SpanMarker repository for documentation and additional information on this library.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Framework versions

• SpanMarker 1.2.4
• Transformers 4.28.1
• Pytorch 1.13.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.2

See also

• lxyuan/span-marker-bert-base-multilingual-cased-multinerd is similar to this model, but trained on 3 epochs instead of 2. It reaches better performance on 7 out of the 10 languages.
• lxyuan/span-marker-bert-base-multilingual-uncased-multinerd is a strong uncased variant of this model, also trained on 3 epochs instead of 2.

Contributions

Many thanks to Simone Tedeschi from Babelscape for his insight when training this model and his involvement in the creation of the training dataset.