metadata

language:
  - es
license: cc-by-4.0
library_name: span-marker
tags:
  - span-marker
  - token-classification
  - ner
  - named-entity-recognition
  - generated_from_span_marker_trainer
datasets:
  - xtreme
metrics:
  - precision
  - recall
  - f1
widget:
  - text: Me llamo Álvaro y vivo en Barcelona (España).
  - text: Marie Curie fue profesora en la Universidad de Paris.
  - text: >-
      La Universidad de Salamanca es la universidad en activo más antigua de
      España.
pipeline_tag: token-classification
base_model: bert-base-multilingual-cased
model-index:
  - name: SpanMarker with bert-base-multilingual-cased on xtreme/PAN-X.es
    results:
      - task:
          type: token-classification
          name: Named Entity Recognition
        dataset:
          name: xtreme/PAN-X.es
          type: xtreme
          split: eval
        metrics:
          - type: f1
            value: 0.9186626746506986
            name: F1
          - type: precision
            value: 0.9231154938993816
            name: Precision
          - type: recall
            value: 0.9142526071842411
            name: Recall

SpanMarker with bert-base-multilingual-cased on xtreme/PAN-X.es

This is a SpanMarker model trained on the xtreme/PAN-X.es dataset that can be used for Named Entity Recognition. This SpanMarker model uses bert-base-multilingual-cased as the underlying encoder.

Model Details

Model Description

Model Type: SpanMarker
Encoder: bert-base-multilingual-cased
Maximum Sequence Length: 512 tokens
Maximum Entity Length: 8 words
Training Dataset: xtreme/PAN-X.es
Languages: es
License: cc-by-4.0

Model Sources

Repository: SpanMarker on GitHub
Thesis: SpanMarker For Named Entity Recognition

Model Labels

Label	Examples
LOC	"Salamanca", "Paris", "Barcelona (España)"
ORG	"ONU", "Fútbol Club Barcelona", "Museo Nacional del Prado"
PER	"Fray Luis de León", "Leo Messi", "Álvaro Bartolomé"

Uses

Direct Use for Inference

from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("alvarobartt/bert-base-multilingual-cased-ner-spanish")
# Run inference
entities = model.predict("Marie Curie fue profesora en la Universidad de Paris.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Sentence length	3	6.4642	64
Entities per sentence	1	1.2375	24

Training Hyperparameters

learning_rate: 5e-05
train_batch_size: 8
eval_batch_size: 4
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: 2

Training Results

Epoch	Step	Validation Loss	Validation Precision	Validation Recall	Validation F1	Validation Accuracy
0.3998	1000	0.0388	0.8761	0.8641	0.8701	0.9223
0.7997	2000	0.0326	0.8995	0.8740	0.8866	0.9341
1.1995	3000	0.0277	0.9076	0.9019	0.9047	0.9424
1.5994	4000	0.0261	0.9143	0.9113	0.9128	0.9473
1.9992	5000	0.0234	0.9231	0.9143	0.9187	0.9502

Framework Versions

Python: 3.10.12
SpanMarker: 1.3.1.dev
Transformers: 4.33.3
PyTorch: 2.0.1+cu118
Datasets: 2.14.5
Tokenizers: 0.13.3

Citation

BibTeX

@software{Aarsen_SpanMarker,
    author = {Aarsen, Tom},
    license = {Apache-2.0},
    title = {{SpanMarker for Named Entity Recognition}},
    url = {https://github.com/tomaarsen/SpanMarkerNER}
}