mrm8488/bert-spanish-cased-finetuned-pos

Spanish BERT (BETO) + POS

This model is a fine-tuned on Spanish CONLL CORPORA version of the Spanish BERT cased (BETO) for POS (Part of Speech tagging) downstream task.

Details of the downstream task (POS) - Dataset

• Dataset: CONLL Corpora ES with data augmentation techniques

I preprocessed the dataset and split it as train / dev (80/20)

Dataset	# Examples
Train	340 K
Dev	50 K

• Fine-tune on NER script provided by Huggingface

• 60 Labels covered:

AO, AQ, CC, CS, DA, DD, DE, DI, DN, DP, DT, Faa, Fat, Fc, Fd, Fe, Fg, Fh, Fia, Fit, Fp, Fpa, Fpt, Fs, Ft, Fx, Fz, I, NC, NP, P0, PD, PI, PN, PP, PR, PT, PX, RG, RN, SP, VAI, VAM, VAN, VAP, VAS, VMG, VMI, VMM, VMN, VMP, VMS, VSG, VSI, VSM, VSN, VSP, VSS, Y and Z

Metrics on evaluation set:

Metric	# score
F1	90.06
Precision	89.46
Recall	90.67

Model in action

Fast usage with pipelines:

from transformers import pipeline

nlp_pos = pipeline(
    "ner",
    model="mrm8488/bert-spanish-cased-finetuned-pos",
    tokenizer=(
        'mrm8488/bert-spanish-cased-finetuned-pos',  
        {"use_fast": False}
))


text = 'Mis amigos están pensando en viajar a Londres este verano'

nlp_pos(text)

#Output:
'''
[{'entity': 'NC', 'score': 0.7792173624038696, 'word': '[CLS]'},
 {'entity': 'DP', 'score': 0.9996283650398254, 'word': 'Mis'},
 {'entity': 'NC', 'score': 0.9999253749847412, 'word': 'amigos'},
 {'entity': 'VMI', 'score': 0.9998560547828674, 'word': 'están'},
 {'entity': 'VMG', 'score': 0.9992249011993408, 'word': 'pensando'},
 {'entity': 'SP', 'score': 0.9999602437019348, 'word': 'en'},
 {'entity': 'VMN', 'score': 0.9998666048049927, 'word': 'viajar'},
 {'entity': 'SP', 'score': 0.9999545216560364, 'word': 'a'},
 {'entity': 'VMN', 'score': 0.8722310662269592, 'word': 'Londres'},
 {'entity': 'DD', 'score': 0.9995203614234924, 'word': 'este'},
 {'entity': 'NC', 'score': 0.9999248385429382, 'word': 'verano'},
 {'entity': 'NC', 'score': 0.8802427649497986, 'word': '[SEP]'}]
 '''

16 POS tags version also available here

Created by Manuel Romero/@mrm8488

Made with ♥ in Spain