Table 13 Results of transformers experiments.

From: Multilingual identification of nuanced dimensions of hope speech in social media texts

 

LM

Model

English

Spanish

German

Binary Hope Speech Detection

Monolingual

BERT

bert-base-uncased

0.8522

-

-

dccuchile/bert-base-spanish-wwm-uncased

-

0.8387

-

bert-base-german-dbmdz-uncased

-

-

0.8704

RoBERTa

roberta-base

0.8511

-

-

bertin-project/bertin-roberta-base-spanish

-

0.8010

-

uklfr/gottbert-base

-

-

0.8625

Multilingual

mBERT

google-bert/bert-base-multilingual-uncased

0.8407

0.8272

0.8541

XLM-RoBERTa

FacebookAI/xlm-roberta-base

0.8623

0.8369

0.8704

Multiclass Hope Speech Detection

Monolingual

BERT

bert-base-uncased

0.7069

-

-

dccuchile/bert-base-spanish-wwm-uncased

-

0.6745

-

bert-base-german-dbmdz-uncased

-

-

0.7007

RoBERTa

roberta-base

0.7081

-

-

bertin-project/bertin-roberta-base-spanish

-

0.6534

-

uklfr/gottbert-base

-

-

0.6977

Multilingual

mBERT

google-bert/bert-base-multilingual-uncased

0.6743

0.6572

0.6646

XLM-RoBERTa

FacebookAI/xlm-roberta-base

0.6907

0.6801

0.6878