Table 4 Comparison of the multilingual...

Table 4

Comparison of the multilingual model for undersampling and oversampling data

Model	Pearson			Time
Model	Train	Val	Test	Time
Undersampling data
Multilingual paraphrase-multilingual-MiniLM-L12-v2	0.72	0.70	0.59	00:15:04
Distilbert-base-multilingual-cased-v1	0.72	0.69	0.59	00:20:18
Oversampling data
Multilingual paraphrase-multilingual-MiniLM-L12-v2	0.62	0.59	0.61	01:17:21
Distilbert-base-multilingual-cased-v1	0.66	0.58	0.63	01:44:45

Model	Pearson			Time
Model	Train	Val	Test	Time
Undersampling data
Multilingual paraphrase-multilingual-MiniLM-L12-v2	0.72	0.70	0.59	00:15:04
Distilbert-base-multilingual-cased-v1	0.72	0.69	0.59	00:20:18
Oversampling data
Multilingual paraphrase-multilingual-MiniLM-L12-v2	0.62	0.59	0.61	01:17:21
Distilbert-base-multilingual-cased-v1	0.66	0.58	0.63	01:44:45

Note(s): The values are italized based on Pearson -Test, i.e. the highest value is the best result

Source(s): Table courtesy of Dhini and Girsang (2023)