Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Flores-200

Benchmarks

Task NameDataset NameSOTA ResultTrend
Machine TranslationFLORES-200 xx→en
XCOMET97.68
46
Machine TranslationFlores-200 Romance group en->xx (test)
BLEU39.83
46
TranslationFLORES-200 it-en (devtest)
sacreBLEU37.0319
35
TranslationFLORES-200 en-it (devtest)
sacreBLEU33.831
35
Machine TranslationFLORES-200 (devtest)
Delta chrF++4.88
26
Machine TranslationFLORES-200 en→xx
XCOMET Score96.93
24
Machine TranslationFLORES-200 CS-EN (test)
BLEU21.9
22
Machine TranslationFLORES-200 (test)
xCOMET (DE)98.04
22
Machine TranslationFLORES-200 Target language
MT Score34.1
16
Tokenization EfficiencyFLORES-200 (dev+devtest)
Mean Brahmic Fertility10.43
11
Machine TranslationFLORES-200 EN-ZH (test)
BLEU24.3
11
Machine TranslationFLORES-200 EN-CS (test)
BLEU12.9
11
Machine TranslationFLORES-200 TR-EN (test)
BLEU19.1
11
Machine TranslationFLORES-200 hye_Armn-English (test)
COMET-2288.8
7
Machine TranslationFlores-200 en-bo
BLEU13.56
6
Machine TranslationFLORES-200 uig_Arab-English (test)
COMET-2286.4
6
Language IdentificationFLORES-200 CLD3 label set 77 languages (test)
F1 Score99.9
5
Machine TranslationFLoRes-200 Korean (test)
BLEU21.7
5
Per-word fertilityFLORES-200
Fertility (En)1.431
4
Machine TranslationFLORES-200 Spanish-Bengali (devtest)
chrF++47.8
4
Machine TranslationFLORES-200 Spanish-Yoruba (devtest)
chrF++24.52
4
Bitext Mining (with hard negatives)FLORES-200 34 languages
d-xsim++32.82
4
Machine TranslationFLORES-200 eng to mri
BLEU25.05
3
Machine TranslationFLORES-200 mri to eng
BLEU30.15
3
Bitext Mining (with hard negatives)FLORES-200 81 languages
xsim++18.65
3
Showing 25 of 32 rows