FLORES
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
FLORES-200 Full (190 labels) (test)
96.8F1 Score
4
FLORES+
0.007FPR (Norwegian Bokmål)
4
Flores+ Ko to En
31.8Score
4
Flores Japanese-English (test)
87.81Direct Assessment
4
Flores German-English (test)
89.29DA Score
4
FLORES-200 34 languages
2.9d-xsim
4
Flores (test)
11.2BLEU
4
Flores (test)
16.5BLEU
4
Flores (test)
28BLEU
4
Flores
15.3ChrF++ Score
4
Flores
42.5ChrF++ Score
4
Flores
53.8ChrF++ Score
4
FLORES Low-resource and High-resource
14.15OTR (Low-Resource)
4
FLORES Low-resource and High-resource
23.26OTR (Low Resource)
4
FLORES (t)
58.28Language Democratization
4
Flores 101 (en-x)
24.9BLEU Score
4
Flores 101 (L-H)
1Zero-Shot Off-Target Rate
4
Flores 101 (H-L)
1Zero-Shot Off-Target Rate
4
Flores 101 (H-H)
3Off-Target Rate (Zero-Shot)
4
FLORES en->xx key noise (test)
-72.46Score (en->de)
4
FLORES 2.5.1 (clean)
67.1chrF (en->de)
4
FLORES clean 2.5.1
69.9chrF (de-en)
4
FLORES key perturbation (test)
-48.8Error (en→de)
4
FLORES dupe perturbation (test)
-3.7Score (en→de)
4
FLORES xx→en, key noise v1.0 (test)
-50.3Score (de→en)
4