Language Identification on UDHR 360 languages (out-of-domain)
90.6F1 ScoreGlotLID-M+ConLID-S
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GlotLID-M+ConLID-STraining Dataset=Combined/Ensemble2025.06 | 90.6 | 0.0208 | |
| ConLID-STraining Dataset=GlotLID-C-train-down2025.06 | 90.12 | 0.0217 | |
| ConLID-S+LID_{CE}Training Dataset=GlotLID-C-train-down2025.06 | 89.98 | 0.0218 | |
| ConLID-HTraining Dataset=GlotLID-C-train-down2025.06 | 89.61 | 0.0221 | |
| LID_{SCL}Training Dataset=GlotLID-C-train-down2025.06 | 89.38 | 0.021 | |
| LID_{CE}Training Dataset=GlotLID-C-train-down2025.06 | 89.25 | 0.0222 | |
| GlotLID-MTraining Dataset=Subsets of GlotLID-C2025.06 | 89.19 | 0.0214 |