Supported Languages

WLDetect supports 148 languages trained on OpenLID-v2 and evaluated on FLORES+.

Performance Summary

Average Accuracy: 92.92%
Macro Precision: 0.9399
Macro Recall: 0.9294
Macro F1: 0.9274
Languages ≥ 95% accuracy: 103/148
Languages ≥ 90% accuracy: 113/148

Language List

Languages sorted by FLORES accuracy (highest to lowest).

Language Code	Accuracy	Precision	F1
asm_Beng	100.00%	1.0000	1.0000
ben_Beng	100.00%	0.9930	0.9965
cmn_Hant	100.00%	0.9379	0.9680
dzo_Tibt	100.00%	1.0000	1.0000
ell_Grek	100.00%	0.9970	0.9985
guj_Gujr	100.00%	1.0000	1.0000
heb_Hebr	100.00%	1.0000	1.0000
hun_Latn	100.00%	0.9477	0.9732
hye_Armn	100.00%	1.0000	1.0000
jpn_Jpan	100.00%	0.9990	0.9995
kan_Knda	100.00%	1.0000	1.0000
kat_Geor	100.00%	1.0000	1.0000
khm_Khmr	100.00%	1.0000	1.0000
kor_Hang	100.00%	1.0000	1.0000
lao_Laoo	100.00%	1.0000	1.0000
mal_Mlym	100.00%	1.0000	1.0000
mya_Mymr	100.00%	1.0000	1.0000
ory_Orya	100.00%	1.0000	1.0000
pan_Guru	100.00%	1.0000	1.0000
pes_Arab	100.00%	0.8692	0.9300
sat_Olck	100.00%	1.0000	1.0000
shn_Mymr	100.00%	1.0000	1.0000
sin_Sinh	100.00%	1.0000	1.0000
snd_Arab	100.00%	0.9970	0.9985
tam_Taml	100.00%	1.0000	1.0000
taq_Tfng	100.00%	1.0000	1.0000
tel_Telu	100.00%	1.0000	1.0000
tha_Thai	100.00%	1.0000	1.0000
uig_Arab	100.00%	0.9990	0.9995
ukr_Cyrl	100.00%	0.9842	0.9920
urd_Arab	100.00%	0.9130	0.9545
vie_Latn	100.00%	0.9891	0.9945
ckb_Arab	99.90%	1.0000	0.9995
hin_Deva	99.90%	0.5605	0.7181
kir_Cyrl	99.90%	0.9891	0.9940
lit_Latn	99.90%	0.9755	0.9871
lvs_Latn	99.90%	0.8078	0.8933
npi_Deva	99.90%	0.9970	0.9980
rus_Cyrl	99.90%	0.9930	0.9960
amh_Ethi	99.80%	0.9531	0.9750
arb_Arab	99.80%	0.4802	0.6484
mar_Deva	99.80%	0.9891	0.9935
ron_Latn	99.80%	0.9698	0.9837
tuk_Latn	99.80%	0.9822	0.9900
tur_Latn	99.80%	0.9679	0.9827
eng_Latn	99.70%	0.8955	0.9435
kik_Latn	99.70%	0.9832	0.9900
pbt_Arab	99.70%	1.0000	0.9985
pol_Latn	99.70%	0.9395	0.9674
als_Latn	99.60%	0.9641	0.9798
bjn_Arab	99.60%	0.9940	0.9950
deu_Latn	99.60%	0.9697	0.9827
khk_Cyrl	99.60%	0.9990	0.9975
mlt_Latn	99.60%	0.9890	0.9925
por_Latn	99.60%	0.9077	0.9498
azj_Latn	99.50%	0.7619	0.8630
bul_Cyrl	99.50%	0.9940	0.9945
fra_Latn	99.50%	0.9026	0.9466
tat_Cyrl	99.40%	0.8528	0.9180
tgk_Cyrl	99.40%	1.0000	0.9970
ekk_Latn	99.30%	0.9252	0.9579
mni_Beng	99.30%	1.0000	0.9965
fin_Latn	99.20%	0.9556	0.9734
kaz_Cyrl	99.20%	0.9940	0.9930
uzn_Latn	99.20%	0.8942	0.9406
ilo_Latn	99.00%	0.7992	0.8844
nld_Latn	99.00%	0.7711	0.8669
slk_Latn	99.00%	0.9164	0.9518
epo_Latn	98.90%	0.9880	0.9885
bel_Cyrl	98.80%	1.0000	0.9939
cym_Latn	98.80%	0.9970	0.9924
mkd_Cyrl	98.80%	0.9572	0.9724
tpi_Latn	98.80%	0.9919	0.9899
hau_Latn	98.70%	0.9619	0.9743
ita_Latn	98.70%	0.8586	0.9183
nus_Latn	98.70%	1.0000	0.9934
eus_Latn	98.50%	0.9590	0.9718
ewe_Latn	98.50%	0.9534	0.9689
ces_Latn	97.99%	0.9939	0.9869
gaz_Latn	97.89%	0.9683	0.9736
swe_Latn	97.89%	0.9597	0.9692
bak_Cyrl	97.79%	1.0000	0.9888
spa_Latn	97.69%	0.9137	0.9443
ceb_Latn	97.59%	0.8935	0.9329
cmn_Hans	97.49%	1.0000	0.9873
slv_Latn	97.29%	0.9327	0.9524
tsn_Latn	97.19%	0.9133	0.9417
afr_Latn	96.89%	0.9244	0.9461
som_Latn	96.79%	0.9718	0.9698
fij_Latn	96.69%	0.9377	0.9521
hat_Latn	96.59%	0.9008	0.9322
gle_Latn	96.39%	0.9049	0.9335
fil_Latn	96.29%	0.9152	0.9384
ind_Latn	96.29%	0.5000	0.6582
lin_Latn	95.89%	0.9775	0.9681
srp_Cyrl	95.89%	0.9927	0.9755
yue_Hant	95.79%	1.0000	0.9785
twi_Latn	95.74%	0.9770	0.9671
ibo_Latn	95.59%	0.9958	0.9754
nya_Latn	95.59%	0.7975	0.8695
sna_Latn	95.39%	0.9342	0.9439
tso_Latn	95.29%	0.8482	0.8975
tir_Ethi	95.09%	0.9979	0.9738
hrv_Latn	94.88%	0.9643	0.9565
swh_Latn	94.18%	0.9418	0.9418
war_Latn	93.58%	0.9648	0.9501
kab_Latn	93.48%	0.9759	0.9549
bem_Latn	92.78%	0.9095	0.9186
run_Latn	92.38%	0.8583	0.8899
kmr_Latn	91.57%	0.9796	0.9466
yor_Latn	91.27%	0.9681	0.9396
nob_Latn	91.22%	0.9182	0.9152
kas_Arab	90.17%	0.9967	0.9468
pag_Latn	89.87%	0.9614	0.9290
pap_Latn	89.77%	0.9179	0.9077
gug_Latn	89.67%	0.8756	0.8860
oci_Latn	88.52%	0.9231	0.9037
lua_Latn	88.47%	0.8991	0.8918
gla_Latn	88.16%	0.9681	0.9228
lus_Latn	87.96%	0.9300	0.9041
quy_Latn	87.26%	0.9285	0.8997
dan_Latn	87.16%	0.8076	0.8384
ktu_Latn	87.06%	0.9538	0.9103
fao_Latn	85.96%	0.8248	0.8418
mos_Latn	85.96%	0.9695	0.9112
fur_Latn	85.36%	0.9092	0.8805
san_Deva	84.85%	1.0000	0.9181
smo_Latn	84.05%	0.9405	0.8877
cat_Latn	83.45%	0.9014	0.8667
isl_Latn	81.44%	0.8817	0.8467
lug_Latn	81.34%	0.9632	0.8820
tum_Latn	80.54%	0.9710	0.8805
zul_Latn	80.34%	0.7629	0.7826
vec_Latn	78.44%	0.9861	0.8737
xho_Latn	78.44%	0.7045	0.7423
jav_Latn	77.03%	0.8321	0.8000
ayr_Latn	76.43%	0.9361	0.8415
plt_Latn	75.93%	0.9895	0.8593
sag_Latn	72.42%	0.9014	0.8031
mri_Latn	71.01%	0.9944	0.8286
ban_Latn	63.59%	0.7955	0.7068
lim_Latn	63.09%	0.9844	0.7689
sun_Latn	55.07%	0.9515	0.6976
knc_Latn	53.56%	1.0000	0.6976
zsm_Latn	51.25%	0.8559	0.6412
knc_Arab	45.84%	1.0000	0.6286
bho_Deva	35.71%	0.9972	0.5258
arz_Arab	27.88%	0.8968	0.4254

Notes

Language Codes: ISO 639-3 language code + ISO 15924 script code
- Format: {lang}_{Script} (e.g., eng_Latn for English in Latin script)
FLORES Evaluation: FLORES+ dev set (1012 sentences per language)
Removed Languages: Languages with high confusion or insufficient training data:
- crh_Latn (Crimean Tatar)
- ltz_Latn (Luxembourgish)