Dependency Parsing on DialectBench benchmark leaderboard · SOTA2 Research

Dependency Parsing on DialectBench

30.8ALN ScoreXLM-R + Alignment + VAÇAÍ-Bowl

Evaluation Results

Method
XLM-R + Alignment + VAÇAÍ-Bowl2026.05	30.8	7.46	3.1	22.95	15.89	37.43	29.37	19.53	29.39	4.75
XLM-R + Alignment2026.05	29.73	6.23	0.81	20.02	16.47	37.68	28.98	19.49	29.67	3.52
mBERT + Alignment + VAÇAÍ-Bowl + TOPPing2026.05	29.47	10.41	1.1	22.05	21.75	43.84	31.3	20.12	36.86	5.21
mBERT + Alignment + TOPPing2026.05	28.92	9.59	1.08	21.82	20.38	43.83	30.7	19.27	36.24	3.98
XLM-R2026.05	28.87	5.34	1.16	21.04	14.8	38.1	29.01	21.12	27.6	2.6
mBERT + TOPPing2026.05	27.93	9.51	2.38	20.36	20.1	43.29	29.82	18.06	35.42	5.51
mBERT + Alignment2026.05	22	7.95	1.35	18.12	14.22	36.9	34.89	14.06	33.18	3.69
mBERT + VAÇAÍ-Bowl2026.05	22	7.05	1.35	18	15.85	40.21	35.25	15.12	35.42	4.59
mBERT + VAÇAÍ-Bowl2026.05	20.79	12.13	2.44	18.79	20.33	43.79	34.89	18.21	37.65	5.2
mBERT2026.05	20.53	10.41	2.32	18.11	20.33	42.29	34.96	17.19	37.05	5.2
mBERT2026.05	20.43	7.54	1.42	15.3	13.17	36.89	31.44	13.35	34	4.59
mBERT + Alignment2026.05	18.21	10.74	2.21	18.45	20.1	43.19	32.61	16.06	34.15	4.9