Cross-jurisdiction Legal Reasoning on KoBLEX benchmark leaderboard · SOTA2 Research

Cross-jurisdiction Legal Reasoning on KoBLEX

40.65Token F1 ScorePARSER (original)

Evaluation Results

Method	Links
PARSER (original)2026.04		40.65	46.24
JurisMA2026.04		34.88	69.72
SP + OR2026.04		32.18	21.5
IRCoT2026.04		31.62	20.42
SP2026.04		30.52	—
FLARE2026.04		29.54	40.64
CoT + OR2026.04		28.39	21.5
ProbTree2026.04		28.38	15.84
CoT2026.04		26.37	—
Self-Ask2026.04		16.59	9.29
BeamAggr2026.04		16.02	14.05