Human Alignment on MultiNLI benchmark leaderboard · SOTA2 Research

Human Alignment on MultiNLI

23R@1MATCHA

Evaluation Results

Method	Links
MATCHA2026.05		23	56.08
BERTScore2026.05		15	48.6
METEOR2026.05		11.5	46.64
R2-F12026.05		10.5	43.53
SimCSE2026.05		10.5	49.13
R1-F12026.05		8	47.88
EmbSim2026.05		8	47
BLEURT2026.05		4	40.38
RL-F12026.05		3	45.86