Share your thoughts, 1 month free Claude Pro on usSee more

Linguistic Minimal Pairs on BLiMP 10% (test)

76.1BLiMP 10% Accuracy

DTG

Updated 16d ago

Evaluation Results

Method	Links
DTG 2026.05		76.1
GiLT-DP 2026.05		76.1
GiLT-PSD 2026.05		75.7
TXL-Large 2026.05		75.4
PLM-Mask 2026.05		75.3
PLM 2026.05		75.1
TXL (baseline) 2026.05		75.1
GiLT-DM 2026.05		74.9
Pushdown-LM 2026.05		74
TG 2026.05		73.5
GiLT-PAS 2026.05		73