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

Commonsense Reasoning on XWinograd

80.7Accuracy

BETR

Updated 1mo ago

Evaluation Results

Method	Links
BETR 2026.04		80.7
NAG_Llama-3.2-3B 2026.04		80.6
NAG_SmolLM3-3B 2026.04		80.6
NAG_SmolLM3-3B 2026.04		80.2
NAG_Qwen3-1.7B 2026.04		80.1
NAG_Qwen3-1.7B 2026.04		79.9
NAG_Llama-3.2-3B 2026.04		79.9
Random 2026.04		76.5
FineWeb-Edu 2026.04		76.2
BETR 2026.04		76.1
1B Model (HQ filtering) 2026.04		68.25
1B Model (No filtering) 2026.04		68.06
1B Model (Multilingual classifier) 2026.04		67.66