List your company

Multilingual Multiple-Choice Reasoning on Global PIQA 116 languages 1.0 (test)

79.31AccuracyQwen3.5-4B
62.555666.905371.25575.6047Mar 12, 2026

Evaluation results

MethodLinks
Qwen3.5-4Bdecoding=greedy, parameters=4B2026.0379.31
Qwen3-4Bdecoding=greedy, parameters=4B2026.0374.6
Gemma3-4Bdecoding=greedy, parameters=4B2026.0370.8
Ministral-3-3Bdecoding=greedy, parameters=3B2026.0370.7
Tiny Aya Globaldecoding=greedy2026.0368.3
SmolLM3-3Bdecoding=greedy, parameters=3B2026.0363.2