List your company

Multilingual Reasoning on MMLU-ProX (1k Stratified Subset Test)

18.7AccuracyTask Arithmetic
16.30816.92917.5518.171Feb 9, 2026

Evaluation results

MethodLinks
Task ArithmeticFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.0218.746.4
Weight AverageFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.0218.546.4
SAE (Local)Fusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct, scope=local2026.0218.247.8
RankmeanFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.0217.615.7
SAE (Global)Fusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct, scope=global2026.021748.4
PSOFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.0216.447.2