Mathematical Reasoning on GSM8K full (test) (Accuracy, Avg.)
79.8AccuracySAE (Global)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SAE (Global)Fusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct, scope=global2026.02 | 79.8 | 48.4 | |
| PSOFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.02 | 78.01 | 47.2 | |
| SAE (Local)Fusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct, scope=local2026.02 | 77.48 | 47.8 | |
| Weight AverageFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.02 | 74.2 | 46.4 | |
| Task ArithmeticFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.02 | 74.1 | 46.4 | |
| RankmeanFusion Setting=Math + Multilingual, Backbone=LLaMA-3.2-3B-Instruct2026.02 | 13.7 | 15.7 |