Cross-lingual Capability Evaluation on Multilingual benchmark suite
40.4Multilingual ScoreBase Model
Evaluation results
| Method | Links | |
|---|---|---|
| Base ModelModel Backbone=Qwen1.5-MoE-A2.7B2025.12 | 40.4 | |
| SFT + CheckpointingModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=Full-Parameter2025.12 | 39.5 | |
| LOMOModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=Full-Parameter2025.12 | 39.3 | |
| GaLoreModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=Full-Parameter2025.12 | 39.2 | |
| DoRAModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=PEFT2025.12 | 38.9 | |
| RevFFNModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=Full-Parameter2025.12 | 38.8 | |
| LoRAModel Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=PEFT2025.12 | 38.5 | |
| (IA)³Model Backbone=Qwen1.5-MoE-A2.7B, Fine-tuning type=PEFT2025.12 | 38.2 |