Multi-task Natural Language Processing on cross-domain multi-task NLP datasets
80.46Average PerformanceALoRA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ALoRABase Model=Qwen2-7B2025.09 | 80.46 | -7.98 | |
| HydraLoRABase Model=Qwen2-7B2025.09 | 80.03 | -7.14 | |
| LoRABase Model=Qwen2-7B2025.09 | 79.6 | -6.78 | |
| MoSLoRABase Model=Qwen2-7B2025.09 | 79.34 | -6.59 | |
| LoHaBase Model=Qwen2-7B2025.09 | 78.35 | -5.27 | |
| AdaLoRABase Model=Qwen2-7B2025.09 | 77.61 | -4.33 | |
| ST BaselineBase Model=Qwen2-7B2025.09 | 76.16 | — | |
| ALoRABase Model=LLaMA2-7B2025.09 | 67.13 | -8.33 | |
| HydraLoRABase Model=LLaMA2-7B2025.09 | 66.45 | -6.39 | |
| MoSLoRABase Model=LLaMA2-7B2025.09 | 66.18 | -6.58 | |
| LoHaBase Model=LLaMA2-7B2025.09 | 65.7 | -5.76 | |
| ST BaselineBase Model=LLaMA2-7B2025.09 | 63.36 | — | |
| AdaLoRABase Model=LLaMA2-7B2025.09 | 61.96 | -0.41 | |
| LoRABase Model=LLaMA2-7B2025.09 | 61.67 | 0.31 |