Common sense on CommonsenseQA (Accuracy)
74AccuracyID-LoRA
Evaluation results
| Method | Links | |
|---|---|---|
| ID-LoRABase Model=Mistral-7B, Trainable Parameters (%)=0.62%2026.02 | 74 | |
| MoELoRABase Model=Mistral-7B, Trainable Parameters (%)=1.21%2026.02 | 71.9 | |
| HydraLoRABase Model=Mistral-7B, Trainable Parameters (%)=1.22%2026.02 | 71.3 | |
| DoRABase Model=Mistral-7B, Trainable Parameters (%)=1.16%2026.02 | 71.1 | |
| LoRABase Model=Mistral-7B, Trainable Parameters (%)=1.15%2026.02 | 70.8 | |
| FFTBase Model=Mistral-7B, Trainable Parameters (%)=100%2026.02 | 62.6 | |
| ID-LoRABase Model=LLaMA-3-8B, Trainable Parameters (%)=0.56%2026.02 | 47.4 | |
| LoRABase Model=LLaMA-3-8B, Trainable Parameters (%)=1.03%2026.02 | 46.9 | |
| HydraLoRABase Model=LLaMA-3-8B, Trainable Parameters (%)=1.10%2026.02 | 44.1 | |
| FFTBase Model=LLaMA-3-8B, Trainable Parameters (%)=100%2026.02 | 38.7 | |
| MoELoRABase Model=LLaMA-3-8B, Trainable Parameters (%)=1.09%2026.02 | 35.4 | |
| DoRABase Model=LLaMA-3-8B, Trainable Parameters (%)=1.05%2026.02 | 35 |