Language Understanding on LLM-adapters Understanding
59.19AccuracyLARS-LP
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LARS-LPBackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.672026.04 | 59.19 | 52.87 | |
| LoRABackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.452026.04 | 57 | 62.73 | |
| LARSBackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.672026.04 | 56.97 | 45.38 | |
| AdaLoRABackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.452026.04 | 53.99 | 52.81 | |
| IA3Backbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.012026.04 | 52.91 | 63.81 | |
| LARS-LPBackbone=Llama-3.2-1B, Trainable Parameters (%)=0.672026.04 | 39.32 | 32.39 | |
| AdaLoRABackbone=Llama-3.2-1B, Trainable Parameters (%)=0.452026.04 | 37.4 | 32.6 | |
| IA3Backbone=Llama-3.2-1B, Trainable Parameters (%)=0.012026.04 | 37.4 | 36.99 | |
| LARSBackbone=Llama-3.2-1B, Trainable Parameters (%)=0.672026.04 | 37.33 | 27.44 | |
| LoRABackbone=Llama-3.2-1B, Trainable Parameters (%)=0.452026.04 | 35.74 | 35.72 | |
| PromptBackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.0032026.04 | 14.71 | 47.39 | |
| PromptBackbone=Llama-3.2-1B, Trainable Parameters (%)=0.0032026.04 | 12.87 | 29.31 | |
| PrefixBackbone=Qwen2.5-7B-Instruct, Trainable Parameters (%)=0.032026.04 | 9.17 | 45.22 | |
| PrefixBackbone=Llama-3.2-1B, Trainable Parameters (%)=0.032026.04 | 9.05 | 25.49 |