Natural Language Understanding on GLUE (CoLA, MRPC, WNLI Subset)
84.5MRPC AccuracyPLoRA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PLoRABase Model=LLAMA-3.1-8B, Configuration=Best LoRA2025.08 | 84.5 | 80 | 73.2 | — | |
| PLoRABase Model=LLAMA-3.2-3B, Configuration=Best LoRA2025.08 | 80.6 | 77.3 | 64.8 | — | |
| Default LoRABase Model=LLAMA-3.1-8B2025.08 | 80.3 | 73.8 | 67.6 | — | |
| Default LoRABase Model=LLAMA-3.2-3B2025.08 | 77.4 | 71.8 | 61.9 | — | |
| Base ModelBase Model=LLAMA-3.1-8B2025.08 | 71.3 | 71.9 | 54.9 | — | |
| Base ModelBase Model=LLAMA-3.2-3B2025.08 | 70.3 | 69.9 | 46.4 | — | |
| PLoRABase Model=QWen-2.5-7B, Configuration=Best LoRA2025.08 | 70 | 80.2 | 84.5 | — | |
| PLoRABase Model=QWen-2.5-3B, Configuration=Best LoRA2025.08 | 67.6 | 77.2 | 73.4 | — | |
| Default LoRABase Model=QWen-2.5-7B2025.08 | 64.7 | 68.4 | 80.1 | — | |
| Base ModelBase Model=QWen-2.5-7B2025.08 | 64.1 | 62.7 | 78.8 | — | |
| Default LoRABase Model=QWen-2.5-3B2025.08 | 62.6 | 53.8 | 66.2 | — | |
| Base ModelBase Model=QWen-2.5-3B2025.08 | 62.4 | 48.8 | 53.5 | — |