Common-sense reasoning on Winogrande Medium (In-Distribution)
83.99AccuracyENS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ENSBackbone=Llama3.1-8B, Fine-tuning method=LoRA2026.07 | 83.99 | 11.17 | 0.55 | |
| MCDBackbone=Llama3.1-8B, Fine-tuning method=LoRA, Monte Carlo samples (M)=102026.07 | 83.18 | 13.05 | 0.64 | |
| LABackbone=Llama3.1-8B, Fine-tuning method=LoRA2026.07 | 82.83 | 3.5 | 0.51 | |
| DALorRABackbone=Llama3.1-8B, Fine-tuning method=LoRA, Monte Carlo samples (M)=102026.07 | 82.68 | 3.25 | 0.51 | |
| MAPBackbone=Llama3.1-8B, Fine-tuning method=LoRA2026.07 | 82.59 | 14.26 | 0.74 | |
| MLEBackbone=Llama3.1-8B, Fine-tuning method=LoRA2026.07 | 82.3 | 13.83 | 0.68 | |
| TFBBackbone=Llama3.1-8B, Fine-tuning method=LoRA, Monte Carlo samples (M)=102026.07 | 82.04 | 3.59 | 0.43 | |
| BLoBBackbone=Llama3.1-8B, Fine-tuning method=LoRA, Monte Carlo samples (M)=102026.07 | 81.64 | 3.42 | 0.47 | |
| C-LoRABackbone=Llama3.1-8B, Fine-tuning method=LoRA, Monte Carlo samples (M)=102026.07 | 81.62 | 9.97 | 0.57 |