Question Answering on Commonsense Reasoning
25.4OBQALLR
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| LLRBackbone=Llama-3.2-1B, Evaluation Protocol=Finetuning2026.05 | 25.4 | 65 | 38.64 | 68.64 | 45.49 | 46.54 | 76.25 | 62.67 | 53.58 | |
| UniformBackbone=Llama-3.2-1B, Evaluation Protocol=Finetuning2026.05 | 25 | 64.8 | 38.82 | 68.14 | 45.29 | 47.54 | 72.25 | 61.67 | 52.94 | |
| Un-tunedBackbone=Llama-3.2-1B, Evaluation Protocol=Zero-shot / Base2026.05 | 24.6 | 59.91 | 35.75 | 68.31 | 44.98 | 41.66 | 74.37 | 55.2 | 49.94 |