Question Answering on ARC Challenge (Accuracy and Sample Size)
90.3AccuracyGrACE-SC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GrACE-SCBackbone=Qwen2.5, Sampling budget T=82025.09 | 90.3 | 8 | |
| GrACE-ESBackbone=Qwen2.5, Sampling budget T=82025.09 | 90.1 | 2.47 | |
| SCBackbone=Qwen2.5, Sampling budget T=82025.09 | 89.5 | 8 | |
| ESCBackbone=Qwen2.5, Sampling budget T=82025.09 | 89.3 | 4.32 | |
| ASCBackbone=Qwen2.5, Sampling budget T=82025.09 | 88.7 | 2.23 | |
| GrACE-SCBackbone=Llama-3.1, Sampling budget T=82025.09 | 87.2 | 8 | |
| GrACE-ESBackbone=Llama-3.1, Sampling budget T=82025.09 | 87.2 | 2.14 | |
| SCBackbone=Llama-3.1, Sampling budget T=82025.09 | 86.3 | 8 | |
| ESCBackbone=Llama-3.1, Sampling budget T=82025.09 | 86.3 | 4.69 | |
| ASCBackbone=Llama-3.1, Sampling budget T=82025.09 | 85.9 | 2.34 |