Science Question Answering on ARC Challenging (LLMcritic calls, Reduction %)
15.65LLMcritic CallsVecCISC + HAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VecCISC + HACBudget=20, Model=Mistral 7B2026.05 | 15.65 | -21.73 | |
| VecCISC + HACBudget=20, Model=Llama 3.1 8B2026.05 | 14.8 | -26 | |
| VecCISC + HACBudget=20, Model=Llama3.3 70B2026.05 | 13.31 | -33.43 | |
| VecCISC + HACBudget=20, Model=Qwen2.5 7B2026.05 | 10.94 | -45.3 | |
| VecCISC + HACBudget=20, Model=GPT-4o mini2026.05 | 10.39 | -48.07 |