Question Answering on HotPotQA (Accuracy, F1-Score, Recall, Precision)
84AccuracyCARE10
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CARE10Model=2.0-flashBase2026.04 | 84 | 82.5 | 75.3 | 91.1 | |
| CARE10Model=LLaMa70bBase2026.04 | 83.3 | 81.5 | 73.8 | 91 | |
| CARE10Model=GPT-4.1Base2026.04 | 82.7 | 81.4 | 75.7 | 88 | |
| CARE10Model=1.5-flash-8b2026.04 | 81.2 | 82.2 | 87.2 | 77.8 | |
| CARE10Model=o4-mini2026.04 | 78.1 | 73.2 | 60 | 94 | |
| DirectModel=LLaMa8b2026.04 | 74.8 | 71.9 | 64.3 | 81.5 | |
| DirectModel=LLaMa70bBase2026.04 | 74.5 | 68.6 | 55.6 | 89.5 | |
| DirectModel=GPT-4.1Base2026.04 | 72 | 65.8 | 54 | 84.4 | |
| DirectModel=1.5-flash-8b2026.04 | 70.8 | 61 | 45.7 | 92 | |
| DirectModel=o4-mini2026.04 | 69.4 | 59 | 44.2 | 88.9 | |
| DirectModel=2.0-flashBase2026.04 | 69.1 | 59 | 44.3 | 88.1 | |
| CARE10Model=LLaMa8b2026.04 | 67.4 | 73.6 | 90.8 | 61.9 | |
| IndirectModel=o4-mini2026.04 | 64.3 | 47.3 | 32.2 | 89.8 | |
| IndirectModel=GPT-4.1Base2026.04 | 64.2 | 47.4 | 32.2 | 89.8 | |
| IndirectModel=LLaMa70bBase2026.04 | 60.5 | 36.8 | 23 | 92 | |
| IndirectModel=2.0-flashBase2026.04 | 59.5 | 34.5 | 21.3 | 90.1 | |
| IndirectModel=1.5-flash-8b2026.04 | 57.9 | 29.4 | 17.5 | 91.4 | |
| IndirectModel=LLaMa8b2026.04 | 56.6 | 44.5 | 34.8 | 61.8 |