Fact-Checking on PubHealth
78.66Balanced AccuracyKG-CRAFTL3.3
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| KG-CRAFTL3.3Backbone=L3.32026.01 | 78.66 | 72.82 | 73.89 | 73.35 | |
| Llama3.3 70BBackbone=Llama3.3 70B2025.02 | 76.9 | — | — | — | |
| Llama3.3 70B2025.02 | 76.9 | — | — | — | |
| Qwen2.5 72BBackbone=Qwen2.5 72B2025.02 | 76.7 | — | — | — | |
| Qwen2.5 72B2025.02 | 76.7 | — | — | — | |
| OpenAI o1Scale=>300B2025.02 | 74.8 | — | — | — | |
| Claude 3.5-SonnetScale=>300B2025.02 | 73.8 | — | — | — | |
| GraphCheck-Qwen2.5 72BBackbone=Qwen2.5 72B2025.02 | 73.6 | — | — | — | |
| GraphCheckBackbone=Llama3.3 70B2025.02 | 73.6 | — | — | — | |
| GraphCheckL3.3Backbone=L3.32026.01 | 73.6 | — | — | — | |
| GPT-4Scale=>300B2025.02 | 73.2 | — | — | — | |
| DeepSeek-V3 671BScale=>300B2025.02 | 72.9 | — | — | — | |
| GraphCheck-Llama3.3 70BBackbone=Llama3.3 70B2025.02 | 71.7 | — | — | — | |
| GraphCheckBackbone=Qwen 72B2025.02 | 71.7 | — | — | — | |
| GraphCheckQwen 72BBackbone=Qwen 72B2026.01 | 71.7 | — | — | — | |
| Llama3 8B2025.02 | 70.7 | — | — | — | |
| GPT-4oScale=>300B2025.02 | 67 | — | — | — | |
| MiniCheck2025.02 | 66.3 | — | — | — | |
| MiniCheck2025.02 | 66.3 | — | — | — | |
| AlignScore2025.02 | 64.4 | — | — | — | |
| AlignScore2025.02 | 64.4 | — | — | — | |
| GraphEval2025.02 | 63.7 | — | — | — | |
| GraphEval2025.02 | 63.7 | — | — | — | |
| ACUEval2025.02 | 62.9 | — | — | — | |
| KG-CRAFTC3.5Backbone=C3.52026.01 | 61.02 | 76.5 | 75.18 | 75.83 | |
| Qwen2.5 7B2025.02 | 59.1 | — | — | — |