Fact-checking on AdvFact
64AccuracyToE
Evaluation Results
| Method | Links | |
|---|---|---|
| ToEBackbone=DeepSeek-V3.22026.06 | 64 | |
| ToEBackbone=GPT-OSS-20B2026.06 | 54 | |
| AdsentBackbone=GPT-OSS-20B2026.06 | 52 | |
| TELLERBackbone=DeepSeek-V3.22026.06 | 40 | |
| ZCoTBackbone=DeepSeek-V3.22026.06 | 36 | |
| TELLERBackbone=GPT-OSS-20B2026.06 | 32 | |
| AdsentBackbone=DeepSeek-V3.22026.06 | 30 | |
| DefGenBackbone=DeepSeek-V3.22026.06 | 28 | |
| DefGenBackbone=GPT-OSS-20B2026.06 | 24 | |
| ZCoTBackbone=GPT-OSS-20B2026.06 | 20 | |
| DirectBackbone=DeepSeek-V3.22026.06 | 16 | |
| STEELBackbone=DeepSeek-V3.22026.06 | 16 | |
| STEELBackbone=GPT-OSS-20B2026.06 | 14 | |
| AFacTABackbone=DeepSeek-V3.22026.06 | 12 | |
| AFacTABackbone=GPT-OSS-20B2026.06 | 12 | |
| DirectBackbone=GPT-OSS-20B2026.06 | 8 |