Claim Verification on ChatGPT (test)
82.4Verification ConfidenceDYDECOMP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DYDECOMPAtomicity=2, Decompose Policy=In-Context Example, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 82.4 | 75 | |
| DYDECOMPAtomicity=1, Decompose Policy=In-Context Example, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 80.9 | 60.9 | |
| DYDECOMPAtomicity=0, Decompose Policy=In-Context Example, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 71.5 | 42.8 | |
| FActScoreAtomicity=0, Decompose Policy=In-Context Example, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 67.7 | 45.4 | |
| DYDECOMPAtomicity=1, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 65.4 | 75.8 | |
| DYDECOMPAtomicity=2, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 64.4 | 65.2 | |
| FActScoreAtomicity=0, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 62.7 | 66.6 | |
| FActScore AtomAtomicity=0, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 61.8 | 79.6 | |
| DYDECOMPAtomicity=0, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 60 | 78.9 | |
| R-NDAtomicity=0, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 55.6 | 44.9 | |
| WICEAtomicity=0, Decompose Policy=Retrieval, Decomposition LLM=Llama3-Inst-70B, Verification LLM=Llama3-Inst-8B2025.03 | 43.1 | 78.2 |