Faithfulness evaluation on FRANK
0.841Pearson CorrQAFactEval
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| QAFactEvalDirection=QA-based2024.07 | 0.841 | — | 0.813 | 0.933 | |
| G-EvalDirection=LLM-based, Backbone=GPT-42024.07 | 0.841 | — | 0.834 | 0.95 | |
| FineSurEDirection=LLM-based, Backbone=GPT-42024.07 | 0.833 | 86.4 | 0.839 | 0.95 | |
| SummaC-ConvDirection=NLI-based2024.07 | 0.828 | — | 0.814 | 0.883 | |
| UniEvalDirection=QA-based2024.07 | 0.743 | — | 0.772 | 0.983 | |
| BARTScoreDirection=Similarity-based2024.07 | 0.717 | — | 0.736 | 0.937 | |
| ROUGE-2Direction=Similarity-based2024.07 | 0.384 | — | 0.315 | 0.947 | |
| ROUGE-1Direction=Similarity-based2024.07 | 0.324 | — | 0.332 | 0.883 | |
| ROUGE-LDirection=Similarity-based2024.07 | 0.175 | — | 0.18 | 0.667 | |
| BERTScoreDirection=Similarity-based2024.07 | 0.008 | — | 0 | -0.133 |