Factual Consistency Evaluation on QAGS-XSum (Spearman correlation)
57.4Spearman CorrelationALIGN-large
Evaluation Results
| Method | Links | |
|---|---|---|
| ALIGN-largeModel Parameters=355M2023.07 | 57.4 | |
| ALIGNModel size=large2023.07 | 57.4 | |
| ALIGNSCORE-largeType=Ours2023.05 | 57.2 | |
| G-EVAL-4Model ID=GPT42023.07 | 53.7 | |
| ALIGN-baseModel Parameters=125M2023.07 | 52.7 | |
| ALIGNModel size=base2023.07 | 52.7 | |
| ALIGNSCORE-baseType=Ours2023.05 | 51.9 | |
| UniEvalType=Misc2023.05 | 50 | |
| UniEval2023.07 | 50 | |
| SummaC-CONVType=NLI2023.05 | 45 | |
| QAFactEvalType=QA2023.05 | 44.1 | |
| QAFactEval2023.07 | 44.1 | |
| SummaC-ZSType=NLI2023.05 | 43.7 | |
| G-EVAL-3.5Model ID=GPT3.5-d032023.07 | 40.6 | |
| DAEType=NLI2023.05 | 37.5 | |
| NER-OverlapType=Similarity Matching2023.05 | 31.2 | |
| CTCType=Misc2023.05 | 30.6 | |
| CTC2023.07 | 30.6 | |
| FactCCType=Misc2023.05 | 28.8 | |
| GPTScoreModel ID=GPT3.5-d032023.07 | 22 | |
| ROUGE-1Type=Similarity Matching2023.05 | 18.1 | |
| ROUGE-2Type=Similarity Matching2023.05 | 17.7 | |
| BARTScoreType=Misc2023.05 | 17 | |
| BARTScore2023.07 | 17 | |
| ROUGE-LType=Similarity Matching2023.05 | 16.5 | |
| BLEURTType=Regression2023.05 | 12.4 | |
| BLEURT2023.07 | 12.4 | |
| QuestEvalType=QA2023.05 | 11.9 | |
| SimCSEType=Similarity Matching2023.05 | 11.2 | |
| BLEUType=Similarity Matching2023.05 | 10.1 | |
| BLANCType=Misc2023.05 | 1.6 | |
| MNLIType=NLI2023.05 | 0.7 | |
| ALIGNSCORE-largeBackbone=ROBERTa (355M)2023.05 | 0.572 | |
| G-EVAL-4Backbone=GPT42023.05 | 0.537 | |
| ALIGNSCORE-baseBackbone=ROBERTa (125M)2023.05 | 0.519 | |
| G-EVAL-3.5Backbone=GPT3.5-d032023.05 | 0.406 | |
| FEQAType=QA2023.05 | -6.5 | |
| BERTScoreType=Similarity Matching2023.05 | -8.9 | |
| BERTScore2023.07 | -8.9 |