Faithfulness Evaluation on CNN/DM
33.2AUPCP-SHAP
Evaluation results
| Method | Links | |
|---|---|---|
| P-SHAPModel=Flan-UL2, Scalarizer=SUMM2024.03 | 33.2 | |
| MExGen C-LIMEModel=Flan-UL2, Scalarizer=SUMM2024.03 | 32.1 | |
| MExGen LOOModel=Flan-UL2, Scalarizer=SUMM2024.03 | 32.1 | |
| MExGen L-SHAPModel=Flan-UL2, Scalarizer=SUMM2024.03 | 32 | |
| MExGen C-LIMEModel=Llama-3-8B-Instruct, Scalarizer=SUMM2024.03 | 26.4 | |
| MExGen L-SHAPModel=Llama-3-8B-Instruct, Scalarizer=SUMM2024.03 | 26.3 | |
| MExGen LOOModel=Llama-3-8B-Instruct, Scalarizer=SUMM2024.03 | 26.1 | |
| P-SHAPModel=Llama-3-8B-Instruct, Scalarizer=SUMM2024.03 | 22.1 | |
| MExGen L-SHAPModel=DistilBART, Scalarizer=Log Probability2024.03 | 14.7 | |
| MExGen C-LIMEModel=DistilBART, Scalarizer=Log Probability2024.03 | 13.5 | |
| MExGen LOOModel=DistilBART, Scalarizer=Log Probability2024.03 | 13.2 | |
| P-SHAPModel=DistilBART, Scalarizer=Log Probability2024.03 | 9.7 |