Long-form Question Answering on ALCE LFQA
38.6ROUGE-LATTR. FIRST_CoT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ATTR. FIRST_CoTEvaluation Protocol=ICL, Chain-of-Thought=true2024.03 | 38.6 | 90.7 | 89.3 | 48.2 | 0 | |
| ATTR. FIRSTEvaluation Protocol=ICL2024.03 | 35.8 | 90.5 | 78.7 | 65.2 | 0 | |
| ALCEEvaluation Protocol=ICL2024.03 | 35.2 | 89.9 | 49.8 | 2,153.3 | 26.9 | |
| GEMINIEvaluation Protocol=ICL2024.03 | 33.1 | 89.6 | — | — | — | |
| PRIMERAEvaluation Protocol=FT2024.03 | 32.2 | 88.8 | — | — | — | |
| ATTR. FIRST_jointEvaluation Protocol=FT, Training=joint2024.03 | 27 | 88.1 | 44.9 | 33.1 | 0 | |
| ATTR. FIRSTEvaluation Protocol=FT2024.03 | 24.7 | 88 | 52.7 | 20.9 | 0 |