Question Answering on ACE-whQA
79.02EMGPT-3.5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-3.5Verifier=true2023.07 | 79.02 | 80.91 | 0.84 | |
| GPT-3.5 + ALIGN VerifierVerifier=ALIGN, Unanswerable threshold selection=SQuAD v2 validation split2023.07 | 79.02 | 80.91 | 0.84 | |
| FLAN T5Verifier=true2023.07 | 75.75 | 77.6 | 0.9 | |
| FLAN T5 + ALIGN VerifierVerifier=ALIGN, Unanswerable threshold selection=SQuAD v2 validation split2023.07 | 75.75 | 77.6 | 0.9 | |
| GPT-3.5Verifier=false2023.07 | 67.98 | 71.98 | 0.77 | |
| GPT-3.5mode=Zero-shot2023.07 | 67.98 | 71.98 | 0.77 | |
| ElectraVerifier=false2023.07 | 52.32 | 55.59 | 0.87 | |
| Flan T5Verifier=false2023.07 | 26.29 | 29.24 | 0.51 | |
| FLAN T5mode=Zero-shot2023.07 | 26.29 | 29.24 | 0.51 |