Open-domain Question Answering on COMPMIX (test)
56.5Exact MatchSPAGHETTI
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SPAGHETTIBackbone=GPT-42024.06 | 56.5 | 70 | 81.9 | |
| SPAGHETTIBackbone=GPT-3.52024.06 | 55.6 | 65.6 | 75.3 | |
| GPT-4Zero-shot=true, Engine=GPT-42024.06 | 52.8 | 62 | 78.4 | |
| SPAGHETTIBackbone=LLaMA-7B2024.06 | 51.7 | 60.5 | 70.4 | |
| GPT-3 (text-davinci-003)Engine=text-davinci-0032024.06 | 50.2 | — | — | |
| EXPLAINGNN2024.06 | 44.2 | — | — | |
| UniK-QA2024.06 | 44 | — | — | |
| CONVINSE2024.06 | 40.7 | — | — | |
| GPT-3.5 (turbo-instruct)Zero-shot=true, Engine=turbo-instruct2024.06 | 36.1 | 54.2 | 69.9 |