Open-domain question answering on Natural Questions (Macro-averaged EM)
11.4Macro EMGPT-3 175B
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-3 175B#Params=500x, C (Text Corpus)=false, Zero-shot=true2022.12 | 11.4 | |
| BM25 + GPT-3 13B#Params=37x, C (Text Corpus)=true, Zero-shot=true2022.12 | 11.2 | |
| BM25 + T5 3B#Params=8.5x, C (Text Corpus)=true, Zero-shot=true2022.12 | 10.8 | |
| NPM#Params=1.0x, C (Text Corpus)=true, Zero-shot=true2022.12 | 10.8 | |
| BM25 + OPT 6.7B#Params=19x, C (Text Corpus)=true, Zero-shot=true2022.12 | 10.7 | |
| BM25 + OPT 13B#Params=37x, C (Text Corpus)=true, Zero-shot=true2022.12 | 10.7 | |
| BM25 + GPT-3 6.7B#Params=19x, C (Text Corpus)=true, Zero-shot=true2022.12 | 8.8 | |
| BM25 + OPT 2.7B#Params=7.6x, C (Text Corpus)=true, Zero-shot=true2022.12 | 8.3 | |
| BM25 + GPT-3 175B#Params=500x, C (Text Corpus)=true, Zero-shot=true2022.12 | 6.4 | |
| BM25 + GPT-3 2.7B#Params=7.6x, C (Text Corpus)=true, Zero-shot=true2022.12 | 6.1 | |
| BM25 + T5#Params=2.2x, C (Text Corpus)=true, Zero-shot=true2022.12 | 5.2 | |
| GPT-3 13B#Params=37x, C (Text Corpus)=false, Zero-shot=true2022.12 | 5.2 | |
| OPT 6.7B#Params=19x, C (Text Corpus)=false, Zero-shot=true2022.12 | 4.2 | |
| OPT 13B#Params=37x, C (Text Corpus)=false, Zero-shot=true2022.12 | 4.2 | |
| GPT-3 6.7B#Params=19x, C (Text Corpus)=false, Zero-shot=true2022.12 | 3.1 | |
| OPT 2.7B#Params=7.6x, C (Text Corpus)=false, Zero-shot=true2022.12 | 2.1 | |
| T5 3B#Params=8.5x, C (Text Corpus)=false, Zero-shot=true2022.12 | 2 | |
| GPT-3 2.7B#Params=7.6x, C (Text Corpus)=false, Zero-shot=true2022.12 | 1.1 | |
| T5#Params=2.2x, C (Text Corpus)=false, Zero-shot=true2022.12 | 0.5 |