Natural language inference on CB (test)
89.3AccuracyALIGN
Evaluation Results
| Method | Links | |
|---|---|---|
| ALIGNModel Variant=large, Parameters=355M, Evaluation Protocol=Zero-shot2023.07 | 89.3 | |
| FLAN-T5Model Variant=large, Parameters=780M, Evaluation Protocol=Zero-shot2023.07 | 87.5 | |
| FLAN-T5Model Variant=base, Parameters=250M, Evaluation Protocol=Zero-shot2023.07 | 82.1 | |
| ALIGNModel Variant=base, Parameters=125M, Evaluation Protocol=Zero-shot2023.07 | 76.8 | |
| Coherence Boosting (GPT-2 XL)alpha=-1.92021.10 | 66.07 | |
| GPT-2 XL (1.6B)alpha=-12021.10 | 51.79 | |
| Coherence Boosting (GPT-2 Small)alpha=-2.42021.10 | 48.21 | |
| GPT-2 XL (1.6B)alpha=02021.10 | 30.36 | |
| Coherence Boosting (GPT-3 175B)alpha=-1.912021.10 | 28.57 | |
| GPT-3 175Balpha=-12021.10 | 25 | |
| GPT-2 Small (125M)alpha=-12021.10 | 23.21 | |
| GPT-2 Small (125M)alpha=02021.10 | 12.5 | |
| GPT-3 175Balpha=02021.10 | 5.36 |