Natural Language Inference on MultiNLI matched/mismatched
92.6AccuracyTuring NLR v5 XXL
Evaluation Results
| Method | Links | |
|---|---|---|
| Turing NLR v5 XXLYear=2021, Parameters=5.4B2024.12 | 92.6 | |
| UnitedSynT5Year=2024, Parameters=3B2024.12 | 92.6 | |
| GPT-4Year=2024, Parameters=1.7T2024.12 | 91.8 | |
| Claude 3Year=2024, Parameters=400B2024.12 | 91.2 | |
| DeBERTa-v3-largeYear=2021, Parameters=375M2024.12 | 91.1 | |
| PaLM-2Year=2023, Parameters=340B*2024.12 | 90.8 | |
| RoBERTa-largeYear=2019, Parameters=355M2024.12 | 90.2 | |
| GPT-3.5Year=2022, Parameters=175B2024.12 | 90.2 | |
| UnitedSynT5Year=2024, Parameters=335M2024.12 | 89.8 | |
| BERT-largeYear=2018, Parameters=340M2024.12 | 86.7 |