Paraphrase Identification on PAWS (test)
94.1AccuracyBART
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BARTTraining Mode=In-distribution2022.10 | 94.1 | 94.1 | |
| RoBERTaTraining Mode=In-distribution2022.10 | 93.6 | 93.5 | |
| BERTTraining Mode=In-distribution2022.10 | 93.2 | 92.7 | |
| GAPXTraining Mode=In-distribution2022.10 | 92.7 | 92.7 | |
| BERT+EPTraining Mode=In-distribution2022.10 | 89.7 | 89.7 | |
| IDPTraining Mode=In-distribution2022.10 | 88.5 | 88.2 | |
| GAPTraining Mode=In-distribution2022.10 | 85.2 | 85.1 | |
| GENICLBackbone=Qwen2.5-3B2025.05 | 81.8 | — | |
| OODPTraining Mode=In-distribution2022.10 | 77.1 | 67.9 | |
| GENICLBackbone=Vicuna-13B2025.05 | 71.7 | — | |
| LLM-RBackbone=Qwen2.5-3B2025.05 | 70.4 | — | |
| GENICLBackbone=LLaMA-3.2-3B2025.05 | 66.2 | — | |
| LLM-RBackbone=Vicuna-13B2025.05 | 64.4 | — | |
| GENICLBackbone=GPT-Neo 2.7B2025.05 | 60.5 | — | |
| Zero-shotBackbone=Qwen2.5-3B2025.05 | 58.8 | — | |
| Zero-shotBackbone=Vicuna-13B2025.05 | 57.2 | — | |
| BOWTraining Mode=In-distribution2022.10 | 56.8 | 48.8 | |
| LLM-RBackbone=LLaMA-3.2-3B2025.05 | 56 | — | |
| LLM-RBackbone=GPT-Neo 2.7B2025.05 | 55.5 | — | |
| Zero-shotBackbone=LLaMA-3.2-3B2025.05 | 54.1 | — | |
| BiLSTMTraining Mode=In-distribution2022.10 | 53 | 43.6 | |
| Zero-shotBackbone=GPT-Neo 2.7B2025.05 | 52.6 | — |