Text Classification on ANLI, MNLI, QNLI, WNLI, RTE, MRPC (val)
43.1ANLI AccuracyT5-base (Fine-Tuned)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| T5-base (Fine-Tuned)Model=T5-base, Model Size=220M, Data=Full, Method=Fine-Tuned2024.03 | 43.1 | 86.6 | 93.7 | 78.8 | 80.1 | 87.5 | 78.3 | |
| LLaMA2Model Size=7B, Method=Few-shot2024.03 | 36 | 41.5 | 55.3 | 53.5 | 62.4 | 65.4 | 52.3 | |
| GOLDModel=LLaMA2 -> T5-base, Model Size=220M, Data=3K2024.03 | 35.7 | 62.5 | 91.7 | 57.7 | 69.6 | 85 | 67.1 | |
| ZeroGenModel=LLaMA2 -> T5-base, Model Size=220M, Data=3K2024.03 | 34.6 | 56.1 | 88.5 | 54.9 | 62.1 | 84.3 | 63.4 | |
| P2ModelModel=LLaMA2 -> T5-base, Model Size=220M, Data=3K2024.03 | 34.4 | 59.5 | 62.2 | 56.3 | 58.8 | 75 | 57.7 | |
| ProGenModel=LLaMA2 -> T5-base, Model Size=220M, Data=3K2024.03 | 34.3 | 55.1 | 85.9 | 57.7 | 66 | 80.3 | 63.2 | |
| T5-base (Pre-Trained)Model=T5-base, Model Size=220M, Data=Full, Method=Pre-Trained2024.03 | 29 | 56.6 | 88.3 | 52.1 | 68.5 | 75 | 61.6 |