Semantic Textual Similarity on SemEval-2022 Task 2 Idiomatic STS (evaluation)
0.69Spearman Rho (All)Adaptive Contrastive Triplet Loss
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Adaptive Contrastive Triplet LossModel Size=558M2024.06 | 0.69 | — | — | 0.548 | 0.716 | |
| YNU-HPCCModel Size=183M2024.06 | 0.665 | — | — | 0.428 | 0.664 | |
| Liu et al.Supervision Setting=Supervised2023.06 | 0.6648 | 0.4277 | 0.6637 | — | — | |
| OURS-ave3Supervision Setting=Unsupervised2023.06 | 0.6613 | 0.4268 | 0.8691 | — | — | |
| OURSSupervision Setting=Unsupervised2023.06 | 0.6531 | 0.4265 | 0.8691 | — | — | |
| PhelpsSupervision Setting=Supervised2023.06 | 0.6504 | 0.4124 | 0.8188 | — | — | |
| drsphelpsModel Size=420M2024.06 | 0.65 | — | — | 0.412 | 0.819 | |
| PhelpsSupervision Setting=Unsupervised2023.06 | 0.6402 | 0.403 | 0.8641 | — | — | |
| Sem-BaseSupervision Setting=Supervised2023.06 | 0.5951 | 0.399 | 0.5961 | — | — | |
| baselineModel Size=110M2024.06 | 0.595 | — | — | 0.399 | 0.596 | |
| E5 largeModel Size=334M2024.06 | 0.514 | — | — | 0.252 | 0.807 | |
| Sem-BaseSupervision Setting=Unsupervised2023.06 | 0.481 | 0.2263 | 0.8311 | — | — | |
| GTE largeModel Size=334M2024.06 | 0.465 | — | — | 0.236 | 0.806 | |
| LLama2Model Size=13,000M2024.06 | 0.399 | — | — | 0.171 | 0.486 |