Relevance Classification on ESCI-JP (test)
69.72AccuracyMPCoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MPCoTModel Category=Teacher Model, Protocol=SFT+DPO2026.01 | 69.72 | 53.69 | |
| Best-SingleCoTModel Category=Teacher Model, Protocol=SFT+DPO2026.01 | 65.39 | 51.14 | |
| MPCoTModel Category=Teacher Model, Protocol=SFT2026.01 | 64.17 | 48.06 | |
| LRKDModel Category=Student Model, Architecture=GAT2026.01 | 63.96 | 47.21 | |
| LRKDModel Category=Student Model, Architecture=Poly2026.01 | 63.44 | 45.27 | |
| CED-KDModel Category=Student Model2026.01 | 62.94 | 43.1 | |
| MKDModel Category=Student Model2026.01 | 62.5 | 43.17 | |
| BERT-multilingual-baseModel Category=Deployable Model Baseline2026.01 | 62.36 | 42.6 | |
| ProgressiveCoTModel Category=Teacher Model, Protocol=SFT2026.01 | 62.33 | 50.01 | |
| LRKDModel Category=Student Model, Architecture=MLP2026.01 | 62.19 | 42.96 | |
| Best-SingleCoTModel Category=Teacher Model, Protocol=SFT2026.01 | 60.11 | 49.97 | |
| Best-SingleCoTModel Category=Teacher Model, Protocol=few-shot2026.01 | 58.72 | 48.9 | |
| MPCoTModel Category=Teacher Model, Protocol=few-shot2026.01 | 55.82 | 45.61 |