Ranking on TREC Deep Learning 2019
73NDCG@10R2ANKER
Evaluation results
| Method | Links | ||
|---|---|---|---|
| R2ANKEREvaluation Protocol=BM25-reranking2022.06 | 73 | — | |
| RobustMaskBase Model=PairLM (BERT-Base), Mask Ratio=30%2025.12 | 71.7 | 84.5 | |
| RobustMaskBase Model=PairLM (BERT-Base), Mask Ratio=60%2025.12 | 71.5 | 83.6 | |
| RocketQAv2Evaluation Protocol=BM25-reranking2022.06 | 71.4 | — | |
| PairLM (BERT-Base)Backbone=BERT-Base, Type=Pairwise ranking2025.12 | 71.2 | 85.7 | |
| BERT-Base2025.12 | 71 | 87.1 | |
| BGEModel Version=BGE-M32025.12 | 67.9 | 85.9 | |
| BERT-Base+PGDTraining Method=Adversarial Training, Attack Algorithm=PGD2025.12 | 66.8 | 82.6 | |
| CertDRBase Model=PairLM (BERT-Base)2025.12 | 66.1 | 77.4 | |
| TKEncoder=Shallow Transformer2025.12 | 65.2 | 75.1 | |
| CertDRBase Model=PairLM (BERT-Base), Data Augmentation=None2025.12 | 54.2 | 66.7 | |
| BM252025.12 | 48.7 | 68.5 |