This model is DPR trained on MS MARCO. The training details and evaluation results are as follows:
| Model | Pretrain Model | Train w/ Marco Title | Marco Dev MRR@10 | BEIR Avg NDCG@10 |
|---|---|---|---|---|
| DPR | bert-base-uncased | w/ | 32.4 | 35.5 |
| BERI Dataset | NDCG@10 |
|---|---|
| TREC-COVID | 58.8 |
| NFCorpus | 23.4 |
| FiQA | 20.6 |
| ArguAna | 39.4 |
| Touché-2020 | 22.3 |
| Quora | 78.0 |
| SCIDOCS | 11.9 |
| SciFact | 49.4 |
| NQ | 43.9 |
| HotpotQA | 45.3 |
| Signal-1M | 20.2 |
| TREC-NEWS | 31.8 |
| DBPedia-entity | 28.7 |
| Fever | 65.0 |
| Climate-Fever | 14.9 |
| BioASQ | 24.1 |
| Robust04 | 32.3 |
| CQADupStack | 28.3 |
The implementation is the same as our EMNLP 2022 paper "Reduce Catastrophic Forgetting of Dense Retrieval Training with Teleportation Negatives". The associated GitHub repository is available at https://github.com/OpenMatch/ANCE-Tele.
@inproceedings{sun2022ancetele,
title={Reduce Catastrophic Forgetting of Dense Retrieval Training with Teleportation Negatives},
author={Si, Sun and Chenyan, Xiong and Yue, Yu and Arnold, Overwijk and Zhiyuan, Liu and Jie, Bao},
booktitle={Proceedings of EMNLP 2022},
year={2022}
}
