dpr-xlm-align-C3-zhen

DPR Model trained from XLM-Align with C3 pretraining and retrieval finetuned on MSMARCO.

This model is meant to serve English queries and Chinese documents.

Please consider citing the following paper:

@inproceedings{sigir2022c3,
    author = {Eugene Yang and Suraj Nair and Ramraj Chandradevan and Rebecca Iglesias-Flores and Douglas W. Oard},
    title = {C3: Continued Pretraining with Contrastive Weak Supervision for Cross Language Ad-Hoc Retrieval},
    booktitle = {Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) (Short Paper)},
    year = {2022},
    url = {https://arxiv.org/abs/2204.11989}
}