PrimeQA/open-nq-colbert-xlmr-large

Model Description

This is a retriever model based on ColBERT v2 with bert-base-uncased language model.
This model was trained with the OpenNQ data.
The architecture of the model and hyper parameters are described in the paper ‘Relevance-guided Supervision for OpenQA with ColBERT’.

Intended uses & limitations

This model uses the xlm-roberta-large LM. Biases associated with the pre-trained language model we used may be present in this ColBERT v2 model.

Usage

This model can be used with PrimeQA’s ColBERT engine.

BibTeX entry and citation info

@article{Khattab2021RelevanceguidedSF,
  title={Relevance-guided Supervision for OpenQA with ColBERT},
  author={O. Khattab and Christopher Potts and Matei A. Zaharia},
  journal={Transactions of the Association for Computational Linguistics},
  year={2021},
}

@article{Lee2019LatentRF,
  title={Latent Retrieval for Weakly Supervised Open Domain Question Answering},
  author={Kenton Lee and Ming-Wei Chang and Kristina Toutanova},
  journal={ACL},
  year={2019}
}