Edit model card


NER-PMR-large is initialized with PMR-large and further fine-tuned with 4 NER training data, namely CoNLL, WNUT17, ACE2004, and ACE2005.

The model performance on the test sets are:

CoNLL WNUT17 ACE2004 ACE2005
RoBERTa-large (single-task model) 92.8 57.1 86.3 87.0
PMR-large (single-task model) 93.6 60.8 87.5 87.4
NER-PMR-large (multi-task model) 92.9 54.7 87.8 88.4

Note that the performance of RoBERTa-large and PMR-large are single-task fine-tuning, while NER-PMR-large is a multi-task fine-tuned model. As it is fine-tuned on multiple datasets, we believe that NER-PMR-large has a better generalization capability to other NER tasks than PMR-large and RoBERTa-large.

How to use

You can try the codes from this repo for both training and inference.

BibTeX entry and citation info

  title={From Clozing to Comprehending: Retrofitting Pre-trained Language Model to Pre-trained Machine Reader},
  author={Xu, Weiwen and Li, Xin and Zhang, Wenxuan and Zhou, Meng and Bing, Lidong and Lam, Wai and Si, Luo},
  journal={arXiv preprint arXiv:2212.04755},
Downloads last month