Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,37 @@
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
4 |
+
# From Clozing to Comprehending: Retrofitting Pre-trained Masked Language Model to Pre-trained Machine Reader
|
5 |
+
Pre-trained Machine Reader (PMR) is pre-trained with 18 million Machine Reading Comprehension (MRC) examples constructed with Wikipedia Hyperlinks.
|
6 |
+
It was introduced in the paper From Clozing to Comprehending: Retrofitting Pre-trained Masked Language Model to Pre-trained Machine Reader by
|
7 |
+
Weiwen Xu, Xin Li, Wenxuan Zhang, Meng Zhou, Wai Lam, Luo Si, Lidong Bing
|
8 |
+
and first released in [this repository](https://github.com/DAMO-NLP-SG/PMR).
|
9 |
+
|
10 |
+
|
11 |
+
|
12 |
+
## Model description
|
13 |
+
This model is initialized with [PMR-large](https://huggingface.co/DAMO-NLP-SG/PMR-large) and further fine-tuned with 4 NER training data, namely [CoNLL](https://huggingface.co/datasets/conll2003), [WNUT17](https://huggingface.co/datasets/wnut_17), [ACE2004](https://paperswithcode.com/sota/nested-named-entity-recognition-on-ace-2004), and [ACE2005](https://paperswithcode.com/sota/nested-named-entity-recognition-on-ace-2005).
|
14 |
+
|
15 |
+
The model performance on the test sets are:
|
16 |
+
|
17 |
+
|| CoNLL | WNUT17 | ACE2004 | ACE2005 |
|
18 |
+
|--|------------|-----------|----------|--|
|
19 |
+
|RoBERTa-large (single-task model)| 92.8 | 57.1 | 86.3|87.0|
|
20 |
+
|PMR-large (single-task model)| 93.6 | 60.8 | 87.5 | 87.4|
|
21 |
+
|NER-PMR-large (multi-task model)| 92.9 | 54.7| 87.8| 88.4|
|
22 |
+
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.
|
23 |
+
|
24 |
+
### How to use
|
25 |
+
You can try the codes from [this repo](https://github.com/DAMO-NLP-SG/PMR/NER).
|
26 |
+
|
27 |
+
|
28 |
+
|
29 |
+
### BibTeX entry and citation info
|
30 |
+
```bibtxt
|
31 |
+
@article{xu2022clozing,
|
32 |
+
title={From Clozing to Comprehending: Retrofitting Pre-trained Language Model to Pre-trained Machine Reader},
|
33 |
+
author={Xu, Weiwen and Li, Xin and Zhang, Wenxuan and Zhou, Meng and Bing, Lidong and Lam, Wai and Si, Luo},
|
34 |
+
journal={arXiv preprint arXiv:2212.04755},
|
35 |
+
year={2022}
|
36 |
+
}
|
37 |
+
```
|