mirrors
/
mmocr
mirror of https://github.com/open-mmlab/mmocr.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
mmocr/configs/ner/bert_softmax
History
Tong Gao 4f7270e574
Fix #282: Support init_cfg & update depreciated configs (#365) 
* update coco ref

* init_cfg for dbnet

* initcfg for mask_rcnn

* textsnake init_cfg

* fix dbnet

* panet initcfg

* psenet initcfg

* fcenet initcfg

* drrg initcfg

* add init_cfg to detectors

* update maskrcnn config file to support mmdet

* fix init_cfg of fce_head

* crnn initcfg

* init_weights in training

* nrtr initcfg

* robust_scanner initcfg

* sar init_cfg

* seg init_cfg

* tps_crnn init_cfg

* sdmgr initcfg

* ner init_cfg

* fix textsnake

* sdmgr initcfg

* move "pretrained" to "init_cfg" for config files

* Moduleslist update

* fix seg

* ner init_cfg

* fix base

* fix encode decode recognizer

* revert dbnet config

* fix crnn

* fix base.py

* fix robust_scanner

* fix panet

* fix test

* remove redundant init_weights() in fcehead

* clean up

* relex mmdet version in workflow

* Add dependency version check

* Update mmocr/models/textdet/dense_heads/pse_head.py

Co-authored-by: Hongbin Sun <hongbin306@gmail.com>

Co-authored-by: Hongbin Sun <hongbin306@gmail.com>
 		2021-07-20 23:18:25 +08:00
..
README.md Ner task (#148) 2021-05-18 11:33:51 +08:00
bert_softmax_cluener_18e.py Fix #282: Support init_cfg & update depreciated configs (#365) 2021-07-20 23:18:25 +08:00

README.md

Chinese Named Entity Recognition using BERT + Softmax

Introduction

[ALGORITHM]

@article{devlin2018bert,
  title={Bert: Pre-training of deep bidirectional transformers for language understanding},
  author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
  journal={arXiv preprint arXiv:1810.04805},
  year={2018}
}

Dataset

Train Dataset

trainset text_num entity_num
CLUENER2020 10748 23338

Test Dataset

testset text_num entity_num
CLUENER2020 1343 2982

Results and models

Method Pretrain Precision Recall F1-Score Download
bert_softmax pretrain 0.7885 0.7998 0.7941 model | log