Skip to content

Code for Label Semantics for Few Shot Named Entity Recognition

Notifications You must be signed in to change notification settings

TianWuYuJiangHenShou/LabelSemantics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

21 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

LabelSemantics

Code for Label Semantics for Few Shot Named Entity Recognition

经过修改,现在的方案已与原paper中思路一样,采用两个独立Bert Encoder分别对token和label进行编码

关于数据集,内部数据不可共享,别再来要数据了!!!。建议使用resume(网上有)以及其他开源数据做个简单实验

代码结构如下

├── fewshot.json
├── LabelSemantics_fewshot.py
    ├── finetune in target dataset
├── LabelSemantics.py
    ├── pretrain in source dataset
├── pretrain.json

实验结果如下

我们在自己的数据集上做了实验,基本效果如下。

K-Shots 5-shot 20-shot 50-shot 100-shot 200-shot 500-shot Full dataset
F1 0.53 0.61 0.64 0.67 0.71 0.76 0.89
epoch 100 100 100 100 100 100 100

About

Code for Label Semantics for Few Shot Named Entity Recognition

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages