Light

nooralahzadeh / dsner-pytorch Goto Github PK

View Code? Open in Web Editor NEW

27.0 3.0 8.0 7.33 MB

Pytorch implementation of :https://github.com/rainarch/DSNER

Python 100.00%

dsner-pytorch's Introduction

DNSER-pytorch

Pytorch implementation of :https://github.com/rainarch/DSNER

python 3.6

pytorch

First: python script/preprocess-EC

main module: python dsner.py

args parameters should be define based on the paper

Result: Dev and Test respectively

Model	Setup	epoch	P	R	F1	P	R	F1
LSTM+CRF	H	562	66.30 %	64.21 %	65.24 %	63.64 %	62.53 %	63.08 %
LSTM+CRF	H+A	769	67.80 %	58.53 %	62.82 %	63.65 %	53.59 %	58.19 %
LSTm+CRF+SL	H+A	538	68.21 %	61.89 %	64.90 %	66.90 %	61.66 %	64.17 %
LSTm+CRF+PA	H+A	652	62.21 %	70.11 %	65.93 %	61.12 %	67.63 %	64.21 %
LSTm+CRF+PA+SL	H+A	370	69.12 %	71.16 %	70.12 %	63.46 %	63.94 %	63.70 %

Another paper which is used RL as a post-Processing for denoising with a different reward formula:

Reinforcement-based denoising of distantly supervised NER with partial annotation : https://www.aclweb.org/anthology/D19-6125/

dsner-pytorch's People

Contributors

Stargazers

Watchers

Forkers

rainarch 90217 thuqinyj16 zhufz xidaline mylv1222 greitzmann zbbwss

dsner-pytorch's Issues

how to run lstm+crf+sl or lstm+crf

if I want to run lstm +crf+sl or lstm +crf ，how to run it. Whether I only use ds_fa data, right? Do I need to change other code.

ner.embed.pth

What's the file: 'ner.embed.pth '. Is it generated by word2vec. The format of it is not like char + vec

NEWS Dataset with low performance

Hello, when I used the NEWS dataset, I got F1 about： 73.
However, In the dataset, it is 79.22.
Have you tried NEWS dataset in your code?

When I change setup to "A+H" and mode to "PA+SL" to run LSTm+CRF+PA+SL model, the cpu memory runs out quickly.

The machine's cpu memory is 57G and GPU is 11G.

Python 3.6.4 :: Anaconda, Inc.

torch 1.3.0+cu92

I don't know why it consumes so much cpu memory. Could you help me about this problem?

Some question about the value of action product by policy network

Hello, after 15 rounds of running your code, I found that the value of a was approaching 1, which led to the selection of all distant supervision data. This is obviously unreasonable, is there any way to solve it?

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.