deepsequence

version number: 1.0.1
author: Decheng Yang

Overview

deep sequence labeling toolkit

Add new features:

1. support bilstm+crf
2. support bert
3. support character level feature: char cnn or char lstm
4. support pos tag feature
5. support user defined domain dictionary (e.g. company)
6. support “BIOLU” format for sequence evaluation
7. easy to customize model by config file, see examples/ner_news_train.py

Installation / Usage

To install use pip:

$ pip install deepsequence

Or clone the repo:

$ git clone http://git.xxxx.com/deepsequence.git
$ python setup.py install

Contributing

TBD

Example

train ner model for news:

see detail in /examples/ner_news_train.py:

use bilstm:
python ner_news_train.py --config bilstm_parameters.json

use bert:
python ner_news_train.py --config bert_parameters.json

TBD