Easy to use BiLSTM+CRF sequence tagging for text.
Project description
https://pypi.org/project/seqtag/
BiLSTM + CRF for sequence tagging
This is adapted from guillaumegenthial 's original implementation and is made configurable and easy to adapt and use.
Requirements:
This code is tested with all tensorflow versions from 1.3.0 to 1.10.0.
tensorflow is not included in setup.py since, it will remove tensorflow-gpu. Separately install tensorflow by following https://www.tensorflow.org/install/ for this module to work.
Download the 300 dimnesional glove vectors from https://nlp.stanford.edu/projects/glove/
Installation:
The stable version can be installed by running "pip install seqtag"
How to Use:
Create a training directory with train.txt and valid.txt (test.txt is optional) set the config parameters as expected in a configuration file.
An example of the configuration file can be found at https://github.com/bedapudi6788/seqtag/blob/master/example_config.json
Training:
from seqtag import trainer
trainer.train(config_path = 'path_to_config_json')
Running Predictions:
from seqtag import predictor
model = predictor.load_model(path_to_config.json)
predictor.predict(model, ['I', 'am', 'Batman'])
['O', 'O', 'B-PER']
For an usage example take a look at https://github.com/bedapudi6788/Deep-Segmentation/ . seqtag is used for sentence segmentation in this repo.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for seqtag-1.0.8-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 72e54add2b33f53878215291d5ab11a2db726aaf46ba300d635ea7e304c28940 |
|
MD5 | ab65a7ebe4a3bf95b65178f18ec20677 |
|
BLAKE2b-256 | da2af354f6aeca37ceebc347d1cd02fd6a861fdb38c09e6a12dcd786de8c779b |