StanfordNLP Docker Image

Author

Teng Fu

Email: teng.fu@teleware.com

Base Image

This is the docker image for NLP task, its baseImage is:

FROM tftwdockerhub/nlp_devel_docker_github:latest

Additional installed packages

Pytorch installed:

PyTorch: 1.0

For NLP packages:

StanfordNLP 0.1.1
Oracle Jave 8
CoreNLP 2018-10-05

CoreNLP installation is inspired by konradstrack/corenlp Java installation is based on java's official dockerfile

All the additional packages are also installed by using Python script in this directory stanfordnlp_model_download.py. Developer can customise the resources that is required in main() function.

There are two Jupyter Notebook in the Repo, provide the demo code for the usage of CoreNLP server with StanfordNLP:

coreNLP_server_start.ipynb for CoreNLP server startup. If developer wish to try the CoreNLP client feature, cmds in this notebook must be run priory to client code.
stanfordNLP_demo.ipynb for StanfordNLP code demo.

Docker Registry Repo

tftwdockerhub/stanfordNLP-gpu:latest

Binder test

Usage

on dsvm-gpu virtual machines

sudo docker pull tftwdockerhub/stanfordNLP-gpu:latest

remember the target port is 8889

sudo nvidia-docker run -it -p 8889:8888 -v \<project-dir-path\>:/app tftwdockerhub/stanfordNLP-gpu:latest

In local browser, remember the target port is 8889 and the token string on CLI screen

http://\<vm-ipaddress-or-dns\>:8889/?token=21c5e12xxxxxx