Multilabel Text Classifier Web Frontend
401
This is a web frontend for users to query a multilabel text classifier to classify texts they input. The classifier backend engine can be either the BERT-based multi-label text classification engine or fastText-based multi-label text classification engine.
You can either build the docker image from the source code or pull the docker images directly from the Docker Hub
You can build the docker image from the source code by running the following command in the project directory as follows:
docker build -t yamai/multilabel-classifier-web .
Without building the docker image from the source code, you can pull the docker image from the Docker Hub as follows:
docker pull yamai/multilabel-classifier-web:latest
You need to create a docker network (e.g., mlc-net) for connecting this Multilabel Text Classifier Web Frontend and the text classifier backend engine as follows:
docker network create mlc-net
Run the text classifier backend engine, e.g., fast-text-multilabel-classifier, on the created docker network:
docker run -v $MODEL_DIR:/model -p 8000:8000 --name backend --network mlc-net yamai/fasttext-multilabel-classifier:serve-latest
Run the Multilabel Text Classifier Web Frontend, on the created docker network:
docker run -e "MLC_ENDPOINT=http://backend:8000/classifier" -e "MLC_HOST_PORT=0.0.0.0:5000" -p 5000:5000 --name frontend --network mlc-net -it yamai/multilabel-classifier-web
You may change the default settings by mapping your customized settings.py in the host (e.g., under /home/mlc/) to /web/settings.py in the container as follows:
docker run -e "MLC_ENDPOINT=http://backend:8000/classifier" -e "MLC_HOST_PORT=0.0.0.0:5000" -p 5000:5000 --name frontend --network mlc-net -v /home/mlc/settings.py:/web/settings.py -it yamai/multilabel-classifier-web
These settings are as follows:
THRESHOLD: the label of which the likelihood is above this threshold will be shown in bold, e.g., 0.5.ENDPOINT: the default endpoint of the multilabel text classifier backend, e.g., http://0.0.0.0:8000/classifier.HOST_PORT: the default host and port for serving this web frontend, e.g., 0.0.0.0:5000.TITLE: the title shown on this web frontend.With the above configuration, you may open the URL https://localhost:5000 on a browser to access the Multilabel Text Classifier Web Frontend.
If you need any supporting resources or consultancy services from YAM AI Machinery, please find us at:
Content type
Image
Digest
Size
38.6 MB
Last updated
about 7 years ago
docker pull yamai/multilabel-classifier-web