simple-tf-use

Simple-TensorFlow-Serving powered Universal Sentence Encoder [Multilingual, Large]

---

Install.

Create a virtual environment or install requirements to your virtual environment. You could pip install -r requirements.txt, but dependencies basically are:

tensorflow==1.13.1 tf-sentencepiece==tf-sentencepiece==0.1.83

Build.

build.py should handle everything:

python build.py

Re-build the container:

cd simple_tensorflow_serving && docker build . --tag stefanondisponibile/simple-tf-use:latest

Run the container:

docker run -dp 8501:8501 stefanondisponibile/simple-tf-use:latest

Test it:

curl -H "Content-Type: application/json" -X POST -d '{"model_name": "default", "model_version": "001", "data": { "text": ["Some text."] }}' http://localhost:8501

Looking good? Push it to DockerHUB:

docker push stefanondisponibile/simple-tf-use