Experiments

#Train and serve a Fashion model with TensorFlow Serving

• Create virtual enviornment

virtualenv venv --python=python3

• Activate virtualenviornment

source venv/bin/activate

• Install Requirements

pip install -r requirements.txt

• Run:

python tf_serving.py

• To test:

  • Open another terminal

    • Activate enviornment

    • nohup tensorflow_model_server --rest_api_port=8501 --model_name=fashion_model --model_base_path="full path to pb folder" >server.log 2>&1

    • Check server log status is success

  • If Server log status is success

  python tf_serving.py -in predict mode in another terminal