Deploy_classification_model

Here I trained a model using TF Hub. The model is pre trained on 7B data. The dataset is on a sentiment classification of food, where -1 means negative review, 0 means nutral and +1 means positive review. The pre trained model will take care of our model.
After downloading the data you can run the whole notebook. The model will be created in model_checkpoint folder, where the best model will be saved along with the weights. The model will be exported in the amazon_review folder, there will be a folder based on the time frame you are running the notebook. You can find it from there.

Beside docker tensorflow serving supports

• Remote Procedure Protocal (gRPC)
• Representational State Transfer (REST)

Default ports

• RPC: 8500
• REST: 8501

Data

The dataset can be found here

Deploy

open a cmd and cd over to the Deploy_classification_model
Run this in cmd

1. Docker
   1. docker pull tensorflow/serving
   2. docker run -p 8500:8500 -p 8501:8501 --mount type=bind,source=/path/amazon_review/,target=/models/amazon_review -e MODEL_NAME=amazon_review -t tensorflow/serving
2. Rest
   python3 tf_serving_rest_client.py
   it will run the model and let us type the comment manually.
   Default port http://{HOST}:{PORT}/v1/models/{MODEL_NAME}
   Specified port http://{HOST}:{PORT}/v1/models/{MODEL_NAME}[/versions/{MODEL_VERSION}]:predict
3. gRPC
   python3 tf_serving_grpc_client.py
   it will run the model and let us type the comment manually.