Exploring Beyond Notebooks

Introduction

We train a model on terminals, notebooks, etc. This week I realized that there are more things beyond the training and validation accuracy. The thing of deploying and adjusting the inference response through UI or API hitting makes me a real deal of work to do.

Just look at this flow below that will make you feel curious and I will explain each

Terms or Tools

• Model saving is your trained model saved in tar in most cases.

• Tensorflow serving service will up your inference model on a server with certain commands on docker.

• Flask service is a bridge between the world and your model, it will interact with the TensorFlow serving and the input data.

• Postman is more sort of API hitter where I provided an input image and it give a better and response after completing the flow.

Future Steps Tweet

Tushar Raj Verma @TrajVoid

I was able to implement the TensorFlow serving for the model inference that is a good practice to use in production In the below flow you will find the ‣ Training steps ‣ Saving and deploying ‣ Posting request ‣ Predictions ‣ Returning desired outputs https://t.co/6chn46jJ5A

Conclusion

The power working beyond notebooks made me realised that we all should have good knowledge about the flow of deploying,I expect that many of us will se it as a next step.

Thanks For Reading!