Auto Hyper-Parameter Optimizer

This is the small implementation to understand the power of MLOps. It is one another approach to mutate the hectic process of designing Ml/Dl model with desired accuracy without human intervention. It is more efficient approach.

I explained all the details of this project and step by step implementation in my previous article : https://www.garudax.id/posts/himani-agarwal-6b5ba418a_mlops-devops-integration-activity-6672573725944246272-oj5

Here i mention only the changes which i made to do this task more effectively.

In this approach I use some basic linux commands to change the values of hyper-parameters and for training model I use Transfer learning.

Hyper-parameters : those parameters which user have to specify and to get more accuracy he has to change them accordingly again and again.

Transfer Learning : when new things come up we just add that in model without re-setting all the old experience or learning is performed without considering past learned knowledge in other tasks.

Here I use vgg16 - pretrined model and add new knowledge to this.

For reference use my git-hub repository - https://github.com/17ejtcs031/ml_update_task3.git

1. Dockerfile

2. After creating job for download the code from SCM the following jobs ,I created

Our core changes has been doing here -

so that's it ,all other things will be same as previous one (article details mentioned above) and yeah it is working great.........!!!