top of page

Curious What is TensorFlow? Understand the background and basics! (Part-2)

To understand the basics of TensorFlow, we have seen and learned about the first task of building a graph. Now, we will focus on the second part of it which is the Running of the Model.





We can do this in different stages and the first stage is, training our model/data. We will run the model on testing data and if we are happy with the results i.e. the model is accurate enough then, we can take our model to the next level by using the real data.


Once the graph is done, we run a session on our graph to get the expected output or the results and to test how our model is… and so on and so forth. We do this on the testing data giving some sample inputs and we expect the desired result out of this model. So, for that, we need to specify EPOCHS i.e. how many times we want our model to run.


Also, we need to specify the learning rate which tells us how values are going to change over time and how quickly they will change.


But if we don’t run it proper times then we might not get the accurate result. This is similar to find the optimal learning rate. It going to take ages and ages to run but, again if it does not learn quickly then, we might get a slower model. It’s not difficult to use, TensorFlow actually makes things really easy providing high-level APIs.





The challenge is to find the most efficient way to go about using it to solve the problems. Also, finding enough data to train and test can be a bit problematic e.g. stuff like images.

We have access to massive libraries and huge datasets that allow us to study the already trained data. We can build something completely new then we have to train the data.

Building the model can be the toughest part and running i.e. training and testing the data can be the most time-consuming part.


These are not warnings to stay away from this, but it is more about telling you the depth of creating a model properly. We need to keep all the ideas in our mind at a time to get through it.


- Sarthak Niwate

info.sarthakfeeds@gmail.com

 
 
 

Comments


Subscribe Form

  • twitter
  • linkedin
  • blogger
  • generic-social-link

©2019 by Content Sphere. Proudly created with Wix.com

bottom of page