How to Pass TensorFlow Developer Certification
TensorFlow Developer Certification is one of the most prestigious certification on machine learning. If you want to be acknowledged by company that you are good with machine learning, this is definitely the certification that you have to take.
You can learn more about tensorflow certification at the tensorflow official site https://www.tensorflow.org/certificate.
I divide this process into four stage that you have to prepare before taking the certification exam.
- Software preparation stage
- Knowledge preparation stage
- Document preparation stage
- Working stage
Software Preparation Stage
First thing first is you have to install python 3.8 and installing pycharm 2020 or 2019 IDE. Python 3.8 can be downloaded here and you can choose the Windows x86–64 executable installer for windows.
After that you don’t need to install python libraries such as numpy, pandas or tensorflow since when you take the examination, the plugin will download all of the necessary library automatically.
Next thing that you have to prepare is google colab, familiarize yourself with google colab since developing the model on google colab is fast and free.
Knowledge Preparation Stage
Learning machine learning can be hard, but thank to tensfoeflow, they already gives you all the necessary things that you need to understand on their coursera course. Coursera: Deeplearning.ai Tensorflow Developer
On my opinion, this is the key to pass the tensorflow developer certification, the question asked on the certification is really similar to the exercise given by the course. The exercise especially hello world in neural network, and classification exercise human or horse, dog or cat, rock paper scissors, fashion_mnist, sarcasm detection, and movie rating classification.
From what I remember the certification itself have 5 question.
Question 1 : Very similar to hello world in neural network
Question 2 : Classification with mnist dataset
Question 3 : Classification with augmentation on real images
Question 4 : Question about text processing and NLP
Question 5 : Time series prediction.
And also if you are Indonesia, you could try dicoding: Simulasi Ujian TensorFlow Developer Certificate
This simulation gives you a really good benefit since the simulation is very similar to the question asked in the certification. Is this cheating ? I don’t know, but I will utilize everything that I could use.
After working on the simulation, don’t delete your code nor your model. Since the examination have a similar question you could re-use your code.
Document Preparation Stage
After buying and redeeming the certification code, we need to validate our identity, this can be done by uploading your passport or government issued ID. After that we need to wait for 2 hours to activate our account.
Working Stage
To start the certification we need to donwload the Tensorflow plugin on pycharm marketplace, but if it didn’t exist, you could download it manually nad put it on pycharm plugin folder. Download it manually from the link below
https://plugins.jetbrains.com/plugin/13812-tensorflow-developer-certificate
Then extracted it to this directory:
C:\Program Files\JetBrains\PyCharm Community Edition 2021.1.1\plugins
From stack overflow discussion.
After activation the plugin, you could start working on the task. Pay attention to the upper right sidebar called assistant.
When we click the assistant and then click our program, the assistant will test our model with score up to 5 in every question. And from what I remember you could past the test if you have accumulated the total of 15 score, but just to be safe, try achieving 20.
Another tips is, you can use google colab to test your model while the assistant is checking our model. If you already developed a good model in google colab, you can download it and put it on the same folder with the question folder in pycharm.
