How I passed the TensorFlow Developer Certification Exam!
At the beginning of 2020, TensorFlow, an open-source framework for developing an end-to-end deep learning model released its official certification program which tests developer abilities to build Computer Vision, Natural language Processing(NLP), and time series models. It also tests abilities to strategize data and modeling best practices such as handling fewer data problems, over-fitting, and under-fitting.
Through this article, I will talk about the following:
What is TensorFlow?
What is TensorFlow Developer Certificate Program?
What does it mean to the career of certificate holder?
How to prepare for the exam?
Yes, prepared and ready to take the exam, what to expect during the exam?
After the exam…
Like any other exam, there is confidential information that will not be shared in this article as I agreed while taking the exam, and I hope you also do so when you take the exam.
Alright, let’s get started!!
What is TensorFlow?
TensorFlow is an end-to-end deep learning framework for developing and training machine learning models. TensorFlow has gained popularity due to its simplicity in building models, a wide range of application deployment scenarios such as mobile devices (iOS and Android), and embedded devices.
Speaking of that, there are variants of TensorFlow APIs such as TensorFlow.js for JavaScript developers, TensorFlow Lite for mobile devices and embedded devices, and Swift for TensorFlow (beta). Also, there are other awesome tools and resources which you can check out here.
But why TensorFlow? It is the center of all Google machine learning applications and it is what power all apps such as YouTube, Google Photos, Search engine, and many more. Being backed by Google, there must be a huge opportunity in harnessing the use of this framework.
Besides being backed by google which can guarantee its long-term development, its community is also supportive which can make it easy to find numerous answers to any problem you can be having in your ML projects.
What is TensorFlow Developer Certificate Program?
As said, the TensorFlow certificate exam is a professional certificate that can help TensorFlow developers to demonstrate that they are proficient in using TensorFlow to build deep learning models.
The certificate costs $100, but you can get a $50 stipend which takes between 4 to 6 weeks from the day you request it. If you plan to request a stipend, do it as soon as you start to prepare for the exam.
The goal of this certificate is to provide everyone in the world the opportunity to showcase their expertise in ML in an increasingly AI-driven global job market. From tensorflow.org
TensorFlow team has also made a network of certificate holders on a global map. You can be the first one in your country to set the pin there!!
What does it mean to the career of certificate holder?
Gaining something to show that you have expertise in using this leading ML framework comes with huge benefits such as yes, such recognition, and can also showcase your skills on social networks such as LinkedIn and GitHub. The most thing I liked about this certificate exam, is that you learn things you would not feel like learning.
Here are other benefits:
It can surely boost your resume especially if you don't have prior ML experiences to show
Being on the professional network of certified developers
It can leap your career.
Whether you take a certificate for having giant recognition, boosting your career, or learning new skills, they are all good reasons.
How to prepare for the exam?
There are numerous resources to help you get ready for the exam. The first thing which may be good to look into is the Candidate Handbook. Since this is updated regularly, check it often before you take the exam.
For example, when I first looked into it months before the exam, the python version to use was 3.7 but when I was about to take the exam, it had been changed to 3.8 (as of now). Luckily, it was the final day of preparation and I updated to Python 3.8 as soon as possible. You do not want to be frustrated by these kinds of things during the exam!
The following is the list of resources to help you get started. They are recommended if you are beginning or if TensorFlow is not your day-to-day framework.
DeepLearning.AI TensorFlow Developer Professional Certificate: This course is the number one to help you pass the exam. It was taught by Laurence Moroney, Lead AI Advocate/Head of TensorFlow Advocacy at Google. By taking this course, you learn how to develop deep learning models for Computer vision, Natural Language Processing, and time series.
Intro to TensorFlow for Deep Learning course: This is a free course taught by the TensorFlow team and Udacity as a practical approach to deep learning with TensorFlow. It is a great course and the curriculum is great as well. This is a substitute for DeepLearning.AI TensorFlow Developer Professional Certificate.
Hands-on Machine Learning for Scikit-Learn and TensorFlow: This is a great book not only to pass the exam but to also understand every single thing in the end-to-end ML works. The repository for the book chapter is open on the author GitHub page.
Intro to Deep Learning MIT: These are free lectures by MIT on Deep Learning. The recommended lecturers are the first three, each 40 minutes long. They are very practical, quick, and helpful in understanding the basics of deep learning, convolutional neural networks, and sequence models.
There are so many resources to look into, and you can find all of them on the TensorFlow website. Even if you may be already good with TensorFlow, you should definitely check it before taking the exam.
The Python Academy also provides in-depth and practical training boot camp for learning how to build effective deep learning models for Computer Vision, NLP, and time series. If you have taken any academy course or would like to learn along with the certified instructors in live settings, where you can ask questions, this training is for you. Learn more about the training here!
Yes, prepared and ready to take the exam, what to expect during the exam?
The exam is 5 hours long and it will be automatically submitted if the time gets to end. The official environment for the exam is a current stable version of PyCharm.
Having limited time means that you are better off prepared to pass the exam if you have some plans in order. Here are what I can say about this part, and I will link to whatever helped.
Be familiar with PyCharm before taking an exam, and do run some of the tutorial models available on TensorFlow Tutorial page.
You are required to build five models, and they become complex from model one to the fifth.
During the exam, you are required to train models which may be slow depending on your machine. It is also advised to run them in Google Colab. I did this in parallel to training in PyCharm and it helped a lot. The point here is that you can build models while others are being trained in Colab or Kaggle kernel.
Be sure to have stable internet for the whole examination period, power, and fewer distractions. I did it at midnight because I know there are no distractions and good internet as well.
If you run into an issue, check what the instructions and Candidate handbook have to say about it.
Read the Candidate Handbook again before taking the exam, and have it and other instructions open during the exam.
After the exam…
After the exam, you receive an email that either says you passed or you didn’t pass. You do not get a score or any details of your results. The only score you get is the ones you get while submitting models for a test such as 3/5 on every single model, and you get this directly in PyCharm not in the email notifying results.
Getting to the end, as Daniel Bourke puts it in his YouTube video, in this world where everyone can declare expertise after finishing an online course, you need some sorts of things to set you apart. This certificate may not be a must to have, but it is nice to have!!
Thank you for reading!
P.S: The original version of this article was posted on the author's blog, TensorFlow Certified Developer and Instructor at The Python Academy.
ความคิดเห็น