I am learning about Deep Learning and have read/watched articles and videos regarding the topic to get a better understanding. I want to practice the things I learnt, but I am not familiar with TensorFlow or PyTorch.
Can anyone suggest which library to choose and resources to get guidance for hands-on practice?
2 Likes
Great to hear that you’re diving into Deep Learning — exploring articles and videos is a solid way to build your foundation.
If you’re taking courses from DeepLearning.AI (like the Deep Learning Specialization), you’ll primarily be working with TensorFlow, since that’s the framework used in most of their assignments and exercises. It’s a great way to get hands-on experience and build confidence as you go.
That said, PyTorch has become the preferred library in both academia and much of the industry due to its flexibility and more intuitive syntax, especially for research and prototyping.
Ultimately, both frameworks are powerful and widely used — the most important thing is to build strong conceptual understanding and get comfortable with one before switching or exploring both. Once you’re confident with one, it’s relatively easy to learn the other.
For practice:
- Stick with the framework used in your current course for consistency.
- Explore beginner tutorials on the official docs (TensorFlow, PyTorch) — both have excellent hands-on guides.
- Try implementing simple models like linear regression, logistic regression, or basic neural nets on small datasets (e.g. MNIST, Iris) as a next step.
Wishing you the best as you continue learning — you’re already on the right track by asking the right questions.
8 Likes
Thanks!
There is one specialization here from DeepLearning.AI that uses PyTorch, which is the GANs Specialization. That is also very interesting in its own right, so taking that once you learn TF from DLS would be a way to get a nice introduction to torch as well.
7 Likes
Do consider google trends to evaluate frameworks based on dimensions of interest.
3 Likes
The other thing worth saying is that if your longer term goal is to work in the field, then your best bet is to learn both TF and torch. Both are widely used and depending on where you end up working, it would be advantageous to know both of them. It’s just like learning programming languages: more is better. Another line item on your resume is a good thing, right? 
3 Likes
Thanks @paulinpaloalto for sharing valuable information. I have started learning Tensorflow as of now and plan to get familiar with PyTorch as well after it, because…you know… as the resume logic dictates, “The more the merrier.” 
1 Like
I am just like you but I’ve started a little earlier. I started my journey with tensorflow by taking courses from DeepLearning.ai. Amazing Framework and good documentation, I wanted to stick with it. But as I keep going on my journey I’ve learned that Pytorch is also used by many people and some research papers are being implemented in Pytorch itself, and so if you want to work with those models it’s better to work in the native framework on which they are built. So now, I’m taking my time to learn the Pytorch by working with projects. Both are great frameworks (Pytorch and Tensorflow). I’d say, it’s better to pick one first and be comfortable with it, then when you think you are ready slowly try the other one. The best way to try is build projects. All the best!!!
4 Likes
Since you’re still learning the subject, you might find Keras a good place to start. It’s a user-friendly interface for building neural networks and is widely used in educational resources. One recommended book is:
Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow (2nd Edition) by Aurélien Géron.
Keras runs on top of backends like TensorFlow or PyTorch, but it simplifies many tasks, making it easier for beginners to use.
1 Like
As a beginner go for tensorflow its far more better and simple to use and its highly efficient too.
1 Like