What are the best resources to learn PyTorch and FastAI for Machine Learning?

Here's a well-written sentence combining the best aspects of both sources for learning Python: In my opinion, the official Pytorch website and a well-regarded YouTube channel are excellent resources. When choosing a YouTube channel, focus on one with a comprehensive curriculum to ensure you cover all the essential topics.

The FastAI course is a good resource for staying current in the world of machine learning. It not only covers essential topics but also keeps its content up-to-date. For instance, it included early preview lessons on Stable Diffusion as early as October 2022, making it one of the first courses to delve into this topic. Additionally, the FastAI website features articles that offer valuable insights and perspectives, making it a comprehensive platform for machine learning enthusiasts.

FastAI offers state of the art material explaining the latest model architectures. For beginners, I would recommend to first start with the basics and take a course at a university or coursera.org etc though.

In my opinion, the best way to learn about creating custom APIs is to walk through an official document and then try to implement your own. This hands-on approach will solidify your understanding of the concepts involved.

Deep Learning for Coders with fastai and PyTorch - jeremy howard jeremy howard is creator of fastai and i follow since my AI journey. in this book you have learn that how to train a model on a wide range of tasks using fastai and PyTorch. You’ll also dive progressively further into deep learning theory to gain a complete understanding of the algorithms behind the scenes.

For me a great way to learn is reading papers and the documentation on pytorch. For each network layer or method they usually link to the original paper and you also get the change to see the git repository, use their code and write comments/report bugs.

Writing code from scratch is an excellent way to gain a deeper understanding of how algorithms work. By building the code yourself, you break down the problem-solving process into smaller steps and gain insights into the logic behind each step. This hands-on approach allows you to see how the algorithm interacts with data and how it arrives at the desired outcome.