The Philosophy for Learning

Many of you want to start learning ML whether it’s for a side project, getting into the field, or just gaining the skills to up your market value. Regardless of your motivation I’m guessing this is how you came to this article:

1. 1- Got interested in ML somehow
2. 2- Started looking into how to start
3. 3- Saved a bunch of resources and guides
4. 4- Haven’t done any of them

Often the hardest parts of learning are the end when you start to reach the really heavy shit, and the complete beginning where you are lost and overwhelmed which leads to procrastination. This is normal but sucks ass so I’m going to tell you how to get past it and actually make progress.

The best model for this is the 10,000 hours mantra Karpathy shills constantly. In short this means that if you put in 10,000 hours into anything you will become good at it, maybe not the best but greatly competent. With this model by putting in 10,000 hours it doesn’t matter where you start as long as it’s the elementary levels because throughout those 10,000 hours you will discover what you need to know that you don’t already, filling in knowledge gaps until you get to the cutting edge.

Now here’s how I’m going to motivate you with this, if you put in 1 hour a day it will take you approximately 30 years for this level of mastery. That’s a long fucking time, but consider this: moving to 3 hours a day brings you down to 10 years, and moving to 10 hours a day brings you to 3. This means that with every extra hour put in every day you are saving years until mastery which gives you more time to make an impact with your skills.

So in relation to ML you need to be putting in the effort and JUST START SOMETHING, literally anything. Put the hours into a resource, practice what you’ve learned with projects, then find knowledge gaps and fill them. Iterate through this process until mastery, but although motivating this still leaves room for confusion which can be used as another procrastination method. So I will take away this friction to your learning now with some places to start.

How to Actually Start

Remember this article is written by me in a place to where I too am just starting my ML and CS journey, the main purpose is to KICKSTART you into your own studies. With that in mind here are someplaces to start with resources linked below. Most of this shit will be plagiarized from moots in tpot who have helped me with my own shit, so without further ado here is some starting places.

Do you know the math for ML (Calculus, Linear Algebra, Stats and Prob)?

Nuh uh:

• “Mathematics for Machine Learning” by Marc Peter Deisenroth, A. Aldo Faisal, Cheng Soon Ong
• Math Academy Grind (Do ML Roadmap there)

はい、パパ:

• “The Hundred Pages Machine Learning Book” by Andriy Burkov
• Andrew Ng’s coursera or youtube lectures

Do you just want to start doing project based shit?

はい、パパ:

• Start learning Pytorch via youtube or ZTM
• Start with random project idea and use LLM/Search Engine to learn things you need to know to build it

Are you reading this not even knowing how to code with Python?

はい、パパ:

• “Automate the Boring Stuff with Python” by Al Sweigart
• Leetgrind

There, now pick something and do it.

Learn -> Apply -> See Knowledge Gaps and Iterate

Thanks for reading and good luck on accelerating anon.