🌱 Seedling Planted: Updated:

Accelerate and you will succeed


What better way to accelerate knowledge?

TL;DR: Learning process from installation to professional use.

My choice: the Mac with Apple Silicon

I have a Mac. Sometimes it sounds complicated and even more so in Argentina where basically everything is Windows. Can you have a Mac with Windows? Yes, but you already know what that would be. I want to make it clear that these are completely different things: Mac is hardware, Windows is software. But for most people it can be enough of a differentiator.

I bought my first one around 2013-2014. It was like being in the future since my previous PC was still on my desk and I had taken it several times to upgrade the RAM (to play video games). Learning that operating system was complicated, but I never wanted to go back to Windows again (now we can be talking about similar things).

And one day it was gone. You don’t come back from a glass of Coca-Cola (now I drink Zero). And I got an M1. Since the previous one had an Intel chip (and at that time I had no knowledge of the difference) I lost the possibility of playing many games that are no longer native on this architecture (now I can speak with more words). After losing my game saves, all that was left was to learn Machine Learning.

What is Machine Learning?

I can’t answer a question with another question, but we humans are used to thinking in context, logic, and seeking a deeper meaning of things before giving an answer. We’re very good at this. On the other hand, Machine Learning differs in that it looks for patterns after learning many (better if there are many-many) examples. Better if we have inputs and outputs already ready to find those patterns (careful, there are several types). Here you will surely find a better definition. But the fact is that we can use Machine Learning to program without writing what the computer has to do. The typical black box model:

Between the input and the output lies the magic. The patterns are learned and these are finally used for other data and it keeps adjusting.

MLX: a similar but different framework

Installation according to correct usage criteria

Tests with small models

Tests with HuggingFace models