3 ms·
It's true that Markov chains are very limited in their capabilities. But one thing I love about them is that they are one of the simplest and most intuitive wa
by unoti 3y ago
It's true that Markov chains are very limited in their capabilities. But one thing I love about them is that they are one of the simplest and most intuitive ways to write code that *learns* from input data.
If you're never written something that *learns*, try it out! Here's a very primitive one I wrote recently to explain the basic idea and explains it along the way.
https://github.com/unoti/markov-basics/blob/main/markov-basics.ipynb https://github.com/unoti/markov-basics/blob/main/markov-basi...
This starts with generating US city names, and ends with generating text based on the Dungeons and Dragons DM guide.
- feoren 3y ago> that they are one of the simplest and most intuitive ways to write code that learns from input data Linear regression is significantly simpler and also learns from input data.
- max_ 3y agoMarkov chains are basically lookup tables. I don't see how linear regression is simpler that that.
- Jensson 3y agoMarkov chains aren't lookup tables... Markov chains is the description of a process where you generate the next step by looking at preceding steps, that can be a lookup table but it can also be more complex like modern LLMs.
- bormaj 3y agoThank you for sharing this! The implementation and examples were super intuitive.