Live data from Hacker News

Markov Chains are the Original Language Models

elijahpotter.dev

11–20 of 177 posts

Re: Markov Chains are the Original Language Models

#11
post #7
post #6

blogpost ends suddenly on “My solution” on ios safari. i’m pretty sure this was unintentionally cut off, OP

I'm also getting a client side error where the demo should be? Not sure if its just on my end.

My goodness. Thank you for letting me know. I'll look into it right away.

Re: Markov Chains are the Original Language Models

#12
post #4

Markov Chains are cool stuff. Back in 2015 in one of my senior class projects, I built a haiku bot that would take a seed word and make a haiku from a corpus of Tweets I collected from the firehose API.

Intriguing. How did you get it to fit the syllable count?

I forget now and can't find the code, but if I recall correctly, I had each tokenized word map to its syllables and used that in calculating the next word.

I wasn't able to get far enough to make each haiku make perfect grammatical sense, but it very often produced interesting results.

Re: Markov Chains are the Original Language Models

#14

This is what I've been saying to people, LLMs aren't much smarter than a Markov Chain - a lot of twitter bots and earlier chat agents were driven by them. What I do see though is the infrastructure around LLMs making a difference.

I've never seen a Markov chain do anything like GPT4. I'm not sure how you can say with a straight face they are basically the same.

Re: Markov Chains are the Original Language Models

#15

This is what I've been saying to people, LLMs aren't much smarter than a Markov Chain - a lot of twitter bots and earlier chat agents were driven by them. What I do see though is the infrastructure around LLMs making a difference.

I've never seen a Markov chain do anything like GPT4. I'm not sure how you can say with a straight face they are basically the same.

LLMs are Markov chains in latent space, it's the latent representation that give them their power, but ultimately there's not as much difference as one would suspect.

Re: Markov Chains are the Original Language Models

#16
post #3

Markov chains were one of the coolest discoveries of my programming career and I spent years using them to make forum and social media bots trained on people’s post histories for them. I think that experience is part of why I’ve been generally unimpressed with a lot of LLM hype. Like yeah, it’s cool and definitely more useful than a markov chain - but for the amount of resources that went into it I’d expect the gap t…

I first learned about Markov chains in the late 90s when I ran across a Markov chain bot on IRC. The person who wrote it was nice and answered my questions about it. I managed to write my own shitty copy but was pleased enough with myself. Same as you I thought it was one of the coolest things I came across in programming.

Also same as you why I've been very cool to LLM hype. Not that LLMs aren't feats of their own but they're not nearly as smart as their hype suggests.

Re: Markov Chains are the Original Language Models

#17

This is what I've been saying to people, LLMs aren't much smarter than a Markov Chain - a lot of twitter bots and earlier chat agents were driven by them. What I do see though is the infrastructure around LLMs making a difference.

I've never seen a Markov chain do anything like GPT4. I'm not sure how you can say with a straight face they are basically the same.

No one has spent 100M on training Markov chains.

Re: Markov Chains are the Original Language Models

#18
post #9

Here's a good post on how effective these can be: https://colab.research.google.com/github/norvig/pytudes/blob...

My favorite intro is https://norvig.com/ngrams/ because it shows a variety of applications using simple code. (and uh since I'm in the acknowledgements, ok.)

(The people saying modern LLMs are just fancy Markov chains in practice are being silly, though.)

Re: Markov Chains are the Original Language Models

#19

This is what I've been saying to people, LLMs aren't much smarter than a Markov Chain - a lot of twitter bots and earlier chat agents were driven by them. What I do see though is the infrastructure around LLMs making a difference.

me too, current LLM are markov chains with context

Re: Markov Chains are the Original Language Models

#20
post #7

Earlier quoted context omitted.

I'm also getting a client side error where the demo should be? Not sure if its just on my end.

My goodness. Thank you for letting me know. I'll look into it right away.

while you're at it, site design on mobile is dreadfully info sparse. recommend stealing some css from any given developer writer you like.
Post reply on HN