Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Liquid Foundation Models: Our First Series of Generative AI Models

liquid.ai

Post reply on HN
Print view View original

51–60 of 159 posts

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#51

Post by jrflowers »

alach11 wrote: post #46
NoCoooode wrote: post #39

Earlier quoted context omitted.

That they can't do this sort of simple question speaks volumes to the entire approach. I don't think generative AI will ever be able to reach AGI, and most people selling LLM today pretend it is AGI

All it speaks to is that tokenization is weird and introduces artifacts to LLM performance. Counting letters is a trivial task when you're staring at words on a screen. It's much harder when you're perceiving vectors based on parts of words. The fact that LLMs find certain things easier/harder than humans is completely unsurprising, and there are much more interesting benchmarks to use to compare one LLM to another.

This is a good point. While LLMs being incapable of reliably doing a simple task that’s been doable by computers since the punch card days is an important consideration for anyone that might be thinking about using them for anything other than as a toy, this fact is uninteresting because of Reasons

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#52

Post by asadm »

coding123 wrote: post #38
zozbot234 wrote: post #30

> How many r's in "strawberry"? Think step by step! What a nice question! Well, let's just count: 1. S 2. T 3. R! So, we have our answer: there are 3 R's in "strawberry". Totally checks out!

Can it write a python function to count the letters?

all of them can.

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#53

Post by swyx »

NoCoooode wrote: post #39
alach11 wrote: post #33

Earlier quoted context omitted.

This is one of the least interesting questions to ask LLMs. I wish it wasn't so popular.

That they can't do this sort of simple question speaks volumes to the entire approach. I don't think generative AI will ever be able to reach AGI, and most people selling LLM today pretend it is AGI

imagine being so confidently wrong about AI

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#54

Post by Spivak »

coding123 wrote: post #38
zozbot234 wrote: post #30

> How many r's in "strawberry"? Think step by step! What a nice question! Well, let's just count: 1. S 2. T 3. R! So, we have our answer: there are 3 R's in "strawberry". Totally checks out!

Can it write a python function to count the letters?

This is how I get it to do correct date calculations. Don't tell me what date next Monday is, write Python code using datetime to calculate next Monday and I'll run it in a sandbox. Works super well.

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#55

Post by simion314 »

vessenes wrote: post #5

So a new set of small models that are competitive with and beat phi-3.5 on some benchmarks is extremely impressive. Context length is the next frontier for models in this range - getting to a true 128-200k token length in a small model would be very very useful. Hallucinations are less of an issue because you can just dump all the source data in, whole codebases can go in for stuff ranging from a refactor to ‘write d…

>Hallucinations are less of an issue because you can just dump all the source data in, whole codebases can go in for stuff ranging from a refactor to ‘write documentation of the API’.

Is there no risk ? I mean say for testing purposes we give the AI a giant CSV file and ask it to make it a json is the chance for error 0% ? Because today we need to double check when we ask AI to transform some data or transform some code, there is the risk of it messing something up but if it is not something that would crash immediately you risk introducing a ton of new bugs by asking an AI to refactor instead of using some good tools.

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#56

Post by moralestapia »

Spivak wrote: post #54
coding123 wrote: post #38

Earlier quoted context omitted.

Can it write a python function to count the letters?

This is how I get it to do correct date calculations. Don't tell me what date next Monday is, write Python code using datetime to calculate next Monday and I'll run it in a sandbox. Works super well.

I kind of also do this, for some queries, and I found it to be a very interesting concept, even from a philosophical pov.

It's some sort of Chain-of-thought constraint on a different layer of abstraction. Does this come with its own advantages?

(And no, AGI is not here either).

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#57

Post by jimbokun »

NoCoooode wrote: post #39
alach11 wrote: post #33

Earlier quoted context omitted.

This is one of the least interesting questions to ask LLMs. I wish it wasn't so popular.

That they can't do this sort of simple question speaks volumes to the entire approach. I don't think generative AI will ever be able to reach AGI, and most people selling LLM today pretend it is AGI

In isolation, probably not.

But it's likely to be an important component in an AGI system. I suppose the interesting question is how to integrate LLMs with more traditional logic and planning systems.

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#58

Post by k__ »

GaggiX wrote: post #40
Jackson__ wrote: post #36

API only model, yet trying to compete with only open models in their benchmark image. Of course it'd be a complete embarrassment to see how hard it gets trounced by GPT4o and Claude 3.5, but that's par for the course if you don't want to release model weights, at least in my opinion.

Yes, I agree, for these small models it's wasted potential to be closed source, they can only be used effectively if they are open. EDIT: HN is rate-limiting me so I will reply here: In my opinion 1B and 3B truly shine on edge devices, if not than it's not worth the effort, you can have much better models for already dirt cheap using an API.

While I'm all for open models; why can't the small models not be used effectively?

Wouldn't they lower the costs compared to big models drastically?

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#59

Post by tossandthrow »

mrbonner wrote: post #32

Let me start counting how many minutes it would take for someone to post a comment blasting this model for reasoning capabilities. For me LLM, is as useful as it is a word calculator, borrowing from Simon Wilson, a frequent poster here and author of the LLM python package I use. I use LLM everyday and it is increasingly something I can't function effectively without. But, look, I am not using any of those reasoning u…

In order to effectively write a good summary, the model needs to have some capacity of reasoning capability.

It is a continuum.

Re: Liquid Foundation Models: Our First Series of Generative AI Models

#60

Post by WhitneyLand »

smokedetector1 wrote: post #48
GaggiX wrote: post #42

Earlier quoted context omitted.

The fact that LLMs are usually trained on tokens and not on characters, doesn't really speak about what generative AI is going to reach or not. >most people selling LLM today pretend it is AGI Who are these "most people"?

ELI5 why are tokens not a single letter?

Such an architecture could be implemented, it could use one token per letter, or one token per word, instead of the typical 0.75 per word we see.

The choice just comes with trade-offs in memory usage, compute, and effectiveness of the model in various scenarios.

So what we ended up with was a pragmatic/engineering decision rather than a theoretical or fundamental constraint.

Post reply on HN