From the person in the video regarding issues with benchmarks in general, and for LLMs. Also their approach. Good article.
Introducing System One Models and Jev
121–130 of 513 posts
Re: Introducing System One Models and Jev
#122It could be used for coding if you gave it an AST. If you work at TypeSafe please try this. Side note: This is probably how LLMs would perform with better encoders and next-latent prediction, so eventually those will beat this architecture out. Still amazing though.
the hard part for coding is actually state engineering (e.g. getting your dependencies in context) - we haven't even tried it yet (because my philosophy is we should automate the easy tasks before the hard and we've been working on getting the model smart on the former)
we do think there's a lot of potential though and do want coding themed releases soon
Re: Introducing System One Models and Jev
#123First, congrats to the team on launching something genuinely interesting and new. Seems like a more accurate title would be "Jev: Trading general purpose generation for fast typed inference" or something like that. This is interesting, but the speed comparison seems misleading? A generative model that can output code in a Turing-complete language can do anything a computer can do. Jev can only generate structured out…
I'm biased but I wouldn't call it misleading - generating text is super awesome and flexible, (we describe that in the blog post - and I personally use string models all the time) but it's true you pay a high tax for autoregressive generation > Also "can't hallucinate" seems wrong? Sure, it can't emit an invalid type, but it can still emit a completely wrong valid value. that is likely true of all ML! perhaps we coul…
Just to be sure that I understand, you're saying that your model "can't hallucinate" because it only outputs a single thing, right? In this way, an LLM can't hallucinate either if I prompt it to do a classification task with a discrete set of possible outputs, right? (Assuming I reject non-conforming output. Actually, maybe what you're saying is that your system can't output non-conforming output?)
Re: Introducing System One Models and Jev
#124After much fumbling around with prompts and evals, this is exactly how I am using LLMs in production, to narrowly make choices and return structured data. Any deterministic work gets pulled out of the prompt and my goal is to narrow the model output to be as clearly defined and as minimal as possible. Jev's focus on structured I/O and confidence scores are game changing. If this does at all what it claims, I think th…
we hope so! the bigger hope is to not just eat LLM market share, but to allow for people to use AI much more in the inner loop of software
Re: Introducing System One Models and Jev
#125We had early access and found it to be pretty useful. Having a second form of verification, where you can ask multiple questions (in the form of Nouls) raised our confidence in the outputs of other models. [0] IMHO This type of model works incredibly well in concert with LLMs, not as a replacement.
[0] https://goodstartlabs.com/research/verification-is-the-bottl...
Re: Introducing System One Models and Jev
#126Earlier quoted context omitted.
Indeed, they talk as skeptics but don’t offer a ton of evidence, other than a couple videos of demos. A live demo would be far more convincing.
They gesture at not using benchmarks for some reason...
https://typesafe.ai/blog/antibenchmaxxing
But also effectively this is a classification model. It excels at specific certain types of workloads, and obviously will fail at others. Not really sure how one benchmarks this tbf. I can see their argument on why this requires a novel specific eval for whatever your usecase is. A consistent "global" benchmark might be hard to do
Re: Introducing System One Models and Jev
#127I was thinking about something similar (maybe) - generally speaking, embeddings for LLMs tend to learn real world concepts - things like 'fruit' or 'France' or 'city' as directions in embeddings.
But in things like programming, most concepts are abstract - 'if hungry eat an apple' in programming terms would look like
'if hunger > 50 {apples--; hunger-=30;}'
and compilers work with 'concept erasure' - to them, tokens (which are like llm tokens) look like
'if var1 > 50 {var2--;var1-=30}'.
They don't care about how these things map to real concepts. So all the embedding directions used to encode real-world concepts are just noise to LLMs when programming. This greatly reduces dimensionality and training costs. So does a token representation tuned for programming constructs, rather than natural language would probably have a more efficient encoding.
Re: Introducing System One Models and Jev
#128I am positive I know exactly how this works, I made something similar a few months back. But the problem is without generation you are extremely limited in the use cases. And while the model can't hallucinate, it can still be wrong. It just can't make up data.
Are you able to share how it works in that case?
I'll tell you this. Output isn't too cheap to meter, there is no decoder.
Re: Introducing System One Models and Jev
#129Okay, so it doesn't output text, that much is understood. What are the inputs like? I'm assuming maybe a text input? maybe an AST definition? Really hard to tell how this works at all from the demos, especially since we can't really try it out.
Re: Introducing System One Models and Jev
#130First, congrats to the team on launching something genuinely interesting and new. Seems like a more accurate title would be "Jev: Trading general purpose generation for fast typed inference" or something like that. This is interesting, but the speed comparison seems misleading? A generative model that can output code in a Turing-complete language can do anything a computer can do. Jev can only generate structured out…
I'm biased but I wouldn't call it misleading - generating text is super awesome and flexible, (we describe that in the blog post - and I personally use string models all the time) but it's true you pay a high tax for autoregressive generation > Also "can't hallucinate" seems wrong? Sure, it can't emit an invalid type, but it can still emit a completely wrong valid value. that is likely true of all ML! perhaps we coul…
> I'm biased but I wouldn't call it misleading
- @CompleteSkeptic
Very strange.