Posted by albelfio 17 hours ago
> [them] Output tokens: FREE (too cheap to meter).
I'm very confused by this.
[0] https://goodstartlabs.com/research/verification-is-the-bottl...
And each question is a separate single token model completion done in parallel
If self driving is red means stop, green means go, and stay in your lane - then it would work great, but having to actually think and test which maneuver is optimal for a given situation while weighting safety, road rules, random unexpected actions and getting to your destination, I think it's a much bigger problem. A bigger model specifically trained on that maybe would do great, but then the output is not the constraint anymore.
But I haven't tried the model, so I 'm just ballparking and could be very wrong.
LLMs are universal simulators, their latents model the world. So I bet if you compare their logprobs with probabilities output by this model, it will be highly correlated.
Someone should do this quick experiment. I bet there won't be enough of a meaningful difference.
Edit: never mind, found https://docs.typesafe.ai/introduction/quickstart by now
Jev input … > Unstructured data (e.g. text) with an emphasis on structured program state.
What pre-training data/model is Jev based on? Surely result effectiveness is dependent (outside of one’s own input as “state”) on that?
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.