Can someone do a ELI5A of how they achieve classification over any user defined list of items? Normal neural networks do a softmax over a known output set to get probabilities
You can achieve open-vocabulary classification by making the final weights in the softmax come from a category encoder instead of being fixed learned weights. So instead of
softmax(encode(input)*learned_weights)
You have
softmax(encode(input)*encode(categories))
I'm not sure if Jev does it this way, but it's how you get open-vocabulary zero-shot image classification with models like CLIP [1].
I can think of two possible approaches
1. Jev limits to 255 distinct options. So they can preprocess your set of options and “tell” the LLM via input tokens 1 = red, 2 = blue, etc then jev need only output softmax over 255 states while benefiting from pretrain of other LLMs
2. You allow the forward pass to output over the total token state but mask over the logits to limit to the user options. Less plausible? bc tricky when input is multi token which they clearly support.
My guess would be option 1. Didn’t read the kev repo here which would also explain
I don't agree. If anything these newer languages have better tooling and new projects are always built from ground up to support open standards like open telemetry
Open telemetry is about how telemetry data is reported, not how it's collected. It's hard to compete with JFR on the breadth and depth of low-overhead, in production telemetry, built into the standard library and the JVM itself.
Has anyone tried formal verification with AI generated code? I can't convince my company to use it but I realise it's very easy to ask Claude to add a verification step locally on my own PRs
At one point I was using TLA+ but it just made the problem "is the spec right?" or "does the code match the spec?". I could ensure that the properties defined in the spec were valid, but that didn't seem to translate into confidence that my code was correct. Maybe I was holding it wrong, it was just an experiment in an area I'm unfamiliar with.
Ultimately I have stuck to the informal verification of defining my expectations and ensuring that tests cover them.
I don't know TLA+ but indeed that seemed to be the limitation when I read about it too. What about languages like Dafny, that allow you to "prove" your programs?
They're probably referring to tech like reasoning models, or agent harnesses for example, which are arguably slowly moving things towards the symbolic end of the spectrum.
LLMs using REPL are one instance of symbols to "bounce" their prediction against domain constraints for verification. Also shout out to Gary Marcus who was right after all (and LLM companies wasting 100s of billions of dollars for years in-between on pure scaling).
I had tried to get long term memory out of Claude by indexing my notes with keywords and putting that in a sqllite database and Claude queries using full text search. Don't know how good it is, it seems to find things alright. My goal was to keep context small and only get Claude to ask for what it needs. Datalog seems like a great idea, will definitely try it out
The rewrite was a lot faster. Actually shaking out the bugs from the rewrite (or just getting enough confidence in the new version from use by early adopters) is probably not going to be drastically faster.
My understanding is: it takes text input and it does one shot classification (no training data)
reply