baobabKoodaa

In Spain they appear to happen rather frequently

As I said, it doubles the probability. That is an easily verifiable fact. Not sure why you are calling it facetious. I feel like you are trying to make some point about small probabilities being effectively indistinguishable from zero, but for some reason you don't want to articulate that?

Anyway, 3% is very very clearly distinguishable from 0% and so is 5%.

You think all people in all European countries work exactly the same amount of hours?

His point, which you pretend to not understand, is that the bar needs to be higher. Both water and fireworks are fairly innocuous items, even though both can be used to inflict harm and even though fireworks can be more easily weaponized than water.

Two decades

"have to" is very aspirational in that sentence

They did not post "valuations", so your hostile comment seems weird and misplaced.

The point is that IPO price is more speculative than the price established by the public market once the stock is freely tradable. When a stock holds its price post IPO, that indicates the IPO was priced more or less correctly (absent any other information).

Yes, it doubles the probability that I win the lotto. Yes, doubling a probability is significant. I'm not sure you're using the term "prior" correctly here, though.

It seems like you talked about investing after the IPO (e.g. on the first day listed on the stock market) but for some reason you phrased everything as investing at the IPO. Weird.

Average and median IPO pop in U.S. is very clearly positive.

Retail should not invest in IPOs but that is due to different reasons that you didn't discuss.

A change from 3% to 5% is significant. It's almost doubling the probability.

It's time to cut the undersea cables

Disgusting. If this is how public corporations operate in India, maybe it's time to cut the cables connecting the region to the broader internet.

Web PKI doesn't work like you think it does. I can use eMudhra's private keys to forge a cert for your website. You don't have a choice in this.

This sounds like a misunderstanding of how PKI on the web works. Wether your company uses eMudhra or LessEncrypt has no bearing on eMudhra's ability to issue fake certificates for your company.

You know the old saying: if you find 8 cockroaches in your house, then a realistic estimate for the total number of cockroaches in your house is exactly 8.

Unreal. Absolutely unreal.

Question back: how do you imagine them "raising the prices" of the tokens generated on my own GPU? That's the implication we're discussing here. Please explain how that works.

What if I want a whole world inside of computers?

Also: who are the investors subsidizing electricity bills for the people who run LLMs on their own GPUs? I'm still waiting for my subsidy cheque over here, so if you can plz send it to me ASAP, thanks.

Don't give him any ideas

No, what you said was:

> What it does was published in arxiv back in 2025

This is false. You made a statement which was false.

If you want to say that the author of the paper made a really bad knockoff of Jev in 2026 after Jev was published, then say that.

Jev can do a subset of LLM things faster/cheaper. Jev is not better than big LLM's at anything.

The comment I was responding to describes Jev's significance as being specifically about embeddings:

> Jev is basically the embeddings side of an LLM

This is how the paper you are referencing is describing the "key contribution" as it relates to embeddings:

> Novel state representation techniques using Azure OpenAI embeddings (3072 dimensions) with sales-specific features

And now you are claiming that the author of the paper created:

> "generalized" version of same

It's a bit hard to guess what you are trying to say, but if I were to steelman your argument, I would guess that you mean: training a model with a vocabulary that has some tokens representing things like "OPTION_A", "OPTION_B" is in your mind the same thing as "generalized version of sales-specific features in embeddings"? Is this what you were trying to say?

Sounds like you are running a tiny toy model if you can get generations in under 5 ms? Typical response times from LLMs for typical "jev-like" queries from OpenAI and Anthropic are 2s-10s. Not milliseconds. Seconds. Same queries from Jev are like 0.2s. and the cost is 1000x.

No, we're not at the point where you could ask Claude to do all of that, unless you have a super easy problem to begin with and/or you don't care about output quality. Feel free to link a counter example.

What you're missing is: cost and speed. Otherwise, it's very much like running an LLM and constraining the output.

I don't know how good Jev's "confidence scores" are, but I would be surprised if they were in any sense better than logits from some good LLM. One advantage of Jev here is that the confidence scores are easy to access. Most LLM API providers don't provide an easy/convenient way to access the logits. But that's a minor point, you could of course build something like this with LLMs (and many people have).

"Models on hf" (unspecified) are "FREE"? Like "free to download"? Sure, but nobody was talking about that. They cost money to run inference on. Unless you are talking about some tiny toy models that are useless for any non-toy problems. You clearly don't have any idea what you're talking about. Just stop, man.

Yes, businesses are solving specific problems, but most businesses have more than 1 problem to solve. No, it is not economical to pay a data scientist to develop a custom model for each of your tiny problems. It is often much more economical to use a general purpose solution, like an LLM, or now, Jev.

No. It can't simultaneously be both general purpose and having task-specific embeddings.