Why is everybody so obsessed with it? There are 2 Jev posts on the front page even now, I feel like I'm taking crazy pills.

Nerd sniping the hackernews/twitter crowd with "large scale transformer-based language model alternatives."

People don't want to believe something as unsatisfying as "Scaling up LLMs" can yield something as profound as AGI/be useful, and just hope that literally anything else can take their mindshare away, and this just happens to be the new rage. Along with clearly-not-frontier-level open source models, non-transformer based architectures, etc.

Beats me also - this feels unreliable, extremely niche, and over-hyped. I don't trust LLMs even when they explain their reasoning; the idea of trusting a black-box classifier like this seems insane.

For certain tasks, it seems much, much more efficient. That's not nothing. People have been using LLMs for various classification tasks.

In my company, and I think in most companies that are using AI at all, one of the first ways it got integrated is as a classifier, to tag orders based on feeding all their data into a prompt and asking for a structured output.

I think demand for tools that are more tailored for this type of integration is high. I don't really understand why Jev is supposed to get my company's decisions right more than an LLM, but regardless of the tech I think people are just excited about the possibility of iterating faster, more explainability, higher-level tools that are specifically created to help hone classifiers etc.

Great, we don't need 15 thousands posts per hour across social media channels. We had classification NN before LLMs as well.

I think the point of Jev is to thread the needle of the gap between non-LLM classifiers and LLMs.

Classifiers like classical NNs require:

- annotated data, potentially a lot of it - training - inference

#2 and #3 aren’t a big deal if you have an ML engineer, but #1 will always be a potential headache no matter who you are. The tradeoff is that they could be quite fast, cheap, and you can get probabilities, not just classes.

With LLMs you get:

- zero shot classification (no dataset or training required) - potentially can use third party model providers like OpenAI off the shelf. Don’t even need to host your own model.

The downside to LLMs is that they are comparatively slow and expensive to traditional classifiers. Historically they also were prone to hallucination or malformed responses, though not as much these days. You also can technically get log-probs back, but these aren’t equivalent to the classifier probabilities.

Jev gets you the zero-shot, zero-infra benefits of LLMs, while being closer to the speed and cost of traditional ML classifiers, as well as both classification and probability responses.

Yeah but they weren't that great, you couldn't ask for arbitrary classifications after the model was trained. You are underestimating what they've done here, even if it does seem a little overhyped.

This is what happens when a society stops believing in God.

I have entirely stopped using "AI" in conversation and now call it "digital god".

AI is evidence we're smarter than our creator.

Damn, I did exactly this already for the M3 and got it working. Was hoping to make a post about it but this kinda scoops it. Bravo!

This article is about the M5 and the A18 Pro so your M3 contributions will be extremely useful!

Share your code please.

I'm currently traveling but I'll post it tomorrow. That said, you could recreate it yourself by just cloning the Asahi repos and asking Astra to make it work on the M3. I followed along with Alyssa Rosenzweig's original series of articles and tried to reproduce the milestones that she did in the same order to sanity check the work. The whole thing took about a week.

I’m going to try getting it going with codex and Claude code this evening.

I’ve just found that having more cousin projects to be inspired by really helps llms. Plus, you never know what future project/agent may find a gem in your project to apply somewhere else.

Just posted code here: https://github.com/SuperDisk/asahi-plus-slop

I don't even put it in a container. I just yolo it. It hasn't done anything bad so far. My logic is that if it nukes my filesystem then I deserve it.

Miguel de Icaza took that code and got it running on Mono back in the day, and a few months ago for kicks I compiled it and it still runs. Quite buggy though.

The industry standard VFX compositing tool Nuke also uses Tcl for its expressions and save format, it's all Tcl terms as I recall. My brother did VFX for a while and I was amused when he was having some trouble with an expression and I was like "it's a Tcl system... I know this" and helped him.

I used to do this to play Minecraft with my cousin without a router.

I swear "bifrost" has to be the most overused name in computing, possibly only behind "yggdrasil." I'm not sure what's so magnetic about those names but I've seen at least 10 different things called that.

I've been waiting over a year trying to get my wife into the country. Looks like I'll be waiting even longer now. The US immigration system is just so utterly cruel and broken.

> It is the opinion of the Board that Large Language Models (LLMs), herein referred to as Slop Generators, are unsuitable for use as software engineering tools, particularly in the Free and Open Source Software movement.

https://asahilinux.org/docs/project/policies/slop/

I wonder how they handle merging upstream clanked stuff.