frontsideair

I was able to confirm that it actually runs on ANE, I'm impressed.

> Apple locked it behind Siri. apfel sets it free

This doesn't feel truthful, it sounds like this tool is a hack that unlocks something. If I understand it correctly, it's using the same FoundationModels framework that powers Apple Intelligence, but for CLI and OpenAI compatible REST endpoint. Which is fine, just the marketing goes hard a bit.

> Runs on Neural Engine

Also unsure if this runs on ANE, when I tried Apple Intelligence I saw that it ran on the GPU (Metal).

Yeah, the initial experience with no colors doesn’t look great. I can implement this when I have some free time, if you feel like doing it please feel free to open a PR. Thanks!

14B Qwen was a good choice, but it became outdated a bit and seems like the new version of 4B surpassed it in benchmarks somehow.

It's a balancing game, how slow a token generation speed can you tolerate? Would you rather get an answer quick, or wait for a few seconds (or sometimes minutes) for reasoning?

For quick answers, Gemma 3 12B is still good. GPT-OSS 20B is pretty quick when reasoning is set to low, which usually doesn't think longer than one sentence. I haven't gotten much use out of Qwen3 4B Thinking (2507) but at least it's fast while reasoning.

Ollama adding a paid cloud version made me postpone this post for a few weeks at least. I don't object them to make money, but it was hard to recommend a tool for local usage and make the first instruction to go to settings and enable airplane mode.

Luckily llama.cpp has come a long way and was at a point that I could easily recommend as the open source option instead.

This is the command probably:

  sudo sysctl iogpu.wired_limit_mb=184320

Source: https://github.com/ggml-org/llama.cpp/discussions/15396

I'm interested in this, my impression was that the newer chips have unified memory and high memory bandwidth. Do you do inference on the CPU or the external GPU?

Good point, let me add a quick note.

Thank you, it was the integral part of the whole post!

According to the benchmarks, this one is improved in every one of them compared to the previous version, some better than 30B-A3B. Definitely worth a try, it’ll easily fit into memory and token generation speed will be pleasantly fast.

This is the first time I’m hearing about Nebula. How does it compare to Tailscale?

I saw from the comments that they provide smaller binaries, which may have worked on my Raspberry Pi. Maybe I’ll give it a try one day.

https://tailscale.com/kb/1207/small-tailscale

Thanks, this could’ve worked for my Raspberry Pi! I would’ve tried it if it was still in commission. Next time?

That’s exactly why I put the spend limit, and this was the first time to confirm that it actually works.

Exactly, I had a spend limit since I didn’t want to break the bank. It’s back up now.

Synthetic benchmarks should be retired. Build a real app, measure real impact. Unless your users insert 1000 rows, of course.

Other than that, I agree with the general premise of React becoming the new IBM, that's totally fair.

This looks like a treasure trove that I somehow missed! I'll give it a more thorough read later, but gave my star for now, thanks!

I'm really interested in `home-manager` honestly, and while I agree about the benefits of declarative package management, I find it easier to type commands at a terminal like `nix profile upgrade pkg` or `nix registry pin nixpkgs`, with the guarantee of rollbacks.

Since when I'm editing a text file to update a package, I have to look for the latest version separately and copy/paste it into my editor. If I somehow mess it up the file is broken, while no harm is done I still find this workflow a bit brittle.

If there are home-manager commands I missed to do this, I'd be eager to give it a try. That kind of workflow would remind me of running `npm upgrade pkg` and have it reflected on a `package.json` file.

If you don't care about political views of presenters, there's no need to bring inclusion to the table; just consider their technical talents and call it a day.

If you actually care about inclusion, you need to think about how including backwards-views that oppose inclusion itself could affect your effort. Banning only physically violent behavior is behind the times, it's obvious there are many ways of harming someone without lifting a finger. If you're supporting inclusion for inclusions's sake, including everyone may make sense. But if you really care about people who depend on you including them, it's obvious some measures must be taken to create a safe environment for the underrepresented.

I'll just put this here: phpsadness.com

I'm starting to think that I should go back to GoboLinux from Ubuntu.

I know this is about the firing of a reddit mod, but I can't shake of the feeling that the people behind this petition care more about the banning of hateful subs. This is a retribution for them.

I really feel disappointed when people think free speech is more important than anything. Yes, it is important, but we defend it for a reason. Because it protects the weak. But when I see people defending it so they can harass or otherwise spew hate speech about minorities, I wonder where did we do wrong. Maybe, just maybe, we treated it like a dogma and then it backfired? It should've been a means to an end, not the other way.

It's been said, but really, it should be opt-in rather than opt-out. If you really need this, just ask. If I'm feeling particularly nice that day, I might even allow it, but when I discover their mischief after they've done it, there's no way I'll be okay with it.

If there's one good thing, they remove the data they've collected from you if you opt-out. (But it'll be already fed to the machine learning system, I guess.)

Author here. I haven't seen SQRL and I'll check it in detail later. It looks like a more complicated thing than what I did and much more work compared to a simple, hackathon-like project.

My project works only on Dropbox and it doesn't require Dropbox to support any other authentication mechanism. It just uses OAuth2 behind the scenes.

Well of course it has no anti-malware features, the only protection it has is that no credentials are sent to the client, just an authorization token, which you can disable using your phone.

I supported Drop-dropbox.com when I first heard about it, but since nothing's changed since then I guess it was unsuccessful.

I'll try to answer to this as best as I can, since I'm new to this feminism thing there's bound to be stuff that I'll miss.

Well first of all it's very nice of her approving the use of image and Playboy too for not going after their rights. But I think it's far fetched to see her approval as a positive message. There are women out there contributing to the objectification of women too, for whatever reason they have. (One weird side of this is people think women are shielded from criticism of their own objectification, but it's a delicate matter to say the least. One has the right to objectify themselves so it's hard to say something without getting in the way of their right to self express.) There are more than one side to this issue; but it's not about her consent, it's about how it might be contributing to boys club image of tech.

It's easy to see that we have a problem of the lack of women in tech. I find this very depressing since they're able as much as men are and it looks wasteful to dismiss half the population. I think we're getting better each day, but it does not happen magically. People fight for it and will keep fighting until there's no discrimination based on sex. I accept that Lena image is one of the minor issues, but I still see this as one of the factors that drives women away. This boys club image of tech gives the implicit message that women are not wanted here.

I know there's no nudity in the image and one needs to research to find its origins in Playboy so it seems unlikely to come across it. But the real reason here is if we're willing to combat sexism, it will give women comfort that we're willing to change tradition to be more welcoming. I believe actions speak louder than words, and if the reason to keep the tradition is not all that important, we should do it.

(I also don't think it's fair to compare it to an image of a naked guy, since it's unlikely to drive boys away from tech. It's not just naked guy vs. naked girl, the context and the message makes a world of difference.)

I understand your point, but I believe some tradition must be put to rest at some point as we grow wiser. On the other hand, Wikipedia says the Lena image was remastered in 2013 so it may lose its historical significance if the remastered version is used from now on. (It's one of those moments where making something better actually makes it worse.)

Oops, my bad for not checking. Thanks for clarification!

Maybe Bellard is reading these comments. I made a very unpopular remark about his use of Lena.jpg and while I believe it's mostly coincidence, he used an imageset without Lena.jpg.

If you're reading, Bellard, hi!