I believe the ARC-AGI benchmark fits that description, it's sort of an IQ test for LLMs, though I would caution against using the word "Intelligence" for LLMs.

Woz is just a nerd, simple as that. And he stayed true to himself and that ethos his whole life.

Critical Skills is spotting the potential bugs before they happen but in order to do that you need to have an extremely acute understanding or a have a lot of experience in the stack, libs and programming language of choice. Something that, ironically, you will not get by "vibe coding".

I tried installing and setting up the project today, it was miserable. I finally got it to work only to find out that the mistral models' tool calling does not work at all for claude code. Also, there is no mention anywhere of what models actually support anthropic level tool calling. If anyone knows if there are some open weight models (deepseek or others) I can host on my infra to get this to work out of the box that would be amazing.

I recently saw an interview without someone on the mac team and what's interesting is that the original Mac team had a lot of friction because of this philosophy. Jobs constantly asked unreasonable design constraints of his engineering team, the team prepared two laptops one with the "esthetic" laptop and one with the pragmatic design. One vastly outperformed the other and Jobs conceded, this was a constant pattern for the engineering teams at Apple. What you see before you is not the esthetic version but the pragmatic version. Sometimes Jobs was right but most of the time he was delusional.

By default it will create hard links for python packages, so it won't consume any more memory (besides the small overhead of hard links).

As far as I can tell, this only applies to the wheel contents, not to the .pyc bytecode cache created by the interpreter. If you use the defaults, Python will just create per-environment copies on demand; if you precompile with `--compile-bytecode`, uv will put the results directly in the installed copy rather than caching them and hard-linking from there.

I plan on offering this kind of cached precompiled bytecode in PAPER, but I know that compiled bytecode includes absolute paths (used for displaying stack traces) that will presumably then refer to the cached copies. I'll want to test as best I can that this doesn't break anything more subtle.

Unless it can't because you happen to have mounted your user cache directory from a different volume in an attempt to debloat your hourly backups.

In that case, you use copy OR what you can can also do, if you really care about disk usage, is use symbolic links between the drives. have a .venv sym link on drive A (raid 1) point to the uv_cache_dir's venv on drive B (raid 0). I have not tested though what happens when you unmount and sync.

... your hourly backups aren't also using the same hard-linking strategy?

They’re ZFS snapshots, so no. That’s why I’m forced to keep my cache directory in a different dataset.

Ironically, if my backup scheme were using hard links, then I could simply exclude the cache directory from backup, so I’d have no reason to do that mountpoint spiel, and uv’s hard links would work normally.

... ZFS snapshots don't produce copies of the underlying file data, do they? (Yeah, I'm still a dinosaur on ext4...)

Nothing wrong with ext4! Having choices and preferences is a good thing.

You’re correct, ZFS snapshots don’t produce copies, at least not at the time they’re being created. They work a little like copy-on-write.

I tried Multi-Account Containers to try isolate my work from my personal but it just wasn't good enough. History, bookmarks, passwords and plugins were all shared. Do I really need twitch emotes and every personal password on my work environment? No. What I found as a solution was `about:profiles` and creating a separate profile with a distinct theme to tell the envs apart, and to sync I had to add a different Firefox account but it all works quite well.

Mozilla shipped an easier-to-use profile manager in Firefox 138. Here are instructions for enabling it:

https://allthings.how/use-the-new-updated-profile-manager-in...

I'm wondering if one of the big reasons that OpenAI is making gpt-4.5 deprecated is not only because it's not cost-effective to host but because they don't want their parent model being used to train competitors' models (like deepseek).

DocLayout-YOLO model is under the AGPL-3.0 license, it's not permissive. You can't have your project under the MIT license and also use copyleft software.

I’m sorry that I didn’t know that detail, thank you so much for letting me know! I’ll read AGPL-3.0 license more carefully and check if it’s okay with MIT. If not, I’ll fix license or change model. really appreciate your help!