Hacker Newsnew | past | comments | ask | show | jobs | submit | linolevan's commentslogin

I believe SmartOS is a distro of Illumos (in the same way that debian is a distro of linux).

I'm not convinced that LLM training is at such a high energy use that it really matters in the big picture. You can train a (terrible) LLM on a laptop[1], and frankly that's less energy efficient than just training it on a rented cloud GPU.

Most of the innovation happening today is in post-training rather than pre-training, which is good for people concerned with energy use because post-training is relatively cheap (I was able to post-train a ~2b model in less than 6 hours on a rented cluster[2]).

[1]: https://github.com/lino-levan/wubus-1 [2]: https://huggingface.co/lino-levan/qwen3-1.7b-smoltalk


Had a small discussion about this on an OP on bsky. A somewhat interesting discussion over there.

https://bsky.app/profile/simonpcouch.com/post/3mcuf3eazzs2c


Tried it within LMStudio on my m4 macbook pro – it feels dramatically worse than gpt-oss-20b. Of the two (code) prompts I've tried so far, it started spitting out invalid code and got stuck in a repeating loop for both. It's possible that LMStudio quantizes the model in such a manner that it explodes, but so far not a great first impression.

Are you using the full BF16 model or a quantized mlx4?

Not sure what the default is – whatever that was. It's probably the quantized mlx4 if I had to guess.

This is a crazy change. I wonder if part of the reasoning is that sites without a robots.txt tend to be very low-quality. Search is a very hard problem and in a world of LLM-generated internet, it's become way harder.

My take: google marketing found a ploy to make "google" look like a better nettizen than the AI companies that hammer away on sites to the level of a DDOS attack.

its not true, just a seo posting stuff and another person thought it was legit

Honestly, not crazy. This is should have been how it always was. Why should search engine crawling be opt-out rather than opt-in?

This is a cool subcommunity! Had no idea there were still open problems that people were working on. Surprised to see human intuition is still around – would have expected a solution through pure brute force.

The state space is much too large for generic brute-forcing. The number of possible patterns in a 16 x 16 grid is already roughly the number of atoms in the universe, or 10^31 years in Planck time units.

In that respect, it reminds me a bit of the busy beaver problem.

I wonder: consider the decision problem of determining whether or not a given still life is glider-constructible. Is this problem known to be undecidable?

It's straightforward to show that an "inverse" of this problem -- given an arbitrary glider construction sequence, does it result in a still life? -- is undecidable, because gliders can construct patterns that behave like arbitrary Turing machines.


My understanding is that the only still-lifes known not to have a glider synthesis are those containing the components listed at [0], which are 'self-forcing' and have no possible predecessors other than themselves. Intuitively, one would think there must be other cases of unsynthesizable still-lifes (given that a still-life can have arbitrary internal complexity, whereas gliders can only access the surface), but that's the only strategy we have to find them so far.

[0] https://conwaylife.com/forums/viewtopic.php?f=2&t=6830&p=201...


> Maybe it's time to try pushing the envelope on this: what's the biggest blobbiest most spacedustful period-4 c/2 orthogonal spaceship that current technology can come up with? Might there be some kind of extensible greyship-like thing that escorts a patch of active agar instead of a stable central region, that might allow an easier proof of non-glider-constructibility?

I always enjoy the absolutely incomprehensible GoL jargon


Is it that easy though? Because the Turing machine constructions we have in the game of life are clearly not still lifes, and I don't know if you can construct a Turing machine which freezes into a still life upon halting.

You can make a Turing machine that contains self-destruct circuitry which destroys all moving parts upon halting. The resulting pattern will be a (pseudo) still life.

Since GoL is Turing Complete,is such an inconstructable pattern an example of godels incompleteness theorem? I feel like I must be confusing some things here.

Aah, but construction in GoL is not limited to gliders...still.

The secret proxy trick is something I expect to become standard at some point in the near future. I first saw this trick being used in Deno Sandboxes (https://docs.deno.com/sandboxes/security/) but it's cheap/easy to implement so I'd be surprised if this doesn't become the standard for a lot of these BaaS platforms.

I'm wondering in what ways is this similar/different to https://github.com/DGoettlich/history-llms?

I saw TimeCapsuleLLM a few months ago, and I'm a big fan of the concept but I feel like the execution really isn't that great. I wish you:

- Released the full, actual dataset (untokenized, why did you pretokenize the small dataset release?)

- Created a reproducible run script so I can try it out myself

- Actually did data curation to remove artifacts in your dataset

- Post-trained the model so it could have some amount of chat-ability

- Released a web demo so that we could try it out (the model is tiny! Easily can run in the web browser without a server)

I may sit down and roll a better iteration myself.


I guess chat-ability would require some chat-like data, so would that mean first coming up with a way to extract chat-like dialogue from the era and then use that to fine-tune the model?

I'm not the OP of the claim (and I love VLC) but maybe they're referring to this early 2018 issue: https://trac.videolan.org/vlc/ticket/19723 which seems to be being actively worked on.

There's also https://code.videolan.org/videolan/vlc/-/issues/25651 but that's an off by one error so likely not really relevant to video playback for the average user.


For what it's worth, I've found the Rust-y explanations well worth the read! Please keep them in, it's really interesting to read about the technicalities.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: