blobbers

5 ms·

1.
by blobbers 9d ago
Not sure why this is being posted today or getting comments. bzip3 not a terribly impressive or useful compression algo. benchmark on enwik: https://www.mattmahoney.net/dc/text.html#1703
2.
by blobbers 23d ago
The top comment on this thread was about AI models being easily distilled being a stroke of luck. This should not be surprising at all. Every new students spends tiny fractions of time learning knowledge that took many lifetimes to discover
3.
by blobbers 26d ago
Does anyone know if this border security inspection is a goon scrolling through your photos, or do they just copy the whole memory of your phone to a real “threat detector”. When I used to work in cell phones and I was debugging them, when
4.
by blobbers 27d ago
I find it slightly creepy that people are reading my past thoughts. Time to look into privacy settings. I just felt like the article wasn't particularly well written. Just because AI can produce slop, doesn't mean humans don'
5.
by blobbers 27d ago
I think that's not the author's point; he's mostly talking that time solidifies ideas as shown by the evolution of points of view. At the same time, his picture of the wave shows that over time the painting feels more distill
6.
by blobbers 27d ago
I'm confused; why is this hitting #2 on HN? I don't see the H or the N in this.
7.
by blobbers 28d ago
I'm not saying I'm better than me without kids; I'm just saying the level of freedom (high) and time management required without kids (low) is shocking and I believe it can only be fully realized in the normcore world by havi
8.
by blobbers 29d ago
I'm shocked by the amount of time people who don't have kids waste. So while PG may believe that people may have their "act already together" pre-kids, I sort of look at the amount of time people without kids waste and
9.
by blobbers 1mo ago
(such as giant automated meeting summaries we all get after an hour long slog).
10.
by blobbers 1mo ago
Yeah, but who wants to deploy a model just like that. I mean, what if that model doesn't handle the same guard rails. There's a million reasons why switching models should not be a snap decision. The back-up model is not a bad i
11.
by blobbers 1mo ago
What does openrouter actually do though? They buy a bunch of tokens from token providers and allow you to easily route queries between token processors? This seems like the kind of thing that you can have Claude write in an afternoon for wh
12.
by blobbers 1mo ago
Interesting thought: what if the idea of an open internet is over. What if we're now moving into a world of strictly KYC. The same way "The Facebook" generated massive revenue by creating a KYC world.
13.
by blobbers 1mo ago
The winner generalize really well. They're just incredibly slow. Most of the neural compressors are trying to squeeze every last bit out by sacrificing runtime. You can speed them up a fair bit and they're pretty good. fast-cmix w
14.
by blobbers 2mo ago
woah woah... a bartender becomes the head of engineering at a high flying start up? This isn't the .com. Did the bartender also go to MIT? Drug fueled founder parties? Orgies? I've clearly been doing silicon valley wrong. Or right
15.
by blobbers 2mo ago
Google abandons a piece of software into decaying crap but their scale and 'free'-ness is so large that we don't get a good replacement for years and years... On a sidenote, is there anything good to replace google news? Si
16.
by blobbers 2mo ago
Ghosting is rude. Companies should at least politely reject you. Chances are if they ghosted you, you didn't do as well as you think. You nailed a tech interview, but you only nailed part 1 of 3 and the interviewer didn't have the
17.
by blobbers 2mo ago
My experience is mostly around either ensembles of weak learners that are bootstrapped to create a strong learner, or around boosted learners where you're training on residuals. With CoT you're sort of adding entropy to your answe
18.
by blobbers 2mo ago
I'm curious if there is measurable value in diversity of thought, and if there's diminishing returns on a single models thought pattern. For example, compute X tokens with model A, then feed those into model B, etc. to get chain o
19.
by blobbers 2mo ago
Okay, I skimmed the article. Is creatine safe to supplement for children?
20.
by blobbers 2mo ago
Just pay your bill Elon.
21.
by blobbers 2mo ago
Do you feel like lobste.rs is a better version of HackerNews?
22.
by blobbers 2mo ago
I'm talking about the Watts used to power the GPUs, not the GPUs themselves.
23.
by blobbers 2mo ago
No, the public utilities aren't the ones taking the debt to fund the power generation for some direct data centers. Companies like Bloom Energy are putting up power plants right next to DCs. If datacenter power demand then these compan
24.
by blobbers 2mo ago
At least if the datacenters usage crashes, we'll have cheap power from all the infra that got built.
25.
by blobbers 2mo ago
I am have an old Pentium 4 with RDRAM, think I could sell it to them? I think it has like 256MB. Haven't turned it on in awhile. Hope the first 640KB still work.
26.
by blobbers 2mo ago
I never played this game! I didn't realize there was a RTS after Crescent Hawk's revenge. I'm looking at the MC2 source code now... it looks like there is a Linux port, meaning could probably get it working for Apple using Fa
27.
by blobbers 2mo ago
madwifi lives on...
28.
by blobbers 3mo ago
How does llama.cpp use the GPU efficiently as opposed to MLX? Is there any way to use MLX and GPU at the same time? Or does memory become a big problem? TBH, I never understood Apple hyping these neural cores because I didn't think any
29.
by blobbers 3mo ago
What happens when you find the models are terrible? The claimed results don't match? My dev cycle tends to be write a test for blah blah, add feature to satisfy test, make sure tests pass.
30.
by blobbers 3mo ago
Great, and here I thought my job was safe.

More ›