tyho
tyho on Hacker News
I googled "history of the chemical industry" and synthetic dyes were mentioned prominently by each of the first five results returned.
It's also stores and releases light, not electricity, a bit like luminous paint, less like a battery or a capacitor.
The SIMD stuff is incredible. I have been having lots of fun with it. You can use LLMs as a scalar to SIMD transpiler, it works amazingly well.
Sure a SIMD expert writing assembly can probably do a better job than an LLM using these new intrinsics, but it’s still massively faster.
What a brutal bug. I'd never entertain a that bug in SQLite could be causing problems in code I wrote.
You can absolutely parse HTML with regex, so long as the document is finite in length. Every finite language is regular, hence can be parsed with regex's.
Eat the gamma cookie, it's the worst option for any, but with gamma it won't matter much whether it's in your hand, pocket or gut.
Pocket the beta emitter, a little bit of shielding will make a big difference.
Hold the alpha emitter, if you hold it with just a pair of fingers you will be able to reduce you dosage a lot compared to holding it tightly.
Wow, what a free society! In the UK if you refuse to unlock your device you can be imprisoned indefinitely! In HK it's just one year!
4chan send their packets to their ISP, not the UK.
> following a highly publicised mass-murder of children which was (entirely wrongly) being blamed on asylum seekers.
The mass-murder was a consequence of the asylum system. Given what is publicly known, parents of Axel Rudakubana were overwhelmingly likely to have been asylum seekers.
What's the easiest way to get these local models browsing the web right now?
FFS, it's too late for me to edit. You are of course correct.
> The 2D map analogy was a nice stepping stone for building intuition but now we need to cast it aside, because embeddings operate in hundreds or thousands of dimensions. It’s impossible for us lowly 3-dimensional creatures to visualize what “distance” looks like in 1000 dimensions. Also, we don’t know what each dimension represents, hence the section heading “Very weird multi-dimensional space”.5 One dimension might represent something close to color. The king - man + woman ≈ queen anecdote suggests that these models contain a dimension with some notion of gender. And so on. Well Dude, we just don’t know.
nit. This suggests that the model contains a direction with some notion of gender, not a dimension. Direction and dimension appear to be inextricably linked by definition, but with some handwavy maths, you find that the number of nearly orthogonal dimensions within n dimensional space is exponential with regards to n. This helps explain why spaces on the order of 1k dimensions can "fit" billions of concepts.
I wrote some code today and I relied on the new behaviour. I felt a little dirty.
The single example given here is incorrect. Since Go 1.22 loop variables have per-iteration scope, the "broken" code is deterministic and correct.
https://go.dev/blog/loopvar-preview
Does anyone really use an X GUI on BSD?
A crude way:
To watermark:
First establish a keyed DRBG.
For every nth token prediction:
read a bit from the DRBG for every possible token to label them red/black.
before selecting the next token, set the logit for black tokens to -Inf, this ensures a red token will be selected.
To detect:
Establish the same DRBG.
Tokenize, for each nth token, determine the red set of tokens in that position.
If you only see red tokens in lots of positions, then you can be confident the content is watermarked with your key.
This would probably take a bit of fiddling to work well, but would be pretty much undetectable. Conceptually it's forcing the LLM to use a "flagged" synonym at key positions. A more sophisticated version of a shiboleth.
In practice you might chose to instead watermark all tokens, less heavy handedly (nudge logits, rather than override), and use highly robust error correcting codes.
There are way better ways to watermark LLM output. It's easy to make it undetectable, which this is'nt.
I tasted the coin cell bitterant when I replaced my Airtag battery. Didn't taste very bitter to me.
I have been using txtar to provide context to LLM's recently.
Is that actually happening? I use o1 a lot for brainstorming etc, but I don't think I could build a program of any significant size without understanding how to program myself so to guide o1 and stitch it's outputs together. It's also not at all uncommon for me to have to do stuff completely myself, it doesn't have great insights into deep problems for instance.
I’m imaging a variable focal length across the image plane, decreasing as the distance to the centre increases.
Yes. Do people not reliably dip beams in the US? that must be horrible!
All psychology research falls in one of two categories:
a) common sense intuitive result
b) does not replicate
There isn't really anything you can do to convince me that your team has the expertise to maintain a browser after this. It doesn't matter that you have fixed it, your team is clearly not capable of writing a secure browser, now or ever.
I think this should be a resigning matter for the CTO.