sarker
ketman hetman
Suddenly I cannot remember the color of your eyes
Or the things we said as we stood together for the last time
User ID: 636
In San Francisco, people who aren't SWEs hate all the "techbros" and your average dead end L5 is hated, or at least considered uninteresting, by all the other average dead end L5s. You might get some respect if you work on something unusual (and you can feel smug around people who live on the peninsula), but converting your work to respect requires some cloutmaxxing (usually by poasting on Twitter, attending the right parties, and name-dropping). Even then, your status is limited to those who travel in the same cloutmaxxing circles. But if you do all this, carefully manage your bubble, and gas up your self conception, sometimes, just maybe, you can look in the mirror and convince yourself that you are high status.
sarker
ketman hetman
7hr ago
It's one thing to say something is in the LLM's memory. It's quite another to say it's doing some kind of lookup in a database or into the training data.
sarker
ketman hetman
20hr ago
What does that memory look like for an LLM if not reference to a database?
The memory is the weights of the model. The first stage of training an LLM is next token prediction - the LLM is shown a block of text and is trained to produce the next word (technically, part of a word, but that's not important). Internally, the model manipulates the numeric representation of the input tokens as points in a high dimensional space. The model produces a kind of probability distribution over all the words it knows for what the next word might be, and the weights of the model are adjusted so that it's more likely to expect the correct word.
The result is that every token, with some context before it, that the model sees leaves some kind of "impression" upon the model. The details are fuzzy, but things like word choice or style are probably represented in some regions of that high dimensional space, which is what lets the model say that something "sounds like" Shakespeare.
Anthropic have some public research around trying to look into a language model to see what's going on in that latent space and how neuron activations relate to concepts. It may be difficult to read without any ML background, but fortunately you can now feed it in to an LLM and have it explain anything you don't understand.
sarker
ketman hetman
1d ago
The most damning blow to the 80/20 interpretation comes from a 2009 blog post by a co-founder, which shows that actual messaging behaviour follows a similar distribution for both genders:
Does this really make it better? That women hold their noses and message perfectly average guys they think are ugly seems like cold comfort for the mids in question.
sarker
ketman hetman
1d ago
If I gave you a snippet of Shakespeare and asked you to guess who wrote it, I expect the Bard would be one of your top choices. How are you doing that if you don't consult Google or your Shakespeare box set?
Each token that the model sees in training updates its view of what sort of things are associated and in what way. Elements of style or topics may be clustered somewhere in the high dimensional latent space with the corresponding authors.
sarker
ketman hetman
1d ago
I don't know if I should mention this, but Costco avocados are kino. Basically never deal with a bad avocado. Just don't get anything from South America. Mexico or USA only.
sarker
ketman hetman
1d ago
Why gø thróügh âll thāt èffôrt whën pūttïng ã búnch óf shît âróūñd thé lēttérß îs súrêly jùst ás gōōd?
sarker
ketman hetman
2d ago
I'll admit I didn't read every word of these articles but they are mostly discussing content blocking that Russia, China, et al have been doing for decades.
I didn't see any evidence for this claim:
Every year they test a rogue version of the TCP/IP stack. It utilizes a Russian National Domain Name System to keep RUNet, which is like a parallel universe to DARPANet back in the days of the Cold War, that’s distinctly different from ICANN’s standards.
It's also not really clear to me why this would even require a "rogue version" of TCP/IP, what a "rogue version" would mean, or why you'd need to reinvent the TCP/IP wheel to force everyone to use your own DNS servers (which btw usually talk over UDP and not TCP).
sarker
ketman hetman
2d ago
Рунет simply refers to the Russian language Internet. Where are you getting this from?
sarker
ketman hetman
2d ago
There's at least one way.
I can believe that the median horse life 130 years ago was net negative. Seeing a horse being beaten literally broke Nietzsche's brain.
Even if it was net positive, unless you'd like to bite the bullet of the repugnant conclusion, I don't find it to be too compelling.
sarker
ketman hetman
4d ago
I'd probably shoot the guys behind the UK weights and measures act who left us with a divergence in the UK vs US fluid ounce and who retardedly redefined a gallon as 160 fl oz instead of 128.
sarker
ketman hetman
4d ago
Twelve years? My next explanation is that you are unc and things are different among the zillennials and zoomers.
sarker
ketman hetman
4d ago
IME the rates of WMIF dating relationships and marriages are roughly the same,
I haven't really found this to be the case. I think you are from the East Coast? Maybe things are different over there.
sarker
ketman hetman
4d ago
Human population decline is already priced in thanks to demographic transition. No AI required.
I'm pretty sure that, if you could ask them, a horse would rather be a random horse today than a random horse 130 years ago.
sarker
ketman hetman
4d ago
From my experience (i.e. the outside looking in), the women do date white guys with some regularity. But when it comes to settling down, Mother and Father have strong opinions about the need to find a good [insert intersection of 2-10 ethnic, religious, and/or astrological criteria] boy to marry. Frequently, the women seem to go along with it, or aren't disagreeable enough to tell the parents to stuff it.
For the men, the pattern I tend to see is little success in the dating market so I assume they maybe meet a girl that matches the criteria above to settle down with via some kind of guided parental arrangement. But this is speculation on my part.
sarker
ketman hetman
4d ago
The youth who actually grow up in the US, immersed around American culture and people and aren't in hard racial enclaves become very similar to everyone else.
They are similar in that they wear blue jeans and speak with an American accent, but the number of Indian kids I grew up with who ended up marrying Indians is staggering.
Similarly, a crown is simply a gold coin with a crown on it. No problem!
There's a basically inexhaustible fund of such examples of all kinds of currencies.
Mark 12:42: "And there came a certain poor widow, and she threw in two mites [lepta], which make a farthing [quadrans]."
There was no such coin as a "mite" in Britain at all, it merely meant "a coin of low value" due to a low-value coin of such a name in the Netherlands - so that's what they used in the translation.
sarker
ketman hetman
4d ago
It does seem that escudo has the currency meaning in Spanish though.
An autistic fixation on accuracy in translating currency names in literature is basically a high modernist project. The based and lindy approach is to pick a reasonable substitute. Matthew 20:2 KJV: "And when he had agreed with the labourers for a penny [orig. dēnarion] a day, he sent them into his vineyard."
That's household income. Top 1% individual income is something like 450k.
More options
Context Copy link