unraveller

10 ms·

31.
by unraveller 2y ago
You can ask it to talk serious without any quips or follow on questions and it's just a more sycophantic chatbot with great interruption flow. The realism is needed to unlock next-gen use-cases. Then you can trust it to be an easily ad
32.
by unraveller 2y ago
GPU saw a 10% improvement over the TPU >The TPU is so inefficient at FTs that the researchers did not use the FFT algorithm on sequences < 4096 elements, instead opting for a quadratic-scaling FT implementation using a pre-computed DF
33.
by unraveller 2y ago
Every row in a spreadsheet used to be a floor in a building full of human computers and typists. Every copy and paste you do into excel is killing hundreds of jobs.
34.
by unraveller 2y ago
Rich nations see risk, rising giants see leverage.
35.
by unraveller 2y ago
Letting algorithms bring you slop of any kind is your own fault. wontfix.
36.
by unraveller 2y ago
https://www.youtube.com/watch?v=Z3yQHYNXPws the official figure yt release vid
37.
by unraveller 2y ago
Taste is being automated too. Architecture choices, UI and broader creative strokes will not be far behind general coding ability. So then what will experience bring to the table? Short context loops and a few agents are the best we can do
38.
by unraveller 2y ago
glitchy page, crashing on ctrl key. needs a "less like this story" button and introduce topics more naturally, article titles are usually clickbait, the opposite of what information dense voice medium is good for.
39.
by unraveller 2y ago
The blackboxing makes it tougher to compare with anything, and means they can say you're holding it wrong if the router brings you a second-rate response.
40.
by unraveller 2y ago
April 2024
41.
by unraveller 2y ago
The hammer factory is not owed anything extra for the derived benefit that their product brings, why should books be different? Authors just want kickbacks for going after the open-source AI competition in their infancy. Even they see more
42.
by unraveller 2y ago
Trained on doesn't mean significant inclusion in the final state. Is it truly a violation of copyright when a user hacks out bits and pieces of easily restyled raw data points from a model to look samey? what about if it takes two mode
43.
by unraveller 2y ago
Could be why OpenAI paid them so much, to go after their open-source competition hardest of all.
44.
by unraveller 2y ago
The way it works counts if you bring prompting into it. It could easily have learned enough style chops of [author] from other sources to mimic/predict those stanzas from raw data points. Whatever the ruling one thing is for sure, plag
45.
by unraveller 2y ago
An insider's trivia game means nothing if they design the test to the trajectory of LLM capabilities and not to the real world that human's value. Let every high score get fresh news coverage to align with their updated timeline s
46.
by unraveller 2y ago
Gotta remember, this is the sloppiest it will ever be.
47.
by unraveller 2y ago
baby duck syndrome
48.
by unraveller 2y ago
What's this stuff about the model catering to ‘80%’ of generative AI tasks? What model do they expect me to use for the other 20% of the time when my question needs reasoning smarts.
49.
by unraveller 2y ago
Only it's not smart to trust an untrustworthy thing for such matters. Better to know of capabilities and judge for yourself. Also, it'd be dumb to push wholly disagreeable propaganda, so cherry-picking from an infinite set doesn&#
50.
by unraveller 2y ago
You can't quantify predictable outcomes of business rivalries so well either, but you probably don't want to do away with them for that reason.
51.
by unraveller 2y ago
So powerful ASI will arrive out of the blue sooner-than-I-thought, we've got to reg cap naow! It will be greater than anyone but it won't be able solve THAT problem or any problem created after 2026, I can tell. CEOs with little f
52.
by unraveller 2y ago
He was obsessed with IBM remaining the world leader in everything they did, changing the org structure and over hiring for research with this aim. This time feels different, you can't corner the resource of best AI engineers as the yea
53.
by unraveller 2y ago
This stolen valor mindset becomes absurd if Bach anonymously open-sources the formula he discovered and dies before releasing any music himself. Whatever music that follows, however hard fought for - even if 1:1 Bach output that he kept in
54.
by unraveller 2y ago
I have been moved by the rare piece of AI genre music. Admittedly, most of Gen AI that gets updooted is repulsive though. This emerging field is not for the faint hearted and has terrible curation chops.
55.
by unraveller 2y ago
"Pushing to help make it a reality" as in they want credit if someone else does it with this lot of LLMs. If they built something that conforms to everyone's definition of AGI I doubt they'd release that thing, unless so
56.
by unraveller 2y ago
AI songs present as a refined end-product but you don't have to treat them as anything more than killing time or B-grade inspiration for something else. Instrument skills will surely be preserved somewhere due to nostalgia, it's t
57.
by unraveller 2y ago
Something must be wrong with the value of an upvote if they are too pervasive and too unwanted.
58.
by unraveller 2y ago
A perfect AI assistant is never coming. A near perfect one still has all the context switching downsides of unlimited free use of the mechanical turk API just faster. Their problem really is in the present, no one should be buying their dan
59.
by unraveller 2y ago
Something tells me that if they did it "properly" i.e. scraped and distilled all the world's libraries directly themselves (google streetview style without storing any individual pages) instead of torrent there'd still b
60.
by unraveller 2y ago
Direct derivative works of a single work is easy to prove by model activation but input/output similarity is much easier to get outrage points. True internal function would show that no-use is required to "distribute" derivat

More ›