kromem

45 ms·

481.
by kromem 3y ago
Did Wolfram spend the 40 years following that PhD in a physics career? Or a math and computer science career kicked off by his interest in cellular automata, which led him initially to his A New Kind of Science and now this? He didn'
482.
by kromem 3y ago
The key is the oft repeated "deeply abstract." The whole thing reminds me of a book written by one of the researchers who was first reviewing the Dead Sea Scrolls. An expert on his field, he eventually writes a book about how ever
483.
by kromem 3y ago
Indeed. Here's his dissertation: https://thesis.library.caltech.edu/2597/ Does it maybe read a bit different from the post above? Roger Penrose has a Nobel Prize in physics. But his consciousness collapse interpre
484.
by kromem 3y ago
You don't think it's implausible? Reading this line didn't set off any red flags? "And that the structure of space and everything in it is just defined by the network of relations between these elements—that we might cal
485.
by kromem 3y ago
It's pretty wild to have gone from watching a lecture from Neil Turok (who is currently my favorite theoretical physicist with a "here's my idea for what physics currently has wrong with its model") and looking at Wolfra
486.
by kromem 3y ago
It's such a small playing field with too few players. I have a slide in a presentation on the topic that's an inverted pyramid, as pretty much the entire LLM field rests on only a few companies, who aren't necessarily even do
487.
by kromem 3y ago
They are keeping the landscape from becoming too centralized around another direct competitor. Open AI is now basically Microsoft. Copilot in all your docs, etc. If only Amazon was the large investor in Anthropic it encourages a similar exc
488.
by kromem 3y ago
To be fair, the 'whales' for X-Twitter were literally the ones that fronted him the cash to buy it in the first place.
489.
by kromem 3y ago
I've been increasingly focused on the parallels between QM and how we build virtual worlds. We have a number of virtual worlds where you have continuous spatial seed functions determining where things should be, but in a number of thos
490.
by kromem 3y ago
Right. So if the effect of the ketamine on depression relates to the conscious experience of taking it, and not simply the unconscious chemical interactions, then isn't scheduling the dose at a time when the person is unconscious a sig
491.
by kromem 3y ago
Hopefully the beginning of finally growing beyond RLHF. While it was useful with the less complex earlier models, at this point it's an approach that's likely holding the SotA back. I'd much rather see systems designed with a
492.
by kromem 3y ago
More correct headline: Ketamine found no better than placebo for depression when both were given to people under general anesthesia .
493.
by kromem 3y ago
Reddit increasingly reminds me of a four year old who reached into a cookie jar and now their hand is stuck because they tried to grab too much at once, but rather than let go of at least some of what they want to grab they've decided
494.
by kromem 3y ago
Another alternative explanation: general anesthesia's disassociation alone combined with an expectation of transformative psychological change results in transformative psychological change similar to ketamine's disassociation com
495.
by kromem 3y ago
I hope he goes into directly producing content. I'd happily subscribe to an unfiltered Jon Stewart's coverage of current events, and I can't imagine I'm the only one. No advertisers, no parent conglomerate.
496.
by kromem 3y ago
They are planning on moving more into the space. I get the sense this was about the China topic though, and that the AI mention is mostly in the story to get clicks.
497.
by kromem 3y ago
Yeah, they are pretty bad. But if you had said they'd be this good by now just three years ago, I'd have considered you suffering from optimism to the point of delusion. The nice thing about AI's scalability is that in anothe
498.
by kromem 3y ago
Max Tegmark has been on a roll lately with papers looking at linear representations in LLMs.
499.
by kromem 3y ago
Literally from the abstract of the accompanying paper: "Overall, we present evidence that language models linearly represent the truth or falsehood of factual statements." You didn't even read anything before commenting just
500.
by kromem 3y ago
I'm guessing that it having a "meaning of life" section could have something to do with that impression?
501.
by kromem 3y ago
I dunno man. Kinda feel like the true Techno-Optimists were the people collectively hacking together solutions under open source principles, and that quite a lot of the libertarian-eque mumbo-jumbo here invisibly stands on the shoulders of
502.
by kromem 3y ago
You jest, but don't forget that to Microsoft leadership their battle in LLMs is likely seen as even more important than all of team Xbox. Getting nerds who grew up on Zork to effectively give free problem solving training data and pay
503.
by kromem 3y ago
Yeah right. Microsoft isn't about to let go of IP rights around the most well known text adventures. But at the same time, I'd be surprised if they don't do anything with them either. The synergy between their investment into
504.
by kromem 3y ago
Actually, in Bing's implementation they added an internal monologue hidden field which is where the model outputs evaluations of the conversation state, such as if it needs to conduct a search or end the conversation. In this case, it&
505.
by kromem 3y ago
Yes, as you feared, possibly out of date. https://www.biorxiv.org/content/10.1101/2022.03.13.483276v2.... Rather than indicating export of genes from the Levant to North Africa, there seems to have been widespread
506.
by kromem 3y ago
No, it isn't. The most advanced system is the raw GPT-4 foundational model, which is not public. What is public is the RLHF instruct/chat/safety fine tuned version of the foundational model. As could be seen with DaVinci vs
507.
by kromem 3y ago
You literally just made their point. "Generative AI is going to replace actors." "Well actually productions are moving back to actors because they are cheaper than digital artists." Yes. And guess what else is going to b
508.
by kromem 3y ago
Nearly every comment is missing his take given the title here. He's not saying everything should be open source. He's saying that the right to have an open source version of anything is what we should be absolutists about. So no
509.
by kromem 3y ago
Clever approach, and as can be seen in the relative lift for GPT-3 to GPT-4, this appears to be an approach that will compound as the underlying model improves at analogy. I just wish there was an appendix with more detailed examples other
510.
by kromem 3y ago
At a certain point, I hope NLP research finally gets that the neurological processes we're attempting to model aren't a single giant unified network either and starts looking at layering models as a way to achieve success and not

More ›