sinenomine

10 ms·

91.
by sinenomine 4y ago
Frankly, a ridiculously overhyped paper. Here is a dose of healthy criticism: https://twitter.com/CharlesMBrenner/status/16135817576344576... Ageing is a very important R&D field, thus it is important to disco
92.
by sinenomine 4y ago
Sorry, since the Pile and C4 https://huggingface.co/datasets/c4 (and, more generally, common crawl) and BigCode https://www.bigcode-project.org/ became available, this argument ceased being the real mo
93.
A Proposal to Accelerate Progress Towards Human Rejuvenation(fightaging.org)
2 points by sinenomine 4y ago|0 comments
94.
by sinenomine 4y ago
Expect a fierce competition of these: a flurry of new apps and old apps reinventing themselves under pressure.
95.
by sinenomine 4y ago
If you write a project for yourself, or develop a startup or an MVP it's often not worth it. Late-binding dynamically typed languages are underrated, if you can hold the mental picture of types in tour working memory, that is.
96.
by sinenomine 4y ago
Why don't people just discuss the significant heritability of big-5 trait Neuroticism and many other traits
97.
by sinenomine 4y ago
It would be more interesting if this paper studied low-level molecular pathology of ageing, i.e. protein crosslinks, advanced glycation end-products, etc. These pathologies are going to remain even after the supposed fix of epigenetic decay
98.
by sinenomine 4y ago
Data isn't the hard part here, plenty is available, even with all the necessary preprocessing.
99.
by sinenomine 4y ago
Yes, you can. It's not rocket science - lambdalabs and preemptible instances on various public clouds are possible choices. And you probably should.
100.
by sinenomine 4y ago
Someone needs to host a simple website with a timeline of every goalpost Gary moves further into public space as the years go on. This website will need ddos and legal protection, because Marcus will be furious and might sue. Just how much
101.
by sinenomine 4y ago
Ironic, since the relatively recently discovered double descent makes it clear that bias-variance tradeoff as we know it from statistical learning theory simply doesn't apply to "overparameterized" deep models. Much of old th
102.
by sinenomine 4y ago
Overproduction of elites is a very real phenomenon.
103.
by sinenomine 4y ago
This might even work, but see how much poison Elon Musk got out of his attempt? For this guild these careers are their lifeblood, and the process of minting these (I.e. ensuring the pathway for leeching from fortune500 profits exists for
104.
by sinenomine 4y ago
For people writing the checks "productive" also means "staying on good terms with the PMC". If you are outed from these circles, your career takes a nosedive. Startups were a breath of fresh air, but as we continue falli
105.
by sinenomine 4y ago
A very underrated take. Consider how the https://en.wikipedia.org/wiki/Professional%E2%80%93manageria... gets a very privileged treatment by the media and what this says about our society. There should be no sinecures
106.
The Stage Is Set for More Progress Towards Human Longevity in the Next Decade(fightaging.org)
2 points by sinenomine 4y ago|0 comments
107.
by sinenomine 4y ago
Cute, but reads like a snarky fearmongering story, a discursive pretext to ban small biotech businesses. Remember the media backlash following the humble glowing plant kickstarter project? Instead of this we need stories and cases of applyi
108.
by sinenomine 4y ago
It is also interesting to consider problems that will likely be still hard even for advanced AIs: Control of (complex) nonlinear dynamical systems over long timescales will likely pose a problem and will require high-level unsavory shortcut
109.
by sinenomine 4y ago
To add to it: a recent attempt at designing a psychometric test encoding human cognitive priors: https://arxiv.org/abs/1911.01547 - the ARC dataset. https://github.com/google/BIG-bench - the most
110.
by sinenomine 4y ago
Chronic diseases break our medical R&D model and our society is prone to normalizing these as if they were some immutable constants of life. Without reconsidering our attitude to research & development & incentives, and correspo
111.
by sinenomine 4y ago
Honestly, the board of M.D.s - the ensemble of upper-tier experts - is going to be the real tough benchmark for the years to come. Expecting to compete with this with only a few forward passes through a 0.5T-scale LLM is almost too hopeful.
112.
by sinenomine 4y ago
Interpolation is a leaky analogy. With scale these models acquire increasingly sparse discrete-like activation patterns, and there are techniques[1] amounting to pinpointing the exact locations of the knowledge and updating it freely. It se
113.
by sinenomine 4y ago
Invasive treatments should be signed off by a real person for the time being.
114.
by sinenomine 4y ago
I agree with the fundamental data/measurement bottleneck. Humans should likely concentrate our effort on scaling and democratizing this, physical, side of the equation. Still, given the apparent usefulness of basic unaided visual analy
115.
by sinenomine 4y ago
If gradient-descent in DNNs indeed has an intrinsic bias for short and simple solutions, like this paper https://arxiv.org/abs/2006.15191 shows, and if the "reasoning" is among the shortest solutions to match
116.
by sinenomine 4y ago
As you said, this is a milestone long overdue. The next frontier of transformative success in the medicine should come from finding novel, effective treatments to common chronic diseases. We need to harness AI/AGI in all possible capac
117.
by sinenomine 4y ago
It's like betting that StableDiffusion's descendants won't be able to draw hands correctly. Sure, there could be a manifold market for that, if someone is truly confident in risking (play)-money.
118.
by sinenomine 4y ago
It is hard to overestimate just how scarce even the median human doctor competency is, at the global scale. If large language models indeed can give billions access to mostly-reliable medical diagnosis ... it will be a huge humanitarian win
119.
by sinenomine 4y ago
This stock feel-good advice has led millions into depression.
120.
by sinenomine 4y ago
The person is refreshingly honest. Spending one's prime years of life as a glorofied ChatGPT instance doing data piping chores is a sad fate. We should be able to do better than that.

More ›