crakenzak

I hate to be that guy, but the nm number is completely unrelated to the actual size of the features printed on the chip. It’s become mainly a "chip manufacturing generation" label. But generation over generation, the features printed on the chips are becoming significantly more compact and advanced, so truly "what a time to be alive"! ;)

Out of curiosity, why has there not been a slight paradigm shift in modern system programming languages to expose more control over the caches?

Interesting! Source?

Why not work with SpaceX and use bandwidth from their LEO Starlink constellation?

Sure, then just do that. No one will hold you at gunpoint and force you to use any kind of tool.

Because you were introducing exploits in your "SSO flow" in 2005. You can't have your cake and eat it too. If you want organizations (that spend billions on, and make this software free of cost) to care about security, there will usually be a slight inconvenience penalty you have to pay to comply with better patterns.

> we are releasing an early version of this model, OpenAI o1-preview, for immediate use in ChatGPT

Awesome!

Seems like their cloud compute offering (what I was curious about) is just a wrapper around Azure:

> JioCloud Compute powered by Microsoft Azure is a simple, scalable, secure and cost-effective solution …

So, can adult stem cells become any type of cell or not?

This reminds me a lot of the Hutter Prize[1]. Funnily enough, the Hutter Prize shifted my thinking 180 degrees towards intelligence ~= compression, because to truly compress information well you must understand its nuanced.

[1]http://prize.hutter1.net/

I keep getting a "Email rate limit exceeded" when attempting to create an account so I can upvote and comment.

"The market can remain irrational longer than you can remain solvent"

Its just the pharmaceutical drug name, similar to Semaglutide with brand name Ozempic/Wegovy. It'll have a brand name once its launched on the market.

> But how does a pure prediction engine make decisions? Make a judgement call? Analyze inconsistencies? Theorize?

My intuition leads me to believe that these are arising properties/characteristics of complex and large prediction engines. A sufficiently good prediction/optimization engine can act in an agentic way, while never had that explicit goal.

I recently read this very interesting piece that dives into this: https://www.lesswrong.com/posts/kpPnReyBC54KESiSn/optimality...

Greg Brockman et al.

Their Novo Nordisk one is really great too!

> leading to an estimated average increase of 17,000 to 29,000 more patents each year

Insane to me that they use this as a measure of innovation, when almost by definition it is the antithesis of innovation.

Can’t wait to see some Phi-3 fine tunes! Will be testing this out locally, such a small model that I can run it without quantization.

Feels incredible to be living in a time with such neck breaking innovations. What are chances we’ll have a <100B parameter GPT4/Claude Opus model in the next 5 years?

I think we need to focus on getting a Cargo-esque tool for C++ established first.

cries in C/C++

Very interesting, thanks for linking the research. Also, huge fan of your website!

Hmm, valid concern but GLP1-RAs reduce fat (primarily) by better glycemic control and satiation.

Unless if you're on Ozempic until you literally dry out every single fat reserve in your body and your brain has to resort to breaking down the myelin sheaths (something that can happen in ultra long distance running, hence why hallucinations sometimes happen on 100+ mile runes), then the GLP1-RAs wouldn't be able to force the body to break down the fat in the brain.

To the best of my knowledge, GLP-1's have 0 direct impact on brain fat metabolism or myelin sheath breakdown.

Side note, GLP1 based therapies are being investigated for improving brain health and potentially helping prevent the development of neurodegenerative illnesses, potentially because of an anti inflammatory feedback mechanism it activates when its able to reach tissues outside the gut. (natural GLP1 is usually broken down ~5 min in the body after release)

The insulin protein promotes glucose (not fats) uptake into the muscles (~75%), liver (~15%), and other tissue (~10%) to aggressively lower blood sugar levels back to a total circulating amount of ~5g.

Yup Casgevy & Lyfgenia! These sickle cell (and now also approved for beta thalassemia) CRISPR therapies basically work in the following way:

- Blood stem cells are removed from the patient and the CRISPR Cas9 protein outside of the body is injected to cut the gene responsible for suppressing fetal hemoglobin production (even people with sickle cell have healthy fetal hemoglobin, their adult hemoglobin gene is what causes the deformed red blood cells)

- Chemotherapy is used to kill all living bone marrow and remove all previous unedited stem cells.

- New edited stem cells are inserted, and patient recovers with new blood production being of healthy red blood cells.

I'd say a huge step forward was FDA and EMA approval, but figuring out a way to remove previous unedited stem cells with chemotherapy would be a step change in the patient experience.

The angle is that by releasing cutting edge AI research to the public openly, the relative difference between open source models/tech and closed source tech shrinks.

Whether or not you think the "value" of AI products is proportional to their performance gap vs the next closest thing or not is up to you. Very interesting PG essay I read recently talks about the opposite of this (Superlinear returns) where if you're half as good as the next competitor, you don't get half the customers, you get 0.

Essay: https://paulgraham.com/superlinear.html

I still, for the life of me, can't understand why Google doesn't just start selling their TPUs to everyone. Nvidia wouldn't be anywhere near their size if they only made H100s available through their DGX cloud, which is what Google is doing only making TPUs available through Google Cloud.

Good hardware, good software support, and market is starving for performant competitors to the H100s (and soon B100s). Would sell like hotcakes.

Is the FDA capturing any of that markup somehow? I fail to see how it would impact their decision process on approving if it had no impact on their incentives.

How about we as governments/society try to hit two birds with one stone, and pour funding into making energy cheap and trend -> ~$0 with nuclear, more efficient solar, better storage.

Then use that near free energy to desalinate whatever we need. Hell, we could regreen the Sahara and Gobi desert while we're at it too.

The CIA is one of the most wicked organizations I’ve ever heard of. And that’s just from the small amount of info we know about them.

Lol nope I’m a normal person. Gimme a captcha and I’ll (hopefully) solve it ;)