gbasin

Yes this feels analogous to ensembling which can help measure variance, also

Very exciting, thanks for sharing all this

Does funding matter? <insert one of a dozen studies here showing little relationship between student funding and outcomes, across and within countries>

The RLHF process performed on that chat models actively confuses their ability to estimate confidence. The raw models are quite well calibrated though

At least in the US, this will only be true before a company goes public. After that point, the shares will all be freely salable and have the same value (except for a small premium for supervoting shares which some founders may retain).

When was that not the case? Only time I can think of is A/B testing facebook.com and maybe the transition to mobile feed?

There is "long range" dependence, it's just only on the prompt: the conversation with the user and the hidden header (e.g. "Answer as ChatGPT, an intelligent AI, state your reasons, be succinct, etc."). That ends up being enough.

ngl would've killed for more alpha back then

Wow I would have killed for this in my algo trading days :) Great looking tool

Ya, luckily we already have one a large bank ready to use it and that's enough to start: https://twitter.com/garybasin/status/1624134955566788608?s=2...

Cofounder here, happy to answer questions

The 3rd one is interesting and something I hadn't considered as much before — may be applicable still in less regulated markets

Not to create unnecessary paranoia, and the odds are low, but there is a possibility of early onset alzheimer's...

some charts here https://twitter.com/Econimica/media

To play devil's advocate, these big institutional buyers moving into the single-family home market after a huge run up just feels like suckers rushing in at the top.

We're possibly at a multi-generational peak in inflation-adjusted returns to homeownership. Total population growth has essentially peaked, and working age population is falling. Expect more inflation over the next 50 years as we have more retirees (pure consumers of goods) to working age folk (who are net producers). On top of that, we've built more housing than "needed" in 99% of the country (look up the numbers if you're curious... construction has outpaced population growth).

ya, also keep everything in cache to be below a few mics

sub 10 in Java was very doable 10 years ago, fwiw. even better than collecting once a day is doing everything off-heap — C-style java :D

Sure ping me on twitter

I founded and ran a quant trading firm for about a decade :)

If your predictions are good, I'd be happy to get you $100 million in assets to manage. It's very unlikely that your predictions are good...

ok and then what? :D you still can't recover original

Nothing a little Stable Diffusion can't fix. Soon, digital memories will be just like real ones — slightly different every time we look back, but with no way of knowing what's changed

the innovator's dilemma

anti-trust law would prevent the most egregious forms of this

Take a look at Manhattan's Financial District — seems like half of the commercial buildings have already been converted to residential. Some layouts are a bit awkward but it works fine

The CoreLogic data shows that what it calls “mega” investors, with a thousand or more homes, bought 3% of houses last year and in 2022, compared with about 1% in previous years, with the bulk of investor purchases made by smaller groups.

Nothing an LCD and some cameras can't solve

Cost is not static. Building more will create economies of scale. Most importantly, overhauling the excessive safety guidelines will give you an immediate savings

While this is also possible, this is not what the statement was referring to

Yep. Some of these are held by custodians and others in smart contracts, depending on the type of "wrapper"