alexlikeits1999

I've gone through the link you posted and the o1 system card and can't see any reference to IMO. Are you sure they were referring to IMO or were they referring to AIME?

Sorry, car accidents won't cut it. Here are the life tables: https://www.cdc.gov/nchs/data/nvsr/nvsr61/nvsr61_03.pdf which link to this spreadsheet: https://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/...

Here are traffic deaths by age group: https://www.cdc.gov/mmwr/volumes/68/wr/mm6806a8.htm. Let's pad those numbers up by a significant amount and assign 20 per 100,000 from birth, and remove those deaths from the life tables (ie let's say noone dies from traffic accidents in the USA). That still only lifts life expectancy by 1 year (from 78.6 to 79.6)

You can keep adding other parts of accidental deaths but: 1) they don't fit the story of USA dying more from driving 2) the leftovers are already below the accidental death rate from countries with much higher life expectancy 3) a big chunk of the rest is accidental falls which predominantly affects very old people so doesn't have a big impact on life expectancy

The best thing I've read about cross-country longevity, and I don't even care about its main point (ie USA healthcare spending). If you have a Patreon account or something similar please post it. I want more of this.

This is from DeepMind. They rarely release source code at the time that the paper is submitted/published. They often release source code later on. While not ideal it seems like a billion times better than the situation where they don't publish at all. I don't think any of their results are fake so if nothing else it's a signal that says "this is possible to do in a way roughly described within".

You don't have to provide them with your real name.

Also, I work in genomics and am a 23andme customer/product (although I was the latter before I the former became true).

Thank you very much for that first link. It seems to be severely under-publicised considering the huge amount of exposure that the topic gets.

Do you happen to know if rs1229984 is linked to any particular group that could make this result problematic? I noticed that the paper mentions that "There was also some evidence for a difference in years of education, and, while the size of the effect was small, this requires further investigation." That sounds like a decent-sized red flag.

That's the list and figure for women. For men you have to see the next graph which is slightly different. Then the next two graphs show you that outside of the US the opiod epidemic is not a big part of the reason. Then when you look at the 2016 estimates table you see that this is all just noise except for the UK and the US.

While I agree with the general sentiment that CR is probably not very useful for humans, I think you are either misinformed or exaggerating for the sake of generating a response.

Ad-libitum feeding is not force feeding. Noone goes and shoves the food into mice. The food is just available.

Also, CR has extended lives in many species, including dogs (see https://avmajournals.avma.org/doi/abs/10.2460/javma.2002.220...), not just mice.

That first paragraph is pretty deceptive. They are not comparing against the results from GATK 1.0.

Just a nitpick, but the original didn't do better than human scores. It does better than their human panel, who seem to suck at space invaders and a whole bunch of other games. In general DQN does worse than expected at space invaders for some reason (compared to, say, galaxian, a very similar game, at which it does very well)

Changing the frameskip is not needed any more since the implementation does max of last two frames before processing (same as what DeepMind does)

EDIT: talking about Sprague's implementation btw, not necessarily OPs.

Regarding the expectation that it would have gone 5-0 or 4-1 to either one, you'd have to have equally predicted a similarly one-sided result in the Deep Blue v Kasparov match. Rate of improvement might end up being completely domain-dependent, and hard to predict a-priori how it will turn out.

Another explanation is that people with bad livers can't handle coffee (slow to break it down), so this is just selection-bias.

Nathan Sprague has a mostly-working implementation at https://github.com/spragunr/deep_q_rl