wegfawefgawefg

Learning piano makes me better at music. Learning the violin is then easier.

i think a lot of people are fixating on the market.

I will share a metaphor you can spread.

I run a mile every morning not because it is the most calorie efficient way to get around, nor because it is the most monetarily productive use of my time, but because it keeps my legs strong and me healthy.

Go read the Intrisic Curiosity Module papers, 1 and 2.

this was voted down, but the poster had expressed doubt that it had still been taught this way, and I was providing information that it was still taught thid eay even at a high end high school tenish years ago.

I thought it reasonable to share.

went to school till 2012, highly esteemed US high school, was still taught then

This is bait.

i used to run debian on an intel macbook air. regular debian. was pretty nice.

sir why dont we dig the trench with spoons instead of shovels, that way we can hire ten times as many men for twice as long!

4o is multimodal, thats the whole point of 4o

sure.

to be honest i dont really care if they work the same or not. I just like that they do work and find it interesting.

i dont even think peoples brains work the same as eachother. half of people cant even visually imagine an apple.

Neural networks seem to notice and remember very small details, as if they have access to signals from early layers. Humans often miss the minor details. Theres probably a lot more signal normalization happening. That limits calorie usage and artifacts the features.

I dont think that this is necessarily a property neural networks cant have. I think it could be engineered in. For now though seems like were making a lot of progress even without efficiency constraints so nobody cares.

There are so many points to consider here im not sure i can address them all.

- Airplanes dont have wings like birds but can fly. and in some ways are superior to birds. (some ways not)

- Human brains may be doing some analogue of sample augmentation which gives you some multiple more equivalent samples of data to train on per real input state of environment. This is done for ml too.

- Whether that input data is text, or embodied is sort of irrelevant to cognition in general, but may be necessary for solving problems in a particular domain. (text only vs sight vs blind)

they take in many samples of touch data

overfitting vs generalizing

also we all know real people who fail to generalize, and overfit. copycats, potentially even with great skill, no creativity.

almost all of the models are wrong about their own architecture. half of them claim to be openai and they arent. you cant trust them about this

"those who have more capital have an advantage"

how about power consumption, cache size, etc?

what i know is that in my particle simulations and multithreaded compilation tasks it is easily 2-10x faster. where before i could choke on 1 million particles, now im choking on 30mill.

i specifically bought this processor for its rust compilation times.

Actually companies can just wait. Multiple times my company has said: "a new model that solves this will probably come out in like 2-4 months anyways, just leave the old one as is for now".

It has been true like ten times in the past two years.

wat... my 9950x cpu that i just bought is way faster than the similarly priced cpu i bought 6 years ago.

The difference is night and day. What are you talking about.

a sexy adventure. i met a hot japanese girl online. i wanted to meet her but during covid japan was closed. so i enrolled in a year at a japanese language school to get a student visa.

well now we are married, baby is due soon. and i can speak japanese.

in a couple years well be making a slightly larger cottage back in the US and then ill be a cottage man again.

whenever we are having a tough day, we just think of waking up to the blue sky and chirping birds in the countryside. but now the vision includes little feet

I acknowledge that industrial farmers go to university, etc, but of the 8 billion people, most of the farmers probably dont really know whats going on. they just know what to sorta do with the plants.

which was true 10,000 years ago before humans even know what cells or photosynthesis was. people had farms then.

so the point is, even without a genuine understanding, if the causal relationship between seeding parts of the ocean and the desired result is achieved, then an understanding wouldnt be necessary.

consider the following hypothetical situation:

lets say your car tires pop.

Person A: "I will paint your car tires red. That will fix them."

Person B: "painting my flat car tires red wont fix them."

Person C: "well youre just being defeatest. we have to do something".

Person B: "..."

you dont really need to understand how something works to know if it works.

theres 8 billion of us because of farming. most of the farmers dont really know how it all works.

woudln't it be cool to make nodules and go drop them in barren parts of the ocean and then they become seeded?

This is a great question, and I dont understand why it was downvoted.

officer she told me she was 4 billion years old

as a once remote cottage guy, yes you can have internet. kids love it out there. you bring your friends.

its better out there man. i miss it.

I was waking up with no alarm to blue sky and chirping birds, and then I could read papers and to pytorch all day. Never heard a car.

this doesnt apply to 90% of programming jobs so im not so sure the money growth expectation disilusion angle is right.

I generally assume it is due to lack of agency.

I can not believe this was voted down. It is simply an assertion of fact. Whether true or not, seems reasonable and most people would agree with it.

Really even this made people get angry?

To whoever downvoted this. I used to come to hacker news years ago and see 90% programming related things: Type system abuse, extreme common lisp code, libraries and tools. It was pretty much all tech debate all the time. The worst days would be 70% tech. Today thats a good day.

Nowadays theres a lot more californians being mad at elon, poetic eulogies, paid articles about "how to write good" or how "you should write even if people dont read it."

People will try to say it hasnt changed but it has changed. It is about half as tech related, and about 10 times as reactionary in the same way the rest of the internet has become more dumb.

You can be mad if you want, but everyone can see it, and thats why some people I know in real life are referring to hacker news as orange reddit in a denigrating way. Reading hacker news doesnt confer programming street cred like it did ten years ago. That isnt because im cynical, or want it to be true. It just is true.

my tech friends in the valley call it "orange reddit".

wish the the engagement on here was generally more sophisticated.

only one part. applies to zero of the rust code i write. formats horrendously on mobile.

you gotta work on this more before putting it on hacker news