Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
sharemywin
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
10 ms
·
91.
▲
by
sharemywin
2y ago
I personally remember sh*tting on doordash...what a stupid idea.
92.
▲
by
sharemywin
2y ago
Could have been one managers bad practice.
93.
▲
They trained a new model that is good at creative writing
(twitter.com)
2 points
by
sharemywin
2y ago
|
1 comments
94.
▲
by
sharemywin
2y ago
I don't think the costs go to zero though for that.
95.
▲
by
sharemywin
2y ago
1. there are limits to natural resources. 2. UBI is basically keeping humans as pets. 3. what value do billionaires bring to the table if their insight and wisdom can be replicated by a machine.
96.
▲
by
sharemywin
2y ago
crypto has that problem. payoff before the result.
97.
▲
by
sharemywin
2y ago
if the government had a sovereign wealth fund it could have bought all the failing companies and "divested" when the market recovered. instead of being massively in debt it could have probably come out on top. granted that incenti
98.
▲
by
sharemywin
2y ago
only in a corrupt system.
99.
▲
by
sharemywin
2y ago
Not that the stock market is the end all be all of society but if a 4T tax cut isn't making the market giddy, I'm not sure anything will. once the reel hurt hits the economy like the 1T that gets sucked out of small town healthcar
100.
▲
by
sharemywin
2y ago
because it's one of those schema to dump a giant pile of sh*t on to the public. government buys reserve price skyrockets. no one uses it so the price plunges American tax payers stuck holding the bag.
101.
▲
by
sharemywin
2y ago
One is the cost of regulation and two, other things like the fear of litigation. I suppose the other way to deal with Enron would be to make it so you can't sue them for fraud. I guess that would have been another way to go.
102.
▲
by
sharemywin
2y ago
so it cost 100 times more to run and chatgpt was losing money. what kind of money are these other companies burning through?
103.
▲
by
sharemywin
2y ago
I think I read 20% of their mineral are under Russia occupation. I know what mineral rights I would be giving the US. you want them you know where they are.
104.
▲
by
sharemywin
2y ago
I'm pretty sure if you set the temp to 0 it will product the exact same output every time. its the sampling that produces the output variation.
105.
▲
by
sharemywin
2y ago
I don't think the reasoning models are LLMs. they have LLMs as a component but they have another layer that learned(reinforcement learning) how to prompt the LLMs(for lack of a better way to describe it)
106.
▲
by
sharemywin
2y ago
but these thinking models aren't LLMs. yes they have an LLM component but they aren't llms they have a component that has "learned"(reinforcement learning) to search through the LLMs concepts/word space for ideas th
107.
▲
by
sharemywin
2y ago
each level above the first is predicting concepts right.
108.
▲
by
sharemywin
2y ago
imagine if any Saas could be built using AI for ~$1.
109.
▲
by
sharemywin
2y ago
maybe they can house some nuke bases there too.
110.
▲
by
sharemywin
2y ago
crypto. and network states. curtis yarvin.
111.
▲
by
sharemywin
2y ago
got 2 words nuclear proliferation.
112.
▲
by
sharemywin
2y ago
crypto why do you think their doing this.
113.
▲
by
sharemywin
2y ago
I know farmers hate it with all those billions of $ in extra cash from selling food that was going to waste anyway. I'm sure the rural communities will hate it too when all that money isn't going to the love communities either.
114.
▲
by
sharemywin
2y ago
while it makes perfect sense in a democracy to argue about funding levels or even if the program should exist and make your voice hear through your representatives. Turning it off without any planning like its the back up server for a socia
115.
▲
by
sharemywin
2y ago
The problem that really needs fixing is not the public employees but the private contractors—and Elon Musk is one of them. exactly - it's the corrupt bidding process. everyone wants to complain about inefficant government. "In 202
116.
▲
by
sharemywin
2y ago
"Wait, but hold on. Let me confirm.” the thing is both r1 and o1 are trained on reinforcement learning so I suspect it's learned through trial and error that it makes mistakes and it should check for mistakes.
117.
▲
by
sharemywin
2y ago
you just carve out us and them circles then just make the circles smaller and smaller. look at the push right now in the US against corrupt foreign aid and the mass deportations seems like the first step.
118.
▲
by
sharemywin
2y ago
I also see RAG playing a role in the dependable fact based side. the imaginative side might benefit from it too. think like thought joggers or some kinds of save prompts that could be iterated through to generate more robust ideas.
119.
▲
by
sharemywin
2y ago
lawn mowers, fans, fireplaces, rippling creeks, waves, snow storms, rain storms, and weird hums at various wavelengths.
120.
▲
by
sharemywin
2y ago
It would be nice to be able to have it read it to me with text to speech. also, would be cool to have some kind of group based version. none of which I would pay for so I'm probably not your target audience. you might be able to create
More ›