samrus

16 ms·

331.
by samrus 8mo ago
Live learning has actually been a pretty interesting idea in ML for a long time that i dont know why doesnt get more effort put into it. Probably cost. But itd be really cool to have an LLM that gets finetuned on your data and RLs from your
332.
by samrus 8mo ago
People have. Microsoft wont since their incentive structure is fucked It actually could be built reliably if the verification is easy
333.
by samrus 8mo ago
Yes. Thats how all advancement in human knowledge happened. Small and incremental forays out of our training distribution. These have been identified as various things. Eureka moments, strokes of genius, out of the box thinking, lateral thi
334.
by samrus 8mo ago
You might have misread it. Texas' model is decriminalizing teens sexting, not criminalizing it
335.
by samrus 8mo ago
Its like how even nuclear reactors still just boil water. Its kinda funny. Whats the solar of monetization?
336.
by samrus 8mo ago
Its interesting. But whats the performance actually like? He said a third of the PRs are outright rejected for being garbage, but it look like thats manual, not automated. But what fraction is accepted?
337.
by samrus 8mo ago
Scaling alone wont get us to AGI. We are in the latter half of this AI summer where the real research has slowed down and even stopped and the MBAs and moguls are doing stupid things For us to take the next step towards AGI, we need an AI w
338.
by samrus 8mo ago
What your describing is a low trust society. If you disregard the social contract like that, then people wont owe the "the world" anythign either. Collaboration and civics goes out the window. If you want to look at what kind of a
339.
by samrus 9mo ago
Wrong. The problem with bitcoin is its volatility itself. If it doubled in value, that means its undergoing hyper deflation, which makes it a terrible store of value. Bitcoin's proponents are its own biggest liability. People who see i
340.
by samrus 9mo ago
Whats your baseline for volatility? 0%? Outliers aside, metals are still the most stable store of value thats widely used And the P/E and product announcements your touting as inherent value are for generating value, not storing it. If
341.
by samrus 9mo ago
One problem is the risk of false positives messing up the "training"
342.
by samrus 9mo ago
I wouldnt describe the 2010s as low pay for devs
343.
by samrus 9mo ago
Python dev is cheaper and faster though. People arent gonna kill velocity by making their backend in c++ so the devs can have seperation of concerns, something that can, and should, be self enforced with discipline
344.
by samrus 9mo ago
Fellas, is it fragmentation to natively support linux?
345.
by samrus 9mo ago
Man i love the idea of this sort of thing. Surviving in civilization but in a more natural/primal manner. Ive heard it be called urban camping or stealth camping. A youtuber i watch who also scratches that itch is steve wallis, who rec
346.
by samrus 9mo ago
This specifically, and llms in general, remind me how apt it is that robot is the slavic word for slave
347.
by samrus 9mo ago
I kinda hate that it settled. I fully understand the plaintiffs not wanting to proceed, but i really wish the sheriff was actually punished for what he did. This sort of power tripping should be a fireable offence
348.
by samrus 9mo ago
I mean problems not worth solving, because theyve already been solved. If you need to just do the grunt work of retrieving the solution to a trite and worn out problem from the models training data, then they work great But if you want to d
349.
by samrus 9mo ago
Not the core foundation model. The foundation model still only predicts the next token in a static way. The reasoning is tacked onto the instructGPT style finetuning step and its done through prompt engineering. Which is the shittiest way a
350.
by samrus 9mo ago
Exactly. I go back to a recent ancestor of LLMs, seq2seq. Its purpose was to translate things. Thats all. That needed representation learning and an attention mechanism, and it lead to some really freaky emergent capabilities, but its train
351.
by samrus 9mo ago
Yeah. I get charging for hosting an app in your store, that requires work to build the ecosystem and security and stuff. But you have to draw the line on a platform and ecosystem and vetting that someone else has done. What did apple do to
352.
by samrus 9mo ago
One argument for the robots is that they can be programmed to drive safer, while humans cant. But that depends on reliability, especially in unforseen (and untrained-upon) circumstances. We'll have to see how they do, but they have bee
353.
by samrus 9mo ago
But would a human be driving at 17 in a school zone during drop off hours? Id argue a human may be slower exactly because of this scenario
354.
by samrus 9mo ago
I hate the classic apple users' "mom" argument. Why are all your moms morons? And why do you want to fuck up the entire mobile landscape to baby proof it for them. Im not gonna ruin my experience with technology because you d
355.
by samrus 9mo ago
This is low even for apple. They havent earned commision on this at all
356.
by samrus 9mo ago
Yeah. Because you'll think you have a flight to beijing when you dont
357.
by samrus 9mo ago
The harder it is to do, the more the targets guard will be down In this case, sending your malicious image through a fake email might get flagged, or even not opened by someone whos been trained in infosec enough to be suspicious of these t
358.
by samrus 9mo ago
It works for hard problems when the person already solves it and just needs the grunt work done It also works for problems that have been solved a thousand times before, which impresses people and makes them think it is actually solving tho
359.
by samrus 9mo ago
Regardless of how addictive it is? So the same argument applies to heroin? Shoukd heroin be legalized and allowed to be sold outside of schools?
360.
by samrus 9mo ago
I think the general incompetence in the decision making behind windows is also a driver

More ›