samrus

22 ms·

361.
by samrus 9mo ago
Yeah i have fantasies of having my own email server and stuff but the spam detection is probably the 3rd thing that would have me crawling back
362.
by samrus 9mo ago
What a do nothing argument. We know where it is now. Lets quickly afapt to this situation and then we'll adapt to where it goes nexf
363.
by samrus 9mo ago
I think the main gripe peopme have is value not flowing the other way when frontier labs use training data. I think this poisoning is intended to be somewhat of a DRM feature, where if you play nice and pay people for their data then you ge
364.
by samrus 9mo ago
The metric used to gauge the efficacy of the usps was rate of ontime delivery, but whe. The same point was made about the canadian postal service, that was never brought up. Plus the trend being fixated on is between 2022 and 2024/2025
365.
by samrus 9mo ago
I believe the LLM is being sycophantic here because its trying to follow a prompt even rhough the basis of the prompt is wrong. Emporers new clothes kind of thing
366.
by samrus 9mo ago
Whats the baseline? How many did a human being get? We need to compare this to baseline to know if its good or bad
367.
by samrus 9mo ago
Yeah but writing that backend app will be much much easier now
368.
by samrus 9mo ago
Linux being opensource helped people alot who werent engineers. Its for the benefit of opensource devs who can make software for normal people to use
369.
by samrus 9mo ago
With llms being hard to test objectively, any claim made about them has to be substantiated with atleast anecdotes. The article presented some backing, if you dont think its enough you gotta present some of your own, or people cant talk you
370.
by samrus 9mo ago
Trying to follow invalid/impossible prompts by producing an invalid/impossible result and pretending its all good is a regression. I would expect a confident coder to point out the prompt/instruction was invalid. This test is
371.
by samrus 9mo ago
To understand why they syck at real world tasks, you gotta realize what LLMs are as a model. What do they model? They model static sequences, an example of which is language, the first thing they were used on. What are those? 'Sequence
372.
by samrus 9mo ago
How does that make sense with this news that cloud gpu costs have risen? That would make the subscription services more expensive as well
373.
by samrus 9mo ago
Its not supposed to generate revenue, its supposed to shift the cost to the corporarions that are abusing this loophole, and hopefully lead to them not abusing it in the future. Like how high taxes on cigarettes arent meant to make money, t
374.
by samrus 9mo ago
Yeah. Another way to collect that metric is surveying people to ask them if cancelling their gym membership was easier I do get the previous commenters hesitance though. It feels like you are holding mambani to a higher stnadard that anyone
375.
by samrus 9mo ago
As someone with psoriotic arthritis, this is just my diet (plus avoiding gluten) and honestly following it has made me feel alot better even aside from preventing the psoriosis Good initiative from the government, i wouldnt have expected th
376.
by samrus 9mo ago
And the New Deal
377.
by samrus 9mo ago
The point is that OAI has no good reason to hide the full chat logs
378.
by samrus 9mo ago
Lets get the full picture on both and let the court decide. We have the testosterone, now lets have OAI cough up the chat logs
379.
by samrus 9mo ago
Yeah but they do actually compute things the way humans did and do. Submarines dont swim the way humans do, and they arent called swimmers, and LLMs srnet intelligent the way humans are but they are marketed as artificial intelligence
380.
by samrus 10mo ago
"Actually we did find weapons of mass destruction"
381.
by samrus 10mo ago
Absolutely agree. The idea that private corporation manage our digital payments is crazy if you ever imagine that happening to physical payments. Imagine if bank of america got to decide if the dollar bill your trying to use is too damaged.
382.
by samrus 10mo ago
Locks arent meant to prevent doors from being opened, theyre meant to make it harder If thats the sort of james bond shit that has to be done to cheat on in person exams, then theres a way higher chance of being caught. Using chatgpt at hom
383.
by samrus 10mo ago
Yeah. There was a recent post about this sprt of gaming happening at Stanford
384.
by samrus 10mo ago
Absoluteky not. Actually having to contruct the flashcards embeds the information in your head to deeper level than 10 reviews could Same with taking notes in class. You can never look at them again but the most benefit comes from having to
385.
by samrus 10mo ago
The pendulum keeps swinging Trauma isnt real -> whats wrong with these kids -> we need to be more sensitive to their suffering -> everyone has trauma -> why are corpos using trauma to shovel snakeoil and coddling down everyones
386.
by samrus 10mo ago
Maybe not this in particular. But the general idea is that this is reputation laundering. Yournot reading something up to forbes' (percieved historical) standards. Your reading something some rando paid to get on forbes
387.
by samrus 10mo ago
All that tinkering is getting you dangerously close to daily driving linux. And the advantage there is that the maker isnt actively trying to get in your way
388.
by samrus 10mo ago
Basically, all human knowledge is an application of either math or philosophy, and law is philosophy, so cant be modeled by math
389.
by samrus 10mo ago
Exactly. This is exactly we get in return for compromising on quality and price with framework. Other tech is cheaper because of planned obsolescence or lock in. Im glad to pay more money to have this freedom
390.
by samrus 10mo ago
There are other men in the arena who arent tripping on their own feet

More ›