> people who use youtube as a sleep aid - don't you get woken up by ads?
I pay for Premium for that reason. I don't have any other TV/streaming services, so while I don't especially like giving Google money, I use it enough that the expense is well-justified.
I throw on video game let's plays or long form retrospectives. If I'm left to myself, I'll end up keeping myself awake thinking about stuff.
Funny enough, I usually spend more time picking what to sleep to than I actually spend awake once it starts playing. Most times I'm dead asleep within the first minute or two after spending five minutes making selections. If it goes quiet, I wake up.
> If I'm left to myself, I'll end up keeping myself awake thinking about stuff
That's the best part of disconnecting and going to bed, being cozy in the darkness with nothing but your thoughts and emotions. Interesting that others try to avoid that, thanks for sharing :)
Depends on the type of thoughts that frequent your mind. Dozing off to sleep thinking about chatting with friends is quite different from worrying about your ailing family member. I assure you that the latter isn't something to look forward to for most people.
They're just outputting tokens that resemble a reasoning process. The underlying tech is still the same LLM it always has been.
I can't deny that doing it that way improves results, but any model could do the same thing if you add extra prompts to encourage the reasoning process, then use that as context for the final solution. People discovered that trick before "reasoning" models became the hot thing. It's the "Work it out step by step" trick but in a dedicated fine-tune.
> They're just outputting tokens that resemble a reasoning process.
Looking at one such process of emulating reasoning (got deepseek-70B locally), I'm starting to wonder how does that differ from actual reasoning? We "think" about something, may make errors in that thinking, look for things that don't make sense and correct ourselves. That "think" step is still a blackbox.
I asked that llm a typical question of gas exchange between containers, it made some errors and noticed some calculations that didn't make sense:
> Moles left A: ~0.0021 mol
> Moles entered B: ~0.008 mol
> But 0.0021 +0.008=0.0101 mol, which doesn't make sense because that would imply a net increase of moles in the system.
Well, that's totally invalid calculation, it should be "-" in there. It also noticed that those quantities should be same in other place.
Eventually, after 102 minutes and 10141 tokens, involving checking answers from different angles multiple times, it outputted approximately correct response.
I made a prototype template for a site and had it exactly how I wanted in about an hour. I mocked it up in Photopea, gave Claude an image, and told it to generate HTML that would make a page with those qualities.
A few tweaks and a couple more prompts later and I had all the main elements I wanted in place, and just need to fill content which I won't use AI for.
Probably saved me several hours because I suck at frontend.
I also struggle with frontend, can you share any more about how you did this? All local? From a service? My usage so Fae is limited to copilot, but I'd like to try other things.
> The purpose is to prevent users from running unauthorized software on the computers they allegedly own.
I've maintained for several years now that the actual corporate wet dream is that they can lock down the average PC architecture/OS to the same degree they have on phones. Because unfortunately, in the phone sector, the market has already shown the majority of users don't care who really owns their devices.
My hope is that Linux gets wide enough adoption to prevent that from becoming a feasible option for them in the future.
Buy a Mac. You'll see that corporate dream come a reality. Immutable OS partition. Security prompts that can no longer be bypassed. Binary signing requirement. It just keeps getting worse and worse, for a power user.
May be "certified UNIX" (when you look at it funny), but it feels like no freedom-loving UNIX-style system I've ever used.
You can turn all that off if you wanted to. OpenCore Legacy Patcher will build you a kernel with the SIP flag mask set to 0xFF, ie, completely disabled.
Same sorts of laws are the reason why it's generally difficult to import really cheap vehicles from overseas unless they're of a certain age, Chinese or not. Classical example that gained publicity are Japanese "Kei Trucks" that can only be imported after they're 25 years old.
And yeah, they cite safety as the reason. But as a motorcycle rider, I put a tank of explosive liquid against my crotch on a vehicle that will send me flying if one tire loses traction too badly with minimal safety mechanisms against it, and in my state I technically don't even have to wear a helmet.
On the other end of the spectrum, the most popular cars in the nation are the size of a light tank, and will smash into you at center mass before throwing you underneath them if you get hit by one.
That wouldn't work anyway. In basically every jurisdiction, operators of vehicles are expected to retain control of the vehicle regardless of self driving status.
I think it’s useful for legislators to do their job and explicitly define and clarify the boundaries of the law. You can’t necessarily just rely on precedent for new things.
While not empirical proof I typically distrust anything that has massive marketing budgets. Nord seems to sponsor every Tom, Dick, and Harry on YouTube to push their product and, as we've seen from many other unmasked operations that do that (Honey, Established Titles), that doesn't bode well.
I don't use Mullvad, but I've never seen them run ads directly, and they've gotten exposure via word of mouth very effectively.
We'll absolutely document a theft of a sack of potatoes if we notice it, because it's never just the sack of potatoes. The vast majority of thieves get emboldened and either up their thefts, or continue to repeat offend at a low level.
The one sack of potatoes is likely actually a sack of potatoes every week, and it doesn't really take long to rack up a dollar amount worth prosecuting over. I can get my local detectives hungry to file a warrant at the $100 mark.
Shoplifting is a bad game where the store has every edge, and only has to win once. The shoplifter has to win every time. If they don't, they usually find the line between getting away with it and jailtime is much thinner than they thought.
I pay for Premium for that reason. I don't have any other TV/streaming services, so while I don't especially like giving Google money, I use it enough that the expense is well-justified.
reply