It explains the fundamentals of a transformer network (which all current LLMs are based on) on a super tiny, down to the basics example network, allowing you to understand what is happening within the network step by step, rather than being confronted with theoretical concepts or tonnes of linear algebra.
It's really nice and as hands on, as these things get.
Get ready for your daily dose of inspiration, fun facts, short stories, and top 10 lists. In today’s video, we’ll be exploring [brief video summary]. Don’t forget to like, comment, and subscribe for more content that sparks curiosity and fuels your mind!
I once had a problem with an ASUS notebook. I think it was the touchpad. So I looked in dmesg and found something like:
“HID something something was configured with flag 1. If this is incorrect, try the command blah blah flag=0.”
Ran the command and it was fixed.
I've never seen such a beautiful error in Windows. And I really lost my respect when I tried to calibrate an external screen on a Mac because that felt like Linux from 2016.
You've got the paint mircophone thing, which requires someone to physically be at (your) location and install hardware, but not encrypted DNS, or not - you know - just posting your current whereabouts to social media?
Someone being physically near you and installing specialized hardware is quite a threat level.
Fucking: "Cars will collect telemetry locally to be shared when you visit auto repair shops." DUDE. They have mobile data in these things. I can't even. Your list is so thoughtless.
And what's second hand clothing gonna do for you, if the people surveying you have eyes and object permanence?
It sounds kinda stupid, but before I got used to my current pans, I always had to add a small drop of water to check whether I could start. As soon as the water began to boil, the pan was hot enough to cook (but wouldn't just burn everything into a crisp).
In theory, it's really, really simple, but to be able to intuit the pans temperature, really felt like a boss move… for the first few times, until it just became second nature. I had to scrape off more than one egg before it clicked.
Its manufacturing requires relatively large quantities of fairly toxic and also very resistant chemicals. This means that when they enter the environment (and they will), they will poison it for a very long time, with consequences that are not yet fully understood.
In addition, it decomposes when heated above 260°C and can detach from the pan and get into your food. Keep in mind that almost all oils start to smoke well before reaching this temperature, so you should be able to tell when you've just ruined your pan.
In my opinion: hot stainless steel and a little butter will prevent food from sticking. This requires a (easy to learn, but still) technique, but once you got it, you can enjoy your stainless steel pan for basically forever.
Lenovo just let me add a 65W USB-C charger at no extra cost. It's GaN too. I know it's probably not actually "gratis", but at least they don't have to ship e-waste with every order.
I learned a while ago that constant thoughts of death and despair before sleep may indicate an anxiety disorder. However, since online diagnoses from strangers are almost always bullshit, I'd like to recommend a weighted blanket. I know it sounds stupid, but the weighted blanket fucks are right. You probably want one.
A bunch of manufacturers implement adaptive refresh rates. So your battery shouldn't drain noticeably faster unless you are scrolling, but that's where you'd want extra frames the most.
More frames, more smooth and it at least feels like I'm a bit quicker with navigation. It's just nice.
I highly recommend this:
https://rti.github.io/gptvis
It explains the fundamentals of a transformer network (which all current LLMs are based on) on a super tiny, down to the basics example network, allowing you to understand what is happening within the network step by step, rather than being confronted with theoretical concepts or tonnes of linear algebra.
It's really nice and as hands on, as these things get.