[-] lets_get_off_lemmy@reddthat.com 24 points 1 week ago* (last edited 1 week ago)

Long, boring, hard to pay attention to. I read philosophy and theory sometimes but it's few and far between for those reasons. I really have to be in a special mood to sit down and read something that dense.

Edit: I'm not the original commenter

Depression for me.

No. I wish I could stay awake forever sometimes.

[-] lets_get_off_lemmy@reddthat.com 34 points 2 months ago

That we're able to dodge the incoming wave of fascism and peacefully skim board into progressive politics.

[-] lets_get_off_lemmy@reddthat.com 15 points 2 months ago

This is such a drunk, stupid tech bro idea.

[-] lets_get_off_lemmy@reddthat.com 1 points 2 months ago

Depends who you are. If you're a person of interest to the Russians for any reason, I wouldn't trust it.

[-] lets_get_off_lemmy@reddthat.com 5 points 2 months ago

Same. I've found I'm most productive from like 3-7 pm, which sucks. I'd like to be productive in the morning or in the early afternoon instead of mostly past regular work hours.

[-] lets_get_off_lemmy@reddthat.com 16 points 2 months ago* (last edited 2 months ago)

I'm an AI researcher and yes, that's basically right. There is no special "lighting mechanism" portion of the network designed before training. Just, after seeing enough images with correct lighting (either for text to image transformer models or GANs), it will understand what correct lighting should look like. It's all about the distribution of the training data. A simple example is this-person-does-not-exist.com. All of the training images are high resolution, close-up, well-lit headshots. If all the training data instead had unrealistic lighting, you would get unrealistic lighting out. If it's something like 50/50, you'll get every part of that spectrum between good lighting and bad lighting at the output.

That's not to say that the overall training scheme of especially something like GPT-4 doesn't include secondary training operations for more complex tasks. But lighting of images is a simple thing to get correct with enough training images.

As an aside, I said that website above is a simple example, but I remember less than 6 years ago when that came out and it was revolutionary, so it's crazy how fast the space has moved forward in such a short time.

Edit: to answer the multiple subjects question: it probably has seen fewer images with multiple subjects and doesn't have enough "knowledge" from it's training data to accurately apply lighting in those scenarios. And you can imagine lighting is more complex in a scene with more subjects so it's more difficult for the model to use a general solution it's seen many times to fit the more complex problem.

[-] lets_get_off_lemmy@reddthat.com 32 points 2 months ago

Hahaha, as someone that works in AI research, good luck to them. The first is a very hard problem that won't just be prompt engineering with your OpenAI account (why not just use 3D blueprints for weapons that already exist?) and the second is certifiably stupid. There are plenty of ways to make bombs already that don't involve training a model that's an expert in chemistry. A bunch of amateur 8chan half-brains probably couldn't follow a Medium article, let alone do ground breaking research.

But like you said, if they want to test the viability of those bombs, I say go for it! Make it in the garage!

[-] lets_get_off_lemmy@reddthat.com 1 points 3 months ago

I don't think it's lane surfing if you're not changing lanes. Anyway, this comment section has made me realize that it always just depends. Drive aware, keep safe distance, don't unnecessarily change lanes, let people pass (on the left) if they're going faster than you, etc.

The best advice I ever got about driving was "be predictable." I think if anyone really takes that to heart empathetically then it would be safer.

[-] lets_get_off_lemmy@reddthat.com 1 points 3 months ago

I find this scenario extremely rare as in most cases I'm envisioning, there is a middle turn lane separating the two opposite lanes. Either from a light or just as a buffer between the flows of traffic.

This is all to say, there aren't any hard and fast rules and there are too many scenarios to cover with a blanket statement like "The left lane is for passing. If you're not passing somebody, move over to the right lane. It's not that hard people" (comment above).

victorz said it more succinctly below.

[-] lets_get_off_lemmy@reddthat.com 3 points 3 months ago

See this is a sensible response to people getting unreasonably PO'd about this. You drive in "the left" (whatever that means to your relative position) until someone faster comes along and they can't move more left than you.

I get upset when some fuckwit is going 15+ over the flow of traffic and then that fuckwit gets pissed when he runs up on someone's ass expecting them to be aware of every dangerous fuckwit out there.

view more: next ›

lets_get_off_lemmy

joined 11 months ago