The spam scale problem should be solvable with a one-shot locally evaluated NN trained to classify behavior/tweets from their massive massive ground truth dataset. If it’s evaluated locally then I don’t see how scale is an issue.
I've come to think of this as just a culture problem. We think of share-everything and recent social changes as innovation and "the future" but not everything that is modified at t+1 vs at time t is necessarily a pareto improvement. I've come to believe something similar to you, that we need to come to our senses (each individually) and converge to more reasonable social norms.
Making claims without scientific backing probably shouldn't be a bannable offence. The spirit of science is about discussion and counter arguments and being open to being wrong or even the idea that the consensus might be invalid, not about coercion.
I think we're not discussing the real point here which isn't whether "full completely unregulated free speech" should be the norm on twitter or not. Elon isn't proposing allowing CP or whatever the worst stuff you see on 8chan is. The battleground is simply in a much subtler grey area of idea space. Elon basically seems to think there's been push to disallow discussion about quite reasonable subjects (which indeed would have been allowed on HN too, e.g. covid vaccine pros and cons) and he'd like it to be allowed again.
Note also that twitter is quite different from traditional forums in the way it's structured. You follow people, you can block people. The analogy to traditional forums needing moderation isn't really one-to-one.
I think this is really cute and wonderful. I think there's some room to improve here, e.g. an AR ghost overlay of previous frames as well as ML extracted frames from pure paper without having to print a template (both of which I believe are feasible). But overall I love that people are experimenting with this stuff (and wouldn't want to react negatively here in any way to discourage this type of work).
Some might argue moving from digital to pen and paper is a step down but I'd argue the opposite. I'd love and look forward to a possible future where we move digital advancements into the analog world in this way. As Theil says you can barely tell the difference between this world and the 80s if you remove cell phones and computers, but things like this are a cool way to hopefully get us closer to computers aiding more natural forms of interaction.
Autonomous cars and fusion are very different creatures than the growth of the internet or computers in general. The latter was widely distributed and had low hanging fruit that just about everyone in society could pluck and have fun with, the former is something a handful of companies and a few hundred researchers will get to develop while the rest of us will have to have our creativity satisfied with the fulfilling act of paying for it. (Even ML to an extent, since it's so data and compute expensive - generally far from a low hanging fruit -, though less so.)
Are you sure? I'd like to hear the take of someone in their early twenties on this. My feeling is that a lot of them are terrified by it themselves too.
Would you say The Recognitions is readible? I've wanted to read it for a while but it felt intimidating and it was lumped into the 'post modern' category in book stores etc which has made me afraid it's one of those samuel beckett-esque books where they just repeat the same word or phrase for pages and pages (maybe meaningful through some lens but also somewhat nonsense and not really worth my limited time). I read a few pages of The Recognitions and it seemed pretty novel-y so I feel like it might be not too bad. What's your take (reviews I've read doesn't give me a satisfying answer)?
Who even decides what's best for the company? One could argue him openly criticizing twitter is in its best interest. Like pouring alcohol in a wound, it hurts and kills off some bad parts of the system but it'll be stronger at the end of it than had you not done it in the first place.
They'll just be A/B tested from out of a collection of alternatives. (After all isn't that what the artistic filter is supposed act as a proxy for in the first place.)
You could just use ffhq/faces, it's fine. You'd think it gets stuck in a local minima but it really doesn't. Transfer learning from any 'finished' checkpoint is much much much faster than doing it from scratch.
I'd recommend sticking with stylegan2 for the moment though since it's much easier (cheaper time/computation wise) to transfer from (and the result is about the same - in fact better at the same time and computation budget).
> "The computation for the 512x512px training set was run for 20 days on an IBM AC922 POWER9 server running RHEL7, utilising all four Nvidia Tesla V100 SXM2 16GB GPUs."
That's super overkill. If you just transfer learn from a good checkpoint you'll get good results in a day or two on a V100 (probably even a P100), especially at just 512x512. (I've done this lots of times and a colab pro account for $10 is more than plenty.)
The correct response here from the artists point of view should be a widespread coming together against their art being used as training data for ML models. With a quickly spread new license on most major art submission sites that explicitly forbids AI algorithms from using their work, artists would effectively starve OpenAI and others from using their own works to put them out of a job.
They're also not censored on the dataset front and thus produce much more interesting outputs.
OpenAI has a low resolution checkpoint for similar functionality as this - called GLIDE - and the output is super boring compared to community driven efforts, in large part because of similar dataset restrictions as this likely has been subjected to.
Yeah if I were in a similar position in a future scenario and I found this story and it ended with public shaming and prison etc, then I'd just burn the thing instead of trying to return it.
Well you got your wish. Gonna be honest and say I’m not sure I’m enjoying HNs move to follow twitter and facebook to micromanage discussion to this degree. I really don’t think we were being uncivil in having a discussion about what we appreciate about this guy.
I don’t think of him as shining. I like that he’s a real person, flaws and all. Not someone to idolize. Many great hackers that I know and in general are total jerks (I suspect the trait helps wrangling a machine somehow but idk). I wouldn’t find it accurate to my experiences if we had some polished perfect optics guy as representative (not that that’s what he is)
I have gone to college with similar overachievers/braggers, and now that you mention it if they were in the spotlight as much as him I'd roll my eyes pretty hard. Still though, having not gone to college with this particular one somehow I can stand it.