infinitepcg @ infinitepcg @lemmy.world

Posts

2

Comments

124

Joined

2 yr. ago

It's not a different discipline, an LLM is an example of a machine learning model.

Nobody knows if and when programming will be automated in a meaningful way. But once we have the tech to do it, we can automate pretty much all work. So I think this will not be a problem for programmers until it's a problem for everyone.

Thank you for writing the explanation! I still think that this doesn't need a blockchain. Instances could broadcast user creation, so each instance could validate user age on its own (or ask other trusted instances when they first "saw" that user).

Fundamentally, blockchain solves the problem that there is no central source of trust, but in the Fediverse people necesarily trust the instance that they sign up, so a blockchain can't add much in my opinion.

I see. I'm not convinced that proving the account creation date makes much of a difference here. Obviously the instance records when you sign up, so you would only need this to protect against malicious instances. But if a spammer is manipulating their instance to allow them to spam more, you have a much bigger problem than reliably knowing their account creation date.

this account holder has this name on that instance

How would that help? A spam bot could just make lots of blockchain wallets.

you get all sorts of unspoofable benefits from that

what are the benefits? I struggle to come up with any benefits.

If the animations look realistic, it’s almost certainly not predetermined

You could do a perfectly realistic simulation, record the path for each outcome and then play one of them.

Or, if the physics simulation is deterministic, you could store a set of starting positions and their outcomes.

I think it's reasonable to not short stocks. I just find it a bit weird to see people confidently proclaim that a company is overvalued, but than not shorting the stock, which would be the rational thing to do.

I'm not an expert on Monte Carlo methods, but reading the Wikipedia article on Markov Chain Monte Carlo, this doesn't fit what WFC does for the reasons I mentioned above. In MCMC, your get a better result by taking more steps, in WFC, the number of steps is given by the map size, it can't be changed.

It's hard to tell how much a platform is worth, arguably the value of Twitter was 44B, since someone was willing to pay that.

The good news is, if you're really certain that Reddit is overvalued, you'll soon be able to short it and get rich if you end up being right!

I don't think WFC can be described as an example of a Monte Carlo method.

In a Monte Carlo experiment, you use randomness to approximate a solution, for example to solve an integral where you don't have a closed form. The more you sample, the more accurate the result.

In WFC, the number of random experiments depends on your map size and is not variable.

it doesn't train or self-improve like ML does

I think the training (or fitting) process is comparable to how a support vector machine is trained. It's not iterative like SGD in deep learning, it's closer to the traditional machine learning techniques.

But I agree that this is a pretty academic discussion, it doesn't matter much in practice.

I don't think the number of bots matters much, there are much more real people on Twitter than on Mastodon. It's not an issue for Twitter because they already are the platform where everyone else is. I'm optimistic about Mastodon, it already has the better UX and the better business model and I think it will slowly attract more users over time and eventually reach the relevance that Twitter had at its peak.

I think these two fields are very closely related and have some overlap. My favorite procgen algorithm, Wavefuncion Collapse, can be described using the framework of machine learning. It has hyperparameters, it has model parameters, it has training data and it does inference. These are all common aspects of modern "AI" techniques.

When the Apple car is released, the EU will invent 350 kW DC fast charging via USB-C 🙏

It works as long as you don't call list() within that function.

The difficult thing is gaining users, not writing the code.

I've been on Mastodon for over a year and I never experienced anything that could be classified as a technical glitch. From a tech / UI perspective it feels very polished to me.

I guess the only exception would be that old posts are sometimes missing on profiles from different servers.

Haha, I completely missed that it's a game.

I wonder how often someone walks in and tells them about the mistake. Do the baristas have a standard response?

The NPR article says the opposite of the headline.

By contrast, the new study found that in a third of societies for which there is data, the women hunt large game. In other words, they do go after the kind of big mammals associated with the stereotype of male hunters.

Yes, women hunted sometimes, but in 40 of the 60 societies they looked at, women didn't participate in big game hunting at all. In the remaining third, they did find at least one female hunter, but they don't say what the ratio is.