Is it the training process that you take issue with or the usage of the resulting model?

We’ve been doing this in RL research with Minecraft as well (see MineDojo). An excerpt from the GitHub page:

MineDojo […] provides open access to an internet-scale knowledge base of 730K YouTube videos, 7K Wiki pages, 340K Reddit posts.

Again, no one has run into legal issues with this yet either, but this also isn’t as ubiquitous compared to Atari, nor has it been around for as long.

Did you mean to respond to a different comment? I have no idea what happened in the VP debate.

The very first response I gave said you just have to reframe state.

And I said “am augmented state space would make it Markovian”. Is that not what you meant by reframing the state? If not, then apologies for the misunderstanding. I do my best, but I understand that falls short sometimes.

Reinforcement learning research has been using Atari games as standard benchmarks for over a decade now and no one has faced legal issues yet.

I’m not familiar with the term “beam” in the context of LLMs, so that’s not factored into my argument in any way. LLMs generate text based on the history of tokens generated thus far, not just the last token. That is by definition non-Markovian. You can argue that an augmented state space would make it Markovian, but you can say that about any stochastic process. Once you start doing that, both become mathematically equivalent. Thinking about this a bit more, I don’t think it really makes sense to talk about a process being Markovian or not without a wider context, so I’ll let this one go.

nitpick that makes communication worse

How many readers do you think know what “Markov” means? How many would know what “stochastic” or “random” means? I’m willing to bet that the former is a strict subset of the latter.

It’s in reference to your complaint about the imprecision of “stochastic process”. I’m not disagreeing that molecular diffusion is a stochastic process. I’m saying that if you want to use “Markov process” to describe a non-Markovian stochastic process, then you no longer have the precision you’re looking for and now molecular diffusion also falls under your new definition of Markov process.

That’s basically like saying that typical smartphones are square because it’s close enough to rectangle and rectangle is too vague of a term. The point of more specific terms is to narrow down the set of possibilities. If you use “square” to mean the set of rectangles, then you lose the ability to do that and now both words are equally vague.

Everyone’s weird in their own ways. It’s just that one of them is trying to convince people that weird is bad while simultaneously trying to court their votes.

Stochastic process

Or maybe had to simultaneously work multiple full time jobs and a weekend job to make ends meet?

Why settle for good enough when you have a term that is both actually correct and more widely understood?

Why does everyone keep calling them Markov chains? They’re missing all the required properties, including the eponymous Markovian property. Wouldn’t it be more correct to call them stochastic processes?

Edit: Correction, turns out the only difference between a stochastic process and a Markov process is the Markovian property. It’s literally defined as “stochastic process but Markovian”.

I find it amusing that everyone is answering the question with the assumption that the premise of OP’s question is correct. You’re all hallucinating the same way that an LLM would.

LLMs are rarely trained on a single source of data exclusively. All the big ones you find will have been trained on a huge dataset including Reddit, research papers, books, letters, government documents, Wikipedia, GitHub, and much more.

Example datasets:

• RedPajama
• The Pile

What’s your definition of power then?

I would argue that there is a bidirectional casual relationship. Having more money gives you more power because you can directly spend that money to do things. More power means you can better influence people to give you their money.

Do you know if there’s anything for people who have no interest in merch or concerts? I’m thinking it would be pretty cool if someone set up a system where you pay a recurring amount and it gets split between artists based on what you listen to in your local library.

We’re not at a point yet where this is a concern, so still on the brainstorming phase of how to do this.

I think the main concern I have is the addictive side of the internet that’s enabled by their recommendation systems and infinite scrolling, so that’s what I would try to block. For example, allow free reign on YouTube, but you have to specifically search for what you want to see.

There’s also the question of privacy, and whether we should be keeping track of and checking their browsing histories. I’m currently leaning towards yes, while also making sure that they’re aware of what we’re doing. There’s value in letting them make their own mistakes and learning from them, but that only applies to things that they can learn and easily recover from.

Immediately afterwards? I’ve been told to wait at least half an hour after eating, or else brushing will have a negative effect on your teeth.

Focus on Dilution, Not Restriction

This is probably the key takeaway for most people. If you want to decrease your Calorie intake, then eat food that is more satiating per Calorie. A bunch of those named diets are based on this idea (e.g. Keto / low-carb, Paleo, Mediterranean)

That said, everyone’s mind and bodies are different. You’ll have to experiment and figure out what works for you. Some people do respond well to things like time-restriction based diets, or straight up Calories counting.