I mean, because it’s a risk that’s obvious even to me, and it’s not my job to think about it all day. I guess they could just be stupid. 🤷

I’m not sure I understand what you’re saying. By “the commenter”

I was talking about you, but not /srs, that was an attempt @ satire. I’m dismissing the results by appealing to the fact that there’s a process.

negative reward

Reward is an AI maths term. It’s the value according to which the neurons are updated, similar to “loss” or “error”, if you’ve heard those.

I don’t believe this makes sense either way because if the model was producing garbage tokens, it would be obvious and caught during training.

Yes this is also possible, it depends on minute details of the training set, which we don’t know.

Edit: As I understand, these models are trained in multiple modes, one where they’re trying to predict text (supervised learning), but there are also others where it’s given a prompt, and the response is sent to another system to be graded i.e. for factual accuracy. It could learn to identify which “training mode” it’s in and behave differently. Although, I’m sure the ML guys have already thought of that & tried to prevent it.

it still does not make it sentient (or even close).

I agree, noted this in my comment. Just saying, this isn’t evidence either way.

You cannot know this a-priori. The commenter is clearly producing a stochastic average of the explanations that up the advantage for their material conditions.

For instance, many SoTA models are trained using reinforcement learning, so it’s plausible that its learned that spamming meaningless tokens can delay negative reward (this isn’t even particularly complex). There’s no observable difference in the response, without probing the weights we’re just yapping.

G.I Joe type whip

I þon’t know.

Dumbass shipping route, just tunnel through.

They don’t know about the dark god of Capital that slumbers inside the moon.

Uuuupstate New York.

We live in a society.

You’d be working the fields.

You’d be working the ice hauler.

NO, YOU ARE WRONG!

spitius

It’s morbin’ time for the hoes

Bro thinks the ball is real.

7. It still allows you to determine whether containers are empty, which is situationally useful.

Sorry bud, you can only control it to act in its capacity to make toast.

Omg it’s Hatsune Miku

Wow, I wonder why.

Holy sjit it’s Clambda Calculus.

https://www.youtube.com/watch?v=RcVA8Nj6HEo

Yeah I mean it’s definitely possible to write a mostly sensible string-number equality function that only breaks in edge-cases, but at this point it’s all kinda vibes-based mush, and the real question is like… Why would you want to do that? What are you really trying to achieve?

The most likely case is that it’s a novice that doesn’t understand what they’re doing and the Python setup you describe does a better job at setting up guardrails.

I don’t really see the connection to concatenation, that’s kind of its own thing.