On: OK, I can partly explain the LLM chess weirdness now

[synarchefriend → OutOfHere]

The model was trained on games in PGN notation. It would be shocking if it found ASCII art easier to understand than what it was actually trained on.

[OutOfHere → synarchefriend]

Well, clearly you're not interested in experimentation, only in assumptions.

[daveguy → OutOfHere]

How does stating the outcome you expect imply you are not interested in experimentation? Hypothesis formation is the very first step in experimentation.

[danielmarkbruce → OutOfHere]

Most people who understand LLMs and how they are trained would be shocked. In practice, that's an objectively true statement.

[BeetleB → OutOfHere]

Please, please show us your experiments.

[OutOfHere → BeetleB]

I am not the one writing and posting useless articles, even harmful articles, also distorting the understanding of LLMs. Ask the ones who do to perform better experiments.

[multjoy → OutOfHere]

You know that the LLM isn't actually your friend, don't you?

[BeetleB → OutOfHere]

So to quote yourself:

> Well, clearly you're not interested in experimentation