Best I've found is from: "Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Models": aclanthology.org/2021.emnlp-m...

"Minority and female group names are singly tokenized less than white and male names. Single tokenization correlates with frequency"

Loved the manga reader and yellow case. Japan looked amazing. Inspires me to get back to improving my Japanese. Though my problem is my kanji is so bad I can only read the most boring kids stories.

I should be doing things but circles. Lots of fun, thanks.
Another good source for losing time: "Morphing: A Guide to Mathematical Transformations for Architects and Designers" (discourse.mcneel.com/t/implementi...)

There's a good paper on why LLMs error on these problems and by design always will. Anything with a large permutation space is too expensive to represent and gets pruned. arxiv.org/pdf/2401.11817
Pretty far from super-intelligence...

The research does not mention the LLM tool used & risks or error rates are not examined. (asabiopreport.substack.com/p/a-multidis...)

The reference to "our models" imply it's "Informed Solutions" own LLM. Which I assume must need to be certified?

Is there any transparency of the LLM?

Happy to help provide some words if a random fan would be helpful. I'm a massive fan of Brindlewood and any carved from Brindlewood game.

Wildsea. It’s not set in a dystopian future, or on the cusp. Human empires failed & nature came back and built a better world. My definition: play starts in a world where the big problems of our time have been fixed. It’s not about building better futures through play but playing in them now.

.

I would vote for one with CHEF one it. It’s very tempting, I can see that first game session and the joy of that hat. Since it’s silly and I have no idea how expensive those hats would be I would pluck £10-15 out of the air. I guess similar to custom dice or gm screens.

Odd, I did not have an issue with more than one no vote selected. Just picked the ones I knew and left most "No Vote"