AI generated documentation?

[magras]

Is there a reference that backs up this claim?

openskill.py/openskill/models/weng_lin/plackett_luce.py

Lines 254 to 260 in 11ac382

	The PlackettLuce model departs from single scalar representations of
	player performance present in simpler models. There is a vector of
	abilities for each player that captures their performance across multiple
	dimensions. The outcome of a match between multiple players depends on
	their abilities in each dimension. By introducing this multidimensional
	aspect, the Plackett-Luce model provides a richer framework for ranking
	players based on their abilities in various dimensions.

[vivekjoshy]

Are you asking for a reference for whether this is what the Plackett-Luce model is? This is a well known and well studied model in statistics. It's also known a exploded logit model. This specific documention might be be referencing when I was working on item covariates and forgot to update it maybe not sure. I'm not sure if it was AI generated or I copied it from wikipedia article tbh. Thank you for pointing this out. Feel free to fix with a PR and make it better 😁!

Here is a source:

https://hturner.github.io/PlackettLuce/articles/PLADMM.html

[magras]

I was confused by multidimensionality described in the documentation, because I do not see it in the source code and several descriptions of Plackett-Luce I found online. It looked a lot like LLM hallucination to me. I'm sorry, I was wrong.

The link is interesting. I'm not going to pretend that I understand everything described in pladmm function, but it looks like there is indeed a possibility to implement multidimensional ranking scoring.

Guess I owe you a PR now. =)