Live data from Hacker News

List animals until failure

rose.systems

1–10 of 198 posts

Re: List animals until failure

#2
It might be an interesting LLM benchmark: how many can they list without breaking the rules (repetition or non-animals). Although I bet that big bucks would be then thrown at pointlessly optimizing for that benchmark, so...

Re: List animals until failure

#5

For anyone wondering: This is based on basic text parsing and a key-value lookup table, no AI involved whatsoever. Here's the table: https://rose.systems/animalist/lower_title_to_id.js

There must be another table. I got "Are you Australian?" for "dingo" and for "cicada" "don't you love their songs?"

edit: https://rose.systems/animalist/eggs.js

Re: List animals until failure

#6
I got 42. I was very impressed by how it handled more and less specific categories. It also understood rotifers were a microscopic animal, which I half expected not to work. Great project.

Re: List animals until failure

#7
post #2

It might be an interesting LLM benchmark: how many can they list without breaking the rules (repetition or non-animals). Although I bet that big bucks would be then thrown at pointlessly optimizing for that benchmark, so...

Might be an interesting problem for understanding how various models perform recollection of prior tokens within the context window. I'm sure they could list animals until their window is full but what I'm not sure of is how much of the window they could fill without repeating.

Re: List animals until failure

#9
post #5

For anyone wondering: This is based on basic text parsing and a key-value lookup table, no AI involved whatsoever. Here's the table: https://rose.systems/animalist/lower_title_to_id.js

There must be another table. I got "Are you Australian?" for "dingo" and for "cicada" "don't you love their songs?" edit: https://rose.systems/animalist/eggs.js

Yea, there's a bunch of easter eggs. And then there's the table for the taxonomy tree.
Post reply on HN