logoalt Hacker News

kouteiheika • today at 4:46 PM • 2 replies • view on HN

Fun, but considering this puts `claude-sonnet-5` at the first place isn't it a little... iffy when it comes to measuring intelligence?


Replies

hp6 • today at 4:57 PM

I was also surprised by this, from a proper benchmarks perspective your right, this is iffy. To make it a legit I would have to increase the number of games significantly as well as understand how much of it is random and how much real signal, not to say make the game more complex.

But all of it would kill the fun)

charcircuit • today at 6:41 PM

I watched a battle and the agent walked up to each other one and just killed them while the others didn't attack even once. There is something wrong with how the agents are setup.