Chinese eval startup claims first third-party model evaluations; Ó hÉigeartaigh questions Beijing AISI precedent

2 people · 3 posts · first post Tue 15:31 UTC · latest 40 min ago

Seán Ó hÉigeartaigh questions a Beijing AISI-linked Chinese eval startup's claim to be first with third-party model evaluations, noting Beijing AISI has evaluated Chinese models for years. The startup responded on issues he raised, with discussion involving Tomek Korbak and Usman Anwar.

All posts · 2, most important first

1. @s_oheigeartaigh policy · Tue 15:32 UTC
   Though Beijing AISI's been doing 3rd part evaluations of Chinese models for a few years? I'm a little confused re: their claim about being the first https://t.co/T5OVig4hS7
   0 likes · 0 reposts
2. @dkokotajlo research · Tue 16:50 UTC
   Thank you!
   0 likes · 0 reposts

Reposted by @s_oheigeartaigh.