thank you for waiting patiently, the official benkraft voter guide is here! come for yes I do have an opinion on the governor's race, stay for footnote 8 with 200 words of my best opinions on CEO pay data
oh yeah and while I'm picking fights with Big Voter Guide, let it be known that said footnote 8 is more accurate than GrowSF's research on the same point
thank you for waiting patiently, the official benkraft voter guide is here! come for yes I do have an opinion on the governor's race, stay for footnote 8 with 200 words of my best opinions on CEO pay data
thank you for waiting patiently, the official benkraft voter guide is here! come for yes I do have an opinion on the governor's race, stay for footnote 8 with 200 words of my best opinions on CEO pay data
@_dylanga FWIW we had 33 PRs affected of ~150 merged during the incident (until we noticed and stopped the queue). It seems like the majority of the impact was on a few dozen orgs our size and up since small repos don't use merge queues.
@MatthewDSun@anjali_shriva oh this is interesting! to some extent I'm just snarking about providers doing the boring details slightly differently -- I mostly do this for internal (devtools) consumption where we generally pay passthrough token rates -- but I do like the idea of pricing according to value
@xuanalogue This also seems like the kind of result that I'd like to see more to measure whether the model "knows" it's being evaluated. LLMs love to make sure the Checkov's Gun gets fired!