web dev + hot dad. enjoy charts, unions, conputer games, philosophy. chicago crespo.business
conputer dipshit
Loading...
both very relevant to my life right now and truly the perfect content for bluesky, a wretched hive of people who remember their high school test scores if there ever was one
I read these critiques whenever I see them in order to put pressure on my natural belief that, whether or not we can measure it well, intelligence is real and it's spectacular
a related post which I enjoy
people were quick to say Fable isn't better at writing, based mostly on its speaking style in chat. but I am finding it very good at editing written prose. "review the prose added for awkward or unclear language. look at the blame on some files for lines I wrote to use as a model". 100% hit rate
and another one
maybe I'm biased because this is standard bigger model expectations. this is probably the first model this large ever. huge models are really subtle at writing. RIP gpt-4.5
