this announcement had such strong gpt-output vibe..
to the "writers": pleeease don't present unedited slop to me. i'm a human, if you want my attention consider using your own voice. i don't want to read what gpt thought would "market" you best.
but this thread did remind me of marimo so that's sweet
i had the exact same behavior back in 2023, it seemed like clearly leakage of user conversations - but it was just a bug with api calls in the software i was using.
is that a method to save context tokens by "baking" the attended in-context-learned into the weights?
Im missing one step of so-what in the paper abstract..
