@theo Patched and merged: https://t.co/zl25ueJuke
I think you just got unlucky I'm ngl. Apparently they are rebuilding with 16.4.x so a few things are unstable according to @usr_bin_roygbiv. Ig I just picked a bad time to recommend it to you.
@usr_bin_roygbiv Makes it harder for the models to screw this stuff up, saves time, and frontend is still kinda something that models will lack signal and/or test coverage over.
@thsottiaux@kr0der Uhhhhhhh
A swarm of GPT Pro xhighs. Basically the kind of TTC that Noam is warning everyone about. All on Cerebras too.
And this is somehow hooked into codex using worktrees and omp style workflowz to make the parallel TTC work for code.
Idk, just imagining things now.
i think these posttraining-automation benchmarks are even more important than they seem
when models cross the threshold of being able posttrain other models, hopefully there will be a cambrian explosion of the types of minds
authoring minds will become an accessible artform