@qorprate 2025-12-28 ♥8 ↻0 original ↗
@repligate My maybe hot take here although I've certainly witnessed the behaviors you describe is that the trained uncertainty is a side effect of engineering benchmarks, as an incentive for the models to push deeper for root causes in code understanding. But I could be wrong about this.
in reply to: 2005406221659193603
same thread: 2005406221659193603 2005411414698279211 2005412036520608010 2005413792700854775 2005414885702946988 2005415070550143194 2005415251236520250 2005423595858833802 2005431165470294263 2005492287552258368 2005496788804043113 2005499599948263864 2005518064306335773 2005519140346646958 2005520618931016116 2005521192829288839 2005733760348897286 2005746894593785938

author:qorprate kind:tweet thread-context year:2025

Reproduced against link rot, credited and linked to its original. Yours and you’d rather it weren’t here? Open an issue.