Elektrine lite

← Feed

@adamchainz@fosstodon.org

Post #2468515

2026-03-31 23:08 UTC

@pamelafox@fosstodon.org I think it comes from models being tuned to minimize the amount of edits or context read 🤷

Replies (1)

  • @pamelafox@fosstodon.org 2026-03-31 23:17

    @adamchainz@fosstodon.org yeah they need to redo that RLHF! @tonybaloney@fosstodon.org is doing cool research around non-optimal Python generated by LLMs, on that note.

    Open ##2468516