‹ BackHN Continuity

Thread

Clef: Open-weight decision models, and new RL fine-tuning platform

637 points · 217 comments · jasondavies

  1. zwaps · · focus · HN ↗
    No mention of calibration. Is it just another llm finetune?
    1. kflansburg · · focus · HN ↗
      > Our post-training utilizes label-smoothed cross-entropy for valid schema outputs paired with a Brier loss to refine probability calibration.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.