RDLTR

Short for Read Later. Save links from any browser or phone. Click to save, read to clear. The full tour

One topic. Every takeSeek and you shall find

Reading up on Alignment Research Center

1 deep · digging since feb 25

  • www.anthropic.com favicon
    The persona selection model

    AI assistants behave human-like because pretraining teaches them to simulate personas from text, and post-training refines this Assistant persona without changing its fundamental nature.