Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    News articles/press releases

    np_max-act · gemini-2.0-flash

    This neuron detects parenthesized metadata in news‐style text—especially location datelines, source or contact tags (e.g. “(City, ST)”, press service labels, email/domain fragments) enclosed in parentheses.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    agas
    -0.08
     downloaded
    -0.07
    ячи
    -0.06
     avoid
    -0.06
     molt
    -0.06
     бл
    -0.06
     Pants
    -0.06
     brill
    -0.06
     Aberdeen
    -0.06
     Tonight
    -0.06
    POSITIVE LOGITS
    ,a
    0.07
    TK
    0.07
    _vc
    0.07
     Gä
    0.06
    ]);
↵
    0.06
    0.06
     Persona
    0.06
     Spec
    0.06
    层
    0.06
    &ZeroWidthSpace
    0.06
    Act Density 0.027%

    No Known Activations