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

    Neuronpedia

    INDEX
    Explanations

    treatable, doable, findable, accessible

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     and
    0.79
     up
    0.78
     with
    0.77
     to
    0.73
     on
    0.65
     p
    0.63
     v
    0.63
     not
    0.61
     but
    0.60
    与
    0.59
    POSITIVE LOGITS
    thed
    0.63
    ar
    0.56
    h
    0.56
     прида
    0.52
    rés
    0.51
    arın
    0.51
    didn
    0.51
    takes
    0.50
    ającego
    0.50
    но
    0.50
    Act Density 0.003%

    No Known Activations