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

    Neuronpedia

    INDEX
    Explanations

    code keywords and explanations

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    ar
    0.95
    are
    0.93
    (=
    0.92
    (),"
    0.91
    and
    0.90
    “,
    0.89
    ([],
    0.86
    the
    0.85
    in
    0.85
    inside
    0.84
    POSITIVE LOGITS
     Cómo
    0.98
     น่า
    0.93
     Elite
    0.93
     Características
    0.91
     Reasons
    0.88
     faisant
    0.88
     impasse
    0.88
     Legends
    0.87
     Tales
    0.87
     mise
    0.86
    Act Density 0.118%

    No Known Activations