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

    Neuronpedia

    INDEX
    Explanations

    almost exclusively or certainly

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     Permits
    0.37
    AND
    0.37
     Typically
    0.37
    andad
    0.37
     AND
    0.36
     Adds
    0.35
     Hence
    0.35
     ভীষণ
    0.35
     Preferably
    0.34
     Including
    0.34
    POSITIVE LOGITS
     almost
    0.57
     limitless
    0.55
     indestruct
    0.55
     unheard
    0.54
     impossible
    0.52
    ほぼ
    0.52
     indestructible
    0.52
     certainly
    0.50
     incompre
    0.50
     useless
    0.49
    Act Density 0.033%

    No Known Activations