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

    Neuronpedia

    INDEX
    Explanations

    attackers could gain power

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     pomoći
    0.42
    નિ
    0.41
     authorize
    0.41
    应该是
    0.41
    忝
    0.40
    неоп
    0.40
    mouseenter
    0.39
    autor
    0.39
     asigur
    0.38
     authorizes
    0.38
    POSITIVE LOGITS
    猖
    0.61
     threatens
    0.59
     threatening
    0.57
     threaten
    0.56
     prow
    0.53
     attacking
    0.51
     menacing
    0.51
     embold
    0.50
     menace
    0.50
     attacks
    0.49
    Act Density 0.018%

    No Known Activations