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

    Neuronpedia

    INDEX
    Explanations

    explicitly prohibit safety guidelines

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     در
    0.32
     ول
    0.29
     precedents
    0.28
    过来
    0.28
    ใน
    0.27
    Sight
    0.27
     از
    0.27
     From
    0.27
    在了
    0.27
     مص
    0.26
    POSITIVE LOGITS
     said
    0.45
     mengatakan
    0.43
     afirmó
    0.43
     dijo
    0.41
     sagte
    0.41
    said
    0.39
     lamented
    0.38
     сказал
    0.38
     contended
    0.37
     sostuvo
    0.37
    Act Density 0.016%

    No Known Activations