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

    Neuronpedia

    INDEX
    Explanations

    indeed, even, fact

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     แต่
    0.64
     but
    0.62
     pero
    0.56
     nonetheless
    0.55
     nhưng
    0.55
     (
    0.51
     ngunit
    0.51
     carefully
    0.50
     ولكن
    0.50
     long
    0.50
    POSITIVE LOGITS
     sogar
    1.65
    甚至
    1.59
     addirittura
    1.52
     навіть
    1.44
     даже
    1.43
     bahkan
    1.41
     thậm
    1.41
     Bahkan
    1.38
    むしろ
    1.37
     nawet
    1.36
    Act Density 0.001%

    No Known Activations