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

    Neuronpedia

    INDEX
    Explanations

    subjects performing actions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    上述
    0.55
    Effective
    0.55
    effective
    0.50
    Percent
    0.50
    Overlap
    0.48
     pernyataan
    0.47
     suelen
    0.46
    必ず
    0.46
    際に
    0.46
    Typically
    0.46
    POSITIVE LOGITS
     began
    1.08
     went
    1.02
     became
    0.99
     knew
    0.97
     laughed
    0.93
     took
    0.92
     awoke
    0.85
     panicked
    0.85
     lasted
    0.85
     hurriedly
    0.85
    Act Density 0.106%

    No Known Activations