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

    Neuronpedia

    INDEX
    Explanations

    breaking things down or apart

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    I
    1.08
    D
    1.08
    T
    1.05
    <0x80>
    1.03
    H
    1.03
    J
    0.99
    G
    0.97
    U
    0.97
    K
    0.95
    A
    0.93
    POSITIVE LOGITS
     breaking
    1.02
     Breaking
    0.96
     breaker
    0.93
     الكسور
    0.89
     breakers
    0.88
    ق
    0.88
     breakpoint
    0.86
     Breaks
    0.85
     Breakdown
    0.84
     régi
    0.83
    Act Density 0.226%

    No Known Activations