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

    Neuronpedia

    INDEX
    Explanations

    ensures function correctness

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     sometimes
    0.87
    もしくは
    0.85
    或
    0.83
     cuando
    0.83
     أو
    0.83
     when
    0.83
     hoặc
    0.79
    あるいは
    0.79
     или
    0.78
     либо
    0.77
    POSITIVE LOGITS
     garantiza
    1.10
     забезпе
    1.06
     ensures
    1.06
     robuste
    1.05
     incroy
    1.05
     prachtige
    1.03
     Ensures
    1.03
    autiful
    1.00
     excellent
    0.99
     обеспечивает
    0.98
    Act Density 0.628%

    No Known Activations