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

    Neuronpedia

    INDEX
    Explanations

    let's address

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     debugging
    0.59
     querying
    0.58
     effecting
    0.57
     impressing
    0.56
     manifesting
    0.56
     wasting
    0.55
     messing
    0.55
     gazing
    0.55
     fooling
    0.55
     donning
    0.54
    POSITIVE LOGITS
     be
    0.92
     make
    0.92
     create
    0.92
     take
    0.88
     provide
    0.86
     keep
    0.85
     communicate
    0.83
     eliminate
    0.82
     generate
    0.80
     respond
    0.79
    Act Density 1.588%

    No Known Activations