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

    Neuronpedia

    INDEX
    Explanations

    thin or thick descriptions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     thin
    1.03
    Thin
    1.02
     Thin
    0.98
    thin
    0.93
    薄
    0.90
     thick
    0.86
     thinner
    0.85
     тон
    0.80
     Thick
    0.77
    thick
    0.72
    POSITIVE LOGITS
     layer
    0.91
    layer
    0.75
    walled
    0.75
     слоя
    0.73
     layers
    0.72
     слой
    0.70
     walled
    0.65
     Layer
    0.63
    Layer
    0.63
     camada
    0.61
    Act Density 0.025%

    No Known Activations