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

    Neuronpedia

    INDEX
    Explanations

    refusing to harm or kill

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     ECONOMIC
    0.48
     Economic
    0.44
     korzyst
    0.43
     buffers
    0.41
     vhod
    0.40
     buffs
    0.39
     buffer
    0.38
    Economic
    0.38
    parsedBlock
    0.38
     consommateurs
    0.38
    POSITIVE LOGITS
     ending
    0.74
    Ending
    0.70
     ended
    0.67
     saving
    0.64
    结束
    0.64
     extinguish
    0.64
     endanger
    0.64
     Ending
    0.63
    ending
    0.63
     extinguishing
    0.62
    Act Density 0.042%

    No Known Activations