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

    Neuronpedia

    INDEX
    Explanations

    really, exactly, merely

    np_max-act-logits · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    ationale
    0.55
     inequities
    0.52
    任意
    0.52
     utils
    0.50
    ולם
    0.49
    bewerken
    0.49
     achieves
    0.48
     infinit
    0.48
     necessitates
    0.48
     establishes
    0.47
    POSITIVE LOGITS
     AGAIN
    0.77
    really
    0.75
     really
    0.74
     Often
    0.73
    often
    0.71
    Often
    0.70
    Really
    0.70
     Really
    0.69
     often
    0.69
     często
    0.68
    Act Density 0.053%

    No Known Activations