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

    Neuronpedia

    INDEX
    Explanations

    contrasting caveats and limitations

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     remarkably
    1.06
     Incidentally
    1.00
     amazingly
    0.98
     unbelievably
    0.96
     невероят
    0.95
    unfortunately
    0.89
    可惜
    0.86
    Sadly
    0.86
    Incidentally
    0.84
    even
    0.81
    POSITIVE LOGITS
     unless
    1.09
     workaround
    1.05
     Nonetheless
    1.04
    unless
    1.03
     needs
    1.01
     Nevertheless
    1.00
     worse
    0.93
    それでも
    0.93
     Unless
    0.93
     inkább
    0.92
    Act Density 0.153%

    No Known Activations