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

    Neuronpedia

    INDEX
    Explanations

    don't minimize failure

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     calibrated
    0.89
     efficiencies
    0.85
     conceptually
    0.80
     infused
    0.79
     precise
    0.76
     tax
    0.75
     khái
    0.75
     introduces
    0.74
     purified
    0.74
     கிட்டத்தட்ட
    0.73
    POSITIVE LOGITS
     Nếu
    0.80
    If
    0.78
     Should
    0.77
    Should
    0.76
    Jika
    0.74
    Unhandled
    0.73
     Failure
    0.72
     Jetzt
    0.72
     Jika
    0.71
     If
    0.71
    Act Density 0.928%

    No Known Activations