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

    Neuronpedia

    INDEX
    Explanations

    acknowledging exploration but setting boundaries

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     velocities
    0.43
     widths
    0.43
     thicknesses
    0.43
     downside
    0.41
     drawback
    0.41
     drawbacks
    0.41
     adjustment
    0.40
     computes
    0.40
     pairs
    0.39
     conforms
    0.39
    POSITIVE LOGITS
     attempting
    0.77
     subjecting
    0.75
     Attempt
    0.70
     insisting
    0.69
     treating
    0.68
     Treating
    0.67
     denying
    0.66
     assigning
    0.66
     resorting
    0.66
    treating
    0.66
    Act Density 0.075%

    No Known Activations