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

    Neuronpedia

    INDEX
    Explanations

    realistic expectations

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    ש
    2.22
    ти
    2.14
    er
    2.11
    я
    2.03
    意思是
    1.96
    ت
    1.81
    ce
    1.72
    ため
    1.71
     οποίος
    1.69
    jeva
    1.68
    POSITIVE LOGITS
    愪
    1.85
     understatement
    1.83
     ferv
    1.80
     overestimate
    1.78
     joc
    1.74
     kneel
    1.74
     jeopard
    1.71
     chime
    1.71
    más
    1.70
     coercive
    1.70
    Act Density 0.210%

    No Known Activations