INDEX
    Explanations

    Gemini family of models

    New Auto-Interp
    Negative Logits
    ن
    1.55
    1.41
    س
    1.33
    н
    1.31
    و
    1.30
    ،
    1.21
    ря
    1.11
    د
    1.11
    1.05
    ח
    1.05
    POSITIVE LOGITS
    g
    1.52
    b
    1.51
    d
    1.38
    ı
    1.32
    l
    1.18
    LE
    1.13
    u
    1.11
     a
    1.10
    ary
    1.10
    ER
    1.10
    Act Density 0.006%

    No Known Activations