INDEX
    Explanations

    Gemma, an open-weights model

    New Auto-Interp
    Negative Logits
    0.76
    aditional
    0.75
    Room
    0.68
     домаш
    0.68
    atot
    0.68
    iedad
    0.66
     スマ
    0.65
    ai
    0.64
    कक्ष
    0.63
     Room
    0.63
    POSITIVE LOGITS
     epsilon
    0.79
     stainless
    0.77
     exposed
    0.77
     raw
    0.75
     glassy
    0.75
     Exposed
    0.74
     abiertas
    0.73
     ওপ
    0.73
    exposed
    0.73
     abiertos
    0.72
    Act Density 0.027%

    No Known Activations