INDEX
    Explanations

    scientific studies

    This neuron activates on Romanian-language text.

    New Auto-Interp
    Negative Logits
     هنوز
    -0.07
    .slide
    -0.07
    Method
    -0.06
     خر
    -0.06
     Product
    -0.06
    _Db
    -0.06
     Cant
    -0.06
    Professor
    -0.06
    ورات
    -0.06
     languages
    -0.06
    POSITIVE LOGITS
     부산
    0.06
     volcano
    0.06
    ungalow
    0.06
    /pop
    0.06
    ۱۹۷
    0.06
     durch
    0.06
    anging
    0.06
    0.06
     bang
    0.06
     defaultProps
    0.06
    Act Density 0.063%

    No Known Activations