INDEX
    Explanations

    legal documents

    The neuron fires on second-person, reader-addressing language (e.g. “your,” “live,” “to [verb]”), essentially picking up on direct calls to action.

    New Auto-Interp
    Negative Logits
     cartoon
    -0.07
     دو
    -0.07
     graft
    -0.06
    italize
    -0.06
     Haiti
    -0.06
    Type
    -0.06
    IDENT
    -0.06
     aDecoder
    -0.06
    anı
    -0.06
    ilation
    -0.06
    POSITIVE LOGITS
    _cp
    0.06
    0.06
     recognized
    0.06
    !↵↵
    0.06
     나오
    0.06
    [item
    0.06
    ----------------------------------------------------------------------------
    0.06
     сегодня
    0.06
     chall
    0.06
    	sn
    0.06
    Act Density 0.454%

    No Known Activations