INDEX
    Explanations

    relative clauses

    The neuron fires on the little linking/action words (“are,” “have,” “that,” etc.) used to introduce descriptive or instructive statements—essentially spotting the common verbs and connectors that kick off points in advice or list-like text.

    New Auto-Interp
    Negative Logits
    DP
    -0.07
    ξι
    -0.07
    DTD
    -0.06
    Reflection
    -0.06
    DN
    -0.06
    11
    -0.06
    Susp
    -0.06
     control
    -0.06
    _PB
    -0.06
     Ci
    -0.06
    POSITIVE LOGITS
     الذي
    0.07
    0.06
     når
    0.06
     इस
    0.06
     +/-
    0.06
    ував
    0.06
     詳細
    0.06
    마다
    0.06
    etc
    0.06
     através
    0.06
    Act Density 0.083%

    No Known Activations