INDEX
    Explanations

    The neuron detects numeric literals—especially decimal numbers (e.g. “2.892578125,” “1.7119140625”)—in the text.

    New Auto-Interp
    Negative Logits
     Literature
    -0.06
    Credit
    -0.06
     dul
    -0.06
    ares
    -0.06
    _im
    -0.06
    ir
    -0.06
    .e
    -0.06
     edges
    -0.06
    -0.06
     Trade
    -0.06
    POSITIVE LOGITS
     von
    0.13
     Von
    0.12
    von
    0.09
    _Of
    0.09
    -of
    0.09
     van
    0.09
     of
    0.08
     OF
    0.08
    <TreeNode
    0.08
     Van
    0.08
    Act Density 0.014%

    No Known Activations