INDEX
    Explanations

    This neuron detects section headers and numbered issue/entitlement labels (e.g. “THE ISSUES,” “1.”, “Entitlement to service connection”) in legal‐style documents.

    New Auto-Interp
    Negative Logits
    Row
    -0.06
    [(
    -0.06
    _configuration
    -0.06
     이는
    -0.06
     vocals
    -0.06
     ΑΠ
    -0.06
    (?:
    -0.06
     Cz
    -0.06
    -associated
    -0.06
    ocrates
    -0.06
    POSITIVE LOGITS
    ίο
    0.07
    -san
    0.07
     milfs
    0.07
    .FormBorderStyle
    0.07
    .socket
    0.06
    -FIRST
    0.06
     sun
    0.06
     Edinburgh
    0.06
    ١
    0.06
     Georgia
    0.06
    Act Density 0.005%

    No Known Activations