INDEX
    Explanations

    collecting evidence

    The neuron fires on language that lays out procedural “documenting/record‐keeping” steps—e.g. numbered list markers (“1.”, “2.”) and imperatives like “Document,” “Record,” “Keep,” “Gather” that instruct you to record evidence or details.

    New Auto-Interp
    Negative Logits
    SCAN
    -0.06
     Matcher
    -0.06
    945
    -0.06
    igin
    -0.06
    _socket
    -0.06
    porno
    -0.06
    ักเร
    -0.06
    .symmetric
    -0.06
    ým
    -0.06
    approve
    -0.06
    POSITIVE LOGITS
     lotion
    0.07
     {@
    0.07
     []↵↵
    0.06
     Mechanical
    0.06
    -right
    0.06
     دي
    0.06
     inlet
    0.06
    	ArrayList
    0.06
     eds
    0.06
     τ
    0.06
    Act Density 0.011%

    No Known Activations