Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    string delimiters and code punctuation

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     '
    3.91
     ‘
    3.43
     '(
    2.91
     '[
    2.90
    『
    2.88
    ='
    2.78
     '#
    2.63
     '.
    2.62
     '/
    2.61
     '_
    2.59
    POSITIVE LOGITS
     {"
    1.90
     "-"
    1.77
     "/"
    1.71
    ("");
    1.71
    ("-
    1.70
    ["
    1.69
    ("/",
    1.69
     (“
    1.68
     "/
    1.67
    ("",
    1.66
    Act Density 0.238%

    No Known Activations