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

    Neuronpedia

    INDEX
    Explanations

    colons and equals signs

    np_max-act · gemini-2.0-flash

    This neuron detects the colon character used as a key–value or field separator in structured/configuration text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     prose
    -0.07
    lob
    -0.06
    impact
    -0.06
    .exit
    -0.06
    .pyplot
    -0.06
    .Rows
    -0.06
     thuisontvangst
    -0.06
     zboží
    -0.06
     dy
    -0.06
    "F
    -0.06
    POSITIVE LOGITS
    (coeffs
    0.07
     Maar
    0.07
     apologise
    0.07
     appealed
    0.06
     كار
    0.06
    allowed
    0.06
     sincerely
    0.06
    된
    0.06
     sia
    0.06
    失败
    0.06
    Act Density 0.026%

    No Known Activations