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

    Neuronpedia

    INDEX
    Explanations

    Code

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on comma and closing‐parenthesis tokens within code, i.e. it detects punctuation used to separate arguments in function or method calls.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    lak
    -0.07
    φορ
    -0.07
    _rd
    -0.07
    лер
    -0.07
    ằm
    -0.07
     약
    -0.06
    andel
    -0.06
    廊
    -0.06
    UNCH
    -0.06
    esters
    -0.06
    POSITIVE LOGITS
    刻
    0.06
     الحرب
    0.06
     Constant
    0.06
     JS
    0.06
    .stack
    0.06
    >Main
    0.06
     validates
    0.06
     mutable
    0.06
    .inspect
    0.06
    ]")]↵
    0.06
    Act Density 0.075%

    No Known Activations