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 activates on tokens representing object‐member accesses (identifiers following a dot, e.g. “.type”, “.direction”, “File.Exists”, etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     As
    -0.07
     whether
    -0.07
     incarceration
    -0.07
     Although
    -0.07
    CAN
    -0.07
     Problems
    -0.07
    Although
    -0.06
    	free
    -0.06
    As
    -0.06
     Jeh
    -0.06
    POSITIVE LOGITS
    .deserialize
    0.07
    ướ
    0.07
    抽
    0.06
     uzak
    0.06
    .vocab
    0.06
    发
    0.06
    ΄
    0.06
     tanı
    0.06
    VD
    0.06
     назад
    0.06
    Act Density 0.099%

    No Known Activations