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

    Neuronpedia

    INDEX
    Explanations

    be

    np_max-act · gemini-2.0-flash

    This neuron detects the special metadata tokens that mark header or segment boundaries (e.g. the “<|start_header_id|>” markers) in the transcript.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    .pk
    -0.07
     Sist
    -0.07
     protr
    -0.06
     Buying
    -0.06
    ,False
    -0.06
    infos
    -0.06
    妖
    -0.06
    일본
    -0.06
    addons
    -0.06
    PBS
    -0.06
    POSITIVE LOGITS
    ながら
    0.07
     τρό
    0.06
     fetal
    0.06
    .ReadString
    0.06
    testing
    0.06
     освіт
    0.06
     كان
    0.06
     вмест
    0.06
    推
    0.06
    اتر
    0.05
    Act Density 0.069%

    No Known Activations