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

    Neuronpedia

    INDEX
    Explanations

    lean

    np_max-act · gemini-2.0-flash

    The neuron activates specifically on the tokens “leaned” plus “in,” i.e. it detects occurrences of the phrase “leaned in.”

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Namen
    -0.08
     pageInfo
    -0.07
    üssen
    -0.07
     时间
    -0.07
    pective
    -0.06
    inspace
    -0.06
    /day
    -0.06
    mach
    -0.06
     chatte
    -0.06
    ayıp
    -0.06
    POSITIVE LOGITS
     leaned
    0.12
     leaning
    0.10
     leans
    0.07
     Κου
    0.07
    jured
    0.07
     driven
    0.06
    efined
    0.06
    Jean
    0.06
     Glenn
    0.06
     XT
    0.06
    Act Density 0.003%

    No Known Activations