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

    Neuronpedia

    INDEX
    Explanations

    digit

    np_max-act · gemini-2.0-flash

    This neuron activates on tokens naming a number’s place value (e.g. “units digit,” “tens digit,” “hundreds digit,” “thousands digit,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    .Author
    -0.07
    Queryable
    -0.07
    Ber
    -0.07
     sempre
    -0.07
    TemplateName
    -0.07
    Incoming
    -0.07
     Pin
    -0.07
     Bethesda
    -0.06
     bullying
    -0.06
     Entertainment
    -0.06
    POSITIVE LOGITS
    0.07
     specifics
    0.07
     Shot
    0.07
    detail
    0.07
     крем
    0.07
    .swift
    0.06
                                    ↵
    0.06
     Ли
    0.06
     �
    0.06
    Ы
    0.06
    Act Density 0.002%

    No Known Activations