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

    Neuronpedia

    INDEX
    Explanations

    code errors

    np_max-act · gemini-2.0-flash

    This neuron responds to numeric version or numeric‐range tokens (e.g. “1.0.0,” “100,” “400”)—essentially detecting numbers and version‐style digit sequences in text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     kaf
    -0.07
    Lua
    -0.07
    �
    -0.06
    seudo
    -0.06
    ':"
    -0.06
    (out
    -0.06
     Αλ
    -0.06
     Santana
    -0.06
    [:]
    -0.06
    (DataType
    -0.06
    POSITIVE LOGITS
     DOC
    0.07
     München
    0.07
     الإسلام
    0.06
     العراق
    0.06
     McMaster
    0.06
     EVP
    0.06
    opoly
    0.06
     illum
    0.06
     prof
    0.06
    358
    0.06
    Act Density 0.027%

    No Known Activations