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

    Neuronpedia

    INDEX
    Explanations

    one

    np_max-act · gemini-2.0-flash

    The neuron detects the special header‐delimiter tokens (e.g. “<|start_header_id|>”) used to mark metadata boundaries in the chat transcript.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    _str
    -0.06
    cccc
    -0.06
    фф
    -0.06
    Specifier
    -0.06
    capability
    -0.06
     lying
    -0.06
     Skinner
    -0.06
     мил
    -0.06
    proxy
    -0.06
     impr
    -0.06
    POSITIVE LOGITS
    BuilderFactory
    0.07
     bylo
    0.07
    redirectTo
    0.07
    μένα
    0.07
     historically
    0.07
    <?>
    0.07
     součástí
    0.06
     zamanda
    0.06
    .Broadcast
    0.06
     courtyard
    0.06
    Act Density 0.056%

    No Known Activations