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

    Neuronpedia

    INDEX
    Explanations

    cap

    np_max-act · gemini-2.0-flash

    The neuron is keyed to the sub‐token “cap,” activating whenever that three‐letter sequence appears (e.g. in “capillary,” “capstan,” “capybara,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Zh
    -0.09
    руж
    -0.07
    Ngoài
    -0.07
     Herbert
    -0.07
     Philosoph
    -0.07
    SQL
    -0.07
     Του
    -0.07
    _THREAD
    -0.07
    rdf
    -0.07
    Zh
    -0.07
    POSITIVE LOGITS
     cap
    0.15
     Cap
    0.12
     caps
    0.12
    -cap
    0.11
     Caps
    0.11
    cap
    0.10
     capped
    0.10
    _cap
    0.10
    Cap
    0.10
     cover
    0.09
    Act Density 0.014%

    No Known Activations