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

    Neuronpedia

    INDEX
    Explanations

    support personnel

    np_max-act · gemini-2.0-flash

    This neuron activates on the “assistant” role token marking the start of an AI assistant’s response.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    Python code related to API requests and handling command strings.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8

    the assistant role/header marker in chat-style transcripts (i.e., metadata indicating an assistant message)

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
    .Internal
    -0.07
    \Url
    -0.07
    weapon
    -0.07
    /fw
    -0.06
     twig
    -0.06
    �
    -0.06
     život
    -0.06
    áním
    -0.06
     ApiService
    -0.06
     Beta
    -0.06
    POSITIVE LOGITS
    sync
    0.06
     PL
    0.06
     trous
    0.06
    sch
    0.06
     follic
    0.06
    chin
    0.06
     grading
    0.06
    163
    0.06
    핑
    0.06
    高速
    0.06
    Act Density 0.068%

    No Known Activations