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

    Neuronpedia

    INDEX
    Explanations

    abilities / communication

    np_max-act · gemini-2.0-flash

    humorous interactions and transformations involving characters in comedic scenarios.

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

    The neuron preferentially lights up on action words—especially verbs and related terms that describe magical or supernatural acts (e.g. “teleports,” “gesture,” “speak,” “obtain,” “powers,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    -thread
    -0.07
    _CID
    -0.06
    >`↵
    -0.06
     cultivated
    -0.06
    Bounding
    -0.06
     đoạn
    -0.06
    olutely
    -0.06
     Alzheimer
    -0.06
    drop
    -0.06
    recover
    -0.06
    POSITIVE LOGITS
    {}.
    0.07
    _undo
    0.07
    ,value
    0.07
     migrationBuilder
    0.07
    ']=
    0.07
    .mod
    0.07
    (main
    0.06
    .UserName
    0.06
    UILT
    0.06
    .da
    0.06
    Act Density 0.034%

    No Known Activations