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

    Neuronpedia

    INDEX
    Explanations

    stability

    np_max-act · gemini-2.0-flash

    This neuron recognizes markup-like or control tokens used to delimit sections (e.g. header IDs, begin/end markers) rather than ordinary words.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     choices
    -0.07
    Auto
    -0.06
     enters
    -0.06
    Avg
    -0.06
    _trajectory
    -0.06
     absorbed
    -0.06
    ories
    -0.06
    effect
    -0.06
    omas
    -0.06
     ContentValues
    -0.06
    POSITIVE LOGITS
     накоп
    0.07
     موجود
    0.07
    �
    0.07
     strategist
    0.07
    xCF
    0.07
    隆
    0.06
    ��
    0.06
     Symposium
    0.06
    ,...
    0.06
    (Utils
    0.06
    Act Density 0.008%

    No Known Activations