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

    Neuronpedia

    INDEX
    Explanations

    attends to tokens indicating a relationship or transition from an earlier context to a later setting

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.11
    1:0.12
    2:0.15
    3:0.08
    4:0.10
    5:0.06
    6:0.12
    7:0.22
    Negative Logits
     tantôt
    -0.26
    wój
    -0.25
     массы
    -0.24
    KeySpec
    -0.24
     Wilson
    -0.24
    étais
    -0.23
     Bois
    -0.23
    كوين
    -0.23
    IRED
    -0.23
     WILSON
    -0.23
    POSITIVE LOGITS
    hoeddwyd
    0.41
    addCriterion
    0.40
    exitRule
    0.40
    oneofs
    0.38
    Rujuakan
    0.38
    دانشنامهٔ
    0.37
     незавершена
    0.37
    fjspx
    0.37
    DockStyle
    0.37
    msgTypes
    0.36
    Act Density 0.395%

    No Known Activations