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

    Neuronpedia

    INDEX
    Explanations

    attends to tokens related to selector actions from tokens related to controls or attributes

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.07
    1:0.15
    2:0.08
    3:0.05
    4:0.17
    5:0.36
    6:0.04
    7:0.05
    Negative Logits
    is
    -0.36
    ga
    -0.35
    at
    -0.32
    études
    -0.31
     translateY
    -0.30
    kir
    -0.30
    1
    -0.30
    U
    -0.29
    RC
    -0.29
    }{\
    -0.29
    POSITIVE LOGITS
    */].
    0.41
     Caz
    0.40
    UserScript
    0.39
     invokevirtual
    0.39
     &___
    0.38
    Datuak
    0.37
     Bb
    0.37
    ImageContext
    0.36
    itecture
    0.36
    NamedQueries
    0.36
    Act Density 0.057%

    No Known Activations