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

    Neuronpedia

    INDEX
    Explanations

    attends to pairs of adjacent tokens that are connected by a punctuation mark from separated tokens that are further apart in the sequence

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.11
    1:0.09
    2:0.07
    3:0.11
    4:0.08
    5:0.07
    6:0.16
    7:0.26
    Negative Logits
     esternos
    -0.40
    AndEndTag
    -0.35
     AssemblyCulture
    -0.35
    SharedDtor
    -0.33
     <=",
    -0.32
    parsedMessage
    -0.31
    AsUp
    -0.31
    RenderAtEndOf
    -0.31
     سكانية
    -0.31
    __*/
    -0.30
    POSITIVE LOGITS
     Reverso
    0.28
     发表于
    0.27
    Vol
    0.24
     Vol
    0.23
     Sil
    0.22
     vol
    0.22
    inot
    0.20
    ↵↵
    0.20
     Mas
    0.20
     i
    0.20
    Act Density 0.001%

    No Known Activations