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

    Neuronpedia

    INDEX
    Explanations

    attends to the token "unique" from tokens marked with closing parentheses

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.10
    1:0.14
    2:0.14
    3:0.14
    4:0.13
    5:0.07
    6:0.12
    7:0.13
    Negative Logits
    essions
    -0.29
    tagHelperRunner
    -0.28
    (!__
    -0.27
    𝗾
    -0.26
    ksesta
    -0.26
    RegistryLite
    -0.25
    Бахар
    -0.25
     gainera
    -0.24
     CWE
    -0.24
     rağmen
    -0.24
    POSITIVE LOGITS
     חיצוניים
    0.34
     aDecoder
    0.34
     محفوظة
    0.28
     virtuel
    0.27
    ford
    0.25
     pare
    0.24
     comuniques
    0.24
     Comprometido
    0.23
     Wedge
    0.23
     comod
    0.23
    Act Density 0.103%

    No Known Activations