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

    Neuronpedia

    INDEX
    Explanations

    attends to tokens describing positive qualities from tokens indicating negative aspects or conditions

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.12
    1:0.08
    2:0.09
    3:0.49
    4:0.05
    5:0.02
    6:0.05
    7:0.07
    Negative Logits
     ſtate
    -0.45
    AndroidJUnit
    -0.44
     Efq
    -0.40
    ImageContext
    -0.39
    ViewImports
    -0.39
     houſe
    -0.38
     perſon
    -0.38
     greateſt
    -0.37
     purpoſe
    -0.36
     Theſe
    -0.36
    POSITIVE LOGITS
     betweenstory
    0.32
     détaillée
    0.31
    AccessorTable
    0.30
     فريبيس
    0.28
    departments
    0.28
     procédure
    0.27
    OrUpdate
    0.27
     diminta
    0.27
     CWE
    0.27
    boton
    0.26
    Act Density 0.603%

    No Known Activations