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

    Neuronpedia

    INDEX
    Explanations

    attends to specific tokens that signal larger concepts or ideas from more general tokens

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.09
    1:0.11
    2:0.10
    3:0.07
    4:0.07
    5:0.02
    6:0.16
    7:0.34
    Negative Logits
     resourceCulture
    -0.30
    Hentet
    -0.26
    expandindo
    -0.24
     ExecuteAsync
    -0.24
     فريبيس
    -0.22
    형
    -0.22
     taj
    -0.22
     quanto
    -0.22
    intur
    -0.21
     ushort
    -0.21
    POSITIVE LOGITS
    +:+
    0.37
    HasAnnotation
    0.32
    COUVER
    0.31
    Referencies
    0.31
     validamos
    0.30
    utilisons
    0.30
    Espèce
    0.29
     HasFactory
    0.29
     nonatomic
    0.29
     betweenstory
    0.29
    Act Density 0.834%

    No Known Activations