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

    Neuronpedia

    INDEX
    Explanations

    attends to numerical tokens representing data or counts from accompanying tokens that indicate references or data indices

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.08
    1:0.13
    2:0.27
    3:0.05
    4:0.08
    5:0.21
    6:0.03
    7:0.11
    Negative Logits
     interest
    -0.28
    キ
    -0.26
     interests
    -0.26
    wh
    -0.24
     last
    -0.24
     sola
    -0.24
    last
    -0.23
     Gil
    -0.23
    i
    -0.23
    フェ
    -0.23
    POSITIVE LOGITS
    fromnode
    0.57
    RegressionTest
    0.51
     GenerationType
    0.49
    CloseOperation
    0.49
    StoryboardSegue
    0.46
    UnusedPrivate
    0.46
    AddTagHelper
    0.46
     Paglinawan
    0.46
     onCreateView
    0.45
     betweenstory
    0.45
    Act Density 0.079%

    No Known Activations