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

    Neuronpedia

    INDEX
    Explanations

    attends to the comma from specific tokens that denote related parts of a sentence or thought, emphasizing contrasts or connections

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.14
    1:0.52
    2:0.10
    3:0.04
    4:0.04
    5:0.03
    6:0.03
    7:0.06
    Negative Logits
    SequentialGroup
    -0.38
    minecraft
    -0.38
    <!--[
    -0.35
    underbrace
    -0.32
    URIComponent
    -0.29
     Patria
    -0.29
    roek
    -0.29
    ——–
    -0.28
     …………
    -0.28
    nesses
    -0.28
    POSITIVE LOGITS
    NOPQRST
    0.50
     librement
    0.44
     automatiques
    0.38
    Vidite
    0.37
     avoient
    0.36
    DebuggerNonUser
    0.35
    fitrión
    0.35
     रेटिंग
    0.35
     Betracht
    0.35
    UnknownFields
    0.34
    Act Density 0.388%

    No Known Activations