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

    Neuronpedia

    INDEX
    Explanations

    attends to punctuation or sentence-ending marks from phrases that pose questions or expressions of uncertainty

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Head Attr Weights
    0:0.11
    1:0.11
    2:0.07
    3:0.13
    4:0.15
    5:0.12
    6:0.15
    7:0.12
    Negative Logits
     Houſe
    -0.45
    ſelf
    -0.43
     Majefty
    -0.42
     myſelf
    -0.42
     themſelves
    -0.40
     houſe
    -0.40
     Anſ
    -0.40
     himſelf
    -0.38
    ſelves
    -0.37
     Jefus
    -0.36
    POSITIVE LOGITS
    BrowserModule
    0.30
     AssemblyProduct
    0.26
    principalTable
    0.24
    jsonPath
    0.24
     parti
    0.23
     Per
    0.23
    .
    0.22
     MERCHANTABILITY
    0.22
    bindParam
    0.21
     per
    0.21
    Act Density 0.007%

    No Known Activations