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

    Neuronpedia

    INDEX
    Explanations

    Avoiding spoilers

    np_max-act · gemini-2.0-flash

    The neuron detects first-person spoiler warnings or “don’t want to spoil/tell you how” style phrases where the reviewer signals withholding plot details.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Tomb
    -0.07
    Laugh
    -0.07
    oki
    -0.06
     getCode
    -0.06
     ix
    -0.06
     BLUE
    -0.06
     marsh
    -0.06
     hog
    -0.06
     dub
    -0.06
    JA
    -0.06
    POSITIVE LOGITS
    .setTextColor
    0.06
    �
    0.06
     blockIdx
    0.06
    _CID
    0.06
    -na
    0.06
     시험
    0.06
     vont
    0.06
    bedtls
    0.06
     XCTestCase
    0.06
    aws
    0.06
    Act Density 0.101%

    No Known Activations