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

    Neuronpedia

    INDEX
    Explanations

    Story scenes, character actions

    np_max-act · gemini-2.0-flash

    The neuron detects speaker‐label tokens (e.g. names or IDs with colons) that mark dialogue turns.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     practical
    -0.08
     Comfort
    -0.07
    �
    -0.07
     piece
    -0.07
     film
    -0.07
    聘
    -0.07
     Healing
    -0.06
    (click
    -0.06
     관계
    -0.06
     GPA
    -0.06
    POSITIVE LOGITS
    .hours
    0.06
    .role
    0.06
    vanished
    0.06
     medios
    0.06
    HasColumnName
    0.06
     Decompiled
    0.06
     στι
    0.06
     companyId
    0.06
    _suite
    0.05
     sund
    0.05
    Act Density 0.085%

    No Known Activations