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

    Neuronpedia

    INDEX
    Explanations

    realm

    np_max-act · gemini-2.0-flash

    This neuron detects anomalous or non-standard symbol sequences—strange unicode/glyph fragments (e.g. “�י�־ו�ו”) that stand out from ordinary text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Car
    -0.08
     congestion
    -0.07
     BUS
    -0.07
     Bus
    -0.07
    uktur
    -0.07
     cars
    -0.07
    posed
    -0.07
    ItemCount
    -0.07
    .codigo
    -0.07
     positive
    -0.07
    POSITIVE LOGITS
     realm
    0.12
     Realm
    0.12
    Realm
    0.11
     realms
    0.10
    alm
    0.08
     Vault
    0.07
    realm
    0.07
     orbs
    0.07
    임
    0.07
    REP
    0.07
    Act Density 0.003%

    No Known Activations