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

    Neuronpedia

    INDEX
    Explanations

    general text

    np_max-act · gemini-2.0-flash

    The neuron selectively activates on multiword or hyphenated compound noun phrases (e.g. “no‐brainer,” “catch‐all,” “knee‐jerk,” “permission slip,” “double bill,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Awake
    -0.07
     accents
    -0.06
     accent
    -0.06
    (Of
    -0.06
     Lazar
    -0.06
     Ober
    -0.05
     plac
    -0.05
     чувств
    -0.05
     obsc
    -0.05
    ().__
    -0.05
    POSITIVE LOGITS
    shoot
    0.08
     피해
    0.07
     IEntity
    0.07
    werk
    0.07
    lights
    0.07
     subscription
    0.07
    AsString
    0.07
    LoginForm
    0.07
    希望
    0.07
    でしょう
    0.06
    Act Density 0.078%

    No Known Activations