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

    Neuronpedia

    INDEX
    Explanations

    success or failure

    np_max-act · gemini-2.0-flash

    This neuron activates on discourse‐structuring and modal/connective tokens (e.g. “if,” “however,” “instead,” “would,” “might,” “in conclusion,” etc.), essentially picking up argument flow markers.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     Package
    -0.06
     Şubat
    -0.06
     cuối
    -0.06
    Sweden
    -0.06
    -archive
    -0.06
    iya
    -0.06
     Feed
    -0.06
     partisan
    -0.06
     culture
    -0.06
     смесь
    -0.06
    POSITIVE LOGITS
     Dealer
    0.07
    gil
    0.07
     Alg
    0.06
     configparser
    0.06
     insp
    0.06
    比
    0.06
    内
    0.06
     Malay
    0.06
    )initWithFrame
    0.06
     Ї
    0.06
    Act Density 0.042%

    No Known Activations