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

    Neuronpedia

    INDEX
    Explanations

    gradient

    np_max-act · gemini-2.0-flash

    This neuron fires on occurrences of “gradient” (especially in “gradient descent”), i.e. it recognizes gradient-related terminology.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     mobs
    -0.08
    voucher
    -0.07
    ubuntu
    -0.07
    umor
    -0.06
    -abs
    -0.06
    告
    -0.06
    165
    -0.06
    richText
    -0.06
    ulators
    -0.06
    766
    -0.06
    POSITIVE LOGITS
     innovative
    0.07
    (Sprite
    0.07
    .HORIZONTAL
    0.07
    �
    0.07
     estamos
    0.06
     dejtings
    0.06
     habil
    0.06
    Coordinates
    0.06
    rou
    0.06
     ослож
    0.06
    Act Density 0.002%

    No Known Activations