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

    Neuronpedia

    INDEX
    Explanations

    ://

    np_max-act · gemini-2.0-flash

    The neuron activates on literal URL fragments (especially the “http://”/“https://” scheme and host parts), i.e. it detects when a token is part of a full web address.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     pruning
    -0.07
    gui
    -0.07
     tiêu
    -0.07
    ibold
    -0.07
    bow
    -0.07
    tool
    -0.06
     productId
    -0.06
     uno
    -0.06
     insol
    -0.06
    primer
    -0.06
    POSITIVE LOGITS
     ischem
    0.07
    }")
    0.07
    ViewController
    0.07
    -security
    0.07
     інститут
    0.07
     Ferguson
    0.06
     грн
    0.06
     توق
    0.06
    illac
    0.06
     TensorFlow
    0.06
    Act Density 0.010%

    No Known Activations